Increased Dynamics of α-Synuclein Fibrils by β-Synuclein Leads to Reduced Seeding and Cytotoxicity

Alpha-synuclein (αS) fibrils are toxic to cells and contribute to the pathogenesis and progression of Parkinson’s disease and other synucleinopathies. β-Synuclein (βS), which co-localizes with αS, has been shown to provide a neuroprotective effect, but the molecular mechanism by which this occurs remains elusive. Here we show that αS fibrils formed in the presence of βS are less cytotoxic, exhibit reduced cell seeding capacity and are more resistant to fibril shedding compared to αS fibrils alone. Using solid-state NMR, we found that the overall structure of the core of αS fibrils when co-incubated with βS is minimally perturbed, however, the dynamics of Lys and Thr residues, located primarily in the imperfect KTKEGV repeats of the αS N-terminus, are increased. Our results suggest that amyloid fibril dynamics may play a key role in modulating toxicity and seeding. Thus, enhancing the dynamics of amyloid fibrils may be a strategy for future therapeutic targeting of neurodegenerative diseases.


Results
Co-incubation with βS induces subtle differences in αS fibril morphology. We studied the differences in the morphology of αS fibrils formed from the incubation of monomeric N-terminally acetylated αS, and αS fibrils formed by co-incubation of monomeric N-terminally acetylated αS with monomeric N-terminally acetylated βS at a 1:3 ratio (called αS/βS co-incubated fibrils). We have used the N-terminally acetylated forms of αS and βS, since this post-translational modification is constitutively present in the native forms of these intrinsically disordered proteins 37,38 . Consistent with our previous work 33, 35 , βS delays αS fibril formation in the Thioflavin T (ThT) aggregation assay (Fig. S1), whereby the co-incubation of αS with βS results in a longer lag time and slower growth kinetics compared with αS by itself. Fibrils formed as the end products of these two monomer aggregation assays display differences in their polymorph composition. Atomic force microscopy (AFM) images show that αS by itself (Fig. 1a) forms long straight or twisted fibril polymorphs, similar to previous reports 13,39,40 , while co-incubated αS/βS forms fibril polymorphs that are shorter and straight (Fig. 1b), with no discernable twisting pattern. On average, the height of αS fibrils (6.0 ± 1.1 nm) tends to be shorter than αS/βS 15 αS/βS co-incubated fibril core structure is maintained while the N-terminal dynamics are increased. We investigated the conformational and dynamics properties of the αS fibril when co-incubated with βS utilizing solid-state NMR (ssNMR) spectroscopy. We first assessed the secondary structure of our fibrils from 2D 13 C-13 C and 15 N-13 C correlation spectra. Figure 2a shows the 2D 13 C-13 C 100 ms DARR spectra and Fig. S5 the 2D 15 N-13 C correlation spectra of αS fibrils in black overlaid with co-incubated αS/βS fibrils in red. This experiment utilizes a cross-polarization period to transfer 1 H to 13 C magnetization, which preferentially detects the rigid residues that make up the core of the fibril and does not detect the dynamic or disordered residues that make up the bulk of the N-and C-terminal regions. The 13 C chemical shifts, 13 C-13 C cross peaks, and 15 N- 13 C cross peaks of the spectra do not show marked differences between αS and αS/βS fibrils, indicating that the core structure of the fibril does not change even when formed in the presence of a stoichiometric excess of βS (Fig. 2a). To ascertain whether our fibrils maintain the common Greek-key motif core structure previously identified 13,39-41 , without conducting a full structure determination by ssNMR, we compared our spectra with the 13 C chemicals shifts of αS fibrils deposited into the Biological Magnetic Resonance Data Bank 42 (BMRB). Out of 15 total entries in the BMRB of αS fibril chemical shifts (8 human WT, 6 human mutants, 1 mouse WT) only one is associated with a high-resolution 3D structural model of human WT αS fibrils (BMRB 25518, PDB 2N0A) 41 .
Using the 13 C chemical shifts from BMRB 25518 we created a 13 C-13 C chemical shift correlation map, shown in green in Fig. 2c, and overlaid it with our αS fibril 13 C-13 C spectrum; we observe relatively good agreement between the published chemical shifts and our αS spectrum. In addition, we created a 13 C-13 C chemical shift correlation map using an average of the 13 (Fig. 2d).  Figure 2. Characterization of the Fibril Core. (a) Overlay of αS fibril (black) and co-incubated αS/βS fibril (red) 2D 13 C-13 C correlation spectra (100 ms DARR), showing that the conformation of the core does not differ significantly between these two fibrils. (b) Expansion of select regions of the 2D 13 C-13 C spectra in (a) that show the major cross peak intensity differences, which are the Lys, Thr and a tentatively assigned Asn residues. (c,d) Overlays of the αS 2D 13 C-13 C correlation spectrum in (a) with chemical shift correlation maps derived from (c) the published solid-state NMR structure of αS fibrils (green, PDB: 2N0A, BMRB: 25518 41 ) and (d) an average of the 13  When overlaid with our αS fibril spectrum, we again find relatively good agreement between the two spectra. This comparison between the previously published 13 C chemical shifts, in particular the 13 C chemical shifts from PDB 2N0A, and our own 13 C spectra suggests that the secondary structure of our αS fibril core is consistent with a Greek-key topology. A full structure determination and peak assignment is currently underway to confirm this assessment.
While the core of our αS and co-incubated αS/βS fibrils are unchanged, there are some subtle differences in peak intensities between αS and αS/βS fibrils. The Cβ-Cα, Cβ-Cγ and Cβ-CO cross-peaks of the threonine residues show marked intensity decreases in the co-incubated fibril (Fig. 2b). We also observed intensity decreases in the cross peaks of lysine Cε-γ and Cβ-γ of N65 of the co-incubated fibril (Fig. 2b); the N65 Cβ-γ assignment is tentative, and could also plausibly be from I88 Cβ-CO or F94 Cβ-CO. The loss of intensity of these peaks could be caused by increased dynamics of these residues in the co-incubated fibril. To further investigate this point we measured the 13 C T 1ρ relaxation time 49 , which reports on μs timescale dynamics, of the threonine Cβ, since these peaks are well resolved from chemical shift overlap of any other residue in the region from ~65-70 ppm ( Fig. 2a and Fig. S4a,b). We found that the 13 C T 1ρ relaxation time of the Thr Cβ's from co-incubated αS/βS fibrils decreased relative to the Thr Cβ's from αS fibrils (Fig. S4c), indicating an increase in dynamics of these residues.
To further characterize any differences between αS and co-incubated αS/βS fibrils, we probed the changes in water accessibility and hydration between the two fibrils. Figure 3 shows the water-edited 2D 13 C-13 C correlation spectra 50,51 of 13 C, 15 N-labeled αS fibrils (Fig. 3a) and co-incubated αS/βS fibrils where only αS is uniformly labeled with 13 C and 15 N (Fig. 3b). The basic premise of this experiment is to observe how the transfer of water 1 H magnetization varies across the fibril. The long water 1 H spin-diffusion (SD) mixing time (100 ms, black) spectra represent a state where the water 1 H magnetization has fully equilibrated across each fibril, while the short water 1 H SD mixing time (3 ms, red) spectra illustrate the fibril residues that are in closest proximity to water. The relative proximity or accessibility of a residue to water is then most easily compared by taking the ratio between these two intensities (Int 3ms /Int 100ms ). Due to sample sensitivity and time constraints, in lieu of obtaining residue specific full SD build up curves, the reason for using the ratio between the short and long mixing times is to provide normalization of the intensities measured in the short mixing time experiment and allow for a relative comparison of the initial spin-diffusion buildup rates between different samples. For example, residues that are far from water or are located in the center of the fibril core will have smaller water-accessibility ratios, while residues that are on the surface of the fibril will have larger water-accessibility ratios. This approach has been used previously to probe the differential hydration environments of amyloid fibrils and other biomolecules [51][52][53][54][55] .
Slices from the water edited 2D 13 C-13 C spectra show decreases in the relative intensities of several of the threonine and lysine cross-peaks of the αS/βS fibrils compared to αS fibrils, while those lysine cross-peaks that remain in the αS/βS fibril spectra have increased water spin-diffusion (i.e. larger Int 3ms /Int 100ms ratios) relative to αS fibrils (Fig. 3c). The increase in water spin-diffusion of the αS/βS fibril lysine peaks indicates that these residues are more water accessible. Conversely, the water spin diffusion ratios of the hydrophobic alanine and valine residues do not change between αS and co-incubated αS/βS fibrils (Fig. 3c), indicating that the hydration environment of these residues does not significantly change between the two fibrils.
Co-incubated fibrils are more sensitive to proteinase K digestion. Proteasomal impairment has been implicated in several neurodegenerative diseases 56 , including PD 57 , and as proteasome activity decreases with age cells become more vulnerable to deleterious protein aggregation 58 . Therefore, an understanding of how synuclein fibrils and aggregates undergo protease degradation and clearance may shed critical light on PD progression. In order to understand the differences in protease degradation and fibril stability between αS fibrils and co-incubated αS/βS fibrils, we carried out a series of digestion assays with increasing concentrations of proteinase K (Fig. 3c). We observed that co-incubated αS/βS fibrils are more sensitive to proteinase K digestion at a concentration of 5 μg/ml. The co-incubated fibrils display an enhanced propensity to be degraded to low molecular weight species (i.e. intense band at ~5 kDa only) compared to αS fibrils, which have a larger proportion of high molecular species (i.e. intense bands at ~10 and ~15 kDa).

co-incubated αS/βS fibrils are less toxic and exhibit reduced seeding and proliferation capacity compared to αS fibrils in neuroblastoma cells.
Having now characterized the conformational and polymorphic differences between αS and co-incubated αS/βS fibrils, we investigated how these changes in protofilament packing and dynamics affects the cytotoxicity of the fibrils. It is common practice to sonicate amyloid fibrils directly before use in cell toxicity assays, in order to obtain a more uniform distribution of fibril lengths and size. However the sonication process also produces smaller aggregates and oligomers 59 , with an unknown distribution of shapes and sizes, that show different toxicities, fibril seeding abilities, and cell incorporation capacities 19 . In order to avoid forming these additional oligomers, we have elected to use un-sonicated fibrils in our assays of cell toxicity and seeding, with polymorph compositions (Fig. 1a,b) and size distributions (Fig. 1c,d) illustrated in our AFM images. As a consequence of using long un-sonicated fibrils, the concentration of fibrils used in our assays (~1 μM) is higher than the nM concentrations used previously for smaller sonicated fibrils, since these smaller sonicated species are more easily incorporated into cells 60,61 . However, our fibril preparations are internalized similarly to sonicated fibrils, showing the characteristic fluorescent punctate-like structures (Fig. S7).
We also investigated the ability of αS and αS/βS fibrils to seed aggregation in vitro using a quiescent thioflavin-T (ThT) fluorescence assay. Figure S3 shows that while preformed seeds of both αS and αS/βS fibrils www.nature.com/scientificreports www.nature.com/scientificreports/ have the ability to induce fibril formation in the presence of αS monomers, the time for the ThT fluorescence curve to plateau takes longer with αS/βS fibril seeds. This indicates that αS/βS fibrils have a reduced capacity to seed further αS aggregation. We confirmed these observations of αS and αS/βS fibril seeding capacity in cell, by assessing the ability of these fibrils to seed aggregation of endogenous αS in SH-SY5Y cells through the analysis of the fluorescence intensities of dyes that specifically bind to αS and amyloid structures (Fig. 4b). Cells were treated with monomeric αS, αS fibrils or αS/βS fibrils for 24 hours before being fixed and stained with purified mouse anti-αS (anti-α-synuclein) antibody, thioflavin S (ThioS), and 4' ,6-diamidino-2-phenylindole (DAPI). Cells were then imaged by confocal fluorescence microscopy, where the anti-αS antibody fluoresces red and indicates the presence of any synuclein species present, ThioS fluoresces green and indicates the formation of amyloid species, and DAPI stains the cell nucleus blue (Fig. 4b). Compared with cells treated with monomeric αS (Fig. 4b, bottom row), cells treated with αS fibrils showed an increase in anti-αS antibody fluorescence of 7.3× (Fig. 4b, top row), while cells treated with αS/βS fibrils showed a smaller increase of 4.4× (Fig. 4b, middle row). ThioS staining indicating amyloid formation showed a similar trend with a 4.8x increase with αS fibrils vs a 3.4x increase with αS/βS fibrils.  www.nature.com/scientificreports www.nature.com/scientificreports/ mature αS fibrils over time 18 . To understand the effect of βS on the stability and equilibrium of αS fibrils, we sought to determine the morphology, toxicity and cell seeding capacities of the oligomers that are shed from αS fibrils and αS/βS fibrils.
We first measured the thermostability of the two fibrils using far-UV circular dichroism (CD) spectroscopy. The CD spectra show that both αS and αS/βS fibrils have the characteristic spectral minimum at 218 nm, indicating the presence of β-sheet structure (Fig. S2). We monitored the change in ellipticity of the 218 nm signal as a function of temperature, and found that change in ellipticity of co-incubated αS/βS fibrils is less than that of αS fibrils as temperature increased, indicating that αS/βS fibrils are more thermostable than αS fibrils (Fig. S2). AFM images show that the oligomers that are shed from αS fibrils (Fig. 5a) primarily adopt small globular www.nature.com/scientificreports www.nature.com/scientificreports/ morphologies, while oligomers shed from αS/βS fibrils tend to adopt short proto-fibril morphologies with some larger globular species also present (Fig. 5b). We next measured the toxicity of the shed oligomers in SH-SY5Y cells. After a 48 hour period of incubation with shed oligomers from either αS or αS/βS fibrils, we found that oligomers shed from αS reduced cell viability by 17% compared to the untreated cells and cells treated with monomeric αS, whereas oligomers shed from αS/βS did not (Fig. 5c). We also assessed the ability of shed oligomers to seed further aggregation in cells, using confocal fluorescence microscopy. Compared with cells treated with monomeric αS (Fig. 5d, bottom row), cells treated with oligomers shed from αS fibrils showed an increase in anti-synuclein antibody fluorescence of 1.6× (Fig. 5d, top row), while cells treated with oligomers shed from αS/βS fibrils showed an increase of 1.3× (Fig. 5d, middle row). ThioS staining indicates that amyloid formation increased by 1.6× in cells treated with oligomers shed from αS fibrils and by 1.3× in cells treated with oligomers shed from αS/βS fibrils.

Discussion
Amyloid fibrils of αS are key pathologic features of PD and have been recognized as contributing to the progression of the disease. These fibrils are thought to contribute to cellular toxicity through their ability to seed further aggregation of endogenous αS, and the ability of the fibrils to "shed" oligomer and protofibril species that may be toxic. The αS fibrils studied in this work should be distinguished from fibrils that are contained within aggresome-like LBs. Fibrils that are formed as the end product of the aggregation pathway of αS (either in vivo or in vitro), and are not yet collected into LBs, exist in a dynamic equilibrium with oligomers, as evidenced by the ability of fibrils to "shed" smaller molecular species 18,63 . Here we have demonstrated that αS fibrils formed in the presence of the natural inhibitor βS, while maintaining similar core structures as αS fibrils alone, exhibit reduced toxicity to neuroblastoma cells, reduced seeding properties, and are in dynamic equilibrium with oligomers that also share reduced toxicity and seeding (Fig. 6).
We have utilized ssNMR and the changes in 13 C chemical shifts to probe how the core residues and dynamics of αS fibrils and co-incubated αS/βS fibrils differ from one another. While we have not yet completed a full assignment and structure determination of our fibrils, 13 C chemical shifts are very sensitive reporters of amino acid type and secondary structure 64 . The cross-polarization based ssNMR experiments used in this work preferentially detect molecules and domains that are rigid, and dynamic and disordered domains are not detected. The dynamic regions of αS fibrils are the very N-terminal (residues ~1-43) and C-terminal (residues ~97-140) domains; what we detect in our spectra are the relatively rigid core of the amyloid fibrils (residues ~44-96). The 13 C resonances observed in our spectra show characteristic β-sheet chemical shifts. Comparison of our spectra with the chemical shift lists (BMRB Entry 25518) and spectra reported by Rienstra and coworkers 41 , who have previously determined the core fibril structure of full-length αS by ssNMR, show relatively good agreement (Fig. 2c)

F i b r i l F o r m a t i o n F i b r i l F o r m a t i o n F ib r il F o r m a t io n F ib r il F o r m a t io n
Non-Toxic Toxic Figure 6. αS and αS/βS fibril toxicities and seeding potentials. αS aggregates and misfolds along a nucleationdependent fibril formation pathway, generating various oligomeric species before finally adopting a characteristic repeating cross-beta amyloid fibril structure. When αS aggregates on its own (left pathway), the resulting fibrils are toxic to cultured human neuroblastoma cells, and the oligomeric species that shed from these fibrils are also toxic to cells. However, if αS is co-incubated with βS and allowed to aggregate (right pathway), then the resulting fibrils are no longer toxic to cells, and oligomer species that shed from these fibrils are also non-toxic. The αS and αS/βS fibrils also display differential seeding capacities (confocal images, bottom). αS fibrils are able to efficiently seed amyloid formation, while co-incubated αS/βS fibrils have reduced propensity to seed further aggregation, as evidenced by the difference in green intensity in confocal images. www.nature.com/scientificreports www.nature.com/scientificreports/ have prepared our fibrils in a similar manner to those used for the published αS fibril structure 41 , we can reasonably assume that our fibrils adopt a similar core structure. αS has a 140 amino acid primary sequence generally described by 3 domains: a 60 residue polyampholyte N-terminal domain, a 35 residue hydrophobic NAC domain, and a 45 residue highly negatively-charged polyelectrolyte C-terminal domain. Lysine and threonine (Fig. 3e) are almost exclusively located in the N-terminal and NAC regions of the αS sequence, where they are clustered into imperfect KTKEGV repeats. These two regions form the "Greek-key" motif of the αS fibril core structures, roughly spanning residues 44-96. Of particular note is the preNAC domain (44)(45)(46)(47)(48)(49)(50)(51)(52)(53)(54)(55)(56)(57)(58)(59)(60), which makes up the packing interface between two αS protofilaments 13,39,40 , and contains two full KTK repeats.
While the overall fold of the core residues does not change between αS and αS/βS fibrils, and is similar to previously determined αS structures, based on the assignments of the Thr and Lys cross-peaks in our 2D 13 C-13 C spectra of the co-incubated αS/βS fibril we do find changes to these residues that make up the preNAC domain's final two KTKEGV repeats. First, from the loss of intensity in our 2D spectra and measurements of T 1ρ relaxation, we have found an increase in the local dynamics of the Thr and Lys residues of co-incubated αS/βS fibrils relative to αS fibrils alone. Second, we observed that the water accessibility of the Lys residues increases in the co-incubated αS/βS fibrils relative to αS fibrils. Our ssNMR measurements of the fibril core structure (residues 44-96), and residue dynamics and water-accessibility highlight that βS does not perturb the core structure of αS fibrils, but instead may have an impact on the residues at the protofilament packing interface.
Recent structures of αS fibrils determined by cryo-electron microscopy have begun to show the importance of protofilament packing to the observed differences in fibril polymorphism, rather than a change in the conformation of the protofilament core 13,39,40 . Our AFM images indicate that the primary distinctions between αS and αS/βS fibrils lie in the average height and length, although these parameters are widely distributed (Fig. 1c,d). The more subtle distinction lies with the change in polymorph composition of the fibrils: αS fibrils show a mixture of twisted and straight polymorphs (Fig. 1a) while αS/βS fibrils appear to only have straight polymorphs (Fig. 1b). These observations suggest that βS modulates the packing of the protofilaments in the mature fibril, and is less likely to be incorporated into the cross-beta structure of the individual protofilament. A change in protofilament packing is also supported by the proteinase K digestion profiles, which indicate that αS/βS fibrils are more easily accessible to cleavage by proteinase K, suggesting that the co-incubated αS/βS fibrils might be more susceptible to degradation in vivo (Fig. 3d). These profiles resemble those from previous work by Miake and coworkers, who established that proteinase K digestion of αS preferentially cleaves the N-and C-terminal portions of αS and leaves the fibril core from residues 31-109 intact 65 . In addition, our results demonstrate that while αS fibrils are indeed toxic to neuroblastoma cells (Fig. 4a), co-incubated αS/βS fibrils are not (Fig. 4a) and have a lesser tendency to cause the formation of synuclein aggregates and amyloid species compared with αS fibrils (Fig. 4b). Therefore, co-incubation of βS with αS results in fibrils that are not toxic to cells and have reduced ability to seed further aggregation in a cellular environment. The reduced seeding ability also suggests that βS interferes with the ability of αS fibrils to catalyze secondary nucleation processes on the fibril surface, proposed previously by Knowles and coworkers 66 . Taken together with our earlier results, these results again suggest that βS is associated with the N-terminal domain and protofilament packing interface along αS fibrils, inducing more dynamic flexibility in the N-terminal portions of the αS fibrils, and may suggest that N-terminal domain dynamics and packing may play a role in this seeding process.
We have also found that αS fibrils are less thermostable (Fig. S2) and shed primarily small globular and amorphous oligomers (Fig. 5a), while co-incubated αS/βS fibrils are more thermostable (Fig. S1) and shed primarily short proto-fibril aggregates (Fig. 5b). The proto-fibril species shed from αS/βS fibrils also show reduced seeding propensity compared to the small globular and amorphous αS oligomer species (Fig. 5d). This finding suggests that the dynamic equilibrium is shifted away from the formation of small toxic oligomers towards less toxic proto-fibrils in the presence of βS (Fig. 5c). Taken together in the context of our previous observations, the protofilament-packing of co-incubated αS/βS fibrils appears to be more stable than αS fibrils, while increasing the local dynamics of the N-terminal domain. This results in a reduced capacity for seeding and shedding of toxic oligomeric species.
βS has previously been identified in studies of transgenic mice as a natural anti-Parkinsonian factor which has the ability to reduce αS inclusion formation 30 . Yet, even though it reduces αS positive inclusions, it does not completely abolish the formation of αS fibrils. We propose that the role of βS as an inhibitor is multifaceted, influencing αS aggregation at multiple points along its fibril-formation pathway. In the earliest stages of αS aggregation, βS can stabilize αS in αS-βS heterodimers 35 , which help to slow down the conversion of αS into higher order aggregates. As αS continues to aggregate, βS has been found to stabilize and eliminate the formation of toxic oligomers 67,68 . In this work, we have now shown that in the last stage of αS aggregation, co-incubation with βS minimizes the toxicity and seeding ability of αS fibrils, and furthermore alters the fibril-oligomer equilibrium. Our findings demonstrate that βS can reduce the effects of toxic αS fibrils in cells without changing the core structure of αS fibrils, and provide insight into how the dynamics and the surface of these fibrils may directly contribute to their toxicity and seeding ability. The multi-pronged targeting of αS by βS highlights the potential of βS as a lead for the future design of inhibitors that provide therapeutic intervention in synucleinopathies at multiple stages of αS aggregation.
The misfolding and aggregation of endogenous αS monomers due to seeding by fibrils is believed to be critical to the progression of synucleinopathies. The mechanism by which mature αS fibrils seed further aggregation is believed to proceed by surface-mediated secondary nucleation [69][70][71] , where the surface properties of αS fibrils govern their interaction with endogenous αS monomers and template further aggregation. The exact details of how additional αS monomers undergo templated conversion are not yet known, but the present work provides some clues. The recent cryo-EM structures of αS fibrils show that a steric-zipper motif in the N-terminal domain mediates the interface between two protofilaments and stabilizes the mature fibril morphology 13,40 . Our results show that co-incubated αS/βS fibrils have increased dynamics and water accessibility of residues in the N-terminal www.nature.com/scientificreports www.nature.com/scientificreports/ domain, particularly in the KTKEVG repeats in the preNAC region, and allow for enhanced protease degradation of the fibril, suggesting that the protofilament interface may be altered and more dynamic. These observations also highlight the importance of dynamics in mediating the seeding ability of αS fibrils: increased dynamics of the N-terminal domain may lead to reduced seeding, as secondary nucleation may necessitate a rigid N-terminal domain for proper templating of αS aggregation. Our results suggest that enhancing amyloid fibril dynamics at templating domains may be an approach for future therapeutic intervention for neurodegenerative diseases.

Materials and Methods
Protein expression and purification. Expression of N-terminally acetylated human αS and βS proteins was performed via co-expression with pNatB plasmid (Addgene #53613) in E. coli BL21(DE3) cells, and protein purification was performed as described previously 72 . Uniformly 13 C, 15 N isotopically labeled αS for ssNMR experiments was expressed in M9 minimal media supplemented with 13 C-glucose and 15 N-ammonium chloride as the sole carbon and nitrogen sources, respectively. Protein molecular weight and purity were assessed by ESI-MS, and stored at −20 °C as a lyophilized powder until use. Atomic Force Microscopy (AFM). Samples (20 μL) were placed onto freshly cleaved mica (Ted Pella Inc., Redding, CA) and incubated for 15 min at room temperature, followed by 3 washes of 200 μL each deionized water as described previously 34 . All images were collected on a NX-10 instrument (Park Systems, Suwon, South Korea) using non-contact mode tips (PPP-NHCR, 42 N/m, 330 kHz; Nanosensors, Neuchatel, Switzerland). Image processing and analysis were carried out in the Gwyddion software package 73 .
Solid-State Nuclear Magnetic Resonance Experiments. All MAS ssNMR experiments were carried out on an Avance III HD 600 MHz (14 T) spectrometer (Bruker BioSpin, Billerica, MA) using a 1.6 mm triple resonance MAS probe (Phoenix NMR, Loveland, CO) tuned to 1 H/ 13 C/ 15 N frequencies. Typical radiofrequency (rf) field strengths were 118 kHz for 13 C, 74 kHz for 15 N, and 100-145 kHz for 1 H. 13 C chemical shifts were referenced to the 13 CH 2 signal of adamantane at 38.48 ppm on the tetramethylsilane (TMS) scale, and 15 N chemical shifts were referenced to the 15 N signal of N-acetylvaline at 122.0 ppm on the liquid ammonia scale. All experiments utilized a MAS rate of 13.333 kHz, and sample temperature was controlled to 25 °C, unless otherwise noted. One-dimensional (1D) 13 C MAS spectra were recorded using a conventional cross-polarization (CP) sequence. Two-dimensional (2D) 13 C-13 C dipolar-assisted rotational-resonance (DARR) experiments 74 utilized a mixing period of 100 ms. A 2D water-edited DARR 50 experiment, with a DARR mixing period of 100 ms, a T 2 -filter of 6 ms, and a 1 H spin-diffusion period of either 3 ms or 100 ms, was used to measure the water-protein 1 H spin diffusion differences between the two fibril samples. 2D 15 N-13 C correlation spectra were measured using a REDOR-based pulse sequence 75 , utilizing a REDOR period of 1.35 ms to observe long range correlations. A standard Bruker CP based pulse sequence was used to measure the 13 C T 1ρ relaxation of the Thr Cβ 49 . A 13 C ω 1 field of 2.25ω r was applied on resonance with the Thr Cβ (70 ppm) during the spin-lock period.

Analysis of fibril composition by ESI-MS.
Mature fibril samples were dissolved in 4 M guanidine hydrochloride overnight, then buffer exchanged with 50 mM ammonium acetate with 0.1% formic acid. Samples were concentrated to 10 µM for ESI-MS analysis.
Neuroblastoma cell culture. Human SH-SY5Y neuroblastoma cells (ATCC, Manassas, VA) were cultured in DMEM/F12 (GE Healthcare, Boston, MA) with 10% fetal bovine serum (Gibco Co., Dublin, Ireland) and kept in a 37 °C, 5% CO 2 humidified atmosphere. Before cell viability assays or immunocytochemistry, cells were plated Preparation of ATTO-550 labelled fibrils. Fibrils were labeled with the fluorescent ATTO550-NHS-ester (ATTO-TEC GmbH) per the manufacturer's procedure. In brief, fibrils formed as described above were incubated with a 2 M excess of ATTO550-NHS-ester in labeling buffer (pH 8.3 PBS/sodium bicarbonate solution) for 1 hour at room temperature. Conjugated ATTO550-fibrils were then separated from unreacted fluorophore by centrifugation at 16k rpm for 30 min and resuspension of the ATTO550-fibril pellet in pH 7.4 PBS; this centrifugation/ resuspension wash was repeated twice.
Statistical analysis. All cell viability experiments were performed in triplicate, and each assay was repeated at least three times. Statistical significance was determined by one-way ANOVA with post-hoc Bonferonni analysis (GraphPad Prism).