Copper-mediated Amyloid-β Toxicity Is Associated with an Intermolecular Histidine Bridge*

Amyloid-β peptide (Aβ) is pivotal to the pathogenesis of Alzheimer disease. Here we report the formation of a toxic Aβ-Cu2+ complex formed via a histidine-bridged dimer, as observed at Cu2+/peptide ratios of >0.6:1 by EPR spectroscopy. The toxicity of the Aβ-Cu2+ complex to cultured primary cortical neurons was attenuated when either the π -or τ-nitrogen of the imidazole side chains of His were methylated, thereby inhibiting formation of the His bridge. Toxicity did not correlate with the ability to form amyloid or perturb the acyl-chain region of a lipid membrane as measured by diphenyl-1,3,5-hexatriene anisotropy, but did correlate with lipid peroxidation and dityrosine formation. 31P magic angle spinning solid-state NMR showed that Aβ and Aβ-Cu2+ complexes interacted at the surface of a lipid membrane. These findings indicate that the generation of the Aβ toxic species is modulated by the Cu2+ concentration and the ability to form an intermolecular His bridge.

bound forms of A␤ and may be central to the pathological mechanism of AD (reviewed by Bush (1)).
The coordination environment of Cu 2ϩ bound to A␤ has been studied by EPR, Raman, and NMR spectroscopies. The consensus is that the coordination of Cu 2ϩ to A␤ is via the three His residues His 6 , His 13 , and His 14 and an as yet undefined fourth ligand; options put forward include tyrosine (Tyr 10 ) (2)(3)(4)(5)(6), the N-terminal nitrogen (7), or an as yet unidentified carboxylate side chain (8,9) (Fig. 1a). Recently, Glu 11 has been identified as providing the carboxylate side chain when Zn 2ϩ is bound to A␤ 1-16 (10). The EPR studies of Curtain et al. (2) showed that in the pH range 6.0 to 7.5 in PBS, an increase in the Cu 2ϩ /peptide molar ratio above ϳ0.3:1 broadened the mononuclear Cu 2ϩ spectrum observed at low ratios. At Cu 2ϩ /peptide molar ratios greater than 0.6:1, a distinctive broad resonance developed near g ϳ 2. This broad resonance was attributed to exchange interactions between two Cu 2ϩ ions ϳ6 Å apart, with the second Cu 2ϩ ion being coordinated to A␤ in a cooperative manner via a His bridge (Fig. 1b), similar to that observed by Ohtsu et al. (11) for Cu 2ϩ bridge imidizolate complexes. Here we report the first observation of resonances at g ϳ 4 in the EPR spectra of A␤-Cu 2ϩ peptides, where the broad g ϳ 2 line is also observed. Such EPR spectra are diagnostic for the existence of binuclear Cu 2ϩ centers where the Cu 2ϩ ions are coupled by dipolar interactions; simulations presented here demonstrate that these atoms are 6.2 Ϯ 0.2 Å apart (12). This distance is consistent with a His bridge between the Cu 2ϩ atoms. To further the understanding of the role played by the His bridge and A␤-lipid interactions in the neurotoxicity of A␤, we examine the toxic mechanism of both metal-free and Cu 2ϩ -bound forms of the A␤ peptide.
The link between His bridge formation and neurotoxicity was explored through experiments using A␤ 1-42 peptides modified by the methylation of the His residues at either theor -nitrogen of the imidazole side chain (Fig. 1c). Correspondingly modified A␤ 1-40 has been shown to be nontoxic, despite being redox-active in a manner similar to the wild type (WT) peptide (4). This lack of toxicity has been attributed to these peptides being unable to bind to cell surface membranes, congruent with the notion that A␤-lipid membrane interactions are central to the cause of neurotoxicity in AD (13)(14)(15)(16)(17). Here we describe the Cu 2ϩ coordination, lipid interactions, and aggregation properties of both the neurotoxic WT A␤ 1-42 peptide and A␤ 1-42 where the His residues have been methylated at either theornitrogen of the imidazole side chain. The Cu 2ϩ -bound WT A␤ 1-42 peptide was shown to be significantly more toxic, as monitored by the viability of primary cortical neurons, under conditions in which the His bridge was formed. Perturbation of this metal coordination mode leads to a loss in toxicity even in the presence of Cu 2ϩ . We also show that toxicity correlates with the ability of the peptides to form dityrosine and induce lipid peroxidation. Toxicity displays a negative correlation with the formation of amyloid material and the ability to perturb the acylchain region of lipid bilayer. These results show that the ability to form a Cu 2ϩ -linked His bridge dimer, resulting ultimately in lipid peroxidation, is central to the metal-meditated toxicity of A␤.

EXPERIMENTAL PROCEDURES
All reagents were purchased from Sigma unless otherwise stated. Peptide Synthesis-The uncommon N im -protected amino acids, Fmoc-L-His(N-1-Me)-OH (His--Me) and Fmoc-L-His(N-3-Me)-OH (His--Me), were purchased from Bachem (Bubendorf, Switzerland); all were used without further purification. All modified peptides were synthesized according to methods outlined previously (18). Due to the large degree of similarity between the peptides, large batches of resins were prepared and split at appropriate points in the synthetic pathway such that each peptide was prepared on a scale of 0.03 mmol. WT A␤ 1-42 was purchased from the W. M. Keck Laboratory (Yale University, New Haven, CT).
Peptide and Aggregate Preparation-PBS is defined as 10 mM sodium phosphate, 150 mM NaCl at pH 7.4. Cu 2ϩ was present as a solution of 1 part CuCl 2 to 6 parts glycine. The addition of a glycine counterion was essential in order to prevent the formation of insoluble phosphate-metal complexes. A 25 mM stock solution was made and diluted in deionized water to the desired concentration as required. Peptide working solutions were prepared by taking a known amount of A␤ 1-42 and then dissolving it in 1,1,1,3,3,3-hexafluoro-2-isopropanol (HFIP). The peptide was incubated at 25°C for 1 h to remove any preformed structure. Known amounts of peptide were then aliquoted out, and HFIP was removed by evaporation. The resulting peptide film was stored at Ϫ80°C until required. The film was dissolved in 20 mM NaOH and then diluted out in deionized water and 10ϫ PBS at a v/v/v ratio of 2:7:1. All solutions were sonicated in a water bath for 10 min and centrifuged for 10 min at 4°C at high speed before the addition of 10ϫ PBS. The supernatant was kept on ice for immediate use. The absorbance value at 214 nm was measured, and the concentration of the peptides was determined using a molar extinction coefficient value of 75,887 liters/mol/ cm, adjusting to the required concentration with buffer A. CuCl 2 /Gly was added to the required concentration from a 25 mM stock. Samples were used immediately or incubated at 37°C with agitation at 200 rpm for either 2 or 24 h. EPR Spectroscopy-Continuous wave EPR spectra were obtained with Bruker ESP380E FT/CW and ECS106 X-band spectrometers. Temperatures around 120 K in both were achieved with a Bruker nitrogen gas flow insert in the standard rectangular TE 012 cavity. Spectra acquired at 2.5 K using the ESP380E were achieved using an Oxford Instruments CF935 cryostat with an ER4118 cylindrical cavity insert. The microwave frequency was measured with an EIP microwave 548A frequency counter, and the g factors were determined with reference to the F ϩ line in CaO (19). Spectrum simulations were performed with the SOPHE software (version 1.1.4) described by Hanson et al. (20). For the EPR measurements, the chloride of 99.99% pure 65 Cu (Cambridge Isotopes) was used. The peptides were dissolved from an HFIP-treated stock to 100 M in deionized water, and 65 CuCl 2 to the desired Cu 2ϩ / peptide molar ratio was added immediately from a 2.0 mM stock solution via a glass microsyringe. The pH was then adjusted to 7.4 by adding concentrated PBS or ethylmorpholine buffer, and after mixing, samples were rapidly transferred to Wilmad "Suprasil" EPR tubes and frozen immediately. Before and after the EPR spectroscopy, metal concentrations were measured by inductively coupled plasma mass spectrometry, and peptide concentrations were determined by quantitative amino acid analysis.
High Performance Immobilized Metal Ion Chromatography-All solutions were delivered to the column at a flow rate of 1.5 ml/min. The high performance immobilized metal ion chromatography column (Amersham Biosciences) was swollen in PBS (pH 7.4), washed in EDTA (50 mM) to remove trace metals, and then washed with deionized H 2 O. The column was charged with Cu 2ϩ by injecting 2 ml of 0.1 M CuCl 2 / glycine(1:6). The excess metal was removed by washing the column with deionized H 2 O. The column was then re-equilibrated in PBS (15 ml). The peptides were prepared at 1 mM in 10% 2Ј,2Ј,2Ј-trifluoroethanol/ PBS, and 5 l was injected onto the column. The peptides were eluted using either a pH gradient (7.4 -2.5) generated by a linear increase (0 -100%) in PBS at pH 2.5 or an imidazole gradient (0 -100% 500 mM imidazole in PBS at pH 7.4).
Primary Neuronal Cultures-Cortical neuronal cultures were prepared as described previously (21). Briefly, embryonic day 14 BL6Jx129sv mouse cortices were removed, dissected free of meninges, and dissociated in 0.025% (w/v) trypsin in Krebs buffer. The dissociated cells were triturated using a filter-plugged fine pipette tip, pelleted, resuspended in plating medium (minimum Eagle's medium, 10% fetal calf serum, 5% horse serum), and counted. Cortical neuronal cells were plated into poly(Llysine)-coated 48-well plates at a density of 125,000 cells/well in plating medium. All cultures were maintained in an incubator set at 37°C with 5% CO 2 . After 2 h, the plating medium was replaced with fresh neurobasal medium containing B 27 supplements, Geneticin, and 0.5 mM glutamine (all tissue culture reagents were purchased from Invitrogen unless otherwise stated). This method resulted in cultures highly enriched for neurons (Ͼ95% purity) with minimal astrocyte and microglial content as determined by immunostaining of culture preparations using specific marker antibodies.
Cell Viability Assay-The neuronal cells were allowed to mature for 6 days in culture before commencing treatment using freshly prepared serum free neurobasal medium plus B 27 supplements minus antioxidants. For the treatment of neuronal cultures, peptide was prepared at 50 M at Cu 2ϩ /peptide molar ratios of 0:1, 0.1:1, 1:1, and 10:1 and used immediately. Peptides were diluted to the final concentration of 5 M in neurobasal medium. The mixtures were then added to neuronal cells for 96 h. Cell survival was monitored by phase-contrast microscopy, and cell viability was quantitated using the MTS assay as described previously (21). Briefly, the medium was replaced with fresh neurobasal and a yet to be identified oxygen atom. b, the imidazole side chain of a His residue bridging between Cu 2ϩ atoms generating A␤ dimer and a metal coordination site similar to the active site of SOD 1 (25). c, methylation of the imidazole side chain of His at either the or position, designed to allow the initial copper binding event to occur but inhibit the subsequent formation of His-bridged structures. medium supplemented with B 27 lacking antioxidants, and 10% (v/v) MTS (Promega, Madison, WI) was added to each well and incubated for 3 h at 37°C in a 5% CO 2 incubator. Plates were gently shaken, and a 150-l aliquot from each well was transferred to separate wells of a 96-well plate. The color change of each well was determined by measuring the absorbance at 490 nm using a Wallac Victor Multireader, and background readings of MTS incubated in cell-free medium were subtracted from each value before calculations. The data were normalized and calculated as a percentage of untreated vehicle control values. Data are shown as mean Ϯ S.E. Statistical comparisons between groups were done using Student's t test.
Quantification of Aggregate and Fibril Formation-For determination of fibril growth end points, a discontinuous assay of fibril growth was used. Peptide samples (20 l) were removed from incubation after 2 or 24 h and added to 600 l of a 20 M thioflavin-T (ThT) solution at pH 7.4 in buffer A. The ThT signal was quantified by averaging the fluorescence emission at 480 nm over 10 s when excited at 444 nm using a PerkinElmer Life Sciences LS55 fluorescence spectrophotometer, slit widths for excitation of 2.5 nm and emission of 12.5 nm at 37°C.
Circular Dichroism Spectroscopy-The peptides were prepared as described above diluting to a final concentration of 20 M in PBS to achieve a measurable CD signal. Cu 2ϩ was titrated into the peptide preparation at 0.5-2.5 peptide mol eq after diluting the peptide, gently mixed, and allowed to equilibrate for 15 min at room temperature. Far UV-CD spectra were collected between 200 and 250 nm at room temperature on a Jasco 810 spectrometer using a quartz cuvette with a 1-cm path length. Measurements were recorded at 50 nm/min with a bandwidth of 1 nm and a response time of 2 s, averaging 10 accumulation scans per measurement. The spectra were background-subtracted, smoothed using the Jasco software (version 32) FFT filter function, and converted to molar ellipticity. All experiments were performed in triplicate.
Dityrosine Formation-Peptide was prepared at 10 M in the presence of 250 M H 2 O 2 at Cu 2ϩ /peptide molar ratios of 1:1. After 24 h of incubation, the reaction was quenched by the addition of 10 l of 12.5 mM EDTA (final concentration 250 M). Dityrosine content was quantified using a PerkinElmer Life Sciences LS55 fluorescence spectrophotometer with an excitation wavelength of 320 nm and scanning the emission wavelengths from 350 to 500 nm. Maximum light signal was observed and recorded at 418 nm (22).
Preparation of Large Unilamellar Vesicles-Large unilamellar vesicles (LUVs) were prepared by initially dissolving equal quantities of 1-palmitoyl-2-oleoyl-sn-glycero-3-phosphocholine (PC) and 1-palmitoyl-2-oleoyl-sn-glycero-3-[phospho-L-serine] (PS) (Avanti Polar Lipids Inc.) in chloroform. The chloroform was evaporated off, and the lipids were resuspended in buffer A at ϳ10 mM and incubated at 37°C for 1 h with agitation at 200 rpm in the presence of glass beads. The lipid mixture was subjected to five freeze-thaw cycles using liquid nitrogen and a 37°C water bath. The sample was then passed through an extruder apparatus using a 100-nm filter, resulting in a uniform solution of LUVs with a mean diameter between 120 and 140 nm.
Steady-state Polarization of DPH-Diphenyl-1,3,5-hexatriene (DPH) was purchased from Molecular Probes. Briefly, DPH was dissolved at 10 mM in Me 2 SO and diluted to 16.6 M in buffer A containing 416 M LUVs. A 60-l aliquot of this sample was mixed with 40 l of A␤ 1-42 at 10 M, giving final concentrations of 10 M probe, 250 M LUVs, and 4 M peptide. Samples were then incubated at 37°C with aggregation at 200 rpm for 20 min prior to reading. The dye was excited at 359 nm, and emission was recorded at 430 nm. Anisotropy was measured on a Varian Eclipses spectrophotometer using internal polarizers and calculated by the equation, r ϭ (Ivv Ϫ G⅐Ivh)/(Ivv ϩ 2G⅐Ivh), where Ivv and Ivh are the fluorescence intensities when the excitation and emission polarizers are set at v (vertical) and h (horizontal), respectively. The grating factor G ϭ Ihv/Ihh. Solid-state NMR-Solid-state NMR experiments were conducted on a Varian Inova 300 spectrometer operating at a resonance frequency of 121.4 MHz for 31 P. 31 P spectra were acquired at 28°C, using a Doty 5-mm double resonance NMR probe with a standard Hahn echo and proton decoupling. Pulse times used were /2 ϭ 5 s pulse with a delay of 2 s under magic angle spinning of ϳ5 kHz. Typically, 10,000 scans were acquired, and an exponential line broadening of 10 Hz was applied. Chemical shift was referenced using 80% phosphoric acid (H 3 PO 4 ) as 0 ppm for 31 P NMR experiments.
Lipid Peroxidation-20 M peptide was prepared as above at a Cu 2ϩ / peptide molar ratio of 0:1 or 1:1. Samples were incubated for 5 min before the addition to 50% PC, 50% PS LUVs and ascorbate. Samples contained a final concentration of 10 M peptide, 1 mM LUV, and 10 M ascorbate and were incubated for 24 h at 37°C with agitation at 200 rpm. CQ was prepared in 100% Me 2 SO and added to final concentrations of 2, 20, and 50 M. The final Me 2 SO concentration was 1% in all cases. The level of lipid peroxidation was quantified via a lipid peroxidation colorimetric assay (Oxford Biomedical Research) as per the manufacturer's instructions.

RESULTS
The His Bridge Dimer as Observed by EPR-Most of the experiments described below were performed at pH 7.4 in PBS solutions, because NaCl at a concentration of ϳ0.15 M is part of the physiological milieu. Under these conditions of pH, peptide concentration, and buffer, the predominant species observed by EPR depended on the Cu 2ϩ /peptide molar ratio. The addition of 65 CuCl 2 to A␤ 1-16, A␤ 1-28, and WT A␤ 1-42 to give Cu 2ϩ /peptide molar ratios of 0 -0.5:1 gave EPR signals analogous to that shown in Fig. 2a. These spectra are similar to those reported previously (2, 3) for low Cu 2ϩ /peptide ratios and show a spectrum mainly due to mononuclear Cu 2ϩ in an approximately axially symmetric square planar environment. A simulated spectrum using the spin-Hamiltonian parameters is given in supplemental Table 1. These parameters are typical for Cu 2ϩ coordinated by 2N2O or 3N1O (23).
At Cu 2ϩ /peptide molar ratios of above 0.5:1, a significantly different resonance developed in the g ϳ 2 region, as reported previously (2,4) and as shown in Fig. 2b. Concurrently, we demonstrate here for the first time that a resonance also develops at g ϳ 4 when very high spectrometer gains are used (Fig. 3). The parallel appearance of the g ϳ 2 and g ϳ 4 resonances is diagnostic for the existence of a binuclear Cu 2ϩ center (i.e. a pair of Cu 2ϩ ions coupled by exchange and/or dipolar interactions) (12). Although the Cu 2ϩ hyperfine structure is not well resolved, its presence is confirmation that the resonance at g ϳ 4 is due to a pair of Cu 2ϩ ions and not to a nonspecific aggregate. At Cu 2ϩ /peptide molar ratios of ϳ1:1 and greater, another broad resonance developed centered at g ϳ 2.14 ( Fig. 2c). No resonance at g ϳ 4 appears to be associated with this new signal, which may be attributed to a number of Cu 2ϩ ions within ϳ10 Å of each other but not structurally related. The broken line spectrum in Fig. 2c represents the simulation of an isotropic resonance at g ϳ 2.14, such as might arise from a number of Cu 2ϩ ions within ϳ10 Å of each other without a specific geometrical relationship and where the anisotropic features are averaged by a combination of dipolar and weak exchange interactions.
Spectra acquired at 2.5 K of A␤ 1-28 (Cu 2ϩ /peptide molar ratio 1:1) exhibited the same resonances as observed at 120 K at g ϳ 2 and g ϳ 4 but with 50 times the intensity, indicating that antiferromagnetic exchange interactions between the Cu 2ϩ ions of the dimer must be Ͻ Ͻ10 cm Ϫ1 . The dimer spectra did not exhibit microwave power saturation up to 25 milliwatts at 2.5 K, indicating the existence of spin-spin relaxation processes as expected for a coupled system. In contrast, the mononuclear spectra exhibited power saturation effects above ϳ1 milliwatt at 2.5 K. To verify that the g ϳ 2 and g ϳ 4 spectra were not due to oxo-or phosphate-bridged binuclear Cu 2ϩ centers or to adventitious Fe 3ϩ , control spectra were acquired at 120 K of 1.5 mM 65 CuCl 2 in PBS, of PBS alone, and of the peptide in PBS prior to the addition of 65 CuCl 2 . At pH 7.4, no resonances were observed near either g ϳ 2 or g ϳ 4.
As a comparison with WT A␤ 1-42, the spectra of A␤ 1-42 containing His--Me or His--Me at positions 6, 13, and 14 were recorded over the range of Cu 2ϩ /peptide molar ratios 0.3-1:1. Only mononuclear spectra were observed, similar to those previously published by Tickler et al. (4). There were no indications of the broad resonance at g ϳ 2 characteristic of the binuclear Cu 2ϩ center or for resonances near g ϳ 4, even at spectrometer settings where the g ϳ 4 resonances were clearly observed in the unmodified peptides. Fig. 2d shows the mononuclear spectra obtained with WT A␤ 1-42 at a Cu 2ϩ /peptide molar ratio of 0.1:1 and the similarly mononuclear spectra of His--Me obtained at a Cu 2ϩ / peptide molar ratio of 1:1 (Fig. 2e).
High pressure immobilized affinity chromatography was employed to ascertain if the methylation of the imidazole side chains had a significant effect on the relative Cu 2ϩ binding affinity compared with WT A␤ 1-42. The strength of peptide-metal ion interactions is usually assessed as a function of the pH required to elute the various peptides from a Cu 2ϩ -loaded immobilized metal ion affinity chromatography column. However, we were unable to elute any of the peptides from the Cu 2ϩcharged immobilized metal ion affinity chromatography column at pH Ͻ2 or by using an imidazole gradient. The peptides could only be eluted by the addition of EDTA to the eluting buffer, indicating that all peptides had a very high affinity for Cu 2ϩ .
The effect of buffer conditions on A␤-Cu 2ϩ coordination was investigated by acquiring spectra of A␤ 1-28 in pH 7.4 ethylmorpholine buffer over a range of Cu 2ϩ /peptide molar ratios, recreating the conditions of Syme et al. (7) (supplemental Fig. 1). At Cu 2ϩ /peptide molar ratios of 0.1:1 and 1:1, only mononuclear spectra where observed. Spectra in the g ϳ 2 region were identical to those reported at Cu 2ϩ /peptide ratios of Յ1:1 by Syme et al. (7). The broad line near g ϳ 2 was not observed, even at Cu 2ϩ concentrations and spectrometer settings that yielded these resonances in PBS. Therefore, A␤-Cu 2ϩ complexes in an ethylmorpholine buffer do not generate the specific binuclear Cu 2ϩ centers observed in PBS at Cu 2ϩ /peptide molar ratios of Յ1:1.
Observation of the His Bridge as Deduced from EPR Spectroscopy-The EPR spectra due to pairs of Cu 2ϩ ions separated by distances of around 3-7 Å were first described 30 years ago (12). These spectra arise from the exchange and/or dipolar coupling of two Cu 2ϩ ions, each with spin S ϭ 1 ⁄ 2, to give a singlet state (S ϭ 0) and a triplet state (S ϭ 1). "Allowed" electron spin transitions within the triplet state,  where the magnetic spin quantum number M changes by 1 (⌬M ϭ Ϯ1), occur near g ϳ 2, whereas "forbidden" or "half-field" transitions, where ⌬M ϭ Ϯ2, occur near g ϳ 4. Thus, the broad line at g ϳ 2 and the resonance at g ϳ 4 are diagnostic for a pair of Cu 2ϩ ions coupled by exchange and/or dipolar interactions. Provided that second order exchange is negligible, the zero field splitting of the triplet state is due to dipolar interactions alone, and its magnitude depends on 1/R 3 , where R is the internuclear distance. The distance between the Cu 2ϩ ions can be calculated from the zero field splitting of the triplet state and can be estimated by the use of computer simulations.
The measurements at 2.5 K show that antiferromagnetic exchange interactions are Ͻ Ͻ10 cm Ϫ1 and therefore that the effect of second order exchange on the zero field splitting of the triplet state can be neglected. The broken line spectra shown in Figs. 2 and 3 were simulated, assuming that the zero field splitting of the triplet state is due to dipolar coupling alone. The parameters giving the best fits are listed in supplemental Table 1. The fits were obtained by successively varying the geometrical and spin Hamiltonian parameters and line widths until agreement between the field positions of peaks in the experimental and computed spectra was within ϳ10 G. Differences between simulated and experimental spectra are due to line width variations, and in the case of the g ϳ 2 spectra to there being an equilibrium between different Cu 2ϩ -bound states of the peptide, with the simulation representing the predominant species. The simulation of the g ϳ 4 resonance is presented in supplemental Fig. 2 with a smaller line width in order to more clearly delineate the positions of the Cu 2ϩ hyperfine peaks. All fitting attempts converged to the same range of parameters required to fit both the ⌬M ϭ 1 and ⌬M ϭ 2 spectra. The values of g ʈ ϭ 2.34 Ϯ 0.01 and g Ќ ϭ 2.12 Ϯ 0.01 are larger than expected for 2-or 3-nitrogen coordination but indicate that the Cu 2ϩ coordination geometry is distorted toward tetrahedral (24).
Therefore, the picture that emerges is of one involving two structurally identical Cu 2ϩ sites, probably 3N-coordinated, in a tetrahedrally distorted square planar configuration. The Cu 2ϩ ions are 6.2 Ϯ 0.2 Å apart, with axially symmetric g and A matrices, but with the symmetry axes tilted toward the internuclear vector and toward each other by 15 Ϯ 5°. The 6.2 Ϯ 0.2-Å distance between the Cu 2ϩ atoms is consistent with our hypothesis that Cu 2ϩ coordination by A␤ at 1:1 Cu 2ϩ /peptide occurs via a His bridge between two metal centers to generate a structure that resembles the active site of superoxide dismutase 1 (25). The formation of such a bridge is inhibited by the methylation of either of the nitrogen atoms from the imidazole side chain.
Toxic Properties of A␤-Cu 2ϩ Complexes-To ascertain the possible physiological significance of the His bridge, A␤ 1-42 (5 M) at Cu 2ϩ / peptide molar ratios of 0:1, 0.1:1, 1:1, and 10:1 was tested for neurotoxic activity on mouse primary cortical cultures (Fig. 4a). Only A␤ 1-42 prepared at a Cu 2ϩ /peptide molar ratio of 1:1 and 10:1 showed neurotoxicity at 5 M, with the 1:1 ratio the most toxic. Toxicity was rescued by the addition of the H 2 O 2 scavenging enzyme catalase, indicating that production of ROS is a plausible toxic insult. It therefore appears that conditions leading to the formation of His-bridged A␤ oligomers are associated with metal ion-mediated toxicity.
It has previously been shown that metal-free forms of the similarly methylated A␤ 1-40 peptide are nontoxic (4). To investigate the effect of the loss of the His bridge on the toxicity of the metal-bound form of the A␤ peptide, WT A␤ 1-42, His--Me, and His--Me were added to mouse primary cortical cultures at Cu 2ϩ /peptide molar ratios of 0:1 and 1:1 (Fig. 4b). Under these conditions, the WT A␤ 1-42 at a Cu 2ϩ /peptide molar ratio of 1:1 reduced cell viability to ϳ70% as compared with the vehicle. In contrast, both His--Me and His--Me were not toxic. The EPR data show that no His bridge was induced by Cu 2ϩ in either His--Me or His--Me at Cu 2ϩ /peptide molar ratios of Յ1:1. The ability of the A␤ peptide to induce toxicity therefore correlates with the formation of the His-bridged dimers of A␤ as observed by EPR.
Cu 2ϩ Modulation of A␤ Amyloid Formation-Our data suggest that the ability of A␤ 1-42 to form a His-bridged Cu 2ϩ -induced dimer was required for metal-mediated toxicity. In order to determine if there is a correlation between the ability of the peptides to form amyloid material and this toxic species, WT A␤ 1-42, His--Me, and His--Me were incubated at 37°C for 2 h at Cu 2ϩ /peptide molar ratios of 0:1 to 25:1. The amyloid content was assessed by use of ThT fluorescence, and the morphology was assessed by electron microscopy. In the absence of added Cu 2ϩ , all peptides showed an increase in ThT fluorescence, indicating the formation of amyloid material (Fig. 5). Comparable ThT intensities and peptide aggregate morphologies where observed after 24 h of incubation (data not shown). It is interesting to note that His--Me produced a 2-fold increase in ThT signal over both WT A␤ 1-42 and His--Me. This increase in signal may be due to subtle changes in the morphology of the fibril leading to a greater binding capacity for ThT or an increase in the total amount of amyloid material. As the Cu 2ϩ concentration was raised, all three peptides displayed a loss in amyloidogenic material, evident as a decrease in ThT fluorescence. The Cu 2ϩinduced shifts in the aggregation profile from amyloid material to amorphous aggregates centered close to a Cu 2ϩ /peptide molar ratio of 1:1 and were evident for all three peptides. Electron microscopy morphological analysis of all aggregates (data not shown) demonstrated that the decrease in ThT signal corresponded to a loss in fibrillar material and a gain in amorphous aggregate material ranging in size from 10 nm to 1 m. His--Me does display a small amount of ThT fluorescence even in the presence of high Cu 2ϩ concentration (Fig. 5), and concurrently small fibrillar-like aggregates were observed by electron microscopy (data not shown). Nevertheless, all three peptides displayed a loss in amyloid material as the Cu 2ϩ concentration rose, and it appears that Cu 2ϩ -induced aggregation of A␤ is not dependent on the ability of the peptide to form a His bridge but on its ability to coordinate the initial Cu 2ϩ . It is also clear that conditions promoting the formation of amyloid material do not lead to metal ion-mediated toxicity. CD spectra were recorded for all peptides to determine whether the modifications significantly altered the structural properties of the peptide in the presence and absence of Cu 2ϩ . All three unaged peptides formed ␤-sheet structures in the presence of Cu 2ϩ , with the two modified peptides having a greater ␤-sheet propensity than the WT peptide (supplemental Fig. 3).
Cu 2ϩ Modulation of A␤ Lipid Interactions-Previous work by our group (4,21,26) and others (13)(14)(15)(16)(17) has indicated that the primary toxic mechanism of A␤ requires an interaction with the plasma membrane of cells. In order to assess the ability of A␤ 1-42 and the variant peptides to penetrate the lipid membrane, the change in fluidity of the acyl-chain region of LUVs was monitored by the anisotropy of DPH. DPH is known to intercalate axially to the acyl-chain region of a lipid bilayer, and it reports on the flexibility of this hydrophobic environment (27). Both aged and unaged WT A␤ 1-42 at Cu 2ϩ /peptide molar ratios of 0:1 and 1:1, corresponding to the absence or presence of the His bridge, were allowed to associate with LUVs composed of 50% PC and 50% PS. The steady-state anisotropy of DPH was then measured after 20 min of incubation (Fig. 6). When unaged WT A␤ 1-42 was allowed to interact with the lipid membrane, there was little change in the fluidity of the acylchain region over buffer alone (Fig. 6a). However, upon aging for 24 h, which resulted in the formation of amyloid material, there was a large decrease in the fluidity of the acyl-chain region as observed by the increase in anisotropy, representing a perturbation of the membrane bilayer by the amyloidogenic material. The nonamyloidogenic Cu 2ϩ aggregates produced at a Cu 2ϩ /peptide ratio of 1:1 did not by contrast perturb this region of the bilayer. A similar pattern of lipid perturbation was observed with His--Me and His--Me, where only the presence of amyloid material leads to a perturbation of the acyl-chain region of the membrane. It is notable that Cu 2ϩ aggregates created by each of the variant peptides do not lead to a change in the fluidity of the acyl-chain region of the membrane. Whereas the magnitude of the fluorescence signal of DPH was slightly quenched by Cu 2ϩ (Ͻ10 -15%), the anisotropy of the fluorophore was not affected by the presence of the metal ion, since the signal recorded in buffer only is identical in the presence or absence of additional Cu 2ϩ ions. Also, because anisotropy is a ratio of the intensity of the polarized light to the total light, it varies only with the motility of the fluorophore. These observations would suggest that mechanical interference of the acyl-chain region of the lipid membrane by the A␤ 1-42 peptide does not contribute to metal ionmediated toxicity.
To ascertain if A␤ and A␤-Cu 2ϩ complexes were interacting with the head group region of the lipid membrane, experiments were carried out by 31 P magic angle spinning solid-state NMR. The 31 P solid-state NMR spectra of the phospholipid head group of the LUV showed a broad  resonance at Ϫ0.75 ppm (Fig. 7a, arrow 2); the addition of A␤ 1-42 resulted in the appearance of a new resonance at 2.51 ppm (Fig. 7b,  arrow 3). The appearance of the second peak indicated the formation of a lipid-peptide-associated phase. When A␤ 1-42-coordinating Cu 2ϩ was added to LUV, the A␤ 1-42-associated lipid signal was reduced significantly and broadened (Fig. 7c), consistent with the paramagnetic Cu 2ϩ being bound to A␤ 1-42 and in close association with the phospholipid head group. The Cu 2ϩ bound to A␤ 1-42 caused paramagnetic relaxation of the peptide-bound peak and T 2 broadening of peaks due to both PC and PS lipids. This indicated that the Cu 2ϩ was interacting preferentially with the A␤-phospholipid associated phase rather than the nonassociated phase. This supports our previous cortical cell culture studies, where lipid-peptide binding appears to be crucial for cell toxicity and suggests that phospholipid head group interaction plays an important role in the pathway for neurotoxicity (4,21,26,28). It is also worth noting that there is a small shift in the peak due to the phosphate buffer upon the addition of A␤, from 1.76 to 1.26 ppm, with a further shift to 1.00 ppm upon the addition of Cu 2ϩ . Such shifts are consistent with the concept that the buffer plays a role in determining the structure of the peptide as outlined above.
Dityrosine Formation Correlates with the Ability to Form a His Bridge-Tyr 10 is a critical residue in the production of ROS by A␤ (29), and the resulting formation of dityrosine adducts has been suggested to play a role in A␤ aggregation and neurotoxicity (22,29). Dityrosine adducts of A␤ are formed as a by-product when the peptide is incubated in the presence of Cu 2ϩ and H 2 O 2 (22,29). The addition of H 2 O 2 strongly promotes dityrosine cross-linking via the increased production of ROS (22,29). The role of the His bridge in the generation of dityrosine adducts was assessed by assaying the formation of dityrosine by WT A␤ 1-42, His--Me, and His--Me at a Cu 2ϩ /peptide molar ratio of 1:1 in the presence of 250 mM H 2 O 2 (Fig. 8). In comparison with WT A␤ 1-42, His--Me and His--Me showed a marked decrease in the ability to generate dityrosine adducts (Fig. 8). This lack of dityrosine adducts reflects a loss in the ability to form the His bridge between Cu 2ϩ atoms.
Since tyrosine radicals are relatively short lived, it is likely that the formation of a His bridge between two Cu 2ϩ atoms bound to two different A␤ peptides facilitates the covalent cross-linking by two tyrosine radicals to generate a dityrosine moiety. The ability of the A␤ peptides to form dityrosine adducts (Fig. 8), therefore, correlates with metaldependent toxicity, as shown in Fig. 4b.
Cu 2ϩ Modulation of A␤-induced Lipid Peroxidation-Lipid peroxidation is a well established mechanism of cellular injury and an indicator of oxidative stress that has been implicated in the pathogenesis of AD (15). Peroxidation of unsaturated fatty acids produces lipid peroxides that are unstable and decompose to form a series of compounds, including reactive carbonyl compounds. To assess whether A␤-Cu 2ϩ complexes can oxidize lipid membranes, we have used a colorimetric assay to monitor the formation of malondialdehyde and 4-hydroxyalkenals, which are generated by the peroxidation of unsaturated phospholipids. WT A␤ 1-42, His--Me, and His--Me at Cu 2ϩ /peptide molar ratios of 0:1 and 1:1 were incubated with 1 mM 50% PC, 50% PS LUVs for 24 h in the presence of 10 mM ascorbate, which is present to act as a reductant to initiate redox cycling of the Cu 2ϩ (30). Lipids incubated in an oxygen atmosphere undergo some degree of peroxidation in the presence of free Cu 2ϩ and ascorbate. However, in the presence of WT A␤ 1-42 at a Cu 2ϩ /peptide molar ratio of 1:1, there was a significant increase in lipid peroxidation compared with the vehicle (Fig. 9a). In contrast, lipid peroxidation was attenuated by both His--Me and His--Me, consistent with the lower neurotoxicity displayed by these peptides at the 1:1 ratio. The metal-attenuating compound clioquinol (CQ) can inhibit A␤ toxicity (29) and amyloid deposition in transgenic mice (31). In a small phase II clinical trial, CQ was efficacious in moderately affected AD subjects (32). CQ coordinates Cu 2ϩ with a ligand/metal stoichiometry of 2:1 (33), and we show here that it can inhibit the lipid peroxidation induced by the WT A␤-Cu 2ϩ complex (Fig. 9b).

DISCUSSION
We have previously postulated that one consequence of Cu 2ϩ coordination by A␤ is the formation of a His bridge between two Cu 2ϩ atoms ( Fig. 1)  (2, 4), a concept that has been challenged (6 -8). Our EPR data presented here and elsewhere (2,4) provide strong support for the existence of a 6.2 Ϯ 0.2-Å binuclear Cu 2ϩ center, consistent with the occurrence of a His bridge, at Cu 2ϩ /peptide molar ratios in the range 0.5-1.0:1 for A␤ 1-28, A␤ 1-40, and A␤ 1-42 in PBS. The absence of any evidence for Cu 2ϩ -bridged dimers at Cu 2ϩ /peptide molar ratios of 1:1 with the His--Me and His--Me establishes the importance of the imidazole side chain in dimer formation. The broad EPR signals reported here have not been reported in a number of other EPR studies of A␤-Cu 2ϩ interactions (6 -8). This discrepancy is due to differences in buffer conditions and the mode of presentation of Cu 2ϩ to the A␤ peptide; no two studies have used the same conditions.  Both Huang et al. (34) and Narayanan and Reif (35) have shown that buffer constituents, such as NaCl, have a marked effect on metal-mediated oligomerization of A␤. We obtained our spectra in PBS at pH ϳ 6.9 -7.5 and presented Cu 2ϩ as CuCl 2 (2). In contrast, Syme et al. (7) used the nonphysiological ethylmorpholine buffers at pH 7.8 and presented Cu 2ϩ as Cu(NO 3 ) 2 . When we acquired a Cu 2ϩ EPR spectrum at Cu 2ϩ /peptide molar ratios of 1:1 in this buffer, we also did not observe line broadening. We have been careful to distinguish between the broad lines due to either His-bridged binuclear Cu 2ϩ centers or aggregates and the line broadening of mononuclear spectra that can arise in different ways. Further, the pleiomorphic nature of A␤ is well documented, and it is not surprising that the buffer conditions used will have a profound influence on the observed structural state.
Whereas the mechanism of A␤ toxicity in AD remains unresolved, a general theme is beginning to emerge whereby the interaction of A␤ with lipid membranes is a necessary step in neurotoxicity (13)(14)(15)(16)(17). These interactions result in changes in membrane fluidity leading to depolarization and disorder (13), pore/channel formation leading to ion channel formation and disrupted calcium homeostasis (14,16), lipid peroxidation via membrane-associated free radical formation (15), and cholesterol oxidation (17). H 2 O 2 and ROS production has been implicated in the neurotoxicity of A␤, and the A␤-Cu 2ϩ complexes are able to produce H 2 O 2 (36 -38). The production of H 2 O 2 has been shown to occur during a short burst at the early stages of aggregation, around the time soluble oligomers are being produced (39). In this study, we demonstrated that toxicity to primary cortical neurons correlates with the formation of a His bridge between Cu 2ϩ ions, which is populated at equimolar Cu 2ϩ /peptide ratios (Figs. 3 and 4). The resulting toxic effect was rescued by the addition of catalase, implicating H 2 O 2 generation. A consequence of ROS production is lipid peroxidation, resulting in sub- FIGURE 9. Lipid peroxidation of synthetic LUVs. a, 1 mM lipid was incubated with 10 M peptide and 10 M ascorbate at Cu 2ϩ /peptide ratios of 0:1 (Ⅺ) and 1:1 (f) for 24 h at 37°C. b, 1 mM lipid was incubated with 10 M WT peptide and 10 mM ascorbate at Cu 2ϩ /peptide ratios of 1:1 for 24 h at 37°C. CQ was added at a final concentration of 2, 20, and 50 M in Me 2 SO. Me 2 SO was at a final concentration of 1% and accounts for the lower observed level of lipid peroxidation. Lipid peroxidation was determined via a colorimetric assay that measures the concentration of peroxidation by-products malondialdehyde (MDA) and 4-hydroxyalkenals (HNE). Data are shown as mean Ϯ S.E. Statistical comparisons between groups were done using Student's t test. p Ͻ 0.05 versus vehicle; each experiment was carried out in triplicate.

TABLE 1
Correlation of toxicity with formation of His-bridged structures and membrane binding sequent cell death, which precedes amyloid plaque formation (15,40). Supporting this mechanism, we have demonstrated that A␤ 1-42 at Cu 2ϩ /peptide molar ratios of 1:1, conditions in which the His bridge was formed, induced lipid peroxidation (Fig. 9). Interestingly, the observed toxicity to primary cortical neurons does not correlate with the ability of the peptides to form amyloid material or to mechanically perturb the acyl-chain region of a lipid membrane, as monitored by the anisotropy of DPH (Figs. 5 and 6). Solid-state NMR studies indicate that the A␤ peptide has an affinity for negatively charged phospholipids (41,42). Our previous 31 P static solid-state NMR data have also shown that metal-bound forms of A␤ increase perturbation of the lipid head groups, mediated through Cu 2ϩ binding, compared with the metal-free form (28). Here we demonstrate the appearance of a lipid-A␤-bound phase and a peptide-free lipid phase (Fig. 7) due to the lipid head group interaction with WT A␤ 1-42. The observation of a lipid-A␤-bound phase indicated that A␤ and A␤-Cu 2ϩ complexes were able to associate with the lipid membrane head group region and that this association would promote lipid peroxidation.
Conditions that lead to the formation of the His-bridged A␤ oligomers result in the population of a highly toxic species. It follows, therefore, that disruption of this structure would abrogate A␤ toxicity. In agreement with this hypothesis, methylation of the three His residues at either theor -nitrogen of the imidazole side chain resulted in the elimination of the His bridge at equimolar Cu 2ϩ /peptide ratios (Figs. 2 and 3). These modifications rendered the peptide nontoxic (Fig. 4); congruently, a loss in lipid peroxidation activity was observed (Fig. 9a). That CQ was able to inhibit the observed lipid peroxidation (Fig. 9b) is consistent with this process being a metal-dependent phenomenon, since previously observed data have shown that CQ inhibits A␤/Cu interactions (31). The loss in toxic activity presumably resulted from the modified A␤ peptides sequestering the redox-active Cu 2ϩ away from the lipid environment. Tickler et al. (4) have shown previously that His--Me-and His--Me-modified A␤ peptides have a lower affinity for lipid membranes, and since ROS are relatively short lived species that react with substrates in their immediate vicinity, less lipid peroxidation is expected and observed for these peptides.
Moreover, eliminating the His bridge also affected the formation of dityrosine adducts (Fig. 8). Tyr 10 is critical in the redox cycling of Cu 2ϩassociated ROS production by A␤-Cu 2ϩ (29). A by-product of the redox reactions is that two tyrosine radicals can react to form dityrosine adducts. The formation of dityrosine adducts by both His--Me and His--Me A␤ was severely inhibited (Fig. 8), correlating with the decrease in lipid peroxidation (Fig. 9) and subsequent toxicity (Fig. 4). This is consistent with a role for either the tyrosine radical or dityrosine in metal-mediated toxicity.
A survey of the literature shows that variations to the A␤ peptide sequence that ablate lipid binding cause a decrease in A␤ neurotoxicity (Table 1). A␤ 1-28, which is capable of binding to lipids, although weakly, has been reported to show a small degree of toxicity (43). Methylation of the His residues at either theor -nitrogen of the imidazole side chain of A␤ 1-40/42 results in a loss in lipid binding and toxicity as well as a loss in the ability to form the His bridge (4). By contrast, A␤ 1-42 containing a M35V mutation shows an increase in toxicity over the WT peptide and displays a higher affinity for lipid membranes (21). In addition, A␤ 1-42 with an oxidized Met 35 is unable to penetrate the lipid membrane but is able to bind to the surface and induce toxicity (26). These results, together with the data presented here, lead to a model whereby binding of A␤ to the surface of a lipid membrane is required to cause Cu 2ϩ -induced toxicity. Taken with the results presented above, the toxic A␤-Cu 2ϩ species requires the formation of the His bridge and imparts its toxic effect via the association with lipids and the generation of ROS, resulting in lipid peroxidation and subsequent cell death. The lipid peroxidation can be inhibited by the use of the metal-attenuating compound CQ, which can disaggregate metal-induced aggregates and retard fibril growth (44) and inhibit amyloid deposition in transgenic mice (31) and was efficacious in a phase II clinical trial on a severely affected AD group (32). Therefore, strategies aimed at inhibiting the formation of His-bridged A␤-Cu 2ϩ complexes, thereby preventing metal-induced A␤ oligomer formation and consequent production of ROS, would limit the oxidative stress associated with AD and be of therapeutic benefit.