INSTITUTE OF PHYSICS PUBLISHING NANOTECHNOLOGY

Nanotechnology 17 (2006) 4003–4009 doi:10.1088/0957-4484/17/16/001

AFM study of glucagon fibrillation viaoligomeric structures resulting ininterwoven fibrilsMingdong Dong1,2, Mads Bruun Hovgaard1,2, Sailong Xu1,2,Daniel Erik Otzen1,3 and Flemming Besenbacher1,2,4

1 Interdisciplinary Nanoscience Center (iNANO), University of Aarhus, DK-8000 Aarhus C,Denmark2 Department of Physics and Astronomy, University of Aarhus, DK-8000 Aarhus C, Denmark3 Centre for Insoluble Protein Structures (inSPIN), Department of Life Sciences,Aalborg University, DK-9000 Aalborg, Denmark

E-mail: fbe@inano.dk

Received 7 July 2006Published 14 July 2006Online at stacks.iop.org/Nano/17/4003

AbstractGlucagon is a 29-residue amphiphatic hormone involved in the regulation ofblood glucose levels in conjunction with insulin. In concentrated aqueoussolutions, glucagon spontaneously aggregates to form amyloid fibrils,destroying its biological activity. In this study we utilize the atomic forcemicroscope (AFM) to elucidate the fibrillation mechanism of glucagon at thenanoscale under acidic conditions (pH 2.0) by visualizing the nanostructuresof fibrils formed at different stages of the incubation. Hollow disc-shapedoligomers form at an early stage in the process and subsequently rearrange tomore solid oligomers. These oligomers co-exist with, and most likely act asprecursors for, protofibrils, which subsequently associate to form at leastthree different classes of higher-order fibrils of different heights. A repeatunit of around 50 nm along the main fibril axis suggests a helicalarrangement of interwoven protofibrils. The diversity of oligomeric andfibrillar arrangements formed at pH 2.0 complements previous spectroscopicanalyses that revealed that fibrils formed under different conditions can differsubstantially in stability and secondary structure.

(Some figures in this article are in colour only in the electronic version)

1. Introduction

The improper aggregation of polypeptide fragments anddeposition in lesions or plaques often result in abnormalphysiological function at the plaque site [1]. For instance, astretch of polyglutamine that repeats in a particular protein hasbeen proposed as the cause for many neurological disorders,including Huntington’s disease and spinal bulbar muscularatrophy, while aggregation of other unrelated proteins are citedas the causes for prion disease (PrP aggregation), Parkinson’sdisease and amyotrophic lateral sclerosis (ALS) (a-synucleinaggregation), dialysis-related amyloidosis (β-2 microglobulinaggregation), corneal dystrophy (kerato-epithelial deposits)

4 Author to whom any correspondence should be addressed.

and aggregation of islet amyloid polypeptide in more than 95%of type II diabetes [2]. Here, we focus on the polypeptideglucagon, which is an amphiphatic hormone, involved in theregulation of blood glucose levels. Glucagon has a strongpropensity to aggregate under a variety of conditions, whichlimits the stability of the hormone. The ensuing fibrils differsignificantly in secondary structure, suggesting an inherentstructural plasticity of the hormone which may also be reflectedat the level of the actual fibrils [3, 4]. When glucagonfibrillation is followed by spectroscopic techniques, such astryptophan fluorescence, binding of the fluorescent probethioflavin T, or changes in the far-UV CD signal, a simpletransition is generally observed [3, 4]. This transition involvesa lag time of several hours followed by a rapid elongation

0957-4484/06/164003+07$30.00 © 2006 IOP Publishing Ltd Printed in the UK 4003

M Dong et al

phase and stabilization at a new spectroscopic plateau. Nointermediate species such as pre-fibrillar oligomers, or nucleiare detected in this manner, either because they do notaccumulate to a sufficient extent or are not spectroscopicallydistinct from the preceding or subsequent states. Occasionally,a small overshoot can be observed at the end of the elongationphase, suggesting that once the assembly of glucagons intoan initial higher-order structure has occurred additional minorrearrangements can occur at the local level. These approaches,however, do not provide a global view of the overall changesin fibrillar structure. Atomic force microscopy (AFM) [5–9]is particularly useful for studying fibrillar ultrastructure, as itpermits visualization of individual biological macromoleculesat the nanometre scale [10–12]. For example, AFM hasbeen successfully used for monitoring individual peptide fibrilgrowth by time-lapse imaging and proved a useful tool to findthe condition that controls the growth and nature of amyloiddeposition in vivo [13].

In this work, we have used AFM to study the in vitrostructures of glucagon aggregation at pH 2.0 in detail togain insight into the fibrillogenesis process. We identifya number of different aggregated states, including disc-shaped oligomers, which we suggest act as precursorsfor the actual protofibrils. This observation emphasizesthe advantage of monitoring association processes at thelevel of individual molecular aggregates and illustrates theconformational richness of the fibrillation process occurring atthe nanoscale.

2. Experimental details

2.1. Sample preparation

Pharmaceutical grade glucagon protein powder (Novo NordiskA/S, Bagsværd, Denmark, purity > 98.9%) was dissolved to aconcentration of 2.5 mg ml−1 at pH 2.0 (10 mM HCl and 1 mMNa2SO4) and then incubated at room temperature.

2.2. Atomic force microscopy

A 1 µl sample was removed from the stock solutionevery hour for 24 h, deposited onto the freshly cleavedmica surfaces, and air dried for 5 min. All theimages were obtained using a commercial Nanoscope IIIaMultiMode SPM (Veeco Instruments, Santa Barbara, CA)under ambient conditions. Ultrasharp silicon cantilevers(triangular, NSC11/Sting, MikroMasch, Spain) were used witha typical resonance frequency of 50 kHz, a spring constant of5 N m−1 and a normal tip radius of 10 nm. AFM imagingwas performed in tapping mode at scan frequencies of 1–2 Hz with minimal loading forces of 100 pN applied andoptimized feedback parameters. The resolution of AFMscans was 512 × 512 pixels. Several images were obtainedfrom separate locations across the mica surfaces to ensurereproducibility. All the images were first flattened using theNanoScopeTM software (Digital Instruments, version 5.31),while excluding the particles from the flattened area, andthen analysed automatically by using the commercial ScanningProbe Image Processor (SPIPTM) software (Image MetrologyApS, version 4.2, Lyngby, Denmark) to yield size histogramsof the images.

3. Results and discussion

To obtain a complete profile of the fibrillogenesis process, atime-lapse study of glucagon fibrillation was conducted usingAFM to visualize the structures and states of fibrillation formedat different incubation times. Upon depositing newly dissolvedglucagon onto a freshly cleaned mica surface, the formationof a homogeneously spread layer was observed, consisting ofmonomeric globular structures (figure 1(a)). In an extendedform (all beta-sheet) the monomer is approximately 10 nm insize (3.5 A/residue × 29 residues). A more compact globularstructure, however, will result in smaller features as observed.After one hour of incubation, a few bigger particles wereobserved on the surface in addition to the monomers (indicatedby white circles in figure 1(b)). These particles are disc shaped,and appear to contain a central channel or hole as shown in theinset image of figure 1(b). The bigger particles suggest theformation of oligomers, which have also been observed for alarge number of other aggregating proteins [14]. At this stage,the glucagon monomers are still the dominant species. After90 min, the oligomer particle structures have, however, becomemore prominent, and the central cavities have disappeared(figures 1(c) and 2(a)). Statistical analysis shows an oligomersize distribution of 1.5 ± 0.5 nm in height and 20.8 ±5.2 nm in width after Gaussian fitting (figures 2(b) and (c)).The oligomers show relatively uniform disc-shaped particles.Repeated AFM imaging reveals that the glucagon retains theparticular and non-fibrillar shape for a period of at least 3 h.After 5 h of incubation, a few straight fibrillar structures aredetected with lengths ranging from short (<100 nm) to long(>1 µm) (figures 1(d), 3(a)). High-resolution images showthat the larger globular oligomers (figure 3(b)) still coexistwith fibrils, suggesting such oligomers as an essential speciesfor the fibril growth. In addition, two kinds of fibrils areindicated by arrow ‘α’ and arrow ‘β’ in figure 3(b). The‘α’ fibril comprises a uniform fibril, which is thinner thanthe ‘β’ fibril. Importantly, the ‘β1’ fibril clearly shows thattwo thin fibrils ‘α1’and ‘α2’ have joined together. The linescan along ‘A–A’ in figure 3(b) shows detailed width andheight differences between these two kinds of fibrils lying onmica. The ‘α’ fibril and ‘β’ fibril have heights of ∼6.05 and∼10.73 nm and widths of 32.9 and 54.9 nm, respectively.The height of the ‘β’ fibril is almost twice the height ofthe ‘α’ fibril. Since the AFM images may be perturbed bythe final size of the AFM tip, often referred to as AFM tipconvolution, the height value is a more accurate measure ofthe fibril size. The additional height errors introduced by tipcompression and surface adsorption forces are, however, nottaken into account, but minimized by always imaging at theleast interaction force needed. The statistical analysis showsa fibrillar size distribution with three height classes, namely6.5 ± 2.9, 10.0 ± 1.8 and 15.0 ± 7.0 nm (all fitted withGaussian curves) and a width of 35.68 ± 14.6 nm (fitted withthe Gaussian curve as shown in figure 3(d)). The statisticalanalysis based on the AFM images suggests that ‘α’ fibrils arethe first fibrillar species (henceforth called protofibrils), the ‘β’fibrils are stacked by two protofibrils called type 1 fibril and ‘ε’fibrils are due to additional accretion of protofibrils or type 1fibrils. Protofibrils remain the predominant nanospecies at thisinitial time point.

4004

AFM study of glucagon fibrillation via oligomeric structures resulting in interwoven fibrils

00:30 (a) (b) (c)

(d) (e) (f)05:00

01:00

07:00

01:30

20:00

500nm 500nm 100nm

1µm 1µm 1µm

Figure 1. AFM images of aggregated glucagon species over time in 10 mM HCl and 1 mM Na2SO4. Incubation times in hours are indicatedin each figure.

100nm

0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.50.00

0.05

0.10

0.15

nm

nm10 20 30 400.00

0.05

0.10

0.15

Fre

qu

en

cy

(a)(b)

(c)

Width

Gauss fit of width

Height

Gauss fit of height

Figure 2. (a) AFM image of glucagon after 90 min of incubation; the glucagon oligomers show ‘globular’ shape structures. (b) The particlewidth distribution fitted to Gaussian curves. (c) The particle height distribution fitted to Gaussian curves.

Following 1 h of additional incubation, figure 4(a) showsa significant number of long fibrils with lengths of up to5 µm extending from protofibrils. In addition to these longfibrils, a significant percentage of smaller globular structuresstill exist. Very interestingly, a high-resolution image (shownin figure 4(b)) shows periodicities of ∼52 nm and ∼55 nmon the B–B and C–C lines, respectively, as shown by the lineprofiles in figure 4(e). Furthermore, the fibril indicated by thearrow ‘δ’ in figure 4(b) is almost twice as high as the adjacentsmooth protofibril indicated by the arrow ‘α’ in figure 4(b)as shown by the line profile A–A in figure 4(e). We call thehigher fibril a type 2 fibril and suggest that it consists of twoprotofibrils with a periodicity due to the protofibrils wrappingup each other. In contrast, figures 4(c) and (d) show smoothfibrils as indicated by arrows ‘γ ’ with a height ∼3 timesgreater than protofibrils, suggesting that it consists of threeinterwoven protofibrils. The smoothness of the fibril without

periodicities is presumably due to a closer and more optimizedwrapping. We term this a type 3 fibril. After incubationup to 7 h, figure 5(a) shows mature fibrils with lengths ofup to 15 µm. The high-magnification image (figure 5(b))supports the rather complicated assembly pathway suggestedso far. Two protofibrils (α1 and α2) assemble into a type 1fibril (β1). Furthermore, three type 1 fibrils (β1, β2 and β3)can intertwine to assemble higher and much more compactfibrils (γ1), which correspond to the triple helical pathway astype 3 fibrils. Occasionally, two type 1 fibrils (β4 and β5)also intertwine to form helical fibrils (δ), which correspondto the same double helical pathway as type 2 fibrils. Clearly,the higher level bundled fibrils can form through two kindsof assembly pathways. Eventually, after overnight incubation(22 h), the mica surface was completely covered with above15 µm long fibrils (figure 1(f)). The mature fibrils showstraight and compact morphology and are stable for several

4005

M Dong et al

A

1

5 10 15 20 25 30

nm

HeightGauss fit of height Gauss fit peak α Gauss fit peak β Gauss fit peak ε

Fre

quen

cy

nm

20 30 40 50 60 70 80

WidthGauss fit of width

III

ε

III

3

A

0 200 400 600 800

10.73nm

6.05nm

nm

Hei

ght (

nm)

(a) (b)

(c)

(d)

β

β

α

α

α

αβ

ε

2

1

22µm 500nm

-3

0

3

6

9

12

0.00

0.05

0.10

0.15

0.00

0.05

0.10

0.15

Figure 3. Glucagon fibrils assemble on mica surface after 5 h of incubation. (a) AFM image showing short protofibrils with broad variation oflength. (b) Higher magnification reveals the coexistence of protofibrils (indicated by arrows α1, α2, and α3), higher fibrils (indicated by arrowsβ1 and β2), oligomers (indicated by white circles), and a fibril cluster (indicated by arrow ε). The β1 fibril shows α1 and α2 joining together.(c) Line scan profile along the line A–A of image (b) demonstrating the type 1 fibril (height:10.73 nm), consisting of two protofibrils(height: 6.05 nm). (d) The fibril width and height distributions fitted with Gaussian curves.

weeks (data not shown). However, the periodicity cannotbe observed, which suggests that the periodicity gets smalleralong the mature fibril, or the fibrillar packing is optimized tosmooth out surface protrusions.

Oligomers and protofibrils are clearly both importantspecies in the fibrillogenesis process. At early stagesof glucagon incubation, disc-shaped oligomer formation isrequired for fibril formation as reported by other amyloidstudies [15–17]. Similar disc-shaped particles were alsoobserved by in vitro studies of the light-chain domain by CongoRed birefringence [18]. The simplest model that incorporatesthese different species posits that the glucagon oligomer is thebuilding block for early-stage protofibrils, although it is likelythat they have to undergo additional structural rearrangementsupon assembly to lead to the smooth and straight features ofthe protofibrils (figure 6). Subsequently, protofibrils assembleinto three major types of fibrils.

Fibrillar polymorphism at the ultrastructural level is ageneral feature for amyloid aggregation, and can be causedby a variety of interfibrillar interactions during the assemblyprocess. The three major pathways of fibrillogenesis can bepropagated during fibril growth. The mature fibrils are verystable and can persist for long periods of time, presumablyallowing them to seed additional quantities of the peptide intoamyloid fibrils.

On the basis of the AFM experiments we cannot uniquelyassign the increase in spectroscopic signal to the formationof a specific species [3]. Due to the rather stochastic natureof protein fibrillation, it is not straightforward to conductparallel spectroscopic and AFM studies on the same speciesincubated under identical conditions. However, we considerit most likely that the protofibrils are responsible for the largeincrease in thioflavin T fluorescence, as oligomers generallydo not show a capability to enhance thioflavin T fluorescence.

4006

AFM study of glucagon fibrillation via oligomeric structures resulting in interwoven fibrils

200nm 1µm(b)(a)

(c)

C

C

B

B

(d)200nm 200nm

D

A A

D

0 100 200 300 400 500 600 700

0 100 200 300 400 500 600 700

0 100 200 300 400 500 600 700

0 100 200 300 400 500 600 700

A-A

B-B

C-C

D-D

(e)

γ

α

α

δ

δ

δ

δ

δγ

γα

α

α

048

12

048

12

048

12

048

12

Hei

ght (

nm)

Position (nm)

Figure 4. Protofibrils assemble into different types of fibrils on mica surface. (a) Formation of glucagon fibrils of up to 5 µm in length after6 h of incubation. ((b)–(d)) High-magnification images showing photofibrils (indicated by arrows α), type 2 fibrils with a regular periodicity(indicated by arrows δ) and type 3 fibrils with straight and compact morphology (indicated by arrows γ ). (e) The line profiles based on theline scan along A–A, B–B, C–C and D–D of (b) and (d).

In fact, using pyrene-labelled glucagon molecules, we haverecently identified a spectroscopically distinct species thataccumulates transiently prior to the major fibrillation step,reaching a maximum just before fibrillation begins anddecaying in step with the ensuing fibril elongation phase(P A Christensen and DEO, unpublished observations).

This observation is consistent with the formation of aprefibrillar aggregate. It is likely that the oligomers haveto rearrange substantially to form the stable protofibrillarstructure, leading to the increase in organized secondarystructure and blue-shift in Trp fluorescence that is observed inparallel [3, 4].

4007

M Dong et al

2

(a) (b)

β1

α1

β2

β3

α

δ 1

β4

γ 1

β5

4 µm 500 nm

Figure 5. (a) Formation of glucagon fibrils with lengths of up to 15 µm after 7 h incubation. (b) High-magnification image showingprotofibrils (indicated by arrows α1 and α2), type 1 fibrils (indicated by arrows β1, β2, β3, β4 and β5), type 3 fibrils (indicated by the arrow γ1)and type 2 fibrils (indicated by the arrow δ1)

Type 2 fibril

Type 1 fibril

Type 3 fibril

6

3

0

Monomer Early oligomer Mature oligomer Protofibril

Type 1 fibril Type 2 fibril Type 3 fibrilProtofibril

1 2 4

5

7

Figure 6. Proposed model of glucagon aggregation based on AFM data. We tentatively include a rearrangement step for the oligomer, whichmay be a necessary precondition for the stable propagation into protofibrils. The fibrils probably form by lateral association of pre-formedprotofibrils, but it is also likely that the fibrils can subsequently grow by elongation of associated protofibrils.

4. Conclusion

We have used AFM to monitor the assembly process ofglucagon aggregation at the nanoscale. Our observationsclearly suggest that oligomers, protofibrils and mature fibrilsare the main players in this process at pH 2, and provide thebasis for a general model (figure 6). The initial glucagon

monomers are visualized as particle-like structures, whosesize increases dramatically with incubation time from anearly stage due to formation of disc-like oligomers. Whilewe cannot speculate on the significance of the cavitiesobserved for early-stage oligomers in detail, it cannot beruled out that these cavities are a natural result of the initialassociation of monomers and are subsequently filled out by

4008

AFM study of glucagon fibrillation via oligomeric structures resulting in interwoven fibrils

the accretion of additional monomers and/or a rearrangementof the oligomer structure. This could then provide anenergetically more favourable precursor for the subsequentformation of protofibrils by oligomer association, as suggestedby the observed co-existence of oligomers and protofibrils. Thelag time may be due to the significant structural rearrangementsrequired for the oligomer to be able to assemble intoprotofibrils. Following protofibril formation, a variety offibrils is observed, which mainly follows three major pathwaysas shown in figure 5 (path 3, 4, 5). Type 1, type 2 andtype 3 represent the two fibrils staking, two fibrils cross-overtwisting, and three fibrils twisting together, respectively. Thus,the protofibrils are essential components for fibril elongationand the rise of fibrillar variability [11]. Occasionally, type Ifibrils can also follow pathway 6 and 7 (shown in figure 5)to propagate to higher fibril levels. Previous experimentshave demonstrated that the overall morphologies of glucagonaggregations depend on the condition in which fibrillogenesistakes place [3]. Here we have visualized different fibrilsobserved in the same experimental condition, where thepolymorphisms are caused by fibrils with a variable numberand arrangement of protofibrils.

Acknowledgments

Our research is supported by the Danish Natural ScienceResearch Council through funding for the iNANO Centre. Weare grateful to Novo Nordisk A/S for providing glucagon.We would like to acknowledge Professor P Hinterdorfer ofUniversity of Linz, Austria, and Professor Vinod Subramaniamof University of Twente, the Netherlands, for valuablediscussions and suggestions.

References

[1] Frokjaer S and Otzen D E 2005 Protein drug stability: aformulation challenge Nat. Rev. Drug Discovery 4 298–306

[2] Caughey B and Lansbury P T 2003 Protofibrils, pores, fibrils,and neurodegeneration: separating the responsible proteinaggregates from the innocent bystanders Annu. Rev.Neurosci. 26 267–98

[3] Pedersen J S, Dikov D, Flink J L, Hjuler H A,Christiansen G and Otzen D E 2006 The changing face of

glucagon fibrillation: structural polymorphism andconformational imprinting J. Mol. Biol. 355 501–23

[4] Pedersen J S, Flink J M, Dikov D and Otzen D E 2006 Sulfatesdramatically stabilize a salt-dependent type of glucagonfibrils Biophys. J. 90 4181–94

[5] Hinterdorfer P and Dufrene Y F 2006 Detection andlocalization of single molecular recognition events usingatomic force microscopy Nat. Methods 3 347–55

[6] Horber J K H and Miles M J 2003 Scanning probe evolution inbiology Science 302 1002–5

[7] Kienberger F, Ebner A, Gruber H J and Hinterdorfer P 2006Molecular recognition imaging and force spectroscopy ofsingle biomolecules Acc. Chem. Res. 39 29–36

[8] Hoyer W, Antony T, Cherny D, Heim G, Jovin T M andSubramaniam V 2002 Dependence of alpha-synucleinaggregate morphology on solution conditions J. Mol. Biol.322 383–93

[9] Andersen Ebbe S, Contera S A, Knudsen B, Damgaard C K,Besenbacher F and Kjems J 2004 Role of the trans-activationresponse element in dimerization of HIV-1 RNA J. Biol.Chem. 279 22243–9

[10] Harper J D, Wong S S, Lieber C M and Lansbury P T 1997Observation of metastable a beta amyloid protofibrils byatomic force microscopy Chem. Biol. 4 119–25

[11] Khurana R et al 2003 A general model for amyloid fibrilassembly based on morphological studies using atomic forcemicroscopy Biophys. J. 85 1135–44

[12] Gunning A P, Kirby A R, Mackie A R, Kroon P,Williamson G and Morris V J 2004 Watching molecularprocesses with the atomic force microscope: dynamics ofpolymer adsorption and desorption at the single moleculelevel J. Microsc. 216 52–6

[13] Goldsbury C, Kistler J, Aebi U, Arvinte T andCooper G J S 1999 Watching amyloid fibrils grow bytime-lapse atomic force microscopy J. Mol. Biol. 285 33–9

[14] Dobson C M 2003 Protein folding and misfolding Nature426 884–90

[15] Lundberg K M, Stenland C J, Cohen F E, Prusiner S B andMillhauser G L 1997 Kinetics and mechanism of amyloidformation by the prion protein H1 peptide as determined bytime-dependent ESR Chem. Biol. 4 345–55

[16] Zhu M, Han S B, Zhou F M, Carter S A and Fink A L 2004Annular oligomeric amyloid intermediates observed by insitu atomic force microscopy J. Biol. Chem. 279 24452–9

[17] Zhu M, Souillac P O, Ionescu-Zanetti C, Carter S A andFink A L 2002 Surface-catalyzed amyloid fibril formationJ. Biol. Chem. 277 50914–22

[18] Raffen R et al 1999 Physicochemical consequences of aminoacid variations that contribute to fibril formation byimmunoglobulin light chains Protein Sci. 8 509–17

4009