Destabilization of human IAPP amyloid fibrils by proline mutations outside of the putative amyloidogenic domain: is there a critical amyloidogenic domain in human IAPP?

Islet amyloid polypeptide (IAPP; amylin) is responsible for amyloid formation in type-2 diabetes. Not all organisms form islet amyloid, and amyloid formation correlates strongly with variations in primary sequence. Studies of human and rodent IAPP have pointed to the amino acid residues 20-29 region as the important amyloid-modulating sequence. The rat 20-29 sequence contains three proline residues and does not form amyloid, while the human sequence contains no proline and readily forms amyloid. This has led to the view that the 20-29 region constitutes a critical amyloidogenic domain that dictates the properties of the entire sequence. The different behavior of human and rat IAPP could be due to differences in the 20-29 region or due simply to the fact that multiple proline residues destabilize amyloid fibrils. We tested how critical the 20-29 region is by studying a variant identical with the human peptide in this segment but with three proline residues outside this region. We designed a variant of the amyloidogenic 8-37 region of human IAPP (hIAPP(8-37) 3xP) with proline substitutions at positions 17, 19 and 30. Compared to the wild-type, the 3xP variant was much easier to synthesize and had dramatically greater solubility. Fourier transform infra red spectroscopy, transmission electron microscopy, Congo red staining and thioflavin-T binding indicate that this variant has a reduced tendency to form beta-sheet structure and forms deposits with much less structural order than the wild-type. Far-UV CD studies show that the small amount of beta-sheet structure developed by hIAPP(8-37) 3xP after long periods of incubation dissociates readily into random-coil structure upon dilution into Tris buffer. The observation that proline substitutions outside the putative core domain effectively abolish amyloid formation indicates that models of IAPP aggregation must consider contributions from other regions.     

Transformation of the amyloidogenic peptide amylin(20-29) into its corresponding peptoid and retropeptoid: access to both an amyloid inhibitor and template for self-assembled supramolecular tapes

The highly amyloidogenic peptide sequence of amylin(20-29) was transformed into its corresponding peptoid and retropeptoid sequences to design a novel class of beta-sheet breaker peptides as amyloid inhibitors. This report describes the synthesis of the chiral peptoid building block of L-isoleucine, the solid phase synthesis of the peptoid and retropeptoid sequences of amylin(20-29), and the structural analysis of these amylin derivatives in solution by infrared spectroscopy, circular dichroism, and transmission electron microscopy. It was found that the peptoid sequence did not form amyloid fibrils or any other secondary structures and was able to inhibit amyloid formation of native amylin(20-29). Although the retropeptoid did not form amyloid fibrils it had only modest amyloid inhibitor properties since supramolecular tapes were formed.     

Backbone-modified amylin derivatives: implications for amyloid inhibitor design and as template for self-assembling bionanomaterials.

This report reviews our approach to the design, synthesis and structural/morphological analysis of backbone-modified amylin(20-29) derivatives. Depending on the position in the peptide backbone and the type of amide bond isostere/modification, the amylin(20-29) peptides behave either as inhibitors of amyloid fibril formation, which are able to retard amyloid formation of native amylin(20-29), or as templates for the formation of self-assembled supramolecular structures. Molecular fine-tuning of the hydrogen-bond accepting/donating properties allows the control over the morphology of the supramolecular aggregation motifs such as helical ribbons and tapes, ribbons progressing to closed peptide nanotubes, (twisted) lamellar sheets or amyloid fibrils.     

Atomic force microscopy reveals defects within mica supported lipid bilayers induced by the amyloidogenic human amylin peptide

To date, over 20 peptides or proteins have been identified that can form amyloid fibrils in the body and are thought to cause disease. The mechanism by which amyloid peptides cause the cytotoxicity observed and disease is not understood. However, one of the major hypotheses is that amyloid peptides cause membrane perturbation. Hence, we have studied the interaction between lipid bilayers and the 37 amino acid residue polypeptide amylin, which is the primary constituent of the pancreatic amyloid associated with type 2 diabetes. Using a dye release assay we confirmed that the amyloidogenic human amylin peptide causes membrane disruption; however, time-lapse atomic force microscopy revealed that this did not occur by the formation of defined pores. On the contrary, the peptide induced the formation of small defects spreading over the lipid surface. We also found that rat amylin, which has 84% identity with human amylin but cannot form amyloid fibrils, could also induce similar lesions to supported lipid bilayers. The effect, however, for rat amylin but not human amylin, was inhibited under high ionic conditions. These data provide an alternative theory to pore formation, and how amyloid peptides may cause membrane disruption and possibly cytotoxicity.     

Design of peptide-based inhibitors of human islet amyloid polypeptide fibrillogenesis

Human islet amyloid polypeptide (IAPP) is the major component of amyloid deposits found in the pancreas of over 90% of all cases of type-2 diabetes. We have generated a series of overlapping hexapeptides to target an amyloidogenic region of IAPP (residues 20-29) and examined their effects on fibril assembly. Peptide fragments corresponding to SNNFGA (residues 20-25) and GAILSST (residues 24-29) were strong inhibitors of the beta-sheet transition and amyloid aggregation. Circular dichroism indicated that even at 1:1 molar ratios, these peptides maintained full-length IAPP (1-37) in a largely random coil conformation. Negative stain electron microscopy revealed that co-incubation of these peptides with IAPP resulted in the formation of only semi-fibrous aggregates and loss of the typical high density and morphology of IAPP fibrils. This inhibitory activity, particularly for the SNNFGA sequence, also correlated with a reduction in IAPP-induced cytotoxicity as determined by cell culture studies. In contrast, the peptide NFGAIL (residues 22-27) enhanced IAPP fibril formation. Conversion to the amyloidogenic beta-sheet was immediate and the accompanying fibrils were more dense and complex than IAPP alone. The remaining peptide fragments either had no detectable effects or were only weakly inhibitory. Specificity of peptide activity was illustrated by the fragments, SSNNFG and AILSST. These differed from the most active inhibitors by only a single amino acid residue but delayed the random-to-beta conformational change only when used at higher molar ratios. This study has identified internal IAPP peptide fragments which can regulate fibrillogenesis and may be of therapeutic use for the treatment of type-2 diabetes.     

Effects of sequential proline substitutions on amyloid formation by human amylin20-29

Amylin, also known as islet amyloid polypeptide (IAPP), is the major protein component of the fibril deposits found in the pancreas of individuals with type II diabetes. The central region of amylin, residues 20-29, has been implicated as a key determinate of amyloid formation. To establish which positions are most important for amyloid formation, the wild-type sequence of the 20-29 fragment and a set of 10 variants have been synthesized in which a proline was placed at each position. Proline is energetically unfavorable in the extended cross-beta structure found in amyloid. If a particular position is critical for amyloid formation, then substitution with a proline should inhibit amyloid formation. A proline substitution at any position inhibited aggregation and amyloid formation. Substitution of Asn22, Gly24, and residues 26-28 had the largest effect. Fourier transform infrared (FTIR) spectroscopy showed little secondary structure in these peptides, and transmission electron microscopy (TEM) showed mostly amorphous material. The peptides were much more soluble than the wild-type sequence, and no birefringence was observed with Congo Red staining. Proline substitutions at the N (residues 20 and 21) and C termini showed the least effect. These peptides showed the classic fibril morphology, a significant amount of beta-sheet structure, and exhibited green birefringence when stained with Congo Red. The results indicate that residues 22, 24, and 26-28 play a key role in formation of amyloid by amylin. Positions 23 and 25 also appear to be important, but may be less critical than positions 22, 24, and 26-28.     

Identification of a penta- and hexapeptide of islet amyloid polypeptide (IAPP) with amyloidogenic and cytotoxic properties

Pancreatic amyloid is found in more than 95 % of type II diabetes patients. Pancreatic amyloid is formed by the aggregation of islet amyloid polypeptide (hIAPP or amylin), which is a 37-residue peptide. Because pancreatic amyloid is cytotoxic, it is believed that its formation is directly associated with the development of the disease. We recently showed that hIAPP amyloid formation follows the nucleation-dependent polymerization mechanism and proceeds via a conformational transition of soluble hIAPP into aggregated beta-sheets. Here, we report that the penta- and hexapeptide sequences, hIAPP(23-27) (FGAIL) and hIAPP(22-27) (NFGAIL) of hIAPP are sufficient for the formation of beta-sheet-containing amyloid fibrils. Although these two peptides differ by only one amino acid residue, they aggregate into completely different fibrillar assemblies. hIAPP(23-27) (FGAIL) fibrils self-assemble laterally into unusually broad ribbons, whereas hIAPP(22-27) (NFGAIL) fibrils coil around each other in a typical amyloid fibril morphology. hIAPP(20-27) (SNNFGAIL) also aggregates into beta-sheet-containing fibrils, whereas no amyloidogenicity is found for hIAPP(24-27) (GAIL), indicating that hIAPP(23-27) (FGAIL) is the shortest fibrillogenic sequence of hIAPP. Insoluble amyloid formation by the partial hIAPP sequences followed kinetics that were consistent with a nucleation-dependent polymerization mechanism. hIAPP(22-27) (NFGAIL), hIAPP(20-27) (SNNFGAIL), and also the known fibrillogenic sequence, hIAPP(20-29) (SNNFGAILSS) exhibited significantly lower kinetic and thermodynamic solubilities than the pentapeptide hIAPP(23-27) (FGAIL). Fibrils formed by all short peptide sequences and also by hIAPP(20-29) were cytotoxic towards the pancreatic cell line RIN5fm, whereas no cytotoxicity was observed for the soluble form of the peptides, a notion that is consistent with hIAPP cytotoxicity. Our results suggest that a penta- and hexapeptide sequence of an appropriate amino acid composition can be sufficient for beta-sheet and amyloid fibril formation and cytotoxicity and may assist in the rational design of inhibitors of pancreatic amyloid formation or other amyloidosis-related diseases.     

Full-length rat amylin forms fibrils following substitution of single residues from human amylin

Pancreatic amyloid deposits, composed of the 37 amino acid residue peptide amylin, represent an integral part of type 2 diabetes mellitus pathology. Human amylin (hA) forms fibrils in vitro and is toxic to cultured pancreatic islet beta-cells. In contrast, rat amylin (rA) which differs from hA by only six amino acid residues in the central region of the peptide, residues 18-29, does not form fibrils and is not cytotoxic. To elucidate the role of individual residues in fibril formation, we have generated a series of full-length rA variants and examined their ability to form fibrils in vitro. Single-residue substitutions with amino acids from corresponding positions of the hA sequence, i.e. R18H, L23F, or V26I, were sufficient to render rA competent for fibril formation albeit at a small yield. Combining two or three of these substitutions generally increased the ability to produce fibrils. Variant rA fibril morphologies were examined by negative stain electron microscopy and found to be similar to those generated by hA itself. Bulk assays, i.e. involving thioflavin-T fluorescence and sedimentation, showed that the amount of fibril formation was relatively small for these rA variants when compared to hA under the same conditions. Fibril growth was demonstrated by time-lapse atomic force microscopy, and MALDI-TOF mass spectrometry was used to verify that fibrils consisted of full-length peptide. Our observations confirm previous reports that the three proline residues play a dominant negative role in fibril formation. However, their presence is not sufficient to completely abolish the ability of rA to form fibrils, as each of the other three implicated residues (i.e. R18, L23 and V26) also has a dominant modulating effect.     

Atomic force microscopy study of human amylin (20-29) fibrils

Here we present atomic force microscopy images of the fibrils formed by human amylin(20-29). This peptide is a fragment of the polypeptide amylin, the major proteinaceous component of amyloid deposits found in cases of type-II diabetes mellitus. Our results demonstrate that the amylin(20-29) peptide fragment forms amyloid-like fibrils that display polymorphic structures. Twisting along the axis of fibrils was often observed in fibrils aged for 6 hours but disappeared in mature fibrils aged for longer time periods.     

The role of protein sequence and amino acid composition in amyloid formation: scrambling and backward reading of IAPP amyloid fibrils

The specific functional structure of natural proteins is determined by the way in which amino acids are sequentially connected in the polypeptide. The tight sequence/structure relationship governing protein folding does not seem to apply to amyloid fibril formation because many proteins without any sequence relationship have been shown to assemble into very similar β-sheet-enriched structures. Here, we have characterized the aggregation kinetics, seeding ability, morphology, conformation, stability, and toxicity of amyloid fibrils formed by a 20-residue domain of the islet amyloid polypeptide (IAPP), as well as of a backward and scrambled version of this peptide. The three IAPP peptides readily aggregate into ordered, β-sheet-enriched, amyloid-like fibrils. However, the mechanism of formation and the structural and functional properties of aggregates formed from these three peptides are different in such a way that they do not cross-seed each other despite sharing a common amino acid composition. The results confirm that, as for globular proteins, highly specific polypeptide sequential traits govern the assembly pathway, final fine structure, and cytotoxic properties of amyloid conformations.     

Neurotoxicity of Human Amylin in Rat Primary Hippocampal Cultures: Similarity to Alzheimer's Disease Amyloid-β Neurotoxicity

Amylin, a 37-amino-acid amyloidogenic peptide, bears biophysical similarities to the amyloid-β peptide (Aβ) deposited in Alzheimer's disease. Using embryonic rat hippocampal cultures we tested whether amylin induces neurotoxicity similar to that previously observed with Aβ(1–40). Treatment with human amylin (1–37) resulted in prominent toxicity as assessed by phasecontrast microscopy and quantification of lactate dehydrogenase in the medium. Amylin-induced neurotoxicity was morphologically similar to that induced by Aβ(1–40). In contrast, the nonamyloidogenic rat amylin showed negligible neurotoxicity despite having 95% sequence similarity to human amylin. Only full-length human amylin was toxic; various amylin peptide fragments including amino acid residues 20–29 were nontoxic at similar concentrations. These studies suggest that unrelated amyloidogenic peptides like human amylin and Aβ can adopt a similar neurotoxic conformation in vitro. Similar conformation-dependent neurotoxicity may drive the prominent neurite degeneration around compacted but not diffuse deposits of Aβ in Alzheimer's disease.     

Conformational preferences of the amylin nucleation site in SDS micelles: an NMR study

Human islet amyloid polypeptide (hIAPP), or amylin, is a 37 amino acid hormone secreted by pancreatic beta-cells. hIAPP constitutes approximately 90% of the amyloid deposits found in type II diabetic patients. It has been shown that the central region of the peptide (hIAPP(20-29)) constitutes the nucleation site for the amyloidogenic process with F23 playing a key role in the formation of the beta-pleated structures. In addition, it has been proposed that an important stage in the cytotoxicity of hIAPP is its interaction with the beta-cell membranes. As a first step toward the characterization of the interaction of hIAPP with cell membranes, we determined conformational preferences of hIAPP(20-29) in membrane-mimicking environments. We found that upon interacting with negatively charged micelles, the dominant conformation of hIAPP(20-29) is a distorted type I beta-turn centered on residues F23 and G24, with F23, A25, and I26 forming a small hydrophobic cluster that may facilitate the interaction of this peptide with the membrane bilayer. Moreover, we were able to elucidate the topological orientation of the peptide that is absorbed on the micelle surface, with the hydrophobic cluster oriented toward the hydrocarbon region of the micelles and both N- and C-termini exposed to the solvent.     

Copper(II) inhibits the formation of amylin amyloid in vitro

The amyloidogenic peptide amylin is found associated with pancreatic islet beta-cells and is implicated in the aetiology of type-2 diabetes mellitus. We have used fluorimetry and transmission electron microscopy to investigate in vitro the influence of Al(III), Fe(III), Zn(II) and Cu(II) on amylin amyloid formation under near-physiological conditions. Cu(II) at 10.0 microM inhibited amylin of 0.4 and 2.0 microM from forming amyloid fibrils while the same concentration of either Al(III) or Zn(II) promoted the formation of beta-pleated sheet structures. If amylin amyloid is cytotoxic to beta-cells then Cu(II) should protect against the degeneration of the islets in type-2 diabetes mellitus.     

Promotion of formation of amyloid fibrils by aluminium adenosine triphosphate (AlATP)

The formation of amyloid fibrils is considered to be an important step in the aetiology of Alzheimer's disease and other amyloidoses. Fibril formation in vitro has been shown to depend on many different factors including modifications to the amino acid profile of fibrillogenic peptides and interactions with both large and small molecules of physiological significance. How these factors might contribute to amyloid fibril formation in vivo is not clear as very little is known about the promotion of fibril formation in undersaturated solutions of amyloidogenic peptides. We have used thioflavin T fluorescence and reverse phase high performance liquid chromatography to show that ATP, and in particular AlATP, promoted the formation of thioflavin T-reactive fibrils of beta amyloid and, an unrelated amyloidogenic peptide, amylin. Evidence is presented that induction of fibril formation followed the complexation of AIATP by one or more monomers of the respective peptide. However, the complex formed could not be identified directly and it is suggested that AlATP might be acting as a chaperone in the assembly of amyloid fibrils. The effect of AlATP was not mimicked by either AlADP or AlAMP. However, it was blocked by suramin, a P2 ATP receptor antagonist, and this has prompted us to speculate that the precursor proteins to beta amyloid and amylin may be substrates or receptors for ATP in vivo.     

The dye SYPRO orange binds to amylin amyloid fibrils but not pre‐fibrillar intermediates

Amyloid deposition underlies a broad range of diseases including multiple neurodegenerative diseases, systemic amyloidosis and type‐2 diabetes. Amyloid sensitive dyes, particularly thioflavin‐T, are widely used to detect ex‐vivo amyloid deposits, to monitor amyloid formation in vitro and to follow the kinetics of amyloid self‐assembly. We show that the dye SYPRO‐orange binds to amyloid fibrils formed by human amylin, the polypeptide responsible for islet amyloid formation in type‐2 diabetes. No fluorescence enhancement is observed in the presence of pre‐fibrillar species or in the presence of non‐amyloidogenic rat amylin. The kinetics of human amylin amyloid formation can be monitored by SYPRO‐orange fluorescence and match the time course determined with thioflavin‐T assays. Thus, SYPRO‐orange offers an alternative to thioflavin‐T assays of amylin amyloid formation. The implications for the interpretation of SYPRO‐orange‐based assays of protein stability and protein‐ligand interactions are discussed.     

Identification and characterization of a novel molecular-recognition and self-assembly domain within the islet amyloid polypeptide

The islet amyloid polypeptide (hIAPP) is a 37 amino acid residue polypeptide that was found to accumulate as amyloid fibrils in the pancreas of individuals with type II diabetes. Previous studies identified various fragments of hIAPP that can form amyloid fibrils in vitro (e.g. hIAPP(8-20), hIAPP(23-27), and hIAPP(30-37)). However, no comparative and systematic information was available on the role of these structural domains (or others) in the process of molecular recognition that mediates fibrillization, in the context of the full-length polypeptide. To systematically map and compare potential recognition domains, we studied the ability of hIAPP to interact with an array of 28 membrane-spotted overlapping peptides that span the entire sequence of hIAPP (i.e. hIAPP(1-10), hIAPP(2-11...), hIAPP(28-37)). Our study clearly identified a major domain of molecular recognition within hIAPP, as the polypeptide was found to bind with high affinity to a defined linear group of peptides ranging from hIAPP(7-16) to hIAPP(12-21). The maximal binding of the full-length polypeptide was to the hIAPP(11-20) peptide fragment (with the sequence RLANFLVHSS). In order to define the minimal fragment, within this apparent recognition motif, that is capable of self-association and thus may serve as the core molecular recognition motif, we examined the ability of truncated analogs of the recognition sequence to self-assemble into amyloid fibrils. The shortest active fragments capable of self-assembly were found to be the pentapeptides FLVHS and NFLVH. The apparent role of this motif in the process of hIAPP self-assembly is consistent with the profile of the hIAAP-binding distribution to the peptide array. The identification of such short recognition motifs is extremely useful in the attempts to develop means to block amyloid fibril formation by hIAPP. It is worth mentioning that this is only the second time in which peptides as short as a pentapeptide were shown to form amyloid fibrils (the other pentapeptide is FGAIL).     

Sensitivity of amyloid formation by human islet amyloid polypeptide to mutations at residue 20

Islet amyloid polypeptide (IAPP, amylin) is responsible for amyloid formation in type 2 diabetes and in islet cell transplants. The only known natural mutation found in mature human IAPP is a Ser20-to-Gly missense mutation, found with small frequency in Chinese and Japanese populations. The mutation appears to be associated with increased risk of early-onset type 2 diabetes. Early measurements in the presence of organic co-solvents showed that S20G-IAPP formed amyloid more quickly than the wild type. We confirm that the mutant accelerates amyloid formation under a range of conditions including in the absence of co-solvents. Ser20 adopts a normal backbone geometry, and the side chain makes no steric clashes in models of IAPP amyloid fibers, suggesting that the increased rate of amyloid formation by the mutant does not result from the relief of steric incompatibility in the fiber state. Transmission electronic microscopy, circular dichroism, and seeding studies were used to probe the structure of the resulting fibers. The S20G-IAPP peptide is toxic to cultured rat INS-1 (transformed rat insulinoma-1) β-cells. The sensitivity of amyloid formation to the identity of residue 20 was exploited to design a variant that is much slower to aggregate and that inhibits amyloid formation by wild-type IAPP. An S20K mutant forms amyloid with an 18-fold longer lag phase in homogeneous solution. Thioflavin T binding assays, together with experiments using a p-cyanophenylalanine (p-cyanoPhe) variant of human IAPP, show that the designed S20K mutant inhibits amyloid formation by human IAPP. The experiments illustrate how p-cyanoPhe can be exploited to monitor amyloid formation even in the presence of other amyloidogenic proteins.     

Morin hydrate inhibits amyloid formation by islet amyloid polypeptide and disaggregates amyloid fibers

The polypeptide hormone Islet Amyloid Polypeptide (IAPP, amylin) is responsible for islet amyloid formation in type-2 diabetes and in islet cell transplants, where it may contribute to graft failure. Human IAPP is extremely amyloidogenic and fewer inhibitors of IAPP amyloid formation have been reported than for the Alzheimer's Aβ peptide or for α-synuclein. The ability of a set of hydroxyflavones to inhibit IAPP amyloid formation was tested. Fluorescence detected thioflavin-T-binding assays are the most popular methods for measuring the kinetics of amyloid formation and for screening potential inhibitors; however, we show that they can lead to false positives with hydroxyflavones. Several of the compounds inhibit thioflavin-T fluorescence, but not amyloid formation; a result which highlights the hazards of relying solely on thioflavin-T assays to screen potential inhibitors. Transmission electron microscopy (TEM) and right-angle light scattering show that Morin hydrate (2',3,4',5,7-Pentahydroxyflavone) inhibits amyloid formation by human IAPP and disaggregates preformed IAPP amyloid fibers. In contrast, Myricetin, Kaempferol, and Quercetin, which differ only in hydroxyl groups on the B-ring, are not effective inhibitors. Morin hydrate represents a new type of IAPP amyloid inhibitor and the results with the other compounds highlight the importance of the substitution pattern on the B-ring.     

Effect of sequence variation on the mechanical response of amyloid fibrils probed by steered molecular dynamics simulation

The mechanical failure of mature amyloid fibers produces fragments that act as seeds for the growth of new fibrils. Fragmentation may also be correlated with cytotoxicity. We have used steered atomistic molecular dynamics simulations to study the mechanical failure of fibrils formed by the amyloidogenic fragment of human amylin hIAPP20-29 subjected to force applied in a variety of directions. By introducing systematic variations to this peptide sequence in silico, we have also investigated the role of the amino-acid sequence in determining the mechanical stability of amyloid fibrils. Our calculations show that the force required to induce mechanical failure depends on the direction of the applied stress and upon the degree of structural order present in the β-sheet assemblies, which in turn depends on the peptide sequence. The results have implications for the importance of sequence-dependent mechanical properties on seeding the growth of new fibrils and the role of breakage events in cytotoxicity.     

Mapping abeta amyloid fibril secondary structure using scanning proline mutagenesis

Although the amyloid fibrils formed from the Alzheimer's disease amyloid peptide Abeta are rich in cross-beta sheet, the peptide likely also exhibits turn and unstructured regions when it becomes incorporated into amyloid. We generated a series of single-proline replacement mutants of Abeta(1-40) and determined the thermodynamic stabilities of amyloid fibrils formed from these mutants to characterize the susceptibility of different residue positions of the Abeta sequence to proline substitution. The results suggest that the Abeta peptide, when engaged in the amyloid fibril, folds into a conformation containing three highly structured segments, consisting of contiguous sequence elements 15-21, 24-28, and 31-36, that are sensitive to proline replacement and likely to include the beta-sheet portions of the fibrils. Residues relatively insensitive to proline replacement fall into two groups: (a) residues 1-14 and 37-40 are likely to exist in relatively unstructured, flexible elements extruded from the beta-sheet-rich amyloid core; (b) residues 22, 23, 29 and 30 are likely to occupy turn positions between these three structured elements. Although destabilized, fibrils formed from Abeta(1-40) proline mutants are very similar in structure to wild-type fibrils, as indicated by hydrogen-deuterium exchange and other analysis. Interestingly, however, some proline mutations destabilize fibrils while at the same time increasing the number of amide protons protected from hydrogen exchange. This suggests that the stability of amyloid fibrils, rather than being driven exclusively by the formation of H-bonded beta-sheet, is achieved, as in globular proteins, through a balance of stabilizing and destabilizing forces. The proline scanning data are most compatible with a model for amyloid protofilament structure loosely resembling the parallel beta-helix folding motif, such that each Abeta(15-36) core region occupies a single layer of a prismatic, H-bonded stack of peptides.