The N terminus of α-synuclein dictates fibril formation

Significance α-Synuclein truncations are prevalent in Parkinson’s disease. While C-terminal truncations are known to accelerate α-synuclein aggregation, the understanding of how N-terminal truncations influence fibril formation lags behind. Here, we show that the removal of N-terminal residues outside the canonical amyloid core weakens the fibril seeding efficiency and hinders propagation. Remarkably, a unique, asymmetric amyloid core is revealed by cryogenic electron microscopy in the N-terminally truncated 41–140 variant, highlighting the strong influence of the N terminus in dictating fibril structure. This work offers insights into the interplay of truncations and full-length proteins; thus, changes in their relative levels could prove to be beneficial for molecular intervention.

Recently, structure determination by cryogenic electron microscopy (cryo-EM) has revealed fibril structures for full-length α-syn (1-140) (24,(30)(31)(32), C-terminal truncations (24,33), phosphorylated Y39 (34), and PD-related mutants, E46K (35,36), H50Q (37), and A53T (38). One striking feature of these fibrils is the eclectic mix of structures, often termed as fibril polymorphism. In fact, it was recently shown that different conformational strains of α-syn fibrils are present in PD and MSA patients (39,40). The outstanding question still remains as to how the same polypeptide chain can produce such a vast number of polymorphic structures. While there are significant structural differences, some features of α-syn fibrils are conserved. All fibrils are formed from a twisting pair of protofilaments with the exception of a H50Q polymorph, which is composed of a single filament. A kernel motif of a bent β-arch appears in all structures. Also, at least one inter-or intramolecular salt bridge between a Lys and Glu is revealed in each structure (24,(30)(31)(32)(33)(34)(35)(36)(37)(38)40), which is not surprising given that there are numerous possibilities for salt bridges between the 14 Lys, 8 Glu, and 2 Asp residues located throughout the first 100 residues in the sequence ( Fig. 1A and SI Appendix, Fig. S1). Generally, residues between 37 and 97 constitute the fibril core with a few exceptions that involve additional residues in the N terminus, which include phosphorylated Y39 fibrils with an extended core of 1-100 (34) and two polymorphs of 1-140 showing interactions of N-terminal β-strands (residues 14-24) (30). Fibrils derived from Significance α-Synuclein truncations are prevalent in Parkinson's disease. While C-terminal truncations are known to accelerate α-synuclein aggregation, the understanding of how N-terminal truncations influence fibril formation lags behind. Here, we show that the removal of N-terminal residues outside the canonical amyloid core weakens the fibril seeding efficiency and hinders propagation. Remarkably, a unique, asymmetric amyloid core is revealed by cryogenic electron microscopy in the N-terminally truncated 41-140 variant, highlighting the strong influence of the N terminus in dictating fibril structure. This work offers insights into the interplay of truncations and full-length proteins; thus, changes in their relative levels could prove to be beneficial for molecular intervention.
brains of MSA patients also indicate additional involvement of the N-terminal region extending to residue 14 (40). Due to the contribution of N-terminal residues in these structures and the fact that C-terminal truncations resulted in modest conformational changes, we hypothesize that N-terminal residues play a greater role in influencing fibril structure.
Here, we sought to understand the role of the N terminus in α-syn fibril formation by removing different N-terminal residues and evaluating their effects on aggregation kinetics, fibril structure, and propagation. Three ΔN-terminal constructs (14-140, 36-140, and 41-140) have been examined, in which the first 13-, 35-, and 40-residues in the N terminus were deleted (Fig. 1A). We specifically chose these sites based on the locations of native Gly residues, which allows us to generate native sequences (i.e., no overhang) upon Tobacco Etch Virus (TEV) protease cleavage of the hexahistidine affinity tag, which facilitates facile protein purification.
All three ΔN-α-syn exhibited different aggregation kinetics and distinct fibril ultrastructural features as determined by thioflavin-T (ThT) fluorescence and transmission electron microscopy (TEM), respectively. In cross-seeding experiments, both fibrillar 36-140 and 41-140 did not seed the full-length (1-140) protein, while 14-140 fared better but less efficient than self-seeding, supportive of the significant impact of removing N-terminal residues in fibril structure. The reverse reaction involving full-length seeds showed that fibril formation of 14-140 and 36-140 but not 41-140 could be accelerated. This observation is explained by the fibril structure adopted by 41-140, which was determined by cryo-EM to an overall resolution of 3.2 Å. Unlike any currently known α-syn structure, the amyloid core is formed by two asymmetric protomers with different amino acid chain lengths, adopting an extended β-hairpin (E61-D98) and the bent β-arch kernel (E46-K96) with a large nonpolar interfilament interface (442 Å 2 ) stabilized by an intermolecular salt bridge between K80 and E83. Collectively, these results establish the important role of N-terminal residues in fibril formation and structure.
Interestingly, lower final ThT intensities were consistently observed for the ΔN-α-syn variants. To determine whether the observed ThT intensity differences are due to relative amounts of aggregated proteins, sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) analysis was performed after ultracentrifugation (SI Appendix, Fig. S2). Densitometric analysis of the aggregation reactions revealed that only for 14-140 does the lower final ThT signal correlate with a reduced amount of aggregated proteins (50% compared to 85% full-length). In contrast, at least 90% of both 36-140 and 41-140 are in the pelleted fraction, which is comparable to 1-140, suggestive of ThT sensitivity to fibril polymorphism.
Evaluation of N-Terminal Truncations on Cellular Viability. The cytotoxicity of 1-140, 14-140, 36-140, and 41-140 α-syn fibrils were examined in two different cell lines: human SH-SY5Y neuroblastoma and rat dopaminergic N27 cells. Cell viability was assessed by mitochondrial activity after 48 h treatment with fibrils (1 μM). Fibrils were prepared by ultracentrifugation, and the pelleted fibrils were quantified prior to treatment. Using a standard 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide assay, impaired mitochondrial activity was consistently observed from all fibril treatments compared to the buffer control (SI Appendix, Fig. S7); however, the uncertainty in our measurements precluded the evaluation whether there were significant differences among the ΔN-variants and the full-length fibrils for either cell line. Based on these results, these ΔN-fibrils have comparable cytotoxicity to cells under the conditions examined.
N-Terminal Residues Are Essential in Fibril Propagation. We next asked whether these ΔN-α-syn fibrils could cross-propagate or seed soluble full-length protein ( Fig. 2A and SI Appendix, Fig. S8). As a reference, self-seeding by preformed 1-140 fibrils (5%) accelerated the aggregation of the full-length protein by abrogating the lag phase and enhancing the final ThT signal. This result is recapitulated at a higher seed concentration (10%, SI Appendix, Fig. S8). By comparison, only the addition of 14-140 fibrils stimulated 1-140 aggregation by shortening the lag time and increasing the final ThT signal, whereas in the presence of either 36-140 or 41-140 fibrils, 1-140 exhibited indistinguishable kinetics to that of itself alone. SDS-PAGE analysis indicated less pelleted material after ultracentrifugation for 14-140 seeded reactions, supporting that 14-140 fibrils are inferior seeds for 1-140 (Fig. 2B). While 14-140 appears to cross-seed on the basis of aggregation kinetics analysis, the resulting TEM images of the 1-140 fibrils formed under all conditions are highly similar to each other (Fig. 2 C-F), providing no evidence for propagation of the 14-140 fibril morphologies.
Having established that 36-140 and 41-140 fibrils cannot crossseed the full-length soluble protein, we investigated whether the reverse reaction would also be prohibitive, which would further substantiate the importance of N-terminal residues in fibril propagation. Efficient seeding was observed in the presence of full-length fibrils for both 14-140 and 36-140, with lag phases abolished and higher final ThT signals (Fig. 3 A and B). In contrast, 1-140 fibrils had no measurable effect on the behavior of 41-140 (Fig. 3C), indicating that seeding did not occur. Even upon doubling the seed concentration, the full-length fibril is unable to seed 41-140, suggestive of structural incompatibility (SI Appendix, Fig. S9). Postaggregation SDS-PAGE analysis revealed that while 14-140 has a higher level of aggregated materials compared to the unseeded control, a similar amount of pelleted protein was obtained for 36-140, cautioning that higher ThT signals do not necessarily correlate to fibril amounts. This cross-seeding by full-length fibrils of 14-140 was corroborated by TEM images taken postaggregation, which revealed more homogeneous fibrils with a half pitch of ∼80 nm (Fig. 3 A, Right), which are not characterized from those of the unseeded reaction (SI Appendix, Fig. S3). Interestingly, while the fibril morphology of 36-140 remained heterogenous (Fig. 3 B, Right), the observed helical pitches of the twisted fibrils were ∼90 and 220 nm, which again are different from those of the unseeded reaction (SI Appendix, Fig. S4), indicating an effect of the full-length seeds on the growth of 36-140 fibrils. It is evident that 1-140 was not able to promote 41-140 fibril growth since the fibrils are highly reminiscent (Fig. 3 C, Right) to those formed from the 41-140 control (SI Appendix, Fig. S5).
Collectively, these cross-seeding experiments suggest that different mechanisms are at play depending on the fibril seeds and soluble proteins used. Since 14-140, but neither 36-140 or 41-140 fibrils, could enhance 1-140 aggregation, residues between 14 and 35 could be important in the productive recruitment of full-length protein. Alternatively, the removal of the N-terminal 35 residues could completely change the fibril structure that is favored only by ΔN-variants. However, when using seeds with an intact N terminus, only 41-140 failed to template, which would indicate that residues 36 through 40 are more important in either scenario, in the replication of the full-length structure and the formation of a different fibril structure.
Fibril Structure of 41-140 Reveals Asymmetric Protofilaments. To gain an understanding to our cross-seeding observations, we sought to obtain fibril structural information by cryo-EM. Out of the three ΔN-constructs studied, 41-140 fibrils had the most homogenous population, and thus, the fibril structure of 41-140 was successfully determined by cryo-EM at near-atomic resolution (SI Appendix, Table S3). The twisted fibrils seen in negative-stain TEM (Fig. 4A) are recapitulated in the cryo-EM micrograph (Fig. 4B). Two-dimensional (2D) class averages indicate a fibril width of ∼6.5 nm (Fig. 4C). Interestingly, the density map of cryo-EM three-dimensional (3D) reconstruction shows two asymmetrical protofilaments intertwining into a left-handed helix with the helical twist of −1.64°with a half pitch ∼52 nm (Fig. 4D), similar to our negative-stain TEM (Fig. 4A). The two protofilaments, designated as chains A and B (Fig. 4E, green and blue), stack along the axis with a helical rise of 4.8 Å as determined by the layer lines from the power spectrum of the micrographs (SI Appendix, Fig. S10). A cross-sectional density map of the fibril is also shown in Fig. 4 E, Bottom Left. The fibril structure was determined at an overall resolution of 3.2 Å (Fig. 4 E, Right and SI Appendix, Fig. S11), with clear sidechain density that permitted unambiguous de novo atomic model building. A different number of residues was resolved for the two asymmetric protofilaments, comprising residues E46-K96 and E61-D98 for chains A and B, respectively. Chain A folds into a β-strand-rich architecture including the β-arch kernel (E57-T72) that is present in other full-length α-syn structures (31)(32)(33). Chain B adopts a dramatically different β-hairpin conformation. Unlike chain A, the lack of density for N-terminal residues G41-K60 suggests conformational flexibility in this region for this protomer. We note that during image processing, a minor second conformation of 41-140 was also observed, showing additional density at the N terminus of chain B (SI Appendix, Fig. S12). While specific residue assignment was not possible, it likely represents residues between G41 and K60. This extra density also results in compaction of residues E46-H50 in chain A that is otherwise more extended in the main polymorph. Similar to published structures (31)(32)(33), there is no visible density for the C terminus (residues 99-140), highlighting its disordered nature in the fibrillar state.
Key Interactions in 41-140 Fibrils. Key interactions that define the structure are highlighted in Fig. 5A. The β-arch motif in chain A is stabilized by a previously documented intramolecular salt bridge between K58 and E61 (24,31,37,38) and hydrogen-bonding between H50 and E83. There is a large interface between the two protofilaments, consisting of residues K60-F94 (chain A) and Q62-I88 (chain B). An intermolecular salt bridge is formed by K80 in chain A and E83 in chain B, an interaction not observed in other structures (SI Appendix, Fig. S1). Additional hydrogenbonding is found between the two Q62 residues. This interface (Fig. 5B) has a calculated buried surface area of 442 Å 2 with an estimated total binding energy of −7.87 kcal/mol which is nearly twice of that of the steric zipper between residues H50 and E57 in the full-length protein (239 Å 2 and −4.49 kcal/mol). We note that K80 from chain B is found buried at the interface, which was also found in the hydrophobic core of the E46K fibril structure, in which it is interacting with a resolved water molecule (36). Although a water molecule is not resolved at the current resolution, there is sufficient space to fit one near the e-amino group. Thus, we surmise that the e-amino group in 41-140 is also hydrogenbonding with a water molecule, similar to that of the E46K fibril structure.
Comparison of 41-140 to Acetylated Full-Length α-syn. A comparison of chains A and B of 41-140 to the acetylated (Ac) 1-140 protomer (Protein Data Bank [PDB] code 6OSJ) is shown in Fig. 5 C and D, respectively. A highly consistent β-arch motif is found between residues E57 through T72 for chain A and Ac1-140 (denoted as β4 and β5 strands in the full-length structure). Clear differences in the region spanning residues G73 to D98 are observed between the two protomers, in which chain A in 41-140 moves away in order to accommodate the insertion of a portion of chain B, which brings E83 of chain B into close proximity to form an intermolecular salt bridge with K80 of chain A. This interaction is unique to 41-140 and has not been observed in other structures (SI Appendix, Fig. S1) (24,(31)(32)(33). In contrast, K80 forms an intramolecular salt bridge with E46 in Ac1-140, bringing β6 and β2 close together (SI Appendix, Fig. S13). In the Ac1-140 structure, β7 is folded back and packs against β6.
The overlay of chain B with Ac1-140 shows striking conformational differences. There are only 13 overlapping residues, A76 to A89, which span strands β6 and β7. This extended hairpin conformation composed of residues G68 to V95 appears unique to this structure thus far. Collectively, these results support the idea that N-terminal residues modulate the mechanism of amyloid assembly and contribute to the molecular interactions that define fibril structure.

Discussion
Fibril polymorphism is a defining feature of pathological amyloids, yet mechanistic insight into how multiple structures are generated from a given polypeptide chain is still an open question. In this work, we explored the impact of the N terminus on α-syn fibril formation by using a series of N-terminal truncations. The removal of 13-, 35-, and 40-residues from the N terminus-influenced aggregation behaviors (Fig. 1), and all three truncations have distinguishing ultrastructural features different from those exhibited by the full-length α-syn fibrils (Fig. 1 C-F and SI Appendix, Figs. S3-S5). Structure determination of 41-140 by cryo-EM revealed a fibril conformation in which two asymmetric protofilaments with different lengths of ordered residues interwind to form a twisted fibril (Fig. 4). A large interface is formed between residues K60 and F94 (chain A) and Q62 and I88 (chain B), and the protomers are oriented in an N-to-C/N-to-C geometry (Fig. 5). An intermolecular salt bridge forms between K80 and E83, exemplifying the ability of the polypeptide chain to utilize the many available Lys and Glu residues as stabilizing factors (Fig. 1A). Including this work, there are currently a total of five unique intramolecular (K45-E57, K58-E57, K58-E61, K80-E46, and K96-E61) and six intermolecular (K45-E46, K45-E57, K58-E35, K60-E28, K80-E61, and K80-E83) Lys-Glu salt bridges found in α-syn fibril structures (SI Appendix, Fig. S1).
Since the N-terminal region (residues 1 through 36) is unresolved and thus presumed to be largely disordered in most full-length α-syn fibril structures, the impact of its removal on the 41-140 structure was initially surprising. However, recent work has demonstrated the participation of N-terminal residues in two 1-140 polymorphs in which residues V16 through E20 form a hydrophobic steric zipper with residues S87 through A91 (30). Also, fibrils derived from MSA patients showed a larger core comprising residues 14-99, with an extended N terminus forming a cross-β hairpin (40). Even our own work on C-terminally truncated Ac1-103 α-syn suggested the presence of additional N-terminal residues (24). Taken together, the N terminus plays a pivotal role in fibril assembly.
Using cross-fibril propagation experiments, we conclusively show the importance of the N terminus in α-syn fibril formation. As the N terminus is removed, ΔN-α-syn fibrils were shown to be poor seeds for soluble full-length α-syn (Fig. 2). This is noteworthy because the removal of C-terminal residues has the opposing effect in which it creates potent seeds and can efficiently propagate soluble full-length α-syn (17,24). That said, the addition of 14-140 fibrils did result in a lag time reduction, suggesting that residues 15 through 40 play a more crucial role. However, due to the unique fibril structure that is adopted by 41-140 (Fig. 4), we speculate that this fibril structure is inaccessible to 1-140 even when a template is provided. This result is surprising given that the structurally resolved core only involves residues 46 and beyond. Moreover, it suggests that the presence of the first 40 residues is prohibitive for propagation. While it is unknown whether 36-140 also adopts a different fibril structure, it is evident that it behaves more like 41-140. Thus, we propose that 1-140 and 14-140 fibrils are more alike compared to 36-140 and 41-140, which may be more similar to each other.
From a mechanistic standpoint, two recent studies also have highlighted the involvement of N-terminal residues in fibril formation. Specifically, residues 1-11 (42) and 1-12 (43) are identified in monomer addition to the ends of seeds (i.e., elongation) and monomer binding to the fibril surface (i.e., secondary nucleation), respectively. These results are in agreement with our cross-propagation experiments using ΔN-α-syn fibrils, in which the effect of the first 13 residues is clearly documented (Fig. 2). However, our data contradicts the conclusion that intermolecular contacts from residues 1 through 35 are essential as lag phases of both 14-140 and 36-140 can be bypassed by the addition of 1-140 seeds (Fig. 3), which indicate that monomers lacking either 1-13 or 1-35 can add to fulllength seeds and form fibrils. Furthermore, since 41-140 could not be seeded by 1-140 fibrils, it points to the involvement of the intervening hydrophobic residues, 37 VLYV 40 , in facilitating the formation of the full-length structure. This postulate is supported by recent work on a Δ36-42 α-syn mutant (28), which suggested that this region modulates fibril formation. Collectively, these data affirm the strong influence of the N terminus and its interactions (or the lack thereof) in driving full-length α-syn amyloid formation. Lastly, the role of the C terminus should not be overlooked due to favorable electrostatic interactions with the N-terminal region, which is supported by the differences in limited-PK digestion among the variants compared to the full-length protein. Thus, it is imperative for future investigations to evaluate the effect of removing both Nand C-terminal residues and to address the interplay between the two termini in how a given α-syn structure is faithfully propagated.
With ΔN-α-syn modulating aggregation kinetics as well as fibril structure and propagation, a central question is whether these species are beneficial or detrimental in a cellular environment. It is known that both N-and C-terminal truncations are prevalent in both PD and healthy individuals (44), and some of their generation can be attributed to lysosomal protease activity (16,17). While 41-140 has not been identified in patients, it is related to another known truncation, 39-140, which can be generated by CtsD activity (16). These proteases have long been implicated in α-syn turnover and given the opposing fibril propagation behaviors of ΔNand ΔC-α-syn, it is imperative that future studies consider contributions from both fibril variants. With the enhanced seeding capabilities of C-terminal truncations (24), one could envision the presence of ΔN-α-syn potentially attenuating this deleterious effect. Such a scenario could occur in the lysosomal environment where levels of protease activity dictate the truncation generated. However, an absence of the N terminus could also reduce α-syn membrane binding, thus shifting the α-syn equilibrium in the cellular milieu and lead to dysfunction. This potential tug of war presents an interesting interplay between different truncation variants, providing future targets in tempering α-syn aggregation.

Materials and Methods
Details regarding reagents, molecular cloning, protein expression and purification, aggregation kinetics, TEM, liquid chromatography-mass spectrometry, cell culture and cell viability assay, cryo-EM sample preparation, and data collection and processing as well as model and refinement are provided in SI Appendix. Black boxes highlight intramolecular interactions of a salt bridge between K58 and E61 and hydrogen-bonding between H50 and E83 in chain A (blue). Magenta boxes indicate intermolecular interactions of a salt bridge between K80 (chain A, blue) and E83 (chain B, green) and hydrogen-bonding between Q62 and Q62 from each protomer. (B) The interface between chain A and B is composed of K60-F94 and Q62-I88, respectively. The buried surface area is calculated to be 442 Å 2 . Overlay of Ac1-140 (PDB 6OSJ, red) with (C) chain A (blue) and (D) chain B (green) with β-strand assignments based on the Ac1-140 structure. Residues E57-T72 (chain A) adopt the conserved β-arch kernel with identical side-chain orientations as the full-length protein. For chain B, a minimal conserved region is found between residues A76 and A89 with their side chains shown.