Modulation of aggregation and structural polymorphisms of β-amyloid fibrils in cellular environments by pyroglutamate-3 variant cross-seeding

Abstract

Amyloidogenic deposition of β-amyloid (Aβ) peptides in human brain involves not only the wild-type Aβ (wt-Aβ) sequences, but also posttranslationally modified Aβ (PTM-Aβ) variants. Recent studies hypothesizes that the PTM-Aβ variants may trigger the deposition of wt-Aβ, which underlies the pathology of Sporadic Alzheimer’s disease. Among PTM-Aβ variants, the pyroglutamate-3-Aβ (_pyroE3-Aβ) has attracted much attention because of their significant abundances and broad distributions in senile plaques and dispersible and soluble oligomers. _pyroE3-specific antibodies are being tested as potential anti-Aβ drugs in clinical trials. However, evidence that support the triggering effect of _pyroE3-Aβ on wt-Aβ in cells remain lacking, which diminishes its pathological relevance. We show here that cross-seeding with _pyroE3-Aβ₄₀ leads to accelerated extracellular and intracellular aggregation of wt-Aβ₄₀ in different neuronal cells. Cytotoxicity levels are elevated through the cross-seeded aggregation, comparing with the self-seeded aggregation of wt-Aβ₄₀ or the static presence of _pyroE3-Aβ₄₀ seeds. For the extracellular deposition in mouse neuroblastoma Neuro2a (N2a) cells, the cytotoxicity elevation correlates positively with the seeding efficiency. Besides aggregation rates, cross-seeding with _pyroE3-Aβ₄₀ also modulates the molecular level structural polymorphisms of the resultant wt-Aβ₄₀ fibrils. Using solid-state nuclear magnetic resonance (ssNMR) spectroscopy, we identified key structural differences between the parent _pyroE3/ΔE3 and wt-Aβ₄₀ fibrils within their fibrillar cores. Structural propagation from seeds to daughter fibrils is demonstrated to be more pronounced in the extracellular seeding in N2a cells by comparing the ssNMR spectra from different seeded wt-Aβ₄₀ fibrils, but less significant in the intracellular seeding process in human neuroblastoma SH-SY5Y cells.

Extracellular deposition of β-amyloid (Aβ) peptides is widely considered as a clinical hallmark and the main pathological processes in Alzheimer’s disease (AD) (1, 2, 3). Recently modifications of the long-lasting amyloid cascade hypothesis emphasize crosstalk between WT Aβ (wt-Aβ) and other amyloidogenic sequences of which the posttranslationally modified Aβ (PTM-Aβ) variants have been attracted increasing attention for the past 2 decades. Immunohistochemistry evidence showed that several PTM-Aβ variants, including phosphorylated-S8-Aβ, nitrated-Y10-Aβ, isomerized-D7-Aβ, pyroglutamate-3 (_pyroE3), and pyroglutamate-11-Aβ, were abundant in senile plaques and concentrated in the centra of plaques surrounded by the wt-Aβ (4, 5, 6). Aggregates formed by phosphorylated-S8, isomerized-D7, and _pyroE3-Aβ have also been detected in earlier stage amyloid deposits than with wt-Aβ aggregates (6, 7, 8, 9, 10). These results support a hypothetical amyloidosis process that PTM-Aβ variants may seed the wt-Aβ aggregation. Further biophysical studies demonstrated that posttranslational modifications of Aβ in general modulated its physicochemical and biological properties. For instance, the critical aggregation concentrations of _pyroE3 and its truncated precursor ΔE3-Aβ were one order of magnitude lower than their wt-Aβ counterparts and their lag periods of fibrillation were ∼50% shorter (11, 12, 13). The _pyroE3/ΔE3-Aβ fibrils also showed enhanced stability against degradation by aminopeptidase (14). Both immunohistochemistry and biophysical studies support a recent hypothesis that PTM-Aβ variants may in general serve as a trigger for pathological Aβ deposits, especially in the sporadic AD cases.

Despite these supporting results, lacking evidence that clearly addresses the pathological relevance and molecular basis of cross-seeding between PTM and wt-Aβ still cause underestimation of the pathological role of PTM-Aβ variants. Particularly, studies that directly explore the impacts of PTM-Aβ variants on wt-Aβ aggregates are rare. _pyroE3-Aβ is one of the most studied PTM-Aβ variants because it is present in human AD brain tissues with large abundance, that is, up to ∼25% of total Aβ (15). _pyroE3-Aβ variants are also widely distributed in several types of Aβ deposits, including senile plaques and dispensable and soluble Aβ oligomers (7). Amyloidosis of _pyroE3-Aβ has been shown to possess higher level of cellular toxicity than the wt-Aβ analogs under same testing conditions, with hypothesized mechanisms of forming stable and soluble low-molecular-weight oligomers or causing membrane disruptions (16, 17, 18, 19). Clinically, it has been reported that passive immunization with Donanemab, a _pyroE3-Aβ-specific antibody, alleviated amyloid plaques and stabilized cognitive deficits in a phase II trial (20). A previous transmission electron microscope (TEM) imaging study demonstrated that in the presence of preformed _pyroE3-Aβ₄₀ seeds, fibrillation of wt-Aβ₄₀ was accelerated, leading to altered morphologies in the resultant filaments (19). This work provided the most direct evidence about the impact of _pyroE3-Aβ variants on wt-Aβ in aqueous solutions. On the other hand, while the molecular structures of wt-Aβ fibrils have been extensively characterized (21), there are much fewer structural studies on PTM-Aβ fibrils. For _pyroE3-Aβ₄₀, a previous solid-state NMR (ssNMR) work reported that it possessed similar ¹³C/¹⁵N chemical shift fingerprints as wt-Aβ₄₀ fibrils grown under the same condition, suggesting that these two fibrils might share certain core segments (11). Similar structural features were also reported in nonfibrillar oligomers formed by _pyroE3/E11-Aβ₄₀ (22). We previously measured the site-specific interstrand distances using ¹³C-PITHIRDs-CT spectroscopy in _pyroE3/ΔE3 and wt-Aβ₄₀ fibrils and showed that residues L17/A21/A30/L34 had similar parallel-in-register interstrand distances in all three fibrils and residues G9/V12 were localized in parallel-in-register β-sheet in _pyroE3/ΔE3 fibrils but not in wt-Aβ₄₀ fibril (23). Thus, the extension of fibrillar cores may be different in PTM versus wt-Aβ₄₀ fibrils.

The present work aims to understand how the preformed _pyroE3/ΔE3-Aβ₄₀ fibrillar seeds may influence the amyloidogenic aggregation of wt-Aβ₄₀ in neuronal cells. To this end, we quantified the time courses of cross-seeded amyloid aggregation of wt-Aβ₄₀ fibrils in mouse and human neuroblastoma cells and the associated elevations of cytotoxicity levels. Amyloid aggregates grown in cellular environments through either self-seeding or cross-seeding were then utilized to further produce isotopically labeled fibril samples for ssNMR measurements to investigate the molecular level structural propagation and variations.

Results

Seeded fibrillation induces Aβ amyloid deposition with cell type–specific locations and time dependence

We quantitatively analyzed the time-dependent aggregation of wt-Aβ₄₀ in the presence of either wt (i.e., self-seeding) or _pyroE3/ΔE3-Aβ₄₀ (i.e., cross-seeding) fibrillar seeds by confocal microscopy and flow cytometry. While the confocal microscopy provides information about all Aβ deposits in the cell culture, the flow cytometry with appropriate gating for cell sizes and morphologies reports the Aβ deposits associated with intact cells. In both experiments, cells were incubated with preformed pyroE3/ΔE3/wt-Aβ₄₀ seeds (nonfluorescently labeled) and monomeric wt-Aβ₄₀ peptides with a conjugated 5(6)-carboxyfluorescein group at its N terminus as a fluorescent reporter for designed time periods from 0 to 24 h. The usage of 5(6)-carboxyfluorescein did not alter the fibrillation rate based on our previous studies (23, 24).

Figure 1, A and B shows the representative fluorescence images for N2a and SH-SY5Y cells incubated with Aβ seeds and monomers for various time periods. Control images (Fig. S1) with only fluorescently labeled wt-Aβ₄₀ monomers showed minimal aggregates within 12-h incubation. For fibrillar-seeds-only controls, aggregates were observed with more heterogeneous distribution in N2a cells but mostly intracellularly in SH-SY5Y cells. All seeded fibrillation, regardless of the types of seeds and cells, produced visible time-dependent increase of fluorescence-active deposits. Most amyloid deposits (green fluorescence) colocalized with the SH-SY5Y cells (bright-field cell contours). However, in N2a, the deposits were distributed more heterogeneously, with abundant aggregates associated peripherally to cells. We confirmed the difference in the location of amyloid deposits using 3D confocal imaging (Fig. S2). Amyloid deposits were found to colocalize with cell nucleus for both N2a and SH-SY5Y, consistent with previous reports of the nonspecific intracellular distribution of Aβ deposits in the human neuroblastoma cell (25, 26). Fluorescence intensities in different seeding systems and cell types were quantified in Figure 1, C and D, where cross-seeding with _pyroE3/ΔE3-Aβ₄₀ in N2a cells (Fig. 1C, middle and bottom panels) led to more rapid deposition of wt-Aβ₄₀ aggregates compared with the self-seeding process. The comparison also agreed with our previous finding in aqueous solutions (23), where the seeded elongation rate constants for _pyroE3/ΔE3-Aβ₄₀ were 2 to 3 times higher than for wt-Aβ₄₀. Considering the fact that the _pyroE3/ΔE3-seeded wt-Aβ₄₀ fibrils possess more extended hydrophobic cores (23), which may indicate more stable structures and slower disassociation of monomers from the ends (27), the monomeric wt-Aβ₄₀ peptides may preferentially bind to _pyroE3/ΔE3-Aβ₄₀ seeds instead of wt-Aβ₄₀ seeds themselves. In SH-SY5Y cells, the seeded wt-Aβ₄₀ deposition occurred rapidly in all three systems and the overall fluorescence intensities reached plateau in about 6 h. Noting that major Aβ depositions occurred extracellularly and intracellularly in N2a and SH-SY5Y, respectively, the difference in aggregation rates suggested that cellular environments also affect the seeding processes.

Figure 1.

Confocal images and analysis of the seededAβ₄₀incells.A and B, representative confocal fluorescence images showing the time-dependent seeded wt-Aβ₄₀ aggregation in N2a (panel A) and SH-SY5Y (panel B) cells. C and D, quantification of the time-dependent increment of fluorescence intensities (box and whisker plots) due to the seeded wt-Aβ₄₀ aggregation in N2a (panel C) and SH-SY5Y (panel D) cells. For each dataset, the average, 25 to 75%, 1 to 99%, and min/max fluorescence intensities were shown as open squares, filled rectangles, crosses, and open circles, respectively. Increase of average fluorescence intensities were fit to single-exponential function (solid lines in C and D) with the following growth time constants: N2a/wt, 5.9 ± 2.2 h; N2a/_pyroE3, 0.8 ± 0.3 h; N2a/ΔE3, 0.2 ± 0.1 h; SH-SY5Y/wt, 0.5 ± 0.3 h; SH-SY5Y/_pyroE3, 0.6 ± 0.3 h; SH-SY5Y/ΔE3, 0.5 ± 0.2 h. Aβ, β-amyloid.

Flow cytometry analysis was done to assess the time-dependent Aβ deposition associated with cells. Both N2a and SH-SY5Y showed instant increase of fluorescence intensities (Fig. S3) upon incubation (e.g., within the first 5 min), suggesting rapid adhesions of fluorophore-labeled wt-Aβ₄₀ monomers to cells. This also agrees with the quick adsorption of exogenous Aβ_1-40 peptides to liposomes demonstrated in our previous works (28, 29). Interestingly, control confocal imaging of fluorescently labeled monomers (Fig. S1) within short incubation time (∼30 min) did not show significant green fluorescence, meaning that this initial cell adhesion of Aβ monomers did not produce large aggregates. Further increase of on-cell fluorescence intensities was observed with long incubation (e.g., 30 min to 24 h, Figure 2, A and B for N2a and SH-SY5Y, respectively), which can be attributed to time-dependent amyloid aggregation and consistent with the fluorescence microscopy. Distributions of fluorescence intensities in N2a cells were broader than the SH-SY5Y cells, agreeing with more heterogeneous spatial deposition of Aβ aggregates. A semiquantitative analysis of the on-cell fluorescence intensities was done by fitting each distribution to one or two Gaussian functions (Fig. S4), with the best-fit population/peak intensities summarized in Table S1. The weighted average on-cell fluorescence intensities from flow cytometry and the average fluorescence intensities from confocal microscopy showed a general positive correlation (Fig. 2, C and D), suggesting a uniform seeded Aβ amyloid deposition process in each individual system (i.e., a particular cell line and a specific type of fibrillar seed). However, the Pearson’s R-values were slightly higher for SH-SY5Y cells (0.96–0.97) than N2a cells (0.86–0.95), which was also consistent with the higher heterogeneity in terms of the on-cell fluorescence and amyloid deposit distribution in N2a cells.

Figure 2.

Flow cytometry and analysis of the seededAβ₄₀incells.A and B, plot of the time-dependent on-cell seeded wt-Aβ₄₀ aggregates fluorescence intensities based on flow cytometry analysis in N2a (panel A) and SH-SY5Y (panel B) cells. C and D, plot the linear correlation between the averaged fluorescence intensities from confocal imaging and flow cytometry measurements in N2a (panel C) and SH-SY5Y (panel D) cells. Linear fitting led to the Pearson’s R values (R = 1 indicates a positive correlation). Aβ, β-amyloid.

Seeded fibrillation induces elevation of cell toxicity levels that correlates with the fibrillation rates

Mature Aβ fibrils have been considered inert and nontoxic to cells in some previous works (30, 31). However, both fragmentized fibrillar seeds that are obtained through sonication of long and mature fibrils and the dynamic aggregation process are known to be cytotoxic (32, 33, 34, 35). In the present work, we explore whether and how the seeded Aβ fibrillation process correlates with the induced cytotoxicity level changes. A previously developed 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT)-based assay was applied to quantitatively assess the time-dependent elevation of cell viability loss. Similar to the confocal and flow cytometry analysis, cells were incubated with preformed pyroE3/ΔE3/wt-Aβ₄₀ seeds and monomeric wt-Aβ₄₀ peptides (nonfluorescently labeled) for 0 to 48 h. Compared with the static presence of various Aβ seeds (dashed lines, Fig. 3, A and B), seeded fibrillation processes (solid lines, Fig. 3, A and B) induced higher level cell viability loss, especially at long-term incubation. As a control, we showed that incubation with only wt-Aβ₄₀ monomers caused much less cell viability loss over time (e.g., ∼10% viability reduction over 48 h, Fig. S5). It is worth noting that the cell viability dropped significantly within short incubation time in the presence of Aβ seeds or seeds/monomer mixtures, but not with monomers only. This may be attributed to rapid association between cells and Aβ seeds, which was seen by confocal imaging (Fig. S1). However, the MTT-based assay requires at least a few hours incubation for the development of UV-active formazan crystals, and therefore does not report instant events. Analyses of the cytotoxicity level elevation (Fig. 3, C and D), in comparison with the time-dependent Aβ aggregation processes in the two cell lines, led to the following conclusions (1): in N2a cells, the cross-seeding with _pyroE3/ΔE3-Aβ₄₀ induced more rapid elevation of cellular toxicity compared with the self-seeding with wt-Aβ₄₀, which correlated positively with their relative in-cell aggregation rates (2); in N2a cells, there was a time-delay between the seeded Aβ aggregation and the toxicity elevation, presumably due to the lagged cell response to the extracellular Aβ aggregation. For instance, in the self-seeding process, the fluorescence-based Aβ aggregation increased significantly between 12 to 24 h, while the cell viability loss only increased rapidly after 24 h. For cross-seeding, the aggregation process reached maximum fluorescence intensities in 6 to 12 h, while the toxicity level elevation continued to increase afterward (3); in SH-SY5Y cells, both self- and cross-seeding processes led to rapid cytotoxicity elevation (e.g., ∼20% within 2-h incubation), and there was no apparent difference between the _pyroE3/ΔE3 and wt-Aβ₄₀ seeds, which was also consistent with the rapid seeded Aβ aggregation shown by confocal imaging and flow cytometry.

Figure 3.

Viability of cells in the presence of seededAβ₄₀. The time courses of cell viability in the presence of only Aβ fibrillar seeds (dashed lines in panel A and B), Aβ fibrillar seeds with monomeric peptides in 1:10 M ratio (solid lines in panel A and B), and the difference between these two sets of samples (panel C and D), determined by the MTT-based fluorescence assay. Percentages of cell viability were normalized to controls with only cells incubated for the same time periods as samples. Color-coding in panels (A–D): black, self-seeding of wt-Aβ₄₀; red, cross-seeding of wt-Aβ₄₀ with _pyroE3-Aβ₄₀ seeds; and blue, cross-seeding of wt-Aβ₄₀ with ΔE3-Aβ₄₀ seeds. Aβ, β-amyloid.

Structural and residue-specific dynamics differences between _pyroE3/ΔE3-Aβ₄₀ and wt-Aβ₄₀ fibrils

To understand the molecular structural differences in parent Aβ₄₀ fibrils that might underlie the seeded aggregation process, we compared the residue-specific ¹³C chemical shifts and inter-residue interactions between _pyroE3/ΔE3- and wt-Aβ₄₀ fibrils using selectively uniformly labeled sequences (Table S2, also with all the primary sequences and chemical structures of pyroglutamation), which covered the entire primary sequence. Intraresidue cross peaks were not observed for residues E3-D7 in cross-polarized–based 2D ¹³C-¹³C spectra with short-mixing time (Figs. 4, A and B and S6), indicating highly dynamic N-terminal segments in all three fibrils. Segments S8-Y10 and G38-V39 showed broadened peaks compared to other residues, meaning that these residues possessed local disorder. It was noted that residues L17, F20, S26, and A30 showed more than one set of Cα/Cβ cross peaks in wt- and/or _pyroE3/ΔE3-Aβ fibrils, suggesting the presence of molecular level structural polymorphisms. Analysis of the residue-specific chemical shift deviation (Fig. 4C and Table S3) demonstrated high-level structural similarity between _pyroE3 and ΔE3-Aβ₄₀ fibrils, but significant difference between the wt-Aβ₄₀ and _pyroE3/ΔE3-Aβ₄₀ fibrils. Particularly, continuous chemical shift deviations were identified in segment D23-A30, suggesting structurally different “loop” regions. In addition, apparent ¹³C chemical shift differences were also observed for residues F19, A21, I32, L34, and V36, agreeing well with our previous results that these residues involved in different inter-residue cross peak patterns in wt-Aβ₄₀ versus _pyroE3/ΔE3-Aβ₄₀ fibrils (23). Further evidence of structural differences was obtained by the long-mixing spin diffusion, which detect inter-residue cross peaks within ∼0.8 nm spatial distances, and frequency-selective ¹³C-¹⁵N REDOR spectroscopy, which quantitatively reported a specifically isotope-labeled ¹³C-¹⁵N nucleus pair on D23 and K28 sidechains. Long-range contacts between residues G25-I31-M35 were present in _pyroE3/ΔE3-Aβ₄₀ fibrils but missing in wt-Aβ₄₀ fibrils (Fig. 5A). Oppositely, inter-residue D23-S26-K28 and V18-S26 contact (Fig. S7) and D23-¹³Cδ/K28-¹⁵Nζ proximity (Fig. 5B) was identified in wt-Aβ₄₀ fibrils but not in _pyroE3/ΔE3-Aβ₄₀ fibrils. Table S4 summarized all inter-residue contacts, where most identified inter-residue contacts in wt-Aβ₄₀ fibrils, including D23-K28, F19-I32/L34, and D23-S26, have been reported in previous structural models (36, 37, 38, 39). On the contrary, multiple sets of inter-residue contacts identified in _pyroE3/ΔE3-Aβ₄₀ have rarely been reported. For instance, L17 was close to S8/Y10 and V39, which were at the edge of fibrillar core based on the broadened ¹³C peaks. In the meanwhile, L17 showed two sets of ¹³C chemical shifts, which also indicated local structural flexibility at this site.

Figure 4.

Chemical shift deviations in the parent wt and variantAβ₄₀fibrils. A and B, representative short-mixing 2D ¹³C-¹³C correlation spectra for the parent wt, _pyroE3 and ΔE3-Aβ₄₀ fibrils. Color boxes highlight the peak shifting in different fibrils, for example, positions of crosses mark the corresponding peaks in wt-Aβ₄₀ fibrils, which were shifted from the peaks in _pyroE3/ΔE3-Aβ₄₀ fibrils. C, comparison of the residue-specific C’/Cα/Cβ ¹³C chemical shifts in different parent fibrils. A chemical shift deviation $\sqrt{{\left(\Delta C^{\prime }\right)}^2+{\left(\Delta C\alpha \right)}^2+{\left(\Delta C\beta \right)}^2}$ > ∼1 ppm is considered significant based on the line widths in 2D ssNMR spectra. Aβ, β-amyloid; ssNMR, solid-state NMR.

Figure 5.

Residue-specificlong-rangecontacts in theparent wt and variant Aβ₄₀fibrils.A, representative long-mixing 2D ¹³C-¹³C correlation spectra for parent fibrils. Colored boxes and lines highlighted the inter-residue cross peaks that are characteristic in wt or _pyroE3/ΔE3-Aβ₄₀ fibrils. B, representative ¹³C-¹⁵N fsREDOR spectra and plot of REDOR dephasing to probe the D23-K28 proximity in different parent fibrils. Dashed lines were simulated using a ¹³C-¹⁵N spin-pair with different internuclear distances by SIMPSON package. Aβ, β-amyloid.

Besides molecular level structural differences, _pyroE3/ΔE3- and wt-Aβ₄₀ fibrils also showed distinct intrinsic flexibility at the selected sites, probed by temperature-dependent ²H ssNMR line shape analysis (Figs. 6, A and B and S8) and relaxation dispersion times (T₂) (Fig. 6C) with the time-domain ²H-QCPMG (40, 41). These techniques are known to be sensitive to slow μs-ms time scale dynamics due to modulations of the deuterium quadrupolar interaction by internal motions (42, 43). The local motions comprising rotameric interconversions of the hydrophobic sidechains of core residues L34, V36, and the N-terminal V12 were analyzed using previously developed models shown in Fig. S10 (44, 45), resulting in the values of rotameric exchange constant k_rot and the weights of the dominant conformers w at individual temperatures (Fig. S9). The temperature dependence of k_rot and w can be assumed to follow the Arrhenius and Boltzmann behaviors, respectively, leading to the values of activation energies E_a and differences in the potential energy between the major and minor conformers ΔE (Fig. 6D). Our results indicate that, in general, the residue-specific dynamics parameters in _pyroE3/ΔE3-Aβ₄₀ fibrils are significantly more like each other than to the wt-Aβ₄₀ fibrils.

Figure 6.

Dynamics in theparent wt and variant Aβ₄₀fibrils.A and B, examples of experimental ²H static solid-state NMR line shapes for the pyroE3-Aβ_1-40 fibrils, collected at 9.4 T using the quadrupolar echo experiment (43). C, ²H NMR relaxation times T₂versus echo spacing delay τ_QCPMG for the G9 site in the three types of fibrils, collected via the time-domain multiple-echo acquisition scheme (40, 41). _pyroE3-Aβ₄₀ fibrils were probed at 9.4 T, while the data for the wt and ΔE3-Aβ₄₀ fibrils are taken from prior work at 14.1 T (40, 48). The solid lines represent the fits to the two-site exchange model of Fig. S10. D, plot of side chain rotameric interconversion energies E_a and ΔE values for parent fibrils at L34, V36, and L12. E, freezing curves of p_boundversus T for the three types of fibrils obtained using the line shape decomposition (46). The solid lines indicate the fits to the sigmoidal curve as described in the text. Aβ, β-amyloid.

Additionally, the disordered N-terminal domain has been suggested to undergo large-scale concerted rearrangements (Fig. S10) (46), which are partially quenched due to transient intramolecular or intermolecular interactions with the structured hydrophobic core. This global motional mode can be modeled by the fraction of “free” and “bound” (p_bound) states in which the free state experiences diffusive-like motions parametrized by the isotropic diffusion coefficient D. The disordered domain participates in the conformational exchange process between the free and the transiently bound state, with the rate constant k_ex. We have found that in the wt-Aβ₄₀ fibrils the extent of the diffusive motion and of the conformational exchange gradually decrease along the N-terminal sequence (46, 47). The G9 site, labeled at the C^αD₂ position, plays a special role in probing variations between different types of fibrils, as its position is most sensitive to variant-specific changes in the dynamics for this mode (48). From the temperature dependence of the ²H NMR static line shapes, one can extract the p_bound as a function of temperature (Figs. 6E and S10), which follows a sigmoidal behavior: $p_{bound}=a+\frac{b-a}{1+\exp \left(\frac{T-T_m}{\sigma }\right)}$. The resulting parameters, such as the mid-point of the freezing curve T_m and the width of the transition σ, parameterize the variations between different types of fibrils. The values of the D and k_ex are most precisely determined for the G9 site from the T₂ QCPMG measurements (Fig. C). The summary of all resulting parameters for the G9 site for the three types of fibrils are shown in Fig. S11. Again, we observe that _pyroE3/ΔE3-Aβ₄₀ fibrils have similar behavior and in general diverge further from wt-Aβ₄₀ fibrils than from each other.

Overall, all ssNMR results indicated that both the molecular structures and the intrinsic residue–specific dynamics of _pyroE3/ΔE3-Aβ₄₀ fibrils are different from wt-Aβ₄₀ fibrils, which potentially form the underlying molecular basis for their different seeding abilities. We have recently also shown that the cross-seeded wt-Aβ₄₀ fibrils can retain many of the essential dynamics features in the original PTM-Aβ₄₀ seeds. Certain features may even be enhanced as the result of the cross-seeding, such as the activation energies of the rotameric motions and large-scale rearrangements of the N-terminal domain (49).

Structural propagation and deviation through the seeded Aβ aggregation in cellular environments

To understand the molecular level structural propagation through seeded Aβ aggregation in cellular environments, we compared the residue-specific ¹³C chemical shifts in all seeded wt-Aβ₄₀ samples in N2a and SH-SY5Y cells. To incorporate the isotope-labeled sequence to prepare ssNMR samples (sequence 4 in Table S2, ¹³C uniformly labeled at E11, V18, F20, D23, S26, N27, K28, and G29), cells were incubated with seeds and monomers for at least 72 h to allow maturation of seeded Aβ₄₀ aggregates. Mixtures of aggregates and cells (mostly cell debris because of the severe cell death at longer incubation time, according to the viability test shown in Fig. 3) were then sonicated extensively, diluted, and isotope-labeled wt-Aβ₄₀ monomers were added. Figure 7A shows the thioflavin-T kinetics curve that indicate the seeing effect in the isotope-labeled fibril growth. The lag periods were significantly shortened in the presence of seeds-contained cell mixture comparing with either monomeric wt-Aβ₄₀ peptide by itself or the presence of cell-only controls. Furthermore, the lag periods were shortened with less dilution fold, which also supported the seeding effect. Sequence 4 was chosen because it showed sound chemical shift differences between the parent _pyroE3/ΔE3 and wt-Aβ₄₀ fibrils. Figure 7B compares the representative spectral regions (Cα/Cβ cross peaks) in the six different seeded wt-Aβ₄₀ fibrils (also see Table S5 for the summary of chemical shifts). The results demonstrated visualizable ¹³C chemical shift deviations between the self-seeded fibrils in N2a and the other ones, especially at residues D23, S26, N27, and K28. More quantitative analyses were done by comparing the pair-wise chemical shift deviations for residue-specific Cα/Cβ (and C’/Ca peaks for G29) cross peaks between two seeded fibrils, as well as between the seeded and parent fibrils. The average pair-wise chemical shift deviation was plotted in Figure 7C as a heat map. The following conclusions can be drawn: first, self-seeding and cross-seeding led to structurally different wt-Aβ₄₀ fibrils in N2a cells (1.65–1.72 average deviations), while cross-seeding with _pyroE3 and ΔE3-Aβ₄₀ results in similar fibril structures (0.29 average deviation). Second, for any seeds, structures of the resultant wt-Aβ₄₀ fibrils were affected by the cellular system, that is, the presence of N2a and SH-SY5Y cells influenced the seeded wt-Aβ₄₀ fibril structures (0.5–1.6 average deviations). Third, the structural propagation from parent wt-, _pyroE3, ΔE3-Aβ₄₀ fibrils to the seeded wt-Aβ₄₀ fibrils are more significant in N2a cells (0.34–0.37 average deviation) compared with SH-SY5Y cells (0.60–1.69 average deviation).

Figure 7.

Deviations in theseeded wt-Aβ₄₀ fibrils incells.A, representative Th-T kinetics of seeded wt-Aβ₄₀ fibril growth using aggregates extracted from cells. B, 2D ¹³C-¹³C short-mixing correlation spectra for six different seeded WT Aβ_1-40 fibrils. The color spheres highlighted the position of residue-specific Cα/Cβ cross peaks in the self-seeded one from N2a cells (upper left spectrum). The peak positions shifted significantly in the other seeded fibrils. C, a heat map that shows the pair-wise comparison of Cα/Cβ chemical shift deviations between two different parent/seeded fibrils. Individual numbers in the heat map were calculated as the average value of Cα/Cβ chemical shift deviations (Cα/C′ for G29) over labeled residues V18, F20, D23, S26, N27, K28, and G29. D, comparison of the residue-specific Cα/Cβ (and Cα/C′ for G29) chemical shifts between the six seeded wt-Aβ₄₀ fibrils (color-coding shown in the insets) and the parent fibrils. A chemical shift deviation $\sqrt{{\left(\Delta C\alpha \right)}^2+{\left(\Delta C\beta \right)}^2}$ > ∼0.7 ppm is considered significant based on the line widths in 2D ssNMR spectra. Residue D23 in certain seeded fibrils were unassigned, shown by asterisks. Aβ, β-amyloid; ssNMR, solid-state NMR; Th-T, thioflavin-T.

Figure 7D further compares residue-specific Cα/Cβ chemical shift deviations between the seeded fibrils and their parent seeds, where a value larger than ∼0.7 ppm may be considered significant based on the ssNMR spectral resolution. Residues V18 and F20, usually considered in β-sheet segments in Aβ₄₀ fibrils, showed minimal chemical shift changes in all the parent and seed fibrils. Especially, F20 possessed two sets of peaks in all fibrils, neither of which shifted significantly. This observation suggests that the typical β-sheet segment L17-A21 may be preserved in both the parent _pyroE3/ΔE3-Aβ₄₀ fibrils and the seeded wt-Aβ₄₀ fibrils. The segment D23-G29, and particularly residues S26-K28, served as a fingerprint of structural propagation. In the presence of N2a cells, the self-seeded fibril (black, solid) showed consistently small deviations to wt-Aβ₄₀ parent fibrils and large deviations to _pyroE3/ΔE3-Aβ₄₀ parent fibrils. On the contrary, the cross-seeded fibrils (red and blue, solid) possessed consistently large deviations to wt-Aβ₄₀ parent fibrils and small deviations to _pyroE3/ΔE3-Aβ₄₀ parent fibrils. Therefore, the molecular structural features in parent Aβ₄₀ fibrils (excluding N-terminal residues such as E11, which showed less clear pattern based on our results) propagate to seeded fibrils in the presence of N2a cells, inducing modulations of the Aβ₄₀ fibrillar structural polymorphisms through the templating effect. However, in the presence of SH-SY5Y cells, such structural propagation becomes less clear. For example, self-seeding with the parent wt-Aβ₄₀ seeds led to a fibrillar structure with significant chemical shift deviations at S26, N27, and K28, suggesting that the seeding mechanisms were affected by different cellular environments.

Discussion

Results from the current work provide insights on the pathologically relevant influences of _pyroE3/ΔE3-Aβ₄₀ on the wt-Aβ₄₀ from two aspects. First, we showed that the presence of _pyroE3/ΔE3-Aβ₄₀ fibrillar seeds induced wt-Aβ₄₀ amyloid aggregates in cellular environments, with a positive correlation between the seeded aggregation rate and the cytotoxicity level elevation. In N2a cells, cross-seeding with _pyroE3/ΔE3-Aβ₄₀ leads to more rapid aggregation than the self-seeding, correlating with more rapid time-dependent cell viability loss. This positive correlation strongly supports the hypothesized triggering effect of _pyroE3/ΔE3-Aβ₄₀: although the _pyroE3-Aβ variant itself was shown to form aggregates more rapidly than wt-Aβ₄₀ in previous studies, this does not necessarily indicate that it would trigger the aggregation of wt-Aβ₄₀, even though they share highly similar primary sequences. A counterexample is that Aβ₄₂, which is known to aggregate more rapidly than Aβ₄₀, does not seed the fibrillation of Aβ₄₀ efficiently (50, 51). Comparison between the seeded aggregation processes and the time courses of cell viability changes suggest that mechanisms of seeded Aβ aggregation in cellular environments may be modulated by cell types, as shown in the schematic Figure 8. For N2a cells, large populations of deposits are peripheral to cellular exteriors, as indicated by the fluorescence microscopy. In this case, Aβ monomers are kept at a low concentration and fibrils may grow through the seeded elongation process in an extracellular and membrane-proximity environment (e.g., extracellular matrix or plasma membranes). Nonspecific membrane disruption may contribute to the mechanism of cell death (e.g., cell necrosis). We showed previously that monomeric wt-Aβ₄₀ peptides could be adsorbed to model phospholipid liposomes and isolated synaptic plasma membranes upon exogenous addition and the following growth of fibrils from membrane-associated Aβ_1-40 led to local disruptive effects such as vesicle content leaking and lipid mixing (28, 29, 52). For SH-SY5Y cells, fluorescence microscopy showed that the predominant aggregation occurred inside of cells within 2 h, and few fluorescently active deposits were found either peripherally or outside of cells. These observations imply a rapid uptake of wt-Aβ₄₀, which were also confirmed in previous studies using the same cell line (25). Consequently, there may be a rapid increase of local Aβ concentrations, because by estimation cells only occupy less than 3% of culture media volume considering ∼1 M SH-SY5Y cells in ∼150 μl culture media. Depending on whether Aβ₄₀ fibrillar seeds can enter SH-SY5Y cells with comparable efficiencies as monomers, the monomeric Aβ₄₀ may aggregate by self-nucleation or secondary nucleation to form oligomers and/or fibrils in intracellular environments. The pH values may also influence the seeded Aβ₄₀ aggregation pathways. It has been shown that Aβ may enter the human cell line through endocytosis (25), where the pH is lower than the extracellular physiological environments. A previous work demonstrated that Aβ₄₀ may form amorphous aggregates rapidly under endosomal pH, which then slowly converted into spherical oligomers and fibrils (53). This finding also agrees with the rapid aggregation rates found in SH-SY5Y cells in present work. These two potentially different seeding processes may also explain the distinct time courses for cellular viability loss for the two cell lines, as we showed by MTT assays. For N2a, the seeded aggregation induces an elevation of cytotoxicity that correlates positively with the seeding efficiency, implying that the cell death is associated with the deposition of aggregates, which happens extracellularly and may cause cell disruption in a longer time course. For SH-SY5Y, fast intracellular Aβ aggregation under high local concentrations may cause rapid elevation of cytotoxicity. Future studies will be done to explore different cell death pathways in these two cell lines and their correlations with the seeded Aβ aggregations.

Figure 8.

Schematic model to show the proposed distinct seeded wt-Aβ₄₀aggregation process in N2a and SH-SY5Y cells. The putative cartoon models for parent wt- and _pyroE3/ΔE3-Aβ₄₀ fibrils are sketched based on the current ssNMR-derived secondary structures and inter-residue contacts. Aβ, β-amyloid; ssNMR, solid-state NMR.

Our results also suggest a second pathological impact for the presence of _pyroE3/ΔE3-Aβ₄₀ fibrils: they may modulate the structural polymorphism of the seeded wt-Aβ₄₀ fibrils in cellular environments. Recent ssNMR and cryogenic TEM studies of ex vivo Aβ fibrils developed from postmortem tissue specimen has proposed the pathological relevance of molecular level structural polymorphism of Aβ fibrils (54, 55, 56). However, it remains unclear how Aβ may produce fibrils with heterogeneous molecular structures. From the diagnostic aspect, it is also useful to understand whether aggregates with certain molecular level structural features may become predominant at specific stages of disease progression. The structural templating effect through seeded fibrillation is more significant in the presence of N2a cells than SH-SY5Y cells. This may be associated with the extracellularly seeded aggregation, where the monomeric wt-Aβ₄₀ concentration is sufficiently low to eliminate the self-nucleation and secondary nucleation (Fig. 8). Along the primary Aβ sequence, the flexible N-terminal domain seems to carry more structural variations through seeding. It has been shown that residue-specific dynamics features of PTM-Aβ₄₀ variants were amplified through the seeding to wt-Aβ₄₀ fibrils (49). Residues that are typically located in β-sheet domains, such as V18 and F20, are structurally preserved through seeding and less sensitive to different cellular environments. Most importantly, our results identified that the loop segment (i.e., E22-G29 in Aβ₄₀) served as a “fingerprint” domain to report the structural propagation and deviation through seeding, because it replicated the structural features of parent seeds when the templating effect is significant (e.g., N2a cells) but adopted deviations when the templating effect is less predominant (e.g., SH-SY5Y cells). Future work will be done to probe structural propagations and deviations in the C-terminal β-sheet domain, which typically forms the quaternary interfaces in previously reported Aβ₄₀ fibrils and is shown to be sensitive to fibril growth environments such as the presence of membrane bilayers (57).

Experimental procedures

Peptide synthesis

All the ¹³C-isotope–labeled and unlabeled Aβ sequences (summarized in Table S1) were synthesized by standard fluorenylmethuloxycabonyl (FMOC)-based solid-phase peptide synthesis protocols, either manually or using a microwave peptide synthesizer (Biotage Initiator+ Alstra) For microwave-based peptide synthesis, the following instrument setup were applied: 75 °C nominal temperature, 10-min coupling time for all residues except for histidine, which used 40 °C nominal temperature and 60-min coupling. Low-substitution ChemMatrix Rink Amide resin (0.22 mmol/g loading, AAPPTec Inc) was utilized for the hydrophobic Aβ sequences. A few segments, such as N27-K28, F19-F20, H13-K16, and F4-S8, were double-coupled (5-fold excess amino acids per coupling cycle) to enhance the yield of synthesis (58). The fluorescent-labeled wt-Aβ₄₀ peptide for confocal microscopy and flow cytometry was synthesized with an additional coupling step of 5(6)-carboxyfluorescein (Sigma-Aldrich) to the N terminus after the last deprotection step. Crude peptides were cleaved from resin using a cocktail of 85% (v/v) trifluoroacetic acid, 5% (v/v) thioanisole, 2.5% (v/v) ethane-1,2-dithiol (all from Sigma-Aldrich Inc), 5% (v/v) H₂O, and 2.5% (m/v) phenol. The reaction was conducted at ambient temperature for 4 h. Peptides were purified using HPLC (Agilent Inc) with a reversed-phase C18 column and linear H₂O-acetonitrile solvent gradient. Products was confirmed using LC-MS with >95% purity (Shimadzu Inc) and stored at −80 °C freezer until usage.

Preparation of parent Aβ fibrils

All parent fibrils (wt-, _pyroE3-, and ΔE3-Aβ₄₀ sequences) were grown using the following protocols: purified peptides were dissolved in hexafluoro-isopropanol (Sigma-Aldrich Inc) to ∼2 mg/ml, briefly bath-sonicated and incubated at ambient temperature for overnight to remove preformed aggregates. Hexafluoro-isopropanol was then removed by N₂ flow and high vacuum for at least 5 h. The resulting films were dissolved in dimethyl sulfoxide (DMSO, Sigma-Aldrich) to ∼5.5 mg/ml and quantified based on the 280 nm absorbance. Aliquots of Aβ/DMSO stock were diluted into 50-fold 10 mM phosphate buffer (pH 7.4, with 150 mM NaCl and 0.01% NaN₃) with the final Aβ concentration ∼25 μM. The Aβ solution was then incubated at 37 °C for at least 7 days with continuous 50 rpm orbital shaking. Formation of fibrils was confirmed using routine thioflavin-T fluorescence with enhancement of emission at 490 nm (e.g., wavelength ∼440 nm). The resultant first-generation Aβ fibril solution was then utilized as seeds to produce second-generation fibrils: 10% (m/m with respect to the amount of monomeric Aβ for the second generation) first-generation Aβ solution was sonicated for 2 min on ice using a probe sonicator (20% duty cycle, 30 s/30 s on/off pulse) and diluted into 10 mM phosphate buffer. Fresh monomers were then treated as described before and then diluted into the solution of seeds. The mixture was incubated at 37 °C quiescently for 72 h. The third- and fourth-generation Aβ fibrils were produced using similar seeding protocols with 24-h incubation time, and the fourth-generation fibrils were utilized as “parent fibrils” for further experiments.

Cell culturing

Both the murine neuroblastoma cell (N2a, CCL-131, ATCC) and the human neuroblastoma cell (SH-SY5Y, 94030304, ECACC) were cultured using recommended protocols. Briefly, N2a cells were grown in 1:1 Dulbecco’s Modified Eagle Medium and Minimal Essential Medium (MEM) supplemented with 10% fetal bovine serum (Sigma-Aldrich) and 2% penicillin-streptomycin (Gibco). SH-SY5Y cells were cultured in 1:1 MEM and Ham’s F12 (Sigma-Aldrich) supplemented with 10% fetal bovine serum, 1% penicillin-streptomycin and 2 mM L-glutamine (Sigma-Aldrich). Both cell lines were incubated at 37 °C and 5% CO₂ atmosphere and passed twice weekly until 90% confluency.

Cell viability assay

For cell viability test, cultured cells were plated in 96-well cell culture plates (Corning, 3595) at a density of 5 k cells per well in 100 μl culture medium. The cells were then incubated at 37 °C, 5% CO₂ atmosphere for 12 h before addition of Aβ fibrillar seeds and/or monomers. For seeded Aβ aggregation, the culture medium was exchanged with 100 μl Opti-MEM with 2% penicillin-streptomycin (Gibco), containing 10 μM wt-Aβ₄₀ and 1 μM preformed fibrillar seeds (the fourth-generation wt-, _pyroE3-, or ΔE3-Aβ₄₀ parent fibrils prepared using the protocols described above). Controls were done by incubating cells with only monomeric wt-Aβ₄₀ peptides or without any additives (Fig. S5). The cells were incubated for the desired time periods (2, 6, 12, 24, and 48 h), followed by the addition of 10 μl MTT (5 mg/ml) in Opti-MEM and further incubation for 2 h to allow development of UV-active formazan crystals. The culture medium was then drained, and the formazan was redissolved in 150 μl DMSO and incubated at ambient temperature for 10 min. The absorbance at 490 nm was recorded using a multimode microplate reader (BioTek Inc).

Confocal laser scanning microscopy

For confocal microscopy, cover glass and slides (VWR Inc) were immersed in 2 N NaOH solution, mixed overnight, washed with water until the pH of rinsing returned to 6. Slides were dipped in 100% ethanol and air-dried. The cover glass was rinsed with 100% ethanol, followed by sterilization by autoclave. A cover glass was placed into each well on a 24-well cell culture plate (VWR Inc) and coated with 150 μl poly-L-lysine hydrobromide solution (100 μg/ml) in PBS (137 mM NaCl, 2.7 mM KCl, 8 mM Na₂HPO₄) for 15 min at ambient temperature. Poly-L-lysine solution was then discarded, cover glass was rinsed three times with 500 μl PBS, and wells were allowed to air-dry.

Cells cultured as described above were plated onto the wells 24 h prior to the start of seeded Aβ aggregation at a density of 200 k cells per well. Cell culture media was discarded, and cells were washed once with PBS once before addition of Aβ seeds/monomers. The following solutions were added to each well: 390 μl Opti-MEM, 60 μl 10 μM preformed Aβ₄₀ seeds (wt-, _pyroE3, or ΔE3), and 150 μl 40 μM 5(6)-carboxyfluorescein-conjugated wt-Aβ₄₀ monomer. The final concentration of monomeric wt-Aβ₄₀ and seeds were 10 μM and 1 μM, respectively. Controls were done by incubating cells with only monomeric wt-Aβ₄₀ peptides or only fluorescent-labeled wt-Aβ₄₀ seeds (Fig. S1). The mixture was incubated at 37 °C for the desired time periods (0.08, 0.5, 2, 6, 12, and 24 h). To stain the nucleus (for 3D confocal imaging), 3 μl SlowFade Diamond Antifade Mountant with 4',6-diamidino-2-phenylindole (DAPI) (Invitrogen) was added to the glass slide. The cover glass was placed with the sample side in contact with the nuclei stain. Clear nail polish was placed on the outer edge of the coverslip to secure it to the glass slide.

Images were taken using a Zeiss LSM 880 confocal microscope, installed with the ZEN Black package for image processing. The following excitation/emission wavelengths were applied: 405/440 to 480 nm for DAPI and 488/500 to 540 for 5(6)-carboxyfluorescein-conjugated wt-Aβ₄₀. All images were obtained with a 40× water immersion objective. For 3D imaging, the z-axis–stacked images were collected with 31 to 40 slices spanning over a 30-μm distance.

Flow cytometry

Cells were cultured using the above protocol, plated onto 24-well cell culture plates without the addition of cover glass, incubated for 24 h at 37 °C, washed once with PBS buffer, and mixed with the Aβ₄₀ seeds (wt-, _pyroE3, or ΔE3) and fluorophore-conjugated wt-Aβ₄₀ monomers for the desired time periods as described above. The final concentration of monomeric wt-Aβ₄₀ and seeds were 10 μM and 1 μM, respectively. Controls were done by incubating cells with only monomeric wt-Aβ₄₀ peptides or without any additive (Fig. S3). Cells were trypsinized with 200 μl 0.05% trypsin-EDTA for 2 min. Four wells of mixture (a total of 800 k cells) were combined, centrifuged at 500g to 800g for 10 min (Eppendorf Centrifuge 5418), fixed with 200 μl 4% PFA in PBS, washed twice with 250 μl PBS, and resuspended in 300 μl PBS for the flow cytometry analysis.

Fluorescence intensities from cell-bound 5(6)-carboxyfluorescein-conjugated wt-Aβ₄₀ aggregates were recorded on a ZE5 Cell Analyzer (Bio-Rad, installed with the Everest software package, https://www.bio-rad.com) with excitation/emission wavelengths set at 488/500 to 550 nm. Cells without the addition of Aβ seeds or monomers were utilized to delineate the nonfluorescent signals and to adjust the voltage settings for the side scatter and forward scatter (FSC). Overall 20k events were collected to generate histograms of fluorescent intensities, which were processed using the FlowJo package (BD Biosciences, https://www.flowjo.com) to select the gating for intact cells based on the 488 nm FSC:side scatter ratio and FSC area-to-height ratio.

MAS ssNMR spectroscopy

For wt-, _pyroE3-, and ΔE3-Aβ₄₀ sequences shown in Table S2, ssNMR samples were prepared by adding isotope-labeled monomers to the parent fibrillar seeds (after four-generation growth as described above, 1:20 w/w seeds-to-monomers ratio), followed by quiescent incubation for at least 72 h. Formation of fibrils was confirmed by CD spectroscopy (Fig. S12) and negatively stained TEM (Fig. S13). Fibrils were collected by ultracentrifugation (Beckmann Coulter, with a TLA-110 fixed-angle rotor, 285,000g, 10 °C for 30 min) and dried on lyophilizer. The pellets were packed into 2.5 mm magic angle spinning (MAS) rotors and rehydrated with 1 μl/mg deionized water.

To prepare seeded wt-Aβ₄₀ fibrils with isotope labeling in cells, N2a and SH-SY5Y cells were cultured as described above to 90% confluency and 10⁶ cells/ml. Ten micromolars monomeric wt-Aβ₄₀ peptides were then added to the cells with 10% (w/w) wt-, _pyroE3, or ΔE3-Aβ₄₀ seeds. The mixtures were further incubated (CO₂ incubator, 37 °C) for 72 h until loss of 80 to 90% cell viability (by MTT assays, described in the result session) and then sonicated extensively on ice for 15 min with a probe sonicator (25% duty cycle, 30 s/30 s on/off pulses). The mixture was diluted 20-fold with 10 mM phosphate buffer (pH 7.4, 0.01% NaN₃) and ∼1 mg isotope-labeled wt-Aβ₄₀ peptides (sequence 4 in Table S2) were then added to a final monomer concentration ∼10 μM. The seeding solution was incubated quiescently for 72 h, pelleted down by ultracentrifuge (285,000g for 30 min), lyophilized, packed into 2.5 mm thin-wall MAS rotors, and rehydrated with 5 μl deionized water.

All MAS ssNMR measurements were done on a 600 MHz Bruker spectrometer with a 2.5 mm TriGamma MAS probe (tuned to ¹H, ³¹P and ¹³C). 2D ¹³C-¹³C spin diffusion spectra was collected with 10 kHz MAS frequency at 280 K. The pulse sequence starts with a 55 kHz ¹H initial π/2 pulse, followed by a 1.5 ms cross-polarization period with 45 kHz/55 kHz ¹H/¹³C fields (with 30% linear ramp applied to ¹H channel). A 10 kHz radiofrequency-assisted diffusion was applied during the mixing period, which was set to 20 ms and 500 ms for intra- and inter-residue cross peaks, respectively. A 95 kHz two-pulse phase-modulation ¹H decoupling was applied through the detection period. Short- and long-mixing 2D spectra were typically collected with 12 to 24 and 36 to 48-h signal averaging and processed using NMRPipe with 50 Hz Gaussian line broadening in both dimensions. All ¹³C chemical shifts were calibrated using an external standard ¹³C’-Ala sample (to 177.95 ppm) for the comparison of chemical shifts.

The ¹³C-¹⁵N frequency–selective REDOR spectroscopy (59) was applied with 1 ms Gaussian pulses on both ¹³C and ¹⁵N channels with carrier frequencies at D23-Cγ (∼181.0 ppm) and K28-Nζ (∼90.5 ppm), respectively. The ¹³C and ¹⁵N π pulses were set to 50 kHz and 45 kHz, respectively. REDOR spectra were collected with the pulsed-spin locking acquisition algorithm (60) to enhance the spectral signal-to-noise ratio. All S₀ and S₁ spectra were processed with 20 Hz Gaussian line broadening and integrated over ∼0.5 ppm width around the peaks for the evaluation of REDOR dephasing.

Sample preparation protocols, NMR spectroscopy, and modeling for the ²H ssNMR experiments followed previously works with the details provided in the supplementary information (40, 42, 61, 62).

Data availability

All data are provided in the manuscript and supporting information.

Supporting information

Experimental procedures, including the ²H ssNMR sample preparation, spectroscopy, and modeling, are provided in the Supporting information. Supplementary figures including additional ¹³C-¹³C and ²H ssNMR spectra, ²H NMR relaxation data, CD and TEM images, confocal images, and flow cytometry data fitting. Supplementary tables summarize the isotope labeling patterns used in the current work and the chemical shift assignments in the parent and seeded Aβ₄₀ fibrils. References (40, 42, 43, 44, 45, 46, 47, 48, 49) are cited in the supporting information.

Conflict of interest

The authors declare that they have no conflicts of interest with the contents of this article.

Reviewed by members of the JBC Editorial Board. Edited by Elizabeth J. Coulson