Topological connection between vesicles and nanotubes in ...

48
HAL Id: hal-03399741 https://hal.archives-ouvertes.fr/hal-03399741 Preprint submitted on 24 Oct 2021 HAL is a multi-disciplinary open access archive for the deposit and dissemination of sci- entific research documents, whether they are pub- lished or not. The documents may come from teaching and research institutions in France or abroad, or from public or private research centers. L’archive ouverte pluridisciplinaire HAL, est destinée au dépôt et à la diffusion de documents scientifiques de niveau recherche, publiés ou non, émanant des établissements d’enseignement et de recherche français ou étrangers, des laboratoires publics ou privés. Topological connection between vesicles and nanotubes in single-component lipid membranes driven by head-tail interactions Niki Baccile, Cédric Lorthioir, Abdoul Ba, Patrick Le Griel, Cristina Diogo, Javier Perez, Wim Soetaert, Sophie Roelants To cite this version: Niki Baccile, Cédric Lorthioir, Abdoul Ba, Patrick Le Griel, Cristina Diogo, et al.. Topological connection between vesicles and nanotubes in single-component lipid membranes driven by head-tail interactions. 2021. hal-03399741

Transcript of Topological connection between vesicles and nanotubes in ...

HAL Id hal-03399741httpshalarchives-ouvertesfrhal-03399741

Preprint submitted on 24 Oct 2021

HAL is a multi-disciplinary open accessarchive for the deposit and dissemination of sci-entific research documents whether they are pub-lished or not The documents may come fromteaching and research institutions in France orabroad or from public or private research centers

Lrsquoarchive ouverte pluridisciplinaire HAL estdestineacutee au deacutepocirct et agrave la diffusion de documentsscientifiques de niveau recherche publieacutes ou noneacutemanant des eacutetablissements drsquoenseignement et derecherche franccedilais ou eacutetrangers des laboratoirespublics ou priveacutes

Topological connection between vesicles and nanotubesin single-component lipid membranes driven by head-tail

interactionsNiki Baccile Ceacutedric Lorthioir Abdoul Ba Patrick Le Griel Cristina Diogo

Javier Perez Wim Soetaert Sophie Roelants

To cite this versionNiki Baccile Ceacutedric Lorthioir Abdoul Ba Patrick Le Griel Cristina Diogo et al Topologicalconnection between vesicles and nanotubes in single-component lipid membranes driven by head-tailinteractions 2021 hal-03399741

1

Topological connection between vesicles and nanotubes in single-1

component lipid membranes driven by head-tail interactions 2

3

Niki Baccilea Ceacutedric Lorthioira Abdoul Aziz Baa Patrick Le Griela Cristina Coelho Diogob 4

Javier Perezc Wim Soetaertd Sophie L K W Roelantsd 5

6

a Sorbonne Universiteacute Centre National de la Recherche Scientifique Laboratoire de Chimie de 7

la Matiegravere Condenseacutee de Paris LCMCP F-75005 Paris France 8

b Institut des Mateacuteriaux de Paris Centre Sorbonne Universiteacute Paris France 9

c Synchrotron Soleil LrsquoOrme des Merisiers Saint-Aubin BP48 91192 Gif-sur-Yvette Cedex 10

France 11

d InBio Department of Biotechnology Ghent University Ghent Belgium 12

13

Corresponding author 14

Dr Niki Baccile 15

E-mail address nikibaccilesorbonne-universitefr 16

Phone +33 1 44 27 56 77 17

2

Abstract 18

Lipid nanotube-vesicle networks are important channels for intercellular communication and 19

transport of matter Experimentally observed in neighboring mammalian cells but also 20

reproduced in model membrane systems a broad consensus exists on their formation and 21

stability Lipid membranes must be composed of at least two components each stabilizing low 22

(generally a phospholipid) and high curvatures Strong anisotropy or enhanced conical shape 23

of the second amphiphilic component is crucial for the formation of nanotunnels Anisotropic 24

driving forces generally favor nanotube protrusions from vesicles In the present work we 25

report the unique case of topologically-connected nanotubes-vesicles obtained in the absence 26

of directional forces in single-component membranes composed of an anisotropic bolaform 27

glucolipid above its melting temperature Tm Cryo-TEM and fluorescence confocal 28

microscopy show the interconnection between vesicles and nanotubes in a single-phase region 29

between 60deg and 90degC under diluted conditions Solid-state NMR necessarily recorded on the 30

dry powder yet demonstrates that the glucolipid can simultaneously assume two distinct 31

configurations head-head and head-tail These arrangements seemingly of comparable energy 32

above the Tm could explain the existence and stability of the topologically-connected vesicles 33

and nanotubes which are generally not observed for classical single-component phospholipid-34

based membranes above their Tm 35

36

Keywords Nanotube vesicle networks Tunnelling nanotubes Block liposomes Liposomes 37

Lipid nanotubes Biosurfactants Microbial glycolipids 38

39

Introduction 40

Topological connections between closed lipidic compartments through nanotubes1ndash3 41

have been shown to play a crucial role in the transfer of matter and communication in 42

neighboring mammalian cells4 These singular nanosystems observed since the lsquo90s as 43

spontaneous non-equilibrium structures in electroformed model liposome membranes56 have 44

since been largely studied both experimentally and theoretically7ndash9 Addressed in the literature 45

by different terms tunnelling nanotubes (TNT)149 block liposomes10ndash12 or nanotube-vesicle 46

networks13ndash16 (instead of tubes some work speak of tethers617) all refer to a similar 47

phenomenon driven by various internal or external forces The latter must overcome the energy 48

barrier needed to bend a phospholipid bilayer from low positive mean and gaussian curvatures 49

(vesicle) to a high mean and zero Gaussian (tube) curvatures 50

3

A large body of both experimental and theoretical work has shown that budding and 51

eventual nanotube formation from an existing membrane can only occur spontaneously for 52

membranes of at least two-components9102425111218ndash23 and below a vesicle-nanotube phase 53

transition event2627 Otherwise external anisotropic driving forces like electroformation5ndash54

82829 osmotic pressure6 laser ldquotweezersrdquo30 or electrodynamics13ndash16 must be employed to obtain 55

similar results with in some cases an impressive degree of 2D and 3D organization13ndash16 The 56

origin of spontaneous nanotube formation has been shown to be related to a nanoscale phase 57

separation between two membrane components stabilizing low and high curvatures 58

respectively10ndash1231 with at least one component being highly anisotropic203233 Theoretically 59

this behavior has been explained by deviations in the elastic properties of membranes due to in-60

plane orientational ordering of membrane inclusions composed of anisotropic amphiphiles 61

these referring to a non-symmetrical shape upon a 90deg tilt along the amphiphile axis8919242533 62

In this work we show unexpected nanotubing of membranes prepared from a single-63

component lipid in the absence of external directional forces and above the lipidrsquos Tm This 64

phenomenon is observed for a novel anisotropic double amphiphile (bolaform amphiphile or 65

bolaamphiphile) a glucolipid composed of β-D-glucose and a C181-cis fatty alcohol (G-66

C181-OH Figure 1) This compound is obtained by microbial fermentation of a genetically-67

modified S bombicola yeast in the presence of oleyl alcohol34 and is developed in the broader 68

context of extending the library of new biobased surfactants and lipids in view of replacing 69

petrochemical low molecular weight amphiphiles35ndash41 The structure of G-C181-OH is 70

analogous to that of other microbial glycolipids developed through genetic engineering4243 71

Topological connections between nanotubes and vesicles are observed by means of 72

cryogenic transmission electron microscopy (cryo-TEM) fluorescence microscopy and wide 73

angle X-ray scattering (WAXS) above the melting temperatures Tm= 483degC under strong 74

dilution (05 wt) Spin diffusion and 1H double-quantum single-quantum (DQ-SQ) 2D solid-75

state nuclear magnetic resonance (ssNMR) spectroscopy under magic angle spinning (MAS) 76

necessarily performed on the dry glucolipid powder provides a spectroscopic ground to 77

understand the vesicle-nanotube coexistence The bolaform glucolipids could be in a head-78

headtail-tail configuration in the vesicles but rather in a head-tail configuration in the 79

nanotubes In the dry glucolipid both configurations are energetically comparable in the single-80

phase region above Tm and their coexistence is evidenced by the splitting of some 13C NMR 81

signals These facts could explain the stability of nanotubes while the following hypotheses are 82

formulated for their formation membrane inclusions with different orientational ordering13ndash16 83

possibly driven by inter-vesicle collisions 84

4

85

86

87

88

89

Figure 1 ndash Non acetylated C181 alcohol glucoside G-C181-OH is obtained by a bioprocess performed 90

with modified S bombicola yeast 91

92

Experimental Section 93

Synthesis of non acetylated C181 Alcohol Glucosides (G-C181-OH) G-C181-OH (Mw= 94

41856 gmol-1) was produced by aerobic whole cell bioprocess with a modified S bombicola 95

strain as described by Van Renterghem et al (Fig S4 in Ref 34) The molecule was purchased 96

from the Bio Base Europe Pilot Plant (Gent Belgium) and has the generalized chemical 97

structure given in figure Figure 1 The HPLC and 1H NMR spectrum (MeOD-d4) with peak 98

assignment are shown in Figure S 1 High purity levels (99) and high degree of uniformity 99

were obtained as can be derived from HPLC-ELSD chromatogram 1H NMR and table of 100

contaminant given in Figure S 1 101

102

1H solution Nuclear Magnetic Resonance (NMR) 1H solution NMR experiments were 103

performed on a Bruker Avance III 300 spectrometer using a 5 mm 1H-X BBFO probe using 104

methanol-d4 as solvent The number of transients is 8 with 3 s recycling delay an acquisition 105

time of 546 s and a receiver gain of 362 The 1H NMR spectrum and relative assignment are 106

shown in Figure S 1 and Table S 1 13C solution NMR were performed on the same probe using 107

DMSO-d6 Corresponding chemical shifts are given in Table 1 Referencing is done with 108

respect to TMS δ1H= 0 ppm δ13C= 0 ppm 109

110

Sample preparation The sample was dissolved in milliQ-grade water at the concentration of 5 111

mgmL (05 wt) Lack of pH-sensitive probes (eg COOH groups) in G-C181-OH as 112

otherwise found in other microbial amphiphiles44 but also the will to avoid ion-specific 113

effects45 exclude the use of buffer The solution was heat at the given temperature using a CH3-114

150 Combitherm-2 dry block heating device For the cryo-TEM experiments the vials were 115

heat at the given temperature next to the cryofixation device and plunged into liquid ethane as 116

fast as possible For temperatures of 25degC and 70degC the sample was heat at 90degC and then 117

Synthesis path

5

cooled at to the desired temperature For the SAXS experiment the sample was heat at 90degC in 118

the block heating device and cooled at to 25degC before analysis For the SANS experiment the 119

sample was heat at 90degC directly in front of the beam using the thermalized sample holder 120

available 121

122

Differential Scanning Calorimetry (DSC) DSC was performed using a DSC Q20 apparatus 123

from TA Instruments equipped with the Advantage for Q Series Version acquisition software 124

(v540) Acquisition was performed on both hydrated and dry powder sample (~ 3-5 mg) sealed 125

in a classical aluminium cup and using an immediate sequence of heating and cooling ramps at 126

a rate of 10degCmin-1 Melting temperatures Tm 1 and 2 Tm1 and Tm2 were taken at the minimum 127

of the endothermic peak 128

129

Small and Wide Angle Scattering experiments Small angle neutron scattering (SANS) 130

experiments were performed at the D11 beamline of Institut Laue Langevin (Grenoble France) 131

during the run No 9-13-778 Four q-ranges have been explored and merged using the following 132

wavelengths λ and sample-to-detector (StD) distances 1) ultra-low q λ= 135Aring StD= 39 m 133

2) low-q λ= 53Aring StD= 39 m 3) mid-q λ= 53Aring StD= 8 m 4) high-q λ= 53Aring StD= 14 m 134

The sample (C= 5 mgmL-1) was prepared in 999 D2O to limit the incoherent background 135

scattering The sample solution was analyzed in standard 1 mm quartz cells Direct beam empty 136

quartz cell H2O (incoherent scatterer) within the quartz cell were recorded and boron carbide 137

(B4C) was used as neutron absorber The sample acquisition was measured at 90degC where 138

temperature was controlled through the controller thermalized sample holder available at the 139

beamline The background sample (D2O) signal was subtracted from the experimental data 140

Absolute values of the scattering intensity were obtained from the direct determination of the 141

number of neutrons in the incident beam and the detector cell solid angle The 2D raw data were 142

corrected for the ambient background and empty cell scattering and normalized to yield an 143

absolute scale (cross section per unit volume) by the neutron flux on the samples The data were 144

then circularly averaged to yield the 1D intensity distribution I(q) The software package Grasp 145

(developed at ILL and available free of charge) was used to integrate the data while the 146

software package SAXSUtilities (developed at ESRF and available free of charge) was used to 147

merge the data acquired at all configurations and subtract the background 148

Wide-angle X-ray scattering (WAXS) was performed under temperature control at the 149

SWING beamline of SOLEIL synchrotron facility (Saint-Aubin France) during the run 150

20201747 (energy 14 keV sample-to-detector distance 05 m) Two sample concentrations in 151

6

H2O were probed C= 5 mgmL-1 and C= 25 mgmL-1 The 2D data were integrated azimuthally 152

at the beamline using the software Foxtrot and in order to obtain the I(q) vs q spectrum after 153

masking the beam stop shadow Silver behenate (d(100) = 5838 Aring) was used as a standard to 154

calibrate the q-scale Sample solutions were inserted in borosilicate capillaries of 15 mm in 155

diameter Capillaries were flame-sealed A capillary oven with controlled temperature (plusmn 05degC) 156

provided at the beamline was used to control the sample temperature between 25degC and 111degC 157

Data were normalized by the transmission and calibrated to the SAXS signal of H2O at large q-158

values (I= 00163 cm-1) in order to obtain an absolute intensity scale The water signal was 159

measured by subtracting the signal of the empty capillary from the signal of a water-filled 160

capillary The signal of (water + capillary) was used as background for the samples and it was 161

subtracted after integration of the 2D data 162

163

Cryogenic transmission electron microscopy (cryo-TEM) Cryo-TEM experiments were carried 164

out on an FEI Tecnai 120 twin microscope operating at 120 kV and equipped with a Gatan 165

Orius CCD numeric camera The sample holder was a Gatan Cryoholder (Gatan 626DH 166

Gatan) Digital Micrograph software was used for image acquisition Cryofixation was 167

performed using a homemade cryofixation device The solutions were deposited on a glow-168

discharged holey carbon coated TEM copper grid (Quantifoil R22 Germany) Excess solution 169

was removed and the grid was immediately plunged into liquid ethane at -180degC before 170

transferring them into liquid nitrogen All grids were kept at liquid nitrogen temperature 171

throughout all experimentation Cryo-TEM images have been treated and analyzed using Fiji 172

(is just ImageJ) software available free of charge at the developerrsquos website46 173

The samplersquos concentration was C= 5 mgmL-1 For technical reasons the temperature 174

during cryofixation could not be controlled For this reason the temperature associated with 175

each cryo-TEM image in the text was related to the equilibration temperature just before 176

cryofixation the device being as close as possible to the dry bath used for temperature control 177

Since temperature control during cryofixation was not possible in our device and practically 178

impossible at high temperatures one should take the T values as indicative of the given value 179

Although the sample transfer between the dry bath and the cryofixation device was performed 180

as fast as possible it is more than likely that the actual temperature of the sample at the moment 181

of cryofixation was lower than the one indicated here although impossible to know precisely 182

183

7

Fluorescence microscopy Images were recorded using a 40x objective on a Leica SP5 upright 184

confocal microscope with 405 nm laser excitation The sample was marked with 6-dodecanoyl-185

NN-dimethyl-2-naphthylamine (Laurdan Mw= 35354 gmol-1) fluorophore as follows A 186

stock solution of 3 mg of Laurdan in 100 microL of acetone was initially prepared (C= 30 mgmL-187

1) A volume of 028 microL of the stock Laurdan solution was diluted in 1 mL of a 5 mgmL-1 G-188

C181-OH water solution temporary set at 110degC in a sealed Eppendorf The latter was 189

important so to enhance inclusion of Laurdan in the membrane constituted by the glucolipid 190

while in its vesicle phase The final Laurdan concentration was approximately 24 micromol with a 191

molar Laurdan-to-G-C181-OH molar ratio of approximately 1-to-500 Fluorophore-to-lipid 192

ratios above 1200 were generally considered as non-interfering with the lipid phase behavior 193

as also verified in this work 194

Laurdan can be excited between 365 nm and 410 nm and its emission wavelength can 195

vary according to the rigidity of the membrane intercalation within a rigid or flexible 196

membrane generates emissions at 440 nm or 490 nm respectively47 In the present experiment 197

we have flame-sealed a solution of G-C181-OH at 5 mgmL-1 containing Laurdan (1-to-500 in 198

lipid-to-Laurdan ratio) in flat optical capillaries of 01 mm thickness Capillaries were set on a 199

Linkam support for temperature control eventually placed under the microscope The sample 200

was excited at 405 nm and detection was made in spectral mode with a 5 nm window sliding 201

every 5 nm from 412 nm to 602 nm in 16 bits which allows quantify the signal and compare 202

the different conditions Quantitative analysis was performed according to the literature4849 203

several ROI (range of interest) have been drawn on several images using the ROI manager of 204

Fiji (is just ImageJ) software The integrated intensity of each ROI at 440 nm and 490 nm was 205

eventually recorded and used to calculate the value of the generalized polarization GP as 206

follows of 207

119866119875 =(120582440 minus 120582490)

120582440 + 120582490 208

A negative GP indicates a flexible membrane while a positive GP indicates a rigid membrane 209

210

Solid-state Nuclear Magnetic Resonance (ssNMR) 1H and 13C ssNMR experiments have been 211

performed using an Avance III HD Bruker 705 T (ν1H= 300 MHz) spectrometer and a 4 mm 212

magic angle spinning (MAS) probe except for the BAck-to-BAck (BABA) experiment which 213

was performed on an Avance III Bruker 1644 T (ν1H= 700 MHz) spectrometer equipped with 214

25 mm MAS probe Temperature was controlled through a Smart cooler BCUIIa BCU-xtreme 215

unit using 10 min of equilibration time after each temperature variation Temperature was 216

8

always varied from high (110degC) to low (70degC) before analysis Due to combined use of fast 217

MAS rates necessary to achieve high resolution spectra and heating close to the boiling point 218

of water neither diluted nor concentrated samples could be used for ssNMR experiments All 219

tests employing up to 50 wt G-C181-OH in water resulted in sample centrifugation drying 220

but also expulsion from the rotor causing sudden stop of rotor spinning For these reasons all 221

attempts to study hydrated G-C181-OH failed and only the dry powder could be analyzed The 222

dry sample powder was analyzed as such in zirconia rotors (4 mm or 25 mm) spinning at MAS= 223

10 kHz (4 mm) or 22 kHz (25 mm) and Chemical shifts were calibrated with respect to 224

adamantane (13C 13C(CH2)= 3848 ppm = 191 ppm) and glycine alpha (13C(CO)= 17603 225

ppm) Details on the acquisition parameters are given in the Supporting Information 226

227

Results and discussion 228

Topologically-connected nanotubes and vesicles 229

G-C181-OH is a biobased glucolipid water-insoluble at room temperature and of 230

which the physicochemical properties are poorly known The DSC thermogram performed at 231

10degCmin of G-C181-OH as a function of the water content (Figure 2a) shows two endothermic 232

peaks on the dry powder (996 wt glucolipid) They are attributed to two melting phenomena 233

occurring at the following temperatures Tm 483degC (Tm1) and 957degC (Tm2) the enthalpy of the 234

former being lower than that of the latter Interestingly DSC also shows that Tm1 is hydration 235

independent and reversible (the associated exothermic peak is systematically observed) while 236

the transition at Tm2 is simply suppressed (here tested up to 130degC) on hydrated powders and 237

never observed on the cooling profile even at slow cooling rates (1degCmin) The thermogram 238

recorded on the dry powder is typical of single-component lipids having an intermediate50 239

metastable phase (often referred to the ripple phase Pβ´ in reference to the periodic undulation 240

of the bilayer)50ndash54 between the classical lamellar gel (Lβ) and liquid crystalline (Lα) phases50ndash241

54 and characterized by regions of both liquid order and disorder5455 The effect of water does 242

not influence the low-temperature crystallization at Tm1 but it contributes to disrupt the ordered 243

array of hydrocarbon chains in the metastable phase 244

On the basis of the DSC thermogram the aqueous phase behavior of G-C181-OH under 245

diluted conditions (5 mgmL-1) was studied at four different temperatures (red marks on Figure 246

2a) below and above Tm1 but also between Tm1 and Tm2 corresponding to the thermogram of 247

the dry powder 248

9

249

Figure 2 ndash a) DSC thermogram of G-C181-OH under dry (04 wt water estimated by thermogravimetric 250

analysis) and hydrated conditions recorded at a heating rate of 10degCmin Red marks indicate the 251

temperatures at which cryo-TEM experiments are performed on the corresponding 5 mgmL aqueous 252

solutions b) Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 130degC 253

90degC 70degC and 25degC T indicates the temperature of the sample solution before cryofixation Tne T T 254

being the temperature of the sample upon plunging in liquid ethane Please refer to the materials and 255

methods section for more information 256

T= 130degC

T= 90degC

T= 70degC

T= 25degC

a)

b)

20 40 60 80 100 120

99 wt

50 wt

25 wt

T degC

En

do

10 wt

Tm1= 483 degC

Tm2= 957 degC

10

257

The series of cryo-TEM images in Figure 2b show an overview of the most relevant 258

structures observed on G-C181-OH solution from T= 130degC to T= 25degC whereas T is 259

indicative of the equilibrium temperature before cryofixation (refer to materials and methods 260

section for more information on the sense and interpretation of T) At Tgt 100degC the sample 261

is essentially composed of polydisperse single wall vesicles massively shown by the 262

complementary pictures presented in Figure S 2 In between Tm1 and Tm2 at T= 90degC and T= 263

70degC (Figure 2b Figure 3 Figure 4 complemented by Figure S 3 and Figure S 4) vesicles are 264

connected by nanotubes At Tlt Tm1 flat crystals are otherwise observed (Figure 2b and Figure 265

S 5) 266

267

268

Figure 3 ndash Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 90degC 269

270

A deeper insight at T= 90degC is given in Figure 3 (more images are provided in Figure 271

S 3) Arrows 1 through 6 mark very clearly the typical continuity in the glucolipid membrane 272

between the vesicle and nanotubes Arrow Ndeg 7 shows that a single nanotube connects the outer 273

membrane of two vesicles several hundred nanometers apart The connection is shown by 274

2

34

5

6

6

1

5

7

A

B

D

C

0 10 20 30 40 50 60

BIn

tensity

au

Profile nm

A

0 10 20 30 40 50 60

Inte

nsity

au

Profile nm

C

D

T= 90degC

11

arrows Ndeg5 and Ndeg6 The corresponding profiles A through D confirm the tubular nature of 275

the elongated objects Profiles B and D are both characterized by typical symmetrical thick 276

walls (low intensity depths highlighted by shaded rectangles) surrounding a hollow region (high 277

intensity peak between the shaded rectangles) The wall thickness varies between 35 and 40 278

nm for profile B and between 41 and 61 nm for profile D while the total nanotube diameter 279

varies between 14 nm at B and 20 nm at D At the same time the thickness of the vesicle wall 280

is about 6 nm at A and 135 nm at C Considerations about the bilayer or interdigitated structure 281

of the membrane will be presented below 282

The connection between nanotubes and vesicles seems to be multiple For instance 283

arrow Ndeg1 (Figure S 3) and Ndeg4 (Figure 3) show a single-wall vesicle of which the membrane 284

is extruded into a nanotube Similarly a closer look at arrows Ndeg1 2 and 3 in Figure 3 show 285

that the vesicle is single-wall and fully closed while the nanotube nucleates at the vesicle outer 286

surface with no apparent topological continuity In many regions of the sample at T= 90degC 287

one can find exvaginations or buds232531 in the nanotube wall This is shown for instance by 288

arrows Ndeg7 and particularly put in evidence in Figure 4 by arrows Ndeg1 through Ndeg4 where the 289

amplitude of the oscillation varies between 20-25 nm (arrow Ndeg1) and 6-8 nm (arrows Ndeg3-4) 290

291

Figure 4 - Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heated at T= 90degC 292

293

Still between Tm2 and Tm1 but at lower temperature (T= 70degC) the nanotubes seem to 294

vary from single threads (arrow Ndeg1 Figure S 4) to nanotubes of diameter below 15 nm both 295

connecting nanotubes of diameter ranging between 50 nm and 100 nm (arrows Ndeg2 Figure S 296

4) The larger tubes are most likely formed by the inflation and flattening of the nanotubes 297

probably after assimilation of surrounding vesicles and fusion of the exvagination discussed 298

above The transition between vesicles tubes and flat membranes is certainly more accentuated 299

Arrow Ndeg3 in Figure S 4 shows distinct vesicles while arrow Ndeg4 identifies a flat structure 300

evidently formed from fused vesicles In particular arrow Ndeg5 points at the fusion of two 301

2

1 132

4

12

vesicles while arrow Ndeg6 points at two vesicles fusing into the flat membrane Additional areas 302

where vesicle fusion is prominent are indicated by the oscillations at the outer skirt of broad 303

lamellae and pointed by arrow Ndeg7 304

Finally below Tm1 at T= 25degC for instance mainly flat crystals with no visible 305

exvaginations or tubes are visible (Figure S 5) by cryo-TEM and the material turns into an 306

insoluble precipitate Massive presence of flat crystalline structures is also clear from the bulk 307

iridescent behavior of a G-C181-OH solution under manual shear 308

In summary despite the uncertainty related to the actual state of equilibrium of the 309

glucolipids at the moment of cryofixation cryo-TEM depicts a vesicle phase at Tgt 100degC a 310

flat lamellar phase below Tm1 and an interconnected vesicle-nanotube network between Tm1 and 311

about 100degC The combination of statistically-relevant fluorescence microscopy and 312

scatteringdiffraction experiments provides an additional proof of the coexistence of 313

vesiclenanotube systems between the two transition temperatures Please note that the 314

experiments reported below have been performed under controlled conditions of temperature 315

which is then indicated by the use of T instead of T 316

13

317

Figure 5 ndash a-c) Confocal fluorescence microscopy images recorded at 90degC and 70degC for a 5 mgmL G-318

C181-OH aqueous solution d-g) Typical fluorescence microscopy images recorded at T= 80degC using 319

Laurdan (refer to materials and methods section for more details on this fluorophore) Excitation 320

wavelength is 405 nm for all the images while emission wavelength is 490 nm in (df) and 440 nm in (eg) 321

h) Full emission spectra obtained from the integration of various fluorescence images recorded from 412 322

nm to 600 nm (step 5 nm) at 90degC 80degC and 70degC i) Box plot of GP defined as 119918119927 =(120640120786120786120782minus120640120786120791120782)

120640120786120786120782+120640120786120791120782 with 120640120786120786120782 323

and 120640120786120791120782 being the emission wavelengths at 440 nm and 490 nm at 90degC 80degC and 70degC Each data point 324

l= 490 nm l= 440 nm

l= 490 nm l= 440 nm

10 μm400 450 500 550 600

00

02

04

06

08

10

No

rma

lize

d I

nte

nsity

l nm

90degC

80degC

80degC

70degC

70degC

400 450 500 550 60000

02

04

06

08

10

No

rma

lize

d I

nte

nsity

l nm

90degC

80degC

80degC

70degC

70degC

d) e)

f) g)

h)

i)

T= 80degC

Laurdan

T= 80degC

T= 80degC T= 80degC

T= 90degC T= 70degC T= 70degC

a) b) c)

1

1

1

14

corresponds to a set of regions of interest (ROIs) (54 at 90degC 65 at 80degC and 75 at 70degC) selected in the 325

corresponding fluorescence images ROIs were representatively selected so to include the entire image but 326

also vesicles alone tubes alone and mixtures of vesicles and tubes 327

328

Figure 5ab shows two typical confocal fluorescence microscopy images recorded on a 329

G-C181-OH solution (5 mgmL) in a flame-sealed capillary heat at T= 90degC and T= 70degC 330

Although the resolution of optical microscopy is not comparable with cryo-TEM the sample is 331

broadly composed of vesicles coexisting with nanotubes whereas in some cases the nanotube 332

starts from the vesicle itself (arrow Ndeg1) in good agreement with cryo-TEM data Furthermore 333

SANS recorded at T= 90degC (Figure S 6a) shows a scattering profile with a pronounced -2 334

dependence of the scattering intensity against the wavevector q in the log-log scale This 335

indicates the massive presence of a flat interface and it is expected typical of bilayers in 336

vesicular morphologies56 However the typical signature of nanotubes cannot be observed in 337

the SANS profile and this can be explained by both an argument related to the relative lower 338

amount of nanotubes with respect to vesicles at 90degC but also to the steeper slope of flat 339

morphologies (-2) compared to tubes (-1) the signal of the former masking the tube signal at 340

low wavevector values The SANS profile is characterized by a diffraction peak at q= 0167 Aring-341

1 of which the assignment is refined by WAXS data 342

Temperature-driven WAXS experiments in Figure S 6bc (5 mgmL-1 and 25 mgmL-1 343

respectively) show a lamellar order possibly attributed to a multilamellar structure of the 344

vesicles (dv Figure S 6d) at high temperature and to the interplanar arrangement of the 345

crystalline flat structures (dL Figure S 6d) at T= 25degC The sharp (001) reflection indicates 346

typical interlamellar distance of 353 Aring at 25degC although this value varies with temperature 347

(400 Aring at 111degC 372 Aring at 88degC) thus confirming its assignment to a hydration-dependent 348

interlamellar order More intriguing is the second broader peak at q= 0403 Aring-1 (d= 156 Aring) 349

and its reflection both clearly visible at 88degC below Tm2 We attribute it to the inter-glucolipid 350

arrangement inside the tubular phase (dt Figure S 6d) coexisting with the vesicle phase 351

between Tm1 and Tm2 and to the planar lipid arrangement (di Figure S 6d) in the rigid flat 352

lamellar structure below Tm2 Its temperature- hence hydration- independent position also 353

confirms its structural origin Finally besides the water-structure peak at about 2 Aring-1 we also 354

observe a broad peak at 144 Aring-1 characteristics of the lateral packing (d Figure S 6d) of lipids 355

in bilayers While its position could be consistent with an Lα phase5758 above Tm2 its broadness 356

and positional invariance as well as the coexistence between vesicles and tubes do not allow a 357

precise assignment 358

15

Does G-C181-OH assume a bilayer or interdigitated structure the latter being the one 359

found for other similar bolaform glucolipids4243 The size of G-C181-OH can be estimated to 360

be at the most 32 Aring given by the length L of the C16 tail (L = 22 Aring from the Tanford formula 361

L= 154 + 1265 times n (n= 16)59) and the sugar (about 10 Aring)4243 However the cis conformation 362

of the C=C bond imposes a ldquokinkrdquo of 60deg in the aliphatic chain which reduces the effective 363

length of the oleic backbone to about 17 Aring for an effective total G-C181-OH length of about 364

27 Aring Cryo-TEM provides a tube thickness varying from 35 Aring to 60 Aring and a vesicle membrane 365

thickness of at least 60 Aring WAXS suggests an interlamellar thickness (membrane and water) 366

between 40 Aring and 37 Aring and an inter-lipid distance in the tube of about 16 Aring In the absence of 367

a clear form factor oscillation profile as found for similar glucolipids4243 one must 368

acknowledge that these data pledge neither in favor of fully interdigitated nor a double layer 369

structure and might suggest a tilted lipid arrangement with partial interdigitation60 370

371

13C ssNMR demonstrates that G-C181-OH can assume two coexisting head-head and 372

head-tail configurations 373

Vesicles are characterized by low positive mean and gaussian curvatures while 374

nanotubes are described by a combination of a high positive mean and zero gaussian curvatures 375

In this regard the coexistence of vesicles and nanotubes requires glucolipids to adapt 376

themselves to both environments at the same time For this reason the formation of nanotube-377

vesicle networks generally requires a specific driving force able to impose a high mean and 378

zero gaussian curvature to the membrane In many systems insertion of at least a second 379

constituent (lipid surfactant protein) in the membrane becomes a necessary internal driving 380

force for the nucleation and growth of so-called membrane inclusions from which 381

topologically-connected nanotubes can form10ndash123261 as explained by orientational ordering 382

arguments62 It has also been shown that external driving forces such as use of laser 383

ldquotweezersrdquo30 osmotic pressure6 electroformation5ndash82829 or controlled suction13ndash16 are 384

otherwise necessary as also supported by theoretical studies19 385

The bending energy of tubes is higher than that of a vesicle membrane8 due to the 386

dependence of the membrane free energy from the principal curvatures For this reason 387

isotropic (rotational symmetry along the main axis) but also anisotropic33 lipids generally 388

stabilize low-curvature shapes like flat membranes or vesicles Lipid nanotubes are also a 389

common phase formed by amphiphiles but generally for lipids with a certain complexity in 390

their molecular structures as exampled by diacetylenic derivatives of phospholipids among 391

others262763ndash65 Negative contributions to the bending energy favoring the simultaneous 392

16

presence of interconnected vesicles and tubes was shown to be driven by the presence of 393

anisotropic lipids in a membrane continuum Spontaneous deviations to the elastic theory in the 394

presence of surfactants lipids and proteins with anisotropic structures favor the formation of 395

protrusions with radius of the order of the fraction of a micron9ndash1219ndash2362 396

In the absence of both internal and external driving forces mentioned above G-C181-397

OH should behave as a standard lipid Its DSC profile (Figure 2)536667 under both hydrated and 398

dry conditions identifies a hydration-independent transition (Tm1) slightly above room 399

temperature this is followed by one main transition at higher temperature (Tm2) in the dry 400

sample50 For such a profile one expects a gel phase (Lβ) below Tm1 and a liquid crystalline 401

structure possibly in the morphology of vesicles (L4 phase) possibly above Tm1 and most likely 402

above Tm231 as experimentally found On the basis of the thermogram recorded on the dry 403

powder one may also expect a ldquoripplerdquo (Pβ) phase between Tm1 and Tm2 characterized by 404

periodic undulations of the bilayer (generally addressed to as regions with both liquid order and 405

disorder)5455 406

In the present work instead of a Lβ phase we observe flat crystals below Tm1 and a 407

vesicle-nanotube network instead of the ldquoripplerdquo phase The latter is particularly unexpected 408

because the membrane is composed of a single-component (G-C181-OH) and no anisotropic 409

external force (osmotic hydrodynamic electric) is used to pull and stabilize the tubes In 410

addition the nanotubes are observed at fairly high temperatures above Tm1 (between about 411

60degC and 90degC) while this morphology is generally favored at low temperature (proportionality 412

with 1kT figure 5 in Ref 62) commonly below a given transition temperature characteristics 413

for each specific amphiphile27 Furthermore nanotubes of G-C181-OH seem to be stable in 414

time thus differing from what is found in electroformed vesicles78 They are also observed 415

away from phase transition events (may them be at Tm1 or Tm2) as otherwise reported before 416

for a diacetylenic derivative of phosphatidylcholine26 and many other systems27 417

A stable nanotube-vesicle network for a single-component phospholipid in the absence 418

of external driving forces is not expected and has never been reported before It could be 419

explained by the coexistence of two micro-separated phases of different rigidity This 420

hypothesis is excluded by confocal microscopy on Laurdan-labelled membranes and magic 421

angle spinning (MAS) solid-state NMR (ssNMR) experiments 422

Laurdan is a well-known fluorophore used to put in evidence the coexistence of rigid 423

and fluid domains in both model and biological membranes47ndash49 Laurdan was used here to label 424

the G-C181-OH membranes between Tm1 and Tm2 and to verify the possible micro-segregation 425

and coexistence of rigid (eg Lβ λem= 440 nm) and fluid (eg Lα λem= 490 nm) domains in 426

17

tubes and vesicles respectively A set of quantitative experiments performed on a 5 mgmL-1 427

solution summarized in Figure 5 shows an excess of fluorescence at only λem= 490 nm for 428

temperature between 90degC and 70degC (Figure 5h) The GP value which quantifies the rigidfluid 429

character of the membrane is systematically negative (Figure 5i) indicating an overall fluid 430

membrane at all temperatures GP is quantified on more than 50 regions of interest (ROIs) at 431

each temperature including vesicles and tubes alone as well as connected nanotube-vesicles 432

(Figure 5d-g) The dispersion of GP at 80degC and 70degC between -08 and -02 could suggest the 433

spurious presence of membranes with various degrees of flexibility although fluidity is always 434

predominant The combination of WAXS (broad peak d= 144 Aring-1 invariant with 435

temperature Figure S 6bc) and confocal fluorescence microscopy using Laurdan confirms the 436

single-phase of the membrane composed of G-C181-OH both in vesicles and nanotubes This 437

is also consistent with the lack of phase transitions in DSC experiments between ~85degC and 438

~60degC especially in hydrated samples We conclude that the vesicle-nanotube transition is not 439

driven by the local segregation of gel phase (Lβ) microdomains nor that the nanotubes are 440

composed of such phase 441

ssNMR is the best-suited approach to investigate the origin of the local driving force for 442

the nucleation of anisotropic inclusions of which the average orientation depends on the local 443

membrane and inclusion curvatures and was shown to promote tubing62 ssNMR can provide 444

information on the local molecular mobility eg through the nuclear spin relaxation behavior 445

but also on intermolecular interactions and stereochemistry eg through variations in the 446

chemical shift68ndash70 447

Ideally ssNMR should be performed on diluted G-C181-OH samples above Tm1 448

However fast MAS necessary for high resolution centrifuges the sample As a first 449

approximation hydrated G-C181-OH samples for instance at 50 wt as classically done for 450

phospholipids71 could be studied Unfortunately permeability of the sample holder combined 451

to heating (close to the boiling point of water) and MAS provoke dehydration and even sample 452

ejection After experiencing both of the above this approach was abandoned As a second 453

approximation the study of the dehydrated G-C181-OH powder avoids centrifugation and 454

evaporation If studying dried lipids may not reflect their exact behaviour under diluted 455

conditions valuable structural information can still be obtained in dehydrated molecular 456

systems like proteins thus illustrating that this approach is not unreasonable Furthermore if 457

DSC (Figure 2a) shows that hydration suppresses the Tm2 it also shows that the value of Tm1 is 458

not affected by water at concentrations as low as 10 wt For this reason performing a 459

structural analysis of dehydrated G-C181-OH by ssNMR above Tm1 may help understanding 460

18

the local arrangement of G-C181-OH in two morphologies vesicles and nanotubes having 461

sensibly different curvatures 462

To explore the local mobility of the most relevant chemical groups of G-C181-OH 463

above Tm2 (T= 110degC) and between Tm2 and Tm1 (70 lt T degC lt 90) we combine a series of 1H 464

and 13C experiments Single-pulse (SP) MAS experiments (1H and 13C) when recorded under 465

quantitative conditions allow the detection of all chemical species while cross-polarization 466

(CP) MAS (13C only) experiments are sensitive only to constrained groups for which the 1H-467

13C dipolar coupling is not averaged to zero by the molecular motions 468

469

Figure 6 ndash a) 1H MAS (10 kHz) ssNMR spectra recorded from 110degC to 70degC b) 13C MAS (10 kHz) ssNMR 470

spectra recorded using high power decoupling single pulse (SP) and cross-polarization (CP) pulse sequences 471

Contact time for CP experiments and corresponding temperatures are given on each spectrum Full spectra 472

are given in Figure S 7 473

The most relevant 1H and 13C spectra are given in Figure 6 (full range in Figure 7 and 474

Figure S 7) As expected the 1H solid-state SPMAS spectra show broader peaks compared to 475

the solution-state NMR spectrum (Figure 6a) reflecting the obvious solid-state character of the 476

sample at any temperature The sugar group and the C=C bond resonate in the 25-6 ppm range 477

and 535 ppm respectively while the OH groups of the sugar moiety are found below 45 ppm 478

(Figure 6a full assignment in Table S 1) Very broad (~1 ppm) at 100degC the position of the 479

OH peak varies with the temperature value and becomes narrower (lt~05 ppm) and visible at 480

δH= 464 ppm at T= 70degC Line narrowing and chemical shift variations of the OH resonance 481

in ssNMR is always a direct sign of H-bonding interactions whereas the strength of the H-482

bonding hence its OHhellipH distance is correlated to a dispersion in chemical shifts72ndash75 All other 483

60 55 50 45 40 35 30 25

ppm

110

90

80

70

910

1rsquo

6rsquo16

1

OH

134 132 130 128 126

ppm110 108 106 104 102 100

ppm

SP (T= 110degC)

SP (T= 70degC)

CP 50 ms (T= 70degC)

CP 3 ms (T= 70degC)

910 1rsquo

1H 13C

1

811

910

CH2 CH2

1rsquo

6rsquo

a) b)

C102

13024

C9101

13003

C92

12869 C1rsquo210382

C1rsquo110443

C1rsquo13035

2rsquo3rsquo4rsquo

T degC

5rsquo

19

resonances broaden from 110degC to 70degC thus indicating slower andor more anisotropic 484

reorientational motions of all chemical groups The specific dynamic behaviour of the sugar 485

headgroup is also reflected by its 1H transverse relaxation time T2 whereas the shorter the T2 486

the higher the rigidity Experiments performed between 110degC and 70degC shown and 487

commented in Figure S 8 in the Supporting Information globally show a single short (lt 1 ms 488

below 100degC) T2 relaxation component for H1rsquo H5rsquo and H6rsquo and two both short and long 489

(between 4 ms and 12 ms) T2 for the acyl chain (CH2 H910 H811) even below 100degC 490

At 110degC 13C SPMAS experiments (Figure 6b full range in Figure 7 and Figure S 7 491

assignments in Table 1) show a single peak at δC= 10335 ppm for C1rsquo and at δC= 13003 ppm 492

for C910 as expected for a symmetrical conformation with two chemically-equivalent 493

carbons76 At 70degC the C1rsquo peak disappears while C910 now shows two peaks around 128 494

ppm and 130 ppm (Table 1 full temperature series in Figure S 7b) attributed to C92 (δC= 12869 495

ppm) and C10 in a non-symmetric configuration respectively for which these two carbons are 496

chemically inequivalent76 More precisely the resonance at 130 ppm is actually the 497

superposition of two close resonances C9101 (δC= 13003 ppm) and C102 (δC= 13024 ppm) 498

where subscripts 1 and 2 stand for two different conformations of the cis-olefin group as 499

reported for various phases of oleic acid in the vicinity of 0degC76 500

501

Table 1 ndash 13C Chemical shifts of G-C181-OH in solution and solid-state Solid-state experiments are 502

recorded at MAS= 10 kHz All spectra are reported in Figure 7 and Figure S 7 The contact time for cross 503

polarization is 3 ms In all ssNMR experiments temperature is first raised at 110degC and then lowered to 504

either 90degC or 70degC Attribution of the 13C resonances in Table 1 is performed on the basis of the solution 505

spectrum (Figure S 1b) Additional peaks observed in both single pulse and cross polarization ssNMR 506

experiments are attributed when possible to a configuration effect of the given atom eg C9101 and C92 507

indicate two configurations of C9 All new peaks observed in solid-state and not detected in solution are 508

highlighted in red for convenience 509

510

511

δ ppm

Peak Liquid (DMSO-d6)

Solid Single Pulse

T= 110degC

Solid Single Pulse

T= 90degC

Solid Single Pulse

T= 70degC

Solid Cross Polarization

T= 70degC

C102

- - 13024 13024

C910C9101 13009 13003 13003 13003 13003

C1C9C10C1rsquo

C6rsquo

20

C92 - - - 12869 12869

C1rsquoC11 10329 10335 10335 - 10443

C12 - - - - 10382

C3 7724 7709 7702 - 7829

- 7634 7629 - 7621

7525

C2 7388 7398 7390 7370 7370

C4 7052 7077 7054 7041 7098

C16 6900 7016 - - 7029

- - - - 6793

C1C5C6 6153 6241 6236 6239 6262

C1C5C6 6117 - - - 5956

C2 3300 3290 3290 3283 3279

CH2(trans) - - - 3198 3196

CH2(gauche)C15 2941 2981 2987 3004 -

C811 2706 2747 2747 2832 2817

CH2 2597 2625 2628 2630 2817

512

13C CPMAS spectra at 70degC (Figure 6b full range in Figure 7 and Figure S 7d) at 513

increasing contact times confirm the presence of two cis-olefin environments one for which 514

C910 are equivalent labelled here C9101 and one for which they are inequivalent labelled 515

C92 C102 Both environments are rigid enough to be observed at short contact time (tc= 50 μs) 516

A similar double environment is also observed for the sugar C1rsquo CPMAS at 70degC shows two 517

peaks (C1rsquo1 at δC= 10443 ppm C1rsquo2 at δC= 10382 ppm Figure 6b Figure S 7d) These are 518

significantly shifted downfield from the C1rsquo peak observed at 110degC and are detected at contact 519

times as short as 50 μs thus indicating their rigid character One also finds that CPMAS is 520

totally inefficient at 100degC for short contact times (50 μs) and poorly efficient even at 3 ms 521

typical of fast-tumbling molecules (Figure S 7c) also both gauche (δC= 3004 ppm) and all-522

trans (δC= 3198 ppm) conformations of the CH2 carbons coexist at 70degC (Figure S 7b)77 523

According to ssNMR one main mobile molecular environment exists at 110degC as 524

shown by the well-resolved 1H spectrum single C=C (C9101) and anomeric (C1rsquo) resonances 525

gauche conformation of the aliphatic CH2 and single T2 of the aliphatic CH2 chain and sugar 526

protons The sugar group is certainly mobile (inefficient 1H13C CPMAS) but still more 527

constrainted than the acyl chain (shorter T2 (1H)) As the temperature is lowered the overall 528

rigidity increases (broadening of 1H NMR peaks faster 1H transverse relaxation and efficient 529

1H13C magnetization transfer during CPMAS) presumably driven by H-bonding 530

interactions (downfield shift and line narrowing of the OH 1H signal very short OH T2 loss of 531

the C1rsquo signal in SPMAS) At the same time a number of spectroscopic data (two concomitant 532

21

resonances for anomeric C1rsquo1 and C1rsquo2 cis-olefin C9101 and C92 C102 and CH2 gauche and 533

all-trans environments) and relaxation behaviour (fast and slow T2 relaxation components) 534

strongly suggests the coexistence of two molecular environments Further proofs of the 535

coexistence of two distinct molecular domains are given by a number of chemical shift 536

variations in the 13C CPMAS spectrum at 70degC compared to the one at 110degC 90degC and 537

solution This is shown in Figure 7 and in the list of 13C chemical shifts in Table 1 538

539

540

Figure 7 - Comparison between solution- and solid-state 13C NMR spectra of G-C181-OH Solid-state 541

experiments are recorded directly on the sample powder at a MAS of 10 kHz (4 mm zirconia rotor) SP 542

Single Pulse CP Cross Polarization In all the ssNMR experiments temperature is first raised at 110degC 543

and then lowered to either 90degC or 70degC Segmented lines are meant to highlight the known peak position 544

in solution thus stressing the strong chemical shift variations in the solid-state 545

546

NMR experiments above Tm2 suggest that vesicles could contain G-C181-OH with a 547

single family of mobile conformations characterized by chemically-equivalent C=C (red 548

Figure 8b) and a single C1rsquo conformation (green Figure 8b) Experiments between Tm1 and Tm2 549

otherwise suggest that two coexisting conformations characterized by chemically-equivalent 550

(red Figure 8b) and inequivalent (magenta Figure 8b) C=C could be associated to nanotubes 551

Two distinct C1rsquo sites (grey Figure 8b) also indicate two configurations of the sugar headgroup 552

A comment on this point will be given in the following paragraphs 553

The local (Aring-scale) variation of molecular mobility below Tm2 drives the coexistence of 554

two distinct molecular environments of the same G-C181-OH molecule This occurs within a 555

single-phase domain (as deduced by WAXS) of equivalent membrane rigidity (as derived from 556

Laurdan probe in confocal microscopy) Although never reported before the difference in the 557

local mobility of the same amphiphile (here G-C181-OH) could satisfy the theoretically-558

140 120 100 80 60 40 20

CP(3 ms)-70degC

SP-70degC

SP-90degC

SP-110degC

ppm

LiquidC910

C1rsquo

C2rsquo

C4rsquo

C3rsquo

C1

C6rsquoC5rsquo

C1

6

C8

11

C2 C

H2

C1

5

CH

2

DM

SO

13C

22

required conditions that induce local lipid segregation and formation of membrane inclusions 559

(Figure 8c) In case of their non-zero average orientation the local membrane curvature 560

increases62 and nanotubes can spontaneously form (Figure 8bc) More intriguing this 561

mechanism occurs at least 10degC above Tm1 although it should not be favored above the Tm27 562

From both experimental and theoretical backgrounds the conditions of topologically-563

connected vesicles-nanotubes systems are met only when either internal (two components) or 564

external driving forces are applied19 In the present single-lipid system of which the phase 565

behavior is summarized in Figure 8a the most plausible driving force could be explained by 566

the NMR results collected on the dry sample G-C181-OH concomitantly occupies two 567

molecular environments characterized by 1) two distinct conformations of the sugar 568

displaying equivalent mobility and 2) two conformations of the acyl chain unsaturation 569

570

23

571

Figure 8 ndash a) Thermotropic phase behavior of G-C181-OH at 5 mgmL-1 Note that Tm2 is observed on the 572

dry powder only b) Possible relationship between the molecular environment as derived from NMR and 573

membrane curvatures All the molecular drawings refer to the same G-C181-OH molecule (bulky ellipsoid 574

equiv sugar headgroup) while each color corresponds to a given configuration Grey of C1rsquo1 and C1rsquo2 indicates 575

the uncertainty about the assignment of the specific NMR peak The curvature C is defined as 1R R being 576

the curvature radius R1 R2 radius of curvature CG Gauss curvature CM Mean curvature c) Free 577

interpretation of the anisotropic high-curvature (CMgt 0 CG= 0) inclusions in the membrane (low CM and 578

CG) continuum driven by inter-sugar H-bonding and reduced intramembrane diffusion for a G-C181-OH 579

glucolipid 580

581

Flat crystalline structure

Nanotube-vesicle network

Vesicles

48degC

120deg

106 105 104 103 102 101

ppm

H9 H10H1rsquo

106 105 104 103 102 101

ppm106 105 104 103 102 101

ppm106 105 104 103 102 101

ppm

C=CC=C

C1rsquo1C1rsquo2

C1rsquo

Tm2(96degC)Tm1

a)

b)

c)

Same molecule different local orientations and sugar conformations

High curvature membrane

Low curvature inclusion

Flip-FlopStrong H-bonding

CG

CM ( + )

CG 0

CM 0 CG 0

CM 0

G-C181-OH

24

Two related questions are unanswered at the moment What actually drives nanotube 582

formation and is the organization of G-C181-OH within the vesicles and tubes the same Lipid 583

nanotubes have been studied for at least three decades over which broad experimental and 584

theoretical insight has been generated Despite this fact the actual understanding is still matter 585

of debate27 Micelle-to-nanotube and vesicle-to-nanotube phase transitions are generally 586

reported for single amphiphiles below a given Tm and intermediate phase composed of twisted 587

and flat fibers is concomitantly reported Initially thought to be a chirality-driven phenomenon 588

recent studies propose a more general mechanism involving crystallization and symmetry 589

breaking in the lipid packing driven by optimal packing27 If twisted ribbons were observed 590

before for a number of microbial glycolipid bolaamphiphiles44 they are not found for G-C181-591

OH This fact as well as the large body of cryo-TEM images and the observation of nanotubes 592

above Tm1 rather pledge in favour of a direct topological change from positive (vesicles) to 593

zero (nanotubes) Gaussian curvatures In this case vesicle-to-nanotube transitions could be 594

driven by external forces56307813ndash162829 595

Considering the absence of external forces in the present system there are only two 596

possibilities left spontaneous budding of the membrane or nanotube pulling after inter-vesicle 597

collisions due to Brownian motion Both are possible as inclusions composed of a different 598

arrangement of the same molecule in the membrane continuum can theoretically explain 599

spontaneous nanotube formation in an isotropic medium672223629ndash1218ndash21 However the second 600

one is probably more likely to prevail because it could explain the large number of nanotubes 601

connecting two distant vesicles Interestingly similar events occur for any phospholipid-based 602

vesicular solution but nanotubes are extremely rare unless at least two lipids of different 603

molecular structure are mixed together The spectroscopic 1H and 13C and relaxation (1H) 604

ssNMR data even if collected on a dry powder demonstrate that the coexistence of two G-605

C181-OH configurations involving both the sugar headgroup and fatty alcohol chain are 606

possible for this molecule As a matter of fact this is not so surprising as more than one energy 607

minima each corresponding to a given conformation are known for glucose its non-reducing 608

derivative methylglucose and carbohydrates in general78ndash80 609

The presence of two configurations could explain the coexistence between two 610

structures of different curvature but the data presented so far do not explain the origin of these 611

two configurations In other words what drives the change in the local structure of the sugar 612

headgroup and fatty alcohol within the membrane 613

The association of known self-assembled structures to spectroscopic 13C ssNMR data is 614

known for years in particular for gluconamides lipids having a linear sugar headgroup81ndash84 It 615

25

was found that gauche effects in some cases up to the γ position in the sugar chain induce 616

upfield chemical shifts while anti and trans conformations result in downfield shifts Each 617

structure (monolayer crystal bilayer crystal or micellar fiber) was then characterized by a given 618

organization of the gluconamide and a characteristic 13C ssNMR fingerprint which could be 619

explained by gauche γ-gauche anti and trans conformations of the headgroup Similar 620

structural-spectroscopic correlations have been collected over the years for much simpler cyclic 621

sugars85 whereas pyranoses can undergo important conformational changes (chair half-chair 622

boat skew) when the appropriate energy barriers are crossed78ndash8086 A quantitative 623

conformational study based on the 13C chemical shift variations shown in Table 1 would be 624

certainly helpful but very delicate in the context of the present state of the art because a 625

rigorous approach requires at least one known crystal structure8182 which is currently missing 626

for this compound A qualitative approach will then be adopted 627

G-C181-OH is a bolaform amphiphile and flip-flop could be a possible mechanism 628

involved in the nanotube stabilization whereas head-tail interactions have long been considered 629

as important in nanotube-forming bolaform amphiphiles6063 From mere considerations based 630

on T2 (1H) relaxation arguments one could exclude head-tail configurations if the R-CH2OH 631

endgroup (H1) was involved in a direct interaction with the sugar group its corresponding T2 632

would probably be comparable to the sugarrsquos one (OH and H1rsquo) but this does not seem to be 633

the case However this hypothesis would be based on the comparison of the relaxation from 634

CH and CH2 protons additionally involved in different kind of motions Besides the relaxation 635

of H1 is influenced by the strong overlap between H1 and H5rsquo For these reasons relaxation 636

arguments are not solid enough and will not be used 637

We then address this question by exploring intermolecular interactions in a direct 638

manner using 2D 1H-1H homonuclear ssNMR correlation spectroscopy At 90degC in the fast-639

tumbling domain of molecular motion (narrow 1H NMR peaks Figure 6a) and corresponding 640

to the vesicle-rich region spin diffusion (NOESY) experiments are preferred over dipolar 641

coupling based experiments because such interaction is averaged out by molecular motion 642

However when the molecular motion is more constrained in the domain of nanotube stability 643

at 70degC (broad 1H spectra Figure 6a) Double-Quantum Single-Quantum (DQ-SQ) correlation 644

experiments are best suited to probe interaction between close molecular neighbors These 645

experiments are reported in Figure 9 and the related interpretation is only briefly reported 646

below while a more detailed discussion can be found on Page 12 in the Supporting Information 647

26

648

Figure 9 ndash a) 2D 1H-1H NOESY NMR experiment performed at MAS= 10 kHz (4 mm probe) on a Bruker 649

300 MHz Avance III HD spectrometer using 300 ms of mixing time Sample temperature is regulated to 650

90degC reached by cooling from 110degC b) 2D 1H-1H DQ-SQ (BAck-to-BAck BABA pulse sequence used for 651

the DQ excitation scheme) NMR experiment performed at MAS= 22 kHz (25 mm probe) on a Bruker 700 652

MHz Avance III spectrometer using a DQ excitation time equal to 2 rotor periods Sample temperature is 653

regulated to 70degC reached by cooling from 110degC The typical schemes illustrating how to interpret these 654

2D experiments are reported above each experimental map Zooms of the sugar region and configurational 655

interpretation are given below each 2D map More information are given in the text below 656

H16-H16

H1-H1

H9-H10

H1rsquo-H1

H1rsquoH910 H16

H1

H1rsquo-H1rsquo

Hxrsquo-Hxrsquo

2

(pp

m)

(ppm)

315

811910 1rsquo

2rsquo2

3rsquo 4rsquo6rsquo

5rsquo

1rsquo2910

asymp asymp asympHDO MeOD Acetone

1

315

811 9101rsquo

2rsquo2

3rsquo4rsquo6rsquo

5rsquo

1rsquo2 9 10

asympasympasympHDOMeODAcetone

H1rsquo-H1

1rsquo

1

(p

pm

)

(ppm)

H1-H2

H1rsquo-H1

H1rsquo

H910

H1H5rsquo

H2

H1rsquo-H5rsquo

315

811910 1rsquo

2rsquo2

3rsquo 4rsquo6rsquo

5rsquo

1rsquo2910

asymp asymp asympHDO MeOD Acetone

1

H2rsquoH3rsquoH4rsquo

H2rsquoH3rsquoH4rsquo

X= 2rsquo3rsquo4rsquo

NOESY - T= 90degC

1H

1H

1H

1H

DQ-SQ - T= 70degC

a) b)

2

(pp

m)

(p

pm

)

(ppm)

A B

AhellipB

AhellipA

BhellipB

NOESY

(ppm)

A B

AhellipB

DQ-SQ

Single-quantum dimension

Do

ub

le-q

ua

ntu

m d

imen

sio

n

AhellipA

BhellipB

BhellipA

BhellipA

315

811910 1rsquo

2rsquo2

3rsquo 4rsquo6rsquo

5rsquo

1rsquo2910

asymp asymp asympHDO MeOD Acetone

1

lt 5 Aring

lt 5 Aring

H1-H2

H811

H15

H16

H6rsquo

H1rsquo

H1

H5rsquo

H2

H1-H2

H1rsquo-H1

H1rsquo-H5rsquo

H1rsquo

H1 H5rsquo

H1rsquo

H1

H5rsquo

H1H2rsquo H3rsquo H4rsquo

H1-H1

H1rsquo-H1

Hxrsquo-HxrsquoX= 2rsquo3rsquo4rsquo

H1-H2

H1rsquo

Zoom sugar region 28 ndash 45 ppmNOESY DQ-SQ

27

Head-tail or head-head configurations could be probed by the existence of H1rsquo-H1 657

interactions whereas H1 (δH= 356 ppm) is associated to the glucolipid tail and H1rsquo (δH= 432 658

ppm) to the sugar headgroup These atoms are more than 15 Aring apart and a direct interaction 659

could not be observed in both NOESY and DQ-SQ experiments NOESY experiments (Figure 660

9a) show that H1rsquo-H1 correlations are not observed Meanwhile clear-cut direct H1rsquo-H1 661

interactions are indeed observed are detected at T= 70degC (DQ cross-peaks related to H1rsquo-H1 662

correlation in DQ-SQ experiment shown in Figure 9b) Interestingly the absence of auto peaks 663

at δH= 432 ppm and δH= 335 ppm in the SQ dimension excludes H1rsquo-H1rsquo and Hxrsquo-Hxrsquo (x= 2 664

3 4) interactions respectively strengthening the idea that sugar headgroups from two nearest-665

neighbor G-C181-OH molecules are not adjacent in the nanotube-vesicle region 666

Homonuclear 2D ssNMR supports the idea according to which G-C181-OH can 667

undergo molecular flip-flops and a head-tail arrangement in the dry product Extrapolating these 668

conclusions in the lower concentration range lack of H1rsquo-H1 proximity at T= 90degC suggests a 669

head-head configuration in the vesicle-rich region and head-tail configuration within the 670

nanotubes It is however still unclear whether flip-flops are driven by spontaneous temperature-671

induced conformational changes in the glucolipid packing stimulating inclusions within the 672

membrane itself672223629ndash1218ndash21 or by defects generated by inter-vesicle collisions In fact a 673

combination of both is not unlikely Nanotubes are not stable when temperature is 20 to 30 674

degrees above Tm1 indicating that inter-vesicle collisions alone cannot explain the unusual 675

phenomenon In an energy landscape approach87 the head-head configuration is certainly 676

favored at high temperature while both head-head and head-tail configurations could be 677

favored in the mid-temperature region corresponding to the range between Tm1 and Tm2 in the 678

dry powder Both statistical and defect-induced flip-flop events contribute to form high mean 679

zero Gaussian curvature morphologies (nanotubes) for which the head-tail conformation 680

limiting the proximity between bulky sugar headgroups is probably more adapted (Figure 8c) 681

The flip-flop interpretation now reasonably explains the strong differences in the sugar and fatty 682

alcohol conformations deduced by the multiple NMR signals attributed to chemically-683

equivalent groups The existence of various low-energy conformations each having specific 684

NMR signature for sugars in solution78ndash8086 and solid-state81ndash84 further supports our 685

interpretation 686

One last point should be clarified Why between Tm1 and Tm2 are two molecular 687

environments able to coexist instead of a single one of average rigidity as found in the case of 688

ldquoripplerdquo phases50ndash54 We believe that this question can be answered based on the NMR results 689

the relatively short T2 values of the H1rsquo and OH protons shows the constrained mobility of the 690

28

sugar headgroup despite a certain flexibility of the chain (long T2 component) This trend may 691

suggest restrictions in translational mobility probably due to the sugar-OH hydrogen bonding 692

(Figure 8c) meaning that lipids are not able to diffuse within the membrane thus keeping two 693

different configurations stable 694

695

Conclusions 696

Topologically-connected nanotube-vesicle lipidic systems are commonly observed for 697

self-assembled membranes composed of at least two amphiphiles and spontaneously driven by 698

anisotropic inclusions andor by external directional forces This work shows that strong 699

differences in membrane curvature can spontaneously occur in single-lipid systems without 700

external anisotropic forces Coexistence and interconnection between nanotubes and vesicles 701

are evidenced via cryo-TEM and fluorescence confocal microscopy in a single-phase region at 702

temperature above a transition temperature Tm1= 483degC determined by DSC on the hydrated 703

sample but below the second transition temperature Tm2 (957degC) determined by DSC on the 704

dry powder X-ray diffraction (WAXS) also shows the coexistence of two lattice periods one 705

attributed to an inter-membrane repeating distance and the other possibly related to the intra-706

nanotube glucolipid arrangement Labelling the systems with Laurdan a fluorophore of which 707

the emission strongly varies with the local membrane rigidity excludes the presence of rigid 708

micro-separated phases (eg Lβ) 709

1H and 13C ssNMR despite the fact that it can only be performed on the dry powder 710

confirm the presence of one main fluid G-C181-OH environment above Tm2 Below Tm2 the 711

splitting of some 13C peaks and the coexistence of short and long 1H relaxation components 712

demonstrate that G-C181-OH can simultaneously pack into structurally and dynamically 713

inequivalent environments Combining advanced 2D 1H-1H correlation experiments based on 714

spin diffusion or double-quanta excitation suggests that vesicles could contain a preferred head-715

head configuration of the bolaform lipid while nanotubes a preferred head-tail assembly 716

The presence of two environments composed of the same molecule characterized by 717

different molecular configurations and rigidity fulfills the theoretical conditions necessary to 718

spontaneously nucleate a membrane inclusion of non-zero average orientation described as 719

necessary to observe topologically-connected nanotube-vesicle structures in the absence of 720

external forces This hypothesis could also be enriched by the possibility of nanotubes to be 721

pulled away between two vesicles after a collision event 722

This work expands the actual theoretical and experimental framework on lipid networks 723

stabilized by a different radius of curvature It could also motivate the synthesis or isolation 724

29

of new lipids providing a higher level of 3D architectural control in soft networks possibly in 725

the range of room temperature 726

727

Acknowledgements 728

Dr Andrea Lassenberger and Dr Sylvain Preacutevost at Institut Laue Langevin (ILL Grenoble 729

France) are kindly acknowledged for their assistance on the SANS experiment Confocal 730

images were performed at the Institut de Biologie Paris Seine (IBPS) imaging facility Jean-731

Franccedilois Gilles (Imaging Core Facility Institut de Biologie Paris Seine (IBPS) CNRS 732

Sorbonne Universiteacute Paris France) is kindly acknowledged for his assistance on the confocal 733

microscopy imaging Dr Guillaume Laurent (LCMCP Sorbonne Universiteacute Paris France) is 734

kindly acknowledged for helpful discussions The French region Ile-de-France SESAME 735

program is acknowledged for financial support (700 MHz NMR 736

spectrometer) Dr Olivier Diat (ICSM Marcoule France) Prof eronikaIglič and Prof V Aleš737

onsul discussi Slovenia) are kindly acknowledged for helpfUniversity of LjubljanaIglič (-Kralj 738

739

Financial support 740

WAXS experiments have been supported by Soleil Light Source Saint Aubin France proposal 741

Ndeg 20201747 SANS experiments have been supported by ILL proposal Ndeg9-13-778 742

743

References 744

1 D Kabaso N Bobrovska W Goacuteźdź E Gongadze V Kralj-Iglič R Zorec and A 745

Iglič Bioelectrochemistry 2012 87 204ndash210 746

2 C R Safinya U Raviv D J Needleman A Zidovska M C Choi M A Ojeda-747

Lopez K K Ewert Y Li H P Miller J Quispe B Carragher C S Potter M W 748

Kim S C Feinstein and L Wilson Adv Mater 2011 23 2260ndash2270 749

3 P Veranič M Lokar G J Schuumltz J Weghuber S Wieser H Haumlgerstrand V Kralj-750

Iglič and A Iglič Biophys J 2008 95 4416ndash4425 751

4 A Rustom R Saffrich I Markovic P Walther and H-H Gerdes Science (80- ) 752

2004 303 1007ndash1010 753

5 M I Angelova S Soleacuteau P Meacuteleacuteard F Faucon and P Bothorel Prog Colloid 754

Polym Sci 1992 89 127ndash131 755

6 L Mathivet S Cribier and P F Devaux Biophys J 1996 70 1112ndash1121 756

7 V Kralj-Iglič G Gomišček J Majhenc V Arrigler and S Svetina Colloids Surfaces 757

A Physicochem Eng Asp 2001 181 315ndash318 758

30

8 V Kralj-Iglič A Iglič G Gomišček F Sevšek V Arrigler and H Haumlgerstrand J 759

Phys A Math Gen 2002 35 1533ndash1549 760

9 M Drab D Stopar V Kralj-Iglič and A Iglič Cells 2019 8 626 761

10 A Zidovska K K Ewert J Quispe B Carragher C S Potter and C R Safinya 762

Langmuir 2009 25 2979ndash2985 763

11 A Zidovska K K Ewert J Quispe B Carragher C S Potter and C R Safinya Soft 764

Matter 2011 7 8363ndash8369 765

12 A Zidovska K K Ewert J Quispe B Carragher C S Potter and C R Safinya 766

Biochim Biophys Acta - Biomembr 2009 1788 1869ndash1876 767

13 M Karlsson K Sott M Davidson A S Cans P Linderholm D Chiu and O Orwar 768

Proc Natl Acad Sci U S A 2002 99 11573ndash11578 769

14 A Karlsson R Karlsson M Karlsson A-S Cans A Stromberg F Ryttsen and O 770

Orwar Nature 2001 409 150ndash152 771

15 M Karlsson K Sott A S Cans A Karlsson R Karlsson and O Orwar Langmuir 772

2001 17 6754ndash6758 773

16 E Evans H Bowman A Leung D Needham and D Tirrell Science (80- ) 1996 774

273 933ndash935 775

17 U Seifert Configurations of fluid membranes and vesicles 1997 vol 46 776

18 J Kaumls and E Sackmann Biophys J 1991 60 825ndash844 777

19 N Bobrovska W Goacuteźdź V Kralj-Iglič and A Iglič PLoS One 2013 8 e73941 778

20 A Iglič B Babnik K Bohinc M Fošnarič H Haumlgerstrand and V Kralj-Iglič J 779

Biomech 2007 40 579ndash585 780

21 K Bohinc T Slivnik A Iglič M Brumen and V Kralj-Iglič J Phys Chem C 2007 781

111 9709ndash9718 782

22 A Iglič H Haumlgerstrand P Veranič A Plemenitaš and V Kralj-Iglič J Theor Biol 783

2006 240 368ndash373 784

23 V Kralj-Iglič and P Veranič Adv Planar Lipid Bilayers Liposomes 2006 5 129ndash785

149 786

24 A Iglic and V Kralj-Iglic Adv Planar Lipid Bilayers Liposomes 2006 4 253ndash279 787

25 S Penič L Mesarec M Fošnarič L Mroacutewczyńska H Haumlgerstrand V Kralj-Iglič and 788

A Iglič Front Phys 2020 8 342 789

26 P Yager and P E Schoen Mol Cryst Liq Cryst 1984 106 371ndash381 790

27 T G Barclay K Constantopoulos and J Matisons Chem Rev 2014 114 10217ndash791

10291 792

31

28 B Božič G Gomišček V Kralj-Iglič S Svetina and B Žekš Eur Biophys J 2002 793

31 487ndash496 794

29 A Iglič H Haumlgerstrand M Bobrowska-Haumlgerstrand V Arrigler and V Kralj-Iglič 795

Phys Lett A 2003 310 493ndash497 796

30 R Bar-Ziv and E Moses Phys Rev Lett 1994 73 1392ndash1395 797

31 S Chiruvolu H E Warriner E Naranjo S H Idziak J O Raumldler R J Plano J A 798

Zasadzinski and C R Safinya Science (80- ) 1994 266 1222ndash1225 799

32 V Kralj-Iglič A Iglič H Haumlgerstrand and P Peterlin Phys Rev E 2000 61 4230ndash800

4234 801

33 L Mesarec M Drab S Penič V Kralj-Iglič and A Iglič Int J Mol Sci 2021 22 802

2348 803

34 L Van Renterghem S L K W Roelants N Baccile K Uyttersprot M C Taelman 804

B Everaert S Mincke S Ledegen S Debrouwer K Scholtens C V Stevens W 805

Soetaert L Van Renterghem S L K W Roelants N Baccile K Uytersprot M C 806

Taelman B Everaert S Mincke S Ledegen S Debrouwer K Scholtens C V 807

Stevens and W Soetaert Biotechnol Bioeng 2018 115 1195ndash1206 808

35 J Clark and F Deswarte Introduction to Chemicals from Biomass Second Edition 809

John Wiley amp Sons 2015 810

36 K V Sajna R Houmlfer R K Sukumaran L D Gottumukkala and A Pandey in 811

Industrial Biorefineries and White Biotechnology eds A Pandey R Houmlfer M 812

Taherzadeh K M Nampoothiri and C Larroche Amsterdam Oxford Waltham 813

Elsevier 2015 pp 499ndash521 814

37 R Marchant and I M Banat Trends Biotechnol 2012 30 558ndash565 815

38 K Hill and C LeHen-Ferrenbach in Sugar-Based Surfactants Fundamentals and 816

Applications ed C C Ruiz Boca Raton CRC Press 2009 pp 1ndash20 817

39 L Rodrigues I M Banat J Teixeira and R Oliveira J Antimicrob Chemother 818

2006 57 609ndash618 819

40 J D Desai and I M Banat Microbiol Mol Biol Rev 1997 61 47ndash64 820

41 S L K W Roelants K Ciesielska S L De Maeseneire H Moens B Everaert S 821

Verweire Q Denon B Vanlerberghe I N A Van Bogaert P Van der Meeren B 822

Devreese and W Soetaert Biotechnol Bioeng 2016 113 550ndash559 823

42 N Baccile M Selmane P Le Griel S Preacutevost J Perez C V Stevens E Delbeke S 824

Zibek M Guenther W Soetaert I N A Van Bogaert and S Roelants Langmuir 825

2016 32 6343ndash6359 826

32

43 N Baccile A-S Cuvier S Preacutevost C V Stevens E Delbeke J Berton W Soetaert 827

I N A Van Bogaert and S Roelants Langmuir 2016 32 10881ndash10894 828

44 N Baccile C Seyrig A Poirier S A Castro S L K W Roelants and S Abel Green 829

Chem 2021 23 3842ndash3944 830

45 A Salis L Cappai C Carucci D F Parsons and M Monduzzi J Phys Chem Lett 831

2020 11 6805ndash6811 832

46 J Schindelin I Arganda-Carreras E Frise V Kaynig M Longair T Pietzsch S 833

Preibisch C Rueden S Saalfeld B Schmid J-Y Tinevez D J White V 834

Hartenstein K Eliceiri P Tomancak and A Cardona Nat Methods 2012 9 676ndash835

682 836

47 T Parasassi and E Gratton J Fluoresc 1995 5 59ndash69 837

48 M M Dodes Traian F L Gonzaacutelez Flecha and V Levi J Lipid Res 2012 53 609ndash838

616 839

49 D M Owen C Rentero A Magenau A Abu-Siniyeh and K Gaus Nat Protoc 840

2012 7 24ndash35 841

50 W Stillwell in An Introduction to Biological Membranes Academic Press 2016 pp 842

135ndash180 843

51 W Stillwell in An Introduction to Biological Membranes Academic Press 2016 pp 844

181ndash220 845

52 B Tenchov Chem Phys Lipids 1991 57 165ndash177 846

53 R Koynova and M Caffrey Biochim Biophys Acta 1998 1376 91ndash145 847

54 F Schmid S Dolezel O Lenz and S Meinhardt J Phys Conf Ser 2014 487 848

012004 849

55 S Meinhardt R L C Vink and F Schmid Proc Natl Acad Sci U S A 2013 110 850

4476ndash4481 851

56 O Glatter and O Kratky Small Angle X-ray Scattering Academic Press London 852

1982 853

57 B A Cunningham A D Brown D H Wolfe W P Williams and A Brain Phys 854

Rev E 1998 58 3662ndash3672 855

58 M Nadler A Steiner T Dvir O Szekely P Szekely A Ginsburg R Asor R Resh 856

C Tamburu M Peres and U Raviv Soft Matter 2011 7 1512ndash1523 857

59 C Tanford The hydrophobic effect Formation of micelles and biological membranes 858

John Wiley amp Sons Inc New York 1973 859

60 M Masuda and T Shimizu Langmuir 2004 20 5969ndash5977 860

33

61 A Zidovska K K Ewert J Quispe B Carragher C S Potter and C R Safinya 861

Block Liposomes Vesicles of Charged Lipids with Distinctly Shaped Nanoscale 862

Sphere- Pear- Tube- or Rod-Segments Elsevier Inc 1st edn 2009 vol 465 863

62 A Iglic V Kralj-Iglic B Božič M Bobrowska-Haumlgerstrand B Isomaa and H 864

Haumlgerstrand Bioelectrochemistry 2000 52 203ndash211 865

63 T Shimizu M Masuda and H Minamikawa Chem Rev 2005 105 1401ndash1444 866

64 J V Selinger and J M Schnur Phys Rev Lett 1993 71 4091ndash4094 867

65 J M Schnur Science (80- ) 1993 262 1669ndash1676 868

66 B Tenchov R Koynova and G Rapp Biophys J 2001 80 1873ndash1890 869

67 R Koynova and B Tenchov Ser Biomech 2017 31 3ndash19 870

68 S J Angyal and R Le Fur Carbohydr Res 1984 126 15ndash26 871

69 J K Whitesell and M A Minton J Am Chem Soc 1987 109 225ndash228 872

70 T Takayama S Ando and I Ando J Mol Struct 1990 220 245ndash250 873

71 K Gawrisch N V Eldho and I V Polozov Chem Phys Lipids 2002 116 135ndash151 874

72 F Tielens C Gervais J F Lambert F Mauri and D Costa Chem Mater 2008 20 875

3336ndash3344 876

73 G A Jeffrey and Y Yeon Acta Crystallogr Sect B 1986 42 410ndash413 877

74 B Berglund and R W Vaughan J Chem Phys 1980 73 2037ndash2043 878

75 C Gervais C Coelho T Azaiumls J Maquet G Laurent F Pourpoint C Bonhomme 879

P Florian B Alonso G Guerrero P H Mutin and F Mauri J Magn Reson 2007 880

187 131ndash140 881

76 C Akita T Kawaguchi F Kaneko H Yamamoto and M Suzuki J Phys Chem B 882

2004 108 4862ndash4868 883

77 R N Purusottam L Seacutenicourt J J Lacapegravere and P Tekely Biochim Biophys Acta - 884

Biomembr 2015 1848 3134ndash3139 885

78 X Biarneacutes A Ardegravevol A Planas C Rovira A Laio and M Parrinello J Am Chem 886

Soc 2007 129 10686ndash10693 887

79 L M J Kroon‐Batenburg and J Kroon Biopolymers 1990 29 1243ndash1248 888

80 K N Kirschner and R J Woods Proc Natl Acad Sci U S A 2001 98 10541ndash889

10545 890

81 S Svenson B Kirste and J H Fuhrhop J Am Chem Soc 1994 116 11969ndash11975 891

82 I Sack S Macholl J H Fuhrhop and G Buntkowsky Phys Chem Chem Phys 892

2000 2 1781ndash1788 893

83 S Svenson J Koening and J H Fuhrhop J Phys Chem 1994 98 1022ndash1028 894

34

84 S Svenson A Schaefer and J H Fuhrhop J Chem Soc Perkin Trans 2 1994 2 895

1023ndash1028 896

85 K Bock and C Pedersen Adv Carbohydr Chem Biochem 1983 41 27ndash66 897

86 T B Grindley Structure and Conformation of Carbohydrates Springer Berlin 898

Heidelberg 2008 899

87 F Tantakitti J Boekhoven X Wang R V Kazantsev T Yu J Li E Zhuang R 900

Zandi J H Ortony C J Newcomb L C Palmer G S Shekhawat M O De La 901

Cruz G C Schatz and S I Stupp Nat Mater 2016 15 469ndash476 902

903

1

Supporting Information 904

905

Topological connection between vesicles and nanotubes in single-906

component lipid membranes driven by head-tail interactions 907

908

Niki Baccilea Ceacutedric Lorthioira Abdoul Aziz Baa Patrick Le Griela Cristina Coelhob Javier 909

Perezc Wim Soetaertd Sophie L K W Roelantsd 910

911

a Sorbonne Universiteacute Centre National de la Recherche Scientifique Laboratoire de Chimie de 912

la Matiegravere Condenseacutee de Paris LCMCP F-75005 Paris France 913

b Institut des Mateacuteriaux de Paris Centre Sorbonne Universiteacute Paris France 914

c Synchrotron Soleil LrsquoOrme des Merisiers Saint-Aubin BP48 91192 Gif-sur-Yvette Cedex 915

France 916

d InBio Department of Biotechnology Ghent University Ghent Belgium 917

918

919

Corresponding author 920

Dr Niki Baccile 921

E-mail address nikibaccilesorbonne-universitefr 922

Phone +33 1 44 27 56 77 923

924

925

926

2

a) 927

Parameter Method

Dry Matter (DM ) 996 Infrared balance 105degC

Glucose () 012 HPLC-Metacarb

Glycerol () nd HPLC-Metacarb

Free Fatty Acid content () lt001 Internal method BBEPP

GC with FID Detector

Oil () lt001 Internal method BBEPP

GC with FID Detector

Moulds (CFUg) lt10 3M-Nordval ndeg16 B

Yeasts (CFUg) lt10 3M-Nordval ndeg16 B

Anaerobic count (CFUg) lt10 SP-VG M005 B

Protein () tbd BCA protein assay

928

Figure S 1 - a) HPLC-ELSD chromatogram and composition table 929

930

931

932

933

3

b) 934

935

Group Ndeg δ ppm

CH2 1 354

CH2 2 153

CH2 3-712-14 132

CH2 811 203

CH 910 534

CH2 15 162

CH2 16 388

CH 1 424

CH 2 316

CH 3 334

CH 4 329

CH 5 354

CH2 6 366

Figure S 1 (next) ndash b) 1H NMR spectrum (MeOD-d4) recorded on the purified G-C181-OH compounds 936

and related assignment 937

938

2

811910 1rsquo

5rsquo1

3rsquo4rsquo

1rsquo1910

asymp asymp asympHDO MeOD Acetone

6rsquo2rsquo

1615

3-712-14

4

939

940

Figure S 2 ndash Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 130degC 941

5

942

943

Figure S 3 - Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 90degC 944

945

90degC

1

22

6

946

947

948

949

Figure S 4 ndash Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 70degC 950

951

21

11

1

2

5

3

4

45

5

6

7

7

7

952

953

954

Figure S 5 ndash Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 25degC 955

956

957

1

1 1

8

958

Figure S 6 ndash a) SANS experiment recorded at 90degC for a 5 mgmL G-C181-OH aqueous solution b-c) 959

WAXS experiments recorded on aqueous solutions of G-C181-OH at b) 5 mgmL and c) 25 mgmL d) 960

Structural interpretation of the repeating distances pointed out in c) 961

962

963

964

001 01 1

q Aring-1

25degC

68degC

88degC

111degC

01 1q Aring

-1

02 04 06 08 1q Aring

-1

001 01

10-1

100

101

Inte

nsity

cm

-1

q Aring-1

-2

T= 90degC

SANS WAXS WAXS

a) b) c)

dv

dL d= 144 Aring-1

di

ddH2O

dt

dvdL

dtdi d

d)

9

965

966

Figure S 7 ndash ssNMR MAS (10 kHz) experiments recorded on a G-C181-OH dry powder a) 1H spectra 967

recorded from 110degC to 70degC b) 13C single pulse (SP) spectra recorded with high power 1H decoupling from 968

110degC to 70degC c) 13C cross polarization (CP) spectra recorded at 110degC and contact time tc of 50 micros and 969

3000 micros d) 13C CP spectra recorded at 70degC and for 20 lt tc micros lt 3000 970

971

1

811

910

CH2 CH2

1rsquo

6rsquo

8 6 4 2 0 ppm

110

90

80

70

910

1rsquo

6rsquo16

1

2rsquo3rsquo4rsquo

811

OH

45 40 35 30 25 20 15

ppm

200 180 160 140 120 100 80 60 40 20 0

ppm140 120 100 80

ppm

CH2-gauche

3004

110

90

80

70

65

60

50

25

C91013003

C9 - 12869

CH2-all-trans

3198

C1rsquo10335

a) b)

1H

T degCT degC

13C Single Pulse MAS

A

B

A

B

70degC70degC

5rsquo

215

200 180 160 140 120 100 80 60 40 20 0

ppm

13C CPMAST= 110degC

200 180 160 140 120 100 80 60 40 20 0

ppm

132 131 130 129 128 127 126

ppm

106 105 104 103 102

ppm

13C CPMAST= 70degC

50

tc micros

3000

20

tc micros

50 500 1000 3000

tc micros

C102

13024 C910113003

C92

12869

C1rsquo213082

C1rsquo110443

A B

106 105 104 103 102

ppm

3000

1000

500

50

20

AB

c)

d)

10

972

Figure S 8 ndash a) Short and d) long T2 components displayed by the 1H transverse relaxation signal c) 1H 973

transverse relaxation signals recorded on a G-C181-OH dry powder from 110degC to 70degC The area of each 974

given resonance is plotted (log-lin) as a function of the echo time (CPMG pulse sequence) for each 975

temperature and fitted using the equation given above T2s and T2l indicate the short and long T2 relaxation 976

times respectively obtained by describing the 1H transverse relaxation signal using a biexponential decay 977

For H1rsquo H5rsquo and H6rsquo the 1H relaxation function could be described by a monoexponential decay 978

- Protons from the sugar group (H1rsquo H5rsquo H6rsquo) show a single T2 relaxation component (Figure 979

6c and Figure S 8) while protons of the acyl chain (CH2 H910 H811) show both short (Figure 980

6c) and long (Figure 6d) T2 relaxation component 981

- The short T2 relaxation time systematically decreases from 110degC to 70degC for all species by a 982

factor ranging from two to seven depending on the specific chemical group Very long T2 are 983

observed at 110degC for all acyl species while nearly unchanged values are measured below 984

90degC without any clear trend 985

- The sugar protons (H1rsquo H5rsquo H6rsquo) have the shortest set of T2 values (lt 1 ms) below 100degC 986

The long component of T2 generally ranges between 4 ms and 12 ms 987

0 1 2 3 4 5 6 7 8

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1 2 3 4 5 6 7 8

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

H1

H8H11

H9H10

CH2

H1rsquo

H6rsquo

CH2

H910H811 H1CH2

H1rsquo H5rsquo6rsquo

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

OH

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45 T= 110degC

T= 90degC

T= 80degC

T= 70degC

Log

In

ten

sity

Time ms

Fit

)

CH2 H910 H811 H1 H56 H1 OH

00

05

10

15

20

25

30

35

40

T2 m

s

T= 110degC

T= 90degC

T= 80degC

T= 70degC

CH2 H910 H811 H1

0

2

4

6

8

10

12

T2 m

s

T= 110degC

T= 90degC

T= 80degC

T= 70degC

Short T2(1H) Long T2(1H)

a) b)

c)

11

Table S 1 ndash 1H chemical shifts of G-C181-OH in solution and in the solid-state Solid-state spectra are 988

recorded at a MAS frequency of 10 kHz in DMSO-d6 broad peak 989

990

δH ppm

Group Ndeg Liquid

(MeOD-d4) Solid

(T= 110degC) Solid

(T= 70degC)

CH2 H1 354 357

CH2 H2 153 156

CH2 H3-712-14 132 135

CH2 H811 203 203

CH H910 534 534

CH2 H15 162 164

CH2 H16 388 383

CH H1 424 432

CH H2 316 335

CH H3 334 335

CH H4 329 335

CH H5 354 351 357

CH2 H6 366 383

OH OH1 432

45 463

OH OH2 492

OH OH3 490

OH OH4 487

OH OH6 445

991

992

H1H9H10H1rsquo

H6rsquo

OH2rsquo

OH1

12

Detailed discussion about the NOESY and DQ-SQ 2D 1H-1H ssNMR experiments presented in 993

Figure 9 994

995

2D 1H-1H ssNMR experiments are a unique tool to explore through-space proximities 996

between 1H nuclei ssNMR offers a large number of possibilities each adapted to a given 997

system The advantage of 2D maps with respect to a dynamics approach is their direct 998

readability in clear two nuclei which exchange their magnetization through-space (whichever 999

mechanism may be responsible for this exchange) will display a specific cross-peak However 1000

the way cross-peaks are presented and interpreted strictly depends on the nature of the pulse 1001

program which in turn is chosen according to the physical nature of the sample 1002

1D temperature-dependent 1H NMR experiments performed on G-C181-OH (Figure 1003

4a in the main text) indicate sharp peaks at 90degC and broad peaks at 70degC Narrow peaks 1004

indicate fast andor large-amplitude molecular tumbling averaging the strong 1H-1H dipolar 1005

interactions This excludes the use of pulse sequences like DQ-SQ which is not efficient under 1006

these conditions For this reason we employ spin-diffusion-based pulse sequences like 1007

NOESY which is more adapted for samples with high molecular mobility When mixing times 1008

are chosen to be short (lt 1 s) one generally explores through-space proximities within about 5 1009

10 Aring On the contrary when molecular motions are restricted dipolar coupling becomes 1010

significant 1H NMR peaks are broader and pulse sequences like DQ-SQ can now be used 1011

with a spatial resolution of less than 5 Aring 1012

Whenever possible DQ-SQ experiments are preferred over spin diffusion-based pulse 1013

sequences because they are more selective in particular for likewise interactions which are 1014

not discriminated in NOESY-type experiments This is illustrated by the typical interaction 1015

schemes for NOESY (top Figure 9a) and DQ-SQ (top Figure 9b) For two given A and B 1016

protons 2D NOESY maps will display two on-diagonal and two square-shaped symmetrical 1017

off-diagonal cross-peaks Off-diagonal cross-peaks illustrate the mutual A-B and B-A 1018

interaction but the on-diagonal peaks are systematic and do not prove any A-A or B-B 1019

interaction Similarly to NOESY 2D DQ-SQ maps display off-diagonal peaks (symmetric 1020

across the diagonal top Figure 9b) upon A-B and B-A interaction but in contrast they will 1021

display on-diagonal peaks only in the presence of self-interaction In the tutorial example of 1022

Figure 9b top the on-diagonal A-A peak indicates mutual A-A interaction while lack of on-1023

diagonal B-B peak evidences the absence of B-B proximities Combining NOESY and DQ-SQ 1024

experiments is capital to study the local arrangement of G-C181-OH and its evolution with 1025

temperature 1026

13

At T= 90degC the H1 and H5rsquo contributions are very close but possible to discriminate 1027

them (δ(H1)= 354 ppm δ(H5rsquo)= 351 ppm) according to 1H NMR chemical shifts in Table S 1028

1 2D 1H-1H NOESY in Figure 9a also discriminates between the two components attributed 1029

on the basis of the interactions with their mutual neighbors (green H1-H2 and H5rsquo-H1rsquo squares 1030

in Figure 9a) On this basis 2D NOESY does not show a direct H1rsquo-H1 correlation (blue red-1031

crossed) but rather an extensive network of intra-sugar interactions These features exclude 1032

suggest intra- and inter-sugar interactions and they exclude head-tail arrangements as a major 1033

molecular configuration at 90degC 1034

At 70degC the 1H NMR peaks are broader indicating a reduction of the T2 (1H) relaxation 1035

times andor residual homonuclear dipolar interactions excluding the use of NOESY In this 1036

case 1H-1H DQ-SQ correlation experiments are best suited Figure 9b presents the 1H-1H DQ-1037

SQ experiment on G-C181-OH at 70degC Peaks are broader than in NOESY and attribution may 1038

be cumbersome However the experiment is performed at 700 MHz and MAS= 22 kHz which 1039

improves spectral resolution with respect to the same experiment performed at 300 MHz and 1040

MAS= 10 kHz The auto-peak at δH= 356 ppm in the SQ dimension is attributed to H1-H1 spin 1041

pairs mainly resulting from intra CH2 interaction while the DQ cross-peaks at δH= 156-356 1042

ppm and δH= 432-356 ppm respectively origin from H2-H1 and H1rsquo-H1 interactions (Figure 1043

9b) DQ-SQ map does not show any auto-peak corresponding to H1rsquo-H1rsquo or Hxrsquo-Hxrsquo (x= 2rsquo 1044

3rsquo 4rsquo) interactions (red-crossed blue Figure 9b) thus excluding any proximity (lt 5 Aring) between 1045

close glucose neighbors On the contrary DQ-SQ maps (recorded both at 700 MHz Figure 9b 1046

or at 300 MHz not shown) systematically show DQ cross-peaks which can only be attributed 1047

to H1rsquo-H1 interactions (δH= 432-356) The presence of these H1rsquo-H1 interactions can only be 1048

explained by a head-tail arrangement of the G-C181-OH molecules as illustrated below the 1049

DQ-SQ map in Figure 9b 1050

1

Topological connection between vesicles and nanotubes in single-1

component lipid membranes driven by head-tail interactions 2

3

Niki Baccilea Ceacutedric Lorthioira Abdoul Aziz Baa Patrick Le Griela Cristina Coelho Diogob 4

Javier Perezc Wim Soetaertd Sophie L K W Roelantsd 5

6

a Sorbonne Universiteacute Centre National de la Recherche Scientifique Laboratoire de Chimie de 7

la Matiegravere Condenseacutee de Paris LCMCP F-75005 Paris France 8

b Institut des Mateacuteriaux de Paris Centre Sorbonne Universiteacute Paris France 9

c Synchrotron Soleil LrsquoOrme des Merisiers Saint-Aubin BP48 91192 Gif-sur-Yvette Cedex 10

France 11

d InBio Department of Biotechnology Ghent University Ghent Belgium 12

13

Corresponding author 14

Dr Niki Baccile 15

E-mail address nikibaccilesorbonne-universitefr 16

Phone +33 1 44 27 56 77 17

2

Abstract 18

Lipid nanotube-vesicle networks are important channels for intercellular communication and 19

transport of matter Experimentally observed in neighboring mammalian cells but also 20

reproduced in model membrane systems a broad consensus exists on their formation and 21

stability Lipid membranes must be composed of at least two components each stabilizing low 22

(generally a phospholipid) and high curvatures Strong anisotropy or enhanced conical shape 23

of the second amphiphilic component is crucial for the formation of nanotunnels Anisotropic 24

driving forces generally favor nanotube protrusions from vesicles In the present work we 25

report the unique case of topologically-connected nanotubes-vesicles obtained in the absence 26

of directional forces in single-component membranes composed of an anisotropic bolaform 27

glucolipid above its melting temperature Tm Cryo-TEM and fluorescence confocal 28

microscopy show the interconnection between vesicles and nanotubes in a single-phase region 29

between 60deg and 90degC under diluted conditions Solid-state NMR necessarily recorded on the 30

dry powder yet demonstrates that the glucolipid can simultaneously assume two distinct 31

configurations head-head and head-tail These arrangements seemingly of comparable energy 32

above the Tm could explain the existence and stability of the topologically-connected vesicles 33

and nanotubes which are generally not observed for classical single-component phospholipid-34

based membranes above their Tm 35

36

Keywords Nanotube vesicle networks Tunnelling nanotubes Block liposomes Liposomes 37

Lipid nanotubes Biosurfactants Microbial glycolipids 38

39

Introduction 40

Topological connections between closed lipidic compartments through nanotubes1ndash3 41

have been shown to play a crucial role in the transfer of matter and communication in 42

neighboring mammalian cells4 These singular nanosystems observed since the lsquo90s as 43

spontaneous non-equilibrium structures in electroformed model liposome membranes56 have 44

since been largely studied both experimentally and theoretically7ndash9 Addressed in the literature 45

by different terms tunnelling nanotubes (TNT)149 block liposomes10ndash12 or nanotube-vesicle 46

networks13ndash16 (instead of tubes some work speak of tethers617) all refer to a similar 47

phenomenon driven by various internal or external forces The latter must overcome the energy 48

barrier needed to bend a phospholipid bilayer from low positive mean and gaussian curvatures 49

(vesicle) to a high mean and zero Gaussian (tube) curvatures 50

3

A large body of both experimental and theoretical work has shown that budding and 51

eventual nanotube formation from an existing membrane can only occur spontaneously for 52

membranes of at least two-components9102425111218ndash23 and below a vesicle-nanotube phase 53

transition event2627 Otherwise external anisotropic driving forces like electroformation5ndash54

82829 osmotic pressure6 laser ldquotweezersrdquo30 or electrodynamics13ndash16 must be employed to obtain 55

similar results with in some cases an impressive degree of 2D and 3D organization13ndash16 The 56

origin of spontaneous nanotube formation has been shown to be related to a nanoscale phase 57

separation between two membrane components stabilizing low and high curvatures 58

respectively10ndash1231 with at least one component being highly anisotropic203233 Theoretically 59

this behavior has been explained by deviations in the elastic properties of membranes due to in-60

plane orientational ordering of membrane inclusions composed of anisotropic amphiphiles 61

these referring to a non-symmetrical shape upon a 90deg tilt along the amphiphile axis8919242533 62

In this work we show unexpected nanotubing of membranes prepared from a single-63

component lipid in the absence of external directional forces and above the lipidrsquos Tm This 64

phenomenon is observed for a novel anisotropic double amphiphile (bolaform amphiphile or 65

bolaamphiphile) a glucolipid composed of β-D-glucose and a C181-cis fatty alcohol (G-66

C181-OH Figure 1) This compound is obtained by microbial fermentation of a genetically-67

modified S bombicola yeast in the presence of oleyl alcohol34 and is developed in the broader 68

context of extending the library of new biobased surfactants and lipids in view of replacing 69

petrochemical low molecular weight amphiphiles35ndash41 The structure of G-C181-OH is 70

analogous to that of other microbial glycolipids developed through genetic engineering4243 71

Topological connections between nanotubes and vesicles are observed by means of 72

cryogenic transmission electron microscopy (cryo-TEM) fluorescence microscopy and wide 73

angle X-ray scattering (WAXS) above the melting temperatures Tm= 483degC under strong 74

dilution (05 wt) Spin diffusion and 1H double-quantum single-quantum (DQ-SQ) 2D solid-75

state nuclear magnetic resonance (ssNMR) spectroscopy under magic angle spinning (MAS) 76

necessarily performed on the dry glucolipid powder provides a spectroscopic ground to 77

understand the vesicle-nanotube coexistence The bolaform glucolipids could be in a head-78

headtail-tail configuration in the vesicles but rather in a head-tail configuration in the 79

nanotubes In the dry glucolipid both configurations are energetically comparable in the single-80

phase region above Tm and their coexistence is evidenced by the splitting of some 13C NMR 81

signals These facts could explain the stability of nanotubes while the following hypotheses are 82

formulated for their formation membrane inclusions with different orientational ordering13ndash16 83

possibly driven by inter-vesicle collisions 84

4

85

86

87

88

89

Figure 1 ndash Non acetylated C181 alcohol glucoside G-C181-OH is obtained by a bioprocess performed 90

with modified S bombicola yeast 91

92

Experimental Section 93

Synthesis of non acetylated C181 Alcohol Glucosides (G-C181-OH) G-C181-OH (Mw= 94

41856 gmol-1) was produced by aerobic whole cell bioprocess with a modified S bombicola 95

strain as described by Van Renterghem et al (Fig S4 in Ref 34) The molecule was purchased 96

from the Bio Base Europe Pilot Plant (Gent Belgium) and has the generalized chemical 97

structure given in figure Figure 1 The HPLC and 1H NMR spectrum (MeOD-d4) with peak 98

assignment are shown in Figure S 1 High purity levels (99) and high degree of uniformity 99

were obtained as can be derived from HPLC-ELSD chromatogram 1H NMR and table of 100

contaminant given in Figure S 1 101

102

1H solution Nuclear Magnetic Resonance (NMR) 1H solution NMR experiments were 103

performed on a Bruker Avance III 300 spectrometer using a 5 mm 1H-X BBFO probe using 104

methanol-d4 as solvent The number of transients is 8 with 3 s recycling delay an acquisition 105

time of 546 s and a receiver gain of 362 The 1H NMR spectrum and relative assignment are 106

shown in Figure S 1 and Table S 1 13C solution NMR were performed on the same probe using 107

DMSO-d6 Corresponding chemical shifts are given in Table 1 Referencing is done with 108

respect to TMS δ1H= 0 ppm δ13C= 0 ppm 109

110

Sample preparation The sample was dissolved in milliQ-grade water at the concentration of 5 111

mgmL (05 wt) Lack of pH-sensitive probes (eg COOH groups) in G-C181-OH as 112

otherwise found in other microbial amphiphiles44 but also the will to avoid ion-specific 113

effects45 exclude the use of buffer The solution was heat at the given temperature using a CH3-114

150 Combitherm-2 dry block heating device For the cryo-TEM experiments the vials were 115

heat at the given temperature next to the cryofixation device and plunged into liquid ethane as 116

fast as possible For temperatures of 25degC and 70degC the sample was heat at 90degC and then 117

Synthesis path

5

cooled at to the desired temperature For the SAXS experiment the sample was heat at 90degC in 118

the block heating device and cooled at to 25degC before analysis For the SANS experiment the 119

sample was heat at 90degC directly in front of the beam using the thermalized sample holder 120

available 121

122

Differential Scanning Calorimetry (DSC) DSC was performed using a DSC Q20 apparatus 123

from TA Instruments equipped with the Advantage for Q Series Version acquisition software 124

(v540) Acquisition was performed on both hydrated and dry powder sample (~ 3-5 mg) sealed 125

in a classical aluminium cup and using an immediate sequence of heating and cooling ramps at 126

a rate of 10degCmin-1 Melting temperatures Tm 1 and 2 Tm1 and Tm2 were taken at the minimum 127

of the endothermic peak 128

129

Small and Wide Angle Scattering experiments Small angle neutron scattering (SANS) 130

experiments were performed at the D11 beamline of Institut Laue Langevin (Grenoble France) 131

during the run No 9-13-778 Four q-ranges have been explored and merged using the following 132

wavelengths λ and sample-to-detector (StD) distances 1) ultra-low q λ= 135Aring StD= 39 m 133

2) low-q λ= 53Aring StD= 39 m 3) mid-q λ= 53Aring StD= 8 m 4) high-q λ= 53Aring StD= 14 m 134

The sample (C= 5 mgmL-1) was prepared in 999 D2O to limit the incoherent background 135

scattering The sample solution was analyzed in standard 1 mm quartz cells Direct beam empty 136

quartz cell H2O (incoherent scatterer) within the quartz cell were recorded and boron carbide 137

(B4C) was used as neutron absorber The sample acquisition was measured at 90degC where 138

temperature was controlled through the controller thermalized sample holder available at the 139

beamline The background sample (D2O) signal was subtracted from the experimental data 140

Absolute values of the scattering intensity were obtained from the direct determination of the 141

number of neutrons in the incident beam and the detector cell solid angle The 2D raw data were 142

corrected for the ambient background and empty cell scattering and normalized to yield an 143

absolute scale (cross section per unit volume) by the neutron flux on the samples The data were 144

then circularly averaged to yield the 1D intensity distribution I(q) The software package Grasp 145

(developed at ILL and available free of charge) was used to integrate the data while the 146

software package SAXSUtilities (developed at ESRF and available free of charge) was used to 147

merge the data acquired at all configurations and subtract the background 148

Wide-angle X-ray scattering (WAXS) was performed under temperature control at the 149

SWING beamline of SOLEIL synchrotron facility (Saint-Aubin France) during the run 150

20201747 (energy 14 keV sample-to-detector distance 05 m) Two sample concentrations in 151

6

H2O were probed C= 5 mgmL-1 and C= 25 mgmL-1 The 2D data were integrated azimuthally 152

at the beamline using the software Foxtrot and in order to obtain the I(q) vs q spectrum after 153

masking the beam stop shadow Silver behenate (d(100) = 5838 Aring) was used as a standard to 154

calibrate the q-scale Sample solutions were inserted in borosilicate capillaries of 15 mm in 155

diameter Capillaries were flame-sealed A capillary oven with controlled temperature (plusmn 05degC) 156

provided at the beamline was used to control the sample temperature between 25degC and 111degC 157

Data were normalized by the transmission and calibrated to the SAXS signal of H2O at large q-158

values (I= 00163 cm-1) in order to obtain an absolute intensity scale The water signal was 159

measured by subtracting the signal of the empty capillary from the signal of a water-filled 160

capillary The signal of (water + capillary) was used as background for the samples and it was 161

subtracted after integration of the 2D data 162

163

Cryogenic transmission electron microscopy (cryo-TEM) Cryo-TEM experiments were carried 164

out on an FEI Tecnai 120 twin microscope operating at 120 kV and equipped with a Gatan 165

Orius CCD numeric camera The sample holder was a Gatan Cryoholder (Gatan 626DH 166

Gatan) Digital Micrograph software was used for image acquisition Cryofixation was 167

performed using a homemade cryofixation device The solutions were deposited on a glow-168

discharged holey carbon coated TEM copper grid (Quantifoil R22 Germany) Excess solution 169

was removed and the grid was immediately plunged into liquid ethane at -180degC before 170

transferring them into liquid nitrogen All grids were kept at liquid nitrogen temperature 171

throughout all experimentation Cryo-TEM images have been treated and analyzed using Fiji 172

(is just ImageJ) software available free of charge at the developerrsquos website46 173

The samplersquos concentration was C= 5 mgmL-1 For technical reasons the temperature 174

during cryofixation could not be controlled For this reason the temperature associated with 175

each cryo-TEM image in the text was related to the equilibration temperature just before 176

cryofixation the device being as close as possible to the dry bath used for temperature control 177

Since temperature control during cryofixation was not possible in our device and practically 178

impossible at high temperatures one should take the T values as indicative of the given value 179

Although the sample transfer between the dry bath and the cryofixation device was performed 180

as fast as possible it is more than likely that the actual temperature of the sample at the moment 181

of cryofixation was lower than the one indicated here although impossible to know precisely 182

183

7

Fluorescence microscopy Images were recorded using a 40x objective on a Leica SP5 upright 184

confocal microscope with 405 nm laser excitation The sample was marked with 6-dodecanoyl-185

NN-dimethyl-2-naphthylamine (Laurdan Mw= 35354 gmol-1) fluorophore as follows A 186

stock solution of 3 mg of Laurdan in 100 microL of acetone was initially prepared (C= 30 mgmL-187

1) A volume of 028 microL of the stock Laurdan solution was diluted in 1 mL of a 5 mgmL-1 G-188

C181-OH water solution temporary set at 110degC in a sealed Eppendorf The latter was 189

important so to enhance inclusion of Laurdan in the membrane constituted by the glucolipid 190

while in its vesicle phase The final Laurdan concentration was approximately 24 micromol with a 191

molar Laurdan-to-G-C181-OH molar ratio of approximately 1-to-500 Fluorophore-to-lipid 192

ratios above 1200 were generally considered as non-interfering with the lipid phase behavior 193

as also verified in this work 194

Laurdan can be excited between 365 nm and 410 nm and its emission wavelength can 195

vary according to the rigidity of the membrane intercalation within a rigid or flexible 196

membrane generates emissions at 440 nm or 490 nm respectively47 In the present experiment 197

we have flame-sealed a solution of G-C181-OH at 5 mgmL-1 containing Laurdan (1-to-500 in 198

lipid-to-Laurdan ratio) in flat optical capillaries of 01 mm thickness Capillaries were set on a 199

Linkam support for temperature control eventually placed under the microscope The sample 200

was excited at 405 nm and detection was made in spectral mode with a 5 nm window sliding 201

every 5 nm from 412 nm to 602 nm in 16 bits which allows quantify the signal and compare 202

the different conditions Quantitative analysis was performed according to the literature4849 203

several ROI (range of interest) have been drawn on several images using the ROI manager of 204

Fiji (is just ImageJ) software The integrated intensity of each ROI at 440 nm and 490 nm was 205

eventually recorded and used to calculate the value of the generalized polarization GP as 206

follows of 207

119866119875 =(120582440 minus 120582490)

120582440 + 120582490 208

A negative GP indicates a flexible membrane while a positive GP indicates a rigid membrane 209

210

Solid-state Nuclear Magnetic Resonance (ssNMR) 1H and 13C ssNMR experiments have been 211

performed using an Avance III HD Bruker 705 T (ν1H= 300 MHz) spectrometer and a 4 mm 212

magic angle spinning (MAS) probe except for the BAck-to-BAck (BABA) experiment which 213

was performed on an Avance III Bruker 1644 T (ν1H= 700 MHz) spectrometer equipped with 214

25 mm MAS probe Temperature was controlled through a Smart cooler BCUIIa BCU-xtreme 215

unit using 10 min of equilibration time after each temperature variation Temperature was 216

8

always varied from high (110degC) to low (70degC) before analysis Due to combined use of fast 217

MAS rates necessary to achieve high resolution spectra and heating close to the boiling point 218

of water neither diluted nor concentrated samples could be used for ssNMR experiments All 219

tests employing up to 50 wt G-C181-OH in water resulted in sample centrifugation drying 220

but also expulsion from the rotor causing sudden stop of rotor spinning For these reasons all 221

attempts to study hydrated G-C181-OH failed and only the dry powder could be analyzed The 222

dry sample powder was analyzed as such in zirconia rotors (4 mm or 25 mm) spinning at MAS= 223

10 kHz (4 mm) or 22 kHz (25 mm) and Chemical shifts were calibrated with respect to 224

adamantane (13C 13C(CH2)= 3848 ppm = 191 ppm) and glycine alpha (13C(CO)= 17603 225

ppm) Details on the acquisition parameters are given in the Supporting Information 226

227

Results and discussion 228

Topologically-connected nanotubes and vesicles 229

G-C181-OH is a biobased glucolipid water-insoluble at room temperature and of 230

which the physicochemical properties are poorly known The DSC thermogram performed at 231

10degCmin of G-C181-OH as a function of the water content (Figure 2a) shows two endothermic 232

peaks on the dry powder (996 wt glucolipid) They are attributed to two melting phenomena 233

occurring at the following temperatures Tm 483degC (Tm1) and 957degC (Tm2) the enthalpy of the 234

former being lower than that of the latter Interestingly DSC also shows that Tm1 is hydration 235

independent and reversible (the associated exothermic peak is systematically observed) while 236

the transition at Tm2 is simply suppressed (here tested up to 130degC) on hydrated powders and 237

never observed on the cooling profile even at slow cooling rates (1degCmin) The thermogram 238

recorded on the dry powder is typical of single-component lipids having an intermediate50 239

metastable phase (often referred to the ripple phase Pβ´ in reference to the periodic undulation 240

of the bilayer)50ndash54 between the classical lamellar gel (Lβ) and liquid crystalline (Lα) phases50ndash241

54 and characterized by regions of both liquid order and disorder5455 The effect of water does 242

not influence the low-temperature crystallization at Tm1 but it contributes to disrupt the ordered 243

array of hydrocarbon chains in the metastable phase 244

On the basis of the DSC thermogram the aqueous phase behavior of G-C181-OH under 245

diluted conditions (5 mgmL-1) was studied at four different temperatures (red marks on Figure 246

2a) below and above Tm1 but also between Tm1 and Tm2 corresponding to the thermogram of 247

the dry powder 248

9

249

Figure 2 ndash a) DSC thermogram of G-C181-OH under dry (04 wt water estimated by thermogravimetric 250

analysis) and hydrated conditions recorded at a heating rate of 10degCmin Red marks indicate the 251

temperatures at which cryo-TEM experiments are performed on the corresponding 5 mgmL aqueous 252

solutions b) Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 130degC 253

90degC 70degC and 25degC T indicates the temperature of the sample solution before cryofixation Tne T T 254

being the temperature of the sample upon plunging in liquid ethane Please refer to the materials and 255

methods section for more information 256

T= 130degC

T= 90degC

T= 70degC

T= 25degC

a)

b)

20 40 60 80 100 120

99 wt

50 wt

25 wt

T degC

En

do

10 wt

Tm1= 483 degC

Tm2= 957 degC

10

257

The series of cryo-TEM images in Figure 2b show an overview of the most relevant 258

structures observed on G-C181-OH solution from T= 130degC to T= 25degC whereas T is 259

indicative of the equilibrium temperature before cryofixation (refer to materials and methods 260

section for more information on the sense and interpretation of T) At Tgt 100degC the sample 261

is essentially composed of polydisperse single wall vesicles massively shown by the 262

complementary pictures presented in Figure S 2 In between Tm1 and Tm2 at T= 90degC and T= 263

70degC (Figure 2b Figure 3 Figure 4 complemented by Figure S 3 and Figure S 4) vesicles are 264

connected by nanotubes At Tlt Tm1 flat crystals are otherwise observed (Figure 2b and Figure 265

S 5) 266

267

268

Figure 3 ndash Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 90degC 269

270

A deeper insight at T= 90degC is given in Figure 3 (more images are provided in Figure 271

S 3) Arrows 1 through 6 mark very clearly the typical continuity in the glucolipid membrane 272

between the vesicle and nanotubes Arrow Ndeg 7 shows that a single nanotube connects the outer 273

membrane of two vesicles several hundred nanometers apart The connection is shown by 274

2

34

5

6

6

1

5

7

A

B

D

C

0 10 20 30 40 50 60

BIn

tensity

au

Profile nm

A

0 10 20 30 40 50 60

Inte

nsity

au

Profile nm

C

D

T= 90degC

11

arrows Ndeg5 and Ndeg6 The corresponding profiles A through D confirm the tubular nature of 275

the elongated objects Profiles B and D are both characterized by typical symmetrical thick 276

walls (low intensity depths highlighted by shaded rectangles) surrounding a hollow region (high 277

intensity peak between the shaded rectangles) The wall thickness varies between 35 and 40 278

nm for profile B and between 41 and 61 nm for profile D while the total nanotube diameter 279

varies between 14 nm at B and 20 nm at D At the same time the thickness of the vesicle wall 280

is about 6 nm at A and 135 nm at C Considerations about the bilayer or interdigitated structure 281

of the membrane will be presented below 282

The connection between nanotubes and vesicles seems to be multiple For instance 283

arrow Ndeg1 (Figure S 3) and Ndeg4 (Figure 3) show a single-wall vesicle of which the membrane 284

is extruded into a nanotube Similarly a closer look at arrows Ndeg1 2 and 3 in Figure 3 show 285

that the vesicle is single-wall and fully closed while the nanotube nucleates at the vesicle outer 286

surface with no apparent topological continuity In many regions of the sample at T= 90degC 287

one can find exvaginations or buds232531 in the nanotube wall This is shown for instance by 288

arrows Ndeg7 and particularly put in evidence in Figure 4 by arrows Ndeg1 through Ndeg4 where the 289

amplitude of the oscillation varies between 20-25 nm (arrow Ndeg1) and 6-8 nm (arrows Ndeg3-4) 290

291

Figure 4 - Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heated at T= 90degC 292

293

Still between Tm2 and Tm1 but at lower temperature (T= 70degC) the nanotubes seem to 294

vary from single threads (arrow Ndeg1 Figure S 4) to nanotubes of diameter below 15 nm both 295

connecting nanotubes of diameter ranging between 50 nm and 100 nm (arrows Ndeg2 Figure S 296

4) The larger tubes are most likely formed by the inflation and flattening of the nanotubes 297

probably after assimilation of surrounding vesicles and fusion of the exvagination discussed 298

above The transition between vesicles tubes and flat membranes is certainly more accentuated 299

Arrow Ndeg3 in Figure S 4 shows distinct vesicles while arrow Ndeg4 identifies a flat structure 300

evidently formed from fused vesicles In particular arrow Ndeg5 points at the fusion of two 301

2

1 132

4

12

vesicles while arrow Ndeg6 points at two vesicles fusing into the flat membrane Additional areas 302

where vesicle fusion is prominent are indicated by the oscillations at the outer skirt of broad 303

lamellae and pointed by arrow Ndeg7 304

Finally below Tm1 at T= 25degC for instance mainly flat crystals with no visible 305

exvaginations or tubes are visible (Figure S 5) by cryo-TEM and the material turns into an 306

insoluble precipitate Massive presence of flat crystalline structures is also clear from the bulk 307

iridescent behavior of a G-C181-OH solution under manual shear 308

In summary despite the uncertainty related to the actual state of equilibrium of the 309

glucolipids at the moment of cryofixation cryo-TEM depicts a vesicle phase at Tgt 100degC a 310

flat lamellar phase below Tm1 and an interconnected vesicle-nanotube network between Tm1 and 311

about 100degC The combination of statistically-relevant fluorescence microscopy and 312

scatteringdiffraction experiments provides an additional proof of the coexistence of 313

vesiclenanotube systems between the two transition temperatures Please note that the 314

experiments reported below have been performed under controlled conditions of temperature 315

which is then indicated by the use of T instead of T 316

13

317

Figure 5 ndash a-c) Confocal fluorescence microscopy images recorded at 90degC and 70degC for a 5 mgmL G-318

C181-OH aqueous solution d-g) Typical fluorescence microscopy images recorded at T= 80degC using 319

Laurdan (refer to materials and methods section for more details on this fluorophore) Excitation 320

wavelength is 405 nm for all the images while emission wavelength is 490 nm in (df) and 440 nm in (eg) 321

h) Full emission spectra obtained from the integration of various fluorescence images recorded from 412 322

nm to 600 nm (step 5 nm) at 90degC 80degC and 70degC i) Box plot of GP defined as 119918119927 =(120640120786120786120782minus120640120786120791120782)

120640120786120786120782+120640120786120791120782 with 120640120786120786120782 323

and 120640120786120791120782 being the emission wavelengths at 440 nm and 490 nm at 90degC 80degC and 70degC Each data point 324

l= 490 nm l= 440 nm

l= 490 nm l= 440 nm

10 μm400 450 500 550 600

00

02

04

06

08

10

No

rma

lize

d I

nte

nsity

l nm

90degC

80degC

80degC

70degC

70degC

400 450 500 550 60000

02

04

06

08

10

No

rma

lize

d I

nte

nsity

l nm

90degC

80degC

80degC

70degC

70degC

d) e)

f) g)

h)

i)

T= 80degC

Laurdan

T= 80degC

T= 80degC T= 80degC

T= 90degC T= 70degC T= 70degC

a) b) c)

1

1

1

14

corresponds to a set of regions of interest (ROIs) (54 at 90degC 65 at 80degC and 75 at 70degC) selected in the 325

corresponding fluorescence images ROIs were representatively selected so to include the entire image but 326

also vesicles alone tubes alone and mixtures of vesicles and tubes 327

328

Figure 5ab shows two typical confocal fluorescence microscopy images recorded on a 329

G-C181-OH solution (5 mgmL) in a flame-sealed capillary heat at T= 90degC and T= 70degC 330

Although the resolution of optical microscopy is not comparable with cryo-TEM the sample is 331

broadly composed of vesicles coexisting with nanotubes whereas in some cases the nanotube 332

starts from the vesicle itself (arrow Ndeg1) in good agreement with cryo-TEM data Furthermore 333

SANS recorded at T= 90degC (Figure S 6a) shows a scattering profile with a pronounced -2 334

dependence of the scattering intensity against the wavevector q in the log-log scale This 335

indicates the massive presence of a flat interface and it is expected typical of bilayers in 336

vesicular morphologies56 However the typical signature of nanotubes cannot be observed in 337

the SANS profile and this can be explained by both an argument related to the relative lower 338

amount of nanotubes with respect to vesicles at 90degC but also to the steeper slope of flat 339

morphologies (-2) compared to tubes (-1) the signal of the former masking the tube signal at 340

low wavevector values The SANS profile is characterized by a diffraction peak at q= 0167 Aring-341

1 of which the assignment is refined by WAXS data 342

Temperature-driven WAXS experiments in Figure S 6bc (5 mgmL-1 and 25 mgmL-1 343

respectively) show a lamellar order possibly attributed to a multilamellar structure of the 344

vesicles (dv Figure S 6d) at high temperature and to the interplanar arrangement of the 345

crystalline flat structures (dL Figure S 6d) at T= 25degC The sharp (001) reflection indicates 346

typical interlamellar distance of 353 Aring at 25degC although this value varies with temperature 347

(400 Aring at 111degC 372 Aring at 88degC) thus confirming its assignment to a hydration-dependent 348

interlamellar order More intriguing is the second broader peak at q= 0403 Aring-1 (d= 156 Aring) 349

and its reflection both clearly visible at 88degC below Tm2 We attribute it to the inter-glucolipid 350

arrangement inside the tubular phase (dt Figure S 6d) coexisting with the vesicle phase 351

between Tm1 and Tm2 and to the planar lipid arrangement (di Figure S 6d) in the rigid flat 352

lamellar structure below Tm2 Its temperature- hence hydration- independent position also 353

confirms its structural origin Finally besides the water-structure peak at about 2 Aring-1 we also 354

observe a broad peak at 144 Aring-1 characteristics of the lateral packing (d Figure S 6d) of lipids 355

in bilayers While its position could be consistent with an Lα phase5758 above Tm2 its broadness 356

and positional invariance as well as the coexistence between vesicles and tubes do not allow a 357

precise assignment 358

15

Does G-C181-OH assume a bilayer or interdigitated structure the latter being the one 359

found for other similar bolaform glucolipids4243 The size of G-C181-OH can be estimated to 360

be at the most 32 Aring given by the length L of the C16 tail (L = 22 Aring from the Tanford formula 361

L= 154 + 1265 times n (n= 16)59) and the sugar (about 10 Aring)4243 However the cis conformation 362

of the C=C bond imposes a ldquokinkrdquo of 60deg in the aliphatic chain which reduces the effective 363

length of the oleic backbone to about 17 Aring for an effective total G-C181-OH length of about 364

27 Aring Cryo-TEM provides a tube thickness varying from 35 Aring to 60 Aring and a vesicle membrane 365

thickness of at least 60 Aring WAXS suggests an interlamellar thickness (membrane and water) 366

between 40 Aring and 37 Aring and an inter-lipid distance in the tube of about 16 Aring In the absence of 367

a clear form factor oscillation profile as found for similar glucolipids4243 one must 368

acknowledge that these data pledge neither in favor of fully interdigitated nor a double layer 369

structure and might suggest a tilted lipid arrangement with partial interdigitation60 370

371

13C ssNMR demonstrates that G-C181-OH can assume two coexisting head-head and 372

head-tail configurations 373

Vesicles are characterized by low positive mean and gaussian curvatures while 374

nanotubes are described by a combination of a high positive mean and zero gaussian curvatures 375

In this regard the coexistence of vesicles and nanotubes requires glucolipids to adapt 376

themselves to both environments at the same time For this reason the formation of nanotube-377

vesicle networks generally requires a specific driving force able to impose a high mean and 378

zero gaussian curvature to the membrane In many systems insertion of at least a second 379

constituent (lipid surfactant protein) in the membrane becomes a necessary internal driving 380

force for the nucleation and growth of so-called membrane inclusions from which 381

topologically-connected nanotubes can form10ndash123261 as explained by orientational ordering 382

arguments62 It has also been shown that external driving forces such as use of laser 383

ldquotweezersrdquo30 osmotic pressure6 electroformation5ndash82829 or controlled suction13ndash16 are 384

otherwise necessary as also supported by theoretical studies19 385

The bending energy of tubes is higher than that of a vesicle membrane8 due to the 386

dependence of the membrane free energy from the principal curvatures For this reason 387

isotropic (rotational symmetry along the main axis) but also anisotropic33 lipids generally 388

stabilize low-curvature shapes like flat membranes or vesicles Lipid nanotubes are also a 389

common phase formed by amphiphiles but generally for lipids with a certain complexity in 390

their molecular structures as exampled by diacetylenic derivatives of phospholipids among 391

others262763ndash65 Negative contributions to the bending energy favoring the simultaneous 392

16

presence of interconnected vesicles and tubes was shown to be driven by the presence of 393

anisotropic lipids in a membrane continuum Spontaneous deviations to the elastic theory in the 394

presence of surfactants lipids and proteins with anisotropic structures favor the formation of 395

protrusions with radius of the order of the fraction of a micron9ndash1219ndash2362 396

In the absence of both internal and external driving forces mentioned above G-C181-397

OH should behave as a standard lipid Its DSC profile (Figure 2)536667 under both hydrated and 398

dry conditions identifies a hydration-independent transition (Tm1) slightly above room 399

temperature this is followed by one main transition at higher temperature (Tm2) in the dry 400

sample50 For such a profile one expects a gel phase (Lβ) below Tm1 and a liquid crystalline 401

structure possibly in the morphology of vesicles (L4 phase) possibly above Tm1 and most likely 402

above Tm231 as experimentally found On the basis of the thermogram recorded on the dry 403

powder one may also expect a ldquoripplerdquo (Pβ) phase between Tm1 and Tm2 characterized by 404

periodic undulations of the bilayer (generally addressed to as regions with both liquid order and 405

disorder)5455 406

In the present work instead of a Lβ phase we observe flat crystals below Tm1 and a 407

vesicle-nanotube network instead of the ldquoripplerdquo phase The latter is particularly unexpected 408

because the membrane is composed of a single-component (G-C181-OH) and no anisotropic 409

external force (osmotic hydrodynamic electric) is used to pull and stabilize the tubes In 410

addition the nanotubes are observed at fairly high temperatures above Tm1 (between about 411

60degC and 90degC) while this morphology is generally favored at low temperature (proportionality 412

with 1kT figure 5 in Ref 62) commonly below a given transition temperature characteristics 413

for each specific amphiphile27 Furthermore nanotubes of G-C181-OH seem to be stable in 414

time thus differing from what is found in electroformed vesicles78 They are also observed 415

away from phase transition events (may them be at Tm1 or Tm2) as otherwise reported before 416

for a diacetylenic derivative of phosphatidylcholine26 and many other systems27 417

A stable nanotube-vesicle network for a single-component phospholipid in the absence 418

of external driving forces is not expected and has never been reported before It could be 419

explained by the coexistence of two micro-separated phases of different rigidity This 420

hypothesis is excluded by confocal microscopy on Laurdan-labelled membranes and magic 421

angle spinning (MAS) solid-state NMR (ssNMR) experiments 422

Laurdan is a well-known fluorophore used to put in evidence the coexistence of rigid 423

and fluid domains in both model and biological membranes47ndash49 Laurdan was used here to label 424

the G-C181-OH membranes between Tm1 and Tm2 and to verify the possible micro-segregation 425

and coexistence of rigid (eg Lβ λem= 440 nm) and fluid (eg Lα λem= 490 nm) domains in 426

17

tubes and vesicles respectively A set of quantitative experiments performed on a 5 mgmL-1 427

solution summarized in Figure 5 shows an excess of fluorescence at only λem= 490 nm for 428

temperature between 90degC and 70degC (Figure 5h) The GP value which quantifies the rigidfluid 429

character of the membrane is systematically negative (Figure 5i) indicating an overall fluid 430

membrane at all temperatures GP is quantified on more than 50 regions of interest (ROIs) at 431

each temperature including vesicles and tubes alone as well as connected nanotube-vesicles 432

(Figure 5d-g) The dispersion of GP at 80degC and 70degC between -08 and -02 could suggest the 433

spurious presence of membranes with various degrees of flexibility although fluidity is always 434

predominant The combination of WAXS (broad peak d= 144 Aring-1 invariant with 435

temperature Figure S 6bc) and confocal fluorescence microscopy using Laurdan confirms the 436

single-phase of the membrane composed of G-C181-OH both in vesicles and nanotubes This 437

is also consistent with the lack of phase transitions in DSC experiments between ~85degC and 438

~60degC especially in hydrated samples We conclude that the vesicle-nanotube transition is not 439

driven by the local segregation of gel phase (Lβ) microdomains nor that the nanotubes are 440

composed of such phase 441

ssNMR is the best-suited approach to investigate the origin of the local driving force for 442

the nucleation of anisotropic inclusions of which the average orientation depends on the local 443

membrane and inclusion curvatures and was shown to promote tubing62 ssNMR can provide 444

information on the local molecular mobility eg through the nuclear spin relaxation behavior 445

but also on intermolecular interactions and stereochemistry eg through variations in the 446

chemical shift68ndash70 447

Ideally ssNMR should be performed on diluted G-C181-OH samples above Tm1 448

However fast MAS necessary for high resolution centrifuges the sample As a first 449

approximation hydrated G-C181-OH samples for instance at 50 wt as classically done for 450

phospholipids71 could be studied Unfortunately permeability of the sample holder combined 451

to heating (close to the boiling point of water) and MAS provoke dehydration and even sample 452

ejection After experiencing both of the above this approach was abandoned As a second 453

approximation the study of the dehydrated G-C181-OH powder avoids centrifugation and 454

evaporation If studying dried lipids may not reflect their exact behaviour under diluted 455

conditions valuable structural information can still be obtained in dehydrated molecular 456

systems like proteins thus illustrating that this approach is not unreasonable Furthermore if 457

DSC (Figure 2a) shows that hydration suppresses the Tm2 it also shows that the value of Tm1 is 458

not affected by water at concentrations as low as 10 wt For this reason performing a 459

structural analysis of dehydrated G-C181-OH by ssNMR above Tm1 may help understanding 460

18

the local arrangement of G-C181-OH in two morphologies vesicles and nanotubes having 461

sensibly different curvatures 462

To explore the local mobility of the most relevant chemical groups of G-C181-OH 463

above Tm2 (T= 110degC) and between Tm2 and Tm1 (70 lt T degC lt 90) we combine a series of 1H 464

and 13C experiments Single-pulse (SP) MAS experiments (1H and 13C) when recorded under 465

quantitative conditions allow the detection of all chemical species while cross-polarization 466

(CP) MAS (13C only) experiments are sensitive only to constrained groups for which the 1H-467

13C dipolar coupling is not averaged to zero by the molecular motions 468

469

Figure 6 ndash a) 1H MAS (10 kHz) ssNMR spectra recorded from 110degC to 70degC b) 13C MAS (10 kHz) ssNMR 470

spectra recorded using high power decoupling single pulse (SP) and cross-polarization (CP) pulse sequences 471

Contact time for CP experiments and corresponding temperatures are given on each spectrum Full spectra 472

are given in Figure S 7 473

The most relevant 1H and 13C spectra are given in Figure 6 (full range in Figure 7 and 474

Figure S 7) As expected the 1H solid-state SPMAS spectra show broader peaks compared to 475

the solution-state NMR spectrum (Figure 6a) reflecting the obvious solid-state character of the 476

sample at any temperature The sugar group and the C=C bond resonate in the 25-6 ppm range 477

and 535 ppm respectively while the OH groups of the sugar moiety are found below 45 ppm 478

(Figure 6a full assignment in Table S 1) Very broad (~1 ppm) at 100degC the position of the 479

OH peak varies with the temperature value and becomes narrower (lt~05 ppm) and visible at 480

δH= 464 ppm at T= 70degC Line narrowing and chemical shift variations of the OH resonance 481

in ssNMR is always a direct sign of H-bonding interactions whereas the strength of the H-482

bonding hence its OHhellipH distance is correlated to a dispersion in chemical shifts72ndash75 All other 483

60 55 50 45 40 35 30 25

ppm

110

90

80

70

910

1rsquo

6rsquo16

1

OH

134 132 130 128 126

ppm110 108 106 104 102 100

ppm

SP (T= 110degC)

SP (T= 70degC)

CP 50 ms (T= 70degC)

CP 3 ms (T= 70degC)

910 1rsquo

1H 13C

1

811

910

CH2 CH2

1rsquo

6rsquo

a) b)

C102

13024

C9101

13003

C92

12869 C1rsquo210382

C1rsquo110443

C1rsquo13035

2rsquo3rsquo4rsquo

T degC

5rsquo

19

resonances broaden from 110degC to 70degC thus indicating slower andor more anisotropic 484

reorientational motions of all chemical groups The specific dynamic behaviour of the sugar 485

headgroup is also reflected by its 1H transverse relaxation time T2 whereas the shorter the T2 486

the higher the rigidity Experiments performed between 110degC and 70degC shown and 487

commented in Figure S 8 in the Supporting Information globally show a single short (lt 1 ms 488

below 100degC) T2 relaxation component for H1rsquo H5rsquo and H6rsquo and two both short and long 489

(between 4 ms and 12 ms) T2 for the acyl chain (CH2 H910 H811) even below 100degC 490

At 110degC 13C SPMAS experiments (Figure 6b full range in Figure 7 and Figure S 7 491

assignments in Table 1) show a single peak at δC= 10335 ppm for C1rsquo and at δC= 13003 ppm 492

for C910 as expected for a symmetrical conformation with two chemically-equivalent 493

carbons76 At 70degC the C1rsquo peak disappears while C910 now shows two peaks around 128 494

ppm and 130 ppm (Table 1 full temperature series in Figure S 7b) attributed to C92 (δC= 12869 495

ppm) and C10 in a non-symmetric configuration respectively for which these two carbons are 496

chemically inequivalent76 More precisely the resonance at 130 ppm is actually the 497

superposition of two close resonances C9101 (δC= 13003 ppm) and C102 (δC= 13024 ppm) 498

where subscripts 1 and 2 stand for two different conformations of the cis-olefin group as 499

reported for various phases of oleic acid in the vicinity of 0degC76 500

501

Table 1 ndash 13C Chemical shifts of G-C181-OH in solution and solid-state Solid-state experiments are 502

recorded at MAS= 10 kHz All spectra are reported in Figure 7 and Figure S 7 The contact time for cross 503

polarization is 3 ms In all ssNMR experiments temperature is first raised at 110degC and then lowered to 504

either 90degC or 70degC Attribution of the 13C resonances in Table 1 is performed on the basis of the solution 505

spectrum (Figure S 1b) Additional peaks observed in both single pulse and cross polarization ssNMR 506

experiments are attributed when possible to a configuration effect of the given atom eg C9101 and C92 507

indicate two configurations of C9 All new peaks observed in solid-state and not detected in solution are 508

highlighted in red for convenience 509

510

511

δ ppm

Peak Liquid (DMSO-d6)

Solid Single Pulse

T= 110degC

Solid Single Pulse

T= 90degC

Solid Single Pulse

T= 70degC

Solid Cross Polarization

T= 70degC

C102

- - 13024 13024

C910C9101 13009 13003 13003 13003 13003

C1C9C10C1rsquo

C6rsquo

20

C92 - - - 12869 12869

C1rsquoC11 10329 10335 10335 - 10443

C12 - - - - 10382

C3 7724 7709 7702 - 7829

- 7634 7629 - 7621

7525

C2 7388 7398 7390 7370 7370

C4 7052 7077 7054 7041 7098

C16 6900 7016 - - 7029

- - - - 6793

C1C5C6 6153 6241 6236 6239 6262

C1C5C6 6117 - - - 5956

C2 3300 3290 3290 3283 3279

CH2(trans) - - - 3198 3196

CH2(gauche)C15 2941 2981 2987 3004 -

C811 2706 2747 2747 2832 2817

CH2 2597 2625 2628 2630 2817

512

13C CPMAS spectra at 70degC (Figure 6b full range in Figure 7 and Figure S 7d) at 513

increasing contact times confirm the presence of two cis-olefin environments one for which 514

C910 are equivalent labelled here C9101 and one for which they are inequivalent labelled 515

C92 C102 Both environments are rigid enough to be observed at short contact time (tc= 50 μs) 516

A similar double environment is also observed for the sugar C1rsquo CPMAS at 70degC shows two 517

peaks (C1rsquo1 at δC= 10443 ppm C1rsquo2 at δC= 10382 ppm Figure 6b Figure S 7d) These are 518

significantly shifted downfield from the C1rsquo peak observed at 110degC and are detected at contact 519

times as short as 50 μs thus indicating their rigid character One also finds that CPMAS is 520

totally inefficient at 100degC for short contact times (50 μs) and poorly efficient even at 3 ms 521

typical of fast-tumbling molecules (Figure S 7c) also both gauche (δC= 3004 ppm) and all-522

trans (δC= 3198 ppm) conformations of the CH2 carbons coexist at 70degC (Figure S 7b)77 523

According to ssNMR one main mobile molecular environment exists at 110degC as 524

shown by the well-resolved 1H spectrum single C=C (C9101) and anomeric (C1rsquo) resonances 525

gauche conformation of the aliphatic CH2 and single T2 of the aliphatic CH2 chain and sugar 526

protons The sugar group is certainly mobile (inefficient 1H13C CPMAS) but still more 527

constrainted than the acyl chain (shorter T2 (1H)) As the temperature is lowered the overall 528

rigidity increases (broadening of 1H NMR peaks faster 1H transverse relaxation and efficient 529

1H13C magnetization transfer during CPMAS) presumably driven by H-bonding 530

interactions (downfield shift and line narrowing of the OH 1H signal very short OH T2 loss of 531

the C1rsquo signal in SPMAS) At the same time a number of spectroscopic data (two concomitant 532

21

resonances for anomeric C1rsquo1 and C1rsquo2 cis-olefin C9101 and C92 C102 and CH2 gauche and 533

all-trans environments) and relaxation behaviour (fast and slow T2 relaxation components) 534

strongly suggests the coexistence of two molecular environments Further proofs of the 535

coexistence of two distinct molecular domains are given by a number of chemical shift 536

variations in the 13C CPMAS spectrum at 70degC compared to the one at 110degC 90degC and 537

solution This is shown in Figure 7 and in the list of 13C chemical shifts in Table 1 538

539

540

Figure 7 - Comparison between solution- and solid-state 13C NMR spectra of G-C181-OH Solid-state 541

experiments are recorded directly on the sample powder at a MAS of 10 kHz (4 mm zirconia rotor) SP 542

Single Pulse CP Cross Polarization In all the ssNMR experiments temperature is first raised at 110degC 543

and then lowered to either 90degC or 70degC Segmented lines are meant to highlight the known peak position 544

in solution thus stressing the strong chemical shift variations in the solid-state 545

546

NMR experiments above Tm2 suggest that vesicles could contain G-C181-OH with a 547

single family of mobile conformations characterized by chemically-equivalent C=C (red 548

Figure 8b) and a single C1rsquo conformation (green Figure 8b) Experiments between Tm1 and Tm2 549

otherwise suggest that two coexisting conformations characterized by chemically-equivalent 550

(red Figure 8b) and inequivalent (magenta Figure 8b) C=C could be associated to nanotubes 551

Two distinct C1rsquo sites (grey Figure 8b) also indicate two configurations of the sugar headgroup 552

A comment on this point will be given in the following paragraphs 553

The local (Aring-scale) variation of molecular mobility below Tm2 drives the coexistence of 554

two distinct molecular environments of the same G-C181-OH molecule This occurs within a 555

single-phase domain (as deduced by WAXS) of equivalent membrane rigidity (as derived from 556

Laurdan probe in confocal microscopy) Although never reported before the difference in the 557

local mobility of the same amphiphile (here G-C181-OH) could satisfy the theoretically-558

140 120 100 80 60 40 20

CP(3 ms)-70degC

SP-70degC

SP-90degC

SP-110degC

ppm

LiquidC910

C1rsquo

C2rsquo

C4rsquo

C3rsquo

C1

C6rsquoC5rsquo

C1

6

C8

11

C2 C

H2

C1

5

CH

2

DM

SO

13C

22

required conditions that induce local lipid segregation and formation of membrane inclusions 559

(Figure 8c) In case of their non-zero average orientation the local membrane curvature 560

increases62 and nanotubes can spontaneously form (Figure 8bc) More intriguing this 561

mechanism occurs at least 10degC above Tm1 although it should not be favored above the Tm27 562

From both experimental and theoretical backgrounds the conditions of topologically-563

connected vesicles-nanotubes systems are met only when either internal (two components) or 564

external driving forces are applied19 In the present single-lipid system of which the phase 565

behavior is summarized in Figure 8a the most plausible driving force could be explained by 566

the NMR results collected on the dry sample G-C181-OH concomitantly occupies two 567

molecular environments characterized by 1) two distinct conformations of the sugar 568

displaying equivalent mobility and 2) two conformations of the acyl chain unsaturation 569

570

23

571

Figure 8 ndash a) Thermotropic phase behavior of G-C181-OH at 5 mgmL-1 Note that Tm2 is observed on the 572

dry powder only b) Possible relationship between the molecular environment as derived from NMR and 573

membrane curvatures All the molecular drawings refer to the same G-C181-OH molecule (bulky ellipsoid 574

equiv sugar headgroup) while each color corresponds to a given configuration Grey of C1rsquo1 and C1rsquo2 indicates 575

the uncertainty about the assignment of the specific NMR peak The curvature C is defined as 1R R being 576

the curvature radius R1 R2 radius of curvature CG Gauss curvature CM Mean curvature c) Free 577

interpretation of the anisotropic high-curvature (CMgt 0 CG= 0) inclusions in the membrane (low CM and 578

CG) continuum driven by inter-sugar H-bonding and reduced intramembrane diffusion for a G-C181-OH 579

glucolipid 580

581

Flat crystalline structure

Nanotube-vesicle network

Vesicles

48degC

120deg

106 105 104 103 102 101

ppm

H9 H10H1rsquo

106 105 104 103 102 101

ppm106 105 104 103 102 101

ppm106 105 104 103 102 101

ppm

C=CC=C

C1rsquo1C1rsquo2

C1rsquo

Tm2(96degC)Tm1

a)

b)

c)

Same molecule different local orientations and sugar conformations

High curvature membrane

Low curvature inclusion

Flip-FlopStrong H-bonding

CG

CM ( + )

CG 0

CM 0 CG 0

CM 0

G-C181-OH

24

Two related questions are unanswered at the moment What actually drives nanotube 582

formation and is the organization of G-C181-OH within the vesicles and tubes the same Lipid 583

nanotubes have been studied for at least three decades over which broad experimental and 584

theoretical insight has been generated Despite this fact the actual understanding is still matter 585

of debate27 Micelle-to-nanotube and vesicle-to-nanotube phase transitions are generally 586

reported for single amphiphiles below a given Tm and intermediate phase composed of twisted 587

and flat fibers is concomitantly reported Initially thought to be a chirality-driven phenomenon 588

recent studies propose a more general mechanism involving crystallization and symmetry 589

breaking in the lipid packing driven by optimal packing27 If twisted ribbons were observed 590

before for a number of microbial glycolipid bolaamphiphiles44 they are not found for G-C181-591

OH This fact as well as the large body of cryo-TEM images and the observation of nanotubes 592

above Tm1 rather pledge in favour of a direct topological change from positive (vesicles) to 593

zero (nanotubes) Gaussian curvatures In this case vesicle-to-nanotube transitions could be 594

driven by external forces56307813ndash162829 595

Considering the absence of external forces in the present system there are only two 596

possibilities left spontaneous budding of the membrane or nanotube pulling after inter-vesicle 597

collisions due to Brownian motion Both are possible as inclusions composed of a different 598

arrangement of the same molecule in the membrane continuum can theoretically explain 599

spontaneous nanotube formation in an isotropic medium672223629ndash1218ndash21 However the second 600

one is probably more likely to prevail because it could explain the large number of nanotubes 601

connecting two distant vesicles Interestingly similar events occur for any phospholipid-based 602

vesicular solution but nanotubes are extremely rare unless at least two lipids of different 603

molecular structure are mixed together The spectroscopic 1H and 13C and relaxation (1H) 604

ssNMR data even if collected on a dry powder demonstrate that the coexistence of two G-605

C181-OH configurations involving both the sugar headgroup and fatty alcohol chain are 606

possible for this molecule As a matter of fact this is not so surprising as more than one energy 607

minima each corresponding to a given conformation are known for glucose its non-reducing 608

derivative methylglucose and carbohydrates in general78ndash80 609

The presence of two configurations could explain the coexistence between two 610

structures of different curvature but the data presented so far do not explain the origin of these 611

two configurations In other words what drives the change in the local structure of the sugar 612

headgroup and fatty alcohol within the membrane 613

The association of known self-assembled structures to spectroscopic 13C ssNMR data is 614

known for years in particular for gluconamides lipids having a linear sugar headgroup81ndash84 It 615

25

was found that gauche effects in some cases up to the γ position in the sugar chain induce 616

upfield chemical shifts while anti and trans conformations result in downfield shifts Each 617

structure (monolayer crystal bilayer crystal or micellar fiber) was then characterized by a given 618

organization of the gluconamide and a characteristic 13C ssNMR fingerprint which could be 619

explained by gauche γ-gauche anti and trans conformations of the headgroup Similar 620

structural-spectroscopic correlations have been collected over the years for much simpler cyclic 621

sugars85 whereas pyranoses can undergo important conformational changes (chair half-chair 622

boat skew) when the appropriate energy barriers are crossed78ndash8086 A quantitative 623

conformational study based on the 13C chemical shift variations shown in Table 1 would be 624

certainly helpful but very delicate in the context of the present state of the art because a 625

rigorous approach requires at least one known crystal structure8182 which is currently missing 626

for this compound A qualitative approach will then be adopted 627

G-C181-OH is a bolaform amphiphile and flip-flop could be a possible mechanism 628

involved in the nanotube stabilization whereas head-tail interactions have long been considered 629

as important in nanotube-forming bolaform amphiphiles6063 From mere considerations based 630

on T2 (1H) relaxation arguments one could exclude head-tail configurations if the R-CH2OH 631

endgroup (H1) was involved in a direct interaction with the sugar group its corresponding T2 632

would probably be comparable to the sugarrsquos one (OH and H1rsquo) but this does not seem to be 633

the case However this hypothesis would be based on the comparison of the relaxation from 634

CH and CH2 protons additionally involved in different kind of motions Besides the relaxation 635

of H1 is influenced by the strong overlap between H1 and H5rsquo For these reasons relaxation 636

arguments are not solid enough and will not be used 637

We then address this question by exploring intermolecular interactions in a direct 638

manner using 2D 1H-1H homonuclear ssNMR correlation spectroscopy At 90degC in the fast-639

tumbling domain of molecular motion (narrow 1H NMR peaks Figure 6a) and corresponding 640

to the vesicle-rich region spin diffusion (NOESY) experiments are preferred over dipolar 641

coupling based experiments because such interaction is averaged out by molecular motion 642

However when the molecular motion is more constrained in the domain of nanotube stability 643

at 70degC (broad 1H spectra Figure 6a) Double-Quantum Single-Quantum (DQ-SQ) correlation 644

experiments are best suited to probe interaction between close molecular neighbors These 645

experiments are reported in Figure 9 and the related interpretation is only briefly reported 646

below while a more detailed discussion can be found on Page 12 in the Supporting Information 647

26

648

Figure 9 ndash a) 2D 1H-1H NOESY NMR experiment performed at MAS= 10 kHz (4 mm probe) on a Bruker 649

300 MHz Avance III HD spectrometer using 300 ms of mixing time Sample temperature is regulated to 650

90degC reached by cooling from 110degC b) 2D 1H-1H DQ-SQ (BAck-to-BAck BABA pulse sequence used for 651

the DQ excitation scheme) NMR experiment performed at MAS= 22 kHz (25 mm probe) on a Bruker 700 652

MHz Avance III spectrometer using a DQ excitation time equal to 2 rotor periods Sample temperature is 653

regulated to 70degC reached by cooling from 110degC The typical schemes illustrating how to interpret these 654

2D experiments are reported above each experimental map Zooms of the sugar region and configurational 655

interpretation are given below each 2D map More information are given in the text below 656

H16-H16

H1-H1

H9-H10

H1rsquo-H1

H1rsquoH910 H16

H1

H1rsquo-H1rsquo

Hxrsquo-Hxrsquo

2

(pp

m)

(ppm)

315

811910 1rsquo

2rsquo2

3rsquo 4rsquo6rsquo

5rsquo

1rsquo2910

asymp asymp asympHDO MeOD Acetone

1

315

811 9101rsquo

2rsquo2

3rsquo4rsquo6rsquo

5rsquo

1rsquo2 9 10

asympasympasympHDOMeODAcetone

H1rsquo-H1

1rsquo

1

(p

pm

)

(ppm)

H1-H2

H1rsquo-H1

H1rsquo

H910

H1H5rsquo

H2

H1rsquo-H5rsquo

315

811910 1rsquo

2rsquo2

3rsquo 4rsquo6rsquo

5rsquo

1rsquo2910

asymp asymp asympHDO MeOD Acetone

1

H2rsquoH3rsquoH4rsquo

H2rsquoH3rsquoH4rsquo

X= 2rsquo3rsquo4rsquo

NOESY - T= 90degC

1H

1H

1H

1H

DQ-SQ - T= 70degC

a) b)

2

(pp

m)

(p

pm

)

(ppm)

A B

AhellipB

AhellipA

BhellipB

NOESY

(ppm)

A B

AhellipB

DQ-SQ

Single-quantum dimension

Do

ub

le-q

ua

ntu

m d

imen

sio

n

AhellipA

BhellipB

BhellipA

BhellipA

315

811910 1rsquo

2rsquo2

3rsquo 4rsquo6rsquo

5rsquo

1rsquo2910

asymp asymp asympHDO MeOD Acetone

1

lt 5 Aring

lt 5 Aring

H1-H2

H811

H15

H16

H6rsquo

H1rsquo

H1

H5rsquo

H2

H1-H2

H1rsquo-H1

H1rsquo-H5rsquo

H1rsquo

H1 H5rsquo

H1rsquo

H1

H5rsquo

H1H2rsquo H3rsquo H4rsquo

H1-H1

H1rsquo-H1

Hxrsquo-HxrsquoX= 2rsquo3rsquo4rsquo

H1-H2

H1rsquo

Zoom sugar region 28 ndash 45 ppmNOESY DQ-SQ

27

Head-tail or head-head configurations could be probed by the existence of H1rsquo-H1 657

interactions whereas H1 (δH= 356 ppm) is associated to the glucolipid tail and H1rsquo (δH= 432 658

ppm) to the sugar headgroup These atoms are more than 15 Aring apart and a direct interaction 659

could not be observed in both NOESY and DQ-SQ experiments NOESY experiments (Figure 660

9a) show that H1rsquo-H1 correlations are not observed Meanwhile clear-cut direct H1rsquo-H1 661

interactions are indeed observed are detected at T= 70degC (DQ cross-peaks related to H1rsquo-H1 662

correlation in DQ-SQ experiment shown in Figure 9b) Interestingly the absence of auto peaks 663

at δH= 432 ppm and δH= 335 ppm in the SQ dimension excludes H1rsquo-H1rsquo and Hxrsquo-Hxrsquo (x= 2 664

3 4) interactions respectively strengthening the idea that sugar headgroups from two nearest-665

neighbor G-C181-OH molecules are not adjacent in the nanotube-vesicle region 666

Homonuclear 2D ssNMR supports the idea according to which G-C181-OH can 667

undergo molecular flip-flops and a head-tail arrangement in the dry product Extrapolating these 668

conclusions in the lower concentration range lack of H1rsquo-H1 proximity at T= 90degC suggests a 669

head-head configuration in the vesicle-rich region and head-tail configuration within the 670

nanotubes It is however still unclear whether flip-flops are driven by spontaneous temperature-671

induced conformational changes in the glucolipid packing stimulating inclusions within the 672

membrane itself672223629ndash1218ndash21 or by defects generated by inter-vesicle collisions In fact a 673

combination of both is not unlikely Nanotubes are not stable when temperature is 20 to 30 674

degrees above Tm1 indicating that inter-vesicle collisions alone cannot explain the unusual 675

phenomenon In an energy landscape approach87 the head-head configuration is certainly 676

favored at high temperature while both head-head and head-tail configurations could be 677

favored in the mid-temperature region corresponding to the range between Tm1 and Tm2 in the 678

dry powder Both statistical and defect-induced flip-flop events contribute to form high mean 679

zero Gaussian curvature morphologies (nanotubes) for which the head-tail conformation 680

limiting the proximity between bulky sugar headgroups is probably more adapted (Figure 8c) 681

The flip-flop interpretation now reasonably explains the strong differences in the sugar and fatty 682

alcohol conformations deduced by the multiple NMR signals attributed to chemically-683

equivalent groups The existence of various low-energy conformations each having specific 684

NMR signature for sugars in solution78ndash8086 and solid-state81ndash84 further supports our 685

interpretation 686

One last point should be clarified Why between Tm1 and Tm2 are two molecular 687

environments able to coexist instead of a single one of average rigidity as found in the case of 688

ldquoripplerdquo phases50ndash54 We believe that this question can be answered based on the NMR results 689

the relatively short T2 values of the H1rsquo and OH protons shows the constrained mobility of the 690

28

sugar headgroup despite a certain flexibility of the chain (long T2 component) This trend may 691

suggest restrictions in translational mobility probably due to the sugar-OH hydrogen bonding 692

(Figure 8c) meaning that lipids are not able to diffuse within the membrane thus keeping two 693

different configurations stable 694

695

Conclusions 696

Topologically-connected nanotube-vesicle lipidic systems are commonly observed for 697

self-assembled membranes composed of at least two amphiphiles and spontaneously driven by 698

anisotropic inclusions andor by external directional forces This work shows that strong 699

differences in membrane curvature can spontaneously occur in single-lipid systems without 700

external anisotropic forces Coexistence and interconnection between nanotubes and vesicles 701

are evidenced via cryo-TEM and fluorescence confocal microscopy in a single-phase region at 702

temperature above a transition temperature Tm1= 483degC determined by DSC on the hydrated 703

sample but below the second transition temperature Tm2 (957degC) determined by DSC on the 704

dry powder X-ray diffraction (WAXS) also shows the coexistence of two lattice periods one 705

attributed to an inter-membrane repeating distance and the other possibly related to the intra-706

nanotube glucolipid arrangement Labelling the systems with Laurdan a fluorophore of which 707

the emission strongly varies with the local membrane rigidity excludes the presence of rigid 708

micro-separated phases (eg Lβ) 709

1H and 13C ssNMR despite the fact that it can only be performed on the dry powder 710

confirm the presence of one main fluid G-C181-OH environment above Tm2 Below Tm2 the 711

splitting of some 13C peaks and the coexistence of short and long 1H relaxation components 712

demonstrate that G-C181-OH can simultaneously pack into structurally and dynamically 713

inequivalent environments Combining advanced 2D 1H-1H correlation experiments based on 714

spin diffusion or double-quanta excitation suggests that vesicles could contain a preferred head-715

head configuration of the bolaform lipid while nanotubes a preferred head-tail assembly 716

The presence of two environments composed of the same molecule characterized by 717

different molecular configurations and rigidity fulfills the theoretical conditions necessary to 718

spontaneously nucleate a membrane inclusion of non-zero average orientation described as 719

necessary to observe topologically-connected nanotube-vesicle structures in the absence of 720

external forces This hypothesis could also be enriched by the possibility of nanotubes to be 721

pulled away between two vesicles after a collision event 722

This work expands the actual theoretical and experimental framework on lipid networks 723

stabilized by a different radius of curvature It could also motivate the synthesis or isolation 724

29

of new lipids providing a higher level of 3D architectural control in soft networks possibly in 725

the range of room temperature 726

727

Acknowledgements 728

Dr Andrea Lassenberger and Dr Sylvain Preacutevost at Institut Laue Langevin (ILL Grenoble 729

France) are kindly acknowledged for their assistance on the SANS experiment Confocal 730

images were performed at the Institut de Biologie Paris Seine (IBPS) imaging facility Jean-731

Franccedilois Gilles (Imaging Core Facility Institut de Biologie Paris Seine (IBPS) CNRS 732

Sorbonne Universiteacute Paris France) is kindly acknowledged for his assistance on the confocal 733

microscopy imaging Dr Guillaume Laurent (LCMCP Sorbonne Universiteacute Paris France) is 734

kindly acknowledged for helpful discussions The French region Ile-de-France SESAME 735

program is acknowledged for financial support (700 MHz NMR 736

spectrometer) Dr Olivier Diat (ICSM Marcoule France) Prof eronikaIglič and Prof V Aleš737

onsul discussi Slovenia) are kindly acknowledged for helpfUniversity of LjubljanaIglič (-Kralj 738

739

Financial support 740

WAXS experiments have been supported by Soleil Light Source Saint Aubin France proposal 741

Ndeg 20201747 SANS experiments have been supported by ILL proposal Ndeg9-13-778 742

743

References 744

1 D Kabaso N Bobrovska W Goacuteźdź E Gongadze V Kralj-Iglič R Zorec and A 745

Iglič Bioelectrochemistry 2012 87 204ndash210 746

2 C R Safinya U Raviv D J Needleman A Zidovska M C Choi M A Ojeda-747

Lopez K K Ewert Y Li H P Miller J Quispe B Carragher C S Potter M W 748

Kim S C Feinstein and L Wilson Adv Mater 2011 23 2260ndash2270 749

3 P Veranič M Lokar G J Schuumltz J Weghuber S Wieser H Haumlgerstrand V Kralj-750

Iglič and A Iglič Biophys J 2008 95 4416ndash4425 751

4 A Rustom R Saffrich I Markovic P Walther and H-H Gerdes Science (80- ) 752

2004 303 1007ndash1010 753

5 M I Angelova S Soleacuteau P Meacuteleacuteard F Faucon and P Bothorel Prog Colloid 754

Polym Sci 1992 89 127ndash131 755

6 L Mathivet S Cribier and P F Devaux Biophys J 1996 70 1112ndash1121 756

7 V Kralj-Iglič G Gomišček J Majhenc V Arrigler and S Svetina Colloids Surfaces 757

A Physicochem Eng Asp 2001 181 315ndash318 758

30

8 V Kralj-Iglič A Iglič G Gomišček F Sevšek V Arrigler and H Haumlgerstrand J 759

Phys A Math Gen 2002 35 1533ndash1549 760

9 M Drab D Stopar V Kralj-Iglič and A Iglič Cells 2019 8 626 761

10 A Zidovska K K Ewert J Quispe B Carragher C S Potter and C R Safinya 762

Langmuir 2009 25 2979ndash2985 763

11 A Zidovska K K Ewert J Quispe B Carragher C S Potter and C R Safinya Soft 764

Matter 2011 7 8363ndash8369 765

12 A Zidovska K K Ewert J Quispe B Carragher C S Potter and C R Safinya 766

Biochim Biophys Acta - Biomembr 2009 1788 1869ndash1876 767

13 M Karlsson K Sott M Davidson A S Cans P Linderholm D Chiu and O Orwar 768

Proc Natl Acad Sci U S A 2002 99 11573ndash11578 769

14 A Karlsson R Karlsson M Karlsson A-S Cans A Stromberg F Ryttsen and O 770

Orwar Nature 2001 409 150ndash152 771

15 M Karlsson K Sott A S Cans A Karlsson R Karlsson and O Orwar Langmuir 772

2001 17 6754ndash6758 773

16 E Evans H Bowman A Leung D Needham and D Tirrell Science (80- ) 1996 774

273 933ndash935 775

17 U Seifert Configurations of fluid membranes and vesicles 1997 vol 46 776

18 J Kaumls and E Sackmann Biophys J 1991 60 825ndash844 777

19 N Bobrovska W Goacuteźdź V Kralj-Iglič and A Iglič PLoS One 2013 8 e73941 778

20 A Iglič B Babnik K Bohinc M Fošnarič H Haumlgerstrand and V Kralj-Iglič J 779

Biomech 2007 40 579ndash585 780

21 K Bohinc T Slivnik A Iglič M Brumen and V Kralj-Iglič J Phys Chem C 2007 781

111 9709ndash9718 782

22 A Iglič H Haumlgerstrand P Veranič A Plemenitaš and V Kralj-Iglič J Theor Biol 783

2006 240 368ndash373 784

23 V Kralj-Iglič and P Veranič Adv Planar Lipid Bilayers Liposomes 2006 5 129ndash785

149 786

24 A Iglic and V Kralj-Iglic Adv Planar Lipid Bilayers Liposomes 2006 4 253ndash279 787

25 S Penič L Mesarec M Fošnarič L Mroacutewczyńska H Haumlgerstrand V Kralj-Iglič and 788

A Iglič Front Phys 2020 8 342 789

26 P Yager and P E Schoen Mol Cryst Liq Cryst 1984 106 371ndash381 790

27 T G Barclay K Constantopoulos and J Matisons Chem Rev 2014 114 10217ndash791

10291 792

31

28 B Božič G Gomišček V Kralj-Iglič S Svetina and B Žekš Eur Biophys J 2002 793

31 487ndash496 794

29 A Iglič H Haumlgerstrand M Bobrowska-Haumlgerstrand V Arrigler and V Kralj-Iglič 795

Phys Lett A 2003 310 493ndash497 796

30 R Bar-Ziv and E Moses Phys Rev Lett 1994 73 1392ndash1395 797

31 S Chiruvolu H E Warriner E Naranjo S H Idziak J O Raumldler R J Plano J A 798

Zasadzinski and C R Safinya Science (80- ) 1994 266 1222ndash1225 799

32 V Kralj-Iglič A Iglič H Haumlgerstrand and P Peterlin Phys Rev E 2000 61 4230ndash800

4234 801

33 L Mesarec M Drab S Penič V Kralj-Iglič and A Iglič Int J Mol Sci 2021 22 802

2348 803

34 L Van Renterghem S L K W Roelants N Baccile K Uyttersprot M C Taelman 804

B Everaert S Mincke S Ledegen S Debrouwer K Scholtens C V Stevens W 805

Soetaert L Van Renterghem S L K W Roelants N Baccile K Uytersprot M C 806

Taelman B Everaert S Mincke S Ledegen S Debrouwer K Scholtens C V 807

Stevens and W Soetaert Biotechnol Bioeng 2018 115 1195ndash1206 808

35 J Clark and F Deswarte Introduction to Chemicals from Biomass Second Edition 809

John Wiley amp Sons 2015 810

36 K V Sajna R Houmlfer R K Sukumaran L D Gottumukkala and A Pandey in 811

Industrial Biorefineries and White Biotechnology eds A Pandey R Houmlfer M 812

Taherzadeh K M Nampoothiri and C Larroche Amsterdam Oxford Waltham 813

Elsevier 2015 pp 499ndash521 814

37 R Marchant and I M Banat Trends Biotechnol 2012 30 558ndash565 815

38 K Hill and C LeHen-Ferrenbach in Sugar-Based Surfactants Fundamentals and 816

Applications ed C C Ruiz Boca Raton CRC Press 2009 pp 1ndash20 817

39 L Rodrigues I M Banat J Teixeira and R Oliveira J Antimicrob Chemother 818

2006 57 609ndash618 819

40 J D Desai and I M Banat Microbiol Mol Biol Rev 1997 61 47ndash64 820

41 S L K W Roelants K Ciesielska S L De Maeseneire H Moens B Everaert S 821

Verweire Q Denon B Vanlerberghe I N A Van Bogaert P Van der Meeren B 822

Devreese and W Soetaert Biotechnol Bioeng 2016 113 550ndash559 823

42 N Baccile M Selmane P Le Griel S Preacutevost J Perez C V Stevens E Delbeke S 824

Zibek M Guenther W Soetaert I N A Van Bogaert and S Roelants Langmuir 825

2016 32 6343ndash6359 826

32

43 N Baccile A-S Cuvier S Preacutevost C V Stevens E Delbeke J Berton W Soetaert 827

I N A Van Bogaert and S Roelants Langmuir 2016 32 10881ndash10894 828

44 N Baccile C Seyrig A Poirier S A Castro S L K W Roelants and S Abel Green 829

Chem 2021 23 3842ndash3944 830

45 A Salis L Cappai C Carucci D F Parsons and M Monduzzi J Phys Chem Lett 831

2020 11 6805ndash6811 832

46 J Schindelin I Arganda-Carreras E Frise V Kaynig M Longair T Pietzsch S 833

Preibisch C Rueden S Saalfeld B Schmid J-Y Tinevez D J White V 834

Hartenstein K Eliceiri P Tomancak and A Cardona Nat Methods 2012 9 676ndash835

682 836

47 T Parasassi and E Gratton J Fluoresc 1995 5 59ndash69 837

48 M M Dodes Traian F L Gonzaacutelez Flecha and V Levi J Lipid Res 2012 53 609ndash838

616 839

49 D M Owen C Rentero A Magenau A Abu-Siniyeh and K Gaus Nat Protoc 840

2012 7 24ndash35 841

50 W Stillwell in An Introduction to Biological Membranes Academic Press 2016 pp 842

135ndash180 843

51 W Stillwell in An Introduction to Biological Membranes Academic Press 2016 pp 844

181ndash220 845

52 B Tenchov Chem Phys Lipids 1991 57 165ndash177 846

53 R Koynova and M Caffrey Biochim Biophys Acta 1998 1376 91ndash145 847

54 F Schmid S Dolezel O Lenz and S Meinhardt J Phys Conf Ser 2014 487 848

012004 849

55 S Meinhardt R L C Vink and F Schmid Proc Natl Acad Sci U S A 2013 110 850

4476ndash4481 851

56 O Glatter and O Kratky Small Angle X-ray Scattering Academic Press London 852

1982 853

57 B A Cunningham A D Brown D H Wolfe W P Williams and A Brain Phys 854

Rev E 1998 58 3662ndash3672 855

58 M Nadler A Steiner T Dvir O Szekely P Szekely A Ginsburg R Asor R Resh 856

C Tamburu M Peres and U Raviv Soft Matter 2011 7 1512ndash1523 857

59 C Tanford The hydrophobic effect Formation of micelles and biological membranes 858

John Wiley amp Sons Inc New York 1973 859

60 M Masuda and T Shimizu Langmuir 2004 20 5969ndash5977 860

33

61 A Zidovska K K Ewert J Quispe B Carragher C S Potter and C R Safinya 861

Block Liposomes Vesicles of Charged Lipids with Distinctly Shaped Nanoscale 862

Sphere- Pear- Tube- or Rod-Segments Elsevier Inc 1st edn 2009 vol 465 863

62 A Iglic V Kralj-Iglic B Božič M Bobrowska-Haumlgerstrand B Isomaa and H 864

Haumlgerstrand Bioelectrochemistry 2000 52 203ndash211 865

63 T Shimizu M Masuda and H Minamikawa Chem Rev 2005 105 1401ndash1444 866

64 J V Selinger and J M Schnur Phys Rev Lett 1993 71 4091ndash4094 867

65 J M Schnur Science (80- ) 1993 262 1669ndash1676 868

66 B Tenchov R Koynova and G Rapp Biophys J 2001 80 1873ndash1890 869

67 R Koynova and B Tenchov Ser Biomech 2017 31 3ndash19 870

68 S J Angyal and R Le Fur Carbohydr Res 1984 126 15ndash26 871

69 J K Whitesell and M A Minton J Am Chem Soc 1987 109 225ndash228 872

70 T Takayama S Ando and I Ando J Mol Struct 1990 220 245ndash250 873

71 K Gawrisch N V Eldho and I V Polozov Chem Phys Lipids 2002 116 135ndash151 874

72 F Tielens C Gervais J F Lambert F Mauri and D Costa Chem Mater 2008 20 875

3336ndash3344 876

73 G A Jeffrey and Y Yeon Acta Crystallogr Sect B 1986 42 410ndash413 877

74 B Berglund and R W Vaughan J Chem Phys 1980 73 2037ndash2043 878

75 C Gervais C Coelho T Azaiumls J Maquet G Laurent F Pourpoint C Bonhomme 879

P Florian B Alonso G Guerrero P H Mutin and F Mauri J Magn Reson 2007 880

187 131ndash140 881

76 C Akita T Kawaguchi F Kaneko H Yamamoto and M Suzuki J Phys Chem B 882

2004 108 4862ndash4868 883

77 R N Purusottam L Seacutenicourt J J Lacapegravere and P Tekely Biochim Biophys Acta - 884

Biomembr 2015 1848 3134ndash3139 885

78 X Biarneacutes A Ardegravevol A Planas C Rovira A Laio and M Parrinello J Am Chem 886

Soc 2007 129 10686ndash10693 887

79 L M J Kroon‐Batenburg and J Kroon Biopolymers 1990 29 1243ndash1248 888

80 K N Kirschner and R J Woods Proc Natl Acad Sci U S A 2001 98 10541ndash889

10545 890

81 S Svenson B Kirste and J H Fuhrhop J Am Chem Soc 1994 116 11969ndash11975 891

82 I Sack S Macholl J H Fuhrhop and G Buntkowsky Phys Chem Chem Phys 892

2000 2 1781ndash1788 893

83 S Svenson J Koening and J H Fuhrhop J Phys Chem 1994 98 1022ndash1028 894

34

84 S Svenson A Schaefer and J H Fuhrhop J Chem Soc Perkin Trans 2 1994 2 895

1023ndash1028 896

85 K Bock and C Pedersen Adv Carbohydr Chem Biochem 1983 41 27ndash66 897

86 T B Grindley Structure and Conformation of Carbohydrates Springer Berlin 898

Heidelberg 2008 899

87 F Tantakitti J Boekhoven X Wang R V Kazantsev T Yu J Li E Zhuang R 900

Zandi J H Ortony C J Newcomb L C Palmer G S Shekhawat M O De La 901

Cruz G C Schatz and S I Stupp Nat Mater 2016 15 469ndash476 902

903

1

Supporting Information 904

905

Topological connection between vesicles and nanotubes in single-906

component lipid membranes driven by head-tail interactions 907

908

Niki Baccilea Ceacutedric Lorthioira Abdoul Aziz Baa Patrick Le Griela Cristina Coelhob Javier 909

Perezc Wim Soetaertd Sophie L K W Roelantsd 910

911

a Sorbonne Universiteacute Centre National de la Recherche Scientifique Laboratoire de Chimie de 912

la Matiegravere Condenseacutee de Paris LCMCP F-75005 Paris France 913

b Institut des Mateacuteriaux de Paris Centre Sorbonne Universiteacute Paris France 914

c Synchrotron Soleil LrsquoOrme des Merisiers Saint-Aubin BP48 91192 Gif-sur-Yvette Cedex 915

France 916

d InBio Department of Biotechnology Ghent University Ghent Belgium 917

918

919

Corresponding author 920

Dr Niki Baccile 921

E-mail address nikibaccilesorbonne-universitefr 922

Phone +33 1 44 27 56 77 923

924

925

926

2

a) 927

Parameter Method

Dry Matter (DM ) 996 Infrared balance 105degC

Glucose () 012 HPLC-Metacarb

Glycerol () nd HPLC-Metacarb

Free Fatty Acid content () lt001 Internal method BBEPP

GC with FID Detector

Oil () lt001 Internal method BBEPP

GC with FID Detector

Moulds (CFUg) lt10 3M-Nordval ndeg16 B

Yeasts (CFUg) lt10 3M-Nordval ndeg16 B

Anaerobic count (CFUg) lt10 SP-VG M005 B

Protein () tbd BCA protein assay

928

Figure S 1 - a) HPLC-ELSD chromatogram and composition table 929

930

931

932

933

3

b) 934

935

Group Ndeg δ ppm

CH2 1 354

CH2 2 153

CH2 3-712-14 132

CH2 811 203

CH 910 534

CH2 15 162

CH2 16 388

CH 1 424

CH 2 316

CH 3 334

CH 4 329

CH 5 354

CH2 6 366

Figure S 1 (next) ndash b) 1H NMR spectrum (MeOD-d4) recorded on the purified G-C181-OH compounds 936

and related assignment 937

938

2

811910 1rsquo

5rsquo1

3rsquo4rsquo

1rsquo1910

asymp asymp asympHDO MeOD Acetone

6rsquo2rsquo

1615

3-712-14

4

939

940

Figure S 2 ndash Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 130degC 941

5

942

943

Figure S 3 - Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 90degC 944

945

90degC

1

22

6

946

947

948

949

Figure S 4 ndash Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 70degC 950

951

21

11

1

2

5

3

4

45

5

6

7

7

7

952

953

954

Figure S 5 ndash Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 25degC 955

956

957

1

1 1

8

958

Figure S 6 ndash a) SANS experiment recorded at 90degC for a 5 mgmL G-C181-OH aqueous solution b-c) 959

WAXS experiments recorded on aqueous solutions of G-C181-OH at b) 5 mgmL and c) 25 mgmL d) 960

Structural interpretation of the repeating distances pointed out in c) 961

962

963

964

001 01 1

q Aring-1

25degC

68degC

88degC

111degC

01 1q Aring

-1

02 04 06 08 1q Aring

-1

001 01

10-1

100

101

Inte

nsity

cm

-1

q Aring-1

-2

T= 90degC

SANS WAXS WAXS

a) b) c)

dv

dL d= 144 Aring-1

di

ddH2O

dt

dvdL

dtdi d

d)

9

965

966

Figure S 7 ndash ssNMR MAS (10 kHz) experiments recorded on a G-C181-OH dry powder a) 1H spectra 967

recorded from 110degC to 70degC b) 13C single pulse (SP) spectra recorded with high power 1H decoupling from 968

110degC to 70degC c) 13C cross polarization (CP) spectra recorded at 110degC and contact time tc of 50 micros and 969

3000 micros d) 13C CP spectra recorded at 70degC and for 20 lt tc micros lt 3000 970

971

1

811

910

CH2 CH2

1rsquo

6rsquo

8 6 4 2 0 ppm

110

90

80

70

910

1rsquo

6rsquo16

1

2rsquo3rsquo4rsquo

811

OH

45 40 35 30 25 20 15

ppm

200 180 160 140 120 100 80 60 40 20 0

ppm140 120 100 80

ppm

CH2-gauche

3004

110

90

80

70

65

60

50

25

C91013003

C9 - 12869

CH2-all-trans

3198

C1rsquo10335

a) b)

1H

T degCT degC

13C Single Pulse MAS

A

B

A

B

70degC70degC

5rsquo

215

200 180 160 140 120 100 80 60 40 20 0

ppm

13C CPMAST= 110degC

200 180 160 140 120 100 80 60 40 20 0

ppm

132 131 130 129 128 127 126

ppm

106 105 104 103 102

ppm

13C CPMAST= 70degC

50

tc micros

3000

20

tc micros

50 500 1000 3000

tc micros

C102

13024 C910113003

C92

12869

C1rsquo213082

C1rsquo110443

A B

106 105 104 103 102

ppm

3000

1000

500

50

20

AB

c)

d)

10

972

Figure S 8 ndash a) Short and d) long T2 components displayed by the 1H transverse relaxation signal c) 1H 973

transverse relaxation signals recorded on a G-C181-OH dry powder from 110degC to 70degC The area of each 974

given resonance is plotted (log-lin) as a function of the echo time (CPMG pulse sequence) for each 975

temperature and fitted using the equation given above T2s and T2l indicate the short and long T2 relaxation 976

times respectively obtained by describing the 1H transverse relaxation signal using a biexponential decay 977

For H1rsquo H5rsquo and H6rsquo the 1H relaxation function could be described by a monoexponential decay 978

- Protons from the sugar group (H1rsquo H5rsquo H6rsquo) show a single T2 relaxation component (Figure 979

6c and Figure S 8) while protons of the acyl chain (CH2 H910 H811) show both short (Figure 980

6c) and long (Figure 6d) T2 relaxation component 981

- The short T2 relaxation time systematically decreases from 110degC to 70degC for all species by a 982

factor ranging from two to seven depending on the specific chemical group Very long T2 are 983

observed at 110degC for all acyl species while nearly unchanged values are measured below 984

90degC without any clear trend 985

- The sugar protons (H1rsquo H5rsquo H6rsquo) have the shortest set of T2 values (lt 1 ms) below 100degC 986

The long component of T2 generally ranges between 4 ms and 12 ms 987

0 1 2 3 4 5 6 7 8

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1 2 3 4 5 6 7 8

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

H1

H8H11

H9H10

CH2

H1rsquo

H6rsquo

CH2

H910H811 H1CH2

H1rsquo H5rsquo6rsquo

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

OH

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45 T= 110degC

T= 90degC

T= 80degC

T= 70degC

Log

In

ten

sity

Time ms

Fit

)

CH2 H910 H811 H1 H56 H1 OH

00

05

10

15

20

25

30

35

40

T2 m

s

T= 110degC

T= 90degC

T= 80degC

T= 70degC

CH2 H910 H811 H1

0

2

4

6

8

10

12

T2 m

s

T= 110degC

T= 90degC

T= 80degC

T= 70degC

Short T2(1H) Long T2(1H)

a) b)

c)

11

Table S 1 ndash 1H chemical shifts of G-C181-OH in solution and in the solid-state Solid-state spectra are 988

recorded at a MAS frequency of 10 kHz in DMSO-d6 broad peak 989

990

δH ppm

Group Ndeg Liquid

(MeOD-d4) Solid

(T= 110degC) Solid

(T= 70degC)

CH2 H1 354 357

CH2 H2 153 156

CH2 H3-712-14 132 135

CH2 H811 203 203

CH H910 534 534

CH2 H15 162 164

CH2 H16 388 383

CH H1 424 432

CH H2 316 335

CH H3 334 335

CH H4 329 335

CH H5 354 351 357

CH2 H6 366 383

OH OH1 432

45 463

OH OH2 492

OH OH3 490

OH OH4 487

OH OH6 445

991

992

H1H9H10H1rsquo

H6rsquo

OH2rsquo

OH1

12

Detailed discussion about the NOESY and DQ-SQ 2D 1H-1H ssNMR experiments presented in 993

Figure 9 994

995

2D 1H-1H ssNMR experiments are a unique tool to explore through-space proximities 996

between 1H nuclei ssNMR offers a large number of possibilities each adapted to a given 997

system The advantage of 2D maps with respect to a dynamics approach is their direct 998

readability in clear two nuclei which exchange their magnetization through-space (whichever 999

mechanism may be responsible for this exchange) will display a specific cross-peak However 1000

the way cross-peaks are presented and interpreted strictly depends on the nature of the pulse 1001

program which in turn is chosen according to the physical nature of the sample 1002

1D temperature-dependent 1H NMR experiments performed on G-C181-OH (Figure 1003

4a in the main text) indicate sharp peaks at 90degC and broad peaks at 70degC Narrow peaks 1004

indicate fast andor large-amplitude molecular tumbling averaging the strong 1H-1H dipolar 1005

interactions This excludes the use of pulse sequences like DQ-SQ which is not efficient under 1006

these conditions For this reason we employ spin-diffusion-based pulse sequences like 1007

NOESY which is more adapted for samples with high molecular mobility When mixing times 1008

are chosen to be short (lt 1 s) one generally explores through-space proximities within about 5 1009

10 Aring On the contrary when molecular motions are restricted dipolar coupling becomes 1010

significant 1H NMR peaks are broader and pulse sequences like DQ-SQ can now be used 1011

with a spatial resolution of less than 5 Aring 1012

Whenever possible DQ-SQ experiments are preferred over spin diffusion-based pulse 1013

sequences because they are more selective in particular for likewise interactions which are 1014

not discriminated in NOESY-type experiments This is illustrated by the typical interaction 1015

schemes for NOESY (top Figure 9a) and DQ-SQ (top Figure 9b) For two given A and B 1016

protons 2D NOESY maps will display two on-diagonal and two square-shaped symmetrical 1017

off-diagonal cross-peaks Off-diagonal cross-peaks illustrate the mutual A-B and B-A 1018

interaction but the on-diagonal peaks are systematic and do not prove any A-A or B-B 1019

interaction Similarly to NOESY 2D DQ-SQ maps display off-diagonal peaks (symmetric 1020

across the diagonal top Figure 9b) upon A-B and B-A interaction but in contrast they will 1021

display on-diagonal peaks only in the presence of self-interaction In the tutorial example of 1022

Figure 9b top the on-diagonal A-A peak indicates mutual A-A interaction while lack of on-1023

diagonal B-B peak evidences the absence of B-B proximities Combining NOESY and DQ-SQ 1024

experiments is capital to study the local arrangement of G-C181-OH and its evolution with 1025

temperature 1026

13

At T= 90degC the H1 and H5rsquo contributions are very close but possible to discriminate 1027

them (δ(H1)= 354 ppm δ(H5rsquo)= 351 ppm) according to 1H NMR chemical shifts in Table S 1028

1 2D 1H-1H NOESY in Figure 9a also discriminates between the two components attributed 1029

on the basis of the interactions with their mutual neighbors (green H1-H2 and H5rsquo-H1rsquo squares 1030

in Figure 9a) On this basis 2D NOESY does not show a direct H1rsquo-H1 correlation (blue red-1031

crossed) but rather an extensive network of intra-sugar interactions These features exclude 1032

suggest intra- and inter-sugar interactions and they exclude head-tail arrangements as a major 1033

molecular configuration at 90degC 1034

At 70degC the 1H NMR peaks are broader indicating a reduction of the T2 (1H) relaxation 1035

times andor residual homonuclear dipolar interactions excluding the use of NOESY In this 1036

case 1H-1H DQ-SQ correlation experiments are best suited Figure 9b presents the 1H-1H DQ-1037

SQ experiment on G-C181-OH at 70degC Peaks are broader than in NOESY and attribution may 1038

be cumbersome However the experiment is performed at 700 MHz and MAS= 22 kHz which 1039

improves spectral resolution with respect to the same experiment performed at 300 MHz and 1040

MAS= 10 kHz The auto-peak at δH= 356 ppm in the SQ dimension is attributed to H1-H1 spin 1041

pairs mainly resulting from intra CH2 interaction while the DQ cross-peaks at δH= 156-356 1042

ppm and δH= 432-356 ppm respectively origin from H2-H1 and H1rsquo-H1 interactions (Figure 1043

9b) DQ-SQ map does not show any auto-peak corresponding to H1rsquo-H1rsquo or Hxrsquo-Hxrsquo (x= 2rsquo 1044

3rsquo 4rsquo) interactions (red-crossed blue Figure 9b) thus excluding any proximity (lt 5 Aring) between 1045

close glucose neighbors On the contrary DQ-SQ maps (recorded both at 700 MHz Figure 9b 1046

or at 300 MHz not shown) systematically show DQ cross-peaks which can only be attributed 1047

to H1rsquo-H1 interactions (δH= 432-356) The presence of these H1rsquo-H1 interactions can only be 1048

explained by a head-tail arrangement of the G-C181-OH molecules as illustrated below the 1049

DQ-SQ map in Figure 9b 1050

2

Abstract 18

Lipid nanotube-vesicle networks are important channels for intercellular communication and 19

transport of matter Experimentally observed in neighboring mammalian cells but also 20

reproduced in model membrane systems a broad consensus exists on their formation and 21

stability Lipid membranes must be composed of at least two components each stabilizing low 22

(generally a phospholipid) and high curvatures Strong anisotropy or enhanced conical shape 23

of the second amphiphilic component is crucial for the formation of nanotunnels Anisotropic 24

driving forces generally favor nanotube protrusions from vesicles In the present work we 25

report the unique case of topologically-connected nanotubes-vesicles obtained in the absence 26

of directional forces in single-component membranes composed of an anisotropic bolaform 27

glucolipid above its melting temperature Tm Cryo-TEM and fluorescence confocal 28

microscopy show the interconnection between vesicles and nanotubes in a single-phase region 29

between 60deg and 90degC under diluted conditions Solid-state NMR necessarily recorded on the 30

dry powder yet demonstrates that the glucolipid can simultaneously assume two distinct 31

configurations head-head and head-tail These arrangements seemingly of comparable energy 32

above the Tm could explain the existence and stability of the topologically-connected vesicles 33

and nanotubes which are generally not observed for classical single-component phospholipid-34

based membranes above their Tm 35

36

Keywords Nanotube vesicle networks Tunnelling nanotubes Block liposomes Liposomes 37

Lipid nanotubes Biosurfactants Microbial glycolipids 38

39

Introduction 40

Topological connections between closed lipidic compartments through nanotubes1ndash3 41

have been shown to play a crucial role in the transfer of matter and communication in 42

neighboring mammalian cells4 These singular nanosystems observed since the lsquo90s as 43

spontaneous non-equilibrium structures in electroformed model liposome membranes56 have 44

since been largely studied both experimentally and theoretically7ndash9 Addressed in the literature 45

by different terms tunnelling nanotubes (TNT)149 block liposomes10ndash12 or nanotube-vesicle 46

networks13ndash16 (instead of tubes some work speak of tethers617) all refer to a similar 47

phenomenon driven by various internal or external forces The latter must overcome the energy 48

barrier needed to bend a phospholipid bilayer from low positive mean and gaussian curvatures 49

(vesicle) to a high mean and zero Gaussian (tube) curvatures 50

3

A large body of both experimental and theoretical work has shown that budding and 51

eventual nanotube formation from an existing membrane can only occur spontaneously for 52

membranes of at least two-components9102425111218ndash23 and below a vesicle-nanotube phase 53

transition event2627 Otherwise external anisotropic driving forces like electroformation5ndash54

82829 osmotic pressure6 laser ldquotweezersrdquo30 or electrodynamics13ndash16 must be employed to obtain 55

similar results with in some cases an impressive degree of 2D and 3D organization13ndash16 The 56

origin of spontaneous nanotube formation has been shown to be related to a nanoscale phase 57

separation between two membrane components stabilizing low and high curvatures 58

respectively10ndash1231 with at least one component being highly anisotropic203233 Theoretically 59

this behavior has been explained by deviations in the elastic properties of membranes due to in-60

plane orientational ordering of membrane inclusions composed of anisotropic amphiphiles 61

these referring to a non-symmetrical shape upon a 90deg tilt along the amphiphile axis8919242533 62

In this work we show unexpected nanotubing of membranes prepared from a single-63

component lipid in the absence of external directional forces and above the lipidrsquos Tm This 64

phenomenon is observed for a novel anisotropic double amphiphile (bolaform amphiphile or 65

bolaamphiphile) a glucolipid composed of β-D-glucose and a C181-cis fatty alcohol (G-66

C181-OH Figure 1) This compound is obtained by microbial fermentation of a genetically-67

modified S bombicola yeast in the presence of oleyl alcohol34 and is developed in the broader 68

context of extending the library of new biobased surfactants and lipids in view of replacing 69

petrochemical low molecular weight amphiphiles35ndash41 The structure of G-C181-OH is 70

analogous to that of other microbial glycolipids developed through genetic engineering4243 71

Topological connections between nanotubes and vesicles are observed by means of 72

cryogenic transmission electron microscopy (cryo-TEM) fluorescence microscopy and wide 73

angle X-ray scattering (WAXS) above the melting temperatures Tm= 483degC under strong 74

dilution (05 wt) Spin diffusion and 1H double-quantum single-quantum (DQ-SQ) 2D solid-75

state nuclear magnetic resonance (ssNMR) spectroscopy under magic angle spinning (MAS) 76

necessarily performed on the dry glucolipid powder provides a spectroscopic ground to 77

understand the vesicle-nanotube coexistence The bolaform glucolipids could be in a head-78

headtail-tail configuration in the vesicles but rather in a head-tail configuration in the 79

nanotubes In the dry glucolipid both configurations are energetically comparable in the single-80

phase region above Tm and their coexistence is evidenced by the splitting of some 13C NMR 81

signals These facts could explain the stability of nanotubes while the following hypotheses are 82

formulated for their formation membrane inclusions with different orientational ordering13ndash16 83

possibly driven by inter-vesicle collisions 84

4

85

86

87

88

89

Figure 1 ndash Non acetylated C181 alcohol glucoside G-C181-OH is obtained by a bioprocess performed 90

with modified S bombicola yeast 91

92

Experimental Section 93

Synthesis of non acetylated C181 Alcohol Glucosides (G-C181-OH) G-C181-OH (Mw= 94

41856 gmol-1) was produced by aerobic whole cell bioprocess with a modified S bombicola 95

strain as described by Van Renterghem et al (Fig S4 in Ref 34) The molecule was purchased 96

from the Bio Base Europe Pilot Plant (Gent Belgium) and has the generalized chemical 97

structure given in figure Figure 1 The HPLC and 1H NMR spectrum (MeOD-d4) with peak 98

assignment are shown in Figure S 1 High purity levels (99) and high degree of uniformity 99

were obtained as can be derived from HPLC-ELSD chromatogram 1H NMR and table of 100

contaminant given in Figure S 1 101

102

1H solution Nuclear Magnetic Resonance (NMR) 1H solution NMR experiments were 103

performed on a Bruker Avance III 300 spectrometer using a 5 mm 1H-X BBFO probe using 104

methanol-d4 as solvent The number of transients is 8 with 3 s recycling delay an acquisition 105

time of 546 s and a receiver gain of 362 The 1H NMR spectrum and relative assignment are 106

shown in Figure S 1 and Table S 1 13C solution NMR were performed on the same probe using 107

DMSO-d6 Corresponding chemical shifts are given in Table 1 Referencing is done with 108

respect to TMS δ1H= 0 ppm δ13C= 0 ppm 109

110

Sample preparation The sample was dissolved in milliQ-grade water at the concentration of 5 111

mgmL (05 wt) Lack of pH-sensitive probes (eg COOH groups) in G-C181-OH as 112

otherwise found in other microbial amphiphiles44 but also the will to avoid ion-specific 113

effects45 exclude the use of buffer The solution was heat at the given temperature using a CH3-114

150 Combitherm-2 dry block heating device For the cryo-TEM experiments the vials were 115

heat at the given temperature next to the cryofixation device and plunged into liquid ethane as 116

fast as possible For temperatures of 25degC and 70degC the sample was heat at 90degC and then 117

Synthesis path

5

cooled at to the desired temperature For the SAXS experiment the sample was heat at 90degC in 118

the block heating device and cooled at to 25degC before analysis For the SANS experiment the 119

sample was heat at 90degC directly in front of the beam using the thermalized sample holder 120

available 121

122

Differential Scanning Calorimetry (DSC) DSC was performed using a DSC Q20 apparatus 123

from TA Instruments equipped with the Advantage for Q Series Version acquisition software 124

(v540) Acquisition was performed on both hydrated and dry powder sample (~ 3-5 mg) sealed 125

in a classical aluminium cup and using an immediate sequence of heating and cooling ramps at 126

a rate of 10degCmin-1 Melting temperatures Tm 1 and 2 Tm1 and Tm2 were taken at the minimum 127

of the endothermic peak 128

129

Small and Wide Angle Scattering experiments Small angle neutron scattering (SANS) 130

experiments were performed at the D11 beamline of Institut Laue Langevin (Grenoble France) 131

during the run No 9-13-778 Four q-ranges have been explored and merged using the following 132

wavelengths λ and sample-to-detector (StD) distances 1) ultra-low q λ= 135Aring StD= 39 m 133

2) low-q λ= 53Aring StD= 39 m 3) mid-q λ= 53Aring StD= 8 m 4) high-q λ= 53Aring StD= 14 m 134

The sample (C= 5 mgmL-1) was prepared in 999 D2O to limit the incoherent background 135

scattering The sample solution was analyzed in standard 1 mm quartz cells Direct beam empty 136

quartz cell H2O (incoherent scatterer) within the quartz cell were recorded and boron carbide 137

(B4C) was used as neutron absorber The sample acquisition was measured at 90degC where 138

temperature was controlled through the controller thermalized sample holder available at the 139

beamline The background sample (D2O) signal was subtracted from the experimental data 140

Absolute values of the scattering intensity were obtained from the direct determination of the 141

number of neutrons in the incident beam and the detector cell solid angle The 2D raw data were 142

corrected for the ambient background and empty cell scattering and normalized to yield an 143

absolute scale (cross section per unit volume) by the neutron flux on the samples The data were 144

then circularly averaged to yield the 1D intensity distribution I(q) The software package Grasp 145

(developed at ILL and available free of charge) was used to integrate the data while the 146

software package SAXSUtilities (developed at ESRF and available free of charge) was used to 147

merge the data acquired at all configurations and subtract the background 148

Wide-angle X-ray scattering (WAXS) was performed under temperature control at the 149

SWING beamline of SOLEIL synchrotron facility (Saint-Aubin France) during the run 150

20201747 (energy 14 keV sample-to-detector distance 05 m) Two sample concentrations in 151

6

H2O were probed C= 5 mgmL-1 and C= 25 mgmL-1 The 2D data were integrated azimuthally 152

at the beamline using the software Foxtrot and in order to obtain the I(q) vs q spectrum after 153

masking the beam stop shadow Silver behenate (d(100) = 5838 Aring) was used as a standard to 154

calibrate the q-scale Sample solutions were inserted in borosilicate capillaries of 15 mm in 155

diameter Capillaries were flame-sealed A capillary oven with controlled temperature (plusmn 05degC) 156

provided at the beamline was used to control the sample temperature between 25degC and 111degC 157

Data were normalized by the transmission and calibrated to the SAXS signal of H2O at large q-158

values (I= 00163 cm-1) in order to obtain an absolute intensity scale The water signal was 159

measured by subtracting the signal of the empty capillary from the signal of a water-filled 160

capillary The signal of (water + capillary) was used as background for the samples and it was 161

subtracted after integration of the 2D data 162

163

Cryogenic transmission electron microscopy (cryo-TEM) Cryo-TEM experiments were carried 164

out on an FEI Tecnai 120 twin microscope operating at 120 kV and equipped with a Gatan 165

Orius CCD numeric camera The sample holder was a Gatan Cryoholder (Gatan 626DH 166

Gatan) Digital Micrograph software was used for image acquisition Cryofixation was 167

performed using a homemade cryofixation device The solutions were deposited on a glow-168

discharged holey carbon coated TEM copper grid (Quantifoil R22 Germany) Excess solution 169

was removed and the grid was immediately plunged into liquid ethane at -180degC before 170

transferring them into liquid nitrogen All grids were kept at liquid nitrogen temperature 171

throughout all experimentation Cryo-TEM images have been treated and analyzed using Fiji 172

(is just ImageJ) software available free of charge at the developerrsquos website46 173

The samplersquos concentration was C= 5 mgmL-1 For technical reasons the temperature 174

during cryofixation could not be controlled For this reason the temperature associated with 175

each cryo-TEM image in the text was related to the equilibration temperature just before 176

cryofixation the device being as close as possible to the dry bath used for temperature control 177

Since temperature control during cryofixation was not possible in our device and practically 178

impossible at high temperatures one should take the T values as indicative of the given value 179

Although the sample transfer between the dry bath and the cryofixation device was performed 180

as fast as possible it is more than likely that the actual temperature of the sample at the moment 181

of cryofixation was lower than the one indicated here although impossible to know precisely 182

183

7

Fluorescence microscopy Images were recorded using a 40x objective on a Leica SP5 upright 184

confocal microscope with 405 nm laser excitation The sample was marked with 6-dodecanoyl-185

NN-dimethyl-2-naphthylamine (Laurdan Mw= 35354 gmol-1) fluorophore as follows A 186

stock solution of 3 mg of Laurdan in 100 microL of acetone was initially prepared (C= 30 mgmL-187

1) A volume of 028 microL of the stock Laurdan solution was diluted in 1 mL of a 5 mgmL-1 G-188

C181-OH water solution temporary set at 110degC in a sealed Eppendorf The latter was 189

important so to enhance inclusion of Laurdan in the membrane constituted by the glucolipid 190

while in its vesicle phase The final Laurdan concentration was approximately 24 micromol with a 191

molar Laurdan-to-G-C181-OH molar ratio of approximately 1-to-500 Fluorophore-to-lipid 192

ratios above 1200 were generally considered as non-interfering with the lipid phase behavior 193

as also verified in this work 194

Laurdan can be excited between 365 nm and 410 nm and its emission wavelength can 195

vary according to the rigidity of the membrane intercalation within a rigid or flexible 196

membrane generates emissions at 440 nm or 490 nm respectively47 In the present experiment 197

we have flame-sealed a solution of G-C181-OH at 5 mgmL-1 containing Laurdan (1-to-500 in 198

lipid-to-Laurdan ratio) in flat optical capillaries of 01 mm thickness Capillaries were set on a 199

Linkam support for temperature control eventually placed under the microscope The sample 200

was excited at 405 nm and detection was made in spectral mode with a 5 nm window sliding 201

every 5 nm from 412 nm to 602 nm in 16 bits which allows quantify the signal and compare 202

the different conditions Quantitative analysis was performed according to the literature4849 203

several ROI (range of interest) have been drawn on several images using the ROI manager of 204

Fiji (is just ImageJ) software The integrated intensity of each ROI at 440 nm and 490 nm was 205

eventually recorded and used to calculate the value of the generalized polarization GP as 206

follows of 207

119866119875 =(120582440 minus 120582490)

120582440 + 120582490 208

A negative GP indicates a flexible membrane while a positive GP indicates a rigid membrane 209

210

Solid-state Nuclear Magnetic Resonance (ssNMR) 1H and 13C ssNMR experiments have been 211

performed using an Avance III HD Bruker 705 T (ν1H= 300 MHz) spectrometer and a 4 mm 212

magic angle spinning (MAS) probe except for the BAck-to-BAck (BABA) experiment which 213

was performed on an Avance III Bruker 1644 T (ν1H= 700 MHz) spectrometer equipped with 214

25 mm MAS probe Temperature was controlled through a Smart cooler BCUIIa BCU-xtreme 215

unit using 10 min of equilibration time after each temperature variation Temperature was 216

8

always varied from high (110degC) to low (70degC) before analysis Due to combined use of fast 217

MAS rates necessary to achieve high resolution spectra and heating close to the boiling point 218

of water neither diluted nor concentrated samples could be used for ssNMR experiments All 219

tests employing up to 50 wt G-C181-OH in water resulted in sample centrifugation drying 220

but also expulsion from the rotor causing sudden stop of rotor spinning For these reasons all 221

attempts to study hydrated G-C181-OH failed and only the dry powder could be analyzed The 222

dry sample powder was analyzed as such in zirconia rotors (4 mm or 25 mm) spinning at MAS= 223

10 kHz (4 mm) or 22 kHz (25 mm) and Chemical shifts were calibrated with respect to 224

adamantane (13C 13C(CH2)= 3848 ppm = 191 ppm) and glycine alpha (13C(CO)= 17603 225

ppm) Details on the acquisition parameters are given in the Supporting Information 226

227

Results and discussion 228

Topologically-connected nanotubes and vesicles 229

G-C181-OH is a biobased glucolipid water-insoluble at room temperature and of 230

which the physicochemical properties are poorly known The DSC thermogram performed at 231

10degCmin of G-C181-OH as a function of the water content (Figure 2a) shows two endothermic 232

peaks on the dry powder (996 wt glucolipid) They are attributed to two melting phenomena 233

occurring at the following temperatures Tm 483degC (Tm1) and 957degC (Tm2) the enthalpy of the 234

former being lower than that of the latter Interestingly DSC also shows that Tm1 is hydration 235

independent and reversible (the associated exothermic peak is systematically observed) while 236

the transition at Tm2 is simply suppressed (here tested up to 130degC) on hydrated powders and 237

never observed on the cooling profile even at slow cooling rates (1degCmin) The thermogram 238

recorded on the dry powder is typical of single-component lipids having an intermediate50 239

metastable phase (often referred to the ripple phase Pβ´ in reference to the periodic undulation 240

of the bilayer)50ndash54 between the classical lamellar gel (Lβ) and liquid crystalline (Lα) phases50ndash241

54 and characterized by regions of both liquid order and disorder5455 The effect of water does 242

not influence the low-temperature crystallization at Tm1 but it contributes to disrupt the ordered 243

array of hydrocarbon chains in the metastable phase 244

On the basis of the DSC thermogram the aqueous phase behavior of G-C181-OH under 245

diluted conditions (5 mgmL-1) was studied at four different temperatures (red marks on Figure 246

2a) below and above Tm1 but also between Tm1 and Tm2 corresponding to the thermogram of 247

the dry powder 248

9

249

Figure 2 ndash a) DSC thermogram of G-C181-OH under dry (04 wt water estimated by thermogravimetric 250

analysis) and hydrated conditions recorded at a heating rate of 10degCmin Red marks indicate the 251

temperatures at which cryo-TEM experiments are performed on the corresponding 5 mgmL aqueous 252

solutions b) Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 130degC 253

90degC 70degC and 25degC T indicates the temperature of the sample solution before cryofixation Tne T T 254

being the temperature of the sample upon plunging in liquid ethane Please refer to the materials and 255

methods section for more information 256

T= 130degC

T= 90degC

T= 70degC

T= 25degC

a)

b)

20 40 60 80 100 120

99 wt

50 wt

25 wt

T degC

En

do

10 wt

Tm1= 483 degC

Tm2= 957 degC

10

257

The series of cryo-TEM images in Figure 2b show an overview of the most relevant 258

structures observed on G-C181-OH solution from T= 130degC to T= 25degC whereas T is 259

indicative of the equilibrium temperature before cryofixation (refer to materials and methods 260

section for more information on the sense and interpretation of T) At Tgt 100degC the sample 261

is essentially composed of polydisperse single wall vesicles massively shown by the 262

complementary pictures presented in Figure S 2 In between Tm1 and Tm2 at T= 90degC and T= 263

70degC (Figure 2b Figure 3 Figure 4 complemented by Figure S 3 and Figure S 4) vesicles are 264

connected by nanotubes At Tlt Tm1 flat crystals are otherwise observed (Figure 2b and Figure 265

S 5) 266

267

268

Figure 3 ndash Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 90degC 269

270

A deeper insight at T= 90degC is given in Figure 3 (more images are provided in Figure 271

S 3) Arrows 1 through 6 mark very clearly the typical continuity in the glucolipid membrane 272

between the vesicle and nanotubes Arrow Ndeg 7 shows that a single nanotube connects the outer 273

membrane of two vesicles several hundred nanometers apart The connection is shown by 274

2

34

5

6

6

1

5

7

A

B

D

C

0 10 20 30 40 50 60

BIn

tensity

au

Profile nm

A

0 10 20 30 40 50 60

Inte

nsity

au

Profile nm

C

D

T= 90degC

11

arrows Ndeg5 and Ndeg6 The corresponding profiles A through D confirm the tubular nature of 275

the elongated objects Profiles B and D are both characterized by typical symmetrical thick 276

walls (low intensity depths highlighted by shaded rectangles) surrounding a hollow region (high 277

intensity peak between the shaded rectangles) The wall thickness varies between 35 and 40 278

nm for profile B and between 41 and 61 nm for profile D while the total nanotube diameter 279

varies between 14 nm at B and 20 nm at D At the same time the thickness of the vesicle wall 280

is about 6 nm at A and 135 nm at C Considerations about the bilayer or interdigitated structure 281

of the membrane will be presented below 282

The connection between nanotubes and vesicles seems to be multiple For instance 283

arrow Ndeg1 (Figure S 3) and Ndeg4 (Figure 3) show a single-wall vesicle of which the membrane 284

is extruded into a nanotube Similarly a closer look at arrows Ndeg1 2 and 3 in Figure 3 show 285

that the vesicle is single-wall and fully closed while the nanotube nucleates at the vesicle outer 286

surface with no apparent topological continuity In many regions of the sample at T= 90degC 287

one can find exvaginations or buds232531 in the nanotube wall This is shown for instance by 288

arrows Ndeg7 and particularly put in evidence in Figure 4 by arrows Ndeg1 through Ndeg4 where the 289

amplitude of the oscillation varies between 20-25 nm (arrow Ndeg1) and 6-8 nm (arrows Ndeg3-4) 290

291

Figure 4 - Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heated at T= 90degC 292

293

Still between Tm2 and Tm1 but at lower temperature (T= 70degC) the nanotubes seem to 294

vary from single threads (arrow Ndeg1 Figure S 4) to nanotubes of diameter below 15 nm both 295

connecting nanotubes of diameter ranging between 50 nm and 100 nm (arrows Ndeg2 Figure S 296

4) The larger tubes are most likely formed by the inflation and flattening of the nanotubes 297

probably after assimilation of surrounding vesicles and fusion of the exvagination discussed 298

above The transition between vesicles tubes and flat membranes is certainly more accentuated 299

Arrow Ndeg3 in Figure S 4 shows distinct vesicles while arrow Ndeg4 identifies a flat structure 300

evidently formed from fused vesicles In particular arrow Ndeg5 points at the fusion of two 301

2

1 132

4

12

vesicles while arrow Ndeg6 points at two vesicles fusing into the flat membrane Additional areas 302

where vesicle fusion is prominent are indicated by the oscillations at the outer skirt of broad 303

lamellae and pointed by arrow Ndeg7 304

Finally below Tm1 at T= 25degC for instance mainly flat crystals with no visible 305

exvaginations or tubes are visible (Figure S 5) by cryo-TEM and the material turns into an 306

insoluble precipitate Massive presence of flat crystalline structures is also clear from the bulk 307

iridescent behavior of a G-C181-OH solution under manual shear 308

In summary despite the uncertainty related to the actual state of equilibrium of the 309

glucolipids at the moment of cryofixation cryo-TEM depicts a vesicle phase at Tgt 100degC a 310

flat lamellar phase below Tm1 and an interconnected vesicle-nanotube network between Tm1 and 311

about 100degC The combination of statistically-relevant fluorescence microscopy and 312

scatteringdiffraction experiments provides an additional proof of the coexistence of 313

vesiclenanotube systems between the two transition temperatures Please note that the 314

experiments reported below have been performed under controlled conditions of temperature 315

which is then indicated by the use of T instead of T 316

13

317

Figure 5 ndash a-c) Confocal fluorescence microscopy images recorded at 90degC and 70degC for a 5 mgmL G-318

C181-OH aqueous solution d-g) Typical fluorescence microscopy images recorded at T= 80degC using 319

Laurdan (refer to materials and methods section for more details on this fluorophore) Excitation 320

wavelength is 405 nm for all the images while emission wavelength is 490 nm in (df) and 440 nm in (eg) 321

h) Full emission spectra obtained from the integration of various fluorescence images recorded from 412 322

nm to 600 nm (step 5 nm) at 90degC 80degC and 70degC i) Box plot of GP defined as 119918119927 =(120640120786120786120782minus120640120786120791120782)

120640120786120786120782+120640120786120791120782 with 120640120786120786120782 323

and 120640120786120791120782 being the emission wavelengths at 440 nm and 490 nm at 90degC 80degC and 70degC Each data point 324

l= 490 nm l= 440 nm

l= 490 nm l= 440 nm

10 μm400 450 500 550 600

00

02

04

06

08

10

No

rma

lize

d I

nte

nsity

l nm

90degC

80degC

80degC

70degC

70degC

400 450 500 550 60000

02

04

06

08

10

No

rma

lize

d I

nte

nsity

l nm

90degC

80degC

80degC

70degC

70degC

d) e)

f) g)

h)

i)

T= 80degC

Laurdan

T= 80degC

T= 80degC T= 80degC

T= 90degC T= 70degC T= 70degC

a) b) c)

1

1

1

14

corresponds to a set of regions of interest (ROIs) (54 at 90degC 65 at 80degC and 75 at 70degC) selected in the 325

corresponding fluorescence images ROIs were representatively selected so to include the entire image but 326

also vesicles alone tubes alone and mixtures of vesicles and tubes 327

328

Figure 5ab shows two typical confocal fluorescence microscopy images recorded on a 329

G-C181-OH solution (5 mgmL) in a flame-sealed capillary heat at T= 90degC and T= 70degC 330

Although the resolution of optical microscopy is not comparable with cryo-TEM the sample is 331

broadly composed of vesicles coexisting with nanotubes whereas in some cases the nanotube 332

starts from the vesicle itself (arrow Ndeg1) in good agreement with cryo-TEM data Furthermore 333

SANS recorded at T= 90degC (Figure S 6a) shows a scattering profile with a pronounced -2 334

dependence of the scattering intensity against the wavevector q in the log-log scale This 335

indicates the massive presence of a flat interface and it is expected typical of bilayers in 336

vesicular morphologies56 However the typical signature of nanotubes cannot be observed in 337

the SANS profile and this can be explained by both an argument related to the relative lower 338

amount of nanotubes with respect to vesicles at 90degC but also to the steeper slope of flat 339

morphologies (-2) compared to tubes (-1) the signal of the former masking the tube signal at 340

low wavevector values The SANS profile is characterized by a diffraction peak at q= 0167 Aring-341

1 of which the assignment is refined by WAXS data 342

Temperature-driven WAXS experiments in Figure S 6bc (5 mgmL-1 and 25 mgmL-1 343

respectively) show a lamellar order possibly attributed to a multilamellar structure of the 344

vesicles (dv Figure S 6d) at high temperature and to the interplanar arrangement of the 345

crystalline flat structures (dL Figure S 6d) at T= 25degC The sharp (001) reflection indicates 346

typical interlamellar distance of 353 Aring at 25degC although this value varies with temperature 347

(400 Aring at 111degC 372 Aring at 88degC) thus confirming its assignment to a hydration-dependent 348

interlamellar order More intriguing is the second broader peak at q= 0403 Aring-1 (d= 156 Aring) 349

and its reflection both clearly visible at 88degC below Tm2 We attribute it to the inter-glucolipid 350

arrangement inside the tubular phase (dt Figure S 6d) coexisting with the vesicle phase 351

between Tm1 and Tm2 and to the planar lipid arrangement (di Figure S 6d) in the rigid flat 352

lamellar structure below Tm2 Its temperature- hence hydration- independent position also 353

confirms its structural origin Finally besides the water-structure peak at about 2 Aring-1 we also 354

observe a broad peak at 144 Aring-1 characteristics of the lateral packing (d Figure S 6d) of lipids 355

in bilayers While its position could be consistent with an Lα phase5758 above Tm2 its broadness 356

and positional invariance as well as the coexistence between vesicles and tubes do not allow a 357

precise assignment 358

15

Does G-C181-OH assume a bilayer or interdigitated structure the latter being the one 359

found for other similar bolaform glucolipids4243 The size of G-C181-OH can be estimated to 360

be at the most 32 Aring given by the length L of the C16 tail (L = 22 Aring from the Tanford formula 361

L= 154 + 1265 times n (n= 16)59) and the sugar (about 10 Aring)4243 However the cis conformation 362

of the C=C bond imposes a ldquokinkrdquo of 60deg in the aliphatic chain which reduces the effective 363

length of the oleic backbone to about 17 Aring for an effective total G-C181-OH length of about 364

27 Aring Cryo-TEM provides a tube thickness varying from 35 Aring to 60 Aring and a vesicle membrane 365

thickness of at least 60 Aring WAXS suggests an interlamellar thickness (membrane and water) 366

between 40 Aring and 37 Aring and an inter-lipid distance in the tube of about 16 Aring In the absence of 367

a clear form factor oscillation profile as found for similar glucolipids4243 one must 368

acknowledge that these data pledge neither in favor of fully interdigitated nor a double layer 369

structure and might suggest a tilted lipid arrangement with partial interdigitation60 370

371

13C ssNMR demonstrates that G-C181-OH can assume two coexisting head-head and 372

head-tail configurations 373

Vesicles are characterized by low positive mean and gaussian curvatures while 374

nanotubes are described by a combination of a high positive mean and zero gaussian curvatures 375

In this regard the coexistence of vesicles and nanotubes requires glucolipids to adapt 376

themselves to both environments at the same time For this reason the formation of nanotube-377

vesicle networks generally requires a specific driving force able to impose a high mean and 378

zero gaussian curvature to the membrane In many systems insertion of at least a second 379

constituent (lipid surfactant protein) in the membrane becomes a necessary internal driving 380

force for the nucleation and growth of so-called membrane inclusions from which 381

topologically-connected nanotubes can form10ndash123261 as explained by orientational ordering 382

arguments62 It has also been shown that external driving forces such as use of laser 383

ldquotweezersrdquo30 osmotic pressure6 electroformation5ndash82829 or controlled suction13ndash16 are 384

otherwise necessary as also supported by theoretical studies19 385

The bending energy of tubes is higher than that of a vesicle membrane8 due to the 386

dependence of the membrane free energy from the principal curvatures For this reason 387

isotropic (rotational symmetry along the main axis) but also anisotropic33 lipids generally 388

stabilize low-curvature shapes like flat membranes or vesicles Lipid nanotubes are also a 389

common phase formed by amphiphiles but generally for lipids with a certain complexity in 390

their molecular structures as exampled by diacetylenic derivatives of phospholipids among 391

others262763ndash65 Negative contributions to the bending energy favoring the simultaneous 392

16

presence of interconnected vesicles and tubes was shown to be driven by the presence of 393

anisotropic lipids in a membrane continuum Spontaneous deviations to the elastic theory in the 394

presence of surfactants lipids and proteins with anisotropic structures favor the formation of 395

protrusions with radius of the order of the fraction of a micron9ndash1219ndash2362 396

In the absence of both internal and external driving forces mentioned above G-C181-397

OH should behave as a standard lipid Its DSC profile (Figure 2)536667 under both hydrated and 398

dry conditions identifies a hydration-independent transition (Tm1) slightly above room 399

temperature this is followed by one main transition at higher temperature (Tm2) in the dry 400

sample50 For such a profile one expects a gel phase (Lβ) below Tm1 and a liquid crystalline 401

structure possibly in the morphology of vesicles (L4 phase) possibly above Tm1 and most likely 402

above Tm231 as experimentally found On the basis of the thermogram recorded on the dry 403

powder one may also expect a ldquoripplerdquo (Pβ) phase between Tm1 and Tm2 characterized by 404

periodic undulations of the bilayer (generally addressed to as regions with both liquid order and 405

disorder)5455 406

In the present work instead of a Lβ phase we observe flat crystals below Tm1 and a 407

vesicle-nanotube network instead of the ldquoripplerdquo phase The latter is particularly unexpected 408

because the membrane is composed of a single-component (G-C181-OH) and no anisotropic 409

external force (osmotic hydrodynamic electric) is used to pull and stabilize the tubes In 410

addition the nanotubes are observed at fairly high temperatures above Tm1 (between about 411

60degC and 90degC) while this morphology is generally favored at low temperature (proportionality 412

with 1kT figure 5 in Ref 62) commonly below a given transition temperature characteristics 413

for each specific amphiphile27 Furthermore nanotubes of G-C181-OH seem to be stable in 414

time thus differing from what is found in electroformed vesicles78 They are also observed 415

away from phase transition events (may them be at Tm1 or Tm2) as otherwise reported before 416

for a diacetylenic derivative of phosphatidylcholine26 and many other systems27 417

A stable nanotube-vesicle network for a single-component phospholipid in the absence 418

of external driving forces is not expected and has never been reported before It could be 419

explained by the coexistence of two micro-separated phases of different rigidity This 420

hypothesis is excluded by confocal microscopy on Laurdan-labelled membranes and magic 421

angle spinning (MAS) solid-state NMR (ssNMR) experiments 422

Laurdan is a well-known fluorophore used to put in evidence the coexistence of rigid 423

and fluid domains in both model and biological membranes47ndash49 Laurdan was used here to label 424

the G-C181-OH membranes between Tm1 and Tm2 and to verify the possible micro-segregation 425

and coexistence of rigid (eg Lβ λem= 440 nm) and fluid (eg Lα λem= 490 nm) domains in 426

17

tubes and vesicles respectively A set of quantitative experiments performed on a 5 mgmL-1 427

solution summarized in Figure 5 shows an excess of fluorescence at only λem= 490 nm for 428

temperature between 90degC and 70degC (Figure 5h) The GP value which quantifies the rigidfluid 429

character of the membrane is systematically negative (Figure 5i) indicating an overall fluid 430

membrane at all temperatures GP is quantified on more than 50 regions of interest (ROIs) at 431

each temperature including vesicles and tubes alone as well as connected nanotube-vesicles 432

(Figure 5d-g) The dispersion of GP at 80degC and 70degC between -08 and -02 could suggest the 433

spurious presence of membranes with various degrees of flexibility although fluidity is always 434

predominant The combination of WAXS (broad peak d= 144 Aring-1 invariant with 435

temperature Figure S 6bc) and confocal fluorescence microscopy using Laurdan confirms the 436

single-phase of the membrane composed of G-C181-OH both in vesicles and nanotubes This 437

is also consistent with the lack of phase transitions in DSC experiments between ~85degC and 438

~60degC especially in hydrated samples We conclude that the vesicle-nanotube transition is not 439

driven by the local segregation of gel phase (Lβ) microdomains nor that the nanotubes are 440

composed of such phase 441

ssNMR is the best-suited approach to investigate the origin of the local driving force for 442

the nucleation of anisotropic inclusions of which the average orientation depends on the local 443

membrane and inclusion curvatures and was shown to promote tubing62 ssNMR can provide 444

information on the local molecular mobility eg through the nuclear spin relaxation behavior 445

but also on intermolecular interactions and stereochemistry eg through variations in the 446

chemical shift68ndash70 447

Ideally ssNMR should be performed on diluted G-C181-OH samples above Tm1 448

However fast MAS necessary for high resolution centrifuges the sample As a first 449

approximation hydrated G-C181-OH samples for instance at 50 wt as classically done for 450

phospholipids71 could be studied Unfortunately permeability of the sample holder combined 451

to heating (close to the boiling point of water) and MAS provoke dehydration and even sample 452

ejection After experiencing both of the above this approach was abandoned As a second 453

approximation the study of the dehydrated G-C181-OH powder avoids centrifugation and 454

evaporation If studying dried lipids may not reflect their exact behaviour under diluted 455

conditions valuable structural information can still be obtained in dehydrated molecular 456

systems like proteins thus illustrating that this approach is not unreasonable Furthermore if 457

DSC (Figure 2a) shows that hydration suppresses the Tm2 it also shows that the value of Tm1 is 458

not affected by water at concentrations as low as 10 wt For this reason performing a 459

structural analysis of dehydrated G-C181-OH by ssNMR above Tm1 may help understanding 460

18

the local arrangement of G-C181-OH in two morphologies vesicles and nanotubes having 461

sensibly different curvatures 462

To explore the local mobility of the most relevant chemical groups of G-C181-OH 463

above Tm2 (T= 110degC) and between Tm2 and Tm1 (70 lt T degC lt 90) we combine a series of 1H 464

and 13C experiments Single-pulse (SP) MAS experiments (1H and 13C) when recorded under 465

quantitative conditions allow the detection of all chemical species while cross-polarization 466

(CP) MAS (13C only) experiments are sensitive only to constrained groups for which the 1H-467

13C dipolar coupling is not averaged to zero by the molecular motions 468

469

Figure 6 ndash a) 1H MAS (10 kHz) ssNMR spectra recorded from 110degC to 70degC b) 13C MAS (10 kHz) ssNMR 470

spectra recorded using high power decoupling single pulse (SP) and cross-polarization (CP) pulse sequences 471

Contact time for CP experiments and corresponding temperatures are given on each spectrum Full spectra 472

are given in Figure S 7 473

The most relevant 1H and 13C spectra are given in Figure 6 (full range in Figure 7 and 474

Figure S 7) As expected the 1H solid-state SPMAS spectra show broader peaks compared to 475

the solution-state NMR spectrum (Figure 6a) reflecting the obvious solid-state character of the 476

sample at any temperature The sugar group and the C=C bond resonate in the 25-6 ppm range 477

and 535 ppm respectively while the OH groups of the sugar moiety are found below 45 ppm 478

(Figure 6a full assignment in Table S 1) Very broad (~1 ppm) at 100degC the position of the 479

OH peak varies with the temperature value and becomes narrower (lt~05 ppm) and visible at 480

δH= 464 ppm at T= 70degC Line narrowing and chemical shift variations of the OH resonance 481

in ssNMR is always a direct sign of H-bonding interactions whereas the strength of the H-482

bonding hence its OHhellipH distance is correlated to a dispersion in chemical shifts72ndash75 All other 483

60 55 50 45 40 35 30 25

ppm

110

90

80

70

910

1rsquo

6rsquo16

1

OH

134 132 130 128 126

ppm110 108 106 104 102 100

ppm

SP (T= 110degC)

SP (T= 70degC)

CP 50 ms (T= 70degC)

CP 3 ms (T= 70degC)

910 1rsquo

1H 13C

1

811

910

CH2 CH2

1rsquo

6rsquo

a) b)

C102

13024

C9101

13003

C92

12869 C1rsquo210382

C1rsquo110443

C1rsquo13035

2rsquo3rsquo4rsquo

T degC

5rsquo

19

resonances broaden from 110degC to 70degC thus indicating slower andor more anisotropic 484

reorientational motions of all chemical groups The specific dynamic behaviour of the sugar 485

headgroup is also reflected by its 1H transverse relaxation time T2 whereas the shorter the T2 486

the higher the rigidity Experiments performed between 110degC and 70degC shown and 487

commented in Figure S 8 in the Supporting Information globally show a single short (lt 1 ms 488

below 100degC) T2 relaxation component for H1rsquo H5rsquo and H6rsquo and two both short and long 489

(between 4 ms and 12 ms) T2 for the acyl chain (CH2 H910 H811) even below 100degC 490

At 110degC 13C SPMAS experiments (Figure 6b full range in Figure 7 and Figure S 7 491

assignments in Table 1) show a single peak at δC= 10335 ppm for C1rsquo and at δC= 13003 ppm 492

for C910 as expected for a symmetrical conformation with two chemically-equivalent 493

carbons76 At 70degC the C1rsquo peak disappears while C910 now shows two peaks around 128 494

ppm and 130 ppm (Table 1 full temperature series in Figure S 7b) attributed to C92 (δC= 12869 495

ppm) and C10 in a non-symmetric configuration respectively for which these two carbons are 496

chemically inequivalent76 More precisely the resonance at 130 ppm is actually the 497

superposition of two close resonances C9101 (δC= 13003 ppm) and C102 (δC= 13024 ppm) 498

where subscripts 1 and 2 stand for two different conformations of the cis-olefin group as 499

reported for various phases of oleic acid in the vicinity of 0degC76 500

501

Table 1 ndash 13C Chemical shifts of G-C181-OH in solution and solid-state Solid-state experiments are 502

recorded at MAS= 10 kHz All spectra are reported in Figure 7 and Figure S 7 The contact time for cross 503

polarization is 3 ms In all ssNMR experiments temperature is first raised at 110degC and then lowered to 504

either 90degC or 70degC Attribution of the 13C resonances in Table 1 is performed on the basis of the solution 505

spectrum (Figure S 1b) Additional peaks observed in both single pulse and cross polarization ssNMR 506

experiments are attributed when possible to a configuration effect of the given atom eg C9101 and C92 507

indicate two configurations of C9 All new peaks observed in solid-state and not detected in solution are 508

highlighted in red for convenience 509

510

511

δ ppm

Peak Liquid (DMSO-d6)

Solid Single Pulse

T= 110degC

Solid Single Pulse

T= 90degC

Solid Single Pulse

T= 70degC

Solid Cross Polarization

T= 70degC

C102

- - 13024 13024

C910C9101 13009 13003 13003 13003 13003

C1C9C10C1rsquo

C6rsquo

20

C92 - - - 12869 12869

C1rsquoC11 10329 10335 10335 - 10443

C12 - - - - 10382

C3 7724 7709 7702 - 7829

- 7634 7629 - 7621

7525

C2 7388 7398 7390 7370 7370

C4 7052 7077 7054 7041 7098

C16 6900 7016 - - 7029

- - - - 6793

C1C5C6 6153 6241 6236 6239 6262

C1C5C6 6117 - - - 5956

C2 3300 3290 3290 3283 3279

CH2(trans) - - - 3198 3196

CH2(gauche)C15 2941 2981 2987 3004 -

C811 2706 2747 2747 2832 2817

CH2 2597 2625 2628 2630 2817

512

13C CPMAS spectra at 70degC (Figure 6b full range in Figure 7 and Figure S 7d) at 513

increasing contact times confirm the presence of two cis-olefin environments one for which 514

C910 are equivalent labelled here C9101 and one for which they are inequivalent labelled 515

C92 C102 Both environments are rigid enough to be observed at short contact time (tc= 50 μs) 516

A similar double environment is also observed for the sugar C1rsquo CPMAS at 70degC shows two 517

peaks (C1rsquo1 at δC= 10443 ppm C1rsquo2 at δC= 10382 ppm Figure 6b Figure S 7d) These are 518

significantly shifted downfield from the C1rsquo peak observed at 110degC and are detected at contact 519

times as short as 50 μs thus indicating their rigid character One also finds that CPMAS is 520

totally inefficient at 100degC for short contact times (50 μs) and poorly efficient even at 3 ms 521

typical of fast-tumbling molecules (Figure S 7c) also both gauche (δC= 3004 ppm) and all-522

trans (δC= 3198 ppm) conformations of the CH2 carbons coexist at 70degC (Figure S 7b)77 523

According to ssNMR one main mobile molecular environment exists at 110degC as 524

shown by the well-resolved 1H spectrum single C=C (C9101) and anomeric (C1rsquo) resonances 525

gauche conformation of the aliphatic CH2 and single T2 of the aliphatic CH2 chain and sugar 526

protons The sugar group is certainly mobile (inefficient 1H13C CPMAS) but still more 527

constrainted than the acyl chain (shorter T2 (1H)) As the temperature is lowered the overall 528

rigidity increases (broadening of 1H NMR peaks faster 1H transverse relaxation and efficient 529

1H13C magnetization transfer during CPMAS) presumably driven by H-bonding 530

interactions (downfield shift and line narrowing of the OH 1H signal very short OH T2 loss of 531

the C1rsquo signal in SPMAS) At the same time a number of spectroscopic data (two concomitant 532

21

resonances for anomeric C1rsquo1 and C1rsquo2 cis-olefin C9101 and C92 C102 and CH2 gauche and 533

all-trans environments) and relaxation behaviour (fast and slow T2 relaxation components) 534

strongly suggests the coexistence of two molecular environments Further proofs of the 535

coexistence of two distinct molecular domains are given by a number of chemical shift 536

variations in the 13C CPMAS spectrum at 70degC compared to the one at 110degC 90degC and 537

solution This is shown in Figure 7 and in the list of 13C chemical shifts in Table 1 538

539

540

Figure 7 - Comparison between solution- and solid-state 13C NMR spectra of G-C181-OH Solid-state 541

experiments are recorded directly on the sample powder at a MAS of 10 kHz (4 mm zirconia rotor) SP 542

Single Pulse CP Cross Polarization In all the ssNMR experiments temperature is first raised at 110degC 543

and then lowered to either 90degC or 70degC Segmented lines are meant to highlight the known peak position 544

in solution thus stressing the strong chemical shift variations in the solid-state 545

546

NMR experiments above Tm2 suggest that vesicles could contain G-C181-OH with a 547

single family of mobile conformations characterized by chemically-equivalent C=C (red 548

Figure 8b) and a single C1rsquo conformation (green Figure 8b) Experiments between Tm1 and Tm2 549

otherwise suggest that two coexisting conformations characterized by chemically-equivalent 550

(red Figure 8b) and inequivalent (magenta Figure 8b) C=C could be associated to nanotubes 551

Two distinct C1rsquo sites (grey Figure 8b) also indicate two configurations of the sugar headgroup 552

A comment on this point will be given in the following paragraphs 553

The local (Aring-scale) variation of molecular mobility below Tm2 drives the coexistence of 554

two distinct molecular environments of the same G-C181-OH molecule This occurs within a 555

single-phase domain (as deduced by WAXS) of equivalent membrane rigidity (as derived from 556

Laurdan probe in confocal microscopy) Although never reported before the difference in the 557

local mobility of the same amphiphile (here G-C181-OH) could satisfy the theoretically-558

140 120 100 80 60 40 20

CP(3 ms)-70degC

SP-70degC

SP-90degC

SP-110degC

ppm

LiquidC910

C1rsquo

C2rsquo

C4rsquo

C3rsquo

C1

C6rsquoC5rsquo

C1

6

C8

11

C2 C

H2

C1

5

CH

2

DM

SO

13C

22

required conditions that induce local lipid segregation and formation of membrane inclusions 559

(Figure 8c) In case of their non-zero average orientation the local membrane curvature 560

increases62 and nanotubes can spontaneously form (Figure 8bc) More intriguing this 561

mechanism occurs at least 10degC above Tm1 although it should not be favored above the Tm27 562

From both experimental and theoretical backgrounds the conditions of topologically-563

connected vesicles-nanotubes systems are met only when either internal (two components) or 564

external driving forces are applied19 In the present single-lipid system of which the phase 565

behavior is summarized in Figure 8a the most plausible driving force could be explained by 566

the NMR results collected on the dry sample G-C181-OH concomitantly occupies two 567

molecular environments characterized by 1) two distinct conformations of the sugar 568

displaying equivalent mobility and 2) two conformations of the acyl chain unsaturation 569

570

23

571

Figure 8 ndash a) Thermotropic phase behavior of G-C181-OH at 5 mgmL-1 Note that Tm2 is observed on the 572

dry powder only b) Possible relationship between the molecular environment as derived from NMR and 573

membrane curvatures All the molecular drawings refer to the same G-C181-OH molecule (bulky ellipsoid 574

equiv sugar headgroup) while each color corresponds to a given configuration Grey of C1rsquo1 and C1rsquo2 indicates 575

the uncertainty about the assignment of the specific NMR peak The curvature C is defined as 1R R being 576

the curvature radius R1 R2 radius of curvature CG Gauss curvature CM Mean curvature c) Free 577

interpretation of the anisotropic high-curvature (CMgt 0 CG= 0) inclusions in the membrane (low CM and 578

CG) continuum driven by inter-sugar H-bonding and reduced intramembrane diffusion for a G-C181-OH 579

glucolipid 580

581

Flat crystalline structure

Nanotube-vesicle network

Vesicles

48degC

120deg

106 105 104 103 102 101

ppm

H9 H10H1rsquo

106 105 104 103 102 101

ppm106 105 104 103 102 101

ppm106 105 104 103 102 101

ppm

C=CC=C

C1rsquo1C1rsquo2

C1rsquo

Tm2(96degC)Tm1

a)

b)

c)

Same molecule different local orientations and sugar conformations

High curvature membrane

Low curvature inclusion

Flip-FlopStrong H-bonding

CG

CM ( + )

CG 0

CM 0 CG 0

CM 0

G-C181-OH

24

Two related questions are unanswered at the moment What actually drives nanotube 582

formation and is the organization of G-C181-OH within the vesicles and tubes the same Lipid 583

nanotubes have been studied for at least three decades over which broad experimental and 584

theoretical insight has been generated Despite this fact the actual understanding is still matter 585

of debate27 Micelle-to-nanotube and vesicle-to-nanotube phase transitions are generally 586

reported for single amphiphiles below a given Tm and intermediate phase composed of twisted 587

and flat fibers is concomitantly reported Initially thought to be a chirality-driven phenomenon 588

recent studies propose a more general mechanism involving crystallization and symmetry 589

breaking in the lipid packing driven by optimal packing27 If twisted ribbons were observed 590

before for a number of microbial glycolipid bolaamphiphiles44 they are not found for G-C181-591

OH This fact as well as the large body of cryo-TEM images and the observation of nanotubes 592

above Tm1 rather pledge in favour of a direct topological change from positive (vesicles) to 593

zero (nanotubes) Gaussian curvatures In this case vesicle-to-nanotube transitions could be 594

driven by external forces56307813ndash162829 595

Considering the absence of external forces in the present system there are only two 596

possibilities left spontaneous budding of the membrane or nanotube pulling after inter-vesicle 597

collisions due to Brownian motion Both are possible as inclusions composed of a different 598

arrangement of the same molecule in the membrane continuum can theoretically explain 599

spontaneous nanotube formation in an isotropic medium672223629ndash1218ndash21 However the second 600

one is probably more likely to prevail because it could explain the large number of nanotubes 601

connecting two distant vesicles Interestingly similar events occur for any phospholipid-based 602

vesicular solution but nanotubes are extremely rare unless at least two lipids of different 603

molecular structure are mixed together The spectroscopic 1H and 13C and relaxation (1H) 604

ssNMR data even if collected on a dry powder demonstrate that the coexistence of two G-605

C181-OH configurations involving both the sugar headgroup and fatty alcohol chain are 606

possible for this molecule As a matter of fact this is not so surprising as more than one energy 607

minima each corresponding to a given conformation are known for glucose its non-reducing 608

derivative methylglucose and carbohydrates in general78ndash80 609

The presence of two configurations could explain the coexistence between two 610

structures of different curvature but the data presented so far do not explain the origin of these 611

two configurations In other words what drives the change in the local structure of the sugar 612

headgroup and fatty alcohol within the membrane 613

The association of known self-assembled structures to spectroscopic 13C ssNMR data is 614

known for years in particular for gluconamides lipids having a linear sugar headgroup81ndash84 It 615

25

was found that gauche effects in some cases up to the γ position in the sugar chain induce 616

upfield chemical shifts while anti and trans conformations result in downfield shifts Each 617

structure (monolayer crystal bilayer crystal or micellar fiber) was then characterized by a given 618

organization of the gluconamide and a characteristic 13C ssNMR fingerprint which could be 619

explained by gauche γ-gauche anti and trans conformations of the headgroup Similar 620

structural-spectroscopic correlations have been collected over the years for much simpler cyclic 621

sugars85 whereas pyranoses can undergo important conformational changes (chair half-chair 622

boat skew) when the appropriate energy barriers are crossed78ndash8086 A quantitative 623

conformational study based on the 13C chemical shift variations shown in Table 1 would be 624

certainly helpful but very delicate in the context of the present state of the art because a 625

rigorous approach requires at least one known crystal structure8182 which is currently missing 626

for this compound A qualitative approach will then be adopted 627

G-C181-OH is a bolaform amphiphile and flip-flop could be a possible mechanism 628

involved in the nanotube stabilization whereas head-tail interactions have long been considered 629

as important in nanotube-forming bolaform amphiphiles6063 From mere considerations based 630

on T2 (1H) relaxation arguments one could exclude head-tail configurations if the R-CH2OH 631

endgroup (H1) was involved in a direct interaction with the sugar group its corresponding T2 632

would probably be comparable to the sugarrsquos one (OH and H1rsquo) but this does not seem to be 633

the case However this hypothesis would be based on the comparison of the relaxation from 634

CH and CH2 protons additionally involved in different kind of motions Besides the relaxation 635

of H1 is influenced by the strong overlap between H1 and H5rsquo For these reasons relaxation 636

arguments are not solid enough and will not be used 637

We then address this question by exploring intermolecular interactions in a direct 638

manner using 2D 1H-1H homonuclear ssNMR correlation spectroscopy At 90degC in the fast-639

tumbling domain of molecular motion (narrow 1H NMR peaks Figure 6a) and corresponding 640

to the vesicle-rich region spin diffusion (NOESY) experiments are preferred over dipolar 641

coupling based experiments because such interaction is averaged out by molecular motion 642

However when the molecular motion is more constrained in the domain of nanotube stability 643

at 70degC (broad 1H spectra Figure 6a) Double-Quantum Single-Quantum (DQ-SQ) correlation 644

experiments are best suited to probe interaction between close molecular neighbors These 645

experiments are reported in Figure 9 and the related interpretation is only briefly reported 646

below while a more detailed discussion can be found on Page 12 in the Supporting Information 647

26

648

Figure 9 ndash a) 2D 1H-1H NOESY NMR experiment performed at MAS= 10 kHz (4 mm probe) on a Bruker 649

300 MHz Avance III HD spectrometer using 300 ms of mixing time Sample temperature is regulated to 650

90degC reached by cooling from 110degC b) 2D 1H-1H DQ-SQ (BAck-to-BAck BABA pulse sequence used for 651

the DQ excitation scheme) NMR experiment performed at MAS= 22 kHz (25 mm probe) on a Bruker 700 652

MHz Avance III spectrometer using a DQ excitation time equal to 2 rotor periods Sample temperature is 653

regulated to 70degC reached by cooling from 110degC The typical schemes illustrating how to interpret these 654

2D experiments are reported above each experimental map Zooms of the sugar region and configurational 655

interpretation are given below each 2D map More information are given in the text below 656

H16-H16

H1-H1

H9-H10

H1rsquo-H1

H1rsquoH910 H16

H1

H1rsquo-H1rsquo

Hxrsquo-Hxrsquo

2

(pp

m)

(ppm)

315

811910 1rsquo

2rsquo2

3rsquo 4rsquo6rsquo

5rsquo

1rsquo2910

asymp asymp asympHDO MeOD Acetone

1

315

811 9101rsquo

2rsquo2

3rsquo4rsquo6rsquo

5rsquo

1rsquo2 9 10

asympasympasympHDOMeODAcetone

H1rsquo-H1

1rsquo

1

(p

pm

)

(ppm)

H1-H2

H1rsquo-H1

H1rsquo

H910

H1H5rsquo

H2

H1rsquo-H5rsquo

315

811910 1rsquo

2rsquo2

3rsquo 4rsquo6rsquo

5rsquo

1rsquo2910

asymp asymp asympHDO MeOD Acetone

1

H2rsquoH3rsquoH4rsquo

H2rsquoH3rsquoH4rsquo

X= 2rsquo3rsquo4rsquo

NOESY - T= 90degC

1H

1H

1H

1H

DQ-SQ - T= 70degC

a) b)

2

(pp

m)

(p

pm

)

(ppm)

A B

AhellipB

AhellipA

BhellipB

NOESY

(ppm)

A B

AhellipB

DQ-SQ

Single-quantum dimension

Do

ub

le-q

ua

ntu

m d

imen

sio

n

AhellipA

BhellipB

BhellipA

BhellipA

315

811910 1rsquo

2rsquo2

3rsquo 4rsquo6rsquo

5rsquo

1rsquo2910

asymp asymp asympHDO MeOD Acetone

1

lt 5 Aring

lt 5 Aring

H1-H2

H811

H15

H16

H6rsquo

H1rsquo

H1

H5rsquo

H2

H1-H2

H1rsquo-H1

H1rsquo-H5rsquo

H1rsquo

H1 H5rsquo

H1rsquo

H1

H5rsquo

H1H2rsquo H3rsquo H4rsquo

H1-H1

H1rsquo-H1

Hxrsquo-HxrsquoX= 2rsquo3rsquo4rsquo

H1-H2

H1rsquo

Zoom sugar region 28 ndash 45 ppmNOESY DQ-SQ

27

Head-tail or head-head configurations could be probed by the existence of H1rsquo-H1 657

interactions whereas H1 (δH= 356 ppm) is associated to the glucolipid tail and H1rsquo (δH= 432 658

ppm) to the sugar headgroup These atoms are more than 15 Aring apart and a direct interaction 659

could not be observed in both NOESY and DQ-SQ experiments NOESY experiments (Figure 660

9a) show that H1rsquo-H1 correlations are not observed Meanwhile clear-cut direct H1rsquo-H1 661

interactions are indeed observed are detected at T= 70degC (DQ cross-peaks related to H1rsquo-H1 662

correlation in DQ-SQ experiment shown in Figure 9b) Interestingly the absence of auto peaks 663

at δH= 432 ppm and δH= 335 ppm in the SQ dimension excludes H1rsquo-H1rsquo and Hxrsquo-Hxrsquo (x= 2 664

3 4) interactions respectively strengthening the idea that sugar headgroups from two nearest-665

neighbor G-C181-OH molecules are not adjacent in the nanotube-vesicle region 666

Homonuclear 2D ssNMR supports the idea according to which G-C181-OH can 667

undergo molecular flip-flops and a head-tail arrangement in the dry product Extrapolating these 668

conclusions in the lower concentration range lack of H1rsquo-H1 proximity at T= 90degC suggests a 669

head-head configuration in the vesicle-rich region and head-tail configuration within the 670

nanotubes It is however still unclear whether flip-flops are driven by spontaneous temperature-671

induced conformational changes in the glucolipid packing stimulating inclusions within the 672

membrane itself672223629ndash1218ndash21 or by defects generated by inter-vesicle collisions In fact a 673

combination of both is not unlikely Nanotubes are not stable when temperature is 20 to 30 674

degrees above Tm1 indicating that inter-vesicle collisions alone cannot explain the unusual 675

phenomenon In an energy landscape approach87 the head-head configuration is certainly 676

favored at high temperature while both head-head and head-tail configurations could be 677

favored in the mid-temperature region corresponding to the range between Tm1 and Tm2 in the 678

dry powder Both statistical and defect-induced flip-flop events contribute to form high mean 679

zero Gaussian curvature morphologies (nanotubes) for which the head-tail conformation 680

limiting the proximity between bulky sugar headgroups is probably more adapted (Figure 8c) 681

The flip-flop interpretation now reasonably explains the strong differences in the sugar and fatty 682

alcohol conformations deduced by the multiple NMR signals attributed to chemically-683

equivalent groups The existence of various low-energy conformations each having specific 684

NMR signature for sugars in solution78ndash8086 and solid-state81ndash84 further supports our 685

interpretation 686

One last point should be clarified Why between Tm1 and Tm2 are two molecular 687

environments able to coexist instead of a single one of average rigidity as found in the case of 688

ldquoripplerdquo phases50ndash54 We believe that this question can be answered based on the NMR results 689

the relatively short T2 values of the H1rsquo and OH protons shows the constrained mobility of the 690

28

sugar headgroup despite a certain flexibility of the chain (long T2 component) This trend may 691

suggest restrictions in translational mobility probably due to the sugar-OH hydrogen bonding 692

(Figure 8c) meaning that lipids are not able to diffuse within the membrane thus keeping two 693

different configurations stable 694

695

Conclusions 696

Topologically-connected nanotube-vesicle lipidic systems are commonly observed for 697

self-assembled membranes composed of at least two amphiphiles and spontaneously driven by 698

anisotropic inclusions andor by external directional forces This work shows that strong 699

differences in membrane curvature can spontaneously occur in single-lipid systems without 700

external anisotropic forces Coexistence and interconnection between nanotubes and vesicles 701

are evidenced via cryo-TEM and fluorescence confocal microscopy in a single-phase region at 702

temperature above a transition temperature Tm1= 483degC determined by DSC on the hydrated 703

sample but below the second transition temperature Tm2 (957degC) determined by DSC on the 704

dry powder X-ray diffraction (WAXS) also shows the coexistence of two lattice periods one 705

attributed to an inter-membrane repeating distance and the other possibly related to the intra-706

nanotube glucolipid arrangement Labelling the systems with Laurdan a fluorophore of which 707

the emission strongly varies with the local membrane rigidity excludes the presence of rigid 708

micro-separated phases (eg Lβ) 709

1H and 13C ssNMR despite the fact that it can only be performed on the dry powder 710

confirm the presence of one main fluid G-C181-OH environment above Tm2 Below Tm2 the 711

splitting of some 13C peaks and the coexistence of short and long 1H relaxation components 712

demonstrate that G-C181-OH can simultaneously pack into structurally and dynamically 713

inequivalent environments Combining advanced 2D 1H-1H correlation experiments based on 714

spin diffusion or double-quanta excitation suggests that vesicles could contain a preferred head-715

head configuration of the bolaform lipid while nanotubes a preferred head-tail assembly 716

The presence of two environments composed of the same molecule characterized by 717

different molecular configurations and rigidity fulfills the theoretical conditions necessary to 718

spontaneously nucleate a membrane inclusion of non-zero average orientation described as 719

necessary to observe topologically-connected nanotube-vesicle structures in the absence of 720

external forces This hypothesis could also be enriched by the possibility of nanotubes to be 721

pulled away between two vesicles after a collision event 722

This work expands the actual theoretical and experimental framework on lipid networks 723

stabilized by a different radius of curvature It could also motivate the synthesis or isolation 724

29

of new lipids providing a higher level of 3D architectural control in soft networks possibly in 725

the range of room temperature 726

727

Acknowledgements 728

Dr Andrea Lassenberger and Dr Sylvain Preacutevost at Institut Laue Langevin (ILL Grenoble 729

France) are kindly acknowledged for their assistance on the SANS experiment Confocal 730

images were performed at the Institut de Biologie Paris Seine (IBPS) imaging facility Jean-731

Franccedilois Gilles (Imaging Core Facility Institut de Biologie Paris Seine (IBPS) CNRS 732

Sorbonne Universiteacute Paris France) is kindly acknowledged for his assistance on the confocal 733

microscopy imaging Dr Guillaume Laurent (LCMCP Sorbonne Universiteacute Paris France) is 734

kindly acknowledged for helpful discussions The French region Ile-de-France SESAME 735

program is acknowledged for financial support (700 MHz NMR 736

spectrometer) Dr Olivier Diat (ICSM Marcoule France) Prof eronikaIglič and Prof V Aleš737

onsul discussi Slovenia) are kindly acknowledged for helpfUniversity of LjubljanaIglič (-Kralj 738

739

Financial support 740

WAXS experiments have been supported by Soleil Light Source Saint Aubin France proposal 741

Ndeg 20201747 SANS experiments have been supported by ILL proposal Ndeg9-13-778 742

743

References 744

1 D Kabaso N Bobrovska W Goacuteźdź E Gongadze V Kralj-Iglič R Zorec and A 745

Iglič Bioelectrochemistry 2012 87 204ndash210 746

2 C R Safinya U Raviv D J Needleman A Zidovska M C Choi M A Ojeda-747

Lopez K K Ewert Y Li H P Miller J Quispe B Carragher C S Potter M W 748

Kim S C Feinstein and L Wilson Adv Mater 2011 23 2260ndash2270 749

3 P Veranič M Lokar G J Schuumltz J Weghuber S Wieser H Haumlgerstrand V Kralj-750

Iglič and A Iglič Biophys J 2008 95 4416ndash4425 751

4 A Rustom R Saffrich I Markovic P Walther and H-H Gerdes Science (80- ) 752

2004 303 1007ndash1010 753

5 M I Angelova S Soleacuteau P Meacuteleacuteard F Faucon and P Bothorel Prog Colloid 754

Polym Sci 1992 89 127ndash131 755

6 L Mathivet S Cribier and P F Devaux Biophys J 1996 70 1112ndash1121 756

7 V Kralj-Iglič G Gomišček J Majhenc V Arrigler and S Svetina Colloids Surfaces 757

A Physicochem Eng Asp 2001 181 315ndash318 758

30

8 V Kralj-Iglič A Iglič G Gomišček F Sevšek V Arrigler and H Haumlgerstrand J 759

Phys A Math Gen 2002 35 1533ndash1549 760

9 M Drab D Stopar V Kralj-Iglič and A Iglič Cells 2019 8 626 761

10 A Zidovska K K Ewert J Quispe B Carragher C S Potter and C R Safinya 762

Langmuir 2009 25 2979ndash2985 763

11 A Zidovska K K Ewert J Quispe B Carragher C S Potter and C R Safinya Soft 764

Matter 2011 7 8363ndash8369 765

12 A Zidovska K K Ewert J Quispe B Carragher C S Potter and C R Safinya 766

Biochim Biophys Acta - Biomembr 2009 1788 1869ndash1876 767

13 M Karlsson K Sott M Davidson A S Cans P Linderholm D Chiu and O Orwar 768

Proc Natl Acad Sci U S A 2002 99 11573ndash11578 769

14 A Karlsson R Karlsson M Karlsson A-S Cans A Stromberg F Ryttsen and O 770

Orwar Nature 2001 409 150ndash152 771

15 M Karlsson K Sott A S Cans A Karlsson R Karlsson and O Orwar Langmuir 772

2001 17 6754ndash6758 773

16 E Evans H Bowman A Leung D Needham and D Tirrell Science (80- ) 1996 774

273 933ndash935 775

17 U Seifert Configurations of fluid membranes and vesicles 1997 vol 46 776

18 J Kaumls and E Sackmann Biophys J 1991 60 825ndash844 777

19 N Bobrovska W Goacuteźdź V Kralj-Iglič and A Iglič PLoS One 2013 8 e73941 778

20 A Iglič B Babnik K Bohinc M Fošnarič H Haumlgerstrand and V Kralj-Iglič J 779

Biomech 2007 40 579ndash585 780

21 K Bohinc T Slivnik A Iglič M Brumen and V Kralj-Iglič J Phys Chem C 2007 781

111 9709ndash9718 782

22 A Iglič H Haumlgerstrand P Veranič A Plemenitaš and V Kralj-Iglič J Theor Biol 783

2006 240 368ndash373 784

23 V Kralj-Iglič and P Veranič Adv Planar Lipid Bilayers Liposomes 2006 5 129ndash785

149 786

24 A Iglic and V Kralj-Iglic Adv Planar Lipid Bilayers Liposomes 2006 4 253ndash279 787

25 S Penič L Mesarec M Fošnarič L Mroacutewczyńska H Haumlgerstrand V Kralj-Iglič and 788

A Iglič Front Phys 2020 8 342 789

26 P Yager and P E Schoen Mol Cryst Liq Cryst 1984 106 371ndash381 790

27 T G Barclay K Constantopoulos and J Matisons Chem Rev 2014 114 10217ndash791

10291 792

31

28 B Božič G Gomišček V Kralj-Iglič S Svetina and B Žekš Eur Biophys J 2002 793

31 487ndash496 794

29 A Iglič H Haumlgerstrand M Bobrowska-Haumlgerstrand V Arrigler and V Kralj-Iglič 795

Phys Lett A 2003 310 493ndash497 796

30 R Bar-Ziv and E Moses Phys Rev Lett 1994 73 1392ndash1395 797

31 S Chiruvolu H E Warriner E Naranjo S H Idziak J O Raumldler R J Plano J A 798

Zasadzinski and C R Safinya Science (80- ) 1994 266 1222ndash1225 799

32 V Kralj-Iglič A Iglič H Haumlgerstrand and P Peterlin Phys Rev E 2000 61 4230ndash800

4234 801

33 L Mesarec M Drab S Penič V Kralj-Iglič and A Iglič Int J Mol Sci 2021 22 802

2348 803

34 L Van Renterghem S L K W Roelants N Baccile K Uyttersprot M C Taelman 804

B Everaert S Mincke S Ledegen S Debrouwer K Scholtens C V Stevens W 805

Soetaert L Van Renterghem S L K W Roelants N Baccile K Uytersprot M C 806

Taelman B Everaert S Mincke S Ledegen S Debrouwer K Scholtens C V 807

Stevens and W Soetaert Biotechnol Bioeng 2018 115 1195ndash1206 808

35 J Clark and F Deswarte Introduction to Chemicals from Biomass Second Edition 809

John Wiley amp Sons 2015 810

36 K V Sajna R Houmlfer R K Sukumaran L D Gottumukkala and A Pandey in 811

Industrial Biorefineries and White Biotechnology eds A Pandey R Houmlfer M 812

Taherzadeh K M Nampoothiri and C Larroche Amsterdam Oxford Waltham 813

Elsevier 2015 pp 499ndash521 814

37 R Marchant and I M Banat Trends Biotechnol 2012 30 558ndash565 815

38 K Hill and C LeHen-Ferrenbach in Sugar-Based Surfactants Fundamentals and 816

Applications ed C C Ruiz Boca Raton CRC Press 2009 pp 1ndash20 817

39 L Rodrigues I M Banat J Teixeira and R Oliveira J Antimicrob Chemother 818

2006 57 609ndash618 819

40 J D Desai and I M Banat Microbiol Mol Biol Rev 1997 61 47ndash64 820

41 S L K W Roelants K Ciesielska S L De Maeseneire H Moens B Everaert S 821

Verweire Q Denon B Vanlerberghe I N A Van Bogaert P Van der Meeren B 822

Devreese and W Soetaert Biotechnol Bioeng 2016 113 550ndash559 823

42 N Baccile M Selmane P Le Griel S Preacutevost J Perez C V Stevens E Delbeke S 824

Zibek M Guenther W Soetaert I N A Van Bogaert and S Roelants Langmuir 825

2016 32 6343ndash6359 826

32

43 N Baccile A-S Cuvier S Preacutevost C V Stevens E Delbeke J Berton W Soetaert 827

I N A Van Bogaert and S Roelants Langmuir 2016 32 10881ndash10894 828

44 N Baccile C Seyrig A Poirier S A Castro S L K W Roelants and S Abel Green 829

Chem 2021 23 3842ndash3944 830

45 A Salis L Cappai C Carucci D F Parsons and M Monduzzi J Phys Chem Lett 831

2020 11 6805ndash6811 832

46 J Schindelin I Arganda-Carreras E Frise V Kaynig M Longair T Pietzsch S 833

Preibisch C Rueden S Saalfeld B Schmid J-Y Tinevez D J White V 834

Hartenstein K Eliceiri P Tomancak and A Cardona Nat Methods 2012 9 676ndash835

682 836

47 T Parasassi and E Gratton J Fluoresc 1995 5 59ndash69 837

48 M M Dodes Traian F L Gonzaacutelez Flecha and V Levi J Lipid Res 2012 53 609ndash838

616 839

49 D M Owen C Rentero A Magenau A Abu-Siniyeh and K Gaus Nat Protoc 840

2012 7 24ndash35 841

50 W Stillwell in An Introduction to Biological Membranes Academic Press 2016 pp 842

135ndash180 843

51 W Stillwell in An Introduction to Biological Membranes Academic Press 2016 pp 844

181ndash220 845

52 B Tenchov Chem Phys Lipids 1991 57 165ndash177 846

53 R Koynova and M Caffrey Biochim Biophys Acta 1998 1376 91ndash145 847

54 F Schmid S Dolezel O Lenz and S Meinhardt J Phys Conf Ser 2014 487 848

012004 849

55 S Meinhardt R L C Vink and F Schmid Proc Natl Acad Sci U S A 2013 110 850

4476ndash4481 851

56 O Glatter and O Kratky Small Angle X-ray Scattering Academic Press London 852

1982 853

57 B A Cunningham A D Brown D H Wolfe W P Williams and A Brain Phys 854

Rev E 1998 58 3662ndash3672 855

58 M Nadler A Steiner T Dvir O Szekely P Szekely A Ginsburg R Asor R Resh 856

C Tamburu M Peres and U Raviv Soft Matter 2011 7 1512ndash1523 857

59 C Tanford The hydrophobic effect Formation of micelles and biological membranes 858

John Wiley amp Sons Inc New York 1973 859

60 M Masuda and T Shimizu Langmuir 2004 20 5969ndash5977 860

33

61 A Zidovska K K Ewert J Quispe B Carragher C S Potter and C R Safinya 861

Block Liposomes Vesicles of Charged Lipids with Distinctly Shaped Nanoscale 862

Sphere- Pear- Tube- or Rod-Segments Elsevier Inc 1st edn 2009 vol 465 863

62 A Iglic V Kralj-Iglic B Božič M Bobrowska-Haumlgerstrand B Isomaa and H 864

Haumlgerstrand Bioelectrochemistry 2000 52 203ndash211 865

63 T Shimizu M Masuda and H Minamikawa Chem Rev 2005 105 1401ndash1444 866

64 J V Selinger and J M Schnur Phys Rev Lett 1993 71 4091ndash4094 867

65 J M Schnur Science (80- ) 1993 262 1669ndash1676 868

66 B Tenchov R Koynova and G Rapp Biophys J 2001 80 1873ndash1890 869

67 R Koynova and B Tenchov Ser Biomech 2017 31 3ndash19 870

68 S J Angyal and R Le Fur Carbohydr Res 1984 126 15ndash26 871

69 J K Whitesell and M A Minton J Am Chem Soc 1987 109 225ndash228 872

70 T Takayama S Ando and I Ando J Mol Struct 1990 220 245ndash250 873

71 K Gawrisch N V Eldho and I V Polozov Chem Phys Lipids 2002 116 135ndash151 874

72 F Tielens C Gervais J F Lambert F Mauri and D Costa Chem Mater 2008 20 875

3336ndash3344 876

73 G A Jeffrey and Y Yeon Acta Crystallogr Sect B 1986 42 410ndash413 877

74 B Berglund and R W Vaughan J Chem Phys 1980 73 2037ndash2043 878

75 C Gervais C Coelho T Azaiumls J Maquet G Laurent F Pourpoint C Bonhomme 879

P Florian B Alonso G Guerrero P H Mutin and F Mauri J Magn Reson 2007 880

187 131ndash140 881

76 C Akita T Kawaguchi F Kaneko H Yamamoto and M Suzuki J Phys Chem B 882

2004 108 4862ndash4868 883

77 R N Purusottam L Seacutenicourt J J Lacapegravere and P Tekely Biochim Biophys Acta - 884

Biomembr 2015 1848 3134ndash3139 885

78 X Biarneacutes A Ardegravevol A Planas C Rovira A Laio and M Parrinello J Am Chem 886

Soc 2007 129 10686ndash10693 887

79 L M J Kroon‐Batenburg and J Kroon Biopolymers 1990 29 1243ndash1248 888

80 K N Kirschner and R J Woods Proc Natl Acad Sci U S A 2001 98 10541ndash889

10545 890

81 S Svenson B Kirste and J H Fuhrhop J Am Chem Soc 1994 116 11969ndash11975 891

82 I Sack S Macholl J H Fuhrhop and G Buntkowsky Phys Chem Chem Phys 892

2000 2 1781ndash1788 893

83 S Svenson J Koening and J H Fuhrhop J Phys Chem 1994 98 1022ndash1028 894

34

84 S Svenson A Schaefer and J H Fuhrhop J Chem Soc Perkin Trans 2 1994 2 895

1023ndash1028 896

85 K Bock and C Pedersen Adv Carbohydr Chem Biochem 1983 41 27ndash66 897

86 T B Grindley Structure and Conformation of Carbohydrates Springer Berlin 898

Heidelberg 2008 899

87 F Tantakitti J Boekhoven X Wang R V Kazantsev T Yu J Li E Zhuang R 900

Zandi J H Ortony C J Newcomb L C Palmer G S Shekhawat M O De La 901

Cruz G C Schatz and S I Stupp Nat Mater 2016 15 469ndash476 902

903

1

Supporting Information 904

905

Topological connection between vesicles and nanotubes in single-906

component lipid membranes driven by head-tail interactions 907

908

Niki Baccilea Ceacutedric Lorthioira Abdoul Aziz Baa Patrick Le Griela Cristina Coelhob Javier 909

Perezc Wim Soetaertd Sophie L K W Roelantsd 910

911

a Sorbonne Universiteacute Centre National de la Recherche Scientifique Laboratoire de Chimie de 912

la Matiegravere Condenseacutee de Paris LCMCP F-75005 Paris France 913

b Institut des Mateacuteriaux de Paris Centre Sorbonne Universiteacute Paris France 914

c Synchrotron Soleil LrsquoOrme des Merisiers Saint-Aubin BP48 91192 Gif-sur-Yvette Cedex 915

France 916

d InBio Department of Biotechnology Ghent University Ghent Belgium 917

918

919

Corresponding author 920

Dr Niki Baccile 921

E-mail address nikibaccilesorbonne-universitefr 922

Phone +33 1 44 27 56 77 923

924

925

926

2

a) 927

Parameter Method

Dry Matter (DM ) 996 Infrared balance 105degC

Glucose () 012 HPLC-Metacarb

Glycerol () nd HPLC-Metacarb

Free Fatty Acid content () lt001 Internal method BBEPP

GC with FID Detector

Oil () lt001 Internal method BBEPP

GC with FID Detector

Moulds (CFUg) lt10 3M-Nordval ndeg16 B

Yeasts (CFUg) lt10 3M-Nordval ndeg16 B

Anaerobic count (CFUg) lt10 SP-VG M005 B

Protein () tbd BCA protein assay

928

Figure S 1 - a) HPLC-ELSD chromatogram and composition table 929

930

931

932

933

3

b) 934

935

Group Ndeg δ ppm

CH2 1 354

CH2 2 153

CH2 3-712-14 132

CH2 811 203

CH 910 534

CH2 15 162

CH2 16 388

CH 1 424

CH 2 316

CH 3 334

CH 4 329

CH 5 354

CH2 6 366

Figure S 1 (next) ndash b) 1H NMR spectrum (MeOD-d4) recorded on the purified G-C181-OH compounds 936

and related assignment 937

938

2

811910 1rsquo

5rsquo1

3rsquo4rsquo

1rsquo1910

asymp asymp asympHDO MeOD Acetone

6rsquo2rsquo

1615

3-712-14

4

939

940

Figure S 2 ndash Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 130degC 941

5

942

943

Figure S 3 - Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 90degC 944

945

90degC

1

22

6

946

947

948

949

Figure S 4 ndash Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 70degC 950

951

21

11

1

2

5

3

4

45

5

6

7

7

7

952

953

954

Figure S 5 ndash Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 25degC 955

956

957

1

1 1

8

958

Figure S 6 ndash a) SANS experiment recorded at 90degC for a 5 mgmL G-C181-OH aqueous solution b-c) 959

WAXS experiments recorded on aqueous solutions of G-C181-OH at b) 5 mgmL and c) 25 mgmL d) 960

Structural interpretation of the repeating distances pointed out in c) 961

962

963

964

001 01 1

q Aring-1

25degC

68degC

88degC

111degC

01 1q Aring

-1

02 04 06 08 1q Aring

-1

001 01

10-1

100

101

Inte

nsity

cm

-1

q Aring-1

-2

T= 90degC

SANS WAXS WAXS

a) b) c)

dv

dL d= 144 Aring-1

di

ddH2O

dt

dvdL

dtdi d

d)

9

965

966

Figure S 7 ndash ssNMR MAS (10 kHz) experiments recorded on a G-C181-OH dry powder a) 1H spectra 967

recorded from 110degC to 70degC b) 13C single pulse (SP) spectra recorded with high power 1H decoupling from 968

110degC to 70degC c) 13C cross polarization (CP) spectra recorded at 110degC and contact time tc of 50 micros and 969

3000 micros d) 13C CP spectra recorded at 70degC and for 20 lt tc micros lt 3000 970

971

1

811

910

CH2 CH2

1rsquo

6rsquo

8 6 4 2 0 ppm

110

90

80

70

910

1rsquo

6rsquo16

1

2rsquo3rsquo4rsquo

811

OH

45 40 35 30 25 20 15

ppm

200 180 160 140 120 100 80 60 40 20 0

ppm140 120 100 80

ppm

CH2-gauche

3004

110

90

80

70

65

60

50

25

C91013003

C9 - 12869

CH2-all-trans

3198

C1rsquo10335

a) b)

1H

T degCT degC

13C Single Pulse MAS

A

B

A

B

70degC70degC

5rsquo

215

200 180 160 140 120 100 80 60 40 20 0

ppm

13C CPMAST= 110degC

200 180 160 140 120 100 80 60 40 20 0

ppm

132 131 130 129 128 127 126

ppm

106 105 104 103 102

ppm

13C CPMAST= 70degC

50

tc micros

3000

20

tc micros

50 500 1000 3000

tc micros

C102

13024 C910113003

C92

12869

C1rsquo213082

C1rsquo110443

A B

106 105 104 103 102

ppm

3000

1000

500

50

20

AB

c)

d)

10

972

Figure S 8 ndash a) Short and d) long T2 components displayed by the 1H transverse relaxation signal c) 1H 973

transverse relaxation signals recorded on a G-C181-OH dry powder from 110degC to 70degC The area of each 974

given resonance is plotted (log-lin) as a function of the echo time (CPMG pulse sequence) for each 975

temperature and fitted using the equation given above T2s and T2l indicate the short and long T2 relaxation 976

times respectively obtained by describing the 1H transverse relaxation signal using a biexponential decay 977

For H1rsquo H5rsquo and H6rsquo the 1H relaxation function could be described by a monoexponential decay 978

- Protons from the sugar group (H1rsquo H5rsquo H6rsquo) show a single T2 relaxation component (Figure 979

6c and Figure S 8) while protons of the acyl chain (CH2 H910 H811) show both short (Figure 980

6c) and long (Figure 6d) T2 relaxation component 981

- The short T2 relaxation time systematically decreases from 110degC to 70degC for all species by a 982

factor ranging from two to seven depending on the specific chemical group Very long T2 are 983

observed at 110degC for all acyl species while nearly unchanged values are measured below 984

90degC without any clear trend 985

- The sugar protons (H1rsquo H5rsquo H6rsquo) have the shortest set of T2 values (lt 1 ms) below 100degC 986

The long component of T2 generally ranges between 4 ms and 12 ms 987

0 1 2 3 4 5 6 7 8

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1 2 3 4 5 6 7 8

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

H1

H8H11

H9H10

CH2

H1rsquo

H6rsquo

CH2

H910H811 H1CH2

H1rsquo H5rsquo6rsquo

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

OH

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45 T= 110degC

T= 90degC

T= 80degC

T= 70degC

Log

In

ten

sity

Time ms

Fit

)

CH2 H910 H811 H1 H56 H1 OH

00

05

10

15

20

25

30

35

40

T2 m

s

T= 110degC

T= 90degC

T= 80degC

T= 70degC

CH2 H910 H811 H1

0

2

4

6

8

10

12

T2 m

s

T= 110degC

T= 90degC

T= 80degC

T= 70degC

Short T2(1H) Long T2(1H)

a) b)

c)

11

Table S 1 ndash 1H chemical shifts of G-C181-OH in solution and in the solid-state Solid-state spectra are 988

recorded at a MAS frequency of 10 kHz in DMSO-d6 broad peak 989

990

δH ppm

Group Ndeg Liquid

(MeOD-d4) Solid

(T= 110degC) Solid

(T= 70degC)

CH2 H1 354 357

CH2 H2 153 156

CH2 H3-712-14 132 135

CH2 H811 203 203

CH H910 534 534

CH2 H15 162 164

CH2 H16 388 383

CH H1 424 432

CH H2 316 335

CH H3 334 335

CH H4 329 335

CH H5 354 351 357

CH2 H6 366 383

OH OH1 432

45 463

OH OH2 492

OH OH3 490

OH OH4 487

OH OH6 445

991

992

H1H9H10H1rsquo

H6rsquo

OH2rsquo

OH1

12

Detailed discussion about the NOESY and DQ-SQ 2D 1H-1H ssNMR experiments presented in 993

Figure 9 994

995

2D 1H-1H ssNMR experiments are a unique tool to explore through-space proximities 996

between 1H nuclei ssNMR offers a large number of possibilities each adapted to a given 997

system The advantage of 2D maps with respect to a dynamics approach is their direct 998

readability in clear two nuclei which exchange their magnetization through-space (whichever 999

mechanism may be responsible for this exchange) will display a specific cross-peak However 1000

the way cross-peaks are presented and interpreted strictly depends on the nature of the pulse 1001

program which in turn is chosen according to the physical nature of the sample 1002

1D temperature-dependent 1H NMR experiments performed on G-C181-OH (Figure 1003

4a in the main text) indicate sharp peaks at 90degC and broad peaks at 70degC Narrow peaks 1004

indicate fast andor large-amplitude molecular tumbling averaging the strong 1H-1H dipolar 1005

interactions This excludes the use of pulse sequences like DQ-SQ which is not efficient under 1006

these conditions For this reason we employ spin-diffusion-based pulse sequences like 1007

NOESY which is more adapted for samples with high molecular mobility When mixing times 1008

are chosen to be short (lt 1 s) one generally explores through-space proximities within about 5 1009

10 Aring On the contrary when molecular motions are restricted dipolar coupling becomes 1010

significant 1H NMR peaks are broader and pulse sequences like DQ-SQ can now be used 1011

with a spatial resolution of less than 5 Aring 1012

Whenever possible DQ-SQ experiments are preferred over spin diffusion-based pulse 1013

sequences because they are more selective in particular for likewise interactions which are 1014

not discriminated in NOESY-type experiments This is illustrated by the typical interaction 1015

schemes for NOESY (top Figure 9a) and DQ-SQ (top Figure 9b) For two given A and B 1016

protons 2D NOESY maps will display two on-diagonal and two square-shaped symmetrical 1017

off-diagonal cross-peaks Off-diagonal cross-peaks illustrate the mutual A-B and B-A 1018

interaction but the on-diagonal peaks are systematic and do not prove any A-A or B-B 1019

interaction Similarly to NOESY 2D DQ-SQ maps display off-diagonal peaks (symmetric 1020

across the diagonal top Figure 9b) upon A-B and B-A interaction but in contrast they will 1021

display on-diagonal peaks only in the presence of self-interaction In the tutorial example of 1022

Figure 9b top the on-diagonal A-A peak indicates mutual A-A interaction while lack of on-1023

diagonal B-B peak evidences the absence of B-B proximities Combining NOESY and DQ-SQ 1024

experiments is capital to study the local arrangement of G-C181-OH and its evolution with 1025

temperature 1026

13

At T= 90degC the H1 and H5rsquo contributions are very close but possible to discriminate 1027

them (δ(H1)= 354 ppm δ(H5rsquo)= 351 ppm) according to 1H NMR chemical shifts in Table S 1028

1 2D 1H-1H NOESY in Figure 9a also discriminates between the two components attributed 1029

on the basis of the interactions with their mutual neighbors (green H1-H2 and H5rsquo-H1rsquo squares 1030

in Figure 9a) On this basis 2D NOESY does not show a direct H1rsquo-H1 correlation (blue red-1031

crossed) but rather an extensive network of intra-sugar interactions These features exclude 1032

suggest intra- and inter-sugar interactions and they exclude head-tail arrangements as a major 1033

molecular configuration at 90degC 1034

At 70degC the 1H NMR peaks are broader indicating a reduction of the T2 (1H) relaxation 1035

times andor residual homonuclear dipolar interactions excluding the use of NOESY In this 1036

case 1H-1H DQ-SQ correlation experiments are best suited Figure 9b presents the 1H-1H DQ-1037

SQ experiment on G-C181-OH at 70degC Peaks are broader than in NOESY and attribution may 1038

be cumbersome However the experiment is performed at 700 MHz and MAS= 22 kHz which 1039

improves spectral resolution with respect to the same experiment performed at 300 MHz and 1040

MAS= 10 kHz The auto-peak at δH= 356 ppm in the SQ dimension is attributed to H1-H1 spin 1041

pairs mainly resulting from intra CH2 interaction while the DQ cross-peaks at δH= 156-356 1042

ppm and δH= 432-356 ppm respectively origin from H2-H1 and H1rsquo-H1 interactions (Figure 1043

9b) DQ-SQ map does not show any auto-peak corresponding to H1rsquo-H1rsquo or Hxrsquo-Hxrsquo (x= 2rsquo 1044

3rsquo 4rsquo) interactions (red-crossed blue Figure 9b) thus excluding any proximity (lt 5 Aring) between 1045

close glucose neighbors On the contrary DQ-SQ maps (recorded both at 700 MHz Figure 9b 1046

or at 300 MHz not shown) systematically show DQ cross-peaks which can only be attributed 1047

to H1rsquo-H1 interactions (δH= 432-356) The presence of these H1rsquo-H1 interactions can only be 1048

explained by a head-tail arrangement of the G-C181-OH molecules as illustrated below the 1049

DQ-SQ map in Figure 9b 1050

3

A large body of both experimental and theoretical work has shown that budding and 51

eventual nanotube formation from an existing membrane can only occur spontaneously for 52

membranes of at least two-components9102425111218ndash23 and below a vesicle-nanotube phase 53

transition event2627 Otherwise external anisotropic driving forces like electroformation5ndash54

82829 osmotic pressure6 laser ldquotweezersrdquo30 or electrodynamics13ndash16 must be employed to obtain 55

similar results with in some cases an impressive degree of 2D and 3D organization13ndash16 The 56

origin of spontaneous nanotube formation has been shown to be related to a nanoscale phase 57

separation between two membrane components stabilizing low and high curvatures 58

respectively10ndash1231 with at least one component being highly anisotropic203233 Theoretically 59

this behavior has been explained by deviations in the elastic properties of membranes due to in-60

plane orientational ordering of membrane inclusions composed of anisotropic amphiphiles 61

these referring to a non-symmetrical shape upon a 90deg tilt along the amphiphile axis8919242533 62

In this work we show unexpected nanotubing of membranes prepared from a single-63

component lipid in the absence of external directional forces and above the lipidrsquos Tm This 64

phenomenon is observed for a novel anisotropic double amphiphile (bolaform amphiphile or 65

bolaamphiphile) a glucolipid composed of β-D-glucose and a C181-cis fatty alcohol (G-66

C181-OH Figure 1) This compound is obtained by microbial fermentation of a genetically-67

modified S bombicola yeast in the presence of oleyl alcohol34 and is developed in the broader 68

context of extending the library of new biobased surfactants and lipids in view of replacing 69

petrochemical low molecular weight amphiphiles35ndash41 The structure of G-C181-OH is 70

analogous to that of other microbial glycolipids developed through genetic engineering4243 71

Topological connections between nanotubes and vesicles are observed by means of 72

cryogenic transmission electron microscopy (cryo-TEM) fluorescence microscopy and wide 73

angle X-ray scattering (WAXS) above the melting temperatures Tm= 483degC under strong 74

dilution (05 wt) Spin diffusion and 1H double-quantum single-quantum (DQ-SQ) 2D solid-75

state nuclear magnetic resonance (ssNMR) spectroscopy under magic angle spinning (MAS) 76

necessarily performed on the dry glucolipid powder provides a spectroscopic ground to 77

understand the vesicle-nanotube coexistence The bolaform glucolipids could be in a head-78

headtail-tail configuration in the vesicles but rather in a head-tail configuration in the 79

nanotubes In the dry glucolipid both configurations are energetically comparable in the single-80

phase region above Tm and their coexistence is evidenced by the splitting of some 13C NMR 81

signals These facts could explain the stability of nanotubes while the following hypotheses are 82

formulated for their formation membrane inclusions with different orientational ordering13ndash16 83

possibly driven by inter-vesicle collisions 84

4

85

86

87

88

89

Figure 1 ndash Non acetylated C181 alcohol glucoside G-C181-OH is obtained by a bioprocess performed 90

with modified S bombicola yeast 91

92

Experimental Section 93

Synthesis of non acetylated C181 Alcohol Glucosides (G-C181-OH) G-C181-OH (Mw= 94

41856 gmol-1) was produced by aerobic whole cell bioprocess with a modified S bombicola 95

strain as described by Van Renterghem et al (Fig S4 in Ref 34) The molecule was purchased 96

from the Bio Base Europe Pilot Plant (Gent Belgium) and has the generalized chemical 97

structure given in figure Figure 1 The HPLC and 1H NMR spectrum (MeOD-d4) with peak 98

assignment are shown in Figure S 1 High purity levels (99) and high degree of uniformity 99

were obtained as can be derived from HPLC-ELSD chromatogram 1H NMR and table of 100

contaminant given in Figure S 1 101

102

1H solution Nuclear Magnetic Resonance (NMR) 1H solution NMR experiments were 103

performed on a Bruker Avance III 300 spectrometer using a 5 mm 1H-X BBFO probe using 104

methanol-d4 as solvent The number of transients is 8 with 3 s recycling delay an acquisition 105

time of 546 s and a receiver gain of 362 The 1H NMR spectrum and relative assignment are 106

shown in Figure S 1 and Table S 1 13C solution NMR were performed on the same probe using 107

DMSO-d6 Corresponding chemical shifts are given in Table 1 Referencing is done with 108

respect to TMS δ1H= 0 ppm δ13C= 0 ppm 109

110

Sample preparation The sample was dissolved in milliQ-grade water at the concentration of 5 111

mgmL (05 wt) Lack of pH-sensitive probes (eg COOH groups) in G-C181-OH as 112

otherwise found in other microbial amphiphiles44 but also the will to avoid ion-specific 113

effects45 exclude the use of buffer The solution was heat at the given temperature using a CH3-114

150 Combitherm-2 dry block heating device For the cryo-TEM experiments the vials were 115

heat at the given temperature next to the cryofixation device and plunged into liquid ethane as 116

fast as possible For temperatures of 25degC and 70degC the sample was heat at 90degC and then 117

Synthesis path

5

cooled at to the desired temperature For the SAXS experiment the sample was heat at 90degC in 118

the block heating device and cooled at to 25degC before analysis For the SANS experiment the 119

sample was heat at 90degC directly in front of the beam using the thermalized sample holder 120

available 121

122

Differential Scanning Calorimetry (DSC) DSC was performed using a DSC Q20 apparatus 123

from TA Instruments equipped with the Advantage for Q Series Version acquisition software 124

(v540) Acquisition was performed on both hydrated and dry powder sample (~ 3-5 mg) sealed 125

in a classical aluminium cup and using an immediate sequence of heating and cooling ramps at 126

a rate of 10degCmin-1 Melting temperatures Tm 1 and 2 Tm1 and Tm2 were taken at the minimum 127

of the endothermic peak 128

129

Small and Wide Angle Scattering experiments Small angle neutron scattering (SANS) 130

experiments were performed at the D11 beamline of Institut Laue Langevin (Grenoble France) 131

during the run No 9-13-778 Four q-ranges have been explored and merged using the following 132

wavelengths λ and sample-to-detector (StD) distances 1) ultra-low q λ= 135Aring StD= 39 m 133

2) low-q λ= 53Aring StD= 39 m 3) mid-q λ= 53Aring StD= 8 m 4) high-q λ= 53Aring StD= 14 m 134

The sample (C= 5 mgmL-1) was prepared in 999 D2O to limit the incoherent background 135

scattering The sample solution was analyzed in standard 1 mm quartz cells Direct beam empty 136

quartz cell H2O (incoherent scatterer) within the quartz cell were recorded and boron carbide 137

(B4C) was used as neutron absorber The sample acquisition was measured at 90degC where 138

temperature was controlled through the controller thermalized sample holder available at the 139

beamline The background sample (D2O) signal was subtracted from the experimental data 140

Absolute values of the scattering intensity were obtained from the direct determination of the 141

number of neutrons in the incident beam and the detector cell solid angle The 2D raw data were 142

corrected for the ambient background and empty cell scattering and normalized to yield an 143

absolute scale (cross section per unit volume) by the neutron flux on the samples The data were 144

then circularly averaged to yield the 1D intensity distribution I(q) The software package Grasp 145

(developed at ILL and available free of charge) was used to integrate the data while the 146

software package SAXSUtilities (developed at ESRF and available free of charge) was used to 147

merge the data acquired at all configurations and subtract the background 148

Wide-angle X-ray scattering (WAXS) was performed under temperature control at the 149

SWING beamline of SOLEIL synchrotron facility (Saint-Aubin France) during the run 150

20201747 (energy 14 keV sample-to-detector distance 05 m) Two sample concentrations in 151

6

H2O were probed C= 5 mgmL-1 and C= 25 mgmL-1 The 2D data were integrated azimuthally 152

at the beamline using the software Foxtrot and in order to obtain the I(q) vs q spectrum after 153

masking the beam stop shadow Silver behenate (d(100) = 5838 Aring) was used as a standard to 154

calibrate the q-scale Sample solutions were inserted in borosilicate capillaries of 15 mm in 155

diameter Capillaries were flame-sealed A capillary oven with controlled temperature (plusmn 05degC) 156

provided at the beamline was used to control the sample temperature between 25degC and 111degC 157

Data were normalized by the transmission and calibrated to the SAXS signal of H2O at large q-158

values (I= 00163 cm-1) in order to obtain an absolute intensity scale The water signal was 159

measured by subtracting the signal of the empty capillary from the signal of a water-filled 160

capillary The signal of (water + capillary) was used as background for the samples and it was 161

subtracted after integration of the 2D data 162

163

Cryogenic transmission electron microscopy (cryo-TEM) Cryo-TEM experiments were carried 164

out on an FEI Tecnai 120 twin microscope operating at 120 kV and equipped with a Gatan 165

Orius CCD numeric camera The sample holder was a Gatan Cryoholder (Gatan 626DH 166

Gatan) Digital Micrograph software was used for image acquisition Cryofixation was 167

performed using a homemade cryofixation device The solutions were deposited on a glow-168

discharged holey carbon coated TEM copper grid (Quantifoil R22 Germany) Excess solution 169

was removed and the grid was immediately plunged into liquid ethane at -180degC before 170

transferring them into liquid nitrogen All grids were kept at liquid nitrogen temperature 171

throughout all experimentation Cryo-TEM images have been treated and analyzed using Fiji 172

(is just ImageJ) software available free of charge at the developerrsquos website46 173

The samplersquos concentration was C= 5 mgmL-1 For technical reasons the temperature 174

during cryofixation could not be controlled For this reason the temperature associated with 175

each cryo-TEM image in the text was related to the equilibration temperature just before 176

cryofixation the device being as close as possible to the dry bath used for temperature control 177

Since temperature control during cryofixation was not possible in our device and practically 178

impossible at high temperatures one should take the T values as indicative of the given value 179

Although the sample transfer between the dry bath and the cryofixation device was performed 180

as fast as possible it is more than likely that the actual temperature of the sample at the moment 181

of cryofixation was lower than the one indicated here although impossible to know precisely 182

183

7

Fluorescence microscopy Images were recorded using a 40x objective on a Leica SP5 upright 184

confocal microscope with 405 nm laser excitation The sample was marked with 6-dodecanoyl-185

NN-dimethyl-2-naphthylamine (Laurdan Mw= 35354 gmol-1) fluorophore as follows A 186

stock solution of 3 mg of Laurdan in 100 microL of acetone was initially prepared (C= 30 mgmL-187

1) A volume of 028 microL of the stock Laurdan solution was diluted in 1 mL of a 5 mgmL-1 G-188

C181-OH water solution temporary set at 110degC in a sealed Eppendorf The latter was 189

important so to enhance inclusion of Laurdan in the membrane constituted by the glucolipid 190

while in its vesicle phase The final Laurdan concentration was approximately 24 micromol with a 191

molar Laurdan-to-G-C181-OH molar ratio of approximately 1-to-500 Fluorophore-to-lipid 192

ratios above 1200 were generally considered as non-interfering with the lipid phase behavior 193

as also verified in this work 194

Laurdan can be excited between 365 nm and 410 nm and its emission wavelength can 195

vary according to the rigidity of the membrane intercalation within a rigid or flexible 196

membrane generates emissions at 440 nm or 490 nm respectively47 In the present experiment 197

we have flame-sealed a solution of G-C181-OH at 5 mgmL-1 containing Laurdan (1-to-500 in 198

lipid-to-Laurdan ratio) in flat optical capillaries of 01 mm thickness Capillaries were set on a 199

Linkam support for temperature control eventually placed under the microscope The sample 200

was excited at 405 nm and detection was made in spectral mode with a 5 nm window sliding 201

every 5 nm from 412 nm to 602 nm in 16 bits which allows quantify the signal and compare 202

the different conditions Quantitative analysis was performed according to the literature4849 203

several ROI (range of interest) have been drawn on several images using the ROI manager of 204

Fiji (is just ImageJ) software The integrated intensity of each ROI at 440 nm and 490 nm was 205

eventually recorded and used to calculate the value of the generalized polarization GP as 206

follows of 207

119866119875 =(120582440 minus 120582490)

120582440 + 120582490 208

A negative GP indicates a flexible membrane while a positive GP indicates a rigid membrane 209

210

Solid-state Nuclear Magnetic Resonance (ssNMR) 1H and 13C ssNMR experiments have been 211

performed using an Avance III HD Bruker 705 T (ν1H= 300 MHz) spectrometer and a 4 mm 212

magic angle spinning (MAS) probe except for the BAck-to-BAck (BABA) experiment which 213

was performed on an Avance III Bruker 1644 T (ν1H= 700 MHz) spectrometer equipped with 214

25 mm MAS probe Temperature was controlled through a Smart cooler BCUIIa BCU-xtreme 215

unit using 10 min of equilibration time after each temperature variation Temperature was 216

8

always varied from high (110degC) to low (70degC) before analysis Due to combined use of fast 217

MAS rates necessary to achieve high resolution spectra and heating close to the boiling point 218

of water neither diluted nor concentrated samples could be used for ssNMR experiments All 219

tests employing up to 50 wt G-C181-OH in water resulted in sample centrifugation drying 220

but also expulsion from the rotor causing sudden stop of rotor spinning For these reasons all 221

attempts to study hydrated G-C181-OH failed and only the dry powder could be analyzed The 222

dry sample powder was analyzed as such in zirconia rotors (4 mm or 25 mm) spinning at MAS= 223

10 kHz (4 mm) or 22 kHz (25 mm) and Chemical shifts were calibrated with respect to 224

adamantane (13C 13C(CH2)= 3848 ppm = 191 ppm) and glycine alpha (13C(CO)= 17603 225

ppm) Details on the acquisition parameters are given in the Supporting Information 226

227

Results and discussion 228

Topologically-connected nanotubes and vesicles 229

G-C181-OH is a biobased glucolipid water-insoluble at room temperature and of 230

which the physicochemical properties are poorly known The DSC thermogram performed at 231

10degCmin of G-C181-OH as a function of the water content (Figure 2a) shows two endothermic 232

peaks on the dry powder (996 wt glucolipid) They are attributed to two melting phenomena 233

occurring at the following temperatures Tm 483degC (Tm1) and 957degC (Tm2) the enthalpy of the 234

former being lower than that of the latter Interestingly DSC also shows that Tm1 is hydration 235

independent and reversible (the associated exothermic peak is systematically observed) while 236

the transition at Tm2 is simply suppressed (here tested up to 130degC) on hydrated powders and 237

never observed on the cooling profile even at slow cooling rates (1degCmin) The thermogram 238

recorded on the dry powder is typical of single-component lipids having an intermediate50 239

metastable phase (often referred to the ripple phase Pβ´ in reference to the periodic undulation 240

of the bilayer)50ndash54 between the classical lamellar gel (Lβ) and liquid crystalline (Lα) phases50ndash241

54 and characterized by regions of both liquid order and disorder5455 The effect of water does 242

not influence the low-temperature crystallization at Tm1 but it contributes to disrupt the ordered 243

array of hydrocarbon chains in the metastable phase 244

On the basis of the DSC thermogram the aqueous phase behavior of G-C181-OH under 245

diluted conditions (5 mgmL-1) was studied at four different temperatures (red marks on Figure 246

2a) below and above Tm1 but also between Tm1 and Tm2 corresponding to the thermogram of 247

the dry powder 248

9

249

Figure 2 ndash a) DSC thermogram of G-C181-OH under dry (04 wt water estimated by thermogravimetric 250

analysis) and hydrated conditions recorded at a heating rate of 10degCmin Red marks indicate the 251

temperatures at which cryo-TEM experiments are performed on the corresponding 5 mgmL aqueous 252

solutions b) Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 130degC 253

90degC 70degC and 25degC T indicates the temperature of the sample solution before cryofixation Tne T T 254

being the temperature of the sample upon plunging in liquid ethane Please refer to the materials and 255

methods section for more information 256

T= 130degC

T= 90degC

T= 70degC

T= 25degC

a)

b)

20 40 60 80 100 120

99 wt

50 wt

25 wt

T degC

En

do

10 wt

Tm1= 483 degC

Tm2= 957 degC

10

257

The series of cryo-TEM images in Figure 2b show an overview of the most relevant 258

structures observed on G-C181-OH solution from T= 130degC to T= 25degC whereas T is 259

indicative of the equilibrium temperature before cryofixation (refer to materials and methods 260

section for more information on the sense and interpretation of T) At Tgt 100degC the sample 261

is essentially composed of polydisperse single wall vesicles massively shown by the 262

complementary pictures presented in Figure S 2 In between Tm1 and Tm2 at T= 90degC and T= 263

70degC (Figure 2b Figure 3 Figure 4 complemented by Figure S 3 and Figure S 4) vesicles are 264

connected by nanotubes At Tlt Tm1 flat crystals are otherwise observed (Figure 2b and Figure 265

S 5) 266

267

268

Figure 3 ndash Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 90degC 269

270

A deeper insight at T= 90degC is given in Figure 3 (more images are provided in Figure 271

S 3) Arrows 1 through 6 mark very clearly the typical continuity in the glucolipid membrane 272

between the vesicle and nanotubes Arrow Ndeg 7 shows that a single nanotube connects the outer 273

membrane of two vesicles several hundred nanometers apart The connection is shown by 274

2

34

5

6

6

1

5

7

A

B

D

C

0 10 20 30 40 50 60

BIn

tensity

au

Profile nm

A

0 10 20 30 40 50 60

Inte

nsity

au

Profile nm

C

D

T= 90degC

11

arrows Ndeg5 and Ndeg6 The corresponding profiles A through D confirm the tubular nature of 275

the elongated objects Profiles B and D are both characterized by typical symmetrical thick 276

walls (low intensity depths highlighted by shaded rectangles) surrounding a hollow region (high 277

intensity peak between the shaded rectangles) The wall thickness varies between 35 and 40 278

nm for profile B and between 41 and 61 nm for profile D while the total nanotube diameter 279

varies between 14 nm at B and 20 nm at D At the same time the thickness of the vesicle wall 280

is about 6 nm at A and 135 nm at C Considerations about the bilayer or interdigitated structure 281

of the membrane will be presented below 282

The connection between nanotubes and vesicles seems to be multiple For instance 283

arrow Ndeg1 (Figure S 3) and Ndeg4 (Figure 3) show a single-wall vesicle of which the membrane 284

is extruded into a nanotube Similarly a closer look at arrows Ndeg1 2 and 3 in Figure 3 show 285

that the vesicle is single-wall and fully closed while the nanotube nucleates at the vesicle outer 286

surface with no apparent topological continuity In many regions of the sample at T= 90degC 287

one can find exvaginations or buds232531 in the nanotube wall This is shown for instance by 288

arrows Ndeg7 and particularly put in evidence in Figure 4 by arrows Ndeg1 through Ndeg4 where the 289

amplitude of the oscillation varies between 20-25 nm (arrow Ndeg1) and 6-8 nm (arrows Ndeg3-4) 290

291

Figure 4 - Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heated at T= 90degC 292

293

Still between Tm2 and Tm1 but at lower temperature (T= 70degC) the nanotubes seem to 294

vary from single threads (arrow Ndeg1 Figure S 4) to nanotubes of diameter below 15 nm both 295

connecting nanotubes of diameter ranging between 50 nm and 100 nm (arrows Ndeg2 Figure S 296

4) The larger tubes are most likely formed by the inflation and flattening of the nanotubes 297

probably after assimilation of surrounding vesicles and fusion of the exvagination discussed 298

above The transition between vesicles tubes and flat membranes is certainly more accentuated 299

Arrow Ndeg3 in Figure S 4 shows distinct vesicles while arrow Ndeg4 identifies a flat structure 300

evidently formed from fused vesicles In particular arrow Ndeg5 points at the fusion of two 301

2

1 132

4

12

vesicles while arrow Ndeg6 points at two vesicles fusing into the flat membrane Additional areas 302

where vesicle fusion is prominent are indicated by the oscillations at the outer skirt of broad 303

lamellae and pointed by arrow Ndeg7 304

Finally below Tm1 at T= 25degC for instance mainly flat crystals with no visible 305

exvaginations or tubes are visible (Figure S 5) by cryo-TEM and the material turns into an 306

insoluble precipitate Massive presence of flat crystalline structures is also clear from the bulk 307

iridescent behavior of a G-C181-OH solution under manual shear 308

In summary despite the uncertainty related to the actual state of equilibrium of the 309

glucolipids at the moment of cryofixation cryo-TEM depicts a vesicle phase at Tgt 100degC a 310

flat lamellar phase below Tm1 and an interconnected vesicle-nanotube network between Tm1 and 311

about 100degC The combination of statistically-relevant fluorescence microscopy and 312

scatteringdiffraction experiments provides an additional proof of the coexistence of 313

vesiclenanotube systems between the two transition temperatures Please note that the 314

experiments reported below have been performed under controlled conditions of temperature 315

which is then indicated by the use of T instead of T 316

13

317

Figure 5 ndash a-c) Confocal fluorescence microscopy images recorded at 90degC and 70degC for a 5 mgmL G-318

C181-OH aqueous solution d-g) Typical fluorescence microscopy images recorded at T= 80degC using 319

Laurdan (refer to materials and methods section for more details on this fluorophore) Excitation 320

wavelength is 405 nm for all the images while emission wavelength is 490 nm in (df) and 440 nm in (eg) 321

h) Full emission spectra obtained from the integration of various fluorescence images recorded from 412 322

nm to 600 nm (step 5 nm) at 90degC 80degC and 70degC i) Box plot of GP defined as 119918119927 =(120640120786120786120782minus120640120786120791120782)

120640120786120786120782+120640120786120791120782 with 120640120786120786120782 323

and 120640120786120791120782 being the emission wavelengths at 440 nm and 490 nm at 90degC 80degC and 70degC Each data point 324

l= 490 nm l= 440 nm

l= 490 nm l= 440 nm

10 μm400 450 500 550 600

00

02

04

06

08

10

No

rma

lize

d I

nte

nsity

l nm

90degC

80degC

80degC

70degC

70degC

400 450 500 550 60000

02

04

06

08

10

No

rma

lize

d I

nte

nsity

l nm

90degC

80degC

80degC

70degC

70degC

d) e)

f) g)

h)

i)

T= 80degC

Laurdan

T= 80degC

T= 80degC T= 80degC

T= 90degC T= 70degC T= 70degC

a) b) c)

1

1

1

14

corresponds to a set of regions of interest (ROIs) (54 at 90degC 65 at 80degC and 75 at 70degC) selected in the 325

corresponding fluorescence images ROIs were representatively selected so to include the entire image but 326

also vesicles alone tubes alone and mixtures of vesicles and tubes 327

328

Figure 5ab shows two typical confocal fluorescence microscopy images recorded on a 329

G-C181-OH solution (5 mgmL) in a flame-sealed capillary heat at T= 90degC and T= 70degC 330

Although the resolution of optical microscopy is not comparable with cryo-TEM the sample is 331

broadly composed of vesicles coexisting with nanotubes whereas in some cases the nanotube 332

starts from the vesicle itself (arrow Ndeg1) in good agreement with cryo-TEM data Furthermore 333

SANS recorded at T= 90degC (Figure S 6a) shows a scattering profile with a pronounced -2 334

dependence of the scattering intensity against the wavevector q in the log-log scale This 335

indicates the massive presence of a flat interface and it is expected typical of bilayers in 336

vesicular morphologies56 However the typical signature of nanotubes cannot be observed in 337

the SANS profile and this can be explained by both an argument related to the relative lower 338

amount of nanotubes with respect to vesicles at 90degC but also to the steeper slope of flat 339

morphologies (-2) compared to tubes (-1) the signal of the former masking the tube signal at 340

low wavevector values The SANS profile is characterized by a diffraction peak at q= 0167 Aring-341

1 of which the assignment is refined by WAXS data 342

Temperature-driven WAXS experiments in Figure S 6bc (5 mgmL-1 and 25 mgmL-1 343

respectively) show a lamellar order possibly attributed to a multilamellar structure of the 344

vesicles (dv Figure S 6d) at high temperature and to the interplanar arrangement of the 345

crystalline flat structures (dL Figure S 6d) at T= 25degC The sharp (001) reflection indicates 346

typical interlamellar distance of 353 Aring at 25degC although this value varies with temperature 347

(400 Aring at 111degC 372 Aring at 88degC) thus confirming its assignment to a hydration-dependent 348

interlamellar order More intriguing is the second broader peak at q= 0403 Aring-1 (d= 156 Aring) 349

and its reflection both clearly visible at 88degC below Tm2 We attribute it to the inter-glucolipid 350

arrangement inside the tubular phase (dt Figure S 6d) coexisting with the vesicle phase 351

between Tm1 and Tm2 and to the planar lipid arrangement (di Figure S 6d) in the rigid flat 352

lamellar structure below Tm2 Its temperature- hence hydration- independent position also 353

confirms its structural origin Finally besides the water-structure peak at about 2 Aring-1 we also 354

observe a broad peak at 144 Aring-1 characteristics of the lateral packing (d Figure S 6d) of lipids 355

in bilayers While its position could be consistent with an Lα phase5758 above Tm2 its broadness 356

and positional invariance as well as the coexistence between vesicles and tubes do not allow a 357

precise assignment 358

15

Does G-C181-OH assume a bilayer or interdigitated structure the latter being the one 359

found for other similar bolaform glucolipids4243 The size of G-C181-OH can be estimated to 360

be at the most 32 Aring given by the length L of the C16 tail (L = 22 Aring from the Tanford formula 361

L= 154 + 1265 times n (n= 16)59) and the sugar (about 10 Aring)4243 However the cis conformation 362

of the C=C bond imposes a ldquokinkrdquo of 60deg in the aliphatic chain which reduces the effective 363

length of the oleic backbone to about 17 Aring for an effective total G-C181-OH length of about 364

27 Aring Cryo-TEM provides a tube thickness varying from 35 Aring to 60 Aring and a vesicle membrane 365

thickness of at least 60 Aring WAXS suggests an interlamellar thickness (membrane and water) 366

between 40 Aring and 37 Aring and an inter-lipid distance in the tube of about 16 Aring In the absence of 367

a clear form factor oscillation profile as found for similar glucolipids4243 one must 368

acknowledge that these data pledge neither in favor of fully interdigitated nor a double layer 369

structure and might suggest a tilted lipid arrangement with partial interdigitation60 370

371

13C ssNMR demonstrates that G-C181-OH can assume two coexisting head-head and 372

head-tail configurations 373

Vesicles are characterized by low positive mean and gaussian curvatures while 374

nanotubes are described by a combination of a high positive mean and zero gaussian curvatures 375

In this regard the coexistence of vesicles and nanotubes requires glucolipids to adapt 376

themselves to both environments at the same time For this reason the formation of nanotube-377

vesicle networks generally requires a specific driving force able to impose a high mean and 378

zero gaussian curvature to the membrane In many systems insertion of at least a second 379

constituent (lipid surfactant protein) in the membrane becomes a necessary internal driving 380

force for the nucleation and growth of so-called membrane inclusions from which 381

topologically-connected nanotubes can form10ndash123261 as explained by orientational ordering 382

arguments62 It has also been shown that external driving forces such as use of laser 383

ldquotweezersrdquo30 osmotic pressure6 electroformation5ndash82829 or controlled suction13ndash16 are 384

otherwise necessary as also supported by theoretical studies19 385

The bending energy of tubes is higher than that of a vesicle membrane8 due to the 386

dependence of the membrane free energy from the principal curvatures For this reason 387

isotropic (rotational symmetry along the main axis) but also anisotropic33 lipids generally 388

stabilize low-curvature shapes like flat membranes or vesicles Lipid nanotubes are also a 389

common phase formed by amphiphiles but generally for lipids with a certain complexity in 390

their molecular structures as exampled by diacetylenic derivatives of phospholipids among 391

others262763ndash65 Negative contributions to the bending energy favoring the simultaneous 392

16

presence of interconnected vesicles and tubes was shown to be driven by the presence of 393

anisotropic lipids in a membrane continuum Spontaneous deviations to the elastic theory in the 394

presence of surfactants lipids and proteins with anisotropic structures favor the formation of 395

protrusions with radius of the order of the fraction of a micron9ndash1219ndash2362 396

In the absence of both internal and external driving forces mentioned above G-C181-397

OH should behave as a standard lipid Its DSC profile (Figure 2)536667 under both hydrated and 398

dry conditions identifies a hydration-independent transition (Tm1) slightly above room 399

temperature this is followed by one main transition at higher temperature (Tm2) in the dry 400

sample50 For such a profile one expects a gel phase (Lβ) below Tm1 and a liquid crystalline 401

structure possibly in the morphology of vesicles (L4 phase) possibly above Tm1 and most likely 402

above Tm231 as experimentally found On the basis of the thermogram recorded on the dry 403

powder one may also expect a ldquoripplerdquo (Pβ) phase between Tm1 and Tm2 characterized by 404

periodic undulations of the bilayer (generally addressed to as regions with both liquid order and 405

disorder)5455 406

In the present work instead of a Lβ phase we observe flat crystals below Tm1 and a 407

vesicle-nanotube network instead of the ldquoripplerdquo phase The latter is particularly unexpected 408

because the membrane is composed of a single-component (G-C181-OH) and no anisotropic 409

external force (osmotic hydrodynamic electric) is used to pull and stabilize the tubes In 410

addition the nanotubes are observed at fairly high temperatures above Tm1 (between about 411

60degC and 90degC) while this morphology is generally favored at low temperature (proportionality 412

with 1kT figure 5 in Ref 62) commonly below a given transition temperature characteristics 413

for each specific amphiphile27 Furthermore nanotubes of G-C181-OH seem to be stable in 414

time thus differing from what is found in electroformed vesicles78 They are also observed 415

away from phase transition events (may them be at Tm1 or Tm2) as otherwise reported before 416

for a diacetylenic derivative of phosphatidylcholine26 and many other systems27 417

A stable nanotube-vesicle network for a single-component phospholipid in the absence 418

of external driving forces is not expected and has never been reported before It could be 419

explained by the coexistence of two micro-separated phases of different rigidity This 420

hypothesis is excluded by confocal microscopy on Laurdan-labelled membranes and magic 421

angle spinning (MAS) solid-state NMR (ssNMR) experiments 422

Laurdan is a well-known fluorophore used to put in evidence the coexistence of rigid 423

and fluid domains in both model and biological membranes47ndash49 Laurdan was used here to label 424

the G-C181-OH membranes between Tm1 and Tm2 and to verify the possible micro-segregation 425

and coexistence of rigid (eg Lβ λem= 440 nm) and fluid (eg Lα λem= 490 nm) domains in 426

17

tubes and vesicles respectively A set of quantitative experiments performed on a 5 mgmL-1 427

solution summarized in Figure 5 shows an excess of fluorescence at only λem= 490 nm for 428

temperature between 90degC and 70degC (Figure 5h) The GP value which quantifies the rigidfluid 429

character of the membrane is systematically negative (Figure 5i) indicating an overall fluid 430

membrane at all temperatures GP is quantified on more than 50 regions of interest (ROIs) at 431

each temperature including vesicles and tubes alone as well as connected nanotube-vesicles 432

(Figure 5d-g) The dispersion of GP at 80degC and 70degC between -08 and -02 could suggest the 433

spurious presence of membranes with various degrees of flexibility although fluidity is always 434

predominant The combination of WAXS (broad peak d= 144 Aring-1 invariant with 435

temperature Figure S 6bc) and confocal fluorescence microscopy using Laurdan confirms the 436

single-phase of the membrane composed of G-C181-OH both in vesicles and nanotubes This 437

is also consistent with the lack of phase transitions in DSC experiments between ~85degC and 438

~60degC especially in hydrated samples We conclude that the vesicle-nanotube transition is not 439

driven by the local segregation of gel phase (Lβ) microdomains nor that the nanotubes are 440

composed of such phase 441

ssNMR is the best-suited approach to investigate the origin of the local driving force for 442

the nucleation of anisotropic inclusions of which the average orientation depends on the local 443

membrane and inclusion curvatures and was shown to promote tubing62 ssNMR can provide 444

information on the local molecular mobility eg through the nuclear spin relaxation behavior 445

but also on intermolecular interactions and stereochemistry eg through variations in the 446

chemical shift68ndash70 447

Ideally ssNMR should be performed on diluted G-C181-OH samples above Tm1 448

However fast MAS necessary for high resolution centrifuges the sample As a first 449

approximation hydrated G-C181-OH samples for instance at 50 wt as classically done for 450

phospholipids71 could be studied Unfortunately permeability of the sample holder combined 451

to heating (close to the boiling point of water) and MAS provoke dehydration and even sample 452

ejection After experiencing both of the above this approach was abandoned As a second 453

approximation the study of the dehydrated G-C181-OH powder avoids centrifugation and 454

evaporation If studying dried lipids may not reflect their exact behaviour under diluted 455

conditions valuable structural information can still be obtained in dehydrated molecular 456

systems like proteins thus illustrating that this approach is not unreasonable Furthermore if 457

DSC (Figure 2a) shows that hydration suppresses the Tm2 it also shows that the value of Tm1 is 458

not affected by water at concentrations as low as 10 wt For this reason performing a 459

structural analysis of dehydrated G-C181-OH by ssNMR above Tm1 may help understanding 460

18

the local arrangement of G-C181-OH in two morphologies vesicles and nanotubes having 461

sensibly different curvatures 462

To explore the local mobility of the most relevant chemical groups of G-C181-OH 463

above Tm2 (T= 110degC) and between Tm2 and Tm1 (70 lt T degC lt 90) we combine a series of 1H 464

and 13C experiments Single-pulse (SP) MAS experiments (1H and 13C) when recorded under 465

quantitative conditions allow the detection of all chemical species while cross-polarization 466

(CP) MAS (13C only) experiments are sensitive only to constrained groups for which the 1H-467

13C dipolar coupling is not averaged to zero by the molecular motions 468

469

Figure 6 ndash a) 1H MAS (10 kHz) ssNMR spectra recorded from 110degC to 70degC b) 13C MAS (10 kHz) ssNMR 470

spectra recorded using high power decoupling single pulse (SP) and cross-polarization (CP) pulse sequences 471

Contact time for CP experiments and corresponding temperatures are given on each spectrum Full spectra 472

are given in Figure S 7 473

The most relevant 1H and 13C spectra are given in Figure 6 (full range in Figure 7 and 474

Figure S 7) As expected the 1H solid-state SPMAS spectra show broader peaks compared to 475

the solution-state NMR spectrum (Figure 6a) reflecting the obvious solid-state character of the 476

sample at any temperature The sugar group and the C=C bond resonate in the 25-6 ppm range 477

and 535 ppm respectively while the OH groups of the sugar moiety are found below 45 ppm 478

(Figure 6a full assignment in Table S 1) Very broad (~1 ppm) at 100degC the position of the 479

OH peak varies with the temperature value and becomes narrower (lt~05 ppm) and visible at 480

δH= 464 ppm at T= 70degC Line narrowing and chemical shift variations of the OH resonance 481

in ssNMR is always a direct sign of H-bonding interactions whereas the strength of the H-482

bonding hence its OHhellipH distance is correlated to a dispersion in chemical shifts72ndash75 All other 483

60 55 50 45 40 35 30 25

ppm

110

90

80

70

910

1rsquo

6rsquo16

1

OH

134 132 130 128 126

ppm110 108 106 104 102 100

ppm

SP (T= 110degC)

SP (T= 70degC)

CP 50 ms (T= 70degC)

CP 3 ms (T= 70degC)

910 1rsquo

1H 13C

1

811

910

CH2 CH2

1rsquo

6rsquo

a) b)

C102

13024

C9101

13003

C92

12869 C1rsquo210382

C1rsquo110443

C1rsquo13035

2rsquo3rsquo4rsquo

T degC

5rsquo

19

resonances broaden from 110degC to 70degC thus indicating slower andor more anisotropic 484

reorientational motions of all chemical groups The specific dynamic behaviour of the sugar 485

headgroup is also reflected by its 1H transverse relaxation time T2 whereas the shorter the T2 486

the higher the rigidity Experiments performed between 110degC and 70degC shown and 487

commented in Figure S 8 in the Supporting Information globally show a single short (lt 1 ms 488

below 100degC) T2 relaxation component for H1rsquo H5rsquo and H6rsquo and two both short and long 489

(between 4 ms and 12 ms) T2 for the acyl chain (CH2 H910 H811) even below 100degC 490

At 110degC 13C SPMAS experiments (Figure 6b full range in Figure 7 and Figure S 7 491

assignments in Table 1) show a single peak at δC= 10335 ppm for C1rsquo and at δC= 13003 ppm 492

for C910 as expected for a symmetrical conformation with two chemically-equivalent 493

carbons76 At 70degC the C1rsquo peak disappears while C910 now shows two peaks around 128 494

ppm and 130 ppm (Table 1 full temperature series in Figure S 7b) attributed to C92 (δC= 12869 495

ppm) and C10 in a non-symmetric configuration respectively for which these two carbons are 496

chemically inequivalent76 More precisely the resonance at 130 ppm is actually the 497

superposition of two close resonances C9101 (δC= 13003 ppm) and C102 (δC= 13024 ppm) 498

where subscripts 1 and 2 stand for two different conformations of the cis-olefin group as 499

reported for various phases of oleic acid in the vicinity of 0degC76 500

501

Table 1 ndash 13C Chemical shifts of G-C181-OH in solution and solid-state Solid-state experiments are 502

recorded at MAS= 10 kHz All spectra are reported in Figure 7 and Figure S 7 The contact time for cross 503

polarization is 3 ms In all ssNMR experiments temperature is first raised at 110degC and then lowered to 504

either 90degC or 70degC Attribution of the 13C resonances in Table 1 is performed on the basis of the solution 505

spectrum (Figure S 1b) Additional peaks observed in both single pulse and cross polarization ssNMR 506

experiments are attributed when possible to a configuration effect of the given atom eg C9101 and C92 507

indicate two configurations of C9 All new peaks observed in solid-state and not detected in solution are 508

highlighted in red for convenience 509

510

511

δ ppm

Peak Liquid (DMSO-d6)

Solid Single Pulse

T= 110degC

Solid Single Pulse

T= 90degC

Solid Single Pulse

T= 70degC

Solid Cross Polarization

T= 70degC

C102

- - 13024 13024

C910C9101 13009 13003 13003 13003 13003

C1C9C10C1rsquo

C6rsquo

20

C92 - - - 12869 12869

C1rsquoC11 10329 10335 10335 - 10443

C12 - - - - 10382

C3 7724 7709 7702 - 7829

- 7634 7629 - 7621

7525

C2 7388 7398 7390 7370 7370

C4 7052 7077 7054 7041 7098

C16 6900 7016 - - 7029

- - - - 6793

C1C5C6 6153 6241 6236 6239 6262

C1C5C6 6117 - - - 5956

C2 3300 3290 3290 3283 3279

CH2(trans) - - - 3198 3196

CH2(gauche)C15 2941 2981 2987 3004 -

C811 2706 2747 2747 2832 2817

CH2 2597 2625 2628 2630 2817

512

13C CPMAS spectra at 70degC (Figure 6b full range in Figure 7 and Figure S 7d) at 513

increasing contact times confirm the presence of two cis-olefin environments one for which 514

C910 are equivalent labelled here C9101 and one for which they are inequivalent labelled 515

C92 C102 Both environments are rigid enough to be observed at short contact time (tc= 50 μs) 516

A similar double environment is also observed for the sugar C1rsquo CPMAS at 70degC shows two 517

peaks (C1rsquo1 at δC= 10443 ppm C1rsquo2 at δC= 10382 ppm Figure 6b Figure S 7d) These are 518

significantly shifted downfield from the C1rsquo peak observed at 110degC and are detected at contact 519

times as short as 50 μs thus indicating their rigid character One also finds that CPMAS is 520

totally inefficient at 100degC for short contact times (50 μs) and poorly efficient even at 3 ms 521

typical of fast-tumbling molecules (Figure S 7c) also both gauche (δC= 3004 ppm) and all-522

trans (δC= 3198 ppm) conformations of the CH2 carbons coexist at 70degC (Figure S 7b)77 523

According to ssNMR one main mobile molecular environment exists at 110degC as 524

shown by the well-resolved 1H spectrum single C=C (C9101) and anomeric (C1rsquo) resonances 525

gauche conformation of the aliphatic CH2 and single T2 of the aliphatic CH2 chain and sugar 526

protons The sugar group is certainly mobile (inefficient 1H13C CPMAS) but still more 527

constrainted than the acyl chain (shorter T2 (1H)) As the temperature is lowered the overall 528

rigidity increases (broadening of 1H NMR peaks faster 1H transverse relaxation and efficient 529

1H13C magnetization transfer during CPMAS) presumably driven by H-bonding 530

interactions (downfield shift and line narrowing of the OH 1H signal very short OH T2 loss of 531

the C1rsquo signal in SPMAS) At the same time a number of spectroscopic data (two concomitant 532

21

resonances for anomeric C1rsquo1 and C1rsquo2 cis-olefin C9101 and C92 C102 and CH2 gauche and 533

all-trans environments) and relaxation behaviour (fast and slow T2 relaxation components) 534

strongly suggests the coexistence of two molecular environments Further proofs of the 535

coexistence of two distinct molecular domains are given by a number of chemical shift 536

variations in the 13C CPMAS spectrum at 70degC compared to the one at 110degC 90degC and 537

solution This is shown in Figure 7 and in the list of 13C chemical shifts in Table 1 538

539

540

Figure 7 - Comparison between solution- and solid-state 13C NMR spectra of G-C181-OH Solid-state 541

experiments are recorded directly on the sample powder at a MAS of 10 kHz (4 mm zirconia rotor) SP 542

Single Pulse CP Cross Polarization In all the ssNMR experiments temperature is first raised at 110degC 543

and then lowered to either 90degC or 70degC Segmented lines are meant to highlight the known peak position 544

in solution thus stressing the strong chemical shift variations in the solid-state 545

546

NMR experiments above Tm2 suggest that vesicles could contain G-C181-OH with a 547

single family of mobile conformations characterized by chemically-equivalent C=C (red 548

Figure 8b) and a single C1rsquo conformation (green Figure 8b) Experiments between Tm1 and Tm2 549

otherwise suggest that two coexisting conformations characterized by chemically-equivalent 550

(red Figure 8b) and inequivalent (magenta Figure 8b) C=C could be associated to nanotubes 551

Two distinct C1rsquo sites (grey Figure 8b) also indicate two configurations of the sugar headgroup 552

A comment on this point will be given in the following paragraphs 553

The local (Aring-scale) variation of molecular mobility below Tm2 drives the coexistence of 554

two distinct molecular environments of the same G-C181-OH molecule This occurs within a 555

single-phase domain (as deduced by WAXS) of equivalent membrane rigidity (as derived from 556

Laurdan probe in confocal microscopy) Although never reported before the difference in the 557

local mobility of the same amphiphile (here G-C181-OH) could satisfy the theoretically-558

140 120 100 80 60 40 20

CP(3 ms)-70degC

SP-70degC

SP-90degC

SP-110degC

ppm

LiquidC910

C1rsquo

C2rsquo

C4rsquo

C3rsquo

C1

C6rsquoC5rsquo

C1

6

C8

11

C2 C

H2

C1

5

CH

2

DM

SO

13C

22

required conditions that induce local lipid segregation and formation of membrane inclusions 559

(Figure 8c) In case of their non-zero average orientation the local membrane curvature 560

increases62 and nanotubes can spontaneously form (Figure 8bc) More intriguing this 561

mechanism occurs at least 10degC above Tm1 although it should not be favored above the Tm27 562

From both experimental and theoretical backgrounds the conditions of topologically-563

connected vesicles-nanotubes systems are met only when either internal (two components) or 564

external driving forces are applied19 In the present single-lipid system of which the phase 565

behavior is summarized in Figure 8a the most plausible driving force could be explained by 566

the NMR results collected on the dry sample G-C181-OH concomitantly occupies two 567

molecular environments characterized by 1) two distinct conformations of the sugar 568

displaying equivalent mobility and 2) two conformations of the acyl chain unsaturation 569

570

23

571

Figure 8 ndash a) Thermotropic phase behavior of G-C181-OH at 5 mgmL-1 Note that Tm2 is observed on the 572

dry powder only b) Possible relationship between the molecular environment as derived from NMR and 573

membrane curvatures All the molecular drawings refer to the same G-C181-OH molecule (bulky ellipsoid 574

equiv sugar headgroup) while each color corresponds to a given configuration Grey of C1rsquo1 and C1rsquo2 indicates 575

the uncertainty about the assignment of the specific NMR peak The curvature C is defined as 1R R being 576

the curvature radius R1 R2 radius of curvature CG Gauss curvature CM Mean curvature c) Free 577

interpretation of the anisotropic high-curvature (CMgt 0 CG= 0) inclusions in the membrane (low CM and 578

CG) continuum driven by inter-sugar H-bonding and reduced intramembrane diffusion for a G-C181-OH 579

glucolipid 580

581

Flat crystalline structure

Nanotube-vesicle network

Vesicles

48degC

120deg

106 105 104 103 102 101

ppm

H9 H10H1rsquo

106 105 104 103 102 101

ppm106 105 104 103 102 101

ppm106 105 104 103 102 101

ppm

C=CC=C

C1rsquo1C1rsquo2

C1rsquo

Tm2(96degC)Tm1

a)

b)

c)

Same molecule different local orientations and sugar conformations

High curvature membrane

Low curvature inclusion

Flip-FlopStrong H-bonding

CG

CM ( + )

CG 0

CM 0 CG 0

CM 0

G-C181-OH

24

Two related questions are unanswered at the moment What actually drives nanotube 582

formation and is the organization of G-C181-OH within the vesicles and tubes the same Lipid 583

nanotubes have been studied for at least three decades over which broad experimental and 584

theoretical insight has been generated Despite this fact the actual understanding is still matter 585

of debate27 Micelle-to-nanotube and vesicle-to-nanotube phase transitions are generally 586

reported for single amphiphiles below a given Tm and intermediate phase composed of twisted 587

and flat fibers is concomitantly reported Initially thought to be a chirality-driven phenomenon 588

recent studies propose a more general mechanism involving crystallization and symmetry 589

breaking in the lipid packing driven by optimal packing27 If twisted ribbons were observed 590

before for a number of microbial glycolipid bolaamphiphiles44 they are not found for G-C181-591

OH This fact as well as the large body of cryo-TEM images and the observation of nanotubes 592

above Tm1 rather pledge in favour of a direct topological change from positive (vesicles) to 593

zero (nanotubes) Gaussian curvatures In this case vesicle-to-nanotube transitions could be 594

driven by external forces56307813ndash162829 595

Considering the absence of external forces in the present system there are only two 596

possibilities left spontaneous budding of the membrane or nanotube pulling after inter-vesicle 597

collisions due to Brownian motion Both are possible as inclusions composed of a different 598

arrangement of the same molecule in the membrane continuum can theoretically explain 599

spontaneous nanotube formation in an isotropic medium672223629ndash1218ndash21 However the second 600

one is probably more likely to prevail because it could explain the large number of nanotubes 601

connecting two distant vesicles Interestingly similar events occur for any phospholipid-based 602

vesicular solution but nanotubes are extremely rare unless at least two lipids of different 603

molecular structure are mixed together The spectroscopic 1H and 13C and relaxation (1H) 604

ssNMR data even if collected on a dry powder demonstrate that the coexistence of two G-605

C181-OH configurations involving both the sugar headgroup and fatty alcohol chain are 606

possible for this molecule As a matter of fact this is not so surprising as more than one energy 607

minima each corresponding to a given conformation are known for glucose its non-reducing 608

derivative methylglucose and carbohydrates in general78ndash80 609

The presence of two configurations could explain the coexistence between two 610

structures of different curvature but the data presented so far do not explain the origin of these 611

two configurations In other words what drives the change in the local structure of the sugar 612

headgroup and fatty alcohol within the membrane 613

The association of known self-assembled structures to spectroscopic 13C ssNMR data is 614

known for years in particular for gluconamides lipids having a linear sugar headgroup81ndash84 It 615

25

was found that gauche effects in some cases up to the γ position in the sugar chain induce 616

upfield chemical shifts while anti and trans conformations result in downfield shifts Each 617

structure (monolayer crystal bilayer crystal or micellar fiber) was then characterized by a given 618

organization of the gluconamide and a characteristic 13C ssNMR fingerprint which could be 619

explained by gauche γ-gauche anti and trans conformations of the headgroup Similar 620

structural-spectroscopic correlations have been collected over the years for much simpler cyclic 621

sugars85 whereas pyranoses can undergo important conformational changes (chair half-chair 622

boat skew) when the appropriate energy barriers are crossed78ndash8086 A quantitative 623

conformational study based on the 13C chemical shift variations shown in Table 1 would be 624

certainly helpful but very delicate in the context of the present state of the art because a 625

rigorous approach requires at least one known crystal structure8182 which is currently missing 626

for this compound A qualitative approach will then be adopted 627

G-C181-OH is a bolaform amphiphile and flip-flop could be a possible mechanism 628

involved in the nanotube stabilization whereas head-tail interactions have long been considered 629

as important in nanotube-forming bolaform amphiphiles6063 From mere considerations based 630

on T2 (1H) relaxation arguments one could exclude head-tail configurations if the R-CH2OH 631

endgroup (H1) was involved in a direct interaction with the sugar group its corresponding T2 632

would probably be comparable to the sugarrsquos one (OH and H1rsquo) but this does not seem to be 633

the case However this hypothesis would be based on the comparison of the relaxation from 634

CH and CH2 protons additionally involved in different kind of motions Besides the relaxation 635

of H1 is influenced by the strong overlap between H1 and H5rsquo For these reasons relaxation 636

arguments are not solid enough and will not be used 637

We then address this question by exploring intermolecular interactions in a direct 638

manner using 2D 1H-1H homonuclear ssNMR correlation spectroscopy At 90degC in the fast-639

tumbling domain of molecular motion (narrow 1H NMR peaks Figure 6a) and corresponding 640

to the vesicle-rich region spin diffusion (NOESY) experiments are preferred over dipolar 641

coupling based experiments because such interaction is averaged out by molecular motion 642

However when the molecular motion is more constrained in the domain of nanotube stability 643

at 70degC (broad 1H spectra Figure 6a) Double-Quantum Single-Quantum (DQ-SQ) correlation 644

experiments are best suited to probe interaction between close molecular neighbors These 645

experiments are reported in Figure 9 and the related interpretation is only briefly reported 646

below while a more detailed discussion can be found on Page 12 in the Supporting Information 647

26

648

Figure 9 ndash a) 2D 1H-1H NOESY NMR experiment performed at MAS= 10 kHz (4 mm probe) on a Bruker 649

300 MHz Avance III HD spectrometer using 300 ms of mixing time Sample temperature is regulated to 650

90degC reached by cooling from 110degC b) 2D 1H-1H DQ-SQ (BAck-to-BAck BABA pulse sequence used for 651

the DQ excitation scheme) NMR experiment performed at MAS= 22 kHz (25 mm probe) on a Bruker 700 652

MHz Avance III spectrometer using a DQ excitation time equal to 2 rotor periods Sample temperature is 653

regulated to 70degC reached by cooling from 110degC The typical schemes illustrating how to interpret these 654

2D experiments are reported above each experimental map Zooms of the sugar region and configurational 655

interpretation are given below each 2D map More information are given in the text below 656

H16-H16

H1-H1

H9-H10

H1rsquo-H1

H1rsquoH910 H16

H1

H1rsquo-H1rsquo

Hxrsquo-Hxrsquo

2

(pp

m)

(ppm)

315

811910 1rsquo

2rsquo2

3rsquo 4rsquo6rsquo

5rsquo

1rsquo2910

asymp asymp asympHDO MeOD Acetone

1

315

811 9101rsquo

2rsquo2

3rsquo4rsquo6rsquo

5rsquo

1rsquo2 9 10

asympasympasympHDOMeODAcetone

H1rsquo-H1

1rsquo

1

(p

pm

)

(ppm)

H1-H2

H1rsquo-H1

H1rsquo

H910

H1H5rsquo

H2

H1rsquo-H5rsquo

315

811910 1rsquo

2rsquo2

3rsquo 4rsquo6rsquo

5rsquo

1rsquo2910

asymp asymp asympHDO MeOD Acetone

1

H2rsquoH3rsquoH4rsquo

H2rsquoH3rsquoH4rsquo

X= 2rsquo3rsquo4rsquo

NOESY - T= 90degC

1H

1H

1H

1H

DQ-SQ - T= 70degC

a) b)

2

(pp

m)

(p

pm

)

(ppm)

A B

AhellipB

AhellipA

BhellipB

NOESY

(ppm)

A B

AhellipB

DQ-SQ

Single-quantum dimension

Do

ub

le-q

ua

ntu

m d

imen

sio

n

AhellipA

BhellipB

BhellipA

BhellipA

315

811910 1rsquo

2rsquo2

3rsquo 4rsquo6rsquo

5rsquo

1rsquo2910

asymp asymp asympHDO MeOD Acetone

1

lt 5 Aring

lt 5 Aring

H1-H2

H811

H15

H16

H6rsquo

H1rsquo

H1

H5rsquo

H2

H1-H2

H1rsquo-H1

H1rsquo-H5rsquo

H1rsquo

H1 H5rsquo

H1rsquo

H1

H5rsquo

H1H2rsquo H3rsquo H4rsquo

H1-H1

H1rsquo-H1

Hxrsquo-HxrsquoX= 2rsquo3rsquo4rsquo

H1-H2

H1rsquo

Zoom sugar region 28 ndash 45 ppmNOESY DQ-SQ

27

Head-tail or head-head configurations could be probed by the existence of H1rsquo-H1 657

interactions whereas H1 (δH= 356 ppm) is associated to the glucolipid tail and H1rsquo (δH= 432 658

ppm) to the sugar headgroup These atoms are more than 15 Aring apart and a direct interaction 659

could not be observed in both NOESY and DQ-SQ experiments NOESY experiments (Figure 660

9a) show that H1rsquo-H1 correlations are not observed Meanwhile clear-cut direct H1rsquo-H1 661

interactions are indeed observed are detected at T= 70degC (DQ cross-peaks related to H1rsquo-H1 662

correlation in DQ-SQ experiment shown in Figure 9b) Interestingly the absence of auto peaks 663

at δH= 432 ppm and δH= 335 ppm in the SQ dimension excludes H1rsquo-H1rsquo and Hxrsquo-Hxrsquo (x= 2 664

3 4) interactions respectively strengthening the idea that sugar headgroups from two nearest-665

neighbor G-C181-OH molecules are not adjacent in the nanotube-vesicle region 666

Homonuclear 2D ssNMR supports the idea according to which G-C181-OH can 667

undergo molecular flip-flops and a head-tail arrangement in the dry product Extrapolating these 668

conclusions in the lower concentration range lack of H1rsquo-H1 proximity at T= 90degC suggests a 669

head-head configuration in the vesicle-rich region and head-tail configuration within the 670

nanotubes It is however still unclear whether flip-flops are driven by spontaneous temperature-671

induced conformational changes in the glucolipid packing stimulating inclusions within the 672

membrane itself672223629ndash1218ndash21 or by defects generated by inter-vesicle collisions In fact a 673

combination of both is not unlikely Nanotubes are not stable when temperature is 20 to 30 674

degrees above Tm1 indicating that inter-vesicle collisions alone cannot explain the unusual 675

phenomenon In an energy landscape approach87 the head-head configuration is certainly 676

favored at high temperature while both head-head and head-tail configurations could be 677

favored in the mid-temperature region corresponding to the range between Tm1 and Tm2 in the 678

dry powder Both statistical and defect-induced flip-flop events contribute to form high mean 679

zero Gaussian curvature morphologies (nanotubes) for which the head-tail conformation 680

limiting the proximity between bulky sugar headgroups is probably more adapted (Figure 8c) 681

The flip-flop interpretation now reasonably explains the strong differences in the sugar and fatty 682

alcohol conformations deduced by the multiple NMR signals attributed to chemically-683

equivalent groups The existence of various low-energy conformations each having specific 684

NMR signature for sugars in solution78ndash8086 and solid-state81ndash84 further supports our 685

interpretation 686

One last point should be clarified Why between Tm1 and Tm2 are two molecular 687

environments able to coexist instead of a single one of average rigidity as found in the case of 688

ldquoripplerdquo phases50ndash54 We believe that this question can be answered based on the NMR results 689

the relatively short T2 values of the H1rsquo and OH protons shows the constrained mobility of the 690

28

sugar headgroup despite a certain flexibility of the chain (long T2 component) This trend may 691

suggest restrictions in translational mobility probably due to the sugar-OH hydrogen bonding 692

(Figure 8c) meaning that lipids are not able to diffuse within the membrane thus keeping two 693

different configurations stable 694

695

Conclusions 696

Topologically-connected nanotube-vesicle lipidic systems are commonly observed for 697

self-assembled membranes composed of at least two amphiphiles and spontaneously driven by 698

anisotropic inclusions andor by external directional forces This work shows that strong 699

differences in membrane curvature can spontaneously occur in single-lipid systems without 700

external anisotropic forces Coexistence and interconnection between nanotubes and vesicles 701

are evidenced via cryo-TEM and fluorescence confocal microscopy in a single-phase region at 702

temperature above a transition temperature Tm1= 483degC determined by DSC on the hydrated 703

sample but below the second transition temperature Tm2 (957degC) determined by DSC on the 704

dry powder X-ray diffraction (WAXS) also shows the coexistence of two lattice periods one 705

attributed to an inter-membrane repeating distance and the other possibly related to the intra-706

nanotube glucolipid arrangement Labelling the systems with Laurdan a fluorophore of which 707

the emission strongly varies with the local membrane rigidity excludes the presence of rigid 708

micro-separated phases (eg Lβ) 709

1H and 13C ssNMR despite the fact that it can only be performed on the dry powder 710

confirm the presence of one main fluid G-C181-OH environment above Tm2 Below Tm2 the 711

splitting of some 13C peaks and the coexistence of short and long 1H relaxation components 712

demonstrate that G-C181-OH can simultaneously pack into structurally and dynamically 713

inequivalent environments Combining advanced 2D 1H-1H correlation experiments based on 714

spin diffusion or double-quanta excitation suggests that vesicles could contain a preferred head-715

head configuration of the bolaform lipid while nanotubes a preferred head-tail assembly 716

The presence of two environments composed of the same molecule characterized by 717

different molecular configurations and rigidity fulfills the theoretical conditions necessary to 718

spontaneously nucleate a membrane inclusion of non-zero average orientation described as 719

necessary to observe topologically-connected nanotube-vesicle structures in the absence of 720

external forces This hypothesis could also be enriched by the possibility of nanotubes to be 721

pulled away between two vesicles after a collision event 722

This work expands the actual theoretical and experimental framework on lipid networks 723

stabilized by a different radius of curvature It could also motivate the synthesis or isolation 724

29

of new lipids providing a higher level of 3D architectural control in soft networks possibly in 725

the range of room temperature 726

727

Acknowledgements 728

Dr Andrea Lassenberger and Dr Sylvain Preacutevost at Institut Laue Langevin (ILL Grenoble 729

France) are kindly acknowledged for their assistance on the SANS experiment Confocal 730

images were performed at the Institut de Biologie Paris Seine (IBPS) imaging facility Jean-731

Franccedilois Gilles (Imaging Core Facility Institut de Biologie Paris Seine (IBPS) CNRS 732

Sorbonne Universiteacute Paris France) is kindly acknowledged for his assistance on the confocal 733

microscopy imaging Dr Guillaume Laurent (LCMCP Sorbonne Universiteacute Paris France) is 734

kindly acknowledged for helpful discussions The French region Ile-de-France SESAME 735

program is acknowledged for financial support (700 MHz NMR 736

spectrometer) Dr Olivier Diat (ICSM Marcoule France) Prof eronikaIglič and Prof V Aleš737

onsul discussi Slovenia) are kindly acknowledged for helpfUniversity of LjubljanaIglič (-Kralj 738

739

Financial support 740

WAXS experiments have been supported by Soleil Light Source Saint Aubin France proposal 741

Ndeg 20201747 SANS experiments have been supported by ILL proposal Ndeg9-13-778 742

743

References 744

1 D Kabaso N Bobrovska W Goacuteźdź E Gongadze V Kralj-Iglič R Zorec and A 745

Iglič Bioelectrochemistry 2012 87 204ndash210 746

2 C R Safinya U Raviv D J Needleman A Zidovska M C Choi M A Ojeda-747

Lopez K K Ewert Y Li H P Miller J Quispe B Carragher C S Potter M W 748

Kim S C Feinstein and L Wilson Adv Mater 2011 23 2260ndash2270 749

3 P Veranič M Lokar G J Schuumltz J Weghuber S Wieser H Haumlgerstrand V Kralj-750

Iglič and A Iglič Biophys J 2008 95 4416ndash4425 751

4 A Rustom R Saffrich I Markovic P Walther and H-H Gerdes Science (80- ) 752

2004 303 1007ndash1010 753

5 M I Angelova S Soleacuteau P Meacuteleacuteard F Faucon and P Bothorel Prog Colloid 754

Polym Sci 1992 89 127ndash131 755

6 L Mathivet S Cribier and P F Devaux Biophys J 1996 70 1112ndash1121 756

7 V Kralj-Iglič G Gomišček J Majhenc V Arrigler and S Svetina Colloids Surfaces 757

A Physicochem Eng Asp 2001 181 315ndash318 758

30

8 V Kralj-Iglič A Iglič G Gomišček F Sevšek V Arrigler and H Haumlgerstrand J 759

Phys A Math Gen 2002 35 1533ndash1549 760

9 M Drab D Stopar V Kralj-Iglič and A Iglič Cells 2019 8 626 761

10 A Zidovska K K Ewert J Quispe B Carragher C S Potter and C R Safinya 762

Langmuir 2009 25 2979ndash2985 763

11 A Zidovska K K Ewert J Quispe B Carragher C S Potter and C R Safinya Soft 764

Matter 2011 7 8363ndash8369 765

12 A Zidovska K K Ewert J Quispe B Carragher C S Potter and C R Safinya 766

Biochim Biophys Acta - Biomembr 2009 1788 1869ndash1876 767

13 M Karlsson K Sott M Davidson A S Cans P Linderholm D Chiu and O Orwar 768

Proc Natl Acad Sci U S A 2002 99 11573ndash11578 769

14 A Karlsson R Karlsson M Karlsson A-S Cans A Stromberg F Ryttsen and O 770

Orwar Nature 2001 409 150ndash152 771

15 M Karlsson K Sott A S Cans A Karlsson R Karlsson and O Orwar Langmuir 772

2001 17 6754ndash6758 773

16 E Evans H Bowman A Leung D Needham and D Tirrell Science (80- ) 1996 774

273 933ndash935 775

17 U Seifert Configurations of fluid membranes and vesicles 1997 vol 46 776

18 J Kaumls and E Sackmann Biophys J 1991 60 825ndash844 777

19 N Bobrovska W Goacuteźdź V Kralj-Iglič and A Iglič PLoS One 2013 8 e73941 778

20 A Iglič B Babnik K Bohinc M Fošnarič H Haumlgerstrand and V Kralj-Iglič J 779

Biomech 2007 40 579ndash585 780

21 K Bohinc T Slivnik A Iglič M Brumen and V Kralj-Iglič J Phys Chem C 2007 781

111 9709ndash9718 782

22 A Iglič H Haumlgerstrand P Veranič A Plemenitaš and V Kralj-Iglič J Theor Biol 783

2006 240 368ndash373 784

23 V Kralj-Iglič and P Veranič Adv Planar Lipid Bilayers Liposomes 2006 5 129ndash785

149 786

24 A Iglic and V Kralj-Iglic Adv Planar Lipid Bilayers Liposomes 2006 4 253ndash279 787

25 S Penič L Mesarec M Fošnarič L Mroacutewczyńska H Haumlgerstrand V Kralj-Iglič and 788

A Iglič Front Phys 2020 8 342 789

26 P Yager and P E Schoen Mol Cryst Liq Cryst 1984 106 371ndash381 790

27 T G Barclay K Constantopoulos and J Matisons Chem Rev 2014 114 10217ndash791

10291 792

31

28 B Božič G Gomišček V Kralj-Iglič S Svetina and B Žekš Eur Biophys J 2002 793

31 487ndash496 794

29 A Iglič H Haumlgerstrand M Bobrowska-Haumlgerstrand V Arrigler and V Kralj-Iglič 795

Phys Lett A 2003 310 493ndash497 796

30 R Bar-Ziv and E Moses Phys Rev Lett 1994 73 1392ndash1395 797

31 S Chiruvolu H E Warriner E Naranjo S H Idziak J O Raumldler R J Plano J A 798

Zasadzinski and C R Safinya Science (80- ) 1994 266 1222ndash1225 799

32 V Kralj-Iglič A Iglič H Haumlgerstrand and P Peterlin Phys Rev E 2000 61 4230ndash800

4234 801

33 L Mesarec M Drab S Penič V Kralj-Iglič and A Iglič Int J Mol Sci 2021 22 802

2348 803

34 L Van Renterghem S L K W Roelants N Baccile K Uyttersprot M C Taelman 804

B Everaert S Mincke S Ledegen S Debrouwer K Scholtens C V Stevens W 805

Soetaert L Van Renterghem S L K W Roelants N Baccile K Uytersprot M C 806

Taelman B Everaert S Mincke S Ledegen S Debrouwer K Scholtens C V 807

Stevens and W Soetaert Biotechnol Bioeng 2018 115 1195ndash1206 808

35 J Clark and F Deswarte Introduction to Chemicals from Biomass Second Edition 809

John Wiley amp Sons 2015 810

36 K V Sajna R Houmlfer R K Sukumaran L D Gottumukkala and A Pandey in 811

Industrial Biorefineries and White Biotechnology eds A Pandey R Houmlfer M 812

Taherzadeh K M Nampoothiri and C Larroche Amsterdam Oxford Waltham 813

Elsevier 2015 pp 499ndash521 814

37 R Marchant and I M Banat Trends Biotechnol 2012 30 558ndash565 815

38 K Hill and C LeHen-Ferrenbach in Sugar-Based Surfactants Fundamentals and 816

Applications ed C C Ruiz Boca Raton CRC Press 2009 pp 1ndash20 817

39 L Rodrigues I M Banat J Teixeira and R Oliveira J Antimicrob Chemother 818

2006 57 609ndash618 819

40 J D Desai and I M Banat Microbiol Mol Biol Rev 1997 61 47ndash64 820

41 S L K W Roelants K Ciesielska S L De Maeseneire H Moens B Everaert S 821

Verweire Q Denon B Vanlerberghe I N A Van Bogaert P Van der Meeren B 822

Devreese and W Soetaert Biotechnol Bioeng 2016 113 550ndash559 823

42 N Baccile M Selmane P Le Griel S Preacutevost J Perez C V Stevens E Delbeke S 824

Zibek M Guenther W Soetaert I N A Van Bogaert and S Roelants Langmuir 825

2016 32 6343ndash6359 826

32

43 N Baccile A-S Cuvier S Preacutevost C V Stevens E Delbeke J Berton W Soetaert 827

I N A Van Bogaert and S Roelants Langmuir 2016 32 10881ndash10894 828

44 N Baccile C Seyrig A Poirier S A Castro S L K W Roelants and S Abel Green 829

Chem 2021 23 3842ndash3944 830

45 A Salis L Cappai C Carucci D F Parsons and M Monduzzi J Phys Chem Lett 831

2020 11 6805ndash6811 832

46 J Schindelin I Arganda-Carreras E Frise V Kaynig M Longair T Pietzsch S 833

Preibisch C Rueden S Saalfeld B Schmid J-Y Tinevez D J White V 834

Hartenstein K Eliceiri P Tomancak and A Cardona Nat Methods 2012 9 676ndash835

682 836

47 T Parasassi and E Gratton J Fluoresc 1995 5 59ndash69 837

48 M M Dodes Traian F L Gonzaacutelez Flecha and V Levi J Lipid Res 2012 53 609ndash838

616 839

49 D M Owen C Rentero A Magenau A Abu-Siniyeh and K Gaus Nat Protoc 840

2012 7 24ndash35 841

50 W Stillwell in An Introduction to Biological Membranes Academic Press 2016 pp 842

135ndash180 843

51 W Stillwell in An Introduction to Biological Membranes Academic Press 2016 pp 844

181ndash220 845

52 B Tenchov Chem Phys Lipids 1991 57 165ndash177 846

53 R Koynova and M Caffrey Biochim Biophys Acta 1998 1376 91ndash145 847

54 F Schmid S Dolezel O Lenz and S Meinhardt J Phys Conf Ser 2014 487 848

012004 849

55 S Meinhardt R L C Vink and F Schmid Proc Natl Acad Sci U S A 2013 110 850

4476ndash4481 851

56 O Glatter and O Kratky Small Angle X-ray Scattering Academic Press London 852

1982 853

57 B A Cunningham A D Brown D H Wolfe W P Williams and A Brain Phys 854

Rev E 1998 58 3662ndash3672 855

58 M Nadler A Steiner T Dvir O Szekely P Szekely A Ginsburg R Asor R Resh 856

C Tamburu M Peres and U Raviv Soft Matter 2011 7 1512ndash1523 857

59 C Tanford The hydrophobic effect Formation of micelles and biological membranes 858

John Wiley amp Sons Inc New York 1973 859

60 M Masuda and T Shimizu Langmuir 2004 20 5969ndash5977 860

33

61 A Zidovska K K Ewert J Quispe B Carragher C S Potter and C R Safinya 861

Block Liposomes Vesicles of Charged Lipids with Distinctly Shaped Nanoscale 862

Sphere- Pear- Tube- or Rod-Segments Elsevier Inc 1st edn 2009 vol 465 863

62 A Iglic V Kralj-Iglic B Božič M Bobrowska-Haumlgerstrand B Isomaa and H 864

Haumlgerstrand Bioelectrochemistry 2000 52 203ndash211 865

63 T Shimizu M Masuda and H Minamikawa Chem Rev 2005 105 1401ndash1444 866

64 J V Selinger and J M Schnur Phys Rev Lett 1993 71 4091ndash4094 867

65 J M Schnur Science (80- ) 1993 262 1669ndash1676 868

66 B Tenchov R Koynova and G Rapp Biophys J 2001 80 1873ndash1890 869

67 R Koynova and B Tenchov Ser Biomech 2017 31 3ndash19 870

68 S J Angyal and R Le Fur Carbohydr Res 1984 126 15ndash26 871

69 J K Whitesell and M A Minton J Am Chem Soc 1987 109 225ndash228 872

70 T Takayama S Ando and I Ando J Mol Struct 1990 220 245ndash250 873

71 K Gawrisch N V Eldho and I V Polozov Chem Phys Lipids 2002 116 135ndash151 874

72 F Tielens C Gervais J F Lambert F Mauri and D Costa Chem Mater 2008 20 875

3336ndash3344 876

73 G A Jeffrey and Y Yeon Acta Crystallogr Sect B 1986 42 410ndash413 877

74 B Berglund and R W Vaughan J Chem Phys 1980 73 2037ndash2043 878

75 C Gervais C Coelho T Azaiumls J Maquet G Laurent F Pourpoint C Bonhomme 879

P Florian B Alonso G Guerrero P H Mutin and F Mauri J Magn Reson 2007 880

187 131ndash140 881

76 C Akita T Kawaguchi F Kaneko H Yamamoto and M Suzuki J Phys Chem B 882

2004 108 4862ndash4868 883

77 R N Purusottam L Seacutenicourt J J Lacapegravere and P Tekely Biochim Biophys Acta - 884

Biomembr 2015 1848 3134ndash3139 885

78 X Biarneacutes A Ardegravevol A Planas C Rovira A Laio and M Parrinello J Am Chem 886

Soc 2007 129 10686ndash10693 887

79 L M J Kroon‐Batenburg and J Kroon Biopolymers 1990 29 1243ndash1248 888

80 K N Kirschner and R J Woods Proc Natl Acad Sci U S A 2001 98 10541ndash889

10545 890

81 S Svenson B Kirste and J H Fuhrhop J Am Chem Soc 1994 116 11969ndash11975 891

82 I Sack S Macholl J H Fuhrhop and G Buntkowsky Phys Chem Chem Phys 892

2000 2 1781ndash1788 893

83 S Svenson J Koening and J H Fuhrhop J Phys Chem 1994 98 1022ndash1028 894

34

84 S Svenson A Schaefer and J H Fuhrhop J Chem Soc Perkin Trans 2 1994 2 895

1023ndash1028 896

85 K Bock and C Pedersen Adv Carbohydr Chem Biochem 1983 41 27ndash66 897

86 T B Grindley Structure and Conformation of Carbohydrates Springer Berlin 898

Heidelberg 2008 899

87 F Tantakitti J Boekhoven X Wang R V Kazantsev T Yu J Li E Zhuang R 900

Zandi J H Ortony C J Newcomb L C Palmer G S Shekhawat M O De La 901

Cruz G C Schatz and S I Stupp Nat Mater 2016 15 469ndash476 902

903

1

Supporting Information 904

905

Topological connection between vesicles and nanotubes in single-906

component lipid membranes driven by head-tail interactions 907

908

Niki Baccilea Ceacutedric Lorthioira Abdoul Aziz Baa Patrick Le Griela Cristina Coelhob Javier 909

Perezc Wim Soetaertd Sophie L K W Roelantsd 910

911

a Sorbonne Universiteacute Centre National de la Recherche Scientifique Laboratoire de Chimie de 912

la Matiegravere Condenseacutee de Paris LCMCP F-75005 Paris France 913

b Institut des Mateacuteriaux de Paris Centre Sorbonne Universiteacute Paris France 914

c Synchrotron Soleil LrsquoOrme des Merisiers Saint-Aubin BP48 91192 Gif-sur-Yvette Cedex 915

France 916

d InBio Department of Biotechnology Ghent University Ghent Belgium 917

918

919

Corresponding author 920

Dr Niki Baccile 921

E-mail address nikibaccilesorbonne-universitefr 922

Phone +33 1 44 27 56 77 923

924

925

926

2

a) 927

Parameter Method

Dry Matter (DM ) 996 Infrared balance 105degC

Glucose () 012 HPLC-Metacarb

Glycerol () nd HPLC-Metacarb

Free Fatty Acid content () lt001 Internal method BBEPP

GC with FID Detector

Oil () lt001 Internal method BBEPP

GC with FID Detector

Moulds (CFUg) lt10 3M-Nordval ndeg16 B

Yeasts (CFUg) lt10 3M-Nordval ndeg16 B

Anaerobic count (CFUg) lt10 SP-VG M005 B

Protein () tbd BCA protein assay

928

Figure S 1 - a) HPLC-ELSD chromatogram and composition table 929

930

931

932

933

3

b) 934

935

Group Ndeg δ ppm

CH2 1 354

CH2 2 153

CH2 3-712-14 132

CH2 811 203

CH 910 534

CH2 15 162

CH2 16 388

CH 1 424

CH 2 316

CH 3 334

CH 4 329

CH 5 354

CH2 6 366

Figure S 1 (next) ndash b) 1H NMR spectrum (MeOD-d4) recorded on the purified G-C181-OH compounds 936

and related assignment 937

938

2

811910 1rsquo

5rsquo1

3rsquo4rsquo

1rsquo1910

asymp asymp asympHDO MeOD Acetone

6rsquo2rsquo

1615

3-712-14

4

939

940

Figure S 2 ndash Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 130degC 941

5

942

943

Figure S 3 - Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 90degC 944

945

90degC

1

22

6

946

947

948

949

Figure S 4 ndash Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 70degC 950

951

21

11

1

2

5

3

4

45

5

6

7

7

7

952

953

954

Figure S 5 ndash Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 25degC 955

956

957

1

1 1

8

958

Figure S 6 ndash a) SANS experiment recorded at 90degC for a 5 mgmL G-C181-OH aqueous solution b-c) 959

WAXS experiments recorded on aqueous solutions of G-C181-OH at b) 5 mgmL and c) 25 mgmL d) 960

Structural interpretation of the repeating distances pointed out in c) 961

962

963

964

001 01 1

q Aring-1

25degC

68degC

88degC

111degC

01 1q Aring

-1

02 04 06 08 1q Aring

-1

001 01

10-1

100

101

Inte

nsity

cm

-1

q Aring-1

-2

T= 90degC

SANS WAXS WAXS

a) b) c)

dv

dL d= 144 Aring-1

di

ddH2O

dt

dvdL

dtdi d

d)

9

965

966

Figure S 7 ndash ssNMR MAS (10 kHz) experiments recorded on a G-C181-OH dry powder a) 1H spectra 967

recorded from 110degC to 70degC b) 13C single pulse (SP) spectra recorded with high power 1H decoupling from 968

110degC to 70degC c) 13C cross polarization (CP) spectra recorded at 110degC and contact time tc of 50 micros and 969

3000 micros d) 13C CP spectra recorded at 70degC and for 20 lt tc micros lt 3000 970

971

1

811

910

CH2 CH2

1rsquo

6rsquo

8 6 4 2 0 ppm

110

90

80

70

910

1rsquo

6rsquo16

1

2rsquo3rsquo4rsquo

811

OH

45 40 35 30 25 20 15

ppm

200 180 160 140 120 100 80 60 40 20 0

ppm140 120 100 80

ppm

CH2-gauche

3004

110

90

80

70

65

60

50

25

C91013003

C9 - 12869

CH2-all-trans

3198

C1rsquo10335

a) b)

1H

T degCT degC

13C Single Pulse MAS

A

B

A

B

70degC70degC

5rsquo

215

200 180 160 140 120 100 80 60 40 20 0

ppm

13C CPMAST= 110degC

200 180 160 140 120 100 80 60 40 20 0

ppm

132 131 130 129 128 127 126

ppm

106 105 104 103 102

ppm

13C CPMAST= 70degC

50

tc micros

3000

20

tc micros

50 500 1000 3000

tc micros

C102

13024 C910113003

C92

12869

C1rsquo213082

C1rsquo110443

A B

106 105 104 103 102

ppm

3000

1000

500

50

20

AB

c)

d)

10

972

Figure S 8 ndash a) Short and d) long T2 components displayed by the 1H transverse relaxation signal c) 1H 973

transverse relaxation signals recorded on a G-C181-OH dry powder from 110degC to 70degC The area of each 974

given resonance is plotted (log-lin) as a function of the echo time (CPMG pulse sequence) for each 975

temperature and fitted using the equation given above T2s and T2l indicate the short and long T2 relaxation 976

times respectively obtained by describing the 1H transverse relaxation signal using a biexponential decay 977

For H1rsquo H5rsquo and H6rsquo the 1H relaxation function could be described by a monoexponential decay 978

- Protons from the sugar group (H1rsquo H5rsquo H6rsquo) show a single T2 relaxation component (Figure 979

6c and Figure S 8) while protons of the acyl chain (CH2 H910 H811) show both short (Figure 980

6c) and long (Figure 6d) T2 relaxation component 981

- The short T2 relaxation time systematically decreases from 110degC to 70degC for all species by a 982

factor ranging from two to seven depending on the specific chemical group Very long T2 are 983

observed at 110degC for all acyl species while nearly unchanged values are measured below 984

90degC without any clear trend 985

- The sugar protons (H1rsquo H5rsquo H6rsquo) have the shortest set of T2 values (lt 1 ms) below 100degC 986

The long component of T2 generally ranges between 4 ms and 12 ms 987

0 1 2 3 4 5 6 7 8

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1 2 3 4 5 6 7 8

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

H1

H8H11

H9H10

CH2

H1rsquo

H6rsquo

CH2

H910H811 H1CH2

H1rsquo H5rsquo6rsquo

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

OH

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45 T= 110degC

T= 90degC

T= 80degC

T= 70degC

Log

In

ten

sity

Time ms

Fit

)

CH2 H910 H811 H1 H56 H1 OH

00

05

10

15

20

25

30

35

40

T2 m

s

T= 110degC

T= 90degC

T= 80degC

T= 70degC

CH2 H910 H811 H1

0

2

4

6

8

10

12

T2 m

s

T= 110degC

T= 90degC

T= 80degC

T= 70degC

Short T2(1H) Long T2(1H)

a) b)

c)

11

Table S 1 ndash 1H chemical shifts of G-C181-OH in solution and in the solid-state Solid-state spectra are 988

recorded at a MAS frequency of 10 kHz in DMSO-d6 broad peak 989

990

δH ppm

Group Ndeg Liquid

(MeOD-d4) Solid

(T= 110degC) Solid

(T= 70degC)

CH2 H1 354 357

CH2 H2 153 156

CH2 H3-712-14 132 135

CH2 H811 203 203

CH H910 534 534

CH2 H15 162 164

CH2 H16 388 383

CH H1 424 432

CH H2 316 335

CH H3 334 335

CH H4 329 335

CH H5 354 351 357

CH2 H6 366 383

OH OH1 432

45 463

OH OH2 492

OH OH3 490

OH OH4 487

OH OH6 445

991

992

H1H9H10H1rsquo

H6rsquo

OH2rsquo

OH1

12

Detailed discussion about the NOESY and DQ-SQ 2D 1H-1H ssNMR experiments presented in 993

Figure 9 994

995

2D 1H-1H ssNMR experiments are a unique tool to explore through-space proximities 996

between 1H nuclei ssNMR offers a large number of possibilities each adapted to a given 997

system The advantage of 2D maps with respect to a dynamics approach is their direct 998

readability in clear two nuclei which exchange their magnetization through-space (whichever 999

mechanism may be responsible for this exchange) will display a specific cross-peak However 1000

the way cross-peaks are presented and interpreted strictly depends on the nature of the pulse 1001

program which in turn is chosen according to the physical nature of the sample 1002

1D temperature-dependent 1H NMR experiments performed on G-C181-OH (Figure 1003

4a in the main text) indicate sharp peaks at 90degC and broad peaks at 70degC Narrow peaks 1004

indicate fast andor large-amplitude molecular tumbling averaging the strong 1H-1H dipolar 1005

interactions This excludes the use of pulse sequences like DQ-SQ which is not efficient under 1006

these conditions For this reason we employ spin-diffusion-based pulse sequences like 1007

NOESY which is more adapted for samples with high molecular mobility When mixing times 1008

are chosen to be short (lt 1 s) one generally explores through-space proximities within about 5 1009

10 Aring On the contrary when molecular motions are restricted dipolar coupling becomes 1010

significant 1H NMR peaks are broader and pulse sequences like DQ-SQ can now be used 1011

with a spatial resolution of less than 5 Aring 1012

Whenever possible DQ-SQ experiments are preferred over spin diffusion-based pulse 1013

sequences because they are more selective in particular for likewise interactions which are 1014

not discriminated in NOESY-type experiments This is illustrated by the typical interaction 1015

schemes for NOESY (top Figure 9a) and DQ-SQ (top Figure 9b) For two given A and B 1016

protons 2D NOESY maps will display two on-diagonal and two square-shaped symmetrical 1017

off-diagonal cross-peaks Off-diagonal cross-peaks illustrate the mutual A-B and B-A 1018

interaction but the on-diagonal peaks are systematic and do not prove any A-A or B-B 1019

interaction Similarly to NOESY 2D DQ-SQ maps display off-diagonal peaks (symmetric 1020

across the diagonal top Figure 9b) upon A-B and B-A interaction but in contrast they will 1021

display on-diagonal peaks only in the presence of self-interaction In the tutorial example of 1022

Figure 9b top the on-diagonal A-A peak indicates mutual A-A interaction while lack of on-1023

diagonal B-B peak evidences the absence of B-B proximities Combining NOESY and DQ-SQ 1024

experiments is capital to study the local arrangement of G-C181-OH and its evolution with 1025

temperature 1026

13

At T= 90degC the H1 and H5rsquo contributions are very close but possible to discriminate 1027

them (δ(H1)= 354 ppm δ(H5rsquo)= 351 ppm) according to 1H NMR chemical shifts in Table S 1028

1 2D 1H-1H NOESY in Figure 9a also discriminates between the two components attributed 1029

on the basis of the interactions with their mutual neighbors (green H1-H2 and H5rsquo-H1rsquo squares 1030

in Figure 9a) On this basis 2D NOESY does not show a direct H1rsquo-H1 correlation (blue red-1031

crossed) but rather an extensive network of intra-sugar interactions These features exclude 1032

suggest intra- and inter-sugar interactions and they exclude head-tail arrangements as a major 1033

molecular configuration at 90degC 1034

At 70degC the 1H NMR peaks are broader indicating a reduction of the T2 (1H) relaxation 1035

times andor residual homonuclear dipolar interactions excluding the use of NOESY In this 1036

case 1H-1H DQ-SQ correlation experiments are best suited Figure 9b presents the 1H-1H DQ-1037

SQ experiment on G-C181-OH at 70degC Peaks are broader than in NOESY and attribution may 1038

be cumbersome However the experiment is performed at 700 MHz and MAS= 22 kHz which 1039

improves spectral resolution with respect to the same experiment performed at 300 MHz and 1040

MAS= 10 kHz The auto-peak at δH= 356 ppm in the SQ dimension is attributed to H1-H1 spin 1041

pairs mainly resulting from intra CH2 interaction while the DQ cross-peaks at δH= 156-356 1042

ppm and δH= 432-356 ppm respectively origin from H2-H1 and H1rsquo-H1 interactions (Figure 1043

9b) DQ-SQ map does not show any auto-peak corresponding to H1rsquo-H1rsquo or Hxrsquo-Hxrsquo (x= 2rsquo 1044

3rsquo 4rsquo) interactions (red-crossed blue Figure 9b) thus excluding any proximity (lt 5 Aring) between 1045

close glucose neighbors On the contrary DQ-SQ maps (recorded both at 700 MHz Figure 9b 1046

or at 300 MHz not shown) systematically show DQ cross-peaks which can only be attributed 1047

to H1rsquo-H1 interactions (δH= 432-356) The presence of these H1rsquo-H1 interactions can only be 1048

explained by a head-tail arrangement of the G-C181-OH molecules as illustrated below the 1049

DQ-SQ map in Figure 9b 1050

4

85

86

87

88

89

Figure 1 ndash Non acetylated C181 alcohol glucoside G-C181-OH is obtained by a bioprocess performed 90

with modified S bombicola yeast 91

92

Experimental Section 93

Synthesis of non acetylated C181 Alcohol Glucosides (G-C181-OH) G-C181-OH (Mw= 94

41856 gmol-1) was produced by aerobic whole cell bioprocess with a modified S bombicola 95

strain as described by Van Renterghem et al (Fig S4 in Ref 34) The molecule was purchased 96

from the Bio Base Europe Pilot Plant (Gent Belgium) and has the generalized chemical 97

structure given in figure Figure 1 The HPLC and 1H NMR spectrum (MeOD-d4) with peak 98

assignment are shown in Figure S 1 High purity levels (99) and high degree of uniformity 99

were obtained as can be derived from HPLC-ELSD chromatogram 1H NMR and table of 100

contaminant given in Figure S 1 101

102

1H solution Nuclear Magnetic Resonance (NMR) 1H solution NMR experiments were 103

performed on a Bruker Avance III 300 spectrometer using a 5 mm 1H-X BBFO probe using 104

methanol-d4 as solvent The number of transients is 8 with 3 s recycling delay an acquisition 105

time of 546 s and a receiver gain of 362 The 1H NMR spectrum and relative assignment are 106

shown in Figure S 1 and Table S 1 13C solution NMR were performed on the same probe using 107

DMSO-d6 Corresponding chemical shifts are given in Table 1 Referencing is done with 108

respect to TMS δ1H= 0 ppm δ13C= 0 ppm 109

110

Sample preparation The sample was dissolved in milliQ-grade water at the concentration of 5 111

mgmL (05 wt) Lack of pH-sensitive probes (eg COOH groups) in G-C181-OH as 112

otherwise found in other microbial amphiphiles44 but also the will to avoid ion-specific 113

effects45 exclude the use of buffer The solution was heat at the given temperature using a CH3-114

150 Combitherm-2 dry block heating device For the cryo-TEM experiments the vials were 115

heat at the given temperature next to the cryofixation device and plunged into liquid ethane as 116

fast as possible For temperatures of 25degC and 70degC the sample was heat at 90degC and then 117

Synthesis path

5

cooled at to the desired temperature For the SAXS experiment the sample was heat at 90degC in 118

the block heating device and cooled at to 25degC before analysis For the SANS experiment the 119

sample was heat at 90degC directly in front of the beam using the thermalized sample holder 120

available 121

122

Differential Scanning Calorimetry (DSC) DSC was performed using a DSC Q20 apparatus 123

from TA Instruments equipped with the Advantage for Q Series Version acquisition software 124

(v540) Acquisition was performed on both hydrated and dry powder sample (~ 3-5 mg) sealed 125

in a classical aluminium cup and using an immediate sequence of heating and cooling ramps at 126

a rate of 10degCmin-1 Melting temperatures Tm 1 and 2 Tm1 and Tm2 were taken at the minimum 127

of the endothermic peak 128

129

Small and Wide Angle Scattering experiments Small angle neutron scattering (SANS) 130

experiments were performed at the D11 beamline of Institut Laue Langevin (Grenoble France) 131

during the run No 9-13-778 Four q-ranges have been explored and merged using the following 132

wavelengths λ and sample-to-detector (StD) distances 1) ultra-low q λ= 135Aring StD= 39 m 133

2) low-q λ= 53Aring StD= 39 m 3) mid-q λ= 53Aring StD= 8 m 4) high-q λ= 53Aring StD= 14 m 134

The sample (C= 5 mgmL-1) was prepared in 999 D2O to limit the incoherent background 135

scattering The sample solution was analyzed in standard 1 mm quartz cells Direct beam empty 136

quartz cell H2O (incoherent scatterer) within the quartz cell were recorded and boron carbide 137

(B4C) was used as neutron absorber The sample acquisition was measured at 90degC where 138

temperature was controlled through the controller thermalized sample holder available at the 139

beamline The background sample (D2O) signal was subtracted from the experimental data 140

Absolute values of the scattering intensity were obtained from the direct determination of the 141

number of neutrons in the incident beam and the detector cell solid angle The 2D raw data were 142

corrected for the ambient background and empty cell scattering and normalized to yield an 143

absolute scale (cross section per unit volume) by the neutron flux on the samples The data were 144

then circularly averaged to yield the 1D intensity distribution I(q) The software package Grasp 145

(developed at ILL and available free of charge) was used to integrate the data while the 146

software package SAXSUtilities (developed at ESRF and available free of charge) was used to 147

merge the data acquired at all configurations and subtract the background 148

Wide-angle X-ray scattering (WAXS) was performed under temperature control at the 149

SWING beamline of SOLEIL synchrotron facility (Saint-Aubin France) during the run 150

20201747 (energy 14 keV sample-to-detector distance 05 m) Two sample concentrations in 151

6

H2O were probed C= 5 mgmL-1 and C= 25 mgmL-1 The 2D data were integrated azimuthally 152

at the beamline using the software Foxtrot and in order to obtain the I(q) vs q spectrum after 153

masking the beam stop shadow Silver behenate (d(100) = 5838 Aring) was used as a standard to 154

calibrate the q-scale Sample solutions were inserted in borosilicate capillaries of 15 mm in 155

diameter Capillaries were flame-sealed A capillary oven with controlled temperature (plusmn 05degC) 156

provided at the beamline was used to control the sample temperature between 25degC and 111degC 157

Data were normalized by the transmission and calibrated to the SAXS signal of H2O at large q-158

values (I= 00163 cm-1) in order to obtain an absolute intensity scale The water signal was 159

measured by subtracting the signal of the empty capillary from the signal of a water-filled 160

capillary The signal of (water + capillary) was used as background for the samples and it was 161

subtracted after integration of the 2D data 162

163

Cryogenic transmission electron microscopy (cryo-TEM) Cryo-TEM experiments were carried 164

out on an FEI Tecnai 120 twin microscope operating at 120 kV and equipped with a Gatan 165

Orius CCD numeric camera The sample holder was a Gatan Cryoholder (Gatan 626DH 166

Gatan) Digital Micrograph software was used for image acquisition Cryofixation was 167

performed using a homemade cryofixation device The solutions were deposited on a glow-168

discharged holey carbon coated TEM copper grid (Quantifoil R22 Germany) Excess solution 169

was removed and the grid was immediately plunged into liquid ethane at -180degC before 170

transferring them into liquid nitrogen All grids were kept at liquid nitrogen temperature 171

throughout all experimentation Cryo-TEM images have been treated and analyzed using Fiji 172

(is just ImageJ) software available free of charge at the developerrsquos website46 173

The samplersquos concentration was C= 5 mgmL-1 For technical reasons the temperature 174

during cryofixation could not be controlled For this reason the temperature associated with 175

each cryo-TEM image in the text was related to the equilibration temperature just before 176

cryofixation the device being as close as possible to the dry bath used for temperature control 177

Since temperature control during cryofixation was not possible in our device and practically 178

impossible at high temperatures one should take the T values as indicative of the given value 179

Although the sample transfer between the dry bath and the cryofixation device was performed 180

as fast as possible it is more than likely that the actual temperature of the sample at the moment 181

of cryofixation was lower than the one indicated here although impossible to know precisely 182

183

7

Fluorescence microscopy Images were recorded using a 40x objective on a Leica SP5 upright 184

confocal microscope with 405 nm laser excitation The sample was marked with 6-dodecanoyl-185

NN-dimethyl-2-naphthylamine (Laurdan Mw= 35354 gmol-1) fluorophore as follows A 186

stock solution of 3 mg of Laurdan in 100 microL of acetone was initially prepared (C= 30 mgmL-187

1) A volume of 028 microL of the stock Laurdan solution was diluted in 1 mL of a 5 mgmL-1 G-188

C181-OH water solution temporary set at 110degC in a sealed Eppendorf The latter was 189

important so to enhance inclusion of Laurdan in the membrane constituted by the glucolipid 190

while in its vesicle phase The final Laurdan concentration was approximately 24 micromol with a 191

molar Laurdan-to-G-C181-OH molar ratio of approximately 1-to-500 Fluorophore-to-lipid 192

ratios above 1200 were generally considered as non-interfering with the lipid phase behavior 193

as also verified in this work 194

Laurdan can be excited between 365 nm and 410 nm and its emission wavelength can 195

vary according to the rigidity of the membrane intercalation within a rigid or flexible 196

membrane generates emissions at 440 nm or 490 nm respectively47 In the present experiment 197

we have flame-sealed a solution of G-C181-OH at 5 mgmL-1 containing Laurdan (1-to-500 in 198

lipid-to-Laurdan ratio) in flat optical capillaries of 01 mm thickness Capillaries were set on a 199

Linkam support for temperature control eventually placed under the microscope The sample 200

was excited at 405 nm and detection was made in spectral mode with a 5 nm window sliding 201

every 5 nm from 412 nm to 602 nm in 16 bits which allows quantify the signal and compare 202

the different conditions Quantitative analysis was performed according to the literature4849 203

several ROI (range of interest) have been drawn on several images using the ROI manager of 204

Fiji (is just ImageJ) software The integrated intensity of each ROI at 440 nm and 490 nm was 205

eventually recorded and used to calculate the value of the generalized polarization GP as 206

follows of 207

119866119875 =(120582440 minus 120582490)

120582440 + 120582490 208

A negative GP indicates a flexible membrane while a positive GP indicates a rigid membrane 209

210

Solid-state Nuclear Magnetic Resonance (ssNMR) 1H and 13C ssNMR experiments have been 211

performed using an Avance III HD Bruker 705 T (ν1H= 300 MHz) spectrometer and a 4 mm 212

magic angle spinning (MAS) probe except for the BAck-to-BAck (BABA) experiment which 213

was performed on an Avance III Bruker 1644 T (ν1H= 700 MHz) spectrometer equipped with 214

25 mm MAS probe Temperature was controlled through a Smart cooler BCUIIa BCU-xtreme 215

unit using 10 min of equilibration time after each temperature variation Temperature was 216

8

always varied from high (110degC) to low (70degC) before analysis Due to combined use of fast 217

MAS rates necessary to achieve high resolution spectra and heating close to the boiling point 218

of water neither diluted nor concentrated samples could be used for ssNMR experiments All 219

tests employing up to 50 wt G-C181-OH in water resulted in sample centrifugation drying 220

but also expulsion from the rotor causing sudden stop of rotor spinning For these reasons all 221

attempts to study hydrated G-C181-OH failed and only the dry powder could be analyzed The 222

dry sample powder was analyzed as such in zirconia rotors (4 mm or 25 mm) spinning at MAS= 223

10 kHz (4 mm) or 22 kHz (25 mm) and Chemical shifts were calibrated with respect to 224

adamantane (13C 13C(CH2)= 3848 ppm = 191 ppm) and glycine alpha (13C(CO)= 17603 225

ppm) Details on the acquisition parameters are given in the Supporting Information 226

227

Results and discussion 228

Topologically-connected nanotubes and vesicles 229

G-C181-OH is a biobased glucolipid water-insoluble at room temperature and of 230

which the physicochemical properties are poorly known The DSC thermogram performed at 231

10degCmin of G-C181-OH as a function of the water content (Figure 2a) shows two endothermic 232

peaks on the dry powder (996 wt glucolipid) They are attributed to two melting phenomena 233

occurring at the following temperatures Tm 483degC (Tm1) and 957degC (Tm2) the enthalpy of the 234

former being lower than that of the latter Interestingly DSC also shows that Tm1 is hydration 235

independent and reversible (the associated exothermic peak is systematically observed) while 236

the transition at Tm2 is simply suppressed (here tested up to 130degC) on hydrated powders and 237

never observed on the cooling profile even at slow cooling rates (1degCmin) The thermogram 238

recorded on the dry powder is typical of single-component lipids having an intermediate50 239

metastable phase (often referred to the ripple phase Pβ´ in reference to the periodic undulation 240

of the bilayer)50ndash54 between the classical lamellar gel (Lβ) and liquid crystalline (Lα) phases50ndash241

54 and characterized by regions of both liquid order and disorder5455 The effect of water does 242

not influence the low-temperature crystallization at Tm1 but it contributes to disrupt the ordered 243

array of hydrocarbon chains in the metastable phase 244

On the basis of the DSC thermogram the aqueous phase behavior of G-C181-OH under 245

diluted conditions (5 mgmL-1) was studied at four different temperatures (red marks on Figure 246

2a) below and above Tm1 but also between Tm1 and Tm2 corresponding to the thermogram of 247

the dry powder 248

9

249

Figure 2 ndash a) DSC thermogram of G-C181-OH under dry (04 wt water estimated by thermogravimetric 250

analysis) and hydrated conditions recorded at a heating rate of 10degCmin Red marks indicate the 251

temperatures at which cryo-TEM experiments are performed on the corresponding 5 mgmL aqueous 252

solutions b) Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 130degC 253

90degC 70degC and 25degC T indicates the temperature of the sample solution before cryofixation Tne T T 254

being the temperature of the sample upon plunging in liquid ethane Please refer to the materials and 255

methods section for more information 256

T= 130degC

T= 90degC

T= 70degC

T= 25degC

a)

b)

20 40 60 80 100 120

99 wt

50 wt

25 wt

T degC

En

do

10 wt

Tm1= 483 degC

Tm2= 957 degC

10

257

The series of cryo-TEM images in Figure 2b show an overview of the most relevant 258

structures observed on G-C181-OH solution from T= 130degC to T= 25degC whereas T is 259

indicative of the equilibrium temperature before cryofixation (refer to materials and methods 260

section for more information on the sense and interpretation of T) At Tgt 100degC the sample 261

is essentially composed of polydisperse single wall vesicles massively shown by the 262

complementary pictures presented in Figure S 2 In between Tm1 and Tm2 at T= 90degC and T= 263

70degC (Figure 2b Figure 3 Figure 4 complemented by Figure S 3 and Figure S 4) vesicles are 264

connected by nanotubes At Tlt Tm1 flat crystals are otherwise observed (Figure 2b and Figure 265

S 5) 266

267

268

Figure 3 ndash Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 90degC 269

270

A deeper insight at T= 90degC is given in Figure 3 (more images are provided in Figure 271

S 3) Arrows 1 through 6 mark very clearly the typical continuity in the glucolipid membrane 272

between the vesicle and nanotubes Arrow Ndeg 7 shows that a single nanotube connects the outer 273

membrane of two vesicles several hundred nanometers apart The connection is shown by 274

2

34

5

6

6

1

5

7

A

B

D

C

0 10 20 30 40 50 60

BIn

tensity

au

Profile nm

A

0 10 20 30 40 50 60

Inte

nsity

au

Profile nm

C

D

T= 90degC

11

arrows Ndeg5 and Ndeg6 The corresponding profiles A through D confirm the tubular nature of 275

the elongated objects Profiles B and D are both characterized by typical symmetrical thick 276

walls (low intensity depths highlighted by shaded rectangles) surrounding a hollow region (high 277

intensity peak between the shaded rectangles) The wall thickness varies between 35 and 40 278

nm for profile B and between 41 and 61 nm for profile D while the total nanotube diameter 279

varies between 14 nm at B and 20 nm at D At the same time the thickness of the vesicle wall 280

is about 6 nm at A and 135 nm at C Considerations about the bilayer or interdigitated structure 281

of the membrane will be presented below 282

The connection between nanotubes and vesicles seems to be multiple For instance 283

arrow Ndeg1 (Figure S 3) and Ndeg4 (Figure 3) show a single-wall vesicle of which the membrane 284

is extruded into a nanotube Similarly a closer look at arrows Ndeg1 2 and 3 in Figure 3 show 285

that the vesicle is single-wall and fully closed while the nanotube nucleates at the vesicle outer 286

surface with no apparent topological continuity In many regions of the sample at T= 90degC 287

one can find exvaginations or buds232531 in the nanotube wall This is shown for instance by 288

arrows Ndeg7 and particularly put in evidence in Figure 4 by arrows Ndeg1 through Ndeg4 where the 289

amplitude of the oscillation varies between 20-25 nm (arrow Ndeg1) and 6-8 nm (arrows Ndeg3-4) 290

291

Figure 4 - Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heated at T= 90degC 292

293

Still between Tm2 and Tm1 but at lower temperature (T= 70degC) the nanotubes seem to 294

vary from single threads (arrow Ndeg1 Figure S 4) to nanotubes of diameter below 15 nm both 295

connecting nanotubes of diameter ranging between 50 nm and 100 nm (arrows Ndeg2 Figure S 296

4) The larger tubes are most likely formed by the inflation and flattening of the nanotubes 297

probably after assimilation of surrounding vesicles and fusion of the exvagination discussed 298

above The transition between vesicles tubes and flat membranes is certainly more accentuated 299

Arrow Ndeg3 in Figure S 4 shows distinct vesicles while arrow Ndeg4 identifies a flat structure 300

evidently formed from fused vesicles In particular arrow Ndeg5 points at the fusion of two 301

2

1 132

4

12

vesicles while arrow Ndeg6 points at two vesicles fusing into the flat membrane Additional areas 302

where vesicle fusion is prominent are indicated by the oscillations at the outer skirt of broad 303

lamellae and pointed by arrow Ndeg7 304

Finally below Tm1 at T= 25degC for instance mainly flat crystals with no visible 305

exvaginations or tubes are visible (Figure S 5) by cryo-TEM and the material turns into an 306

insoluble precipitate Massive presence of flat crystalline structures is also clear from the bulk 307

iridescent behavior of a G-C181-OH solution under manual shear 308

In summary despite the uncertainty related to the actual state of equilibrium of the 309

glucolipids at the moment of cryofixation cryo-TEM depicts a vesicle phase at Tgt 100degC a 310

flat lamellar phase below Tm1 and an interconnected vesicle-nanotube network between Tm1 and 311

about 100degC The combination of statistically-relevant fluorescence microscopy and 312

scatteringdiffraction experiments provides an additional proof of the coexistence of 313

vesiclenanotube systems between the two transition temperatures Please note that the 314

experiments reported below have been performed under controlled conditions of temperature 315

which is then indicated by the use of T instead of T 316

13

317

Figure 5 ndash a-c) Confocal fluorescence microscopy images recorded at 90degC and 70degC for a 5 mgmL G-318

C181-OH aqueous solution d-g) Typical fluorescence microscopy images recorded at T= 80degC using 319

Laurdan (refer to materials and methods section for more details on this fluorophore) Excitation 320

wavelength is 405 nm for all the images while emission wavelength is 490 nm in (df) and 440 nm in (eg) 321

h) Full emission spectra obtained from the integration of various fluorescence images recorded from 412 322

nm to 600 nm (step 5 nm) at 90degC 80degC and 70degC i) Box plot of GP defined as 119918119927 =(120640120786120786120782minus120640120786120791120782)

120640120786120786120782+120640120786120791120782 with 120640120786120786120782 323

and 120640120786120791120782 being the emission wavelengths at 440 nm and 490 nm at 90degC 80degC and 70degC Each data point 324

l= 490 nm l= 440 nm

l= 490 nm l= 440 nm

10 μm400 450 500 550 600

00

02

04

06

08

10

No

rma

lize

d I

nte

nsity

l nm

90degC

80degC

80degC

70degC

70degC

400 450 500 550 60000

02

04

06

08

10

No

rma

lize

d I

nte

nsity

l nm

90degC

80degC

80degC

70degC

70degC

d) e)

f) g)

h)

i)

T= 80degC

Laurdan

T= 80degC

T= 80degC T= 80degC

T= 90degC T= 70degC T= 70degC

a) b) c)

1

1

1

14

corresponds to a set of regions of interest (ROIs) (54 at 90degC 65 at 80degC and 75 at 70degC) selected in the 325

corresponding fluorescence images ROIs were representatively selected so to include the entire image but 326

also vesicles alone tubes alone and mixtures of vesicles and tubes 327

328

Figure 5ab shows two typical confocal fluorescence microscopy images recorded on a 329

G-C181-OH solution (5 mgmL) in a flame-sealed capillary heat at T= 90degC and T= 70degC 330

Although the resolution of optical microscopy is not comparable with cryo-TEM the sample is 331

broadly composed of vesicles coexisting with nanotubes whereas in some cases the nanotube 332

starts from the vesicle itself (arrow Ndeg1) in good agreement with cryo-TEM data Furthermore 333

SANS recorded at T= 90degC (Figure S 6a) shows a scattering profile with a pronounced -2 334

dependence of the scattering intensity against the wavevector q in the log-log scale This 335

indicates the massive presence of a flat interface and it is expected typical of bilayers in 336

vesicular morphologies56 However the typical signature of nanotubes cannot be observed in 337

the SANS profile and this can be explained by both an argument related to the relative lower 338

amount of nanotubes with respect to vesicles at 90degC but also to the steeper slope of flat 339

morphologies (-2) compared to tubes (-1) the signal of the former masking the tube signal at 340

low wavevector values The SANS profile is characterized by a diffraction peak at q= 0167 Aring-341

1 of which the assignment is refined by WAXS data 342

Temperature-driven WAXS experiments in Figure S 6bc (5 mgmL-1 and 25 mgmL-1 343

respectively) show a lamellar order possibly attributed to a multilamellar structure of the 344

vesicles (dv Figure S 6d) at high temperature and to the interplanar arrangement of the 345

crystalline flat structures (dL Figure S 6d) at T= 25degC The sharp (001) reflection indicates 346

typical interlamellar distance of 353 Aring at 25degC although this value varies with temperature 347

(400 Aring at 111degC 372 Aring at 88degC) thus confirming its assignment to a hydration-dependent 348

interlamellar order More intriguing is the second broader peak at q= 0403 Aring-1 (d= 156 Aring) 349

and its reflection both clearly visible at 88degC below Tm2 We attribute it to the inter-glucolipid 350

arrangement inside the tubular phase (dt Figure S 6d) coexisting with the vesicle phase 351

between Tm1 and Tm2 and to the planar lipid arrangement (di Figure S 6d) in the rigid flat 352

lamellar structure below Tm2 Its temperature- hence hydration- independent position also 353

confirms its structural origin Finally besides the water-structure peak at about 2 Aring-1 we also 354

observe a broad peak at 144 Aring-1 characteristics of the lateral packing (d Figure S 6d) of lipids 355

in bilayers While its position could be consistent with an Lα phase5758 above Tm2 its broadness 356

and positional invariance as well as the coexistence between vesicles and tubes do not allow a 357

precise assignment 358

15

Does G-C181-OH assume a bilayer or interdigitated structure the latter being the one 359

found for other similar bolaform glucolipids4243 The size of G-C181-OH can be estimated to 360

be at the most 32 Aring given by the length L of the C16 tail (L = 22 Aring from the Tanford formula 361

L= 154 + 1265 times n (n= 16)59) and the sugar (about 10 Aring)4243 However the cis conformation 362

of the C=C bond imposes a ldquokinkrdquo of 60deg in the aliphatic chain which reduces the effective 363

length of the oleic backbone to about 17 Aring for an effective total G-C181-OH length of about 364

27 Aring Cryo-TEM provides a tube thickness varying from 35 Aring to 60 Aring and a vesicle membrane 365

thickness of at least 60 Aring WAXS suggests an interlamellar thickness (membrane and water) 366

between 40 Aring and 37 Aring and an inter-lipid distance in the tube of about 16 Aring In the absence of 367

a clear form factor oscillation profile as found for similar glucolipids4243 one must 368

acknowledge that these data pledge neither in favor of fully interdigitated nor a double layer 369

structure and might suggest a tilted lipid arrangement with partial interdigitation60 370

371

13C ssNMR demonstrates that G-C181-OH can assume two coexisting head-head and 372

head-tail configurations 373

Vesicles are characterized by low positive mean and gaussian curvatures while 374

nanotubes are described by a combination of a high positive mean and zero gaussian curvatures 375

In this regard the coexistence of vesicles and nanotubes requires glucolipids to adapt 376

themselves to both environments at the same time For this reason the formation of nanotube-377

vesicle networks generally requires a specific driving force able to impose a high mean and 378

zero gaussian curvature to the membrane In many systems insertion of at least a second 379

constituent (lipid surfactant protein) in the membrane becomes a necessary internal driving 380

force for the nucleation and growth of so-called membrane inclusions from which 381

topologically-connected nanotubes can form10ndash123261 as explained by orientational ordering 382

arguments62 It has also been shown that external driving forces such as use of laser 383

ldquotweezersrdquo30 osmotic pressure6 electroformation5ndash82829 or controlled suction13ndash16 are 384

otherwise necessary as also supported by theoretical studies19 385

The bending energy of tubes is higher than that of a vesicle membrane8 due to the 386

dependence of the membrane free energy from the principal curvatures For this reason 387

isotropic (rotational symmetry along the main axis) but also anisotropic33 lipids generally 388

stabilize low-curvature shapes like flat membranes or vesicles Lipid nanotubes are also a 389

common phase formed by amphiphiles but generally for lipids with a certain complexity in 390

their molecular structures as exampled by diacetylenic derivatives of phospholipids among 391

others262763ndash65 Negative contributions to the bending energy favoring the simultaneous 392

16

presence of interconnected vesicles and tubes was shown to be driven by the presence of 393

anisotropic lipids in a membrane continuum Spontaneous deviations to the elastic theory in the 394

presence of surfactants lipids and proteins with anisotropic structures favor the formation of 395

protrusions with radius of the order of the fraction of a micron9ndash1219ndash2362 396

In the absence of both internal and external driving forces mentioned above G-C181-397

OH should behave as a standard lipid Its DSC profile (Figure 2)536667 under both hydrated and 398

dry conditions identifies a hydration-independent transition (Tm1) slightly above room 399

temperature this is followed by one main transition at higher temperature (Tm2) in the dry 400

sample50 For such a profile one expects a gel phase (Lβ) below Tm1 and a liquid crystalline 401

structure possibly in the morphology of vesicles (L4 phase) possibly above Tm1 and most likely 402

above Tm231 as experimentally found On the basis of the thermogram recorded on the dry 403

powder one may also expect a ldquoripplerdquo (Pβ) phase between Tm1 and Tm2 characterized by 404

periodic undulations of the bilayer (generally addressed to as regions with both liquid order and 405

disorder)5455 406

In the present work instead of a Lβ phase we observe flat crystals below Tm1 and a 407

vesicle-nanotube network instead of the ldquoripplerdquo phase The latter is particularly unexpected 408

because the membrane is composed of a single-component (G-C181-OH) and no anisotropic 409

external force (osmotic hydrodynamic electric) is used to pull and stabilize the tubes In 410

addition the nanotubes are observed at fairly high temperatures above Tm1 (between about 411

60degC and 90degC) while this morphology is generally favored at low temperature (proportionality 412

with 1kT figure 5 in Ref 62) commonly below a given transition temperature characteristics 413

for each specific amphiphile27 Furthermore nanotubes of G-C181-OH seem to be stable in 414

time thus differing from what is found in electroformed vesicles78 They are also observed 415

away from phase transition events (may them be at Tm1 or Tm2) as otherwise reported before 416

for a diacetylenic derivative of phosphatidylcholine26 and many other systems27 417

A stable nanotube-vesicle network for a single-component phospholipid in the absence 418

of external driving forces is not expected and has never been reported before It could be 419

explained by the coexistence of two micro-separated phases of different rigidity This 420

hypothesis is excluded by confocal microscopy on Laurdan-labelled membranes and magic 421

angle spinning (MAS) solid-state NMR (ssNMR) experiments 422

Laurdan is a well-known fluorophore used to put in evidence the coexistence of rigid 423

and fluid domains in both model and biological membranes47ndash49 Laurdan was used here to label 424

the G-C181-OH membranes between Tm1 and Tm2 and to verify the possible micro-segregation 425

and coexistence of rigid (eg Lβ λem= 440 nm) and fluid (eg Lα λem= 490 nm) domains in 426

17

tubes and vesicles respectively A set of quantitative experiments performed on a 5 mgmL-1 427

solution summarized in Figure 5 shows an excess of fluorescence at only λem= 490 nm for 428

temperature between 90degC and 70degC (Figure 5h) The GP value which quantifies the rigidfluid 429

character of the membrane is systematically negative (Figure 5i) indicating an overall fluid 430

membrane at all temperatures GP is quantified on more than 50 regions of interest (ROIs) at 431

each temperature including vesicles and tubes alone as well as connected nanotube-vesicles 432

(Figure 5d-g) The dispersion of GP at 80degC and 70degC between -08 and -02 could suggest the 433

spurious presence of membranes with various degrees of flexibility although fluidity is always 434

predominant The combination of WAXS (broad peak d= 144 Aring-1 invariant with 435

temperature Figure S 6bc) and confocal fluorescence microscopy using Laurdan confirms the 436

single-phase of the membrane composed of G-C181-OH both in vesicles and nanotubes This 437

is also consistent with the lack of phase transitions in DSC experiments between ~85degC and 438

~60degC especially in hydrated samples We conclude that the vesicle-nanotube transition is not 439

driven by the local segregation of gel phase (Lβ) microdomains nor that the nanotubes are 440

composed of such phase 441

ssNMR is the best-suited approach to investigate the origin of the local driving force for 442

the nucleation of anisotropic inclusions of which the average orientation depends on the local 443

membrane and inclusion curvatures and was shown to promote tubing62 ssNMR can provide 444

information on the local molecular mobility eg through the nuclear spin relaxation behavior 445

but also on intermolecular interactions and stereochemistry eg through variations in the 446

chemical shift68ndash70 447

Ideally ssNMR should be performed on diluted G-C181-OH samples above Tm1 448

However fast MAS necessary for high resolution centrifuges the sample As a first 449

approximation hydrated G-C181-OH samples for instance at 50 wt as classically done for 450

phospholipids71 could be studied Unfortunately permeability of the sample holder combined 451

to heating (close to the boiling point of water) and MAS provoke dehydration and even sample 452

ejection After experiencing both of the above this approach was abandoned As a second 453

approximation the study of the dehydrated G-C181-OH powder avoids centrifugation and 454

evaporation If studying dried lipids may not reflect their exact behaviour under diluted 455

conditions valuable structural information can still be obtained in dehydrated molecular 456

systems like proteins thus illustrating that this approach is not unreasonable Furthermore if 457

DSC (Figure 2a) shows that hydration suppresses the Tm2 it also shows that the value of Tm1 is 458

not affected by water at concentrations as low as 10 wt For this reason performing a 459

structural analysis of dehydrated G-C181-OH by ssNMR above Tm1 may help understanding 460

18

the local arrangement of G-C181-OH in two morphologies vesicles and nanotubes having 461

sensibly different curvatures 462

To explore the local mobility of the most relevant chemical groups of G-C181-OH 463

above Tm2 (T= 110degC) and between Tm2 and Tm1 (70 lt T degC lt 90) we combine a series of 1H 464

and 13C experiments Single-pulse (SP) MAS experiments (1H and 13C) when recorded under 465

quantitative conditions allow the detection of all chemical species while cross-polarization 466

(CP) MAS (13C only) experiments are sensitive only to constrained groups for which the 1H-467

13C dipolar coupling is not averaged to zero by the molecular motions 468

469

Figure 6 ndash a) 1H MAS (10 kHz) ssNMR spectra recorded from 110degC to 70degC b) 13C MAS (10 kHz) ssNMR 470

spectra recorded using high power decoupling single pulse (SP) and cross-polarization (CP) pulse sequences 471

Contact time for CP experiments and corresponding temperatures are given on each spectrum Full spectra 472

are given in Figure S 7 473

The most relevant 1H and 13C spectra are given in Figure 6 (full range in Figure 7 and 474

Figure S 7) As expected the 1H solid-state SPMAS spectra show broader peaks compared to 475

the solution-state NMR spectrum (Figure 6a) reflecting the obvious solid-state character of the 476

sample at any temperature The sugar group and the C=C bond resonate in the 25-6 ppm range 477

and 535 ppm respectively while the OH groups of the sugar moiety are found below 45 ppm 478

(Figure 6a full assignment in Table S 1) Very broad (~1 ppm) at 100degC the position of the 479

OH peak varies with the temperature value and becomes narrower (lt~05 ppm) and visible at 480

δH= 464 ppm at T= 70degC Line narrowing and chemical shift variations of the OH resonance 481

in ssNMR is always a direct sign of H-bonding interactions whereas the strength of the H-482

bonding hence its OHhellipH distance is correlated to a dispersion in chemical shifts72ndash75 All other 483

60 55 50 45 40 35 30 25

ppm

110

90

80

70

910

1rsquo

6rsquo16

1

OH

134 132 130 128 126

ppm110 108 106 104 102 100

ppm

SP (T= 110degC)

SP (T= 70degC)

CP 50 ms (T= 70degC)

CP 3 ms (T= 70degC)

910 1rsquo

1H 13C

1

811

910

CH2 CH2

1rsquo

6rsquo

a) b)

C102

13024

C9101

13003

C92

12869 C1rsquo210382

C1rsquo110443

C1rsquo13035

2rsquo3rsquo4rsquo

T degC

5rsquo

19

resonances broaden from 110degC to 70degC thus indicating slower andor more anisotropic 484

reorientational motions of all chemical groups The specific dynamic behaviour of the sugar 485

headgroup is also reflected by its 1H transverse relaxation time T2 whereas the shorter the T2 486

the higher the rigidity Experiments performed between 110degC and 70degC shown and 487

commented in Figure S 8 in the Supporting Information globally show a single short (lt 1 ms 488

below 100degC) T2 relaxation component for H1rsquo H5rsquo and H6rsquo and two both short and long 489

(between 4 ms and 12 ms) T2 for the acyl chain (CH2 H910 H811) even below 100degC 490

At 110degC 13C SPMAS experiments (Figure 6b full range in Figure 7 and Figure S 7 491

assignments in Table 1) show a single peak at δC= 10335 ppm for C1rsquo and at δC= 13003 ppm 492

for C910 as expected for a symmetrical conformation with two chemically-equivalent 493

carbons76 At 70degC the C1rsquo peak disappears while C910 now shows two peaks around 128 494

ppm and 130 ppm (Table 1 full temperature series in Figure S 7b) attributed to C92 (δC= 12869 495

ppm) and C10 in a non-symmetric configuration respectively for which these two carbons are 496

chemically inequivalent76 More precisely the resonance at 130 ppm is actually the 497

superposition of two close resonances C9101 (δC= 13003 ppm) and C102 (δC= 13024 ppm) 498

where subscripts 1 and 2 stand for two different conformations of the cis-olefin group as 499

reported for various phases of oleic acid in the vicinity of 0degC76 500

501

Table 1 ndash 13C Chemical shifts of G-C181-OH in solution and solid-state Solid-state experiments are 502

recorded at MAS= 10 kHz All spectra are reported in Figure 7 and Figure S 7 The contact time for cross 503

polarization is 3 ms In all ssNMR experiments temperature is first raised at 110degC and then lowered to 504

either 90degC or 70degC Attribution of the 13C resonances in Table 1 is performed on the basis of the solution 505

spectrum (Figure S 1b) Additional peaks observed in both single pulse and cross polarization ssNMR 506

experiments are attributed when possible to a configuration effect of the given atom eg C9101 and C92 507

indicate two configurations of C9 All new peaks observed in solid-state and not detected in solution are 508

highlighted in red for convenience 509

510

511

δ ppm

Peak Liquid (DMSO-d6)

Solid Single Pulse

T= 110degC

Solid Single Pulse

T= 90degC

Solid Single Pulse

T= 70degC

Solid Cross Polarization

T= 70degC

C102

- - 13024 13024

C910C9101 13009 13003 13003 13003 13003

C1C9C10C1rsquo

C6rsquo

20

C92 - - - 12869 12869

C1rsquoC11 10329 10335 10335 - 10443

C12 - - - - 10382

C3 7724 7709 7702 - 7829

- 7634 7629 - 7621

7525

C2 7388 7398 7390 7370 7370

C4 7052 7077 7054 7041 7098

C16 6900 7016 - - 7029

- - - - 6793

C1C5C6 6153 6241 6236 6239 6262

C1C5C6 6117 - - - 5956

C2 3300 3290 3290 3283 3279

CH2(trans) - - - 3198 3196

CH2(gauche)C15 2941 2981 2987 3004 -

C811 2706 2747 2747 2832 2817

CH2 2597 2625 2628 2630 2817

512

13C CPMAS spectra at 70degC (Figure 6b full range in Figure 7 and Figure S 7d) at 513

increasing contact times confirm the presence of two cis-olefin environments one for which 514

C910 are equivalent labelled here C9101 and one for which they are inequivalent labelled 515

C92 C102 Both environments are rigid enough to be observed at short contact time (tc= 50 μs) 516

A similar double environment is also observed for the sugar C1rsquo CPMAS at 70degC shows two 517

peaks (C1rsquo1 at δC= 10443 ppm C1rsquo2 at δC= 10382 ppm Figure 6b Figure S 7d) These are 518

significantly shifted downfield from the C1rsquo peak observed at 110degC and are detected at contact 519

times as short as 50 μs thus indicating their rigid character One also finds that CPMAS is 520

totally inefficient at 100degC for short contact times (50 μs) and poorly efficient even at 3 ms 521

typical of fast-tumbling molecules (Figure S 7c) also both gauche (δC= 3004 ppm) and all-522

trans (δC= 3198 ppm) conformations of the CH2 carbons coexist at 70degC (Figure S 7b)77 523

According to ssNMR one main mobile molecular environment exists at 110degC as 524

shown by the well-resolved 1H spectrum single C=C (C9101) and anomeric (C1rsquo) resonances 525

gauche conformation of the aliphatic CH2 and single T2 of the aliphatic CH2 chain and sugar 526

protons The sugar group is certainly mobile (inefficient 1H13C CPMAS) but still more 527

constrainted than the acyl chain (shorter T2 (1H)) As the temperature is lowered the overall 528

rigidity increases (broadening of 1H NMR peaks faster 1H transverse relaxation and efficient 529

1H13C magnetization transfer during CPMAS) presumably driven by H-bonding 530

interactions (downfield shift and line narrowing of the OH 1H signal very short OH T2 loss of 531

the C1rsquo signal in SPMAS) At the same time a number of spectroscopic data (two concomitant 532

21

resonances for anomeric C1rsquo1 and C1rsquo2 cis-olefin C9101 and C92 C102 and CH2 gauche and 533

all-trans environments) and relaxation behaviour (fast and slow T2 relaxation components) 534

strongly suggests the coexistence of two molecular environments Further proofs of the 535

coexistence of two distinct molecular domains are given by a number of chemical shift 536

variations in the 13C CPMAS spectrum at 70degC compared to the one at 110degC 90degC and 537

solution This is shown in Figure 7 and in the list of 13C chemical shifts in Table 1 538

539

540

Figure 7 - Comparison between solution- and solid-state 13C NMR spectra of G-C181-OH Solid-state 541

experiments are recorded directly on the sample powder at a MAS of 10 kHz (4 mm zirconia rotor) SP 542

Single Pulse CP Cross Polarization In all the ssNMR experiments temperature is first raised at 110degC 543

and then lowered to either 90degC or 70degC Segmented lines are meant to highlight the known peak position 544

in solution thus stressing the strong chemical shift variations in the solid-state 545

546

NMR experiments above Tm2 suggest that vesicles could contain G-C181-OH with a 547

single family of mobile conformations characterized by chemically-equivalent C=C (red 548

Figure 8b) and a single C1rsquo conformation (green Figure 8b) Experiments between Tm1 and Tm2 549

otherwise suggest that two coexisting conformations characterized by chemically-equivalent 550

(red Figure 8b) and inequivalent (magenta Figure 8b) C=C could be associated to nanotubes 551

Two distinct C1rsquo sites (grey Figure 8b) also indicate two configurations of the sugar headgroup 552

A comment on this point will be given in the following paragraphs 553

The local (Aring-scale) variation of molecular mobility below Tm2 drives the coexistence of 554

two distinct molecular environments of the same G-C181-OH molecule This occurs within a 555

single-phase domain (as deduced by WAXS) of equivalent membrane rigidity (as derived from 556

Laurdan probe in confocal microscopy) Although never reported before the difference in the 557

local mobility of the same amphiphile (here G-C181-OH) could satisfy the theoretically-558

140 120 100 80 60 40 20

CP(3 ms)-70degC

SP-70degC

SP-90degC

SP-110degC

ppm

LiquidC910

C1rsquo

C2rsquo

C4rsquo

C3rsquo

C1

C6rsquoC5rsquo

C1

6

C8

11

C2 C

H2

C1

5

CH

2

DM

SO

13C

22

required conditions that induce local lipid segregation and formation of membrane inclusions 559

(Figure 8c) In case of their non-zero average orientation the local membrane curvature 560

increases62 and nanotubes can spontaneously form (Figure 8bc) More intriguing this 561

mechanism occurs at least 10degC above Tm1 although it should not be favored above the Tm27 562

From both experimental and theoretical backgrounds the conditions of topologically-563

connected vesicles-nanotubes systems are met only when either internal (two components) or 564

external driving forces are applied19 In the present single-lipid system of which the phase 565

behavior is summarized in Figure 8a the most plausible driving force could be explained by 566

the NMR results collected on the dry sample G-C181-OH concomitantly occupies two 567

molecular environments characterized by 1) two distinct conformations of the sugar 568

displaying equivalent mobility and 2) two conformations of the acyl chain unsaturation 569

570

23

571

Figure 8 ndash a) Thermotropic phase behavior of G-C181-OH at 5 mgmL-1 Note that Tm2 is observed on the 572

dry powder only b) Possible relationship between the molecular environment as derived from NMR and 573

membrane curvatures All the molecular drawings refer to the same G-C181-OH molecule (bulky ellipsoid 574

equiv sugar headgroup) while each color corresponds to a given configuration Grey of C1rsquo1 and C1rsquo2 indicates 575

the uncertainty about the assignment of the specific NMR peak The curvature C is defined as 1R R being 576

the curvature radius R1 R2 radius of curvature CG Gauss curvature CM Mean curvature c) Free 577

interpretation of the anisotropic high-curvature (CMgt 0 CG= 0) inclusions in the membrane (low CM and 578

CG) continuum driven by inter-sugar H-bonding and reduced intramembrane diffusion for a G-C181-OH 579

glucolipid 580

581

Flat crystalline structure

Nanotube-vesicle network

Vesicles

48degC

120deg

106 105 104 103 102 101

ppm

H9 H10H1rsquo

106 105 104 103 102 101

ppm106 105 104 103 102 101

ppm106 105 104 103 102 101

ppm

C=CC=C

C1rsquo1C1rsquo2

C1rsquo

Tm2(96degC)Tm1

a)

b)

c)

Same molecule different local orientations and sugar conformations

High curvature membrane

Low curvature inclusion

Flip-FlopStrong H-bonding

CG

CM ( + )

CG 0

CM 0 CG 0

CM 0

G-C181-OH

24

Two related questions are unanswered at the moment What actually drives nanotube 582

formation and is the organization of G-C181-OH within the vesicles and tubes the same Lipid 583

nanotubes have been studied for at least three decades over which broad experimental and 584

theoretical insight has been generated Despite this fact the actual understanding is still matter 585

of debate27 Micelle-to-nanotube and vesicle-to-nanotube phase transitions are generally 586

reported for single amphiphiles below a given Tm and intermediate phase composed of twisted 587

and flat fibers is concomitantly reported Initially thought to be a chirality-driven phenomenon 588

recent studies propose a more general mechanism involving crystallization and symmetry 589

breaking in the lipid packing driven by optimal packing27 If twisted ribbons were observed 590

before for a number of microbial glycolipid bolaamphiphiles44 they are not found for G-C181-591

OH This fact as well as the large body of cryo-TEM images and the observation of nanotubes 592

above Tm1 rather pledge in favour of a direct topological change from positive (vesicles) to 593

zero (nanotubes) Gaussian curvatures In this case vesicle-to-nanotube transitions could be 594

driven by external forces56307813ndash162829 595

Considering the absence of external forces in the present system there are only two 596

possibilities left spontaneous budding of the membrane or nanotube pulling after inter-vesicle 597

collisions due to Brownian motion Both are possible as inclusions composed of a different 598

arrangement of the same molecule in the membrane continuum can theoretically explain 599

spontaneous nanotube formation in an isotropic medium672223629ndash1218ndash21 However the second 600

one is probably more likely to prevail because it could explain the large number of nanotubes 601

connecting two distant vesicles Interestingly similar events occur for any phospholipid-based 602

vesicular solution but nanotubes are extremely rare unless at least two lipids of different 603

molecular structure are mixed together The spectroscopic 1H and 13C and relaxation (1H) 604

ssNMR data even if collected on a dry powder demonstrate that the coexistence of two G-605

C181-OH configurations involving both the sugar headgroup and fatty alcohol chain are 606

possible for this molecule As a matter of fact this is not so surprising as more than one energy 607

minima each corresponding to a given conformation are known for glucose its non-reducing 608

derivative methylglucose and carbohydrates in general78ndash80 609

The presence of two configurations could explain the coexistence between two 610

structures of different curvature but the data presented so far do not explain the origin of these 611

two configurations In other words what drives the change in the local structure of the sugar 612

headgroup and fatty alcohol within the membrane 613

The association of known self-assembled structures to spectroscopic 13C ssNMR data is 614

known for years in particular for gluconamides lipids having a linear sugar headgroup81ndash84 It 615

25

was found that gauche effects in some cases up to the γ position in the sugar chain induce 616

upfield chemical shifts while anti and trans conformations result in downfield shifts Each 617

structure (monolayer crystal bilayer crystal or micellar fiber) was then characterized by a given 618

organization of the gluconamide and a characteristic 13C ssNMR fingerprint which could be 619

explained by gauche γ-gauche anti and trans conformations of the headgroup Similar 620

structural-spectroscopic correlations have been collected over the years for much simpler cyclic 621

sugars85 whereas pyranoses can undergo important conformational changes (chair half-chair 622

boat skew) when the appropriate energy barriers are crossed78ndash8086 A quantitative 623

conformational study based on the 13C chemical shift variations shown in Table 1 would be 624

certainly helpful but very delicate in the context of the present state of the art because a 625

rigorous approach requires at least one known crystal structure8182 which is currently missing 626

for this compound A qualitative approach will then be adopted 627

G-C181-OH is a bolaform amphiphile and flip-flop could be a possible mechanism 628

involved in the nanotube stabilization whereas head-tail interactions have long been considered 629

as important in nanotube-forming bolaform amphiphiles6063 From mere considerations based 630

on T2 (1H) relaxation arguments one could exclude head-tail configurations if the R-CH2OH 631

endgroup (H1) was involved in a direct interaction with the sugar group its corresponding T2 632

would probably be comparable to the sugarrsquos one (OH and H1rsquo) but this does not seem to be 633

the case However this hypothesis would be based on the comparison of the relaxation from 634

CH and CH2 protons additionally involved in different kind of motions Besides the relaxation 635

of H1 is influenced by the strong overlap between H1 and H5rsquo For these reasons relaxation 636

arguments are not solid enough and will not be used 637

We then address this question by exploring intermolecular interactions in a direct 638

manner using 2D 1H-1H homonuclear ssNMR correlation spectroscopy At 90degC in the fast-639

tumbling domain of molecular motion (narrow 1H NMR peaks Figure 6a) and corresponding 640

to the vesicle-rich region spin diffusion (NOESY) experiments are preferred over dipolar 641

coupling based experiments because such interaction is averaged out by molecular motion 642

However when the molecular motion is more constrained in the domain of nanotube stability 643

at 70degC (broad 1H spectra Figure 6a) Double-Quantum Single-Quantum (DQ-SQ) correlation 644

experiments are best suited to probe interaction between close molecular neighbors These 645

experiments are reported in Figure 9 and the related interpretation is only briefly reported 646

below while a more detailed discussion can be found on Page 12 in the Supporting Information 647

26

648

Figure 9 ndash a) 2D 1H-1H NOESY NMR experiment performed at MAS= 10 kHz (4 mm probe) on a Bruker 649

300 MHz Avance III HD spectrometer using 300 ms of mixing time Sample temperature is regulated to 650

90degC reached by cooling from 110degC b) 2D 1H-1H DQ-SQ (BAck-to-BAck BABA pulse sequence used for 651

the DQ excitation scheme) NMR experiment performed at MAS= 22 kHz (25 mm probe) on a Bruker 700 652

MHz Avance III spectrometer using a DQ excitation time equal to 2 rotor periods Sample temperature is 653

regulated to 70degC reached by cooling from 110degC The typical schemes illustrating how to interpret these 654

2D experiments are reported above each experimental map Zooms of the sugar region and configurational 655

interpretation are given below each 2D map More information are given in the text below 656

H16-H16

H1-H1

H9-H10

H1rsquo-H1

H1rsquoH910 H16

H1

H1rsquo-H1rsquo

Hxrsquo-Hxrsquo

2

(pp

m)

(ppm)

315

811910 1rsquo

2rsquo2

3rsquo 4rsquo6rsquo

5rsquo

1rsquo2910

asymp asymp asympHDO MeOD Acetone

1

315

811 9101rsquo

2rsquo2

3rsquo4rsquo6rsquo

5rsquo

1rsquo2 9 10

asympasympasympHDOMeODAcetone

H1rsquo-H1

1rsquo

1

(p

pm

)

(ppm)

H1-H2

H1rsquo-H1

H1rsquo

H910

H1H5rsquo

H2

H1rsquo-H5rsquo

315

811910 1rsquo

2rsquo2

3rsquo 4rsquo6rsquo

5rsquo

1rsquo2910

asymp asymp asympHDO MeOD Acetone

1

H2rsquoH3rsquoH4rsquo

H2rsquoH3rsquoH4rsquo

X= 2rsquo3rsquo4rsquo

NOESY - T= 90degC

1H

1H

1H

1H

DQ-SQ - T= 70degC

a) b)

2

(pp

m)

(p

pm

)

(ppm)

A B

AhellipB

AhellipA

BhellipB

NOESY

(ppm)

A B

AhellipB

DQ-SQ

Single-quantum dimension

Do

ub

le-q

ua

ntu

m d

imen

sio

n

AhellipA

BhellipB

BhellipA

BhellipA

315

811910 1rsquo

2rsquo2

3rsquo 4rsquo6rsquo

5rsquo

1rsquo2910

asymp asymp asympHDO MeOD Acetone

1

lt 5 Aring

lt 5 Aring

H1-H2

H811

H15

H16

H6rsquo

H1rsquo

H1

H5rsquo

H2

H1-H2

H1rsquo-H1

H1rsquo-H5rsquo

H1rsquo

H1 H5rsquo

H1rsquo

H1

H5rsquo

H1H2rsquo H3rsquo H4rsquo

H1-H1

H1rsquo-H1

Hxrsquo-HxrsquoX= 2rsquo3rsquo4rsquo

H1-H2

H1rsquo

Zoom sugar region 28 ndash 45 ppmNOESY DQ-SQ

27

Head-tail or head-head configurations could be probed by the existence of H1rsquo-H1 657

interactions whereas H1 (δH= 356 ppm) is associated to the glucolipid tail and H1rsquo (δH= 432 658

ppm) to the sugar headgroup These atoms are more than 15 Aring apart and a direct interaction 659

could not be observed in both NOESY and DQ-SQ experiments NOESY experiments (Figure 660

9a) show that H1rsquo-H1 correlations are not observed Meanwhile clear-cut direct H1rsquo-H1 661

interactions are indeed observed are detected at T= 70degC (DQ cross-peaks related to H1rsquo-H1 662

correlation in DQ-SQ experiment shown in Figure 9b) Interestingly the absence of auto peaks 663

at δH= 432 ppm and δH= 335 ppm in the SQ dimension excludes H1rsquo-H1rsquo and Hxrsquo-Hxrsquo (x= 2 664

3 4) interactions respectively strengthening the idea that sugar headgroups from two nearest-665

neighbor G-C181-OH molecules are not adjacent in the nanotube-vesicle region 666

Homonuclear 2D ssNMR supports the idea according to which G-C181-OH can 667

undergo molecular flip-flops and a head-tail arrangement in the dry product Extrapolating these 668

conclusions in the lower concentration range lack of H1rsquo-H1 proximity at T= 90degC suggests a 669

head-head configuration in the vesicle-rich region and head-tail configuration within the 670

nanotubes It is however still unclear whether flip-flops are driven by spontaneous temperature-671

induced conformational changes in the glucolipid packing stimulating inclusions within the 672

membrane itself672223629ndash1218ndash21 or by defects generated by inter-vesicle collisions In fact a 673

combination of both is not unlikely Nanotubes are not stable when temperature is 20 to 30 674

degrees above Tm1 indicating that inter-vesicle collisions alone cannot explain the unusual 675

phenomenon In an energy landscape approach87 the head-head configuration is certainly 676

favored at high temperature while both head-head and head-tail configurations could be 677

favored in the mid-temperature region corresponding to the range between Tm1 and Tm2 in the 678

dry powder Both statistical and defect-induced flip-flop events contribute to form high mean 679

zero Gaussian curvature morphologies (nanotubes) for which the head-tail conformation 680

limiting the proximity between bulky sugar headgroups is probably more adapted (Figure 8c) 681

The flip-flop interpretation now reasonably explains the strong differences in the sugar and fatty 682

alcohol conformations deduced by the multiple NMR signals attributed to chemically-683

equivalent groups The existence of various low-energy conformations each having specific 684

NMR signature for sugars in solution78ndash8086 and solid-state81ndash84 further supports our 685

interpretation 686

One last point should be clarified Why between Tm1 and Tm2 are two molecular 687

environments able to coexist instead of a single one of average rigidity as found in the case of 688

ldquoripplerdquo phases50ndash54 We believe that this question can be answered based on the NMR results 689

the relatively short T2 values of the H1rsquo and OH protons shows the constrained mobility of the 690

28

sugar headgroup despite a certain flexibility of the chain (long T2 component) This trend may 691

suggest restrictions in translational mobility probably due to the sugar-OH hydrogen bonding 692

(Figure 8c) meaning that lipids are not able to diffuse within the membrane thus keeping two 693

different configurations stable 694

695

Conclusions 696

Topologically-connected nanotube-vesicle lipidic systems are commonly observed for 697

self-assembled membranes composed of at least two amphiphiles and spontaneously driven by 698

anisotropic inclusions andor by external directional forces This work shows that strong 699

differences in membrane curvature can spontaneously occur in single-lipid systems without 700

external anisotropic forces Coexistence and interconnection between nanotubes and vesicles 701

are evidenced via cryo-TEM and fluorescence confocal microscopy in a single-phase region at 702

temperature above a transition temperature Tm1= 483degC determined by DSC on the hydrated 703

sample but below the second transition temperature Tm2 (957degC) determined by DSC on the 704

dry powder X-ray diffraction (WAXS) also shows the coexistence of two lattice periods one 705

attributed to an inter-membrane repeating distance and the other possibly related to the intra-706

nanotube glucolipid arrangement Labelling the systems with Laurdan a fluorophore of which 707

the emission strongly varies with the local membrane rigidity excludes the presence of rigid 708

micro-separated phases (eg Lβ) 709

1H and 13C ssNMR despite the fact that it can only be performed on the dry powder 710

confirm the presence of one main fluid G-C181-OH environment above Tm2 Below Tm2 the 711

splitting of some 13C peaks and the coexistence of short and long 1H relaxation components 712

demonstrate that G-C181-OH can simultaneously pack into structurally and dynamically 713

inequivalent environments Combining advanced 2D 1H-1H correlation experiments based on 714

spin diffusion or double-quanta excitation suggests that vesicles could contain a preferred head-715

head configuration of the bolaform lipid while nanotubes a preferred head-tail assembly 716

The presence of two environments composed of the same molecule characterized by 717

different molecular configurations and rigidity fulfills the theoretical conditions necessary to 718

spontaneously nucleate a membrane inclusion of non-zero average orientation described as 719

necessary to observe topologically-connected nanotube-vesicle structures in the absence of 720

external forces This hypothesis could also be enriched by the possibility of nanotubes to be 721

pulled away between two vesicles after a collision event 722

This work expands the actual theoretical and experimental framework on lipid networks 723

stabilized by a different radius of curvature It could also motivate the synthesis or isolation 724

29

of new lipids providing a higher level of 3D architectural control in soft networks possibly in 725

the range of room temperature 726

727

Acknowledgements 728

Dr Andrea Lassenberger and Dr Sylvain Preacutevost at Institut Laue Langevin (ILL Grenoble 729

France) are kindly acknowledged for their assistance on the SANS experiment Confocal 730

images were performed at the Institut de Biologie Paris Seine (IBPS) imaging facility Jean-731

Franccedilois Gilles (Imaging Core Facility Institut de Biologie Paris Seine (IBPS) CNRS 732

Sorbonne Universiteacute Paris France) is kindly acknowledged for his assistance on the confocal 733

microscopy imaging Dr Guillaume Laurent (LCMCP Sorbonne Universiteacute Paris France) is 734

kindly acknowledged for helpful discussions The French region Ile-de-France SESAME 735

program is acknowledged for financial support (700 MHz NMR 736

spectrometer) Dr Olivier Diat (ICSM Marcoule France) Prof eronikaIglič and Prof V Aleš737

onsul discussi Slovenia) are kindly acknowledged for helpfUniversity of LjubljanaIglič (-Kralj 738

739

Financial support 740

WAXS experiments have been supported by Soleil Light Source Saint Aubin France proposal 741

Ndeg 20201747 SANS experiments have been supported by ILL proposal Ndeg9-13-778 742

743

References 744

1 D Kabaso N Bobrovska W Goacuteźdź E Gongadze V Kralj-Iglič R Zorec and A 745

Iglič Bioelectrochemistry 2012 87 204ndash210 746

2 C R Safinya U Raviv D J Needleman A Zidovska M C Choi M A Ojeda-747

Lopez K K Ewert Y Li H P Miller J Quispe B Carragher C S Potter M W 748

Kim S C Feinstein and L Wilson Adv Mater 2011 23 2260ndash2270 749

3 P Veranič M Lokar G J Schuumltz J Weghuber S Wieser H Haumlgerstrand V Kralj-750

Iglič and A Iglič Biophys J 2008 95 4416ndash4425 751

4 A Rustom R Saffrich I Markovic P Walther and H-H Gerdes Science (80- ) 752

2004 303 1007ndash1010 753

5 M I Angelova S Soleacuteau P Meacuteleacuteard F Faucon and P Bothorel Prog Colloid 754

Polym Sci 1992 89 127ndash131 755

6 L Mathivet S Cribier and P F Devaux Biophys J 1996 70 1112ndash1121 756

7 V Kralj-Iglič G Gomišček J Majhenc V Arrigler and S Svetina Colloids Surfaces 757

A Physicochem Eng Asp 2001 181 315ndash318 758

30

8 V Kralj-Iglič A Iglič G Gomišček F Sevšek V Arrigler and H Haumlgerstrand J 759

Phys A Math Gen 2002 35 1533ndash1549 760

9 M Drab D Stopar V Kralj-Iglič and A Iglič Cells 2019 8 626 761

10 A Zidovska K K Ewert J Quispe B Carragher C S Potter and C R Safinya 762

Langmuir 2009 25 2979ndash2985 763

11 A Zidovska K K Ewert J Quispe B Carragher C S Potter and C R Safinya Soft 764

Matter 2011 7 8363ndash8369 765

12 A Zidovska K K Ewert J Quispe B Carragher C S Potter and C R Safinya 766

Biochim Biophys Acta - Biomembr 2009 1788 1869ndash1876 767

13 M Karlsson K Sott M Davidson A S Cans P Linderholm D Chiu and O Orwar 768

Proc Natl Acad Sci U S A 2002 99 11573ndash11578 769

14 A Karlsson R Karlsson M Karlsson A-S Cans A Stromberg F Ryttsen and O 770

Orwar Nature 2001 409 150ndash152 771

15 M Karlsson K Sott A S Cans A Karlsson R Karlsson and O Orwar Langmuir 772

2001 17 6754ndash6758 773

16 E Evans H Bowman A Leung D Needham and D Tirrell Science (80- ) 1996 774

273 933ndash935 775

17 U Seifert Configurations of fluid membranes and vesicles 1997 vol 46 776

18 J Kaumls and E Sackmann Biophys J 1991 60 825ndash844 777

19 N Bobrovska W Goacuteźdź V Kralj-Iglič and A Iglič PLoS One 2013 8 e73941 778

20 A Iglič B Babnik K Bohinc M Fošnarič H Haumlgerstrand and V Kralj-Iglič J 779

Biomech 2007 40 579ndash585 780

21 K Bohinc T Slivnik A Iglič M Brumen and V Kralj-Iglič J Phys Chem C 2007 781

111 9709ndash9718 782

22 A Iglič H Haumlgerstrand P Veranič A Plemenitaš and V Kralj-Iglič J Theor Biol 783

2006 240 368ndash373 784

23 V Kralj-Iglič and P Veranič Adv Planar Lipid Bilayers Liposomes 2006 5 129ndash785

149 786

24 A Iglic and V Kralj-Iglic Adv Planar Lipid Bilayers Liposomes 2006 4 253ndash279 787

25 S Penič L Mesarec M Fošnarič L Mroacutewczyńska H Haumlgerstrand V Kralj-Iglič and 788

A Iglič Front Phys 2020 8 342 789

26 P Yager and P E Schoen Mol Cryst Liq Cryst 1984 106 371ndash381 790

27 T G Barclay K Constantopoulos and J Matisons Chem Rev 2014 114 10217ndash791

10291 792

31

28 B Božič G Gomišček V Kralj-Iglič S Svetina and B Žekš Eur Biophys J 2002 793

31 487ndash496 794

29 A Iglič H Haumlgerstrand M Bobrowska-Haumlgerstrand V Arrigler and V Kralj-Iglič 795

Phys Lett A 2003 310 493ndash497 796

30 R Bar-Ziv and E Moses Phys Rev Lett 1994 73 1392ndash1395 797

31 S Chiruvolu H E Warriner E Naranjo S H Idziak J O Raumldler R J Plano J A 798

Zasadzinski and C R Safinya Science (80- ) 1994 266 1222ndash1225 799

32 V Kralj-Iglič A Iglič H Haumlgerstrand and P Peterlin Phys Rev E 2000 61 4230ndash800

4234 801

33 L Mesarec M Drab S Penič V Kralj-Iglič and A Iglič Int J Mol Sci 2021 22 802

2348 803

34 L Van Renterghem S L K W Roelants N Baccile K Uyttersprot M C Taelman 804

B Everaert S Mincke S Ledegen S Debrouwer K Scholtens C V Stevens W 805

Soetaert L Van Renterghem S L K W Roelants N Baccile K Uytersprot M C 806

Taelman B Everaert S Mincke S Ledegen S Debrouwer K Scholtens C V 807

Stevens and W Soetaert Biotechnol Bioeng 2018 115 1195ndash1206 808

35 J Clark and F Deswarte Introduction to Chemicals from Biomass Second Edition 809

John Wiley amp Sons 2015 810

36 K V Sajna R Houmlfer R K Sukumaran L D Gottumukkala and A Pandey in 811

Industrial Biorefineries and White Biotechnology eds A Pandey R Houmlfer M 812

Taherzadeh K M Nampoothiri and C Larroche Amsterdam Oxford Waltham 813

Elsevier 2015 pp 499ndash521 814

37 R Marchant and I M Banat Trends Biotechnol 2012 30 558ndash565 815

38 K Hill and C LeHen-Ferrenbach in Sugar-Based Surfactants Fundamentals and 816

Applications ed C C Ruiz Boca Raton CRC Press 2009 pp 1ndash20 817

39 L Rodrigues I M Banat J Teixeira and R Oliveira J Antimicrob Chemother 818

2006 57 609ndash618 819

40 J D Desai and I M Banat Microbiol Mol Biol Rev 1997 61 47ndash64 820

41 S L K W Roelants K Ciesielska S L De Maeseneire H Moens B Everaert S 821

Verweire Q Denon B Vanlerberghe I N A Van Bogaert P Van der Meeren B 822

Devreese and W Soetaert Biotechnol Bioeng 2016 113 550ndash559 823

42 N Baccile M Selmane P Le Griel S Preacutevost J Perez C V Stevens E Delbeke S 824

Zibek M Guenther W Soetaert I N A Van Bogaert and S Roelants Langmuir 825

2016 32 6343ndash6359 826

32

43 N Baccile A-S Cuvier S Preacutevost C V Stevens E Delbeke J Berton W Soetaert 827

I N A Van Bogaert and S Roelants Langmuir 2016 32 10881ndash10894 828

44 N Baccile C Seyrig A Poirier S A Castro S L K W Roelants and S Abel Green 829

Chem 2021 23 3842ndash3944 830

45 A Salis L Cappai C Carucci D F Parsons and M Monduzzi J Phys Chem Lett 831

2020 11 6805ndash6811 832

46 J Schindelin I Arganda-Carreras E Frise V Kaynig M Longair T Pietzsch S 833

Preibisch C Rueden S Saalfeld B Schmid J-Y Tinevez D J White V 834

Hartenstein K Eliceiri P Tomancak and A Cardona Nat Methods 2012 9 676ndash835

682 836

47 T Parasassi and E Gratton J Fluoresc 1995 5 59ndash69 837

48 M M Dodes Traian F L Gonzaacutelez Flecha and V Levi J Lipid Res 2012 53 609ndash838

616 839

49 D M Owen C Rentero A Magenau A Abu-Siniyeh and K Gaus Nat Protoc 840

2012 7 24ndash35 841

50 W Stillwell in An Introduction to Biological Membranes Academic Press 2016 pp 842

135ndash180 843

51 W Stillwell in An Introduction to Biological Membranes Academic Press 2016 pp 844

181ndash220 845

52 B Tenchov Chem Phys Lipids 1991 57 165ndash177 846

53 R Koynova and M Caffrey Biochim Biophys Acta 1998 1376 91ndash145 847

54 F Schmid S Dolezel O Lenz and S Meinhardt J Phys Conf Ser 2014 487 848

012004 849

55 S Meinhardt R L C Vink and F Schmid Proc Natl Acad Sci U S A 2013 110 850

4476ndash4481 851

56 O Glatter and O Kratky Small Angle X-ray Scattering Academic Press London 852

1982 853

57 B A Cunningham A D Brown D H Wolfe W P Williams and A Brain Phys 854

Rev E 1998 58 3662ndash3672 855

58 M Nadler A Steiner T Dvir O Szekely P Szekely A Ginsburg R Asor R Resh 856

C Tamburu M Peres and U Raviv Soft Matter 2011 7 1512ndash1523 857

59 C Tanford The hydrophobic effect Formation of micelles and biological membranes 858

John Wiley amp Sons Inc New York 1973 859

60 M Masuda and T Shimizu Langmuir 2004 20 5969ndash5977 860

33

61 A Zidovska K K Ewert J Quispe B Carragher C S Potter and C R Safinya 861

Block Liposomes Vesicles of Charged Lipids with Distinctly Shaped Nanoscale 862

Sphere- Pear- Tube- or Rod-Segments Elsevier Inc 1st edn 2009 vol 465 863

62 A Iglic V Kralj-Iglic B Božič M Bobrowska-Haumlgerstrand B Isomaa and H 864

Haumlgerstrand Bioelectrochemistry 2000 52 203ndash211 865

63 T Shimizu M Masuda and H Minamikawa Chem Rev 2005 105 1401ndash1444 866

64 J V Selinger and J M Schnur Phys Rev Lett 1993 71 4091ndash4094 867

65 J M Schnur Science (80- ) 1993 262 1669ndash1676 868

66 B Tenchov R Koynova and G Rapp Biophys J 2001 80 1873ndash1890 869

67 R Koynova and B Tenchov Ser Biomech 2017 31 3ndash19 870

68 S J Angyal and R Le Fur Carbohydr Res 1984 126 15ndash26 871

69 J K Whitesell and M A Minton J Am Chem Soc 1987 109 225ndash228 872

70 T Takayama S Ando and I Ando J Mol Struct 1990 220 245ndash250 873

71 K Gawrisch N V Eldho and I V Polozov Chem Phys Lipids 2002 116 135ndash151 874

72 F Tielens C Gervais J F Lambert F Mauri and D Costa Chem Mater 2008 20 875

3336ndash3344 876

73 G A Jeffrey and Y Yeon Acta Crystallogr Sect B 1986 42 410ndash413 877

74 B Berglund and R W Vaughan J Chem Phys 1980 73 2037ndash2043 878

75 C Gervais C Coelho T Azaiumls J Maquet G Laurent F Pourpoint C Bonhomme 879

P Florian B Alonso G Guerrero P H Mutin and F Mauri J Magn Reson 2007 880

187 131ndash140 881

76 C Akita T Kawaguchi F Kaneko H Yamamoto and M Suzuki J Phys Chem B 882

2004 108 4862ndash4868 883

77 R N Purusottam L Seacutenicourt J J Lacapegravere and P Tekely Biochim Biophys Acta - 884

Biomembr 2015 1848 3134ndash3139 885

78 X Biarneacutes A Ardegravevol A Planas C Rovira A Laio and M Parrinello J Am Chem 886

Soc 2007 129 10686ndash10693 887

79 L M J Kroon‐Batenburg and J Kroon Biopolymers 1990 29 1243ndash1248 888

80 K N Kirschner and R J Woods Proc Natl Acad Sci U S A 2001 98 10541ndash889

10545 890

81 S Svenson B Kirste and J H Fuhrhop J Am Chem Soc 1994 116 11969ndash11975 891

82 I Sack S Macholl J H Fuhrhop and G Buntkowsky Phys Chem Chem Phys 892

2000 2 1781ndash1788 893

83 S Svenson J Koening and J H Fuhrhop J Phys Chem 1994 98 1022ndash1028 894

34

84 S Svenson A Schaefer and J H Fuhrhop J Chem Soc Perkin Trans 2 1994 2 895

1023ndash1028 896

85 K Bock and C Pedersen Adv Carbohydr Chem Biochem 1983 41 27ndash66 897

86 T B Grindley Structure and Conformation of Carbohydrates Springer Berlin 898

Heidelberg 2008 899

87 F Tantakitti J Boekhoven X Wang R V Kazantsev T Yu J Li E Zhuang R 900

Zandi J H Ortony C J Newcomb L C Palmer G S Shekhawat M O De La 901

Cruz G C Schatz and S I Stupp Nat Mater 2016 15 469ndash476 902

903

1

Supporting Information 904

905

Topological connection between vesicles and nanotubes in single-906

component lipid membranes driven by head-tail interactions 907

908

Niki Baccilea Ceacutedric Lorthioira Abdoul Aziz Baa Patrick Le Griela Cristina Coelhob Javier 909

Perezc Wim Soetaertd Sophie L K W Roelantsd 910

911

a Sorbonne Universiteacute Centre National de la Recherche Scientifique Laboratoire de Chimie de 912

la Matiegravere Condenseacutee de Paris LCMCP F-75005 Paris France 913

b Institut des Mateacuteriaux de Paris Centre Sorbonne Universiteacute Paris France 914

c Synchrotron Soleil LrsquoOrme des Merisiers Saint-Aubin BP48 91192 Gif-sur-Yvette Cedex 915

France 916

d InBio Department of Biotechnology Ghent University Ghent Belgium 917

918

919

Corresponding author 920

Dr Niki Baccile 921

E-mail address nikibaccilesorbonne-universitefr 922

Phone +33 1 44 27 56 77 923

924

925

926

2

a) 927

Parameter Method

Dry Matter (DM ) 996 Infrared balance 105degC

Glucose () 012 HPLC-Metacarb

Glycerol () nd HPLC-Metacarb

Free Fatty Acid content () lt001 Internal method BBEPP

GC with FID Detector

Oil () lt001 Internal method BBEPP

GC with FID Detector

Moulds (CFUg) lt10 3M-Nordval ndeg16 B

Yeasts (CFUg) lt10 3M-Nordval ndeg16 B

Anaerobic count (CFUg) lt10 SP-VG M005 B

Protein () tbd BCA protein assay

928

Figure S 1 - a) HPLC-ELSD chromatogram and composition table 929

930

931

932

933

3

b) 934

935

Group Ndeg δ ppm

CH2 1 354

CH2 2 153

CH2 3-712-14 132

CH2 811 203

CH 910 534

CH2 15 162

CH2 16 388

CH 1 424

CH 2 316

CH 3 334

CH 4 329

CH 5 354

CH2 6 366

Figure S 1 (next) ndash b) 1H NMR spectrum (MeOD-d4) recorded on the purified G-C181-OH compounds 936

and related assignment 937

938

2

811910 1rsquo

5rsquo1

3rsquo4rsquo

1rsquo1910

asymp asymp asympHDO MeOD Acetone

6rsquo2rsquo

1615

3-712-14

4

939

940

Figure S 2 ndash Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 130degC 941

5

942

943

Figure S 3 - Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 90degC 944

945

90degC

1

22

6

946

947

948

949

Figure S 4 ndash Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 70degC 950

951

21

11

1

2

5

3

4

45

5

6

7

7

7

952

953

954

Figure S 5 ndash Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 25degC 955

956

957

1

1 1

8

958

Figure S 6 ndash a) SANS experiment recorded at 90degC for a 5 mgmL G-C181-OH aqueous solution b-c) 959

WAXS experiments recorded on aqueous solutions of G-C181-OH at b) 5 mgmL and c) 25 mgmL d) 960

Structural interpretation of the repeating distances pointed out in c) 961

962

963

964

001 01 1

q Aring-1

25degC

68degC

88degC

111degC

01 1q Aring

-1

02 04 06 08 1q Aring

-1

001 01

10-1

100

101

Inte

nsity

cm

-1

q Aring-1

-2

T= 90degC

SANS WAXS WAXS

a) b) c)

dv

dL d= 144 Aring-1

di

ddH2O

dt

dvdL

dtdi d

d)

9

965

966

Figure S 7 ndash ssNMR MAS (10 kHz) experiments recorded on a G-C181-OH dry powder a) 1H spectra 967

recorded from 110degC to 70degC b) 13C single pulse (SP) spectra recorded with high power 1H decoupling from 968

110degC to 70degC c) 13C cross polarization (CP) spectra recorded at 110degC and contact time tc of 50 micros and 969

3000 micros d) 13C CP spectra recorded at 70degC and for 20 lt tc micros lt 3000 970

971

1

811

910

CH2 CH2

1rsquo

6rsquo

8 6 4 2 0 ppm

110

90

80

70

910

1rsquo

6rsquo16

1

2rsquo3rsquo4rsquo

811

OH

45 40 35 30 25 20 15

ppm

200 180 160 140 120 100 80 60 40 20 0

ppm140 120 100 80

ppm

CH2-gauche

3004

110

90

80

70

65

60

50

25

C91013003

C9 - 12869

CH2-all-trans

3198

C1rsquo10335

a) b)

1H

T degCT degC

13C Single Pulse MAS

A

B

A

B

70degC70degC

5rsquo

215

200 180 160 140 120 100 80 60 40 20 0

ppm

13C CPMAST= 110degC

200 180 160 140 120 100 80 60 40 20 0

ppm

132 131 130 129 128 127 126

ppm

106 105 104 103 102

ppm

13C CPMAST= 70degC

50

tc micros

3000

20

tc micros

50 500 1000 3000

tc micros

C102

13024 C910113003

C92

12869

C1rsquo213082

C1rsquo110443

A B

106 105 104 103 102

ppm

3000

1000

500

50

20

AB

c)

d)

10

972

Figure S 8 ndash a) Short and d) long T2 components displayed by the 1H transverse relaxation signal c) 1H 973

transverse relaxation signals recorded on a G-C181-OH dry powder from 110degC to 70degC The area of each 974

given resonance is plotted (log-lin) as a function of the echo time (CPMG pulse sequence) for each 975

temperature and fitted using the equation given above T2s and T2l indicate the short and long T2 relaxation 976

times respectively obtained by describing the 1H transverse relaxation signal using a biexponential decay 977

For H1rsquo H5rsquo and H6rsquo the 1H relaxation function could be described by a monoexponential decay 978

- Protons from the sugar group (H1rsquo H5rsquo H6rsquo) show a single T2 relaxation component (Figure 979

6c and Figure S 8) while protons of the acyl chain (CH2 H910 H811) show both short (Figure 980

6c) and long (Figure 6d) T2 relaxation component 981

- The short T2 relaxation time systematically decreases from 110degC to 70degC for all species by a 982

factor ranging from two to seven depending on the specific chemical group Very long T2 are 983

observed at 110degC for all acyl species while nearly unchanged values are measured below 984

90degC without any clear trend 985

- The sugar protons (H1rsquo H5rsquo H6rsquo) have the shortest set of T2 values (lt 1 ms) below 100degC 986

The long component of T2 generally ranges between 4 ms and 12 ms 987

0 1 2 3 4 5 6 7 8

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1 2 3 4 5 6 7 8

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

H1

H8H11

H9H10

CH2

H1rsquo

H6rsquo

CH2

H910H811 H1CH2

H1rsquo H5rsquo6rsquo

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

OH

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45 T= 110degC

T= 90degC

T= 80degC

T= 70degC

Log

In

ten

sity

Time ms

Fit

)

CH2 H910 H811 H1 H56 H1 OH

00

05

10

15

20

25

30

35

40

T2 m

s

T= 110degC

T= 90degC

T= 80degC

T= 70degC

CH2 H910 H811 H1

0

2

4

6

8

10

12

T2 m

s

T= 110degC

T= 90degC

T= 80degC

T= 70degC

Short T2(1H) Long T2(1H)

a) b)

c)

11

Table S 1 ndash 1H chemical shifts of G-C181-OH in solution and in the solid-state Solid-state spectra are 988

recorded at a MAS frequency of 10 kHz in DMSO-d6 broad peak 989

990

δH ppm

Group Ndeg Liquid

(MeOD-d4) Solid

(T= 110degC) Solid

(T= 70degC)

CH2 H1 354 357

CH2 H2 153 156

CH2 H3-712-14 132 135

CH2 H811 203 203

CH H910 534 534

CH2 H15 162 164

CH2 H16 388 383

CH H1 424 432

CH H2 316 335

CH H3 334 335

CH H4 329 335

CH H5 354 351 357

CH2 H6 366 383

OH OH1 432

45 463

OH OH2 492

OH OH3 490

OH OH4 487

OH OH6 445

991

992

H1H9H10H1rsquo

H6rsquo

OH2rsquo

OH1

12

Detailed discussion about the NOESY and DQ-SQ 2D 1H-1H ssNMR experiments presented in 993

Figure 9 994

995

2D 1H-1H ssNMR experiments are a unique tool to explore through-space proximities 996

between 1H nuclei ssNMR offers a large number of possibilities each adapted to a given 997

system The advantage of 2D maps with respect to a dynamics approach is their direct 998

readability in clear two nuclei which exchange their magnetization through-space (whichever 999

mechanism may be responsible for this exchange) will display a specific cross-peak However 1000

the way cross-peaks are presented and interpreted strictly depends on the nature of the pulse 1001

program which in turn is chosen according to the physical nature of the sample 1002

1D temperature-dependent 1H NMR experiments performed on G-C181-OH (Figure 1003

4a in the main text) indicate sharp peaks at 90degC and broad peaks at 70degC Narrow peaks 1004

indicate fast andor large-amplitude molecular tumbling averaging the strong 1H-1H dipolar 1005

interactions This excludes the use of pulse sequences like DQ-SQ which is not efficient under 1006

these conditions For this reason we employ spin-diffusion-based pulse sequences like 1007

NOESY which is more adapted for samples with high molecular mobility When mixing times 1008

are chosen to be short (lt 1 s) one generally explores through-space proximities within about 5 1009

10 Aring On the contrary when molecular motions are restricted dipolar coupling becomes 1010

significant 1H NMR peaks are broader and pulse sequences like DQ-SQ can now be used 1011

with a spatial resolution of less than 5 Aring 1012

Whenever possible DQ-SQ experiments are preferred over spin diffusion-based pulse 1013

sequences because they are more selective in particular for likewise interactions which are 1014

not discriminated in NOESY-type experiments This is illustrated by the typical interaction 1015

schemes for NOESY (top Figure 9a) and DQ-SQ (top Figure 9b) For two given A and B 1016

protons 2D NOESY maps will display two on-diagonal and two square-shaped symmetrical 1017

off-diagonal cross-peaks Off-diagonal cross-peaks illustrate the mutual A-B and B-A 1018

interaction but the on-diagonal peaks are systematic and do not prove any A-A or B-B 1019

interaction Similarly to NOESY 2D DQ-SQ maps display off-diagonal peaks (symmetric 1020

across the diagonal top Figure 9b) upon A-B and B-A interaction but in contrast they will 1021

display on-diagonal peaks only in the presence of self-interaction In the tutorial example of 1022

Figure 9b top the on-diagonal A-A peak indicates mutual A-A interaction while lack of on-1023

diagonal B-B peak evidences the absence of B-B proximities Combining NOESY and DQ-SQ 1024

experiments is capital to study the local arrangement of G-C181-OH and its evolution with 1025

temperature 1026

13

At T= 90degC the H1 and H5rsquo contributions are very close but possible to discriminate 1027

them (δ(H1)= 354 ppm δ(H5rsquo)= 351 ppm) according to 1H NMR chemical shifts in Table S 1028

1 2D 1H-1H NOESY in Figure 9a also discriminates between the two components attributed 1029

on the basis of the interactions with their mutual neighbors (green H1-H2 and H5rsquo-H1rsquo squares 1030

in Figure 9a) On this basis 2D NOESY does not show a direct H1rsquo-H1 correlation (blue red-1031

crossed) but rather an extensive network of intra-sugar interactions These features exclude 1032

suggest intra- and inter-sugar interactions and they exclude head-tail arrangements as a major 1033

molecular configuration at 90degC 1034

At 70degC the 1H NMR peaks are broader indicating a reduction of the T2 (1H) relaxation 1035

times andor residual homonuclear dipolar interactions excluding the use of NOESY In this 1036

case 1H-1H DQ-SQ correlation experiments are best suited Figure 9b presents the 1H-1H DQ-1037

SQ experiment on G-C181-OH at 70degC Peaks are broader than in NOESY and attribution may 1038

be cumbersome However the experiment is performed at 700 MHz and MAS= 22 kHz which 1039

improves spectral resolution with respect to the same experiment performed at 300 MHz and 1040

MAS= 10 kHz The auto-peak at δH= 356 ppm in the SQ dimension is attributed to H1-H1 spin 1041

pairs mainly resulting from intra CH2 interaction while the DQ cross-peaks at δH= 156-356 1042

ppm and δH= 432-356 ppm respectively origin from H2-H1 and H1rsquo-H1 interactions (Figure 1043

9b) DQ-SQ map does not show any auto-peak corresponding to H1rsquo-H1rsquo or Hxrsquo-Hxrsquo (x= 2rsquo 1044

3rsquo 4rsquo) interactions (red-crossed blue Figure 9b) thus excluding any proximity (lt 5 Aring) between 1045

close glucose neighbors On the contrary DQ-SQ maps (recorded both at 700 MHz Figure 9b 1046

or at 300 MHz not shown) systematically show DQ cross-peaks which can only be attributed 1047

to H1rsquo-H1 interactions (δH= 432-356) The presence of these H1rsquo-H1 interactions can only be 1048

explained by a head-tail arrangement of the G-C181-OH molecules as illustrated below the 1049

DQ-SQ map in Figure 9b 1050

5

cooled at to the desired temperature For the SAXS experiment the sample was heat at 90degC in 118

the block heating device and cooled at to 25degC before analysis For the SANS experiment the 119

sample was heat at 90degC directly in front of the beam using the thermalized sample holder 120

available 121

122

Differential Scanning Calorimetry (DSC) DSC was performed using a DSC Q20 apparatus 123

from TA Instruments equipped with the Advantage for Q Series Version acquisition software 124

(v540) Acquisition was performed on both hydrated and dry powder sample (~ 3-5 mg) sealed 125

in a classical aluminium cup and using an immediate sequence of heating and cooling ramps at 126

a rate of 10degCmin-1 Melting temperatures Tm 1 and 2 Tm1 and Tm2 were taken at the minimum 127

of the endothermic peak 128

129

Small and Wide Angle Scattering experiments Small angle neutron scattering (SANS) 130

experiments were performed at the D11 beamline of Institut Laue Langevin (Grenoble France) 131

during the run No 9-13-778 Four q-ranges have been explored and merged using the following 132

wavelengths λ and sample-to-detector (StD) distances 1) ultra-low q λ= 135Aring StD= 39 m 133

2) low-q λ= 53Aring StD= 39 m 3) mid-q λ= 53Aring StD= 8 m 4) high-q λ= 53Aring StD= 14 m 134

The sample (C= 5 mgmL-1) was prepared in 999 D2O to limit the incoherent background 135

scattering The sample solution was analyzed in standard 1 mm quartz cells Direct beam empty 136

quartz cell H2O (incoherent scatterer) within the quartz cell were recorded and boron carbide 137

(B4C) was used as neutron absorber The sample acquisition was measured at 90degC where 138

temperature was controlled through the controller thermalized sample holder available at the 139

beamline The background sample (D2O) signal was subtracted from the experimental data 140

Absolute values of the scattering intensity were obtained from the direct determination of the 141

number of neutrons in the incident beam and the detector cell solid angle The 2D raw data were 142

corrected for the ambient background and empty cell scattering and normalized to yield an 143

absolute scale (cross section per unit volume) by the neutron flux on the samples The data were 144

then circularly averaged to yield the 1D intensity distribution I(q) The software package Grasp 145

(developed at ILL and available free of charge) was used to integrate the data while the 146

software package SAXSUtilities (developed at ESRF and available free of charge) was used to 147

merge the data acquired at all configurations and subtract the background 148

Wide-angle X-ray scattering (WAXS) was performed under temperature control at the 149

SWING beamline of SOLEIL synchrotron facility (Saint-Aubin France) during the run 150

20201747 (energy 14 keV sample-to-detector distance 05 m) Two sample concentrations in 151

6

H2O were probed C= 5 mgmL-1 and C= 25 mgmL-1 The 2D data were integrated azimuthally 152

at the beamline using the software Foxtrot and in order to obtain the I(q) vs q spectrum after 153

masking the beam stop shadow Silver behenate (d(100) = 5838 Aring) was used as a standard to 154

calibrate the q-scale Sample solutions were inserted in borosilicate capillaries of 15 mm in 155

diameter Capillaries were flame-sealed A capillary oven with controlled temperature (plusmn 05degC) 156

provided at the beamline was used to control the sample temperature between 25degC and 111degC 157

Data were normalized by the transmission and calibrated to the SAXS signal of H2O at large q-158

values (I= 00163 cm-1) in order to obtain an absolute intensity scale The water signal was 159

measured by subtracting the signal of the empty capillary from the signal of a water-filled 160

capillary The signal of (water + capillary) was used as background for the samples and it was 161

subtracted after integration of the 2D data 162

163

Cryogenic transmission electron microscopy (cryo-TEM) Cryo-TEM experiments were carried 164

out on an FEI Tecnai 120 twin microscope operating at 120 kV and equipped with a Gatan 165

Orius CCD numeric camera The sample holder was a Gatan Cryoholder (Gatan 626DH 166

Gatan) Digital Micrograph software was used for image acquisition Cryofixation was 167

performed using a homemade cryofixation device The solutions were deposited on a glow-168

discharged holey carbon coated TEM copper grid (Quantifoil R22 Germany) Excess solution 169

was removed and the grid was immediately plunged into liquid ethane at -180degC before 170

transferring them into liquid nitrogen All grids were kept at liquid nitrogen temperature 171

throughout all experimentation Cryo-TEM images have been treated and analyzed using Fiji 172

(is just ImageJ) software available free of charge at the developerrsquos website46 173

The samplersquos concentration was C= 5 mgmL-1 For technical reasons the temperature 174

during cryofixation could not be controlled For this reason the temperature associated with 175

each cryo-TEM image in the text was related to the equilibration temperature just before 176

cryofixation the device being as close as possible to the dry bath used for temperature control 177

Since temperature control during cryofixation was not possible in our device and practically 178

impossible at high temperatures one should take the T values as indicative of the given value 179

Although the sample transfer between the dry bath and the cryofixation device was performed 180

as fast as possible it is more than likely that the actual temperature of the sample at the moment 181

of cryofixation was lower than the one indicated here although impossible to know precisely 182

183

7

Fluorescence microscopy Images were recorded using a 40x objective on a Leica SP5 upright 184

confocal microscope with 405 nm laser excitation The sample was marked with 6-dodecanoyl-185

NN-dimethyl-2-naphthylamine (Laurdan Mw= 35354 gmol-1) fluorophore as follows A 186

stock solution of 3 mg of Laurdan in 100 microL of acetone was initially prepared (C= 30 mgmL-187

1) A volume of 028 microL of the stock Laurdan solution was diluted in 1 mL of a 5 mgmL-1 G-188

C181-OH water solution temporary set at 110degC in a sealed Eppendorf The latter was 189

important so to enhance inclusion of Laurdan in the membrane constituted by the glucolipid 190

while in its vesicle phase The final Laurdan concentration was approximately 24 micromol with a 191

molar Laurdan-to-G-C181-OH molar ratio of approximately 1-to-500 Fluorophore-to-lipid 192

ratios above 1200 were generally considered as non-interfering with the lipid phase behavior 193

as also verified in this work 194

Laurdan can be excited between 365 nm and 410 nm and its emission wavelength can 195

vary according to the rigidity of the membrane intercalation within a rigid or flexible 196

membrane generates emissions at 440 nm or 490 nm respectively47 In the present experiment 197

we have flame-sealed a solution of G-C181-OH at 5 mgmL-1 containing Laurdan (1-to-500 in 198

lipid-to-Laurdan ratio) in flat optical capillaries of 01 mm thickness Capillaries were set on a 199

Linkam support for temperature control eventually placed under the microscope The sample 200

was excited at 405 nm and detection was made in spectral mode with a 5 nm window sliding 201

every 5 nm from 412 nm to 602 nm in 16 bits which allows quantify the signal and compare 202

the different conditions Quantitative analysis was performed according to the literature4849 203

several ROI (range of interest) have been drawn on several images using the ROI manager of 204

Fiji (is just ImageJ) software The integrated intensity of each ROI at 440 nm and 490 nm was 205

eventually recorded and used to calculate the value of the generalized polarization GP as 206

follows of 207

119866119875 =(120582440 minus 120582490)

120582440 + 120582490 208

A negative GP indicates a flexible membrane while a positive GP indicates a rigid membrane 209

210

Solid-state Nuclear Magnetic Resonance (ssNMR) 1H and 13C ssNMR experiments have been 211

performed using an Avance III HD Bruker 705 T (ν1H= 300 MHz) spectrometer and a 4 mm 212

magic angle spinning (MAS) probe except for the BAck-to-BAck (BABA) experiment which 213

was performed on an Avance III Bruker 1644 T (ν1H= 700 MHz) spectrometer equipped with 214

25 mm MAS probe Temperature was controlled through a Smart cooler BCUIIa BCU-xtreme 215

unit using 10 min of equilibration time after each temperature variation Temperature was 216

8

always varied from high (110degC) to low (70degC) before analysis Due to combined use of fast 217

MAS rates necessary to achieve high resolution spectra and heating close to the boiling point 218

of water neither diluted nor concentrated samples could be used for ssNMR experiments All 219

tests employing up to 50 wt G-C181-OH in water resulted in sample centrifugation drying 220

but also expulsion from the rotor causing sudden stop of rotor spinning For these reasons all 221

attempts to study hydrated G-C181-OH failed and only the dry powder could be analyzed The 222

dry sample powder was analyzed as such in zirconia rotors (4 mm or 25 mm) spinning at MAS= 223

10 kHz (4 mm) or 22 kHz (25 mm) and Chemical shifts were calibrated with respect to 224

adamantane (13C 13C(CH2)= 3848 ppm = 191 ppm) and glycine alpha (13C(CO)= 17603 225

ppm) Details on the acquisition parameters are given in the Supporting Information 226

227

Results and discussion 228

Topologically-connected nanotubes and vesicles 229

G-C181-OH is a biobased glucolipid water-insoluble at room temperature and of 230

which the physicochemical properties are poorly known The DSC thermogram performed at 231

10degCmin of G-C181-OH as a function of the water content (Figure 2a) shows two endothermic 232

peaks on the dry powder (996 wt glucolipid) They are attributed to two melting phenomena 233

occurring at the following temperatures Tm 483degC (Tm1) and 957degC (Tm2) the enthalpy of the 234

former being lower than that of the latter Interestingly DSC also shows that Tm1 is hydration 235

independent and reversible (the associated exothermic peak is systematically observed) while 236

the transition at Tm2 is simply suppressed (here tested up to 130degC) on hydrated powders and 237

never observed on the cooling profile even at slow cooling rates (1degCmin) The thermogram 238

recorded on the dry powder is typical of single-component lipids having an intermediate50 239

metastable phase (often referred to the ripple phase Pβ´ in reference to the periodic undulation 240

of the bilayer)50ndash54 between the classical lamellar gel (Lβ) and liquid crystalline (Lα) phases50ndash241

54 and characterized by regions of both liquid order and disorder5455 The effect of water does 242

not influence the low-temperature crystallization at Tm1 but it contributes to disrupt the ordered 243

array of hydrocarbon chains in the metastable phase 244

On the basis of the DSC thermogram the aqueous phase behavior of G-C181-OH under 245

diluted conditions (5 mgmL-1) was studied at four different temperatures (red marks on Figure 246

2a) below and above Tm1 but also between Tm1 and Tm2 corresponding to the thermogram of 247

the dry powder 248

9

249

Figure 2 ndash a) DSC thermogram of G-C181-OH under dry (04 wt water estimated by thermogravimetric 250

analysis) and hydrated conditions recorded at a heating rate of 10degCmin Red marks indicate the 251

temperatures at which cryo-TEM experiments are performed on the corresponding 5 mgmL aqueous 252

solutions b) Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 130degC 253

90degC 70degC and 25degC T indicates the temperature of the sample solution before cryofixation Tne T T 254

being the temperature of the sample upon plunging in liquid ethane Please refer to the materials and 255

methods section for more information 256

T= 130degC

T= 90degC

T= 70degC

T= 25degC

a)

b)

20 40 60 80 100 120

99 wt

50 wt

25 wt

T degC

En

do

10 wt

Tm1= 483 degC

Tm2= 957 degC

10

257

The series of cryo-TEM images in Figure 2b show an overview of the most relevant 258

structures observed on G-C181-OH solution from T= 130degC to T= 25degC whereas T is 259

indicative of the equilibrium temperature before cryofixation (refer to materials and methods 260

section for more information on the sense and interpretation of T) At Tgt 100degC the sample 261

is essentially composed of polydisperse single wall vesicles massively shown by the 262

complementary pictures presented in Figure S 2 In between Tm1 and Tm2 at T= 90degC and T= 263

70degC (Figure 2b Figure 3 Figure 4 complemented by Figure S 3 and Figure S 4) vesicles are 264

connected by nanotubes At Tlt Tm1 flat crystals are otherwise observed (Figure 2b and Figure 265

S 5) 266

267

268

Figure 3 ndash Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 90degC 269

270

A deeper insight at T= 90degC is given in Figure 3 (more images are provided in Figure 271

S 3) Arrows 1 through 6 mark very clearly the typical continuity in the glucolipid membrane 272

between the vesicle and nanotubes Arrow Ndeg 7 shows that a single nanotube connects the outer 273

membrane of two vesicles several hundred nanometers apart The connection is shown by 274

2

34

5

6

6

1

5

7

A

B

D

C

0 10 20 30 40 50 60

BIn

tensity

au

Profile nm

A

0 10 20 30 40 50 60

Inte

nsity

au

Profile nm

C

D

T= 90degC

11

arrows Ndeg5 and Ndeg6 The corresponding profiles A through D confirm the tubular nature of 275

the elongated objects Profiles B and D are both characterized by typical symmetrical thick 276

walls (low intensity depths highlighted by shaded rectangles) surrounding a hollow region (high 277

intensity peak between the shaded rectangles) The wall thickness varies between 35 and 40 278

nm for profile B and between 41 and 61 nm for profile D while the total nanotube diameter 279

varies between 14 nm at B and 20 nm at D At the same time the thickness of the vesicle wall 280

is about 6 nm at A and 135 nm at C Considerations about the bilayer or interdigitated structure 281

of the membrane will be presented below 282

The connection between nanotubes and vesicles seems to be multiple For instance 283

arrow Ndeg1 (Figure S 3) and Ndeg4 (Figure 3) show a single-wall vesicle of which the membrane 284

is extruded into a nanotube Similarly a closer look at arrows Ndeg1 2 and 3 in Figure 3 show 285

that the vesicle is single-wall and fully closed while the nanotube nucleates at the vesicle outer 286

surface with no apparent topological continuity In many regions of the sample at T= 90degC 287

one can find exvaginations or buds232531 in the nanotube wall This is shown for instance by 288

arrows Ndeg7 and particularly put in evidence in Figure 4 by arrows Ndeg1 through Ndeg4 where the 289

amplitude of the oscillation varies between 20-25 nm (arrow Ndeg1) and 6-8 nm (arrows Ndeg3-4) 290

291

Figure 4 - Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heated at T= 90degC 292

293

Still between Tm2 and Tm1 but at lower temperature (T= 70degC) the nanotubes seem to 294

vary from single threads (arrow Ndeg1 Figure S 4) to nanotubes of diameter below 15 nm both 295

connecting nanotubes of diameter ranging between 50 nm and 100 nm (arrows Ndeg2 Figure S 296

4) The larger tubes are most likely formed by the inflation and flattening of the nanotubes 297

probably after assimilation of surrounding vesicles and fusion of the exvagination discussed 298

above The transition between vesicles tubes and flat membranes is certainly more accentuated 299

Arrow Ndeg3 in Figure S 4 shows distinct vesicles while arrow Ndeg4 identifies a flat structure 300

evidently formed from fused vesicles In particular arrow Ndeg5 points at the fusion of two 301

2

1 132

4

12

vesicles while arrow Ndeg6 points at two vesicles fusing into the flat membrane Additional areas 302

where vesicle fusion is prominent are indicated by the oscillations at the outer skirt of broad 303

lamellae and pointed by arrow Ndeg7 304

Finally below Tm1 at T= 25degC for instance mainly flat crystals with no visible 305

exvaginations or tubes are visible (Figure S 5) by cryo-TEM and the material turns into an 306

insoluble precipitate Massive presence of flat crystalline structures is also clear from the bulk 307

iridescent behavior of a G-C181-OH solution under manual shear 308

In summary despite the uncertainty related to the actual state of equilibrium of the 309

glucolipids at the moment of cryofixation cryo-TEM depicts a vesicle phase at Tgt 100degC a 310

flat lamellar phase below Tm1 and an interconnected vesicle-nanotube network between Tm1 and 311

about 100degC The combination of statistically-relevant fluorescence microscopy and 312

scatteringdiffraction experiments provides an additional proof of the coexistence of 313

vesiclenanotube systems between the two transition temperatures Please note that the 314

experiments reported below have been performed under controlled conditions of temperature 315

which is then indicated by the use of T instead of T 316

13

317

Figure 5 ndash a-c) Confocal fluorescence microscopy images recorded at 90degC and 70degC for a 5 mgmL G-318

C181-OH aqueous solution d-g) Typical fluorescence microscopy images recorded at T= 80degC using 319

Laurdan (refer to materials and methods section for more details on this fluorophore) Excitation 320

wavelength is 405 nm for all the images while emission wavelength is 490 nm in (df) and 440 nm in (eg) 321

h) Full emission spectra obtained from the integration of various fluorescence images recorded from 412 322

nm to 600 nm (step 5 nm) at 90degC 80degC and 70degC i) Box plot of GP defined as 119918119927 =(120640120786120786120782minus120640120786120791120782)

120640120786120786120782+120640120786120791120782 with 120640120786120786120782 323

and 120640120786120791120782 being the emission wavelengths at 440 nm and 490 nm at 90degC 80degC and 70degC Each data point 324

l= 490 nm l= 440 nm

l= 490 nm l= 440 nm

10 μm400 450 500 550 600

00

02

04

06

08

10

No

rma

lize

d I

nte

nsity

l nm

90degC

80degC

80degC

70degC

70degC

400 450 500 550 60000

02

04

06

08

10

No

rma

lize

d I

nte

nsity

l nm

90degC

80degC

80degC

70degC

70degC

d) e)

f) g)

h)

i)

T= 80degC

Laurdan

T= 80degC

T= 80degC T= 80degC

T= 90degC T= 70degC T= 70degC

a) b) c)

1

1

1

14

corresponds to a set of regions of interest (ROIs) (54 at 90degC 65 at 80degC and 75 at 70degC) selected in the 325

corresponding fluorescence images ROIs were representatively selected so to include the entire image but 326

also vesicles alone tubes alone and mixtures of vesicles and tubes 327

328

Figure 5ab shows two typical confocal fluorescence microscopy images recorded on a 329

G-C181-OH solution (5 mgmL) in a flame-sealed capillary heat at T= 90degC and T= 70degC 330

Although the resolution of optical microscopy is not comparable with cryo-TEM the sample is 331

broadly composed of vesicles coexisting with nanotubes whereas in some cases the nanotube 332

starts from the vesicle itself (arrow Ndeg1) in good agreement with cryo-TEM data Furthermore 333

SANS recorded at T= 90degC (Figure S 6a) shows a scattering profile with a pronounced -2 334

dependence of the scattering intensity against the wavevector q in the log-log scale This 335

indicates the massive presence of a flat interface and it is expected typical of bilayers in 336

vesicular morphologies56 However the typical signature of nanotubes cannot be observed in 337

the SANS profile and this can be explained by both an argument related to the relative lower 338

amount of nanotubes with respect to vesicles at 90degC but also to the steeper slope of flat 339

morphologies (-2) compared to tubes (-1) the signal of the former masking the tube signal at 340

low wavevector values The SANS profile is characterized by a diffraction peak at q= 0167 Aring-341

1 of which the assignment is refined by WAXS data 342

Temperature-driven WAXS experiments in Figure S 6bc (5 mgmL-1 and 25 mgmL-1 343

respectively) show a lamellar order possibly attributed to a multilamellar structure of the 344

vesicles (dv Figure S 6d) at high temperature and to the interplanar arrangement of the 345

crystalline flat structures (dL Figure S 6d) at T= 25degC The sharp (001) reflection indicates 346

typical interlamellar distance of 353 Aring at 25degC although this value varies with temperature 347

(400 Aring at 111degC 372 Aring at 88degC) thus confirming its assignment to a hydration-dependent 348

interlamellar order More intriguing is the second broader peak at q= 0403 Aring-1 (d= 156 Aring) 349

and its reflection both clearly visible at 88degC below Tm2 We attribute it to the inter-glucolipid 350

arrangement inside the tubular phase (dt Figure S 6d) coexisting with the vesicle phase 351

between Tm1 and Tm2 and to the planar lipid arrangement (di Figure S 6d) in the rigid flat 352

lamellar structure below Tm2 Its temperature- hence hydration- independent position also 353

confirms its structural origin Finally besides the water-structure peak at about 2 Aring-1 we also 354

observe a broad peak at 144 Aring-1 characteristics of the lateral packing (d Figure S 6d) of lipids 355

in bilayers While its position could be consistent with an Lα phase5758 above Tm2 its broadness 356

and positional invariance as well as the coexistence between vesicles and tubes do not allow a 357

precise assignment 358

15

Does G-C181-OH assume a bilayer or interdigitated structure the latter being the one 359

found for other similar bolaform glucolipids4243 The size of G-C181-OH can be estimated to 360

be at the most 32 Aring given by the length L of the C16 tail (L = 22 Aring from the Tanford formula 361

L= 154 + 1265 times n (n= 16)59) and the sugar (about 10 Aring)4243 However the cis conformation 362

of the C=C bond imposes a ldquokinkrdquo of 60deg in the aliphatic chain which reduces the effective 363

length of the oleic backbone to about 17 Aring for an effective total G-C181-OH length of about 364

27 Aring Cryo-TEM provides a tube thickness varying from 35 Aring to 60 Aring and a vesicle membrane 365

thickness of at least 60 Aring WAXS suggests an interlamellar thickness (membrane and water) 366

between 40 Aring and 37 Aring and an inter-lipid distance in the tube of about 16 Aring In the absence of 367

a clear form factor oscillation profile as found for similar glucolipids4243 one must 368

acknowledge that these data pledge neither in favor of fully interdigitated nor a double layer 369

structure and might suggest a tilted lipid arrangement with partial interdigitation60 370

371

13C ssNMR demonstrates that G-C181-OH can assume two coexisting head-head and 372

head-tail configurations 373

Vesicles are characterized by low positive mean and gaussian curvatures while 374

nanotubes are described by a combination of a high positive mean and zero gaussian curvatures 375

In this regard the coexistence of vesicles and nanotubes requires glucolipids to adapt 376

themselves to both environments at the same time For this reason the formation of nanotube-377

vesicle networks generally requires a specific driving force able to impose a high mean and 378

zero gaussian curvature to the membrane In many systems insertion of at least a second 379

constituent (lipid surfactant protein) in the membrane becomes a necessary internal driving 380

force for the nucleation and growth of so-called membrane inclusions from which 381

topologically-connected nanotubes can form10ndash123261 as explained by orientational ordering 382

arguments62 It has also been shown that external driving forces such as use of laser 383

ldquotweezersrdquo30 osmotic pressure6 electroformation5ndash82829 or controlled suction13ndash16 are 384

otherwise necessary as also supported by theoretical studies19 385

The bending energy of tubes is higher than that of a vesicle membrane8 due to the 386

dependence of the membrane free energy from the principal curvatures For this reason 387

isotropic (rotational symmetry along the main axis) but also anisotropic33 lipids generally 388

stabilize low-curvature shapes like flat membranes or vesicles Lipid nanotubes are also a 389

common phase formed by amphiphiles but generally for lipids with a certain complexity in 390

their molecular structures as exampled by diacetylenic derivatives of phospholipids among 391

others262763ndash65 Negative contributions to the bending energy favoring the simultaneous 392

16

presence of interconnected vesicles and tubes was shown to be driven by the presence of 393

anisotropic lipids in a membrane continuum Spontaneous deviations to the elastic theory in the 394

presence of surfactants lipids and proteins with anisotropic structures favor the formation of 395

protrusions with radius of the order of the fraction of a micron9ndash1219ndash2362 396

In the absence of both internal and external driving forces mentioned above G-C181-397

OH should behave as a standard lipid Its DSC profile (Figure 2)536667 under both hydrated and 398

dry conditions identifies a hydration-independent transition (Tm1) slightly above room 399

temperature this is followed by one main transition at higher temperature (Tm2) in the dry 400

sample50 For such a profile one expects a gel phase (Lβ) below Tm1 and a liquid crystalline 401

structure possibly in the morphology of vesicles (L4 phase) possibly above Tm1 and most likely 402

above Tm231 as experimentally found On the basis of the thermogram recorded on the dry 403

powder one may also expect a ldquoripplerdquo (Pβ) phase between Tm1 and Tm2 characterized by 404

periodic undulations of the bilayer (generally addressed to as regions with both liquid order and 405

disorder)5455 406

In the present work instead of a Lβ phase we observe flat crystals below Tm1 and a 407

vesicle-nanotube network instead of the ldquoripplerdquo phase The latter is particularly unexpected 408

because the membrane is composed of a single-component (G-C181-OH) and no anisotropic 409

external force (osmotic hydrodynamic electric) is used to pull and stabilize the tubes In 410

addition the nanotubes are observed at fairly high temperatures above Tm1 (between about 411

60degC and 90degC) while this morphology is generally favored at low temperature (proportionality 412

with 1kT figure 5 in Ref 62) commonly below a given transition temperature characteristics 413

for each specific amphiphile27 Furthermore nanotubes of G-C181-OH seem to be stable in 414

time thus differing from what is found in electroformed vesicles78 They are also observed 415

away from phase transition events (may them be at Tm1 or Tm2) as otherwise reported before 416

for a diacetylenic derivative of phosphatidylcholine26 and many other systems27 417

A stable nanotube-vesicle network for a single-component phospholipid in the absence 418

of external driving forces is not expected and has never been reported before It could be 419

explained by the coexistence of two micro-separated phases of different rigidity This 420

hypothesis is excluded by confocal microscopy on Laurdan-labelled membranes and magic 421

angle spinning (MAS) solid-state NMR (ssNMR) experiments 422

Laurdan is a well-known fluorophore used to put in evidence the coexistence of rigid 423

and fluid domains in both model and biological membranes47ndash49 Laurdan was used here to label 424

the G-C181-OH membranes between Tm1 and Tm2 and to verify the possible micro-segregation 425

and coexistence of rigid (eg Lβ λem= 440 nm) and fluid (eg Lα λem= 490 nm) domains in 426

17

tubes and vesicles respectively A set of quantitative experiments performed on a 5 mgmL-1 427

solution summarized in Figure 5 shows an excess of fluorescence at only λem= 490 nm for 428

temperature between 90degC and 70degC (Figure 5h) The GP value which quantifies the rigidfluid 429

character of the membrane is systematically negative (Figure 5i) indicating an overall fluid 430

membrane at all temperatures GP is quantified on more than 50 regions of interest (ROIs) at 431

each temperature including vesicles and tubes alone as well as connected nanotube-vesicles 432

(Figure 5d-g) The dispersion of GP at 80degC and 70degC between -08 and -02 could suggest the 433

spurious presence of membranes with various degrees of flexibility although fluidity is always 434

predominant The combination of WAXS (broad peak d= 144 Aring-1 invariant with 435

temperature Figure S 6bc) and confocal fluorescence microscopy using Laurdan confirms the 436

single-phase of the membrane composed of G-C181-OH both in vesicles and nanotubes This 437

is also consistent with the lack of phase transitions in DSC experiments between ~85degC and 438

~60degC especially in hydrated samples We conclude that the vesicle-nanotube transition is not 439

driven by the local segregation of gel phase (Lβ) microdomains nor that the nanotubes are 440

composed of such phase 441

ssNMR is the best-suited approach to investigate the origin of the local driving force for 442

the nucleation of anisotropic inclusions of which the average orientation depends on the local 443

membrane and inclusion curvatures and was shown to promote tubing62 ssNMR can provide 444

information on the local molecular mobility eg through the nuclear spin relaxation behavior 445

but also on intermolecular interactions and stereochemistry eg through variations in the 446

chemical shift68ndash70 447

Ideally ssNMR should be performed on diluted G-C181-OH samples above Tm1 448

However fast MAS necessary for high resolution centrifuges the sample As a first 449

approximation hydrated G-C181-OH samples for instance at 50 wt as classically done for 450

phospholipids71 could be studied Unfortunately permeability of the sample holder combined 451

to heating (close to the boiling point of water) and MAS provoke dehydration and even sample 452

ejection After experiencing both of the above this approach was abandoned As a second 453

approximation the study of the dehydrated G-C181-OH powder avoids centrifugation and 454

evaporation If studying dried lipids may not reflect their exact behaviour under diluted 455

conditions valuable structural information can still be obtained in dehydrated molecular 456

systems like proteins thus illustrating that this approach is not unreasonable Furthermore if 457

DSC (Figure 2a) shows that hydration suppresses the Tm2 it also shows that the value of Tm1 is 458

not affected by water at concentrations as low as 10 wt For this reason performing a 459

structural analysis of dehydrated G-C181-OH by ssNMR above Tm1 may help understanding 460

18

the local arrangement of G-C181-OH in two morphologies vesicles and nanotubes having 461

sensibly different curvatures 462

To explore the local mobility of the most relevant chemical groups of G-C181-OH 463

above Tm2 (T= 110degC) and between Tm2 and Tm1 (70 lt T degC lt 90) we combine a series of 1H 464

and 13C experiments Single-pulse (SP) MAS experiments (1H and 13C) when recorded under 465

quantitative conditions allow the detection of all chemical species while cross-polarization 466

(CP) MAS (13C only) experiments are sensitive only to constrained groups for which the 1H-467

13C dipolar coupling is not averaged to zero by the molecular motions 468

469

Figure 6 ndash a) 1H MAS (10 kHz) ssNMR spectra recorded from 110degC to 70degC b) 13C MAS (10 kHz) ssNMR 470

spectra recorded using high power decoupling single pulse (SP) and cross-polarization (CP) pulse sequences 471

Contact time for CP experiments and corresponding temperatures are given on each spectrum Full spectra 472

are given in Figure S 7 473

The most relevant 1H and 13C spectra are given in Figure 6 (full range in Figure 7 and 474

Figure S 7) As expected the 1H solid-state SPMAS spectra show broader peaks compared to 475

the solution-state NMR spectrum (Figure 6a) reflecting the obvious solid-state character of the 476

sample at any temperature The sugar group and the C=C bond resonate in the 25-6 ppm range 477

and 535 ppm respectively while the OH groups of the sugar moiety are found below 45 ppm 478

(Figure 6a full assignment in Table S 1) Very broad (~1 ppm) at 100degC the position of the 479

OH peak varies with the temperature value and becomes narrower (lt~05 ppm) and visible at 480

δH= 464 ppm at T= 70degC Line narrowing and chemical shift variations of the OH resonance 481

in ssNMR is always a direct sign of H-bonding interactions whereas the strength of the H-482

bonding hence its OHhellipH distance is correlated to a dispersion in chemical shifts72ndash75 All other 483

60 55 50 45 40 35 30 25

ppm

110

90

80

70

910

1rsquo

6rsquo16

1

OH

134 132 130 128 126

ppm110 108 106 104 102 100

ppm

SP (T= 110degC)

SP (T= 70degC)

CP 50 ms (T= 70degC)

CP 3 ms (T= 70degC)

910 1rsquo

1H 13C

1

811

910

CH2 CH2

1rsquo

6rsquo

a) b)

C102

13024

C9101

13003

C92

12869 C1rsquo210382

C1rsquo110443

C1rsquo13035

2rsquo3rsquo4rsquo

T degC

5rsquo

19

resonances broaden from 110degC to 70degC thus indicating slower andor more anisotropic 484

reorientational motions of all chemical groups The specific dynamic behaviour of the sugar 485

headgroup is also reflected by its 1H transverse relaxation time T2 whereas the shorter the T2 486

the higher the rigidity Experiments performed between 110degC and 70degC shown and 487

commented in Figure S 8 in the Supporting Information globally show a single short (lt 1 ms 488

below 100degC) T2 relaxation component for H1rsquo H5rsquo and H6rsquo and two both short and long 489

(between 4 ms and 12 ms) T2 for the acyl chain (CH2 H910 H811) even below 100degC 490

At 110degC 13C SPMAS experiments (Figure 6b full range in Figure 7 and Figure S 7 491

assignments in Table 1) show a single peak at δC= 10335 ppm for C1rsquo and at δC= 13003 ppm 492

for C910 as expected for a symmetrical conformation with two chemically-equivalent 493

carbons76 At 70degC the C1rsquo peak disappears while C910 now shows two peaks around 128 494

ppm and 130 ppm (Table 1 full temperature series in Figure S 7b) attributed to C92 (δC= 12869 495

ppm) and C10 in a non-symmetric configuration respectively for which these two carbons are 496

chemically inequivalent76 More precisely the resonance at 130 ppm is actually the 497

superposition of two close resonances C9101 (δC= 13003 ppm) and C102 (δC= 13024 ppm) 498

where subscripts 1 and 2 stand for two different conformations of the cis-olefin group as 499

reported for various phases of oleic acid in the vicinity of 0degC76 500

501

Table 1 ndash 13C Chemical shifts of G-C181-OH in solution and solid-state Solid-state experiments are 502

recorded at MAS= 10 kHz All spectra are reported in Figure 7 and Figure S 7 The contact time for cross 503

polarization is 3 ms In all ssNMR experiments temperature is first raised at 110degC and then lowered to 504

either 90degC or 70degC Attribution of the 13C resonances in Table 1 is performed on the basis of the solution 505

spectrum (Figure S 1b) Additional peaks observed in both single pulse and cross polarization ssNMR 506

experiments are attributed when possible to a configuration effect of the given atom eg C9101 and C92 507

indicate two configurations of C9 All new peaks observed in solid-state and not detected in solution are 508

highlighted in red for convenience 509

510

511

δ ppm

Peak Liquid (DMSO-d6)

Solid Single Pulse

T= 110degC

Solid Single Pulse

T= 90degC

Solid Single Pulse

T= 70degC

Solid Cross Polarization

T= 70degC

C102

- - 13024 13024

C910C9101 13009 13003 13003 13003 13003

C1C9C10C1rsquo

C6rsquo

20

C92 - - - 12869 12869

C1rsquoC11 10329 10335 10335 - 10443

C12 - - - - 10382

C3 7724 7709 7702 - 7829

- 7634 7629 - 7621

7525

C2 7388 7398 7390 7370 7370

C4 7052 7077 7054 7041 7098

C16 6900 7016 - - 7029

- - - - 6793

C1C5C6 6153 6241 6236 6239 6262

C1C5C6 6117 - - - 5956

C2 3300 3290 3290 3283 3279

CH2(trans) - - - 3198 3196

CH2(gauche)C15 2941 2981 2987 3004 -

C811 2706 2747 2747 2832 2817

CH2 2597 2625 2628 2630 2817

512

13C CPMAS spectra at 70degC (Figure 6b full range in Figure 7 and Figure S 7d) at 513

increasing contact times confirm the presence of two cis-olefin environments one for which 514

C910 are equivalent labelled here C9101 and one for which they are inequivalent labelled 515

C92 C102 Both environments are rigid enough to be observed at short contact time (tc= 50 μs) 516

A similar double environment is also observed for the sugar C1rsquo CPMAS at 70degC shows two 517

peaks (C1rsquo1 at δC= 10443 ppm C1rsquo2 at δC= 10382 ppm Figure 6b Figure S 7d) These are 518

significantly shifted downfield from the C1rsquo peak observed at 110degC and are detected at contact 519

times as short as 50 μs thus indicating their rigid character One also finds that CPMAS is 520

totally inefficient at 100degC for short contact times (50 μs) and poorly efficient even at 3 ms 521

typical of fast-tumbling molecules (Figure S 7c) also both gauche (δC= 3004 ppm) and all-522

trans (δC= 3198 ppm) conformations of the CH2 carbons coexist at 70degC (Figure S 7b)77 523

According to ssNMR one main mobile molecular environment exists at 110degC as 524

shown by the well-resolved 1H spectrum single C=C (C9101) and anomeric (C1rsquo) resonances 525

gauche conformation of the aliphatic CH2 and single T2 of the aliphatic CH2 chain and sugar 526

protons The sugar group is certainly mobile (inefficient 1H13C CPMAS) but still more 527

constrainted than the acyl chain (shorter T2 (1H)) As the temperature is lowered the overall 528

rigidity increases (broadening of 1H NMR peaks faster 1H transverse relaxation and efficient 529

1H13C magnetization transfer during CPMAS) presumably driven by H-bonding 530

interactions (downfield shift and line narrowing of the OH 1H signal very short OH T2 loss of 531

the C1rsquo signal in SPMAS) At the same time a number of spectroscopic data (two concomitant 532

21

resonances for anomeric C1rsquo1 and C1rsquo2 cis-olefin C9101 and C92 C102 and CH2 gauche and 533

all-trans environments) and relaxation behaviour (fast and slow T2 relaxation components) 534

strongly suggests the coexistence of two molecular environments Further proofs of the 535

coexistence of two distinct molecular domains are given by a number of chemical shift 536

variations in the 13C CPMAS spectrum at 70degC compared to the one at 110degC 90degC and 537

solution This is shown in Figure 7 and in the list of 13C chemical shifts in Table 1 538

539

540

Figure 7 - Comparison between solution- and solid-state 13C NMR spectra of G-C181-OH Solid-state 541

experiments are recorded directly on the sample powder at a MAS of 10 kHz (4 mm zirconia rotor) SP 542

Single Pulse CP Cross Polarization In all the ssNMR experiments temperature is first raised at 110degC 543

and then lowered to either 90degC or 70degC Segmented lines are meant to highlight the known peak position 544

in solution thus stressing the strong chemical shift variations in the solid-state 545

546

NMR experiments above Tm2 suggest that vesicles could contain G-C181-OH with a 547

single family of mobile conformations characterized by chemically-equivalent C=C (red 548

Figure 8b) and a single C1rsquo conformation (green Figure 8b) Experiments between Tm1 and Tm2 549

otherwise suggest that two coexisting conformations characterized by chemically-equivalent 550

(red Figure 8b) and inequivalent (magenta Figure 8b) C=C could be associated to nanotubes 551

Two distinct C1rsquo sites (grey Figure 8b) also indicate two configurations of the sugar headgroup 552

A comment on this point will be given in the following paragraphs 553

The local (Aring-scale) variation of molecular mobility below Tm2 drives the coexistence of 554

two distinct molecular environments of the same G-C181-OH molecule This occurs within a 555

single-phase domain (as deduced by WAXS) of equivalent membrane rigidity (as derived from 556

Laurdan probe in confocal microscopy) Although never reported before the difference in the 557

local mobility of the same amphiphile (here G-C181-OH) could satisfy the theoretically-558

140 120 100 80 60 40 20

CP(3 ms)-70degC

SP-70degC

SP-90degC

SP-110degC

ppm

LiquidC910

C1rsquo

C2rsquo

C4rsquo

C3rsquo

C1

C6rsquoC5rsquo

C1

6

C8

11

C2 C

H2

C1

5

CH

2

DM

SO

13C

22

required conditions that induce local lipid segregation and formation of membrane inclusions 559

(Figure 8c) In case of their non-zero average orientation the local membrane curvature 560

increases62 and nanotubes can spontaneously form (Figure 8bc) More intriguing this 561

mechanism occurs at least 10degC above Tm1 although it should not be favored above the Tm27 562

From both experimental and theoretical backgrounds the conditions of topologically-563

connected vesicles-nanotubes systems are met only when either internal (two components) or 564

external driving forces are applied19 In the present single-lipid system of which the phase 565

behavior is summarized in Figure 8a the most plausible driving force could be explained by 566

the NMR results collected on the dry sample G-C181-OH concomitantly occupies two 567

molecular environments characterized by 1) two distinct conformations of the sugar 568

displaying equivalent mobility and 2) two conformations of the acyl chain unsaturation 569

570

23

571

Figure 8 ndash a) Thermotropic phase behavior of G-C181-OH at 5 mgmL-1 Note that Tm2 is observed on the 572

dry powder only b) Possible relationship between the molecular environment as derived from NMR and 573

membrane curvatures All the molecular drawings refer to the same G-C181-OH molecule (bulky ellipsoid 574

equiv sugar headgroup) while each color corresponds to a given configuration Grey of C1rsquo1 and C1rsquo2 indicates 575

the uncertainty about the assignment of the specific NMR peak The curvature C is defined as 1R R being 576

the curvature radius R1 R2 radius of curvature CG Gauss curvature CM Mean curvature c) Free 577

interpretation of the anisotropic high-curvature (CMgt 0 CG= 0) inclusions in the membrane (low CM and 578

CG) continuum driven by inter-sugar H-bonding and reduced intramembrane diffusion for a G-C181-OH 579

glucolipid 580

581

Flat crystalline structure

Nanotube-vesicle network

Vesicles

48degC

120deg

106 105 104 103 102 101

ppm

H9 H10H1rsquo

106 105 104 103 102 101

ppm106 105 104 103 102 101

ppm106 105 104 103 102 101

ppm

C=CC=C

C1rsquo1C1rsquo2

C1rsquo

Tm2(96degC)Tm1

a)

b)

c)

Same molecule different local orientations and sugar conformations

High curvature membrane

Low curvature inclusion

Flip-FlopStrong H-bonding

CG

CM ( + )

CG 0

CM 0 CG 0

CM 0

G-C181-OH

24

Two related questions are unanswered at the moment What actually drives nanotube 582

formation and is the organization of G-C181-OH within the vesicles and tubes the same Lipid 583

nanotubes have been studied for at least three decades over which broad experimental and 584

theoretical insight has been generated Despite this fact the actual understanding is still matter 585

of debate27 Micelle-to-nanotube and vesicle-to-nanotube phase transitions are generally 586

reported for single amphiphiles below a given Tm and intermediate phase composed of twisted 587

and flat fibers is concomitantly reported Initially thought to be a chirality-driven phenomenon 588

recent studies propose a more general mechanism involving crystallization and symmetry 589

breaking in the lipid packing driven by optimal packing27 If twisted ribbons were observed 590

before for a number of microbial glycolipid bolaamphiphiles44 they are not found for G-C181-591

OH This fact as well as the large body of cryo-TEM images and the observation of nanotubes 592

above Tm1 rather pledge in favour of a direct topological change from positive (vesicles) to 593

zero (nanotubes) Gaussian curvatures In this case vesicle-to-nanotube transitions could be 594

driven by external forces56307813ndash162829 595

Considering the absence of external forces in the present system there are only two 596

possibilities left spontaneous budding of the membrane or nanotube pulling after inter-vesicle 597

collisions due to Brownian motion Both are possible as inclusions composed of a different 598

arrangement of the same molecule in the membrane continuum can theoretically explain 599

spontaneous nanotube formation in an isotropic medium672223629ndash1218ndash21 However the second 600

one is probably more likely to prevail because it could explain the large number of nanotubes 601

connecting two distant vesicles Interestingly similar events occur for any phospholipid-based 602

vesicular solution but nanotubes are extremely rare unless at least two lipids of different 603

molecular structure are mixed together The spectroscopic 1H and 13C and relaxation (1H) 604

ssNMR data even if collected on a dry powder demonstrate that the coexistence of two G-605

C181-OH configurations involving both the sugar headgroup and fatty alcohol chain are 606

possible for this molecule As a matter of fact this is not so surprising as more than one energy 607

minima each corresponding to a given conformation are known for glucose its non-reducing 608

derivative methylglucose and carbohydrates in general78ndash80 609

The presence of two configurations could explain the coexistence between two 610

structures of different curvature but the data presented so far do not explain the origin of these 611

two configurations In other words what drives the change in the local structure of the sugar 612

headgroup and fatty alcohol within the membrane 613

The association of known self-assembled structures to spectroscopic 13C ssNMR data is 614

known for years in particular for gluconamides lipids having a linear sugar headgroup81ndash84 It 615

25

was found that gauche effects in some cases up to the γ position in the sugar chain induce 616

upfield chemical shifts while anti and trans conformations result in downfield shifts Each 617

structure (monolayer crystal bilayer crystal or micellar fiber) was then characterized by a given 618

organization of the gluconamide and a characteristic 13C ssNMR fingerprint which could be 619

explained by gauche γ-gauche anti and trans conformations of the headgroup Similar 620

structural-spectroscopic correlations have been collected over the years for much simpler cyclic 621

sugars85 whereas pyranoses can undergo important conformational changes (chair half-chair 622

boat skew) when the appropriate energy barriers are crossed78ndash8086 A quantitative 623

conformational study based on the 13C chemical shift variations shown in Table 1 would be 624

certainly helpful but very delicate in the context of the present state of the art because a 625

rigorous approach requires at least one known crystal structure8182 which is currently missing 626

for this compound A qualitative approach will then be adopted 627

G-C181-OH is a bolaform amphiphile and flip-flop could be a possible mechanism 628

involved in the nanotube stabilization whereas head-tail interactions have long been considered 629

as important in nanotube-forming bolaform amphiphiles6063 From mere considerations based 630

on T2 (1H) relaxation arguments one could exclude head-tail configurations if the R-CH2OH 631

endgroup (H1) was involved in a direct interaction with the sugar group its corresponding T2 632

would probably be comparable to the sugarrsquos one (OH and H1rsquo) but this does not seem to be 633

the case However this hypothesis would be based on the comparison of the relaxation from 634

CH and CH2 protons additionally involved in different kind of motions Besides the relaxation 635

of H1 is influenced by the strong overlap between H1 and H5rsquo For these reasons relaxation 636

arguments are not solid enough and will not be used 637

We then address this question by exploring intermolecular interactions in a direct 638

manner using 2D 1H-1H homonuclear ssNMR correlation spectroscopy At 90degC in the fast-639

tumbling domain of molecular motion (narrow 1H NMR peaks Figure 6a) and corresponding 640

to the vesicle-rich region spin diffusion (NOESY) experiments are preferred over dipolar 641

coupling based experiments because such interaction is averaged out by molecular motion 642

However when the molecular motion is more constrained in the domain of nanotube stability 643

at 70degC (broad 1H spectra Figure 6a) Double-Quantum Single-Quantum (DQ-SQ) correlation 644

experiments are best suited to probe interaction between close molecular neighbors These 645

experiments are reported in Figure 9 and the related interpretation is only briefly reported 646

below while a more detailed discussion can be found on Page 12 in the Supporting Information 647

26

648

Figure 9 ndash a) 2D 1H-1H NOESY NMR experiment performed at MAS= 10 kHz (4 mm probe) on a Bruker 649

300 MHz Avance III HD spectrometer using 300 ms of mixing time Sample temperature is regulated to 650

90degC reached by cooling from 110degC b) 2D 1H-1H DQ-SQ (BAck-to-BAck BABA pulse sequence used for 651

the DQ excitation scheme) NMR experiment performed at MAS= 22 kHz (25 mm probe) on a Bruker 700 652

MHz Avance III spectrometer using a DQ excitation time equal to 2 rotor periods Sample temperature is 653

regulated to 70degC reached by cooling from 110degC The typical schemes illustrating how to interpret these 654

2D experiments are reported above each experimental map Zooms of the sugar region and configurational 655

interpretation are given below each 2D map More information are given in the text below 656

H16-H16

H1-H1

H9-H10

H1rsquo-H1

H1rsquoH910 H16

H1

H1rsquo-H1rsquo

Hxrsquo-Hxrsquo

2

(pp

m)

(ppm)

315

811910 1rsquo

2rsquo2

3rsquo 4rsquo6rsquo

5rsquo

1rsquo2910

asymp asymp asympHDO MeOD Acetone

1

315

811 9101rsquo

2rsquo2

3rsquo4rsquo6rsquo

5rsquo

1rsquo2 9 10

asympasympasympHDOMeODAcetone

H1rsquo-H1

1rsquo

1

(p

pm

)

(ppm)

H1-H2

H1rsquo-H1

H1rsquo

H910

H1H5rsquo

H2

H1rsquo-H5rsquo

315

811910 1rsquo

2rsquo2

3rsquo 4rsquo6rsquo

5rsquo

1rsquo2910

asymp asymp asympHDO MeOD Acetone

1

H2rsquoH3rsquoH4rsquo

H2rsquoH3rsquoH4rsquo

X= 2rsquo3rsquo4rsquo

NOESY - T= 90degC

1H

1H

1H

1H

DQ-SQ - T= 70degC

a) b)

2

(pp

m)

(p

pm

)

(ppm)

A B

AhellipB

AhellipA

BhellipB

NOESY

(ppm)

A B

AhellipB

DQ-SQ

Single-quantum dimension

Do

ub

le-q

ua

ntu

m d

imen

sio

n

AhellipA

BhellipB

BhellipA

BhellipA

315

811910 1rsquo

2rsquo2

3rsquo 4rsquo6rsquo

5rsquo

1rsquo2910

asymp asymp asympHDO MeOD Acetone

1

lt 5 Aring

lt 5 Aring

H1-H2

H811

H15

H16

H6rsquo

H1rsquo

H1

H5rsquo

H2

H1-H2

H1rsquo-H1

H1rsquo-H5rsquo

H1rsquo

H1 H5rsquo

H1rsquo

H1

H5rsquo

H1H2rsquo H3rsquo H4rsquo

H1-H1

H1rsquo-H1

Hxrsquo-HxrsquoX= 2rsquo3rsquo4rsquo

H1-H2

H1rsquo

Zoom sugar region 28 ndash 45 ppmNOESY DQ-SQ

27

Head-tail or head-head configurations could be probed by the existence of H1rsquo-H1 657

interactions whereas H1 (δH= 356 ppm) is associated to the glucolipid tail and H1rsquo (δH= 432 658

ppm) to the sugar headgroup These atoms are more than 15 Aring apart and a direct interaction 659

could not be observed in both NOESY and DQ-SQ experiments NOESY experiments (Figure 660

9a) show that H1rsquo-H1 correlations are not observed Meanwhile clear-cut direct H1rsquo-H1 661

interactions are indeed observed are detected at T= 70degC (DQ cross-peaks related to H1rsquo-H1 662

correlation in DQ-SQ experiment shown in Figure 9b) Interestingly the absence of auto peaks 663

at δH= 432 ppm and δH= 335 ppm in the SQ dimension excludes H1rsquo-H1rsquo and Hxrsquo-Hxrsquo (x= 2 664

3 4) interactions respectively strengthening the idea that sugar headgroups from two nearest-665

neighbor G-C181-OH molecules are not adjacent in the nanotube-vesicle region 666

Homonuclear 2D ssNMR supports the idea according to which G-C181-OH can 667

undergo molecular flip-flops and a head-tail arrangement in the dry product Extrapolating these 668

conclusions in the lower concentration range lack of H1rsquo-H1 proximity at T= 90degC suggests a 669

head-head configuration in the vesicle-rich region and head-tail configuration within the 670

nanotubes It is however still unclear whether flip-flops are driven by spontaneous temperature-671

induced conformational changes in the glucolipid packing stimulating inclusions within the 672

membrane itself672223629ndash1218ndash21 or by defects generated by inter-vesicle collisions In fact a 673

combination of both is not unlikely Nanotubes are not stable when temperature is 20 to 30 674

degrees above Tm1 indicating that inter-vesicle collisions alone cannot explain the unusual 675

phenomenon In an energy landscape approach87 the head-head configuration is certainly 676

favored at high temperature while both head-head and head-tail configurations could be 677

favored in the mid-temperature region corresponding to the range between Tm1 and Tm2 in the 678

dry powder Both statistical and defect-induced flip-flop events contribute to form high mean 679

zero Gaussian curvature morphologies (nanotubes) for which the head-tail conformation 680

limiting the proximity between bulky sugar headgroups is probably more adapted (Figure 8c) 681

The flip-flop interpretation now reasonably explains the strong differences in the sugar and fatty 682

alcohol conformations deduced by the multiple NMR signals attributed to chemically-683

equivalent groups The existence of various low-energy conformations each having specific 684

NMR signature for sugars in solution78ndash8086 and solid-state81ndash84 further supports our 685

interpretation 686

One last point should be clarified Why between Tm1 and Tm2 are two molecular 687

environments able to coexist instead of a single one of average rigidity as found in the case of 688

ldquoripplerdquo phases50ndash54 We believe that this question can be answered based on the NMR results 689

the relatively short T2 values of the H1rsquo and OH protons shows the constrained mobility of the 690

28

sugar headgroup despite a certain flexibility of the chain (long T2 component) This trend may 691

suggest restrictions in translational mobility probably due to the sugar-OH hydrogen bonding 692

(Figure 8c) meaning that lipids are not able to diffuse within the membrane thus keeping two 693

different configurations stable 694

695

Conclusions 696

Topologically-connected nanotube-vesicle lipidic systems are commonly observed for 697

self-assembled membranes composed of at least two amphiphiles and spontaneously driven by 698

anisotropic inclusions andor by external directional forces This work shows that strong 699

differences in membrane curvature can spontaneously occur in single-lipid systems without 700

external anisotropic forces Coexistence and interconnection between nanotubes and vesicles 701

are evidenced via cryo-TEM and fluorescence confocal microscopy in a single-phase region at 702

temperature above a transition temperature Tm1= 483degC determined by DSC on the hydrated 703

sample but below the second transition temperature Tm2 (957degC) determined by DSC on the 704

dry powder X-ray diffraction (WAXS) also shows the coexistence of two lattice periods one 705

attributed to an inter-membrane repeating distance and the other possibly related to the intra-706

nanotube glucolipid arrangement Labelling the systems with Laurdan a fluorophore of which 707

the emission strongly varies with the local membrane rigidity excludes the presence of rigid 708

micro-separated phases (eg Lβ) 709

1H and 13C ssNMR despite the fact that it can only be performed on the dry powder 710

confirm the presence of one main fluid G-C181-OH environment above Tm2 Below Tm2 the 711

splitting of some 13C peaks and the coexistence of short and long 1H relaxation components 712

demonstrate that G-C181-OH can simultaneously pack into structurally and dynamically 713

inequivalent environments Combining advanced 2D 1H-1H correlation experiments based on 714

spin diffusion or double-quanta excitation suggests that vesicles could contain a preferred head-715

head configuration of the bolaform lipid while nanotubes a preferred head-tail assembly 716

The presence of two environments composed of the same molecule characterized by 717

different molecular configurations and rigidity fulfills the theoretical conditions necessary to 718

spontaneously nucleate a membrane inclusion of non-zero average orientation described as 719

necessary to observe topologically-connected nanotube-vesicle structures in the absence of 720

external forces This hypothesis could also be enriched by the possibility of nanotubes to be 721

pulled away between two vesicles after a collision event 722

This work expands the actual theoretical and experimental framework on lipid networks 723

stabilized by a different radius of curvature It could also motivate the synthesis or isolation 724

29

of new lipids providing a higher level of 3D architectural control in soft networks possibly in 725

the range of room temperature 726

727

Acknowledgements 728

Dr Andrea Lassenberger and Dr Sylvain Preacutevost at Institut Laue Langevin (ILL Grenoble 729

France) are kindly acknowledged for their assistance on the SANS experiment Confocal 730

images were performed at the Institut de Biologie Paris Seine (IBPS) imaging facility Jean-731

Franccedilois Gilles (Imaging Core Facility Institut de Biologie Paris Seine (IBPS) CNRS 732

Sorbonne Universiteacute Paris France) is kindly acknowledged for his assistance on the confocal 733

microscopy imaging Dr Guillaume Laurent (LCMCP Sorbonne Universiteacute Paris France) is 734

kindly acknowledged for helpful discussions The French region Ile-de-France SESAME 735

program is acknowledged for financial support (700 MHz NMR 736

spectrometer) Dr Olivier Diat (ICSM Marcoule France) Prof eronikaIglič and Prof V Aleš737

onsul discussi Slovenia) are kindly acknowledged for helpfUniversity of LjubljanaIglič (-Kralj 738

739

Financial support 740

WAXS experiments have been supported by Soleil Light Source Saint Aubin France proposal 741

Ndeg 20201747 SANS experiments have been supported by ILL proposal Ndeg9-13-778 742

743

References 744

1 D Kabaso N Bobrovska W Goacuteźdź E Gongadze V Kralj-Iglič R Zorec and A 745

Iglič Bioelectrochemistry 2012 87 204ndash210 746

2 C R Safinya U Raviv D J Needleman A Zidovska M C Choi M A Ojeda-747

Lopez K K Ewert Y Li H P Miller J Quispe B Carragher C S Potter M W 748

Kim S C Feinstein and L Wilson Adv Mater 2011 23 2260ndash2270 749

3 P Veranič M Lokar G J Schuumltz J Weghuber S Wieser H Haumlgerstrand V Kralj-750

Iglič and A Iglič Biophys J 2008 95 4416ndash4425 751

4 A Rustom R Saffrich I Markovic P Walther and H-H Gerdes Science (80- ) 752

2004 303 1007ndash1010 753

5 M I Angelova S Soleacuteau P Meacuteleacuteard F Faucon and P Bothorel Prog Colloid 754

Polym Sci 1992 89 127ndash131 755

6 L Mathivet S Cribier and P F Devaux Biophys J 1996 70 1112ndash1121 756

7 V Kralj-Iglič G Gomišček J Majhenc V Arrigler and S Svetina Colloids Surfaces 757

A Physicochem Eng Asp 2001 181 315ndash318 758

30

8 V Kralj-Iglič A Iglič G Gomišček F Sevšek V Arrigler and H Haumlgerstrand J 759

Phys A Math Gen 2002 35 1533ndash1549 760

9 M Drab D Stopar V Kralj-Iglič and A Iglič Cells 2019 8 626 761

10 A Zidovska K K Ewert J Quispe B Carragher C S Potter and C R Safinya 762

Langmuir 2009 25 2979ndash2985 763

11 A Zidovska K K Ewert J Quispe B Carragher C S Potter and C R Safinya Soft 764

Matter 2011 7 8363ndash8369 765

12 A Zidovska K K Ewert J Quispe B Carragher C S Potter and C R Safinya 766

Biochim Biophys Acta - Biomembr 2009 1788 1869ndash1876 767

13 M Karlsson K Sott M Davidson A S Cans P Linderholm D Chiu and O Orwar 768

Proc Natl Acad Sci U S A 2002 99 11573ndash11578 769

14 A Karlsson R Karlsson M Karlsson A-S Cans A Stromberg F Ryttsen and O 770

Orwar Nature 2001 409 150ndash152 771

15 M Karlsson K Sott A S Cans A Karlsson R Karlsson and O Orwar Langmuir 772

2001 17 6754ndash6758 773

16 E Evans H Bowman A Leung D Needham and D Tirrell Science (80- ) 1996 774

273 933ndash935 775

17 U Seifert Configurations of fluid membranes and vesicles 1997 vol 46 776

18 J Kaumls and E Sackmann Biophys J 1991 60 825ndash844 777

19 N Bobrovska W Goacuteźdź V Kralj-Iglič and A Iglič PLoS One 2013 8 e73941 778

20 A Iglič B Babnik K Bohinc M Fošnarič H Haumlgerstrand and V Kralj-Iglič J 779

Biomech 2007 40 579ndash585 780

21 K Bohinc T Slivnik A Iglič M Brumen and V Kralj-Iglič J Phys Chem C 2007 781

111 9709ndash9718 782

22 A Iglič H Haumlgerstrand P Veranič A Plemenitaš and V Kralj-Iglič J Theor Biol 783

2006 240 368ndash373 784

23 V Kralj-Iglič and P Veranič Adv Planar Lipid Bilayers Liposomes 2006 5 129ndash785

149 786

24 A Iglic and V Kralj-Iglic Adv Planar Lipid Bilayers Liposomes 2006 4 253ndash279 787

25 S Penič L Mesarec M Fošnarič L Mroacutewczyńska H Haumlgerstrand V Kralj-Iglič and 788

A Iglič Front Phys 2020 8 342 789

26 P Yager and P E Schoen Mol Cryst Liq Cryst 1984 106 371ndash381 790

27 T G Barclay K Constantopoulos and J Matisons Chem Rev 2014 114 10217ndash791

10291 792

31

28 B Božič G Gomišček V Kralj-Iglič S Svetina and B Žekš Eur Biophys J 2002 793

31 487ndash496 794

29 A Iglič H Haumlgerstrand M Bobrowska-Haumlgerstrand V Arrigler and V Kralj-Iglič 795

Phys Lett A 2003 310 493ndash497 796

30 R Bar-Ziv and E Moses Phys Rev Lett 1994 73 1392ndash1395 797

31 S Chiruvolu H E Warriner E Naranjo S H Idziak J O Raumldler R J Plano J A 798

Zasadzinski and C R Safinya Science (80- ) 1994 266 1222ndash1225 799

32 V Kralj-Iglič A Iglič H Haumlgerstrand and P Peterlin Phys Rev E 2000 61 4230ndash800

4234 801

33 L Mesarec M Drab S Penič V Kralj-Iglič and A Iglič Int J Mol Sci 2021 22 802

2348 803

34 L Van Renterghem S L K W Roelants N Baccile K Uyttersprot M C Taelman 804

B Everaert S Mincke S Ledegen S Debrouwer K Scholtens C V Stevens W 805

Soetaert L Van Renterghem S L K W Roelants N Baccile K Uytersprot M C 806

Taelman B Everaert S Mincke S Ledegen S Debrouwer K Scholtens C V 807

Stevens and W Soetaert Biotechnol Bioeng 2018 115 1195ndash1206 808

35 J Clark and F Deswarte Introduction to Chemicals from Biomass Second Edition 809

John Wiley amp Sons 2015 810

36 K V Sajna R Houmlfer R K Sukumaran L D Gottumukkala and A Pandey in 811

Industrial Biorefineries and White Biotechnology eds A Pandey R Houmlfer M 812

Taherzadeh K M Nampoothiri and C Larroche Amsterdam Oxford Waltham 813

Elsevier 2015 pp 499ndash521 814

37 R Marchant and I M Banat Trends Biotechnol 2012 30 558ndash565 815

38 K Hill and C LeHen-Ferrenbach in Sugar-Based Surfactants Fundamentals and 816

Applications ed C C Ruiz Boca Raton CRC Press 2009 pp 1ndash20 817

39 L Rodrigues I M Banat J Teixeira and R Oliveira J Antimicrob Chemother 818

2006 57 609ndash618 819

40 J D Desai and I M Banat Microbiol Mol Biol Rev 1997 61 47ndash64 820

41 S L K W Roelants K Ciesielska S L De Maeseneire H Moens B Everaert S 821

Verweire Q Denon B Vanlerberghe I N A Van Bogaert P Van der Meeren B 822

Devreese and W Soetaert Biotechnol Bioeng 2016 113 550ndash559 823

42 N Baccile M Selmane P Le Griel S Preacutevost J Perez C V Stevens E Delbeke S 824

Zibek M Guenther W Soetaert I N A Van Bogaert and S Roelants Langmuir 825

2016 32 6343ndash6359 826

32

43 N Baccile A-S Cuvier S Preacutevost C V Stevens E Delbeke J Berton W Soetaert 827

I N A Van Bogaert and S Roelants Langmuir 2016 32 10881ndash10894 828

44 N Baccile C Seyrig A Poirier S A Castro S L K W Roelants and S Abel Green 829

Chem 2021 23 3842ndash3944 830

45 A Salis L Cappai C Carucci D F Parsons and M Monduzzi J Phys Chem Lett 831

2020 11 6805ndash6811 832

46 J Schindelin I Arganda-Carreras E Frise V Kaynig M Longair T Pietzsch S 833

Preibisch C Rueden S Saalfeld B Schmid J-Y Tinevez D J White V 834

Hartenstein K Eliceiri P Tomancak and A Cardona Nat Methods 2012 9 676ndash835

682 836

47 T Parasassi and E Gratton J Fluoresc 1995 5 59ndash69 837

48 M M Dodes Traian F L Gonzaacutelez Flecha and V Levi J Lipid Res 2012 53 609ndash838

616 839

49 D M Owen C Rentero A Magenau A Abu-Siniyeh and K Gaus Nat Protoc 840

2012 7 24ndash35 841

50 W Stillwell in An Introduction to Biological Membranes Academic Press 2016 pp 842

135ndash180 843

51 W Stillwell in An Introduction to Biological Membranes Academic Press 2016 pp 844

181ndash220 845

52 B Tenchov Chem Phys Lipids 1991 57 165ndash177 846

53 R Koynova and M Caffrey Biochim Biophys Acta 1998 1376 91ndash145 847

54 F Schmid S Dolezel O Lenz and S Meinhardt J Phys Conf Ser 2014 487 848

012004 849

55 S Meinhardt R L C Vink and F Schmid Proc Natl Acad Sci U S A 2013 110 850

4476ndash4481 851

56 O Glatter and O Kratky Small Angle X-ray Scattering Academic Press London 852

1982 853

57 B A Cunningham A D Brown D H Wolfe W P Williams and A Brain Phys 854

Rev E 1998 58 3662ndash3672 855

58 M Nadler A Steiner T Dvir O Szekely P Szekely A Ginsburg R Asor R Resh 856

C Tamburu M Peres and U Raviv Soft Matter 2011 7 1512ndash1523 857

59 C Tanford The hydrophobic effect Formation of micelles and biological membranes 858

John Wiley amp Sons Inc New York 1973 859

60 M Masuda and T Shimizu Langmuir 2004 20 5969ndash5977 860

33

61 A Zidovska K K Ewert J Quispe B Carragher C S Potter and C R Safinya 861

Block Liposomes Vesicles of Charged Lipids with Distinctly Shaped Nanoscale 862

Sphere- Pear- Tube- or Rod-Segments Elsevier Inc 1st edn 2009 vol 465 863

62 A Iglic V Kralj-Iglic B Božič M Bobrowska-Haumlgerstrand B Isomaa and H 864

Haumlgerstrand Bioelectrochemistry 2000 52 203ndash211 865

63 T Shimizu M Masuda and H Minamikawa Chem Rev 2005 105 1401ndash1444 866

64 J V Selinger and J M Schnur Phys Rev Lett 1993 71 4091ndash4094 867

65 J M Schnur Science (80- ) 1993 262 1669ndash1676 868

66 B Tenchov R Koynova and G Rapp Biophys J 2001 80 1873ndash1890 869

67 R Koynova and B Tenchov Ser Biomech 2017 31 3ndash19 870

68 S J Angyal and R Le Fur Carbohydr Res 1984 126 15ndash26 871

69 J K Whitesell and M A Minton J Am Chem Soc 1987 109 225ndash228 872

70 T Takayama S Ando and I Ando J Mol Struct 1990 220 245ndash250 873

71 K Gawrisch N V Eldho and I V Polozov Chem Phys Lipids 2002 116 135ndash151 874

72 F Tielens C Gervais J F Lambert F Mauri and D Costa Chem Mater 2008 20 875

3336ndash3344 876

73 G A Jeffrey and Y Yeon Acta Crystallogr Sect B 1986 42 410ndash413 877

74 B Berglund and R W Vaughan J Chem Phys 1980 73 2037ndash2043 878

75 C Gervais C Coelho T Azaiumls J Maquet G Laurent F Pourpoint C Bonhomme 879

P Florian B Alonso G Guerrero P H Mutin and F Mauri J Magn Reson 2007 880

187 131ndash140 881

76 C Akita T Kawaguchi F Kaneko H Yamamoto and M Suzuki J Phys Chem B 882

2004 108 4862ndash4868 883

77 R N Purusottam L Seacutenicourt J J Lacapegravere and P Tekely Biochim Biophys Acta - 884

Biomembr 2015 1848 3134ndash3139 885

78 X Biarneacutes A Ardegravevol A Planas C Rovira A Laio and M Parrinello J Am Chem 886

Soc 2007 129 10686ndash10693 887

79 L M J Kroon‐Batenburg and J Kroon Biopolymers 1990 29 1243ndash1248 888

80 K N Kirschner and R J Woods Proc Natl Acad Sci U S A 2001 98 10541ndash889

10545 890

81 S Svenson B Kirste and J H Fuhrhop J Am Chem Soc 1994 116 11969ndash11975 891

82 I Sack S Macholl J H Fuhrhop and G Buntkowsky Phys Chem Chem Phys 892

2000 2 1781ndash1788 893

83 S Svenson J Koening and J H Fuhrhop J Phys Chem 1994 98 1022ndash1028 894

34

84 S Svenson A Schaefer and J H Fuhrhop J Chem Soc Perkin Trans 2 1994 2 895

1023ndash1028 896

85 K Bock and C Pedersen Adv Carbohydr Chem Biochem 1983 41 27ndash66 897

86 T B Grindley Structure and Conformation of Carbohydrates Springer Berlin 898

Heidelberg 2008 899

87 F Tantakitti J Boekhoven X Wang R V Kazantsev T Yu J Li E Zhuang R 900

Zandi J H Ortony C J Newcomb L C Palmer G S Shekhawat M O De La 901

Cruz G C Schatz and S I Stupp Nat Mater 2016 15 469ndash476 902

903

1

Supporting Information 904

905

Topological connection between vesicles and nanotubes in single-906

component lipid membranes driven by head-tail interactions 907

908

Niki Baccilea Ceacutedric Lorthioira Abdoul Aziz Baa Patrick Le Griela Cristina Coelhob Javier 909

Perezc Wim Soetaertd Sophie L K W Roelantsd 910

911

a Sorbonne Universiteacute Centre National de la Recherche Scientifique Laboratoire de Chimie de 912

la Matiegravere Condenseacutee de Paris LCMCP F-75005 Paris France 913

b Institut des Mateacuteriaux de Paris Centre Sorbonne Universiteacute Paris France 914

c Synchrotron Soleil LrsquoOrme des Merisiers Saint-Aubin BP48 91192 Gif-sur-Yvette Cedex 915

France 916

d InBio Department of Biotechnology Ghent University Ghent Belgium 917

918

919

Corresponding author 920

Dr Niki Baccile 921

E-mail address nikibaccilesorbonne-universitefr 922

Phone +33 1 44 27 56 77 923

924

925

926

2

a) 927

Parameter Method

Dry Matter (DM ) 996 Infrared balance 105degC

Glucose () 012 HPLC-Metacarb

Glycerol () nd HPLC-Metacarb

Free Fatty Acid content () lt001 Internal method BBEPP

GC with FID Detector

Oil () lt001 Internal method BBEPP

GC with FID Detector

Moulds (CFUg) lt10 3M-Nordval ndeg16 B

Yeasts (CFUg) lt10 3M-Nordval ndeg16 B

Anaerobic count (CFUg) lt10 SP-VG M005 B

Protein () tbd BCA protein assay

928

Figure S 1 - a) HPLC-ELSD chromatogram and composition table 929

930

931

932

933

3

b) 934

935

Group Ndeg δ ppm

CH2 1 354

CH2 2 153

CH2 3-712-14 132

CH2 811 203

CH 910 534

CH2 15 162

CH2 16 388

CH 1 424

CH 2 316

CH 3 334

CH 4 329

CH 5 354

CH2 6 366

Figure S 1 (next) ndash b) 1H NMR spectrum (MeOD-d4) recorded on the purified G-C181-OH compounds 936

and related assignment 937

938

2

811910 1rsquo

5rsquo1

3rsquo4rsquo

1rsquo1910

asymp asymp asympHDO MeOD Acetone

6rsquo2rsquo

1615

3-712-14

4

939

940

Figure S 2 ndash Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 130degC 941

5

942

943

Figure S 3 - Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 90degC 944

945

90degC

1

22

6

946

947

948

949

Figure S 4 ndash Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 70degC 950

951

21

11

1

2

5

3

4

45

5

6

7

7

7

952

953

954

Figure S 5 ndash Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 25degC 955

956

957

1

1 1

8

958

Figure S 6 ndash a) SANS experiment recorded at 90degC for a 5 mgmL G-C181-OH aqueous solution b-c) 959

WAXS experiments recorded on aqueous solutions of G-C181-OH at b) 5 mgmL and c) 25 mgmL d) 960

Structural interpretation of the repeating distances pointed out in c) 961

962

963

964

001 01 1

q Aring-1

25degC

68degC

88degC

111degC

01 1q Aring

-1

02 04 06 08 1q Aring

-1

001 01

10-1

100

101

Inte

nsity

cm

-1

q Aring-1

-2

T= 90degC

SANS WAXS WAXS

a) b) c)

dv

dL d= 144 Aring-1

di

ddH2O

dt

dvdL

dtdi d

d)

9

965

966

Figure S 7 ndash ssNMR MAS (10 kHz) experiments recorded on a G-C181-OH dry powder a) 1H spectra 967

recorded from 110degC to 70degC b) 13C single pulse (SP) spectra recorded with high power 1H decoupling from 968

110degC to 70degC c) 13C cross polarization (CP) spectra recorded at 110degC and contact time tc of 50 micros and 969

3000 micros d) 13C CP spectra recorded at 70degC and for 20 lt tc micros lt 3000 970

971

1

811

910

CH2 CH2

1rsquo

6rsquo

8 6 4 2 0 ppm

110

90

80

70

910

1rsquo

6rsquo16

1

2rsquo3rsquo4rsquo

811

OH

45 40 35 30 25 20 15

ppm

200 180 160 140 120 100 80 60 40 20 0

ppm140 120 100 80

ppm

CH2-gauche

3004

110

90

80

70

65

60

50

25

C91013003

C9 - 12869

CH2-all-trans

3198

C1rsquo10335

a) b)

1H

T degCT degC

13C Single Pulse MAS

A

B

A

B

70degC70degC

5rsquo

215

200 180 160 140 120 100 80 60 40 20 0

ppm

13C CPMAST= 110degC

200 180 160 140 120 100 80 60 40 20 0

ppm

132 131 130 129 128 127 126

ppm

106 105 104 103 102

ppm

13C CPMAST= 70degC

50

tc micros

3000

20

tc micros

50 500 1000 3000

tc micros

C102

13024 C910113003

C92

12869

C1rsquo213082

C1rsquo110443

A B

106 105 104 103 102

ppm

3000

1000

500

50

20

AB

c)

d)

10

972

Figure S 8 ndash a) Short and d) long T2 components displayed by the 1H transverse relaxation signal c) 1H 973

transverse relaxation signals recorded on a G-C181-OH dry powder from 110degC to 70degC The area of each 974

given resonance is plotted (log-lin) as a function of the echo time (CPMG pulse sequence) for each 975

temperature and fitted using the equation given above T2s and T2l indicate the short and long T2 relaxation 976

times respectively obtained by describing the 1H transverse relaxation signal using a biexponential decay 977

For H1rsquo H5rsquo and H6rsquo the 1H relaxation function could be described by a monoexponential decay 978

- Protons from the sugar group (H1rsquo H5rsquo H6rsquo) show a single T2 relaxation component (Figure 979

6c and Figure S 8) while protons of the acyl chain (CH2 H910 H811) show both short (Figure 980

6c) and long (Figure 6d) T2 relaxation component 981

- The short T2 relaxation time systematically decreases from 110degC to 70degC for all species by a 982

factor ranging from two to seven depending on the specific chemical group Very long T2 are 983

observed at 110degC for all acyl species while nearly unchanged values are measured below 984

90degC without any clear trend 985

- The sugar protons (H1rsquo H5rsquo H6rsquo) have the shortest set of T2 values (lt 1 ms) below 100degC 986

The long component of T2 generally ranges between 4 ms and 12 ms 987

0 1 2 3 4 5 6 7 8

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1 2 3 4 5 6 7 8

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

H1

H8H11

H9H10

CH2

H1rsquo

H6rsquo

CH2

H910H811 H1CH2

H1rsquo H5rsquo6rsquo

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

OH

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45 T= 110degC

T= 90degC

T= 80degC

T= 70degC

Log

In

ten

sity

Time ms

Fit

)

CH2 H910 H811 H1 H56 H1 OH

00

05

10

15

20

25

30

35

40

T2 m

s

T= 110degC

T= 90degC

T= 80degC

T= 70degC

CH2 H910 H811 H1

0

2

4

6

8

10

12

T2 m

s

T= 110degC

T= 90degC

T= 80degC

T= 70degC

Short T2(1H) Long T2(1H)

a) b)

c)

11

Table S 1 ndash 1H chemical shifts of G-C181-OH in solution and in the solid-state Solid-state spectra are 988

recorded at a MAS frequency of 10 kHz in DMSO-d6 broad peak 989

990

δH ppm

Group Ndeg Liquid

(MeOD-d4) Solid

(T= 110degC) Solid

(T= 70degC)

CH2 H1 354 357

CH2 H2 153 156

CH2 H3-712-14 132 135

CH2 H811 203 203

CH H910 534 534

CH2 H15 162 164

CH2 H16 388 383

CH H1 424 432

CH H2 316 335

CH H3 334 335

CH H4 329 335

CH H5 354 351 357

CH2 H6 366 383

OH OH1 432

45 463

OH OH2 492

OH OH3 490

OH OH4 487

OH OH6 445

991

992

H1H9H10H1rsquo

H6rsquo

OH2rsquo

OH1

12

Detailed discussion about the NOESY and DQ-SQ 2D 1H-1H ssNMR experiments presented in 993

Figure 9 994

995

2D 1H-1H ssNMR experiments are a unique tool to explore through-space proximities 996

between 1H nuclei ssNMR offers a large number of possibilities each adapted to a given 997

system The advantage of 2D maps with respect to a dynamics approach is their direct 998

readability in clear two nuclei which exchange their magnetization through-space (whichever 999

mechanism may be responsible for this exchange) will display a specific cross-peak However 1000

the way cross-peaks are presented and interpreted strictly depends on the nature of the pulse 1001

program which in turn is chosen according to the physical nature of the sample 1002

1D temperature-dependent 1H NMR experiments performed on G-C181-OH (Figure 1003

4a in the main text) indicate sharp peaks at 90degC and broad peaks at 70degC Narrow peaks 1004

indicate fast andor large-amplitude molecular tumbling averaging the strong 1H-1H dipolar 1005

interactions This excludes the use of pulse sequences like DQ-SQ which is not efficient under 1006

these conditions For this reason we employ spin-diffusion-based pulse sequences like 1007

NOESY which is more adapted for samples with high molecular mobility When mixing times 1008

are chosen to be short (lt 1 s) one generally explores through-space proximities within about 5 1009

10 Aring On the contrary when molecular motions are restricted dipolar coupling becomes 1010

significant 1H NMR peaks are broader and pulse sequences like DQ-SQ can now be used 1011

with a spatial resolution of less than 5 Aring 1012

Whenever possible DQ-SQ experiments are preferred over spin diffusion-based pulse 1013

sequences because they are more selective in particular for likewise interactions which are 1014

not discriminated in NOESY-type experiments This is illustrated by the typical interaction 1015

schemes for NOESY (top Figure 9a) and DQ-SQ (top Figure 9b) For two given A and B 1016

protons 2D NOESY maps will display two on-diagonal and two square-shaped symmetrical 1017

off-diagonal cross-peaks Off-diagonal cross-peaks illustrate the mutual A-B and B-A 1018

interaction but the on-diagonal peaks are systematic and do not prove any A-A or B-B 1019

interaction Similarly to NOESY 2D DQ-SQ maps display off-diagonal peaks (symmetric 1020

across the diagonal top Figure 9b) upon A-B and B-A interaction but in contrast they will 1021

display on-diagonal peaks only in the presence of self-interaction In the tutorial example of 1022

Figure 9b top the on-diagonal A-A peak indicates mutual A-A interaction while lack of on-1023

diagonal B-B peak evidences the absence of B-B proximities Combining NOESY and DQ-SQ 1024

experiments is capital to study the local arrangement of G-C181-OH and its evolution with 1025

temperature 1026

13

At T= 90degC the H1 and H5rsquo contributions are very close but possible to discriminate 1027

them (δ(H1)= 354 ppm δ(H5rsquo)= 351 ppm) according to 1H NMR chemical shifts in Table S 1028

1 2D 1H-1H NOESY in Figure 9a also discriminates between the two components attributed 1029

on the basis of the interactions with their mutual neighbors (green H1-H2 and H5rsquo-H1rsquo squares 1030

in Figure 9a) On this basis 2D NOESY does not show a direct H1rsquo-H1 correlation (blue red-1031

crossed) but rather an extensive network of intra-sugar interactions These features exclude 1032

suggest intra- and inter-sugar interactions and they exclude head-tail arrangements as a major 1033

molecular configuration at 90degC 1034

At 70degC the 1H NMR peaks are broader indicating a reduction of the T2 (1H) relaxation 1035

times andor residual homonuclear dipolar interactions excluding the use of NOESY In this 1036

case 1H-1H DQ-SQ correlation experiments are best suited Figure 9b presents the 1H-1H DQ-1037

SQ experiment on G-C181-OH at 70degC Peaks are broader than in NOESY and attribution may 1038

be cumbersome However the experiment is performed at 700 MHz and MAS= 22 kHz which 1039

improves spectral resolution with respect to the same experiment performed at 300 MHz and 1040

MAS= 10 kHz The auto-peak at δH= 356 ppm in the SQ dimension is attributed to H1-H1 spin 1041

pairs mainly resulting from intra CH2 interaction while the DQ cross-peaks at δH= 156-356 1042

ppm and δH= 432-356 ppm respectively origin from H2-H1 and H1rsquo-H1 interactions (Figure 1043

9b) DQ-SQ map does not show any auto-peak corresponding to H1rsquo-H1rsquo or Hxrsquo-Hxrsquo (x= 2rsquo 1044

3rsquo 4rsquo) interactions (red-crossed blue Figure 9b) thus excluding any proximity (lt 5 Aring) between 1045

close glucose neighbors On the contrary DQ-SQ maps (recorded both at 700 MHz Figure 9b 1046

or at 300 MHz not shown) systematically show DQ cross-peaks which can only be attributed 1047

to H1rsquo-H1 interactions (δH= 432-356) The presence of these H1rsquo-H1 interactions can only be 1048

explained by a head-tail arrangement of the G-C181-OH molecules as illustrated below the 1049

DQ-SQ map in Figure 9b 1050

6

H2O were probed C= 5 mgmL-1 and C= 25 mgmL-1 The 2D data were integrated azimuthally 152

at the beamline using the software Foxtrot and in order to obtain the I(q) vs q spectrum after 153

masking the beam stop shadow Silver behenate (d(100) = 5838 Aring) was used as a standard to 154

calibrate the q-scale Sample solutions were inserted in borosilicate capillaries of 15 mm in 155

diameter Capillaries were flame-sealed A capillary oven with controlled temperature (plusmn 05degC) 156

provided at the beamline was used to control the sample temperature between 25degC and 111degC 157

Data were normalized by the transmission and calibrated to the SAXS signal of H2O at large q-158

values (I= 00163 cm-1) in order to obtain an absolute intensity scale The water signal was 159

measured by subtracting the signal of the empty capillary from the signal of a water-filled 160

capillary The signal of (water + capillary) was used as background for the samples and it was 161

subtracted after integration of the 2D data 162

163

Cryogenic transmission electron microscopy (cryo-TEM) Cryo-TEM experiments were carried 164

out on an FEI Tecnai 120 twin microscope operating at 120 kV and equipped with a Gatan 165

Orius CCD numeric camera The sample holder was a Gatan Cryoholder (Gatan 626DH 166

Gatan) Digital Micrograph software was used for image acquisition Cryofixation was 167

performed using a homemade cryofixation device The solutions were deposited on a glow-168

discharged holey carbon coated TEM copper grid (Quantifoil R22 Germany) Excess solution 169

was removed and the grid was immediately plunged into liquid ethane at -180degC before 170

transferring them into liquid nitrogen All grids were kept at liquid nitrogen temperature 171

throughout all experimentation Cryo-TEM images have been treated and analyzed using Fiji 172

(is just ImageJ) software available free of charge at the developerrsquos website46 173

The samplersquos concentration was C= 5 mgmL-1 For technical reasons the temperature 174

during cryofixation could not be controlled For this reason the temperature associated with 175

each cryo-TEM image in the text was related to the equilibration temperature just before 176

cryofixation the device being as close as possible to the dry bath used for temperature control 177

Since temperature control during cryofixation was not possible in our device and practically 178

impossible at high temperatures one should take the T values as indicative of the given value 179

Although the sample transfer between the dry bath and the cryofixation device was performed 180

as fast as possible it is more than likely that the actual temperature of the sample at the moment 181

of cryofixation was lower than the one indicated here although impossible to know precisely 182

183

7

Fluorescence microscopy Images were recorded using a 40x objective on a Leica SP5 upright 184

confocal microscope with 405 nm laser excitation The sample was marked with 6-dodecanoyl-185

NN-dimethyl-2-naphthylamine (Laurdan Mw= 35354 gmol-1) fluorophore as follows A 186

stock solution of 3 mg of Laurdan in 100 microL of acetone was initially prepared (C= 30 mgmL-187

1) A volume of 028 microL of the stock Laurdan solution was diluted in 1 mL of a 5 mgmL-1 G-188

C181-OH water solution temporary set at 110degC in a sealed Eppendorf The latter was 189

important so to enhance inclusion of Laurdan in the membrane constituted by the glucolipid 190

while in its vesicle phase The final Laurdan concentration was approximately 24 micromol with a 191

molar Laurdan-to-G-C181-OH molar ratio of approximately 1-to-500 Fluorophore-to-lipid 192

ratios above 1200 were generally considered as non-interfering with the lipid phase behavior 193

as also verified in this work 194

Laurdan can be excited between 365 nm and 410 nm and its emission wavelength can 195

vary according to the rigidity of the membrane intercalation within a rigid or flexible 196

membrane generates emissions at 440 nm or 490 nm respectively47 In the present experiment 197

we have flame-sealed a solution of G-C181-OH at 5 mgmL-1 containing Laurdan (1-to-500 in 198

lipid-to-Laurdan ratio) in flat optical capillaries of 01 mm thickness Capillaries were set on a 199

Linkam support for temperature control eventually placed under the microscope The sample 200

was excited at 405 nm and detection was made in spectral mode with a 5 nm window sliding 201

every 5 nm from 412 nm to 602 nm in 16 bits which allows quantify the signal and compare 202

the different conditions Quantitative analysis was performed according to the literature4849 203

several ROI (range of interest) have been drawn on several images using the ROI manager of 204

Fiji (is just ImageJ) software The integrated intensity of each ROI at 440 nm and 490 nm was 205

eventually recorded and used to calculate the value of the generalized polarization GP as 206

follows of 207

119866119875 =(120582440 minus 120582490)

120582440 + 120582490 208

A negative GP indicates a flexible membrane while a positive GP indicates a rigid membrane 209

210

Solid-state Nuclear Magnetic Resonance (ssNMR) 1H and 13C ssNMR experiments have been 211

performed using an Avance III HD Bruker 705 T (ν1H= 300 MHz) spectrometer and a 4 mm 212

magic angle spinning (MAS) probe except for the BAck-to-BAck (BABA) experiment which 213

was performed on an Avance III Bruker 1644 T (ν1H= 700 MHz) spectrometer equipped with 214

25 mm MAS probe Temperature was controlled through a Smart cooler BCUIIa BCU-xtreme 215

unit using 10 min of equilibration time after each temperature variation Temperature was 216

8

always varied from high (110degC) to low (70degC) before analysis Due to combined use of fast 217

MAS rates necessary to achieve high resolution spectra and heating close to the boiling point 218

of water neither diluted nor concentrated samples could be used for ssNMR experiments All 219

tests employing up to 50 wt G-C181-OH in water resulted in sample centrifugation drying 220

but also expulsion from the rotor causing sudden stop of rotor spinning For these reasons all 221

attempts to study hydrated G-C181-OH failed and only the dry powder could be analyzed The 222

dry sample powder was analyzed as such in zirconia rotors (4 mm or 25 mm) spinning at MAS= 223

10 kHz (4 mm) or 22 kHz (25 mm) and Chemical shifts were calibrated with respect to 224

adamantane (13C 13C(CH2)= 3848 ppm = 191 ppm) and glycine alpha (13C(CO)= 17603 225

ppm) Details on the acquisition parameters are given in the Supporting Information 226

227

Results and discussion 228

Topologically-connected nanotubes and vesicles 229

G-C181-OH is a biobased glucolipid water-insoluble at room temperature and of 230

which the physicochemical properties are poorly known The DSC thermogram performed at 231

10degCmin of G-C181-OH as a function of the water content (Figure 2a) shows two endothermic 232

peaks on the dry powder (996 wt glucolipid) They are attributed to two melting phenomena 233

occurring at the following temperatures Tm 483degC (Tm1) and 957degC (Tm2) the enthalpy of the 234

former being lower than that of the latter Interestingly DSC also shows that Tm1 is hydration 235

independent and reversible (the associated exothermic peak is systematically observed) while 236

the transition at Tm2 is simply suppressed (here tested up to 130degC) on hydrated powders and 237

never observed on the cooling profile even at slow cooling rates (1degCmin) The thermogram 238

recorded on the dry powder is typical of single-component lipids having an intermediate50 239

metastable phase (often referred to the ripple phase Pβ´ in reference to the periodic undulation 240

of the bilayer)50ndash54 between the classical lamellar gel (Lβ) and liquid crystalline (Lα) phases50ndash241

54 and characterized by regions of both liquid order and disorder5455 The effect of water does 242

not influence the low-temperature crystallization at Tm1 but it contributes to disrupt the ordered 243

array of hydrocarbon chains in the metastable phase 244

On the basis of the DSC thermogram the aqueous phase behavior of G-C181-OH under 245

diluted conditions (5 mgmL-1) was studied at four different temperatures (red marks on Figure 246

2a) below and above Tm1 but also between Tm1 and Tm2 corresponding to the thermogram of 247

the dry powder 248

9

249

Figure 2 ndash a) DSC thermogram of G-C181-OH under dry (04 wt water estimated by thermogravimetric 250

analysis) and hydrated conditions recorded at a heating rate of 10degCmin Red marks indicate the 251

temperatures at which cryo-TEM experiments are performed on the corresponding 5 mgmL aqueous 252

solutions b) Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 130degC 253

90degC 70degC and 25degC T indicates the temperature of the sample solution before cryofixation Tne T T 254

being the temperature of the sample upon plunging in liquid ethane Please refer to the materials and 255

methods section for more information 256

T= 130degC

T= 90degC

T= 70degC

T= 25degC

a)

b)

20 40 60 80 100 120

99 wt

50 wt

25 wt

T degC

En

do

10 wt

Tm1= 483 degC

Tm2= 957 degC

10

257

The series of cryo-TEM images in Figure 2b show an overview of the most relevant 258

structures observed on G-C181-OH solution from T= 130degC to T= 25degC whereas T is 259

indicative of the equilibrium temperature before cryofixation (refer to materials and methods 260

section for more information on the sense and interpretation of T) At Tgt 100degC the sample 261

is essentially composed of polydisperse single wall vesicles massively shown by the 262

complementary pictures presented in Figure S 2 In between Tm1 and Tm2 at T= 90degC and T= 263

70degC (Figure 2b Figure 3 Figure 4 complemented by Figure S 3 and Figure S 4) vesicles are 264

connected by nanotubes At Tlt Tm1 flat crystals are otherwise observed (Figure 2b and Figure 265

S 5) 266

267

268

Figure 3 ndash Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 90degC 269

270

A deeper insight at T= 90degC is given in Figure 3 (more images are provided in Figure 271

S 3) Arrows 1 through 6 mark very clearly the typical continuity in the glucolipid membrane 272

between the vesicle and nanotubes Arrow Ndeg 7 shows that a single nanotube connects the outer 273

membrane of two vesicles several hundred nanometers apart The connection is shown by 274

2

34

5

6

6

1

5

7

A

B

D

C

0 10 20 30 40 50 60

BIn

tensity

au

Profile nm

A

0 10 20 30 40 50 60

Inte

nsity

au

Profile nm

C

D

T= 90degC

11

arrows Ndeg5 and Ndeg6 The corresponding profiles A through D confirm the tubular nature of 275

the elongated objects Profiles B and D are both characterized by typical symmetrical thick 276

walls (low intensity depths highlighted by shaded rectangles) surrounding a hollow region (high 277

intensity peak between the shaded rectangles) The wall thickness varies between 35 and 40 278

nm for profile B and between 41 and 61 nm for profile D while the total nanotube diameter 279

varies between 14 nm at B and 20 nm at D At the same time the thickness of the vesicle wall 280

is about 6 nm at A and 135 nm at C Considerations about the bilayer or interdigitated structure 281

of the membrane will be presented below 282

The connection between nanotubes and vesicles seems to be multiple For instance 283

arrow Ndeg1 (Figure S 3) and Ndeg4 (Figure 3) show a single-wall vesicle of which the membrane 284

is extruded into a nanotube Similarly a closer look at arrows Ndeg1 2 and 3 in Figure 3 show 285

that the vesicle is single-wall and fully closed while the nanotube nucleates at the vesicle outer 286

surface with no apparent topological continuity In many regions of the sample at T= 90degC 287

one can find exvaginations or buds232531 in the nanotube wall This is shown for instance by 288

arrows Ndeg7 and particularly put in evidence in Figure 4 by arrows Ndeg1 through Ndeg4 where the 289

amplitude of the oscillation varies between 20-25 nm (arrow Ndeg1) and 6-8 nm (arrows Ndeg3-4) 290

291

Figure 4 - Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heated at T= 90degC 292

293

Still between Tm2 and Tm1 but at lower temperature (T= 70degC) the nanotubes seem to 294

vary from single threads (arrow Ndeg1 Figure S 4) to nanotubes of diameter below 15 nm both 295

connecting nanotubes of diameter ranging between 50 nm and 100 nm (arrows Ndeg2 Figure S 296

4) The larger tubes are most likely formed by the inflation and flattening of the nanotubes 297

probably after assimilation of surrounding vesicles and fusion of the exvagination discussed 298

above The transition between vesicles tubes and flat membranes is certainly more accentuated 299

Arrow Ndeg3 in Figure S 4 shows distinct vesicles while arrow Ndeg4 identifies a flat structure 300

evidently formed from fused vesicles In particular arrow Ndeg5 points at the fusion of two 301

2

1 132

4

12

vesicles while arrow Ndeg6 points at two vesicles fusing into the flat membrane Additional areas 302

where vesicle fusion is prominent are indicated by the oscillations at the outer skirt of broad 303

lamellae and pointed by arrow Ndeg7 304

Finally below Tm1 at T= 25degC for instance mainly flat crystals with no visible 305

exvaginations or tubes are visible (Figure S 5) by cryo-TEM and the material turns into an 306

insoluble precipitate Massive presence of flat crystalline structures is also clear from the bulk 307

iridescent behavior of a G-C181-OH solution under manual shear 308

In summary despite the uncertainty related to the actual state of equilibrium of the 309

glucolipids at the moment of cryofixation cryo-TEM depicts a vesicle phase at Tgt 100degC a 310

flat lamellar phase below Tm1 and an interconnected vesicle-nanotube network between Tm1 and 311

about 100degC The combination of statistically-relevant fluorescence microscopy and 312

scatteringdiffraction experiments provides an additional proof of the coexistence of 313

vesiclenanotube systems between the two transition temperatures Please note that the 314

experiments reported below have been performed under controlled conditions of temperature 315

which is then indicated by the use of T instead of T 316

13

317

Figure 5 ndash a-c) Confocal fluorescence microscopy images recorded at 90degC and 70degC for a 5 mgmL G-318

C181-OH aqueous solution d-g) Typical fluorescence microscopy images recorded at T= 80degC using 319

Laurdan (refer to materials and methods section for more details on this fluorophore) Excitation 320

wavelength is 405 nm for all the images while emission wavelength is 490 nm in (df) and 440 nm in (eg) 321

h) Full emission spectra obtained from the integration of various fluorescence images recorded from 412 322

nm to 600 nm (step 5 nm) at 90degC 80degC and 70degC i) Box plot of GP defined as 119918119927 =(120640120786120786120782minus120640120786120791120782)

120640120786120786120782+120640120786120791120782 with 120640120786120786120782 323

and 120640120786120791120782 being the emission wavelengths at 440 nm and 490 nm at 90degC 80degC and 70degC Each data point 324

l= 490 nm l= 440 nm

l= 490 nm l= 440 nm

10 μm400 450 500 550 600

00

02

04

06

08

10

No

rma

lize

d I

nte

nsity

l nm

90degC

80degC

80degC

70degC

70degC

400 450 500 550 60000

02

04

06

08

10

No

rma

lize

d I

nte

nsity

l nm

90degC

80degC

80degC

70degC

70degC

d) e)

f) g)

h)

i)

T= 80degC

Laurdan

T= 80degC

T= 80degC T= 80degC

T= 90degC T= 70degC T= 70degC

a) b) c)

1

1

1

14

corresponds to a set of regions of interest (ROIs) (54 at 90degC 65 at 80degC and 75 at 70degC) selected in the 325

corresponding fluorescence images ROIs were representatively selected so to include the entire image but 326

also vesicles alone tubes alone and mixtures of vesicles and tubes 327

328

Figure 5ab shows two typical confocal fluorescence microscopy images recorded on a 329

G-C181-OH solution (5 mgmL) in a flame-sealed capillary heat at T= 90degC and T= 70degC 330

Although the resolution of optical microscopy is not comparable with cryo-TEM the sample is 331

broadly composed of vesicles coexisting with nanotubes whereas in some cases the nanotube 332

starts from the vesicle itself (arrow Ndeg1) in good agreement with cryo-TEM data Furthermore 333

SANS recorded at T= 90degC (Figure S 6a) shows a scattering profile with a pronounced -2 334

dependence of the scattering intensity against the wavevector q in the log-log scale This 335

indicates the massive presence of a flat interface and it is expected typical of bilayers in 336

vesicular morphologies56 However the typical signature of nanotubes cannot be observed in 337

the SANS profile and this can be explained by both an argument related to the relative lower 338

amount of nanotubes with respect to vesicles at 90degC but also to the steeper slope of flat 339

morphologies (-2) compared to tubes (-1) the signal of the former masking the tube signal at 340

low wavevector values The SANS profile is characterized by a diffraction peak at q= 0167 Aring-341

1 of which the assignment is refined by WAXS data 342

Temperature-driven WAXS experiments in Figure S 6bc (5 mgmL-1 and 25 mgmL-1 343

respectively) show a lamellar order possibly attributed to a multilamellar structure of the 344

vesicles (dv Figure S 6d) at high temperature and to the interplanar arrangement of the 345

crystalline flat structures (dL Figure S 6d) at T= 25degC The sharp (001) reflection indicates 346

typical interlamellar distance of 353 Aring at 25degC although this value varies with temperature 347

(400 Aring at 111degC 372 Aring at 88degC) thus confirming its assignment to a hydration-dependent 348

interlamellar order More intriguing is the second broader peak at q= 0403 Aring-1 (d= 156 Aring) 349

and its reflection both clearly visible at 88degC below Tm2 We attribute it to the inter-glucolipid 350

arrangement inside the tubular phase (dt Figure S 6d) coexisting with the vesicle phase 351

between Tm1 and Tm2 and to the planar lipid arrangement (di Figure S 6d) in the rigid flat 352

lamellar structure below Tm2 Its temperature- hence hydration- independent position also 353

confirms its structural origin Finally besides the water-structure peak at about 2 Aring-1 we also 354

observe a broad peak at 144 Aring-1 characteristics of the lateral packing (d Figure S 6d) of lipids 355

in bilayers While its position could be consistent with an Lα phase5758 above Tm2 its broadness 356

and positional invariance as well as the coexistence between vesicles and tubes do not allow a 357

precise assignment 358

15

Does G-C181-OH assume a bilayer or interdigitated structure the latter being the one 359

found for other similar bolaform glucolipids4243 The size of G-C181-OH can be estimated to 360

be at the most 32 Aring given by the length L of the C16 tail (L = 22 Aring from the Tanford formula 361

L= 154 + 1265 times n (n= 16)59) and the sugar (about 10 Aring)4243 However the cis conformation 362

of the C=C bond imposes a ldquokinkrdquo of 60deg in the aliphatic chain which reduces the effective 363

length of the oleic backbone to about 17 Aring for an effective total G-C181-OH length of about 364

27 Aring Cryo-TEM provides a tube thickness varying from 35 Aring to 60 Aring and a vesicle membrane 365

thickness of at least 60 Aring WAXS suggests an interlamellar thickness (membrane and water) 366

between 40 Aring and 37 Aring and an inter-lipid distance in the tube of about 16 Aring In the absence of 367

a clear form factor oscillation profile as found for similar glucolipids4243 one must 368

acknowledge that these data pledge neither in favor of fully interdigitated nor a double layer 369

structure and might suggest a tilted lipid arrangement with partial interdigitation60 370

371

13C ssNMR demonstrates that G-C181-OH can assume two coexisting head-head and 372

head-tail configurations 373

Vesicles are characterized by low positive mean and gaussian curvatures while 374

nanotubes are described by a combination of a high positive mean and zero gaussian curvatures 375

In this regard the coexistence of vesicles and nanotubes requires glucolipids to adapt 376

themselves to both environments at the same time For this reason the formation of nanotube-377

vesicle networks generally requires a specific driving force able to impose a high mean and 378

zero gaussian curvature to the membrane In many systems insertion of at least a second 379

constituent (lipid surfactant protein) in the membrane becomes a necessary internal driving 380

force for the nucleation and growth of so-called membrane inclusions from which 381

topologically-connected nanotubes can form10ndash123261 as explained by orientational ordering 382

arguments62 It has also been shown that external driving forces such as use of laser 383

ldquotweezersrdquo30 osmotic pressure6 electroformation5ndash82829 or controlled suction13ndash16 are 384

otherwise necessary as also supported by theoretical studies19 385

The bending energy of tubes is higher than that of a vesicle membrane8 due to the 386

dependence of the membrane free energy from the principal curvatures For this reason 387

isotropic (rotational symmetry along the main axis) but also anisotropic33 lipids generally 388

stabilize low-curvature shapes like flat membranes or vesicles Lipid nanotubes are also a 389

common phase formed by amphiphiles but generally for lipids with a certain complexity in 390

their molecular structures as exampled by diacetylenic derivatives of phospholipids among 391

others262763ndash65 Negative contributions to the bending energy favoring the simultaneous 392

16

presence of interconnected vesicles and tubes was shown to be driven by the presence of 393

anisotropic lipids in a membrane continuum Spontaneous deviations to the elastic theory in the 394

presence of surfactants lipids and proteins with anisotropic structures favor the formation of 395

protrusions with radius of the order of the fraction of a micron9ndash1219ndash2362 396

In the absence of both internal and external driving forces mentioned above G-C181-397

OH should behave as a standard lipid Its DSC profile (Figure 2)536667 under both hydrated and 398

dry conditions identifies a hydration-independent transition (Tm1) slightly above room 399

temperature this is followed by one main transition at higher temperature (Tm2) in the dry 400

sample50 For such a profile one expects a gel phase (Lβ) below Tm1 and a liquid crystalline 401

structure possibly in the morphology of vesicles (L4 phase) possibly above Tm1 and most likely 402

above Tm231 as experimentally found On the basis of the thermogram recorded on the dry 403

powder one may also expect a ldquoripplerdquo (Pβ) phase between Tm1 and Tm2 characterized by 404

periodic undulations of the bilayer (generally addressed to as regions with both liquid order and 405

disorder)5455 406

In the present work instead of a Lβ phase we observe flat crystals below Tm1 and a 407

vesicle-nanotube network instead of the ldquoripplerdquo phase The latter is particularly unexpected 408

because the membrane is composed of a single-component (G-C181-OH) and no anisotropic 409

external force (osmotic hydrodynamic electric) is used to pull and stabilize the tubes In 410

addition the nanotubes are observed at fairly high temperatures above Tm1 (between about 411

60degC and 90degC) while this morphology is generally favored at low temperature (proportionality 412

with 1kT figure 5 in Ref 62) commonly below a given transition temperature characteristics 413

for each specific amphiphile27 Furthermore nanotubes of G-C181-OH seem to be stable in 414

time thus differing from what is found in electroformed vesicles78 They are also observed 415

away from phase transition events (may them be at Tm1 or Tm2) as otherwise reported before 416

for a diacetylenic derivative of phosphatidylcholine26 and many other systems27 417

A stable nanotube-vesicle network for a single-component phospholipid in the absence 418

of external driving forces is not expected and has never been reported before It could be 419

explained by the coexistence of two micro-separated phases of different rigidity This 420

hypothesis is excluded by confocal microscopy on Laurdan-labelled membranes and magic 421

angle spinning (MAS) solid-state NMR (ssNMR) experiments 422

Laurdan is a well-known fluorophore used to put in evidence the coexistence of rigid 423

and fluid domains in both model and biological membranes47ndash49 Laurdan was used here to label 424

the G-C181-OH membranes between Tm1 and Tm2 and to verify the possible micro-segregation 425

and coexistence of rigid (eg Lβ λem= 440 nm) and fluid (eg Lα λem= 490 nm) domains in 426

17

tubes and vesicles respectively A set of quantitative experiments performed on a 5 mgmL-1 427

solution summarized in Figure 5 shows an excess of fluorescence at only λem= 490 nm for 428

temperature between 90degC and 70degC (Figure 5h) The GP value which quantifies the rigidfluid 429

character of the membrane is systematically negative (Figure 5i) indicating an overall fluid 430

membrane at all temperatures GP is quantified on more than 50 regions of interest (ROIs) at 431

each temperature including vesicles and tubes alone as well as connected nanotube-vesicles 432

(Figure 5d-g) The dispersion of GP at 80degC and 70degC between -08 and -02 could suggest the 433

spurious presence of membranes with various degrees of flexibility although fluidity is always 434

predominant The combination of WAXS (broad peak d= 144 Aring-1 invariant with 435

temperature Figure S 6bc) and confocal fluorescence microscopy using Laurdan confirms the 436

single-phase of the membrane composed of G-C181-OH both in vesicles and nanotubes This 437

is also consistent with the lack of phase transitions in DSC experiments between ~85degC and 438

~60degC especially in hydrated samples We conclude that the vesicle-nanotube transition is not 439

driven by the local segregation of gel phase (Lβ) microdomains nor that the nanotubes are 440

composed of such phase 441

ssNMR is the best-suited approach to investigate the origin of the local driving force for 442

the nucleation of anisotropic inclusions of which the average orientation depends on the local 443

membrane and inclusion curvatures and was shown to promote tubing62 ssNMR can provide 444

information on the local molecular mobility eg through the nuclear spin relaxation behavior 445

but also on intermolecular interactions and stereochemistry eg through variations in the 446

chemical shift68ndash70 447

Ideally ssNMR should be performed on diluted G-C181-OH samples above Tm1 448

However fast MAS necessary for high resolution centrifuges the sample As a first 449

approximation hydrated G-C181-OH samples for instance at 50 wt as classically done for 450

phospholipids71 could be studied Unfortunately permeability of the sample holder combined 451

to heating (close to the boiling point of water) and MAS provoke dehydration and even sample 452

ejection After experiencing both of the above this approach was abandoned As a second 453

approximation the study of the dehydrated G-C181-OH powder avoids centrifugation and 454

evaporation If studying dried lipids may not reflect their exact behaviour under diluted 455

conditions valuable structural information can still be obtained in dehydrated molecular 456

systems like proteins thus illustrating that this approach is not unreasonable Furthermore if 457

DSC (Figure 2a) shows that hydration suppresses the Tm2 it also shows that the value of Tm1 is 458

not affected by water at concentrations as low as 10 wt For this reason performing a 459

structural analysis of dehydrated G-C181-OH by ssNMR above Tm1 may help understanding 460

18

the local arrangement of G-C181-OH in two morphologies vesicles and nanotubes having 461

sensibly different curvatures 462

To explore the local mobility of the most relevant chemical groups of G-C181-OH 463

above Tm2 (T= 110degC) and between Tm2 and Tm1 (70 lt T degC lt 90) we combine a series of 1H 464

and 13C experiments Single-pulse (SP) MAS experiments (1H and 13C) when recorded under 465

quantitative conditions allow the detection of all chemical species while cross-polarization 466

(CP) MAS (13C only) experiments are sensitive only to constrained groups for which the 1H-467

13C dipolar coupling is not averaged to zero by the molecular motions 468

469

Figure 6 ndash a) 1H MAS (10 kHz) ssNMR spectra recorded from 110degC to 70degC b) 13C MAS (10 kHz) ssNMR 470

spectra recorded using high power decoupling single pulse (SP) and cross-polarization (CP) pulse sequences 471

Contact time for CP experiments and corresponding temperatures are given on each spectrum Full spectra 472

are given in Figure S 7 473

The most relevant 1H and 13C spectra are given in Figure 6 (full range in Figure 7 and 474

Figure S 7) As expected the 1H solid-state SPMAS spectra show broader peaks compared to 475

the solution-state NMR spectrum (Figure 6a) reflecting the obvious solid-state character of the 476

sample at any temperature The sugar group and the C=C bond resonate in the 25-6 ppm range 477

and 535 ppm respectively while the OH groups of the sugar moiety are found below 45 ppm 478

(Figure 6a full assignment in Table S 1) Very broad (~1 ppm) at 100degC the position of the 479

OH peak varies with the temperature value and becomes narrower (lt~05 ppm) and visible at 480

δH= 464 ppm at T= 70degC Line narrowing and chemical shift variations of the OH resonance 481

in ssNMR is always a direct sign of H-bonding interactions whereas the strength of the H-482

bonding hence its OHhellipH distance is correlated to a dispersion in chemical shifts72ndash75 All other 483

60 55 50 45 40 35 30 25

ppm

110

90

80

70

910

1rsquo

6rsquo16

1

OH

134 132 130 128 126

ppm110 108 106 104 102 100

ppm

SP (T= 110degC)

SP (T= 70degC)

CP 50 ms (T= 70degC)

CP 3 ms (T= 70degC)

910 1rsquo

1H 13C

1

811

910

CH2 CH2

1rsquo

6rsquo

a) b)

C102

13024

C9101

13003

C92

12869 C1rsquo210382

C1rsquo110443

C1rsquo13035

2rsquo3rsquo4rsquo

T degC

5rsquo

19

resonances broaden from 110degC to 70degC thus indicating slower andor more anisotropic 484

reorientational motions of all chemical groups The specific dynamic behaviour of the sugar 485

headgroup is also reflected by its 1H transverse relaxation time T2 whereas the shorter the T2 486

the higher the rigidity Experiments performed between 110degC and 70degC shown and 487

commented in Figure S 8 in the Supporting Information globally show a single short (lt 1 ms 488

below 100degC) T2 relaxation component for H1rsquo H5rsquo and H6rsquo and two both short and long 489

(between 4 ms and 12 ms) T2 for the acyl chain (CH2 H910 H811) even below 100degC 490

At 110degC 13C SPMAS experiments (Figure 6b full range in Figure 7 and Figure S 7 491

assignments in Table 1) show a single peak at δC= 10335 ppm for C1rsquo and at δC= 13003 ppm 492

for C910 as expected for a symmetrical conformation with two chemically-equivalent 493

carbons76 At 70degC the C1rsquo peak disappears while C910 now shows two peaks around 128 494

ppm and 130 ppm (Table 1 full temperature series in Figure S 7b) attributed to C92 (δC= 12869 495

ppm) and C10 in a non-symmetric configuration respectively for which these two carbons are 496

chemically inequivalent76 More precisely the resonance at 130 ppm is actually the 497

superposition of two close resonances C9101 (δC= 13003 ppm) and C102 (δC= 13024 ppm) 498

where subscripts 1 and 2 stand for two different conformations of the cis-olefin group as 499

reported for various phases of oleic acid in the vicinity of 0degC76 500

501

Table 1 ndash 13C Chemical shifts of G-C181-OH in solution and solid-state Solid-state experiments are 502

recorded at MAS= 10 kHz All spectra are reported in Figure 7 and Figure S 7 The contact time for cross 503

polarization is 3 ms In all ssNMR experiments temperature is first raised at 110degC and then lowered to 504

either 90degC or 70degC Attribution of the 13C resonances in Table 1 is performed on the basis of the solution 505

spectrum (Figure S 1b) Additional peaks observed in both single pulse and cross polarization ssNMR 506

experiments are attributed when possible to a configuration effect of the given atom eg C9101 and C92 507

indicate two configurations of C9 All new peaks observed in solid-state and not detected in solution are 508

highlighted in red for convenience 509

510

511

δ ppm

Peak Liquid (DMSO-d6)

Solid Single Pulse

T= 110degC

Solid Single Pulse

T= 90degC

Solid Single Pulse

T= 70degC

Solid Cross Polarization

T= 70degC

C102

- - 13024 13024

C910C9101 13009 13003 13003 13003 13003

C1C9C10C1rsquo

C6rsquo

20

C92 - - - 12869 12869

C1rsquoC11 10329 10335 10335 - 10443

C12 - - - - 10382

C3 7724 7709 7702 - 7829

- 7634 7629 - 7621

7525

C2 7388 7398 7390 7370 7370

C4 7052 7077 7054 7041 7098

C16 6900 7016 - - 7029

- - - - 6793

C1C5C6 6153 6241 6236 6239 6262

C1C5C6 6117 - - - 5956

C2 3300 3290 3290 3283 3279

CH2(trans) - - - 3198 3196

CH2(gauche)C15 2941 2981 2987 3004 -

C811 2706 2747 2747 2832 2817

CH2 2597 2625 2628 2630 2817

512

13C CPMAS spectra at 70degC (Figure 6b full range in Figure 7 and Figure S 7d) at 513

increasing contact times confirm the presence of two cis-olefin environments one for which 514

C910 are equivalent labelled here C9101 and one for which they are inequivalent labelled 515

C92 C102 Both environments are rigid enough to be observed at short contact time (tc= 50 μs) 516

A similar double environment is also observed for the sugar C1rsquo CPMAS at 70degC shows two 517

peaks (C1rsquo1 at δC= 10443 ppm C1rsquo2 at δC= 10382 ppm Figure 6b Figure S 7d) These are 518

significantly shifted downfield from the C1rsquo peak observed at 110degC and are detected at contact 519

times as short as 50 μs thus indicating their rigid character One also finds that CPMAS is 520

totally inefficient at 100degC for short contact times (50 μs) and poorly efficient even at 3 ms 521

typical of fast-tumbling molecules (Figure S 7c) also both gauche (δC= 3004 ppm) and all-522

trans (δC= 3198 ppm) conformations of the CH2 carbons coexist at 70degC (Figure S 7b)77 523

According to ssNMR one main mobile molecular environment exists at 110degC as 524

shown by the well-resolved 1H spectrum single C=C (C9101) and anomeric (C1rsquo) resonances 525

gauche conformation of the aliphatic CH2 and single T2 of the aliphatic CH2 chain and sugar 526

protons The sugar group is certainly mobile (inefficient 1H13C CPMAS) but still more 527

constrainted than the acyl chain (shorter T2 (1H)) As the temperature is lowered the overall 528

rigidity increases (broadening of 1H NMR peaks faster 1H transverse relaxation and efficient 529

1H13C magnetization transfer during CPMAS) presumably driven by H-bonding 530

interactions (downfield shift and line narrowing of the OH 1H signal very short OH T2 loss of 531

the C1rsquo signal in SPMAS) At the same time a number of spectroscopic data (two concomitant 532

21

resonances for anomeric C1rsquo1 and C1rsquo2 cis-olefin C9101 and C92 C102 and CH2 gauche and 533

all-trans environments) and relaxation behaviour (fast and slow T2 relaxation components) 534

strongly suggests the coexistence of two molecular environments Further proofs of the 535

coexistence of two distinct molecular domains are given by a number of chemical shift 536

variations in the 13C CPMAS spectrum at 70degC compared to the one at 110degC 90degC and 537

solution This is shown in Figure 7 and in the list of 13C chemical shifts in Table 1 538

539

540

Figure 7 - Comparison between solution- and solid-state 13C NMR spectra of G-C181-OH Solid-state 541

experiments are recorded directly on the sample powder at a MAS of 10 kHz (4 mm zirconia rotor) SP 542

Single Pulse CP Cross Polarization In all the ssNMR experiments temperature is first raised at 110degC 543

and then lowered to either 90degC or 70degC Segmented lines are meant to highlight the known peak position 544

in solution thus stressing the strong chemical shift variations in the solid-state 545

546

NMR experiments above Tm2 suggest that vesicles could contain G-C181-OH with a 547

single family of mobile conformations characterized by chemically-equivalent C=C (red 548

Figure 8b) and a single C1rsquo conformation (green Figure 8b) Experiments between Tm1 and Tm2 549

otherwise suggest that two coexisting conformations characterized by chemically-equivalent 550

(red Figure 8b) and inequivalent (magenta Figure 8b) C=C could be associated to nanotubes 551

Two distinct C1rsquo sites (grey Figure 8b) also indicate two configurations of the sugar headgroup 552

A comment on this point will be given in the following paragraphs 553

The local (Aring-scale) variation of molecular mobility below Tm2 drives the coexistence of 554

two distinct molecular environments of the same G-C181-OH molecule This occurs within a 555

single-phase domain (as deduced by WAXS) of equivalent membrane rigidity (as derived from 556

Laurdan probe in confocal microscopy) Although never reported before the difference in the 557

local mobility of the same amphiphile (here G-C181-OH) could satisfy the theoretically-558

140 120 100 80 60 40 20

CP(3 ms)-70degC

SP-70degC

SP-90degC

SP-110degC

ppm

LiquidC910

C1rsquo

C2rsquo

C4rsquo

C3rsquo

C1

C6rsquoC5rsquo

C1

6

C8

11

C2 C

H2

C1

5

CH

2

DM

SO

13C

22

required conditions that induce local lipid segregation and formation of membrane inclusions 559

(Figure 8c) In case of their non-zero average orientation the local membrane curvature 560

increases62 and nanotubes can spontaneously form (Figure 8bc) More intriguing this 561

mechanism occurs at least 10degC above Tm1 although it should not be favored above the Tm27 562

From both experimental and theoretical backgrounds the conditions of topologically-563

connected vesicles-nanotubes systems are met only when either internal (two components) or 564

external driving forces are applied19 In the present single-lipid system of which the phase 565

behavior is summarized in Figure 8a the most plausible driving force could be explained by 566

the NMR results collected on the dry sample G-C181-OH concomitantly occupies two 567

molecular environments characterized by 1) two distinct conformations of the sugar 568

displaying equivalent mobility and 2) two conformations of the acyl chain unsaturation 569

570

23

571

Figure 8 ndash a) Thermotropic phase behavior of G-C181-OH at 5 mgmL-1 Note that Tm2 is observed on the 572

dry powder only b) Possible relationship between the molecular environment as derived from NMR and 573

membrane curvatures All the molecular drawings refer to the same G-C181-OH molecule (bulky ellipsoid 574

equiv sugar headgroup) while each color corresponds to a given configuration Grey of C1rsquo1 and C1rsquo2 indicates 575

the uncertainty about the assignment of the specific NMR peak The curvature C is defined as 1R R being 576

the curvature radius R1 R2 radius of curvature CG Gauss curvature CM Mean curvature c) Free 577

interpretation of the anisotropic high-curvature (CMgt 0 CG= 0) inclusions in the membrane (low CM and 578

CG) continuum driven by inter-sugar H-bonding and reduced intramembrane diffusion for a G-C181-OH 579

glucolipid 580

581

Flat crystalline structure

Nanotube-vesicle network

Vesicles

48degC

120deg

106 105 104 103 102 101

ppm

H9 H10H1rsquo

106 105 104 103 102 101

ppm106 105 104 103 102 101

ppm106 105 104 103 102 101

ppm

C=CC=C

C1rsquo1C1rsquo2

C1rsquo

Tm2(96degC)Tm1

a)

b)

c)

Same molecule different local orientations and sugar conformations

High curvature membrane

Low curvature inclusion

Flip-FlopStrong H-bonding

CG

CM ( + )

CG 0

CM 0 CG 0

CM 0

G-C181-OH

24

Two related questions are unanswered at the moment What actually drives nanotube 582

formation and is the organization of G-C181-OH within the vesicles and tubes the same Lipid 583

nanotubes have been studied for at least three decades over which broad experimental and 584

theoretical insight has been generated Despite this fact the actual understanding is still matter 585

of debate27 Micelle-to-nanotube and vesicle-to-nanotube phase transitions are generally 586

reported for single amphiphiles below a given Tm and intermediate phase composed of twisted 587

and flat fibers is concomitantly reported Initially thought to be a chirality-driven phenomenon 588

recent studies propose a more general mechanism involving crystallization and symmetry 589

breaking in the lipid packing driven by optimal packing27 If twisted ribbons were observed 590

before for a number of microbial glycolipid bolaamphiphiles44 they are not found for G-C181-591

OH This fact as well as the large body of cryo-TEM images and the observation of nanotubes 592

above Tm1 rather pledge in favour of a direct topological change from positive (vesicles) to 593

zero (nanotubes) Gaussian curvatures In this case vesicle-to-nanotube transitions could be 594

driven by external forces56307813ndash162829 595

Considering the absence of external forces in the present system there are only two 596

possibilities left spontaneous budding of the membrane or nanotube pulling after inter-vesicle 597

collisions due to Brownian motion Both are possible as inclusions composed of a different 598

arrangement of the same molecule in the membrane continuum can theoretically explain 599

spontaneous nanotube formation in an isotropic medium672223629ndash1218ndash21 However the second 600

one is probably more likely to prevail because it could explain the large number of nanotubes 601

connecting two distant vesicles Interestingly similar events occur for any phospholipid-based 602

vesicular solution but nanotubes are extremely rare unless at least two lipids of different 603

molecular structure are mixed together The spectroscopic 1H and 13C and relaxation (1H) 604

ssNMR data even if collected on a dry powder demonstrate that the coexistence of two G-605

C181-OH configurations involving both the sugar headgroup and fatty alcohol chain are 606

possible for this molecule As a matter of fact this is not so surprising as more than one energy 607

minima each corresponding to a given conformation are known for glucose its non-reducing 608

derivative methylglucose and carbohydrates in general78ndash80 609

The presence of two configurations could explain the coexistence between two 610

structures of different curvature but the data presented so far do not explain the origin of these 611

two configurations In other words what drives the change in the local structure of the sugar 612

headgroup and fatty alcohol within the membrane 613

The association of known self-assembled structures to spectroscopic 13C ssNMR data is 614

known for years in particular for gluconamides lipids having a linear sugar headgroup81ndash84 It 615

25

was found that gauche effects in some cases up to the γ position in the sugar chain induce 616

upfield chemical shifts while anti and trans conformations result in downfield shifts Each 617

structure (monolayer crystal bilayer crystal or micellar fiber) was then characterized by a given 618

organization of the gluconamide and a characteristic 13C ssNMR fingerprint which could be 619

explained by gauche γ-gauche anti and trans conformations of the headgroup Similar 620

structural-spectroscopic correlations have been collected over the years for much simpler cyclic 621

sugars85 whereas pyranoses can undergo important conformational changes (chair half-chair 622

boat skew) when the appropriate energy barriers are crossed78ndash8086 A quantitative 623

conformational study based on the 13C chemical shift variations shown in Table 1 would be 624

certainly helpful but very delicate in the context of the present state of the art because a 625

rigorous approach requires at least one known crystal structure8182 which is currently missing 626

for this compound A qualitative approach will then be adopted 627

G-C181-OH is a bolaform amphiphile and flip-flop could be a possible mechanism 628

involved in the nanotube stabilization whereas head-tail interactions have long been considered 629

as important in nanotube-forming bolaform amphiphiles6063 From mere considerations based 630

on T2 (1H) relaxation arguments one could exclude head-tail configurations if the R-CH2OH 631

endgroup (H1) was involved in a direct interaction with the sugar group its corresponding T2 632

would probably be comparable to the sugarrsquos one (OH and H1rsquo) but this does not seem to be 633

the case However this hypothesis would be based on the comparison of the relaxation from 634

CH and CH2 protons additionally involved in different kind of motions Besides the relaxation 635

of H1 is influenced by the strong overlap between H1 and H5rsquo For these reasons relaxation 636

arguments are not solid enough and will not be used 637

We then address this question by exploring intermolecular interactions in a direct 638

manner using 2D 1H-1H homonuclear ssNMR correlation spectroscopy At 90degC in the fast-639

tumbling domain of molecular motion (narrow 1H NMR peaks Figure 6a) and corresponding 640

to the vesicle-rich region spin diffusion (NOESY) experiments are preferred over dipolar 641

coupling based experiments because such interaction is averaged out by molecular motion 642

However when the molecular motion is more constrained in the domain of nanotube stability 643

at 70degC (broad 1H spectra Figure 6a) Double-Quantum Single-Quantum (DQ-SQ) correlation 644

experiments are best suited to probe interaction between close molecular neighbors These 645

experiments are reported in Figure 9 and the related interpretation is only briefly reported 646

below while a more detailed discussion can be found on Page 12 in the Supporting Information 647

26

648

Figure 9 ndash a) 2D 1H-1H NOESY NMR experiment performed at MAS= 10 kHz (4 mm probe) on a Bruker 649

300 MHz Avance III HD spectrometer using 300 ms of mixing time Sample temperature is regulated to 650

90degC reached by cooling from 110degC b) 2D 1H-1H DQ-SQ (BAck-to-BAck BABA pulse sequence used for 651

the DQ excitation scheme) NMR experiment performed at MAS= 22 kHz (25 mm probe) on a Bruker 700 652

MHz Avance III spectrometer using a DQ excitation time equal to 2 rotor periods Sample temperature is 653

regulated to 70degC reached by cooling from 110degC The typical schemes illustrating how to interpret these 654

2D experiments are reported above each experimental map Zooms of the sugar region and configurational 655

interpretation are given below each 2D map More information are given in the text below 656

H16-H16

H1-H1

H9-H10

H1rsquo-H1

H1rsquoH910 H16

H1

H1rsquo-H1rsquo

Hxrsquo-Hxrsquo

2

(pp

m)

(ppm)

315

811910 1rsquo

2rsquo2

3rsquo 4rsquo6rsquo

5rsquo

1rsquo2910

asymp asymp asympHDO MeOD Acetone

1

315

811 9101rsquo

2rsquo2

3rsquo4rsquo6rsquo

5rsquo

1rsquo2 9 10

asympasympasympHDOMeODAcetone

H1rsquo-H1

1rsquo

1

(p

pm

)

(ppm)

H1-H2

H1rsquo-H1

H1rsquo

H910

H1H5rsquo

H2

H1rsquo-H5rsquo

315

811910 1rsquo

2rsquo2

3rsquo 4rsquo6rsquo

5rsquo

1rsquo2910

asymp asymp asympHDO MeOD Acetone

1

H2rsquoH3rsquoH4rsquo

H2rsquoH3rsquoH4rsquo

X= 2rsquo3rsquo4rsquo

NOESY - T= 90degC

1H

1H

1H

1H

DQ-SQ - T= 70degC

a) b)

2

(pp

m)

(p

pm

)

(ppm)

A B

AhellipB

AhellipA

BhellipB

NOESY

(ppm)

A B

AhellipB

DQ-SQ

Single-quantum dimension

Do

ub

le-q

ua

ntu

m d

imen

sio

n

AhellipA

BhellipB

BhellipA

BhellipA

315

811910 1rsquo

2rsquo2

3rsquo 4rsquo6rsquo

5rsquo

1rsquo2910

asymp asymp asympHDO MeOD Acetone

1

lt 5 Aring

lt 5 Aring

H1-H2

H811

H15

H16

H6rsquo

H1rsquo

H1

H5rsquo

H2

H1-H2

H1rsquo-H1

H1rsquo-H5rsquo

H1rsquo

H1 H5rsquo

H1rsquo

H1

H5rsquo

H1H2rsquo H3rsquo H4rsquo

H1-H1

H1rsquo-H1

Hxrsquo-HxrsquoX= 2rsquo3rsquo4rsquo

H1-H2

H1rsquo

Zoom sugar region 28 ndash 45 ppmNOESY DQ-SQ

27

Head-tail or head-head configurations could be probed by the existence of H1rsquo-H1 657

interactions whereas H1 (δH= 356 ppm) is associated to the glucolipid tail and H1rsquo (δH= 432 658

ppm) to the sugar headgroup These atoms are more than 15 Aring apart and a direct interaction 659

could not be observed in both NOESY and DQ-SQ experiments NOESY experiments (Figure 660

9a) show that H1rsquo-H1 correlations are not observed Meanwhile clear-cut direct H1rsquo-H1 661

interactions are indeed observed are detected at T= 70degC (DQ cross-peaks related to H1rsquo-H1 662

correlation in DQ-SQ experiment shown in Figure 9b) Interestingly the absence of auto peaks 663

at δH= 432 ppm and δH= 335 ppm in the SQ dimension excludes H1rsquo-H1rsquo and Hxrsquo-Hxrsquo (x= 2 664

3 4) interactions respectively strengthening the idea that sugar headgroups from two nearest-665

neighbor G-C181-OH molecules are not adjacent in the nanotube-vesicle region 666

Homonuclear 2D ssNMR supports the idea according to which G-C181-OH can 667

undergo molecular flip-flops and a head-tail arrangement in the dry product Extrapolating these 668

conclusions in the lower concentration range lack of H1rsquo-H1 proximity at T= 90degC suggests a 669

head-head configuration in the vesicle-rich region and head-tail configuration within the 670

nanotubes It is however still unclear whether flip-flops are driven by spontaneous temperature-671

induced conformational changes in the glucolipid packing stimulating inclusions within the 672

membrane itself672223629ndash1218ndash21 or by defects generated by inter-vesicle collisions In fact a 673

combination of both is not unlikely Nanotubes are not stable when temperature is 20 to 30 674

degrees above Tm1 indicating that inter-vesicle collisions alone cannot explain the unusual 675

phenomenon In an energy landscape approach87 the head-head configuration is certainly 676

favored at high temperature while both head-head and head-tail configurations could be 677

favored in the mid-temperature region corresponding to the range between Tm1 and Tm2 in the 678

dry powder Both statistical and defect-induced flip-flop events contribute to form high mean 679

zero Gaussian curvature morphologies (nanotubes) for which the head-tail conformation 680

limiting the proximity between bulky sugar headgroups is probably more adapted (Figure 8c) 681

The flip-flop interpretation now reasonably explains the strong differences in the sugar and fatty 682

alcohol conformations deduced by the multiple NMR signals attributed to chemically-683

equivalent groups The existence of various low-energy conformations each having specific 684

NMR signature for sugars in solution78ndash8086 and solid-state81ndash84 further supports our 685

interpretation 686

One last point should be clarified Why between Tm1 and Tm2 are two molecular 687

environments able to coexist instead of a single one of average rigidity as found in the case of 688

ldquoripplerdquo phases50ndash54 We believe that this question can be answered based on the NMR results 689

the relatively short T2 values of the H1rsquo and OH protons shows the constrained mobility of the 690

28

sugar headgroup despite a certain flexibility of the chain (long T2 component) This trend may 691

suggest restrictions in translational mobility probably due to the sugar-OH hydrogen bonding 692

(Figure 8c) meaning that lipids are not able to diffuse within the membrane thus keeping two 693

different configurations stable 694

695

Conclusions 696

Topologically-connected nanotube-vesicle lipidic systems are commonly observed for 697

self-assembled membranes composed of at least two amphiphiles and spontaneously driven by 698

anisotropic inclusions andor by external directional forces This work shows that strong 699

differences in membrane curvature can spontaneously occur in single-lipid systems without 700

external anisotropic forces Coexistence and interconnection between nanotubes and vesicles 701

are evidenced via cryo-TEM and fluorescence confocal microscopy in a single-phase region at 702

temperature above a transition temperature Tm1= 483degC determined by DSC on the hydrated 703

sample but below the second transition temperature Tm2 (957degC) determined by DSC on the 704

dry powder X-ray diffraction (WAXS) also shows the coexistence of two lattice periods one 705

attributed to an inter-membrane repeating distance and the other possibly related to the intra-706

nanotube glucolipid arrangement Labelling the systems with Laurdan a fluorophore of which 707

the emission strongly varies with the local membrane rigidity excludes the presence of rigid 708

micro-separated phases (eg Lβ) 709

1H and 13C ssNMR despite the fact that it can only be performed on the dry powder 710

confirm the presence of one main fluid G-C181-OH environment above Tm2 Below Tm2 the 711

splitting of some 13C peaks and the coexistence of short and long 1H relaxation components 712

demonstrate that G-C181-OH can simultaneously pack into structurally and dynamically 713

inequivalent environments Combining advanced 2D 1H-1H correlation experiments based on 714

spin diffusion or double-quanta excitation suggests that vesicles could contain a preferred head-715

head configuration of the bolaform lipid while nanotubes a preferred head-tail assembly 716

The presence of two environments composed of the same molecule characterized by 717

different molecular configurations and rigidity fulfills the theoretical conditions necessary to 718

spontaneously nucleate a membrane inclusion of non-zero average orientation described as 719

necessary to observe topologically-connected nanotube-vesicle structures in the absence of 720

external forces This hypothesis could also be enriched by the possibility of nanotubes to be 721

pulled away between two vesicles after a collision event 722

This work expands the actual theoretical and experimental framework on lipid networks 723

stabilized by a different radius of curvature It could also motivate the synthesis or isolation 724

29

of new lipids providing a higher level of 3D architectural control in soft networks possibly in 725

the range of room temperature 726

727

Acknowledgements 728

Dr Andrea Lassenberger and Dr Sylvain Preacutevost at Institut Laue Langevin (ILL Grenoble 729

France) are kindly acknowledged for their assistance on the SANS experiment Confocal 730

images were performed at the Institut de Biologie Paris Seine (IBPS) imaging facility Jean-731

Franccedilois Gilles (Imaging Core Facility Institut de Biologie Paris Seine (IBPS) CNRS 732

Sorbonne Universiteacute Paris France) is kindly acknowledged for his assistance on the confocal 733

microscopy imaging Dr Guillaume Laurent (LCMCP Sorbonne Universiteacute Paris France) is 734

kindly acknowledged for helpful discussions The French region Ile-de-France SESAME 735

program is acknowledged for financial support (700 MHz NMR 736

spectrometer) Dr Olivier Diat (ICSM Marcoule France) Prof eronikaIglič and Prof V Aleš737

onsul discussi Slovenia) are kindly acknowledged for helpfUniversity of LjubljanaIglič (-Kralj 738

739

Financial support 740

WAXS experiments have been supported by Soleil Light Source Saint Aubin France proposal 741

Ndeg 20201747 SANS experiments have been supported by ILL proposal Ndeg9-13-778 742

743

References 744

1 D Kabaso N Bobrovska W Goacuteźdź E Gongadze V Kralj-Iglič R Zorec and A 745

Iglič Bioelectrochemistry 2012 87 204ndash210 746

2 C R Safinya U Raviv D J Needleman A Zidovska M C Choi M A Ojeda-747

Lopez K K Ewert Y Li H P Miller J Quispe B Carragher C S Potter M W 748

Kim S C Feinstein and L Wilson Adv Mater 2011 23 2260ndash2270 749

3 P Veranič M Lokar G J Schuumltz J Weghuber S Wieser H Haumlgerstrand V Kralj-750

Iglič and A Iglič Biophys J 2008 95 4416ndash4425 751

4 A Rustom R Saffrich I Markovic P Walther and H-H Gerdes Science (80- ) 752

2004 303 1007ndash1010 753

5 M I Angelova S Soleacuteau P Meacuteleacuteard F Faucon and P Bothorel Prog Colloid 754

Polym Sci 1992 89 127ndash131 755

6 L Mathivet S Cribier and P F Devaux Biophys J 1996 70 1112ndash1121 756

7 V Kralj-Iglič G Gomišček J Majhenc V Arrigler and S Svetina Colloids Surfaces 757

A Physicochem Eng Asp 2001 181 315ndash318 758

30

8 V Kralj-Iglič A Iglič G Gomišček F Sevšek V Arrigler and H Haumlgerstrand J 759

Phys A Math Gen 2002 35 1533ndash1549 760

9 M Drab D Stopar V Kralj-Iglič and A Iglič Cells 2019 8 626 761

10 A Zidovska K K Ewert J Quispe B Carragher C S Potter and C R Safinya 762

Langmuir 2009 25 2979ndash2985 763

11 A Zidovska K K Ewert J Quispe B Carragher C S Potter and C R Safinya Soft 764

Matter 2011 7 8363ndash8369 765

12 A Zidovska K K Ewert J Quispe B Carragher C S Potter and C R Safinya 766

Biochim Biophys Acta - Biomembr 2009 1788 1869ndash1876 767

13 M Karlsson K Sott M Davidson A S Cans P Linderholm D Chiu and O Orwar 768

Proc Natl Acad Sci U S A 2002 99 11573ndash11578 769

14 A Karlsson R Karlsson M Karlsson A-S Cans A Stromberg F Ryttsen and O 770

Orwar Nature 2001 409 150ndash152 771

15 M Karlsson K Sott A S Cans A Karlsson R Karlsson and O Orwar Langmuir 772

2001 17 6754ndash6758 773

16 E Evans H Bowman A Leung D Needham and D Tirrell Science (80- ) 1996 774

273 933ndash935 775

17 U Seifert Configurations of fluid membranes and vesicles 1997 vol 46 776

18 J Kaumls and E Sackmann Biophys J 1991 60 825ndash844 777

19 N Bobrovska W Goacuteźdź V Kralj-Iglič and A Iglič PLoS One 2013 8 e73941 778

20 A Iglič B Babnik K Bohinc M Fošnarič H Haumlgerstrand and V Kralj-Iglič J 779

Biomech 2007 40 579ndash585 780

21 K Bohinc T Slivnik A Iglič M Brumen and V Kralj-Iglič J Phys Chem C 2007 781

111 9709ndash9718 782

22 A Iglič H Haumlgerstrand P Veranič A Plemenitaš and V Kralj-Iglič J Theor Biol 783

2006 240 368ndash373 784

23 V Kralj-Iglič and P Veranič Adv Planar Lipid Bilayers Liposomes 2006 5 129ndash785

149 786

24 A Iglic and V Kralj-Iglic Adv Planar Lipid Bilayers Liposomes 2006 4 253ndash279 787

25 S Penič L Mesarec M Fošnarič L Mroacutewczyńska H Haumlgerstrand V Kralj-Iglič and 788

A Iglič Front Phys 2020 8 342 789

26 P Yager and P E Schoen Mol Cryst Liq Cryst 1984 106 371ndash381 790

27 T G Barclay K Constantopoulos and J Matisons Chem Rev 2014 114 10217ndash791

10291 792

31

28 B Božič G Gomišček V Kralj-Iglič S Svetina and B Žekš Eur Biophys J 2002 793

31 487ndash496 794

29 A Iglič H Haumlgerstrand M Bobrowska-Haumlgerstrand V Arrigler and V Kralj-Iglič 795

Phys Lett A 2003 310 493ndash497 796

30 R Bar-Ziv and E Moses Phys Rev Lett 1994 73 1392ndash1395 797

31 S Chiruvolu H E Warriner E Naranjo S H Idziak J O Raumldler R J Plano J A 798

Zasadzinski and C R Safinya Science (80- ) 1994 266 1222ndash1225 799

32 V Kralj-Iglič A Iglič H Haumlgerstrand and P Peterlin Phys Rev E 2000 61 4230ndash800

4234 801

33 L Mesarec M Drab S Penič V Kralj-Iglič and A Iglič Int J Mol Sci 2021 22 802

2348 803

34 L Van Renterghem S L K W Roelants N Baccile K Uyttersprot M C Taelman 804

B Everaert S Mincke S Ledegen S Debrouwer K Scholtens C V Stevens W 805

Soetaert L Van Renterghem S L K W Roelants N Baccile K Uytersprot M C 806

Taelman B Everaert S Mincke S Ledegen S Debrouwer K Scholtens C V 807

Stevens and W Soetaert Biotechnol Bioeng 2018 115 1195ndash1206 808

35 J Clark and F Deswarte Introduction to Chemicals from Biomass Second Edition 809

John Wiley amp Sons 2015 810

36 K V Sajna R Houmlfer R K Sukumaran L D Gottumukkala and A Pandey in 811

Industrial Biorefineries and White Biotechnology eds A Pandey R Houmlfer M 812

Taherzadeh K M Nampoothiri and C Larroche Amsterdam Oxford Waltham 813

Elsevier 2015 pp 499ndash521 814

37 R Marchant and I M Banat Trends Biotechnol 2012 30 558ndash565 815

38 K Hill and C LeHen-Ferrenbach in Sugar-Based Surfactants Fundamentals and 816

Applications ed C C Ruiz Boca Raton CRC Press 2009 pp 1ndash20 817

39 L Rodrigues I M Banat J Teixeira and R Oliveira J Antimicrob Chemother 818

2006 57 609ndash618 819

40 J D Desai and I M Banat Microbiol Mol Biol Rev 1997 61 47ndash64 820

41 S L K W Roelants K Ciesielska S L De Maeseneire H Moens B Everaert S 821

Verweire Q Denon B Vanlerberghe I N A Van Bogaert P Van der Meeren B 822

Devreese and W Soetaert Biotechnol Bioeng 2016 113 550ndash559 823

42 N Baccile M Selmane P Le Griel S Preacutevost J Perez C V Stevens E Delbeke S 824

Zibek M Guenther W Soetaert I N A Van Bogaert and S Roelants Langmuir 825

2016 32 6343ndash6359 826

32

43 N Baccile A-S Cuvier S Preacutevost C V Stevens E Delbeke J Berton W Soetaert 827

I N A Van Bogaert and S Roelants Langmuir 2016 32 10881ndash10894 828

44 N Baccile C Seyrig A Poirier S A Castro S L K W Roelants and S Abel Green 829

Chem 2021 23 3842ndash3944 830

45 A Salis L Cappai C Carucci D F Parsons and M Monduzzi J Phys Chem Lett 831

2020 11 6805ndash6811 832

46 J Schindelin I Arganda-Carreras E Frise V Kaynig M Longair T Pietzsch S 833

Preibisch C Rueden S Saalfeld B Schmid J-Y Tinevez D J White V 834

Hartenstein K Eliceiri P Tomancak and A Cardona Nat Methods 2012 9 676ndash835

682 836

47 T Parasassi and E Gratton J Fluoresc 1995 5 59ndash69 837

48 M M Dodes Traian F L Gonzaacutelez Flecha and V Levi J Lipid Res 2012 53 609ndash838

616 839

49 D M Owen C Rentero A Magenau A Abu-Siniyeh and K Gaus Nat Protoc 840

2012 7 24ndash35 841

50 W Stillwell in An Introduction to Biological Membranes Academic Press 2016 pp 842

135ndash180 843

51 W Stillwell in An Introduction to Biological Membranes Academic Press 2016 pp 844

181ndash220 845

52 B Tenchov Chem Phys Lipids 1991 57 165ndash177 846

53 R Koynova and M Caffrey Biochim Biophys Acta 1998 1376 91ndash145 847

54 F Schmid S Dolezel O Lenz and S Meinhardt J Phys Conf Ser 2014 487 848

012004 849

55 S Meinhardt R L C Vink and F Schmid Proc Natl Acad Sci U S A 2013 110 850

4476ndash4481 851

56 O Glatter and O Kratky Small Angle X-ray Scattering Academic Press London 852

1982 853

57 B A Cunningham A D Brown D H Wolfe W P Williams and A Brain Phys 854

Rev E 1998 58 3662ndash3672 855

58 M Nadler A Steiner T Dvir O Szekely P Szekely A Ginsburg R Asor R Resh 856

C Tamburu M Peres and U Raviv Soft Matter 2011 7 1512ndash1523 857

59 C Tanford The hydrophobic effect Formation of micelles and biological membranes 858

John Wiley amp Sons Inc New York 1973 859

60 M Masuda and T Shimizu Langmuir 2004 20 5969ndash5977 860

33

61 A Zidovska K K Ewert J Quispe B Carragher C S Potter and C R Safinya 861

Block Liposomes Vesicles of Charged Lipids with Distinctly Shaped Nanoscale 862

Sphere- Pear- Tube- or Rod-Segments Elsevier Inc 1st edn 2009 vol 465 863

62 A Iglic V Kralj-Iglic B Božič M Bobrowska-Haumlgerstrand B Isomaa and H 864

Haumlgerstrand Bioelectrochemistry 2000 52 203ndash211 865

63 T Shimizu M Masuda and H Minamikawa Chem Rev 2005 105 1401ndash1444 866

64 J V Selinger and J M Schnur Phys Rev Lett 1993 71 4091ndash4094 867

65 J M Schnur Science (80- ) 1993 262 1669ndash1676 868

66 B Tenchov R Koynova and G Rapp Biophys J 2001 80 1873ndash1890 869

67 R Koynova and B Tenchov Ser Biomech 2017 31 3ndash19 870

68 S J Angyal and R Le Fur Carbohydr Res 1984 126 15ndash26 871

69 J K Whitesell and M A Minton J Am Chem Soc 1987 109 225ndash228 872

70 T Takayama S Ando and I Ando J Mol Struct 1990 220 245ndash250 873

71 K Gawrisch N V Eldho and I V Polozov Chem Phys Lipids 2002 116 135ndash151 874

72 F Tielens C Gervais J F Lambert F Mauri and D Costa Chem Mater 2008 20 875

3336ndash3344 876

73 G A Jeffrey and Y Yeon Acta Crystallogr Sect B 1986 42 410ndash413 877

74 B Berglund and R W Vaughan J Chem Phys 1980 73 2037ndash2043 878

75 C Gervais C Coelho T Azaiumls J Maquet G Laurent F Pourpoint C Bonhomme 879

P Florian B Alonso G Guerrero P H Mutin and F Mauri J Magn Reson 2007 880

187 131ndash140 881

76 C Akita T Kawaguchi F Kaneko H Yamamoto and M Suzuki J Phys Chem B 882

2004 108 4862ndash4868 883

77 R N Purusottam L Seacutenicourt J J Lacapegravere and P Tekely Biochim Biophys Acta - 884

Biomembr 2015 1848 3134ndash3139 885

78 X Biarneacutes A Ardegravevol A Planas C Rovira A Laio and M Parrinello J Am Chem 886

Soc 2007 129 10686ndash10693 887

79 L M J Kroon‐Batenburg and J Kroon Biopolymers 1990 29 1243ndash1248 888

80 K N Kirschner and R J Woods Proc Natl Acad Sci U S A 2001 98 10541ndash889

10545 890

81 S Svenson B Kirste and J H Fuhrhop J Am Chem Soc 1994 116 11969ndash11975 891

82 I Sack S Macholl J H Fuhrhop and G Buntkowsky Phys Chem Chem Phys 892

2000 2 1781ndash1788 893

83 S Svenson J Koening and J H Fuhrhop J Phys Chem 1994 98 1022ndash1028 894

34

84 S Svenson A Schaefer and J H Fuhrhop J Chem Soc Perkin Trans 2 1994 2 895

1023ndash1028 896

85 K Bock and C Pedersen Adv Carbohydr Chem Biochem 1983 41 27ndash66 897

86 T B Grindley Structure and Conformation of Carbohydrates Springer Berlin 898

Heidelberg 2008 899

87 F Tantakitti J Boekhoven X Wang R V Kazantsev T Yu J Li E Zhuang R 900

Zandi J H Ortony C J Newcomb L C Palmer G S Shekhawat M O De La 901

Cruz G C Schatz and S I Stupp Nat Mater 2016 15 469ndash476 902

903

1

Supporting Information 904

905

Topological connection between vesicles and nanotubes in single-906

component lipid membranes driven by head-tail interactions 907

908

Niki Baccilea Ceacutedric Lorthioira Abdoul Aziz Baa Patrick Le Griela Cristina Coelhob Javier 909

Perezc Wim Soetaertd Sophie L K W Roelantsd 910

911

a Sorbonne Universiteacute Centre National de la Recherche Scientifique Laboratoire de Chimie de 912

la Matiegravere Condenseacutee de Paris LCMCP F-75005 Paris France 913

b Institut des Mateacuteriaux de Paris Centre Sorbonne Universiteacute Paris France 914

c Synchrotron Soleil LrsquoOrme des Merisiers Saint-Aubin BP48 91192 Gif-sur-Yvette Cedex 915

France 916

d InBio Department of Biotechnology Ghent University Ghent Belgium 917

918

919

Corresponding author 920

Dr Niki Baccile 921

E-mail address nikibaccilesorbonne-universitefr 922

Phone +33 1 44 27 56 77 923

924

925

926

2

a) 927

Parameter Method

Dry Matter (DM ) 996 Infrared balance 105degC

Glucose () 012 HPLC-Metacarb

Glycerol () nd HPLC-Metacarb

Free Fatty Acid content () lt001 Internal method BBEPP

GC with FID Detector

Oil () lt001 Internal method BBEPP

GC with FID Detector

Moulds (CFUg) lt10 3M-Nordval ndeg16 B

Yeasts (CFUg) lt10 3M-Nordval ndeg16 B

Anaerobic count (CFUg) lt10 SP-VG M005 B

Protein () tbd BCA protein assay

928

Figure S 1 - a) HPLC-ELSD chromatogram and composition table 929

930

931

932

933

3

b) 934

935

Group Ndeg δ ppm

CH2 1 354

CH2 2 153

CH2 3-712-14 132

CH2 811 203

CH 910 534

CH2 15 162

CH2 16 388

CH 1 424

CH 2 316

CH 3 334

CH 4 329

CH 5 354

CH2 6 366

Figure S 1 (next) ndash b) 1H NMR spectrum (MeOD-d4) recorded on the purified G-C181-OH compounds 936

and related assignment 937

938

2

811910 1rsquo

5rsquo1

3rsquo4rsquo

1rsquo1910

asymp asymp asympHDO MeOD Acetone

6rsquo2rsquo

1615

3-712-14

4

939

940

Figure S 2 ndash Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 130degC 941

5

942

943

Figure S 3 - Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 90degC 944

945

90degC

1

22

6

946

947

948

949

Figure S 4 ndash Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 70degC 950

951

21

11

1

2

5

3

4

45

5

6

7

7

7

952

953

954

Figure S 5 ndash Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 25degC 955

956

957

1

1 1

8

958

Figure S 6 ndash a) SANS experiment recorded at 90degC for a 5 mgmL G-C181-OH aqueous solution b-c) 959

WAXS experiments recorded on aqueous solutions of G-C181-OH at b) 5 mgmL and c) 25 mgmL d) 960

Structural interpretation of the repeating distances pointed out in c) 961

962

963

964

001 01 1

q Aring-1

25degC

68degC

88degC

111degC

01 1q Aring

-1

02 04 06 08 1q Aring

-1

001 01

10-1

100

101

Inte

nsity

cm

-1

q Aring-1

-2

T= 90degC

SANS WAXS WAXS

a) b) c)

dv

dL d= 144 Aring-1

di

ddH2O

dt

dvdL

dtdi d

d)

9

965

966

Figure S 7 ndash ssNMR MAS (10 kHz) experiments recorded on a G-C181-OH dry powder a) 1H spectra 967

recorded from 110degC to 70degC b) 13C single pulse (SP) spectra recorded with high power 1H decoupling from 968

110degC to 70degC c) 13C cross polarization (CP) spectra recorded at 110degC and contact time tc of 50 micros and 969

3000 micros d) 13C CP spectra recorded at 70degC and for 20 lt tc micros lt 3000 970

971

1

811

910

CH2 CH2

1rsquo

6rsquo

8 6 4 2 0 ppm

110

90

80

70

910

1rsquo

6rsquo16

1

2rsquo3rsquo4rsquo

811

OH

45 40 35 30 25 20 15

ppm

200 180 160 140 120 100 80 60 40 20 0

ppm140 120 100 80

ppm

CH2-gauche

3004

110

90

80

70

65

60

50

25

C91013003

C9 - 12869

CH2-all-trans

3198

C1rsquo10335

a) b)

1H

T degCT degC

13C Single Pulse MAS

A

B

A

B

70degC70degC

5rsquo

215

200 180 160 140 120 100 80 60 40 20 0

ppm

13C CPMAST= 110degC

200 180 160 140 120 100 80 60 40 20 0

ppm

132 131 130 129 128 127 126

ppm

106 105 104 103 102

ppm

13C CPMAST= 70degC

50

tc micros

3000

20

tc micros

50 500 1000 3000

tc micros

C102

13024 C910113003

C92

12869

C1rsquo213082

C1rsquo110443

A B

106 105 104 103 102

ppm

3000

1000

500

50

20

AB

c)

d)

10

972

Figure S 8 ndash a) Short and d) long T2 components displayed by the 1H transverse relaxation signal c) 1H 973

transverse relaxation signals recorded on a G-C181-OH dry powder from 110degC to 70degC The area of each 974

given resonance is plotted (log-lin) as a function of the echo time (CPMG pulse sequence) for each 975

temperature and fitted using the equation given above T2s and T2l indicate the short and long T2 relaxation 976

times respectively obtained by describing the 1H transverse relaxation signal using a biexponential decay 977

For H1rsquo H5rsquo and H6rsquo the 1H relaxation function could be described by a monoexponential decay 978

- Protons from the sugar group (H1rsquo H5rsquo H6rsquo) show a single T2 relaxation component (Figure 979

6c and Figure S 8) while protons of the acyl chain (CH2 H910 H811) show both short (Figure 980

6c) and long (Figure 6d) T2 relaxation component 981

- The short T2 relaxation time systematically decreases from 110degC to 70degC for all species by a 982

factor ranging from two to seven depending on the specific chemical group Very long T2 are 983

observed at 110degC for all acyl species while nearly unchanged values are measured below 984

90degC without any clear trend 985

- The sugar protons (H1rsquo H5rsquo H6rsquo) have the shortest set of T2 values (lt 1 ms) below 100degC 986

The long component of T2 generally ranges between 4 ms and 12 ms 987

0 1 2 3 4 5 6 7 8

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1 2 3 4 5 6 7 8

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

H1

H8H11

H9H10

CH2

H1rsquo

H6rsquo

CH2

H910H811 H1CH2

H1rsquo H5rsquo6rsquo

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

OH

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45 T= 110degC

T= 90degC

T= 80degC

T= 70degC

Log

In

ten

sity

Time ms

Fit

)

CH2 H910 H811 H1 H56 H1 OH

00

05

10

15

20

25

30

35

40

T2 m

s

T= 110degC

T= 90degC

T= 80degC

T= 70degC

CH2 H910 H811 H1

0

2

4

6

8

10

12

T2 m

s

T= 110degC

T= 90degC

T= 80degC

T= 70degC

Short T2(1H) Long T2(1H)

a) b)

c)

11

Table S 1 ndash 1H chemical shifts of G-C181-OH in solution and in the solid-state Solid-state spectra are 988

recorded at a MAS frequency of 10 kHz in DMSO-d6 broad peak 989

990

δH ppm

Group Ndeg Liquid

(MeOD-d4) Solid

(T= 110degC) Solid

(T= 70degC)

CH2 H1 354 357

CH2 H2 153 156

CH2 H3-712-14 132 135

CH2 H811 203 203

CH H910 534 534

CH2 H15 162 164

CH2 H16 388 383

CH H1 424 432

CH H2 316 335

CH H3 334 335

CH H4 329 335

CH H5 354 351 357

CH2 H6 366 383

OH OH1 432

45 463

OH OH2 492

OH OH3 490

OH OH4 487

OH OH6 445

991

992

H1H9H10H1rsquo

H6rsquo

OH2rsquo

OH1

12

Detailed discussion about the NOESY and DQ-SQ 2D 1H-1H ssNMR experiments presented in 993

Figure 9 994

995

2D 1H-1H ssNMR experiments are a unique tool to explore through-space proximities 996

between 1H nuclei ssNMR offers a large number of possibilities each adapted to a given 997

system The advantage of 2D maps with respect to a dynamics approach is their direct 998

readability in clear two nuclei which exchange their magnetization through-space (whichever 999

mechanism may be responsible for this exchange) will display a specific cross-peak However 1000

the way cross-peaks are presented and interpreted strictly depends on the nature of the pulse 1001

program which in turn is chosen according to the physical nature of the sample 1002

1D temperature-dependent 1H NMR experiments performed on G-C181-OH (Figure 1003

4a in the main text) indicate sharp peaks at 90degC and broad peaks at 70degC Narrow peaks 1004

indicate fast andor large-amplitude molecular tumbling averaging the strong 1H-1H dipolar 1005

interactions This excludes the use of pulse sequences like DQ-SQ which is not efficient under 1006

these conditions For this reason we employ spin-diffusion-based pulse sequences like 1007

NOESY which is more adapted for samples with high molecular mobility When mixing times 1008

are chosen to be short (lt 1 s) one generally explores through-space proximities within about 5 1009

10 Aring On the contrary when molecular motions are restricted dipolar coupling becomes 1010

significant 1H NMR peaks are broader and pulse sequences like DQ-SQ can now be used 1011

with a spatial resolution of less than 5 Aring 1012

Whenever possible DQ-SQ experiments are preferred over spin diffusion-based pulse 1013

sequences because they are more selective in particular for likewise interactions which are 1014

not discriminated in NOESY-type experiments This is illustrated by the typical interaction 1015

schemes for NOESY (top Figure 9a) and DQ-SQ (top Figure 9b) For two given A and B 1016

protons 2D NOESY maps will display two on-diagonal and two square-shaped symmetrical 1017

off-diagonal cross-peaks Off-diagonal cross-peaks illustrate the mutual A-B and B-A 1018

interaction but the on-diagonal peaks are systematic and do not prove any A-A or B-B 1019

interaction Similarly to NOESY 2D DQ-SQ maps display off-diagonal peaks (symmetric 1020

across the diagonal top Figure 9b) upon A-B and B-A interaction but in contrast they will 1021

display on-diagonal peaks only in the presence of self-interaction In the tutorial example of 1022

Figure 9b top the on-diagonal A-A peak indicates mutual A-A interaction while lack of on-1023

diagonal B-B peak evidences the absence of B-B proximities Combining NOESY and DQ-SQ 1024

experiments is capital to study the local arrangement of G-C181-OH and its evolution with 1025

temperature 1026

13

At T= 90degC the H1 and H5rsquo contributions are very close but possible to discriminate 1027

them (δ(H1)= 354 ppm δ(H5rsquo)= 351 ppm) according to 1H NMR chemical shifts in Table S 1028

1 2D 1H-1H NOESY in Figure 9a also discriminates between the two components attributed 1029

on the basis of the interactions with their mutual neighbors (green H1-H2 and H5rsquo-H1rsquo squares 1030

in Figure 9a) On this basis 2D NOESY does not show a direct H1rsquo-H1 correlation (blue red-1031

crossed) but rather an extensive network of intra-sugar interactions These features exclude 1032

suggest intra- and inter-sugar interactions and they exclude head-tail arrangements as a major 1033

molecular configuration at 90degC 1034

At 70degC the 1H NMR peaks are broader indicating a reduction of the T2 (1H) relaxation 1035

times andor residual homonuclear dipolar interactions excluding the use of NOESY In this 1036

case 1H-1H DQ-SQ correlation experiments are best suited Figure 9b presents the 1H-1H DQ-1037

SQ experiment on G-C181-OH at 70degC Peaks are broader than in NOESY and attribution may 1038

be cumbersome However the experiment is performed at 700 MHz and MAS= 22 kHz which 1039

improves spectral resolution with respect to the same experiment performed at 300 MHz and 1040

MAS= 10 kHz The auto-peak at δH= 356 ppm in the SQ dimension is attributed to H1-H1 spin 1041

pairs mainly resulting from intra CH2 interaction while the DQ cross-peaks at δH= 156-356 1042

ppm and δH= 432-356 ppm respectively origin from H2-H1 and H1rsquo-H1 interactions (Figure 1043

9b) DQ-SQ map does not show any auto-peak corresponding to H1rsquo-H1rsquo or Hxrsquo-Hxrsquo (x= 2rsquo 1044

3rsquo 4rsquo) interactions (red-crossed blue Figure 9b) thus excluding any proximity (lt 5 Aring) between 1045

close glucose neighbors On the contrary DQ-SQ maps (recorded both at 700 MHz Figure 9b 1046

or at 300 MHz not shown) systematically show DQ cross-peaks which can only be attributed 1047

to H1rsquo-H1 interactions (δH= 432-356) The presence of these H1rsquo-H1 interactions can only be 1048

explained by a head-tail arrangement of the G-C181-OH molecules as illustrated below the 1049

DQ-SQ map in Figure 9b 1050

7

Fluorescence microscopy Images were recorded using a 40x objective on a Leica SP5 upright 184

confocal microscope with 405 nm laser excitation The sample was marked with 6-dodecanoyl-185

NN-dimethyl-2-naphthylamine (Laurdan Mw= 35354 gmol-1) fluorophore as follows A 186

stock solution of 3 mg of Laurdan in 100 microL of acetone was initially prepared (C= 30 mgmL-187

1) A volume of 028 microL of the stock Laurdan solution was diluted in 1 mL of a 5 mgmL-1 G-188

C181-OH water solution temporary set at 110degC in a sealed Eppendorf The latter was 189

important so to enhance inclusion of Laurdan in the membrane constituted by the glucolipid 190

while in its vesicle phase The final Laurdan concentration was approximately 24 micromol with a 191

molar Laurdan-to-G-C181-OH molar ratio of approximately 1-to-500 Fluorophore-to-lipid 192

ratios above 1200 were generally considered as non-interfering with the lipid phase behavior 193

as also verified in this work 194

Laurdan can be excited between 365 nm and 410 nm and its emission wavelength can 195

vary according to the rigidity of the membrane intercalation within a rigid or flexible 196

membrane generates emissions at 440 nm or 490 nm respectively47 In the present experiment 197

we have flame-sealed a solution of G-C181-OH at 5 mgmL-1 containing Laurdan (1-to-500 in 198

lipid-to-Laurdan ratio) in flat optical capillaries of 01 mm thickness Capillaries were set on a 199

Linkam support for temperature control eventually placed under the microscope The sample 200

was excited at 405 nm and detection was made in spectral mode with a 5 nm window sliding 201

every 5 nm from 412 nm to 602 nm in 16 bits which allows quantify the signal and compare 202

the different conditions Quantitative analysis was performed according to the literature4849 203

several ROI (range of interest) have been drawn on several images using the ROI manager of 204

Fiji (is just ImageJ) software The integrated intensity of each ROI at 440 nm and 490 nm was 205

eventually recorded and used to calculate the value of the generalized polarization GP as 206

follows of 207

119866119875 =(120582440 minus 120582490)

120582440 + 120582490 208

A negative GP indicates a flexible membrane while a positive GP indicates a rigid membrane 209

210

Solid-state Nuclear Magnetic Resonance (ssNMR) 1H and 13C ssNMR experiments have been 211

performed using an Avance III HD Bruker 705 T (ν1H= 300 MHz) spectrometer and a 4 mm 212

magic angle spinning (MAS) probe except for the BAck-to-BAck (BABA) experiment which 213

was performed on an Avance III Bruker 1644 T (ν1H= 700 MHz) spectrometer equipped with 214

25 mm MAS probe Temperature was controlled through a Smart cooler BCUIIa BCU-xtreme 215

unit using 10 min of equilibration time after each temperature variation Temperature was 216

8

always varied from high (110degC) to low (70degC) before analysis Due to combined use of fast 217

MAS rates necessary to achieve high resolution spectra and heating close to the boiling point 218

of water neither diluted nor concentrated samples could be used for ssNMR experiments All 219

tests employing up to 50 wt G-C181-OH in water resulted in sample centrifugation drying 220

but also expulsion from the rotor causing sudden stop of rotor spinning For these reasons all 221

attempts to study hydrated G-C181-OH failed and only the dry powder could be analyzed The 222

dry sample powder was analyzed as such in zirconia rotors (4 mm or 25 mm) spinning at MAS= 223

10 kHz (4 mm) or 22 kHz (25 mm) and Chemical shifts were calibrated with respect to 224

adamantane (13C 13C(CH2)= 3848 ppm = 191 ppm) and glycine alpha (13C(CO)= 17603 225

ppm) Details on the acquisition parameters are given in the Supporting Information 226

227

Results and discussion 228

Topologically-connected nanotubes and vesicles 229

G-C181-OH is a biobased glucolipid water-insoluble at room temperature and of 230

which the physicochemical properties are poorly known The DSC thermogram performed at 231

10degCmin of G-C181-OH as a function of the water content (Figure 2a) shows two endothermic 232

peaks on the dry powder (996 wt glucolipid) They are attributed to two melting phenomena 233

occurring at the following temperatures Tm 483degC (Tm1) and 957degC (Tm2) the enthalpy of the 234

former being lower than that of the latter Interestingly DSC also shows that Tm1 is hydration 235

independent and reversible (the associated exothermic peak is systematically observed) while 236

the transition at Tm2 is simply suppressed (here tested up to 130degC) on hydrated powders and 237

never observed on the cooling profile even at slow cooling rates (1degCmin) The thermogram 238

recorded on the dry powder is typical of single-component lipids having an intermediate50 239

metastable phase (often referred to the ripple phase Pβ´ in reference to the periodic undulation 240

of the bilayer)50ndash54 between the classical lamellar gel (Lβ) and liquid crystalline (Lα) phases50ndash241

54 and characterized by regions of both liquid order and disorder5455 The effect of water does 242

not influence the low-temperature crystallization at Tm1 but it contributes to disrupt the ordered 243

array of hydrocarbon chains in the metastable phase 244

On the basis of the DSC thermogram the aqueous phase behavior of G-C181-OH under 245

diluted conditions (5 mgmL-1) was studied at four different temperatures (red marks on Figure 246

2a) below and above Tm1 but also between Tm1 and Tm2 corresponding to the thermogram of 247

the dry powder 248

9

249

Figure 2 ndash a) DSC thermogram of G-C181-OH under dry (04 wt water estimated by thermogravimetric 250

analysis) and hydrated conditions recorded at a heating rate of 10degCmin Red marks indicate the 251

temperatures at which cryo-TEM experiments are performed on the corresponding 5 mgmL aqueous 252

solutions b) Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 130degC 253

90degC 70degC and 25degC T indicates the temperature of the sample solution before cryofixation Tne T T 254

being the temperature of the sample upon plunging in liquid ethane Please refer to the materials and 255

methods section for more information 256

T= 130degC

T= 90degC

T= 70degC

T= 25degC

a)

b)

20 40 60 80 100 120

99 wt

50 wt

25 wt

T degC

En

do

10 wt

Tm1= 483 degC

Tm2= 957 degC

10

257

The series of cryo-TEM images in Figure 2b show an overview of the most relevant 258

structures observed on G-C181-OH solution from T= 130degC to T= 25degC whereas T is 259

indicative of the equilibrium temperature before cryofixation (refer to materials and methods 260

section for more information on the sense and interpretation of T) At Tgt 100degC the sample 261

is essentially composed of polydisperse single wall vesicles massively shown by the 262

complementary pictures presented in Figure S 2 In between Tm1 and Tm2 at T= 90degC and T= 263

70degC (Figure 2b Figure 3 Figure 4 complemented by Figure S 3 and Figure S 4) vesicles are 264

connected by nanotubes At Tlt Tm1 flat crystals are otherwise observed (Figure 2b and Figure 265

S 5) 266

267

268

Figure 3 ndash Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 90degC 269

270

A deeper insight at T= 90degC is given in Figure 3 (more images are provided in Figure 271

S 3) Arrows 1 through 6 mark very clearly the typical continuity in the glucolipid membrane 272

between the vesicle and nanotubes Arrow Ndeg 7 shows that a single nanotube connects the outer 273

membrane of two vesicles several hundred nanometers apart The connection is shown by 274

2

34

5

6

6

1

5

7

A

B

D

C

0 10 20 30 40 50 60

BIn

tensity

au

Profile nm

A

0 10 20 30 40 50 60

Inte

nsity

au

Profile nm

C

D

T= 90degC

11

arrows Ndeg5 and Ndeg6 The corresponding profiles A through D confirm the tubular nature of 275

the elongated objects Profiles B and D are both characterized by typical symmetrical thick 276

walls (low intensity depths highlighted by shaded rectangles) surrounding a hollow region (high 277

intensity peak between the shaded rectangles) The wall thickness varies between 35 and 40 278

nm for profile B and between 41 and 61 nm for profile D while the total nanotube diameter 279

varies between 14 nm at B and 20 nm at D At the same time the thickness of the vesicle wall 280

is about 6 nm at A and 135 nm at C Considerations about the bilayer or interdigitated structure 281

of the membrane will be presented below 282

The connection between nanotubes and vesicles seems to be multiple For instance 283

arrow Ndeg1 (Figure S 3) and Ndeg4 (Figure 3) show a single-wall vesicle of which the membrane 284

is extruded into a nanotube Similarly a closer look at arrows Ndeg1 2 and 3 in Figure 3 show 285

that the vesicle is single-wall and fully closed while the nanotube nucleates at the vesicle outer 286

surface with no apparent topological continuity In many regions of the sample at T= 90degC 287

one can find exvaginations or buds232531 in the nanotube wall This is shown for instance by 288

arrows Ndeg7 and particularly put in evidence in Figure 4 by arrows Ndeg1 through Ndeg4 where the 289

amplitude of the oscillation varies between 20-25 nm (arrow Ndeg1) and 6-8 nm (arrows Ndeg3-4) 290

291

Figure 4 - Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heated at T= 90degC 292

293

Still between Tm2 and Tm1 but at lower temperature (T= 70degC) the nanotubes seem to 294

vary from single threads (arrow Ndeg1 Figure S 4) to nanotubes of diameter below 15 nm both 295

connecting nanotubes of diameter ranging between 50 nm and 100 nm (arrows Ndeg2 Figure S 296

4) The larger tubes are most likely formed by the inflation and flattening of the nanotubes 297

probably after assimilation of surrounding vesicles and fusion of the exvagination discussed 298

above The transition between vesicles tubes and flat membranes is certainly more accentuated 299

Arrow Ndeg3 in Figure S 4 shows distinct vesicles while arrow Ndeg4 identifies a flat structure 300

evidently formed from fused vesicles In particular arrow Ndeg5 points at the fusion of two 301

2

1 132

4

12

vesicles while arrow Ndeg6 points at two vesicles fusing into the flat membrane Additional areas 302

where vesicle fusion is prominent are indicated by the oscillations at the outer skirt of broad 303

lamellae and pointed by arrow Ndeg7 304

Finally below Tm1 at T= 25degC for instance mainly flat crystals with no visible 305

exvaginations or tubes are visible (Figure S 5) by cryo-TEM and the material turns into an 306

insoluble precipitate Massive presence of flat crystalline structures is also clear from the bulk 307

iridescent behavior of a G-C181-OH solution under manual shear 308

In summary despite the uncertainty related to the actual state of equilibrium of the 309

glucolipids at the moment of cryofixation cryo-TEM depicts a vesicle phase at Tgt 100degC a 310

flat lamellar phase below Tm1 and an interconnected vesicle-nanotube network between Tm1 and 311

about 100degC The combination of statistically-relevant fluorescence microscopy and 312

scatteringdiffraction experiments provides an additional proof of the coexistence of 313

vesiclenanotube systems between the two transition temperatures Please note that the 314

experiments reported below have been performed under controlled conditions of temperature 315

which is then indicated by the use of T instead of T 316

13

317

Figure 5 ndash a-c) Confocal fluorescence microscopy images recorded at 90degC and 70degC for a 5 mgmL G-318

C181-OH aqueous solution d-g) Typical fluorescence microscopy images recorded at T= 80degC using 319

Laurdan (refer to materials and methods section for more details on this fluorophore) Excitation 320

wavelength is 405 nm for all the images while emission wavelength is 490 nm in (df) and 440 nm in (eg) 321

h) Full emission spectra obtained from the integration of various fluorescence images recorded from 412 322

nm to 600 nm (step 5 nm) at 90degC 80degC and 70degC i) Box plot of GP defined as 119918119927 =(120640120786120786120782minus120640120786120791120782)

120640120786120786120782+120640120786120791120782 with 120640120786120786120782 323

and 120640120786120791120782 being the emission wavelengths at 440 nm and 490 nm at 90degC 80degC and 70degC Each data point 324

l= 490 nm l= 440 nm

l= 490 nm l= 440 nm

10 μm400 450 500 550 600

00

02

04

06

08

10

No

rma

lize

d I

nte

nsity

l nm

90degC

80degC

80degC

70degC

70degC

400 450 500 550 60000

02

04

06

08

10

No

rma

lize

d I

nte

nsity

l nm

90degC

80degC

80degC

70degC

70degC

d) e)

f) g)

h)

i)

T= 80degC

Laurdan

T= 80degC

T= 80degC T= 80degC

T= 90degC T= 70degC T= 70degC

a) b) c)

1

1

1

14

corresponds to a set of regions of interest (ROIs) (54 at 90degC 65 at 80degC and 75 at 70degC) selected in the 325

corresponding fluorescence images ROIs were representatively selected so to include the entire image but 326

also vesicles alone tubes alone and mixtures of vesicles and tubes 327

328

Figure 5ab shows two typical confocal fluorescence microscopy images recorded on a 329

G-C181-OH solution (5 mgmL) in a flame-sealed capillary heat at T= 90degC and T= 70degC 330

Although the resolution of optical microscopy is not comparable with cryo-TEM the sample is 331

broadly composed of vesicles coexisting with nanotubes whereas in some cases the nanotube 332

starts from the vesicle itself (arrow Ndeg1) in good agreement with cryo-TEM data Furthermore 333

SANS recorded at T= 90degC (Figure S 6a) shows a scattering profile with a pronounced -2 334

dependence of the scattering intensity against the wavevector q in the log-log scale This 335

indicates the massive presence of a flat interface and it is expected typical of bilayers in 336

vesicular morphologies56 However the typical signature of nanotubes cannot be observed in 337

the SANS profile and this can be explained by both an argument related to the relative lower 338

amount of nanotubes with respect to vesicles at 90degC but also to the steeper slope of flat 339

morphologies (-2) compared to tubes (-1) the signal of the former masking the tube signal at 340

low wavevector values The SANS profile is characterized by a diffraction peak at q= 0167 Aring-341

1 of which the assignment is refined by WAXS data 342

Temperature-driven WAXS experiments in Figure S 6bc (5 mgmL-1 and 25 mgmL-1 343

respectively) show a lamellar order possibly attributed to a multilamellar structure of the 344

vesicles (dv Figure S 6d) at high temperature and to the interplanar arrangement of the 345

crystalline flat structures (dL Figure S 6d) at T= 25degC The sharp (001) reflection indicates 346

typical interlamellar distance of 353 Aring at 25degC although this value varies with temperature 347

(400 Aring at 111degC 372 Aring at 88degC) thus confirming its assignment to a hydration-dependent 348

interlamellar order More intriguing is the second broader peak at q= 0403 Aring-1 (d= 156 Aring) 349

and its reflection both clearly visible at 88degC below Tm2 We attribute it to the inter-glucolipid 350

arrangement inside the tubular phase (dt Figure S 6d) coexisting with the vesicle phase 351

between Tm1 and Tm2 and to the planar lipid arrangement (di Figure S 6d) in the rigid flat 352

lamellar structure below Tm2 Its temperature- hence hydration- independent position also 353

confirms its structural origin Finally besides the water-structure peak at about 2 Aring-1 we also 354

observe a broad peak at 144 Aring-1 characteristics of the lateral packing (d Figure S 6d) of lipids 355

in bilayers While its position could be consistent with an Lα phase5758 above Tm2 its broadness 356

and positional invariance as well as the coexistence between vesicles and tubes do not allow a 357

precise assignment 358

15

Does G-C181-OH assume a bilayer or interdigitated structure the latter being the one 359

found for other similar bolaform glucolipids4243 The size of G-C181-OH can be estimated to 360

be at the most 32 Aring given by the length L of the C16 tail (L = 22 Aring from the Tanford formula 361

L= 154 + 1265 times n (n= 16)59) and the sugar (about 10 Aring)4243 However the cis conformation 362

of the C=C bond imposes a ldquokinkrdquo of 60deg in the aliphatic chain which reduces the effective 363

length of the oleic backbone to about 17 Aring for an effective total G-C181-OH length of about 364

27 Aring Cryo-TEM provides a tube thickness varying from 35 Aring to 60 Aring and a vesicle membrane 365

thickness of at least 60 Aring WAXS suggests an interlamellar thickness (membrane and water) 366

between 40 Aring and 37 Aring and an inter-lipid distance in the tube of about 16 Aring In the absence of 367

a clear form factor oscillation profile as found for similar glucolipids4243 one must 368

acknowledge that these data pledge neither in favor of fully interdigitated nor a double layer 369

structure and might suggest a tilted lipid arrangement with partial interdigitation60 370

371

13C ssNMR demonstrates that G-C181-OH can assume two coexisting head-head and 372

head-tail configurations 373

Vesicles are characterized by low positive mean and gaussian curvatures while 374

nanotubes are described by a combination of a high positive mean and zero gaussian curvatures 375

In this regard the coexistence of vesicles and nanotubes requires glucolipids to adapt 376

themselves to both environments at the same time For this reason the formation of nanotube-377

vesicle networks generally requires a specific driving force able to impose a high mean and 378

zero gaussian curvature to the membrane In many systems insertion of at least a second 379

constituent (lipid surfactant protein) in the membrane becomes a necessary internal driving 380

force for the nucleation and growth of so-called membrane inclusions from which 381

topologically-connected nanotubes can form10ndash123261 as explained by orientational ordering 382

arguments62 It has also been shown that external driving forces such as use of laser 383

ldquotweezersrdquo30 osmotic pressure6 electroformation5ndash82829 or controlled suction13ndash16 are 384

otherwise necessary as also supported by theoretical studies19 385

The bending energy of tubes is higher than that of a vesicle membrane8 due to the 386

dependence of the membrane free energy from the principal curvatures For this reason 387

isotropic (rotational symmetry along the main axis) but also anisotropic33 lipids generally 388

stabilize low-curvature shapes like flat membranes or vesicles Lipid nanotubes are also a 389

common phase formed by amphiphiles but generally for lipids with a certain complexity in 390

their molecular structures as exampled by diacetylenic derivatives of phospholipids among 391

others262763ndash65 Negative contributions to the bending energy favoring the simultaneous 392

16

presence of interconnected vesicles and tubes was shown to be driven by the presence of 393

anisotropic lipids in a membrane continuum Spontaneous deviations to the elastic theory in the 394

presence of surfactants lipids and proteins with anisotropic structures favor the formation of 395

protrusions with radius of the order of the fraction of a micron9ndash1219ndash2362 396

In the absence of both internal and external driving forces mentioned above G-C181-397

OH should behave as a standard lipid Its DSC profile (Figure 2)536667 under both hydrated and 398

dry conditions identifies a hydration-independent transition (Tm1) slightly above room 399

temperature this is followed by one main transition at higher temperature (Tm2) in the dry 400

sample50 For such a profile one expects a gel phase (Lβ) below Tm1 and a liquid crystalline 401

structure possibly in the morphology of vesicles (L4 phase) possibly above Tm1 and most likely 402

above Tm231 as experimentally found On the basis of the thermogram recorded on the dry 403

powder one may also expect a ldquoripplerdquo (Pβ) phase between Tm1 and Tm2 characterized by 404

periodic undulations of the bilayer (generally addressed to as regions with both liquid order and 405

disorder)5455 406

In the present work instead of a Lβ phase we observe flat crystals below Tm1 and a 407

vesicle-nanotube network instead of the ldquoripplerdquo phase The latter is particularly unexpected 408

because the membrane is composed of a single-component (G-C181-OH) and no anisotropic 409

external force (osmotic hydrodynamic electric) is used to pull and stabilize the tubes In 410

addition the nanotubes are observed at fairly high temperatures above Tm1 (between about 411

60degC and 90degC) while this morphology is generally favored at low temperature (proportionality 412

with 1kT figure 5 in Ref 62) commonly below a given transition temperature characteristics 413

for each specific amphiphile27 Furthermore nanotubes of G-C181-OH seem to be stable in 414

time thus differing from what is found in electroformed vesicles78 They are also observed 415

away from phase transition events (may them be at Tm1 or Tm2) as otherwise reported before 416

for a diacetylenic derivative of phosphatidylcholine26 and many other systems27 417

A stable nanotube-vesicle network for a single-component phospholipid in the absence 418

of external driving forces is not expected and has never been reported before It could be 419

explained by the coexistence of two micro-separated phases of different rigidity This 420

hypothesis is excluded by confocal microscopy on Laurdan-labelled membranes and magic 421

angle spinning (MAS) solid-state NMR (ssNMR) experiments 422

Laurdan is a well-known fluorophore used to put in evidence the coexistence of rigid 423

and fluid domains in both model and biological membranes47ndash49 Laurdan was used here to label 424

the G-C181-OH membranes between Tm1 and Tm2 and to verify the possible micro-segregation 425

and coexistence of rigid (eg Lβ λem= 440 nm) and fluid (eg Lα λem= 490 nm) domains in 426

17

tubes and vesicles respectively A set of quantitative experiments performed on a 5 mgmL-1 427

solution summarized in Figure 5 shows an excess of fluorescence at only λem= 490 nm for 428

temperature between 90degC and 70degC (Figure 5h) The GP value which quantifies the rigidfluid 429

character of the membrane is systematically negative (Figure 5i) indicating an overall fluid 430

membrane at all temperatures GP is quantified on more than 50 regions of interest (ROIs) at 431

each temperature including vesicles and tubes alone as well as connected nanotube-vesicles 432

(Figure 5d-g) The dispersion of GP at 80degC and 70degC between -08 and -02 could suggest the 433

spurious presence of membranes with various degrees of flexibility although fluidity is always 434

predominant The combination of WAXS (broad peak d= 144 Aring-1 invariant with 435

temperature Figure S 6bc) and confocal fluorescence microscopy using Laurdan confirms the 436

single-phase of the membrane composed of G-C181-OH both in vesicles and nanotubes This 437

is also consistent with the lack of phase transitions in DSC experiments between ~85degC and 438

~60degC especially in hydrated samples We conclude that the vesicle-nanotube transition is not 439

driven by the local segregation of gel phase (Lβ) microdomains nor that the nanotubes are 440

composed of such phase 441

ssNMR is the best-suited approach to investigate the origin of the local driving force for 442

the nucleation of anisotropic inclusions of which the average orientation depends on the local 443

membrane and inclusion curvatures and was shown to promote tubing62 ssNMR can provide 444

information on the local molecular mobility eg through the nuclear spin relaxation behavior 445

but also on intermolecular interactions and stereochemistry eg through variations in the 446

chemical shift68ndash70 447

Ideally ssNMR should be performed on diluted G-C181-OH samples above Tm1 448

However fast MAS necessary for high resolution centrifuges the sample As a first 449

approximation hydrated G-C181-OH samples for instance at 50 wt as classically done for 450

phospholipids71 could be studied Unfortunately permeability of the sample holder combined 451

to heating (close to the boiling point of water) and MAS provoke dehydration and even sample 452

ejection After experiencing both of the above this approach was abandoned As a second 453

approximation the study of the dehydrated G-C181-OH powder avoids centrifugation and 454

evaporation If studying dried lipids may not reflect their exact behaviour under diluted 455

conditions valuable structural information can still be obtained in dehydrated molecular 456

systems like proteins thus illustrating that this approach is not unreasonable Furthermore if 457

DSC (Figure 2a) shows that hydration suppresses the Tm2 it also shows that the value of Tm1 is 458

not affected by water at concentrations as low as 10 wt For this reason performing a 459

structural analysis of dehydrated G-C181-OH by ssNMR above Tm1 may help understanding 460

18

the local arrangement of G-C181-OH in two morphologies vesicles and nanotubes having 461

sensibly different curvatures 462

To explore the local mobility of the most relevant chemical groups of G-C181-OH 463

above Tm2 (T= 110degC) and between Tm2 and Tm1 (70 lt T degC lt 90) we combine a series of 1H 464

and 13C experiments Single-pulse (SP) MAS experiments (1H and 13C) when recorded under 465

quantitative conditions allow the detection of all chemical species while cross-polarization 466

(CP) MAS (13C only) experiments are sensitive only to constrained groups for which the 1H-467

13C dipolar coupling is not averaged to zero by the molecular motions 468

469

Figure 6 ndash a) 1H MAS (10 kHz) ssNMR spectra recorded from 110degC to 70degC b) 13C MAS (10 kHz) ssNMR 470

spectra recorded using high power decoupling single pulse (SP) and cross-polarization (CP) pulse sequences 471

Contact time for CP experiments and corresponding temperatures are given on each spectrum Full spectra 472

are given in Figure S 7 473

The most relevant 1H and 13C spectra are given in Figure 6 (full range in Figure 7 and 474

Figure S 7) As expected the 1H solid-state SPMAS spectra show broader peaks compared to 475

the solution-state NMR spectrum (Figure 6a) reflecting the obvious solid-state character of the 476

sample at any temperature The sugar group and the C=C bond resonate in the 25-6 ppm range 477

and 535 ppm respectively while the OH groups of the sugar moiety are found below 45 ppm 478

(Figure 6a full assignment in Table S 1) Very broad (~1 ppm) at 100degC the position of the 479

OH peak varies with the temperature value and becomes narrower (lt~05 ppm) and visible at 480

δH= 464 ppm at T= 70degC Line narrowing and chemical shift variations of the OH resonance 481

in ssNMR is always a direct sign of H-bonding interactions whereas the strength of the H-482

bonding hence its OHhellipH distance is correlated to a dispersion in chemical shifts72ndash75 All other 483

60 55 50 45 40 35 30 25

ppm

110

90

80

70

910

1rsquo

6rsquo16

1

OH

134 132 130 128 126

ppm110 108 106 104 102 100

ppm

SP (T= 110degC)

SP (T= 70degC)

CP 50 ms (T= 70degC)

CP 3 ms (T= 70degC)

910 1rsquo

1H 13C

1

811

910

CH2 CH2

1rsquo

6rsquo

a) b)

C102

13024

C9101

13003

C92

12869 C1rsquo210382

C1rsquo110443

C1rsquo13035

2rsquo3rsquo4rsquo

T degC

5rsquo

19

resonances broaden from 110degC to 70degC thus indicating slower andor more anisotropic 484

reorientational motions of all chemical groups The specific dynamic behaviour of the sugar 485

headgroup is also reflected by its 1H transverse relaxation time T2 whereas the shorter the T2 486

the higher the rigidity Experiments performed between 110degC and 70degC shown and 487

commented in Figure S 8 in the Supporting Information globally show a single short (lt 1 ms 488

below 100degC) T2 relaxation component for H1rsquo H5rsquo and H6rsquo and two both short and long 489

(between 4 ms and 12 ms) T2 for the acyl chain (CH2 H910 H811) even below 100degC 490

At 110degC 13C SPMAS experiments (Figure 6b full range in Figure 7 and Figure S 7 491

assignments in Table 1) show a single peak at δC= 10335 ppm for C1rsquo and at δC= 13003 ppm 492

for C910 as expected for a symmetrical conformation with two chemically-equivalent 493

carbons76 At 70degC the C1rsquo peak disappears while C910 now shows two peaks around 128 494

ppm and 130 ppm (Table 1 full temperature series in Figure S 7b) attributed to C92 (δC= 12869 495

ppm) and C10 in a non-symmetric configuration respectively for which these two carbons are 496

chemically inequivalent76 More precisely the resonance at 130 ppm is actually the 497

superposition of two close resonances C9101 (δC= 13003 ppm) and C102 (δC= 13024 ppm) 498

where subscripts 1 and 2 stand for two different conformations of the cis-olefin group as 499

reported for various phases of oleic acid in the vicinity of 0degC76 500

501

Table 1 ndash 13C Chemical shifts of G-C181-OH in solution and solid-state Solid-state experiments are 502

recorded at MAS= 10 kHz All spectra are reported in Figure 7 and Figure S 7 The contact time for cross 503

polarization is 3 ms In all ssNMR experiments temperature is first raised at 110degC and then lowered to 504

either 90degC or 70degC Attribution of the 13C resonances in Table 1 is performed on the basis of the solution 505

spectrum (Figure S 1b) Additional peaks observed in both single pulse and cross polarization ssNMR 506

experiments are attributed when possible to a configuration effect of the given atom eg C9101 and C92 507

indicate two configurations of C9 All new peaks observed in solid-state and not detected in solution are 508

highlighted in red for convenience 509

510

511

δ ppm

Peak Liquid (DMSO-d6)

Solid Single Pulse

T= 110degC

Solid Single Pulse

T= 90degC

Solid Single Pulse

T= 70degC

Solid Cross Polarization

T= 70degC

C102

- - 13024 13024

C910C9101 13009 13003 13003 13003 13003

C1C9C10C1rsquo

C6rsquo

20

C92 - - - 12869 12869

C1rsquoC11 10329 10335 10335 - 10443

C12 - - - - 10382

C3 7724 7709 7702 - 7829

- 7634 7629 - 7621

7525

C2 7388 7398 7390 7370 7370

C4 7052 7077 7054 7041 7098

C16 6900 7016 - - 7029

- - - - 6793

C1C5C6 6153 6241 6236 6239 6262

C1C5C6 6117 - - - 5956

C2 3300 3290 3290 3283 3279

CH2(trans) - - - 3198 3196

CH2(gauche)C15 2941 2981 2987 3004 -

C811 2706 2747 2747 2832 2817

CH2 2597 2625 2628 2630 2817

512

13C CPMAS spectra at 70degC (Figure 6b full range in Figure 7 and Figure S 7d) at 513

increasing contact times confirm the presence of two cis-olefin environments one for which 514

C910 are equivalent labelled here C9101 and one for which they are inequivalent labelled 515

C92 C102 Both environments are rigid enough to be observed at short contact time (tc= 50 μs) 516

A similar double environment is also observed for the sugar C1rsquo CPMAS at 70degC shows two 517

peaks (C1rsquo1 at δC= 10443 ppm C1rsquo2 at δC= 10382 ppm Figure 6b Figure S 7d) These are 518

significantly shifted downfield from the C1rsquo peak observed at 110degC and are detected at contact 519

times as short as 50 μs thus indicating their rigid character One also finds that CPMAS is 520

totally inefficient at 100degC for short contact times (50 μs) and poorly efficient even at 3 ms 521

typical of fast-tumbling molecules (Figure S 7c) also both gauche (δC= 3004 ppm) and all-522

trans (δC= 3198 ppm) conformations of the CH2 carbons coexist at 70degC (Figure S 7b)77 523

According to ssNMR one main mobile molecular environment exists at 110degC as 524

shown by the well-resolved 1H spectrum single C=C (C9101) and anomeric (C1rsquo) resonances 525

gauche conformation of the aliphatic CH2 and single T2 of the aliphatic CH2 chain and sugar 526

protons The sugar group is certainly mobile (inefficient 1H13C CPMAS) but still more 527

constrainted than the acyl chain (shorter T2 (1H)) As the temperature is lowered the overall 528

rigidity increases (broadening of 1H NMR peaks faster 1H transverse relaxation and efficient 529

1H13C magnetization transfer during CPMAS) presumably driven by H-bonding 530

interactions (downfield shift and line narrowing of the OH 1H signal very short OH T2 loss of 531

the C1rsquo signal in SPMAS) At the same time a number of spectroscopic data (two concomitant 532

21

resonances for anomeric C1rsquo1 and C1rsquo2 cis-olefin C9101 and C92 C102 and CH2 gauche and 533

all-trans environments) and relaxation behaviour (fast and slow T2 relaxation components) 534

strongly suggests the coexistence of two molecular environments Further proofs of the 535

coexistence of two distinct molecular domains are given by a number of chemical shift 536

variations in the 13C CPMAS spectrum at 70degC compared to the one at 110degC 90degC and 537

solution This is shown in Figure 7 and in the list of 13C chemical shifts in Table 1 538

539

540

Figure 7 - Comparison between solution- and solid-state 13C NMR spectra of G-C181-OH Solid-state 541

experiments are recorded directly on the sample powder at a MAS of 10 kHz (4 mm zirconia rotor) SP 542

Single Pulse CP Cross Polarization In all the ssNMR experiments temperature is first raised at 110degC 543

and then lowered to either 90degC or 70degC Segmented lines are meant to highlight the known peak position 544

in solution thus stressing the strong chemical shift variations in the solid-state 545

546

NMR experiments above Tm2 suggest that vesicles could contain G-C181-OH with a 547

single family of mobile conformations characterized by chemically-equivalent C=C (red 548

Figure 8b) and a single C1rsquo conformation (green Figure 8b) Experiments between Tm1 and Tm2 549

otherwise suggest that two coexisting conformations characterized by chemically-equivalent 550

(red Figure 8b) and inequivalent (magenta Figure 8b) C=C could be associated to nanotubes 551

Two distinct C1rsquo sites (grey Figure 8b) also indicate two configurations of the sugar headgroup 552

A comment on this point will be given in the following paragraphs 553

The local (Aring-scale) variation of molecular mobility below Tm2 drives the coexistence of 554

two distinct molecular environments of the same G-C181-OH molecule This occurs within a 555

single-phase domain (as deduced by WAXS) of equivalent membrane rigidity (as derived from 556

Laurdan probe in confocal microscopy) Although never reported before the difference in the 557

local mobility of the same amphiphile (here G-C181-OH) could satisfy the theoretically-558

140 120 100 80 60 40 20

CP(3 ms)-70degC

SP-70degC

SP-90degC

SP-110degC

ppm

LiquidC910

C1rsquo

C2rsquo

C4rsquo

C3rsquo

C1

C6rsquoC5rsquo

C1

6

C8

11

C2 C

H2

C1

5

CH

2

DM

SO

13C

22

required conditions that induce local lipid segregation and formation of membrane inclusions 559

(Figure 8c) In case of their non-zero average orientation the local membrane curvature 560

increases62 and nanotubes can spontaneously form (Figure 8bc) More intriguing this 561

mechanism occurs at least 10degC above Tm1 although it should not be favored above the Tm27 562

From both experimental and theoretical backgrounds the conditions of topologically-563

connected vesicles-nanotubes systems are met only when either internal (two components) or 564

external driving forces are applied19 In the present single-lipid system of which the phase 565

behavior is summarized in Figure 8a the most plausible driving force could be explained by 566

the NMR results collected on the dry sample G-C181-OH concomitantly occupies two 567

molecular environments characterized by 1) two distinct conformations of the sugar 568

displaying equivalent mobility and 2) two conformations of the acyl chain unsaturation 569

570

23

571

Figure 8 ndash a) Thermotropic phase behavior of G-C181-OH at 5 mgmL-1 Note that Tm2 is observed on the 572

dry powder only b) Possible relationship between the molecular environment as derived from NMR and 573

membrane curvatures All the molecular drawings refer to the same G-C181-OH molecule (bulky ellipsoid 574

equiv sugar headgroup) while each color corresponds to a given configuration Grey of C1rsquo1 and C1rsquo2 indicates 575

the uncertainty about the assignment of the specific NMR peak The curvature C is defined as 1R R being 576

the curvature radius R1 R2 radius of curvature CG Gauss curvature CM Mean curvature c) Free 577

interpretation of the anisotropic high-curvature (CMgt 0 CG= 0) inclusions in the membrane (low CM and 578

CG) continuum driven by inter-sugar H-bonding and reduced intramembrane diffusion for a G-C181-OH 579

glucolipid 580

581

Flat crystalline structure

Nanotube-vesicle network

Vesicles

48degC

120deg

106 105 104 103 102 101

ppm

H9 H10H1rsquo

106 105 104 103 102 101

ppm106 105 104 103 102 101

ppm106 105 104 103 102 101

ppm

C=CC=C

C1rsquo1C1rsquo2

C1rsquo

Tm2(96degC)Tm1

a)

b)

c)

Same molecule different local orientations and sugar conformations

High curvature membrane

Low curvature inclusion

Flip-FlopStrong H-bonding

CG

CM ( + )

CG 0

CM 0 CG 0

CM 0

G-C181-OH

24

Two related questions are unanswered at the moment What actually drives nanotube 582

formation and is the organization of G-C181-OH within the vesicles and tubes the same Lipid 583

nanotubes have been studied for at least three decades over which broad experimental and 584

theoretical insight has been generated Despite this fact the actual understanding is still matter 585

of debate27 Micelle-to-nanotube and vesicle-to-nanotube phase transitions are generally 586

reported for single amphiphiles below a given Tm and intermediate phase composed of twisted 587

and flat fibers is concomitantly reported Initially thought to be a chirality-driven phenomenon 588

recent studies propose a more general mechanism involving crystallization and symmetry 589

breaking in the lipid packing driven by optimal packing27 If twisted ribbons were observed 590

before for a number of microbial glycolipid bolaamphiphiles44 they are not found for G-C181-591

OH This fact as well as the large body of cryo-TEM images and the observation of nanotubes 592

above Tm1 rather pledge in favour of a direct topological change from positive (vesicles) to 593

zero (nanotubes) Gaussian curvatures In this case vesicle-to-nanotube transitions could be 594

driven by external forces56307813ndash162829 595

Considering the absence of external forces in the present system there are only two 596

possibilities left spontaneous budding of the membrane or nanotube pulling after inter-vesicle 597

collisions due to Brownian motion Both are possible as inclusions composed of a different 598

arrangement of the same molecule in the membrane continuum can theoretically explain 599

spontaneous nanotube formation in an isotropic medium672223629ndash1218ndash21 However the second 600

one is probably more likely to prevail because it could explain the large number of nanotubes 601

connecting two distant vesicles Interestingly similar events occur for any phospholipid-based 602

vesicular solution but nanotubes are extremely rare unless at least two lipids of different 603

molecular structure are mixed together The spectroscopic 1H and 13C and relaxation (1H) 604

ssNMR data even if collected on a dry powder demonstrate that the coexistence of two G-605

C181-OH configurations involving both the sugar headgroup and fatty alcohol chain are 606

possible for this molecule As a matter of fact this is not so surprising as more than one energy 607

minima each corresponding to a given conformation are known for glucose its non-reducing 608

derivative methylglucose and carbohydrates in general78ndash80 609

The presence of two configurations could explain the coexistence between two 610

structures of different curvature but the data presented so far do not explain the origin of these 611

two configurations In other words what drives the change in the local structure of the sugar 612

headgroup and fatty alcohol within the membrane 613

The association of known self-assembled structures to spectroscopic 13C ssNMR data is 614

known for years in particular for gluconamides lipids having a linear sugar headgroup81ndash84 It 615

25

was found that gauche effects in some cases up to the γ position in the sugar chain induce 616

upfield chemical shifts while anti and trans conformations result in downfield shifts Each 617

structure (monolayer crystal bilayer crystal or micellar fiber) was then characterized by a given 618

organization of the gluconamide and a characteristic 13C ssNMR fingerprint which could be 619

explained by gauche γ-gauche anti and trans conformations of the headgroup Similar 620

structural-spectroscopic correlations have been collected over the years for much simpler cyclic 621

sugars85 whereas pyranoses can undergo important conformational changes (chair half-chair 622

boat skew) when the appropriate energy barriers are crossed78ndash8086 A quantitative 623

conformational study based on the 13C chemical shift variations shown in Table 1 would be 624

certainly helpful but very delicate in the context of the present state of the art because a 625

rigorous approach requires at least one known crystal structure8182 which is currently missing 626

for this compound A qualitative approach will then be adopted 627

G-C181-OH is a bolaform amphiphile and flip-flop could be a possible mechanism 628

involved in the nanotube stabilization whereas head-tail interactions have long been considered 629

as important in nanotube-forming bolaform amphiphiles6063 From mere considerations based 630

on T2 (1H) relaxation arguments one could exclude head-tail configurations if the R-CH2OH 631

endgroup (H1) was involved in a direct interaction with the sugar group its corresponding T2 632

would probably be comparable to the sugarrsquos one (OH and H1rsquo) but this does not seem to be 633

the case However this hypothesis would be based on the comparison of the relaxation from 634

CH and CH2 protons additionally involved in different kind of motions Besides the relaxation 635

of H1 is influenced by the strong overlap between H1 and H5rsquo For these reasons relaxation 636

arguments are not solid enough and will not be used 637

We then address this question by exploring intermolecular interactions in a direct 638

manner using 2D 1H-1H homonuclear ssNMR correlation spectroscopy At 90degC in the fast-639

tumbling domain of molecular motion (narrow 1H NMR peaks Figure 6a) and corresponding 640

to the vesicle-rich region spin diffusion (NOESY) experiments are preferred over dipolar 641

coupling based experiments because such interaction is averaged out by molecular motion 642

However when the molecular motion is more constrained in the domain of nanotube stability 643

at 70degC (broad 1H spectra Figure 6a) Double-Quantum Single-Quantum (DQ-SQ) correlation 644

experiments are best suited to probe interaction between close molecular neighbors These 645

experiments are reported in Figure 9 and the related interpretation is only briefly reported 646

below while a more detailed discussion can be found on Page 12 in the Supporting Information 647

26

648

Figure 9 ndash a) 2D 1H-1H NOESY NMR experiment performed at MAS= 10 kHz (4 mm probe) on a Bruker 649

300 MHz Avance III HD spectrometer using 300 ms of mixing time Sample temperature is regulated to 650

90degC reached by cooling from 110degC b) 2D 1H-1H DQ-SQ (BAck-to-BAck BABA pulse sequence used for 651

the DQ excitation scheme) NMR experiment performed at MAS= 22 kHz (25 mm probe) on a Bruker 700 652

MHz Avance III spectrometer using a DQ excitation time equal to 2 rotor periods Sample temperature is 653

regulated to 70degC reached by cooling from 110degC The typical schemes illustrating how to interpret these 654

2D experiments are reported above each experimental map Zooms of the sugar region and configurational 655

interpretation are given below each 2D map More information are given in the text below 656

H16-H16

H1-H1

H9-H10

H1rsquo-H1

H1rsquoH910 H16

H1

H1rsquo-H1rsquo

Hxrsquo-Hxrsquo

2

(pp

m)

(ppm)

315

811910 1rsquo

2rsquo2

3rsquo 4rsquo6rsquo

5rsquo

1rsquo2910

asymp asymp asympHDO MeOD Acetone

1

315

811 9101rsquo

2rsquo2

3rsquo4rsquo6rsquo

5rsquo

1rsquo2 9 10

asympasympasympHDOMeODAcetone

H1rsquo-H1

1rsquo

1

(p

pm

)

(ppm)

H1-H2

H1rsquo-H1

H1rsquo

H910

H1H5rsquo

H2

H1rsquo-H5rsquo

315

811910 1rsquo

2rsquo2

3rsquo 4rsquo6rsquo

5rsquo

1rsquo2910

asymp asymp asympHDO MeOD Acetone

1

H2rsquoH3rsquoH4rsquo

H2rsquoH3rsquoH4rsquo

X= 2rsquo3rsquo4rsquo

NOESY - T= 90degC

1H

1H

1H

1H

DQ-SQ - T= 70degC

a) b)

2

(pp

m)

(p

pm

)

(ppm)

A B

AhellipB

AhellipA

BhellipB

NOESY

(ppm)

A B

AhellipB

DQ-SQ

Single-quantum dimension

Do

ub

le-q

ua

ntu

m d

imen

sio

n

AhellipA

BhellipB

BhellipA

BhellipA

315

811910 1rsquo

2rsquo2

3rsquo 4rsquo6rsquo

5rsquo

1rsquo2910

asymp asymp asympHDO MeOD Acetone

1

lt 5 Aring

lt 5 Aring

H1-H2

H811

H15

H16

H6rsquo

H1rsquo

H1

H5rsquo

H2

H1-H2

H1rsquo-H1

H1rsquo-H5rsquo

H1rsquo

H1 H5rsquo

H1rsquo

H1

H5rsquo

H1H2rsquo H3rsquo H4rsquo

H1-H1

H1rsquo-H1

Hxrsquo-HxrsquoX= 2rsquo3rsquo4rsquo

H1-H2

H1rsquo

Zoom sugar region 28 ndash 45 ppmNOESY DQ-SQ

27

Head-tail or head-head configurations could be probed by the existence of H1rsquo-H1 657

interactions whereas H1 (δH= 356 ppm) is associated to the glucolipid tail and H1rsquo (δH= 432 658

ppm) to the sugar headgroup These atoms are more than 15 Aring apart and a direct interaction 659

could not be observed in both NOESY and DQ-SQ experiments NOESY experiments (Figure 660

9a) show that H1rsquo-H1 correlations are not observed Meanwhile clear-cut direct H1rsquo-H1 661

interactions are indeed observed are detected at T= 70degC (DQ cross-peaks related to H1rsquo-H1 662

correlation in DQ-SQ experiment shown in Figure 9b) Interestingly the absence of auto peaks 663

at δH= 432 ppm and δH= 335 ppm in the SQ dimension excludes H1rsquo-H1rsquo and Hxrsquo-Hxrsquo (x= 2 664

3 4) interactions respectively strengthening the idea that sugar headgroups from two nearest-665

neighbor G-C181-OH molecules are not adjacent in the nanotube-vesicle region 666

Homonuclear 2D ssNMR supports the idea according to which G-C181-OH can 667

undergo molecular flip-flops and a head-tail arrangement in the dry product Extrapolating these 668

conclusions in the lower concentration range lack of H1rsquo-H1 proximity at T= 90degC suggests a 669

head-head configuration in the vesicle-rich region and head-tail configuration within the 670

nanotubes It is however still unclear whether flip-flops are driven by spontaneous temperature-671

induced conformational changes in the glucolipid packing stimulating inclusions within the 672

membrane itself672223629ndash1218ndash21 or by defects generated by inter-vesicle collisions In fact a 673

combination of both is not unlikely Nanotubes are not stable when temperature is 20 to 30 674

degrees above Tm1 indicating that inter-vesicle collisions alone cannot explain the unusual 675

phenomenon In an energy landscape approach87 the head-head configuration is certainly 676

favored at high temperature while both head-head and head-tail configurations could be 677

favored in the mid-temperature region corresponding to the range between Tm1 and Tm2 in the 678

dry powder Both statistical and defect-induced flip-flop events contribute to form high mean 679

zero Gaussian curvature morphologies (nanotubes) for which the head-tail conformation 680

limiting the proximity between bulky sugar headgroups is probably more adapted (Figure 8c) 681

The flip-flop interpretation now reasonably explains the strong differences in the sugar and fatty 682

alcohol conformations deduced by the multiple NMR signals attributed to chemically-683

equivalent groups The existence of various low-energy conformations each having specific 684

NMR signature for sugars in solution78ndash8086 and solid-state81ndash84 further supports our 685

interpretation 686

One last point should be clarified Why between Tm1 and Tm2 are two molecular 687

environments able to coexist instead of a single one of average rigidity as found in the case of 688

ldquoripplerdquo phases50ndash54 We believe that this question can be answered based on the NMR results 689

the relatively short T2 values of the H1rsquo and OH protons shows the constrained mobility of the 690

28

sugar headgroup despite a certain flexibility of the chain (long T2 component) This trend may 691

suggest restrictions in translational mobility probably due to the sugar-OH hydrogen bonding 692

(Figure 8c) meaning that lipids are not able to diffuse within the membrane thus keeping two 693

different configurations stable 694

695

Conclusions 696

Topologically-connected nanotube-vesicle lipidic systems are commonly observed for 697

self-assembled membranes composed of at least two amphiphiles and spontaneously driven by 698

anisotropic inclusions andor by external directional forces This work shows that strong 699

differences in membrane curvature can spontaneously occur in single-lipid systems without 700

external anisotropic forces Coexistence and interconnection between nanotubes and vesicles 701

are evidenced via cryo-TEM and fluorescence confocal microscopy in a single-phase region at 702

temperature above a transition temperature Tm1= 483degC determined by DSC on the hydrated 703

sample but below the second transition temperature Tm2 (957degC) determined by DSC on the 704

dry powder X-ray diffraction (WAXS) also shows the coexistence of two lattice periods one 705

attributed to an inter-membrane repeating distance and the other possibly related to the intra-706

nanotube glucolipid arrangement Labelling the systems with Laurdan a fluorophore of which 707

the emission strongly varies with the local membrane rigidity excludes the presence of rigid 708

micro-separated phases (eg Lβ) 709

1H and 13C ssNMR despite the fact that it can only be performed on the dry powder 710

confirm the presence of one main fluid G-C181-OH environment above Tm2 Below Tm2 the 711

splitting of some 13C peaks and the coexistence of short and long 1H relaxation components 712

demonstrate that G-C181-OH can simultaneously pack into structurally and dynamically 713

inequivalent environments Combining advanced 2D 1H-1H correlation experiments based on 714

spin diffusion or double-quanta excitation suggests that vesicles could contain a preferred head-715

head configuration of the bolaform lipid while nanotubes a preferred head-tail assembly 716

The presence of two environments composed of the same molecule characterized by 717

different molecular configurations and rigidity fulfills the theoretical conditions necessary to 718

spontaneously nucleate a membrane inclusion of non-zero average orientation described as 719

necessary to observe topologically-connected nanotube-vesicle structures in the absence of 720

external forces This hypothesis could also be enriched by the possibility of nanotubes to be 721

pulled away between two vesicles after a collision event 722

This work expands the actual theoretical and experimental framework on lipid networks 723

stabilized by a different radius of curvature It could also motivate the synthesis or isolation 724

29

of new lipids providing a higher level of 3D architectural control in soft networks possibly in 725

the range of room temperature 726

727

Acknowledgements 728

Dr Andrea Lassenberger and Dr Sylvain Preacutevost at Institut Laue Langevin (ILL Grenoble 729

France) are kindly acknowledged for their assistance on the SANS experiment Confocal 730

images were performed at the Institut de Biologie Paris Seine (IBPS) imaging facility Jean-731

Franccedilois Gilles (Imaging Core Facility Institut de Biologie Paris Seine (IBPS) CNRS 732

Sorbonne Universiteacute Paris France) is kindly acknowledged for his assistance on the confocal 733

microscopy imaging Dr Guillaume Laurent (LCMCP Sorbonne Universiteacute Paris France) is 734

kindly acknowledged for helpful discussions The French region Ile-de-France SESAME 735

program is acknowledged for financial support (700 MHz NMR 736

spectrometer) Dr Olivier Diat (ICSM Marcoule France) Prof eronikaIglič and Prof V Aleš737

onsul discussi Slovenia) are kindly acknowledged for helpfUniversity of LjubljanaIglič (-Kralj 738

739

Financial support 740

WAXS experiments have been supported by Soleil Light Source Saint Aubin France proposal 741

Ndeg 20201747 SANS experiments have been supported by ILL proposal Ndeg9-13-778 742

743

References 744

1 D Kabaso N Bobrovska W Goacuteźdź E Gongadze V Kralj-Iglič R Zorec and A 745

Iglič Bioelectrochemistry 2012 87 204ndash210 746

2 C R Safinya U Raviv D J Needleman A Zidovska M C Choi M A Ojeda-747

Lopez K K Ewert Y Li H P Miller J Quispe B Carragher C S Potter M W 748

Kim S C Feinstein and L Wilson Adv Mater 2011 23 2260ndash2270 749

3 P Veranič M Lokar G J Schuumltz J Weghuber S Wieser H Haumlgerstrand V Kralj-750

Iglič and A Iglič Biophys J 2008 95 4416ndash4425 751

4 A Rustom R Saffrich I Markovic P Walther and H-H Gerdes Science (80- ) 752

2004 303 1007ndash1010 753

5 M I Angelova S Soleacuteau P Meacuteleacuteard F Faucon and P Bothorel Prog Colloid 754

Polym Sci 1992 89 127ndash131 755

6 L Mathivet S Cribier and P F Devaux Biophys J 1996 70 1112ndash1121 756

7 V Kralj-Iglič G Gomišček J Majhenc V Arrigler and S Svetina Colloids Surfaces 757

A Physicochem Eng Asp 2001 181 315ndash318 758

30

8 V Kralj-Iglič A Iglič G Gomišček F Sevšek V Arrigler and H Haumlgerstrand J 759

Phys A Math Gen 2002 35 1533ndash1549 760

9 M Drab D Stopar V Kralj-Iglič and A Iglič Cells 2019 8 626 761

10 A Zidovska K K Ewert J Quispe B Carragher C S Potter and C R Safinya 762

Langmuir 2009 25 2979ndash2985 763

11 A Zidovska K K Ewert J Quispe B Carragher C S Potter and C R Safinya Soft 764

Matter 2011 7 8363ndash8369 765

12 A Zidovska K K Ewert J Quispe B Carragher C S Potter and C R Safinya 766

Biochim Biophys Acta - Biomembr 2009 1788 1869ndash1876 767

13 M Karlsson K Sott M Davidson A S Cans P Linderholm D Chiu and O Orwar 768

Proc Natl Acad Sci U S A 2002 99 11573ndash11578 769

14 A Karlsson R Karlsson M Karlsson A-S Cans A Stromberg F Ryttsen and O 770

Orwar Nature 2001 409 150ndash152 771

15 M Karlsson K Sott A S Cans A Karlsson R Karlsson and O Orwar Langmuir 772

2001 17 6754ndash6758 773

16 E Evans H Bowman A Leung D Needham and D Tirrell Science (80- ) 1996 774

273 933ndash935 775

17 U Seifert Configurations of fluid membranes and vesicles 1997 vol 46 776

18 J Kaumls and E Sackmann Biophys J 1991 60 825ndash844 777

19 N Bobrovska W Goacuteźdź V Kralj-Iglič and A Iglič PLoS One 2013 8 e73941 778

20 A Iglič B Babnik K Bohinc M Fošnarič H Haumlgerstrand and V Kralj-Iglič J 779

Biomech 2007 40 579ndash585 780

21 K Bohinc T Slivnik A Iglič M Brumen and V Kralj-Iglič J Phys Chem C 2007 781

111 9709ndash9718 782

22 A Iglič H Haumlgerstrand P Veranič A Plemenitaš and V Kralj-Iglič J Theor Biol 783

2006 240 368ndash373 784

23 V Kralj-Iglič and P Veranič Adv Planar Lipid Bilayers Liposomes 2006 5 129ndash785

149 786

24 A Iglic and V Kralj-Iglic Adv Planar Lipid Bilayers Liposomes 2006 4 253ndash279 787

25 S Penič L Mesarec M Fošnarič L Mroacutewczyńska H Haumlgerstrand V Kralj-Iglič and 788

A Iglič Front Phys 2020 8 342 789

26 P Yager and P E Schoen Mol Cryst Liq Cryst 1984 106 371ndash381 790

27 T G Barclay K Constantopoulos and J Matisons Chem Rev 2014 114 10217ndash791

10291 792

31

28 B Božič G Gomišček V Kralj-Iglič S Svetina and B Žekš Eur Biophys J 2002 793

31 487ndash496 794

29 A Iglič H Haumlgerstrand M Bobrowska-Haumlgerstrand V Arrigler and V Kralj-Iglič 795

Phys Lett A 2003 310 493ndash497 796

30 R Bar-Ziv and E Moses Phys Rev Lett 1994 73 1392ndash1395 797

31 S Chiruvolu H E Warriner E Naranjo S H Idziak J O Raumldler R J Plano J A 798

Zasadzinski and C R Safinya Science (80- ) 1994 266 1222ndash1225 799

32 V Kralj-Iglič A Iglič H Haumlgerstrand and P Peterlin Phys Rev E 2000 61 4230ndash800

4234 801

33 L Mesarec M Drab S Penič V Kralj-Iglič and A Iglič Int J Mol Sci 2021 22 802

2348 803

34 L Van Renterghem S L K W Roelants N Baccile K Uyttersprot M C Taelman 804

B Everaert S Mincke S Ledegen S Debrouwer K Scholtens C V Stevens W 805

Soetaert L Van Renterghem S L K W Roelants N Baccile K Uytersprot M C 806

Taelman B Everaert S Mincke S Ledegen S Debrouwer K Scholtens C V 807

Stevens and W Soetaert Biotechnol Bioeng 2018 115 1195ndash1206 808

35 J Clark and F Deswarte Introduction to Chemicals from Biomass Second Edition 809

John Wiley amp Sons 2015 810

36 K V Sajna R Houmlfer R K Sukumaran L D Gottumukkala and A Pandey in 811

Industrial Biorefineries and White Biotechnology eds A Pandey R Houmlfer M 812

Taherzadeh K M Nampoothiri and C Larroche Amsterdam Oxford Waltham 813

Elsevier 2015 pp 499ndash521 814

37 R Marchant and I M Banat Trends Biotechnol 2012 30 558ndash565 815

38 K Hill and C LeHen-Ferrenbach in Sugar-Based Surfactants Fundamentals and 816

Applications ed C C Ruiz Boca Raton CRC Press 2009 pp 1ndash20 817

39 L Rodrigues I M Banat J Teixeira and R Oliveira J Antimicrob Chemother 818

2006 57 609ndash618 819

40 J D Desai and I M Banat Microbiol Mol Biol Rev 1997 61 47ndash64 820

41 S L K W Roelants K Ciesielska S L De Maeseneire H Moens B Everaert S 821

Verweire Q Denon B Vanlerberghe I N A Van Bogaert P Van der Meeren B 822

Devreese and W Soetaert Biotechnol Bioeng 2016 113 550ndash559 823

42 N Baccile M Selmane P Le Griel S Preacutevost J Perez C V Stevens E Delbeke S 824

Zibek M Guenther W Soetaert I N A Van Bogaert and S Roelants Langmuir 825

2016 32 6343ndash6359 826

32

43 N Baccile A-S Cuvier S Preacutevost C V Stevens E Delbeke J Berton W Soetaert 827

I N A Van Bogaert and S Roelants Langmuir 2016 32 10881ndash10894 828

44 N Baccile C Seyrig A Poirier S A Castro S L K W Roelants and S Abel Green 829

Chem 2021 23 3842ndash3944 830

45 A Salis L Cappai C Carucci D F Parsons and M Monduzzi J Phys Chem Lett 831

2020 11 6805ndash6811 832

46 J Schindelin I Arganda-Carreras E Frise V Kaynig M Longair T Pietzsch S 833

Preibisch C Rueden S Saalfeld B Schmid J-Y Tinevez D J White V 834

Hartenstein K Eliceiri P Tomancak and A Cardona Nat Methods 2012 9 676ndash835

682 836

47 T Parasassi and E Gratton J Fluoresc 1995 5 59ndash69 837

48 M M Dodes Traian F L Gonzaacutelez Flecha and V Levi J Lipid Res 2012 53 609ndash838

616 839

49 D M Owen C Rentero A Magenau A Abu-Siniyeh and K Gaus Nat Protoc 840

2012 7 24ndash35 841

50 W Stillwell in An Introduction to Biological Membranes Academic Press 2016 pp 842

135ndash180 843

51 W Stillwell in An Introduction to Biological Membranes Academic Press 2016 pp 844

181ndash220 845

52 B Tenchov Chem Phys Lipids 1991 57 165ndash177 846

53 R Koynova and M Caffrey Biochim Biophys Acta 1998 1376 91ndash145 847

54 F Schmid S Dolezel O Lenz and S Meinhardt J Phys Conf Ser 2014 487 848

012004 849

55 S Meinhardt R L C Vink and F Schmid Proc Natl Acad Sci U S A 2013 110 850

4476ndash4481 851

56 O Glatter and O Kratky Small Angle X-ray Scattering Academic Press London 852

1982 853

57 B A Cunningham A D Brown D H Wolfe W P Williams and A Brain Phys 854

Rev E 1998 58 3662ndash3672 855

58 M Nadler A Steiner T Dvir O Szekely P Szekely A Ginsburg R Asor R Resh 856

C Tamburu M Peres and U Raviv Soft Matter 2011 7 1512ndash1523 857

59 C Tanford The hydrophobic effect Formation of micelles and biological membranes 858

John Wiley amp Sons Inc New York 1973 859

60 M Masuda and T Shimizu Langmuir 2004 20 5969ndash5977 860

33

61 A Zidovska K K Ewert J Quispe B Carragher C S Potter and C R Safinya 861

Block Liposomes Vesicles of Charged Lipids with Distinctly Shaped Nanoscale 862

Sphere- Pear- Tube- or Rod-Segments Elsevier Inc 1st edn 2009 vol 465 863

62 A Iglic V Kralj-Iglic B Božič M Bobrowska-Haumlgerstrand B Isomaa and H 864

Haumlgerstrand Bioelectrochemistry 2000 52 203ndash211 865

63 T Shimizu M Masuda and H Minamikawa Chem Rev 2005 105 1401ndash1444 866

64 J V Selinger and J M Schnur Phys Rev Lett 1993 71 4091ndash4094 867

65 J M Schnur Science (80- ) 1993 262 1669ndash1676 868

66 B Tenchov R Koynova and G Rapp Biophys J 2001 80 1873ndash1890 869

67 R Koynova and B Tenchov Ser Biomech 2017 31 3ndash19 870

68 S J Angyal and R Le Fur Carbohydr Res 1984 126 15ndash26 871

69 J K Whitesell and M A Minton J Am Chem Soc 1987 109 225ndash228 872

70 T Takayama S Ando and I Ando J Mol Struct 1990 220 245ndash250 873

71 K Gawrisch N V Eldho and I V Polozov Chem Phys Lipids 2002 116 135ndash151 874

72 F Tielens C Gervais J F Lambert F Mauri and D Costa Chem Mater 2008 20 875

3336ndash3344 876

73 G A Jeffrey and Y Yeon Acta Crystallogr Sect B 1986 42 410ndash413 877

74 B Berglund and R W Vaughan J Chem Phys 1980 73 2037ndash2043 878

75 C Gervais C Coelho T Azaiumls J Maquet G Laurent F Pourpoint C Bonhomme 879

P Florian B Alonso G Guerrero P H Mutin and F Mauri J Magn Reson 2007 880

187 131ndash140 881

76 C Akita T Kawaguchi F Kaneko H Yamamoto and M Suzuki J Phys Chem B 882

2004 108 4862ndash4868 883

77 R N Purusottam L Seacutenicourt J J Lacapegravere and P Tekely Biochim Biophys Acta - 884

Biomembr 2015 1848 3134ndash3139 885

78 X Biarneacutes A Ardegravevol A Planas C Rovira A Laio and M Parrinello J Am Chem 886

Soc 2007 129 10686ndash10693 887

79 L M J Kroon‐Batenburg and J Kroon Biopolymers 1990 29 1243ndash1248 888

80 K N Kirschner and R J Woods Proc Natl Acad Sci U S A 2001 98 10541ndash889

10545 890

81 S Svenson B Kirste and J H Fuhrhop J Am Chem Soc 1994 116 11969ndash11975 891

82 I Sack S Macholl J H Fuhrhop and G Buntkowsky Phys Chem Chem Phys 892

2000 2 1781ndash1788 893

83 S Svenson J Koening and J H Fuhrhop J Phys Chem 1994 98 1022ndash1028 894

34

84 S Svenson A Schaefer and J H Fuhrhop J Chem Soc Perkin Trans 2 1994 2 895

1023ndash1028 896

85 K Bock and C Pedersen Adv Carbohydr Chem Biochem 1983 41 27ndash66 897

86 T B Grindley Structure and Conformation of Carbohydrates Springer Berlin 898

Heidelberg 2008 899

87 F Tantakitti J Boekhoven X Wang R V Kazantsev T Yu J Li E Zhuang R 900

Zandi J H Ortony C J Newcomb L C Palmer G S Shekhawat M O De La 901

Cruz G C Schatz and S I Stupp Nat Mater 2016 15 469ndash476 902

903

1

Supporting Information 904

905

Topological connection between vesicles and nanotubes in single-906

component lipid membranes driven by head-tail interactions 907

908

Niki Baccilea Ceacutedric Lorthioira Abdoul Aziz Baa Patrick Le Griela Cristina Coelhob Javier 909

Perezc Wim Soetaertd Sophie L K W Roelantsd 910

911

a Sorbonne Universiteacute Centre National de la Recherche Scientifique Laboratoire de Chimie de 912

la Matiegravere Condenseacutee de Paris LCMCP F-75005 Paris France 913

b Institut des Mateacuteriaux de Paris Centre Sorbonne Universiteacute Paris France 914

c Synchrotron Soleil LrsquoOrme des Merisiers Saint-Aubin BP48 91192 Gif-sur-Yvette Cedex 915

France 916

d InBio Department of Biotechnology Ghent University Ghent Belgium 917

918

919

Corresponding author 920

Dr Niki Baccile 921

E-mail address nikibaccilesorbonne-universitefr 922

Phone +33 1 44 27 56 77 923

924

925

926

2

a) 927

Parameter Method

Dry Matter (DM ) 996 Infrared balance 105degC

Glucose () 012 HPLC-Metacarb

Glycerol () nd HPLC-Metacarb

Free Fatty Acid content () lt001 Internal method BBEPP

GC with FID Detector

Oil () lt001 Internal method BBEPP

GC with FID Detector

Moulds (CFUg) lt10 3M-Nordval ndeg16 B

Yeasts (CFUg) lt10 3M-Nordval ndeg16 B

Anaerobic count (CFUg) lt10 SP-VG M005 B

Protein () tbd BCA protein assay

928

Figure S 1 - a) HPLC-ELSD chromatogram and composition table 929

930

931

932

933

3

b) 934

935

Group Ndeg δ ppm

CH2 1 354

CH2 2 153

CH2 3-712-14 132

CH2 811 203

CH 910 534

CH2 15 162

CH2 16 388

CH 1 424

CH 2 316

CH 3 334

CH 4 329

CH 5 354

CH2 6 366

Figure S 1 (next) ndash b) 1H NMR spectrum (MeOD-d4) recorded on the purified G-C181-OH compounds 936

and related assignment 937

938

2

811910 1rsquo

5rsquo1

3rsquo4rsquo

1rsquo1910

asymp asymp asympHDO MeOD Acetone

6rsquo2rsquo

1615

3-712-14

4

939

940

Figure S 2 ndash Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 130degC 941

5

942

943

Figure S 3 - Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 90degC 944

945

90degC

1

22

6

946

947

948

949

Figure S 4 ndash Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 70degC 950

951

21

11

1

2

5

3

4

45

5

6

7

7

7

952

953

954

Figure S 5 ndash Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 25degC 955

956

957

1

1 1

8

958

Figure S 6 ndash a) SANS experiment recorded at 90degC for a 5 mgmL G-C181-OH aqueous solution b-c) 959

WAXS experiments recorded on aqueous solutions of G-C181-OH at b) 5 mgmL and c) 25 mgmL d) 960

Structural interpretation of the repeating distances pointed out in c) 961

962

963

964

001 01 1

q Aring-1

25degC

68degC

88degC

111degC

01 1q Aring

-1

02 04 06 08 1q Aring

-1

001 01

10-1

100

101

Inte

nsity

cm

-1

q Aring-1

-2

T= 90degC

SANS WAXS WAXS

a) b) c)

dv

dL d= 144 Aring-1

di

ddH2O

dt

dvdL

dtdi d

d)

9

965

966

Figure S 7 ndash ssNMR MAS (10 kHz) experiments recorded on a G-C181-OH dry powder a) 1H spectra 967

recorded from 110degC to 70degC b) 13C single pulse (SP) spectra recorded with high power 1H decoupling from 968

110degC to 70degC c) 13C cross polarization (CP) spectra recorded at 110degC and contact time tc of 50 micros and 969

3000 micros d) 13C CP spectra recorded at 70degC and for 20 lt tc micros lt 3000 970

971

1

811

910

CH2 CH2

1rsquo

6rsquo

8 6 4 2 0 ppm

110

90

80

70

910

1rsquo

6rsquo16

1

2rsquo3rsquo4rsquo

811

OH

45 40 35 30 25 20 15

ppm

200 180 160 140 120 100 80 60 40 20 0

ppm140 120 100 80

ppm

CH2-gauche

3004

110

90

80

70

65

60

50

25

C91013003

C9 - 12869

CH2-all-trans

3198

C1rsquo10335

a) b)

1H

T degCT degC

13C Single Pulse MAS

A

B

A

B

70degC70degC

5rsquo

215

200 180 160 140 120 100 80 60 40 20 0

ppm

13C CPMAST= 110degC

200 180 160 140 120 100 80 60 40 20 0

ppm

132 131 130 129 128 127 126

ppm

106 105 104 103 102

ppm

13C CPMAST= 70degC

50

tc micros

3000

20

tc micros

50 500 1000 3000

tc micros

C102

13024 C910113003

C92

12869

C1rsquo213082

C1rsquo110443

A B

106 105 104 103 102

ppm

3000

1000

500

50

20

AB

c)

d)

10

972

Figure S 8 ndash a) Short and d) long T2 components displayed by the 1H transverse relaxation signal c) 1H 973

transverse relaxation signals recorded on a G-C181-OH dry powder from 110degC to 70degC The area of each 974

given resonance is plotted (log-lin) as a function of the echo time (CPMG pulse sequence) for each 975

temperature and fitted using the equation given above T2s and T2l indicate the short and long T2 relaxation 976

times respectively obtained by describing the 1H transverse relaxation signal using a biexponential decay 977

For H1rsquo H5rsquo and H6rsquo the 1H relaxation function could be described by a monoexponential decay 978

- Protons from the sugar group (H1rsquo H5rsquo H6rsquo) show a single T2 relaxation component (Figure 979

6c and Figure S 8) while protons of the acyl chain (CH2 H910 H811) show both short (Figure 980

6c) and long (Figure 6d) T2 relaxation component 981

- The short T2 relaxation time systematically decreases from 110degC to 70degC for all species by a 982

factor ranging from two to seven depending on the specific chemical group Very long T2 are 983

observed at 110degC for all acyl species while nearly unchanged values are measured below 984

90degC without any clear trend 985

- The sugar protons (H1rsquo H5rsquo H6rsquo) have the shortest set of T2 values (lt 1 ms) below 100degC 986

The long component of T2 generally ranges between 4 ms and 12 ms 987

0 1 2 3 4 5 6 7 8

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1 2 3 4 5 6 7 8

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

H1

H8H11

H9H10

CH2

H1rsquo

H6rsquo

CH2

H910H811 H1CH2

H1rsquo H5rsquo6rsquo

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

OH

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45 T= 110degC

T= 90degC

T= 80degC

T= 70degC

Log

In

ten

sity

Time ms

Fit

)

CH2 H910 H811 H1 H56 H1 OH

00

05

10

15

20

25

30

35

40

T2 m

s

T= 110degC

T= 90degC

T= 80degC

T= 70degC

CH2 H910 H811 H1

0

2

4

6

8

10

12

T2 m

s

T= 110degC

T= 90degC

T= 80degC

T= 70degC

Short T2(1H) Long T2(1H)

a) b)

c)

11

Table S 1 ndash 1H chemical shifts of G-C181-OH in solution and in the solid-state Solid-state spectra are 988

recorded at a MAS frequency of 10 kHz in DMSO-d6 broad peak 989

990

δH ppm

Group Ndeg Liquid

(MeOD-d4) Solid

(T= 110degC) Solid

(T= 70degC)

CH2 H1 354 357

CH2 H2 153 156

CH2 H3-712-14 132 135

CH2 H811 203 203

CH H910 534 534

CH2 H15 162 164

CH2 H16 388 383

CH H1 424 432

CH H2 316 335

CH H3 334 335

CH H4 329 335

CH H5 354 351 357

CH2 H6 366 383

OH OH1 432

45 463

OH OH2 492

OH OH3 490

OH OH4 487

OH OH6 445

991

992

H1H9H10H1rsquo

H6rsquo

OH2rsquo

OH1

12

Detailed discussion about the NOESY and DQ-SQ 2D 1H-1H ssNMR experiments presented in 993

Figure 9 994

995

2D 1H-1H ssNMR experiments are a unique tool to explore through-space proximities 996

between 1H nuclei ssNMR offers a large number of possibilities each adapted to a given 997

system The advantage of 2D maps with respect to a dynamics approach is their direct 998

readability in clear two nuclei which exchange their magnetization through-space (whichever 999

mechanism may be responsible for this exchange) will display a specific cross-peak However 1000

the way cross-peaks are presented and interpreted strictly depends on the nature of the pulse 1001

program which in turn is chosen according to the physical nature of the sample 1002

1D temperature-dependent 1H NMR experiments performed on G-C181-OH (Figure 1003

4a in the main text) indicate sharp peaks at 90degC and broad peaks at 70degC Narrow peaks 1004

indicate fast andor large-amplitude molecular tumbling averaging the strong 1H-1H dipolar 1005

interactions This excludes the use of pulse sequences like DQ-SQ which is not efficient under 1006

these conditions For this reason we employ spin-diffusion-based pulse sequences like 1007

NOESY which is more adapted for samples with high molecular mobility When mixing times 1008

are chosen to be short (lt 1 s) one generally explores through-space proximities within about 5 1009

10 Aring On the contrary when molecular motions are restricted dipolar coupling becomes 1010

significant 1H NMR peaks are broader and pulse sequences like DQ-SQ can now be used 1011

with a spatial resolution of less than 5 Aring 1012

Whenever possible DQ-SQ experiments are preferred over spin diffusion-based pulse 1013

sequences because they are more selective in particular for likewise interactions which are 1014

not discriminated in NOESY-type experiments This is illustrated by the typical interaction 1015

schemes for NOESY (top Figure 9a) and DQ-SQ (top Figure 9b) For two given A and B 1016

protons 2D NOESY maps will display two on-diagonal and two square-shaped symmetrical 1017

off-diagonal cross-peaks Off-diagonal cross-peaks illustrate the mutual A-B and B-A 1018

interaction but the on-diagonal peaks are systematic and do not prove any A-A or B-B 1019

interaction Similarly to NOESY 2D DQ-SQ maps display off-diagonal peaks (symmetric 1020

across the diagonal top Figure 9b) upon A-B and B-A interaction but in contrast they will 1021

display on-diagonal peaks only in the presence of self-interaction In the tutorial example of 1022

Figure 9b top the on-diagonal A-A peak indicates mutual A-A interaction while lack of on-1023

diagonal B-B peak evidences the absence of B-B proximities Combining NOESY and DQ-SQ 1024

experiments is capital to study the local arrangement of G-C181-OH and its evolution with 1025

temperature 1026

13

At T= 90degC the H1 and H5rsquo contributions are very close but possible to discriminate 1027

them (δ(H1)= 354 ppm δ(H5rsquo)= 351 ppm) according to 1H NMR chemical shifts in Table S 1028

1 2D 1H-1H NOESY in Figure 9a also discriminates between the two components attributed 1029

on the basis of the interactions with their mutual neighbors (green H1-H2 and H5rsquo-H1rsquo squares 1030

in Figure 9a) On this basis 2D NOESY does not show a direct H1rsquo-H1 correlation (blue red-1031

crossed) but rather an extensive network of intra-sugar interactions These features exclude 1032

suggest intra- and inter-sugar interactions and they exclude head-tail arrangements as a major 1033

molecular configuration at 90degC 1034

At 70degC the 1H NMR peaks are broader indicating a reduction of the T2 (1H) relaxation 1035

times andor residual homonuclear dipolar interactions excluding the use of NOESY In this 1036

case 1H-1H DQ-SQ correlation experiments are best suited Figure 9b presents the 1H-1H DQ-1037

SQ experiment on G-C181-OH at 70degC Peaks are broader than in NOESY and attribution may 1038

be cumbersome However the experiment is performed at 700 MHz and MAS= 22 kHz which 1039

improves spectral resolution with respect to the same experiment performed at 300 MHz and 1040

MAS= 10 kHz The auto-peak at δH= 356 ppm in the SQ dimension is attributed to H1-H1 spin 1041

pairs mainly resulting from intra CH2 interaction while the DQ cross-peaks at δH= 156-356 1042

ppm and δH= 432-356 ppm respectively origin from H2-H1 and H1rsquo-H1 interactions (Figure 1043

9b) DQ-SQ map does not show any auto-peak corresponding to H1rsquo-H1rsquo or Hxrsquo-Hxrsquo (x= 2rsquo 1044

3rsquo 4rsquo) interactions (red-crossed blue Figure 9b) thus excluding any proximity (lt 5 Aring) between 1045

close glucose neighbors On the contrary DQ-SQ maps (recorded both at 700 MHz Figure 9b 1046

or at 300 MHz not shown) systematically show DQ cross-peaks which can only be attributed 1047

to H1rsquo-H1 interactions (δH= 432-356) The presence of these H1rsquo-H1 interactions can only be 1048

explained by a head-tail arrangement of the G-C181-OH molecules as illustrated below the 1049

DQ-SQ map in Figure 9b 1050

8

always varied from high (110degC) to low (70degC) before analysis Due to combined use of fast 217

MAS rates necessary to achieve high resolution spectra and heating close to the boiling point 218

of water neither diluted nor concentrated samples could be used for ssNMR experiments All 219

tests employing up to 50 wt G-C181-OH in water resulted in sample centrifugation drying 220

but also expulsion from the rotor causing sudden stop of rotor spinning For these reasons all 221

attempts to study hydrated G-C181-OH failed and only the dry powder could be analyzed The 222

dry sample powder was analyzed as such in zirconia rotors (4 mm or 25 mm) spinning at MAS= 223

10 kHz (4 mm) or 22 kHz (25 mm) and Chemical shifts were calibrated with respect to 224

adamantane (13C 13C(CH2)= 3848 ppm = 191 ppm) and glycine alpha (13C(CO)= 17603 225

ppm) Details on the acquisition parameters are given in the Supporting Information 226

227

Results and discussion 228

Topologically-connected nanotubes and vesicles 229

G-C181-OH is a biobased glucolipid water-insoluble at room temperature and of 230

which the physicochemical properties are poorly known The DSC thermogram performed at 231

10degCmin of G-C181-OH as a function of the water content (Figure 2a) shows two endothermic 232

peaks on the dry powder (996 wt glucolipid) They are attributed to two melting phenomena 233

occurring at the following temperatures Tm 483degC (Tm1) and 957degC (Tm2) the enthalpy of the 234

former being lower than that of the latter Interestingly DSC also shows that Tm1 is hydration 235

independent and reversible (the associated exothermic peak is systematically observed) while 236

the transition at Tm2 is simply suppressed (here tested up to 130degC) on hydrated powders and 237

never observed on the cooling profile even at slow cooling rates (1degCmin) The thermogram 238

recorded on the dry powder is typical of single-component lipids having an intermediate50 239

metastable phase (often referred to the ripple phase Pβ´ in reference to the periodic undulation 240

of the bilayer)50ndash54 between the classical lamellar gel (Lβ) and liquid crystalline (Lα) phases50ndash241

54 and characterized by regions of both liquid order and disorder5455 The effect of water does 242

not influence the low-temperature crystallization at Tm1 but it contributes to disrupt the ordered 243

array of hydrocarbon chains in the metastable phase 244

On the basis of the DSC thermogram the aqueous phase behavior of G-C181-OH under 245

diluted conditions (5 mgmL-1) was studied at four different temperatures (red marks on Figure 246

2a) below and above Tm1 but also between Tm1 and Tm2 corresponding to the thermogram of 247

the dry powder 248

9

249

Figure 2 ndash a) DSC thermogram of G-C181-OH under dry (04 wt water estimated by thermogravimetric 250

analysis) and hydrated conditions recorded at a heating rate of 10degCmin Red marks indicate the 251

temperatures at which cryo-TEM experiments are performed on the corresponding 5 mgmL aqueous 252

solutions b) Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 130degC 253

90degC 70degC and 25degC T indicates the temperature of the sample solution before cryofixation Tne T T 254

being the temperature of the sample upon plunging in liquid ethane Please refer to the materials and 255

methods section for more information 256

T= 130degC

T= 90degC

T= 70degC

T= 25degC

a)

b)

20 40 60 80 100 120

99 wt

50 wt

25 wt

T degC

En

do

10 wt

Tm1= 483 degC

Tm2= 957 degC

10

257

The series of cryo-TEM images in Figure 2b show an overview of the most relevant 258

structures observed on G-C181-OH solution from T= 130degC to T= 25degC whereas T is 259

indicative of the equilibrium temperature before cryofixation (refer to materials and methods 260

section for more information on the sense and interpretation of T) At Tgt 100degC the sample 261

is essentially composed of polydisperse single wall vesicles massively shown by the 262

complementary pictures presented in Figure S 2 In between Tm1 and Tm2 at T= 90degC and T= 263

70degC (Figure 2b Figure 3 Figure 4 complemented by Figure S 3 and Figure S 4) vesicles are 264

connected by nanotubes At Tlt Tm1 flat crystals are otherwise observed (Figure 2b and Figure 265

S 5) 266

267

268

Figure 3 ndash Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 90degC 269

270

A deeper insight at T= 90degC is given in Figure 3 (more images are provided in Figure 271

S 3) Arrows 1 through 6 mark very clearly the typical continuity in the glucolipid membrane 272

between the vesicle and nanotubes Arrow Ndeg 7 shows that a single nanotube connects the outer 273

membrane of two vesicles several hundred nanometers apart The connection is shown by 274

2

34

5

6

6

1

5

7

A

B

D

C

0 10 20 30 40 50 60

BIn

tensity

au

Profile nm

A

0 10 20 30 40 50 60

Inte

nsity

au

Profile nm

C

D

T= 90degC

11

arrows Ndeg5 and Ndeg6 The corresponding profiles A through D confirm the tubular nature of 275

the elongated objects Profiles B and D are both characterized by typical symmetrical thick 276

walls (low intensity depths highlighted by shaded rectangles) surrounding a hollow region (high 277

intensity peak between the shaded rectangles) The wall thickness varies between 35 and 40 278

nm for profile B and between 41 and 61 nm for profile D while the total nanotube diameter 279

varies between 14 nm at B and 20 nm at D At the same time the thickness of the vesicle wall 280

is about 6 nm at A and 135 nm at C Considerations about the bilayer or interdigitated structure 281

of the membrane will be presented below 282

The connection between nanotubes and vesicles seems to be multiple For instance 283

arrow Ndeg1 (Figure S 3) and Ndeg4 (Figure 3) show a single-wall vesicle of which the membrane 284

is extruded into a nanotube Similarly a closer look at arrows Ndeg1 2 and 3 in Figure 3 show 285

that the vesicle is single-wall and fully closed while the nanotube nucleates at the vesicle outer 286

surface with no apparent topological continuity In many regions of the sample at T= 90degC 287

one can find exvaginations or buds232531 in the nanotube wall This is shown for instance by 288

arrows Ndeg7 and particularly put in evidence in Figure 4 by arrows Ndeg1 through Ndeg4 where the 289

amplitude of the oscillation varies between 20-25 nm (arrow Ndeg1) and 6-8 nm (arrows Ndeg3-4) 290

291

Figure 4 - Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heated at T= 90degC 292

293

Still between Tm2 and Tm1 but at lower temperature (T= 70degC) the nanotubes seem to 294

vary from single threads (arrow Ndeg1 Figure S 4) to nanotubes of diameter below 15 nm both 295

connecting nanotubes of diameter ranging between 50 nm and 100 nm (arrows Ndeg2 Figure S 296

4) The larger tubes are most likely formed by the inflation and flattening of the nanotubes 297

probably after assimilation of surrounding vesicles and fusion of the exvagination discussed 298

above The transition between vesicles tubes and flat membranes is certainly more accentuated 299

Arrow Ndeg3 in Figure S 4 shows distinct vesicles while arrow Ndeg4 identifies a flat structure 300

evidently formed from fused vesicles In particular arrow Ndeg5 points at the fusion of two 301

2

1 132

4

12

vesicles while arrow Ndeg6 points at two vesicles fusing into the flat membrane Additional areas 302

where vesicle fusion is prominent are indicated by the oscillations at the outer skirt of broad 303

lamellae and pointed by arrow Ndeg7 304

Finally below Tm1 at T= 25degC for instance mainly flat crystals with no visible 305

exvaginations or tubes are visible (Figure S 5) by cryo-TEM and the material turns into an 306

insoluble precipitate Massive presence of flat crystalline structures is also clear from the bulk 307

iridescent behavior of a G-C181-OH solution under manual shear 308

In summary despite the uncertainty related to the actual state of equilibrium of the 309

glucolipids at the moment of cryofixation cryo-TEM depicts a vesicle phase at Tgt 100degC a 310

flat lamellar phase below Tm1 and an interconnected vesicle-nanotube network between Tm1 and 311

about 100degC The combination of statistically-relevant fluorescence microscopy and 312

scatteringdiffraction experiments provides an additional proof of the coexistence of 313

vesiclenanotube systems between the two transition temperatures Please note that the 314

experiments reported below have been performed under controlled conditions of temperature 315

which is then indicated by the use of T instead of T 316

13

317

Figure 5 ndash a-c) Confocal fluorescence microscopy images recorded at 90degC and 70degC for a 5 mgmL G-318

C181-OH aqueous solution d-g) Typical fluorescence microscopy images recorded at T= 80degC using 319

Laurdan (refer to materials and methods section for more details on this fluorophore) Excitation 320

wavelength is 405 nm for all the images while emission wavelength is 490 nm in (df) and 440 nm in (eg) 321

h) Full emission spectra obtained from the integration of various fluorescence images recorded from 412 322

nm to 600 nm (step 5 nm) at 90degC 80degC and 70degC i) Box plot of GP defined as 119918119927 =(120640120786120786120782minus120640120786120791120782)

120640120786120786120782+120640120786120791120782 with 120640120786120786120782 323

and 120640120786120791120782 being the emission wavelengths at 440 nm and 490 nm at 90degC 80degC and 70degC Each data point 324

l= 490 nm l= 440 nm

l= 490 nm l= 440 nm

10 μm400 450 500 550 600

00

02

04

06

08

10

No

rma

lize

d I

nte

nsity

l nm

90degC

80degC

80degC

70degC

70degC

400 450 500 550 60000

02

04

06

08

10

No

rma

lize

d I

nte

nsity

l nm

90degC

80degC

80degC

70degC

70degC

d) e)

f) g)

h)

i)

T= 80degC

Laurdan

T= 80degC

T= 80degC T= 80degC

T= 90degC T= 70degC T= 70degC

a) b) c)

1

1

1

14

corresponds to a set of regions of interest (ROIs) (54 at 90degC 65 at 80degC and 75 at 70degC) selected in the 325

corresponding fluorescence images ROIs were representatively selected so to include the entire image but 326

also vesicles alone tubes alone and mixtures of vesicles and tubes 327

328

Figure 5ab shows two typical confocal fluorescence microscopy images recorded on a 329

G-C181-OH solution (5 mgmL) in a flame-sealed capillary heat at T= 90degC and T= 70degC 330

Although the resolution of optical microscopy is not comparable with cryo-TEM the sample is 331

broadly composed of vesicles coexisting with nanotubes whereas in some cases the nanotube 332

starts from the vesicle itself (arrow Ndeg1) in good agreement with cryo-TEM data Furthermore 333

SANS recorded at T= 90degC (Figure S 6a) shows a scattering profile with a pronounced -2 334

dependence of the scattering intensity against the wavevector q in the log-log scale This 335

indicates the massive presence of a flat interface and it is expected typical of bilayers in 336

vesicular morphologies56 However the typical signature of nanotubes cannot be observed in 337

the SANS profile and this can be explained by both an argument related to the relative lower 338

amount of nanotubes with respect to vesicles at 90degC but also to the steeper slope of flat 339

morphologies (-2) compared to tubes (-1) the signal of the former masking the tube signal at 340

low wavevector values The SANS profile is characterized by a diffraction peak at q= 0167 Aring-341

1 of which the assignment is refined by WAXS data 342

Temperature-driven WAXS experiments in Figure S 6bc (5 mgmL-1 and 25 mgmL-1 343

respectively) show a lamellar order possibly attributed to a multilamellar structure of the 344

vesicles (dv Figure S 6d) at high temperature and to the interplanar arrangement of the 345

crystalline flat structures (dL Figure S 6d) at T= 25degC The sharp (001) reflection indicates 346

typical interlamellar distance of 353 Aring at 25degC although this value varies with temperature 347

(400 Aring at 111degC 372 Aring at 88degC) thus confirming its assignment to a hydration-dependent 348

interlamellar order More intriguing is the second broader peak at q= 0403 Aring-1 (d= 156 Aring) 349

and its reflection both clearly visible at 88degC below Tm2 We attribute it to the inter-glucolipid 350

arrangement inside the tubular phase (dt Figure S 6d) coexisting with the vesicle phase 351

between Tm1 and Tm2 and to the planar lipid arrangement (di Figure S 6d) in the rigid flat 352

lamellar structure below Tm2 Its temperature- hence hydration- independent position also 353

confirms its structural origin Finally besides the water-structure peak at about 2 Aring-1 we also 354

observe a broad peak at 144 Aring-1 characteristics of the lateral packing (d Figure S 6d) of lipids 355

in bilayers While its position could be consistent with an Lα phase5758 above Tm2 its broadness 356

and positional invariance as well as the coexistence between vesicles and tubes do not allow a 357

precise assignment 358

15

Does G-C181-OH assume a bilayer or interdigitated structure the latter being the one 359

found for other similar bolaform glucolipids4243 The size of G-C181-OH can be estimated to 360

be at the most 32 Aring given by the length L of the C16 tail (L = 22 Aring from the Tanford formula 361

L= 154 + 1265 times n (n= 16)59) and the sugar (about 10 Aring)4243 However the cis conformation 362

of the C=C bond imposes a ldquokinkrdquo of 60deg in the aliphatic chain which reduces the effective 363

length of the oleic backbone to about 17 Aring for an effective total G-C181-OH length of about 364

27 Aring Cryo-TEM provides a tube thickness varying from 35 Aring to 60 Aring and a vesicle membrane 365

thickness of at least 60 Aring WAXS suggests an interlamellar thickness (membrane and water) 366

between 40 Aring and 37 Aring and an inter-lipid distance in the tube of about 16 Aring In the absence of 367

a clear form factor oscillation profile as found for similar glucolipids4243 one must 368

acknowledge that these data pledge neither in favor of fully interdigitated nor a double layer 369

structure and might suggest a tilted lipid arrangement with partial interdigitation60 370

371

13C ssNMR demonstrates that G-C181-OH can assume two coexisting head-head and 372

head-tail configurations 373

Vesicles are characterized by low positive mean and gaussian curvatures while 374

nanotubes are described by a combination of a high positive mean and zero gaussian curvatures 375

In this regard the coexistence of vesicles and nanotubes requires glucolipids to adapt 376

themselves to both environments at the same time For this reason the formation of nanotube-377

vesicle networks generally requires a specific driving force able to impose a high mean and 378

zero gaussian curvature to the membrane In many systems insertion of at least a second 379

constituent (lipid surfactant protein) in the membrane becomes a necessary internal driving 380

force for the nucleation and growth of so-called membrane inclusions from which 381

topologically-connected nanotubes can form10ndash123261 as explained by orientational ordering 382

arguments62 It has also been shown that external driving forces such as use of laser 383

ldquotweezersrdquo30 osmotic pressure6 electroformation5ndash82829 or controlled suction13ndash16 are 384

otherwise necessary as also supported by theoretical studies19 385

The bending energy of tubes is higher than that of a vesicle membrane8 due to the 386

dependence of the membrane free energy from the principal curvatures For this reason 387

isotropic (rotational symmetry along the main axis) but also anisotropic33 lipids generally 388

stabilize low-curvature shapes like flat membranes or vesicles Lipid nanotubes are also a 389

common phase formed by amphiphiles but generally for lipids with a certain complexity in 390

their molecular structures as exampled by diacetylenic derivatives of phospholipids among 391

others262763ndash65 Negative contributions to the bending energy favoring the simultaneous 392

16

presence of interconnected vesicles and tubes was shown to be driven by the presence of 393

anisotropic lipids in a membrane continuum Spontaneous deviations to the elastic theory in the 394

presence of surfactants lipids and proteins with anisotropic structures favor the formation of 395

protrusions with radius of the order of the fraction of a micron9ndash1219ndash2362 396

In the absence of both internal and external driving forces mentioned above G-C181-397

OH should behave as a standard lipid Its DSC profile (Figure 2)536667 under both hydrated and 398

dry conditions identifies a hydration-independent transition (Tm1) slightly above room 399

temperature this is followed by one main transition at higher temperature (Tm2) in the dry 400

sample50 For such a profile one expects a gel phase (Lβ) below Tm1 and a liquid crystalline 401

structure possibly in the morphology of vesicles (L4 phase) possibly above Tm1 and most likely 402

above Tm231 as experimentally found On the basis of the thermogram recorded on the dry 403

powder one may also expect a ldquoripplerdquo (Pβ) phase between Tm1 and Tm2 characterized by 404

periodic undulations of the bilayer (generally addressed to as regions with both liquid order and 405

disorder)5455 406

In the present work instead of a Lβ phase we observe flat crystals below Tm1 and a 407

vesicle-nanotube network instead of the ldquoripplerdquo phase The latter is particularly unexpected 408

because the membrane is composed of a single-component (G-C181-OH) and no anisotropic 409

external force (osmotic hydrodynamic electric) is used to pull and stabilize the tubes In 410

addition the nanotubes are observed at fairly high temperatures above Tm1 (between about 411

60degC and 90degC) while this morphology is generally favored at low temperature (proportionality 412

with 1kT figure 5 in Ref 62) commonly below a given transition temperature characteristics 413

for each specific amphiphile27 Furthermore nanotubes of G-C181-OH seem to be stable in 414

time thus differing from what is found in electroformed vesicles78 They are also observed 415

away from phase transition events (may them be at Tm1 or Tm2) as otherwise reported before 416

for a diacetylenic derivative of phosphatidylcholine26 and many other systems27 417

A stable nanotube-vesicle network for a single-component phospholipid in the absence 418

of external driving forces is not expected and has never been reported before It could be 419

explained by the coexistence of two micro-separated phases of different rigidity This 420

hypothesis is excluded by confocal microscopy on Laurdan-labelled membranes and magic 421

angle spinning (MAS) solid-state NMR (ssNMR) experiments 422

Laurdan is a well-known fluorophore used to put in evidence the coexistence of rigid 423

and fluid domains in both model and biological membranes47ndash49 Laurdan was used here to label 424

the G-C181-OH membranes between Tm1 and Tm2 and to verify the possible micro-segregation 425

and coexistence of rigid (eg Lβ λem= 440 nm) and fluid (eg Lα λem= 490 nm) domains in 426

17

tubes and vesicles respectively A set of quantitative experiments performed on a 5 mgmL-1 427

solution summarized in Figure 5 shows an excess of fluorescence at only λem= 490 nm for 428

temperature between 90degC and 70degC (Figure 5h) The GP value which quantifies the rigidfluid 429

character of the membrane is systematically negative (Figure 5i) indicating an overall fluid 430

membrane at all temperatures GP is quantified on more than 50 regions of interest (ROIs) at 431

each temperature including vesicles and tubes alone as well as connected nanotube-vesicles 432

(Figure 5d-g) The dispersion of GP at 80degC and 70degC between -08 and -02 could suggest the 433

spurious presence of membranes with various degrees of flexibility although fluidity is always 434

predominant The combination of WAXS (broad peak d= 144 Aring-1 invariant with 435

temperature Figure S 6bc) and confocal fluorescence microscopy using Laurdan confirms the 436

single-phase of the membrane composed of G-C181-OH both in vesicles and nanotubes This 437

is also consistent with the lack of phase transitions in DSC experiments between ~85degC and 438

~60degC especially in hydrated samples We conclude that the vesicle-nanotube transition is not 439

driven by the local segregation of gel phase (Lβ) microdomains nor that the nanotubes are 440

composed of such phase 441

ssNMR is the best-suited approach to investigate the origin of the local driving force for 442

the nucleation of anisotropic inclusions of which the average orientation depends on the local 443

membrane and inclusion curvatures and was shown to promote tubing62 ssNMR can provide 444

information on the local molecular mobility eg through the nuclear spin relaxation behavior 445

but also on intermolecular interactions and stereochemistry eg through variations in the 446

chemical shift68ndash70 447

Ideally ssNMR should be performed on diluted G-C181-OH samples above Tm1 448

However fast MAS necessary for high resolution centrifuges the sample As a first 449

approximation hydrated G-C181-OH samples for instance at 50 wt as classically done for 450

phospholipids71 could be studied Unfortunately permeability of the sample holder combined 451

to heating (close to the boiling point of water) and MAS provoke dehydration and even sample 452

ejection After experiencing both of the above this approach was abandoned As a second 453

approximation the study of the dehydrated G-C181-OH powder avoids centrifugation and 454

evaporation If studying dried lipids may not reflect their exact behaviour under diluted 455

conditions valuable structural information can still be obtained in dehydrated molecular 456

systems like proteins thus illustrating that this approach is not unreasonable Furthermore if 457

DSC (Figure 2a) shows that hydration suppresses the Tm2 it also shows that the value of Tm1 is 458

not affected by water at concentrations as low as 10 wt For this reason performing a 459

structural analysis of dehydrated G-C181-OH by ssNMR above Tm1 may help understanding 460

18

the local arrangement of G-C181-OH in two morphologies vesicles and nanotubes having 461

sensibly different curvatures 462

To explore the local mobility of the most relevant chemical groups of G-C181-OH 463

above Tm2 (T= 110degC) and between Tm2 and Tm1 (70 lt T degC lt 90) we combine a series of 1H 464

and 13C experiments Single-pulse (SP) MAS experiments (1H and 13C) when recorded under 465

quantitative conditions allow the detection of all chemical species while cross-polarization 466

(CP) MAS (13C only) experiments are sensitive only to constrained groups for which the 1H-467

13C dipolar coupling is not averaged to zero by the molecular motions 468

469

Figure 6 ndash a) 1H MAS (10 kHz) ssNMR spectra recorded from 110degC to 70degC b) 13C MAS (10 kHz) ssNMR 470

spectra recorded using high power decoupling single pulse (SP) and cross-polarization (CP) pulse sequences 471

Contact time for CP experiments and corresponding temperatures are given on each spectrum Full spectra 472

are given in Figure S 7 473

The most relevant 1H and 13C spectra are given in Figure 6 (full range in Figure 7 and 474

Figure S 7) As expected the 1H solid-state SPMAS spectra show broader peaks compared to 475

the solution-state NMR spectrum (Figure 6a) reflecting the obvious solid-state character of the 476

sample at any temperature The sugar group and the C=C bond resonate in the 25-6 ppm range 477

and 535 ppm respectively while the OH groups of the sugar moiety are found below 45 ppm 478

(Figure 6a full assignment in Table S 1) Very broad (~1 ppm) at 100degC the position of the 479

OH peak varies with the temperature value and becomes narrower (lt~05 ppm) and visible at 480

δH= 464 ppm at T= 70degC Line narrowing and chemical shift variations of the OH resonance 481

in ssNMR is always a direct sign of H-bonding interactions whereas the strength of the H-482

bonding hence its OHhellipH distance is correlated to a dispersion in chemical shifts72ndash75 All other 483

60 55 50 45 40 35 30 25

ppm

110

90

80

70

910

1rsquo

6rsquo16

1

OH

134 132 130 128 126

ppm110 108 106 104 102 100

ppm

SP (T= 110degC)

SP (T= 70degC)

CP 50 ms (T= 70degC)

CP 3 ms (T= 70degC)

910 1rsquo

1H 13C

1

811

910

CH2 CH2

1rsquo

6rsquo

a) b)

C102

13024

C9101

13003

C92

12869 C1rsquo210382

C1rsquo110443

C1rsquo13035

2rsquo3rsquo4rsquo

T degC

5rsquo

19

resonances broaden from 110degC to 70degC thus indicating slower andor more anisotropic 484

reorientational motions of all chemical groups The specific dynamic behaviour of the sugar 485

headgroup is also reflected by its 1H transverse relaxation time T2 whereas the shorter the T2 486

the higher the rigidity Experiments performed between 110degC and 70degC shown and 487

commented in Figure S 8 in the Supporting Information globally show a single short (lt 1 ms 488

below 100degC) T2 relaxation component for H1rsquo H5rsquo and H6rsquo and two both short and long 489

(between 4 ms and 12 ms) T2 for the acyl chain (CH2 H910 H811) even below 100degC 490

At 110degC 13C SPMAS experiments (Figure 6b full range in Figure 7 and Figure S 7 491

assignments in Table 1) show a single peak at δC= 10335 ppm for C1rsquo and at δC= 13003 ppm 492

for C910 as expected for a symmetrical conformation with two chemically-equivalent 493

carbons76 At 70degC the C1rsquo peak disappears while C910 now shows two peaks around 128 494

ppm and 130 ppm (Table 1 full temperature series in Figure S 7b) attributed to C92 (δC= 12869 495

ppm) and C10 in a non-symmetric configuration respectively for which these two carbons are 496

chemically inequivalent76 More precisely the resonance at 130 ppm is actually the 497

superposition of two close resonances C9101 (δC= 13003 ppm) and C102 (δC= 13024 ppm) 498

where subscripts 1 and 2 stand for two different conformations of the cis-olefin group as 499

reported for various phases of oleic acid in the vicinity of 0degC76 500

501

Table 1 ndash 13C Chemical shifts of G-C181-OH in solution and solid-state Solid-state experiments are 502

recorded at MAS= 10 kHz All spectra are reported in Figure 7 and Figure S 7 The contact time for cross 503

polarization is 3 ms In all ssNMR experiments temperature is first raised at 110degC and then lowered to 504

either 90degC or 70degC Attribution of the 13C resonances in Table 1 is performed on the basis of the solution 505

spectrum (Figure S 1b) Additional peaks observed in both single pulse and cross polarization ssNMR 506

experiments are attributed when possible to a configuration effect of the given atom eg C9101 and C92 507

indicate two configurations of C9 All new peaks observed in solid-state and not detected in solution are 508

highlighted in red for convenience 509

510

511

δ ppm

Peak Liquid (DMSO-d6)

Solid Single Pulse

T= 110degC

Solid Single Pulse

T= 90degC

Solid Single Pulse

T= 70degC

Solid Cross Polarization

T= 70degC

C102

- - 13024 13024

C910C9101 13009 13003 13003 13003 13003

C1C9C10C1rsquo

C6rsquo

20

C92 - - - 12869 12869

C1rsquoC11 10329 10335 10335 - 10443

C12 - - - - 10382

C3 7724 7709 7702 - 7829

- 7634 7629 - 7621

7525

C2 7388 7398 7390 7370 7370

C4 7052 7077 7054 7041 7098

C16 6900 7016 - - 7029

- - - - 6793

C1C5C6 6153 6241 6236 6239 6262

C1C5C6 6117 - - - 5956

C2 3300 3290 3290 3283 3279

CH2(trans) - - - 3198 3196

CH2(gauche)C15 2941 2981 2987 3004 -

C811 2706 2747 2747 2832 2817

CH2 2597 2625 2628 2630 2817

512

13C CPMAS spectra at 70degC (Figure 6b full range in Figure 7 and Figure S 7d) at 513

increasing contact times confirm the presence of two cis-olefin environments one for which 514

C910 are equivalent labelled here C9101 and one for which they are inequivalent labelled 515

C92 C102 Both environments are rigid enough to be observed at short contact time (tc= 50 μs) 516

A similar double environment is also observed for the sugar C1rsquo CPMAS at 70degC shows two 517

peaks (C1rsquo1 at δC= 10443 ppm C1rsquo2 at δC= 10382 ppm Figure 6b Figure S 7d) These are 518

significantly shifted downfield from the C1rsquo peak observed at 110degC and are detected at contact 519

times as short as 50 μs thus indicating their rigid character One also finds that CPMAS is 520

totally inefficient at 100degC for short contact times (50 μs) and poorly efficient even at 3 ms 521

typical of fast-tumbling molecules (Figure S 7c) also both gauche (δC= 3004 ppm) and all-522

trans (δC= 3198 ppm) conformations of the CH2 carbons coexist at 70degC (Figure S 7b)77 523

According to ssNMR one main mobile molecular environment exists at 110degC as 524

shown by the well-resolved 1H spectrum single C=C (C9101) and anomeric (C1rsquo) resonances 525

gauche conformation of the aliphatic CH2 and single T2 of the aliphatic CH2 chain and sugar 526

protons The sugar group is certainly mobile (inefficient 1H13C CPMAS) but still more 527

constrainted than the acyl chain (shorter T2 (1H)) As the temperature is lowered the overall 528

rigidity increases (broadening of 1H NMR peaks faster 1H transverse relaxation and efficient 529

1H13C magnetization transfer during CPMAS) presumably driven by H-bonding 530

interactions (downfield shift and line narrowing of the OH 1H signal very short OH T2 loss of 531

the C1rsquo signal in SPMAS) At the same time a number of spectroscopic data (two concomitant 532

21

resonances for anomeric C1rsquo1 and C1rsquo2 cis-olefin C9101 and C92 C102 and CH2 gauche and 533

all-trans environments) and relaxation behaviour (fast and slow T2 relaxation components) 534

strongly suggests the coexistence of two molecular environments Further proofs of the 535

coexistence of two distinct molecular domains are given by a number of chemical shift 536

variations in the 13C CPMAS spectrum at 70degC compared to the one at 110degC 90degC and 537

solution This is shown in Figure 7 and in the list of 13C chemical shifts in Table 1 538

539

540

Figure 7 - Comparison between solution- and solid-state 13C NMR spectra of G-C181-OH Solid-state 541

experiments are recorded directly on the sample powder at a MAS of 10 kHz (4 mm zirconia rotor) SP 542

Single Pulse CP Cross Polarization In all the ssNMR experiments temperature is first raised at 110degC 543

and then lowered to either 90degC or 70degC Segmented lines are meant to highlight the known peak position 544

in solution thus stressing the strong chemical shift variations in the solid-state 545

546

NMR experiments above Tm2 suggest that vesicles could contain G-C181-OH with a 547

single family of mobile conformations characterized by chemically-equivalent C=C (red 548

Figure 8b) and a single C1rsquo conformation (green Figure 8b) Experiments between Tm1 and Tm2 549

otherwise suggest that two coexisting conformations characterized by chemically-equivalent 550

(red Figure 8b) and inequivalent (magenta Figure 8b) C=C could be associated to nanotubes 551

Two distinct C1rsquo sites (grey Figure 8b) also indicate two configurations of the sugar headgroup 552

A comment on this point will be given in the following paragraphs 553

The local (Aring-scale) variation of molecular mobility below Tm2 drives the coexistence of 554

two distinct molecular environments of the same G-C181-OH molecule This occurs within a 555

single-phase domain (as deduced by WAXS) of equivalent membrane rigidity (as derived from 556

Laurdan probe in confocal microscopy) Although never reported before the difference in the 557

local mobility of the same amphiphile (here G-C181-OH) could satisfy the theoretically-558

140 120 100 80 60 40 20

CP(3 ms)-70degC

SP-70degC

SP-90degC

SP-110degC

ppm

LiquidC910

C1rsquo

C2rsquo

C4rsquo

C3rsquo

C1

C6rsquoC5rsquo

C1

6

C8

11

C2 C

H2

C1

5

CH

2

DM

SO

13C

22

required conditions that induce local lipid segregation and formation of membrane inclusions 559

(Figure 8c) In case of their non-zero average orientation the local membrane curvature 560

increases62 and nanotubes can spontaneously form (Figure 8bc) More intriguing this 561

mechanism occurs at least 10degC above Tm1 although it should not be favored above the Tm27 562

From both experimental and theoretical backgrounds the conditions of topologically-563

connected vesicles-nanotubes systems are met only when either internal (two components) or 564

external driving forces are applied19 In the present single-lipid system of which the phase 565

behavior is summarized in Figure 8a the most plausible driving force could be explained by 566

the NMR results collected on the dry sample G-C181-OH concomitantly occupies two 567

molecular environments characterized by 1) two distinct conformations of the sugar 568

displaying equivalent mobility and 2) two conformations of the acyl chain unsaturation 569

570

23

571

Figure 8 ndash a) Thermotropic phase behavior of G-C181-OH at 5 mgmL-1 Note that Tm2 is observed on the 572

dry powder only b) Possible relationship between the molecular environment as derived from NMR and 573

membrane curvatures All the molecular drawings refer to the same G-C181-OH molecule (bulky ellipsoid 574

equiv sugar headgroup) while each color corresponds to a given configuration Grey of C1rsquo1 and C1rsquo2 indicates 575

the uncertainty about the assignment of the specific NMR peak The curvature C is defined as 1R R being 576

the curvature radius R1 R2 radius of curvature CG Gauss curvature CM Mean curvature c) Free 577

interpretation of the anisotropic high-curvature (CMgt 0 CG= 0) inclusions in the membrane (low CM and 578

CG) continuum driven by inter-sugar H-bonding and reduced intramembrane diffusion for a G-C181-OH 579

glucolipid 580

581

Flat crystalline structure

Nanotube-vesicle network

Vesicles

48degC

120deg

106 105 104 103 102 101

ppm

H9 H10H1rsquo

106 105 104 103 102 101

ppm106 105 104 103 102 101

ppm106 105 104 103 102 101

ppm

C=CC=C

C1rsquo1C1rsquo2

C1rsquo

Tm2(96degC)Tm1

a)

b)

c)

Same molecule different local orientations and sugar conformations

High curvature membrane

Low curvature inclusion

Flip-FlopStrong H-bonding

CG

CM ( + )

CG 0

CM 0 CG 0

CM 0

G-C181-OH

24

Two related questions are unanswered at the moment What actually drives nanotube 582

formation and is the organization of G-C181-OH within the vesicles and tubes the same Lipid 583

nanotubes have been studied for at least three decades over which broad experimental and 584

theoretical insight has been generated Despite this fact the actual understanding is still matter 585

of debate27 Micelle-to-nanotube and vesicle-to-nanotube phase transitions are generally 586

reported for single amphiphiles below a given Tm and intermediate phase composed of twisted 587

and flat fibers is concomitantly reported Initially thought to be a chirality-driven phenomenon 588

recent studies propose a more general mechanism involving crystallization and symmetry 589

breaking in the lipid packing driven by optimal packing27 If twisted ribbons were observed 590

before for a number of microbial glycolipid bolaamphiphiles44 they are not found for G-C181-591

OH This fact as well as the large body of cryo-TEM images and the observation of nanotubes 592

above Tm1 rather pledge in favour of a direct topological change from positive (vesicles) to 593

zero (nanotubes) Gaussian curvatures In this case vesicle-to-nanotube transitions could be 594

driven by external forces56307813ndash162829 595

Considering the absence of external forces in the present system there are only two 596

possibilities left spontaneous budding of the membrane or nanotube pulling after inter-vesicle 597

collisions due to Brownian motion Both are possible as inclusions composed of a different 598

arrangement of the same molecule in the membrane continuum can theoretically explain 599

spontaneous nanotube formation in an isotropic medium672223629ndash1218ndash21 However the second 600

one is probably more likely to prevail because it could explain the large number of nanotubes 601

connecting two distant vesicles Interestingly similar events occur for any phospholipid-based 602

vesicular solution but nanotubes are extremely rare unless at least two lipids of different 603

molecular structure are mixed together The spectroscopic 1H and 13C and relaxation (1H) 604

ssNMR data even if collected on a dry powder demonstrate that the coexistence of two G-605

C181-OH configurations involving both the sugar headgroup and fatty alcohol chain are 606

possible for this molecule As a matter of fact this is not so surprising as more than one energy 607

minima each corresponding to a given conformation are known for glucose its non-reducing 608

derivative methylglucose and carbohydrates in general78ndash80 609

The presence of two configurations could explain the coexistence between two 610

structures of different curvature but the data presented so far do not explain the origin of these 611

two configurations In other words what drives the change in the local structure of the sugar 612

headgroup and fatty alcohol within the membrane 613

The association of known self-assembled structures to spectroscopic 13C ssNMR data is 614

known for years in particular for gluconamides lipids having a linear sugar headgroup81ndash84 It 615

25

was found that gauche effects in some cases up to the γ position in the sugar chain induce 616

upfield chemical shifts while anti and trans conformations result in downfield shifts Each 617

structure (monolayer crystal bilayer crystal or micellar fiber) was then characterized by a given 618

organization of the gluconamide and a characteristic 13C ssNMR fingerprint which could be 619

explained by gauche γ-gauche anti and trans conformations of the headgroup Similar 620

structural-spectroscopic correlations have been collected over the years for much simpler cyclic 621

sugars85 whereas pyranoses can undergo important conformational changes (chair half-chair 622

boat skew) when the appropriate energy barriers are crossed78ndash8086 A quantitative 623

conformational study based on the 13C chemical shift variations shown in Table 1 would be 624

certainly helpful but very delicate in the context of the present state of the art because a 625

rigorous approach requires at least one known crystal structure8182 which is currently missing 626

for this compound A qualitative approach will then be adopted 627

G-C181-OH is a bolaform amphiphile and flip-flop could be a possible mechanism 628

involved in the nanotube stabilization whereas head-tail interactions have long been considered 629

as important in nanotube-forming bolaform amphiphiles6063 From mere considerations based 630

on T2 (1H) relaxation arguments one could exclude head-tail configurations if the R-CH2OH 631

endgroup (H1) was involved in a direct interaction with the sugar group its corresponding T2 632

would probably be comparable to the sugarrsquos one (OH and H1rsquo) but this does not seem to be 633

the case However this hypothesis would be based on the comparison of the relaxation from 634

CH and CH2 protons additionally involved in different kind of motions Besides the relaxation 635

of H1 is influenced by the strong overlap between H1 and H5rsquo For these reasons relaxation 636

arguments are not solid enough and will not be used 637

We then address this question by exploring intermolecular interactions in a direct 638

manner using 2D 1H-1H homonuclear ssNMR correlation spectroscopy At 90degC in the fast-639

tumbling domain of molecular motion (narrow 1H NMR peaks Figure 6a) and corresponding 640

to the vesicle-rich region spin diffusion (NOESY) experiments are preferred over dipolar 641

coupling based experiments because such interaction is averaged out by molecular motion 642

However when the molecular motion is more constrained in the domain of nanotube stability 643

at 70degC (broad 1H spectra Figure 6a) Double-Quantum Single-Quantum (DQ-SQ) correlation 644

experiments are best suited to probe interaction between close molecular neighbors These 645

experiments are reported in Figure 9 and the related interpretation is only briefly reported 646

below while a more detailed discussion can be found on Page 12 in the Supporting Information 647

26

648

Figure 9 ndash a) 2D 1H-1H NOESY NMR experiment performed at MAS= 10 kHz (4 mm probe) on a Bruker 649

300 MHz Avance III HD spectrometer using 300 ms of mixing time Sample temperature is regulated to 650

90degC reached by cooling from 110degC b) 2D 1H-1H DQ-SQ (BAck-to-BAck BABA pulse sequence used for 651

the DQ excitation scheme) NMR experiment performed at MAS= 22 kHz (25 mm probe) on a Bruker 700 652

MHz Avance III spectrometer using a DQ excitation time equal to 2 rotor periods Sample temperature is 653

regulated to 70degC reached by cooling from 110degC The typical schemes illustrating how to interpret these 654

2D experiments are reported above each experimental map Zooms of the sugar region and configurational 655

interpretation are given below each 2D map More information are given in the text below 656

H16-H16

H1-H1

H9-H10

H1rsquo-H1

H1rsquoH910 H16

H1

H1rsquo-H1rsquo

Hxrsquo-Hxrsquo

2

(pp

m)

(ppm)

315

811910 1rsquo

2rsquo2

3rsquo 4rsquo6rsquo

5rsquo

1rsquo2910

asymp asymp asympHDO MeOD Acetone

1

315

811 9101rsquo

2rsquo2

3rsquo4rsquo6rsquo

5rsquo

1rsquo2 9 10

asympasympasympHDOMeODAcetone

H1rsquo-H1

1rsquo

1

(p

pm

)

(ppm)

H1-H2

H1rsquo-H1

H1rsquo

H910

H1H5rsquo

H2

H1rsquo-H5rsquo

315

811910 1rsquo

2rsquo2

3rsquo 4rsquo6rsquo

5rsquo

1rsquo2910

asymp asymp asympHDO MeOD Acetone

1

H2rsquoH3rsquoH4rsquo

H2rsquoH3rsquoH4rsquo

X= 2rsquo3rsquo4rsquo

NOESY - T= 90degC

1H

1H

1H

1H

DQ-SQ - T= 70degC

a) b)

2

(pp

m)

(p

pm

)

(ppm)

A B

AhellipB

AhellipA

BhellipB

NOESY

(ppm)

A B

AhellipB

DQ-SQ

Single-quantum dimension

Do

ub

le-q

ua

ntu

m d

imen

sio

n

AhellipA

BhellipB

BhellipA

BhellipA

315

811910 1rsquo

2rsquo2

3rsquo 4rsquo6rsquo

5rsquo

1rsquo2910

asymp asymp asympHDO MeOD Acetone

1

lt 5 Aring

lt 5 Aring

H1-H2

H811

H15

H16

H6rsquo

H1rsquo

H1

H5rsquo

H2

H1-H2

H1rsquo-H1

H1rsquo-H5rsquo

H1rsquo

H1 H5rsquo

H1rsquo

H1

H5rsquo

H1H2rsquo H3rsquo H4rsquo

H1-H1

H1rsquo-H1

Hxrsquo-HxrsquoX= 2rsquo3rsquo4rsquo

H1-H2

H1rsquo

Zoom sugar region 28 ndash 45 ppmNOESY DQ-SQ

27

Head-tail or head-head configurations could be probed by the existence of H1rsquo-H1 657

interactions whereas H1 (δH= 356 ppm) is associated to the glucolipid tail and H1rsquo (δH= 432 658

ppm) to the sugar headgroup These atoms are more than 15 Aring apart and a direct interaction 659

could not be observed in both NOESY and DQ-SQ experiments NOESY experiments (Figure 660

9a) show that H1rsquo-H1 correlations are not observed Meanwhile clear-cut direct H1rsquo-H1 661

interactions are indeed observed are detected at T= 70degC (DQ cross-peaks related to H1rsquo-H1 662

correlation in DQ-SQ experiment shown in Figure 9b) Interestingly the absence of auto peaks 663

at δH= 432 ppm and δH= 335 ppm in the SQ dimension excludes H1rsquo-H1rsquo and Hxrsquo-Hxrsquo (x= 2 664

3 4) interactions respectively strengthening the idea that sugar headgroups from two nearest-665

neighbor G-C181-OH molecules are not adjacent in the nanotube-vesicle region 666

Homonuclear 2D ssNMR supports the idea according to which G-C181-OH can 667

undergo molecular flip-flops and a head-tail arrangement in the dry product Extrapolating these 668

conclusions in the lower concentration range lack of H1rsquo-H1 proximity at T= 90degC suggests a 669

head-head configuration in the vesicle-rich region and head-tail configuration within the 670

nanotubes It is however still unclear whether flip-flops are driven by spontaneous temperature-671

induced conformational changes in the glucolipid packing stimulating inclusions within the 672

membrane itself672223629ndash1218ndash21 or by defects generated by inter-vesicle collisions In fact a 673

combination of both is not unlikely Nanotubes are not stable when temperature is 20 to 30 674

degrees above Tm1 indicating that inter-vesicle collisions alone cannot explain the unusual 675

phenomenon In an energy landscape approach87 the head-head configuration is certainly 676

favored at high temperature while both head-head and head-tail configurations could be 677

favored in the mid-temperature region corresponding to the range between Tm1 and Tm2 in the 678

dry powder Both statistical and defect-induced flip-flop events contribute to form high mean 679

zero Gaussian curvature morphologies (nanotubes) for which the head-tail conformation 680

limiting the proximity between bulky sugar headgroups is probably more adapted (Figure 8c) 681

The flip-flop interpretation now reasonably explains the strong differences in the sugar and fatty 682

alcohol conformations deduced by the multiple NMR signals attributed to chemically-683

equivalent groups The existence of various low-energy conformations each having specific 684

NMR signature for sugars in solution78ndash8086 and solid-state81ndash84 further supports our 685

interpretation 686

One last point should be clarified Why between Tm1 and Tm2 are two molecular 687

environments able to coexist instead of a single one of average rigidity as found in the case of 688

ldquoripplerdquo phases50ndash54 We believe that this question can be answered based on the NMR results 689

the relatively short T2 values of the H1rsquo and OH protons shows the constrained mobility of the 690

28

sugar headgroup despite a certain flexibility of the chain (long T2 component) This trend may 691

suggest restrictions in translational mobility probably due to the sugar-OH hydrogen bonding 692

(Figure 8c) meaning that lipids are not able to diffuse within the membrane thus keeping two 693

different configurations stable 694

695

Conclusions 696

Topologically-connected nanotube-vesicle lipidic systems are commonly observed for 697

self-assembled membranes composed of at least two amphiphiles and spontaneously driven by 698

anisotropic inclusions andor by external directional forces This work shows that strong 699

differences in membrane curvature can spontaneously occur in single-lipid systems without 700

external anisotropic forces Coexistence and interconnection between nanotubes and vesicles 701

are evidenced via cryo-TEM and fluorescence confocal microscopy in a single-phase region at 702

temperature above a transition temperature Tm1= 483degC determined by DSC on the hydrated 703

sample but below the second transition temperature Tm2 (957degC) determined by DSC on the 704

dry powder X-ray diffraction (WAXS) also shows the coexistence of two lattice periods one 705

attributed to an inter-membrane repeating distance and the other possibly related to the intra-706

nanotube glucolipid arrangement Labelling the systems with Laurdan a fluorophore of which 707

the emission strongly varies with the local membrane rigidity excludes the presence of rigid 708

micro-separated phases (eg Lβ) 709

1H and 13C ssNMR despite the fact that it can only be performed on the dry powder 710

confirm the presence of one main fluid G-C181-OH environment above Tm2 Below Tm2 the 711

splitting of some 13C peaks and the coexistence of short and long 1H relaxation components 712

demonstrate that G-C181-OH can simultaneously pack into structurally and dynamically 713

inequivalent environments Combining advanced 2D 1H-1H correlation experiments based on 714

spin diffusion or double-quanta excitation suggests that vesicles could contain a preferred head-715

head configuration of the bolaform lipid while nanotubes a preferred head-tail assembly 716

The presence of two environments composed of the same molecule characterized by 717

different molecular configurations and rigidity fulfills the theoretical conditions necessary to 718

spontaneously nucleate a membrane inclusion of non-zero average orientation described as 719

necessary to observe topologically-connected nanotube-vesicle structures in the absence of 720

external forces This hypothesis could also be enriched by the possibility of nanotubes to be 721

pulled away between two vesicles after a collision event 722

This work expands the actual theoretical and experimental framework on lipid networks 723

stabilized by a different radius of curvature It could also motivate the synthesis or isolation 724

29

of new lipids providing a higher level of 3D architectural control in soft networks possibly in 725

the range of room temperature 726

727

Acknowledgements 728

Dr Andrea Lassenberger and Dr Sylvain Preacutevost at Institut Laue Langevin (ILL Grenoble 729

France) are kindly acknowledged for their assistance on the SANS experiment Confocal 730

images were performed at the Institut de Biologie Paris Seine (IBPS) imaging facility Jean-731

Franccedilois Gilles (Imaging Core Facility Institut de Biologie Paris Seine (IBPS) CNRS 732

Sorbonne Universiteacute Paris France) is kindly acknowledged for his assistance on the confocal 733

microscopy imaging Dr Guillaume Laurent (LCMCP Sorbonne Universiteacute Paris France) is 734

kindly acknowledged for helpful discussions The French region Ile-de-France SESAME 735

program is acknowledged for financial support (700 MHz NMR 736

spectrometer) Dr Olivier Diat (ICSM Marcoule France) Prof eronikaIglič and Prof V Aleš737

onsul discussi Slovenia) are kindly acknowledged for helpfUniversity of LjubljanaIglič (-Kralj 738

739

Financial support 740

WAXS experiments have been supported by Soleil Light Source Saint Aubin France proposal 741

Ndeg 20201747 SANS experiments have been supported by ILL proposal Ndeg9-13-778 742

743

References 744

1 D Kabaso N Bobrovska W Goacuteźdź E Gongadze V Kralj-Iglič R Zorec and A 745

Iglič Bioelectrochemistry 2012 87 204ndash210 746

2 C R Safinya U Raviv D J Needleman A Zidovska M C Choi M A Ojeda-747

Lopez K K Ewert Y Li H P Miller J Quispe B Carragher C S Potter M W 748

Kim S C Feinstein and L Wilson Adv Mater 2011 23 2260ndash2270 749

3 P Veranič M Lokar G J Schuumltz J Weghuber S Wieser H Haumlgerstrand V Kralj-750

Iglič and A Iglič Biophys J 2008 95 4416ndash4425 751

4 A Rustom R Saffrich I Markovic P Walther and H-H Gerdes Science (80- ) 752

2004 303 1007ndash1010 753

5 M I Angelova S Soleacuteau P Meacuteleacuteard F Faucon and P Bothorel Prog Colloid 754

Polym Sci 1992 89 127ndash131 755

6 L Mathivet S Cribier and P F Devaux Biophys J 1996 70 1112ndash1121 756

7 V Kralj-Iglič G Gomišček J Majhenc V Arrigler and S Svetina Colloids Surfaces 757

A Physicochem Eng Asp 2001 181 315ndash318 758

30

8 V Kralj-Iglič A Iglič G Gomišček F Sevšek V Arrigler and H Haumlgerstrand J 759

Phys A Math Gen 2002 35 1533ndash1549 760

9 M Drab D Stopar V Kralj-Iglič and A Iglič Cells 2019 8 626 761

10 A Zidovska K K Ewert J Quispe B Carragher C S Potter and C R Safinya 762

Langmuir 2009 25 2979ndash2985 763

11 A Zidovska K K Ewert J Quispe B Carragher C S Potter and C R Safinya Soft 764

Matter 2011 7 8363ndash8369 765

12 A Zidovska K K Ewert J Quispe B Carragher C S Potter and C R Safinya 766

Biochim Biophys Acta - Biomembr 2009 1788 1869ndash1876 767

13 M Karlsson K Sott M Davidson A S Cans P Linderholm D Chiu and O Orwar 768

Proc Natl Acad Sci U S A 2002 99 11573ndash11578 769

14 A Karlsson R Karlsson M Karlsson A-S Cans A Stromberg F Ryttsen and O 770

Orwar Nature 2001 409 150ndash152 771

15 M Karlsson K Sott A S Cans A Karlsson R Karlsson and O Orwar Langmuir 772

2001 17 6754ndash6758 773

16 E Evans H Bowman A Leung D Needham and D Tirrell Science (80- ) 1996 774

273 933ndash935 775

17 U Seifert Configurations of fluid membranes and vesicles 1997 vol 46 776

18 J Kaumls and E Sackmann Biophys J 1991 60 825ndash844 777

19 N Bobrovska W Goacuteźdź V Kralj-Iglič and A Iglič PLoS One 2013 8 e73941 778

20 A Iglič B Babnik K Bohinc M Fošnarič H Haumlgerstrand and V Kralj-Iglič J 779

Biomech 2007 40 579ndash585 780

21 K Bohinc T Slivnik A Iglič M Brumen and V Kralj-Iglič J Phys Chem C 2007 781

111 9709ndash9718 782

22 A Iglič H Haumlgerstrand P Veranič A Plemenitaš and V Kralj-Iglič J Theor Biol 783

2006 240 368ndash373 784

23 V Kralj-Iglič and P Veranič Adv Planar Lipid Bilayers Liposomes 2006 5 129ndash785

149 786

24 A Iglic and V Kralj-Iglic Adv Planar Lipid Bilayers Liposomes 2006 4 253ndash279 787

25 S Penič L Mesarec M Fošnarič L Mroacutewczyńska H Haumlgerstrand V Kralj-Iglič and 788

A Iglič Front Phys 2020 8 342 789

26 P Yager and P E Schoen Mol Cryst Liq Cryst 1984 106 371ndash381 790

27 T G Barclay K Constantopoulos and J Matisons Chem Rev 2014 114 10217ndash791

10291 792

31

28 B Božič G Gomišček V Kralj-Iglič S Svetina and B Žekš Eur Biophys J 2002 793

31 487ndash496 794

29 A Iglič H Haumlgerstrand M Bobrowska-Haumlgerstrand V Arrigler and V Kralj-Iglič 795

Phys Lett A 2003 310 493ndash497 796

30 R Bar-Ziv and E Moses Phys Rev Lett 1994 73 1392ndash1395 797

31 S Chiruvolu H E Warriner E Naranjo S H Idziak J O Raumldler R J Plano J A 798

Zasadzinski and C R Safinya Science (80- ) 1994 266 1222ndash1225 799

32 V Kralj-Iglič A Iglič H Haumlgerstrand and P Peterlin Phys Rev E 2000 61 4230ndash800

4234 801

33 L Mesarec M Drab S Penič V Kralj-Iglič and A Iglič Int J Mol Sci 2021 22 802

2348 803

34 L Van Renterghem S L K W Roelants N Baccile K Uyttersprot M C Taelman 804

B Everaert S Mincke S Ledegen S Debrouwer K Scholtens C V Stevens W 805

Soetaert L Van Renterghem S L K W Roelants N Baccile K Uytersprot M C 806

Taelman B Everaert S Mincke S Ledegen S Debrouwer K Scholtens C V 807

Stevens and W Soetaert Biotechnol Bioeng 2018 115 1195ndash1206 808

35 J Clark and F Deswarte Introduction to Chemicals from Biomass Second Edition 809

John Wiley amp Sons 2015 810

36 K V Sajna R Houmlfer R K Sukumaran L D Gottumukkala and A Pandey in 811

Industrial Biorefineries and White Biotechnology eds A Pandey R Houmlfer M 812

Taherzadeh K M Nampoothiri and C Larroche Amsterdam Oxford Waltham 813

Elsevier 2015 pp 499ndash521 814

37 R Marchant and I M Banat Trends Biotechnol 2012 30 558ndash565 815

38 K Hill and C LeHen-Ferrenbach in Sugar-Based Surfactants Fundamentals and 816

Applications ed C C Ruiz Boca Raton CRC Press 2009 pp 1ndash20 817

39 L Rodrigues I M Banat J Teixeira and R Oliveira J Antimicrob Chemother 818

2006 57 609ndash618 819

40 J D Desai and I M Banat Microbiol Mol Biol Rev 1997 61 47ndash64 820

41 S L K W Roelants K Ciesielska S L De Maeseneire H Moens B Everaert S 821

Verweire Q Denon B Vanlerberghe I N A Van Bogaert P Van der Meeren B 822

Devreese and W Soetaert Biotechnol Bioeng 2016 113 550ndash559 823

42 N Baccile M Selmane P Le Griel S Preacutevost J Perez C V Stevens E Delbeke S 824

Zibek M Guenther W Soetaert I N A Van Bogaert and S Roelants Langmuir 825

2016 32 6343ndash6359 826

32

43 N Baccile A-S Cuvier S Preacutevost C V Stevens E Delbeke J Berton W Soetaert 827

I N A Van Bogaert and S Roelants Langmuir 2016 32 10881ndash10894 828

44 N Baccile C Seyrig A Poirier S A Castro S L K W Roelants and S Abel Green 829

Chem 2021 23 3842ndash3944 830

45 A Salis L Cappai C Carucci D F Parsons and M Monduzzi J Phys Chem Lett 831

2020 11 6805ndash6811 832

46 J Schindelin I Arganda-Carreras E Frise V Kaynig M Longair T Pietzsch S 833

Preibisch C Rueden S Saalfeld B Schmid J-Y Tinevez D J White V 834

Hartenstein K Eliceiri P Tomancak and A Cardona Nat Methods 2012 9 676ndash835

682 836

47 T Parasassi and E Gratton J Fluoresc 1995 5 59ndash69 837

48 M M Dodes Traian F L Gonzaacutelez Flecha and V Levi J Lipid Res 2012 53 609ndash838

616 839

49 D M Owen C Rentero A Magenau A Abu-Siniyeh and K Gaus Nat Protoc 840

2012 7 24ndash35 841

50 W Stillwell in An Introduction to Biological Membranes Academic Press 2016 pp 842

135ndash180 843

51 W Stillwell in An Introduction to Biological Membranes Academic Press 2016 pp 844

181ndash220 845

52 B Tenchov Chem Phys Lipids 1991 57 165ndash177 846

53 R Koynova and M Caffrey Biochim Biophys Acta 1998 1376 91ndash145 847

54 F Schmid S Dolezel O Lenz and S Meinhardt J Phys Conf Ser 2014 487 848

012004 849

55 S Meinhardt R L C Vink and F Schmid Proc Natl Acad Sci U S A 2013 110 850

4476ndash4481 851

56 O Glatter and O Kratky Small Angle X-ray Scattering Academic Press London 852

1982 853

57 B A Cunningham A D Brown D H Wolfe W P Williams and A Brain Phys 854

Rev E 1998 58 3662ndash3672 855

58 M Nadler A Steiner T Dvir O Szekely P Szekely A Ginsburg R Asor R Resh 856

C Tamburu M Peres and U Raviv Soft Matter 2011 7 1512ndash1523 857

59 C Tanford The hydrophobic effect Formation of micelles and biological membranes 858

John Wiley amp Sons Inc New York 1973 859

60 M Masuda and T Shimizu Langmuir 2004 20 5969ndash5977 860

33

61 A Zidovska K K Ewert J Quispe B Carragher C S Potter and C R Safinya 861

Block Liposomes Vesicles of Charged Lipids with Distinctly Shaped Nanoscale 862

Sphere- Pear- Tube- or Rod-Segments Elsevier Inc 1st edn 2009 vol 465 863

62 A Iglic V Kralj-Iglic B Božič M Bobrowska-Haumlgerstrand B Isomaa and H 864

Haumlgerstrand Bioelectrochemistry 2000 52 203ndash211 865

63 T Shimizu M Masuda and H Minamikawa Chem Rev 2005 105 1401ndash1444 866

64 J V Selinger and J M Schnur Phys Rev Lett 1993 71 4091ndash4094 867

65 J M Schnur Science (80- ) 1993 262 1669ndash1676 868

66 B Tenchov R Koynova and G Rapp Biophys J 2001 80 1873ndash1890 869

67 R Koynova and B Tenchov Ser Biomech 2017 31 3ndash19 870

68 S J Angyal and R Le Fur Carbohydr Res 1984 126 15ndash26 871

69 J K Whitesell and M A Minton J Am Chem Soc 1987 109 225ndash228 872

70 T Takayama S Ando and I Ando J Mol Struct 1990 220 245ndash250 873

71 K Gawrisch N V Eldho and I V Polozov Chem Phys Lipids 2002 116 135ndash151 874

72 F Tielens C Gervais J F Lambert F Mauri and D Costa Chem Mater 2008 20 875

3336ndash3344 876

73 G A Jeffrey and Y Yeon Acta Crystallogr Sect B 1986 42 410ndash413 877

74 B Berglund and R W Vaughan J Chem Phys 1980 73 2037ndash2043 878

75 C Gervais C Coelho T Azaiumls J Maquet G Laurent F Pourpoint C Bonhomme 879

P Florian B Alonso G Guerrero P H Mutin and F Mauri J Magn Reson 2007 880

187 131ndash140 881

76 C Akita T Kawaguchi F Kaneko H Yamamoto and M Suzuki J Phys Chem B 882

2004 108 4862ndash4868 883

77 R N Purusottam L Seacutenicourt J J Lacapegravere and P Tekely Biochim Biophys Acta - 884

Biomembr 2015 1848 3134ndash3139 885

78 X Biarneacutes A Ardegravevol A Planas C Rovira A Laio and M Parrinello J Am Chem 886

Soc 2007 129 10686ndash10693 887

79 L M J Kroon‐Batenburg and J Kroon Biopolymers 1990 29 1243ndash1248 888

80 K N Kirschner and R J Woods Proc Natl Acad Sci U S A 2001 98 10541ndash889

10545 890

81 S Svenson B Kirste and J H Fuhrhop J Am Chem Soc 1994 116 11969ndash11975 891

82 I Sack S Macholl J H Fuhrhop and G Buntkowsky Phys Chem Chem Phys 892

2000 2 1781ndash1788 893

83 S Svenson J Koening and J H Fuhrhop J Phys Chem 1994 98 1022ndash1028 894

34

84 S Svenson A Schaefer and J H Fuhrhop J Chem Soc Perkin Trans 2 1994 2 895

1023ndash1028 896

85 K Bock and C Pedersen Adv Carbohydr Chem Biochem 1983 41 27ndash66 897

86 T B Grindley Structure and Conformation of Carbohydrates Springer Berlin 898

Heidelberg 2008 899

87 F Tantakitti J Boekhoven X Wang R V Kazantsev T Yu J Li E Zhuang R 900

Zandi J H Ortony C J Newcomb L C Palmer G S Shekhawat M O De La 901

Cruz G C Schatz and S I Stupp Nat Mater 2016 15 469ndash476 902

903

1

Supporting Information 904

905

Topological connection between vesicles and nanotubes in single-906

component lipid membranes driven by head-tail interactions 907

908

Niki Baccilea Ceacutedric Lorthioira Abdoul Aziz Baa Patrick Le Griela Cristina Coelhob Javier 909

Perezc Wim Soetaertd Sophie L K W Roelantsd 910

911

a Sorbonne Universiteacute Centre National de la Recherche Scientifique Laboratoire de Chimie de 912

la Matiegravere Condenseacutee de Paris LCMCP F-75005 Paris France 913

b Institut des Mateacuteriaux de Paris Centre Sorbonne Universiteacute Paris France 914

c Synchrotron Soleil LrsquoOrme des Merisiers Saint-Aubin BP48 91192 Gif-sur-Yvette Cedex 915

France 916

d InBio Department of Biotechnology Ghent University Ghent Belgium 917

918

919

Corresponding author 920

Dr Niki Baccile 921

E-mail address nikibaccilesorbonne-universitefr 922

Phone +33 1 44 27 56 77 923

924

925

926

2

a) 927

Parameter Method

Dry Matter (DM ) 996 Infrared balance 105degC

Glucose () 012 HPLC-Metacarb

Glycerol () nd HPLC-Metacarb

Free Fatty Acid content () lt001 Internal method BBEPP

GC with FID Detector

Oil () lt001 Internal method BBEPP

GC with FID Detector

Moulds (CFUg) lt10 3M-Nordval ndeg16 B

Yeasts (CFUg) lt10 3M-Nordval ndeg16 B

Anaerobic count (CFUg) lt10 SP-VG M005 B

Protein () tbd BCA protein assay

928

Figure S 1 - a) HPLC-ELSD chromatogram and composition table 929

930

931

932

933

3

b) 934

935

Group Ndeg δ ppm

CH2 1 354

CH2 2 153

CH2 3-712-14 132

CH2 811 203

CH 910 534

CH2 15 162

CH2 16 388

CH 1 424

CH 2 316

CH 3 334

CH 4 329

CH 5 354

CH2 6 366

Figure S 1 (next) ndash b) 1H NMR spectrum (MeOD-d4) recorded on the purified G-C181-OH compounds 936

and related assignment 937

938

2

811910 1rsquo

5rsquo1

3rsquo4rsquo

1rsquo1910

asymp asymp asympHDO MeOD Acetone

6rsquo2rsquo

1615

3-712-14

4

939

940

Figure S 2 ndash Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 130degC 941

5

942

943

Figure S 3 - Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 90degC 944

945

90degC

1

22

6

946

947

948

949

Figure S 4 ndash Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 70degC 950

951

21

11

1

2

5

3

4

45

5

6

7

7

7

952

953

954

Figure S 5 ndash Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 25degC 955

956

957

1

1 1

8

958

Figure S 6 ndash a) SANS experiment recorded at 90degC for a 5 mgmL G-C181-OH aqueous solution b-c) 959

WAXS experiments recorded on aqueous solutions of G-C181-OH at b) 5 mgmL and c) 25 mgmL d) 960

Structural interpretation of the repeating distances pointed out in c) 961

962

963

964

001 01 1

q Aring-1

25degC

68degC

88degC

111degC

01 1q Aring

-1

02 04 06 08 1q Aring

-1

001 01

10-1

100

101

Inte

nsity

cm

-1

q Aring-1

-2

T= 90degC

SANS WAXS WAXS

a) b) c)

dv

dL d= 144 Aring-1

di

ddH2O

dt

dvdL

dtdi d

d)

9

965

966

Figure S 7 ndash ssNMR MAS (10 kHz) experiments recorded on a G-C181-OH dry powder a) 1H spectra 967

recorded from 110degC to 70degC b) 13C single pulse (SP) spectra recorded with high power 1H decoupling from 968

110degC to 70degC c) 13C cross polarization (CP) spectra recorded at 110degC and contact time tc of 50 micros and 969

3000 micros d) 13C CP spectra recorded at 70degC and for 20 lt tc micros lt 3000 970

971

1

811

910

CH2 CH2

1rsquo

6rsquo

8 6 4 2 0 ppm

110

90

80

70

910

1rsquo

6rsquo16

1

2rsquo3rsquo4rsquo

811

OH

45 40 35 30 25 20 15

ppm

200 180 160 140 120 100 80 60 40 20 0

ppm140 120 100 80

ppm

CH2-gauche

3004

110

90

80

70

65

60

50

25

C91013003

C9 - 12869

CH2-all-trans

3198

C1rsquo10335

a) b)

1H

T degCT degC

13C Single Pulse MAS

A

B

A

B

70degC70degC

5rsquo

215

200 180 160 140 120 100 80 60 40 20 0

ppm

13C CPMAST= 110degC

200 180 160 140 120 100 80 60 40 20 0

ppm

132 131 130 129 128 127 126

ppm

106 105 104 103 102

ppm

13C CPMAST= 70degC

50

tc micros

3000

20

tc micros

50 500 1000 3000

tc micros

C102

13024 C910113003

C92

12869

C1rsquo213082

C1rsquo110443

A B

106 105 104 103 102

ppm

3000

1000

500

50

20

AB

c)

d)

10

972

Figure S 8 ndash a) Short and d) long T2 components displayed by the 1H transverse relaxation signal c) 1H 973

transverse relaxation signals recorded on a G-C181-OH dry powder from 110degC to 70degC The area of each 974

given resonance is plotted (log-lin) as a function of the echo time (CPMG pulse sequence) for each 975

temperature and fitted using the equation given above T2s and T2l indicate the short and long T2 relaxation 976

times respectively obtained by describing the 1H transverse relaxation signal using a biexponential decay 977

For H1rsquo H5rsquo and H6rsquo the 1H relaxation function could be described by a monoexponential decay 978

- Protons from the sugar group (H1rsquo H5rsquo H6rsquo) show a single T2 relaxation component (Figure 979

6c and Figure S 8) while protons of the acyl chain (CH2 H910 H811) show both short (Figure 980

6c) and long (Figure 6d) T2 relaxation component 981

- The short T2 relaxation time systematically decreases from 110degC to 70degC for all species by a 982

factor ranging from two to seven depending on the specific chemical group Very long T2 are 983

observed at 110degC for all acyl species while nearly unchanged values are measured below 984

90degC without any clear trend 985

- The sugar protons (H1rsquo H5rsquo H6rsquo) have the shortest set of T2 values (lt 1 ms) below 100degC 986

The long component of T2 generally ranges between 4 ms and 12 ms 987

0 1 2 3 4 5 6 7 8

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1 2 3 4 5 6 7 8

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

H1

H8H11

H9H10

CH2

H1rsquo

H6rsquo

CH2

H910H811 H1CH2

H1rsquo H5rsquo6rsquo

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

OH

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45 T= 110degC

T= 90degC

T= 80degC

T= 70degC

Log

In

ten

sity

Time ms

Fit

)

CH2 H910 H811 H1 H56 H1 OH

00

05

10

15

20

25

30

35

40

T2 m

s

T= 110degC

T= 90degC

T= 80degC

T= 70degC

CH2 H910 H811 H1

0

2

4

6

8

10

12

T2 m

s

T= 110degC

T= 90degC

T= 80degC

T= 70degC

Short T2(1H) Long T2(1H)

a) b)

c)

11

Table S 1 ndash 1H chemical shifts of G-C181-OH in solution and in the solid-state Solid-state spectra are 988

recorded at a MAS frequency of 10 kHz in DMSO-d6 broad peak 989

990

δH ppm

Group Ndeg Liquid

(MeOD-d4) Solid

(T= 110degC) Solid

(T= 70degC)

CH2 H1 354 357

CH2 H2 153 156

CH2 H3-712-14 132 135

CH2 H811 203 203

CH H910 534 534

CH2 H15 162 164

CH2 H16 388 383

CH H1 424 432

CH H2 316 335

CH H3 334 335

CH H4 329 335

CH H5 354 351 357

CH2 H6 366 383

OH OH1 432

45 463

OH OH2 492

OH OH3 490

OH OH4 487

OH OH6 445

991

992

H1H9H10H1rsquo

H6rsquo

OH2rsquo

OH1

12

Detailed discussion about the NOESY and DQ-SQ 2D 1H-1H ssNMR experiments presented in 993

Figure 9 994

995

2D 1H-1H ssNMR experiments are a unique tool to explore through-space proximities 996

between 1H nuclei ssNMR offers a large number of possibilities each adapted to a given 997

system The advantage of 2D maps with respect to a dynamics approach is their direct 998

readability in clear two nuclei which exchange their magnetization through-space (whichever 999

mechanism may be responsible for this exchange) will display a specific cross-peak However 1000

the way cross-peaks are presented and interpreted strictly depends on the nature of the pulse 1001

program which in turn is chosen according to the physical nature of the sample 1002

1D temperature-dependent 1H NMR experiments performed on G-C181-OH (Figure 1003

4a in the main text) indicate sharp peaks at 90degC and broad peaks at 70degC Narrow peaks 1004

indicate fast andor large-amplitude molecular tumbling averaging the strong 1H-1H dipolar 1005

interactions This excludes the use of pulse sequences like DQ-SQ which is not efficient under 1006

these conditions For this reason we employ spin-diffusion-based pulse sequences like 1007

NOESY which is more adapted for samples with high molecular mobility When mixing times 1008

are chosen to be short (lt 1 s) one generally explores through-space proximities within about 5 1009

10 Aring On the contrary when molecular motions are restricted dipolar coupling becomes 1010

significant 1H NMR peaks are broader and pulse sequences like DQ-SQ can now be used 1011

with a spatial resolution of less than 5 Aring 1012

Whenever possible DQ-SQ experiments are preferred over spin diffusion-based pulse 1013

sequences because they are more selective in particular for likewise interactions which are 1014

not discriminated in NOESY-type experiments This is illustrated by the typical interaction 1015

schemes for NOESY (top Figure 9a) and DQ-SQ (top Figure 9b) For two given A and B 1016

protons 2D NOESY maps will display two on-diagonal and two square-shaped symmetrical 1017

off-diagonal cross-peaks Off-diagonal cross-peaks illustrate the mutual A-B and B-A 1018

interaction but the on-diagonal peaks are systematic and do not prove any A-A or B-B 1019

interaction Similarly to NOESY 2D DQ-SQ maps display off-diagonal peaks (symmetric 1020

across the diagonal top Figure 9b) upon A-B and B-A interaction but in contrast they will 1021

display on-diagonal peaks only in the presence of self-interaction In the tutorial example of 1022

Figure 9b top the on-diagonal A-A peak indicates mutual A-A interaction while lack of on-1023

diagonal B-B peak evidences the absence of B-B proximities Combining NOESY and DQ-SQ 1024

experiments is capital to study the local arrangement of G-C181-OH and its evolution with 1025

temperature 1026

13

At T= 90degC the H1 and H5rsquo contributions are very close but possible to discriminate 1027

them (δ(H1)= 354 ppm δ(H5rsquo)= 351 ppm) according to 1H NMR chemical shifts in Table S 1028

1 2D 1H-1H NOESY in Figure 9a also discriminates between the two components attributed 1029

on the basis of the interactions with their mutual neighbors (green H1-H2 and H5rsquo-H1rsquo squares 1030

in Figure 9a) On this basis 2D NOESY does not show a direct H1rsquo-H1 correlation (blue red-1031

crossed) but rather an extensive network of intra-sugar interactions These features exclude 1032

suggest intra- and inter-sugar interactions and they exclude head-tail arrangements as a major 1033

molecular configuration at 90degC 1034

At 70degC the 1H NMR peaks are broader indicating a reduction of the T2 (1H) relaxation 1035

times andor residual homonuclear dipolar interactions excluding the use of NOESY In this 1036

case 1H-1H DQ-SQ correlation experiments are best suited Figure 9b presents the 1H-1H DQ-1037

SQ experiment on G-C181-OH at 70degC Peaks are broader than in NOESY and attribution may 1038

be cumbersome However the experiment is performed at 700 MHz and MAS= 22 kHz which 1039

improves spectral resolution with respect to the same experiment performed at 300 MHz and 1040

MAS= 10 kHz The auto-peak at δH= 356 ppm in the SQ dimension is attributed to H1-H1 spin 1041

pairs mainly resulting from intra CH2 interaction while the DQ cross-peaks at δH= 156-356 1042

ppm and δH= 432-356 ppm respectively origin from H2-H1 and H1rsquo-H1 interactions (Figure 1043

9b) DQ-SQ map does not show any auto-peak corresponding to H1rsquo-H1rsquo or Hxrsquo-Hxrsquo (x= 2rsquo 1044

3rsquo 4rsquo) interactions (red-crossed blue Figure 9b) thus excluding any proximity (lt 5 Aring) between 1045

close glucose neighbors On the contrary DQ-SQ maps (recorded both at 700 MHz Figure 9b 1046

or at 300 MHz not shown) systematically show DQ cross-peaks which can only be attributed 1047

to H1rsquo-H1 interactions (δH= 432-356) The presence of these H1rsquo-H1 interactions can only be 1048

explained by a head-tail arrangement of the G-C181-OH molecules as illustrated below the 1049

DQ-SQ map in Figure 9b 1050

9

249

Figure 2 ndash a) DSC thermogram of G-C181-OH under dry (04 wt water estimated by thermogravimetric 250

analysis) and hydrated conditions recorded at a heating rate of 10degCmin Red marks indicate the 251

temperatures at which cryo-TEM experiments are performed on the corresponding 5 mgmL aqueous 252

solutions b) Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 130degC 253

90degC 70degC and 25degC T indicates the temperature of the sample solution before cryofixation Tne T T 254

being the temperature of the sample upon plunging in liquid ethane Please refer to the materials and 255

methods section for more information 256

T= 130degC

T= 90degC

T= 70degC

T= 25degC

a)

b)

20 40 60 80 100 120

99 wt

50 wt

25 wt

T degC

En

do

10 wt

Tm1= 483 degC

Tm2= 957 degC

10

257

The series of cryo-TEM images in Figure 2b show an overview of the most relevant 258

structures observed on G-C181-OH solution from T= 130degC to T= 25degC whereas T is 259

indicative of the equilibrium temperature before cryofixation (refer to materials and methods 260

section for more information on the sense and interpretation of T) At Tgt 100degC the sample 261

is essentially composed of polydisperse single wall vesicles massively shown by the 262

complementary pictures presented in Figure S 2 In between Tm1 and Tm2 at T= 90degC and T= 263

70degC (Figure 2b Figure 3 Figure 4 complemented by Figure S 3 and Figure S 4) vesicles are 264

connected by nanotubes At Tlt Tm1 flat crystals are otherwise observed (Figure 2b and Figure 265

S 5) 266

267

268

Figure 3 ndash Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 90degC 269

270

A deeper insight at T= 90degC is given in Figure 3 (more images are provided in Figure 271

S 3) Arrows 1 through 6 mark very clearly the typical continuity in the glucolipid membrane 272

between the vesicle and nanotubes Arrow Ndeg 7 shows that a single nanotube connects the outer 273

membrane of two vesicles several hundred nanometers apart The connection is shown by 274

2

34

5

6

6

1

5

7

A

B

D

C

0 10 20 30 40 50 60

BIn

tensity

au

Profile nm

A

0 10 20 30 40 50 60

Inte

nsity

au

Profile nm

C

D

T= 90degC

11

arrows Ndeg5 and Ndeg6 The corresponding profiles A through D confirm the tubular nature of 275

the elongated objects Profiles B and D are both characterized by typical symmetrical thick 276

walls (low intensity depths highlighted by shaded rectangles) surrounding a hollow region (high 277

intensity peak between the shaded rectangles) The wall thickness varies between 35 and 40 278

nm for profile B and between 41 and 61 nm for profile D while the total nanotube diameter 279

varies between 14 nm at B and 20 nm at D At the same time the thickness of the vesicle wall 280

is about 6 nm at A and 135 nm at C Considerations about the bilayer or interdigitated structure 281

of the membrane will be presented below 282

The connection between nanotubes and vesicles seems to be multiple For instance 283

arrow Ndeg1 (Figure S 3) and Ndeg4 (Figure 3) show a single-wall vesicle of which the membrane 284

is extruded into a nanotube Similarly a closer look at arrows Ndeg1 2 and 3 in Figure 3 show 285

that the vesicle is single-wall and fully closed while the nanotube nucleates at the vesicle outer 286

surface with no apparent topological continuity In many regions of the sample at T= 90degC 287

one can find exvaginations or buds232531 in the nanotube wall This is shown for instance by 288

arrows Ndeg7 and particularly put in evidence in Figure 4 by arrows Ndeg1 through Ndeg4 where the 289

amplitude of the oscillation varies between 20-25 nm (arrow Ndeg1) and 6-8 nm (arrows Ndeg3-4) 290

291

Figure 4 - Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heated at T= 90degC 292

293

Still between Tm2 and Tm1 but at lower temperature (T= 70degC) the nanotubes seem to 294

vary from single threads (arrow Ndeg1 Figure S 4) to nanotubes of diameter below 15 nm both 295

connecting nanotubes of diameter ranging between 50 nm and 100 nm (arrows Ndeg2 Figure S 296

4) The larger tubes are most likely formed by the inflation and flattening of the nanotubes 297

probably after assimilation of surrounding vesicles and fusion of the exvagination discussed 298

above The transition between vesicles tubes and flat membranes is certainly more accentuated 299

Arrow Ndeg3 in Figure S 4 shows distinct vesicles while arrow Ndeg4 identifies a flat structure 300

evidently formed from fused vesicles In particular arrow Ndeg5 points at the fusion of two 301

2

1 132

4

12

vesicles while arrow Ndeg6 points at two vesicles fusing into the flat membrane Additional areas 302

where vesicle fusion is prominent are indicated by the oscillations at the outer skirt of broad 303

lamellae and pointed by arrow Ndeg7 304

Finally below Tm1 at T= 25degC for instance mainly flat crystals with no visible 305

exvaginations or tubes are visible (Figure S 5) by cryo-TEM and the material turns into an 306

insoluble precipitate Massive presence of flat crystalline structures is also clear from the bulk 307

iridescent behavior of a G-C181-OH solution under manual shear 308

In summary despite the uncertainty related to the actual state of equilibrium of the 309

glucolipids at the moment of cryofixation cryo-TEM depicts a vesicle phase at Tgt 100degC a 310

flat lamellar phase below Tm1 and an interconnected vesicle-nanotube network between Tm1 and 311

about 100degC The combination of statistically-relevant fluorescence microscopy and 312

scatteringdiffraction experiments provides an additional proof of the coexistence of 313

vesiclenanotube systems between the two transition temperatures Please note that the 314

experiments reported below have been performed under controlled conditions of temperature 315

which is then indicated by the use of T instead of T 316

13

317

Figure 5 ndash a-c) Confocal fluorescence microscopy images recorded at 90degC and 70degC for a 5 mgmL G-318

C181-OH aqueous solution d-g) Typical fluorescence microscopy images recorded at T= 80degC using 319

Laurdan (refer to materials and methods section for more details on this fluorophore) Excitation 320

wavelength is 405 nm for all the images while emission wavelength is 490 nm in (df) and 440 nm in (eg) 321

h) Full emission spectra obtained from the integration of various fluorescence images recorded from 412 322

nm to 600 nm (step 5 nm) at 90degC 80degC and 70degC i) Box plot of GP defined as 119918119927 =(120640120786120786120782minus120640120786120791120782)

120640120786120786120782+120640120786120791120782 with 120640120786120786120782 323

and 120640120786120791120782 being the emission wavelengths at 440 nm and 490 nm at 90degC 80degC and 70degC Each data point 324

l= 490 nm l= 440 nm

l= 490 nm l= 440 nm

10 μm400 450 500 550 600

00

02

04

06

08

10

No

rma

lize

d I

nte

nsity

l nm

90degC

80degC

80degC

70degC

70degC

400 450 500 550 60000

02

04

06

08

10

No

rma

lize

d I

nte

nsity

l nm

90degC

80degC

80degC

70degC

70degC

d) e)

f) g)

h)

i)

T= 80degC

Laurdan

T= 80degC

T= 80degC T= 80degC

T= 90degC T= 70degC T= 70degC

a) b) c)

1

1

1

14

corresponds to a set of regions of interest (ROIs) (54 at 90degC 65 at 80degC and 75 at 70degC) selected in the 325

corresponding fluorescence images ROIs were representatively selected so to include the entire image but 326

also vesicles alone tubes alone and mixtures of vesicles and tubes 327

328

Figure 5ab shows two typical confocal fluorescence microscopy images recorded on a 329

G-C181-OH solution (5 mgmL) in a flame-sealed capillary heat at T= 90degC and T= 70degC 330

Although the resolution of optical microscopy is not comparable with cryo-TEM the sample is 331

broadly composed of vesicles coexisting with nanotubes whereas in some cases the nanotube 332

starts from the vesicle itself (arrow Ndeg1) in good agreement with cryo-TEM data Furthermore 333

SANS recorded at T= 90degC (Figure S 6a) shows a scattering profile with a pronounced -2 334

dependence of the scattering intensity against the wavevector q in the log-log scale This 335

indicates the massive presence of a flat interface and it is expected typical of bilayers in 336

vesicular morphologies56 However the typical signature of nanotubes cannot be observed in 337

the SANS profile and this can be explained by both an argument related to the relative lower 338

amount of nanotubes with respect to vesicles at 90degC but also to the steeper slope of flat 339

morphologies (-2) compared to tubes (-1) the signal of the former masking the tube signal at 340

low wavevector values The SANS profile is characterized by a diffraction peak at q= 0167 Aring-341

1 of which the assignment is refined by WAXS data 342

Temperature-driven WAXS experiments in Figure S 6bc (5 mgmL-1 and 25 mgmL-1 343

respectively) show a lamellar order possibly attributed to a multilamellar structure of the 344

vesicles (dv Figure S 6d) at high temperature and to the interplanar arrangement of the 345

crystalline flat structures (dL Figure S 6d) at T= 25degC The sharp (001) reflection indicates 346

typical interlamellar distance of 353 Aring at 25degC although this value varies with temperature 347

(400 Aring at 111degC 372 Aring at 88degC) thus confirming its assignment to a hydration-dependent 348

interlamellar order More intriguing is the second broader peak at q= 0403 Aring-1 (d= 156 Aring) 349

and its reflection both clearly visible at 88degC below Tm2 We attribute it to the inter-glucolipid 350

arrangement inside the tubular phase (dt Figure S 6d) coexisting with the vesicle phase 351

between Tm1 and Tm2 and to the planar lipid arrangement (di Figure S 6d) in the rigid flat 352

lamellar structure below Tm2 Its temperature- hence hydration- independent position also 353

confirms its structural origin Finally besides the water-structure peak at about 2 Aring-1 we also 354

observe a broad peak at 144 Aring-1 characteristics of the lateral packing (d Figure S 6d) of lipids 355

in bilayers While its position could be consistent with an Lα phase5758 above Tm2 its broadness 356

and positional invariance as well as the coexistence between vesicles and tubes do not allow a 357

precise assignment 358

15

Does G-C181-OH assume a bilayer or interdigitated structure the latter being the one 359

found for other similar bolaform glucolipids4243 The size of G-C181-OH can be estimated to 360

be at the most 32 Aring given by the length L of the C16 tail (L = 22 Aring from the Tanford formula 361

L= 154 + 1265 times n (n= 16)59) and the sugar (about 10 Aring)4243 However the cis conformation 362

of the C=C bond imposes a ldquokinkrdquo of 60deg in the aliphatic chain which reduces the effective 363

length of the oleic backbone to about 17 Aring for an effective total G-C181-OH length of about 364

27 Aring Cryo-TEM provides a tube thickness varying from 35 Aring to 60 Aring and a vesicle membrane 365

thickness of at least 60 Aring WAXS suggests an interlamellar thickness (membrane and water) 366

between 40 Aring and 37 Aring and an inter-lipid distance in the tube of about 16 Aring In the absence of 367

a clear form factor oscillation profile as found for similar glucolipids4243 one must 368

acknowledge that these data pledge neither in favor of fully interdigitated nor a double layer 369

structure and might suggest a tilted lipid arrangement with partial interdigitation60 370

371

13C ssNMR demonstrates that G-C181-OH can assume two coexisting head-head and 372

head-tail configurations 373

Vesicles are characterized by low positive mean and gaussian curvatures while 374

nanotubes are described by a combination of a high positive mean and zero gaussian curvatures 375

In this regard the coexistence of vesicles and nanotubes requires glucolipids to adapt 376

themselves to both environments at the same time For this reason the formation of nanotube-377

vesicle networks generally requires a specific driving force able to impose a high mean and 378

zero gaussian curvature to the membrane In many systems insertion of at least a second 379

constituent (lipid surfactant protein) in the membrane becomes a necessary internal driving 380

force for the nucleation and growth of so-called membrane inclusions from which 381

topologically-connected nanotubes can form10ndash123261 as explained by orientational ordering 382

arguments62 It has also been shown that external driving forces such as use of laser 383

ldquotweezersrdquo30 osmotic pressure6 electroformation5ndash82829 or controlled suction13ndash16 are 384

otherwise necessary as also supported by theoretical studies19 385

The bending energy of tubes is higher than that of a vesicle membrane8 due to the 386

dependence of the membrane free energy from the principal curvatures For this reason 387

isotropic (rotational symmetry along the main axis) but also anisotropic33 lipids generally 388

stabilize low-curvature shapes like flat membranes or vesicles Lipid nanotubes are also a 389

common phase formed by amphiphiles but generally for lipids with a certain complexity in 390

their molecular structures as exampled by diacetylenic derivatives of phospholipids among 391

others262763ndash65 Negative contributions to the bending energy favoring the simultaneous 392

16

presence of interconnected vesicles and tubes was shown to be driven by the presence of 393

anisotropic lipids in a membrane continuum Spontaneous deviations to the elastic theory in the 394

presence of surfactants lipids and proteins with anisotropic structures favor the formation of 395

protrusions with radius of the order of the fraction of a micron9ndash1219ndash2362 396

In the absence of both internal and external driving forces mentioned above G-C181-397

OH should behave as a standard lipid Its DSC profile (Figure 2)536667 under both hydrated and 398

dry conditions identifies a hydration-independent transition (Tm1) slightly above room 399

temperature this is followed by one main transition at higher temperature (Tm2) in the dry 400

sample50 For such a profile one expects a gel phase (Lβ) below Tm1 and a liquid crystalline 401

structure possibly in the morphology of vesicles (L4 phase) possibly above Tm1 and most likely 402

above Tm231 as experimentally found On the basis of the thermogram recorded on the dry 403

powder one may also expect a ldquoripplerdquo (Pβ) phase between Tm1 and Tm2 characterized by 404

periodic undulations of the bilayer (generally addressed to as regions with both liquid order and 405

disorder)5455 406

In the present work instead of a Lβ phase we observe flat crystals below Tm1 and a 407

vesicle-nanotube network instead of the ldquoripplerdquo phase The latter is particularly unexpected 408

because the membrane is composed of a single-component (G-C181-OH) and no anisotropic 409

external force (osmotic hydrodynamic electric) is used to pull and stabilize the tubes In 410

addition the nanotubes are observed at fairly high temperatures above Tm1 (between about 411

60degC and 90degC) while this morphology is generally favored at low temperature (proportionality 412

with 1kT figure 5 in Ref 62) commonly below a given transition temperature characteristics 413

for each specific amphiphile27 Furthermore nanotubes of G-C181-OH seem to be stable in 414

time thus differing from what is found in electroformed vesicles78 They are also observed 415

away from phase transition events (may them be at Tm1 or Tm2) as otherwise reported before 416

for a diacetylenic derivative of phosphatidylcholine26 and many other systems27 417

A stable nanotube-vesicle network for a single-component phospholipid in the absence 418

of external driving forces is not expected and has never been reported before It could be 419

explained by the coexistence of two micro-separated phases of different rigidity This 420

hypothesis is excluded by confocal microscopy on Laurdan-labelled membranes and magic 421

angle spinning (MAS) solid-state NMR (ssNMR) experiments 422

Laurdan is a well-known fluorophore used to put in evidence the coexistence of rigid 423

and fluid domains in both model and biological membranes47ndash49 Laurdan was used here to label 424

the G-C181-OH membranes between Tm1 and Tm2 and to verify the possible micro-segregation 425

and coexistence of rigid (eg Lβ λem= 440 nm) and fluid (eg Lα λem= 490 nm) domains in 426

17

tubes and vesicles respectively A set of quantitative experiments performed on a 5 mgmL-1 427

solution summarized in Figure 5 shows an excess of fluorescence at only λem= 490 nm for 428

temperature between 90degC and 70degC (Figure 5h) The GP value which quantifies the rigidfluid 429

character of the membrane is systematically negative (Figure 5i) indicating an overall fluid 430

membrane at all temperatures GP is quantified on more than 50 regions of interest (ROIs) at 431

each temperature including vesicles and tubes alone as well as connected nanotube-vesicles 432

(Figure 5d-g) The dispersion of GP at 80degC and 70degC between -08 and -02 could suggest the 433

spurious presence of membranes with various degrees of flexibility although fluidity is always 434

predominant The combination of WAXS (broad peak d= 144 Aring-1 invariant with 435

temperature Figure S 6bc) and confocal fluorescence microscopy using Laurdan confirms the 436

single-phase of the membrane composed of G-C181-OH both in vesicles and nanotubes This 437

is also consistent with the lack of phase transitions in DSC experiments between ~85degC and 438

~60degC especially in hydrated samples We conclude that the vesicle-nanotube transition is not 439

driven by the local segregation of gel phase (Lβ) microdomains nor that the nanotubes are 440

composed of such phase 441

ssNMR is the best-suited approach to investigate the origin of the local driving force for 442

the nucleation of anisotropic inclusions of which the average orientation depends on the local 443

membrane and inclusion curvatures and was shown to promote tubing62 ssNMR can provide 444

information on the local molecular mobility eg through the nuclear spin relaxation behavior 445

but also on intermolecular interactions and stereochemistry eg through variations in the 446

chemical shift68ndash70 447

Ideally ssNMR should be performed on diluted G-C181-OH samples above Tm1 448

However fast MAS necessary for high resolution centrifuges the sample As a first 449

approximation hydrated G-C181-OH samples for instance at 50 wt as classically done for 450

phospholipids71 could be studied Unfortunately permeability of the sample holder combined 451

to heating (close to the boiling point of water) and MAS provoke dehydration and even sample 452

ejection After experiencing both of the above this approach was abandoned As a second 453

approximation the study of the dehydrated G-C181-OH powder avoids centrifugation and 454

evaporation If studying dried lipids may not reflect their exact behaviour under diluted 455

conditions valuable structural information can still be obtained in dehydrated molecular 456

systems like proteins thus illustrating that this approach is not unreasonable Furthermore if 457

DSC (Figure 2a) shows that hydration suppresses the Tm2 it also shows that the value of Tm1 is 458

not affected by water at concentrations as low as 10 wt For this reason performing a 459

structural analysis of dehydrated G-C181-OH by ssNMR above Tm1 may help understanding 460

18

the local arrangement of G-C181-OH in two morphologies vesicles and nanotubes having 461

sensibly different curvatures 462

To explore the local mobility of the most relevant chemical groups of G-C181-OH 463

above Tm2 (T= 110degC) and between Tm2 and Tm1 (70 lt T degC lt 90) we combine a series of 1H 464

and 13C experiments Single-pulse (SP) MAS experiments (1H and 13C) when recorded under 465

quantitative conditions allow the detection of all chemical species while cross-polarization 466

(CP) MAS (13C only) experiments are sensitive only to constrained groups for which the 1H-467

13C dipolar coupling is not averaged to zero by the molecular motions 468

469

Figure 6 ndash a) 1H MAS (10 kHz) ssNMR spectra recorded from 110degC to 70degC b) 13C MAS (10 kHz) ssNMR 470

spectra recorded using high power decoupling single pulse (SP) and cross-polarization (CP) pulse sequences 471

Contact time for CP experiments and corresponding temperatures are given on each spectrum Full spectra 472

are given in Figure S 7 473

The most relevant 1H and 13C spectra are given in Figure 6 (full range in Figure 7 and 474

Figure S 7) As expected the 1H solid-state SPMAS spectra show broader peaks compared to 475

the solution-state NMR spectrum (Figure 6a) reflecting the obvious solid-state character of the 476

sample at any temperature The sugar group and the C=C bond resonate in the 25-6 ppm range 477

and 535 ppm respectively while the OH groups of the sugar moiety are found below 45 ppm 478

(Figure 6a full assignment in Table S 1) Very broad (~1 ppm) at 100degC the position of the 479

OH peak varies with the temperature value and becomes narrower (lt~05 ppm) and visible at 480

δH= 464 ppm at T= 70degC Line narrowing and chemical shift variations of the OH resonance 481

in ssNMR is always a direct sign of H-bonding interactions whereas the strength of the H-482

bonding hence its OHhellipH distance is correlated to a dispersion in chemical shifts72ndash75 All other 483

60 55 50 45 40 35 30 25

ppm

110

90

80

70

910

1rsquo

6rsquo16

1

OH

134 132 130 128 126

ppm110 108 106 104 102 100

ppm

SP (T= 110degC)

SP (T= 70degC)

CP 50 ms (T= 70degC)

CP 3 ms (T= 70degC)

910 1rsquo

1H 13C

1

811

910

CH2 CH2

1rsquo

6rsquo

a) b)

C102

13024

C9101

13003

C92

12869 C1rsquo210382

C1rsquo110443

C1rsquo13035

2rsquo3rsquo4rsquo

T degC

5rsquo

19

resonances broaden from 110degC to 70degC thus indicating slower andor more anisotropic 484

reorientational motions of all chemical groups The specific dynamic behaviour of the sugar 485

headgroup is also reflected by its 1H transverse relaxation time T2 whereas the shorter the T2 486

the higher the rigidity Experiments performed between 110degC and 70degC shown and 487

commented in Figure S 8 in the Supporting Information globally show a single short (lt 1 ms 488

below 100degC) T2 relaxation component for H1rsquo H5rsquo and H6rsquo and two both short and long 489

(between 4 ms and 12 ms) T2 for the acyl chain (CH2 H910 H811) even below 100degC 490

At 110degC 13C SPMAS experiments (Figure 6b full range in Figure 7 and Figure S 7 491

assignments in Table 1) show a single peak at δC= 10335 ppm for C1rsquo and at δC= 13003 ppm 492

for C910 as expected for a symmetrical conformation with two chemically-equivalent 493

carbons76 At 70degC the C1rsquo peak disappears while C910 now shows two peaks around 128 494

ppm and 130 ppm (Table 1 full temperature series in Figure S 7b) attributed to C92 (δC= 12869 495

ppm) and C10 in a non-symmetric configuration respectively for which these two carbons are 496

chemically inequivalent76 More precisely the resonance at 130 ppm is actually the 497

superposition of two close resonances C9101 (δC= 13003 ppm) and C102 (δC= 13024 ppm) 498

where subscripts 1 and 2 stand for two different conformations of the cis-olefin group as 499

reported for various phases of oleic acid in the vicinity of 0degC76 500

501

Table 1 ndash 13C Chemical shifts of G-C181-OH in solution and solid-state Solid-state experiments are 502

recorded at MAS= 10 kHz All spectra are reported in Figure 7 and Figure S 7 The contact time for cross 503

polarization is 3 ms In all ssNMR experiments temperature is first raised at 110degC and then lowered to 504

either 90degC or 70degC Attribution of the 13C resonances in Table 1 is performed on the basis of the solution 505

spectrum (Figure S 1b) Additional peaks observed in both single pulse and cross polarization ssNMR 506

experiments are attributed when possible to a configuration effect of the given atom eg C9101 and C92 507

indicate two configurations of C9 All new peaks observed in solid-state and not detected in solution are 508

highlighted in red for convenience 509

510

511

δ ppm

Peak Liquid (DMSO-d6)

Solid Single Pulse

T= 110degC

Solid Single Pulse

T= 90degC

Solid Single Pulse

T= 70degC

Solid Cross Polarization

T= 70degC

C102

- - 13024 13024

C910C9101 13009 13003 13003 13003 13003

C1C9C10C1rsquo

C6rsquo

20

C92 - - - 12869 12869

C1rsquoC11 10329 10335 10335 - 10443

C12 - - - - 10382

C3 7724 7709 7702 - 7829

- 7634 7629 - 7621

7525

C2 7388 7398 7390 7370 7370

C4 7052 7077 7054 7041 7098

C16 6900 7016 - - 7029

- - - - 6793

C1C5C6 6153 6241 6236 6239 6262

C1C5C6 6117 - - - 5956

C2 3300 3290 3290 3283 3279

CH2(trans) - - - 3198 3196

CH2(gauche)C15 2941 2981 2987 3004 -

C811 2706 2747 2747 2832 2817

CH2 2597 2625 2628 2630 2817

512

13C CPMAS spectra at 70degC (Figure 6b full range in Figure 7 and Figure S 7d) at 513

increasing contact times confirm the presence of two cis-olefin environments one for which 514

C910 are equivalent labelled here C9101 and one for which they are inequivalent labelled 515

C92 C102 Both environments are rigid enough to be observed at short contact time (tc= 50 μs) 516

A similar double environment is also observed for the sugar C1rsquo CPMAS at 70degC shows two 517

peaks (C1rsquo1 at δC= 10443 ppm C1rsquo2 at δC= 10382 ppm Figure 6b Figure S 7d) These are 518

significantly shifted downfield from the C1rsquo peak observed at 110degC and are detected at contact 519

times as short as 50 μs thus indicating their rigid character One also finds that CPMAS is 520

totally inefficient at 100degC for short contact times (50 μs) and poorly efficient even at 3 ms 521

typical of fast-tumbling molecules (Figure S 7c) also both gauche (δC= 3004 ppm) and all-522

trans (δC= 3198 ppm) conformations of the CH2 carbons coexist at 70degC (Figure S 7b)77 523

According to ssNMR one main mobile molecular environment exists at 110degC as 524

shown by the well-resolved 1H spectrum single C=C (C9101) and anomeric (C1rsquo) resonances 525

gauche conformation of the aliphatic CH2 and single T2 of the aliphatic CH2 chain and sugar 526

protons The sugar group is certainly mobile (inefficient 1H13C CPMAS) but still more 527

constrainted than the acyl chain (shorter T2 (1H)) As the temperature is lowered the overall 528

rigidity increases (broadening of 1H NMR peaks faster 1H transverse relaxation and efficient 529

1H13C magnetization transfer during CPMAS) presumably driven by H-bonding 530

interactions (downfield shift and line narrowing of the OH 1H signal very short OH T2 loss of 531

the C1rsquo signal in SPMAS) At the same time a number of spectroscopic data (two concomitant 532

21

resonances for anomeric C1rsquo1 and C1rsquo2 cis-olefin C9101 and C92 C102 and CH2 gauche and 533

all-trans environments) and relaxation behaviour (fast and slow T2 relaxation components) 534

strongly suggests the coexistence of two molecular environments Further proofs of the 535

coexistence of two distinct molecular domains are given by a number of chemical shift 536

variations in the 13C CPMAS spectrum at 70degC compared to the one at 110degC 90degC and 537

solution This is shown in Figure 7 and in the list of 13C chemical shifts in Table 1 538

539

540

Figure 7 - Comparison between solution- and solid-state 13C NMR spectra of G-C181-OH Solid-state 541

experiments are recorded directly on the sample powder at a MAS of 10 kHz (4 mm zirconia rotor) SP 542

Single Pulse CP Cross Polarization In all the ssNMR experiments temperature is first raised at 110degC 543

and then lowered to either 90degC or 70degC Segmented lines are meant to highlight the known peak position 544

in solution thus stressing the strong chemical shift variations in the solid-state 545

546

NMR experiments above Tm2 suggest that vesicles could contain G-C181-OH with a 547

single family of mobile conformations characterized by chemically-equivalent C=C (red 548

Figure 8b) and a single C1rsquo conformation (green Figure 8b) Experiments between Tm1 and Tm2 549

otherwise suggest that two coexisting conformations characterized by chemically-equivalent 550

(red Figure 8b) and inequivalent (magenta Figure 8b) C=C could be associated to nanotubes 551

Two distinct C1rsquo sites (grey Figure 8b) also indicate two configurations of the sugar headgroup 552

A comment on this point will be given in the following paragraphs 553

The local (Aring-scale) variation of molecular mobility below Tm2 drives the coexistence of 554

two distinct molecular environments of the same G-C181-OH molecule This occurs within a 555

single-phase domain (as deduced by WAXS) of equivalent membrane rigidity (as derived from 556

Laurdan probe in confocal microscopy) Although never reported before the difference in the 557

local mobility of the same amphiphile (here G-C181-OH) could satisfy the theoretically-558

140 120 100 80 60 40 20

CP(3 ms)-70degC

SP-70degC

SP-90degC

SP-110degC

ppm

LiquidC910

C1rsquo

C2rsquo

C4rsquo

C3rsquo

C1

C6rsquoC5rsquo

C1

6

C8

11

C2 C

H2

C1

5

CH

2

DM

SO

13C

22

required conditions that induce local lipid segregation and formation of membrane inclusions 559

(Figure 8c) In case of their non-zero average orientation the local membrane curvature 560

increases62 and nanotubes can spontaneously form (Figure 8bc) More intriguing this 561

mechanism occurs at least 10degC above Tm1 although it should not be favored above the Tm27 562

From both experimental and theoretical backgrounds the conditions of topologically-563

connected vesicles-nanotubes systems are met only when either internal (two components) or 564

external driving forces are applied19 In the present single-lipid system of which the phase 565

behavior is summarized in Figure 8a the most plausible driving force could be explained by 566

the NMR results collected on the dry sample G-C181-OH concomitantly occupies two 567

molecular environments characterized by 1) two distinct conformations of the sugar 568

displaying equivalent mobility and 2) two conformations of the acyl chain unsaturation 569

570

23

571

Figure 8 ndash a) Thermotropic phase behavior of G-C181-OH at 5 mgmL-1 Note that Tm2 is observed on the 572

dry powder only b) Possible relationship between the molecular environment as derived from NMR and 573

membrane curvatures All the molecular drawings refer to the same G-C181-OH molecule (bulky ellipsoid 574

equiv sugar headgroup) while each color corresponds to a given configuration Grey of C1rsquo1 and C1rsquo2 indicates 575

the uncertainty about the assignment of the specific NMR peak The curvature C is defined as 1R R being 576

the curvature radius R1 R2 radius of curvature CG Gauss curvature CM Mean curvature c) Free 577

interpretation of the anisotropic high-curvature (CMgt 0 CG= 0) inclusions in the membrane (low CM and 578

CG) continuum driven by inter-sugar H-bonding and reduced intramembrane diffusion for a G-C181-OH 579

glucolipid 580

581

Flat crystalline structure

Nanotube-vesicle network

Vesicles

48degC

120deg

106 105 104 103 102 101

ppm

H9 H10H1rsquo

106 105 104 103 102 101

ppm106 105 104 103 102 101

ppm106 105 104 103 102 101

ppm

C=CC=C

C1rsquo1C1rsquo2

C1rsquo

Tm2(96degC)Tm1

a)

b)

c)

Same molecule different local orientations and sugar conformations

High curvature membrane

Low curvature inclusion

Flip-FlopStrong H-bonding

CG

CM ( + )

CG 0

CM 0 CG 0

CM 0

G-C181-OH

24

Two related questions are unanswered at the moment What actually drives nanotube 582

formation and is the organization of G-C181-OH within the vesicles and tubes the same Lipid 583

nanotubes have been studied for at least three decades over which broad experimental and 584

theoretical insight has been generated Despite this fact the actual understanding is still matter 585

of debate27 Micelle-to-nanotube and vesicle-to-nanotube phase transitions are generally 586

reported for single amphiphiles below a given Tm and intermediate phase composed of twisted 587

and flat fibers is concomitantly reported Initially thought to be a chirality-driven phenomenon 588

recent studies propose a more general mechanism involving crystallization and symmetry 589

breaking in the lipid packing driven by optimal packing27 If twisted ribbons were observed 590

before for a number of microbial glycolipid bolaamphiphiles44 they are not found for G-C181-591

OH This fact as well as the large body of cryo-TEM images and the observation of nanotubes 592

above Tm1 rather pledge in favour of a direct topological change from positive (vesicles) to 593

zero (nanotubes) Gaussian curvatures In this case vesicle-to-nanotube transitions could be 594

driven by external forces56307813ndash162829 595

Considering the absence of external forces in the present system there are only two 596

possibilities left spontaneous budding of the membrane or nanotube pulling after inter-vesicle 597

collisions due to Brownian motion Both are possible as inclusions composed of a different 598

arrangement of the same molecule in the membrane continuum can theoretically explain 599

spontaneous nanotube formation in an isotropic medium672223629ndash1218ndash21 However the second 600

one is probably more likely to prevail because it could explain the large number of nanotubes 601

connecting two distant vesicles Interestingly similar events occur for any phospholipid-based 602

vesicular solution but nanotubes are extremely rare unless at least two lipids of different 603

molecular structure are mixed together The spectroscopic 1H and 13C and relaxation (1H) 604

ssNMR data even if collected on a dry powder demonstrate that the coexistence of two G-605

C181-OH configurations involving both the sugar headgroup and fatty alcohol chain are 606

possible for this molecule As a matter of fact this is not so surprising as more than one energy 607

minima each corresponding to a given conformation are known for glucose its non-reducing 608

derivative methylglucose and carbohydrates in general78ndash80 609

The presence of two configurations could explain the coexistence between two 610

structures of different curvature but the data presented so far do not explain the origin of these 611

two configurations In other words what drives the change in the local structure of the sugar 612

headgroup and fatty alcohol within the membrane 613

The association of known self-assembled structures to spectroscopic 13C ssNMR data is 614

known for years in particular for gluconamides lipids having a linear sugar headgroup81ndash84 It 615

25

was found that gauche effects in some cases up to the γ position in the sugar chain induce 616

upfield chemical shifts while anti and trans conformations result in downfield shifts Each 617

structure (monolayer crystal bilayer crystal or micellar fiber) was then characterized by a given 618

organization of the gluconamide and a characteristic 13C ssNMR fingerprint which could be 619

explained by gauche γ-gauche anti and trans conformations of the headgroup Similar 620

structural-spectroscopic correlations have been collected over the years for much simpler cyclic 621

sugars85 whereas pyranoses can undergo important conformational changes (chair half-chair 622

boat skew) when the appropriate energy barriers are crossed78ndash8086 A quantitative 623

conformational study based on the 13C chemical shift variations shown in Table 1 would be 624

certainly helpful but very delicate in the context of the present state of the art because a 625

rigorous approach requires at least one known crystal structure8182 which is currently missing 626

for this compound A qualitative approach will then be adopted 627

G-C181-OH is a bolaform amphiphile and flip-flop could be a possible mechanism 628

involved in the nanotube stabilization whereas head-tail interactions have long been considered 629

as important in nanotube-forming bolaform amphiphiles6063 From mere considerations based 630

on T2 (1H) relaxation arguments one could exclude head-tail configurations if the R-CH2OH 631

endgroup (H1) was involved in a direct interaction with the sugar group its corresponding T2 632

would probably be comparable to the sugarrsquos one (OH and H1rsquo) but this does not seem to be 633

the case However this hypothesis would be based on the comparison of the relaxation from 634

CH and CH2 protons additionally involved in different kind of motions Besides the relaxation 635

of H1 is influenced by the strong overlap between H1 and H5rsquo For these reasons relaxation 636

arguments are not solid enough and will not be used 637

We then address this question by exploring intermolecular interactions in a direct 638

manner using 2D 1H-1H homonuclear ssNMR correlation spectroscopy At 90degC in the fast-639

tumbling domain of molecular motion (narrow 1H NMR peaks Figure 6a) and corresponding 640

to the vesicle-rich region spin diffusion (NOESY) experiments are preferred over dipolar 641

coupling based experiments because such interaction is averaged out by molecular motion 642

However when the molecular motion is more constrained in the domain of nanotube stability 643

at 70degC (broad 1H spectra Figure 6a) Double-Quantum Single-Quantum (DQ-SQ) correlation 644

experiments are best suited to probe interaction between close molecular neighbors These 645

experiments are reported in Figure 9 and the related interpretation is only briefly reported 646

below while a more detailed discussion can be found on Page 12 in the Supporting Information 647

26

648

Figure 9 ndash a) 2D 1H-1H NOESY NMR experiment performed at MAS= 10 kHz (4 mm probe) on a Bruker 649

300 MHz Avance III HD spectrometer using 300 ms of mixing time Sample temperature is regulated to 650

90degC reached by cooling from 110degC b) 2D 1H-1H DQ-SQ (BAck-to-BAck BABA pulse sequence used for 651

the DQ excitation scheme) NMR experiment performed at MAS= 22 kHz (25 mm probe) on a Bruker 700 652

MHz Avance III spectrometer using a DQ excitation time equal to 2 rotor periods Sample temperature is 653

regulated to 70degC reached by cooling from 110degC The typical schemes illustrating how to interpret these 654

2D experiments are reported above each experimental map Zooms of the sugar region and configurational 655

interpretation are given below each 2D map More information are given in the text below 656

H16-H16

H1-H1

H9-H10

H1rsquo-H1

H1rsquoH910 H16

H1

H1rsquo-H1rsquo

Hxrsquo-Hxrsquo

2

(pp

m)

(ppm)

315

811910 1rsquo

2rsquo2

3rsquo 4rsquo6rsquo

5rsquo

1rsquo2910

asymp asymp asympHDO MeOD Acetone

1

315

811 9101rsquo

2rsquo2

3rsquo4rsquo6rsquo

5rsquo

1rsquo2 9 10

asympasympasympHDOMeODAcetone

H1rsquo-H1

1rsquo

1

(p

pm

)

(ppm)

H1-H2

H1rsquo-H1

H1rsquo

H910

H1H5rsquo

H2

H1rsquo-H5rsquo

315

811910 1rsquo

2rsquo2

3rsquo 4rsquo6rsquo

5rsquo

1rsquo2910

asymp asymp asympHDO MeOD Acetone

1

H2rsquoH3rsquoH4rsquo

H2rsquoH3rsquoH4rsquo

X= 2rsquo3rsquo4rsquo

NOESY - T= 90degC

1H

1H

1H

1H

DQ-SQ - T= 70degC

a) b)

2

(pp

m)

(p

pm

)

(ppm)

A B

AhellipB

AhellipA

BhellipB

NOESY

(ppm)

A B

AhellipB

DQ-SQ

Single-quantum dimension

Do

ub

le-q

ua

ntu

m d

imen

sio

n

AhellipA

BhellipB

BhellipA

BhellipA

315

811910 1rsquo

2rsquo2

3rsquo 4rsquo6rsquo

5rsquo

1rsquo2910

asymp asymp asympHDO MeOD Acetone

1

lt 5 Aring

lt 5 Aring

H1-H2

H811

H15

H16

H6rsquo

H1rsquo

H1

H5rsquo

H2

H1-H2

H1rsquo-H1

H1rsquo-H5rsquo

H1rsquo

H1 H5rsquo

H1rsquo

H1

H5rsquo

H1H2rsquo H3rsquo H4rsquo

H1-H1

H1rsquo-H1

Hxrsquo-HxrsquoX= 2rsquo3rsquo4rsquo

H1-H2

H1rsquo

Zoom sugar region 28 ndash 45 ppmNOESY DQ-SQ

27

Head-tail or head-head configurations could be probed by the existence of H1rsquo-H1 657

interactions whereas H1 (δH= 356 ppm) is associated to the glucolipid tail and H1rsquo (δH= 432 658

ppm) to the sugar headgroup These atoms are more than 15 Aring apart and a direct interaction 659

could not be observed in both NOESY and DQ-SQ experiments NOESY experiments (Figure 660

9a) show that H1rsquo-H1 correlations are not observed Meanwhile clear-cut direct H1rsquo-H1 661

interactions are indeed observed are detected at T= 70degC (DQ cross-peaks related to H1rsquo-H1 662

correlation in DQ-SQ experiment shown in Figure 9b) Interestingly the absence of auto peaks 663

at δH= 432 ppm and δH= 335 ppm in the SQ dimension excludes H1rsquo-H1rsquo and Hxrsquo-Hxrsquo (x= 2 664

3 4) interactions respectively strengthening the idea that sugar headgroups from two nearest-665

neighbor G-C181-OH molecules are not adjacent in the nanotube-vesicle region 666

Homonuclear 2D ssNMR supports the idea according to which G-C181-OH can 667

undergo molecular flip-flops and a head-tail arrangement in the dry product Extrapolating these 668

conclusions in the lower concentration range lack of H1rsquo-H1 proximity at T= 90degC suggests a 669

head-head configuration in the vesicle-rich region and head-tail configuration within the 670

nanotubes It is however still unclear whether flip-flops are driven by spontaneous temperature-671

induced conformational changes in the glucolipid packing stimulating inclusions within the 672

membrane itself672223629ndash1218ndash21 or by defects generated by inter-vesicle collisions In fact a 673

combination of both is not unlikely Nanotubes are not stable when temperature is 20 to 30 674

degrees above Tm1 indicating that inter-vesicle collisions alone cannot explain the unusual 675

phenomenon In an energy landscape approach87 the head-head configuration is certainly 676

favored at high temperature while both head-head and head-tail configurations could be 677

favored in the mid-temperature region corresponding to the range between Tm1 and Tm2 in the 678

dry powder Both statistical and defect-induced flip-flop events contribute to form high mean 679

zero Gaussian curvature morphologies (nanotubes) for which the head-tail conformation 680

limiting the proximity between bulky sugar headgroups is probably more adapted (Figure 8c) 681

The flip-flop interpretation now reasonably explains the strong differences in the sugar and fatty 682

alcohol conformations deduced by the multiple NMR signals attributed to chemically-683

equivalent groups The existence of various low-energy conformations each having specific 684

NMR signature for sugars in solution78ndash8086 and solid-state81ndash84 further supports our 685

interpretation 686

One last point should be clarified Why between Tm1 and Tm2 are two molecular 687

environments able to coexist instead of a single one of average rigidity as found in the case of 688

ldquoripplerdquo phases50ndash54 We believe that this question can be answered based on the NMR results 689

the relatively short T2 values of the H1rsquo and OH protons shows the constrained mobility of the 690

28

sugar headgroup despite a certain flexibility of the chain (long T2 component) This trend may 691

suggest restrictions in translational mobility probably due to the sugar-OH hydrogen bonding 692

(Figure 8c) meaning that lipids are not able to diffuse within the membrane thus keeping two 693

different configurations stable 694

695

Conclusions 696

Topologically-connected nanotube-vesicle lipidic systems are commonly observed for 697

self-assembled membranes composed of at least two amphiphiles and spontaneously driven by 698

anisotropic inclusions andor by external directional forces This work shows that strong 699

differences in membrane curvature can spontaneously occur in single-lipid systems without 700

external anisotropic forces Coexistence and interconnection between nanotubes and vesicles 701

are evidenced via cryo-TEM and fluorescence confocal microscopy in a single-phase region at 702

temperature above a transition temperature Tm1= 483degC determined by DSC on the hydrated 703

sample but below the second transition temperature Tm2 (957degC) determined by DSC on the 704

dry powder X-ray diffraction (WAXS) also shows the coexistence of two lattice periods one 705

attributed to an inter-membrane repeating distance and the other possibly related to the intra-706

nanotube glucolipid arrangement Labelling the systems with Laurdan a fluorophore of which 707

the emission strongly varies with the local membrane rigidity excludes the presence of rigid 708

micro-separated phases (eg Lβ) 709

1H and 13C ssNMR despite the fact that it can only be performed on the dry powder 710

confirm the presence of one main fluid G-C181-OH environment above Tm2 Below Tm2 the 711

splitting of some 13C peaks and the coexistence of short and long 1H relaxation components 712

demonstrate that G-C181-OH can simultaneously pack into structurally and dynamically 713

inequivalent environments Combining advanced 2D 1H-1H correlation experiments based on 714

spin diffusion or double-quanta excitation suggests that vesicles could contain a preferred head-715

head configuration of the bolaform lipid while nanotubes a preferred head-tail assembly 716

The presence of two environments composed of the same molecule characterized by 717

different molecular configurations and rigidity fulfills the theoretical conditions necessary to 718

spontaneously nucleate a membrane inclusion of non-zero average orientation described as 719

necessary to observe topologically-connected nanotube-vesicle structures in the absence of 720

external forces This hypothesis could also be enriched by the possibility of nanotubes to be 721

pulled away between two vesicles after a collision event 722

This work expands the actual theoretical and experimental framework on lipid networks 723

stabilized by a different radius of curvature It could also motivate the synthesis or isolation 724

29

of new lipids providing a higher level of 3D architectural control in soft networks possibly in 725

the range of room temperature 726

727

Acknowledgements 728

Dr Andrea Lassenberger and Dr Sylvain Preacutevost at Institut Laue Langevin (ILL Grenoble 729

France) are kindly acknowledged for their assistance on the SANS experiment Confocal 730

images were performed at the Institut de Biologie Paris Seine (IBPS) imaging facility Jean-731

Franccedilois Gilles (Imaging Core Facility Institut de Biologie Paris Seine (IBPS) CNRS 732

Sorbonne Universiteacute Paris France) is kindly acknowledged for his assistance on the confocal 733

microscopy imaging Dr Guillaume Laurent (LCMCP Sorbonne Universiteacute Paris France) is 734

kindly acknowledged for helpful discussions The French region Ile-de-France SESAME 735

program is acknowledged for financial support (700 MHz NMR 736

spectrometer) Dr Olivier Diat (ICSM Marcoule France) Prof eronikaIglič and Prof V Aleš737

onsul discussi Slovenia) are kindly acknowledged for helpfUniversity of LjubljanaIglič (-Kralj 738

739

Financial support 740

WAXS experiments have been supported by Soleil Light Source Saint Aubin France proposal 741

Ndeg 20201747 SANS experiments have been supported by ILL proposal Ndeg9-13-778 742

743

References 744

1 D Kabaso N Bobrovska W Goacuteźdź E Gongadze V Kralj-Iglič R Zorec and A 745

Iglič Bioelectrochemistry 2012 87 204ndash210 746

2 C R Safinya U Raviv D J Needleman A Zidovska M C Choi M A Ojeda-747

Lopez K K Ewert Y Li H P Miller J Quispe B Carragher C S Potter M W 748

Kim S C Feinstein and L Wilson Adv Mater 2011 23 2260ndash2270 749

3 P Veranič M Lokar G J Schuumltz J Weghuber S Wieser H Haumlgerstrand V Kralj-750

Iglič and A Iglič Biophys J 2008 95 4416ndash4425 751

4 A Rustom R Saffrich I Markovic P Walther and H-H Gerdes Science (80- ) 752

2004 303 1007ndash1010 753

5 M I Angelova S Soleacuteau P Meacuteleacuteard F Faucon and P Bothorel Prog Colloid 754

Polym Sci 1992 89 127ndash131 755

6 L Mathivet S Cribier and P F Devaux Biophys J 1996 70 1112ndash1121 756

7 V Kralj-Iglič G Gomišček J Majhenc V Arrigler and S Svetina Colloids Surfaces 757

A Physicochem Eng Asp 2001 181 315ndash318 758

30

8 V Kralj-Iglič A Iglič G Gomišček F Sevšek V Arrigler and H Haumlgerstrand J 759

Phys A Math Gen 2002 35 1533ndash1549 760

9 M Drab D Stopar V Kralj-Iglič and A Iglič Cells 2019 8 626 761

10 A Zidovska K K Ewert J Quispe B Carragher C S Potter and C R Safinya 762

Langmuir 2009 25 2979ndash2985 763

11 A Zidovska K K Ewert J Quispe B Carragher C S Potter and C R Safinya Soft 764

Matter 2011 7 8363ndash8369 765

12 A Zidovska K K Ewert J Quispe B Carragher C S Potter and C R Safinya 766

Biochim Biophys Acta - Biomembr 2009 1788 1869ndash1876 767

13 M Karlsson K Sott M Davidson A S Cans P Linderholm D Chiu and O Orwar 768

Proc Natl Acad Sci U S A 2002 99 11573ndash11578 769

14 A Karlsson R Karlsson M Karlsson A-S Cans A Stromberg F Ryttsen and O 770

Orwar Nature 2001 409 150ndash152 771

15 M Karlsson K Sott A S Cans A Karlsson R Karlsson and O Orwar Langmuir 772

2001 17 6754ndash6758 773

16 E Evans H Bowman A Leung D Needham and D Tirrell Science (80- ) 1996 774

273 933ndash935 775

17 U Seifert Configurations of fluid membranes and vesicles 1997 vol 46 776

18 J Kaumls and E Sackmann Biophys J 1991 60 825ndash844 777

19 N Bobrovska W Goacuteźdź V Kralj-Iglič and A Iglič PLoS One 2013 8 e73941 778

20 A Iglič B Babnik K Bohinc M Fošnarič H Haumlgerstrand and V Kralj-Iglič J 779

Biomech 2007 40 579ndash585 780

21 K Bohinc T Slivnik A Iglič M Brumen and V Kralj-Iglič J Phys Chem C 2007 781

111 9709ndash9718 782

22 A Iglič H Haumlgerstrand P Veranič A Plemenitaš and V Kralj-Iglič J Theor Biol 783

2006 240 368ndash373 784

23 V Kralj-Iglič and P Veranič Adv Planar Lipid Bilayers Liposomes 2006 5 129ndash785

149 786

24 A Iglic and V Kralj-Iglic Adv Planar Lipid Bilayers Liposomes 2006 4 253ndash279 787

25 S Penič L Mesarec M Fošnarič L Mroacutewczyńska H Haumlgerstrand V Kralj-Iglič and 788

A Iglič Front Phys 2020 8 342 789

26 P Yager and P E Schoen Mol Cryst Liq Cryst 1984 106 371ndash381 790

27 T G Barclay K Constantopoulos and J Matisons Chem Rev 2014 114 10217ndash791

10291 792

31

28 B Božič G Gomišček V Kralj-Iglič S Svetina and B Žekš Eur Biophys J 2002 793

31 487ndash496 794

29 A Iglič H Haumlgerstrand M Bobrowska-Haumlgerstrand V Arrigler and V Kralj-Iglič 795

Phys Lett A 2003 310 493ndash497 796

30 R Bar-Ziv and E Moses Phys Rev Lett 1994 73 1392ndash1395 797

31 S Chiruvolu H E Warriner E Naranjo S H Idziak J O Raumldler R J Plano J A 798

Zasadzinski and C R Safinya Science (80- ) 1994 266 1222ndash1225 799

32 V Kralj-Iglič A Iglič H Haumlgerstrand and P Peterlin Phys Rev E 2000 61 4230ndash800

4234 801

33 L Mesarec M Drab S Penič V Kralj-Iglič and A Iglič Int J Mol Sci 2021 22 802

2348 803

34 L Van Renterghem S L K W Roelants N Baccile K Uyttersprot M C Taelman 804

B Everaert S Mincke S Ledegen S Debrouwer K Scholtens C V Stevens W 805

Soetaert L Van Renterghem S L K W Roelants N Baccile K Uytersprot M C 806

Taelman B Everaert S Mincke S Ledegen S Debrouwer K Scholtens C V 807

Stevens and W Soetaert Biotechnol Bioeng 2018 115 1195ndash1206 808

35 J Clark and F Deswarte Introduction to Chemicals from Biomass Second Edition 809

John Wiley amp Sons 2015 810

36 K V Sajna R Houmlfer R K Sukumaran L D Gottumukkala and A Pandey in 811

Industrial Biorefineries and White Biotechnology eds A Pandey R Houmlfer M 812

Taherzadeh K M Nampoothiri and C Larroche Amsterdam Oxford Waltham 813

Elsevier 2015 pp 499ndash521 814

37 R Marchant and I M Banat Trends Biotechnol 2012 30 558ndash565 815

38 K Hill and C LeHen-Ferrenbach in Sugar-Based Surfactants Fundamentals and 816

Applications ed C C Ruiz Boca Raton CRC Press 2009 pp 1ndash20 817

39 L Rodrigues I M Banat J Teixeira and R Oliveira J Antimicrob Chemother 818

2006 57 609ndash618 819

40 J D Desai and I M Banat Microbiol Mol Biol Rev 1997 61 47ndash64 820

41 S L K W Roelants K Ciesielska S L De Maeseneire H Moens B Everaert S 821

Verweire Q Denon B Vanlerberghe I N A Van Bogaert P Van der Meeren B 822

Devreese and W Soetaert Biotechnol Bioeng 2016 113 550ndash559 823

42 N Baccile M Selmane P Le Griel S Preacutevost J Perez C V Stevens E Delbeke S 824

Zibek M Guenther W Soetaert I N A Van Bogaert and S Roelants Langmuir 825

2016 32 6343ndash6359 826

32

43 N Baccile A-S Cuvier S Preacutevost C V Stevens E Delbeke J Berton W Soetaert 827

I N A Van Bogaert and S Roelants Langmuir 2016 32 10881ndash10894 828

44 N Baccile C Seyrig A Poirier S A Castro S L K W Roelants and S Abel Green 829

Chem 2021 23 3842ndash3944 830

45 A Salis L Cappai C Carucci D F Parsons and M Monduzzi J Phys Chem Lett 831

2020 11 6805ndash6811 832

46 J Schindelin I Arganda-Carreras E Frise V Kaynig M Longair T Pietzsch S 833

Preibisch C Rueden S Saalfeld B Schmid J-Y Tinevez D J White V 834

Hartenstein K Eliceiri P Tomancak and A Cardona Nat Methods 2012 9 676ndash835

682 836

47 T Parasassi and E Gratton J Fluoresc 1995 5 59ndash69 837

48 M M Dodes Traian F L Gonzaacutelez Flecha and V Levi J Lipid Res 2012 53 609ndash838

616 839

49 D M Owen C Rentero A Magenau A Abu-Siniyeh and K Gaus Nat Protoc 840

2012 7 24ndash35 841

50 W Stillwell in An Introduction to Biological Membranes Academic Press 2016 pp 842

135ndash180 843

51 W Stillwell in An Introduction to Biological Membranes Academic Press 2016 pp 844

181ndash220 845

52 B Tenchov Chem Phys Lipids 1991 57 165ndash177 846

53 R Koynova and M Caffrey Biochim Biophys Acta 1998 1376 91ndash145 847

54 F Schmid S Dolezel O Lenz and S Meinhardt J Phys Conf Ser 2014 487 848

012004 849

55 S Meinhardt R L C Vink and F Schmid Proc Natl Acad Sci U S A 2013 110 850

4476ndash4481 851

56 O Glatter and O Kratky Small Angle X-ray Scattering Academic Press London 852

1982 853

57 B A Cunningham A D Brown D H Wolfe W P Williams and A Brain Phys 854

Rev E 1998 58 3662ndash3672 855

58 M Nadler A Steiner T Dvir O Szekely P Szekely A Ginsburg R Asor R Resh 856

C Tamburu M Peres and U Raviv Soft Matter 2011 7 1512ndash1523 857

59 C Tanford The hydrophobic effect Formation of micelles and biological membranes 858

John Wiley amp Sons Inc New York 1973 859

60 M Masuda and T Shimizu Langmuir 2004 20 5969ndash5977 860

33

61 A Zidovska K K Ewert J Quispe B Carragher C S Potter and C R Safinya 861

Block Liposomes Vesicles of Charged Lipids with Distinctly Shaped Nanoscale 862

Sphere- Pear- Tube- or Rod-Segments Elsevier Inc 1st edn 2009 vol 465 863

62 A Iglic V Kralj-Iglic B Božič M Bobrowska-Haumlgerstrand B Isomaa and H 864

Haumlgerstrand Bioelectrochemistry 2000 52 203ndash211 865

63 T Shimizu M Masuda and H Minamikawa Chem Rev 2005 105 1401ndash1444 866

64 J V Selinger and J M Schnur Phys Rev Lett 1993 71 4091ndash4094 867

65 J M Schnur Science (80- ) 1993 262 1669ndash1676 868

66 B Tenchov R Koynova and G Rapp Biophys J 2001 80 1873ndash1890 869

67 R Koynova and B Tenchov Ser Biomech 2017 31 3ndash19 870

68 S J Angyal and R Le Fur Carbohydr Res 1984 126 15ndash26 871

69 J K Whitesell and M A Minton J Am Chem Soc 1987 109 225ndash228 872

70 T Takayama S Ando and I Ando J Mol Struct 1990 220 245ndash250 873

71 K Gawrisch N V Eldho and I V Polozov Chem Phys Lipids 2002 116 135ndash151 874

72 F Tielens C Gervais J F Lambert F Mauri and D Costa Chem Mater 2008 20 875

3336ndash3344 876

73 G A Jeffrey and Y Yeon Acta Crystallogr Sect B 1986 42 410ndash413 877

74 B Berglund and R W Vaughan J Chem Phys 1980 73 2037ndash2043 878

75 C Gervais C Coelho T Azaiumls J Maquet G Laurent F Pourpoint C Bonhomme 879

P Florian B Alonso G Guerrero P H Mutin and F Mauri J Magn Reson 2007 880

187 131ndash140 881

76 C Akita T Kawaguchi F Kaneko H Yamamoto and M Suzuki J Phys Chem B 882

2004 108 4862ndash4868 883

77 R N Purusottam L Seacutenicourt J J Lacapegravere and P Tekely Biochim Biophys Acta - 884

Biomembr 2015 1848 3134ndash3139 885

78 X Biarneacutes A Ardegravevol A Planas C Rovira A Laio and M Parrinello J Am Chem 886

Soc 2007 129 10686ndash10693 887

79 L M J Kroon‐Batenburg and J Kroon Biopolymers 1990 29 1243ndash1248 888

80 K N Kirschner and R J Woods Proc Natl Acad Sci U S A 2001 98 10541ndash889

10545 890

81 S Svenson B Kirste and J H Fuhrhop J Am Chem Soc 1994 116 11969ndash11975 891

82 I Sack S Macholl J H Fuhrhop and G Buntkowsky Phys Chem Chem Phys 892

2000 2 1781ndash1788 893

83 S Svenson J Koening and J H Fuhrhop J Phys Chem 1994 98 1022ndash1028 894

34

84 S Svenson A Schaefer and J H Fuhrhop J Chem Soc Perkin Trans 2 1994 2 895

1023ndash1028 896

85 K Bock and C Pedersen Adv Carbohydr Chem Biochem 1983 41 27ndash66 897

86 T B Grindley Structure and Conformation of Carbohydrates Springer Berlin 898

Heidelberg 2008 899

87 F Tantakitti J Boekhoven X Wang R V Kazantsev T Yu J Li E Zhuang R 900

Zandi J H Ortony C J Newcomb L C Palmer G S Shekhawat M O De La 901

Cruz G C Schatz and S I Stupp Nat Mater 2016 15 469ndash476 902

903

1

Supporting Information 904

905

Topological connection between vesicles and nanotubes in single-906

component lipid membranes driven by head-tail interactions 907

908

Niki Baccilea Ceacutedric Lorthioira Abdoul Aziz Baa Patrick Le Griela Cristina Coelhob Javier 909

Perezc Wim Soetaertd Sophie L K W Roelantsd 910

911

a Sorbonne Universiteacute Centre National de la Recherche Scientifique Laboratoire de Chimie de 912

la Matiegravere Condenseacutee de Paris LCMCP F-75005 Paris France 913

b Institut des Mateacuteriaux de Paris Centre Sorbonne Universiteacute Paris France 914

c Synchrotron Soleil LrsquoOrme des Merisiers Saint-Aubin BP48 91192 Gif-sur-Yvette Cedex 915

France 916

d InBio Department of Biotechnology Ghent University Ghent Belgium 917

918

919

Corresponding author 920

Dr Niki Baccile 921

E-mail address nikibaccilesorbonne-universitefr 922

Phone +33 1 44 27 56 77 923

924

925

926

2

a) 927

Parameter Method

Dry Matter (DM ) 996 Infrared balance 105degC

Glucose () 012 HPLC-Metacarb

Glycerol () nd HPLC-Metacarb

Free Fatty Acid content () lt001 Internal method BBEPP

GC with FID Detector

Oil () lt001 Internal method BBEPP

GC with FID Detector

Moulds (CFUg) lt10 3M-Nordval ndeg16 B

Yeasts (CFUg) lt10 3M-Nordval ndeg16 B

Anaerobic count (CFUg) lt10 SP-VG M005 B

Protein () tbd BCA protein assay

928

Figure S 1 - a) HPLC-ELSD chromatogram and composition table 929

930

931

932

933

3

b) 934

935

Group Ndeg δ ppm

CH2 1 354

CH2 2 153

CH2 3-712-14 132

CH2 811 203

CH 910 534

CH2 15 162

CH2 16 388

CH 1 424

CH 2 316

CH 3 334

CH 4 329

CH 5 354

CH2 6 366

Figure S 1 (next) ndash b) 1H NMR spectrum (MeOD-d4) recorded on the purified G-C181-OH compounds 936

and related assignment 937

938

2

811910 1rsquo

5rsquo1

3rsquo4rsquo

1rsquo1910

asymp asymp asympHDO MeOD Acetone

6rsquo2rsquo

1615

3-712-14

4

939

940

Figure S 2 ndash Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 130degC 941

5

942

943

Figure S 3 - Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 90degC 944

945

90degC

1

22

6

946

947

948

949

Figure S 4 ndash Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 70degC 950

951

21

11

1

2

5

3

4

45

5

6

7

7

7

952

953

954

Figure S 5 ndash Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 25degC 955

956

957

1

1 1

8

958

Figure S 6 ndash a) SANS experiment recorded at 90degC for a 5 mgmL G-C181-OH aqueous solution b-c) 959

WAXS experiments recorded on aqueous solutions of G-C181-OH at b) 5 mgmL and c) 25 mgmL d) 960

Structural interpretation of the repeating distances pointed out in c) 961

962

963

964

001 01 1

q Aring-1

25degC

68degC

88degC

111degC

01 1q Aring

-1

02 04 06 08 1q Aring

-1

001 01

10-1

100

101

Inte

nsity

cm

-1

q Aring-1

-2

T= 90degC

SANS WAXS WAXS

a) b) c)

dv

dL d= 144 Aring-1

di

ddH2O

dt

dvdL

dtdi d

d)

9

965

966

Figure S 7 ndash ssNMR MAS (10 kHz) experiments recorded on a G-C181-OH dry powder a) 1H spectra 967

recorded from 110degC to 70degC b) 13C single pulse (SP) spectra recorded with high power 1H decoupling from 968

110degC to 70degC c) 13C cross polarization (CP) spectra recorded at 110degC and contact time tc of 50 micros and 969

3000 micros d) 13C CP spectra recorded at 70degC and for 20 lt tc micros lt 3000 970

971

1

811

910

CH2 CH2

1rsquo

6rsquo

8 6 4 2 0 ppm

110

90

80

70

910

1rsquo

6rsquo16

1

2rsquo3rsquo4rsquo

811

OH

45 40 35 30 25 20 15

ppm

200 180 160 140 120 100 80 60 40 20 0

ppm140 120 100 80

ppm

CH2-gauche

3004

110

90

80

70

65

60

50

25

C91013003

C9 - 12869

CH2-all-trans

3198

C1rsquo10335

a) b)

1H

T degCT degC

13C Single Pulse MAS

A

B

A

B

70degC70degC

5rsquo

215

200 180 160 140 120 100 80 60 40 20 0

ppm

13C CPMAST= 110degC

200 180 160 140 120 100 80 60 40 20 0

ppm

132 131 130 129 128 127 126

ppm

106 105 104 103 102

ppm

13C CPMAST= 70degC

50

tc micros

3000

20

tc micros

50 500 1000 3000

tc micros

C102

13024 C910113003

C92

12869

C1rsquo213082

C1rsquo110443

A B

106 105 104 103 102

ppm

3000

1000

500

50

20

AB

c)

d)

10

972

Figure S 8 ndash a) Short and d) long T2 components displayed by the 1H transverse relaxation signal c) 1H 973

transverse relaxation signals recorded on a G-C181-OH dry powder from 110degC to 70degC The area of each 974

given resonance is plotted (log-lin) as a function of the echo time (CPMG pulse sequence) for each 975

temperature and fitted using the equation given above T2s and T2l indicate the short and long T2 relaxation 976

times respectively obtained by describing the 1H transverse relaxation signal using a biexponential decay 977

For H1rsquo H5rsquo and H6rsquo the 1H relaxation function could be described by a monoexponential decay 978

- Protons from the sugar group (H1rsquo H5rsquo H6rsquo) show a single T2 relaxation component (Figure 979

6c and Figure S 8) while protons of the acyl chain (CH2 H910 H811) show both short (Figure 980

6c) and long (Figure 6d) T2 relaxation component 981

- The short T2 relaxation time systematically decreases from 110degC to 70degC for all species by a 982

factor ranging from two to seven depending on the specific chemical group Very long T2 are 983

observed at 110degC for all acyl species while nearly unchanged values are measured below 984

90degC without any clear trend 985

- The sugar protons (H1rsquo H5rsquo H6rsquo) have the shortest set of T2 values (lt 1 ms) below 100degC 986

The long component of T2 generally ranges between 4 ms and 12 ms 987

0 1 2 3 4 5 6 7 8

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1 2 3 4 5 6 7 8

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

H1

H8H11

H9H10

CH2

H1rsquo

H6rsquo

CH2

H910H811 H1CH2

H1rsquo H5rsquo6rsquo

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

OH

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45 T= 110degC

T= 90degC

T= 80degC

T= 70degC

Log

In

ten

sity

Time ms

Fit

)

CH2 H910 H811 H1 H56 H1 OH

00

05

10

15

20

25

30

35

40

T2 m

s

T= 110degC

T= 90degC

T= 80degC

T= 70degC

CH2 H910 H811 H1

0

2

4

6

8

10

12

T2 m

s

T= 110degC

T= 90degC

T= 80degC

T= 70degC

Short T2(1H) Long T2(1H)

a) b)

c)

11

Table S 1 ndash 1H chemical shifts of G-C181-OH in solution and in the solid-state Solid-state spectra are 988

recorded at a MAS frequency of 10 kHz in DMSO-d6 broad peak 989

990

δH ppm

Group Ndeg Liquid

(MeOD-d4) Solid

(T= 110degC) Solid

(T= 70degC)

CH2 H1 354 357

CH2 H2 153 156

CH2 H3-712-14 132 135

CH2 H811 203 203

CH H910 534 534

CH2 H15 162 164

CH2 H16 388 383

CH H1 424 432

CH H2 316 335

CH H3 334 335

CH H4 329 335

CH H5 354 351 357

CH2 H6 366 383

OH OH1 432

45 463

OH OH2 492

OH OH3 490

OH OH4 487

OH OH6 445

991

992

H1H9H10H1rsquo

H6rsquo

OH2rsquo

OH1

12

Detailed discussion about the NOESY and DQ-SQ 2D 1H-1H ssNMR experiments presented in 993

Figure 9 994

995

2D 1H-1H ssNMR experiments are a unique tool to explore through-space proximities 996

between 1H nuclei ssNMR offers a large number of possibilities each adapted to a given 997

system The advantage of 2D maps with respect to a dynamics approach is their direct 998

readability in clear two nuclei which exchange their magnetization through-space (whichever 999

mechanism may be responsible for this exchange) will display a specific cross-peak However 1000

the way cross-peaks are presented and interpreted strictly depends on the nature of the pulse 1001

program which in turn is chosen according to the physical nature of the sample 1002

1D temperature-dependent 1H NMR experiments performed on G-C181-OH (Figure 1003

4a in the main text) indicate sharp peaks at 90degC and broad peaks at 70degC Narrow peaks 1004

indicate fast andor large-amplitude molecular tumbling averaging the strong 1H-1H dipolar 1005

interactions This excludes the use of pulse sequences like DQ-SQ which is not efficient under 1006

these conditions For this reason we employ spin-diffusion-based pulse sequences like 1007

NOESY which is more adapted for samples with high molecular mobility When mixing times 1008

are chosen to be short (lt 1 s) one generally explores through-space proximities within about 5 1009

10 Aring On the contrary when molecular motions are restricted dipolar coupling becomes 1010

significant 1H NMR peaks are broader and pulse sequences like DQ-SQ can now be used 1011

with a spatial resolution of less than 5 Aring 1012

Whenever possible DQ-SQ experiments are preferred over spin diffusion-based pulse 1013

sequences because they are more selective in particular for likewise interactions which are 1014

not discriminated in NOESY-type experiments This is illustrated by the typical interaction 1015

schemes for NOESY (top Figure 9a) and DQ-SQ (top Figure 9b) For two given A and B 1016

protons 2D NOESY maps will display two on-diagonal and two square-shaped symmetrical 1017

off-diagonal cross-peaks Off-diagonal cross-peaks illustrate the mutual A-B and B-A 1018

interaction but the on-diagonal peaks are systematic and do not prove any A-A or B-B 1019

interaction Similarly to NOESY 2D DQ-SQ maps display off-diagonal peaks (symmetric 1020

across the diagonal top Figure 9b) upon A-B and B-A interaction but in contrast they will 1021

display on-diagonal peaks only in the presence of self-interaction In the tutorial example of 1022

Figure 9b top the on-diagonal A-A peak indicates mutual A-A interaction while lack of on-1023

diagonal B-B peak evidences the absence of B-B proximities Combining NOESY and DQ-SQ 1024

experiments is capital to study the local arrangement of G-C181-OH and its evolution with 1025

temperature 1026

13

At T= 90degC the H1 and H5rsquo contributions are very close but possible to discriminate 1027

them (δ(H1)= 354 ppm δ(H5rsquo)= 351 ppm) according to 1H NMR chemical shifts in Table S 1028

1 2D 1H-1H NOESY in Figure 9a also discriminates between the two components attributed 1029

on the basis of the interactions with their mutual neighbors (green H1-H2 and H5rsquo-H1rsquo squares 1030

in Figure 9a) On this basis 2D NOESY does not show a direct H1rsquo-H1 correlation (blue red-1031

crossed) but rather an extensive network of intra-sugar interactions These features exclude 1032

suggest intra- and inter-sugar interactions and they exclude head-tail arrangements as a major 1033

molecular configuration at 90degC 1034

At 70degC the 1H NMR peaks are broader indicating a reduction of the T2 (1H) relaxation 1035

times andor residual homonuclear dipolar interactions excluding the use of NOESY In this 1036

case 1H-1H DQ-SQ correlation experiments are best suited Figure 9b presents the 1H-1H DQ-1037

SQ experiment on G-C181-OH at 70degC Peaks are broader than in NOESY and attribution may 1038

be cumbersome However the experiment is performed at 700 MHz and MAS= 22 kHz which 1039

improves spectral resolution with respect to the same experiment performed at 300 MHz and 1040

MAS= 10 kHz The auto-peak at δH= 356 ppm in the SQ dimension is attributed to H1-H1 spin 1041

pairs mainly resulting from intra CH2 interaction while the DQ cross-peaks at δH= 156-356 1042

ppm and δH= 432-356 ppm respectively origin from H2-H1 and H1rsquo-H1 interactions (Figure 1043

9b) DQ-SQ map does not show any auto-peak corresponding to H1rsquo-H1rsquo or Hxrsquo-Hxrsquo (x= 2rsquo 1044

3rsquo 4rsquo) interactions (red-crossed blue Figure 9b) thus excluding any proximity (lt 5 Aring) between 1045

close glucose neighbors On the contrary DQ-SQ maps (recorded both at 700 MHz Figure 9b 1046

or at 300 MHz not shown) systematically show DQ cross-peaks which can only be attributed 1047

to H1rsquo-H1 interactions (δH= 432-356) The presence of these H1rsquo-H1 interactions can only be 1048

explained by a head-tail arrangement of the G-C181-OH molecules as illustrated below the 1049

DQ-SQ map in Figure 9b 1050

10

257

The series of cryo-TEM images in Figure 2b show an overview of the most relevant 258

structures observed on G-C181-OH solution from T= 130degC to T= 25degC whereas T is 259

indicative of the equilibrium temperature before cryofixation (refer to materials and methods 260

section for more information on the sense and interpretation of T) At Tgt 100degC the sample 261

is essentially composed of polydisperse single wall vesicles massively shown by the 262

complementary pictures presented in Figure S 2 In between Tm1 and Tm2 at T= 90degC and T= 263

70degC (Figure 2b Figure 3 Figure 4 complemented by Figure S 3 and Figure S 4) vesicles are 264

connected by nanotubes At Tlt Tm1 flat crystals are otherwise observed (Figure 2b and Figure 265

S 5) 266

267

268

Figure 3 ndash Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 90degC 269

270

A deeper insight at T= 90degC is given in Figure 3 (more images are provided in Figure 271

S 3) Arrows 1 through 6 mark very clearly the typical continuity in the glucolipid membrane 272

between the vesicle and nanotubes Arrow Ndeg 7 shows that a single nanotube connects the outer 273

membrane of two vesicles several hundred nanometers apart The connection is shown by 274

2

34

5

6

6

1

5

7

A

B

D

C

0 10 20 30 40 50 60

BIn

tensity

au

Profile nm

A

0 10 20 30 40 50 60

Inte

nsity

au

Profile nm

C

D

T= 90degC

11

arrows Ndeg5 and Ndeg6 The corresponding profiles A through D confirm the tubular nature of 275

the elongated objects Profiles B and D are both characterized by typical symmetrical thick 276

walls (low intensity depths highlighted by shaded rectangles) surrounding a hollow region (high 277

intensity peak between the shaded rectangles) The wall thickness varies between 35 and 40 278

nm for profile B and between 41 and 61 nm for profile D while the total nanotube diameter 279

varies between 14 nm at B and 20 nm at D At the same time the thickness of the vesicle wall 280

is about 6 nm at A and 135 nm at C Considerations about the bilayer or interdigitated structure 281

of the membrane will be presented below 282

The connection between nanotubes and vesicles seems to be multiple For instance 283

arrow Ndeg1 (Figure S 3) and Ndeg4 (Figure 3) show a single-wall vesicle of which the membrane 284

is extruded into a nanotube Similarly a closer look at arrows Ndeg1 2 and 3 in Figure 3 show 285

that the vesicle is single-wall and fully closed while the nanotube nucleates at the vesicle outer 286

surface with no apparent topological continuity In many regions of the sample at T= 90degC 287

one can find exvaginations or buds232531 in the nanotube wall This is shown for instance by 288

arrows Ndeg7 and particularly put in evidence in Figure 4 by arrows Ndeg1 through Ndeg4 where the 289

amplitude of the oscillation varies between 20-25 nm (arrow Ndeg1) and 6-8 nm (arrows Ndeg3-4) 290

291

Figure 4 - Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heated at T= 90degC 292

293

Still between Tm2 and Tm1 but at lower temperature (T= 70degC) the nanotubes seem to 294

vary from single threads (arrow Ndeg1 Figure S 4) to nanotubes of diameter below 15 nm both 295

connecting nanotubes of diameter ranging between 50 nm and 100 nm (arrows Ndeg2 Figure S 296

4) The larger tubes are most likely formed by the inflation and flattening of the nanotubes 297

probably after assimilation of surrounding vesicles and fusion of the exvagination discussed 298

above The transition between vesicles tubes and flat membranes is certainly more accentuated 299

Arrow Ndeg3 in Figure S 4 shows distinct vesicles while arrow Ndeg4 identifies a flat structure 300

evidently formed from fused vesicles In particular arrow Ndeg5 points at the fusion of two 301

2

1 132

4

12

vesicles while arrow Ndeg6 points at two vesicles fusing into the flat membrane Additional areas 302

where vesicle fusion is prominent are indicated by the oscillations at the outer skirt of broad 303

lamellae and pointed by arrow Ndeg7 304

Finally below Tm1 at T= 25degC for instance mainly flat crystals with no visible 305

exvaginations or tubes are visible (Figure S 5) by cryo-TEM and the material turns into an 306

insoluble precipitate Massive presence of flat crystalline structures is also clear from the bulk 307

iridescent behavior of a G-C181-OH solution under manual shear 308

In summary despite the uncertainty related to the actual state of equilibrium of the 309

glucolipids at the moment of cryofixation cryo-TEM depicts a vesicle phase at Tgt 100degC a 310

flat lamellar phase below Tm1 and an interconnected vesicle-nanotube network between Tm1 and 311

about 100degC The combination of statistically-relevant fluorescence microscopy and 312

scatteringdiffraction experiments provides an additional proof of the coexistence of 313

vesiclenanotube systems between the two transition temperatures Please note that the 314

experiments reported below have been performed under controlled conditions of temperature 315

which is then indicated by the use of T instead of T 316

13

317

Figure 5 ndash a-c) Confocal fluorescence microscopy images recorded at 90degC and 70degC for a 5 mgmL G-318

C181-OH aqueous solution d-g) Typical fluorescence microscopy images recorded at T= 80degC using 319

Laurdan (refer to materials and methods section for more details on this fluorophore) Excitation 320

wavelength is 405 nm for all the images while emission wavelength is 490 nm in (df) and 440 nm in (eg) 321

h) Full emission spectra obtained from the integration of various fluorescence images recorded from 412 322

nm to 600 nm (step 5 nm) at 90degC 80degC and 70degC i) Box plot of GP defined as 119918119927 =(120640120786120786120782minus120640120786120791120782)

120640120786120786120782+120640120786120791120782 with 120640120786120786120782 323

and 120640120786120791120782 being the emission wavelengths at 440 nm and 490 nm at 90degC 80degC and 70degC Each data point 324

l= 490 nm l= 440 nm

l= 490 nm l= 440 nm

10 μm400 450 500 550 600

00

02

04

06

08

10

No

rma

lize

d I

nte

nsity

l nm

90degC

80degC

80degC

70degC

70degC

400 450 500 550 60000

02

04

06

08

10

No

rma

lize

d I

nte

nsity

l nm

90degC

80degC

80degC

70degC

70degC

d) e)

f) g)

h)

i)

T= 80degC

Laurdan

T= 80degC

T= 80degC T= 80degC

T= 90degC T= 70degC T= 70degC

a) b) c)

1

1

1

14

corresponds to a set of regions of interest (ROIs) (54 at 90degC 65 at 80degC and 75 at 70degC) selected in the 325

corresponding fluorescence images ROIs were representatively selected so to include the entire image but 326

also vesicles alone tubes alone and mixtures of vesicles and tubes 327

328

Figure 5ab shows two typical confocal fluorescence microscopy images recorded on a 329

G-C181-OH solution (5 mgmL) in a flame-sealed capillary heat at T= 90degC and T= 70degC 330

Although the resolution of optical microscopy is not comparable with cryo-TEM the sample is 331

broadly composed of vesicles coexisting with nanotubes whereas in some cases the nanotube 332

starts from the vesicle itself (arrow Ndeg1) in good agreement with cryo-TEM data Furthermore 333

SANS recorded at T= 90degC (Figure S 6a) shows a scattering profile with a pronounced -2 334

dependence of the scattering intensity against the wavevector q in the log-log scale This 335

indicates the massive presence of a flat interface and it is expected typical of bilayers in 336

vesicular morphologies56 However the typical signature of nanotubes cannot be observed in 337

the SANS profile and this can be explained by both an argument related to the relative lower 338

amount of nanotubes with respect to vesicles at 90degC but also to the steeper slope of flat 339

morphologies (-2) compared to tubes (-1) the signal of the former masking the tube signal at 340

low wavevector values The SANS profile is characterized by a diffraction peak at q= 0167 Aring-341

1 of which the assignment is refined by WAXS data 342

Temperature-driven WAXS experiments in Figure S 6bc (5 mgmL-1 and 25 mgmL-1 343

respectively) show a lamellar order possibly attributed to a multilamellar structure of the 344

vesicles (dv Figure S 6d) at high temperature and to the interplanar arrangement of the 345

crystalline flat structures (dL Figure S 6d) at T= 25degC The sharp (001) reflection indicates 346

typical interlamellar distance of 353 Aring at 25degC although this value varies with temperature 347

(400 Aring at 111degC 372 Aring at 88degC) thus confirming its assignment to a hydration-dependent 348

interlamellar order More intriguing is the second broader peak at q= 0403 Aring-1 (d= 156 Aring) 349

and its reflection both clearly visible at 88degC below Tm2 We attribute it to the inter-glucolipid 350

arrangement inside the tubular phase (dt Figure S 6d) coexisting with the vesicle phase 351

between Tm1 and Tm2 and to the planar lipid arrangement (di Figure S 6d) in the rigid flat 352

lamellar structure below Tm2 Its temperature- hence hydration- independent position also 353

confirms its structural origin Finally besides the water-structure peak at about 2 Aring-1 we also 354

observe a broad peak at 144 Aring-1 characteristics of the lateral packing (d Figure S 6d) of lipids 355

in bilayers While its position could be consistent with an Lα phase5758 above Tm2 its broadness 356

and positional invariance as well as the coexistence between vesicles and tubes do not allow a 357

precise assignment 358

15

Does G-C181-OH assume a bilayer or interdigitated structure the latter being the one 359

found for other similar bolaform glucolipids4243 The size of G-C181-OH can be estimated to 360

be at the most 32 Aring given by the length L of the C16 tail (L = 22 Aring from the Tanford formula 361

L= 154 + 1265 times n (n= 16)59) and the sugar (about 10 Aring)4243 However the cis conformation 362

of the C=C bond imposes a ldquokinkrdquo of 60deg in the aliphatic chain which reduces the effective 363

length of the oleic backbone to about 17 Aring for an effective total G-C181-OH length of about 364

27 Aring Cryo-TEM provides a tube thickness varying from 35 Aring to 60 Aring and a vesicle membrane 365

thickness of at least 60 Aring WAXS suggests an interlamellar thickness (membrane and water) 366

between 40 Aring and 37 Aring and an inter-lipid distance in the tube of about 16 Aring In the absence of 367

a clear form factor oscillation profile as found for similar glucolipids4243 one must 368

acknowledge that these data pledge neither in favor of fully interdigitated nor a double layer 369

structure and might suggest a tilted lipid arrangement with partial interdigitation60 370

371

13C ssNMR demonstrates that G-C181-OH can assume two coexisting head-head and 372

head-tail configurations 373

Vesicles are characterized by low positive mean and gaussian curvatures while 374

nanotubes are described by a combination of a high positive mean and zero gaussian curvatures 375

In this regard the coexistence of vesicles and nanotubes requires glucolipids to adapt 376

themselves to both environments at the same time For this reason the formation of nanotube-377

vesicle networks generally requires a specific driving force able to impose a high mean and 378

zero gaussian curvature to the membrane In many systems insertion of at least a second 379

constituent (lipid surfactant protein) in the membrane becomes a necessary internal driving 380

force for the nucleation and growth of so-called membrane inclusions from which 381

topologically-connected nanotubes can form10ndash123261 as explained by orientational ordering 382

arguments62 It has also been shown that external driving forces such as use of laser 383

ldquotweezersrdquo30 osmotic pressure6 electroformation5ndash82829 or controlled suction13ndash16 are 384

otherwise necessary as also supported by theoretical studies19 385

The bending energy of tubes is higher than that of a vesicle membrane8 due to the 386

dependence of the membrane free energy from the principal curvatures For this reason 387

isotropic (rotational symmetry along the main axis) but also anisotropic33 lipids generally 388

stabilize low-curvature shapes like flat membranes or vesicles Lipid nanotubes are also a 389

common phase formed by amphiphiles but generally for lipids with a certain complexity in 390

their molecular structures as exampled by diacetylenic derivatives of phospholipids among 391

others262763ndash65 Negative contributions to the bending energy favoring the simultaneous 392

16

presence of interconnected vesicles and tubes was shown to be driven by the presence of 393

anisotropic lipids in a membrane continuum Spontaneous deviations to the elastic theory in the 394

presence of surfactants lipids and proteins with anisotropic structures favor the formation of 395

protrusions with radius of the order of the fraction of a micron9ndash1219ndash2362 396

In the absence of both internal and external driving forces mentioned above G-C181-397

OH should behave as a standard lipid Its DSC profile (Figure 2)536667 under both hydrated and 398

dry conditions identifies a hydration-independent transition (Tm1) slightly above room 399

temperature this is followed by one main transition at higher temperature (Tm2) in the dry 400

sample50 For such a profile one expects a gel phase (Lβ) below Tm1 and a liquid crystalline 401

structure possibly in the morphology of vesicles (L4 phase) possibly above Tm1 and most likely 402

above Tm231 as experimentally found On the basis of the thermogram recorded on the dry 403

powder one may also expect a ldquoripplerdquo (Pβ) phase between Tm1 and Tm2 characterized by 404

periodic undulations of the bilayer (generally addressed to as regions with both liquid order and 405

disorder)5455 406

In the present work instead of a Lβ phase we observe flat crystals below Tm1 and a 407

vesicle-nanotube network instead of the ldquoripplerdquo phase The latter is particularly unexpected 408

because the membrane is composed of a single-component (G-C181-OH) and no anisotropic 409

external force (osmotic hydrodynamic electric) is used to pull and stabilize the tubes In 410

addition the nanotubes are observed at fairly high temperatures above Tm1 (between about 411

60degC and 90degC) while this morphology is generally favored at low temperature (proportionality 412

with 1kT figure 5 in Ref 62) commonly below a given transition temperature characteristics 413

for each specific amphiphile27 Furthermore nanotubes of G-C181-OH seem to be stable in 414

time thus differing from what is found in electroformed vesicles78 They are also observed 415

away from phase transition events (may them be at Tm1 or Tm2) as otherwise reported before 416

for a diacetylenic derivative of phosphatidylcholine26 and many other systems27 417

A stable nanotube-vesicle network for a single-component phospholipid in the absence 418

of external driving forces is not expected and has never been reported before It could be 419

explained by the coexistence of two micro-separated phases of different rigidity This 420

hypothesis is excluded by confocal microscopy on Laurdan-labelled membranes and magic 421

angle spinning (MAS) solid-state NMR (ssNMR) experiments 422

Laurdan is a well-known fluorophore used to put in evidence the coexistence of rigid 423

and fluid domains in both model and biological membranes47ndash49 Laurdan was used here to label 424

the G-C181-OH membranes between Tm1 and Tm2 and to verify the possible micro-segregation 425

and coexistence of rigid (eg Lβ λem= 440 nm) and fluid (eg Lα λem= 490 nm) domains in 426

17

tubes and vesicles respectively A set of quantitative experiments performed on a 5 mgmL-1 427

solution summarized in Figure 5 shows an excess of fluorescence at only λem= 490 nm for 428

temperature between 90degC and 70degC (Figure 5h) The GP value which quantifies the rigidfluid 429

character of the membrane is systematically negative (Figure 5i) indicating an overall fluid 430

membrane at all temperatures GP is quantified on more than 50 regions of interest (ROIs) at 431

each temperature including vesicles and tubes alone as well as connected nanotube-vesicles 432

(Figure 5d-g) The dispersion of GP at 80degC and 70degC between -08 and -02 could suggest the 433

spurious presence of membranes with various degrees of flexibility although fluidity is always 434

predominant The combination of WAXS (broad peak d= 144 Aring-1 invariant with 435

temperature Figure S 6bc) and confocal fluorescence microscopy using Laurdan confirms the 436

single-phase of the membrane composed of G-C181-OH both in vesicles and nanotubes This 437

is also consistent with the lack of phase transitions in DSC experiments between ~85degC and 438

~60degC especially in hydrated samples We conclude that the vesicle-nanotube transition is not 439

driven by the local segregation of gel phase (Lβ) microdomains nor that the nanotubes are 440

composed of such phase 441

ssNMR is the best-suited approach to investigate the origin of the local driving force for 442

the nucleation of anisotropic inclusions of which the average orientation depends on the local 443

membrane and inclusion curvatures and was shown to promote tubing62 ssNMR can provide 444

information on the local molecular mobility eg through the nuclear spin relaxation behavior 445

but also on intermolecular interactions and stereochemistry eg through variations in the 446

chemical shift68ndash70 447

Ideally ssNMR should be performed on diluted G-C181-OH samples above Tm1 448

However fast MAS necessary for high resolution centrifuges the sample As a first 449

approximation hydrated G-C181-OH samples for instance at 50 wt as classically done for 450

phospholipids71 could be studied Unfortunately permeability of the sample holder combined 451

to heating (close to the boiling point of water) and MAS provoke dehydration and even sample 452

ejection After experiencing both of the above this approach was abandoned As a second 453

approximation the study of the dehydrated G-C181-OH powder avoids centrifugation and 454

evaporation If studying dried lipids may not reflect their exact behaviour under diluted 455

conditions valuable structural information can still be obtained in dehydrated molecular 456

systems like proteins thus illustrating that this approach is not unreasonable Furthermore if 457

DSC (Figure 2a) shows that hydration suppresses the Tm2 it also shows that the value of Tm1 is 458

not affected by water at concentrations as low as 10 wt For this reason performing a 459

structural analysis of dehydrated G-C181-OH by ssNMR above Tm1 may help understanding 460

18

the local arrangement of G-C181-OH in two morphologies vesicles and nanotubes having 461

sensibly different curvatures 462

To explore the local mobility of the most relevant chemical groups of G-C181-OH 463

above Tm2 (T= 110degC) and between Tm2 and Tm1 (70 lt T degC lt 90) we combine a series of 1H 464

and 13C experiments Single-pulse (SP) MAS experiments (1H and 13C) when recorded under 465

quantitative conditions allow the detection of all chemical species while cross-polarization 466

(CP) MAS (13C only) experiments are sensitive only to constrained groups for which the 1H-467

13C dipolar coupling is not averaged to zero by the molecular motions 468

469

Figure 6 ndash a) 1H MAS (10 kHz) ssNMR spectra recorded from 110degC to 70degC b) 13C MAS (10 kHz) ssNMR 470

spectra recorded using high power decoupling single pulse (SP) and cross-polarization (CP) pulse sequences 471

Contact time for CP experiments and corresponding temperatures are given on each spectrum Full spectra 472

are given in Figure S 7 473

The most relevant 1H and 13C spectra are given in Figure 6 (full range in Figure 7 and 474

Figure S 7) As expected the 1H solid-state SPMAS spectra show broader peaks compared to 475

the solution-state NMR spectrum (Figure 6a) reflecting the obvious solid-state character of the 476

sample at any temperature The sugar group and the C=C bond resonate in the 25-6 ppm range 477

and 535 ppm respectively while the OH groups of the sugar moiety are found below 45 ppm 478

(Figure 6a full assignment in Table S 1) Very broad (~1 ppm) at 100degC the position of the 479

OH peak varies with the temperature value and becomes narrower (lt~05 ppm) and visible at 480

δH= 464 ppm at T= 70degC Line narrowing and chemical shift variations of the OH resonance 481

in ssNMR is always a direct sign of H-bonding interactions whereas the strength of the H-482

bonding hence its OHhellipH distance is correlated to a dispersion in chemical shifts72ndash75 All other 483

60 55 50 45 40 35 30 25

ppm

110

90

80

70

910

1rsquo

6rsquo16

1

OH

134 132 130 128 126

ppm110 108 106 104 102 100

ppm

SP (T= 110degC)

SP (T= 70degC)

CP 50 ms (T= 70degC)

CP 3 ms (T= 70degC)

910 1rsquo

1H 13C

1

811

910

CH2 CH2

1rsquo

6rsquo

a) b)

C102

13024

C9101

13003

C92

12869 C1rsquo210382

C1rsquo110443

C1rsquo13035

2rsquo3rsquo4rsquo

T degC

5rsquo

19

resonances broaden from 110degC to 70degC thus indicating slower andor more anisotropic 484

reorientational motions of all chemical groups The specific dynamic behaviour of the sugar 485

headgroup is also reflected by its 1H transverse relaxation time T2 whereas the shorter the T2 486

the higher the rigidity Experiments performed between 110degC and 70degC shown and 487

commented in Figure S 8 in the Supporting Information globally show a single short (lt 1 ms 488

below 100degC) T2 relaxation component for H1rsquo H5rsquo and H6rsquo and two both short and long 489

(between 4 ms and 12 ms) T2 for the acyl chain (CH2 H910 H811) even below 100degC 490

At 110degC 13C SPMAS experiments (Figure 6b full range in Figure 7 and Figure S 7 491

assignments in Table 1) show a single peak at δC= 10335 ppm for C1rsquo and at δC= 13003 ppm 492

for C910 as expected for a symmetrical conformation with two chemically-equivalent 493

carbons76 At 70degC the C1rsquo peak disappears while C910 now shows two peaks around 128 494

ppm and 130 ppm (Table 1 full temperature series in Figure S 7b) attributed to C92 (δC= 12869 495

ppm) and C10 in a non-symmetric configuration respectively for which these two carbons are 496

chemically inequivalent76 More precisely the resonance at 130 ppm is actually the 497

superposition of two close resonances C9101 (δC= 13003 ppm) and C102 (δC= 13024 ppm) 498

where subscripts 1 and 2 stand for two different conformations of the cis-olefin group as 499

reported for various phases of oleic acid in the vicinity of 0degC76 500

501

Table 1 ndash 13C Chemical shifts of G-C181-OH in solution and solid-state Solid-state experiments are 502

recorded at MAS= 10 kHz All spectra are reported in Figure 7 and Figure S 7 The contact time for cross 503

polarization is 3 ms In all ssNMR experiments temperature is first raised at 110degC and then lowered to 504

either 90degC or 70degC Attribution of the 13C resonances in Table 1 is performed on the basis of the solution 505

spectrum (Figure S 1b) Additional peaks observed in both single pulse and cross polarization ssNMR 506

experiments are attributed when possible to a configuration effect of the given atom eg C9101 and C92 507

indicate two configurations of C9 All new peaks observed in solid-state and not detected in solution are 508

highlighted in red for convenience 509

510

511

δ ppm

Peak Liquid (DMSO-d6)

Solid Single Pulse

T= 110degC

Solid Single Pulse

T= 90degC

Solid Single Pulse

T= 70degC

Solid Cross Polarization

T= 70degC

C102

- - 13024 13024

C910C9101 13009 13003 13003 13003 13003

C1C9C10C1rsquo

C6rsquo

20

C92 - - - 12869 12869

C1rsquoC11 10329 10335 10335 - 10443

C12 - - - - 10382

C3 7724 7709 7702 - 7829

- 7634 7629 - 7621

7525

C2 7388 7398 7390 7370 7370

C4 7052 7077 7054 7041 7098

C16 6900 7016 - - 7029

- - - - 6793

C1C5C6 6153 6241 6236 6239 6262

C1C5C6 6117 - - - 5956

C2 3300 3290 3290 3283 3279

CH2(trans) - - - 3198 3196

CH2(gauche)C15 2941 2981 2987 3004 -

C811 2706 2747 2747 2832 2817

CH2 2597 2625 2628 2630 2817

512

13C CPMAS spectra at 70degC (Figure 6b full range in Figure 7 and Figure S 7d) at 513

increasing contact times confirm the presence of two cis-olefin environments one for which 514

C910 are equivalent labelled here C9101 and one for which they are inequivalent labelled 515

C92 C102 Both environments are rigid enough to be observed at short contact time (tc= 50 μs) 516

A similar double environment is also observed for the sugar C1rsquo CPMAS at 70degC shows two 517

peaks (C1rsquo1 at δC= 10443 ppm C1rsquo2 at δC= 10382 ppm Figure 6b Figure S 7d) These are 518

significantly shifted downfield from the C1rsquo peak observed at 110degC and are detected at contact 519

times as short as 50 μs thus indicating their rigid character One also finds that CPMAS is 520

totally inefficient at 100degC for short contact times (50 μs) and poorly efficient even at 3 ms 521

typical of fast-tumbling molecules (Figure S 7c) also both gauche (δC= 3004 ppm) and all-522

trans (δC= 3198 ppm) conformations of the CH2 carbons coexist at 70degC (Figure S 7b)77 523

According to ssNMR one main mobile molecular environment exists at 110degC as 524

shown by the well-resolved 1H spectrum single C=C (C9101) and anomeric (C1rsquo) resonances 525

gauche conformation of the aliphatic CH2 and single T2 of the aliphatic CH2 chain and sugar 526

protons The sugar group is certainly mobile (inefficient 1H13C CPMAS) but still more 527

constrainted than the acyl chain (shorter T2 (1H)) As the temperature is lowered the overall 528

rigidity increases (broadening of 1H NMR peaks faster 1H transverse relaxation and efficient 529

1H13C magnetization transfer during CPMAS) presumably driven by H-bonding 530

interactions (downfield shift and line narrowing of the OH 1H signal very short OH T2 loss of 531

the C1rsquo signal in SPMAS) At the same time a number of spectroscopic data (two concomitant 532

21

resonances for anomeric C1rsquo1 and C1rsquo2 cis-olefin C9101 and C92 C102 and CH2 gauche and 533

all-trans environments) and relaxation behaviour (fast and slow T2 relaxation components) 534

strongly suggests the coexistence of two molecular environments Further proofs of the 535

coexistence of two distinct molecular domains are given by a number of chemical shift 536

variations in the 13C CPMAS spectrum at 70degC compared to the one at 110degC 90degC and 537

solution This is shown in Figure 7 and in the list of 13C chemical shifts in Table 1 538

539

540

Figure 7 - Comparison between solution- and solid-state 13C NMR spectra of G-C181-OH Solid-state 541

experiments are recorded directly on the sample powder at a MAS of 10 kHz (4 mm zirconia rotor) SP 542

Single Pulse CP Cross Polarization In all the ssNMR experiments temperature is first raised at 110degC 543

and then lowered to either 90degC or 70degC Segmented lines are meant to highlight the known peak position 544

in solution thus stressing the strong chemical shift variations in the solid-state 545

546

NMR experiments above Tm2 suggest that vesicles could contain G-C181-OH with a 547

single family of mobile conformations characterized by chemically-equivalent C=C (red 548

Figure 8b) and a single C1rsquo conformation (green Figure 8b) Experiments between Tm1 and Tm2 549

otherwise suggest that two coexisting conformations characterized by chemically-equivalent 550

(red Figure 8b) and inequivalent (magenta Figure 8b) C=C could be associated to nanotubes 551

Two distinct C1rsquo sites (grey Figure 8b) also indicate two configurations of the sugar headgroup 552

A comment on this point will be given in the following paragraphs 553

The local (Aring-scale) variation of molecular mobility below Tm2 drives the coexistence of 554

two distinct molecular environments of the same G-C181-OH molecule This occurs within a 555

single-phase domain (as deduced by WAXS) of equivalent membrane rigidity (as derived from 556

Laurdan probe in confocal microscopy) Although never reported before the difference in the 557

local mobility of the same amphiphile (here G-C181-OH) could satisfy the theoretically-558

140 120 100 80 60 40 20

CP(3 ms)-70degC

SP-70degC

SP-90degC

SP-110degC

ppm

LiquidC910

C1rsquo

C2rsquo

C4rsquo

C3rsquo

C1

C6rsquoC5rsquo

C1

6

C8

11

C2 C

H2

C1

5

CH

2

DM

SO

13C

22

required conditions that induce local lipid segregation and formation of membrane inclusions 559

(Figure 8c) In case of their non-zero average orientation the local membrane curvature 560

increases62 and nanotubes can spontaneously form (Figure 8bc) More intriguing this 561

mechanism occurs at least 10degC above Tm1 although it should not be favored above the Tm27 562

From both experimental and theoretical backgrounds the conditions of topologically-563

connected vesicles-nanotubes systems are met only when either internal (two components) or 564

external driving forces are applied19 In the present single-lipid system of which the phase 565

behavior is summarized in Figure 8a the most plausible driving force could be explained by 566

the NMR results collected on the dry sample G-C181-OH concomitantly occupies two 567

molecular environments characterized by 1) two distinct conformations of the sugar 568

displaying equivalent mobility and 2) two conformations of the acyl chain unsaturation 569

570

23

571

Figure 8 ndash a) Thermotropic phase behavior of G-C181-OH at 5 mgmL-1 Note that Tm2 is observed on the 572

dry powder only b) Possible relationship between the molecular environment as derived from NMR and 573

membrane curvatures All the molecular drawings refer to the same G-C181-OH molecule (bulky ellipsoid 574

equiv sugar headgroup) while each color corresponds to a given configuration Grey of C1rsquo1 and C1rsquo2 indicates 575

the uncertainty about the assignment of the specific NMR peak The curvature C is defined as 1R R being 576

the curvature radius R1 R2 radius of curvature CG Gauss curvature CM Mean curvature c) Free 577

interpretation of the anisotropic high-curvature (CMgt 0 CG= 0) inclusions in the membrane (low CM and 578

CG) continuum driven by inter-sugar H-bonding and reduced intramembrane diffusion for a G-C181-OH 579

glucolipid 580

581

Flat crystalline structure

Nanotube-vesicle network

Vesicles

48degC

120deg

106 105 104 103 102 101

ppm

H9 H10H1rsquo

106 105 104 103 102 101

ppm106 105 104 103 102 101

ppm106 105 104 103 102 101

ppm

C=CC=C

C1rsquo1C1rsquo2

C1rsquo

Tm2(96degC)Tm1

a)

b)

c)

Same molecule different local orientations and sugar conformations

High curvature membrane

Low curvature inclusion

Flip-FlopStrong H-bonding

CG

CM ( + )

CG 0

CM 0 CG 0

CM 0

G-C181-OH

24

Two related questions are unanswered at the moment What actually drives nanotube 582

formation and is the organization of G-C181-OH within the vesicles and tubes the same Lipid 583

nanotubes have been studied for at least three decades over which broad experimental and 584

theoretical insight has been generated Despite this fact the actual understanding is still matter 585

of debate27 Micelle-to-nanotube and vesicle-to-nanotube phase transitions are generally 586

reported for single amphiphiles below a given Tm and intermediate phase composed of twisted 587

and flat fibers is concomitantly reported Initially thought to be a chirality-driven phenomenon 588

recent studies propose a more general mechanism involving crystallization and symmetry 589

breaking in the lipid packing driven by optimal packing27 If twisted ribbons were observed 590

before for a number of microbial glycolipid bolaamphiphiles44 they are not found for G-C181-591

OH This fact as well as the large body of cryo-TEM images and the observation of nanotubes 592

above Tm1 rather pledge in favour of a direct topological change from positive (vesicles) to 593

zero (nanotubes) Gaussian curvatures In this case vesicle-to-nanotube transitions could be 594

driven by external forces56307813ndash162829 595

Considering the absence of external forces in the present system there are only two 596

possibilities left spontaneous budding of the membrane or nanotube pulling after inter-vesicle 597

collisions due to Brownian motion Both are possible as inclusions composed of a different 598

arrangement of the same molecule in the membrane continuum can theoretically explain 599

spontaneous nanotube formation in an isotropic medium672223629ndash1218ndash21 However the second 600

one is probably more likely to prevail because it could explain the large number of nanotubes 601

connecting two distant vesicles Interestingly similar events occur for any phospholipid-based 602

vesicular solution but nanotubes are extremely rare unless at least two lipids of different 603

molecular structure are mixed together The spectroscopic 1H and 13C and relaxation (1H) 604

ssNMR data even if collected on a dry powder demonstrate that the coexistence of two G-605

C181-OH configurations involving both the sugar headgroup and fatty alcohol chain are 606

possible for this molecule As a matter of fact this is not so surprising as more than one energy 607

minima each corresponding to a given conformation are known for glucose its non-reducing 608

derivative methylglucose and carbohydrates in general78ndash80 609

The presence of two configurations could explain the coexistence between two 610

structures of different curvature but the data presented so far do not explain the origin of these 611

two configurations In other words what drives the change in the local structure of the sugar 612

headgroup and fatty alcohol within the membrane 613

The association of known self-assembled structures to spectroscopic 13C ssNMR data is 614

known for years in particular for gluconamides lipids having a linear sugar headgroup81ndash84 It 615

25

was found that gauche effects in some cases up to the γ position in the sugar chain induce 616

upfield chemical shifts while anti and trans conformations result in downfield shifts Each 617

structure (monolayer crystal bilayer crystal or micellar fiber) was then characterized by a given 618

organization of the gluconamide and a characteristic 13C ssNMR fingerprint which could be 619

explained by gauche γ-gauche anti and trans conformations of the headgroup Similar 620

structural-spectroscopic correlations have been collected over the years for much simpler cyclic 621

sugars85 whereas pyranoses can undergo important conformational changes (chair half-chair 622

boat skew) when the appropriate energy barriers are crossed78ndash8086 A quantitative 623

conformational study based on the 13C chemical shift variations shown in Table 1 would be 624

certainly helpful but very delicate in the context of the present state of the art because a 625

rigorous approach requires at least one known crystal structure8182 which is currently missing 626

for this compound A qualitative approach will then be adopted 627

G-C181-OH is a bolaform amphiphile and flip-flop could be a possible mechanism 628

involved in the nanotube stabilization whereas head-tail interactions have long been considered 629

as important in nanotube-forming bolaform amphiphiles6063 From mere considerations based 630

on T2 (1H) relaxation arguments one could exclude head-tail configurations if the R-CH2OH 631

endgroup (H1) was involved in a direct interaction with the sugar group its corresponding T2 632

would probably be comparable to the sugarrsquos one (OH and H1rsquo) but this does not seem to be 633

the case However this hypothesis would be based on the comparison of the relaxation from 634

CH and CH2 protons additionally involved in different kind of motions Besides the relaxation 635

of H1 is influenced by the strong overlap between H1 and H5rsquo For these reasons relaxation 636

arguments are not solid enough and will not be used 637

We then address this question by exploring intermolecular interactions in a direct 638

manner using 2D 1H-1H homonuclear ssNMR correlation spectroscopy At 90degC in the fast-639

tumbling domain of molecular motion (narrow 1H NMR peaks Figure 6a) and corresponding 640

to the vesicle-rich region spin diffusion (NOESY) experiments are preferred over dipolar 641

coupling based experiments because such interaction is averaged out by molecular motion 642

However when the molecular motion is more constrained in the domain of nanotube stability 643

at 70degC (broad 1H spectra Figure 6a) Double-Quantum Single-Quantum (DQ-SQ) correlation 644

experiments are best suited to probe interaction between close molecular neighbors These 645

experiments are reported in Figure 9 and the related interpretation is only briefly reported 646

below while a more detailed discussion can be found on Page 12 in the Supporting Information 647

26

648

Figure 9 ndash a) 2D 1H-1H NOESY NMR experiment performed at MAS= 10 kHz (4 mm probe) on a Bruker 649

300 MHz Avance III HD spectrometer using 300 ms of mixing time Sample temperature is regulated to 650

90degC reached by cooling from 110degC b) 2D 1H-1H DQ-SQ (BAck-to-BAck BABA pulse sequence used for 651

the DQ excitation scheme) NMR experiment performed at MAS= 22 kHz (25 mm probe) on a Bruker 700 652

MHz Avance III spectrometer using a DQ excitation time equal to 2 rotor periods Sample temperature is 653

regulated to 70degC reached by cooling from 110degC The typical schemes illustrating how to interpret these 654

2D experiments are reported above each experimental map Zooms of the sugar region and configurational 655

interpretation are given below each 2D map More information are given in the text below 656

H16-H16

H1-H1

H9-H10

H1rsquo-H1

H1rsquoH910 H16

H1

H1rsquo-H1rsquo

Hxrsquo-Hxrsquo

2

(pp

m)

(ppm)

315

811910 1rsquo

2rsquo2

3rsquo 4rsquo6rsquo

5rsquo

1rsquo2910

asymp asymp asympHDO MeOD Acetone

1

315

811 9101rsquo

2rsquo2

3rsquo4rsquo6rsquo

5rsquo

1rsquo2 9 10

asympasympasympHDOMeODAcetone

H1rsquo-H1

1rsquo

1

(p

pm

)

(ppm)

H1-H2

H1rsquo-H1

H1rsquo

H910

H1H5rsquo

H2

H1rsquo-H5rsquo

315

811910 1rsquo

2rsquo2

3rsquo 4rsquo6rsquo

5rsquo

1rsquo2910

asymp asymp asympHDO MeOD Acetone

1

H2rsquoH3rsquoH4rsquo

H2rsquoH3rsquoH4rsquo

X= 2rsquo3rsquo4rsquo

NOESY - T= 90degC

1H

1H

1H

1H

DQ-SQ - T= 70degC

a) b)

2

(pp

m)

(p

pm

)

(ppm)

A B

AhellipB

AhellipA

BhellipB

NOESY

(ppm)

A B

AhellipB

DQ-SQ

Single-quantum dimension

Do

ub

le-q

ua

ntu

m d

imen

sio

n

AhellipA

BhellipB

BhellipA

BhellipA

315

811910 1rsquo

2rsquo2

3rsquo 4rsquo6rsquo

5rsquo

1rsquo2910

asymp asymp asympHDO MeOD Acetone

1

lt 5 Aring

lt 5 Aring

H1-H2

H811

H15

H16

H6rsquo

H1rsquo

H1

H5rsquo

H2

H1-H2

H1rsquo-H1

H1rsquo-H5rsquo

H1rsquo

H1 H5rsquo

H1rsquo

H1

H5rsquo

H1H2rsquo H3rsquo H4rsquo

H1-H1

H1rsquo-H1

Hxrsquo-HxrsquoX= 2rsquo3rsquo4rsquo

H1-H2

H1rsquo

Zoom sugar region 28 ndash 45 ppmNOESY DQ-SQ

27

Head-tail or head-head configurations could be probed by the existence of H1rsquo-H1 657

interactions whereas H1 (δH= 356 ppm) is associated to the glucolipid tail and H1rsquo (δH= 432 658

ppm) to the sugar headgroup These atoms are more than 15 Aring apart and a direct interaction 659

could not be observed in both NOESY and DQ-SQ experiments NOESY experiments (Figure 660

9a) show that H1rsquo-H1 correlations are not observed Meanwhile clear-cut direct H1rsquo-H1 661

interactions are indeed observed are detected at T= 70degC (DQ cross-peaks related to H1rsquo-H1 662

correlation in DQ-SQ experiment shown in Figure 9b) Interestingly the absence of auto peaks 663

at δH= 432 ppm and δH= 335 ppm in the SQ dimension excludes H1rsquo-H1rsquo and Hxrsquo-Hxrsquo (x= 2 664

3 4) interactions respectively strengthening the idea that sugar headgroups from two nearest-665

neighbor G-C181-OH molecules are not adjacent in the nanotube-vesicle region 666

Homonuclear 2D ssNMR supports the idea according to which G-C181-OH can 667

undergo molecular flip-flops and a head-tail arrangement in the dry product Extrapolating these 668

conclusions in the lower concentration range lack of H1rsquo-H1 proximity at T= 90degC suggests a 669

head-head configuration in the vesicle-rich region and head-tail configuration within the 670

nanotubes It is however still unclear whether flip-flops are driven by spontaneous temperature-671

induced conformational changes in the glucolipid packing stimulating inclusions within the 672

membrane itself672223629ndash1218ndash21 or by defects generated by inter-vesicle collisions In fact a 673

combination of both is not unlikely Nanotubes are not stable when temperature is 20 to 30 674

degrees above Tm1 indicating that inter-vesicle collisions alone cannot explain the unusual 675

phenomenon In an energy landscape approach87 the head-head configuration is certainly 676

favored at high temperature while both head-head and head-tail configurations could be 677

favored in the mid-temperature region corresponding to the range between Tm1 and Tm2 in the 678

dry powder Both statistical and defect-induced flip-flop events contribute to form high mean 679

zero Gaussian curvature morphologies (nanotubes) for which the head-tail conformation 680

limiting the proximity between bulky sugar headgroups is probably more adapted (Figure 8c) 681

The flip-flop interpretation now reasonably explains the strong differences in the sugar and fatty 682

alcohol conformations deduced by the multiple NMR signals attributed to chemically-683

equivalent groups The existence of various low-energy conformations each having specific 684

NMR signature for sugars in solution78ndash8086 and solid-state81ndash84 further supports our 685

interpretation 686

One last point should be clarified Why between Tm1 and Tm2 are two molecular 687

environments able to coexist instead of a single one of average rigidity as found in the case of 688

ldquoripplerdquo phases50ndash54 We believe that this question can be answered based on the NMR results 689

the relatively short T2 values of the H1rsquo and OH protons shows the constrained mobility of the 690

28

sugar headgroup despite a certain flexibility of the chain (long T2 component) This trend may 691

suggest restrictions in translational mobility probably due to the sugar-OH hydrogen bonding 692

(Figure 8c) meaning that lipids are not able to diffuse within the membrane thus keeping two 693

different configurations stable 694

695

Conclusions 696

Topologically-connected nanotube-vesicle lipidic systems are commonly observed for 697

self-assembled membranes composed of at least two amphiphiles and spontaneously driven by 698

anisotropic inclusions andor by external directional forces This work shows that strong 699

differences in membrane curvature can spontaneously occur in single-lipid systems without 700

external anisotropic forces Coexistence and interconnection between nanotubes and vesicles 701

are evidenced via cryo-TEM and fluorescence confocal microscopy in a single-phase region at 702

temperature above a transition temperature Tm1= 483degC determined by DSC on the hydrated 703

sample but below the second transition temperature Tm2 (957degC) determined by DSC on the 704

dry powder X-ray diffraction (WAXS) also shows the coexistence of two lattice periods one 705

attributed to an inter-membrane repeating distance and the other possibly related to the intra-706

nanotube glucolipid arrangement Labelling the systems with Laurdan a fluorophore of which 707

the emission strongly varies with the local membrane rigidity excludes the presence of rigid 708

micro-separated phases (eg Lβ) 709

1H and 13C ssNMR despite the fact that it can only be performed on the dry powder 710

confirm the presence of one main fluid G-C181-OH environment above Tm2 Below Tm2 the 711

splitting of some 13C peaks and the coexistence of short and long 1H relaxation components 712

demonstrate that G-C181-OH can simultaneously pack into structurally and dynamically 713

inequivalent environments Combining advanced 2D 1H-1H correlation experiments based on 714

spin diffusion or double-quanta excitation suggests that vesicles could contain a preferred head-715

head configuration of the bolaform lipid while nanotubes a preferred head-tail assembly 716

The presence of two environments composed of the same molecule characterized by 717

different molecular configurations and rigidity fulfills the theoretical conditions necessary to 718

spontaneously nucleate a membrane inclusion of non-zero average orientation described as 719

necessary to observe topologically-connected nanotube-vesicle structures in the absence of 720

external forces This hypothesis could also be enriched by the possibility of nanotubes to be 721

pulled away between two vesicles after a collision event 722

This work expands the actual theoretical and experimental framework on lipid networks 723

stabilized by a different radius of curvature It could also motivate the synthesis or isolation 724

29

of new lipids providing a higher level of 3D architectural control in soft networks possibly in 725

the range of room temperature 726

727

Acknowledgements 728

Dr Andrea Lassenberger and Dr Sylvain Preacutevost at Institut Laue Langevin (ILL Grenoble 729

France) are kindly acknowledged for their assistance on the SANS experiment Confocal 730

images were performed at the Institut de Biologie Paris Seine (IBPS) imaging facility Jean-731

Franccedilois Gilles (Imaging Core Facility Institut de Biologie Paris Seine (IBPS) CNRS 732

Sorbonne Universiteacute Paris France) is kindly acknowledged for his assistance on the confocal 733

microscopy imaging Dr Guillaume Laurent (LCMCP Sorbonne Universiteacute Paris France) is 734

kindly acknowledged for helpful discussions The French region Ile-de-France SESAME 735

program is acknowledged for financial support (700 MHz NMR 736

spectrometer) Dr Olivier Diat (ICSM Marcoule France) Prof eronikaIglič and Prof V Aleš737

onsul discussi Slovenia) are kindly acknowledged for helpfUniversity of LjubljanaIglič (-Kralj 738

739

Financial support 740

WAXS experiments have been supported by Soleil Light Source Saint Aubin France proposal 741

Ndeg 20201747 SANS experiments have been supported by ILL proposal Ndeg9-13-778 742

743

References 744

1 D Kabaso N Bobrovska W Goacuteźdź E Gongadze V Kralj-Iglič R Zorec and A 745

Iglič Bioelectrochemistry 2012 87 204ndash210 746

2 C R Safinya U Raviv D J Needleman A Zidovska M C Choi M A Ojeda-747

Lopez K K Ewert Y Li H P Miller J Quispe B Carragher C S Potter M W 748

Kim S C Feinstein and L Wilson Adv Mater 2011 23 2260ndash2270 749

3 P Veranič M Lokar G J Schuumltz J Weghuber S Wieser H Haumlgerstrand V Kralj-750

Iglič and A Iglič Biophys J 2008 95 4416ndash4425 751

4 A Rustom R Saffrich I Markovic P Walther and H-H Gerdes Science (80- ) 752

2004 303 1007ndash1010 753

5 M I Angelova S Soleacuteau P Meacuteleacuteard F Faucon and P Bothorel Prog Colloid 754

Polym Sci 1992 89 127ndash131 755

6 L Mathivet S Cribier and P F Devaux Biophys J 1996 70 1112ndash1121 756

7 V Kralj-Iglič G Gomišček J Majhenc V Arrigler and S Svetina Colloids Surfaces 757

A Physicochem Eng Asp 2001 181 315ndash318 758

30

8 V Kralj-Iglič A Iglič G Gomišček F Sevšek V Arrigler and H Haumlgerstrand J 759

Phys A Math Gen 2002 35 1533ndash1549 760

9 M Drab D Stopar V Kralj-Iglič and A Iglič Cells 2019 8 626 761

10 A Zidovska K K Ewert J Quispe B Carragher C S Potter and C R Safinya 762

Langmuir 2009 25 2979ndash2985 763

11 A Zidovska K K Ewert J Quispe B Carragher C S Potter and C R Safinya Soft 764

Matter 2011 7 8363ndash8369 765

12 A Zidovska K K Ewert J Quispe B Carragher C S Potter and C R Safinya 766

Biochim Biophys Acta - Biomembr 2009 1788 1869ndash1876 767

13 M Karlsson K Sott M Davidson A S Cans P Linderholm D Chiu and O Orwar 768

Proc Natl Acad Sci U S A 2002 99 11573ndash11578 769

14 A Karlsson R Karlsson M Karlsson A-S Cans A Stromberg F Ryttsen and O 770

Orwar Nature 2001 409 150ndash152 771

15 M Karlsson K Sott A S Cans A Karlsson R Karlsson and O Orwar Langmuir 772

2001 17 6754ndash6758 773

16 E Evans H Bowman A Leung D Needham and D Tirrell Science (80- ) 1996 774

273 933ndash935 775

17 U Seifert Configurations of fluid membranes and vesicles 1997 vol 46 776

18 J Kaumls and E Sackmann Biophys J 1991 60 825ndash844 777

19 N Bobrovska W Goacuteźdź V Kralj-Iglič and A Iglič PLoS One 2013 8 e73941 778

20 A Iglič B Babnik K Bohinc M Fošnarič H Haumlgerstrand and V Kralj-Iglič J 779

Biomech 2007 40 579ndash585 780

21 K Bohinc T Slivnik A Iglič M Brumen and V Kralj-Iglič J Phys Chem C 2007 781

111 9709ndash9718 782

22 A Iglič H Haumlgerstrand P Veranič A Plemenitaš and V Kralj-Iglič J Theor Biol 783

2006 240 368ndash373 784

23 V Kralj-Iglič and P Veranič Adv Planar Lipid Bilayers Liposomes 2006 5 129ndash785

149 786

24 A Iglic and V Kralj-Iglic Adv Planar Lipid Bilayers Liposomes 2006 4 253ndash279 787

25 S Penič L Mesarec M Fošnarič L Mroacutewczyńska H Haumlgerstrand V Kralj-Iglič and 788

A Iglič Front Phys 2020 8 342 789

26 P Yager and P E Schoen Mol Cryst Liq Cryst 1984 106 371ndash381 790

27 T G Barclay K Constantopoulos and J Matisons Chem Rev 2014 114 10217ndash791

10291 792

31

28 B Božič G Gomišček V Kralj-Iglič S Svetina and B Žekš Eur Biophys J 2002 793

31 487ndash496 794

29 A Iglič H Haumlgerstrand M Bobrowska-Haumlgerstrand V Arrigler and V Kralj-Iglič 795

Phys Lett A 2003 310 493ndash497 796

30 R Bar-Ziv and E Moses Phys Rev Lett 1994 73 1392ndash1395 797

31 S Chiruvolu H E Warriner E Naranjo S H Idziak J O Raumldler R J Plano J A 798

Zasadzinski and C R Safinya Science (80- ) 1994 266 1222ndash1225 799

32 V Kralj-Iglič A Iglič H Haumlgerstrand and P Peterlin Phys Rev E 2000 61 4230ndash800

4234 801

33 L Mesarec M Drab S Penič V Kralj-Iglič and A Iglič Int J Mol Sci 2021 22 802

2348 803

34 L Van Renterghem S L K W Roelants N Baccile K Uyttersprot M C Taelman 804

B Everaert S Mincke S Ledegen S Debrouwer K Scholtens C V Stevens W 805

Soetaert L Van Renterghem S L K W Roelants N Baccile K Uytersprot M C 806

Taelman B Everaert S Mincke S Ledegen S Debrouwer K Scholtens C V 807

Stevens and W Soetaert Biotechnol Bioeng 2018 115 1195ndash1206 808

35 J Clark and F Deswarte Introduction to Chemicals from Biomass Second Edition 809

John Wiley amp Sons 2015 810

36 K V Sajna R Houmlfer R K Sukumaran L D Gottumukkala and A Pandey in 811

Industrial Biorefineries and White Biotechnology eds A Pandey R Houmlfer M 812

Taherzadeh K M Nampoothiri and C Larroche Amsterdam Oxford Waltham 813

Elsevier 2015 pp 499ndash521 814

37 R Marchant and I M Banat Trends Biotechnol 2012 30 558ndash565 815

38 K Hill and C LeHen-Ferrenbach in Sugar-Based Surfactants Fundamentals and 816

Applications ed C C Ruiz Boca Raton CRC Press 2009 pp 1ndash20 817

39 L Rodrigues I M Banat J Teixeira and R Oliveira J Antimicrob Chemother 818

2006 57 609ndash618 819

40 J D Desai and I M Banat Microbiol Mol Biol Rev 1997 61 47ndash64 820

41 S L K W Roelants K Ciesielska S L De Maeseneire H Moens B Everaert S 821

Verweire Q Denon B Vanlerberghe I N A Van Bogaert P Van der Meeren B 822

Devreese and W Soetaert Biotechnol Bioeng 2016 113 550ndash559 823

42 N Baccile M Selmane P Le Griel S Preacutevost J Perez C V Stevens E Delbeke S 824

Zibek M Guenther W Soetaert I N A Van Bogaert and S Roelants Langmuir 825

2016 32 6343ndash6359 826

32

43 N Baccile A-S Cuvier S Preacutevost C V Stevens E Delbeke J Berton W Soetaert 827

I N A Van Bogaert and S Roelants Langmuir 2016 32 10881ndash10894 828

44 N Baccile C Seyrig A Poirier S A Castro S L K W Roelants and S Abel Green 829

Chem 2021 23 3842ndash3944 830

45 A Salis L Cappai C Carucci D F Parsons and M Monduzzi J Phys Chem Lett 831

2020 11 6805ndash6811 832

46 J Schindelin I Arganda-Carreras E Frise V Kaynig M Longair T Pietzsch S 833

Preibisch C Rueden S Saalfeld B Schmid J-Y Tinevez D J White V 834

Hartenstein K Eliceiri P Tomancak and A Cardona Nat Methods 2012 9 676ndash835

682 836

47 T Parasassi and E Gratton J Fluoresc 1995 5 59ndash69 837

48 M M Dodes Traian F L Gonzaacutelez Flecha and V Levi J Lipid Res 2012 53 609ndash838

616 839

49 D M Owen C Rentero A Magenau A Abu-Siniyeh and K Gaus Nat Protoc 840

2012 7 24ndash35 841

50 W Stillwell in An Introduction to Biological Membranes Academic Press 2016 pp 842

135ndash180 843

51 W Stillwell in An Introduction to Biological Membranes Academic Press 2016 pp 844

181ndash220 845

52 B Tenchov Chem Phys Lipids 1991 57 165ndash177 846

53 R Koynova and M Caffrey Biochim Biophys Acta 1998 1376 91ndash145 847

54 F Schmid S Dolezel O Lenz and S Meinhardt J Phys Conf Ser 2014 487 848

012004 849

55 S Meinhardt R L C Vink and F Schmid Proc Natl Acad Sci U S A 2013 110 850

4476ndash4481 851

56 O Glatter and O Kratky Small Angle X-ray Scattering Academic Press London 852

1982 853

57 B A Cunningham A D Brown D H Wolfe W P Williams and A Brain Phys 854

Rev E 1998 58 3662ndash3672 855

58 M Nadler A Steiner T Dvir O Szekely P Szekely A Ginsburg R Asor R Resh 856

C Tamburu M Peres and U Raviv Soft Matter 2011 7 1512ndash1523 857

59 C Tanford The hydrophobic effect Formation of micelles and biological membranes 858

John Wiley amp Sons Inc New York 1973 859

60 M Masuda and T Shimizu Langmuir 2004 20 5969ndash5977 860

33

61 A Zidovska K K Ewert J Quispe B Carragher C S Potter and C R Safinya 861

Block Liposomes Vesicles of Charged Lipids with Distinctly Shaped Nanoscale 862

Sphere- Pear- Tube- or Rod-Segments Elsevier Inc 1st edn 2009 vol 465 863

62 A Iglic V Kralj-Iglic B Božič M Bobrowska-Haumlgerstrand B Isomaa and H 864

Haumlgerstrand Bioelectrochemistry 2000 52 203ndash211 865

63 T Shimizu M Masuda and H Minamikawa Chem Rev 2005 105 1401ndash1444 866

64 J V Selinger and J M Schnur Phys Rev Lett 1993 71 4091ndash4094 867

65 J M Schnur Science (80- ) 1993 262 1669ndash1676 868

66 B Tenchov R Koynova and G Rapp Biophys J 2001 80 1873ndash1890 869

67 R Koynova and B Tenchov Ser Biomech 2017 31 3ndash19 870

68 S J Angyal and R Le Fur Carbohydr Res 1984 126 15ndash26 871

69 J K Whitesell and M A Minton J Am Chem Soc 1987 109 225ndash228 872

70 T Takayama S Ando and I Ando J Mol Struct 1990 220 245ndash250 873

71 K Gawrisch N V Eldho and I V Polozov Chem Phys Lipids 2002 116 135ndash151 874

72 F Tielens C Gervais J F Lambert F Mauri and D Costa Chem Mater 2008 20 875

3336ndash3344 876

73 G A Jeffrey and Y Yeon Acta Crystallogr Sect B 1986 42 410ndash413 877

74 B Berglund and R W Vaughan J Chem Phys 1980 73 2037ndash2043 878

75 C Gervais C Coelho T Azaiumls J Maquet G Laurent F Pourpoint C Bonhomme 879

P Florian B Alonso G Guerrero P H Mutin and F Mauri J Magn Reson 2007 880

187 131ndash140 881

76 C Akita T Kawaguchi F Kaneko H Yamamoto and M Suzuki J Phys Chem B 882

2004 108 4862ndash4868 883

77 R N Purusottam L Seacutenicourt J J Lacapegravere and P Tekely Biochim Biophys Acta - 884

Biomembr 2015 1848 3134ndash3139 885

78 X Biarneacutes A Ardegravevol A Planas C Rovira A Laio and M Parrinello J Am Chem 886

Soc 2007 129 10686ndash10693 887

79 L M J Kroon‐Batenburg and J Kroon Biopolymers 1990 29 1243ndash1248 888

80 K N Kirschner and R J Woods Proc Natl Acad Sci U S A 2001 98 10541ndash889

10545 890

81 S Svenson B Kirste and J H Fuhrhop J Am Chem Soc 1994 116 11969ndash11975 891

82 I Sack S Macholl J H Fuhrhop and G Buntkowsky Phys Chem Chem Phys 892

2000 2 1781ndash1788 893

83 S Svenson J Koening and J H Fuhrhop J Phys Chem 1994 98 1022ndash1028 894

34

84 S Svenson A Schaefer and J H Fuhrhop J Chem Soc Perkin Trans 2 1994 2 895

1023ndash1028 896

85 K Bock and C Pedersen Adv Carbohydr Chem Biochem 1983 41 27ndash66 897

86 T B Grindley Structure and Conformation of Carbohydrates Springer Berlin 898

Heidelberg 2008 899

87 F Tantakitti J Boekhoven X Wang R V Kazantsev T Yu J Li E Zhuang R 900

Zandi J H Ortony C J Newcomb L C Palmer G S Shekhawat M O De La 901

Cruz G C Schatz and S I Stupp Nat Mater 2016 15 469ndash476 902

903

1

Supporting Information 904

905

Topological connection between vesicles and nanotubes in single-906

component lipid membranes driven by head-tail interactions 907

908

Niki Baccilea Ceacutedric Lorthioira Abdoul Aziz Baa Patrick Le Griela Cristina Coelhob Javier 909

Perezc Wim Soetaertd Sophie L K W Roelantsd 910

911

a Sorbonne Universiteacute Centre National de la Recherche Scientifique Laboratoire de Chimie de 912

la Matiegravere Condenseacutee de Paris LCMCP F-75005 Paris France 913

b Institut des Mateacuteriaux de Paris Centre Sorbonne Universiteacute Paris France 914

c Synchrotron Soleil LrsquoOrme des Merisiers Saint-Aubin BP48 91192 Gif-sur-Yvette Cedex 915

France 916

d InBio Department of Biotechnology Ghent University Ghent Belgium 917

918

919

Corresponding author 920

Dr Niki Baccile 921

E-mail address nikibaccilesorbonne-universitefr 922

Phone +33 1 44 27 56 77 923

924

925

926

2

a) 927

Parameter Method

Dry Matter (DM ) 996 Infrared balance 105degC

Glucose () 012 HPLC-Metacarb

Glycerol () nd HPLC-Metacarb

Free Fatty Acid content () lt001 Internal method BBEPP

GC with FID Detector

Oil () lt001 Internal method BBEPP

GC with FID Detector

Moulds (CFUg) lt10 3M-Nordval ndeg16 B

Yeasts (CFUg) lt10 3M-Nordval ndeg16 B

Anaerobic count (CFUg) lt10 SP-VG M005 B

Protein () tbd BCA protein assay

928

Figure S 1 - a) HPLC-ELSD chromatogram and composition table 929

930

931

932

933

3

b) 934

935

Group Ndeg δ ppm

CH2 1 354

CH2 2 153

CH2 3-712-14 132

CH2 811 203

CH 910 534

CH2 15 162

CH2 16 388

CH 1 424

CH 2 316

CH 3 334

CH 4 329

CH 5 354

CH2 6 366

Figure S 1 (next) ndash b) 1H NMR spectrum (MeOD-d4) recorded on the purified G-C181-OH compounds 936

and related assignment 937

938

2

811910 1rsquo

5rsquo1

3rsquo4rsquo

1rsquo1910

asymp asymp asympHDO MeOD Acetone

6rsquo2rsquo

1615

3-712-14

4

939

940

Figure S 2 ndash Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 130degC 941

5

942

943

Figure S 3 - Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 90degC 944

945

90degC

1

22

6

946

947

948

949

Figure S 4 ndash Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 70degC 950

951

21

11

1

2

5

3

4

45

5

6

7

7

7

952

953

954

Figure S 5 ndash Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 25degC 955

956

957

1

1 1

8

958

Figure S 6 ndash a) SANS experiment recorded at 90degC for a 5 mgmL G-C181-OH aqueous solution b-c) 959

WAXS experiments recorded on aqueous solutions of G-C181-OH at b) 5 mgmL and c) 25 mgmL d) 960

Structural interpretation of the repeating distances pointed out in c) 961

962

963

964

001 01 1

q Aring-1

25degC

68degC

88degC

111degC

01 1q Aring

-1

02 04 06 08 1q Aring

-1

001 01

10-1

100

101

Inte

nsity

cm

-1

q Aring-1

-2

T= 90degC

SANS WAXS WAXS

a) b) c)

dv

dL d= 144 Aring-1

di

ddH2O

dt

dvdL

dtdi d

d)

9

965

966

Figure S 7 ndash ssNMR MAS (10 kHz) experiments recorded on a G-C181-OH dry powder a) 1H spectra 967

recorded from 110degC to 70degC b) 13C single pulse (SP) spectra recorded with high power 1H decoupling from 968

110degC to 70degC c) 13C cross polarization (CP) spectra recorded at 110degC and contact time tc of 50 micros and 969

3000 micros d) 13C CP spectra recorded at 70degC and for 20 lt tc micros lt 3000 970

971

1

811

910

CH2 CH2

1rsquo

6rsquo

8 6 4 2 0 ppm

110

90

80

70

910

1rsquo

6rsquo16

1

2rsquo3rsquo4rsquo

811

OH

45 40 35 30 25 20 15

ppm

200 180 160 140 120 100 80 60 40 20 0

ppm140 120 100 80

ppm

CH2-gauche

3004

110

90

80

70

65

60

50

25

C91013003

C9 - 12869

CH2-all-trans

3198

C1rsquo10335

a) b)

1H

T degCT degC

13C Single Pulse MAS

A

B

A

B

70degC70degC

5rsquo

215

200 180 160 140 120 100 80 60 40 20 0

ppm

13C CPMAST= 110degC

200 180 160 140 120 100 80 60 40 20 0

ppm

132 131 130 129 128 127 126

ppm

106 105 104 103 102

ppm

13C CPMAST= 70degC

50

tc micros

3000

20

tc micros

50 500 1000 3000

tc micros

C102

13024 C910113003

C92

12869

C1rsquo213082

C1rsquo110443

A B

106 105 104 103 102

ppm

3000

1000

500

50

20

AB

c)

d)

10

972

Figure S 8 ndash a) Short and d) long T2 components displayed by the 1H transverse relaxation signal c) 1H 973

transverse relaxation signals recorded on a G-C181-OH dry powder from 110degC to 70degC The area of each 974

given resonance is plotted (log-lin) as a function of the echo time (CPMG pulse sequence) for each 975

temperature and fitted using the equation given above T2s and T2l indicate the short and long T2 relaxation 976

times respectively obtained by describing the 1H transverse relaxation signal using a biexponential decay 977

For H1rsquo H5rsquo and H6rsquo the 1H relaxation function could be described by a monoexponential decay 978

- Protons from the sugar group (H1rsquo H5rsquo H6rsquo) show a single T2 relaxation component (Figure 979

6c and Figure S 8) while protons of the acyl chain (CH2 H910 H811) show both short (Figure 980

6c) and long (Figure 6d) T2 relaxation component 981

- The short T2 relaxation time systematically decreases from 110degC to 70degC for all species by a 982

factor ranging from two to seven depending on the specific chemical group Very long T2 are 983

observed at 110degC for all acyl species while nearly unchanged values are measured below 984

90degC without any clear trend 985

- The sugar protons (H1rsquo H5rsquo H6rsquo) have the shortest set of T2 values (lt 1 ms) below 100degC 986

The long component of T2 generally ranges between 4 ms and 12 ms 987

0 1 2 3 4 5 6 7 8

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1 2 3 4 5 6 7 8

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

H1

H8H11

H9H10

CH2

H1rsquo

H6rsquo

CH2

H910H811 H1CH2

H1rsquo H5rsquo6rsquo

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

OH

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45 T= 110degC

T= 90degC

T= 80degC

T= 70degC

Log

In

ten

sity

Time ms

Fit

)

CH2 H910 H811 H1 H56 H1 OH

00

05

10

15

20

25

30

35

40

T2 m

s

T= 110degC

T= 90degC

T= 80degC

T= 70degC

CH2 H910 H811 H1

0

2

4

6

8

10

12

T2 m

s

T= 110degC

T= 90degC

T= 80degC

T= 70degC

Short T2(1H) Long T2(1H)

a) b)

c)

11

Table S 1 ndash 1H chemical shifts of G-C181-OH in solution and in the solid-state Solid-state spectra are 988

recorded at a MAS frequency of 10 kHz in DMSO-d6 broad peak 989

990

δH ppm

Group Ndeg Liquid

(MeOD-d4) Solid

(T= 110degC) Solid

(T= 70degC)

CH2 H1 354 357

CH2 H2 153 156

CH2 H3-712-14 132 135

CH2 H811 203 203

CH H910 534 534

CH2 H15 162 164

CH2 H16 388 383

CH H1 424 432

CH H2 316 335

CH H3 334 335

CH H4 329 335

CH H5 354 351 357

CH2 H6 366 383

OH OH1 432

45 463

OH OH2 492

OH OH3 490

OH OH4 487

OH OH6 445

991

992

H1H9H10H1rsquo

H6rsquo

OH2rsquo

OH1

12

Detailed discussion about the NOESY and DQ-SQ 2D 1H-1H ssNMR experiments presented in 993

Figure 9 994

995

2D 1H-1H ssNMR experiments are a unique tool to explore through-space proximities 996

between 1H nuclei ssNMR offers a large number of possibilities each adapted to a given 997

system The advantage of 2D maps with respect to a dynamics approach is their direct 998

readability in clear two nuclei which exchange their magnetization through-space (whichever 999

mechanism may be responsible for this exchange) will display a specific cross-peak However 1000

the way cross-peaks are presented and interpreted strictly depends on the nature of the pulse 1001

program which in turn is chosen according to the physical nature of the sample 1002

1D temperature-dependent 1H NMR experiments performed on G-C181-OH (Figure 1003

4a in the main text) indicate sharp peaks at 90degC and broad peaks at 70degC Narrow peaks 1004

indicate fast andor large-amplitude molecular tumbling averaging the strong 1H-1H dipolar 1005

interactions This excludes the use of pulse sequences like DQ-SQ which is not efficient under 1006

these conditions For this reason we employ spin-diffusion-based pulse sequences like 1007

NOESY which is more adapted for samples with high molecular mobility When mixing times 1008

are chosen to be short (lt 1 s) one generally explores through-space proximities within about 5 1009

10 Aring On the contrary when molecular motions are restricted dipolar coupling becomes 1010

significant 1H NMR peaks are broader and pulse sequences like DQ-SQ can now be used 1011

with a spatial resolution of less than 5 Aring 1012

Whenever possible DQ-SQ experiments are preferred over spin diffusion-based pulse 1013

sequences because they are more selective in particular for likewise interactions which are 1014

not discriminated in NOESY-type experiments This is illustrated by the typical interaction 1015

schemes for NOESY (top Figure 9a) and DQ-SQ (top Figure 9b) For two given A and B 1016

protons 2D NOESY maps will display two on-diagonal and two square-shaped symmetrical 1017

off-diagonal cross-peaks Off-diagonal cross-peaks illustrate the mutual A-B and B-A 1018

interaction but the on-diagonal peaks are systematic and do not prove any A-A or B-B 1019

interaction Similarly to NOESY 2D DQ-SQ maps display off-diagonal peaks (symmetric 1020

across the diagonal top Figure 9b) upon A-B and B-A interaction but in contrast they will 1021

display on-diagonal peaks only in the presence of self-interaction In the tutorial example of 1022

Figure 9b top the on-diagonal A-A peak indicates mutual A-A interaction while lack of on-1023

diagonal B-B peak evidences the absence of B-B proximities Combining NOESY and DQ-SQ 1024

experiments is capital to study the local arrangement of G-C181-OH and its evolution with 1025

temperature 1026

13

At T= 90degC the H1 and H5rsquo contributions are very close but possible to discriminate 1027

them (δ(H1)= 354 ppm δ(H5rsquo)= 351 ppm) according to 1H NMR chemical shifts in Table S 1028

1 2D 1H-1H NOESY in Figure 9a also discriminates between the two components attributed 1029

on the basis of the interactions with their mutual neighbors (green H1-H2 and H5rsquo-H1rsquo squares 1030

in Figure 9a) On this basis 2D NOESY does not show a direct H1rsquo-H1 correlation (blue red-1031

crossed) but rather an extensive network of intra-sugar interactions These features exclude 1032

suggest intra- and inter-sugar interactions and they exclude head-tail arrangements as a major 1033

molecular configuration at 90degC 1034

At 70degC the 1H NMR peaks are broader indicating a reduction of the T2 (1H) relaxation 1035

times andor residual homonuclear dipolar interactions excluding the use of NOESY In this 1036

case 1H-1H DQ-SQ correlation experiments are best suited Figure 9b presents the 1H-1H DQ-1037

SQ experiment on G-C181-OH at 70degC Peaks are broader than in NOESY and attribution may 1038

be cumbersome However the experiment is performed at 700 MHz and MAS= 22 kHz which 1039

improves spectral resolution with respect to the same experiment performed at 300 MHz and 1040

MAS= 10 kHz The auto-peak at δH= 356 ppm in the SQ dimension is attributed to H1-H1 spin 1041

pairs mainly resulting from intra CH2 interaction while the DQ cross-peaks at δH= 156-356 1042

ppm and δH= 432-356 ppm respectively origin from H2-H1 and H1rsquo-H1 interactions (Figure 1043

9b) DQ-SQ map does not show any auto-peak corresponding to H1rsquo-H1rsquo or Hxrsquo-Hxrsquo (x= 2rsquo 1044

3rsquo 4rsquo) interactions (red-crossed blue Figure 9b) thus excluding any proximity (lt 5 Aring) between 1045

close glucose neighbors On the contrary DQ-SQ maps (recorded both at 700 MHz Figure 9b 1046

or at 300 MHz not shown) systematically show DQ cross-peaks which can only be attributed 1047

to H1rsquo-H1 interactions (δH= 432-356) The presence of these H1rsquo-H1 interactions can only be 1048

explained by a head-tail arrangement of the G-C181-OH molecules as illustrated below the 1049

DQ-SQ map in Figure 9b 1050

11

arrows Ndeg5 and Ndeg6 The corresponding profiles A through D confirm the tubular nature of 275

the elongated objects Profiles B and D are both characterized by typical symmetrical thick 276

walls (low intensity depths highlighted by shaded rectangles) surrounding a hollow region (high 277

intensity peak between the shaded rectangles) The wall thickness varies between 35 and 40 278

nm for profile B and between 41 and 61 nm for profile D while the total nanotube diameter 279

varies between 14 nm at B and 20 nm at D At the same time the thickness of the vesicle wall 280

is about 6 nm at A and 135 nm at C Considerations about the bilayer or interdigitated structure 281

of the membrane will be presented below 282

The connection between nanotubes and vesicles seems to be multiple For instance 283

arrow Ndeg1 (Figure S 3) and Ndeg4 (Figure 3) show a single-wall vesicle of which the membrane 284

is extruded into a nanotube Similarly a closer look at arrows Ndeg1 2 and 3 in Figure 3 show 285

that the vesicle is single-wall and fully closed while the nanotube nucleates at the vesicle outer 286

surface with no apparent topological continuity In many regions of the sample at T= 90degC 287

one can find exvaginations or buds232531 in the nanotube wall This is shown for instance by 288

arrows Ndeg7 and particularly put in evidence in Figure 4 by arrows Ndeg1 through Ndeg4 where the 289

amplitude of the oscillation varies between 20-25 nm (arrow Ndeg1) and 6-8 nm (arrows Ndeg3-4) 290

291

Figure 4 - Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heated at T= 90degC 292

293

Still between Tm2 and Tm1 but at lower temperature (T= 70degC) the nanotubes seem to 294

vary from single threads (arrow Ndeg1 Figure S 4) to nanotubes of diameter below 15 nm both 295

connecting nanotubes of diameter ranging between 50 nm and 100 nm (arrows Ndeg2 Figure S 296

4) The larger tubes are most likely formed by the inflation and flattening of the nanotubes 297

probably after assimilation of surrounding vesicles and fusion of the exvagination discussed 298

above The transition between vesicles tubes and flat membranes is certainly more accentuated 299

Arrow Ndeg3 in Figure S 4 shows distinct vesicles while arrow Ndeg4 identifies a flat structure 300

evidently formed from fused vesicles In particular arrow Ndeg5 points at the fusion of two 301

2

1 132

4

12

vesicles while arrow Ndeg6 points at two vesicles fusing into the flat membrane Additional areas 302

where vesicle fusion is prominent are indicated by the oscillations at the outer skirt of broad 303

lamellae and pointed by arrow Ndeg7 304

Finally below Tm1 at T= 25degC for instance mainly flat crystals with no visible 305

exvaginations or tubes are visible (Figure S 5) by cryo-TEM and the material turns into an 306

insoluble precipitate Massive presence of flat crystalline structures is also clear from the bulk 307

iridescent behavior of a G-C181-OH solution under manual shear 308

In summary despite the uncertainty related to the actual state of equilibrium of the 309

glucolipids at the moment of cryofixation cryo-TEM depicts a vesicle phase at Tgt 100degC a 310

flat lamellar phase below Tm1 and an interconnected vesicle-nanotube network between Tm1 and 311

about 100degC The combination of statistically-relevant fluorescence microscopy and 312

scatteringdiffraction experiments provides an additional proof of the coexistence of 313

vesiclenanotube systems between the two transition temperatures Please note that the 314

experiments reported below have been performed under controlled conditions of temperature 315

which is then indicated by the use of T instead of T 316

13

317

Figure 5 ndash a-c) Confocal fluorescence microscopy images recorded at 90degC and 70degC for a 5 mgmL G-318

C181-OH aqueous solution d-g) Typical fluorescence microscopy images recorded at T= 80degC using 319

Laurdan (refer to materials and methods section for more details on this fluorophore) Excitation 320

wavelength is 405 nm for all the images while emission wavelength is 490 nm in (df) and 440 nm in (eg) 321

h) Full emission spectra obtained from the integration of various fluorescence images recorded from 412 322

nm to 600 nm (step 5 nm) at 90degC 80degC and 70degC i) Box plot of GP defined as 119918119927 =(120640120786120786120782minus120640120786120791120782)

120640120786120786120782+120640120786120791120782 with 120640120786120786120782 323

and 120640120786120791120782 being the emission wavelengths at 440 nm and 490 nm at 90degC 80degC and 70degC Each data point 324

l= 490 nm l= 440 nm

l= 490 nm l= 440 nm

10 μm400 450 500 550 600

00

02

04

06

08

10

No

rma

lize

d I

nte

nsity

l nm

90degC

80degC

80degC

70degC

70degC

400 450 500 550 60000

02

04

06

08

10

No

rma

lize

d I

nte

nsity

l nm

90degC

80degC

80degC

70degC

70degC

d) e)

f) g)

h)

i)

T= 80degC

Laurdan

T= 80degC

T= 80degC T= 80degC

T= 90degC T= 70degC T= 70degC

a) b) c)

1

1

1

14

corresponds to a set of regions of interest (ROIs) (54 at 90degC 65 at 80degC and 75 at 70degC) selected in the 325

corresponding fluorescence images ROIs were representatively selected so to include the entire image but 326

also vesicles alone tubes alone and mixtures of vesicles and tubes 327

328

Figure 5ab shows two typical confocal fluorescence microscopy images recorded on a 329

G-C181-OH solution (5 mgmL) in a flame-sealed capillary heat at T= 90degC and T= 70degC 330

Although the resolution of optical microscopy is not comparable with cryo-TEM the sample is 331

broadly composed of vesicles coexisting with nanotubes whereas in some cases the nanotube 332

starts from the vesicle itself (arrow Ndeg1) in good agreement with cryo-TEM data Furthermore 333

SANS recorded at T= 90degC (Figure S 6a) shows a scattering profile with a pronounced -2 334

dependence of the scattering intensity against the wavevector q in the log-log scale This 335

indicates the massive presence of a flat interface and it is expected typical of bilayers in 336

vesicular morphologies56 However the typical signature of nanotubes cannot be observed in 337

the SANS profile and this can be explained by both an argument related to the relative lower 338

amount of nanotubes with respect to vesicles at 90degC but also to the steeper slope of flat 339

morphologies (-2) compared to tubes (-1) the signal of the former masking the tube signal at 340

low wavevector values The SANS profile is characterized by a diffraction peak at q= 0167 Aring-341

1 of which the assignment is refined by WAXS data 342

Temperature-driven WAXS experiments in Figure S 6bc (5 mgmL-1 and 25 mgmL-1 343

respectively) show a lamellar order possibly attributed to a multilamellar structure of the 344

vesicles (dv Figure S 6d) at high temperature and to the interplanar arrangement of the 345

crystalline flat structures (dL Figure S 6d) at T= 25degC The sharp (001) reflection indicates 346

typical interlamellar distance of 353 Aring at 25degC although this value varies with temperature 347

(400 Aring at 111degC 372 Aring at 88degC) thus confirming its assignment to a hydration-dependent 348

interlamellar order More intriguing is the second broader peak at q= 0403 Aring-1 (d= 156 Aring) 349

and its reflection both clearly visible at 88degC below Tm2 We attribute it to the inter-glucolipid 350

arrangement inside the tubular phase (dt Figure S 6d) coexisting with the vesicle phase 351

between Tm1 and Tm2 and to the planar lipid arrangement (di Figure S 6d) in the rigid flat 352

lamellar structure below Tm2 Its temperature- hence hydration- independent position also 353

confirms its structural origin Finally besides the water-structure peak at about 2 Aring-1 we also 354

observe a broad peak at 144 Aring-1 characteristics of the lateral packing (d Figure S 6d) of lipids 355

in bilayers While its position could be consistent with an Lα phase5758 above Tm2 its broadness 356

and positional invariance as well as the coexistence between vesicles and tubes do not allow a 357

precise assignment 358

15

Does G-C181-OH assume a bilayer or interdigitated structure the latter being the one 359

found for other similar bolaform glucolipids4243 The size of G-C181-OH can be estimated to 360

be at the most 32 Aring given by the length L of the C16 tail (L = 22 Aring from the Tanford formula 361

L= 154 + 1265 times n (n= 16)59) and the sugar (about 10 Aring)4243 However the cis conformation 362

of the C=C bond imposes a ldquokinkrdquo of 60deg in the aliphatic chain which reduces the effective 363

length of the oleic backbone to about 17 Aring for an effective total G-C181-OH length of about 364

27 Aring Cryo-TEM provides a tube thickness varying from 35 Aring to 60 Aring and a vesicle membrane 365

thickness of at least 60 Aring WAXS suggests an interlamellar thickness (membrane and water) 366

between 40 Aring and 37 Aring and an inter-lipid distance in the tube of about 16 Aring In the absence of 367

a clear form factor oscillation profile as found for similar glucolipids4243 one must 368

acknowledge that these data pledge neither in favor of fully interdigitated nor a double layer 369

structure and might suggest a tilted lipid arrangement with partial interdigitation60 370

371

13C ssNMR demonstrates that G-C181-OH can assume two coexisting head-head and 372

head-tail configurations 373

Vesicles are characterized by low positive mean and gaussian curvatures while 374

nanotubes are described by a combination of a high positive mean and zero gaussian curvatures 375

In this regard the coexistence of vesicles and nanotubes requires glucolipids to adapt 376

themselves to both environments at the same time For this reason the formation of nanotube-377

vesicle networks generally requires a specific driving force able to impose a high mean and 378

zero gaussian curvature to the membrane In many systems insertion of at least a second 379

constituent (lipid surfactant protein) in the membrane becomes a necessary internal driving 380

force for the nucleation and growth of so-called membrane inclusions from which 381

topologically-connected nanotubes can form10ndash123261 as explained by orientational ordering 382

arguments62 It has also been shown that external driving forces such as use of laser 383

ldquotweezersrdquo30 osmotic pressure6 electroformation5ndash82829 or controlled suction13ndash16 are 384

otherwise necessary as also supported by theoretical studies19 385

The bending energy of tubes is higher than that of a vesicle membrane8 due to the 386

dependence of the membrane free energy from the principal curvatures For this reason 387

isotropic (rotational symmetry along the main axis) but also anisotropic33 lipids generally 388

stabilize low-curvature shapes like flat membranes or vesicles Lipid nanotubes are also a 389

common phase formed by amphiphiles but generally for lipids with a certain complexity in 390

their molecular structures as exampled by diacetylenic derivatives of phospholipids among 391

others262763ndash65 Negative contributions to the bending energy favoring the simultaneous 392

16

presence of interconnected vesicles and tubes was shown to be driven by the presence of 393

anisotropic lipids in a membrane continuum Spontaneous deviations to the elastic theory in the 394

presence of surfactants lipids and proteins with anisotropic structures favor the formation of 395

protrusions with radius of the order of the fraction of a micron9ndash1219ndash2362 396

In the absence of both internal and external driving forces mentioned above G-C181-397

OH should behave as a standard lipid Its DSC profile (Figure 2)536667 under both hydrated and 398

dry conditions identifies a hydration-independent transition (Tm1) slightly above room 399

temperature this is followed by one main transition at higher temperature (Tm2) in the dry 400

sample50 For such a profile one expects a gel phase (Lβ) below Tm1 and a liquid crystalline 401

structure possibly in the morphology of vesicles (L4 phase) possibly above Tm1 and most likely 402

above Tm231 as experimentally found On the basis of the thermogram recorded on the dry 403

powder one may also expect a ldquoripplerdquo (Pβ) phase between Tm1 and Tm2 characterized by 404

periodic undulations of the bilayer (generally addressed to as regions with both liquid order and 405

disorder)5455 406

In the present work instead of a Lβ phase we observe flat crystals below Tm1 and a 407

vesicle-nanotube network instead of the ldquoripplerdquo phase The latter is particularly unexpected 408

because the membrane is composed of a single-component (G-C181-OH) and no anisotropic 409

external force (osmotic hydrodynamic electric) is used to pull and stabilize the tubes In 410

addition the nanotubes are observed at fairly high temperatures above Tm1 (between about 411

60degC and 90degC) while this morphology is generally favored at low temperature (proportionality 412

with 1kT figure 5 in Ref 62) commonly below a given transition temperature characteristics 413

for each specific amphiphile27 Furthermore nanotubes of G-C181-OH seem to be stable in 414

time thus differing from what is found in electroformed vesicles78 They are also observed 415

away from phase transition events (may them be at Tm1 or Tm2) as otherwise reported before 416

for a diacetylenic derivative of phosphatidylcholine26 and many other systems27 417

A stable nanotube-vesicle network for a single-component phospholipid in the absence 418

of external driving forces is not expected and has never been reported before It could be 419

explained by the coexistence of two micro-separated phases of different rigidity This 420

hypothesis is excluded by confocal microscopy on Laurdan-labelled membranes and magic 421

angle spinning (MAS) solid-state NMR (ssNMR) experiments 422

Laurdan is a well-known fluorophore used to put in evidence the coexistence of rigid 423

and fluid domains in both model and biological membranes47ndash49 Laurdan was used here to label 424

the G-C181-OH membranes between Tm1 and Tm2 and to verify the possible micro-segregation 425

and coexistence of rigid (eg Lβ λem= 440 nm) and fluid (eg Lα λem= 490 nm) domains in 426

17

tubes and vesicles respectively A set of quantitative experiments performed on a 5 mgmL-1 427

solution summarized in Figure 5 shows an excess of fluorescence at only λem= 490 nm for 428

temperature between 90degC and 70degC (Figure 5h) The GP value which quantifies the rigidfluid 429

character of the membrane is systematically negative (Figure 5i) indicating an overall fluid 430

membrane at all temperatures GP is quantified on more than 50 regions of interest (ROIs) at 431

each temperature including vesicles and tubes alone as well as connected nanotube-vesicles 432

(Figure 5d-g) The dispersion of GP at 80degC and 70degC between -08 and -02 could suggest the 433

spurious presence of membranes with various degrees of flexibility although fluidity is always 434

predominant The combination of WAXS (broad peak d= 144 Aring-1 invariant with 435

temperature Figure S 6bc) and confocal fluorescence microscopy using Laurdan confirms the 436

single-phase of the membrane composed of G-C181-OH both in vesicles and nanotubes This 437

is also consistent with the lack of phase transitions in DSC experiments between ~85degC and 438

~60degC especially in hydrated samples We conclude that the vesicle-nanotube transition is not 439

driven by the local segregation of gel phase (Lβ) microdomains nor that the nanotubes are 440

composed of such phase 441

ssNMR is the best-suited approach to investigate the origin of the local driving force for 442

the nucleation of anisotropic inclusions of which the average orientation depends on the local 443

membrane and inclusion curvatures and was shown to promote tubing62 ssNMR can provide 444

information on the local molecular mobility eg through the nuclear spin relaxation behavior 445

but also on intermolecular interactions and stereochemistry eg through variations in the 446

chemical shift68ndash70 447

Ideally ssNMR should be performed on diluted G-C181-OH samples above Tm1 448

However fast MAS necessary for high resolution centrifuges the sample As a first 449

approximation hydrated G-C181-OH samples for instance at 50 wt as classically done for 450

phospholipids71 could be studied Unfortunately permeability of the sample holder combined 451

to heating (close to the boiling point of water) and MAS provoke dehydration and even sample 452

ejection After experiencing both of the above this approach was abandoned As a second 453

approximation the study of the dehydrated G-C181-OH powder avoids centrifugation and 454

evaporation If studying dried lipids may not reflect their exact behaviour under diluted 455

conditions valuable structural information can still be obtained in dehydrated molecular 456

systems like proteins thus illustrating that this approach is not unreasonable Furthermore if 457

DSC (Figure 2a) shows that hydration suppresses the Tm2 it also shows that the value of Tm1 is 458

not affected by water at concentrations as low as 10 wt For this reason performing a 459

structural analysis of dehydrated G-C181-OH by ssNMR above Tm1 may help understanding 460

18

the local arrangement of G-C181-OH in two morphologies vesicles and nanotubes having 461

sensibly different curvatures 462

To explore the local mobility of the most relevant chemical groups of G-C181-OH 463

above Tm2 (T= 110degC) and between Tm2 and Tm1 (70 lt T degC lt 90) we combine a series of 1H 464

and 13C experiments Single-pulse (SP) MAS experiments (1H and 13C) when recorded under 465

quantitative conditions allow the detection of all chemical species while cross-polarization 466

(CP) MAS (13C only) experiments are sensitive only to constrained groups for which the 1H-467

13C dipolar coupling is not averaged to zero by the molecular motions 468

469

Figure 6 ndash a) 1H MAS (10 kHz) ssNMR spectra recorded from 110degC to 70degC b) 13C MAS (10 kHz) ssNMR 470

spectra recorded using high power decoupling single pulse (SP) and cross-polarization (CP) pulse sequences 471

Contact time for CP experiments and corresponding temperatures are given on each spectrum Full spectra 472

are given in Figure S 7 473

The most relevant 1H and 13C spectra are given in Figure 6 (full range in Figure 7 and 474

Figure S 7) As expected the 1H solid-state SPMAS spectra show broader peaks compared to 475

the solution-state NMR spectrum (Figure 6a) reflecting the obvious solid-state character of the 476

sample at any temperature The sugar group and the C=C bond resonate in the 25-6 ppm range 477

and 535 ppm respectively while the OH groups of the sugar moiety are found below 45 ppm 478

(Figure 6a full assignment in Table S 1) Very broad (~1 ppm) at 100degC the position of the 479

OH peak varies with the temperature value and becomes narrower (lt~05 ppm) and visible at 480

δH= 464 ppm at T= 70degC Line narrowing and chemical shift variations of the OH resonance 481

in ssNMR is always a direct sign of H-bonding interactions whereas the strength of the H-482

bonding hence its OHhellipH distance is correlated to a dispersion in chemical shifts72ndash75 All other 483

60 55 50 45 40 35 30 25

ppm

110

90

80

70

910

1rsquo

6rsquo16

1

OH

134 132 130 128 126

ppm110 108 106 104 102 100

ppm

SP (T= 110degC)

SP (T= 70degC)

CP 50 ms (T= 70degC)

CP 3 ms (T= 70degC)

910 1rsquo

1H 13C

1

811

910

CH2 CH2

1rsquo

6rsquo

a) b)

C102

13024

C9101

13003

C92

12869 C1rsquo210382

C1rsquo110443

C1rsquo13035

2rsquo3rsquo4rsquo

T degC

5rsquo

19

resonances broaden from 110degC to 70degC thus indicating slower andor more anisotropic 484

reorientational motions of all chemical groups The specific dynamic behaviour of the sugar 485

headgroup is also reflected by its 1H transverse relaxation time T2 whereas the shorter the T2 486

the higher the rigidity Experiments performed between 110degC and 70degC shown and 487

commented in Figure S 8 in the Supporting Information globally show a single short (lt 1 ms 488

below 100degC) T2 relaxation component for H1rsquo H5rsquo and H6rsquo and two both short and long 489

(between 4 ms and 12 ms) T2 for the acyl chain (CH2 H910 H811) even below 100degC 490

At 110degC 13C SPMAS experiments (Figure 6b full range in Figure 7 and Figure S 7 491

assignments in Table 1) show a single peak at δC= 10335 ppm for C1rsquo and at δC= 13003 ppm 492

for C910 as expected for a symmetrical conformation with two chemically-equivalent 493

carbons76 At 70degC the C1rsquo peak disappears while C910 now shows two peaks around 128 494

ppm and 130 ppm (Table 1 full temperature series in Figure S 7b) attributed to C92 (δC= 12869 495

ppm) and C10 in a non-symmetric configuration respectively for which these two carbons are 496

chemically inequivalent76 More precisely the resonance at 130 ppm is actually the 497

superposition of two close resonances C9101 (δC= 13003 ppm) and C102 (δC= 13024 ppm) 498

where subscripts 1 and 2 stand for two different conformations of the cis-olefin group as 499

reported for various phases of oleic acid in the vicinity of 0degC76 500

501

Table 1 ndash 13C Chemical shifts of G-C181-OH in solution and solid-state Solid-state experiments are 502

recorded at MAS= 10 kHz All spectra are reported in Figure 7 and Figure S 7 The contact time for cross 503

polarization is 3 ms In all ssNMR experiments temperature is first raised at 110degC and then lowered to 504

either 90degC or 70degC Attribution of the 13C resonances in Table 1 is performed on the basis of the solution 505

spectrum (Figure S 1b) Additional peaks observed in both single pulse and cross polarization ssNMR 506

experiments are attributed when possible to a configuration effect of the given atom eg C9101 and C92 507

indicate two configurations of C9 All new peaks observed in solid-state and not detected in solution are 508

highlighted in red for convenience 509

510

511

δ ppm

Peak Liquid (DMSO-d6)

Solid Single Pulse

T= 110degC

Solid Single Pulse

T= 90degC

Solid Single Pulse

T= 70degC

Solid Cross Polarization

T= 70degC

C102

- - 13024 13024

C910C9101 13009 13003 13003 13003 13003

C1C9C10C1rsquo

C6rsquo

20

C92 - - - 12869 12869

C1rsquoC11 10329 10335 10335 - 10443

C12 - - - - 10382

C3 7724 7709 7702 - 7829

- 7634 7629 - 7621

7525

C2 7388 7398 7390 7370 7370

C4 7052 7077 7054 7041 7098

C16 6900 7016 - - 7029

- - - - 6793

C1C5C6 6153 6241 6236 6239 6262

C1C5C6 6117 - - - 5956

C2 3300 3290 3290 3283 3279

CH2(trans) - - - 3198 3196

CH2(gauche)C15 2941 2981 2987 3004 -

C811 2706 2747 2747 2832 2817

CH2 2597 2625 2628 2630 2817

512

13C CPMAS spectra at 70degC (Figure 6b full range in Figure 7 and Figure S 7d) at 513

increasing contact times confirm the presence of two cis-olefin environments one for which 514

C910 are equivalent labelled here C9101 and one for which they are inequivalent labelled 515

C92 C102 Both environments are rigid enough to be observed at short contact time (tc= 50 μs) 516

A similar double environment is also observed for the sugar C1rsquo CPMAS at 70degC shows two 517

peaks (C1rsquo1 at δC= 10443 ppm C1rsquo2 at δC= 10382 ppm Figure 6b Figure S 7d) These are 518

significantly shifted downfield from the C1rsquo peak observed at 110degC and are detected at contact 519

times as short as 50 μs thus indicating their rigid character One also finds that CPMAS is 520

totally inefficient at 100degC for short contact times (50 μs) and poorly efficient even at 3 ms 521

typical of fast-tumbling molecules (Figure S 7c) also both gauche (δC= 3004 ppm) and all-522

trans (δC= 3198 ppm) conformations of the CH2 carbons coexist at 70degC (Figure S 7b)77 523

According to ssNMR one main mobile molecular environment exists at 110degC as 524

shown by the well-resolved 1H spectrum single C=C (C9101) and anomeric (C1rsquo) resonances 525

gauche conformation of the aliphatic CH2 and single T2 of the aliphatic CH2 chain and sugar 526

protons The sugar group is certainly mobile (inefficient 1H13C CPMAS) but still more 527

constrainted than the acyl chain (shorter T2 (1H)) As the temperature is lowered the overall 528

rigidity increases (broadening of 1H NMR peaks faster 1H transverse relaxation and efficient 529

1H13C magnetization transfer during CPMAS) presumably driven by H-bonding 530

interactions (downfield shift and line narrowing of the OH 1H signal very short OH T2 loss of 531

the C1rsquo signal in SPMAS) At the same time a number of spectroscopic data (two concomitant 532

21

resonances for anomeric C1rsquo1 and C1rsquo2 cis-olefin C9101 and C92 C102 and CH2 gauche and 533

all-trans environments) and relaxation behaviour (fast and slow T2 relaxation components) 534

strongly suggests the coexistence of two molecular environments Further proofs of the 535

coexistence of two distinct molecular domains are given by a number of chemical shift 536

variations in the 13C CPMAS spectrum at 70degC compared to the one at 110degC 90degC and 537

solution This is shown in Figure 7 and in the list of 13C chemical shifts in Table 1 538

539

540

Figure 7 - Comparison between solution- and solid-state 13C NMR spectra of G-C181-OH Solid-state 541

experiments are recorded directly on the sample powder at a MAS of 10 kHz (4 mm zirconia rotor) SP 542

Single Pulse CP Cross Polarization In all the ssNMR experiments temperature is first raised at 110degC 543

and then lowered to either 90degC or 70degC Segmented lines are meant to highlight the known peak position 544

in solution thus stressing the strong chemical shift variations in the solid-state 545

546

NMR experiments above Tm2 suggest that vesicles could contain G-C181-OH with a 547

single family of mobile conformations characterized by chemically-equivalent C=C (red 548

Figure 8b) and a single C1rsquo conformation (green Figure 8b) Experiments between Tm1 and Tm2 549

otherwise suggest that two coexisting conformations characterized by chemically-equivalent 550

(red Figure 8b) and inequivalent (magenta Figure 8b) C=C could be associated to nanotubes 551

Two distinct C1rsquo sites (grey Figure 8b) also indicate two configurations of the sugar headgroup 552

A comment on this point will be given in the following paragraphs 553

The local (Aring-scale) variation of molecular mobility below Tm2 drives the coexistence of 554

two distinct molecular environments of the same G-C181-OH molecule This occurs within a 555

single-phase domain (as deduced by WAXS) of equivalent membrane rigidity (as derived from 556

Laurdan probe in confocal microscopy) Although never reported before the difference in the 557

local mobility of the same amphiphile (here G-C181-OH) could satisfy the theoretically-558

140 120 100 80 60 40 20

CP(3 ms)-70degC

SP-70degC

SP-90degC

SP-110degC

ppm

LiquidC910

C1rsquo

C2rsquo

C4rsquo

C3rsquo

C1

C6rsquoC5rsquo

C1

6

C8

11

C2 C

H2

C1

5

CH

2

DM

SO

13C

22

required conditions that induce local lipid segregation and formation of membrane inclusions 559

(Figure 8c) In case of their non-zero average orientation the local membrane curvature 560

increases62 and nanotubes can spontaneously form (Figure 8bc) More intriguing this 561

mechanism occurs at least 10degC above Tm1 although it should not be favored above the Tm27 562

From both experimental and theoretical backgrounds the conditions of topologically-563

connected vesicles-nanotubes systems are met only when either internal (two components) or 564

external driving forces are applied19 In the present single-lipid system of which the phase 565

behavior is summarized in Figure 8a the most plausible driving force could be explained by 566

the NMR results collected on the dry sample G-C181-OH concomitantly occupies two 567

molecular environments characterized by 1) two distinct conformations of the sugar 568

displaying equivalent mobility and 2) two conformations of the acyl chain unsaturation 569

570

23

571

Figure 8 ndash a) Thermotropic phase behavior of G-C181-OH at 5 mgmL-1 Note that Tm2 is observed on the 572

dry powder only b) Possible relationship between the molecular environment as derived from NMR and 573

membrane curvatures All the molecular drawings refer to the same G-C181-OH molecule (bulky ellipsoid 574

equiv sugar headgroup) while each color corresponds to a given configuration Grey of C1rsquo1 and C1rsquo2 indicates 575

the uncertainty about the assignment of the specific NMR peak The curvature C is defined as 1R R being 576

the curvature radius R1 R2 radius of curvature CG Gauss curvature CM Mean curvature c) Free 577

interpretation of the anisotropic high-curvature (CMgt 0 CG= 0) inclusions in the membrane (low CM and 578

CG) continuum driven by inter-sugar H-bonding and reduced intramembrane diffusion for a G-C181-OH 579

glucolipid 580

581

Flat crystalline structure

Nanotube-vesicle network

Vesicles

48degC

120deg

106 105 104 103 102 101

ppm

H9 H10H1rsquo

106 105 104 103 102 101

ppm106 105 104 103 102 101

ppm106 105 104 103 102 101

ppm

C=CC=C

C1rsquo1C1rsquo2

C1rsquo

Tm2(96degC)Tm1

a)

b)

c)

Same molecule different local orientations and sugar conformations

High curvature membrane

Low curvature inclusion

Flip-FlopStrong H-bonding

CG

CM ( + )

CG 0

CM 0 CG 0

CM 0

G-C181-OH

24

Two related questions are unanswered at the moment What actually drives nanotube 582

formation and is the organization of G-C181-OH within the vesicles and tubes the same Lipid 583

nanotubes have been studied for at least three decades over which broad experimental and 584

theoretical insight has been generated Despite this fact the actual understanding is still matter 585

of debate27 Micelle-to-nanotube and vesicle-to-nanotube phase transitions are generally 586

reported for single amphiphiles below a given Tm and intermediate phase composed of twisted 587

and flat fibers is concomitantly reported Initially thought to be a chirality-driven phenomenon 588

recent studies propose a more general mechanism involving crystallization and symmetry 589

breaking in the lipid packing driven by optimal packing27 If twisted ribbons were observed 590

before for a number of microbial glycolipid bolaamphiphiles44 they are not found for G-C181-591

OH This fact as well as the large body of cryo-TEM images and the observation of nanotubes 592

above Tm1 rather pledge in favour of a direct topological change from positive (vesicles) to 593

zero (nanotubes) Gaussian curvatures In this case vesicle-to-nanotube transitions could be 594

driven by external forces56307813ndash162829 595

Considering the absence of external forces in the present system there are only two 596

possibilities left spontaneous budding of the membrane or nanotube pulling after inter-vesicle 597

collisions due to Brownian motion Both are possible as inclusions composed of a different 598

arrangement of the same molecule in the membrane continuum can theoretically explain 599

spontaneous nanotube formation in an isotropic medium672223629ndash1218ndash21 However the second 600

one is probably more likely to prevail because it could explain the large number of nanotubes 601

connecting two distant vesicles Interestingly similar events occur for any phospholipid-based 602

vesicular solution but nanotubes are extremely rare unless at least two lipids of different 603

molecular structure are mixed together The spectroscopic 1H and 13C and relaxation (1H) 604

ssNMR data even if collected on a dry powder demonstrate that the coexistence of two G-605

C181-OH configurations involving both the sugar headgroup and fatty alcohol chain are 606

possible for this molecule As a matter of fact this is not so surprising as more than one energy 607

minima each corresponding to a given conformation are known for glucose its non-reducing 608

derivative methylglucose and carbohydrates in general78ndash80 609

The presence of two configurations could explain the coexistence between two 610

structures of different curvature but the data presented so far do not explain the origin of these 611

two configurations In other words what drives the change in the local structure of the sugar 612

headgroup and fatty alcohol within the membrane 613

The association of known self-assembled structures to spectroscopic 13C ssNMR data is 614

known for years in particular for gluconamides lipids having a linear sugar headgroup81ndash84 It 615

25

was found that gauche effects in some cases up to the γ position in the sugar chain induce 616

upfield chemical shifts while anti and trans conformations result in downfield shifts Each 617

structure (monolayer crystal bilayer crystal or micellar fiber) was then characterized by a given 618

organization of the gluconamide and a characteristic 13C ssNMR fingerprint which could be 619

explained by gauche γ-gauche anti and trans conformations of the headgroup Similar 620

structural-spectroscopic correlations have been collected over the years for much simpler cyclic 621

sugars85 whereas pyranoses can undergo important conformational changes (chair half-chair 622

boat skew) when the appropriate energy barriers are crossed78ndash8086 A quantitative 623

conformational study based on the 13C chemical shift variations shown in Table 1 would be 624

certainly helpful but very delicate in the context of the present state of the art because a 625

rigorous approach requires at least one known crystal structure8182 which is currently missing 626

for this compound A qualitative approach will then be adopted 627

G-C181-OH is a bolaform amphiphile and flip-flop could be a possible mechanism 628

involved in the nanotube stabilization whereas head-tail interactions have long been considered 629

as important in nanotube-forming bolaform amphiphiles6063 From mere considerations based 630

on T2 (1H) relaxation arguments one could exclude head-tail configurations if the R-CH2OH 631

endgroup (H1) was involved in a direct interaction with the sugar group its corresponding T2 632

would probably be comparable to the sugarrsquos one (OH and H1rsquo) but this does not seem to be 633

the case However this hypothesis would be based on the comparison of the relaxation from 634

CH and CH2 protons additionally involved in different kind of motions Besides the relaxation 635

of H1 is influenced by the strong overlap between H1 and H5rsquo For these reasons relaxation 636

arguments are not solid enough and will not be used 637

We then address this question by exploring intermolecular interactions in a direct 638

manner using 2D 1H-1H homonuclear ssNMR correlation spectroscopy At 90degC in the fast-639

tumbling domain of molecular motion (narrow 1H NMR peaks Figure 6a) and corresponding 640

to the vesicle-rich region spin diffusion (NOESY) experiments are preferred over dipolar 641

coupling based experiments because such interaction is averaged out by molecular motion 642

However when the molecular motion is more constrained in the domain of nanotube stability 643

at 70degC (broad 1H spectra Figure 6a) Double-Quantum Single-Quantum (DQ-SQ) correlation 644

experiments are best suited to probe interaction between close molecular neighbors These 645

experiments are reported in Figure 9 and the related interpretation is only briefly reported 646

below while a more detailed discussion can be found on Page 12 in the Supporting Information 647

26

648

Figure 9 ndash a) 2D 1H-1H NOESY NMR experiment performed at MAS= 10 kHz (4 mm probe) on a Bruker 649

300 MHz Avance III HD spectrometer using 300 ms of mixing time Sample temperature is regulated to 650

90degC reached by cooling from 110degC b) 2D 1H-1H DQ-SQ (BAck-to-BAck BABA pulse sequence used for 651

the DQ excitation scheme) NMR experiment performed at MAS= 22 kHz (25 mm probe) on a Bruker 700 652

MHz Avance III spectrometer using a DQ excitation time equal to 2 rotor periods Sample temperature is 653

regulated to 70degC reached by cooling from 110degC The typical schemes illustrating how to interpret these 654

2D experiments are reported above each experimental map Zooms of the sugar region and configurational 655

interpretation are given below each 2D map More information are given in the text below 656

H16-H16

H1-H1

H9-H10

H1rsquo-H1

H1rsquoH910 H16

H1

H1rsquo-H1rsquo

Hxrsquo-Hxrsquo

2

(pp

m)

(ppm)

315

811910 1rsquo

2rsquo2

3rsquo 4rsquo6rsquo

5rsquo

1rsquo2910

asymp asymp asympHDO MeOD Acetone

1

315

811 9101rsquo

2rsquo2

3rsquo4rsquo6rsquo

5rsquo

1rsquo2 9 10

asympasympasympHDOMeODAcetone

H1rsquo-H1

1rsquo

1

(p

pm

)

(ppm)

H1-H2

H1rsquo-H1

H1rsquo

H910

H1H5rsquo

H2

H1rsquo-H5rsquo

315

811910 1rsquo

2rsquo2

3rsquo 4rsquo6rsquo

5rsquo

1rsquo2910

asymp asymp asympHDO MeOD Acetone

1

H2rsquoH3rsquoH4rsquo

H2rsquoH3rsquoH4rsquo

X= 2rsquo3rsquo4rsquo

NOESY - T= 90degC

1H

1H

1H

1H

DQ-SQ - T= 70degC

a) b)

2

(pp

m)

(p

pm

)

(ppm)

A B

AhellipB

AhellipA

BhellipB

NOESY

(ppm)

A B

AhellipB

DQ-SQ

Single-quantum dimension

Do

ub

le-q

ua

ntu

m d

imen

sio

n

AhellipA

BhellipB

BhellipA

BhellipA

315

811910 1rsquo

2rsquo2

3rsquo 4rsquo6rsquo

5rsquo

1rsquo2910

asymp asymp asympHDO MeOD Acetone

1

lt 5 Aring

lt 5 Aring

H1-H2

H811

H15

H16

H6rsquo

H1rsquo

H1

H5rsquo

H2

H1-H2

H1rsquo-H1

H1rsquo-H5rsquo

H1rsquo

H1 H5rsquo

H1rsquo

H1

H5rsquo

H1H2rsquo H3rsquo H4rsquo

H1-H1

H1rsquo-H1

Hxrsquo-HxrsquoX= 2rsquo3rsquo4rsquo

H1-H2

H1rsquo

Zoom sugar region 28 ndash 45 ppmNOESY DQ-SQ

27

Head-tail or head-head configurations could be probed by the existence of H1rsquo-H1 657

interactions whereas H1 (δH= 356 ppm) is associated to the glucolipid tail and H1rsquo (δH= 432 658

ppm) to the sugar headgroup These atoms are more than 15 Aring apart and a direct interaction 659

could not be observed in both NOESY and DQ-SQ experiments NOESY experiments (Figure 660

9a) show that H1rsquo-H1 correlations are not observed Meanwhile clear-cut direct H1rsquo-H1 661

interactions are indeed observed are detected at T= 70degC (DQ cross-peaks related to H1rsquo-H1 662

correlation in DQ-SQ experiment shown in Figure 9b) Interestingly the absence of auto peaks 663

at δH= 432 ppm and δH= 335 ppm in the SQ dimension excludes H1rsquo-H1rsquo and Hxrsquo-Hxrsquo (x= 2 664

3 4) interactions respectively strengthening the idea that sugar headgroups from two nearest-665

neighbor G-C181-OH molecules are not adjacent in the nanotube-vesicle region 666

Homonuclear 2D ssNMR supports the idea according to which G-C181-OH can 667

undergo molecular flip-flops and a head-tail arrangement in the dry product Extrapolating these 668

conclusions in the lower concentration range lack of H1rsquo-H1 proximity at T= 90degC suggests a 669

head-head configuration in the vesicle-rich region and head-tail configuration within the 670

nanotubes It is however still unclear whether flip-flops are driven by spontaneous temperature-671

induced conformational changes in the glucolipid packing stimulating inclusions within the 672

membrane itself672223629ndash1218ndash21 or by defects generated by inter-vesicle collisions In fact a 673

combination of both is not unlikely Nanotubes are not stable when temperature is 20 to 30 674

degrees above Tm1 indicating that inter-vesicle collisions alone cannot explain the unusual 675

phenomenon In an energy landscape approach87 the head-head configuration is certainly 676

favored at high temperature while both head-head and head-tail configurations could be 677

favored in the mid-temperature region corresponding to the range between Tm1 and Tm2 in the 678

dry powder Both statistical and defect-induced flip-flop events contribute to form high mean 679

zero Gaussian curvature morphologies (nanotubes) for which the head-tail conformation 680

limiting the proximity between bulky sugar headgroups is probably more adapted (Figure 8c) 681

The flip-flop interpretation now reasonably explains the strong differences in the sugar and fatty 682

alcohol conformations deduced by the multiple NMR signals attributed to chemically-683

equivalent groups The existence of various low-energy conformations each having specific 684

NMR signature for sugars in solution78ndash8086 and solid-state81ndash84 further supports our 685

interpretation 686

One last point should be clarified Why between Tm1 and Tm2 are two molecular 687

environments able to coexist instead of a single one of average rigidity as found in the case of 688

ldquoripplerdquo phases50ndash54 We believe that this question can be answered based on the NMR results 689

the relatively short T2 values of the H1rsquo and OH protons shows the constrained mobility of the 690

28

sugar headgroup despite a certain flexibility of the chain (long T2 component) This trend may 691

suggest restrictions in translational mobility probably due to the sugar-OH hydrogen bonding 692

(Figure 8c) meaning that lipids are not able to diffuse within the membrane thus keeping two 693

different configurations stable 694

695

Conclusions 696

Topologically-connected nanotube-vesicle lipidic systems are commonly observed for 697

self-assembled membranes composed of at least two amphiphiles and spontaneously driven by 698

anisotropic inclusions andor by external directional forces This work shows that strong 699

differences in membrane curvature can spontaneously occur in single-lipid systems without 700

external anisotropic forces Coexistence and interconnection between nanotubes and vesicles 701

are evidenced via cryo-TEM and fluorescence confocal microscopy in a single-phase region at 702

temperature above a transition temperature Tm1= 483degC determined by DSC on the hydrated 703

sample but below the second transition temperature Tm2 (957degC) determined by DSC on the 704

dry powder X-ray diffraction (WAXS) also shows the coexistence of two lattice periods one 705

attributed to an inter-membrane repeating distance and the other possibly related to the intra-706

nanotube glucolipid arrangement Labelling the systems with Laurdan a fluorophore of which 707

the emission strongly varies with the local membrane rigidity excludes the presence of rigid 708

micro-separated phases (eg Lβ) 709

1H and 13C ssNMR despite the fact that it can only be performed on the dry powder 710

confirm the presence of one main fluid G-C181-OH environment above Tm2 Below Tm2 the 711

splitting of some 13C peaks and the coexistence of short and long 1H relaxation components 712

demonstrate that G-C181-OH can simultaneously pack into structurally and dynamically 713

inequivalent environments Combining advanced 2D 1H-1H correlation experiments based on 714

spin diffusion or double-quanta excitation suggests that vesicles could contain a preferred head-715

head configuration of the bolaform lipid while nanotubes a preferred head-tail assembly 716

The presence of two environments composed of the same molecule characterized by 717

different molecular configurations and rigidity fulfills the theoretical conditions necessary to 718

spontaneously nucleate a membrane inclusion of non-zero average orientation described as 719

necessary to observe topologically-connected nanotube-vesicle structures in the absence of 720

external forces This hypothesis could also be enriched by the possibility of nanotubes to be 721

pulled away between two vesicles after a collision event 722

This work expands the actual theoretical and experimental framework on lipid networks 723

stabilized by a different radius of curvature It could also motivate the synthesis or isolation 724

29

of new lipids providing a higher level of 3D architectural control in soft networks possibly in 725

the range of room temperature 726

727

Acknowledgements 728

Dr Andrea Lassenberger and Dr Sylvain Preacutevost at Institut Laue Langevin (ILL Grenoble 729

France) are kindly acknowledged for their assistance on the SANS experiment Confocal 730

images were performed at the Institut de Biologie Paris Seine (IBPS) imaging facility Jean-731

Franccedilois Gilles (Imaging Core Facility Institut de Biologie Paris Seine (IBPS) CNRS 732

Sorbonne Universiteacute Paris France) is kindly acknowledged for his assistance on the confocal 733

microscopy imaging Dr Guillaume Laurent (LCMCP Sorbonne Universiteacute Paris France) is 734

kindly acknowledged for helpful discussions The French region Ile-de-France SESAME 735

program is acknowledged for financial support (700 MHz NMR 736

spectrometer) Dr Olivier Diat (ICSM Marcoule France) Prof eronikaIglič and Prof V Aleš737

onsul discussi Slovenia) are kindly acknowledged for helpfUniversity of LjubljanaIglič (-Kralj 738

739

Financial support 740

WAXS experiments have been supported by Soleil Light Source Saint Aubin France proposal 741

Ndeg 20201747 SANS experiments have been supported by ILL proposal Ndeg9-13-778 742

743

References 744

1 D Kabaso N Bobrovska W Goacuteźdź E Gongadze V Kralj-Iglič R Zorec and A 745

Iglič Bioelectrochemistry 2012 87 204ndash210 746

2 C R Safinya U Raviv D J Needleman A Zidovska M C Choi M A Ojeda-747

Lopez K K Ewert Y Li H P Miller J Quispe B Carragher C S Potter M W 748

Kim S C Feinstein and L Wilson Adv Mater 2011 23 2260ndash2270 749

3 P Veranič M Lokar G J Schuumltz J Weghuber S Wieser H Haumlgerstrand V Kralj-750

Iglič and A Iglič Biophys J 2008 95 4416ndash4425 751

4 A Rustom R Saffrich I Markovic P Walther and H-H Gerdes Science (80- ) 752

2004 303 1007ndash1010 753

5 M I Angelova S Soleacuteau P Meacuteleacuteard F Faucon and P Bothorel Prog Colloid 754

Polym Sci 1992 89 127ndash131 755

6 L Mathivet S Cribier and P F Devaux Biophys J 1996 70 1112ndash1121 756

7 V Kralj-Iglič G Gomišček J Majhenc V Arrigler and S Svetina Colloids Surfaces 757

A Physicochem Eng Asp 2001 181 315ndash318 758

30

8 V Kralj-Iglič A Iglič G Gomišček F Sevšek V Arrigler and H Haumlgerstrand J 759

Phys A Math Gen 2002 35 1533ndash1549 760

9 M Drab D Stopar V Kralj-Iglič and A Iglič Cells 2019 8 626 761

10 A Zidovska K K Ewert J Quispe B Carragher C S Potter and C R Safinya 762

Langmuir 2009 25 2979ndash2985 763

11 A Zidovska K K Ewert J Quispe B Carragher C S Potter and C R Safinya Soft 764

Matter 2011 7 8363ndash8369 765

12 A Zidovska K K Ewert J Quispe B Carragher C S Potter and C R Safinya 766

Biochim Biophys Acta - Biomembr 2009 1788 1869ndash1876 767

13 M Karlsson K Sott M Davidson A S Cans P Linderholm D Chiu and O Orwar 768

Proc Natl Acad Sci U S A 2002 99 11573ndash11578 769

14 A Karlsson R Karlsson M Karlsson A-S Cans A Stromberg F Ryttsen and O 770

Orwar Nature 2001 409 150ndash152 771

15 M Karlsson K Sott A S Cans A Karlsson R Karlsson and O Orwar Langmuir 772

2001 17 6754ndash6758 773

16 E Evans H Bowman A Leung D Needham and D Tirrell Science (80- ) 1996 774

273 933ndash935 775

17 U Seifert Configurations of fluid membranes and vesicles 1997 vol 46 776

18 J Kaumls and E Sackmann Biophys J 1991 60 825ndash844 777

19 N Bobrovska W Goacuteźdź V Kralj-Iglič and A Iglič PLoS One 2013 8 e73941 778

20 A Iglič B Babnik K Bohinc M Fošnarič H Haumlgerstrand and V Kralj-Iglič J 779

Biomech 2007 40 579ndash585 780

21 K Bohinc T Slivnik A Iglič M Brumen and V Kralj-Iglič J Phys Chem C 2007 781

111 9709ndash9718 782

22 A Iglič H Haumlgerstrand P Veranič A Plemenitaš and V Kralj-Iglič J Theor Biol 783

2006 240 368ndash373 784

23 V Kralj-Iglič and P Veranič Adv Planar Lipid Bilayers Liposomes 2006 5 129ndash785

149 786

24 A Iglic and V Kralj-Iglic Adv Planar Lipid Bilayers Liposomes 2006 4 253ndash279 787

25 S Penič L Mesarec M Fošnarič L Mroacutewczyńska H Haumlgerstrand V Kralj-Iglič and 788

A Iglič Front Phys 2020 8 342 789

26 P Yager and P E Schoen Mol Cryst Liq Cryst 1984 106 371ndash381 790

27 T G Barclay K Constantopoulos and J Matisons Chem Rev 2014 114 10217ndash791

10291 792

31

28 B Božič G Gomišček V Kralj-Iglič S Svetina and B Žekš Eur Biophys J 2002 793

31 487ndash496 794

29 A Iglič H Haumlgerstrand M Bobrowska-Haumlgerstrand V Arrigler and V Kralj-Iglič 795

Phys Lett A 2003 310 493ndash497 796

30 R Bar-Ziv and E Moses Phys Rev Lett 1994 73 1392ndash1395 797

31 S Chiruvolu H E Warriner E Naranjo S H Idziak J O Raumldler R J Plano J A 798

Zasadzinski and C R Safinya Science (80- ) 1994 266 1222ndash1225 799

32 V Kralj-Iglič A Iglič H Haumlgerstrand and P Peterlin Phys Rev E 2000 61 4230ndash800

4234 801

33 L Mesarec M Drab S Penič V Kralj-Iglič and A Iglič Int J Mol Sci 2021 22 802

2348 803

34 L Van Renterghem S L K W Roelants N Baccile K Uyttersprot M C Taelman 804

B Everaert S Mincke S Ledegen S Debrouwer K Scholtens C V Stevens W 805

Soetaert L Van Renterghem S L K W Roelants N Baccile K Uytersprot M C 806

Taelman B Everaert S Mincke S Ledegen S Debrouwer K Scholtens C V 807

Stevens and W Soetaert Biotechnol Bioeng 2018 115 1195ndash1206 808

35 J Clark and F Deswarte Introduction to Chemicals from Biomass Second Edition 809

John Wiley amp Sons 2015 810

36 K V Sajna R Houmlfer R K Sukumaran L D Gottumukkala and A Pandey in 811

Industrial Biorefineries and White Biotechnology eds A Pandey R Houmlfer M 812

Taherzadeh K M Nampoothiri and C Larroche Amsterdam Oxford Waltham 813

Elsevier 2015 pp 499ndash521 814

37 R Marchant and I M Banat Trends Biotechnol 2012 30 558ndash565 815

38 K Hill and C LeHen-Ferrenbach in Sugar-Based Surfactants Fundamentals and 816

Applications ed C C Ruiz Boca Raton CRC Press 2009 pp 1ndash20 817

39 L Rodrigues I M Banat J Teixeira and R Oliveira J Antimicrob Chemother 818

2006 57 609ndash618 819

40 J D Desai and I M Banat Microbiol Mol Biol Rev 1997 61 47ndash64 820

41 S L K W Roelants K Ciesielska S L De Maeseneire H Moens B Everaert S 821

Verweire Q Denon B Vanlerberghe I N A Van Bogaert P Van der Meeren B 822

Devreese and W Soetaert Biotechnol Bioeng 2016 113 550ndash559 823

42 N Baccile M Selmane P Le Griel S Preacutevost J Perez C V Stevens E Delbeke S 824

Zibek M Guenther W Soetaert I N A Van Bogaert and S Roelants Langmuir 825

2016 32 6343ndash6359 826

32

43 N Baccile A-S Cuvier S Preacutevost C V Stevens E Delbeke J Berton W Soetaert 827

I N A Van Bogaert and S Roelants Langmuir 2016 32 10881ndash10894 828

44 N Baccile C Seyrig A Poirier S A Castro S L K W Roelants and S Abel Green 829

Chem 2021 23 3842ndash3944 830

45 A Salis L Cappai C Carucci D F Parsons and M Monduzzi J Phys Chem Lett 831

2020 11 6805ndash6811 832

46 J Schindelin I Arganda-Carreras E Frise V Kaynig M Longair T Pietzsch S 833

Preibisch C Rueden S Saalfeld B Schmid J-Y Tinevez D J White V 834

Hartenstein K Eliceiri P Tomancak and A Cardona Nat Methods 2012 9 676ndash835

682 836

47 T Parasassi and E Gratton J Fluoresc 1995 5 59ndash69 837

48 M M Dodes Traian F L Gonzaacutelez Flecha and V Levi J Lipid Res 2012 53 609ndash838

616 839

49 D M Owen C Rentero A Magenau A Abu-Siniyeh and K Gaus Nat Protoc 840

2012 7 24ndash35 841

50 W Stillwell in An Introduction to Biological Membranes Academic Press 2016 pp 842

135ndash180 843

51 W Stillwell in An Introduction to Biological Membranes Academic Press 2016 pp 844

181ndash220 845

52 B Tenchov Chem Phys Lipids 1991 57 165ndash177 846

53 R Koynova and M Caffrey Biochim Biophys Acta 1998 1376 91ndash145 847

54 F Schmid S Dolezel O Lenz and S Meinhardt J Phys Conf Ser 2014 487 848

012004 849

55 S Meinhardt R L C Vink and F Schmid Proc Natl Acad Sci U S A 2013 110 850

4476ndash4481 851

56 O Glatter and O Kratky Small Angle X-ray Scattering Academic Press London 852

1982 853

57 B A Cunningham A D Brown D H Wolfe W P Williams and A Brain Phys 854

Rev E 1998 58 3662ndash3672 855

58 M Nadler A Steiner T Dvir O Szekely P Szekely A Ginsburg R Asor R Resh 856

C Tamburu M Peres and U Raviv Soft Matter 2011 7 1512ndash1523 857

59 C Tanford The hydrophobic effect Formation of micelles and biological membranes 858

John Wiley amp Sons Inc New York 1973 859

60 M Masuda and T Shimizu Langmuir 2004 20 5969ndash5977 860

33

61 A Zidovska K K Ewert J Quispe B Carragher C S Potter and C R Safinya 861

Block Liposomes Vesicles of Charged Lipids with Distinctly Shaped Nanoscale 862

Sphere- Pear- Tube- or Rod-Segments Elsevier Inc 1st edn 2009 vol 465 863

62 A Iglic V Kralj-Iglic B Božič M Bobrowska-Haumlgerstrand B Isomaa and H 864

Haumlgerstrand Bioelectrochemistry 2000 52 203ndash211 865

63 T Shimizu M Masuda and H Minamikawa Chem Rev 2005 105 1401ndash1444 866

64 J V Selinger and J M Schnur Phys Rev Lett 1993 71 4091ndash4094 867

65 J M Schnur Science (80- ) 1993 262 1669ndash1676 868

66 B Tenchov R Koynova and G Rapp Biophys J 2001 80 1873ndash1890 869

67 R Koynova and B Tenchov Ser Biomech 2017 31 3ndash19 870

68 S J Angyal and R Le Fur Carbohydr Res 1984 126 15ndash26 871

69 J K Whitesell and M A Minton J Am Chem Soc 1987 109 225ndash228 872

70 T Takayama S Ando and I Ando J Mol Struct 1990 220 245ndash250 873

71 K Gawrisch N V Eldho and I V Polozov Chem Phys Lipids 2002 116 135ndash151 874

72 F Tielens C Gervais J F Lambert F Mauri and D Costa Chem Mater 2008 20 875

3336ndash3344 876

73 G A Jeffrey and Y Yeon Acta Crystallogr Sect B 1986 42 410ndash413 877

74 B Berglund and R W Vaughan J Chem Phys 1980 73 2037ndash2043 878

75 C Gervais C Coelho T Azaiumls J Maquet G Laurent F Pourpoint C Bonhomme 879

P Florian B Alonso G Guerrero P H Mutin and F Mauri J Magn Reson 2007 880

187 131ndash140 881

76 C Akita T Kawaguchi F Kaneko H Yamamoto and M Suzuki J Phys Chem B 882

2004 108 4862ndash4868 883

77 R N Purusottam L Seacutenicourt J J Lacapegravere and P Tekely Biochim Biophys Acta - 884

Biomembr 2015 1848 3134ndash3139 885

78 X Biarneacutes A Ardegravevol A Planas C Rovira A Laio and M Parrinello J Am Chem 886

Soc 2007 129 10686ndash10693 887

79 L M J Kroon‐Batenburg and J Kroon Biopolymers 1990 29 1243ndash1248 888

80 K N Kirschner and R J Woods Proc Natl Acad Sci U S A 2001 98 10541ndash889

10545 890

81 S Svenson B Kirste and J H Fuhrhop J Am Chem Soc 1994 116 11969ndash11975 891

82 I Sack S Macholl J H Fuhrhop and G Buntkowsky Phys Chem Chem Phys 892

2000 2 1781ndash1788 893

83 S Svenson J Koening and J H Fuhrhop J Phys Chem 1994 98 1022ndash1028 894

34

84 S Svenson A Schaefer and J H Fuhrhop J Chem Soc Perkin Trans 2 1994 2 895

1023ndash1028 896

85 K Bock and C Pedersen Adv Carbohydr Chem Biochem 1983 41 27ndash66 897

86 T B Grindley Structure and Conformation of Carbohydrates Springer Berlin 898

Heidelberg 2008 899

87 F Tantakitti J Boekhoven X Wang R V Kazantsev T Yu J Li E Zhuang R 900

Zandi J H Ortony C J Newcomb L C Palmer G S Shekhawat M O De La 901

Cruz G C Schatz and S I Stupp Nat Mater 2016 15 469ndash476 902

903

1

Supporting Information 904

905

Topological connection between vesicles and nanotubes in single-906

component lipid membranes driven by head-tail interactions 907

908

Niki Baccilea Ceacutedric Lorthioira Abdoul Aziz Baa Patrick Le Griela Cristina Coelhob Javier 909

Perezc Wim Soetaertd Sophie L K W Roelantsd 910

911

a Sorbonne Universiteacute Centre National de la Recherche Scientifique Laboratoire de Chimie de 912

la Matiegravere Condenseacutee de Paris LCMCP F-75005 Paris France 913

b Institut des Mateacuteriaux de Paris Centre Sorbonne Universiteacute Paris France 914

c Synchrotron Soleil LrsquoOrme des Merisiers Saint-Aubin BP48 91192 Gif-sur-Yvette Cedex 915

France 916

d InBio Department of Biotechnology Ghent University Ghent Belgium 917

918

919

Corresponding author 920

Dr Niki Baccile 921

E-mail address nikibaccilesorbonne-universitefr 922

Phone +33 1 44 27 56 77 923

924

925

926

2

a) 927

Parameter Method

Dry Matter (DM ) 996 Infrared balance 105degC

Glucose () 012 HPLC-Metacarb

Glycerol () nd HPLC-Metacarb

Free Fatty Acid content () lt001 Internal method BBEPP

GC with FID Detector

Oil () lt001 Internal method BBEPP

GC with FID Detector

Moulds (CFUg) lt10 3M-Nordval ndeg16 B

Yeasts (CFUg) lt10 3M-Nordval ndeg16 B

Anaerobic count (CFUg) lt10 SP-VG M005 B

Protein () tbd BCA protein assay

928

Figure S 1 - a) HPLC-ELSD chromatogram and composition table 929

930

931

932

933

3

b) 934

935

Group Ndeg δ ppm

CH2 1 354

CH2 2 153

CH2 3-712-14 132

CH2 811 203

CH 910 534

CH2 15 162

CH2 16 388

CH 1 424

CH 2 316

CH 3 334

CH 4 329

CH 5 354

CH2 6 366

Figure S 1 (next) ndash b) 1H NMR spectrum (MeOD-d4) recorded on the purified G-C181-OH compounds 936

and related assignment 937

938

2

811910 1rsquo

5rsquo1

3rsquo4rsquo

1rsquo1910

asymp asymp asympHDO MeOD Acetone

6rsquo2rsquo

1615

3-712-14

4

939

940

Figure S 2 ndash Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 130degC 941

5

942

943

Figure S 3 - Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 90degC 944

945

90degC

1

22

6

946

947

948

949

Figure S 4 ndash Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 70degC 950

951

21

11

1

2

5

3

4

45

5

6

7

7

7

952

953

954

Figure S 5 ndash Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 25degC 955

956

957

1

1 1

8

958

Figure S 6 ndash a) SANS experiment recorded at 90degC for a 5 mgmL G-C181-OH aqueous solution b-c) 959

WAXS experiments recorded on aqueous solutions of G-C181-OH at b) 5 mgmL and c) 25 mgmL d) 960

Structural interpretation of the repeating distances pointed out in c) 961

962

963

964

001 01 1

q Aring-1

25degC

68degC

88degC

111degC

01 1q Aring

-1

02 04 06 08 1q Aring

-1

001 01

10-1

100

101

Inte

nsity

cm

-1

q Aring-1

-2

T= 90degC

SANS WAXS WAXS

a) b) c)

dv

dL d= 144 Aring-1

di

ddH2O

dt

dvdL

dtdi d

d)

9

965

966

Figure S 7 ndash ssNMR MAS (10 kHz) experiments recorded on a G-C181-OH dry powder a) 1H spectra 967

recorded from 110degC to 70degC b) 13C single pulse (SP) spectra recorded with high power 1H decoupling from 968

110degC to 70degC c) 13C cross polarization (CP) spectra recorded at 110degC and contact time tc of 50 micros and 969

3000 micros d) 13C CP spectra recorded at 70degC and for 20 lt tc micros lt 3000 970

971

1

811

910

CH2 CH2

1rsquo

6rsquo

8 6 4 2 0 ppm

110

90

80

70

910

1rsquo

6rsquo16

1

2rsquo3rsquo4rsquo

811

OH

45 40 35 30 25 20 15

ppm

200 180 160 140 120 100 80 60 40 20 0

ppm140 120 100 80

ppm

CH2-gauche

3004

110

90

80

70

65

60

50

25

C91013003

C9 - 12869

CH2-all-trans

3198

C1rsquo10335

a) b)

1H

T degCT degC

13C Single Pulse MAS

A

B

A

B

70degC70degC

5rsquo

215

200 180 160 140 120 100 80 60 40 20 0

ppm

13C CPMAST= 110degC

200 180 160 140 120 100 80 60 40 20 0

ppm

132 131 130 129 128 127 126

ppm

106 105 104 103 102

ppm

13C CPMAST= 70degC

50

tc micros

3000

20

tc micros

50 500 1000 3000

tc micros

C102

13024 C910113003

C92

12869

C1rsquo213082

C1rsquo110443

A B

106 105 104 103 102

ppm

3000

1000

500

50

20

AB

c)

d)

10

972

Figure S 8 ndash a) Short and d) long T2 components displayed by the 1H transverse relaxation signal c) 1H 973

transverse relaxation signals recorded on a G-C181-OH dry powder from 110degC to 70degC The area of each 974

given resonance is plotted (log-lin) as a function of the echo time (CPMG pulse sequence) for each 975

temperature and fitted using the equation given above T2s and T2l indicate the short and long T2 relaxation 976

times respectively obtained by describing the 1H transverse relaxation signal using a biexponential decay 977

For H1rsquo H5rsquo and H6rsquo the 1H relaxation function could be described by a monoexponential decay 978

- Protons from the sugar group (H1rsquo H5rsquo H6rsquo) show a single T2 relaxation component (Figure 979

6c and Figure S 8) while protons of the acyl chain (CH2 H910 H811) show both short (Figure 980

6c) and long (Figure 6d) T2 relaxation component 981

- The short T2 relaxation time systematically decreases from 110degC to 70degC for all species by a 982

factor ranging from two to seven depending on the specific chemical group Very long T2 are 983

observed at 110degC for all acyl species while nearly unchanged values are measured below 984

90degC without any clear trend 985

- The sugar protons (H1rsquo H5rsquo H6rsquo) have the shortest set of T2 values (lt 1 ms) below 100degC 986

The long component of T2 generally ranges between 4 ms and 12 ms 987

0 1 2 3 4 5 6 7 8

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1 2 3 4 5 6 7 8

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

H1

H8H11

H9H10

CH2

H1rsquo

H6rsquo

CH2

H910H811 H1CH2

H1rsquo H5rsquo6rsquo

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

OH

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45 T= 110degC

T= 90degC

T= 80degC

T= 70degC

Log

In

ten

sity

Time ms

Fit

)

CH2 H910 H811 H1 H56 H1 OH

00

05

10

15

20

25

30

35

40

T2 m

s

T= 110degC

T= 90degC

T= 80degC

T= 70degC

CH2 H910 H811 H1

0

2

4

6

8

10

12

T2 m

s

T= 110degC

T= 90degC

T= 80degC

T= 70degC

Short T2(1H) Long T2(1H)

a) b)

c)

11

Table S 1 ndash 1H chemical shifts of G-C181-OH in solution and in the solid-state Solid-state spectra are 988

recorded at a MAS frequency of 10 kHz in DMSO-d6 broad peak 989

990

δH ppm

Group Ndeg Liquid

(MeOD-d4) Solid

(T= 110degC) Solid

(T= 70degC)

CH2 H1 354 357

CH2 H2 153 156

CH2 H3-712-14 132 135

CH2 H811 203 203

CH H910 534 534

CH2 H15 162 164

CH2 H16 388 383

CH H1 424 432

CH H2 316 335

CH H3 334 335

CH H4 329 335

CH H5 354 351 357

CH2 H6 366 383

OH OH1 432

45 463

OH OH2 492

OH OH3 490

OH OH4 487

OH OH6 445

991

992

H1H9H10H1rsquo

H6rsquo

OH2rsquo

OH1

12

Detailed discussion about the NOESY and DQ-SQ 2D 1H-1H ssNMR experiments presented in 993

Figure 9 994

995

2D 1H-1H ssNMR experiments are a unique tool to explore through-space proximities 996

between 1H nuclei ssNMR offers a large number of possibilities each adapted to a given 997

system The advantage of 2D maps with respect to a dynamics approach is their direct 998

readability in clear two nuclei which exchange their magnetization through-space (whichever 999

mechanism may be responsible for this exchange) will display a specific cross-peak However 1000

the way cross-peaks are presented and interpreted strictly depends on the nature of the pulse 1001

program which in turn is chosen according to the physical nature of the sample 1002

1D temperature-dependent 1H NMR experiments performed on G-C181-OH (Figure 1003

4a in the main text) indicate sharp peaks at 90degC and broad peaks at 70degC Narrow peaks 1004

indicate fast andor large-amplitude molecular tumbling averaging the strong 1H-1H dipolar 1005

interactions This excludes the use of pulse sequences like DQ-SQ which is not efficient under 1006

these conditions For this reason we employ spin-diffusion-based pulse sequences like 1007

NOESY which is more adapted for samples with high molecular mobility When mixing times 1008

are chosen to be short (lt 1 s) one generally explores through-space proximities within about 5 1009

10 Aring On the contrary when molecular motions are restricted dipolar coupling becomes 1010

significant 1H NMR peaks are broader and pulse sequences like DQ-SQ can now be used 1011

with a spatial resolution of less than 5 Aring 1012

Whenever possible DQ-SQ experiments are preferred over spin diffusion-based pulse 1013

sequences because they are more selective in particular for likewise interactions which are 1014

not discriminated in NOESY-type experiments This is illustrated by the typical interaction 1015

schemes for NOESY (top Figure 9a) and DQ-SQ (top Figure 9b) For two given A and B 1016

protons 2D NOESY maps will display two on-diagonal and two square-shaped symmetrical 1017

off-diagonal cross-peaks Off-diagonal cross-peaks illustrate the mutual A-B and B-A 1018

interaction but the on-diagonal peaks are systematic and do not prove any A-A or B-B 1019

interaction Similarly to NOESY 2D DQ-SQ maps display off-diagonal peaks (symmetric 1020

across the diagonal top Figure 9b) upon A-B and B-A interaction but in contrast they will 1021

display on-diagonal peaks only in the presence of self-interaction In the tutorial example of 1022

Figure 9b top the on-diagonal A-A peak indicates mutual A-A interaction while lack of on-1023

diagonal B-B peak evidences the absence of B-B proximities Combining NOESY and DQ-SQ 1024

experiments is capital to study the local arrangement of G-C181-OH and its evolution with 1025

temperature 1026

13

At T= 90degC the H1 and H5rsquo contributions are very close but possible to discriminate 1027

them (δ(H1)= 354 ppm δ(H5rsquo)= 351 ppm) according to 1H NMR chemical shifts in Table S 1028

1 2D 1H-1H NOESY in Figure 9a also discriminates between the two components attributed 1029

on the basis of the interactions with their mutual neighbors (green H1-H2 and H5rsquo-H1rsquo squares 1030

in Figure 9a) On this basis 2D NOESY does not show a direct H1rsquo-H1 correlation (blue red-1031

crossed) but rather an extensive network of intra-sugar interactions These features exclude 1032

suggest intra- and inter-sugar interactions and they exclude head-tail arrangements as a major 1033

molecular configuration at 90degC 1034

At 70degC the 1H NMR peaks are broader indicating a reduction of the T2 (1H) relaxation 1035

times andor residual homonuclear dipolar interactions excluding the use of NOESY In this 1036

case 1H-1H DQ-SQ correlation experiments are best suited Figure 9b presents the 1H-1H DQ-1037

SQ experiment on G-C181-OH at 70degC Peaks are broader than in NOESY and attribution may 1038

be cumbersome However the experiment is performed at 700 MHz and MAS= 22 kHz which 1039

improves spectral resolution with respect to the same experiment performed at 300 MHz and 1040

MAS= 10 kHz The auto-peak at δH= 356 ppm in the SQ dimension is attributed to H1-H1 spin 1041

pairs mainly resulting from intra CH2 interaction while the DQ cross-peaks at δH= 156-356 1042

ppm and δH= 432-356 ppm respectively origin from H2-H1 and H1rsquo-H1 interactions (Figure 1043

9b) DQ-SQ map does not show any auto-peak corresponding to H1rsquo-H1rsquo or Hxrsquo-Hxrsquo (x= 2rsquo 1044

3rsquo 4rsquo) interactions (red-crossed blue Figure 9b) thus excluding any proximity (lt 5 Aring) between 1045

close glucose neighbors On the contrary DQ-SQ maps (recorded both at 700 MHz Figure 9b 1046

or at 300 MHz not shown) systematically show DQ cross-peaks which can only be attributed 1047

to H1rsquo-H1 interactions (δH= 432-356) The presence of these H1rsquo-H1 interactions can only be 1048

explained by a head-tail arrangement of the G-C181-OH molecules as illustrated below the 1049

DQ-SQ map in Figure 9b 1050

12

vesicles while arrow Ndeg6 points at two vesicles fusing into the flat membrane Additional areas 302

where vesicle fusion is prominent are indicated by the oscillations at the outer skirt of broad 303

lamellae and pointed by arrow Ndeg7 304

Finally below Tm1 at T= 25degC for instance mainly flat crystals with no visible 305

exvaginations or tubes are visible (Figure S 5) by cryo-TEM and the material turns into an 306

insoluble precipitate Massive presence of flat crystalline structures is also clear from the bulk 307

iridescent behavior of a G-C181-OH solution under manual shear 308

In summary despite the uncertainty related to the actual state of equilibrium of the 309

glucolipids at the moment of cryofixation cryo-TEM depicts a vesicle phase at Tgt 100degC a 310

flat lamellar phase below Tm1 and an interconnected vesicle-nanotube network between Tm1 and 311

about 100degC The combination of statistically-relevant fluorescence microscopy and 312

scatteringdiffraction experiments provides an additional proof of the coexistence of 313

vesiclenanotube systems between the two transition temperatures Please note that the 314

experiments reported below have been performed under controlled conditions of temperature 315

which is then indicated by the use of T instead of T 316

13

317

Figure 5 ndash a-c) Confocal fluorescence microscopy images recorded at 90degC and 70degC for a 5 mgmL G-318

C181-OH aqueous solution d-g) Typical fluorescence microscopy images recorded at T= 80degC using 319

Laurdan (refer to materials and methods section for more details on this fluorophore) Excitation 320

wavelength is 405 nm for all the images while emission wavelength is 490 nm in (df) and 440 nm in (eg) 321

h) Full emission spectra obtained from the integration of various fluorescence images recorded from 412 322

nm to 600 nm (step 5 nm) at 90degC 80degC and 70degC i) Box plot of GP defined as 119918119927 =(120640120786120786120782minus120640120786120791120782)

120640120786120786120782+120640120786120791120782 with 120640120786120786120782 323

and 120640120786120791120782 being the emission wavelengths at 440 nm and 490 nm at 90degC 80degC and 70degC Each data point 324

l= 490 nm l= 440 nm

l= 490 nm l= 440 nm

10 μm400 450 500 550 600

00

02

04

06

08

10

No

rma

lize

d I

nte

nsity

l nm

90degC

80degC

80degC

70degC

70degC

400 450 500 550 60000

02

04

06

08

10

No

rma

lize

d I

nte

nsity

l nm

90degC

80degC

80degC

70degC

70degC

d) e)

f) g)

h)

i)

T= 80degC

Laurdan

T= 80degC

T= 80degC T= 80degC

T= 90degC T= 70degC T= 70degC

a) b) c)

1

1

1

14

corresponds to a set of regions of interest (ROIs) (54 at 90degC 65 at 80degC and 75 at 70degC) selected in the 325

corresponding fluorescence images ROIs were representatively selected so to include the entire image but 326

also vesicles alone tubes alone and mixtures of vesicles and tubes 327

328

Figure 5ab shows two typical confocal fluorescence microscopy images recorded on a 329

G-C181-OH solution (5 mgmL) in a flame-sealed capillary heat at T= 90degC and T= 70degC 330

Although the resolution of optical microscopy is not comparable with cryo-TEM the sample is 331

broadly composed of vesicles coexisting with nanotubes whereas in some cases the nanotube 332

starts from the vesicle itself (arrow Ndeg1) in good agreement with cryo-TEM data Furthermore 333

SANS recorded at T= 90degC (Figure S 6a) shows a scattering profile with a pronounced -2 334

dependence of the scattering intensity against the wavevector q in the log-log scale This 335

indicates the massive presence of a flat interface and it is expected typical of bilayers in 336

vesicular morphologies56 However the typical signature of nanotubes cannot be observed in 337

the SANS profile and this can be explained by both an argument related to the relative lower 338

amount of nanotubes with respect to vesicles at 90degC but also to the steeper slope of flat 339

morphologies (-2) compared to tubes (-1) the signal of the former masking the tube signal at 340

low wavevector values The SANS profile is characterized by a diffraction peak at q= 0167 Aring-341

1 of which the assignment is refined by WAXS data 342

Temperature-driven WAXS experiments in Figure S 6bc (5 mgmL-1 and 25 mgmL-1 343

respectively) show a lamellar order possibly attributed to a multilamellar structure of the 344

vesicles (dv Figure S 6d) at high temperature and to the interplanar arrangement of the 345

crystalline flat structures (dL Figure S 6d) at T= 25degC The sharp (001) reflection indicates 346

typical interlamellar distance of 353 Aring at 25degC although this value varies with temperature 347

(400 Aring at 111degC 372 Aring at 88degC) thus confirming its assignment to a hydration-dependent 348

interlamellar order More intriguing is the second broader peak at q= 0403 Aring-1 (d= 156 Aring) 349

and its reflection both clearly visible at 88degC below Tm2 We attribute it to the inter-glucolipid 350

arrangement inside the tubular phase (dt Figure S 6d) coexisting with the vesicle phase 351

between Tm1 and Tm2 and to the planar lipid arrangement (di Figure S 6d) in the rigid flat 352

lamellar structure below Tm2 Its temperature- hence hydration- independent position also 353

confirms its structural origin Finally besides the water-structure peak at about 2 Aring-1 we also 354

observe a broad peak at 144 Aring-1 characteristics of the lateral packing (d Figure S 6d) of lipids 355

in bilayers While its position could be consistent with an Lα phase5758 above Tm2 its broadness 356

and positional invariance as well as the coexistence between vesicles and tubes do not allow a 357

precise assignment 358

15

Does G-C181-OH assume a bilayer or interdigitated structure the latter being the one 359

found for other similar bolaform glucolipids4243 The size of G-C181-OH can be estimated to 360

be at the most 32 Aring given by the length L of the C16 tail (L = 22 Aring from the Tanford formula 361

L= 154 + 1265 times n (n= 16)59) and the sugar (about 10 Aring)4243 However the cis conformation 362

of the C=C bond imposes a ldquokinkrdquo of 60deg in the aliphatic chain which reduces the effective 363

length of the oleic backbone to about 17 Aring for an effective total G-C181-OH length of about 364

27 Aring Cryo-TEM provides a tube thickness varying from 35 Aring to 60 Aring and a vesicle membrane 365

thickness of at least 60 Aring WAXS suggests an interlamellar thickness (membrane and water) 366

between 40 Aring and 37 Aring and an inter-lipid distance in the tube of about 16 Aring In the absence of 367

a clear form factor oscillation profile as found for similar glucolipids4243 one must 368

acknowledge that these data pledge neither in favor of fully interdigitated nor a double layer 369

structure and might suggest a tilted lipid arrangement with partial interdigitation60 370

371

13C ssNMR demonstrates that G-C181-OH can assume two coexisting head-head and 372

head-tail configurations 373

Vesicles are characterized by low positive mean and gaussian curvatures while 374

nanotubes are described by a combination of a high positive mean and zero gaussian curvatures 375

In this regard the coexistence of vesicles and nanotubes requires glucolipids to adapt 376

themselves to both environments at the same time For this reason the formation of nanotube-377

vesicle networks generally requires a specific driving force able to impose a high mean and 378

zero gaussian curvature to the membrane In many systems insertion of at least a second 379

constituent (lipid surfactant protein) in the membrane becomes a necessary internal driving 380

force for the nucleation and growth of so-called membrane inclusions from which 381

topologically-connected nanotubes can form10ndash123261 as explained by orientational ordering 382

arguments62 It has also been shown that external driving forces such as use of laser 383

ldquotweezersrdquo30 osmotic pressure6 electroformation5ndash82829 or controlled suction13ndash16 are 384

otherwise necessary as also supported by theoretical studies19 385

The bending energy of tubes is higher than that of a vesicle membrane8 due to the 386

dependence of the membrane free energy from the principal curvatures For this reason 387

isotropic (rotational symmetry along the main axis) but also anisotropic33 lipids generally 388

stabilize low-curvature shapes like flat membranes or vesicles Lipid nanotubes are also a 389

common phase formed by amphiphiles but generally for lipids with a certain complexity in 390

their molecular structures as exampled by diacetylenic derivatives of phospholipids among 391

others262763ndash65 Negative contributions to the bending energy favoring the simultaneous 392

16

presence of interconnected vesicles and tubes was shown to be driven by the presence of 393

anisotropic lipids in a membrane continuum Spontaneous deviations to the elastic theory in the 394

presence of surfactants lipids and proteins with anisotropic structures favor the formation of 395

protrusions with radius of the order of the fraction of a micron9ndash1219ndash2362 396

In the absence of both internal and external driving forces mentioned above G-C181-397

OH should behave as a standard lipid Its DSC profile (Figure 2)536667 under both hydrated and 398

dry conditions identifies a hydration-independent transition (Tm1) slightly above room 399

temperature this is followed by one main transition at higher temperature (Tm2) in the dry 400

sample50 For such a profile one expects a gel phase (Lβ) below Tm1 and a liquid crystalline 401

structure possibly in the morphology of vesicles (L4 phase) possibly above Tm1 and most likely 402

above Tm231 as experimentally found On the basis of the thermogram recorded on the dry 403

powder one may also expect a ldquoripplerdquo (Pβ) phase between Tm1 and Tm2 characterized by 404

periodic undulations of the bilayer (generally addressed to as regions with both liquid order and 405

disorder)5455 406

In the present work instead of a Lβ phase we observe flat crystals below Tm1 and a 407

vesicle-nanotube network instead of the ldquoripplerdquo phase The latter is particularly unexpected 408

because the membrane is composed of a single-component (G-C181-OH) and no anisotropic 409

external force (osmotic hydrodynamic electric) is used to pull and stabilize the tubes In 410

addition the nanotubes are observed at fairly high temperatures above Tm1 (between about 411

60degC and 90degC) while this morphology is generally favored at low temperature (proportionality 412

with 1kT figure 5 in Ref 62) commonly below a given transition temperature characteristics 413

for each specific amphiphile27 Furthermore nanotubes of G-C181-OH seem to be stable in 414

time thus differing from what is found in electroformed vesicles78 They are also observed 415

away from phase transition events (may them be at Tm1 or Tm2) as otherwise reported before 416

for a diacetylenic derivative of phosphatidylcholine26 and many other systems27 417

A stable nanotube-vesicle network for a single-component phospholipid in the absence 418

of external driving forces is not expected and has never been reported before It could be 419

explained by the coexistence of two micro-separated phases of different rigidity This 420

hypothesis is excluded by confocal microscopy on Laurdan-labelled membranes and magic 421

angle spinning (MAS) solid-state NMR (ssNMR) experiments 422

Laurdan is a well-known fluorophore used to put in evidence the coexistence of rigid 423

and fluid domains in both model and biological membranes47ndash49 Laurdan was used here to label 424

the G-C181-OH membranes between Tm1 and Tm2 and to verify the possible micro-segregation 425

and coexistence of rigid (eg Lβ λem= 440 nm) and fluid (eg Lα λem= 490 nm) domains in 426

17

tubes and vesicles respectively A set of quantitative experiments performed on a 5 mgmL-1 427

solution summarized in Figure 5 shows an excess of fluorescence at only λem= 490 nm for 428

temperature between 90degC and 70degC (Figure 5h) The GP value which quantifies the rigidfluid 429

character of the membrane is systematically negative (Figure 5i) indicating an overall fluid 430

membrane at all temperatures GP is quantified on more than 50 regions of interest (ROIs) at 431

each temperature including vesicles and tubes alone as well as connected nanotube-vesicles 432

(Figure 5d-g) The dispersion of GP at 80degC and 70degC between -08 and -02 could suggest the 433

spurious presence of membranes with various degrees of flexibility although fluidity is always 434

predominant The combination of WAXS (broad peak d= 144 Aring-1 invariant with 435

temperature Figure S 6bc) and confocal fluorescence microscopy using Laurdan confirms the 436

single-phase of the membrane composed of G-C181-OH both in vesicles and nanotubes This 437

is also consistent with the lack of phase transitions in DSC experiments between ~85degC and 438

~60degC especially in hydrated samples We conclude that the vesicle-nanotube transition is not 439

driven by the local segregation of gel phase (Lβ) microdomains nor that the nanotubes are 440

composed of such phase 441

ssNMR is the best-suited approach to investigate the origin of the local driving force for 442

the nucleation of anisotropic inclusions of which the average orientation depends on the local 443

membrane and inclusion curvatures and was shown to promote tubing62 ssNMR can provide 444

information on the local molecular mobility eg through the nuclear spin relaxation behavior 445

but also on intermolecular interactions and stereochemistry eg through variations in the 446

chemical shift68ndash70 447

Ideally ssNMR should be performed on diluted G-C181-OH samples above Tm1 448

However fast MAS necessary for high resolution centrifuges the sample As a first 449

approximation hydrated G-C181-OH samples for instance at 50 wt as classically done for 450

phospholipids71 could be studied Unfortunately permeability of the sample holder combined 451

to heating (close to the boiling point of water) and MAS provoke dehydration and even sample 452

ejection After experiencing both of the above this approach was abandoned As a second 453

approximation the study of the dehydrated G-C181-OH powder avoids centrifugation and 454

evaporation If studying dried lipids may not reflect their exact behaviour under diluted 455

conditions valuable structural information can still be obtained in dehydrated molecular 456

systems like proteins thus illustrating that this approach is not unreasonable Furthermore if 457

DSC (Figure 2a) shows that hydration suppresses the Tm2 it also shows that the value of Tm1 is 458

not affected by water at concentrations as low as 10 wt For this reason performing a 459

structural analysis of dehydrated G-C181-OH by ssNMR above Tm1 may help understanding 460

18

the local arrangement of G-C181-OH in two morphologies vesicles and nanotubes having 461

sensibly different curvatures 462

To explore the local mobility of the most relevant chemical groups of G-C181-OH 463

above Tm2 (T= 110degC) and between Tm2 and Tm1 (70 lt T degC lt 90) we combine a series of 1H 464

and 13C experiments Single-pulse (SP) MAS experiments (1H and 13C) when recorded under 465

quantitative conditions allow the detection of all chemical species while cross-polarization 466

(CP) MAS (13C only) experiments are sensitive only to constrained groups for which the 1H-467

13C dipolar coupling is not averaged to zero by the molecular motions 468

469

Figure 6 ndash a) 1H MAS (10 kHz) ssNMR spectra recorded from 110degC to 70degC b) 13C MAS (10 kHz) ssNMR 470

spectra recorded using high power decoupling single pulse (SP) and cross-polarization (CP) pulse sequences 471

Contact time for CP experiments and corresponding temperatures are given on each spectrum Full spectra 472

are given in Figure S 7 473

The most relevant 1H and 13C spectra are given in Figure 6 (full range in Figure 7 and 474

Figure S 7) As expected the 1H solid-state SPMAS spectra show broader peaks compared to 475

the solution-state NMR spectrum (Figure 6a) reflecting the obvious solid-state character of the 476

sample at any temperature The sugar group and the C=C bond resonate in the 25-6 ppm range 477

and 535 ppm respectively while the OH groups of the sugar moiety are found below 45 ppm 478

(Figure 6a full assignment in Table S 1) Very broad (~1 ppm) at 100degC the position of the 479

OH peak varies with the temperature value and becomes narrower (lt~05 ppm) and visible at 480

δH= 464 ppm at T= 70degC Line narrowing and chemical shift variations of the OH resonance 481

in ssNMR is always a direct sign of H-bonding interactions whereas the strength of the H-482

bonding hence its OHhellipH distance is correlated to a dispersion in chemical shifts72ndash75 All other 483

60 55 50 45 40 35 30 25

ppm

110

90

80

70

910

1rsquo

6rsquo16

1

OH

134 132 130 128 126

ppm110 108 106 104 102 100

ppm

SP (T= 110degC)

SP (T= 70degC)

CP 50 ms (T= 70degC)

CP 3 ms (T= 70degC)

910 1rsquo

1H 13C

1

811

910

CH2 CH2

1rsquo

6rsquo

a) b)

C102

13024

C9101

13003

C92

12869 C1rsquo210382

C1rsquo110443

C1rsquo13035

2rsquo3rsquo4rsquo

T degC

5rsquo

19

resonances broaden from 110degC to 70degC thus indicating slower andor more anisotropic 484

reorientational motions of all chemical groups The specific dynamic behaviour of the sugar 485

headgroup is also reflected by its 1H transverse relaxation time T2 whereas the shorter the T2 486

the higher the rigidity Experiments performed between 110degC and 70degC shown and 487

commented in Figure S 8 in the Supporting Information globally show a single short (lt 1 ms 488

below 100degC) T2 relaxation component for H1rsquo H5rsquo and H6rsquo and two both short and long 489

(between 4 ms and 12 ms) T2 for the acyl chain (CH2 H910 H811) even below 100degC 490

At 110degC 13C SPMAS experiments (Figure 6b full range in Figure 7 and Figure S 7 491

assignments in Table 1) show a single peak at δC= 10335 ppm for C1rsquo and at δC= 13003 ppm 492

for C910 as expected for a symmetrical conformation with two chemically-equivalent 493

carbons76 At 70degC the C1rsquo peak disappears while C910 now shows two peaks around 128 494

ppm and 130 ppm (Table 1 full temperature series in Figure S 7b) attributed to C92 (δC= 12869 495

ppm) and C10 in a non-symmetric configuration respectively for which these two carbons are 496

chemically inequivalent76 More precisely the resonance at 130 ppm is actually the 497

superposition of two close resonances C9101 (δC= 13003 ppm) and C102 (δC= 13024 ppm) 498

where subscripts 1 and 2 stand for two different conformations of the cis-olefin group as 499

reported for various phases of oleic acid in the vicinity of 0degC76 500

501

Table 1 ndash 13C Chemical shifts of G-C181-OH in solution and solid-state Solid-state experiments are 502

recorded at MAS= 10 kHz All spectra are reported in Figure 7 and Figure S 7 The contact time for cross 503

polarization is 3 ms In all ssNMR experiments temperature is first raised at 110degC and then lowered to 504

either 90degC or 70degC Attribution of the 13C resonances in Table 1 is performed on the basis of the solution 505

spectrum (Figure S 1b) Additional peaks observed in both single pulse and cross polarization ssNMR 506

experiments are attributed when possible to a configuration effect of the given atom eg C9101 and C92 507

indicate two configurations of C9 All new peaks observed in solid-state and not detected in solution are 508

highlighted in red for convenience 509

510

511

δ ppm

Peak Liquid (DMSO-d6)

Solid Single Pulse

T= 110degC

Solid Single Pulse

T= 90degC

Solid Single Pulse

T= 70degC

Solid Cross Polarization

T= 70degC

C102

- - 13024 13024

C910C9101 13009 13003 13003 13003 13003

C1C9C10C1rsquo

C6rsquo

20

C92 - - - 12869 12869

C1rsquoC11 10329 10335 10335 - 10443

C12 - - - - 10382

C3 7724 7709 7702 - 7829

- 7634 7629 - 7621

7525

C2 7388 7398 7390 7370 7370

C4 7052 7077 7054 7041 7098

C16 6900 7016 - - 7029

- - - - 6793

C1C5C6 6153 6241 6236 6239 6262

C1C5C6 6117 - - - 5956

C2 3300 3290 3290 3283 3279

CH2(trans) - - - 3198 3196

CH2(gauche)C15 2941 2981 2987 3004 -

C811 2706 2747 2747 2832 2817

CH2 2597 2625 2628 2630 2817

512

13C CPMAS spectra at 70degC (Figure 6b full range in Figure 7 and Figure S 7d) at 513

increasing contact times confirm the presence of two cis-olefin environments one for which 514

C910 are equivalent labelled here C9101 and one for which they are inequivalent labelled 515

C92 C102 Both environments are rigid enough to be observed at short contact time (tc= 50 μs) 516

A similar double environment is also observed for the sugar C1rsquo CPMAS at 70degC shows two 517

peaks (C1rsquo1 at δC= 10443 ppm C1rsquo2 at δC= 10382 ppm Figure 6b Figure S 7d) These are 518

significantly shifted downfield from the C1rsquo peak observed at 110degC and are detected at contact 519

times as short as 50 μs thus indicating their rigid character One also finds that CPMAS is 520

totally inefficient at 100degC for short contact times (50 μs) and poorly efficient even at 3 ms 521

typical of fast-tumbling molecules (Figure S 7c) also both gauche (δC= 3004 ppm) and all-522

trans (δC= 3198 ppm) conformations of the CH2 carbons coexist at 70degC (Figure S 7b)77 523

According to ssNMR one main mobile molecular environment exists at 110degC as 524

shown by the well-resolved 1H spectrum single C=C (C9101) and anomeric (C1rsquo) resonances 525

gauche conformation of the aliphatic CH2 and single T2 of the aliphatic CH2 chain and sugar 526

protons The sugar group is certainly mobile (inefficient 1H13C CPMAS) but still more 527

constrainted than the acyl chain (shorter T2 (1H)) As the temperature is lowered the overall 528

rigidity increases (broadening of 1H NMR peaks faster 1H transverse relaxation and efficient 529

1H13C magnetization transfer during CPMAS) presumably driven by H-bonding 530

interactions (downfield shift and line narrowing of the OH 1H signal very short OH T2 loss of 531

the C1rsquo signal in SPMAS) At the same time a number of spectroscopic data (two concomitant 532

21

resonances for anomeric C1rsquo1 and C1rsquo2 cis-olefin C9101 and C92 C102 and CH2 gauche and 533

all-trans environments) and relaxation behaviour (fast and slow T2 relaxation components) 534

strongly suggests the coexistence of two molecular environments Further proofs of the 535

coexistence of two distinct molecular domains are given by a number of chemical shift 536

variations in the 13C CPMAS spectrum at 70degC compared to the one at 110degC 90degC and 537

solution This is shown in Figure 7 and in the list of 13C chemical shifts in Table 1 538

539

540

Figure 7 - Comparison between solution- and solid-state 13C NMR spectra of G-C181-OH Solid-state 541

experiments are recorded directly on the sample powder at a MAS of 10 kHz (4 mm zirconia rotor) SP 542

Single Pulse CP Cross Polarization In all the ssNMR experiments temperature is first raised at 110degC 543

and then lowered to either 90degC or 70degC Segmented lines are meant to highlight the known peak position 544

in solution thus stressing the strong chemical shift variations in the solid-state 545

546

NMR experiments above Tm2 suggest that vesicles could contain G-C181-OH with a 547

single family of mobile conformations characterized by chemically-equivalent C=C (red 548

Figure 8b) and a single C1rsquo conformation (green Figure 8b) Experiments between Tm1 and Tm2 549

otherwise suggest that two coexisting conformations characterized by chemically-equivalent 550

(red Figure 8b) and inequivalent (magenta Figure 8b) C=C could be associated to nanotubes 551

Two distinct C1rsquo sites (grey Figure 8b) also indicate two configurations of the sugar headgroup 552

A comment on this point will be given in the following paragraphs 553

The local (Aring-scale) variation of molecular mobility below Tm2 drives the coexistence of 554

two distinct molecular environments of the same G-C181-OH molecule This occurs within a 555

single-phase domain (as deduced by WAXS) of equivalent membrane rigidity (as derived from 556

Laurdan probe in confocal microscopy) Although never reported before the difference in the 557

local mobility of the same amphiphile (here G-C181-OH) could satisfy the theoretically-558

140 120 100 80 60 40 20

CP(3 ms)-70degC

SP-70degC

SP-90degC

SP-110degC

ppm

LiquidC910

C1rsquo

C2rsquo

C4rsquo

C3rsquo

C1

C6rsquoC5rsquo

C1

6

C8

11

C2 C

H2

C1

5

CH

2

DM

SO

13C

22

required conditions that induce local lipid segregation and formation of membrane inclusions 559

(Figure 8c) In case of their non-zero average orientation the local membrane curvature 560

increases62 and nanotubes can spontaneously form (Figure 8bc) More intriguing this 561

mechanism occurs at least 10degC above Tm1 although it should not be favored above the Tm27 562

From both experimental and theoretical backgrounds the conditions of topologically-563

connected vesicles-nanotubes systems are met only when either internal (two components) or 564

external driving forces are applied19 In the present single-lipid system of which the phase 565

behavior is summarized in Figure 8a the most plausible driving force could be explained by 566

the NMR results collected on the dry sample G-C181-OH concomitantly occupies two 567

molecular environments characterized by 1) two distinct conformations of the sugar 568

displaying equivalent mobility and 2) two conformations of the acyl chain unsaturation 569

570

23

571

Figure 8 ndash a) Thermotropic phase behavior of G-C181-OH at 5 mgmL-1 Note that Tm2 is observed on the 572

dry powder only b) Possible relationship between the molecular environment as derived from NMR and 573

membrane curvatures All the molecular drawings refer to the same G-C181-OH molecule (bulky ellipsoid 574

equiv sugar headgroup) while each color corresponds to a given configuration Grey of C1rsquo1 and C1rsquo2 indicates 575

the uncertainty about the assignment of the specific NMR peak The curvature C is defined as 1R R being 576

the curvature radius R1 R2 radius of curvature CG Gauss curvature CM Mean curvature c) Free 577

interpretation of the anisotropic high-curvature (CMgt 0 CG= 0) inclusions in the membrane (low CM and 578

CG) continuum driven by inter-sugar H-bonding and reduced intramembrane diffusion for a G-C181-OH 579

glucolipid 580

581

Flat crystalline structure

Nanotube-vesicle network

Vesicles

48degC

120deg

106 105 104 103 102 101

ppm

H9 H10H1rsquo

106 105 104 103 102 101

ppm106 105 104 103 102 101

ppm106 105 104 103 102 101

ppm

C=CC=C

C1rsquo1C1rsquo2

C1rsquo

Tm2(96degC)Tm1

a)

b)

c)

Same molecule different local orientations and sugar conformations

High curvature membrane

Low curvature inclusion

Flip-FlopStrong H-bonding

CG

CM ( + )

CG 0

CM 0 CG 0

CM 0

G-C181-OH

24

Two related questions are unanswered at the moment What actually drives nanotube 582

formation and is the organization of G-C181-OH within the vesicles and tubes the same Lipid 583

nanotubes have been studied for at least three decades over which broad experimental and 584

theoretical insight has been generated Despite this fact the actual understanding is still matter 585

of debate27 Micelle-to-nanotube and vesicle-to-nanotube phase transitions are generally 586

reported for single amphiphiles below a given Tm and intermediate phase composed of twisted 587

and flat fibers is concomitantly reported Initially thought to be a chirality-driven phenomenon 588

recent studies propose a more general mechanism involving crystallization and symmetry 589

breaking in the lipid packing driven by optimal packing27 If twisted ribbons were observed 590

before for a number of microbial glycolipid bolaamphiphiles44 they are not found for G-C181-591

OH This fact as well as the large body of cryo-TEM images and the observation of nanotubes 592

above Tm1 rather pledge in favour of a direct topological change from positive (vesicles) to 593

zero (nanotubes) Gaussian curvatures In this case vesicle-to-nanotube transitions could be 594

driven by external forces56307813ndash162829 595

Considering the absence of external forces in the present system there are only two 596

possibilities left spontaneous budding of the membrane or nanotube pulling after inter-vesicle 597

collisions due to Brownian motion Both are possible as inclusions composed of a different 598

arrangement of the same molecule in the membrane continuum can theoretically explain 599

spontaneous nanotube formation in an isotropic medium672223629ndash1218ndash21 However the second 600

one is probably more likely to prevail because it could explain the large number of nanotubes 601

connecting two distant vesicles Interestingly similar events occur for any phospholipid-based 602

vesicular solution but nanotubes are extremely rare unless at least two lipids of different 603

molecular structure are mixed together The spectroscopic 1H and 13C and relaxation (1H) 604

ssNMR data even if collected on a dry powder demonstrate that the coexistence of two G-605

C181-OH configurations involving both the sugar headgroup and fatty alcohol chain are 606

possible for this molecule As a matter of fact this is not so surprising as more than one energy 607

minima each corresponding to a given conformation are known for glucose its non-reducing 608

derivative methylglucose and carbohydrates in general78ndash80 609

The presence of two configurations could explain the coexistence between two 610

structures of different curvature but the data presented so far do not explain the origin of these 611

two configurations In other words what drives the change in the local structure of the sugar 612

headgroup and fatty alcohol within the membrane 613

The association of known self-assembled structures to spectroscopic 13C ssNMR data is 614

known for years in particular for gluconamides lipids having a linear sugar headgroup81ndash84 It 615

25

was found that gauche effects in some cases up to the γ position in the sugar chain induce 616

upfield chemical shifts while anti and trans conformations result in downfield shifts Each 617

structure (monolayer crystal bilayer crystal or micellar fiber) was then characterized by a given 618

organization of the gluconamide and a characteristic 13C ssNMR fingerprint which could be 619

explained by gauche γ-gauche anti and trans conformations of the headgroup Similar 620

structural-spectroscopic correlations have been collected over the years for much simpler cyclic 621

sugars85 whereas pyranoses can undergo important conformational changes (chair half-chair 622

boat skew) when the appropriate energy barriers are crossed78ndash8086 A quantitative 623

conformational study based on the 13C chemical shift variations shown in Table 1 would be 624

certainly helpful but very delicate in the context of the present state of the art because a 625

rigorous approach requires at least one known crystal structure8182 which is currently missing 626

for this compound A qualitative approach will then be adopted 627

G-C181-OH is a bolaform amphiphile and flip-flop could be a possible mechanism 628

involved in the nanotube stabilization whereas head-tail interactions have long been considered 629

as important in nanotube-forming bolaform amphiphiles6063 From mere considerations based 630

on T2 (1H) relaxation arguments one could exclude head-tail configurations if the R-CH2OH 631

endgroup (H1) was involved in a direct interaction with the sugar group its corresponding T2 632

would probably be comparable to the sugarrsquos one (OH and H1rsquo) but this does not seem to be 633

the case However this hypothesis would be based on the comparison of the relaxation from 634

CH and CH2 protons additionally involved in different kind of motions Besides the relaxation 635

of H1 is influenced by the strong overlap between H1 and H5rsquo For these reasons relaxation 636

arguments are not solid enough and will not be used 637

We then address this question by exploring intermolecular interactions in a direct 638

manner using 2D 1H-1H homonuclear ssNMR correlation spectroscopy At 90degC in the fast-639

tumbling domain of molecular motion (narrow 1H NMR peaks Figure 6a) and corresponding 640

to the vesicle-rich region spin diffusion (NOESY) experiments are preferred over dipolar 641

coupling based experiments because such interaction is averaged out by molecular motion 642

However when the molecular motion is more constrained in the domain of nanotube stability 643

at 70degC (broad 1H spectra Figure 6a) Double-Quantum Single-Quantum (DQ-SQ) correlation 644

experiments are best suited to probe interaction between close molecular neighbors These 645

experiments are reported in Figure 9 and the related interpretation is only briefly reported 646

below while a more detailed discussion can be found on Page 12 in the Supporting Information 647

26

648

Figure 9 ndash a) 2D 1H-1H NOESY NMR experiment performed at MAS= 10 kHz (4 mm probe) on a Bruker 649

300 MHz Avance III HD spectrometer using 300 ms of mixing time Sample temperature is regulated to 650

90degC reached by cooling from 110degC b) 2D 1H-1H DQ-SQ (BAck-to-BAck BABA pulse sequence used for 651

the DQ excitation scheme) NMR experiment performed at MAS= 22 kHz (25 mm probe) on a Bruker 700 652

MHz Avance III spectrometer using a DQ excitation time equal to 2 rotor periods Sample temperature is 653

regulated to 70degC reached by cooling from 110degC The typical schemes illustrating how to interpret these 654

2D experiments are reported above each experimental map Zooms of the sugar region and configurational 655

interpretation are given below each 2D map More information are given in the text below 656

H16-H16

H1-H1

H9-H10

H1rsquo-H1

H1rsquoH910 H16

H1

H1rsquo-H1rsquo

Hxrsquo-Hxrsquo

2

(pp

m)

(ppm)

315

811910 1rsquo

2rsquo2

3rsquo 4rsquo6rsquo

5rsquo

1rsquo2910

asymp asymp asympHDO MeOD Acetone

1

315

811 9101rsquo

2rsquo2

3rsquo4rsquo6rsquo

5rsquo

1rsquo2 9 10

asympasympasympHDOMeODAcetone

H1rsquo-H1

1rsquo

1

(p

pm

)

(ppm)

H1-H2

H1rsquo-H1

H1rsquo

H910

H1H5rsquo

H2

H1rsquo-H5rsquo

315

811910 1rsquo

2rsquo2

3rsquo 4rsquo6rsquo

5rsquo

1rsquo2910

asymp asymp asympHDO MeOD Acetone

1

H2rsquoH3rsquoH4rsquo

H2rsquoH3rsquoH4rsquo

X= 2rsquo3rsquo4rsquo

NOESY - T= 90degC

1H

1H

1H

1H

DQ-SQ - T= 70degC

a) b)

2

(pp

m)

(p

pm

)

(ppm)

A B

AhellipB

AhellipA

BhellipB

NOESY

(ppm)

A B

AhellipB

DQ-SQ

Single-quantum dimension

Do

ub

le-q

ua

ntu

m d

imen

sio

n

AhellipA

BhellipB

BhellipA

BhellipA

315

811910 1rsquo

2rsquo2

3rsquo 4rsquo6rsquo

5rsquo

1rsquo2910

asymp asymp asympHDO MeOD Acetone

1

lt 5 Aring

lt 5 Aring

H1-H2

H811

H15

H16

H6rsquo

H1rsquo

H1

H5rsquo

H2

H1-H2

H1rsquo-H1

H1rsquo-H5rsquo

H1rsquo

H1 H5rsquo

H1rsquo

H1

H5rsquo

H1H2rsquo H3rsquo H4rsquo

H1-H1

H1rsquo-H1

Hxrsquo-HxrsquoX= 2rsquo3rsquo4rsquo

H1-H2

H1rsquo

Zoom sugar region 28 ndash 45 ppmNOESY DQ-SQ

27

Head-tail or head-head configurations could be probed by the existence of H1rsquo-H1 657

interactions whereas H1 (δH= 356 ppm) is associated to the glucolipid tail and H1rsquo (δH= 432 658

ppm) to the sugar headgroup These atoms are more than 15 Aring apart and a direct interaction 659

could not be observed in both NOESY and DQ-SQ experiments NOESY experiments (Figure 660

9a) show that H1rsquo-H1 correlations are not observed Meanwhile clear-cut direct H1rsquo-H1 661

interactions are indeed observed are detected at T= 70degC (DQ cross-peaks related to H1rsquo-H1 662

correlation in DQ-SQ experiment shown in Figure 9b) Interestingly the absence of auto peaks 663

at δH= 432 ppm and δH= 335 ppm in the SQ dimension excludes H1rsquo-H1rsquo and Hxrsquo-Hxrsquo (x= 2 664

3 4) interactions respectively strengthening the idea that sugar headgroups from two nearest-665

neighbor G-C181-OH molecules are not adjacent in the nanotube-vesicle region 666

Homonuclear 2D ssNMR supports the idea according to which G-C181-OH can 667

undergo molecular flip-flops and a head-tail arrangement in the dry product Extrapolating these 668

conclusions in the lower concentration range lack of H1rsquo-H1 proximity at T= 90degC suggests a 669

head-head configuration in the vesicle-rich region and head-tail configuration within the 670

nanotubes It is however still unclear whether flip-flops are driven by spontaneous temperature-671

induced conformational changes in the glucolipid packing stimulating inclusions within the 672

membrane itself672223629ndash1218ndash21 or by defects generated by inter-vesicle collisions In fact a 673

combination of both is not unlikely Nanotubes are not stable when temperature is 20 to 30 674

degrees above Tm1 indicating that inter-vesicle collisions alone cannot explain the unusual 675

phenomenon In an energy landscape approach87 the head-head configuration is certainly 676

favored at high temperature while both head-head and head-tail configurations could be 677

favored in the mid-temperature region corresponding to the range between Tm1 and Tm2 in the 678

dry powder Both statistical and defect-induced flip-flop events contribute to form high mean 679

zero Gaussian curvature morphologies (nanotubes) for which the head-tail conformation 680

limiting the proximity between bulky sugar headgroups is probably more adapted (Figure 8c) 681

The flip-flop interpretation now reasonably explains the strong differences in the sugar and fatty 682

alcohol conformations deduced by the multiple NMR signals attributed to chemically-683

equivalent groups The existence of various low-energy conformations each having specific 684

NMR signature for sugars in solution78ndash8086 and solid-state81ndash84 further supports our 685

interpretation 686

One last point should be clarified Why between Tm1 and Tm2 are two molecular 687

environments able to coexist instead of a single one of average rigidity as found in the case of 688

ldquoripplerdquo phases50ndash54 We believe that this question can be answered based on the NMR results 689

the relatively short T2 values of the H1rsquo and OH protons shows the constrained mobility of the 690

28

sugar headgroup despite a certain flexibility of the chain (long T2 component) This trend may 691

suggest restrictions in translational mobility probably due to the sugar-OH hydrogen bonding 692

(Figure 8c) meaning that lipids are not able to diffuse within the membrane thus keeping two 693

different configurations stable 694

695

Conclusions 696

Topologically-connected nanotube-vesicle lipidic systems are commonly observed for 697

self-assembled membranes composed of at least two amphiphiles and spontaneously driven by 698

anisotropic inclusions andor by external directional forces This work shows that strong 699

differences in membrane curvature can spontaneously occur in single-lipid systems without 700

external anisotropic forces Coexistence and interconnection between nanotubes and vesicles 701

are evidenced via cryo-TEM and fluorescence confocal microscopy in a single-phase region at 702

temperature above a transition temperature Tm1= 483degC determined by DSC on the hydrated 703

sample but below the second transition temperature Tm2 (957degC) determined by DSC on the 704

dry powder X-ray diffraction (WAXS) also shows the coexistence of two lattice periods one 705

attributed to an inter-membrane repeating distance and the other possibly related to the intra-706

nanotube glucolipid arrangement Labelling the systems with Laurdan a fluorophore of which 707

the emission strongly varies with the local membrane rigidity excludes the presence of rigid 708

micro-separated phases (eg Lβ) 709

1H and 13C ssNMR despite the fact that it can only be performed on the dry powder 710

confirm the presence of one main fluid G-C181-OH environment above Tm2 Below Tm2 the 711

splitting of some 13C peaks and the coexistence of short and long 1H relaxation components 712

demonstrate that G-C181-OH can simultaneously pack into structurally and dynamically 713

inequivalent environments Combining advanced 2D 1H-1H correlation experiments based on 714

spin diffusion or double-quanta excitation suggests that vesicles could contain a preferred head-715

head configuration of the bolaform lipid while nanotubes a preferred head-tail assembly 716

The presence of two environments composed of the same molecule characterized by 717

different molecular configurations and rigidity fulfills the theoretical conditions necessary to 718

spontaneously nucleate a membrane inclusion of non-zero average orientation described as 719

necessary to observe topologically-connected nanotube-vesicle structures in the absence of 720

external forces This hypothesis could also be enriched by the possibility of nanotubes to be 721

pulled away between two vesicles after a collision event 722

This work expands the actual theoretical and experimental framework on lipid networks 723

stabilized by a different radius of curvature It could also motivate the synthesis or isolation 724

29

of new lipids providing a higher level of 3D architectural control in soft networks possibly in 725

the range of room temperature 726

727

Acknowledgements 728

Dr Andrea Lassenberger and Dr Sylvain Preacutevost at Institut Laue Langevin (ILL Grenoble 729

France) are kindly acknowledged for their assistance on the SANS experiment Confocal 730

images were performed at the Institut de Biologie Paris Seine (IBPS) imaging facility Jean-731

Franccedilois Gilles (Imaging Core Facility Institut de Biologie Paris Seine (IBPS) CNRS 732

Sorbonne Universiteacute Paris France) is kindly acknowledged for his assistance on the confocal 733

microscopy imaging Dr Guillaume Laurent (LCMCP Sorbonne Universiteacute Paris France) is 734

kindly acknowledged for helpful discussions The French region Ile-de-France SESAME 735

program is acknowledged for financial support (700 MHz NMR 736

spectrometer) Dr Olivier Diat (ICSM Marcoule France) Prof eronikaIglič and Prof V Aleš737

onsul discussi Slovenia) are kindly acknowledged for helpfUniversity of LjubljanaIglič (-Kralj 738

739

Financial support 740

WAXS experiments have been supported by Soleil Light Source Saint Aubin France proposal 741

Ndeg 20201747 SANS experiments have been supported by ILL proposal Ndeg9-13-778 742

743

References 744

1 D Kabaso N Bobrovska W Goacuteźdź E Gongadze V Kralj-Iglič R Zorec and A 745

Iglič Bioelectrochemistry 2012 87 204ndash210 746

2 C R Safinya U Raviv D J Needleman A Zidovska M C Choi M A Ojeda-747

Lopez K K Ewert Y Li H P Miller J Quispe B Carragher C S Potter M W 748

Kim S C Feinstein and L Wilson Adv Mater 2011 23 2260ndash2270 749

3 P Veranič M Lokar G J Schuumltz J Weghuber S Wieser H Haumlgerstrand V Kralj-750

Iglič and A Iglič Biophys J 2008 95 4416ndash4425 751

4 A Rustom R Saffrich I Markovic P Walther and H-H Gerdes Science (80- ) 752

2004 303 1007ndash1010 753

5 M I Angelova S Soleacuteau P Meacuteleacuteard F Faucon and P Bothorel Prog Colloid 754

Polym Sci 1992 89 127ndash131 755

6 L Mathivet S Cribier and P F Devaux Biophys J 1996 70 1112ndash1121 756

7 V Kralj-Iglič G Gomišček J Majhenc V Arrigler and S Svetina Colloids Surfaces 757

A Physicochem Eng Asp 2001 181 315ndash318 758

30

8 V Kralj-Iglič A Iglič G Gomišček F Sevšek V Arrigler and H Haumlgerstrand J 759

Phys A Math Gen 2002 35 1533ndash1549 760

9 M Drab D Stopar V Kralj-Iglič and A Iglič Cells 2019 8 626 761

10 A Zidovska K K Ewert J Quispe B Carragher C S Potter and C R Safinya 762

Langmuir 2009 25 2979ndash2985 763

11 A Zidovska K K Ewert J Quispe B Carragher C S Potter and C R Safinya Soft 764

Matter 2011 7 8363ndash8369 765

12 A Zidovska K K Ewert J Quispe B Carragher C S Potter and C R Safinya 766

Biochim Biophys Acta - Biomembr 2009 1788 1869ndash1876 767

13 M Karlsson K Sott M Davidson A S Cans P Linderholm D Chiu and O Orwar 768

Proc Natl Acad Sci U S A 2002 99 11573ndash11578 769

14 A Karlsson R Karlsson M Karlsson A-S Cans A Stromberg F Ryttsen and O 770

Orwar Nature 2001 409 150ndash152 771

15 M Karlsson K Sott A S Cans A Karlsson R Karlsson and O Orwar Langmuir 772

2001 17 6754ndash6758 773

16 E Evans H Bowman A Leung D Needham and D Tirrell Science (80- ) 1996 774

273 933ndash935 775

17 U Seifert Configurations of fluid membranes and vesicles 1997 vol 46 776

18 J Kaumls and E Sackmann Biophys J 1991 60 825ndash844 777

19 N Bobrovska W Goacuteźdź V Kralj-Iglič and A Iglič PLoS One 2013 8 e73941 778

20 A Iglič B Babnik K Bohinc M Fošnarič H Haumlgerstrand and V Kralj-Iglič J 779

Biomech 2007 40 579ndash585 780

21 K Bohinc T Slivnik A Iglič M Brumen and V Kralj-Iglič J Phys Chem C 2007 781

111 9709ndash9718 782

22 A Iglič H Haumlgerstrand P Veranič A Plemenitaš and V Kralj-Iglič J Theor Biol 783

2006 240 368ndash373 784

23 V Kralj-Iglič and P Veranič Adv Planar Lipid Bilayers Liposomes 2006 5 129ndash785

149 786

24 A Iglic and V Kralj-Iglic Adv Planar Lipid Bilayers Liposomes 2006 4 253ndash279 787

25 S Penič L Mesarec M Fošnarič L Mroacutewczyńska H Haumlgerstrand V Kralj-Iglič and 788

A Iglič Front Phys 2020 8 342 789

26 P Yager and P E Schoen Mol Cryst Liq Cryst 1984 106 371ndash381 790

27 T G Barclay K Constantopoulos and J Matisons Chem Rev 2014 114 10217ndash791

10291 792

31

28 B Božič G Gomišček V Kralj-Iglič S Svetina and B Žekš Eur Biophys J 2002 793

31 487ndash496 794

29 A Iglič H Haumlgerstrand M Bobrowska-Haumlgerstrand V Arrigler and V Kralj-Iglič 795

Phys Lett A 2003 310 493ndash497 796

30 R Bar-Ziv and E Moses Phys Rev Lett 1994 73 1392ndash1395 797

31 S Chiruvolu H E Warriner E Naranjo S H Idziak J O Raumldler R J Plano J A 798

Zasadzinski and C R Safinya Science (80- ) 1994 266 1222ndash1225 799

32 V Kralj-Iglič A Iglič H Haumlgerstrand and P Peterlin Phys Rev E 2000 61 4230ndash800

4234 801

33 L Mesarec M Drab S Penič V Kralj-Iglič and A Iglič Int J Mol Sci 2021 22 802

2348 803

34 L Van Renterghem S L K W Roelants N Baccile K Uyttersprot M C Taelman 804

B Everaert S Mincke S Ledegen S Debrouwer K Scholtens C V Stevens W 805

Soetaert L Van Renterghem S L K W Roelants N Baccile K Uytersprot M C 806

Taelman B Everaert S Mincke S Ledegen S Debrouwer K Scholtens C V 807

Stevens and W Soetaert Biotechnol Bioeng 2018 115 1195ndash1206 808

35 J Clark and F Deswarte Introduction to Chemicals from Biomass Second Edition 809

John Wiley amp Sons 2015 810

36 K V Sajna R Houmlfer R K Sukumaran L D Gottumukkala and A Pandey in 811

Industrial Biorefineries and White Biotechnology eds A Pandey R Houmlfer M 812

Taherzadeh K M Nampoothiri and C Larroche Amsterdam Oxford Waltham 813

Elsevier 2015 pp 499ndash521 814

37 R Marchant and I M Banat Trends Biotechnol 2012 30 558ndash565 815

38 K Hill and C LeHen-Ferrenbach in Sugar-Based Surfactants Fundamentals and 816

Applications ed C C Ruiz Boca Raton CRC Press 2009 pp 1ndash20 817

39 L Rodrigues I M Banat J Teixeira and R Oliveira J Antimicrob Chemother 818

2006 57 609ndash618 819

40 J D Desai and I M Banat Microbiol Mol Biol Rev 1997 61 47ndash64 820

41 S L K W Roelants K Ciesielska S L De Maeseneire H Moens B Everaert S 821

Verweire Q Denon B Vanlerberghe I N A Van Bogaert P Van der Meeren B 822

Devreese and W Soetaert Biotechnol Bioeng 2016 113 550ndash559 823

42 N Baccile M Selmane P Le Griel S Preacutevost J Perez C V Stevens E Delbeke S 824

Zibek M Guenther W Soetaert I N A Van Bogaert and S Roelants Langmuir 825

2016 32 6343ndash6359 826

32

43 N Baccile A-S Cuvier S Preacutevost C V Stevens E Delbeke J Berton W Soetaert 827

I N A Van Bogaert and S Roelants Langmuir 2016 32 10881ndash10894 828

44 N Baccile C Seyrig A Poirier S A Castro S L K W Roelants and S Abel Green 829

Chem 2021 23 3842ndash3944 830

45 A Salis L Cappai C Carucci D F Parsons and M Monduzzi J Phys Chem Lett 831

2020 11 6805ndash6811 832

46 J Schindelin I Arganda-Carreras E Frise V Kaynig M Longair T Pietzsch S 833

Preibisch C Rueden S Saalfeld B Schmid J-Y Tinevez D J White V 834

Hartenstein K Eliceiri P Tomancak and A Cardona Nat Methods 2012 9 676ndash835

682 836

47 T Parasassi and E Gratton J Fluoresc 1995 5 59ndash69 837

48 M M Dodes Traian F L Gonzaacutelez Flecha and V Levi J Lipid Res 2012 53 609ndash838

616 839

49 D M Owen C Rentero A Magenau A Abu-Siniyeh and K Gaus Nat Protoc 840

2012 7 24ndash35 841

50 W Stillwell in An Introduction to Biological Membranes Academic Press 2016 pp 842

135ndash180 843

51 W Stillwell in An Introduction to Biological Membranes Academic Press 2016 pp 844

181ndash220 845

52 B Tenchov Chem Phys Lipids 1991 57 165ndash177 846

53 R Koynova and M Caffrey Biochim Biophys Acta 1998 1376 91ndash145 847

54 F Schmid S Dolezel O Lenz and S Meinhardt J Phys Conf Ser 2014 487 848

012004 849

55 S Meinhardt R L C Vink and F Schmid Proc Natl Acad Sci U S A 2013 110 850

4476ndash4481 851

56 O Glatter and O Kratky Small Angle X-ray Scattering Academic Press London 852

1982 853

57 B A Cunningham A D Brown D H Wolfe W P Williams and A Brain Phys 854

Rev E 1998 58 3662ndash3672 855

58 M Nadler A Steiner T Dvir O Szekely P Szekely A Ginsburg R Asor R Resh 856

C Tamburu M Peres and U Raviv Soft Matter 2011 7 1512ndash1523 857

59 C Tanford The hydrophobic effect Formation of micelles and biological membranes 858

John Wiley amp Sons Inc New York 1973 859

60 M Masuda and T Shimizu Langmuir 2004 20 5969ndash5977 860

33

61 A Zidovska K K Ewert J Quispe B Carragher C S Potter and C R Safinya 861

Block Liposomes Vesicles of Charged Lipids with Distinctly Shaped Nanoscale 862

Sphere- Pear- Tube- or Rod-Segments Elsevier Inc 1st edn 2009 vol 465 863

62 A Iglic V Kralj-Iglic B Božič M Bobrowska-Haumlgerstrand B Isomaa and H 864

Haumlgerstrand Bioelectrochemistry 2000 52 203ndash211 865

63 T Shimizu M Masuda and H Minamikawa Chem Rev 2005 105 1401ndash1444 866

64 J V Selinger and J M Schnur Phys Rev Lett 1993 71 4091ndash4094 867

65 J M Schnur Science (80- ) 1993 262 1669ndash1676 868

66 B Tenchov R Koynova and G Rapp Biophys J 2001 80 1873ndash1890 869

67 R Koynova and B Tenchov Ser Biomech 2017 31 3ndash19 870

68 S J Angyal and R Le Fur Carbohydr Res 1984 126 15ndash26 871

69 J K Whitesell and M A Minton J Am Chem Soc 1987 109 225ndash228 872

70 T Takayama S Ando and I Ando J Mol Struct 1990 220 245ndash250 873

71 K Gawrisch N V Eldho and I V Polozov Chem Phys Lipids 2002 116 135ndash151 874

72 F Tielens C Gervais J F Lambert F Mauri and D Costa Chem Mater 2008 20 875

3336ndash3344 876

73 G A Jeffrey and Y Yeon Acta Crystallogr Sect B 1986 42 410ndash413 877

74 B Berglund and R W Vaughan J Chem Phys 1980 73 2037ndash2043 878

75 C Gervais C Coelho T Azaiumls J Maquet G Laurent F Pourpoint C Bonhomme 879

P Florian B Alonso G Guerrero P H Mutin and F Mauri J Magn Reson 2007 880

187 131ndash140 881

76 C Akita T Kawaguchi F Kaneko H Yamamoto and M Suzuki J Phys Chem B 882

2004 108 4862ndash4868 883

77 R N Purusottam L Seacutenicourt J J Lacapegravere and P Tekely Biochim Biophys Acta - 884

Biomembr 2015 1848 3134ndash3139 885

78 X Biarneacutes A Ardegravevol A Planas C Rovira A Laio and M Parrinello J Am Chem 886

Soc 2007 129 10686ndash10693 887

79 L M J Kroon‐Batenburg and J Kroon Biopolymers 1990 29 1243ndash1248 888

80 K N Kirschner and R J Woods Proc Natl Acad Sci U S A 2001 98 10541ndash889

10545 890

81 S Svenson B Kirste and J H Fuhrhop J Am Chem Soc 1994 116 11969ndash11975 891

82 I Sack S Macholl J H Fuhrhop and G Buntkowsky Phys Chem Chem Phys 892

2000 2 1781ndash1788 893

83 S Svenson J Koening and J H Fuhrhop J Phys Chem 1994 98 1022ndash1028 894

34

84 S Svenson A Schaefer and J H Fuhrhop J Chem Soc Perkin Trans 2 1994 2 895

1023ndash1028 896

85 K Bock and C Pedersen Adv Carbohydr Chem Biochem 1983 41 27ndash66 897

86 T B Grindley Structure and Conformation of Carbohydrates Springer Berlin 898

Heidelberg 2008 899

87 F Tantakitti J Boekhoven X Wang R V Kazantsev T Yu J Li E Zhuang R 900

Zandi J H Ortony C J Newcomb L C Palmer G S Shekhawat M O De La 901

Cruz G C Schatz and S I Stupp Nat Mater 2016 15 469ndash476 902

903

1

Supporting Information 904

905

Topological connection between vesicles and nanotubes in single-906

component lipid membranes driven by head-tail interactions 907

908

Niki Baccilea Ceacutedric Lorthioira Abdoul Aziz Baa Patrick Le Griela Cristina Coelhob Javier 909

Perezc Wim Soetaertd Sophie L K W Roelantsd 910

911

a Sorbonne Universiteacute Centre National de la Recherche Scientifique Laboratoire de Chimie de 912

la Matiegravere Condenseacutee de Paris LCMCP F-75005 Paris France 913

b Institut des Mateacuteriaux de Paris Centre Sorbonne Universiteacute Paris France 914

c Synchrotron Soleil LrsquoOrme des Merisiers Saint-Aubin BP48 91192 Gif-sur-Yvette Cedex 915

France 916

d InBio Department of Biotechnology Ghent University Ghent Belgium 917

918

919

Corresponding author 920

Dr Niki Baccile 921

E-mail address nikibaccilesorbonne-universitefr 922

Phone +33 1 44 27 56 77 923

924

925

926

2

a) 927

Parameter Method

Dry Matter (DM ) 996 Infrared balance 105degC

Glucose () 012 HPLC-Metacarb

Glycerol () nd HPLC-Metacarb

Free Fatty Acid content () lt001 Internal method BBEPP

GC with FID Detector

Oil () lt001 Internal method BBEPP

GC with FID Detector

Moulds (CFUg) lt10 3M-Nordval ndeg16 B

Yeasts (CFUg) lt10 3M-Nordval ndeg16 B

Anaerobic count (CFUg) lt10 SP-VG M005 B

Protein () tbd BCA protein assay

928

Figure S 1 - a) HPLC-ELSD chromatogram and composition table 929

930

931

932

933

3

b) 934

935

Group Ndeg δ ppm

CH2 1 354

CH2 2 153

CH2 3-712-14 132

CH2 811 203

CH 910 534

CH2 15 162

CH2 16 388

CH 1 424

CH 2 316

CH 3 334

CH 4 329

CH 5 354

CH2 6 366

Figure S 1 (next) ndash b) 1H NMR spectrum (MeOD-d4) recorded on the purified G-C181-OH compounds 936

and related assignment 937

938

2

811910 1rsquo

5rsquo1

3rsquo4rsquo

1rsquo1910

asymp asymp asympHDO MeOD Acetone

6rsquo2rsquo

1615

3-712-14

4

939

940

Figure S 2 ndash Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 130degC 941

5

942

943

Figure S 3 - Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 90degC 944

945

90degC

1

22

6

946

947

948

949

Figure S 4 ndash Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 70degC 950

951

21

11

1

2

5

3

4

45

5

6

7

7

7

952

953

954

Figure S 5 ndash Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 25degC 955

956

957

1

1 1

8

958

Figure S 6 ndash a) SANS experiment recorded at 90degC for a 5 mgmL G-C181-OH aqueous solution b-c) 959

WAXS experiments recorded on aqueous solutions of G-C181-OH at b) 5 mgmL and c) 25 mgmL d) 960

Structural interpretation of the repeating distances pointed out in c) 961

962

963

964

001 01 1

q Aring-1

25degC

68degC

88degC

111degC

01 1q Aring

-1

02 04 06 08 1q Aring

-1

001 01

10-1

100

101

Inte

nsity

cm

-1

q Aring-1

-2

T= 90degC

SANS WAXS WAXS

a) b) c)

dv

dL d= 144 Aring-1

di

ddH2O

dt

dvdL

dtdi d

d)

9

965

966

Figure S 7 ndash ssNMR MAS (10 kHz) experiments recorded on a G-C181-OH dry powder a) 1H spectra 967

recorded from 110degC to 70degC b) 13C single pulse (SP) spectra recorded with high power 1H decoupling from 968

110degC to 70degC c) 13C cross polarization (CP) spectra recorded at 110degC and contact time tc of 50 micros and 969

3000 micros d) 13C CP spectra recorded at 70degC and for 20 lt tc micros lt 3000 970

971

1

811

910

CH2 CH2

1rsquo

6rsquo

8 6 4 2 0 ppm

110

90

80

70

910

1rsquo

6rsquo16

1

2rsquo3rsquo4rsquo

811

OH

45 40 35 30 25 20 15

ppm

200 180 160 140 120 100 80 60 40 20 0

ppm140 120 100 80

ppm

CH2-gauche

3004

110

90

80

70

65

60

50

25

C91013003

C9 - 12869

CH2-all-trans

3198

C1rsquo10335

a) b)

1H

T degCT degC

13C Single Pulse MAS

A

B

A

B

70degC70degC

5rsquo

215

200 180 160 140 120 100 80 60 40 20 0

ppm

13C CPMAST= 110degC

200 180 160 140 120 100 80 60 40 20 0

ppm

132 131 130 129 128 127 126

ppm

106 105 104 103 102

ppm

13C CPMAST= 70degC

50

tc micros

3000

20

tc micros

50 500 1000 3000

tc micros

C102

13024 C910113003

C92

12869

C1rsquo213082

C1rsquo110443

A B

106 105 104 103 102

ppm

3000

1000

500

50

20

AB

c)

d)

10

972

Figure S 8 ndash a) Short and d) long T2 components displayed by the 1H transverse relaxation signal c) 1H 973

transverse relaxation signals recorded on a G-C181-OH dry powder from 110degC to 70degC The area of each 974

given resonance is plotted (log-lin) as a function of the echo time (CPMG pulse sequence) for each 975

temperature and fitted using the equation given above T2s and T2l indicate the short and long T2 relaxation 976

times respectively obtained by describing the 1H transverse relaxation signal using a biexponential decay 977

For H1rsquo H5rsquo and H6rsquo the 1H relaxation function could be described by a monoexponential decay 978

- Protons from the sugar group (H1rsquo H5rsquo H6rsquo) show a single T2 relaxation component (Figure 979

6c and Figure S 8) while protons of the acyl chain (CH2 H910 H811) show both short (Figure 980

6c) and long (Figure 6d) T2 relaxation component 981

- The short T2 relaxation time systematically decreases from 110degC to 70degC for all species by a 982

factor ranging from two to seven depending on the specific chemical group Very long T2 are 983

observed at 110degC for all acyl species while nearly unchanged values are measured below 984

90degC without any clear trend 985

- The sugar protons (H1rsquo H5rsquo H6rsquo) have the shortest set of T2 values (lt 1 ms) below 100degC 986

The long component of T2 generally ranges between 4 ms and 12 ms 987

0 1 2 3 4 5 6 7 8

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1 2 3 4 5 6 7 8

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

H1

H8H11

H9H10

CH2

H1rsquo

H6rsquo

CH2

H910H811 H1CH2

H1rsquo H5rsquo6rsquo

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

OH

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45 T= 110degC

T= 90degC

T= 80degC

T= 70degC

Log

In

ten

sity

Time ms

Fit

)

CH2 H910 H811 H1 H56 H1 OH

00

05

10

15

20

25

30

35

40

T2 m

s

T= 110degC

T= 90degC

T= 80degC

T= 70degC

CH2 H910 H811 H1

0

2

4

6

8

10

12

T2 m

s

T= 110degC

T= 90degC

T= 80degC

T= 70degC

Short T2(1H) Long T2(1H)

a) b)

c)

11

Table S 1 ndash 1H chemical shifts of G-C181-OH in solution and in the solid-state Solid-state spectra are 988

recorded at a MAS frequency of 10 kHz in DMSO-d6 broad peak 989

990

δH ppm

Group Ndeg Liquid

(MeOD-d4) Solid

(T= 110degC) Solid

(T= 70degC)

CH2 H1 354 357

CH2 H2 153 156

CH2 H3-712-14 132 135

CH2 H811 203 203

CH H910 534 534

CH2 H15 162 164

CH2 H16 388 383

CH H1 424 432

CH H2 316 335

CH H3 334 335

CH H4 329 335

CH H5 354 351 357

CH2 H6 366 383

OH OH1 432

45 463

OH OH2 492

OH OH3 490

OH OH4 487

OH OH6 445

991

992

H1H9H10H1rsquo

H6rsquo

OH2rsquo

OH1

12

Detailed discussion about the NOESY and DQ-SQ 2D 1H-1H ssNMR experiments presented in 993

Figure 9 994

995

2D 1H-1H ssNMR experiments are a unique tool to explore through-space proximities 996

between 1H nuclei ssNMR offers a large number of possibilities each adapted to a given 997

system The advantage of 2D maps with respect to a dynamics approach is their direct 998

readability in clear two nuclei which exchange their magnetization through-space (whichever 999

mechanism may be responsible for this exchange) will display a specific cross-peak However 1000

the way cross-peaks are presented and interpreted strictly depends on the nature of the pulse 1001

program which in turn is chosen according to the physical nature of the sample 1002

1D temperature-dependent 1H NMR experiments performed on G-C181-OH (Figure 1003

4a in the main text) indicate sharp peaks at 90degC and broad peaks at 70degC Narrow peaks 1004

indicate fast andor large-amplitude molecular tumbling averaging the strong 1H-1H dipolar 1005

interactions This excludes the use of pulse sequences like DQ-SQ which is not efficient under 1006

these conditions For this reason we employ spin-diffusion-based pulse sequences like 1007

NOESY which is more adapted for samples with high molecular mobility When mixing times 1008

are chosen to be short (lt 1 s) one generally explores through-space proximities within about 5 1009

10 Aring On the contrary when molecular motions are restricted dipolar coupling becomes 1010

significant 1H NMR peaks are broader and pulse sequences like DQ-SQ can now be used 1011

with a spatial resolution of less than 5 Aring 1012

Whenever possible DQ-SQ experiments are preferred over spin diffusion-based pulse 1013

sequences because they are more selective in particular for likewise interactions which are 1014

not discriminated in NOESY-type experiments This is illustrated by the typical interaction 1015

schemes for NOESY (top Figure 9a) and DQ-SQ (top Figure 9b) For two given A and B 1016

protons 2D NOESY maps will display two on-diagonal and two square-shaped symmetrical 1017

off-diagonal cross-peaks Off-diagonal cross-peaks illustrate the mutual A-B and B-A 1018

interaction but the on-diagonal peaks are systematic and do not prove any A-A or B-B 1019

interaction Similarly to NOESY 2D DQ-SQ maps display off-diagonal peaks (symmetric 1020

across the diagonal top Figure 9b) upon A-B and B-A interaction but in contrast they will 1021

display on-diagonal peaks only in the presence of self-interaction In the tutorial example of 1022

Figure 9b top the on-diagonal A-A peak indicates mutual A-A interaction while lack of on-1023

diagonal B-B peak evidences the absence of B-B proximities Combining NOESY and DQ-SQ 1024

experiments is capital to study the local arrangement of G-C181-OH and its evolution with 1025

temperature 1026

13

At T= 90degC the H1 and H5rsquo contributions are very close but possible to discriminate 1027

them (δ(H1)= 354 ppm δ(H5rsquo)= 351 ppm) according to 1H NMR chemical shifts in Table S 1028

1 2D 1H-1H NOESY in Figure 9a also discriminates between the two components attributed 1029

on the basis of the interactions with their mutual neighbors (green H1-H2 and H5rsquo-H1rsquo squares 1030

in Figure 9a) On this basis 2D NOESY does not show a direct H1rsquo-H1 correlation (blue red-1031

crossed) but rather an extensive network of intra-sugar interactions These features exclude 1032

suggest intra- and inter-sugar interactions and they exclude head-tail arrangements as a major 1033

molecular configuration at 90degC 1034

At 70degC the 1H NMR peaks are broader indicating a reduction of the T2 (1H) relaxation 1035

times andor residual homonuclear dipolar interactions excluding the use of NOESY In this 1036

case 1H-1H DQ-SQ correlation experiments are best suited Figure 9b presents the 1H-1H DQ-1037

SQ experiment on G-C181-OH at 70degC Peaks are broader than in NOESY and attribution may 1038

be cumbersome However the experiment is performed at 700 MHz and MAS= 22 kHz which 1039

improves spectral resolution with respect to the same experiment performed at 300 MHz and 1040

MAS= 10 kHz The auto-peak at δH= 356 ppm in the SQ dimension is attributed to H1-H1 spin 1041

pairs mainly resulting from intra CH2 interaction while the DQ cross-peaks at δH= 156-356 1042

ppm and δH= 432-356 ppm respectively origin from H2-H1 and H1rsquo-H1 interactions (Figure 1043

9b) DQ-SQ map does not show any auto-peak corresponding to H1rsquo-H1rsquo or Hxrsquo-Hxrsquo (x= 2rsquo 1044

3rsquo 4rsquo) interactions (red-crossed blue Figure 9b) thus excluding any proximity (lt 5 Aring) between 1045

close glucose neighbors On the contrary DQ-SQ maps (recorded both at 700 MHz Figure 9b 1046

or at 300 MHz not shown) systematically show DQ cross-peaks which can only be attributed 1047

to H1rsquo-H1 interactions (δH= 432-356) The presence of these H1rsquo-H1 interactions can only be 1048

explained by a head-tail arrangement of the G-C181-OH molecules as illustrated below the 1049

DQ-SQ map in Figure 9b 1050

13

317

Figure 5 ndash a-c) Confocal fluorescence microscopy images recorded at 90degC and 70degC for a 5 mgmL G-318

C181-OH aqueous solution d-g) Typical fluorescence microscopy images recorded at T= 80degC using 319

Laurdan (refer to materials and methods section for more details on this fluorophore) Excitation 320

wavelength is 405 nm for all the images while emission wavelength is 490 nm in (df) and 440 nm in (eg) 321

h) Full emission spectra obtained from the integration of various fluorescence images recorded from 412 322

nm to 600 nm (step 5 nm) at 90degC 80degC and 70degC i) Box plot of GP defined as 119918119927 =(120640120786120786120782minus120640120786120791120782)

120640120786120786120782+120640120786120791120782 with 120640120786120786120782 323

and 120640120786120791120782 being the emission wavelengths at 440 nm and 490 nm at 90degC 80degC and 70degC Each data point 324

l= 490 nm l= 440 nm

l= 490 nm l= 440 nm

10 μm400 450 500 550 600

00

02

04

06

08

10

No

rma

lize

d I

nte

nsity

l nm

90degC

80degC

80degC

70degC

70degC

400 450 500 550 60000

02

04

06

08

10

No

rma

lize

d I

nte

nsity

l nm

90degC

80degC

80degC

70degC

70degC

d) e)

f) g)

h)

i)

T= 80degC

Laurdan

T= 80degC

T= 80degC T= 80degC

T= 90degC T= 70degC T= 70degC

a) b) c)

1

1

1

14

corresponds to a set of regions of interest (ROIs) (54 at 90degC 65 at 80degC and 75 at 70degC) selected in the 325

corresponding fluorescence images ROIs were representatively selected so to include the entire image but 326

also vesicles alone tubes alone and mixtures of vesicles and tubes 327

328

Figure 5ab shows two typical confocal fluorescence microscopy images recorded on a 329

G-C181-OH solution (5 mgmL) in a flame-sealed capillary heat at T= 90degC and T= 70degC 330

Although the resolution of optical microscopy is not comparable with cryo-TEM the sample is 331

broadly composed of vesicles coexisting with nanotubes whereas in some cases the nanotube 332

starts from the vesicle itself (arrow Ndeg1) in good agreement with cryo-TEM data Furthermore 333

SANS recorded at T= 90degC (Figure S 6a) shows a scattering profile with a pronounced -2 334

dependence of the scattering intensity against the wavevector q in the log-log scale This 335

indicates the massive presence of a flat interface and it is expected typical of bilayers in 336

vesicular morphologies56 However the typical signature of nanotubes cannot be observed in 337

the SANS profile and this can be explained by both an argument related to the relative lower 338

amount of nanotubes with respect to vesicles at 90degC but also to the steeper slope of flat 339

morphologies (-2) compared to tubes (-1) the signal of the former masking the tube signal at 340

low wavevector values The SANS profile is characterized by a diffraction peak at q= 0167 Aring-341

1 of which the assignment is refined by WAXS data 342

Temperature-driven WAXS experiments in Figure S 6bc (5 mgmL-1 and 25 mgmL-1 343

respectively) show a lamellar order possibly attributed to a multilamellar structure of the 344

vesicles (dv Figure S 6d) at high temperature and to the interplanar arrangement of the 345

crystalline flat structures (dL Figure S 6d) at T= 25degC The sharp (001) reflection indicates 346

typical interlamellar distance of 353 Aring at 25degC although this value varies with temperature 347

(400 Aring at 111degC 372 Aring at 88degC) thus confirming its assignment to a hydration-dependent 348

interlamellar order More intriguing is the second broader peak at q= 0403 Aring-1 (d= 156 Aring) 349

and its reflection both clearly visible at 88degC below Tm2 We attribute it to the inter-glucolipid 350

arrangement inside the tubular phase (dt Figure S 6d) coexisting with the vesicle phase 351

between Tm1 and Tm2 and to the planar lipid arrangement (di Figure S 6d) in the rigid flat 352

lamellar structure below Tm2 Its temperature- hence hydration- independent position also 353

confirms its structural origin Finally besides the water-structure peak at about 2 Aring-1 we also 354

observe a broad peak at 144 Aring-1 characteristics of the lateral packing (d Figure S 6d) of lipids 355

in bilayers While its position could be consistent with an Lα phase5758 above Tm2 its broadness 356

and positional invariance as well as the coexistence between vesicles and tubes do not allow a 357

precise assignment 358

15

Does G-C181-OH assume a bilayer or interdigitated structure the latter being the one 359

found for other similar bolaform glucolipids4243 The size of G-C181-OH can be estimated to 360

be at the most 32 Aring given by the length L of the C16 tail (L = 22 Aring from the Tanford formula 361

L= 154 + 1265 times n (n= 16)59) and the sugar (about 10 Aring)4243 However the cis conformation 362

of the C=C bond imposes a ldquokinkrdquo of 60deg in the aliphatic chain which reduces the effective 363

length of the oleic backbone to about 17 Aring for an effective total G-C181-OH length of about 364

27 Aring Cryo-TEM provides a tube thickness varying from 35 Aring to 60 Aring and a vesicle membrane 365

thickness of at least 60 Aring WAXS suggests an interlamellar thickness (membrane and water) 366

between 40 Aring and 37 Aring and an inter-lipid distance in the tube of about 16 Aring In the absence of 367

a clear form factor oscillation profile as found for similar glucolipids4243 one must 368

acknowledge that these data pledge neither in favor of fully interdigitated nor a double layer 369

structure and might suggest a tilted lipid arrangement with partial interdigitation60 370

371

13C ssNMR demonstrates that G-C181-OH can assume two coexisting head-head and 372

head-tail configurations 373

Vesicles are characterized by low positive mean and gaussian curvatures while 374

nanotubes are described by a combination of a high positive mean and zero gaussian curvatures 375

In this regard the coexistence of vesicles and nanotubes requires glucolipids to adapt 376

themselves to both environments at the same time For this reason the formation of nanotube-377

vesicle networks generally requires a specific driving force able to impose a high mean and 378

zero gaussian curvature to the membrane In many systems insertion of at least a second 379

constituent (lipid surfactant protein) in the membrane becomes a necessary internal driving 380

force for the nucleation and growth of so-called membrane inclusions from which 381

topologically-connected nanotubes can form10ndash123261 as explained by orientational ordering 382

arguments62 It has also been shown that external driving forces such as use of laser 383

ldquotweezersrdquo30 osmotic pressure6 electroformation5ndash82829 or controlled suction13ndash16 are 384

otherwise necessary as also supported by theoretical studies19 385

The bending energy of tubes is higher than that of a vesicle membrane8 due to the 386

dependence of the membrane free energy from the principal curvatures For this reason 387

isotropic (rotational symmetry along the main axis) but also anisotropic33 lipids generally 388

stabilize low-curvature shapes like flat membranes or vesicles Lipid nanotubes are also a 389

common phase formed by amphiphiles but generally for lipids with a certain complexity in 390

their molecular structures as exampled by diacetylenic derivatives of phospholipids among 391

others262763ndash65 Negative contributions to the bending energy favoring the simultaneous 392

16

presence of interconnected vesicles and tubes was shown to be driven by the presence of 393

anisotropic lipids in a membrane continuum Spontaneous deviations to the elastic theory in the 394

presence of surfactants lipids and proteins with anisotropic structures favor the formation of 395

protrusions with radius of the order of the fraction of a micron9ndash1219ndash2362 396

In the absence of both internal and external driving forces mentioned above G-C181-397

OH should behave as a standard lipid Its DSC profile (Figure 2)536667 under both hydrated and 398

dry conditions identifies a hydration-independent transition (Tm1) slightly above room 399

temperature this is followed by one main transition at higher temperature (Tm2) in the dry 400

sample50 For such a profile one expects a gel phase (Lβ) below Tm1 and a liquid crystalline 401

structure possibly in the morphology of vesicles (L4 phase) possibly above Tm1 and most likely 402

above Tm231 as experimentally found On the basis of the thermogram recorded on the dry 403

powder one may also expect a ldquoripplerdquo (Pβ) phase between Tm1 and Tm2 characterized by 404

periodic undulations of the bilayer (generally addressed to as regions with both liquid order and 405

disorder)5455 406

In the present work instead of a Lβ phase we observe flat crystals below Tm1 and a 407

vesicle-nanotube network instead of the ldquoripplerdquo phase The latter is particularly unexpected 408

because the membrane is composed of a single-component (G-C181-OH) and no anisotropic 409

external force (osmotic hydrodynamic electric) is used to pull and stabilize the tubes In 410

addition the nanotubes are observed at fairly high temperatures above Tm1 (between about 411

60degC and 90degC) while this morphology is generally favored at low temperature (proportionality 412

with 1kT figure 5 in Ref 62) commonly below a given transition temperature characteristics 413

for each specific amphiphile27 Furthermore nanotubes of G-C181-OH seem to be stable in 414

time thus differing from what is found in electroformed vesicles78 They are also observed 415

away from phase transition events (may them be at Tm1 or Tm2) as otherwise reported before 416

for a diacetylenic derivative of phosphatidylcholine26 and many other systems27 417

A stable nanotube-vesicle network for a single-component phospholipid in the absence 418

of external driving forces is not expected and has never been reported before It could be 419

explained by the coexistence of two micro-separated phases of different rigidity This 420

hypothesis is excluded by confocal microscopy on Laurdan-labelled membranes and magic 421

angle spinning (MAS) solid-state NMR (ssNMR) experiments 422

Laurdan is a well-known fluorophore used to put in evidence the coexistence of rigid 423

and fluid domains in both model and biological membranes47ndash49 Laurdan was used here to label 424

the G-C181-OH membranes between Tm1 and Tm2 and to verify the possible micro-segregation 425

and coexistence of rigid (eg Lβ λem= 440 nm) and fluid (eg Lα λem= 490 nm) domains in 426

17

tubes and vesicles respectively A set of quantitative experiments performed on a 5 mgmL-1 427

solution summarized in Figure 5 shows an excess of fluorescence at only λem= 490 nm for 428

temperature between 90degC and 70degC (Figure 5h) The GP value which quantifies the rigidfluid 429

character of the membrane is systematically negative (Figure 5i) indicating an overall fluid 430

membrane at all temperatures GP is quantified on more than 50 regions of interest (ROIs) at 431

each temperature including vesicles and tubes alone as well as connected nanotube-vesicles 432

(Figure 5d-g) The dispersion of GP at 80degC and 70degC between -08 and -02 could suggest the 433

spurious presence of membranes with various degrees of flexibility although fluidity is always 434

predominant The combination of WAXS (broad peak d= 144 Aring-1 invariant with 435

temperature Figure S 6bc) and confocal fluorescence microscopy using Laurdan confirms the 436

single-phase of the membrane composed of G-C181-OH both in vesicles and nanotubes This 437

is also consistent with the lack of phase transitions in DSC experiments between ~85degC and 438

~60degC especially in hydrated samples We conclude that the vesicle-nanotube transition is not 439

driven by the local segregation of gel phase (Lβ) microdomains nor that the nanotubes are 440

composed of such phase 441

ssNMR is the best-suited approach to investigate the origin of the local driving force for 442

the nucleation of anisotropic inclusions of which the average orientation depends on the local 443

membrane and inclusion curvatures and was shown to promote tubing62 ssNMR can provide 444

information on the local molecular mobility eg through the nuclear spin relaxation behavior 445

but also on intermolecular interactions and stereochemistry eg through variations in the 446

chemical shift68ndash70 447

Ideally ssNMR should be performed on diluted G-C181-OH samples above Tm1 448

However fast MAS necessary for high resolution centrifuges the sample As a first 449

approximation hydrated G-C181-OH samples for instance at 50 wt as classically done for 450

phospholipids71 could be studied Unfortunately permeability of the sample holder combined 451

to heating (close to the boiling point of water) and MAS provoke dehydration and even sample 452

ejection After experiencing both of the above this approach was abandoned As a second 453

approximation the study of the dehydrated G-C181-OH powder avoids centrifugation and 454

evaporation If studying dried lipids may not reflect their exact behaviour under diluted 455

conditions valuable structural information can still be obtained in dehydrated molecular 456

systems like proteins thus illustrating that this approach is not unreasonable Furthermore if 457

DSC (Figure 2a) shows that hydration suppresses the Tm2 it also shows that the value of Tm1 is 458

not affected by water at concentrations as low as 10 wt For this reason performing a 459

structural analysis of dehydrated G-C181-OH by ssNMR above Tm1 may help understanding 460

18

the local arrangement of G-C181-OH in two morphologies vesicles and nanotubes having 461

sensibly different curvatures 462

To explore the local mobility of the most relevant chemical groups of G-C181-OH 463

above Tm2 (T= 110degC) and between Tm2 and Tm1 (70 lt T degC lt 90) we combine a series of 1H 464

and 13C experiments Single-pulse (SP) MAS experiments (1H and 13C) when recorded under 465

quantitative conditions allow the detection of all chemical species while cross-polarization 466

(CP) MAS (13C only) experiments are sensitive only to constrained groups for which the 1H-467

13C dipolar coupling is not averaged to zero by the molecular motions 468

469

Figure 6 ndash a) 1H MAS (10 kHz) ssNMR spectra recorded from 110degC to 70degC b) 13C MAS (10 kHz) ssNMR 470

spectra recorded using high power decoupling single pulse (SP) and cross-polarization (CP) pulse sequences 471

Contact time for CP experiments and corresponding temperatures are given on each spectrum Full spectra 472

are given in Figure S 7 473

The most relevant 1H and 13C spectra are given in Figure 6 (full range in Figure 7 and 474

Figure S 7) As expected the 1H solid-state SPMAS spectra show broader peaks compared to 475

the solution-state NMR spectrum (Figure 6a) reflecting the obvious solid-state character of the 476

sample at any temperature The sugar group and the C=C bond resonate in the 25-6 ppm range 477

and 535 ppm respectively while the OH groups of the sugar moiety are found below 45 ppm 478

(Figure 6a full assignment in Table S 1) Very broad (~1 ppm) at 100degC the position of the 479

OH peak varies with the temperature value and becomes narrower (lt~05 ppm) and visible at 480

δH= 464 ppm at T= 70degC Line narrowing and chemical shift variations of the OH resonance 481

in ssNMR is always a direct sign of H-bonding interactions whereas the strength of the H-482

bonding hence its OHhellipH distance is correlated to a dispersion in chemical shifts72ndash75 All other 483

60 55 50 45 40 35 30 25

ppm

110

90

80

70

910

1rsquo

6rsquo16

1

OH

134 132 130 128 126

ppm110 108 106 104 102 100

ppm

SP (T= 110degC)

SP (T= 70degC)

CP 50 ms (T= 70degC)

CP 3 ms (T= 70degC)

910 1rsquo

1H 13C

1

811

910

CH2 CH2

1rsquo

6rsquo

a) b)

C102

13024

C9101

13003

C92

12869 C1rsquo210382

C1rsquo110443

C1rsquo13035

2rsquo3rsquo4rsquo

T degC

5rsquo

19

resonances broaden from 110degC to 70degC thus indicating slower andor more anisotropic 484

reorientational motions of all chemical groups The specific dynamic behaviour of the sugar 485

headgroup is also reflected by its 1H transverse relaxation time T2 whereas the shorter the T2 486

the higher the rigidity Experiments performed between 110degC and 70degC shown and 487

commented in Figure S 8 in the Supporting Information globally show a single short (lt 1 ms 488

below 100degC) T2 relaxation component for H1rsquo H5rsquo and H6rsquo and two both short and long 489

(between 4 ms and 12 ms) T2 for the acyl chain (CH2 H910 H811) even below 100degC 490

At 110degC 13C SPMAS experiments (Figure 6b full range in Figure 7 and Figure S 7 491

assignments in Table 1) show a single peak at δC= 10335 ppm for C1rsquo and at δC= 13003 ppm 492

for C910 as expected for a symmetrical conformation with two chemically-equivalent 493

carbons76 At 70degC the C1rsquo peak disappears while C910 now shows two peaks around 128 494

ppm and 130 ppm (Table 1 full temperature series in Figure S 7b) attributed to C92 (δC= 12869 495

ppm) and C10 in a non-symmetric configuration respectively for which these two carbons are 496

chemically inequivalent76 More precisely the resonance at 130 ppm is actually the 497

superposition of two close resonances C9101 (δC= 13003 ppm) and C102 (δC= 13024 ppm) 498

where subscripts 1 and 2 stand for two different conformations of the cis-olefin group as 499

reported for various phases of oleic acid in the vicinity of 0degC76 500

501

Table 1 ndash 13C Chemical shifts of G-C181-OH in solution and solid-state Solid-state experiments are 502

recorded at MAS= 10 kHz All spectra are reported in Figure 7 and Figure S 7 The contact time for cross 503

polarization is 3 ms In all ssNMR experiments temperature is first raised at 110degC and then lowered to 504

either 90degC or 70degC Attribution of the 13C resonances in Table 1 is performed on the basis of the solution 505

spectrum (Figure S 1b) Additional peaks observed in both single pulse and cross polarization ssNMR 506

experiments are attributed when possible to a configuration effect of the given atom eg C9101 and C92 507

indicate two configurations of C9 All new peaks observed in solid-state and not detected in solution are 508

highlighted in red for convenience 509

510

511

δ ppm

Peak Liquid (DMSO-d6)

Solid Single Pulse

T= 110degC

Solid Single Pulse

T= 90degC

Solid Single Pulse

T= 70degC

Solid Cross Polarization

T= 70degC

C102

- - 13024 13024

C910C9101 13009 13003 13003 13003 13003

C1C9C10C1rsquo

C6rsquo

20

C92 - - - 12869 12869

C1rsquoC11 10329 10335 10335 - 10443

C12 - - - - 10382

C3 7724 7709 7702 - 7829

- 7634 7629 - 7621

7525

C2 7388 7398 7390 7370 7370

C4 7052 7077 7054 7041 7098

C16 6900 7016 - - 7029

- - - - 6793

C1C5C6 6153 6241 6236 6239 6262

C1C5C6 6117 - - - 5956

C2 3300 3290 3290 3283 3279

CH2(trans) - - - 3198 3196

CH2(gauche)C15 2941 2981 2987 3004 -

C811 2706 2747 2747 2832 2817

CH2 2597 2625 2628 2630 2817

512

13C CPMAS spectra at 70degC (Figure 6b full range in Figure 7 and Figure S 7d) at 513

increasing contact times confirm the presence of two cis-olefin environments one for which 514

C910 are equivalent labelled here C9101 and one for which they are inequivalent labelled 515

C92 C102 Both environments are rigid enough to be observed at short contact time (tc= 50 μs) 516

A similar double environment is also observed for the sugar C1rsquo CPMAS at 70degC shows two 517

peaks (C1rsquo1 at δC= 10443 ppm C1rsquo2 at δC= 10382 ppm Figure 6b Figure S 7d) These are 518

significantly shifted downfield from the C1rsquo peak observed at 110degC and are detected at contact 519

times as short as 50 μs thus indicating their rigid character One also finds that CPMAS is 520

totally inefficient at 100degC for short contact times (50 μs) and poorly efficient even at 3 ms 521

typical of fast-tumbling molecules (Figure S 7c) also both gauche (δC= 3004 ppm) and all-522

trans (δC= 3198 ppm) conformations of the CH2 carbons coexist at 70degC (Figure S 7b)77 523

According to ssNMR one main mobile molecular environment exists at 110degC as 524

shown by the well-resolved 1H spectrum single C=C (C9101) and anomeric (C1rsquo) resonances 525

gauche conformation of the aliphatic CH2 and single T2 of the aliphatic CH2 chain and sugar 526

protons The sugar group is certainly mobile (inefficient 1H13C CPMAS) but still more 527

constrainted than the acyl chain (shorter T2 (1H)) As the temperature is lowered the overall 528

rigidity increases (broadening of 1H NMR peaks faster 1H transverse relaxation and efficient 529

1H13C magnetization transfer during CPMAS) presumably driven by H-bonding 530

interactions (downfield shift and line narrowing of the OH 1H signal very short OH T2 loss of 531

the C1rsquo signal in SPMAS) At the same time a number of spectroscopic data (two concomitant 532

21

resonances for anomeric C1rsquo1 and C1rsquo2 cis-olefin C9101 and C92 C102 and CH2 gauche and 533

all-trans environments) and relaxation behaviour (fast and slow T2 relaxation components) 534

strongly suggests the coexistence of two molecular environments Further proofs of the 535

coexistence of two distinct molecular domains are given by a number of chemical shift 536

variations in the 13C CPMAS spectrum at 70degC compared to the one at 110degC 90degC and 537

solution This is shown in Figure 7 and in the list of 13C chemical shifts in Table 1 538

539

540

Figure 7 - Comparison between solution- and solid-state 13C NMR spectra of G-C181-OH Solid-state 541

experiments are recorded directly on the sample powder at a MAS of 10 kHz (4 mm zirconia rotor) SP 542

Single Pulse CP Cross Polarization In all the ssNMR experiments temperature is first raised at 110degC 543

and then lowered to either 90degC or 70degC Segmented lines are meant to highlight the known peak position 544

in solution thus stressing the strong chemical shift variations in the solid-state 545

546

NMR experiments above Tm2 suggest that vesicles could contain G-C181-OH with a 547

single family of mobile conformations characterized by chemically-equivalent C=C (red 548

Figure 8b) and a single C1rsquo conformation (green Figure 8b) Experiments between Tm1 and Tm2 549

otherwise suggest that two coexisting conformations characterized by chemically-equivalent 550

(red Figure 8b) and inequivalent (magenta Figure 8b) C=C could be associated to nanotubes 551

Two distinct C1rsquo sites (grey Figure 8b) also indicate two configurations of the sugar headgroup 552

A comment on this point will be given in the following paragraphs 553

The local (Aring-scale) variation of molecular mobility below Tm2 drives the coexistence of 554

two distinct molecular environments of the same G-C181-OH molecule This occurs within a 555

single-phase domain (as deduced by WAXS) of equivalent membrane rigidity (as derived from 556

Laurdan probe in confocal microscopy) Although never reported before the difference in the 557

local mobility of the same amphiphile (here G-C181-OH) could satisfy the theoretically-558

140 120 100 80 60 40 20

CP(3 ms)-70degC

SP-70degC

SP-90degC

SP-110degC

ppm

LiquidC910

C1rsquo

C2rsquo

C4rsquo

C3rsquo

C1

C6rsquoC5rsquo

C1

6

C8

11

C2 C

H2

C1

5

CH

2

DM

SO

13C

22

required conditions that induce local lipid segregation and formation of membrane inclusions 559

(Figure 8c) In case of their non-zero average orientation the local membrane curvature 560

increases62 and nanotubes can spontaneously form (Figure 8bc) More intriguing this 561

mechanism occurs at least 10degC above Tm1 although it should not be favored above the Tm27 562

From both experimental and theoretical backgrounds the conditions of topologically-563

connected vesicles-nanotubes systems are met only when either internal (two components) or 564

external driving forces are applied19 In the present single-lipid system of which the phase 565

behavior is summarized in Figure 8a the most plausible driving force could be explained by 566

the NMR results collected on the dry sample G-C181-OH concomitantly occupies two 567

molecular environments characterized by 1) two distinct conformations of the sugar 568

displaying equivalent mobility and 2) two conformations of the acyl chain unsaturation 569

570

23

571

Figure 8 ndash a) Thermotropic phase behavior of G-C181-OH at 5 mgmL-1 Note that Tm2 is observed on the 572

dry powder only b) Possible relationship between the molecular environment as derived from NMR and 573

membrane curvatures All the molecular drawings refer to the same G-C181-OH molecule (bulky ellipsoid 574

equiv sugar headgroup) while each color corresponds to a given configuration Grey of C1rsquo1 and C1rsquo2 indicates 575

the uncertainty about the assignment of the specific NMR peak The curvature C is defined as 1R R being 576

the curvature radius R1 R2 radius of curvature CG Gauss curvature CM Mean curvature c) Free 577

interpretation of the anisotropic high-curvature (CMgt 0 CG= 0) inclusions in the membrane (low CM and 578

CG) continuum driven by inter-sugar H-bonding and reduced intramembrane diffusion for a G-C181-OH 579

glucolipid 580

581

Flat crystalline structure

Nanotube-vesicle network

Vesicles

48degC

120deg

106 105 104 103 102 101

ppm

H9 H10H1rsquo

106 105 104 103 102 101

ppm106 105 104 103 102 101

ppm106 105 104 103 102 101

ppm

C=CC=C

C1rsquo1C1rsquo2

C1rsquo

Tm2(96degC)Tm1

a)

b)

c)

Same molecule different local orientations and sugar conformations

High curvature membrane

Low curvature inclusion

Flip-FlopStrong H-bonding

CG

CM ( + )

CG 0

CM 0 CG 0

CM 0

G-C181-OH

24

Two related questions are unanswered at the moment What actually drives nanotube 582

formation and is the organization of G-C181-OH within the vesicles and tubes the same Lipid 583

nanotubes have been studied for at least three decades over which broad experimental and 584

theoretical insight has been generated Despite this fact the actual understanding is still matter 585

of debate27 Micelle-to-nanotube and vesicle-to-nanotube phase transitions are generally 586

reported for single amphiphiles below a given Tm and intermediate phase composed of twisted 587

and flat fibers is concomitantly reported Initially thought to be a chirality-driven phenomenon 588

recent studies propose a more general mechanism involving crystallization and symmetry 589

breaking in the lipid packing driven by optimal packing27 If twisted ribbons were observed 590

before for a number of microbial glycolipid bolaamphiphiles44 they are not found for G-C181-591

OH This fact as well as the large body of cryo-TEM images and the observation of nanotubes 592

above Tm1 rather pledge in favour of a direct topological change from positive (vesicles) to 593

zero (nanotubes) Gaussian curvatures In this case vesicle-to-nanotube transitions could be 594

driven by external forces56307813ndash162829 595

Considering the absence of external forces in the present system there are only two 596

possibilities left spontaneous budding of the membrane or nanotube pulling after inter-vesicle 597

collisions due to Brownian motion Both are possible as inclusions composed of a different 598

arrangement of the same molecule in the membrane continuum can theoretically explain 599

spontaneous nanotube formation in an isotropic medium672223629ndash1218ndash21 However the second 600

one is probably more likely to prevail because it could explain the large number of nanotubes 601

connecting two distant vesicles Interestingly similar events occur for any phospholipid-based 602

vesicular solution but nanotubes are extremely rare unless at least two lipids of different 603

molecular structure are mixed together The spectroscopic 1H and 13C and relaxation (1H) 604

ssNMR data even if collected on a dry powder demonstrate that the coexistence of two G-605

C181-OH configurations involving both the sugar headgroup and fatty alcohol chain are 606

possible for this molecule As a matter of fact this is not so surprising as more than one energy 607

minima each corresponding to a given conformation are known for glucose its non-reducing 608

derivative methylglucose and carbohydrates in general78ndash80 609

The presence of two configurations could explain the coexistence between two 610

structures of different curvature but the data presented so far do not explain the origin of these 611

two configurations In other words what drives the change in the local structure of the sugar 612

headgroup and fatty alcohol within the membrane 613

The association of known self-assembled structures to spectroscopic 13C ssNMR data is 614

known for years in particular for gluconamides lipids having a linear sugar headgroup81ndash84 It 615

25

was found that gauche effects in some cases up to the γ position in the sugar chain induce 616

upfield chemical shifts while anti and trans conformations result in downfield shifts Each 617

structure (monolayer crystal bilayer crystal or micellar fiber) was then characterized by a given 618

organization of the gluconamide and a characteristic 13C ssNMR fingerprint which could be 619

explained by gauche γ-gauche anti and trans conformations of the headgroup Similar 620

structural-spectroscopic correlations have been collected over the years for much simpler cyclic 621

sugars85 whereas pyranoses can undergo important conformational changes (chair half-chair 622

boat skew) when the appropriate energy barriers are crossed78ndash8086 A quantitative 623

conformational study based on the 13C chemical shift variations shown in Table 1 would be 624

certainly helpful but very delicate in the context of the present state of the art because a 625

rigorous approach requires at least one known crystal structure8182 which is currently missing 626

for this compound A qualitative approach will then be adopted 627

G-C181-OH is a bolaform amphiphile and flip-flop could be a possible mechanism 628

involved in the nanotube stabilization whereas head-tail interactions have long been considered 629

as important in nanotube-forming bolaform amphiphiles6063 From mere considerations based 630

on T2 (1H) relaxation arguments one could exclude head-tail configurations if the R-CH2OH 631

endgroup (H1) was involved in a direct interaction with the sugar group its corresponding T2 632

would probably be comparable to the sugarrsquos one (OH and H1rsquo) but this does not seem to be 633

the case However this hypothesis would be based on the comparison of the relaxation from 634

CH and CH2 protons additionally involved in different kind of motions Besides the relaxation 635

of H1 is influenced by the strong overlap between H1 and H5rsquo For these reasons relaxation 636

arguments are not solid enough and will not be used 637

We then address this question by exploring intermolecular interactions in a direct 638

manner using 2D 1H-1H homonuclear ssNMR correlation spectroscopy At 90degC in the fast-639

tumbling domain of molecular motion (narrow 1H NMR peaks Figure 6a) and corresponding 640

to the vesicle-rich region spin diffusion (NOESY) experiments are preferred over dipolar 641

coupling based experiments because such interaction is averaged out by molecular motion 642

However when the molecular motion is more constrained in the domain of nanotube stability 643

at 70degC (broad 1H spectra Figure 6a) Double-Quantum Single-Quantum (DQ-SQ) correlation 644

experiments are best suited to probe interaction between close molecular neighbors These 645

experiments are reported in Figure 9 and the related interpretation is only briefly reported 646

below while a more detailed discussion can be found on Page 12 in the Supporting Information 647

26

648

Figure 9 ndash a) 2D 1H-1H NOESY NMR experiment performed at MAS= 10 kHz (4 mm probe) on a Bruker 649

300 MHz Avance III HD spectrometer using 300 ms of mixing time Sample temperature is regulated to 650

90degC reached by cooling from 110degC b) 2D 1H-1H DQ-SQ (BAck-to-BAck BABA pulse sequence used for 651

the DQ excitation scheme) NMR experiment performed at MAS= 22 kHz (25 mm probe) on a Bruker 700 652

MHz Avance III spectrometer using a DQ excitation time equal to 2 rotor periods Sample temperature is 653

regulated to 70degC reached by cooling from 110degC The typical schemes illustrating how to interpret these 654

2D experiments are reported above each experimental map Zooms of the sugar region and configurational 655

interpretation are given below each 2D map More information are given in the text below 656

H16-H16

H1-H1

H9-H10

H1rsquo-H1

H1rsquoH910 H16

H1

H1rsquo-H1rsquo

Hxrsquo-Hxrsquo

2

(pp

m)

(ppm)

315

811910 1rsquo

2rsquo2

3rsquo 4rsquo6rsquo

5rsquo

1rsquo2910

asymp asymp asympHDO MeOD Acetone

1

315

811 9101rsquo

2rsquo2

3rsquo4rsquo6rsquo

5rsquo

1rsquo2 9 10

asympasympasympHDOMeODAcetone

H1rsquo-H1

1rsquo

1

(p

pm

)

(ppm)

H1-H2

H1rsquo-H1

H1rsquo

H910

H1H5rsquo

H2

H1rsquo-H5rsquo

315

811910 1rsquo

2rsquo2

3rsquo 4rsquo6rsquo

5rsquo

1rsquo2910

asymp asymp asympHDO MeOD Acetone

1

H2rsquoH3rsquoH4rsquo

H2rsquoH3rsquoH4rsquo

X= 2rsquo3rsquo4rsquo

NOESY - T= 90degC

1H

1H

1H

1H

DQ-SQ - T= 70degC

a) b)

2

(pp

m)

(p

pm

)

(ppm)

A B

AhellipB

AhellipA

BhellipB

NOESY

(ppm)

A B

AhellipB

DQ-SQ

Single-quantum dimension

Do

ub

le-q

ua

ntu

m d

imen

sio

n

AhellipA

BhellipB

BhellipA

BhellipA

315

811910 1rsquo

2rsquo2

3rsquo 4rsquo6rsquo

5rsquo

1rsquo2910

asymp asymp asympHDO MeOD Acetone

1

lt 5 Aring

lt 5 Aring

H1-H2

H811

H15

H16

H6rsquo

H1rsquo

H1

H5rsquo

H2

H1-H2

H1rsquo-H1

H1rsquo-H5rsquo

H1rsquo

H1 H5rsquo

H1rsquo

H1

H5rsquo

H1H2rsquo H3rsquo H4rsquo

H1-H1

H1rsquo-H1

Hxrsquo-HxrsquoX= 2rsquo3rsquo4rsquo

H1-H2

H1rsquo

Zoom sugar region 28 ndash 45 ppmNOESY DQ-SQ

27

Head-tail or head-head configurations could be probed by the existence of H1rsquo-H1 657

interactions whereas H1 (δH= 356 ppm) is associated to the glucolipid tail and H1rsquo (δH= 432 658

ppm) to the sugar headgroup These atoms are more than 15 Aring apart and a direct interaction 659

could not be observed in both NOESY and DQ-SQ experiments NOESY experiments (Figure 660

9a) show that H1rsquo-H1 correlations are not observed Meanwhile clear-cut direct H1rsquo-H1 661

interactions are indeed observed are detected at T= 70degC (DQ cross-peaks related to H1rsquo-H1 662

correlation in DQ-SQ experiment shown in Figure 9b) Interestingly the absence of auto peaks 663

at δH= 432 ppm and δH= 335 ppm in the SQ dimension excludes H1rsquo-H1rsquo and Hxrsquo-Hxrsquo (x= 2 664

3 4) interactions respectively strengthening the idea that sugar headgroups from two nearest-665

neighbor G-C181-OH molecules are not adjacent in the nanotube-vesicle region 666

Homonuclear 2D ssNMR supports the idea according to which G-C181-OH can 667

undergo molecular flip-flops and a head-tail arrangement in the dry product Extrapolating these 668

conclusions in the lower concentration range lack of H1rsquo-H1 proximity at T= 90degC suggests a 669

head-head configuration in the vesicle-rich region and head-tail configuration within the 670

nanotubes It is however still unclear whether flip-flops are driven by spontaneous temperature-671

induced conformational changes in the glucolipid packing stimulating inclusions within the 672

membrane itself672223629ndash1218ndash21 or by defects generated by inter-vesicle collisions In fact a 673

combination of both is not unlikely Nanotubes are not stable when temperature is 20 to 30 674

degrees above Tm1 indicating that inter-vesicle collisions alone cannot explain the unusual 675

phenomenon In an energy landscape approach87 the head-head configuration is certainly 676

favored at high temperature while both head-head and head-tail configurations could be 677

favored in the mid-temperature region corresponding to the range between Tm1 and Tm2 in the 678

dry powder Both statistical and defect-induced flip-flop events contribute to form high mean 679

zero Gaussian curvature morphologies (nanotubes) for which the head-tail conformation 680

limiting the proximity between bulky sugar headgroups is probably more adapted (Figure 8c) 681

The flip-flop interpretation now reasonably explains the strong differences in the sugar and fatty 682

alcohol conformations deduced by the multiple NMR signals attributed to chemically-683

equivalent groups The existence of various low-energy conformations each having specific 684

NMR signature for sugars in solution78ndash8086 and solid-state81ndash84 further supports our 685

interpretation 686

One last point should be clarified Why between Tm1 and Tm2 are two molecular 687

environments able to coexist instead of a single one of average rigidity as found in the case of 688

ldquoripplerdquo phases50ndash54 We believe that this question can be answered based on the NMR results 689

the relatively short T2 values of the H1rsquo and OH protons shows the constrained mobility of the 690

28

sugar headgroup despite a certain flexibility of the chain (long T2 component) This trend may 691

suggest restrictions in translational mobility probably due to the sugar-OH hydrogen bonding 692

(Figure 8c) meaning that lipids are not able to diffuse within the membrane thus keeping two 693

different configurations stable 694

695

Conclusions 696

Topologically-connected nanotube-vesicle lipidic systems are commonly observed for 697

self-assembled membranes composed of at least two amphiphiles and spontaneously driven by 698

anisotropic inclusions andor by external directional forces This work shows that strong 699

differences in membrane curvature can spontaneously occur in single-lipid systems without 700

external anisotropic forces Coexistence and interconnection between nanotubes and vesicles 701

are evidenced via cryo-TEM and fluorescence confocal microscopy in a single-phase region at 702

temperature above a transition temperature Tm1= 483degC determined by DSC on the hydrated 703

sample but below the second transition temperature Tm2 (957degC) determined by DSC on the 704

dry powder X-ray diffraction (WAXS) also shows the coexistence of two lattice periods one 705

attributed to an inter-membrane repeating distance and the other possibly related to the intra-706

nanotube glucolipid arrangement Labelling the systems with Laurdan a fluorophore of which 707

the emission strongly varies with the local membrane rigidity excludes the presence of rigid 708

micro-separated phases (eg Lβ) 709

1H and 13C ssNMR despite the fact that it can only be performed on the dry powder 710

confirm the presence of one main fluid G-C181-OH environment above Tm2 Below Tm2 the 711

splitting of some 13C peaks and the coexistence of short and long 1H relaxation components 712

demonstrate that G-C181-OH can simultaneously pack into structurally and dynamically 713

inequivalent environments Combining advanced 2D 1H-1H correlation experiments based on 714

spin diffusion or double-quanta excitation suggests that vesicles could contain a preferred head-715

head configuration of the bolaform lipid while nanotubes a preferred head-tail assembly 716

The presence of two environments composed of the same molecule characterized by 717

different molecular configurations and rigidity fulfills the theoretical conditions necessary to 718

spontaneously nucleate a membrane inclusion of non-zero average orientation described as 719

necessary to observe topologically-connected nanotube-vesicle structures in the absence of 720

external forces This hypothesis could also be enriched by the possibility of nanotubes to be 721

pulled away between two vesicles after a collision event 722

This work expands the actual theoretical and experimental framework on lipid networks 723

stabilized by a different radius of curvature It could also motivate the synthesis or isolation 724

29

of new lipids providing a higher level of 3D architectural control in soft networks possibly in 725

the range of room temperature 726

727

Acknowledgements 728

Dr Andrea Lassenberger and Dr Sylvain Preacutevost at Institut Laue Langevin (ILL Grenoble 729

France) are kindly acknowledged for their assistance on the SANS experiment Confocal 730

images were performed at the Institut de Biologie Paris Seine (IBPS) imaging facility Jean-731

Franccedilois Gilles (Imaging Core Facility Institut de Biologie Paris Seine (IBPS) CNRS 732

Sorbonne Universiteacute Paris France) is kindly acknowledged for his assistance on the confocal 733

microscopy imaging Dr Guillaume Laurent (LCMCP Sorbonne Universiteacute Paris France) is 734

kindly acknowledged for helpful discussions The French region Ile-de-France SESAME 735

program is acknowledged for financial support (700 MHz NMR 736

spectrometer) Dr Olivier Diat (ICSM Marcoule France) Prof eronikaIglič and Prof V Aleš737

onsul discussi Slovenia) are kindly acknowledged for helpfUniversity of LjubljanaIglič (-Kralj 738

739

Financial support 740

WAXS experiments have been supported by Soleil Light Source Saint Aubin France proposal 741

Ndeg 20201747 SANS experiments have been supported by ILL proposal Ndeg9-13-778 742

743

References 744

1 D Kabaso N Bobrovska W Goacuteźdź E Gongadze V Kralj-Iglič R Zorec and A 745

Iglič Bioelectrochemistry 2012 87 204ndash210 746

2 C R Safinya U Raviv D J Needleman A Zidovska M C Choi M A Ojeda-747

Lopez K K Ewert Y Li H P Miller J Quispe B Carragher C S Potter M W 748

Kim S C Feinstein and L Wilson Adv Mater 2011 23 2260ndash2270 749

3 P Veranič M Lokar G J Schuumltz J Weghuber S Wieser H Haumlgerstrand V Kralj-750

Iglič and A Iglič Biophys J 2008 95 4416ndash4425 751

4 A Rustom R Saffrich I Markovic P Walther and H-H Gerdes Science (80- ) 752

2004 303 1007ndash1010 753

5 M I Angelova S Soleacuteau P Meacuteleacuteard F Faucon and P Bothorel Prog Colloid 754

Polym Sci 1992 89 127ndash131 755

6 L Mathivet S Cribier and P F Devaux Biophys J 1996 70 1112ndash1121 756

7 V Kralj-Iglič G Gomišček J Majhenc V Arrigler and S Svetina Colloids Surfaces 757

A Physicochem Eng Asp 2001 181 315ndash318 758

30

8 V Kralj-Iglič A Iglič G Gomišček F Sevšek V Arrigler and H Haumlgerstrand J 759

Phys A Math Gen 2002 35 1533ndash1549 760

9 M Drab D Stopar V Kralj-Iglič and A Iglič Cells 2019 8 626 761

10 A Zidovska K K Ewert J Quispe B Carragher C S Potter and C R Safinya 762

Langmuir 2009 25 2979ndash2985 763

11 A Zidovska K K Ewert J Quispe B Carragher C S Potter and C R Safinya Soft 764

Matter 2011 7 8363ndash8369 765

12 A Zidovska K K Ewert J Quispe B Carragher C S Potter and C R Safinya 766

Biochim Biophys Acta - Biomembr 2009 1788 1869ndash1876 767

13 M Karlsson K Sott M Davidson A S Cans P Linderholm D Chiu and O Orwar 768

Proc Natl Acad Sci U S A 2002 99 11573ndash11578 769

14 A Karlsson R Karlsson M Karlsson A-S Cans A Stromberg F Ryttsen and O 770

Orwar Nature 2001 409 150ndash152 771

15 M Karlsson K Sott A S Cans A Karlsson R Karlsson and O Orwar Langmuir 772

2001 17 6754ndash6758 773

16 E Evans H Bowman A Leung D Needham and D Tirrell Science (80- ) 1996 774

273 933ndash935 775

17 U Seifert Configurations of fluid membranes and vesicles 1997 vol 46 776

18 J Kaumls and E Sackmann Biophys J 1991 60 825ndash844 777

19 N Bobrovska W Goacuteźdź V Kralj-Iglič and A Iglič PLoS One 2013 8 e73941 778

20 A Iglič B Babnik K Bohinc M Fošnarič H Haumlgerstrand and V Kralj-Iglič J 779

Biomech 2007 40 579ndash585 780

21 K Bohinc T Slivnik A Iglič M Brumen and V Kralj-Iglič J Phys Chem C 2007 781

111 9709ndash9718 782

22 A Iglič H Haumlgerstrand P Veranič A Plemenitaš and V Kralj-Iglič J Theor Biol 783

2006 240 368ndash373 784

23 V Kralj-Iglič and P Veranič Adv Planar Lipid Bilayers Liposomes 2006 5 129ndash785

149 786

24 A Iglic and V Kralj-Iglic Adv Planar Lipid Bilayers Liposomes 2006 4 253ndash279 787

25 S Penič L Mesarec M Fošnarič L Mroacutewczyńska H Haumlgerstrand V Kralj-Iglič and 788

A Iglič Front Phys 2020 8 342 789

26 P Yager and P E Schoen Mol Cryst Liq Cryst 1984 106 371ndash381 790

27 T G Barclay K Constantopoulos and J Matisons Chem Rev 2014 114 10217ndash791

10291 792

31

28 B Božič G Gomišček V Kralj-Iglič S Svetina and B Žekš Eur Biophys J 2002 793

31 487ndash496 794

29 A Iglič H Haumlgerstrand M Bobrowska-Haumlgerstrand V Arrigler and V Kralj-Iglič 795

Phys Lett A 2003 310 493ndash497 796

30 R Bar-Ziv and E Moses Phys Rev Lett 1994 73 1392ndash1395 797

31 S Chiruvolu H E Warriner E Naranjo S H Idziak J O Raumldler R J Plano J A 798

Zasadzinski and C R Safinya Science (80- ) 1994 266 1222ndash1225 799

32 V Kralj-Iglič A Iglič H Haumlgerstrand and P Peterlin Phys Rev E 2000 61 4230ndash800

4234 801

33 L Mesarec M Drab S Penič V Kralj-Iglič and A Iglič Int J Mol Sci 2021 22 802

2348 803

34 L Van Renterghem S L K W Roelants N Baccile K Uyttersprot M C Taelman 804

B Everaert S Mincke S Ledegen S Debrouwer K Scholtens C V Stevens W 805

Soetaert L Van Renterghem S L K W Roelants N Baccile K Uytersprot M C 806

Taelman B Everaert S Mincke S Ledegen S Debrouwer K Scholtens C V 807

Stevens and W Soetaert Biotechnol Bioeng 2018 115 1195ndash1206 808

35 J Clark and F Deswarte Introduction to Chemicals from Biomass Second Edition 809

John Wiley amp Sons 2015 810

36 K V Sajna R Houmlfer R K Sukumaran L D Gottumukkala and A Pandey in 811

Industrial Biorefineries and White Biotechnology eds A Pandey R Houmlfer M 812

Taherzadeh K M Nampoothiri and C Larroche Amsterdam Oxford Waltham 813

Elsevier 2015 pp 499ndash521 814

37 R Marchant and I M Banat Trends Biotechnol 2012 30 558ndash565 815

38 K Hill and C LeHen-Ferrenbach in Sugar-Based Surfactants Fundamentals and 816

Applications ed C C Ruiz Boca Raton CRC Press 2009 pp 1ndash20 817

39 L Rodrigues I M Banat J Teixeira and R Oliveira J Antimicrob Chemother 818

2006 57 609ndash618 819

40 J D Desai and I M Banat Microbiol Mol Biol Rev 1997 61 47ndash64 820

41 S L K W Roelants K Ciesielska S L De Maeseneire H Moens B Everaert S 821

Verweire Q Denon B Vanlerberghe I N A Van Bogaert P Van der Meeren B 822

Devreese and W Soetaert Biotechnol Bioeng 2016 113 550ndash559 823

42 N Baccile M Selmane P Le Griel S Preacutevost J Perez C V Stevens E Delbeke S 824

Zibek M Guenther W Soetaert I N A Van Bogaert and S Roelants Langmuir 825

2016 32 6343ndash6359 826

32

43 N Baccile A-S Cuvier S Preacutevost C V Stevens E Delbeke J Berton W Soetaert 827

I N A Van Bogaert and S Roelants Langmuir 2016 32 10881ndash10894 828

44 N Baccile C Seyrig A Poirier S A Castro S L K W Roelants and S Abel Green 829

Chem 2021 23 3842ndash3944 830

45 A Salis L Cappai C Carucci D F Parsons and M Monduzzi J Phys Chem Lett 831

2020 11 6805ndash6811 832

46 J Schindelin I Arganda-Carreras E Frise V Kaynig M Longair T Pietzsch S 833

Preibisch C Rueden S Saalfeld B Schmid J-Y Tinevez D J White V 834

Hartenstein K Eliceiri P Tomancak and A Cardona Nat Methods 2012 9 676ndash835

682 836

47 T Parasassi and E Gratton J Fluoresc 1995 5 59ndash69 837

48 M M Dodes Traian F L Gonzaacutelez Flecha and V Levi J Lipid Res 2012 53 609ndash838

616 839

49 D M Owen C Rentero A Magenau A Abu-Siniyeh and K Gaus Nat Protoc 840

2012 7 24ndash35 841

50 W Stillwell in An Introduction to Biological Membranes Academic Press 2016 pp 842

135ndash180 843

51 W Stillwell in An Introduction to Biological Membranes Academic Press 2016 pp 844

181ndash220 845

52 B Tenchov Chem Phys Lipids 1991 57 165ndash177 846

53 R Koynova and M Caffrey Biochim Biophys Acta 1998 1376 91ndash145 847

54 F Schmid S Dolezel O Lenz and S Meinhardt J Phys Conf Ser 2014 487 848

012004 849

55 S Meinhardt R L C Vink and F Schmid Proc Natl Acad Sci U S A 2013 110 850

4476ndash4481 851

56 O Glatter and O Kratky Small Angle X-ray Scattering Academic Press London 852

1982 853

57 B A Cunningham A D Brown D H Wolfe W P Williams and A Brain Phys 854

Rev E 1998 58 3662ndash3672 855

58 M Nadler A Steiner T Dvir O Szekely P Szekely A Ginsburg R Asor R Resh 856

C Tamburu M Peres and U Raviv Soft Matter 2011 7 1512ndash1523 857

59 C Tanford The hydrophobic effect Formation of micelles and biological membranes 858

John Wiley amp Sons Inc New York 1973 859

60 M Masuda and T Shimizu Langmuir 2004 20 5969ndash5977 860

33

61 A Zidovska K K Ewert J Quispe B Carragher C S Potter and C R Safinya 861

Block Liposomes Vesicles of Charged Lipids with Distinctly Shaped Nanoscale 862

Sphere- Pear- Tube- or Rod-Segments Elsevier Inc 1st edn 2009 vol 465 863

62 A Iglic V Kralj-Iglic B Božič M Bobrowska-Haumlgerstrand B Isomaa and H 864

Haumlgerstrand Bioelectrochemistry 2000 52 203ndash211 865

63 T Shimizu M Masuda and H Minamikawa Chem Rev 2005 105 1401ndash1444 866

64 J V Selinger and J M Schnur Phys Rev Lett 1993 71 4091ndash4094 867

65 J M Schnur Science (80- ) 1993 262 1669ndash1676 868

66 B Tenchov R Koynova and G Rapp Biophys J 2001 80 1873ndash1890 869

67 R Koynova and B Tenchov Ser Biomech 2017 31 3ndash19 870

68 S J Angyal and R Le Fur Carbohydr Res 1984 126 15ndash26 871

69 J K Whitesell and M A Minton J Am Chem Soc 1987 109 225ndash228 872

70 T Takayama S Ando and I Ando J Mol Struct 1990 220 245ndash250 873

71 K Gawrisch N V Eldho and I V Polozov Chem Phys Lipids 2002 116 135ndash151 874

72 F Tielens C Gervais J F Lambert F Mauri and D Costa Chem Mater 2008 20 875

3336ndash3344 876

73 G A Jeffrey and Y Yeon Acta Crystallogr Sect B 1986 42 410ndash413 877

74 B Berglund and R W Vaughan J Chem Phys 1980 73 2037ndash2043 878

75 C Gervais C Coelho T Azaiumls J Maquet G Laurent F Pourpoint C Bonhomme 879

P Florian B Alonso G Guerrero P H Mutin and F Mauri J Magn Reson 2007 880

187 131ndash140 881

76 C Akita T Kawaguchi F Kaneko H Yamamoto and M Suzuki J Phys Chem B 882

2004 108 4862ndash4868 883

77 R N Purusottam L Seacutenicourt J J Lacapegravere and P Tekely Biochim Biophys Acta - 884

Biomembr 2015 1848 3134ndash3139 885

78 X Biarneacutes A Ardegravevol A Planas C Rovira A Laio and M Parrinello J Am Chem 886

Soc 2007 129 10686ndash10693 887

79 L M J Kroon‐Batenburg and J Kroon Biopolymers 1990 29 1243ndash1248 888

80 K N Kirschner and R J Woods Proc Natl Acad Sci U S A 2001 98 10541ndash889

10545 890

81 S Svenson B Kirste and J H Fuhrhop J Am Chem Soc 1994 116 11969ndash11975 891

82 I Sack S Macholl J H Fuhrhop and G Buntkowsky Phys Chem Chem Phys 892

2000 2 1781ndash1788 893

83 S Svenson J Koening and J H Fuhrhop J Phys Chem 1994 98 1022ndash1028 894

34

84 S Svenson A Schaefer and J H Fuhrhop J Chem Soc Perkin Trans 2 1994 2 895

1023ndash1028 896

85 K Bock and C Pedersen Adv Carbohydr Chem Biochem 1983 41 27ndash66 897

86 T B Grindley Structure and Conformation of Carbohydrates Springer Berlin 898

Heidelberg 2008 899

87 F Tantakitti J Boekhoven X Wang R V Kazantsev T Yu J Li E Zhuang R 900

Zandi J H Ortony C J Newcomb L C Palmer G S Shekhawat M O De La 901

Cruz G C Schatz and S I Stupp Nat Mater 2016 15 469ndash476 902

903

1

Supporting Information 904

905

Topological connection between vesicles and nanotubes in single-906

component lipid membranes driven by head-tail interactions 907

908

Niki Baccilea Ceacutedric Lorthioira Abdoul Aziz Baa Patrick Le Griela Cristina Coelhob Javier 909

Perezc Wim Soetaertd Sophie L K W Roelantsd 910

911

a Sorbonne Universiteacute Centre National de la Recherche Scientifique Laboratoire de Chimie de 912

la Matiegravere Condenseacutee de Paris LCMCP F-75005 Paris France 913

b Institut des Mateacuteriaux de Paris Centre Sorbonne Universiteacute Paris France 914

c Synchrotron Soleil LrsquoOrme des Merisiers Saint-Aubin BP48 91192 Gif-sur-Yvette Cedex 915

France 916

d InBio Department of Biotechnology Ghent University Ghent Belgium 917

918

919

Corresponding author 920

Dr Niki Baccile 921

E-mail address nikibaccilesorbonne-universitefr 922

Phone +33 1 44 27 56 77 923

924

925

926

2

a) 927

Parameter Method

Dry Matter (DM ) 996 Infrared balance 105degC

Glucose () 012 HPLC-Metacarb

Glycerol () nd HPLC-Metacarb

Free Fatty Acid content () lt001 Internal method BBEPP

GC with FID Detector

Oil () lt001 Internal method BBEPP

GC with FID Detector

Moulds (CFUg) lt10 3M-Nordval ndeg16 B

Yeasts (CFUg) lt10 3M-Nordval ndeg16 B

Anaerobic count (CFUg) lt10 SP-VG M005 B

Protein () tbd BCA protein assay

928

Figure S 1 - a) HPLC-ELSD chromatogram and composition table 929

930

931

932

933

3

b) 934

935

Group Ndeg δ ppm

CH2 1 354

CH2 2 153

CH2 3-712-14 132

CH2 811 203

CH 910 534

CH2 15 162

CH2 16 388

CH 1 424

CH 2 316

CH 3 334

CH 4 329

CH 5 354

CH2 6 366

Figure S 1 (next) ndash b) 1H NMR spectrum (MeOD-d4) recorded on the purified G-C181-OH compounds 936

and related assignment 937

938

2

811910 1rsquo

5rsquo1

3rsquo4rsquo

1rsquo1910

asymp asymp asympHDO MeOD Acetone

6rsquo2rsquo

1615

3-712-14

4

939

940

Figure S 2 ndash Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 130degC 941

5

942

943

Figure S 3 - Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 90degC 944

945

90degC

1

22

6

946

947

948

949

Figure S 4 ndash Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 70degC 950

951

21

11

1

2

5

3

4

45

5

6

7

7

7

952

953

954

Figure S 5 ndash Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 25degC 955

956

957

1

1 1

8

958

Figure S 6 ndash a) SANS experiment recorded at 90degC for a 5 mgmL G-C181-OH aqueous solution b-c) 959

WAXS experiments recorded on aqueous solutions of G-C181-OH at b) 5 mgmL and c) 25 mgmL d) 960

Structural interpretation of the repeating distances pointed out in c) 961

962

963

964

001 01 1

q Aring-1

25degC

68degC

88degC

111degC

01 1q Aring

-1

02 04 06 08 1q Aring

-1

001 01

10-1

100

101

Inte

nsity

cm

-1

q Aring-1

-2

T= 90degC

SANS WAXS WAXS

a) b) c)

dv

dL d= 144 Aring-1

di

ddH2O

dt

dvdL

dtdi d

d)

9

965

966

Figure S 7 ndash ssNMR MAS (10 kHz) experiments recorded on a G-C181-OH dry powder a) 1H spectra 967

recorded from 110degC to 70degC b) 13C single pulse (SP) spectra recorded with high power 1H decoupling from 968

110degC to 70degC c) 13C cross polarization (CP) spectra recorded at 110degC and contact time tc of 50 micros and 969

3000 micros d) 13C CP spectra recorded at 70degC and for 20 lt tc micros lt 3000 970

971

1

811

910

CH2 CH2

1rsquo

6rsquo

8 6 4 2 0 ppm

110

90

80

70

910

1rsquo

6rsquo16

1

2rsquo3rsquo4rsquo

811

OH

45 40 35 30 25 20 15

ppm

200 180 160 140 120 100 80 60 40 20 0

ppm140 120 100 80

ppm

CH2-gauche

3004

110

90

80

70

65

60

50

25

C91013003

C9 - 12869

CH2-all-trans

3198

C1rsquo10335

a) b)

1H

T degCT degC

13C Single Pulse MAS

A

B

A

B

70degC70degC

5rsquo

215

200 180 160 140 120 100 80 60 40 20 0

ppm

13C CPMAST= 110degC

200 180 160 140 120 100 80 60 40 20 0

ppm

132 131 130 129 128 127 126

ppm

106 105 104 103 102

ppm

13C CPMAST= 70degC

50

tc micros

3000

20

tc micros

50 500 1000 3000

tc micros

C102

13024 C910113003

C92

12869

C1rsquo213082

C1rsquo110443

A B

106 105 104 103 102

ppm

3000

1000

500

50

20

AB

c)

d)

10

972

Figure S 8 ndash a) Short and d) long T2 components displayed by the 1H transverse relaxation signal c) 1H 973

transverse relaxation signals recorded on a G-C181-OH dry powder from 110degC to 70degC The area of each 974

given resonance is plotted (log-lin) as a function of the echo time (CPMG pulse sequence) for each 975

temperature and fitted using the equation given above T2s and T2l indicate the short and long T2 relaxation 976

times respectively obtained by describing the 1H transverse relaxation signal using a biexponential decay 977

For H1rsquo H5rsquo and H6rsquo the 1H relaxation function could be described by a monoexponential decay 978

- Protons from the sugar group (H1rsquo H5rsquo H6rsquo) show a single T2 relaxation component (Figure 979

6c and Figure S 8) while protons of the acyl chain (CH2 H910 H811) show both short (Figure 980

6c) and long (Figure 6d) T2 relaxation component 981

- The short T2 relaxation time systematically decreases from 110degC to 70degC for all species by a 982

factor ranging from two to seven depending on the specific chemical group Very long T2 are 983

observed at 110degC for all acyl species while nearly unchanged values are measured below 984

90degC without any clear trend 985

- The sugar protons (H1rsquo H5rsquo H6rsquo) have the shortest set of T2 values (lt 1 ms) below 100degC 986

The long component of T2 generally ranges between 4 ms and 12 ms 987

0 1 2 3 4 5 6 7 8

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1 2 3 4 5 6 7 8

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

H1

H8H11

H9H10

CH2

H1rsquo

H6rsquo

CH2

H910H811 H1CH2

H1rsquo H5rsquo6rsquo

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

OH

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45 T= 110degC

T= 90degC

T= 80degC

T= 70degC

Log

In

ten

sity

Time ms

Fit

)

CH2 H910 H811 H1 H56 H1 OH

00

05

10

15

20

25

30

35

40

T2 m

s

T= 110degC

T= 90degC

T= 80degC

T= 70degC

CH2 H910 H811 H1

0

2

4

6

8

10

12

T2 m

s

T= 110degC

T= 90degC

T= 80degC

T= 70degC

Short T2(1H) Long T2(1H)

a) b)

c)

11

Table S 1 ndash 1H chemical shifts of G-C181-OH in solution and in the solid-state Solid-state spectra are 988

recorded at a MAS frequency of 10 kHz in DMSO-d6 broad peak 989

990

δH ppm

Group Ndeg Liquid

(MeOD-d4) Solid

(T= 110degC) Solid

(T= 70degC)

CH2 H1 354 357

CH2 H2 153 156

CH2 H3-712-14 132 135

CH2 H811 203 203

CH H910 534 534

CH2 H15 162 164

CH2 H16 388 383

CH H1 424 432

CH H2 316 335

CH H3 334 335

CH H4 329 335

CH H5 354 351 357

CH2 H6 366 383

OH OH1 432

45 463

OH OH2 492

OH OH3 490

OH OH4 487

OH OH6 445

991

992

H1H9H10H1rsquo

H6rsquo

OH2rsquo

OH1

12

Detailed discussion about the NOESY and DQ-SQ 2D 1H-1H ssNMR experiments presented in 993

Figure 9 994

995

2D 1H-1H ssNMR experiments are a unique tool to explore through-space proximities 996

between 1H nuclei ssNMR offers a large number of possibilities each adapted to a given 997

system The advantage of 2D maps with respect to a dynamics approach is their direct 998

readability in clear two nuclei which exchange their magnetization through-space (whichever 999

mechanism may be responsible for this exchange) will display a specific cross-peak However 1000

the way cross-peaks are presented and interpreted strictly depends on the nature of the pulse 1001

program which in turn is chosen according to the physical nature of the sample 1002

1D temperature-dependent 1H NMR experiments performed on G-C181-OH (Figure 1003

4a in the main text) indicate sharp peaks at 90degC and broad peaks at 70degC Narrow peaks 1004

indicate fast andor large-amplitude molecular tumbling averaging the strong 1H-1H dipolar 1005

interactions This excludes the use of pulse sequences like DQ-SQ which is not efficient under 1006

these conditions For this reason we employ spin-diffusion-based pulse sequences like 1007

NOESY which is more adapted for samples with high molecular mobility When mixing times 1008

are chosen to be short (lt 1 s) one generally explores through-space proximities within about 5 1009

10 Aring On the contrary when molecular motions are restricted dipolar coupling becomes 1010

significant 1H NMR peaks are broader and pulse sequences like DQ-SQ can now be used 1011

with a spatial resolution of less than 5 Aring 1012

Whenever possible DQ-SQ experiments are preferred over spin diffusion-based pulse 1013

sequences because they are more selective in particular for likewise interactions which are 1014

not discriminated in NOESY-type experiments This is illustrated by the typical interaction 1015

schemes for NOESY (top Figure 9a) and DQ-SQ (top Figure 9b) For two given A and B 1016

protons 2D NOESY maps will display two on-diagonal and two square-shaped symmetrical 1017

off-diagonal cross-peaks Off-diagonal cross-peaks illustrate the mutual A-B and B-A 1018

interaction but the on-diagonal peaks are systematic and do not prove any A-A or B-B 1019

interaction Similarly to NOESY 2D DQ-SQ maps display off-diagonal peaks (symmetric 1020

across the diagonal top Figure 9b) upon A-B and B-A interaction but in contrast they will 1021

display on-diagonal peaks only in the presence of self-interaction In the tutorial example of 1022

Figure 9b top the on-diagonal A-A peak indicates mutual A-A interaction while lack of on-1023

diagonal B-B peak evidences the absence of B-B proximities Combining NOESY and DQ-SQ 1024

experiments is capital to study the local arrangement of G-C181-OH and its evolution with 1025

temperature 1026

13

At T= 90degC the H1 and H5rsquo contributions are very close but possible to discriminate 1027

them (δ(H1)= 354 ppm δ(H5rsquo)= 351 ppm) according to 1H NMR chemical shifts in Table S 1028

1 2D 1H-1H NOESY in Figure 9a also discriminates between the two components attributed 1029

on the basis of the interactions with their mutual neighbors (green H1-H2 and H5rsquo-H1rsquo squares 1030

in Figure 9a) On this basis 2D NOESY does not show a direct H1rsquo-H1 correlation (blue red-1031

crossed) but rather an extensive network of intra-sugar interactions These features exclude 1032

suggest intra- and inter-sugar interactions and they exclude head-tail arrangements as a major 1033

molecular configuration at 90degC 1034

At 70degC the 1H NMR peaks are broader indicating a reduction of the T2 (1H) relaxation 1035

times andor residual homonuclear dipolar interactions excluding the use of NOESY In this 1036

case 1H-1H DQ-SQ correlation experiments are best suited Figure 9b presents the 1H-1H DQ-1037

SQ experiment on G-C181-OH at 70degC Peaks are broader than in NOESY and attribution may 1038

be cumbersome However the experiment is performed at 700 MHz and MAS= 22 kHz which 1039

improves spectral resolution with respect to the same experiment performed at 300 MHz and 1040

MAS= 10 kHz The auto-peak at δH= 356 ppm in the SQ dimension is attributed to H1-H1 spin 1041

pairs mainly resulting from intra CH2 interaction while the DQ cross-peaks at δH= 156-356 1042

ppm and δH= 432-356 ppm respectively origin from H2-H1 and H1rsquo-H1 interactions (Figure 1043

9b) DQ-SQ map does not show any auto-peak corresponding to H1rsquo-H1rsquo or Hxrsquo-Hxrsquo (x= 2rsquo 1044

3rsquo 4rsquo) interactions (red-crossed blue Figure 9b) thus excluding any proximity (lt 5 Aring) between 1045

close glucose neighbors On the contrary DQ-SQ maps (recorded both at 700 MHz Figure 9b 1046

or at 300 MHz not shown) systematically show DQ cross-peaks which can only be attributed 1047

to H1rsquo-H1 interactions (δH= 432-356) The presence of these H1rsquo-H1 interactions can only be 1048

explained by a head-tail arrangement of the G-C181-OH molecules as illustrated below the 1049

DQ-SQ map in Figure 9b 1050

14

corresponds to a set of regions of interest (ROIs) (54 at 90degC 65 at 80degC and 75 at 70degC) selected in the 325

corresponding fluorescence images ROIs were representatively selected so to include the entire image but 326

also vesicles alone tubes alone and mixtures of vesicles and tubes 327

328

Figure 5ab shows two typical confocal fluorescence microscopy images recorded on a 329

G-C181-OH solution (5 mgmL) in a flame-sealed capillary heat at T= 90degC and T= 70degC 330

Although the resolution of optical microscopy is not comparable with cryo-TEM the sample is 331

broadly composed of vesicles coexisting with nanotubes whereas in some cases the nanotube 332

starts from the vesicle itself (arrow Ndeg1) in good agreement with cryo-TEM data Furthermore 333

SANS recorded at T= 90degC (Figure S 6a) shows a scattering profile with a pronounced -2 334

dependence of the scattering intensity against the wavevector q in the log-log scale This 335

indicates the massive presence of a flat interface and it is expected typical of bilayers in 336

vesicular morphologies56 However the typical signature of nanotubes cannot be observed in 337

the SANS profile and this can be explained by both an argument related to the relative lower 338

amount of nanotubes with respect to vesicles at 90degC but also to the steeper slope of flat 339

morphologies (-2) compared to tubes (-1) the signal of the former masking the tube signal at 340

low wavevector values The SANS profile is characterized by a diffraction peak at q= 0167 Aring-341

1 of which the assignment is refined by WAXS data 342

Temperature-driven WAXS experiments in Figure S 6bc (5 mgmL-1 and 25 mgmL-1 343

respectively) show a lamellar order possibly attributed to a multilamellar structure of the 344

vesicles (dv Figure S 6d) at high temperature and to the interplanar arrangement of the 345

crystalline flat structures (dL Figure S 6d) at T= 25degC The sharp (001) reflection indicates 346

typical interlamellar distance of 353 Aring at 25degC although this value varies with temperature 347

(400 Aring at 111degC 372 Aring at 88degC) thus confirming its assignment to a hydration-dependent 348

interlamellar order More intriguing is the second broader peak at q= 0403 Aring-1 (d= 156 Aring) 349

and its reflection both clearly visible at 88degC below Tm2 We attribute it to the inter-glucolipid 350

arrangement inside the tubular phase (dt Figure S 6d) coexisting with the vesicle phase 351

between Tm1 and Tm2 and to the planar lipid arrangement (di Figure S 6d) in the rigid flat 352

lamellar structure below Tm2 Its temperature- hence hydration- independent position also 353

confirms its structural origin Finally besides the water-structure peak at about 2 Aring-1 we also 354

observe a broad peak at 144 Aring-1 characteristics of the lateral packing (d Figure S 6d) of lipids 355

in bilayers While its position could be consistent with an Lα phase5758 above Tm2 its broadness 356

and positional invariance as well as the coexistence between vesicles and tubes do not allow a 357

precise assignment 358

15

Does G-C181-OH assume a bilayer or interdigitated structure the latter being the one 359

found for other similar bolaform glucolipids4243 The size of G-C181-OH can be estimated to 360

be at the most 32 Aring given by the length L of the C16 tail (L = 22 Aring from the Tanford formula 361

L= 154 + 1265 times n (n= 16)59) and the sugar (about 10 Aring)4243 However the cis conformation 362

of the C=C bond imposes a ldquokinkrdquo of 60deg in the aliphatic chain which reduces the effective 363

length of the oleic backbone to about 17 Aring for an effective total G-C181-OH length of about 364

27 Aring Cryo-TEM provides a tube thickness varying from 35 Aring to 60 Aring and a vesicle membrane 365

thickness of at least 60 Aring WAXS suggests an interlamellar thickness (membrane and water) 366

between 40 Aring and 37 Aring and an inter-lipid distance in the tube of about 16 Aring In the absence of 367

a clear form factor oscillation profile as found for similar glucolipids4243 one must 368

acknowledge that these data pledge neither in favor of fully interdigitated nor a double layer 369

structure and might suggest a tilted lipid arrangement with partial interdigitation60 370

371

13C ssNMR demonstrates that G-C181-OH can assume two coexisting head-head and 372

head-tail configurations 373

Vesicles are characterized by low positive mean and gaussian curvatures while 374

nanotubes are described by a combination of a high positive mean and zero gaussian curvatures 375

In this regard the coexistence of vesicles and nanotubes requires glucolipids to adapt 376

themselves to both environments at the same time For this reason the formation of nanotube-377

vesicle networks generally requires a specific driving force able to impose a high mean and 378

zero gaussian curvature to the membrane In many systems insertion of at least a second 379

constituent (lipid surfactant protein) in the membrane becomes a necessary internal driving 380

force for the nucleation and growth of so-called membrane inclusions from which 381

topologically-connected nanotubes can form10ndash123261 as explained by orientational ordering 382

arguments62 It has also been shown that external driving forces such as use of laser 383

ldquotweezersrdquo30 osmotic pressure6 electroformation5ndash82829 or controlled suction13ndash16 are 384

otherwise necessary as also supported by theoretical studies19 385

The bending energy of tubes is higher than that of a vesicle membrane8 due to the 386

dependence of the membrane free energy from the principal curvatures For this reason 387

isotropic (rotational symmetry along the main axis) but also anisotropic33 lipids generally 388

stabilize low-curvature shapes like flat membranes or vesicles Lipid nanotubes are also a 389

common phase formed by amphiphiles but generally for lipids with a certain complexity in 390

their molecular structures as exampled by diacetylenic derivatives of phospholipids among 391

others262763ndash65 Negative contributions to the bending energy favoring the simultaneous 392

16

presence of interconnected vesicles and tubes was shown to be driven by the presence of 393

anisotropic lipids in a membrane continuum Spontaneous deviations to the elastic theory in the 394

presence of surfactants lipids and proteins with anisotropic structures favor the formation of 395

protrusions with radius of the order of the fraction of a micron9ndash1219ndash2362 396

In the absence of both internal and external driving forces mentioned above G-C181-397

OH should behave as a standard lipid Its DSC profile (Figure 2)536667 under both hydrated and 398

dry conditions identifies a hydration-independent transition (Tm1) slightly above room 399

temperature this is followed by one main transition at higher temperature (Tm2) in the dry 400

sample50 For such a profile one expects a gel phase (Lβ) below Tm1 and a liquid crystalline 401

structure possibly in the morphology of vesicles (L4 phase) possibly above Tm1 and most likely 402

above Tm231 as experimentally found On the basis of the thermogram recorded on the dry 403

powder one may also expect a ldquoripplerdquo (Pβ) phase between Tm1 and Tm2 characterized by 404

periodic undulations of the bilayer (generally addressed to as regions with both liquid order and 405

disorder)5455 406

In the present work instead of a Lβ phase we observe flat crystals below Tm1 and a 407

vesicle-nanotube network instead of the ldquoripplerdquo phase The latter is particularly unexpected 408

because the membrane is composed of a single-component (G-C181-OH) and no anisotropic 409

external force (osmotic hydrodynamic electric) is used to pull and stabilize the tubes In 410

addition the nanotubes are observed at fairly high temperatures above Tm1 (between about 411

60degC and 90degC) while this morphology is generally favored at low temperature (proportionality 412

with 1kT figure 5 in Ref 62) commonly below a given transition temperature characteristics 413

for each specific amphiphile27 Furthermore nanotubes of G-C181-OH seem to be stable in 414

time thus differing from what is found in electroformed vesicles78 They are also observed 415

away from phase transition events (may them be at Tm1 or Tm2) as otherwise reported before 416

for a diacetylenic derivative of phosphatidylcholine26 and many other systems27 417

A stable nanotube-vesicle network for a single-component phospholipid in the absence 418

of external driving forces is not expected and has never been reported before It could be 419

explained by the coexistence of two micro-separated phases of different rigidity This 420

hypothesis is excluded by confocal microscopy on Laurdan-labelled membranes and magic 421

angle spinning (MAS) solid-state NMR (ssNMR) experiments 422

Laurdan is a well-known fluorophore used to put in evidence the coexistence of rigid 423

and fluid domains in both model and biological membranes47ndash49 Laurdan was used here to label 424

the G-C181-OH membranes between Tm1 and Tm2 and to verify the possible micro-segregation 425

and coexistence of rigid (eg Lβ λem= 440 nm) and fluid (eg Lα λem= 490 nm) domains in 426

17

tubes and vesicles respectively A set of quantitative experiments performed on a 5 mgmL-1 427

solution summarized in Figure 5 shows an excess of fluorescence at only λem= 490 nm for 428

temperature between 90degC and 70degC (Figure 5h) The GP value which quantifies the rigidfluid 429

character of the membrane is systematically negative (Figure 5i) indicating an overall fluid 430

membrane at all temperatures GP is quantified on more than 50 regions of interest (ROIs) at 431

each temperature including vesicles and tubes alone as well as connected nanotube-vesicles 432

(Figure 5d-g) The dispersion of GP at 80degC and 70degC between -08 and -02 could suggest the 433

spurious presence of membranes with various degrees of flexibility although fluidity is always 434

predominant The combination of WAXS (broad peak d= 144 Aring-1 invariant with 435

temperature Figure S 6bc) and confocal fluorescence microscopy using Laurdan confirms the 436

single-phase of the membrane composed of G-C181-OH both in vesicles and nanotubes This 437

is also consistent with the lack of phase transitions in DSC experiments between ~85degC and 438

~60degC especially in hydrated samples We conclude that the vesicle-nanotube transition is not 439

driven by the local segregation of gel phase (Lβ) microdomains nor that the nanotubes are 440

composed of such phase 441

ssNMR is the best-suited approach to investigate the origin of the local driving force for 442

the nucleation of anisotropic inclusions of which the average orientation depends on the local 443

membrane and inclusion curvatures and was shown to promote tubing62 ssNMR can provide 444

information on the local molecular mobility eg through the nuclear spin relaxation behavior 445

but also on intermolecular interactions and stereochemistry eg through variations in the 446

chemical shift68ndash70 447

Ideally ssNMR should be performed on diluted G-C181-OH samples above Tm1 448

However fast MAS necessary for high resolution centrifuges the sample As a first 449

approximation hydrated G-C181-OH samples for instance at 50 wt as classically done for 450

phospholipids71 could be studied Unfortunately permeability of the sample holder combined 451

to heating (close to the boiling point of water) and MAS provoke dehydration and even sample 452

ejection After experiencing both of the above this approach was abandoned As a second 453

approximation the study of the dehydrated G-C181-OH powder avoids centrifugation and 454

evaporation If studying dried lipids may not reflect their exact behaviour under diluted 455

conditions valuable structural information can still be obtained in dehydrated molecular 456

systems like proteins thus illustrating that this approach is not unreasonable Furthermore if 457

DSC (Figure 2a) shows that hydration suppresses the Tm2 it also shows that the value of Tm1 is 458

not affected by water at concentrations as low as 10 wt For this reason performing a 459

structural analysis of dehydrated G-C181-OH by ssNMR above Tm1 may help understanding 460

18

the local arrangement of G-C181-OH in two morphologies vesicles and nanotubes having 461

sensibly different curvatures 462

To explore the local mobility of the most relevant chemical groups of G-C181-OH 463

above Tm2 (T= 110degC) and between Tm2 and Tm1 (70 lt T degC lt 90) we combine a series of 1H 464

and 13C experiments Single-pulse (SP) MAS experiments (1H and 13C) when recorded under 465

quantitative conditions allow the detection of all chemical species while cross-polarization 466

(CP) MAS (13C only) experiments are sensitive only to constrained groups for which the 1H-467

13C dipolar coupling is not averaged to zero by the molecular motions 468

469

Figure 6 ndash a) 1H MAS (10 kHz) ssNMR spectra recorded from 110degC to 70degC b) 13C MAS (10 kHz) ssNMR 470

spectra recorded using high power decoupling single pulse (SP) and cross-polarization (CP) pulse sequences 471

Contact time for CP experiments and corresponding temperatures are given on each spectrum Full spectra 472

are given in Figure S 7 473

The most relevant 1H and 13C spectra are given in Figure 6 (full range in Figure 7 and 474

Figure S 7) As expected the 1H solid-state SPMAS spectra show broader peaks compared to 475

the solution-state NMR spectrum (Figure 6a) reflecting the obvious solid-state character of the 476

sample at any temperature The sugar group and the C=C bond resonate in the 25-6 ppm range 477

and 535 ppm respectively while the OH groups of the sugar moiety are found below 45 ppm 478

(Figure 6a full assignment in Table S 1) Very broad (~1 ppm) at 100degC the position of the 479

OH peak varies with the temperature value and becomes narrower (lt~05 ppm) and visible at 480

δH= 464 ppm at T= 70degC Line narrowing and chemical shift variations of the OH resonance 481

in ssNMR is always a direct sign of H-bonding interactions whereas the strength of the H-482

bonding hence its OHhellipH distance is correlated to a dispersion in chemical shifts72ndash75 All other 483

60 55 50 45 40 35 30 25

ppm

110

90

80

70

910

1rsquo

6rsquo16

1

OH

134 132 130 128 126

ppm110 108 106 104 102 100

ppm

SP (T= 110degC)

SP (T= 70degC)

CP 50 ms (T= 70degC)

CP 3 ms (T= 70degC)

910 1rsquo

1H 13C

1

811

910

CH2 CH2

1rsquo

6rsquo

a) b)

C102

13024

C9101

13003

C92

12869 C1rsquo210382

C1rsquo110443

C1rsquo13035

2rsquo3rsquo4rsquo

T degC

5rsquo

19

resonances broaden from 110degC to 70degC thus indicating slower andor more anisotropic 484

reorientational motions of all chemical groups The specific dynamic behaviour of the sugar 485

headgroup is also reflected by its 1H transverse relaxation time T2 whereas the shorter the T2 486

the higher the rigidity Experiments performed between 110degC and 70degC shown and 487

commented in Figure S 8 in the Supporting Information globally show a single short (lt 1 ms 488

below 100degC) T2 relaxation component for H1rsquo H5rsquo and H6rsquo and two both short and long 489

(between 4 ms and 12 ms) T2 for the acyl chain (CH2 H910 H811) even below 100degC 490

At 110degC 13C SPMAS experiments (Figure 6b full range in Figure 7 and Figure S 7 491

assignments in Table 1) show a single peak at δC= 10335 ppm for C1rsquo and at δC= 13003 ppm 492

for C910 as expected for a symmetrical conformation with two chemically-equivalent 493

carbons76 At 70degC the C1rsquo peak disappears while C910 now shows two peaks around 128 494

ppm and 130 ppm (Table 1 full temperature series in Figure S 7b) attributed to C92 (δC= 12869 495

ppm) and C10 in a non-symmetric configuration respectively for which these two carbons are 496

chemically inequivalent76 More precisely the resonance at 130 ppm is actually the 497

superposition of two close resonances C9101 (δC= 13003 ppm) and C102 (δC= 13024 ppm) 498

where subscripts 1 and 2 stand for two different conformations of the cis-olefin group as 499

reported for various phases of oleic acid in the vicinity of 0degC76 500

501

Table 1 ndash 13C Chemical shifts of G-C181-OH in solution and solid-state Solid-state experiments are 502

recorded at MAS= 10 kHz All spectra are reported in Figure 7 and Figure S 7 The contact time for cross 503

polarization is 3 ms In all ssNMR experiments temperature is first raised at 110degC and then lowered to 504

either 90degC or 70degC Attribution of the 13C resonances in Table 1 is performed on the basis of the solution 505

spectrum (Figure S 1b) Additional peaks observed in both single pulse and cross polarization ssNMR 506

experiments are attributed when possible to a configuration effect of the given atom eg C9101 and C92 507

indicate two configurations of C9 All new peaks observed in solid-state and not detected in solution are 508

highlighted in red for convenience 509

510

511

δ ppm

Peak Liquid (DMSO-d6)

Solid Single Pulse

T= 110degC

Solid Single Pulse

T= 90degC

Solid Single Pulse

T= 70degC

Solid Cross Polarization

T= 70degC

C102

- - 13024 13024

C910C9101 13009 13003 13003 13003 13003

C1C9C10C1rsquo

C6rsquo

20

C92 - - - 12869 12869

C1rsquoC11 10329 10335 10335 - 10443

C12 - - - - 10382

C3 7724 7709 7702 - 7829

- 7634 7629 - 7621

7525

C2 7388 7398 7390 7370 7370

C4 7052 7077 7054 7041 7098

C16 6900 7016 - - 7029

- - - - 6793

C1C5C6 6153 6241 6236 6239 6262

C1C5C6 6117 - - - 5956

C2 3300 3290 3290 3283 3279

CH2(trans) - - - 3198 3196

CH2(gauche)C15 2941 2981 2987 3004 -

C811 2706 2747 2747 2832 2817

CH2 2597 2625 2628 2630 2817

512

13C CPMAS spectra at 70degC (Figure 6b full range in Figure 7 and Figure S 7d) at 513

increasing contact times confirm the presence of two cis-olefin environments one for which 514

C910 are equivalent labelled here C9101 and one for which they are inequivalent labelled 515

C92 C102 Both environments are rigid enough to be observed at short contact time (tc= 50 μs) 516

A similar double environment is also observed for the sugar C1rsquo CPMAS at 70degC shows two 517

peaks (C1rsquo1 at δC= 10443 ppm C1rsquo2 at δC= 10382 ppm Figure 6b Figure S 7d) These are 518

significantly shifted downfield from the C1rsquo peak observed at 110degC and are detected at contact 519

times as short as 50 μs thus indicating their rigid character One also finds that CPMAS is 520

totally inefficient at 100degC for short contact times (50 μs) and poorly efficient even at 3 ms 521

typical of fast-tumbling molecules (Figure S 7c) also both gauche (δC= 3004 ppm) and all-522

trans (δC= 3198 ppm) conformations of the CH2 carbons coexist at 70degC (Figure S 7b)77 523

According to ssNMR one main mobile molecular environment exists at 110degC as 524

shown by the well-resolved 1H spectrum single C=C (C9101) and anomeric (C1rsquo) resonances 525

gauche conformation of the aliphatic CH2 and single T2 of the aliphatic CH2 chain and sugar 526

protons The sugar group is certainly mobile (inefficient 1H13C CPMAS) but still more 527

constrainted than the acyl chain (shorter T2 (1H)) As the temperature is lowered the overall 528

rigidity increases (broadening of 1H NMR peaks faster 1H transverse relaxation and efficient 529

1H13C magnetization transfer during CPMAS) presumably driven by H-bonding 530

interactions (downfield shift and line narrowing of the OH 1H signal very short OH T2 loss of 531

the C1rsquo signal in SPMAS) At the same time a number of spectroscopic data (two concomitant 532

21

resonances for anomeric C1rsquo1 and C1rsquo2 cis-olefin C9101 and C92 C102 and CH2 gauche and 533

all-trans environments) and relaxation behaviour (fast and slow T2 relaxation components) 534

strongly suggests the coexistence of two molecular environments Further proofs of the 535

coexistence of two distinct molecular domains are given by a number of chemical shift 536

variations in the 13C CPMAS spectrum at 70degC compared to the one at 110degC 90degC and 537

solution This is shown in Figure 7 and in the list of 13C chemical shifts in Table 1 538

539

540

Figure 7 - Comparison between solution- and solid-state 13C NMR spectra of G-C181-OH Solid-state 541

experiments are recorded directly on the sample powder at a MAS of 10 kHz (4 mm zirconia rotor) SP 542

Single Pulse CP Cross Polarization In all the ssNMR experiments temperature is first raised at 110degC 543

and then lowered to either 90degC or 70degC Segmented lines are meant to highlight the known peak position 544

in solution thus stressing the strong chemical shift variations in the solid-state 545

546

NMR experiments above Tm2 suggest that vesicles could contain G-C181-OH with a 547

single family of mobile conformations characterized by chemically-equivalent C=C (red 548

Figure 8b) and a single C1rsquo conformation (green Figure 8b) Experiments between Tm1 and Tm2 549

otherwise suggest that two coexisting conformations characterized by chemically-equivalent 550

(red Figure 8b) and inequivalent (magenta Figure 8b) C=C could be associated to nanotubes 551

Two distinct C1rsquo sites (grey Figure 8b) also indicate two configurations of the sugar headgroup 552

A comment on this point will be given in the following paragraphs 553

The local (Aring-scale) variation of molecular mobility below Tm2 drives the coexistence of 554

two distinct molecular environments of the same G-C181-OH molecule This occurs within a 555

single-phase domain (as deduced by WAXS) of equivalent membrane rigidity (as derived from 556

Laurdan probe in confocal microscopy) Although never reported before the difference in the 557

local mobility of the same amphiphile (here G-C181-OH) could satisfy the theoretically-558

140 120 100 80 60 40 20

CP(3 ms)-70degC

SP-70degC

SP-90degC

SP-110degC

ppm

LiquidC910

C1rsquo

C2rsquo

C4rsquo

C3rsquo

C1

C6rsquoC5rsquo

C1

6

C8

11

C2 C

H2

C1

5

CH

2

DM

SO

13C

22

required conditions that induce local lipid segregation and formation of membrane inclusions 559

(Figure 8c) In case of their non-zero average orientation the local membrane curvature 560

increases62 and nanotubes can spontaneously form (Figure 8bc) More intriguing this 561

mechanism occurs at least 10degC above Tm1 although it should not be favored above the Tm27 562

From both experimental and theoretical backgrounds the conditions of topologically-563

connected vesicles-nanotubes systems are met only when either internal (two components) or 564

external driving forces are applied19 In the present single-lipid system of which the phase 565

behavior is summarized in Figure 8a the most plausible driving force could be explained by 566

the NMR results collected on the dry sample G-C181-OH concomitantly occupies two 567

molecular environments characterized by 1) two distinct conformations of the sugar 568

displaying equivalent mobility and 2) two conformations of the acyl chain unsaturation 569

570

23

571

Figure 8 ndash a) Thermotropic phase behavior of G-C181-OH at 5 mgmL-1 Note that Tm2 is observed on the 572

dry powder only b) Possible relationship between the molecular environment as derived from NMR and 573

membrane curvatures All the molecular drawings refer to the same G-C181-OH molecule (bulky ellipsoid 574

equiv sugar headgroup) while each color corresponds to a given configuration Grey of C1rsquo1 and C1rsquo2 indicates 575

the uncertainty about the assignment of the specific NMR peak The curvature C is defined as 1R R being 576

the curvature radius R1 R2 radius of curvature CG Gauss curvature CM Mean curvature c) Free 577

interpretation of the anisotropic high-curvature (CMgt 0 CG= 0) inclusions in the membrane (low CM and 578

CG) continuum driven by inter-sugar H-bonding and reduced intramembrane diffusion for a G-C181-OH 579

glucolipid 580

581

Flat crystalline structure

Nanotube-vesicle network

Vesicles

48degC

120deg

106 105 104 103 102 101

ppm

H9 H10H1rsquo

106 105 104 103 102 101

ppm106 105 104 103 102 101

ppm106 105 104 103 102 101

ppm

C=CC=C

C1rsquo1C1rsquo2

C1rsquo

Tm2(96degC)Tm1

a)

b)

c)

Same molecule different local orientations and sugar conformations

High curvature membrane

Low curvature inclusion

Flip-FlopStrong H-bonding

CG

CM ( + )

CG 0

CM 0 CG 0

CM 0

G-C181-OH

24

Two related questions are unanswered at the moment What actually drives nanotube 582

formation and is the organization of G-C181-OH within the vesicles and tubes the same Lipid 583

nanotubes have been studied for at least three decades over which broad experimental and 584

theoretical insight has been generated Despite this fact the actual understanding is still matter 585

of debate27 Micelle-to-nanotube and vesicle-to-nanotube phase transitions are generally 586

reported for single amphiphiles below a given Tm and intermediate phase composed of twisted 587

and flat fibers is concomitantly reported Initially thought to be a chirality-driven phenomenon 588

recent studies propose a more general mechanism involving crystallization and symmetry 589

breaking in the lipid packing driven by optimal packing27 If twisted ribbons were observed 590

before for a number of microbial glycolipid bolaamphiphiles44 they are not found for G-C181-591

OH This fact as well as the large body of cryo-TEM images and the observation of nanotubes 592

above Tm1 rather pledge in favour of a direct topological change from positive (vesicles) to 593

zero (nanotubes) Gaussian curvatures In this case vesicle-to-nanotube transitions could be 594

driven by external forces56307813ndash162829 595

Considering the absence of external forces in the present system there are only two 596

possibilities left spontaneous budding of the membrane or nanotube pulling after inter-vesicle 597

collisions due to Brownian motion Both are possible as inclusions composed of a different 598

arrangement of the same molecule in the membrane continuum can theoretically explain 599

spontaneous nanotube formation in an isotropic medium672223629ndash1218ndash21 However the second 600

one is probably more likely to prevail because it could explain the large number of nanotubes 601

connecting two distant vesicles Interestingly similar events occur for any phospholipid-based 602

vesicular solution but nanotubes are extremely rare unless at least two lipids of different 603

molecular structure are mixed together The spectroscopic 1H and 13C and relaxation (1H) 604

ssNMR data even if collected on a dry powder demonstrate that the coexistence of two G-605

C181-OH configurations involving both the sugar headgroup and fatty alcohol chain are 606

possible for this molecule As a matter of fact this is not so surprising as more than one energy 607

minima each corresponding to a given conformation are known for glucose its non-reducing 608

derivative methylglucose and carbohydrates in general78ndash80 609

The presence of two configurations could explain the coexistence between two 610

structures of different curvature but the data presented so far do not explain the origin of these 611

two configurations In other words what drives the change in the local structure of the sugar 612

headgroup and fatty alcohol within the membrane 613

The association of known self-assembled structures to spectroscopic 13C ssNMR data is 614

known for years in particular for gluconamides lipids having a linear sugar headgroup81ndash84 It 615

25

was found that gauche effects in some cases up to the γ position in the sugar chain induce 616

upfield chemical shifts while anti and trans conformations result in downfield shifts Each 617

structure (monolayer crystal bilayer crystal or micellar fiber) was then characterized by a given 618

organization of the gluconamide and a characteristic 13C ssNMR fingerprint which could be 619

explained by gauche γ-gauche anti and trans conformations of the headgroup Similar 620

structural-spectroscopic correlations have been collected over the years for much simpler cyclic 621

sugars85 whereas pyranoses can undergo important conformational changes (chair half-chair 622

boat skew) when the appropriate energy barriers are crossed78ndash8086 A quantitative 623

conformational study based on the 13C chemical shift variations shown in Table 1 would be 624

certainly helpful but very delicate in the context of the present state of the art because a 625

rigorous approach requires at least one known crystal structure8182 which is currently missing 626

for this compound A qualitative approach will then be adopted 627

G-C181-OH is a bolaform amphiphile and flip-flop could be a possible mechanism 628

involved in the nanotube stabilization whereas head-tail interactions have long been considered 629

as important in nanotube-forming bolaform amphiphiles6063 From mere considerations based 630

on T2 (1H) relaxation arguments one could exclude head-tail configurations if the R-CH2OH 631

endgroup (H1) was involved in a direct interaction with the sugar group its corresponding T2 632

would probably be comparable to the sugarrsquos one (OH and H1rsquo) but this does not seem to be 633

the case However this hypothesis would be based on the comparison of the relaxation from 634

CH and CH2 protons additionally involved in different kind of motions Besides the relaxation 635

of H1 is influenced by the strong overlap between H1 and H5rsquo For these reasons relaxation 636

arguments are not solid enough and will not be used 637

We then address this question by exploring intermolecular interactions in a direct 638

manner using 2D 1H-1H homonuclear ssNMR correlation spectroscopy At 90degC in the fast-639

tumbling domain of molecular motion (narrow 1H NMR peaks Figure 6a) and corresponding 640

to the vesicle-rich region spin diffusion (NOESY) experiments are preferred over dipolar 641

coupling based experiments because such interaction is averaged out by molecular motion 642

However when the molecular motion is more constrained in the domain of nanotube stability 643

at 70degC (broad 1H spectra Figure 6a) Double-Quantum Single-Quantum (DQ-SQ) correlation 644

experiments are best suited to probe interaction between close molecular neighbors These 645

experiments are reported in Figure 9 and the related interpretation is only briefly reported 646

below while a more detailed discussion can be found on Page 12 in the Supporting Information 647

26

648

Figure 9 ndash a) 2D 1H-1H NOESY NMR experiment performed at MAS= 10 kHz (4 mm probe) on a Bruker 649

300 MHz Avance III HD spectrometer using 300 ms of mixing time Sample temperature is regulated to 650

90degC reached by cooling from 110degC b) 2D 1H-1H DQ-SQ (BAck-to-BAck BABA pulse sequence used for 651

the DQ excitation scheme) NMR experiment performed at MAS= 22 kHz (25 mm probe) on a Bruker 700 652

MHz Avance III spectrometer using a DQ excitation time equal to 2 rotor periods Sample temperature is 653

regulated to 70degC reached by cooling from 110degC The typical schemes illustrating how to interpret these 654

2D experiments are reported above each experimental map Zooms of the sugar region and configurational 655

interpretation are given below each 2D map More information are given in the text below 656

H16-H16

H1-H1

H9-H10

H1rsquo-H1

H1rsquoH910 H16

H1

H1rsquo-H1rsquo

Hxrsquo-Hxrsquo

2

(pp

m)

(ppm)

315

811910 1rsquo

2rsquo2

3rsquo 4rsquo6rsquo

5rsquo

1rsquo2910

asymp asymp asympHDO MeOD Acetone

1

315

811 9101rsquo

2rsquo2

3rsquo4rsquo6rsquo

5rsquo

1rsquo2 9 10

asympasympasympHDOMeODAcetone

H1rsquo-H1

1rsquo

1

(p

pm

)

(ppm)

H1-H2

H1rsquo-H1

H1rsquo

H910

H1H5rsquo

H2

H1rsquo-H5rsquo

315

811910 1rsquo

2rsquo2

3rsquo 4rsquo6rsquo

5rsquo

1rsquo2910

asymp asymp asympHDO MeOD Acetone

1

H2rsquoH3rsquoH4rsquo

H2rsquoH3rsquoH4rsquo

X= 2rsquo3rsquo4rsquo

NOESY - T= 90degC

1H

1H

1H

1H

DQ-SQ - T= 70degC

a) b)

2

(pp

m)

(p

pm

)

(ppm)

A B

AhellipB

AhellipA

BhellipB

NOESY

(ppm)

A B

AhellipB

DQ-SQ

Single-quantum dimension

Do

ub

le-q

ua

ntu

m d

imen

sio

n

AhellipA

BhellipB

BhellipA

BhellipA

315

811910 1rsquo

2rsquo2

3rsquo 4rsquo6rsquo

5rsquo

1rsquo2910

asymp asymp asympHDO MeOD Acetone

1

lt 5 Aring

lt 5 Aring

H1-H2

H811

H15

H16

H6rsquo

H1rsquo

H1

H5rsquo

H2

H1-H2

H1rsquo-H1

H1rsquo-H5rsquo

H1rsquo

H1 H5rsquo

H1rsquo

H1

H5rsquo

H1H2rsquo H3rsquo H4rsquo

H1-H1

H1rsquo-H1

Hxrsquo-HxrsquoX= 2rsquo3rsquo4rsquo

H1-H2

H1rsquo

Zoom sugar region 28 ndash 45 ppmNOESY DQ-SQ

27

Head-tail or head-head configurations could be probed by the existence of H1rsquo-H1 657

interactions whereas H1 (δH= 356 ppm) is associated to the glucolipid tail and H1rsquo (δH= 432 658

ppm) to the sugar headgroup These atoms are more than 15 Aring apart and a direct interaction 659

could not be observed in both NOESY and DQ-SQ experiments NOESY experiments (Figure 660

9a) show that H1rsquo-H1 correlations are not observed Meanwhile clear-cut direct H1rsquo-H1 661

interactions are indeed observed are detected at T= 70degC (DQ cross-peaks related to H1rsquo-H1 662

correlation in DQ-SQ experiment shown in Figure 9b) Interestingly the absence of auto peaks 663

at δH= 432 ppm and δH= 335 ppm in the SQ dimension excludes H1rsquo-H1rsquo and Hxrsquo-Hxrsquo (x= 2 664

3 4) interactions respectively strengthening the idea that sugar headgroups from two nearest-665

neighbor G-C181-OH molecules are not adjacent in the nanotube-vesicle region 666

Homonuclear 2D ssNMR supports the idea according to which G-C181-OH can 667

undergo molecular flip-flops and a head-tail arrangement in the dry product Extrapolating these 668

conclusions in the lower concentration range lack of H1rsquo-H1 proximity at T= 90degC suggests a 669

head-head configuration in the vesicle-rich region and head-tail configuration within the 670

nanotubes It is however still unclear whether flip-flops are driven by spontaneous temperature-671

induced conformational changes in the glucolipid packing stimulating inclusions within the 672

membrane itself672223629ndash1218ndash21 or by defects generated by inter-vesicle collisions In fact a 673

combination of both is not unlikely Nanotubes are not stable when temperature is 20 to 30 674

degrees above Tm1 indicating that inter-vesicle collisions alone cannot explain the unusual 675

phenomenon In an energy landscape approach87 the head-head configuration is certainly 676

favored at high temperature while both head-head and head-tail configurations could be 677

favored in the mid-temperature region corresponding to the range between Tm1 and Tm2 in the 678

dry powder Both statistical and defect-induced flip-flop events contribute to form high mean 679

zero Gaussian curvature morphologies (nanotubes) for which the head-tail conformation 680

limiting the proximity between bulky sugar headgroups is probably more adapted (Figure 8c) 681

The flip-flop interpretation now reasonably explains the strong differences in the sugar and fatty 682

alcohol conformations deduced by the multiple NMR signals attributed to chemically-683

equivalent groups The existence of various low-energy conformations each having specific 684

NMR signature for sugars in solution78ndash8086 and solid-state81ndash84 further supports our 685

interpretation 686

One last point should be clarified Why between Tm1 and Tm2 are two molecular 687

environments able to coexist instead of a single one of average rigidity as found in the case of 688

ldquoripplerdquo phases50ndash54 We believe that this question can be answered based on the NMR results 689

the relatively short T2 values of the H1rsquo and OH protons shows the constrained mobility of the 690

28

sugar headgroup despite a certain flexibility of the chain (long T2 component) This trend may 691

suggest restrictions in translational mobility probably due to the sugar-OH hydrogen bonding 692

(Figure 8c) meaning that lipids are not able to diffuse within the membrane thus keeping two 693

different configurations stable 694

695

Conclusions 696

Topologically-connected nanotube-vesicle lipidic systems are commonly observed for 697

self-assembled membranes composed of at least two amphiphiles and spontaneously driven by 698

anisotropic inclusions andor by external directional forces This work shows that strong 699

differences in membrane curvature can spontaneously occur in single-lipid systems without 700

external anisotropic forces Coexistence and interconnection between nanotubes and vesicles 701

are evidenced via cryo-TEM and fluorescence confocal microscopy in a single-phase region at 702

temperature above a transition temperature Tm1= 483degC determined by DSC on the hydrated 703

sample but below the second transition temperature Tm2 (957degC) determined by DSC on the 704

dry powder X-ray diffraction (WAXS) also shows the coexistence of two lattice periods one 705

attributed to an inter-membrane repeating distance and the other possibly related to the intra-706

nanotube glucolipid arrangement Labelling the systems with Laurdan a fluorophore of which 707

the emission strongly varies with the local membrane rigidity excludes the presence of rigid 708

micro-separated phases (eg Lβ) 709

1H and 13C ssNMR despite the fact that it can only be performed on the dry powder 710

confirm the presence of one main fluid G-C181-OH environment above Tm2 Below Tm2 the 711

splitting of some 13C peaks and the coexistence of short and long 1H relaxation components 712

demonstrate that G-C181-OH can simultaneously pack into structurally and dynamically 713

inequivalent environments Combining advanced 2D 1H-1H correlation experiments based on 714

spin diffusion or double-quanta excitation suggests that vesicles could contain a preferred head-715

head configuration of the bolaform lipid while nanotubes a preferred head-tail assembly 716

The presence of two environments composed of the same molecule characterized by 717

different molecular configurations and rigidity fulfills the theoretical conditions necessary to 718

spontaneously nucleate a membrane inclusion of non-zero average orientation described as 719

necessary to observe topologically-connected nanotube-vesicle structures in the absence of 720

external forces This hypothesis could also be enriched by the possibility of nanotubes to be 721

pulled away between two vesicles after a collision event 722

This work expands the actual theoretical and experimental framework on lipid networks 723

stabilized by a different radius of curvature It could also motivate the synthesis or isolation 724

29

of new lipids providing a higher level of 3D architectural control in soft networks possibly in 725

the range of room temperature 726

727

Acknowledgements 728

Dr Andrea Lassenberger and Dr Sylvain Preacutevost at Institut Laue Langevin (ILL Grenoble 729

France) are kindly acknowledged for their assistance on the SANS experiment Confocal 730

images were performed at the Institut de Biologie Paris Seine (IBPS) imaging facility Jean-731

Franccedilois Gilles (Imaging Core Facility Institut de Biologie Paris Seine (IBPS) CNRS 732

Sorbonne Universiteacute Paris France) is kindly acknowledged for his assistance on the confocal 733

microscopy imaging Dr Guillaume Laurent (LCMCP Sorbonne Universiteacute Paris France) is 734

kindly acknowledged for helpful discussions The French region Ile-de-France SESAME 735

program is acknowledged for financial support (700 MHz NMR 736

spectrometer) Dr Olivier Diat (ICSM Marcoule France) Prof eronikaIglič and Prof V Aleš737

onsul discussi Slovenia) are kindly acknowledged for helpfUniversity of LjubljanaIglič (-Kralj 738

739

Financial support 740

WAXS experiments have been supported by Soleil Light Source Saint Aubin France proposal 741

Ndeg 20201747 SANS experiments have been supported by ILL proposal Ndeg9-13-778 742

743

References 744

1 D Kabaso N Bobrovska W Goacuteźdź E Gongadze V Kralj-Iglič R Zorec and A 745

Iglič Bioelectrochemistry 2012 87 204ndash210 746

2 C R Safinya U Raviv D J Needleman A Zidovska M C Choi M A Ojeda-747

Lopez K K Ewert Y Li H P Miller J Quispe B Carragher C S Potter M W 748

Kim S C Feinstein and L Wilson Adv Mater 2011 23 2260ndash2270 749

3 P Veranič M Lokar G J Schuumltz J Weghuber S Wieser H Haumlgerstrand V Kralj-750

Iglič and A Iglič Biophys J 2008 95 4416ndash4425 751

4 A Rustom R Saffrich I Markovic P Walther and H-H Gerdes Science (80- ) 752

2004 303 1007ndash1010 753

5 M I Angelova S Soleacuteau P Meacuteleacuteard F Faucon and P Bothorel Prog Colloid 754

Polym Sci 1992 89 127ndash131 755

6 L Mathivet S Cribier and P F Devaux Biophys J 1996 70 1112ndash1121 756

7 V Kralj-Iglič G Gomišček J Majhenc V Arrigler and S Svetina Colloids Surfaces 757

A Physicochem Eng Asp 2001 181 315ndash318 758

30

8 V Kralj-Iglič A Iglič G Gomišček F Sevšek V Arrigler and H Haumlgerstrand J 759

Phys A Math Gen 2002 35 1533ndash1549 760

9 M Drab D Stopar V Kralj-Iglič and A Iglič Cells 2019 8 626 761

10 A Zidovska K K Ewert J Quispe B Carragher C S Potter and C R Safinya 762

Langmuir 2009 25 2979ndash2985 763

11 A Zidovska K K Ewert J Quispe B Carragher C S Potter and C R Safinya Soft 764

Matter 2011 7 8363ndash8369 765

12 A Zidovska K K Ewert J Quispe B Carragher C S Potter and C R Safinya 766

Biochim Biophys Acta - Biomembr 2009 1788 1869ndash1876 767

13 M Karlsson K Sott M Davidson A S Cans P Linderholm D Chiu and O Orwar 768

Proc Natl Acad Sci U S A 2002 99 11573ndash11578 769

14 A Karlsson R Karlsson M Karlsson A-S Cans A Stromberg F Ryttsen and O 770

Orwar Nature 2001 409 150ndash152 771

15 M Karlsson K Sott A S Cans A Karlsson R Karlsson and O Orwar Langmuir 772

2001 17 6754ndash6758 773

16 E Evans H Bowman A Leung D Needham and D Tirrell Science (80- ) 1996 774

273 933ndash935 775

17 U Seifert Configurations of fluid membranes and vesicles 1997 vol 46 776

18 J Kaumls and E Sackmann Biophys J 1991 60 825ndash844 777

19 N Bobrovska W Goacuteźdź V Kralj-Iglič and A Iglič PLoS One 2013 8 e73941 778

20 A Iglič B Babnik K Bohinc M Fošnarič H Haumlgerstrand and V Kralj-Iglič J 779

Biomech 2007 40 579ndash585 780

21 K Bohinc T Slivnik A Iglič M Brumen and V Kralj-Iglič J Phys Chem C 2007 781

111 9709ndash9718 782

22 A Iglič H Haumlgerstrand P Veranič A Plemenitaš and V Kralj-Iglič J Theor Biol 783

2006 240 368ndash373 784

23 V Kralj-Iglič and P Veranič Adv Planar Lipid Bilayers Liposomes 2006 5 129ndash785

149 786

24 A Iglic and V Kralj-Iglic Adv Planar Lipid Bilayers Liposomes 2006 4 253ndash279 787

25 S Penič L Mesarec M Fošnarič L Mroacutewczyńska H Haumlgerstrand V Kralj-Iglič and 788

A Iglič Front Phys 2020 8 342 789

26 P Yager and P E Schoen Mol Cryst Liq Cryst 1984 106 371ndash381 790

27 T G Barclay K Constantopoulos and J Matisons Chem Rev 2014 114 10217ndash791

10291 792

31

28 B Božič G Gomišček V Kralj-Iglič S Svetina and B Žekš Eur Biophys J 2002 793

31 487ndash496 794

29 A Iglič H Haumlgerstrand M Bobrowska-Haumlgerstrand V Arrigler and V Kralj-Iglič 795

Phys Lett A 2003 310 493ndash497 796

30 R Bar-Ziv and E Moses Phys Rev Lett 1994 73 1392ndash1395 797

31 S Chiruvolu H E Warriner E Naranjo S H Idziak J O Raumldler R J Plano J A 798

Zasadzinski and C R Safinya Science (80- ) 1994 266 1222ndash1225 799

32 V Kralj-Iglič A Iglič H Haumlgerstrand and P Peterlin Phys Rev E 2000 61 4230ndash800

4234 801

33 L Mesarec M Drab S Penič V Kralj-Iglič and A Iglič Int J Mol Sci 2021 22 802

2348 803

34 L Van Renterghem S L K W Roelants N Baccile K Uyttersprot M C Taelman 804

B Everaert S Mincke S Ledegen S Debrouwer K Scholtens C V Stevens W 805

Soetaert L Van Renterghem S L K W Roelants N Baccile K Uytersprot M C 806

Taelman B Everaert S Mincke S Ledegen S Debrouwer K Scholtens C V 807

Stevens and W Soetaert Biotechnol Bioeng 2018 115 1195ndash1206 808

35 J Clark and F Deswarte Introduction to Chemicals from Biomass Second Edition 809

John Wiley amp Sons 2015 810

36 K V Sajna R Houmlfer R K Sukumaran L D Gottumukkala and A Pandey in 811

Industrial Biorefineries and White Biotechnology eds A Pandey R Houmlfer M 812

Taherzadeh K M Nampoothiri and C Larroche Amsterdam Oxford Waltham 813

Elsevier 2015 pp 499ndash521 814

37 R Marchant and I M Banat Trends Biotechnol 2012 30 558ndash565 815

38 K Hill and C LeHen-Ferrenbach in Sugar-Based Surfactants Fundamentals and 816

Applications ed C C Ruiz Boca Raton CRC Press 2009 pp 1ndash20 817

39 L Rodrigues I M Banat J Teixeira and R Oliveira J Antimicrob Chemother 818

2006 57 609ndash618 819

40 J D Desai and I M Banat Microbiol Mol Biol Rev 1997 61 47ndash64 820

41 S L K W Roelants K Ciesielska S L De Maeseneire H Moens B Everaert S 821

Verweire Q Denon B Vanlerberghe I N A Van Bogaert P Van der Meeren B 822

Devreese and W Soetaert Biotechnol Bioeng 2016 113 550ndash559 823

42 N Baccile M Selmane P Le Griel S Preacutevost J Perez C V Stevens E Delbeke S 824

Zibek M Guenther W Soetaert I N A Van Bogaert and S Roelants Langmuir 825

2016 32 6343ndash6359 826

32

43 N Baccile A-S Cuvier S Preacutevost C V Stevens E Delbeke J Berton W Soetaert 827

I N A Van Bogaert and S Roelants Langmuir 2016 32 10881ndash10894 828

44 N Baccile C Seyrig A Poirier S A Castro S L K W Roelants and S Abel Green 829

Chem 2021 23 3842ndash3944 830

45 A Salis L Cappai C Carucci D F Parsons and M Monduzzi J Phys Chem Lett 831

2020 11 6805ndash6811 832

46 J Schindelin I Arganda-Carreras E Frise V Kaynig M Longair T Pietzsch S 833

Preibisch C Rueden S Saalfeld B Schmid J-Y Tinevez D J White V 834

Hartenstein K Eliceiri P Tomancak and A Cardona Nat Methods 2012 9 676ndash835

682 836

47 T Parasassi and E Gratton J Fluoresc 1995 5 59ndash69 837

48 M M Dodes Traian F L Gonzaacutelez Flecha and V Levi J Lipid Res 2012 53 609ndash838

616 839

49 D M Owen C Rentero A Magenau A Abu-Siniyeh and K Gaus Nat Protoc 840

2012 7 24ndash35 841

50 W Stillwell in An Introduction to Biological Membranes Academic Press 2016 pp 842

135ndash180 843

51 W Stillwell in An Introduction to Biological Membranes Academic Press 2016 pp 844

181ndash220 845

52 B Tenchov Chem Phys Lipids 1991 57 165ndash177 846

53 R Koynova and M Caffrey Biochim Biophys Acta 1998 1376 91ndash145 847

54 F Schmid S Dolezel O Lenz and S Meinhardt J Phys Conf Ser 2014 487 848

012004 849

55 S Meinhardt R L C Vink and F Schmid Proc Natl Acad Sci U S A 2013 110 850

4476ndash4481 851

56 O Glatter and O Kratky Small Angle X-ray Scattering Academic Press London 852

1982 853

57 B A Cunningham A D Brown D H Wolfe W P Williams and A Brain Phys 854

Rev E 1998 58 3662ndash3672 855

58 M Nadler A Steiner T Dvir O Szekely P Szekely A Ginsburg R Asor R Resh 856

C Tamburu M Peres and U Raviv Soft Matter 2011 7 1512ndash1523 857

59 C Tanford The hydrophobic effect Formation of micelles and biological membranes 858

John Wiley amp Sons Inc New York 1973 859

60 M Masuda and T Shimizu Langmuir 2004 20 5969ndash5977 860

33

61 A Zidovska K K Ewert J Quispe B Carragher C S Potter and C R Safinya 861

Block Liposomes Vesicles of Charged Lipids with Distinctly Shaped Nanoscale 862

Sphere- Pear- Tube- or Rod-Segments Elsevier Inc 1st edn 2009 vol 465 863

62 A Iglic V Kralj-Iglic B Božič M Bobrowska-Haumlgerstrand B Isomaa and H 864

Haumlgerstrand Bioelectrochemistry 2000 52 203ndash211 865

63 T Shimizu M Masuda and H Minamikawa Chem Rev 2005 105 1401ndash1444 866

64 J V Selinger and J M Schnur Phys Rev Lett 1993 71 4091ndash4094 867

65 J M Schnur Science (80- ) 1993 262 1669ndash1676 868

66 B Tenchov R Koynova and G Rapp Biophys J 2001 80 1873ndash1890 869

67 R Koynova and B Tenchov Ser Biomech 2017 31 3ndash19 870

68 S J Angyal and R Le Fur Carbohydr Res 1984 126 15ndash26 871

69 J K Whitesell and M A Minton J Am Chem Soc 1987 109 225ndash228 872

70 T Takayama S Ando and I Ando J Mol Struct 1990 220 245ndash250 873

71 K Gawrisch N V Eldho and I V Polozov Chem Phys Lipids 2002 116 135ndash151 874

72 F Tielens C Gervais J F Lambert F Mauri and D Costa Chem Mater 2008 20 875

3336ndash3344 876

73 G A Jeffrey and Y Yeon Acta Crystallogr Sect B 1986 42 410ndash413 877

74 B Berglund and R W Vaughan J Chem Phys 1980 73 2037ndash2043 878

75 C Gervais C Coelho T Azaiumls J Maquet G Laurent F Pourpoint C Bonhomme 879

P Florian B Alonso G Guerrero P H Mutin and F Mauri J Magn Reson 2007 880

187 131ndash140 881

76 C Akita T Kawaguchi F Kaneko H Yamamoto and M Suzuki J Phys Chem B 882

2004 108 4862ndash4868 883

77 R N Purusottam L Seacutenicourt J J Lacapegravere and P Tekely Biochim Biophys Acta - 884

Biomembr 2015 1848 3134ndash3139 885

78 X Biarneacutes A Ardegravevol A Planas C Rovira A Laio and M Parrinello J Am Chem 886

Soc 2007 129 10686ndash10693 887

79 L M J Kroon‐Batenburg and J Kroon Biopolymers 1990 29 1243ndash1248 888

80 K N Kirschner and R J Woods Proc Natl Acad Sci U S A 2001 98 10541ndash889

10545 890

81 S Svenson B Kirste and J H Fuhrhop J Am Chem Soc 1994 116 11969ndash11975 891

82 I Sack S Macholl J H Fuhrhop and G Buntkowsky Phys Chem Chem Phys 892

2000 2 1781ndash1788 893

83 S Svenson J Koening and J H Fuhrhop J Phys Chem 1994 98 1022ndash1028 894

34

84 S Svenson A Schaefer and J H Fuhrhop J Chem Soc Perkin Trans 2 1994 2 895

1023ndash1028 896

85 K Bock and C Pedersen Adv Carbohydr Chem Biochem 1983 41 27ndash66 897

86 T B Grindley Structure and Conformation of Carbohydrates Springer Berlin 898

Heidelberg 2008 899

87 F Tantakitti J Boekhoven X Wang R V Kazantsev T Yu J Li E Zhuang R 900

Zandi J H Ortony C J Newcomb L C Palmer G S Shekhawat M O De La 901

Cruz G C Schatz and S I Stupp Nat Mater 2016 15 469ndash476 902

903

1

Supporting Information 904

905

Topological connection between vesicles and nanotubes in single-906

component lipid membranes driven by head-tail interactions 907

908

Niki Baccilea Ceacutedric Lorthioira Abdoul Aziz Baa Patrick Le Griela Cristina Coelhob Javier 909

Perezc Wim Soetaertd Sophie L K W Roelantsd 910

911

a Sorbonne Universiteacute Centre National de la Recherche Scientifique Laboratoire de Chimie de 912

la Matiegravere Condenseacutee de Paris LCMCP F-75005 Paris France 913

b Institut des Mateacuteriaux de Paris Centre Sorbonne Universiteacute Paris France 914

c Synchrotron Soleil LrsquoOrme des Merisiers Saint-Aubin BP48 91192 Gif-sur-Yvette Cedex 915

France 916

d InBio Department of Biotechnology Ghent University Ghent Belgium 917

918

919

Corresponding author 920

Dr Niki Baccile 921

E-mail address nikibaccilesorbonne-universitefr 922

Phone +33 1 44 27 56 77 923

924

925

926

2

a) 927

Parameter Method

Dry Matter (DM ) 996 Infrared balance 105degC

Glucose () 012 HPLC-Metacarb

Glycerol () nd HPLC-Metacarb

Free Fatty Acid content () lt001 Internal method BBEPP

GC with FID Detector

Oil () lt001 Internal method BBEPP

GC with FID Detector

Moulds (CFUg) lt10 3M-Nordval ndeg16 B

Yeasts (CFUg) lt10 3M-Nordval ndeg16 B

Anaerobic count (CFUg) lt10 SP-VG M005 B

Protein () tbd BCA protein assay

928

Figure S 1 - a) HPLC-ELSD chromatogram and composition table 929

930

931

932

933

3

b) 934

935

Group Ndeg δ ppm

CH2 1 354

CH2 2 153

CH2 3-712-14 132

CH2 811 203

CH 910 534

CH2 15 162

CH2 16 388

CH 1 424

CH 2 316

CH 3 334

CH 4 329

CH 5 354

CH2 6 366

Figure S 1 (next) ndash b) 1H NMR spectrum (MeOD-d4) recorded on the purified G-C181-OH compounds 936

and related assignment 937

938

2

811910 1rsquo

5rsquo1

3rsquo4rsquo

1rsquo1910

asymp asymp asympHDO MeOD Acetone

6rsquo2rsquo

1615

3-712-14

4

939

940

Figure S 2 ndash Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 130degC 941

5

942

943

Figure S 3 - Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 90degC 944

945

90degC

1

22

6

946

947

948

949

Figure S 4 ndash Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 70degC 950

951

21

11

1

2

5

3

4

45

5

6

7

7

7

952

953

954

Figure S 5 ndash Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 25degC 955

956

957

1

1 1

8

958

Figure S 6 ndash a) SANS experiment recorded at 90degC for a 5 mgmL G-C181-OH aqueous solution b-c) 959

WAXS experiments recorded on aqueous solutions of G-C181-OH at b) 5 mgmL and c) 25 mgmL d) 960

Structural interpretation of the repeating distances pointed out in c) 961

962

963

964

001 01 1

q Aring-1

25degC

68degC

88degC

111degC

01 1q Aring

-1

02 04 06 08 1q Aring

-1

001 01

10-1

100

101

Inte

nsity

cm

-1

q Aring-1

-2

T= 90degC

SANS WAXS WAXS

a) b) c)

dv

dL d= 144 Aring-1

di

ddH2O

dt

dvdL

dtdi d

d)

9

965

966

Figure S 7 ndash ssNMR MAS (10 kHz) experiments recorded on a G-C181-OH dry powder a) 1H spectra 967

recorded from 110degC to 70degC b) 13C single pulse (SP) spectra recorded with high power 1H decoupling from 968

110degC to 70degC c) 13C cross polarization (CP) spectra recorded at 110degC and contact time tc of 50 micros and 969

3000 micros d) 13C CP spectra recorded at 70degC and for 20 lt tc micros lt 3000 970

971

1

811

910

CH2 CH2

1rsquo

6rsquo

8 6 4 2 0 ppm

110

90

80

70

910

1rsquo

6rsquo16

1

2rsquo3rsquo4rsquo

811

OH

45 40 35 30 25 20 15

ppm

200 180 160 140 120 100 80 60 40 20 0

ppm140 120 100 80

ppm

CH2-gauche

3004

110

90

80

70

65

60

50

25

C91013003

C9 - 12869

CH2-all-trans

3198

C1rsquo10335

a) b)

1H

T degCT degC

13C Single Pulse MAS

A

B

A

B

70degC70degC

5rsquo

215

200 180 160 140 120 100 80 60 40 20 0

ppm

13C CPMAST= 110degC

200 180 160 140 120 100 80 60 40 20 0

ppm

132 131 130 129 128 127 126

ppm

106 105 104 103 102

ppm

13C CPMAST= 70degC

50

tc micros

3000

20

tc micros

50 500 1000 3000

tc micros

C102

13024 C910113003

C92

12869

C1rsquo213082

C1rsquo110443

A B

106 105 104 103 102

ppm

3000

1000

500

50

20

AB

c)

d)

10

972

Figure S 8 ndash a) Short and d) long T2 components displayed by the 1H transverse relaxation signal c) 1H 973

transverse relaxation signals recorded on a G-C181-OH dry powder from 110degC to 70degC The area of each 974

given resonance is plotted (log-lin) as a function of the echo time (CPMG pulse sequence) for each 975

temperature and fitted using the equation given above T2s and T2l indicate the short and long T2 relaxation 976

times respectively obtained by describing the 1H transverse relaxation signal using a biexponential decay 977

For H1rsquo H5rsquo and H6rsquo the 1H relaxation function could be described by a monoexponential decay 978

- Protons from the sugar group (H1rsquo H5rsquo H6rsquo) show a single T2 relaxation component (Figure 979

6c and Figure S 8) while protons of the acyl chain (CH2 H910 H811) show both short (Figure 980

6c) and long (Figure 6d) T2 relaxation component 981

- The short T2 relaxation time systematically decreases from 110degC to 70degC for all species by a 982

factor ranging from two to seven depending on the specific chemical group Very long T2 are 983

observed at 110degC for all acyl species while nearly unchanged values are measured below 984

90degC without any clear trend 985

- The sugar protons (H1rsquo H5rsquo H6rsquo) have the shortest set of T2 values (lt 1 ms) below 100degC 986

The long component of T2 generally ranges between 4 ms and 12 ms 987

0 1 2 3 4 5 6 7 8

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1 2 3 4 5 6 7 8

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

H1

H8H11

H9H10

CH2

H1rsquo

H6rsquo

CH2

H910H811 H1CH2

H1rsquo H5rsquo6rsquo

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

OH

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45 T= 110degC

T= 90degC

T= 80degC

T= 70degC

Log

In

ten

sity

Time ms

Fit

)

CH2 H910 H811 H1 H56 H1 OH

00

05

10

15

20

25

30

35

40

T2 m

s

T= 110degC

T= 90degC

T= 80degC

T= 70degC

CH2 H910 H811 H1

0

2

4

6

8

10

12

T2 m

s

T= 110degC

T= 90degC

T= 80degC

T= 70degC

Short T2(1H) Long T2(1H)

a) b)

c)

11

Table S 1 ndash 1H chemical shifts of G-C181-OH in solution and in the solid-state Solid-state spectra are 988

recorded at a MAS frequency of 10 kHz in DMSO-d6 broad peak 989

990

δH ppm

Group Ndeg Liquid

(MeOD-d4) Solid

(T= 110degC) Solid

(T= 70degC)

CH2 H1 354 357

CH2 H2 153 156

CH2 H3-712-14 132 135

CH2 H811 203 203

CH H910 534 534

CH2 H15 162 164

CH2 H16 388 383

CH H1 424 432

CH H2 316 335

CH H3 334 335

CH H4 329 335

CH H5 354 351 357

CH2 H6 366 383

OH OH1 432

45 463

OH OH2 492

OH OH3 490

OH OH4 487

OH OH6 445

991

992

H1H9H10H1rsquo

H6rsquo

OH2rsquo

OH1

12

Detailed discussion about the NOESY and DQ-SQ 2D 1H-1H ssNMR experiments presented in 993

Figure 9 994

995

2D 1H-1H ssNMR experiments are a unique tool to explore through-space proximities 996

between 1H nuclei ssNMR offers a large number of possibilities each adapted to a given 997

system The advantage of 2D maps with respect to a dynamics approach is their direct 998

readability in clear two nuclei which exchange their magnetization through-space (whichever 999

mechanism may be responsible for this exchange) will display a specific cross-peak However 1000

the way cross-peaks are presented and interpreted strictly depends on the nature of the pulse 1001

program which in turn is chosen according to the physical nature of the sample 1002

1D temperature-dependent 1H NMR experiments performed on G-C181-OH (Figure 1003

4a in the main text) indicate sharp peaks at 90degC and broad peaks at 70degC Narrow peaks 1004

indicate fast andor large-amplitude molecular tumbling averaging the strong 1H-1H dipolar 1005

interactions This excludes the use of pulse sequences like DQ-SQ which is not efficient under 1006

these conditions For this reason we employ spin-diffusion-based pulse sequences like 1007

NOESY which is more adapted for samples with high molecular mobility When mixing times 1008

are chosen to be short (lt 1 s) one generally explores through-space proximities within about 5 1009

10 Aring On the contrary when molecular motions are restricted dipolar coupling becomes 1010

significant 1H NMR peaks are broader and pulse sequences like DQ-SQ can now be used 1011

with a spatial resolution of less than 5 Aring 1012

Whenever possible DQ-SQ experiments are preferred over spin diffusion-based pulse 1013

sequences because they are more selective in particular for likewise interactions which are 1014

not discriminated in NOESY-type experiments This is illustrated by the typical interaction 1015

schemes for NOESY (top Figure 9a) and DQ-SQ (top Figure 9b) For two given A and B 1016

protons 2D NOESY maps will display two on-diagonal and two square-shaped symmetrical 1017

off-diagonal cross-peaks Off-diagonal cross-peaks illustrate the mutual A-B and B-A 1018

interaction but the on-diagonal peaks are systematic and do not prove any A-A or B-B 1019

interaction Similarly to NOESY 2D DQ-SQ maps display off-diagonal peaks (symmetric 1020

across the diagonal top Figure 9b) upon A-B and B-A interaction but in contrast they will 1021

display on-diagonal peaks only in the presence of self-interaction In the tutorial example of 1022

Figure 9b top the on-diagonal A-A peak indicates mutual A-A interaction while lack of on-1023

diagonal B-B peak evidences the absence of B-B proximities Combining NOESY and DQ-SQ 1024

experiments is capital to study the local arrangement of G-C181-OH and its evolution with 1025

temperature 1026

13

At T= 90degC the H1 and H5rsquo contributions are very close but possible to discriminate 1027

them (δ(H1)= 354 ppm δ(H5rsquo)= 351 ppm) according to 1H NMR chemical shifts in Table S 1028

1 2D 1H-1H NOESY in Figure 9a also discriminates between the two components attributed 1029

on the basis of the interactions with their mutual neighbors (green H1-H2 and H5rsquo-H1rsquo squares 1030

in Figure 9a) On this basis 2D NOESY does not show a direct H1rsquo-H1 correlation (blue red-1031

crossed) but rather an extensive network of intra-sugar interactions These features exclude 1032

suggest intra- and inter-sugar interactions and they exclude head-tail arrangements as a major 1033

molecular configuration at 90degC 1034

At 70degC the 1H NMR peaks are broader indicating a reduction of the T2 (1H) relaxation 1035

times andor residual homonuclear dipolar interactions excluding the use of NOESY In this 1036

case 1H-1H DQ-SQ correlation experiments are best suited Figure 9b presents the 1H-1H DQ-1037

SQ experiment on G-C181-OH at 70degC Peaks are broader than in NOESY and attribution may 1038

be cumbersome However the experiment is performed at 700 MHz and MAS= 22 kHz which 1039

improves spectral resolution with respect to the same experiment performed at 300 MHz and 1040

MAS= 10 kHz The auto-peak at δH= 356 ppm in the SQ dimension is attributed to H1-H1 spin 1041

pairs mainly resulting from intra CH2 interaction while the DQ cross-peaks at δH= 156-356 1042

ppm and δH= 432-356 ppm respectively origin from H2-H1 and H1rsquo-H1 interactions (Figure 1043

9b) DQ-SQ map does not show any auto-peak corresponding to H1rsquo-H1rsquo or Hxrsquo-Hxrsquo (x= 2rsquo 1044

3rsquo 4rsquo) interactions (red-crossed blue Figure 9b) thus excluding any proximity (lt 5 Aring) between 1045

close glucose neighbors On the contrary DQ-SQ maps (recorded both at 700 MHz Figure 9b 1046

or at 300 MHz not shown) systematically show DQ cross-peaks which can only be attributed 1047

to H1rsquo-H1 interactions (δH= 432-356) The presence of these H1rsquo-H1 interactions can only be 1048

explained by a head-tail arrangement of the G-C181-OH molecules as illustrated below the 1049

DQ-SQ map in Figure 9b 1050

15

Does G-C181-OH assume a bilayer or interdigitated structure the latter being the one 359

found for other similar bolaform glucolipids4243 The size of G-C181-OH can be estimated to 360

be at the most 32 Aring given by the length L of the C16 tail (L = 22 Aring from the Tanford formula 361

L= 154 + 1265 times n (n= 16)59) and the sugar (about 10 Aring)4243 However the cis conformation 362

of the C=C bond imposes a ldquokinkrdquo of 60deg in the aliphatic chain which reduces the effective 363

length of the oleic backbone to about 17 Aring for an effective total G-C181-OH length of about 364

27 Aring Cryo-TEM provides a tube thickness varying from 35 Aring to 60 Aring and a vesicle membrane 365

thickness of at least 60 Aring WAXS suggests an interlamellar thickness (membrane and water) 366

between 40 Aring and 37 Aring and an inter-lipid distance in the tube of about 16 Aring In the absence of 367

a clear form factor oscillation profile as found for similar glucolipids4243 one must 368

acknowledge that these data pledge neither in favor of fully interdigitated nor a double layer 369

structure and might suggest a tilted lipid arrangement with partial interdigitation60 370

371

13C ssNMR demonstrates that G-C181-OH can assume two coexisting head-head and 372

head-tail configurations 373

Vesicles are characterized by low positive mean and gaussian curvatures while 374

nanotubes are described by a combination of a high positive mean and zero gaussian curvatures 375

In this regard the coexistence of vesicles and nanotubes requires glucolipids to adapt 376

themselves to both environments at the same time For this reason the formation of nanotube-377

vesicle networks generally requires a specific driving force able to impose a high mean and 378

zero gaussian curvature to the membrane In many systems insertion of at least a second 379

constituent (lipid surfactant protein) in the membrane becomes a necessary internal driving 380

force for the nucleation and growth of so-called membrane inclusions from which 381

topologically-connected nanotubes can form10ndash123261 as explained by orientational ordering 382

arguments62 It has also been shown that external driving forces such as use of laser 383

ldquotweezersrdquo30 osmotic pressure6 electroformation5ndash82829 or controlled suction13ndash16 are 384

otherwise necessary as also supported by theoretical studies19 385

The bending energy of tubes is higher than that of a vesicle membrane8 due to the 386

dependence of the membrane free energy from the principal curvatures For this reason 387

isotropic (rotational symmetry along the main axis) but also anisotropic33 lipids generally 388

stabilize low-curvature shapes like flat membranes or vesicles Lipid nanotubes are also a 389

common phase formed by amphiphiles but generally for lipids with a certain complexity in 390

their molecular structures as exampled by diacetylenic derivatives of phospholipids among 391

others262763ndash65 Negative contributions to the bending energy favoring the simultaneous 392

16

presence of interconnected vesicles and tubes was shown to be driven by the presence of 393

anisotropic lipids in a membrane continuum Spontaneous deviations to the elastic theory in the 394

presence of surfactants lipids and proteins with anisotropic structures favor the formation of 395

protrusions with radius of the order of the fraction of a micron9ndash1219ndash2362 396

In the absence of both internal and external driving forces mentioned above G-C181-397

OH should behave as a standard lipid Its DSC profile (Figure 2)536667 under both hydrated and 398

dry conditions identifies a hydration-independent transition (Tm1) slightly above room 399

temperature this is followed by one main transition at higher temperature (Tm2) in the dry 400

sample50 For such a profile one expects a gel phase (Lβ) below Tm1 and a liquid crystalline 401

structure possibly in the morphology of vesicles (L4 phase) possibly above Tm1 and most likely 402

above Tm231 as experimentally found On the basis of the thermogram recorded on the dry 403

powder one may also expect a ldquoripplerdquo (Pβ) phase between Tm1 and Tm2 characterized by 404

periodic undulations of the bilayer (generally addressed to as regions with both liquid order and 405

disorder)5455 406

In the present work instead of a Lβ phase we observe flat crystals below Tm1 and a 407

vesicle-nanotube network instead of the ldquoripplerdquo phase The latter is particularly unexpected 408

because the membrane is composed of a single-component (G-C181-OH) and no anisotropic 409

external force (osmotic hydrodynamic electric) is used to pull and stabilize the tubes In 410

addition the nanotubes are observed at fairly high temperatures above Tm1 (between about 411

60degC and 90degC) while this morphology is generally favored at low temperature (proportionality 412

with 1kT figure 5 in Ref 62) commonly below a given transition temperature characteristics 413

for each specific amphiphile27 Furthermore nanotubes of G-C181-OH seem to be stable in 414

time thus differing from what is found in electroformed vesicles78 They are also observed 415

away from phase transition events (may them be at Tm1 or Tm2) as otherwise reported before 416

for a diacetylenic derivative of phosphatidylcholine26 and many other systems27 417

A stable nanotube-vesicle network for a single-component phospholipid in the absence 418

of external driving forces is not expected and has never been reported before It could be 419

explained by the coexistence of two micro-separated phases of different rigidity This 420

hypothesis is excluded by confocal microscopy on Laurdan-labelled membranes and magic 421

angle spinning (MAS) solid-state NMR (ssNMR) experiments 422

Laurdan is a well-known fluorophore used to put in evidence the coexistence of rigid 423

and fluid domains in both model and biological membranes47ndash49 Laurdan was used here to label 424

the G-C181-OH membranes between Tm1 and Tm2 and to verify the possible micro-segregation 425

and coexistence of rigid (eg Lβ λem= 440 nm) and fluid (eg Lα λem= 490 nm) domains in 426

17

tubes and vesicles respectively A set of quantitative experiments performed on a 5 mgmL-1 427

solution summarized in Figure 5 shows an excess of fluorescence at only λem= 490 nm for 428

temperature between 90degC and 70degC (Figure 5h) The GP value which quantifies the rigidfluid 429

character of the membrane is systematically negative (Figure 5i) indicating an overall fluid 430

membrane at all temperatures GP is quantified on more than 50 regions of interest (ROIs) at 431

each temperature including vesicles and tubes alone as well as connected nanotube-vesicles 432

(Figure 5d-g) The dispersion of GP at 80degC and 70degC between -08 and -02 could suggest the 433

spurious presence of membranes with various degrees of flexibility although fluidity is always 434

predominant The combination of WAXS (broad peak d= 144 Aring-1 invariant with 435

temperature Figure S 6bc) and confocal fluorescence microscopy using Laurdan confirms the 436

single-phase of the membrane composed of G-C181-OH both in vesicles and nanotubes This 437

is also consistent with the lack of phase transitions in DSC experiments between ~85degC and 438

~60degC especially in hydrated samples We conclude that the vesicle-nanotube transition is not 439

driven by the local segregation of gel phase (Lβ) microdomains nor that the nanotubes are 440

composed of such phase 441

ssNMR is the best-suited approach to investigate the origin of the local driving force for 442

the nucleation of anisotropic inclusions of which the average orientation depends on the local 443

membrane and inclusion curvatures and was shown to promote tubing62 ssNMR can provide 444

information on the local molecular mobility eg through the nuclear spin relaxation behavior 445

but also on intermolecular interactions and stereochemistry eg through variations in the 446

chemical shift68ndash70 447

Ideally ssNMR should be performed on diluted G-C181-OH samples above Tm1 448

However fast MAS necessary for high resolution centrifuges the sample As a first 449

approximation hydrated G-C181-OH samples for instance at 50 wt as classically done for 450

phospholipids71 could be studied Unfortunately permeability of the sample holder combined 451

to heating (close to the boiling point of water) and MAS provoke dehydration and even sample 452

ejection After experiencing both of the above this approach was abandoned As a second 453

approximation the study of the dehydrated G-C181-OH powder avoids centrifugation and 454

evaporation If studying dried lipids may not reflect their exact behaviour under diluted 455

conditions valuable structural information can still be obtained in dehydrated molecular 456

systems like proteins thus illustrating that this approach is not unreasonable Furthermore if 457

DSC (Figure 2a) shows that hydration suppresses the Tm2 it also shows that the value of Tm1 is 458

not affected by water at concentrations as low as 10 wt For this reason performing a 459

structural analysis of dehydrated G-C181-OH by ssNMR above Tm1 may help understanding 460

18

the local arrangement of G-C181-OH in two morphologies vesicles and nanotubes having 461

sensibly different curvatures 462

To explore the local mobility of the most relevant chemical groups of G-C181-OH 463

above Tm2 (T= 110degC) and between Tm2 and Tm1 (70 lt T degC lt 90) we combine a series of 1H 464

and 13C experiments Single-pulse (SP) MAS experiments (1H and 13C) when recorded under 465

quantitative conditions allow the detection of all chemical species while cross-polarization 466

(CP) MAS (13C only) experiments are sensitive only to constrained groups for which the 1H-467

13C dipolar coupling is not averaged to zero by the molecular motions 468

469

Figure 6 ndash a) 1H MAS (10 kHz) ssNMR spectra recorded from 110degC to 70degC b) 13C MAS (10 kHz) ssNMR 470

spectra recorded using high power decoupling single pulse (SP) and cross-polarization (CP) pulse sequences 471

Contact time for CP experiments and corresponding temperatures are given on each spectrum Full spectra 472

are given in Figure S 7 473

The most relevant 1H and 13C spectra are given in Figure 6 (full range in Figure 7 and 474

Figure S 7) As expected the 1H solid-state SPMAS spectra show broader peaks compared to 475

the solution-state NMR spectrum (Figure 6a) reflecting the obvious solid-state character of the 476

sample at any temperature The sugar group and the C=C bond resonate in the 25-6 ppm range 477

and 535 ppm respectively while the OH groups of the sugar moiety are found below 45 ppm 478

(Figure 6a full assignment in Table S 1) Very broad (~1 ppm) at 100degC the position of the 479

OH peak varies with the temperature value and becomes narrower (lt~05 ppm) and visible at 480

δH= 464 ppm at T= 70degC Line narrowing and chemical shift variations of the OH resonance 481

in ssNMR is always a direct sign of H-bonding interactions whereas the strength of the H-482

bonding hence its OHhellipH distance is correlated to a dispersion in chemical shifts72ndash75 All other 483

60 55 50 45 40 35 30 25

ppm

110

90

80

70

910

1rsquo

6rsquo16

1

OH

134 132 130 128 126

ppm110 108 106 104 102 100

ppm

SP (T= 110degC)

SP (T= 70degC)

CP 50 ms (T= 70degC)

CP 3 ms (T= 70degC)

910 1rsquo

1H 13C

1

811

910

CH2 CH2

1rsquo

6rsquo

a) b)

C102

13024

C9101

13003

C92

12869 C1rsquo210382

C1rsquo110443

C1rsquo13035

2rsquo3rsquo4rsquo

T degC

5rsquo

19

resonances broaden from 110degC to 70degC thus indicating slower andor more anisotropic 484

reorientational motions of all chemical groups The specific dynamic behaviour of the sugar 485

headgroup is also reflected by its 1H transverse relaxation time T2 whereas the shorter the T2 486

the higher the rigidity Experiments performed between 110degC and 70degC shown and 487

commented in Figure S 8 in the Supporting Information globally show a single short (lt 1 ms 488

below 100degC) T2 relaxation component for H1rsquo H5rsquo and H6rsquo and two both short and long 489

(between 4 ms and 12 ms) T2 for the acyl chain (CH2 H910 H811) even below 100degC 490

At 110degC 13C SPMAS experiments (Figure 6b full range in Figure 7 and Figure S 7 491

assignments in Table 1) show a single peak at δC= 10335 ppm for C1rsquo and at δC= 13003 ppm 492

for C910 as expected for a symmetrical conformation with two chemically-equivalent 493

carbons76 At 70degC the C1rsquo peak disappears while C910 now shows two peaks around 128 494

ppm and 130 ppm (Table 1 full temperature series in Figure S 7b) attributed to C92 (δC= 12869 495

ppm) and C10 in a non-symmetric configuration respectively for which these two carbons are 496

chemically inequivalent76 More precisely the resonance at 130 ppm is actually the 497

superposition of two close resonances C9101 (δC= 13003 ppm) and C102 (δC= 13024 ppm) 498

where subscripts 1 and 2 stand for two different conformations of the cis-olefin group as 499

reported for various phases of oleic acid in the vicinity of 0degC76 500

501

Table 1 ndash 13C Chemical shifts of G-C181-OH in solution and solid-state Solid-state experiments are 502

recorded at MAS= 10 kHz All spectra are reported in Figure 7 and Figure S 7 The contact time for cross 503

polarization is 3 ms In all ssNMR experiments temperature is first raised at 110degC and then lowered to 504

either 90degC or 70degC Attribution of the 13C resonances in Table 1 is performed on the basis of the solution 505

spectrum (Figure S 1b) Additional peaks observed in both single pulse and cross polarization ssNMR 506

experiments are attributed when possible to a configuration effect of the given atom eg C9101 and C92 507

indicate two configurations of C9 All new peaks observed in solid-state and not detected in solution are 508

highlighted in red for convenience 509

510

511

δ ppm

Peak Liquid (DMSO-d6)

Solid Single Pulse

T= 110degC

Solid Single Pulse

T= 90degC

Solid Single Pulse

T= 70degC

Solid Cross Polarization

T= 70degC

C102

- - 13024 13024

C910C9101 13009 13003 13003 13003 13003

C1C9C10C1rsquo

C6rsquo

20

C92 - - - 12869 12869

C1rsquoC11 10329 10335 10335 - 10443

C12 - - - - 10382

C3 7724 7709 7702 - 7829

- 7634 7629 - 7621

7525

C2 7388 7398 7390 7370 7370

C4 7052 7077 7054 7041 7098

C16 6900 7016 - - 7029

- - - - 6793

C1C5C6 6153 6241 6236 6239 6262

C1C5C6 6117 - - - 5956

C2 3300 3290 3290 3283 3279

CH2(trans) - - - 3198 3196

CH2(gauche)C15 2941 2981 2987 3004 -

C811 2706 2747 2747 2832 2817

CH2 2597 2625 2628 2630 2817

512

13C CPMAS spectra at 70degC (Figure 6b full range in Figure 7 and Figure S 7d) at 513

increasing contact times confirm the presence of two cis-olefin environments one for which 514

C910 are equivalent labelled here C9101 and one for which they are inequivalent labelled 515

C92 C102 Both environments are rigid enough to be observed at short contact time (tc= 50 μs) 516

A similar double environment is also observed for the sugar C1rsquo CPMAS at 70degC shows two 517

peaks (C1rsquo1 at δC= 10443 ppm C1rsquo2 at δC= 10382 ppm Figure 6b Figure S 7d) These are 518

significantly shifted downfield from the C1rsquo peak observed at 110degC and are detected at contact 519

times as short as 50 μs thus indicating their rigid character One also finds that CPMAS is 520

totally inefficient at 100degC for short contact times (50 μs) and poorly efficient even at 3 ms 521

typical of fast-tumbling molecules (Figure S 7c) also both gauche (δC= 3004 ppm) and all-522

trans (δC= 3198 ppm) conformations of the CH2 carbons coexist at 70degC (Figure S 7b)77 523

According to ssNMR one main mobile molecular environment exists at 110degC as 524

shown by the well-resolved 1H spectrum single C=C (C9101) and anomeric (C1rsquo) resonances 525

gauche conformation of the aliphatic CH2 and single T2 of the aliphatic CH2 chain and sugar 526

protons The sugar group is certainly mobile (inefficient 1H13C CPMAS) but still more 527

constrainted than the acyl chain (shorter T2 (1H)) As the temperature is lowered the overall 528

rigidity increases (broadening of 1H NMR peaks faster 1H transverse relaxation and efficient 529

1H13C magnetization transfer during CPMAS) presumably driven by H-bonding 530

interactions (downfield shift and line narrowing of the OH 1H signal very short OH T2 loss of 531

the C1rsquo signal in SPMAS) At the same time a number of spectroscopic data (two concomitant 532

21

resonances for anomeric C1rsquo1 and C1rsquo2 cis-olefin C9101 and C92 C102 and CH2 gauche and 533

all-trans environments) and relaxation behaviour (fast and slow T2 relaxation components) 534

strongly suggests the coexistence of two molecular environments Further proofs of the 535

coexistence of two distinct molecular domains are given by a number of chemical shift 536

variations in the 13C CPMAS spectrum at 70degC compared to the one at 110degC 90degC and 537

solution This is shown in Figure 7 and in the list of 13C chemical shifts in Table 1 538

539

540

Figure 7 - Comparison between solution- and solid-state 13C NMR spectra of G-C181-OH Solid-state 541

experiments are recorded directly on the sample powder at a MAS of 10 kHz (4 mm zirconia rotor) SP 542

Single Pulse CP Cross Polarization In all the ssNMR experiments temperature is first raised at 110degC 543

and then lowered to either 90degC or 70degC Segmented lines are meant to highlight the known peak position 544

in solution thus stressing the strong chemical shift variations in the solid-state 545

546

NMR experiments above Tm2 suggest that vesicles could contain G-C181-OH with a 547

single family of mobile conformations characterized by chemically-equivalent C=C (red 548

Figure 8b) and a single C1rsquo conformation (green Figure 8b) Experiments between Tm1 and Tm2 549

otherwise suggest that two coexisting conformations characterized by chemically-equivalent 550

(red Figure 8b) and inequivalent (magenta Figure 8b) C=C could be associated to nanotubes 551

Two distinct C1rsquo sites (grey Figure 8b) also indicate two configurations of the sugar headgroup 552

A comment on this point will be given in the following paragraphs 553

The local (Aring-scale) variation of molecular mobility below Tm2 drives the coexistence of 554

two distinct molecular environments of the same G-C181-OH molecule This occurs within a 555

single-phase domain (as deduced by WAXS) of equivalent membrane rigidity (as derived from 556

Laurdan probe in confocal microscopy) Although never reported before the difference in the 557

local mobility of the same amphiphile (here G-C181-OH) could satisfy the theoretically-558

140 120 100 80 60 40 20

CP(3 ms)-70degC

SP-70degC

SP-90degC

SP-110degC

ppm

LiquidC910

C1rsquo

C2rsquo

C4rsquo

C3rsquo

C1

C6rsquoC5rsquo

C1

6

C8

11

C2 C

H2

C1

5

CH

2

DM

SO

13C

22

required conditions that induce local lipid segregation and formation of membrane inclusions 559

(Figure 8c) In case of their non-zero average orientation the local membrane curvature 560

increases62 and nanotubes can spontaneously form (Figure 8bc) More intriguing this 561

mechanism occurs at least 10degC above Tm1 although it should not be favored above the Tm27 562

From both experimental and theoretical backgrounds the conditions of topologically-563

connected vesicles-nanotubes systems are met only when either internal (two components) or 564

external driving forces are applied19 In the present single-lipid system of which the phase 565

behavior is summarized in Figure 8a the most plausible driving force could be explained by 566

the NMR results collected on the dry sample G-C181-OH concomitantly occupies two 567

molecular environments characterized by 1) two distinct conformations of the sugar 568

displaying equivalent mobility and 2) two conformations of the acyl chain unsaturation 569

570

23

571

Figure 8 ndash a) Thermotropic phase behavior of G-C181-OH at 5 mgmL-1 Note that Tm2 is observed on the 572

dry powder only b) Possible relationship between the molecular environment as derived from NMR and 573

membrane curvatures All the molecular drawings refer to the same G-C181-OH molecule (bulky ellipsoid 574

equiv sugar headgroup) while each color corresponds to a given configuration Grey of C1rsquo1 and C1rsquo2 indicates 575

the uncertainty about the assignment of the specific NMR peak The curvature C is defined as 1R R being 576

the curvature radius R1 R2 radius of curvature CG Gauss curvature CM Mean curvature c) Free 577

interpretation of the anisotropic high-curvature (CMgt 0 CG= 0) inclusions in the membrane (low CM and 578

CG) continuum driven by inter-sugar H-bonding and reduced intramembrane diffusion for a G-C181-OH 579

glucolipid 580

581

Flat crystalline structure

Nanotube-vesicle network

Vesicles

48degC

120deg

106 105 104 103 102 101

ppm

H9 H10H1rsquo

106 105 104 103 102 101

ppm106 105 104 103 102 101

ppm106 105 104 103 102 101

ppm

C=CC=C

C1rsquo1C1rsquo2

C1rsquo

Tm2(96degC)Tm1

a)

b)

c)

Same molecule different local orientations and sugar conformations

High curvature membrane

Low curvature inclusion

Flip-FlopStrong H-bonding

CG

CM ( + )

CG 0

CM 0 CG 0

CM 0

G-C181-OH

24

Two related questions are unanswered at the moment What actually drives nanotube 582

formation and is the organization of G-C181-OH within the vesicles and tubes the same Lipid 583

nanotubes have been studied for at least three decades over which broad experimental and 584

theoretical insight has been generated Despite this fact the actual understanding is still matter 585

of debate27 Micelle-to-nanotube and vesicle-to-nanotube phase transitions are generally 586

reported for single amphiphiles below a given Tm and intermediate phase composed of twisted 587

and flat fibers is concomitantly reported Initially thought to be a chirality-driven phenomenon 588

recent studies propose a more general mechanism involving crystallization and symmetry 589

breaking in the lipid packing driven by optimal packing27 If twisted ribbons were observed 590

before for a number of microbial glycolipid bolaamphiphiles44 they are not found for G-C181-591

OH This fact as well as the large body of cryo-TEM images and the observation of nanotubes 592

above Tm1 rather pledge in favour of a direct topological change from positive (vesicles) to 593

zero (nanotubes) Gaussian curvatures In this case vesicle-to-nanotube transitions could be 594

driven by external forces56307813ndash162829 595

Considering the absence of external forces in the present system there are only two 596

possibilities left spontaneous budding of the membrane or nanotube pulling after inter-vesicle 597

collisions due to Brownian motion Both are possible as inclusions composed of a different 598

arrangement of the same molecule in the membrane continuum can theoretically explain 599

spontaneous nanotube formation in an isotropic medium672223629ndash1218ndash21 However the second 600

one is probably more likely to prevail because it could explain the large number of nanotubes 601

connecting two distant vesicles Interestingly similar events occur for any phospholipid-based 602

vesicular solution but nanotubes are extremely rare unless at least two lipids of different 603

molecular structure are mixed together The spectroscopic 1H and 13C and relaxation (1H) 604

ssNMR data even if collected on a dry powder demonstrate that the coexistence of two G-605

C181-OH configurations involving both the sugar headgroup and fatty alcohol chain are 606

possible for this molecule As a matter of fact this is not so surprising as more than one energy 607

minima each corresponding to a given conformation are known for glucose its non-reducing 608

derivative methylglucose and carbohydrates in general78ndash80 609

The presence of two configurations could explain the coexistence between two 610

structures of different curvature but the data presented so far do not explain the origin of these 611

two configurations In other words what drives the change in the local structure of the sugar 612

headgroup and fatty alcohol within the membrane 613

The association of known self-assembled structures to spectroscopic 13C ssNMR data is 614

known for years in particular for gluconamides lipids having a linear sugar headgroup81ndash84 It 615

25

was found that gauche effects in some cases up to the γ position in the sugar chain induce 616

upfield chemical shifts while anti and trans conformations result in downfield shifts Each 617

structure (monolayer crystal bilayer crystal or micellar fiber) was then characterized by a given 618

organization of the gluconamide and a characteristic 13C ssNMR fingerprint which could be 619

explained by gauche γ-gauche anti and trans conformations of the headgroup Similar 620

structural-spectroscopic correlations have been collected over the years for much simpler cyclic 621

sugars85 whereas pyranoses can undergo important conformational changes (chair half-chair 622

boat skew) when the appropriate energy barriers are crossed78ndash8086 A quantitative 623

conformational study based on the 13C chemical shift variations shown in Table 1 would be 624

certainly helpful but very delicate in the context of the present state of the art because a 625

rigorous approach requires at least one known crystal structure8182 which is currently missing 626

for this compound A qualitative approach will then be adopted 627

G-C181-OH is a bolaform amphiphile and flip-flop could be a possible mechanism 628

involved in the nanotube stabilization whereas head-tail interactions have long been considered 629

as important in nanotube-forming bolaform amphiphiles6063 From mere considerations based 630

on T2 (1H) relaxation arguments one could exclude head-tail configurations if the R-CH2OH 631

endgroup (H1) was involved in a direct interaction with the sugar group its corresponding T2 632

would probably be comparable to the sugarrsquos one (OH and H1rsquo) but this does not seem to be 633

the case However this hypothesis would be based on the comparison of the relaxation from 634

CH and CH2 protons additionally involved in different kind of motions Besides the relaxation 635

of H1 is influenced by the strong overlap between H1 and H5rsquo For these reasons relaxation 636

arguments are not solid enough and will not be used 637

We then address this question by exploring intermolecular interactions in a direct 638

manner using 2D 1H-1H homonuclear ssNMR correlation spectroscopy At 90degC in the fast-639

tumbling domain of molecular motion (narrow 1H NMR peaks Figure 6a) and corresponding 640

to the vesicle-rich region spin diffusion (NOESY) experiments are preferred over dipolar 641

coupling based experiments because such interaction is averaged out by molecular motion 642

However when the molecular motion is more constrained in the domain of nanotube stability 643

at 70degC (broad 1H spectra Figure 6a) Double-Quantum Single-Quantum (DQ-SQ) correlation 644

experiments are best suited to probe interaction between close molecular neighbors These 645

experiments are reported in Figure 9 and the related interpretation is only briefly reported 646

below while a more detailed discussion can be found on Page 12 in the Supporting Information 647

26

648

Figure 9 ndash a) 2D 1H-1H NOESY NMR experiment performed at MAS= 10 kHz (4 mm probe) on a Bruker 649

300 MHz Avance III HD spectrometer using 300 ms of mixing time Sample temperature is regulated to 650

90degC reached by cooling from 110degC b) 2D 1H-1H DQ-SQ (BAck-to-BAck BABA pulse sequence used for 651

the DQ excitation scheme) NMR experiment performed at MAS= 22 kHz (25 mm probe) on a Bruker 700 652

MHz Avance III spectrometer using a DQ excitation time equal to 2 rotor periods Sample temperature is 653

regulated to 70degC reached by cooling from 110degC The typical schemes illustrating how to interpret these 654

2D experiments are reported above each experimental map Zooms of the sugar region and configurational 655

interpretation are given below each 2D map More information are given in the text below 656

H16-H16

H1-H1

H9-H10

H1rsquo-H1

H1rsquoH910 H16

H1

H1rsquo-H1rsquo

Hxrsquo-Hxrsquo

2

(pp

m)

(ppm)

315

811910 1rsquo

2rsquo2

3rsquo 4rsquo6rsquo

5rsquo

1rsquo2910

asymp asymp asympHDO MeOD Acetone

1

315

811 9101rsquo

2rsquo2

3rsquo4rsquo6rsquo

5rsquo

1rsquo2 9 10

asympasympasympHDOMeODAcetone

H1rsquo-H1

1rsquo

1

(p

pm

)

(ppm)

H1-H2

H1rsquo-H1

H1rsquo

H910

H1H5rsquo

H2

H1rsquo-H5rsquo

315

811910 1rsquo

2rsquo2

3rsquo 4rsquo6rsquo

5rsquo

1rsquo2910

asymp asymp asympHDO MeOD Acetone

1

H2rsquoH3rsquoH4rsquo

H2rsquoH3rsquoH4rsquo

X= 2rsquo3rsquo4rsquo

NOESY - T= 90degC

1H

1H

1H

1H

DQ-SQ - T= 70degC

a) b)

2

(pp

m)

(p

pm

)

(ppm)

A B

AhellipB

AhellipA

BhellipB

NOESY

(ppm)

A B

AhellipB

DQ-SQ

Single-quantum dimension

Do

ub

le-q

ua

ntu

m d

imen

sio

n

AhellipA

BhellipB

BhellipA

BhellipA

315

811910 1rsquo

2rsquo2

3rsquo 4rsquo6rsquo

5rsquo

1rsquo2910

asymp asymp asympHDO MeOD Acetone

1

lt 5 Aring

lt 5 Aring

H1-H2

H811

H15

H16

H6rsquo

H1rsquo

H1

H5rsquo

H2

H1-H2

H1rsquo-H1

H1rsquo-H5rsquo

H1rsquo

H1 H5rsquo

H1rsquo

H1

H5rsquo

H1H2rsquo H3rsquo H4rsquo

H1-H1

H1rsquo-H1

Hxrsquo-HxrsquoX= 2rsquo3rsquo4rsquo

H1-H2

H1rsquo

Zoom sugar region 28 ndash 45 ppmNOESY DQ-SQ

27

Head-tail or head-head configurations could be probed by the existence of H1rsquo-H1 657

interactions whereas H1 (δH= 356 ppm) is associated to the glucolipid tail and H1rsquo (δH= 432 658

ppm) to the sugar headgroup These atoms are more than 15 Aring apart and a direct interaction 659

could not be observed in both NOESY and DQ-SQ experiments NOESY experiments (Figure 660

9a) show that H1rsquo-H1 correlations are not observed Meanwhile clear-cut direct H1rsquo-H1 661

interactions are indeed observed are detected at T= 70degC (DQ cross-peaks related to H1rsquo-H1 662

correlation in DQ-SQ experiment shown in Figure 9b) Interestingly the absence of auto peaks 663

at δH= 432 ppm and δH= 335 ppm in the SQ dimension excludes H1rsquo-H1rsquo and Hxrsquo-Hxrsquo (x= 2 664

3 4) interactions respectively strengthening the idea that sugar headgroups from two nearest-665

neighbor G-C181-OH molecules are not adjacent in the nanotube-vesicle region 666

Homonuclear 2D ssNMR supports the idea according to which G-C181-OH can 667

undergo molecular flip-flops and a head-tail arrangement in the dry product Extrapolating these 668

conclusions in the lower concentration range lack of H1rsquo-H1 proximity at T= 90degC suggests a 669

head-head configuration in the vesicle-rich region and head-tail configuration within the 670

nanotubes It is however still unclear whether flip-flops are driven by spontaneous temperature-671

induced conformational changes in the glucolipid packing stimulating inclusions within the 672

membrane itself672223629ndash1218ndash21 or by defects generated by inter-vesicle collisions In fact a 673

combination of both is not unlikely Nanotubes are not stable when temperature is 20 to 30 674

degrees above Tm1 indicating that inter-vesicle collisions alone cannot explain the unusual 675

phenomenon In an energy landscape approach87 the head-head configuration is certainly 676

favored at high temperature while both head-head and head-tail configurations could be 677

favored in the mid-temperature region corresponding to the range between Tm1 and Tm2 in the 678

dry powder Both statistical and defect-induced flip-flop events contribute to form high mean 679

zero Gaussian curvature morphologies (nanotubes) for which the head-tail conformation 680

limiting the proximity between bulky sugar headgroups is probably more adapted (Figure 8c) 681

The flip-flop interpretation now reasonably explains the strong differences in the sugar and fatty 682

alcohol conformations deduced by the multiple NMR signals attributed to chemically-683

equivalent groups The existence of various low-energy conformations each having specific 684

NMR signature for sugars in solution78ndash8086 and solid-state81ndash84 further supports our 685

interpretation 686

One last point should be clarified Why between Tm1 and Tm2 are two molecular 687

environments able to coexist instead of a single one of average rigidity as found in the case of 688

ldquoripplerdquo phases50ndash54 We believe that this question can be answered based on the NMR results 689

the relatively short T2 values of the H1rsquo and OH protons shows the constrained mobility of the 690

28

sugar headgroup despite a certain flexibility of the chain (long T2 component) This trend may 691

suggest restrictions in translational mobility probably due to the sugar-OH hydrogen bonding 692

(Figure 8c) meaning that lipids are not able to diffuse within the membrane thus keeping two 693

different configurations stable 694

695

Conclusions 696

Topologically-connected nanotube-vesicle lipidic systems are commonly observed for 697

self-assembled membranes composed of at least two amphiphiles and spontaneously driven by 698

anisotropic inclusions andor by external directional forces This work shows that strong 699

differences in membrane curvature can spontaneously occur in single-lipid systems without 700

external anisotropic forces Coexistence and interconnection between nanotubes and vesicles 701

are evidenced via cryo-TEM and fluorescence confocal microscopy in a single-phase region at 702

temperature above a transition temperature Tm1= 483degC determined by DSC on the hydrated 703

sample but below the second transition temperature Tm2 (957degC) determined by DSC on the 704

dry powder X-ray diffraction (WAXS) also shows the coexistence of two lattice periods one 705

attributed to an inter-membrane repeating distance and the other possibly related to the intra-706

nanotube glucolipid arrangement Labelling the systems with Laurdan a fluorophore of which 707

the emission strongly varies with the local membrane rigidity excludes the presence of rigid 708

micro-separated phases (eg Lβ) 709

1H and 13C ssNMR despite the fact that it can only be performed on the dry powder 710

confirm the presence of one main fluid G-C181-OH environment above Tm2 Below Tm2 the 711

splitting of some 13C peaks and the coexistence of short and long 1H relaxation components 712

demonstrate that G-C181-OH can simultaneously pack into structurally and dynamically 713

inequivalent environments Combining advanced 2D 1H-1H correlation experiments based on 714

spin diffusion or double-quanta excitation suggests that vesicles could contain a preferred head-715

head configuration of the bolaform lipid while nanotubes a preferred head-tail assembly 716

The presence of two environments composed of the same molecule characterized by 717

different molecular configurations and rigidity fulfills the theoretical conditions necessary to 718

spontaneously nucleate a membrane inclusion of non-zero average orientation described as 719

necessary to observe topologically-connected nanotube-vesicle structures in the absence of 720

external forces This hypothesis could also be enriched by the possibility of nanotubes to be 721

pulled away between two vesicles after a collision event 722

This work expands the actual theoretical and experimental framework on lipid networks 723

stabilized by a different radius of curvature It could also motivate the synthesis or isolation 724

29

of new lipids providing a higher level of 3D architectural control in soft networks possibly in 725

the range of room temperature 726

727

Acknowledgements 728

Dr Andrea Lassenberger and Dr Sylvain Preacutevost at Institut Laue Langevin (ILL Grenoble 729

France) are kindly acknowledged for their assistance on the SANS experiment Confocal 730

images were performed at the Institut de Biologie Paris Seine (IBPS) imaging facility Jean-731

Franccedilois Gilles (Imaging Core Facility Institut de Biologie Paris Seine (IBPS) CNRS 732

Sorbonne Universiteacute Paris France) is kindly acknowledged for his assistance on the confocal 733

microscopy imaging Dr Guillaume Laurent (LCMCP Sorbonne Universiteacute Paris France) is 734

kindly acknowledged for helpful discussions The French region Ile-de-France SESAME 735

program is acknowledged for financial support (700 MHz NMR 736

spectrometer) Dr Olivier Diat (ICSM Marcoule France) Prof eronikaIglič and Prof V Aleš737

onsul discussi Slovenia) are kindly acknowledged for helpfUniversity of LjubljanaIglič (-Kralj 738

739

Financial support 740

WAXS experiments have been supported by Soleil Light Source Saint Aubin France proposal 741

Ndeg 20201747 SANS experiments have been supported by ILL proposal Ndeg9-13-778 742

743

References 744

1 D Kabaso N Bobrovska W Goacuteźdź E Gongadze V Kralj-Iglič R Zorec and A 745

Iglič Bioelectrochemistry 2012 87 204ndash210 746

2 C R Safinya U Raviv D J Needleman A Zidovska M C Choi M A Ojeda-747

Lopez K K Ewert Y Li H P Miller J Quispe B Carragher C S Potter M W 748

Kim S C Feinstein and L Wilson Adv Mater 2011 23 2260ndash2270 749

3 P Veranič M Lokar G J Schuumltz J Weghuber S Wieser H Haumlgerstrand V Kralj-750

Iglič and A Iglič Biophys J 2008 95 4416ndash4425 751

4 A Rustom R Saffrich I Markovic P Walther and H-H Gerdes Science (80- ) 752

2004 303 1007ndash1010 753

5 M I Angelova S Soleacuteau P Meacuteleacuteard F Faucon and P Bothorel Prog Colloid 754

Polym Sci 1992 89 127ndash131 755

6 L Mathivet S Cribier and P F Devaux Biophys J 1996 70 1112ndash1121 756

7 V Kralj-Iglič G Gomišček J Majhenc V Arrigler and S Svetina Colloids Surfaces 757

A Physicochem Eng Asp 2001 181 315ndash318 758

30

8 V Kralj-Iglič A Iglič G Gomišček F Sevšek V Arrigler and H Haumlgerstrand J 759

Phys A Math Gen 2002 35 1533ndash1549 760

9 M Drab D Stopar V Kralj-Iglič and A Iglič Cells 2019 8 626 761

10 A Zidovska K K Ewert J Quispe B Carragher C S Potter and C R Safinya 762

Langmuir 2009 25 2979ndash2985 763

11 A Zidovska K K Ewert J Quispe B Carragher C S Potter and C R Safinya Soft 764

Matter 2011 7 8363ndash8369 765

12 A Zidovska K K Ewert J Quispe B Carragher C S Potter and C R Safinya 766

Biochim Biophys Acta - Biomembr 2009 1788 1869ndash1876 767

13 M Karlsson K Sott M Davidson A S Cans P Linderholm D Chiu and O Orwar 768

Proc Natl Acad Sci U S A 2002 99 11573ndash11578 769

14 A Karlsson R Karlsson M Karlsson A-S Cans A Stromberg F Ryttsen and O 770

Orwar Nature 2001 409 150ndash152 771

15 M Karlsson K Sott A S Cans A Karlsson R Karlsson and O Orwar Langmuir 772

2001 17 6754ndash6758 773

16 E Evans H Bowman A Leung D Needham and D Tirrell Science (80- ) 1996 774

273 933ndash935 775

17 U Seifert Configurations of fluid membranes and vesicles 1997 vol 46 776

18 J Kaumls and E Sackmann Biophys J 1991 60 825ndash844 777

19 N Bobrovska W Goacuteźdź V Kralj-Iglič and A Iglič PLoS One 2013 8 e73941 778

20 A Iglič B Babnik K Bohinc M Fošnarič H Haumlgerstrand and V Kralj-Iglič J 779

Biomech 2007 40 579ndash585 780

21 K Bohinc T Slivnik A Iglič M Brumen and V Kralj-Iglič J Phys Chem C 2007 781

111 9709ndash9718 782

22 A Iglič H Haumlgerstrand P Veranič A Plemenitaš and V Kralj-Iglič J Theor Biol 783

2006 240 368ndash373 784

23 V Kralj-Iglič and P Veranič Adv Planar Lipid Bilayers Liposomes 2006 5 129ndash785

149 786

24 A Iglic and V Kralj-Iglic Adv Planar Lipid Bilayers Liposomes 2006 4 253ndash279 787

25 S Penič L Mesarec M Fošnarič L Mroacutewczyńska H Haumlgerstrand V Kralj-Iglič and 788

A Iglič Front Phys 2020 8 342 789

26 P Yager and P E Schoen Mol Cryst Liq Cryst 1984 106 371ndash381 790

27 T G Barclay K Constantopoulos and J Matisons Chem Rev 2014 114 10217ndash791

10291 792

31

28 B Božič G Gomišček V Kralj-Iglič S Svetina and B Žekš Eur Biophys J 2002 793

31 487ndash496 794

29 A Iglič H Haumlgerstrand M Bobrowska-Haumlgerstrand V Arrigler and V Kralj-Iglič 795

Phys Lett A 2003 310 493ndash497 796

30 R Bar-Ziv and E Moses Phys Rev Lett 1994 73 1392ndash1395 797

31 S Chiruvolu H E Warriner E Naranjo S H Idziak J O Raumldler R J Plano J A 798

Zasadzinski and C R Safinya Science (80- ) 1994 266 1222ndash1225 799

32 V Kralj-Iglič A Iglič H Haumlgerstrand and P Peterlin Phys Rev E 2000 61 4230ndash800

4234 801

33 L Mesarec M Drab S Penič V Kralj-Iglič and A Iglič Int J Mol Sci 2021 22 802

2348 803

34 L Van Renterghem S L K W Roelants N Baccile K Uyttersprot M C Taelman 804

B Everaert S Mincke S Ledegen S Debrouwer K Scholtens C V Stevens W 805

Soetaert L Van Renterghem S L K W Roelants N Baccile K Uytersprot M C 806

Taelman B Everaert S Mincke S Ledegen S Debrouwer K Scholtens C V 807

Stevens and W Soetaert Biotechnol Bioeng 2018 115 1195ndash1206 808

35 J Clark and F Deswarte Introduction to Chemicals from Biomass Second Edition 809

John Wiley amp Sons 2015 810

36 K V Sajna R Houmlfer R K Sukumaran L D Gottumukkala and A Pandey in 811

Industrial Biorefineries and White Biotechnology eds A Pandey R Houmlfer M 812

Taherzadeh K M Nampoothiri and C Larroche Amsterdam Oxford Waltham 813

Elsevier 2015 pp 499ndash521 814

37 R Marchant and I M Banat Trends Biotechnol 2012 30 558ndash565 815

38 K Hill and C LeHen-Ferrenbach in Sugar-Based Surfactants Fundamentals and 816

Applications ed C C Ruiz Boca Raton CRC Press 2009 pp 1ndash20 817

39 L Rodrigues I M Banat J Teixeira and R Oliveira J Antimicrob Chemother 818

2006 57 609ndash618 819

40 J D Desai and I M Banat Microbiol Mol Biol Rev 1997 61 47ndash64 820

41 S L K W Roelants K Ciesielska S L De Maeseneire H Moens B Everaert S 821

Verweire Q Denon B Vanlerberghe I N A Van Bogaert P Van der Meeren B 822

Devreese and W Soetaert Biotechnol Bioeng 2016 113 550ndash559 823

42 N Baccile M Selmane P Le Griel S Preacutevost J Perez C V Stevens E Delbeke S 824

Zibek M Guenther W Soetaert I N A Van Bogaert and S Roelants Langmuir 825

2016 32 6343ndash6359 826

32

43 N Baccile A-S Cuvier S Preacutevost C V Stevens E Delbeke J Berton W Soetaert 827

I N A Van Bogaert and S Roelants Langmuir 2016 32 10881ndash10894 828

44 N Baccile C Seyrig A Poirier S A Castro S L K W Roelants and S Abel Green 829

Chem 2021 23 3842ndash3944 830

45 A Salis L Cappai C Carucci D F Parsons and M Monduzzi J Phys Chem Lett 831

2020 11 6805ndash6811 832

46 J Schindelin I Arganda-Carreras E Frise V Kaynig M Longair T Pietzsch S 833

Preibisch C Rueden S Saalfeld B Schmid J-Y Tinevez D J White V 834

Hartenstein K Eliceiri P Tomancak and A Cardona Nat Methods 2012 9 676ndash835

682 836

47 T Parasassi and E Gratton J Fluoresc 1995 5 59ndash69 837

48 M M Dodes Traian F L Gonzaacutelez Flecha and V Levi J Lipid Res 2012 53 609ndash838

616 839

49 D M Owen C Rentero A Magenau A Abu-Siniyeh and K Gaus Nat Protoc 840

2012 7 24ndash35 841

50 W Stillwell in An Introduction to Biological Membranes Academic Press 2016 pp 842

135ndash180 843

51 W Stillwell in An Introduction to Biological Membranes Academic Press 2016 pp 844

181ndash220 845

52 B Tenchov Chem Phys Lipids 1991 57 165ndash177 846

53 R Koynova and M Caffrey Biochim Biophys Acta 1998 1376 91ndash145 847

54 F Schmid S Dolezel O Lenz and S Meinhardt J Phys Conf Ser 2014 487 848

012004 849

55 S Meinhardt R L C Vink and F Schmid Proc Natl Acad Sci U S A 2013 110 850

4476ndash4481 851

56 O Glatter and O Kratky Small Angle X-ray Scattering Academic Press London 852

1982 853

57 B A Cunningham A D Brown D H Wolfe W P Williams and A Brain Phys 854

Rev E 1998 58 3662ndash3672 855

58 M Nadler A Steiner T Dvir O Szekely P Szekely A Ginsburg R Asor R Resh 856

C Tamburu M Peres and U Raviv Soft Matter 2011 7 1512ndash1523 857

59 C Tanford The hydrophobic effect Formation of micelles and biological membranes 858

John Wiley amp Sons Inc New York 1973 859

60 M Masuda and T Shimizu Langmuir 2004 20 5969ndash5977 860

33

61 A Zidovska K K Ewert J Quispe B Carragher C S Potter and C R Safinya 861

Block Liposomes Vesicles of Charged Lipids with Distinctly Shaped Nanoscale 862

Sphere- Pear- Tube- or Rod-Segments Elsevier Inc 1st edn 2009 vol 465 863

62 A Iglic V Kralj-Iglic B Božič M Bobrowska-Haumlgerstrand B Isomaa and H 864

Haumlgerstrand Bioelectrochemistry 2000 52 203ndash211 865

63 T Shimizu M Masuda and H Minamikawa Chem Rev 2005 105 1401ndash1444 866

64 J V Selinger and J M Schnur Phys Rev Lett 1993 71 4091ndash4094 867

65 J M Schnur Science (80- ) 1993 262 1669ndash1676 868

66 B Tenchov R Koynova and G Rapp Biophys J 2001 80 1873ndash1890 869

67 R Koynova and B Tenchov Ser Biomech 2017 31 3ndash19 870

68 S J Angyal and R Le Fur Carbohydr Res 1984 126 15ndash26 871

69 J K Whitesell and M A Minton J Am Chem Soc 1987 109 225ndash228 872

70 T Takayama S Ando and I Ando J Mol Struct 1990 220 245ndash250 873

71 K Gawrisch N V Eldho and I V Polozov Chem Phys Lipids 2002 116 135ndash151 874

72 F Tielens C Gervais J F Lambert F Mauri and D Costa Chem Mater 2008 20 875

3336ndash3344 876

73 G A Jeffrey and Y Yeon Acta Crystallogr Sect B 1986 42 410ndash413 877

74 B Berglund and R W Vaughan J Chem Phys 1980 73 2037ndash2043 878

75 C Gervais C Coelho T Azaiumls J Maquet G Laurent F Pourpoint C Bonhomme 879

P Florian B Alonso G Guerrero P H Mutin and F Mauri J Magn Reson 2007 880

187 131ndash140 881

76 C Akita T Kawaguchi F Kaneko H Yamamoto and M Suzuki J Phys Chem B 882

2004 108 4862ndash4868 883

77 R N Purusottam L Seacutenicourt J J Lacapegravere and P Tekely Biochim Biophys Acta - 884

Biomembr 2015 1848 3134ndash3139 885

78 X Biarneacutes A Ardegravevol A Planas C Rovira A Laio and M Parrinello J Am Chem 886

Soc 2007 129 10686ndash10693 887

79 L M J Kroon‐Batenburg and J Kroon Biopolymers 1990 29 1243ndash1248 888

80 K N Kirschner and R J Woods Proc Natl Acad Sci U S A 2001 98 10541ndash889

10545 890

81 S Svenson B Kirste and J H Fuhrhop J Am Chem Soc 1994 116 11969ndash11975 891

82 I Sack S Macholl J H Fuhrhop and G Buntkowsky Phys Chem Chem Phys 892

2000 2 1781ndash1788 893

83 S Svenson J Koening and J H Fuhrhop J Phys Chem 1994 98 1022ndash1028 894

34

84 S Svenson A Schaefer and J H Fuhrhop J Chem Soc Perkin Trans 2 1994 2 895

1023ndash1028 896

85 K Bock and C Pedersen Adv Carbohydr Chem Biochem 1983 41 27ndash66 897

86 T B Grindley Structure and Conformation of Carbohydrates Springer Berlin 898

Heidelberg 2008 899

87 F Tantakitti J Boekhoven X Wang R V Kazantsev T Yu J Li E Zhuang R 900

Zandi J H Ortony C J Newcomb L C Palmer G S Shekhawat M O De La 901

Cruz G C Schatz and S I Stupp Nat Mater 2016 15 469ndash476 902

903

1

Supporting Information 904

905

Topological connection between vesicles and nanotubes in single-906

component lipid membranes driven by head-tail interactions 907

908

Niki Baccilea Ceacutedric Lorthioira Abdoul Aziz Baa Patrick Le Griela Cristina Coelhob Javier 909

Perezc Wim Soetaertd Sophie L K W Roelantsd 910

911

a Sorbonne Universiteacute Centre National de la Recherche Scientifique Laboratoire de Chimie de 912

la Matiegravere Condenseacutee de Paris LCMCP F-75005 Paris France 913

b Institut des Mateacuteriaux de Paris Centre Sorbonne Universiteacute Paris France 914

c Synchrotron Soleil LrsquoOrme des Merisiers Saint-Aubin BP48 91192 Gif-sur-Yvette Cedex 915

France 916

d InBio Department of Biotechnology Ghent University Ghent Belgium 917

918

919

Corresponding author 920

Dr Niki Baccile 921

E-mail address nikibaccilesorbonne-universitefr 922

Phone +33 1 44 27 56 77 923

924

925

926

2

a) 927

Parameter Method

Dry Matter (DM ) 996 Infrared balance 105degC

Glucose () 012 HPLC-Metacarb

Glycerol () nd HPLC-Metacarb

Free Fatty Acid content () lt001 Internal method BBEPP

GC with FID Detector

Oil () lt001 Internal method BBEPP

GC with FID Detector

Moulds (CFUg) lt10 3M-Nordval ndeg16 B

Yeasts (CFUg) lt10 3M-Nordval ndeg16 B

Anaerobic count (CFUg) lt10 SP-VG M005 B

Protein () tbd BCA protein assay

928

Figure S 1 - a) HPLC-ELSD chromatogram and composition table 929

930

931

932

933

3

b) 934

935

Group Ndeg δ ppm

CH2 1 354

CH2 2 153

CH2 3-712-14 132

CH2 811 203

CH 910 534

CH2 15 162

CH2 16 388

CH 1 424

CH 2 316

CH 3 334

CH 4 329

CH 5 354

CH2 6 366

Figure S 1 (next) ndash b) 1H NMR spectrum (MeOD-d4) recorded on the purified G-C181-OH compounds 936

and related assignment 937

938

2

811910 1rsquo

5rsquo1

3rsquo4rsquo

1rsquo1910

asymp asymp asympHDO MeOD Acetone

6rsquo2rsquo

1615

3-712-14

4

939

940

Figure S 2 ndash Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 130degC 941

5

942

943

Figure S 3 - Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 90degC 944

945

90degC

1

22

6

946

947

948

949

Figure S 4 ndash Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 70degC 950

951

21

11

1

2

5

3

4

45

5

6

7

7

7

952

953

954

Figure S 5 ndash Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 25degC 955

956

957

1

1 1

8

958

Figure S 6 ndash a) SANS experiment recorded at 90degC for a 5 mgmL G-C181-OH aqueous solution b-c) 959

WAXS experiments recorded on aqueous solutions of G-C181-OH at b) 5 mgmL and c) 25 mgmL d) 960

Structural interpretation of the repeating distances pointed out in c) 961

962

963

964

001 01 1

q Aring-1

25degC

68degC

88degC

111degC

01 1q Aring

-1

02 04 06 08 1q Aring

-1

001 01

10-1

100

101

Inte

nsity

cm

-1

q Aring-1

-2

T= 90degC

SANS WAXS WAXS

a) b) c)

dv

dL d= 144 Aring-1

di

ddH2O

dt

dvdL

dtdi d

d)

9

965

966

Figure S 7 ndash ssNMR MAS (10 kHz) experiments recorded on a G-C181-OH dry powder a) 1H spectra 967

recorded from 110degC to 70degC b) 13C single pulse (SP) spectra recorded with high power 1H decoupling from 968

110degC to 70degC c) 13C cross polarization (CP) spectra recorded at 110degC and contact time tc of 50 micros and 969

3000 micros d) 13C CP spectra recorded at 70degC and for 20 lt tc micros lt 3000 970

971

1

811

910

CH2 CH2

1rsquo

6rsquo

8 6 4 2 0 ppm

110

90

80

70

910

1rsquo

6rsquo16

1

2rsquo3rsquo4rsquo

811

OH

45 40 35 30 25 20 15

ppm

200 180 160 140 120 100 80 60 40 20 0

ppm140 120 100 80

ppm

CH2-gauche

3004

110

90

80

70

65

60

50

25

C91013003

C9 - 12869

CH2-all-trans

3198

C1rsquo10335

a) b)

1H

T degCT degC

13C Single Pulse MAS

A

B

A

B

70degC70degC

5rsquo

215

200 180 160 140 120 100 80 60 40 20 0

ppm

13C CPMAST= 110degC

200 180 160 140 120 100 80 60 40 20 0

ppm

132 131 130 129 128 127 126

ppm

106 105 104 103 102

ppm

13C CPMAST= 70degC

50

tc micros

3000

20

tc micros

50 500 1000 3000

tc micros

C102

13024 C910113003

C92

12869

C1rsquo213082

C1rsquo110443

A B

106 105 104 103 102

ppm

3000

1000

500

50

20

AB

c)

d)

10

972

Figure S 8 ndash a) Short and d) long T2 components displayed by the 1H transverse relaxation signal c) 1H 973

transverse relaxation signals recorded on a G-C181-OH dry powder from 110degC to 70degC The area of each 974

given resonance is plotted (log-lin) as a function of the echo time (CPMG pulse sequence) for each 975

temperature and fitted using the equation given above T2s and T2l indicate the short and long T2 relaxation 976

times respectively obtained by describing the 1H transverse relaxation signal using a biexponential decay 977

For H1rsquo H5rsquo and H6rsquo the 1H relaxation function could be described by a monoexponential decay 978

- Protons from the sugar group (H1rsquo H5rsquo H6rsquo) show a single T2 relaxation component (Figure 979

6c and Figure S 8) while protons of the acyl chain (CH2 H910 H811) show both short (Figure 980

6c) and long (Figure 6d) T2 relaxation component 981

- The short T2 relaxation time systematically decreases from 110degC to 70degC for all species by a 982

factor ranging from two to seven depending on the specific chemical group Very long T2 are 983

observed at 110degC for all acyl species while nearly unchanged values are measured below 984

90degC without any clear trend 985

- The sugar protons (H1rsquo H5rsquo H6rsquo) have the shortest set of T2 values (lt 1 ms) below 100degC 986

The long component of T2 generally ranges between 4 ms and 12 ms 987

0 1 2 3 4 5 6 7 8

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1 2 3 4 5 6 7 8

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

H1

H8H11

H9H10

CH2

H1rsquo

H6rsquo

CH2

H910H811 H1CH2

H1rsquo H5rsquo6rsquo

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

OH

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45 T= 110degC

T= 90degC

T= 80degC

T= 70degC

Log

In

ten

sity

Time ms

Fit

)

CH2 H910 H811 H1 H56 H1 OH

00

05

10

15

20

25

30

35

40

T2 m

s

T= 110degC

T= 90degC

T= 80degC

T= 70degC

CH2 H910 H811 H1

0

2

4

6

8

10

12

T2 m

s

T= 110degC

T= 90degC

T= 80degC

T= 70degC

Short T2(1H) Long T2(1H)

a) b)

c)

11

Table S 1 ndash 1H chemical shifts of G-C181-OH in solution and in the solid-state Solid-state spectra are 988

recorded at a MAS frequency of 10 kHz in DMSO-d6 broad peak 989

990

δH ppm

Group Ndeg Liquid

(MeOD-d4) Solid

(T= 110degC) Solid

(T= 70degC)

CH2 H1 354 357

CH2 H2 153 156

CH2 H3-712-14 132 135

CH2 H811 203 203

CH H910 534 534

CH2 H15 162 164

CH2 H16 388 383

CH H1 424 432

CH H2 316 335

CH H3 334 335

CH H4 329 335

CH H5 354 351 357

CH2 H6 366 383

OH OH1 432

45 463

OH OH2 492

OH OH3 490

OH OH4 487

OH OH6 445

991

992

H1H9H10H1rsquo

H6rsquo

OH2rsquo

OH1

12

Detailed discussion about the NOESY and DQ-SQ 2D 1H-1H ssNMR experiments presented in 993

Figure 9 994

995

2D 1H-1H ssNMR experiments are a unique tool to explore through-space proximities 996

between 1H nuclei ssNMR offers a large number of possibilities each adapted to a given 997

system The advantage of 2D maps with respect to a dynamics approach is their direct 998

readability in clear two nuclei which exchange their magnetization through-space (whichever 999

mechanism may be responsible for this exchange) will display a specific cross-peak However 1000

the way cross-peaks are presented and interpreted strictly depends on the nature of the pulse 1001

program which in turn is chosen according to the physical nature of the sample 1002

1D temperature-dependent 1H NMR experiments performed on G-C181-OH (Figure 1003

4a in the main text) indicate sharp peaks at 90degC and broad peaks at 70degC Narrow peaks 1004

indicate fast andor large-amplitude molecular tumbling averaging the strong 1H-1H dipolar 1005

interactions This excludes the use of pulse sequences like DQ-SQ which is not efficient under 1006

these conditions For this reason we employ spin-diffusion-based pulse sequences like 1007

NOESY which is more adapted for samples with high molecular mobility When mixing times 1008

are chosen to be short (lt 1 s) one generally explores through-space proximities within about 5 1009

10 Aring On the contrary when molecular motions are restricted dipolar coupling becomes 1010

significant 1H NMR peaks are broader and pulse sequences like DQ-SQ can now be used 1011

with a spatial resolution of less than 5 Aring 1012

Whenever possible DQ-SQ experiments are preferred over spin diffusion-based pulse 1013

sequences because they are more selective in particular for likewise interactions which are 1014

not discriminated in NOESY-type experiments This is illustrated by the typical interaction 1015

schemes for NOESY (top Figure 9a) and DQ-SQ (top Figure 9b) For two given A and B 1016

protons 2D NOESY maps will display two on-diagonal and two square-shaped symmetrical 1017

off-diagonal cross-peaks Off-diagonal cross-peaks illustrate the mutual A-B and B-A 1018

interaction but the on-diagonal peaks are systematic and do not prove any A-A or B-B 1019

interaction Similarly to NOESY 2D DQ-SQ maps display off-diagonal peaks (symmetric 1020

across the diagonal top Figure 9b) upon A-B and B-A interaction but in contrast they will 1021

display on-diagonal peaks only in the presence of self-interaction In the tutorial example of 1022

Figure 9b top the on-diagonal A-A peak indicates mutual A-A interaction while lack of on-1023

diagonal B-B peak evidences the absence of B-B proximities Combining NOESY and DQ-SQ 1024

experiments is capital to study the local arrangement of G-C181-OH and its evolution with 1025

temperature 1026

13

At T= 90degC the H1 and H5rsquo contributions are very close but possible to discriminate 1027

them (δ(H1)= 354 ppm δ(H5rsquo)= 351 ppm) according to 1H NMR chemical shifts in Table S 1028

1 2D 1H-1H NOESY in Figure 9a also discriminates between the two components attributed 1029

on the basis of the interactions with their mutual neighbors (green H1-H2 and H5rsquo-H1rsquo squares 1030

in Figure 9a) On this basis 2D NOESY does not show a direct H1rsquo-H1 correlation (blue red-1031

crossed) but rather an extensive network of intra-sugar interactions These features exclude 1032

suggest intra- and inter-sugar interactions and they exclude head-tail arrangements as a major 1033

molecular configuration at 90degC 1034

At 70degC the 1H NMR peaks are broader indicating a reduction of the T2 (1H) relaxation 1035

times andor residual homonuclear dipolar interactions excluding the use of NOESY In this 1036

case 1H-1H DQ-SQ correlation experiments are best suited Figure 9b presents the 1H-1H DQ-1037

SQ experiment on G-C181-OH at 70degC Peaks are broader than in NOESY and attribution may 1038

be cumbersome However the experiment is performed at 700 MHz and MAS= 22 kHz which 1039

improves spectral resolution with respect to the same experiment performed at 300 MHz and 1040

MAS= 10 kHz The auto-peak at δH= 356 ppm in the SQ dimension is attributed to H1-H1 spin 1041

pairs mainly resulting from intra CH2 interaction while the DQ cross-peaks at δH= 156-356 1042

ppm and δH= 432-356 ppm respectively origin from H2-H1 and H1rsquo-H1 interactions (Figure 1043

9b) DQ-SQ map does not show any auto-peak corresponding to H1rsquo-H1rsquo or Hxrsquo-Hxrsquo (x= 2rsquo 1044

3rsquo 4rsquo) interactions (red-crossed blue Figure 9b) thus excluding any proximity (lt 5 Aring) between 1045

close glucose neighbors On the contrary DQ-SQ maps (recorded both at 700 MHz Figure 9b 1046

or at 300 MHz not shown) systematically show DQ cross-peaks which can only be attributed 1047

to H1rsquo-H1 interactions (δH= 432-356) The presence of these H1rsquo-H1 interactions can only be 1048

explained by a head-tail arrangement of the G-C181-OH molecules as illustrated below the 1049

DQ-SQ map in Figure 9b 1050

16

presence of interconnected vesicles and tubes was shown to be driven by the presence of 393

anisotropic lipids in a membrane continuum Spontaneous deviations to the elastic theory in the 394

presence of surfactants lipids and proteins with anisotropic structures favor the formation of 395

protrusions with radius of the order of the fraction of a micron9ndash1219ndash2362 396

In the absence of both internal and external driving forces mentioned above G-C181-397

OH should behave as a standard lipid Its DSC profile (Figure 2)536667 under both hydrated and 398

dry conditions identifies a hydration-independent transition (Tm1) slightly above room 399

temperature this is followed by one main transition at higher temperature (Tm2) in the dry 400

sample50 For such a profile one expects a gel phase (Lβ) below Tm1 and a liquid crystalline 401

structure possibly in the morphology of vesicles (L4 phase) possibly above Tm1 and most likely 402

above Tm231 as experimentally found On the basis of the thermogram recorded on the dry 403

powder one may also expect a ldquoripplerdquo (Pβ) phase between Tm1 and Tm2 characterized by 404

periodic undulations of the bilayer (generally addressed to as regions with both liquid order and 405

disorder)5455 406

In the present work instead of a Lβ phase we observe flat crystals below Tm1 and a 407

vesicle-nanotube network instead of the ldquoripplerdquo phase The latter is particularly unexpected 408

because the membrane is composed of a single-component (G-C181-OH) and no anisotropic 409

external force (osmotic hydrodynamic electric) is used to pull and stabilize the tubes In 410

addition the nanotubes are observed at fairly high temperatures above Tm1 (between about 411

60degC and 90degC) while this morphology is generally favored at low temperature (proportionality 412

with 1kT figure 5 in Ref 62) commonly below a given transition temperature characteristics 413

for each specific amphiphile27 Furthermore nanotubes of G-C181-OH seem to be stable in 414

time thus differing from what is found in electroformed vesicles78 They are also observed 415

away from phase transition events (may them be at Tm1 or Tm2) as otherwise reported before 416

for a diacetylenic derivative of phosphatidylcholine26 and many other systems27 417

A stable nanotube-vesicle network for a single-component phospholipid in the absence 418

of external driving forces is not expected and has never been reported before It could be 419

explained by the coexistence of two micro-separated phases of different rigidity This 420

hypothesis is excluded by confocal microscopy on Laurdan-labelled membranes and magic 421

angle spinning (MAS) solid-state NMR (ssNMR) experiments 422

Laurdan is a well-known fluorophore used to put in evidence the coexistence of rigid 423

and fluid domains in both model and biological membranes47ndash49 Laurdan was used here to label 424

the G-C181-OH membranes between Tm1 and Tm2 and to verify the possible micro-segregation 425

and coexistence of rigid (eg Lβ λem= 440 nm) and fluid (eg Lα λem= 490 nm) domains in 426

17

tubes and vesicles respectively A set of quantitative experiments performed on a 5 mgmL-1 427

solution summarized in Figure 5 shows an excess of fluorescence at only λem= 490 nm for 428

temperature between 90degC and 70degC (Figure 5h) The GP value which quantifies the rigidfluid 429

character of the membrane is systematically negative (Figure 5i) indicating an overall fluid 430

membrane at all temperatures GP is quantified on more than 50 regions of interest (ROIs) at 431

each temperature including vesicles and tubes alone as well as connected nanotube-vesicles 432

(Figure 5d-g) The dispersion of GP at 80degC and 70degC between -08 and -02 could suggest the 433

spurious presence of membranes with various degrees of flexibility although fluidity is always 434

predominant The combination of WAXS (broad peak d= 144 Aring-1 invariant with 435

temperature Figure S 6bc) and confocal fluorescence microscopy using Laurdan confirms the 436

single-phase of the membrane composed of G-C181-OH both in vesicles and nanotubes This 437

is also consistent with the lack of phase transitions in DSC experiments between ~85degC and 438

~60degC especially in hydrated samples We conclude that the vesicle-nanotube transition is not 439

driven by the local segregation of gel phase (Lβ) microdomains nor that the nanotubes are 440

composed of such phase 441

ssNMR is the best-suited approach to investigate the origin of the local driving force for 442

the nucleation of anisotropic inclusions of which the average orientation depends on the local 443

membrane and inclusion curvatures and was shown to promote tubing62 ssNMR can provide 444

information on the local molecular mobility eg through the nuclear spin relaxation behavior 445

but also on intermolecular interactions and stereochemistry eg through variations in the 446

chemical shift68ndash70 447

Ideally ssNMR should be performed on diluted G-C181-OH samples above Tm1 448

However fast MAS necessary for high resolution centrifuges the sample As a first 449

approximation hydrated G-C181-OH samples for instance at 50 wt as classically done for 450

phospholipids71 could be studied Unfortunately permeability of the sample holder combined 451

to heating (close to the boiling point of water) and MAS provoke dehydration and even sample 452

ejection After experiencing both of the above this approach was abandoned As a second 453

approximation the study of the dehydrated G-C181-OH powder avoids centrifugation and 454

evaporation If studying dried lipids may not reflect their exact behaviour under diluted 455

conditions valuable structural information can still be obtained in dehydrated molecular 456

systems like proteins thus illustrating that this approach is not unreasonable Furthermore if 457

DSC (Figure 2a) shows that hydration suppresses the Tm2 it also shows that the value of Tm1 is 458

not affected by water at concentrations as low as 10 wt For this reason performing a 459

structural analysis of dehydrated G-C181-OH by ssNMR above Tm1 may help understanding 460

18

the local arrangement of G-C181-OH in two morphologies vesicles and nanotubes having 461

sensibly different curvatures 462

To explore the local mobility of the most relevant chemical groups of G-C181-OH 463

above Tm2 (T= 110degC) and between Tm2 and Tm1 (70 lt T degC lt 90) we combine a series of 1H 464

and 13C experiments Single-pulse (SP) MAS experiments (1H and 13C) when recorded under 465

quantitative conditions allow the detection of all chemical species while cross-polarization 466

(CP) MAS (13C only) experiments are sensitive only to constrained groups for which the 1H-467

13C dipolar coupling is not averaged to zero by the molecular motions 468

469

Figure 6 ndash a) 1H MAS (10 kHz) ssNMR spectra recorded from 110degC to 70degC b) 13C MAS (10 kHz) ssNMR 470

spectra recorded using high power decoupling single pulse (SP) and cross-polarization (CP) pulse sequences 471

Contact time for CP experiments and corresponding temperatures are given on each spectrum Full spectra 472

are given in Figure S 7 473

The most relevant 1H and 13C spectra are given in Figure 6 (full range in Figure 7 and 474

Figure S 7) As expected the 1H solid-state SPMAS spectra show broader peaks compared to 475

the solution-state NMR spectrum (Figure 6a) reflecting the obvious solid-state character of the 476

sample at any temperature The sugar group and the C=C bond resonate in the 25-6 ppm range 477

and 535 ppm respectively while the OH groups of the sugar moiety are found below 45 ppm 478

(Figure 6a full assignment in Table S 1) Very broad (~1 ppm) at 100degC the position of the 479

OH peak varies with the temperature value and becomes narrower (lt~05 ppm) and visible at 480

δH= 464 ppm at T= 70degC Line narrowing and chemical shift variations of the OH resonance 481

in ssNMR is always a direct sign of H-bonding interactions whereas the strength of the H-482

bonding hence its OHhellipH distance is correlated to a dispersion in chemical shifts72ndash75 All other 483

60 55 50 45 40 35 30 25

ppm

110

90

80

70

910

1rsquo

6rsquo16

1

OH

134 132 130 128 126

ppm110 108 106 104 102 100

ppm

SP (T= 110degC)

SP (T= 70degC)

CP 50 ms (T= 70degC)

CP 3 ms (T= 70degC)

910 1rsquo

1H 13C

1

811

910

CH2 CH2

1rsquo

6rsquo

a) b)

C102

13024

C9101

13003

C92

12869 C1rsquo210382

C1rsquo110443

C1rsquo13035

2rsquo3rsquo4rsquo

T degC

5rsquo

19

resonances broaden from 110degC to 70degC thus indicating slower andor more anisotropic 484

reorientational motions of all chemical groups The specific dynamic behaviour of the sugar 485

headgroup is also reflected by its 1H transverse relaxation time T2 whereas the shorter the T2 486

the higher the rigidity Experiments performed between 110degC and 70degC shown and 487

commented in Figure S 8 in the Supporting Information globally show a single short (lt 1 ms 488

below 100degC) T2 relaxation component for H1rsquo H5rsquo and H6rsquo and two both short and long 489

(between 4 ms and 12 ms) T2 for the acyl chain (CH2 H910 H811) even below 100degC 490

At 110degC 13C SPMAS experiments (Figure 6b full range in Figure 7 and Figure S 7 491

assignments in Table 1) show a single peak at δC= 10335 ppm for C1rsquo and at δC= 13003 ppm 492

for C910 as expected for a symmetrical conformation with two chemically-equivalent 493

carbons76 At 70degC the C1rsquo peak disappears while C910 now shows two peaks around 128 494

ppm and 130 ppm (Table 1 full temperature series in Figure S 7b) attributed to C92 (δC= 12869 495

ppm) and C10 in a non-symmetric configuration respectively for which these two carbons are 496

chemically inequivalent76 More precisely the resonance at 130 ppm is actually the 497

superposition of two close resonances C9101 (δC= 13003 ppm) and C102 (δC= 13024 ppm) 498

where subscripts 1 and 2 stand for two different conformations of the cis-olefin group as 499

reported for various phases of oleic acid in the vicinity of 0degC76 500

501

Table 1 ndash 13C Chemical shifts of G-C181-OH in solution and solid-state Solid-state experiments are 502

recorded at MAS= 10 kHz All spectra are reported in Figure 7 and Figure S 7 The contact time for cross 503

polarization is 3 ms In all ssNMR experiments temperature is first raised at 110degC and then lowered to 504

either 90degC or 70degC Attribution of the 13C resonances in Table 1 is performed on the basis of the solution 505

spectrum (Figure S 1b) Additional peaks observed in both single pulse and cross polarization ssNMR 506

experiments are attributed when possible to a configuration effect of the given atom eg C9101 and C92 507

indicate two configurations of C9 All new peaks observed in solid-state and not detected in solution are 508

highlighted in red for convenience 509

510

511

δ ppm

Peak Liquid (DMSO-d6)

Solid Single Pulse

T= 110degC

Solid Single Pulse

T= 90degC

Solid Single Pulse

T= 70degC

Solid Cross Polarization

T= 70degC

C102

- - 13024 13024

C910C9101 13009 13003 13003 13003 13003

C1C9C10C1rsquo

C6rsquo

20

C92 - - - 12869 12869

C1rsquoC11 10329 10335 10335 - 10443

C12 - - - - 10382

C3 7724 7709 7702 - 7829

- 7634 7629 - 7621

7525

C2 7388 7398 7390 7370 7370

C4 7052 7077 7054 7041 7098

C16 6900 7016 - - 7029

- - - - 6793

C1C5C6 6153 6241 6236 6239 6262

C1C5C6 6117 - - - 5956

C2 3300 3290 3290 3283 3279

CH2(trans) - - - 3198 3196

CH2(gauche)C15 2941 2981 2987 3004 -

C811 2706 2747 2747 2832 2817

CH2 2597 2625 2628 2630 2817

512

13C CPMAS spectra at 70degC (Figure 6b full range in Figure 7 and Figure S 7d) at 513

increasing contact times confirm the presence of two cis-olefin environments one for which 514

C910 are equivalent labelled here C9101 and one for which they are inequivalent labelled 515

C92 C102 Both environments are rigid enough to be observed at short contact time (tc= 50 μs) 516

A similar double environment is also observed for the sugar C1rsquo CPMAS at 70degC shows two 517

peaks (C1rsquo1 at δC= 10443 ppm C1rsquo2 at δC= 10382 ppm Figure 6b Figure S 7d) These are 518

significantly shifted downfield from the C1rsquo peak observed at 110degC and are detected at contact 519

times as short as 50 μs thus indicating their rigid character One also finds that CPMAS is 520

totally inefficient at 100degC for short contact times (50 μs) and poorly efficient even at 3 ms 521

typical of fast-tumbling molecules (Figure S 7c) also both gauche (δC= 3004 ppm) and all-522

trans (δC= 3198 ppm) conformations of the CH2 carbons coexist at 70degC (Figure S 7b)77 523

According to ssNMR one main mobile molecular environment exists at 110degC as 524

shown by the well-resolved 1H spectrum single C=C (C9101) and anomeric (C1rsquo) resonances 525

gauche conformation of the aliphatic CH2 and single T2 of the aliphatic CH2 chain and sugar 526

protons The sugar group is certainly mobile (inefficient 1H13C CPMAS) but still more 527

constrainted than the acyl chain (shorter T2 (1H)) As the temperature is lowered the overall 528

rigidity increases (broadening of 1H NMR peaks faster 1H transverse relaxation and efficient 529

1H13C magnetization transfer during CPMAS) presumably driven by H-bonding 530

interactions (downfield shift and line narrowing of the OH 1H signal very short OH T2 loss of 531

the C1rsquo signal in SPMAS) At the same time a number of spectroscopic data (two concomitant 532

21

resonances for anomeric C1rsquo1 and C1rsquo2 cis-olefin C9101 and C92 C102 and CH2 gauche and 533

all-trans environments) and relaxation behaviour (fast and slow T2 relaxation components) 534

strongly suggests the coexistence of two molecular environments Further proofs of the 535

coexistence of two distinct molecular domains are given by a number of chemical shift 536

variations in the 13C CPMAS spectrum at 70degC compared to the one at 110degC 90degC and 537

solution This is shown in Figure 7 and in the list of 13C chemical shifts in Table 1 538

539

540

Figure 7 - Comparison between solution- and solid-state 13C NMR spectra of G-C181-OH Solid-state 541

experiments are recorded directly on the sample powder at a MAS of 10 kHz (4 mm zirconia rotor) SP 542

Single Pulse CP Cross Polarization In all the ssNMR experiments temperature is first raised at 110degC 543

and then lowered to either 90degC or 70degC Segmented lines are meant to highlight the known peak position 544

in solution thus stressing the strong chemical shift variations in the solid-state 545

546

NMR experiments above Tm2 suggest that vesicles could contain G-C181-OH with a 547

single family of mobile conformations characterized by chemically-equivalent C=C (red 548

Figure 8b) and a single C1rsquo conformation (green Figure 8b) Experiments between Tm1 and Tm2 549

otherwise suggest that two coexisting conformations characterized by chemically-equivalent 550

(red Figure 8b) and inequivalent (magenta Figure 8b) C=C could be associated to nanotubes 551

Two distinct C1rsquo sites (grey Figure 8b) also indicate two configurations of the sugar headgroup 552

A comment on this point will be given in the following paragraphs 553

The local (Aring-scale) variation of molecular mobility below Tm2 drives the coexistence of 554

two distinct molecular environments of the same G-C181-OH molecule This occurs within a 555

single-phase domain (as deduced by WAXS) of equivalent membrane rigidity (as derived from 556

Laurdan probe in confocal microscopy) Although never reported before the difference in the 557

local mobility of the same amphiphile (here G-C181-OH) could satisfy the theoretically-558

140 120 100 80 60 40 20

CP(3 ms)-70degC

SP-70degC

SP-90degC

SP-110degC

ppm

LiquidC910

C1rsquo

C2rsquo

C4rsquo

C3rsquo

C1

C6rsquoC5rsquo

C1

6

C8

11

C2 C

H2

C1

5

CH

2

DM

SO

13C

22

required conditions that induce local lipid segregation and formation of membrane inclusions 559

(Figure 8c) In case of their non-zero average orientation the local membrane curvature 560

increases62 and nanotubes can spontaneously form (Figure 8bc) More intriguing this 561

mechanism occurs at least 10degC above Tm1 although it should not be favored above the Tm27 562

From both experimental and theoretical backgrounds the conditions of topologically-563

connected vesicles-nanotubes systems are met only when either internal (two components) or 564

external driving forces are applied19 In the present single-lipid system of which the phase 565

behavior is summarized in Figure 8a the most plausible driving force could be explained by 566

the NMR results collected on the dry sample G-C181-OH concomitantly occupies two 567

molecular environments characterized by 1) two distinct conformations of the sugar 568

displaying equivalent mobility and 2) two conformations of the acyl chain unsaturation 569

570

23

571

Figure 8 ndash a) Thermotropic phase behavior of G-C181-OH at 5 mgmL-1 Note that Tm2 is observed on the 572

dry powder only b) Possible relationship between the molecular environment as derived from NMR and 573

membrane curvatures All the molecular drawings refer to the same G-C181-OH molecule (bulky ellipsoid 574

equiv sugar headgroup) while each color corresponds to a given configuration Grey of C1rsquo1 and C1rsquo2 indicates 575

the uncertainty about the assignment of the specific NMR peak The curvature C is defined as 1R R being 576

the curvature radius R1 R2 radius of curvature CG Gauss curvature CM Mean curvature c) Free 577

interpretation of the anisotropic high-curvature (CMgt 0 CG= 0) inclusions in the membrane (low CM and 578

CG) continuum driven by inter-sugar H-bonding and reduced intramembrane diffusion for a G-C181-OH 579

glucolipid 580

581

Flat crystalline structure

Nanotube-vesicle network

Vesicles

48degC

120deg

106 105 104 103 102 101

ppm

H9 H10H1rsquo

106 105 104 103 102 101

ppm106 105 104 103 102 101

ppm106 105 104 103 102 101

ppm

C=CC=C

C1rsquo1C1rsquo2

C1rsquo

Tm2(96degC)Tm1

a)

b)

c)

Same molecule different local orientations and sugar conformations

High curvature membrane

Low curvature inclusion

Flip-FlopStrong H-bonding

CG

CM ( + )

CG 0

CM 0 CG 0

CM 0

G-C181-OH

24

Two related questions are unanswered at the moment What actually drives nanotube 582

formation and is the organization of G-C181-OH within the vesicles and tubes the same Lipid 583

nanotubes have been studied for at least three decades over which broad experimental and 584

theoretical insight has been generated Despite this fact the actual understanding is still matter 585

of debate27 Micelle-to-nanotube and vesicle-to-nanotube phase transitions are generally 586

reported for single amphiphiles below a given Tm and intermediate phase composed of twisted 587

and flat fibers is concomitantly reported Initially thought to be a chirality-driven phenomenon 588

recent studies propose a more general mechanism involving crystallization and symmetry 589

breaking in the lipid packing driven by optimal packing27 If twisted ribbons were observed 590

before for a number of microbial glycolipid bolaamphiphiles44 they are not found for G-C181-591

OH This fact as well as the large body of cryo-TEM images and the observation of nanotubes 592

above Tm1 rather pledge in favour of a direct topological change from positive (vesicles) to 593

zero (nanotubes) Gaussian curvatures In this case vesicle-to-nanotube transitions could be 594

driven by external forces56307813ndash162829 595

Considering the absence of external forces in the present system there are only two 596

possibilities left spontaneous budding of the membrane or nanotube pulling after inter-vesicle 597

collisions due to Brownian motion Both are possible as inclusions composed of a different 598

arrangement of the same molecule in the membrane continuum can theoretically explain 599

spontaneous nanotube formation in an isotropic medium672223629ndash1218ndash21 However the second 600

one is probably more likely to prevail because it could explain the large number of nanotubes 601

connecting two distant vesicles Interestingly similar events occur for any phospholipid-based 602

vesicular solution but nanotubes are extremely rare unless at least two lipids of different 603

molecular structure are mixed together The spectroscopic 1H and 13C and relaxation (1H) 604

ssNMR data even if collected on a dry powder demonstrate that the coexistence of two G-605

C181-OH configurations involving both the sugar headgroup and fatty alcohol chain are 606

possible for this molecule As a matter of fact this is not so surprising as more than one energy 607

minima each corresponding to a given conformation are known for glucose its non-reducing 608

derivative methylglucose and carbohydrates in general78ndash80 609

The presence of two configurations could explain the coexistence between two 610

structures of different curvature but the data presented so far do not explain the origin of these 611

two configurations In other words what drives the change in the local structure of the sugar 612

headgroup and fatty alcohol within the membrane 613

The association of known self-assembled structures to spectroscopic 13C ssNMR data is 614

known for years in particular for gluconamides lipids having a linear sugar headgroup81ndash84 It 615

25

was found that gauche effects in some cases up to the γ position in the sugar chain induce 616

upfield chemical shifts while anti and trans conformations result in downfield shifts Each 617

structure (monolayer crystal bilayer crystal or micellar fiber) was then characterized by a given 618

organization of the gluconamide and a characteristic 13C ssNMR fingerprint which could be 619

explained by gauche γ-gauche anti and trans conformations of the headgroup Similar 620

structural-spectroscopic correlations have been collected over the years for much simpler cyclic 621

sugars85 whereas pyranoses can undergo important conformational changes (chair half-chair 622

boat skew) when the appropriate energy barriers are crossed78ndash8086 A quantitative 623

conformational study based on the 13C chemical shift variations shown in Table 1 would be 624

certainly helpful but very delicate in the context of the present state of the art because a 625

rigorous approach requires at least one known crystal structure8182 which is currently missing 626

for this compound A qualitative approach will then be adopted 627

G-C181-OH is a bolaform amphiphile and flip-flop could be a possible mechanism 628

involved in the nanotube stabilization whereas head-tail interactions have long been considered 629

as important in nanotube-forming bolaform amphiphiles6063 From mere considerations based 630

on T2 (1H) relaxation arguments one could exclude head-tail configurations if the R-CH2OH 631

endgroup (H1) was involved in a direct interaction with the sugar group its corresponding T2 632

would probably be comparable to the sugarrsquos one (OH and H1rsquo) but this does not seem to be 633

the case However this hypothesis would be based on the comparison of the relaxation from 634

CH and CH2 protons additionally involved in different kind of motions Besides the relaxation 635

of H1 is influenced by the strong overlap between H1 and H5rsquo For these reasons relaxation 636

arguments are not solid enough and will not be used 637

We then address this question by exploring intermolecular interactions in a direct 638

manner using 2D 1H-1H homonuclear ssNMR correlation spectroscopy At 90degC in the fast-639

tumbling domain of molecular motion (narrow 1H NMR peaks Figure 6a) and corresponding 640

to the vesicle-rich region spin diffusion (NOESY) experiments are preferred over dipolar 641

coupling based experiments because such interaction is averaged out by molecular motion 642

However when the molecular motion is more constrained in the domain of nanotube stability 643

at 70degC (broad 1H spectra Figure 6a) Double-Quantum Single-Quantum (DQ-SQ) correlation 644

experiments are best suited to probe interaction between close molecular neighbors These 645

experiments are reported in Figure 9 and the related interpretation is only briefly reported 646

below while a more detailed discussion can be found on Page 12 in the Supporting Information 647

26

648

Figure 9 ndash a) 2D 1H-1H NOESY NMR experiment performed at MAS= 10 kHz (4 mm probe) on a Bruker 649

300 MHz Avance III HD spectrometer using 300 ms of mixing time Sample temperature is regulated to 650

90degC reached by cooling from 110degC b) 2D 1H-1H DQ-SQ (BAck-to-BAck BABA pulse sequence used for 651

the DQ excitation scheme) NMR experiment performed at MAS= 22 kHz (25 mm probe) on a Bruker 700 652

MHz Avance III spectrometer using a DQ excitation time equal to 2 rotor periods Sample temperature is 653

regulated to 70degC reached by cooling from 110degC The typical schemes illustrating how to interpret these 654

2D experiments are reported above each experimental map Zooms of the sugar region and configurational 655

interpretation are given below each 2D map More information are given in the text below 656

H16-H16

H1-H1

H9-H10

H1rsquo-H1

H1rsquoH910 H16

H1

H1rsquo-H1rsquo

Hxrsquo-Hxrsquo

2

(pp

m)

(ppm)

315

811910 1rsquo

2rsquo2

3rsquo 4rsquo6rsquo

5rsquo

1rsquo2910

asymp asymp asympHDO MeOD Acetone

1

315

811 9101rsquo

2rsquo2

3rsquo4rsquo6rsquo

5rsquo

1rsquo2 9 10

asympasympasympHDOMeODAcetone

H1rsquo-H1

1rsquo

1

(p

pm

)

(ppm)

H1-H2

H1rsquo-H1

H1rsquo

H910

H1H5rsquo

H2

H1rsquo-H5rsquo

315

811910 1rsquo

2rsquo2

3rsquo 4rsquo6rsquo

5rsquo

1rsquo2910

asymp asymp asympHDO MeOD Acetone

1

H2rsquoH3rsquoH4rsquo

H2rsquoH3rsquoH4rsquo

X= 2rsquo3rsquo4rsquo

NOESY - T= 90degC

1H

1H

1H

1H

DQ-SQ - T= 70degC

a) b)

2

(pp

m)

(p

pm

)

(ppm)

A B

AhellipB

AhellipA

BhellipB

NOESY

(ppm)

A B

AhellipB

DQ-SQ

Single-quantum dimension

Do

ub

le-q

ua

ntu

m d

imen

sio

n

AhellipA

BhellipB

BhellipA

BhellipA

315

811910 1rsquo

2rsquo2

3rsquo 4rsquo6rsquo

5rsquo

1rsquo2910

asymp asymp asympHDO MeOD Acetone

1

lt 5 Aring

lt 5 Aring

H1-H2

H811

H15

H16

H6rsquo

H1rsquo

H1

H5rsquo

H2

H1-H2

H1rsquo-H1

H1rsquo-H5rsquo

H1rsquo

H1 H5rsquo

H1rsquo

H1

H5rsquo

H1H2rsquo H3rsquo H4rsquo

H1-H1

H1rsquo-H1

Hxrsquo-HxrsquoX= 2rsquo3rsquo4rsquo

H1-H2

H1rsquo

Zoom sugar region 28 ndash 45 ppmNOESY DQ-SQ

27

Head-tail or head-head configurations could be probed by the existence of H1rsquo-H1 657

interactions whereas H1 (δH= 356 ppm) is associated to the glucolipid tail and H1rsquo (δH= 432 658

ppm) to the sugar headgroup These atoms are more than 15 Aring apart and a direct interaction 659

could not be observed in both NOESY and DQ-SQ experiments NOESY experiments (Figure 660

9a) show that H1rsquo-H1 correlations are not observed Meanwhile clear-cut direct H1rsquo-H1 661

interactions are indeed observed are detected at T= 70degC (DQ cross-peaks related to H1rsquo-H1 662

correlation in DQ-SQ experiment shown in Figure 9b) Interestingly the absence of auto peaks 663

at δH= 432 ppm and δH= 335 ppm in the SQ dimension excludes H1rsquo-H1rsquo and Hxrsquo-Hxrsquo (x= 2 664

3 4) interactions respectively strengthening the idea that sugar headgroups from two nearest-665

neighbor G-C181-OH molecules are not adjacent in the nanotube-vesicle region 666

Homonuclear 2D ssNMR supports the idea according to which G-C181-OH can 667

undergo molecular flip-flops and a head-tail arrangement in the dry product Extrapolating these 668

conclusions in the lower concentration range lack of H1rsquo-H1 proximity at T= 90degC suggests a 669

head-head configuration in the vesicle-rich region and head-tail configuration within the 670

nanotubes It is however still unclear whether flip-flops are driven by spontaneous temperature-671

induced conformational changes in the glucolipid packing stimulating inclusions within the 672

membrane itself672223629ndash1218ndash21 or by defects generated by inter-vesicle collisions In fact a 673

combination of both is not unlikely Nanotubes are not stable when temperature is 20 to 30 674

degrees above Tm1 indicating that inter-vesicle collisions alone cannot explain the unusual 675

phenomenon In an energy landscape approach87 the head-head configuration is certainly 676

favored at high temperature while both head-head and head-tail configurations could be 677

favored in the mid-temperature region corresponding to the range between Tm1 and Tm2 in the 678

dry powder Both statistical and defect-induced flip-flop events contribute to form high mean 679

zero Gaussian curvature morphologies (nanotubes) for which the head-tail conformation 680

limiting the proximity between bulky sugar headgroups is probably more adapted (Figure 8c) 681

The flip-flop interpretation now reasonably explains the strong differences in the sugar and fatty 682

alcohol conformations deduced by the multiple NMR signals attributed to chemically-683

equivalent groups The existence of various low-energy conformations each having specific 684

NMR signature for sugars in solution78ndash8086 and solid-state81ndash84 further supports our 685

interpretation 686

One last point should be clarified Why between Tm1 and Tm2 are two molecular 687

environments able to coexist instead of a single one of average rigidity as found in the case of 688

ldquoripplerdquo phases50ndash54 We believe that this question can be answered based on the NMR results 689

the relatively short T2 values of the H1rsquo and OH protons shows the constrained mobility of the 690

28

sugar headgroup despite a certain flexibility of the chain (long T2 component) This trend may 691

suggest restrictions in translational mobility probably due to the sugar-OH hydrogen bonding 692

(Figure 8c) meaning that lipids are not able to diffuse within the membrane thus keeping two 693

different configurations stable 694

695

Conclusions 696

Topologically-connected nanotube-vesicle lipidic systems are commonly observed for 697

self-assembled membranes composed of at least two amphiphiles and spontaneously driven by 698

anisotropic inclusions andor by external directional forces This work shows that strong 699

differences in membrane curvature can spontaneously occur in single-lipid systems without 700

external anisotropic forces Coexistence and interconnection between nanotubes and vesicles 701

are evidenced via cryo-TEM and fluorescence confocal microscopy in a single-phase region at 702

temperature above a transition temperature Tm1= 483degC determined by DSC on the hydrated 703

sample but below the second transition temperature Tm2 (957degC) determined by DSC on the 704

dry powder X-ray diffraction (WAXS) also shows the coexistence of two lattice periods one 705

attributed to an inter-membrane repeating distance and the other possibly related to the intra-706

nanotube glucolipid arrangement Labelling the systems with Laurdan a fluorophore of which 707

the emission strongly varies with the local membrane rigidity excludes the presence of rigid 708

micro-separated phases (eg Lβ) 709

1H and 13C ssNMR despite the fact that it can only be performed on the dry powder 710

confirm the presence of one main fluid G-C181-OH environment above Tm2 Below Tm2 the 711

splitting of some 13C peaks and the coexistence of short and long 1H relaxation components 712

demonstrate that G-C181-OH can simultaneously pack into structurally and dynamically 713

inequivalent environments Combining advanced 2D 1H-1H correlation experiments based on 714

spin diffusion or double-quanta excitation suggests that vesicles could contain a preferred head-715

head configuration of the bolaform lipid while nanotubes a preferred head-tail assembly 716

The presence of two environments composed of the same molecule characterized by 717

different molecular configurations and rigidity fulfills the theoretical conditions necessary to 718

spontaneously nucleate a membrane inclusion of non-zero average orientation described as 719

necessary to observe topologically-connected nanotube-vesicle structures in the absence of 720

external forces This hypothesis could also be enriched by the possibility of nanotubes to be 721

pulled away between two vesicles after a collision event 722

This work expands the actual theoretical and experimental framework on lipid networks 723

stabilized by a different radius of curvature It could also motivate the synthesis or isolation 724

29

of new lipids providing a higher level of 3D architectural control in soft networks possibly in 725

the range of room temperature 726

727

Acknowledgements 728

Dr Andrea Lassenberger and Dr Sylvain Preacutevost at Institut Laue Langevin (ILL Grenoble 729

France) are kindly acknowledged for their assistance on the SANS experiment Confocal 730

images were performed at the Institut de Biologie Paris Seine (IBPS) imaging facility Jean-731

Franccedilois Gilles (Imaging Core Facility Institut de Biologie Paris Seine (IBPS) CNRS 732

Sorbonne Universiteacute Paris France) is kindly acknowledged for his assistance on the confocal 733

microscopy imaging Dr Guillaume Laurent (LCMCP Sorbonne Universiteacute Paris France) is 734

kindly acknowledged for helpful discussions The French region Ile-de-France SESAME 735

program is acknowledged for financial support (700 MHz NMR 736

spectrometer) Dr Olivier Diat (ICSM Marcoule France) Prof eronikaIglič and Prof V Aleš737

onsul discussi Slovenia) are kindly acknowledged for helpfUniversity of LjubljanaIglič (-Kralj 738

739

Financial support 740

WAXS experiments have been supported by Soleil Light Source Saint Aubin France proposal 741

Ndeg 20201747 SANS experiments have been supported by ILL proposal Ndeg9-13-778 742

743

References 744

1 D Kabaso N Bobrovska W Goacuteźdź E Gongadze V Kralj-Iglič R Zorec and A 745

Iglič Bioelectrochemistry 2012 87 204ndash210 746

2 C R Safinya U Raviv D J Needleman A Zidovska M C Choi M A Ojeda-747

Lopez K K Ewert Y Li H P Miller J Quispe B Carragher C S Potter M W 748

Kim S C Feinstein and L Wilson Adv Mater 2011 23 2260ndash2270 749

3 P Veranič M Lokar G J Schuumltz J Weghuber S Wieser H Haumlgerstrand V Kralj-750

Iglič and A Iglič Biophys J 2008 95 4416ndash4425 751

4 A Rustom R Saffrich I Markovic P Walther and H-H Gerdes Science (80- ) 752

2004 303 1007ndash1010 753

5 M I Angelova S Soleacuteau P Meacuteleacuteard F Faucon and P Bothorel Prog Colloid 754

Polym Sci 1992 89 127ndash131 755

6 L Mathivet S Cribier and P F Devaux Biophys J 1996 70 1112ndash1121 756

7 V Kralj-Iglič G Gomišček J Majhenc V Arrigler and S Svetina Colloids Surfaces 757

A Physicochem Eng Asp 2001 181 315ndash318 758

30

8 V Kralj-Iglič A Iglič G Gomišček F Sevšek V Arrigler and H Haumlgerstrand J 759

Phys A Math Gen 2002 35 1533ndash1549 760

9 M Drab D Stopar V Kralj-Iglič and A Iglič Cells 2019 8 626 761

10 A Zidovska K K Ewert J Quispe B Carragher C S Potter and C R Safinya 762

Langmuir 2009 25 2979ndash2985 763

11 A Zidovska K K Ewert J Quispe B Carragher C S Potter and C R Safinya Soft 764

Matter 2011 7 8363ndash8369 765

12 A Zidovska K K Ewert J Quispe B Carragher C S Potter and C R Safinya 766

Biochim Biophys Acta - Biomembr 2009 1788 1869ndash1876 767

13 M Karlsson K Sott M Davidson A S Cans P Linderholm D Chiu and O Orwar 768

Proc Natl Acad Sci U S A 2002 99 11573ndash11578 769

14 A Karlsson R Karlsson M Karlsson A-S Cans A Stromberg F Ryttsen and O 770

Orwar Nature 2001 409 150ndash152 771

15 M Karlsson K Sott A S Cans A Karlsson R Karlsson and O Orwar Langmuir 772

2001 17 6754ndash6758 773

16 E Evans H Bowman A Leung D Needham and D Tirrell Science (80- ) 1996 774

273 933ndash935 775

17 U Seifert Configurations of fluid membranes and vesicles 1997 vol 46 776

18 J Kaumls and E Sackmann Biophys J 1991 60 825ndash844 777

19 N Bobrovska W Goacuteźdź V Kralj-Iglič and A Iglič PLoS One 2013 8 e73941 778

20 A Iglič B Babnik K Bohinc M Fošnarič H Haumlgerstrand and V Kralj-Iglič J 779

Biomech 2007 40 579ndash585 780

21 K Bohinc T Slivnik A Iglič M Brumen and V Kralj-Iglič J Phys Chem C 2007 781

111 9709ndash9718 782

22 A Iglič H Haumlgerstrand P Veranič A Plemenitaš and V Kralj-Iglič J Theor Biol 783

2006 240 368ndash373 784

23 V Kralj-Iglič and P Veranič Adv Planar Lipid Bilayers Liposomes 2006 5 129ndash785

149 786

24 A Iglic and V Kralj-Iglic Adv Planar Lipid Bilayers Liposomes 2006 4 253ndash279 787

25 S Penič L Mesarec M Fošnarič L Mroacutewczyńska H Haumlgerstrand V Kralj-Iglič and 788

A Iglič Front Phys 2020 8 342 789

26 P Yager and P E Schoen Mol Cryst Liq Cryst 1984 106 371ndash381 790

27 T G Barclay K Constantopoulos and J Matisons Chem Rev 2014 114 10217ndash791

10291 792

31

28 B Božič G Gomišček V Kralj-Iglič S Svetina and B Žekš Eur Biophys J 2002 793

31 487ndash496 794

29 A Iglič H Haumlgerstrand M Bobrowska-Haumlgerstrand V Arrigler and V Kralj-Iglič 795

Phys Lett A 2003 310 493ndash497 796

30 R Bar-Ziv and E Moses Phys Rev Lett 1994 73 1392ndash1395 797

31 S Chiruvolu H E Warriner E Naranjo S H Idziak J O Raumldler R J Plano J A 798

Zasadzinski and C R Safinya Science (80- ) 1994 266 1222ndash1225 799

32 V Kralj-Iglič A Iglič H Haumlgerstrand and P Peterlin Phys Rev E 2000 61 4230ndash800

4234 801

33 L Mesarec M Drab S Penič V Kralj-Iglič and A Iglič Int J Mol Sci 2021 22 802

2348 803

34 L Van Renterghem S L K W Roelants N Baccile K Uyttersprot M C Taelman 804

B Everaert S Mincke S Ledegen S Debrouwer K Scholtens C V Stevens W 805

Soetaert L Van Renterghem S L K W Roelants N Baccile K Uytersprot M C 806

Taelman B Everaert S Mincke S Ledegen S Debrouwer K Scholtens C V 807

Stevens and W Soetaert Biotechnol Bioeng 2018 115 1195ndash1206 808

35 J Clark and F Deswarte Introduction to Chemicals from Biomass Second Edition 809

John Wiley amp Sons 2015 810

36 K V Sajna R Houmlfer R K Sukumaran L D Gottumukkala and A Pandey in 811

Industrial Biorefineries and White Biotechnology eds A Pandey R Houmlfer M 812

Taherzadeh K M Nampoothiri and C Larroche Amsterdam Oxford Waltham 813

Elsevier 2015 pp 499ndash521 814

37 R Marchant and I M Banat Trends Biotechnol 2012 30 558ndash565 815

38 K Hill and C LeHen-Ferrenbach in Sugar-Based Surfactants Fundamentals and 816

Applications ed C C Ruiz Boca Raton CRC Press 2009 pp 1ndash20 817

39 L Rodrigues I M Banat J Teixeira and R Oliveira J Antimicrob Chemother 818

2006 57 609ndash618 819

40 J D Desai and I M Banat Microbiol Mol Biol Rev 1997 61 47ndash64 820

41 S L K W Roelants K Ciesielska S L De Maeseneire H Moens B Everaert S 821

Verweire Q Denon B Vanlerberghe I N A Van Bogaert P Van der Meeren B 822

Devreese and W Soetaert Biotechnol Bioeng 2016 113 550ndash559 823

42 N Baccile M Selmane P Le Griel S Preacutevost J Perez C V Stevens E Delbeke S 824

Zibek M Guenther W Soetaert I N A Van Bogaert and S Roelants Langmuir 825

2016 32 6343ndash6359 826

32

43 N Baccile A-S Cuvier S Preacutevost C V Stevens E Delbeke J Berton W Soetaert 827

I N A Van Bogaert and S Roelants Langmuir 2016 32 10881ndash10894 828

44 N Baccile C Seyrig A Poirier S A Castro S L K W Roelants and S Abel Green 829

Chem 2021 23 3842ndash3944 830

45 A Salis L Cappai C Carucci D F Parsons and M Monduzzi J Phys Chem Lett 831

2020 11 6805ndash6811 832

46 J Schindelin I Arganda-Carreras E Frise V Kaynig M Longair T Pietzsch S 833

Preibisch C Rueden S Saalfeld B Schmid J-Y Tinevez D J White V 834

Hartenstein K Eliceiri P Tomancak and A Cardona Nat Methods 2012 9 676ndash835

682 836

47 T Parasassi and E Gratton J Fluoresc 1995 5 59ndash69 837

48 M M Dodes Traian F L Gonzaacutelez Flecha and V Levi J Lipid Res 2012 53 609ndash838

616 839

49 D M Owen C Rentero A Magenau A Abu-Siniyeh and K Gaus Nat Protoc 840

2012 7 24ndash35 841

50 W Stillwell in An Introduction to Biological Membranes Academic Press 2016 pp 842

135ndash180 843

51 W Stillwell in An Introduction to Biological Membranes Academic Press 2016 pp 844

181ndash220 845

52 B Tenchov Chem Phys Lipids 1991 57 165ndash177 846

53 R Koynova and M Caffrey Biochim Biophys Acta 1998 1376 91ndash145 847

54 F Schmid S Dolezel O Lenz and S Meinhardt J Phys Conf Ser 2014 487 848

012004 849

55 S Meinhardt R L C Vink and F Schmid Proc Natl Acad Sci U S A 2013 110 850

4476ndash4481 851

56 O Glatter and O Kratky Small Angle X-ray Scattering Academic Press London 852

1982 853

57 B A Cunningham A D Brown D H Wolfe W P Williams and A Brain Phys 854

Rev E 1998 58 3662ndash3672 855

58 M Nadler A Steiner T Dvir O Szekely P Szekely A Ginsburg R Asor R Resh 856

C Tamburu M Peres and U Raviv Soft Matter 2011 7 1512ndash1523 857

59 C Tanford The hydrophobic effect Formation of micelles and biological membranes 858

John Wiley amp Sons Inc New York 1973 859

60 M Masuda and T Shimizu Langmuir 2004 20 5969ndash5977 860

33

61 A Zidovska K K Ewert J Quispe B Carragher C S Potter and C R Safinya 861

Block Liposomes Vesicles of Charged Lipids with Distinctly Shaped Nanoscale 862

Sphere- Pear- Tube- or Rod-Segments Elsevier Inc 1st edn 2009 vol 465 863

62 A Iglic V Kralj-Iglic B Božič M Bobrowska-Haumlgerstrand B Isomaa and H 864

Haumlgerstrand Bioelectrochemistry 2000 52 203ndash211 865

63 T Shimizu M Masuda and H Minamikawa Chem Rev 2005 105 1401ndash1444 866

64 J V Selinger and J M Schnur Phys Rev Lett 1993 71 4091ndash4094 867

65 J M Schnur Science (80- ) 1993 262 1669ndash1676 868

66 B Tenchov R Koynova and G Rapp Biophys J 2001 80 1873ndash1890 869

67 R Koynova and B Tenchov Ser Biomech 2017 31 3ndash19 870

68 S J Angyal and R Le Fur Carbohydr Res 1984 126 15ndash26 871

69 J K Whitesell and M A Minton J Am Chem Soc 1987 109 225ndash228 872

70 T Takayama S Ando and I Ando J Mol Struct 1990 220 245ndash250 873

71 K Gawrisch N V Eldho and I V Polozov Chem Phys Lipids 2002 116 135ndash151 874

72 F Tielens C Gervais J F Lambert F Mauri and D Costa Chem Mater 2008 20 875

3336ndash3344 876

73 G A Jeffrey and Y Yeon Acta Crystallogr Sect B 1986 42 410ndash413 877

74 B Berglund and R W Vaughan J Chem Phys 1980 73 2037ndash2043 878

75 C Gervais C Coelho T Azaiumls J Maquet G Laurent F Pourpoint C Bonhomme 879

P Florian B Alonso G Guerrero P H Mutin and F Mauri J Magn Reson 2007 880

187 131ndash140 881

76 C Akita T Kawaguchi F Kaneko H Yamamoto and M Suzuki J Phys Chem B 882

2004 108 4862ndash4868 883

77 R N Purusottam L Seacutenicourt J J Lacapegravere and P Tekely Biochim Biophys Acta - 884

Biomembr 2015 1848 3134ndash3139 885

78 X Biarneacutes A Ardegravevol A Planas C Rovira A Laio and M Parrinello J Am Chem 886

Soc 2007 129 10686ndash10693 887

79 L M J Kroon‐Batenburg and J Kroon Biopolymers 1990 29 1243ndash1248 888

80 K N Kirschner and R J Woods Proc Natl Acad Sci U S A 2001 98 10541ndash889

10545 890

81 S Svenson B Kirste and J H Fuhrhop J Am Chem Soc 1994 116 11969ndash11975 891

82 I Sack S Macholl J H Fuhrhop and G Buntkowsky Phys Chem Chem Phys 892

2000 2 1781ndash1788 893

83 S Svenson J Koening and J H Fuhrhop J Phys Chem 1994 98 1022ndash1028 894

34

84 S Svenson A Schaefer and J H Fuhrhop J Chem Soc Perkin Trans 2 1994 2 895

1023ndash1028 896

85 K Bock and C Pedersen Adv Carbohydr Chem Biochem 1983 41 27ndash66 897

86 T B Grindley Structure and Conformation of Carbohydrates Springer Berlin 898

Heidelberg 2008 899

87 F Tantakitti J Boekhoven X Wang R V Kazantsev T Yu J Li E Zhuang R 900

Zandi J H Ortony C J Newcomb L C Palmer G S Shekhawat M O De La 901

Cruz G C Schatz and S I Stupp Nat Mater 2016 15 469ndash476 902

903

1

Supporting Information 904

905

Topological connection between vesicles and nanotubes in single-906

component lipid membranes driven by head-tail interactions 907

908

Niki Baccilea Ceacutedric Lorthioira Abdoul Aziz Baa Patrick Le Griela Cristina Coelhob Javier 909

Perezc Wim Soetaertd Sophie L K W Roelantsd 910

911

a Sorbonne Universiteacute Centre National de la Recherche Scientifique Laboratoire de Chimie de 912

la Matiegravere Condenseacutee de Paris LCMCP F-75005 Paris France 913

b Institut des Mateacuteriaux de Paris Centre Sorbonne Universiteacute Paris France 914

c Synchrotron Soleil LrsquoOrme des Merisiers Saint-Aubin BP48 91192 Gif-sur-Yvette Cedex 915

France 916

d InBio Department of Biotechnology Ghent University Ghent Belgium 917

918

919

Corresponding author 920

Dr Niki Baccile 921

E-mail address nikibaccilesorbonne-universitefr 922

Phone +33 1 44 27 56 77 923

924

925

926

2

a) 927

Parameter Method

Dry Matter (DM ) 996 Infrared balance 105degC

Glucose () 012 HPLC-Metacarb

Glycerol () nd HPLC-Metacarb

Free Fatty Acid content () lt001 Internal method BBEPP

GC with FID Detector

Oil () lt001 Internal method BBEPP

GC with FID Detector

Moulds (CFUg) lt10 3M-Nordval ndeg16 B

Yeasts (CFUg) lt10 3M-Nordval ndeg16 B

Anaerobic count (CFUg) lt10 SP-VG M005 B

Protein () tbd BCA protein assay

928

Figure S 1 - a) HPLC-ELSD chromatogram and composition table 929

930

931

932

933

3

b) 934

935

Group Ndeg δ ppm

CH2 1 354

CH2 2 153

CH2 3-712-14 132

CH2 811 203

CH 910 534

CH2 15 162

CH2 16 388

CH 1 424

CH 2 316

CH 3 334

CH 4 329

CH 5 354

CH2 6 366

Figure S 1 (next) ndash b) 1H NMR spectrum (MeOD-d4) recorded on the purified G-C181-OH compounds 936

and related assignment 937

938

2

811910 1rsquo

5rsquo1

3rsquo4rsquo

1rsquo1910

asymp asymp asympHDO MeOD Acetone

6rsquo2rsquo

1615

3-712-14

4

939

940

Figure S 2 ndash Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 130degC 941

5

942

943

Figure S 3 - Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 90degC 944

945

90degC

1

22

6

946

947

948

949

Figure S 4 ndash Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 70degC 950

951

21

11

1

2

5

3

4

45

5

6

7

7

7

952

953

954

Figure S 5 ndash Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 25degC 955

956

957

1

1 1

8

958

Figure S 6 ndash a) SANS experiment recorded at 90degC for a 5 mgmL G-C181-OH aqueous solution b-c) 959

WAXS experiments recorded on aqueous solutions of G-C181-OH at b) 5 mgmL and c) 25 mgmL d) 960

Structural interpretation of the repeating distances pointed out in c) 961

962

963

964

001 01 1

q Aring-1

25degC

68degC

88degC

111degC

01 1q Aring

-1

02 04 06 08 1q Aring

-1

001 01

10-1

100

101

Inte

nsity

cm

-1

q Aring-1

-2

T= 90degC

SANS WAXS WAXS

a) b) c)

dv

dL d= 144 Aring-1

di

ddH2O

dt

dvdL

dtdi d

d)

9

965

966

Figure S 7 ndash ssNMR MAS (10 kHz) experiments recorded on a G-C181-OH dry powder a) 1H spectra 967

recorded from 110degC to 70degC b) 13C single pulse (SP) spectra recorded with high power 1H decoupling from 968

110degC to 70degC c) 13C cross polarization (CP) spectra recorded at 110degC and contact time tc of 50 micros and 969

3000 micros d) 13C CP spectra recorded at 70degC and for 20 lt tc micros lt 3000 970

971

1

811

910

CH2 CH2

1rsquo

6rsquo

8 6 4 2 0 ppm

110

90

80

70

910

1rsquo

6rsquo16

1

2rsquo3rsquo4rsquo

811

OH

45 40 35 30 25 20 15

ppm

200 180 160 140 120 100 80 60 40 20 0

ppm140 120 100 80

ppm

CH2-gauche

3004

110

90

80

70

65

60

50

25

C91013003

C9 - 12869

CH2-all-trans

3198

C1rsquo10335

a) b)

1H

T degCT degC

13C Single Pulse MAS

A

B

A

B

70degC70degC

5rsquo

215

200 180 160 140 120 100 80 60 40 20 0

ppm

13C CPMAST= 110degC

200 180 160 140 120 100 80 60 40 20 0

ppm

132 131 130 129 128 127 126

ppm

106 105 104 103 102

ppm

13C CPMAST= 70degC

50

tc micros

3000

20

tc micros

50 500 1000 3000

tc micros

C102

13024 C910113003

C92

12869

C1rsquo213082

C1rsquo110443

A B

106 105 104 103 102

ppm

3000

1000

500

50

20

AB

c)

d)

10

972

Figure S 8 ndash a) Short and d) long T2 components displayed by the 1H transverse relaxation signal c) 1H 973

transverse relaxation signals recorded on a G-C181-OH dry powder from 110degC to 70degC The area of each 974

given resonance is plotted (log-lin) as a function of the echo time (CPMG pulse sequence) for each 975

temperature and fitted using the equation given above T2s and T2l indicate the short and long T2 relaxation 976

times respectively obtained by describing the 1H transverse relaxation signal using a biexponential decay 977

For H1rsquo H5rsquo and H6rsquo the 1H relaxation function could be described by a monoexponential decay 978

- Protons from the sugar group (H1rsquo H5rsquo H6rsquo) show a single T2 relaxation component (Figure 979

6c and Figure S 8) while protons of the acyl chain (CH2 H910 H811) show both short (Figure 980

6c) and long (Figure 6d) T2 relaxation component 981

- The short T2 relaxation time systematically decreases from 110degC to 70degC for all species by a 982

factor ranging from two to seven depending on the specific chemical group Very long T2 are 983

observed at 110degC for all acyl species while nearly unchanged values are measured below 984

90degC without any clear trend 985

- The sugar protons (H1rsquo H5rsquo H6rsquo) have the shortest set of T2 values (lt 1 ms) below 100degC 986

The long component of T2 generally ranges between 4 ms and 12 ms 987

0 1 2 3 4 5 6 7 8

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1 2 3 4 5 6 7 8

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

H1

H8H11

H9H10

CH2

H1rsquo

H6rsquo

CH2

H910H811 H1CH2

H1rsquo H5rsquo6rsquo

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

OH

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45 T= 110degC

T= 90degC

T= 80degC

T= 70degC

Log

In

ten

sity

Time ms

Fit

)

CH2 H910 H811 H1 H56 H1 OH

00

05

10

15

20

25

30

35

40

T2 m

s

T= 110degC

T= 90degC

T= 80degC

T= 70degC

CH2 H910 H811 H1

0

2

4

6

8

10

12

T2 m

s

T= 110degC

T= 90degC

T= 80degC

T= 70degC

Short T2(1H) Long T2(1H)

a) b)

c)

11

Table S 1 ndash 1H chemical shifts of G-C181-OH in solution and in the solid-state Solid-state spectra are 988

recorded at a MAS frequency of 10 kHz in DMSO-d6 broad peak 989

990

δH ppm

Group Ndeg Liquid

(MeOD-d4) Solid

(T= 110degC) Solid

(T= 70degC)

CH2 H1 354 357

CH2 H2 153 156

CH2 H3-712-14 132 135

CH2 H811 203 203

CH H910 534 534

CH2 H15 162 164

CH2 H16 388 383

CH H1 424 432

CH H2 316 335

CH H3 334 335

CH H4 329 335

CH H5 354 351 357

CH2 H6 366 383

OH OH1 432

45 463

OH OH2 492

OH OH3 490

OH OH4 487

OH OH6 445

991

992

H1H9H10H1rsquo

H6rsquo

OH2rsquo

OH1

12

Detailed discussion about the NOESY and DQ-SQ 2D 1H-1H ssNMR experiments presented in 993

Figure 9 994

995

2D 1H-1H ssNMR experiments are a unique tool to explore through-space proximities 996

between 1H nuclei ssNMR offers a large number of possibilities each adapted to a given 997

system The advantage of 2D maps with respect to a dynamics approach is their direct 998

readability in clear two nuclei which exchange their magnetization through-space (whichever 999

mechanism may be responsible for this exchange) will display a specific cross-peak However 1000

the way cross-peaks are presented and interpreted strictly depends on the nature of the pulse 1001

program which in turn is chosen according to the physical nature of the sample 1002

1D temperature-dependent 1H NMR experiments performed on G-C181-OH (Figure 1003

4a in the main text) indicate sharp peaks at 90degC and broad peaks at 70degC Narrow peaks 1004

indicate fast andor large-amplitude molecular tumbling averaging the strong 1H-1H dipolar 1005

interactions This excludes the use of pulse sequences like DQ-SQ which is not efficient under 1006

these conditions For this reason we employ spin-diffusion-based pulse sequences like 1007

NOESY which is more adapted for samples with high molecular mobility When mixing times 1008

are chosen to be short (lt 1 s) one generally explores through-space proximities within about 5 1009

10 Aring On the contrary when molecular motions are restricted dipolar coupling becomes 1010

significant 1H NMR peaks are broader and pulse sequences like DQ-SQ can now be used 1011

with a spatial resolution of less than 5 Aring 1012

Whenever possible DQ-SQ experiments are preferred over spin diffusion-based pulse 1013

sequences because they are more selective in particular for likewise interactions which are 1014

not discriminated in NOESY-type experiments This is illustrated by the typical interaction 1015

schemes for NOESY (top Figure 9a) and DQ-SQ (top Figure 9b) For two given A and B 1016

protons 2D NOESY maps will display two on-diagonal and two square-shaped symmetrical 1017

off-diagonal cross-peaks Off-diagonal cross-peaks illustrate the mutual A-B and B-A 1018

interaction but the on-diagonal peaks are systematic and do not prove any A-A or B-B 1019

interaction Similarly to NOESY 2D DQ-SQ maps display off-diagonal peaks (symmetric 1020

across the diagonal top Figure 9b) upon A-B and B-A interaction but in contrast they will 1021

display on-diagonal peaks only in the presence of self-interaction In the tutorial example of 1022

Figure 9b top the on-diagonal A-A peak indicates mutual A-A interaction while lack of on-1023

diagonal B-B peak evidences the absence of B-B proximities Combining NOESY and DQ-SQ 1024

experiments is capital to study the local arrangement of G-C181-OH and its evolution with 1025

temperature 1026

13

At T= 90degC the H1 and H5rsquo contributions are very close but possible to discriminate 1027

them (δ(H1)= 354 ppm δ(H5rsquo)= 351 ppm) according to 1H NMR chemical shifts in Table S 1028

1 2D 1H-1H NOESY in Figure 9a also discriminates between the two components attributed 1029

on the basis of the interactions with their mutual neighbors (green H1-H2 and H5rsquo-H1rsquo squares 1030

in Figure 9a) On this basis 2D NOESY does not show a direct H1rsquo-H1 correlation (blue red-1031

crossed) but rather an extensive network of intra-sugar interactions These features exclude 1032

suggest intra- and inter-sugar interactions and they exclude head-tail arrangements as a major 1033

molecular configuration at 90degC 1034

At 70degC the 1H NMR peaks are broader indicating a reduction of the T2 (1H) relaxation 1035

times andor residual homonuclear dipolar interactions excluding the use of NOESY In this 1036

case 1H-1H DQ-SQ correlation experiments are best suited Figure 9b presents the 1H-1H DQ-1037

SQ experiment on G-C181-OH at 70degC Peaks are broader than in NOESY and attribution may 1038

be cumbersome However the experiment is performed at 700 MHz and MAS= 22 kHz which 1039

improves spectral resolution with respect to the same experiment performed at 300 MHz and 1040

MAS= 10 kHz The auto-peak at δH= 356 ppm in the SQ dimension is attributed to H1-H1 spin 1041

pairs mainly resulting from intra CH2 interaction while the DQ cross-peaks at δH= 156-356 1042

ppm and δH= 432-356 ppm respectively origin from H2-H1 and H1rsquo-H1 interactions (Figure 1043

9b) DQ-SQ map does not show any auto-peak corresponding to H1rsquo-H1rsquo or Hxrsquo-Hxrsquo (x= 2rsquo 1044

3rsquo 4rsquo) interactions (red-crossed blue Figure 9b) thus excluding any proximity (lt 5 Aring) between 1045

close glucose neighbors On the contrary DQ-SQ maps (recorded both at 700 MHz Figure 9b 1046

or at 300 MHz not shown) systematically show DQ cross-peaks which can only be attributed 1047

to H1rsquo-H1 interactions (δH= 432-356) The presence of these H1rsquo-H1 interactions can only be 1048

explained by a head-tail arrangement of the G-C181-OH molecules as illustrated below the 1049

DQ-SQ map in Figure 9b 1050

17

tubes and vesicles respectively A set of quantitative experiments performed on a 5 mgmL-1 427

solution summarized in Figure 5 shows an excess of fluorescence at only λem= 490 nm for 428

temperature between 90degC and 70degC (Figure 5h) The GP value which quantifies the rigidfluid 429

character of the membrane is systematically negative (Figure 5i) indicating an overall fluid 430

membrane at all temperatures GP is quantified on more than 50 regions of interest (ROIs) at 431

each temperature including vesicles and tubes alone as well as connected nanotube-vesicles 432

(Figure 5d-g) The dispersion of GP at 80degC and 70degC between -08 and -02 could suggest the 433

spurious presence of membranes with various degrees of flexibility although fluidity is always 434

predominant The combination of WAXS (broad peak d= 144 Aring-1 invariant with 435

temperature Figure S 6bc) and confocal fluorescence microscopy using Laurdan confirms the 436

single-phase of the membrane composed of G-C181-OH both in vesicles and nanotubes This 437

is also consistent with the lack of phase transitions in DSC experiments between ~85degC and 438

~60degC especially in hydrated samples We conclude that the vesicle-nanotube transition is not 439

driven by the local segregation of gel phase (Lβ) microdomains nor that the nanotubes are 440

composed of such phase 441

ssNMR is the best-suited approach to investigate the origin of the local driving force for 442

the nucleation of anisotropic inclusions of which the average orientation depends on the local 443

membrane and inclusion curvatures and was shown to promote tubing62 ssNMR can provide 444

information on the local molecular mobility eg through the nuclear spin relaxation behavior 445

but also on intermolecular interactions and stereochemistry eg through variations in the 446

chemical shift68ndash70 447

Ideally ssNMR should be performed on diluted G-C181-OH samples above Tm1 448

However fast MAS necessary for high resolution centrifuges the sample As a first 449

approximation hydrated G-C181-OH samples for instance at 50 wt as classically done for 450

phospholipids71 could be studied Unfortunately permeability of the sample holder combined 451

to heating (close to the boiling point of water) and MAS provoke dehydration and even sample 452

ejection After experiencing both of the above this approach was abandoned As a second 453

approximation the study of the dehydrated G-C181-OH powder avoids centrifugation and 454

evaporation If studying dried lipids may not reflect their exact behaviour under diluted 455

conditions valuable structural information can still be obtained in dehydrated molecular 456

systems like proteins thus illustrating that this approach is not unreasonable Furthermore if 457

DSC (Figure 2a) shows that hydration suppresses the Tm2 it also shows that the value of Tm1 is 458

not affected by water at concentrations as low as 10 wt For this reason performing a 459

structural analysis of dehydrated G-C181-OH by ssNMR above Tm1 may help understanding 460

18

the local arrangement of G-C181-OH in two morphologies vesicles and nanotubes having 461

sensibly different curvatures 462

To explore the local mobility of the most relevant chemical groups of G-C181-OH 463

above Tm2 (T= 110degC) and between Tm2 and Tm1 (70 lt T degC lt 90) we combine a series of 1H 464

and 13C experiments Single-pulse (SP) MAS experiments (1H and 13C) when recorded under 465

quantitative conditions allow the detection of all chemical species while cross-polarization 466

(CP) MAS (13C only) experiments are sensitive only to constrained groups for which the 1H-467

13C dipolar coupling is not averaged to zero by the molecular motions 468

469

Figure 6 ndash a) 1H MAS (10 kHz) ssNMR spectra recorded from 110degC to 70degC b) 13C MAS (10 kHz) ssNMR 470

spectra recorded using high power decoupling single pulse (SP) and cross-polarization (CP) pulse sequences 471

Contact time for CP experiments and corresponding temperatures are given on each spectrum Full spectra 472

are given in Figure S 7 473

The most relevant 1H and 13C spectra are given in Figure 6 (full range in Figure 7 and 474

Figure S 7) As expected the 1H solid-state SPMAS spectra show broader peaks compared to 475

the solution-state NMR spectrum (Figure 6a) reflecting the obvious solid-state character of the 476

sample at any temperature The sugar group and the C=C bond resonate in the 25-6 ppm range 477

and 535 ppm respectively while the OH groups of the sugar moiety are found below 45 ppm 478

(Figure 6a full assignment in Table S 1) Very broad (~1 ppm) at 100degC the position of the 479

OH peak varies with the temperature value and becomes narrower (lt~05 ppm) and visible at 480

δH= 464 ppm at T= 70degC Line narrowing and chemical shift variations of the OH resonance 481

in ssNMR is always a direct sign of H-bonding interactions whereas the strength of the H-482

bonding hence its OHhellipH distance is correlated to a dispersion in chemical shifts72ndash75 All other 483

60 55 50 45 40 35 30 25

ppm

110

90

80

70

910

1rsquo

6rsquo16

1

OH

134 132 130 128 126

ppm110 108 106 104 102 100

ppm

SP (T= 110degC)

SP (T= 70degC)

CP 50 ms (T= 70degC)

CP 3 ms (T= 70degC)

910 1rsquo

1H 13C

1

811

910

CH2 CH2

1rsquo

6rsquo

a) b)

C102

13024

C9101

13003

C92

12869 C1rsquo210382

C1rsquo110443

C1rsquo13035

2rsquo3rsquo4rsquo

T degC

5rsquo

19

resonances broaden from 110degC to 70degC thus indicating slower andor more anisotropic 484

reorientational motions of all chemical groups The specific dynamic behaviour of the sugar 485

headgroup is also reflected by its 1H transverse relaxation time T2 whereas the shorter the T2 486

the higher the rigidity Experiments performed between 110degC and 70degC shown and 487

commented in Figure S 8 in the Supporting Information globally show a single short (lt 1 ms 488

below 100degC) T2 relaxation component for H1rsquo H5rsquo and H6rsquo and two both short and long 489

(between 4 ms and 12 ms) T2 for the acyl chain (CH2 H910 H811) even below 100degC 490

At 110degC 13C SPMAS experiments (Figure 6b full range in Figure 7 and Figure S 7 491

assignments in Table 1) show a single peak at δC= 10335 ppm for C1rsquo and at δC= 13003 ppm 492

for C910 as expected for a symmetrical conformation with two chemically-equivalent 493

carbons76 At 70degC the C1rsquo peak disappears while C910 now shows two peaks around 128 494

ppm and 130 ppm (Table 1 full temperature series in Figure S 7b) attributed to C92 (δC= 12869 495

ppm) and C10 in a non-symmetric configuration respectively for which these two carbons are 496

chemically inequivalent76 More precisely the resonance at 130 ppm is actually the 497

superposition of two close resonances C9101 (δC= 13003 ppm) and C102 (δC= 13024 ppm) 498

where subscripts 1 and 2 stand for two different conformations of the cis-olefin group as 499

reported for various phases of oleic acid in the vicinity of 0degC76 500

501

Table 1 ndash 13C Chemical shifts of G-C181-OH in solution and solid-state Solid-state experiments are 502

recorded at MAS= 10 kHz All spectra are reported in Figure 7 and Figure S 7 The contact time for cross 503

polarization is 3 ms In all ssNMR experiments temperature is first raised at 110degC and then lowered to 504

either 90degC or 70degC Attribution of the 13C resonances in Table 1 is performed on the basis of the solution 505

spectrum (Figure S 1b) Additional peaks observed in both single pulse and cross polarization ssNMR 506

experiments are attributed when possible to a configuration effect of the given atom eg C9101 and C92 507

indicate two configurations of C9 All new peaks observed in solid-state and not detected in solution are 508

highlighted in red for convenience 509

510

511

δ ppm

Peak Liquid (DMSO-d6)

Solid Single Pulse

T= 110degC

Solid Single Pulse

T= 90degC

Solid Single Pulse

T= 70degC

Solid Cross Polarization

T= 70degC

C102

- - 13024 13024

C910C9101 13009 13003 13003 13003 13003

C1C9C10C1rsquo

C6rsquo

20

C92 - - - 12869 12869

C1rsquoC11 10329 10335 10335 - 10443

C12 - - - - 10382

C3 7724 7709 7702 - 7829

- 7634 7629 - 7621

7525

C2 7388 7398 7390 7370 7370

C4 7052 7077 7054 7041 7098

C16 6900 7016 - - 7029

- - - - 6793

C1C5C6 6153 6241 6236 6239 6262

C1C5C6 6117 - - - 5956

C2 3300 3290 3290 3283 3279

CH2(trans) - - - 3198 3196

CH2(gauche)C15 2941 2981 2987 3004 -

C811 2706 2747 2747 2832 2817

CH2 2597 2625 2628 2630 2817

512

13C CPMAS spectra at 70degC (Figure 6b full range in Figure 7 and Figure S 7d) at 513

increasing contact times confirm the presence of two cis-olefin environments one for which 514

C910 are equivalent labelled here C9101 and one for which they are inequivalent labelled 515

C92 C102 Both environments are rigid enough to be observed at short contact time (tc= 50 μs) 516

A similar double environment is also observed for the sugar C1rsquo CPMAS at 70degC shows two 517

peaks (C1rsquo1 at δC= 10443 ppm C1rsquo2 at δC= 10382 ppm Figure 6b Figure S 7d) These are 518

significantly shifted downfield from the C1rsquo peak observed at 110degC and are detected at contact 519

times as short as 50 μs thus indicating their rigid character One also finds that CPMAS is 520

totally inefficient at 100degC for short contact times (50 μs) and poorly efficient even at 3 ms 521

typical of fast-tumbling molecules (Figure S 7c) also both gauche (δC= 3004 ppm) and all-522

trans (δC= 3198 ppm) conformations of the CH2 carbons coexist at 70degC (Figure S 7b)77 523

According to ssNMR one main mobile molecular environment exists at 110degC as 524

shown by the well-resolved 1H spectrum single C=C (C9101) and anomeric (C1rsquo) resonances 525

gauche conformation of the aliphatic CH2 and single T2 of the aliphatic CH2 chain and sugar 526

protons The sugar group is certainly mobile (inefficient 1H13C CPMAS) but still more 527

constrainted than the acyl chain (shorter T2 (1H)) As the temperature is lowered the overall 528

rigidity increases (broadening of 1H NMR peaks faster 1H transverse relaxation and efficient 529

1H13C magnetization transfer during CPMAS) presumably driven by H-bonding 530

interactions (downfield shift and line narrowing of the OH 1H signal very short OH T2 loss of 531

the C1rsquo signal in SPMAS) At the same time a number of spectroscopic data (two concomitant 532

21

resonances for anomeric C1rsquo1 and C1rsquo2 cis-olefin C9101 and C92 C102 and CH2 gauche and 533

all-trans environments) and relaxation behaviour (fast and slow T2 relaxation components) 534

strongly suggests the coexistence of two molecular environments Further proofs of the 535

coexistence of two distinct molecular domains are given by a number of chemical shift 536

variations in the 13C CPMAS spectrum at 70degC compared to the one at 110degC 90degC and 537

solution This is shown in Figure 7 and in the list of 13C chemical shifts in Table 1 538

539

540

Figure 7 - Comparison between solution- and solid-state 13C NMR spectra of G-C181-OH Solid-state 541

experiments are recorded directly on the sample powder at a MAS of 10 kHz (4 mm zirconia rotor) SP 542

Single Pulse CP Cross Polarization In all the ssNMR experiments temperature is first raised at 110degC 543

and then lowered to either 90degC or 70degC Segmented lines are meant to highlight the known peak position 544

in solution thus stressing the strong chemical shift variations in the solid-state 545

546

NMR experiments above Tm2 suggest that vesicles could contain G-C181-OH with a 547

single family of mobile conformations characterized by chemically-equivalent C=C (red 548

Figure 8b) and a single C1rsquo conformation (green Figure 8b) Experiments between Tm1 and Tm2 549

otherwise suggest that two coexisting conformations characterized by chemically-equivalent 550

(red Figure 8b) and inequivalent (magenta Figure 8b) C=C could be associated to nanotubes 551

Two distinct C1rsquo sites (grey Figure 8b) also indicate two configurations of the sugar headgroup 552

A comment on this point will be given in the following paragraphs 553

The local (Aring-scale) variation of molecular mobility below Tm2 drives the coexistence of 554

two distinct molecular environments of the same G-C181-OH molecule This occurs within a 555

single-phase domain (as deduced by WAXS) of equivalent membrane rigidity (as derived from 556

Laurdan probe in confocal microscopy) Although never reported before the difference in the 557

local mobility of the same amphiphile (here G-C181-OH) could satisfy the theoretically-558

140 120 100 80 60 40 20

CP(3 ms)-70degC

SP-70degC

SP-90degC

SP-110degC

ppm

LiquidC910

C1rsquo

C2rsquo

C4rsquo

C3rsquo

C1

C6rsquoC5rsquo

C1

6

C8

11

C2 C

H2

C1

5

CH

2

DM

SO

13C

22

required conditions that induce local lipid segregation and formation of membrane inclusions 559

(Figure 8c) In case of their non-zero average orientation the local membrane curvature 560

increases62 and nanotubes can spontaneously form (Figure 8bc) More intriguing this 561

mechanism occurs at least 10degC above Tm1 although it should not be favored above the Tm27 562

From both experimental and theoretical backgrounds the conditions of topologically-563

connected vesicles-nanotubes systems are met only when either internal (two components) or 564

external driving forces are applied19 In the present single-lipid system of which the phase 565

behavior is summarized in Figure 8a the most plausible driving force could be explained by 566

the NMR results collected on the dry sample G-C181-OH concomitantly occupies two 567

molecular environments characterized by 1) two distinct conformations of the sugar 568

displaying equivalent mobility and 2) two conformations of the acyl chain unsaturation 569

570

23

571

Figure 8 ndash a) Thermotropic phase behavior of G-C181-OH at 5 mgmL-1 Note that Tm2 is observed on the 572

dry powder only b) Possible relationship between the molecular environment as derived from NMR and 573

membrane curvatures All the molecular drawings refer to the same G-C181-OH molecule (bulky ellipsoid 574

equiv sugar headgroup) while each color corresponds to a given configuration Grey of C1rsquo1 and C1rsquo2 indicates 575

the uncertainty about the assignment of the specific NMR peak The curvature C is defined as 1R R being 576

the curvature radius R1 R2 radius of curvature CG Gauss curvature CM Mean curvature c) Free 577

interpretation of the anisotropic high-curvature (CMgt 0 CG= 0) inclusions in the membrane (low CM and 578

CG) continuum driven by inter-sugar H-bonding and reduced intramembrane diffusion for a G-C181-OH 579

glucolipid 580

581

Flat crystalline structure

Nanotube-vesicle network

Vesicles

48degC

120deg

106 105 104 103 102 101

ppm

H9 H10H1rsquo

106 105 104 103 102 101

ppm106 105 104 103 102 101

ppm106 105 104 103 102 101

ppm

C=CC=C

C1rsquo1C1rsquo2

C1rsquo

Tm2(96degC)Tm1

a)

b)

c)

Same molecule different local orientations and sugar conformations

High curvature membrane

Low curvature inclusion

Flip-FlopStrong H-bonding

CG

CM ( + )

CG 0

CM 0 CG 0

CM 0

G-C181-OH

24

Two related questions are unanswered at the moment What actually drives nanotube 582

formation and is the organization of G-C181-OH within the vesicles and tubes the same Lipid 583

nanotubes have been studied for at least three decades over which broad experimental and 584

theoretical insight has been generated Despite this fact the actual understanding is still matter 585

of debate27 Micelle-to-nanotube and vesicle-to-nanotube phase transitions are generally 586

reported for single amphiphiles below a given Tm and intermediate phase composed of twisted 587

and flat fibers is concomitantly reported Initially thought to be a chirality-driven phenomenon 588

recent studies propose a more general mechanism involving crystallization and symmetry 589

breaking in the lipid packing driven by optimal packing27 If twisted ribbons were observed 590

before for a number of microbial glycolipid bolaamphiphiles44 they are not found for G-C181-591

OH This fact as well as the large body of cryo-TEM images and the observation of nanotubes 592

above Tm1 rather pledge in favour of a direct topological change from positive (vesicles) to 593

zero (nanotubes) Gaussian curvatures In this case vesicle-to-nanotube transitions could be 594

driven by external forces56307813ndash162829 595

Considering the absence of external forces in the present system there are only two 596

possibilities left spontaneous budding of the membrane or nanotube pulling after inter-vesicle 597

collisions due to Brownian motion Both are possible as inclusions composed of a different 598

arrangement of the same molecule in the membrane continuum can theoretically explain 599

spontaneous nanotube formation in an isotropic medium672223629ndash1218ndash21 However the second 600

one is probably more likely to prevail because it could explain the large number of nanotubes 601

connecting two distant vesicles Interestingly similar events occur for any phospholipid-based 602

vesicular solution but nanotubes are extremely rare unless at least two lipids of different 603

molecular structure are mixed together The spectroscopic 1H and 13C and relaxation (1H) 604

ssNMR data even if collected on a dry powder demonstrate that the coexistence of two G-605

C181-OH configurations involving both the sugar headgroup and fatty alcohol chain are 606

possible for this molecule As a matter of fact this is not so surprising as more than one energy 607

minima each corresponding to a given conformation are known for glucose its non-reducing 608

derivative methylglucose and carbohydrates in general78ndash80 609

The presence of two configurations could explain the coexistence between two 610

structures of different curvature but the data presented so far do not explain the origin of these 611

two configurations In other words what drives the change in the local structure of the sugar 612

headgroup and fatty alcohol within the membrane 613

The association of known self-assembled structures to spectroscopic 13C ssNMR data is 614

known for years in particular for gluconamides lipids having a linear sugar headgroup81ndash84 It 615

25

was found that gauche effects in some cases up to the γ position in the sugar chain induce 616

upfield chemical shifts while anti and trans conformations result in downfield shifts Each 617

structure (monolayer crystal bilayer crystal or micellar fiber) was then characterized by a given 618

organization of the gluconamide and a characteristic 13C ssNMR fingerprint which could be 619

explained by gauche γ-gauche anti and trans conformations of the headgroup Similar 620

structural-spectroscopic correlations have been collected over the years for much simpler cyclic 621

sugars85 whereas pyranoses can undergo important conformational changes (chair half-chair 622

boat skew) when the appropriate energy barriers are crossed78ndash8086 A quantitative 623

conformational study based on the 13C chemical shift variations shown in Table 1 would be 624

certainly helpful but very delicate in the context of the present state of the art because a 625

rigorous approach requires at least one known crystal structure8182 which is currently missing 626

for this compound A qualitative approach will then be adopted 627

G-C181-OH is a bolaform amphiphile and flip-flop could be a possible mechanism 628

involved in the nanotube stabilization whereas head-tail interactions have long been considered 629

as important in nanotube-forming bolaform amphiphiles6063 From mere considerations based 630

on T2 (1H) relaxation arguments one could exclude head-tail configurations if the R-CH2OH 631

endgroup (H1) was involved in a direct interaction with the sugar group its corresponding T2 632

would probably be comparable to the sugarrsquos one (OH and H1rsquo) but this does not seem to be 633

the case However this hypothesis would be based on the comparison of the relaxation from 634

CH and CH2 protons additionally involved in different kind of motions Besides the relaxation 635

of H1 is influenced by the strong overlap between H1 and H5rsquo For these reasons relaxation 636

arguments are not solid enough and will not be used 637

We then address this question by exploring intermolecular interactions in a direct 638

manner using 2D 1H-1H homonuclear ssNMR correlation spectroscopy At 90degC in the fast-639

tumbling domain of molecular motion (narrow 1H NMR peaks Figure 6a) and corresponding 640

to the vesicle-rich region spin diffusion (NOESY) experiments are preferred over dipolar 641

coupling based experiments because such interaction is averaged out by molecular motion 642

However when the molecular motion is more constrained in the domain of nanotube stability 643

at 70degC (broad 1H spectra Figure 6a) Double-Quantum Single-Quantum (DQ-SQ) correlation 644

experiments are best suited to probe interaction between close molecular neighbors These 645

experiments are reported in Figure 9 and the related interpretation is only briefly reported 646

below while a more detailed discussion can be found on Page 12 in the Supporting Information 647

26

648

Figure 9 ndash a) 2D 1H-1H NOESY NMR experiment performed at MAS= 10 kHz (4 mm probe) on a Bruker 649

300 MHz Avance III HD spectrometer using 300 ms of mixing time Sample temperature is regulated to 650

90degC reached by cooling from 110degC b) 2D 1H-1H DQ-SQ (BAck-to-BAck BABA pulse sequence used for 651

the DQ excitation scheme) NMR experiment performed at MAS= 22 kHz (25 mm probe) on a Bruker 700 652

MHz Avance III spectrometer using a DQ excitation time equal to 2 rotor periods Sample temperature is 653

regulated to 70degC reached by cooling from 110degC The typical schemes illustrating how to interpret these 654

2D experiments are reported above each experimental map Zooms of the sugar region and configurational 655

interpretation are given below each 2D map More information are given in the text below 656

H16-H16

H1-H1

H9-H10

H1rsquo-H1

H1rsquoH910 H16

H1

H1rsquo-H1rsquo

Hxrsquo-Hxrsquo

2

(pp

m)

(ppm)

315

811910 1rsquo

2rsquo2

3rsquo 4rsquo6rsquo

5rsquo

1rsquo2910

asymp asymp asympHDO MeOD Acetone

1

315

811 9101rsquo

2rsquo2

3rsquo4rsquo6rsquo

5rsquo

1rsquo2 9 10

asympasympasympHDOMeODAcetone

H1rsquo-H1

1rsquo

1

(p

pm

)

(ppm)

H1-H2

H1rsquo-H1

H1rsquo

H910

H1H5rsquo

H2

H1rsquo-H5rsquo

315

811910 1rsquo

2rsquo2

3rsquo 4rsquo6rsquo

5rsquo

1rsquo2910

asymp asymp asympHDO MeOD Acetone

1

H2rsquoH3rsquoH4rsquo

H2rsquoH3rsquoH4rsquo

X= 2rsquo3rsquo4rsquo

NOESY - T= 90degC

1H

1H

1H

1H

DQ-SQ - T= 70degC

a) b)

2

(pp

m)

(p

pm

)

(ppm)

A B

AhellipB

AhellipA

BhellipB

NOESY

(ppm)

A B

AhellipB

DQ-SQ

Single-quantum dimension

Do

ub

le-q

ua

ntu

m d

imen

sio

n

AhellipA

BhellipB

BhellipA

BhellipA

315

811910 1rsquo

2rsquo2

3rsquo 4rsquo6rsquo

5rsquo

1rsquo2910

asymp asymp asympHDO MeOD Acetone

1

lt 5 Aring

lt 5 Aring

H1-H2

H811

H15

H16

H6rsquo

H1rsquo

H1

H5rsquo

H2

H1-H2

H1rsquo-H1

H1rsquo-H5rsquo

H1rsquo

H1 H5rsquo

H1rsquo

H1

H5rsquo

H1H2rsquo H3rsquo H4rsquo

H1-H1

H1rsquo-H1

Hxrsquo-HxrsquoX= 2rsquo3rsquo4rsquo

H1-H2

H1rsquo

Zoom sugar region 28 ndash 45 ppmNOESY DQ-SQ

27

Head-tail or head-head configurations could be probed by the existence of H1rsquo-H1 657

interactions whereas H1 (δH= 356 ppm) is associated to the glucolipid tail and H1rsquo (δH= 432 658

ppm) to the sugar headgroup These atoms are more than 15 Aring apart and a direct interaction 659

could not be observed in both NOESY and DQ-SQ experiments NOESY experiments (Figure 660

9a) show that H1rsquo-H1 correlations are not observed Meanwhile clear-cut direct H1rsquo-H1 661

interactions are indeed observed are detected at T= 70degC (DQ cross-peaks related to H1rsquo-H1 662

correlation in DQ-SQ experiment shown in Figure 9b) Interestingly the absence of auto peaks 663

at δH= 432 ppm and δH= 335 ppm in the SQ dimension excludes H1rsquo-H1rsquo and Hxrsquo-Hxrsquo (x= 2 664

3 4) interactions respectively strengthening the idea that sugar headgroups from two nearest-665

neighbor G-C181-OH molecules are not adjacent in the nanotube-vesicle region 666

Homonuclear 2D ssNMR supports the idea according to which G-C181-OH can 667

undergo molecular flip-flops and a head-tail arrangement in the dry product Extrapolating these 668

conclusions in the lower concentration range lack of H1rsquo-H1 proximity at T= 90degC suggests a 669

head-head configuration in the vesicle-rich region and head-tail configuration within the 670

nanotubes It is however still unclear whether flip-flops are driven by spontaneous temperature-671

induced conformational changes in the glucolipid packing stimulating inclusions within the 672

membrane itself672223629ndash1218ndash21 or by defects generated by inter-vesicle collisions In fact a 673

combination of both is not unlikely Nanotubes are not stable when temperature is 20 to 30 674

degrees above Tm1 indicating that inter-vesicle collisions alone cannot explain the unusual 675

phenomenon In an energy landscape approach87 the head-head configuration is certainly 676

favored at high temperature while both head-head and head-tail configurations could be 677

favored in the mid-temperature region corresponding to the range between Tm1 and Tm2 in the 678

dry powder Both statistical and defect-induced flip-flop events contribute to form high mean 679

zero Gaussian curvature morphologies (nanotubes) for which the head-tail conformation 680

limiting the proximity between bulky sugar headgroups is probably more adapted (Figure 8c) 681

The flip-flop interpretation now reasonably explains the strong differences in the sugar and fatty 682

alcohol conformations deduced by the multiple NMR signals attributed to chemically-683

equivalent groups The existence of various low-energy conformations each having specific 684

NMR signature for sugars in solution78ndash8086 and solid-state81ndash84 further supports our 685

interpretation 686

One last point should be clarified Why between Tm1 and Tm2 are two molecular 687

environments able to coexist instead of a single one of average rigidity as found in the case of 688

ldquoripplerdquo phases50ndash54 We believe that this question can be answered based on the NMR results 689

the relatively short T2 values of the H1rsquo and OH protons shows the constrained mobility of the 690

28

sugar headgroup despite a certain flexibility of the chain (long T2 component) This trend may 691

suggest restrictions in translational mobility probably due to the sugar-OH hydrogen bonding 692

(Figure 8c) meaning that lipids are not able to diffuse within the membrane thus keeping two 693

different configurations stable 694

695

Conclusions 696

Topologically-connected nanotube-vesicle lipidic systems are commonly observed for 697

self-assembled membranes composed of at least two amphiphiles and spontaneously driven by 698

anisotropic inclusions andor by external directional forces This work shows that strong 699

differences in membrane curvature can spontaneously occur in single-lipid systems without 700

external anisotropic forces Coexistence and interconnection between nanotubes and vesicles 701

are evidenced via cryo-TEM and fluorescence confocal microscopy in a single-phase region at 702

temperature above a transition temperature Tm1= 483degC determined by DSC on the hydrated 703

sample but below the second transition temperature Tm2 (957degC) determined by DSC on the 704

dry powder X-ray diffraction (WAXS) also shows the coexistence of two lattice periods one 705

attributed to an inter-membrane repeating distance and the other possibly related to the intra-706

nanotube glucolipid arrangement Labelling the systems with Laurdan a fluorophore of which 707

the emission strongly varies with the local membrane rigidity excludes the presence of rigid 708

micro-separated phases (eg Lβ) 709

1H and 13C ssNMR despite the fact that it can only be performed on the dry powder 710

confirm the presence of one main fluid G-C181-OH environment above Tm2 Below Tm2 the 711

splitting of some 13C peaks and the coexistence of short and long 1H relaxation components 712

demonstrate that G-C181-OH can simultaneously pack into structurally and dynamically 713

inequivalent environments Combining advanced 2D 1H-1H correlation experiments based on 714

spin diffusion or double-quanta excitation suggests that vesicles could contain a preferred head-715

head configuration of the bolaform lipid while nanotubes a preferred head-tail assembly 716

The presence of two environments composed of the same molecule characterized by 717

different molecular configurations and rigidity fulfills the theoretical conditions necessary to 718

spontaneously nucleate a membrane inclusion of non-zero average orientation described as 719

necessary to observe topologically-connected nanotube-vesicle structures in the absence of 720

external forces This hypothesis could also be enriched by the possibility of nanotubes to be 721

pulled away between two vesicles after a collision event 722

This work expands the actual theoretical and experimental framework on lipid networks 723

stabilized by a different radius of curvature It could also motivate the synthesis or isolation 724

29

of new lipids providing a higher level of 3D architectural control in soft networks possibly in 725

the range of room temperature 726

727

Acknowledgements 728

Dr Andrea Lassenberger and Dr Sylvain Preacutevost at Institut Laue Langevin (ILL Grenoble 729

France) are kindly acknowledged for their assistance on the SANS experiment Confocal 730

images were performed at the Institut de Biologie Paris Seine (IBPS) imaging facility Jean-731

Franccedilois Gilles (Imaging Core Facility Institut de Biologie Paris Seine (IBPS) CNRS 732

Sorbonne Universiteacute Paris France) is kindly acknowledged for his assistance on the confocal 733

microscopy imaging Dr Guillaume Laurent (LCMCP Sorbonne Universiteacute Paris France) is 734

kindly acknowledged for helpful discussions The French region Ile-de-France SESAME 735

program is acknowledged for financial support (700 MHz NMR 736

spectrometer) Dr Olivier Diat (ICSM Marcoule France) Prof eronikaIglič and Prof V Aleš737

onsul discussi Slovenia) are kindly acknowledged for helpfUniversity of LjubljanaIglič (-Kralj 738

739

Financial support 740

WAXS experiments have been supported by Soleil Light Source Saint Aubin France proposal 741

Ndeg 20201747 SANS experiments have been supported by ILL proposal Ndeg9-13-778 742

743

References 744

1 D Kabaso N Bobrovska W Goacuteźdź E Gongadze V Kralj-Iglič R Zorec and A 745

Iglič Bioelectrochemistry 2012 87 204ndash210 746

2 C R Safinya U Raviv D J Needleman A Zidovska M C Choi M A Ojeda-747

Lopez K K Ewert Y Li H P Miller J Quispe B Carragher C S Potter M W 748

Kim S C Feinstein and L Wilson Adv Mater 2011 23 2260ndash2270 749

3 P Veranič M Lokar G J Schuumltz J Weghuber S Wieser H Haumlgerstrand V Kralj-750

Iglič and A Iglič Biophys J 2008 95 4416ndash4425 751

4 A Rustom R Saffrich I Markovic P Walther and H-H Gerdes Science (80- ) 752

2004 303 1007ndash1010 753

5 M I Angelova S Soleacuteau P Meacuteleacuteard F Faucon and P Bothorel Prog Colloid 754

Polym Sci 1992 89 127ndash131 755

6 L Mathivet S Cribier and P F Devaux Biophys J 1996 70 1112ndash1121 756

7 V Kralj-Iglič G Gomišček J Majhenc V Arrigler and S Svetina Colloids Surfaces 757

A Physicochem Eng Asp 2001 181 315ndash318 758

30

8 V Kralj-Iglič A Iglič G Gomišček F Sevšek V Arrigler and H Haumlgerstrand J 759

Phys A Math Gen 2002 35 1533ndash1549 760

9 M Drab D Stopar V Kralj-Iglič and A Iglič Cells 2019 8 626 761

10 A Zidovska K K Ewert J Quispe B Carragher C S Potter and C R Safinya 762

Langmuir 2009 25 2979ndash2985 763

11 A Zidovska K K Ewert J Quispe B Carragher C S Potter and C R Safinya Soft 764

Matter 2011 7 8363ndash8369 765

12 A Zidovska K K Ewert J Quispe B Carragher C S Potter and C R Safinya 766

Biochim Biophys Acta - Biomembr 2009 1788 1869ndash1876 767

13 M Karlsson K Sott M Davidson A S Cans P Linderholm D Chiu and O Orwar 768

Proc Natl Acad Sci U S A 2002 99 11573ndash11578 769

14 A Karlsson R Karlsson M Karlsson A-S Cans A Stromberg F Ryttsen and O 770

Orwar Nature 2001 409 150ndash152 771

15 M Karlsson K Sott A S Cans A Karlsson R Karlsson and O Orwar Langmuir 772

2001 17 6754ndash6758 773

16 E Evans H Bowman A Leung D Needham and D Tirrell Science (80- ) 1996 774

273 933ndash935 775

17 U Seifert Configurations of fluid membranes and vesicles 1997 vol 46 776

18 J Kaumls and E Sackmann Biophys J 1991 60 825ndash844 777

19 N Bobrovska W Goacuteźdź V Kralj-Iglič and A Iglič PLoS One 2013 8 e73941 778

20 A Iglič B Babnik K Bohinc M Fošnarič H Haumlgerstrand and V Kralj-Iglič J 779

Biomech 2007 40 579ndash585 780

21 K Bohinc T Slivnik A Iglič M Brumen and V Kralj-Iglič J Phys Chem C 2007 781

111 9709ndash9718 782

22 A Iglič H Haumlgerstrand P Veranič A Plemenitaš and V Kralj-Iglič J Theor Biol 783

2006 240 368ndash373 784

23 V Kralj-Iglič and P Veranič Adv Planar Lipid Bilayers Liposomes 2006 5 129ndash785

149 786

24 A Iglic and V Kralj-Iglic Adv Planar Lipid Bilayers Liposomes 2006 4 253ndash279 787

25 S Penič L Mesarec M Fošnarič L Mroacutewczyńska H Haumlgerstrand V Kralj-Iglič and 788

A Iglič Front Phys 2020 8 342 789

26 P Yager and P E Schoen Mol Cryst Liq Cryst 1984 106 371ndash381 790

27 T G Barclay K Constantopoulos and J Matisons Chem Rev 2014 114 10217ndash791

10291 792

31

28 B Božič G Gomišček V Kralj-Iglič S Svetina and B Žekš Eur Biophys J 2002 793

31 487ndash496 794

29 A Iglič H Haumlgerstrand M Bobrowska-Haumlgerstrand V Arrigler and V Kralj-Iglič 795

Phys Lett A 2003 310 493ndash497 796

30 R Bar-Ziv and E Moses Phys Rev Lett 1994 73 1392ndash1395 797

31 S Chiruvolu H E Warriner E Naranjo S H Idziak J O Raumldler R J Plano J A 798

Zasadzinski and C R Safinya Science (80- ) 1994 266 1222ndash1225 799

32 V Kralj-Iglič A Iglič H Haumlgerstrand and P Peterlin Phys Rev E 2000 61 4230ndash800

4234 801

33 L Mesarec M Drab S Penič V Kralj-Iglič and A Iglič Int J Mol Sci 2021 22 802

2348 803

34 L Van Renterghem S L K W Roelants N Baccile K Uyttersprot M C Taelman 804

B Everaert S Mincke S Ledegen S Debrouwer K Scholtens C V Stevens W 805

Soetaert L Van Renterghem S L K W Roelants N Baccile K Uytersprot M C 806

Taelman B Everaert S Mincke S Ledegen S Debrouwer K Scholtens C V 807

Stevens and W Soetaert Biotechnol Bioeng 2018 115 1195ndash1206 808

35 J Clark and F Deswarte Introduction to Chemicals from Biomass Second Edition 809

John Wiley amp Sons 2015 810

36 K V Sajna R Houmlfer R K Sukumaran L D Gottumukkala and A Pandey in 811

Industrial Biorefineries and White Biotechnology eds A Pandey R Houmlfer M 812

Taherzadeh K M Nampoothiri and C Larroche Amsterdam Oxford Waltham 813

Elsevier 2015 pp 499ndash521 814

37 R Marchant and I M Banat Trends Biotechnol 2012 30 558ndash565 815

38 K Hill and C LeHen-Ferrenbach in Sugar-Based Surfactants Fundamentals and 816

Applications ed C C Ruiz Boca Raton CRC Press 2009 pp 1ndash20 817

39 L Rodrigues I M Banat J Teixeira and R Oliveira J Antimicrob Chemother 818

2006 57 609ndash618 819

40 J D Desai and I M Banat Microbiol Mol Biol Rev 1997 61 47ndash64 820

41 S L K W Roelants K Ciesielska S L De Maeseneire H Moens B Everaert S 821

Verweire Q Denon B Vanlerberghe I N A Van Bogaert P Van der Meeren B 822

Devreese and W Soetaert Biotechnol Bioeng 2016 113 550ndash559 823

42 N Baccile M Selmane P Le Griel S Preacutevost J Perez C V Stevens E Delbeke S 824

Zibek M Guenther W Soetaert I N A Van Bogaert and S Roelants Langmuir 825

2016 32 6343ndash6359 826

32

43 N Baccile A-S Cuvier S Preacutevost C V Stevens E Delbeke J Berton W Soetaert 827

I N A Van Bogaert and S Roelants Langmuir 2016 32 10881ndash10894 828

44 N Baccile C Seyrig A Poirier S A Castro S L K W Roelants and S Abel Green 829

Chem 2021 23 3842ndash3944 830

45 A Salis L Cappai C Carucci D F Parsons and M Monduzzi J Phys Chem Lett 831

2020 11 6805ndash6811 832

46 J Schindelin I Arganda-Carreras E Frise V Kaynig M Longair T Pietzsch S 833

Preibisch C Rueden S Saalfeld B Schmid J-Y Tinevez D J White V 834

Hartenstein K Eliceiri P Tomancak and A Cardona Nat Methods 2012 9 676ndash835

682 836

47 T Parasassi and E Gratton J Fluoresc 1995 5 59ndash69 837

48 M M Dodes Traian F L Gonzaacutelez Flecha and V Levi J Lipid Res 2012 53 609ndash838

616 839

49 D M Owen C Rentero A Magenau A Abu-Siniyeh and K Gaus Nat Protoc 840

2012 7 24ndash35 841

50 W Stillwell in An Introduction to Biological Membranes Academic Press 2016 pp 842

135ndash180 843

51 W Stillwell in An Introduction to Biological Membranes Academic Press 2016 pp 844

181ndash220 845

52 B Tenchov Chem Phys Lipids 1991 57 165ndash177 846

53 R Koynova and M Caffrey Biochim Biophys Acta 1998 1376 91ndash145 847

54 F Schmid S Dolezel O Lenz and S Meinhardt J Phys Conf Ser 2014 487 848

012004 849

55 S Meinhardt R L C Vink and F Schmid Proc Natl Acad Sci U S A 2013 110 850

4476ndash4481 851

56 O Glatter and O Kratky Small Angle X-ray Scattering Academic Press London 852

1982 853

57 B A Cunningham A D Brown D H Wolfe W P Williams and A Brain Phys 854

Rev E 1998 58 3662ndash3672 855

58 M Nadler A Steiner T Dvir O Szekely P Szekely A Ginsburg R Asor R Resh 856

C Tamburu M Peres and U Raviv Soft Matter 2011 7 1512ndash1523 857

59 C Tanford The hydrophobic effect Formation of micelles and biological membranes 858

John Wiley amp Sons Inc New York 1973 859

60 M Masuda and T Shimizu Langmuir 2004 20 5969ndash5977 860

33

61 A Zidovska K K Ewert J Quispe B Carragher C S Potter and C R Safinya 861

Block Liposomes Vesicles of Charged Lipids with Distinctly Shaped Nanoscale 862

Sphere- Pear- Tube- or Rod-Segments Elsevier Inc 1st edn 2009 vol 465 863

62 A Iglic V Kralj-Iglic B Božič M Bobrowska-Haumlgerstrand B Isomaa and H 864

Haumlgerstrand Bioelectrochemistry 2000 52 203ndash211 865

63 T Shimizu M Masuda and H Minamikawa Chem Rev 2005 105 1401ndash1444 866

64 J V Selinger and J M Schnur Phys Rev Lett 1993 71 4091ndash4094 867

65 J M Schnur Science (80- ) 1993 262 1669ndash1676 868

66 B Tenchov R Koynova and G Rapp Biophys J 2001 80 1873ndash1890 869

67 R Koynova and B Tenchov Ser Biomech 2017 31 3ndash19 870

68 S J Angyal and R Le Fur Carbohydr Res 1984 126 15ndash26 871

69 J K Whitesell and M A Minton J Am Chem Soc 1987 109 225ndash228 872

70 T Takayama S Ando and I Ando J Mol Struct 1990 220 245ndash250 873

71 K Gawrisch N V Eldho and I V Polozov Chem Phys Lipids 2002 116 135ndash151 874

72 F Tielens C Gervais J F Lambert F Mauri and D Costa Chem Mater 2008 20 875

3336ndash3344 876

73 G A Jeffrey and Y Yeon Acta Crystallogr Sect B 1986 42 410ndash413 877

74 B Berglund and R W Vaughan J Chem Phys 1980 73 2037ndash2043 878

75 C Gervais C Coelho T Azaiumls J Maquet G Laurent F Pourpoint C Bonhomme 879

P Florian B Alonso G Guerrero P H Mutin and F Mauri J Magn Reson 2007 880

187 131ndash140 881

76 C Akita T Kawaguchi F Kaneko H Yamamoto and M Suzuki J Phys Chem B 882

2004 108 4862ndash4868 883

77 R N Purusottam L Seacutenicourt J J Lacapegravere and P Tekely Biochim Biophys Acta - 884

Biomembr 2015 1848 3134ndash3139 885

78 X Biarneacutes A Ardegravevol A Planas C Rovira A Laio and M Parrinello J Am Chem 886

Soc 2007 129 10686ndash10693 887

79 L M J Kroon‐Batenburg and J Kroon Biopolymers 1990 29 1243ndash1248 888

80 K N Kirschner and R J Woods Proc Natl Acad Sci U S A 2001 98 10541ndash889

10545 890

81 S Svenson B Kirste and J H Fuhrhop J Am Chem Soc 1994 116 11969ndash11975 891

82 I Sack S Macholl J H Fuhrhop and G Buntkowsky Phys Chem Chem Phys 892

2000 2 1781ndash1788 893

83 S Svenson J Koening and J H Fuhrhop J Phys Chem 1994 98 1022ndash1028 894

34

84 S Svenson A Schaefer and J H Fuhrhop J Chem Soc Perkin Trans 2 1994 2 895

1023ndash1028 896

85 K Bock and C Pedersen Adv Carbohydr Chem Biochem 1983 41 27ndash66 897

86 T B Grindley Structure and Conformation of Carbohydrates Springer Berlin 898

Heidelberg 2008 899

87 F Tantakitti J Boekhoven X Wang R V Kazantsev T Yu J Li E Zhuang R 900

Zandi J H Ortony C J Newcomb L C Palmer G S Shekhawat M O De La 901

Cruz G C Schatz and S I Stupp Nat Mater 2016 15 469ndash476 902

903

1

Supporting Information 904

905

Topological connection between vesicles and nanotubes in single-906

component lipid membranes driven by head-tail interactions 907

908

Niki Baccilea Ceacutedric Lorthioira Abdoul Aziz Baa Patrick Le Griela Cristina Coelhob Javier 909

Perezc Wim Soetaertd Sophie L K W Roelantsd 910

911

a Sorbonne Universiteacute Centre National de la Recherche Scientifique Laboratoire de Chimie de 912

la Matiegravere Condenseacutee de Paris LCMCP F-75005 Paris France 913

b Institut des Mateacuteriaux de Paris Centre Sorbonne Universiteacute Paris France 914

c Synchrotron Soleil LrsquoOrme des Merisiers Saint-Aubin BP48 91192 Gif-sur-Yvette Cedex 915

France 916

d InBio Department of Biotechnology Ghent University Ghent Belgium 917

918

919

Corresponding author 920

Dr Niki Baccile 921

E-mail address nikibaccilesorbonne-universitefr 922

Phone +33 1 44 27 56 77 923

924

925

926

2

a) 927

Parameter Method

Dry Matter (DM ) 996 Infrared balance 105degC

Glucose () 012 HPLC-Metacarb

Glycerol () nd HPLC-Metacarb

Free Fatty Acid content () lt001 Internal method BBEPP

GC with FID Detector

Oil () lt001 Internal method BBEPP

GC with FID Detector

Moulds (CFUg) lt10 3M-Nordval ndeg16 B

Yeasts (CFUg) lt10 3M-Nordval ndeg16 B

Anaerobic count (CFUg) lt10 SP-VG M005 B

Protein () tbd BCA protein assay

928

Figure S 1 - a) HPLC-ELSD chromatogram and composition table 929

930

931

932

933

3

b) 934

935

Group Ndeg δ ppm

CH2 1 354

CH2 2 153

CH2 3-712-14 132

CH2 811 203

CH 910 534

CH2 15 162

CH2 16 388

CH 1 424

CH 2 316

CH 3 334

CH 4 329

CH 5 354

CH2 6 366

Figure S 1 (next) ndash b) 1H NMR spectrum (MeOD-d4) recorded on the purified G-C181-OH compounds 936

and related assignment 937

938

2

811910 1rsquo

5rsquo1

3rsquo4rsquo

1rsquo1910

asymp asymp asympHDO MeOD Acetone

6rsquo2rsquo

1615

3-712-14

4

939

940

Figure S 2 ndash Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 130degC 941

5

942

943

Figure S 3 - Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 90degC 944

945

90degC

1

22

6

946

947

948

949

Figure S 4 ndash Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 70degC 950

951

21

11

1

2

5

3

4

45

5

6

7

7

7

952

953

954

Figure S 5 ndash Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 25degC 955

956

957

1

1 1

8

958

Figure S 6 ndash a) SANS experiment recorded at 90degC for a 5 mgmL G-C181-OH aqueous solution b-c) 959

WAXS experiments recorded on aqueous solutions of G-C181-OH at b) 5 mgmL and c) 25 mgmL d) 960

Structural interpretation of the repeating distances pointed out in c) 961

962

963

964

001 01 1

q Aring-1

25degC

68degC

88degC

111degC

01 1q Aring

-1

02 04 06 08 1q Aring

-1

001 01

10-1

100

101

Inte

nsity

cm

-1

q Aring-1

-2

T= 90degC

SANS WAXS WAXS

a) b) c)

dv

dL d= 144 Aring-1

di

ddH2O

dt

dvdL

dtdi d

d)

9

965

966

Figure S 7 ndash ssNMR MAS (10 kHz) experiments recorded on a G-C181-OH dry powder a) 1H spectra 967

recorded from 110degC to 70degC b) 13C single pulse (SP) spectra recorded with high power 1H decoupling from 968

110degC to 70degC c) 13C cross polarization (CP) spectra recorded at 110degC and contact time tc of 50 micros and 969

3000 micros d) 13C CP spectra recorded at 70degC and for 20 lt tc micros lt 3000 970

971

1

811

910

CH2 CH2

1rsquo

6rsquo

8 6 4 2 0 ppm

110

90

80

70

910

1rsquo

6rsquo16

1

2rsquo3rsquo4rsquo

811

OH

45 40 35 30 25 20 15

ppm

200 180 160 140 120 100 80 60 40 20 0

ppm140 120 100 80

ppm

CH2-gauche

3004

110

90

80

70

65

60

50

25

C91013003

C9 - 12869

CH2-all-trans

3198

C1rsquo10335

a) b)

1H

T degCT degC

13C Single Pulse MAS

A

B

A

B

70degC70degC

5rsquo

215

200 180 160 140 120 100 80 60 40 20 0

ppm

13C CPMAST= 110degC

200 180 160 140 120 100 80 60 40 20 0

ppm

132 131 130 129 128 127 126

ppm

106 105 104 103 102

ppm

13C CPMAST= 70degC

50

tc micros

3000

20

tc micros

50 500 1000 3000

tc micros

C102

13024 C910113003

C92

12869

C1rsquo213082

C1rsquo110443

A B

106 105 104 103 102

ppm

3000

1000

500

50

20

AB

c)

d)

10

972

Figure S 8 ndash a) Short and d) long T2 components displayed by the 1H transverse relaxation signal c) 1H 973

transverse relaxation signals recorded on a G-C181-OH dry powder from 110degC to 70degC The area of each 974

given resonance is plotted (log-lin) as a function of the echo time (CPMG pulse sequence) for each 975

temperature and fitted using the equation given above T2s and T2l indicate the short and long T2 relaxation 976

times respectively obtained by describing the 1H transverse relaxation signal using a biexponential decay 977

For H1rsquo H5rsquo and H6rsquo the 1H relaxation function could be described by a monoexponential decay 978

- Protons from the sugar group (H1rsquo H5rsquo H6rsquo) show a single T2 relaxation component (Figure 979

6c and Figure S 8) while protons of the acyl chain (CH2 H910 H811) show both short (Figure 980

6c) and long (Figure 6d) T2 relaxation component 981

- The short T2 relaxation time systematically decreases from 110degC to 70degC for all species by a 982

factor ranging from two to seven depending on the specific chemical group Very long T2 are 983

observed at 110degC for all acyl species while nearly unchanged values are measured below 984

90degC without any clear trend 985

- The sugar protons (H1rsquo H5rsquo H6rsquo) have the shortest set of T2 values (lt 1 ms) below 100degC 986

The long component of T2 generally ranges between 4 ms and 12 ms 987

0 1 2 3 4 5 6 7 8

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1 2 3 4 5 6 7 8

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

H1

H8H11

H9H10

CH2

H1rsquo

H6rsquo

CH2

H910H811 H1CH2

H1rsquo H5rsquo6rsquo

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

OH

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45 T= 110degC

T= 90degC

T= 80degC

T= 70degC

Log

In

ten

sity

Time ms

Fit

)

CH2 H910 H811 H1 H56 H1 OH

00

05

10

15

20

25

30

35

40

T2 m

s

T= 110degC

T= 90degC

T= 80degC

T= 70degC

CH2 H910 H811 H1

0

2

4

6

8

10

12

T2 m

s

T= 110degC

T= 90degC

T= 80degC

T= 70degC

Short T2(1H) Long T2(1H)

a) b)

c)

11

Table S 1 ndash 1H chemical shifts of G-C181-OH in solution and in the solid-state Solid-state spectra are 988

recorded at a MAS frequency of 10 kHz in DMSO-d6 broad peak 989

990

δH ppm

Group Ndeg Liquid

(MeOD-d4) Solid

(T= 110degC) Solid

(T= 70degC)

CH2 H1 354 357

CH2 H2 153 156

CH2 H3-712-14 132 135

CH2 H811 203 203

CH H910 534 534

CH2 H15 162 164

CH2 H16 388 383

CH H1 424 432

CH H2 316 335

CH H3 334 335

CH H4 329 335

CH H5 354 351 357

CH2 H6 366 383

OH OH1 432

45 463

OH OH2 492

OH OH3 490

OH OH4 487

OH OH6 445

991

992

H1H9H10H1rsquo

H6rsquo

OH2rsquo

OH1

12

Detailed discussion about the NOESY and DQ-SQ 2D 1H-1H ssNMR experiments presented in 993

Figure 9 994

995

2D 1H-1H ssNMR experiments are a unique tool to explore through-space proximities 996

between 1H nuclei ssNMR offers a large number of possibilities each adapted to a given 997

system The advantage of 2D maps with respect to a dynamics approach is their direct 998

readability in clear two nuclei which exchange their magnetization through-space (whichever 999

mechanism may be responsible for this exchange) will display a specific cross-peak However 1000

the way cross-peaks are presented and interpreted strictly depends on the nature of the pulse 1001

program which in turn is chosen according to the physical nature of the sample 1002

1D temperature-dependent 1H NMR experiments performed on G-C181-OH (Figure 1003

4a in the main text) indicate sharp peaks at 90degC and broad peaks at 70degC Narrow peaks 1004

indicate fast andor large-amplitude molecular tumbling averaging the strong 1H-1H dipolar 1005

interactions This excludes the use of pulse sequences like DQ-SQ which is not efficient under 1006

these conditions For this reason we employ spin-diffusion-based pulse sequences like 1007

NOESY which is more adapted for samples with high molecular mobility When mixing times 1008

are chosen to be short (lt 1 s) one generally explores through-space proximities within about 5 1009

10 Aring On the contrary when molecular motions are restricted dipolar coupling becomes 1010

significant 1H NMR peaks are broader and pulse sequences like DQ-SQ can now be used 1011

with a spatial resolution of less than 5 Aring 1012

Whenever possible DQ-SQ experiments are preferred over spin diffusion-based pulse 1013

sequences because they are more selective in particular for likewise interactions which are 1014

not discriminated in NOESY-type experiments This is illustrated by the typical interaction 1015

schemes for NOESY (top Figure 9a) and DQ-SQ (top Figure 9b) For two given A and B 1016

protons 2D NOESY maps will display two on-diagonal and two square-shaped symmetrical 1017

off-diagonal cross-peaks Off-diagonal cross-peaks illustrate the mutual A-B and B-A 1018

interaction but the on-diagonal peaks are systematic and do not prove any A-A or B-B 1019

interaction Similarly to NOESY 2D DQ-SQ maps display off-diagonal peaks (symmetric 1020

across the diagonal top Figure 9b) upon A-B and B-A interaction but in contrast they will 1021

display on-diagonal peaks only in the presence of self-interaction In the tutorial example of 1022

Figure 9b top the on-diagonal A-A peak indicates mutual A-A interaction while lack of on-1023

diagonal B-B peak evidences the absence of B-B proximities Combining NOESY and DQ-SQ 1024

experiments is capital to study the local arrangement of G-C181-OH and its evolution with 1025

temperature 1026

13

At T= 90degC the H1 and H5rsquo contributions are very close but possible to discriminate 1027

them (δ(H1)= 354 ppm δ(H5rsquo)= 351 ppm) according to 1H NMR chemical shifts in Table S 1028

1 2D 1H-1H NOESY in Figure 9a also discriminates between the two components attributed 1029

on the basis of the interactions with their mutual neighbors (green H1-H2 and H5rsquo-H1rsquo squares 1030

in Figure 9a) On this basis 2D NOESY does not show a direct H1rsquo-H1 correlation (blue red-1031

crossed) but rather an extensive network of intra-sugar interactions These features exclude 1032

suggest intra- and inter-sugar interactions and they exclude head-tail arrangements as a major 1033

molecular configuration at 90degC 1034

At 70degC the 1H NMR peaks are broader indicating a reduction of the T2 (1H) relaxation 1035

times andor residual homonuclear dipolar interactions excluding the use of NOESY In this 1036

case 1H-1H DQ-SQ correlation experiments are best suited Figure 9b presents the 1H-1H DQ-1037

SQ experiment on G-C181-OH at 70degC Peaks are broader than in NOESY and attribution may 1038

be cumbersome However the experiment is performed at 700 MHz and MAS= 22 kHz which 1039

improves spectral resolution with respect to the same experiment performed at 300 MHz and 1040

MAS= 10 kHz The auto-peak at δH= 356 ppm in the SQ dimension is attributed to H1-H1 spin 1041

pairs mainly resulting from intra CH2 interaction while the DQ cross-peaks at δH= 156-356 1042

ppm and δH= 432-356 ppm respectively origin from H2-H1 and H1rsquo-H1 interactions (Figure 1043

9b) DQ-SQ map does not show any auto-peak corresponding to H1rsquo-H1rsquo or Hxrsquo-Hxrsquo (x= 2rsquo 1044

3rsquo 4rsquo) interactions (red-crossed blue Figure 9b) thus excluding any proximity (lt 5 Aring) between 1045

close glucose neighbors On the contrary DQ-SQ maps (recorded both at 700 MHz Figure 9b 1046

or at 300 MHz not shown) systematically show DQ cross-peaks which can only be attributed 1047

to H1rsquo-H1 interactions (δH= 432-356) The presence of these H1rsquo-H1 interactions can only be 1048

explained by a head-tail arrangement of the G-C181-OH molecules as illustrated below the 1049

DQ-SQ map in Figure 9b 1050

18

the local arrangement of G-C181-OH in two morphologies vesicles and nanotubes having 461

sensibly different curvatures 462

To explore the local mobility of the most relevant chemical groups of G-C181-OH 463

above Tm2 (T= 110degC) and between Tm2 and Tm1 (70 lt T degC lt 90) we combine a series of 1H 464

and 13C experiments Single-pulse (SP) MAS experiments (1H and 13C) when recorded under 465

quantitative conditions allow the detection of all chemical species while cross-polarization 466

(CP) MAS (13C only) experiments are sensitive only to constrained groups for which the 1H-467

13C dipolar coupling is not averaged to zero by the molecular motions 468

469

Figure 6 ndash a) 1H MAS (10 kHz) ssNMR spectra recorded from 110degC to 70degC b) 13C MAS (10 kHz) ssNMR 470

spectra recorded using high power decoupling single pulse (SP) and cross-polarization (CP) pulse sequences 471

Contact time for CP experiments and corresponding temperatures are given on each spectrum Full spectra 472

are given in Figure S 7 473

The most relevant 1H and 13C spectra are given in Figure 6 (full range in Figure 7 and 474

Figure S 7) As expected the 1H solid-state SPMAS spectra show broader peaks compared to 475

the solution-state NMR spectrum (Figure 6a) reflecting the obvious solid-state character of the 476

sample at any temperature The sugar group and the C=C bond resonate in the 25-6 ppm range 477

and 535 ppm respectively while the OH groups of the sugar moiety are found below 45 ppm 478

(Figure 6a full assignment in Table S 1) Very broad (~1 ppm) at 100degC the position of the 479

OH peak varies with the temperature value and becomes narrower (lt~05 ppm) and visible at 480

δH= 464 ppm at T= 70degC Line narrowing and chemical shift variations of the OH resonance 481

in ssNMR is always a direct sign of H-bonding interactions whereas the strength of the H-482

bonding hence its OHhellipH distance is correlated to a dispersion in chemical shifts72ndash75 All other 483

60 55 50 45 40 35 30 25

ppm

110

90

80

70

910

1rsquo

6rsquo16

1

OH

134 132 130 128 126

ppm110 108 106 104 102 100

ppm

SP (T= 110degC)

SP (T= 70degC)

CP 50 ms (T= 70degC)

CP 3 ms (T= 70degC)

910 1rsquo

1H 13C

1

811

910

CH2 CH2

1rsquo

6rsquo

a) b)

C102

13024

C9101

13003

C92

12869 C1rsquo210382

C1rsquo110443

C1rsquo13035

2rsquo3rsquo4rsquo

T degC

5rsquo

19

resonances broaden from 110degC to 70degC thus indicating slower andor more anisotropic 484

reorientational motions of all chemical groups The specific dynamic behaviour of the sugar 485

headgroup is also reflected by its 1H transverse relaxation time T2 whereas the shorter the T2 486

the higher the rigidity Experiments performed between 110degC and 70degC shown and 487

commented in Figure S 8 in the Supporting Information globally show a single short (lt 1 ms 488

below 100degC) T2 relaxation component for H1rsquo H5rsquo and H6rsquo and two both short and long 489

(between 4 ms and 12 ms) T2 for the acyl chain (CH2 H910 H811) even below 100degC 490

At 110degC 13C SPMAS experiments (Figure 6b full range in Figure 7 and Figure S 7 491

assignments in Table 1) show a single peak at δC= 10335 ppm for C1rsquo and at δC= 13003 ppm 492

for C910 as expected for a symmetrical conformation with two chemically-equivalent 493

carbons76 At 70degC the C1rsquo peak disappears while C910 now shows two peaks around 128 494

ppm and 130 ppm (Table 1 full temperature series in Figure S 7b) attributed to C92 (δC= 12869 495

ppm) and C10 in a non-symmetric configuration respectively for which these two carbons are 496

chemically inequivalent76 More precisely the resonance at 130 ppm is actually the 497

superposition of two close resonances C9101 (δC= 13003 ppm) and C102 (δC= 13024 ppm) 498

where subscripts 1 and 2 stand for two different conformations of the cis-olefin group as 499

reported for various phases of oleic acid in the vicinity of 0degC76 500

501

Table 1 ndash 13C Chemical shifts of G-C181-OH in solution and solid-state Solid-state experiments are 502

recorded at MAS= 10 kHz All spectra are reported in Figure 7 and Figure S 7 The contact time for cross 503

polarization is 3 ms In all ssNMR experiments temperature is first raised at 110degC and then lowered to 504

either 90degC or 70degC Attribution of the 13C resonances in Table 1 is performed on the basis of the solution 505

spectrum (Figure S 1b) Additional peaks observed in both single pulse and cross polarization ssNMR 506

experiments are attributed when possible to a configuration effect of the given atom eg C9101 and C92 507

indicate two configurations of C9 All new peaks observed in solid-state and not detected in solution are 508

highlighted in red for convenience 509

510

511

δ ppm

Peak Liquid (DMSO-d6)

Solid Single Pulse

T= 110degC

Solid Single Pulse

T= 90degC

Solid Single Pulse

T= 70degC

Solid Cross Polarization

T= 70degC

C102

- - 13024 13024

C910C9101 13009 13003 13003 13003 13003

C1C9C10C1rsquo

C6rsquo

20

C92 - - - 12869 12869

C1rsquoC11 10329 10335 10335 - 10443

C12 - - - - 10382

C3 7724 7709 7702 - 7829

- 7634 7629 - 7621

7525

C2 7388 7398 7390 7370 7370

C4 7052 7077 7054 7041 7098

C16 6900 7016 - - 7029

- - - - 6793

C1C5C6 6153 6241 6236 6239 6262

C1C5C6 6117 - - - 5956

C2 3300 3290 3290 3283 3279

CH2(trans) - - - 3198 3196

CH2(gauche)C15 2941 2981 2987 3004 -

C811 2706 2747 2747 2832 2817

CH2 2597 2625 2628 2630 2817

512

13C CPMAS spectra at 70degC (Figure 6b full range in Figure 7 and Figure S 7d) at 513

increasing contact times confirm the presence of two cis-olefin environments one for which 514

C910 are equivalent labelled here C9101 and one for which they are inequivalent labelled 515

C92 C102 Both environments are rigid enough to be observed at short contact time (tc= 50 μs) 516

A similar double environment is also observed for the sugar C1rsquo CPMAS at 70degC shows two 517

peaks (C1rsquo1 at δC= 10443 ppm C1rsquo2 at δC= 10382 ppm Figure 6b Figure S 7d) These are 518

significantly shifted downfield from the C1rsquo peak observed at 110degC and are detected at contact 519

times as short as 50 μs thus indicating their rigid character One also finds that CPMAS is 520

totally inefficient at 100degC for short contact times (50 μs) and poorly efficient even at 3 ms 521

typical of fast-tumbling molecules (Figure S 7c) also both gauche (δC= 3004 ppm) and all-522

trans (δC= 3198 ppm) conformations of the CH2 carbons coexist at 70degC (Figure S 7b)77 523

According to ssNMR one main mobile molecular environment exists at 110degC as 524

shown by the well-resolved 1H spectrum single C=C (C9101) and anomeric (C1rsquo) resonances 525

gauche conformation of the aliphatic CH2 and single T2 of the aliphatic CH2 chain and sugar 526

protons The sugar group is certainly mobile (inefficient 1H13C CPMAS) but still more 527

constrainted than the acyl chain (shorter T2 (1H)) As the temperature is lowered the overall 528

rigidity increases (broadening of 1H NMR peaks faster 1H transverse relaxation and efficient 529

1H13C magnetization transfer during CPMAS) presumably driven by H-bonding 530

interactions (downfield shift and line narrowing of the OH 1H signal very short OH T2 loss of 531

the C1rsquo signal in SPMAS) At the same time a number of spectroscopic data (two concomitant 532

21

resonances for anomeric C1rsquo1 and C1rsquo2 cis-olefin C9101 and C92 C102 and CH2 gauche and 533

all-trans environments) and relaxation behaviour (fast and slow T2 relaxation components) 534

strongly suggests the coexistence of two molecular environments Further proofs of the 535

coexistence of two distinct molecular domains are given by a number of chemical shift 536

variations in the 13C CPMAS spectrum at 70degC compared to the one at 110degC 90degC and 537

solution This is shown in Figure 7 and in the list of 13C chemical shifts in Table 1 538

539

540

Figure 7 - Comparison between solution- and solid-state 13C NMR spectra of G-C181-OH Solid-state 541

experiments are recorded directly on the sample powder at a MAS of 10 kHz (4 mm zirconia rotor) SP 542

Single Pulse CP Cross Polarization In all the ssNMR experiments temperature is first raised at 110degC 543

and then lowered to either 90degC or 70degC Segmented lines are meant to highlight the known peak position 544

in solution thus stressing the strong chemical shift variations in the solid-state 545

546

NMR experiments above Tm2 suggest that vesicles could contain G-C181-OH with a 547

single family of mobile conformations characterized by chemically-equivalent C=C (red 548

Figure 8b) and a single C1rsquo conformation (green Figure 8b) Experiments between Tm1 and Tm2 549

otherwise suggest that two coexisting conformations characterized by chemically-equivalent 550

(red Figure 8b) and inequivalent (magenta Figure 8b) C=C could be associated to nanotubes 551

Two distinct C1rsquo sites (grey Figure 8b) also indicate two configurations of the sugar headgroup 552

A comment on this point will be given in the following paragraphs 553

The local (Aring-scale) variation of molecular mobility below Tm2 drives the coexistence of 554

two distinct molecular environments of the same G-C181-OH molecule This occurs within a 555

single-phase domain (as deduced by WAXS) of equivalent membrane rigidity (as derived from 556

Laurdan probe in confocal microscopy) Although never reported before the difference in the 557

local mobility of the same amphiphile (here G-C181-OH) could satisfy the theoretically-558

140 120 100 80 60 40 20

CP(3 ms)-70degC

SP-70degC

SP-90degC

SP-110degC

ppm

LiquidC910

C1rsquo

C2rsquo

C4rsquo

C3rsquo

C1

C6rsquoC5rsquo

C1

6

C8

11

C2 C

H2

C1

5

CH

2

DM

SO

13C

22

required conditions that induce local lipid segregation and formation of membrane inclusions 559

(Figure 8c) In case of their non-zero average orientation the local membrane curvature 560

increases62 and nanotubes can spontaneously form (Figure 8bc) More intriguing this 561

mechanism occurs at least 10degC above Tm1 although it should not be favored above the Tm27 562

From both experimental and theoretical backgrounds the conditions of topologically-563

connected vesicles-nanotubes systems are met only when either internal (two components) or 564

external driving forces are applied19 In the present single-lipid system of which the phase 565

behavior is summarized in Figure 8a the most plausible driving force could be explained by 566

the NMR results collected on the dry sample G-C181-OH concomitantly occupies two 567

molecular environments characterized by 1) two distinct conformations of the sugar 568

displaying equivalent mobility and 2) two conformations of the acyl chain unsaturation 569

570

23

571

Figure 8 ndash a) Thermotropic phase behavior of G-C181-OH at 5 mgmL-1 Note that Tm2 is observed on the 572

dry powder only b) Possible relationship between the molecular environment as derived from NMR and 573

membrane curvatures All the molecular drawings refer to the same G-C181-OH molecule (bulky ellipsoid 574

equiv sugar headgroup) while each color corresponds to a given configuration Grey of C1rsquo1 and C1rsquo2 indicates 575

the uncertainty about the assignment of the specific NMR peak The curvature C is defined as 1R R being 576

the curvature radius R1 R2 radius of curvature CG Gauss curvature CM Mean curvature c) Free 577

interpretation of the anisotropic high-curvature (CMgt 0 CG= 0) inclusions in the membrane (low CM and 578

CG) continuum driven by inter-sugar H-bonding and reduced intramembrane diffusion for a G-C181-OH 579

glucolipid 580

581

Flat crystalline structure

Nanotube-vesicle network

Vesicles

48degC

120deg

106 105 104 103 102 101

ppm

H9 H10H1rsquo

106 105 104 103 102 101

ppm106 105 104 103 102 101

ppm106 105 104 103 102 101

ppm

C=CC=C

C1rsquo1C1rsquo2

C1rsquo

Tm2(96degC)Tm1

a)

b)

c)

Same molecule different local orientations and sugar conformations

High curvature membrane

Low curvature inclusion

Flip-FlopStrong H-bonding

CG

CM ( + )

CG 0

CM 0 CG 0

CM 0

G-C181-OH

24

Two related questions are unanswered at the moment What actually drives nanotube 582

formation and is the organization of G-C181-OH within the vesicles and tubes the same Lipid 583

nanotubes have been studied for at least three decades over which broad experimental and 584

theoretical insight has been generated Despite this fact the actual understanding is still matter 585

of debate27 Micelle-to-nanotube and vesicle-to-nanotube phase transitions are generally 586

reported for single amphiphiles below a given Tm and intermediate phase composed of twisted 587

and flat fibers is concomitantly reported Initially thought to be a chirality-driven phenomenon 588

recent studies propose a more general mechanism involving crystallization and symmetry 589

breaking in the lipid packing driven by optimal packing27 If twisted ribbons were observed 590

before for a number of microbial glycolipid bolaamphiphiles44 they are not found for G-C181-591

OH This fact as well as the large body of cryo-TEM images and the observation of nanotubes 592

above Tm1 rather pledge in favour of a direct topological change from positive (vesicles) to 593

zero (nanotubes) Gaussian curvatures In this case vesicle-to-nanotube transitions could be 594

driven by external forces56307813ndash162829 595

Considering the absence of external forces in the present system there are only two 596

possibilities left spontaneous budding of the membrane or nanotube pulling after inter-vesicle 597

collisions due to Brownian motion Both are possible as inclusions composed of a different 598

arrangement of the same molecule in the membrane continuum can theoretically explain 599

spontaneous nanotube formation in an isotropic medium672223629ndash1218ndash21 However the second 600

one is probably more likely to prevail because it could explain the large number of nanotubes 601

connecting two distant vesicles Interestingly similar events occur for any phospholipid-based 602

vesicular solution but nanotubes are extremely rare unless at least two lipids of different 603

molecular structure are mixed together The spectroscopic 1H and 13C and relaxation (1H) 604

ssNMR data even if collected on a dry powder demonstrate that the coexistence of two G-605

C181-OH configurations involving both the sugar headgroup and fatty alcohol chain are 606

possible for this molecule As a matter of fact this is not so surprising as more than one energy 607

minima each corresponding to a given conformation are known for glucose its non-reducing 608

derivative methylglucose and carbohydrates in general78ndash80 609

The presence of two configurations could explain the coexistence between two 610

structures of different curvature but the data presented so far do not explain the origin of these 611

two configurations In other words what drives the change in the local structure of the sugar 612

headgroup and fatty alcohol within the membrane 613

The association of known self-assembled structures to spectroscopic 13C ssNMR data is 614

known for years in particular for gluconamides lipids having a linear sugar headgroup81ndash84 It 615

25

was found that gauche effects in some cases up to the γ position in the sugar chain induce 616

upfield chemical shifts while anti and trans conformations result in downfield shifts Each 617

structure (monolayer crystal bilayer crystal or micellar fiber) was then characterized by a given 618

organization of the gluconamide and a characteristic 13C ssNMR fingerprint which could be 619

explained by gauche γ-gauche anti and trans conformations of the headgroup Similar 620

structural-spectroscopic correlations have been collected over the years for much simpler cyclic 621

sugars85 whereas pyranoses can undergo important conformational changes (chair half-chair 622

boat skew) when the appropriate energy barriers are crossed78ndash8086 A quantitative 623

conformational study based on the 13C chemical shift variations shown in Table 1 would be 624

certainly helpful but very delicate in the context of the present state of the art because a 625

rigorous approach requires at least one known crystal structure8182 which is currently missing 626

for this compound A qualitative approach will then be adopted 627

G-C181-OH is a bolaform amphiphile and flip-flop could be a possible mechanism 628

involved in the nanotube stabilization whereas head-tail interactions have long been considered 629

as important in nanotube-forming bolaform amphiphiles6063 From mere considerations based 630

on T2 (1H) relaxation arguments one could exclude head-tail configurations if the R-CH2OH 631

endgroup (H1) was involved in a direct interaction with the sugar group its corresponding T2 632

would probably be comparable to the sugarrsquos one (OH and H1rsquo) but this does not seem to be 633

the case However this hypothesis would be based on the comparison of the relaxation from 634

CH and CH2 protons additionally involved in different kind of motions Besides the relaxation 635

of H1 is influenced by the strong overlap between H1 and H5rsquo For these reasons relaxation 636

arguments are not solid enough and will not be used 637

We then address this question by exploring intermolecular interactions in a direct 638

manner using 2D 1H-1H homonuclear ssNMR correlation spectroscopy At 90degC in the fast-639

tumbling domain of molecular motion (narrow 1H NMR peaks Figure 6a) and corresponding 640

to the vesicle-rich region spin diffusion (NOESY) experiments are preferred over dipolar 641

coupling based experiments because such interaction is averaged out by molecular motion 642

However when the molecular motion is more constrained in the domain of nanotube stability 643

at 70degC (broad 1H spectra Figure 6a) Double-Quantum Single-Quantum (DQ-SQ) correlation 644

experiments are best suited to probe interaction between close molecular neighbors These 645

experiments are reported in Figure 9 and the related interpretation is only briefly reported 646

below while a more detailed discussion can be found on Page 12 in the Supporting Information 647

26

648

Figure 9 ndash a) 2D 1H-1H NOESY NMR experiment performed at MAS= 10 kHz (4 mm probe) on a Bruker 649

300 MHz Avance III HD spectrometer using 300 ms of mixing time Sample temperature is regulated to 650

90degC reached by cooling from 110degC b) 2D 1H-1H DQ-SQ (BAck-to-BAck BABA pulse sequence used for 651

the DQ excitation scheme) NMR experiment performed at MAS= 22 kHz (25 mm probe) on a Bruker 700 652

MHz Avance III spectrometer using a DQ excitation time equal to 2 rotor periods Sample temperature is 653

regulated to 70degC reached by cooling from 110degC The typical schemes illustrating how to interpret these 654

2D experiments are reported above each experimental map Zooms of the sugar region and configurational 655

interpretation are given below each 2D map More information are given in the text below 656

H16-H16

H1-H1

H9-H10

H1rsquo-H1

H1rsquoH910 H16

H1

H1rsquo-H1rsquo

Hxrsquo-Hxrsquo

2

(pp

m)

(ppm)

315

811910 1rsquo

2rsquo2

3rsquo 4rsquo6rsquo

5rsquo

1rsquo2910

asymp asymp asympHDO MeOD Acetone

1

315

811 9101rsquo

2rsquo2

3rsquo4rsquo6rsquo

5rsquo

1rsquo2 9 10

asympasympasympHDOMeODAcetone

H1rsquo-H1

1rsquo

1

(p

pm

)

(ppm)

H1-H2

H1rsquo-H1

H1rsquo

H910

H1H5rsquo

H2

H1rsquo-H5rsquo

315

811910 1rsquo

2rsquo2

3rsquo 4rsquo6rsquo

5rsquo

1rsquo2910

asymp asymp asympHDO MeOD Acetone

1

H2rsquoH3rsquoH4rsquo

H2rsquoH3rsquoH4rsquo

X= 2rsquo3rsquo4rsquo

NOESY - T= 90degC

1H

1H

1H

1H

DQ-SQ - T= 70degC

a) b)

2

(pp

m)

(p

pm

)

(ppm)

A B

AhellipB

AhellipA

BhellipB

NOESY

(ppm)

A B

AhellipB

DQ-SQ

Single-quantum dimension

Do

ub

le-q

ua

ntu

m d

imen

sio

n

AhellipA

BhellipB

BhellipA

BhellipA

315

811910 1rsquo

2rsquo2

3rsquo 4rsquo6rsquo

5rsquo

1rsquo2910

asymp asymp asympHDO MeOD Acetone

1

lt 5 Aring

lt 5 Aring

H1-H2

H811

H15

H16

H6rsquo

H1rsquo

H1

H5rsquo

H2

H1-H2

H1rsquo-H1

H1rsquo-H5rsquo

H1rsquo

H1 H5rsquo

H1rsquo

H1

H5rsquo

H1H2rsquo H3rsquo H4rsquo

H1-H1

H1rsquo-H1

Hxrsquo-HxrsquoX= 2rsquo3rsquo4rsquo

H1-H2

H1rsquo

Zoom sugar region 28 ndash 45 ppmNOESY DQ-SQ

27

Head-tail or head-head configurations could be probed by the existence of H1rsquo-H1 657

interactions whereas H1 (δH= 356 ppm) is associated to the glucolipid tail and H1rsquo (δH= 432 658

ppm) to the sugar headgroup These atoms are more than 15 Aring apart and a direct interaction 659

could not be observed in both NOESY and DQ-SQ experiments NOESY experiments (Figure 660

9a) show that H1rsquo-H1 correlations are not observed Meanwhile clear-cut direct H1rsquo-H1 661

interactions are indeed observed are detected at T= 70degC (DQ cross-peaks related to H1rsquo-H1 662

correlation in DQ-SQ experiment shown in Figure 9b) Interestingly the absence of auto peaks 663

at δH= 432 ppm and δH= 335 ppm in the SQ dimension excludes H1rsquo-H1rsquo and Hxrsquo-Hxrsquo (x= 2 664

3 4) interactions respectively strengthening the idea that sugar headgroups from two nearest-665

neighbor G-C181-OH molecules are not adjacent in the nanotube-vesicle region 666

Homonuclear 2D ssNMR supports the idea according to which G-C181-OH can 667

undergo molecular flip-flops and a head-tail arrangement in the dry product Extrapolating these 668

conclusions in the lower concentration range lack of H1rsquo-H1 proximity at T= 90degC suggests a 669

head-head configuration in the vesicle-rich region and head-tail configuration within the 670

nanotubes It is however still unclear whether flip-flops are driven by spontaneous temperature-671

induced conformational changes in the glucolipid packing stimulating inclusions within the 672

membrane itself672223629ndash1218ndash21 or by defects generated by inter-vesicle collisions In fact a 673

combination of both is not unlikely Nanotubes are not stable when temperature is 20 to 30 674

degrees above Tm1 indicating that inter-vesicle collisions alone cannot explain the unusual 675

phenomenon In an energy landscape approach87 the head-head configuration is certainly 676

favored at high temperature while both head-head and head-tail configurations could be 677

favored in the mid-temperature region corresponding to the range between Tm1 and Tm2 in the 678

dry powder Both statistical and defect-induced flip-flop events contribute to form high mean 679

zero Gaussian curvature morphologies (nanotubes) for which the head-tail conformation 680

limiting the proximity between bulky sugar headgroups is probably more adapted (Figure 8c) 681

The flip-flop interpretation now reasonably explains the strong differences in the sugar and fatty 682

alcohol conformations deduced by the multiple NMR signals attributed to chemically-683

equivalent groups The existence of various low-energy conformations each having specific 684

NMR signature for sugars in solution78ndash8086 and solid-state81ndash84 further supports our 685

interpretation 686

One last point should be clarified Why between Tm1 and Tm2 are two molecular 687

environments able to coexist instead of a single one of average rigidity as found in the case of 688

ldquoripplerdquo phases50ndash54 We believe that this question can be answered based on the NMR results 689

the relatively short T2 values of the H1rsquo and OH protons shows the constrained mobility of the 690

28

sugar headgroup despite a certain flexibility of the chain (long T2 component) This trend may 691

suggest restrictions in translational mobility probably due to the sugar-OH hydrogen bonding 692

(Figure 8c) meaning that lipids are not able to diffuse within the membrane thus keeping two 693

different configurations stable 694

695

Conclusions 696

Topologically-connected nanotube-vesicle lipidic systems are commonly observed for 697

self-assembled membranes composed of at least two amphiphiles and spontaneously driven by 698

anisotropic inclusions andor by external directional forces This work shows that strong 699

differences in membrane curvature can spontaneously occur in single-lipid systems without 700

external anisotropic forces Coexistence and interconnection between nanotubes and vesicles 701

are evidenced via cryo-TEM and fluorescence confocal microscopy in a single-phase region at 702

temperature above a transition temperature Tm1= 483degC determined by DSC on the hydrated 703

sample but below the second transition temperature Tm2 (957degC) determined by DSC on the 704

dry powder X-ray diffraction (WAXS) also shows the coexistence of two lattice periods one 705

attributed to an inter-membrane repeating distance and the other possibly related to the intra-706

nanotube glucolipid arrangement Labelling the systems with Laurdan a fluorophore of which 707

the emission strongly varies with the local membrane rigidity excludes the presence of rigid 708

micro-separated phases (eg Lβ) 709

1H and 13C ssNMR despite the fact that it can only be performed on the dry powder 710

confirm the presence of one main fluid G-C181-OH environment above Tm2 Below Tm2 the 711

splitting of some 13C peaks and the coexistence of short and long 1H relaxation components 712

demonstrate that G-C181-OH can simultaneously pack into structurally and dynamically 713

inequivalent environments Combining advanced 2D 1H-1H correlation experiments based on 714

spin diffusion or double-quanta excitation suggests that vesicles could contain a preferred head-715

head configuration of the bolaform lipid while nanotubes a preferred head-tail assembly 716

The presence of two environments composed of the same molecule characterized by 717

different molecular configurations and rigidity fulfills the theoretical conditions necessary to 718

spontaneously nucleate a membrane inclusion of non-zero average orientation described as 719

necessary to observe topologically-connected nanotube-vesicle structures in the absence of 720

external forces This hypothesis could also be enriched by the possibility of nanotubes to be 721

pulled away between two vesicles after a collision event 722

This work expands the actual theoretical and experimental framework on lipid networks 723

stabilized by a different radius of curvature It could also motivate the synthesis or isolation 724

29

of new lipids providing a higher level of 3D architectural control in soft networks possibly in 725

the range of room temperature 726

727

Acknowledgements 728

Dr Andrea Lassenberger and Dr Sylvain Preacutevost at Institut Laue Langevin (ILL Grenoble 729

France) are kindly acknowledged for their assistance on the SANS experiment Confocal 730

images were performed at the Institut de Biologie Paris Seine (IBPS) imaging facility Jean-731

Franccedilois Gilles (Imaging Core Facility Institut de Biologie Paris Seine (IBPS) CNRS 732

Sorbonne Universiteacute Paris France) is kindly acknowledged for his assistance on the confocal 733

microscopy imaging Dr Guillaume Laurent (LCMCP Sorbonne Universiteacute Paris France) is 734

kindly acknowledged for helpful discussions The French region Ile-de-France SESAME 735

program is acknowledged for financial support (700 MHz NMR 736

spectrometer) Dr Olivier Diat (ICSM Marcoule France) Prof eronikaIglič and Prof V Aleš737

onsul discussi Slovenia) are kindly acknowledged for helpfUniversity of LjubljanaIglič (-Kralj 738

739

Financial support 740

WAXS experiments have been supported by Soleil Light Source Saint Aubin France proposal 741

Ndeg 20201747 SANS experiments have been supported by ILL proposal Ndeg9-13-778 742

743

References 744

1 D Kabaso N Bobrovska W Goacuteźdź E Gongadze V Kralj-Iglič R Zorec and A 745

Iglič Bioelectrochemistry 2012 87 204ndash210 746

2 C R Safinya U Raviv D J Needleman A Zidovska M C Choi M A Ojeda-747

Lopez K K Ewert Y Li H P Miller J Quispe B Carragher C S Potter M W 748

Kim S C Feinstein and L Wilson Adv Mater 2011 23 2260ndash2270 749

3 P Veranič M Lokar G J Schuumltz J Weghuber S Wieser H Haumlgerstrand V Kralj-750

Iglič and A Iglič Biophys J 2008 95 4416ndash4425 751

4 A Rustom R Saffrich I Markovic P Walther and H-H Gerdes Science (80- ) 752

2004 303 1007ndash1010 753

5 M I Angelova S Soleacuteau P Meacuteleacuteard F Faucon and P Bothorel Prog Colloid 754

Polym Sci 1992 89 127ndash131 755

6 L Mathivet S Cribier and P F Devaux Biophys J 1996 70 1112ndash1121 756

7 V Kralj-Iglič G Gomišček J Majhenc V Arrigler and S Svetina Colloids Surfaces 757

A Physicochem Eng Asp 2001 181 315ndash318 758

30

8 V Kralj-Iglič A Iglič G Gomišček F Sevšek V Arrigler and H Haumlgerstrand J 759

Phys A Math Gen 2002 35 1533ndash1549 760

9 M Drab D Stopar V Kralj-Iglič and A Iglič Cells 2019 8 626 761

10 A Zidovska K K Ewert J Quispe B Carragher C S Potter and C R Safinya 762

Langmuir 2009 25 2979ndash2985 763

11 A Zidovska K K Ewert J Quispe B Carragher C S Potter and C R Safinya Soft 764

Matter 2011 7 8363ndash8369 765

12 A Zidovska K K Ewert J Quispe B Carragher C S Potter and C R Safinya 766

Biochim Biophys Acta - Biomembr 2009 1788 1869ndash1876 767

13 M Karlsson K Sott M Davidson A S Cans P Linderholm D Chiu and O Orwar 768

Proc Natl Acad Sci U S A 2002 99 11573ndash11578 769

14 A Karlsson R Karlsson M Karlsson A-S Cans A Stromberg F Ryttsen and O 770

Orwar Nature 2001 409 150ndash152 771

15 M Karlsson K Sott A S Cans A Karlsson R Karlsson and O Orwar Langmuir 772

2001 17 6754ndash6758 773

16 E Evans H Bowman A Leung D Needham and D Tirrell Science (80- ) 1996 774

273 933ndash935 775

17 U Seifert Configurations of fluid membranes and vesicles 1997 vol 46 776

18 J Kaumls and E Sackmann Biophys J 1991 60 825ndash844 777

19 N Bobrovska W Goacuteźdź V Kralj-Iglič and A Iglič PLoS One 2013 8 e73941 778

20 A Iglič B Babnik K Bohinc M Fošnarič H Haumlgerstrand and V Kralj-Iglič J 779

Biomech 2007 40 579ndash585 780

21 K Bohinc T Slivnik A Iglič M Brumen and V Kralj-Iglič J Phys Chem C 2007 781

111 9709ndash9718 782

22 A Iglič H Haumlgerstrand P Veranič A Plemenitaš and V Kralj-Iglič J Theor Biol 783

2006 240 368ndash373 784

23 V Kralj-Iglič and P Veranič Adv Planar Lipid Bilayers Liposomes 2006 5 129ndash785

149 786

24 A Iglic and V Kralj-Iglic Adv Planar Lipid Bilayers Liposomes 2006 4 253ndash279 787

25 S Penič L Mesarec M Fošnarič L Mroacutewczyńska H Haumlgerstrand V Kralj-Iglič and 788

A Iglič Front Phys 2020 8 342 789

26 P Yager and P E Schoen Mol Cryst Liq Cryst 1984 106 371ndash381 790

27 T G Barclay K Constantopoulos and J Matisons Chem Rev 2014 114 10217ndash791

10291 792

31

28 B Božič G Gomišček V Kralj-Iglič S Svetina and B Žekš Eur Biophys J 2002 793

31 487ndash496 794

29 A Iglič H Haumlgerstrand M Bobrowska-Haumlgerstrand V Arrigler and V Kralj-Iglič 795

Phys Lett A 2003 310 493ndash497 796

30 R Bar-Ziv and E Moses Phys Rev Lett 1994 73 1392ndash1395 797

31 S Chiruvolu H E Warriner E Naranjo S H Idziak J O Raumldler R J Plano J A 798

Zasadzinski and C R Safinya Science (80- ) 1994 266 1222ndash1225 799

32 V Kralj-Iglič A Iglič H Haumlgerstrand and P Peterlin Phys Rev E 2000 61 4230ndash800

4234 801

33 L Mesarec M Drab S Penič V Kralj-Iglič and A Iglič Int J Mol Sci 2021 22 802

2348 803

34 L Van Renterghem S L K W Roelants N Baccile K Uyttersprot M C Taelman 804

B Everaert S Mincke S Ledegen S Debrouwer K Scholtens C V Stevens W 805

Soetaert L Van Renterghem S L K W Roelants N Baccile K Uytersprot M C 806

Taelman B Everaert S Mincke S Ledegen S Debrouwer K Scholtens C V 807

Stevens and W Soetaert Biotechnol Bioeng 2018 115 1195ndash1206 808

35 J Clark and F Deswarte Introduction to Chemicals from Biomass Second Edition 809

John Wiley amp Sons 2015 810

36 K V Sajna R Houmlfer R K Sukumaran L D Gottumukkala and A Pandey in 811

Industrial Biorefineries and White Biotechnology eds A Pandey R Houmlfer M 812

Taherzadeh K M Nampoothiri and C Larroche Amsterdam Oxford Waltham 813

Elsevier 2015 pp 499ndash521 814

37 R Marchant and I M Banat Trends Biotechnol 2012 30 558ndash565 815

38 K Hill and C LeHen-Ferrenbach in Sugar-Based Surfactants Fundamentals and 816

Applications ed C C Ruiz Boca Raton CRC Press 2009 pp 1ndash20 817

39 L Rodrigues I M Banat J Teixeira and R Oliveira J Antimicrob Chemother 818

2006 57 609ndash618 819

40 J D Desai and I M Banat Microbiol Mol Biol Rev 1997 61 47ndash64 820

41 S L K W Roelants K Ciesielska S L De Maeseneire H Moens B Everaert S 821

Verweire Q Denon B Vanlerberghe I N A Van Bogaert P Van der Meeren B 822

Devreese and W Soetaert Biotechnol Bioeng 2016 113 550ndash559 823

42 N Baccile M Selmane P Le Griel S Preacutevost J Perez C V Stevens E Delbeke S 824

Zibek M Guenther W Soetaert I N A Van Bogaert and S Roelants Langmuir 825

2016 32 6343ndash6359 826

32

43 N Baccile A-S Cuvier S Preacutevost C V Stevens E Delbeke J Berton W Soetaert 827

I N A Van Bogaert and S Roelants Langmuir 2016 32 10881ndash10894 828

44 N Baccile C Seyrig A Poirier S A Castro S L K W Roelants and S Abel Green 829

Chem 2021 23 3842ndash3944 830

45 A Salis L Cappai C Carucci D F Parsons and M Monduzzi J Phys Chem Lett 831

2020 11 6805ndash6811 832

46 J Schindelin I Arganda-Carreras E Frise V Kaynig M Longair T Pietzsch S 833

Preibisch C Rueden S Saalfeld B Schmid J-Y Tinevez D J White V 834

Hartenstein K Eliceiri P Tomancak and A Cardona Nat Methods 2012 9 676ndash835

682 836

47 T Parasassi and E Gratton J Fluoresc 1995 5 59ndash69 837

48 M M Dodes Traian F L Gonzaacutelez Flecha and V Levi J Lipid Res 2012 53 609ndash838

616 839

49 D M Owen C Rentero A Magenau A Abu-Siniyeh and K Gaus Nat Protoc 840

2012 7 24ndash35 841

50 W Stillwell in An Introduction to Biological Membranes Academic Press 2016 pp 842

135ndash180 843

51 W Stillwell in An Introduction to Biological Membranes Academic Press 2016 pp 844

181ndash220 845

52 B Tenchov Chem Phys Lipids 1991 57 165ndash177 846

53 R Koynova and M Caffrey Biochim Biophys Acta 1998 1376 91ndash145 847

54 F Schmid S Dolezel O Lenz and S Meinhardt J Phys Conf Ser 2014 487 848

012004 849

55 S Meinhardt R L C Vink and F Schmid Proc Natl Acad Sci U S A 2013 110 850

4476ndash4481 851

56 O Glatter and O Kratky Small Angle X-ray Scattering Academic Press London 852

1982 853

57 B A Cunningham A D Brown D H Wolfe W P Williams and A Brain Phys 854

Rev E 1998 58 3662ndash3672 855

58 M Nadler A Steiner T Dvir O Szekely P Szekely A Ginsburg R Asor R Resh 856

C Tamburu M Peres and U Raviv Soft Matter 2011 7 1512ndash1523 857

59 C Tanford The hydrophobic effect Formation of micelles and biological membranes 858

John Wiley amp Sons Inc New York 1973 859

60 M Masuda and T Shimizu Langmuir 2004 20 5969ndash5977 860

33

61 A Zidovska K K Ewert J Quispe B Carragher C S Potter and C R Safinya 861

Block Liposomes Vesicles of Charged Lipids with Distinctly Shaped Nanoscale 862

Sphere- Pear- Tube- or Rod-Segments Elsevier Inc 1st edn 2009 vol 465 863

62 A Iglic V Kralj-Iglic B Božič M Bobrowska-Haumlgerstrand B Isomaa and H 864

Haumlgerstrand Bioelectrochemistry 2000 52 203ndash211 865

63 T Shimizu M Masuda and H Minamikawa Chem Rev 2005 105 1401ndash1444 866

64 J V Selinger and J M Schnur Phys Rev Lett 1993 71 4091ndash4094 867

65 J M Schnur Science (80- ) 1993 262 1669ndash1676 868

66 B Tenchov R Koynova and G Rapp Biophys J 2001 80 1873ndash1890 869

67 R Koynova and B Tenchov Ser Biomech 2017 31 3ndash19 870

68 S J Angyal and R Le Fur Carbohydr Res 1984 126 15ndash26 871

69 J K Whitesell and M A Minton J Am Chem Soc 1987 109 225ndash228 872

70 T Takayama S Ando and I Ando J Mol Struct 1990 220 245ndash250 873

71 K Gawrisch N V Eldho and I V Polozov Chem Phys Lipids 2002 116 135ndash151 874

72 F Tielens C Gervais J F Lambert F Mauri and D Costa Chem Mater 2008 20 875

3336ndash3344 876

73 G A Jeffrey and Y Yeon Acta Crystallogr Sect B 1986 42 410ndash413 877

74 B Berglund and R W Vaughan J Chem Phys 1980 73 2037ndash2043 878

75 C Gervais C Coelho T Azaiumls J Maquet G Laurent F Pourpoint C Bonhomme 879

P Florian B Alonso G Guerrero P H Mutin and F Mauri J Magn Reson 2007 880

187 131ndash140 881

76 C Akita T Kawaguchi F Kaneko H Yamamoto and M Suzuki J Phys Chem B 882

2004 108 4862ndash4868 883

77 R N Purusottam L Seacutenicourt J J Lacapegravere and P Tekely Biochim Biophys Acta - 884

Biomembr 2015 1848 3134ndash3139 885

78 X Biarneacutes A Ardegravevol A Planas C Rovira A Laio and M Parrinello J Am Chem 886

Soc 2007 129 10686ndash10693 887

79 L M J Kroon‐Batenburg and J Kroon Biopolymers 1990 29 1243ndash1248 888

80 K N Kirschner and R J Woods Proc Natl Acad Sci U S A 2001 98 10541ndash889

10545 890

81 S Svenson B Kirste and J H Fuhrhop J Am Chem Soc 1994 116 11969ndash11975 891

82 I Sack S Macholl J H Fuhrhop and G Buntkowsky Phys Chem Chem Phys 892

2000 2 1781ndash1788 893

83 S Svenson J Koening and J H Fuhrhop J Phys Chem 1994 98 1022ndash1028 894

34

84 S Svenson A Schaefer and J H Fuhrhop J Chem Soc Perkin Trans 2 1994 2 895

1023ndash1028 896

85 K Bock and C Pedersen Adv Carbohydr Chem Biochem 1983 41 27ndash66 897

86 T B Grindley Structure and Conformation of Carbohydrates Springer Berlin 898

Heidelberg 2008 899

87 F Tantakitti J Boekhoven X Wang R V Kazantsev T Yu J Li E Zhuang R 900

Zandi J H Ortony C J Newcomb L C Palmer G S Shekhawat M O De La 901

Cruz G C Schatz and S I Stupp Nat Mater 2016 15 469ndash476 902

903

1

Supporting Information 904

905

Topological connection between vesicles and nanotubes in single-906

component lipid membranes driven by head-tail interactions 907

908

Niki Baccilea Ceacutedric Lorthioira Abdoul Aziz Baa Patrick Le Griela Cristina Coelhob Javier 909

Perezc Wim Soetaertd Sophie L K W Roelantsd 910

911

a Sorbonne Universiteacute Centre National de la Recherche Scientifique Laboratoire de Chimie de 912

la Matiegravere Condenseacutee de Paris LCMCP F-75005 Paris France 913

b Institut des Mateacuteriaux de Paris Centre Sorbonne Universiteacute Paris France 914

c Synchrotron Soleil LrsquoOrme des Merisiers Saint-Aubin BP48 91192 Gif-sur-Yvette Cedex 915

France 916

d InBio Department of Biotechnology Ghent University Ghent Belgium 917

918

919

Corresponding author 920

Dr Niki Baccile 921

E-mail address nikibaccilesorbonne-universitefr 922

Phone +33 1 44 27 56 77 923

924

925

926

2

a) 927

Parameter Method

Dry Matter (DM ) 996 Infrared balance 105degC

Glucose () 012 HPLC-Metacarb

Glycerol () nd HPLC-Metacarb

Free Fatty Acid content () lt001 Internal method BBEPP

GC with FID Detector

Oil () lt001 Internal method BBEPP

GC with FID Detector

Moulds (CFUg) lt10 3M-Nordval ndeg16 B

Yeasts (CFUg) lt10 3M-Nordval ndeg16 B

Anaerobic count (CFUg) lt10 SP-VG M005 B

Protein () tbd BCA protein assay

928

Figure S 1 - a) HPLC-ELSD chromatogram and composition table 929

930

931

932

933

3

b) 934

935

Group Ndeg δ ppm

CH2 1 354

CH2 2 153

CH2 3-712-14 132

CH2 811 203

CH 910 534

CH2 15 162

CH2 16 388

CH 1 424

CH 2 316

CH 3 334

CH 4 329

CH 5 354

CH2 6 366

Figure S 1 (next) ndash b) 1H NMR spectrum (MeOD-d4) recorded on the purified G-C181-OH compounds 936

and related assignment 937

938

2

811910 1rsquo

5rsquo1

3rsquo4rsquo

1rsquo1910

asymp asymp asympHDO MeOD Acetone

6rsquo2rsquo

1615

3-712-14

4

939

940

Figure S 2 ndash Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 130degC 941

5

942

943

Figure S 3 - Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 90degC 944

945

90degC

1

22

6

946

947

948

949

Figure S 4 ndash Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 70degC 950

951

21

11

1

2

5

3

4

45

5

6

7

7

7

952

953

954

Figure S 5 ndash Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 25degC 955

956

957

1

1 1

8

958

Figure S 6 ndash a) SANS experiment recorded at 90degC for a 5 mgmL G-C181-OH aqueous solution b-c) 959

WAXS experiments recorded on aqueous solutions of G-C181-OH at b) 5 mgmL and c) 25 mgmL d) 960

Structural interpretation of the repeating distances pointed out in c) 961

962

963

964

001 01 1

q Aring-1

25degC

68degC

88degC

111degC

01 1q Aring

-1

02 04 06 08 1q Aring

-1

001 01

10-1

100

101

Inte

nsity

cm

-1

q Aring-1

-2

T= 90degC

SANS WAXS WAXS

a) b) c)

dv

dL d= 144 Aring-1

di

ddH2O

dt

dvdL

dtdi d

d)

9

965

966

Figure S 7 ndash ssNMR MAS (10 kHz) experiments recorded on a G-C181-OH dry powder a) 1H spectra 967

recorded from 110degC to 70degC b) 13C single pulse (SP) spectra recorded with high power 1H decoupling from 968

110degC to 70degC c) 13C cross polarization (CP) spectra recorded at 110degC and contact time tc of 50 micros and 969

3000 micros d) 13C CP spectra recorded at 70degC and for 20 lt tc micros lt 3000 970

971

1

811

910

CH2 CH2

1rsquo

6rsquo

8 6 4 2 0 ppm

110

90

80

70

910

1rsquo

6rsquo16

1

2rsquo3rsquo4rsquo

811

OH

45 40 35 30 25 20 15

ppm

200 180 160 140 120 100 80 60 40 20 0

ppm140 120 100 80

ppm

CH2-gauche

3004

110

90

80

70

65

60

50

25

C91013003

C9 - 12869

CH2-all-trans

3198

C1rsquo10335

a) b)

1H

T degCT degC

13C Single Pulse MAS

A

B

A

B

70degC70degC

5rsquo

215

200 180 160 140 120 100 80 60 40 20 0

ppm

13C CPMAST= 110degC

200 180 160 140 120 100 80 60 40 20 0

ppm

132 131 130 129 128 127 126

ppm

106 105 104 103 102

ppm

13C CPMAST= 70degC

50

tc micros

3000

20

tc micros

50 500 1000 3000

tc micros

C102

13024 C910113003

C92

12869

C1rsquo213082

C1rsquo110443

A B

106 105 104 103 102

ppm

3000

1000

500

50

20

AB

c)

d)

10

972

Figure S 8 ndash a) Short and d) long T2 components displayed by the 1H transverse relaxation signal c) 1H 973

transverse relaxation signals recorded on a G-C181-OH dry powder from 110degC to 70degC The area of each 974

given resonance is plotted (log-lin) as a function of the echo time (CPMG pulse sequence) for each 975

temperature and fitted using the equation given above T2s and T2l indicate the short and long T2 relaxation 976

times respectively obtained by describing the 1H transverse relaxation signal using a biexponential decay 977

For H1rsquo H5rsquo and H6rsquo the 1H relaxation function could be described by a monoexponential decay 978

- Protons from the sugar group (H1rsquo H5rsquo H6rsquo) show a single T2 relaxation component (Figure 979

6c and Figure S 8) while protons of the acyl chain (CH2 H910 H811) show both short (Figure 980

6c) and long (Figure 6d) T2 relaxation component 981

- The short T2 relaxation time systematically decreases from 110degC to 70degC for all species by a 982

factor ranging from two to seven depending on the specific chemical group Very long T2 are 983

observed at 110degC for all acyl species while nearly unchanged values are measured below 984

90degC without any clear trend 985

- The sugar protons (H1rsquo H5rsquo H6rsquo) have the shortest set of T2 values (lt 1 ms) below 100degC 986

The long component of T2 generally ranges between 4 ms and 12 ms 987

0 1 2 3 4 5 6 7 8

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1 2 3 4 5 6 7 8

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

H1

H8H11

H9H10

CH2

H1rsquo

H6rsquo

CH2

H910H811 H1CH2

H1rsquo H5rsquo6rsquo

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

OH

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45 T= 110degC

T= 90degC

T= 80degC

T= 70degC

Log

In

ten

sity

Time ms

Fit

)

CH2 H910 H811 H1 H56 H1 OH

00

05

10

15

20

25

30

35

40

T2 m

s

T= 110degC

T= 90degC

T= 80degC

T= 70degC

CH2 H910 H811 H1

0

2

4

6

8

10

12

T2 m

s

T= 110degC

T= 90degC

T= 80degC

T= 70degC

Short T2(1H) Long T2(1H)

a) b)

c)

11

Table S 1 ndash 1H chemical shifts of G-C181-OH in solution and in the solid-state Solid-state spectra are 988

recorded at a MAS frequency of 10 kHz in DMSO-d6 broad peak 989

990

δH ppm

Group Ndeg Liquid

(MeOD-d4) Solid

(T= 110degC) Solid

(T= 70degC)

CH2 H1 354 357

CH2 H2 153 156

CH2 H3-712-14 132 135

CH2 H811 203 203

CH H910 534 534

CH2 H15 162 164

CH2 H16 388 383

CH H1 424 432

CH H2 316 335

CH H3 334 335

CH H4 329 335

CH H5 354 351 357

CH2 H6 366 383

OH OH1 432

45 463

OH OH2 492

OH OH3 490

OH OH4 487

OH OH6 445

991

992

H1H9H10H1rsquo

H6rsquo

OH2rsquo

OH1

12

Detailed discussion about the NOESY and DQ-SQ 2D 1H-1H ssNMR experiments presented in 993

Figure 9 994

995

2D 1H-1H ssNMR experiments are a unique tool to explore through-space proximities 996

between 1H nuclei ssNMR offers a large number of possibilities each adapted to a given 997

system The advantage of 2D maps with respect to a dynamics approach is their direct 998

readability in clear two nuclei which exchange their magnetization through-space (whichever 999

mechanism may be responsible for this exchange) will display a specific cross-peak However 1000

the way cross-peaks are presented and interpreted strictly depends on the nature of the pulse 1001

program which in turn is chosen according to the physical nature of the sample 1002

1D temperature-dependent 1H NMR experiments performed on G-C181-OH (Figure 1003

4a in the main text) indicate sharp peaks at 90degC and broad peaks at 70degC Narrow peaks 1004

indicate fast andor large-amplitude molecular tumbling averaging the strong 1H-1H dipolar 1005

interactions This excludes the use of pulse sequences like DQ-SQ which is not efficient under 1006

these conditions For this reason we employ spin-diffusion-based pulse sequences like 1007

NOESY which is more adapted for samples with high molecular mobility When mixing times 1008

are chosen to be short (lt 1 s) one generally explores through-space proximities within about 5 1009

10 Aring On the contrary when molecular motions are restricted dipolar coupling becomes 1010

significant 1H NMR peaks are broader and pulse sequences like DQ-SQ can now be used 1011

with a spatial resolution of less than 5 Aring 1012

Whenever possible DQ-SQ experiments are preferred over spin diffusion-based pulse 1013

sequences because they are more selective in particular for likewise interactions which are 1014

not discriminated in NOESY-type experiments This is illustrated by the typical interaction 1015

schemes for NOESY (top Figure 9a) and DQ-SQ (top Figure 9b) For two given A and B 1016

protons 2D NOESY maps will display two on-diagonal and two square-shaped symmetrical 1017

off-diagonal cross-peaks Off-diagonal cross-peaks illustrate the mutual A-B and B-A 1018

interaction but the on-diagonal peaks are systematic and do not prove any A-A or B-B 1019

interaction Similarly to NOESY 2D DQ-SQ maps display off-diagonal peaks (symmetric 1020

across the diagonal top Figure 9b) upon A-B and B-A interaction but in contrast they will 1021

display on-diagonal peaks only in the presence of self-interaction In the tutorial example of 1022

Figure 9b top the on-diagonal A-A peak indicates mutual A-A interaction while lack of on-1023

diagonal B-B peak evidences the absence of B-B proximities Combining NOESY and DQ-SQ 1024

experiments is capital to study the local arrangement of G-C181-OH and its evolution with 1025

temperature 1026

13

At T= 90degC the H1 and H5rsquo contributions are very close but possible to discriminate 1027

them (δ(H1)= 354 ppm δ(H5rsquo)= 351 ppm) according to 1H NMR chemical shifts in Table S 1028

1 2D 1H-1H NOESY in Figure 9a also discriminates between the two components attributed 1029

on the basis of the interactions with their mutual neighbors (green H1-H2 and H5rsquo-H1rsquo squares 1030

in Figure 9a) On this basis 2D NOESY does not show a direct H1rsquo-H1 correlation (blue red-1031

crossed) but rather an extensive network of intra-sugar interactions These features exclude 1032

suggest intra- and inter-sugar interactions and they exclude head-tail arrangements as a major 1033

molecular configuration at 90degC 1034

At 70degC the 1H NMR peaks are broader indicating a reduction of the T2 (1H) relaxation 1035

times andor residual homonuclear dipolar interactions excluding the use of NOESY In this 1036

case 1H-1H DQ-SQ correlation experiments are best suited Figure 9b presents the 1H-1H DQ-1037

SQ experiment on G-C181-OH at 70degC Peaks are broader than in NOESY and attribution may 1038

be cumbersome However the experiment is performed at 700 MHz and MAS= 22 kHz which 1039

improves spectral resolution with respect to the same experiment performed at 300 MHz and 1040

MAS= 10 kHz The auto-peak at δH= 356 ppm in the SQ dimension is attributed to H1-H1 spin 1041

pairs mainly resulting from intra CH2 interaction while the DQ cross-peaks at δH= 156-356 1042

ppm and δH= 432-356 ppm respectively origin from H2-H1 and H1rsquo-H1 interactions (Figure 1043

9b) DQ-SQ map does not show any auto-peak corresponding to H1rsquo-H1rsquo or Hxrsquo-Hxrsquo (x= 2rsquo 1044

3rsquo 4rsquo) interactions (red-crossed blue Figure 9b) thus excluding any proximity (lt 5 Aring) between 1045

close glucose neighbors On the contrary DQ-SQ maps (recorded both at 700 MHz Figure 9b 1046

or at 300 MHz not shown) systematically show DQ cross-peaks which can only be attributed 1047

to H1rsquo-H1 interactions (δH= 432-356) The presence of these H1rsquo-H1 interactions can only be 1048

explained by a head-tail arrangement of the G-C181-OH molecules as illustrated below the 1049

DQ-SQ map in Figure 9b 1050

19

resonances broaden from 110degC to 70degC thus indicating slower andor more anisotropic 484

reorientational motions of all chemical groups The specific dynamic behaviour of the sugar 485

headgroup is also reflected by its 1H transverse relaxation time T2 whereas the shorter the T2 486

the higher the rigidity Experiments performed between 110degC and 70degC shown and 487

commented in Figure S 8 in the Supporting Information globally show a single short (lt 1 ms 488

below 100degC) T2 relaxation component for H1rsquo H5rsquo and H6rsquo and two both short and long 489

(between 4 ms and 12 ms) T2 for the acyl chain (CH2 H910 H811) even below 100degC 490

At 110degC 13C SPMAS experiments (Figure 6b full range in Figure 7 and Figure S 7 491

assignments in Table 1) show a single peak at δC= 10335 ppm for C1rsquo and at δC= 13003 ppm 492

for C910 as expected for a symmetrical conformation with two chemically-equivalent 493

carbons76 At 70degC the C1rsquo peak disappears while C910 now shows two peaks around 128 494

ppm and 130 ppm (Table 1 full temperature series in Figure S 7b) attributed to C92 (δC= 12869 495

ppm) and C10 in a non-symmetric configuration respectively for which these two carbons are 496

chemically inequivalent76 More precisely the resonance at 130 ppm is actually the 497

superposition of two close resonances C9101 (δC= 13003 ppm) and C102 (δC= 13024 ppm) 498

where subscripts 1 and 2 stand for two different conformations of the cis-olefin group as 499

reported for various phases of oleic acid in the vicinity of 0degC76 500

501

Table 1 ndash 13C Chemical shifts of G-C181-OH in solution and solid-state Solid-state experiments are 502

recorded at MAS= 10 kHz All spectra are reported in Figure 7 and Figure S 7 The contact time for cross 503

polarization is 3 ms In all ssNMR experiments temperature is first raised at 110degC and then lowered to 504

either 90degC or 70degC Attribution of the 13C resonances in Table 1 is performed on the basis of the solution 505

spectrum (Figure S 1b) Additional peaks observed in both single pulse and cross polarization ssNMR 506

experiments are attributed when possible to a configuration effect of the given atom eg C9101 and C92 507

indicate two configurations of C9 All new peaks observed in solid-state and not detected in solution are 508

highlighted in red for convenience 509

510

511

δ ppm

Peak Liquid (DMSO-d6)

Solid Single Pulse

T= 110degC

Solid Single Pulse

T= 90degC

Solid Single Pulse

T= 70degC

Solid Cross Polarization

T= 70degC

C102

- - 13024 13024

C910C9101 13009 13003 13003 13003 13003

C1C9C10C1rsquo

C6rsquo

20

C92 - - - 12869 12869

C1rsquoC11 10329 10335 10335 - 10443

C12 - - - - 10382

C3 7724 7709 7702 - 7829

- 7634 7629 - 7621

7525

C2 7388 7398 7390 7370 7370

C4 7052 7077 7054 7041 7098

C16 6900 7016 - - 7029

- - - - 6793

C1C5C6 6153 6241 6236 6239 6262

C1C5C6 6117 - - - 5956

C2 3300 3290 3290 3283 3279

CH2(trans) - - - 3198 3196

CH2(gauche)C15 2941 2981 2987 3004 -

C811 2706 2747 2747 2832 2817

CH2 2597 2625 2628 2630 2817

512

13C CPMAS spectra at 70degC (Figure 6b full range in Figure 7 and Figure S 7d) at 513

increasing contact times confirm the presence of two cis-olefin environments one for which 514

C910 are equivalent labelled here C9101 and one for which they are inequivalent labelled 515

C92 C102 Both environments are rigid enough to be observed at short contact time (tc= 50 μs) 516

A similar double environment is also observed for the sugar C1rsquo CPMAS at 70degC shows two 517

peaks (C1rsquo1 at δC= 10443 ppm C1rsquo2 at δC= 10382 ppm Figure 6b Figure S 7d) These are 518

significantly shifted downfield from the C1rsquo peak observed at 110degC and are detected at contact 519

times as short as 50 μs thus indicating their rigid character One also finds that CPMAS is 520

totally inefficient at 100degC for short contact times (50 μs) and poorly efficient even at 3 ms 521

typical of fast-tumbling molecules (Figure S 7c) also both gauche (δC= 3004 ppm) and all-522

trans (δC= 3198 ppm) conformations of the CH2 carbons coexist at 70degC (Figure S 7b)77 523

According to ssNMR one main mobile molecular environment exists at 110degC as 524

shown by the well-resolved 1H spectrum single C=C (C9101) and anomeric (C1rsquo) resonances 525

gauche conformation of the aliphatic CH2 and single T2 of the aliphatic CH2 chain and sugar 526

protons The sugar group is certainly mobile (inefficient 1H13C CPMAS) but still more 527

constrainted than the acyl chain (shorter T2 (1H)) As the temperature is lowered the overall 528

rigidity increases (broadening of 1H NMR peaks faster 1H transverse relaxation and efficient 529

1H13C magnetization transfer during CPMAS) presumably driven by H-bonding 530

interactions (downfield shift and line narrowing of the OH 1H signal very short OH T2 loss of 531

the C1rsquo signal in SPMAS) At the same time a number of spectroscopic data (two concomitant 532

21

resonances for anomeric C1rsquo1 and C1rsquo2 cis-olefin C9101 and C92 C102 and CH2 gauche and 533

all-trans environments) and relaxation behaviour (fast and slow T2 relaxation components) 534

strongly suggests the coexistence of two molecular environments Further proofs of the 535

coexistence of two distinct molecular domains are given by a number of chemical shift 536

variations in the 13C CPMAS spectrum at 70degC compared to the one at 110degC 90degC and 537

solution This is shown in Figure 7 and in the list of 13C chemical shifts in Table 1 538

539

540

Figure 7 - Comparison between solution- and solid-state 13C NMR spectra of G-C181-OH Solid-state 541

experiments are recorded directly on the sample powder at a MAS of 10 kHz (4 mm zirconia rotor) SP 542

Single Pulse CP Cross Polarization In all the ssNMR experiments temperature is first raised at 110degC 543

and then lowered to either 90degC or 70degC Segmented lines are meant to highlight the known peak position 544

in solution thus stressing the strong chemical shift variations in the solid-state 545

546

NMR experiments above Tm2 suggest that vesicles could contain G-C181-OH with a 547

single family of mobile conformations characterized by chemically-equivalent C=C (red 548

Figure 8b) and a single C1rsquo conformation (green Figure 8b) Experiments between Tm1 and Tm2 549

otherwise suggest that two coexisting conformations characterized by chemically-equivalent 550

(red Figure 8b) and inequivalent (magenta Figure 8b) C=C could be associated to nanotubes 551

Two distinct C1rsquo sites (grey Figure 8b) also indicate two configurations of the sugar headgroup 552

A comment on this point will be given in the following paragraphs 553

The local (Aring-scale) variation of molecular mobility below Tm2 drives the coexistence of 554

two distinct molecular environments of the same G-C181-OH molecule This occurs within a 555

single-phase domain (as deduced by WAXS) of equivalent membrane rigidity (as derived from 556

Laurdan probe in confocal microscopy) Although never reported before the difference in the 557

local mobility of the same amphiphile (here G-C181-OH) could satisfy the theoretically-558

140 120 100 80 60 40 20

CP(3 ms)-70degC

SP-70degC

SP-90degC

SP-110degC

ppm

LiquidC910

C1rsquo

C2rsquo

C4rsquo

C3rsquo

C1

C6rsquoC5rsquo

C1

6

C8

11

C2 C

H2

C1

5

CH

2

DM

SO

13C

22

required conditions that induce local lipid segregation and formation of membrane inclusions 559

(Figure 8c) In case of their non-zero average orientation the local membrane curvature 560

increases62 and nanotubes can spontaneously form (Figure 8bc) More intriguing this 561

mechanism occurs at least 10degC above Tm1 although it should not be favored above the Tm27 562

From both experimental and theoretical backgrounds the conditions of topologically-563

connected vesicles-nanotubes systems are met only when either internal (two components) or 564

external driving forces are applied19 In the present single-lipid system of which the phase 565

behavior is summarized in Figure 8a the most plausible driving force could be explained by 566

the NMR results collected on the dry sample G-C181-OH concomitantly occupies two 567

molecular environments characterized by 1) two distinct conformations of the sugar 568

displaying equivalent mobility and 2) two conformations of the acyl chain unsaturation 569

570

23

571

Figure 8 ndash a) Thermotropic phase behavior of G-C181-OH at 5 mgmL-1 Note that Tm2 is observed on the 572

dry powder only b) Possible relationship between the molecular environment as derived from NMR and 573

membrane curvatures All the molecular drawings refer to the same G-C181-OH molecule (bulky ellipsoid 574

equiv sugar headgroup) while each color corresponds to a given configuration Grey of C1rsquo1 and C1rsquo2 indicates 575

the uncertainty about the assignment of the specific NMR peak The curvature C is defined as 1R R being 576

the curvature radius R1 R2 radius of curvature CG Gauss curvature CM Mean curvature c) Free 577

interpretation of the anisotropic high-curvature (CMgt 0 CG= 0) inclusions in the membrane (low CM and 578

CG) continuum driven by inter-sugar H-bonding and reduced intramembrane diffusion for a G-C181-OH 579

glucolipid 580

581

Flat crystalline structure

Nanotube-vesicle network

Vesicles

48degC

120deg

106 105 104 103 102 101

ppm

H9 H10H1rsquo

106 105 104 103 102 101

ppm106 105 104 103 102 101

ppm106 105 104 103 102 101

ppm

C=CC=C

C1rsquo1C1rsquo2

C1rsquo

Tm2(96degC)Tm1

a)

b)

c)

Same molecule different local orientations and sugar conformations

High curvature membrane

Low curvature inclusion

Flip-FlopStrong H-bonding

CG

CM ( + )

CG 0

CM 0 CG 0

CM 0

G-C181-OH

24

Two related questions are unanswered at the moment What actually drives nanotube 582

formation and is the organization of G-C181-OH within the vesicles and tubes the same Lipid 583

nanotubes have been studied for at least three decades over which broad experimental and 584

theoretical insight has been generated Despite this fact the actual understanding is still matter 585

of debate27 Micelle-to-nanotube and vesicle-to-nanotube phase transitions are generally 586

reported for single amphiphiles below a given Tm and intermediate phase composed of twisted 587

and flat fibers is concomitantly reported Initially thought to be a chirality-driven phenomenon 588

recent studies propose a more general mechanism involving crystallization and symmetry 589

breaking in the lipid packing driven by optimal packing27 If twisted ribbons were observed 590

before for a number of microbial glycolipid bolaamphiphiles44 they are not found for G-C181-591

OH This fact as well as the large body of cryo-TEM images and the observation of nanotubes 592

above Tm1 rather pledge in favour of a direct topological change from positive (vesicles) to 593

zero (nanotubes) Gaussian curvatures In this case vesicle-to-nanotube transitions could be 594

driven by external forces56307813ndash162829 595

Considering the absence of external forces in the present system there are only two 596

possibilities left spontaneous budding of the membrane or nanotube pulling after inter-vesicle 597

collisions due to Brownian motion Both are possible as inclusions composed of a different 598

arrangement of the same molecule in the membrane continuum can theoretically explain 599

spontaneous nanotube formation in an isotropic medium672223629ndash1218ndash21 However the second 600

one is probably more likely to prevail because it could explain the large number of nanotubes 601

connecting two distant vesicles Interestingly similar events occur for any phospholipid-based 602

vesicular solution but nanotubes are extremely rare unless at least two lipids of different 603

molecular structure are mixed together The spectroscopic 1H and 13C and relaxation (1H) 604

ssNMR data even if collected on a dry powder demonstrate that the coexistence of two G-605

C181-OH configurations involving both the sugar headgroup and fatty alcohol chain are 606

possible for this molecule As a matter of fact this is not so surprising as more than one energy 607

minima each corresponding to a given conformation are known for glucose its non-reducing 608

derivative methylglucose and carbohydrates in general78ndash80 609

The presence of two configurations could explain the coexistence between two 610

structures of different curvature but the data presented so far do not explain the origin of these 611

two configurations In other words what drives the change in the local structure of the sugar 612

headgroup and fatty alcohol within the membrane 613

The association of known self-assembled structures to spectroscopic 13C ssNMR data is 614

known for years in particular for gluconamides lipids having a linear sugar headgroup81ndash84 It 615

25

was found that gauche effects in some cases up to the γ position in the sugar chain induce 616

upfield chemical shifts while anti and trans conformations result in downfield shifts Each 617

structure (monolayer crystal bilayer crystal or micellar fiber) was then characterized by a given 618

organization of the gluconamide and a characteristic 13C ssNMR fingerprint which could be 619

explained by gauche γ-gauche anti and trans conformations of the headgroup Similar 620

structural-spectroscopic correlations have been collected over the years for much simpler cyclic 621

sugars85 whereas pyranoses can undergo important conformational changes (chair half-chair 622

boat skew) when the appropriate energy barriers are crossed78ndash8086 A quantitative 623

conformational study based on the 13C chemical shift variations shown in Table 1 would be 624

certainly helpful but very delicate in the context of the present state of the art because a 625

rigorous approach requires at least one known crystal structure8182 which is currently missing 626

for this compound A qualitative approach will then be adopted 627

G-C181-OH is a bolaform amphiphile and flip-flop could be a possible mechanism 628

involved in the nanotube stabilization whereas head-tail interactions have long been considered 629

as important in nanotube-forming bolaform amphiphiles6063 From mere considerations based 630

on T2 (1H) relaxation arguments one could exclude head-tail configurations if the R-CH2OH 631

endgroup (H1) was involved in a direct interaction with the sugar group its corresponding T2 632

would probably be comparable to the sugarrsquos one (OH and H1rsquo) but this does not seem to be 633

the case However this hypothesis would be based on the comparison of the relaxation from 634

CH and CH2 protons additionally involved in different kind of motions Besides the relaxation 635

of H1 is influenced by the strong overlap between H1 and H5rsquo For these reasons relaxation 636

arguments are not solid enough and will not be used 637

We then address this question by exploring intermolecular interactions in a direct 638

manner using 2D 1H-1H homonuclear ssNMR correlation spectroscopy At 90degC in the fast-639

tumbling domain of molecular motion (narrow 1H NMR peaks Figure 6a) and corresponding 640

to the vesicle-rich region spin diffusion (NOESY) experiments are preferred over dipolar 641

coupling based experiments because such interaction is averaged out by molecular motion 642

However when the molecular motion is more constrained in the domain of nanotube stability 643

at 70degC (broad 1H spectra Figure 6a) Double-Quantum Single-Quantum (DQ-SQ) correlation 644

experiments are best suited to probe interaction between close molecular neighbors These 645

experiments are reported in Figure 9 and the related interpretation is only briefly reported 646

below while a more detailed discussion can be found on Page 12 in the Supporting Information 647

26

648

Figure 9 ndash a) 2D 1H-1H NOESY NMR experiment performed at MAS= 10 kHz (4 mm probe) on a Bruker 649

300 MHz Avance III HD spectrometer using 300 ms of mixing time Sample temperature is regulated to 650

90degC reached by cooling from 110degC b) 2D 1H-1H DQ-SQ (BAck-to-BAck BABA pulse sequence used for 651

the DQ excitation scheme) NMR experiment performed at MAS= 22 kHz (25 mm probe) on a Bruker 700 652

MHz Avance III spectrometer using a DQ excitation time equal to 2 rotor periods Sample temperature is 653

regulated to 70degC reached by cooling from 110degC The typical schemes illustrating how to interpret these 654

2D experiments are reported above each experimental map Zooms of the sugar region and configurational 655

interpretation are given below each 2D map More information are given in the text below 656

H16-H16

H1-H1

H9-H10

H1rsquo-H1

H1rsquoH910 H16

H1

H1rsquo-H1rsquo

Hxrsquo-Hxrsquo

2

(pp

m)

(ppm)

315

811910 1rsquo

2rsquo2

3rsquo 4rsquo6rsquo

5rsquo

1rsquo2910

asymp asymp asympHDO MeOD Acetone

1

315

811 9101rsquo

2rsquo2

3rsquo4rsquo6rsquo

5rsquo

1rsquo2 9 10

asympasympasympHDOMeODAcetone

H1rsquo-H1

1rsquo

1

(p

pm

)

(ppm)

H1-H2

H1rsquo-H1

H1rsquo

H910

H1H5rsquo

H2

H1rsquo-H5rsquo

315

811910 1rsquo

2rsquo2

3rsquo 4rsquo6rsquo

5rsquo

1rsquo2910

asymp asymp asympHDO MeOD Acetone

1

H2rsquoH3rsquoH4rsquo

H2rsquoH3rsquoH4rsquo

X= 2rsquo3rsquo4rsquo

NOESY - T= 90degC

1H

1H

1H

1H

DQ-SQ - T= 70degC

a) b)

2

(pp

m)

(p

pm

)

(ppm)

A B

AhellipB

AhellipA

BhellipB

NOESY

(ppm)

A B

AhellipB

DQ-SQ

Single-quantum dimension

Do

ub

le-q

ua

ntu

m d

imen

sio

n

AhellipA

BhellipB

BhellipA

BhellipA

315

811910 1rsquo

2rsquo2

3rsquo 4rsquo6rsquo

5rsquo

1rsquo2910

asymp asymp asympHDO MeOD Acetone

1

lt 5 Aring

lt 5 Aring

H1-H2

H811

H15

H16

H6rsquo

H1rsquo

H1

H5rsquo

H2

H1-H2

H1rsquo-H1

H1rsquo-H5rsquo

H1rsquo

H1 H5rsquo

H1rsquo

H1

H5rsquo

H1H2rsquo H3rsquo H4rsquo

H1-H1

H1rsquo-H1

Hxrsquo-HxrsquoX= 2rsquo3rsquo4rsquo

H1-H2

H1rsquo

Zoom sugar region 28 ndash 45 ppmNOESY DQ-SQ

27

Head-tail or head-head configurations could be probed by the existence of H1rsquo-H1 657

interactions whereas H1 (δH= 356 ppm) is associated to the glucolipid tail and H1rsquo (δH= 432 658

ppm) to the sugar headgroup These atoms are more than 15 Aring apart and a direct interaction 659

could not be observed in both NOESY and DQ-SQ experiments NOESY experiments (Figure 660

9a) show that H1rsquo-H1 correlations are not observed Meanwhile clear-cut direct H1rsquo-H1 661

interactions are indeed observed are detected at T= 70degC (DQ cross-peaks related to H1rsquo-H1 662

correlation in DQ-SQ experiment shown in Figure 9b) Interestingly the absence of auto peaks 663

at δH= 432 ppm and δH= 335 ppm in the SQ dimension excludes H1rsquo-H1rsquo and Hxrsquo-Hxrsquo (x= 2 664

3 4) interactions respectively strengthening the idea that sugar headgroups from two nearest-665

neighbor G-C181-OH molecules are not adjacent in the nanotube-vesicle region 666

Homonuclear 2D ssNMR supports the idea according to which G-C181-OH can 667

undergo molecular flip-flops and a head-tail arrangement in the dry product Extrapolating these 668

conclusions in the lower concentration range lack of H1rsquo-H1 proximity at T= 90degC suggests a 669

head-head configuration in the vesicle-rich region and head-tail configuration within the 670

nanotubes It is however still unclear whether flip-flops are driven by spontaneous temperature-671

induced conformational changes in the glucolipid packing stimulating inclusions within the 672

membrane itself672223629ndash1218ndash21 or by defects generated by inter-vesicle collisions In fact a 673

combination of both is not unlikely Nanotubes are not stable when temperature is 20 to 30 674

degrees above Tm1 indicating that inter-vesicle collisions alone cannot explain the unusual 675

phenomenon In an energy landscape approach87 the head-head configuration is certainly 676

favored at high temperature while both head-head and head-tail configurations could be 677

favored in the mid-temperature region corresponding to the range between Tm1 and Tm2 in the 678

dry powder Both statistical and defect-induced flip-flop events contribute to form high mean 679

zero Gaussian curvature morphologies (nanotubes) for which the head-tail conformation 680

limiting the proximity between bulky sugar headgroups is probably more adapted (Figure 8c) 681

The flip-flop interpretation now reasonably explains the strong differences in the sugar and fatty 682

alcohol conformations deduced by the multiple NMR signals attributed to chemically-683

equivalent groups The existence of various low-energy conformations each having specific 684

NMR signature for sugars in solution78ndash8086 and solid-state81ndash84 further supports our 685

interpretation 686

One last point should be clarified Why between Tm1 and Tm2 are two molecular 687

environments able to coexist instead of a single one of average rigidity as found in the case of 688

ldquoripplerdquo phases50ndash54 We believe that this question can be answered based on the NMR results 689

the relatively short T2 values of the H1rsquo and OH protons shows the constrained mobility of the 690

28

sugar headgroup despite a certain flexibility of the chain (long T2 component) This trend may 691

suggest restrictions in translational mobility probably due to the sugar-OH hydrogen bonding 692

(Figure 8c) meaning that lipids are not able to diffuse within the membrane thus keeping two 693

different configurations stable 694

695

Conclusions 696

Topologically-connected nanotube-vesicle lipidic systems are commonly observed for 697

self-assembled membranes composed of at least two amphiphiles and spontaneously driven by 698

anisotropic inclusions andor by external directional forces This work shows that strong 699

differences in membrane curvature can spontaneously occur in single-lipid systems without 700

external anisotropic forces Coexistence and interconnection between nanotubes and vesicles 701

are evidenced via cryo-TEM and fluorescence confocal microscopy in a single-phase region at 702

temperature above a transition temperature Tm1= 483degC determined by DSC on the hydrated 703

sample but below the second transition temperature Tm2 (957degC) determined by DSC on the 704

dry powder X-ray diffraction (WAXS) also shows the coexistence of two lattice periods one 705

attributed to an inter-membrane repeating distance and the other possibly related to the intra-706

nanotube glucolipid arrangement Labelling the systems with Laurdan a fluorophore of which 707

the emission strongly varies with the local membrane rigidity excludes the presence of rigid 708

micro-separated phases (eg Lβ) 709

1H and 13C ssNMR despite the fact that it can only be performed on the dry powder 710

confirm the presence of one main fluid G-C181-OH environment above Tm2 Below Tm2 the 711

splitting of some 13C peaks and the coexistence of short and long 1H relaxation components 712

demonstrate that G-C181-OH can simultaneously pack into structurally and dynamically 713

inequivalent environments Combining advanced 2D 1H-1H correlation experiments based on 714

spin diffusion or double-quanta excitation suggests that vesicles could contain a preferred head-715

head configuration of the bolaform lipid while nanotubes a preferred head-tail assembly 716

The presence of two environments composed of the same molecule characterized by 717

different molecular configurations and rigidity fulfills the theoretical conditions necessary to 718

spontaneously nucleate a membrane inclusion of non-zero average orientation described as 719

necessary to observe topologically-connected nanotube-vesicle structures in the absence of 720

external forces This hypothesis could also be enriched by the possibility of nanotubes to be 721

pulled away between two vesicles after a collision event 722

This work expands the actual theoretical and experimental framework on lipid networks 723

stabilized by a different radius of curvature It could also motivate the synthesis or isolation 724

29

of new lipids providing a higher level of 3D architectural control in soft networks possibly in 725

the range of room temperature 726

727

Acknowledgements 728

Dr Andrea Lassenberger and Dr Sylvain Preacutevost at Institut Laue Langevin (ILL Grenoble 729

France) are kindly acknowledged for their assistance on the SANS experiment Confocal 730

images were performed at the Institut de Biologie Paris Seine (IBPS) imaging facility Jean-731

Franccedilois Gilles (Imaging Core Facility Institut de Biologie Paris Seine (IBPS) CNRS 732

Sorbonne Universiteacute Paris France) is kindly acknowledged for his assistance on the confocal 733

microscopy imaging Dr Guillaume Laurent (LCMCP Sorbonne Universiteacute Paris France) is 734

kindly acknowledged for helpful discussions The French region Ile-de-France SESAME 735

program is acknowledged for financial support (700 MHz NMR 736

spectrometer) Dr Olivier Diat (ICSM Marcoule France) Prof eronikaIglič and Prof V Aleš737

onsul discussi Slovenia) are kindly acknowledged for helpfUniversity of LjubljanaIglič (-Kralj 738

739

Financial support 740

WAXS experiments have been supported by Soleil Light Source Saint Aubin France proposal 741

Ndeg 20201747 SANS experiments have been supported by ILL proposal Ndeg9-13-778 742

743

References 744

1 D Kabaso N Bobrovska W Goacuteźdź E Gongadze V Kralj-Iglič R Zorec and A 745

Iglič Bioelectrochemistry 2012 87 204ndash210 746

2 C R Safinya U Raviv D J Needleman A Zidovska M C Choi M A Ojeda-747

Lopez K K Ewert Y Li H P Miller J Quispe B Carragher C S Potter M W 748

Kim S C Feinstein and L Wilson Adv Mater 2011 23 2260ndash2270 749

3 P Veranič M Lokar G J Schuumltz J Weghuber S Wieser H Haumlgerstrand V Kralj-750

Iglič and A Iglič Biophys J 2008 95 4416ndash4425 751

4 A Rustom R Saffrich I Markovic P Walther and H-H Gerdes Science (80- ) 752

2004 303 1007ndash1010 753

5 M I Angelova S Soleacuteau P Meacuteleacuteard F Faucon and P Bothorel Prog Colloid 754

Polym Sci 1992 89 127ndash131 755

6 L Mathivet S Cribier and P F Devaux Biophys J 1996 70 1112ndash1121 756

7 V Kralj-Iglič G Gomišček J Majhenc V Arrigler and S Svetina Colloids Surfaces 757

A Physicochem Eng Asp 2001 181 315ndash318 758

30

8 V Kralj-Iglič A Iglič G Gomišček F Sevšek V Arrigler and H Haumlgerstrand J 759

Phys A Math Gen 2002 35 1533ndash1549 760

9 M Drab D Stopar V Kralj-Iglič and A Iglič Cells 2019 8 626 761

10 A Zidovska K K Ewert J Quispe B Carragher C S Potter and C R Safinya 762

Langmuir 2009 25 2979ndash2985 763

11 A Zidovska K K Ewert J Quispe B Carragher C S Potter and C R Safinya Soft 764

Matter 2011 7 8363ndash8369 765

12 A Zidovska K K Ewert J Quispe B Carragher C S Potter and C R Safinya 766

Biochim Biophys Acta - Biomembr 2009 1788 1869ndash1876 767

13 M Karlsson K Sott M Davidson A S Cans P Linderholm D Chiu and O Orwar 768

Proc Natl Acad Sci U S A 2002 99 11573ndash11578 769

14 A Karlsson R Karlsson M Karlsson A-S Cans A Stromberg F Ryttsen and O 770

Orwar Nature 2001 409 150ndash152 771

15 M Karlsson K Sott A S Cans A Karlsson R Karlsson and O Orwar Langmuir 772

2001 17 6754ndash6758 773

16 E Evans H Bowman A Leung D Needham and D Tirrell Science (80- ) 1996 774

273 933ndash935 775

17 U Seifert Configurations of fluid membranes and vesicles 1997 vol 46 776

18 J Kaumls and E Sackmann Biophys J 1991 60 825ndash844 777

19 N Bobrovska W Goacuteźdź V Kralj-Iglič and A Iglič PLoS One 2013 8 e73941 778

20 A Iglič B Babnik K Bohinc M Fošnarič H Haumlgerstrand and V Kralj-Iglič J 779

Biomech 2007 40 579ndash585 780

21 K Bohinc T Slivnik A Iglič M Brumen and V Kralj-Iglič J Phys Chem C 2007 781

111 9709ndash9718 782

22 A Iglič H Haumlgerstrand P Veranič A Plemenitaš and V Kralj-Iglič J Theor Biol 783

2006 240 368ndash373 784

23 V Kralj-Iglič and P Veranič Adv Planar Lipid Bilayers Liposomes 2006 5 129ndash785

149 786

24 A Iglic and V Kralj-Iglic Adv Planar Lipid Bilayers Liposomes 2006 4 253ndash279 787

25 S Penič L Mesarec M Fošnarič L Mroacutewczyńska H Haumlgerstrand V Kralj-Iglič and 788

A Iglič Front Phys 2020 8 342 789

26 P Yager and P E Schoen Mol Cryst Liq Cryst 1984 106 371ndash381 790

27 T G Barclay K Constantopoulos and J Matisons Chem Rev 2014 114 10217ndash791

10291 792

31

28 B Božič G Gomišček V Kralj-Iglič S Svetina and B Žekš Eur Biophys J 2002 793

31 487ndash496 794

29 A Iglič H Haumlgerstrand M Bobrowska-Haumlgerstrand V Arrigler and V Kralj-Iglič 795

Phys Lett A 2003 310 493ndash497 796

30 R Bar-Ziv and E Moses Phys Rev Lett 1994 73 1392ndash1395 797

31 S Chiruvolu H E Warriner E Naranjo S H Idziak J O Raumldler R J Plano J A 798

Zasadzinski and C R Safinya Science (80- ) 1994 266 1222ndash1225 799

32 V Kralj-Iglič A Iglič H Haumlgerstrand and P Peterlin Phys Rev E 2000 61 4230ndash800

4234 801

33 L Mesarec M Drab S Penič V Kralj-Iglič and A Iglič Int J Mol Sci 2021 22 802

2348 803

34 L Van Renterghem S L K W Roelants N Baccile K Uyttersprot M C Taelman 804

B Everaert S Mincke S Ledegen S Debrouwer K Scholtens C V Stevens W 805

Soetaert L Van Renterghem S L K W Roelants N Baccile K Uytersprot M C 806

Taelman B Everaert S Mincke S Ledegen S Debrouwer K Scholtens C V 807

Stevens and W Soetaert Biotechnol Bioeng 2018 115 1195ndash1206 808

35 J Clark and F Deswarte Introduction to Chemicals from Biomass Second Edition 809

John Wiley amp Sons 2015 810

36 K V Sajna R Houmlfer R K Sukumaran L D Gottumukkala and A Pandey in 811

Industrial Biorefineries and White Biotechnology eds A Pandey R Houmlfer M 812

Taherzadeh K M Nampoothiri and C Larroche Amsterdam Oxford Waltham 813

Elsevier 2015 pp 499ndash521 814

37 R Marchant and I M Banat Trends Biotechnol 2012 30 558ndash565 815

38 K Hill and C LeHen-Ferrenbach in Sugar-Based Surfactants Fundamentals and 816

Applications ed C C Ruiz Boca Raton CRC Press 2009 pp 1ndash20 817

39 L Rodrigues I M Banat J Teixeira and R Oliveira J Antimicrob Chemother 818

2006 57 609ndash618 819

40 J D Desai and I M Banat Microbiol Mol Biol Rev 1997 61 47ndash64 820

41 S L K W Roelants K Ciesielska S L De Maeseneire H Moens B Everaert S 821

Verweire Q Denon B Vanlerberghe I N A Van Bogaert P Van der Meeren B 822

Devreese and W Soetaert Biotechnol Bioeng 2016 113 550ndash559 823

42 N Baccile M Selmane P Le Griel S Preacutevost J Perez C V Stevens E Delbeke S 824

Zibek M Guenther W Soetaert I N A Van Bogaert and S Roelants Langmuir 825

2016 32 6343ndash6359 826

32

43 N Baccile A-S Cuvier S Preacutevost C V Stevens E Delbeke J Berton W Soetaert 827

I N A Van Bogaert and S Roelants Langmuir 2016 32 10881ndash10894 828

44 N Baccile C Seyrig A Poirier S A Castro S L K W Roelants and S Abel Green 829

Chem 2021 23 3842ndash3944 830

45 A Salis L Cappai C Carucci D F Parsons and M Monduzzi J Phys Chem Lett 831

2020 11 6805ndash6811 832

46 J Schindelin I Arganda-Carreras E Frise V Kaynig M Longair T Pietzsch S 833

Preibisch C Rueden S Saalfeld B Schmid J-Y Tinevez D J White V 834

Hartenstein K Eliceiri P Tomancak and A Cardona Nat Methods 2012 9 676ndash835

682 836

47 T Parasassi and E Gratton J Fluoresc 1995 5 59ndash69 837

48 M M Dodes Traian F L Gonzaacutelez Flecha and V Levi J Lipid Res 2012 53 609ndash838

616 839

49 D M Owen C Rentero A Magenau A Abu-Siniyeh and K Gaus Nat Protoc 840

2012 7 24ndash35 841

50 W Stillwell in An Introduction to Biological Membranes Academic Press 2016 pp 842

135ndash180 843

51 W Stillwell in An Introduction to Biological Membranes Academic Press 2016 pp 844

181ndash220 845

52 B Tenchov Chem Phys Lipids 1991 57 165ndash177 846

53 R Koynova and M Caffrey Biochim Biophys Acta 1998 1376 91ndash145 847

54 F Schmid S Dolezel O Lenz and S Meinhardt J Phys Conf Ser 2014 487 848

012004 849

55 S Meinhardt R L C Vink and F Schmid Proc Natl Acad Sci U S A 2013 110 850

4476ndash4481 851

56 O Glatter and O Kratky Small Angle X-ray Scattering Academic Press London 852

1982 853

57 B A Cunningham A D Brown D H Wolfe W P Williams and A Brain Phys 854

Rev E 1998 58 3662ndash3672 855

58 M Nadler A Steiner T Dvir O Szekely P Szekely A Ginsburg R Asor R Resh 856

C Tamburu M Peres and U Raviv Soft Matter 2011 7 1512ndash1523 857

59 C Tanford The hydrophobic effect Formation of micelles and biological membranes 858

John Wiley amp Sons Inc New York 1973 859

60 M Masuda and T Shimizu Langmuir 2004 20 5969ndash5977 860

33

61 A Zidovska K K Ewert J Quispe B Carragher C S Potter and C R Safinya 861

Block Liposomes Vesicles of Charged Lipids with Distinctly Shaped Nanoscale 862

Sphere- Pear- Tube- or Rod-Segments Elsevier Inc 1st edn 2009 vol 465 863

62 A Iglic V Kralj-Iglic B Božič M Bobrowska-Haumlgerstrand B Isomaa and H 864

Haumlgerstrand Bioelectrochemistry 2000 52 203ndash211 865

63 T Shimizu M Masuda and H Minamikawa Chem Rev 2005 105 1401ndash1444 866

64 J V Selinger and J M Schnur Phys Rev Lett 1993 71 4091ndash4094 867

65 J M Schnur Science (80- ) 1993 262 1669ndash1676 868

66 B Tenchov R Koynova and G Rapp Biophys J 2001 80 1873ndash1890 869

67 R Koynova and B Tenchov Ser Biomech 2017 31 3ndash19 870

68 S J Angyal and R Le Fur Carbohydr Res 1984 126 15ndash26 871

69 J K Whitesell and M A Minton J Am Chem Soc 1987 109 225ndash228 872

70 T Takayama S Ando and I Ando J Mol Struct 1990 220 245ndash250 873

71 K Gawrisch N V Eldho and I V Polozov Chem Phys Lipids 2002 116 135ndash151 874

72 F Tielens C Gervais J F Lambert F Mauri and D Costa Chem Mater 2008 20 875

3336ndash3344 876

73 G A Jeffrey and Y Yeon Acta Crystallogr Sect B 1986 42 410ndash413 877

74 B Berglund and R W Vaughan J Chem Phys 1980 73 2037ndash2043 878

75 C Gervais C Coelho T Azaiumls J Maquet G Laurent F Pourpoint C Bonhomme 879

P Florian B Alonso G Guerrero P H Mutin and F Mauri J Magn Reson 2007 880

187 131ndash140 881

76 C Akita T Kawaguchi F Kaneko H Yamamoto and M Suzuki J Phys Chem B 882

2004 108 4862ndash4868 883

77 R N Purusottam L Seacutenicourt J J Lacapegravere and P Tekely Biochim Biophys Acta - 884

Biomembr 2015 1848 3134ndash3139 885

78 X Biarneacutes A Ardegravevol A Planas C Rovira A Laio and M Parrinello J Am Chem 886

Soc 2007 129 10686ndash10693 887

79 L M J Kroon‐Batenburg and J Kroon Biopolymers 1990 29 1243ndash1248 888

80 K N Kirschner and R J Woods Proc Natl Acad Sci U S A 2001 98 10541ndash889

10545 890

81 S Svenson B Kirste and J H Fuhrhop J Am Chem Soc 1994 116 11969ndash11975 891

82 I Sack S Macholl J H Fuhrhop and G Buntkowsky Phys Chem Chem Phys 892

2000 2 1781ndash1788 893

83 S Svenson J Koening and J H Fuhrhop J Phys Chem 1994 98 1022ndash1028 894

34

84 S Svenson A Schaefer and J H Fuhrhop J Chem Soc Perkin Trans 2 1994 2 895

1023ndash1028 896

85 K Bock and C Pedersen Adv Carbohydr Chem Biochem 1983 41 27ndash66 897

86 T B Grindley Structure and Conformation of Carbohydrates Springer Berlin 898

Heidelberg 2008 899

87 F Tantakitti J Boekhoven X Wang R V Kazantsev T Yu J Li E Zhuang R 900

Zandi J H Ortony C J Newcomb L C Palmer G S Shekhawat M O De La 901

Cruz G C Schatz and S I Stupp Nat Mater 2016 15 469ndash476 902

903

1

Supporting Information 904

905

Topological connection between vesicles and nanotubes in single-906

component lipid membranes driven by head-tail interactions 907

908

Niki Baccilea Ceacutedric Lorthioira Abdoul Aziz Baa Patrick Le Griela Cristina Coelhob Javier 909

Perezc Wim Soetaertd Sophie L K W Roelantsd 910

911

a Sorbonne Universiteacute Centre National de la Recherche Scientifique Laboratoire de Chimie de 912

la Matiegravere Condenseacutee de Paris LCMCP F-75005 Paris France 913

b Institut des Mateacuteriaux de Paris Centre Sorbonne Universiteacute Paris France 914

c Synchrotron Soleil LrsquoOrme des Merisiers Saint-Aubin BP48 91192 Gif-sur-Yvette Cedex 915

France 916

d InBio Department of Biotechnology Ghent University Ghent Belgium 917

918

919

Corresponding author 920

Dr Niki Baccile 921

E-mail address nikibaccilesorbonne-universitefr 922

Phone +33 1 44 27 56 77 923

924

925

926

2

a) 927

Parameter Method

Dry Matter (DM ) 996 Infrared balance 105degC

Glucose () 012 HPLC-Metacarb

Glycerol () nd HPLC-Metacarb

Free Fatty Acid content () lt001 Internal method BBEPP

GC with FID Detector

Oil () lt001 Internal method BBEPP

GC with FID Detector

Moulds (CFUg) lt10 3M-Nordval ndeg16 B

Yeasts (CFUg) lt10 3M-Nordval ndeg16 B

Anaerobic count (CFUg) lt10 SP-VG M005 B

Protein () tbd BCA protein assay

928

Figure S 1 - a) HPLC-ELSD chromatogram and composition table 929

930

931

932

933

3

b) 934

935

Group Ndeg δ ppm

CH2 1 354

CH2 2 153

CH2 3-712-14 132

CH2 811 203

CH 910 534

CH2 15 162

CH2 16 388

CH 1 424

CH 2 316

CH 3 334

CH 4 329

CH 5 354

CH2 6 366

Figure S 1 (next) ndash b) 1H NMR spectrum (MeOD-d4) recorded on the purified G-C181-OH compounds 936

and related assignment 937

938

2

811910 1rsquo

5rsquo1

3rsquo4rsquo

1rsquo1910

asymp asymp asympHDO MeOD Acetone

6rsquo2rsquo

1615

3-712-14

4

939

940

Figure S 2 ndash Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 130degC 941

5

942

943

Figure S 3 - Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 90degC 944

945

90degC

1

22

6

946

947

948

949

Figure S 4 ndash Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 70degC 950

951

21

11

1

2

5

3

4

45

5

6

7

7

7

952

953

954

Figure S 5 ndash Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 25degC 955

956

957

1

1 1

8

958

Figure S 6 ndash a) SANS experiment recorded at 90degC for a 5 mgmL G-C181-OH aqueous solution b-c) 959

WAXS experiments recorded on aqueous solutions of G-C181-OH at b) 5 mgmL and c) 25 mgmL d) 960

Structural interpretation of the repeating distances pointed out in c) 961

962

963

964

001 01 1

q Aring-1

25degC

68degC

88degC

111degC

01 1q Aring

-1

02 04 06 08 1q Aring

-1

001 01

10-1

100

101

Inte

nsity

cm

-1

q Aring-1

-2

T= 90degC

SANS WAXS WAXS

a) b) c)

dv

dL d= 144 Aring-1

di

ddH2O

dt

dvdL

dtdi d

d)

9

965

966

Figure S 7 ndash ssNMR MAS (10 kHz) experiments recorded on a G-C181-OH dry powder a) 1H spectra 967

recorded from 110degC to 70degC b) 13C single pulse (SP) spectra recorded with high power 1H decoupling from 968

110degC to 70degC c) 13C cross polarization (CP) spectra recorded at 110degC and contact time tc of 50 micros and 969

3000 micros d) 13C CP spectra recorded at 70degC and for 20 lt tc micros lt 3000 970

971

1

811

910

CH2 CH2

1rsquo

6rsquo

8 6 4 2 0 ppm

110

90

80

70

910

1rsquo

6rsquo16

1

2rsquo3rsquo4rsquo

811

OH

45 40 35 30 25 20 15

ppm

200 180 160 140 120 100 80 60 40 20 0

ppm140 120 100 80

ppm

CH2-gauche

3004

110

90

80

70

65

60

50

25

C91013003

C9 - 12869

CH2-all-trans

3198

C1rsquo10335

a) b)

1H

T degCT degC

13C Single Pulse MAS

A

B

A

B

70degC70degC

5rsquo

215

200 180 160 140 120 100 80 60 40 20 0

ppm

13C CPMAST= 110degC

200 180 160 140 120 100 80 60 40 20 0

ppm

132 131 130 129 128 127 126

ppm

106 105 104 103 102

ppm

13C CPMAST= 70degC

50

tc micros

3000

20

tc micros

50 500 1000 3000

tc micros

C102

13024 C910113003

C92

12869

C1rsquo213082

C1rsquo110443

A B

106 105 104 103 102

ppm

3000

1000

500

50

20

AB

c)

d)

10

972

Figure S 8 ndash a) Short and d) long T2 components displayed by the 1H transverse relaxation signal c) 1H 973

transverse relaxation signals recorded on a G-C181-OH dry powder from 110degC to 70degC The area of each 974

given resonance is plotted (log-lin) as a function of the echo time (CPMG pulse sequence) for each 975

temperature and fitted using the equation given above T2s and T2l indicate the short and long T2 relaxation 976

times respectively obtained by describing the 1H transverse relaxation signal using a biexponential decay 977

For H1rsquo H5rsquo and H6rsquo the 1H relaxation function could be described by a monoexponential decay 978

- Protons from the sugar group (H1rsquo H5rsquo H6rsquo) show a single T2 relaxation component (Figure 979

6c and Figure S 8) while protons of the acyl chain (CH2 H910 H811) show both short (Figure 980

6c) and long (Figure 6d) T2 relaxation component 981

- The short T2 relaxation time systematically decreases from 110degC to 70degC for all species by a 982

factor ranging from two to seven depending on the specific chemical group Very long T2 are 983

observed at 110degC for all acyl species while nearly unchanged values are measured below 984

90degC without any clear trend 985

- The sugar protons (H1rsquo H5rsquo H6rsquo) have the shortest set of T2 values (lt 1 ms) below 100degC 986

The long component of T2 generally ranges between 4 ms and 12 ms 987

0 1 2 3 4 5 6 7 8

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1 2 3 4 5 6 7 8

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

H1

H8H11

H9H10

CH2

H1rsquo

H6rsquo

CH2

H910H811 H1CH2

H1rsquo H5rsquo6rsquo

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

OH

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45 T= 110degC

T= 90degC

T= 80degC

T= 70degC

Log

In

ten

sity

Time ms

Fit

)

CH2 H910 H811 H1 H56 H1 OH

00

05

10

15

20

25

30

35

40

T2 m

s

T= 110degC

T= 90degC

T= 80degC

T= 70degC

CH2 H910 H811 H1

0

2

4

6

8

10

12

T2 m

s

T= 110degC

T= 90degC

T= 80degC

T= 70degC

Short T2(1H) Long T2(1H)

a) b)

c)

11

Table S 1 ndash 1H chemical shifts of G-C181-OH in solution and in the solid-state Solid-state spectra are 988

recorded at a MAS frequency of 10 kHz in DMSO-d6 broad peak 989

990

δH ppm

Group Ndeg Liquid

(MeOD-d4) Solid

(T= 110degC) Solid

(T= 70degC)

CH2 H1 354 357

CH2 H2 153 156

CH2 H3-712-14 132 135

CH2 H811 203 203

CH H910 534 534

CH2 H15 162 164

CH2 H16 388 383

CH H1 424 432

CH H2 316 335

CH H3 334 335

CH H4 329 335

CH H5 354 351 357

CH2 H6 366 383

OH OH1 432

45 463

OH OH2 492

OH OH3 490

OH OH4 487

OH OH6 445

991

992

H1H9H10H1rsquo

H6rsquo

OH2rsquo

OH1

12

Detailed discussion about the NOESY and DQ-SQ 2D 1H-1H ssNMR experiments presented in 993

Figure 9 994

995

2D 1H-1H ssNMR experiments are a unique tool to explore through-space proximities 996

between 1H nuclei ssNMR offers a large number of possibilities each adapted to a given 997

system The advantage of 2D maps with respect to a dynamics approach is their direct 998

readability in clear two nuclei which exchange their magnetization through-space (whichever 999

mechanism may be responsible for this exchange) will display a specific cross-peak However 1000

the way cross-peaks are presented and interpreted strictly depends on the nature of the pulse 1001

program which in turn is chosen according to the physical nature of the sample 1002

1D temperature-dependent 1H NMR experiments performed on G-C181-OH (Figure 1003

4a in the main text) indicate sharp peaks at 90degC and broad peaks at 70degC Narrow peaks 1004

indicate fast andor large-amplitude molecular tumbling averaging the strong 1H-1H dipolar 1005

interactions This excludes the use of pulse sequences like DQ-SQ which is not efficient under 1006

these conditions For this reason we employ spin-diffusion-based pulse sequences like 1007

NOESY which is more adapted for samples with high molecular mobility When mixing times 1008

are chosen to be short (lt 1 s) one generally explores through-space proximities within about 5 1009

10 Aring On the contrary when molecular motions are restricted dipolar coupling becomes 1010

significant 1H NMR peaks are broader and pulse sequences like DQ-SQ can now be used 1011

with a spatial resolution of less than 5 Aring 1012

Whenever possible DQ-SQ experiments are preferred over spin diffusion-based pulse 1013

sequences because they are more selective in particular for likewise interactions which are 1014

not discriminated in NOESY-type experiments This is illustrated by the typical interaction 1015

schemes for NOESY (top Figure 9a) and DQ-SQ (top Figure 9b) For two given A and B 1016

protons 2D NOESY maps will display two on-diagonal and two square-shaped symmetrical 1017

off-diagonal cross-peaks Off-diagonal cross-peaks illustrate the mutual A-B and B-A 1018

interaction but the on-diagonal peaks are systematic and do not prove any A-A or B-B 1019

interaction Similarly to NOESY 2D DQ-SQ maps display off-diagonal peaks (symmetric 1020

across the diagonal top Figure 9b) upon A-B and B-A interaction but in contrast they will 1021

display on-diagonal peaks only in the presence of self-interaction In the tutorial example of 1022

Figure 9b top the on-diagonal A-A peak indicates mutual A-A interaction while lack of on-1023

diagonal B-B peak evidences the absence of B-B proximities Combining NOESY and DQ-SQ 1024

experiments is capital to study the local arrangement of G-C181-OH and its evolution with 1025

temperature 1026

13

At T= 90degC the H1 and H5rsquo contributions are very close but possible to discriminate 1027

them (δ(H1)= 354 ppm δ(H5rsquo)= 351 ppm) according to 1H NMR chemical shifts in Table S 1028

1 2D 1H-1H NOESY in Figure 9a also discriminates between the two components attributed 1029

on the basis of the interactions with their mutual neighbors (green H1-H2 and H5rsquo-H1rsquo squares 1030

in Figure 9a) On this basis 2D NOESY does not show a direct H1rsquo-H1 correlation (blue red-1031

crossed) but rather an extensive network of intra-sugar interactions These features exclude 1032

suggest intra- and inter-sugar interactions and they exclude head-tail arrangements as a major 1033

molecular configuration at 90degC 1034

At 70degC the 1H NMR peaks are broader indicating a reduction of the T2 (1H) relaxation 1035

times andor residual homonuclear dipolar interactions excluding the use of NOESY In this 1036

case 1H-1H DQ-SQ correlation experiments are best suited Figure 9b presents the 1H-1H DQ-1037

SQ experiment on G-C181-OH at 70degC Peaks are broader than in NOESY and attribution may 1038

be cumbersome However the experiment is performed at 700 MHz and MAS= 22 kHz which 1039

improves spectral resolution with respect to the same experiment performed at 300 MHz and 1040

MAS= 10 kHz The auto-peak at δH= 356 ppm in the SQ dimension is attributed to H1-H1 spin 1041

pairs mainly resulting from intra CH2 interaction while the DQ cross-peaks at δH= 156-356 1042

ppm and δH= 432-356 ppm respectively origin from H2-H1 and H1rsquo-H1 interactions (Figure 1043

9b) DQ-SQ map does not show any auto-peak corresponding to H1rsquo-H1rsquo or Hxrsquo-Hxrsquo (x= 2rsquo 1044

3rsquo 4rsquo) interactions (red-crossed blue Figure 9b) thus excluding any proximity (lt 5 Aring) between 1045

close glucose neighbors On the contrary DQ-SQ maps (recorded both at 700 MHz Figure 9b 1046

or at 300 MHz not shown) systematically show DQ cross-peaks which can only be attributed 1047

to H1rsquo-H1 interactions (δH= 432-356) The presence of these H1rsquo-H1 interactions can only be 1048

explained by a head-tail arrangement of the G-C181-OH molecules as illustrated below the 1049

DQ-SQ map in Figure 9b 1050

20

C92 - - - 12869 12869

C1rsquoC11 10329 10335 10335 - 10443

C12 - - - - 10382

C3 7724 7709 7702 - 7829

- 7634 7629 - 7621

7525

C2 7388 7398 7390 7370 7370

C4 7052 7077 7054 7041 7098

C16 6900 7016 - - 7029

- - - - 6793

C1C5C6 6153 6241 6236 6239 6262

C1C5C6 6117 - - - 5956

C2 3300 3290 3290 3283 3279

CH2(trans) - - - 3198 3196

CH2(gauche)C15 2941 2981 2987 3004 -

C811 2706 2747 2747 2832 2817

CH2 2597 2625 2628 2630 2817

512

13C CPMAS spectra at 70degC (Figure 6b full range in Figure 7 and Figure S 7d) at 513

increasing contact times confirm the presence of two cis-olefin environments one for which 514

C910 are equivalent labelled here C9101 and one for which they are inequivalent labelled 515

C92 C102 Both environments are rigid enough to be observed at short contact time (tc= 50 μs) 516

A similar double environment is also observed for the sugar C1rsquo CPMAS at 70degC shows two 517

peaks (C1rsquo1 at δC= 10443 ppm C1rsquo2 at δC= 10382 ppm Figure 6b Figure S 7d) These are 518

significantly shifted downfield from the C1rsquo peak observed at 110degC and are detected at contact 519

times as short as 50 μs thus indicating their rigid character One also finds that CPMAS is 520

totally inefficient at 100degC for short contact times (50 μs) and poorly efficient even at 3 ms 521

typical of fast-tumbling molecules (Figure S 7c) also both gauche (δC= 3004 ppm) and all-522

trans (δC= 3198 ppm) conformations of the CH2 carbons coexist at 70degC (Figure S 7b)77 523

According to ssNMR one main mobile molecular environment exists at 110degC as 524

shown by the well-resolved 1H spectrum single C=C (C9101) and anomeric (C1rsquo) resonances 525

gauche conformation of the aliphatic CH2 and single T2 of the aliphatic CH2 chain and sugar 526

protons The sugar group is certainly mobile (inefficient 1H13C CPMAS) but still more 527

constrainted than the acyl chain (shorter T2 (1H)) As the temperature is lowered the overall 528

rigidity increases (broadening of 1H NMR peaks faster 1H transverse relaxation and efficient 529

1H13C magnetization transfer during CPMAS) presumably driven by H-bonding 530

interactions (downfield shift and line narrowing of the OH 1H signal very short OH T2 loss of 531

the C1rsquo signal in SPMAS) At the same time a number of spectroscopic data (two concomitant 532

21

resonances for anomeric C1rsquo1 and C1rsquo2 cis-olefin C9101 and C92 C102 and CH2 gauche and 533

all-trans environments) and relaxation behaviour (fast and slow T2 relaxation components) 534

strongly suggests the coexistence of two molecular environments Further proofs of the 535

coexistence of two distinct molecular domains are given by a number of chemical shift 536

variations in the 13C CPMAS spectrum at 70degC compared to the one at 110degC 90degC and 537

solution This is shown in Figure 7 and in the list of 13C chemical shifts in Table 1 538

539

540

Figure 7 - Comparison between solution- and solid-state 13C NMR spectra of G-C181-OH Solid-state 541

experiments are recorded directly on the sample powder at a MAS of 10 kHz (4 mm zirconia rotor) SP 542

Single Pulse CP Cross Polarization In all the ssNMR experiments temperature is first raised at 110degC 543

and then lowered to either 90degC or 70degC Segmented lines are meant to highlight the known peak position 544

in solution thus stressing the strong chemical shift variations in the solid-state 545

546

NMR experiments above Tm2 suggest that vesicles could contain G-C181-OH with a 547

single family of mobile conformations characterized by chemically-equivalent C=C (red 548

Figure 8b) and a single C1rsquo conformation (green Figure 8b) Experiments between Tm1 and Tm2 549

otherwise suggest that two coexisting conformations characterized by chemically-equivalent 550

(red Figure 8b) and inequivalent (magenta Figure 8b) C=C could be associated to nanotubes 551

Two distinct C1rsquo sites (grey Figure 8b) also indicate two configurations of the sugar headgroup 552

A comment on this point will be given in the following paragraphs 553

The local (Aring-scale) variation of molecular mobility below Tm2 drives the coexistence of 554

two distinct molecular environments of the same G-C181-OH molecule This occurs within a 555

single-phase domain (as deduced by WAXS) of equivalent membrane rigidity (as derived from 556

Laurdan probe in confocal microscopy) Although never reported before the difference in the 557

local mobility of the same amphiphile (here G-C181-OH) could satisfy the theoretically-558

140 120 100 80 60 40 20

CP(3 ms)-70degC

SP-70degC

SP-90degC

SP-110degC

ppm

LiquidC910

C1rsquo

C2rsquo

C4rsquo

C3rsquo

C1

C6rsquoC5rsquo

C1

6

C8

11

C2 C

H2

C1

5

CH

2

DM

SO

13C

22

required conditions that induce local lipid segregation and formation of membrane inclusions 559

(Figure 8c) In case of their non-zero average orientation the local membrane curvature 560

increases62 and nanotubes can spontaneously form (Figure 8bc) More intriguing this 561

mechanism occurs at least 10degC above Tm1 although it should not be favored above the Tm27 562

From both experimental and theoretical backgrounds the conditions of topologically-563

connected vesicles-nanotubes systems are met only when either internal (two components) or 564

external driving forces are applied19 In the present single-lipid system of which the phase 565

behavior is summarized in Figure 8a the most plausible driving force could be explained by 566

the NMR results collected on the dry sample G-C181-OH concomitantly occupies two 567

molecular environments characterized by 1) two distinct conformations of the sugar 568

displaying equivalent mobility and 2) two conformations of the acyl chain unsaturation 569

570

23

571

Figure 8 ndash a) Thermotropic phase behavior of G-C181-OH at 5 mgmL-1 Note that Tm2 is observed on the 572

dry powder only b) Possible relationship between the molecular environment as derived from NMR and 573

membrane curvatures All the molecular drawings refer to the same G-C181-OH molecule (bulky ellipsoid 574

equiv sugar headgroup) while each color corresponds to a given configuration Grey of C1rsquo1 and C1rsquo2 indicates 575

the uncertainty about the assignment of the specific NMR peak The curvature C is defined as 1R R being 576

the curvature radius R1 R2 radius of curvature CG Gauss curvature CM Mean curvature c) Free 577

interpretation of the anisotropic high-curvature (CMgt 0 CG= 0) inclusions in the membrane (low CM and 578

CG) continuum driven by inter-sugar H-bonding and reduced intramembrane diffusion for a G-C181-OH 579

glucolipid 580

581

Flat crystalline structure

Nanotube-vesicle network

Vesicles

48degC

120deg

106 105 104 103 102 101

ppm

H9 H10H1rsquo

106 105 104 103 102 101

ppm106 105 104 103 102 101

ppm106 105 104 103 102 101

ppm

C=CC=C

C1rsquo1C1rsquo2

C1rsquo

Tm2(96degC)Tm1

a)

b)

c)

Same molecule different local orientations and sugar conformations

High curvature membrane

Low curvature inclusion

Flip-FlopStrong H-bonding

CG

CM ( + )

CG 0

CM 0 CG 0

CM 0

G-C181-OH

24

Two related questions are unanswered at the moment What actually drives nanotube 582

formation and is the organization of G-C181-OH within the vesicles and tubes the same Lipid 583

nanotubes have been studied for at least three decades over which broad experimental and 584

theoretical insight has been generated Despite this fact the actual understanding is still matter 585

of debate27 Micelle-to-nanotube and vesicle-to-nanotube phase transitions are generally 586

reported for single amphiphiles below a given Tm and intermediate phase composed of twisted 587

and flat fibers is concomitantly reported Initially thought to be a chirality-driven phenomenon 588

recent studies propose a more general mechanism involving crystallization and symmetry 589

breaking in the lipid packing driven by optimal packing27 If twisted ribbons were observed 590

before for a number of microbial glycolipid bolaamphiphiles44 they are not found for G-C181-591

OH This fact as well as the large body of cryo-TEM images and the observation of nanotubes 592

above Tm1 rather pledge in favour of a direct topological change from positive (vesicles) to 593

zero (nanotubes) Gaussian curvatures In this case vesicle-to-nanotube transitions could be 594

driven by external forces56307813ndash162829 595

Considering the absence of external forces in the present system there are only two 596

possibilities left spontaneous budding of the membrane or nanotube pulling after inter-vesicle 597

collisions due to Brownian motion Both are possible as inclusions composed of a different 598

arrangement of the same molecule in the membrane continuum can theoretically explain 599

spontaneous nanotube formation in an isotropic medium672223629ndash1218ndash21 However the second 600

one is probably more likely to prevail because it could explain the large number of nanotubes 601

connecting two distant vesicles Interestingly similar events occur for any phospholipid-based 602

vesicular solution but nanotubes are extremely rare unless at least two lipids of different 603

molecular structure are mixed together The spectroscopic 1H and 13C and relaxation (1H) 604

ssNMR data even if collected on a dry powder demonstrate that the coexistence of two G-605

C181-OH configurations involving both the sugar headgroup and fatty alcohol chain are 606

possible for this molecule As a matter of fact this is not so surprising as more than one energy 607

minima each corresponding to a given conformation are known for glucose its non-reducing 608

derivative methylglucose and carbohydrates in general78ndash80 609

The presence of two configurations could explain the coexistence between two 610

structures of different curvature but the data presented so far do not explain the origin of these 611

two configurations In other words what drives the change in the local structure of the sugar 612

headgroup and fatty alcohol within the membrane 613

The association of known self-assembled structures to spectroscopic 13C ssNMR data is 614

known for years in particular for gluconamides lipids having a linear sugar headgroup81ndash84 It 615

25

was found that gauche effects in some cases up to the γ position in the sugar chain induce 616

upfield chemical shifts while anti and trans conformations result in downfield shifts Each 617

structure (monolayer crystal bilayer crystal or micellar fiber) was then characterized by a given 618

organization of the gluconamide and a characteristic 13C ssNMR fingerprint which could be 619

explained by gauche γ-gauche anti and trans conformations of the headgroup Similar 620

structural-spectroscopic correlations have been collected over the years for much simpler cyclic 621

sugars85 whereas pyranoses can undergo important conformational changes (chair half-chair 622

boat skew) when the appropriate energy barriers are crossed78ndash8086 A quantitative 623

conformational study based on the 13C chemical shift variations shown in Table 1 would be 624

certainly helpful but very delicate in the context of the present state of the art because a 625

rigorous approach requires at least one known crystal structure8182 which is currently missing 626

for this compound A qualitative approach will then be adopted 627

G-C181-OH is a bolaform amphiphile and flip-flop could be a possible mechanism 628

involved in the nanotube stabilization whereas head-tail interactions have long been considered 629

as important in nanotube-forming bolaform amphiphiles6063 From mere considerations based 630

on T2 (1H) relaxation arguments one could exclude head-tail configurations if the R-CH2OH 631

endgroup (H1) was involved in a direct interaction with the sugar group its corresponding T2 632

would probably be comparable to the sugarrsquos one (OH and H1rsquo) but this does not seem to be 633

the case However this hypothesis would be based on the comparison of the relaxation from 634

CH and CH2 protons additionally involved in different kind of motions Besides the relaxation 635

of H1 is influenced by the strong overlap between H1 and H5rsquo For these reasons relaxation 636

arguments are not solid enough and will not be used 637

We then address this question by exploring intermolecular interactions in a direct 638

manner using 2D 1H-1H homonuclear ssNMR correlation spectroscopy At 90degC in the fast-639

tumbling domain of molecular motion (narrow 1H NMR peaks Figure 6a) and corresponding 640

to the vesicle-rich region spin diffusion (NOESY) experiments are preferred over dipolar 641

coupling based experiments because such interaction is averaged out by molecular motion 642

However when the molecular motion is more constrained in the domain of nanotube stability 643

at 70degC (broad 1H spectra Figure 6a) Double-Quantum Single-Quantum (DQ-SQ) correlation 644

experiments are best suited to probe interaction between close molecular neighbors These 645

experiments are reported in Figure 9 and the related interpretation is only briefly reported 646

below while a more detailed discussion can be found on Page 12 in the Supporting Information 647

26

648

Figure 9 ndash a) 2D 1H-1H NOESY NMR experiment performed at MAS= 10 kHz (4 mm probe) on a Bruker 649

300 MHz Avance III HD spectrometer using 300 ms of mixing time Sample temperature is regulated to 650

90degC reached by cooling from 110degC b) 2D 1H-1H DQ-SQ (BAck-to-BAck BABA pulse sequence used for 651

the DQ excitation scheme) NMR experiment performed at MAS= 22 kHz (25 mm probe) on a Bruker 700 652

MHz Avance III spectrometer using a DQ excitation time equal to 2 rotor periods Sample temperature is 653

regulated to 70degC reached by cooling from 110degC The typical schemes illustrating how to interpret these 654

2D experiments are reported above each experimental map Zooms of the sugar region and configurational 655

interpretation are given below each 2D map More information are given in the text below 656

H16-H16

H1-H1

H9-H10

H1rsquo-H1

H1rsquoH910 H16

H1

H1rsquo-H1rsquo

Hxrsquo-Hxrsquo

2

(pp

m)

(ppm)

315

811910 1rsquo

2rsquo2

3rsquo 4rsquo6rsquo

5rsquo

1rsquo2910

asymp asymp asympHDO MeOD Acetone

1

315

811 9101rsquo

2rsquo2

3rsquo4rsquo6rsquo

5rsquo

1rsquo2 9 10

asympasympasympHDOMeODAcetone

H1rsquo-H1

1rsquo

1

(p

pm

)

(ppm)

H1-H2

H1rsquo-H1

H1rsquo

H910

H1H5rsquo

H2

H1rsquo-H5rsquo

315

811910 1rsquo

2rsquo2

3rsquo 4rsquo6rsquo

5rsquo

1rsquo2910

asymp asymp asympHDO MeOD Acetone

1

H2rsquoH3rsquoH4rsquo

H2rsquoH3rsquoH4rsquo

X= 2rsquo3rsquo4rsquo

NOESY - T= 90degC

1H

1H

1H

1H

DQ-SQ - T= 70degC

a) b)

2

(pp

m)

(p

pm

)

(ppm)

A B

AhellipB

AhellipA

BhellipB

NOESY

(ppm)

A B

AhellipB

DQ-SQ

Single-quantum dimension

Do

ub

le-q

ua

ntu

m d

imen

sio

n

AhellipA

BhellipB

BhellipA

BhellipA

315

811910 1rsquo

2rsquo2

3rsquo 4rsquo6rsquo

5rsquo

1rsquo2910

asymp asymp asympHDO MeOD Acetone

1

lt 5 Aring

lt 5 Aring

H1-H2

H811

H15

H16

H6rsquo

H1rsquo

H1

H5rsquo

H2

H1-H2

H1rsquo-H1

H1rsquo-H5rsquo

H1rsquo

H1 H5rsquo

H1rsquo

H1

H5rsquo

H1H2rsquo H3rsquo H4rsquo

H1-H1

H1rsquo-H1

Hxrsquo-HxrsquoX= 2rsquo3rsquo4rsquo

H1-H2

H1rsquo

Zoom sugar region 28 ndash 45 ppmNOESY DQ-SQ

27

Head-tail or head-head configurations could be probed by the existence of H1rsquo-H1 657

interactions whereas H1 (δH= 356 ppm) is associated to the glucolipid tail and H1rsquo (δH= 432 658

ppm) to the sugar headgroup These atoms are more than 15 Aring apart and a direct interaction 659

could not be observed in both NOESY and DQ-SQ experiments NOESY experiments (Figure 660

9a) show that H1rsquo-H1 correlations are not observed Meanwhile clear-cut direct H1rsquo-H1 661

interactions are indeed observed are detected at T= 70degC (DQ cross-peaks related to H1rsquo-H1 662

correlation in DQ-SQ experiment shown in Figure 9b) Interestingly the absence of auto peaks 663

at δH= 432 ppm and δH= 335 ppm in the SQ dimension excludes H1rsquo-H1rsquo and Hxrsquo-Hxrsquo (x= 2 664

3 4) interactions respectively strengthening the idea that sugar headgroups from two nearest-665

neighbor G-C181-OH molecules are not adjacent in the nanotube-vesicle region 666

Homonuclear 2D ssNMR supports the idea according to which G-C181-OH can 667

undergo molecular flip-flops and a head-tail arrangement in the dry product Extrapolating these 668

conclusions in the lower concentration range lack of H1rsquo-H1 proximity at T= 90degC suggests a 669

head-head configuration in the vesicle-rich region and head-tail configuration within the 670

nanotubes It is however still unclear whether flip-flops are driven by spontaneous temperature-671

induced conformational changes in the glucolipid packing stimulating inclusions within the 672

membrane itself672223629ndash1218ndash21 or by defects generated by inter-vesicle collisions In fact a 673

combination of both is not unlikely Nanotubes are not stable when temperature is 20 to 30 674

degrees above Tm1 indicating that inter-vesicle collisions alone cannot explain the unusual 675

phenomenon In an energy landscape approach87 the head-head configuration is certainly 676

favored at high temperature while both head-head and head-tail configurations could be 677

favored in the mid-temperature region corresponding to the range between Tm1 and Tm2 in the 678

dry powder Both statistical and defect-induced flip-flop events contribute to form high mean 679

zero Gaussian curvature morphologies (nanotubes) for which the head-tail conformation 680

limiting the proximity between bulky sugar headgroups is probably more adapted (Figure 8c) 681

The flip-flop interpretation now reasonably explains the strong differences in the sugar and fatty 682

alcohol conformations deduced by the multiple NMR signals attributed to chemically-683

equivalent groups The existence of various low-energy conformations each having specific 684

NMR signature for sugars in solution78ndash8086 and solid-state81ndash84 further supports our 685

interpretation 686

One last point should be clarified Why between Tm1 and Tm2 are two molecular 687

environments able to coexist instead of a single one of average rigidity as found in the case of 688

ldquoripplerdquo phases50ndash54 We believe that this question can be answered based on the NMR results 689

the relatively short T2 values of the H1rsquo and OH protons shows the constrained mobility of the 690

28

sugar headgroup despite a certain flexibility of the chain (long T2 component) This trend may 691

suggest restrictions in translational mobility probably due to the sugar-OH hydrogen bonding 692

(Figure 8c) meaning that lipids are not able to diffuse within the membrane thus keeping two 693

different configurations stable 694

695

Conclusions 696

Topologically-connected nanotube-vesicle lipidic systems are commonly observed for 697

self-assembled membranes composed of at least two amphiphiles and spontaneously driven by 698

anisotropic inclusions andor by external directional forces This work shows that strong 699

differences in membrane curvature can spontaneously occur in single-lipid systems without 700

external anisotropic forces Coexistence and interconnection between nanotubes and vesicles 701

are evidenced via cryo-TEM and fluorescence confocal microscopy in a single-phase region at 702

temperature above a transition temperature Tm1= 483degC determined by DSC on the hydrated 703

sample but below the second transition temperature Tm2 (957degC) determined by DSC on the 704

dry powder X-ray diffraction (WAXS) also shows the coexistence of two lattice periods one 705

attributed to an inter-membrane repeating distance and the other possibly related to the intra-706

nanotube glucolipid arrangement Labelling the systems with Laurdan a fluorophore of which 707

the emission strongly varies with the local membrane rigidity excludes the presence of rigid 708

micro-separated phases (eg Lβ) 709

1H and 13C ssNMR despite the fact that it can only be performed on the dry powder 710

confirm the presence of one main fluid G-C181-OH environment above Tm2 Below Tm2 the 711

splitting of some 13C peaks and the coexistence of short and long 1H relaxation components 712

demonstrate that G-C181-OH can simultaneously pack into structurally and dynamically 713

inequivalent environments Combining advanced 2D 1H-1H correlation experiments based on 714

spin diffusion or double-quanta excitation suggests that vesicles could contain a preferred head-715

head configuration of the bolaform lipid while nanotubes a preferred head-tail assembly 716

The presence of two environments composed of the same molecule characterized by 717

different molecular configurations and rigidity fulfills the theoretical conditions necessary to 718

spontaneously nucleate a membrane inclusion of non-zero average orientation described as 719

necessary to observe topologically-connected nanotube-vesicle structures in the absence of 720

external forces This hypothesis could also be enriched by the possibility of nanotubes to be 721

pulled away between two vesicles after a collision event 722

This work expands the actual theoretical and experimental framework on lipid networks 723

stabilized by a different radius of curvature It could also motivate the synthesis or isolation 724

29

of new lipids providing a higher level of 3D architectural control in soft networks possibly in 725

the range of room temperature 726

727

Acknowledgements 728

Dr Andrea Lassenberger and Dr Sylvain Preacutevost at Institut Laue Langevin (ILL Grenoble 729

France) are kindly acknowledged for their assistance on the SANS experiment Confocal 730

images were performed at the Institut de Biologie Paris Seine (IBPS) imaging facility Jean-731

Franccedilois Gilles (Imaging Core Facility Institut de Biologie Paris Seine (IBPS) CNRS 732

Sorbonne Universiteacute Paris France) is kindly acknowledged for his assistance on the confocal 733

microscopy imaging Dr Guillaume Laurent (LCMCP Sorbonne Universiteacute Paris France) is 734

kindly acknowledged for helpful discussions The French region Ile-de-France SESAME 735

program is acknowledged for financial support (700 MHz NMR 736

spectrometer) Dr Olivier Diat (ICSM Marcoule France) Prof eronikaIglič and Prof V Aleš737

onsul discussi Slovenia) are kindly acknowledged for helpfUniversity of LjubljanaIglič (-Kralj 738

739

Financial support 740

WAXS experiments have been supported by Soleil Light Source Saint Aubin France proposal 741

Ndeg 20201747 SANS experiments have been supported by ILL proposal Ndeg9-13-778 742

743

References 744

1 D Kabaso N Bobrovska W Goacuteźdź E Gongadze V Kralj-Iglič R Zorec and A 745

Iglič Bioelectrochemistry 2012 87 204ndash210 746

2 C R Safinya U Raviv D J Needleman A Zidovska M C Choi M A Ojeda-747

Lopez K K Ewert Y Li H P Miller J Quispe B Carragher C S Potter M W 748

Kim S C Feinstein and L Wilson Adv Mater 2011 23 2260ndash2270 749

3 P Veranič M Lokar G J Schuumltz J Weghuber S Wieser H Haumlgerstrand V Kralj-750

Iglič and A Iglič Biophys J 2008 95 4416ndash4425 751

4 A Rustom R Saffrich I Markovic P Walther and H-H Gerdes Science (80- ) 752

2004 303 1007ndash1010 753

5 M I Angelova S Soleacuteau P Meacuteleacuteard F Faucon and P Bothorel Prog Colloid 754

Polym Sci 1992 89 127ndash131 755

6 L Mathivet S Cribier and P F Devaux Biophys J 1996 70 1112ndash1121 756

7 V Kralj-Iglič G Gomišček J Majhenc V Arrigler and S Svetina Colloids Surfaces 757

A Physicochem Eng Asp 2001 181 315ndash318 758

30

8 V Kralj-Iglič A Iglič G Gomišček F Sevšek V Arrigler and H Haumlgerstrand J 759

Phys A Math Gen 2002 35 1533ndash1549 760

9 M Drab D Stopar V Kralj-Iglič and A Iglič Cells 2019 8 626 761

10 A Zidovska K K Ewert J Quispe B Carragher C S Potter and C R Safinya 762

Langmuir 2009 25 2979ndash2985 763

11 A Zidovska K K Ewert J Quispe B Carragher C S Potter and C R Safinya Soft 764

Matter 2011 7 8363ndash8369 765

12 A Zidovska K K Ewert J Quispe B Carragher C S Potter and C R Safinya 766

Biochim Biophys Acta - Biomembr 2009 1788 1869ndash1876 767

13 M Karlsson K Sott M Davidson A S Cans P Linderholm D Chiu and O Orwar 768

Proc Natl Acad Sci U S A 2002 99 11573ndash11578 769

14 A Karlsson R Karlsson M Karlsson A-S Cans A Stromberg F Ryttsen and O 770

Orwar Nature 2001 409 150ndash152 771

15 M Karlsson K Sott A S Cans A Karlsson R Karlsson and O Orwar Langmuir 772

2001 17 6754ndash6758 773

16 E Evans H Bowman A Leung D Needham and D Tirrell Science (80- ) 1996 774

273 933ndash935 775

17 U Seifert Configurations of fluid membranes and vesicles 1997 vol 46 776

18 J Kaumls and E Sackmann Biophys J 1991 60 825ndash844 777

19 N Bobrovska W Goacuteźdź V Kralj-Iglič and A Iglič PLoS One 2013 8 e73941 778

20 A Iglič B Babnik K Bohinc M Fošnarič H Haumlgerstrand and V Kralj-Iglič J 779

Biomech 2007 40 579ndash585 780

21 K Bohinc T Slivnik A Iglič M Brumen and V Kralj-Iglič J Phys Chem C 2007 781

111 9709ndash9718 782

22 A Iglič H Haumlgerstrand P Veranič A Plemenitaš and V Kralj-Iglič J Theor Biol 783

2006 240 368ndash373 784

23 V Kralj-Iglič and P Veranič Adv Planar Lipid Bilayers Liposomes 2006 5 129ndash785

149 786

24 A Iglic and V Kralj-Iglic Adv Planar Lipid Bilayers Liposomes 2006 4 253ndash279 787

25 S Penič L Mesarec M Fošnarič L Mroacutewczyńska H Haumlgerstrand V Kralj-Iglič and 788

A Iglič Front Phys 2020 8 342 789

26 P Yager and P E Schoen Mol Cryst Liq Cryst 1984 106 371ndash381 790

27 T G Barclay K Constantopoulos and J Matisons Chem Rev 2014 114 10217ndash791

10291 792

31

28 B Božič G Gomišček V Kralj-Iglič S Svetina and B Žekš Eur Biophys J 2002 793

31 487ndash496 794

29 A Iglič H Haumlgerstrand M Bobrowska-Haumlgerstrand V Arrigler and V Kralj-Iglič 795

Phys Lett A 2003 310 493ndash497 796

30 R Bar-Ziv and E Moses Phys Rev Lett 1994 73 1392ndash1395 797

31 S Chiruvolu H E Warriner E Naranjo S H Idziak J O Raumldler R J Plano J A 798

Zasadzinski and C R Safinya Science (80- ) 1994 266 1222ndash1225 799

32 V Kralj-Iglič A Iglič H Haumlgerstrand and P Peterlin Phys Rev E 2000 61 4230ndash800

4234 801

33 L Mesarec M Drab S Penič V Kralj-Iglič and A Iglič Int J Mol Sci 2021 22 802

2348 803

34 L Van Renterghem S L K W Roelants N Baccile K Uyttersprot M C Taelman 804

B Everaert S Mincke S Ledegen S Debrouwer K Scholtens C V Stevens W 805

Soetaert L Van Renterghem S L K W Roelants N Baccile K Uytersprot M C 806

Taelman B Everaert S Mincke S Ledegen S Debrouwer K Scholtens C V 807

Stevens and W Soetaert Biotechnol Bioeng 2018 115 1195ndash1206 808

35 J Clark and F Deswarte Introduction to Chemicals from Biomass Second Edition 809

John Wiley amp Sons 2015 810

36 K V Sajna R Houmlfer R K Sukumaran L D Gottumukkala and A Pandey in 811

Industrial Biorefineries and White Biotechnology eds A Pandey R Houmlfer M 812

Taherzadeh K M Nampoothiri and C Larroche Amsterdam Oxford Waltham 813

Elsevier 2015 pp 499ndash521 814

37 R Marchant and I M Banat Trends Biotechnol 2012 30 558ndash565 815

38 K Hill and C LeHen-Ferrenbach in Sugar-Based Surfactants Fundamentals and 816

Applications ed C C Ruiz Boca Raton CRC Press 2009 pp 1ndash20 817

39 L Rodrigues I M Banat J Teixeira and R Oliveira J Antimicrob Chemother 818

2006 57 609ndash618 819

40 J D Desai and I M Banat Microbiol Mol Biol Rev 1997 61 47ndash64 820

41 S L K W Roelants K Ciesielska S L De Maeseneire H Moens B Everaert S 821

Verweire Q Denon B Vanlerberghe I N A Van Bogaert P Van der Meeren B 822

Devreese and W Soetaert Biotechnol Bioeng 2016 113 550ndash559 823

42 N Baccile M Selmane P Le Griel S Preacutevost J Perez C V Stevens E Delbeke S 824

Zibek M Guenther W Soetaert I N A Van Bogaert and S Roelants Langmuir 825

2016 32 6343ndash6359 826

32

43 N Baccile A-S Cuvier S Preacutevost C V Stevens E Delbeke J Berton W Soetaert 827

I N A Van Bogaert and S Roelants Langmuir 2016 32 10881ndash10894 828

44 N Baccile C Seyrig A Poirier S A Castro S L K W Roelants and S Abel Green 829

Chem 2021 23 3842ndash3944 830

45 A Salis L Cappai C Carucci D F Parsons and M Monduzzi J Phys Chem Lett 831

2020 11 6805ndash6811 832

46 J Schindelin I Arganda-Carreras E Frise V Kaynig M Longair T Pietzsch S 833

Preibisch C Rueden S Saalfeld B Schmid J-Y Tinevez D J White V 834

Hartenstein K Eliceiri P Tomancak and A Cardona Nat Methods 2012 9 676ndash835

682 836

47 T Parasassi and E Gratton J Fluoresc 1995 5 59ndash69 837

48 M M Dodes Traian F L Gonzaacutelez Flecha and V Levi J Lipid Res 2012 53 609ndash838

616 839

49 D M Owen C Rentero A Magenau A Abu-Siniyeh and K Gaus Nat Protoc 840

2012 7 24ndash35 841

50 W Stillwell in An Introduction to Biological Membranes Academic Press 2016 pp 842

135ndash180 843

51 W Stillwell in An Introduction to Biological Membranes Academic Press 2016 pp 844

181ndash220 845

52 B Tenchov Chem Phys Lipids 1991 57 165ndash177 846

53 R Koynova and M Caffrey Biochim Biophys Acta 1998 1376 91ndash145 847

54 F Schmid S Dolezel O Lenz and S Meinhardt J Phys Conf Ser 2014 487 848

012004 849

55 S Meinhardt R L C Vink and F Schmid Proc Natl Acad Sci U S A 2013 110 850

4476ndash4481 851

56 O Glatter and O Kratky Small Angle X-ray Scattering Academic Press London 852

1982 853

57 B A Cunningham A D Brown D H Wolfe W P Williams and A Brain Phys 854

Rev E 1998 58 3662ndash3672 855

58 M Nadler A Steiner T Dvir O Szekely P Szekely A Ginsburg R Asor R Resh 856

C Tamburu M Peres and U Raviv Soft Matter 2011 7 1512ndash1523 857

59 C Tanford The hydrophobic effect Formation of micelles and biological membranes 858

John Wiley amp Sons Inc New York 1973 859

60 M Masuda and T Shimizu Langmuir 2004 20 5969ndash5977 860

33

61 A Zidovska K K Ewert J Quispe B Carragher C S Potter and C R Safinya 861

Block Liposomes Vesicles of Charged Lipids with Distinctly Shaped Nanoscale 862

Sphere- Pear- Tube- or Rod-Segments Elsevier Inc 1st edn 2009 vol 465 863

62 A Iglic V Kralj-Iglic B Božič M Bobrowska-Haumlgerstrand B Isomaa and H 864

Haumlgerstrand Bioelectrochemistry 2000 52 203ndash211 865

63 T Shimizu M Masuda and H Minamikawa Chem Rev 2005 105 1401ndash1444 866

64 J V Selinger and J M Schnur Phys Rev Lett 1993 71 4091ndash4094 867

65 J M Schnur Science (80- ) 1993 262 1669ndash1676 868

66 B Tenchov R Koynova and G Rapp Biophys J 2001 80 1873ndash1890 869

67 R Koynova and B Tenchov Ser Biomech 2017 31 3ndash19 870

68 S J Angyal and R Le Fur Carbohydr Res 1984 126 15ndash26 871

69 J K Whitesell and M A Minton J Am Chem Soc 1987 109 225ndash228 872

70 T Takayama S Ando and I Ando J Mol Struct 1990 220 245ndash250 873

71 K Gawrisch N V Eldho and I V Polozov Chem Phys Lipids 2002 116 135ndash151 874

72 F Tielens C Gervais J F Lambert F Mauri and D Costa Chem Mater 2008 20 875

3336ndash3344 876

73 G A Jeffrey and Y Yeon Acta Crystallogr Sect B 1986 42 410ndash413 877

74 B Berglund and R W Vaughan J Chem Phys 1980 73 2037ndash2043 878

75 C Gervais C Coelho T Azaiumls J Maquet G Laurent F Pourpoint C Bonhomme 879

P Florian B Alonso G Guerrero P H Mutin and F Mauri J Magn Reson 2007 880

187 131ndash140 881

76 C Akita T Kawaguchi F Kaneko H Yamamoto and M Suzuki J Phys Chem B 882

2004 108 4862ndash4868 883

77 R N Purusottam L Seacutenicourt J J Lacapegravere and P Tekely Biochim Biophys Acta - 884

Biomembr 2015 1848 3134ndash3139 885

78 X Biarneacutes A Ardegravevol A Planas C Rovira A Laio and M Parrinello J Am Chem 886

Soc 2007 129 10686ndash10693 887

79 L M J Kroon‐Batenburg and J Kroon Biopolymers 1990 29 1243ndash1248 888

80 K N Kirschner and R J Woods Proc Natl Acad Sci U S A 2001 98 10541ndash889

10545 890

81 S Svenson B Kirste and J H Fuhrhop J Am Chem Soc 1994 116 11969ndash11975 891

82 I Sack S Macholl J H Fuhrhop and G Buntkowsky Phys Chem Chem Phys 892

2000 2 1781ndash1788 893

83 S Svenson J Koening and J H Fuhrhop J Phys Chem 1994 98 1022ndash1028 894

34

84 S Svenson A Schaefer and J H Fuhrhop J Chem Soc Perkin Trans 2 1994 2 895

1023ndash1028 896

85 K Bock and C Pedersen Adv Carbohydr Chem Biochem 1983 41 27ndash66 897

86 T B Grindley Structure and Conformation of Carbohydrates Springer Berlin 898

Heidelberg 2008 899

87 F Tantakitti J Boekhoven X Wang R V Kazantsev T Yu J Li E Zhuang R 900

Zandi J H Ortony C J Newcomb L C Palmer G S Shekhawat M O De La 901

Cruz G C Schatz and S I Stupp Nat Mater 2016 15 469ndash476 902

903

1

Supporting Information 904

905

Topological connection between vesicles and nanotubes in single-906

component lipid membranes driven by head-tail interactions 907

908

Niki Baccilea Ceacutedric Lorthioira Abdoul Aziz Baa Patrick Le Griela Cristina Coelhob Javier 909

Perezc Wim Soetaertd Sophie L K W Roelantsd 910

911

a Sorbonne Universiteacute Centre National de la Recherche Scientifique Laboratoire de Chimie de 912

la Matiegravere Condenseacutee de Paris LCMCP F-75005 Paris France 913

b Institut des Mateacuteriaux de Paris Centre Sorbonne Universiteacute Paris France 914

c Synchrotron Soleil LrsquoOrme des Merisiers Saint-Aubin BP48 91192 Gif-sur-Yvette Cedex 915

France 916

d InBio Department of Biotechnology Ghent University Ghent Belgium 917

918

919

Corresponding author 920

Dr Niki Baccile 921

E-mail address nikibaccilesorbonne-universitefr 922

Phone +33 1 44 27 56 77 923

924

925

926

2

a) 927

Parameter Method

Dry Matter (DM ) 996 Infrared balance 105degC

Glucose () 012 HPLC-Metacarb

Glycerol () nd HPLC-Metacarb

Free Fatty Acid content () lt001 Internal method BBEPP

GC with FID Detector

Oil () lt001 Internal method BBEPP

GC with FID Detector

Moulds (CFUg) lt10 3M-Nordval ndeg16 B

Yeasts (CFUg) lt10 3M-Nordval ndeg16 B

Anaerobic count (CFUg) lt10 SP-VG M005 B

Protein () tbd BCA protein assay

928

Figure S 1 - a) HPLC-ELSD chromatogram and composition table 929

930

931

932

933

3

b) 934

935

Group Ndeg δ ppm

CH2 1 354

CH2 2 153

CH2 3-712-14 132

CH2 811 203

CH 910 534

CH2 15 162

CH2 16 388

CH 1 424

CH 2 316

CH 3 334

CH 4 329

CH 5 354

CH2 6 366

Figure S 1 (next) ndash b) 1H NMR spectrum (MeOD-d4) recorded on the purified G-C181-OH compounds 936

and related assignment 937

938

2

811910 1rsquo

5rsquo1

3rsquo4rsquo

1rsquo1910

asymp asymp asympHDO MeOD Acetone

6rsquo2rsquo

1615

3-712-14

4

939

940

Figure S 2 ndash Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 130degC 941

5

942

943

Figure S 3 - Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 90degC 944

945

90degC

1

22

6

946

947

948

949

Figure S 4 ndash Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 70degC 950

951

21

11

1

2

5

3

4

45

5

6

7

7

7

952

953

954

Figure S 5 ndash Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 25degC 955

956

957

1

1 1

8

958

Figure S 6 ndash a) SANS experiment recorded at 90degC for a 5 mgmL G-C181-OH aqueous solution b-c) 959

WAXS experiments recorded on aqueous solutions of G-C181-OH at b) 5 mgmL and c) 25 mgmL d) 960

Structural interpretation of the repeating distances pointed out in c) 961

962

963

964

001 01 1

q Aring-1

25degC

68degC

88degC

111degC

01 1q Aring

-1

02 04 06 08 1q Aring

-1

001 01

10-1

100

101

Inte

nsity

cm

-1

q Aring-1

-2

T= 90degC

SANS WAXS WAXS

a) b) c)

dv

dL d= 144 Aring-1

di

ddH2O

dt

dvdL

dtdi d

d)

9

965

966

Figure S 7 ndash ssNMR MAS (10 kHz) experiments recorded on a G-C181-OH dry powder a) 1H spectra 967

recorded from 110degC to 70degC b) 13C single pulse (SP) spectra recorded with high power 1H decoupling from 968

110degC to 70degC c) 13C cross polarization (CP) spectra recorded at 110degC and contact time tc of 50 micros and 969

3000 micros d) 13C CP spectra recorded at 70degC and for 20 lt tc micros lt 3000 970

971

1

811

910

CH2 CH2

1rsquo

6rsquo

8 6 4 2 0 ppm

110

90

80

70

910

1rsquo

6rsquo16

1

2rsquo3rsquo4rsquo

811

OH

45 40 35 30 25 20 15

ppm

200 180 160 140 120 100 80 60 40 20 0

ppm140 120 100 80

ppm

CH2-gauche

3004

110

90

80

70

65

60

50

25

C91013003

C9 - 12869

CH2-all-trans

3198

C1rsquo10335

a) b)

1H

T degCT degC

13C Single Pulse MAS

A

B

A

B

70degC70degC

5rsquo

215

200 180 160 140 120 100 80 60 40 20 0

ppm

13C CPMAST= 110degC

200 180 160 140 120 100 80 60 40 20 0

ppm

132 131 130 129 128 127 126

ppm

106 105 104 103 102

ppm

13C CPMAST= 70degC

50

tc micros

3000

20

tc micros

50 500 1000 3000

tc micros

C102

13024 C910113003

C92

12869

C1rsquo213082

C1rsquo110443

A B

106 105 104 103 102

ppm

3000

1000

500

50

20

AB

c)

d)

10

972

Figure S 8 ndash a) Short and d) long T2 components displayed by the 1H transverse relaxation signal c) 1H 973

transverse relaxation signals recorded on a G-C181-OH dry powder from 110degC to 70degC The area of each 974

given resonance is plotted (log-lin) as a function of the echo time (CPMG pulse sequence) for each 975

temperature and fitted using the equation given above T2s and T2l indicate the short and long T2 relaxation 976

times respectively obtained by describing the 1H transverse relaxation signal using a biexponential decay 977

For H1rsquo H5rsquo and H6rsquo the 1H relaxation function could be described by a monoexponential decay 978

- Protons from the sugar group (H1rsquo H5rsquo H6rsquo) show a single T2 relaxation component (Figure 979

6c and Figure S 8) while protons of the acyl chain (CH2 H910 H811) show both short (Figure 980

6c) and long (Figure 6d) T2 relaxation component 981

- The short T2 relaxation time systematically decreases from 110degC to 70degC for all species by a 982

factor ranging from two to seven depending on the specific chemical group Very long T2 are 983

observed at 110degC for all acyl species while nearly unchanged values are measured below 984

90degC without any clear trend 985

- The sugar protons (H1rsquo H5rsquo H6rsquo) have the shortest set of T2 values (lt 1 ms) below 100degC 986

The long component of T2 generally ranges between 4 ms and 12 ms 987

0 1 2 3 4 5 6 7 8

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1 2 3 4 5 6 7 8

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

H1

H8H11

H9H10

CH2

H1rsquo

H6rsquo

CH2

H910H811 H1CH2

H1rsquo H5rsquo6rsquo

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

OH

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45 T= 110degC

T= 90degC

T= 80degC

T= 70degC

Log

In

ten

sity

Time ms

Fit

)

CH2 H910 H811 H1 H56 H1 OH

00

05

10

15

20

25

30

35

40

T2 m

s

T= 110degC

T= 90degC

T= 80degC

T= 70degC

CH2 H910 H811 H1

0

2

4

6

8

10

12

T2 m

s

T= 110degC

T= 90degC

T= 80degC

T= 70degC

Short T2(1H) Long T2(1H)

a) b)

c)

11

Table S 1 ndash 1H chemical shifts of G-C181-OH in solution and in the solid-state Solid-state spectra are 988

recorded at a MAS frequency of 10 kHz in DMSO-d6 broad peak 989

990

δH ppm

Group Ndeg Liquid

(MeOD-d4) Solid

(T= 110degC) Solid

(T= 70degC)

CH2 H1 354 357

CH2 H2 153 156

CH2 H3-712-14 132 135

CH2 H811 203 203

CH H910 534 534

CH2 H15 162 164

CH2 H16 388 383

CH H1 424 432

CH H2 316 335

CH H3 334 335

CH H4 329 335

CH H5 354 351 357

CH2 H6 366 383

OH OH1 432

45 463

OH OH2 492

OH OH3 490

OH OH4 487

OH OH6 445

991

992

H1H9H10H1rsquo

H6rsquo

OH2rsquo

OH1

12

Detailed discussion about the NOESY and DQ-SQ 2D 1H-1H ssNMR experiments presented in 993

Figure 9 994

995

2D 1H-1H ssNMR experiments are a unique tool to explore through-space proximities 996

between 1H nuclei ssNMR offers a large number of possibilities each adapted to a given 997

system The advantage of 2D maps with respect to a dynamics approach is their direct 998

readability in clear two nuclei which exchange their magnetization through-space (whichever 999

mechanism may be responsible for this exchange) will display a specific cross-peak However 1000

the way cross-peaks are presented and interpreted strictly depends on the nature of the pulse 1001

program which in turn is chosen according to the physical nature of the sample 1002

1D temperature-dependent 1H NMR experiments performed on G-C181-OH (Figure 1003

4a in the main text) indicate sharp peaks at 90degC and broad peaks at 70degC Narrow peaks 1004

indicate fast andor large-amplitude molecular tumbling averaging the strong 1H-1H dipolar 1005

interactions This excludes the use of pulse sequences like DQ-SQ which is not efficient under 1006

these conditions For this reason we employ spin-diffusion-based pulse sequences like 1007

NOESY which is more adapted for samples with high molecular mobility When mixing times 1008

are chosen to be short (lt 1 s) one generally explores through-space proximities within about 5 1009

10 Aring On the contrary when molecular motions are restricted dipolar coupling becomes 1010

significant 1H NMR peaks are broader and pulse sequences like DQ-SQ can now be used 1011

with a spatial resolution of less than 5 Aring 1012

Whenever possible DQ-SQ experiments are preferred over spin diffusion-based pulse 1013

sequences because they are more selective in particular for likewise interactions which are 1014

not discriminated in NOESY-type experiments This is illustrated by the typical interaction 1015

schemes for NOESY (top Figure 9a) and DQ-SQ (top Figure 9b) For two given A and B 1016

protons 2D NOESY maps will display two on-diagonal and two square-shaped symmetrical 1017

off-diagonal cross-peaks Off-diagonal cross-peaks illustrate the mutual A-B and B-A 1018

interaction but the on-diagonal peaks are systematic and do not prove any A-A or B-B 1019

interaction Similarly to NOESY 2D DQ-SQ maps display off-diagonal peaks (symmetric 1020

across the diagonal top Figure 9b) upon A-B and B-A interaction but in contrast they will 1021

display on-diagonal peaks only in the presence of self-interaction In the tutorial example of 1022

Figure 9b top the on-diagonal A-A peak indicates mutual A-A interaction while lack of on-1023

diagonal B-B peak evidences the absence of B-B proximities Combining NOESY and DQ-SQ 1024

experiments is capital to study the local arrangement of G-C181-OH and its evolution with 1025

temperature 1026

13

At T= 90degC the H1 and H5rsquo contributions are very close but possible to discriminate 1027

them (δ(H1)= 354 ppm δ(H5rsquo)= 351 ppm) according to 1H NMR chemical shifts in Table S 1028

1 2D 1H-1H NOESY in Figure 9a also discriminates between the two components attributed 1029

on the basis of the interactions with their mutual neighbors (green H1-H2 and H5rsquo-H1rsquo squares 1030

in Figure 9a) On this basis 2D NOESY does not show a direct H1rsquo-H1 correlation (blue red-1031

crossed) but rather an extensive network of intra-sugar interactions These features exclude 1032

suggest intra- and inter-sugar interactions and they exclude head-tail arrangements as a major 1033

molecular configuration at 90degC 1034

At 70degC the 1H NMR peaks are broader indicating a reduction of the T2 (1H) relaxation 1035

times andor residual homonuclear dipolar interactions excluding the use of NOESY In this 1036

case 1H-1H DQ-SQ correlation experiments are best suited Figure 9b presents the 1H-1H DQ-1037

SQ experiment on G-C181-OH at 70degC Peaks are broader than in NOESY and attribution may 1038

be cumbersome However the experiment is performed at 700 MHz and MAS= 22 kHz which 1039

improves spectral resolution with respect to the same experiment performed at 300 MHz and 1040

MAS= 10 kHz The auto-peak at δH= 356 ppm in the SQ dimension is attributed to H1-H1 spin 1041

pairs mainly resulting from intra CH2 interaction while the DQ cross-peaks at δH= 156-356 1042

ppm and δH= 432-356 ppm respectively origin from H2-H1 and H1rsquo-H1 interactions (Figure 1043

9b) DQ-SQ map does not show any auto-peak corresponding to H1rsquo-H1rsquo or Hxrsquo-Hxrsquo (x= 2rsquo 1044

3rsquo 4rsquo) interactions (red-crossed blue Figure 9b) thus excluding any proximity (lt 5 Aring) between 1045

close glucose neighbors On the contrary DQ-SQ maps (recorded both at 700 MHz Figure 9b 1046

or at 300 MHz not shown) systematically show DQ cross-peaks which can only be attributed 1047

to H1rsquo-H1 interactions (δH= 432-356) The presence of these H1rsquo-H1 interactions can only be 1048

explained by a head-tail arrangement of the G-C181-OH molecules as illustrated below the 1049

DQ-SQ map in Figure 9b 1050

21

resonances for anomeric C1rsquo1 and C1rsquo2 cis-olefin C9101 and C92 C102 and CH2 gauche and 533

all-trans environments) and relaxation behaviour (fast and slow T2 relaxation components) 534

strongly suggests the coexistence of two molecular environments Further proofs of the 535

coexistence of two distinct molecular domains are given by a number of chemical shift 536

variations in the 13C CPMAS spectrum at 70degC compared to the one at 110degC 90degC and 537

solution This is shown in Figure 7 and in the list of 13C chemical shifts in Table 1 538

539

540

Figure 7 - Comparison between solution- and solid-state 13C NMR spectra of G-C181-OH Solid-state 541

experiments are recorded directly on the sample powder at a MAS of 10 kHz (4 mm zirconia rotor) SP 542

Single Pulse CP Cross Polarization In all the ssNMR experiments temperature is first raised at 110degC 543

and then lowered to either 90degC or 70degC Segmented lines are meant to highlight the known peak position 544

in solution thus stressing the strong chemical shift variations in the solid-state 545

546

NMR experiments above Tm2 suggest that vesicles could contain G-C181-OH with a 547

single family of mobile conformations characterized by chemically-equivalent C=C (red 548

Figure 8b) and a single C1rsquo conformation (green Figure 8b) Experiments between Tm1 and Tm2 549

otherwise suggest that two coexisting conformations characterized by chemically-equivalent 550

(red Figure 8b) and inequivalent (magenta Figure 8b) C=C could be associated to nanotubes 551

Two distinct C1rsquo sites (grey Figure 8b) also indicate two configurations of the sugar headgroup 552

A comment on this point will be given in the following paragraphs 553

The local (Aring-scale) variation of molecular mobility below Tm2 drives the coexistence of 554

two distinct molecular environments of the same G-C181-OH molecule This occurs within a 555

single-phase domain (as deduced by WAXS) of equivalent membrane rigidity (as derived from 556

Laurdan probe in confocal microscopy) Although never reported before the difference in the 557

local mobility of the same amphiphile (here G-C181-OH) could satisfy the theoretically-558

140 120 100 80 60 40 20

CP(3 ms)-70degC

SP-70degC

SP-90degC

SP-110degC

ppm

LiquidC910

C1rsquo

C2rsquo

C4rsquo

C3rsquo

C1

C6rsquoC5rsquo

C1

6

C8

11

C2 C

H2

C1

5

CH

2

DM

SO

13C

22

required conditions that induce local lipid segregation and formation of membrane inclusions 559

(Figure 8c) In case of their non-zero average orientation the local membrane curvature 560

increases62 and nanotubes can spontaneously form (Figure 8bc) More intriguing this 561

mechanism occurs at least 10degC above Tm1 although it should not be favored above the Tm27 562

From both experimental and theoretical backgrounds the conditions of topologically-563

connected vesicles-nanotubes systems are met only when either internal (two components) or 564

external driving forces are applied19 In the present single-lipid system of which the phase 565

behavior is summarized in Figure 8a the most plausible driving force could be explained by 566

the NMR results collected on the dry sample G-C181-OH concomitantly occupies two 567

molecular environments characterized by 1) two distinct conformations of the sugar 568

displaying equivalent mobility and 2) two conformations of the acyl chain unsaturation 569

570

23

571

Figure 8 ndash a) Thermotropic phase behavior of G-C181-OH at 5 mgmL-1 Note that Tm2 is observed on the 572

dry powder only b) Possible relationship between the molecular environment as derived from NMR and 573

membrane curvatures All the molecular drawings refer to the same G-C181-OH molecule (bulky ellipsoid 574

equiv sugar headgroup) while each color corresponds to a given configuration Grey of C1rsquo1 and C1rsquo2 indicates 575

the uncertainty about the assignment of the specific NMR peak The curvature C is defined as 1R R being 576

the curvature radius R1 R2 radius of curvature CG Gauss curvature CM Mean curvature c) Free 577

interpretation of the anisotropic high-curvature (CMgt 0 CG= 0) inclusions in the membrane (low CM and 578

CG) continuum driven by inter-sugar H-bonding and reduced intramembrane diffusion for a G-C181-OH 579

glucolipid 580

581

Flat crystalline structure

Nanotube-vesicle network

Vesicles

48degC

120deg

106 105 104 103 102 101

ppm

H9 H10H1rsquo

106 105 104 103 102 101

ppm106 105 104 103 102 101

ppm106 105 104 103 102 101

ppm

C=CC=C

C1rsquo1C1rsquo2

C1rsquo

Tm2(96degC)Tm1

a)

b)

c)

Same molecule different local orientations and sugar conformations

High curvature membrane

Low curvature inclusion

Flip-FlopStrong H-bonding

CG

CM ( + )

CG 0

CM 0 CG 0

CM 0

G-C181-OH

24

Two related questions are unanswered at the moment What actually drives nanotube 582

formation and is the organization of G-C181-OH within the vesicles and tubes the same Lipid 583

nanotubes have been studied for at least three decades over which broad experimental and 584

theoretical insight has been generated Despite this fact the actual understanding is still matter 585

of debate27 Micelle-to-nanotube and vesicle-to-nanotube phase transitions are generally 586

reported for single amphiphiles below a given Tm and intermediate phase composed of twisted 587

and flat fibers is concomitantly reported Initially thought to be a chirality-driven phenomenon 588

recent studies propose a more general mechanism involving crystallization and symmetry 589

breaking in the lipid packing driven by optimal packing27 If twisted ribbons were observed 590

before for a number of microbial glycolipid bolaamphiphiles44 they are not found for G-C181-591

OH This fact as well as the large body of cryo-TEM images and the observation of nanotubes 592

above Tm1 rather pledge in favour of a direct topological change from positive (vesicles) to 593

zero (nanotubes) Gaussian curvatures In this case vesicle-to-nanotube transitions could be 594

driven by external forces56307813ndash162829 595

Considering the absence of external forces in the present system there are only two 596

possibilities left spontaneous budding of the membrane or nanotube pulling after inter-vesicle 597

collisions due to Brownian motion Both are possible as inclusions composed of a different 598

arrangement of the same molecule in the membrane continuum can theoretically explain 599

spontaneous nanotube formation in an isotropic medium672223629ndash1218ndash21 However the second 600

one is probably more likely to prevail because it could explain the large number of nanotubes 601

connecting two distant vesicles Interestingly similar events occur for any phospholipid-based 602

vesicular solution but nanotubes are extremely rare unless at least two lipids of different 603

molecular structure are mixed together The spectroscopic 1H and 13C and relaxation (1H) 604

ssNMR data even if collected on a dry powder demonstrate that the coexistence of two G-605

C181-OH configurations involving both the sugar headgroup and fatty alcohol chain are 606

possible for this molecule As a matter of fact this is not so surprising as more than one energy 607

minima each corresponding to a given conformation are known for glucose its non-reducing 608

derivative methylglucose and carbohydrates in general78ndash80 609

The presence of two configurations could explain the coexistence between two 610

structures of different curvature but the data presented so far do not explain the origin of these 611

two configurations In other words what drives the change in the local structure of the sugar 612

headgroup and fatty alcohol within the membrane 613

The association of known self-assembled structures to spectroscopic 13C ssNMR data is 614

known for years in particular for gluconamides lipids having a linear sugar headgroup81ndash84 It 615

25

was found that gauche effects in some cases up to the γ position in the sugar chain induce 616

upfield chemical shifts while anti and trans conformations result in downfield shifts Each 617

structure (monolayer crystal bilayer crystal or micellar fiber) was then characterized by a given 618

organization of the gluconamide and a characteristic 13C ssNMR fingerprint which could be 619

explained by gauche γ-gauche anti and trans conformations of the headgroup Similar 620

structural-spectroscopic correlations have been collected over the years for much simpler cyclic 621

sugars85 whereas pyranoses can undergo important conformational changes (chair half-chair 622

boat skew) when the appropriate energy barriers are crossed78ndash8086 A quantitative 623

conformational study based on the 13C chemical shift variations shown in Table 1 would be 624

certainly helpful but very delicate in the context of the present state of the art because a 625

rigorous approach requires at least one known crystal structure8182 which is currently missing 626

for this compound A qualitative approach will then be adopted 627

G-C181-OH is a bolaform amphiphile and flip-flop could be a possible mechanism 628

involved in the nanotube stabilization whereas head-tail interactions have long been considered 629

as important in nanotube-forming bolaform amphiphiles6063 From mere considerations based 630

on T2 (1H) relaxation arguments one could exclude head-tail configurations if the R-CH2OH 631

endgroup (H1) was involved in a direct interaction with the sugar group its corresponding T2 632

would probably be comparable to the sugarrsquos one (OH and H1rsquo) but this does not seem to be 633

the case However this hypothesis would be based on the comparison of the relaxation from 634

CH and CH2 protons additionally involved in different kind of motions Besides the relaxation 635

of H1 is influenced by the strong overlap between H1 and H5rsquo For these reasons relaxation 636

arguments are not solid enough and will not be used 637

We then address this question by exploring intermolecular interactions in a direct 638

manner using 2D 1H-1H homonuclear ssNMR correlation spectroscopy At 90degC in the fast-639

tumbling domain of molecular motion (narrow 1H NMR peaks Figure 6a) and corresponding 640

to the vesicle-rich region spin diffusion (NOESY) experiments are preferred over dipolar 641

coupling based experiments because such interaction is averaged out by molecular motion 642

However when the molecular motion is more constrained in the domain of nanotube stability 643

at 70degC (broad 1H spectra Figure 6a) Double-Quantum Single-Quantum (DQ-SQ) correlation 644

experiments are best suited to probe interaction between close molecular neighbors These 645

experiments are reported in Figure 9 and the related interpretation is only briefly reported 646

below while a more detailed discussion can be found on Page 12 in the Supporting Information 647

26

648

Figure 9 ndash a) 2D 1H-1H NOESY NMR experiment performed at MAS= 10 kHz (4 mm probe) on a Bruker 649

300 MHz Avance III HD spectrometer using 300 ms of mixing time Sample temperature is regulated to 650

90degC reached by cooling from 110degC b) 2D 1H-1H DQ-SQ (BAck-to-BAck BABA pulse sequence used for 651

the DQ excitation scheme) NMR experiment performed at MAS= 22 kHz (25 mm probe) on a Bruker 700 652

MHz Avance III spectrometer using a DQ excitation time equal to 2 rotor periods Sample temperature is 653

regulated to 70degC reached by cooling from 110degC The typical schemes illustrating how to interpret these 654

2D experiments are reported above each experimental map Zooms of the sugar region and configurational 655

interpretation are given below each 2D map More information are given in the text below 656

H16-H16

H1-H1

H9-H10

H1rsquo-H1

H1rsquoH910 H16

H1

H1rsquo-H1rsquo

Hxrsquo-Hxrsquo

2

(pp

m)

(ppm)

315

811910 1rsquo

2rsquo2

3rsquo 4rsquo6rsquo

5rsquo

1rsquo2910

asymp asymp asympHDO MeOD Acetone

1

315

811 9101rsquo

2rsquo2

3rsquo4rsquo6rsquo

5rsquo

1rsquo2 9 10

asympasympasympHDOMeODAcetone

H1rsquo-H1

1rsquo

1

(p

pm

)

(ppm)

H1-H2

H1rsquo-H1

H1rsquo

H910

H1H5rsquo

H2

H1rsquo-H5rsquo

315

811910 1rsquo

2rsquo2

3rsquo 4rsquo6rsquo

5rsquo

1rsquo2910

asymp asymp asympHDO MeOD Acetone

1

H2rsquoH3rsquoH4rsquo

H2rsquoH3rsquoH4rsquo

X= 2rsquo3rsquo4rsquo

NOESY - T= 90degC

1H

1H

1H

1H

DQ-SQ - T= 70degC

a) b)

2

(pp

m)

(p

pm

)

(ppm)

A B

AhellipB

AhellipA

BhellipB

NOESY

(ppm)

A B

AhellipB

DQ-SQ

Single-quantum dimension

Do

ub

le-q

ua

ntu

m d

imen

sio

n

AhellipA

BhellipB

BhellipA

BhellipA

315

811910 1rsquo

2rsquo2

3rsquo 4rsquo6rsquo

5rsquo

1rsquo2910

asymp asymp asympHDO MeOD Acetone

1

lt 5 Aring

lt 5 Aring

H1-H2

H811

H15

H16

H6rsquo

H1rsquo

H1

H5rsquo

H2

H1-H2

H1rsquo-H1

H1rsquo-H5rsquo

H1rsquo

H1 H5rsquo

H1rsquo

H1

H5rsquo

H1H2rsquo H3rsquo H4rsquo

H1-H1

H1rsquo-H1

Hxrsquo-HxrsquoX= 2rsquo3rsquo4rsquo

H1-H2

H1rsquo

Zoom sugar region 28 ndash 45 ppmNOESY DQ-SQ

27

Head-tail or head-head configurations could be probed by the existence of H1rsquo-H1 657

interactions whereas H1 (δH= 356 ppm) is associated to the glucolipid tail and H1rsquo (δH= 432 658

ppm) to the sugar headgroup These atoms are more than 15 Aring apart and a direct interaction 659

could not be observed in both NOESY and DQ-SQ experiments NOESY experiments (Figure 660

9a) show that H1rsquo-H1 correlations are not observed Meanwhile clear-cut direct H1rsquo-H1 661

interactions are indeed observed are detected at T= 70degC (DQ cross-peaks related to H1rsquo-H1 662

correlation in DQ-SQ experiment shown in Figure 9b) Interestingly the absence of auto peaks 663

at δH= 432 ppm and δH= 335 ppm in the SQ dimension excludes H1rsquo-H1rsquo and Hxrsquo-Hxrsquo (x= 2 664

3 4) interactions respectively strengthening the idea that sugar headgroups from two nearest-665

neighbor G-C181-OH molecules are not adjacent in the nanotube-vesicle region 666

Homonuclear 2D ssNMR supports the idea according to which G-C181-OH can 667

undergo molecular flip-flops and a head-tail arrangement in the dry product Extrapolating these 668

conclusions in the lower concentration range lack of H1rsquo-H1 proximity at T= 90degC suggests a 669

head-head configuration in the vesicle-rich region and head-tail configuration within the 670

nanotubes It is however still unclear whether flip-flops are driven by spontaneous temperature-671

induced conformational changes in the glucolipid packing stimulating inclusions within the 672

membrane itself672223629ndash1218ndash21 or by defects generated by inter-vesicle collisions In fact a 673

combination of both is not unlikely Nanotubes are not stable when temperature is 20 to 30 674

degrees above Tm1 indicating that inter-vesicle collisions alone cannot explain the unusual 675

phenomenon In an energy landscape approach87 the head-head configuration is certainly 676

favored at high temperature while both head-head and head-tail configurations could be 677

favored in the mid-temperature region corresponding to the range between Tm1 and Tm2 in the 678

dry powder Both statistical and defect-induced flip-flop events contribute to form high mean 679

zero Gaussian curvature morphologies (nanotubes) for which the head-tail conformation 680

limiting the proximity between bulky sugar headgroups is probably more adapted (Figure 8c) 681

The flip-flop interpretation now reasonably explains the strong differences in the sugar and fatty 682

alcohol conformations deduced by the multiple NMR signals attributed to chemically-683

equivalent groups The existence of various low-energy conformations each having specific 684

NMR signature for sugars in solution78ndash8086 and solid-state81ndash84 further supports our 685

interpretation 686

One last point should be clarified Why between Tm1 and Tm2 are two molecular 687

environments able to coexist instead of a single one of average rigidity as found in the case of 688

ldquoripplerdquo phases50ndash54 We believe that this question can be answered based on the NMR results 689

the relatively short T2 values of the H1rsquo and OH protons shows the constrained mobility of the 690

28

sugar headgroup despite a certain flexibility of the chain (long T2 component) This trend may 691

suggest restrictions in translational mobility probably due to the sugar-OH hydrogen bonding 692

(Figure 8c) meaning that lipids are not able to diffuse within the membrane thus keeping two 693

different configurations stable 694

695

Conclusions 696

Topologically-connected nanotube-vesicle lipidic systems are commonly observed for 697

self-assembled membranes composed of at least two amphiphiles and spontaneously driven by 698

anisotropic inclusions andor by external directional forces This work shows that strong 699

differences in membrane curvature can spontaneously occur in single-lipid systems without 700

external anisotropic forces Coexistence and interconnection between nanotubes and vesicles 701

are evidenced via cryo-TEM and fluorescence confocal microscopy in a single-phase region at 702

temperature above a transition temperature Tm1= 483degC determined by DSC on the hydrated 703

sample but below the second transition temperature Tm2 (957degC) determined by DSC on the 704

dry powder X-ray diffraction (WAXS) also shows the coexistence of two lattice periods one 705

attributed to an inter-membrane repeating distance and the other possibly related to the intra-706

nanotube glucolipid arrangement Labelling the systems with Laurdan a fluorophore of which 707

the emission strongly varies with the local membrane rigidity excludes the presence of rigid 708

micro-separated phases (eg Lβ) 709

1H and 13C ssNMR despite the fact that it can only be performed on the dry powder 710

confirm the presence of one main fluid G-C181-OH environment above Tm2 Below Tm2 the 711

splitting of some 13C peaks and the coexistence of short and long 1H relaxation components 712

demonstrate that G-C181-OH can simultaneously pack into structurally and dynamically 713

inequivalent environments Combining advanced 2D 1H-1H correlation experiments based on 714

spin diffusion or double-quanta excitation suggests that vesicles could contain a preferred head-715

head configuration of the bolaform lipid while nanotubes a preferred head-tail assembly 716

The presence of two environments composed of the same molecule characterized by 717

different molecular configurations and rigidity fulfills the theoretical conditions necessary to 718

spontaneously nucleate a membrane inclusion of non-zero average orientation described as 719

necessary to observe topologically-connected nanotube-vesicle structures in the absence of 720

external forces This hypothesis could also be enriched by the possibility of nanotubes to be 721

pulled away between two vesicles after a collision event 722

This work expands the actual theoretical and experimental framework on lipid networks 723

stabilized by a different radius of curvature It could also motivate the synthesis or isolation 724

29

of new lipids providing a higher level of 3D architectural control in soft networks possibly in 725

the range of room temperature 726

727

Acknowledgements 728

Dr Andrea Lassenberger and Dr Sylvain Preacutevost at Institut Laue Langevin (ILL Grenoble 729

France) are kindly acknowledged for their assistance on the SANS experiment Confocal 730

images were performed at the Institut de Biologie Paris Seine (IBPS) imaging facility Jean-731

Franccedilois Gilles (Imaging Core Facility Institut de Biologie Paris Seine (IBPS) CNRS 732

Sorbonne Universiteacute Paris France) is kindly acknowledged for his assistance on the confocal 733

microscopy imaging Dr Guillaume Laurent (LCMCP Sorbonne Universiteacute Paris France) is 734

kindly acknowledged for helpful discussions The French region Ile-de-France SESAME 735

program is acknowledged for financial support (700 MHz NMR 736

spectrometer) Dr Olivier Diat (ICSM Marcoule France) Prof eronikaIglič and Prof V Aleš737

onsul discussi Slovenia) are kindly acknowledged for helpfUniversity of LjubljanaIglič (-Kralj 738

739

Financial support 740

WAXS experiments have been supported by Soleil Light Source Saint Aubin France proposal 741

Ndeg 20201747 SANS experiments have been supported by ILL proposal Ndeg9-13-778 742

743

References 744

1 D Kabaso N Bobrovska W Goacuteźdź E Gongadze V Kralj-Iglič R Zorec and A 745

Iglič Bioelectrochemistry 2012 87 204ndash210 746

2 C R Safinya U Raviv D J Needleman A Zidovska M C Choi M A Ojeda-747

Lopez K K Ewert Y Li H P Miller J Quispe B Carragher C S Potter M W 748

Kim S C Feinstein and L Wilson Adv Mater 2011 23 2260ndash2270 749

3 P Veranič M Lokar G J Schuumltz J Weghuber S Wieser H Haumlgerstrand V Kralj-750

Iglič and A Iglič Biophys J 2008 95 4416ndash4425 751

4 A Rustom R Saffrich I Markovic P Walther and H-H Gerdes Science (80- ) 752

2004 303 1007ndash1010 753

5 M I Angelova S Soleacuteau P Meacuteleacuteard F Faucon and P Bothorel Prog Colloid 754

Polym Sci 1992 89 127ndash131 755

6 L Mathivet S Cribier and P F Devaux Biophys J 1996 70 1112ndash1121 756

7 V Kralj-Iglič G Gomišček J Majhenc V Arrigler and S Svetina Colloids Surfaces 757

A Physicochem Eng Asp 2001 181 315ndash318 758

30

8 V Kralj-Iglič A Iglič G Gomišček F Sevšek V Arrigler and H Haumlgerstrand J 759

Phys A Math Gen 2002 35 1533ndash1549 760

9 M Drab D Stopar V Kralj-Iglič and A Iglič Cells 2019 8 626 761

10 A Zidovska K K Ewert J Quispe B Carragher C S Potter and C R Safinya 762

Langmuir 2009 25 2979ndash2985 763

11 A Zidovska K K Ewert J Quispe B Carragher C S Potter and C R Safinya Soft 764

Matter 2011 7 8363ndash8369 765

12 A Zidovska K K Ewert J Quispe B Carragher C S Potter and C R Safinya 766

Biochim Biophys Acta - Biomembr 2009 1788 1869ndash1876 767

13 M Karlsson K Sott M Davidson A S Cans P Linderholm D Chiu and O Orwar 768

Proc Natl Acad Sci U S A 2002 99 11573ndash11578 769

14 A Karlsson R Karlsson M Karlsson A-S Cans A Stromberg F Ryttsen and O 770

Orwar Nature 2001 409 150ndash152 771

15 M Karlsson K Sott A S Cans A Karlsson R Karlsson and O Orwar Langmuir 772

2001 17 6754ndash6758 773

16 E Evans H Bowman A Leung D Needham and D Tirrell Science (80- ) 1996 774

273 933ndash935 775

17 U Seifert Configurations of fluid membranes and vesicles 1997 vol 46 776

18 J Kaumls and E Sackmann Biophys J 1991 60 825ndash844 777

19 N Bobrovska W Goacuteźdź V Kralj-Iglič and A Iglič PLoS One 2013 8 e73941 778

20 A Iglič B Babnik K Bohinc M Fošnarič H Haumlgerstrand and V Kralj-Iglič J 779

Biomech 2007 40 579ndash585 780

21 K Bohinc T Slivnik A Iglič M Brumen and V Kralj-Iglič J Phys Chem C 2007 781

111 9709ndash9718 782

22 A Iglič H Haumlgerstrand P Veranič A Plemenitaš and V Kralj-Iglič J Theor Biol 783

2006 240 368ndash373 784

23 V Kralj-Iglič and P Veranič Adv Planar Lipid Bilayers Liposomes 2006 5 129ndash785

149 786

24 A Iglic and V Kralj-Iglic Adv Planar Lipid Bilayers Liposomes 2006 4 253ndash279 787

25 S Penič L Mesarec M Fošnarič L Mroacutewczyńska H Haumlgerstrand V Kralj-Iglič and 788

A Iglič Front Phys 2020 8 342 789

26 P Yager and P E Schoen Mol Cryst Liq Cryst 1984 106 371ndash381 790

27 T G Barclay K Constantopoulos and J Matisons Chem Rev 2014 114 10217ndash791

10291 792

31

28 B Božič G Gomišček V Kralj-Iglič S Svetina and B Žekš Eur Biophys J 2002 793

31 487ndash496 794

29 A Iglič H Haumlgerstrand M Bobrowska-Haumlgerstrand V Arrigler and V Kralj-Iglič 795

Phys Lett A 2003 310 493ndash497 796

30 R Bar-Ziv and E Moses Phys Rev Lett 1994 73 1392ndash1395 797

31 S Chiruvolu H E Warriner E Naranjo S H Idziak J O Raumldler R J Plano J A 798

Zasadzinski and C R Safinya Science (80- ) 1994 266 1222ndash1225 799

32 V Kralj-Iglič A Iglič H Haumlgerstrand and P Peterlin Phys Rev E 2000 61 4230ndash800

4234 801

33 L Mesarec M Drab S Penič V Kralj-Iglič and A Iglič Int J Mol Sci 2021 22 802

2348 803

34 L Van Renterghem S L K W Roelants N Baccile K Uyttersprot M C Taelman 804

B Everaert S Mincke S Ledegen S Debrouwer K Scholtens C V Stevens W 805

Soetaert L Van Renterghem S L K W Roelants N Baccile K Uytersprot M C 806

Taelman B Everaert S Mincke S Ledegen S Debrouwer K Scholtens C V 807

Stevens and W Soetaert Biotechnol Bioeng 2018 115 1195ndash1206 808

35 J Clark and F Deswarte Introduction to Chemicals from Biomass Second Edition 809

John Wiley amp Sons 2015 810

36 K V Sajna R Houmlfer R K Sukumaran L D Gottumukkala and A Pandey in 811

Industrial Biorefineries and White Biotechnology eds A Pandey R Houmlfer M 812

Taherzadeh K M Nampoothiri and C Larroche Amsterdam Oxford Waltham 813

Elsevier 2015 pp 499ndash521 814

37 R Marchant and I M Banat Trends Biotechnol 2012 30 558ndash565 815

38 K Hill and C LeHen-Ferrenbach in Sugar-Based Surfactants Fundamentals and 816

Applications ed C C Ruiz Boca Raton CRC Press 2009 pp 1ndash20 817

39 L Rodrigues I M Banat J Teixeira and R Oliveira J Antimicrob Chemother 818

2006 57 609ndash618 819

40 J D Desai and I M Banat Microbiol Mol Biol Rev 1997 61 47ndash64 820

41 S L K W Roelants K Ciesielska S L De Maeseneire H Moens B Everaert S 821

Verweire Q Denon B Vanlerberghe I N A Van Bogaert P Van der Meeren B 822

Devreese and W Soetaert Biotechnol Bioeng 2016 113 550ndash559 823

42 N Baccile M Selmane P Le Griel S Preacutevost J Perez C V Stevens E Delbeke S 824

Zibek M Guenther W Soetaert I N A Van Bogaert and S Roelants Langmuir 825

2016 32 6343ndash6359 826

32

43 N Baccile A-S Cuvier S Preacutevost C V Stevens E Delbeke J Berton W Soetaert 827

I N A Van Bogaert and S Roelants Langmuir 2016 32 10881ndash10894 828

44 N Baccile C Seyrig A Poirier S A Castro S L K W Roelants and S Abel Green 829

Chem 2021 23 3842ndash3944 830

45 A Salis L Cappai C Carucci D F Parsons and M Monduzzi J Phys Chem Lett 831

2020 11 6805ndash6811 832

46 J Schindelin I Arganda-Carreras E Frise V Kaynig M Longair T Pietzsch S 833

Preibisch C Rueden S Saalfeld B Schmid J-Y Tinevez D J White V 834

Hartenstein K Eliceiri P Tomancak and A Cardona Nat Methods 2012 9 676ndash835

682 836

47 T Parasassi and E Gratton J Fluoresc 1995 5 59ndash69 837

48 M M Dodes Traian F L Gonzaacutelez Flecha and V Levi J Lipid Res 2012 53 609ndash838

616 839

49 D M Owen C Rentero A Magenau A Abu-Siniyeh and K Gaus Nat Protoc 840

2012 7 24ndash35 841

50 W Stillwell in An Introduction to Biological Membranes Academic Press 2016 pp 842

135ndash180 843

51 W Stillwell in An Introduction to Biological Membranes Academic Press 2016 pp 844

181ndash220 845

52 B Tenchov Chem Phys Lipids 1991 57 165ndash177 846

53 R Koynova and M Caffrey Biochim Biophys Acta 1998 1376 91ndash145 847

54 F Schmid S Dolezel O Lenz and S Meinhardt J Phys Conf Ser 2014 487 848

012004 849

55 S Meinhardt R L C Vink and F Schmid Proc Natl Acad Sci U S A 2013 110 850

4476ndash4481 851

56 O Glatter and O Kratky Small Angle X-ray Scattering Academic Press London 852

1982 853

57 B A Cunningham A D Brown D H Wolfe W P Williams and A Brain Phys 854

Rev E 1998 58 3662ndash3672 855

58 M Nadler A Steiner T Dvir O Szekely P Szekely A Ginsburg R Asor R Resh 856

C Tamburu M Peres and U Raviv Soft Matter 2011 7 1512ndash1523 857

59 C Tanford The hydrophobic effect Formation of micelles and biological membranes 858

John Wiley amp Sons Inc New York 1973 859

60 M Masuda and T Shimizu Langmuir 2004 20 5969ndash5977 860

33

61 A Zidovska K K Ewert J Quispe B Carragher C S Potter and C R Safinya 861

Block Liposomes Vesicles of Charged Lipids with Distinctly Shaped Nanoscale 862

Sphere- Pear- Tube- or Rod-Segments Elsevier Inc 1st edn 2009 vol 465 863

62 A Iglic V Kralj-Iglic B Božič M Bobrowska-Haumlgerstrand B Isomaa and H 864

Haumlgerstrand Bioelectrochemistry 2000 52 203ndash211 865

63 T Shimizu M Masuda and H Minamikawa Chem Rev 2005 105 1401ndash1444 866

64 J V Selinger and J M Schnur Phys Rev Lett 1993 71 4091ndash4094 867

65 J M Schnur Science (80- ) 1993 262 1669ndash1676 868

66 B Tenchov R Koynova and G Rapp Biophys J 2001 80 1873ndash1890 869

67 R Koynova and B Tenchov Ser Biomech 2017 31 3ndash19 870

68 S J Angyal and R Le Fur Carbohydr Res 1984 126 15ndash26 871

69 J K Whitesell and M A Minton J Am Chem Soc 1987 109 225ndash228 872

70 T Takayama S Ando and I Ando J Mol Struct 1990 220 245ndash250 873

71 K Gawrisch N V Eldho and I V Polozov Chem Phys Lipids 2002 116 135ndash151 874

72 F Tielens C Gervais J F Lambert F Mauri and D Costa Chem Mater 2008 20 875

3336ndash3344 876

73 G A Jeffrey and Y Yeon Acta Crystallogr Sect B 1986 42 410ndash413 877

74 B Berglund and R W Vaughan J Chem Phys 1980 73 2037ndash2043 878

75 C Gervais C Coelho T Azaiumls J Maquet G Laurent F Pourpoint C Bonhomme 879

P Florian B Alonso G Guerrero P H Mutin and F Mauri J Magn Reson 2007 880

187 131ndash140 881

76 C Akita T Kawaguchi F Kaneko H Yamamoto and M Suzuki J Phys Chem B 882

2004 108 4862ndash4868 883

77 R N Purusottam L Seacutenicourt J J Lacapegravere and P Tekely Biochim Biophys Acta - 884

Biomembr 2015 1848 3134ndash3139 885

78 X Biarneacutes A Ardegravevol A Planas C Rovira A Laio and M Parrinello J Am Chem 886

Soc 2007 129 10686ndash10693 887

79 L M J Kroon‐Batenburg and J Kroon Biopolymers 1990 29 1243ndash1248 888

80 K N Kirschner and R J Woods Proc Natl Acad Sci U S A 2001 98 10541ndash889

10545 890

81 S Svenson B Kirste and J H Fuhrhop J Am Chem Soc 1994 116 11969ndash11975 891

82 I Sack S Macholl J H Fuhrhop and G Buntkowsky Phys Chem Chem Phys 892

2000 2 1781ndash1788 893

83 S Svenson J Koening and J H Fuhrhop J Phys Chem 1994 98 1022ndash1028 894

34

84 S Svenson A Schaefer and J H Fuhrhop J Chem Soc Perkin Trans 2 1994 2 895

1023ndash1028 896

85 K Bock and C Pedersen Adv Carbohydr Chem Biochem 1983 41 27ndash66 897

86 T B Grindley Structure and Conformation of Carbohydrates Springer Berlin 898

Heidelberg 2008 899

87 F Tantakitti J Boekhoven X Wang R V Kazantsev T Yu J Li E Zhuang R 900

Zandi J H Ortony C J Newcomb L C Palmer G S Shekhawat M O De La 901

Cruz G C Schatz and S I Stupp Nat Mater 2016 15 469ndash476 902

903

1

Supporting Information 904

905

Topological connection between vesicles and nanotubes in single-906

component lipid membranes driven by head-tail interactions 907

908

Niki Baccilea Ceacutedric Lorthioira Abdoul Aziz Baa Patrick Le Griela Cristina Coelhob Javier 909

Perezc Wim Soetaertd Sophie L K W Roelantsd 910

911

a Sorbonne Universiteacute Centre National de la Recherche Scientifique Laboratoire de Chimie de 912

la Matiegravere Condenseacutee de Paris LCMCP F-75005 Paris France 913

b Institut des Mateacuteriaux de Paris Centre Sorbonne Universiteacute Paris France 914

c Synchrotron Soleil LrsquoOrme des Merisiers Saint-Aubin BP48 91192 Gif-sur-Yvette Cedex 915

France 916

d InBio Department of Biotechnology Ghent University Ghent Belgium 917

918

919

Corresponding author 920

Dr Niki Baccile 921

E-mail address nikibaccilesorbonne-universitefr 922

Phone +33 1 44 27 56 77 923

924

925

926

2

a) 927

Parameter Method

Dry Matter (DM ) 996 Infrared balance 105degC

Glucose () 012 HPLC-Metacarb

Glycerol () nd HPLC-Metacarb

Free Fatty Acid content () lt001 Internal method BBEPP

GC with FID Detector

Oil () lt001 Internal method BBEPP

GC with FID Detector

Moulds (CFUg) lt10 3M-Nordval ndeg16 B

Yeasts (CFUg) lt10 3M-Nordval ndeg16 B

Anaerobic count (CFUg) lt10 SP-VG M005 B

Protein () tbd BCA protein assay

928

Figure S 1 - a) HPLC-ELSD chromatogram and composition table 929

930

931

932

933

3

b) 934

935

Group Ndeg δ ppm

CH2 1 354

CH2 2 153

CH2 3-712-14 132

CH2 811 203

CH 910 534

CH2 15 162

CH2 16 388

CH 1 424

CH 2 316

CH 3 334

CH 4 329

CH 5 354

CH2 6 366

Figure S 1 (next) ndash b) 1H NMR spectrum (MeOD-d4) recorded on the purified G-C181-OH compounds 936

and related assignment 937

938

2

811910 1rsquo

5rsquo1

3rsquo4rsquo

1rsquo1910

asymp asymp asympHDO MeOD Acetone

6rsquo2rsquo

1615

3-712-14

4

939

940

Figure S 2 ndash Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 130degC 941

5

942

943

Figure S 3 - Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 90degC 944

945

90degC

1

22

6

946

947

948

949

Figure S 4 ndash Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 70degC 950

951

21

11

1

2

5

3

4

45

5

6

7

7

7

952

953

954

Figure S 5 ndash Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 25degC 955

956

957

1

1 1

8

958

Figure S 6 ndash a) SANS experiment recorded at 90degC for a 5 mgmL G-C181-OH aqueous solution b-c) 959

WAXS experiments recorded on aqueous solutions of G-C181-OH at b) 5 mgmL and c) 25 mgmL d) 960

Structural interpretation of the repeating distances pointed out in c) 961

962

963

964

001 01 1

q Aring-1

25degC

68degC

88degC

111degC

01 1q Aring

-1

02 04 06 08 1q Aring

-1

001 01

10-1

100

101

Inte

nsity

cm

-1

q Aring-1

-2

T= 90degC

SANS WAXS WAXS

a) b) c)

dv

dL d= 144 Aring-1

di

ddH2O

dt

dvdL

dtdi d

d)

9

965

966

Figure S 7 ndash ssNMR MAS (10 kHz) experiments recorded on a G-C181-OH dry powder a) 1H spectra 967

recorded from 110degC to 70degC b) 13C single pulse (SP) spectra recorded with high power 1H decoupling from 968

110degC to 70degC c) 13C cross polarization (CP) spectra recorded at 110degC and contact time tc of 50 micros and 969

3000 micros d) 13C CP spectra recorded at 70degC and for 20 lt tc micros lt 3000 970

971

1

811

910

CH2 CH2

1rsquo

6rsquo

8 6 4 2 0 ppm

110

90

80

70

910

1rsquo

6rsquo16

1

2rsquo3rsquo4rsquo

811

OH

45 40 35 30 25 20 15

ppm

200 180 160 140 120 100 80 60 40 20 0

ppm140 120 100 80

ppm

CH2-gauche

3004

110

90

80

70

65

60

50

25

C91013003

C9 - 12869

CH2-all-trans

3198

C1rsquo10335

a) b)

1H

T degCT degC

13C Single Pulse MAS

A

B

A

B

70degC70degC

5rsquo

215

200 180 160 140 120 100 80 60 40 20 0

ppm

13C CPMAST= 110degC

200 180 160 140 120 100 80 60 40 20 0

ppm

132 131 130 129 128 127 126

ppm

106 105 104 103 102

ppm

13C CPMAST= 70degC

50

tc micros

3000

20

tc micros

50 500 1000 3000

tc micros

C102

13024 C910113003

C92

12869

C1rsquo213082

C1rsquo110443

A B

106 105 104 103 102

ppm

3000

1000

500

50

20

AB

c)

d)

10

972

Figure S 8 ndash a) Short and d) long T2 components displayed by the 1H transverse relaxation signal c) 1H 973

transverse relaxation signals recorded on a G-C181-OH dry powder from 110degC to 70degC The area of each 974

given resonance is plotted (log-lin) as a function of the echo time (CPMG pulse sequence) for each 975

temperature and fitted using the equation given above T2s and T2l indicate the short and long T2 relaxation 976

times respectively obtained by describing the 1H transverse relaxation signal using a biexponential decay 977

For H1rsquo H5rsquo and H6rsquo the 1H relaxation function could be described by a monoexponential decay 978

- Protons from the sugar group (H1rsquo H5rsquo H6rsquo) show a single T2 relaxation component (Figure 979

6c and Figure S 8) while protons of the acyl chain (CH2 H910 H811) show both short (Figure 980

6c) and long (Figure 6d) T2 relaxation component 981

- The short T2 relaxation time systematically decreases from 110degC to 70degC for all species by a 982

factor ranging from two to seven depending on the specific chemical group Very long T2 are 983

observed at 110degC for all acyl species while nearly unchanged values are measured below 984

90degC without any clear trend 985

- The sugar protons (H1rsquo H5rsquo H6rsquo) have the shortest set of T2 values (lt 1 ms) below 100degC 986

The long component of T2 generally ranges between 4 ms and 12 ms 987

0 1 2 3 4 5 6 7 8

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1 2 3 4 5 6 7 8

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

H1

H8H11

H9H10

CH2

H1rsquo

H6rsquo

CH2

H910H811 H1CH2

H1rsquo H5rsquo6rsquo

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

OH

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45 T= 110degC

T= 90degC

T= 80degC

T= 70degC

Log

In

ten

sity

Time ms

Fit

)

CH2 H910 H811 H1 H56 H1 OH

00

05

10

15

20

25

30

35

40

T2 m

s

T= 110degC

T= 90degC

T= 80degC

T= 70degC

CH2 H910 H811 H1

0

2

4

6

8

10

12

T2 m

s

T= 110degC

T= 90degC

T= 80degC

T= 70degC

Short T2(1H) Long T2(1H)

a) b)

c)

11

Table S 1 ndash 1H chemical shifts of G-C181-OH in solution and in the solid-state Solid-state spectra are 988

recorded at a MAS frequency of 10 kHz in DMSO-d6 broad peak 989

990

δH ppm

Group Ndeg Liquid

(MeOD-d4) Solid

(T= 110degC) Solid

(T= 70degC)

CH2 H1 354 357

CH2 H2 153 156

CH2 H3-712-14 132 135

CH2 H811 203 203

CH H910 534 534

CH2 H15 162 164

CH2 H16 388 383

CH H1 424 432

CH H2 316 335

CH H3 334 335

CH H4 329 335

CH H5 354 351 357

CH2 H6 366 383

OH OH1 432

45 463

OH OH2 492

OH OH3 490

OH OH4 487

OH OH6 445

991

992

H1H9H10H1rsquo

H6rsquo

OH2rsquo

OH1

12

Detailed discussion about the NOESY and DQ-SQ 2D 1H-1H ssNMR experiments presented in 993

Figure 9 994

995

2D 1H-1H ssNMR experiments are a unique tool to explore through-space proximities 996

between 1H nuclei ssNMR offers a large number of possibilities each adapted to a given 997

system The advantage of 2D maps with respect to a dynamics approach is their direct 998

readability in clear two nuclei which exchange their magnetization through-space (whichever 999

mechanism may be responsible for this exchange) will display a specific cross-peak However 1000

the way cross-peaks are presented and interpreted strictly depends on the nature of the pulse 1001

program which in turn is chosen according to the physical nature of the sample 1002

1D temperature-dependent 1H NMR experiments performed on G-C181-OH (Figure 1003

4a in the main text) indicate sharp peaks at 90degC and broad peaks at 70degC Narrow peaks 1004

indicate fast andor large-amplitude molecular tumbling averaging the strong 1H-1H dipolar 1005

interactions This excludes the use of pulse sequences like DQ-SQ which is not efficient under 1006

these conditions For this reason we employ spin-diffusion-based pulse sequences like 1007

NOESY which is more adapted for samples with high molecular mobility When mixing times 1008

are chosen to be short (lt 1 s) one generally explores through-space proximities within about 5 1009

10 Aring On the contrary when molecular motions are restricted dipolar coupling becomes 1010

significant 1H NMR peaks are broader and pulse sequences like DQ-SQ can now be used 1011

with a spatial resolution of less than 5 Aring 1012

Whenever possible DQ-SQ experiments are preferred over spin diffusion-based pulse 1013

sequences because they are more selective in particular for likewise interactions which are 1014

not discriminated in NOESY-type experiments This is illustrated by the typical interaction 1015

schemes for NOESY (top Figure 9a) and DQ-SQ (top Figure 9b) For two given A and B 1016

protons 2D NOESY maps will display two on-diagonal and two square-shaped symmetrical 1017

off-diagonal cross-peaks Off-diagonal cross-peaks illustrate the mutual A-B and B-A 1018

interaction but the on-diagonal peaks are systematic and do not prove any A-A or B-B 1019

interaction Similarly to NOESY 2D DQ-SQ maps display off-diagonal peaks (symmetric 1020

across the diagonal top Figure 9b) upon A-B and B-A interaction but in contrast they will 1021

display on-diagonal peaks only in the presence of self-interaction In the tutorial example of 1022

Figure 9b top the on-diagonal A-A peak indicates mutual A-A interaction while lack of on-1023

diagonal B-B peak evidences the absence of B-B proximities Combining NOESY and DQ-SQ 1024

experiments is capital to study the local arrangement of G-C181-OH and its evolution with 1025

temperature 1026

13

At T= 90degC the H1 and H5rsquo contributions are very close but possible to discriminate 1027

them (δ(H1)= 354 ppm δ(H5rsquo)= 351 ppm) according to 1H NMR chemical shifts in Table S 1028

1 2D 1H-1H NOESY in Figure 9a also discriminates between the two components attributed 1029

on the basis of the interactions with their mutual neighbors (green H1-H2 and H5rsquo-H1rsquo squares 1030

in Figure 9a) On this basis 2D NOESY does not show a direct H1rsquo-H1 correlation (blue red-1031

crossed) but rather an extensive network of intra-sugar interactions These features exclude 1032

suggest intra- and inter-sugar interactions and they exclude head-tail arrangements as a major 1033

molecular configuration at 90degC 1034

At 70degC the 1H NMR peaks are broader indicating a reduction of the T2 (1H) relaxation 1035

times andor residual homonuclear dipolar interactions excluding the use of NOESY In this 1036

case 1H-1H DQ-SQ correlation experiments are best suited Figure 9b presents the 1H-1H DQ-1037

SQ experiment on G-C181-OH at 70degC Peaks are broader than in NOESY and attribution may 1038

be cumbersome However the experiment is performed at 700 MHz and MAS= 22 kHz which 1039

improves spectral resolution with respect to the same experiment performed at 300 MHz and 1040

MAS= 10 kHz The auto-peak at δH= 356 ppm in the SQ dimension is attributed to H1-H1 spin 1041

pairs mainly resulting from intra CH2 interaction while the DQ cross-peaks at δH= 156-356 1042

ppm and δH= 432-356 ppm respectively origin from H2-H1 and H1rsquo-H1 interactions (Figure 1043

9b) DQ-SQ map does not show any auto-peak corresponding to H1rsquo-H1rsquo or Hxrsquo-Hxrsquo (x= 2rsquo 1044

3rsquo 4rsquo) interactions (red-crossed blue Figure 9b) thus excluding any proximity (lt 5 Aring) between 1045

close glucose neighbors On the contrary DQ-SQ maps (recorded both at 700 MHz Figure 9b 1046

or at 300 MHz not shown) systematically show DQ cross-peaks which can only be attributed 1047

to H1rsquo-H1 interactions (δH= 432-356) The presence of these H1rsquo-H1 interactions can only be 1048

explained by a head-tail arrangement of the G-C181-OH molecules as illustrated below the 1049

DQ-SQ map in Figure 9b 1050

22

required conditions that induce local lipid segregation and formation of membrane inclusions 559

(Figure 8c) In case of their non-zero average orientation the local membrane curvature 560

increases62 and nanotubes can spontaneously form (Figure 8bc) More intriguing this 561

mechanism occurs at least 10degC above Tm1 although it should not be favored above the Tm27 562

From both experimental and theoretical backgrounds the conditions of topologically-563

connected vesicles-nanotubes systems are met only when either internal (two components) or 564

external driving forces are applied19 In the present single-lipid system of which the phase 565

behavior is summarized in Figure 8a the most plausible driving force could be explained by 566

the NMR results collected on the dry sample G-C181-OH concomitantly occupies two 567

molecular environments characterized by 1) two distinct conformations of the sugar 568

displaying equivalent mobility and 2) two conformations of the acyl chain unsaturation 569

570

23

571

Figure 8 ndash a) Thermotropic phase behavior of G-C181-OH at 5 mgmL-1 Note that Tm2 is observed on the 572

dry powder only b) Possible relationship between the molecular environment as derived from NMR and 573

membrane curvatures All the molecular drawings refer to the same G-C181-OH molecule (bulky ellipsoid 574

equiv sugar headgroup) while each color corresponds to a given configuration Grey of C1rsquo1 and C1rsquo2 indicates 575

the uncertainty about the assignment of the specific NMR peak The curvature C is defined as 1R R being 576

the curvature radius R1 R2 radius of curvature CG Gauss curvature CM Mean curvature c) Free 577

interpretation of the anisotropic high-curvature (CMgt 0 CG= 0) inclusions in the membrane (low CM and 578

CG) continuum driven by inter-sugar H-bonding and reduced intramembrane diffusion for a G-C181-OH 579

glucolipid 580

581

Flat crystalline structure

Nanotube-vesicle network

Vesicles

48degC

120deg

106 105 104 103 102 101

ppm

H9 H10H1rsquo

106 105 104 103 102 101

ppm106 105 104 103 102 101

ppm106 105 104 103 102 101

ppm

C=CC=C

C1rsquo1C1rsquo2

C1rsquo

Tm2(96degC)Tm1

a)

b)

c)

Same molecule different local orientations and sugar conformations

High curvature membrane

Low curvature inclusion

Flip-FlopStrong H-bonding

CG

CM ( + )

CG 0

CM 0 CG 0

CM 0

G-C181-OH

24

Two related questions are unanswered at the moment What actually drives nanotube 582

formation and is the organization of G-C181-OH within the vesicles and tubes the same Lipid 583

nanotubes have been studied for at least three decades over which broad experimental and 584

theoretical insight has been generated Despite this fact the actual understanding is still matter 585

of debate27 Micelle-to-nanotube and vesicle-to-nanotube phase transitions are generally 586

reported for single amphiphiles below a given Tm and intermediate phase composed of twisted 587

and flat fibers is concomitantly reported Initially thought to be a chirality-driven phenomenon 588

recent studies propose a more general mechanism involving crystallization and symmetry 589

breaking in the lipid packing driven by optimal packing27 If twisted ribbons were observed 590

before for a number of microbial glycolipid bolaamphiphiles44 they are not found for G-C181-591

OH This fact as well as the large body of cryo-TEM images and the observation of nanotubes 592

above Tm1 rather pledge in favour of a direct topological change from positive (vesicles) to 593

zero (nanotubes) Gaussian curvatures In this case vesicle-to-nanotube transitions could be 594

driven by external forces56307813ndash162829 595

Considering the absence of external forces in the present system there are only two 596

possibilities left spontaneous budding of the membrane or nanotube pulling after inter-vesicle 597

collisions due to Brownian motion Both are possible as inclusions composed of a different 598

arrangement of the same molecule in the membrane continuum can theoretically explain 599

spontaneous nanotube formation in an isotropic medium672223629ndash1218ndash21 However the second 600

one is probably more likely to prevail because it could explain the large number of nanotubes 601

connecting two distant vesicles Interestingly similar events occur for any phospholipid-based 602

vesicular solution but nanotubes are extremely rare unless at least two lipids of different 603

molecular structure are mixed together The spectroscopic 1H and 13C and relaxation (1H) 604

ssNMR data even if collected on a dry powder demonstrate that the coexistence of two G-605

C181-OH configurations involving both the sugar headgroup and fatty alcohol chain are 606

possible for this molecule As a matter of fact this is not so surprising as more than one energy 607

minima each corresponding to a given conformation are known for glucose its non-reducing 608

derivative methylglucose and carbohydrates in general78ndash80 609

The presence of two configurations could explain the coexistence between two 610

structures of different curvature but the data presented so far do not explain the origin of these 611

two configurations In other words what drives the change in the local structure of the sugar 612

headgroup and fatty alcohol within the membrane 613

The association of known self-assembled structures to spectroscopic 13C ssNMR data is 614

known for years in particular for gluconamides lipids having a linear sugar headgroup81ndash84 It 615

25

was found that gauche effects in some cases up to the γ position in the sugar chain induce 616

upfield chemical shifts while anti and trans conformations result in downfield shifts Each 617

structure (monolayer crystal bilayer crystal or micellar fiber) was then characterized by a given 618

organization of the gluconamide and a characteristic 13C ssNMR fingerprint which could be 619

explained by gauche γ-gauche anti and trans conformations of the headgroup Similar 620

structural-spectroscopic correlations have been collected over the years for much simpler cyclic 621

sugars85 whereas pyranoses can undergo important conformational changes (chair half-chair 622

boat skew) when the appropriate energy barriers are crossed78ndash8086 A quantitative 623

conformational study based on the 13C chemical shift variations shown in Table 1 would be 624

certainly helpful but very delicate in the context of the present state of the art because a 625

rigorous approach requires at least one known crystal structure8182 which is currently missing 626

for this compound A qualitative approach will then be adopted 627

G-C181-OH is a bolaform amphiphile and flip-flop could be a possible mechanism 628

involved in the nanotube stabilization whereas head-tail interactions have long been considered 629

as important in nanotube-forming bolaform amphiphiles6063 From mere considerations based 630

on T2 (1H) relaxation arguments one could exclude head-tail configurations if the R-CH2OH 631

endgroup (H1) was involved in a direct interaction with the sugar group its corresponding T2 632

would probably be comparable to the sugarrsquos one (OH and H1rsquo) but this does not seem to be 633

the case However this hypothesis would be based on the comparison of the relaxation from 634

CH and CH2 protons additionally involved in different kind of motions Besides the relaxation 635

of H1 is influenced by the strong overlap between H1 and H5rsquo For these reasons relaxation 636

arguments are not solid enough and will not be used 637

We then address this question by exploring intermolecular interactions in a direct 638

manner using 2D 1H-1H homonuclear ssNMR correlation spectroscopy At 90degC in the fast-639

tumbling domain of molecular motion (narrow 1H NMR peaks Figure 6a) and corresponding 640

to the vesicle-rich region spin diffusion (NOESY) experiments are preferred over dipolar 641

coupling based experiments because such interaction is averaged out by molecular motion 642

However when the molecular motion is more constrained in the domain of nanotube stability 643

at 70degC (broad 1H spectra Figure 6a) Double-Quantum Single-Quantum (DQ-SQ) correlation 644

experiments are best suited to probe interaction between close molecular neighbors These 645

experiments are reported in Figure 9 and the related interpretation is only briefly reported 646

below while a more detailed discussion can be found on Page 12 in the Supporting Information 647

26

648

Figure 9 ndash a) 2D 1H-1H NOESY NMR experiment performed at MAS= 10 kHz (4 mm probe) on a Bruker 649

300 MHz Avance III HD spectrometer using 300 ms of mixing time Sample temperature is regulated to 650

90degC reached by cooling from 110degC b) 2D 1H-1H DQ-SQ (BAck-to-BAck BABA pulse sequence used for 651

the DQ excitation scheme) NMR experiment performed at MAS= 22 kHz (25 mm probe) on a Bruker 700 652

MHz Avance III spectrometer using a DQ excitation time equal to 2 rotor periods Sample temperature is 653

regulated to 70degC reached by cooling from 110degC The typical schemes illustrating how to interpret these 654

2D experiments are reported above each experimental map Zooms of the sugar region and configurational 655

interpretation are given below each 2D map More information are given in the text below 656

H16-H16

H1-H1

H9-H10

H1rsquo-H1

H1rsquoH910 H16

H1

H1rsquo-H1rsquo

Hxrsquo-Hxrsquo

2

(pp

m)

(ppm)

315

811910 1rsquo

2rsquo2

3rsquo 4rsquo6rsquo

5rsquo

1rsquo2910

asymp asymp asympHDO MeOD Acetone

1

315

811 9101rsquo

2rsquo2

3rsquo4rsquo6rsquo

5rsquo

1rsquo2 9 10

asympasympasympHDOMeODAcetone

H1rsquo-H1

1rsquo

1

(p

pm

)

(ppm)

H1-H2

H1rsquo-H1

H1rsquo

H910

H1H5rsquo

H2

H1rsquo-H5rsquo

315

811910 1rsquo

2rsquo2

3rsquo 4rsquo6rsquo

5rsquo

1rsquo2910

asymp asymp asympHDO MeOD Acetone

1

H2rsquoH3rsquoH4rsquo

H2rsquoH3rsquoH4rsquo

X= 2rsquo3rsquo4rsquo

NOESY - T= 90degC

1H

1H

1H

1H

DQ-SQ - T= 70degC

a) b)

2

(pp

m)

(p

pm

)

(ppm)

A B

AhellipB

AhellipA

BhellipB

NOESY

(ppm)

A B

AhellipB

DQ-SQ

Single-quantum dimension

Do

ub

le-q

ua

ntu

m d

imen

sio

n

AhellipA

BhellipB

BhellipA

BhellipA

315

811910 1rsquo

2rsquo2

3rsquo 4rsquo6rsquo

5rsquo

1rsquo2910

asymp asymp asympHDO MeOD Acetone

1

lt 5 Aring

lt 5 Aring

H1-H2

H811

H15

H16

H6rsquo

H1rsquo

H1

H5rsquo

H2

H1-H2

H1rsquo-H1

H1rsquo-H5rsquo

H1rsquo

H1 H5rsquo

H1rsquo

H1

H5rsquo

H1H2rsquo H3rsquo H4rsquo

H1-H1

H1rsquo-H1

Hxrsquo-HxrsquoX= 2rsquo3rsquo4rsquo

H1-H2

H1rsquo

Zoom sugar region 28 ndash 45 ppmNOESY DQ-SQ

27

Head-tail or head-head configurations could be probed by the existence of H1rsquo-H1 657

interactions whereas H1 (δH= 356 ppm) is associated to the glucolipid tail and H1rsquo (δH= 432 658

ppm) to the sugar headgroup These atoms are more than 15 Aring apart and a direct interaction 659

could not be observed in both NOESY and DQ-SQ experiments NOESY experiments (Figure 660

9a) show that H1rsquo-H1 correlations are not observed Meanwhile clear-cut direct H1rsquo-H1 661

interactions are indeed observed are detected at T= 70degC (DQ cross-peaks related to H1rsquo-H1 662

correlation in DQ-SQ experiment shown in Figure 9b) Interestingly the absence of auto peaks 663

at δH= 432 ppm and δH= 335 ppm in the SQ dimension excludes H1rsquo-H1rsquo and Hxrsquo-Hxrsquo (x= 2 664

3 4) interactions respectively strengthening the idea that sugar headgroups from two nearest-665

neighbor G-C181-OH molecules are not adjacent in the nanotube-vesicle region 666

Homonuclear 2D ssNMR supports the idea according to which G-C181-OH can 667

undergo molecular flip-flops and a head-tail arrangement in the dry product Extrapolating these 668

conclusions in the lower concentration range lack of H1rsquo-H1 proximity at T= 90degC suggests a 669

head-head configuration in the vesicle-rich region and head-tail configuration within the 670

nanotubes It is however still unclear whether flip-flops are driven by spontaneous temperature-671

induced conformational changes in the glucolipid packing stimulating inclusions within the 672

membrane itself672223629ndash1218ndash21 or by defects generated by inter-vesicle collisions In fact a 673

combination of both is not unlikely Nanotubes are not stable when temperature is 20 to 30 674

degrees above Tm1 indicating that inter-vesicle collisions alone cannot explain the unusual 675

phenomenon In an energy landscape approach87 the head-head configuration is certainly 676

favored at high temperature while both head-head and head-tail configurations could be 677

favored in the mid-temperature region corresponding to the range between Tm1 and Tm2 in the 678

dry powder Both statistical and defect-induced flip-flop events contribute to form high mean 679

zero Gaussian curvature morphologies (nanotubes) for which the head-tail conformation 680

limiting the proximity between bulky sugar headgroups is probably more adapted (Figure 8c) 681

The flip-flop interpretation now reasonably explains the strong differences in the sugar and fatty 682

alcohol conformations deduced by the multiple NMR signals attributed to chemically-683

equivalent groups The existence of various low-energy conformations each having specific 684

NMR signature for sugars in solution78ndash8086 and solid-state81ndash84 further supports our 685

interpretation 686

One last point should be clarified Why between Tm1 and Tm2 are two molecular 687

environments able to coexist instead of a single one of average rigidity as found in the case of 688

ldquoripplerdquo phases50ndash54 We believe that this question can be answered based on the NMR results 689

the relatively short T2 values of the H1rsquo and OH protons shows the constrained mobility of the 690

28

sugar headgroup despite a certain flexibility of the chain (long T2 component) This trend may 691

suggest restrictions in translational mobility probably due to the sugar-OH hydrogen bonding 692

(Figure 8c) meaning that lipids are not able to diffuse within the membrane thus keeping two 693

different configurations stable 694

695

Conclusions 696

Topologically-connected nanotube-vesicle lipidic systems are commonly observed for 697

self-assembled membranes composed of at least two amphiphiles and spontaneously driven by 698

anisotropic inclusions andor by external directional forces This work shows that strong 699

differences in membrane curvature can spontaneously occur in single-lipid systems without 700

external anisotropic forces Coexistence and interconnection between nanotubes and vesicles 701

are evidenced via cryo-TEM and fluorescence confocal microscopy in a single-phase region at 702

temperature above a transition temperature Tm1= 483degC determined by DSC on the hydrated 703

sample but below the second transition temperature Tm2 (957degC) determined by DSC on the 704

dry powder X-ray diffraction (WAXS) also shows the coexistence of two lattice periods one 705

attributed to an inter-membrane repeating distance and the other possibly related to the intra-706

nanotube glucolipid arrangement Labelling the systems with Laurdan a fluorophore of which 707

the emission strongly varies with the local membrane rigidity excludes the presence of rigid 708

micro-separated phases (eg Lβ) 709

1H and 13C ssNMR despite the fact that it can only be performed on the dry powder 710

confirm the presence of one main fluid G-C181-OH environment above Tm2 Below Tm2 the 711

splitting of some 13C peaks and the coexistence of short and long 1H relaxation components 712

demonstrate that G-C181-OH can simultaneously pack into structurally and dynamically 713

inequivalent environments Combining advanced 2D 1H-1H correlation experiments based on 714

spin diffusion or double-quanta excitation suggests that vesicles could contain a preferred head-715

head configuration of the bolaform lipid while nanotubes a preferred head-tail assembly 716

The presence of two environments composed of the same molecule characterized by 717

different molecular configurations and rigidity fulfills the theoretical conditions necessary to 718

spontaneously nucleate a membrane inclusion of non-zero average orientation described as 719

necessary to observe topologically-connected nanotube-vesicle structures in the absence of 720

external forces This hypothesis could also be enriched by the possibility of nanotubes to be 721

pulled away between two vesicles after a collision event 722

This work expands the actual theoretical and experimental framework on lipid networks 723

stabilized by a different radius of curvature It could also motivate the synthesis or isolation 724

29

of new lipids providing a higher level of 3D architectural control in soft networks possibly in 725

the range of room temperature 726

727

Acknowledgements 728

Dr Andrea Lassenberger and Dr Sylvain Preacutevost at Institut Laue Langevin (ILL Grenoble 729

France) are kindly acknowledged for their assistance on the SANS experiment Confocal 730

images were performed at the Institut de Biologie Paris Seine (IBPS) imaging facility Jean-731

Franccedilois Gilles (Imaging Core Facility Institut de Biologie Paris Seine (IBPS) CNRS 732

Sorbonne Universiteacute Paris France) is kindly acknowledged for his assistance on the confocal 733

microscopy imaging Dr Guillaume Laurent (LCMCP Sorbonne Universiteacute Paris France) is 734

kindly acknowledged for helpful discussions The French region Ile-de-France SESAME 735

program is acknowledged for financial support (700 MHz NMR 736

spectrometer) Dr Olivier Diat (ICSM Marcoule France) Prof eronikaIglič and Prof V Aleš737

onsul discussi Slovenia) are kindly acknowledged for helpfUniversity of LjubljanaIglič (-Kralj 738

739

Financial support 740

WAXS experiments have been supported by Soleil Light Source Saint Aubin France proposal 741

Ndeg 20201747 SANS experiments have been supported by ILL proposal Ndeg9-13-778 742

743

References 744

1 D Kabaso N Bobrovska W Goacuteźdź E Gongadze V Kralj-Iglič R Zorec and A 745

Iglič Bioelectrochemistry 2012 87 204ndash210 746

2 C R Safinya U Raviv D J Needleman A Zidovska M C Choi M A Ojeda-747

Lopez K K Ewert Y Li H P Miller J Quispe B Carragher C S Potter M W 748

Kim S C Feinstein and L Wilson Adv Mater 2011 23 2260ndash2270 749

3 P Veranič M Lokar G J Schuumltz J Weghuber S Wieser H Haumlgerstrand V Kralj-750

Iglič and A Iglič Biophys J 2008 95 4416ndash4425 751

4 A Rustom R Saffrich I Markovic P Walther and H-H Gerdes Science (80- ) 752

2004 303 1007ndash1010 753

5 M I Angelova S Soleacuteau P Meacuteleacuteard F Faucon and P Bothorel Prog Colloid 754

Polym Sci 1992 89 127ndash131 755

6 L Mathivet S Cribier and P F Devaux Biophys J 1996 70 1112ndash1121 756

7 V Kralj-Iglič G Gomišček J Majhenc V Arrigler and S Svetina Colloids Surfaces 757

A Physicochem Eng Asp 2001 181 315ndash318 758

30

8 V Kralj-Iglič A Iglič G Gomišček F Sevšek V Arrigler and H Haumlgerstrand J 759

Phys A Math Gen 2002 35 1533ndash1549 760

9 M Drab D Stopar V Kralj-Iglič and A Iglič Cells 2019 8 626 761

10 A Zidovska K K Ewert J Quispe B Carragher C S Potter and C R Safinya 762

Langmuir 2009 25 2979ndash2985 763

11 A Zidovska K K Ewert J Quispe B Carragher C S Potter and C R Safinya Soft 764

Matter 2011 7 8363ndash8369 765

12 A Zidovska K K Ewert J Quispe B Carragher C S Potter and C R Safinya 766

Biochim Biophys Acta - Biomembr 2009 1788 1869ndash1876 767

13 M Karlsson K Sott M Davidson A S Cans P Linderholm D Chiu and O Orwar 768

Proc Natl Acad Sci U S A 2002 99 11573ndash11578 769

14 A Karlsson R Karlsson M Karlsson A-S Cans A Stromberg F Ryttsen and O 770

Orwar Nature 2001 409 150ndash152 771

15 M Karlsson K Sott A S Cans A Karlsson R Karlsson and O Orwar Langmuir 772

2001 17 6754ndash6758 773

16 E Evans H Bowman A Leung D Needham and D Tirrell Science (80- ) 1996 774

273 933ndash935 775

17 U Seifert Configurations of fluid membranes and vesicles 1997 vol 46 776

18 J Kaumls and E Sackmann Biophys J 1991 60 825ndash844 777

19 N Bobrovska W Goacuteźdź V Kralj-Iglič and A Iglič PLoS One 2013 8 e73941 778

20 A Iglič B Babnik K Bohinc M Fošnarič H Haumlgerstrand and V Kralj-Iglič J 779

Biomech 2007 40 579ndash585 780

21 K Bohinc T Slivnik A Iglič M Brumen and V Kralj-Iglič J Phys Chem C 2007 781

111 9709ndash9718 782

22 A Iglič H Haumlgerstrand P Veranič A Plemenitaš and V Kralj-Iglič J Theor Biol 783

2006 240 368ndash373 784

23 V Kralj-Iglič and P Veranič Adv Planar Lipid Bilayers Liposomes 2006 5 129ndash785

149 786

24 A Iglic and V Kralj-Iglic Adv Planar Lipid Bilayers Liposomes 2006 4 253ndash279 787

25 S Penič L Mesarec M Fošnarič L Mroacutewczyńska H Haumlgerstrand V Kralj-Iglič and 788

A Iglič Front Phys 2020 8 342 789

26 P Yager and P E Schoen Mol Cryst Liq Cryst 1984 106 371ndash381 790

27 T G Barclay K Constantopoulos and J Matisons Chem Rev 2014 114 10217ndash791

10291 792

31

28 B Božič G Gomišček V Kralj-Iglič S Svetina and B Žekš Eur Biophys J 2002 793

31 487ndash496 794

29 A Iglič H Haumlgerstrand M Bobrowska-Haumlgerstrand V Arrigler and V Kralj-Iglič 795

Phys Lett A 2003 310 493ndash497 796

30 R Bar-Ziv and E Moses Phys Rev Lett 1994 73 1392ndash1395 797

31 S Chiruvolu H E Warriner E Naranjo S H Idziak J O Raumldler R J Plano J A 798

Zasadzinski and C R Safinya Science (80- ) 1994 266 1222ndash1225 799

32 V Kralj-Iglič A Iglič H Haumlgerstrand and P Peterlin Phys Rev E 2000 61 4230ndash800

4234 801

33 L Mesarec M Drab S Penič V Kralj-Iglič and A Iglič Int J Mol Sci 2021 22 802

2348 803

34 L Van Renterghem S L K W Roelants N Baccile K Uyttersprot M C Taelman 804

B Everaert S Mincke S Ledegen S Debrouwer K Scholtens C V Stevens W 805

Soetaert L Van Renterghem S L K W Roelants N Baccile K Uytersprot M C 806

Taelman B Everaert S Mincke S Ledegen S Debrouwer K Scholtens C V 807

Stevens and W Soetaert Biotechnol Bioeng 2018 115 1195ndash1206 808

35 J Clark and F Deswarte Introduction to Chemicals from Biomass Second Edition 809

John Wiley amp Sons 2015 810

36 K V Sajna R Houmlfer R K Sukumaran L D Gottumukkala and A Pandey in 811

Industrial Biorefineries and White Biotechnology eds A Pandey R Houmlfer M 812

Taherzadeh K M Nampoothiri and C Larroche Amsterdam Oxford Waltham 813

Elsevier 2015 pp 499ndash521 814

37 R Marchant and I M Banat Trends Biotechnol 2012 30 558ndash565 815

38 K Hill and C LeHen-Ferrenbach in Sugar-Based Surfactants Fundamentals and 816

Applications ed C C Ruiz Boca Raton CRC Press 2009 pp 1ndash20 817

39 L Rodrigues I M Banat J Teixeira and R Oliveira J Antimicrob Chemother 818

2006 57 609ndash618 819

40 J D Desai and I M Banat Microbiol Mol Biol Rev 1997 61 47ndash64 820

41 S L K W Roelants K Ciesielska S L De Maeseneire H Moens B Everaert S 821

Verweire Q Denon B Vanlerberghe I N A Van Bogaert P Van der Meeren B 822

Devreese and W Soetaert Biotechnol Bioeng 2016 113 550ndash559 823

42 N Baccile M Selmane P Le Griel S Preacutevost J Perez C V Stevens E Delbeke S 824

Zibek M Guenther W Soetaert I N A Van Bogaert and S Roelants Langmuir 825

2016 32 6343ndash6359 826

32

43 N Baccile A-S Cuvier S Preacutevost C V Stevens E Delbeke J Berton W Soetaert 827

I N A Van Bogaert and S Roelants Langmuir 2016 32 10881ndash10894 828

44 N Baccile C Seyrig A Poirier S A Castro S L K W Roelants and S Abel Green 829

Chem 2021 23 3842ndash3944 830

45 A Salis L Cappai C Carucci D F Parsons and M Monduzzi J Phys Chem Lett 831

2020 11 6805ndash6811 832

46 J Schindelin I Arganda-Carreras E Frise V Kaynig M Longair T Pietzsch S 833

Preibisch C Rueden S Saalfeld B Schmid J-Y Tinevez D J White V 834

Hartenstein K Eliceiri P Tomancak and A Cardona Nat Methods 2012 9 676ndash835

682 836

47 T Parasassi and E Gratton J Fluoresc 1995 5 59ndash69 837

48 M M Dodes Traian F L Gonzaacutelez Flecha and V Levi J Lipid Res 2012 53 609ndash838

616 839

49 D M Owen C Rentero A Magenau A Abu-Siniyeh and K Gaus Nat Protoc 840

2012 7 24ndash35 841

50 W Stillwell in An Introduction to Biological Membranes Academic Press 2016 pp 842

135ndash180 843

51 W Stillwell in An Introduction to Biological Membranes Academic Press 2016 pp 844

181ndash220 845

52 B Tenchov Chem Phys Lipids 1991 57 165ndash177 846

53 R Koynova and M Caffrey Biochim Biophys Acta 1998 1376 91ndash145 847

54 F Schmid S Dolezel O Lenz and S Meinhardt J Phys Conf Ser 2014 487 848

012004 849

55 S Meinhardt R L C Vink and F Schmid Proc Natl Acad Sci U S A 2013 110 850

4476ndash4481 851

56 O Glatter and O Kratky Small Angle X-ray Scattering Academic Press London 852

1982 853

57 B A Cunningham A D Brown D H Wolfe W P Williams and A Brain Phys 854

Rev E 1998 58 3662ndash3672 855

58 M Nadler A Steiner T Dvir O Szekely P Szekely A Ginsburg R Asor R Resh 856

C Tamburu M Peres and U Raviv Soft Matter 2011 7 1512ndash1523 857

59 C Tanford The hydrophobic effect Formation of micelles and biological membranes 858

John Wiley amp Sons Inc New York 1973 859

60 M Masuda and T Shimizu Langmuir 2004 20 5969ndash5977 860

33

61 A Zidovska K K Ewert J Quispe B Carragher C S Potter and C R Safinya 861

Block Liposomes Vesicles of Charged Lipids with Distinctly Shaped Nanoscale 862

Sphere- Pear- Tube- or Rod-Segments Elsevier Inc 1st edn 2009 vol 465 863

62 A Iglic V Kralj-Iglic B Božič M Bobrowska-Haumlgerstrand B Isomaa and H 864

Haumlgerstrand Bioelectrochemistry 2000 52 203ndash211 865

63 T Shimizu M Masuda and H Minamikawa Chem Rev 2005 105 1401ndash1444 866

64 J V Selinger and J M Schnur Phys Rev Lett 1993 71 4091ndash4094 867

65 J M Schnur Science (80- ) 1993 262 1669ndash1676 868

66 B Tenchov R Koynova and G Rapp Biophys J 2001 80 1873ndash1890 869

67 R Koynova and B Tenchov Ser Biomech 2017 31 3ndash19 870

68 S J Angyal and R Le Fur Carbohydr Res 1984 126 15ndash26 871

69 J K Whitesell and M A Minton J Am Chem Soc 1987 109 225ndash228 872

70 T Takayama S Ando and I Ando J Mol Struct 1990 220 245ndash250 873

71 K Gawrisch N V Eldho and I V Polozov Chem Phys Lipids 2002 116 135ndash151 874

72 F Tielens C Gervais J F Lambert F Mauri and D Costa Chem Mater 2008 20 875

3336ndash3344 876

73 G A Jeffrey and Y Yeon Acta Crystallogr Sect B 1986 42 410ndash413 877

74 B Berglund and R W Vaughan J Chem Phys 1980 73 2037ndash2043 878

75 C Gervais C Coelho T Azaiumls J Maquet G Laurent F Pourpoint C Bonhomme 879

P Florian B Alonso G Guerrero P H Mutin and F Mauri J Magn Reson 2007 880

187 131ndash140 881

76 C Akita T Kawaguchi F Kaneko H Yamamoto and M Suzuki J Phys Chem B 882

2004 108 4862ndash4868 883

77 R N Purusottam L Seacutenicourt J J Lacapegravere and P Tekely Biochim Biophys Acta - 884

Biomembr 2015 1848 3134ndash3139 885

78 X Biarneacutes A Ardegravevol A Planas C Rovira A Laio and M Parrinello J Am Chem 886

Soc 2007 129 10686ndash10693 887

79 L M J Kroon‐Batenburg and J Kroon Biopolymers 1990 29 1243ndash1248 888

80 K N Kirschner and R J Woods Proc Natl Acad Sci U S A 2001 98 10541ndash889

10545 890

81 S Svenson B Kirste and J H Fuhrhop J Am Chem Soc 1994 116 11969ndash11975 891

82 I Sack S Macholl J H Fuhrhop and G Buntkowsky Phys Chem Chem Phys 892

2000 2 1781ndash1788 893

83 S Svenson J Koening and J H Fuhrhop J Phys Chem 1994 98 1022ndash1028 894

34

84 S Svenson A Schaefer and J H Fuhrhop J Chem Soc Perkin Trans 2 1994 2 895

1023ndash1028 896

85 K Bock and C Pedersen Adv Carbohydr Chem Biochem 1983 41 27ndash66 897

86 T B Grindley Structure and Conformation of Carbohydrates Springer Berlin 898

Heidelberg 2008 899

87 F Tantakitti J Boekhoven X Wang R V Kazantsev T Yu J Li E Zhuang R 900

Zandi J H Ortony C J Newcomb L C Palmer G S Shekhawat M O De La 901

Cruz G C Schatz and S I Stupp Nat Mater 2016 15 469ndash476 902

903

1

Supporting Information 904

905

Topological connection between vesicles and nanotubes in single-906

component lipid membranes driven by head-tail interactions 907

908

Niki Baccilea Ceacutedric Lorthioira Abdoul Aziz Baa Patrick Le Griela Cristina Coelhob Javier 909

Perezc Wim Soetaertd Sophie L K W Roelantsd 910

911

a Sorbonne Universiteacute Centre National de la Recherche Scientifique Laboratoire de Chimie de 912

la Matiegravere Condenseacutee de Paris LCMCP F-75005 Paris France 913

b Institut des Mateacuteriaux de Paris Centre Sorbonne Universiteacute Paris France 914

c Synchrotron Soleil LrsquoOrme des Merisiers Saint-Aubin BP48 91192 Gif-sur-Yvette Cedex 915

France 916

d InBio Department of Biotechnology Ghent University Ghent Belgium 917

918

919

Corresponding author 920

Dr Niki Baccile 921

E-mail address nikibaccilesorbonne-universitefr 922

Phone +33 1 44 27 56 77 923

924

925

926

2

a) 927

Parameter Method

Dry Matter (DM ) 996 Infrared balance 105degC

Glucose () 012 HPLC-Metacarb

Glycerol () nd HPLC-Metacarb

Free Fatty Acid content () lt001 Internal method BBEPP

GC with FID Detector

Oil () lt001 Internal method BBEPP

GC with FID Detector

Moulds (CFUg) lt10 3M-Nordval ndeg16 B

Yeasts (CFUg) lt10 3M-Nordval ndeg16 B

Anaerobic count (CFUg) lt10 SP-VG M005 B

Protein () tbd BCA protein assay

928

Figure S 1 - a) HPLC-ELSD chromatogram and composition table 929

930

931

932

933

3

b) 934

935

Group Ndeg δ ppm

CH2 1 354

CH2 2 153

CH2 3-712-14 132

CH2 811 203

CH 910 534

CH2 15 162

CH2 16 388

CH 1 424

CH 2 316

CH 3 334

CH 4 329

CH 5 354

CH2 6 366

Figure S 1 (next) ndash b) 1H NMR spectrum (MeOD-d4) recorded on the purified G-C181-OH compounds 936

and related assignment 937

938

2

811910 1rsquo

5rsquo1

3rsquo4rsquo

1rsquo1910

asymp asymp asympHDO MeOD Acetone

6rsquo2rsquo

1615

3-712-14

4

939

940

Figure S 2 ndash Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 130degC 941

5

942

943

Figure S 3 - Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 90degC 944

945

90degC

1

22

6

946

947

948

949

Figure S 4 ndash Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 70degC 950

951

21

11

1

2

5

3

4

45

5

6

7

7

7

952

953

954

Figure S 5 ndash Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 25degC 955

956

957

1

1 1

8

958

Figure S 6 ndash a) SANS experiment recorded at 90degC for a 5 mgmL G-C181-OH aqueous solution b-c) 959

WAXS experiments recorded on aqueous solutions of G-C181-OH at b) 5 mgmL and c) 25 mgmL d) 960

Structural interpretation of the repeating distances pointed out in c) 961

962

963

964

001 01 1

q Aring-1

25degC

68degC

88degC

111degC

01 1q Aring

-1

02 04 06 08 1q Aring

-1

001 01

10-1

100

101

Inte

nsity

cm

-1

q Aring-1

-2

T= 90degC

SANS WAXS WAXS

a) b) c)

dv

dL d= 144 Aring-1

di

ddH2O

dt

dvdL

dtdi d

d)

9

965

966

Figure S 7 ndash ssNMR MAS (10 kHz) experiments recorded on a G-C181-OH dry powder a) 1H spectra 967

recorded from 110degC to 70degC b) 13C single pulse (SP) spectra recorded with high power 1H decoupling from 968

110degC to 70degC c) 13C cross polarization (CP) spectra recorded at 110degC and contact time tc of 50 micros and 969

3000 micros d) 13C CP spectra recorded at 70degC and for 20 lt tc micros lt 3000 970

971

1

811

910

CH2 CH2

1rsquo

6rsquo

8 6 4 2 0 ppm

110

90

80

70

910

1rsquo

6rsquo16

1

2rsquo3rsquo4rsquo

811

OH

45 40 35 30 25 20 15

ppm

200 180 160 140 120 100 80 60 40 20 0

ppm140 120 100 80

ppm

CH2-gauche

3004

110

90

80

70

65

60

50

25

C91013003

C9 - 12869

CH2-all-trans

3198

C1rsquo10335

a) b)

1H

T degCT degC

13C Single Pulse MAS

A

B

A

B

70degC70degC

5rsquo

215

200 180 160 140 120 100 80 60 40 20 0

ppm

13C CPMAST= 110degC

200 180 160 140 120 100 80 60 40 20 0

ppm

132 131 130 129 128 127 126

ppm

106 105 104 103 102

ppm

13C CPMAST= 70degC

50

tc micros

3000

20

tc micros

50 500 1000 3000

tc micros

C102

13024 C910113003

C92

12869

C1rsquo213082

C1rsquo110443

A B

106 105 104 103 102

ppm

3000

1000

500

50

20

AB

c)

d)

10

972

Figure S 8 ndash a) Short and d) long T2 components displayed by the 1H transverse relaxation signal c) 1H 973

transverse relaxation signals recorded on a G-C181-OH dry powder from 110degC to 70degC The area of each 974

given resonance is plotted (log-lin) as a function of the echo time (CPMG pulse sequence) for each 975

temperature and fitted using the equation given above T2s and T2l indicate the short and long T2 relaxation 976

times respectively obtained by describing the 1H transverse relaxation signal using a biexponential decay 977

For H1rsquo H5rsquo and H6rsquo the 1H relaxation function could be described by a monoexponential decay 978

- Protons from the sugar group (H1rsquo H5rsquo H6rsquo) show a single T2 relaxation component (Figure 979

6c and Figure S 8) while protons of the acyl chain (CH2 H910 H811) show both short (Figure 980

6c) and long (Figure 6d) T2 relaxation component 981

- The short T2 relaxation time systematically decreases from 110degC to 70degC for all species by a 982

factor ranging from two to seven depending on the specific chemical group Very long T2 are 983

observed at 110degC for all acyl species while nearly unchanged values are measured below 984

90degC without any clear trend 985

- The sugar protons (H1rsquo H5rsquo H6rsquo) have the shortest set of T2 values (lt 1 ms) below 100degC 986

The long component of T2 generally ranges between 4 ms and 12 ms 987

0 1 2 3 4 5 6 7 8

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1 2 3 4 5 6 7 8

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

H1

H8H11

H9H10

CH2

H1rsquo

H6rsquo

CH2

H910H811 H1CH2

H1rsquo H5rsquo6rsquo

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

OH

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45 T= 110degC

T= 90degC

T= 80degC

T= 70degC

Log

In

ten

sity

Time ms

Fit

)

CH2 H910 H811 H1 H56 H1 OH

00

05

10

15

20

25

30

35

40

T2 m

s

T= 110degC

T= 90degC

T= 80degC

T= 70degC

CH2 H910 H811 H1

0

2

4

6

8

10

12

T2 m

s

T= 110degC

T= 90degC

T= 80degC

T= 70degC

Short T2(1H) Long T2(1H)

a) b)

c)

11

Table S 1 ndash 1H chemical shifts of G-C181-OH in solution and in the solid-state Solid-state spectra are 988

recorded at a MAS frequency of 10 kHz in DMSO-d6 broad peak 989

990

δH ppm

Group Ndeg Liquid

(MeOD-d4) Solid

(T= 110degC) Solid

(T= 70degC)

CH2 H1 354 357

CH2 H2 153 156

CH2 H3-712-14 132 135

CH2 H811 203 203

CH H910 534 534

CH2 H15 162 164

CH2 H16 388 383

CH H1 424 432

CH H2 316 335

CH H3 334 335

CH H4 329 335

CH H5 354 351 357

CH2 H6 366 383

OH OH1 432

45 463

OH OH2 492

OH OH3 490

OH OH4 487

OH OH6 445

991

992

H1H9H10H1rsquo

H6rsquo

OH2rsquo

OH1

12

Detailed discussion about the NOESY and DQ-SQ 2D 1H-1H ssNMR experiments presented in 993

Figure 9 994

995

2D 1H-1H ssNMR experiments are a unique tool to explore through-space proximities 996

between 1H nuclei ssNMR offers a large number of possibilities each adapted to a given 997

system The advantage of 2D maps with respect to a dynamics approach is their direct 998

readability in clear two nuclei which exchange their magnetization through-space (whichever 999

mechanism may be responsible for this exchange) will display a specific cross-peak However 1000

the way cross-peaks are presented and interpreted strictly depends on the nature of the pulse 1001

program which in turn is chosen according to the physical nature of the sample 1002

1D temperature-dependent 1H NMR experiments performed on G-C181-OH (Figure 1003

4a in the main text) indicate sharp peaks at 90degC and broad peaks at 70degC Narrow peaks 1004

indicate fast andor large-amplitude molecular tumbling averaging the strong 1H-1H dipolar 1005

interactions This excludes the use of pulse sequences like DQ-SQ which is not efficient under 1006

these conditions For this reason we employ spin-diffusion-based pulse sequences like 1007

NOESY which is more adapted for samples with high molecular mobility When mixing times 1008

are chosen to be short (lt 1 s) one generally explores through-space proximities within about 5 1009

10 Aring On the contrary when molecular motions are restricted dipolar coupling becomes 1010

significant 1H NMR peaks are broader and pulse sequences like DQ-SQ can now be used 1011

with a spatial resolution of less than 5 Aring 1012

Whenever possible DQ-SQ experiments are preferred over spin diffusion-based pulse 1013

sequences because they are more selective in particular for likewise interactions which are 1014

not discriminated in NOESY-type experiments This is illustrated by the typical interaction 1015

schemes for NOESY (top Figure 9a) and DQ-SQ (top Figure 9b) For two given A and B 1016

protons 2D NOESY maps will display two on-diagonal and two square-shaped symmetrical 1017

off-diagonal cross-peaks Off-diagonal cross-peaks illustrate the mutual A-B and B-A 1018

interaction but the on-diagonal peaks are systematic and do not prove any A-A or B-B 1019

interaction Similarly to NOESY 2D DQ-SQ maps display off-diagonal peaks (symmetric 1020

across the diagonal top Figure 9b) upon A-B and B-A interaction but in contrast they will 1021

display on-diagonal peaks only in the presence of self-interaction In the tutorial example of 1022

Figure 9b top the on-diagonal A-A peak indicates mutual A-A interaction while lack of on-1023

diagonal B-B peak evidences the absence of B-B proximities Combining NOESY and DQ-SQ 1024

experiments is capital to study the local arrangement of G-C181-OH and its evolution with 1025

temperature 1026

13

At T= 90degC the H1 and H5rsquo contributions are very close but possible to discriminate 1027

them (δ(H1)= 354 ppm δ(H5rsquo)= 351 ppm) according to 1H NMR chemical shifts in Table S 1028

1 2D 1H-1H NOESY in Figure 9a also discriminates between the two components attributed 1029

on the basis of the interactions with their mutual neighbors (green H1-H2 and H5rsquo-H1rsquo squares 1030

in Figure 9a) On this basis 2D NOESY does not show a direct H1rsquo-H1 correlation (blue red-1031

crossed) but rather an extensive network of intra-sugar interactions These features exclude 1032

suggest intra- and inter-sugar interactions and they exclude head-tail arrangements as a major 1033

molecular configuration at 90degC 1034

At 70degC the 1H NMR peaks are broader indicating a reduction of the T2 (1H) relaxation 1035

times andor residual homonuclear dipolar interactions excluding the use of NOESY In this 1036

case 1H-1H DQ-SQ correlation experiments are best suited Figure 9b presents the 1H-1H DQ-1037

SQ experiment on G-C181-OH at 70degC Peaks are broader than in NOESY and attribution may 1038

be cumbersome However the experiment is performed at 700 MHz and MAS= 22 kHz which 1039

improves spectral resolution with respect to the same experiment performed at 300 MHz and 1040

MAS= 10 kHz The auto-peak at δH= 356 ppm in the SQ dimension is attributed to H1-H1 spin 1041

pairs mainly resulting from intra CH2 interaction while the DQ cross-peaks at δH= 156-356 1042

ppm and δH= 432-356 ppm respectively origin from H2-H1 and H1rsquo-H1 interactions (Figure 1043

9b) DQ-SQ map does not show any auto-peak corresponding to H1rsquo-H1rsquo or Hxrsquo-Hxrsquo (x= 2rsquo 1044

3rsquo 4rsquo) interactions (red-crossed blue Figure 9b) thus excluding any proximity (lt 5 Aring) between 1045

close glucose neighbors On the contrary DQ-SQ maps (recorded both at 700 MHz Figure 9b 1046

or at 300 MHz not shown) systematically show DQ cross-peaks which can only be attributed 1047

to H1rsquo-H1 interactions (δH= 432-356) The presence of these H1rsquo-H1 interactions can only be 1048

explained by a head-tail arrangement of the G-C181-OH molecules as illustrated below the 1049

DQ-SQ map in Figure 9b 1050

23

571

Figure 8 ndash a) Thermotropic phase behavior of G-C181-OH at 5 mgmL-1 Note that Tm2 is observed on the 572

dry powder only b) Possible relationship between the molecular environment as derived from NMR and 573

membrane curvatures All the molecular drawings refer to the same G-C181-OH molecule (bulky ellipsoid 574

equiv sugar headgroup) while each color corresponds to a given configuration Grey of C1rsquo1 and C1rsquo2 indicates 575

the uncertainty about the assignment of the specific NMR peak The curvature C is defined as 1R R being 576

the curvature radius R1 R2 radius of curvature CG Gauss curvature CM Mean curvature c) Free 577

interpretation of the anisotropic high-curvature (CMgt 0 CG= 0) inclusions in the membrane (low CM and 578

CG) continuum driven by inter-sugar H-bonding and reduced intramembrane diffusion for a G-C181-OH 579

glucolipid 580

581

Flat crystalline structure

Nanotube-vesicle network

Vesicles

48degC

120deg

106 105 104 103 102 101

ppm

H9 H10H1rsquo

106 105 104 103 102 101

ppm106 105 104 103 102 101

ppm106 105 104 103 102 101

ppm

C=CC=C

C1rsquo1C1rsquo2

C1rsquo

Tm2(96degC)Tm1

a)

b)

c)

Same molecule different local orientations and sugar conformations

High curvature membrane

Low curvature inclusion

Flip-FlopStrong H-bonding

CG

CM ( + )

CG 0

CM 0 CG 0

CM 0

G-C181-OH

24

Two related questions are unanswered at the moment What actually drives nanotube 582

formation and is the organization of G-C181-OH within the vesicles and tubes the same Lipid 583

nanotubes have been studied for at least three decades over which broad experimental and 584

theoretical insight has been generated Despite this fact the actual understanding is still matter 585

of debate27 Micelle-to-nanotube and vesicle-to-nanotube phase transitions are generally 586

reported for single amphiphiles below a given Tm and intermediate phase composed of twisted 587

and flat fibers is concomitantly reported Initially thought to be a chirality-driven phenomenon 588

recent studies propose a more general mechanism involving crystallization and symmetry 589

breaking in the lipid packing driven by optimal packing27 If twisted ribbons were observed 590

before for a number of microbial glycolipid bolaamphiphiles44 they are not found for G-C181-591

OH This fact as well as the large body of cryo-TEM images and the observation of nanotubes 592

above Tm1 rather pledge in favour of a direct topological change from positive (vesicles) to 593

zero (nanotubes) Gaussian curvatures In this case vesicle-to-nanotube transitions could be 594

driven by external forces56307813ndash162829 595

Considering the absence of external forces in the present system there are only two 596

possibilities left spontaneous budding of the membrane or nanotube pulling after inter-vesicle 597

collisions due to Brownian motion Both are possible as inclusions composed of a different 598

arrangement of the same molecule in the membrane continuum can theoretically explain 599

spontaneous nanotube formation in an isotropic medium672223629ndash1218ndash21 However the second 600

one is probably more likely to prevail because it could explain the large number of nanotubes 601

connecting two distant vesicles Interestingly similar events occur for any phospholipid-based 602

vesicular solution but nanotubes are extremely rare unless at least two lipids of different 603

molecular structure are mixed together The spectroscopic 1H and 13C and relaxation (1H) 604

ssNMR data even if collected on a dry powder demonstrate that the coexistence of two G-605

C181-OH configurations involving both the sugar headgroup and fatty alcohol chain are 606

possible for this molecule As a matter of fact this is not so surprising as more than one energy 607

minima each corresponding to a given conformation are known for glucose its non-reducing 608

derivative methylglucose and carbohydrates in general78ndash80 609

The presence of two configurations could explain the coexistence between two 610

structures of different curvature but the data presented so far do not explain the origin of these 611

two configurations In other words what drives the change in the local structure of the sugar 612

headgroup and fatty alcohol within the membrane 613

The association of known self-assembled structures to spectroscopic 13C ssNMR data is 614

known for years in particular for gluconamides lipids having a linear sugar headgroup81ndash84 It 615

25

was found that gauche effects in some cases up to the γ position in the sugar chain induce 616

upfield chemical shifts while anti and trans conformations result in downfield shifts Each 617

structure (monolayer crystal bilayer crystal or micellar fiber) was then characterized by a given 618

organization of the gluconamide and a characteristic 13C ssNMR fingerprint which could be 619

explained by gauche γ-gauche anti and trans conformations of the headgroup Similar 620

structural-spectroscopic correlations have been collected over the years for much simpler cyclic 621

sugars85 whereas pyranoses can undergo important conformational changes (chair half-chair 622

boat skew) when the appropriate energy barriers are crossed78ndash8086 A quantitative 623

conformational study based on the 13C chemical shift variations shown in Table 1 would be 624

certainly helpful but very delicate in the context of the present state of the art because a 625

rigorous approach requires at least one known crystal structure8182 which is currently missing 626

for this compound A qualitative approach will then be adopted 627

G-C181-OH is a bolaform amphiphile and flip-flop could be a possible mechanism 628

involved in the nanotube stabilization whereas head-tail interactions have long been considered 629

as important in nanotube-forming bolaform amphiphiles6063 From mere considerations based 630

on T2 (1H) relaxation arguments one could exclude head-tail configurations if the R-CH2OH 631

endgroup (H1) was involved in a direct interaction with the sugar group its corresponding T2 632

would probably be comparable to the sugarrsquos one (OH and H1rsquo) but this does not seem to be 633

the case However this hypothesis would be based on the comparison of the relaxation from 634

CH and CH2 protons additionally involved in different kind of motions Besides the relaxation 635

of H1 is influenced by the strong overlap between H1 and H5rsquo For these reasons relaxation 636

arguments are not solid enough and will not be used 637

We then address this question by exploring intermolecular interactions in a direct 638

manner using 2D 1H-1H homonuclear ssNMR correlation spectroscopy At 90degC in the fast-639

tumbling domain of molecular motion (narrow 1H NMR peaks Figure 6a) and corresponding 640

to the vesicle-rich region spin diffusion (NOESY) experiments are preferred over dipolar 641

coupling based experiments because such interaction is averaged out by molecular motion 642

However when the molecular motion is more constrained in the domain of nanotube stability 643

at 70degC (broad 1H spectra Figure 6a) Double-Quantum Single-Quantum (DQ-SQ) correlation 644

experiments are best suited to probe interaction between close molecular neighbors These 645

experiments are reported in Figure 9 and the related interpretation is only briefly reported 646

below while a more detailed discussion can be found on Page 12 in the Supporting Information 647

26

648

Figure 9 ndash a) 2D 1H-1H NOESY NMR experiment performed at MAS= 10 kHz (4 mm probe) on a Bruker 649

300 MHz Avance III HD spectrometer using 300 ms of mixing time Sample temperature is regulated to 650

90degC reached by cooling from 110degC b) 2D 1H-1H DQ-SQ (BAck-to-BAck BABA pulse sequence used for 651

the DQ excitation scheme) NMR experiment performed at MAS= 22 kHz (25 mm probe) on a Bruker 700 652

MHz Avance III spectrometer using a DQ excitation time equal to 2 rotor periods Sample temperature is 653

regulated to 70degC reached by cooling from 110degC The typical schemes illustrating how to interpret these 654

2D experiments are reported above each experimental map Zooms of the sugar region and configurational 655

interpretation are given below each 2D map More information are given in the text below 656

H16-H16

H1-H1

H9-H10

H1rsquo-H1

H1rsquoH910 H16

H1

H1rsquo-H1rsquo

Hxrsquo-Hxrsquo

2

(pp

m)

(ppm)

315

811910 1rsquo

2rsquo2

3rsquo 4rsquo6rsquo

5rsquo

1rsquo2910

asymp asymp asympHDO MeOD Acetone

1

315

811 9101rsquo

2rsquo2

3rsquo4rsquo6rsquo

5rsquo

1rsquo2 9 10

asympasympasympHDOMeODAcetone

H1rsquo-H1

1rsquo

1

(p

pm

)

(ppm)

H1-H2

H1rsquo-H1

H1rsquo

H910

H1H5rsquo

H2

H1rsquo-H5rsquo

315

811910 1rsquo

2rsquo2

3rsquo 4rsquo6rsquo

5rsquo

1rsquo2910

asymp asymp asympHDO MeOD Acetone

1

H2rsquoH3rsquoH4rsquo

H2rsquoH3rsquoH4rsquo

X= 2rsquo3rsquo4rsquo

NOESY - T= 90degC

1H

1H

1H

1H

DQ-SQ - T= 70degC

a) b)

2

(pp

m)

(p

pm

)

(ppm)

A B

AhellipB

AhellipA

BhellipB

NOESY

(ppm)

A B

AhellipB

DQ-SQ

Single-quantum dimension

Do

ub

le-q

ua

ntu

m d

imen

sio

n

AhellipA

BhellipB

BhellipA

BhellipA

315

811910 1rsquo

2rsquo2

3rsquo 4rsquo6rsquo

5rsquo

1rsquo2910

asymp asymp asympHDO MeOD Acetone

1

lt 5 Aring

lt 5 Aring

H1-H2

H811

H15

H16

H6rsquo

H1rsquo

H1

H5rsquo

H2

H1-H2

H1rsquo-H1

H1rsquo-H5rsquo

H1rsquo

H1 H5rsquo

H1rsquo

H1

H5rsquo

H1H2rsquo H3rsquo H4rsquo

H1-H1

H1rsquo-H1

Hxrsquo-HxrsquoX= 2rsquo3rsquo4rsquo

H1-H2

H1rsquo

Zoom sugar region 28 ndash 45 ppmNOESY DQ-SQ

27

Head-tail or head-head configurations could be probed by the existence of H1rsquo-H1 657

interactions whereas H1 (δH= 356 ppm) is associated to the glucolipid tail and H1rsquo (δH= 432 658

ppm) to the sugar headgroup These atoms are more than 15 Aring apart and a direct interaction 659

could not be observed in both NOESY and DQ-SQ experiments NOESY experiments (Figure 660

9a) show that H1rsquo-H1 correlations are not observed Meanwhile clear-cut direct H1rsquo-H1 661

interactions are indeed observed are detected at T= 70degC (DQ cross-peaks related to H1rsquo-H1 662

correlation in DQ-SQ experiment shown in Figure 9b) Interestingly the absence of auto peaks 663

at δH= 432 ppm and δH= 335 ppm in the SQ dimension excludes H1rsquo-H1rsquo and Hxrsquo-Hxrsquo (x= 2 664

3 4) interactions respectively strengthening the idea that sugar headgroups from two nearest-665

neighbor G-C181-OH molecules are not adjacent in the nanotube-vesicle region 666

Homonuclear 2D ssNMR supports the idea according to which G-C181-OH can 667

undergo molecular flip-flops and a head-tail arrangement in the dry product Extrapolating these 668

conclusions in the lower concentration range lack of H1rsquo-H1 proximity at T= 90degC suggests a 669

head-head configuration in the vesicle-rich region and head-tail configuration within the 670

nanotubes It is however still unclear whether flip-flops are driven by spontaneous temperature-671

induced conformational changes in the glucolipid packing stimulating inclusions within the 672

membrane itself672223629ndash1218ndash21 or by defects generated by inter-vesicle collisions In fact a 673

combination of both is not unlikely Nanotubes are not stable when temperature is 20 to 30 674

degrees above Tm1 indicating that inter-vesicle collisions alone cannot explain the unusual 675

phenomenon In an energy landscape approach87 the head-head configuration is certainly 676

favored at high temperature while both head-head and head-tail configurations could be 677

favored in the mid-temperature region corresponding to the range between Tm1 and Tm2 in the 678

dry powder Both statistical and defect-induced flip-flop events contribute to form high mean 679

zero Gaussian curvature morphologies (nanotubes) for which the head-tail conformation 680

limiting the proximity between bulky sugar headgroups is probably more adapted (Figure 8c) 681

The flip-flop interpretation now reasonably explains the strong differences in the sugar and fatty 682

alcohol conformations deduced by the multiple NMR signals attributed to chemically-683

equivalent groups The existence of various low-energy conformations each having specific 684

NMR signature for sugars in solution78ndash8086 and solid-state81ndash84 further supports our 685

interpretation 686

One last point should be clarified Why between Tm1 and Tm2 are two molecular 687

environments able to coexist instead of a single one of average rigidity as found in the case of 688

ldquoripplerdquo phases50ndash54 We believe that this question can be answered based on the NMR results 689

the relatively short T2 values of the H1rsquo and OH protons shows the constrained mobility of the 690

28

sugar headgroup despite a certain flexibility of the chain (long T2 component) This trend may 691

suggest restrictions in translational mobility probably due to the sugar-OH hydrogen bonding 692

(Figure 8c) meaning that lipids are not able to diffuse within the membrane thus keeping two 693

different configurations stable 694

695

Conclusions 696

Topologically-connected nanotube-vesicle lipidic systems are commonly observed for 697

self-assembled membranes composed of at least two amphiphiles and spontaneously driven by 698

anisotropic inclusions andor by external directional forces This work shows that strong 699

differences in membrane curvature can spontaneously occur in single-lipid systems without 700

external anisotropic forces Coexistence and interconnection between nanotubes and vesicles 701

are evidenced via cryo-TEM and fluorescence confocal microscopy in a single-phase region at 702

temperature above a transition temperature Tm1= 483degC determined by DSC on the hydrated 703

sample but below the second transition temperature Tm2 (957degC) determined by DSC on the 704

dry powder X-ray diffraction (WAXS) also shows the coexistence of two lattice periods one 705

attributed to an inter-membrane repeating distance and the other possibly related to the intra-706

nanotube glucolipid arrangement Labelling the systems with Laurdan a fluorophore of which 707

the emission strongly varies with the local membrane rigidity excludes the presence of rigid 708

micro-separated phases (eg Lβ) 709

1H and 13C ssNMR despite the fact that it can only be performed on the dry powder 710

confirm the presence of one main fluid G-C181-OH environment above Tm2 Below Tm2 the 711

splitting of some 13C peaks and the coexistence of short and long 1H relaxation components 712

demonstrate that G-C181-OH can simultaneously pack into structurally and dynamically 713

inequivalent environments Combining advanced 2D 1H-1H correlation experiments based on 714

spin diffusion or double-quanta excitation suggests that vesicles could contain a preferred head-715

head configuration of the bolaform lipid while nanotubes a preferred head-tail assembly 716

The presence of two environments composed of the same molecule characterized by 717

different molecular configurations and rigidity fulfills the theoretical conditions necessary to 718

spontaneously nucleate a membrane inclusion of non-zero average orientation described as 719

necessary to observe topologically-connected nanotube-vesicle structures in the absence of 720

external forces This hypothesis could also be enriched by the possibility of nanotubes to be 721

pulled away between two vesicles after a collision event 722

This work expands the actual theoretical and experimental framework on lipid networks 723

stabilized by a different radius of curvature It could also motivate the synthesis or isolation 724

29

of new lipids providing a higher level of 3D architectural control in soft networks possibly in 725

the range of room temperature 726

727

Acknowledgements 728

Dr Andrea Lassenberger and Dr Sylvain Preacutevost at Institut Laue Langevin (ILL Grenoble 729

France) are kindly acknowledged for their assistance on the SANS experiment Confocal 730

images were performed at the Institut de Biologie Paris Seine (IBPS) imaging facility Jean-731

Franccedilois Gilles (Imaging Core Facility Institut de Biologie Paris Seine (IBPS) CNRS 732

Sorbonne Universiteacute Paris France) is kindly acknowledged for his assistance on the confocal 733

microscopy imaging Dr Guillaume Laurent (LCMCP Sorbonne Universiteacute Paris France) is 734

kindly acknowledged for helpful discussions The French region Ile-de-France SESAME 735

program is acknowledged for financial support (700 MHz NMR 736

spectrometer) Dr Olivier Diat (ICSM Marcoule France) Prof eronikaIglič and Prof V Aleš737

onsul discussi Slovenia) are kindly acknowledged for helpfUniversity of LjubljanaIglič (-Kralj 738

739

Financial support 740

WAXS experiments have been supported by Soleil Light Source Saint Aubin France proposal 741

Ndeg 20201747 SANS experiments have been supported by ILL proposal Ndeg9-13-778 742

743

References 744

1 D Kabaso N Bobrovska W Goacuteźdź E Gongadze V Kralj-Iglič R Zorec and A 745

Iglič Bioelectrochemistry 2012 87 204ndash210 746

2 C R Safinya U Raviv D J Needleman A Zidovska M C Choi M A Ojeda-747

Lopez K K Ewert Y Li H P Miller J Quispe B Carragher C S Potter M W 748

Kim S C Feinstein and L Wilson Adv Mater 2011 23 2260ndash2270 749

3 P Veranič M Lokar G J Schuumltz J Weghuber S Wieser H Haumlgerstrand V Kralj-750

Iglič and A Iglič Biophys J 2008 95 4416ndash4425 751

4 A Rustom R Saffrich I Markovic P Walther and H-H Gerdes Science (80- ) 752

2004 303 1007ndash1010 753

5 M I Angelova S Soleacuteau P Meacuteleacuteard F Faucon and P Bothorel Prog Colloid 754

Polym Sci 1992 89 127ndash131 755

6 L Mathivet S Cribier and P F Devaux Biophys J 1996 70 1112ndash1121 756

7 V Kralj-Iglič G Gomišček J Majhenc V Arrigler and S Svetina Colloids Surfaces 757

A Physicochem Eng Asp 2001 181 315ndash318 758

30

8 V Kralj-Iglič A Iglič G Gomišček F Sevšek V Arrigler and H Haumlgerstrand J 759

Phys A Math Gen 2002 35 1533ndash1549 760

9 M Drab D Stopar V Kralj-Iglič and A Iglič Cells 2019 8 626 761

10 A Zidovska K K Ewert J Quispe B Carragher C S Potter and C R Safinya 762

Langmuir 2009 25 2979ndash2985 763

11 A Zidovska K K Ewert J Quispe B Carragher C S Potter and C R Safinya Soft 764

Matter 2011 7 8363ndash8369 765

12 A Zidovska K K Ewert J Quispe B Carragher C S Potter and C R Safinya 766

Biochim Biophys Acta - Biomembr 2009 1788 1869ndash1876 767

13 M Karlsson K Sott M Davidson A S Cans P Linderholm D Chiu and O Orwar 768

Proc Natl Acad Sci U S A 2002 99 11573ndash11578 769

14 A Karlsson R Karlsson M Karlsson A-S Cans A Stromberg F Ryttsen and O 770

Orwar Nature 2001 409 150ndash152 771

15 M Karlsson K Sott A S Cans A Karlsson R Karlsson and O Orwar Langmuir 772

2001 17 6754ndash6758 773

16 E Evans H Bowman A Leung D Needham and D Tirrell Science (80- ) 1996 774

273 933ndash935 775

17 U Seifert Configurations of fluid membranes and vesicles 1997 vol 46 776

18 J Kaumls and E Sackmann Biophys J 1991 60 825ndash844 777

19 N Bobrovska W Goacuteźdź V Kralj-Iglič and A Iglič PLoS One 2013 8 e73941 778

20 A Iglič B Babnik K Bohinc M Fošnarič H Haumlgerstrand and V Kralj-Iglič J 779

Biomech 2007 40 579ndash585 780

21 K Bohinc T Slivnik A Iglič M Brumen and V Kralj-Iglič J Phys Chem C 2007 781

111 9709ndash9718 782

22 A Iglič H Haumlgerstrand P Veranič A Plemenitaš and V Kralj-Iglič J Theor Biol 783

2006 240 368ndash373 784

23 V Kralj-Iglič and P Veranič Adv Planar Lipid Bilayers Liposomes 2006 5 129ndash785

149 786

24 A Iglic and V Kralj-Iglic Adv Planar Lipid Bilayers Liposomes 2006 4 253ndash279 787

25 S Penič L Mesarec M Fošnarič L Mroacutewczyńska H Haumlgerstrand V Kralj-Iglič and 788

A Iglič Front Phys 2020 8 342 789

26 P Yager and P E Schoen Mol Cryst Liq Cryst 1984 106 371ndash381 790

27 T G Barclay K Constantopoulos and J Matisons Chem Rev 2014 114 10217ndash791

10291 792

31

28 B Božič G Gomišček V Kralj-Iglič S Svetina and B Žekš Eur Biophys J 2002 793

31 487ndash496 794

29 A Iglič H Haumlgerstrand M Bobrowska-Haumlgerstrand V Arrigler and V Kralj-Iglič 795

Phys Lett A 2003 310 493ndash497 796

30 R Bar-Ziv and E Moses Phys Rev Lett 1994 73 1392ndash1395 797

31 S Chiruvolu H E Warriner E Naranjo S H Idziak J O Raumldler R J Plano J A 798

Zasadzinski and C R Safinya Science (80- ) 1994 266 1222ndash1225 799

32 V Kralj-Iglič A Iglič H Haumlgerstrand and P Peterlin Phys Rev E 2000 61 4230ndash800

4234 801

33 L Mesarec M Drab S Penič V Kralj-Iglič and A Iglič Int J Mol Sci 2021 22 802

2348 803

34 L Van Renterghem S L K W Roelants N Baccile K Uyttersprot M C Taelman 804

B Everaert S Mincke S Ledegen S Debrouwer K Scholtens C V Stevens W 805

Soetaert L Van Renterghem S L K W Roelants N Baccile K Uytersprot M C 806

Taelman B Everaert S Mincke S Ledegen S Debrouwer K Scholtens C V 807

Stevens and W Soetaert Biotechnol Bioeng 2018 115 1195ndash1206 808

35 J Clark and F Deswarte Introduction to Chemicals from Biomass Second Edition 809

John Wiley amp Sons 2015 810

36 K V Sajna R Houmlfer R K Sukumaran L D Gottumukkala and A Pandey in 811

Industrial Biorefineries and White Biotechnology eds A Pandey R Houmlfer M 812

Taherzadeh K M Nampoothiri and C Larroche Amsterdam Oxford Waltham 813

Elsevier 2015 pp 499ndash521 814

37 R Marchant and I M Banat Trends Biotechnol 2012 30 558ndash565 815

38 K Hill and C LeHen-Ferrenbach in Sugar-Based Surfactants Fundamentals and 816

Applications ed C C Ruiz Boca Raton CRC Press 2009 pp 1ndash20 817

39 L Rodrigues I M Banat J Teixeira and R Oliveira J Antimicrob Chemother 818

2006 57 609ndash618 819

40 J D Desai and I M Banat Microbiol Mol Biol Rev 1997 61 47ndash64 820

41 S L K W Roelants K Ciesielska S L De Maeseneire H Moens B Everaert S 821

Verweire Q Denon B Vanlerberghe I N A Van Bogaert P Van der Meeren B 822

Devreese and W Soetaert Biotechnol Bioeng 2016 113 550ndash559 823

42 N Baccile M Selmane P Le Griel S Preacutevost J Perez C V Stevens E Delbeke S 824

Zibek M Guenther W Soetaert I N A Van Bogaert and S Roelants Langmuir 825

2016 32 6343ndash6359 826

32

43 N Baccile A-S Cuvier S Preacutevost C V Stevens E Delbeke J Berton W Soetaert 827

I N A Van Bogaert and S Roelants Langmuir 2016 32 10881ndash10894 828

44 N Baccile C Seyrig A Poirier S A Castro S L K W Roelants and S Abel Green 829

Chem 2021 23 3842ndash3944 830

45 A Salis L Cappai C Carucci D F Parsons and M Monduzzi J Phys Chem Lett 831

2020 11 6805ndash6811 832

46 J Schindelin I Arganda-Carreras E Frise V Kaynig M Longair T Pietzsch S 833

Preibisch C Rueden S Saalfeld B Schmid J-Y Tinevez D J White V 834

Hartenstein K Eliceiri P Tomancak and A Cardona Nat Methods 2012 9 676ndash835

682 836

47 T Parasassi and E Gratton J Fluoresc 1995 5 59ndash69 837

48 M M Dodes Traian F L Gonzaacutelez Flecha and V Levi J Lipid Res 2012 53 609ndash838

616 839

49 D M Owen C Rentero A Magenau A Abu-Siniyeh and K Gaus Nat Protoc 840

2012 7 24ndash35 841

50 W Stillwell in An Introduction to Biological Membranes Academic Press 2016 pp 842

135ndash180 843

51 W Stillwell in An Introduction to Biological Membranes Academic Press 2016 pp 844

181ndash220 845

52 B Tenchov Chem Phys Lipids 1991 57 165ndash177 846

53 R Koynova and M Caffrey Biochim Biophys Acta 1998 1376 91ndash145 847

54 F Schmid S Dolezel O Lenz and S Meinhardt J Phys Conf Ser 2014 487 848

012004 849

55 S Meinhardt R L C Vink and F Schmid Proc Natl Acad Sci U S A 2013 110 850

4476ndash4481 851

56 O Glatter and O Kratky Small Angle X-ray Scattering Academic Press London 852

1982 853

57 B A Cunningham A D Brown D H Wolfe W P Williams and A Brain Phys 854

Rev E 1998 58 3662ndash3672 855

58 M Nadler A Steiner T Dvir O Szekely P Szekely A Ginsburg R Asor R Resh 856

C Tamburu M Peres and U Raviv Soft Matter 2011 7 1512ndash1523 857

59 C Tanford The hydrophobic effect Formation of micelles and biological membranes 858

John Wiley amp Sons Inc New York 1973 859

60 M Masuda and T Shimizu Langmuir 2004 20 5969ndash5977 860

33

61 A Zidovska K K Ewert J Quispe B Carragher C S Potter and C R Safinya 861

Block Liposomes Vesicles of Charged Lipids with Distinctly Shaped Nanoscale 862

Sphere- Pear- Tube- or Rod-Segments Elsevier Inc 1st edn 2009 vol 465 863

62 A Iglic V Kralj-Iglic B Božič M Bobrowska-Haumlgerstrand B Isomaa and H 864

Haumlgerstrand Bioelectrochemistry 2000 52 203ndash211 865

63 T Shimizu M Masuda and H Minamikawa Chem Rev 2005 105 1401ndash1444 866

64 J V Selinger and J M Schnur Phys Rev Lett 1993 71 4091ndash4094 867

65 J M Schnur Science (80- ) 1993 262 1669ndash1676 868

66 B Tenchov R Koynova and G Rapp Biophys J 2001 80 1873ndash1890 869

67 R Koynova and B Tenchov Ser Biomech 2017 31 3ndash19 870

68 S J Angyal and R Le Fur Carbohydr Res 1984 126 15ndash26 871

69 J K Whitesell and M A Minton J Am Chem Soc 1987 109 225ndash228 872

70 T Takayama S Ando and I Ando J Mol Struct 1990 220 245ndash250 873

71 K Gawrisch N V Eldho and I V Polozov Chem Phys Lipids 2002 116 135ndash151 874

72 F Tielens C Gervais J F Lambert F Mauri and D Costa Chem Mater 2008 20 875

3336ndash3344 876

73 G A Jeffrey and Y Yeon Acta Crystallogr Sect B 1986 42 410ndash413 877

74 B Berglund and R W Vaughan J Chem Phys 1980 73 2037ndash2043 878

75 C Gervais C Coelho T Azaiumls J Maquet G Laurent F Pourpoint C Bonhomme 879

P Florian B Alonso G Guerrero P H Mutin and F Mauri J Magn Reson 2007 880

187 131ndash140 881

76 C Akita T Kawaguchi F Kaneko H Yamamoto and M Suzuki J Phys Chem B 882

2004 108 4862ndash4868 883

77 R N Purusottam L Seacutenicourt J J Lacapegravere and P Tekely Biochim Biophys Acta - 884

Biomembr 2015 1848 3134ndash3139 885

78 X Biarneacutes A Ardegravevol A Planas C Rovira A Laio and M Parrinello J Am Chem 886

Soc 2007 129 10686ndash10693 887

79 L M J Kroon‐Batenburg and J Kroon Biopolymers 1990 29 1243ndash1248 888

80 K N Kirschner and R J Woods Proc Natl Acad Sci U S A 2001 98 10541ndash889

10545 890

81 S Svenson B Kirste and J H Fuhrhop J Am Chem Soc 1994 116 11969ndash11975 891

82 I Sack S Macholl J H Fuhrhop and G Buntkowsky Phys Chem Chem Phys 892

2000 2 1781ndash1788 893

83 S Svenson J Koening and J H Fuhrhop J Phys Chem 1994 98 1022ndash1028 894

34

84 S Svenson A Schaefer and J H Fuhrhop J Chem Soc Perkin Trans 2 1994 2 895

1023ndash1028 896

85 K Bock and C Pedersen Adv Carbohydr Chem Biochem 1983 41 27ndash66 897

86 T B Grindley Structure and Conformation of Carbohydrates Springer Berlin 898

Heidelberg 2008 899

87 F Tantakitti J Boekhoven X Wang R V Kazantsev T Yu J Li E Zhuang R 900

Zandi J H Ortony C J Newcomb L C Palmer G S Shekhawat M O De La 901

Cruz G C Schatz and S I Stupp Nat Mater 2016 15 469ndash476 902

903

1

Supporting Information 904

905

Topological connection between vesicles and nanotubes in single-906

component lipid membranes driven by head-tail interactions 907

908

Niki Baccilea Ceacutedric Lorthioira Abdoul Aziz Baa Patrick Le Griela Cristina Coelhob Javier 909

Perezc Wim Soetaertd Sophie L K W Roelantsd 910

911

a Sorbonne Universiteacute Centre National de la Recherche Scientifique Laboratoire de Chimie de 912

la Matiegravere Condenseacutee de Paris LCMCP F-75005 Paris France 913

b Institut des Mateacuteriaux de Paris Centre Sorbonne Universiteacute Paris France 914

c Synchrotron Soleil LrsquoOrme des Merisiers Saint-Aubin BP48 91192 Gif-sur-Yvette Cedex 915

France 916

d InBio Department of Biotechnology Ghent University Ghent Belgium 917

918

919

Corresponding author 920

Dr Niki Baccile 921

E-mail address nikibaccilesorbonne-universitefr 922

Phone +33 1 44 27 56 77 923

924

925

926

2

a) 927

Parameter Method

Dry Matter (DM ) 996 Infrared balance 105degC

Glucose () 012 HPLC-Metacarb

Glycerol () nd HPLC-Metacarb

Free Fatty Acid content () lt001 Internal method BBEPP

GC with FID Detector

Oil () lt001 Internal method BBEPP

GC with FID Detector

Moulds (CFUg) lt10 3M-Nordval ndeg16 B

Yeasts (CFUg) lt10 3M-Nordval ndeg16 B

Anaerobic count (CFUg) lt10 SP-VG M005 B

Protein () tbd BCA protein assay

928

Figure S 1 - a) HPLC-ELSD chromatogram and composition table 929

930

931

932

933

3

b) 934

935

Group Ndeg δ ppm

CH2 1 354

CH2 2 153

CH2 3-712-14 132

CH2 811 203

CH 910 534

CH2 15 162

CH2 16 388

CH 1 424

CH 2 316

CH 3 334

CH 4 329

CH 5 354

CH2 6 366

Figure S 1 (next) ndash b) 1H NMR spectrum (MeOD-d4) recorded on the purified G-C181-OH compounds 936

and related assignment 937

938

2

811910 1rsquo

5rsquo1

3rsquo4rsquo

1rsquo1910

asymp asymp asympHDO MeOD Acetone

6rsquo2rsquo

1615

3-712-14

4

939

940

Figure S 2 ndash Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 130degC 941

5

942

943

Figure S 3 - Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 90degC 944

945

90degC

1

22

6

946

947

948

949

Figure S 4 ndash Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 70degC 950

951

21

11

1

2

5

3

4

45

5

6

7

7

7

952

953

954

Figure S 5 ndash Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 25degC 955

956

957

1

1 1

8

958

Figure S 6 ndash a) SANS experiment recorded at 90degC for a 5 mgmL G-C181-OH aqueous solution b-c) 959

WAXS experiments recorded on aqueous solutions of G-C181-OH at b) 5 mgmL and c) 25 mgmL d) 960

Structural interpretation of the repeating distances pointed out in c) 961

962

963

964

001 01 1

q Aring-1

25degC

68degC

88degC

111degC

01 1q Aring

-1

02 04 06 08 1q Aring

-1

001 01

10-1

100

101

Inte

nsity

cm

-1

q Aring-1

-2

T= 90degC

SANS WAXS WAXS

a) b) c)

dv

dL d= 144 Aring-1

di

ddH2O

dt

dvdL

dtdi d

d)

9

965

966

Figure S 7 ndash ssNMR MAS (10 kHz) experiments recorded on a G-C181-OH dry powder a) 1H spectra 967

recorded from 110degC to 70degC b) 13C single pulse (SP) spectra recorded with high power 1H decoupling from 968

110degC to 70degC c) 13C cross polarization (CP) spectra recorded at 110degC and contact time tc of 50 micros and 969

3000 micros d) 13C CP spectra recorded at 70degC and for 20 lt tc micros lt 3000 970

971

1

811

910

CH2 CH2

1rsquo

6rsquo

8 6 4 2 0 ppm

110

90

80

70

910

1rsquo

6rsquo16

1

2rsquo3rsquo4rsquo

811

OH

45 40 35 30 25 20 15

ppm

200 180 160 140 120 100 80 60 40 20 0

ppm140 120 100 80

ppm

CH2-gauche

3004

110

90

80

70

65

60

50

25

C91013003

C9 - 12869

CH2-all-trans

3198

C1rsquo10335

a) b)

1H

T degCT degC

13C Single Pulse MAS

A

B

A

B

70degC70degC

5rsquo

215

200 180 160 140 120 100 80 60 40 20 0

ppm

13C CPMAST= 110degC

200 180 160 140 120 100 80 60 40 20 0

ppm

132 131 130 129 128 127 126

ppm

106 105 104 103 102

ppm

13C CPMAST= 70degC

50

tc micros

3000

20

tc micros

50 500 1000 3000

tc micros

C102

13024 C910113003

C92

12869

C1rsquo213082

C1rsquo110443

A B

106 105 104 103 102

ppm

3000

1000

500

50

20

AB

c)

d)

10

972

Figure S 8 ndash a) Short and d) long T2 components displayed by the 1H transverse relaxation signal c) 1H 973

transverse relaxation signals recorded on a G-C181-OH dry powder from 110degC to 70degC The area of each 974

given resonance is plotted (log-lin) as a function of the echo time (CPMG pulse sequence) for each 975

temperature and fitted using the equation given above T2s and T2l indicate the short and long T2 relaxation 976

times respectively obtained by describing the 1H transverse relaxation signal using a biexponential decay 977

For H1rsquo H5rsquo and H6rsquo the 1H relaxation function could be described by a monoexponential decay 978

- Protons from the sugar group (H1rsquo H5rsquo H6rsquo) show a single T2 relaxation component (Figure 979

6c and Figure S 8) while protons of the acyl chain (CH2 H910 H811) show both short (Figure 980

6c) and long (Figure 6d) T2 relaxation component 981

- The short T2 relaxation time systematically decreases from 110degC to 70degC for all species by a 982

factor ranging from two to seven depending on the specific chemical group Very long T2 are 983

observed at 110degC for all acyl species while nearly unchanged values are measured below 984

90degC without any clear trend 985

- The sugar protons (H1rsquo H5rsquo H6rsquo) have the shortest set of T2 values (lt 1 ms) below 100degC 986

The long component of T2 generally ranges between 4 ms and 12 ms 987

0 1 2 3 4 5 6 7 8

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1 2 3 4 5 6 7 8

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

H1

H8H11

H9H10

CH2

H1rsquo

H6rsquo

CH2

H910H811 H1CH2

H1rsquo H5rsquo6rsquo

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

OH

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45 T= 110degC

T= 90degC

T= 80degC

T= 70degC

Log

In

ten

sity

Time ms

Fit

)

CH2 H910 H811 H1 H56 H1 OH

00

05

10

15

20

25

30

35

40

T2 m

s

T= 110degC

T= 90degC

T= 80degC

T= 70degC

CH2 H910 H811 H1

0

2

4

6

8

10

12

T2 m

s

T= 110degC

T= 90degC

T= 80degC

T= 70degC

Short T2(1H) Long T2(1H)

a) b)

c)

11

Table S 1 ndash 1H chemical shifts of G-C181-OH in solution and in the solid-state Solid-state spectra are 988

recorded at a MAS frequency of 10 kHz in DMSO-d6 broad peak 989

990

δH ppm

Group Ndeg Liquid

(MeOD-d4) Solid

(T= 110degC) Solid

(T= 70degC)

CH2 H1 354 357

CH2 H2 153 156

CH2 H3-712-14 132 135

CH2 H811 203 203

CH H910 534 534

CH2 H15 162 164

CH2 H16 388 383

CH H1 424 432

CH H2 316 335

CH H3 334 335

CH H4 329 335

CH H5 354 351 357

CH2 H6 366 383

OH OH1 432

45 463

OH OH2 492

OH OH3 490

OH OH4 487

OH OH6 445

991

992

H1H9H10H1rsquo

H6rsquo

OH2rsquo

OH1

12

Detailed discussion about the NOESY and DQ-SQ 2D 1H-1H ssNMR experiments presented in 993

Figure 9 994

995

2D 1H-1H ssNMR experiments are a unique tool to explore through-space proximities 996

between 1H nuclei ssNMR offers a large number of possibilities each adapted to a given 997

system The advantage of 2D maps with respect to a dynamics approach is their direct 998

readability in clear two nuclei which exchange their magnetization through-space (whichever 999

mechanism may be responsible for this exchange) will display a specific cross-peak However 1000

the way cross-peaks are presented and interpreted strictly depends on the nature of the pulse 1001

program which in turn is chosen according to the physical nature of the sample 1002

1D temperature-dependent 1H NMR experiments performed on G-C181-OH (Figure 1003

4a in the main text) indicate sharp peaks at 90degC and broad peaks at 70degC Narrow peaks 1004

indicate fast andor large-amplitude molecular tumbling averaging the strong 1H-1H dipolar 1005

interactions This excludes the use of pulse sequences like DQ-SQ which is not efficient under 1006

these conditions For this reason we employ spin-diffusion-based pulse sequences like 1007

NOESY which is more adapted for samples with high molecular mobility When mixing times 1008

are chosen to be short (lt 1 s) one generally explores through-space proximities within about 5 1009

10 Aring On the contrary when molecular motions are restricted dipolar coupling becomes 1010

significant 1H NMR peaks are broader and pulse sequences like DQ-SQ can now be used 1011

with a spatial resolution of less than 5 Aring 1012

Whenever possible DQ-SQ experiments are preferred over spin diffusion-based pulse 1013

sequences because they are more selective in particular for likewise interactions which are 1014

not discriminated in NOESY-type experiments This is illustrated by the typical interaction 1015

schemes for NOESY (top Figure 9a) and DQ-SQ (top Figure 9b) For two given A and B 1016

protons 2D NOESY maps will display two on-diagonal and two square-shaped symmetrical 1017

off-diagonal cross-peaks Off-diagonal cross-peaks illustrate the mutual A-B and B-A 1018

interaction but the on-diagonal peaks are systematic and do not prove any A-A or B-B 1019

interaction Similarly to NOESY 2D DQ-SQ maps display off-diagonal peaks (symmetric 1020

across the diagonal top Figure 9b) upon A-B and B-A interaction but in contrast they will 1021

display on-diagonal peaks only in the presence of self-interaction In the tutorial example of 1022

Figure 9b top the on-diagonal A-A peak indicates mutual A-A interaction while lack of on-1023

diagonal B-B peak evidences the absence of B-B proximities Combining NOESY and DQ-SQ 1024

experiments is capital to study the local arrangement of G-C181-OH and its evolution with 1025

temperature 1026

13

At T= 90degC the H1 and H5rsquo contributions are very close but possible to discriminate 1027

them (δ(H1)= 354 ppm δ(H5rsquo)= 351 ppm) according to 1H NMR chemical shifts in Table S 1028

1 2D 1H-1H NOESY in Figure 9a also discriminates between the two components attributed 1029

on the basis of the interactions with their mutual neighbors (green H1-H2 and H5rsquo-H1rsquo squares 1030

in Figure 9a) On this basis 2D NOESY does not show a direct H1rsquo-H1 correlation (blue red-1031

crossed) but rather an extensive network of intra-sugar interactions These features exclude 1032

suggest intra- and inter-sugar interactions and they exclude head-tail arrangements as a major 1033

molecular configuration at 90degC 1034

At 70degC the 1H NMR peaks are broader indicating a reduction of the T2 (1H) relaxation 1035

times andor residual homonuclear dipolar interactions excluding the use of NOESY In this 1036

case 1H-1H DQ-SQ correlation experiments are best suited Figure 9b presents the 1H-1H DQ-1037

SQ experiment on G-C181-OH at 70degC Peaks are broader than in NOESY and attribution may 1038

be cumbersome However the experiment is performed at 700 MHz and MAS= 22 kHz which 1039

improves spectral resolution with respect to the same experiment performed at 300 MHz and 1040

MAS= 10 kHz The auto-peak at δH= 356 ppm in the SQ dimension is attributed to H1-H1 spin 1041

pairs mainly resulting from intra CH2 interaction while the DQ cross-peaks at δH= 156-356 1042

ppm and δH= 432-356 ppm respectively origin from H2-H1 and H1rsquo-H1 interactions (Figure 1043

9b) DQ-SQ map does not show any auto-peak corresponding to H1rsquo-H1rsquo or Hxrsquo-Hxrsquo (x= 2rsquo 1044

3rsquo 4rsquo) interactions (red-crossed blue Figure 9b) thus excluding any proximity (lt 5 Aring) between 1045

close glucose neighbors On the contrary DQ-SQ maps (recorded both at 700 MHz Figure 9b 1046

or at 300 MHz not shown) systematically show DQ cross-peaks which can only be attributed 1047

to H1rsquo-H1 interactions (δH= 432-356) The presence of these H1rsquo-H1 interactions can only be 1048

explained by a head-tail arrangement of the G-C181-OH molecules as illustrated below the 1049

DQ-SQ map in Figure 9b 1050

24

Two related questions are unanswered at the moment What actually drives nanotube 582

formation and is the organization of G-C181-OH within the vesicles and tubes the same Lipid 583

nanotubes have been studied for at least three decades over which broad experimental and 584

theoretical insight has been generated Despite this fact the actual understanding is still matter 585

of debate27 Micelle-to-nanotube and vesicle-to-nanotube phase transitions are generally 586

reported for single amphiphiles below a given Tm and intermediate phase composed of twisted 587

and flat fibers is concomitantly reported Initially thought to be a chirality-driven phenomenon 588

recent studies propose a more general mechanism involving crystallization and symmetry 589

breaking in the lipid packing driven by optimal packing27 If twisted ribbons were observed 590

before for a number of microbial glycolipid bolaamphiphiles44 they are not found for G-C181-591

OH This fact as well as the large body of cryo-TEM images and the observation of nanotubes 592

above Tm1 rather pledge in favour of a direct topological change from positive (vesicles) to 593

zero (nanotubes) Gaussian curvatures In this case vesicle-to-nanotube transitions could be 594

driven by external forces56307813ndash162829 595

Considering the absence of external forces in the present system there are only two 596

possibilities left spontaneous budding of the membrane or nanotube pulling after inter-vesicle 597

collisions due to Brownian motion Both are possible as inclusions composed of a different 598

arrangement of the same molecule in the membrane continuum can theoretically explain 599

spontaneous nanotube formation in an isotropic medium672223629ndash1218ndash21 However the second 600

one is probably more likely to prevail because it could explain the large number of nanotubes 601

connecting two distant vesicles Interestingly similar events occur for any phospholipid-based 602

vesicular solution but nanotubes are extremely rare unless at least two lipids of different 603

molecular structure are mixed together The spectroscopic 1H and 13C and relaxation (1H) 604

ssNMR data even if collected on a dry powder demonstrate that the coexistence of two G-605

C181-OH configurations involving both the sugar headgroup and fatty alcohol chain are 606

possible for this molecule As a matter of fact this is not so surprising as more than one energy 607

minima each corresponding to a given conformation are known for glucose its non-reducing 608

derivative methylglucose and carbohydrates in general78ndash80 609

The presence of two configurations could explain the coexistence between two 610

structures of different curvature but the data presented so far do not explain the origin of these 611

two configurations In other words what drives the change in the local structure of the sugar 612

headgroup and fatty alcohol within the membrane 613

The association of known self-assembled structures to spectroscopic 13C ssNMR data is 614

known for years in particular for gluconamides lipids having a linear sugar headgroup81ndash84 It 615

25

was found that gauche effects in some cases up to the γ position in the sugar chain induce 616

upfield chemical shifts while anti and trans conformations result in downfield shifts Each 617

structure (monolayer crystal bilayer crystal or micellar fiber) was then characterized by a given 618

organization of the gluconamide and a characteristic 13C ssNMR fingerprint which could be 619

explained by gauche γ-gauche anti and trans conformations of the headgroup Similar 620

structural-spectroscopic correlations have been collected over the years for much simpler cyclic 621

sugars85 whereas pyranoses can undergo important conformational changes (chair half-chair 622

boat skew) when the appropriate energy barriers are crossed78ndash8086 A quantitative 623

conformational study based on the 13C chemical shift variations shown in Table 1 would be 624

certainly helpful but very delicate in the context of the present state of the art because a 625

rigorous approach requires at least one known crystal structure8182 which is currently missing 626

for this compound A qualitative approach will then be adopted 627

G-C181-OH is a bolaform amphiphile and flip-flop could be a possible mechanism 628

involved in the nanotube stabilization whereas head-tail interactions have long been considered 629

as important in nanotube-forming bolaform amphiphiles6063 From mere considerations based 630

on T2 (1H) relaxation arguments one could exclude head-tail configurations if the R-CH2OH 631

endgroup (H1) was involved in a direct interaction with the sugar group its corresponding T2 632

would probably be comparable to the sugarrsquos one (OH and H1rsquo) but this does not seem to be 633

the case However this hypothesis would be based on the comparison of the relaxation from 634

CH and CH2 protons additionally involved in different kind of motions Besides the relaxation 635

of H1 is influenced by the strong overlap between H1 and H5rsquo For these reasons relaxation 636

arguments are not solid enough and will not be used 637

We then address this question by exploring intermolecular interactions in a direct 638

manner using 2D 1H-1H homonuclear ssNMR correlation spectroscopy At 90degC in the fast-639

tumbling domain of molecular motion (narrow 1H NMR peaks Figure 6a) and corresponding 640

to the vesicle-rich region spin diffusion (NOESY) experiments are preferred over dipolar 641

coupling based experiments because such interaction is averaged out by molecular motion 642

However when the molecular motion is more constrained in the domain of nanotube stability 643

at 70degC (broad 1H spectra Figure 6a) Double-Quantum Single-Quantum (DQ-SQ) correlation 644

experiments are best suited to probe interaction between close molecular neighbors These 645

experiments are reported in Figure 9 and the related interpretation is only briefly reported 646

below while a more detailed discussion can be found on Page 12 in the Supporting Information 647

26

648

Figure 9 ndash a) 2D 1H-1H NOESY NMR experiment performed at MAS= 10 kHz (4 mm probe) on a Bruker 649

300 MHz Avance III HD spectrometer using 300 ms of mixing time Sample temperature is regulated to 650

90degC reached by cooling from 110degC b) 2D 1H-1H DQ-SQ (BAck-to-BAck BABA pulse sequence used for 651

the DQ excitation scheme) NMR experiment performed at MAS= 22 kHz (25 mm probe) on a Bruker 700 652

MHz Avance III spectrometer using a DQ excitation time equal to 2 rotor periods Sample temperature is 653

regulated to 70degC reached by cooling from 110degC The typical schemes illustrating how to interpret these 654

2D experiments are reported above each experimental map Zooms of the sugar region and configurational 655

interpretation are given below each 2D map More information are given in the text below 656

H16-H16

H1-H1

H9-H10

H1rsquo-H1

H1rsquoH910 H16

H1

H1rsquo-H1rsquo

Hxrsquo-Hxrsquo

2

(pp

m)

(ppm)

315

811910 1rsquo

2rsquo2

3rsquo 4rsquo6rsquo

5rsquo

1rsquo2910

asymp asymp asympHDO MeOD Acetone

1

315

811 9101rsquo

2rsquo2

3rsquo4rsquo6rsquo

5rsquo

1rsquo2 9 10

asympasympasympHDOMeODAcetone

H1rsquo-H1

1rsquo

1

(p

pm

)

(ppm)

H1-H2

H1rsquo-H1

H1rsquo

H910

H1H5rsquo

H2

H1rsquo-H5rsquo

315

811910 1rsquo

2rsquo2

3rsquo 4rsquo6rsquo

5rsquo

1rsquo2910

asymp asymp asympHDO MeOD Acetone

1

H2rsquoH3rsquoH4rsquo

H2rsquoH3rsquoH4rsquo

X= 2rsquo3rsquo4rsquo

NOESY - T= 90degC

1H

1H

1H

1H

DQ-SQ - T= 70degC

a) b)

2

(pp

m)

(p

pm

)

(ppm)

A B

AhellipB

AhellipA

BhellipB

NOESY

(ppm)

A B

AhellipB

DQ-SQ

Single-quantum dimension

Do

ub

le-q

ua

ntu

m d

imen

sio

n

AhellipA

BhellipB

BhellipA

BhellipA

315

811910 1rsquo

2rsquo2

3rsquo 4rsquo6rsquo

5rsquo

1rsquo2910

asymp asymp asympHDO MeOD Acetone

1

lt 5 Aring

lt 5 Aring

H1-H2

H811

H15

H16

H6rsquo

H1rsquo

H1

H5rsquo

H2

H1-H2

H1rsquo-H1

H1rsquo-H5rsquo

H1rsquo

H1 H5rsquo

H1rsquo

H1

H5rsquo

H1H2rsquo H3rsquo H4rsquo

H1-H1

H1rsquo-H1

Hxrsquo-HxrsquoX= 2rsquo3rsquo4rsquo

H1-H2

H1rsquo

Zoom sugar region 28 ndash 45 ppmNOESY DQ-SQ

27

Head-tail or head-head configurations could be probed by the existence of H1rsquo-H1 657

interactions whereas H1 (δH= 356 ppm) is associated to the glucolipid tail and H1rsquo (δH= 432 658

ppm) to the sugar headgroup These atoms are more than 15 Aring apart and a direct interaction 659

could not be observed in both NOESY and DQ-SQ experiments NOESY experiments (Figure 660

9a) show that H1rsquo-H1 correlations are not observed Meanwhile clear-cut direct H1rsquo-H1 661

interactions are indeed observed are detected at T= 70degC (DQ cross-peaks related to H1rsquo-H1 662

correlation in DQ-SQ experiment shown in Figure 9b) Interestingly the absence of auto peaks 663

at δH= 432 ppm and δH= 335 ppm in the SQ dimension excludes H1rsquo-H1rsquo and Hxrsquo-Hxrsquo (x= 2 664

3 4) interactions respectively strengthening the idea that sugar headgroups from two nearest-665

neighbor G-C181-OH molecules are not adjacent in the nanotube-vesicle region 666

Homonuclear 2D ssNMR supports the idea according to which G-C181-OH can 667

undergo molecular flip-flops and a head-tail arrangement in the dry product Extrapolating these 668

conclusions in the lower concentration range lack of H1rsquo-H1 proximity at T= 90degC suggests a 669

head-head configuration in the vesicle-rich region and head-tail configuration within the 670

nanotubes It is however still unclear whether flip-flops are driven by spontaneous temperature-671

induced conformational changes in the glucolipid packing stimulating inclusions within the 672

membrane itself672223629ndash1218ndash21 or by defects generated by inter-vesicle collisions In fact a 673

combination of both is not unlikely Nanotubes are not stable when temperature is 20 to 30 674

degrees above Tm1 indicating that inter-vesicle collisions alone cannot explain the unusual 675

phenomenon In an energy landscape approach87 the head-head configuration is certainly 676

favored at high temperature while both head-head and head-tail configurations could be 677

favored in the mid-temperature region corresponding to the range between Tm1 and Tm2 in the 678

dry powder Both statistical and defect-induced flip-flop events contribute to form high mean 679

zero Gaussian curvature morphologies (nanotubes) for which the head-tail conformation 680

limiting the proximity between bulky sugar headgroups is probably more adapted (Figure 8c) 681

The flip-flop interpretation now reasonably explains the strong differences in the sugar and fatty 682

alcohol conformations deduced by the multiple NMR signals attributed to chemically-683

equivalent groups The existence of various low-energy conformations each having specific 684

NMR signature for sugars in solution78ndash8086 and solid-state81ndash84 further supports our 685

interpretation 686

One last point should be clarified Why between Tm1 and Tm2 are two molecular 687

environments able to coexist instead of a single one of average rigidity as found in the case of 688

ldquoripplerdquo phases50ndash54 We believe that this question can be answered based on the NMR results 689

the relatively short T2 values of the H1rsquo and OH protons shows the constrained mobility of the 690

28

sugar headgroup despite a certain flexibility of the chain (long T2 component) This trend may 691

suggest restrictions in translational mobility probably due to the sugar-OH hydrogen bonding 692

(Figure 8c) meaning that lipids are not able to diffuse within the membrane thus keeping two 693

different configurations stable 694

695

Conclusions 696

Topologically-connected nanotube-vesicle lipidic systems are commonly observed for 697

self-assembled membranes composed of at least two amphiphiles and spontaneously driven by 698

anisotropic inclusions andor by external directional forces This work shows that strong 699

differences in membrane curvature can spontaneously occur in single-lipid systems without 700

external anisotropic forces Coexistence and interconnection between nanotubes and vesicles 701

are evidenced via cryo-TEM and fluorescence confocal microscopy in a single-phase region at 702

temperature above a transition temperature Tm1= 483degC determined by DSC on the hydrated 703

sample but below the second transition temperature Tm2 (957degC) determined by DSC on the 704

dry powder X-ray diffraction (WAXS) also shows the coexistence of two lattice periods one 705

attributed to an inter-membrane repeating distance and the other possibly related to the intra-706

nanotube glucolipid arrangement Labelling the systems with Laurdan a fluorophore of which 707

the emission strongly varies with the local membrane rigidity excludes the presence of rigid 708

micro-separated phases (eg Lβ) 709

1H and 13C ssNMR despite the fact that it can only be performed on the dry powder 710

confirm the presence of one main fluid G-C181-OH environment above Tm2 Below Tm2 the 711

splitting of some 13C peaks and the coexistence of short and long 1H relaxation components 712

demonstrate that G-C181-OH can simultaneously pack into structurally and dynamically 713

inequivalent environments Combining advanced 2D 1H-1H correlation experiments based on 714

spin diffusion or double-quanta excitation suggests that vesicles could contain a preferred head-715

head configuration of the bolaform lipid while nanotubes a preferred head-tail assembly 716

The presence of two environments composed of the same molecule characterized by 717

different molecular configurations and rigidity fulfills the theoretical conditions necessary to 718

spontaneously nucleate a membrane inclusion of non-zero average orientation described as 719

necessary to observe topologically-connected nanotube-vesicle structures in the absence of 720

external forces This hypothesis could also be enriched by the possibility of nanotubes to be 721

pulled away between two vesicles after a collision event 722

This work expands the actual theoretical and experimental framework on lipid networks 723

stabilized by a different radius of curvature It could also motivate the synthesis or isolation 724

29

of new lipids providing a higher level of 3D architectural control in soft networks possibly in 725

the range of room temperature 726

727

Acknowledgements 728

Dr Andrea Lassenberger and Dr Sylvain Preacutevost at Institut Laue Langevin (ILL Grenoble 729

France) are kindly acknowledged for their assistance on the SANS experiment Confocal 730

images were performed at the Institut de Biologie Paris Seine (IBPS) imaging facility Jean-731

Franccedilois Gilles (Imaging Core Facility Institut de Biologie Paris Seine (IBPS) CNRS 732

Sorbonne Universiteacute Paris France) is kindly acknowledged for his assistance on the confocal 733

microscopy imaging Dr Guillaume Laurent (LCMCP Sorbonne Universiteacute Paris France) is 734

kindly acknowledged for helpful discussions The French region Ile-de-France SESAME 735

program is acknowledged for financial support (700 MHz NMR 736

spectrometer) Dr Olivier Diat (ICSM Marcoule France) Prof eronikaIglič and Prof V Aleš737

onsul discussi Slovenia) are kindly acknowledged for helpfUniversity of LjubljanaIglič (-Kralj 738

739

Financial support 740

WAXS experiments have been supported by Soleil Light Source Saint Aubin France proposal 741

Ndeg 20201747 SANS experiments have been supported by ILL proposal Ndeg9-13-778 742

743

References 744

1 D Kabaso N Bobrovska W Goacuteźdź E Gongadze V Kralj-Iglič R Zorec and A 745

Iglič Bioelectrochemistry 2012 87 204ndash210 746

2 C R Safinya U Raviv D J Needleman A Zidovska M C Choi M A Ojeda-747

Lopez K K Ewert Y Li H P Miller J Quispe B Carragher C S Potter M W 748

Kim S C Feinstein and L Wilson Adv Mater 2011 23 2260ndash2270 749

3 P Veranič M Lokar G J Schuumltz J Weghuber S Wieser H Haumlgerstrand V Kralj-750

Iglič and A Iglič Biophys J 2008 95 4416ndash4425 751

4 A Rustom R Saffrich I Markovic P Walther and H-H Gerdes Science (80- ) 752

2004 303 1007ndash1010 753

5 M I Angelova S Soleacuteau P Meacuteleacuteard F Faucon and P Bothorel Prog Colloid 754

Polym Sci 1992 89 127ndash131 755

6 L Mathivet S Cribier and P F Devaux Biophys J 1996 70 1112ndash1121 756

7 V Kralj-Iglič G Gomišček J Majhenc V Arrigler and S Svetina Colloids Surfaces 757

A Physicochem Eng Asp 2001 181 315ndash318 758

30

8 V Kralj-Iglič A Iglič G Gomišček F Sevšek V Arrigler and H Haumlgerstrand J 759

Phys A Math Gen 2002 35 1533ndash1549 760

9 M Drab D Stopar V Kralj-Iglič and A Iglič Cells 2019 8 626 761

10 A Zidovska K K Ewert J Quispe B Carragher C S Potter and C R Safinya 762

Langmuir 2009 25 2979ndash2985 763

11 A Zidovska K K Ewert J Quispe B Carragher C S Potter and C R Safinya Soft 764

Matter 2011 7 8363ndash8369 765

12 A Zidovska K K Ewert J Quispe B Carragher C S Potter and C R Safinya 766

Biochim Biophys Acta - Biomembr 2009 1788 1869ndash1876 767

13 M Karlsson K Sott M Davidson A S Cans P Linderholm D Chiu and O Orwar 768

Proc Natl Acad Sci U S A 2002 99 11573ndash11578 769

14 A Karlsson R Karlsson M Karlsson A-S Cans A Stromberg F Ryttsen and O 770

Orwar Nature 2001 409 150ndash152 771

15 M Karlsson K Sott A S Cans A Karlsson R Karlsson and O Orwar Langmuir 772

2001 17 6754ndash6758 773

16 E Evans H Bowman A Leung D Needham and D Tirrell Science (80- ) 1996 774

273 933ndash935 775

17 U Seifert Configurations of fluid membranes and vesicles 1997 vol 46 776

18 J Kaumls and E Sackmann Biophys J 1991 60 825ndash844 777

19 N Bobrovska W Goacuteźdź V Kralj-Iglič and A Iglič PLoS One 2013 8 e73941 778

20 A Iglič B Babnik K Bohinc M Fošnarič H Haumlgerstrand and V Kralj-Iglič J 779

Biomech 2007 40 579ndash585 780

21 K Bohinc T Slivnik A Iglič M Brumen and V Kralj-Iglič J Phys Chem C 2007 781

111 9709ndash9718 782

22 A Iglič H Haumlgerstrand P Veranič A Plemenitaš and V Kralj-Iglič J Theor Biol 783

2006 240 368ndash373 784

23 V Kralj-Iglič and P Veranič Adv Planar Lipid Bilayers Liposomes 2006 5 129ndash785

149 786

24 A Iglic and V Kralj-Iglic Adv Planar Lipid Bilayers Liposomes 2006 4 253ndash279 787

25 S Penič L Mesarec M Fošnarič L Mroacutewczyńska H Haumlgerstrand V Kralj-Iglič and 788

A Iglič Front Phys 2020 8 342 789

26 P Yager and P E Schoen Mol Cryst Liq Cryst 1984 106 371ndash381 790

27 T G Barclay K Constantopoulos and J Matisons Chem Rev 2014 114 10217ndash791

10291 792

31

28 B Božič G Gomišček V Kralj-Iglič S Svetina and B Žekš Eur Biophys J 2002 793

31 487ndash496 794

29 A Iglič H Haumlgerstrand M Bobrowska-Haumlgerstrand V Arrigler and V Kralj-Iglič 795

Phys Lett A 2003 310 493ndash497 796

30 R Bar-Ziv and E Moses Phys Rev Lett 1994 73 1392ndash1395 797

31 S Chiruvolu H E Warriner E Naranjo S H Idziak J O Raumldler R J Plano J A 798

Zasadzinski and C R Safinya Science (80- ) 1994 266 1222ndash1225 799

32 V Kralj-Iglič A Iglič H Haumlgerstrand and P Peterlin Phys Rev E 2000 61 4230ndash800

4234 801

33 L Mesarec M Drab S Penič V Kralj-Iglič and A Iglič Int J Mol Sci 2021 22 802

2348 803

34 L Van Renterghem S L K W Roelants N Baccile K Uyttersprot M C Taelman 804

B Everaert S Mincke S Ledegen S Debrouwer K Scholtens C V Stevens W 805

Soetaert L Van Renterghem S L K W Roelants N Baccile K Uytersprot M C 806

Taelman B Everaert S Mincke S Ledegen S Debrouwer K Scholtens C V 807

Stevens and W Soetaert Biotechnol Bioeng 2018 115 1195ndash1206 808

35 J Clark and F Deswarte Introduction to Chemicals from Biomass Second Edition 809

John Wiley amp Sons 2015 810

36 K V Sajna R Houmlfer R K Sukumaran L D Gottumukkala and A Pandey in 811

Industrial Biorefineries and White Biotechnology eds A Pandey R Houmlfer M 812

Taherzadeh K M Nampoothiri and C Larroche Amsterdam Oxford Waltham 813

Elsevier 2015 pp 499ndash521 814

37 R Marchant and I M Banat Trends Biotechnol 2012 30 558ndash565 815

38 K Hill and C LeHen-Ferrenbach in Sugar-Based Surfactants Fundamentals and 816

Applications ed C C Ruiz Boca Raton CRC Press 2009 pp 1ndash20 817

39 L Rodrigues I M Banat J Teixeira and R Oliveira J Antimicrob Chemother 818

2006 57 609ndash618 819

40 J D Desai and I M Banat Microbiol Mol Biol Rev 1997 61 47ndash64 820

41 S L K W Roelants K Ciesielska S L De Maeseneire H Moens B Everaert S 821

Verweire Q Denon B Vanlerberghe I N A Van Bogaert P Van der Meeren B 822

Devreese and W Soetaert Biotechnol Bioeng 2016 113 550ndash559 823

42 N Baccile M Selmane P Le Griel S Preacutevost J Perez C V Stevens E Delbeke S 824

Zibek M Guenther W Soetaert I N A Van Bogaert and S Roelants Langmuir 825

2016 32 6343ndash6359 826

32

43 N Baccile A-S Cuvier S Preacutevost C V Stevens E Delbeke J Berton W Soetaert 827

I N A Van Bogaert and S Roelants Langmuir 2016 32 10881ndash10894 828

44 N Baccile C Seyrig A Poirier S A Castro S L K W Roelants and S Abel Green 829

Chem 2021 23 3842ndash3944 830

45 A Salis L Cappai C Carucci D F Parsons and M Monduzzi J Phys Chem Lett 831

2020 11 6805ndash6811 832

46 J Schindelin I Arganda-Carreras E Frise V Kaynig M Longair T Pietzsch S 833

Preibisch C Rueden S Saalfeld B Schmid J-Y Tinevez D J White V 834

Hartenstein K Eliceiri P Tomancak and A Cardona Nat Methods 2012 9 676ndash835

682 836

47 T Parasassi and E Gratton J Fluoresc 1995 5 59ndash69 837

48 M M Dodes Traian F L Gonzaacutelez Flecha and V Levi J Lipid Res 2012 53 609ndash838

616 839

49 D M Owen C Rentero A Magenau A Abu-Siniyeh and K Gaus Nat Protoc 840

2012 7 24ndash35 841

50 W Stillwell in An Introduction to Biological Membranes Academic Press 2016 pp 842

135ndash180 843

51 W Stillwell in An Introduction to Biological Membranes Academic Press 2016 pp 844

181ndash220 845

52 B Tenchov Chem Phys Lipids 1991 57 165ndash177 846

53 R Koynova and M Caffrey Biochim Biophys Acta 1998 1376 91ndash145 847

54 F Schmid S Dolezel O Lenz and S Meinhardt J Phys Conf Ser 2014 487 848

012004 849

55 S Meinhardt R L C Vink and F Schmid Proc Natl Acad Sci U S A 2013 110 850

4476ndash4481 851

56 O Glatter and O Kratky Small Angle X-ray Scattering Academic Press London 852

1982 853

57 B A Cunningham A D Brown D H Wolfe W P Williams and A Brain Phys 854

Rev E 1998 58 3662ndash3672 855

58 M Nadler A Steiner T Dvir O Szekely P Szekely A Ginsburg R Asor R Resh 856

C Tamburu M Peres and U Raviv Soft Matter 2011 7 1512ndash1523 857

59 C Tanford The hydrophobic effect Formation of micelles and biological membranes 858

John Wiley amp Sons Inc New York 1973 859

60 M Masuda and T Shimizu Langmuir 2004 20 5969ndash5977 860

33

61 A Zidovska K K Ewert J Quispe B Carragher C S Potter and C R Safinya 861

Block Liposomes Vesicles of Charged Lipids with Distinctly Shaped Nanoscale 862

Sphere- Pear- Tube- or Rod-Segments Elsevier Inc 1st edn 2009 vol 465 863

62 A Iglic V Kralj-Iglic B Božič M Bobrowska-Haumlgerstrand B Isomaa and H 864

Haumlgerstrand Bioelectrochemistry 2000 52 203ndash211 865

63 T Shimizu M Masuda and H Minamikawa Chem Rev 2005 105 1401ndash1444 866

64 J V Selinger and J M Schnur Phys Rev Lett 1993 71 4091ndash4094 867

65 J M Schnur Science (80- ) 1993 262 1669ndash1676 868

66 B Tenchov R Koynova and G Rapp Biophys J 2001 80 1873ndash1890 869

67 R Koynova and B Tenchov Ser Biomech 2017 31 3ndash19 870

68 S J Angyal and R Le Fur Carbohydr Res 1984 126 15ndash26 871

69 J K Whitesell and M A Minton J Am Chem Soc 1987 109 225ndash228 872

70 T Takayama S Ando and I Ando J Mol Struct 1990 220 245ndash250 873

71 K Gawrisch N V Eldho and I V Polozov Chem Phys Lipids 2002 116 135ndash151 874

72 F Tielens C Gervais J F Lambert F Mauri and D Costa Chem Mater 2008 20 875

3336ndash3344 876

73 G A Jeffrey and Y Yeon Acta Crystallogr Sect B 1986 42 410ndash413 877

74 B Berglund and R W Vaughan J Chem Phys 1980 73 2037ndash2043 878

75 C Gervais C Coelho T Azaiumls J Maquet G Laurent F Pourpoint C Bonhomme 879

P Florian B Alonso G Guerrero P H Mutin and F Mauri J Magn Reson 2007 880

187 131ndash140 881

76 C Akita T Kawaguchi F Kaneko H Yamamoto and M Suzuki J Phys Chem B 882

2004 108 4862ndash4868 883

77 R N Purusottam L Seacutenicourt J J Lacapegravere and P Tekely Biochim Biophys Acta - 884

Biomembr 2015 1848 3134ndash3139 885

78 X Biarneacutes A Ardegravevol A Planas C Rovira A Laio and M Parrinello J Am Chem 886

Soc 2007 129 10686ndash10693 887

79 L M J Kroon‐Batenburg and J Kroon Biopolymers 1990 29 1243ndash1248 888

80 K N Kirschner and R J Woods Proc Natl Acad Sci U S A 2001 98 10541ndash889

10545 890

81 S Svenson B Kirste and J H Fuhrhop J Am Chem Soc 1994 116 11969ndash11975 891

82 I Sack S Macholl J H Fuhrhop and G Buntkowsky Phys Chem Chem Phys 892

2000 2 1781ndash1788 893

83 S Svenson J Koening and J H Fuhrhop J Phys Chem 1994 98 1022ndash1028 894

34

84 S Svenson A Schaefer and J H Fuhrhop J Chem Soc Perkin Trans 2 1994 2 895

1023ndash1028 896

85 K Bock and C Pedersen Adv Carbohydr Chem Biochem 1983 41 27ndash66 897

86 T B Grindley Structure and Conformation of Carbohydrates Springer Berlin 898

Heidelberg 2008 899

87 F Tantakitti J Boekhoven X Wang R V Kazantsev T Yu J Li E Zhuang R 900

Zandi J H Ortony C J Newcomb L C Palmer G S Shekhawat M O De La 901

Cruz G C Schatz and S I Stupp Nat Mater 2016 15 469ndash476 902

903

1

Supporting Information 904

905

Topological connection between vesicles and nanotubes in single-906

component lipid membranes driven by head-tail interactions 907

908

Niki Baccilea Ceacutedric Lorthioira Abdoul Aziz Baa Patrick Le Griela Cristina Coelhob Javier 909

Perezc Wim Soetaertd Sophie L K W Roelantsd 910

911

a Sorbonne Universiteacute Centre National de la Recherche Scientifique Laboratoire de Chimie de 912

la Matiegravere Condenseacutee de Paris LCMCP F-75005 Paris France 913

b Institut des Mateacuteriaux de Paris Centre Sorbonne Universiteacute Paris France 914

c Synchrotron Soleil LrsquoOrme des Merisiers Saint-Aubin BP48 91192 Gif-sur-Yvette Cedex 915

France 916

d InBio Department of Biotechnology Ghent University Ghent Belgium 917

918

919

Corresponding author 920

Dr Niki Baccile 921

E-mail address nikibaccilesorbonne-universitefr 922

Phone +33 1 44 27 56 77 923

924

925

926

2

a) 927

Parameter Method

Dry Matter (DM ) 996 Infrared balance 105degC

Glucose () 012 HPLC-Metacarb

Glycerol () nd HPLC-Metacarb

Free Fatty Acid content () lt001 Internal method BBEPP

GC with FID Detector

Oil () lt001 Internal method BBEPP

GC with FID Detector

Moulds (CFUg) lt10 3M-Nordval ndeg16 B

Yeasts (CFUg) lt10 3M-Nordval ndeg16 B

Anaerobic count (CFUg) lt10 SP-VG M005 B

Protein () tbd BCA protein assay

928

Figure S 1 - a) HPLC-ELSD chromatogram and composition table 929

930

931

932

933

3

b) 934

935

Group Ndeg δ ppm

CH2 1 354

CH2 2 153

CH2 3-712-14 132

CH2 811 203

CH 910 534

CH2 15 162

CH2 16 388

CH 1 424

CH 2 316

CH 3 334

CH 4 329

CH 5 354

CH2 6 366

Figure S 1 (next) ndash b) 1H NMR spectrum (MeOD-d4) recorded on the purified G-C181-OH compounds 936

and related assignment 937

938

2

811910 1rsquo

5rsquo1

3rsquo4rsquo

1rsquo1910

asymp asymp asympHDO MeOD Acetone

6rsquo2rsquo

1615

3-712-14

4

939

940

Figure S 2 ndash Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 130degC 941

5

942

943

Figure S 3 - Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 90degC 944

945

90degC

1

22

6

946

947

948

949

Figure S 4 ndash Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 70degC 950

951

21

11

1

2

5

3

4

45

5

6

7

7

7

952

953

954

Figure S 5 ndash Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 25degC 955

956

957

1

1 1

8

958

Figure S 6 ndash a) SANS experiment recorded at 90degC for a 5 mgmL G-C181-OH aqueous solution b-c) 959

WAXS experiments recorded on aqueous solutions of G-C181-OH at b) 5 mgmL and c) 25 mgmL d) 960

Structural interpretation of the repeating distances pointed out in c) 961

962

963

964

001 01 1

q Aring-1

25degC

68degC

88degC

111degC

01 1q Aring

-1

02 04 06 08 1q Aring

-1

001 01

10-1

100

101

Inte

nsity

cm

-1

q Aring-1

-2

T= 90degC

SANS WAXS WAXS

a) b) c)

dv

dL d= 144 Aring-1

di

ddH2O

dt

dvdL

dtdi d

d)

9

965

966

Figure S 7 ndash ssNMR MAS (10 kHz) experiments recorded on a G-C181-OH dry powder a) 1H spectra 967

recorded from 110degC to 70degC b) 13C single pulse (SP) spectra recorded with high power 1H decoupling from 968

110degC to 70degC c) 13C cross polarization (CP) spectra recorded at 110degC and contact time tc of 50 micros and 969

3000 micros d) 13C CP spectra recorded at 70degC and for 20 lt tc micros lt 3000 970

971

1

811

910

CH2 CH2

1rsquo

6rsquo

8 6 4 2 0 ppm

110

90

80

70

910

1rsquo

6rsquo16

1

2rsquo3rsquo4rsquo

811

OH

45 40 35 30 25 20 15

ppm

200 180 160 140 120 100 80 60 40 20 0

ppm140 120 100 80

ppm

CH2-gauche

3004

110

90

80

70

65

60

50

25

C91013003

C9 - 12869

CH2-all-trans

3198

C1rsquo10335

a) b)

1H

T degCT degC

13C Single Pulse MAS

A

B

A

B

70degC70degC

5rsquo

215

200 180 160 140 120 100 80 60 40 20 0

ppm

13C CPMAST= 110degC

200 180 160 140 120 100 80 60 40 20 0

ppm

132 131 130 129 128 127 126

ppm

106 105 104 103 102

ppm

13C CPMAST= 70degC

50

tc micros

3000

20

tc micros

50 500 1000 3000

tc micros

C102

13024 C910113003

C92

12869

C1rsquo213082

C1rsquo110443

A B

106 105 104 103 102

ppm

3000

1000

500

50

20

AB

c)

d)

10

972

Figure S 8 ndash a) Short and d) long T2 components displayed by the 1H transverse relaxation signal c) 1H 973

transverse relaxation signals recorded on a G-C181-OH dry powder from 110degC to 70degC The area of each 974

given resonance is plotted (log-lin) as a function of the echo time (CPMG pulse sequence) for each 975

temperature and fitted using the equation given above T2s and T2l indicate the short and long T2 relaxation 976

times respectively obtained by describing the 1H transverse relaxation signal using a biexponential decay 977

For H1rsquo H5rsquo and H6rsquo the 1H relaxation function could be described by a monoexponential decay 978

- Protons from the sugar group (H1rsquo H5rsquo H6rsquo) show a single T2 relaxation component (Figure 979

6c and Figure S 8) while protons of the acyl chain (CH2 H910 H811) show both short (Figure 980

6c) and long (Figure 6d) T2 relaxation component 981

- The short T2 relaxation time systematically decreases from 110degC to 70degC for all species by a 982

factor ranging from two to seven depending on the specific chemical group Very long T2 are 983

observed at 110degC for all acyl species while nearly unchanged values are measured below 984

90degC without any clear trend 985

- The sugar protons (H1rsquo H5rsquo H6rsquo) have the shortest set of T2 values (lt 1 ms) below 100degC 986

The long component of T2 generally ranges between 4 ms and 12 ms 987

0 1 2 3 4 5 6 7 8

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1 2 3 4 5 6 7 8

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

H1

H8H11

H9H10

CH2

H1rsquo

H6rsquo

CH2

H910H811 H1CH2

H1rsquo H5rsquo6rsquo

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

OH

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45 T= 110degC

T= 90degC

T= 80degC

T= 70degC

Log

In

ten

sity

Time ms

Fit

)

CH2 H910 H811 H1 H56 H1 OH

00

05

10

15

20

25

30

35

40

T2 m

s

T= 110degC

T= 90degC

T= 80degC

T= 70degC

CH2 H910 H811 H1

0

2

4

6

8

10

12

T2 m

s

T= 110degC

T= 90degC

T= 80degC

T= 70degC

Short T2(1H) Long T2(1H)

a) b)

c)

11

Table S 1 ndash 1H chemical shifts of G-C181-OH in solution and in the solid-state Solid-state spectra are 988

recorded at a MAS frequency of 10 kHz in DMSO-d6 broad peak 989

990

δH ppm

Group Ndeg Liquid

(MeOD-d4) Solid

(T= 110degC) Solid

(T= 70degC)

CH2 H1 354 357

CH2 H2 153 156

CH2 H3-712-14 132 135

CH2 H811 203 203

CH H910 534 534

CH2 H15 162 164

CH2 H16 388 383

CH H1 424 432

CH H2 316 335

CH H3 334 335

CH H4 329 335

CH H5 354 351 357

CH2 H6 366 383

OH OH1 432

45 463

OH OH2 492

OH OH3 490

OH OH4 487

OH OH6 445

991

992

H1H9H10H1rsquo

H6rsquo

OH2rsquo

OH1

12

Detailed discussion about the NOESY and DQ-SQ 2D 1H-1H ssNMR experiments presented in 993

Figure 9 994

995

2D 1H-1H ssNMR experiments are a unique tool to explore through-space proximities 996

between 1H nuclei ssNMR offers a large number of possibilities each adapted to a given 997

system The advantage of 2D maps with respect to a dynamics approach is their direct 998

readability in clear two nuclei which exchange their magnetization through-space (whichever 999

mechanism may be responsible for this exchange) will display a specific cross-peak However 1000

the way cross-peaks are presented and interpreted strictly depends on the nature of the pulse 1001

program which in turn is chosen according to the physical nature of the sample 1002

1D temperature-dependent 1H NMR experiments performed on G-C181-OH (Figure 1003

4a in the main text) indicate sharp peaks at 90degC and broad peaks at 70degC Narrow peaks 1004

indicate fast andor large-amplitude molecular tumbling averaging the strong 1H-1H dipolar 1005

interactions This excludes the use of pulse sequences like DQ-SQ which is not efficient under 1006

these conditions For this reason we employ spin-diffusion-based pulse sequences like 1007

NOESY which is more adapted for samples with high molecular mobility When mixing times 1008

are chosen to be short (lt 1 s) one generally explores through-space proximities within about 5 1009

10 Aring On the contrary when molecular motions are restricted dipolar coupling becomes 1010

significant 1H NMR peaks are broader and pulse sequences like DQ-SQ can now be used 1011

with a spatial resolution of less than 5 Aring 1012

Whenever possible DQ-SQ experiments are preferred over spin diffusion-based pulse 1013

sequences because they are more selective in particular for likewise interactions which are 1014

not discriminated in NOESY-type experiments This is illustrated by the typical interaction 1015

schemes for NOESY (top Figure 9a) and DQ-SQ (top Figure 9b) For two given A and B 1016

protons 2D NOESY maps will display two on-diagonal and two square-shaped symmetrical 1017

off-diagonal cross-peaks Off-diagonal cross-peaks illustrate the mutual A-B and B-A 1018

interaction but the on-diagonal peaks are systematic and do not prove any A-A or B-B 1019

interaction Similarly to NOESY 2D DQ-SQ maps display off-diagonal peaks (symmetric 1020

across the diagonal top Figure 9b) upon A-B and B-A interaction but in contrast they will 1021

display on-diagonal peaks only in the presence of self-interaction In the tutorial example of 1022

Figure 9b top the on-diagonal A-A peak indicates mutual A-A interaction while lack of on-1023

diagonal B-B peak evidences the absence of B-B proximities Combining NOESY and DQ-SQ 1024

experiments is capital to study the local arrangement of G-C181-OH and its evolution with 1025

temperature 1026

13

At T= 90degC the H1 and H5rsquo contributions are very close but possible to discriminate 1027

them (δ(H1)= 354 ppm δ(H5rsquo)= 351 ppm) according to 1H NMR chemical shifts in Table S 1028

1 2D 1H-1H NOESY in Figure 9a also discriminates between the two components attributed 1029

on the basis of the interactions with their mutual neighbors (green H1-H2 and H5rsquo-H1rsquo squares 1030

in Figure 9a) On this basis 2D NOESY does not show a direct H1rsquo-H1 correlation (blue red-1031

crossed) but rather an extensive network of intra-sugar interactions These features exclude 1032

suggest intra- and inter-sugar interactions and they exclude head-tail arrangements as a major 1033

molecular configuration at 90degC 1034

At 70degC the 1H NMR peaks are broader indicating a reduction of the T2 (1H) relaxation 1035

times andor residual homonuclear dipolar interactions excluding the use of NOESY In this 1036

case 1H-1H DQ-SQ correlation experiments are best suited Figure 9b presents the 1H-1H DQ-1037

SQ experiment on G-C181-OH at 70degC Peaks are broader than in NOESY and attribution may 1038

be cumbersome However the experiment is performed at 700 MHz and MAS= 22 kHz which 1039

improves spectral resolution with respect to the same experiment performed at 300 MHz and 1040

MAS= 10 kHz The auto-peak at δH= 356 ppm in the SQ dimension is attributed to H1-H1 spin 1041

pairs mainly resulting from intra CH2 interaction while the DQ cross-peaks at δH= 156-356 1042

ppm and δH= 432-356 ppm respectively origin from H2-H1 and H1rsquo-H1 interactions (Figure 1043

9b) DQ-SQ map does not show any auto-peak corresponding to H1rsquo-H1rsquo or Hxrsquo-Hxrsquo (x= 2rsquo 1044

3rsquo 4rsquo) interactions (red-crossed blue Figure 9b) thus excluding any proximity (lt 5 Aring) between 1045

close glucose neighbors On the contrary DQ-SQ maps (recorded both at 700 MHz Figure 9b 1046

or at 300 MHz not shown) systematically show DQ cross-peaks which can only be attributed 1047

to H1rsquo-H1 interactions (δH= 432-356) The presence of these H1rsquo-H1 interactions can only be 1048

explained by a head-tail arrangement of the G-C181-OH molecules as illustrated below the 1049

DQ-SQ map in Figure 9b 1050

25

was found that gauche effects in some cases up to the γ position in the sugar chain induce 616

upfield chemical shifts while anti and trans conformations result in downfield shifts Each 617

structure (monolayer crystal bilayer crystal or micellar fiber) was then characterized by a given 618

organization of the gluconamide and a characteristic 13C ssNMR fingerprint which could be 619

explained by gauche γ-gauche anti and trans conformations of the headgroup Similar 620

structural-spectroscopic correlations have been collected over the years for much simpler cyclic 621

sugars85 whereas pyranoses can undergo important conformational changes (chair half-chair 622

boat skew) when the appropriate energy barriers are crossed78ndash8086 A quantitative 623

conformational study based on the 13C chemical shift variations shown in Table 1 would be 624

certainly helpful but very delicate in the context of the present state of the art because a 625

rigorous approach requires at least one known crystal structure8182 which is currently missing 626

for this compound A qualitative approach will then be adopted 627

G-C181-OH is a bolaform amphiphile and flip-flop could be a possible mechanism 628

involved in the nanotube stabilization whereas head-tail interactions have long been considered 629

as important in nanotube-forming bolaform amphiphiles6063 From mere considerations based 630

on T2 (1H) relaxation arguments one could exclude head-tail configurations if the R-CH2OH 631

endgroup (H1) was involved in a direct interaction with the sugar group its corresponding T2 632

would probably be comparable to the sugarrsquos one (OH and H1rsquo) but this does not seem to be 633

the case However this hypothesis would be based on the comparison of the relaxation from 634

CH and CH2 protons additionally involved in different kind of motions Besides the relaxation 635

of H1 is influenced by the strong overlap between H1 and H5rsquo For these reasons relaxation 636

arguments are not solid enough and will not be used 637

We then address this question by exploring intermolecular interactions in a direct 638

manner using 2D 1H-1H homonuclear ssNMR correlation spectroscopy At 90degC in the fast-639

tumbling domain of molecular motion (narrow 1H NMR peaks Figure 6a) and corresponding 640

to the vesicle-rich region spin diffusion (NOESY) experiments are preferred over dipolar 641

coupling based experiments because such interaction is averaged out by molecular motion 642

However when the molecular motion is more constrained in the domain of nanotube stability 643

at 70degC (broad 1H spectra Figure 6a) Double-Quantum Single-Quantum (DQ-SQ) correlation 644

experiments are best suited to probe interaction between close molecular neighbors These 645

experiments are reported in Figure 9 and the related interpretation is only briefly reported 646

below while a more detailed discussion can be found on Page 12 in the Supporting Information 647

26

648

Figure 9 ndash a) 2D 1H-1H NOESY NMR experiment performed at MAS= 10 kHz (4 mm probe) on a Bruker 649

300 MHz Avance III HD spectrometer using 300 ms of mixing time Sample temperature is regulated to 650

90degC reached by cooling from 110degC b) 2D 1H-1H DQ-SQ (BAck-to-BAck BABA pulse sequence used for 651

the DQ excitation scheme) NMR experiment performed at MAS= 22 kHz (25 mm probe) on a Bruker 700 652

MHz Avance III spectrometer using a DQ excitation time equal to 2 rotor periods Sample temperature is 653

regulated to 70degC reached by cooling from 110degC The typical schemes illustrating how to interpret these 654

2D experiments are reported above each experimental map Zooms of the sugar region and configurational 655

interpretation are given below each 2D map More information are given in the text below 656

H16-H16

H1-H1

H9-H10

H1rsquo-H1

H1rsquoH910 H16

H1

H1rsquo-H1rsquo

Hxrsquo-Hxrsquo

2

(pp

m)

(ppm)

315

811910 1rsquo

2rsquo2

3rsquo 4rsquo6rsquo

5rsquo

1rsquo2910

asymp asymp asympHDO MeOD Acetone

1

315

811 9101rsquo

2rsquo2

3rsquo4rsquo6rsquo

5rsquo

1rsquo2 9 10

asympasympasympHDOMeODAcetone

H1rsquo-H1

1rsquo

1

(p

pm

)

(ppm)

H1-H2

H1rsquo-H1

H1rsquo

H910

H1H5rsquo

H2

H1rsquo-H5rsquo

315

811910 1rsquo

2rsquo2

3rsquo 4rsquo6rsquo

5rsquo

1rsquo2910

asymp asymp asympHDO MeOD Acetone

1

H2rsquoH3rsquoH4rsquo

H2rsquoH3rsquoH4rsquo

X= 2rsquo3rsquo4rsquo

NOESY - T= 90degC

1H

1H

1H

1H

DQ-SQ - T= 70degC

a) b)

2

(pp

m)

(p

pm

)

(ppm)

A B

AhellipB

AhellipA

BhellipB

NOESY

(ppm)

A B

AhellipB

DQ-SQ

Single-quantum dimension

Do

ub

le-q

ua

ntu

m d

imen

sio

n

AhellipA

BhellipB

BhellipA

BhellipA

315

811910 1rsquo

2rsquo2

3rsquo 4rsquo6rsquo

5rsquo

1rsquo2910

asymp asymp asympHDO MeOD Acetone

1

lt 5 Aring

lt 5 Aring

H1-H2

H811

H15

H16

H6rsquo

H1rsquo

H1

H5rsquo

H2

H1-H2

H1rsquo-H1

H1rsquo-H5rsquo

H1rsquo

H1 H5rsquo

H1rsquo

H1

H5rsquo

H1H2rsquo H3rsquo H4rsquo

H1-H1

H1rsquo-H1

Hxrsquo-HxrsquoX= 2rsquo3rsquo4rsquo

H1-H2

H1rsquo

Zoom sugar region 28 ndash 45 ppmNOESY DQ-SQ

27

Head-tail or head-head configurations could be probed by the existence of H1rsquo-H1 657

interactions whereas H1 (δH= 356 ppm) is associated to the glucolipid tail and H1rsquo (δH= 432 658

ppm) to the sugar headgroup These atoms are more than 15 Aring apart and a direct interaction 659

could not be observed in both NOESY and DQ-SQ experiments NOESY experiments (Figure 660

9a) show that H1rsquo-H1 correlations are not observed Meanwhile clear-cut direct H1rsquo-H1 661

interactions are indeed observed are detected at T= 70degC (DQ cross-peaks related to H1rsquo-H1 662

correlation in DQ-SQ experiment shown in Figure 9b) Interestingly the absence of auto peaks 663

at δH= 432 ppm and δH= 335 ppm in the SQ dimension excludes H1rsquo-H1rsquo and Hxrsquo-Hxrsquo (x= 2 664

3 4) interactions respectively strengthening the idea that sugar headgroups from two nearest-665

neighbor G-C181-OH molecules are not adjacent in the nanotube-vesicle region 666

Homonuclear 2D ssNMR supports the idea according to which G-C181-OH can 667

undergo molecular flip-flops and a head-tail arrangement in the dry product Extrapolating these 668

conclusions in the lower concentration range lack of H1rsquo-H1 proximity at T= 90degC suggests a 669

head-head configuration in the vesicle-rich region and head-tail configuration within the 670

nanotubes It is however still unclear whether flip-flops are driven by spontaneous temperature-671

induced conformational changes in the glucolipid packing stimulating inclusions within the 672

membrane itself672223629ndash1218ndash21 or by defects generated by inter-vesicle collisions In fact a 673

combination of both is not unlikely Nanotubes are not stable when temperature is 20 to 30 674

degrees above Tm1 indicating that inter-vesicle collisions alone cannot explain the unusual 675

phenomenon In an energy landscape approach87 the head-head configuration is certainly 676

favored at high temperature while both head-head and head-tail configurations could be 677

favored in the mid-temperature region corresponding to the range between Tm1 and Tm2 in the 678

dry powder Both statistical and defect-induced flip-flop events contribute to form high mean 679

zero Gaussian curvature morphologies (nanotubes) for which the head-tail conformation 680

limiting the proximity between bulky sugar headgroups is probably more adapted (Figure 8c) 681

The flip-flop interpretation now reasonably explains the strong differences in the sugar and fatty 682

alcohol conformations deduced by the multiple NMR signals attributed to chemically-683

equivalent groups The existence of various low-energy conformations each having specific 684

NMR signature for sugars in solution78ndash8086 and solid-state81ndash84 further supports our 685

interpretation 686

One last point should be clarified Why between Tm1 and Tm2 are two molecular 687

environments able to coexist instead of a single one of average rigidity as found in the case of 688

ldquoripplerdquo phases50ndash54 We believe that this question can be answered based on the NMR results 689

the relatively short T2 values of the H1rsquo and OH protons shows the constrained mobility of the 690

28

sugar headgroup despite a certain flexibility of the chain (long T2 component) This trend may 691

suggest restrictions in translational mobility probably due to the sugar-OH hydrogen bonding 692

(Figure 8c) meaning that lipids are not able to diffuse within the membrane thus keeping two 693

different configurations stable 694

695

Conclusions 696

Topologically-connected nanotube-vesicle lipidic systems are commonly observed for 697

self-assembled membranes composed of at least two amphiphiles and spontaneously driven by 698

anisotropic inclusions andor by external directional forces This work shows that strong 699

differences in membrane curvature can spontaneously occur in single-lipid systems without 700

external anisotropic forces Coexistence and interconnection between nanotubes and vesicles 701

are evidenced via cryo-TEM and fluorescence confocal microscopy in a single-phase region at 702

temperature above a transition temperature Tm1= 483degC determined by DSC on the hydrated 703

sample but below the second transition temperature Tm2 (957degC) determined by DSC on the 704

dry powder X-ray diffraction (WAXS) also shows the coexistence of two lattice periods one 705

attributed to an inter-membrane repeating distance and the other possibly related to the intra-706

nanotube glucolipid arrangement Labelling the systems with Laurdan a fluorophore of which 707

the emission strongly varies with the local membrane rigidity excludes the presence of rigid 708

micro-separated phases (eg Lβ) 709

1H and 13C ssNMR despite the fact that it can only be performed on the dry powder 710

confirm the presence of one main fluid G-C181-OH environment above Tm2 Below Tm2 the 711

splitting of some 13C peaks and the coexistence of short and long 1H relaxation components 712

demonstrate that G-C181-OH can simultaneously pack into structurally and dynamically 713

inequivalent environments Combining advanced 2D 1H-1H correlation experiments based on 714

spin diffusion or double-quanta excitation suggests that vesicles could contain a preferred head-715

head configuration of the bolaform lipid while nanotubes a preferred head-tail assembly 716

The presence of two environments composed of the same molecule characterized by 717

different molecular configurations and rigidity fulfills the theoretical conditions necessary to 718

spontaneously nucleate a membrane inclusion of non-zero average orientation described as 719

necessary to observe topologically-connected nanotube-vesicle structures in the absence of 720

external forces This hypothesis could also be enriched by the possibility of nanotubes to be 721

pulled away between two vesicles after a collision event 722

This work expands the actual theoretical and experimental framework on lipid networks 723

stabilized by a different radius of curvature It could also motivate the synthesis or isolation 724

29

of new lipids providing a higher level of 3D architectural control in soft networks possibly in 725

the range of room temperature 726

727

Acknowledgements 728

Dr Andrea Lassenberger and Dr Sylvain Preacutevost at Institut Laue Langevin (ILL Grenoble 729

France) are kindly acknowledged for their assistance on the SANS experiment Confocal 730

images were performed at the Institut de Biologie Paris Seine (IBPS) imaging facility Jean-731

Franccedilois Gilles (Imaging Core Facility Institut de Biologie Paris Seine (IBPS) CNRS 732

Sorbonne Universiteacute Paris France) is kindly acknowledged for his assistance on the confocal 733

microscopy imaging Dr Guillaume Laurent (LCMCP Sorbonne Universiteacute Paris France) is 734

kindly acknowledged for helpful discussions The French region Ile-de-France SESAME 735

program is acknowledged for financial support (700 MHz NMR 736

spectrometer) Dr Olivier Diat (ICSM Marcoule France) Prof eronikaIglič and Prof V Aleš737

onsul discussi Slovenia) are kindly acknowledged for helpfUniversity of LjubljanaIglič (-Kralj 738

739

Financial support 740

WAXS experiments have been supported by Soleil Light Source Saint Aubin France proposal 741

Ndeg 20201747 SANS experiments have been supported by ILL proposal Ndeg9-13-778 742

743

References 744

1 D Kabaso N Bobrovska W Goacuteźdź E Gongadze V Kralj-Iglič R Zorec and A 745

Iglič Bioelectrochemistry 2012 87 204ndash210 746

2 C R Safinya U Raviv D J Needleman A Zidovska M C Choi M A Ojeda-747

Lopez K K Ewert Y Li H P Miller J Quispe B Carragher C S Potter M W 748

Kim S C Feinstein and L Wilson Adv Mater 2011 23 2260ndash2270 749

3 P Veranič M Lokar G J Schuumltz J Weghuber S Wieser H Haumlgerstrand V Kralj-750

Iglič and A Iglič Biophys J 2008 95 4416ndash4425 751

4 A Rustom R Saffrich I Markovic P Walther and H-H Gerdes Science (80- ) 752

2004 303 1007ndash1010 753

5 M I Angelova S Soleacuteau P Meacuteleacuteard F Faucon and P Bothorel Prog Colloid 754

Polym Sci 1992 89 127ndash131 755

6 L Mathivet S Cribier and P F Devaux Biophys J 1996 70 1112ndash1121 756

7 V Kralj-Iglič G Gomišček J Majhenc V Arrigler and S Svetina Colloids Surfaces 757

A Physicochem Eng Asp 2001 181 315ndash318 758

30

8 V Kralj-Iglič A Iglič G Gomišček F Sevšek V Arrigler and H Haumlgerstrand J 759

Phys A Math Gen 2002 35 1533ndash1549 760

9 M Drab D Stopar V Kralj-Iglič and A Iglič Cells 2019 8 626 761

10 A Zidovska K K Ewert J Quispe B Carragher C S Potter and C R Safinya 762

Langmuir 2009 25 2979ndash2985 763

11 A Zidovska K K Ewert J Quispe B Carragher C S Potter and C R Safinya Soft 764

Matter 2011 7 8363ndash8369 765

12 A Zidovska K K Ewert J Quispe B Carragher C S Potter and C R Safinya 766

Biochim Biophys Acta - Biomembr 2009 1788 1869ndash1876 767

13 M Karlsson K Sott M Davidson A S Cans P Linderholm D Chiu and O Orwar 768

Proc Natl Acad Sci U S A 2002 99 11573ndash11578 769

14 A Karlsson R Karlsson M Karlsson A-S Cans A Stromberg F Ryttsen and O 770

Orwar Nature 2001 409 150ndash152 771

15 M Karlsson K Sott A S Cans A Karlsson R Karlsson and O Orwar Langmuir 772

2001 17 6754ndash6758 773

16 E Evans H Bowman A Leung D Needham and D Tirrell Science (80- ) 1996 774

273 933ndash935 775

17 U Seifert Configurations of fluid membranes and vesicles 1997 vol 46 776

18 J Kaumls and E Sackmann Biophys J 1991 60 825ndash844 777

19 N Bobrovska W Goacuteźdź V Kralj-Iglič and A Iglič PLoS One 2013 8 e73941 778

20 A Iglič B Babnik K Bohinc M Fošnarič H Haumlgerstrand and V Kralj-Iglič J 779

Biomech 2007 40 579ndash585 780

21 K Bohinc T Slivnik A Iglič M Brumen and V Kralj-Iglič J Phys Chem C 2007 781

111 9709ndash9718 782

22 A Iglič H Haumlgerstrand P Veranič A Plemenitaš and V Kralj-Iglič J Theor Biol 783

2006 240 368ndash373 784

23 V Kralj-Iglič and P Veranič Adv Planar Lipid Bilayers Liposomes 2006 5 129ndash785

149 786

24 A Iglic and V Kralj-Iglic Adv Planar Lipid Bilayers Liposomes 2006 4 253ndash279 787

25 S Penič L Mesarec M Fošnarič L Mroacutewczyńska H Haumlgerstrand V Kralj-Iglič and 788

A Iglič Front Phys 2020 8 342 789

26 P Yager and P E Schoen Mol Cryst Liq Cryst 1984 106 371ndash381 790

27 T G Barclay K Constantopoulos and J Matisons Chem Rev 2014 114 10217ndash791

10291 792

31

28 B Božič G Gomišček V Kralj-Iglič S Svetina and B Žekš Eur Biophys J 2002 793

31 487ndash496 794

29 A Iglič H Haumlgerstrand M Bobrowska-Haumlgerstrand V Arrigler and V Kralj-Iglič 795

Phys Lett A 2003 310 493ndash497 796

30 R Bar-Ziv and E Moses Phys Rev Lett 1994 73 1392ndash1395 797

31 S Chiruvolu H E Warriner E Naranjo S H Idziak J O Raumldler R J Plano J A 798

Zasadzinski and C R Safinya Science (80- ) 1994 266 1222ndash1225 799

32 V Kralj-Iglič A Iglič H Haumlgerstrand and P Peterlin Phys Rev E 2000 61 4230ndash800

4234 801

33 L Mesarec M Drab S Penič V Kralj-Iglič and A Iglič Int J Mol Sci 2021 22 802

2348 803

34 L Van Renterghem S L K W Roelants N Baccile K Uyttersprot M C Taelman 804

B Everaert S Mincke S Ledegen S Debrouwer K Scholtens C V Stevens W 805

Soetaert L Van Renterghem S L K W Roelants N Baccile K Uytersprot M C 806

Taelman B Everaert S Mincke S Ledegen S Debrouwer K Scholtens C V 807

Stevens and W Soetaert Biotechnol Bioeng 2018 115 1195ndash1206 808

35 J Clark and F Deswarte Introduction to Chemicals from Biomass Second Edition 809

John Wiley amp Sons 2015 810

36 K V Sajna R Houmlfer R K Sukumaran L D Gottumukkala and A Pandey in 811

Industrial Biorefineries and White Biotechnology eds A Pandey R Houmlfer M 812

Taherzadeh K M Nampoothiri and C Larroche Amsterdam Oxford Waltham 813

Elsevier 2015 pp 499ndash521 814

37 R Marchant and I M Banat Trends Biotechnol 2012 30 558ndash565 815

38 K Hill and C LeHen-Ferrenbach in Sugar-Based Surfactants Fundamentals and 816

Applications ed C C Ruiz Boca Raton CRC Press 2009 pp 1ndash20 817

39 L Rodrigues I M Banat J Teixeira and R Oliveira J Antimicrob Chemother 818

2006 57 609ndash618 819

40 J D Desai and I M Banat Microbiol Mol Biol Rev 1997 61 47ndash64 820

41 S L K W Roelants K Ciesielska S L De Maeseneire H Moens B Everaert S 821

Verweire Q Denon B Vanlerberghe I N A Van Bogaert P Van der Meeren B 822

Devreese and W Soetaert Biotechnol Bioeng 2016 113 550ndash559 823

42 N Baccile M Selmane P Le Griel S Preacutevost J Perez C V Stevens E Delbeke S 824

Zibek M Guenther W Soetaert I N A Van Bogaert and S Roelants Langmuir 825

2016 32 6343ndash6359 826

32

43 N Baccile A-S Cuvier S Preacutevost C V Stevens E Delbeke J Berton W Soetaert 827

I N A Van Bogaert and S Roelants Langmuir 2016 32 10881ndash10894 828

44 N Baccile C Seyrig A Poirier S A Castro S L K W Roelants and S Abel Green 829

Chem 2021 23 3842ndash3944 830

45 A Salis L Cappai C Carucci D F Parsons and M Monduzzi J Phys Chem Lett 831

2020 11 6805ndash6811 832

46 J Schindelin I Arganda-Carreras E Frise V Kaynig M Longair T Pietzsch S 833

Preibisch C Rueden S Saalfeld B Schmid J-Y Tinevez D J White V 834

Hartenstein K Eliceiri P Tomancak and A Cardona Nat Methods 2012 9 676ndash835

682 836

47 T Parasassi and E Gratton J Fluoresc 1995 5 59ndash69 837

48 M M Dodes Traian F L Gonzaacutelez Flecha and V Levi J Lipid Res 2012 53 609ndash838

616 839

49 D M Owen C Rentero A Magenau A Abu-Siniyeh and K Gaus Nat Protoc 840

2012 7 24ndash35 841

50 W Stillwell in An Introduction to Biological Membranes Academic Press 2016 pp 842

135ndash180 843

51 W Stillwell in An Introduction to Biological Membranes Academic Press 2016 pp 844

181ndash220 845

52 B Tenchov Chem Phys Lipids 1991 57 165ndash177 846

53 R Koynova and M Caffrey Biochim Biophys Acta 1998 1376 91ndash145 847

54 F Schmid S Dolezel O Lenz and S Meinhardt J Phys Conf Ser 2014 487 848

012004 849

55 S Meinhardt R L C Vink and F Schmid Proc Natl Acad Sci U S A 2013 110 850

4476ndash4481 851

56 O Glatter and O Kratky Small Angle X-ray Scattering Academic Press London 852

1982 853

57 B A Cunningham A D Brown D H Wolfe W P Williams and A Brain Phys 854

Rev E 1998 58 3662ndash3672 855

58 M Nadler A Steiner T Dvir O Szekely P Szekely A Ginsburg R Asor R Resh 856

C Tamburu M Peres and U Raviv Soft Matter 2011 7 1512ndash1523 857

59 C Tanford The hydrophobic effect Formation of micelles and biological membranes 858

John Wiley amp Sons Inc New York 1973 859

60 M Masuda and T Shimizu Langmuir 2004 20 5969ndash5977 860

33

61 A Zidovska K K Ewert J Quispe B Carragher C S Potter and C R Safinya 861

Block Liposomes Vesicles of Charged Lipids with Distinctly Shaped Nanoscale 862

Sphere- Pear- Tube- or Rod-Segments Elsevier Inc 1st edn 2009 vol 465 863

62 A Iglic V Kralj-Iglic B Božič M Bobrowska-Haumlgerstrand B Isomaa and H 864

Haumlgerstrand Bioelectrochemistry 2000 52 203ndash211 865

63 T Shimizu M Masuda and H Minamikawa Chem Rev 2005 105 1401ndash1444 866

64 J V Selinger and J M Schnur Phys Rev Lett 1993 71 4091ndash4094 867

65 J M Schnur Science (80- ) 1993 262 1669ndash1676 868

66 B Tenchov R Koynova and G Rapp Biophys J 2001 80 1873ndash1890 869

67 R Koynova and B Tenchov Ser Biomech 2017 31 3ndash19 870

68 S J Angyal and R Le Fur Carbohydr Res 1984 126 15ndash26 871

69 J K Whitesell and M A Minton J Am Chem Soc 1987 109 225ndash228 872

70 T Takayama S Ando and I Ando J Mol Struct 1990 220 245ndash250 873

71 K Gawrisch N V Eldho and I V Polozov Chem Phys Lipids 2002 116 135ndash151 874

72 F Tielens C Gervais J F Lambert F Mauri and D Costa Chem Mater 2008 20 875

3336ndash3344 876

73 G A Jeffrey and Y Yeon Acta Crystallogr Sect B 1986 42 410ndash413 877

74 B Berglund and R W Vaughan J Chem Phys 1980 73 2037ndash2043 878

75 C Gervais C Coelho T Azaiumls J Maquet G Laurent F Pourpoint C Bonhomme 879

P Florian B Alonso G Guerrero P H Mutin and F Mauri J Magn Reson 2007 880

187 131ndash140 881

76 C Akita T Kawaguchi F Kaneko H Yamamoto and M Suzuki J Phys Chem B 882

2004 108 4862ndash4868 883

77 R N Purusottam L Seacutenicourt J J Lacapegravere and P Tekely Biochim Biophys Acta - 884

Biomembr 2015 1848 3134ndash3139 885

78 X Biarneacutes A Ardegravevol A Planas C Rovira A Laio and M Parrinello J Am Chem 886

Soc 2007 129 10686ndash10693 887

79 L M J Kroon‐Batenburg and J Kroon Biopolymers 1990 29 1243ndash1248 888

80 K N Kirschner and R J Woods Proc Natl Acad Sci U S A 2001 98 10541ndash889

10545 890

81 S Svenson B Kirste and J H Fuhrhop J Am Chem Soc 1994 116 11969ndash11975 891

82 I Sack S Macholl J H Fuhrhop and G Buntkowsky Phys Chem Chem Phys 892

2000 2 1781ndash1788 893

83 S Svenson J Koening and J H Fuhrhop J Phys Chem 1994 98 1022ndash1028 894

34

84 S Svenson A Schaefer and J H Fuhrhop J Chem Soc Perkin Trans 2 1994 2 895

1023ndash1028 896

85 K Bock and C Pedersen Adv Carbohydr Chem Biochem 1983 41 27ndash66 897

86 T B Grindley Structure and Conformation of Carbohydrates Springer Berlin 898

Heidelberg 2008 899

87 F Tantakitti J Boekhoven X Wang R V Kazantsev T Yu J Li E Zhuang R 900

Zandi J H Ortony C J Newcomb L C Palmer G S Shekhawat M O De La 901

Cruz G C Schatz and S I Stupp Nat Mater 2016 15 469ndash476 902

903

1

Supporting Information 904

905

Topological connection between vesicles and nanotubes in single-906

component lipid membranes driven by head-tail interactions 907

908

Niki Baccilea Ceacutedric Lorthioira Abdoul Aziz Baa Patrick Le Griela Cristina Coelhob Javier 909

Perezc Wim Soetaertd Sophie L K W Roelantsd 910

911

a Sorbonne Universiteacute Centre National de la Recherche Scientifique Laboratoire de Chimie de 912

la Matiegravere Condenseacutee de Paris LCMCP F-75005 Paris France 913

b Institut des Mateacuteriaux de Paris Centre Sorbonne Universiteacute Paris France 914

c Synchrotron Soleil LrsquoOrme des Merisiers Saint-Aubin BP48 91192 Gif-sur-Yvette Cedex 915

France 916

d InBio Department of Biotechnology Ghent University Ghent Belgium 917

918

919

Corresponding author 920

Dr Niki Baccile 921

E-mail address nikibaccilesorbonne-universitefr 922

Phone +33 1 44 27 56 77 923

924

925

926

2

a) 927

Parameter Method

Dry Matter (DM ) 996 Infrared balance 105degC

Glucose () 012 HPLC-Metacarb

Glycerol () nd HPLC-Metacarb

Free Fatty Acid content () lt001 Internal method BBEPP

GC with FID Detector

Oil () lt001 Internal method BBEPP

GC with FID Detector

Moulds (CFUg) lt10 3M-Nordval ndeg16 B

Yeasts (CFUg) lt10 3M-Nordval ndeg16 B

Anaerobic count (CFUg) lt10 SP-VG M005 B

Protein () tbd BCA protein assay

928

Figure S 1 - a) HPLC-ELSD chromatogram and composition table 929

930

931

932

933

3

b) 934

935

Group Ndeg δ ppm

CH2 1 354

CH2 2 153

CH2 3-712-14 132

CH2 811 203

CH 910 534

CH2 15 162

CH2 16 388

CH 1 424

CH 2 316

CH 3 334

CH 4 329

CH 5 354

CH2 6 366

Figure S 1 (next) ndash b) 1H NMR spectrum (MeOD-d4) recorded on the purified G-C181-OH compounds 936

and related assignment 937

938

2

811910 1rsquo

5rsquo1

3rsquo4rsquo

1rsquo1910

asymp asymp asympHDO MeOD Acetone

6rsquo2rsquo

1615

3-712-14

4

939

940

Figure S 2 ndash Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 130degC 941

5

942

943

Figure S 3 - Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 90degC 944

945

90degC

1

22

6

946

947

948

949

Figure S 4 ndash Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 70degC 950

951

21

11

1

2

5

3

4

45

5

6

7

7

7

952

953

954

Figure S 5 ndash Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 25degC 955

956

957

1

1 1

8

958

Figure S 6 ndash a) SANS experiment recorded at 90degC for a 5 mgmL G-C181-OH aqueous solution b-c) 959

WAXS experiments recorded on aqueous solutions of G-C181-OH at b) 5 mgmL and c) 25 mgmL d) 960

Structural interpretation of the repeating distances pointed out in c) 961

962

963

964

001 01 1

q Aring-1

25degC

68degC

88degC

111degC

01 1q Aring

-1

02 04 06 08 1q Aring

-1

001 01

10-1

100

101

Inte

nsity

cm

-1

q Aring-1

-2

T= 90degC

SANS WAXS WAXS

a) b) c)

dv

dL d= 144 Aring-1

di

ddH2O

dt

dvdL

dtdi d

d)

9

965

966

Figure S 7 ndash ssNMR MAS (10 kHz) experiments recorded on a G-C181-OH dry powder a) 1H spectra 967

recorded from 110degC to 70degC b) 13C single pulse (SP) spectra recorded with high power 1H decoupling from 968

110degC to 70degC c) 13C cross polarization (CP) spectra recorded at 110degC and contact time tc of 50 micros and 969

3000 micros d) 13C CP spectra recorded at 70degC and for 20 lt tc micros lt 3000 970

971

1

811

910

CH2 CH2

1rsquo

6rsquo

8 6 4 2 0 ppm

110

90

80

70

910

1rsquo

6rsquo16

1

2rsquo3rsquo4rsquo

811

OH

45 40 35 30 25 20 15

ppm

200 180 160 140 120 100 80 60 40 20 0

ppm140 120 100 80

ppm

CH2-gauche

3004

110

90

80

70

65

60

50

25

C91013003

C9 - 12869

CH2-all-trans

3198

C1rsquo10335

a) b)

1H

T degCT degC

13C Single Pulse MAS

A

B

A

B

70degC70degC

5rsquo

215

200 180 160 140 120 100 80 60 40 20 0

ppm

13C CPMAST= 110degC

200 180 160 140 120 100 80 60 40 20 0

ppm

132 131 130 129 128 127 126

ppm

106 105 104 103 102

ppm

13C CPMAST= 70degC

50

tc micros

3000

20

tc micros

50 500 1000 3000

tc micros

C102

13024 C910113003

C92

12869

C1rsquo213082

C1rsquo110443

A B

106 105 104 103 102

ppm

3000

1000

500

50

20

AB

c)

d)

10

972

Figure S 8 ndash a) Short and d) long T2 components displayed by the 1H transverse relaxation signal c) 1H 973

transverse relaxation signals recorded on a G-C181-OH dry powder from 110degC to 70degC The area of each 974

given resonance is plotted (log-lin) as a function of the echo time (CPMG pulse sequence) for each 975

temperature and fitted using the equation given above T2s and T2l indicate the short and long T2 relaxation 976

times respectively obtained by describing the 1H transverse relaxation signal using a biexponential decay 977

For H1rsquo H5rsquo and H6rsquo the 1H relaxation function could be described by a monoexponential decay 978

- Protons from the sugar group (H1rsquo H5rsquo H6rsquo) show a single T2 relaxation component (Figure 979

6c and Figure S 8) while protons of the acyl chain (CH2 H910 H811) show both short (Figure 980

6c) and long (Figure 6d) T2 relaxation component 981

- The short T2 relaxation time systematically decreases from 110degC to 70degC for all species by a 982

factor ranging from two to seven depending on the specific chemical group Very long T2 are 983

observed at 110degC for all acyl species while nearly unchanged values are measured below 984

90degC without any clear trend 985

- The sugar protons (H1rsquo H5rsquo H6rsquo) have the shortest set of T2 values (lt 1 ms) below 100degC 986

The long component of T2 generally ranges between 4 ms and 12 ms 987

0 1 2 3 4 5 6 7 8

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1 2 3 4 5 6 7 8

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

H1

H8H11

H9H10

CH2

H1rsquo

H6rsquo

CH2

H910H811 H1CH2

H1rsquo H5rsquo6rsquo

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

OH

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45 T= 110degC

T= 90degC

T= 80degC

T= 70degC

Log

In

ten

sity

Time ms

Fit

)

CH2 H910 H811 H1 H56 H1 OH

00

05

10

15

20

25

30

35

40

T2 m

s

T= 110degC

T= 90degC

T= 80degC

T= 70degC

CH2 H910 H811 H1

0

2

4

6

8

10

12

T2 m

s

T= 110degC

T= 90degC

T= 80degC

T= 70degC

Short T2(1H) Long T2(1H)

a) b)

c)

11

Table S 1 ndash 1H chemical shifts of G-C181-OH in solution and in the solid-state Solid-state spectra are 988

recorded at a MAS frequency of 10 kHz in DMSO-d6 broad peak 989

990

δH ppm

Group Ndeg Liquid

(MeOD-d4) Solid

(T= 110degC) Solid

(T= 70degC)

CH2 H1 354 357

CH2 H2 153 156

CH2 H3-712-14 132 135

CH2 H811 203 203

CH H910 534 534

CH2 H15 162 164

CH2 H16 388 383

CH H1 424 432

CH H2 316 335

CH H3 334 335

CH H4 329 335

CH H5 354 351 357

CH2 H6 366 383

OH OH1 432

45 463

OH OH2 492

OH OH3 490

OH OH4 487

OH OH6 445

991

992

H1H9H10H1rsquo

H6rsquo

OH2rsquo

OH1

12

Detailed discussion about the NOESY and DQ-SQ 2D 1H-1H ssNMR experiments presented in 993

Figure 9 994

995

2D 1H-1H ssNMR experiments are a unique tool to explore through-space proximities 996

between 1H nuclei ssNMR offers a large number of possibilities each adapted to a given 997

system The advantage of 2D maps with respect to a dynamics approach is their direct 998

readability in clear two nuclei which exchange their magnetization through-space (whichever 999

mechanism may be responsible for this exchange) will display a specific cross-peak However 1000

the way cross-peaks are presented and interpreted strictly depends on the nature of the pulse 1001

program which in turn is chosen according to the physical nature of the sample 1002

1D temperature-dependent 1H NMR experiments performed on G-C181-OH (Figure 1003

4a in the main text) indicate sharp peaks at 90degC and broad peaks at 70degC Narrow peaks 1004

indicate fast andor large-amplitude molecular tumbling averaging the strong 1H-1H dipolar 1005

interactions This excludes the use of pulse sequences like DQ-SQ which is not efficient under 1006

these conditions For this reason we employ spin-diffusion-based pulse sequences like 1007

NOESY which is more adapted for samples with high molecular mobility When mixing times 1008

are chosen to be short (lt 1 s) one generally explores through-space proximities within about 5 1009

10 Aring On the contrary when molecular motions are restricted dipolar coupling becomes 1010

significant 1H NMR peaks are broader and pulse sequences like DQ-SQ can now be used 1011

with a spatial resolution of less than 5 Aring 1012

Whenever possible DQ-SQ experiments are preferred over spin diffusion-based pulse 1013

sequences because they are more selective in particular for likewise interactions which are 1014

not discriminated in NOESY-type experiments This is illustrated by the typical interaction 1015

schemes for NOESY (top Figure 9a) and DQ-SQ (top Figure 9b) For two given A and B 1016

protons 2D NOESY maps will display two on-diagonal and two square-shaped symmetrical 1017

off-diagonal cross-peaks Off-diagonal cross-peaks illustrate the mutual A-B and B-A 1018

interaction but the on-diagonal peaks are systematic and do not prove any A-A or B-B 1019

interaction Similarly to NOESY 2D DQ-SQ maps display off-diagonal peaks (symmetric 1020

across the diagonal top Figure 9b) upon A-B and B-A interaction but in contrast they will 1021

display on-diagonal peaks only in the presence of self-interaction In the tutorial example of 1022

Figure 9b top the on-diagonal A-A peak indicates mutual A-A interaction while lack of on-1023

diagonal B-B peak evidences the absence of B-B proximities Combining NOESY and DQ-SQ 1024

experiments is capital to study the local arrangement of G-C181-OH and its evolution with 1025

temperature 1026

13

At T= 90degC the H1 and H5rsquo contributions are very close but possible to discriminate 1027

them (δ(H1)= 354 ppm δ(H5rsquo)= 351 ppm) according to 1H NMR chemical shifts in Table S 1028

1 2D 1H-1H NOESY in Figure 9a also discriminates between the two components attributed 1029

on the basis of the interactions with their mutual neighbors (green H1-H2 and H5rsquo-H1rsquo squares 1030

in Figure 9a) On this basis 2D NOESY does not show a direct H1rsquo-H1 correlation (blue red-1031

crossed) but rather an extensive network of intra-sugar interactions These features exclude 1032

suggest intra- and inter-sugar interactions and they exclude head-tail arrangements as a major 1033

molecular configuration at 90degC 1034

At 70degC the 1H NMR peaks are broader indicating a reduction of the T2 (1H) relaxation 1035

times andor residual homonuclear dipolar interactions excluding the use of NOESY In this 1036

case 1H-1H DQ-SQ correlation experiments are best suited Figure 9b presents the 1H-1H DQ-1037

SQ experiment on G-C181-OH at 70degC Peaks are broader than in NOESY and attribution may 1038

be cumbersome However the experiment is performed at 700 MHz and MAS= 22 kHz which 1039

improves spectral resolution with respect to the same experiment performed at 300 MHz and 1040

MAS= 10 kHz The auto-peak at δH= 356 ppm in the SQ dimension is attributed to H1-H1 spin 1041

pairs mainly resulting from intra CH2 interaction while the DQ cross-peaks at δH= 156-356 1042

ppm and δH= 432-356 ppm respectively origin from H2-H1 and H1rsquo-H1 interactions (Figure 1043

9b) DQ-SQ map does not show any auto-peak corresponding to H1rsquo-H1rsquo or Hxrsquo-Hxrsquo (x= 2rsquo 1044

3rsquo 4rsquo) interactions (red-crossed blue Figure 9b) thus excluding any proximity (lt 5 Aring) between 1045

close glucose neighbors On the contrary DQ-SQ maps (recorded both at 700 MHz Figure 9b 1046

or at 300 MHz not shown) systematically show DQ cross-peaks which can only be attributed 1047

to H1rsquo-H1 interactions (δH= 432-356) The presence of these H1rsquo-H1 interactions can only be 1048

explained by a head-tail arrangement of the G-C181-OH molecules as illustrated below the 1049

DQ-SQ map in Figure 9b 1050

26

648

Figure 9 ndash a) 2D 1H-1H NOESY NMR experiment performed at MAS= 10 kHz (4 mm probe) on a Bruker 649

300 MHz Avance III HD spectrometer using 300 ms of mixing time Sample temperature is regulated to 650

90degC reached by cooling from 110degC b) 2D 1H-1H DQ-SQ (BAck-to-BAck BABA pulse sequence used for 651

the DQ excitation scheme) NMR experiment performed at MAS= 22 kHz (25 mm probe) on a Bruker 700 652

MHz Avance III spectrometer using a DQ excitation time equal to 2 rotor periods Sample temperature is 653

regulated to 70degC reached by cooling from 110degC The typical schemes illustrating how to interpret these 654

2D experiments are reported above each experimental map Zooms of the sugar region and configurational 655

interpretation are given below each 2D map More information are given in the text below 656

H16-H16

H1-H1

H9-H10

H1rsquo-H1

H1rsquoH910 H16

H1

H1rsquo-H1rsquo

Hxrsquo-Hxrsquo

2

(pp

m)

(ppm)

315

811910 1rsquo

2rsquo2

3rsquo 4rsquo6rsquo

5rsquo

1rsquo2910

asymp asymp asympHDO MeOD Acetone

1

315

811 9101rsquo

2rsquo2

3rsquo4rsquo6rsquo

5rsquo

1rsquo2 9 10

asympasympasympHDOMeODAcetone

H1rsquo-H1

1rsquo

1

(p

pm

)

(ppm)

H1-H2

H1rsquo-H1

H1rsquo

H910

H1H5rsquo

H2

H1rsquo-H5rsquo

315

811910 1rsquo

2rsquo2

3rsquo 4rsquo6rsquo

5rsquo

1rsquo2910

asymp asymp asympHDO MeOD Acetone

1

H2rsquoH3rsquoH4rsquo

H2rsquoH3rsquoH4rsquo

X= 2rsquo3rsquo4rsquo

NOESY - T= 90degC

1H

1H

1H

1H

DQ-SQ - T= 70degC

a) b)

2

(pp

m)

(p

pm

)

(ppm)

A B

AhellipB

AhellipA

BhellipB

NOESY

(ppm)

A B

AhellipB

DQ-SQ

Single-quantum dimension

Do

ub

le-q

ua

ntu

m d

imen

sio

n

AhellipA

BhellipB

BhellipA

BhellipA

315

811910 1rsquo

2rsquo2

3rsquo 4rsquo6rsquo

5rsquo

1rsquo2910

asymp asymp asympHDO MeOD Acetone

1

lt 5 Aring

lt 5 Aring

H1-H2

H811

H15

H16

H6rsquo

H1rsquo

H1

H5rsquo

H2

H1-H2

H1rsquo-H1

H1rsquo-H5rsquo

H1rsquo

H1 H5rsquo

H1rsquo

H1

H5rsquo

H1H2rsquo H3rsquo H4rsquo

H1-H1

H1rsquo-H1

Hxrsquo-HxrsquoX= 2rsquo3rsquo4rsquo

H1-H2

H1rsquo

Zoom sugar region 28 ndash 45 ppmNOESY DQ-SQ

27

Head-tail or head-head configurations could be probed by the existence of H1rsquo-H1 657

interactions whereas H1 (δH= 356 ppm) is associated to the glucolipid tail and H1rsquo (δH= 432 658

ppm) to the sugar headgroup These atoms are more than 15 Aring apart and a direct interaction 659

could not be observed in both NOESY and DQ-SQ experiments NOESY experiments (Figure 660

9a) show that H1rsquo-H1 correlations are not observed Meanwhile clear-cut direct H1rsquo-H1 661

interactions are indeed observed are detected at T= 70degC (DQ cross-peaks related to H1rsquo-H1 662

correlation in DQ-SQ experiment shown in Figure 9b) Interestingly the absence of auto peaks 663

at δH= 432 ppm and δH= 335 ppm in the SQ dimension excludes H1rsquo-H1rsquo and Hxrsquo-Hxrsquo (x= 2 664

3 4) interactions respectively strengthening the idea that sugar headgroups from two nearest-665

neighbor G-C181-OH molecules are not adjacent in the nanotube-vesicle region 666

Homonuclear 2D ssNMR supports the idea according to which G-C181-OH can 667

undergo molecular flip-flops and a head-tail arrangement in the dry product Extrapolating these 668

conclusions in the lower concentration range lack of H1rsquo-H1 proximity at T= 90degC suggests a 669

head-head configuration in the vesicle-rich region and head-tail configuration within the 670

nanotubes It is however still unclear whether flip-flops are driven by spontaneous temperature-671

induced conformational changes in the glucolipid packing stimulating inclusions within the 672

membrane itself672223629ndash1218ndash21 or by defects generated by inter-vesicle collisions In fact a 673

combination of both is not unlikely Nanotubes are not stable when temperature is 20 to 30 674

degrees above Tm1 indicating that inter-vesicle collisions alone cannot explain the unusual 675

phenomenon In an energy landscape approach87 the head-head configuration is certainly 676

favored at high temperature while both head-head and head-tail configurations could be 677

favored in the mid-temperature region corresponding to the range between Tm1 and Tm2 in the 678

dry powder Both statistical and defect-induced flip-flop events contribute to form high mean 679

zero Gaussian curvature morphologies (nanotubes) for which the head-tail conformation 680

limiting the proximity between bulky sugar headgroups is probably more adapted (Figure 8c) 681

The flip-flop interpretation now reasonably explains the strong differences in the sugar and fatty 682

alcohol conformations deduced by the multiple NMR signals attributed to chemically-683

equivalent groups The existence of various low-energy conformations each having specific 684

NMR signature for sugars in solution78ndash8086 and solid-state81ndash84 further supports our 685

interpretation 686

One last point should be clarified Why between Tm1 and Tm2 are two molecular 687

environments able to coexist instead of a single one of average rigidity as found in the case of 688

ldquoripplerdquo phases50ndash54 We believe that this question can be answered based on the NMR results 689

the relatively short T2 values of the H1rsquo and OH protons shows the constrained mobility of the 690

28

sugar headgroup despite a certain flexibility of the chain (long T2 component) This trend may 691

suggest restrictions in translational mobility probably due to the sugar-OH hydrogen bonding 692

(Figure 8c) meaning that lipids are not able to diffuse within the membrane thus keeping two 693

different configurations stable 694

695

Conclusions 696

Topologically-connected nanotube-vesicle lipidic systems are commonly observed for 697

self-assembled membranes composed of at least two amphiphiles and spontaneously driven by 698

anisotropic inclusions andor by external directional forces This work shows that strong 699

differences in membrane curvature can spontaneously occur in single-lipid systems without 700

external anisotropic forces Coexistence and interconnection between nanotubes and vesicles 701

are evidenced via cryo-TEM and fluorescence confocal microscopy in a single-phase region at 702

temperature above a transition temperature Tm1= 483degC determined by DSC on the hydrated 703

sample but below the second transition temperature Tm2 (957degC) determined by DSC on the 704

dry powder X-ray diffraction (WAXS) also shows the coexistence of two lattice periods one 705

attributed to an inter-membrane repeating distance and the other possibly related to the intra-706

nanotube glucolipid arrangement Labelling the systems with Laurdan a fluorophore of which 707

the emission strongly varies with the local membrane rigidity excludes the presence of rigid 708

micro-separated phases (eg Lβ) 709

1H and 13C ssNMR despite the fact that it can only be performed on the dry powder 710

confirm the presence of one main fluid G-C181-OH environment above Tm2 Below Tm2 the 711

splitting of some 13C peaks and the coexistence of short and long 1H relaxation components 712

demonstrate that G-C181-OH can simultaneously pack into structurally and dynamically 713

inequivalent environments Combining advanced 2D 1H-1H correlation experiments based on 714

spin diffusion or double-quanta excitation suggests that vesicles could contain a preferred head-715

head configuration of the bolaform lipid while nanotubes a preferred head-tail assembly 716

The presence of two environments composed of the same molecule characterized by 717

different molecular configurations and rigidity fulfills the theoretical conditions necessary to 718

spontaneously nucleate a membrane inclusion of non-zero average orientation described as 719

necessary to observe topologically-connected nanotube-vesicle structures in the absence of 720

external forces This hypothesis could also be enriched by the possibility of nanotubes to be 721

pulled away between two vesicles after a collision event 722

This work expands the actual theoretical and experimental framework on lipid networks 723

stabilized by a different radius of curvature It could also motivate the synthesis or isolation 724

29

of new lipids providing a higher level of 3D architectural control in soft networks possibly in 725

the range of room temperature 726

727

Acknowledgements 728

Dr Andrea Lassenberger and Dr Sylvain Preacutevost at Institut Laue Langevin (ILL Grenoble 729

France) are kindly acknowledged for their assistance on the SANS experiment Confocal 730

images were performed at the Institut de Biologie Paris Seine (IBPS) imaging facility Jean-731

Franccedilois Gilles (Imaging Core Facility Institut de Biologie Paris Seine (IBPS) CNRS 732

Sorbonne Universiteacute Paris France) is kindly acknowledged for his assistance on the confocal 733

microscopy imaging Dr Guillaume Laurent (LCMCP Sorbonne Universiteacute Paris France) is 734

kindly acknowledged for helpful discussions The French region Ile-de-France SESAME 735

program is acknowledged for financial support (700 MHz NMR 736

spectrometer) Dr Olivier Diat (ICSM Marcoule France) Prof eronikaIglič and Prof V Aleš737

onsul discussi Slovenia) are kindly acknowledged for helpfUniversity of LjubljanaIglič (-Kralj 738

739

Financial support 740

WAXS experiments have been supported by Soleil Light Source Saint Aubin France proposal 741

Ndeg 20201747 SANS experiments have been supported by ILL proposal Ndeg9-13-778 742

743

References 744

1 D Kabaso N Bobrovska W Goacuteźdź E Gongadze V Kralj-Iglič R Zorec and A 745

Iglič Bioelectrochemistry 2012 87 204ndash210 746

2 C R Safinya U Raviv D J Needleman A Zidovska M C Choi M A Ojeda-747

Lopez K K Ewert Y Li H P Miller J Quispe B Carragher C S Potter M W 748

Kim S C Feinstein and L Wilson Adv Mater 2011 23 2260ndash2270 749

3 P Veranič M Lokar G J Schuumltz J Weghuber S Wieser H Haumlgerstrand V Kralj-750

Iglič and A Iglič Biophys J 2008 95 4416ndash4425 751

4 A Rustom R Saffrich I Markovic P Walther and H-H Gerdes Science (80- ) 752

2004 303 1007ndash1010 753

5 M I Angelova S Soleacuteau P Meacuteleacuteard F Faucon and P Bothorel Prog Colloid 754

Polym Sci 1992 89 127ndash131 755

6 L Mathivet S Cribier and P F Devaux Biophys J 1996 70 1112ndash1121 756

7 V Kralj-Iglič G Gomišček J Majhenc V Arrigler and S Svetina Colloids Surfaces 757

A Physicochem Eng Asp 2001 181 315ndash318 758

30

8 V Kralj-Iglič A Iglič G Gomišček F Sevšek V Arrigler and H Haumlgerstrand J 759

Phys A Math Gen 2002 35 1533ndash1549 760

9 M Drab D Stopar V Kralj-Iglič and A Iglič Cells 2019 8 626 761

10 A Zidovska K K Ewert J Quispe B Carragher C S Potter and C R Safinya 762

Langmuir 2009 25 2979ndash2985 763

11 A Zidovska K K Ewert J Quispe B Carragher C S Potter and C R Safinya Soft 764

Matter 2011 7 8363ndash8369 765

12 A Zidovska K K Ewert J Quispe B Carragher C S Potter and C R Safinya 766

Biochim Biophys Acta - Biomembr 2009 1788 1869ndash1876 767

13 M Karlsson K Sott M Davidson A S Cans P Linderholm D Chiu and O Orwar 768

Proc Natl Acad Sci U S A 2002 99 11573ndash11578 769

14 A Karlsson R Karlsson M Karlsson A-S Cans A Stromberg F Ryttsen and O 770

Orwar Nature 2001 409 150ndash152 771

15 M Karlsson K Sott A S Cans A Karlsson R Karlsson and O Orwar Langmuir 772

2001 17 6754ndash6758 773

16 E Evans H Bowman A Leung D Needham and D Tirrell Science (80- ) 1996 774

273 933ndash935 775

17 U Seifert Configurations of fluid membranes and vesicles 1997 vol 46 776

18 J Kaumls and E Sackmann Biophys J 1991 60 825ndash844 777

19 N Bobrovska W Goacuteźdź V Kralj-Iglič and A Iglič PLoS One 2013 8 e73941 778

20 A Iglič B Babnik K Bohinc M Fošnarič H Haumlgerstrand and V Kralj-Iglič J 779

Biomech 2007 40 579ndash585 780

21 K Bohinc T Slivnik A Iglič M Brumen and V Kralj-Iglič J Phys Chem C 2007 781

111 9709ndash9718 782

22 A Iglič H Haumlgerstrand P Veranič A Plemenitaš and V Kralj-Iglič J Theor Biol 783

2006 240 368ndash373 784

23 V Kralj-Iglič and P Veranič Adv Planar Lipid Bilayers Liposomes 2006 5 129ndash785

149 786

24 A Iglic and V Kralj-Iglic Adv Planar Lipid Bilayers Liposomes 2006 4 253ndash279 787

25 S Penič L Mesarec M Fošnarič L Mroacutewczyńska H Haumlgerstrand V Kralj-Iglič and 788

A Iglič Front Phys 2020 8 342 789

26 P Yager and P E Schoen Mol Cryst Liq Cryst 1984 106 371ndash381 790

27 T G Barclay K Constantopoulos and J Matisons Chem Rev 2014 114 10217ndash791

10291 792

31

28 B Božič G Gomišček V Kralj-Iglič S Svetina and B Žekš Eur Biophys J 2002 793

31 487ndash496 794

29 A Iglič H Haumlgerstrand M Bobrowska-Haumlgerstrand V Arrigler and V Kralj-Iglič 795

Phys Lett A 2003 310 493ndash497 796

30 R Bar-Ziv and E Moses Phys Rev Lett 1994 73 1392ndash1395 797

31 S Chiruvolu H E Warriner E Naranjo S H Idziak J O Raumldler R J Plano J A 798

Zasadzinski and C R Safinya Science (80- ) 1994 266 1222ndash1225 799

32 V Kralj-Iglič A Iglič H Haumlgerstrand and P Peterlin Phys Rev E 2000 61 4230ndash800

4234 801

33 L Mesarec M Drab S Penič V Kralj-Iglič and A Iglič Int J Mol Sci 2021 22 802

2348 803

34 L Van Renterghem S L K W Roelants N Baccile K Uyttersprot M C Taelman 804

B Everaert S Mincke S Ledegen S Debrouwer K Scholtens C V Stevens W 805

Soetaert L Van Renterghem S L K W Roelants N Baccile K Uytersprot M C 806

Taelman B Everaert S Mincke S Ledegen S Debrouwer K Scholtens C V 807

Stevens and W Soetaert Biotechnol Bioeng 2018 115 1195ndash1206 808

35 J Clark and F Deswarte Introduction to Chemicals from Biomass Second Edition 809

John Wiley amp Sons 2015 810

36 K V Sajna R Houmlfer R K Sukumaran L D Gottumukkala and A Pandey in 811

Industrial Biorefineries and White Biotechnology eds A Pandey R Houmlfer M 812

Taherzadeh K M Nampoothiri and C Larroche Amsterdam Oxford Waltham 813

Elsevier 2015 pp 499ndash521 814

37 R Marchant and I M Banat Trends Biotechnol 2012 30 558ndash565 815

38 K Hill and C LeHen-Ferrenbach in Sugar-Based Surfactants Fundamentals and 816

Applications ed C C Ruiz Boca Raton CRC Press 2009 pp 1ndash20 817

39 L Rodrigues I M Banat J Teixeira and R Oliveira J Antimicrob Chemother 818

2006 57 609ndash618 819

40 J D Desai and I M Banat Microbiol Mol Biol Rev 1997 61 47ndash64 820

41 S L K W Roelants K Ciesielska S L De Maeseneire H Moens B Everaert S 821

Verweire Q Denon B Vanlerberghe I N A Van Bogaert P Van der Meeren B 822

Devreese and W Soetaert Biotechnol Bioeng 2016 113 550ndash559 823

42 N Baccile M Selmane P Le Griel S Preacutevost J Perez C V Stevens E Delbeke S 824

Zibek M Guenther W Soetaert I N A Van Bogaert and S Roelants Langmuir 825

2016 32 6343ndash6359 826

32

43 N Baccile A-S Cuvier S Preacutevost C V Stevens E Delbeke J Berton W Soetaert 827

I N A Van Bogaert and S Roelants Langmuir 2016 32 10881ndash10894 828

44 N Baccile C Seyrig A Poirier S A Castro S L K W Roelants and S Abel Green 829

Chem 2021 23 3842ndash3944 830

45 A Salis L Cappai C Carucci D F Parsons and M Monduzzi J Phys Chem Lett 831

2020 11 6805ndash6811 832

46 J Schindelin I Arganda-Carreras E Frise V Kaynig M Longair T Pietzsch S 833

Preibisch C Rueden S Saalfeld B Schmid J-Y Tinevez D J White V 834

Hartenstein K Eliceiri P Tomancak and A Cardona Nat Methods 2012 9 676ndash835

682 836

47 T Parasassi and E Gratton J Fluoresc 1995 5 59ndash69 837

48 M M Dodes Traian F L Gonzaacutelez Flecha and V Levi J Lipid Res 2012 53 609ndash838

616 839

49 D M Owen C Rentero A Magenau A Abu-Siniyeh and K Gaus Nat Protoc 840

2012 7 24ndash35 841

50 W Stillwell in An Introduction to Biological Membranes Academic Press 2016 pp 842

135ndash180 843

51 W Stillwell in An Introduction to Biological Membranes Academic Press 2016 pp 844

181ndash220 845

52 B Tenchov Chem Phys Lipids 1991 57 165ndash177 846

53 R Koynova and M Caffrey Biochim Biophys Acta 1998 1376 91ndash145 847

54 F Schmid S Dolezel O Lenz and S Meinhardt J Phys Conf Ser 2014 487 848

012004 849

55 S Meinhardt R L C Vink and F Schmid Proc Natl Acad Sci U S A 2013 110 850

4476ndash4481 851

56 O Glatter and O Kratky Small Angle X-ray Scattering Academic Press London 852

1982 853

57 B A Cunningham A D Brown D H Wolfe W P Williams and A Brain Phys 854

Rev E 1998 58 3662ndash3672 855

58 M Nadler A Steiner T Dvir O Szekely P Szekely A Ginsburg R Asor R Resh 856

C Tamburu M Peres and U Raviv Soft Matter 2011 7 1512ndash1523 857

59 C Tanford The hydrophobic effect Formation of micelles and biological membranes 858

John Wiley amp Sons Inc New York 1973 859

60 M Masuda and T Shimizu Langmuir 2004 20 5969ndash5977 860

33

61 A Zidovska K K Ewert J Quispe B Carragher C S Potter and C R Safinya 861

Block Liposomes Vesicles of Charged Lipids with Distinctly Shaped Nanoscale 862

Sphere- Pear- Tube- or Rod-Segments Elsevier Inc 1st edn 2009 vol 465 863

62 A Iglic V Kralj-Iglic B Božič M Bobrowska-Haumlgerstrand B Isomaa and H 864

Haumlgerstrand Bioelectrochemistry 2000 52 203ndash211 865

63 T Shimizu M Masuda and H Minamikawa Chem Rev 2005 105 1401ndash1444 866

64 J V Selinger and J M Schnur Phys Rev Lett 1993 71 4091ndash4094 867

65 J M Schnur Science (80- ) 1993 262 1669ndash1676 868

66 B Tenchov R Koynova and G Rapp Biophys J 2001 80 1873ndash1890 869

67 R Koynova and B Tenchov Ser Biomech 2017 31 3ndash19 870

68 S J Angyal and R Le Fur Carbohydr Res 1984 126 15ndash26 871

69 J K Whitesell and M A Minton J Am Chem Soc 1987 109 225ndash228 872

70 T Takayama S Ando and I Ando J Mol Struct 1990 220 245ndash250 873

71 K Gawrisch N V Eldho and I V Polozov Chem Phys Lipids 2002 116 135ndash151 874

72 F Tielens C Gervais J F Lambert F Mauri and D Costa Chem Mater 2008 20 875

3336ndash3344 876

73 G A Jeffrey and Y Yeon Acta Crystallogr Sect B 1986 42 410ndash413 877

74 B Berglund and R W Vaughan J Chem Phys 1980 73 2037ndash2043 878

75 C Gervais C Coelho T Azaiumls J Maquet G Laurent F Pourpoint C Bonhomme 879

P Florian B Alonso G Guerrero P H Mutin and F Mauri J Magn Reson 2007 880

187 131ndash140 881

76 C Akita T Kawaguchi F Kaneko H Yamamoto and M Suzuki J Phys Chem B 882

2004 108 4862ndash4868 883

77 R N Purusottam L Seacutenicourt J J Lacapegravere and P Tekely Biochim Biophys Acta - 884

Biomembr 2015 1848 3134ndash3139 885

78 X Biarneacutes A Ardegravevol A Planas C Rovira A Laio and M Parrinello J Am Chem 886

Soc 2007 129 10686ndash10693 887

79 L M J Kroon‐Batenburg and J Kroon Biopolymers 1990 29 1243ndash1248 888

80 K N Kirschner and R J Woods Proc Natl Acad Sci U S A 2001 98 10541ndash889

10545 890

81 S Svenson B Kirste and J H Fuhrhop J Am Chem Soc 1994 116 11969ndash11975 891

82 I Sack S Macholl J H Fuhrhop and G Buntkowsky Phys Chem Chem Phys 892

2000 2 1781ndash1788 893

83 S Svenson J Koening and J H Fuhrhop J Phys Chem 1994 98 1022ndash1028 894

34

84 S Svenson A Schaefer and J H Fuhrhop J Chem Soc Perkin Trans 2 1994 2 895

1023ndash1028 896

85 K Bock and C Pedersen Adv Carbohydr Chem Biochem 1983 41 27ndash66 897

86 T B Grindley Structure and Conformation of Carbohydrates Springer Berlin 898

Heidelberg 2008 899

87 F Tantakitti J Boekhoven X Wang R V Kazantsev T Yu J Li E Zhuang R 900

Zandi J H Ortony C J Newcomb L C Palmer G S Shekhawat M O De La 901

Cruz G C Schatz and S I Stupp Nat Mater 2016 15 469ndash476 902

903

1

Supporting Information 904

905

Topological connection between vesicles and nanotubes in single-906

component lipid membranes driven by head-tail interactions 907

908

Niki Baccilea Ceacutedric Lorthioira Abdoul Aziz Baa Patrick Le Griela Cristina Coelhob Javier 909

Perezc Wim Soetaertd Sophie L K W Roelantsd 910

911

a Sorbonne Universiteacute Centre National de la Recherche Scientifique Laboratoire de Chimie de 912

la Matiegravere Condenseacutee de Paris LCMCP F-75005 Paris France 913

b Institut des Mateacuteriaux de Paris Centre Sorbonne Universiteacute Paris France 914

c Synchrotron Soleil LrsquoOrme des Merisiers Saint-Aubin BP48 91192 Gif-sur-Yvette Cedex 915

France 916

d InBio Department of Biotechnology Ghent University Ghent Belgium 917

918

919

Corresponding author 920

Dr Niki Baccile 921

E-mail address nikibaccilesorbonne-universitefr 922

Phone +33 1 44 27 56 77 923

924

925

926

2

a) 927

Parameter Method

Dry Matter (DM ) 996 Infrared balance 105degC

Glucose () 012 HPLC-Metacarb

Glycerol () nd HPLC-Metacarb

Free Fatty Acid content () lt001 Internal method BBEPP

GC with FID Detector

Oil () lt001 Internal method BBEPP

GC with FID Detector

Moulds (CFUg) lt10 3M-Nordval ndeg16 B

Yeasts (CFUg) lt10 3M-Nordval ndeg16 B

Anaerobic count (CFUg) lt10 SP-VG M005 B

Protein () tbd BCA protein assay

928

Figure S 1 - a) HPLC-ELSD chromatogram and composition table 929

930

931

932

933

3

b) 934

935

Group Ndeg δ ppm

CH2 1 354

CH2 2 153

CH2 3-712-14 132

CH2 811 203

CH 910 534

CH2 15 162

CH2 16 388

CH 1 424

CH 2 316

CH 3 334

CH 4 329

CH 5 354

CH2 6 366

Figure S 1 (next) ndash b) 1H NMR spectrum (MeOD-d4) recorded on the purified G-C181-OH compounds 936

and related assignment 937

938

2

811910 1rsquo

5rsquo1

3rsquo4rsquo

1rsquo1910

asymp asymp asympHDO MeOD Acetone

6rsquo2rsquo

1615

3-712-14

4

939

940

Figure S 2 ndash Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 130degC 941

5

942

943

Figure S 3 - Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 90degC 944

945

90degC

1

22

6

946

947

948

949

Figure S 4 ndash Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 70degC 950

951

21

11

1

2

5

3

4

45

5

6

7

7

7

952

953

954

Figure S 5 ndash Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 25degC 955

956

957

1

1 1

8

958

Figure S 6 ndash a) SANS experiment recorded at 90degC for a 5 mgmL G-C181-OH aqueous solution b-c) 959

WAXS experiments recorded on aqueous solutions of G-C181-OH at b) 5 mgmL and c) 25 mgmL d) 960

Structural interpretation of the repeating distances pointed out in c) 961

962

963

964

001 01 1

q Aring-1

25degC

68degC

88degC

111degC

01 1q Aring

-1

02 04 06 08 1q Aring

-1

001 01

10-1

100

101

Inte

nsity

cm

-1

q Aring-1

-2

T= 90degC

SANS WAXS WAXS

a) b) c)

dv

dL d= 144 Aring-1

di

ddH2O

dt

dvdL

dtdi d

d)

9

965

966

Figure S 7 ndash ssNMR MAS (10 kHz) experiments recorded on a G-C181-OH dry powder a) 1H spectra 967

recorded from 110degC to 70degC b) 13C single pulse (SP) spectra recorded with high power 1H decoupling from 968

110degC to 70degC c) 13C cross polarization (CP) spectra recorded at 110degC and contact time tc of 50 micros and 969

3000 micros d) 13C CP spectra recorded at 70degC and for 20 lt tc micros lt 3000 970

971

1

811

910

CH2 CH2

1rsquo

6rsquo

8 6 4 2 0 ppm

110

90

80

70

910

1rsquo

6rsquo16

1

2rsquo3rsquo4rsquo

811

OH

45 40 35 30 25 20 15

ppm

200 180 160 140 120 100 80 60 40 20 0

ppm140 120 100 80

ppm

CH2-gauche

3004

110

90

80

70

65

60

50

25

C91013003

C9 - 12869

CH2-all-trans

3198

C1rsquo10335

a) b)

1H

T degCT degC

13C Single Pulse MAS

A

B

A

B

70degC70degC

5rsquo

215

200 180 160 140 120 100 80 60 40 20 0

ppm

13C CPMAST= 110degC

200 180 160 140 120 100 80 60 40 20 0

ppm

132 131 130 129 128 127 126

ppm

106 105 104 103 102

ppm

13C CPMAST= 70degC

50

tc micros

3000

20

tc micros

50 500 1000 3000

tc micros

C102

13024 C910113003

C92

12869

C1rsquo213082

C1rsquo110443

A B

106 105 104 103 102

ppm

3000

1000

500

50

20

AB

c)

d)

10

972

Figure S 8 ndash a) Short and d) long T2 components displayed by the 1H transverse relaxation signal c) 1H 973

transverse relaxation signals recorded on a G-C181-OH dry powder from 110degC to 70degC The area of each 974

given resonance is plotted (log-lin) as a function of the echo time (CPMG pulse sequence) for each 975

temperature and fitted using the equation given above T2s and T2l indicate the short and long T2 relaxation 976

times respectively obtained by describing the 1H transverse relaxation signal using a biexponential decay 977

For H1rsquo H5rsquo and H6rsquo the 1H relaxation function could be described by a monoexponential decay 978

- Protons from the sugar group (H1rsquo H5rsquo H6rsquo) show a single T2 relaxation component (Figure 979

6c and Figure S 8) while protons of the acyl chain (CH2 H910 H811) show both short (Figure 980

6c) and long (Figure 6d) T2 relaxation component 981

- The short T2 relaxation time systematically decreases from 110degC to 70degC for all species by a 982

factor ranging from two to seven depending on the specific chemical group Very long T2 are 983

observed at 110degC for all acyl species while nearly unchanged values are measured below 984

90degC without any clear trend 985

- The sugar protons (H1rsquo H5rsquo H6rsquo) have the shortest set of T2 values (lt 1 ms) below 100degC 986

The long component of T2 generally ranges between 4 ms and 12 ms 987

0 1 2 3 4 5 6 7 8

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1 2 3 4 5 6 7 8

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

H1

H8H11

H9H10

CH2

H1rsquo

H6rsquo

CH2

H910H811 H1CH2

H1rsquo H5rsquo6rsquo

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

OH

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45 T= 110degC

T= 90degC

T= 80degC

T= 70degC

Log

In

ten

sity

Time ms

Fit

)

CH2 H910 H811 H1 H56 H1 OH

00

05

10

15

20

25

30

35

40

T2 m

s

T= 110degC

T= 90degC

T= 80degC

T= 70degC

CH2 H910 H811 H1

0

2

4

6

8

10

12

T2 m

s

T= 110degC

T= 90degC

T= 80degC

T= 70degC

Short T2(1H) Long T2(1H)

a) b)

c)

11

Table S 1 ndash 1H chemical shifts of G-C181-OH in solution and in the solid-state Solid-state spectra are 988

recorded at a MAS frequency of 10 kHz in DMSO-d6 broad peak 989

990

δH ppm

Group Ndeg Liquid

(MeOD-d4) Solid

(T= 110degC) Solid

(T= 70degC)

CH2 H1 354 357

CH2 H2 153 156

CH2 H3-712-14 132 135

CH2 H811 203 203

CH H910 534 534

CH2 H15 162 164

CH2 H16 388 383

CH H1 424 432

CH H2 316 335

CH H3 334 335

CH H4 329 335

CH H5 354 351 357

CH2 H6 366 383

OH OH1 432

45 463

OH OH2 492

OH OH3 490

OH OH4 487

OH OH6 445

991

992

H1H9H10H1rsquo

H6rsquo

OH2rsquo

OH1

12

Detailed discussion about the NOESY and DQ-SQ 2D 1H-1H ssNMR experiments presented in 993

Figure 9 994

995

2D 1H-1H ssNMR experiments are a unique tool to explore through-space proximities 996

between 1H nuclei ssNMR offers a large number of possibilities each adapted to a given 997

system The advantage of 2D maps with respect to a dynamics approach is their direct 998

readability in clear two nuclei which exchange their magnetization through-space (whichever 999

mechanism may be responsible for this exchange) will display a specific cross-peak However 1000

the way cross-peaks are presented and interpreted strictly depends on the nature of the pulse 1001

program which in turn is chosen according to the physical nature of the sample 1002

1D temperature-dependent 1H NMR experiments performed on G-C181-OH (Figure 1003

4a in the main text) indicate sharp peaks at 90degC and broad peaks at 70degC Narrow peaks 1004

indicate fast andor large-amplitude molecular tumbling averaging the strong 1H-1H dipolar 1005

interactions This excludes the use of pulse sequences like DQ-SQ which is not efficient under 1006

these conditions For this reason we employ spin-diffusion-based pulse sequences like 1007

NOESY which is more adapted for samples with high molecular mobility When mixing times 1008

are chosen to be short (lt 1 s) one generally explores through-space proximities within about 5 1009

10 Aring On the contrary when molecular motions are restricted dipolar coupling becomes 1010

significant 1H NMR peaks are broader and pulse sequences like DQ-SQ can now be used 1011

with a spatial resolution of less than 5 Aring 1012

Whenever possible DQ-SQ experiments are preferred over spin diffusion-based pulse 1013

sequences because they are more selective in particular for likewise interactions which are 1014

not discriminated in NOESY-type experiments This is illustrated by the typical interaction 1015

schemes for NOESY (top Figure 9a) and DQ-SQ (top Figure 9b) For two given A and B 1016

protons 2D NOESY maps will display two on-diagonal and two square-shaped symmetrical 1017

off-diagonal cross-peaks Off-diagonal cross-peaks illustrate the mutual A-B and B-A 1018

interaction but the on-diagonal peaks are systematic and do not prove any A-A or B-B 1019

interaction Similarly to NOESY 2D DQ-SQ maps display off-diagonal peaks (symmetric 1020

across the diagonal top Figure 9b) upon A-B and B-A interaction but in contrast they will 1021

display on-diagonal peaks only in the presence of self-interaction In the tutorial example of 1022

Figure 9b top the on-diagonal A-A peak indicates mutual A-A interaction while lack of on-1023

diagonal B-B peak evidences the absence of B-B proximities Combining NOESY and DQ-SQ 1024

experiments is capital to study the local arrangement of G-C181-OH and its evolution with 1025

temperature 1026

13

At T= 90degC the H1 and H5rsquo contributions are very close but possible to discriminate 1027

them (δ(H1)= 354 ppm δ(H5rsquo)= 351 ppm) according to 1H NMR chemical shifts in Table S 1028

1 2D 1H-1H NOESY in Figure 9a also discriminates between the two components attributed 1029

on the basis of the interactions with their mutual neighbors (green H1-H2 and H5rsquo-H1rsquo squares 1030

in Figure 9a) On this basis 2D NOESY does not show a direct H1rsquo-H1 correlation (blue red-1031

crossed) but rather an extensive network of intra-sugar interactions These features exclude 1032

suggest intra- and inter-sugar interactions and they exclude head-tail arrangements as a major 1033

molecular configuration at 90degC 1034

At 70degC the 1H NMR peaks are broader indicating a reduction of the T2 (1H) relaxation 1035

times andor residual homonuclear dipolar interactions excluding the use of NOESY In this 1036

case 1H-1H DQ-SQ correlation experiments are best suited Figure 9b presents the 1H-1H DQ-1037

SQ experiment on G-C181-OH at 70degC Peaks are broader than in NOESY and attribution may 1038

be cumbersome However the experiment is performed at 700 MHz and MAS= 22 kHz which 1039

improves spectral resolution with respect to the same experiment performed at 300 MHz and 1040

MAS= 10 kHz The auto-peak at δH= 356 ppm in the SQ dimension is attributed to H1-H1 spin 1041

pairs mainly resulting from intra CH2 interaction while the DQ cross-peaks at δH= 156-356 1042

ppm and δH= 432-356 ppm respectively origin from H2-H1 and H1rsquo-H1 interactions (Figure 1043

9b) DQ-SQ map does not show any auto-peak corresponding to H1rsquo-H1rsquo or Hxrsquo-Hxrsquo (x= 2rsquo 1044

3rsquo 4rsquo) interactions (red-crossed blue Figure 9b) thus excluding any proximity (lt 5 Aring) between 1045

close glucose neighbors On the contrary DQ-SQ maps (recorded both at 700 MHz Figure 9b 1046

or at 300 MHz not shown) systematically show DQ cross-peaks which can only be attributed 1047

to H1rsquo-H1 interactions (δH= 432-356) The presence of these H1rsquo-H1 interactions can only be 1048

explained by a head-tail arrangement of the G-C181-OH molecules as illustrated below the 1049

DQ-SQ map in Figure 9b 1050

27

Head-tail or head-head configurations could be probed by the existence of H1rsquo-H1 657

interactions whereas H1 (δH= 356 ppm) is associated to the glucolipid tail and H1rsquo (δH= 432 658

ppm) to the sugar headgroup These atoms are more than 15 Aring apart and a direct interaction 659

could not be observed in both NOESY and DQ-SQ experiments NOESY experiments (Figure 660

9a) show that H1rsquo-H1 correlations are not observed Meanwhile clear-cut direct H1rsquo-H1 661

interactions are indeed observed are detected at T= 70degC (DQ cross-peaks related to H1rsquo-H1 662

correlation in DQ-SQ experiment shown in Figure 9b) Interestingly the absence of auto peaks 663

at δH= 432 ppm and δH= 335 ppm in the SQ dimension excludes H1rsquo-H1rsquo and Hxrsquo-Hxrsquo (x= 2 664

3 4) interactions respectively strengthening the idea that sugar headgroups from two nearest-665

neighbor G-C181-OH molecules are not adjacent in the nanotube-vesicle region 666

Homonuclear 2D ssNMR supports the idea according to which G-C181-OH can 667

undergo molecular flip-flops and a head-tail arrangement in the dry product Extrapolating these 668

conclusions in the lower concentration range lack of H1rsquo-H1 proximity at T= 90degC suggests a 669

head-head configuration in the vesicle-rich region and head-tail configuration within the 670

nanotubes It is however still unclear whether flip-flops are driven by spontaneous temperature-671

induced conformational changes in the glucolipid packing stimulating inclusions within the 672

membrane itself672223629ndash1218ndash21 or by defects generated by inter-vesicle collisions In fact a 673

combination of both is not unlikely Nanotubes are not stable when temperature is 20 to 30 674

degrees above Tm1 indicating that inter-vesicle collisions alone cannot explain the unusual 675

phenomenon In an energy landscape approach87 the head-head configuration is certainly 676

favored at high temperature while both head-head and head-tail configurations could be 677

favored in the mid-temperature region corresponding to the range between Tm1 and Tm2 in the 678

dry powder Both statistical and defect-induced flip-flop events contribute to form high mean 679

zero Gaussian curvature morphologies (nanotubes) for which the head-tail conformation 680

limiting the proximity between bulky sugar headgroups is probably more adapted (Figure 8c) 681

The flip-flop interpretation now reasonably explains the strong differences in the sugar and fatty 682

alcohol conformations deduced by the multiple NMR signals attributed to chemically-683

equivalent groups The existence of various low-energy conformations each having specific 684

NMR signature for sugars in solution78ndash8086 and solid-state81ndash84 further supports our 685

interpretation 686

One last point should be clarified Why between Tm1 and Tm2 are two molecular 687

environments able to coexist instead of a single one of average rigidity as found in the case of 688

ldquoripplerdquo phases50ndash54 We believe that this question can be answered based on the NMR results 689

the relatively short T2 values of the H1rsquo and OH protons shows the constrained mobility of the 690

28

sugar headgroup despite a certain flexibility of the chain (long T2 component) This trend may 691

suggest restrictions in translational mobility probably due to the sugar-OH hydrogen bonding 692

(Figure 8c) meaning that lipids are not able to diffuse within the membrane thus keeping two 693

different configurations stable 694

695

Conclusions 696

Topologically-connected nanotube-vesicle lipidic systems are commonly observed for 697

self-assembled membranes composed of at least two amphiphiles and spontaneously driven by 698

anisotropic inclusions andor by external directional forces This work shows that strong 699

differences in membrane curvature can spontaneously occur in single-lipid systems without 700

external anisotropic forces Coexistence and interconnection between nanotubes and vesicles 701

are evidenced via cryo-TEM and fluorescence confocal microscopy in a single-phase region at 702

temperature above a transition temperature Tm1= 483degC determined by DSC on the hydrated 703

sample but below the second transition temperature Tm2 (957degC) determined by DSC on the 704

dry powder X-ray diffraction (WAXS) also shows the coexistence of two lattice periods one 705

attributed to an inter-membrane repeating distance and the other possibly related to the intra-706

nanotube glucolipid arrangement Labelling the systems with Laurdan a fluorophore of which 707

the emission strongly varies with the local membrane rigidity excludes the presence of rigid 708

micro-separated phases (eg Lβ) 709

1H and 13C ssNMR despite the fact that it can only be performed on the dry powder 710

confirm the presence of one main fluid G-C181-OH environment above Tm2 Below Tm2 the 711

splitting of some 13C peaks and the coexistence of short and long 1H relaxation components 712

demonstrate that G-C181-OH can simultaneously pack into structurally and dynamically 713

inequivalent environments Combining advanced 2D 1H-1H correlation experiments based on 714

spin diffusion or double-quanta excitation suggests that vesicles could contain a preferred head-715

head configuration of the bolaform lipid while nanotubes a preferred head-tail assembly 716

The presence of two environments composed of the same molecule characterized by 717

different molecular configurations and rigidity fulfills the theoretical conditions necessary to 718

spontaneously nucleate a membrane inclusion of non-zero average orientation described as 719

necessary to observe topologically-connected nanotube-vesicle structures in the absence of 720

external forces This hypothesis could also be enriched by the possibility of nanotubes to be 721

pulled away between two vesicles after a collision event 722

This work expands the actual theoretical and experimental framework on lipid networks 723

stabilized by a different radius of curvature It could also motivate the synthesis or isolation 724

29

of new lipids providing a higher level of 3D architectural control in soft networks possibly in 725

the range of room temperature 726

727

Acknowledgements 728

Dr Andrea Lassenberger and Dr Sylvain Preacutevost at Institut Laue Langevin (ILL Grenoble 729

France) are kindly acknowledged for their assistance on the SANS experiment Confocal 730

images were performed at the Institut de Biologie Paris Seine (IBPS) imaging facility Jean-731

Franccedilois Gilles (Imaging Core Facility Institut de Biologie Paris Seine (IBPS) CNRS 732

Sorbonne Universiteacute Paris France) is kindly acknowledged for his assistance on the confocal 733

microscopy imaging Dr Guillaume Laurent (LCMCP Sorbonne Universiteacute Paris France) is 734

kindly acknowledged for helpful discussions The French region Ile-de-France SESAME 735

program is acknowledged for financial support (700 MHz NMR 736

spectrometer) Dr Olivier Diat (ICSM Marcoule France) Prof eronikaIglič and Prof V Aleš737

onsul discussi Slovenia) are kindly acknowledged for helpfUniversity of LjubljanaIglič (-Kralj 738

739

Financial support 740

WAXS experiments have been supported by Soleil Light Source Saint Aubin France proposal 741

Ndeg 20201747 SANS experiments have been supported by ILL proposal Ndeg9-13-778 742

743

References 744

1 D Kabaso N Bobrovska W Goacuteźdź E Gongadze V Kralj-Iglič R Zorec and A 745

Iglič Bioelectrochemistry 2012 87 204ndash210 746

2 C R Safinya U Raviv D J Needleman A Zidovska M C Choi M A Ojeda-747

Lopez K K Ewert Y Li H P Miller J Quispe B Carragher C S Potter M W 748

Kim S C Feinstein and L Wilson Adv Mater 2011 23 2260ndash2270 749

3 P Veranič M Lokar G J Schuumltz J Weghuber S Wieser H Haumlgerstrand V Kralj-750

Iglič and A Iglič Biophys J 2008 95 4416ndash4425 751

4 A Rustom R Saffrich I Markovic P Walther and H-H Gerdes Science (80- ) 752

2004 303 1007ndash1010 753

5 M I Angelova S Soleacuteau P Meacuteleacuteard F Faucon and P Bothorel Prog Colloid 754

Polym Sci 1992 89 127ndash131 755

6 L Mathivet S Cribier and P F Devaux Biophys J 1996 70 1112ndash1121 756

7 V Kralj-Iglič G Gomišček J Majhenc V Arrigler and S Svetina Colloids Surfaces 757

A Physicochem Eng Asp 2001 181 315ndash318 758

30

8 V Kralj-Iglič A Iglič G Gomišček F Sevšek V Arrigler and H Haumlgerstrand J 759

Phys A Math Gen 2002 35 1533ndash1549 760

9 M Drab D Stopar V Kralj-Iglič and A Iglič Cells 2019 8 626 761

10 A Zidovska K K Ewert J Quispe B Carragher C S Potter and C R Safinya 762

Langmuir 2009 25 2979ndash2985 763

11 A Zidovska K K Ewert J Quispe B Carragher C S Potter and C R Safinya Soft 764

Matter 2011 7 8363ndash8369 765

12 A Zidovska K K Ewert J Quispe B Carragher C S Potter and C R Safinya 766

Biochim Biophys Acta - Biomembr 2009 1788 1869ndash1876 767

13 M Karlsson K Sott M Davidson A S Cans P Linderholm D Chiu and O Orwar 768

Proc Natl Acad Sci U S A 2002 99 11573ndash11578 769

14 A Karlsson R Karlsson M Karlsson A-S Cans A Stromberg F Ryttsen and O 770

Orwar Nature 2001 409 150ndash152 771

15 M Karlsson K Sott A S Cans A Karlsson R Karlsson and O Orwar Langmuir 772

2001 17 6754ndash6758 773

16 E Evans H Bowman A Leung D Needham and D Tirrell Science (80- ) 1996 774

273 933ndash935 775

17 U Seifert Configurations of fluid membranes and vesicles 1997 vol 46 776

18 J Kaumls and E Sackmann Biophys J 1991 60 825ndash844 777

19 N Bobrovska W Goacuteźdź V Kralj-Iglič and A Iglič PLoS One 2013 8 e73941 778

20 A Iglič B Babnik K Bohinc M Fošnarič H Haumlgerstrand and V Kralj-Iglič J 779

Biomech 2007 40 579ndash585 780

21 K Bohinc T Slivnik A Iglič M Brumen and V Kralj-Iglič J Phys Chem C 2007 781

111 9709ndash9718 782

22 A Iglič H Haumlgerstrand P Veranič A Plemenitaš and V Kralj-Iglič J Theor Biol 783

2006 240 368ndash373 784

23 V Kralj-Iglič and P Veranič Adv Planar Lipid Bilayers Liposomes 2006 5 129ndash785

149 786

24 A Iglic and V Kralj-Iglic Adv Planar Lipid Bilayers Liposomes 2006 4 253ndash279 787

25 S Penič L Mesarec M Fošnarič L Mroacutewczyńska H Haumlgerstrand V Kralj-Iglič and 788

A Iglič Front Phys 2020 8 342 789

26 P Yager and P E Schoen Mol Cryst Liq Cryst 1984 106 371ndash381 790

27 T G Barclay K Constantopoulos and J Matisons Chem Rev 2014 114 10217ndash791

10291 792

31

28 B Božič G Gomišček V Kralj-Iglič S Svetina and B Žekš Eur Biophys J 2002 793

31 487ndash496 794

29 A Iglič H Haumlgerstrand M Bobrowska-Haumlgerstrand V Arrigler and V Kralj-Iglič 795

Phys Lett A 2003 310 493ndash497 796

30 R Bar-Ziv and E Moses Phys Rev Lett 1994 73 1392ndash1395 797

31 S Chiruvolu H E Warriner E Naranjo S H Idziak J O Raumldler R J Plano J A 798

Zasadzinski and C R Safinya Science (80- ) 1994 266 1222ndash1225 799

32 V Kralj-Iglič A Iglič H Haumlgerstrand and P Peterlin Phys Rev E 2000 61 4230ndash800

4234 801

33 L Mesarec M Drab S Penič V Kralj-Iglič and A Iglič Int J Mol Sci 2021 22 802

2348 803

34 L Van Renterghem S L K W Roelants N Baccile K Uyttersprot M C Taelman 804

B Everaert S Mincke S Ledegen S Debrouwer K Scholtens C V Stevens W 805

Soetaert L Van Renterghem S L K W Roelants N Baccile K Uytersprot M C 806

Taelman B Everaert S Mincke S Ledegen S Debrouwer K Scholtens C V 807

Stevens and W Soetaert Biotechnol Bioeng 2018 115 1195ndash1206 808

35 J Clark and F Deswarte Introduction to Chemicals from Biomass Second Edition 809

John Wiley amp Sons 2015 810

36 K V Sajna R Houmlfer R K Sukumaran L D Gottumukkala and A Pandey in 811

Industrial Biorefineries and White Biotechnology eds A Pandey R Houmlfer M 812

Taherzadeh K M Nampoothiri and C Larroche Amsterdam Oxford Waltham 813

Elsevier 2015 pp 499ndash521 814

37 R Marchant and I M Banat Trends Biotechnol 2012 30 558ndash565 815

38 K Hill and C LeHen-Ferrenbach in Sugar-Based Surfactants Fundamentals and 816

Applications ed C C Ruiz Boca Raton CRC Press 2009 pp 1ndash20 817

39 L Rodrigues I M Banat J Teixeira and R Oliveira J Antimicrob Chemother 818

2006 57 609ndash618 819

40 J D Desai and I M Banat Microbiol Mol Biol Rev 1997 61 47ndash64 820

41 S L K W Roelants K Ciesielska S L De Maeseneire H Moens B Everaert S 821

Verweire Q Denon B Vanlerberghe I N A Van Bogaert P Van der Meeren B 822

Devreese and W Soetaert Biotechnol Bioeng 2016 113 550ndash559 823

42 N Baccile M Selmane P Le Griel S Preacutevost J Perez C V Stevens E Delbeke S 824

Zibek M Guenther W Soetaert I N A Van Bogaert and S Roelants Langmuir 825

2016 32 6343ndash6359 826

32

43 N Baccile A-S Cuvier S Preacutevost C V Stevens E Delbeke J Berton W Soetaert 827

I N A Van Bogaert and S Roelants Langmuir 2016 32 10881ndash10894 828

44 N Baccile C Seyrig A Poirier S A Castro S L K W Roelants and S Abel Green 829

Chem 2021 23 3842ndash3944 830

45 A Salis L Cappai C Carucci D F Parsons and M Monduzzi J Phys Chem Lett 831

2020 11 6805ndash6811 832

46 J Schindelin I Arganda-Carreras E Frise V Kaynig M Longair T Pietzsch S 833

Preibisch C Rueden S Saalfeld B Schmid J-Y Tinevez D J White V 834

Hartenstein K Eliceiri P Tomancak and A Cardona Nat Methods 2012 9 676ndash835

682 836

47 T Parasassi and E Gratton J Fluoresc 1995 5 59ndash69 837

48 M M Dodes Traian F L Gonzaacutelez Flecha and V Levi J Lipid Res 2012 53 609ndash838

616 839

49 D M Owen C Rentero A Magenau A Abu-Siniyeh and K Gaus Nat Protoc 840

2012 7 24ndash35 841

50 W Stillwell in An Introduction to Biological Membranes Academic Press 2016 pp 842

135ndash180 843

51 W Stillwell in An Introduction to Biological Membranes Academic Press 2016 pp 844

181ndash220 845

52 B Tenchov Chem Phys Lipids 1991 57 165ndash177 846

53 R Koynova and M Caffrey Biochim Biophys Acta 1998 1376 91ndash145 847

54 F Schmid S Dolezel O Lenz and S Meinhardt J Phys Conf Ser 2014 487 848

012004 849

55 S Meinhardt R L C Vink and F Schmid Proc Natl Acad Sci U S A 2013 110 850

4476ndash4481 851

56 O Glatter and O Kratky Small Angle X-ray Scattering Academic Press London 852

1982 853

57 B A Cunningham A D Brown D H Wolfe W P Williams and A Brain Phys 854

Rev E 1998 58 3662ndash3672 855

58 M Nadler A Steiner T Dvir O Szekely P Szekely A Ginsburg R Asor R Resh 856

C Tamburu M Peres and U Raviv Soft Matter 2011 7 1512ndash1523 857

59 C Tanford The hydrophobic effect Formation of micelles and biological membranes 858

John Wiley amp Sons Inc New York 1973 859

60 M Masuda and T Shimizu Langmuir 2004 20 5969ndash5977 860

33

61 A Zidovska K K Ewert J Quispe B Carragher C S Potter and C R Safinya 861

Block Liposomes Vesicles of Charged Lipids with Distinctly Shaped Nanoscale 862

Sphere- Pear- Tube- or Rod-Segments Elsevier Inc 1st edn 2009 vol 465 863

62 A Iglic V Kralj-Iglic B Božič M Bobrowska-Haumlgerstrand B Isomaa and H 864

Haumlgerstrand Bioelectrochemistry 2000 52 203ndash211 865

63 T Shimizu M Masuda and H Minamikawa Chem Rev 2005 105 1401ndash1444 866

64 J V Selinger and J M Schnur Phys Rev Lett 1993 71 4091ndash4094 867

65 J M Schnur Science (80- ) 1993 262 1669ndash1676 868

66 B Tenchov R Koynova and G Rapp Biophys J 2001 80 1873ndash1890 869

67 R Koynova and B Tenchov Ser Biomech 2017 31 3ndash19 870

68 S J Angyal and R Le Fur Carbohydr Res 1984 126 15ndash26 871

69 J K Whitesell and M A Minton J Am Chem Soc 1987 109 225ndash228 872

70 T Takayama S Ando and I Ando J Mol Struct 1990 220 245ndash250 873

71 K Gawrisch N V Eldho and I V Polozov Chem Phys Lipids 2002 116 135ndash151 874

72 F Tielens C Gervais J F Lambert F Mauri and D Costa Chem Mater 2008 20 875

3336ndash3344 876

73 G A Jeffrey and Y Yeon Acta Crystallogr Sect B 1986 42 410ndash413 877

74 B Berglund and R W Vaughan J Chem Phys 1980 73 2037ndash2043 878

75 C Gervais C Coelho T Azaiumls J Maquet G Laurent F Pourpoint C Bonhomme 879

P Florian B Alonso G Guerrero P H Mutin and F Mauri J Magn Reson 2007 880

187 131ndash140 881

76 C Akita T Kawaguchi F Kaneko H Yamamoto and M Suzuki J Phys Chem B 882

2004 108 4862ndash4868 883

77 R N Purusottam L Seacutenicourt J J Lacapegravere and P Tekely Biochim Biophys Acta - 884

Biomembr 2015 1848 3134ndash3139 885

78 X Biarneacutes A Ardegravevol A Planas C Rovira A Laio and M Parrinello J Am Chem 886

Soc 2007 129 10686ndash10693 887

79 L M J Kroon‐Batenburg and J Kroon Biopolymers 1990 29 1243ndash1248 888

80 K N Kirschner and R J Woods Proc Natl Acad Sci U S A 2001 98 10541ndash889

10545 890

81 S Svenson B Kirste and J H Fuhrhop J Am Chem Soc 1994 116 11969ndash11975 891

82 I Sack S Macholl J H Fuhrhop and G Buntkowsky Phys Chem Chem Phys 892

2000 2 1781ndash1788 893

83 S Svenson J Koening and J H Fuhrhop J Phys Chem 1994 98 1022ndash1028 894

34

84 S Svenson A Schaefer and J H Fuhrhop J Chem Soc Perkin Trans 2 1994 2 895

1023ndash1028 896

85 K Bock and C Pedersen Adv Carbohydr Chem Biochem 1983 41 27ndash66 897

86 T B Grindley Structure and Conformation of Carbohydrates Springer Berlin 898

Heidelberg 2008 899

87 F Tantakitti J Boekhoven X Wang R V Kazantsev T Yu J Li E Zhuang R 900

Zandi J H Ortony C J Newcomb L C Palmer G S Shekhawat M O De La 901

Cruz G C Schatz and S I Stupp Nat Mater 2016 15 469ndash476 902

903

1

Supporting Information 904

905

Topological connection between vesicles and nanotubes in single-906

component lipid membranes driven by head-tail interactions 907

908

Niki Baccilea Ceacutedric Lorthioira Abdoul Aziz Baa Patrick Le Griela Cristina Coelhob Javier 909

Perezc Wim Soetaertd Sophie L K W Roelantsd 910

911

a Sorbonne Universiteacute Centre National de la Recherche Scientifique Laboratoire de Chimie de 912

la Matiegravere Condenseacutee de Paris LCMCP F-75005 Paris France 913

b Institut des Mateacuteriaux de Paris Centre Sorbonne Universiteacute Paris France 914

c Synchrotron Soleil LrsquoOrme des Merisiers Saint-Aubin BP48 91192 Gif-sur-Yvette Cedex 915

France 916

d InBio Department of Biotechnology Ghent University Ghent Belgium 917

918

919

Corresponding author 920

Dr Niki Baccile 921

E-mail address nikibaccilesorbonne-universitefr 922

Phone +33 1 44 27 56 77 923

924

925

926

2

a) 927

Parameter Method

Dry Matter (DM ) 996 Infrared balance 105degC

Glucose () 012 HPLC-Metacarb

Glycerol () nd HPLC-Metacarb

Free Fatty Acid content () lt001 Internal method BBEPP

GC with FID Detector

Oil () lt001 Internal method BBEPP

GC with FID Detector

Moulds (CFUg) lt10 3M-Nordval ndeg16 B

Yeasts (CFUg) lt10 3M-Nordval ndeg16 B

Anaerobic count (CFUg) lt10 SP-VG M005 B

Protein () tbd BCA protein assay

928

Figure S 1 - a) HPLC-ELSD chromatogram and composition table 929

930

931

932

933

3

b) 934

935

Group Ndeg δ ppm

CH2 1 354

CH2 2 153

CH2 3-712-14 132

CH2 811 203

CH 910 534

CH2 15 162

CH2 16 388

CH 1 424

CH 2 316

CH 3 334

CH 4 329

CH 5 354

CH2 6 366

Figure S 1 (next) ndash b) 1H NMR spectrum (MeOD-d4) recorded on the purified G-C181-OH compounds 936

and related assignment 937

938

2

811910 1rsquo

5rsquo1

3rsquo4rsquo

1rsquo1910

asymp asymp asympHDO MeOD Acetone

6rsquo2rsquo

1615

3-712-14

4

939

940

Figure S 2 ndash Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 130degC 941

5

942

943

Figure S 3 - Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 90degC 944

945

90degC

1

22

6

946

947

948

949

Figure S 4 ndash Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 70degC 950

951

21

11

1

2

5

3

4

45

5

6

7

7

7

952

953

954

Figure S 5 ndash Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 25degC 955

956

957

1

1 1

8

958

Figure S 6 ndash a) SANS experiment recorded at 90degC for a 5 mgmL G-C181-OH aqueous solution b-c) 959

WAXS experiments recorded on aqueous solutions of G-C181-OH at b) 5 mgmL and c) 25 mgmL d) 960

Structural interpretation of the repeating distances pointed out in c) 961

962

963

964

001 01 1

q Aring-1

25degC

68degC

88degC

111degC

01 1q Aring

-1

02 04 06 08 1q Aring

-1

001 01

10-1

100

101

Inte

nsity

cm

-1

q Aring-1

-2

T= 90degC

SANS WAXS WAXS

a) b) c)

dv

dL d= 144 Aring-1

di

ddH2O

dt

dvdL

dtdi d

d)

9

965

966

Figure S 7 ndash ssNMR MAS (10 kHz) experiments recorded on a G-C181-OH dry powder a) 1H spectra 967

recorded from 110degC to 70degC b) 13C single pulse (SP) spectra recorded with high power 1H decoupling from 968

110degC to 70degC c) 13C cross polarization (CP) spectra recorded at 110degC and contact time tc of 50 micros and 969

3000 micros d) 13C CP spectra recorded at 70degC and for 20 lt tc micros lt 3000 970

971

1

811

910

CH2 CH2

1rsquo

6rsquo

8 6 4 2 0 ppm

110

90

80

70

910

1rsquo

6rsquo16

1

2rsquo3rsquo4rsquo

811

OH

45 40 35 30 25 20 15

ppm

200 180 160 140 120 100 80 60 40 20 0

ppm140 120 100 80

ppm

CH2-gauche

3004

110

90

80

70

65

60

50

25

C91013003

C9 - 12869

CH2-all-trans

3198

C1rsquo10335

a) b)

1H

T degCT degC

13C Single Pulse MAS

A

B

A

B

70degC70degC

5rsquo

215

200 180 160 140 120 100 80 60 40 20 0

ppm

13C CPMAST= 110degC

200 180 160 140 120 100 80 60 40 20 0

ppm

132 131 130 129 128 127 126

ppm

106 105 104 103 102

ppm

13C CPMAST= 70degC

50

tc micros

3000

20

tc micros

50 500 1000 3000

tc micros

C102

13024 C910113003

C92

12869

C1rsquo213082

C1rsquo110443

A B

106 105 104 103 102

ppm

3000

1000

500

50

20

AB

c)

d)

10

972

Figure S 8 ndash a) Short and d) long T2 components displayed by the 1H transverse relaxation signal c) 1H 973

transverse relaxation signals recorded on a G-C181-OH dry powder from 110degC to 70degC The area of each 974

given resonance is plotted (log-lin) as a function of the echo time (CPMG pulse sequence) for each 975

temperature and fitted using the equation given above T2s and T2l indicate the short and long T2 relaxation 976

times respectively obtained by describing the 1H transverse relaxation signal using a biexponential decay 977

For H1rsquo H5rsquo and H6rsquo the 1H relaxation function could be described by a monoexponential decay 978

- Protons from the sugar group (H1rsquo H5rsquo H6rsquo) show a single T2 relaxation component (Figure 979

6c and Figure S 8) while protons of the acyl chain (CH2 H910 H811) show both short (Figure 980

6c) and long (Figure 6d) T2 relaxation component 981

- The short T2 relaxation time systematically decreases from 110degC to 70degC for all species by a 982

factor ranging from two to seven depending on the specific chemical group Very long T2 are 983

observed at 110degC for all acyl species while nearly unchanged values are measured below 984

90degC without any clear trend 985

- The sugar protons (H1rsquo H5rsquo H6rsquo) have the shortest set of T2 values (lt 1 ms) below 100degC 986

The long component of T2 generally ranges between 4 ms and 12 ms 987

0 1 2 3 4 5 6 7 8

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1 2 3 4 5 6 7 8

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

H1

H8H11

H9H10

CH2

H1rsquo

H6rsquo

CH2

H910H811 H1CH2

H1rsquo H5rsquo6rsquo

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

OH

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45 T= 110degC

T= 90degC

T= 80degC

T= 70degC

Log

In

ten

sity

Time ms

Fit

)

CH2 H910 H811 H1 H56 H1 OH

00

05

10

15

20

25

30

35

40

T2 m

s

T= 110degC

T= 90degC

T= 80degC

T= 70degC

CH2 H910 H811 H1

0

2

4

6

8

10

12

T2 m

s

T= 110degC

T= 90degC

T= 80degC

T= 70degC

Short T2(1H) Long T2(1H)

a) b)

c)

11

Table S 1 ndash 1H chemical shifts of G-C181-OH in solution and in the solid-state Solid-state spectra are 988

recorded at a MAS frequency of 10 kHz in DMSO-d6 broad peak 989

990

δH ppm

Group Ndeg Liquid

(MeOD-d4) Solid

(T= 110degC) Solid

(T= 70degC)

CH2 H1 354 357

CH2 H2 153 156

CH2 H3-712-14 132 135

CH2 H811 203 203

CH H910 534 534

CH2 H15 162 164

CH2 H16 388 383

CH H1 424 432

CH H2 316 335

CH H3 334 335

CH H4 329 335

CH H5 354 351 357

CH2 H6 366 383

OH OH1 432

45 463

OH OH2 492

OH OH3 490

OH OH4 487

OH OH6 445

991

992

H1H9H10H1rsquo

H6rsquo

OH2rsquo

OH1

12

Detailed discussion about the NOESY and DQ-SQ 2D 1H-1H ssNMR experiments presented in 993

Figure 9 994

995

2D 1H-1H ssNMR experiments are a unique tool to explore through-space proximities 996

between 1H nuclei ssNMR offers a large number of possibilities each adapted to a given 997

system The advantage of 2D maps with respect to a dynamics approach is their direct 998

readability in clear two nuclei which exchange their magnetization through-space (whichever 999

mechanism may be responsible for this exchange) will display a specific cross-peak However 1000

the way cross-peaks are presented and interpreted strictly depends on the nature of the pulse 1001

program which in turn is chosen according to the physical nature of the sample 1002

1D temperature-dependent 1H NMR experiments performed on G-C181-OH (Figure 1003

4a in the main text) indicate sharp peaks at 90degC and broad peaks at 70degC Narrow peaks 1004

indicate fast andor large-amplitude molecular tumbling averaging the strong 1H-1H dipolar 1005

interactions This excludes the use of pulse sequences like DQ-SQ which is not efficient under 1006

these conditions For this reason we employ spin-diffusion-based pulse sequences like 1007

NOESY which is more adapted for samples with high molecular mobility When mixing times 1008

are chosen to be short (lt 1 s) one generally explores through-space proximities within about 5 1009

10 Aring On the contrary when molecular motions are restricted dipolar coupling becomes 1010

significant 1H NMR peaks are broader and pulse sequences like DQ-SQ can now be used 1011

with a spatial resolution of less than 5 Aring 1012

Whenever possible DQ-SQ experiments are preferred over spin diffusion-based pulse 1013

sequences because they are more selective in particular for likewise interactions which are 1014

not discriminated in NOESY-type experiments This is illustrated by the typical interaction 1015

schemes for NOESY (top Figure 9a) and DQ-SQ (top Figure 9b) For two given A and B 1016

protons 2D NOESY maps will display two on-diagonal and two square-shaped symmetrical 1017

off-diagonal cross-peaks Off-diagonal cross-peaks illustrate the mutual A-B and B-A 1018

interaction but the on-diagonal peaks are systematic and do not prove any A-A or B-B 1019

interaction Similarly to NOESY 2D DQ-SQ maps display off-diagonal peaks (symmetric 1020

across the diagonal top Figure 9b) upon A-B and B-A interaction but in contrast they will 1021

display on-diagonal peaks only in the presence of self-interaction In the tutorial example of 1022

Figure 9b top the on-diagonal A-A peak indicates mutual A-A interaction while lack of on-1023

diagonal B-B peak evidences the absence of B-B proximities Combining NOESY and DQ-SQ 1024

experiments is capital to study the local arrangement of G-C181-OH and its evolution with 1025

temperature 1026

13

At T= 90degC the H1 and H5rsquo contributions are very close but possible to discriminate 1027

them (δ(H1)= 354 ppm δ(H5rsquo)= 351 ppm) according to 1H NMR chemical shifts in Table S 1028

1 2D 1H-1H NOESY in Figure 9a also discriminates between the two components attributed 1029

on the basis of the interactions with their mutual neighbors (green H1-H2 and H5rsquo-H1rsquo squares 1030

in Figure 9a) On this basis 2D NOESY does not show a direct H1rsquo-H1 correlation (blue red-1031

crossed) but rather an extensive network of intra-sugar interactions These features exclude 1032

suggest intra- and inter-sugar interactions and they exclude head-tail arrangements as a major 1033

molecular configuration at 90degC 1034

At 70degC the 1H NMR peaks are broader indicating a reduction of the T2 (1H) relaxation 1035

times andor residual homonuclear dipolar interactions excluding the use of NOESY In this 1036

case 1H-1H DQ-SQ correlation experiments are best suited Figure 9b presents the 1H-1H DQ-1037

SQ experiment on G-C181-OH at 70degC Peaks are broader than in NOESY and attribution may 1038

be cumbersome However the experiment is performed at 700 MHz and MAS= 22 kHz which 1039

improves spectral resolution with respect to the same experiment performed at 300 MHz and 1040

MAS= 10 kHz The auto-peak at δH= 356 ppm in the SQ dimension is attributed to H1-H1 spin 1041

pairs mainly resulting from intra CH2 interaction while the DQ cross-peaks at δH= 156-356 1042

ppm and δH= 432-356 ppm respectively origin from H2-H1 and H1rsquo-H1 interactions (Figure 1043

9b) DQ-SQ map does not show any auto-peak corresponding to H1rsquo-H1rsquo or Hxrsquo-Hxrsquo (x= 2rsquo 1044

3rsquo 4rsquo) interactions (red-crossed blue Figure 9b) thus excluding any proximity (lt 5 Aring) between 1045

close glucose neighbors On the contrary DQ-SQ maps (recorded both at 700 MHz Figure 9b 1046

or at 300 MHz not shown) systematically show DQ cross-peaks which can only be attributed 1047

to H1rsquo-H1 interactions (δH= 432-356) The presence of these H1rsquo-H1 interactions can only be 1048

explained by a head-tail arrangement of the G-C181-OH molecules as illustrated below the 1049

DQ-SQ map in Figure 9b 1050

28

sugar headgroup despite a certain flexibility of the chain (long T2 component) This trend may 691

suggest restrictions in translational mobility probably due to the sugar-OH hydrogen bonding 692

(Figure 8c) meaning that lipids are not able to diffuse within the membrane thus keeping two 693

different configurations stable 694

695

Conclusions 696

Topologically-connected nanotube-vesicle lipidic systems are commonly observed for 697

self-assembled membranes composed of at least two amphiphiles and spontaneously driven by 698

anisotropic inclusions andor by external directional forces This work shows that strong 699

differences in membrane curvature can spontaneously occur in single-lipid systems without 700

external anisotropic forces Coexistence and interconnection between nanotubes and vesicles 701

are evidenced via cryo-TEM and fluorescence confocal microscopy in a single-phase region at 702

temperature above a transition temperature Tm1= 483degC determined by DSC on the hydrated 703

sample but below the second transition temperature Tm2 (957degC) determined by DSC on the 704

dry powder X-ray diffraction (WAXS) also shows the coexistence of two lattice periods one 705

attributed to an inter-membrane repeating distance and the other possibly related to the intra-706

nanotube glucolipid arrangement Labelling the systems with Laurdan a fluorophore of which 707

the emission strongly varies with the local membrane rigidity excludes the presence of rigid 708

micro-separated phases (eg Lβ) 709

1H and 13C ssNMR despite the fact that it can only be performed on the dry powder 710

confirm the presence of one main fluid G-C181-OH environment above Tm2 Below Tm2 the 711

splitting of some 13C peaks and the coexistence of short and long 1H relaxation components 712

demonstrate that G-C181-OH can simultaneously pack into structurally and dynamically 713

inequivalent environments Combining advanced 2D 1H-1H correlation experiments based on 714

spin diffusion or double-quanta excitation suggests that vesicles could contain a preferred head-715

head configuration of the bolaform lipid while nanotubes a preferred head-tail assembly 716

The presence of two environments composed of the same molecule characterized by 717

different molecular configurations and rigidity fulfills the theoretical conditions necessary to 718

spontaneously nucleate a membrane inclusion of non-zero average orientation described as 719

necessary to observe topologically-connected nanotube-vesicle structures in the absence of 720

external forces This hypothesis could also be enriched by the possibility of nanotubes to be 721

pulled away between two vesicles after a collision event 722

This work expands the actual theoretical and experimental framework on lipid networks 723

stabilized by a different radius of curvature It could also motivate the synthesis or isolation 724

29

of new lipids providing a higher level of 3D architectural control in soft networks possibly in 725

the range of room temperature 726

727

Acknowledgements 728

Dr Andrea Lassenberger and Dr Sylvain Preacutevost at Institut Laue Langevin (ILL Grenoble 729

France) are kindly acknowledged for their assistance on the SANS experiment Confocal 730

images were performed at the Institut de Biologie Paris Seine (IBPS) imaging facility Jean-731

Franccedilois Gilles (Imaging Core Facility Institut de Biologie Paris Seine (IBPS) CNRS 732

Sorbonne Universiteacute Paris France) is kindly acknowledged for his assistance on the confocal 733

microscopy imaging Dr Guillaume Laurent (LCMCP Sorbonne Universiteacute Paris France) is 734

kindly acknowledged for helpful discussions The French region Ile-de-France SESAME 735

program is acknowledged for financial support (700 MHz NMR 736

spectrometer) Dr Olivier Diat (ICSM Marcoule France) Prof eronikaIglič and Prof V Aleš737

onsul discussi Slovenia) are kindly acknowledged for helpfUniversity of LjubljanaIglič (-Kralj 738

739

Financial support 740

WAXS experiments have been supported by Soleil Light Source Saint Aubin France proposal 741

Ndeg 20201747 SANS experiments have been supported by ILL proposal Ndeg9-13-778 742

743

References 744

1 D Kabaso N Bobrovska W Goacuteźdź E Gongadze V Kralj-Iglič R Zorec and A 745

Iglič Bioelectrochemistry 2012 87 204ndash210 746

2 C R Safinya U Raviv D J Needleman A Zidovska M C Choi M A Ojeda-747

Lopez K K Ewert Y Li H P Miller J Quispe B Carragher C S Potter M W 748

Kim S C Feinstein and L Wilson Adv Mater 2011 23 2260ndash2270 749

3 P Veranič M Lokar G J Schuumltz J Weghuber S Wieser H Haumlgerstrand V Kralj-750

Iglič and A Iglič Biophys J 2008 95 4416ndash4425 751

4 A Rustom R Saffrich I Markovic P Walther and H-H Gerdes Science (80- ) 752

2004 303 1007ndash1010 753

5 M I Angelova S Soleacuteau P Meacuteleacuteard F Faucon and P Bothorel Prog Colloid 754

Polym Sci 1992 89 127ndash131 755

6 L Mathivet S Cribier and P F Devaux Biophys J 1996 70 1112ndash1121 756

7 V Kralj-Iglič G Gomišček J Majhenc V Arrigler and S Svetina Colloids Surfaces 757

A Physicochem Eng Asp 2001 181 315ndash318 758

30

8 V Kralj-Iglič A Iglič G Gomišček F Sevšek V Arrigler and H Haumlgerstrand J 759

Phys A Math Gen 2002 35 1533ndash1549 760

9 M Drab D Stopar V Kralj-Iglič and A Iglič Cells 2019 8 626 761

10 A Zidovska K K Ewert J Quispe B Carragher C S Potter and C R Safinya 762

Langmuir 2009 25 2979ndash2985 763

11 A Zidovska K K Ewert J Quispe B Carragher C S Potter and C R Safinya Soft 764

Matter 2011 7 8363ndash8369 765

12 A Zidovska K K Ewert J Quispe B Carragher C S Potter and C R Safinya 766

Biochim Biophys Acta - Biomembr 2009 1788 1869ndash1876 767

13 M Karlsson K Sott M Davidson A S Cans P Linderholm D Chiu and O Orwar 768

Proc Natl Acad Sci U S A 2002 99 11573ndash11578 769

14 A Karlsson R Karlsson M Karlsson A-S Cans A Stromberg F Ryttsen and O 770

Orwar Nature 2001 409 150ndash152 771

15 M Karlsson K Sott A S Cans A Karlsson R Karlsson and O Orwar Langmuir 772

2001 17 6754ndash6758 773

16 E Evans H Bowman A Leung D Needham and D Tirrell Science (80- ) 1996 774

273 933ndash935 775

17 U Seifert Configurations of fluid membranes and vesicles 1997 vol 46 776

18 J Kaumls and E Sackmann Biophys J 1991 60 825ndash844 777

19 N Bobrovska W Goacuteźdź V Kralj-Iglič and A Iglič PLoS One 2013 8 e73941 778

20 A Iglič B Babnik K Bohinc M Fošnarič H Haumlgerstrand and V Kralj-Iglič J 779

Biomech 2007 40 579ndash585 780

21 K Bohinc T Slivnik A Iglič M Brumen and V Kralj-Iglič J Phys Chem C 2007 781

111 9709ndash9718 782

22 A Iglič H Haumlgerstrand P Veranič A Plemenitaš and V Kralj-Iglič J Theor Biol 783

2006 240 368ndash373 784

23 V Kralj-Iglič and P Veranič Adv Planar Lipid Bilayers Liposomes 2006 5 129ndash785

149 786

24 A Iglic and V Kralj-Iglic Adv Planar Lipid Bilayers Liposomes 2006 4 253ndash279 787

25 S Penič L Mesarec M Fošnarič L Mroacutewczyńska H Haumlgerstrand V Kralj-Iglič and 788

A Iglič Front Phys 2020 8 342 789

26 P Yager and P E Schoen Mol Cryst Liq Cryst 1984 106 371ndash381 790

27 T G Barclay K Constantopoulos and J Matisons Chem Rev 2014 114 10217ndash791

10291 792

31

28 B Božič G Gomišček V Kralj-Iglič S Svetina and B Žekš Eur Biophys J 2002 793

31 487ndash496 794

29 A Iglič H Haumlgerstrand M Bobrowska-Haumlgerstrand V Arrigler and V Kralj-Iglič 795

Phys Lett A 2003 310 493ndash497 796

30 R Bar-Ziv and E Moses Phys Rev Lett 1994 73 1392ndash1395 797

31 S Chiruvolu H E Warriner E Naranjo S H Idziak J O Raumldler R J Plano J A 798

Zasadzinski and C R Safinya Science (80- ) 1994 266 1222ndash1225 799

32 V Kralj-Iglič A Iglič H Haumlgerstrand and P Peterlin Phys Rev E 2000 61 4230ndash800

4234 801

33 L Mesarec M Drab S Penič V Kralj-Iglič and A Iglič Int J Mol Sci 2021 22 802

2348 803

34 L Van Renterghem S L K W Roelants N Baccile K Uyttersprot M C Taelman 804

B Everaert S Mincke S Ledegen S Debrouwer K Scholtens C V Stevens W 805

Soetaert L Van Renterghem S L K W Roelants N Baccile K Uytersprot M C 806

Taelman B Everaert S Mincke S Ledegen S Debrouwer K Scholtens C V 807

Stevens and W Soetaert Biotechnol Bioeng 2018 115 1195ndash1206 808

35 J Clark and F Deswarte Introduction to Chemicals from Biomass Second Edition 809

John Wiley amp Sons 2015 810

36 K V Sajna R Houmlfer R K Sukumaran L D Gottumukkala and A Pandey in 811

Industrial Biorefineries and White Biotechnology eds A Pandey R Houmlfer M 812

Taherzadeh K M Nampoothiri and C Larroche Amsterdam Oxford Waltham 813

Elsevier 2015 pp 499ndash521 814

37 R Marchant and I M Banat Trends Biotechnol 2012 30 558ndash565 815

38 K Hill and C LeHen-Ferrenbach in Sugar-Based Surfactants Fundamentals and 816

Applications ed C C Ruiz Boca Raton CRC Press 2009 pp 1ndash20 817

39 L Rodrigues I M Banat J Teixeira and R Oliveira J Antimicrob Chemother 818

2006 57 609ndash618 819

40 J D Desai and I M Banat Microbiol Mol Biol Rev 1997 61 47ndash64 820

41 S L K W Roelants K Ciesielska S L De Maeseneire H Moens B Everaert S 821

Verweire Q Denon B Vanlerberghe I N A Van Bogaert P Van der Meeren B 822

Devreese and W Soetaert Biotechnol Bioeng 2016 113 550ndash559 823

42 N Baccile M Selmane P Le Griel S Preacutevost J Perez C V Stevens E Delbeke S 824

Zibek M Guenther W Soetaert I N A Van Bogaert and S Roelants Langmuir 825

2016 32 6343ndash6359 826

32

43 N Baccile A-S Cuvier S Preacutevost C V Stevens E Delbeke J Berton W Soetaert 827

I N A Van Bogaert and S Roelants Langmuir 2016 32 10881ndash10894 828

44 N Baccile C Seyrig A Poirier S A Castro S L K W Roelants and S Abel Green 829

Chem 2021 23 3842ndash3944 830

45 A Salis L Cappai C Carucci D F Parsons and M Monduzzi J Phys Chem Lett 831

2020 11 6805ndash6811 832

46 J Schindelin I Arganda-Carreras E Frise V Kaynig M Longair T Pietzsch S 833

Preibisch C Rueden S Saalfeld B Schmid J-Y Tinevez D J White V 834

Hartenstein K Eliceiri P Tomancak and A Cardona Nat Methods 2012 9 676ndash835

682 836

47 T Parasassi and E Gratton J Fluoresc 1995 5 59ndash69 837

48 M M Dodes Traian F L Gonzaacutelez Flecha and V Levi J Lipid Res 2012 53 609ndash838

616 839

49 D M Owen C Rentero A Magenau A Abu-Siniyeh and K Gaus Nat Protoc 840

2012 7 24ndash35 841

50 W Stillwell in An Introduction to Biological Membranes Academic Press 2016 pp 842

135ndash180 843

51 W Stillwell in An Introduction to Biological Membranes Academic Press 2016 pp 844

181ndash220 845

52 B Tenchov Chem Phys Lipids 1991 57 165ndash177 846

53 R Koynova and M Caffrey Biochim Biophys Acta 1998 1376 91ndash145 847

54 F Schmid S Dolezel O Lenz and S Meinhardt J Phys Conf Ser 2014 487 848

012004 849

55 S Meinhardt R L C Vink and F Schmid Proc Natl Acad Sci U S A 2013 110 850

4476ndash4481 851

56 O Glatter and O Kratky Small Angle X-ray Scattering Academic Press London 852

1982 853

57 B A Cunningham A D Brown D H Wolfe W P Williams and A Brain Phys 854

Rev E 1998 58 3662ndash3672 855

58 M Nadler A Steiner T Dvir O Szekely P Szekely A Ginsburg R Asor R Resh 856

C Tamburu M Peres and U Raviv Soft Matter 2011 7 1512ndash1523 857

59 C Tanford The hydrophobic effect Formation of micelles and biological membranes 858

John Wiley amp Sons Inc New York 1973 859

60 M Masuda and T Shimizu Langmuir 2004 20 5969ndash5977 860

33

61 A Zidovska K K Ewert J Quispe B Carragher C S Potter and C R Safinya 861

Block Liposomes Vesicles of Charged Lipids with Distinctly Shaped Nanoscale 862

Sphere- Pear- Tube- or Rod-Segments Elsevier Inc 1st edn 2009 vol 465 863

62 A Iglic V Kralj-Iglic B Božič M Bobrowska-Haumlgerstrand B Isomaa and H 864

Haumlgerstrand Bioelectrochemistry 2000 52 203ndash211 865

63 T Shimizu M Masuda and H Minamikawa Chem Rev 2005 105 1401ndash1444 866

64 J V Selinger and J M Schnur Phys Rev Lett 1993 71 4091ndash4094 867

65 J M Schnur Science (80- ) 1993 262 1669ndash1676 868

66 B Tenchov R Koynova and G Rapp Biophys J 2001 80 1873ndash1890 869

67 R Koynova and B Tenchov Ser Biomech 2017 31 3ndash19 870

68 S J Angyal and R Le Fur Carbohydr Res 1984 126 15ndash26 871

69 J K Whitesell and M A Minton J Am Chem Soc 1987 109 225ndash228 872

70 T Takayama S Ando and I Ando J Mol Struct 1990 220 245ndash250 873

71 K Gawrisch N V Eldho and I V Polozov Chem Phys Lipids 2002 116 135ndash151 874

72 F Tielens C Gervais J F Lambert F Mauri and D Costa Chem Mater 2008 20 875

3336ndash3344 876

73 G A Jeffrey and Y Yeon Acta Crystallogr Sect B 1986 42 410ndash413 877

74 B Berglund and R W Vaughan J Chem Phys 1980 73 2037ndash2043 878

75 C Gervais C Coelho T Azaiumls J Maquet G Laurent F Pourpoint C Bonhomme 879

P Florian B Alonso G Guerrero P H Mutin and F Mauri J Magn Reson 2007 880

187 131ndash140 881

76 C Akita T Kawaguchi F Kaneko H Yamamoto and M Suzuki J Phys Chem B 882

2004 108 4862ndash4868 883

77 R N Purusottam L Seacutenicourt J J Lacapegravere and P Tekely Biochim Biophys Acta - 884

Biomembr 2015 1848 3134ndash3139 885

78 X Biarneacutes A Ardegravevol A Planas C Rovira A Laio and M Parrinello J Am Chem 886

Soc 2007 129 10686ndash10693 887

79 L M J Kroon‐Batenburg and J Kroon Biopolymers 1990 29 1243ndash1248 888

80 K N Kirschner and R J Woods Proc Natl Acad Sci U S A 2001 98 10541ndash889

10545 890

81 S Svenson B Kirste and J H Fuhrhop J Am Chem Soc 1994 116 11969ndash11975 891

82 I Sack S Macholl J H Fuhrhop and G Buntkowsky Phys Chem Chem Phys 892

2000 2 1781ndash1788 893

83 S Svenson J Koening and J H Fuhrhop J Phys Chem 1994 98 1022ndash1028 894

34

84 S Svenson A Schaefer and J H Fuhrhop J Chem Soc Perkin Trans 2 1994 2 895

1023ndash1028 896

85 K Bock and C Pedersen Adv Carbohydr Chem Biochem 1983 41 27ndash66 897

86 T B Grindley Structure and Conformation of Carbohydrates Springer Berlin 898

Heidelberg 2008 899

87 F Tantakitti J Boekhoven X Wang R V Kazantsev T Yu J Li E Zhuang R 900

Zandi J H Ortony C J Newcomb L C Palmer G S Shekhawat M O De La 901

Cruz G C Schatz and S I Stupp Nat Mater 2016 15 469ndash476 902

903

1

Supporting Information 904

905

Topological connection between vesicles and nanotubes in single-906

component lipid membranes driven by head-tail interactions 907

908

Niki Baccilea Ceacutedric Lorthioira Abdoul Aziz Baa Patrick Le Griela Cristina Coelhob Javier 909

Perezc Wim Soetaertd Sophie L K W Roelantsd 910

911

a Sorbonne Universiteacute Centre National de la Recherche Scientifique Laboratoire de Chimie de 912

la Matiegravere Condenseacutee de Paris LCMCP F-75005 Paris France 913

b Institut des Mateacuteriaux de Paris Centre Sorbonne Universiteacute Paris France 914

c Synchrotron Soleil LrsquoOrme des Merisiers Saint-Aubin BP48 91192 Gif-sur-Yvette Cedex 915

France 916

d InBio Department of Biotechnology Ghent University Ghent Belgium 917

918

919

Corresponding author 920

Dr Niki Baccile 921

E-mail address nikibaccilesorbonne-universitefr 922

Phone +33 1 44 27 56 77 923

924

925

926

2

a) 927

Parameter Method

Dry Matter (DM ) 996 Infrared balance 105degC

Glucose () 012 HPLC-Metacarb

Glycerol () nd HPLC-Metacarb

Free Fatty Acid content () lt001 Internal method BBEPP

GC with FID Detector

Oil () lt001 Internal method BBEPP

GC with FID Detector

Moulds (CFUg) lt10 3M-Nordval ndeg16 B

Yeasts (CFUg) lt10 3M-Nordval ndeg16 B

Anaerobic count (CFUg) lt10 SP-VG M005 B

Protein () tbd BCA protein assay

928

Figure S 1 - a) HPLC-ELSD chromatogram and composition table 929

930

931

932

933

3

b) 934

935

Group Ndeg δ ppm

CH2 1 354

CH2 2 153

CH2 3-712-14 132

CH2 811 203

CH 910 534

CH2 15 162

CH2 16 388

CH 1 424

CH 2 316

CH 3 334

CH 4 329

CH 5 354

CH2 6 366

Figure S 1 (next) ndash b) 1H NMR spectrum (MeOD-d4) recorded on the purified G-C181-OH compounds 936

and related assignment 937

938

2

811910 1rsquo

5rsquo1

3rsquo4rsquo

1rsquo1910

asymp asymp asympHDO MeOD Acetone

6rsquo2rsquo

1615

3-712-14

4

939

940

Figure S 2 ndash Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 130degC 941

5

942

943

Figure S 3 - Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 90degC 944

945

90degC

1

22

6

946

947

948

949

Figure S 4 ndash Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 70degC 950

951

21

11

1

2

5

3

4

45

5

6

7

7

7

952

953

954

Figure S 5 ndash Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 25degC 955

956

957

1

1 1

8

958

Figure S 6 ndash a) SANS experiment recorded at 90degC for a 5 mgmL G-C181-OH aqueous solution b-c) 959

WAXS experiments recorded on aqueous solutions of G-C181-OH at b) 5 mgmL and c) 25 mgmL d) 960

Structural interpretation of the repeating distances pointed out in c) 961

962

963

964

001 01 1

q Aring-1

25degC

68degC

88degC

111degC

01 1q Aring

-1

02 04 06 08 1q Aring

-1

001 01

10-1

100

101

Inte

nsity

cm

-1

q Aring-1

-2

T= 90degC

SANS WAXS WAXS

a) b) c)

dv

dL d= 144 Aring-1

di

ddH2O

dt

dvdL

dtdi d

d)

9

965

966

Figure S 7 ndash ssNMR MAS (10 kHz) experiments recorded on a G-C181-OH dry powder a) 1H spectra 967

recorded from 110degC to 70degC b) 13C single pulse (SP) spectra recorded with high power 1H decoupling from 968

110degC to 70degC c) 13C cross polarization (CP) spectra recorded at 110degC and contact time tc of 50 micros and 969

3000 micros d) 13C CP spectra recorded at 70degC and for 20 lt tc micros lt 3000 970

971

1

811

910

CH2 CH2

1rsquo

6rsquo

8 6 4 2 0 ppm

110

90

80

70

910

1rsquo

6rsquo16

1

2rsquo3rsquo4rsquo

811

OH

45 40 35 30 25 20 15

ppm

200 180 160 140 120 100 80 60 40 20 0

ppm140 120 100 80

ppm

CH2-gauche

3004

110

90

80

70

65

60

50

25

C91013003

C9 - 12869

CH2-all-trans

3198

C1rsquo10335

a) b)

1H

T degCT degC

13C Single Pulse MAS

A

B

A

B

70degC70degC

5rsquo

215

200 180 160 140 120 100 80 60 40 20 0

ppm

13C CPMAST= 110degC

200 180 160 140 120 100 80 60 40 20 0

ppm

132 131 130 129 128 127 126

ppm

106 105 104 103 102

ppm

13C CPMAST= 70degC

50

tc micros

3000

20

tc micros

50 500 1000 3000

tc micros

C102

13024 C910113003

C92

12869

C1rsquo213082

C1rsquo110443

A B

106 105 104 103 102

ppm

3000

1000

500

50

20

AB

c)

d)

10

972

Figure S 8 ndash a) Short and d) long T2 components displayed by the 1H transverse relaxation signal c) 1H 973

transverse relaxation signals recorded on a G-C181-OH dry powder from 110degC to 70degC The area of each 974

given resonance is plotted (log-lin) as a function of the echo time (CPMG pulse sequence) for each 975

temperature and fitted using the equation given above T2s and T2l indicate the short and long T2 relaxation 976

times respectively obtained by describing the 1H transverse relaxation signal using a biexponential decay 977

For H1rsquo H5rsquo and H6rsquo the 1H relaxation function could be described by a monoexponential decay 978

- Protons from the sugar group (H1rsquo H5rsquo H6rsquo) show a single T2 relaxation component (Figure 979

6c and Figure S 8) while protons of the acyl chain (CH2 H910 H811) show both short (Figure 980

6c) and long (Figure 6d) T2 relaxation component 981

- The short T2 relaxation time systematically decreases from 110degC to 70degC for all species by a 982

factor ranging from two to seven depending on the specific chemical group Very long T2 are 983

observed at 110degC for all acyl species while nearly unchanged values are measured below 984

90degC without any clear trend 985

- The sugar protons (H1rsquo H5rsquo H6rsquo) have the shortest set of T2 values (lt 1 ms) below 100degC 986

The long component of T2 generally ranges between 4 ms and 12 ms 987

0 1 2 3 4 5 6 7 8

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1 2 3 4 5 6 7 8

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

H1

H8H11

H9H10

CH2

H1rsquo

H6rsquo

CH2

H910H811 H1CH2

H1rsquo H5rsquo6rsquo

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

OH

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45 T= 110degC

T= 90degC

T= 80degC

T= 70degC

Log

In

ten

sity

Time ms

Fit

)

CH2 H910 H811 H1 H56 H1 OH

00

05

10

15

20

25

30

35

40

T2 m

s

T= 110degC

T= 90degC

T= 80degC

T= 70degC

CH2 H910 H811 H1

0

2

4

6

8

10

12

T2 m

s

T= 110degC

T= 90degC

T= 80degC

T= 70degC

Short T2(1H) Long T2(1H)

a) b)

c)

11

Table S 1 ndash 1H chemical shifts of G-C181-OH in solution and in the solid-state Solid-state spectra are 988

recorded at a MAS frequency of 10 kHz in DMSO-d6 broad peak 989

990

δH ppm

Group Ndeg Liquid

(MeOD-d4) Solid

(T= 110degC) Solid

(T= 70degC)

CH2 H1 354 357

CH2 H2 153 156

CH2 H3-712-14 132 135

CH2 H811 203 203

CH H910 534 534

CH2 H15 162 164

CH2 H16 388 383

CH H1 424 432

CH H2 316 335

CH H3 334 335

CH H4 329 335

CH H5 354 351 357

CH2 H6 366 383

OH OH1 432

45 463

OH OH2 492

OH OH3 490

OH OH4 487

OH OH6 445

991

992

H1H9H10H1rsquo

H6rsquo

OH2rsquo

OH1

12

Detailed discussion about the NOESY and DQ-SQ 2D 1H-1H ssNMR experiments presented in 993

Figure 9 994

995

2D 1H-1H ssNMR experiments are a unique tool to explore through-space proximities 996

between 1H nuclei ssNMR offers a large number of possibilities each adapted to a given 997

system The advantage of 2D maps with respect to a dynamics approach is their direct 998

readability in clear two nuclei which exchange their magnetization through-space (whichever 999

mechanism may be responsible for this exchange) will display a specific cross-peak However 1000

the way cross-peaks are presented and interpreted strictly depends on the nature of the pulse 1001

program which in turn is chosen according to the physical nature of the sample 1002

1D temperature-dependent 1H NMR experiments performed on G-C181-OH (Figure 1003

4a in the main text) indicate sharp peaks at 90degC and broad peaks at 70degC Narrow peaks 1004

indicate fast andor large-amplitude molecular tumbling averaging the strong 1H-1H dipolar 1005

interactions This excludes the use of pulse sequences like DQ-SQ which is not efficient under 1006

these conditions For this reason we employ spin-diffusion-based pulse sequences like 1007

NOESY which is more adapted for samples with high molecular mobility When mixing times 1008

are chosen to be short (lt 1 s) one generally explores through-space proximities within about 5 1009

10 Aring On the contrary when molecular motions are restricted dipolar coupling becomes 1010

significant 1H NMR peaks are broader and pulse sequences like DQ-SQ can now be used 1011

with a spatial resolution of less than 5 Aring 1012

Whenever possible DQ-SQ experiments are preferred over spin diffusion-based pulse 1013

sequences because they are more selective in particular for likewise interactions which are 1014

not discriminated in NOESY-type experiments This is illustrated by the typical interaction 1015

schemes for NOESY (top Figure 9a) and DQ-SQ (top Figure 9b) For two given A and B 1016

protons 2D NOESY maps will display two on-diagonal and two square-shaped symmetrical 1017

off-diagonal cross-peaks Off-diagonal cross-peaks illustrate the mutual A-B and B-A 1018

interaction but the on-diagonal peaks are systematic and do not prove any A-A or B-B 1019

interaction Similarly to NOESY 2D DQ-SQ maps display off-diagonal peaks (symmetric 1020

across the diagonal top Figure 9b) upon A-B and B-A interaction but in contrast they will 1021

display on-diagonal peaks only in the presence of self-interaction In the tutorial example of 1022

Figure 9b top the on-diagonal A-A peak indicates mutual A-A interaction while lack of on-1023

diagonal B-B peak evidences the absence of B-B proximities Combining NOESY and DQ-SQ 1024

experiments is capital to study the local arrangement of G-C181-OH and its evolution with 1025

temperature 1026

13

At T= 90degC the H1 and H5rsquo contributions are very close but possible to discriminate 1027

them (δ(H1)= 354 ppm δ(H5rsquo)= 351 ppm) according to 1H NMR chemical shifts in Table S 1028

1 2D 1H-1H NOESY in Figure 9a also discriminates between the two components attributed 1029

on the basis of the interactions with their mutual neighbors (green H1-H2 and H5rsquo-H1rsquo squares 1030

in Figure 9a) On this basis 2D NOESY does not show a direct H1rsquo-H1 correlation (blue red-1031

crossed) but rather an extensive network of intra-sugar interactions These features exclude 1032

suggest intra- and inter-sugar interactions and they exclude head-tail arrangements as a major 1033

molecular configuration at 90degC 1034

At 70degC the 1H NMR peaks are broader indicating a reduction of the T2 (1H) relaxation 1035

times andor residual homonuclear dipolar interactions excluding the use of NOESY In this 1036

case 1H-1H DQ-SQ correlation experiments are best suited Figure 9b presents the 1H-1H DQ-1037

SQ experiment on G-C181-OH at 70degC Peaks are broader than in NOESY and attribution may 1038

be cumbersome However the experiment is performed at 700 MHz and MAS= 22 kHz which 1039

improves spectral resolution with respect to the same experiment performed at 300 MHz and 1040

MAS= 10 kHz The auto-peak at δH= 356 ppm in the SQ dimension is attributed to H1-H1 spin 1041

pairs mainly resulting from intra CH2 interaction while the DQ cross-peaks at δH= 156-356 1042

ppm and δH= 432-356 ppm respectively origin from H2-H1 and H1rsquo-H1 interactions (Figure 1043

9b) DQ-SQ map does not show any auto-peak corresponding to H1rsquo-H1rsquo or Hxrsquo-Hxrsquo (x= 2rsquo 1044

3rsquo 4rsquo) interactions (red-crossed blue Figure 9b) thus excluding any proximity (lt 5 Aring) between 1045

close glucose neighbors On the contrary DQ-SQ maps (recorded both at 700 MHz Figure 9b 1046

or at 300 MHz not shown) systematically show DQ cross-peaks which can only be attributed 1047

to H1rsquo-H1 interactions (δH= 432-356) The presence of these H1rsquo-H1 interactions can only be 1048

explained by a head-tail arrangement of the G-C181-OH molecules as illustrated below the 1049

DQ-SQ map in Figure 9b 1050

29

of new lipids providing a higher level of 3D architectural control in soft networks possibly in 725

the range of room temperature 726

727

Acknowledgements 728

Dr Andrea Lassenberger and Dr Sylvain Preacutevost at Institut Laue Langevin (ILL Grenoble 729

France) are kindly acknowledged for their assistance on the SANS experiment Confocal 730

images were performed at the Institut de Biologie Paris Seine (IBPS) imaging facility Jean-731

Franccedilois Gilles (Imaging Core Facility Institut de Biologie Paris Seine (IBPS) CNRS 732

Sorbonne Universiteacute Paris France) is kindly acknowledged for his assistance on the confocal 733

microscopy imaging Dr Guillaume Laurent (LCMCP Sorbonne Universiteacute Paris France) is 734

kindly acknowledged for helpful discussions The French region Ile-de-France SESAME 735

program is acknowledged for financial support (700 MHz NMR 736

spectrometer) Dr Olivier Diat (ICSM Marcoule France) Prof eronikaIglič and Prof V Aleš737

onsul discussi Slovenia) are kindly acknowledged for helpfUniversity of LjubljanaIglič (-Kralj 738

739

Financial support 740

WAXS experiments have been supported by Soleil Light Source Saint Aubin France proposal 741

Ndeg 20201747 SANS experiments have been supported by ILL proposal Ndeg9-13-778 742

743

References 744

1 D Kabaso N Bobrovska W Goacuteźdź E Gongadze V Kralj-Iglič R Zorec and A 745

Iglič Bioelectrochemistry 2012 87 204ndash210 746

2 C R Safinya U Raviv D J Needleman A Zidovska M C Choi M A Ojeda-747

Lopez K K Ewert Y Li H P Miller J Quispe B Carragher C S Potter M W 748

Kim S C Feinstein and L Wilson Adv Mater 2011 23 2260ndash2270 749

3 P Veranič M Lokar G J Schuumltz J Weghuber S Wieser H Haumlgerstrand V Kralj-750

Iglič and A Iglič Biophys J 2008 95 4416ndash4425 751

4 A Rustom R Saffrich I Markovic P Walther and H-H Gerdes Science (80- ) 752

2004 303 1007ndash1010 753

5 M I Angelova S Soleacuteau P Meacuteleacuteard F Faucon and P Bothorel Prog Colloid 754

Polym Sci 1992 89 127ndash131 755

6 L Mathivet S Cribier and P F Devaux Biophys J 1996 70 1112ndash1121 756

7 V Kralj-Iglič G Gomišček J Majhenc V Arrigler and S Svetina Colloids Surfaces 757

A Physicochem Eng Asp 2001 181 315ndash318 758

30

8 V Kralj-Iglič A Iglič G Gomišček F Sevšek V Arrigler and H Haumlgerstrand J 759

Phys A Math Gen 2002 35 1533ndash1549 760

9 M Drab D Stopar V Kralj-Iglič and A Iglič Cells 2019 8 626 761

10 A Zidovska K K Ewert J Quispe B Carragher C S Potter and C R Safinya 762

Langmuir 2009 25 2979ndash2985 763

11 A Zidovska K K Ewert J Quispe B Carragher C S Potter and C R Safinya Soft 764

Matter 2011 7 8363ndash8369 765

12 A Zidovska K K Ewert J Quispe B Carragher C S Potter and C R Safinya 766

Biochim Biophys Acta - Biomembr 2009 1788 1869ndash1876 767

13 M Karlsson K Sott M Davidson A S Cans P Linderholm D Chiu and O Orwar 768

Proc Natl Acad Sci U S A 2002 99 11573ndash11578 769

14 A Karlsson R Karlsson M Karlsson A-S Cans A Stromberg F Ryttsen and O 770

Orwar Nature 2001 409 150ndash152 771

15 M Karlsson K Sott A S Cans A Karlsson R Karlsson and O Orwar Langmuir 772

2001 17 6754ndash6758 773

16 E Evans H Bowman A Leung D Needham and D Tirrell Science (80- ) 1996 774

273 933ndash935 775

17 U Seifert Configurations of fluid membranes and vesicles 1997 vol 46 776

18 J Kaumls and E Sackmann Biophys J 1991 60 825ndash844 777

19 N Bobrovska W Goacuteźdź V Kralj-Iglič and A Iglič PLoS One 2013 8 e73941 778

20 A Iglič B Babnik K Bohinc M Fošnarič H Haumlgerstrand and V Kralj-Iglič J 779

Biomech 2007 40 579ndash585 780

21 K Bohinc T Slivnik A Iglič M Brumen and V Kralj-Iglič J Phys Chem C 2007 781

111 9709ndash9718 782

22 A Iglič H Haumlgerstrand P Veranič A Plemenitaš and V Kralj-Iglič J Theor Biol 783

2006 240 368ndash373 784

23 V Kralj-Iglič and P Veranič Adv Planar Lipid Bilayers Liposomes 2006 5 129ndash785

149 786

24 A Iglic and V Kralj-Iglic Adv Planar Lipid Bilayers Liposomes 2006 4 253ndash279 787

25 S Penič L Mesarec M Fošnarič L Mroacutewczyńska H Haumlgerstrand V Kralj-Iglič and 788

A Iglič Front Phys 2020 8 342 789

26 P Yager and P E Schoen Mol Cryst Liq Cryst 1984 106 371ndash381 790

27 T G Barclay K Constantopoulos and J Matisons Chem Rev 2014 114 10217ndash791

10291 792

31

28 B Božič G Gomišček V Kralj-Iglič S Svetina and B Žekš Eur Biophys J 2002 793

31 487ndash496 794

29 A Iglič H Haumlgerstrand M Bobrowska-Haumlgerstrand V Arrigler and V Kralj-Iglič 795

Phys Lett A 2003 310 493ndash497 796

30 R Bar-Ziv and E Moses Phys Rev Lett 1994 73 1392ndash1395 797

31 S Chiruvolu H E Warriner E Naranjo S H Idziak J O Raumldler R J Plano J A 798

Zasadzinski and C R Safinya Science (80- ) 1994 266 1222ndash1225 799

32 V Kralj-Iglič A Iglič H Haumlgerstrand and P Peterlin Phys Rev E 2000 61 4230ndash800

4234 801

33 L Mesarec M Drab S Penič V Kralj-Iglič and A Iglič Int J Mol Sci 2021 22 802

2348 803

34 L Van Renterghem S L K W Roelants N Baccile K Uyttersprot M C Taelman 804

B Everaert S Mincke S Ledegen S Debrouwer K Scholtens C V Stevens W 805

Soetaert L Van Renterghem S L K W Roelants N Baccile K Uytersprot M C 806

Taelman B Everaert S Mincke S Ledegen S Debrouwer K Scholtens C V 807

Stevens and W Soetaert Biotechnol Bioeng 2018 115 1195ndash1206 808

35 J Clark and F Deswarte Introduction to Chemicals from Biomass Second Edition 809

John Wiley amp Sons 2015 810

36 K V Sajna R Houmlfer R K Sukumaran L D Gottumukkala and A Pandey in 811

Industrial Biorefineries and White Biotechnology eds A Pandey R Houmlfer M 812

Taherzadeh K M Nampoothiri and C Larroche Amsterdam Oxford Waltham 813

Elsevier 2015 pp 499ndash521 814

37 R Marchant and I M Banat Trends Biotechnol 2012 30 558ndash565 815

38 K Hill and C LeHen-Ferrenbach in Sugar-Based Surfactants Fundamentals and 816

Applications ed C C Ruiz Boca Raton CRC Press 2009 pp 1ndash20 817

39 L Rodrigues I M Banat J Teixeira and R Oliveira J Antimicrob Chemother 818

2006 57 609ndash618 819

40 J D Desai and I M Banat Microbiol Mol Biol Rev 1997 61 47ndash64 820

41 S L K W Roelants K Ciesielska S L De Maeseneire H Moens B Everaert S 821

Verweire Q Denon B Vanlerberghe I N A Van Bogaert P Van der Meeren B 822

Devreese and W Soetaert Biotechnol Bioeng 2016 113 550ndash559 823

42 N Baccile M Selmane P Le Griel S Preacutevost J Perez C V Stevens E Delbeke S 824

Zibek M Guenther W Soetaert I N A Van Bogaert and S Roelants Langmuir 825

2016 32 6343ndash6359 826

32

43 N Baccile A-S Cuvier S Preacutevost C V Stevens E Delbeke J Berton W Soetaert 827

I N A Van Bogaert and S Roelants Langmuir 2016 32 10881ndash10894 828

44 N Baccile C Seyrig A Poirier S A Castro S L K W Roelants and S Abel Green 829

Chem 2021 23 3842ndash3944 830

45 A Salis L Cappai C Carucci D F Parsons and M Monduzzi J Phys Chem Lett 831

2020 11 6805ndash6811 832

46 J Schindelin I Arganda-Carreras E Frise V Kaynig M Longair T Pietzsch S 833

Preibisch C Rueden S Saalfeld B Schmid J-Y Tinevez D J White V 834

Hartenstein K Eliceiri P Tomancak and A Cardona Nat Methods 2012 9 676ndash835

682 836

47 T Parasassi and E Gratton J Fluoresc 1995 5 59ndash69 837

48 M M Dodes Traian F L Gonzaacutelez Flecha and V Levi J Lipid Res 2012 53 609ndash838

616 839

49 D M Owen C Rentero A Magenau A Abu-Siniyeh and K Gaus Nat Protoc 840

2012 7 24ndash35 841

50 W Stillwell in An Introduction to Biological Membranes Academic Press 2016 pp 842

135ndash180 843

51 W Stillwell in An Introduction to Biological Membranes Academic Press 2016 pp 844

181ndash220 845

52 B Tenchov Chem Phys Lipids 1991 57 165ndash177 846

53 R Koynova and M Caffrey Biochim Biophys Acta 1998 1376 91ndash145 847

54 F Schmid S Dolezel O Lenz and S Meinhardt J Phys Conf Ser 2014 487 848

012004 849

55 S Meinhardt R L C Vink and F Schmid Proc Natl Acad Sci U S A 2013 110 850

4476ndash4481 851

56 O Glatter and O Kratky Small Angle X-ray Scattering Academic Press London 852

1982 853

57 B A Cunningham A D Brown D H Wolfe W P Williams and A Brain Phys 854

Rev E 1998 58 3662ndash3672 855

58 M Nadler A Steiner T Dvir O Szekely P Szekely A Ginsburg R Asor R Resh 856

C Tamburu M Peres and U Raviv Soft Matter 2011 7 1512ndash1523 857

59 C Tanford The hydrophobic effect Formation of micelles and biological membranes 858

John Wiley amp Sons Inc New York 1973 859

60 M Masuda and T Shimizu Langmuir 2004 20 5969ndash5977 860

33

61 A Zidovska K K Ewert J Quispe B Carragher C S Potter and C R Safinya 861

Block Liposomes Vesicles of Charged Lipids with Distinctly Shaped Nanoscale 862

Sphere- Pear- Tube- or Rod-Segments Elsevier Inc 1st edn 2009 vol 465 863

62 A Iglic V Kralj-Iglic B Božič M Bobrowska-Haumlgerstrand B Isomaa and H 864

Haumlgerstrand Bioelectrochemistry 2000 52 203ndash211 865

63 T Shimizu M Masuda and H Minamikawa Chem Rev 2005 105 1401ndash1444 866

64 J V Selinger and J M Schnur Phys Rev Lett 1993 71 4091ndash4094 867

65 J M Schnur Science (80- ) 1993 262 1669ndash1676 868

66 B Tenchov R Koynova and G Rapp Biophys J 2001 80 1873ndash1890 869

67 R Koynova and B Tenchov Ser Biomech 2017 31 3ndash19 870

68 S J Angyal and R Le Fur Carbohydr Res 1984 126 15ndash26 871

69 J K Whitesell and M A Minton J Am Chem Soc 1987 109 225ndash228 872

70 T Takayama S Ando and I Ando J Mol Struct 1990 220 245ndash250 873

71 K Gawrisch N V Eldho and I V Polozov Chem Phys Lipids 2002 116 135ndash151 874

72 F Tielens C Gervais J F Lambert F Mauri and D Costa Chem Mater 2008 20 875

3336ndash3344 876

73 G A Jeffrey and Y Yeon Acta Crystallogr Sect B 1986 42 410ndash413 877

74 B Berglund and R W Vaughan J Chem Phys 1980 73 2037ndash2043 878

75 C Gervais C Coelho T Azaiumls J Maquet G Laurent F Pourpoint C Bonhomme 879

P Florian B Alonso G Guerrero P H Mutin and F Mauri J Magn Reson 2007 880

187 131ndash140 881

76 C Akita T Kawaguchi F Kaneko H Yamamoto and M Suzuki J Phys Chem B 882

2004 108 4862ndash4868 883

77 R N Purusottam L Seacutenicourt J J Lacapegravere and P Tekely Biochim Biophys Acta - 884

Biomembr 2015 1848 3134ndash3139 885

78 X Biarneacutes A Ardegravevol A Planas C Rovira A Laio and M Parrinello J Am Chem 886

Soc 2007 129 10686ndash10693 887

79 L M J Kroon‐Batenburg and J Kroon Biopolymers 1990 29 1243ndash1248 888

80 K N Kirschner and R J Woods Proc Natl Acad Sci U S A 2001 98 10541ndash889

10545 890

81 S Svenson B Kirste and J H Fuhrhop J Am Chem Soc 1994 116 11969ndash11975 891

82 I Sack S Macholl J H Fuhrhop and G Buntkowsky Phys Chem Chem Phys 892

2000 2 1781ndash1788 893

83 S Svenson J Koening and J H Fuhrhop J Phys Chem 1994 98 1022ndash1028 894

34

84 S Svenson A Schaefer and J H Fuhrhop J Chem Soc Perkin Trans 2 1994 2 895

1023ndash1028 896

85 K Bock and C Pedersen Adv Carbohydr Chem Biochem 1983 41 27ndash66 897

86 T B Grindley Structure and Conformation of Carbohydrates Springer Berlin 898

Heidelberg 2008 899

87 F Tantakitti J Boekhoven X Wang R V Kazantsev T Yu J Li E Zhuang R 900

Zandi J H Ortony C J Newcomb L C Palmer G S Shekhawat M O De La 901

Cruz G C Schatz and S I Stupp Nat Mater 2016 15 469ndash476 902

903

1

Supporting Information 904

905

Topological connection between vesicles and nanotubes in single-906

component lipid membranes driven by head-tail interactions 907

908

Niki Baccilea Ceacutedric Lorthioira Abdoul Aziz Baa Patrick Le Griela Cristina Coelhob Javier 909

Perezc Wim Soetaertd Sophie L K W Roelantsd 910

911

a Sorbonne Universiteacute Centre National de la Recherche Scientifique Laboratoire de Chimie de 912

la Matiegravere Condenseacutee de Paris LCMCP F-75005 Paris France 913

b Institut des Mateacuteriaux de Paris Centre Sorbonne Universiteacute Paris France 914

c Synchrotron Soleil LrsquoOrme des Merisiers Saint-Aubin BP48 91192 Gif-sur-Yvette Cedex 915

France 916

d InBio Department of Biotechnology Ghent University Ghent Belgium 917

918

919

Corresponding author 920

Dr Niki Baccile 921

E-mail address nikibaccilesorbonne-universitefr 922

Phone +33 1 44 27 56 77 923

924

925

926

2

a) 927

Parameter Method

Dry Matter (DM ) 996 Infrared balance 105degC

Glucose () 012 HPLC-Metacarb

Glycerol () nd HPLC-Metacarb

Free Fatty Acid content () lt001 Internal method BBEPP

GC with FID Detector

Oil () lt001 Internal method BBEPP

GC with FID Detector

Moulds (CFUg) lt10 3M-Nordval ndeg16 B

Yeasts (CFUg) lt10 3M-Nordval ndeg16 B

Anaerobic count (CFUg) lt10 SP-VG M005 B

Protein () tbd BCA protein assay

928

Figure S 1 - a) HPLC-ELSD chromatogram and composition table 929

930

931

932

933

3

b) 934

935

Group Ndeg δ ppm

CH2 1 354

CH2 2 153

CH2 3-712-14 132

CH2 811 203

CH 910 534

CH2 15 162

CH2 16 388

CH 1 424

CH 2 316

CH 3 334

CH 4 329

CH 5 354

CH2 6 366

Figure S 1 (next) ndash b) 1H NMR spectrum (MeOD-d4) recorded on the purified G-C181-OH compounds 936

and related assignment 937

938

2

811910 1rsquo

5rsquo1

3rsquo4rsquo

1rsquo1910

asymp asymp asympHDO MeOD Acetone

6rsquo2rsquo

1615

3-712-14

4

939

940

Figure S 2 ndash Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 130degC 941

5

942

943

Figure S 3 - Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 90degC 944

945

90degC

1

22

6

946

947

948

949

Figure S 4 ndash Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 70degC 950

951

21

11

1

2

5

3

4

45

5

6

7

7

7

952

953

954

Figure S 5 ndash Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 25degC 955

956

957

1

1 1

8

958

Figure S 6 ndash a) SANS experiment recorded at 90degC for a 5 mgmL G-C181-OH aqueous solution b-c) 959

WAXS experiments recorded on aqueous solutions of G-C181-OH at b) 5 mgmL and c) 25 mgmL d) 960

Structural interpretation of the repeating distances pointed out in c) 961

962

963

964

001 01 1

q Aring-1

25degC

68degC

88degC

111degC

01 1q Aring

-1

02 04 06 08 1q Aring

-1

001 01

10-1

100

101

Inte

nsity

cm

-1

q Aring-1

-2

T= 90degC

SANS WAXS WAXS

a) b) c)

dv

dL d= 144 Aring-1

di

ddH2O

dt

dvdL

dtdi d

d)

9

965

966

Figure S 7 ndash ssNMR MAS (10 kHz) experiments recorded on a G-C181-OH dry powder a) 1H spectra 967

recorded from 110degC to 70degC b) 13C single pulse (SP) spectra recorded with high power 1H decoupling from 968

110degC to 70degC c) 13C cross polarization (CP) spectra recorded at 110degC and contact time tc of 50 micros and 969

3000 micros d) 13C CP spectra recorded at 70degC and for 20 lt tc micros lt 3000 970

971

1

811

910

CH2 CH2

1rsquo

6rsquo

8 6 4 2 0 ppm

110

90

80

70

910

1rsquo

6rsquo16

1

2rsquo3rsquo4rsquo

811

OH

45 40 35 30 25 20 15

ppm

200 180 160 140 120 100 80 60 40 20 0

ppm140 120 100 80

ppm

CH2-gauche

3004

110

90

80

70

65

60

50

25

C91013003

C9 - 12869

CH2-all-trans

3198

C1rsquo10335

a) b)

1H

T degCT degC

13C Single Pulse MAS

A

B

A

B

70degC70degC

5rsquo

215

200 180 160 140 120 100 80 60 40 20 0

ppm

13C CPMAST= 110degC

200 180 160 140 120 100 80 60 40 20 0

ppm

132 131 130 129 128 127 126

ppm

106 105 104 103 102

ppm

13C CPMAST= 70degC

50

tc micros

3000

20

tc micros

50 500 1000 3000

tc micros

C102

13024 C910113003

C92

12869

C1rsquo213082

C1rsquo110443

A B

106 105 104 103 102

ppm

3000

1000

500

50

20

AB

c)

d)

10

972

Figure S 8 ndash a) Short and d) long T2 components displayed by the 1H transverse relaxation signal c) 1H 973

transverse relaxation signals recorded on a G-C181-OH dry powder from 110degC to 70degC The area of each 974

given resonance is plotted (log-lin) as a function of the echo time (CPMG pulse sequence) for each 975

temperature and fitted using the equation given above T2s and T2l indicate the short and long T2 relaxation 976

times respectively obtained by describing the 1H transverse relaxation signal using a biexponential decay 977

For H1rsquo H5rsquo and H6rsquo the 1H relaxation function could be described by a monoexponential decay 978

- Protons from the sugar group (H1rsquo H5rsquo H6rsquo) show a single T2 relaxation component (Figure 979

6c and Figure S 8) while protons of the acyl chain (CH2 H910 H811) show both short (Figure 980

6c) and long (Figure 6d) T2 relaxation component 981

- The short T2 relaxation time systematically decreases from 110degC to 70degC for all species by a 982

factor ranging from two to seven depending on the specific chemical group Very long T2 are 983

observed at 110degC for all acyl species while nearly unchanged values are measured below 984

90degC without any clear trend 985

- The sugar protons (H1rsquo H5rsquo H6rsquo) have the shortest set of T2 values (lt 1 ms) below 100degC 986

The long component of T2 generally ranges between 4 ms and 12 ms 987

0 1 2 3 4 5 6 7 8

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1 2 3 4 5 6 7 8

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

H1

H8H11

H9H10

CH2

H1rsquo

H6rsquo

CH2

H910H811 H1CH2

H1rsquo H5rsquo6rsquo

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

OH

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45 T= 110degC

T= 90degC

T= 80degC

T= 70degC

Log

In

ten

sity

Time ms

Fit

)

CH2 H910 H811 H1 H56 H1 OH

00

05

10

15

20

25

30

35

40

T2 m

s

T= 110degC

T= 90degC

T= 80degC

T= 70degC

CH2 H910 H811 H1

0

2

4

6

8

10

12

T2 m

s

T= 110degC

T= 90degC

T= 80degC

T= 70degC

Short T2(1H) Long T2(1H)

a) b)

c)

11

Table S 1 ndash 1H chemical shifts of G-C181-OH in solution and in the solid-state Solid-state spectra are 988

recorded at a MAS frequency of 10 kHz in DMSO-d6 broad peak 989

990

δH ppm

Group Ndeg Liquid

(MeOD-d4) Solid

(T= 110degC) Solid

(T= 70degC)

CH2 H1 354 357

CH2 H2 153 156

CH2 H3-712-14 132 135

CH2 H811 203 203

CH H910 534 534

CH2 H15 162 164

CH2 H16 388 383

CH H1 424 432

CH H2 316 335

CH H3 334 335

CH H4 329 335

CH H5 354 351 357

CH2 H6 366 383

OH OH1 432

45 463

OH OH2 492

OH OH3 490

OH OH4 487

OH OH6 445

991

992

H1H9H10H1rsquo

H6rsquo

OH2rsquo

OH1

12

Detailed discussion about the NOESY and DQ-SQ 2D 1H-1H ssNMR experiments presented in 993

Figure 9 994

995

2D 1H-1H ssNMR experiments are a unique tool to explore through-space proximities 996

between 1H nuclei ssNMR offers a large number of possibilities each adapted to a given 997

system The advantage of 2D maps with respect to a dynamics approach is their direct 998

readability in clear two nuclei which exchange their magnetization through-space (whichever 999

mechanism may be responsible for this exchange) will display a specific cross-peak However 1000

the way cross-peaks are presented and interpreted strictly depends on the nature of the pulse 1001

program which in turn is chosen according to the physical nature of the sample 1002

1D temperature-dependent 1H NMR experiments performed on G-C181-OH (Figure 1003

4a in the main text) indicate sharp peaks at 90degC and broad peaks at 70degC Narrow peaks 1004

indicate fast andor large-amplitude molecular tumbling averaging the strong 1H-1H dipolar 1005

interactions This excludes the use of pulse sequences like DQ-SQ which is not efficient under 1006

these conditions For this reason we employ spin-diffusion-based pulse sequences like 1007

NOESY which is more adapted for samples with high molecular mobility When mixing times 1008

are chosen to be short (lt 1 s) one generally explores through-space proximities within about 5 1009

10 Aring On the contrary when molecular motions are restricted dipolar coupling becomes 1010

significant 1H NMR peaks are broader and pulse sequences like DQ-SQ can now be used 1011

with a spatial resolution of less than 5 Aring 1012

Whenever possible DQ-SQ experiments are preferred over spin diffusion-based pulse 1013

sequences because they are more selective in particular for likewise interactions which are 1014

not discriminated in NOESY-type experiments This is illustrated by the typical interaction 1015

schemes for NOESY (top Figure 9a) and DQ-SQ (top Figure 9b) For two given A and B 1016

protons 2D NOESY maps will display two on-diagonal and two square-shaped symmetrical 1017

off-diagonal cross-peaks Off-diagonal cross-peaks illustrate the mutual A-B and B-A 1018

interaction but the on-diagonal peaks are systematic and do not prove any A-A or B-B 1019

interaction Similarly to NOESY 2D DQ-SQ maps display off-diagonal peaks (symmetric 1020

across the diagonal top Figure 9b) upon A-B and B-A interaction but in contrast they will 1021

display on-diagonal peaks only in the presence of self-interaction In the tutorial example of 1022

Figure 9b top the on-diagonal A-A peak indicates mutual A-A interaction while lack of on-1023

diagonal B-B peak evidences the absence of B-B proximities Combining NOESY and DQ-SQ 1024

experiments is capital to study the local arrangement of G-C181-OH and its evolution with 1025

temperature 1026

13

At T= 90degC the H1 and H5rsquo contributions are very close but possible to discriminate 1027

them (δ(H1)= 354 ppm δ(H5rsquo)= 351 ppm) according to 1H NMR chemical shifts in Table S 1028

1 2D 1H-1H NOESY in Figure 9a also discriminates between the two components attributed 1029

on the basis of the interactions with their mutual neighbors (green H1-H2 and H5rsquo-H1rsquo squares 1030

in Figure 9a) On this basis 2D NOESY does not show a direct H1rsquo-H1 correlation (blue red-1031

crossed) but rather an extensive network of intra-sugar interactions These features exclude 1032

suggest intra- and inter-sugar interactions and they exclude head-tail arrangements as a major 1033

molecular configuration at 90degC 1034

At 70degC the 1H NMR peaks are broader indicating a reduction of the T2 (1H) relaxation 1035

times andor residual homonuclear dipolar interactions excluding the use of NOESY In this 1036

case 1H-1H DQ-SQ correlation experiments are best suited Figure 9b presents the 1H-1H DQ-1037

SQ experiment on G-C181-OH at 70degC Peaks are broader than in NOESY and attribution may 1038

be cumbersome However the experiment is performed at 700 MHz and MAS= 22 kHz which 1039

improves spectral resolution with respect to the same experiment performed at 300 MHz and 1040

MAS= 10 kHz The auto-peak at δH= 356 ppm in the SQ dimension is attributed to H1-H1 spin 1041

pairs mainly resulting from intra CH2 interaction while the DQ cross-peaks at δH= 156-356 1042

ppm and δH= 432-356 ppm respectively origin from H2-H1 and H1rsquo-H1 interactions (Figure 1043

9b) DQ-SQ map does not show any auto-peak corresponding to H1rsquo-H1rsquo or Hxrsquo-Hxrsquo (x= 2rsquo 1044

3rsquo 4rsquo) interactions (red-crossed blue Figure 9b) thus excluding any proximity (lt 5 Aring) between 1045

close glucose neighbors On the contrary DQ-SQ maps (recorded both at 700 MHz Figure 9b 1046

or at 300 MHz not shown) systematically show DQ cross-peaks which can only be attributed 1047

to H1rsquo-H1 interactions (δH= 432-356) The presence of these H1rsquo-H1 interactions can only be 1048

explained by a head-tail arrangement of the G-C181-OH molecules as illustrated below the 1049

DQ-SQ map in Figure 9b 1050

30

8 V Kralj-Iglič A Iglič G Gomišček F Sevšek V Arrigler and H Haumlgerstrand J 759

Phys A Math Gen 2002 35 1533ndash1549 760

9 M Drab D Stopar V Kralj-Iglič and A Iglič Cells 2019 8 626 761

10 A Zidovska K K Ewert J Quispe B Carragher C S Potter and C R Safinya 762

Langmuir 2009 25 2979ndash2985 763

11 A Zidovska K K Ewert J Quispe B Carragher C S Potter and C R Safinya Soft 764

Matter 2011 7 8363ndash8369 765

12 A Zidovska K K Ewert J Quispe B Carragher C S Potter and C R Safinya 766

Biochim Biophys Acta - Biomembr 2009 1788 1869ndash1876 767

13 M Karlsson K Sott M Davidson A S Cans P Linderholm D Chiu and O Orwar 768

Proc Natl Acad Sci U S A 2002 99 11573ndash11578 769

14 A Karlsson R Karlsson M Karlsson A-S Cans A Stromberg F Ryttsen and O 770

Orwar Nature 2001 409 150ndash152 771

15 M Karlsson K Sott A S Cans A Karlsson R Karlsson and O Orwar Langmuir 772

2001 17 6754ndash6758 773

16 E Evans H Bowman A Leung D Needham and D Tirrell Science (80- ) 1996 774

273 933ndash935 775

17 U Seifert Configurations of fluid membranes and vesicles 1997 vol 46 776

18 J Kaumls and E Sackmann Biophys J 1991 60 825ndash844 777

19 N Bobrovska W Goacuteźdź V Kralj-Iglič and A Iglič PLoS One 2013 8 e73941 778

20 A Iglič B Babnik K Bohinc M Fošnarič H Haumlgerstrand and V Kralj-Iglič J 779

Biomech 2007 40 579ndash585 780

21 K Bohinc T Slivnik A Iglič M Brumen and V Kralj-Iglič J Phys Chem C 2007 781

111 9709ndash9718 782

22 A Iglič H Haumlgerstrand P Veranič A Plemenitaš and V Kralj-Iglič J Theor Biol 783

2006 240 368ndash373 784

23 V Kralj-Iglič and P Veranič Adv Planar Lipid Bilayers Liposomes 2006 5 129ndash785

149 786

24 A Iglic and V Kralj-Iglic Adv Planar Lipid Bilayers Liposomes 2006 4 253ndash279 787

25 S Penič L Mesarec M Fošnarič L Mroacutewczyńska H Haumlgerstrand V Kralj-Iglič and 788

A Iglič Front Phys 2020 8 342 789

26 P Yager and P E Schoen Mol Cryst Liq Cryst 1984 106 371ndash381 790

27 T G Barclay K Constantopoulos and J Matisons Chem Rev 2014 114 10217ndash791

10291 792

31

28 B Božič G Gomišček V Kralj-Iglič S Svetina and B Žekš Eur Biophys J 2002 793

31 487ndash496 794

29 A Iglič H Haumlgerstrand M Bobrowska-Haumlgerstrand V Arrigler and V Kralj-Iglič 795

Phys Lett A 2003 310 493ndash497 796

30 R Bar-Ziv and E Moses Phys Rev Lett 1994 73 1392ndash1395 797

31 S Chiruvolu H E Warriner E Naranjo S H Idziak J O Raumldler R J Plano J A 798

Zasadzinski and C R Safinya Science (80- ) 1994 266 1222ndash1225 799

32 V Kralj-Iglič A Iglič H Haumlgerstrand and P Peterlin Phys Rev E 2000 61 4230ndash800

4234 801

33 L Mesarec M Drab S Penič V Kralj-Iglič and A Iglič Int J Mol Sci 2021 22 802

2348 803

34 L Van Renterghem S L K W Roelants N Baccile K Uyttersprot M C Taelman 804

B Everaert S Mincke S Ledegen S Debrouwer K Scholtens C V Stevens W 805

Soetaert L Van Renterghem S L K W Roelants N Baccile K Uytersprot M C 806

Taelman B Everaert S Mincke S Ledegen S Debrouwer K Scholtens C V 807

Stevens and W Soetaert Biotechnol Bioeng 2018 115 1195ndash1206 808

35 J Clark and F Deswarte Introduction to Chemicals from Biomass Second Edition 809

John Wiley amp Sons 2015 810

36 K V Sajna R Houmlfer R K Sukumaran L D Gottumukkala and A Pandey in 811

Industrial Biorefineries and White Biotechnology eds A Pandey R Houmlfer M 812

Taherzadeh K M Nampoothiri and C Larroche Amsterdam Oxford Waltham 813

Elsevier 2015 pp 499ndash521 814

37 R Marchant and I M Banat Trends Biotechnol 2012 30 558ndash565 815

38 K Hill and C LeHen-Ferrenbach in Sugar-Based Surfactants Fundamentals and 816

Applications ed C C Ruiz Boca Raton CRC Press 2009 pp 1ndash20 817

39 L Rodrigues I M Banat J Teixeira and R Oliveira J Antimicrob Chemother 818

2006 57 609ndash618 819

40 J D Desai and I M Banat Microbiol Mol Biol Rev 1997 61 47ndash64 820

41 S L K W Roelants K Ciesielska S L De Maeseneire H Moens B Everaert S 821

Verweire Q Denon B Vanlerberghe I N A Van Bogaert P Van der Meeren B 822

Devreese and W Soetaert Biotechnol Bioeng 2016 113 550ndash559 823

42 N Baccile M Selmane P Le Griel S Preacutevost J Perez C V Stevens E Delbeke S 824

Zibek M Guenther W Soetaert I N A Van Bogaert and S Roelants Langmuir 825

2016 32 6343ndash6359 826

32

43 N Baccile A-S Cuvier S Preacutevost C V Stevens E Delbeke J Berton W Soetaert 827

I N A Van Bogaert and S Roelants Langmuir 2016 32 10881ndash10894 828

44 N Baccile C Seyrig A Poirier S A Castro S L K W Roelants and S Abel Green 829

Chem 2021 23 3842ndash3944 830

45 A Salis L Cappai C Carucci D F Parsons and M Monduzzi J Phys Chem Lett 831

2020 11 6805ndash6811 832

46 J Schindelin I Arganda-Carreras E Frise V Kaynig M Longair T Pietzsch S 833

Preibisch C Rueden S Saalfeld B Schmid J-Y Tinevez D J White V 834

Hartenstein K Eliceiri P Tomancak and A Cardona Nat Methods 2012 9 676ndash835

682 836

47 T Parasassi and E Gratton J Fluoresc 1995 5 59ndash69 837

48 M M Dodes Traian F L Gonzaacutelez Flecha and V Levi J Lipid Res 2012 53 609ndash838

616 839

49 D M Owen C Rentero A Magenau A Abu-Siniyeh and K Gaus Nat Protoc 840

2012 7 24ndash35 841

50 W Stillwell in An Introduction to Biological Membranes Academic Press 2016 pp 842

135ndash180 843

51 W Stillwell in An Introduction to Biological Membranes Academic Press 2016 pp 844

181ndash220 845

52 B Tenchov Chem Phys Lipids 1991 57 165ndash177 846

53 R Koynova and M Caffrey Biochim Biophys Acta 1998 1376 91ndash145 847

54 F Schmid S Dolezel O Lenz and S Meinhardt J Phys Conf Ser 2014 487 848

012004 849

55 S Meinhardt R L C Vink and F Schmid Proc Natl Acad Sci U S A 2013 110 850

4476ndash4481 851

56 O Glatter and O Kratky Small Angle X-ray Scattering Academic Press London 852

1982 853

57 B A Cunningham A D Brown D H Wolfe W P Williams and A Brain Phys 854

Rev E 1998 58 3662ndash3672 855

58 M Nadler A Steiner T Dvir O Szekely P Szekely A Ginsburg R Asor R Resh 856

C Tamburu M Peres and U Raviv Soft Matter 2011 7 1512ndash1523 857

59 C Tanford The hydrophobic effect Formation of micelles and biological membranes 858

John Wiley amp Sons Inc New York 1973 859

60 M Masuda and T Shimizu Langmuir 2004 20 5969ndash5977 860

33

61 A Zidovska K K Ewert J Quispe B Carragher C S Potter and C R Safinya 861

Block Liposomes Vesicles of Charged Lipids with Distinctly Shaped Nanoscale 862

Sphere- Pear- Tube- or Rod-Segments Elsevier Inc 1st edn 2009 vol 465 863

62 A Iglic V Kralj-Iglic B Božič M Bobrowska-Haumlgerstrand B Isomaa and H 864

Haumlgerstrand Bioelectrochemistry 2000 52 203ndash211 865

63 T Shimizu M Masuda and H Minamikawa Chem Rev 2005 105 1401ndash1444 866

64 J V Selinger and J M Schnur Phys Rev Lett 1993 71 4091ndash4094 867

65 J M Schnur Science (80- ) 1993 262 1669ndash1676 868

66 B Tenchov R Koynova and G Rapp Biophys J 2001 80 1873ndash1890 869

67 R Koynova and B Tenchov Ser Biomech 2017 31 3ndash19 870

68 S J Angyal and R Le Fur Carbohydr Res 1984 126 15ndash26 871

69 J K Whitesell and M A Minton J Am Chem Soc 1987 109 225ndash228 872

70 T Takayama S Ando and I Ando J Mol Struct 1990 220 245ndash250 873

71 K Gawrisch N V Eldho and I V Polozov Chem Phys Lipids 2002 116 135ndash151 874

72 F Tielens C Gervais J F Lambert F Mauri and D Costa Chem Mater 2008 20 875

3336ndash3344 876

73 G A Jeffrey and Y Yeon Acta Crystallogr Sect B 1986 42 410ndash413 877

74 B Berglund and R W Vaughan J Chem Phys 1980 73 2037ndash2043 878

75 C Gervais C Coelho T Azaiumls J Maquet G Laurent F Pourpoint C Bonhomme 879

P Florian B Alonso G Guerrero P H Mutin and F Mauri J Magn Reson 2007 880

187 131ndash140 881

76 C Akita T Kawaguchi F Kaneko H Yamamoto and M Suzuki J Phys Chem B 882

2004 108 4862ndash4868 883

77 R N Purusottam L Seacutenicourt J J Lacapegravere and P Tekely Biochim Biophys Acta - 884

Biomembr 2015 1848 3134ndash3139 885

78 X Biarneacutes A Ardegravevol A Planas C Rovira A Laio and M Parrinello J Am Chem 886

Soc 2007 129 10686ndash10693 887

79 L M J Kroon‐Batenburg and J Kroon Biopolymers 1990 29 1243ndash1248 888

80 K N Kirschner and R J Woods Proc Natl Acad Sci U S A 2001 98 10541ndash889

10545 890

81 S Svenson B Kirste and J H Fuhrhop J Am Chem Soc 1994 116 11969ndash11975 891

82 I Sack S Macholl J H Fuhrhop and G Buntkowsky Phys Chem Chem Phys 892

2000 2 1781ndash1788 893

83 S Svenson J Koening and J H Fuhrhop J Phys Chem 1994 98 1022ndash1028 894

34

84 S Svenson A Schaefer and J H Fuhrhop J Chem Soc Perkin Trans 2 1994 2 895

1023ndash1028 896

85 K Bock and C Pedersen Adv Carbohydr Chem Biochem 1983 41 27ndash66 897

86 T B Grindley Structure and Conformation of Carbohydrates Springer Berlin 898

Heidelberg 2008 899

87 F Tantakitti J Boekhoven X Wang R V Kazantsev T Yu J Li E Zhuang R 900

Zandi J H Ortony C J Newcomb L C Palmer G S Shekhawat M O De La 901

Cruz G C Schatz and S I Stupp Nat Mater 2016 15 469ndash476 902

903

1

Supporting Information 904

905

Topological connection between vesicles and nanotubes in single-906

component lipid membranes driven by head-tail interactions 907

908

Niki Baccilea Ceacutedric Lorthioira Abdoul Aziz Baa Patrick Le Griela Cristina Coelhob Javier 909

Perezc Wim Soetaertd Sophie L K W Roelantsd 910

911

a Sorbonne Universiteacute Centre National de la Recherche Scientifique Laboratoire de Chimie de 912

la Matiegravere Condenseacutee de Paris LCMCP F-75005 Paris France 913

b Institut des Mateacuteriaux de Paris Centre Sorbonne Universiteacute Paris France 914

c Synchrotron Soleil LrsquoOrme des Merisiers Saint-Aubin BP48 91192 Gif-sur-Yvette Cedex 915

France 916

d InBio Department of Biotechnology Ghent University Ghent Belgium 917

918

919

Corresponding author 920

Dr Niki Baccile 921

E-mail address nikibaccilesorbonne-universitefr 922

Phone +33 1 44 27 56 77 923

924

925

926

2

a) 927

Parameter Method

Dry Matter (DM ) 996 Infrared balance 105degC

Glucose () 012 HPLC-Metacarb

Glycerol () nd HPLC-Metacarb

Free Fatty Acid content () lt001 Internal method BBEPP

GC with FID Detector

Oil () lt001 Internal method BBEPP

GC with FID Detector

Moulds (CFUg) lt10 3M-Nordval ndeg16 B

Yeasts (CFUg) lt10 3M-Nordval ndeg16 B

Anaerobic count (CFUg) lt10 SP-VG M005 B

Protein () tbd BCA protein assay

928

Figure S 1 - a) HPLC-ELSD chromatogram and composition table 929

930

931

932

933

3

b) 934

935

Group Ndeg δ ppm

CH2 1 354

CH2 2 153

CH2 3-712-14 132

CH2 811 203

CH 910 534

CH2 15 162

CH2 16 388

CH 1 424

CH 2 316

CH 3 334

CH 4 329

CH 5 354

CH2 6 366

Figure S 1 (next) ndash b) 1H NMR spectrum (MeOD-d4) recorded on the purified G-C181-OH compounds 936

and related assignment 937

938

2

811910 1rsquo

5rsquo1

3rsquo4rsquo

1rsquo1910

asymp asymp asympHDO MeOD Acetone

6rsquo2rsquo

1615

3-712-14

4

939

940

Figure S 2 ndash Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 130degC 941

5

942

943

Figure S 3 - Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 90degC 944

945

90degC

1

22

6

946

947

948

949

Figure S 4 ndash Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 70degC 950

951

21

11

1

2

5

3

4

45

5

6

7

7

7

952

953

954

Figure S 5 ndash Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 25degC 955

956

957

1

1 1

8

958

Figure S 6 ndash a) SANS experiment recorded at 90degC for a 5 mgmL G-C181-OH aqueous solution b-c) 959

WAXS experiments recorded on aqueous solutions of G-C181-OH at b) 5 mgmL and c) 25 mgmL d) 960

Structural interpretation of the repeating distances pointed out in c) 961

962

963

964

001 01 1

q Aring-1

25degC

68degC

88degC

111degC

01 1q Aring

-1

02 04 06 08 1q Aring

-1

001 01

10-1

100

101

Inte

nsity

cm

-1

q Aring-1

-2

T= 90degC

SANS WAXS WAXS

a) b) c)

dv

dL d= 144 Aring-1

di

ddH2O

dt

dvdL

dtdi d

d)

9

965

966

Figure S 7 ndash ssNMR MAS (10 kHz) experiments recorded on a G-C181-OH dry powder a) 1H spectra 967

recorded from 110degC to 70degC b) 13C single pulse (SP) spectra recorded with high power 1H decoupling from 968

110degC to 70degC c) 13C cross polarization (CP) spectra recorded at 110degC and contact time tc of 50 micros and 969

3000 micros d) 13C CP spectra recorded at 70degC and for 20 lt tc micros lt 3000 970

971

1

811

910

CH2 CH2

1rsquo

6rsquo

8 6 4 2 0 ppm

110

90

80

70

910

1rsquo

6rsquo16

1

2rsquo3rsquo4rsquo

811

OH

45 40 35 30 25 20 15

ppm

200 180 160 140 120 100 80 60 40 20 0

ppm140 120 100 80

ppm

CH2-gauche

3004

110

90

80

70

65

60

50

25

C91013003

C9 - 12869

CH2-all-trans

3198

C1rsquo10335

a) b)

1H

T degCT degC

13C Single Pulse MAS

A

B

A

B

70degC70degC

5rsquo

215

200 180 160 140 120 100 80 60 40 20 0

ppm

13C CPMAST= 110degC

200 180 160 140 120 100 80 60 40 20 0

ppm

132 131 130 129 128 127 126

ppm

106 105 104 103 102

ppm

13C CPMAST= 70degC

50

tc micros

3000

20

tc micros

50 500 1000 3000

tc micros

C102

13024 C910113003

C92

12869

C1rsquo213082

C1rsquo110443

A B

106 105 104 103 102

ppm

3000

1000

500

50

20

AB

c)

d)

10

972

Figure S 8 ndash a) Short and d) long T2 components displayed by the 1H transverse relaxation signal c) 1H 973

transverse relaxation signals recorded on a G-C181-OH dry powder from 110degC to 70degC The area of each 974

given resonance is plotted (log-lin) as a function of the echo time (CPMG pulse sequence) for each 975

temperature and fitted using the equation given above T2s and T2l indicate the short and long T2 relaxation 976

times respectively obtained by describing the 1H transverse relaxation signal using a biexponential decay 977

For H1rsquo H5rsquo and H6rsquo the 1H relaxation function could be described by a monoexponential decay 978

- Protons from the sugar group (H1rsquo H5rsquo H6rsquo) show a single T2 relaxation component (Figure 979

6c and Figure S 8) while protons of the acyl chain (CH2 H910 H811) show both short (Figure 980

6c) and long (Figure 6d) T2 relaxation component 981

- The short T2 relaxation time systematically decreases from 110degC to 70degC for all species by a 982

factor ranging from two to seven depending on the specific chemical group Very long T2 are 983

observed at 110degC for all acyl species while nearly unchanged values are measured below 984

90degC without any clear trend 985

- The sugar protons (H1rsquo H5rsquo H6rsquo) have the shortest set of T2 values (lt 1 ms) below 100degC 986

The long component of T2 generally ranges between 4 ms and 12 ms 987

0 1 2 3 4 5 6 7 8

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1 2 3 4 5 6 7 8

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

H1

H8H11

H9H10

CH2

H1rsquo

H6rsquo

CH2

H910H811 H1CH2

H1rsquo H5rsquo6rsquo

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

OH

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45 T= 110degC

T= 90degC

T= 80degC

T= 70degC

Log

In

ten

sity

Time ms

Fit

)

CH2 H910 H811 H1 H56 H1 OH

00

05

10

15

20

25

30

35

40

T2 m

s

T= 110degC

T= 90degC

T= 80degC

T= 70degC

CH2 H910 H811 H1

0

2

4

6

8

10

12

T2 m

s

T= 110degC

T= 90degC

T= 80degC

T= 70degC

Short T2(1H) Long T2(1H)

a) b)

c)

11

Table S 1 ndash 1H chemical shifts of G-C181-OH in solution and in the solid-state Solid-state spectra are 988

recorded at a MAS frequency of 10 kHz in DMSO-d6 broad peak 989

990

δH ppm

Group Ndeg Liquid

(MeOD-d4) Solid

(T= 110degC) Solid

(T= 70degC)

CH2 H1 354 357

CH2 H2 153 156

CH2 H3-712-14 132 135

CH2 H811 203 203

CH H910 534 534

CH2 H15 162 164

CH2 H16 388 383

CH H1 424 432

CH H2 316 335

CH H3 334 335

CH H4 329 335

CH H5 354 351 357

CH2 H6 366 383

OH OH1 432

45 463

OH OH2 492

OH OH3 490

OH OH4 487

OH OH6 445

991

992

H1H9H10H1rsquo

H6rsquo

OH2rsquo

OH1

12

Detailed discussion about the NOESY and DQ-SQ 2D 1H-1H ssNMR experiments presented in 993

Figure 9 994

995

2D 1H-1H ssNMR experiments are a unique tool to explore through-space proximities 996

between 1H nuclei ssNMR offers a large number of possibilities each adapted to a given 997

system The advantage of 2D maps with respect to a dynamics approach is their direct 998

readability in clear two nuclei which exchange their magnetization through-space (whichever 999

mechanism may be responsible for this exchange) will display a specific cross-peak However 1000

the way cross-peaks are presented and interpreted strictly depends on the nature of the pulse 1001

program which in turn is chosen according to the physical nature of the sample 1002

1D temperature-dependent 1H NMR experiments performed on G-C181-OH (Figure 1003

4a in the main text) indicate sharp peaks at 90degC and broad peaks at 70degC Narrow peaks 1004

indicate fast andor large-amplitude molecular tumbling averaging the strong 1H-1H dipolar 1005

interactions This excludes the use of pulse sequences like DQ-SQ which is not efficient under 1006

these conditions For this reason we employ spin-diffusion-based pulse sequences like 1007

NOESY which is more adapted for samples with high molecular mobility When mixing times 1008

are chosen to be short (lt 1 s) one generally explores through-space proximities within about 5 1009

10 Aring On the contrary when molecular motions are restricted dipolar coupling becomes 1010

significant 1H NMR peaks are broader and pulse sequences like DQ-SQ can now be used 1011

with a spatial resolution of less than 5 Aring 1012

Whenever possible DQ-SQ experiments are preferred over spin diffusion-based pulse 1013

sequences because they are more selective in particular for likewise interactions which are 1014

not discriminated in NOESY-type experiments This is illustrated by the typical interaction 1015

schemes for NOESY (top Figure 9a) and DQ-SQ (top Figure 9b) For two given A and B 1016

protons 2D NOESY maps will display two on-diagonal and two square-shaped symmetrical 1017

off-diagonal cross-peaks Off-diagonal cross-peaks illustrate the mutual A-B and B-A 1018

interaction but the on-diagonal peaks are systematic and do not prove any A-A or B-B 1019

interaction Similarly to NOESY 2D DQ-SQ maps display off-diagonal peaks (symmetric 1020

across the diagonal top Figure 9b) upon A-B and B-A interaction but in contrast they will 1021

display on-diagonal peaks only in the presence of self-interaction In the tutorial example of 1022

Figure 9b top the on-diagonal A-A peak indicates mutual A-A interaction while lack of on-1023

diagonal B-B peak evidences the absence of B-B proximities Combining NOESY and DQ-SQ 1024

experiments is capital to study the local arrangement of G-C181-OH and its evolution with 1025

temperature 1026

13

At T= 90degC the H1 and H5rsquo contributions are very close but possible to discriminate 1027

them (δ(H1)= 354 ppm δ(H5rsquo)= 351 ppm) according to 1H NMR chemical shifts in Table S 1028

1 2D 1H-1H NOESY in Figure 9a also discriminates between the two components attributed 1029

on the basis of the interactions with their mutual neighbors (green H1-H2 and H5rsquo-H1rsquo squares 1030

in Figure 9a) On this basis 2D NOESY does not show a direct H1rsquo-H1 correlation (blue red-1031

crossed) but rather an extensive network of intra-sugar interactions These features exclude 1032

suggest intra- and inter-sugar interactions and they exclude head-tail arrangements as a major 1033

molecular configuration at 90degC 1034

At 70degC the 1H NMR peaks are broader indicating a reduction of the T2 (1H) relaxation 1035

times andor residual homonuclear dipolar interactions excluding the use of NOESY In this 1036

case 1H-1H DQ-SQ correlation experiments are best suited Figure 9b presents the 1H-1H DQ-1037

SQ experiment on G-C181-OH at 70degC Peaks are broader than in NOESY and attribution may 1038

be cumbersome However the experiment is performed at 700 MHz and MAS= 22 kHz which 1039

improves spectral resolution with respect to the same experiment performed at 300 MHz and 1040

MAS= 10 kHz The auto-peak at δH= 356 ppm in the SQ dimension is attributed to H1-H1 spin 1041

pairs mainly resulting from intra CH2 interaction while the DQ cross-peaks at δH= 156-356 1042

ppm and δH= 432-356 ppm respectively origin from H2-H1 and H1rsquo-H1 interactions (Figure 1043

9b) DQ-SQ map does not show any auto-peak corresponding to H1rsquo-H1rsquo or Hxrsquo-Hxrsquo (x= 2rsquo 1044

3rsquo 4rsquo) interactions (red-crossed blue Figure 9b) thus excluding any proximity (lt 5 Aring) between 1045

close glucose neighbors On the contrary DQ-SQ maps (recorded both at 700 MHz Figure 9b 1046

or at 300 MHz not shown) systematically show DQ cross-peaks which can only be attributed 1047

to H1rsquo-H1 interactions (δH= 432-356) The presence of these H1rsquo-H1 interactions can only be 1048

explained by a head-tail arrangement of the G-C181-OH molecules as illustrated below the 1049

DQ-SQ map in Figure 9b 1050

31

28 B Božič G Gomišček V Kralj-Iglič S Svetina and B Žekš Eur Biophys J 2002 793

31 487ndash496 794

29 A Iglič H Haumlgerstrand M Bobrowska-Haumlgerstrand V Arrigler and V Kralj-Iglič 795

Phys Lett A 2003 310 493ndash497 796

30 R Bar-Ziv and E Moses Phys Rev Lett 1994 73 1392ndash1395 797

31 S Chiruvolu H E Warriner E Naranjo S H Idziak J O Raumldler R J Plano J A 798

Zasadzinski and C R Safinya Science (80- ) 1994 266 1222ndash1225 799

32 V Kralj-Iglič A Iglič H Haumlgerstrand and P Peterlin Phys Rev E 2000 61 4230ndash800

4234 801

33 L Mesarec M Drab S Penič V Kralj-Iglič and A Iglič Int J Mol Sci 2021 22 802

2348 803

34 L Van Renterghem S L K W Roelants N Baccile K Uyttersprot M C Taelman 804

B Everaert S Mincke S Ledegen S Debrouwer K Scholtens C V Stevens W 805

Soetaert L Van Renterghem S L K W Roelants N Baccile K Uytersprot M C 806

Taelman B Everaert S Mincke S Ledegen S Debrouwer K Scholtens C V 807

Stevens and W Soetaert Biotechnol Bioeng 2018 115 1195ndash1206 808

35 J Clark and F Deswarte Introduction to Chemicals from Biomass Second Edition 809

John Wiley amp Sons 2015 810

36 K V Sajna R Houmlfer R K Sukumaran L D Gottumukkala and A Pandey in 811

Industrial Biorefineries and White Biotechnology eds A Pandey R Houmlfer M 812

Taherzadeh K M Nampoothiri and C Larroche Amsterdam Oxford Waltham 813

Elsevier 2015 pp 499ndash521 814

37 R Marchant and I M Banat Trends Biotechnol 2012 30 558ndash565 815

38 K Hill and C LeHen-Ferrenbach in Sugar-Based Surfactants Fundamentals and 816

Applications ed C C Ruiz Boca Raton CRC Press 2009 pp 1ndash20 817

39 L Rodrigues I M Banat J Teixeira and R Oliveira J Antimicrob Chemother 818

2006 57 609ndash618 819

40 J D Desai and I M Banat Microbiol Mol Biol Rev 1997 61 47ndash64 820

41 S L K W Roelants K Ciesielska S L De Maeseneire H Moens B Everaert S 821

Verweire Q Denon B Vanlerberghe I N A Van Bogaert P Van der Meeren B 822

Devreese and W Soetaert Biotechnol Bioeng 2016 113 550ndash559 823

42 N Baccile M Selmane P Le Griel S Preacutevost J Perez C V Stevens E Delbeke S 824

Zibek M Guenther W Soetaert I N A Van Bogaert and S Roelants Langmuir 825

2016 32 6343ndash6359 826

32

43 N Baccile A-S Cuvier S Preacutevost C V Stevens E Delbeke J Berton W Soetaert 827

I N A Van Bogaert and S Roelants Langmuir 2016 32 10881ndash10894 828

44 N Baccile C Seyrig A Poirier S A Castro S L K W Roelants and S Abel Green 829

Chem 2021 23 3842ndash3944 830

45 A Salis L Cappai C Carucci D F Parsons and M Monduzzi J Phys Chem Lett 831

2020 11 6805ndash6811 832

46 J Schindelin I Arganda-Carreras E Frise V Kaynig M Longair T Pietzsch S 833

Preibisch C Rueden S Saalfeld B Schmid J-Y Tinevez D J White V 834

Hartenstein K Eliceiri P Tomancak and A Cardona Nat Methods 2012 9 676ndash835

682 836

47 T Parasassi and E Gratton J Fluoresc 1995 5 59ndash69 837

48 M M Dodes Traian F L Gonzaacutelez Flecha and V Levi J Lipid Res 2012 53 609ndash838

616 839

49 D M Owen C Rentero A Magenau A Abu-Siniyeh and K Gaus Nat Protoc 840

2012 7 24ndash35 841

50 W Stillwell in An Introduction to Biological Membranes Academic Press 2016 pp 842

135ndash180 843

51 W Stillwell in An Introduction to Biological Membranes Academic Press 2016 pp 844

181ndash220 845

52 B Tenchov Chem Phys Lipids 1991 57 165ndash177 846

53 R Koynova and M Caffrey Biochim Biophys Acta 1998 1376 91ndash145 847

54 F Schmid S Dolezel O Lenz and S Meinhardt J Phys Conf Ser 2014 487 848

012004 849

55 S Meinhardt R L C Vink and F Schmid Proc Natl Acad Sci U S A 2013 110 850

4476ndash4481 851

56 O Glatter and O Kratky Small Angle X-ray Scattering Academic Press London 852

1982 853

57 B A Cunningham A D Brown D H Wolfe W P Williams and A Brain Phys 854

Rev E 1998 58 3662ndash3672 855

58 M Nadler A Steiner T Dvir O Szekely P Szekely A Ginsburg R Asor R Resh 856

C Tamburu M Peres and U Raviv Soft Matter 2011 7 1512ndash1523 857

59 C Tanford The hydrophobic effect Formation of micelles and biological membranes 858

John Wiley amp Sons Inc New York 1973 859

60 M Masuda and T Shimizu Langmuir 2004 20 5969ndash5977 860

33

61 A Zidovska K K Ewert J Quispe B Carragher C S Potter and C R Safinya 861

Block Liposomes Vesicles of Charged Lipids with Distinctly Shaped Nanoscale 862

Sphere- Pear- Tube- or Rod-Segments Elsevier Inc 1st edn 2009 vol 465 863

62 A Iglic V Kralj-Iglic B Božič M Bobrowska-Haumlgerstrand B Isomaa and H 864

Haumlgerstrand Bioelectrochemistry 2000 52 203ndash211 865

63 T Shimizu M Masuda and H Minamikawa Chem Rev 2005 105 1401ndash1444 866

64 J V Selinger and J M Schnur Phys Rev Lett 1993 71 4091ndash4094 867

65 J M Schnur Science (80- ) 1993 262 1669ndash1676 868

66 B Tenchov R Koynova and G Rapp Biophys J 2001 80 1873ndash1890 869

67 R Koynova and B Tenchov Ser Biomech 2017 31 3ndash19 870

68 S J Angyal and R Le Fur Carbohydr Res 1984 126 15ndash26 871

69 J K Whitesell and M A Minton J Am Chem Soc 1987 109 225ndash228 872

70 T Takayama S Ando and I Ando J Mol Struct 1990 220 245ndash250 873

71 K Gawrisch N V Eldho and I V Polozov Chem Phys Lipids 2002 116 135ndash151 874

72 F Tielens C Gervais J F Lambert F Mauri and D Costa Chem Mater 2008 20 875

3336ndash3344 876

73 G A Jeffrey and Y Yeon Acta Crystallogr Sect B 1986 42 410ndash413 877

74 B Berglund and R W Vaughan J Chem Phys 1980 73 2037ndash2043 878

75 C Gervais C Coelho T Azaiumls J Maquet G Laurent F Pourpoint C Bonhomme 879

P Florian B Alonso G Guerrero P H Mutin and F Mauri J Magn Reson 2007 880

187 131ndash140 881

76 C Akita T Kawaguchi F Kaneko H Yamamoto and M Suzuki J Phys Chem B 882

2004 108 4862ndash4868 883

77 R N Purusottam L Seacutenicourt J J Lacapegravere and P Tekely Biochim Biophys Acta - 884

Biomembr 2015 1848 3134ndash3139 885

78 X Biarneacutes A Ardegravevol A Planas C Rovira A Laio and M Parrinello J Am Chem 886

Soc 2007 129 10686ndash10693 887

79 L M J Kroon‐Batenburg and J Kroon Biopolymers 1990 29 1243ndash1248 888

80 K N Kirschner and R J Woods Proc Natl Acad Sci U S A 2001 98 10541ndash889

10545 890

81 S Svenson B Kirste and J H Fuhrhop J Am Chem Soc 1994 116 11969ndash11975 891

82 I Sack S Macholl J H Fuhrhop and G Buntkowsky Phys Chem Chem Phys 892

2000 2 1781ndash1788 893

83 S Svenson J Koening and J H Fuhrhop J Phys Chem 1994 98 1022ndash1028 894

34

84 S Svenson A Schaefer and J H Fuhrhop J Chem Soc Perkin Trans 2 1994 2 895

1023ndash1028 896

85 K Bock and C Pedersen Adv Carbohydr Chem Biochem 1983 41 27ndash66 897

86 T B Grindley Structure and Conformation of Carbohydrates Springer Berlin 898

Heidelberg 2008 899

87 F Tantakitti J Boekhoven X Wang R V Kazantsev T Yu J Li E Zhuang R 900

Zandi J H Ortony C J Newcomb L C Palmer G S Shekhawat M O De La 901

Cruz G C Schatz and S I Stupp Nat Mater 2016 15 469ndash476 902

903

1

Supporting Information 904

905

Topological connection between vesicles and nanotubes in single-906

component lipid membranes driven by head-tail interactions 907

908

Niki Baccilea Ceacutedric Lorthioira Abdoul Aziz Baa Patrick Le Griela Cristina Coelhob Javier 909

Perezc Wim Soetaertd Sophie L K W Roelantsd 910

911

a Sorbonne Universiteacute Centre National de la Recherche Scientifique Laboratoire de Chimie de 912

la Matiegravere Condenseacutee de Paris LCMCP F-75005 Paris France 913

b Institut des Mateacuteriaux de Paris Centre Sorbonne Universiteacute Paris France 914

c Synchrotron Soleil LrsquoOrme des Merisiers Saint-Aubin BP48 91192 Gif-sur-Yvette Cedex 915

France 916

d InBio Department of Biotechnology Ghent University Ghent Belgium 917

918

919

Corresponding author 920

Dr Niki Baccile 921

E-mail address nikibaccilesorbonne-universitefr 922

Phone +33 1 44 27 56 77 923

924

925

926

2

a) 927

Parameter Method

Dry Matter (DM ) 996 Infrared balance 105degC

Glucose () 012 HPLC-Metacarb

Glycerol () nd HPLC-Metacarb

Free Fatty Acid content () lt001 Internal method BBEPP

GC with FID Detector

Oil () lt001 Internal method BBEPP

GC with FID Detector

Moulds (CFUg) lt10 3M-Nordval ndeg16 B

Yeasts (CFUg) lt10 3M-Nordval ndeg16 B

Anaerobic count (CFUg) lt10 SP-VG M005 B

Protein () tbd BCA protein assay

928

Figure S 1 - a) HPLC-ELSD chromatogram and composition table 929

930

931

932

933

3

b) 934

935

Group Ndeg δ ppm

CH2 1 354

CH2 2 153

CH2 3-712-14 132

CH2 811 203

CH 910 534

CH2 15 162

CH2 16 388

CH 1 424

CH 2 316

CH 3 334

CH 4 329

CH 5 354

CH2 6 366

Figure S 1 (next) ndash b) 1H NMR spectrum (MeOD-d4) recorded on the purified G-C181-OH compounds 936

and related assignment 937

938

2

811910 1rsquo

5rsquo1

3rsquo4rsquo

1rsquo1910

asymp asymp asympHDO MeOD Acetone

6rsquo2rsquo

1615

3-712-14

4

939

940

Figure S 2 ndash Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 130degC 941

5

942

943

Figure S 3 - Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 90degC 944

945

90degC

1

22

6

946

947

948

949

Figure S 4 ndash Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 70degC 950

951

21

11

1

2

5

3

4

45

5

6

7

7

7

952

953

954

Figure S 5 ndash Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 25degC 955

956

957

1

1 1

8

958

Figure S 6 ndash a) SANS experiment recorded at 90degC for a 5 mgmL G-C181-OH aqueous solution b-c) 959

WAXS experiments recorded on aqueous solutions of G-C181-OH at b) 5 mgmL and c) 25 mgmL d) 960

Structural interpretation of the repeating distances pointed out in c) 961

962

963

964

001 01 1

q Aring-1

25degC

68degC

88degC

111degC

01 1q Aring

-1

02 04 06 08 1q Aring

-1

001 01

10-1

100

101

Inte

nsity

cm

-1

q Aring-1

-2

T= 90degC

SANS WAXS WAXS

a) b) c)

dv

dL d= 144 Aring-1

di

ddH2O

dt

dvdL

dtdi d

d)

9

965

966

Figure S 7 ndash ssNMR MAS (10 kHz) experiments recorded on a G-C181-OH dry powder a) 1H spectra 967

recorded from 110degC to 70degC b) 13C single pulse (SP) spectra recorded with high power 1H decoupling from 968

110degC to 70degC c) 13C cross polarization (CP) spectra recorded at 110degC and contact time tc of 50 micros and 969

3000 micros d) 13C CP spectra recorded at 70degC and for 20 lt tc micros lt 3000 970

971

1

811

910

CH2 CH2

1rsquo

6rsquo

8 6 4 2 0 ppm

110

90

80

70

910

1rsquo

6rsquo16

1

2rsquo3rsquo4rsquo

811

OH

45 40 35 30 25 20 15

ppm

200 180 160 140 120 100 80 60 40 20 0

ppm140 120 100 80

ppm

CH2-gauche

3004

110

90

80

70

65

60

50

25

C91013003

C9 - 12869

CH2-all-trans

3198

C1rsquo10335

a) b)

1H

T degCT degC

13C Single Pulse MAS

A

B

A

B

70degC70degC

5rsquo

215

200 180 160 140 120 100 80 60 40 20 0

ppm

13C CPMAST= 110degC

200 180 160 140 120 100 80 60 40 20 0

ppm

132 131 130 129 128 127 126

ppm

106 105 104 103 102

ppm

13C CPMAST= 70degC

50

tc micros

3000

20

tc micros

50 500 1000 3000

tc micros

C102

13024 C910113003

C92

12869

C1rsquo213082

C1rsquo110443

A B

106 105 104 103 102

ppm

3000

1000

500

50

20

AB

c)

d)

10

972

Figure S 8 ndash a) Short and d) long T2 components displayed by the 1H transverse relaxation signal c) 1H 973

transverse relaxation signals recorded on a G-C181-OH dry powder from 110degC to 70degC The area of each 974

given resonance is plotted (log-lin) as a function of the echo time (CPMG pulse sequence) for each 975

temperature and fitted using the equation given above T2s and T2l indicate the short and long T2 relaxation 976

times respectively obtained by describing the 1H transverse relaxation signal using a biexponential decay 977

For H1rsquo H5rsquo and H6rsquo the 1H relaxation function could be described by a monoexponential decay 978

- Protons from the sugar group (H1rsquo H5rsquo H6rsquo) show a single T2 relaxation component (Figure 979

6c and Figure S 8) while protons of the acyl chain (CH2 H910 H811) show both short (Figure 980

6c) and long (Figure 6d) T2 relaxation component 981

- The short T2 relaxation time systematically decreases from 110degC to 70degC for all species by a 982

factor ranging from two to seven depending on the specific chemical group Very long T2 are 983

observed at 110degC for all acyl species while nearly unchanged values are measured below 984

90degC without any clear trend 985

- The sugar protons (H1rsquo H5rsquo H6rsquo) have the shortest set of T2 values (lt 1 ms) below 100degC 986

The long component of T2 generally ranges between 4 ms and 12 ms 987

0 1 2 3 4 5 6 7 8

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1 2 3 4 5 6 7 8

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

H1

H8H11

H9H10

CH2

H1rsquo

H6rsquo

CH2

H910H811 H1CH2

H1rsquo H5rsquo6rsquo

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

OH

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45 T= 110degC

T= 90degC

T= 80degC

T= 70degC

Log

In

ten

sity

Time ms

Fit

)

CH2 H910 H811 H1 H56 H1 OH

00

05

10

15

20

25

30

35

40

T2 m

s

T= 110degC

T= 90degC

T= 80degC

T= 70degC

CH2 H910 H811 H1

0

2

4

6

8

10

12

T2 m

s

T= 110degC

T= 90degC

T= 80degC

T= 70degC

Short T2(1H) Long T2(1H)

a) b)

c)

11

Table S 1 ndash 1H chemical shifts of G-C181-OH in solution and in the solid-state Solid-state spectra are 988

recorded at a MAS frequency of 10 kHz in DMSO-d6 broad peak 989

990

δH ppm

Group Ndeg Liquid

(MeOD-d4) Solid

(T= 110degC) Solid

(T= 70degC)

CH2 H1 354 357

CH2 H2 153 156

CH2 H3-712-14 132 135

CH2 H811 203 203

CH H910 534 534

CH2 H15 162 164

CH2 H16 388 383

CH H1 424 432

CH H2 316 335

CH H3 334 335

CH H4 329 335

CH H5 354 351 357

CH2 H6 366 383

OH OH1 432

45 463

OH OH2 492

OH OH3 490

OH OH4 487

OH OH6 445

991

992

H1H9H10H1rsquo

H6rsquo

OH2rsquo

OH1

12

Detailed discussion about the NOESY and DQ-SQ 2D 1H-1H ssNMR experiments presented in 993

Figure 9 994

995

2D 1H-1H ssNMR experiments are a unique tool to explore through-space proximities 996

between 1H nuclei ssNMR offers a large number of possibilities each adapted to a given 997

system The advantage of 2D maps with respect to a dynamics approach is their direct 998

readability in clear two nuclei which exchange their magnetization through-space (whichever 999

mechanism may be responsible for this exchange) will display a specific cross-peak However 1000

the way cross-peaks are presented and interpreted strictly depends on the nature of the pulse 1001

program which in turn is chosen according to the physical nature of the sample 1002

1D temperature-dependent 1H NMR experiments performed on G-C181-OH (Figure 1003

4a in the main text) indicate sharp peaks at 90degC and broad peaks at 70degC Narrow peaks 1004

indicate fast andor large-amplitude molecular tumbling averaging the strong 1H-1H dipolar 1005

interactions This excludes the use of pulse sequences like DQ-SQ which is not efficient under 1006

these conditions For this reason we employ spin-diffusion-based pulse sequences like 1007

NOESY which is more adapted for samples with high molecular mobility When mixing times 1008

are chosen to be short (lt 1 s) one generally explores through-space proximities within about 5 1009

10 Aring On the contrary when molecular motions are restricted dipolar coupling becomes 1010

significant 1H NMR peaks are broader and pulse sequences like DQ-SQ can now be used 1011

with a spatial resolution of less than 5 Aring 1012

Whenever possible DQ-SQ experiments are preferred over spin diffusion-based pulse 1013

sequences because they are more selective in particular for likewise interactions which are 1014

not discriminated in NOESY-type experiments This is illustrated by the typical interaction 1015

schemes for NOESY (top Figure 9a) and DQ-SQ (top Figure 9b) For two given A and B 1016

protons 2D NOESY maps will display two on-diagonal and two square-shaped symmetrical 1017

off-diagonal cross-peaks Off-diagonal cross-peaks illustrate the mutual A-B and B-A 1018

interaction but the on-diagonal peaks are systematic and do not prove any A-A or B-B 1019

interaction Similarly to NOESY 2D DQ-SQ maps display off-diagonal peaks (symmetric 1020

across the diagonal top Figure 9b) upon A-B and B-A interaction but in contrast they will 1021

display on-diagonal peaks only in the presence of self-interaction In the tutorial example of 1022

Figure 9b top the on-diagonal A-A peak indicates mutual A-A interaction while lack of on-1023

diagonal B-B peak evidences the absence of B-B proximities Combining NOESY and DQ-SQ 1024

experiments is capital to study the local arrangement of G-C181-OH and its evolution with 1025

temperature 1026

13

At T= 90degC the H1 and H5rsquo contributions are very close but possible to discriminate 1027

them (δ(H1)= 354 ppm δ(H5rsquo)= 351 ppm) according to 1H NMR chemical shifts in Table S 1028

1 2D 1H-1H NOESY in Figure 9a also discriminates between the two components attributed 1029

on the basis of the interactions with their mutual neighbors (green H1-H2 and H5rsquo-H1rsquo squares 1030

in Figure 9a) On this basis 2D NOESY does not show a direct H1rsquo-H1 correlation (blue red-1031

crossed) but rather an extensive network of intra-sugar interactions These features exclude 1032

suggest intra- and inter-sugar interactions and they exclude head-tail arrangements as a major 1033

molecular configuration at 90degC 1034

At 70degC the 1H NMR peaks are broader indicating a reduction of the T2 (1H) relaxation 1035

times andor residual homonuclear dipolar interactions excluding the use of NOESY In this 1036

case 1H-1H DQ-SQ correlation experiments are best suited Figure 9b presents the 1H-1H DQ-1037

SQ experiment on G-C181-OH at 70degC Peaks are broader than in NOESY and attribution may 1038

be cumbersome However the experiment is performed at 700 MHz and MAS= 22 kHz which 1039

improves spectral resolution with respect to the same experiment performed at 300 MHz and 1040

MAS= 10 kHz The auto-peak at δH= 356 ppm in the SQ dimension is attributed to H1-H1 spin 1041

pairs mainly resulting from intra CH2 interaction while the DQ cross-peaks at δH= 156-356 1042

ppm and δH= 432-356 ppm respectively origin from H2-H1 and H1rsquo-H1 interactions (Figure 1043

9b) DQ-SQ map does not show any auto-peak corresponding to H1rsquo-H1rsquo or Hxrsquo-Hxrsquo (x= 2rsquo 1044

3rsquo 4rsquo) interactions (red-crossed blue Figure 9b) thus excluding any proximity (lt 5 Aring) between 1045

close glucose neighbors On the contrary DQ-SQ maps (recorded both at 700 MHz Figure 9b 1046

or at 300 MHz not shown) systematically show DQ cross-peaks which can only be attributed 1047

to H1rsquo-H1 interactions (δH= 432-356) The presence of these H1rsquo-H1 interactions can only be 1048

explained by a head-tail arrangement of the G-C181-OH molecules as illustrated below the 1049

DQ-SQ map in Figure 9b 1050

32

43 N Baccile A-S Cuvier S Preacutevost C V Stevens E Delbeke J Berton W Soetaert 827

I N A Van Bogaert and S Roelants Langmuir 2016 32 10881ndash10894 828

44 N Baccile C Seyrig A Poirier S A Castro S L K W Roelants and S Abel Green 829

Chem 2021 23 3842ndash3944 830

45 A Salis L Cappai C Carucci D F Parsons and M Monduzzi J Phys Chem Lett 831

2020 11 6805ndash6811 832

46 J Schindelin I Arganda-Carreras E Frise V Kaynig M Longair T Pietzsch S 833

Preibisch C Rueden S Saalfeld B Schmid J-Y Tinevez D J White V 834

Hartenstein K Eliceiri P Tomancak and A Cardona Nat Methods 2012 9 676ndash835

682 836

47 T Parasassi and E Gratton J Fluoresc 1995 5 59ndash69 837

48 M M Dodes Traian F L Gonzaacutelez Flecha and V Levi J Lipid Res 2012 53 609ndash838

616 839

49 D M Owen C Rentero A Magenau A Abu-Siniyeh and K Gaus Nat Protoc 840

2012 7 24ndash35 841

50 W Stillwell in An Introduction to Biological Membranes Academic Press 2016 pp 842

135ndash180 843

51 W Stillwell in An Introduction to Biological Membranes Academic Press 2016 pp 844

181ndash220 845

52 B Tenchov Chem Phys Lipids 1991 57 165ndash177 846

53 R Koynova and M Caffrey Biochim Biophys Acta 1998 1376 91ndash145 847

54 F Schmid S Dolezel O Lenz and S Meinhardt J Phys Conf Ser 2014 487 848

012004 849

55 S Meinhardt R L C Vink and F Schmid Proc Natl Acad Sci U S A 2013 110 850

4476ndash4481 851

56 O Glatter and O Kratky Small Angle X-ray Scattering Academic Press London 852

1982 853

57 B A Cunningham A D Brown D H Wolfe W P Williams and A Brain Phys 854

Rev E 1998 58 3662ndash3672 855

58 M Nadler A Steiner T Dvir O Szekely P Szekely A Ginsburg R Asor R Resh 856

C Tamburu M Peres and U Raviv Soft Matter 2011 7 1512ndash1523 857

59 C Tanford The hydrophobic effect Formation of micelles and biological membranes 858

John Wiley amp Sons Inc New York 1973 859

60 M Masuda and T Shimizu Langmuir 2004 20 5969ndash5977 860

33

61 A Zidovska K K Ewert J Quispe B Carragher C S Potter and C R Safinya 861

Block Liposomes Vesicles of Charged Lipids with Distinctly Shaped Nanoscale 862

Sphere- Pear- Tube- or Rod-Segments Elsevier Inc 1st edn 2009 vol 465 863

62 A Iglic V Kralj-Iglic B Božič M Bobrowska-Haumlgerstrand B Isomaa and H 864

Haumlgerstrand Bioelectrochemistry 2000 52 203ndash211 865

63 T Shimizu M Masuda and H Minamikawa Chem Rev 2005 105 1401ndash1444 866

64 J V Selinger and J M Schnur Phys Rev Lett 1993 71 4091ndash4094 867

65 J M Schnur Science (80- ) 1993 262 1669ndash1676 868

66 B Tenchov R Koynova and G Rapp Biophys J 2001 80 1873ndash1890 869

67 R Koynova and B Tenchov Ser Biomech 2017 31 3ndash19 870

68 S J Angyal and R Le Fur Carbohydr Res 1984 126 15ndash26 871

69 J K Whitesell and M A Minton J Am Chem Soc 1987 109 225ndash228 872

70 T Takayama S Ando and I Ando J Mol Struct 1990 220 245ndash250 873

71 K Gawrisch N V Eldho and I V Polozov Chem Phys Lipids 2002 116 135ndash151 874

72 F Tielens C Gervais J F Lambert F Mauri and D Costa Chem Mater 2008 20 875

3336ndash3344 876

73 G A Jeffrey and Y Yeon Acta Crystallogr Sect B 1986 42 410ndash413 877

74 B Berglund and R W Vaughan J Chem Phys 1980 73 2037ndash2043 878

75 C Gervais C Coelho T Azaiumls J Maquet G Laurent F Pourpoint C Bonhomme 879

P Florian B Alonso G Guerrero P H Mutin and F Mauri J Magn Reson 2007 880

187 131ndash140 881

76 C Akita T Kawaguchi F Kaneko H Yamamoto and M Suzuki J Phys Chem B 882

2004 108 4862ndash4868 883

77 R N Purusottam L Seacutenicourt J J Lacapegravere and P Tekely Biochim Biophys Acta - 884

Biomembr 2015 1848 3134ndash3139 885

78 X Biarneacutes A Ardegravevol A Planas C Rovira A Laio and M Parrinello J Am Chem 886

Soc 2007 129 10686ndash10693 887

79 L M J Kroon‐Batenburg and J Kroon Biopolymers 1990 29 1243ndash1248 888

80 K N Kirschner and R J Woods Proc Natl Acad Sci U S A 2001 98 10541ndash889

10545 890

81 S Svenson B Kirste and J H Fuhrhop J Am Chem Soc 1994 116 11969ndash11975 891

82 I Sack S Macholl J H Fuhrhop and G Buntkowsky Phys Chem Chem Phys 892

2000 2 1781ndash1788 893

83 S Svenson J Koening and J H Fuhrhop J Phys Chem 1994 98 1022ndash1028 894

34

84 S Svenson A Schaefer and J H Fuhrhop J Chem Soc Perkin Trans 2 1994 2 895

1023ndash1028 896

85 K Bock and C Pedersen Adv Carbohydr Chem Biochem 1983 41 27ndash66 897

86 T B Grindley Structure and Conformation of Carbohydrates Springer Berlin 898

Heidelberg 2008 899

87 F Tantakitti J Boekhoven X Wang R V Kazantsev T Yu J Li E Zhuang R 900

Zandi J H Ortony C J Newcomb L C Palmer G S Shekhawat M O De La 901

Cruz G C Schatz and S I Stupp Nat Mater 2016 15 469ndash476 902

903

1

Supporting Information 904

905

Topological connection between vesicles and nanotubes in single-906

component lipid membranes driven by head-tail interactions 907

908

Niki Baccilea Ceacutedric Lorthioira Abdoul Aziz Baa Patrick Le Griela Cristina Coelhob Javier 909

Perezc Wim Soetaertd Sophie L K W Roelantsd 910

911

a Sorbonne Universiteacute Centre National de la Recherche Scientifique Laboratoire de Chimie de 912

la Matiegravere Condenseacutee de Paris LCMCP F-75005 Paris France 913

b Institut des Mateacuteriaux de Paris Centre Sorbonne Universiteacute Paris France 914

c Synchrotron Soleil LrsquoOrme des Merisiers Saint-Aubin BP48 91192 Gif-sur-Yvette Cedex 915

France 916

d InBio Department of Biotechnology Ghent University Ghent Belgium 917

918

919

Corresponding author 920

Dr Niki Baccile 921

E-mail address nikibaccilesorbonne-universitefr 922

Phone +33 1 44 27 56 77 923

924

925

926

2

a) 927

Parameter Method

Dry Matter (DM ) 996 Infrared balance 105degC

Glucose () 012 HPLC-Metacarb

Glycerol () nd HPLC-Metacarb

Free Fatty Acid content () lt001 Internal method BBEPP

GC with FID Detector

Oil () lt001 Internal method BBEPP

GC with FID Detector

Moulds (CFUg) lt10 3M-Nordval ndeg16 B

Yeasts (CFUg) lt10 3M-Nordval ndeg16 B

Anaerobic count (CFUg) lt10 SP-VG M005 B

Protein () tbd BCA protein assay

928

Figure S 1 - a) HPLC-ELSD chromatogram and composition table 929

930

931

932

933

3

b) 934

935

Group Ndeg δ ppm

CH2 1 354

CH2 2 153

CH2 3-712-14 132

CH2 811 203

CH 910 534

CH2 15 162

CH2 16 388

CH 1 424

CH 2 316

CH 3 334

CH 4 329

CH 5 354

CH2 6 366

Figure S 1 (next) ndash b) 1H NMR spectrum (MeOD-d4) recorded on the purified G-C181-OH compounds 936

and related assignment 937

938

2

811910 1rsquo

5rsquo1

3rsquo4rsquo

1rsquo1910

asymp asymp asympHDO MeOD Acetone

6rsquo2rsquo

1615

3-712-14

4

939

940

Figure S 2 ndash Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 130degC 941

5

942

943

Figure S 3 - Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 90degC 944

945

90degC

1

22

6

946

947

948

949

Figure S 4 ndash Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 70degC 950

951

21

11

1

2

5

3

4

45

5

6

7

7

7

952

953

954

Figure S 5 ndash Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 25degC 955

956

957

1

1 1

8

958

Figure S 6 ndash a) SANS experiment recorded at 90degC for a 5 mgmL G-C181-OH aqueous solution b-c) 959

WAXS experiments recorded on aqueous solutions of G-C181-OH at b) 5 mgmL and c) 25 mgmL d) 960

Structural interpretation of the repeating distances pointed out in c) 961

962

963

964

001 01 1

q Aring-1

25degC

68degC

88degC

111degC

01 1q Aring

-1

02 04 06 08 1q Aring

-1

001 01

10-1

100

101

Inte

nsity

cm

-1

q Aring-1

-2

T= 90degC

SANS WAXS WAXS

a) b) c)

dv

dL d= 144 Aring-1

di

ddH2O

dt

dvdL

dtdi d

d)

9

965

966

Figure S 7 ndash ssNMR MAS (10 kHz) experiments recorded on a G-C181-OH dry powder a) 1H spectra 967

recorded from 110degC to 70degC b) 13C single pulse (SP) spectra recorded with high power 1H decoupling from 968

110degC to 70degC c) 13C cross polarization (CP) spectra recorded at 110degC and contact time tc of 50 micros and 969

3000 micros d) 13C CP spectra recorded at 70degC and for 20 lt tc micros lt 3000 970

971

1

811

910

CH2 CH2

1rsquo

6rsquo

8 6 4 2 0 ppm

110

90

80

70

910

1rsquo

6rsquo16

1

2rsquo3rsquo4rsquo

811

OH

45 40 35 30 25 20 15

ppm

200 180 160 140 120 100 80 60 40 20 0

ppm140 120 100 80

ppm

CH2-gauche

3004

110

90

80

70

65

60

50

25

C91013003

C9 - 12869

CH2-all-trans

3198

C1rsquo10335

a) b)

1H

T degCT degC

13C Single Pulse MAS

A

B

A

B

70degC70degC

5rsquo

215

200 180 160 140 120 100 80 60 40 20 0

ppm

13C CPMAST= 110degC

200 180 160 140 120 100 80 60 40 20 0

ppm

132 131 130 129 128 127 126

ppm

106 105 104 103 102

ppm

13C CPMAST= 70degC

50

tc micros

3000

20

tc micros

50 500 1000 3000

tc micros

C102

13024 C910113003

C92

12869

C1rsquo213082

C1rsquo110443

A B

106 105 104 103 102

ppm

3000

1000

500

50

20

AB

c)

d)

10

972

Figure S 8 ndash a) Short and d) long T2 components displayed by the 1H transverse relaxation signal c) 1H 973

transverse relaxation signals recorded on a G-C181-OH dry powder from 110degC to 70degC The area of each 974

given resonance is plotted (log-lin) as a function of the echo time (CPMG pulse sequence) for each 975

temperature and fitted using the equation given above T2s and T2l indicate the short and long T2 relaxation 976

times respectively obtained by describing the 1H transverse relaxation signal using a biexponential decay 977

For H1rsquo H5rsquo and H6rsquo the 1H relaxation function could be described by a monoexponential decay 978

- Protons from the sugar group (H1rsquo H5rsquo H6rsquo) show a single T2 relaxation component (Figure 979

6c and Figure S 8) while protons of the acyl chain (CH2 H910 H811) show both short (Figure 980

6c) and long (Figure 6d) T2 relaxation component 981

- The short T2 relaxation time systematically decreases from 110degC to 70degC for all species by a 982

factor ranging from two to seven depending on the specific chemical group Very long T2 are 983

observed at 110degC for all acyl species while nearly unchanged values are measured below 984

90degC without any clear trend 985

- The sugar protons (H1rsquo H5rsquo H6rsquo) have the shortest set of T2 values (lt 1 ms) below 100degC 986

The long component of T2 generally ranges between 4 ms and 12 ms 987

0 1 2 3 4 5 6 7 8

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1 2 3 4 5 6 7 8

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

H1

H8H11

H9H10

CH2

H1rsquo

H6rsquo

CH2

H910H811 H1CH2

H1rsquo H5rsquo6rsquo

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

OH

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45 T= 110degC

T= 90degC

T= 80degC

T= 70degC

Log

In

ten

sity

Time ms

Fit

)

CH2 H910 H811 H1 H56 H1 OH

00

05

10

15

20

25

30

35

40

T2 m

s

T= 110degC

T= 90degC

T= 80degC

T= 70degC

CH2 H910 H811 H1

0

2

4

6

8

10

12

T2 m

s

T= 110degC

T= 90degC

T= 80degC

T= 70degC

Short T2(1H) Long T2(1H)

a) b)

c)

11

Table S 1 ndash 1H chemical shifts of G-C181-OH in solution and in the solid-state Solid-state spectra are 988

recorded at a MAS frequency of 10 kHz in DMSO-d6 broad peak 989

990

δH ppm

Group Ndeg Liquid

(MeOD-d4) Solid

(T= 110degC) Solid

(T= 70degC)

CH2 H1 354 357

CH2 H2 153 156

CH2 H3-712-14 132 135

CH2 H811 203 203

CH H910 534 534

CH2 H15 162 164

CH2 H16 388 383

CH H1 424 432

CH H2 316 335

CH H3 334 335

CH H4 329 335

CH H5 354 351 357

CH2 H6 366 383

OH OH1 432

45 463

OH OH2 492

OH OH3 490

OH OH4 487

OH OH6 445

991

992

H1H9H10H1rsquo

H6rsquo

OH2rsquo

OH1

12

Detailed discussion about the NOESY and DQ-SQ 2D 1H-1H ssNMR experiments presented in 993

Figure 9 994

995

2D 1H-1H ssNMR experiments are a unique tool to explore through-space proximities 996

between 1H nuclei ssNMR offers a large number of possibilities each adapted to a given 997

system The advantage of 2D maps with respect to a dynamics approach is their direct 998

readability in clear two nuclei which exchange their magnetization through-space (whichever 999

mechanism may be responsible for this exchange) will display a specific cross-peak However 1000

the way cross-peaks are presented and interpreted strictly depends on the nature of the pulse 1001

program which in turn is chosen according to the physical nature of the sample 1002

1D temperature-dependent 1H NMR experiments performed on G-C181-OH (Figure 1003

4a in the main text) indicate sharp peaks at 90degC and broad peaks at 70degC Narrow peaks 1004

indicate fast andor large-amplitude molecular tumbling averaging the strong 1H-1H dipolar 1005

interactions This excludes the use of pulse sequences like DQ-SQ which is not efficient under 1006

these conditions For this reason we employ spin-diffusion-based pulse sequences like 1007

NOESY which is more adapted for samples with high molecular mobility When mixing times 1008

are chosen to be short (lt 1 s) one generally explores through-space proximities within about 5 1009

10 Aring On the contrary when molecular motions are restricted dipolar coupling becomes 1010

significant 1H NMR peaks are broader and pulse sequences like DQ-SQ can now be used 1011

with a spatial resolution of less than 5 Aring 1012

Whenever possible DQ-SQ experiments are preferred over spin diffusion-based pulse 1013

sequences because they are more selective in particular for likewise interactions which are 1014

not discriminated in NOESY-type experiments This is illustrated by the typical interaction 1015

schemes for NOESY (top Figure 9a) and DQ-SQ (top Figure 9b) For two given A and B 1016

protons 2D NOESY maps will display two on-diagonal and two square-shaped symmetrical 1017

off-diagonal cross-peaks Off-diagonal cross-peaks illustrate the mutual A-B and B-A 1018

interaction but the on-diagonal peaks are systematic and do not prove any A-A or B-B 1019

interaction Similarly to NOESY 2D DQ-SQ maps display off-diagonal peaks (symmetric 1020

across the diagonal top Figure 9b) upon A-B and B-A interaction but in contrast they will 1021

display on-diagonal peaks only in the presence of self-interaction In the tutorial example of 1022

Figure 9b top the on-diagonal A-A peak indicates mutual A-A interaction while lack of on-1023

diagonal B-B peak evidences the absence of B-B proximities Combining NOESY and DQ-SQ 1024

experiments is capital to study the local arrangement of G-C181-OH and its evolution with 1025

temperature 1026

13

At T= 90degC the H1 and H5rsquo contributions are very close but possible to discriminate 1027

them (δ(H1)= 354 ppm δ(H5rsquo)= 351 ppm) according to 1H NMR chemical shifts in Table S 1028

1 2D 1H-1H NOESY in Figure 9a also discriminates between the two components attributed 1029

on the basis of the interactions with their mutual neighbors (green H1-H2 and H5rsquo-H1rsquo squares 1030

in Figure 9a) On this basis 2D NOESY does not show a direct H1rsquo-H1 correlation (blue red-1031

crossed) but rather an extensive network of intra-sugar interactions These features exclude 1032

suggest intra- and inter-sugar interactions and they exclude head-tail arrangements as a major 1033

molecular configuration at 90degC 1034

At 70degC the 1H NMR peaks are broader indicating a reduction of the T2 (1H) relaxation 1035

times andor residual homonuclear dipolar interactions excluding the use of NOESY In this 1036

case 1H-1H DQ-SQ correlation experiments are best suited Figure 9b presents the 1H-1H DQ-1037

SQ experiment on G-C181-OH at 70degC Peaks are broader than in NOESY and attribution may 1038

be cumbersome However the experiment is performed at 700 MHz and MAS= 22 kHz which 1039

improves spectral resolution with respect to the same experiment performed at 300 MHz and 1040

MAS= 10 kHz The auto-peak at δH= 356 ppm in the SQ dimension is attributed to H1-H1 spin 1041

pairs mainly resulting from intra CH2 interaction while the DQ cross-peaks at δH= 156-356 1042

ppm and δH= 432-356 ppm respectively origin from H2-H1 and H1rsquo-H1 interactions (Figure 1043

9b) DQ-SQ map does not show any auto-peak corresponding to H1rsquo-H1rsquo or Hxrsquo-Hxrsquo (x= 2rsquo 1044

3rsquo 4rsquo) interactions (red-crossed blue Figure 9b) thus excluding any proximity (lt 5 Aring) between 1045

close glucose neighbors On the contrary DQ-SQ maps (recorded both at 700 MHz Figure 9b 1046

or at 300 MHz not shown) systematically show DQ cross-peaks which can only be attributed 1047

to H1rsquo-H1 interactions (δH= 432-356) The presence of these H1rsquo-H1 interactions can only be 1048

explained by a head-tail arrangement of the G-C181-OH molecules as illustrated below the 1049

DQ-SQ map in Figure 9b 1050

33

61 A Zidovska K K Ewert J Quispe B Carragher C S Potter and C R Safinya 861

Block Liposomes Vesicles of Charged Lipids with Distinctly Shaped Nanoscale 862

Sphere- Pear- Tube- or Rod-Segments Elsevier Inc 1st edn 2009 vol 465 863

62 A Iglic V Kralj-Iglic B Božič M Bobrowska-Haumlgerstrand B Isomaa and H 864

Haumlgerstrand Bioelectrochemistry 2000 52 203ndash211 865

63 T Shimizu M Masuda and H Minamikawa Chem Rev 2005 105 1401ndash1444 866

64 J V Selinger and J M Schnur Phys Rev Lett 1993 71 4091ndash4094 867

65 J M Schnur Science (80- ) 1993 262 1669ndash1676 868

66 B Tenchov R Koynova and G Rapp Biophys J 2001 80 1873ndash1890 869

67 R Koynova and B Tenchov Ser Biomech 2017 31 3ndash19 870

68 S J Angyal and R Le Fur Carbohydr Res 1984 126 15ndash26 871

69 J K Whitesell and M A Minton J Am Chem Soc 1987 109 225ndash228 872

70 T Takayama S Ando and I Ando J Mol Struct 1990 220 245ndash250 873

71 K Gawrisch N V Eldho and I V Polozov Chem Phys Lipids 2002 116 135ndash151 874

72 F Tielens C Gervais J F Lambert F Mauri and D Costa Chem Mater 2008 20 875

3336ndash3344 876

73 G A Jeffrey and Y Yeon Acta Crystallogr Sect B 1986 42 410ndash413 877

74 B Berglund and R W Vaughan J Chem Phys 1980 73 2037ndash2043 878

75 C Gervais C Coelho T Azaiumls J Maquet G Laurent F Pourpoint C Bonhomme 879

P Florian B Alonso G Guerrero P H Mutin and F Mauri J Magn Reson 2007 880

187 131ndash140 881

76 C Akita T Kawaguchi F Kaneko H Yamamoto and M Suzuki J Phys Chem B 882

2004 108 4862ndash4868 883

77 R N Purusottam L Seacutenicourt J J Lacapegravere and P Tekely Biochim Biophys Acta - 884

Biomembr 2015 1848 3134ndash3139 885

78 X Biarneacutes A Ardegravevol A Planas C Rovira A Laio and M Parrinello J Am Chem 886

Soc 2007 129 10686ndash10693 887

79 L M J Kroon‐Batenburg and J Kroon Biopolymers 1990 29 1243ndash1248 888

80 K N Kirschner and R J Woods Proc Natl Acad Sci U S A 2001 98 10541ndash889

10545 890

81 S Svenson B Kirste and J H Fuhrhop J Am Chem Soc 1994 116 11969ndash11975 891

82 I Sack S Macholl J H Fuhrhop and G Buntkowsky Phys Chem Chem Phys 892

2000 2 1781ndash1788 893

83 S Svenson J Koening and J H Fuhrhop J Phys Chem 1994 98 1022ndash1028 894

34

84 S Svenson A Schaefer and J H Fuhrhop J Chem Soc Perkin Trans 2 1994 2 895

1023ndash1028 896

85 K Bock and C Pedersen Adv Carbohydr Chem Biochem 1983 41 27ndash66 897

86 T B Grindley Structure and Conformation of Carbohydrates Springer Berlin 898

Heidelberg 2008 899

87 F Tantakitti J Boekhoven X Wang R V Kazantsev T Yu J Li E Zhuang R 900

Zandi J H Ortony C J Newcomb L C Palmer G S Shekhawat M O De La 901

Cruz G C Schatz and S I Stupp Nat Mater 2016 15 469ndash476 902

903

1

Supporting Information 904

905

Topological connection between vesicles and nanotubes in single-906

component lipid membranes driven by head-tail interactions 907

908

Niki Baccilea Ceacutedric Lorthioira Abdoul Aziz Baa Patrick Le Griela Cristina Coelhob Javier 909

Perezc Wim Soetaertd Sophie L K W Roelantsd 910

911

a Sorbonne Universiteacute Centre National de la Recherche Scientifique Laboratoire de Chimie de 912

la Matiegravere Condenseacutee de Paris LCMCP F-75005 Paris France 913

b Institut des Mateacuteriaux de Paris Centre Sorbonne Universiteacute Paris France 914

c Synchrotron Soleil LrsquoOrme des Merisiers Saint-Aubin BP48 91192 Gif-sur-Yvette Cedex 915

France 916

d InBio Department of Biotechnology Ghent University Ghent Belgium 917

918

919

Corresponding author 920

Dr Niki Baccile 921

E-mail address nikibaccilesorbonne-universitefr 922

Phone +33 1 44 27 56 77 923

924

925

926

2

a) 927

Parameter Method

Dry Matter (DM ) 996 Infrared balance 105degC

Glucose () 012 HPLC-Metacarb

Glycerol () nd HPLC-Metacarb

Free Fatty Acid content () lt001 Internal method BBEPP

GC with FID Detector

Oil () lt001 Internal method BBEPP

GC with FID Detector

Moulds (CFUg) lt10 3M-Nordval ndeg16 B

Yeasts (CFUg) lt10 3M-Nordval ndeg16 B

Anaerobic count (CFUg) lt10 SP-VG M005 B

Protein () tbd BCA protein assay

928

Figure S 1 - a) HPLC-ELSD chromatogram and composition table 929

930

931

932

933

3

b) 934

935

Group Ndeg δ ppm

CH2 1 354

CH2 2 153

CH2 3-712-14 132

CH2 811 203

CH 910 534

CH2 15 162

CH2 16 388

CH 1 424

CH 2 316

CH 3 334

CH 4 329

CH 5 354

CH2 6 366

Figure S 1 (next) ndash b) 1H NMR spectrum (MeOD-d4) recorded on the purified G-C181-OH compounds 936

and related assignment 937

938

2

811910 1rsquo

5rsquo1

3rsquo4rsquo

1rsquo1910

asymp asymp asympHDO MeOD Acetone

6rsquo2rsquo

1615

3-712-14

4

939

940

Figure S 2 ndash Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 130degC 941

5

942

943

Figure S 3 - Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 90degC 944

945

90degC

1

22

6

946

947

948

949

Figure S 4 ndash Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 70degC 950

951

21

11

1

2

5

3

4

45

5

6

7

7

7

952

953

954

Figure S 5 ndash Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 25degC 955

956

957

1

1 1

8

958

Figure S 6 ndash a) SANS experiment recorded at 90degC for a 5 mgmL G-C181-OH aqueous solution b-c) 959

WAXS experiments recorded on aqueous solutions of G-C181-OH at b) 5 mgmL and c) 25 mgmL d) 960

Structural interpretation of the repeating distances pointed out in c) 961

962

963

964

001 01 1

q Aring-1

25degC

68degC

88degC

111degC

01 1q Aring

-1

02 04 06 08 1q Aring

-1

001 01

10-1

100

101

Inte

nsity

cm

-1

q Aring-1

-2

T= 90degC

SANS WAXS WAXS

a) b) c)

dv

dL d= 144 Aring-1

di

ddH2O

dt

dvdL

dtdi d

d)

9

965

966

Figure S 7 ndash ssNMR MAS (10 kHz) experiments recorded on a G-C181-OH dry powder a) 1H spectra 967

recorded from 110degC to 70degC b) 13C single pulse (SP) spectra recorded with high power 1H decoupling from 968

110degC to 70degC c) 13C cross polarization (CP) spectra recorded at 110degC and contact time tc of 50 micros and 969

3000 micros d) 13C CP spectra recorded at 70degC and for 20 lt tc micros lt 3000 970

971

1

811

910

CH2 CH2

1rsquo

6rsquo

8 6 4 2 0 ppm

110

90

80

70

910

1rsquo

6rsquo16

1

2rsquo3rsquo4rsquo

811

OH

45 40 35 30 25 20 15

ppm

200 180 160 140 120 100 80 60 40 20 0

ppm140 120 100 80

ppm

CH2-gauche

3004

110

90

80

70

65

60

50

25

C91013003

C9 - 12869

CH2-all-trans

3198

C1rsquo10335

a) b)

1H

T degCT degC

13C Single Pulse MAS

A

B

A

B

70degC70degC

5rsquo

215

200 180 160 140 120 100 80 60 40 20 0

ppm

13C CPMAST= 110degC

200 180 160 140 120 100 80 60 40 20 0

ppm

132 131 130 129 128 127 126

ppm

106 105 104 103 102

ppm

13C CPMAST= 70degC

50

tc micros

3000

20

tc micros

50 500 1000 3000

tc micros

C102

13024 C910113003

C92

12869

C1rsquo213082

C1rsquo110443

A B

106 105 104 103 102

ppm

3000

1000

500

50

20

AB

c)

d)

10

972

Figure S 8 ndash a) Short and d) long T2 components displayed by the 1H transverse relaxation signal c) 1H 973

transverse relaxation signals recorded on a G-C181-OH dry powder from 110degC to 70degC The area of each 974

given resonance is plotted (log-lin) as a function of the echo time (CPMG pulse sequence) for each 975

temperature and fitted using the equation given above T2s and T2l indicate the short and long T2 relaxation 976

times respectively obtained by describing the 1H transverse relaxation signal using a biexponential decay 977

For H1rsquo H5rsquo and H6rsquo the 1H relaxation function could be described by a monoexponential decay 978

- Protons from the sugar group (H1rsquo H5rsquo H6rsquo) show a single T2 relaxation component (Figure 979

6c and Figure S 8) while protons of the acyl chain (CH2 H910 H811) show both short (Figure 980

6c) and long (Figure 6d) T2 relaxation component 981

- The short T2 relaxation time systematically decreases from 110degC to 70degC for all species by a 982

factor ranging from two to seven depending on the specific chemical group Very long T2 are 983

observed at 110degC for all acyl species while nearly unchanged values are measured below 984

90degC without any clear trend 985

- The sugar protons (H1rsquo H5rsquo H6rsquo) have the shortest set of T2 values (lt 1 ms) below 100degC 986

The long component of T2 generally ranges between 4 ms and 12 ms 987

0 1 2 3 4 5 6 7 8

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1 2 3 4 5 6 7 8

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

H1

H8H11

H9H10

CH2

H1rsquo

H6rsquo

CH2

H910H811 H1CH2

H1rsquo H5rsquo6rsquo

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

OH

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45 T= 110degC

T= 90degC

T= 80degC

T= 70degC

Log

In

ten

sity

Time ms

Fit

)

CH2 H910 H811 H1 H56 H1 OH

00

05

10

15

20

25

30

35

40

T2 m

s

T= 110degC

T= 90degC

T= 80degC

T= 70degC

CH2 H910 H811 H1

0

2

4

6

8

10

12

T2 m

s

T= 110degC

T= 90degC

T= 80degC

T= 70degC

Short T2(1H) Long T2(1H)

a) b)

c)

11

Table S 1 ndash 1H chemical shifts of G-C181-OH in solution and in the solid-state Solid-state spectra are 988

recorded at a MAS frequency of 10 kHz in DMSO-d6 broad peak 989

990

δH ppm

Group Ndeg Liquid

(MeOD-d4) Solid

(T= 110degC) Solid

(T= 70degC)

CH2 H1 354 357

CH2 H2 153 156

CH2 H3-712-14 132 135

CH2 H811 203 203

CH H910 534 534

CH2 H15 162 164

CH2 H16 388 383

CH H1 424 432

CH H2 316 335

CH H3 334 335

CH H4 329 335

CH H5 354 351 357

CH2 H6 366 383

OH OH1 432

45 463

OH OH2 492

OH OH3 490

OH OH4 487

OH OH6 445

991

992

H1H9H10H1rsquo

H6rsquo

OH2rsquo

OH1

12

Detailed discussion about the NOESY and DQ-SQ 2D 1H-1H ssNMR experiments presented in 993

Figure 9 994

995

2D 1H-1H ssNMR experiments are a unique tool to explore through-space proximities 996

between 1H nuclei ssNMR offers a large number of possibilities each adapted to a given 997

system The advantage of 2D maps with respect to a dynamics approach is their direct 998

readability in clear two nuclei which exchange their magnetization through-space (whichever 999

mechanism may be responsible for this exchange) will display a specific cross-peak However 1000

the way cross-peaks are presented and interpreted strictly depends on the nature of the pulse 1001

program which in turn is chosen according to the physical nature of the sample 1002

1D temperature-dependent 1H NMR experiments performed on G-C181-OH (Figure 1003

4a in the main text) indicate sharp peaks at 90degC and broad peaks at 70degC Narrow peaks 1004

indicate fast andor large-amplitude molecular tumbling averaging the strong 1H-1H dipolar 1005

interactions This excludes the use of pulse sequences like DQ-SQ which is not efficient under 1006

these conditions For this reason we employ spin-diffusion-based pulse sequences like 1007

NOESY which is more adapted for samples with high molecular mobility When mixing times 1008

are chosen to be short (lt 1 s) one generally explores through-space proximities within about 5 1009

10 Aring On the contrary when molecular motions are restricted dipolar coupling becomes 1010

significant 1H NMR peaks are broader and pulse sequences like DQ-SQ can now be used 1011

with a spatial resolution of less than 5 Aring 1012

Whenever possible DQ-SQ experiments are preferred over spin diffusion-based pulse 1013

sequences because they are more selective in particular for likewise interactions which are 1014

not discriminated in NOESY-type experiments This is illustrated by the typical interaction 1015

schemes for NOESY (top Figure 9a) and DQ-SQ (top Figure 9b) For two given A and B 1016

protons 2D NOESY maps will display two on-diagonal and two square-shaped symmetrical 1017

off-diagonal cross-peaks Off-diagonal cross-peaks illustrate the mutual A-B and B-A 1018

interaction but the on-diagonal peaks are systematic and do not prove any A-A or B-B 1019

interaction Similarly to NOESY 2D DQ-SQ maps display off-diagonal peaks (symmetric 1020

across the diagonal top Figure 9b) upon A-B and B-A interaction but in contrast they will 1021

display on-diagonal peaks only in the presence of self-interaction In the tutorial example of 1022

Figure 9b top the on-diagonal A-A peak indicates mutual A-A interaction while lack of on-1023

diagonal B-B peak evidences the absence of B-B proximities Combining NOESY and DQ-SQ 1024

experiments is capital to study the local arrangement of G-C181-OH and its evolution with 1025

temperature 1026

13

At T= 90degC the H1 and H5rsquo contributions are very close but possible to discriminate 1027

them (δ(H1)= 354 ppm δ(H5rsquo)= 351 ppm) according to 1H NMR chemical shifts in Table S 1028

1 2D 1H-1H NOESY in Figure 9a also discriminates between the two components attributed 1029

on the basis of the interactions with their mutual neighbors (green H1-H2 and H5rsquo-H1rsquo squares 1030

in Figure 9a) On this basis 2D NOESY does not show a direct H1rsquo-H1 correlation (blue red-1031

crossed) but rather an extensive network of intra-sugar interactions These features exclude 1032

suggest intra- and inter-sugar interactions and they exclude head-tail arrangements as a major 1033

molecular configuration at 90degC 1034

At 70degC the 1H NMR peaks are broader indicating a reduction of the T2 (1H) relaxation 1035

times andor residual homonuclear dipolar interactions excluding the use of NOESY In this 1036

case 1H-1H DQ-SQ correlation experiments are best suited Figure 9b presents the 1H-1H DQ-1037

SQ experiment on G-C181-OH at 70degC Peaks are broader than in NOESY and attribution may 1038

be cumbersome However the experiment is performed at 700 MHz and MAS= 22 kHz which 1039

improves spectral resolution with respect to the same experiment performed at 300 MHz and 1040

MAS= 10 kHz The auto-peak at δH= 356 ppm in the SQ dimension is attributed to H1-H1 spin 1041

pairs mainly resulting from intra CH2 interaction while the DQ cross-peaks at δH= 156-356 1042

ppm and δH= 432-356 ppm respectively origin from H2-H1 and H1rsquo-H1 interactions (Figure 1043

9b) DQ-SQ map does not show any auto-peak corresponding to H1rsquo-H1rsquo or Hxrsquo-Hxrsquo (x= 2rsquo 1044

3rsquo 4rsquo) interactions (red-crossed blue Figure 9b) thus excluding any proximity (lt 5 Aring) between 1045

close glucose neighbors On the contrary DQ-SQ maps (recorded both at 700 MHz Figure 9b 1046

or at 300 MHz not shown) systematically show DQ cross-peaks which can only be attributed 1047

to H1rsquo-H1 interactions (δH= 432-356) The presence of these H1rsquo-H1 interactions can only be 1048

explained by a head-tail arrangement of the G-C181-OH molecules as illustrated below the 1049

DQ-SQ map in Figure 9b 1050

34

84 S Svenson A Schaefer and J H Fuhrhop J Chem Soc Perkin Trans 2 1994 2 895

1023ndash1028 896

85 K Bock and C Pedersen Adv Carbohydr Chem Biochem 1983 41 27ndash66 897

86 T B Grindley Structure and Conformation of Carbohydrates Springer Berlin 898

Heidelberg 2008 899

87 F Tantakitti J Boekhoven X Wang R V Kazantsev T Yu J Li E Zhuang R 900

Zandi J H Ortony C J Newcomb L C Palmer G S Shekhawat M O De La 901

Cruz G C Schatz and S I Stupp Nat Mater 2016 15 469ndash476 902

903

1

Supporting Information 904

905

Topological connection between vesicles and nanotubes in single-906

component lipid membranes driven by head-tail interactions 907

908

Niki Baccilea Ceacutedric Lorthioira Abdoul Aziz Baa Patrick Le Griela Cristina Coelhob Javier 909

Perezc Wim Soetaertd Sophie L K W Roelantsd 910

911

a Sorbonne Universiteacute Centre National de la Recherche Scientifique Laboratoire de Chimie de 912

la Matiegravere Condenseacutee de Paris LCMCP F-75005 Paris France 913

b Institut des Mateacuteriaux de Paris Centre Sorbonne Universiteacute Paris France 914

c Synchrotron Soleil LrsquoOrme des Merisiers Saint-Aubin BP48 91192 Gif-sur-Yvette Cedex 915

France 916

d InBio Department of Biotechnology Ghent University Ghent Belgium 917

918

919

Corresponding author 920

Dr Niki Baccile 921

E-mail address nikibaccilesorbonne-universitefr 922

Phone +33 1 44 27 56 77 923

924

925

926

2

a) 927

Parameter Method

Dry Matter (DM ) 996 Infrared balance 105degC

Glucose () 012 HPLC-Metacarb

Glycerol () nd HPLC-Metacarb

Free Fatty Acid content () lt001 Internal method BBEPP

GC with FID Detector

Oil () lt001 Internal method BBEPP

GC with FID Detector

Moulds (CFUg) lt10 3M-Nordval ndeg16 B

Yeasts (CFUg) lt10 3M-Nordval ndeg16 B

Anaerobic count (CFUg) lt10 SP-VG M005 B

Protein () tbd BCA protein assay

928

Figure S 1 - a) HPLC-ELSD chromatogram and composition table 929

930

931

932

933

3

b) 934

935

Group Ndeg δ ppm

CH2 1 354

CH2 2 153

CH2 3-712-14 132

CH2 811 203

CH 910 534

CH2 15 162

CH2 16 388

CH 1 424

CH 2 316

CH 3 334

CH 4 329

CH 5 354

CH2 6 366

Figure S 1 (next) ndash b) 1H NMR spectrum (MeOD-d4) recorded on the purified G-C181-OH compounds 936

and related assignment 937

938

2

811910 1rsquo

5rsquo1

3rsquo4rsquo

1rsquo1910

asymp asymp asympHDO MeOD Acetone

6rsquo2rsquo

1615

3-712-14

4

939

940

Figure S 2 ndash Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 130degC 941

5

942

943

Figure S 3 - Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 90degC 944

945

90degC

1

22

6

946

947

948

949

Figure S 4 ndash Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 70degC 950

951

21

11

1

2

5

3

4

45

5

6

7

7

7

952

953

954

Figure S 5 ndash Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 25degC 955

956

957

1

1 1

8

958

Figure S 6 ndash a) SANS experiment recorded at 90degC for a 5 mgmL G-C181-OH aqueous solution b-c) 959

WAXS experiments recorded on aqueous solutions of G-C181-OH at b) 5 mgmL and c) 25 mgmL d) 960

Structural interpretation of the repeating distances pointed out in c) 961

962

963

964

001 01 1

q Aring-1

25degC

68degC

88degC

111degC

01 1q Aring

-1

02 04 06 08 1q Aring

-1

001 01

10-1

100

101

Inte

nsity

cm

-1

q Aring-1

-2

T= 90degC

SANS WAXS WAXS

a) b) c)

dv

dL d= 144 Aring-1

di

ddH2O

dt

dvdL

dtdi d

d)

9

965

966

Figure S 7 ndash ssNMR MAS (10 kHz) experiments recorded on a G-C181-OH dry powder a) 1H spectra 967

recorded from 110degC to 70degC b) 13C single pulse (SP) spectra recorded with high power 1H decoupling from 968

110degC to 70degC c) 13C cross polarization (CP) spectra recorded at 110degC and contact time tc of 50 micros and 969

3000 micros d) 13C CP spectra recorded at 70degC and for 20 lt tc micros lt 3000 970

971

1

811

910

CH2 CH2

1rsquo

6rsquo

8 6 4 2 0 ppm

110

90

80

70

910

1rsquo

6rsquo16

1

2rsquo3rsquo4rsquo

811

OH

45 40 35 30 25 20 15

ppm

200 180 160 140 120 100 80 60 40 20 0

ppm140 120 100 80

ppm

CH2-gauche

3004

110

90

80

70

65

60

50

25

C91013003

C9 - 12869

CH2-all-trans

3198

C1rsquo10335

a) b)

1H

T degCT degC

13C Single Pulse MAS

A

B

A

B

70degC70degC

5rsquo

215

200 180 160 140 120 100 80 60 40 20 0

ppm

13C CPMAST= 110degC

200 180 160 140 120 100 80 60 40 20 0

ppm

132 131 130 129 128 127 126

ppm

106 105 104 103 102

ppm

13C CPMAST= 70degC

50

tc micros

3000

20

tc micros

50 500 1000 3000

tc micros

C102

13024 C910113003

C92

12869

C1rsquo213082

C1rsquo110443

A B

106 105 104 103 102

ppm

3000

1000

500

50

20

AB

c)

d)

10

972

Figure S 8 ndash a) Short and d) long T2 components displayed by the 1H transverse relaxation signal c) 1H 973

transverse relaxation signals recorded on a G-C181-OH dry powder from 110degC to 70degC The area of each 974

given resonance is plotted (log-lin) as a function of the echo time (CPMG pulse sequence) for each 975

temperature and fitted using the equation given above T2s and T2l indicate the short and long T2 relaxation 976

times respectively obtained by describing the 1H transverse relaxation signal using a biexponential decay 977

For H1rsquo H5rsquo and H6rsquo the 1H relaxation function could be described by a monoexponential decay 978

- Protons from the sugar group (H1rsquo H5rsquo H6rsquo) show a single T2 relaxation component (Figure 979

6c and Figure S 8) while protons of the acyl chain (CH2 H910 H811) show both short (Figure 980

6c) and long (Figure 6d) T2 relaxation component 981

- The short T2 relaxation time systematically decreases from 110degC to 70degC for all species by a 982

factor ranging from two to seven depending on the specific chemical group Very long T2 are 983

observed at 110degC for all acyl species while nearly unchanged values are measured below 984

90degC without any clear trend 985

- The sugar protons (H1rsquo H5rsquo H6rsquo) have the shortest set of T2 values (lt 1 ms) below 100degC 986

The long component of T2 generally ranges between 4 ms and 12 ms 987

0 1 2 3 4 5 6 7 8

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1 2 3 4 5 6 7 8

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

H1

H8H11

H9H10

CH2

H1rsquo

H6rsquo

CH2

H910H811 H1CH2

H1rsquo H5rsquo6rsquo

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

OH

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45 T= 110degC

T= 90degC

T= 80degC

T= 70degC

Log

In

ten

sity

Time ms

Fit

)

CH2 H910 H811 H1 H56 H1 OH

00

05

10

15

20

25

30

35

40

T2 m

s

T= 110degC

T= 90degC

T= 80degC

T= 70degC

CH2 H910 H811 H1

0

2

4

6

8

10

12

T2 m

s

T= 110degC

T= 90degC

T= 80degC

T= 70degC

Short T2(1H) Long T2(1H)

a) b)

c)

11

Table S 1 ndash 1H chemical shifts of G-C181-OH in solution and in the solid-state Solid-state spectra are 988

recorded at a MAS frequency of 10 kHz in DMSO-d6 broad peak 989

990

δH ppm

Group Ndeg Liquid

(MeOD-d4) Solid

(T= 110degC) Solid

(T= 70degC)

CH2 H1 354 357

CH2 H2 153 156

CH2 H3-712-14 132 135

CH2 H811 203 203

CH H910 534 534

CH2 H15 162 164

CH2 H16 388 383

CH H1 424 432

CH H2 316 335

CH H3 334 335

CH H4 329 335

CH H5 354 351 357

CH2 H6 366 383

OH OH1 432

45 463

OH OH2 492

OH OH3 490

OH OH4 487

OH OH6 445

991

992

H1H9H10H1rsquo

H6rsquo

OH2rsquo

OH1

12

Detailed discussion about the NOESY and DQ-SQ 2D 1H-1H ssNMR experiments presented in 993

Figure 9 994

995

2D 1H-1H ssNMR experiments are a unique tool to explore through-space proximities 996

between 1H nuclei ssNMR offers a large number of possibilities each adapted to a given 997

system The advantage of 2D maps with respect to a dynamics approach is their direct 998

readability in clear two nuclei which exchange their magnetization through-space (whichever 999

mechanism may be responsible for this exchange) will display a specific cross-peak However 1000

the way cross-peaks are presented and interpreted strictly depends on the nature of the pulse 1001

program which in turn is chosen according to the physical nature of the sample 1002

1D temperature-dependent 1H NMR experiments performed on G-C181-OH (Figure 1003

4a in the main text) indicate sharp peaks at 90degC and broad peaks at 70degC Narrow peaks 1004

indicate fast andor large-amplitude molecular tumbling averaging the strong 1H-1H dipolar 1005

interactions This excludes the use of pulse sequences like DQ-SQ which is not efficient under 1006

these conditions For this reason we employ spin-diffusion-based pulse sequences like 1007

NOESY which is more adapted for samples with high molecular mobility When mixing times 1008

are chosen to be short (lt 1 s) one generally explores through-space proximities within about 5 1009

10 Aring On the contrary when molecular motions are restricted dipolar coupling becomes 1010

significant 1H NMR peaks are broader and pulse sequences like DQ-SQ can now be used 1011

with a spatial resolution of less than 5 Aring 1012

Whenever possible DQ-SQ experiments are preferred over spin diffusion-based pulse 1013

sequences because they are more selective in particular for likewise interactions which are 1014

not discriminated in NOESY-type experiments This is illustrated by the typical interaction 1015

schemes for NOESY (top Figure 9a) and DQ-SQ (top Figure 9b) For two given A and B 1016

protons 2D NOESY maps will display two on-diagonal and two square-shaped symmetrical 1017

off-diagonal cross-peaks Off-diagonal cross-peaks illustrate the mutual A-B and B-A 1018

interaction but the on-diagonal peaks are systematic and do not prove any A-A or B-B 1019

interaction Similarly to NOESY 2D DQ-SQ maps display off-diagonal peaks (symmetric 1020

across the diagonal top Figure 9b) upon A-B and B-A interaction but in contrast they will 1021

display on-diagonal peaks only in the presence of self-interaction In the tutorial example of 1022

Figure 9b top the on-diagonal A-A peak indicates mutual A-A interaction while lack of on-1023

diagonal B-B peak evidences the absence of B-B proximities Combining NOESY and DQ-SQ 1024

experiments is capital to study the local arrangement of G-C181-OH and its evolution with 1025

temperature 1026

13

At T= 90degC the H1 and H5rsquo contributions are very close but possible to discriminate 1027

them (δ(H1)= 354 ppm δ(H5rsquo)= 351 ppm) according to 1H NMR chemical shifts in Table S 1028

1 2D 1H-1H NOESY in Figure 9a also discriminates between the two components attributed 1029

on the basis of the interactions with their mutual neighbors (green H1-H2 and H5rsquo-H1rsquo squares 1030

in Figure 9a) On this basis 2D NOESY does not show a direct H1rsquo-H1 correlation (blue red-1031

crossed) but rather an extensive network of intra-sugar interactions These features exclude 1032

suggest intra- and inter-sugar interactions and they exclude head-tail arrangements as a major 1033

molecular configuration at 90degC 1034

At 70degC the 1H NMR peaks are broader indicating a reduction of the T2 (1H) relaxation 1035

times andor residual homonuclear dipolar interactions excluding the use of NOESY In this 1036

case 1H-1H DQ-SQ correlation experiments are best suited Figure 9b presents the 1H-1H DQ-1037

SQ experiment on G-C181-OH at 70degC Peaks are broader than in NOESY and attribution may 1038

be cumbersome However the experiment is performed at 700 MHz and MAS= 22 kHz which 1039

improves spectral resolution with respect to the same experiment performed at 300 MHz and 1040

MAS= 10 kHz The auto-peak at δH= 356 ppm in the SQ dimension is attributed to H1-H1 spin 1041

pairs mainly resulting from intra CH2 interaction while the DQ cross-peaks at δH= 156-356 1042

ppm and δH= 432-356 ppm respectively origin from H2-H1 and H1rsquo-H1 interactions (Figure 1043

9b) DQ-SQ map does not show any auto-peak corresponding to H1rsquo-H1rsquo or Hxrsquo-Hxrsquo (x= 2rsquo 1044

3rsquo 4rsquo) interactions (red-crossed blue Figure 9b) thus excluding any proximity (lt 5 Aring) between 1045

close glucose neighbors On the contrary DQ-SQ maps (recorded both at 700 MHz Figure 9b 1046

or at 300 MHz not shown) systematically show DQ cross-peaks which can only be attributed 1047

to H1rsquo-H1 interactions (δH= 432-356) The presence of these H1rsquo-H1 interactions can only be 1048

explained by a head-tail arrangement of the G-C181-OH molecules as illustrated below the 1049

DQ-SQ map in Figure 9b 1050

1

Supporting Information 904

905

Topological connection between vesicles and nanotubes in single-906

component lipid membranes driven by head-tail interactions 907

908

Niki Baccilea Ceacutedric Lorthioira Abdoul Aziz Baa Patrick Le Griela Cristina Coelhob Javier 909

Perezc Wim Soetaertd Sophie L K W Roelantsd 910

911

a Sorbonne Universiteacute Centre National de la Recherche Scientifique Laboratoire de Chimie de 912

la Matiegravere Condenseacutee de Paris LCMCP F-75005 Paris France 913

b Institut des Mateacuteriaux de Paris Centre Sorbonne Universiteacute Paris France 914

c Synchrotron Soleil LrsquoOrme des Merisiers Saint-Aubin BP48 91192 Gif-sur-Yvette Cedex 915

France 916

d InBio Department of Biotechnology Ghent University Ghent Belgium 917

918

919

Corresponding author 920

Dr Niki Baccile 921

E-mail address nikibaccilesorbonne-universitefr 922

Phone +33 1 44 27 56 77 923

924

925

926

2

a) 927

Parameter Method

Dry Matter (DM ) 996 Infrared balance 105degC

Glucose () 012 HPLC-Metacarb

Glycerol () nd HPLC-Metacarb

Free Fatty Acid content () lt001 Internal method BBEPP

GC with FID Detector

Oil () lt001 Internal method BBEPP

GC with FID Detector

Moulds (CFUg) lt10 3M-Nordval ndeg16 B

Yeasts (CFUg) lt10 3M-Nordval ndeg16 B

Anaerobic count (CFUg) lt10 SP-VG M005 B

Protein () tbd BCA protein assay

928

Figure S 1 - a) HPLC-ELSD chromatogram and composition table 929

930

931

932

933

3

b) 934

935

Group Ndeg δ ppm

CH2 1 354

CH2 2 153

CH2 3-712-14 132

CH2 811 203

CH 910 534

CH2 15 162

CH2 16 388

CH 1 424

CH 2 316

CH 3 334

CH 4 329

CH 5 354

CH2 6 366

Figure S 1 (next) ndash b) 1H NMR spectrum (MeOD-d4) recorded on the purified G-C181-OH compounds 936

and related assignment 937

938

2

811910 1rsquo

5rsquo1

3rsquo4rsquo

1rsquo1910

asymp asymp asympHDO MeOD Acetone

6rsquo2rsquo

1615

3-712-14

4

939

940

Figure S 2 ndash Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 130degC 941

5

942

943

Figure S 3 - Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 90degC 944

945

90degC

1

22

6

946

947

948

949

Figure S 4 ndash Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 70degC 950

951

21

11

1

2

5

3

4

45

5

6

7

7

7

952

953

954

Figure S 5 ndash Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 25degC 955

956

957

1

1 1

8

958

Figure S 6 ndash a) SANS experiment recorded at 90degC for a 5 mgmL G-C181-OH aqueous solution b-c) 959

WAXS experiments recorded on aqueous solutions of G-C181-OH at b) 5 mgmL and c) 25 mgmL d) 960

Structural interpretation of the repeating distances pointed out in c) 961

962

963

964

001 01 1

q Aring-1

25degC

68degC

88degC

111degC

01 1q Aring

-1

02 04 06 08 1q Aring

-1

001 01

10-1

100

101

Inte

nsity

cm

-1

q Aring-1

-2

T= 90degC

SANS WAXS WAXS

a) b) c)

dv

dL d= 144 Aring-1

di

ddH2O

dt

dvdL

dtdi d

d)

9

965

966

Figure S 7 ndash ssNMR MAS (10 kHz) experiments recorded on a G-C181-OH dry powder a) 1H spectra 967

recorded from 110degC to 70degC b) 13C single pulse (SP) spectra recorded with high power 1H decoupling from 968

110degC to 70degC c) 13C cross polarization (CP) spectra recorded at 110degC and contact time tc of 50 micros and 969

3000 micros d) 13C CP spectra recorded at 70degC and for 20 lt tc micros lt 3000 970

971

1

811

910

CH2 CH2

1rsquo

6rsquo

8 6 4 2 0 ppm

110

90

80

70

910

1rsquo

6rsquo16

1

2rsquo3rsquo4rsquo

811

OH

45 40 35 30 25 20 15

ppm

200 180 160 140 120 100 80 60 40 20 0

ppm140 120 100 80

ppm

CH2-gauche

3004

110

90

80

70

65

60

50

25

C91013003

C9 - 12869

CH2-all-trans

3198

C1rsquo10335

a) b)

1H

T degCT degC

13C Single Pulse MAS

A

B

A

B

70degC70degC

5rsquo

215

200 180 160 140 120 100 80 60 40 20 0

ppm

13C CPMAST= 110degC

200 180 160 140 120 100 80 60 40 20 0

ppm

132 131 130 129 128 127 126

ppm

106 105 104 103 102

ppm

13C CPMAST= 70degC

50

tc micros

3000

20

tc micros

50 500 1000 3000

tc micros

C102

13024 C910113003

C92

12869

C1rsquo213082

C1rsquo110443

A B

106 105 104 103 102

ppm

3000

1000

500

50

20

AB

c)

d)

10

972

Figure S 8 ndash a) Short and d) long T2 components displayed by the 1H transverse relaxation signal c) 1H 973

transverse relaxation signals recorded on a G-C181-OH dry powder from 110degC to 70degC The area of each 974

given resonance is plotted (log-lin) as a function of the echo time (CPMG pulse sequence) for each 975

temperature and fitted using the equation given above T2s and T2l indicate the short and long T2 relaxation 976

times respectively obtained by describing the 1H transverse relaxation signal using a biexponential decay 977

For H1rsquo H5rsquo and H6rsquo the 1H relaxation function could be described by a monoexponential decay 978

- Protons from the sugar group (H1rsquo H5rsquo H6rsquo) show a single T2 relaxation component (Figure 979

6c and Figure S 8) while protons of the acyl chain (CH2 H910 H811) show both short (Figure 980

6c) and long (Figure 6d) T2 relaxation component 981

- The short T2 relaxation time systematically decreases from 110degC to 70degC for all species by a 982

factor ranging from two to seven depending on the specific chemical group Very long T2 are 983

observed at 110degC for all acyl species while nearly unchanged values are measured below 984

90degC without any clear trend 985

- The sugar protons (H1rsquo H5rsquo H6rsquo) have the shortest set of T2 values (lt 1 ms) below 100degC 986

The long component of T2 generally ranges between 4 ms and 12 ms 987

0 1 2 3 4 5 6 7 8

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1 2 3 4 5 6 7 8

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

H1

H8H11

H9H10

CH2

H1rsquo

H6rsquo

CH2

H910H811 H1CH2

H1rsquo H5rsquo6rsquo

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

OH

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45 T= 110degC

T= 90degC

T= 80degC

T= 70degC

Log

In

ten

sity

Time ms

Fit

)

CH2 H910 H811 H1 H56 H1 OH

00

05

10

15

20

25

30

35

40

T2 m

s

T= 110degC

T= 90degC

T= 80degC

T= 70degC

CH2 H910 H811 H1

0

2

4

6

8

10

12

T2 m

s

T= 110degC

T= 90degC

T= 80degC

T= 70degC

Short T2(1H) Long T2(1H)

a) b)

c)

11

Table S 1 ndash 1H chemical shifts of G-C181-OH in solution and in the solid-state Solid-state spectra are 988

recorded at a MAS frequency of 10 kHz in DMSO-d6 broad peak 989

990

δH ppm

Group Ndeg Liquid

(MeOD-d4) Solid

(T= 110degC) Solid

(T= 70degC)

CH2 H1 354 357

CH2 H2 153 156

CH2 H3-712-14 132 135

CH2 H811 203 203

CH H910 534 534

CH2 H15 162 164

CH2 H16 388 383

CH H1 424 432

CH H2 316 335

CH H3 334 335

CH H4 329 335

CH H5 354 351 357

CH2 H6 366 383

OH OH1 432

45 463

OH OH2 492

OH OH3 490

OH OH4 487

OH OH6 445

991

992

H1H9H10H1rsquo

H6rsquo

OH2rsquo

OH1

12

Detailed discussion about the NOESY and DQ-SQ 2D 1H-1H ssNMR experiments presented in 993

Figure 9 994

995

2D 1H-1H ssNMR experiments are a unique tool to explore through-space proximities 996

between 1H nuclei ssNMR offers a large number of possibilities each adapted to a given 997

system The advantage of 2D maps with respect to a dynamics approach is their direct 998

readability in clear two nuclei which exchange their magnetization through-space (whichever 999

mechanism may be responsible for this exchange) will display a specific cross-peak However 1000

the way cross-peaks are presented and interpreted strictly depends on the nature of the pulse 1001

program which in turn is chosen according to the physical nature of the sample 1002

1D temperature-dependent 1H NMR experiments performed on G-C181-OH (Figure 1003

4a in the main text) indicate sharp peaks at 90degC and broad peaks at 70degC Narrow peaks 1004

indicate fast andor large-amplitude molecular tumbling averaging the strong 1H-1H dipolar 1005

interactions This excludes the use of pulse sequences like DQ-SQ which is not efficient under 1006

these conditions For this reason we employ spin-diffusion-based pulse sequences like 1007

NOESY which is more adapted for samples with high molecular mobility When mixing times 1008

are chosen to be short (lt 1 s) one generally explores through-space proximities within about 5 1009

10 Aring On the contrary when molecular motions are restricted dipolar coupling becomes 1010

significant 1H NMR peaks are broader and pulse sequences like DQ-SQ can now be used 1011

with a spatial resolution of less than 5 Aring 1012

Whenever possible DQ-SQ experiments are preferred over spin diffusion-based pulse 1013

sequences because they are more selective in particular for likewise interactions which are 1014

not discriminated in NOESY-type experiments This is illustrated by the typical interaction 1015

schemes for NOESY (top Figure 9a) and DQ-SQ (top Figure 9b) For two given A and B 1016

protons 2D NOESY maps will display two on-diagonal and two square-shaped symmetrical 1017

off-diagonal cross-peaks Off-diagonal cross-peaks illustrate the mutual A-B and B-A 1018

interaction but the on-diagonal peaks are systematic and do not prove any A-A or B-B 1019

interaction Similarly to NOESY 2D DQ-SQ maps display off-diagonal peaks (symmetric 1020

across the diagonal top Figure 9b) upon A-B and B-A interaction but in contrast they will 1021

display on-diagonal peaks only in the presence of self-interaction In the tutorial example of 1022

Figure 9b top the on-diagonal A-A peak indicates mutual A-A interaction while lack of on-1023

diagonal B-B peak evidences the absence of B-B proximities Combining NOESY and DQ-SQ 1024

experiments is capital to study the local arrangement of G-C181-OH and its evolution with 1025

temperature 1026

13

At T= 90degC the H1 and H5rsquo contributions are very close but possible to discriminate 1027

them (δ(H1)= 354 ppm δ(H5rsquo)= 351 ppm) according to 1H NMR chemical shifts in Table S 1028

1 2D 1H-1H NOESY in Figure 9a also discriminates between the two components attributed 1029

on the basis of the interactions with their mutual neighbors (green H1-H2 and H5rsquo-H1rsquo squares 1030

in Figure 9a) On this basis 2D NOESY does not show a direct H1rsquo-H1 correlation (blue red-1031

crossed) but rather an extensive network of intra-sugar interactions These features exclude 1032

suggest intra- and inter-sugar interactions and they exclude head-tail arrangements as a major 1033

molecular configuration at 90degC 1034

At 70degC the 1H NMR peaks are broader indicating a reduction of the T2 (1H) relaxation 1035

times andor residual homonuclear dipolar interactions excluding the use of NOESY In this 1036

case 1H-1H DQ-SQ correlation experiments are best suited Figure 9b presents the 1H-1H DQ-1037

SQ experiment on G-C181-OH at 70degC Peaks are broader than in NOESY and attribution may 1038

be cumbersome However the experiment is performed at 700 MHz and MAS= 22 kHz which 1039

improves spectral resolution with respect to the same experiment performed at 300 MHz and 1040

MAS= 10 kHz The auto-peak at δH= 356 ppm in the SQ dimension is attributed to H1-H1 spin 1041

pairs mainly resulting from intra CH2 interaction while the DQ cross-peaks at δH= 156-356 1042

ppm and δH= 432-356 ppm respectively origin from H2-H1 and H1rsquo-H1 interactions (Figure 1043

9b) DQ-SQ map does not show any auto-peak corresponding to H1rsquo-H1rsquo or Hxrsquo-Hxrsquo (x= 2rsquo 1044

3rsquo 4rsquo) interactions (red-crossed blue Figure 9b) thus excluding any proximity (lt 5 Aring) between 1045

close glucose neighbors On the contrary DQ-SQ maps (recorded both at 700 MHz Figure 9b 1046

or at 300 MHz not shown) systematically show DQ cross-peaks which can only be attributed 1047

to H1rsquo-H1 interactions (δH= 432-356) The presence of these H1rsquo-H1 interactions can only be 1048

explained by a head-tail arrangement of the G-C181-OH molecules as illustrated below the 1049

DQ-SQ map in Figure 9b 1050

2

a) 927

Parameter Method

Dry Matter (DM ) 996 Infrared balance 105degC

Glucose () 012 HPLC-Metacarb

Glycerol () nd HPLC-Metacarb

Free Fatty Acid content () lt001 Internal method BBEPP

GC with FID Detector

Oil () lt001 Internal method BBEPP

GC with FID Detector

Moulds (CFUg) lt10 3M-Nordval ndeg16 B

Yeasts (CFUg) lt10 3M-Nordval ndeg16 B

Anaerobic count (CFUg) lt10 SP-VG M005 B

Protein () tbd BCA protein assay

928

Figure S 1 - a) HPLC-ELSD chromatogram and composition table 929

930

931

932

933

3

b) 934

935

Group Ndeg δ ppm

CH2 1 354

CH2 2 153

CH2 3-712-14 132

CH2 811 203

CH 910 534

CH2 15 162

CH2 16 388

CH 1 424

CH 2 316

CH 3 334

CH 4 329

CH 5 354

CH2 6 366

Figure S 1 (next) ndash b) 1H NMR spectrum (MeOD-d4) recorded on the purified G-C181-OH compounds 936

and related assignment 937

938

2

811910 1rsquo

5rsquo1

3rsquo4rsquo

1rsquo1910

asymp asymp asympHDO MeOD Acetone

6rsquo2rsquo

1615

3-712-14

4

939

940

Figure S 2 ndash Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 130degC 941

5

942

943

Figure S 3 - Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 90degC 944

945

90degC

1

22

6

946

947

948

949

Figure S 4 ndash Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 70degC 950

951

21

11

1

2

5

3

4

45

5

6

7

7

7

952

953

954

Figure S 5 ndash Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 25degC 955

956

957

1

1 1

8

958

Figure S 6 ndash a) SANS experiment recorded at 90degC for a 5 mgmL G-C181-OH aqueous solution b-c) 959

WAXS experiments recorded on aqueous solutions of G-C181-OH at b) 5 mgmL and c) 25 mgmL d) 960

Structural interpretation of the repeating distances pointed out in c) 961

962

963

964

001 01 1

q Aring-1

25degC

68degC

88degC

111degC

01 1q Aring

-1

02 04 06 08 1q Aring

-1

001 01

10-1

100

101

Inte

nsity

cm

-1

q Aring-1

-2

T= 90degC

SANS WAXS WAXS

a) b) c)

dv

dL d= 144 Aring-1

di

ddH2O

dt

dvdL

dtdi d

d)

9

965

966

Figure S 7 ndash ssNMR MAS (10 kHz) experiments recorded on a G-C181-OH dry powder a) 1H spectra 967

recorded from 110degC to 70degC b) 13C single pulse (SP) spectra recorded with high power 1H decoupling from 968

110degC to 70degC c) 13C cross polarization (CP) spectra recorded at 110degC and contact time tc of 50 micros and 969

3000 micros d) 13C CP spectra recorded at 70degC and for 20 lt tc micros lt 3000 970

971

1

811

910

CH2 CH2

1rsquo

6rsquo

8 6 4 2 0 ppm

110

90

80

70

910

1rsquo

6rsquo16

1

2rsquo3rsquo4rsquo

811

OH

45 40 35 30 25 20 15

ppm

200 180 160 140 120 100 80 60 40 20 0

ppm140 120 100 80

ppm

CH2-gauche

3004

110

90

80

70

65

60

50

25

C91013003

C9 - 12869

CH2-all-trans

3198

C1rsquo10335

a) b)

1H

T degCT degC

13C Single Pulse MAS

A

B

A

B

70degC70degC

5rsquo

215

200 180 160 140 120 100 80 60 40 20 0

ppm

13C CPMAST= 110degC

200 180 160 140 120 100 80 60 40 20 0

ppm

132 131 130 129 128 127 126

ppm

106 105 104 103 102

ppm

13C CPMAST= 70degC

50

tc micros

3000

20

tc micros

50 500 1000 3000

tc micros

C102

13024 C910113003

C92

12869

C1rsquo213082

C1rsquo110443

A B

106 105 104 103 102

ppm

3000

1000

500

50

20

AB

c)

d)

10

972

Figure S 8 ndash a) Short and d) long T2 components displayed by the 1H transverse relaxation signal c) 1H 973

transverse relaxation signals recorded on a G-C181-OH dry powder from 110degC to 70degC The area of each 974

given resonance is plotted (log-lin) as a function of the echo time (CPMG pulse sequence) for each 975

temperature and fitted using the equation given above T2s and T2l indicate the short and long T2 relaxation 976

times respectively obtained by describing the 1H transverse relaxation signal using a biexponential decay 977

For H1rsquo H5rsquo and H6rsquo the 1H relaxation function could be described by a monoexponential decay 978

- Protons from the sugar group (H1rsquo H5rsquo H6rsquo) show a single T2 relaxation component (Figure 979

6c and Figure S 8) while protons of the acyl chain (CH2 H910 H811) show both short (Figure 980

6c) and long (Figure 6d) T2 relaxation component 981

- The short T2 relaxation time systematically decreases from 110degC to 70degC for all species by a 982

factor ranging from two to seven depending on the specific chemical group Very long T2 are 983

observed at 110degC for all acyl species while nearly unchanged values are measured below 984

90degC without any clear trend 985

- The sugar protons (H1rsquo H5rsquo H6rsquo) have the shortest set of T2 values (lt 1 ms) below 100degC 986

The long component of T2 generally ranges between 4 ms and 12 ms 987

0 1 2 3 4 5 6 7 8

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1 2 3 4 5 6 7 8

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

H1

H8H11

H9H10

CH2

H1rsquo

H6rsquo

CH2

H910H811 H1CH2

H1rsquo H5rsquo6rsquo

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

OH

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45 T= 110degC

T= 90degC

T= 80degC

T= 70degC

Log

In

ten

sity

Time ms

Fit

)

CH2 H910 H811 H1 H56 H1 OH

00

05

10

15

20

25

30

35

40

T2 m

s

T= 110degC

T= 90degC

T= 80degC

T= 70degC

CH2 H910 H811 H1

0

2

4

6

8

10

12

T2 m

s

T= 110degC

T= 90degC

T= 80degC

T= 70degC

Short T2(1H) Long T2(1H)

a) b)

c)

11

Table S 1 ndash 1H chemical shifts of G-C181-OH in solution and in the solid-state Solid-state spectra are 988

recorded at a MAS frequency of 10 kHz in DMSO-d6 broad peak 989

990

δH ppm

Group Ndeg Liquid

(MeOD-d4) Solid

(T= 110degC) Solid

(T= 70degC)

CH2 H1 354 357

CH2 H2 153 156

CH2 H3-712-14 132 135

CH2 H811 203 203

CH H910 534 534

CH2 H15 162 164

CH2 H16 388 383

CH H1 424 432

CH H2 316 335

CH H3 334 335

CH H4 329 335

CH H5 354 351 357

CH2 H6 366 383

OH OH1 432

45 463

OH OH2 492

OH OH3 490

OH OH4 487

OH OH6 445

991

992

H1H9H10H1rsquo

H6rsquo

OH2rsquo

OH1

12

Detailed discussion about the NOESY and DQ-SQ 2D 1H-1H ssNMR experiments presented in 993

Figure 9 994

995

2D 1H-1H ssNMR experiments are a unique tool to explore through-space proximities 996

between 1H nuclei ssNMR offers a large number of possibilities each adapted to a given 997

system The advantage of 2D maps with respect to a dynamics approach is their direct 998

readability in clear two nuclei which exchange their magnetization through-space (whichever 999

mechanism may be responsible for this exchange) will display a specific cross-peak However 1000

the way cross-peaks are presented and interpreted strictly depends on the nature of the pulse 1001

program which in turn is chosen according to the physical nature of the sample 1002

1D temperature-dependent 1H NMR experiments performed on G-C181-OH (Figure 1003

4a in the main text) indicate sharp peaks at 90degC and broad peaks at 70degC Narrow peaks 1004

indicate fast andor large-amplitude molecular tumbling averaging the strong 1H-1H dipolar 1005

interactions This excludes the use of pulse sequences like DQ-SQ which is not efficient under 1006

these conditions For this reason we employ spin-diffusion-based pulse sequences like 1007

NOESY which is more adapted for samples with high molecular mobility When mixing times 1008

are chosen to be short (lt 1 s) one generally explores through-space proximities within about 5 1009

10 Aring On the contrary when molecular motions are restricted dipolar coupling becomes 1010

significant 1H NMR peaks are broader and pulse sequences like DQ-SQ can now be used 1011

with a spatial resolution of less than 5 Aring 1012

Whenever possible DQ-SQ experiments are preferred over spin diffusion-based pulse 1013

sequences because they are more selective in particular for likewise interactions which are 1014

not discriminated in NOESY-type experiments This is illustrated by the typical interaction 1015

schemes for NOESY (top Figure 9a) and DQ-SQ (top Figure 9b) For two given A and B 1016

protons 2D NOESY maps will display two on-diagonal and two square-shaped symmetrical 1017

off-diagonal cross-peaks Off-diagonal cross-peaks illustrate the mutual A-B and B-A 1018

interaction but the on-diagonal peaks are systematic and do not prove any A-A or B-B 1019

interaction Similarly to NOESY 2D DQ-SQ maps display off-diagonal peaks (symmetric 1020

across the diagonal top Figure 9b) upon A-B and B-A interaction but in contrast they will 1021

display on-diagonal peaks only in the presence of self-interaction In the tutorial example of 1022

Figure 9b top the on-diagonal A-A peak indicates mutual A-A interaction while lack of on-1023

diagonal B-B peak evidences the absence of B-B proximities Combining NOESY and DQ-SQ 1024

experiments is capital to study the local arrangement of G-C181-OH and its evolution with 1025

temperature 1026

13

At T= 90degC the H1 and H5rsquo contributions are very close but possible to discriminate 1027

them (δ(H1)= 354 ppm δ(H5rsquo)= 351 ppm) according to 1H NMR chemical shifts in Table S 1028

1 2D 1H-1H NOESY in Figure 9a also discriminates between the two components attributed 1029

on the basis of the interactions with their mutual neighbors (green H1-H2 and H5rsquo-H1rsquo squares 1030

in Figure 9a) On this basis 2D NOESY does not show a direct H1rsquo-H1 correlation (blue red-1031

crossed) but rather an extensive network of intra-sugar interactions These features exclude 1032

suggest intra- and inter-sugar interactions and they exclude head-tail arrangements as a major 1033

molecular configuration at 90degC 1034

At 70degC the 1H NMR peaks are broader indicating a reduction of the T2 (1H) relaxation 1035

times andor residual homonuclear dipolar interactions excluding the use of NOESY In this 1036

case 1H-1H DQ-SQ correlation experiments are best suited Figure 9b presents the 1H-1H DQ-1037

SQ experiment on G-C181-OH at 70degC Peaks are broader than in NOESY and attribution may 1038

be cumbersome However the experiment is performed at 700 MHz and MAS= 22 kHz which 1039

improves spectral resolution with respect to the same experiment performed at 300 MHz and 1040

MAS= 10 kHz The auto-peak at δH= 356 ppm in the SQ dimension is attributed to H1-H1 spin 1041

pairs mainly resulting from intra CH2 interaction while the DQ cross-peaks at δH= 156-356 1042

ppm and δH= 432-356 ppm respectively origin from H2-H1 and H1rsquo-H1 interactions (Figure 1043

9b) DQ-SQ map does not show any auto-peak corresponding to H1rsquo-H1rsquo or Hxrsquo-Hxrsquo (x= 2rsquo 1044

3rsquo 4rsquo) interactions (red-crossed blue Figure 9b) thus excluding any proximity (lt 5 Aring) between 1045

close glucose neighbors On the contrary DQ-SQ maps (recorded both at 700 MHz Figure 9b 1046

or at 300 MHz not shown) systematically show DQ cross-peaks which can only be attributed 1047

to H1rsquo-H1 interactions (δH= 432-356) The presence of these H1rsquo-H1 interactions can only be 1048

explained by a head-tail arrangement of the G-C181-OH molecules as illustrated below the 1049

DQ-SQ map in Figure 9b 1050

3

b) 934

935

Group Ndeg δ ppm

CH2 1 354

CH2 2 153

CH2 3-712-14 132

CH2 811 203

CH 910 534

CH2 15 162

CH2 16 388

CH 1 424

CH 2 316

CH 3 334

CH 4 329

CH 5 354

CH2 6 366

Figure S 1 (next) ndash b) 1H NMR spectrum (MeOD-d4) recorded on the purified G-C181-OH compounds 936

and related assignment 937

938

2

811910 1rsquo

5rsquo1

3rsquo4rsquo

1rsquo1910

asymp asymp asympHDO MeOD Acetone

6rsquo2rsquo

1615

3-712-14

4

939

940

Figure S 2 ndash Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 130degC 941

5

942

943

Figure S 3 - Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 90degC 944

945

90degC

1

22

6

946

947

948

949

Figure S 4 ndash Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 70degC 950

951

21

11

1

2

5

3

4

45

5

6

7

7

7

952

953

954

Figure S 5 ndash Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 25degC 955

956

957

1

1 1

8

958

Figure S 6 ndash a) SANS experiment recorded at 90degC for a 5 mgmL G-C181-OH aqueous solution b-c) 959

WAXS experiments recorded on aqueous solutions of G-C181-OH at b) 5 mgmL and c) 25 mgmL d) 960

Structural interpretation of the repeating distances pointed out in c) 961

962

963

964

001 01 1

q Aring-1

25degC

68degC

88degC

111degC

01 1q Aring

-1

02 04 06 08 1q Aring

-1

001 01

10-1

100

101

Inte

nsity

cm

-1

q Aring-1

-2

T= 90degC

SANS WAXS WAXS

a) b) c)

dv

dL d= 144 Aring-1

di

ddH2O

dt

dvdL

dtdi d

d)

9

965

966

Figure S 7 ndash ssNMR MAS (10 kHz) experiments recorded on a G-C181-OH dry powder a) 1H spectra 967

recorded from 110degC to 70degC b) 13C single pulse (SP) spectra recorded with high power 1H decoupling from 968

110degC to 70degC c) 13C cross polarization (CP) spectra recorded at 110degC and contact time tc of 50 micros and 969

3000 micros d) 13C CP spectra recorded at 70degC and for 20 lt tc micros lt 3000 970

971

1

811

910

CH2 CH2

1rsquo

6rsquo

8 6 4 2 0 ppm

110

90

80

70

910

1rsquo

6rsquo16

1

2rsquo3rsquo4rsquo

811

OH

45 40 35 30 25 20 15

ppm

200 180 160 140 120 100 80 60 40 20 0

ppm140 120 100 80

ppm

CH2-gauche

3004

110

90

80

70

65

60

50

25

C91013003

C9 - 12869

CH2-all-trans

3198

C1rsquo10335

a) b)

1H

T degCT degC

13C Single Pulse MAS

A

B

A

B

70degC70degC

5rsquo

215

200 180 160 140 120 100 80 60 40 20 0

ppm

13C CPMAST= 110degC

200 180 160 140 120 100 80 60 40 20 0

ppm

132 131 130 129 128 127 126

ppm

106 105 104 103 102

ppm

13C CPMAST= 70degC

50

tc micros

3000

20

tc micros

50 500 1000 3000

tc micros

C102

13024 C910113003

C92

12869

C1rsquo213082

C1rsquo110443

A B

106 105 104 103 102

ppm

3000

1000

500

50

20

AB

c)

d)

10

972

Figure S 8 ndash a) Short and d) long T2 components displayed by the 1H transverse relaxation signal c) 1H 973

transverse relaxation signals recorded on a G-C181-OH dry powder from 110degC to 70degC The area of each 974

given resonance is plotted (log-lin) as a function of the echo time (CPMG pulse sequence) for each 975

temperature and fitted using the equation given above T2s and T2l indicate the short and long T2 relaxation 976

times respectively obtained by describing the 1H transverse relaxation signal using a biexponential decay 977

For H1rsquo H5rsquo and H6rsquo the 1H relaxation function could be described by a monoexponential decay 978

- Protons from the sugar group (H1rsquo H5rsquo H6rsquo) show a single T2 relaxation component (Figure 979

6c and Figure S 8) while protons of the acyl chain (CH2 H910 H811) show both short (Figure 980

6c) and long (Figure 6d) T2 relaxation component 981

- The short T2 relaxation time systematically decreases from 110degC to 70degC for all species by a 982

factor ranging from two to seven depending on the specific chemical group Very long T2 are 983

observed at 110degC for all acyl species while nearly unchanged values are measured below 984

90degC without any clear trend 985

- The sugar protons (H1rsquo H5rsquo H6rsquo) have the shortest set of T2 values (lt 1 ms) below 100degC 986

The long component of T2 generally ranges between 4 ms and 12 ms 987

0 1 2 3 4 5 6 7 8

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1 2 3 4 5 6 7 8

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

H1

H8H11

H9H10

CH2

H1rsquo

H6rsquo

CH2

H910H811 H1CH2

H1rsquo H5rsquo6rsquo

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

OH

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45 T= 110degC

T= 90degC

T= 80degC

T= 70degC

Log

In

ten

sity

Time ms

Fit

)

CH2 H910 H811 H1 H56 H1 OH

00

05

10

15

20

25

30

35

40

T2 m

s

T= 110degC

T= 90degC

T= 80degC

T= 70degC

CH2 H910 H811 H1

0

2

4

6

8

10

12

T2 m

s

T= 110degC

T= 90degC

T= 80degC

T= 70degC

Short T2(1H) Long T2(1H)

a) b)

c)

11

Table S 1 ndash 1H chemical shifts of G-C181-OH in solution and in the solid-state Solid-state spectra are 988

recorded at a MAS frequency of 10 kHz in DMSO-d6 broad peak 989

990

δH ppm

Group Ndeg Liquid

(MeOD-d4) Solid

(T= 110degC) Solid

(T= 70degC)

CH2 H1 354 357

CH2 H2 153 156

CH2 H3-712-14 132 135

CH2 H811 203 203

CH H910 534 534

CH2 H15 162 164

CH2 H16 388 383

CH H1 424 432

CH H2 316 335

CH H3 334 335

CH H4 329 335

CH H5 354 351 357

CH2 H6 366 383

OH OH1 432

45 463

OH OH2 492

OH OH3 490

OH OH4 487

OH OH6 445

991

992

H1H9H10H1rsquo

H6rsquo

OH2rsquo

OH1

12

Detailed discussion about the NOESY and DQ-SQ 2D 1H-1H ssNMR experiments presented in 993

Figure 9 994

995

2D 1H-1H ssNMR experiments are a unique tool to explore through-space proximities 996

between 1H nuclei ssNMR offers a large number of possibilities each adapted to a given 997

system The advantage of 2D maps with respect to a dynamics approach is their direct 998

readability in clear two nuclei which exchange their magnetization through-space (whichever 999

mechanism may be responsible for this exchange) will display a specific cross-peak However 1000

the way cross-peaks are presented and interpreted strictly depends on the nature of the pulse 1001

program which in turn is chosen according to the physical nature of the sample 1002

1D temperature-dependent 1H NMR experiments performed on G-C181-OH (Figure 1003

4a in the main text) indicate sharp peaks at 90degC and broad peaks at 70degC Narrow peaks 1004

indicate fast andor large-amplitude molecular tumbling averaging the strong 1H-1H dipolar 1005

interactions This excludes the use of pulse sequences like DQ-SQ which is not efficient under 1006

these conditions For this reason we employ spin-diffusion-based pulse sequences like 1007

NOESY which is more adapted for samples with high molecular mobility When mixing times 1008

are chosen to be short (lt 1 s) one generally explores through-space proximities within about 5 1009

10 Aring On the contrary when molecular motions are restricted dipolar coupling becomes 1010

significant 1H NMR peaks are broader and pulse sequences like DQ-SQ can now be used 1011

with a spatial resolution of less than 5 Aring 1012

Whenever possible DQ-SQ experiments are preferred over spin diffusion-based pulse 1013

sequences because they are more selective in particular for likewise interactions which are 1014

not discriminated in NOESY-type experiments This is illustrated by the typical interaction 1015

schemes for NOESY (top Figure 9a) and DQ-SQ (top Figure 9b) For two given A and B 1016

protons 2D NOESY maps will display two on-diagonal and two square-shaped symmetrical 1017

off-diagonal cross-peaks Off-diagonal cross-peaks illustrate the mutual A-B and B-A 1018

interaction but the on-diagonal peaks are systematic and do not prove any A-A or B-B 1019

interaction Similarly to NOESY 2D DQ-SQ maps display off-diagonal peaks (symmetric 1020

across the diagonal top Figure 9b) upon A-B and B-A interaction but in contrast they will 1021

display on-diagonal peaks only in the presence of self-interaction In the tutorial example of 1022

Figure 9b top the on-diagonal A-A peak indicates mutual A-A interaction while lack of on-1023

diagonal B-B peak evidences the absence of B-B proximities Combining NOESY and DQ-SQ 1024

experiments is capital to study the local arrangement of G-C181-OH and its evolution with 1025

temperature 1026

13

At T= 90degC the H1 and H5rsquo contributions are very close but possible to discriminate 1027

them (δ(H1)= 354 ppm δ(H5rsquo)= 351 ppm) according to 1H NMR chemical shifts in Table S 1028

1 2D 1H-1H NOESY in Figure 9a also discriminates between the two components attributed 1029

on the basis of the interactions with their mutual neighbors (green H1-H2 and H5rsquo-H1rsquo squares 1030

in Figure 9a) On this basis 2D NOESY does not show a direct H1rsquo-H1 correlation (blue red-1031

crossed) but rather an extensive network of intra-sugar interactions These features exclude 1032

suggest intra- and inter-sugar interactions and they exclude head-tail arrangements as a major 1033

molecular configuration at 90degC 1034

At 70degC the 1H NMR peaks are broader indicating a reduction of the T2 (1H) relaxation 1035

times andor residual homonuclear dipolar interactions excluding the use of NOESY In this 1036

case 1H-1H DQ-SQ correlation experiments are best suited Figure 9b presents the 1H-1H DQ-1037

SQ experiment on G-C181-OH at 70degC Peaks are broader than in NOESY and attribution may 1038

be cumbersome However the experiment is performed at 700 MHz and MAS= 22 kHz which 1039

improves spectral resolution with respect to the same experiment performed at 300 MHz and 1040

MAS= 10 kHz The auto-peak at δH= 356 ppm in the SQ dimension is attributed to H1-H1 spin 1041

pairs mainly resulting from intra CH2 interaction while the DQ cross-peaks at δH= 156-356 1042

ppm and δH= 432-356 ppm respectively origin from H2-H1 and H1rsquo-H1 interactions (Figure 1043

9b) DQ-SQ map does not show any auto-peak corresponding to H1rsquo-H1rsquo or Hxrsquo-Hxrsquo (x= 2rsquo 1044

3rsquo 4rsquo) interactions (red-crossed blue Figure 9b) thus excluding any proximity (lt 5 Aring) between 1045

close glucose neighbors On the contrary DQ-SQ maps (recorded both at 700 MHz Figure 9b 1046

or at 300 MHz not shown) systematically show DQ cross-peaks which can only be attributed 1047

to H1rsquo-H1 interactions (δH= 432-356) The presence of these H1rsquo-H1 interactions can only be 1048

explained by a head-tail arrangement of the G-C181-OH molecules as illustrated below the 1049

DQ-SQ map in Figure 9b 1050

4

939

940

Figure S 2 ndash Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 130degC 941

5

942

943

Figure S 3 - Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 90degC 944

945

90degC

1

22

6

946

947

948

949

Figure S 4 ndash Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 70degC 950

951

21

11

1

2

5

3

4

45

5

6

7

7

7

952

953

954

Figure S 5 ndash Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 25degC 955

956

957

1

1 1

8

958

Figure S 6 ndash a) SANS experiment recorded at 90degC for a 5 mgmL G-C181-OH aqueous solution b-c) 959

WAXS experiments recorded on aqueous solutions of G-C181-OH at b) 5 mgmL and c) 25 mgmL d) 960

Structural interpretation of the repeating distances pointed out in c) 961

962

963

964

001 01 1

q Aring-1

25degC

68degC

88degC

111degC

01 1q Aring

-1

02 04 06 08 1q Aring

-1

001 01

10-1

100

101

Inte

nsity

cm

-1

q Aring-1

-2

T= 90degC

SANS WAXS WAXS

a) b) c)

dv

dL d= 144 Aring-1

di

ddH2O

dt

dvdL

dtdi d

d)

9

965

966

Figure S 7 ndash ssNMR MAS (10 kHz) experiments recorded on a G-C181-OH dry powder a) 1H spectra 967

recorded from 110degC to 70degC b) 13C single pulse (SP) spectra recorded with high power 1H decoupling from 968

110degC to 70degC c) 13C cross polarization (CP) spectra recorded at 110degC and contact time tc of 50 micros and 969

3000 micros d) 13C CP spectra recorded at 70degC and for 20 lt tc micros lt 3000 970

971

1

811

910

CH2 CH2

1rsquo

6rsquo

8 6 4 2 0 ppm

110

90

80

70

910

1rsquo

6rsquo16

1

2rsquo3rsquo4rsquo

811

OH

45 40 35 30 25 20 15

ppm

200 180 160 140 120 100 80 60 40 20 0

ppm140 120 100 80

ppm

CH2-gauche

3004

110

90

80

70

65

60

50

25

C91013003

C9 - 12869

CH2-all-trans

3198

C1rsquo10335

a) b)

1H

T degCT degC

13C Single Pulse MAS

A

B

A

B

70degC70degC

5rsquo

215

200 180 160 140 120 100 80 60 40 20 0

ppm

13C CPMAST= 110degC

200 180 160 140 120 100 80 60 40 20 0

ppm

132 131 130 129 128 127 126

ppm

106 105 104 103 102

ppm

13C CPMAST= 70degC

50

tc micros

3000

20

tc micros

50 500 1000 3000

tc micros

C102

13024 C910113003

C92

12869

C1rsquo213082

C1rsquo110443

A B

106 105 104 103 102

ppm

3000

1000

500

50

20

AB

c)

d)

10

972

Figure S 8 ndash a) Short and d) long T2 components displayed by the 1H transverse relaxation signal c) 1H 973

transverse relaxation signals recorded on a G-C181-OH dry powder from 110degC to 70degC The area of each 974

given resonance is plotted (log-lin) as a function of the echo time (CPMG pulse sequence) for each 975

temperature and fitted using the equation given above T2s and T2l indicate the short and long T2 relaxation 976

times respectively obtained by describing the 1H transverse relaxation signal using a biexponential decay 977

For H1rsquo H5rsquo and H6rsquo the 1H relaxation function could be described by a monoexponential decay 978

- Protons from the sugar group (H1rsquo H5rsquo H6rsquo) show a single T2 relaxation component (Figure 979

6c and Figure S 8) while protons of the acyl chain (CH2 H910 H811) show both short (Figure 980

6c) and long (Figure 6d) T2 relaxation component 981

- The short T2 relaxation time systematically decreases from 110degC to 70degC for all species by a 982

factor ranging from two to seven depending on the specific chemical group Very long T2 are 983

observed at 110degC for all acyl species while nearly unchanged values are measured below 984

90degC without any clear trend 985

- The sugar protons (H1rsquo H5rsquo H6rsquo) have the shortest set of T2 values (lt 1 ms) below 100degC 986

The long component of T2 generally ranges between 4 ms and 12 ms 987

0 1 2 3 4 5 6 7 8

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1 2 3 4 5 6 7 8

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

H1

H8H11

H9H10

CH2

H1rsquo

H6rsquo

CH2

H910H811 H1CH2

H1rsquo H5rsquo6rsquo

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

OH

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45 T= 110degC

T= 90degC

T= 80degC

T= 70degC

Log

In

ten

sity

Time ms

Fit

)

CH2 H910 H811 H1 H56 H1 OH

00

05

10

15

20

25

30

35

40

T2 m

s

T= 110degC

T= 90degC

T= 80degC

T= 70degC

CH2 H910 H811 H1

0

2

4

6

8

10

12

T2 m

s

T= 110degC

T= 90degC

T= 80degC

T= 70degC

Short T2(1H) Long T2(1H)

a) b)

c)

11

Table S 1 ndash 1H chemical shifts of G-C181-OH in solution and in the solid-state Solid-state spectra are 988

recorded at a MAS frequency of 10 kHz in DMSO-d6 broad peak 989

990

δH ppm

Group Ndeg Liquid

(MeOD-d4) Solid

(T= 110degC) Solid

(T= 70degC)

CH2 H1 354 357

CH2 H2 153 156

CH2 H3-712-14 132 135

CH2 H811 203 203

CH H910 534 534

CH2 H15 162 164

CH2 H16 388 383

CH H1 424 432

CH H2 316 335

CH H3 334 335

CH H4 329 335

CH H5 354 351 357

CH2 H6 366 383

OH OH1 432

45 463

OH OH2 492

OH OH3 490

OH OH4 487

OH OH6 445

991

992

H1H9H10H1rsquo

H6rsquo

OH2rsquo

OH1

12

Detailed discussion about the NOESY and DQ-SQ 2D 1H-1H ssNMR experiments presented in 993

Figure 9 994

995

2D 1H-1H ssNMR experiments are a unique tool to explore through-space proximities 996

between 1H nuclei ssNMR offers a large number of possibilities each adapted to a given 997

system The advantage of 2D maps with respect to a dynamics approach is their direct 998

readability in clear two nuclei which exchange their magnetization through-space (whichever 999

mechanism may be responsible for this exchange) will display a specific cross-peak However 1000

the way cross-peaks are presented and interpreted strictly depends on the nature of the pulse 1001

program which in turn is chosen according to the physical nature of the sample 1002

1D temperature-dependent 1H NMR experiments performed on G-C181-OH (Figure 1003

4a in the main text) indicate sharp peaks at 90degC and broad peaks at 70degC Narrow peaks 1004

indicate fast andor large-amplitude molecular tumbling averaging the strong 1H-1H dipolar 1005

interactions This excludes the use of pulse sequences like DQ-SQ which is not efficient under 1006

these conditions For this reason we employ spin-diffusion-based pulse sequences like 1007

NOESY which is more adapted for samples with high molecular mobility When mixing times 1008

are chosen to be short (lt 1 s) one generally explores through-space proximities within about 5 1009

10 Aring On the contrary when molecular motions are restricted dipolar coupling becomes 1010

significant 1H NMR peaks are broader and pulse sequences like DQ-SQ can now be used 1011

with a spatial resolution of less than 5 Aring 1012

Whenever possible DQ-SQ experiments are preferred over spin diffusion-based pulse 1013

sequences because they are more selective in particular for likewise interactions which are 1014

not discriminated in NOESY-type experiments This is illustrated by the typical interaction 1015

schemes for NOESY (top Figure 9a) and DQ-SQ (top Figure 9b) For two given A and B 1016

protons 2D NOESY maps will display two on-diagonal and two square-shaped symmetrical 1017

off-diagonal cross-peaks Off-diagonal cross-peaks illustrate the mutual A-B and B-A 1018

interaction but the on-diagonal peaks are systematic and do not prove any A-A or B-B 1019

interaction Similarly to NOESY 2D DQ-SQ maps display off-diagonal peaks (symmetric 1020

across the diagonal top Figure 9b) upon A-B and B-A interaction but in contrast they will 1021

display on-diagonal peaks only in the presence of self-interaction In the tutorial example of 1022

Figure 9b top the on-diagonal A-A peak indicates mutual A-A interaction while lack of on-1023

diagonal B-B peak evidences the absence of B-B proximities Combining NOESY and DQ-SQ 1024

experiments is capital to study the local arrangement of G-C181-OH and its evolution with 1025

temperature 1026

13

At T= 90degC the H1 and H5rsquo contributions are very close but possible to discriminate 1027

them (δ(H1)= 354 ppm δ(H5rsquo)= 351 ppm) according to 1H NMR chemical shifts in Table S 1028

1 2D 1H-1H NOESY in Figure 9a also discriminates between the two components attributed 1029

on the basis of the interactions with their mutual neighbors (green H1-H2 and H5rsquo-H1rsquo squares 1030

in Figure 9a) On this basis 2D NOESY does not show a direct H1rsquo-H1 correlation (blue red-1031

crossed) but rather an extensive network of intra-sugar interactions These features exclude 1032

suggest intra- and inter-sugar interactions and they exclude head-tail arrangements as a major 1033

molecular configuration at 90degC 1034

At 70degC the 1H NMR peaks are broader indicating a reduction of the T2 (1H) relaxation 1035

times andor residual homonuclear dipolar interactions excluding the use of NOESY In this 1036

case 1H-1H DQ-SQ correlation experiments are best suited Figure 9b presents the 1H-1H DQ-1037

SQ experiment on G-C181-OH at 70degC Peaks are broader than in NOESY and attribution may 1038

be cumbersome However the experiment is performed at 700 MHz and MAS= 22 kHz which 1039

improves spectral resolution with respect to the same experiment performed at 300 MHz and 1040

MAS= 10 kHz The auto-peak at δH= 356 ppm in the SQ dimension is attributed to H1-H1 spin 1041

pairs mainly resulting from intra CH2 interaction while the DQ cross-peaks at δH= 156-356 1042

ppm and δH= 432-356 ppm respectively origin from H2-H1 and H1rsquo-H1 interactions (Figure 1043

9b) DQ-SQ map does not show any auto-peak corresponding to H1rsquo-H1rsquo or Hxrsquo-Hxrsquo (x= 2rsquo 1044

3rsquo 4rsquo) interactions (red-crossed blue Figure 9b) thus excluding any proximity (lt 5 Aring) between 1045

close glucose neighbors On the contrary DQ-SQ maps (recorded both at 700 MHz Figure 9b 1046

or at 300 MHz not shown) systematically show DQ cross-peaks which can only be attributed 1047

to H1rsquo-H1 interactions (δH= 432-356) The presence of these H1rsquo-H1 interactions can only be 1048

explained by a head-tail arrangement of the G-C181-OH molecules as illustrated below the 1049

DQ-SQ map in Figure 9b 1050

5

942

943

Figure S 3 - Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 90degC 944

945

90degC

1

22

6

946

947

948

949

Figure S 4 ndash Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 70degC 950

951

21

11

1

2

5

3

4

45

5

6

7

7

7

952

953

954

Figure S 5 ndash Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 25degC 955

956

957

1

1 1

8

958

Figure S 6 ndash a) SANS experiment recorded at 90degC for a 5 mgmL G-C181-OH aqueous solution b-c) 959

WAXS experiments recorded on aqueous solutions of G-C181-OH at b) 5 mgmL and c) 25 mgmL d) 960

Structural interpretation of the repeating distances pointed out in c) 961

962

963

964

001 01 1

q Aring-1

25degC

68degC

88degC

111degC

01 1q Aring

-1

02 04 06 08 1q Aring

-1

001 01

10-1

100

101

Inte

nsity

cm

-1

q Aring-1

-2

T= 90degC

SANS WAXS WAXS

a) b) c)

dv

dL d= 144 Aring-1

di

ddH2O

dt

dvdL

dtdi d

d)

9

965

966

Figure S 7 ndash ssNMR MAS (10 kHz) experiments recorded on a G-C181-OH dry powder a) 1H spectra 967

recorded from 110degC to 70degC b) 13C single pulse (SP) spectra recorded with high power 1H decoupling from 968

110degC to 70degC c) 13C cross polarization (CP) spectra recorded at 110degC and contact time tc of 50 micros and 969

3000 micros d) 13C CP spectra recorded at 70degC and for 20 lt tc micros lt 3000 970

971

1

811

910

CH2 CH2

1rsquo

6rsquo

8 6 4 2 0 ppm

110

90

80

70

910

1rsquo

6rsquo16

1

2rsquo3rsquo4rsquo

811

OH

45 40 35 30 25 20 15

ppm

200 180 160 140 120 100 80 60 40 20 0

ppm140 120 100 80

ppm

CH2-gauche

3004

110

90

80

70

65

60

50

25

C91013003

C9 - 12869

CH2-all-trans

3198

C1rsquo10335

a) b)

1H

T degCT degC

13C Single Pulse MAS

A

B

A

B

70degC70degC

5rsquo

215

200 180 160 140 120 100 80 60 40 20 0

ppm

13C CPMAST= 110degC

200 180 160 140 120 100 80 60 40 20 0

ppm

132 131 130 129 128 127 126

ppm

106 105 104 103 102

ppm

13C CPMAST= 70degC

50

tc micros

3000

20

tc micros

50 500 1000 3000

tc micros

C102

13024 C910113003

C92

12869

C1rsquo213082

C1rsquo110443

A B

106 105 104 103 102

ppm

3000

1000

500

50

20

AB

c)

d)

10

972

Figure S 8 ndash a) Short and d) long T2 components displayed by the 1H transverse relaxation signal c) 1H 973

transverse relaxation signals recorded on a G-C181-OH dry powder from 110degC to 70degC The area of each 974

given resonance is plotted (log-lin) as a function of the echo time (CPMG pulse sequence) for each 975

temperature and fitted using the equation given above T2s and T2l indicate the short and long T2 relaxation 976

times respectively obtained by describing the 1H transverse relaxation signal using a biexponential decay 977

For H1rsquo H5rsquo and H6rsquo the 1H relaxation function could be described by a monoexponential decay 978

- Protons from the sugar group (H1rsquo H5rsquo H6rsquo) show a single T2 relaxation component (Figure 979

6c and Figure S 8) while protons of the acyl chain (CH2 H910 H811) show both short (Figure 980

6c) and long (Figure 6d) T2 relaxation component 981

- The short T2 relaxation time systematically decreases from 110degC to 70degC for all species by a 982

factor ranging from two to seven depending on the specific chemical group Very long T2 are 983

observed at 110degC for all acyl species while nearly unchanged values are measured below 984

90degC without any clear trend 985

- The sugar protons (H1rsquo H5rsquo H6rsquo) have the shortest set of T2 values (lt 1 ms) below 100degC 986

The long component of T2 generally ranges between 4 ms and 12 ms 987

0 1 2 3 4 5 6 7 8

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1 2 3 4 5 6 7 8

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

H1

H8H11

H9H10

CH2

H1rsquo

H6rsquo

CH2

H910H811 H1CH2

H1rsquo H5rsquo6rsquo

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

OH

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45 T= 110degC

T= 90degC

T= 80degC

T= 70degC

Log

In

ten

sity

Time ms

Fit

)

CH2 H910 H811 H1 H56 H1 OH

00

05

10

15

20

25

30

35

40

T2 m

s

T= 110degC

T= 90degC

T= 80degC

T= 70degC

CH2 H910 H811 H1

0

2

4

6

8

10

12

T2 m

s

T= 110degC

T= 90degC

T= 80degC

T= 70degC

Short T2(1H) Long T2(1H)

a) b)

c)

11

Table S 1 ndash 1H chemical shifts of G-C181-OH in solution and in the solid-state Solid-state spectra are 988

recorded at a MAS frequency of 10 kHz in DMSO-d6 broad peak 989

990

δH ppm

Group Ndeg Liquid

(MeOD-d4) Solid

(T= 110degC) Solid

(T= 70degC)

CH2 H1 354 357

CH2 H2 153 156

CH2 H3-712-14 132 135

CH2 H811 203 203

CH H910 534 534

CH2 H15 162 164

CH2 H16 388 383

CH H1 424 432

CH H2 316 335

CH H3 334 335

CH H4 329 335

CH H5 354 351 357

CH2 H6 366 383

OH OH1 432

45 463

OH OH2 492

OH OH3 490

OH OH4 487

OH OH6 445

991

992

H1H9H10H1rsquo

H6rsquo

OH2rsquo

OH1

12

Detailed discussion about the NOESY and DQ-SQ 2D 1H-1H ssNMR experiments presented in 993

Figure 9 994

995

2D 1H-1H ssNMR experiments are a unique tool to explore through-space proximities 996

between 1H nuclei ssNMR offers a large number of possibilities each adapted to a given 997

system The advantage of 2D maps with respect to a dynamics approach is their direct 998

readability in clear two nuclei which exchange their magnetization through-space (whichever 999

mechanism may be responsible for this exchange) will display a specific cross-peak However 1000

the way cross-peaks are presented and interpreted strictly depends on the nature of the pulse 1001

program which in turn is chosen according to the physical nature of the sample 1002

1D temperature-dependent 1H NMR experiments performed on G-C181-OH (Figure 1003

4a in the main text) indicate sharp peaks at 90degC and broad peaks at 70degC Narrow peaks 1004

indicate fast andor large-amplitude molecular tumbling averaging the strong 1H-1H dipolar 1005

interactions This excludes the use of pulse sequences like DQ-SQ which is not efficient under 1006

these conditions For this reason we employ spin-diffusion-based pulse sequences like 1007

NOESY which is more adapted for samples with high molecular mobility When mixing times 1008

are chosen to be short (lt 1 s) one generally explores through-space proximities within about 5 1009

10 Aring On the contrary when molecular motions are restricted dipolar coupling becomes 1010

significant 1H NMR peaks are broader and pulse sequences like DQ-SQ can now be used 1011

with a spatial resolution of less than 5 Aring 1012

Whenever possible DQ-SQ experiments are preferred over spin diffusion-based pulse 1013

sequences because they are more selective in particular for likewise interactions which are 1014

not discriminated in NOESY-type experiments This is illustrated by the typical interaction 1015

schemes for NOESY (top Figure 9a) and DQ-SQ (top Figure 9b) For two given A and B 1016

protons 2D NOESY maps will display two on-diagonal and two square-shaped symmetrical 1017

off-diagonal cross-peaks Off-diagonal cross-peaks illustrate the mutual A-B and B-A 1018

interaction but the on-diagonal peaks are systematic and do not prove any A-A or B-B 1019

interaction Similarly to NOESY 2D DQ-SQ maps display off-diagonal peaks (symmetric 1020

across the diagonal top Figure 9b) upon A-B and B-A interaction but in contrast they will 1021

display on-diagonal peaks only in the presence of self-interaction In the tutorial example of 1022

Figure 9b top the on-diagonal A-A peak indicates mutual A-A interaction while lack of on-1023

diagonal B-B peak evidences the absence of B-B proximities Combining NOESY and DQ-SQ 1024

experiments is capital to study the local arrangement of G-C181-OH and its evolution with 1025

temperature 1026

13

At T= 90degC the H1 and H5rsquo contributions are very close but possible to discriminate 1027

them (δ(H1)= 354 ppm δ(H5rsquo)= 351 ppm) according to 1H NMR chemical shifts in Table S 1028

1 2D 1H-1H NOESY in Figure 9a also discriminates between the two components attributed 1029

on the basis of the interactions with their mutual neighbors (green H1-H2 and H5rsquo-H1rsquo squares 1030

in Figure 9a) On this basis 2D NOESY does not show a direct H1rsquo-H1 correlation (blue red-1031

crossed) but rather an extensive network of intra-sugar interactions These features exclude 1032

suggest intra- and inter-sugar interactions and they exclude head-tail arrangements as a major 1033

molecular configuration at 90degC 1034

At 70degC the 1H NMR peaks are broader indicating a reduction of the T2 (1H) relaxation 1035

times andor residual homonuclear dipolar interactions excluding the use of NOESY In this 1036

case 1H-1H DQ-SQ correlation experiments are best suited Figure 9b presents the 1H-1H DQ-1037

SQ experiment on G-C181-OH at 70degC Peaks are broader than in NOESY and attribution may 1038

be cumbersome However the experiment is performed at 700 MHz and MAS= 22 kHz which 1039

improves spectral resolution with respect to the same experiment performed at 300 MHz and 1040

MAS= 10 kHz The auto-peak at δH= 356 ppm in the SQ dimension is attributed to H1-H1 spin 1041

pairs mainly resulting from intra CH2 interaction while the DQ cross-peaks at δH= 156-356 1042

ppm and δH= 432-356 ppm respectively origin from H2-H1 and H1rsquo-H1 interactions (Figure 1043

9b) DQ-SQ map does not show any auto-peak corresponding to H1rsquo-H1rsquo or Hxrsquo-Hxrsquo (x= 2rsquo 1044

3rsquo 4rsquo) interactions (red-crossed blue Figure 9b) thus excluding any proximity (lt 5 Aring) between 1045

close glucose neighbors On the contrary DQ-SQ maps (recorded both at 700 MHz Figure 9b 1046

or at 300 MHz not shown) systematically show DQ cross-peaks which can only be attributed 1047

to H1rsquo-H1 interactions (δH= 432-356) The presence of these H1rsquo-H1 interactions can only be 1048

explained by a head-tail arrangement of the G-C181-OH molecules as illustrated below the 1049

DQ-SQ map in Figure 9b 1050

6

946

947

948

949

Figure S 4 ndash Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 70degC 950

951

21

11

1

2

5

3

4

45

5

6

7

7

7

952

953

954

Figure S 5 ndash Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 25degC 955

956

957

1

1 1

8

958

Figure S 6 ndash a) SANS experiment recorded at 90degC for a 5 mgmL G-C181-OH aqueous solution b-c) 959

WAXS experiments recorded on aqueous solutions of G-C181-OH at b) 5 mgmL and c) 25 mgmL d) 960

Structural interpretation of the repeating distances pointed out in c) 961

962

963

964

001 01 1

q Aring-1

25degC

68degC

88degC

111degC

01 1q Aring

-1

02 04 06 08 1q Aring

-1

001 01

10-1

100

101

Inte

nsity

cm

-1

q Aring-1

-2

T= 90degC

SANS WAXS WAXS

a) b) c)

dv

dL d= 144 Aring-1

di

ddH2O

dt

dvdL

dtdi d

d)

9

965

966

Figure S 7 ndash ssNMR MAS (10 kHz) experiments recorded on a G-C181-OH dry powder a) 1H spectra 967

recorded from 110degC to 70degC b) 13C single pulse (SP) spectra recorded with high power 1H decoupling from 968

110degC to 70degC c) 13C cross polarization (CP) spectra recorded at 110degC and contact time tc of 50 micros and 969

3000 micros d) 13C CP spectra recorded at 70degC and for 20 lt tc micros lt 3000 970

971

1

811

910

CH2 CH2

1rsquo

6rsquo

8 6 4 2 0 ppm

110

90

80

70

910

1rsquo

6rsquo16

1

2rsquo3rsquo4rsquo

811

OH

45 40 35 30 25 20 15

ppm

200 180 160 140 120 100 80 60 40 20 0

ppm140 120 100 80

ppm

CH2-gauche

3004

110

90

80

70

65

60

50

25

C91013003

C9 - 12869

CH2-all-trans

3198

C1rsquo10335

a) b)

1H

T degCT degC

13C Single Pulse MAS

A

B

A

B

70degC70degC

5rsquo

215

200 180 160 140 120 100 80 60 40 20 0

ppm

13C CPMAST= 110degC

200 180 160 140 120 100 80 60 40 20 0

ppm

132 131 130 129 128 127 126

ppm

106 105 104 103 102

ppm

13C CPMAST= 70degC

50

tc micros

3000

20

tc micros

50 500 1000 3000

tc micros

C102

13024 C910113003

C92

12869

C1rsquo213082

C1rsquo110443

A B

106 105 104 103 102

ppm

3000

1000

500

50

20

AB

c)

d)

10

972

Figure S 8 ndash a) Short and d) long T2 components displayed by the 1H transverse relaxation signal c) 1H 973

transverse relaxation signals recorded on a G-C181-OH dry powder from 110degC to 70degC The area of each 974

given resonance is plotted (log-lin) as a function of the echo time (CPMG pulse sequence) for each 975

temperature and fitted using the equation given above T2s and T2l indicate the short and long T2 relaxation 976

times respectively obtained by describing the 1H transverse relaxation signal using a biexponential decay 977

For H1rsquo H5rsquo and H6rsquo the 1H relaxation function could be described by a monoexponential decay 978

- Protons from the sugar group (H1rsquo H5rsquo H6rsquo) show a single T2 relaxation component (Figure 979

6c and Figure S 8) while protons of the acyl chain (CH2 H910 H811) show both short (Figure 980

6c) and long (Figure 6d) T2 relaxation component 981

- The short T2 relaxation time systematically decreases from 110degC to 70degC for all species by a 982

factor ranging from two to seven depending on the specific chemical group Very long T2 are 983

observed at 110degC for all acyl species while nearly unchanged values are measured below 984

90degC without any clear trend 985

- The sugar protons (H1rsquo H5rsquo H6rsquo) have the shortest set of T2 values (lt 1 ms) below 100degC 986

The long component of T2 generally ranges between 4 ms and 12 ms 987

0 1 2 3 4 5 6 7 8

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1 2 3 4 5 6 7 8

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

H1

H8H11

H9H10

CH2

H1rsquo

H6rsquo

CH2

H910H811 H1CH2

H1rsquo H5rsquo6rsquo

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

OH

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45 T= 110degC

T= 90degC

T= 80degC

T= 70degC

Log

In

ten

sity

Time ms

Fit

)

CH2 H910 H811 H1 H56 H1 OH

00

05

10

15

20

25

30

35

40

T2 m

s

T= 110degC

T= 90degC

T= 80degC

T= 70degC

CH2 H910 H811 H1

0

2

4

6

8

10

12

T2 m

s

T= 110degC

T= 90degC

T= 80degC

T= 70degC

Short T2(1H) Long T2(1H)

a) b)

c)

11

Table S 1 ndash 1H chemical shifts of G-C181-OH in solution and in the solid-state Solid-state spectra are 988

recorded at a MAS frequency of 10 kHz in DMSO-d6 broad peak 989

990

δH ppm

Group Ndeg Liquid

(MeOD-d4) Solid

(T= 110degC) Solid

(T= 70degC)

CH2 H1 354 357

CH2 H2 153 156

CH2 H3-712-14 132 135

CH2 H811 203 203

CH H910 534 534

CH2 H15 162 164

CH2 H16 388 383

CH H1 424 432

CH H2 316 335

CH H3 334 335

CH H4 329 335

CH H5 354 351 357

CH2 H6 366 383

OH OH1 432

45 463

OH OH2 492

OH OH3 490

OH OH4 487

OH OH6 445

991

992

H1H9H10H1rsquo

H6rsquo

OH2rsquo

OH1

12

Detailed discussion about the NOESY and DQ-SQ 2D 1H-1H ssNMR experiments presented in 993

Figure 9 994

995

2D 1H-1H ssNMR experiments are a unique tool to explore through-space proximities 996

between 1H nuclei ssNMR offers a large number of possibilities each adapted to a given 997

system The advantage of 2D maps with respect to a dynamics approach is their direct 998

readability in clear two nuclei which exchange their magnetization through-space (whichever 999

mechanism may be responsible for this exchange) will display a specific cross-peak However 1000

the way cross-peaks are presented and interpreted strictly depends on the nature of the pulse 1001

program which in turn is chosen according to the physical nature of the sample 1002

1D temperature-dependent 1H NMR experiments performed on G-C181-OH (Figure 1003

4a in the main text) indicate sharp peaks at 90degC and broad peaks at 70degC Narrow peaks 1004

indicate fast andor large-amplitude molecular tumbling averaging the strong 1H-1H dipolar 1005

interactions This excludes the use of pulse sequences like DQ-SQ which is not efficient under 1006

these conditions For this reason we employ spin-diffusion-based pulse sequences like 1007

NOESY which is more adapted for samples with high molecular mobility When mixing times 1008

are chosen to be short (lt 1 s) one generally explores through-space proximities within about 5 1009

10 Aring On the contrary when molecular motions are restricted dipolar coupling becomes 1010

significant 1H NMR peaks are broader and pulse sequences like DQ-SQ can now be used 1011

with a spatial resolution of less than 5 Aring 1012

Whenever possible DQ-SQ experiments are preferred over spin diffusion-based pulse 1013

sequences because they are more selective in particular for likewise interactions which are 1014

not discriminated in NOESY-type experiments This is illustrated by the typical interaction 1015

schemes for NOESY (top Figure 9a) and DQ-SQ (top Figure 9b) For two given A and B 1016

protons 2D NOESY maps will display two on-diagonal and two square-shaped symmetrical 1017

off-diagonal cross-peaks Off-diagonal cross-peaks illustrate the mutual A-B and B-A 1018

interaction but the on-diagonal peaks are systematic and do not prove any A-A or B-B 1019

interaction Similarly to NOESY 2D DQ-SQ maps display off-diagonal peaks (symmetric 1020

across the diagonal top Figure 9b) upon A-B and B-A interaction but in contrast they will 1021

display on-diagonal peaks only in the presence of self-interaction In the tutorial example of 1022

Figure 9b top the on-diagonal A-A peak indicates mutual A-A interaction while lack of on-1023

diagonal B-B peak evidences the absence of B-B proximities Combining NOESY and DQ-SQ 1024

experiments is capital to study the local arrangement of G-C181-OH and its evolution with 1025

temperature 1026

13

At T= 90degC the H1 and H5rsquo contributions are very close but possible to discriminate 1027

them (δ(H1)= 354 ppm δ(H5rsquo)= 351 ppm) according to 1H NMR chemical shifts in Table S 1028

1 2D 1H-1H NOESY in Figure 9a also discriminates between the two components attributed 1029

on the basis of the interactions with their mutual neighbors (green H1-H2 and H5rsquo-H1rsquo squares 1030

in Figure 9a) On this basis 2D NOESY does not show a direct H1rsquo-H1 correlation (blue red-1031

crossed) but rather an extensive network of intra-sugar interactions These features exclude 1032

suggest intra- and inter-sugar interactions and they exclude head-tail arrangements as a major 1033

molecular configuration at 90degC 1034

At 70degC the 1H NMR peaks are broader indicating a reduction of the T2 (1H) relaxation 1035

times andor residual homonuclear dipolar interactions excluding the use of NOESY In this 1036

case 1H-1H DQ-SQ correlation experiments are best suited Figure 9b presents the 1H-1H DQ-1037

SQ experiment on G-C181-OH at 70degC Peaks are broader than in NOESY and attribution may 1038

be cumbersome However the experiment is performed at 700 MHz and MAS= 22 kHz which 1039

improves spectral resolution with respect to the same experiment performed at 300 MHz and 1040

MAS= 10 kHz The auto-peak at δH= 356 ppm in the SQ dimension is attributed to H1-H1 spin 1041

pairs mainly resulting from intra CH2 interaction while the DQ cross-peaks at δH= 156-356 1042

ppm and δH= 432-356 ppm respectively origin from H2-H1 and H1rsquo-H1 interactions (Figure 1043

9b) DQ-SQ map does not show any auto-peak corresponding to H1rsquo-H1rsquo or Hxrsquo-Hxrsquo (x= 2rsquo 1044

3rsquo 4rsquo) interactions (red-crossed blue Figure 9b) thus excluding any proximity (lt 5 Aring) between 1045

close glucose neighbors On the contrary DQ-SQ maps (recorded both at 700 MHz Figure 9b 1046

or at 300 MHz not shown) systematically show DQ cross-peaks which can only be attributed 1047

to H1rsquo-H1 interactions (δH= 432-356) The presence of these H1rsquo-H1 interactions can only be 1048

explained by a head-tail arrangement of the G-C181-OH molecules as illustrated below the 1049

DQ-SQ map in Figure 9b 1050

7

952

953

954

Figure S 5 ndash Cryo-TEM images recorded for a 5 mgmL G-C181-OH aqueous solution heat at T= 25degC 955

956

957

1

1 1

8

958

Figure S 6 ndash a) SANS experiment recorded at 90degC for a 5 mgmL G-C181-OH aqueous solution b-c) 959

WAXS experiments recorded on aqueous solutions of G-C181-OH at b) 5 mgmL and c) 25 mgmL d) 960

Structural interpretation of the repeating distances pointed out in c) 961

962

963

964

001 01 1

q Aring-1

25degC

68degC

88degC

111degC

01 1q Aring

-1

02 04 06 08 1q Aring

-1

001 01

10-1

100

101

Inte

nsity

cm

-1

q Aring-1

-2

T= 90degC

SANS WAXS WAXS

a) b) c)

dv

dL d= 144 Aring-1

di

ddH2O

dt

dvdL

dtdi d

d)

9

965

966

Figure S 7 ndash ssNMR MAS (10 kHz) experiments recorded on a G-C181-OH dry powder a) 1H spectra 967

recorded from 110degC to 70degC b) 13C single pulse (SP) spectra recorded with high power 1H decoupling from 968

110degC to 70degC c) 13C cross polarization (CP) spectra recorded at 110degC and contact time tc of 50 micros and 969

3000 micros d) 13C CP spectra recorded at 70degC and for 20 lt tc micros lt 3000 970

971

1

811

910

CH2 CH2

1rsquo

6rsquo

8 6 4 2 0 ppm

110

90

80

70

910

1rsquo

6rsquo16

1

2rsquo3rsquo4rsquo

811

OH

45 40 35 30 25 20 15

ppm

200 180 160 140 120 100 80 60 40 20 0

ppm140 120 100 80

ppm

CH2-gauche

3004

110

90

80

70

65

60

50

25

C91013003

C9 - 12869

CH2-all-trans

3198

C1rsquo10335

a) b)

1H

T degCT degC

13C Single Pulse MAS

A

B

A

B

70degC70degC

5rsquo

215

200 180 160 140 120 100 80 60 40 20 0

ppm

13C CPMAST= 110degC

200 180 160 140 120 100 80 60 40 20 0

ppm

132 131 130 129 128 127 126

ppm

106 105 104 103 102

ppm

13C CPMAST= 70degC

50

tc micros

3000

20

tc micros

50 500 1000 3000

tc micros

C102

13024 C910113003

C92

12869

C1rsquo213082

C1rsquo110443

A B

106 105 104 103 102

ppm

3000

1000

500

50

20

AB

c)

d)

10

972

Figure S 8 ndash a) Short and d) long T2 components displayed by the 1H transverse relaxation signal c) 1H 973

transverse relaxation signals recorded on a G-C181-OH dry powder from 110degC to 70degC The area of each 974

given resonance is plotted (log-lin) as a function of the echo time (CPMG pulse sequence) for each 975

temperature and fitted using the equation given above T2s and T2l indicate the short and long T2 relaxation 976

times respectively obtained by describing the 1H transverse relaxation signal using a biexponential decay 977

For H1rsquo H5rsquo and H6rsquo the 1H relaxation function could be described by a monoexponential decay 978

- Protons from the sugar group (H1rsquo H5rsquo H6rsquo) show a single T2 relaxation component (Figure 979

6c and Figure S 8) while protons of the acyl chain (CH2 H910 H811) show both short (Figure 980

6c) and long (Figure 6d) T2 relaxation component 981

- The short T2 relaxation time systematically decreases from 110degC to 70degC for all species by a 982

factor ranging from two to seven depending on the specific chemical group Very long T2 are 983

observed at 110degC for all acyl species while nearly unchanged values are measured below 984

90degC without any clear trend 985

- The sugar protons (H1rsquo H5rsquo H6rsquo) have the shortest set of T2 values (lt 1 ms) below 100degC 986

The long component of T2 generally ranges between 4 ms and 12 ms 987

0 1 2 3 4 5 6 7 8

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1 2 3 4 5 6 7 8

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

H1

H8H11

H9H10

CH2

H1rsquo

H6rsquo

CH2

H910H811 H1CH2

H1rsquo H5rsquo6rsquo

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

OH

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45 T= 110degC

T= 90degC

T= 80degC

T= 70degC

Log

In

ten

sity

Time ms

Fit

)

CH2 H910 H811 H1 H56 H1 OH

00

05

10

15

20

25

30

35

40

T2 m

s

T= 110degC

T= 90degC

T= 80degC

T= 70degC

CH2 H910 H811 H1

0

2

4

6

8

10

12

T2 m

s

T= 110degC

T= 90degC

T= 80degC

T= 70degC

Short T2(1H) Long T2(1H)

a) b)

c)

11

Table S 1 ndash 1H chemical shifts of G-C181-OH in solution and in the solid-state Solid-state spectra are 988

recorded at a MAS frequency of 10 kHz in DMSO-d6 broad peak 989

990

δH ppm

Group Ndeg Liquid

(MeOD-d4) Solid

(T= 110degC) Solid

(T= 70degC)

CH2 H1 354 357

CH2 H2 153 156

CH2 H3-712-14 132 135

CH2 H811 203 203

CH H910 534 534

CH2 H15 162 164

CH2 H16 388 383

CH H1 424 432

CH H2 316 335

CH H3 334 335

CH H4 329 335

CH H5 354 351 357

CH2 H6 366 383

OH OH1 432

45 463

OH OH2 492

OH OH3 490

OH OH4 487

OH OH6 445

991

992

H1H9H10H1rsquo

H6rsquo

OH2rsquo

OH1

12

Detailed discussion about the NOESY and DQ-SQ 2D 1H-1H ssNMR experiments presented in 993

Figure 9 994

995

2D 1H-1H ssNMR experiments are a unique tool to explore through-space proximities 996

between 1H nuclei ssNMR offers a large number of possibilities each adapted to a given 997

system The advantage of 2D maps with respect to a dynamics approach is their direct 998

readability in clear two nuclei which exchange their magnetization through-space (whichever 999

mechanism may be responsible for this exchange) will display a specific cross-peak However 1000

the way cross-peaks are presented and interpreted strictly depends on the nature of the pulse 1001

program which in turn is chosen according to the physical nature of the sample 1002

1D temperature-dependent 1H NMR experiments performed on G-C181-OH (Figure 1003

4a in the main text) indicate sharp peaks at 90degC and broad peaks at 70degC Narrow peaks 1004

indicate fast andor large-amplitude molecular tumbling averaging the strong 1H-1H dipolar 1005

interactions This excludes the use of pulse sequences like DQ-SQ which is not efficient under 1006

these conditions For this reason we employ spin-diffusion-based pulse sequences like 1007

NOESY which is more adapted for samples with high molecular mobility When mixing times 1008

are chosen to be short (lt 1 s) one generally explores through-space proximities within about 5 1009

10 Aring On the contrary when molecular motions are restricted dipolar coupling becomes 1010

significant 1H NMR peaks are broader and pulse sequences like DQ-SQ can now be used 1011

with a spatial resolution of less than 5 Aring 1012

Whenever possible DQ-SQ experiments are preferred over spin diffusion-based pulse 1013

sequences because they are more selective in particular for likewise interactions which are 1014

not discriminated in NOESY-type experiments This is illustrated by the typical interaction 1015

schemes for NOESY (top Figure 9a) and DQ-SQ (top Figure 9b) For two given A and B 1016

protons 2D NOESY maps will display two on-diagonal and two square-shaped symmetrical 1017

off-diagonal cross-peaks Off-diagonal cross-peaks illustrate the mutual A-B and B-A 1018

interaction but the on-diagonal peaks are systematic and do not prove any A-A or B-B 1019

interaction Similarly to NOESY 2D DQ-SQ maps display off-diagonal peaks (symmetric 1020

across the diagonal top Figure 9b) upon A-B and B-A interaction but in contrast they will 1021

display on-diagonal peaks only in the presence of self-interaction In the tutorial example of 1022

Figure 9b top the on-diagonal A-A peak indicates mutual A-A interaction while lack of on-1023

diagonal B-B peak evidences the absence of B-B proximities Combining NOESY and DQ-SQ 1024

experiments is capital to study the local arrangement of G-C181-OH and its evolution with 1025

temperature 1026

13

At T= 90degC the H1 and H5rsquo contributions are very close but possible to discriminate 1027

them (δ(H1)= 354 ppm δ(H5rsquo)= 351 ppm) according to 1H NMR chemical shifts in Table S 1028

1 2D 1H-1H NOESY in Figure 9a also discriminates between the two components attributed 1029

on the basis of the interactions with their mutual neighbors (green H1-H2 and H5rsquo-H1rsquo squares 1030

in Figure 9a) On this basis 2D NOESY does not show a direct H1rsquo-H1 correlation (blue red-1031

crossed) but rather an extensive network of intra-sugar interactions These features exclude 1032

suggest intra- and inter-sugar interactions and they exclude head-tail arrangements as a major 1033

molecular configuration at 90degC 1034

At 70degC the 1H NMR peaks are broader indicating a reduction of the T2 (1H) relaxation 1035

times andor residual homonuclear dipolar interactions excluding the use of NOESY In this 1036

case 1H-1H DQ-SQ correlation experiments are best suited Figure 9b presents the 1H-1H DQ-1037

SQ experiment on G-C181-OH at 70degC Peaks are broader than in NOESY and attribution may 1038

be cumbersome However the experiment is performed at 700 MHz and MAS= 22 kHz which 1039

improves spectral resolution with respect to the same experiment performed at 300 MHz and 1040

MAS= 10 kHz The auto-peak at δH= 356 ppm in the SQ dimension is attributed to H1-H1 spin 1041

pairs mainly resulting from intra CH2 interaction while the DQ cross-peaks at δH= 156-356 1042

ppm and δH= 432-356 ppm respectively origin from H2-H1 and H1rsquo-H1 interactions (Figure 1043

9b) DQ-SQ map does not show any auto-peak corresponding to H1rsquo-H1rsquo or Hxrsquo-Hxrsquo (x= 2rsquo 1044

3rsquo 4rsquo) interactions (red-crossed blue Figure 9b) thus excluding any proximity (lt 5 Aring) between 1045

close glucose neighbors On the contrary DQ-SQ maps (recorded both at 700 MHz Figure 9b 1046

or at 300 MHz not shown) systematically show DQ cross-peaks which can only be attributed 1047

to H1rsquo-H1 interactions (δH= 432-356) The presence of these H1rsquo-H1 interactions can only be 1048

explained by a head-tail arrangement of the G-C181-OH molecules as illustrated below the 1049

DQ-SQ map in Figure 9b 1050

8

958

Figure S 6 ndash a) SANS experiment recorded at 90degC for a 5 mgmL G-C181-OH aqueous solution b-c) 959

WAXS experiments recorded on aqueous solutions of G-C181-OH at b) 5 mgmL and c) 25 mgmL d) 960

Structural interpretation of the repeating distances pointed out in c) 961

962

963

964

001 01 1

q Aring-1

25degC

68degC

88degC

111degC

01 1q Aring

-1

02 04 06 08 1q Aring

-1

001 01

10-1

100

101

Inte

nsity

cm

-1

q Aring-1

-2

T= 90degC

SANS WAXS WAXS

a) b) c)

dv

dL d= 144 Aring-1

di

ddH2O

dt

dvdL

dtdi d

d)

9

965

966

Figure S 7 ndash ssNMR MAS (10 kHz) experiments recorded on a G-C181-OH dry powder a) 1H spectra 967

recorded from 110degC to 70degC b) 13C single pulse (SP) spectra recorded with high power 1H decoupling from 968

110degC to 70degC c) 13C cross polarization (CP) spectra recorded at 110degC and contact time tc of 50 micros and 969

3000 micros d) 13C CP spectra recorded at 70degC and for 20 lt tc micros lt 3000 970

971

1

811

910

CH2 CH2

1rsquo

6rsquo

8 6 4 2 0 ppm

110

90

80

70

910

1rsquo

6rsquo16

1

2rsquo3rsquo4rsquo

811

OH

45 40 35 30 25 20 15

ppm

200 180 160 140 120 100 80 60 40 20 0

ppm140 120 100 80

ppm

CH2-gauche

3004

110

90

80

70

65

60

50

25

C91013003

C9 - 12869

CH2-all-trans

3198

C1rsquo10335

a) b)

1H

T degCT degC

13C Single Pulse MAS

A

B

A

B

70degC70degC

5rsquo

215

200 180 160 140 120 100 80 60 40 20 0

ppm

13C CPMAST= 110degC

200 180 160 140 120 100 80 60 40 20 0

ppm

132 131 130 129 128 127 126

ppm

106 105 104 103 102

ppm

13C CPMAST= 70degC

50

tc micros

3000

20

tc micros

50 500 1000 3000

tc micros

C102

13024 C910113003

C92

12869

C1rsquo213082

C1rsquo110443

A B

106 105 104 103 102

ppm

3000

1000

500

50

20

AB

c)

d)

10

972

Figure S 8 ndash a) Short and d) long T2 components displayed by the 1H transverse relaxation signal c) 1H 973

transverse relaxation signals recorded on a G-C181-OH dry powder from 110degC to 70degC The area of each 974

given resonance is plotted (log-lin) as a function of the echo time (CPMG pulse sequence) for each 975

temperature and fitted using the equation given above T2s and T2l indicate the short and long T2 relaxation 976

times respectively obtained by describing the 1H transverse relaxation signal using a biexponential decay 977

For H1rsquo H5rsquo and H6rsquo the 1H relaxation function could be described by a monoexponential decay 978

- Protons from the sugar group (H1rsquo H5rsquo H6rsquo) show a single T2 relaxation component (Figure 979

6c and Figure S 8) while protons of the acyl chain (CH2 H910 H811) show both short (Figure 980

6c) and long (Figure 6d) T2 relaxation component 981

- The short T2 relaxation time systematically decreases from 110degC to 70degC for all species by a 982

factor ranging from two to seven depending on the specific chemical group Very long T2 are 983

observed at 110degC for all acyl species while nearly unchanged values are measured below 984

90degC without any clear trend 985

- The sugar protons (H1rsquo H5rsquo H6rsquo) have the shortest set of T2 values (lt 1 ms) below 100degC 986

The long component of T2 generally ranges between 4 ms and 12 ms 987

0 1 2 3 4 5 6 7 8

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1 2 3 4 5 6 7 8

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

H1

H8H11

H9H10

CH2

H1rsquo

H6rsquo

CH2

H910H811 H1CH2

H1rsquo H5rsquo6rsquo

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

OH

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45 T= 110degC

T= 90degC

T= 80degC

T= 70degC

Log

In

ten

sity

Time ms

Fit

)

CH2 H910 H811 H1 H56 H1 OH

00

05

10

15

20

25

30

35

40

T2 m

s

T= 110degC

T= 90degC

T= 80degC

T= 70degC

CH2 H910 H811 H1

0

2

4

6

8

10

12

T2 m

s

T= 110degC

T= 90degC

T= 80degC

T= 70degC

Short T2(1H) Long T2(1H)

a) b)

c)

11

Table S 1 ndash 1H chemical shifts of G-C181-OH in solution and in the solid-state Solid-state spectra are 988

recorded at a MAS frequency of 10 kHz in DMSO-d6 broad peak 989

990

δH ppm

Group Ndeg Liquid

(MeOD-d4) Solid

(T= 110degC) Solid

(T= 70degC)

CH2 H1 354 357

CH2 H2 153 156

CH2 H3-712-14 132 135

CH2 H811 203 203

CH H910 534 534

CH2 H15 162 164

CH2 H16 388 383

CH H1 424 432

CH H2 316 335

CH H3 334 335

CH H4 329 335

CH H5 354 351 357

CH2 H6 366 383

OH OH1 432

45 463

OH OH2 492

OH OH3 490

OH OH4 487

OH OH6 445

991

992

H1H9H10H1rsquo

H6rsquo

OH2rsquo

OH1

12

Detailed discussion about the NOESY and DQ-SQ 2D 1H-1H ssNMR experiments presented in 993

Figure 9 994

995

2D 1H-1H ssNMR experiments are a unique tool to explore through-space proximities 996

between 1H nuclei ssNMR offers a large number of possibilities each adapted to a given 997

system The advantage of 2D maps with respect to a dynamics approach is their direct 998

readability in clear two nuclei which exchange their magnetization through-space (whichever 999

mechanism may be responsible for this exchange) will display a specific cross-peak However 1000

the way cross-peaks are presented and interpreted strictly depends on the nature of the pulse 1001

program which in turn is chosen according to the physical nature of the sample 1002

1D temperature-dependent 1H NMR experiments performed on G-C181-OH (Figure 1003

4a in the main text) indicate sharp peaks at 90degC and broad peaks at 70degC Narrow peaks 1004

indicate fast andor large-amplitude molecular tumbling averaging the strong 1H-1H dipolar 1005

interactions This excludes the use of pulse sequences like DQ-SQ which is not efficient under 1006

these conditions For this reason we employ spin-diffusion-based pulse sequences like 1007

NOESY which is more adapted for samples with high molecular mobility When mixing times 1008

are chosen to be short (lt 1 s) one generally explores through-space proximities within about 5 1009

10 Aring On the contrary when molecular motions are restricted dipolar coupling becomes 1010

significant 1H NMR peaks are broader and pulse sequences like DQ-SQ can now be used 1011

with a spatial resolution of less than 5 Aring 1012

Whenever possible DQ-SQ experiments are preferred over spin diffusion-based pulse 1013

sequences because they are more selective in particular for likewise interactions which are 1014

not discriminated in NOESY-type experiments This is illustrated by the typical interaction 1015

schemes for NOESY (top Figure 9a) and DQ-SQ (top Figure 9b) For two given A and B 1016

protons 2D NOESY maps will display two on-diagonal and two square-shaped symmetrical 1017

off-diagonal cross-peaks Off-diagonal cross-peaks illustrate the mutual A-B and B-A 1018

interaction but the on-diagonal peaks are systematic and do not prove any A-A or B-B 1019

interaction Similarly to NOESY 2D DQ-SQ maps display off-diagonal peaks (symmetric 1020

across the diagonal top Figure 9b) upon A-B and B-A interaction but in contrast they will 1021

display on-diagonal peaks only in the presence of self-interaction In the tutorial example of 1022

Figure 9b top the on-diagonal A-A peak indicates mutual A-A interaction while lack of on-1023

diagonal B-B peak evidences the absence of B-B proximities Combining NOESY and DQ-SQ 1024

experiments is capital to study the local arrangement of G-C181-OH and its evolution with 1025

temperature 1026

13

At T= 90degC the H1 and H5rsquo contributions are very close but possible to discriminate 1027

them (δ(H1)= 354 ppm δ(H5rsquo)= 351 ppm) according to 1H NMR chemical shifts in Table S 1028

1 2D 1H-1H NOESY in Figure 9a also discriminates between the two components attributed 1029

on the basis of the interactions with their mutual neighbors (green H1-H2 and H5rsquo-H1rsquo squares 1030

in Figure 9a) On this basis 2D NOESY does not show a direct H1rsquo-H1 correlation (blue red-1031

crossed) but rather an extensive network of intra-sugar interactions These features exclude 1032

suggest intra- and inter-sugar interactions and they exclude head-tail arrangements as a major 1033

molecular configuration at 90degC 1034

At 70degC the 1H NMR peaks are broader indicating a reduction of the T2 (1H) relaxation 1035

times andor residual homonuclear dipolar interactions excluding the use of NOESY In this 1036

case 1H-1H DQ-SQ correlation experiments are best suited Figure 9b presents the 1H-1H DQ-1037

SQ experiment on G-C181-OH at 70degC Peaks are broader than in NOESY and attribution may 1038

be cumbersome However the experiment is performed at 700 MHz and MAS= 22 kHz which 1039

improves spectral resolution with respect to the same experiment performed at 300 MHz and 1040

MAS= 10 kHz The auto-peak at δH= 356 ppm in the SQ dimension is attributed to H1-H1 spin 1041

pairs mainly resulting from intra CH2 interaction while the DQ cross-peaks at δH= 156-356 1042

ppm and δH= 432-356 ppm respectively origin from H2-H1 and H1rsquo-H1 interactions (Figure 1043

9b) DQ-SQ map does not show any auto-peak corresponding to H1rsquo-H1rsquo or Hxrsquo-Hxrsquo (x= 2rsquo 1044

3rsquo 4rsquo) interactions (red-crossed blue Figure 9b) thus excluding any proximity (lt 5 Aring) between 1045

close glucose neighbors On the contrary DQ-SQ maps (recorded both at 700 MHz Figure 9b 1046

or at 300 MHz not shown) systematically show DQ cross-peaks which can only be attributed 1047

to H1rsquo-H1 interactions (δH= 432-356) The presence of these H1rsquo-H1 interactions can only be 1048

explained by a head-tail arrangement of the G-C181-OH molecules as illustrated below the 1049

DQ-SQ map in Figure 9b 1050

9

965

966

Figure S 7 ndash ssNMR MAS (10 kHz) experiments recorded on a G-C181-OH dry powder a) 1H spectra 967

recorded from 110degC to 70degC b) 13C single pulse (SP) spectra recorded with high power 1H decoupling from 968

110degC to 70degC c) 13C cross polarization (CP) spectra recorded at 110degC and contact time tc of 50 micros and 969

3000 micros d) 13C CP spectra recorded at 70degC and for 20 lt tc micros lt 3000 970

971

1

811

910

CH2 CH2

1rsquo

6rsquo

8 6 4 2 0 ppm

110

90

80

70

910

1rsquo

6rsquo16

1

2rsquo3rsquo4rsquo

811

OH

45 40 35 30 25 20 15

ppm

200 180 160 140 120 100 80 60 40 20 0

ppm140 120 100 80

ppm

CH2-gauche

3004

110

90

80

70

65

60

50

25

C91013003

C9 - 12869

CH2-all-trans

3198

C1rsquo10335

a) b)

1H

T degCT degC

13C Single Pulse MAS

A

B

A

B

70degC70degC

5rsquo

215

200 180 160 140 120 100 80 60 40 20 0

ppm

13C CPMAST= 110degC

200 180 160 140 120 100 80 60 40 20 0

ppm

132 131 130 129 128 127 126

ppm

106 105 104 103 102

ppm

13C CPMAST= 70degC

50

tc micros

3000

20

tc micros

50 500 1000 3000

tc micros

C102

13024 C910113003

C92

12869

C1rsquo213082

C1rsquo110443

A B

106 105 104 103 102

ppm

3000

1000

500

50

20

AB

c)

d)

10

972

Figure S 8 ndash a) Short and d) long T2 components displayed by the 1H transverse relaxation signal c) 1H 973

transverse relaxation signals recorded on a G-C181-OH dry powder from 110degC to 70degC The area of each 974

given resonance is plotted (log-lin) as a function of the echo time (CPMG pulse sequence) for each 975

temperature and fitted using the equation given above T2s and T2l indicate the short and long T2 relaxation 976

times respectively obtained by describing the 1H transverse relaxation signal using a biexponential decay 977

For H1rsquo H5rsquo and H6rsquo the 1H relaxation function could be described by a monoexponential decay 978

- Protons from the sugar group (H1rsquo H5rsquo H6rsquo) show a single T2 relaxation component (Figure 979

6c and Figure S 8) while protons of the acyl chain (CH2 H910 H811) show both short (Figure 980

6c) and long (Figure 6d) T2 relaxation component 981

- The short T2 relaxation time systematically decreases from 110degC to 70degC for all species by a 982

factor ranging from two to seven depending on the specific chemical group Very long T2 are 983

observed at 110degC for all acyl species while nearly unchanged values are measured below 984

90degC without any clear trend 985

- The sugar protons (H1rsquo H5rsquo H6rsquo) have the shortest set of T2 values (lt 1 ms) below 100degC 986

The long component of T2 generally ranges between 4 ms and 12 ms 987

0 1 2 3 4 5 6 7 8

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1 2 3 4 5 6 7 8

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

H1

H8H11

H9H10

CH2

H1rsquo

H6rsquo

CH2

H910H811 H1CH2

H1rsquo H5rsquo6rsquo

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

OH

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45 T= 110degC

T= 90degC

T= 80degC

T= 70degC

Log

In

ten

sity

Time ms

Fit

)

CH2 H910 H811 H1 H56 H1 OH

00

05

10

15

20

25

30

35

40

T2 m

s

T= 110degC

T= 90degC

T= 80degC

T= 70degC

CH2 H910 H811 H1

0

2

4

6

8

10

12

T2 m

s

T= 110degC

T= 90degC

T= 80degC

T= 70degC

Short T2(1H) Long T2(1H)

a) b)

c)

11

Table S 1 ndash 1H chemical shifts of G-C181-OH in solution and in the solid-state Solid-state spectra are 988

recorded at a MAS frequency of 10 kHz in DMSO-d6 broad peak 989

990

δH ppm

Group Ndeg Liquid

(MeOD-d4) Solid

(T= 110degC) Solid

(T= 70degC)

CH2 H1 354 357

CH2 H2 153 156

CH2 H3-712-14 132 135

CH2 H811 203 203

CH H910 534 534

CH2 H15 162 164

CH2 H16 388 383

CH H1 424 432

CH H2 316 335

CH H3 334 335

CH H4 329 335

CH H5 354 351 357

CH2 H6 366 383

OH OH1 432

45 463

OH OH2 492

OH OH3 490

OH OH4 487

OH OH6 445

991

992

H1H9H10H1rsquo

H6rsquo

OH2rsquo

OH1

12

Detailed discussion about the NOESY and DQ-SQ 2D 1H-1H ssNMR experiments presented in 993

Figure 9 994

995

2D 1H-1H ssNMR experiments are a unique tool to explore through-space proximities 996

between 1H nuclei ssNMR offers a large number of possibilities each adapted to a given 997

system The advantage of 2D maps with respect to a dynamics approach is their direct 998

readability in clear two nuclei which exchange their magnetization through-space (whichever 999

mechanism may be responsible for this exchange) will display a specific cross-peak However 1000

the way cross-peaks are presented and interpreted strictly depends on the nature of the pulse 1001

program which in turn is chosen according to the physical nature of the sample 1002

1D temperature-dependent 1H NMR experiments performed on G-C181-OH (Figure 1003

4a in the main text) indicate sharp peaks at 90degC and broad peaks at 70degC Narrow peaks 1004

indicate fast andor large-amplitude molecular tumbling averaging the strong 1H-1H dipolar 1005

interactions This excludes the use of pulse sequences like DQ-SQ which is not efficient under 1006

these conditions For this reason we employ spin-diffusion-based pulse sequences like 1007

NOESY which is more adapted for samples with high molecular mobility When mixing times 1008

are chosen to be short (lt 1 s) one generally explores through-space proximities within about 5 1009

10 Aring On the contrary when molecular motions are restricted dipolar coupling becomes 1010

significant 1H NMR peaks are broader and pulse sequences like DQ-SQ can now be used 1011

with a spatial resolution of less than 5 Aring 1012

Whenever possible DQ-SQ experiments are preferred over spin diffusion-based pulse 1013

sequences because they are more selective in particular for likewise interactions which are 1014

not discriminated in NOESY-type experiments This is illustrated by the typical interaction 1015

schemes for NOESY (top Figure 9a) and DQ-SQ (top Figure 9b) For two given A and B 1016

protons 2D NOESY maps will display two on-diagonal and two square-shaped symmetrical 1017

off-diagonal cross-peaks Off-diagonal cross-peaks illustrate the mutual A-B and B-A 1018

interaction but the on-diagonal peaks are systematic and do not prove any A-A or B-B 1019

interaction Similarly to NOESY 2D DQ-SQ maps display off-diagonal peaks (symmetric 1020

across the diagonal top Figure 9b) upon A-B and B-A interaction but in contrast they will 1021

display on-diagonal peaks only in the presence of self-interaction In the tutorial example of 1022

Figure 9b top the on-diagonal A-A peak indicates mutual A-A interaction while lack of on-1023

diagonal B-B peak evidences the absence of B-B proximities Combining NOESY and DQ-SQ 1024

experiments is capital to study the local arrangement of G-C181-OH and its evolution with 1025

temperature 1026

13

At T= 90degC the H1 and H5rsquo contributions are very close but possible to discriminate 1027

them (δ(H1)= 354 ppm δ(H5rsquo)= 351 ppm) according to 1H NMR chemical shifts in Table S 1028

1 2D 1H-1H NOESY in Figure 9a also discriminates between the two components attributed 1029

on the basis of the interactions with their mutual neighbors (green H1-H2 and H5rsquo-H1rsquo squares 1030

in Figure 9a) On this basis 2D NOESY does not show a direct H1rsquo-H1 correlation (blue red-1031

crossed) but rather an extensive network of intra-sugar interactions These features exclude 1032

suggest intra- and inter-sugar interactions and they exclude head-tail arrangements as a major 1033

molecular configuration at 90degC 1034

At 70degC the 1H NMR peaks are broader indicating a reduction of the T2 (1H) relaxation 1035

times andor residual homonuclear dipolar interactions excluding the use of NOESY In this 1036

case 1H-1H DQ-SQ correlation experiments are best suited Figure 9b presents the 1H-1H DQ-1037

SQ experiment on G-C181-OH at 70degC Peaks are broader than in NOESY and attribution may 1038

be cumbersome However the experiment is performed at 700 MHz and MAS= 22 kHz which 1039

improves spectral resolution with respect to the same experiment performed at 300 MHz and 1040

MAS= 10 kHz The auto-peak at δH= 356 ppm in the SQ dimension is attributed to H1-H1 spin 1041

pairs mainly resulting from intra CH2 interaction while the DQ cross-peaks at δH= 156-356 1042

ppm and δH= 432-356 ppm respectively origin from H2-H1 and H1rsquo-H1 interactions (Figure 1043

9b) DQ-SQ map does not show any auto-peak corresponding to H1rsquo-H1rsquo or Hxrsquo-Hxrsquo (x= 2rsquo 1044

3rsquo 4rsquo) interactions (red-crossed blue Figure 9b) thus excluding any proximity (lt 5 Aring) between 1045

close glucose neighbors On the contrary DQ-SQ maps (recorded both at 700 MHz Figure 9b 1046

or at 300 MHz not shown) systematically show DQ cross-peaks which can only be attributed 1047

to H1rsquo-H1 interactions (δH= 432-356) The presence of these H1rsquo-H1 interactions can only be 1048

explained by a head-tail arrangement of the G-C181-OH molecules as illustrated below the 1049

DQ-SQ map in Figure 9b 1050

10

972

Figure S 8 ndash a) Short and d) long T2 components displayed by the 1H transverse relaxation signal c) 1H 973

transverse relaxation signals recorded on a G-C181-OH dry powder from 110degC to 70degC The area of each 974

given resonance is plotted (log-lin) as a function of the echo time (CPMG pulse sequence) for each 975

temperature and fitted using the equation given above T2s and T2l indicate the short and long T2 relaxation 976

times respectively obtained by describing the 1H transverse relaxation signal using a biexponential decay 977

For H1rsquo H5rsquo and H6rsquo the 1H relaxation function could be described by a monoexponential decay 978

- Protons from the sugar group (H1rsquo H5rsquo H6rsquo) show a single T2 relaxation component (Figure 979

6c and Figure S 8) while protons of the acyl chain (CH2 H910 H811) show both short (Figure 980

6c) and long (Figure 6d) T2 relaxation component 981

- The short T2 relaxation time systematically decreases from 110degC to 70degC for all species by a 982

factor ranging from two to seven depending on the specific chemical group Very long T2 are 983

observed at 110degC for all acyl species while nearly unchanged values are measured below 984

90degC without any clear trend 985

- The sugar protons (H1rsquo H5rsquo H6rsquo) have the shortest set of T2 values (lt 1 ms) below 100degC 986

The long component of T2 generally ranges between 4 ms and 12 ms 987

0 1 2 3 4 5 6 7 8

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1 2 3 4 5 6 7 8

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

H1

H8H11

H9H10

CH2

H1rsquo

H6rsquo

CH2

H910H811 H1CH2

H1rsquo H5rsquo6rsquo

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45

Log

In

ten

sity

Time ms

OH

0 1000 2000 3000 4000 5000 6000 7000 8000

00

05

10

15

20

25

30

35

40

45 T= 110degC

T= 90degC

T= 80degC

T= 70degC

Log

In

ten

sity

Time ms

Fit

)

CH2 H910 H811 H1 H56 H1 OH

00

05

10

15

20

25

30

35

40

T2 m

s

T= 110degC

T= 90degC

T= 80degC

T= 70degC

CH2 H910 H811 H1

0

2

4

6

8

10

12

T2 m

s

T= 110degC

T= 90degC

T= 80degC

T= 70degC

Short T2(1H) Long T2(1H)

a) b)

c)

11

Table S 1 ndash 1H chemical shifts of G-C181-OH in solution and in the solid-state Solid-state spectra are 988

recorded at a MAS frequency of 10 kHz in DMSO-d6 broad peak 989

990

δH ppm

Group Ndeg Liquid

(MeOD-d4) Solid

(T= 110degC) Solid

(T= 70degC)

CH2 H1 354 357

CH2 H2 153 156

CH2 H3-712-14 132 135

CH2 H811 203 203

CH H910 534 534

CH2 H15 162 164

CH2 H16 388 383

CH H1 424 432

CH H2 316 335

CH H3 334 335

CH H4 329 335

CH H5 354 351 357

CH2 H6 366 383

OH OH1 432

45 463

OH OH2 492

OH OH3 490

OH OH4 487

OH OH6 445

991

992

H1H9H10H1rsquo

H6rsquo

OH2rsquo

OH1

12

Detailed discussion about the NOESY and DQ-SQ 2D 1H-1H ssNMR experiments presented in 993

Figure 9 994

995

2D 1H-1H ssNMR experiments are a unique tool to explore through-space proximities 996

between 1H nuclei ssNMR offers a large number of possibilities each adapted to a given 997

system The advantage of 2D maps with respect to a dynamics approach is their direct 998

readability in clear two nuclei which exchange their magnetization through-space (whichever 999

mechanism may be responsible for this exchange) will display a specific cross-peak However 1000

the way cross-peaks are presented and interpreted strictly depends on the nature of the pulse 1001

program which in turn is chosen according to the physical nature of the sample 1002

1D temperature-dependent 1H NMR experiments performed on G-C181-OH (Figure 1003

4a in the main text) indicate sharp peaks at 90degC and broad peaks at 70degC Narrow peaks 1004

indicate fast andor large-amplitude molecular tumbling averaging the strong 1H-1H dipolar 1005

interactions This excludes the use of pulse sequences like DQ-SQ which is not efficient under 1006

these conditions For this reason we employ spin-diffusion-based pulse sequences like 1007

NOESY which is more adapted for samples with high molecular mobility When mixing times 1008

are chosen to be short (lt 1 s) one generally explores through-space proximities within about 5 1009

10 Aring On the contrary when molecular motions are restricted dipolar coupling becomes 1010

significant 1H NMR peaks are broader and pulse sequences like DQ-SQ can now be used 1011

with a spatial resolution of less than 5 Aring 1012

Whenever possible DQ-SQ experiments are preferred over spin diffusion-based pulse 1013

sequences because they are more selective in particular for likewise interactions which are 1014

not discriminated in NOESY-type experiments This is illustrated by the typical interaction 1015

schemes for NOESY (top Figure 9a) and DQ-SQ (top Figure 9b) For two given A and B 1016

protons 2D NOESY maps will display two on-diagonal and two square-shaped symmetrical 1017

off-diagonal cross-peaks Off-diagonal cross-peaks illustrate the mutual A-B and B-A 1018

interaction but the on-diagonal peaks are systematic and do not prove any A-A or B-B 1019

interaction Similarly to NOESY 2D DQ-SQ maps display off-diagonal peaks (symmetric 1020

across the diagonal top Figure 9b) upon A-B and B-A interaction but in contrast they will 1021

display on-diagonal peaks only in the presence of self-interaction In the tutorial example of 1022

Figure 9b top the on-diagonal A-A peak indicates mutual A-A interaction while lack of on-1023

diagonal B-B peak evidences the absence of B-B proximities Combining NOESY and DQ-SQ 1024

experiments is capital to study the local arrangement of G-C181-OH and its evolution with 1025

temperature 1026

13

At T= 90degC the H1 and H5rsquo contributions are very close but possible to discriminate 1027

them (δ(H1)= 354 ppm δ(H5rsquo)= 351 ppm) according to 1H NMR chemical shifts in Table S 1028

1 2D 1H-1H NOESY in Figure 9a also discriminates between the two components attributed 1029

on the basis of the interactions with their mutual neighbors (green H1-H2 and H5rsquo-H1rsquo squares 1030

in Figure 9a) On this basis 2D NOESY does not show a direct H1rsquo-H1 correlation (blue red-1031

crossed) but rather an extensive network of intra-sugar interactions These features exclude 1032

suggest intra- and inter-sugar interactions and they exclude head-tail arrangements as a major 1033

molecular configuration at 90degC 1034

At 70degC the 1H NMR peaks are broader indicating a reduction of the T2 (1H) relaxation 1035

times andor residual homonuclear dipolar interactions excluding the use of NOESY In this 1036

case 1H-1H DQ-SQ correlation experiments are best suited Figure 9b presents the 1H-1H DQ-1037

SQ experiment on G-C181-OH at 70degC Peaks are broader than in NOESY and attribution may 1038

be cumbersome However the experiment is performed at 700 MHz and MAS= 22 kHz which 1039

improves spectral resolution with respect to the same experiment performed at 300 MHz and 1040

MAS= 10 kHz The auto-peak at δH= 356 ppm in the SQ dimension is attributed to H1-H1 spin 1041

pairs mainly resulting from intra CH2 interaction while the DQ cross-peaks at δH= 156-356 1042

ppm and δH= 432-356 ppm respectively origin from H2-H1 and H1rsquo-H1 interactions (Figure 1043

9b) DQ-SQ map does not show any auto-peak corresponding to H1rsquo-H1rsquo or Hxrsquo-Hxrsquo (x= 2rsquo 1044

3rsquo 4rsquo) interactions (red-crossed blue Figure 9b) thus excluding any proximity (lt 5 Aring) between 1045

close glucose neighbors On the contrary DQ-SQ maps (recorded both at 700 MHz Figure 9b 1046

or at 300 MHz not shown) systematically show DQ cross-peaks which can only be attributed 1047

to H1rsquo-H1 interactions (δH= 432-356) The presence of these H1rsquo-H1 interactions can only be 1048

explained by a head-tail arrangement of the G-C181-OH molecules as illustrated below the 1049

DQ-SQ map in Figure 9b 1050

11

Table S 1 ndash 1H chemical shifts of G-C181-OH in solution and in the solid-state Solid-state spectra are 988

recorded at a MAS frequency of 10 kHz in DMSO-d6 broad peak 989

990

δH ppm

Group Ndeg Liquid

(MeOD-d4) Solid

(T= 110degC) Solid

(T= 70degC)

CH2 H1 354 357

CH2 H2 153 156

CH2 H3-712-14 132 135

CH2 H811 203 203

CH H910 534 534

CH2 H15 162 164

CH2 H16 388 383

CH H1 424 432

CH H2 316 335

CH H3 334 335

CH H4 329 335

CH H5 354 351 357

CH2 H6 366 383

OH OH1 432

45 463

OH OH2 492

OH OH3 490

OH OH4 487

OH OH6 445

991

992

H1H9H10H1rsquo

H6rsquo

OH2rsquo

OH1

12

Detailed discussion about the NOESY and DQ-SQ 2D 1H-1H ssNMR experiments presented in 993

Figure 9 994

995

2D 1H-1H ssNMR experiments are a unique tool to explore through-space proximities 996

between 1H nuclei ssNMR offers a large number of possibilities each adapted to a given 997

system The advantage of 2D maps with respect to a dynamics approach is their direct 998

readability in clear two nuclei which exchange their magnetization through-space (whichever 999

mechanism may be responsible for this exchange) will display a specific cross-peak However 1000

the way cross-peaks are presented and interpreted strictly depends on the nature of the pulse 1001

program which in turn is chosen according to the physical nature of the sample 1002

1D temperature-dependent 1H NMR experiments performed on G-C181-OH (Figure 1003

4a in the main text) indicate sharp peaks at 90degC and broad peaks at 70degC Narrow peaks 1004

indicate fast andor large-amplitude molecular tumbling averaging the strong 1H-1H dipolar 1005

interactions This excludes the use of pulse sequences like DQ-SQ which is not efficient under 1006

these conditions For this reason we employ spin-diffusion-based pulse sequences like 1007

NOESY which is more adapted for samples with high molecular mobility When mixing times 1008

are chosen to be short (lt 1 s) one generally explores through-space proximities within about 5 1009

10 Aring On the contrary when molecular motions are restricted dipolar coupling becomes 1010

significant 1H NMR peaks are broader and pulse sequences like DQ-SQ can now be used 1011

with a spatial resolution of less than 5 Aring 1012

Whenever possible DQ-SQ experiments are preferred over spin diffusion-based pulse 1013

sequences because they are more selective in particular for likewise interactions which are 1014

not discriminated in NOESY-type experiments This is illustrated by the typical interaction 1015

schemes for NOESY (top Figure 9a) and DQ-SQ (top Figure 9b) For two given A and B 1016

protons 2D NOESY maps will display two on-diagonal and two square-shaped symmetrical 1017

off-diagonal cross-peaks Off-diagonal cross-peaks illustrate the mutual A-B and B-A 1018

interaction but the on-diagonal peaks are systematic and do not prove any A-A or B-B 1019

interaction Similarly to NOESY 2D DQ-SQ maps display off-diagonal peaks (symmetric 1020

across the diagonal top Figure 9b) upon A-B and B-A interaction but in contrast they will 1021

display on-diagonal peaks only in the presence of self-interaction In the tutorial example of 1022

Figure 9b top the on-diagonal A-A peak indicates mutual A-A interaction while lack of on-1023

diagonal B-B peak evidences the absence of B-B proximities Combining NOESY and DQ-SQ 1024

experiments is capital to study the local arrangement of G-C181-OH and its evolution with 1025

temperature 1026

13

At T= 90degC the H1 and H5rsquo contributions are very close but possible to discriminate 1027

them (δ(H1)= 354 ppm δ(H5rsquo)= 351 ppm) according to 1H NMR chemical shifts in Table S 1028

1 2D 1H-1H NOESY in Figure 9a also discriminates between the two components attributed 1029

on the basis of the interactions with their mutual neighbors (green H1-H2 and H5rsquo-H1rsquo squares 1030

in Figure 9a) On this basis 2D NOESY does not show a direct H1rsquo-H1 correlation (blue red-1031

crossed) but rather an extensive network of intra-sugar interactions These features exclude 1032

suggest intra- and inter-sugar interactions and they exclude head-tail arrangements as a major 1033

molecular configuration at 90degC 1034

At 70degC the 1H NMR peaks are broader indicating a reduction of the T2 (1H) relaxation 1035

times andor residual homonuclear dipolar interactions excluding the use of NOESY In this 1036

case 1H-1H DQ-SQ correlation experiments are best suited Figure 9b presents the 1H-1H DQ-1037

SQ experiment on G-C181-OH at 70degC Peaks are broader than in NOESY and attribution may 1038

be cumbersome However the experiment is performed at 700 MHz and MAS= 22 kHz which 1039

improves spectral resolution with respect to the same experiment performed at 300 MHz and 1040

MAS= 10 kHz The auto-peak at δH= 356 ppm in the SQ dimension is attributed to H1-H1 spin 1041

pairs mainly resulting from intra CH2 interaction while the DQ cross-peaks at δH= 156-356 1042

ppm and δH= 432-356 ppm respectively origin from H2-H1 and H1rsquo-H1 interactions (Figure 1043

9b) DQ-SQ map does not show any auto-peak corresponding to H1rsquo-H1rsquo or Hxrsquo-Hxrsquo (x= 2rsquo 1044

3rsquo 4rsquo) interactions (red-crossed blue Figure 9b) thus excluding any proximity (lt 5 Aring) between 1045

close glucose neighbors On the contrary DQ-SQ maps (recorded both at 700 MHz Figure 9b 1046

or at 300 MHz not shown) systematically show DQ cross-peaks which can only be attributed 1047

to H1rsquo-H1 interactions (δH= 432-356) The presence of these H1rsquo-H1 interactions can only be 1048

explained by a head-tail arrangement of the G-C181-OH molecules as illustrated below the 1049

DQ-SQ map in Figure 9b 1050

12

Detailed discussion about the NOESY and DQ-SQ 2D 1H-1H ssNMR experiments presented in 993

Figure 9 994

995

2D 1H-1H ssNMR experiments are a unique tool to explore through-space proximities 996

between 1H nuclei ssNMR offers a large number of possibilities each adapted to a given 997

system The advantage of 2D maps with respect to a dynamics approach is their direct 998

readability in clear two nuclei which exchange their magnetization through-space (whichever 999

mechanism may be responsible for this exchange) will display a specific cross-peak However 1000

the way cross-peaks are presented and interpreted strictly depends on the nature of the pulse 1001

program which in turn is chosen according to the physical nature of the sample 1002

1D temperature-dependent 1H NMR experiments performed on G-C181-OH (Figure 1003

4a in the main text) indicate sharp peaks at 90degC and broad peaks at 70degC Narrow peaks 1004

indicate fast andor large-amplitude molecular tumbling averaging the strong 1H-1H dipolar 1005

interactions This excludes the use of pulse sequences like DQ-SQ which is not efficient under 1006

these conditions For this reason we employ spin-diffusion-based pulse sequences like 1007

NOESY which is more adapted for samples with high molecular mobility When mixing times 1008

are chosen to be short (lt 1 s) one generally explores through-space proximities within about 5 1009

10 Aring On the contrary when molecular motions are restricted dipolar coupling becomes 1010

significant 1H NMR peaks are broader and pulse sequences like DQ-SQ can now be used 1011

with a spatial resolution of less than 5 Aring 1012

Whenever possible DQ-SQ experiments are preferred over spin diffusion-based pulse 1013

sequences because they are more selective in particular for likewise interactions which are 1014

not discriminated in NOESY-type experiments This is illustrated by the typical interaction 1015

schemes for NOESY (top Figure 9a) and DQ-SQ (top Figure 9b) For two given A and B 1016

protons 2D NOESY maps will display two on-diagonal and two square-shaped symmetrical 1017

off-diagonal cross-peaks Off-diagonal cross-peaks illustrate the mutual A-B and B-A 1018

interaction but the on-diagonal peaks are systematic and do not prove any A-A or B-B 1019

interaction Similarly to NOESY 2D DQ-SQ maps display off-diagonal peaks (symmetric 1020

across the diagonal top Figure 9b) upon A-B and B-A interaction but in contrast they will 1021

display on-diagonal peaks only in the presence of self-interaction In the tutorial example of 1022

Figure 9b top the on-diagonal A-A peak indicates mutual A-A interaction while lack of on-1023

diagonal B-B peak evidences the absence of B-B proximities Combining NOESY and DQ-SQ 1024

experiments is capital to study the local arrangement of G-C181-OH and its evolution with 1025

temperature 1026

13

At T= 90degC the H1 and H5rsquo contributions are very close but possible to discriminate 1027

them (δ(H1)= 354 ppm δ(H5rsquo)= 351 ppm) according to 1H NMR chemical shifts in Table S 1028

1 2D 1H-1H NOESY in Figure 9a also discriminates between the two components attributed 1029

on the basis of the interactions with their mutual neighbors (green H1-H2 and H5rsquo-H1rsquo squares 1030

in Figure 9a) On this basis 2D NOESY does not show a direct H1rsquo-H1 correlation (blue red-1031

crossed) but rather an extensive network of intra-sugar interactions These features exclude 1032

suggest intra- and inter-sugar interactions and they exclude head-tail arrangements as a major 1033

molecular configuration at 90degC 1034

At 70degC the 1H NMR peaks are broader indicating a reduction of the T2 (1H) relaxation 1035

times andor residual homonuclear dipolar interactions excluding the use of NOESY In this 1036

case 1H-1H DQ-SQ correlation experiments are best suited Figure 9b presents the 1H-1H DQ-1037

SQ experiment on G-C181-OH at 70degC Peaks are broader than in NOESY and attribution may 1038

be cumbersome However the experiment is performed at 700 MHz and MAS= 22 kHz which 1039

improves spectral resolution with respect to the same experiment performed at 300 MHz and 1040

MAS= 10 kHz The auto-peak at δH= 356 ppm in the SQ dimension is attributed to H1-H1 spin 1041

pairs mainly resulting from intra CH2 interaction while the DQ cross-peaks at δH= 156-356 1042

ppm and δH= 432-356 ppm respectively origin from H2-H1 and H1rsquo-H1 interactions (Figure 1043

9b) DQ-SQ map does not show any auto-peak corresponding to H1rsquo-H1rsquo or Hxrsquo-Hxrsquo (x= 2rsquo 1044

3rsquo 4rsquo) interactions (red-crossed blue Figure 9b) thus excluding any proximity (lt 5 Aring) between 1045

close glucose neighbors On the contrary DQ-SQ maps (recorded both at 700 MHz Figure 9b 1046

or at 300 MHz not shown) systematically show DQ cross-peaks which can only be attributed 1047

to H1rsquo-H1 interactions (δH= 432-356) The presence of these H1rsquo-H1 interactions can only be 1048

explained by a head-tail arrangement of the G-C181-OH molecules as illustrated below the 1049

DQ-SQ map in Figure 9b 1050

13

At T= 90degC the H1 and H5rsquo contributions are very close but possible to discriminate 1027

them (δ(H1)= 354 ppm δ(H5rsquo)= 351 ppm) according to 1H NMR chemical shifts in Table S 1028

1 2D 1H-1H NOESY in Figure 9a also discriminates between the two components attributed 1029

on the basis of the interactions with their mutual neighbors (green H1-H2 and H5rsquo-H1rsquo squares 1030

in Figure 9a) On this basis 2D NOESY does not show a direct H1rsquo-H1 correlation (blue red-1031

crossed) but rather an extensive network of intra-sugar interactions These features exclude 1032

suggest intra- and inter-sugar interactions and they exclude head-tail arrangements as a major 1033

molecular configuration at 90degC 1034

At 70degC the 1H NMR peaks are broader indicating a reduction of the T2 (1H) relaxation 1035

times andor residual homonuclear dipolar interactions excluding the use of NOESY In this 1036

case 1H-1H DQ-SQ correlation experiments are best suited Figure 9b presents the 1H-1H DQ-1037

SQ experiment on G-C181-OH at 70degC Peaks are broader than in NOESY and attribution may 1038

be cumbersome However the experiment is performed at 700 MHz and MAS= 22 kHz which 1039

improves spectral resolution with respect to the same experiment performed at 300 MHz and 1040

MAS= 10 kHz The auto-peak at δH= 356 ppm in the SQ dimension is attributed to H1-H1 spin 1041

pairs mainly resulting from intra CH2 interaction while the DQ cross-peaks at δH= 156-356 1042

ppm and δH= 432-356 ppm respectively origin from H2-H1 and H1rsquo-H1 interactions (Figure 1043

9b) DQ-SQ map does not show any auto-peak corresponding to H1rsquo-H1rsquo or Hxrsquo-Hxrsquo (x= 2rsquo 1044

3rsquo 4rsquo) interactions (red-crossed blue Figure 9b) thus excluding any proximity (lt 5 Aring) between 1045

close glucose neighbors On the contrary DQ-SQ maps (recorded both at 700 MHz Figure 9b 1046

or at 300 MHz not shown) systematically show DQ cross-peaks which can only be attributed 1047

to H1rsquo-H1 interactions (δH= 432-356) The presence of these H1rsquo-H1 interactions can only be 1048

explained by a head-tail arrangement of the G-C181-OH molecules as illustrated below the 1049

DQ-SQ map in Figure 9b 1050