Enhanced Toxicity and Cellular Binding of a Modified Amyloid β Peptide with a Methionine to Valine Substitution*

The amyloid β peptide (Aβ) is toxic to neuronal cells, and it is probable that this toxicity is responsible for the progressive cognitive decline associated with Alzheimer's disease. However, the nature of the toxic Aβ species and its precise mechanism of action remain to be determined. It has been reported that the methionine residue at position 35 has a pivotal role to play in the toxicity of Aβ. We examined the effect of mutating the methionine to valine in Aβ42 (AβM35V). The neurotoxic activity of AβM35V on primary mouse neuronal cortical cells was enhanced, and this diminished cell viability occurred at an accelerated rate compared with Aβ42. AβM35V binds Cu2+ and produces similar amounts of H2O2 as Aβ42 in vitro, and the neurotoxic activity was attenuated by the H2O2 scavenger catalase. The increased toxicity of AβM35V was associated with increased binding of this mutated peptide to cortical cells. The M35V mutation altered the interaction between Aβ and copper in a lipid environment as shown by EPR analysis, which indicated that the valine substitution made the peptide less rigid in the bilayer region with a resulting higher affinity for the bilayer. Circular dichroism spectroscopy showed that both Aβ42 and AβM35V displayed a mixture of α-helical and β-sheet conformations. These findings provide further evidence that the toxicity of Aβ is regulated by binding to neuronal cells.

Genetic evidence derived from early onset cases of Alzheimer's disease (AD) 1 indicated that the metabolism of the amyloid ␤ peptide (A␤) is clearly linked to the pathogenesis of this disease (1). There is a clear correlation found between the levels of soluble A␤, synaptic damage, and cognitive impairment in AD patients (2)(3)(4). It is well established that synthetic A␤ is toxic to neuronal cells in culture (5)(6)(7)(8). Moreover, naturally secreted oligomers of A␤ can potently inhibit neuronal long term potentiation, a measure of synaptic plasticity in vivo (9,10). Although the precise mechanism of A␤-induced neuronal toxicity remains unclear, the interaction of A␤ with membranes and/or membrane proteins appears to play an important role in A␤-mediated neurotoxicity. Because A␤ contains part of the transmembrane domain of the amyloid precursor protein (APP), it is not surprising that A␤ interacts with cell membranes and lipoproteins. Numerous reports have described the effects of A␤ on membranes and lipid systems and their possible roles in A␤ neurotoxicity. These include changes in membrane fluidity leading to membrane depolarization and disorder (11,12), pore/channel formation that could affect calcium homeostasis (13)(14)(15), and lipid peroxidation via membrane-associated free radical formation (16 -19). A␤ can decrease the fluidity of both artificial unilamellar liposomes and mouse brain membranes (12) and human frontal cortex membranes (11). A␤ has an amplifying effect on cellular calcium signaling. At a low concentration, A␤ acts by stimulating endogenous calcium conductance pathways (20), whereas at higher concentrations A␤ disrupts the membrane to allow calcium flux in a channel-like fashion (15,21). Our previous studies indicated that the interaction between large unilamellar vesicles (LUVs) and A␤ in the presence of copper and zinc resulted in ␣-helical oligomeric channel-like structures being formed (22,23).
The Met-35 residue in the A␤42 sequence has been reported to play a key role in A␤ toxicity because A␤s with Met-35 substituted by either norleucine or cysteine are non-toxic and unable to induce oxidative stress responses in cells (24,25). In contrast, we reported that Met(O)A␤ has significant toxicity to neuronal cells in primary culture (26). Many of the proposed mechanisms of toxicity listed above are dependent on A␤ aggregation, whether as dimers or trimers (9), larger aggregates such as protofibrils (27), or the megadalton fully formed fibrils (28,29). The oxidation of methionine affects the aggregation properties of the peptide leading to an altered oligomer size distribution (30) and inhibition of fibril formation (31). In addition, Met(O)A␤ induces changes in peptide structure resulting in modified aggregation properties and diminished interactions with lipid membranes by inhibiting membrane penetration (26). Nonetheless, these modifications did not significantly hinder the toxic effects of the peptide suggesting that A␤ toxicity via these mechanisms is of secondary importance. To further probe the role of Met-35 in A␤ toxicity we investigated a novel mutant A␤42 where the Met-35 was substituted with a Val residue (A␤M35V) and studied its neurotoxic, structural, and biochemical properties compared with its wild type A␤42 equivalent.

EXPERIMENTAL PROCEDURES
Peptide Preparation-Synthetic A␤42 was purchased from W. M. Keck Laboratory (Yale University, New Haven, CT). Recombinant A␤42 and A␤M35V were purchased from Recombinant Peptide Technologies (Athens, GA). Peptide working stock solutions were prepared by making a 250 M concentration by weight, dissolving the peptide in 20 mM NaOH, and then diluting out in water and 10ϫ PBS in a v/v ratio of 2:7:1. All A␤ solutions were sonicated in a water bath for 5 min and centrifuged for 5 min at high speed, and supernatant was used for the toxicity experiments. The absorbance value at 214 nm was measured, and an accurate concentration of the peptides was determined using a molar extinction coefficient value of 75,887 liters/mol/cm and adjusting the final concentration to 200 M. This method of A␤ preparation is associated with increased peptide solubility and reproducibility between experiments. All chemicals were purchased from Sigma unless otherwise stated.
Primary Neuronal Cultures-Cortical neuronal cultures were prepared as described previously under sterile conditions (26). Briefly, embryonic day 14 BL6Jϫ129sv mouse cortices were removed, dissected free of meninges, and dissociated in 0.025% (w/v) trypsin (Sigma) in Krebs buffer. The dissociated cells were triturated using a filterplugged 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(L-lysine)-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 B27 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 contamination as determined by immunostaining of culture preparations using specific marker antibodies (data not shown).
Cell Viability Assays-The neuronal cells were allowed to mature for 6 days in culture before commencing treatment using freshly prepared neurobasal medium plus B27 supplements minus antioxidants. For the treatment of neuronal cultures, freshly prepared soluble A␤ stock solutions were diluted to the final concentration in neurobasal medium. The mixtures were then added to neuronal cells for up to 4 days. Cell survival was monitored by phase contrast microscopy, and cell viability was quantitated using the MTS assay as described previously (26). Briefly, the medium was replaced with fresh neurobasal medium supplemented with B27 lacking antioxidants, and 10% v/v MTS (Promega, Madison, WI) was added to each well and incubated for 3-4 h at 37°C in a 5% CO 2 incubator. Plates were gently shaken, and a 100-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.
Electron Microscopy-200 M stock solutions of A␤42 and A␤M35V were prepared as described above and diluted to 5 M in 1ϫ PBS to replicate the concentration of peptide that is routinely used for toxicity experiments. An initial aliquot (day 0) and further aliquots were taken at days 1, 2, and 3 following incubation at 37°C. Aliquots were then serially diluted at 1:100 four times in water. A small drop was then spotted in the center of a carbon-coated grid (ProSciTech, Queensland, Australia) and allowed to dry by placing the grid upside down on a 3% agar dish for up to 1 h at room temperature. The peptides were negatively stained with 0.5% uranyl acetate for 3 min and then washed several times with water. Images of the samples were taken using a Siemens ELMI-SKOP 102 electron microscope. and horseradish peroxidase (0.1 M) incubated at 37°C. There was a 4:3 molar excess of peptide to copper to ensure that there would be no free copper ions in the solution. The reaction was carried out at 200 nM A␤42 to reflect the soluble concentration of A␤ in the brain. Studies were performed on the day of reagent preparation. The fluorescence signal specific for H 2 O 2 was decreased in parallel samples co-incubated with catalase (40 units/ml; 100 nM). Fluorescence readings were recorded on a PerkinElmer Life Sciences LS55 plate reader (excitation, 485 nm; emission, 530 nm) against a standard curve of reagent grade H 2 O 2 in PBS. All experiments were performed in triplicate.
Preparation of LUVs-Large unilaminar vesicles were prepared by initially dissolving equal quantities of palmitoyl phosphatidyl serine (Avanti Polar Lipids Inc.) and synthetic palmitoyl phosphatidyl choline in chloroform. The chloroform was evaporated off, and the lipids were resuspended in 10 mM phosphate buffer, pH 6.8, to 50 mg/ml. Following the addition of some glass beads, the mixture was shaken for 1 h at 37°C. The lipid mixture was then decanted from the glass beads, subjected to freeze-thaw cycles using liquid nitrogen and a 37°C water bath, and finally passed through an extruder apparatus, resulting in a uniform solution of LUV having a mean diameter between 120 and 140 nm. LUVs were stored at 4°C and used within 48 h of preparation.
EPR Spectroscopy-The acidic phospholipid spin label 16NPS was synthesized according to Hubbell and McConnell (32). The water-soluble spin label Tempo choline chloride (TCC) was obtained from Molecular Probes. All spin probes were checked for purity to ensure that their number of spins/mol were Ͼ90% of theory (33). LUVs were prepared by freezing and thawing of the lipid suspension six times and then extruding the suspension through a 22-m Nucleopore filter (Pleasanton, NJ) (34). Lyophilized peptides were weighed and dissolved at the desired concentration in a suspension of LUV in buffer, and the mixture was vortexed under N 2 for 10 min at 305 K. Stock solutions of 100 mM concentration of analytical reagent grade CuCl 2 were prepared, and the desired amount was added to the LUV after addition of the peptide. The metal chelator EGTA (50 mM) was added to all control solutions to counter the possible effects of any trace metals present. For copper EPR measurements, 99.99% pure 65 Cu (Cambridge Isotopes) was used. Metal concentrations were measured by inductively coupled plasmamass spectroscopy (Model 700, Varian), and peptide concentrations were determined by quantitative amino acid analysis.
Continuous wave X-band EPR spectra were obtained using a Bruker ECS106 spectrometer equipped with a temperature controller and flowthrough cryostat. Cu 2ϩ spectra were collected at 110 K from samples contained in 4-mm inner diameter "Suprasil" quartz EPR tubes (Wilmad). Labeled lipid samples (25 l) were contained in 0.8-mm inner diameter quartz capillaries (Wilmad) and handled as described by Gordon et al. (35) to ensure reproducibility. Other procedures, including adding to the spin-labeled LUV a small amount of the water-soluble, non-membrane-penetrant spin probe Tempo choline chloride to ensure x axis reproducibility, have been described previously (22,36,37).
The analysis of EPR spectra of peptide/lipid mixtures was carried out using the spectral subtraction and addition methods described by Marsh (38). As a check on the validity of these procedures as applied to our system, where a motion-restricted lipid component was observed, the lipid spin label spectra were simulated using the modified Bloch equations as described by Davoust and Devaux (39). Their Model I was used in the simulations, where the unique director orientation in the fluid component, corresponding to the nitroxide axes being aligned along the membrane normal in the motion-restricted lipid component, was assumed to be preserved on exchange.
CD Spectroscopy-The LUV mixture was subjected to five freezethaw cycles and then extruded 11 times through a 0.1-m pore filter (Millipore, Australia) using an Avanti "mini-extruder" apparatus. A␤42 and A␤M35V were dissolved in LUVs (50 mg/ml, giving a molar ratio of 60:1 LUV/peptide) in the presence of Cu 2ϩ (molar ratio of 1:1 Cu 2ϩ / peptide). The samples were then diluted in phosphate buffer, pH 6.8, to a final peptide concentration of 10 M. The CD spectra were obtained using a Jasco 810 spectropolarimeter at 37°C. Far-UV CD spectra were obtained from 190 to 260 nm with a 1-cm path length. The base line acquired in the absence of peptide was subtracted, and the resulting spectra were smoothed using a Fourier transform.
Peptide Binding and Cell Uptake-5 M A␤42 and A␤M35V were added to 6-day-old cortical neuronal cells in culture for 1 h. The conditioned medium was then collected and centrifuged, and the supernatant was transferred to a new tube. The cortical neurons in the wells were briefly washed in cold PBS or ice-cold 0.1 M sodium carbonate (carbonate wash); the cells were scraped up from the wells, and proteins were extracted from the cells in 100 l of extraction buffer. The conditioned medium and cell extracts were analyzed by Western blot to confirm the presence of peptide uptake by cells. Aliquots of each sample were diluted in sample buffer (100 mM Tris, pH 6.8, 8% SDS, 50% glycerol, 25% ␤-mercaptoethanol, 0.1% bromphenol blue) and heated to 90°C for 5 min, and samples were separated by SDS-PAGE using 16.5% T, 3% C Tris-Tricine gels and then transferred to 0.2-m pore nitrocellulose membrane (Bio-Rad). Immunoblots were boiled for 5 min in PBS and blocked for 1 h in TBST (10 mM Tris-HCl pH 8.0, 150 mM NaCl, 0.1% Tween 20) containing 5% skim milk. The immunoblots were then probed for 2 h at room temperature with an anti-A␤ monoclonal antibody (WO2, targeting A␤ residues 5-8) diluted 1:50 from culture medium in PBS, 0.1% Tween 20, washed three times in 10-min intervals in TBST, probed with a peroxidase sheep anti-mouse immunoglobulin diluted 1:5000 in TBST for 1 h at room temperature, and then washed several times again. Blots were developed with ECL reagent (Amersham Biosciences) for 1 min per the manufacturer's instruction, and the chemiluminescent signals from immunoblots were quantitated using a charge-coupled device camera imaging system and GeneTool analysis software (GeneGnome, Syngene, Cambridge, UK).

RESULTS
A␤M35V Is More Neurotoxic than Wild Type A␤42-The neurotoxic activity of A␤M35V was tested to determine whether it, like the wild type A␤42, is toxic to neuronal cells. A␤M35V had a significantly higher degree of toxicity on neuronal cortical cultures than A␤42 after 96 h at all of the concentrations tested (Fig. 1A). Because A␤M35V is potently toxic to cells at 5 M after 96 h of treatment, we examined the cell viability of the cortical neurons at shorter treatment times of 24, 48, and 72 h (Fig. 1B). The cell viability of the cortical neurons treated with wild type A␤42 was significantly decreased after 72 and 48 h but not after 24 h. In comparison, A␤M35V was 2-2.5-fold more toxic to the neuronal cells compared with A␤42 after 72 and 48 h. A 24-h treatment with A␤M35V caused a significant decrease in cell viability (Fig.  1B). To determine whether the toxicity of A␤M35V was mediated via H 2 O 2 , we used a selective H 2 O 2 -scavenging enzyme catalase as reported previously (26). As expected, the catalase completely rescued the toxicity induced by A␤42. However, it only caused total rescue of A␤M35V at the lowest concentration examined (1.25 M), and partial rescue was observed at the higher doses tested (Fig. 1A). This indicates that A␤M35V utilized H 2 O 2 -like wild type A␤42, but at the higher concentrations of A␤M35V tested, the high levels of cell toxicity far exceeded the capacity of catalase to exert its protective properties.
The Aggregation Profiles of A␤42 and A␤M35V Were Similar in Electron Micrographs-A␤ will form oligomers, protofibrils, or fibrils when dissolved and allowed to aggregate (40). A␤42 and A␤M35V were prepared under identical conditions, and the sizes of the oligomeric species formed after 0 and 3 days of incubation were examined by electron microscopy (Fig. 2). The results show that the day 0 samples for both A␤42 and A␤M35V had aggregated with diameters generally less than 10 nm (Fig. 2, Day 0). There were no quantifiable differences in the observed sizes of the oligomers for A␤M35V and A␤42 at day 0 or after day 3 (Fig. 2).
Catalytic Production of H 2 O 2 Is Similar between A␤42 and A␤M35V-The ability of catalase to rescue A␤ toxicity implicates the generation of H 2 O 2 in the toxic pathway. We have reported previously that A␤s in the presence of the redox-active metal copper and iron are able to catalytically generate H 2 O 2 . The catalytic activity of both A␤42 and A␤M35V to produce H 2 O 2 was determined using a DCF assay (41). The V max and K m of H 2 O 2 production were determined by successively increasing the amount of the substrate ascorbic acid used in the reaction (Fig. 3A). The Lineweaver-Burk plots yielded straight lines for both A␤42 and A␤M35V (Fig. 3B). The calculated V max (63 Ϯ 4 and 45 Ϯ 1 nM/min) and K m (3.6 Ϯ 0.4 and 1.4 Ϯ 0.1 M) values were similar for A␤42 and A␤M35V, respectively, indicating that the presence of Met-35 is not critical for the production of H 2 O 2 in the presence of a substrate such as ascorbate.

EPR Spectroscopy Reveals That A␤M35V Has a Slower Copper-reducing Activity and Is Less Rigid in a Lipid Bilayer-The
Cu 2ϩ EPR spectra of the two peptide complexes at pH 6.9 were determined and plotted in Fig. 4A. The spectra are characteristic of type 2 square planar coordination of the Cu 2ϩ (42) and are identical, indicating that substituting the Met to Val does not alter the copper coordination site of A␤42. Both samples were allowed to stand at 295 K for a series of time intervals (0 -24 h); then the temperature was reduced to 110 K, and their spectra were recorded. The amount of Cu 2ϩ visible in the spectra was determined by double integration and plotted in Fig.  4B. This shows that the reduction of Cu 2ϩ to the EPR-silent Cu ϩ is much slower with A␤M35V, confirming earlier observations (22) that in the absence of a substrate the Met-35 residue in A␤42 is able to initiate redox chemical reactions of copper/ iron bound to the peptide in solution.
To determine whether the interaction between A␤ and copper in a lipid environment is altered for A␤M35V, we obtained a series of EPR spectra of 16NPS in LUVs containing 0.05 M PBS pH 6.9, and after adding 0.03 mol of peptide/lipid of both A␤42 and A␤M35V coordinated with 0.3 mol equivalents of Cu 2ϩ per mole of peptide. All spectra were recorded at 303 K. Both peptides in the presence of Cu 2ϩ give the motion-restricted lipid component characteristic of systems with a rigid peptide segment inserted into the lipid bilayer (Fig. 4C). However, there were significant differences in the line width. A␤M35V showed a narrower line width for the partially immobilized component of the spectrum suggesting that the valine substitution has made this mutant peptide less rigid in the bilayer region when probed by 16NPS. These spectra could be simulated as described by Davoust and Devaux (39), discounting long range perturbation effects on the bulk lipid (43) by the peptide at the relative high peptide/lipid ratios used in our experiments. The on-and off-rate constants were estimated from the simulation experiments, with the values for the former being in the range of 6.0 -5.5 ϫ 10 6 (s Ϫ1 ) and for the latter being 5.5-4.5 ϫ 10 6 (s Ϫ1 ). These values are similar to those published for other lipid/protein systems (44) and show that A␤M35V has a higher affinity for the lipid bilayer as compared with A␤42.
CD Spectroscopy Reveals That A␤M35V Has a Higher Proportion of ␤-Sheet Structure and Random Coil than A␤42-A CD spectroscopy analysis was obtained to examine whether the secondary structure of A␤M35V in a lipid environment is altered (Fig. 5). In preparing the samples the mutant peptide was observed to dissolve more rapidly in the LUV solution compared with A␤42, suggesting that A␤M35V has a higher affinity for the lipids. Although the presence of LUV in solution increased the signal to noise ratio in the lower wavelengths and prevented accurate measurements of helical content, the spectra could be qualitatively compared with each other. The CD spectra of the wild type A␤42 in the presence of LUV and Cu 2ϩ showed a double minima at 222 nm and just above 210 nm, plus a positive reading at 200 nm. Taken together, these features indicate that the secondary structure is predominantly ␣-helical, with some ␤-sheet present. The CD spectra for A␤M35V showed a minimum at around 213 nm, which on its own would indicate a predominance of ␤-sheet structure; however, the inflections in the region between 220 and 225 nm indicate that there is possibly a small amount of ␣-helix present. At wavelengths below 205 nm, the light scatter because of the presence of the LUV makes it impossible to identify the positive peaks in Cu 2ϩ . Instrument settings were as follows: width of spectrum, 1100 gauss; midfield, 3750 gauss; microwave frequency, 9.6 GHz; microwave power, 20 mW; modulation frequency, 50 KHz; and modulation amplitude, 1 gauss. B, comparison of the rate of reduction of Cu 2ϩ to the EPR-silent Cu ϩ by A␤42 (diamonds) and A␤M35V (squares). C, 16NPS spin probe in LUV containing 0.03 M peptide/lipid of either A␤42 and A␤M35V or no peptide (Control). A␤42-control and A␤M35V-control are difference spectra between control, A␤42, and A␤M35V, respectively. Instrument settings were as follows: width of spectrum, 100 gauss; midfield, 3425 gauss; microwave frequency, 9.6 GHz; microwave power, 20 mW; modulation frequency, 100 KHz; and modulation amplitude, 1 gauss. this region of the spectra. Thus in the presence of LUV and Cu 2ϩ the wild type A␤42 contained a higher proportion of ␣-helical structure.
A␤M35V Has a Higher Affinity for Binding to Cortical Neuronal Cells in Culture-The increased toxicity of A␤M35V could be because of increased cell membrane binding of the peptide as suggested by the higher affinity of A␤M35V for the synthetic lipid bilayers. To test this observation, cortical cells were treated with the peptides for a short duration (1 h) so as to minimize the incidence of cell death. The amount of peptide bound to cortical cell extracts was quantitated by Western blotting, and the results were expressed as a percentage of the initial peptide concentration. Densitometry of the bands from Western blots shows that twice as much A␤M35V compared with wild type A␤ bound to the cortical cells after a 1-h cell exposure (p Ͻ 0.05, Fig. 6). Washing the cells with a mild carbonate buffer to remove any loosely bound peptide from the plasma membrane, there was an even greater differential with approximately 5ϫ more A␤M35V present in cell extracts compared with wild type A␤42 (p Ͻ 0.05).

DISCUSSION
The Met-35 residue in A␤42 has been suggested to play a key role in the toxicity of A␤. The substitution of Met-35 in the A␤42 peptide sequence to either norleucine or cysteine has been reported to result in the peptides being non-toxic and unable to induce oxidative stress responses in cells (24). The same group reported that the modified A␤, Met(O)A␤, was also non-toxic in their cell culture models (25). In contrast, when we treated our cortical neuronal cell cultures for a longer period of time (96 h compared with 24 h (24, 25)), we showed that Met(O)A␤ caused significant decreases in cell viability, and this was rescued by catalase and attenuated by the metal chelator clioquinol (26). To clarify the role of Met-35 in A␤ toxicity, we substituted the Met-35 residue for a valine residue and compared its behavior with wild type A␤42. Strikingly, A␤M35V was more toxic to neuronal cells. Even at low micromolar concentrations (1.25 M) A␤M35V treatment resulted in significantly decreased cell viability. Interestingly, catalase was only effective at rescuing the diminished cell viability of A␤M35V at the lowest dose used (1.25 M) and showed only partial rescue when higher concentrations of this peptide were examined. Because the higher concentrations of A␤M35V were associated with greater than 40% cell death (Fig. 1A), the effectiveness of the amount of catalase used in these experiments (2000 units/ml) to totally rescue cell toxicity at these very high levels of cell death is diminished. The neurotoxic concentration of soluble A␤M35V in our neuronal cell culture models approaches those used in other reports with A␤42 for ADDLS (45) and spheroids (46) and in modified cell culture conditions (5). Not only was A␤M35V neurotoxicity greater on a molar basis compared with A␤42, but also cell viability decreased faster as seen with a 24-h treatment, which is comparable with other cell culture models using Locke's buffer instead of neurobasal medium (5), transformed cell lines (47)(48)(49), and brain slice cultures (10,50). Our results definitively show that the A␤M35V mutant peptide has enhanced toxicity to neuronal cells in primary culture, an effect that was found to occur at a significantly faster rate compared with the wild type peptide.
Having established that A␤M35V is more toxic than wild type A␤42, we set out to identify the basis for this increased toxicity. Among the proposed mechanisms of toxicity, there is increasing evidence that the biophysical state of the peptide will influence its toxic activity. A␤ can aggregate into different oligomeric species whether as dimers or trimers (9), larger aggregates such as protofibrils (27), or the megadalton fully formed fibrils (28,29). Electron microscopy images of the aggregation profile of the peptides in culture conditions replicating those used for our toxicity experiments showed A␤42 and A␤M35V over a 3-day sampling period to be similar (Fig. 2). The sizes of the spherical aggregates are between 10 and 30 nm in diameter (data not shown) with no detectable fibrils present in any of the electron microscopy images. Bitan et al. (51) have reported previously that spherical aggregates of this size are made up of A␤ oligomers ranging from a monomer to a dodecamer. This similarity in the aggregation of the two peptides suggests that the size of the peptide species does not contribute to the increased toxicity of the A␤M35V mutant peptide. H 2 O 2 production has been implicated previously in the neurotoxicity of A␤ (5,47,52), and the A␤-copper complex in the presence of a reducing substrate, such as ascorbate or dopamine, is able to produce H 2 O 2 (5,52). The generation of reactive oxygen species is facilitated by the presence of a redox-active metal ion such as Cu 2ϩ or Fe 3ϩ , and in vitro A␤ interacts with either of these metals to generate reactive oxygen species, an event that gives rise to Fenton chemistry (52)(53)(54). When A␤ binds Cu 2ϩ or Fe 3ϩ , extensive redox chemistry occurs reducing the oxidation state of both metals and producing H 2 O 2 from O 2 in a catalytic manner (52). The ability of A␤ to coordinate and reduce Cu 2ϩ in vitro results in the peptide being oxidized (26) with the most likely site of oxidation being the sulfur atom of Met-35 (55). Surprisingly A␤M35V could reduce Cu 2ϩ , albeit much slower compared with Met(O)A␤ and wild type peptides (Fig. 4A) (26). This small reduction of Cu 2ϩ to Cu ϩ suggests that there may be other unknown contributing factors to this reaction step such as the attenuation of conversion to Cu ϩ and clearly illustrates the critical role of the Met residue in reducing Cu 2ϩ . Although this decreased copper-reducing reaction would suggest poor H 2 O 2 production, A␤M35V and A␤42 produced similar amounts of H 2 O 2 (Fig. 3). In addition, we had shown previously that Met(O)A␤42 and A␤42 produced similar amounts of H 2 O 2 (26). Taken together, the observations of a decrease in copper-reducing activity are not a strong predictor of H 2 O 2 production. What may be important is the high affinity of A␤ for copper (56) and the rapid and complete reduction of Cu 2ϩ to Cu ϩ by A␤ (8). Because metal binding occurs near the N terminus of A␤ (57), we have identified recently tyrosine 10 as a pivotal residue to drive the catalytic production of H 2 O 2 by A␤-copper whereby mutation of the tyrosine residue to alanine resulted in an inhibition of H 2 O 2 production (41). Therefore these results suggest that any modification of Met-35 has a negligible effect on the production of H 2 O 2 .
Another plausible mechanism for 〈␤ neurotoxicity is based on 〈␤ interacting with membranes and/or membrane proteins because it contains part of the putative transmembrane domain of the amyloid precursor protein (APP). The effects of 〈␤ on membranes and lipid systems and their possible roles in 〈␤ neurotoxicity include changes in membrane fluidity leading to membrane depolarization and disorder (11,12), pore/channel formation that could affect calcium homeostasis (13)(14)(15), and lipid peroxidation via membrane-associated free radical formation (16 -19). 〈␤ can also decrease the fluidity of both artificial unilamellar liposomes and mouse brain membranes (12) and human frontal cortex membranes (11). The interaction between large unilamellar vesicles and 〈␤ in the presence of copper and zinc resulted in ␣-helical oligomeric channel-like structures being formed (22,23). The M35V substitution alters the secondary structure of this peptide in the presence of a lipid environment (Fig. 5) resulting in a higher proportion of ␤-sheet structure and less ␣-helix. Wild type A␤42 contained a higher proportion of ␣-helical structure, whereas Met(O)A␤ was shown to have random coil structure (26). In the presence of Cu 2ϩ , both the wild type peptide and A␤M35V resulted in the motion-restricted lipid component characteristic of systems with a rigid peptide segment inserted into the lipid bilayer. However, the narrower line width of the partially immobilized component of the A␤M35V spectrum suggests that the valine substitution has made this peptide less rigid in the bilayer region. The EPR spectra highlights that A␤M35V has a higher affinity for the lipid bilayer as compared with the wild type peptide. This result is in agreement with the observation of A␤M35V being far more soluble in a LUV milieu compared with A␤42 (data not shown). In contrast, the addition of metal ions to Met(O)A␤ did not result in such a structural transition, and the peptide was unable to penetrate the lipid membrane (26). The less rigid state of the mutant A␤M35V in the lipid bilayer may facilitate increased oligomerization of A␤M35V as well as increased and/or faster binding to lipid membranes causing A␤M35V to have increased toxic properties compared with A␤42.
The plasma membrane provides a protective barrier shielding the cytosolic contents, as well as providing a conduit for communication with the extracellular environment. The regulation of the plasma membrane composition of protein and lipid contents is essential for maintenance of neuronal cell viability. There is increasing evidence linking altered cholesterol levels in the brain with AD (for review, see Ref. 59) and the viability of A␤-treated neurons in culture (58). The EPR and structural data suggest that A␤M35V has an increased affinity for lipids. This was supported by the finding that there was a 2-fold higher amount of A␤M35V bound to the neuronal cells in culture compared with A␤42 (Fig. 6). These results add further evidence that the mutant peptide having a higher affinity of the lipid membranes is responsible for the decreased cell viability. The implication is that toxicity is not solely because of the production of H 2 O 2 in the medium but that some sitespecific effects are likely to be in play that are exaggerated in A␤M35V compared with A␤42.
In summary our current and previous results (22,23,26) suggest that the wild type A␤, its oxidized form, Met(O)A␤, and now the mutant peptide A␤M35V induce cell death via similar pathways that are metal-dependent and involve H 2 O 2 generation in the absence of a methionine residue and argue against fibril formation as the toxic species responsible for cell death.
Our data support the model that soluble A␤ species are the most likely candidates responsible for the synaptic damage and cognitive impairment present in AD. Finally, this mutant peptide A␤M35V provides us with a new research tool with the advantage that it acts more rapidly and with increased potency at killing cells compared with the wild type A␤42.