New Insights on How Metals Disrupt Amyloid β-Aggregation and Their Effects on Amyloid-β Cytotoxicity

Amyloid-β protein (Aβ) aggregates in the brain to form senile plaques. By using thioflavin T, a dye that specifically binds to fibrillar structures, we found that metals such as Zn(II) and Cu(II) normally inhibit amyloid β-aggregation. Another method for detecting Aβ, which does not distinguish the types of aggregates, showed that these metals induce a non-β-sheeted aggregation, as reported previously. Secondary structural analysis and microscopic studies revealed that metals induced Aβ to make non-fibrillar aggregates by disrupting β-sheet formation. These non-fibrillar Aβ aggregates displayed much weaker Congo Red birefringence, and in separate cell culture experiments, were less toxic than self β-aggregates, as demonstrated by 3-[4,5-dimethylthiazol-2-yl]-2,5-diphenyltetrazolium bromide assay. The toxicity of soluble Aβ was enhanced in the presence of Cu(II), which suggests the previously hypothesized role of Aβ in generating oxidative stress. Finally, under an acidic condition, similar to that in the inflammation associated with senile plaques, β-aggregation was robustly facilitated at one specific concentration of Zn(II) in the presence of heparin. However, because a higher concentration of Zn(II) virtually abolished this abnormal phenomenon, and at normal pH any concentrations strongly inhibit β-aggregation and its associated cytotoxicity, including its anti-oxidative nature we suggest that Zn(II) has an overall protective effect against β-amyloid toxicity.

reported to cause neuronal cell death in primary rat hippocampal cultures (3), and soluble monomeric species of A␤ are relatively nontoxic as compared with fibrillar A␤ (4). Thus, these in vitro studies suggest that the degree of ␤-aggregation is particularly important for neurotoxicity to occur (5)(6)(7)(8). However, many controversial results from in vivo studies have been reported concerning the pathological role of plaque formation in AD. Irizarry et al. (9) reported that transgenic (TG) mice expressing human A␤ failed to exhibit neuronal loss despite depositing substantial amounts of A␤. On the other hand, TG mice that express Swedish mutant amyloid precursor protein (APP) formed plaques that were detected by both an anti-APP antibody and a ␤-sheet specific dye (10). Moreover, these APP TG mice also displayed memory deficits. Taken together, these results indicate that, although the plaque assembly process may require further investigation, amyloid ␤-aggregation certainly is an essential event in the pathogenesis of AD.
Based on these lines of evidence, the search for a compound that interrupts ␤-aggregation and thus protects against its neurotoxicity is of great interest. One such promising family of compounds is metals. Initially, however, metals attracted attention in a negative way. It was reported that zinc induced the aggregation of A␤ (11)(12)(13), while copper dramatically induced A␤ aggregation under acidic conditions (14). Additionally, aluminum and iron were reported to promote the aggregation of A␤ at physiological concentrations (15). As a result, most of these metal ions, at one time or another, have been proposed as risk factors for the development of AD (16 -18), rather than potentially beneficial compounds to encounter AD. Pilot studies in our laboratory, however, provided new evidence that certain metals may in fact inhibit the formation of ␤-aggregates. This finding prompted us to conduct a more thorough examination of this metal-induced inhibition of ␤-aggregation.
In the present study, we sought to identify and characterize the different types of A␤ aggregates produced in the presence of different metal ions. We also assessed the cytotoxicity of these aggregates. Although previous studies describing the effects of metals on A␤-associated aggregation measured overall aggregation (11)(12)(13)(14), here, we used thioflavin T (ThT), a fluorescent dye that binds with fibrillar structures (19,20), to specifically measure the degree of ␤-aggregation (21). Along with results using this specific dye, as well as results showing structural differences of aggregates, revealed by circular dichroism (CD) spectroscopy, Congo Red staining (22)(23)(24)(25), and electron microscopy, we report that certain metals normally prevent A␤ from participating in fibrillogenesis by forming metal-induced aggregates, whose structural characteristics are distinct from fibrous self ␤-aggregates. Moreover, this inhibitory effect of metals on ␤-aggregation via conformational enforcement was directly correlated with their protective effects against cell toxicity, measured by the reducing rate of 3-[4,5-dimethylthiazol-2-yl]-2,5-diphenyltetrazolium bromide (MTT) (26 -29). Because we also recognize that Cu(II) is a redox-active metal, we performed experiments that shed new light on two kinds of effects Cu(II) has on A␤ cytotoxicity in its soluble state (30 -33) versus its aggregated state. Finally, we report that under an experimental condition that models the acidotic inflammatory response with the involvement of heparan sulfate proteoglycan (HSPG) (36 -38) present around senile plaques (14,34,35), ␤-aggregation is strongly induced in the presence of Zn(II) at a specific concentration. We propose a hypothetical mechanism of inflammation-associated amyloidogenesis to account for these data. Nonetheless, since a higher concentration of Zn(II) diminished this phenomenon, we suggest that Zn(II) has therapeutic possibilities.
A␤ Sample Preparation-A stock solution of A␤ was prepared by dissolving powdered A␤ peptide in 100% HFIP to a final concentration of 1 mg/ml. After shaking for 2 h at 4°C, the A␤ stock solution was aliquoted into siliconized tubes and stored at Ϫ20°C. Just prior to each experiment, aliquots were spin-vacuumed using an Integrated Speed-Vac system (Savant) and then dissolved either in a Tris-buffered solution or in serum-free 5% Neurobasal medium (Life Technologies, Inc.).
Fluorescence Spectroscopy (ThT Assay)-The degree of ␤-aggregation was determined using the fluorescent dye, ThT, which specifically binds to fibrous structures (19,20). First, A␤ stock solution (see above) was diluted with either 20 mM Tris-HCl alone or with 20 mM Tris-HCl and different concentrations of different metals. ThT was added to each test sample to a final concentration of 10 M. Each sample was prepared in Black Cliniplates (Labsystems) and shaken for 10 s prior to each measurement. Measurements were carried out every 20 min.
The relative degree of ␤-aggregation was assessed in terms of fluorescence intensity, which was measured at 37°C using a Fluoroskan Ascent FL spectrophotometer (Labsystems). Measurements were performed at an excitation wavelength of 444 nm and an emission of 485 nm, wavelengths that result in the optimum detection of bound ThT. To account for background fluorescence, the fluorescence intensity measured from each control solution without A␤ was subtracted from that of each solution containing A␤. The fluorescence spectra of A␤40 from different commercial sources and from different lots were in good agreement.
UV Spectroscopy (OD 214 Assay)-The effect of different metals on total A␤ aggregation was assessed using an optical density (OD) assay. We adopted a modified procedure described previously (14). A␤ stock solutions were first diluted with 20 mM Tris-HCl (pH 7.4) and centrifuged at 10,000 ϫ g for 10 min at 4°C. The supernatant was collected for use in the aggregation experiments. The supernatant containing soluble A␤ was mixed with the same volume of either Tris buffer alone or buffer containing different concentrations of one of various metals (e.g. CuCl 2 , ZnCl 2 ). The concentration of A␤ in each test sample was ϳ10 M. At this point each sample was divided into two equal volumes; 1) one volume was diluted 10 times with buffer and termed "before," and 2) the remaining volume was incubated for 30 min at 37°C and termed "after." At this point, to assess A␤ concentration, the optical density (OD) of the "before" samples (OD before ) was measured using a DU640 UV spectrophotometer (Beckman) set at a wavelength of 214 nm.
After 30 min the "after" samples were ultracentrifuged at 100,000 ϫ g for 10 min at 4°C using an Optima TL Ultracentrifuge (Beckman). Supernatants were collected, diluted 10 times with buffer, and absorbance was measured (OD after ) as described above. The percentage of total aggregation was calculated as the ratio of the difference between OD before and OD after divided by OD before as follows.
͑OD before Ϫ OD after ) Ϭ OD before ϫ 100% (Eq. 1) Note that in all the samples containing metal ions, background signals were determined by measuring the OD of "blanks" containing only metals, and were subtracted from the OD of all test samples containing both A␤ and metals prior to calculating the percentage of total aggregation. Circular Dichroism (CD) Spectroscopy-Metal-induced changes in A␤ structure were determined as follows. First, 100 g of A␤ peptide dissolved in 100 l of HFIP was evaporated under nitrogen gas, then dissolved in 460 l of 20 mM Tris-HCl (pH 7.4) with or without 100 M metal ions (i.e. CuCl 2 , ZnCl 2 ). At this point the concentration of A␤ in each test sample was 50 M. Second, half of the A␤ sample was poured into a 1-mm pathlength quartz cuvette and promptly scanned with a J-720WI CD spectropolarimeter (Jasco). The remaining halves were incubated for 12 h at 37°C prior to being scanned. All CD measurements were carried out at 37°C using the following parameters: 1-nm bandwidth, 5 nm/min run speed, 1-nm step size, 2-s response time, and averaged over three runs. The molecular ellipticities, [⌰], of each sample were calculated using algorithms provided by the manufacturer. The CDs of the buffers used, and when applicable the CDs of each metal solution, were subtracted from the appropriate CDs of each sample prior to the calculation of molecular ellipticities.
Congo Red Staining-Congo Red staining was used to examine the morphology of metal-induced A␤ aggregates. A␤ stock solution was incubated with either 100 mM Tris-HCl (pH 7.4) alone or Tris buffer containing 20 M CuCl 2 or 20 M ZnCl 2 overnight at 37°C. Congo Red was then added to each sample and incubated for 20 min at room temperature. Samples were applied within a circular area delimited by a PAP Pen (Daido Sangyo) on APS pre-coated glass slides (Matsunami), and covered with cover glass (Matsunami). The slides were examined with an Eclipse E600 (Nikon) light microscope equipped with a polarizing filter.
Electron Microscopy-The ultrastructural characteristics of metalinduced A␤ aggregates were examined as follows. A␤40 was incubated for 13 h at 37°C with or without either 4 mM CuCl 2 or 4 mM ZnCl 2 as described above (see "ThT Assay"). Each sample was mixed for 10 s every 20 min. Three l of each sample was applied to 300-mesh copper grids with Formvar supporting membrane, blotted with filter paper, and stained with 1% (w/v) neutralized tungstophosphoric acid (Nacalai tesque, Inc., Kyoto, Japan). The specimens were examined on a Leo 912AB electron microscope (Leo, Oberkochen, Germany), at an accelerating voltage of 100 kV.
Cell Cultures-Two lines of cells, human embryonic kidney (HEK) 293 cells and neuronal hippocampal cells, were used to test the toxic effects of metal-induced A␤ aggregation as assessed in the MTT assay (see below). These cells were cultured as follows. HEK 293 cells were grown in Dulbecco's modified Eagle's medium (DMEM) (Sigma-Aldrich) containing 10% fetal bovine serum (HyClone) and incubated in a humidified chamber (85% humidity) containing 5% CO 2 at 37°C. On the morning of A␤ treatment, the cell culture medium was replaced with serum-free DMEM, and the cells were plated onto a 96-well coated plate (Corning) at a final cell count of 20,000 cells/well. The cell viability was subsequently assessed using the MTT assay (see below).
MTT Assay-The toxic effects of spontaneous ␤-aggregation and metal-induced ␤-aggregation of A␤ on cell cultures were assessed. Solutions containing either A␤ alone or A␤ and metal ions were prepared as described for the ThT assay. For the experiments using soluble A␤ samples, ␤-aggregation was assessed using the ThT assay, and then the samples were transferred immediately into wells containing the HEK 293 cell (see "Cell Cultures"). After certain periods, MTT was added to each well and the plate was kept in a CO 2 incubator for an additional 2-4 h. The cells were then lysed by the addition of a lysis solution (50% dimethylformamide, 20% SDS, pH 4.7) and were incubated overnight or longer. The degree of MTT reduction in each sample was subsequently assessed by measuring absorption at 570 nm at 37°C using a Bio Kinetics EL340 reader (Bio-Tek Instruments). Background absorbance values, as assessed from cell-free wells, were subtracted from the absorption values of each test sample. Percentage of MTT reduction was calculated by taking the average of values from a condition without both A␤ and metal as 100% in each experiment.
For the experiments using pre-incubated A␤ samples, the relative degree of ␤-aggregation in each sample was regularly monitored using the ThT assay until ␤-aggregation reached a maximum. At this point, the samples were transferred into wells containing either the HEK 293 cell cultures or the hippocampal neuronal cultures prepared as described above. In experiments carried out in neuronal cultures, serumfree Neurobasal medium (final concentration: 50%) without B-27 supplement was the condition of medium. Cell viability was assessed with the MTT assay as described above for the experiments using soluble A␤.

RESULTS
Using ThT, a fluorescent dye that specifically binds with fibrous structures (19,20), we determined the time course of ␤-aggregation formed by A␤40 and A␤42 in the presence of various metals (Fig. 1). As predicted, the ␤-aggregation kinetics were sigmoidal. The final values were remarkable, however, because they were apparently opposite to those in previous reports (11,14). Unexpectedly, the maximum ␤-aggregation occurred in conditions that did not contain metals (Fig. 1A), and lesser amount of ␤-aggregation occurred in conditions that contained the 20 M metal concentration. Additionally, the rank order of metal efficacy on ␤-aggregation, was nearly opposite to that reported previously (11) in which aggregation was measured using different methods.
To determine whether the particular metal concentration we initially chose contributed to these unexpected results, the ThT experiments were repeated with various concentrations of the metals. The final ␤-aggregation values shown in Fig. 2 reveal that both Zn(II) and Cu(II) inhibited the ␤-aggregation of both A␤40 and A␤42 in a concentration-dependent manner. The same results were obtained when the experiment was repeated using different commercial sources and lots of A␤. Using a different assay to assess aggregation, the OD 214 absorbance assay, we found, in contrast to the ThT results, both Zn(II) and Cu(II) effectively induced A␤40 aggregation (Fig. 3), although Cu(II) ranked second to Zn(II) in efficacy. These results using the absorbance assay are generally consistent with previous reports (14), allowing us to exclude the possibility that our unexpected results using the ThT assay were artifactual perhaps from using inappropriately prepared A␤, for example. In order to explain the apparently discordant results using the OD 214 or ThT assay, we considered the hypothesis that they measure different types of amyloid aggregates: the ThT assay directly measures only ␤-aggregation (19,20) because of the specificity of thioflavin T for fibrous structures, whereas the OD 214 assay indirectly measures all types of aggregation (11,14) because the optical density of the supernatant reflects the amount of remaining A␤ after centrifugation of unspecific aggregation compared with the amount before incubation. Moreover, because metal ions induced all types of aggregation of A␤40 (i.e. as measured in the OD 214 assay) but inhibited ␤-aggregation (i.e. as measured in the ThT assay), we hypothesized that metal ions must induce the formation of a second non-␤sheeted type of aggregate. This hypothesis was addressed by examining the secondary structure of A␤40 in either the absence or presence of various metal ions using CD spectroscopy.
The spectral curve of A␤ in the absence of Zn(II) or Cu(II) before incubation has a minimum ellipticity at around 200 nm (Fig. 4), indicating that the secondary structure of A␤ under these conditions is a random coil structure (39). After 12 h of incubation, the spectral curve for A␤ alone shifted to the right, displaying a minimum at around 210 nm (Fig. 4). This kind of curve is characteristic of the ␤-sheet structure. In contrast, the spectral curves of A␤ incubated with either Zn(II) or Cu(II) before and after incubation were completely different from either of these patterns (Fig. 4), indicating neither the ␤-sheet nor the random coil conformations were present. Both of these curves had slightly negative ellipticities from 220 to 230 nm and positive ellipticities around 203 nm. This positive ellipticity is especially conspicuous in the curve for the A␤ sample containing Cu(II) ions. These curves most closely resemble the type II ␤-turn conformation, although it is insufficient to make a definitive assignment using only CD spectra. Nonetheless, the differences in the spectroscopic pattern from ␤-sheet con- formation suggest that A␤ in the presence of metals take non-␤-sheeted conformation.
To confirm this notion experimentally, we determined whether Zn(II) and Cu(II) inhibit ␤-sheet formation of A␤. Congo Red, another ␤-sheet-specific dye, was added to samples containing either pre-incubated A␤40 alone or pre-incubated A␤40 treated with metals (Fig. 5). Congo Red binds ␤-sheeted A␤ fibrils to produce a characteristic "apple green" birefringence when viewed under polarized light (22)(23)(24)(25). Pre-incubated A␤ alone exhibited strong birefringence as expected (Fig.  5A). On the other hand, aggregates from samples incubated with Zn(II) were predominantly stained tinctorial red (Fig. 5C), as reported previously (11), displaying weaker birefringence. Congo Red-stained aggregates from A␤40 samples co-incubated with Cu(II) were light pink in color, not quite apple green (Fig.  5E). Together, these data show that Zn(II) and, to a lesser extent, Cu(II) prevent A␤ from forming ␤-sheet conformations in an as-yet-undetermined manner.
The ultrastructural differences between these aggregates were examined electron microscopically. When incubated alone, A␤40 formed fibrillar precipitates bearing extensive aggregations arranged in lateral arrays of fibers (Fig. 5B). The overall appearance, as well as the diameter of these fibers, are quite similar to those of fibrillar A␤ reported previously (40). Fig. 5 (D and F) shows the A␤40 aggregates in the presence of Zn(II) and Cu(II), respectively. A␤ aggregates from Zn(II)-treated samples are more spread out on the grid than those from the Cu(II)-treated samples. Nevertheless, both types of aggregates have granular rather than fibrillar structures. Taken together, all of these data suggest that metals prevent individual A␤ molecules from aggregating in a single direction, thus preventing fibrillogenesis (i.e. ␤-aggregation). Although the precise mechanism is unknown, it is quite possible that such metals exert their effects at the stoichiometric level, i.e. on peptide conformation.
Metals forced A␤ to form aggregates that were apparently not characterized by ␤-sheet fibrils. In the next set of experiments, we sought to determine whether non-␤ aggregates formed during Zn(II) or Cu(II) treatment were toxic to cells. The effects on cell viability were first assessed in a simplified model using HEK 293 cell cultures. After measuring the degree of ␤-aggregation in various concentrations of Zn(II) and Cu(II), each A␤ sample was transferred to cell cultures. Cell viability was indirectly measured as a function of the percentage of MTT reduced (Fig. 6). Since the dye MTT is known to be converted into a purple formazan by mitochondrial redox activity, MTT reduction assay has been widely used to measure cellular redox activity (26). Greater ␤-aggregation resulted in greater toxicity and clearly decreased cell viability. With concentrations greater than 10 M, both Zn(II) (Fig. 6A) and Cu(II) (Fig. 6B) effectively suppressed ␤-aggregation, resulting in an increase in cell viability.
Cultured rat primary hippocampal neurons also were less viable when ␤-aggregated samples of either A␤40 or A␤42 were introduced (Fig. 7). Neurotoxicity was reduced when the A␤ sample was pre-incubated with either Zn(II) or Cu(II), in parallel with the decreased ␤-aggregation level (Fig. 7, A and B). concentration of pre-incubated A␤40 (and, by inference, the concentration of ␤-aggregate) that was added to the test culture. With a 50 nM concentration of pre-incubated A␤40, the viability of neurons dropped to ϳ50%, dramatically illustrating the significance of fibrillogenesis for the induction of neurotoxicity.
Zn(II) and Cu(II) diminish the ability of A␤ to form ␤-aggregates, which results in protection against associated cytotoxicity. Although our experiments have focused on the role of these metals in disrupting an inherent cytotoxic property of ␤-aggregates, others have reported results on the roles of these metals as anti-and pro-oxidants in directly reducing or promoting cytotoxicity associated with A␤-generated oxidative stress (30 -33, 41, 42). In these reports, soluble A␤ was directly added to cell cultures, whereas in our latter two experiments (i.e. Figs. 6  and 7), pre-incubated samples of A␤ were added to the cultures. The question arises, then, about the relative roles of metals with respect to A␤-associated cell death and the soluble state of A␤, and whether there might be differing effects depending on this state.
This question motivated us to next compare the viability of cells after soluble A␤40 or pre-incubated A␤40, either with or without Cu(II), was added to HEK 293 cell cultures. Cytotoxicity was directly correlated with the attenuated ␤-aggregation level (Fig. 8, A and B). However, Cu(II) mixed with soluble A␤ also resulted in cell death (Fig. 8B). Therefore, comparison of the absolute values of the MTT assays from the four conditions suggests that 10 M Cu(II) in the presence of soluble A␤ may initially accelerates cell death via a possible oxidative stress mechanism by increasing the toxicity of soluble A␤ (30,31), but in parallel ameliorated the toxicity associated with ␤-aggregates.
Amyloid ␤-aggregation was most effectively inhibited in our experiments by Zn(II). Because it is not a redox-active metal, it also has a well characterized protective function against A␤generated oxidative stress, which had been documented in vitro (32,33,42). Given these results suggesting that Zn(II) is a doubly protective agent, it is puzzling that zinc has been reported to reside in senile plaques of AD brain (43,44) as well as in A␤ deposits in APP TG mice (45); its role in plaque assembly has been suggested before (43)(44)(45). One possible explanation for these in vivo findings is a unique condition we observed using the ThT assay in which ␤-aggregation of A␤40 was robustly induced in the presence of Zn(II) (Fig. 9). Unlike all the previous experiments in which the pH was maintained at 7.4, when pH is lowered to 6.8 in the presence of 200 M heparin, 20 M Zn(II) results in disproportionate ␤-aggregation, as shown in Fig. 9B. Because both a higher concentration and lower concentrations of Zn(II), under both pH 6.8 and 7.4 conditions, result in much less ␤-aggregation, this may represent an aberrant phenomenon, thus perhaps suggesting the pathological mechanism. Although it is curious that this increase was peculiar to the range of pH associated with tissue inflammation where the function of HSPG is focused, it requires more study to make a more confident statement about the physiological significance of this phenomenon. Overall, our data suggest that Zn(II) has predominately a strong protective role against ␤-amyloid toxicity. DISCUSSION After being processed from its precursor protein, ␤-amyloid changes its conformation to form aggregates, which are eventually deposited as senile plaques, one of the key pathological hallmarks of AD (1,2). Based on findings about this process, intense energy has been focused on discovering a way to interrupt this process at each stage. Initially motivated by anecdotal observations, the role of metals has been extensively examined from a variety of approaches and at systems, cellular, and macromolecular levels. At the protein level, Zn(II) and Cu(II) were shown to induce the aggregation of A␤, leading many to suspect a central, detrimental role of these metals (11)(12)(13)(14). At the cellular level, data from culture studies suggest the possible source of A␤ neurotoxicity is oxidative stress, which is potentially enhanced by redox-active metals such as copper and iron, and suppressed by the redox-inactive form of zinc (30 -33, 41, 42). Somewhat paradoxically, both zinc and copper are constitutively found in the brain and are suggested to participate in plaque assembly (43)(44)(45). Obviously, the question of whether these metals are beneficial or harmful needs further analysis. In this report, we propose a novel concept that different types of aggregates can be formed by A␤. We also describe the effects of metals on amyloidogenesis and the associated cytotoxicity.
Different Forms of Aggregates-Results from the ThT and OD 214 assays lead us to hypothesize that, in addition to the ␤-aggregate, another type of aggregate is formed in the presence of metals. This idea logically follows because these metals increased overall aggregation, but decreased ␤-aggregation. Data from conformational and microscopic analyses support this hypothesis. Although CD spectroscopic pattern may suggest that metals induce A␤ to take on ␤-turn conformation, more specific method such as NMR would certainly be necessary for the precise identification. Nonetheless, the secondary structure of metal-induced non-␤-aggregate is clearly distinct from the typical ␤-sheet structure inherent to self ␤-aggregates. Consistent with these spectral data, these aggregates exhibited weaker Congo Red birefringence than ␤-sheet fibrils formed in the absence of metals. Furthermore, ultrastructural examination of these aggregates revealed that they had a granular rather than fibrous appearance in contrast to aggregates formed in the absence of metals (i.e. ␤-aggregates).
Several reports suggest that metals have some role in the dimerization of A␤ (46 -48). Brain extracted A␤ species that are dimers formed granule-like aggregates after incubation, whereas monomers formed fibrous structures (49). Although the size of particles observed in our metal-induced aggregates is slightly larger than that of these dimer-originated aggregates, it is certainly tempting to suggest that the granule-like particles formed in the presence of metals are in fact constituted from metal-associated A␤ dimers. Further studies are required to resolve this issue.
The Role of Zinc and Copper in ␤-Aggregation and Cytotoxicity-The Zn(II)-and Cu(II)-induced decreases in ␤-aggregation, and thus fibrillogenesis, were directly correlated with decreased cytotoxicity in HEK 293 cell and rat primary neuronal cultures. At higher metal concentrations, ␤-aggregation was suppressed and the viability of cells was proportionately preserved. The deleterious effect of pre-incubated A␤ on neuronal cells was remarkable, even at submicromolar concentrations (see Fig. 7C), indicating that neurons are extremely susceptible to the toxic effects of A␤ fibers or ␤-aggregates. Although a direct comparison between A␤40 and A␤42 could not be made because of different treatment periods, we did observe that MTT-reducing activity was decreased more in samples containing A␤40 compared with those containing A␤42. These results were consistent with their degree of ␤-aggregation. This does not necessarily indicate that, physiologically, A␤42 is less toxic than A␤40. A␤42 is known, however, to be a more disease-relevant species because it has a more rapid nucleating activity than does A␤40 (50). This was clearly evident with the ThT assay in which A␤42 began to form ␤-aggregates almost instantly (Fig. 1B). The final amount of ␤-aggregates from A␤42, however, was less than that of A␤40, probably because more A␤42 molecules were consumed for nucleation and only a small amount remained for the growth phase. Therefore, in an in vivo environment in which the levels of different A␤ species are presumably not as limited as in this experimental set-up, A␤42 would certainly be more aggregative and thus more disease-relevant than A␤40.
The Dual Role of Copper in A␤-induced Cytotoxicity-The preponderance of our data clearly shows that ␤-aggregates are cytotoxic and that, regardless of the A␤ species (A␤40 or A␤42) involved, both Zn(II) and Cu(II) reduce cytotoxicity by inhibiting ␤-aggregation. On the other hand, we also observed, as have others, that soluble A␤ promotes cytotoxicity in the presence of Cu(II) (Fig. 8B). One explanation for this latter finding is that Cu(II) is capable of promoting potentially toxic, prooxidative reactions (30,31,41,42). Assuming that both of these apparently contradictory results represent events occurring in vivo, Cu(II) both facilitates oxidative stress, and presumably cell death, through its interaction with soluble A␤, and results in the formation of a precipitated complex that is less toxic than A␤ fibers. If this were indeed the case, one would then expect to see in the normal brain either the catabolism of newly secreted, soluble A␤ before it can react with redox-active metals like Cu(II) or the deposition of this A␤ as non-toxic aggregates instead of toxic ␤-aggregates. It is plausible that a combination of these processes may also occur.
A Hypothesis for the Role of Zinc in the Pathophysiology of AD-Similar to Cu(II), Zn(II) also inhibits ␤-aggregation by inducing the formation of non-fibrillar, non-toxic aggregates. Because zinc is a redox-inactive metal, in contrast to Cu(II), Zn(II) protects against soluble A␤-generated oxidative stress and its associated cytotoxicity (32,33).
Given these results, it is perhaps surprising that Zn(II) has also been suggested to play a role in plaque assembly in vivo (43)(44)(45). In order to provide some insight into these apparently contradictory findings and to explain the potential pathological roles of Zn(II), we performed an experiment that aimed to model the in vivo conditions reported to be present in the milieu around senile plaques of AD brain. With a slightly acidic pH of 6.8 and in the presence of heparin, we found that the unique concentration of 20 M Zn(II) robustly induced ␤-aggregation. Interestingly, this is the identical intraluminal concen- tration of Zn(II) found within cortical vasculature (51). Moreover, it is also well known that a lowering of pH can occur at sites of inflammation (14,35), as well as in the post-mortem AD brain (34).
The slightly acidic experimental condition together with heparin in our final experiment suggests a hypothetical mechanism of inflammation-associated amyloidogenesis. In this mechanism, HSPG was implicated (36,53) since it was observed in senile plaques and cerebrovascular amyloid (20,54) where inflammatory responses, such as activated microglia, were detected (55). As mentioned previously, slightly acidic conditions also prevail around inflammation (14,35). Interestingly, the proposed Zn(II)-binding amino acid residues include His-13 and His-14 (46,48,58), which can be highly protonated by the slight lowering of pH (from its pK a value of around 6.0), are also included within the potential binding site of HSPG (52,59). It is also relevant to note that more acidic pH values destabilize the bond between zinc and A␤, resulting in less aggregation (56,57), which is supposed to mean primarily zinc-induced, non-␤-sheeted aggregation. Therefore, at normal pH zinc promote the formation of non-␤-sheeted A␤ aggregates (11), while suppressing the fibrillogenesis of A␤ as shown in this report. When pH lowers slightly, the zinc-promoted, non-␤-sheet aggregates are destabilized (56,57) and in return the fibrillogenesis takes place only where the optimal conditions meet such as heparin and a certain stoichiometric level of zinc. Given these facts, we propose that this aberrant increase in ␤-aggregation in the presence of zinc may represent the pathological mechanism by which acidosis induces the liberation of A␤ from Zn(II). "Freed" A␤ may form ␤-aggregates or interact with redox-active metals such as Cu(II), resulting in the acidosis-promoted neurotoxicity around inflammation-associated senile plaque. Because our data show that higher concentrations of Zn(II) diminish even this phenomenon, we propose that Zn(II) could be one such candidate for natural protection against A␤ toxicity. Further investigations on the efficacy of Zn(II) to ameliorate the A␤ neurotoxicity associated with acidosis, especially in vivo, may lead to the identification of environmental factors to be targeted for overcoming AD.