Zinc-induced Alzheimer’s Aβ1–40 Aggregation Is Mediated by Conformational Factors*

The heterogeneous precipitates of Aβ that accumulate in the brain cortex in Alzheimer’s disease possess varying degrees of resistance to resolubilization. We previously found that Aβ1–40 is rapidly precipitated in vitro by physiological concentrations of zinc, a neurochemical that is highly abundant in brain compartments where Aβ is most likely to precipitate. We now present evidence that the zinc-induced precipitation of Aβ is mediated by a peptide dimer and favored by conditions that promote α-helical and diminish β-sheet conformations. The manner in which the synthetic peptide is solubilized was critical to its behaviorin vitro. Zinc-induced Aβ aggregation was dependent upon the presence of NaCl, was enhanced by α-helical-promoting solvents, but was abolished when the peptide stock solution was stored frozen. The Aβ aggregates induced by zinc were reversible by chelation, but could then be reprecipitated by zinc for several cycles, indicating that the peptide’s conformation is probably preserved in the zinc-mediated assembly. In contrast, Aβ aggregates induced by low pH (5.5) were not resolubilized by returning the pH milieu to 7.4. The zinc-Aβ interaction exhibits features resembling the gelation process of zinc-mediated fibrin assembly, suggesting that, in events such as clot formation or injury, reversible Aβ assembly could be physiologically purposive. Such a mechanism is contemplated in the early evolution of diffuse plaques in Alzheimer’s disease and suggests a possible therapeutic strategy for the resolubilization of some forms of Aβ deposit in the disease.

The pathological hallmark of Alzheimer's disease is the abundant accumulation in the brain of A␤, a 39 -43-amino acid peptide, as morphologically heterogeneous deposits in the neuropil (senile plaques) and cerebral blood vessels (congophilic angiopathy) (1,2). A␤ is a soluble component of cerebrospinal fluid where it is found in concentrations in the low nanomolar range (3)(4)(5). Many studies now indicate that synthetic A␤ becomes toxic to cultured neuronal cells when in a specific ␤-sheet conformation that involves incubating the synthetic peptide in a aqueous solution over time periods of days to weeks (6 -14). Whereas the ␤-sheet conformation of the peptide is found in highly insoluble fibrillar amyloid deposits in Alzheimer's disease, a substantial proportion of A␤ precipitates into nonfibrillar deposits that can be resolubilized by extraction into aqueous solvents (15).
We have recently reported that A␤ itself specifically and saturably binds zinc, manifesting high affinity binding (K D ϭ 107 nM) with a 1:1 (zinc:A␤) stoichiometry and low affinity binding (K D ϭ 5.2 M) with a 2:1 stoichiometry (16). This binding is probably histidine-mediated since it is abolished by acidic pH (no binding at pH 6). The zinc-binding site was mapped to a stretch of contiguous residues between positions 6 -28 of the A␤ sequence. Occupation of the zinc-binding site, which straddles the lysine 16 position of ␣-secretase cleavage (17,18), inhibits ␣-secretase type (tryptic) cleavage, and so may influence the generation of A␤ from amyloid protein precursor (APP) and may increase the biological half-life of A␤ by protecting the peptide from proteolytic attack (16).
We also found that concentrations of zinc Ն1 M rapidly destabilize human A␤1-40 solutions, inducing rapid A␤ precipitation (19) that is highly specific for zinc, although both copper(II) and iron(II) can induce partial aggregation at equivalent concentrations (20). Meanwhile, rat A␤1-40 (with substitutions of Arg 3 Gly, Tyr 3 Phe, and His 3 Arg at positions 5, 10, and 13, respectively) binds zinc less avidly (K a ϭ 3.8 M, with 1:1 stoichiometry) and is unaffected by zinc at these concentrations, perhaps explaining the scarcity with which these animals form cerebral A␤ amyloid (21,22). In the absence of zinc, the solubilities of the rat and the human A␤ species are indistinguishable (19). We observed that iodinating the peptide on the 10th residue of tyrosine attenuated zinc-mediated precipitation (19), and since this residue is substituted with a phenylalanine in the rat species, we concluded that this tyrosine is critical in coordinating zinc to the human peptide. These observations are important because zinc is abundant in the same neocortical regions where A␤ deposits are most commonly found, and high micromolar zinc concentrations are achieved during glutamatergic neurotransmission (24,25), suggesting an explanation for the propensity of A␤ to deposit close to the neocortical synaptic vicinity. Recently, we have also found evidence that zinc mediates the assembly of a significant fraction of A␤ deposits in Alzheimer-affected postmortem brain tissue. 1 Hence, an elaboration of the interactions between A␤ and zinc in vitro may be germane to the pathology of Alzheimer's disease.
The concentration of zinc required to precipitate A␤1-40 in vitro has been in disagreement with results reported recently by Esler et al. (26), who have claimed that concentrations no less than 100 M are required to demonstrate appreciable precipitation of the peptide. In particular, the validity of the filtration assay for A␤ aggregation used in our previous study was challenged by these workers who contended that 125 I-A␤1-40 is a suitable tracer for monitoring the interaction of A␤ with zinc, at variance with our findings (26). Rodriguez et al. (27) have recently reported that zinc concentrations below 100 M induce the abundant and immediate precipitation of soluble A␤1-40, in agreement with our initial reports (16,19,20) and in disagreement with the findings of Esler et al. (26). This debate is important for two reasons. First, if the concentration of zinc required to precipitate A␤ is, in fact, over 100 M, then it is very unlikely that this interaction is of any neurobiological significance. Much of the zinc that is released by synaptic transmission (24,25) may not be available for exchange with A␤ since it will in large part be sequestered by macromolecules and other ligands (28). If Ͼ100 M zinc is required to induce A␤ precipitation, the majority of the zinc released during neurotransmission would need to exchange with A␤ if synaptic zinc were to be a factor in the peptide's accumulation. This is unlikely. If, on the other hand, low micromolar concentrations of zinc are required to precipitate A␤, then only Ϸ1% of the total zinc released during glutamatergic neurotransmission would be required to induce A␤ assembly, making such an interaction far more likely. Second, our findings (19) questioned the use of 125 I-A␤ as a valid tracer for unmodified A␤ behavior, and in our hands, the effect of low micromolar zinc upon the precipitation of A␤ differentiated the iodinated peptide from unmodified A␤. This is important since radiolabeling of A␤ by iodination is a common means of creating a marker for the peptide.
To explore the reasons for the variance between our findings and those of Esler et al. (26), we studied physicochemical factors that influence the interaction of zinc with synthetic A␤1-40, and used turbidometry as a highly specific, although relatively insensitive, method for monitoring peptide aggregation. We now report that the striking precipitation of A␤1-40 by low micromolar concentrations of zinc is sensitive to complex factors in the buffer milieu that impact upon the peptide's conformation and polymerization state. These factors may explain the variance that exist between our earlier findings and the report of Esler et al. (26). Our findings also indicate that the conformational state of the peptide may not be perturbed by precipitation with zinc. These data may provide insights into neurobiological factors that influence the solubility of A␤ in the pathophysiological environment of the brain in Alzheimer's disease.

EXPERIMENTAL PROCEDURES
Reagents and Preparation-Human A␤1-40 amyloid peptide was used in all experiments, synthesized, purified, and characterized by HPLC 2 analysis, amino acid analysis and mass spectroscopy by W. M. Keck Foundation Biotechnology Resource Laboratory (Yale University, New Haven, CT). The HPLC elution profile of A␤1-40 peptide in preparations used in these experiments was identified as a sole peak in the eluate. Amino acid analysis of the synthetic peptide indicated that there were no apparent chemical modifications at amino acid residues. Mass spectroscopy was performed on each batch of peptide as a further confirmation.
Milli-Q water (Millipore Corp., Milford, MA) was used for A␤ solubilization and other stock reagent dilution. Buffers were treated with Chelex-100 resin (Bio-Rad) to minimize trace metal contamination, and filtered through a 0.22-m cellulose acetate filter unit (Corning Costar Corporation, Cambridge, MA). The background zinc concentrations in the buffers were measured at Ͻ0.1 M by ion-coupled plasma-atomic emission spectroscopy. Standard zinc stock solution (10 mg/ml in 10% HCl, U. S. National Institute of Standards and Technology, Gaithersburg, MD) was used in all experiments. All other reagents were at least analytical grade.
Synthetic A␤ peptide solutions were prepared on the day of the experiment according to the protocol of Evans et al. (29), except where indicated. Lyophilized peptide was first solubilized in water to reach 500 M and then indirectly sonicated for 3 min (30 s on, 10 s off) through a water bath to avoid frothing. The peptide preparation was then filtered through a water-washed Spin-X cellulose acetate filter unit (0.22 , Corning Costar Corporation). Sonication and filtration are considered to be critical procedures to remove any trace of peptide microparticulate matter. Concentrations of A␤1-40 were determined by BCA protein assay (Pierce). The validity of the BCA assay for the measurement of A␤ peptide concentrations in these solutions was confirmed by amino acid analysis.
Turbidometric Assay of Zinc-induced A␤1-40 Aggregation-Turbidity measurement as an assay for aggregation was performed according to established protocols (29 -32) with minor modifications. The reactions were performed at room temperature in a flat-bottom 96-well microtiter plate (Corning Costar Corporation), and absorbances (405 nm) were measured using a V max kinetic microplate reader directed by Softmax version 2.32 software (Molecular Devices Corporation). Automatic 30-s plate agitation mode was selected for the plate reader to evenly suspend the aggregates in the wells before all readings.
To investigate the reversibility of zinc-induced A␤1-40 aggregation, 25 M zinc and 25 M A␤1-40 were mixed in 150 mM NaCl, 50 mM HEPES, pH 7.4 (200 l), and turbidity measurements were taken at four 1-min intervals using a 96-well plate reader. Subsequently, 20-l aliquots of 10 mM EDTA or 10 mM zinc (prepared in incubation buffer) were added into the wells alternately, and following a 2-min delay, a further four readings were taken at 1-min intervals. After the final EDTA addition and turbidity reading, the mixtures were incubated for an additional 30 min before taking final readings.
To investigate the reversibility of pH 5.5-induced A␤1-40 aggregation, A␤1-40 was brought to 25 M in 150 mM NaCl, 50 mM HEPES, pH 7.4 (200 l), and its absorbance was read at 405 nm as the background reading. The pH of the solution was then brought to 5.5 by the addition of concentrated HCl (5.5 l), and turbidity measurements were taken at four 1-min intervals. Subsequently, concentrated NaOH (7.5 l) was added into the wells to adjust the pH back to 7.4, and following a 2-min delay, a further four measurements were taken at 1-min intervals. These cycles were repeated as indicated, and the pH of the mixture was constantly monitored with a pH probe.
To further determine the state of aggregation of the incubated peptides in the reversibility experiments, replicate samples (300 l) representing various zinc-containing or chelated conditions were removed from the incubation tray at various time points, pelleted (10,000 ϫ g for 15 min), and either the supernatant was measured for remaining peptide content before and after centrifugation using the BCA assay, or the pellet was stained with 50 l of Congo Red (1% in 50% ethanol for 5 min). Pellets were washed twice with 50% ethanol (100 l) before being resuspended in 20 l of HEPES buffer. An aliquot (3 l) of each resuspension was placed on a microscope slide uniformly for microscopic analysis under polarized light.

RESULTS AND DISCUSSION
To characterize synthetic A␤1-40 in neutral buffered saline, gel-filtration chromatography was performed. A␤1-40 solution (2.3 M) was freshly prepared in TBS from its lyophilized powder (see "Experimental Procedures") on the day of the experiment. The peptide solution was loaded onto the gel-filtration column within 1 h of preparation, and eluted as a single sharp peak corresponding to 8.6 kDa compared to molecular size markers, and at an estimated concentration of 470 nM (Fig.  1), compatible with the peptide being in a dimeric state. Other groups have reported the presence of dimeric A␤ using gelfiltration chromatography (33,34), and our data are in agreement with the recent report that A␤1-40 is predominantly dimeric upon gel-filtration in neutral buffered saline (27). However, in the absence of NaCl, we found that the peptide's elution profile became too broad to allow its relative molecular size to be resolved (data not shown), suggesting that the apparent dimerization of the peptide is dependent upon the presence of NaCl. Since we wished to achieve a basic data set that describes the behavior of the most abundant species of A␤ under conditions that approach a physiologically plausible milieu, we proceeded to study the behavior of the peptide in isotonic neutral buffered saline, mindful that the concentration of NaCl appears to enact significant conformational effects upon the peptide.
Some studies have reported chromatographic profiles of A␤ peptides that appear nondimeric (monomeric or oligomeric). Apart from differences in the composition of the solvent system used for the chromatographic procedure, these alternative results may have been due to differences in the behaviors of the specific subspecies of A␤ peptide that were studied, reported differences in the preparation of the A␤ peptide, and variations in the experimental procedure (35)(36)(37). For example, in our previous study of A␤1-40 by gel-filtration chromatography, we observed that the peptide migrated mainly as an apparent dimer (65%) together with minor apparent polymer (30%) and monomer (5%) peaks (16). The difference between our current result and the previous result can be explained as due to two newly introduced variables. First, we have introduced sonication in the preparation of A␤1-40 peptide solution, which may have contributed to the dissolution of A␤1-40 polymer into dimeric or monomeric species. Second, in the current study only 5 g of A␤1-40 were applied to the column compared with 55 g in the earlier study. Therefore, absorbance readings at 214 nm in this study are much closer to the base-line buffer absorbance reading, and the absorbance may not be sufficient to demonstrate a significant monomeric peak, since, based upon our earlier data, its proportion is expected to be small in the total A␤ peptide eluent. Therefore, in the current study the presence of a small proportion of monomeric peptide cannot be excluded.
To confirm whether the concentration of zinc required to induce A␤1-40 precipitation is in the low or high micromolar range, we studied the behavior of the peptide by turbidometry, because it is a well established method that has been used to study the aggregation state of A␤ (29 -32). We first determined that the absorbance (405 nm) value of freshly prepared A␤1-40 (10 M) in TBS was equivalent to the absorbance of the experimental buffers used (data not shown), indicating that the presence of the soluble A␤ peptide does not contribute to turbidity in this system at the time frame studied. We chose the minimal zinc binding domain (ZBD) at the amino terminus of APP, which has a K a for zinc binding of Ϸ750 nM (ZBD, residues 179 -189 of APP) (38), as a zinc-binding control peptide for comparison to the behavior of A␤1-40 in the experiments. The background absorbance turbidity of ZBD (10 M in TBS, pH 7.4) was found to be of the same as that of A␤1-40 in TBS alone (data not shown), and incubation of ZBD solutions with the zinc concentrations used in this study did not increase their turbidity, indicating that the ZBD peptide does not aggregate in the presence of zinc.
We proceeded to study conditions representing concentrations of peptide (10 M) and zinc (Ͻ100 M) for which Esler et al. (26) could find no evidence of aggregation. There were negligible changes of absorbance readings for solution of 10 M A␤1-40 mixed with 0.1, 0.5 and 1 M zinc, compared with 10 M A␤1-40 alone. However, the absorbance increases of A␤1-40 with 5 and 10 M zinc were substantially above background ( Fig. 2A), corroborating our earlier findings (16,19,20). Although we had previously found a significant degree of A␤1-40 aggregation induced by 1 M zinc as measured by an absorbance-monitored filtration assay (19), the turbidometry assay used here did not detect A␤1-40 aggregation at zinc concentrations below 5 M. This result may be due to the sensitivity limitation of the turbidometry technique. After A␤ solutions were incubated with zinc concentrations below 5 M, the reaction may have produced particles of insufficient caliber to alter the light scattering properties of the starting solution. Whereas a strength of the turbidometric approach is that a positive signal is a valid indicator of aggregation, a weakness of the approach is its lack of sensitivity in that a negative signal (no change in absorbance at 405 nm) does not necessarily indicate that the solution is free of microparticles.
Zinc-induced A␤ aggregation was instantaneous, and remained constant for 24 h (data not shown). Zinc-induced turbidity was abolished by the presence of the divalent metal ion chelator EDTA (1 mM), and was substantially reduced by the presence of 10 M ZBD ( Fig. 2A), indicating that the ZBD effectively acts as a chelator in this system. Since these data show that concentrations of zinc less than or equal to 10 M clearly induce significant A␤ aggregation, they are in disagreement with the conclusions of Esler et al. (26), who did not employ turbidometry in their assays.
Incubation of A␤1-40 (10 M) with higher (50 -300 M) zinc concentrations indicated that the degree of turbidity engendered by incubation with zinc matched the degree of turbidity engendered by incubating the peptide at pH 5.5 (Fig. 2B). pH 5.5 is the calculated pI of the peptide, and incubation of the peptide at this pH is a precipitation stress for A␤ (39) that has been previously validated by turbidometry (40), and induces ␤-sheet formation (41). Turbidity changes induced by pH 5.5 and by zinc concentrations Ն50 M, saturated at approximately 0.12 absorbance unit. As a further corroborative study, the corresponding mixtures were pelleted by centrifugation (10,000 ϫ g, 10 min), and the proportion of the peptide remaining in the supernatant as determined by protein assay was Ͻ10%, indicating that most of the starting peptide had precipitated (data not shown).
We investigated factors in the experimental incubation conditions to determine which variables could affect the response of synthetic A␤ to zinc. We found that "aged" A␤1-40, which has been kept in water at Ϫ20°C for more than 2 months, was strikingly unable to be precipitated by zinc ( Fig. 2A), despite exhibiting no increase in turbidity after being thawed, gently mixed, and refiltered. A␤ stock solutions that were stored for as little as 7 days were also found to have an attenuated response to zinc. These data indicate that to appreciate the maximum effects of low micromolar concentrations of zinc upon the solubility of A␤, freshly prepared peptide should be used for every experiment, a procedural practice that was not employed by Esler et al. (26), who stored their stock peptide solutions at Ϫ20°C for unreported lengths of time before use. Since timedependent loss of helical content of A␤ within an unstable region of the peptide that is within the zinc-binding site, which spans residue 6 to 28 (16), has been previously described (42), we surmised that the loss of interaction with zinc that we observed upon peptide storage may be due to a conformational change, so we proceeded to study other conformation-influencing factors in the microenvironment of the peptide that may impact upon A␤ interaction with zinc.
To investigate effects of NaCl upon zinc-induced A␤1-40 aggregation, 10 M zinc and 10 M A␤ were incubated in HEPES buffers in the presence of various NaCl concentrations. We observed that the turbidity of these solutions strongly depended upon the concentration of NaCl, so that in the absence of NaCl, zinc-induced precipitation of A␤ was markedly attenuated (Fig. 3). The degree of precipitation induced under these conditions was also confirmed by centrifugation of the samples (10,000 ϫ g, 10 min) and measuring the amount of peptide remaining in the supernatant by protein assay. These findings are again at variance with those of Esler et al. (26), who reported that the presence of NaCl had no effect upon the degree of A␤ aggregation induced by zinc. Since the presence of 150 mM NaCl increases the ionic strength of the solution, these data again support the likelihood that a change in structure of the peptide is responsible for the variance between our reports and that of Esler et al. Because the apparent dimerization of the peptide (see Fig. 1) and the zinc-mediated precipitation of A␤ are both dependent upon the presence of NaCl, the likelihood exists that zinc-mediated A␤ assembly may be coordinated by the peptide as a dimeric subunit.
Because our data pointed toward a specific conformation of the peptide as mediating the interaction with zinc, we next studied the effect of solvents that promote helical conformation (TFE) or reduce ␤-sheet conformation (Me 2 SO) upon zinc-mediated A␤ aggregation. The effects of these solvents upon A␤ secondary structure have been characterized (43,44). We observed that the presence of these solvents either within the reaction (10 -30% TFE) or limited to the preparation of the stock solution (75% Me 2 SO, 25% HFIP), induced a substantial increase in zinc-induced aggregation of A␤1-40 compared with the amount of precipitation induced by zinc in the absence of solvent (Fig. 4). Although these solvents would normally attenuate the fibrillization of the peptide (43,44), we find that zinc-induced A␤ aggregation is promoted by preserving the ␣-helical conformation.
We tested the zinc-induced A␤ aggregates for reversibility by chelation to determine whether precipitation by zinc causes the peptide to preserve its secondary structure or changes it into a thermodynamically irreversible ␤-sheet conformer. We observed that zinc-induced turbid aggregates are totally reversed by removal of the zinc and that the peptide may alternate between a precipitated and nonprecipitated state for many cycles of fluctuating zinc concentration (Fig. 5A). This result indicates that the peptide does not lose conformational energy, once it interacts with zinc, and probably retains its original conformation without entering into a lower energy ␤-sheet conformation. To test this hypothesis further, we exposed the peptide to incubation at pH 5.5, an environment known to induce ␤-sheet formation (41). Under these circumstances the peptide solutions became turbid, and the turbidity of the preparation would not return to baseline even after the pH of the incubation was returned to pH 7.4 (Fig. 5B). One previous study indicated that A␤1-28 and A␤1-39 but not A␤1-42 exhibit reversible pH-induced aggregation. In that study, 0.1 M sodium acetate buffer at pH 5.0 was used to induce the aggregation of iodinated A␤ species (A␤1-28, A␤1-39, and A␤1-42), the peptide aggregates were resuspended in 0.1 M MOPS buffer of pH 7.4, and their reversibility was quantified by radioactivity counting (39). In the current study, pH-induced A␤1-40 aggregation was appraised by turbidometry using nonmodified A␤1-40 in HEPES-buffered saline. Two important variables may explain the discrepant results between the two studies. First, we studied the behavior of an A␤ species, A␤1-40, that is more hydrophobic than the A␤1-28 and A␤1-39 studied by Burdick et al. (39). The solubility of A␤ and its propensity to form the irreversible ␤-sheet is related to the length of its hydrophobic carboxyl terminus (30,32). More likely, however, is the possibility that the modification of the A␤ peptide by , and turbidity measurements were taken at four 1-min intervals. Subsequently, 20-l aliquots of 10 mM EDTA or 10 mM zinc (prepared in incubation buffer) were added into the wells alternatively, and following a 2-min delay, a further four readings were taken at 1-min intervals. After the final EDTA addition and turbidity reading, the mixtures were incubated for an additional 30 min before taking final readings. The data indicate the mean (ϮS.D., n ϭ 3) absorbance (405 nm) changes against the absorbance reading of the incubation buffer alone. B, reversibility of pH 5.5-induced A␤1-40 aggregation. A␤1-40 was brought to 25 M in 150 mM NaCl, 50 mM HEPES, pH 7.4 (200 l), and its absorbance read at 405 nm as the background reading. The pH of the solution was then brought to 5.5 by the addition of concentrated HCl (5.5 l), and turbidity measurements were taken at four 1-min intervals. Subsequently, concentrated NaOH (7.5 l) was added into the wells to adjust the pH back to 7.4, and, following a 2-min delay, a further four measurements were taken at 1-min intervals. These cycles were repeated as indicated. The data indicate the mean (ϮS.D., n ϭ 3) absorbance (405 nm) changes against the absorbance reading of the initial incubation buffer. iodination has rendered the peptide more easily resolubilized by alkalinization, perhaps by making it more difficult for the peptide subunits to align in an antiparallel ␤-sheet conformation that might be expected as the conditions induce the peptide to adopt the ␤-sheet conformation. Importantly, our findings that irreversible aggregation of A␤1-40 is induced by low pH are in agreement with another study that employed the same turbidometric technique (40). Our experiment was conducted with only a 5-min precipitation phase (pH 5.5), which induced instantaneous turbidity, and a 30-min resolubilization phase (pH 7.4), which may have been insufficient to achieve sufficient equilibrium to promote resolubilization. However, the study of Wood et al. (40) showed that similar pH-induced aggregation of A␤1-40 cannot be reversed, even after returning the pH to neutral for 2 days, supporting the likelihood that irreversible precipitation of the peptide differentiates zinc-induced from pH-induced A␤ aggregation. Taken together, these reports indicate that irreversibility of pH-induced A␤ aggregation varies between different A␤ species and different peptide modifications. Results of A␤1-40 aggregation induced by pH show that, once the A␤1-40 peptide has lost conformational energy and precipitated in a ␤-sheet conformation, the thermodynamic barrier to a soluble conformation is difficult to reverse, again suggesting that zinc-mediated aggregation is not due to ␤-sheet formation since it is readily reversed by chelation.
To confirm that zinc-mediated A␤ aggregation is not due to ␤-sheet formation, we sampled the precipitate formed in these reactions, stained it with Congo Red, and visualized the product by microscopy under polarized light. Although the peptide formed abundant congophilic precipitates as a consequence of both zinc and pH 5.5 incubations, only the A␤ precipitate formed at pH 5.5 exhibited positive birefringence typical of the ␤-sheet conformation of amyloid. Therefore, the A␤ precipitate induced by zinc under these specific conditions is not tinctorial amyloid. To confirm that the loss of turbidity that occurred when the zinc-induced A␤ suspension was treated with chelator was, in fact, due to resolubilization rather than due to an alteration in the light-scattering properties of a suspension, a sample of zinc-induced A␤ suspension that had been treated with EDTA was stained with Congo Red, pelleted, and visualized by microscopy. No particulate matter was seen, suggesting that the reversal of turbidity caused by zinc chelation reflected resolubilization of the peptide precipitates (data not shown).
We had previously observed that incubation of A␤1-40 with zinc induced a congophilic precipitate that exhibited positive birefringence under polarized light, meeting one of the criteria for the morphology of amyloid (19). In that report, we also demonstrated that up to 32% of the zinc-induced A␤1-40 precipitate could be resolubilized by EDTA chelation. In the current report, we demonstrate that, under these conditions, 100% of the zinc-induced A␤1-40 precipitate could be resolubilized by EDTA chelation. We hypothesized that this variance was due to a procedural difference for A␤1-40 peptide preparation implemented since our earlier report. In the current report, we prepared A␤1-40 solutions by sonication of the lyophilized peptide suspension, so generating a solution that migrates as an apparent dimer in TBS, pH 7.4 (Fig. 1). In the previous report, we did not sonicate the peptide, although in both reports the peptide stocks were filtered. Nonsonicated A␤1-40 migrates on gel-filtration chromatography as both an apparent dimer and a multimer (16). We consider that the presence of multimeric A␤1-40 probably reflects incomplete dissolution and that the propagated energy provided to the peptide through thermal heating by sonication has most likely facilitated the complete dissolution of the peptide into dimeric units, allowed the peptide to assume a high energy ␣-helical confor-mation, and is probably a better reflection of the physiological state of the peptide since the ␣-helical conformer is most stable in solution. The presence of multimeric A␤ in the peptide solution may alter the thermodynamics of A␤ assembly and possibly help bring about conditions that favor irreversible fibrillization. To test this hypothesis, we repeated the study of the effects of zinc and chelation cycles upon A␤ aggregation (Fig.  5A) using peptide that had been freshly solubilized and filtered, but not sonicated. This preparation, while readily exhibiting turbidity when precipitated by zinc, did not then appreciably resolubilize with EDTA chelation, and staining with Congo Red revealed the presence of tinctorial amyloid particles (data not shown), confirming our earlier report (19). These data confirm that the presence of multimeric A␤1-40 in the starting peptide preparation favors the irreversible formation of amyloid-containing particles by zinc by a mechanism that is still unclear.
The degree of zinc-induced turbidity in the chelation reversibility experiment increased following subsequent cycles of chelation (Fig. 5A), despite the stoichiometric ratio of free zinc to chelator remaining Յ1 in each zinc-induced cycle. Therefore, it is possible that the ionic interaction with zinc may induce conformational changes in the peptide that subsequently alter further interaction with zinc. Conversely, the reversal of the turbidity of zinc-treated peptide suspensions induced by EDTA chelation exhibited a time-dependence in the order of 30 min to reach complete resolubilization (Fig. 5A). Hence, although the association of zinc with A␤ to form insoluble assemblies is near-instantaneous, the dissociation appears to possess slower kinetics. This may be due, in part, to the time required for the EDTA to access the interstices of the zinc-induced aggregates.
Our data, in agreement with those of Garzon-Rodriguez et al. (27), suggest that the major interacting soluble A␤1-40 species recruited during zinc-induced aggregation are dimeric. The obligatory binding sequence for zinc mapped to a contiguous stretch of residues between 6 and 28 (19). However, if A␤ dimerization is a prerequisite for zinc binding, this domain (residues 6 -28) may actually reflect a region of the peptide responsible for homotypic dimerization. Our data also show that zinc-induced assembly of A␤ is not mediated by ␤-sheet formation, suggesting that the assemblies of zinc and A␤, which precipitate in 1:1 stoichiometry (19), are likely to be integrated by ionic interactions within the charged region of the zinc-binding site. This site is within an unstable region of the peptide that is in equilibrium between ␣-helix and ␤-sheet (42). The variance between our findings and the findings of Esler et al. (26) may therefore be explained by factors, such as those we have described in this report, which could alter the conformational equilibrium in this region.
The striking versatility of the solubility state of A␤ is unexpected, since the presence of the peptide as a collection in Alzheimer's disease pathology has always suggested that its precipitation is pathological. However, A␤ also collects as diffuse deposits, which presumably are more readily resolubilized than mature amyloid, in the brains of individuals who do not have Alzheimer's disease (45) and following head injury (46). We hypothesize that the presence of A␤ precipitates in the brain in these conditions may reflect a physiological purpose. The zinc-mediated reversible precipitation of A␤ can be compared with the interaction of zinc with fibrin whose solubility state is dependent upon zinc for physiological purposes (47). A␤ is found in platelets (48) where it may colocalize with high concentrations (Ͼ300 M) of zinc as well as the APP (49), fibrinogen, and other coagulation factors (50) in the ␣-secretory granule. Platelet zinc release modulates the propagation of the coagulation cascade through its action on protease activities, such as the effect of zinc at micromolar concentrations in se-lectively enhancing the inhibition of coagulation factor XIa by protease nexin-2/APP (51), probably through binding to the zinc-binding site on the amino terminus of APP (38). Zinc release also modulates the assembly of clotting structural elements such as fibrin in the vicinity of the activated platelet (47). In the current report we have found that zinc appears to mediate a reversible aggregation response from A␤, similar to the gelation response of fibrin incubated with low micromolar concentrations of zinc (47). A␤ could conceivably be recruited by high regional concentrations of zinc into the hemostatic plug or a reversible assembly that may play a role in maintaining tissue integrity in injury.
The abundance of zinc that is released by neurons (24), and housed by neuroglia (52), could potentially interact with A␤ during an injury response to induce the diffuse assembly of the peptide while maintaining the ␣-helical conformation of the peptide. Under these conditions, A␤ assemblies would not be expected to be neurotoxic since they would lack the ␤-sheet conformation.
The instantaneous assembly of 10 M A␤1-40 by 10 M zinc in neutral buffered saline is such a robust feature of the peptide's behavior that we suggest this provocation as a test of the patency of the synthetic peptide and its ␣-helical content. The propensity of A␤ to precipitate in the brain neocortex and not peripherally may reflect the high regional zinc concentrations in the neocortex (28), and we contemplate a possible role for zinc-mediated A␤ assembly in the evolution of Alzheimer's disease plaque pathology. Reversible zinc-mediated A␤ assembly may be reflected in the A␤ that is extractable by aqueous solvent from Alzheimer's disease-affected cortex (15); however, if these precipitates undergo a further physicochemical stress that induces a loss of ␣-helical content, they may then evolve into the more insoluble amyloid deposits. The enrichment of A␤1-42 in cortical amyloid may reflect a liability that the more insoluble A␤1-42 species has to develop the irreversible ␤sheet conformation. We have previously found that soluble A␤1-42 is as sensitive to zinc-mediated precipitation as A␤1-40 (data not shown), but the reversibility of zinc-induced A␤1-42 aggregates remains to be determined.