Co-aggregation of α-Synuclein With Amyloid-β Stabilizes β-Sheet-rich Oligomers and Enhances the Formation of β-Barrels

Abstract

Alzheimer's disease (AD) and Parkinson's disease (PD) are the two most common neurodegenerative diseases with markedly different pathological features of β-amyloid (Aβ) plaques and α-synuclein (αS) Lewy bodies (LBs), respectively. However, clinical overlaps in symptoms and pathologies between AD and PD are commonly observed caused by the cross-interaction between Aβ and αS. To uncover the molecular mechanisms behind their overlapping symptoms and pathologies, we computationally investigated the impact of αS on an Aβ monomer and dimerization using atomistic discrete molecular dynamics simulations (DMD). Our results revealed that αS could directly interact with Aβ monomers and dimers, thus forming β-sheet-rich oligomers, including potentially toxic β-barrel intermediates. The binding hotspot involved the second half of the N-terminal domain and NAC region in αS, along with residues 10-21 and 31-42 in Aβ. In their hetero-complex, the binding hotspot primarily assumed a β-sheet core buried inside, which was dynamically shielded by the highly charged, amyloid-resistant C-terminus of αS. Because the amyloid prion region was the same as the binding hotspot being buried, their fibrillization may be delayed, causing the toxic oligomers to increase. This study sheds light on the intricate relationship between Aβ and αS and provides insights into the overlapping pathology of AD and PD.

Graphical Abstract

The co-aggregation of Aβ and α-synuclein formed β-sheet-rich oligomers, including potentially toxic β-barrel intermediates. The C-terminus of α-synuclein dynamically capped the β-sheet core, which might delay Aβ fibrillization and enhance the population of toxic oligomers.

Introduction

Alzheimer's disease (AD) and Parkinson's disease (PD) are the two most common neurodegenerative diseases with markedly different pathological features of β-amyloid (Aβ) plaques¹ and α-synuclein (αS) Lewy bodies (LBs)², respectively. The AD pathology affects the cerebral cortex and hippocampus³, while the PD pathology mainly affects the substantia nigra⁴. However, overlaps of symptoms and pathologies in AD and PD patients have widely been clinically observed. More than 50% of AD cases exhibit significant PD pathology, which is known as dementia with Lewy bodies (DLB),^{5, 6} alongside Aβ plaques^{7, 8}. Patients with LBs also frequently exhibit AD pathology, which is denoted as PD diagnosed with dementia (PDD)⁹. In addition, the presence of co-pathology in neurodegenerative brains often results in a more rapid decline in cognition and motor performance¹⁰. PDD patients pathologically featuring Aβ plaques have more severe dysfunctions than typical AD patients^{8, 11}. The observations of αS aggregates in AD patients and Aβ plaques in LBs as well as studies using a transgenic mouse model¹² suggest a possible cross-interaction between Aβ and αS. Although Aβ and αS are primarily extracellular and intracellular, respectively, mounting evidence indicates that Aβ can also accumulate intraneurons¹³ and αS can be excreted from cells¹⁴. For example, the NAC (non-amyloid-β component) fragment of αS has been identified in the plaques of AD patients, causing the DLB disease^{11, 15}. Understanding the cross-interaction and co-aggregation between Aβ and αS, therefore, may help uncover the molecular mechanisms in the overlapping symptoms and pathologies of AD and PD.

Both Aβ and αS are intrinsically disordered amyloidogenic proteins. Cleaved off from the amyloid precursor protein by β- and γ-secretases¹, Aβ has two major isoforms of Aβ40 and Aβ42 in senile plaques, where Aβ42 is more amyloidogenic and cytotoxic¹. The 140-residue αS is composed of three distinct regions: a positively charged N-terminal region (residues 1‒60), hydrophobic non-amyloid-β component (NAC) region (residues 61‒95), and negatively charged C-terminal region (residues 96‒140)². The amyloid aggregation kinetics of both Aβ and αS feature a common sigmoidal curve with three phases corresponding to the nucleation of monomers into oligomers and proto-fibrils, followed by the elongation and saturation of proto-fibrils into mature fibrils with cross-β spine cores¹⁶. Mounting experimental evidence reveals that soluble oligomers, particularly those rich in β-sheets and formed during the early aggregation stage, are more cytotoxic than mature fibrils^{16, 17}. Numerous computational and experimental studies have suggested that β-barrel intermediates might constitute the cytotoxic oligomers of amyloidosis^{16, 18-22}.

The effects of Aβ and αS on each other’s aggregations have been extensively studied in vitro^23-25. Fibrils of Aβ and αS act as seeds and could promote each other’s amyloid aggregation^{23, 24}. The seeding effects of αS fibrils are even more potent than the Aβ fibrils in the aggregation of Aβ²³. Oligomerizations, rather than the fibrillization, of Aβ were enhanced in the presence of soluble αS species^{23, 25}. The co-aggregation of αS monomers and oligomers with Aβ promotes Aβ oligomerization and stabilizes preformed Aβ oligomers in vitro^{23, 25}. Intra-cerebral injections of soluble αS into AD mice increase the population of soluble oligomers and decrease the Aβ fibrillar plaques,²⁶ supporting the idea that soluble αS species promote the oligomerization and delay fibrillization of Aβ in vivo^{25, 27}. Clinically, patients with AD and PD co-pathologies display a higher level of soluble oligomers and a lower level of amyloid plaque load in the brain than with AD alone²⁸. The observation that PDD patients often experience more severe symptoms than typical AD patients^{8, 11} is, therefore, consistent with the finding that soluble oligomers are usually much more toxic than mature fibrils^{16, 17}. However, oligomers that emerge during the initial stages of aggregation are typically diverse and unstable, posing significant experimental challenges to their isolation, quantification, and structural elucidation²⁹.

To understand the pathological cascade of overlapping AD and PD at the molecular level, we computationally investigated the binding between Aβ and αS monomers and also the effect of αS on Aβ dimerization, capturing the early aggregation in silico. We used atomistic discrete molecular dynamic (DMD)³⁰ simulations with an implicit solvent model, a rapid and predictive molecular dynamics algorithm widely used to study protein folding and amyloid aggregation^{30, 31}. Our results revealed that αS could directly interact with Aβ monomers and dimers, forming β-sheet-rich oligomers, including potentially toxic β-barrel intermediates^{16, 17, 22, 32}. The binding hotspot involved the second half of the N-terminal domain and NAC region in αS, along with residues 10-21 and 31-42 in Aβ. Interestingly, these cross-interaction hotspot regions overlapped with the amyloidogenic core regions of αS^{31, 33} and Aβ¹⁶. In their hetero-complex, the binding hotspot primarily assumed a β-sheet core buried inside, which was dynamically wrapped by the highly charged and amyloid-resistant C-terminus ^{31, 34} of αS. Because the amyloid prion region was buried inside, their fibrillization may be delayed, causing the toxic oligomers to increase. Our findings are consistent with prior observations of soluble αS, enhancing oligomerization but suppressing fibrillization of Aβ^{23, 25}, and also support clinical observations of AD and PD co-pathologies, showing higher soluble oligomers but lower amyloid plaque load compared to AD alone²⁸. Overall, our study sheds light on the intricate relationship between Aβ and αS and provides insights into the overlapping pathology of AD and PD.

Materials and methods

Molecular system

The amino acid sequences of full-length αS and Aβ used in our simulation are shown in Figs. S1. The initial structure of the full-length structure of αS (PDB: 1xq8) used in our simulation is determined by solution NMR spectroscopy measurement in the micelle-bound form (Fig. S1b)³⁵. The initial structure of Aβ (PDB: 1zoq) was taken from the NMR structures solved in an aqueous solution (Fig. S1d)³⁶. Molecular systems, including one αS monomer mixed with one and two isolated Aβ monomers, were simulated to investigate the conformational effects of αS on the monomeric and dimeric structures of Aβ. For each molecular system, 30 independent DMD simulations were performed starting from different initial states (i.e., coordinates, orientations, and velocities). The minimum inter-peptide distance in each initial structure was more than 1.5 nm. The duration time of every single trajectory was up to 500 and 600 ns for the αS mixed with one and two Aβ systems, respectively. Three additional control molecular systems, including pure one αS, one Aβ, and two isolated Aβ, were performed with 30 independent DMD simulations. For each one- and two-peptide system, we performed 500 and 600 ns independent simulations.

Discrete molecular dynamics simulations

All simulations were performed utilizing the DMD algorithm³⁷ with the Medusa force field at 300K. The all-atom DMD force field has been benchmarked for the accurate prediction of protein stability change upon mutation and protein–ligand binding affinity^{38, 39}. With significantly enhanced sampling efficiency, DMD has been widely used by our group to study protein folding and aggregation^{16, 18, 31, 40-42} and by others^43-46. A comprehensive description of the atomistic DMD algorithm can be found in previous studies^{30, 37, 47}. Similar to the standard MD, both bonded interactions (i.e., covalent bonds, bond angles, and dihedrals) and nonbonded interactions (i.e., van der Waals, solvation, hydrogen bonds, and electrostatic terms) were explicitly considered in the DMD. The major difference between the DMD and standard MD approaches is in the form of potential interaction functions. Inter-atomic interactions in DMD are approximated by step functions instead of continuous potential functions in classical MD. The system’s dynamics are dictated by a series of collision events at which two atoms meet at an energy step and change their velocities according to conservation laws. Hydrogen bond formation is explicitly modeled using a reaction-like algorithm³⁰. To reduce the computational cost, the implicit solvent model of the effective energy function (EEF1) for proteins in the solution proposed by Lazaridis and Karplus is used to model solvation effects⁴⁸. The screened electrostatic interactions were calculated using the Debye–Hückel approximation with the Debye length set to 10 Å, corresponding to ~100 mM NaCl. DMD software is available at Molecules In Action (www.moleculesinaction.com). The units of mass, time, length, and energy used in our united-atom with implicit water model were 1 Da, ~50 fs, 1 Å, and 1 kcal/mol, respectively.

The DMD simulations of the Medusa force field with the EEF1 implicit solvation model have demonstrated predictive power in capturing native states in protein folding³⁰, producing consistent conformational ensembles that match experimental data from single-molecule FRET measurements in multi-domain protein dynamics^{49, 50}. The accuracy of the Medusa force field with the EEF1 implicit solvation model has also been benchmarked in aggregating functional amyloid suckerin peptides⁵¹ and pathological hIAPP peptides⁵² using standard MD simulations with an explicit solvent model in our previous studies. Moreover, our recent simulations of amyloid aggregation, including calcitonin peptides (hCT, sCT, phCT, and TL-hCT)⁵³ and amylin peptides (hIAPP, hIAPP(S20G), and rIAPP)⁵², successfully replicated experimentally observed variations in amyloid tendencies and secondary structures. DMD simulations employing the Medusa force field protocol have consistently demonstrated their predictive capabilities in our studies of the individual aggregation of Aβ¹⁶ and αS³¹. Thus, we chose to employ DMD with the Medusa force field to investigate the co-aggregation of Aβ and αS in this study.

Analysis methods:

The secondary structure was determined by applying the DSSP program⁵⁴. One hydrogen bond was considered to be formed when the N⋯O distance was within 3.5 Å, and the N─H⋯O angle was more than 120°. According to prior protein folding studies^{45, 55}, a pairwise-residue contact was defined as the distance between the heavy atoms from the main chain or side chain of two non-sequential residues within 0.65 nm. A two-dimensional (2D) free-energy surface was constructed using –RTlnP(x,y), where P(x,y) is the probability of a conformation with particular parameter values of x and y. The solvent accessible surface area is calculated using the rolling ball algorithm developed by Shrake and Rupley⁵⁶. If the β-strand segments of an oligomer could form a closed cycle with every β-strand connected by two β-strand neighbors through at least two hydrogen bonds, this oligomer was treated as a β-barrel oligomer^{16, 18, 40, 41}.

Results and discussion

One Aβ and one αS monomer readily co-aggregated into a β-sheet-rich heterodimer

To gain insights into the pathological cascade of overlapping AD and PD resulting from the co-aggregation between Aβ and αS, we initially conducted 30 independent 500 ns DMD simulations involving interactions between one Aβ peptide and one αS peptide. The dynamics of hetero-dimerization were tracked by observing the time-dependent changes in the secondary structure of each residue and by monitoring the count of inter-peptide hydrogen bonds and contacts, along with representative snapshots (Fig. 1a-c). The analysis of co-aggregation dynamics for three randomly selected trajectories from the pool of 30 independent DMD simulations indicated a preference for the isolated Aβ and αS peptides to bind to each other, spontaneously forming a complex rich in β-sheet structures. Within these hetero-dimers, Aβ predominantly adopted β-hairpin conformations, often enveloped by αS, resulting in the establishment of inter-peptide β-sheets held together by a network of extensive backbone hydrogen bonds. Although β-sheet structures were observed in both Aβ and αS prior to the formation of inter-peptide hydrogen bonds and contacts, these β-sheets exhibited dynamic behavior with frequent conformational fluctuations. Once Aβ and αS co-aggregated into a hetero-dimer with a certain number of inter-peptide contacts and hydrogen bonds, the stability of their β-sheets was notably enhanced.

Fig. 1. The hetero-dimerization dynamics analysis of the Aβ and αS monomer.

The hetero-dimerization dynamics of one Aβ and one αS monomer are monitored by observing the time evolution of the secondary structure of each residue; assessing the number of inter-peptide backbone hydrogen bonds (depicted in red) and heavy atom contacts (depicted in grey); and capturing snapshots at 300, 400, and 500 ns (a-c). Three trajectories are randomly chosen from the pool of 30 independent DMD simulations. The average propensity of each residue from αS and Aβ peptides to assume unstructured, β-sheet, and helix formations is shown (d). All the conformations from the last 100 ns of the 30 independent 500 ns DMD simulations are utilized for the secondary structure analysis. For clarity, the N-terminal, NAC, C-terminal regions, and Aβ peptide are color-coded as purple, yellow, red, and green, respectively.

An examination of three out of the 30 independent simulations indicated that our simulation system reached a steady state in the last 100 ns, showing no significant changes in terms of secondary structure contents and the number of intermolecular hydrogen bonds and contacts (Fig. 1a-c). Hence, we used the last 100 ns trajectories of all 30 independent 500-ns DMD simulations to compute the equilibrium secondary structure propensities of each residue in the hetero-dimer (Fig. 1d). For the αS in the hetero-dimer, the first half of the N-terminal domain (residues 8-32) was mainly helical, and the second half of the N-terminal domain (residues 35–56) and the NAC region (residues 61–95) preferred to form β-sheets, while the negatively charged C-terminus (residues 96-140) was mostly unstructured with some short helixes (e.g., residues 96-105). The consistent secondary structure preferences of αS have also been documented in previous computational and experimental investigations. For instance, the crystal structure of αS1–72 segment fused to the maltose-binding protein featured two helices around residues 1–13 and 20–34⁵⁷. An NMR spectroscopy study also suggested that the N-terminal residues 6–37 populated with helical formations in the full-length αS⁵⁸. Previous studies showed that residues from the second half of the N-terminal domain and NAC region had strong β-sheet tendencies^{42, 59}. In experimentally characterized αS fibril models, the in-register parallel β-sheet core was primarily composed of residues 38–95, aligning with our observed β-sheet populated region^60-62. The C-terminus of αS was experimentally shown to be very flexible and to lack a stable structured conformation⁶³. Furthermore, similar secondary structure tendencies were observed in our previous study of αS monomer and dimer ³¹. Aβ in hetero-dimer predominantly adopted β-sheet conformations with β-strand mainly formed by residues 10-20 and 30-42. These same residues were also present as β-strands in Aβ oligomers and fibrils^{16, 47, 64}. Overall, our simulation results suggested that αS and Aβ monomers readily interacted with each other, assembling into β-sheet-rich hetero-complexes.

The binding hotspots between Aβ and αS involved residues 1-18, 32-60, and 61-80 in αS, and residues 10-21 and 31-42 in Aβ

To identify the critical interactions driving the co-aggregation between Aβ and αS, we proceeded to characterize the residue-wise intra-peptide interactions (Fig. S2) and the inter-peptide interactions (Figs. 2a and S3-S5) that stabilized the αS and Aβ hetero-dimer. The intra-peptide contact frequency map of αS within the hetero-dimer (Fig. S3a) revealed a prominent helical contact pattern along the diagonal in the first half of the N-terminal domain (e.g., corresponding structural elements shown in Fig. S3a), in agreement with the secondary structure analysis (Fig. 1d) and previous studies^{42, 58, 59}. In the second half of the N-terminal domain, a distinctive β-hairpin contact pattern perpendicular and proximal to the diagonal was established between the β-strands of residues 35–44 and 47–56. Intriguingly, a comparable extended β-hairpin structure formed by residues 37–54, with β-strands encompassing residues 38–43 and 48–53, and stabilized by β-wrap AS69 proteins was also demonstrated in a previous experimental study⁵⁹. Several overlapping β-hairpin contact patterns were observed around the NAC region, also known as the amyloidogenic core of αS^{42, 65}. Additionally, a β-sheet interaction was observed between the second half of the N-terminal domain and NAC, implying its potential involvement in αS fibrillization along with the NAC region. Numerous previous studies have emphasized the substantial role of the second half of the N-terminal domain in αS aggregation^{33, 65, 66}. For example, a prior experimental study revealed that a fragment out of the second half of the N-terminal domain could promote the fibrillization of full-length αS⁶⁵. C-terminal residues featured several short helical contact patterns. The presence of dynamic short helices within the C-terminal region agreed with the previously observed NMR Cα chemical shift measurements⁵⁸. Some weak β-sheet contact patterns of the C-terminus pairing with the second half of the N-terminal domain were observed. For Aβ in the hetero-dimer (Fig. S3b), the high-frequency contact patterns and their corresponding structured motifs in the hetero-dimer suggested that Aβ peptide predominantly adopted an extended β-hairpin structure between residues 10-21 and 31-42. Two short β-hairpins with weaker probabilities were observed at both termini for residues 1-17 and 28-42. By comparing with experimentally determined J-coupling measurements, earlier MD simulations using the OPLS force field indicated the presence of β-hairpin-rich conformations within the structural ensemble of Aβ^{67, 68}. Such secondary structure propensities of Aβ sequences in forming β-hairpins during aggregation agreed with previous experimental and computational studies^{16, 69, 70}.

Fig. 2. Identification of the hotspots for interpeptide interactions between Aβ and αS monomers.

The residue-pairwise intermolecular contact frequency map between Aβ and αS (a). The average number of inter-molecular contacts of residues from Aβ and αS is calculated by integrating the corresponding pairwise contact frequency map. The radial distribution function (RDF) of Cα atoms from each region of the αS and Aβ corresponding to the geometry center (b). Representative structures with strong residue-pairwise contact frequencies in the Aβ-αS hetero-dimers (c).

The inter-peptide contact frequency map between Aβ and αS (Figs. 2a and S3-S5) indicated that the hetero-dimer Aβ residues 10-20 and 31-42 had a strong tendency to interact with the N-terminal and NAC domains of αS, forming parallel and anti-parallel inter-molecular β-sheets (Figs. S3 and S4). The C-terminus of αS was able to form relatively weak β-sheets with Aβ residues 10-20 and 31-42 (Fig. S5). The binding hotspots for the cross-interaction between Aβ and αS were identified by computing the average total number of inter-peptide contacts per residue (Fig. 2a). The analysis revealed that residues 1-18, 32-60, and 61-80 in αS, along with residues 10-21 and 31-42 in Aβ, acted as binding hotspots, exhibiting a notably higher propensity for inter-peptide contact compared to other residues. The αS segment of residues 9-19 consistently displayed isolated monomers rather than aggregated β-sheet structures in silico³³, eliminating or adding the 9-30 residues segment either enhanced or inhibited αS aggregation in vitro⁷¹, which suggests that residues 1-18 did not belong to the amyloidogenic regions and may even counteract αS aggregation. However, the second half of the N-terminal domain and the NAC domain of αS, and segments comprising residues 10-21 and 31-42 of Aβ, have previously been identified as amyloidogenic regions in studies on αS^{31, 33} and Aβ¹⁶.

The dynamic capping of the αS C-terminus onto the β-sheet core of binding hotspots may hinder Aβ fibrillization

The radius distribution functions of Cα atoms from different regions of αS and from Aβ (Fig. 2b) along with representative snapshots (Fig. 2c) suggest that Aβ, the second half of the N-terminal domain and NAC of αS were buried inside the heterodimer by forming stable β-sheets. The C-terminus of αS was positioned at the peripherical of the complex, resulting from the weak and thus dynamic β-sheet pairing with Aβ residues (Fig. S5) or with the other parts of αS (e.g., the second half of the N-terminal, as shown in Fig. S2a) in the core. The change in solvent accessible surface area (SASA) for both Aβ and αS upon forming the hetero-complexes was examined by calculating the ratio of SASA changes with respect to their isolated monomers (Fig. S6). The interaction between Aβ and αS significantly decreased the SASA of residues in the binding hotspots, including residues 1-18, 32-60, and 61-80 of αS and residues 10-21 and 31-42 of Aβ. The complexation of Aβ and αS did not significantly affect the SASA of αS C-terminal residues (Fig. S6) because the αS C-terminal was on the surface of the heterodimer. Previous experimental studies have demonstrated that disrupting the interaction between the C-terminus and the N-terminus or NAC regions, either through calcium binding or C-terminus truncation, can expedite the aggregation of αS ^{34, 72, 73}, suggesting that the presence of the αS C-terminus around the amyloid prion regions may have a suppressive effect on amyloid aggregation. In our recent simulation study on homo-dimerization of αS, we demonstrated that the interaction of the C-terminus with the second half of the N-terminal domain and NAC regions of αS could prevent αS dimerization³¹. Based on our computational observation, the aggregation-delaying effect of the C-terminal might result from its dynamic capping of the cross-β core during fibrillization. Hence, residues with a high prior aggregation tendency of Aβ and αS mostly buried inside may hinder the fibrillization of Aβ (Figs. 2c and S6c).

The αS monomer exhibited the capability to envelop Aβ monomers and dimers forming β-sheet-rich hetero-aggregates:

To further investigate the impact of αS on Aβ oligomerization, we conducted 30 independent 600-ns DMD simulations by combining two Aβ peptides with one αS peptide. The inter-peptide distance in each independent initial structure was set to be greater than a minimum of 1.5 nm to mitigate potential biases from the initial state. The time evolution of the inter-peptide backbone hydrogen bonds of Aβ-Aβ and Aβ-αS, as well as snapshots, showed that the three peptides readily co-aggregated into β-sheet-rich hetero-trimers (Fig. 3a-c). Two out of the three peptides initially formed either an Aβ-αS hetero-dimer (Figs. 3a and b) or an Aβ-Aβ homo-dimer (Fig. 3c), followed by the incorporation of the remaining peptide. This sequential assembly eventually led to the creation of stable hetero-trimers, characterized by the prevalence of inter-peptide β-sheet structures (Fig. 3a-c). These results indicated that αS could directly interact with both Aβ monomers and dimers. The initial formation of Aβ-αS hetero-dimer could be accompanied by an increase in Aβ-αS hydrogen bonds, forming an Aβ_M-αS_M-Aβ_M hetero-trimer (Fig. 3a), or Aβ-Aβ hydrogen bonds, forming an Aβ_D-αS_M hetero-trimer (Fig. 3b). The former had more Aβ-αS hydrogen bonds, while the latter featured more Aβ-Aβ hydrogen bonds. To gain a deeper insight into the conformational characteristics of the hetero-complexes, we conducted additional calculations of the potential mean force (PMF) based on the number of Aβ-αS inter-peptide contacts plotted against the number of Aβ-Aβ inter-peptide contacts or hydrogen bonds (Fig. 3d). Only the equilibrated simulation data of the last 100 ns from all 30 independent DMD trajectories were used for the conformational PMF analysis. The energetically favorable conformations of the hetero-trimers were primarily marked by the prevalence of Aβ-αS inter-peptide contacts, accompanied by distinct Aβ-Aβ inter-peptide contacts and hydrogen bond arrangements. Specifically, we observed low free energy conformations for both individual Aβ units and their association with αS. These conformations exhibited differing counts of Aβ-Aβ inter-peptide contacts and hydrogen bonds, with one scenario having limited interactions and the other displaying a higher number. Overall, our results suggest that soluble αS monomers can envelop both Aβ monomers and dimers, forming β-sheet-rich hetero-aggregates.

Fig. 3. The co-aggregation dynamics and conformational free energy landscape analysis of two Aβ mixed with one αS.

The time evolution of the number of inter-molecular hydrogen bonds between two Aβ monomers (red) and between Aβ and αS (gray) forming Aβ_M-αS_M-Aβ_M (a) and Aβ_D-αS_M (b and c) structures. Representative structures as shown at every 100 ns during the last 300 ns. The potential mean forces as a function of the number of inter-chain Aβ-Aβ contacts and hydrogen bonds against αS-Aβ contacts are calculated using the last 100 ns conformations from 30 independent DMD simulations (d). Three representative structures featuring low free energy (labeled as 1-3 on the free energy landscape surface) corresponding to three different local minimum states in both PMFs are also presented.

The dynamic interaction of the αS C-terminus with the β-sheet core shielded the exposed surface of amyloidogenic regions, resulting in a delay of further aggregation

The average secondary structure tendency of each residue within the hetero-trimer (composed of two Aβ and one αS, Figs. 4a and b) closely resembled that observed in the hetero-dimer (consisting of one Aβ and one αS, Fig. 1d). Notably, the β-sheet structures remained prominent in the second half of the N-terminus and NAC regions in αS, as well as in residues 10-20 and 30-42 of Aβ within the hetero-trimer (Figs. 4a and b). Various intra- and inter-peptide β-sheet patterns involving these regions were evident in the residue-pairwise contact frequency map (Fig. S7). Helix structures were mainly present in residues 8-32, encompassing the first half of the N-terminal domain of αS. Although the αS C-terminus still primarily adopted unstructured conformations with faint indications of short helix segments, weak β-sheet conformations resulted from its capping to the second half of the N-terminal and NAC regions of αS, and in residues 10-20 and 30-42 of Aβ within the hetero-trimer (Figs. 4b and S7). Similar to their hetero-dimer (Fig. 2b), the radius distribution functions of Cα atoms from different regions of αS and from Aβ within their hetero-trimer showed that the β-sheet populated regions were buried inside. Moreover, the highly polar and charged αS C-terminus was more likely to be located on the outer layer (Fig. 4c).

Fig. 4. The structural analysis for the hetero-trimer aggregated by one αS and two Aβ.

The average propensity of each residue adopted each type of secondary structure (including a random coil, bend, β-sheet, and helix) of αS (a) and Aβ (b) in their co-aggregates. The radial distribution function (RDF) of Cα atoms from the different regions of αS and Aβ in their hetero-trimer c). Heterotrimers from the last 100 ns of 30 independent DMD trajectories are used for the structural analysis.

The average number of inter-peptide Aβ-αS contacts for each residue within the hetero-trimer indicated that these β-sheet populated regions (including the second half of the N-terminal and NAC regions of αS, and residues 10-20 and 30-42 of Aβ) exhibited larger contact numbers than the other regions (Figs. 5a and c). This suggested that these β-sheet populated regions served as the binding hotspots between Aβ and αS in their hetero-trimer, which was observed in their hetero-dimer (Fig. 2a). When comparing the secondary structure propensity of each residue in αS between the systems of pure αS monomer and αS mixed with one or two Aβ peptides, it was found that binding with either an Aβ monomer or dimer slightly enhanced only the β-sheet propensity of αS around its Aβ-binding hotspot regions but did not affect the secondary structure propensities of other regions (Fig. S8). The difference ratio of per residue SASA of αS in the presence and absence of two Aβ peptides showed that their co-aggregation significantly decreased the exposed surface area of residues around binding hotspots but did not induce significant changes in the highly polar and charged αS C-terminus. Because the binding hotspots between Aβ and αS overlapped with the amyloidogenic core regions of both Aβ¹⁶ and αS^{31, 33}, which were mostly buried by the C-terminus, the exposed surface area of their hetero-aggregates was highly charged (Figs. 5 and S7). Interestingly, a similar 'negative design' strategy involving the strategic placement of charged residues at the peripheries of β-sheets has found application in the development of amyloid inhibitors^{74, 75}. Previous studies have demonstrated that burying the amyloidogenic surfaces of Aβ using charged peptide segments could suppress Aβ aggregation^{70, 76}. Hence, the dynamic interaction of the highly charged C-terminus with β-sheet core regions may play a role in delaying further aggregation and fibrillization of Aβ^{23, 25}. This result horts the experimental observation that intracerebral injections of soluble αS into AD mice resulted in a higher population of soluble oligomers and fewer Aβ fibrillar plaques ²⁶.

Fig. 5. The inter-peptide contact and exposed surface area analysis for the hetero-aggregates formed by two Aβ and one αS.

The average number of inter-chain contacts per residue formed by heavy atoms between Aβ and αS in their heterotrimer (a and c). Average difference ratios of accessible surface area per residue of two Aβ peptides in the presence and absence of αS (b). The difference ratio of accessible surface area per residue of αS in the presence and absence of two Aβ peptides (d). The data used in the above structural analysis include the last 100 ns from 30 independent 600 ns DMD trajectories for two Aβ mixed with one αS and the last 100 ns from 30 independent 500 ns DMD trajectories for pure two Aβ and pure one αS.

The interaction between αS and Aβ promoted the formation of β-barrel oligomers

The β-barrel oligomers, first structurally characterized in the aggregation of an αB-crystalline fragment, have been postulated as potentially cytotoxic oligomers of amyloidosis¹⁹. The formation of β-barrel intermediates during aggregation has been supported experimentally and computationally for both toxic fragments and full-length amyloid peptides (e.g., hIAPP and Aβ)^{16, 18, 20, 22, 52}. For example, the population of β-barrel intermediates in wild-type Aβ was higher than that of the AD-protective A2T substitution but lower than the AD-causative mutations of D7N and E22G¹⁶. The S20G substitution in hIAPP increased its amyloidogenicity and cytotoxicity and significantly boosted the formation of β-barrels when compared to the wild-type peptide^{18, 52}. In our simulations, the isolated αS monomer exhibited a low level of β-barrel formation (Fig. 6), while Aβ dimers did not form β-barrels. However, the hetero-complexes displayed significantly higher occurrences of β-barrel intermediate structures than those of αS and Aβ alone (Fig. 6). Representative snapshots of β-barrels formed by one αS and two Aβ showed that the second half of the αS N-terminal domain and NAC combined with Aβ peptides form a closed single β-sheet cylinder surrounded by the helical head of the N-terminus and unstructured C-terminus of αS (Fig. 6b). Hence, the interaction between αS and Aβ promoted the formation of toxic β-barrel oligomers. This result is consistent with the clinical observations that patient brains with co-pathology of Aβ and αS displayed a more rapid decrease in cognition and motor performance¹⁰. Although the formation of β-barrel pores leading to membrane leakage, initiated from isolated monomers, was observed in atomistic MD simulations of short amyloid fragment peptides⁷⁷, it is essential to conduct future aggregation simulations involving full-length αS and Aβ, both individually and in combination within a membrane environment, to fully elucidate the broader roles of β-barrel oligomers in amyloid toxicity.

Fig. 6. The analysis of β-barrel formations.

The frequency of β-barrel oligomers observed in each simulation trajectory of the molecular system (including isolated αS and αS with one and two Aβ) is sorted in a descending order based on the probability of β-barrel formation (a). Only the top 15 trajectories are displayed. In addition, two representative β-barrel structures for each system are provided in side and top views (b-d). The structural diversity in β-barrel formations makes them unsuitable for the structural cluster analysis. Hence, two representative β-barrels are randomly selected from the two most populated β-barrel trajectories. Owing to the presence of β-barrel intermediates before reaching saturation stages, the entire trajectory is used in calculating the β-barrel propensity.

Conclusions

In summary, our computational study demonstrated that the interaction between αS and Aβ led to the formation of β-sheet-rich hetero-oligomers and enhanced the potential for toxic β-barrel intermediates. Within the hetero-complex, β-sheets were mainly formed around their binding hotspots, which included the second half of the N-terminal domain and NAC in αS, as well as residues 10-21 and 31-42 in Aβ. These binding hotspots, overlapping with the amyloidogenic core regions of both Aβ¹⁶ and αS^{31, 33}, formed a β-sheet core. The β-sheet core was buried within and encased by the highly charged C-terminus of αS, likely contributing to the delay in fibrillization and an increase in the population of toxic oligomeric species. Moreover, our computational results supported experimental observations indicating that the presence of soluble αS delayed Aβ fibrillization, facilitated the formation of soluble oligomers, and correlated with clinical findings of more severe symptoms in cases with combined Aβ and αS pathologies than in those with Aβ or αS pathologies alone.