A NMR Study of the Interaction of a Three-domain Construct of ATP7A with Copper(I) and Copper(I)-HAH1

ATP7A is a P-type ATPase involved in copper(I) homeostasis in humans. It possesses a long N-terminal tail protruding into the cytosol and containing six copper(I)-binding domains, which are individually folded and capable of binding one copper(I) ion. ATP7A receives copper from a soluble protein, the metallochaperone HAH1. The exact role and interplay of the six soluble domains is still quite unclear, as it has been extensively demonstrated that they are strongly redundant with respect to copper(I) transport in vivo. In the present work, a three-domain (fourth to sixth, MNK456) construct has been investigated in solution by NMR, in the absence and presence of copper(I). In addition, the interaction of MNK456 with copper(I)-HAH1 has been studied. It is proposed that the fourth domain is the preferential site for the initial interaction with the partner. A significant dependence of the overall domain dynamics on the metallation state and on the presence of HAH1 is observed. This dependence could constitute the molecular mechanism to trigger copper(I) translocation and/or ATP7A relocalization from the trans-Golgi network to the plasmatic membrane.

Copper, an essential trace metal, is utilized as a cofactor in a variety of redox and hydrolytic proteins, which in eukaryotes are found in various cellular locations (1). However, copper can be potentially toxic in vivo, and thus its intracellular concentration is presumably strictly controlled (2-4). Disruption of copper homeostasis leads to illness, such as in Menkes or Wilson diseases (5). These latter diseases are caused by mutations in ATP7A and ATP7B, respectively, which translocate copper in the trans-Golgi network or across the plasma membrane (3,4), depending on cellular conditions (6). Even though ATP7A and ATP7B contain six cytosolic copper(I)-binding domains, it appears that one domain (7,8), or even no soluble domain if copper is abundant (9), is sufficient for copper translocation. In addition, only one of the two most C-terminal metal-binding domains is sufficient for correct protein relocalization (10), whereas both C-terminal metal-binding domains (i.e. the fifth and sixth domains) are needed to regulate copper affinity and protein phosphorylation rates (11). In humans, the function of HAH1 is to transfer copper(I) to the ATP7A and ATP7B proteins (12). This proc-ess has been investigated through a variety of low-resolution techniques, suggesting that there may be preferential interaction sites within the cytoplasmic tail (13,14). NMR data are available for the interaction of individual ATP7A soluble domains (2, 5, and 6) with HAH1, showing copper(I) transfer and a slow kinetics of interaction (15).
The possible differentiation in vivo among the six domains, their possible reciprocal interaction(s), and the interactions with HAH1 and with copper(I) are all open matters of discussion. Here, these matters are addressed through the NMR investigation of a three-domain ATP7A construct, spanning domains 4 through 6 (MNK456 hereafter). 2 This construct presents several points of interest: (i) it contains the last two domains, which are important for catalytic phosphorylation of ATP7B (11); (ii) it contains domain 4, one of the proposed favored interaction sites with HAH1 (14); and (iii) domains 4 and 5 are connected by a 41-residue linker, whereas domains 5 and 6 are connected by a 6-residue linker (shortest interdomain linker in ATP7A). This combination provides insight into the global dynamics and domain interplay of the N-terminal tail of the enzyme and thus information on possible molecular mechanisms in vivo.

MATERIALS AND METHODS
The three C-terminal metal-binding domains of ATP7A have been expressed as a single construct in Escherichia coli and purified using affinity chromatography based on the use of a His 6 tag. The protein has been produced in M9 minimal medium and enriched in 15 N or 15 N and 13 C for NMR studies. The thermal stability of apoMNK456 has been investigated through 1 H-15 N HSQC spectra (16) at various temperatures, from 298 K up to 323 K. The protein was metallated by titrating it with an acetonitrile complex of copper(I) after reduction with a 5-fold excess of DTT. The titration was followed through 1 H-15 N HSQC spectra. The MNK56 and MNK4 constructs have been similarly produced in E. coli from rich medium cultures.
The protein showed significant precipitation over a span of 1-2 days at concentrations above 500 M. To minimize sample aggregation and obtain the best quality spectra, NMR samples were typically 200 -300 M in a 100 mM sodium phosphate buffer at pH 7.0. All spectra were acquired using cryoprobe technology at 500 (for protein dynamics measurements) or 800/900 MHz (for experiments for resonance assignment) spectrometers. The NMR frequencies of backbone nuclei have been assigned using a standard approach based on triple resonance * This work was supported by Ministero dell'Istruzione, dell'Universitá e della Ricerca (MIUR)-COFIN 2003, MIUR-Fondo Integrativo Speciale Ricerca, Ente Cassa di Risparmio di Firenze (Project "PROMELAB"), and the European Commission (Contract QLG2-CT-2002-00988). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. □ S The on-line version of this article (available at http://www.jbc.org) contains supplemental Figs. S1 and S2 and supplemental Tables S1 and S2. 1  experiments and are given in the supplementary material for both the apo and the fully metallated state (Cu 3 MNK456) (supplemental Tables S1 and S2). The backbone dynamics of MNK456 was studied in the apo form as well as in Cu 3 MNK456 through the analysis of 15 N R 1 and R 2 relaxation rates and heteronuclear 1 H-15 N NOEs (17). The interaction with the human copper(I) chaperone HAH1 has been studied through NMR by titrating isotopically enriched apoMNK456 with unlabeled copper-(I) HAH1 up to a HAH1:MNK456 molar ratio of 8:1. In addition, the backbone dynamics of MNK456 was also studied in the presence of HAH1 at various molar ratios (1:1, 2:1, 4:1). ApoHAH1 was purified from rich medium cultures as described previously (18), reduced with 2-fold DTT, and then metallated by adding a slight excess of [Cu(CH 3 CN) 4 ] ϩ in a N 2 atmosphere chamber. Samples were then passed on a desalting column to remove DTT and acetonitrile. The degree of metallation of HAH1 was checked by NMR and found to be essentially 100%. For titrations, copper(I)-HAH1 was added to reduced apoMNK456 samples from which DTT had been removed using a desalting column, in a N 2 atmosphere chamber. Relative ratios of HAH1:MNK456 were measured from the intensity of well resolved NMR signals in the methyl region.
For chemical labeling experiments, 400 pmol of apoMNK456 alone or in the presence of copper(I)-HAH1 were incubated with a 50-fold molar excess of the cysteine-directed reagent 7-diethylamino-3-(4Ј maleimidylphenyl)-4-methylcoumarin (CPM, Sigma) for 2.5 min in the dark under anaerobic conditions. The reaction was quenched with a 10-fold molar excess of ␤-mercaptoethanol over CPM. The same experimental setup has been used for MNK56 and MNK4. To investigate the role of the various domains within MNK456, the latter construct was partially proteolyzed with trypsin at a 1:2000 (w/w) ratio for 3 h at room temperature. The reaction was stopped with the addition of 2 mM phenylmethylsulfonyl fluoride (Sigma) protease inhibitor. The proteolyzed MNK456 fragments were separated on a 15% Tricine gel, and the separation of the CPM-labeled peptides was monitored under UV light using a Gel-Doc system. The protein fragments were also stained with Coomassie R250. The present protocol is essentially the same as that described elsewhere (13). In the case of the MNK56 and MNK4 constructs, no digestion was performed.

RESULTS AND DISCUSSION
The NMR spectra of MNK456 are quite crowded and showed broadening due to conformational exchange processes, thus complicating the analysis and interpretation of spectra (supplemental Fig. S1). The observed broadening experienced a significant enhancement with increasing protein concentration. The problem was particularly evident for domain 5, as also observed for the isolated domain (15). The complexity and behavior of MNK456 in solution prevented a de novo structural characterization of the protein. As the NMR structure of all of the individual domains is available (15,19), backbone amide chemical shifts, which are excellent reporters of tertiary structure changes (20), could be compared. Fig. 1 indeed shows that the structure of the three metalbinding domains is conserved in apoMNK456, as significant variations are localized only next to linker regions and are a consequence of the peptide bonds at the N and C termini of each domain (see Equation 1 below).
After the addition of a 1.0 equivalent of copper(I) to apoMNK456, signals from the residues in the loop containing the metal-binding cysteines (loop I) of the metallated fourth and sixth domains were both already observable (supplemental Fig. S2). In the fifth domain, the signals from the residues in loop I are not observed in the apo or in the copper(I) form. The relative intensity of the signals of the metallated fourth and sixth domains indicates that they have similar affinity for copper (within 20%). When increasing the Cu:MNK456 ratio, broadening of several signals within and close to the metal-binding loops of all three domains is observed. In the fifth domain most signals, even far from the metal-binding region, become extremely weak or undetectable.
Based on 15 N relaxation data, a distinct dynamic behavior is observed in different MNK456 regions (Fig. 2). Each of the three domains appears to be individually rigid in solution on the subnanosecond time scale, as shown by the relatively uniform 15 N R 1 and NOE values within each domain, whereas the first linker region experiences extensive dynamics on the same time scale. The overall dynamics of the multidomain construct is somewhat complex. Domain 4 reorients in solution more freely than the other two domains, being allowed to do so by the long flexible linker I (shown in Fig. 2 by the higher R 1 and lower R 2 values). Indeed, the correlation time for tumbling in solution of domain 4, as calculated from R 2 /R 1 values, is significantly lower than for domains 5 and 6 (7.1 Ϯ 0.3, 12.8 Ϯ 1.3, and 10.8 Ϯ 1.3 ns for domains 4, 5, and 6 respectively). Note that the above values for the correlation times should be regarded as estimates because they are based on the assumption of isotropic reorientation in solution. Conformational equilibria are present in domains 5 and 6, which tumble in solution with similar rates. The similar tumbling rate of the two C-terminal domains is in agreement with what observed for a Bacillus subtilis homolog of ATP7A (21) and can be ascribed to the shortness of linker II. A small degree of uncoupling in the motion of domains 5 and 6 is possibly reflected in the slightly higher correlation time of the former, even though the difference is hardly statistically significant. Loading apoMNK456 with copper(I) induces a constraining of the motional freedom of domain 4 (Fig. 3A), in which apparent correlation time increases from 7.1 Ϯ 0.3 to 8.3 Ϯ 0.6 ns. The mobility of domain 6 appears, instead, globally unaltered. It was not possible to estimate the mobility of domain 5 in the presence of copper(I) because of the aforementioned difficulty in observing signals from its nuclei.
When apoMNK456 is presented with copper(I)-HAH1, the metal ion is transferred from the chaperone to the ATPase. As shown in Fig. 4, signals of Cys 17 and Val 18 in the metal-binding loop (loop I) of the apo form of domain 4 disappear upon the addition of copper(I)-HAH1 at a HAH1:MNK456 ratio slightly exceeding 1.0 (i.e. when only one copper(I) ion is available per MNK456 molecule), suggesting a slow kinetics of exchange similar to isolated MNK2 and MNK5 domains (15). A reduction of the intensity of signals from the metal-binding loop of domains 5 and 6 is also observed during the titration, but this is much shallower than for domain 4, suggesting that in the initial stages (HAH1: MNK456 ratio up to 1.0) of the titration copper is preferentially transferred to domain 4. The chemical shift variations observed during the titration are very small. Signals from copper(I) domain 6 appear at a HAH1:MNK456 ratio of 2:1, showing that copper(I) eventually is loaded in this domain. To analyze independently the possible preferential interaction of copper(I)-HAH1 with a single domain within the present three-domain construct, an experiment based on chemical labeling with a fluorophore and proteolysis, as applied to the Wilson protein (13), was performed (Fig. 5). Cysteines become labeled by the fluorophore only when they are not engaged in any bond, i.e. when they are reduced and not bound to the metal. Therefore, the detection of the modification by the chemical label, or absence thereof, is indicative of the copper(I)binding state (13). The experiment shows that copper(I)-HAH1 does transfer the metal to MNK456, as HAH1 is (partly) labeled by the fluorophore (separate experiments, not shown, demonstrate that copper(I)-HAH1 does not react with the chemical label). In the presence of excess copper(I)-HAH1 (Fig. 5, rightmost lane), one of the bands from the MNK456 digest disappears, demonstrating preferential protection from the cysteine-directed probe caused by copper(I) transfer (13). The molecular weight of the disappearing band is about 8,000 Da, indicating that it contains a single domain. As discussed previously for the Wilson protein (13), this experiment shows that copper(I)-HAH1 selects an individual domain in the present three-domain construct as the initial interaction site. The results of NMR experiments (Fig. 4) are fully con-  sistent with this picture and clearly suggest domain 4 as the preferential interaction site. To assess independently the identity of the domain constituting the preferential site for HAH1 docking, we separately produced MNK56 and MNK4 and repeated the chemical labeling experiments (Fig. 6). Fig. 6 clearly shows that copper(I)-HAH1 can transfer its metal cargo to MNK4 but not to MNK56. Transfer to MNK4 is demonstrated by the disappearance of the MNK4 band and the appearance of a lower molecular weight band due to apoHAH1 (Fig. 6, compare the fourth and third lanes). Instead, MNK56 reacts with the fluorophore both in absence and in presence of copper(I)-HAH1 (Fig. 6, first two  lanes), demonstrating that it is unable to uptake the metal ion from the chaperone.
NMR shows that the presence of copper(I)-HAH1 affects protein dynamics with respect to isolated apoMNK456 (Fig. 3B), inducing slower tumbling in domain 4 (but to a smaller extent than Cu 3 MNK456) as well as enhanced conformational averaging. Instead, in domains 5 and 6 copper(I)-HAH1 induces faster tumbling (tumbling correlation times reduced respectively by 1.4 Ϯ 2.4 and 1.0 Ϯ 1.9 ns). In addition, the presence of HAH1 causes conformational exchange in the linker I region, so that signals from residues in this region are broadened beyond detection.
The present data are consistent with a mechanism for copper(I) transfer to MNK456 from HAH1, where the first entry point for the metal ion into the ATPase is provided by domain 4. This is presumably because of MNK4 having a better electrostatic and/or steric complementarity to the physiological partner. Shielding of domains 5 and 6 by domain 4 in MNK456 is unlikely to play a role in tuning the interaction, as the two-domain construct MNK56 does not interact with HAH1 (Fig.  6). Domain 4 has a surface electrostatic potential relatively similar to that of the two metal-binding domains of Ccc2, the yeast homolog of ATP7A, which form a relatively stable adduct with their partner (22,23). It is also relevant that the pI of domain 4 in ATP7A, as well as in ATP7B, is the most similar to that of both domains of yeast Ccc2 (4.0 -4.5). Other ATP7A domains, in which the surface is somewhat dissimilar from Ccc2, do not form such an adduct and feature a slow kinetic of interaction with HAH1 (15). A role for domain 4 as the preferential interaction site for HAH1 in ATP7B has been proposed recently (14), possibly together with domain 2, whereas earlier work identified only domain 2 (13). Notably, domain 2 has a pI slightly higher than 8.0 in ATP7A and ATP7B, thus quite different from domain 4, but the electrostatic surface at the putative interaction site is still compatible with that of HAH1 (13). The present data show that copper(I) is anyway distributed over the various domains. It is not possible to ascertain whether transfer to domains other than the fourth happens by way of direct interaction with domain 4 or HAH1 or through the release of free copper(I) in solution. The latter mechanism appears quite unlikely to be physiologically relevant.
A second noteworthy conclusion from the present work is that the dynamics of the multidomain MNK456 construct is quite elaborate, depending significantly on the presence of copper(I) and HAH1. This can be the result both of a modulation of interdomain contacts and of the variation of dynamics in the linker I region. There is also some unexpected modulation in the dynamic processes involving the interaction between domains 5 and 6, permitted by the relatively short linker II. It can be suggested that the dependence of domain dynamics on copper(I) and HAH1 (which might be associated also to changes of the time-averaged protein conformation) constitutes a molecular mechanism to signal (internally to the other domains and/or externally to other biomolecules) the status of copper(I) loading of the cytoplasmic tail. Indeed, the present data provide direct experimental support for previous proposals that ATP7A multiple domains could modulate protein relocalization and/or the affinity for copper(I) of the metal binding sites in the trans-membrane region through copper(I)-dependent variations in the conformation and/or dynamics of the cytoplasmic tail (11,24).
In summary, it is proposed that when interacting with the MNK456 triple domain construct, copper(I)-HAH1 preferentially donates its cargo to domain 4. The present work thus reinforces the view that interaction of ATP7A or ATP7B with its partner metallochaperone does not involve all of the domains equally, with some domains actually interacting more readily than others. This preferential interaction is presumably under kinetic control. Thermodynamically all domains   have a similar affinity for copper(I) (25). The metal donated by HAH1 is indeed found to be distributed over all domains in MNK456. It might be supposed that in vivo this (re)distribution happens via interdomain contacts. Loading the cytoplasmic tail of ATP7A with copper(I) and the interaction with the partner induce a significant change in ATP7A domain dynamics. This change (or the associated variation of the timeaveraged protein conformation) may be relevant to activate copper(I) translocation across the membrane and/or ATP7A relocalization from the trans-Golgi to the cytoplasmic membrane.