Molecular Structure of Cu(II)-Bound Amyloid-β Monomer Implicated in Inhibition of Peptide Self-Assembly in Alzheimer's Disease

Abstract

Metal ions, such as copper and zinc ions, have been shown to strongly modulate the self-assembly of the amyloid-β (Aβ) peptide into insoluble fibrils, and elevated concentrations of metal ions have been found in amyloid plaques of Alzheimer's patients. Among the physiological transition metal ions, Cu(II) ions play an outstanding role since they can trigger production of neurotoxic reactive oxygen species. In contrast, structural insights into Cu(II) coordination of Aβ have been challenging due to the paramagnetic nature of Cu(II). Here, we employed specifically tailored paramagnetic NMR experiments to determine NMR structures of Cu(II) bound to monomeric Aβ. We found that monomeric Aβ binds Cu(II) in the N-terminus and combined with molecular dynamics simulations, we could identify two prevalent coordination modes of Cu(II). For these, we report here the NMR structures of the Cu(II)-bound Aβ complex, exhibiting heavy backbone RMSD values of 1.9 and 2.1 Å, respectively. Further, applying aggregation kinetics assays, we identified the specific effect of Cu(II) binding on the Aβ nucleation process. Our results show that Cu(II) efficiently retards Aβ fibrillization by predominately reducing the rate of fibril-end elongation at substoichiometric ratios. A detailed kinetic analysis suggests that this specific effect results in enhanced Aβ oligomer generation promoted by Cu(II). These results can quantitatively be understood by Cu(II) interaction with the Aβ monomer, forming an aggregation inert complex. In fact, this mechanism is strikingly similar to other transition metal ions, suggesting a common mechanism of action of retarding Aβ self-assembly, where the metal ion binding to monomeric Aβ is a key determinant.

Figure 1

Cu(II) binds to the N-terminus of Aβ40 and forms a compact complex. (A) ¹H-¹⁵N HSQC spectra of 75 μM Aβ40 in 10 mM HEPES, pH 7.2, with (blue) and without (red) 100 μM Cu(II) at 281 K. (B) Concentration-dependent attenuation of N-terminal signals upon titration of 40 (light green), 60 (green), 75 (cyan), and 100 μM (blue) Cu(II) concentrations. (C) The translational diffusion coefficient of Aβ40 increases with increasing Cu(II) concentration, indicating a smaller hydrodynamic radius of the Cu(II)-bound state.

Figure 2

Paramagnetic NMR experiments decrease the blind sphere around the paramagnetic center. (A) ¹H_N-R₂ rates of Aβ40 with (blue triangles) and without (red dots) 100 μM Cu(II) and PRE corresponding to the difference of these two rates (shown as bars). (B) CON experiment of Aβ with (blue, dia-state) and without (red, apo-state) 100 μM Cu(II). (C) CON optimized for paramagnetic agents (green, para-state) exhibits the recovery of three cross-peaks that are lost in the diamagnetic version (blue).

Figure 3

NMR structures and MD simulations of the Cu(II)-bound Aβ complex. (A) Overview about all constraints used for structure calculations. Binding ligands are colored in magenta, nuclei exhibiting PRE in cyan, nuclei within a 4.5 to 9.0 Å (for ¹³C atoms) or 6.5 to 9.0 Å (for H^N-atoms) shell in green, and nuclei within the blind sphere of NMR para-experiments (<6.5 Å) in orange. (B) Binding ligands of Cu(II) visualizing the two different chirality modes of the binding ligand D1 here shown for H14 as the fourth ligand. Both chirality modes also exist for the alternative coordination with H13 as the fourth ligand. (C) MD simulations for the chirality modes a and b compared to the experimental constraints for residues 1 to 15 for the coordination with H14 as fourth ligand (for H13 see Figure S8), suggesting chirality mode a as the preferred coordination. (D) High-resolution structures of the first 23 N-terminal residues of Aβ40 encapsulating the Cu(II) ion for the coordinations of chirality mode a with H13 and H14 as fourth ligand, respectively.

Figure 4

Cu(II) inhibits Aβ42 aggregation by specifically retarding fibril-end elongation. (A) Aggregation kinetics of 3 μM Aβ42 in 10 mM sodium-phosphate buffer, pH 8, during quiescent conditions in the presence of different Cu(II) concentrations from 0 (violet) to 10 μM Cu(II) (red). The aggregation traces were individually fitted with a nucleation model, including primary and secondary nucleation as well as fibril-end elongation. (B) Aggregation half times, τ_1/2, obtained from sigmoidal fits of the aggregation traces in panel (A). (C) Relative combined rate constant k_nk₊ (black) and k₊k₂ (pink), related to primary and secondary nucleation processes, respectively, as obtained from fits shown in panel (A). While the parameter k₊k₂ only shows a small variation with Cu(II) concentration, the parameter k_nk₊ changes over several orders of magnitude, indicating that Cu(II) primarily affects primary nucleation and/or fibril elongation. (D–F) Global fit analysis of aggregation traces where the fit parameters were constrained such that only one nucleation rate constant is the sole fitting parameter, i.e., k_n in panel (D), k₂ in panel (E), and k₊ in panel (F), revealing the best fit for k₊. (G) Highly seeded aggregation kinetics of 3.2 μM Aβ42 in the presence of 1.5 μM pre-formed fibrils. Under these conditions, the initial slope (inserted graph as zoom of the first 0.4 h) of the aggregation traces is proportional to k₊. (H) Relative elongation rate as obtained from seeded aggregation kinetics (panel G, red) and from global fit analysis (panel F, blue) as a function of the Cu(II):Aβ ratio. The data could be fitted to a model describing monomer attachment to the fibril ends, revealing an apparent dissociation constant of K_D= 0.7 ± 0.3 μM, where the gray curves represent the error range, which gives the error of K_D. (I) Relative number of nucleation units at different Cu(II) concentrations obtained from the integral of the nucleation rate (inserted graph). The nucleation rate is calculated from the parameters of the kinetic analysis and given in units of M s^–1 ×10^–14.

Figure 5

Final ThT intensity and TEM images of Cu(II)-Aβ42 aggregates. (A) Final ThT intensity of fibrils without Cu(II) (red) and upon addition of 10 μM Cu(II) (blue) and 5 mM EDTA (green) (left panel) or co-aggregated in the presence of 10 μM Cu(II) (cyan, right panel) and subsequent addition of 10 mM EDTA (light green, right panel), showing recovery of signal intensity upon EDTA addition. (B) TEM images were recorded without further storage at the end-points of the aggregation kinetics experiments, revealing clear fibril morphologies without and with low Cu(II) concentrations. In contrast, at 200 μM Cu(II) mostly amorphous structures are visible, which only exhibit very low ThT fluorescence.

Figure 6

Model for Aβ inhibition mechanism of transition metal ions. (A) Global fit of the relative translational diffusion coefficients of Aβ40 in the presence of Ag(I) (green), Zn(II) (blue), or Cu(II) ions (violet) to a two-state model (free and bound state), resulting in a global fit parameter D_B/D_free of 1.094 ± 0.005. (B) Population of the metal ion-bound state as determined from diffusion data in panel (A). (C) Apparent dissociation constant K_D of Cu(II) binding to Aβ monomers from fit to relative elongation rates for Cu(II) (Figure 4H) and from previous results. (D) Schematic model for mechanism of action of inhibition of Aβ self-assembly by Cu(II). The metal ion-bound state is inert to aggregation, resulting in a predominate retardation of fibril-end elongation, which promotes the generation of new oligomers. (E) Free energy diagram visualizing the reaction scheme of amyloid fibrils formation, where increasing Aβ monomer concentration enhances the generation rate of amyloid formation. In contrast, high Cu(II) concentrations promote the formation of amorphous aggregates. Remarkably, this model is also qualitatively applicable to Zn(II) and Ag(I) ions, suggesting a common mechanism of action for these transition metal ions.