The Native Copper- and Zinc- Binding Protein Metallothionein Blocks Copper-Mediated Aβ Aggregation and Toxicity in Rat Cortical Neurons

Background A major pathological hallmark of AD is the deposition of insoluble extracellular β-amyloid (Aβ) plaques. There are compelling data suggesting that Aβ aggregation is catalysed by reaction with the metals zinc and copper. Methodology/Principal Findings We now report that the major human-expressed metallothionein (MT) subtype, MT-2A, is capable of preventing the in vitro copper-mediated aggregation of Aβ1–40 and Aβ1–42. This action of MT-2A appears to involve a metal-swap between Zn7MT-2A and Cu(II)-Aβ, since neither Cu10MT-2A or carboxymethylated MT-2A blocked Cu(II)-Aβ aggregation. Furthermore, Zn7MT-2A blocked Cu(II)-Aβ induced changes in ionic homeostasis and subsequent neurotoxicity of cultured cortical neurons. Conclusions/Significance These results indicate that MTs of the type represented by MT-2A are capable of protecting against Aβ aggregation and toxicity. Given the recent interest in metal-chelation therapies for AD that remove metal from Aβ leaving a metal-free Aβ that can readily bind metals again, we believe that MT-2A might represent a different therapeutic approach as the metal exchange between MT and Aβ leaves the Aβ in a Zn-bound, relatively inert form.


Introduction
Alzheimer's disease (AD) is the most common form of dementia within the ageing population. AD accounts for between 50% and 60% of dementia cases [1]. The pathological hallmarks of the disease include extracellular b-amyloid (Ab) plaques, intracellular neurofibrillary tangles and dystrophic neurites [2]. All of these hallmarks arise from the abnormal and unregulated overproduction of insoluble proteinaceous structures. These proteinaceous structures are responsible for the disruption to normal cellular functioning ultimately leading to cell death.
The main constituents of Ab plaques are the 40-and 42-mer peptides, Ab 1-40 and Ab 1-42 [3], which associate to form abnormal extracellular deposits of fibrils and amorphous aggregates [4]. Ab is derived from the b-amyloid precursor peptide (APP), which is a normal protein [5,6] that is produced by neuronal and nonneuronal cells [7,8]. APP is processed by a combination of a-, b-, and/or csecretases to form numerous protein products. The plaque-forming Ab  and Ab 1-40 arise from the uncommon band c-secretase cleavage of the APP.
The mechanisms underlying the aggregation of Ab have been the subject of intense investigation. There is compelling data to suggest that the aggregation of Ab is catalysed by reaction with the metals zinc and copper. Cu-mediated aggregation of Ab leads to the formation of copper-bound, SDS-insoluble Ab aggregates, that are neurotoxic due to the ability of Ab-bound copper to undergo redox reactions at the cell membrane to generate reactive oxygen species [9,10]. Of note, Ab plaques deposited in the AD brain are found to be enriched with metals, and in particular copper [11,12].
Metallothioneins (MTs) are the major endogenous zinc-and copper-binding protein within the brain. The MT-1/2 isoforms (exemplified by the highly expressed member, MT-2A) are characterised as highly neuroprotective proteins essential for brain repair [13][14][15]. The metal-binding properties of MT-1/2 have been well investigated, and it is recognised that these proteins are capable of binding 7 divalent (Zn 2+ ) and up to 12 monovalent (Cu + ) metal ions in vivo through two distinct metal-thiolate clusters, termed the aand bdomains [16]. The importance of MT in maintaining metal homeostasis is clearly demonstrated in studies involving exposure to heavy metals (either by diet or environment) in MT-1/2 knockout mice, which leads to metal toxicity, while MT-1/2 overexpressing mice are relatively protected from heavy metal toxicity [see review , 17]. MTs also have important roles in copper homeostasis, evidenced by the crossing of a mouse model of Menkes disease (a copper efflux disease) with MT-1/2 knockout mice, which results in embryonic lethality [18].
Given the strong metal-binding properties of MT, we predict that these proteins may be involved in regulating the metalbinding and subsequent aggregation of Ab. Indeed, there is substantial literature supporting a role for MTs in the pathophysiology of AD. These proteins are expressed by astrocytes, and their expression is significantly upregulated in regions of Ab plaque pathology in the pre-clinical [19] and clinical AD brain [20,21], as well as in the brains of transgenic AD mice [22]. Notably, Meloni et al [23] recently reported that the Zn(II)-MT-3 isoform is capable of exchanging metals with Cu(II)-Ab to prevent the formation of SDS-insoluble Cu(II)-Ab aggregates. The goal of this study is to evaluate whether MT-2A, via its zinc-and copperbinding properties, also represents an endogenous protective mechanism against Ab aggregation and toxicity.

MT-2A prevents copper-mediated formation of SDSinsoluble Ab aggregates
To stimulate Ab aggregation, 25mM of Ab 1-40 was mixed with an equimolar concentration of copper and 200mM ascorbate, and shaken at 300rpm for three days at 37uC. The resultant protein aggregates were collected by ultracentrifugation, resuspended in SDS-PAGE loading buffer and electrophoresed. Under these conditions, no Ab 1-40 was visualised on SDS-PAGE ( Figure 1A), as a consequence of the formation of copper-bound SDS-insoluble aggregates (IA). In contrast, in the presence of Zn 7 MT-2A at the range of 2.5-25mM, Ab 1-40 formed aggregates (SA)( Figure 1B), which dissolved in SDS and resolved as a single band on SDS-PAGE of approximately 3-4kDa representing monomeric Ab 1-40 ( Figure 1A). As shown recently by Meloni et al [23] the structurally related metallothionein isoform Zn 7 MT-3 was also capable   , in the presence of 25mM copper and 200mM ascorbate, was incubated at 37uC for three days with shaking at 300rpm. This resulted in the formation of SDS-insoluble Ab aggregates (IA), which were not visualised on SDS-PAGE (A). The presence of Zn 7 -MT2A (5-25mM) prevented formation of SDS-insoluble Ab aggregates (A). Instead, SDS-soluble aggregates (SA) were formed, that were resolved as a single protein band of approximately 3-4kDa size, representing monomeric Ab 1-40 peptide (A). A range of different MT forms were tested for the ability to promote formation of SDS-soluble aggregates (SA)(B). Zn 7 MT-2A promoted formation of SDS-SA, and Zn 7 MT-3 had a similar effect but at 10-fold higher concentration (B). The ability of MT-2A to prevent formation of SDS-IA was linked to metal-binding properties, as different metallated forms of MT-2A had different effects upon Ab aggregation (B). When Cu-Ab 1-40 or Cu-Ab 1-42 aggregates were generated by three days of incubation, and then incubated with shaking in the presence or absence of 25mM Zn 7 MT-2A for up to three days, this was unable de-aggregate either Cu-Ab 1-40 or Cu-Ab 1-42 pre-formed aggregates (C). doi:10.1371/journal.pone.0012030.g001 of preventing formation of SDS-insoluble Ab 1-40 aggregates, but required a 10-fold higher concentration than Zn 7 MT-2A ( Figure 1B). We also investigated whether Zn 7 MT-2A can prevent the Cu-mediated aggregation of Ab   . Under the same  experimental conditions, 25mM Zn 7 MT2A was able to completely  prevent Cu-Ab 1-42 forming SDS-insoluble aggregates (results not  shown).
We also investigated whether MT-2A can de-aggregate preformed Cu-Ab aggregates. Cu-Ab 1-40 and Cu-Ab 1-42 aggregates were produced as described above (three day incubation), after which time 25mM of Zn 7 MT-2A was added. However, incubation with Zn 7 MT-2A for up to three days was unable to de-aggregate either Cu-Ab 1-40 or Cu-Ab 1-42 pre-formed aggregates ( Figure 1C).
To establish whether the metallation state of MT is responsible for the ability of MT-2A to prevent aggregation of Ab 1-40 into SDS-IA, several different metallated forms of MT-2A were used. We found that Cu 10 MT-2A was not able to prevent the formation of SDS-insoluble Ab 1-40 aggregates ( Figure 1). Furthermore, chemical modification of MT-2A to block metal binding, by means of carboxymethylation of cysteine residues (CaMeMT-2A) abolished the ability of MT-2A to prevent formation of insoluble Ab 1-40 aggregates ( Figure 1). Finally, metal free (apo) MT-2A was also unable to prevent copper-mediated formation of insoluble Ab  . Based upon these observations, we predict that the zinc bound to MT is required for the ability of MT to block coppermediated Ab 1-40 aggregation.
Evidence that Zn 7 MT-2A prevents copper-mediated Ab aggregation via metal exchange One possible explanation for our observations is that there is a metal exchange of copper and zinc between Zn 7 MT-2A and Cu(II)Ab with the subsequent formation of Zn-bound Ab aggregates and soluble Cu-bound MT. To test this hypothesis, the amount of copper and zinc present in the aggregated Ab samples was determined using inductively coupled plasma mass spectrometry (ICP-MS). When Ab 1-40 was aggregated for three days in the presence of copper and ascorbate, the amount of copper/zinc detected within the pellet fraction following ultracentrifugation was 89.2/27.1ng (Table 1). However, when Ab  was aggregated in the presence of Zn 7 MT-2A, the amount of copper present in the pellet fraction was significantly reduced to 47.5ng (approximately 53% decrease), while there was a concomitant increase in zinc to 81.4ng (Table 1). There was no change in copper/zinc levels however when CaMeMT-2A was used (Table 1). This provides further evidence that a copper-zinc exchange has occurred between Cu(II)-Ab and Zn 7 MT-2A. We therefore predict that the action of MT-2A to prevent the formation of insoluble Ab aggregates is due to metal exchange of copper and zinc between Zn 7 MT-2A and Cu(II)Ab with the subsequent formation of Zn-bound Ab which only forms soluble protein aggregates.  [24]. However, differences were observed in the composition and affinities of individual copperthiolate clusters of MT-2 and MT-3. In ESI-MS studies in the presence of 10mM DTT, we observed that copper was bound to MT-3 in a mixture of major Cu 12 MT-3 and minor Cu 10 MT-3 forms ( Figure 2B), while we have recently reported that MT-2A is predominantly found in Cu 10 MT-2A form [24]. Addition of the high affinity Cu(I)-binding chelator DETC at 0.5 mM concentration was sufficient to convert Cu 12 MT-3 to a predominant Cu 6 MT-3 form, which could be further demetallated to apo-MT-3 at 3 mM DETC ( Figure 2B). These observations indicate that MT-3 binds copper in two distinct hexacopper clusters exposing different Cu(I)-binding affinities. Conversely however, Cu 10 MT-2A was stable at up to 1.0 mM DETC, and that 1.5mM DETC was required to partially demetallate Cu 10 MT-2A to Cu 6 MT-2A [24]. This suggests that tetracopper-thiolate cluster in Cu 10 MT-2A has higher Cu(I)-binding affinity than low-affinity hexacopperthiolate cluster in Cu 12 MT-3, and may account for the greater ability of MT-2A to prevent Cu(II)-Ab aggregation compared to MT-3. We predict from the amino acid composition of MT-2A and MT-3 that due to their high sequence homology that the Nterminal b-domains will share similar metal-binding properties ( Figure 2C), most likely corresponding to the hexacopper-thiolate cluster that is demetallated by treatment with 3mM DETC. Subsequently, the C-terminal a-domain is likely to represent the metal thiolate cluster that differs between the two isoforms and which may contribute to the different copper-binding properties of MT-2A and MT-3. To test this hypothesis we switched the domains between isoforms to form chimeric recombinant MT-3b2a and MT-2b3a proteins (note the the beta-domains of each chimera are at the N-terminus, which is the same as native MTs). Indeed, Zn 7 MT-3b2a was capable of preventing copper mediated Ab aggregation to a similar degree as Zn 7 MT-2A, while the Zn 7 MT-2b3a chimeric protein had a similar activity to Zn 7 MT-3 ( Figure 2D).

MT-2A protects against Ab toxicity in cultured cortical neurons
It has been reported previously that Cu(II)-Ab 1-40 but not Zn(II)-Ab 1-40 is toxic to cultured neurons [12,25]. The combination of equimolar concentrations of Ab 1-40 and Cu(II) ions results in all free copper becoming rapidly bound to Ab 1-40 to form Cu(II)-Ab 1-40 [23]. We maintained rat cortical neurons for three days in vitro (3DIV) followed by treatment with 40mM Cu(II)-Ab 1-40 and 300mM ascorbate. After 24 hours, this treatment resulted in about an 80% reduction in cell viability, as measured using an alamarBlueH cell viability assay ( Figure 3A). In parallel experiments we performed direct cell counts to validate the alamarBlueH data, and observed a Table 1. 25mM Ab 1-40 was mixed with 25mM copper and 200mM ascorbate (rapidly forming Cu(II)-Ab), and incubated with shaking at 37uC for 72 hours, resulting in the formation of aggregated Ab 1-40 .

Reaction
Copper (ng) Zinc (ng) Treatment with ascorbate or Ab 1-40 alone had no effect upon viability (results not shown). Note that addition of MT-3 to neuronal cultures can have powerful neurotoxic effects in its own right, if applied in the presence of a brain derived extract or serum [26,27], neither of which are present here. In contrast, in the absence of these agents, MT-3 alone has no discernible neurotoxic effect.
To further investigate whether it is Cu(II)-Ab and not free copper that is responsible for neurotoxicity, we fractionated a 40mM Cu(II)-Ab solution on a PD MidiTrap TM G-25 column and determined the protein (A280) and metal (ICP-MS) content of each fraction. Only those fractions containing peptide also contained copper ( Figure 4A), indicating that the Cu(II)-Ab solution did not contain any free copper ions. The neurotoxicity of all fractions was subsequently tested, and only those containing Cu(II)-Ab exhibited neurotoxic activity ( Figure 4B), demonstrating that Cu(II)-Ab is responsible for the neurotoxicity that we have observed in our experiments.
To investigate whether MT-2A acts protectively via the zinccopper metal exchange between Zn 7 MT-2A and Cu(II)-Ab described above, Cu 10 MT-2A was used in the place of Zn 7 MT-2A, which resulted in no neuroprotection against Cu(II)-Ab ( Figure 5A). This suggests that the metal exchange between MT-2A and Cu(II)-Ab is required for neuroprotection. Notably, Cu 10 MT-2A alone was not toxic to neurons ( Figure 5A), confirming that when copper is bound to MT it is unable to produce ROS. In parallel experiments, we found that Zn 7 MT-2A was only able to mildly block the neurotoxicity of H 2 O 2 when applied to cultured neurons ( Figure 5B). The amount of H 2 O 2 directly applied to the neurons was physiologically relevant as Huang and colleagues [28] found that 10mM Ab 1-40 or Ab 1-42 may generate up to 25mM H 2 O 2 in 1 hour in the presence of substoichiometric amounts of Cu(II), depending on the oxygen tension. Since H 2 O 2 is the direct ROS product of the interaction of Cu(II)-Ab with cells [23], this suggests that the protective action of Zn 7 MT-2A is primarily upstream of ROS production. This is consistent with our results suggesting that Zn 7 MT-2A acts via a metal swap with Cu(II)-Ab to prevent ROS formation.
To further confirm that Zn 7 MT-2A is blocking the detrimental effects of oxidative stress induced by Cu(II)-Ab, we have measured changes in ionic homeostasis of neurons in response to Cu(II)-Ab using a microelectrode ion flux measuring (MIFE) technique. The MIFE technique allows direct measurement in changes in K + and Ca 2+ fluxes in response to Cu(II)-Ab. Treatment with Cu(II)Ab 1-40 (in the presence of ascorbate) induced a rapid efflux of K + out of neurons, peaking at 4.260.65 min after Cu(II)-Ab application with K + outflow continuing over a period of 20 minutes ( Figure 6A). The treatment also led to a moderate influx of Ca 2+ , peaking at 6.4561.35 min after the Cu(II)-Ab application (indicated by crossing zero line) with Ca 2+ uptake continuing over the experimental period ( Figure 6C). Treatment with Zn 7 MT-2A completely blocked Cu(II)-Ab induced changes in K + and Ca 2+ fluxes, at a concentration of .5mM ( Figure 6B, D). Individual treatments with either Ab 1-40 or Zn 7 MT-2A alone (in the presence of ascorbate) had no effect upon any of the ions measured (results not shown).

Discussion
The major findings of this study are that the major humanexpressed subtype of metallothionein, MT-2A, is capable of preventing the formation of the toxic Cu-mediated aggregates of It is well established that MT-1/2 expression is elevated in response to almost all forms of stress to the brain, including traumatic-, ischaemic or chemical-brain injury [25,29,30], or in neurodegenerative conditions such as ALS or EAE (an experimental animal model of multiple sclerosis). The common consensus is that MT-1/2 act neuroprotectively, via intracellular functions such as metal detoxification and quenching of oxidative free radicals. More recently, it has been demonstrated that MT-1/2 can be actively secreted by astrocytes under certain pathophysiological conditions [15], and subsequently act from an extracellular location directly upon neurons to activate intracellular neuroprotective pathways [15,31,32]. However, an unexpected and specific protective function of MT in the AD brain has recently been proposed by Meloni and colleagues [23], who have found that Zn 7 MT-3 is able to prevent copper-mediated aggregation of Ab in vitro, and protect a neuronal cell line from soluble Ab 1-40 toxicity. There is some controversy over the level of expression of MT-3 in the AD brain, although it appears that expression is downregulated in AD. Furthermore, the expression of MT-3 is not induced by metals or oxidative stress, and this protein displays neurotoxic actions under some conditions [14,26], suggesting that it is unlikely that this protein contributes greatly to a protective mechanism against Ab aggregation and toxicity. The MT-1/2 isoforms however (as exemplified by MT-2A) are broadly expressed within the adult brain, and greatly elevated levels of expression of these proteins have been noted in the AD brain [20][21][22]. We now report that MT-2A is also capable of preventing copper-mediated Ab aggregation (both Ab 1-40 and Ab 1-42 ), and that this involves a specific metal exchange interaction. Hence, Zn 7 MT-2A (but not Cu 10 MT-2A or CaMeMT-2A) prevents soluble Cu(II)-Ab 1-40 from forming SDS-insoluble Ab  aggregates. This observation might be explained by the wellcharacterised ability of higher binding affinity metals (ie: copper) to displace lower affinity metals (ie: zinc) within MT [for review see 33]. Similarly, published reports indicate that Ab has a Cu(II) binding affinity of between 10 26 to 10 211 [34,35], while MT-2A has been reported to have a binding affinity for Cu(II) of 10 219 [36]. This supports the proposition that MT-2A is capable of removing Cu(II) ions from Cu(II)-Ab. Interestingly, we found that apo-MT-2A (metal free MT-2A) was unable to prevent soluble Cu(II)-Ab from forming SDS-insoluble Ab aggregates. Apo-MT-2A has a high affinity for free copper initially suggesting that it might be able to extract Cu from Ab aggregates. It is possible that the apo-MT rapidly becomes oxidised (via disulphide-bond formation between cysteine residues), which would prevent apo-MT from binding to metals. Or it might be that the ability of MT to participate in an intermolecular interaction with Ab under biological conditions is dependent on not just the relative copper binding affinity between the two proteins, but also on the tertiary structure of the zinc-metallated form of metallothionein, which is fundamentally different to the uncoordinated structure found in the apo-thionein. Finally, ICP-MS demonstrated that there was almost 40% less copper in the pellet fraction when Cu(II)-Ab was aggregated in the presence of Zn 7 MT-2A. In summary, we provide strong evidence that a metal exchange between Cu(II)-Ab and Zn 7 MT-2A is responsible for blocking the aggregation of Ab into an insoluble form. In our studies, we noted that Zn 7 MT-2A was effective at 10-fold lower concentrations than Zn 7 MT-3 in preventing coppermediated Ab 1-40 aggregation. We predict that this difference might be related to the relative copper binding affinities of MT-2A and MT-3, and based upon our experiments with chimeric MT proteins we believe that the a-domain of MT-2A is particularly  important in this activity. We note that the relative affinity of the aand bdomains of MT for copper means that the activity of MT-2A and MT-3 towards copper-induced Ab aggregation will probably depend on the molar ratio of MT to Ab. In this regard, it is generally considered that the b-domain of MT fills with copper first, followed by the a-domain. We show that MT-2A will prevent aggregation of Cu(II)-Ab even at low relative levels of this isoform (eg 0.5-5 mM MT to 25 mM Ab), whereas MT-3 requires a higher MT:Ab ratio (25 mM MT to 25 mM Ab) for activity. This most likely also explains the difference between MT-2A and MT-3 in the neurotoxicity data, in which we used 40mM Cu(II)-Ab to 20mM MT. We predict that when the ratio of Ab to MT is equimolar that it will be primarily the b-domains of MT that will be filled with copper, and that in this scenario MT-2A and MT-3 will protect equally against Cu(II)-Ab neurotoxicity.
We also found that Zn 7 MT-2A protects neurons against Cu(II)-Ab toxicity. We believe that this involves a zinc/copper metalexchange between Zn 7 MT-2A and Cu(II)-Ab that subsequently prevents Ab-bound copper from participating in redox-reactions and producing reactive oxygen species [as suggested in 23]. Hence, only Zn 7 MT-2A but not Cu 10 MT-2A or CaMeMT-2A was capable of protecting cortical neurons from soluble Cu(II)-Ab neurotoxicity. Notably, Cu 10 MT-2A itself was not toxic to neurons, indicating that when copper is bound to MT (for instance following metal-swap with Cu(II)-Ab) that it is unable to produce ROS. As evidence that MT-2A is not acting downstream by scavenging the H 2 O 2 generated by Cu(II)-Ab, we found that Zn 7 MT-2A provided only a small degree of protection against direct H 2 O 2 neurotoxicity. This suggests that Zn 7 MT-2A neuroprotection against Cu(II)-Ab is primarily afforded through a zinc/copper metal swap and subsequent inhibition of H 2 O 2 generation. We do note that it is possible that when Cu(II)-Ab is applied to neurons that some copper is released from the complex, and that this free copper is also partly responsible for neurotoxicity. Although, as noted by Meloni et al [23], we think that only a small amount of copper is released from Cu(II)-Ab upon addition into the culture medium and the primary cause of neurotoxicity is the Cu(II)-Ab complex.
Finally, using a direct and sensitive technique, MIFE, to measure the kinetics of ion fluxes, we were able to determine that oxidative stress (H 2 O 2 ) generated by soluble Cu(II)-Ab induces a rapid net efflux of K + and mild net influx of Ca 2+ into cortical neurons most likely via increased oxidative stress. Notably, K + efflux from cells is a well established trigger of apoptosis. These observations are also in accordance with the results of other groups who have demonstrated that H 2 O 2 induces substantial dysregulation in neuronal K + channel conductance [37] and calcium homeostasis [38]. Importantly, we found that treatment with Zn 7 MT-2A completely abolished the Cu(II)-Ab-induced changes in K + and Ca 2+ net ion fluxes. Taken together, our data suggests that the neuroprotective actions of MT-2A lie in its ability to exchange metals with Cu(II)-Ab to stop production of ROS, and preventing subsequent detrimental changes in ionic balance within neurons.
We believe that our results may reflect a physiological action of MT-2A in the Alzheimer's brain. For instance, MT-1/2 levels in the adult human brain have been reported to be approximately 40mg/g [39]. Expression is primarily by astrocytes, with very low levels of MT-2A expressed in neurons. We have recently reported that secretion of MT-2A by cultured astrocytes can be induced under certain physiological conditions and that extracellular MT-2A can be detected in the site of a physical injury to the brain [15]. Hence, it is conceivable that under stressful situations MT-2A may be secreted from astrocytes into the synaptic vicinity and reach the levels that we have demonstrated are capable of modulating Cumediated Ab aggregation.
Because of the considerable published data linking metalbinding to the aggregation of Ab, metal-chelation drugs have been proposed as a potential therapy for AD [40,41]. An excellent example of this approach is the administration of the copper-and zinc-chelating drug clioquinol, which has been reported to prevent plaque formation in transgenic AD mice [42]. The use of such metal-chelating drugs might not only reduce metal-mediated aggregation of Ab, but also limit the formation of Cu(II)-Ab and thus prevent the generation of oxidative stress and subsequent neurotoxicity. One criticism of metal-chelation therapies for AD is that these chelating agents remove metal from Ab leaving a metalfree Ab that could feasibly readily bind metals again. Hence, more recently, metal redistribution has been proposed as a more appropriate goal of metal-targeted strategies for AD [41]. In this regard, we believe that MT-2A might represent a possible candidate as a metal-redistribution therapeutic agent, as the metal exchange between MT and Ab leaves the Ab in a Zn-bound nontoxic form, and redistributes copper into an inert Cu-MT form.
In summary we provide compelling evidence that MT-2A can protect against copper-induced Ab aggregation and neurotoxicity. This action of MT-2A appears to involve a metal-swap between Zn 7 MT-2A and Cu(II)-Ab Furthermore .. MT-2A can block Cu(II)-Ab induced changes in ion homeostasis and neurotoxicity of cultured cortical neurons. We propose that there is therapeutic potential in a MT-2A based approach to reducing Ab deposition in AD. All MT proteins were provided directly from Bestenbalt in lyophilised form in either a metal free (apo), or in Zn 7 MT or C 10 MT state, Metallation of MT-2A was prepared as we have described previously for Zn 7 MT-3 [43]. Briefly, the protein was dissolved in 20 mM Tris-HCl, pH 8 and the pH was lowered to 2.5. Ten equivalents of Zn 2+ or 12 equivalents of Cu + was added and the pH raised to 8. The buffer was exchanged to 10 mM ammonium bicarbonate, and the solution frozen at 280uC and freeze dried. The lyophilised proteins were reconstituted in Milli-Q water (pH 7.4) immediately prior to use.

In vitro copper-mediated Ab aggregation assay
Synthetic monomeric Ab 1-40 and Ab 1-42 were purchased from EZBiolab (US). The dried Ab peptide contains trifluoroacetate (TFA) as a counterion, and in analysis it was found that the synthetic Ab used in this study contained approximately 5-10% TFA. For in vitro Ab aggregation studies, purified Ab 1-40 and Ab 1-42 (EZBiolab) was dissolved in Tris buffer (20mM Tris-HCl, 100mM NaCl, pH 7.4) to give a 25mM solution, followed by addition of an equimolar concentration of CuCl 2 and 200mM ascorbate. The solution was incubated at 37uC for 72hrs with shaking (300 rpm). Ab aggregation was assessed by gel electrophoresis. Briefly, aggregates were collected by ultracentrifugation at 20,000G for 1 hour then resuspended in LDS (lithium dodecyl sulfate) sample buffer (Invitrogen) and electrophoresed under reducing conditions (b-mercaptoethanol in sample and Invitrogen anti-oxidant supplement in the running buffer) on a 10% Nu-Page Bis-Tris gel (Invitrogen) at 200V for 30 minutes. Protein bands were visualized using Coomassie brilliant blue stain. Aggregates were assessed as SDS-soluble when the Ab protein was observed in monomeric form on the gel and as SDS-insoluble when not observed on the gel.

Electrospray ionisation mass spectrometry (ESI-MS) metal binding analysis
Copper binding affinity of MT-3 was determined in strictly similar conditions and by identical approach, which we have recently elaborated for MT-2A [24]. Briefly where I CunMT-3 denotes the intensity of the Cu n MT-3 peak in the ESI-MS spectra. The fractional occupancy of Cu(I) binding sites in MT-3 was correlated with the concentration of free Cu(I) ions in the sample calculated using the apparent dissociation constant for DETC that we have recently determined using the same techniques (K Cu = 13.8 fM) [24]. The obtained binding curve for MT-3, presented in Fig. 2C, was fitted nonlinearly with the Hill equation (equation 2) and also linearly to the linear version of the Hill equation with the program ''Origin 6.1'' (OriginLab Corporation, USA).

Y~C u(I)
The nonlinear fitting presented in Fig. 2C [24]. A Hill coefficient close to 1 indicates that there exists only weak apparent positive cooperativity in the binding of Cu(I) ions to MT-3, whereas strong positive cooperativity (n = 3.3) exists in case of MT-2, which is demetallated in very narrow range of free Cu(I) ions [24]. As seen from Fig. 2B there are two metal-thiolate clusters in Cu 12 MT-3, both composed from 6 Cu(I) ions. The first hexacopper-thiolate cluster dissociates readily in the presence of 0.5 mM DETC, whereas DETC can not dissociate copper from Cu 10 MT-2 even at 1 mM concentration [24]. The second hexacopper-thiolate cluster of MT-3 is half desaturated at 3 mM DETC, which is similar to the behaviour of hexacopper-thiolate cluster in MT-2 [24].

Inductively coupled plasma mass spectrometry (ICP-MS) metal content analysis
In some cases, the Ab aggregation pellet and supernatant fractions were collected post-ultracentrifugation and their metal content determined by Inductively Coupled Plasma Mass Spectrometry (ICP-MS). Prior to analysis samples were further diluted 106 with ultra-pure water (.18 MOhm) with nitric acid and Indium (as an internal standard) addition to final concentrations 1% and 100 ppb, respectively. Analysis was undertaken using an ELEMENT High Resolution ICP-MS operating in medium resolution mode, enabling 63 Cu and 66 Zn isotopes to be monitored free from overlapping spectral interferences. Both elements were quantified using external calibration methodology, with blank subtraction. Typical analytical protocols have been presented previously [44].

Rodent cortical neural cell cultures
All animal procedures were performed in accordance with the animal ethics guidelines of the University of Tasmania Animal Ethics Committee. Neural cultures were prepared as reported previously [45], and briefly involved the removal of cortices from embryonic day 17 Hooded Wistar embryos, which were incubated with 0.1% trypsin in HEPES buffer at 37uC for 20 minutes. After three washes with warmed HEPES, the tissue was triturated and plated at 5610 4 cells per coverslip onto 13mm 2 glass coverslips in Neurobasal medium (Gibco) and maintained at 37uC in humidified air containing 5% CO 2 .

Rat cortical neuron toxicity assay
To induce cortical neuron toxicity, 40mM soluble Ab 1-40 was applied to neurons in the presence of 40mM CuCl 2 and 300mM ascorbate. Under these specific conditions, it has been demonstrated that all free copper is rapidly bound by Ab 1-40 [23]. The presence of physiological levels of ascorbate permits cycling between copper (II) and (I) oxidation states, as occurs within cellular environments allowing copper to bind to proteins in either Cu(I) or Cu(II) oxidation states. After 24 hours, neuronal viability was measured by the degree of cellular metabolic reduction of alamarBlueH, determined by fluorescence (excitation 535nm, emission 595nm), and was expressed as the percentage of the signal obtained from the vehicle-treated culture. Ab 1-40 was used in this study because there is a wider differential in the relative toxicity of copper vs zinc forms of Ab 1-40 compared to Ab 1-42 (ie: only Cu-Ab 1-40 and not Zn-Ab 1-40 is neurotoxic, while both Cu-and Zn-Ab 1-42 are neurotoxic). Using Ab 1-40 thus maximises the effect of the hypothesised metal swap between ZnMT-2A and Cu-Ab 1-40 , and removes the additional but complicating possibility that metallothionein can independently protect against Zn-Ab 1-42 toxicity.

Statistical analyses of tissue culture experiments
For each experiment unless otherwise stated, a minimum of four wells from at least three separate cultures (derived from different animals), were used for quantification, blinded to conditions. Statistical analysis was completed using SPSS 16.0 (SPSS). When data was unequally distributed, data was transformed so that the residuals were approximately normally distributed. Statistical significance was calculated using One-Way and Two-Way ANOVA with Tukey's Post Hoc Test. All graphical data is presented as mean 6 SEM, significance p,0.05.

Ion-selective flux measurements
The theory of non-invasive microelectrode ion flux (MIFE) measurements was reviewed recently [45] and the complete experimental procedure including ion-selective microelectrode fabrication and cell preparation and immobilisation are given elsewhere [46,47]. Cortical neurons for the MIFE measurements were grown for three days at a 1610 5 cells/well on poly-L-lysine cover slips as described above. By day three a dense monolayer of neurons had developed. Cells were washed in and adapted to the MIFE artificial CSF (aCSF) for one hour prior to experiments. The composition of the aCSF was: 150mM NaCl, 0.5mM KCl, 0.5mM CaCl 2 , 1.5mM MgCl 2 , 1.25mM NaH 2 PO 4 , 5mM NaH-CO 3 , 25mM glucose, pH 7.2. Data was acquired at a rate of 15 samples/sec and later averaged over 10 second intervals. Each experiment was repeated upon at least four different coverslips from three different neuronal cultures. Figure S1 Direct cell counting revealed that treatment of 3DIV rat cortical neuron cultures with 40mM Cu(II)Ab1-40 resulted in significant neuronal death, which could be blocked by the coaddition of 20mM of Zn7MT-2A. Error bars represent standard error of the mean calculated from at least three different experiments. * -p,0.05 (One-Way ANOVA). Found at: doi:10.1371/journal.pone.0012030.s001 (3.21 MB TIF)