Characterization of the Interaction between the Wilson and Menkes Disease Proteins and the Cytoplasmic Copper Chaperone, HAH1p*

Wilson disease (WD) and Menkes disease (MNK) are inherited disorders of copper metabolism. The genes that mutate to give rise to these disorders encode highly homologous copper transporting ATPases. We use yeast and mammalian two-hybrid systems, along with anin vitro assay to demonstrate a specific, copper-dependent interaction between the six metal-binding domains of the WD and MNK ATPases and the cytoplasmic copper chaperone HAH1. We demonstrate that several metal-binding domains interact independently or in combination with HAH1p, although notably domains five and six of WDp do not. Alteration of either the Met or Thr residue of the HAH1p MTCXXC motif has no observable effect on the copper-dependent interaction, whereas alteration of either of the two Cys residues abolishes the interaction. Mutation of any one of the HAH1p C-terminal Lys residues (Lys56, Lys57, or Lys60) to Gly does not affect the interaction, although deletion of the 15 C-terminal residues abolishes the interaction. We show that apo-HAH1p can bind in vitroto copper-loaded WDp, suggesting reversibility of copper transfer from HAH1p to WD/MNKp. The in vitro HAH1/WDp interaction is metalospecific; HAH1 preincubated with Cu2+ or Hg+ but not with Zn2+, Cd2+, Co2+, Ni3+, Fe3+, or Cr3+ interacted with WDp. Finally, we model the protein-protein interaction and present a theoretical representation of the HAH1p·Cu·WD/MNKp complex.

excess, both cupric and cuprous ions are highly toxic, because they act as electron transfer intermediates and catalyze the formation of hydroxyl radicals. This copper-induced production of reactive oxygen species results in DNA damage, as evidenced by strand breaks and by base oxidation of guanosine within cellular DNA and in lipid peroxidation of membranes, especially in mitochondria and lysosomes (7). Thus, proper copper trafficking is essential to cell vitality.
Dietary copper is absorbed into the body through the intestinal mucosa where it joins recycled endogenous copper secreted into the gastrointestinal tract from other tissues. In general, dietary copper absorption is dependent upon the amount of copper reabsorbed from the fluids of other tissues. Newly absorbed copper is transported to body tissues in two phases. First, albumin, transcuprein, amino acids, and a group of uncharacterized low molecular weight proteins transport the majority of exchangeable copper to the liver. After traversing the basolateral membrane of hepatocytes, copper is distributed to endogenous copper-requiring enzymes and to secreted cuproproteins such as ceruloplasmin, which is thereafter released to plasma for delivery of copper to other tissues. Any excess copper is excreted to the bile by the way of the canalicular (apical) plasma membrane (see Refs. 3 and 4 for reviews).
Two P-type ATPases have recently been characterized in humans, ATP7A and ATP7B (8 -12). Mutations in these two genes lead to disorders of copper starvation (Menkes disease) and copper toxicity (Wilson disease), respectively. ATP7A is expressed in all tissues except liver, and mutations in this gene prevent the normal absorption and distribution of copper throughout the body. The resulting copper depletion leads to multi-system disorder and death in childhood. The ATP7B gene is expressed in most tissues but predominantly in the liver. Mutations in this gene lead to excessive copper buildup in the liver. A characteristic feature of WD 1 is the abnormally low levels of active (copper-bound) ceruloplasmin. Recent studies show that the molecular components of copper trafficking pathways are highly conserved between yeast and humans. In Saccharomyces cerevisiae, the pathway begins with Cu ϩ uptake through the action of copper transporters CTR1p and CTR3p (13,14). In the cytoplasm, copper is bound by cytoplasmic copper chaperones that then deliver the metal to specific target enzymes (15)(16)(17)(18)(19)(20)(21). The ATX1p copper chaperone appears to deliver copper to CCC2p, the yeast homologue of the WD and MNK ATPases (15)(16)(17). CCC2p is then required to transfer copper from the cytosol to the lumen of the trans-Golgi network where the multi-copper ferrooxidase Fet3p, yeast homologue of ceruloplasmin, is loaded (22,23). Lin and Culotta (15) provided the first biochemical evidence for interaction between ATX1p and CCC2p. Based on their subcellular localization, Lin and Culotta (15) suggested that ATX1p works as a cytoplasmic copper carrier protein that delivers copper from CTR1p to CCC2p. Recently, Pufahl et al. (17) used the yeast two-hybrid protocol to demonstrate a copperdependent interaction between ATX1p and CCC2p.
In this study we show that human homologues of the yeast proteins ATX1p and CCC2p interact in a copper-dependent manner. Using three independent assays we characterize the interaction of HAH1p (Human ATX1-like Homolog) with the homologous Wilson disease (WDp) and Menkes disease (MNKp) proteins. The WD and MNK genes encode six MTCXXC motifs in the N-terminal portion of their proteins, whereas their prokaryotic and yeast counterparts typically encode one or two such motifs (24 -26). Using deletion constructs of the WDp metal-binding domain, we provide evidence that a single MTCXXC motif is sufficient to interact with the HAH1p chaperone. Mutagenesis analysis demonstrates that the interaction is dependent on the co-coordination of copper ions by cysteine residues within the MTCXXC motifs of the interacting HAH1 and WD proteins. The data suggest a mechanism for the intermolecular transfer of copper ions similar to the "bucket brigade" model previously proposed for the trafficking of toxic ions in bacterial mercury detoxification (27,28).
Yeast Two-hybrid Assay-The yeast two-hybrid assay was performed according to the protocol supplied by the yeast two-hybrid system supplier (CLONTECH).
Mammalian Two-hybrid Assay-HepG2 cells (American Type Culture Collection) were grown in Dulbecco's modified Eagle's medium (Life Technologies, Inc.) containing 10% fetal calf serum, 100 M nonessential amino acids, 100 units/ml penicillin, and 100 g/ml streptomycin, at 37°C in 5% CO 2 . Cells were seeded at a density of 3.2 ϫ 10 6 /100-mm tissue culture dish the day before transfection. Transfection of HepG2 cells with plasmid DNA was performed using the DOSPER liposomal transfection reagent (Roche Molecular Biochemicals) according to the manufacturer's instructions. 12.5 g of both the pM-and pVP16-based plasmids, 2.5 g of reporter plasmid pG5CAT, and 100 l of DOSPER reagent were used per dish. Four days after transfection the cells were lysed, and CAT expression was assayed using the CAT enzyme-linked immunosorbent assay kit (Roche Molecular Biochemicals).
In Vitro Interaction Assay-Escherichia coli SG20050 cells were transformed with the plasmids pHTb-HAH1, pHTb-WM, pHTb-WT, pHTb-WC1, pHTb-WC2, pHTb-W15N, and pHTb-W15C. Proteins were expressed using the ProEx-HT Prokaryotic Expression system (Life Technologies, Inc.) and purified using the nickel nitrilotriacetic acid spin kit (Qiagen). The amylose resin bound fusion product, maltosebinding protein-WDp metal-binding domain, was purified essentially as described by Lutsenko et al. (29). Maltose-binding protein was obtained from New England Biolabs. Protein concentration was measured by the Bradford method using a Bio-Rad protein assay kit with bovine serum albumin as a standard. The histidine tag was removed by cleavage with TEV protease (Life Technologies, Inc.) followed by chromatography on nickel nitrilotriacetic acid spin columns and dialysis. Purified proteins were loaded with copper by incubation in a solution containing 15 M cupric chloride in TBS-DTT (Tris-buffered saline with 25 mM Tris-HCl, 150 mM NaCl, 1 mM DTT, pH 7.5) at 25°C for 4 h. Excess copper was removed by overnight dialysis in TBS-DTT at 4°C. 200 g of HAH1p (normal or mutated, loaded or not loaded with copper) was incubated with 1 ml of TBS-DTT containing either 200 g of amylose resin-bound fusion product of maltose-binding protein-metal-binding domain of the WDp or 100 g of amylose resin-bound maltose-binding protein at 25°C for 3 h with shaking. The resin was washed four times with TBS-DTT, and proteins were eluted with 10 mM maltose in TBS-DTT. Equal aliquots of eluted proteins were analyzed by SDS-PAGE in a Tris-Tricine buffer system. The gel system consisted of a 16.5% separating gel, a 10% spacer gel, and a 4% stacking gel, each made with a 32:1 ratio of acrylamide:bisacrylamide. After gel separation, proteins were stained with Coomassie Brilliant Blue G-250.

RESULTS
Yeast Two-hybrid Analysis of the WDp Metal-binding Domain and HAH1p-To test whether HAH1p transfers copper directly to WDp, we used the yeast two-hybrid assay. The entire 68-amino acid coding region of HAH1 cDNA was fused in frame to the activation domain of GAL4 (in pACT2 vector), whereas a fragment of WD cDNA encoding 623 N-terminal amino acids was fused in frame to the GAL4 DNA-binding domain (in pAS2-1 vector). The WD protein is predicted to possess eight membrane-spanning domains along with four cytoplasmic domains. The six MTCXXC metal-binding motifs are all encoded within the 623-amino acid N-terminal cytoplasmic domain. The two-hybrid plasmids were co-introduced into S. cerevisiae Y187 cells, which contain both the lacZ and his3 reporter genes with upstream GAL4-binding sites. Interaction of the two fusion proteins is necessary to juxtapose the GAL4 DNA-binding and activation domains, which then activate transcription of the reporter genes. Results of these experiments are shown in Fig. 1 (Fig. 1, lane 4).
These results indicate that HAH1p interacts directly with the N-terminal domain of WDp. The strength of signal (2.72 ␤-galactosidase units) was low relative to positive control plasmids (176.5 ␤-galactosidase units; data not shown) but much higher than negative controls (0.01-0.04 ␤-galactosidase units; Fig. 1, lanes 1-3 and 5). We observed the same strength of interaction when the two proteins were cloned in opposite vectors (data not shown). All results obtained from the ␤-galactosidase assay using o-nitrophenyl-␤-D-galactopyranoside as substrate were confirmed both by substituting 5-bromo-4-chloro-3-indolyl ␤-Dgalactopyranoside as substrate and in yeast mating experiments that measure activation of the alternative reporter gene, his3 (data not shown). The presence of 3 mM BCA in the growth medium abolishes the interaction of WDp and HAH1p (Fig. 1,  lane 6), suggesting that copper is required for this interaction.
Characterization of the Six WDp Metal-binding Domains for the Ability to Interact with HAH1p-In the next set of experiments, we sought to determine which MTCXXC motifs, or combination of motifs, are required for interaction with HAH1p. Results of the yeast two-hybrid analysis are shown in Characterization of the HAH1p MTCXXC Motif-In the next set of experiments, we used direct mutagenesis to systematically alter each of the four conserved amino residues in the single MTCXXC motif of the HAH1 protein and then assayed the mutant proteins for copper mediated interaction with WDp. Fig. 3 shows that normal HAH1p (lane 2) and HAH1p with amino acid alterations at MTCXXC (lane 4) and MTCXXC (lane 5) interact with the metal-binding domain of WDp. By contrast, alterations at MTCXXC and MTCXXC residues (lanes 6, 7, 12, and 13) abolish the interaction. We also see in Fig. 3 that omission of the N-terminal (lane 3) or C-terminal (lane 8) 15 amino acid residues of HAH1p likewise disrupts the interaction. In addition to the MTCXXC motif at the N terminus of HAH1p, this protein also harbors three lysine residues at its C terminus, which are highly conserved among eukaryotes. In Fig. 3 we show that mutations Lys 56 3 Gly, Lys 57 3 Gly, and Lys 60 3 Gly do not affect the interaction between HAH1p and WDp as measured by the yeast two-hybrid system (lanes 9 -11).
Assay of HAH1p-Cu-WDp Interaction in Transformed Liver Cells-To better approximate the in vivo environment of a putative copper trafficking interaction between the HAH1 and WD proteins, we next measured the interaction in HepG2 hepatoma cells. Fragments of the WD cDNA encoding either 623 or 499 N-terminal amino acid residues were cloned into the pM vector to produce fusion products of six and four WDp metal-binding domains, respectively, with the GAL4 DNAbinding domain. The entire coding region of the HAH1 gene was cloned into the pVP16 vector to produce a fusion product consisting of the activation domain of the herpes simplex virus VP16 protein and HAH1p. We co-transfected HepG2 cells with pM and pVP16-based plasmids plus a reporter plasmid (pCAT) containing a CAT gene downstream of four GAL4-binding sites and minimum adenovirus E1b promoter. Interestingly, fusion proteins containing all six metal-binding motifs interact minimally, if at all, with HAH1p (Fig. 4, lane 2), whereas protein encoding the four N-terminal most motifs interact more strongly (lanes [3][4][5]. Interaction between WDp and HAH1p is blocked by altering either cysteine group (lanes 6 and 7) or by deleting either the 15 N-terminal most (lane 8) or 15 C-terminal most (lane 9) amino acid residues of HAH1p. Alteration of the methionine or threonine residue of the MTCXXC motif of HAH1p does not disrupt the interaction (lanes 4 and 5).
In Vitro Assay of the Interaction between the WDp Metalbinding Domain and HAH1p-Expression constructs were generated that encode first the entire HAH1 coding segment and second a fusion product consisting of the WDp metalbinding domain and the maltose-binding protein (MBP). E. coli SG20050 cells were transformed with these constructs, and the protein products were purified. HAH1p was subsequently incubated with 15 M CuCl 2 in TBS-DTT, followed by removal of unbound copper by dialysis. The MBP-WDp fusion products (Fig. 5A, lanes 4 and 5) or MBP alone (lanes 2 and 3) were bound to amylose resin and incubated with preparations of HAH1p that were either loaded (lanes 3 and 5) or not loaded (lanes 2 and 4) with copper. Bound protein was then eluted with 10 mM maltose and analyzed by SDS-PAGE. Fig. 5A shows that the metal-binding domain of the WD protein binds specifically to HAH1 protein preloaded with copper (lane 5).
In the next experiment (Fig. 5B)  In this experiment, HAH1p constructs were loaded with copper, whereas resin-bound MBP-Dp fusion protein was copper deficient (see "Experimental Procedures"). The results of the mutagenesis experiments suggest that both Cys residues of the HAH1p MTCXXC motif (copper donor) are required for interaction with WDp (copper acceptor). The inverse experiment with copper-deficient HAH1p mutants and copper loaded resinbound MBP-WDp fusion generated identical results (Fig. 5C), suggesting that both Cys residues of the HAH1p MTCXXC motif (in this case, copper acceptor) are required for interaction with WDp (copper donor). Thus, four Cys residues, two from the copper donor and two from the copper acceptor, are required for the copper-dependent interaction.
Yeast Two-hybrid Analysis of the HAH1p/MNKp Interaction-The MNK and WD proteins are highly homologous, sharing 55% amino acid identity (30). Both proteins contain six metal-binding MTCXXC motifs in their N-terminal portion, as well as a signature inter-membrane CPC motif and all characteristic motifs of P-type ATPases. In this set of experiments, we tested whether MNKp interacts with the HAH1 protein using the yeast two-hybrid assay. Fig. 6 (lane 2) shows that the strength of the HAH1p/MNKp interaction is similar to that of the HAH1p/WDp interaction (Fig. 1, lane 4). Addition of 3 mM BCA to the growth medium abolishes the protein-protein interaction (Fig. 6, lane 3), suggesting that copper is required for this interaction. The pattern of MNKp interaction with altered variants of HAH1p (Fig. 6, lanes 4 -14) is the same as for the WD protein (Fig. 3). Amino acid alterations at MTCXXC (Fig.  6, lane 5 Fig. 7 shows that the metal-binding domain of WDp binds to HAH1 protein that has been preincubated with either Cu 2ϩ or Hg 2ϩ but not with Zn 2ϩ , Cd 2ϩ , Co 2ϩ , Ni 2ϩ , Fe 3ϩ , or Cr 3ϩ .  (23,(31)(32)(33) where the proteins are presumably engaged in the transfer of cytoplasmic copper to copper-requiring proteins.
This study addresses the mechanism of copper trafficking whereby cytoplasmic copper is delivered to the WD and MNK ATPases. The elaboration of copper trafficking in humans has followed from experiments with highly homologous proteins in the yeast, S. cerevisiae. Recent studies in yeast have shown that copper is taken up by cells through the action of CTR1p and then transferred either directly or indirectly to the cytoplasmic chaperone ATX1p. ATX1p is thought to transfer bound copper to CCC2p, which in turn conveys copper to Fet3p. In this study we show that the human homologue of ATX1p, HAH1p, interacts directly with WDp and MNKp using both yeast and mammalian cell assays and in vitro analysis. We further show that two cysteine residues of the HAH1p MTCXXC motif are necessary for the interaction, that a single MTCXXC containing domain of the WD ATPase is capable of interaction with HAH1p, and that copper ion is required for the interaction. We also show that Cu 2ϩ and Hg 2ϩ but not Zn 2ϩ , Cd 2ϩ , Co 2ϩ , Ni 2ϩ , Fe 3ϩ , or Cr 3ϩ can mediate the interaction between the Wilson disease and HAH1 proteins.
Our experiments directly assess the contribution of individual MTCXXC motifs in the interaction between the WD protein and HAH1p. The main conclusion from these studies is that several of the individual WD copper-binding motifs are capable of independent interaction with the HAH1 protein. Metal-binding domains 1, 2, and 3 are each sufficient to interact with HAH1p, and various combinations of domains 1-4 appear to interact with roughly equal efficiency. Interestingly, the last two WD metal-binding domains (5 and 6) fail to interact with HAH1p, and the truncated 1-4 domain fragment interacts more strongly than the intact 1-6 domain fragment. The simplest explanation for these results is: first, that the hybrid protein containing WD metal-binding domains 5 and 6 is folded such that the MTCXXC motifs are not exposed on the globule surface and thus are inaccessible for interaction with HAH1p, and second, that under our experimental conditions, the 5-6 fragment masks MTCXXC motifs from the 1-4 fragment. The results indicate that the fifth and sixth copper-binding motifs do not participate directly in the exchange of copper with HAH1p and thus suggest a different function for these motifs.
The interaction of apo-HAH1p with copper-loaded WDp suggests that copper transfer between these proteins is a reversible process. We think that at equilibrium, the HAH1p-Cu-WDp complex can dissociate with an equal chance of forming either HAH1p-Cu ϩWDp or HAH1p ϩ Cu-WDp. It follows that removal of the Cu-WDp moiety would shift the equilibrium toward the formation of Cu-WDp, whereas removal of HAH1p-Cu would shift the equilibrium toward the transfer of copper from WDp to HAH1p. In vivo, removal of the Cu:WDp complex might be achieved by the transport of copper across the membrane in conjunction with ATP hydrolysis. Thus, our data raise the possibility that HAH1p may function in some circumstances to remove copper from the WD and MNK ATPases, presumably in response to intracellular cues.
In our yeast two-hybrid experiments, HAH1p mutations Lys 56 3 Gly, Lys 57 3 Gly, and Lys 60 3 Gly did not affect interaction between the copper chaperone and either WDp or MNKp. Recent yeast two-hybrid analyses of the yeast HAH1 homologue, ATX1, showed that mutation of the corresponding residues, Lys 65 3 Glu and Lys 61 3 Glu ϩ Lys 62 3 Glu, as well as other compound lysine to glutamate mutations, abolish interaction between the copper chaperon and CCC2p, the WD/ MNK homologue (35). As the x-ray crystal structures for ATX1p (36) and the fourth metal-binding domain of MNKp (MNK4) have recently been reported (37), we attempted to reconcile these differences by modeling the WDp/MNKp -HAH1p interaction. As shown in Fig. 8, our model indicates that the HAH1p/MNK4 interaction is stabilized by two salt bridges formed between the positively charged HAH1p lysine residues Lsy 57 and Lys 25 , and the negatively charged MNK4 residues Glu 22 and Asp 63 , respectively. According to this model, a Lys 57 3 Gly mutation would remove one of the stabilizing salt bridges, whereas a Lys 57 3 Glu mutation would both disrupt the salt bridge and introduce a repulsive force that disrupts the protein-protein interaction. The Lys 56 residue is nearly completely exposed to solvent and thus is unlikely to affect protein-protein interaction. Lys 60 is close to the copperbinding site as predicted from the ATX1p x-ray structure (36), where it may provide an electrostatic potential gradient that favors the movement of the positively charged copper ion from HAH1p to MNK4. The Lys 60 3 Glu mutation would reverse the gradient and thus trap the copper at the HAH1p copper-binding motif, whereas a Lys 60 3 Gly mutation would predictably have less effect.
Proteins containing the MTCXXC motif are present in such evolutionary distant organisms as bacteria and man. The motif consisting of two cysteine residues separated by any two amino acid residues is absolutely conserved among these proteins from diverse phylogenetic origins. By contrast, the threonine residue is often substituted, and less frequently, MTCXXC containing proteins (for example the CopP protein of Helicobacter pylori) lack a methionine residue at this site. A glycine residue often precedes the MTCXXC motif. Functionally characterized MTCXXC proteins are involved in the transport of either copper or mercury, where the cysteine residues are directly involved in the binding of the metal ions. In all three of our test systems mutation of either Met or Thr did not affect interaction between HAH1p and the metalbinding domain of the WD protein, whereas alteration of either Cys residues abolished the interaction. Interestingly, in the work of Hung et al. (34), mutations in both Cys residues of the MTCXXC motif of ATX1p were required to eliminate copper incorporation into Fet3p. It is possible that the Met residue can provide a second sulfur atom for bi-coordination of copper in ATX1 proteins containing a single Cys in the metal-binding motif.
It is noteworthy that the copper trafficking mechanism first proposed by Lin and Culotta (15) and Pufal et al. (17) from yeast studies and supported by this study of human proteins is closely analogous to the mechanism proposed for mercury detoxification in bacteria (27,28). In bacteria, the periplasmic protein, MerP, binds Hg 2ϩ with its MTCXXC motif and then relays it to the integral membrane proteins MerT and/or MerC (both contain a pair of cysteine residues believed to be involved in Hg 2ϩ binding) whose function is to convey Hg 2ϩ to the HAH1p was preincubated with different metal ions, and then its interaction with the amylose resin bound fusion construct maltose-binding protein-metal-binding domain of WDp was studied as described under "Experimental Procedures." Proteins bound to the resin were eluted with maltose and analyzed by SDS-PAGE. cytoplasmic mercury reductase, MerA. MerA catalyzes the reduction of Hg 2ϩ to the volatile and less toxic Hg 0 ion. There is an obvious analogy between the MerP 3 MerT 3 MerA pathway and the HAH1p 3 WDp 3 ceruloplasmin pathway (or ATX1p 3 CCC2p 3 Fet3p pathway in S. cerevisiae). Because unbound copper is highly toxic, it is reasonable to propose a bucket brigade like mechanism for copper trafficking.
We do not yet know whether other copper chaperons (39) mediate the transfer of cytoplasmic copper to the MNK and WD ATPases. We do believe that the copper chaperon proteins can, to some extent, exchange copper among themselves. The cytoplasmic copper chaperon Cox 17p is believed to traffic copper to the mitochondria for insertion into cytochrome c oxidase, the terminal oxidase of the respiratory chain (20). Our yeast twohybrid experiments indicate that HAH1p interacts with Cox17p and that Cox17p, but not HAH1p, interacts with a Cox2p fragment that is presumably located in the mitochondrial intramembrane space (data not shown). Likewise, it is not known how the MNK and WD ATPases transfer copper from the metal-binding domain across the membrane or what role the multiple metal-binding domains play in this regard. The multiple metal-binding domains might serve as traps for copper ions that effectively creates a high local concentration of copper at the intramembrane metal-binding site. Another possibility is that copper binding induces conformational changes in the metal-binding domain that subsequently renders the intramembrane metal-binding site accessible to the cytoplasmic surface. Functional studies will be required to address these questions.