E22G Aβ40 fibril structure and kinetics illuminate how Aβ40 rather than Aβ42 triggers familial Alzheimer’s

Tehrani, Mohammad Jafar; Matsuda, Isamu; Yamagata, Atsushi; Kodama, Yu; Matsunaga, Tatsuya; Sato, Mayuko; Toyooka, Kiminori; McElheny, Dan; Kobayashi, Naohiro; Shirouzu, Mikako; Ishii, Yoshitaka

doi:10.1038/s41467-024-51294-w

E22G Aβ40 fibril structure and kinetics illuminate how Aβ40 rather than Aβ42 triggers familial Alzheimer’s

Article
Open access
Published: 15 August 2024

Volume 15, article number 7045, (2024)
Cite this article

Download PDF

You have full access to this open access article

From

View current issue

E22G Aβ40 fibril structure and kinetics illuminate how Aβ40 rather than Aβ42 triggers familial Alzheimer’s

Download PDF

185 Accesses
Explore all metrics

Abstract

Arctic (E22G) mutation in amyloid-β (Aβ enhances Aβ40 fibril accumulation in Alzheimer’s disease (AD). Unlike sporadic AD, familial AD (FAD) patients with the mutation exhibit more Aβ40 in the plaque core. However, structural details of E22G Aβ40 fibrils remain elusive, hindering therapeutic progress. Here, we determine a distinctive W-shaped parallel β-sheet structure through co-analysis by cryo-electron microscopy (cryoEM) and solid-state nuclear magnetic resonance (SSNMR) of in-vitro-prepared E22G Aβ40 fibrils. The E22G Aβ40 fibrils displays typical amyloid features in cotton-wool plaques in the FAD, such as low thioflavin-T fluorescence and a less compact unbundled morphology. Furthermore, kinetic and MD studies reveal previously unidentified in-vitro evidence that E22G Aβ40, rather than Aβ42, may trigger Aβ misfolding in the FAD, and prompt subsequent misfolding of wild-type (WT) Aβ40/Aβ42 via cross-seeding. The results provide insight into how the Arctic mutation promotes AD via Aβ40 accumulation and cross-propagation.

Aβ(1–42) fibril structure illuminates self-recognition and replication of amyloid in Alzheimer's disease

Article 04 May 2015

Conformational strains of pathogenic amyloid proteins in neurodegenerative diseases

Article 30 May 2022

A new era for understanding amyloid structures and disease

Article 20 September 2018

Introduction

Alzheimer’s disease (AD) is marked by the accumulation of extracellular senile plaques and intracellular tau-containing neurofibrillary tangles within brain parenchyma. The primary components of senile plaques are amyloid-β (Aβ) peptides in a fibrillar aggregated form known as amyloid fibrils^1,2. Misfolding of Aβ isoforms ranging from 38 to 43 residues plays a central role in AD^1,2. Although 40-residue Aβ (Aβ40) is the most abundant Aβ species in the brain, a major part of Aβ fibrils in amyloid plaques are comprised of 42-residue Aβ (Aβ42), which is more prone to aggregation and more cytotoxic than Aβ40^1,2,3. Structural divergence of Aβ fibrils has been associated with their varied cytotoxicity³ and contrasting clinical and pathological characteristics of AD^4,5,6,7. Thus, structural elucidation of amyloid fibrils of Aβ could illuminate the disease mechanism and assist the development of therapeutic or preventive approaches.

Some pathogenic mutations of Aβ are associated with familial AD (FAD) and familial cerebral amyloid angiopathy (CAA), the latter of which is characterized by symptoms such as cerebrovascular microhemorrhage. They typically promote the aggregation kinetics of Aβ, rendering it significantly more fibrillogenic, while some also facilitate the formation of toxic oligomer species^8,9. Several FAD-causing mutations are concentrated at residues 21–23 in the hydrophobic core of the Aβ sequence (Fig. 1a underlined). These mutations include Flemish (A21G), Osaka (E22Δ), and Arctic (E22G) variants^8,10. These mutations lead to cognitive impairments at the early stages of their lives (40s to 60s), while other mutations in the region such as Iowa (D23N), Dutch (E22Q), and Italian (E22K) lead to familial CAA (Fig. 1a)^8,11,12,13.

**Fig. 1: Seeded E22G Aβ40 quiescent fibrils display unbundled and disperse monomorphic features with low ThT fluorescence.**

The Arctic (E22G) mutation promotes the assembly of Aβ into oligomers and fibrils, resulting in early-onset FAD^9,10,14. Unlike other mutations at the same residue, such as E22K and E22Q, which primarily induce CAA, E22G is the only known mutation with typical clinical characteristics of sporadic AD. E22G carriers exhibit senile plaques and other neurofibrillary pathologies with progressive cognitive decline, and in some cases moderate cerebrovascular events^11,15,16. Cotton-wool plaques, which are rare in sporadic AD, are more pronounced in the reported cases of Arctic FAD patients^17,18,19. These large ball-like plaques with a diameter up to ~120 μm and without a densely packed amyloid core are often observed in AD patients with exon-9 deletion mutation in the gene encoding presenilin 1, a subunit of γ-secretase^16,18. Amyloid plaques in sporadic AD, typically with a diameter of ~10 μm, are mainly comprised of the Aβ42 isoform^20,21. Interestingly, cotton-wool plaques in patients with the E22G mutation have greater accumulation of Aβ40, especially in the core of the plaque. The core is often surrounded by both Aβ42 and Aβ40 fibrils, forming a ring-like structure^17,18. Cotton-wool plaques in Arctic FAD patients exhibit a very weak affinity for Pittsburgh compound, a thioflavin-T (ThT) analog, and low staining with thioflavin-S and Congo Red^18,19. In the pathological sense, the Arctic FAD thus can be considered a distinctive subtype of AD^5,6. Similar cotton-wool plaques were also reported for late-onset sporadic AD with mild to marked CAA²². Despite various speculations, these variations from typical plaque have been a pathological mystery in AD and there has been no molecular-level understanding of the issues.

Structures of amyloid fibrils with FAD-associated mutations have attracted great interest because studying Aβ for E22G and other FAD-related mutants can explain how these mutations promote the onset of AD at the molecular level. Such fibril structures may also provide insights into the mechanisms of sporadic AD. Recent solid-state nuclear magnetic resonance spectroscopy (SSNMR) and cryo-electron microscopy (cryoEM) studies on wild-type (WT) Aβ40 and Aβ42 fibrils demonstrated similar structures with β-sheet motifs arranged in a U-shape^{23,24,25,26,27} and S-shape^{28,29,30,31,32,33}, respectively. More recent cryoEM and SSNMR studies for Aβ40 and Aβ42 fibrils isolated from AD patients reported distinctive conformers for which β-sheets are arranged in I-shape³⁴ and ν-/υ-shapes³⁵, respectively; considerable deviations of the structures from the U- and S-shape structures suggest notable structural variations of brain-derived Aβ fibrils in AD. SSNMR analysis has been also effective in defining structures of other amyloid fibrils^36,37.

Nevertheless, structural details of FAD-related Aβ fibrils are scarce for in vitro or brain-derived systems. Until recently, the only atomic structures of FAD-related Aβ fibrils have been reported for E22Δ and D23N Aβ40, which bear C-shaped and U-shaped β-sheet motifs, respectively^38,39,40. Thus, various efforts have been made to study the structures of E22G Aβ40 fibrils. Early transmission electron microscopy (TEM), scanning TEM (STEM), atomic force microscopy (AFM), and SSNMR studies reported at least five distinct fibril morphologies for E22G Aβ40¹⁰. Earlier studies by us and others showed that E22G Aβ40 exhibits high plasticity and its fibrils are likely to adopt both an Aβ42-like S-shaped β-sheet motif as well as an Aβ40-like U-shaped β-sheet motif but without any structural details^41,42. Very recent cryoEM analysis of brain-derived E22G Aβ indicated that the major form of E22G Aβ fibrils is comprised of two pairs of protofilaments that contain U-shaped motifs (E11–G37) and S-shaped motifs (V12–V40)⁴³. Although it was presumed that fibrils are likely made of E22G Aβ(x–40) or Aβ (x–37) based on mass spectroscopy, it appears that E22G Aβ(x–42) was almost completely depleted in the mass spectrometry analysis, probably due to the insolubility of the Aβ (x–42) system under the mild conditions employed. It should be noted that Aβ (x–y) denotes the Aβ sequence from residue x to y. These results also contradict the abundance of Aβ (x–42) from previous mass spectrometry and immunofluorescence studies¹⁸. Thus, it is unclear what isoform(s) are incorporated into the system observed in the cryoEM structure, which potentially lacked density for the C-terminal residues. Also, the high propensity of E22G Aβ to form polymorphic amyloid fibrils (i.e., polymorphs) is likely to allow for different structures of E22G Aβ fibrils for other patients.

More critically, previous structural studies of E22G Aβ fibrils have not revealed any connections between fibril structures with distinctive pathological and clinical features of the Arctic FAD. Recent cryoEM structures of E22G Aβ fibrils derived from Arctic FAD patients, for example, did not explain whether these features of cotton-wool plaques in the FAD can be attributed to any of the resultant fibril structures. Due to the inherent structural and chemical heterogeneity of brain-derived fibrils, it still poses a great challenge to relate the atomic-level structures of individual fibrils with the macroscopic and pathological characteristics of amyloid plaques in the notable subtype of AD.

Another critical but elusive aspect of FAD-related mutant Aβ fibrils is their misfolding kinetics. Mutations of Aβ in FAD are typically heterozygous; WT and mutant Aβ proteins should be produced in an equimolar ratio. Since Aβ40 and Aβ42 are produced in approximately a 9 to 1 ratio^44,45, FAD-related mutant Aβ40 and Aβ42 isoforms are likely produced in a similar ratio, resulting in an Aβ40-rich (both WT and mutant) mixture among the four isoforms. Based on the data for sporadic AD, it is often presumed that mutant Aβ42 may be responsible for the onset of FAD, but there is currently no experimental evidence. For the Arctic mutation, however, greater accumulations of Aβ40 fibrils in amyloid plaques have been reported^17,18. Moreover, the mechanism of cross-seeding between mutant Aβ and WT Aβ isoforms may promote plaque formation involving Aβ isoforms^42,46. Nevertheless, it is largely unexplored how such cross-seeding can impact fibril and plaque formation for Arctic and other FAD mutants. Again, to reveal connections of such cross-seeding properties with the fibril structure, a monomorphic fibril is demanded.

In the present study, we developed a methodology to prepare a monomorphic E22G Aβ40 fibril sample suitable for structural elucidation. More importantly, the bottom-up synthetic-biology approach of building a fibril strain with a defined structure allows us to correlate macroscopic and kinetic features of the fibril with pathological features of the FAD without involving any complex brain materials. Unlike typical densely bundled WT Aβ42 and Aβ40 fibrils, the obtained dispersed and unbundled amyloid fibrils of E22G Aβ40 formed a less compact precipitate; the fibrils also indicated low ThT fluorescence. These features are consistent with those observed for the Aβ40-rich core of cotton-wool plaques of FAD patients with the Arctic mutation^16,19,47. Using SSNMR and cryoEM, we determined an atomic structure of FAD-related E22G Aβ40 fibrils. The structure reveals the presence of two Aβ molecules forming two inner layers with slightly distorted C₂ symmetry; each Aβ molecule adopts a previously unidentified W-shaped parallel β-sheet motif. The cryoEM density also suggests two additional short β-sheet outer layers above and below the inner layers, but SSNMR data unverified the structures. Our data demonstrate that the distinctive pathological features found in the FAD or cotton-wool plaques may originate from the distinct Aβ fibril structure. We also investigated the misfolding kinetics of E22G and WT Aβ40/Aβ42 monomers and the cross-seeding properties of E22G Aβ40 fibrils with WT Aβs using ThT fluorescence and MD simulations. We revealed that among four isoforms of E22G and WT Aβ40/Aβ42, E22G Aβ40 first self-assembled into fibrils under conditions reflecting a physiological concentration ratio of Aβ40 to Aβ42 (~9:1). It was also demonstrated that E22G Aβ40 fibrils can trigger further fibril formation of WT Aβ42 and WT Aβ40 via cross-seeding. Our MD simulations indicated that W-shaped chimera Aβ fibrils comprised of E22G Aβ40 and WT Aβ40/Aβ42 are stable, suggesting that the fibril structure is likely responsible for the cross-seeding properties. These findings imply a possible role for E22G Aβ40 rather than E22G Aβ42 as the initial driving force and the trigger for the cross propagation for the onset of FAD with the Arctic mutation.

Results

Homogeneous E22G Aβ40 fibrils mimic amyloids in cotton-wool plaque

E22G Aβ40 fibrils have a high propensity to form multiple polymorphs with considerable heterogeneity^10,41, and this long precluded their atomic structural analysis. Indeed, heterogeneous short-bundled fibrils were observed in a TEM image of E22G Aβ40 fibrils prepared by incubating ~30 μM Aβ monomer at 37 °C under constant agitation, following a previously published protocol^28,42 (Fig. 1b). Although adding ‘seed’ fibrils to Aβ monomers during incubation was reported to improve homogeneity^28,48, a sample prepared by this successive seeding process over three generations (Fig. 1c, G3) or five generations (Fig. 1d, G5) produced short, straight filaments containing some striated ribbon-like fibrils with high heterogeneity. Here, we denoted fibril samples by incubating Aβ monomers with seed fibrils from the previous generation in the n-th successive generations as Gn samples, and unseeded fibril samples as G0. Importantly, heavily bundled morphologies practically prohibit single-particle analysis of fibrils by cryoEM owing to lack of morphological regularity, even when the structure of each protofilament is homogeneous. To achieve a reproducible preparation of highly homogenous E22G Aβ40 fibrils for structural analysis by cryoEM and SSNMR, we thoroughly optimized the incubation conditions as well as the purification protocol, sample concentration, and seed fragmentation methods (see Supplementary Figs. 1–3 for comparison with less successful attempts). Structurally homogeneous amyloid fibrils of E22G Aβ40 were obtained by incubating an E22G Aβ40 monomer solution with 5% (w/w) seed fibrils in a quiescent manner (i.e., without agitation) at physiological temperature (37 °C), which yielded a more homogeneous sample than incubation under constant agitation at 37 °C. By repeating this method from initial quiescent unseeded G0 fibrils, seeded fibrils were obtained from up to five successive seedings. Figure 1e–j shows a comparison of TEM images for (e) G0 and (f–j) G1–G5 seeded fibrils. The effectiveness of seeding was confirmed by ThT fluorescence assay as mode switching of kinetic curves from a sigmoidal curve for unseeded samples (Fig. 1k) to exponential curves for seeded samples (Fig. 1l), which reflects the first order kinetics that converts monomers into fibrils using seed fibrils as a template (comparison with agitated samples in Supplementary Fig. 4). A mixture of long, straight and twisted fibrils was observed for the unseeded G0 sample (Fig. 1e), as previously observed¹⁰. A considerable improvement in homogeneity was observed for the G3 sample (Fig. 1h). For the seeded G5 sample, the morphology switched to highly homogenous and long, twisted fibrils (Fig. 1j). The dominant twisted fibrils have a cross-over distance of 35–45 nm, a width of 10–15 nm between cross-overs, and a width of 4–5 nm at cross-overs (yellow arrows in Fig. 1j). This unbundled, highly homogeneous fibril morphology is ideal for structural analysis by cryoEM. Interestingly, we observed a dramatic decrease in the maximum ThT fluorescence intensity to ~20% for the G0 sample (Fig. 1l) for latter generations in G3–G5. Turbidity measurements at \({{{\rm{\lambda }}}}\) = 600 nm revealed no decrease in the amount of fibrils for each generation (Supplementary Fig. 5). The fluorescence was also much lower than that for agitated fibril samples (Supplementary Fig. 4). Thus, we attributed the low ThT fluorescence to fundamental structural changes in fibrils, which are consistent with morphological changes in TEM images (Fig. 1e–j). Indeed, previous clinical studies on FAD patients with the Arctic mutation reported a low cortex PET signal from ¹¹C-labeled or ³H-labeled Pittsburgh compound, an analog of ThT^19,47, despite the reported abundance of amyloid plaques in Arctic FAD patients¹⁸. While detailed TEM analyses of amyloids have not been reported for plaques of Arctic FAD, the dispersed, unbundled morphology observed in TEM images (Fig. 1j) is also consistent with the fact that Arctic FAD patients often exhibit cotton-wool plaques, large spherical senile plaques of ~50 μm in diameter, for which densely packed amyloids are typically missing in the core^18,49, where Aβ(x–40) is abundant. We also confirmed the less compact nature of fibrils by monitoring the size of a Congo Red-stained precipitate of quiescent E22G Aβ40 fibrils observed after mild centrifugation at 4000 × g for 5 min, which was approximately 12 times larger in volume than that of WT Aβ40 fibrils (Supplementary Fig. 6). These observations are consistent with the larger plaque size and poor staining by Congo Red for FAD cotton-wool plaques. It is also noteworthy that more Aβ40 than Aβ42 is available for plaque formation in a brain (i.e. [Aβ40]/[Aβ42] ~ 9). Thus, the interesting structural and kinetic features of E22G Aβ40 fibrils reported in this study can explain some uncommon characteristics of the cotton-wool plaques observed in FAD patients, as discussed below.

SSNMR indicates a homogeneous conformer for E22G Aβ40 fibril

SSNMR experiments were carried out on E22G Aβ40 fibrils that were uniformly ¹³C and ¹⁵N isotope-labeled at selected amino acid residues to examine site-specific heterogeneities and the presence of multiple conformations^28,48 not detectable by TEM or H/D exchange methods. 2D ¹³C-¹³C (right) and ¹³C-¹H (left) correlation SSNMR spectra (Fig. 2a–d) for (a, c) seeded G6 quiescent fibrils and (b, d) agitated E22G Aβ40 fibrils as a control (all seeded G6) were obtained for four separate samples uniformly ¹³C- and ¹⁵N-labeled at (a, b) L17, F20, G25, I32, V36 and (c, d) F20, A21, V24, G29 and I31. Spectra for agitated fibrils (Fig. 2b, d) confirmed the existence of 3 to 4 distinct conformers, as indicated by blue (Val) and magenta (Ile) arrows for V36 and I32 (Fig. 2b), and V24 and I31 (Fig. 2d). By contrast, data for quiescent fibrils (Fig. 2a, c) largely showed a single peak for each adjacent ¹³C-¹³C pair, indicating the presence of a predominantly single homogeneous conformer (signal assignments; Supplementary Table 1), consistent with the high homogeneity observed in TEM. ¹³C full spectral line widths at half-height were in a range of 1.1–2.0 ppm. Interestingly, weaker peaks were observed for some residues (Supplementary Table 2). For example, for A21 and F20, the 2D SSNMR data (Fig. 2c) exhibited one strong peak (black arrows) and one weak cross peak (black dotted arrows). The relative intensities for the minor peaks were 35 ± 6% and 24 ± 13% for A21 and F20, respectively. Close examination of these and additional SSNMR data (Supplementary Fig. S7a, c, e) indicated the presence of minor species for quiescent fibrils for F20, A21, S26, K28, I32, G33 and M35, with the average relative intensity of 57 ± 5% (from ¹³C-¹³C correlation data; Supplementary Table 3), which cannot be attributed to impurities. Minor peaks might be present for other residues, but signals were not resolved or detected.

**Fig. 2: SSNMR results indicate a largely homogeneous major conformer for quiescent fibrils.**

CryoEM and SSNMR reveal a distinctive W-shaped motif for E22G Aβ40 fibrils

The G5 and G6 of the seeded quiescent fibrils were used for cryoEM and SSNMR analyses, hereafter described. The final structural models for the E22G Aβ40 fibril sample are shown in Fig. 3a, b, d. CryoEM images of the E22G Aβ40 fibrils revealed high structural homogeneity for the twisted morphology, with a width of ~10 nm and a cross-over distance of ~35 nm (Fig. 3f). The 2D class averages displayed clear rungs along the direction of the fibril axis, indicating the typical cross-β structure of the amyloid fibrils (Supplementary Fig. 8). The cryoEM map (Fig. 3b) was reconstructed at a nominal resolution of 2.5 Å based on the 0.143 Fourier shell correlation (FSC) criterion (Supplementary Fig. 8). While we also tested C2 and pseudo-C2₁ symmetry (Supplementary Fig. 9) besides the adopted C1 symmetry, the reconstructed cryo-EM maps for the C2 and C2₁ symmetry resulted in a similar quality (2.86–2.89 Å resolution); nevertheless, the resolution and densities did not significantly improve compared with those for the C1-symmetry map. Since the symmetry operation did not improve the cryo-EM densities, we adopted the C1-symmetry to prevent inducing artificial effects. The handedness of E22G Aβ40 fibrils was determined using SEM, which confirmed a left-handed twist (Supplementary Fig. 10) and no right-handed fibrils; the left-handed helical twist is consistent with the other in vitro-produced Aβ fibrils^31,34.

**Fig. 3: CryoEM analysis reveals W-shaped core-layer structures for quiescent E22G Aβ40 fibrils.**

The density map for the inner two core layers was well defined (green in Fig. 3b and yellow and green in Fig. 3d), allowing the modeling without H/D exchange or other conformational data, which are often required for cryoEM-based structural determination of amyloid fibrils especially for the data with limited resolution. The first 11 N-terminal residues were undetectable in the cryoEM map, presumably due to the high flexibility. The subsequent residues 12–40 were unambiguously modeled (Fig. 3b). The density maps also suggested the presence of two outer layers, for which we tentatively modeled residues from N27 to V36 (blue and cyan in Fig. 3c), based on the fitting with the side chain densities (Supplementary Fig. 11). However, as discussed below, the structure in the outer layers was not consistent with the SSNMR data and therefore, the outer layer models were omitted (Fig. 3a, b, d and Supplementary Table 6).

The core layers of E22G Aβ40 fibrils have an extended W-shaped motif (yellow in Fig. 3a, c) unlike in vitro-prepared Aβ40 fibrils, which typically have a U-shaped β-turn-β motif ^23,24,25. The W-shaped motif is comprised of a short stretch of β-sheet (K16–F20) and a longer kinked β-sheet (A30–G38), which are connected by a curved turn/coil region (A21–G29). The W-shaped (yellow in Fig. 3c) and slightly different inverted W-shaped (green in Fig. 3c) motifs are stacked against each other; the interface of the two layers is largely filled with hydrophobic residues (green residues in Fig. 3a); and hydrophilic residues (cyan) and positively charged (blue) and negatively charged (red) residues are generally exposed to the surface (Fig. 3a). Interestingly, the turn region from D23 to K28 is a cluster of hydrophilic and charged residues. This curved turn region appears to be stabilized by the flexibility introduced by the E22G mutation and hydrophobic interlayer contacts such as I31–G33, G25–L34 and G25–V36, and intralayer contacts such as F20–V24, D23–S26, D23–K28, N27–I31, N27–L34 and I32–M35 (Fig. 3e).

At a glance at various Aβ fibril structures (Supplementary Fig. 12), the extended W-shaped motif may appear to be similar to that observed in the recently reported WT Aβ40 fibril structure isolated from an AD patient³⁴ (Supplementary Fig. 12b). However, unlike brain-derived WT Aβ40 fibrils, which have an extended, straight, I-shaped motif, E22G Aβ40 fibrils have the extended W-shaped motif. This distinction can be explained by the presence of an additional turn at G22 and a shift in the sidechain orientation from the mutation site to the N-terminus. The bulky and charged sidechain of Glu22 in WT Aβ40 is typically oriented towards the fibril’s outer surface³⁴. By contrast, the E22G mutation introduces flexibility around the mutated Gly site, which facilitates F20–V24 and V18–F20 hydrophobic interactions. Limited exposure of the hydrophobic sidechains (green residues in Fig. 3a) on the surface is consistent with the unusually weak ThT fluorescence⁵⁰ and lack of lateral stacking of the fibrils. In summary, the E22G Aβ40 fibrils exhibit a distinctive structural motif not previously reported, as further discussed below.

SSNMR co-analysis confirms only the core-layer E22G Aβ40 structure

We carried out a co-analysis of the system by SSNMR to examine the initial cryoEM structure in Fig. 3c. A single set of peaks is expected for residues 12 to 40 for the core layer, assuming structural differences between the two core layers are small or negligible. Another set of cross peaks is expected for the outer layer models encompassing the residues 27 to 36. The peak intensities for the inner and outer layers should be equal for each ¹³C site. Indeed, we observed a set of cross peaks from L17 to V39 for isotope-labeled residues (Supplementary Table 1), which we attributed to the peaks for the core layers. We observed another set of cross peaks (Supplementary Table 2) with noticeable chemical shift differences from the first set. However, as discussed above, the peak intensities were ~57% of the peaks for the core layers on average. Also, this minor set of cross peaks was observed for residues 20 to 35, rather than for residues 27–36. Due to the discrepancies, we decided to examine the validity of the outer layer structure using NMR data. For this purpose, we compared ¹³C secondary chemical shifts (Δδ) obtained from the experimental SSNMR chemical shifts (δ_exp; Supplementary Tables 1 and 2) and the corresponding values calculated using ShiftX2⁵¹ software from the cryoEM structure for the core layer (Fig. 4a) and the outer layer (Fig. 4b). The Δδ values represent deviations from the corresponding random coil chemical shifts (δ_rc; i.e., Δδ = δ_exp − δ_rc)⁵². If the experimental and calculated values agree, all data points should align with the dotted line. Data for the core layers (Fig. 4a) showed excellent agreement. Pearson’s correlation coefficient (r) for ¹³CO, ¹³C_β and ¹³Cβ was calculated to be 0.91 (r = 1 indicates a perfect correlation). On the other hand, data for the outer layers (Fig. 4b) showed considerable deviation; the calculated r value was 0.38, indicating major structural differences. Based on the results, we rejected the structures for the outer layers proposed by the cryoEM analysis. It is most likely that these minor SSNMR signals are attributed to the outer layers; the chain assignment of the outer layers in the cryoEM data needs to be revisited with additional data in our future research.

**Fig. 4: SSNMR co-analysis excludes the outer layer structure of E22G Aβ40 fibrils.**

For the core layers, we further validated the proposed model by identifying long-range inter-residue contacts by SSNMR using 2D ¹³C-¹³C correlation SSNMR experiments with extended ¹³C-¹³C DARR⁵³ mixing periods of 200 or 300 ms (red spectra in Fig. 5 and Supplementary Fig. 13). Long-range inter-residue contacts, for example, were identified for G25–V36 (Fig. 5a) and G25–L34, N27–I31 and N27–L34 (Fig. 5b). The long-range contacts observed in the SSNMR data (Supplementary Table 5) are all consistent with the cryoEM structure and suggested contacts (Fig. 3e). No long-range cross peaks were observed for the minor species. Secondary structure analysis from ¹³C chemical shifts using TALOS-N⁵⁴ (Supplementary Fig. 14a, b and Supplementary Table 4) indicated the existence of two β-strands from V18 to A21 and from K28 to V36, consistent with the β-strand regions in Fig. 3c. The torsion angles from the NMR data also showed some deviations from a regular β-sheet conformation at G33, consistent with the kink of the β-sheet around G33 for the core layers in Fig. 3a (also shown in Supplementary Fig. 14c, d).

**Fig. 5: SSNMR ¹³C-¹³C long-range contact data further confirm the core-layer structure.**

NMR-based comparison highlights a distinctive fold of E22G Aβ40 fibril

¹³C backbone chemical shifts for major species of E22G Aβ40 fibrils were compared with those from previously reported structural studies for WT Aβ40^{23,25,27,34,48}, Aβ42^{28,31,55,56,57}, and E22Δ Aβ40³⁸ fibrils. Pearson’s correlation coefficients (r; Fig. 6) were <0.8 for all (a) brain-derived I-shaped Aβ40³⁴, (b) in vitro U-shaped Aβ40²³, and (c) in vitro S-shaped Aβ42²⁸ fibrils, suggesting notable structural differences from E22G Aβ40 fibrils. Indeed, no similarities were noted between the molecular conformations of E22G Aβ40 fibrils and WT Aβ40/Aβ42 fibrils (Supplementary Fig. 12c, d) or E22G Aβ fibrils from FAD human brain tissue⁴⁷ (Supplementary Fig. 12e). Additional chemical shift analysis (Supplementary Fig. 15a) showed a high r value of 0.91 between E22G Aβ40 fibril (major form) and a polymorph of Aβ42 fibrils recently identified in our lab⁵⁷, for which detailed structure and motif information is still unavailable. For other Aβ40/Aβ42 fibrils, direct ¹³C chemical-shift comparison (Supplementary Fig. 15a) or comparison with ¹³C chemical shifts predicted from the structures (Supplementary Fig. 15b) show limited similarity (r < 0.86) for the major species. In summary, the analysis clarified the distinctive structural features of the W-shaped E22G Aβ40 fibrils.

**Fig. 6: ¹³C chemical shift comparison confirms distinctive structural features of W-shaped E22G Aβ40 fibrils.**

Kinetic data suggest that E22G Aβ40 may trigger misfolding in Arctic FAD

It is widely believed that Aβ42 triggers the onset of AD through its self-assembly or misfolding. However, for FAD involving heterozygous Arctic Aβ mutation, notable accumulation of Aβ40 in amyloid plaques has been reported^17,18. Thus, through kinetic studies, we evaluated a previously untested hypothesis that Aβ40 rather than Aβ42 may trigger Aβ self-assembly in Arctic FAD. As the ratio of Aβ40 and Aβ42 was reported to be 9 to 1^44,45, the concentration difference between these isoforms was incorporated into the kinetic analysis. We compared the incubation time dependence of ThT fluorescence (Supplementary Fig. 16) for (a) E22G Aβ40, (b) E22G Aβ42, (c) WT Aβ40 and (d) WT Aβ42, where Aβ concentrations were adjusted to (a, c) ~35 μM and (b, d) ~3 μM. Clearly, E22G Aβ40 exhibited the shortest average lag time of 12.5 ± 2.1 h before fibril formation (Fig. 7a). E22G Aβ42 also showed fibril formation after 30.7 ± 1.8 h. By contrast, none of the data for WT Aβ40 or WT Aβ42 showed any signs of fibril formation within 20 h (Supplementary Fig. 16). Similar results were obtained with a high salt buffer mimicking an extra-cellular solution (see Methods) in the incubation of the Aβ isoforms (WT/E22G Aβ40 ~ 15 µM and WT/E22G Aβ42 ~ 1.2–1.5 μM); E22G Aβ40 exhibited the shortest lag time of 3.3 ± 0.4 h despite the lower Aβ concentration (see Supplementary Fig. 17). Thus, these results are consistent with our hypothesis that E22G Aβ40, rather than E22G or WT Aβ42, is a primary driving force that triggers the self-assembly of Aβ in FAD involving the Arctic mutation. The result also explains findings from recent immunohistochemistry studies showing that the cores of cotton-wool amyloid plaques observed for FAD patients with the Arctic mutation are often dominated by Aβ(x–40) and contain little Aβ(x–42)¹⁸. Although the cores of cotton-wool plaques, which are barely visible by Congo Red staining, are often considered less dense than typical amyloids, the lack of bundled fibrils in Fig. 1j and the low ThT fluorescence of the W-shaped E22G Aβ40 fibrils suggest that traditional staining methods for densely bundled amyloid fibrils may not be effective for detecting amyloid plaques comprised of the more loosely packed W-shaped fibrils.

**Fig. 7: Kinetic and MD simulation data for E22G and WT Aβ isoforms suggest that E22G Aβ40 may be a major driving force of amyloid propagation in Arctic FAD.**

Cross-propagation of E22G Aβ40 fibril structure to WT Aβ40 and Aβ42

Finally, we examined the cross-seeding properties of the homogeneous E22G Aβ40 fibrils prepared for the structural studies. Examination of the W-shaped E22G Aβ40 fibril structure suggests that the main-chain arrangement around V40 may accommodate the association of both monomeric Aβ42 and Aβ40 to E22G Aβ40 fibrils by cross-seeding. Thus, the ability of the E22G Aβ40 fibril structure to cross-propagate to WT Aβ40 and Aβ42 was investigated by incubating WT Aβ40 and Aβ42 monomers at ~45 μM with E22G Aβ40 quiescent fibrils (G5) as seeds at 10% (w/w) of total fibril amount (Fig. 7b, c) under agitating conditions. In the absence of any seeds, our kinetics data demonstrated a sigmoidal or double-sigmoidal profile for WT Aβ40 and Aβ42 fibril formation, with lag times of 3.1 ± 0.5 h and 2.3 ± 0.3 h, respectively (Supplementary Fig. 18). The addition of E22G Aβ40 seed fibrils indeed diminished the lag time, altering the kinetic profiles of both isoforms to an exponential curve (Fig. 7b, c). The cross-seeding effect was more pronounced for WT Aβ42, for which fibrillization was complete within ~1 h. These results suggest that once E22G Aβ40 forms amyloid fibrils via self-assembly during the early stages of FAD, the assembled fibrils can trigger further amyloid formation of WT Aβ42 and WT Aβ40 via the cross-seeding mechanism. These results are consistent with the formation of a ring-like Aβ(x–42) amyloid deposit around a thinly deposited Aβ(x–40) core in the cotton-wool plaque found in Arctic FAD patients.

The stability of the secondary structures for the cross-seeded fibrils in a timescale of ~1 μs was also evaluated through preliminary MD simulation analyses (Fig. 7d–f; see also Supplementary Figs. 19–23), where the W-shaped structure determined in this work was adopted as the initial structure for (d) E22G Aβ40 fibril. We replaced half of the E22G Aβ40 molecules with (e) WT Aβ40 and (f) WT Aβ42 to convert the fibril into chimera fibrils. Then, the stability or compatibility of the cross-seed fibrils (see Methods and Supplementary Information for the details) was examined through the MD simulations for 500 ns. It was confirmed that for (d) the unmodified E22G Aβ40 fibril and (e) the WT Aβ40–E22G Aβ40 chimera fibril, the twisted fibril structures were largely unchanged for both E22G Aβ (yellow) and WT Aβ40 (magenta). For (f), the WT Aβ42–E22G Aβ40 chimera fibril, the structure was also largely intact as we expected, despite some minor destabilization on the WT Aβ42 side (cyan). The β-sheet fractions for the WT Aβ40 and Aβ42 sides of the chimera fibrils (38–45%) were also similar to those of the control (40–44%) (see also Supplementary Fig. 21a–c). The root-mean-squared deviation (RMSD) values of C_α positions of the core region (residues 22–36) exhibited small value (~2 Å) for all the W-shaped chimera fibrils in the MD simulation trajectories (Supplementary Fig. 19d–f). Relatively high fluctuations (>1.6 Å) were restricted to the outer parts of the W-shaped β-sheets (residues 12–17 and 37–40/42) by the per-residue root-mean-square fluctuation (RMSF) analysis, confirming the stability (Supplementary Fig. 23). As a negative control, we also performed corresponding MD simulations for S-shaped chimera fibril of WT Aβ42–WT Aβ40 (Fig. 7g), as the S-shaped Aβ42 fibril seeds do not promote fibril formation of WT Aβ40 monomers in cross-seeding as reported previously⁵⁴. The MD results demonstrate that the S-shaped β-sheet structures for WT Aβ40 (magenta) were disrupted while the S-shaped structures were largely unchanged for WT Aβ42 (cyan) (see also Supplementary Fig. 20d). For some of the trials, β-sheets at the Aβ40 side were dissociated, as indicated by the RMSD values and the structures (Supplementary Fig. 22). The β-sheet fractions of the WT Aβ40 side in the S-shaped chimera fibrils (18–34%) were notably smaller than those of the Aβ42 side (29–47%) (see also Supplementary Fig. 21d). These results confirmed the incompatibility of WT Aβ40 to the S-shaped WT Aβ42 fibril structure⁵⁴. In summary, the MD results for the W-shaped chimera fibrils provided supporting evidence and molecular-level insight into our kinetic results that the W-shaped E22G Aβ40 fibril seed can propagate misfolding of both WT Aβ40 and Aβ42. Our kinetic and MD data provide the initial direct experimental evidence that FAD mutations on Aβ can modulate their cross-seeding properties by altering the atomic structures of the fibril.

Discussion

Despite much interest in amyloid structures associated with FAD, elucidating the atomic-level structure of E22G Aβ fibrils by SSNMR and/or cryoEM and connecting the structure with major pathological characteristics of the relevant fibrils/plaques has proved challenging due to the unusually high heterogeneity¹⁰. This work has reported the initial successful example of isolating a monomorphic E22G Aβ40 fibril sample with twisted morphology by optimization of various incubation conditions. Importantly, the synthetic-biology approach of building a fibril strain with a defined structure allowed us to correlate the fibril properties with distinctive pathological features of the FAD without involving complex brain materials. The atomic-level structural model for in-vitro prepared E22G Aβ40 fibril was determined using cryoEM and SSNMR data. Our study showed that co-analysis by cryoEM and SSNMR can provide a more reliable structure for amyloid fibrils because SSNMR can sensitively detect the presence of multiple conformations that coexist for the same protofilament, which is generally difficult by other methods such as H/D exchange. The obtained structure reveals three remarkable features of E22G Aβ40 fibrils and their relevance to FAD when accompanied by the results of the kinetic studies.

First, the extended W-shaped structure of E22G Aβ40 fibrils differs greatly from those previously reported for WT Aβ40 and WT Aβ42 fibrils, which are known to form U-shaped^{23,24,25,26,27} (Supplementary Fig. 12c)/C-shaped⁵⁸ and S-shaped^{28,29,30,31,32,33} (Supplementary Fig. 12d) parallel β-sheet motifs, respectively. The structure is also different from those of fibrils for FAD-related mutant Aβ such as the C-shaped β-sheet motif found for E22Δ Aβ40³⁸. The extended structure of E22G Aβ40 shares some common features with that of the most prevalent WT Aβ40 fibril derived from the cortex of an AD patient³⁴, which has been an anomaly among reported Aβ fibril structures. Nevertheless, they also have distinct structural features. Compared with the extended straight conformation observed for brain-derived WT Aβ40 (I-shaped, Supplementary Fig. 12b), the E22G Aβ40 fibril in the present study has a highly curved and extended W-shaped β-sheet arrangement across the inner layers. This curved motif is established by the addition of a turn at the G22 mutation site and stabilized by resultant hydrophobic contacts including V18–F20 and F20–V24 contacts, the latter of which is missing in brain-derived WT Aβ40. Due to the additional turn, the mode of inter-layer contacts between the two core layers is also altered. For example, G25–V36, G25–L34, and N27–L34 contacts observed in E22G Aβ40 fibrils (Supplementary Fig. 12a) were not observed in brain-derived WT Aβ40 (Supplementary Fig. 12b). Thus, our results show that despite a few sequence differences from WT/mutant Aβ40 or Aβ42, E22G Aβ40 can adopt a distinctive extended W-shaped motif. The notable structure is likely to be responsible for faster self-assembly of E22G Aβ40, as well as the cross-seeding properties, as discussed below. While it remains to be examined whether this form of E22G Aβ40 fibril is relevant to fibrils present in FAD patients, this study strikingly demonstrated that a distinctive structure of the synthetic E22G Aβ40 fibril rather than other elements originating from a brain can be responsible for reproducing various “mysterious” features of the cotton-wool plaques in the FAD patients. The study also indicates that E22G Aβ40 amyloid fibrils having similar features/structures to those of the W-shaped Aβ fibrils may be involved in the cotton-wool plaque formation. Together with the atomic details, distinctive structure, and highly homogeneous morphology, the E22G Aβ40 fibril system may also present an excellent target for developing antibody therapies for FAD and sporadic AD.

Secondly, our kinetic studies revealed that E22G Aβ40 can form amyloid fibrils faster than E22G/WT Aβ42 and WT Aβ40 under conditions reflecting physiological concentration differences of monomeric Aβ isoforms (i.e., [Aβ40] / [Aβ42] ~ 9). These results are consistent with previous results from immunohistochemical studies showing that more Aβ species carrying Aβ(x–40) and Aβ(1–5) as epitopes are present in the core of amyloid plaques from FAD patients with the E22G Arctic mutation and that almost no Aβ(x–42) was identified in the core¹⁸. The meaning of this clinical evidence is difficult to comprehend; hence it was interpreted as a lack of amyloid in the core. However, our current results indicate that E22G Aβ40 is preferably assembled into the core of amyloid plaques for patients with the Arctic mutation. Low thioflavin-S affinity of cotton-wool plaques is also consistent with the low ThT fluorescence observed for the E22G Aβ40 fibrils, and likewise the low cortex PET signal from ¹¹C-labeled Pittsburgh compound, an analog of ThT⁴⁷. Dispersed and less compact fibril features observed in this study also can explain the larger plaque size and low staining by Congo Red observed for cotton-wool plaques of FAD patients with the Arctic mutation. Therefore, our results provide in-vitro indication that the Arctic mutation may switch the driving force of AD onset to Aβ40 from Aβ42. To our surprise, the various features of E22G Aβ40 fibrils are consistent with observations for the amyloid core of cotton-wool plaques in FAD patients. Since our fibril established in the bottom-up approach did not involve any brain materials, the results provide strong evidence that the distinctive Aβ fibril structure is likely responsible for many of the features of the cotton-wool plaques, which define distinct pathological features of the Arctic FAD, a notable subtype of AD.

Thirdly, the E22G Aβ40 fibril structure revealed that the main-chain arrangements following G22 and near the V40 terminus are likely to accommodate the association of monomeric Aβ42 and Aβ40 via cross-seeding. Indeed, our kinetic studies revealed that E22G Aβ40 fibrils adopting the extended W-shape conformation promote fibrilization of WT Aβ42 and Aβ40 via cross-seeding. It has been shown that WT Aβ42 fibrils do not cross-seed Aβ40 aggregation due to the inability of Aβ40 monomers to template the structure of S-shaped Aβ42 fibrils^25,28,42. However, the extended conformation of E22G Aβ40 fibrils revealed herein explains why the structure of E22G Aβ40 fibrils can promote fibrilization of both WT Aβ40 and Aβ42 monomers. The results were confirmed by our preliminary MD simulations for chimera fibrils comprised of E22G Aβ40 and WT Aβ40/Aβ42. These kinetic and MD results are the initial indication that the atomic structure of Aβ fibril alters the compatibility of cross-propagation with FAD mutation. The cross-seeding mechanism provides a reasonable explanation for the clinical finding that Aβ species with the Aβ(x–40) epitope are distributed homogeneously over the core of cotton-wool plaques, but those with the Aβ(x–42) epitope are distributed in a ring-like structure with a void in the core. It is also noteworthy that similar cotton-wool plaques have also been reported for late-onset sporadic AD with mild to marked CAA²². The Aβ40-rich core and ring-like accumulation of Aβ42 were also reported for cotton-wool plaques in AD cases that were not linked with known mutations²⁹. Although more studies are needed to establish evidence for brain-derived systems, our structural and kinetic analysis of monomorphic E22G Aβ40 fibrils established in the bottom-up approach presents mechanistic evidence that the mutation associated with FAD can modulate plaque formation pathways and patterns via cross-seeding.

Finally, co-analysis by cryoEM and SSNMR was found to be the most effective approach for elucidating the structural features of E22G Aβ40 amyloid fibrils at the atomic scale compared with other biophysical techniques or cryoEM/SSNMR alone. Even at near-atomic overall resolution, the cryoEM densities sometimes involve poorly defined regions, for which additional structural information may be needed to elucidate the correct target structures. In the case of the current analysis, the cryoEM map and the local resolution estimation suggest that the outer layers and the terminal parts of the core layers show higher flexibility than the central part of the core layer. It is also difficult to elucidate the extended W-shaped motif with substantial interlayer contacts by SSNMR alone. Our results demonstrate that co-analysis by cryoEM and SSNMR provides highly reliable results for atomic-level structural analysis of amyloid fibril systems and for analyzing subtypes of AD at molecular levels.

Methods

Materials

N,N-dimethylformamide (DMF, for peptide synthesis, Cat. No. 047-25453), dichloromethane (DCM, for peptide synthesis, Cat. No. 044-25583), piperidine (for peptide synthesis, Cat. No. 161-18811), N,N-diisopropylethylamine (DIEA, for peptide synthesis, Cat. No. 054-06321), and Congo red (Cat. No. 039-23192) were purchased from Fujifilm Wako Pure Chemical (Osaka, Japan). Trifluoroacetic acid (TFA, 99%, Cat. No. T6508), 1,2-ethanedithiol (EDT, ≥98.0%, Cat. No. 8.00795), thioanisole (≥99.0%, Cat. No. T28002), phenol (for molecular biology, ≥99.0%, Cat. No. P1037), dimethyl sulfoxide (DMSO, anhydrous ≥99.9%, Cat. No. 276855), glycine (≥99%, Cat. No. G7126), sodium chloride (≥99.5%, Cat. No. 28-2270), sodium dihydrogen phosphate (≥99%, Cat. No. 28-3809), sodium phosphate dibasic (anhydrous ≥99%, Cat. No. 795410), thioflavin T (ThT, Cat. No. T3516), sodium hydroxide (anhydrous, ≥98%, Cat. No. S5881) sodium azide (≥99.5%, Cat. No. S2002), potassium chloride (anhydrous, ≥99%, Cat. No. 746436), magnesium sulfate (anhydrous, ≥99.5%, Cat. No. 746452), calcium chloride (anhydrous, ≥97%, Cat. No. 746495), and D-(+)-glucose (≥98%, Cat. No. 07-0680) were purchased from Sigma-Aldrich (St. Louis, U.S.A). 2-[4-(2-hydroxyethyl)-1-piperazinyl]ethanesulfonic acid (HEPES, Cat. No. 340-01371) was purchased from Dojindo (Kumamoto, Japan). Acetonitrile (ACN, for high-pressure liquid chromatography (HPLC), ≥99.9%, Cat. No. 01033-7B), 2-propanol (for HPLC, ≥99.7%, Cat. No. 32435-2B), methanol (for HPLC, ≥99.8%, Cat. No. 25183-2B), and diethyl ether (dehydrated, for organic synthesis, ≥99.5%, Cat. No. 14547-05) were purchased from Kanto Chemical Corporation (Tokyo, Japan). Acetic anhydride (≥99.0%, Cat. No. A2036) and 1,1,1,3,3,3-hexafluoro-2-propanol (HFIP, ≥99.0%, Cat. No. H0424) were purchased from Tokyo Chemical Industry (Tokyo, Japan). O-(1H-6-chlorobenzotriazole-1-yl)-1,1,3,3-tetramethyluronium hexa-fluorophosphate (HCTU, Cat. No. 8.51012), Fmoc-amino acids (Novabiochem tradename), Fmoc-Val-NovaSyn TGT resin (Cat. No. 8.56051), and Fmoc-Ala-NovaSyn TGT resin (Cat. No. 8.56125) were purchased from Merck Millipore (Burlington, U.S.A). Isotope labeled Fmoc-amino acids were purchased from Sigma-Aldrich (u-¹³C, 98% and u-¹⁵N, 98%) and Cambridge Isotope Laboratories (Tewksbury, U.S.A; u-¹³C, 99% and u-¹⁵N, 99%). Hydrochloric acid (6 N, Cat. No. 18321) and copper (II) disodium ethylenediaminetetraacetate (Cu-EDTA, anhydrous, Cat. No. 15119-52) were purchased from Nacalai Tesque (Kyoto, Japan). Methylamine tungstate negative stain (Nano-W, 2% solution in water, Cat. No. 2018) was purchased from Nanoprobes (Yaphank, U.S.A). α-cyano-4-hydroxycinnamic acid (CHCA) MALDI matrix (ultrapure, single-use, Cat. No. 90031) was purchased from Thermo Fisher Scientific (Waltham U.S.A). E22G Aβ42 peptide (Cat. No. 4035371) was purchased from Bachem (Bubendorf, Switzerland) and used without further purification. All aqueous solutions were prepared with Milli-Q (MQ) water purified with Merck Millipore Ultrapure (Milli-Q) system.

Sample preparation

All the Aβ peptides utilized in this work, except for E22G Aβ42 which was purchased from Bachem (Bubendorf, Switzerland), were synthesized in our lab with a PTI PurePep Chorus peptide synthesizer (Gyros Protein Technologies, Tucson, U.S.A) using standard Fmoc chemistry. Monomeric Aβs were prepared as previously described^28,42.

All E22G Aβ40 monomers for this study were first dissolved in anhydrous DMSO to a concentration of 3 mM for 15–20 min at ambient temperature. The solution was then diluted to a target concentration of 60 μM at pH 7.4 with a 10.5 mM phosphate buffer containing 0.02 % NaN₃ and 5 mM NaCl (referred to as PB) on ice. Then, it was filtered through an Amicon Ultra-4 centrifugal filter with a 50-kDa molecular-weight-cut-off (Millipore) at 4000 × g for 5–15 min (depending on the volume of the sample solution) at 4 °C using a Beckman Coulter Allegra X-30R centrifuge. The filtrate concentration was determined by UV-VIS absorbance at 280 nm with a molar absorption coefficient of 1490 M^–1 cm^–1 by Hitachi U-3900 UV-VIS spectrophotometer (Hitachi High-tech Science, Tokyo, Japan). The measured concentration was typically 35–45 μM. The concentration was then adjusted to 30 μM, and the final concentration was evaluated ranging from 25 to 35 μM. For kinetics studies using ThT fluorescence assay (Fig. 7b, c and Supplementary Fig. 18), WT Aβs were first dissolved in 10 mM NaOH solution to a concentration of 600 µM, instead of DMSO as previously described^28,42, and then the Aβ solution was diluted with PB as mentioned above. The final concentration was ~45 µM. The quiescent seeded E22G fibril was added to the WT Aβ monomers at 10% (w/w). The seed E22G fibril was sonicated using a Branson SFX 150 homogenizer, as described below. For kinetic studies with ThT fluorescence assay in Supplementary Figs. 16 and 17, Aβ monomers were prepared in the same manner with the following modifications. For experiments in Supplementary Fig. 16, PB was used to dilute the peptides, while for experiments in Supplementary Fig. 17, extracellular buffer⁵⁹ (20 mM HEPES, 140 mM NaCl, 5 mM KCl, 1 mM MgSO₄, 1 mM CaCl₂, 1 mM NaH₂PO₄, 5 mM D-(+)-glucose, 0.02% NaN₃, pH 7.5) was used instead of PB. E22G Aβ42 monomers were dissolved in anhydrous DMSO at a concentration of 0.5 mM instead of 3 mM and then diluted to the target concentration of 20 μM at pH 7.4 with PB or extracellular buffer. The final concentration after filtration was typically 25–45 µM for E22G Aβ40, WT Aβ40, and WT Aβ42 and 2–5 µM for E22G Aβ42. The peptide solution was diluted with the respective buffers (PB or extracellular buffer) to the final concentration (see captions). For the cross-seeding experiments in Supplementary Figs. 16 and 17, the quiescent seeded E22G fibril was added to the WT Aβ monomers at 10% (w/w). The seed E22G fibril was sonicated using a Branson SFX 150 homogenizer, as described below.

The monomer solutions were then incubated for 7 days to produce mature fibrils. Agitated fibrils were incubated at ~37 °C on a Multi Bio RS-24 shaker (Funakoshi, Japan) with 7 rpm rotation. Quiescent fibrils were incubated at ~37 °C without agitation. The samples prepared without addition of seed fibrils were denoted as G0.

To prepare seeded fibril samples, mature G0 fibrils were fragmented using a Branson SFX 150 homogenizer for a total time of ~5 min at a duty factor of 50% and a power of 60% or using a Branson CPX 2800 sonic bath in ice-water for ~5 min. Fragmented seed fibrils were then added to fresh Aβ monomer solutions at 5% (w/w) and incubated for 3–4 days in the same manner described above. This sample was denoted as G1. Successive seeded samples, denoted as G2 to Gn, were prepared in the same way using fibrils from the previous generation G(n − 1) as seed, according to the previously published protocol²⁸. E22G Aβ40 fibril sample used for cryoEM and SEM analysis was the quiescent G5 sample in which homogenizer was utilized for seeding.

For SSNMR studies, E22G Aβ40 monomers uniformly ¹³C- and ¹⁵N-labeled at selected residues were seeded with G5 samples. The fibrils were harvested by centrifugation at 4000 × g for 30 min for agitated samples and at 10,000 × g for 90 min for quiescent samples using a Beckman Coulter Allegra X-30R centrifuge (Beckman Coulter, Brea, U.S.A). The supernatant was then removed, and the pellet was lyophilized.

MALDI-TOF mass spectrometry

Nanogram quantities of Aβ peptide was dissolved in 50% ACN solution containing 0.1% TFA and mixed with a MALDI matrix solution (10 mg CHCA in 100 μL 50% ACN solution containing 0.1% TFA) at 1:1 (v/v). The sample mixture (1 μL) was spotted onto a standard MTP 384 target plate ground steel BC and air-dried. The samples were analyzed with a Bruker Ultraflex MALDI TOF/TOF mass spectrometer (Bruker, Billerica, U.S.A). All Aβ peptides prepared and used in this work showed purity ≥90%, except for E22G Aβ42 (>80%), which was purchased from Bachem (Bubendorf, Switzerland).

ThT fluorescence spectroscopy

Misfolding kinetics of E22G Aβ40 monomers was monitored by incubation-time dependence of ThT fluorescence as described previously^28,42 with minor modifications. For the kinetic data in Fig. 1k, l and Supplementary Figs. 1a, 2a and 4, a 10-μL aliquot of a sample solution was mixed with 10 μL of 500 μM ThT and 980 μL of 50 mM glycine buffer containing 4 mM NaOH was measured on a Hitachi F-7100 fluorescence spectrophotometer in triplicates (N = 3). The fluorescence values were measured for three distinctive samples at each time point in order to obtain the average value and the standard deviation. Curve fittings including the error estimation were performed by LMFIT⁶⁰ using sigmoidal equation for unseeded samples and first-order kinetics equation for seeded samples, based on the methods previously described^28,42.

The kinetics data in Fig. 7a–c and Supplementary Figs. 16–18 were acquired on a BMG Labtech FLUOstar Omega plate reader (BMG Labtech, Ortenberg, Germany) as demonstrated previously⁴². The Aβ samples were loaded on Greiner 96-well flat-bottom plates (Greiner Bio-one, Kremsmünster, Austria) and incubated at 37 °C with ThT so that the final ThT concentration was 30 μM. For agitated experiments, double orbital shaking for 5 min at 700 rpm was used after each measurement taken every 15 min. For quiescent experiments, no shaking was applied. The curve fitting was performed as described above. To estimate the lag time for sigmoidal curves, the average values of the lag times for the 4 trials were calculated with their standard deviation. For Fig. 7a–c and Supplementary Figs. 16 and 18, the Aβ monomer samples were incubated in a 10.5 mM phosphate buffer containing 0.02 % NaN₃ and 5 mM NaCl. For Supplementary Fig. 17, the Aβ monomer samples were incubated in a high-salt extracellular buffer that mimics an extra-cellular solution. The compositions of the extracellular buffer are 20 mM HEPES, 140 mM NaCl, 5 mM KCl, 1 mM MgSO₄, 1 mM CaCl₂, 1 mM NaH₂PO₄, 5 mM D-(+)-glucose, 0.02% NaN₃, pH 7.5, as described above.

TEM analysis

TEM images were captured on a JEOL 1400 Plus electron microscope operated at an acceleration voltage of 80 kV with a JEOL 14810 Ruby CCD camera. Samples for TEM analysis were prepared on formvar carbon supported copper grids with mesh count of 200 (EM Japan). For each sample, 5 μL of mature fibril sample was deposited on the grid. Then, 5 μL of Nano-W solution was added immediately on the grid and mixed by gentle pipetting. Then, the grid was blotted with a filter paper after 1.5 min and air-dried for several minutes. The grids were stored in a dry box until it was used.

SSNMR spectroscopy

To perform 2D ¹³C-¹³C correlation SSNMR experiments using magic-angle spinning (MAS), a 400 MHz NMR system equipped with Bruker Avance NEO spectrometer at Tokyo Institute of Technology was used with a Bruker 2.5-mm ¹H, ¹³C, ¹⁵N triple-resonance CPMAS probe. To perform 2D ¹³C-¹H correlation SSNMR experiments, a high-field 900 MHz NMR system equipped with a JEOL Resonance JNM-ECZR spectrometer at RIKEN NMR facilities was employed with a JEOL 1.0-mm ¹H, ¹³C, ¹⁵N triple-resonance CPMAS probe. Lyophilized E22G Aβ40 fibril samples (2–5 mg) were packed into 2.5-mm zirconia rotors for 2D ¹³C-¹³C correlation experiments, while 0.3–0.5 mg of the lyophilized fibril samples were packed into 1.0-mm zirconia rotors for ¹³C-¹H correlation experiments. All the samples were hydrated with 200 mM Cu-EDTA solution at ~1 μL per mg of the fibril sample. MAS rates in all the ¹³C-¹³C correlation experiments were set to 20,000 ± 5 Hz while MAS rates of 70,000 ± 5 Hz were employed for all the 2D¹³C-¹H correlation experiments. The sample temperature was kept at 20–30 °C by applying VT air temperature of −10 °C and −37 °C in ¹³C-¹³C and ¹³C-¹H correlation experiments, respectively.

During the cross-polarization (CP) period in all 2D ¹³C-¹³C correlation experiments, ¹H RF amplitude was linearly swept from 100 to 80% at the average radio-frequency (RF) amplitude of ~65 kHz, while ¹³C RF amplitude was kept constant at ~47 kHz with a contact time of 1.5–1.8 ms. Dipolar-assisted rotational resonance (DARR) sequence⁵³ was used for ¹³C-¹³C mixing to collect 2D ¹³C-¹³C correlation SSNMR spectra in Figs. 2 and 5 and Supplementary Figs. 3, 7 and 13. A constant amplitude ¹H RF at 20 kHz, which was matched with the MAS rate, was applied during the ¹³C–¹³C mixing period. Acquisition time of 10.24 ms was used in the direct dimension and 256 t₁ complex points with a t₁ increment of 50 μs were collected in the indirect dimension. A SPINAL-64 ¹H decoupling sequence at 100 kHz was implemented during the acquisition time and the t₁ period. The mixing time of 50 ms was used to observe short-range ¹³C-¹³C dipolar correlations while 200 or 300 ms was used for long-range contacts. To confirm reproducibility for the measured signal intensity ratios of the major and minor species summarized in Supplementary Table 3, we separately collected 2D ¹³C-¹³C correlation SSNMR spectra with ¹³C-¹³C mixing sequence using finite-pulse radio-frequency-driven recoupling (fpRFDR) sequence⁶¹ on the E22G Aβ fibril sample that was uniformly ¹³C, ¹⁵N-labeled at D23, S26, K28, I32, and M35. The spectral intensity ratios were largely consistent with those from the DARR data used for Supplementary Table 3. The fpRFDR sequence was performed with mixing time of 1.6 ms using ¹³C RF amplitude of 33 kHz for each π-pulse. The same acquisition parameters described above were also applied during fpRFDR experiments. The natural abundance ¹³CH₂ signal of adamantane was used as a secondary reference (40.49 ppm) to reference ¹³C chemical shifts to 3-(trimethylsilyl)-1-propanesulfonate (DSS, 5% in D₂O).

For 2D ¹³C-¹H correlation SSNMR experiments, ¹H-detected ¹³C/¹H double CP sequence was employed using the pulse sequence in ref. ⁶². Acquisition time of 10.24 ms was used in the direct dimension and 64 t₁ complex points with a t₁ increment of 63.16 μs were collected in the indirect dimension. During the first CP transfer (¹H to ¹³C), ¹H RF amplitude wase kept constant at 19 kHz, while ¹³C RF amplitude was swept adiabatically following an adiabatic passage Hartman-Hahn at an average amplitude of 53 kHz, and the contact time was 500 μs. For the second CP transfer (¹³C to ¹H), ¹H and ¹³C RF amplitudes of 22 and average of 51 kHz were applied with the same shapes mentioned above for the contact time of 800 μs⁶². A variable-length SLAP sequence⁶² at 10 kHz was implemented for the duration of 60 ms to suppress solvent ¹H signals. Here, a SPINAL-64 low-power ¹H decoupling sequence at an RF amplitude of 12 kHz was used during the t₁ period and a WALTZ-65 ¹³C decoupling sequence at an RF amplitude of 18 kHz was used during the acquisition period⁶².

All 2D SSNMR data were processed with NMRPipe⁶³. In all ¹³C-¹³C correlation spectra, the data were apodized with a Lorenz-to-Gauss function with an inverse exponential narrowing (IEN) width of 30 Hz and Gaussian broadening (GB) width of 100 Hz in the t₂ domain. The location of Gaussian function’s maximum was set to 0.1 times FID. Lorenz-to-Gauss function with IEN width of 30 Hz and GB width of 120 Hz was applied in the t₁ domain and the first point of FID was scaled to 0.5 before Fourier transform. For all ¹³C-¹H correlation spectra, the data were apodized by applying an exponential function with line broadening (LB) width of 100 Hz in the t₂ domain. Lorenz-to-Gauss function with IEN width of 30 Hz and GB width of 120 Hz was used in the t₁ domain and the first point of FID was scaled to 0.5 before Fourier transform. All the ¹³C assignments in Supplementary Tables 1 and 2 were implemented manually.

Calculations

Pearson’s correlation coefficients for the chemical shift analyses in Supplementary Figs. 4, 6 and 15 were calculated by Python using SciPy package (v. 1.12.0). ShiftX2⁵¹ was utilized to calculate backbone ¹³C chemical shifts (Fig. 4 and Supplementary Fig. 15b) from the atomic coordinates.

CryoEM analysis

A 3.0-µL aliquot of the sample solution was applied on a glow-discharged holey Quantifoil R1.2/1.3 grid with 300 mesh copper. The grid was blotted for 3 s by filter paper at 4 °C and 100% relative humidity, and plunge-frozen in liquid ethane by using Vitrobot Mark IV from Thermo Fisher Scientific. CryoEM images were acquired by using Thermo Fisher Scientific Titan Krios, equipped with a Gatan K3-Summit detector, with an acceleration voltage of 300 kV and slit width of 15 Å at RIKEN Center for Biosystems Dynamics Research. A total of 2500 movies of 50 frames were acquired at a pixel size of 0.8285 Å and a total dose of 50 electrons per Å² with the defocus values in the range of −0.5 and −2.5 µm.

Acquired movie frames were motion corrected and dose-weighted by using MotionCor2⁶⁴. CTFFIND 4.1⁶⁵ was used to estimate the contrast transfer function (CTF). Helical reconstruction was performed by using RELION 3.1⁶⁶. Fibril segments were picked manually and extracted with an inter-box distance of 10% (24.8 Å) and box size of 300 pixels (248 Å). 478,295 fibril segments were extracted from 36,474 fibrils, then downscaled to 150-pixel boxes. Reference-free 2D classification was performed on CTF-corrected images by applying mask diameter of 200 Å and regularization value of T = 4. 2D class averages (258,228 particles) showing the clear rungs were selected and used for reconstruction. An initial 3D reference encompassing the cross-over distance was reconstructed de novo from multiple 2D class averages with relion_helix_inimodel2d, using the cross-over distance of 350 Å in the micrographs. Then, the initial 3D reference map was low-pass filtered to 60 Å and subjected to a 3D classification. By assuming helical rise of 4.75 Å, a helical twist of 2° was estimated from the cross-over distance of approximately 350 Å. First round of 3D classification was performed with K = 3 and T = 10. The best class (151,818 particles) was re-extracted with the original pixels. Second round of 3D classification was performed with K = 1 and T = 10 without image alignment. The resulting map was low-pass filtered to 15 Å and auto-refined with C₁ symmetry. Starting from the obtained cryo-EM map, we tested C1, C2, and pseudo-C21 symmetry (Supplementary Fig. 9). First, we performed 3D auto-refinement followed by Postprocessing with C1 symmetry, providing the high-quality cryo-EM map with a resolution of 2.86 Å. Second, the cryo-EM map with C2 symmetry was reconstructed to 2.86 Å resolution by the same procedure except for an imposed C2 symmetry. The map showed a similar quality to that with the C1 symmetry, but the cryo-EM densities for the lateral layers appeared to be blurrier. Third, we reconstructed the cryo-EM map with pseudo-C21 symmetry giving a rise value of 2.38 Å and a twist angle of 178.935° (left-handed model). The reconstructed cryo-EM map resulted in a similar quality (2.89 Å resolution) but did not significantly improve compared with the C1-symmetry map. As far as the cryo-EM densities were not improved by symmetry operation, we decided not to impose the symmetry operation to prevent induced artificial effects. CTF refinement was carried out with fitting of defocus set per-particle and fitting of astigmatism set per-micrograph. Post-processing was performed using a mask with the low-pass filter of 15 Å and an auto-estimation of the B-factor with the original pixel size. Bayesian polishing was performed, which was followed by auto-refinement with 8 Å low-pass filter. The final map was inverted by relion_image_handler to make it left-handed, and the final helical twist and rise were optimized to −2.17°and 4.77 Å, respectively. The handedness of the reconstruction was determined in comparison with the SEM images, which showed left-handed twist for the observed fibrils.

An initial model of E22G Aβ40 molecules in one of the cross-β layers was manually constructed in COOT⁶⁷. Amino acids with large (e.g., Ile, Leu, Lys, Met, Phe) and small (Gly) sidechains were used to align E22G Aβ40 molecules. Five copies of the refined molecules were generated to produce 5-layered model. Each layer was manually translated and rotated to roughly fit into the map. The density map was extracted around the fragment of the fibril consisting of 5 layers with phenix.map_box in PHENIX 1.19.1⁶⁸. The fragment of the fibril was refined in real space by applying phenix.real_space_refine. Steric clashes and Ramachandran and rotamer outliers were corrected manually, and the model was re-refined in PHENIX 1.19.1. The final model evaluation was done by using MolProbity⁶⁹. Statistics are shown in Supplementary Table 6.

SEM analysis

A 3.0-µL aliquot of the sample solution was applied on glow-discharged holey Quantifoil R1.2/1.3 grid with 300 mesh copper for 1 min and it was washed with 50 µL MQ water twice. The grids were blotted with filter paper to remove excess solution then air-dried. The grids were fixed on a glass plate with the carbon paste and osmium-coated with HPC-1SW osmium coater from Vacuum Device for 1.2 s. SEM images of amyloid fibrils were obtained using HITACHI SU-8220 field emission SEM with a secondary electron detector at an acceleration voltage of 10 kV and magnification of ×90,000.

Visualization

Aβ fibrils were illustrated by VMD⁷⁰ (Figs. 3a, c and 6 and Supplementary Figs. 14c, 20, 22c–f and 23), COOT⁶⁷ (Fig. 3b and Supplementary Fig. 11a), ChimeraX⁷¹ (Fig. 3d, e and Supplementary Fig. 11b).

MD simulations

To examine the stability of W-shaped and S-shaped Aβ fibrils, 500-ns molecular dynamics (MD) simulations were performed using AMBER22 package³⁶ with Amber22 Tools³⁷, accelerated by GPUs on TSUBAME 3 (supercomputer facility at Tokyo Institute of Technology using pmemd). Each node of TSUBAME 3 provides four NVIDIA Tesla P100 GPUs with CUDA11 middleware, which accelerate MD simulation using Amber22 PMEMD.cuda. PMEMD.cuda is also supported by SPFP (Single Precision Fixed Precision hybrid) method to minimize errors when a longer time step (2.0 fs) is employed for long MD simulations (100–1000 ns). In rare cases that simulations encounter GPU acceleration errors, MPI parallel processing calculations by pmemd. MPI with Intel Xeon E5-2680 V4 CPU 28 T (14-core) were performed. As the initial structures for the W-shaped Aβ fibril, the structure determined for the E22G Aβ40 fibril in this work was adopted with 10 pairs of the central core Aβ molecules (i.e., a total of 20 Aβ molecules). To construct W-shaped chimeric fibrils comprised of 10 E22G Aβ40 molecules and 10 WT Aβ40 or Aβ42 molecules, 5 successive pairs of E22G Aβ40 at one side (the side with smaller chain numbers in pdb) were replaced by WT Aβ40/WT Aβ42. For the E22G mutatant, all hydrogen atoms were reconstructed using the automated completion functionality for missing atoms in LEaP. For reverting the E22G mutation, the structure on residue 22 was also reconstructed in LEaP, to change the mutated Gly back to Glu on residue 22. To generate WT Aβ42 in the chimera fibril, the last two residues for WT Aβ42 were added by the LEaP program. To construct S-shaped chimeric fibrils comprised of 7 WT Aβ42 molecules and 7 WT Aβ40 or Aβ42 molecules, 7 successive pairs of WT Aβ42 at one side (the side with larger chain numbers in pdb) were replaced by WT Aβ40. For the replacement to Aβ40, all atoms in the residues 41 and 42 were removed. All hydrogen atoms in the S-shaped chimeric fibrils were also reconstructed in LEaP.

For all the explicit water MD simulations, the W-shaped fibrils were solvated with TIP3P rigid-body water in a solvation box of (153.186 Å, 103.391 Å, 103.216 Å) using periodic boundary conditions. Charge-neutralized counter ions (20 Cl^– for the E22G Aβ40 fibril and 10 Cl^– for the chimera fibrils), and 39 Na⁺ and Cl^̶ ions (assuming the salt concentration ~40 mM) were introduced in each solventbox. In the constructed molecular system, the total number of water molecules was 43,631–43,738. The S-shaped fibrils were solvated with TIP3P rigid-body water in a solvation box of (103.148 Å, 99.368 Å, 123.177 Å) using periodic boundary conditions. Charge-neutralized counter ions (14 Na⁺ for the chimera fibrils), and 30 Na⁺ and Cl^̶ ions (assuming the salt concentration ~40 mM) were introduced in each solventbox. In the constructed molecular system, the total number of water molecules was 33,310. The MD simulations were performed with Amber ff14SB force field. Particle Mesh Ewald (PME) method was applied to handle the long-range electrostatic interactions and the non-bonded cut-off was set to 10 Å for all calculations, and the SHAKE algorithm was used to constrain geometries of all covalent bonds between heavy atom and hydrogen atoms.

The MD simulation procedure is composed of the following 7 stages: (1) Initial minimization stage, (2) NVT heating stage, (3) system minimization stage, (4) system heating stage by NVT stage, (5) water relaxation stage by NPT, (6) system equilibration stage by NPT, and (7) production stage by NPT. Only stages 1 and 2 were performed to water atoms by position restriction of protein and ionic atoms, so as to remove unrealistic atomic collisions and vacuum sites inside of the protein. In stages 4 and 5, only protein atomic positions were restricted to avoid significant changes in the global structure before starting the MD simulation without restrictions in stage 6. In stage 5, MD simulation was performed to stabilize the whole system by switching NVT to NPT mode while making the density of water molecules realistic (~0.995 g/cm³, 300 K, 1 atom). In stage 6, the system was equilibrated after the restrictions of the atomic position were released. All of the weights used for restriction of atomic position were set to 10.0 kcal/mol at all residues of protein molecule in stages 4 and 5, while ionic atoms were restricted in the same manner in stages 1 and 2.

In all the minimization stages, 2000 steps using the steepest descent method were followed by 18,000 steps with the conjugate gradient method. For the NPT calculations, the pressure control by the Barostat method was set to 1.0 bar. The NPT calculations were performed for 0.5 ns, 5 ns and 500 ns, respectively at stages 5, 6, and 7. To obtain statistical trajectory data, the five MD simulations were run with different random seeds but using the identical protocol for each system. For an analysis of the MD trajectories, cpptraj in the Amber22 Tools was employed³⁸. The RMSD values in Supplementary Figs. 19 and 22 and extended β-sheet fractions in Supplementary Fig. 21 were calculated from the trajectories with cpptraj.

To estimate the motional fluctuation for the W-shaped Aβ fibril in Supplementary Fig. 23, the root-mean-square fluctuation (RMSF) values for the C_α atom of each residue were calculated as follows. To estimate the RMSF values, the average structure of trajectories in the production simulation is usually used as the reference structure. However, since the model used here involved considerable fluctuations in the N- and C-terminal regions, the core part with low RMSF values was selected by calculating the average structure and RMS fitting to the average coordinates using the following procedure. (i) First, we set the target residues for RMS fitting to all the residues. The first 20 ns of the trajectory was eliminated, and the production initial frame structure after the elimination (i.e., the structure at 20 ns) was set as the reference structure. (ii) Then, we performed RMS fitting of the trajectory using the reference structure. The average structure 1 was then calculated. (iii) Next, we performed RMS fitting of the trajectory to the average structure 1. Then, the average structure 2 was calculated, and we performed RMS fitting of the trajectory to average structure 2. (iv) Next, we calculated the RMSF value of C_α for all residues and we set the target residues for RMS fitting to the residues for which the RMSF value obtained in the previous step was less than 2 Å. (v) We repeated steps (iii) to (iv) until the RMSF value converged. (vi) If the RMSF values in (v) converged, we calculated the average value of the RMSF values for each residue in Trials 1 to 5. (vii) Residues for which the average value of the RMSF values determined in (vi) is less than 0.8 Å for all isoforms are defined as the core residues.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Data availability

The cryo-EM maps have been deposited into the Electron Microscopy Data Bank under accession numbers EMD-35972. The coordinates have been in the RCSB Protein Data Bank (PDB) accession codes 8J47. MD simulation input files are available as Supplementary Data 1. Source data are provided with this paper.

References

Selkoe, D. J. Cell biology of protein misfolding: the examples of Alzheimer’s and Parkinson’s diseases. Nat. Cell. Biol. 6, 1054–1061 (2004).
Article CAS PubMed Google Scholar
Savonenko, A. V., Wong, P. C. & Li, T. Alzheimer diseases. in Neurobiology of Brain Disorders, (eds Zigmond, M. J., Wiley, C. A. & Chesselet, M.-F.) 2nd Ed., 313–336 (Elsevier, 2023).
Kuperstein, I. et al. Neurotoxicity of Alzheimer’s disease A beta peptides is induced by small changes in the A beta(42) to A beta(40) ratio. EMBO J. 29, 3408–3420 (2010).
Article CAS PubMed PubMed Central Google Scholar
Cohen, M. L. et al. Rapidly progressive Alzheimer’s disease features distinct structures of amyloid-β. Brain 138, 1009–1022 (2015).
Article PubMed PubMed Central Google Scholar
Rasmussen, J. et al. Amyloid polymorphisms constitute distinct clouds of conformational variants in different etiological subtypes of Alzheimer’s disease. Proc. Natl Acad. Sci. USA 114, 13018–13023 (2017).
Article ADS CAS PubMed PubMed Central Google Scholar
Di Fede, G. et al. Molecular subtypes of Alzheimer’s disease. Sci. Rep. 8, 3269 (2018).
Article ADS PubMed PubMed Central Google Scholar
Condello, C. et al. Structural heterogeneity and intersubject variability of Aβ in familial and sporadic Alzheimer’s disease. Proc. Natl Acad. Sci. USA 115, E782–E791 (2018).
Article CAS PubMed PubMed Central Google Scholar
Cacace, R., Sleegers, K. & Van Broeckhoven, C. Molecular genetics of early-onset Alzheimer’s disease revisited. Alzheimers Dement. 12, 733–748 (2016).
Article PubMed Google Scholar
Nilsberth, C. et al. The ‘Arctic’ APP mutation (E693G) causes Alzheimer’s disease by enhanced Abeta protofibril formation. Nat. Neurosci. 4, 887–893 (2001).
Article CAS PubMed Google Scholar
Norlin, N. et al. Aggregation and fibril morphology of the Arctic mutation of Alzheimer’s Aβ peptide by CD, TEM, STEM and in situ AFM. J. Struct. Biol. 180, 174–189 (2012).
Article CAS PubMed PubMed Central Google Scholar
Revesz, T. et al. Genetics and molecular pathogenesis of sporadic and hereditary cerebral amyloid angiopathies. Acta Neuropathol. 118, 115–130 (2009).
Article CAS PubMed PubMed Central Google Scholar
Biffi, A. & Greenberg, S. M. Cerebral amyloid angiopathy: a systematic review. J. Clin. Neurol. 7, 1–9 (2011).
Article PubMed PubMed Central Google Scholar
Xu, F. et al. Cerebral vascular amyloid seeds drive amyloid β-protein fibril assembly with a distinct anti-parallel structure. Nat. Commun. 7, 13527 (2016).
Article ADS CAS PubMed PubMed Central Google Scholar
Lashuel, H. A. et al. Mixtures of wild-type and a pathogenic (E22G) form of Abeta40 in vitro accumulate protofibrils, including amyloid pores. J. Mol. Biol. 332, 795–808 (2003).
Article CAS PubMed Google Scholar
Zhang-Nunes, S. X. et al. The cerebral beta-amyloid angiopathies: hereditary and sporadic. Brain Pathol. 16, 30–39 (2006).
Article CAS PubMed PubMed Central Google Scholar
Basun, H. et al. Clinical and neuropathological features of the arctic APP gene mutation causing early-onset Alzheimer disease. Arch. Neurol. 65, 499–505 (2008).
Article ADS PubMed PubMed Central Google Scholar
Philipson, O. et al. The Arctic amyloid-β precursor protein (AβPP) mutation results in distinct plaques and accumulation of N- and C-truncated Aβ. Neurobiol. Aging 33, 1010.e1–13 (2012).
Article CAS PubMed Google Scholar
Kalimo, H. et al. The Arctic AβPP mutation leads to Alzheimer’s disease pathology with highly variable topographic deposition of differentially truncated Aβ. Acta Neuropathol. Commun. 1, 60 (2013).
Article PubMed PubMed Central Google Scholar
Lemoine, L. et al. Amyloid, tau, and astrocyte pathology in autosomal-dominant Alzheimer’s disease variants: AβPParc and PSEN1DE9. Mol. Psychiatry 26, 5609–5619 (2021).
Article CAS PubMed Google Scholar
Selkoe, D. J. Alzheimer’s disease: genes, proteins, and therapy. Physiol. Rev. 81, 741–766 (2001).
Article CAS PubMed Google Scholar
Boon, B. D. C. et al. The coarse-grained plaque: a divergent Aβ plaque-type in early-onset Alzheimer’s disease. Acta Neuropathol. 140, 811–830 (2020).
Article CAS PubMed PubMed Central Google Scholar
Le, T. V., Crook, R., Hardy, J. & Dickson, D. W. Cotton wool plaques in non-familial late-onset Alzheimer disease. J. Neuropathol. Exp. Neurol. 60, 1051–1061 (2001).
Article CAS PubMed Google Scholar
Petkova, A. T. et al. A structural model for Alzheimer’s beta-amyloid fibrils based on experimental constraints from solid state NMR. Proc. Natl Acad. Sci. USA 99, 16742–16747 (2002).
Article ADS CAS PubMed PubMed Central Google Scholar
Petkova, A. T., Yau, W.-M. & Tycko, R. Experimental constraints on quaternary structure in Alzheimer’s beta-amyloid fibrils. Biochemistry 45, 498–512 (2006).
Article CAS PubMed Google Scholar
Paravastu, A. K., Leapman, R. D., Yau, W.-M. & Tycko, R. Molecular structural basis for polymorphism in Alzheimer’s beta-amyloid fibrils. Proc. Natl Acad. Sci. USA 105, 18349–18354 (2008).
Article ADS CAS PubMed PubMed Central Google Scholar
Bertini, I., Gonnelli, L., Luchinat, C., Mao, J. & Nesi, A. A new structural model of Aβ40 fibrils. J. Am. Chem. Soc. 133, 16013–16022 (2011).
Article CAS PubMed Google Scholar
Lu, J.-X. et al. Molecular structure of β-amyloid fibrils in Alzheimer’s disease brain tissue. Cell 154, 1257–1268 (2013).
Article CAS PubMed Google Scholar
Xiao, Y. et al. Aβ(1-42) fibril structure illuminates self-recognition and replication of amyloid in Alzheimer’s disease. Nat. Struct. Mol. Biol. 22, 499–505 (2015).
Article CAS PubMed PubMed Central Google Scholar
Colvin, M. T. et al. Atomic resolution structure of monomorphic Aβ42 amyloid fibrils. J. Am. Chem. Soc. 138, 9663–9674 (2016).
Article CAS PubMed PubMed Central Google Scholar
Wälti, M. A. et al. Atomic-resolution structure of a disease-relevant Aβ(1-42) amyloid fibril. Proc. Natl Acad. Sci. USA 113, E4976–E4984 (2016).
Article PubMed PubMed Central Google Scholar
Gremer, L. et al. Fibril structure of amyloid-β(1-42) by cryo-electron microscopy. Science 358, 116–119 (2017).
Article ADS CAS PubMed PubMed Central Google Scholar
Hu, Z.-W. et al. Molecular structure of an N-terminal phosphorylated β-amyloid fibril. Proc. Natl Acad. Sci. USA 116, 11253–11258 (2019).
Article ADS CAS PubMed PubMed Central Google Scholar
Yang, Y. et al. Cryo-EM structures of amyloid-β 42 filaments from human brains. Science 375, 167–172 (2022).
Article ADS CAS PubMed PubMed Central Google Scholar
Ghosh, U., Thurber, K. R., Yau, W.-M. & Tycko, R. Molecular structure of a prevalent amyloid-β fibril polymorph from Alzheimer’s disease brain tissue. Proc. Natl Acad. Sci. USA 118, e2023089118 (2021).
Article CAS PubMed PubMed Central Google Scholar
Lee, M., Yau, W.-M., Louis, J. M. & Tycko, R. Structures of brain-derived 42-residue amyloid-beta fibril polymorphs with unusual molecular conformations and intermolecular interactions. Proc. Natl Acad. Sci. USA 120, e2218831120 (2023).
Article CAS PubMed PubMed Central Google Scholar
Loquet, A. et al. Atomic model of the type III secretion system needle. Nature 486, 276–279 (2012).
Article ADS CAS PubMed PubMed Central Google Scholar
Weingarth, M. & Baldus, M. Solid-state NMR-based approaches for supramolecular structure elucidation. Acc. Chem. Res. 46, 2037–2046 (2013).
Article CAS PubMed Google Scholar
Schütz, A. K. et al. Atomic-resolution three-dimensional structure of amyloid β fibrils bearing the Osaka mutation. Angew. Chem. 54, 331–335 (2015).
Article Google Scholar
Qiang, W., Yau, W.-M., Luo, Y., Mattson, M. P. & Tycko, R. Antiparallel beta-sheet architecture in Iowa-mutant beta-amyloid fibrils. Proc. Natl Acad. Sci. USA 109, 4443–4448 (2012).
Article ADS CAS PubMed PubMed Central Google Scholar
Sgourakis, N. G., Yau, W. M. & Qiang, W. Modeling an in-register, parallel “Iowa” A beta fibril structure using solid-state NMR Data from labeled samples with Rosetta. Structure 23, 216–227 (2015).
Article CAS PubMed Google Scholar
Elkins, M. R. et al. Structural polymorphism of Alzheimer’s β-amyloid fibrils as controlled by an E22 switch: a solid-state NMR study. J. Am. Chem. Soc. 138, 9840–9852 (2016).
Article CAS PubMed PubMed Central Google Scholar
Yoo, B. K., Xiao, Y., McElheny, D. & Ishii, Y. E22G pathogenic mutation of β-amyloid (Aβ) enhances misfolding of Aβ40 by unexpected prion-like cross talk between Aβ42 and Aβ40. J. Am. Chem. Soc. 140, 2781–2784 (2018).
Article CAS PubMed PubMed Central Google Scholar
Yang, Y. et al. Cryo-EM structures of amyloid-β filaments with the Arctic mutation (E22G) from human and mouse brains. Acta Neuropathol. 145, 325–333 (2023).
Article CAS PubMed PubMed Central Google Scholar
Hellström-Lindahl, E., Viitanen, M. & Marutle, A. Comparison of Abeta levels in the brain of familial and sporadic Alzheimer’s disease. Neurochem. Int. 55, 243–252 (2009).
Article PubMed PubMed Central Google Scholar
Kuperstein, I. et al. Neurotoxicity of Alzheimer’s disease Aβ peptides is induced by small changes in the Aβ42 to Aβ40 ratio. EMBO J. 29, 3408–3420 (2010).
Article CAS PubMed PubMed Central Google Scholar
Liang, R., Tian, Y. & Viles, J. H. Cross-seeding of WT amyloid-β with Arctic but not Italian familial mutants accelerates fibril formation in Alzheimer’s disease. J. Biol. Chem. 298, 102071 (2022).
Article CAS PubMed PubMed Central Google Scholar
Schöll, M. et al. Low PiB PET retention in presence of pathologic CSF biomarkers in Arctic APP mutation carriers. Neurology 79, 229–236 (2012).
Article PubMed Google Scholar
Petkova, A. T. et al. Self-propagating, molecular-level polymorphism in Alzheimer’s beta-amyloid fibrils. Science 307, 262–265 (2005).
Article ADS CAS PubMed Google Scholar
Shepherd, C., McCann, H. & Halliday, G. M. Variations in the neuropathology of familial Alzheimer’s disease. Acta Neuropathol. 118, 37–52 (2009).
Article CAS PubMed Google Scholar
Biancalana, M. & Koide, S. Molecular mechanism of Thioflavin-T binding to amyloid fibrils. Biochim. Biophys. Acta 1804, 1405–1412 (2010).
Article CAS PubMed PubMed Central Google Scholar
Han, B., Liu, Y., Ginzinger, S. W. & Wishart, D. S. SHIFTX2: significantly improved protein chemical shift prediction. J. Biomol. NMR 50, 43–57 (2011).
Article CAS PubMed PubMed Central Google Scholar
Wishart, D. S., Bigam, C. G., Holm, A., Hodges, R. S. & Sykes, B. D. 1H, 13C and 15N random coil NMR chemical shifts of the common amino acids. I. Investigations of nearest-neighbor effects. J. Biomol. NMR 5, 67–81 (1995).
Article CAS PubMed Google Scholar
Takegoshi, K., Nakamura, S. & Terao, T. 13C–1H dipolar-driven 13C–13C recoupling without 13C rf irradiation in nuclear magnetic resonance of rotating solids. J. Chem. Phys. 118, 2325–2341 (2003).
Article ADS CAS Google Scholar
Shen, Y. & Bax, A. Protein backbone and sidechain torsion angles predicted from NMR chemical shifts using artificial neural networks. J. Biomol. NMR 56, 227–241 (2013).
Article CAS PubMed PubMed Central Google Scholar
Colvin, M. T. et al. High resolution structural characterization of Aβ42 amyloid fibrils by magic angle spinning NMR. J. Am. Chem. Soc. 137, 7509–7518 (2015).
Article CAS PubMed PubMed Central Google Scholar
Ravotti, F. et al. Solid-state NMR sequential assignment of an Amyloid-β(1-42) fibril polymorph. Biomol. NMR Assign. 10, 269–276 (2016).
Article CAS PubMed Google Scholar
Wickramasinghe, A. et al. Sensitivity-enhanced solid-state NMR detection of structural differences and unique polymorphs in pico- to nanomolar amounts of brain-derived and synthetic 42-residue amyloid-β fibrils. J. Am. Chem. Soc. 143, 11462–11472 (2021).
Article CAS PubMed PubMed Central Google Scholar
Kollmer, M. et al. Cryo-EM structure and polymorphism of Aβ amyloid fibrils purified from Alzheimer’s brain tissue. Nat. Commun. 10, 4760 (2019).
Article ADS PubMed PubMed Central Google Scholar
Cold Spring Harbor Laboratory Press. Extracellular buffer. in Cold Spring Harbor protocols, (ed. Montenegro-Montero, A.) (Cold Spring Harbor Laboratory Press, 2013).
Newville, M., Stensitzki, T., Allen, D. B. & Ingargiola, A. LMFIT: non-linear least-square minimization and curve-fitting for Python. (2014).
Ishii, Y. 13C-13C dipolar recoupling under very fast magic angle spinning in solid-state NMR: applications to distance measurements, spectral assignments, and high-throughput secondary-structure elucidation. J. Chem. Phys. 114, 8473–8483 (2001).
Article ADS CAS Google Scholar
Matsunaga, T., Okabe, R. & Ishii, Y. Efficient solvent suppression with adiabatic inversion for 1H-detected solid-state NMR. J. Biomol. NMR 75, 365–370 (2021).
Article CAS PubMed Google Scholar
Delaglio, F. et al. NMRPipe: a multidimensional spectral processing system based on UNIX pipes. J. Biomol. NMR 6, 277–293 (1995).
Article CAS PubMed Google Scholar
Zheng, S. Q. et al. MotionCor2: anisotropic correction of beam-induced motion for improved cryo-electron microscopy. Nat. Methods 14, 331–332 (2017).
Article CAS PubMed PubMed Central Google Scholar
Rohou, A. & Grigorieff, N. CTFFIND4: fast and accurate defocus estimation from electron micrographs. J. Struct. Biol. 192, 216–221 (2015).
Article PubMed PubMed Central Google Scholar
Scheres, S. H. W. Amyloid structure determination in RELION-3.1. Acta Crystallogr. D Struct. Biol.76, 94–101 (2020).
Article ADS CAS PubMed PubMed Central Google Scholar
Emsley, P., Lohkamp, B., Scott, W. G. & Cowtan, K. Features and development of Coot. Acta Crystallogr. D Biol. Crystallogr. 66, 486–501 (2010).
Article ADS CAS PubMed PubMed Central Google Scholar
Adams, P. D. et al. PHENIX: a comprehensive Python-based system for macromolecular structure solution. Acta Crystallogr. D Biol. Crystallogr. 66, 213–221 (2010).
Article ADS CAS PubMed PubMed Central Google Scholar
Williams, C. J. et al. MolProbity: more and better reference data for improved all-atom structure validation. Protein Sci. 27, 293–315 (2018).
Article CAS PubMed Google Scholar
Humphrey, W., Dalke, A. & Schulten, K. VMD: visual molecular dynamics. J. Mol. Graph. 14, 33–38 (1996).
Article CAS PubMed Google Scholar
Pettersen, E. F. et al. UCSF ChimeraX: structure visualization for researchers, educators, and developers. Protein Sci. 30, 70–82 (2021).
Article CAS PubMed Google Scholar

Download references

Acknowledgements

The structural analysis of the E22G Aβ40 fibrils was mainly supported by an NIH U01 grant (5U01GM098033) to Y.I. The development of the high-field SSNMR analysis methods for the amyloid samples was supported by the JST-Mirai Program (grant No. JPMJMI17A2, Japan) to Y.I. Part of the analysis was performed using instruments supported by JSPS KAKENHI grants (No. JP22K18328, Japan) to Y.I. The authors also thank the RIKEN NMR facility in Yokohama, Japan, and its staff members. CryoEM data in this study were collected at the cryo-EM facility of the RIKEN Center for Biosystems Dynamics Research (Yokohama). We thank H. Ehara, T. Uchikubo, and R. Akasaka for their help with cryo-EM data collection and analysis. This work was partly supported by grants from JSPS/MEXT KAKENHI (JP21H05733 to A.Y.) and RIKEN (the pioneering projects “Structural Cell Biology project” and “Long-timescale chronobiology”). Y.I. thanks the former lab members at the Department of Chemistry, University of Illinois at Chicago who were involved in past E22G Aβ40 fibril projects, including Dr. Sandra Chimon Rogers, who passed away this year.

Author information

These authors contributed equally: Mohammad Jafar Tehrani, Isamu Matsuda.

Authors and Affiliations

School of Life Science and Technology, Tokyo Institute of Technology, 4259 Nagatsuta-cho, Midori-ku, Yokohama, Kanagawa, 226-8503, Japan
Mohammad Jafar Tehrani, Isamu Matsuda, Yu Kodama, Tatsuya Matsunaga & Yoshitaka Ishii
RIKEN Center for Biosystems Dynamics Research, 1-7-22 Suehiro-cho, Tsurumi-ku, Yokohama, Kanagawa, 230-0045, Japan
Atsushi Yamagata, Tatsuya Matsunaga, Naohiro Kobayashi, Mikako Shirouzu & Yoshitaka Ishii
RIKEN Center for Sustainable Resource Science, 1-7-22 Suehiro-cho, Tsurumi-ku, Yokohama, Kanagawa, 230-0045, Japan
Mayuko Sato & Kiminori Toyooka
Department of Chemistry, University of Illinois at Chicago, 845 W Taylor St, Chicago, IL, 60607, USA
Dan McElheny

Authors

Mohammad Jafar Tehrani
View author publications
You can also search for this author in PubMed Google Scholar
Isamu Matsuda
View author publications
You can also search for this author in PubMed Google Scholar
Atsushi Yamagata
View author publications
You can also search for this author in PubMed Google Scholar
Yu Kodama
View author publications
You can also search for this author in PubMed Google Scholar
Tatsuya Matsunaga
View author publications
You can also search for this author in PubMed Google Scholar
Mayuko Sato
View author publications
You can also search for this author in PubMed Google Scholar
Kiminori Toyooka
View author publications
You can also search for this author in PubMed Google Scholar
Dan McElheny
View author publications
You can also search for this author in PubMed Google Scholar
Naohiro Kobayashi
View author publications
You can also search for this author in PubMed Google Scholar
Mikako Shirouzu
View author publications
You can also search for this author in PubMed Google Scholar
Yoshitaka Ishii
View author publications
You can also search for this author in PubMed Google Scholar

Contributions

Y.I. conceived the idea of the study, coordinated the project, and interpreted all the data. M.J.T. optimized and prepared E22G Aβ40 fibril samples used for this study and carried out TEM, kinetics, and ¹³C SSNMR experiments together with SSNMR-based structural validation under supervision of Y.I. and I.M. T.M. performed ¹H-detected SSNMR experiments with M.J.T. A.Y. and I.M. performed cryoEM experiments with the aid of M.Sh. A.Y. and I.M. also constructed the initial cryoEM atomic structure of the E22G Aβ40 fibril. K.T., M.Sa., and A.Y. performed SEM analyses. I.M. performed cross-seeding kinetics experiments; I.M. also performed Congo-Red staining experiments. Y.K., N.K., and D.M. performed MD simulations; Y.I., N.K., and Y.K. mainly analyzed the data. Y.I., M.J.T., I.M., A.Y., K.T., and M.Sh. wrote the manuscript. All the authors approved the final version of the manuscript.

Corresponding author

Correspondence to Yoshitaka Ishii.

Ethics declarations

Competing interests

The authors declare no competing interests.

Peer review

Peer review information

Nature Communications thanks Kenji Iwasaki and the other, anonymous, reviewers for their contribution to the peer review of this work. A peer review file is available.

Additional information

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary information

Supplementary Information

Peer Review File

Description of Additional Supplementary Files

Supplementary Data 1

Reporting Summary

Source data

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc-nd/4.0/.

Reprints and permissions

About this article

Cite this article

Tehrani, M.J., Matsuda, I., Yamagata, A. et al. E22G Aβ40 fibril structure and kinetics illuminate how Aβ40 rather than Aβ42 triggers familial Alzheimer’s. Nat Commun 15, 7045 (2024). https://doi.org/10.1038/s41467-024-51294-w

Download citation

Received: 22 December 2023
Accepted: 05 August 2024
Published: 15 August 2024
DOI: https://doi.org/10.1038/s41467-024-51294-w
Springer Nature Limited

E22G Aβ40 fibril structure and kinetics illuminate how Aβ40 rather than Aβ42 triggers familial Alzheimer’s

Abstract

Similar content being viewed by others

Introduction

Results

Homogeneous E22G Aβ40 fibrils mimic amyloids in cotton-wool plaque

SSNMR indicates a homogeneous conformer for E22G Aβ40 fibril

CryoEM and SSNMR reveal a distinctive W-shaped motif for E22G Aβ40 fibrils

SSNMR co-analysis confirms only the core-layer E22G Aβ40 structure

NMR-based comparison highlights a distinctive fold of E22G Aβ40 fibril

Kinetic data suggest that E22G Aβ40 may trigger misfolding in Arctic FAD

Cross-propagation of E22G Aβ40 fibril structure to WT Aβ40 and Aβ42

Discussion

Methods

Materials

Sample preparation

MALDI-TOF mass spectrometry

ThT fluorescence spectroscopy

TEM analysis

SSNMR spectroscopy

Calculations

CryoEM analysis

SEM analysis

Visualization

MD simulations

Reporting summary

Data availability

References

Acknowledgements

Author information

Authors and Affiliations

Contributions

Corresponding author

Ethics declarations

Competing interests

Peer review

Peer review information

Additional information

Supplementary information

Source data

Rights and permissions

About this article

Cite this article

Share this article

Search

Navigation