Báo cáo khoa học: Amyloid structure – one but not the same: the many levels of fibrillar polymorphism

M I N I R E V I E W

Amyloid structure – one but not the same: the many levels of fibrillar polymorphism Jesper S. Pedersen1, Christian B. Andersen2,3 and Daniel E. Otzen4

1 Department of Biochemistry, Molecular Biology, and Cell Biology, Rice Institute for Biomedical Research, Northwestern University, Evanston, IL, USA 2 Protein Structure and Biophysics, Novo Nordisk A ⁄ S, Ma˚ løv, Denmark 3 Institute of Biophysics, National Research Council, CNR, Palermo, Italy 4 Department of Molecular Biology, Center for Insoluble Protein Structures, Interdisciplinary Nanoscience Centre, University of Aarhus, Denmark

Keywords aggregation; amyloid; fibrillar polymorphism; glucagon; mechanism; protein folding

Correspondence J. S. Pedersen, Department of Biochemistry, Molecular Biology, and Cell Biology, Rice Institute for Biomedical Research, Northwestern University, 2205 Tech Drive, Hogan 2-100, Evanston, IL 60208, USA Tel: +1 847 881 6617 E-mail: jsp@phage.dk D. Otzen, Department of Molecular Biology, Center for Insoluble Protein Structures, Interdisciplinary Nanoscience Centre, University of Aarhus, Gustav Wieds Vej 10, DK-8000 Aarhus C, Denmark Fax: +45 8612 3178 Tel: +45 8942 5046 E-mail: dao@inano.au.dk

(Received 16 April 2010, revised 2 September 2010, accepted 17 September 2010)

Many proteins and peptides can form amyloid-like structures both in vivo and in vitro. Although strikingly similar fibrillar structures can be observed across a variety of amino acid sequences, the fibrils formed often exhibit a stunning wealth of polymorphisms at the level of electron or atomic force microscopy. This appears to violate the Anfinsen principle seen for globu- lar proteins, where each protein sequence codes for just one well-defined fold. To a large extent, polymorphism reflects variable packing of a single protofilament structure in the mature fibrils. However, we and others have recently demonstrated that polymorphism can also reflect real structural differences in the molecular packing of the polypeptide chains leading to several possible protofilament structures and diverse mature fibrillar struc- tures. Glucagon has been a particularly useful model system for studying the fibrillogenesis mechanisms that lead to the formation of structural poly- morphism, thanks to its single tryptophan residue and the availability of large quantities at pharmaceutical-grade quality. Combinations of struc- tural investigations and seed extension experiments have revealed the repro- ducible formation of at least five different self-propagating fibril types from subtle variations in growth conditions. These reflect the underlying com- plexity of the peptide conformational landscape and provide a link to natively disordered proteins, where structure is dictated by context in the form of different binding partners. Here we review some of the latest advances in the study of glucagon fibrillar polymorphism and their implica- tions for mechanisms of fibril formation in general.

doi:10.1111/j.1742-4658.2010.07888.x

Amyloid in disease and functional fibrillar structures

Amyloid and amyloid-like structures are normally associated with several debilitating diseases, including Alzheimer’s, Parkinson’s, Huntington’s and transmissible

spongiform encephalopathies [1]. In these cases, the formation of amyloid and other aberrant aggregates represents an alternative type of folding that makes

Abbreviations AFM, atomic force microscopy; EM, electron microscopy; ITC, isothermic titration calorimetry; SAXS, small angle X-ray scattering; ThT, thioflavin T.

FEBS Journal 277 (2010) 4591–4601 ª 2010 The Authors Journal compilation ª 2010 FEBS

4591

J. S. Pedersen et al.

The structural ambiguity of glucagon amyloids

the protein become toxic and ⁄ or lose its natural func- tion [2–4]. In systematic amyloidoses, it appears to be the massive accumulation of amyloid per se that is pathological, leading to organ failure and eventual death [5–7]. Smaller amounts of amyloids accumulated in sensitive locations, such as the cornea, can also lead to severe functional impairment [8]. In most neurode- negerative diseases, however, prefibrillar aggregates are the toxic species, as detailed by Stefani in the review in this series [9].

types,

It has been suggested that the ability to form amy- loid-like structures, given appropriate conditions, is a general property of the polypeptide chain [10,11]. This ability not only has pathological consequences, but can be turned into good use as functional amyloid, serving as structural anchoring material for a range of pur- poses ranging from biofilm formation to a matrix to anchor melanin as well as for the reduction of interfa- cial tension, coating of spores and many other as yet unknown functions (see [12] and references herein). Several different protein and peptide hormones, includ- ing insulin and glucagon [13,14], have long been known to readily form fibrils, and it has recently been proposed that functional amyloid structures could serve as in vivo intracellular storage of these hormones in pituitary secretory granules, stabilized by interac- tions with glucosaminoglycans [15,16]. Upon release from the cells, these amyloid structures would gradu- ally dissociate into active monomers in the blood- the amyloid stream [17], which would imply that structures formed by these hormones in vivo should have evolved to be relatively unstable.

Fig. 1. Comparison of representative EM morphologies of the five type A [27,44,56,63], Bunagitated different glucagon fibril [44,53,56], Bagitated [23,27,71,73], D [27,73] and S [43,61,73], where combinations of spectroscopic, thermostability and seed extension kinetic data indicate distinct protofilament structures. Scale bar = 50 nm. The emission intensity of ThT staining of type A is app values > 40-fold higher than that observed for types B and D. Tm represent thermal melting midpoints during 90 (cid:2)CÆh)1 temperature ramping of 25 mgÆL)1 fibrils in 25 mM glycine ⁄ HCl (pH 2.67). The CD spectrum shows the reproducible unique fingerprint features of fibrils after sonication that can be used to distinguish between the types [27,43,56].

Fibrillar polymorphisms reflect structural ambiguity in amyloid fibrils

morphology are directly correlated to the number of protofilaments they contain [21]. Strikingly similar fibrillar morphologies can arise from proteins and peptides with very diverse amino acid sequences [22– 24]. In other cases, particular morphologies can be generated by manipulating the specific conditions for fibril formation [25–27].

structure

Initially it was assumed that the protein fold found in amyloid structures was a single energy minimized structure following Anfinsen’s single-fold principle for native proteins [28], with the different morphologies representing different lateral associations of a single, lowest-energy protofilament [22,29]. The only exception from this rule was thought to be the prions, which were shown to form several variants [30,31], so-called self-propagating strains, with different

The key to the properties of amyloid assemblies lies in the regularity and repetitiveness of their underlying molecular architecture of intermolecular b-strands stacked perpendicular to the fibril axis [18], often organized in several b-sheets parallel to the axis. Although atomic resolution structures of the molecu- lar packing of peptides in amyloid-like structures are beginning to emerge [19], much of our understanding of the structure of amyloid-like fibrils comes from electron microscopy (EM) techniques and atomic force microscopy (AFM) [20]. Images produced by these techniques often show stunningly beautiful, per- fectly ordered fibrils, some with regular twisting, oth- ers seemingly smooth and still others as parallel bundles of two or more protofilaments (Fig. 1). Fibril preparations often contain several different morpho- logies, and for some of these the turn lengths and

FEBS Journal 277 (2010) 4591–4601 ª 2010 The Authors Journal compilation ª 2010 FEBS

4592

J. S. Pedersen et al.

The structural ambiguity of glucagon amyloids

Breaking or branching fibril structures allow exponential growth

[23]

the experiment

characteristics,

(> 1 gÆL)1), whereas

(Fig. 2A). Branching in these fibrils

When glucagon powder is dissolved in acidic buffer, the fibril formation follows a highly reproducible sig- moidal curve [47]. Sigmoidal curves are observed in a number of biological systems, such as during exponen- tial growth of bacteria in a flask or exponential ampli- the obvious fication of DNA in PCR. Despite similarities, a common misconception (as reviewed by Roberts [48]) is that the lag time before the detection of fibrils can be taken as a direct indicator of the pro- pensity to slowly form a stable nucleus. For glucagon, this is illustrated by simple seeding experiments, which clearly demonstrate that the exponential growth phase extends through the apparent lag all the way to the (Fig. 2C), which beginning of means that the observed lag time depends inversely on the growth rate of fibrils, as well as the nucleation rate. A compilation of kinetic data from several pro- teins shows a clear inverse correlation between the apparent lag time and growth rates [49,50], suggesting that fibril nucleation may generally take place through- out the apparent lag [51], but exponential amplification from early nucleation events will lessen the effect of later nucleation events. The linear nature of fibrils implies that they grow linearly by the addition of pro- tein molecules to their ends [52]. Exponential growth can be attributed to the presence of secondary path- ways, which continuously increase the number of fibril ends in proportion to the fibril mass present [51]. Recently, we have demonstrated using total internal reflection fluorescence microscopy that a specific type of glucagon fibril, which we refer to as type Bunagitated [23], grows by branching under unagitated conditions [53] is also observed directly in high-resolution EM images [44], which demonstrate that the fibrils consist of two or more protofilaments that twist regularly. We speculate that the twists may be necessary for fibrils to send out branches or that cavities created on the surface of these structures catalyze the formation of new fibril nuclei. Fibrillogenesis of glucagon as well as a number including insulin [54], Ab1–40 [35] of other proteins, and prions [55], can be greatly accelerated by agitation. In the case of glucagon, clues to the mechanism of this acceleration can be observed in the morphology of the resulting fibrils, with agitation generally producing type Bagitated fibrils, which are short, nontwisted and nonbranched parallel bundles of two or more filaments [27]. We speculate that the type Bagitated fibrils are brit- tle fibrils that readily break during agitation, thereby doubling the number of fibril ends that can accept

molecular packing of the prion domain into the amy- loid structure [32]. However, the discovery of self- propagating variants of insulin [33], b2-microglobulin [34], Ab [35] and glucagon [27] fibrils demonstrated that strain behavior was not limited to proteins with a prion-like sequence. Recently it has been demonstrated that even small peptides, including the amylin frag- ment SNNFGAILSS, have the ability to form amyloid with both parallel and antiparallel b-sheet structures [36,37] depending on the structure of the seeds. The peptide GNNQQNY(7-13) forms two different types of quasifibrillar crystal [38] and three different types of amyloid according to solid-state NMR [39] that differ from one another in subtle but distinct ways, such as variations in the mobility of the aromatic Tyr ring. There is a growing number of similar observations from other peptide systems [40,41]. Remarkably, the prevailing structure of insulin fibrils formed during to vary randomly between two agitation appears optically distinct polymorphisms [42], indicating that indeterminism in early nucleation events dictates the structure. We have previously demon- final fibril strated that glucagon is able to form at least five different fibrillar structures that can be propagated by seeding in a strain-specific manner [23,27,43,44]. Each type can be identified by a unique combination of variable including thioflavin T (ThT) staining, CD spectrum fingerprints, thermosta- bility and morphology in EM (see Fig. 1). Type A, Bagitated, Bunagitated all form in the same 50 mm gly- cine ⁄ HCl buffer at pH 2.5, whereas types D and S only accumulate when (cid:2) 200 mm Cl) and 1 mm SO2(cid:3) are added, respectively. Type A forms at high 4 the glucagon concentrations other types form at low concentrations (< 0.5 gÆL)1). We have recently suggested that the prevalence of different glucagon fibril structures is the result of a multitude of aggregation pathways, where even small shifts in environmental conditions can impact the outcome of the struggle for monomers between sev- eral types of fibril due to modulation of their nucle- ation and exponential growth rates [23]. Interestingly, another class of proteins for which environmental conditions are critical in defining their structure is the group of intrinsically disordered proteins that are designed to show great conformational flexibility, allowing them to interact with multiple binding partners types of that can often induce different structure [45,46]. In these cases, however, the poly- morphism has been systematically optimized to facili- tate heterogeneous contacts with different proteins rather than the homogeneous assemblies illustrated by the amyloid folds.

FEBS Journal 277 (2010) 4591–4601 ª 2010 The Authors Journal compilation ª 2010 FEBS

4593

J. S. Pedersen et al.

The structural ambiguity of glucagon amyloids

Fig. 2. Secondary pathways in glucagon fibrillogenesis result in exponential growth of mature fibrils. (A) Under unagitated conditions, TIRF microscopy directly demonstrates that type Bunagitated fibrils increase the number of fibril ends by branching, leading to exponential growth. Adapted from [53]. (B) Under agitated conditions, type Bagitated fibrils have selective growth advantage because they continuously break, (C) Seed extension kinetics of 1 gÆL)1 glucagon in 50 mM thereby exposing new fibril ends that adsorb monomeric glucagon [27]. glycine ⁄ HCl pH 2.5 with agitation confirm that type Bagitated fibrils grow exponentially, and indicate that a small fraction ((cid:2) 1 in 105) of glucagon molecules initiate spontaneous nucleation as soon as glucagon is dissolved under the given conditions [23].

in the two forms is very similar, but that the quiescent form has a triangular cross-section with three protofil- aments with a narrow cavity in the middle, whereas the agitated striated ribbon form consists of two proto- filaments with a tight interaction between the flat sur- faces between them. It has recently been demonstrated that specific types of the Ab1–40 fibril can also grow by branching [60]. Investigations on the similarities and differences between the type Bunagitated and Bagitated glucagon fibril forms are currently being conducted.

Transient off-pathway formation of monofilament type A fibrils allows growth of type B fibrils at high glucagon concentrations

formation of

the

Based on time-lapse EM and AFM studies alone, it has been proposed that complex multifilament structures of mature fibrils may proceed by lateral assembly of preformed protofilaments or pro- tofibrils [29,61,62]. Even though EM and AFM provide the fibril structures high-quality information about

monomers (Fig. 2B) and providing them with a selec- tive growth advantage over type Bunagitated fibrils under agitated conditions [23]. The two types of fibril have very similar CD spectra with an unusual positive peak around 203 nm (Fig. 1) and two distinct b-sheet peaks in FTIR spectra and a shoulder at 1664 cm)1, indicating the presence of b-turns, suggesting that the molecular packing could be very similar [27,56]. Agita- tion-dependent molecular-level polymorphisms have also been reported for Ab1–40 and insulin fibrils [35,57]. Moreover, different types of prion fibril form under shaking and rotating conditions, indicating that the mode of agitation can also influence the prevailing pathway of fibrillogenesis [58]. Quiescent (unagitated) and agitated forms of Ab1–40 fibrils exhibit twisted and striated ribbon morphologies similar to the type Bunagitated and Bagitated glucagon fibrils, respectively. With rounds of seeded growth it has been possible to generate homogeneous samples of quiescent and agi- tated Ab1–40 fibrils that allowed a solid-state NMR study of underlying structural differences [59]. The structures reveal that the secondary structure of Ab1–40

FEBS Journal 277 (2010) 4591–4601 ª 2010 The Authors Journal compilation ª 2010 FEBS

4594

J. S. Pedersen et al.

The structural ambiguity of glucagon amyloids

Fig. 3. A proposed mechanism for the conversion of type A fibrils into type B. (A) Seeding experiments demonstrate that seeds of type A fibrils can grow at both high and low glucagon concentrations. In contrast, seeds of type Bunagitated fibrils grow exponentially only at low glucagon concentrations, possibly due to inhibition of either elongation or branching by a-helical trimers at high concentrations [44]. (B) Once type A fibrils have consumed > 95% of the monomers, the concentration is low enough to allow exponential growth of type Bunagitated fibrils. A rapid subsequent equilibrium between monomers and the relatively unstable type A fibrils ensures that glucagon monomer concentrations are maintained at low enough levels to support the growth of type Bunagitated fibrils. Data for the graph were taken from the recent SAXS study [63].

that both types exhibit

in EM and AFM, occurs

at

structural differences between type A and Bunagitated [56]: unlike type B fibrils, type A fibrils have an unusu- ally strong b-sheet CD spectrum (Fig. 1) and an FTIR spectrum with only one b-sheet peak. Linear dichroism of aligned fibrils indicates that type Bunagitated fibrils are less ordered than type A [56]. X-ray diffraction patterns the classical 4.76 A˚ reveal meridional reflection typical for amyloid-like structures [18], but whereas type Bunagitated only contains the clas- sical 9.8 A˚ equatorial reflection, type A fibrils exhibit a number of periodic reflections similar to those of a cyl- inder with well-defined edges. This suggests that the simple structure of type A fibrils can be aligned more orderly than the branched type Bunagitated fibrils [56]. Limited proteolysis results in the release of a different spectrum of peptides, further substantiating the struc- tural differences between the two types of fibril [56].

The two fibril types also differ in terms of the mecha- nisms leading to their formation, as evident from kinetic cross-seeding experiments: Fractionated seeds of both type A and Bunagitated fibrils can grow exponen- tially at low concentrations, but type Bunagitated fibrils have a faster exponential growth (a more shallow slope

formed, many images need to be analyzed to quantify the amounts of various structures formed. Moreover, the images do not provide information about differ- ences in the molecular packing of fibrils and it is virtu- ally impossible using these techniques alone to quantify the fraction of the protein that has become converted to fibrils at the given time point. The hierarchical build-up model seems to conflict with recent data that suggest that different types of glucagon fibril grow exponentially via their own distinct pathways into structurally distinct entities [23,27,43,44,53,56]. Several studies have revealed that the formation of monofila- ment type A fibrils (Fig. 1), which appear as single fila- ments glucagon concentrations above 1 gÆL)1 [27,44,63]. When agitated at 1 or 2 gÆL)1, transient formation of type A fibrils can be observed as a peak in ThT emission, which dis- appears as type Bagitated fibrils form [27]. However, under unagitated conditions at higher concentrations, the type A fibrils form a thick gel [47] that appears to be stable for longer periods, which makes it possible to study the properties of these otherwise metastable fibrils [44]. A recent study summarized the evidence for

FEBS Journal 277 (2010) 4591–4601 ª 2010 The Authors Journal compilation ª 2010 FEBS

4595

J. S. Pedersen et al.

The structural ambiguity of glucagon amyloids

life of only 1 h, their shedding of monomers apparently keeps concentrations at a sufficient level to facilitate rapid growth of type Bunagitated fibrils. Thus, it appears that the transition from type A to B fibrils probably occurs via shedding and adsorption of monomers.

that

It has been reported that multiple distinct assembly pathways may be responsible for the formation of pro- tofibrillar and mature fibrillar structures of Ab [65], b2-microglobulin [66] and Sup35NM [67]. Clearly, future studies should include kinetics experiments and structural data before concluding that fibrils form via a hierarchical build-up mechanism [68]. Nevertheless, it is unlikely that there will be a single unifying mecha- nism for the build-up of fibrils from its constituents. The diversity of possible interactions due to different protein sequences is simply too great [29]. There are cases where preformed oligomers can be demonstrated to be incorporated directly into the fibrils [69], and kinetic data from SAXS also support insulin fibrils could be built from preformed oligomeric build- ing blocks [70], although the mechanism that leads to a lag time before the accumulation of the building block has not yet been described in detail. As a further example, our SAXS studies of the fibrillation of a-syn- uclein under agitated conditions identify three species, namely a monomer ⁄ dimer state, an oligomer with a central channel and an extended fibril (L. Giehm, D. Svergun, D. E. Otzen and B. Vestergaard, submit- ted). Structurally and mechanistically this oligomer appears to be a direct precursor to the fibril.

Charge neutrality in type B fibrils

type A fibrils grow until

on a lag time versus log[seed] plot) [44]. In contrast, only type A fibrils seed exponential growth at high con- centrations, possibly because either the branching or elongation of type Bunagitated seeds is inhibited by the a-helical trimers that form in equilibrium with mono- mers at these concentrations [64] (Fig. 3A). Consistent with this, the apparent lag time for the formation of type Bunagitated fibrils actually increases with increasing glucagon concentrations above 0.3 gÆL)1 [44]. All of the abovementioned differences make it difficult to propose that type Bagitated fibrils could be assembled by simple lateral associations of several type A protofibrils – the properties of the structures are simply too different. We therefore hypothesize that conversion from type A to mature type Bunagitated fibrils over time could occur by gradual shedding of monomers from unstable type A fibrils that subsequently adsorb to the more stable exponentially growing type Bagitated fibril structure. Data from thermal melting suggest that type A fibrils are relatively unstable compared with type Bagitated fibrils, with apparent thermal melting midpoints (Tapp m ) of < 32 and 55 (cid:2)C, respectively [27]. Moreover, linear extrapolation of urea dissociation kinetics indicates that type A fibrils have a much faster dissociation rate of (cid:2) 0.69 h)1 compared with the 0.03 h)1 observed for type Bagitated fibrils [27]. This corresponds to a half-life of only 1 h if type A fibrils were diluted infinitely in buffer. Proteolysis with pepsin, which continuously degrades flexible monomers more readily than fibrils, shows nearly the same value [27]. Recent developments in small angle X-ray scattering (SAXS) allows noninva- sive quantitative analysis of the relative amounts of fibrils consisting of single (type A) and multiple proto- filaments (type Bunagitated) (green and orange curves in Fig. 3B, respectively) [63], which is very difficult if not impossible to achieve using combinations of ThT and Trp fluorescence alone. Data from the SAXS study they have indicate that the glucagon (blue curve consumed nearly all of in Fig. 3B) before the exponential growth of type Bunagitated fibrils reaches detectable levels. Interestingly, the SAXS data show that the lag time for the forma- tion of type Bunagitated increases from 18 h (5 gÆL)1) to 35 h (10 gÆL)1) [63], indicating that the growth of type Bunagitated depends on the remaining nonfibrillated glucagon concentration rather than on the amount of type A fibrils formed before them [63]. This is inconsis- tent with type A being a structural prerequisite for the formation of type Bunagitated fibrils, but consistent with the quantitative cross-seeding data [44] that suggests type Bunagitated fibrils are unable to grow before mono- mer concentrations are sufficiently low (i.e. due to the formation of type A). Because type A fibrils have a half-

At the acidic pH used for the fibrillogenesis of gluca- gon, histidine residue 1 (His1), the three aspartic acid residues (Asp9, Asp15 and Asp21) and the C-terminus exist mostly in the protonated state. This means that glucagon has a net charge of +5 (N-terminus, His1, Lys12, Arg17 and Arg18). If left unshielded, this would lead to high static repulsion, which is irreconcilable with the close packing of glucagon molecules that occurs in fibrils. There are two possible mechanisms that would allow glucagon fibrils to exist at low pH: either the pKa values are shifted so that glucagon mole- cules in fibrils lose some of these charges or counter ions from the solution shield the positive charges. The protonation state of type Bagitated glucagon fibrils has been investigated by isothermic titration calorimetry (ITC) during extension of seeds [71]. Using a series of buffers with different protonation enthalpies, it was possible to measure how many protons were exchanged with the buffer upon incorporation of a glucagon mole- cule into a seed. The data obtained are consistent with

FEBS Journal 277 (2010) 4591–4601 ª 2010 The Authors Journal compilation ª 2010 FEBS

4596

J. S. Pedersen et al.

The structural ambiguity of glucagon amyloids

tration increases Tapp m with 22 (cid:2)C for, implying that salts are critical to stability of Type S fibrils [43]. This explains why adding as little as 1 mm SO2(cid:3) 4 gives a selec- tive growth advantage to type S fibrils. We have previ- ously speculated that the negative charges on SO2(cid:3) 4 could allow packing of positively charged glucagon mol- ecules into fibrils, as well as lateral associations between several positively charged fibrils or protofilaments thereby increasing stability [43].

A comparison of

Fig. 4. Data from ITC experiments during seed extension indicate that each monomer releases five protons at pH 2.5 and takes up one proton at pH 9.5 upon addition to a type Bagitated fibril end [71].

m

the X-ray diffraction patterns reveals slight differences in the interstrand distance of type B, S and D fibrils with meridionals at 4.7, 4.8 and 4.9 A˚ , respectively [73]. Each fibril type also has their own signature of equatorial reflections. Moreover, Trp25 appears significantly more exposed to acrylamide quenching in type Bagitated fibrils compared with type D and S [73]. Convincing evidence for structural differ- ences between the three types of fibril also comes from cross-seeding experiments under identical conditions (e.g. 1 gÆL)1 monomeric glucagon in 50 mm glycine pH 2.5), which lead to propagation of the structure of the seed [27,43]. We have used ITC to extend this charac- terization with a thermodynamic comparison of type Bagitated, S and D fibrils [73]. By measuring the enthalpy change, DH, for seeded fibril elongation at a series of temperatures, it is possible to estimate the change in heat capacity for fibril formation (DCp) for the three fibril types [74]. Remarkably, the DCp values for the fibril extension of the three types are significantly dif- ferent, with positive values for type D and negative val- ues for type S (Fig. 5). For type Bagitated, the enthalpy

the release of five protons upon monomer addition to a fibril end, indicating that glucagon is charge neutral in the type Bagitated fibrillated state [71] (Fig. 4). Consis- tently, the stability of type Bagitated depends on pH, with fibrils dissociating instantly at pH 1.1 and Tapp increasing from 40 (cid:2)C at pH 2.1 to 61 (cid:2)C at pH 3.2 [27]. Glucagon can also form fibrils in glycine ⁄ NaOH buffer at pH 9.5 [72], where monomeric glucagon is expected to have a net charge of (cid:2) )1 [73], and accord- ing to data from ITC experiments, the molecules also become charge neutral when incorporated into these fibrils [71]. Because of the charge neutrality of fibrils at both high and low pH, it is conceivable that the type of molecular packing of glucagon in fibrils formed at these very different pH values could be identical. The obser- vation that seeds of fibrils formed at pH 9.5 can grow exponentially at pH 2.5, with kinetics that are virtually superimposable to kinetics of type Bagitated seeding sug- gests that type Bagitated fibrils can also form at pH 9.5 (J. S. Pedersen, unpublished data).

Shielding of charges by ions allows fibrils with alternative molecular structures

Fig. 5. Thermodynamic analysis of fibrillar polymorphism. Enthalpy change during extension of various types of glucagon fibril as a function of temperature. The slope of these curves corresponds to DCp values, which are strikingly different for the three types of fibril. Adapted from [73].

Because of the +5 charge on monomeric glucagon mol- ecules at acidic pH, shielding of charges by anions can increase the rate of fibril formation by relieving the charge repulsion between monomers [43]. However, salts appear to favor the growth of fibrils with alterna- tive properties: in the presence of 150–250 mm Cl), type D fibrils appear to have a selective growth advantage over type Bagitated fibrils, even under agitated conditions [27], possibly because the salts also stabilize type Bagitated fibrils, making them less prone to break [27]. The diva- is 125-fold more stabilizing than Cl) for lent anion SO2(cid:3) 4 type S fibrils, and every 10-fold increase in salt concen-

FEBS Journal 277 (2010) 4591–4601 ª 2010 The Authors Journal compilation ª 2010 FEBS

4597

J. S. Pedersen et al.

The structural ambiguity of glucagon amyloids

is dominated by buffer deprotonation, making the intrinsic DCp essentially zero. It is difficult to identify a simple structural basis for this remarkable variation in DCp values. Clearly the predicted change in solvent accessible area, which correlates strongly with DCp for globular proteins [75], is not a useful predictor of fibril- lar DCp. It is possible that strong backbone interactions lead to the unfavorable burial of polar side residues, water and ⁄ or charged groups, which can all have major influence on the change in DCp.

Future perspectives – toxicity of alternative protein folds

So far we have demonstrated that glucagon is able to form at least five types of amyloid fibril that appear to differ at the level of their molecular packing of gluca- gon. Based on some of the structures observed in EM, it is very likely that several other types of fibril with unique properties remain to be discovered. Judging by seed extension kinetics and the overall sigmoidal shape of the growth curves, it appears logical to assume that they all grow exponentially by monomer addition from rare thermodynamic nuclei that start to form when glucagon is dissolved (Fig. 6).

In the current research on protein aggregation and amyloid formation, interest in prefibrillar intermediate structures and other oligomers is growing, given that the toxicity of these species appears to surpass that of

studies

future

mature fibrils [76]. It is possible that the toxicity of aggregates is simply correlated directly with the surface to mass ratio, implying that smaller structures, which have a high surface to mass ratio, are more toxic than large aggregates, which have a small surface to mass ratio. However, it is becoming evident that specific folds in oligomers can be significantly more toxic than others [77]. It has been reported that mishandling of glucagon solutions of > 2 gÆL)1 leads to the formation of toxic aggregates [78]. Moreover, a recent report comparing the toxicity of amyloid structures of several protein hormones indicated that fibril preparations formed during a 14 day incubation at 37 (cid:2)C with slight agitation of 2 gÆL)1 glucagon in the presence of 0.4 mm low relative molecular mass heparin and 5% d-mannitol at pH 5.5 are particularly toxic, resulting in 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl-tetrazolium bromide reduction exceeding that caused by Ab1–42 and Ab1–40 aggregates [16]. Unfortunately, the study did not reveal what specific type(s) of glucagon aggre- gate(s) cause toxicity. Several studies have been aimed at characterizing prefibrillar intermediate species of glucagon and oligomeric structures have been reported in AFM studies [61,62]. However, according to data from field flow fractionation [79], NMR [80], SAXS the benign [63] and dynamic light scattering [44], a-helical trimer, which is in rapid equilibrium with monomers [44,64,81] and crystallizes readily [14,82], is the only oligomeric structure that forms at detectable levels at pH 2.5. It is possible that a shift to pH 5.5, where the molecules have an average charge of +1, could allow glucagon to form more stable toxic oligo- meric species. Another possibility is that the aggre- gated species responsible for toxicity is a type of fibril, which raises the question of what type of fibril is responsible for toxicity. With its fibrillation mecha- nisms and fibrillar polymorphisms being so well under- stood, glucagon appears to be an excellent model system for to further our under- standing of the relationship between protein aggregate structures and toxicity.

Acknowledgements

equilibrium between

trimers

The

Fig. 6. Summary of the five different types of glucagon fibril inves- tigated in detail and the proposed nucleation-dependent pathways that lead to their formation. On each pathway monomers are in equilibrium with individual thermodynamic nuclei, which are the most unstable transition state between monomers and fibrils. According to the monomer concentration dependence of seeded types grow by monomer addition fibril elongation, all fibril [27,43,44]. and a-helical monomers inhibits exponential growth and ⁄ or nucleation of type Bunagitated. It is possible that the growth of other types of mature fibril could similarly be inhibited by a-helical trimers.

We gratefully acknowledge Dr Hans Aage Hjuler and coworkers at Novo Nordisk A ⁄ S for extensive fund- ing over the years as well as generously providing unlimited amounts of the highest possible quality of glucagon samples. We are also grateful to Drs Chris- tian Rischel, Peter Westh and James Flink for fruitful collaborations and stimulating discussions. JSP is sup- ported by the Carlsberg Research Foundation. DEO from the Danish Research acknowledges

support

FEBS Journal 277 (2010) 4591–4601 ª 2010 The Authors Journal compilation ª 2010 FEBS

4598

J. S. Pedersen et al.

The structural ambiguity of glucagon amyloids

Foundation (inSPIN). CBA is supported by a postdoc- toral fellowship financed by The Benzon Foundation and Novo Nordisk.

14 Staub A, Sinn L & Behrens OK (1955) Purification and crystallization of glucagon. J Biol Chem 214, 619–632.

References

15 Nielsen SB, Franzmann M, Basaiawmoit RV, Wimmer R, Mikkelsen JD & Otzen DE (2010) beta-Sheet aggregation of kisspeptin-10 is stimulated by heparin but inhibited by amphiphiles. Biopolymers 93, 678–689. 16 Maji SK, Perrin MH, Sawaya MR, Jessberger S, 1 Soto C (2003) Unfolding the role of protein misfolding in neurodegenerative diseases. Nat Rev Neurosci 4, 49– 60. 2 Lambert MP, Barlow AK, Chromy BA, Edwards C,

Vadodaria K, Rissman RA, Singru PS, Nilsson KP, Simon R, Schubert D et al. (2009) Functional amyloids as natural storage of peptide hormones in pituitary secretory granules. Science 325, 328–332. 17 Maji SK, Schubert D, Rivier C, Lee S, Rivier JE &

Riek R (2008) Amyloid as a depot for the formulation of long-acting drugs. PLoS Biol 6, e17. Freed R, Liosatos M, Morgan TE, Rozovsky I, Trom- mer B, Viola KL et al. (1998) Diffusible, nonfibrillar ligands derived from Abeta1-42 are potent central ner- vous system neurotoxins. Proc Natl Acad Sci USA 95, 6448–6453. 18 Sunde M, Serpell LC, Bartlam M, Fraser PE, Pepys 3 Caughey B & Lansbury PT (2003) Protofibrils, pores,

MB & Blake CC (1997) Common core structure of amyloid fibrils by synchrotron X-ray diffraction. J Mol Biol 273, 729–739. fibrils, and neurodegeneration: separating the responsi- ble protein aggregates from the innocent bystanders. Annu Rev Neurosci 26, 267–298. 19 Tycko R (2006) Molecular structure of amyloid fibrils: 4 Sakono M & Zako T (2010) Amyloid oligomers: forma-

tion and toxicity of Abeta oligomers. FEBS J 277, 1348–1358.

insights from solid-state NMR. Q Rev Biophys 39, 1–55. 20 Fandrich M, Meinhardt J & Grigorieff N (2009) Struc- tural polymorphism of Alzheimer Abeta and other amy- loid fibrils. Prion 3, 89–93.

5 Pepys MB, Hawkins PH, Booth DR, Vigushin DM, Tennent GA, Soutar AK, Totty N, Bguyen O, Blake CC, Terry CJ et al. (1993) Human lysozyme gene muta- tions cause hereditary systemic amyloidosis. Nature 362, 553–557. 21 Meinhardt J, Sachse C, Hortschansky P, Grigorieff N & Fandrich M (2009) Abeta(1-40) fibril polymorphism implies diverse interaction patterns in amyloid fibrils. J Mol Biol 386, 869–877.

22 Jimenez JL, Nettleton EJ, Bouchard M, Robinson CV, Dobson CM & Saibil HR (2002) The protofilament structure of insulin amyloid fibrils. Proc Natl Acad Sci USA 99, 9196–9201. 6 Benson MD, Liepnieks JJ, Yazaki M, Yamashita T, Asl KH, Guenther B & Kluve-Beckerman B (2001) A new human hereditary amyloidosis: the result of a stop- codon mutation in the apolipoprotein AII gene. Genom- ics 72, 272–277. 23 Pedersen JS & Otzen DE (2008) Amyloid a state in 7 Suhr OB, Svendsen IH, Andersson R, Danielsson A˚ ,

many guises: survival of the fittest fibril fold. Protein Sci 17, 2–10. Holmgren G & Ranløv PJ (2003) Hereditary transthyre- tin amyloidosis from a Scandinavian perspective. J Intern Med 254, 225–235. 8 Klintworth GK (1999) Advances in the molecular

genetics of corneal dystrophies. Am J Ophthalmol 128, 747–754. 24 Goldsbury CS, Wirtz S, Muller SA, Sunderji S, Wicki P, Aebi U & Frey P (2000) Studies on the in vitro assembly of a beta 1-40: implications for the search for a beta fibril formation inhibitors. J Struct Biol 130, 217–231.

9 Stefani M (2010) Biochemical and biophysical features of both oligomer ⁄ fibril and cell membrane in amyloid cytotoxicity. FEBS J 277, 4602–4613. 10 Dobson CM (1999) Protein misfolding, evolution and 25 Kumar S & Udgaonkar JB (2009) Structurally distinct amyloid protofibrils form on separate pathways of aggregation of a small protein. Biochemistry 48, 6441– 6449. disease. Trends Biochem Sci 24, 329–332.

26 Milton NG & Harris JR (2009) Polymorphism of amy- loid-beta fibrils and its effects on human erythrocyte catalase binding. Micron 40, 800–810. 11 Fandrich M & Dobson CM (2002) The behaviour of polyamino acids reveals an inverse side chain effect in amyloid structure formation. EMBO J 21, 5682– 5690.

12 Otzen DE & Nielsen PH (2008) We find them here, we find them there: functional bacterial amyloid. Cell Mol Life Sci 65, 910–927. 27 Pedersen JS, Dikov D, Flink JL, Hjuler HA, Christian- sen G & Otzen DE (2006) The changing face of gluca- gon fibrillation: structural polymorphism and conformational imprinting. J Mol Biol 355, 501–523. 28 Anfinsen CB (1973) Principles that govern the folding 13 Waugh DF, Wilhelmson DF, Commerford SL & of protein chains. Science 181, 223–230.

FEBS Journal 277 (2010) 4591–4601 ª 2010 The Authors Journal compilation ª 2010 FEBS

4599

29 Khurana R, Ionescu-Zanetti C, Pope M, Li J, Nielson L, Ramirez-Alvarado M, Regan L, Fink AL & Carter Sackler ML (1953) Studies of the nucleation and growth reactions of selected types of insulin fibrils. J Am Chem Soc 75, 2582–2600.

J. S. Pedersen et al.

The structural ambiguity of glucagon amyloids

terminism in protein misfolding: chiral amplification on amyloidogenic pathway of insulin. J Am Chem Soc 129, 7517–7522. SA (2003) A general model for amyloid fibril assembly based on morphological studies using atomic force microscopy. Biophys J 85, 1135–1144. 30 Bessen RA, Kocisko DA, Raymond GJ, Nandan S, 43 Pedersen JS, Flink JM, Dikov D & Otzen DE (2006)

Sulfates dramatically stabilize a salt dependent type of glucagon fibrils. Biophys J 90, 4181–4194. Lansbury PT & Caughey B (1995) Non-genetic propa- gation of strain-specific properties of scrapie prion pro- tein. Nature 375, 698–700.

44 Andersen CB, Otzen D, Christiansen G & Rischel C (2007) Glucagon amyloid-like fibril morphology is selected via morphology-dependent growth inhibition. Biochemistry 46, 7314–7324. 31 Tanaka M, Chien P, Naber N, Cooke R & Weissman JS (2004) Conformational variations in an infectious protein determine prion strain differences. Nature 428, 323–328. 32 Toyama BH, Kelly MJ, Gross JD & Weissman JS 45 Uversky VN & Dunker AK (2010) Understanding pro- tein non-folding. Biochim Biophys Acta 1804, 1231– 1264. (2007) The structural basis of yeast prion strain vari- ants. Nature 449, 233–237.

46 Uversky VN (2009) Intrinsically disordered proteins and their environment: effects of strong denaturants, temperature, pH, counter ions, membranes, binding partners, osmolytes, and macromolecular crowding. Protein J 28, 305–325. 47 Beaven GH, Gratzer WB & Davies HG (1969) Forma-

33 Dzwolak W, Smirnovas V, Jansen R & Winter R (2004) Insulin forms amyloid in a strain-dependent manner: an FT-IR spectroscopic study. Protein Sci 13, 1927–1932. 34 Yamaguchi K, Takahashi S, Kawai T, Naiki H & Goto Y (2005) Seeding-dependent propagation and matura- tion of amyloid fibril conformation. J Mol Biol 352, 952–960. tion and structure of gels and fibrils from glucagon. Eur J Biochem 11, 37–42. 48 Roberts CJ (2007) Non-native protein aggregation kinetics. Biotechnol Bioeng 98, 927–938. 49 Fandrich M (2007) Absolute correlation between lag 35 Petkova AT, Leapman RD, Guo Z, Yau WM, Mattson MP & Tycko R (2005) Self-propagating, molecular-level polymorphism in Alzheimer’s beta-amyloid fibrils. Sci- ence 307, 262–265.

time and growth rate in the spontaneous formation of several amyloid-like aggregates and fibrils. J Mol Biol 365, 1266–1270.

36 Madine J, Jack E, Stockley PG, Radford SE, Serpell LC & Middleton DA (2008) Structural insights into the polymorphism of amyloid-like fibrils formed by region 20–29 of amylin revealed by solid-state NMR and X-ray fiber diffraction. J Am Chem Soc 130, 14990–15001. 50 Munishkina LA, Henriques J, Uversky VN & Fink AL (2004) Role of protein–water interactions and electro- statics in alpha-synuclein fibril formation. Biochemistry 43, 3289–3300. 51 Ferrone F (1999) Analysis of protein aggregation kinet- 37 Nielsen JT, Bjerring M, Jeppesen M, Pedersen RO, ics. Methods Enzymol 309, 256–274. 52 Ban T, Hamada D, Hasegawa K, Naiki H & Goto Y

(2003) Direct observation of amyloid fibril growth mon- itored by thioflavin T fluorescence. J Biol Chem 278, 16462–16465. Pedersen JM, Hein KL, Vosegaard T, Skrydstrup TS, Otzen DE & Nielsen NC (2009) Unique identification of supramolecular structures in amyloid fibrils by solid-state NMR. Angew Chem Int Ed Engl 48, 2118–2121. 38 Nelson R, Sawaya MR, Balbirnie M, Madsen AØ,

Riekel C, Grothe R & Eisenberg D (2005) Structure of the cross-beta spine of amyloid-like fibrils. Nature 435, 773–778. 39 Van der Wel PCA, Lewandowski JR & Griffin RG 53 Andersen CB, Yagi H, Manno M, Martorana V, Ban T, Christiansen G, Otzen DE, Goto Y & Rischel C (2009) Branching in amyloid fibril growth. Biophys J 96, 1529–1536.

54 Nielsen L, Khurana R, Coats A, Frokjaer S, Brange J, Vyas S, Uversky VN & Fink AL (2001) Effect of envi- ronmental factors on the kinetics of insulin fibril forma- tion: elucidation of the molecular mechanism. Biochemistry 40, 6036–6046.

(2007) Solid-state NMR study of amyloid nanocrystals and fibrils formed by the peptide GNNQQNY from yeast prion protein Sup35p. J Am Chem Soc 129, 5117–5130. 40 Verel R, Tomka IT, Bertozzi C, Cadalbert R, Kammer- er RA, Steinmetz MO & Meier BH (2008) Polymor- phism in an amyloid-like fibril-forming model peptide. Angew Chem Int Ed Engl 47, 5842–5845. 41 Steinmetz MO, Gattin Z, Verel R, Ciani B, Stromer T, 55 Collins SR, Douglass A, Vale RD & Weissman JS (2004) Mechanism of prion propagation: amyloid growth occurs by monomer addition. PLoS Biol 2, e321.

FEBS Journal 277 (2010) 4591–4601 ª 2010 The Authors Journal compilation ª 2010 FEBS

4600

Green JM, Tittmann P, Schulze-Briese C, Gross H, Van Gunsteren WF et al. (2007) Atomic models of de novo designed cc beta-Met amyloid-like fibrils. J Mol Biol 376, 898–912. 42 Dzwolak W, Loksztejn A, Galinska-Rakoczy A, Adachi R, Goto Y & Rupnicki L (2007) Conformational inde- 56 Andersen CB, Hicks MR, Vetri V, Vandahl B, Rahbek- Nielsen H, Thogersen H, Thogersen IB, Enghild JJ, Serpell LC, Rischel C et al. (2010) Glucagon fibril poly- morphism reflects differences in protofilament backbone structure. J Mol Biol 397, 932–946.

J. S. Pedersen et al.

The structural ambiguity of glucagon amyloids

Svergun DI (2007) A helical structural nucleus is the primary elongating unit of insulin amyloid fibrils. PLoS Biol 5, e134. 57 Loksztejn A & Dzwolak W (2010) Vortex-induced for- mation of insulin amyloid superstructures probed by time-lapse atomic force microscopy and circular dichro- ism spectroscopy. J Mol Biol 395, 643–655.

71 Jeppesen MD, Westh P & Otzen DE (2010) The role of protonation in protein fibrillation. FEBS Lett 584, 780– 784. 72 Pedersen JS, Dikov D & Otzen DE (2006) N- and C- 58 Makarava N & Baskakov IV (2008) The same primary structure of the prion protein yields two distinct self- propagating states. J Biol Chem 283, 15988–15996. 59 Paravastu AK, Leapman RD, Yau WM & Tycko R terminal hydrophobic patches are involved in fibrillation of glucagon. Biochemistry 45, 14503–14512.

(2008) Molecular structural basis for polymorphism in Alzheimer’s beta-amyloid fibrils. Proc Natl Acad Sci USA 105, 18349–18354. 73 Jeppesen MD, Hein KL, Nissen P, Westh P & Otzen D (2010) A thermodynamic analysis of fibrillar polymor- phism. Biophys Chem 149, 40–46. 74 Kardos J, Yamamoto K, Hasegawa K, Naiki H &

Goto Y (2004) Direct measurement of the thermody- namic parameters of amyloid formation by isothermal titration calorimetry. J Biol Chem 279, 55308–55314. 75 Myers JK, Pace CN & Scholtz JM (1995) Denaturant 60 Yagi H, Ban T, Morigaki K, Naiki H & Goto Y (2007) Visualization and classification of amyloid beta supra- molecular assemblies. Biochemistry 46, 15009–15017. 61 Dong M, Hovgaard MB, Xu S, Otzen DE & Besenb- acher F (2006) AFM study of glucagon fibrillation via oligomeric structures resulting in interwoven fibrils. Nanotechnology 17, 4003.

m values and heat capacity changes: relation to changes in accessible surface areas of protein unfolding. Protein Sci 4, 2138–2148.

62 De Jong KL, Incledon B, Yip CM & Defelippis MR (2006) Amyloid fibrils of glucagon characterized by high-resolution atomic force microscopy. Biophys J 91, 1905–1914.

76 Kayed R, Head E, Thompson JL, McIntire TM, Milton SC, Cotman CW & Glabe CG (2003) Common struc- ture of soluble amyloid oligomers implies common mechanism of pathogenesis. Science 300, 486–489. 77 Campioni S, Mannini B, Zampagni M, Pensalfini A, 63 Oliveira CL, Behrens MA, Pedersen JS, Erlacher K, Otzen D & Pedersen JS (2009) A SAXS study of glucagon fibrillation. J Mol Biol 387, 147–161. 64 Wu CS & Yang JT (1980) Helical conformation of

glucagon in surfactant solutions. Biochemistry 19, 2117– 2122. Parrini C, Evangelisti E, Relini A, Stefani M, Dobson CM, Cecchi C et al. (2010) A causative link between the structure of aberrant protein oligomers and their toxicity. Nat Chem Biol 6, 140–147. 78 Onoue S, Ohshima K, Debari K, Koh K, Shioda S,

65 Goldsbury C, Frey P, Olivieri V, Aebi U & Muller SA (2005) Multiple assembly pathways underlie amyloid- beta fibril polymorphisms. J Mol Biol 352, 282–298. 66 Gosal WS, Morten IJ, Hewitt EW, Smith DA, Thom- son NH & Radford SE (2005) Competing pathways determine fibril morphology in the self-assembly of beta2-microglobulin into amyloid. J Mol Biol 351, 850– 864. Iwasa S, Kashimoto K & Yajima T (2004) Mishandling of the therapeutic peptide glucagon generates cytotoxic amyloidogenic fibrils. Pharm Res 21, 1274–1283. 79 Hoppe CC, Nguyen LT, Kirsch LE & Wiencek JM (2008) Characterization of seed nuclei in glucagon aggregation using light scattering methods and field- flow fractionation. J Biol Eng 2, 10. 80 Svane AS, Jahn K, Deva T, Malmendal A, Otzen DE,

67 Hess S, Lindquist SL & Scheibel T (2007) Alternative assembly pathways of the amyloidogenic yeast prion determinant Sup35-NM. EMBO Rep 8, 1196–1201. 68 Kodali R & Wetzel R (2007) Polymorphism in the Dittmer J & Nielsen NC (2008) Early stages of amyloid fibril formation studied by liquid-state NMR: the peptide hormone glucagon. Biophys J 95, 366–377. 81 Formisano S, Johnson ML & Edelhoch H (1977) intermediates and products of amyloid assembly. Curr Opin Struct Biol 17, 48–57. 69 Plakoutsi G, Bemporad F, Calamai M, Taddei N, Thermodynamics of the self-association of glucagon. Proc Natl Acad Sci USA 74, 3340–3344.

FEBS Journal 277 (2010) 4591–4601 ª 2010 The Authors Journal compilation ª 2010 FEBS

4601

Dobson CM & Chiti F (2005) Evidence for a mechanism of amyloid formation involving molecular reorganisation within native-like precursor aggregates. J Mol Biol 351, 910–922. 82 Sasaki K, Dockerill S, Adamiak DA, Tickle IJ & Blundell T (1975) X-ray analysis of glucagon and its relationship to receptor binding. Nature 257, 751–757. 70 Vestergaard B, Groenning M, Roessle M, Kastrup JS, van de Weert M, Flink JM, Frokjaer S, Gajhede M &