Molecular structure of an N-terminal phosphorylated β-amyloid fibril

Significance

This work reports the molecular structure formed by a pathologically relevant posttranslational modified β-amyloid (Aβ), the Ser-8-phosphorylated Aβ₄₀ (pS8-Aβ₄₀). The N-terminal structures in pS8-Aβ₄₀ fibril differ significantly from all known wild-type Aβ₄₀ fibrils, with strong intra-strand interactions that make the N terminus associated closely with the amyloid core. The pS8-Aβ₄₀ fibril possesses strong cross-seeding ability to wild-type Aβ₄₀ monomers, while the propagated fibrillar structure shows higher thermodynamic stability and core rigidity compared to the fibrils formed by the self-nucleation of wild-type Aβ₄₀. Our finding may shed light on the origin of structural polymorphisms in Aβ amyloids.

Abstract

The structural polymorphism in β-amyloid (Aβ) plaques from Alzheimer disease (AD) has been recognized as an important pathological factor. Plaques from sporadic AD patients contain fibrillar deposits of various amyloid proteins/peptides, including posttranslational modified Aβ (PTM-Aβ) subtypes. Although many PTM-Aβs were shown to accelerate the fibrillation process, increase neuronal cytotoxicity of aggregates, or enhance the stability of fibrils, the contribution of PTM-Aβs to structural polymorphisms and their pathological roles remains unclear. We report here the NMR-based structure for the Ser-8-phosphorylated 40-residue Aβ (pS8-Aβ₄₀) fibrils, which shows significant difference to the wild-type fibrils, with higher cross-seeding efficiency and thermodynamic stability. Given these physicochemical properties, the structures originated from pS8-Aβ₄₀ fibrils may potentially dominate the polymorphisms in the mixture of wild-type and phosphorylated Aβ deposits. Our results imply that Aβ subtypes with “seeding-prone” properties may influence the polymorphisms of amyloid plaques through the cross-seeding process.

Recent structural analysis of the β-amyloid (Aβ) fibrils derived from plaques from Alzheimer disease (AD) patients shed light on the pathological significance of structural polymorphisms in amyloid deposits (1). Particularly, structural polymorphisms in 40-residue Aβ (Aβ₄₀) fibrils were shown to correlate with clinical symptoms in different AD cases and the lengths of disease progression (1, 2). The ex vivo brain-seeded fibrillation of Aβ elucidated that fibrils were produced when seeds were extracted from AD patients, but not from non-AD controls (3, 4). Meanwhile, exogenous injection of AD brain extracts was shown to induce the formation of plaques in transgenic mice models in vivo, with the inducibility correlated with the stability, the deposition pattern, and the composition of the extracts (5–7). These findings implied that certain fibrillar structures in the pathological plaques might have higher efficiency to amplify the amyloid deposition compared to those in the nonpathological plaques. Recently, it has been shown that the amyloid deposits form conformation-distinct “clouds” that could be targeted using specific fluorescence-based molecular probes (8). It is possible that certain plaques contain fibrillar aggregates with more seeding-prone polymorphisms, and therefore could more actively induce aggregation of other amyloid species.

A fundamental question that remains unsolved is whether such seeding-prone structures originated from the wild-type (wt) Aβ peptides. Although wt-Aβ₄₀ and wt-Aβ₄₂ are the most abundant components in plaques, the answer to this question is not necessarily straightforward because the molecular structures of Aβ fibrils could propagate through cross-seeding between different amyloid proteins (9–11). Fibrillar structures originally from other amyloidosis species with seeding-prone properties may dominate the polymorphisms after long-time evolution. Several types of posttranslational modified Aβ (PTM-Aβ) fibrils have been identified in pathological amyloid plaques (12). Immuno-blotting assays showed that a few critical PTM-Aβ subtypes, including the E3- and E11-pyroglutarmation (13, 14), the D7-isomerization (15), the S8-phosphorylation (13, 16), the S26-phosphorylation (17), and the Y10-nitration of Aβ (18), either concentrated in the plaque core or formed fibrils earlier than the wt-Aβ peptides. In vitro studies have also demonstrated that certain PTM-Aβs grew fibrils more rapidly than their wt-Aβ analogs and possessed higher levels of neuronal cytotoxicity (9, 18, 19). Certain PTM-Aβ subtypes were thought to trigger or accelerate the fibrillation of wt-Aβ peptides (20, 21). However, the structural basis of the pathological roles of PTM-Aβs has not been determined.

Results

The Molecular Structure of Ser-8-Phosphorylated Aβ₄₀ Fibril.

We previously demonstrated that the S8-phosphorylated Aβ₄₀ (pS8-Aβ₄₀), a PTM-Aβ identified in pathological AD plaques, possessed high cross-seeding efficiencies to multiple Aβ subtypes, including the wt-Aβ₄₀, wt-Aβ_42, and E3-pyroglutamated Aβ₄₀ (22). This work reports the molecular structure of pS8-Aβ₄₀ fibril (Fig. 1A) and investigates its potential to dominate structural polymorphisms in a mixed fibrillar system. The pS8-Aβ₄₀ fibrils were grown from synthetic peptides in physiological buffer (i.e., pH 7.4 and 37 °C) with continuous gentle agitation, followed by generation seeding to ensure structural homogeneity. Structural modeling was performed using Xplor-NIH (23) with experimental constraints from solid-state NMR (ssNMR) spectroscopy and transmission electron microscopy (TEM). The fibrils showed striated-ribbon morphology similar to the twofold wt-Aβ₄₀ (wt2-Aβ₄₀) fibrils grown under similar conditions (Fig. 1B and SI Appendix, Fig. S1) (24, 25). The molecular structure for pS8-Aβ₄₀ fibrils illustrated the width ∼8.5 nm across the interface perpendicular to the fibril axis, larger than the width ∼6.0 nm for wt2-Aβ₄₀ fibrils (25). Measurements of the mass-per-length (MPL) (26) indicated that the fibrils possessed twofold symmetry (a minimum MPL ∼19 kDa/nm and multiples, Fig. 1C). Three segments of β strands were identified at G9-V12, L17-E22, and I31-V39, connected by a shorter loop around H13-K16 and a longer loop around D23-A30. The N-terminal segment D1-D7 was disordered and showed broadened ssNMR peaks (Fig. 1 D and E and SI Appendix, Figs. S2–S4 and Table S2). Crucial long-range structural constraints were obtained using multiple types of ssNMR methods (SI Appendix, Figs. S5–S8 and Table S3). Interestingly, side-chain interactions between residues in G9-V12 and L17-E22 strands (Fig. 2A) were intramolecular, confirmed by 2D ¹³C-¹³C spin-diffusion spectroscopy with samples containing isotope-labeled and unlabeled pS8-Aβ₄₀ in a 1:1 molar ratio, while the interactions between L17-E22 and I31-V39 strands were intermolecular (Fig. 2B). Parallel-in-register β sheets from G9 to V39 were confirmed by ¹³C-PITHIRDs-CT experiments (27), while A2 showed longer interstrand distance ∼7.0 Å (Fig. 2C and SI Appendix, Fig. S9).

Fig. 1.

The molecular structure of pS8-Aβ₄₀ fibrils. (A) Structural models shown in surface mode (Left) that highlight two subunits in orange and blue, and stick mode (Right) that highlights the hydrophobic, polar, and charged residues in red, blue, and green, respectively. Side chains of the residues that form key hydrophobic contacts in cores are shown in red, and the phosphate groups at S8 are shown in orange. (B, Top) A representative negatively stained TEM image with parallel filaments highlighted using red arrows. (B, Bottom) Plots of the apparent fibril widths vs. number of filaments. Error bar reflects the SD from three measurements on the fibril widths (indicated by columns) for each number of parallel filaments. (C, Left) A representative dark-field TEM image for the MPL measurement with single and bundled filaments highlighted using yellow arrows. (C, Right) The histogram plot for the analysis of MPL with Gaussian fitting. The minimum 19.5 kDa/nm corresponds to the MPL of a single twofold Aβ filament. (D) Representative 2D ssNMR spectra with lines highlighting individual intraresidue cross peaks. (E) Plots of the residue-specific differences in the secondary chemical shifts of Cα and Cβ. Negative values indicate β strand conformation (shown by red arrows on top of the bars).

Fig. 2.

Structural restraints for pS8-Aβ₄₀ fibrils. (A and B) Representative 2D spin diffusion spectra to show interresidue cross peaks with samples that were fully isotope labeled (red, no dilution) and 50% isotope labeled (blue, 1:1 dilution). The cartoon model in B highlighted that the V12-F20 interaction was intrastrand (side chains shown in the same color) and the F19-L34 interaction was interstrand (side chains shown in different colors). (C, Bottom) Representative ¹³C-PITHIRDs-CT decay curves and simulated curves for different ¹³C-¹³C proximities. (C, Top) The cartoon model highlighting the full ¹³C-labeling sites for PITHIRDs measurements. Noting that A2 was not in parallel β sheet. (D) Representative ¹³C-³¹P REDOR original (S₀) and difference (ΔS) spectra to show the close contact between pS8 phosphate and V24/G25. (E, Top Left) ³¹P-PITHIRDs-CT spectra on pS8-Aβ₄₀ fibrils. (E, Bottom) Plots of ³¹P dephasing and simulated ³¹P decay curves for different ³¹P-³¹P proximities. (E, Top Right) The cartoon model highlighting the proximities between adjacent ³¹Ps (in blue) in pS8-Aβ₄₀ fibrils. (F) The 2D spin diffusion spectra on samples with and without dilution to highlight the side-chain contacts in the quaternary structural interface (Bottom cartoon models). In C and E, error bars reflect the spectral noise levels in corresponding PITHIRDs experiments and were obtained by integrating noises in individual spectra.

The pS8-Aβ₄₀ fibril showed distinct structural features compared to wt-Aβ fibrils (24, 25). Most strikingly, the N-terminal conformation was unique. The wt2-Aβ₄₀ fibrils showed highly dynamic N terminus, where residues D1-S8 were disordered and G9-Q15 were not involved in the fibrillar core (24). Disordered N-terminal segments were also reported for the threefold wt-Aβ₄₀ (wt3-Aβ₄₀) fibril, grown in vitro from synthetic peptides under quiescent incubation (25). The ex vivo brain-seeded Aβ₄₀ fibril possessed well-defined parallel-in-register β-sheet through the entire sequence; however, with intermolecular contacts between the N terminus of one strand and the E22-A30 loop of the adjacent strand (4). In contrast, the pS8-Aβ₄₀ fibril showed strong intrastrand interactions between the N terminus and the rest of amyloid core. For instance, close proximities were detected between phosphate group of pS8 and V24-G25 (Fig. 2D), and between the side chains of E3/F4 and V24/S26 (SI Appendix, Figs. S5 and S6), suggesting that the N-terminal residues might participate in the initial aggregation to form protofibrils. The ³¹P-³¹P interstrand distance was ∼6.5 Å (³¹P-PITHIRDs-CT, Fig. 2E), meaning that pS8 did not belong to the parallel-in-register β sheets (the H-bond network started from G9). All phosphate side chains on pS8 pointed toward the amyloid core, consistent with our previous finding that the fibrillar pS8-Aβ₄₀ aggregates were more resistant to calf intestinal alkaline phosphatase than the monomeric peptides (22). Residues D1-D7 were clearly more dynamic than the rest of pS8-Aβ₄₀ sequence; however, there still existed certain intermolecular structural restraints, e.g., the methyl groups of A2 adopted ∼7.0 Å interstrand distance (Fig. 2C). Furthermore, compared to the wt2-Aβ₄₀, the phosphorylated fibril (Fig. 2F) had hydrophobic side chain packing involving nine residues in its quaternary structural interface (i.e., I31 to V39), while the wild-type fibrils had only six residues (i.e., I31 to V36) (24, 25).

Quantitative Difference in Seeding Abilities Between pS8 and wt2-Aβ₄₀ Fibrils.

We next specifically compare the seeding abilities of the pS8-Aβ₄₀ and wt2-Aβ₄₀ fibrils, which were grown under similar conditions. The thioflavin-T (ThT) fluorescence assay (Fig. 3A and SI Appendix, Fig. S10) on fibrillation showed kinetics that fit to single-exponential increases without lag phase. Little fluorescence increase was observed in the absence of seeds within the same period (i.e., ∼10 h, SI Appendix, Fig. S11), confirming that self-nucleation from Aβ₄₀ monomers was eliminated under the present seeding conditions. Plot of the initial ThT build-up rates as a function of seed concentrations (Fig. 3B) concluded that the elongation rate constant for the seeded fibrillation with pS8-Aβ₄₀ was ∼5 times higher than wt2-Aβ₄₀. Noting that it was the wt2-Aβ₄₀ seeds rather than the pS8-Aβ₄₀ seeds that had the same primary sequence as monomeric peptides in the seeding system, it was clear that the seeding efficiencies were determined by molecular structures rather than the primary sequences. The ThT assays were also applied to the mixed pS8-Aβ₄₀ and wt2-Aβ₄₀ seeds. The results (Fig. 3 C and D and SI Appendix, Fig. S12) showed that with a constant total seed concentration and seed-to-monomer ratio, the initial ThT build-up rates increased linearly and the ThT emission intensities decreased linearly with the percentage of pS8-Aβ₄₀ seeds in the mixture. These results highlighted: (i) the relative seeding efficiencies of pS8-Aβ₄₀ and wt2-Aβ₄₀ fibrils retained quantitatively in mixtures; and (ii) distinct morphologies of individual seeds, characterized by different levels of ThT intensities (28), were propagated to the seeded fibrils.

Fig. 3.

ThT fluorescence kinetics assay of seeded fibrillation. (A) Representative ThT traces of seeded Aβ₄₀ fibrillation with 5 mol % wt2-Aβ₄₀ (black circles), 15 mol % wt2-Aβ₄₀ (black diamonds), 5 mol % pS8-Aβ₄₀ (red circles), and 15 mol % pS8-Aβ₄₀. (B) Plots of the initial ThT build-up rates (extracted from single-exponential fitting of curves in A) vs. seed concentrations. The slopes indicate the elongation rate constants. (C) Representative ThT traces of seeded Ab40 fibrillation with mixed wt2-Aβ₄₀ and pS8-Aβ₄₀ seeds. The mole fraction of pS8-Aβ₄₀ seeds were 1.0, 0.8, 0.5, 0.2, and 0.0 for curves in black, red, green, blue, and purple, respectively. (D) Plots of the initial ThT build-up rates (black, left vertical axis) and ThT emission at plateau (red, right vertical axis) vs. the mole fraction of pS8-Aβ₄₀ seeds. Error bars in A and C were drawn from five independent measurements for each condition. Error bars in B and D were derived from fitting of independent corresponding ThT kinetic traces, respectively.

The pS8-Aβ₄₀ Fibril Possesses Higher Stability and Core Rigidity.

We then asked which fibril possessed higher stability, as mixed fibrillar structures with different stabilities were shown to slowly convert to the more stable one (29). Denaturing guanidine hydrochloride (GndHCl) assay (Fig. 4A) showed that the pS8-Aβ₄₀ fibrils were more resistant to denaturing reagents, and therefore might possess higher thermodynamic stability. Dynamics at selected sites in amyloid cores, which indicated the core rigidity, were probed using ²H ssNMR spectroscopy. The temperature-dependence of ²H quadrupolar line shapes and longitudinal relaxation revealed mainly the activation energies for the side chain rotameric (L17, L34, M35, and V36) and ring-flipping (F19) motions (30, 31). Our results showed the ²H line shapes for L34, M35, and V36 (Fig. 4B) and the ²H relaxation for F19 (Fig. 4C) were distinct between pS8-Aβ₄₀ and wt2-Aβ₄₀ fibrils. Little difference in ²H line shapes was observed for L17 (SI Appendix, Fig. S13). Noting that L17 was located at the edge of β sheets in both fibrils (25, 32), it seemed that these two structures had more distinct dynamic features in the center of amyloid core. Quantitative fitting of both ²H line shapes and relaxation to motional models (SI Appendix, Figs. S14–S17 and Tables S4–S6) led to the conclusion that the activation energies for large-scale motions, including the methyl axes rotameric jumps in L34, M35, and V36 and the ring flips in F19 (33), were larger in pS8-Aβ₄₀ fibrils compared to wt2-Aβ₄₀ fibrils (Fig. 4D), suggesting a more rigid fibrillar core for the modified fibril and in parallel with the higher thermodynamic stability in the phosphorylated fibril shown by denaturing assays.

Fig. 4.

Stabilities of wt2-Aβ₄₀ and pS8-Aβ₄₀ fibrils. (A) Denaturing gel electrophoresis assay (Bottom, representative gel bands) to determine the stability of pS8-Aβ₄₀ (Top, black columns) and wt2-Aβ₄₀ (Top, red columns) fibrils. The average normalized intensities (shown in columns) were obtained from three independent repeats (in cross symbols). (B) Experimental ²H NMR static line shapes for L34 (Left), V36 (Middle), and M35 (Right) in the wt2-Aβ₄₀ (red) and pS8-Aβ₄₀ (black) fibrils. (C, Top) ²H relaxation times $T_{1Z}^{eff}$ vs. 1000/T on semilog scale obtained for ±20 kHz (squares) and ±60 kHz (circles) positions of the powder patterns for the F19 ring in wt2-Aβ₄₀ (red) and pS8-Aβ₄₀ (black) fibrils. (C, Bottom) ²H quadrupolar echo line shapes at 310 K for the two fibrils. (D) Activation energies for the large-scale motions (rotametic jumps for methyl groups and ring flips for aromatic groups) in pS8-Aβ₄₀ (black) and wt2-Aβ₄₀ (red) fibrils. Error bars in C and D are SEs of data fitting to SI Appendix, Eq. S1 and the Arrhenius Equation, respectively.

The Presence of pS8-Aβ₄₀ Seeds May Alter the Monomeric Aβ₄₀ Pathway in Seeding.

The combination of ThT fluorescence, GndHCl denaturing assay, and ²H NMR results led to the conclusion that the fibril structure formed by pS8-Aβ₄₀ was more seeding-prone compared to wt2-Aβ₄₀ because it induced more rapid cross-seeding to wt-Aβ₄₀ peptides and the propagated structures possessed higher stability and rigidity. Thus, the structure propagated from pS8-Aβ₄₀ seeds would dominate the polymorphism in Aβ deposits over time if only these two types of seeds existed, although the pS8-Aβ₄₀ might not be the predominant species. Molecular structures of pS8-Aβ₄₀ and wt2-Aβ₄₀ fibrils had the most significant difference at their N termini (25, 32). We showed using the tyrosine fluorescence-quenching assay that the presence of these different seeds potentially altered the folding pathway of N termini of wt-Aβ peptides. Quenching of the Y10 side chain fluorescence with water-soluble acrylamide was sensitive to its local environment (Fig. 5A) (34), and therefore the slopes of quenching showed measurable differences between wt-Aβ₄₀ peptides, wt2-Aβ₄₀ fibrils, and pS8-Aβ₄₀ fibrils. We then monitored the time-dependence of fluorescence quenching slopes in systems that contained the same concentration of wt-Aβ₄₀ monomers with 10 mol % either wt2-Aβ₄₀ or pS8-Aβ₄₀ seeds. As shown in Fig. 5 B and C, the slopes decreased from the characteristic values for wt-Aβ monomers to the values for fibrils in both cases in the time course of fibrillation (monitored using ThT assay under the same seeding condition). Interestingly, the time dependence of quenching slopes matched qualitatively the kinetics of seeded fibrillation; i.e., the Y10 side chain folded into a water-shielded conformation more rapidly in the presence of pS8-Aβ₄₀ seeds. Results from the fluorescence-quenching assay suggested that the more structurally ordered N terminus in pS8-Aβ₄₀ fibril would enhance its seeding efficiency by accelerating the formation of well-ordered N termini in the “docked” monomers.

Fig. 5.

Tyrosine fluorescence quenching assay on seeded Aβ₄₀ fibrillation. (A) Fluorescence quenching vs. acrylamide concentrations for Aβ₄₀ monomers (black), wt2-Aβ₄₀ fibrils (red), and pS8-Aβ₄₀ fibrils (blue). (B and C) Plots of fluorescence quenching slopes (purple, obtained from linear fitting in A, right vertical axes in both panels) vs. fibrillation time for pS8-Aβ₄₀ (B) and wt2-Aβ₄₀ (C) seeded fibrillation. Increments of ThT fluorescence within the same time scale were also shown in the same plot (blue and red, left vertical axes in both panels). Error bars in A indicates three independent measurements on Tyr10 fluorescence intensities under each condition. Error bars in B and C were obtained through linear fitting of individual data sets from A.

Discussion

The molecular structure of pS8-Aβ₄₀ fibril and its seeding-prone feature have the following implications: First, the molecular architectures of Aβ pathological plaques, and the underlying fibrillar structures, may be different among specific AD subtypes (1), and different molecular strains of Aβ may propagate their features in brains (8). Here, we showed an example of how certain fibrillar structures initiated from less abundant Aβ species might dominate the polymorphisms in amyloid deposits. If a particular amyloid plaque contains the mixture of pS8-Aβ₄₀ and wt2-Aβ₄₀ fibrils, and both fibrils are potential pools of seeds to induce further amyloid deposition, it is clear from our results that the structure originated from pS8-Aβ₄₀ will become dominant after long evolution. We showed previously the pS8-Aβ₄₀ could cross-seed not only wt-Aβ₄₀, but also wt-Aβ₄₂ and pyroglutamate-E3-Aβ₄₀ (23); meaning, that the molecular structure originated from pS8-Aβ₄₀ may propagate to multiple Aβ subtypes, which potentially enhance its dominancy in plaques. Given the highly heterogeneous compositions in pathological AD plaques and the cross-seeding nature of amyloid proteins, it is reasonable to suspect that a dominant Aβ fibril structure extracted from postmortem plaques might be originated from less-abundant morphologies (either a structure formed by a sparse amyloid species or a rare structure formed by the abundant wt-Aβ₄₀ or wt-Aβ₄₂), as long as such morphologies are seeding-prone. Our results on the pS8-Aβ₄₀ suggested that this PTM-Aβ subtype could be a potential source for such seeding-prone morphology. Since other PTM-Aβ subtypes, especially the E3-pyroglutarmated Aβ, also introduce modifications to the N terminus, they may influence the seeding properties of resultant fibrillar morphologies similarly (9).

Second, our results from current and previous works (23) suggest that the quantities of certain Aβ species do not necessarily correlate with their neuronal cellular toxicities. We have shown previously that the pS8-Aβ₄₀ fibrils possessed a higher level of cellular toxicity compared to the wt2-Aβ₄₀ fibrils, and such difference in cytotoxicity was retained in the wt-Aβ₄₀ fibrils grown from these two seeds (22). Therefore, for two Aβ plaques that contain both morphologies from pS8-Aβ₄₀ and wt2-Aβ₄₀, their relative cytotoxicity should in principle correlate with the population of the morphology that has been propagated from the original pS8-Aβ₄₀ fibril structures, rather than the absolute quantities of either wt-Aβ₄₀ or pS8-Aβ₄₀ peptides. Therefore, it may be critical to utilize structure-specific molecular probes (8, 35) and imaging techniques to quantify pathological AD plaques to illustrate their potential correlation with the disease progression.

Third, the molecular structure of pS8-Aβ₄₀ fibril highlighted the significance of N terminus in manipulating both the fibrils structures and the aggregation processes. The presence of an ordered N-terminal segment in Aβ fibril structures may be biologically relevant, because most recent structural studies on the ex vivo brain-extracted Aβ₄₀ fibrils (4) and more pathological Aβ₄₂ fibrils (36–38) both yielded models with ordered N termini. Interestingly, a recent molecular structural model for Aβ₄₂ fibril showed that its N terminus formed intramolecular interactions with C-terminal loop segments, similar to the present pS8-Aβ₄₀ fibril structure. The difference was that intermolecular interactions served as the main restraints for the association between N termini and fibrillar core in those structures, while intramolecular interactions play the key role in pS8-Aβ₄₀ fibrils. This structural feature suggests that the nucleus formed by pS8-Aβ₄₀ fibril may contain a more structurally ordered N terminus compared to the wild-type fibrils. Elongation from such a well-ordered nucleus leads to a more stable and rigid fibrillar structure, which was proved by the denaturing assay and 2H NMR spectroscopy (Fig. 4), and further consistent with the observation that the fibrillar morphologies of pS8-Aβ₄₀ did not vary significantly upon generation seeding (SI Appendix, Fig. S1). Several PTM-Aβ subtypes with modified N termini were shown to possess accelerated fibrillation kinetics and/or more stable fibril structures (9, 12, 18, 19, 39, 40). It will be worth investigating whether the presence of more ordered structures at the N termini is a common feature for PTM-Aβ fibrils and how it would correlate with the changes in the physicochemical properties of fibrillation.

Materials and Methods

Peptide Synthesis and Purification.

All peptides (with different selectively labeling schemes summarized in SI Appendix, Table S1) were synthesized manually using routine solid-phase peptide synthesis protocols. The experimental details were provided in SI Appendix.

Fibrillation.

The pS8-Aβ₄₀ fibrils used in the present work were grown from generation seeding protocols to ensure homogeneous structure (25). The details were provided in SI Appendix. The twofold wt-Aβ₄₀ (wt2-Aβ₄₀) fibrils were prepared using literature protocols (24), and featured morphologies (i.e., straited ribbon) were confirmed using negatively stained TEM.

TEM.

Negatively stained TEM was utilized to observe the morphology of fibrils. Briefly, a 10 μL aliquot of fibril solution was placed on a glow-discharged carbon film, supported by lacey carbon on a copper TEM grid (300 mesh; Ted Pella Inc.) and adsorbed for 2 min. The grid was blotted, rinsed with 10 mL of deionized water, blotted, rinsed again, blotted, then stained with 2% uranyl acetate for 30 s, blotted, and dried in air. TEM images were recorded on a FEI Morgagni microscope, operating at 80 kV, equipped with a side-mounted Advantage HR camera (Advanced Microscopy Techniques). Images were recorded with 44,000× to 89,000× magnifications. Dark-field TEM images of an unstained grid were recorded for MPL measurements, as previously described (26). To analyze the MPL, integrations of intensities were measured for selected regions in fibrils, the internal standard Tobacco Mosaic Virus (TMV), and the background of TEM grid. For the pS8-Aβ₄₀ fibrils, which showed high tendency to form aggregated bundles, the segments with clearly separated filaments (with both single and parallel straited-ribbon filaments) were carefully selected for analysis. Over 500 segments on fibrils were analyzed from ∼70 dark-field TEM images to generate the histograms shown in Fig. 1C, where individual MPL value was calculated using the equation: $MPL=\left(I_{TMV}-I_{B,TMV}\right)/\left(I_{Fibril}-I_{B,Fibril}\right)\times 131kDa/nm$, where $I_{TMV}$ and $I_{Fibril}$ were integrated intensities in defined areas on TMV and fibrils, and $I_{B,TMV}$ and $I_{B,Fibril}$ were integrated intensities for the size of areas of background adjacent to the selected TMV and fibrils respectively.

Thioflavin-T (ThT) Fluorescence Kinetics.

The kinetics of seeded wt-Aβ₄₀ fibrillation in the presence of the pS8-Aβ₄₀ seeds, the wt2-Aβ₄₀ seeds (Fig. 3A), and mixtures of pS8-Aβ₄₀/wt2-Aβ₄₀ seeds (Fig. 3C) were recorded by ThT fluorescence assay on a multimode plate reader (BioTek Inc.). Monomeric pS8-Aβ₄₀/wt-Aβ₄₀ and parent fibrils (both wt2 and pS8, 50 µM) were prepared using the protocols introduced in the previous section, and seeds were generated by sonicating the parent fibrils continuously for 30 s on ice-bath. Different volumes of seeds, monomeric wt-Aβ₄₀ and ThT stock solution were mixed, and the final concentrations of wt-Aβ₄₀ and ThT were 50 µM and 25 µM, respectively. Kinetics were recorded using a 96-well plate (Corning Inc.) incubated quiescently at 37 °C. Each well contains 100 µL mixed solution of fibrils and ThT stock. ThT fluorescence emission was recorded every 10 min with 10 s slow orbital shaking right before reading. Excitation and emission wavelengths were set to 450 ± 25 nm and 500 ± 13 nm, respectively, using filter sets. Three to five parallel experiments were performed for each seeded fibrillation system. Repetitions of ThT fluorescence traces were shown in SI Appendix, Figs. S10 and S12. Representative averaged fluorescence traces for pS8-Aβ₄₀, wt2-Aβ₄₀, and mixture-seeded fibrillation were reported in Fig. 3 A and C.

Thermostability Test of Fibrils Using SDS/PAGE.

A 100 μL aliquot of pS8-Aβ₄₀ or wt2-Aβ₄₀ parent fibrils was mixed with 1 mL guanidine hydrochloride (GdnHCl) with various concentrations from 1 M to 6 M. The final concentrations of GdnHCl were from ∼0.9 M to ∼5.5 M, as shown in Fig. 4A. The detailed procedure was provided in SI Appendix.

Solid-State NMR Spectroscopy.

Static ²H solid-state NMR spectroscopy and ¹³C and ³¹P Magic Angle Spinning (MAS) solid-state NMR spectroscopy were applied to determine the core rigidity and structure of pS8-Aβ₄₀ fibrils, respectively. Details on NMR sample preparations, data acquisition, spectral modeling, and structural modeling (using Xplor-NIH package) are included in SI Appendix.

Supplementary Materials

Coordinates from structural modeling have been deposited into the Protein Data Bank (PDB ID 6OC9) and the assigned chemical shifts have been deposited in the Biological Magnetic Resonance Data Bank (BMRB ID 30596). Detailed descriptions of experimental procedures, including peptide synthesis, ssNMR spectroscopy and structural modeling, are provided in SI Appendix. Additional TEM images, ssNMR spectra, ThT kinetic traces, 2H relaxation data, as well as tables that summarizes the labeling schemes, chemical shifts, structural restraints, and fitting parameters from dynamic measurements, are also included.