Familial Mutations May Switch Conformational Preferences in α‑Synuclein Fibrils

Abstract

The pathogenesis of Parkinson’s disease is closely associated with the aggregation of the α-synuclein protein. Several familial mutants have been identified and shown to affect the aggregation kinetics of α-synuclein through distinct molecular mechanisms. Quantitative evaluation of the relative stabilities of the wild type and mutant fibrils is crucial for understanding the aggregation process and identifying the key component steps. In this work, we examined two topologically different α-synuclein fibril structures that are either determined by solid-state NMR method or modeled based on solid-state NMR data, and characterized their conformational properties and thermodynamic stabilities using molecular dynamics simulations. We show that the two fibril morphologies have comparable size, solvent exposure, secondary structures, and similar molecule/peptide binding modes; but different stabilities. Familial mutations do not significantly alter the overall fibril structures but shift their relative stabilities. Distinct mutations display altered fibril conformational behavior, suggesting different propagation preferences, reminiscent of cross-seeding among prion strains and tau deletion mutants. The simulations quantify the hydrophobic and electrostatic interactions, as well as N-terminal dynamics, that may contribute to the divergent aggregation kinetics that has been observed experimentally. Our results indicate that small molecule and peptide inhibitors may share the same binding region, providing molecular recognition that is independent of fibril conformation.

Graphical Abstract

INTRODUCTION

The presence of Lewy bodies and Lewy neurites in the patient’s brain is a prominent neuropathological hallmark of Parkinson’s disease (PD).^1–4 Their major constituent is α-synuclein (α-Syn) fibrils, the final stage of aggregation by α-Syn protein.^5–7 α-Syn has been implicated in synaptic plasticity, vesicle dynamics, dopamine metabolism, suppression of apoptosis, regulation of glucose activity, modulation of calmodulin activity, and ATP synthase function.^8–13 Increasing evidence has shown that α-Syn aggregates, especially oligomers, may play a key role in triggering neurotoxicity,¹⁴ possibly by interacting with and disrupting membranes,¹⁵ and impairing protein degradation and function of several organelles like the mitochondria and endoplasmic reticulum.^14–18 α-Syn oligomers can spread between interconnected brain regions, and thereby act as seeds for prion-like pathology.^19–21

The identification of a series of rare familial mutations, A30P,²² E46K,²³ H50Q,^24,25 G51D,^26,27 A53E,²⁸ and A53T,²⁹ in the SNCA gene that encodes the α-Syn protein further establishes it as causative for the pathogenesis of related neurodegenerative disorders such as PD and dementia with Lewy bodies.³⁰ The effects of these mutations on α-Syn aggregation kinetics and membrane binding have been investigated in in vitro experiments.^31–47 However, aggregation is sensitive to experimental conditions such as solution pH, temperature, protein concentration, and concentration of metal ions, and so it can be difficult to obtain consistent results for specific mutants. Here we give an overview of the conclusions relating to the effects of point mutations on the kinetics of aggregation of α-Syn into oligomers (on- or off-pathway) and fibrils. Compared to wild type α-Syn, the A30P variant accelerates nonfibrillar aggregate formation (oligomers or protofibrils) but significantly reduces the formation of mature fibrils.^31,33,38,44 However, recently A30P was also shown to retard the formation of both oligomers and fibrils.⁴⁵ The E46K mutant increases membrane affinity,⁴⁸ and accelerates α-Syn aggregation and fibrillation.^41,42 H50Q mutation dramatically accelerates α-Syn aggregation into fibrils and strongly stabilizes fibrils.³⁶ The G51D mutation significantly slows down the rate of α-Syn aggregation into oligomers and fibrils, but forms amorphous aggregates.^27,49 The A53E mutation significantly reduces aggregation in vitro and at cell membranes, but can still form very thin amyloid fibrils with similar morphology as wild type.^40,43 The A53T mutant considerably accelerates α-Syn aggregation, and fibrillation is faster than wild type.^31,46,50 It has been suggested that these mutations could alter the initiation and proliferation steps.³⁷ These studies provide insights into the effects of familial mutations on the kinetics of aggregation, but little is known about the influence of each mutant on the relative thermodynamic stability of α-Syn fibrils.

α-Syn monomer predominantly displays α-helical conformations in the presence of biological micelles or micelles of detergent sodium lauroylsarcosinate,^51,52 but it is an intrinsically disordered protein (IDP) under normal physiological conditions, adopting a random coil conformation.^53,54 The core of α-Syn fibril is organized in a parallel in-register structure rich in cross β-sheet conformation,^55,56 a characteristic of fibrils such as Alzheimer’s disease-associated amyloid β(Aβ).^57–59 The full sequence of α-Syn contains 140 residues, and can be divided into three distinct domains: the amphipathic N-terminal region (residues 1–60), which is believed to be responsible for membrane binding;^60,61 the highly hydrophobic nonamyloid-β component (NAC) domain (residues 61–95), which is assumed to be essential for α-Syn aggregation;⁶² and the acidic (negatively charged) C-terminal domain (residues 96–140), which is critical for the chaperone-like activity of α-Syn.^63,64 In addition to interactions with membranes, the N-terminal also plays an important role in the formation of α-Syn oligomers,⁴⁵ and all familial variants occur in this domain, indicating the critical role of this region in modulating the aggregation tendency of α-Syn. Also note that E46 is in one highly conserved hexameric repeat of K⁴³T⁴⁴K⁴⁵E⁴⁶G⁴⁷V⁴⁸. This N-terminal repeat domain (XKTKEGVXXXX) has been shown to mediate the aggregation kinetics of α-Syn.^65,66 However, the underlying mechanism of how the N-terminus tunes the aggregation kinetics is not well understood. More importantly, α-Syn aggregates, oligomers or fibrils, are potential targets for therapeutic intervention. Recently, the human monoclonal antibody, aducanumab was shown to selectively interact with Aβ aggregates and reduce Aβ plaques in the brains of Alzheimer’s patients (AD).⁶⁷ As such, elucidation of the effects of familial mutations on the conformation and relative stability of α-Syn fibrils should provide important information on modulating α-Syn aggregate formation.

Amyloid polymorphism can affect seed and fibril propagation in neurodegenerative diseases.⁶⁸ Single mutations can modulate the distribution of different conformations, thereby influencing fibril growth, as in the case of α-Syn,⁶⁹ tau protein amyloid formation,⁷⁰ and prion species.⁷¹ On the basis of solid-state NMR (ssNMR) experimental results,^72,73 there are at least two polymorphic variants of α-Syn fibrils (Figures 1 and 2). A pathogenic fibril of full-length human α-Syn was recently determined by ssNMR method, in which the core residues are arranged in parallel, in-register β-sheets with a Greek key topology (G-Key)⁷³ (Figure 1). Based on the Riek model,⁶³ the structure of NAC was predicted by Miller’s group;⁷⁴ and a different fibril structure of α-Syn(30–110) was also predicted,⁷⁵ which has many residue contacts (for example Val³⁷-Ala⁵³ and His⁵⁰-Asn⁶⁵) that are not possible in the G-Key structure. Both ssNMR results⁶³ and computational⁷⁵ studies indicated that the α-Syn(30–110) may have five β-strands packed in the conventional β-strand-loop-β-strand fold (5-Fold) (Figure 2). Therefore, it is important to compare the stabilities of the two human α-synuclein fibril structures and investigate the possible effects of known mutations on their relative stabilities.

Figure 1.

Representative conformations of α-Syn fibrils in the Greek key fold (G-Key). Starting (t = 0 ns) WT structure and structures after 200 ns are shown for WT and all mutants. Mutation sites are shown in CPK representation, as well as transparent surface areas colored according to the residue types: red for acidic residue; blue for basic residue; green for polar residue; and white for nonpolar residue. For the sake of clarity, G51 is not highlighted in the WT structures.

Figure 2.

Representative conformations of α-Syn fibrils in the 5-Fold. Representation and color are the same as shown in Figure 1.

In this study, we characterized the conformational dynamics of the two α-Syn fibril structures using extensive molecular dynamics (MD) simulations in aqueous solution. The effects of each point mutation (A30P, E46K, H50Q, G51D, A53E, and A53T) on fibril structure and stability were compared. The data from our systematic simulation study reveals that these two fibril structures, although exhibiting quite different folds, are similar in their overall structural properties and interaction with small molecules/peptides. The point mutations do not significantly alter the general folds of the two fibril structures, but there are structural changes in regions far away from the mutation sites. We find that α-Syn fibril is more stable in the G-Key fold than in the 5-Fold. However, the mutants display different conformational preferences. Computed fibril stabilities do not correlate with experientally observed aggregation kinetics, indicating that the divergent aggregation propensity of these mutants is more likely due to the early aggregation stage that is modulated by hydrophobic and electrostatic interactions, as well as the N-terminal dynamics of α-Syn proteins.

RESULTS

Two Morphologies of α-Syn Fibrils Have Comparable Structural Properties.

We first compare the structural similarity and difference of the two α-Syn fibril structures. The folds are distinctly different (Figure 1 and Figure 2), yet the computed molecular dynamics structures reveal that they are comparable in overall fibril size, solvent-exposed area (Figure 3), and secondary structure (Figure 4). The average radii of gyration for G-Key and 5-Fold fibrils are 27.2 ± 0.1 and 27.7 ± 0.1 Å, respectively (Figure S1). Moreover, both fibrils have a core dimension of ∼45 Å. The calculated values of solvent-exposed surface area (SASA) are 25 522 ± 28 Ų and 25 857 ± 175 Ų for G-Key and 5-Fold fibrils, respectively. The values of SASA for the component monomers in G-Key and 5-Fold fibrils are ∼4034 ± 10 and ∼4206 ±17 Ų, respectively. Such results seem to indicate that 5-Fold fibril is slightly more exposed to solvent. However, the relative differences in the SASA of fibril and monomer are only 1% and 4%, respectively, illustrating how rapidly the difference in SASA of the two fibril morphologies would decrease with increasing number of monomers. Both fibril cores are rich in β structures (Figure 4) but the distribution of these β structures differs from each other (Figures 5 and 6). Notably, the NAC region (residues 61–95) in both fibril models has a high tendency to form β structures. The average percentages of β structure in G-Key and 5-Fold are 28% and 33%, respectively, implying comparable contents of β structure in the two fibrils with different morphologies. The high degree of similarity in populations of other secondary structures (random coil, bend, helix, and turn) can also be seen in Figure 4, with the largest difference less than 5% in turn structures. Both fibril structures display helical conformations (predominantly α-helix) within the N-terminal region containing residues 20–40, with Val²⁶-Ala³⁰ showing high propensity to form helix conformations (Figures 5 and 6). The relative stability of these two fibril structures was estimated by their calculated conformational energies, as shown in Figure 7. The α-Syn fibril structure in G-Key fold has a signifcantly lower energy (−10021 ± 17 kcal/mol) than that in 5-Fold (−9684±22 kcal/mol). Torsional energies contribute the most to this difference (Figure S4), suggesting that β-sheet-turn-β-sheet (U-turn) structures in the 5-Fold fibril may induce greater backbone strain. This strain could be compensated for, and the 5-Fold structure made stable, under specific experimental conditions, or with additional optimization of its internal interactions (for example, mutation-triggered shifts, as discussed later). The different folds make the interactions between NT and NAC regions, and between CT and NAC regions, more favorable in the G-Key model than in the 5-Fold model (Figure 8), which may contribute to the stability of α-Syn fibril structure in G-Key fold. In addition, no direct interactions were observed between NT and CT regions in both fibril structures.

Figure 3.

(A) Values of radius of gyration (R_g), and solvent accessible surface area (SASA) for (B) fibril and (C) monomer. The SASA of (buried) monomer was averaged over the SASA of the three inner monomers.

Figure 4.

Overall percentage of secondary structures for WT and mutants of α-Syn fibrils. The helix structures include α-helix, π-helix, and 310-helix, but α-helix is the major contribution.

Figure 5.

(A) Residue-specific population of helical conformation. (B) Residue-specific β-strand conformation of WT α-Syn fibrils in G-Key fold. (C–H) Difference in residue-specific β-strand conformation of α-Syn familial mutants relative to the WT.

Figure 6.

(A) Residue-specific population of helical conformation. (B) Residue-specific β-strand conformation of WT α-Syn fibrils in 5-Fold. (C–H) Difference in residue-specific β-strand conformation of α-Syn familial mutants relative to the WT.

Figure 7.

(A) Conformational energy (enthalpy) of WT and familial mutants of α-Syn fibrils. (B) Conformational energy difference of WT and familial mutants of α-Syn fibrils (G-Key fold relative to 5-Fold).

Figure 8.

Interaction energy between different regions (N-terminal, NT; C-terminal, CT; and NAC) of WT and familial mutants of α-Syn fibrils.

We also compared the ligand binding propensity of the α-Syn fibril structure in the different folds by performing molecular docking. Small molecule epigallocatechin-3-gallate (EGCG) and short peptides (WWP, cyclo-WWP), which could prevent the formation of α-Syn fibril,^76–78 were blindly docked to the α-Syn fibril structure using the method outlined in our previous study.⁷⁵ The docking results are summaried in Table S3. EGCG predominantly binds to the NAC domain and the N-terminal of α-Syn fibril structure, independent of the type of fold. The short β-peptides preferentially bind to the NAC domain, and the cyclo-peptide interacts with the α-Syn C-terminal. The above docking results indicate that small molecule/peptide inhibitors could potentially bind to the same region of α-Syn fibrils in both morphologies.

Familial Mutations Do Not Significantly Alter WT α-Syn Fibril Structures.

Having characterized the WT α-Syn fibril structures, we now examine the influence of each N-terminal mutation on the fibril structures. We find no significant alterations in the overall structural properties, but subtle mutation-specific changes are found. Figures 3 and S1 show that the radius of gyration of all mutants varies from 26.3 to 28.8 Å, and from 27.7 to 28.5 Å for α-Syn fibrils in G-Key and 5-Fold, respectively, comparable to WT fibrils. Computed SASA ranges for WT and mutants were also very similar, for fibril and monomer structures in the two different folds. These findings suggest that the single-site mutations do not give a notable difference in the overall size and solvent exposure of the α-Syn fibrils.

The changes in the contents of secondary structures vary by mutant (Figure 4), and all changes are within the same order of magnitude relative to that of WT fibril structures. The same mutants in the 5-Fold structure possess comparable or slightly higher contents of β-strand compared to those in the G-key motif. The same mutant in different folds might have different effects on the secondary structures. For example, slight (2%) decrease and increase in β content was observed for G51D in G-Key and 5-Fold, respectively. In addition, no significant mutation-triggered alterations in the overall helical conformations were observed. The residue-specific distributions of helix and β-strand for mutants in G-Key and 5-Fold are shown in Figures 5 and 6, respectively. For all but A30P, mutants in the G-Key model increase the helical structures differently in the same N-terminal region compared to the WT fibril. Lower helix content was also observed in the C-terminal in variants E46K and H50Q (Figure 5). Similarly, most mutants (E46K, H50Q, A53E, and A53T) in the 5-Fold structure promote the formation of helical conformation in the same N-terminal region as WT (Figure 6). All mutants in the G-Key fold increase the propensity for β structure around Val³⁷ and Leu,³⁸ and decrease the β-structure tendency of regions Lys⁴⁵-Val⁵⁵ and Lys⁶⁰-Val⁶⁶ to varying extents (Figure 5). Note that both these two regions run into highly conserved hexameric repeats K⁴³T⁴⁴K⁴⁵E⁴⁶G⁴⁷V⁴⁸ and K⁵⁸T⁵⁹K⁶⁰E⁶¹Q⁶²V⁶³, respectively. Here, we show that charge swap of −1 → + 1 in this region (E46K) or charge alteration of 0 → −1 (G51D and A53E) in regions close to the K⁵⁸T⁵⁹K⁶⁰E⁶¹Q⁶²V⁶³ motif disrupt the secondary structures of these repeats. Moreover, discrete regions involving residues 78–98 that are far away from the mutation sites are also affected, but such perturbations are generally small except that residues Lys⁸⁰ and Thr⁸¹ in A53E display increased β structure by 73% and 55% compared to WT, respectively (Figure 5).

In all cases of α-Syn mutants in 5-Fold (Figure 6), the mutation sites are in the flexible N-terminal (A30), the first β strand (E46K), or the second β strand (H50, G51, and A53). As expected, point mutations not only affect the propensity of β structure of their neighboring residues, but could influence residues of other β strands through triggered interactions in adjacent β strands. In addition to direct propagation by intramolecular interactions, long-range effects including electrostatic interactions and allosteric effects could also contribute to the divergent alterations of secondary structures caused by point mutations.⁷⁹ A30P illustrates such long-range effects. The changes in β structures are on the same order of magnitude as for the other mutants even though this mutation site is in the dynamic N-terminal. Taken together, our simulation results suggest that residue-specific changes allosterically spread through the fibril, even though point mutations do not significantly alter the overall structures of α-Syn fibrils.

Conformational Preference of Mutant α-Syn Fibril Structures.

The conformational energy for each mutant was calculated and summarized in Figure 7. Note that the difference in conformational energy is about 337 kcal/mol, indicating that WT α-Syn fibril may predominantly assume the G-Key conformation. However, the conformational energy difference for the mutants decreases to about 44, 11, 101, 70, 61, and 104 kcal/mol for A30P, E46K, H50Q, G51D, A53E, and A53T, respectively. The computed relatively small difference in the conformational energy of E46K (11 kcal/mol) suggests that the E46K fibril is similarly stable in both folds. We note that the E46K mutant behaves differently than the other mutants. In the G-Key morphology, E46 originally establishes a salt-bridge with K80, and replacement of E with K results in unfavorable electrostatic repulsion between K46 and K80 (Figure 1), destabilizing the E46K fibril structure in G-Key fold. However, in the 5-Fold topology, we observed that one of the C-terminal residues E110 interacts with K46 by forming a salt-bridge (Figure 2), which seems to stabilize the E46K mutant. (The possible conformation switch between G-Key and 5-Fold structure is provided below.) The largest difference in the conformational energy (104 kcal/mol) indicates that A53T fibril strongly prefers G-Key conformation. Similarly, we found that A53E prefers G-Key fold, while A30P, H50Q, and G51D are less strongly penalized in the 5-Fold conformation. This finding suggests that single point mutation in α-Syn fibril leads to different conformational preference, and thereby different populations under specific conditions, which is consistent with the known polymorphism of α-Syn fibrils.^56,80

E46K Mutant in 5-Fold Morphology Prefers Extended Conformation.

In order to further examine the contribution of the interactions between K46 and E110 to the stability of E46K mutant in 5-Fold, we constructed another E46K mutant (henceforth referred to as E46K_II) where all K46 residues interact with corresponding E110 residues within the same monomers (Figure 9), and performed a 100 ns MD simulation without constraints after minimization and equilibration with harmonic constraints between K46 and E110 residues. Only three E110 residues persistently interact with five K46 residues, and the other two E110 residues move far away from K46 in aqueous solution, indicating that the E46K_II mutant might not predominate. The E46K_II fibril structure shows a reduced radius of gyration of (∼25.4 Å) compared to the extended E46K (∼27.9 Å) (Figure 3). No significant differences in SASAs of both fibril and monomer are observed between the compact E46K_II and extended E46K mutant (Figure 3). As expected, the interactions between NAC and C-terminal regions increase due to the formation of salt-bridges (Figure 8), and E46K has a slightly better time-averaged conformational energy in the extended conformation (−9863 ± 8 kcal/mol for E46K, and −9789 ± 42 kcal/mol for E46K_II), suggesting that the compact E46K morphology is less favorable than the extended morphology (Figure 9). Further analyses of the contribution of each component to the total energy suggest that compact E46K_II has improved van der Waals contacts (by 1475 kcal/mol) with an associated loss in solvation of 2097 kcal/mol, indicating that a delicate balance between hydrophobic interactions and solvation energy could determine whether E46K will adopt an extended or compact morphology. The relatively small difference in electrostatic interactions (3128 ± 4 kcal/mol and 3096 ± 5 kcal/mol for E46K and E46K_II, respectively) implies that the contribution of salt-bridges beween K46 and E110 to the stability of fibril structures are not so significant compared to the other components like van der Waals and solvation energies caused by the alteration of the organization of the C-terminal.

Figure 9.

(A) Conformations of E46K_II at the starting (t = 0 ns) and 100 ns of MD simulations. (B) Conformational energy of two polymorphs of E46K mutant, as well as different contributions to the total energy.

Note that the formation of different morphology of α-Syn fibrils is sensitive to solution conditions.^81,82 The fibril structure in G-Key was grown in 50 mM sodium phosphate buffer,⁷³ whereas the fibril structure of 5-Fold was grown in 0.01% sodium azide.⁷² In most of our simulations we simply added Cl⁻ ions to neutralize the net protein charge in each system (8 mM Cl⁻ and 10 mM Cl⁻ for the G-Key and 5-Fold structures, respectively). To further probe the effect of ion concentrations on the weak salt-bridge that formed in the E46K system, we performed another 100 ns-MD simulation of E46K in physiological strength 150 mM NaCl, and found that the average number of salt-bridges decreases from 3.3 to 0.8 due to electrostatic screening by the salt ions. However, no significant changes in the general structure properties and conformational energies were observed (Table S2).

DISCUSSION

Familial Mutations Contribute to the Polymorphism of α-Syn Fibrils.

The polymorphism of amyloid fibrils has been well established.^56,80 For example, Aβ42 fibrils, which are one of the hallmarks of AD, have four NMR structures available in the Protein Data Bank (PDB ID: 2BEG,⁵⁷ 2MXU,⁵⁸ 5KK3,⁵⁹ and 2NAO⁸³), showing distinctive morphologies (U-turn containing two β-strands, S-shape containing three β-strands, and S-shape containing four β-strands) under different experimental conditions. Solid-state NMR experiments suggested that the landscape of α-Syn fibrils may be more complex.^7,56 Previous studies have characterized two α-Syn strains with different structures, levels of toxicity, seeding, and propagation properties,^84,85 but to date only one ssNMR structure has been reported (PDB ID: 2N0A)⁷³ for α-Syn fibril (G-Key). The 5-Fold morphology of α-Syn fibril was suggested in earlier ssNMR experimental studies,^55,72 but the atomic molecular structure was only recently proposed from computational models.⁷⁵ Based on the atomic structures of G-Key and 5-Fold, we characterized the structural properties of both WT and six familial mutants. It is interesting to find that the two morphologies have comparable size, solvent exposure, and secondary structures, implying that the morphological changes in α-Syn fibrils may originate from different folds (tertiary and/or quaternary, Figure S5), consistent with recent experimental observations.⁸⁶ Moreover, no significant alterations in the overall size, solvent exposure, and secondary structures were observed between WT and mutants in the two polymorphs, suggesting that mutants could exhibit the same structure as either of the WT forms. However, fibrils could consist of various number of monomers in in vitro experiments, and so caution must be used in comparing the fibril size estimated here using the radius of gyration with experimental data.⁸⁷

Experimentally, A30P has been observed to show a conserved β-sheet core and adopt the WT fibril structure.³³ However, A30P could also form a fibril structure different from WT.³⁸ H50Q, A53E, and A53T have been suggested to have minor secondary structure perturbations.^{39,40,43,46,88} By contrast, E46K could have significant effects, and may change the overall structural arrangement of the fibrils.⁴⁶ E46K has also been reported to enhance the interactions of the C-terminal with either the NAC or the N-terminal, resulting in a more compact structure compared to WT fibrils.^35,41,45 In line with these experimental observations, we found that E46K has the highest tendency to shift between 5-Fold and G-Key fold morphologies due to electrostatic repulsion between proximate K46 and K80 sites. In the proposed structural model E46K_II, we observed interactions of the N-terminal residues K46 with the C-terminal residues E110, which could contribute to enhanced interactions between the N- and the C-terminal regions. Thus, the E46K mutant is highly polymorphic, and could induce new morphologies by rearranging its C-terminal region. Recently, the Arctic E22G mutant of Aβ40 has been identified to yield a highly polymorphic Aβ40 fibril landscape.⁸⁹ Thus, introduction of an additional charge, or electrostatic interactions in general, can have profound effects on amyloid polymorphs.

Cross-seeding experiments suggest that A30P monomers cannot bind to A53T fibril seeds, and A53T monomers cannot bind to A30P fibril seeds either.⁶⁹ Consistent with this finding, our results show that A30P fibril has a preference for 5-Fold morphology, but A53T fibril prefers G-Key fold. In addition, formation of helical structures in the N-terminal (residues 20–30) of all fibril models is consistent with experimental observations of α-helices at residues 20–34.⁹⁰ This helical segment may play a role in binding to cell membranes.⁹¹ However, direct correlation of the membrane binding affinity to the contents of helical conformations is not feasible because the first 20 N-terminal residues, which display high tendency to form helices, are not included (by necessity, see Methods) in the present fibril structures.

Familial Mutations Alter the Conformational Preferences of α-Syn Fibrils.

The effects of PD-related familial mutations of α-Syn on the aggregation thermodynamics, kinetics, and the underlying molecular mechanism has been intensively studied. Differential scanning calorimetry experiments suggested a positive enthalpy change (heat absorption) in α-Syn fibril formation.⁹² Recent studies revealed that point mutations dramatically affect the rate of lipid-induced fibril formation and secondary nucleation.³⁷ The presence of nonfibrillar α-Syn oligomers has been suggested in the aggregation process of amyloid.^47,93 These oligomers may hinder elongation of α-Syn fibrils and slow down the kinetics of amyloid formation in a concentration-dependent manner.⁹⁴ On the other hand, certain α-Syn oligomers that are kinetically trapped during the formation of fibrils have been characterized and found to have a much lower tendency to elongate due to the difference in the arrangement of β-strands in the oligomers (antiparallel) and fibrils (parallel).⁹⁵ The rearrangement of β-strands involved in the formation of fibrils in different folds from those oligomers seems rather slow. These minor populations of relatively unstable species prevent amyloid formation by increasing the energy barrier for amyloid assembly.⁹⁶ We thus speculate that the mutations could alter the relative population of oligomeric states that determine the aggregation kinetics of α-Syn.

The coexistence of polymorphs of the same mutants or redistribution of the ensemble of conformers of α-Syn fibrils has been suggested in NMR studies of A30P and A53T mutations.⁹⁷ Based on the calculated conformational energies, we further show that WT α-Syn fibril preferentially adopts G-Key conformation. But different mutants shift the relative stability (and relative population) of the same α-Syn fibril in both morphologies. Direct correlation of our findings with experimental aggregation kinetics is not straightforward as we found there is no clear relationship between the computed conformational preference of α-Syn fibrils and their known aggregation propensity. A30P has been reported to form fibrils more slowly than WT,^{31,33,38,44,45} and our results indicate that the A30P mutant becomes more stable when adopting 5-Fold morphology. Fibrillation in H50Q and A53T is more rapid,^31,36,46,50 and their fibrils are more prone to form 5-Fold and G-Key morphology, respectively. Both G51D and A53E have excess negative charge, and have been reported to form fibrils significantly slower than WT,^27,40,43,49 and their fibrils are respectively found to be more stable in 5-Fold and G-Key. E46K accelerates fibrillation^41,42 and is comparably stable in both morphologies. From this comparison, it appears that subtle changes in α-Syn sequence have more profound effects on the early stage of aggregation of oligomers rather than on the final fibril conformation. Previous experimental studies suggested that long-range intramolecular interactions within the α-Syn monomer could stabilize those conformations that inhibit aggregation.⁹⁸ Familial mutations are likely to enhance or disrupt such tertiary interactions, leading to conformations with opposite aggregation tendencies. In this respect, the introduction of the more hydrophobic residue Q/T may accelerate the initial assembly of α-Syn monomers. The introduction of a negatively charged residue D/E may slow α-Syn aggregation due to intermolecular electrostatic repulsion, resulting in a high energy barrier that would need to be surmounted to form amyloids. The switch of charge from E to K in the E46K mutant has an opposite effect, which may contribute to its fast aggregation. The Pro residue is known to preferentially display β-turn rather than α-helix or β-sheet in soluble proteins.⁹⁹ The conformational constraints induced by A30P in the N-terminal may retard formation of oligomers or conversion of oligomers to fibrils.

In summary, here we provide molecular details of the effect of familial mutations on the structure and stability of α-Syn fibrils. Two distinct morphologies of WT α-Syn fibrils display comparable size, solvent exposure, secondary structure contents, and ligand-binding mode, but different stabilities. Familial mutants shift the relative stabilities (and relative populations) of these fibrils without significantly altering their overall structures. All mutations are in the N-terminal region, but their influence on fibril structures is intrinsically long-ranged. In particular, A30P, H50Q, and G51D fibrils exhibit a calculate preference for the 5-Fold conformation, whereas WT, A53E, and A53T prefer G-Key and E46K fibrils show comparable stability in both conformations. Formation of oligomers appears more important in modulating aggregation kinetics of α-Syn mutants, resulting in divergent effects on the conformational preference of the fibrils. Nevertheless, recognition of the fibril by small molecule/peptide drugs may share a common binding region and appears to be conformation-independent.

METHODS

Based on available ssNMR data,^72,73 we constructed two α-Syn fibril structures with each fibril structure consisting of five monomers. The full-length α-Syn fibril that has been determined by ssNMR structure was used to build the G-Key model.⁷³ The five β-strand model was derived from ssNMR data in our previous study.⁷⁵ Approximately 20 N-terminal residues and 30 C-terminal residues of α-Syn fibril are highly unstructured, and the core region of α-Syn fibril comprises ∼70 residues,^{72,88,100–102} and so our fibril models are composed of residues 20–110. The six familial mutants (A30P, E46K, H50Q, G51D, A53E, and A53T) were constructed by replacement of A30, E46, H50, G51, and A53 with Pro, Glu, Gln, Asp, and Glu/Thr, respectively.

MD simulations were performed at constant pressure–temperature ensemble (NPT) using the NAMD2 program.¹⁰³ Each fibril structure was represented by the CHARMM27 force field with CMAP corrections.^104,105 TIP3P water molecules were used to solvate each fibril structure in a cubic water box.¹⁰⁶ The minimum distance between protein and the edge of water box was at least 12 Å. Each system was neutralized by adding counterions. The Langevin piston method and Langevin dynamics were applied to control the pressure at 1 atm and the temperature at 310 K.¹⁰⁷ The long-range electrostatics were treated using the Particle Mesh Ewald (PME) method,¹⁰⁸ and the van der Waals interactions were calculated using a switching function with a twin cutoff of 10 and 12 Å. The SHAKE algorithm was applied to constrain the hydrogen bonds.¹⁰⁹ An integration time step of 2 fs was used and structures were saved every 5000 steps (10 ps). Two WT fibril structures (G-Key and 5-Fold) were first energy minimized for 10 000 steps, and then heated to 310 K gradually. The two structures were equilibrated for 2 ns before running production dynamics at 1 atm and 310 K. MD simulations of structures in G-Key and 5-Fold were carried out for 50 and 100 ns, respectively. The conformations of WT at 50 or 100 ns were taken as the starting conformations for MD simulations of mutants. The production runs of MD simulations of WT and mutants in either G-Key or 5-Fold were performed for 200 ns each. The time evolution of the radius of gyration was monitored and shown in Figure S1. Extending MD simulations of wild types to 400 ns gives comparable results (Figure S2 and Table S1), indicating that both systems seem to reach an equilibrium. The cumulative average of the percentage of β-strand was also calculated and shown in Figure S3. The changes of the radius of gyration and the content of β-strand suggest that each system could reach an equilibrium within 200 ns MD simulations. Consequently, the last 50 ns trajectory of each system was used for analysis.

The relative stability of each fibril model was evaluated in terms of the conformational energy, which includes molecular mechanics energy (MM, sum of bonded energy terms), nonbonded energy terms (electrostatics and van der Waals), and solvation energy calculated by the generalized Born model using the molecular volume (GBMV) method implemented in CHARMM program (c40b2).^104,110 To obtain the bonded and nonbonded energy contribution, each fibril structure was first minimized for 200 steps using the steepest decedent method. The standard parameters in the GBMV II algorithm were applied to obtain the solvent free energy of the fibril structures after energy minimization. The average values and standard deviations in all data analysis were obtained using the block average method, i.e., the last 50 ns trajectory was divided into two parts and an average value was calculated for each part; the final average value and standard deviation were obtained from the average values of two parts. It should be noted that the current GBMV method uses implicit continuum solvation models to calculate protein–solvent interactions. Emerging results suggest that H-bonds are extraordinarily stable in amyloid fibrils dues to hyperpolarization and hypercooperativity effects,¹¹¹ and future models should include more explicit treatment of water-mediated H-bonds for more precise calculation of conformational stabilities.