Local Structure and Global Patterning of Cu²⁺ Binding in Fibrillar Amyloid-β [Aβ(1-40)] Protein

Abstract

The amyloid-β (Aβ) protein forms fibrils and higher-order plaque aggegrates in Alzheimer's disease (AD) brain. The copper ion, Cu²⁺, is found at high concentrations in plaques, but its role in AD etiology is unclear. We use high-resolution pulsed-electron paramagnetic resonance (EPR) spectroscopy to characterize the coordination structure of Cu²⁺ in the fibrillar form of full-length Aβ(1-40). The results reveal a bis-cis-histidine (His) equatorial Cu²⁺ coordination geometry, and participation of all three N-terminal His residues in Cu²⁺ binding. A model is proposed, in which Cu²⁺–His6/His13 and Cu²⁺–His6/His14 sites alternate along the fibril axis, on opposite sides of the β-sheet fibril structure. The local intra-β-strand coordination structure is not conducive to Cu²⁺/Cu¹⁺ redox-linked coordination changes, and the global arrangement of Cu sites precludes facile multi-electron and bridged-metal site reactivity. This indicates that the fibrillar form of Aβ suppresses Cu redox cycling and reactive oxygen species (ROS) production. The insulator configuration suggests application of Cu²⁺-Aβ fibrils as an amyloid architecture for switchable electron charge/spin coupling and redox reactivity.

INTRODUCTION

Alzheimer's disease (AD) is a neurodegenerative disorder of present and acute future human impact.¹ The AD brain is characterized by extracellular histopathological lesions, or plaques.² The primary component of the plaques is the amyloid-β (Aβ) peptide. The Aβ peptides are n=39-42 amino acids in length, and are denoted Aβ(1-n). Aβ in the plaques is predominantly in the form of amyloid fibrils, which aggregate to form higher-order structures. High-resolution solid-state nuclear magnetic resonance (SS-NMR) studies have led to molecular models for the structure of residues 9-40 of the de-metallated Aβ(1-40) fibril prepared under different conditions,^3-5 and the Aβ(1-42) fibril,⁶ in which β-strand structure in central and C-terminal regions of the peptide promotes stacking of the β-turn β-fold in a parallel, in-register β-sheet arrangement. The β-sheet structure can extend along the fibril axis to micrometer lengths. The metal ion, Cu²⁺, is found at high concentrations of up to 400 μM in plaques, and it is directly coordinated by the Aβ peptide.^7-10 Reactive oxygen species (ROS) generation by Cu²⁺-Aβ, and ensuing oxidative damage to cellular components, has been proposed to contribute to the etiology of AD,^8,11,12 but a specific role of fibril-bound Cu²⁺ (promoting or ameliorating) has remained unclear.^7,8,11,13,14 Mounting refined evidence that relatively low molecular mass oligomers are cytotoxic forms of Aβ,^6,15,16 possibly abetted by enhanced Cu²⁺-promoted ROS generation,^14,17 re-invigorates the question: What is the role of fibrillar Cu²⁺-Aβ in AD?

General features of Cu²⁺ coordination in the soluble and fibrillar forms of Aβ(1-40) or Aβ(1-42) at pH 7.2-7.4 have been addressed by using continuous-wave EPR^18,19 and SS-NMR¹⁰ spectroscopies. The EPR parameters, g_⊥, g_∥, and the copper hyperfine coupling constant, A_∥, of Cu²⁺ complexed with soluble Aβ(1-16), Aβ(1-28), and monomeric Aβ(1-40) and Aβ(1-42) are the same to within experimental uncertainty.^18,19 The EPR line shape of the initially monomeric Cu²⁺-Aβ(1-40) complex did not change¹⁸ during fibrillization in the presence of Cu²⁺. Transmission electron microscopy (TEM) and SS-NMR showed that the presence of Cu²⁺ during Aβ(1-40) fibril formation did not influence the fibril morphology or reorganize the parallel, in-register β-sheet structure in Aβ(1-40) fibrils.¹⁰ These results show that the Cu²⁺ sites in soluble monomeric and fibrillar forms of Aβ are geometrically comparable, may involve identical or chemically-similar ligands, and that the structure of the N-terminal domain that binds Cu²⁺, and the distal β-sheet-forming domains of full-length Aβ, are independent.

The detailed coordination structure of Cu²⁺ in the soluble (non-aggregating, non-fibril forming), monomeric complexes of Cu²⁺ with the truncated Aβ-peptides, Aβ(1-16)^20-25 and Aβ (1-28),^19,26 or both,²⁷ has been addressed by using electron spin echo envelope modulation (ESEEM) and hyperfine sublevel correlation (HYSCORE) spectroscopies, in conjunction with isotopic labeling, His-to-Ala mutations, and chemical modification. The N-terminal metal binding domain in the Aβ peptide contains three histidine (His) residues, His6, His13, and His14, all of which have been implicated in Cu²⁺ coordination in soluble Aβ(1-16) and Aβ(1-28). A model of a two-component (Component Ia, Ib; Scheme 1), approximately equimolar mixture of His6/His13- and His6/His14-Cu²⁺ coordination in soluble Aβ(1-16) was proposed,²² and later revised and extended by additional spectroscopic studies.^23-25,28

Scheme 1.

We have used high-resolution pulsed-electron paramagnetic resonance (EPR) spectroscopy to probe histidine (His) coordination of Cu²⁺ (electron spin, S=1/2; nuclear spin, I=3/2) in frozen solution samples of fibrillar Aβ(1-40), to determine the local coordination structure and global patterning of Cu²⁺ binding at the N-terminus of fibrillar Aβ(1-40). The results provide molecular-level insights into the chemistry and physiological impact of Cu²⁺ binding by Aβ fibrils, and suggest applications of Cu²⁺-Aβ(1-40) as an amyloid architecture²⁹ for designed, switchable electron charge/spin conduction and redox reactivity.

MATERIALS AND METHODS

Sample Preparation

Wild type Aβ(1–16) and Aβ(1–40) peptides were purchased from rPeptide (Athens, GA) or Bachem (King of Prussia, PA). Glycerol, sodium phosphate (NaP_i), sodium chloride, phosphotungstic acid (PTA), and 1,1,1,3,3,3 hexafluro-2-proponal (HFIP) were purchased from Fisher Scientific (Pittsburgh, PA). Cu²⁺-Aβ samples were prepared as described previously.¹⁸ Briefly, Aβ(1–40) peptides were monomerized with HFIP and stored at −80 °C. An aliquot of the stock solution was removed for determination of the concentration, by using a bovine serum albumin calibration curve. Fibrillar Aβ(1-40) were prepared by using 100 μM Aβ(1-40) peptide in 50 mM NaP_i and 75 mM NaCl (pH 7.2), which was incubated at 37 °C for 7-14 days under quiescent conditions in the presence of equimolar Cu²⁺. Progress of fibril formation was assayed by TEM. Samples for EPR experiments were separated from the supernatant and concentrated by centrifugation of the pooled volumes in a microfuge (60 min, 16000 rcf), resuspended and washed with buffer, centrifuged a second time (30 min, 16000 rcf) and resuspended in a buffer of 50 mM NaP_i and 75 mM NaCl (pH 7.2) containing 50% glycerol (v/v). Samples of the soluble Aβ(1-16) were prepared by resuspending dried peptide in buffer containing 50 mM NaP_i, 75 mM NaCl (pH 7.2), and 50% glycerol (v/v). An equimolar amount of Cu²⁺ was added to the sample prior to freezing in liquid nitrogen.

EPR Spectroscopy and Simulations

CW-EPR spectroscopic measurements were made by using a Bruker ELEXSYS E500 EPR spectrometer with an ER 4123SHQE X-band cavity resonator and a Bruker ER 4131VT liquid nitrogen flow temperature control system. Spectra were acquired at 120 K under non-saturating conditions.

CW-EPR simulations were performed by using the SpinCount software package (M. Hendrich, Carnegie Mellon University: www.chem.cmu.edu/groups/hendrich/facilities/index.html) by diagonalization of the following spin Hamiltonian:

$$\mathit{H}={\beta }_e\mathit{B}\cdot \mathit{g}\cdot S+hS\cdot \mathit{A}\cdot I$$ (1)

with S= 1/2 and I = 3/2 for Cu²⁺. In Eq. 1, S and I are the electron and nuclear spin operators, g and A are the electron g and Cu²⁺ hyperfine tensors, β_e is the Bohr magneton, B is the applied magnetic field, and h is Planck's constant.

ESEEM Spectroscopy and Simulations

ESEEM was collected at 6 K on a home-designed and constructed pulsed EPR spectrometer using the three-pulse (π/2-τ-π/2-T-π/2-τ-echo) microwave pulse sequence.³⁰ ESEEM waveforms were cosine Fourier transformed to generate ESEEM frequency spectra. Data processing and analysis was performed by using codes written in Matlab (Mathworks, Natick, MA). All spectra were recorded under the same conditions, with B₀ = 303.0 mT and τ = 310 ns. The τ value of 310 ns was chosen to suppress the undesired modulation from ¹H shf coupling, and to enhance the contribution of the ¹⁴N Δm_I = ±2 (2ν_dq) combination modulation.

Numerical simulations of the ESEEM were performed by using the OPTESIM software package.³¹ The spin Hamiltonian for the Cu²⁺–¹⁴N interaction is:

$$\mathit{H}=\text{-}{g}_n{\beta }_n\mathit{B}\cdot I+hS\cdot \mathit{A}\cdot I+I’\cdot \mathit{Q}\cdot I’$$ (2)

where A is the shf coupling tensor, Q is the nqi tensor, g_n is the nuclear g factor, β_n is the nuclear magneton, B is the applied magnetic field, and h is Planck's constant. The traceless tensor Q is related to the nuclear quadrupole coupling constant, e²qQ/h, and the electric field gradient asymmetry parameter, η, as follows:

$$Q_{zz}=\frac{e^{2}qQ}{2I(2I-1)h}$$ (3)

$$\eta =\frac{Q_{xx}-Q_{yy}}{Q_{zz}}$$ (4)

where e is the elementary charge, q is the electric field gradient, and Q is the nuclear quadrupole moment. The relative orientation of the PAS of the nqi and shf tensors is defined by the Euler angles [α_Q, β_Q, γ_Q]. For systems with >1 coupled ¹⁴N nucleus, the orientation of the shf tensors is defined by the Euler angles, [α_A, β_A, γ_A]. Hybrid optimization in the simulations was performed by sequential application of genetic and Nelder-Mead simplex algorithms.³¹

EPR spectroscopy of natural isotopic abundance fibrillar Cu²⁺-Aβ(1-40) and soluble Cu²⁺-Aβ(1-16)

Continuous-wave (CW)-EPR spectra of the soluble Cu²⁺-Aβ(1-16) and fibrillar Cu²⁺-Aβ(1-40) complexes are shown in Figure 1. The EPR spectra of the two complexes are closely similar, as quantified by the EPR simulations (Figure 1, A and B; dashed lines), and simulation parameters, which are presented in Table 1. The single set of g_⊥ , g_∥ and A_⊥, A_∥ values required in the simulations are the same for the Aβ(1-16) and Aβ(1-40) complexes, to within one standard deviation. The g_∥ and A_∥ parameters in Table 1 are typical of a tetragonal, type-2 Cu²⁺ center.³² A three-nitrogen/one-oxygen (3N1O) equatorial coordination of Cu²⁺ is suggested by the Peisach-Blumberg correlation of g_∥ and A_∥.³³

Figure 1.

CW-EPR spectra of Cu²⁺-Aβ complexes. (A) Soluble Cu²⁺-Aβ(1-16), experiment and simulation (dashed line), (B) Fibrillar Cu²⁺-Aβ(1-40), experiment and simulation (dashed line), (C) ¹³C,¹⁵N-His13-Cu²⁺-Aβ(1-40), (D) ¹³C,¹⁵N-His14-Cu²⁺-Aβ(1-40). Acquisition parameters: microwave frequency = 9.45 GHz; microwave power = 2 mW; modulation amplitude = 1 mT; modulation frequency = 100 kHz; time constant = 10.24 ms; conversion time = 81.92 ms; T = 120 K. Spectra A and B are an average of 10 scans, and spectra C and D are an average of 25 scans. Simulation parameters for spectra A and B are presented in Table 1.

Table 1.

CW-EPR simulation parameters for the soluble, monomelic Cu²⁺-Aβ(1-16) and fibrillar Cu²⁺-Aβ(1-40) complexes.

Parametera	Cu2+-Aβ(1-16)	Cu2+-Aβ(1-40)
g ⊥	2.066 ±0.003	2.065 ±0.003
g ∥	2.277 ±0.003	2.277 ±0.003
σg ⊥	0.030	0.031
σg ∥	0.025	0.026
A∥ (MHz)b	524 ±5	527 ±5

^aSimulations are for S = 1/2, I = 3/2, and have a Gaussian line width of 1.0 mT. Standard deviations are given.

^bThe copper hyperfine parameter (A_∥) is assumed to be from the ⁶⁵Cu isotope.

The EPR simulation parameters (g_∥=2.272, g_⊥=2.056, A_∥=527 MHz) reported here for the fibrillar Cu²⁺-Aβ(1-40) complex are comparable with values reported previously for Cu²⁺ in soluble Aβ(1-40) (g_∥=2.268, g_⊥=2.061, A_∥=534 MHz)¹⁸ and in fibrillar Aβ(1-40/42) (g_∥=2.268, g_⊥=2.061, A_∥=534 MHz).^18,19 These parameters are also comparable to those for Component I of Cu²⁺-Aβ(1-16). Our results, in agreement with the previous EPR studies,^18,19 thus indicate that the Cu²⁺ site in fibrillar Aβ(1-40) has a comparable geometry and set of equatorial ligand atoms, as for Component I in Cu²⁺-Aβ(1-16).

The CW-EPR spectra of the fibrillar ¹³C,¹⁵N-His13-Cu²⁺-Aβ(1-40) and ¹³C,¹⁵N-His14-Cu²⁺-Aβ(1-40) complexes are shown in Figure 1, C and D. The EPR line shapes of the isotope-labeled complexes are comparable to the line shapes for the natural isotopic abundance spectra. Although hyperfine coupling with the I=3/2 copper nucleus is displayed at g_∥, an influence of the change from ¹²C (I=0) to ¹³C (I=1/2), and from ¹⁴N (I=1) to ¹⁵N (I=1/2), on the EPR line shape is not apparent, owing to inhomogeneous broadening. High resolution techniques of EPR spectroscopy, such as ESEEM, are necessary to resolve the ligand superhyperfine coupling.

¹⁴N ESEEM spectroscopy of natural isotopic abundance fibrillar Cu²⁺-Aβ(1-40) and soluble Cu²⁺-Aβ(1-16)

Three-pulse ESEEM waveforms and corresponding cosine Fourier transform (FT) spectra of four Cu²⁺-Aβ complexes, are shown in Figures 2 and 3, respectively. The spectral features are from the “remote,” non-coordinated ¹⁴N of His imidazole, and are characteristic of the “exact cancellation” condition³⁴ for coupling of the unpaired electron spin from Cu²⁺ with ¹⁴N. The ν₀, ν₋ and ν₊ “pure” nuclear quadrupole interaction (nqi) frequencies, and the frequencies of the Δm_I = ±1, and Δm_I = ±2 (double quantum, ν_dq) transitions, that originate from the non-cancellation electron spin manifold, and that include both nqi and superhyperfine (shf) contributions, are identified in Figure 3. Multiple ¹⁴N couplings result in the presence of weak modulation at sums of the fundamental modulation periods.^34,35 The “combination lines” are denoted in the FT spectra in Figure 3. In addition to the combination lines at the nuclear quadrupole frequencies, a broad feature centered at 7.8 MHz is observed, that corresponds to the 2ν_dq, double quantum combination line.³¹ Combination features of order >2 are not observed. The spectra in Figure 3 thus indicate that two histidines coordinate Cu²⁺ in fibrillar Aβ(1-40).

Figure 2.

Three-pulse ESEEM waveforms of Cu²⁺-Aβ complexes. (A) Soluble Cu²⁺-Aβ(1-16), experiment and simulation (dashed line), (B) Fibrillar Cu²⁺-Aβ(1-40), experiment and simulation (dashed line), (C) ¹³C,¹⁵N-His13-Cu²⁺-Aβ(1-40), (D) ¹³C,¹⁵N-His14-Cu²⁺-Aβ(1-40). The vertical scale bar corresponds to 25% of the echo amplitude at τ + T = 8 μs. Acquisition Parameters: microwave frequency, 8.750 GHz, B₀ = 303.0 mT, T = 6 K, τ = 310 ns. Spectrum A is an average of 10 scans, spectrum B is an average of 16 scans, and spectra C and D are an average of 25 scans. Simulation parameters for waveforms A and B are presented in Table 2.

Figure 3.

FT spectra of the three-pulse ESEEM waveforms of Cu²⁺-Aβ complexes. (A) Soluble Cu²⁺-Aβ(1-16), experiment and simulation (dashed line), (B) Fibrillar Cu²⁺-Aβ(1-40), experiment and simulation (dashed line), (C) ¹³C,¹⁵N-His13-Cu²⁺-Aβ(1-40), (D) ¹³C,¹⁵N-His14-Cu²⁺-Aβ(1-40). Inset: Expanded view of double quantum harmonic region, for (A) soluble Cu²⁺-Aβ(1-16) and (B) fibrillar Cu²⁺-Aβ(1-40). Acquisition Parameters: microwave frequency, 8.750 GHz, B₀ = 303 mT, T = 6 K, τ = 310 ns. Spectrum A is an average of 10 scans, spectrum B is an average of 16 scans, and spectra C and D are an average of 25 scans. Simulation parameters for spectra A and B are presented in Table 2.

¹⁴N ESEEM simulations for the fibrillar Cu²⁺-Aβ(1-40) and soluble Cu²⁺-Aβ(1-16) complexes are shown as dashed lines in Figure 3. Simulations were obtained by using the hybrid optimization protocol in the OPTESIM software.³¹ Simulation parameters are presented in Table 2. Simulation of the ESEEM was optimized by using a minimal set of two coupled ¹⁴N. The simulation parameters for the fibrillar Cu²⁺-Aβ(1-40) and soluble Cu²⁺-Aβ(1-16) complexes are the same, to within the 99% confidence interval, with the exception of the A_xx, parameter, which deviates from overlap of the confidence regions by 0.015 MHz, or 1.3%. The simulations indicate that interactions of the coordinating bis-His-imidazole side chains with Cu²⁺ in the fibrillar Aβ(1-40) and soluble Aβ(1-16) complexes are comparable, and suggest that the physical-chemical properties of the other ligands involved in Cu²⁺ coordination are also comparable.

Table 2.

ESEEM simulation parameters for the Cu²⁺ coupling to the histidine imidazole remote ¹⁴N nuclei in the soluble Aβ(1-16) and fibrillar Aβ(1-40) complexes.

Parameters	Cu2+-Aβ(1–16)		Cu2+-Aβ(1–40)
	Values	99% Confidence Interval	Values	99% Confidence Interval
Axx (MHz)	1.14	1.09-1.19	1.30	1.21-1.39
Ayy (MHz)	1.74	1.70-1.76	1.76	1.70-1.80
Azz (MHz)	2.36	2.35-2.39	2.40	2.36-2.45
e2Qq/h (MHz)	1.59	1.57-1.59	1.57	1.56-1.59
η	0.71	0.70-0.71	0.71	0.70-0.72
αQ (°)	357	354-359	0	−5.40-16.6
βQ (°)	31.4	30.9-33.9	30.7	26.8-34.8
γQ (°)	246	243-251	250	245-256
αA (°)	71.0	61.7-80.1	71.7	57.1-83.7
βA (°)	98.2	87.7-106	89.1	77.9-101.0
γA (°)	1.70	−6.1-6.2	0	−24.0-13.8

The line shape of the 2ν_dq feature depends sensitively on the coordination geometry and relative orientation of the His imidazole rings.³¹ The Euler angles, [α_A β_A γ_A], for the relative orientation of the two ¹⁴N shf principal axis systems (PAS), specify cis-coordination of Cu²⁺ by the two His imidazole in fibrillar Cu²⁺-Aβ(1-40), and also directly confirm the bis-cis-His coordination proposed for soluble Cu²⁺-Aβ(1-16) Component I.^21,22,24 This is illustrated in Figure 4 by the approximately orthogonal x-axes. The x-axis of the dipolar shf PAS lies in the imidazole ring plane, and is directed at an angle of 13°, relative to the imidazole N-N axis.³⁶ The x-axis thus points roughly toward the coordinated Cu²⁺. The z-axis is directed approximately perpendicular to the imidazole ring plane, at an angle of 4° with respect to the ring plane normal.³⁶ Thus, Figure 4 also indicates that the imidazole rings are tilted by about 90°, with respect to each other. The robustness of the bis-cis-His assignment is shown by the dramatically different waveforms and line shapes of the 2ν_dq feature in different geometries (Supporting Information, Figures S1 and S2).

Figure 4.

Model for the mutual orientation of the histidine imidazole remote ¹⁴N dipolar superhyperfine principal axis systems in (A) soluble Cu²⁺-Aβ(1-16) and (B) fibrillar Cu²⁺-Aβ(1-40). Figure axes represent the direction cosines, relative to the reference x, y, z-axis system (brown). The dotted surfaces show the 99% confidence interval, that corresponds to the Euler angles, [α_A, β_A, γ_A].

Influence of Selective ¹⁵N-Substitution of Histidine Residues on ¹⁴N ESEEM from Cu²⁺ Bound to Fibrillar Aβ(1-40)

Figures 2 and 3 show the ESEEM waveforms and cosine FT spectra for the Cu²⁺-Aβ(1-40) complexes with ¹⁵N-imidazole ring-labeled His13 and His14. Under the exact cancellation condition [ν_N=A/2, where ν_N is the nuclear Larmor frequency and A is the shf coupling constant], substitution of ¹⁴N (I=1; nqi present) with ¹⁵N (I=1/2; no nqi) results in a dramatic change in the ESEEM from the coupled nitrogen.³⁴ The dominant isotropic contribution to the shf coupling of the remote imidazole nitrogen is expected to lead to a doublet line shape in the FT spectrum upon ¹⁵N-substitution, with dipolar shf-broadened features near v_N−A/2≈0 and v_N+A/2≈2v_N, that replace the narrow, large-amplitude ν₀, ν₋ and ν₊ nqi lines, and distinctive ν_dq line, that are characteristic of the ¹⁴N shf coupling. In addition, ESEEM amplitudes are dependent on the degree of nuclear state mixing,³⁷ which is strong for ¹⁴N coupling, because of the combined nuclear quadrupole and shf interactions, with generally non-coincident PAS.³⁴ In contrast, mixing is relatively weaker for ¹⁵N coupling. Thus, the ¹⁵N ESEEM may not be resolved at modest signal-to-noise ratios, in the presence of exact cancellation ESEEM from other coupled, ¹⁴N-nitrogen nuclei.³⁸ This is the case in Figures 2 and 3. The predominant effect of ¹⁵N-substitution is, therefore, to eliminate the strong exact cancellation ESEEM from the ¹⁴N that is replaced.³⁸ Thus, the qualitatively comparable ESEEM for the ¹⁵N-imidazole-labeled and natural abundance fibrillar Cu²⁺-Aβ(1-40) complexes eliminates a single equatorial bis-His coordination mode for Cu²⁺, because all ¹⁴N combination features would be lost upon ¹⁵N-substitution on His13 or His14. This is exemplified by simulations of single- and bis-cis- imidazole coordinated ¹⁴N in Supporting Information, Figures S3 and S4. Therefore, a single bis-His coordination mode by the following combinations is ruled out: His6/His13, His6/His14, or His13/His14.

The effects of ¹⁵N-substitution are revealed by division of the waveform from the ¹⁵N-labeled samples by the all-¹⁴N-His waveform. As shown in Figure 5, weak modulation in the quotient waveform is present, with peaks and troughs at positions that correspond to the features in the individual ESEEM waveforms, but with opposite phase. This quotient waveform is consistent with the ¹⁵N-substitution, because the waveform in the numerator of the envelope division includes reduced contributions from coupled ¹⁴N, relative to the waveform in the denominator. The shf and nqi simulation parameters, obtained for the natural abundance Cu²⁺-Aβ(1-40) complex in Table 2, were used to perform an ESEEM simulation analysis of the effect of ¹⁵N-substitution on a minimal model of two equimolar pairs of bis-cis-His-Cu²⁺ sites. The simulated quotient waveform in Supporting Information, Figure S5, which is obtained from the division of a 1:1 mixture of [¹⁵N, ¹⁴N] and [¹⁴N, ¹⁴N] sites by the all-¹⁴N model, shows weak, negative-phase modulation, with opposite-phase peaks and troughs at positions that correspond to features in the individual simulated ESEEM waveforms. The quotient simulation thus agrees with the experimental quotient ESEEM in Figure 5.

Figure 5.

Three-pulse ESEEM waveforms of (A) Fibrillar Cu²⁺-Aβ(1-40), (B) ¹³C,¹⁵N-His13-Cu²⁺-Aβ(1-40), (C) ¹³C,¹⁵N-His14-Cu²⁺-Aβ(1-40), (D) envelope division; B divided by A, (E) envelope division; C divided by A. The vertical scale bar corresponds to 25% of the echo amplitude at τ + T = 8 μs. Aquisition Parameters: microwave frequency, 8.750 GHz, B₀ = 303 mT, T = 6 K, τ = 310 ns.

The results and quotient ESEEM analysis are consistent with the approximately equimolar mixture of Cu²⁺-coordinating bis-His pairs, [His6/His13, His6/His14], in the fibril. This is consistent with the approximately equimolar His6/His13 and His6/H14 contributions in the Component Ia, Ib model proposed for soluble Cu²⁺-Aβ(1-16),²² and other soluble Cu²⁺-Aβ complexes.^24,26

Model for Cu²⁺ coordination in fibrillar Aβ(1-40)

We propose a [His6/His13, His6/His14] model for His coordination of Cu²⁺ in fibrillar Aβ(1-40), as illustrated in Figure 6. The structure of residues 9-40 in the model is based on the Aβ(1-40) fibril structure determined by SS-NMR (Protein Data Bank ID: 2LMN).³ All atoms of residues 15-40 and backbone atoms of residues 9-14 were fixed in the model. Residues 1-8 are disordered (absent) in the 2LMN structure.³ Therefore, the dihedral angles of residues 1-8, and side chain dihedral angles of residues 1-14, were adjusted at the N-termini of the β-strands of 2LMN, according to the following considerations: (1) equatorial coordination by bis-cis-histidine, terminal-amine, and Ala2 carbonyl (as in Scheme 1), and (2) adherence to local steric constraints. The molecular visualization program, PyMOL (www.pymol.org), was used in model building. At the local level of the Cu²⁺ sites, the model is consistent with the paramagnetic quenching of the imidazole ring ¹³Cε and ¹³Cδ features in the CPMAS SS-NMR studies of either ¹³C-labeled- His13 or His14, which was reported as 30-60% (corresponding to 40-70% of narrow-line amplitude maintained).¹⁰ In addition, the model incorporates the symmetry of the imidazole ring ¹³C quenching determined by SS-NMR, which was Cε=Cδ>Cγ, and which indicated that His13 and His14 coordinate Cu²⁺ through the imidazole ring Nε.¹⁰ Although we cannot rigorously rule out Cu²⁺ coordination by a His13/His14 sub-population, as proposed for soluble Aβ(1-16),²¹ a His13–Cu²⁺–His14 complex would severely distort the β-strand structure of the liganding peptide around residues 13 and 14. This is because His13 and His14 are part of the β-strand region in the fibrils, and therefore, the imidazole rings extend from opposite sides of the β-sheet³. This is supported by the conclusions of the SS-NMR study,¹⁰ and TEM and CW-EPR studies,¹⁸ that Cu²⁺ binding in fibrils does not alter the parallel, in-register β-sheet structure.

Figure 6.

Model for the coordination of Cu²⁺ in the fibrillar Aβ(1-40) peptide. The model is based on the quaternary structure for the ordered residues 9-40 of Aβ(1-40), which was determined by SS-NMR³ (Protein Data Bank ID, 2LMN). (A) Protein secondary structure cartoon, showing global view down the fibril axis. The Cu²⁺ ions are represented as orange spheres. (B) Side-on view of the N-terminal region, showing the patterning of Cu²⁺ sites along the fibril axis. Purple and green loops represent His6/His13 and His6/His14 sites, respectively. (C) Local Cu²⁺ coordination site geometry.

The model in Figure 6 is consistent with previous studies,^18,19 which concluded that the Cu²⁺-binding, N-terminal domain of Aβ was structurally independent from the β-sheet-forming central and C-terminal regions. Thus, the coordination geometry and atom-types of the non-His equatorial ligands of the Cu²⁺ site in the soluble complexes with Aβ(1-16) and Aβ(1-28), appear to be maintained in fibrils. We again note that the His13 and His14 side chains are constrained by the β-strand structure of the fibrils to protrude from opposite sides of the β-sheet.^3-5 Therefore, the bis-cis-His6/His13-Cu²⁺ and bis-cis-His6/His14-Cu²⁺ sites are positioned on different β-sheet faces, and this pattern alternates along the fibril axis, as shown in Figure 6. The proposed alternating arrangement of sites separates Cu²⁺ beyond the ~7 Å limit of detectable dipolar broadening of the EPR spectrum.¹⁹ There is no evidence in the CW-EPR spectra for electron spin exchange-induced broadening. The shortest Cu²⁺–Cu²⁺ separation distance in the model is 11 Å, along the direction of the fibril axis. As depicted in Figure 6, this arrangement also suppresses the inter-β-strand coordination of Cu²⁺, which is consistent with earlier evidence against this binding mode,¹⁹ and with the general absence of Cu²⁺ binding effects on the structure of the demetallated fibril,¹⁸ Thus, in our view, the approximately isoenergetic Component Ia and Ib Cu²⁺ binding modes, that are identified in the soluble Aβ(1-16) peptide fragment,^22,23 are a manifestation of the alternating site structure of Cu²⁺ binding in fibrillar full-length Aβ. Owing to the parallel, in-register stacking of the β-turn-β fold of Aβ(1-40) peptides in all reported high resolution fibril structures,^3-5 the alternating pattern of N-terminal Cu²⁺ sites displayed in Figure 6 is expected to be a characteristic feature of Cu²⁺–Aβ(1-40) fibril polymorphs.

Implications of Cu²⁺-His coordination geometry and the patterning of Cu sites in fibrillar Aβ for dioxygen reactivity and ROS generation

Reported pro- and anti-oxidant properties of Aβ are dependent upon the specific measurement conditions and the peptide concentration.^12,14 Any cytotoxicity of the Cu-Aβ complex, relative to the demetallated peptide, is thought to be caused by Cu¹⁺-mediated reduction of dioxygen (O₂), to produce hydrogen peroxide (H₂O₂) and other ROS, which is accompanied by Cu²⁺/Cu¹⁺ redox cycling.^8,12,14,39 The coordination geometry of Cu¹⁺ is bis-trans-His in the soluble monomeric Aβ(1-16) and Aβ(1-40) complexes⁴⁰ and in smaller soluble Aβ peptides.⁴¹ A His13–Cu¹⁺–His14 dyad was assigned.^40,41 The linear coordination mode is consistent with the preference for Cu¹⁺ bis-trans-His geometry in the histidyl-histidine dipeptide⁴² and model imidazole complexes.^43-45 The bis-cis-His Cu coordination by sequence-distant His pairs, in the model in Figure 6, precludes facile Cu¹⁺ coordination by the sequence-adjacent His13-His14 dyad. Thus, Cu²⁺ reduction to linear Cu¹⁺ can be achieved only at large inner sphere^40,46 and outer sphere reorganization energies, which would substantially decrease the rate of reduction. The model in Figure 6 also spatially segregates the His imidazole side chains in each peptide into either [His6/His13, His14] or [His6/His14, His13]. This prevents facile achievement of the distorted-trigonal (or “pseudo-T-shaped”) interaction of a third imidazole ligand with the dyad.^42,43 The distorted-trigonal interaction enhances the sluggish O₂ reactivity of linear His-Cu¹⁺-His in models⁴⁵ and in monomeric Aβ.¹⁷ Thus, factors that would promote reoxidation of a linear His-Cu¹⁺-His, if formed at equilibrium in fibrillar Aβ(1-40),⁴⁷ are absent. As shown in Figure 6, the alternating pattern of Cu sites in fibrillar Aβ(1-40) also constrains formation of inter-peptide ligation of Cu, and maintains a minimum Cu-Cu distance of approximately 11 Å. This eliminates multi-electron redox reactivity and bridged-metal site chemistry.⁴¹ For example, a two-electron, two-proton process is proposed for ROS generation by Cu in soluble Aβ(1-16),¹⁴ and Cu-Cu distances in oxygen-activating enzymes are typically approximately 4.0-4.5 Å.⁴⁸ Thus, in addition to the local Cu site constraints, the global arrangement of Cu sites also acts to suppress Cu²⁺/Cu¹⁺ redox cycling and ROS production. In general, reactivity of the soluble monomer, or non-fibrillar oligomers, of Aβ appears to depend on the prevalent dynamics in these forms.^39,49 Extensive configurational sampling^39,50 will be suppressed in fibrillar forms of full-length Aβ. Based on the model in Figure 5, and the body of results considered above, we propose that the general physiological impact of fibrillar Aβ, at the levels of both the local Cu site coordination structure,^17,40 and global Cu site arrangement, is suppression of Cu redox cycling and ROS production.

The amyloid protein structure provides a robust nanomaterial scaffold for templating ordered arrays of molecules to produce collective functions, including energy and charge transfer.^29,51 Metal deposition on engineered amyloid frameworks has created nanowires with relatively high conductivity.^52,53 The native Cu²⁺-Aβ(1-40) fibril accommodates site-specific, high-density metal ion loading, restricts electronic coupling between metal sites. Mutational and chemical modifications in residues 1-8 could be used to promote metal-metal interations, by introducting accessible alternative conformations, or to integrate additional metal ion binding sites. This suggests applications of Cu²⁺-Aβ fibrils as tunable, and potentially switchable (enabled, for example, by metal ion concentration or pH changes), electron charge/spin conductivity and redox reactivity architectures.

CONCLUSION

We have used ¹⁴N ESEEM spectroscopy to address His coordination of Cu²⁺ in frozen solution samples of fibrillar Aβ(1-40), by measuring the superhyperfine interaction of the unpaired electron spin with the remote imidazole nitrogen nucleus. Simulation analysis of the ligand geometry-dependent 2ν_dq combination line shape reveals a bis-cis-His coordination of Cu²⁺. The ¹⁴N ESEEM from native and selectively ¹⁵N-His-labeled Aβ(1-40) indicates an approximately equimolar mixture of Cu²⁺-coordinating bis-His pairs, [His6/His13, His6/His14], in the fibril. This is consistent with the approximately equimolar His6/His13 and His6/H14 contributions in the Component Ia, Ib model proposed for soluble Cu²⁺-Aβ(1-16),²² and other soluble Cu²⁺-Aβ complexes.^24,26 The ESEEM results for fibrillar Cu²⁺-Aβ(1-40), and control Cu²⁺-Aβ(1-16), in agreement with the previous EPR studies,^18,19 thus indicate that the Cu²⁺ site in fibrillar Aβ(1-40) has a comparable geometry and set of equatorial ligand atoms, as for Component I in Cu²⁺-Aβ(1-16). We propose a [His6/His13, His6/His14] model for His coordination of Cu²⁺ in fibrillar Aβ(1-40), in which residues 1-8 are configured in accord with the EPR and ESEEM-derived constraints, and residues 9-40 assume the β-sheet structure determined by SS-NMR.³ The fibril context dictates that the Cu²⁺-bis-cis-His6/His13 and Cu²⁺-bis-cis-His6/His14 sites are positioned on different β-sheet faces, and that this pattern alternates along the fibril axis. The alternating arrangement of sites separates Cu²⁺ beyond the ~7 Å limit of detectable dipolar broadening of the EPR spectrum.¹⁹ The local intra-β-strand coordination structure is not conducive to Cu²⁺/Cu¹⁺ redox-linked coordination changes, and the global arrangement of Cu sites precludes facile multi-electron and bridged-metal site reactivity. It is therefore concluded that the fibrillar form of Aβ suppresses Cu redox cycling and reactive oxygen species (ROS) production.

Abbreviations

Aβ

amyloid-β protein

AD

Alzheimer's disease

CPMAS

Carr-Purcell magic angle spinning

CW

continuous-wave

EPR

electron paramagnetic resonance

ESEEM

electron spin echo envelope modulation

EXAFS

Extended X-ray absorption fine structure

FT

Fourier transform

HYSCORE

hyperfine sublevel correlation

nqi

nuclear quadrupole interaction

OPTESIM

optimization of ESEEM simulation

PAS

principal axis system

shf

superhyperfine

ROS

reactive oxygen species

SS-NMR

solid-state nuclear magnetic resonance

TEM

transmission electron microscopy