Samples for which Cryo-EM data was analyzed in this study.

The outlines around sets of samples indicate those that were prepared on the same day and aggregated on the same rotator or same 96-well plate. Abbreviations: AMP=adenosine monophosphate, ADP= adenosine diphosphate, NADP= Nicotinamide adenine dinucleotide phosphate, PIP2= Phosphatidylinositol 4,5-bisphosphate PGE= Prostaglandin E1, DP = Dipyridamole, PT = Pemetrexed, rotǪ = samples during which the rotator stopped, PRLB = low-binding 96- well plate, PRUV = UV transparent 96-well plate, O-450 = orbital shaking at 450 rpm, ND = resolution too low to determine symmetry, NT = non-twisted or too few twisted for 3D analysis, HBS = 25 mM HEPES with 100 mM NaCl, PBS = phosphate buffered saline 1 Samples 2G-36 are 4 pairs of samples from days 4 and 6 of 4 independent aggregations. 2 Samples 56+57 are from days 8 and 12 of the same aggregation 3 Samples 51-53 are Era 2 with 300 kD filter 4 Sample 5G has two distinct 1A polymorphs 5 (5B) in parentheses indicates that it is clearly identified in micrographs but too few segments or irregular twist hindered further analysis 6 The 1m fold variant in sample 17 is does not have the N-terminal strand (residues 10-17) 7 The numbers in parentheses are FSC values without postprocessing for those datasets whose resolution was stuck at > 4.7 Å. 8 The filled circles indicate that coordinates and/or maps have been deposited. Entry codes for deposited data are in Table S1

Purity of α-Syn is correlated with amyloid polymorph formation.

(A) Analytical HPLC traces of SP-pure (red) HPLC-pure (black dashed) and NΔ4 α-Syn (grey) with inlay zoomed in on region around the main peak. (B) Dot plot of the polymorph outcomes for 38 independent aggregation experiments at pH 7.0 or 7.4 with α-Syn prepared with either the standard protocol or with additional HPLC purification. Each column represents one experiment with its Cryo-EM-verified polymorph outcome represented by a colored dot: purple = type 1, blue = type 2 and maroon = type 5. A black circle around the dot refers to samples that were previously reported in (REF) and those with a black point in the middle are samples for which a small molecule additive was included as part of a binding screen.

The type 5 polymorph requires an intact N-terminus and is inhibited by trace truncations.

(A) The Cα traces of the type 5 assembly variants 5A, 5B and 5m showing the extent of the sequence coverage from residues 1-97 and, in the case of some 5A (not shown here) and for all 5B and 5m, an additional detached β-strand whose identity could not be determined. The first 4 residues which are truncated in the NΔ4 variant are highlighted to illustrate their location in the fold. (B) On the right, a view down the fibril axis of the Type 5A interface showing Cα trace with sidechains along with its EM density illustrates the extent of the hydrophobic interactions between Met1, Val3, Leu38, Val40 within the fold and for the N-acetylation at the 5A interface. On the left, a view rotated by 90° depicts three layers on one half of the 5A interface with the middle layer highlighted. For clarity, only the four residues mentioned above are colored while others are white. Dotted lines indicate the H-bonds between backbone atoms as well as those for the amide of the acetylated N-terminal methionine. (C) Z-slices (∼5 Å thick and 150 Å in diameter) for each Cryo-EM reconstruction found in the aggregated samples composed of 0%, 5%, 50% and 100% NΔ4. The colored dots are labeled with the polymorph types. (D) The two novel asymmetric polymorphs 1/7 and 8/9 depicted as Cα traces color coded by polymorph class and with the terminal residues of each chain indicated.

Multiple coexisting polymorphs appear on single micrographs.

(A) A single micrograph from the HPLC-pure α-Syn sample from the Era 1 pH 7.0 (sample 45) overlayed with projections of the initial 3D-models from relion_helix_inimodel2d aligned to fibrils of the same polymorph. The z-slices from the final refined 3D maps are ∼5 Å thick and labelled with the approximate relative amount of that polymorph in the dataset. Below the micrograph are three sets of 2D classes from the same dataset representing three additional fibril polymorphs that were not further analyzed due to their low abundance. (B) A single micrograph from an Era 1 pH 7.0 HPLC-pure α-Syn sample overlayed with the identified coexisting polymorphs, as in A.

Agitation protocol influences polymorph outcomes.

(A) The color-coded regions below the pH scale indicate the polymorph types that are preferred at each pH range (purple = 1, blue = 2, green = 3 ) as described in (Frey et al. 2024). The three colored circles on the pH scale indicate the occurrence of polymorphs under alternative agitation regimes and are connected by lines to their respective Cryo-EM map z-slices. The z-slices (∼5 Å thick and 150 Å in diameter) of the multiple coexisting polymorphs are grouped for each sample with the classes indicated in the color-coded circles. (B) Structural comparison of the type 6A (green carbon atoms) and 6B (orange carbon atoms) assembly variants. On the left, an overlay of a single subunit from each assembly variant highlights their distinct interfaces and unusually large inter-filament distances. The locations of their respective pseudo-C21 axes are indicated as are the positions of the N- and C-terminal residues. The position of the GlcNAc-modified Ser87 is indicated with the “*”. On the right is a superposition of a single unit of the 6A (pdb: 8JEX) and 6B showing the high degree of similarity, including the position of the additional isolated but unidentified strand (here shown in faded colors).

Under reduced oligomer conditions pH reasserts as dominant polymorph selector.

The aggregation outcomes for HPLC-pure full-length and NΔ4 samples prepared under “Era 2” preparation conditions are shown as z-slices of the Cryo-EM maps (∼5 Å thick and 150 Å in diameter) with numbered color-coded balls to indicate the polymorph. The background shading of each aggregation experiment is indicative of the NΔ4%. The names of the type 2-NS variants that lack the N-terminal strand are underlined.

Correlating aggregation kinetics to polymorph outcomes hints at seed-dependence.

(A) The aggregation kinetics of three independent HPLC-pure α-Syn samples followed by simultaneous measurement of absorbance at 340 nm (upper plot) and tyrosine fluorescence polarization (lower plot). The three samples were prepared by serial dilutions from a single 300 μM stock which had only been passed over a 300 kD filter (no 100 kD filter). The polymorph outcomes of the 300 μM and 150 μM aggregations (samples 51 and 52) were analyzed by Cryo-EM and noted as colored balls. (B) and (C) Aggregation kinetics and the polymorphs outcomes like in A for the Era 1 (samples 45-4G) and Era 2 (samples 53-62) mixtures of full-length and NΔ4 HPLC-pure α-Syn. The time axes are all at the same scale for easier comparison and the darker shades of blue represent a higher percentage of NΔ4 as indicated in the figure.

Structural variants within the types 1 and 2 polymorph classes.

Overlay of type 1 structures from the same sample (A) and between samples (B)-(D) including the new “layer-traversing” fold (E) reveal significant variability in the type 1 fold within and between samples. The alignments of the type 1 structures was performed using all residues for A and C and for residues 50-63 for B and D. (F)-(M) Overlays of type 2-NS and 2 structures (using alignments of residues 40-50) reveal fewer variations between three core folds. (F) Five independent type 2 structures from 2A and 2B fibrils are the same within and between samples. (G) The type 2-NS structures are different between 2A-NS and 2B-NS fibrils derived from the same sample. (H) The type 2-NS structures are the same between 2A-NS and 2E-NS fibrils. (I) The Cα trace of the novel 2E-NS polymorph with the residues that make up its larger and more hydrophobic interface indicated. (J) Overlay of 2A and 2A-NS structures shows their shared interface and the differences in the C-terminal portion of the fold with residue Ile88 highlighted to show its shifted but inward facing location in the last strand of the 2-NS fold. The overlays were performed using only residues in the upper monomer for this and the remainder of the panels. (K) Overlay of 2A-NS structures from different samples shows the lack of variability in this polymorph. (L) Overlay of 2B and 2B-NS structures shows their shared interface and the differences angle between the 2nd and 3rd beta strands and in the C-terminal portion of the fold. (M) Overlay of 2B-NS structures from different samples shows the lack of variability in this polymorph. The color of each Cα trace is paired with the sample or polymorph from which it is derived.

Overview of all helical reconstructions reported in this work.

The sample numbers refer to the listing in Table 1 and the images show an approximately a 5 Å thick slice of the cryo-EM map in a 200 Å diameter region that is centered on the helical axis. The contiguous background colors connect samples that were prepared and aggregated at the same time while the black outlines indicate individual samples, some of which contain multiple helical reconstructions. The two samples with red and green circles around the z-slices are to highlight the two examples discussed in the manuscript for which the helical symmetry was somewhat ambiguous. The green circles are around the reconstruction that gave the best resolution/or map quality. The reported FSC resolutions are those before postprocessing in RELION.

Peak intensities of aSyn and its truncation products identified by LC-MS qTOF analysis.

The 25k g pellet fraction of a SP-purity aSyn aggregation that produced type 1m fibrils was solubilized in 6M guanidine and then injected on an Agilent Eclipse Plus C18 column and eluted with a gradient of CH3CN with formic acid as an ion pairing agent for direct analysis on a Bruker Daltonics maXis ESI-ǪTOF instrument. The peaks whose mass fell within 5 ppm of the calculated mass of an aSyn fragment (the largest or second largest peak of a calculated isotope envelope) were tabulated according to their relative intensity. All combinations of acetylated, oxidized and unmodified masses were searched. The green bars represent the full-length aSyn and the black and grey are the N- and C-terminal truncations. The mass of each peak in kD are listed above each bar. The light blue lines connect fragments that are formed from a single cleavage of the full-length protein. Peaks that likely represent protein with an oxidized Methionine are noted with “-Ox”

Sequence coverage of α-Syn endopeptidase GluC peptides in NΔ4 sample.

The green bars represent peptides identified by LC–MS/MS analysis as having an extra CH2 mass, while the amino acid highlighted in red indicates the modified residue.

Structural conservation of type 1 fold within coexisting interface variants.

(A) The type 1D fold from sample 60 is shown as a CA trace with its two chains lighter and darker shades of grey. (B) The structural similarity of the type 1m and 1D coexisting polymorphs is shown in an overlay of the three chains (chains 1 and 2 from type 1D with 1m). The 1D chains are colored as in A and the 1m chain is in light purple. (C) The structural similarity of the type 1A and 1/7 coexisting polymorphs in sample 45 is shown in an overlay with the 1A chains in darker purple and the two chains from 1/7 colored light purple (type 1) and light blue (type 7). The alignment of the structure was performed using only residues 40-90 of the type 1 chain from the 1/7 polymorph. The sidechains for residues 50-58 are shown in the type 7 chain and the type 1 chain with which it coincides in order to show the similarity between the two interfaces. (D) The type 1/7 polymorph from sample 45 (colored as in C) is overlayed with that of sample 48 (in darker shades of purple and blue) to show the conservation of the core fold with the variability in the N and C termini of the type 1 chain.

Fibrils with 5B-like morphology.

Eight selected micrographs with fibrils that display a 5B-like morphology. The unique cross-section of the 5B polymorph makes its identification in micrographs relatively easy and the suspected 5B fibrils are highlighted by maroon dotted lines on each side. The widest section of a 5B-like fibril in the upper right micrograph is overlayed with a maroon scale bar representing 180 Å, matching the end-end longest dimension of the 5B dimer. The z-slice of a 5B map from another dataset is overlaid as a reference with the vertical maroon line also scaled to 180 Å.

Refinement trajectory for a single set of type 5A helical segments.

The ambiguous helical symmetry of sample 24 is highlighted with each set of three images showing the central layer of the final half-maps and the combined final map from a Refind3D job in RELION. The arrows indicate the jobs from which input data where derived. The helical symmetry that was applied during each reconstruction is indicated by the brackets to the left of the images with C1 being used exclusively until after the Bayesian polishing step. The FSC resolutions before and, in parentheses, after postprocessing are indicated. While the best map appeared for the C2 symmetry, its improvement over that of the 2-start symmetry was not large, and the half-maps during the C1 refinements sometimes displayed more of a 2-start symmetry than a C2 symmetry. Even the half-maps with a single iteration did not have the same apparent symmetry (top left job) and these disagreements between half-maps was variable from iteration to iteration and job to job. For example, after the Polish job, the two identical refinements run in C1 led first to a 2-start-like output map (left) and then later to a clear C2 symmetry (right), the latter having a higher FSC resolution. The output of the former of these two jobs was used in both C2 and 2-start helical refinements that each ended with the same final FSC resolution. The refinement with the applied C2 symmetry gave a slightly improved map quality and so this was chosen as the “correct” symmetry for this dataset.

The revised model of type 1m from sample 15 overlayed with its cryo-EM density contoured at 4σ.

The termini of the modeled chains are labelled as well as the positions of Lys10, Lys12 and Gly31 as discussed in the text.

PDB and EMDB entry codes.

The dataset which have been deposited in the PBD and EMDB databases are listed. 1The data from the 5A polymorphs was reconstructed under three different helical symmetries as illustrated in Figure S6. All three maps have been deposited as an example of ambiguous symmetry. 2The information here is redundant with the main text and Table 1 but is included to help orient the reader to the unique features of certain structures.