My research project is focused on deciphering the structural basis of toxicity in neurodegenerative diseases including Parkinson’s and Lou Gehrig’s (ALS) disease. Even though large proteinacious deposits that are amyloid in nature are found in these diseases, scientific evidence in recent years has tilted the opinion in favor of small oligomers as the more toxic species. However, the transient nature of these small oligomers has made their study extremely difficult. The Eisenberg lab has focused on characterizing amyloid aggregates from a structural perspective. We have deciphered the atomic resolution structure of segments from these proteins in their amyloid-like conformation and used the structures to design inhibitors that prevent aggregation. We have now focused on characterizing these transient oligomers. In my research, I have worked on two amyloid-forming proteins namely Superoxide Dismutase 1 (SOD1) and Alpha Synuclein (α-syn) and have attempted to understand the basis of their toxicity. My results suggest two different modes of action for these proteins. For SOD1, moderately sized oligomers

seem to be more toxic than insoluble amyloid aggregates, but for α-syn the fibers are more toxic. Using the atomic structure of the spine of α-syn fibrils, I have developed a new line of inhibitors that prevent fibril growth and propagation. My results reveal a novel therapeutic approach targeted at the spread and progression that may be applicable for PD and related synucleinopathies.

In the last decade, there has been increasing research into the function and the mechanism of pathogenicity of membraneless organelles (MLOs). MLOs are essential for RNA metabolism including RNA transportation, transcription, translation, and degradation. Ribonucleoproteins (RNPs) participate in these processes and are key components of MLOs. For example, the RNPs hnRNPA2 and hnRNPF bind myelin basic protein mRNA, and this complex associates with a type of MLO called an RNA transport granule, which facilitates transportation to oligodendrocytes for translation. Similarly, the RNPs hnRNPU and LSM14 bind and localize RNA, and this complex associates with a type of MLO called a processing body, which facilitates RNA destabilization and decay. Another example is that the RNPs FUS and TDP-43 bind mRNA, and this complex associates with at type of MLO called a stress granule which forms and disappears in response to stress stimuli. Increasing studies are performed to span our knowledge of the relationship between MLOs and neurodegenerative diseases. RNPs often contain two types of functional domains: an RNA binding domain (RBD) for recognizing RNA and a low complexity domain (LCD) to facilitate MLO formation; most attention in neurodegeneration research has focused on the LCD. LCDs are composed of a limited repertoire of amino acid types such Glycine, Tyrosine, and polar residues. LCDs of many RNPs, including FUS and LSM4, form a separate, dense phase in vitro similar to cellular MLOs and can further self-assemble into stable amyloid-like fibrils with cross-β structures. Their propensity to phase separate implicates the LCDs of RNPs as being principle drivers of the formation and dynamics of MLOs. Mutations in the LCDs can accelerate their self-assembly rates, leading to hyper-stable MLOs which contribute to diseases. Over 50 mutations are found to be located in the LCDs and convert functional amyloid-like fibrils into pathogenic amyloid, but only a few atomic structures of the LCDs are known due to their lack of tertiary structures. Knowing the structures of the LCDs can help us better understand their functions and pathogenic mechanisms. So far, over 40 amyloid fibril structures are determined to atomic resolutions, but none before the work described here gives a clear structural explanation of its pathogenic mechanism. To fill this gap, we used cryoEM and determined the atomic structures of both wildtype and the variant hnRNPA2 LCD, an RNP component of RNA transport granules and stress granules. The wildtype structure of hnRNPA2 LCD shares intrinsic structural and energetic properties with other RNPs such as FUS: it has only one fibril morphology with a single protofilament and it is stabilized by hydrogen bonding and polar interactions. Wildtype hnRNPA2 has a poor solvation energy value, caused by the scarcity of residues that adopt the β-sheet conformation in the structure, which thereby weakens fibril stability. The variant structure turns out to be very different from wildtype in three ways: 1) It has at least 6 morphologies with two or more protofilaments. 2) It has a solvation energy value close to other pathogenic amyloids such as Tau. 3) More residues adopt β-sheet conformations. Compellingly, the variant structures of hnRNPA2 LCD offer evidence for its pathogenic mechanism. A loss of function arises because the variant’s PY nuclear localization signal (PY-NLS) is hidden in the fibril core, disrupting its ability to bind to transporters and be localized into the nucleus. Over half of the inward facing residues in the wildtype structure turn outward facing in the variant structure, exposing new surface epitopes that could potentially contribute to the variant’s toxicity. Thermostability and phase separation assays also confirm our structural results that the variant hnRNPA2 LCD fibrils are more stable than wildtype fibrils. We believe that the variant hnRNPA2 LCD fibril structure is unusual among other pathogenic amyloid fibrils because it suggests an obvious mechanism for pathogenicity and may shed light on analogous mutational conversions occurring in other neurological diseases.

Alzheimer’s disease (AD) and Parkinson’s disease (PD) are the most common neurodegenerative diseases. Although AD is a disease of dementia and PD is predominantly a motor disease, both diseases are characterized by histological hallmarks formed from aggregates of amyloid-forming proteins. Amyloid plaques and neurofibrillary tangles, composed of amyloid-β and tau respectively, are the hallmarks of AD while Lewy bodies, composed of α-synuclein (α-syn), are the hallmarks of PD. Amyloid-forming proteins such as amyloid-β, tau, and α-syn are soluble and functional in their monomeric state. They can misfold and aggregate into fibrils, which themselves aggregate to form amyloid plaques, neurofibrillary tangles, and Lewy bodies. There is a tight correlation between neurofibrillary tangle formation and progression of AD, and between Lewy body formation and progression of PD, so it has long been hypothesized that amyloid fibrils are toxic and contribute to the pathogenesis of AD and PD. Supporting this hypothesis are studies demonstrating that existing amyloid fibrils can propagate or “seed” the formation of additional fibrils among cultured cells and in mice. Using the wealth of atomic resolution structures of amyloid fibrils determined by x-ray crystallography and cryo-electron microscopy, the Eisenberg group has designed peptides, antibodies, and small molecules that target tau and α-syn fibrils. In this dissertation research, I focus on characterizing three of these structure-based designs as potential diagnostics and therapeutics for AD and PD. First, I characterize magnetic nanoparticles functionalized with an α-syn-targeting peptide and determine that they can be used as an MRI contrast agent to distinguish mice with α-syn pathology from wild-type control mice. Second, I characterize a bivalent nanobody that can inhibit seeding by post-mortem brain extracts from AD patients and determine it can cross the blood-brain barrier in mice. Third, I characterize three small molecules that can disaggregate AD brain-extracted fibrils and determine that they can each reduce levels of aggregated tau in mice with tau pathology. Together, the studies in this dissertation demonstrate the potential of using structure-based design of diagnostics and therapeutics for diseases caused by the misfolding of amyloid-forming proteins.

Amyloids are structured aggregates formed by misfolded proteins. Research has shown 25+ diseases associated with amyloid deposits in vivo. No cure has yet been developed for the majority of these conditions and the development of disease-modifying drugs has been disappointing. In the past year, several pharmaceutical companies have simultaneously withdrawn drug trials because of the ineffective clinical results. However, these treatments have been developed based on the molecular understanding of amyloid proteins from a few decades earlier. The amyloid field has made promising advancements in the past several years that offer potential for new therapeutic strategies. We now know that a subset of amyloidosis is caused by the deposition of the mature fibrils. Others are caused by the low molecular weight oligomers formed along the fibrillation pathway. Atomic views of these amyloid species were made available with the first steric zipper structure published in 2005 revealing the fibril spine and a cylindrin structure published in 2012, revealing the core of a toxic oligomer. These structures provide opportunities for new therapeutic strategies against amyloid diseases. In the course of my PhD work, I have investigated two amyloid protein systems. The first project focused on amyotrophic lateral sclerosis, a progressive motor neuron disease caused by oligomer formation of the amyloid protein superoxide dismutase I. I identified a region in the protein that is particularly important for the cytotoxicity, and solved its crystal structure. This work revealed a potential drug target for amyotrophic lateral sclerosis, a painful disease with no available treatment option. The second project focuses on an extracellular peptide whose fibril formation drastically enhances HIV infection. I developed inhibitors that prevented fibril formation and reduced HIV infection rate in a functional assay. Importantly, these inhibitors were designed based on the steric zipper structure, common to many amyloid fibrils. As such, this approach has the potential to be adapted to the other 25+ amyloid systems.

I began my research in the Eisenberg laboratory by studying the polymorphic nature of amyloid proteins Islet Amyloid Polypeptide (IAPP), a protein whose plaques are implicated in Type II Diabetes, and Amyloid-beta (Abeta), whose aggregates were implicated in Alzheimer's Disease. Both polypeptides are cleavage products of precursor proteins, are intrinsically disordered, and contain a highly amyloidogenic C-terminus. Later, I expanded the study of polymorphism to tumor suppressor protein p53, whose aggregation has recently been associated with tumor progression. The last part of my dissertation involves the studies on what may constitute the toxic species of amyloid, involving work with segments from alpha-B-crystallin, IAPP, and paralogs of 53.

This dissertation begins with work on structural and kinetic characterization of IAPP using transmission electron microscopy (TEM) and thioflavin T dye-binding assays. Here, I worked under Jed Wiltzius, who had solved several crystal structures of segments of the polypeptide each of which formed in-register steric zippers. I aided in his studies by performing EM on several of the segments and confirmed they indeed formed fibrils in vitro. I also performed EM and kinetic assays on full-length mutant and wild-type human IAPP, providing evidence that IAPP is capable of forming two distinct fibril polymorphs originating from two different steric zipper spines. These results that illustrate the molecular basis for fibril polymorphism of IAPP suggests a mechanism of protein-only encoded information transfer of different prion strains.

To further understand the polymorphic nature of amyloid proteins, I then focused on structural characterization of Abeta. To elucidate Abeta polymorphism in atomic detail, my colleagues Jacques-Philippe Colletier, Arthur Laganowsky, Meytal Landau and I determined eight new micro-crystal structures of fibril-forming segments of Abeta. These structures, all of various forms of steric zippers, reveal a variety of modes of self-association of Abeta. Combining these atomic structures with previous nuclear magnetic resonance and electron tomography studies, we propose several fiber models, offering molecular models that further illustrate the polydispersity of Abeta assemblies. These structures and molecular models contribute fundamental information for understanding Abeta polymorphic nature and pathogenesis. We furthermore suggest that steric zipper interactions are also the core of protafilaments binding together, explaining the immense heterogeneity in fibril morphologies as visualized under EM and various other characterization methods.

Structural characterization of fibril formation was carried to a third protein, tumor suppressor p53. It had recently been suggested that amyloid aggregation of mutant p53 may account for its gain of toxic function in cancer cells. Working with Alice Soragni, we elucidated the atomic details of the spine of p53 fibrils by identifying the aggregation-prone region and crystallizing two overlapping segments within the region. I also characterized a third segment that appears to exhibit a different type of steric zipper packing than other two segments. Results show that this short region within p53 displays the amyloid fibril polymorphism exhibited by Abeta and IAPP. In addition, these structures provide the basis for structure-based design of inhibitors of p53 aggregation as a potential cancer therapeutic.

A recent structure of a toxic amyloid oligomer, termed cylindrin, led me to also focus on a preliminary analysis of the mechanism of toxicity of this segment from alpha-B-crystallin. This was work done in collaboration with Arthur Laganowsky. I performed liposome disruption assays on the peptide, which suggests that the mechanism of toxicity of cylindrin may not be through membrane disruption. In addition, in collaboration with Professor Alex Van der Bliek, I attempted to transgenically express the peptide in C. elegans, as an in vivo model to examine toxicity. It appears cylindrin expression in C. elegans may induce slight toxicity, as it induces autophagosome accumulation and a slightly longer lifespan and larger brood size in the worms.

Finally, motivated by the extreme difficulty in crystallizing segments of amyloid proteins longer than eight residues, I helped in developing a methodology that has the potential to improve the chances of crystallizing proteins whose structure has remained elusive. In collaboration with Arthur Laganowsky, Minglei Zhao and Professor Todd Yeates, we developed a new crystallization approach, termed metal-mediated synthetic symmetrization, that introduces pairs of histidine or cysteine mutations onto the surface of target proteins, and, upon coordination with metal, generates novel crystal lattice contacts or oligomeric assemblies, thus producing a variety of new crystal forms, and increasing the chances of growing diffraction-quality crystals. We examined the method on two model fusion proteins, T4 lysozyme (T4L) and maltose-binding protein (MBP), and the approach resulted in 16 new crystal structures displaying a variety of oligomeric assemblies and packing modes, representing new and distinct crystal forms for these proteins. The results suggest this method has potential utlility for crystallizing target proteins of unknown structure through either direct mutations on the target protein or fusion of the target protein to metal-site mutants of T4L or MBP, which could serve as crystallization chaperones.

Current work involves exploring non-typical steric zipper interactions. I have recently solved 3 more crystal structures of various segments of IAPP, one of which forms an out-of-register steric zipper. I have also solved 2 more out-of-register zipper structures of segments within p63 and p73, both paralogs of p53 and suggested to co-aggregate with mutant p53. Analysis of these out-of-register structures show that there is no weak interface among the hydrogen bonding interactions, unlike other structures that have displayed out-of-register packing. Interestingly, cell viability assays showed that these peptides are not very toxic, suggesting the importance of these weak interfaces in amyloid toxicity. This work further confirms the polymorphic nature of amyloids.

The results embodied in this dissertation have assisted in advancing our understanding of molecular basis for amyloid fibril polymorphism and provides a preliminary characterization of the potential toxic amyloid oligomer cylindrin. In addition, the new crystallization methodology described in this work has the potential to improve the chances of crystallizing longer amyloid segments and additional proteins of unknown structure. Greater comprehension of the structural details of amyloid proteins not only can shed light into amyloid-aggregation mechanisms, but can also offer insight into the mechanisms of toxicity and aid in the development of therapeutics that target amyloid fibrillization and block aggregation.

Tau protein aggregates represent a type of amyloid that are the pathological hallmark of over 20 neurodegenerative diseases, termed tauopathies. These include Alzheimer's disease, progressive supranuclear palsy, and frontotemporal dementia and parkinsonism linked to chromosome 17 (FTDP-17). Given the association between the presence of tau aggregates and pathology, the field has a clear interest in determining whether said aggregates can be prevented, and if doing so would be a viable therapeutic strategy. Yet, much emerging evidence implicates soluble oligomers formed from the same protein to be the toxic species in many amyloid diseases. There is thus also a substantial effort devoted to identifying and characterizing these oligomers.

To see whether a rationally-designed peptidic inhibitor of tau fibril formation works as designed, I adapted and refined a kinetic fluorescence-based assay to test candidate inhibitors in two model systems. Using Nanogold particles, I also developed a new negative-stain electron microscopy-based method to ascertain whether an effective inhibitor actually blocks aggregation as designed, and employed mass-per-unit-length measurements from scanning transmission electron microscopy to support the findings. Using covalently-labeled fibrillogenic peptides, I also tried to create a FRET-based assay to screen for compounds that dissolve pre-formed fibrils. Shifting from working exclusively on insoluble assemblies, I later characterized segments from tau protein that we hypothesized would form toxic oligomers. I did this using a combination of cytotoxicity assays and biochemical methods.

This dissertation describes my findings. I determined that the inhibitors specifically block tau fibril formation, and that they indeed do so by capping and preventing elongation of nascent filaments. I also was able to show that a FRET-based disaggregation assay is viable, although though no effective compounds were identified in a pilot screen. And finally, I identified a segment of tau protein that exhibits characteristics suggesting that it is the region responsible for tau's forming toxic oligomers.

The formation of elongated, unbranched fibrillar protein aggregates, termed amyloid, has been linked to neurodegenerative diseases including Alzheimer’s Disease, Parkinson’s Disease and Amyotrophic Lateral Sclerosis (ALS). For decades, scientists have sought to determine the structures and polymorphic features of these aggregates to better understand their complex assembly and role in disease. Here, we show structures from three amyloid proteins, TAR DNA Binding Protein 43 (TDP-43), Superoxide Dismutase 1 (SOD1), and α-synuclein, offering insight into amyloid and protein aggregation in general.

First, we investigated TDP-43 as a model for the polymorphic capabilities of pathological amyloid aggregation. From x-ray diffraction, electron diffraction and cryoEM, we show that the segment, 247DLIIKGISVHI257, from the second RNA recognition motif forms an array of amyloid polymorphs. These associations include seven distinct interfaces displaying five different symmetry classes of steric zippers. The polymorphic nature of this segment illustrates at the molecular level how amyloid proteins can form diverse fibril structures.

In the second section, we investigated the role of the low complexity domain (LCD) of TDP-43 in reversible stress granule aggregation and irreversible pathogenic aggregation. We determined six atomic resolution segment structures from the LCD of TDP-43 that offer insight into regions of the protein responsible for aggregation. We illustrate how three of these structures, 300GNNQGSN306, 370GNNSYS375, and 396GFNGGFG402 form fibrils and exhibit characteristic steric zippers, indicative of pathogenic aggregation. We present an atomic structure for segment 312NFGAFS317, which reveals a fibril with a kinked interface that is reversible and may play a role in stress granule assembly. We present a model that combines previous literature with our new structural findings, to illustrate how stress granule assembly can lead to pathogenic aggregation.

Third, we structurally characterized peptide segments from SOD1 because pathological deposition of mutated SOD1 accounts for ~20% of the familial ALS (fALS) cases. We used a computational approach to discover four segments from the protein that form fibril-like aggregates and subsequently solved the structure of three of those segments: 101DSVISLS107, 147GVIGIAQ153 and 147GVTGIAQ153. We found, through the use of proline mutations, that two of these 101DSVISLS107 and 147GVIGIAQ153 are likely to trigger the aggregation of full-length SOD1, suggesting common molecular determinants of fALS and sALS.

In the fourth section, we investigated whether amyloid oligomers or fibrils were responsible for SOD1 toxicity in ALS. We determined the corkscrew-like structure of a cytotoxic segment of SOD1 in its oligomeric state. Through time course cytotoxicity assays, we demonstrated that the toxicity is likely a property of soluble oligomers and not large insoluble aggregates. Our work adds to evidence that the toxic oligomeric entities in protein aggregation diseases contain antiparallel, out-of-register β-sheet structures and identifies a target for structure-based therapeutics in ALS.

In the final section, we characterized the amyloidogenic core of α-synuclein, the main component of Lewy bodies, the neuron-associated aggregates seen in Parkinson’s disease. We utilized a new technique known as Micro-electron diffraction (MicroED) to solve the structure of crystals of an 11-residue segment we term NACore that are invisible by optical microscopy. The structure reveals protofibrils built of pairs of face-to-face β-sheets that exhibit a fiber diffraction pattern similar to full-length α-synuclein. We present a model of the toxic, full-length α-synuclein fibril, incorporating the NACore and an additional segment.

En masse, the structures in this dissertation offer broad insights into pathological and reversible protein aggregation. The structures provide a starting point for additional studies aimed at developing therapeutics for amyloid formation in neurodegenerative disease.

Human neurodegenerative diseases cause the gradual loss of neurons and progressive decline in memory, cognitive ability, movement and behavior. These devastating conditions afflict millions of patients worldwide and their families, however the molecular mechanisms of pathogenesis are not well understood, resulting in the lack of effective treatments and early diagnostic tools. Gross analyses of post-mortem brains using immunohistochemistry has identified abnormal deposits of amyloid proteins as defining features of neurodegenerative diseases. Amyloid proteins, like prions, can template aggregation of monomers into highly ordered, stable fibrils composed of tightly interdigitating β-sheets. Recent advances in cryogenic-electron microscopy (cryo-EM) have enabled the structural determination of amyloid fibrils to near-atomic resolution. These structures provide insight into the formation and propagation of amyloid fibrils, and can guide the design of therapies that prevent, delay, or reverse the aggregation of proteins in neurodegenerative diseases. Moreover, previous work has shown that the structures of brain-extracted fibrils differ from the structures of fibrils assembled in vitro, underscoring the importance of studying patient-derived samples. In my thesis research, I extracted amyloid fibrils from autopsied brains of patients with frontotemporal lobar degeneration with TDP-43 inclusions (FTLD-TDP) and determined their structures using cryo-EM. Surprisingly, the fibrils examined were not formed by TDP-43 but by a fragment of transmembrane protein 106B (TMEM106B), revealing a previously unsuspected amyloid protein. In addition, I used a biosensor cell line, which detects proteopathic TDP-43 aggregates from sarkosyl-insoluble brains extracts, to study cohorts of TDP-43 proteinopathy cases including amyotrophic lateral sclerosis (ALS), FTLD-TDP and Alzheimer’s disease (AD). My work expands the insight into the molecular basis of amyloid disorders, and inspires future research to address newly raised questions about TMEM106B and TDP-43 in neurodegenerative diseases.

Many diseases are characterized by the pathologic accumulation of aggregated proteins. Known as amyloid, these fibrillar aggregates are present in many neurodegenerative diseases, including Alzheimer’s and Parkinson’s disease (PD), and amyotrophic lateral sclerosis (ALS). The development and spread of amyloid fibrils within the brain correlates with disease onset and progression, and inhibiting their formation is a possible route towards therapeutic development. Advances in structural biology, namely micro-crystal x-ray diffraction, micro-electron diffraction (MicroED), cryo-electron microscopy (cryoEM) and solid-state NMR spectroscopy (ssNMR) have enabled the determination of amyloid fibril structures to atomic-level resolutions, improving the possibility of structure-based inhibitor design. In Chapter 1, we use these amyloid structures to design inhibitors which bind to the ends of fibrils, “capping” them so as to prevent further growth. Applying recent breakthroughs in de novo protein design, we describe a computational approach to develop mini-protein inhibitors of 35-48 residues which target the amyloid structures of tau, Aβ (found in Alzheimer’s disease) and αSyn (found in Parkinson’s disease). Biophysical characterization of the in silico designed inhibitors shows they form stable folds, with no sequence homology to naturally occurring proteins, and specifically prevent the aggregation of their targeted amyloid-prone proteins in vitro. The inhibitors also prevent the seeded aggregation and toxicity of fibrils in cells. In vivo evaluation reveals their ability to reduce aggregation and rescue motor deficits in C. elegans models of PD and AD.

In Chapter 2, we apply a similar design strategy to generate inhibitors of a different form of protein aggregation. Proteins with low complexity segments engage in liquid-liquid phase separation (LLPS) during normal cell processes, but aberrant LLPS leads to the eventual aggregation of such proteins into amyloid fibrils. This presents inhibition of LLPS as another potential therapeutic target. As LLPS appears to be a precursor state to fibrillization, we test if inhibitors targeting an amyloid fibril structure can also reduce LLPS of the same protein. To accomplish this, we design de novo miniprotein inhibitors which target the fibril structure of the FUS low-complexity domain, a protein known to undergo LLPS and eventually aggregate in ALS, frontotemporal dementia (FTD) and other dementias. Several designs are able to reduce FUS LLPS in vitro, as well as FUS fibrilization. Additionally, the same inhibitors reduce stress granule formation in cells, a form of LLPS. Mutations to improve binding to the FUS fibril interface improve stress granule reduction, while steric clashes introduced into the interface abolish the inhibitor effects. The top inhibitor construct iFUS-G specifically inhibits FUS LLPS and has no effect on phase separation of low complexity segments from TDP43 or hnRNPA2, aggregation prone proteins similar in sequence composition to FUS. These findings present a rational design strategy to specifically inhibit the phase separation of low complexity proteins and have implications for the structural underpinnings of protein LLPS.

Next, we transition from therapeutically targeting known amyloid proteins to identifying new ones. In addition to FUS, many proteins including hnRNPA1, hnRNPA2, and TDP-43 have been established to undergo aggregation into amyloid-like fibrils through interactions of their low-complexity domains. Mutations in the low-complexity domains of the same proteins can lead to irreversible amyloid aggregation and disease. In Chapter 3 we introduce a computational procedure to identify mutations in low-complexity domains of disease-related proteins that are predicted to increase their propensity for amyloid aggregation. This procedure found several disease-related mutations in a low complexity region of the intermediate filament protein Keratin-8 (KRT8). Atomic structures of wild-type and mutant KRT8 segments confirm the transition of a highly extended strand to a pleated strand capable of amyloid formation. Biochemical analysis of KRT8 reveals the protein forms amyloid aggregates and that the identified mutations promote aggregation. Aggregated KRT8 is found in Mallory-Denk bodies, often observed in the hepatocytes of livers with alcoholic steatohepatitis (ASH). We demonstrate that ethanol promotes KRT8 aggregation, and KRT8 amyloid structures co-crystallize with alcohol. We also observe that KRT8 aggregation can be seeded with ASH patient liver extract, consistent with the amyloid nature of KRT8 aggregates.

Lastly, in Chapter 4 we explore structural characteristics that distinguish amyloid proteins known to undergo reversible versus irreversible aggregation. While all amyloid fibrils are primarily composed of repeating layers of beta-sheets, we observe that fibril structures of proteins known to reversibly aggregate have an enrichment of highly extended non-ideal beta-sheets. Quantum calculations of pleated- and extended-beta sheet amyloid structures show that extended backbones decrease the energy required to separate strand pairs. Non-covalent interaction analysis shows that the extended beta-sheets may be stabilized by interactions between the amide proton and carbonyl oxygen of the same residue, known as C5 hydrogen-bonding. These findings identify a key structural element that may regulate reversible amyloid assembly.

This body of work offers insight into new ways to therapeutically target protein aggregation, identifies novel amyloid pathologies, and explores the structural underpinnings that may distinguish different forms of amyloid aggregation.