Solid-state NMR reveals a comprehensive view of the dynamics of the flexible, disordered N-terminal domain of amyloid-β fibrils

Abstract

Amyloid fibril deposits observed in Alzheimer's disease comprise amyloid-β (Aβ) protein possessing a structured hydrophobic core and a disordered N-terminal domain (residues 1-16). The internal flexibility of the disordered domain is likely essential for Aβ aggregation. Here, we used ²H static solid-state NMR methods to probe the dynamics of selected side chains of the N-terminal domain of Aβ_1-40 fibrils. Line shape and relaxation data suggested a two-state model in which the domain's free state undergoes a diffusive motion that is quenched in the bound state, likely because of transient interactions with the structured C-terminal domain. At 37 °C, we observed freezing of the dynamics progressively along the Aβ sequence, with the fraction of the bound state increasing and the rate of diffusion decreasing. We also found that without solvation, the diffusive motion is quenched. The solvent acted as a plasticizer reminiscent of its role in the onset of global dynamics in globular proteins. As the temperature was lowered, the fraction of the bound state exhibited sigmoidal behavior. The midpoint of the freezing curve coincided with the bulk solvent freezing for the N-terminal residues and increased further along the sequence. Using ²H R_1ρ measurements, we determined the conformational exchange rate constant between the free and bound states under physiological conditions. Zinc-induced aggregation leads to the enhancement of the dynamics, manifested by the faster conformational exchange, faster diffusion, and lower freezing-curve midpoints.

Keywords: R1ρ relaxation; amyloid; fibril; intrinsically disordered protein; neurodegeneration; nuclear magnetic resonance (NMR); protein dynamics; protein misfolding; solid state NMR; zinc.

Figure 1.

Schematic representation of Aβ_1–40 protein with the amino acid side chains probed in this work labeled. A, the structure of the monomer for residues 9 and beyond is taken from Protein Data Bank code 2LMP (77), whereas the rest of the N-terminal domain is shown schematically as a line. B, 3-fold symmetric fibril structure, top view (77). C, a typical negatively stained TEM image of the resulting morphology, shown for fibrils labeled at the Phe-4 site.

Figure 2.

Representative ²H static solid-state NMR line shape data in the hydrated fibril samples of Aβ_1–40 in the 3-fold symmetric polymorph at several temperatures. Also shown (lowest right panel, in red) is the line shape of the isolated N-terminal domain (residues 1–16) labeled at the F4-ring-d₅ position, for which the narrow peak around 0 kHz is attributed to the HOD signal. Note the intrinsic differences in the quadrupolar interactions for different residues, specified in Table 2.

Figure 3.

²H longitudinal relaxation times T₁versus 1000/T recorded at 17.6-T field strength in the dry (black) and hydrated (blue) fibril samples of Aβ_1–40 for the major singularities of the powder pattern spectra: 0 kHz (squares), ±20 kHz for Ala-2, His-6, and Val-12 (circles), and ±60 kHz for Phe-4 and Gly-9 (circles). Error bars smaller than the size of the symbol are not shown. Note the different T₁ scales for the different residues. Error bars shown represent the S.E. of the fits of the decay curves.

Figure 4.

Comparison of the normalized ²H static solid-state NMR line shapes for the dry (black) and hydrated (blue) states of the Aβ_1–40 fibrils in the 3-fold symmetric polymorph.

Figure 5.

Schematic representation of the model in which the disordered N-terminal domain (curved line) transiently interacts with the structured C-terminal domain (blue rectangle). In the free state, the N-terminal domain is assumed to undergo isotropic diffusion, as represented by the gray sphere, whereas in the bound state, the interactions quench this mode. The parameters of the models are shown as the corresponding symbols.

Figure 6.

A, fraction of the bound state (p_bound) derived from the ²H line shape analysis, either determined directly from line shape data at 37 °C (blue) or as the upper temperature baseline parameter from the fits to the sigmoidal curves of Equation 1 (black). Note that the fit to Equation 1 was not performed for Val-12. B, p_bound as a function of the temperature. The solid lines represent the fits to the data according to Equation 1. C, diffusion coefficient D fitted based on the ²H line shapes at 37 °C according to the isotropic diffusion approximation.

Figure 7.

A and B, experimental ²H T_1ρ^fast = 1/R_1ρ^fast relaxation times (corresponding to the fast component of the double-exponential fit) (A) and T_1ρ^slow versus spin-lock field ω_SL (B) at 37 (red circles) and 25 °C (blue squares) at 14 T for hydrated Aβ_1–40 fibrils in the 3-fold symmetric polymorph labeled at the A2-C^βD₃ position. Solid lines represent the best-fit simulated data to the two-site exchange model described in the text. C, examples of the partially relaxed line shapes at 37 °C for the R_1ρ measurements at several values of spin-lock times for the 15-kHz spin-lock field, indicated on the graph. D, experimental magnetization decay curves for the 15-kHz field at 37 °C. Peak intensities integrated over the −1 to +1-kHz region (which is approximately the width at the half-height of the peak shown in C) are in arbitrary units versus time (circles); the solid line shows the double-exponential fit. Error bars shown represent the S.E. of the fits of the decay curves.

Figure 8.

Comparison of the dynamics between the hydrated 3-fold Aβ_1–40 fibrils (blue) and hydrated Zn²⁺-induced aggregates. A, ²H static solid-state NMR line shapes at 37 °C and representative spectra around the freezing transition region. B, midpoint melting temperatures T_m derived from fitting the fractions of the bound states (supporting information SI10) to Equation 1. C, ²H T_1ρ^fast relaxation times (corresponding to the fast component of the double-exponential fit) versus the spin-lock field ω_SL at 37 °C at 14 T for the A2-C^βD₃ site. Solid lines represent the best-fit simulated data to the two-site exchange model described in the text. Error bars shown represent the S.E. of the fits of the decay curves.