Ultramicroscopy ] (]]]]) ]]]–]]]

Contents lists available at ScienceDirect

Ultramicroscopy

0304-39

doi:10.1

n Corr

Aarhus

fax: +4

E-m1 Pr

Univers2 Pr

Nationa

Tehran,

Pleas

journal homepage: www.elsevier.com/locate/ultramic

Immobilisation of living bacteria for AFM imaging underphysiological conditions

Rikke Louise Meyer a,b,n, Xingfei Zhou a,1, Lone Tang a,b, Ayyoob Arpanaei a,2, Peter Kingshott a,Flemming Besenbacher a,c

a Interdisciplinary Nanoscience Center (iNANO), Aarhus University, DK-8000 Aarhus C, Denmarkb Department of Biological Sciences, Aarhus University, DK-8000 Aarhus C, Denmarkc Department of Physics and Astronomy, Aarhus University, DK-8000 Aarhus C, Denmark

a r t i c l e i n f o

Article history:

Received 30 October 2009

Received in revised form

18 May 2010

Accepted 23 June 2010

Keywords:

Immobilisation

Gelatin

Polyethyleneimine

Cell-Tak

Poly-L-lysine

91/$ - see front matter & 2010 Elsevier B.V. A

016/j.ultramic.2010.06.010

esponding author at: Interdisciplinary Na

University, DK-8000 Aarhus C, Denmark. Tel.

5 8942 2722.

ail address: rikke.meyer@inano.au.dk (R. Loui

esent address: Department of Physics,

ity, Ningbo, China.

esent address: Industrial and Environmenta

l Institute of Genetic Engineering and Biotech

Iran.

e cite this article as: R. Louise Meye

a b s t r a c t

Atomic force microscopy (AFM) holds great potential for studying the nanoscale surface structures of

living cells, and to measure their interactions with abiotic surfaces, other cells, or specific biomolecules.

However, the application of AFM in microbiology is challenging due to the difficulty of immobilising

bacterial cells to a flat surface without changing the cell surface properties or cell viability. We have

performed an extensive and thorough study of how to functionalise surfaces in order to immobilise

living bacteria for AFM studies in liquid environments. Our aim was to develop a scheme which allows

bacterial cells to be immobilised to a flat surface with sufficient strength to avoid detachment during

the AFM scanning, and without affecting cell surface chemistry, structure, and viability. We compare

and evaluate published methods, and present a new, reproducible, and generally applicable scheme for

immobilising bacteria cells for an AFM imaging.

Bacterial cells were immobilised to modified glass surfaces by physical confinement of cells in

microwells, physisorption to positively charged surfaces, covalent binding to amine- or carboxyl-

terminated surfaces, and adsorption to surfaces coated with highly adhesive polyphenolic proteins

originating from the mussel Mytilus edulis. Living cells could be immobilised with all of these

approaches, but many cells detached when immobilised by electrostatic interactions and imaged in

buffers like PBS or MOPS. Cells were more firmly attached when immobilised by covalent binding,

although some cells still detached during AFM imaging. The most successful method revealed was

immobilisation by polyphenolic proteins, which facilitated firm immobilisation of the cells.

Furthermore, the cell viability was not affected by this immobilisation scheme, and adhesive proteins

thus provide a fast, reproducible, and generally applicable scheme for immobilising living bacteria for

an AFM imaging.

& 2010 Elsevier B.V. All rights reserved.

1. Introduction

Since its invention two decades ago, atomic force microscopy(AFM) has become a widely used technique for imaging thesurfaces of materials, coatings, biomolecules and entire cells atlength scale ranging from nanometer to microns. The advantage

ll rights reserved.

noscience Center (iNANO),

: +45 6020 2794;

se Meyer).

School of Science, Ningbo

l Biotechnology Department,

nology, P.O. Box: 14965/161,

r, et al., Ultramicroscopy (2

of an AFM over many other high-resolution imaging techniques,such as an electron microscopy, is that cells and biomolecules canbe imaged under in situ physiological conditions i.e. in liquid,without any staining. Imaging living cells with nanoscale resolu-tion is an exciting leap forward in the field of microscopy, as itenables detailed studies of dynamics events, such as cell growth[1,2] and morphological responses to external stimuli or stresses[3]. As the design of AFM instruments have improved toaccommodate the handling of biological samples, application ofAFM in life science areas has increased dramatically [4]. Althoughfirst introduced as an imaging technique, the power of an AFMas an analytical tool for measuring intermolecular interactionforces with high sensitivity has become evident. For example,chemical force microscopy can resolve the chemical propertiesand interactions of living bacteria with nanoscale resolution,and this method has for instance been used to study the

010), doi:10.1016/j.ultramic.2010.06.010

R. Louise Meyer et al. / Ultramicroscopy ] (]]]]) ]]]–]]]2

interactions between the bacterial cell surface and antimicrobialdrugs [5–7].

In microbiology, an AFM imaging has provided detailedstructural information about the bacterial cell surface [8,9]. Manystudies have dealt with dried bacteria, as sample preparation inthis case is easy, and the resolution obtained when imaging in airis in general higher than when imaging is carried out in liquidenvironments [10]. However, the most exciting applications ofAFM in microbiology lie when the imaging of living cells is carriedunder in liquid environments, where interactions betweenbacteria and their surroundings can be studied. The cell–surfaceinteractions involved in bacterial adhesion, cell–cell interactionsoccurring during biofilm formation, and cell–host interactionsresponsible for infections are just a few examples.

A prerequisite for an AFM analysis of bacteria is theimmobilisation of the cells to a flat substrate, and the failure toimmobilise cells sufficiently to withstand the lateral force exertedby the cantilever during scanning, continues to be a challenge.This is particularly problematic when imaging bacteria, as the sizeand shape of bacterial cells only provide a small contact areabetween the cell and the underlying substrate. Bacteria are oftenimmobilised by letting a drop of cell culture dry out on a surfacebefore re-hydrating and imaging in water or buffer [11,12].However, even transient drying of cells may affect cell viability ortrigger a biological response that alters the cell surface before theAFM imaging. The most commonly used immobilisation methodsthat do not require drying are physical entrapment in a porousmembrane [13] or immobilisation by electrostatic interactions tosurfaces that have been coated with positively charged substance,such as polyethyleneimine (PEI) [14], poly-L-lysine [11], orgelatine [15]. Physical entrapment in membranes is only suitablefor coccoid cells, and there are indications of mechanical stressbeing exerted by the entrapment process [16]. While electrostaticimmobilisation does not exert mechanical stress, the strength ofthe interaction is weakened at high ionic strength [17], makingimaging most successful in distilled water, which exerts osmoticstress on the cells. It has been shown that even rinsing cells indistilled water prior to imaging in buffer can destabilise extra-cellular structures, such as the bacterial capsule [18]. Currently nomethod exists for immobilising living bacteria for AFM studiesunder physiological conditions, which can be applied to cells withvarying size, shape, and surface properties. A robust, reproduciblescheme that exerts minimal chemical or mechanical stress duringsample preparation and the subsequent imaging is thus needed toexploit the full potential of an AFM in microbiology. Here, weprovide an overview of existing approaches to immobilise livingbacteria for an AFM analysis under physiological conditions, andpresent a new, easy, and generally applicable method forimmobilisation of bacteria based on highly adhesive polyphenolicproteins.

2. Materials and methods

2.1. Cell cultures

Staphylococcus sciuri (DSM 20345) was used as a Grampositive, coccid model organism to evaluate the differentimmobilisation techniques. Members of the S. sciuri group arewidespread and can be isolated from animals and animal products[19–21]. They form biofilms and constitute up to 4.3% ofcoagulase-negative Staphylococci isolated from clinical samplesand are frequently found in wound infections [22,23]. Immobi-lisation to Cell-TakTM was further evaluated for cells with othercell morphologies and surface properties. The bacteria used forthis purpose were the Gram positive Bacillus subtilis (DSM 10), the

Please cite this article as: R. Louise Meyer, et al., Ultramicroscopy (2

Gram negative Escherichia coli (DSM 498), and Mycobacterium sp.(donated by Dr. K. Finster, Aarhus University), which is Grampositive, but has a characteristic hydrophobic cell wall rich inmycolic acids. Cell suspensions were prepared by inoculating cellsfrom agar plates into 50 mL 5 g/l Tryptic Soy Broth and growingthem overnight at room temperature, with shaking at 120 rpm.Cells were harvested by centrifugation at 5000g for 5 min, washedthree times in PBS, and finally resuspended in PBS or deionisedwater (depending on the subsequent immobilisation technique).For all immobilisation techniques, we found it important to washthe cells thoroughly, and to not resuspend them to an opticaldensity at 600 nm of more than 0.5.

2.2. Immobilisation of bacteria by physical confinement

Glass microscope slides were cleaned by sonication in acetone,alcohol, and deionised water for 10 min each. This cleaningmethod was employed for all microscope slides before any of thesurface modification methods applied in this study.

The principle of immobilisation of bacteria by physicalconfinement was the random capture of cells in microwellsmade by colloid lithography (Fig. 1a). A colloidal monolayermask was assembled on the substrate by electrostatic selfassembly. This approach included positive charging of thesubstrate surface by an electrostatic deposition of sequentiallayers of polyelectrolytes: (1) poly(diallyldimethylammoniumchloride) (PDDA, MW 200,000–350,000, Sigma Aldrich), (2)poly(sodium 4-styrenesulfonate) (PSS, MW 70,000, SigmaAldrich), and (3) aluminium chloride hydroxide (ACH, Reheis).A solution of 3 mm sulphate-modified polystyrene colloids(IDC) was deposited on the surface, and a dispersed colloidalmonolayer was obtained after drying with compressed nitrogen.Thin gold films (about 600–700 nm) were then deposited byelectron-beam evaporation. Finally, tape was gently placed onthe thin gold films, and the dispersed colloids on the glasssubstrate were removed by pulling off the tape, exposing themicro-wells once colloids were removed [24,25].

Bacteria were immobilised in the microwells by placing a dropof S. sciuri in PBS on the slide and incubating for 20 min beforewashing gently with PBS. Samples were kept submerged in PBSuntil AFM imaging was performed in PBS.

2.3. Immobilisation of bacteria by physicochemical interactions and

covalent binding of cells

2.3.1. Preparation of surfaces

Glass surfaces were modified with either carboxylic acid or aminegroups before immobilising cells to the surface with physisorptionor covalent binding (Table 1, Fig. 1b–e). Five different approacheswere used to coat glass surfaces with amine groups. Surfaces wereeither treated by N-(2-aminoethyl)-3-aminopropyltrimethoxy-silane(EDS, Sigma) or 3-aminopropyltriethoxysilane (APTES, Sigma) to formself assembled monolayers, or coated with polyethyleneimine (PEI,Sigma), poly-L-lysine (PLL, Sigma) or gelatin (Sigma).

EDS was added by placing slides in 1% (v/v) solution of an EDScontaining 94% methanol, 1% acetic acid, and 4% distilled water for30 min, shaking occasionally (a modified method from Vermetteet al. [26]). EDS-coated slides were then washed three times withmethanol, heated for 1 min at 1201C, washed with deionisedwater, and dried under a nitrogen stream.

APTES was applied by an immersion of slides in 5% APTES inethanol for 1 h at room temperature. Slides were then washedwith ethanol and distilled water before drying under a nitrogenstream.

010), doi:10.1016/j.ultramic.2010.06.010

Table 1Methods used for chemical modification of glass surfaces.

Coating Chemical group Used in immobilisation by physisorption Used in covalent immobilisation

EDS amine X X

APTES amine X

PEI amine X

PLL amine X X

Gelatin mixed X

PolysineTM slides amine X X

PAA COOH X

Fig. 1. Schematic representation of the immobilisation methods employed. (a) physical confinement by capture in microwells, (b) attractive electrostatic interactions,

(c) covalent binding to amine-functionalised surfaces by EDC–NHS, (d) covalent binding to carboxyl-functionalised surfaces by EDC–NHS, (e) covalent binding to amine-

functionalised surfaces by glutaraldehyde, and (f) attachment to polyphenolic adhesive protein from the mussel Mytilus edulis.

R. Louise Meyer et al. / Ultramicroscopy ] (]]]]) ]]]–]]] 3

Coating with polyethyleneimine (PEI) was obtained by im-mersing PAA-coated glass slides (see below) for 15 min in PBSwith 1 mg/ml 1-ethyl-3-(3-dimethylaminopropyl) carbodiimidehydrochloride (EDC, Sigma) and 0.1 mg/ml N-hydroxysuccinimide(NHS, Sigma), and then transferring slides to a solution of 1 mg/mlPEI in PBS for 30 min, while shaking occasionally. Finally, PEI-coated slides were rinsed with PBS and deionised water beforedrying under a nitrogen stream.

Two procedures were used to coat glass slides with poly-L-lysine. In the first procedure, PAA-coated glass slides (see below)were immersed in PBS with 1 mg/ml EDC and 0.1 mg/ml NHS for15 min. Slides were then transferred to 1% (w/v) PLL in PBS andincubated for 30 min, while shaking and finally washed with PBS,and then deionised water. In the second procedure, PEI-coatedsurfaces were immersed in 2% (v/v) glutaraldehyde overnight andtransferred to 1% (w/v) PLL in PBS for 30 min, while shaking.Finally, slides were washed with PBS and deionised water beforedrying under stream of nitrogen.

Gelatin coated slides were prepared by dissolving 0.5 g gelatin(Sigma) in 100 ml distilled water at 90 1C and cooling down to60 1C before briefly immersing glass microscope slides in thissolution and air drying overnight by standing the slides upright ina laminar flow bench.

Please cite this article as: R. Louise Meyer, et al., Ultramicroscopy (2

Functionalisation with carboxylic acid (COOH) groups wasobtained by binding polyacrylic acid (PAA) covalently to the EDSamine groups on the previously prepared surfaces using amodified version of the method suggested by Vermette et al.[26]. A solution of 1 mg/ml PAA (MW 250,000, Sigma) in PBS wasprepared. EDC and NHS were then added to a final concentrationof 1 and 0.1 mg/ml, respectively, and the amine-coated glassslides were submerged in this solution immediately. The beakerwas shaken for the first hour, and then kept overnight. Finally, thesurfaces were rinsed repeatedly with MilliQ water and dried by anitrogen stream.

2.3.2. Immobilisation of bacteria by physicochemical interactions

S. sciuri suspended in distilled water or PBS were immobilisedby electrostatic interactions to the EDS, APTES, PEI, PLL, andgelatin coated slides plus commercial microscope slides chemi-cally modified to resemble poly-L-lysine (PolysineTM, Menzel). Adrop (0.1–0.5 mL) of cell suspension was placed on the modifiedglass substrates and incubated for 20–30 min (without dryingout), before rinsing gently with PBS and storing in PBS until anAFM imaging. Cells were imaged first in PBS and thereafter indistilled water.

010), doi:10.1016/j.ultramic.2010.06.010

R. Louise Meyer et al. / Ultramicroscopy ] (]]]]) ]]]–]]]4

2.3.3. Immobilisation of bacteria by covalent binding

Covalent binding was achieved in two ways: (1) by linkingamine and carboxyl groups with EDC–NHS (Fig. 1c+d), or (2) bylinking amine to amine groups with glutaraldehyde (Fig. 1e).

EDC–NHS was used to covalently bind cells to glass slidesmodified with carboxyl groups (PAA coated slides) and amine groups(EDS, PLL, or PolysineTM). EDC (1 mg/ml) and NHS (0.1 mg/ml) wasdissolved in PBS and immediately used for resuspending bacteria thathad been washed in PBS and centrifuged to a pellet. A drop of thissolution was then placed on the modified glass surfaces andincubated for 30 min before rinsing with PBS. Slides were stored inPBS until an AFM imaging.

Covalent binding with glutaraldehyde was obtained by placingslides in a 2% (v/v) glutaraldehyde overnight, and rinsingthoroughly before placing a drop of cells suspended in PBS onthe slide and incubating for 30 min. Slides were then gentlywashed with PBS and stored in PBS before AFM imaging indeionised water, PBS, 20 mM HEPES, and 20 mM MOPS buffers(all pH 7).

2.4. Immobilisation of bacteria with adhesive polyphenolic proteins

Cell-TakTM (BD Diagnostics) is a strongly adhesive polyphe-nolic protein extract from Mytilus edulisi, and we here use theproteins to facilitate a firm contact between bacterial cells and aglass surface (Fig. 1f). These adhesive proteins were applied to aglass surface by mixing 1 ml Cell-TakTM with 28.5 ml 0.1 M sodiumbicarbonate and 0.5 ml 1 M NaOH. The solution was immediatelytransferred to an area of approximately 1 cm2 on a circularcoverslip, and left for 20 min at room temperature. The surfacewas then submerged in deionised water, air-dried, subsequentlystored at 4 1C for up to two weeks before immobilisation ofbacteria. Cells were immobilised by adding a drop of the cellssuspended in PBS to the protein coated surface, incubating for5 min, and removing non-adhering bacteria by rinsing vigorouslywith the solution subsequently used for AFM imaging. The surfacewas immediately placed in a BioCellTM and imaged. In addition toS. sciuri, B. subtilis, E. coli and Mycobacterium sp. were also imagedin 20 mM MOPS.

2.5. AFM imaging

Cells immobilised by physical confinement, physisorption, orcovalent binding were imaged by an AFM in PBS and in deionisedwater. Cells immobilised on adsorbed protein layers were imagedin either deionised water, 2 g/l NaCl, PBS, 20 mM HEPES, 20 mMMOPS, or 5 g/l nutrient broth (all pH 7). When interactionsbetween the cantilever tip and the sample surface was too strongto allow use of intermittent contact mode, cells were imaged incontact mode.

A NanoWizardII atomic force microscope (JPK Instruments,Germany) combined with an inverted optical microscope (ZeissAxiovert 200 M, Zeiss, Germany) was used to record AFM imagesat 256 or 512 pixels per line, scanning at 0.2–2 Hz. Cells wereimaged in contact mode using Silicon nitride cantilevers (NP orNP-S, Veeco, USA) with a spring constant of 0.03 N/m, and Siliconcantilevers (CSC38/noAl, Mikromasch, USA) with a spring con-stant of 0.03 N/m. The Silicon cantilevers were taller and thereforeproduced less tip artifacts when imaging cells taller than 1.5 mm.Intermittent contact mode imaging was performed with Siliconcantilevers (NSC19/CrAu, Mikromasch, USA) with a spring con-stant of 0.6 N/m and a resonant frequency of 80 kHz in air (drivefrequency in liquid was 42 kHz). Cells were viewed by opticalmicroscopy simultaneously with an AFM imaging to assesswhether any cells were detached during scanning.

Please cite this article as: R. Louise Meyer, et al., Ultramicroscopy (2

2.6. Assessment of cell viability

The viability of immobilised cells was assessed by testing thecell membrane permeability using the Live/Dead BacLightTM

staining kit (Invitrogen) containing SYTO9 and propidium iodide.S. sciuri was harvested in the mid-exponential growth phase asdescribed above, washed three times in PBS, resuspended in PBS,and immobilised to surfaces as described above by (1) physisorp-tion to PolysineTM slides (Fig. 1b), (2) covalent binding toPolysineTM slides using EDC–NHS (Fig. 1c) or (3) covalent bindingto PolysineTM slides using glutaraldehyde (Fig. 1e), or(4) adsorption to Cell-TakTM coated slides (Fig. 1f). Slides wereincubated for 4 h in PBS before staining according to themanufacturer’s protocol. Samples were gently rinsed beforevisualization and image acquisition on a Zeiss Axiovert 200 Mepifluorescence microscope.

3. Results and discussion

Methods for immobilising bacteria from a suspension on asurface can roughly be divided into four categories: physicalconfinement, physisorption, covalent binding, and binding viaadhesive proteins (Table 2). We were able to immobilise cells in aphysically confined space by random capture in 0.5 mm deep and1.5 mm wide microwells (Fig. 2a and b). However, the success ofcell capture was not very reproducible and large areas had to bescanned to locate cells by an AFM. The frequency of capturingcells in the wells may be increased by modifying the chemistryinside wells to combine physical entrapment with electrostaticinteractions as demonstrated by Rowan et al. [27].

A general advantage of immobilising bacteria in holes or wellscompared to a flat substrate is that the cells do not fully protrudefrom these holes. Thereby tip artifacts caused by large sampleheight are minimised [1]. Kasas and Ikai [13] developed a simpleand elegant method for physical confinement of spherical cells byfiltering a cell suspension through a polycarbonate filter andtrapping cells in the filter pores. Unfortunately filtering exertsmechanical stress on the cells [16,28], and only spherical cellswith a cell diameter slightly larger than the filter pore size arecaptured. Although filtering is probably the most reproducibleand commonly used immobilisation scheme, it is best suited forGram-positive coccid bacteria, and less applicable to rod-shapedor filamentous Gram-negative bacteria. Random capture inmicrowells may exert minimal mechanical stress during immo-bilisation, but it does not circumvent the problem of not beingable to immobilise cells of variable size and shape. That can onlyoccur on a flat substrate.

On flat substrates bacteria were easily immobilised byphysisorption to chemically modified glass surfaces. As modelsystems, we have studied cells immobilised to gelatin- andAPTES-coated surfaces (Fig. 2c and d). The immobilisation ofbacteria to APTES or gelatin coated surfaces is thought to bemediated by a mixture of electrostatic and hydrophobic interac-tions. Gelatin possesses a number of different chemical groups,and hydrophobic interactions may also play a role here. Whileimaging of cells immobilised to APTES or gelatin coated surfaceswas relatively straight forward in deionised water, all cellsdetached when imaged in PBS (data not shown). Stukalov et al.[18] also observed detachment of electrostatically immobilisedcells during AFM imaging in buffer compared to deionised water.This observation is presumably due to weakening of electrostaticinteractions at higher ionic strength [17]. We used Type A gelatin,which has an isoelectric point of 7–9, and therefore carries a netpositive charge at lower pH. The net positive charge of gelatinmay thus have decreased as the buffer pH approached its

010), doi:10.1016/j.ultramic.2010.06.010

Table 2Overview of the methods used for immobilisation of bacteria for AFM analysis in liquid.

Method References Comments

1. Physical confinementCapture in polycarbonate filters [3,13,16,38–40] and others. Best suited for coccoid Gram positive cells.

Can cause deformation of cells.

No chemical modification of cells.

Random cell capture in microwells [1], This study Applicable to cells of a specific size range.

No deformation of cells.

No chemical modification of cells.

2. PhysisorptionGelatin coating [15,41], This study Applicable to cells of different shape and size.

No deformation of cells.

No chemical modification of cells.

EDS coating This study Immobilisation by electrostatic interactions is less

strong at high ionic strength.

APTES coating This study

PEI coating [14,42–44]

Poly-L-lysine or PolysineTM [18,45,46] This study

Poly-dopamine [32]

3. Covalent bindingEDC/NHS crosslinking of COOH groups to NH2 groups [28,47], This study Applicable to cells of different shape and size.

Glutaraldehyde crosslinking of NH2 groups [31,34,48–51] This study No deformation of cells.

Cells may become chemically modified if an EDC–NHS or glutaraldehyde

is added directly to the cell suspension.

4. Immobilisation with adhesive proteins

Adhesive proteins (Cell-TakTM) This study Applicable to cells of different shapes and sizes.

Poly-dopamine coating [32] No deformation of cells and no chemical modification.

Sugar-binding proteins (lectins) [52] Immobilisation with lectins is only applicable to cells containing

specific sugars on the cell surface.

R. Louise Meyer et al. / Ultramicroscopy ] (]]]]) ]]]–]]] 5

isoelectric point. Furthermore, it is not easy to control the tip–sample interactions in PBS. Strong tip–cell interactions andunintentional application of higher force under these conditionsmay also have exacerbated the problem. Immobilisation byphysisorption that involves electrostatic interactions is thereforebest suited for imaging at low ionic strength [29].

A much stronger interaction was obtained by covalentlylinking carboxyl or amine groups on the cell to carboxyl or aminegroups on the substrate. The methods used here linked carboxylgroups to amine groups using EDC/NHS (Fig. 1c and d) oramine groups to amine groups using glutaraldehyde (Fig. 1e).Both approaches successfully immobilised cells on amine mod-ified glass surfaces, and AFM imaging could be performed in PBS(Fig. 2e and f), HEPES, MOPS, and deionised water (data notshown). Tip artifacts were observed when imaging in PBS (Fig. 2eand f), as the cells were substantially higher when imaged in PBS(1.5–2 mm) compared to deionised water (1–1.5 mm). Suchtip artifacts are not unusual when the cell height exceeds1.5 mm [30].

Among the amine-functionalised surfaces, commercial Poly-sineTM glass slides were excellent substrates for covalent bindingof bacteria using glutaraldehyde or an EDC–NHS (Fig. 2e and f).Whereas cells rarely detached from this substrate, detachmentoccurred frequently from EDS-, APTES- and poly-L-lysine-coatedsurfaces. The difference in cell immobilisation efficiency betweenPolysineTM and the other substrates may be explained bydifference in conformation of the chemical groups, betterattachment of amine groups on the commercial slides, or a lowerdensity of amine groups on EDS- and APTES-coated surfaces. Thiscould however not be verified.

Covalent binding of bacteria to carboxyl groups on PAA-coatedglass slides did not result in successful immobilisation, and allcells detached during AFM imaging (data not shown). The reasonfor the failed immobilisation is not yet fully understood. It couldbe caused by low abundance of free amine groups on the cell

Please cite this article as: R. Louise Meyer, et al., Ultramicroscopy (2

surface, low abundance of carboxyl groups on the PAA coatedsubstrate, or that PAA was not efficiently immobilised to the glasssubstrate.

AFM analysis of living cells under physiological conditionsshould ideally be performed on cells that have not been stressedchemically or mechanically. Immobilisation of cells by EDC–NHSinevitably exerts some degree of chemical modification of the cellsurface, as EDC–NHS is mixed directly into the cell suspension andmay crosslink chemical groups on the cell surface. Activatingchemical groups on the glass substrate instead of the cells couldprovide an alternative approach to covalent linking, which onlyaffects cell surface chemistry on the part of the cell in contactwith the substrate. This was explored by activating amine groupson the substrate with glutaraldehyde, followed by removal ofunbound glutaraldehyde before addition of cells. Bacteria havepreviously been immobilised using glutaraldehyde to crosslinkamine groups on the cell surface to amine groups on a PEI coatedAFM cantilever [31]. However, the authors suspended bacteriadirectly in glutaraldehyde, leading to cross-linking of proteins onthe entire cell surface, which affects cell elasticity, adhesionproperties, and viability [14,32,33]. Sullivan [34] modified theapproach and activated APTES coated mica with glutaraldehydefor immobilisation of spheroplasts, and this inspired us to use asimilar approach to immobilise intact cells to amine-functiona-lised glass surfaces. This approach allowed successful immobilisa-tion (Fig. 2f), although some cells did detach during AFM imaging,most likely due to the low density of amine groups present on theAPTES coated glass, which limits the degree of cross-linking.

Highly efficient immobilisation of bacteria to glass wasmediated by adhesive proteins without the need for covalentbinding. The mussel Mytilus edulis excretes polyphenolic proteinsto adhere strongly to surfaces in the marine environment, and theunique properties of these proteins have gained attention as anefficient adhesive in aqueous conditions [35]. An extract of theseadhesive proteins are now available as a commercial product,

010), doi:10.1016/j.ultramic.2010.06.010

Fig. 2. An AFM image of S. sciuri immobilised by random capture in microwells, electrostatic interactions, and covalent binding. (a) Microwells before immobilisation of

cells, (b) two cells caught in a microwell (imaged in deionised water), (c) cells immobilised by physisorption to gelatin (imaged in deionised water), (d) cells immobilised by

physisorption to APTES (imaged in deionised water), (e) cells immobilised covalently to PolysineTM slides by EDC–NHS (imaged in PBS), (f) cells immobilised covalently to

PolysineTM slides by glutaraldehyde (imaged in PBS). Image height is 1 mm in (a), 0.5 mm in (b), and 3 mm in (c–f). Scalebar in all images¼5 mm.

R. Louise Meyer et al. / Ultramicroscopy ] (]]]]) ]]]–]]]6

Cell-TakTM, and bacteria immobilised on glass surfaces coatedwith these proteins could easily be imaged in deionised water,NaCl, physiological buffers (PBS, MOPS, and HEPES), and nutrientbroth (Fig. 3a–h). Imaging could be performed in contact mode aswell as intermittent contact mode (Fig. 3i). Consecutive scans ofthe same area showed that a 1–2 cells detached during each scanin e.g. MOPS buffer (Fig. 3d–f). Continuous scanning in nutrientbroth could be performed with only few cells detaching for thefirst 3 h (Fig. 3h), after which all cells suddenly detached (data notshown). The detachment after 3 h might be linked to cell growthor turnover of cell membrane proteins that could destabilise thecontact between the cell and the substrate.

Immobilisation of cells with variable morphology and surfaceproperties was investigated using Cell-TakTM, and it appeared thatboth Gram positive and Gram negative species could beimmobilised and imaged under physiological conditions(Fig. 4a–c). The long rods of B. subtilis and Mycobacterium sp.

were not always immobilised sufficiently, and only partial

Please cite this article as: R. Louise Meyer, et al., Ultramicroscopy (2

immobilisation of cells was recognised as distortion of theimage rather than detachment during imaging (Fig. 4b and c).

A potential draw-back of immobilisation with polyphenolicproteins is the risk of contaminating the cantilever with theseproteins during imaging. We did observe tip contamination fromtime to time, but this contamination may also arise from contactwith the cells, and it did not appear to occur more frequentlyduring imaging of cells immobilised by Cell-TakTM compared toother methods. Scanning of Cell-TakTM coated surfaces in theabsence of cells did not indicate tip contamination (data notshown). It should be noted that the adhesive properties of theprotein is pH dependent, and imaging at low pH may detach theprotein.

The adhesive properties of the proteins excreted by Mytilus

edulis are linked to the presence of the amino acid 3,4-dihydroxy-L-phenylalanine (DOPA), and Lee et al. [36] developed a biomi-metic adhesive polymer, poly-dopamine, inspired from thecomposition and properties of these proteins. Kang and Elimelech

010), doi:10.1016/j.ultramic.2010.06.010

Fig. 3. S. sciuri immobilised by an attachment to polyphenolic adhesive proteins and imaged in contact mode in (a) deionised water, (b) 2 g/L NaCl, (c) PBS, (d–f) 20 mM

MOPS, (g) 20 mM HEPES, and (h) 5 g/L nutrient broth. The images in MOPS (d–f) are three consecutive scans of the same area over a period of 30 min. Arrows indicate

detachment of cells. S. sciuri was also imaged in tapping mode in deionised water (i). Cantilevers NP-S were used for images (a, b, d, e, f), cantilever CSC38noAl was used for

images (c, g, h), and cantilever CSC19CrAu was used for the tapping mode image (i). The taller CSC cantilevers alleviated the tip artifacts seen when imaging S. sciuri in

buffer or NaCl with the NP-S cantilever. All images are 20�20 mm2, and the height is 2 mm.

Fig. 4. E. coli (a), B. subtilis (b), and Mycobacterium sp. (c) imaged by contact mode in 20 mM HEPES after immobilisation by an attachment to polyphenolic adhesive

proteins. All images are 20�20 mm2, and the height scale is 2 mm in (a, c) and 1.5 mm in (b). Arrows indicate image distortion by insufficiently immobilised cells.

R. Louise Meyer et al. / Ultramicroscopy ] (]]]]) ]]]–]]] 7

[32] recently demonstrated that poly-dopamine is also suitablefor immobilising bacterial on the tip of an AFM cantilever withoutaffecting cell viability or surface properties [32]. Similar to

Please cite this article as: R. Louise Meyer, et al., Ultramicroscopy (2

immobilisation to Cell-TakTM coated surfaces, this approachappears very promising for providing a generally applicableimmobilisation method, which can be used for bacteria of all

010), doi:10.1016/j.ultramic.2010.06.010

Fig. 5. Live/dead (green/red) staining of S. sciuri immobilised (a) electrostatically, (b) covalently by EDC–NHS, (b) covalently by glutaraldehyde, and (c) by an attachment to

polyphenolic adhesive proteins.

R. Louise Meyer et al. / Ultramicroscopy ] (]]]]) ]]]–]]]8

shapes and sizes, and does not require exposure to chemicals,distilled water, or drying—all of which may affect cell viability,cell surface properties, or trigger an undesirable biologicalresponse.

To minimise interaction with the cells, Cell-TakTM was firstadded to the glass substrate, and unbound proteins were removedby rinsing and drying before bringing cells into contact with thesurface. We found no indication that any of the immobilisationtechniques used in this study affected cell viability (Fig. 5).Viability was in this case measured as cell membrane integrity,and the result does not exclude that cell physiology changed uponimmobilisation. This has previously been shown afterimmobilisation to for instance Poly-L-lysine coated surfaces [37].Bacteria – especially those able to form biofilms – will almostcertainly respond physiologically to become immobilised on asurface. This biological response must be considered whenperforming AFM on live cells, especially if the immobilisationtechnique includes antimicrobial compounds, such as poly-L-lysine and glutaraldehyde.

The approaches to cell immobilisation evaluated in this studyall have different advantages and disadvantages with respect toease of sample preparation, reproducibility, and applicability todifferent cell types. No single method provides a one-fits-allapproach matching the cell type, substrate type, analysis condi-tions, and instrumental set-up of any experiment. For example,the combination of AFM with an inverted optical microscoperequires a transparent substrate, and therefore rules outimmobilisation by physical confinement on polycarbonate filtersor silicon wafers. Variability in cell surface charge and desiredbuffer strength during analysis will affect efficiency of immobi-lisation by electrostatic interactions. Likewise, the surface densityof carboxyl or amine groups on the cell surface will affect whether

Please cite this article as: R. Louise Meyer, et al., Ultramicroscopy (2

EDC–NHS or glutaraldehyde is most appropriate for covalentbinding of cells to a substrate. Among the methods tested, wefound that immobilisation by adhesive proteins provided themost reproducible approach, while at the same time providingfast and easy sample preparation and high flexibility with regardto the experimental set-up.

Acknowledgements

We thank The Danish Natural Science Research Council forsupporting Rikke Louise Meyer’s Steno fellowship (sagsnr. 272-05-0020) and for the support to iNANO, the Danish Dairy Boardand Svineafgiftsfonden for supporting Xingfei Zhou, and TheDanish Food Industry Agency for supporting Ayyoob Arpanaei,and the Carlsberg Foundation. Special thanks to Jorg Barner fromJPK Instruments for providing inspiration and discussion, parti-cularly regarding application of Cell-TakTM.

References

[1] L. Kailas, E.C. Ratcliffe, E.J. Hayhurst, M.G. Walker, S.J. Foster, J.K. Hobbs,Ultramicroscopy 109 (7) (2009) 775–780.

[2] A. Touhami, M.H. Jericho, T.J. Beveridge, J. Bacteriol. 186 (11) (2004) 3286–3295.[3] G. Francius, O. Domenech, M.P. Mingeot-Leclercq, Y.F. Dufrene, J. Bacteriol.

190 (24) (2008) 7904–7909.[4] D.J., Muller, Y.F. Dufrene, 3 (5) (2008) 261–269.[5] D. Alsteens, E. Dague, P.G. Rouxhet, A.R. Baulard, Y.F. Dufrene, Langmuir 23

(24) (2007) 11977–11979.[6] E. Dague, D. Alsteens, J.P. Latge, C. Verbelen, D. Raze, A.R. Baulard,

Y.F. Dufrene, Nano Lett. 7 (10) (2007) 3026–3030.[7] D. Alsteens, C. Verbelen, E. Dague, D. Raze, A.R. Baulard, Y.F. Dufrene, Pflugers

Arch. 456 (1) (2008) 117–125.[8] Y.F. Dufrene, Nat. Rev. Microbiol. 6 (9) (2008) 674–680.[9] Y.F. Dufrene, Nat. Rev. Microbiol. 2 (6) (2004) 451–460.

010), doi:10.1016/j.ultramic.2010.06.010

R. Louise Meyer et al. / Ultramicroscopy ] (]]]]) ]]]–]]] 9

[10] T. Fukuma, K. Kobayashi, K. Matsushige, H. Yamada, Appl. Phys. Lett. 86 (19)(2005) 1–3.

[11] A.V. Bolshakova, O.I. Kiselyova, A.S. Filonov, O.Y. Frolova, Y.L. Lyubchenko,I.V. Yaminsky, Ultramicroscopy 86 (1–2) (2001) 121–128.

[12] A. Mendez-Vilas, A.M. Gallardo-Moreno, R. Calzado-Montero, M.L. Gonzalez-Martın, Colloids Surf. B Biointerfaces 63 (1) (2008) 101–109.

[13] S. Kasas, A. Ikai, Biophys. J. 68 (5) (1995) 1678–1680.[14] S.B. Velegol, B.E. Logan, Langmuir 18 (13) (2002) 5256–5262.[15] M.J. Doktycz, C.J. Sullivan, P.R. Hoyt, D.A. Pelletier, S. Wu, D.P. Allison,

Ultramicroscopy 97 (1–4) (2003) 209–216.[16] A. Mendez-Vilas, A.M. Gallardo-Moreno, M.L. Gonzalez-Martın, Microsc.

Microanal. 13 (1) (2007) 55–64.[17] M. Hermansson, Colloids Surf. B Biointerfaces 14 (1–4) (1999) 105–119.[18] O. Stukalov, A. Korenevsky, T.J. Beveridge, J.R. Dutcher, Appl. Environ.

Microbiol. 74 (17) (2008) 5457–5465.[19] T., Hauschild, S. Schwarz, 50 (5) (2003) 241–246.[20] D., Simeoni, L. Rizzotti, P. Cocconcelli, S. Gazzola, F. Dellaglio, S. Torriani, 25

(1) (2008) 196–201.[21] S., Stepanovic, V. Dimitrijevic, D. Vukovic, I. Dakic, B. Savic, M. Svabic-Vlahovic,

82 (2) (2001) 177–185.[22] S., Stepanovic, P. Jezek, D. Vukovic, I. Dakic, P. Petras, 41 (11) (2003) 5262–5264.[23] B., Guirguitzova, D. Chankova, B. Zozikov, 36 (5) (2002) 341–347.[24] P. Hanarp, D.S. Sutherland, J. Gold, B. Kasemo, Colloids Surf. A Physicochem.

Eng. Asp 214 (2003) 23–36.[25] H. Agheli, D.S. Sutherland, IEEE Trans. Nanobiosci. 5 (1) (2006) 9–14.[26] P. Vermette, L. Meagher, Langmuir 18 (26) (2002) 10137–10145.[27] B. Rowan, M.A. Wheeler, R.M. Crooks, Langmuir 18 (25) (2002) 9914–9917.[28] T.A. Camesano, M.J. Natan, B.E. Logan, Langmuir 16 (10) (2000) 4563–4572.[29] H. Wang, R. Bash, J.G. Yodh, G.L. Hager, D. Lohr, S.M. Lindsay, Biophys. J. 83 (6)

(2002) 3619–3625.[30] S.B. Velegol, S. Pardi, X. Li, D. Velegol, B.E. Logan, Langmuir 19 (3) (2003)

851–857.[31] A., Razatos, Y.L. Ong, M.M. Sharma, G. Georgiou, 95 (19) (1998) 11059–11064.[32] S. Kang, M. Elimelech, Langmuir 25 (17) (2009) 9656–9659.

Please cite this article as: R. Louise Meyer, et al., Ultramicroscopy (2

[33] V. Vadillo-Rodriguez, T.J. Beveridge, J.R. Dutcher, J. Bacteriol. 190 (12) (2008)4225–4232.

[34] C.J. Sullivan, S. Venkataraman, S.T. Retterer, D.P. Allison, M.J. Doktycz,Ultramicroscopy 107 (10–11) (2007) 934–942.

[35] H. Lee, N.F. Scherer, P.B. Messersmith, Proc. Natl. Acad. Sci. USA 103 (35)(2006) 12999–13003.

[36] H. Lee, B.P. Lee, P.B. Messersmith, Nature 448 (7151) (2007) 338–341.[37] K., Colville, N. Tompkins, A.D. Rutenberg, M.H. Jericho26 (4) 2639–2644.[38] Y.F. Dufrene, C.J.P. Boonaert, P.A. Gerin, M. Asther, P.G. Rouxhet, J. Bacteriol.

181 (17) (1999) 5350–5354.[39] H.C. Van Der Mei, H.J. Busscher, R. Bos, J. De Vries, C.J.P. Boonaert,

Y.F. DufrA %ane, Biophys. J. 78 (5) (2000) 2668–2674.

[40] Y.F. Dufrene, Nat. Protoc 3 (7) (2008) 1132–1138.[41] M.A. Beckmann, S. Venkataraman, M.J. Doktycz, J.P. Nataro, C.J. Sullivan,

J.L. Morrell-Falvey, D.P. Allison, Ultramicroscopy 106 (8–9) (2006) 695–702.[42] A.E. Pelling, Y. Li, W. Shi, J.K. Gimzewski, Proc. Natl. Acad. Sci. USA 102 (18)

(2005) 6484–6489.[43] F. Gaboriaud, B.S. Parcha, M.L. Gee, J.A. Holden, R.A. Strugnell, Colloids Surf. B

Biointerfaces 62 (2) (2008) 206–213.[44] G. Francius, B. Tesson, E. Dague, V. Martin-Jezequel, Y.F. Dufrene, Environ.

Microbiol. 10 (5) (2008) 1344–1356.[45] P., Schar-Zammaretti, J. Ubbink, 85 (6) (2003) 4076–4092.[46] V. Vadillo-Rodrıguez, H.J. Busscher, H.C. Van Der Mei, J. De Vries, W. Norde,

Colloids Surf. B Biointerfaces 41 (1) (2005) 33–41.[47] J. Strauss, N.A. Burnham, T.A. Camesano, J. Mol. Recognition 22 (5) (2009)

347–355.[48] V. Vadillo-Rodriguez, H.J. Busscher, W. Norde, J. de Vries, R.J.B. Dijkstra,

I. Stokroos, H.C. van der Mei, Appl. Environ. Microbiol. 70 (9) (2004) 5441–5446.[49] S. Xiaoxia, T.Y. Peng, P.S. Olavi, Water Sci. Technol. 54 (2006) 17–25.[50] T. Cao, H. Tang, X. Liang, A. Wang, G.W. Auner, S.O. Salley, K.Y.S. Ng,

Biotechnol. Bioeng. 94 (1) (2006) 167–176.[51] X. Sheng, Y.P. Ting, S.O. Pehkonen, J. Colloid Interface Sci. 310 (2) (2007)

661–669.[52] M. Gad, A. Itoh, A. Ikai, Cell Biol. Int. 21 (11) (1997) 697–706.

010), doi:10.1016/j.ultramic.2010.06.010