Established by various experimental approaches, the α-syn fibril core consists of residues 30 – 100 with disordered N- and C-terminal regions (Figure 1a) (8–11). While outside the amyloid core, in the soluble state, the N- and C-termini participate in intra- and inter-molecular interactions (12–14). Fluorescence spectroscopy has been effective for studying α-syn conformation (15–18) and aggregation (19–24); however, kinetic details at the residue level during amyloid formation remain limited. Towards this objective, we sought to define residue-specific behaviors of the N- versus C-terminus during protein aggregation by introducing environmentally-sensitive dansyl (Dns) (25) fluorescent probes at multiple sites (Figure 1b).

We prepared four single-Cys α-syn mutants in the disordered (G7C and Y136C) and proximal fibril core sites (V26C and L100C) at N- and C-terminal regions and derivatized them with the thiol-reactive Dns precursor, 5-((((2-iodoacetyl)amino)ethyl)amino)-naphthalene-1-sulfonic acid. Minimal Dns-protein (1.5%) was used to monitor wild-type (WT) protein aggregation (1.5 μM Dns-protein and 100 μM WT). Because α-syn fibril formation generally exhibits sigmoidal kinetics, we made frequent fluorescence and laser light scattering (LS) measurements during the lag, exponential growth, and mature phases (Figure 1c, d) (26–28).

Similar to the WT protein, we find that Dns-labeled proteins are predominantly disordered in solution, adopt helical structure in the presence of sodium dodecyl sulfate micelles (29), and form β-sheets upon aggregation (Figures S1,2) (30). Consistent with the circular dichroism data for the soluble protein, steady-state and excited-state fluorescence properties of the Dns probe were site independent, reflecting water-exposed fluorophores (mean wavelength, <λ> ~ 525 – 528 nm, Table S1, average excited state lifetime, <τ> ~ 10 ns, data not shown). In addition, morphology of fibrils containing mixtures of WT and Dns-labeled proteins is indistinguishable from WT alone (Figure S3). Using N-acetylcysteine derivative of Dns, we ascertained that the fluorophore itself is not a fibril probe (Figure S4). These results show that the Dns-labeled proteins are not measurably different from the WT protein alone and can be used as reporters of α-syn amyloid formation.

Upon aggregation, Dns fluorescence exhibits dramatic spectroscopic and residue-specific sensitivity (Figures 1c, S5). We observe overall intensity increases (1.6 – 3 fold) and spectral blue shifts (Δ<λ> = 16 – 42 nm) indicating that all Dns sites are sequestered from an aqueous to a more hydrophobic local environment. Unexpectedly, Dns7 and Dns136 (Δ<λ> = 42 and 29 nm, respectively), residues outside the amyloid core were more responsive probes than that of residues proximal to the core, Dns26 and Dns100 (Δ<λ> = 16 and 20 nm, respectively); Dns7 is in the most hydrophobic surrounding whereas Dns26 and Dns100 appear to be markedly more polar. Anisotropy and fluorescence decay data also show increased immobilization (Table S1) and lifetimes (<τ> ~ 12 – 16 ns, data not shown).

To assess whether Dns probes are sensitive to early aggregation events (23, 24), kinetics derived from <λ> were compared to LS measurements which report on macroscopic aggregates (size detection limit ≥ 100 nm). From independent experiments, we confirmed that similar midpoint transitions were obtained for LS and the standard thioflavin T (ThT) assay in detecting fibril formation (Figure S6). Because of inherent sample-to-sample variations in the lag phase (31), during which time little changes in fibril concentration can be detected, we elected to present a full representative data set for the mixture of WT and Dns136 (Figure 1d) to exemplify that despite the uncertainties in the lag times (20 – 30 h), spectroscopic data (<λ>) and respective trend [t₅₀(Δ<λ>) vs. t₅₀(LS), the midpoint transition times] are consistent and reproducible (see Figure S7 for other sites). To ascertain that the Dns probe has little effect on WT aggregation kinetics, we performed concentration dependence studies (1.5 – 9 μM Dns136) and found no apparent differences (Figure S8). We note that while a variety of fluorophores can be chosen for and used to probe α-syn aggregation, larger, more hydrophobic molecules can perturb the kinetics (21, 23, 24).

When monitored by Dns fluorescence, both Dns7 and Dns136 exhibit earlier aggregation kinetics compared to Dns26 and Dns100 (Figure S7). Additionally, Dns7- and Dns136-monitored kinetics preceded the LS curves whereas in contrast, nearly identical behaviors were found for Dns26 and Dns100 compared to that of LS. To quantify the relative residue-specific trend, we used an established analysis method to scale the aggregation time (32, 33). For each set of aggregation data, LS-kinetics were fit to sigmoidal functions and the resulting t₅₀(LS) were used to scale the time axis (t/t₅₀) for all kinetics data (n ≥ 3) (Figure 2). In accord with the unscaled data, smaller scaled t₅₀ (t₅₀^scaled) values were observed when monitored by both Dns7 and Dns136 fluorescence (t₅₀^scaled(Δ< λ>) = 0.870(3)) compared to LS (t₅₀^scaled(LS) = 1.000(3), Table S2) while probes at positions 26 and 100 fully recapitulate the LS data [t₅₀^scaled(Δ< λ>) = 0.990(3) ~ t₅₀^scaled(LS) = 1.000(5)]. Our data suggest that the pathway for amyloid formation for the disordered N- and C-terminal regions develop initially from the ends, followed by residues towards the amyloid core (34).

The increased sensitivity of the N- and C-terminal distal sites may be coupled to the observation that in solution, these regions are involved in transient interactions (12–14, 35, 36). Particularly, if the preferred solution configuration is, either intraprotein (12, 13) (C-to-NAC region, central hydrophobic region that is essential for aggregation) or antiparallel interprotein interaction (14) (N-to-C/C-to-N), then for α-syn to adopt a cross-β fold, where the β-strand residues are suggested to be parallel-in-register (N-to-C/N-to-C) (8, 11, 37), conformational rearrangement at the N- and C-termini must occur.

To shed light on the specific role of Dns7 and Dns136, we examined if the residue-specificity and sensitivity would be retained if we accelerated the lag phase by seeding with 3% WT fibrils (See Supporting Information for experimental details). Since α-syn aggregation kinetics can be described as a nucleation and nucleation dependent elongation mechanism (26–28), the observed early transition could reflect formation of intermediates or oligomers as well as initial filament elongation processes. Upon seeding, the lag phase will be significantly reduced and sometimes even abolished (38); thus, if the detected early events are related to nucleation, the observed differences between t₅₀(Δ<λ>) and t₅₀(LS) should diminish.

Consistent with spontaneous aggregation, the seeded samples showed earlier transitions for residues 7 and 136 (t₅₀^scaled(Δ<λ>) = 0.58(6) and 0.39(3), respectively (39), t₅₀^scaled(LS) = 1.00(1), Figure 2 inset and Table S2). However, when monitored by Dns26 and Dns100 both showed identical fluorescence and LS kinetics data. Accordingly, we propose that these early conformational rearrangements likely occur after nucleation and represent filament formation or elongation processes.

In summary, our study has provided site-specific information on the role of the α-syn N- and C-termini in amyloid formation. Kinetics obtained for Dns fluorophores in the disordered (7 and 136) regions precede proximal (26 and 100) amyloid core sites. Both seeded and spontaneous aggregation kinetics suggest that residues 7 and 136 exhibit local conformational and environmental changes prior to, whereas changes for residues 26 and 100 occur concomitantly with, macroscopic fibril formation. Our results support the hypothesis that local structural reorganization at the N- and C-termini are necessary for α-syn to break the conformational constraints from either intra- or inter-polypeptide electrostatics attraction and thus, favor the formation of parallel, in-register, β-sheet fibrils.

1_si_001