The Distinct Functional Properties of the Nucleotide-binding Domain of ATP7B, the Human Copper-transporting ATPase* ANALYSIS OF THE WILSON DISEASE MUTATIONS

Copper transport by the P 1 -ATPase ATP7B, or Wilson disease protein (WNDP), 1 is essential for human metab-olism. Perturbation of WNDP function causes intracel-lular copper accumulation and severe pathology, known as Wilson disease (WD). Several WD mutations are clus-tered within the WNDP nucleotide-binding domain (N-domain), where they are predicted to disrupt ATP binding. The mechanism by which the N-domain coordinates ATP is presently unknown, because residues important for nucleotide binding in the better characterized P 2 - ATPases are not conserved within the P 1 -ATPase subfamily. To gain insight into nucleotide binding under normal and disease conditions, we generated the recombinant WNDP N-domain and several WD mutants. Using isothermal titration calorimetry, we demonstrate that the N-domain binds ATP in a Mg 2 (cid:1) -independent manner with a relatively high affinity of 75 (cid:2) M , compared with millimolar affinities observed for the P 2 -ATPase N-do- mains. The WNDP N-domain shows minimal discrimina-tion between ATP, ADP, and AMP, yet discriminates well between ATP and GTP. -thiogalactopyranoside chitin resin-bound fusion chitin M resin intein cleavage m M m M Tris Cleavage buffer intein-mediated excision of the N-domain from the fusion, allowing elution of non-tagged N-domain from the resin. The purified WNDP N-domains contained four non-native amino-terminal residues (AGHM) derived from the pTYB12 vector; the MNKP N-domain contained no non-native amino acids. Expression and purification of the MNKP N-domain was carried out as described for the WNDP N-domain, except that the with isopropyl- (cid:2) - D -thioga-lactopyranoside was carried out at 16 to improve solubility. Typical yields of wt and mutant N-domains from 2–6 mg protein per liter of cell culture. and mutant N-domains Molecular Modeling and Molecular Dynamics Simulations— A spa- tial model of the ATP-binding domain of WNDP (residues was built via homology modeling based on the x-ray structure of the nucleotide-binding domain (residues Ala-320–Lys-758) of Ca 2 (cid:1) -ATPase in the E1 (“open”) state (PDB entry 1EUL (13)). The details of the modeling experiments and validation criteria were described previously (24). In the current work, minor conformational changes were intro- duced into the loop 1062–1071 of the model using the following proce-dure. First, twenty models with different conformations of this loop were generated. Two models in which the side chains of Glu-1064 and His-1069 point toward the adenine binding cleft in the N-domain were selected for future studies. Both models were subjected to 5-ns molecular dynamics simulations in explicit water. The conformers extracted from the equilibrium parts of molecular dynamics trajectories were then employed in docking simulations with the ATP molecule as de- scribed previously (24). The molecular surfaces were mapped according to their hydrophobic properties using the molecular hydrophobicity potential approach (25).

The Wilson disease protein (WNDP) is a key regulator of copper homeostasis in a number of tissues, particularly the liver, brain, and kidneys (1,2). WNDP transports copper from cytosol across cell membranes, using the energy of ATP hydrolysis. Under basal conditions, WNDP delivers copper to enzymes within the secretory pathway; it is also essential for cellular copper excretion when copper concentrations are elevated (1)(2)(3). Mutations in WNDP result in marked accumulation of copper in the cytosol and a severe hepatoneurological disorder known as Wilson disease (WD) (4,5). The clinical manifestations of WD are diverse (6); however, the specific contributions of various WD mutations to phenotypic diversity remain poorly understood (7). Elucidating the consequences of mutations on WNDP structure and function is the first step toward a better understanding of molecular mechanisms underlying WD. In addition, an intriguing connection has recently been made between overexpression of WNDP and increased resistance of cells to the anticancer drug cisplatin (8 -10). These findings point to a role for WNDP as a potential pharmacological target and further emphasize the need for a better understanding of the protein's structure, function, and regulation. At present, such structural and biochemical information on WNDP is very limited.
WNDP belongs to the large family of the P-type ATPases and displays the key catalytic properties expected for the members of this family. In particular, WNDP hydrolyzes ATP to form a transient phosphorylated intermediate at the invariant aspartate located in the DKTG sequence, a signature motif of the P-type ATPase (Fig. 1). Copper, the transported ion, markedly stimulates the reaction (11). The first crystal structure of a P-type ATPase, Ca 2ϩ -ATPase of sarcoplasmic reticulum, was recently solved (12,13). The structure yielded important information about the organization and conformational flexibility of this class of transporters. Ca 2ϩ -ATPase was shown to be composed of several functional domains: the ATP-binding domain; the transmembrane domain, containing sites for transported ions; and the actuator domain, which is critical for phosphatase activity and conformational transitions during the catalytic cycle (14,15). The ATP-binding domain of the P-type ATPases was shown to consist of two domains: the phosphorylation domain (the P-domain) containing the DKTG sequence and the nucleotide-binding domain (the N-domain), which contributes to ATP coordination.
The P-type ATPases are divided into five subfamilies based on their ion specificity and structural characteristics (16). WNDP is a member of the P 1B -ATPase subfamily (16,17). The P 1 -ATPases share very limited sequence homology with Ca 2ϩ -ATPase or other P 2 -ATPases. In fact, residues strictly conserved among all P-type ATPases are almost exclusively lim-ited to the P-domain and the actuator domain (Fig. 1A). The P-domain and actuator domain play essential roles in catalysis; thus, it is not surprising that these two domains contain the sequence motifs found in all P-type ATPases. The N-domain is equally important, because it is involved in nucleotide binding. Therefore, it is intriguing that the amino acid residues known to participate in ATP coordination in the N-domain of Ca 2ϩ -ATPase and other P 2 -ATPases are not conserved in the structure of the P 1 -ATPases. This observation suggests that the coordination environment of the nucleotide in these two P-type ATPase subfamilies could be quite different.
Although there is little sequence similarity between the Ndomains of the P 1 -and P 2 -ATPases, the sequence of the Ndomain is well conserved within the P 1 -ATPase subgroup. A number of the residues are highly conserved or invariant (Fig.  1B), suggesting their potentially important roles in the structure or function of the P 1 -ATPase N-domains. In the WNDP sequence, these residues include Glu-1064, Ser-1067, His-1069, Pro-1070, Gly-1101, and Gly-1103. It is noteworthy that invariant His-1069 is the site of the most frequent disease-causing mutation, an observation that further emphasizes the functional significance of this residue.
Previous studies characterized the effect of mutations of His-1069, or equivalent histidines, on the functional activity of WNDP and several homologous P 1 -ATPases (18 -22). The results confirmed the important role of this residue for the P-type ATPase function and suggested that His-1069 may participate in the positioning of ATP within the catalytic site. However, direct evidence for the role of His-1069 in ATP binding has been lacking. Likewise, a number of other WD-causing mutations have been identified in the N-domain of WNDP, but their effects on nucleotide binding have not been characterized.
Experiments using site-directed mutagenesis of the fulllength P 1 -ATPases demonstrate that the dissection of the functional role of various residues could be impeded by additional effects of mutations on enzyme conformation (19). The specific role of the amino acid residues in nucleotide coordination is particularly difficult to assess, because the nucleotide binding characteristics of the P 1 -type ATPase mutants have been analyzed indirectly by monitoring their ability to use ATP to form a phosphorylated intermediate or to transport copper. Phosphorylation and transport are several steps removed from the initial nucleotide-binding event, making conclusions regarding nucleotide binding tenuous. Therefore, to directly investigate FIG. 1. Schematic of WNDP structural organization and sequence alignment of the P 1 -ATPase N-domains. A, TGEA, DKTG, TGDN, and GDGxND represent sequence motifs conserved in all P-type ATPases. The N-domain is independently folded and inserted into the P-domain; arrows indicate the domain's boundaries. Letters CxxC indicate six copper-binding repeats in the N-terminal domain. Glu-1064, His-1069, Cys-1104, and Arg-1151 (in bold) show relative positions of the residues mutagenized in this work. B, in the alignment, the asterisks indicate strictly conserved residues, dots indicate conserved residues, and divergent regions are in parentheses. The alignment was generated using ClustalW (www.ebi.ac.uk/clustalw/). Protein data base accession numbers are ATP7B_HUMAN, WNDP, P35670; ATP7A_HUMAN, MNKP, Q04656; ATZN_ECOLI, ZntA, P37617; ATU2_YEAST, CCC2, P38995; and CADA_STAUU, CadA, P20021. the role of several residues in nucleotide binding by WNDP, we generated the recombinant WNDP N-domain and N-domain variants carrying known disease mutations, including H1069Q. In addition, we generated the recombinant N-domain of the Menkes disease protein (MNKP), the human coppertransporting ATPase ATP7A. MNKP is homologous to WNDP and is another member of the P 1 -ATPase subfamily. Comparison of these two proteins helped us to dissect the commonalities in their nucleotide-binding properties. Using direct ligand binding measurements, we demonstrate that 1) the N-domains of WNDP and MNKP contain a single nucleotide-binding site, 2) the properties of these sites are distinct from those of the P 2 -ATPases, and 3) the invariant residues Glu-1064 and His-1069 of WNDP are important for nucleotide coordination, whereas the less conserved Cys-1104 and Arg-1151 residues are not essential for nucleotide-binding.

MATERIALS AND METHODS
Expression Constructs for the Wild-type and Mutant N-domains of WNDP-To generate the expression constructs for wt N-domain of WNDP (amino acid residues Val-1036 -Asp-1196 of the full-length protein), the corresponding cDNA region was amplified via PCR using pCDNA3.1(ϩ)-WNDP plasmid as a template and the following primers: WT.ND-fwd, 5Ј-ACATATGGTCCCCAGGGTCATG-3Ј, and WT.ND-rev, 5Ј-AGAATTC TTAGTCTGCGATTGCGATC-3Ј. The PCR product, with 5Ј-NdeI and 3Ј-EcoRI endonuclease restriction sites flanking the coding sequence, was subcloned into the pCRII-Blunt TOPO vector (Invitrogen). The resulting pCRII-WT.ND plasmid was then used as a template for the generation of E1064A, H1069Q, C1104F, C1104A, and R1151H mutants by site-directed mutagenesis.
The N-domain of MNKP-The expression construct for MNKP Ndomain (residues Thr-1048 -Asp-1230 of the full-length protein) was generated by PCR of the corresponding cDNA region of ATP7A using the following primers: MNKP-ND fwd, 5Ј-ATGAATGCTACCATTA CT-CACGGAAC-3Ј, and MNKP-ND rev, 5Ј-ATGAATTCTTAGTCTGCAAT-GGCTATC-3Ј. The PCR product, containing 5Ј-BsmI and 3Ј-EcoRI endonuclease restriction sites flanking the coding sequence, was then cloned into the pTYB12 expression vector. The nucleotide sequence of the pTYB12-MNKP.ND was verified by automated DNA sequencing.
Expression and Purification of the Recombinant N-domains-All Ndomains were expressed as fusion proteins with a chitin-binding domain and an intein protein. In brief, Escherichia coli BL21 (DE3) cells were grown in ampicillin-supplemented (100 g/ml) Luria-Bertani liquid media at 37°C until an A 600 of ϳ0.6 was reached. The expression was then induced by addition of isopropyl-␤-D-thiogalactopyranoside to a final concentration of 0.5 mM, with shaking at 250 rpm for ϳ12 h at 23°C. After harvest by centrifugation, cell pellets were re-suspended in 50 mM Tris, pH 8.2, and 500 mM NaCl, with complete EDTA-free protease inhibitor mixture (Roche) and disrupted by passing the suspension through a French press twice at 15,000 p.s.i. The lysate was centrifuged for 45 min at 15,000 rpm, and the soluble fraction was passed over a chitin resin. Purified resin-bound fusion protein was obtained after washing the chitin resin (New England Biolabs) with approximately 30 column volumes of 50 mM Tris, pH 8.2, and 500 mM NaCl. After the wash, the resin was equilibrated with the intein cleavage buffer containing 50 mM DTT, 50 mM Tris pH 8.2, and 150 mM NaCl. Cleavage buffer promotes intein-mediated excision of the N-domain from the fusion, allowing elution of non-tagged N-domain from the resin. The purified WNDP N-domains contained four non-native aminoterminal residues (AGHM) derived from the pTYB12 vector; the MNKP N-domain contained no non-native amino acids. Expression and purification of the MNKP N-domain was carried out as described for the WNDP N-domain, except that the induction with isopropyl-␤-D-thiogalactopyranoside was carried out at 16°C to improve solubility. Typical yields of wt and mutant N-domains ranged from 2-6 mg protein per liter of cell culture.
CD Spectroscopy-Purified wt and mutant N-domains were dialyzed extensively against 50 mM NaH 2 PO 4 , pH 7.0, buffer and concentrated to 0.4 mg/ml using Amicon Ultra PL-10 centrifugal filters (Millipore) with a 10-kDa molecular mass cutoff. Concentrated samples were then centrifuged at 90,000 ϫ g for 45 min to remove insoluble particles. Protein concentrations were initially determined spectrophotometrically using a theoretical extinction coefficient of 8850 M Ϫ1 cm Ϫ1 at 280 nm; subsequent amino acid analysis confirmed the spectrophotometrically determined concentrations. CD spectra were measured using an AVIV CD model 215 spectrometer; deconvolution of the spectra was carried out with the program CDDN1 (23). The CD measurements were performed for three independent protein preparations of each mutant, and the resulting secondary structure values were averaged.
Thermal denaturation was used to further compare folding of the wt and mutant N-domains. Temperature-dependent structural transitions, measured as molar ellipticity changes at 222 nm, were monitored from 25 to 80°C using CD spectroscopy. Spectra were collected with the AVIV CD 215 spectrometer using a 0.1-cm rectangular CD cell. Molar ellipticity was monitored in 0.5-nm increments with 30-s equilibrations between measurements.
Nucleotide Binding Measurements-Isothermal titration calorimetry (ITC) was used to analyze the nucleotide-binding properties of wt WNDP and MNKP N-domains and the E1064A, H1069Q, C1104A, and R1151H N-domain mutants. Purified proteins were dialyzed extensively against 50 mM NaH 2 PO 4 , pH 7.0, buffer and concentrated to 116 M (ϳ2 mg/ml). Nucleotide solutions of 4 mM ATP disodium salt (Sigma), ATP monosodium salt (Acros Organics), AMP (Acros Organics), or GTP dilithium salt (Alexis Biochemicals) were prepared in the dialysis buffer immediately before each titration. For titrations with ATP-Mg 2ϩ complex, 12 mM MgCl 2 (Sigma) was included in the 4 mM ATP disodium salt solution.
To ensure that the ITC data do not include enthalpy changes caused by hydrolysis of nucleotides by the N-domain, 1 mM ATP in 50 mM MES buffer, pH 6.0, was incubated with or without 1 mM N-domain at room temperature for up to 6 h. The amount of free P i resulting from ATP hydrolysis (10 M P i , 1% of total ATP) was determined by EnzChek phosphate assay kit (Molecular Probes) and was found to be identical in the presence and absence of the N-domain.
ITC experiments were performed with the VP-ITC titration calorimeter (Microcal, Inc.). Protein and nucleotide solutions were degassed by vacuum aspiration for 5 min before loading the samples into the ITC cell and syringe, respectively. All titrations were carried out at 25°C, with a stirring speed of 300 r.p.m. and a 180-s duration between each 8-l injection; thermal power was monitored every 8 s. Wild-type and mutant N-domains were titrated to saturation, where possible, with a 4 mM stock solution of nucleotide. Parallel experiments were performed by injecting nucleotide into buffer or buffer into N-domain to determine the heats of dilution. The heats of dilution were then subtracted from their respective N-domain nucleotide titrations before data analysis. Thermogram analysis was performed with the Origin 5.0 data analysis software provided with the VP-ITC instrument (OriginLab Corp). For each protein-nucleotide interaction, at least three titrations were performed on independently prepared N-domains. Titration thermograms were analyzed independently, and the obtained K d values were averaged.
To characterize binding of ATP using changes in the intrinsic tryptophan fluorescence, 11.5 M WNDP N-domain in 50 mM NaH 2 PO 4 , pH 7.0, was titrated with increasing concentrations of ATP (0 -3 mM) or the same volume of buffer. Using a model PTI-QM1 (Photon Technology International) fluorimeter, protein sample was excited at 295 nm (2-nm slit width), and fluorescence was monitored at an emission wavelength of 320 nm (1-nm slit width), which corresponds to the maximum emission wavelength for the N-domain. The final increase in protein sample volume caused by ATP addition did not exceed 1% and was accounted for in final calculation. Data were plotted as ratio of F/F 0 versus concentration of ATP.
Molecular Modeling and Molecular Dynamics Simulations-A spatial model of the ATP-binding domain of WNDP (residues M996-R1322) was built via homology modeling based on the x-ray structure of the nucleotide-binding domain (residues Ala-320 -Lys-758) of Ca 2ϩ -ATPase in the E1 ("open") state (PDB entry 1EUL (13)). The details of the modeling experiments and validation criteria were described previously (24). In the current work, minor conformational changes were introduced into the loop 1062-1071 of the model using the following procedure. First, twenty models with different conformations of this loop were generated. Two models in which the side chains of Glu-1064 and His-1069 point toward the adenine binding cleft in the N-domain were selected for future studies. Both models were subjected to 5-ns molecular dynamics simulations in explicit water. The conformers extracted from the equilibrium parts of molecular dynamics trajectories were then employed in docking simulations with the ATP molecule as described previously (24). The molecular surfaces were mapped according to their hydrophobic properties using the molecular hydrophobicity potential approach (25).

RESULTS
The recombinant wt and mutant N-domains were expressed in E. coli and isolated after affinity chromatography on chitin beads and intein-mediated self-cleavage, as described under "Experimental Procedures." The expression levels, protein yields, and purities were similar for all N-domains.
Folding and Nucleotide-binding Properties of the wt WNDP N-domain-Previous NMR studies of the N-domain of Na,K-ATPase, a P 2 -ATPase, revealed that the domain is folded and can bind ATP (26). However, the apparent affinity of the Ndomain for the nucleotides was extremely low (5-20 mM) compared with the high affinity of the full-length protein (0.3-1 M). Likewise, the analysis of ATP binding by the proteolytic fragment of Ca 2ϩ -ATPase, containing the N-domain, yielded a high apparent K d value of ϳ0.7 mM (27), whereas characterization of the recombinant N-domain of Ca 2ϩ -ATPase produced a K d value of ϳ2.4 mM (28). These results suggested that the formation of the high affinity site for ATP in P 2 -ATPases is a result of interaction between the N-domain and other regions of the proteins. Because the sequences of the P 1 -and P 2 -ATPase N-domains are dissimilar, we were interested in testing whether the isolated N-domain of WNDP, a P 1 -ATPase, can alone form a high affinity site for the nucleotide.
Before nucleotide binding measurements, the folding of the recombinant wt N-domain of WNDP was examined. Analysis of the secondary structure of the N-domain using CD spectroscopy (Fig. 2) revealed that the N-domain of WNDP has the following secondary structure composition: ϳ26% ␣-helix, ϳ18.5% ␤-turn, and ϳ22% ␤-sheet (Table I). Furthermore, thermal denaturation experiments demonstrated a sharp transition curve typical of a well folded soluble protein, with a T m of ϳ52°C (Fig. 2, inset). Finally, the N-domain was found to be resistant to treatment with mild concentrations of trypsin (data not shown). All these results indicated that the recombinant N-domain of WNDP is well folded.
The ability of the N-domain to bind nucleotides was examined using ITC. The N-domain does not contain the catalytic aspartate and is not expected to hydrolyze ATP. To verify this prediction and ensure that ITC measures only nucleotide-binding events and not hydrolysis of the nucleotides, we determined hydrolytic activity of the N-domain. The ATP hydrolysis in the presence of the N-domain did not exceed the spontaneous rate of ATP decay in the buffer even after prolonged (6 h) incubation. Thus, the enthalpy changes observed in the ITC experiments are caused by binding of the nucleotides to the N-domain.
A representative thermogram for the calorimetric titration of wt N-domain with ATP is presented in Fig. 3A. The exothermic evolution of heat upon ATP injections, shown at the top, illustrates saturable nucleotide binding by the N-domain. Calculation of the enthalpy changes at various molar ratios of ATP to the N-domain (Fig. 3A, bottom) revealed that the best fit for the data corresponds to the presence of one ligand-binding site in the N-domain. In contrast, titration with GTP demonstrated no significant protein-nucleotide interactions. Enthalpy changes in this case were indistinguishable from the results of the nucleotide titration into buffer. Thus, the N-domain of ATP7B discriminates extremely well between adenosine and guanosine moieties, indicating selectivity for the nucleotides.
Calculations of the apparent affinity for ATP yielded a K d of 75.30 Ϯ 3.62 M, an unexpectedly low value, comparing with 0.7-5 mM K d values reported earlier for the N-domains of the P 2 -ATPases (26 -28). The ITC result was confirmed by measuring changes in the intrinsic tryptophan fluorescence of the N-domain upon addition of ATP (Fig. 3B), which yielded similar K d value (110 Ϯ 15 M). This relatively high affinity enabled us to elucidate the nucleotide binding characteristics of the N-domain in more detail. Based on known properties of the P 2 -ATPases, we expected the following relative affinities of the N-domain for the nucleotides: ATP Ͼ ADP Ͼ Ͼ AMP (see, for example, Ref. 29). However, the experiments revealed that the nucleotide specificity of the WNDP N-domain did not follow this trend. As shown in Fig. 4, there is very little difference in the affinity of the N-domain for ATP, ADP, and AMP (K d values summarized in Table II). In fact, the N-domain binds ATP with a slightly lower affinity than ADP or AMP.
To test whether the observed nucleotide-binding characteristics are unique for WNDP or representative of the P 1 -ATPases, we expressed and purified the N-domain of MNKP, another copper-transporting ATPase (Fig. 4B, inset). The MNKP N-domain is 50% identical to the WNDP domain; however, it is larger (20 kDa versus 17 kDa) and contains a unique sequence insert (3). The ITC experiments demonstrated that the nucleotide-binding properties of the MNKP N-domain are very similar to those of WNDP (Fig. 4). In particular, MNKP binds ATP and ADP with high affinity (Table I) and discriminates poorly between these nucleotides. It is interesting that the K d value for ATP is higher than for ADP (83.00 Ϯ 7.10 M and 44.50 Ϯ 0.70 M, respectively), similar to what was observed for the N-domain of WNDP. The N-domain of MNKP did not interact with GTP, confirming its selectivity toward the nucleotides containing the adenosine moiety.
Magnesium plays a critical role in the catalytic cycle of the P-type ATPase. The presence of Mg 2ϩ is required for the hydrolysis of ATP by the full-length WNDP (our data), suggesting that ATP-Mg 2ϩ is a substrate for this reaction. Although the N-domain lacks the catalytic aspartate required for the hydrolysis of ATP, it is unknown whether Mg 2ϩ plays a role in the docking of ATP to the N-domain of WNDP. To address this issue, we examined the effect of Mg 2ϩ on interaction between wt WNDP N-domain and ATP. The results shown in Fig. 4A indicate that Mg 2ϩ increases the affinity of the N-domain for ATP, but only slightly (Table II). In addition, no interactions were observed when the N-domain was titrated with magnesium in the absence of ATP. We conclude that the nucleotide binding by the N-domain is a magnesium-independent event.
Residues Glu-1064 and His-1069 Play Important Roles in Nucleotide Binding-Numerous disease-causing mutations have been identified in the N-domain of WNDP. The E1064A and H1069Q mutations affect amino acid residues that are highly conserved in P 1 -ATPases, whereas C1104F and R1151H substitutions alter less conserved residues (Fig. 1B). Therefore, we hypothesized that Glu-1064 and His-1069 could be essential for nucleotide coordination, whereas the roles of Cys-1104 and Arg-1151 could be more structural.
To examine the involvement of His-1069 and Glu-1064 in ATP binding, we introduced the above mutations into the Ndomain and tested the effects of the mutations on folding and function. The secondary structure of the H1069Q and E1064A N-domains seemed unaltered by these mutations, as evidenced by essentially identical CD data for the mutants and wt Ndomain (Fig. 5A, Table I). However, both mutations had a marked effect on nucleotide binding. As shown in Fig. 5B and Table II, E1064A lost the ability to bind ATP entirely, whereas the H1069Q mutant showed minimal binding, which did not reach saturation under the final conditions of our assay (i.e. at 600 M ATP).
Efforts were made to evaluate the role of Glu-1064 in nucleotide binding in more detail. We generated the N-domain mutant with a more conservative E1064D substitution, preserving the charge of the residue at this position. We were surprised to find that the E1064D mutation had a significant negative effect on protein folding, as evidenced by CD spectroscopy and thermal stability experiments (data not shown). The nucleotide-binding properties of this mis-folded mutant were not investigated.
Cys-1104 Is Not Required for Nucleotide Binding-The residue corresponding to Cys-1104 in WNDP is not conserved and can be replaced by alanine or glycine in other P 1 -ATPases (Fig.  1B). At the same time, in WNDP, the mutation of Cys-1104 to Phe results in a WD phenotype. To better understand the role of Cys-1104 in WNDP folding and function we generated the C1104F mutant of the N-domain and characterized its biochemical properties. Unlike H1069Q and E1064A, the C1104F substitution had a marked effect on folding of the N-domain (Fig. 6). Phenylalanine is a bulky residue, and it is not surprising that the Cys-to-Phe mutation alters the N-domain structure. Therefore, to further elucidate the role of Cys-1104 in protein folding and nucleotide binding, we substituted cysteine with the less bulky alanine.
In sharp contrast to the C1104F mutant, the secondary structure of the C1104A N-domain was very similar to that of the wt N-domain (Fig. 6A). This result suggested that the amino acid residues with a small side-chain are well tolerated at the 1104 position. Thermal denaturation experiments performed on the wt, C1104F, and C1104A N-domains confirmed this conclusion. As shown in Fig. 6B, the C1104A mutant exhibits a temperature-induced structural transition very similar to that of the wt protein. In contrast, the C1104F mutant did not undergo such a transition. Finally, the role of Cys-1104 in nucleotide binding was tested by measuring the ability of the C1104A mutant to bind ATP. As shown in Table II The Effect of the R1151H Substitution on Nucleotide Binding-Arg-1151 is another non-conserved residue, but substitution with histidine at this position results in a WD phenotype (30). The CD spectrum of the R1151H mutant is indistinguishable from the wt N-domain (Table I), suggesting that the effect of this mutation on the secondary structure of the protein is insignificant. However, the nucleotide-binding properties of the N-domain are altered by the R1151H substitution (Table II). The K d value for ATP is increased by 25-30%, suggesting that the affinity for the nucleotide was diminished but not greatly affected. Similar effects were observed for ADP and AMP. It is noteworthy that binding of AMP was somewhat less affected, suggesting that the only effect of the mutation could be in the vicinity of ATP's ␤and ␥-phosphates.
Molecular Modeling-To better understand the basis for the higher nucleotide-binding affinity of the WNDP N-domain and visualize the location of the mutant residues with respect to ATP, we used a molecular modeling approach. This approach included fine-tuning of a recently generated 3D model of the WNDP ATP binding domain (24), molecular dynamics simulations, and ATP docking experiments. The modeling studies predict that all amino acid residues characterized in our work are situated near the nucleotide-binding cleft (Fig. 7). However, only Glu-1064 and His-1069 are in immediate proximity to the adenosine moiety of ATP. This prediction is consistent with our experimental data demonstrating a drastic effect of the E1064A and H1069Q substitutions on affinity of the N-domain for ATP. In the isolated N-domain, all mutated residues are also fairly exposed. This observation may explain why most of the mutations, except for the bulky C1104F substitution, have little effect on folding of the N-domain.
Comparison of the WNDP N-domain model (Fig. 7) with the high resolution structure of Ca 2ϩ -ATPase revealed that, upon ATP docking, the changes in accessible surface area are ϳ100 -150 Å 2 for Ca 2ϩ -ATPase and ϳ250 Å 2 for WNDP, although the spatial dimensions of the binding sites are quite similar. Thus, it seems that ATP is buried deeper within the WNDP's nucleotide binding cleft relative to Ca 2ϩ -ATPase. Furthermore, the hydrophobic surface of the WNDP N-domain exhibits greater complementarity with the nucleotide's adenine ring; i.e. the hydrophobic and hydrophilic atoms of the protein cover a larger fraction of the hydrophobic and hydrophilic surfaces of ATP, respectively. Analysis of several protein-ATP complexes with known high resolution structure suggests that the higher degree of complementarity correlates well with the higher affinity of binding. 2 Taken together, these results are consistent with the idea that WNDP has a higher nucleotide binding affinity than the P 2 -ATPases.

DISCUSSION
Within the large family of P-type ATPases, the P 1 -and P 2 -ATPases have distinct ion-specificities, characteristic transmembrane topologies, and unique sequence motifs (17,31). The overall sequence similarity between the mammalian coppertransporting ATPases (P 1 -ATPases) and Ca 2ϩ -ATPase or Na,K-ATPase (P 2 -ATPases) is less than 5%. Nevertheless, these proteins perform the same general function: coupling ATP binding and hydrolysis with the transport of ions across membranes. The marked sequence dissimilarity between the P 1 -and P 2 -ATPases may indicate that very few amino acid residues are essential for ATP binding and catalysis. On the other hand, similar steps in the enzymatic cycle of the P 1 -and P 2 -ATPases could be achieved via different structural means. Our characterization of the N-domains of two human coppertransporting P 1 -ATPases, WNDP and MNKP, supports this latter hypothesis and identifies several key differences between the nucleotide-binding domains of the P 1 -and P 2 -ATPases.  Table II. Direct ligand binding measurements performed in this work indicate that the isolated N-domains of P 1 -ATPases bind ATP with much higher affinity than the P 2 -ATPase N-domains (70 -80 M versus 0.7-5 mM). In contrast, the full-length P 1and P 2 -ATPases have similarly high apparent affinities for ATP (0.3-1 M). These findings suggest that, compared with the P 2 -ATPases, the formation of the nucleotide-binding site by WNDP and MNKP is significantly less dependent on interactions of the N-domain with other portions of the protein. Although interactions with the P-domain and/or actuator domain are still needed to obtain the low micromolar affinities for ATP observed during catalytic phosphorylation of WNDP, it seems clear that the nucleotide-binding site is largely pre-organized within the N-domain.
Our modeling studies suggest that favorable hydrophobic contacts and H-bonds between the adenosine moiety and the residues forming the binding cleft are likely to play a major role in higher affinity of the P 1 -ATPase N-domain for the nucleotides. The significant hydrophobicity and the size of the binding cleft also help to explain the lack of interactions between the N-domain of WNDP and GTP, because in the WNDP N-domain, the additional amino group of GTP would be placed in an unfavorable environment with sterical hindrances.
Another interesting difference between the P 1 -and P 2 -Ndomains relates to their selectivity for adenosine-based nucleotides. The Na,K-ATPase N-domain binds ATP and ADP with K d values of 5.1 Ϯ 0.4 and 24.2 Ϯ 5.3 mM, respectively (26). Neither the full-length Na,K-ATPase nor the ATP-binding domain bind AMP with appreciable affinity (29). In contrast, the N-domains of WNDP and MNKP were found to bind ATP, ADP, and AMP, with very similar K d values. Clearly, neither the ␤-

Nucleotide-binding Domain of ATP7B
nor the ␥-phosphates of ATP contribute significantly to the binding of nucleotide to the P 1 -ATPase N-domains. It seems likely that the phosphorylation domain, which contains Asp-1027, the acceptor of ␥-phosphate during the catalytic cycle, could play a major role in the modulation of WNDP and MNKP nucleotide selectivity. An unexpected but reproducible finding was that the K d values for ATP were slightly higher than those for ADP and AMP. Although the differences were not large, they were beyond error and observed repeatedly in both MNKP and WNDP N-domains. This result is consistent with the notion that in the isolated N-domain, the ␥-phosphate of ATP experiences weak repulsive interactions. In support of this conclusion, the addition of Mg ϩ2 increases the binding affinity for ATP, bringing the K d value near those for ADP. It has been proposed that ATP binding facilitates movement of the N-domain toward the Pdomain in the P-type ATPases (14). It is tempting to speculate that the observed repulsive interactions between the N-domain and ATP's ␥-phosphate may contribute to such a movement. The structural analysis of the N-domain in a complex with the nucleotide would directly test this hypothesis; these experiments are currently underway in our laboratory.
We have previously expressed and characterized the ATPbinding domain of WNDP, which included both the N-domain and the P-domain (32). These earlier studies used indirect measurements of nucleotide binding based on competition with the fluorescent ATP derivative TNP-ATP. The experiments demonstrated that the isolated ATP-binding domain could bind the nucleotides at two sites (32). Our current studies illustrate that the N-domain has only one site. Therefore, the second set of the nucleotide-coordinating residues in ATP-binding domain is likely to be formed by the residues in the P-domain or by the P-domain and the linker region. This conclusion is consistent with the above notion that the P-domain could be essential for providing selectivity to ATP in the full-length WNDP.
Our studies demonstrated that the recombinant N-domain of WNDP represents an excellent tool for dissecting specific consequences of various disease mutations and identification of residues directly involved in nucleotide binding. A priori, it is difficult to predict the effect of a given mutation on nucleotide coordination, protein folding, domain-domain interactions, or a combination thereof. Utilization of the purified N-domain enabled us to separate these various effects. We demonstrate that despite close proximity of all analyzed residues to the putative ATP-binding cleft (Fig. 7), only Glu-1064 and His-1069 were important for nucleotide binding.
The strictly conserved nature of Glu-1064 and a complete loss of nucleotide binding upon mutation of this residue suggest that Glu-1064 is directly involved in ATP coordination in P 1 -ATPases. This conclusion is consistent with the earlier results of the experiments on zinc-transporting ATPase, ZntA. In these studies, the ZntA mutation E470A, which corresponds to the E1064A WD mutation, severely disrupted the enzyme's function. Specifically, the ATPase activity of E460A ZntA was not detected in 0.45 mM ATP and was reduced by ϳ90% in the presence of 4 mM ATP (20). The important role of Glu-1064 and its analogs in ATP coordination is also suggested by the predicted location of this residue near the adenosine moiety of ATP (Fig. 7).
Our data provide evidence that the invariant His-1069 is also important for nucleotide binding by the WNDP N-domain. However, the role of this residue seems to go beyond direct involvement in nucleotide coordination. Because of very weak ligand-substrate interactions, we were unable to precisely calculate the nucleotide affinity of the H1069Q disease mutant. However, based on the ratio of the ITC signals of wt and H1069Q titrations, we estimate the H1069Q mutant's affinity for ATP to be reduced by at least 15-fold. The marked effect of equivalent His-to-Gln substitution on apparent affinity for ATP was also reported for MNKP using measurements of the ATPdependent transport of copper (21). The ZntA H475Q mutant, an equivalent of the WNDP H1069Q mutation, was shown to have the ATPase activity that was 4 and 43% of wt in 0.45 and 4 mM ATP, respectively (20). Altogether, these data indicate that the invariant histidine in the N-domain of the P 1 -ATPase is important for ATP binding.
However, the mutation of His-1069 in the full-length WNDP also affects reactions that are independent of nucleotide binding, such as phosphorylation from inorganic phosphate (22). Similar effects were observed for the equivalent mutation in ZntA (20), suggesting that in the full-length protein, this histidine is located near the phosphorylation domain and possibly the actuator domain. Thus, it seems that His-1069 and the equivalent residues in other P 1 -type ATPases are strategically positioned to regulate the nucleotide-binding characteristics of the N-domain during the catalytic cycle when several functional domains come together or move apart. The fairly exposed location of His-1069, facing the P-domain (Fig. 7), is consistent with such a regulatory role. Future high resolution structural studies will determine whether this histidine residue is a direct ligand for ATP or regulates the affinity of the nucleotide-binding site via interaction with other domains.
In contrast to the critical roles of the strictly conserved residues in nucleotide binding, mutations of the non-conserved Cys-1104 and Arg-1151 seem to affect the N-domain function through different mechanisms. In particular, substitution of Cys-1104 with Phe disrupts the N-domain folding. In ZntA, another P 1 -ATPase, the position corresponding to Cys-1104 is occupied by alanine (Ala-508) indicating that the presence of Cys in this position is not essential for the ATPase function. Consistent with this conclusion, we found that substitution of Cys-1104 with alanine in WNDP had no effect on either protein folding or nucleotide affinity. Thus, the C1104F WD phenotype probably results from folding defects, rather than loss of an important contact between the cysteine residue and ATP.
In yet another example of the variable effects of WD mutations, the R1151H mutation had very modest effects on either folding or affinity for ATP (Tables I and II). These findings suggest that the residue is likely to play a role in domaindomain interactions or conformational transitions of the fulllength protein. Such a role would help to explain why a mutation that has negligible effects on protein folding or nucleotide binding, manifests as a disease mutation.
In summary, we have used direct nucleotide binding measurements to investigate the functional properties of the WNDP N-domain and several WD mutants. Analysis of the mutations unambiguously established their effects within the context of nucleotide binding. The observed spectrum of consequences on the N-domain structure and function contribute to our understanding of the phenotypic diversity of WD. In addition, the results demonstrate that the P 1 -ATPase N-domains possess functional properties distinct from the P 2 -ATPases.