A Structural-Dynamical Characterization of Human Cox17*

Human Cox17 is a key mitochondrial copper chaperone responsible for supplying copper ions, through the assistance of Sco1, Sco2, and Cox11, to cytochrome c oxidase, the terminal enzyme of the mitochondrial energy transducing respiratory chain. A structural and dynamical characterization of human Cox17 in its various functional metallated and redox states is presented here. The NMR solution structure of the partially oxidized Cox17 (Cox172S-S) consists of a coiled coil-helix-coiled coil-helix domain stabilized by two disulfide bonds involving Cys25-Cys54 and Cys35-Cys44, preceded by a flexible and completely unstructured N-terminal tail. In human Cu(I)Cox172S-S the copper(I) ion is coordinated by the sulfurs of Cys22 and Cys23, and this is the first example of a Cys-Cys binding motif in copper proteins. Copper(I) binding as well as the formation of a third disulfide involving Cys22 and Cys23 cause structural and dynamical changes only restricted to the metal-binding region. Redox properties of the disulfides of human Cox17, here investigated, strongly support the current hypothesis that the unstructured fully reduced Cox17 protein is present in the cytoplasm and enters the intermembrane space (IMS) where is then oxidized by Mia40 to Cox172S-S, thus becoming partially structured and trapped into the IMS. Cox172S-S is the functional species in the IMS, it can bind only one copper(I) ion and is then ready to enter the pathway of copper delivery to cytochrome c oxidase. The copper(I) form of Cox172S-S has features specific for copper chaperones.

Eukaryotic cytochrome c oxidase (CcO), 2 the terminal enzyme of the energy transducing respiratory chain of cells, is embedded within the mitochondrial inner membrane with a portion of the molecule protruding into the intermembrane space (IMS) (37 Å) and a portion extending into the matrix (32 Å) (1,2). Mammalian CcO is composed of 13 subunits, and its assembly is dependent on the insertion of several cofactors necessary for function, including two hemes, three copper ions, zinc, magnesium, and sodium ions (3). Subunit 1 (Cox1) contains two heme a cofactors, one of which interacts with a mononuclear copper site (designated Cu B ) forming a heterobimetallic active site (heme a 3 -Cu B ), whereas subunit 2 (Cox2) contains two copper ions in a binuclear center (designated Cu A ) (1).
Over 30 accessory proteins are necessary for the proper assembly of the enzyme (4,5). The functional role of these accessory factors in the assembly of CcO concerns the formation and insertion of heme a, the delivery and insertion of metal ions to corresponding binding sites, and the final maturation of this multi-subunit enzyme. Six proteins (Cox11, Cox17, Cox19, Cox23, Sco2, and Sco1) have been identified to be implicated in the delivery and insertion of copper ions into CcO (3).
Cox17 is the copper metallochaperone within the IMS acting as the donor of Cu(I) to both Sco1 and Cox11 (6). Cox17, which contains six conserved cysteines, can in principle exist in the IMS in three different oxidation states: from the fully oxidized protein with three disulfide bonds to a partially oxidized form with two disulfide bonds or to a fully reduced state where no disulfide bonds are present (7,8). These forms vary in terms of structural features and of metal binding ability. The partially oxidized state can bind one Cu(I) ion (Cu 1 (I)Cox17 2S-S hereafter), whereas the fully oxidized state is not able to bind copper (7). The structure of Cox17 from Saccharomyces cerevisiae was recently determined (9,10). From this characterization it results that Cox17 has a structural organization with an ␣-helical hairpin domain preceded by an unstructured N-terminal segment (9). The fully reduced form is on the contrary present in a molten globule state where the six free cysteines cooperatively bind four Cu(I) ions forming a tetracopper-thiolate cluster (Cu(I) 4 Cox17 hereafter) (7,9,11).
Recently, it has been found in yeast that Cox17 import into the IMS is catalyzed by a disulfide relay system involving Mia40 and Erv1 proteins, which favor the formation of the partially oxidized Cox17 2S-S state (12). It has been also found in vitro that Cu(I) 1 Cox17 2S-S and not Cu(I) 4 Cox17 transfers Cu(I) to apoWT-HSco1 (13). These data suggest that the Cox17 2S-S form is the active state in the copper transfer within the IMS.
In this paper we report the structural characterization of human Cox17 in the partially (HCox17 2S-S hereafter) and fully oxidized (HCox17 3S-S hereafter) states. The effect of reducing agents on HCox17 3S-S was also investigated. Finally, the structure and backbone dynamics of Cu(I) 1 HCox17 2S-S are reported and compared with those of apoHCox17 2S-S . Two consecutive cysteines binds copper(I) ion, and this is the first example of a Cys-Cys binding motif in copper proteins. Until now, the structure of the partially oxidized form (i.e. with two disulfides) is available only for the demetallated form of the yeast homologue (9).

EXPERIMENTAL PROCEDURES
Protein Production and Copper Binding-The hcox17 gene was amplified by PCR from a pET-11c expression vector already containing the human Cox17 cDNA (8). The hcox17 gene was cloned into the Gateway Entry vector pENTR/tobacco etch virus/D-topoisomerase (Invitrogen), and subcloned into pETG-30A (European Molecular Biology Laboratory Protein Expression and Purification Facility) by Gateway LR reaction to generate an N-terminal, His-GST fused protein. The protein is expressed in Escherichia coli BL21-Origami(DE3) cells (Stratagene), which were grown in Luria-Bertani and minimal medium (( 15 NH 4 ) 2 SO 4 and/or [ 13 C]glucose) for the production of labeled samples. Protein expression was induced with 0.7 mM isopropyl ␤-D-thiogalactopyranoside for 16 h at 298 K. Purification was performed by using a HiTrap chelating HP column (Amersham Biosciences) charged with Zn(II) ions. His-GST tag was cleaved with AcTEV proteases. The digested protein was concentrated by ultrafiltration and loaded in a 16/60 Superdex 75 chromatographic column (Amersham Biosciences) to separate HCox17 from the N-terminal His-GST domain. The fractions showing a single component by SDS/PAGE were collected, and the protein concentration was measured using the Bradford protein assay (14). The pure protein was obtained in the apo form as checked by inductively coupled plasma MS and ESI-MS. HCox17 protein expressed following the above mentioned procedure contains four additional amino acids (GSFT), corresponding to the TEV protease recognition site, at the N terminus. Recombinant native human Cox17, where the first Met is processed away posttranslationally, was used for ESI MS/MS analysis and was produced as already described (8). The numbering of both HCox17 constructs follows the isolated, functional mammalian Cox17 sequences (15,16) where the first Met is processed away posttranslationally. Therefore, the HCox17 sequence numbering starts proline 1.
The reduction of disulfide formed by the two consecutive cysteines, Cys 22 and Cys 23 , is almost instantaneous upon addition of 1 mM DTT under nitrogen atmosphere. The apoHCox17 2S-S protein was then exchanged under anaeroboic conditions into 50 mM phosphate, 1 mM DTT buffer at pH 7.2 using a PD-10 desalting column (Amersham Biosciences) and then concentrated by ultrafiltration to produce the final NMR sample. The Cu(I) form was obtained under anaeroboic conditions by addition of a slight excess of (Cu(I)(CH 3 CN) 4 )PF 6 directly to the final NMR apoHCox17 2S-S sample (0.5-1 mM). Copper excess was then removed dialyzing the sample against NMR buffer or through PD-10 desalting column. The NMR sample of the HCox17 3S-S form was obtained by air oxidation in about 2 days after the removal of 1 mM DTT, through PD-10 desalting column, from the final apoHCox17 2S-S sample.
Far-UV CD spectra (190 -260 nm) on the various forms of HCox17 were recorded on JASCO J-810 spectropolarimeter. Each spectrum was obtained as the average of four scans and corrected by subtracting the contributions from the buffer. A 0.6 mM apoHCox17 2S-S sample was divided in different fraction and each one was incubated for 1 h under anaerobic atmosphere with different DTT concentrations (0, 15, and 20 mM). Each sample was then diluted in 10 mM phosphate buffer, pH 7.2, to obtain a 15-30 M final protein concentration, and CD spectra were recorded. All of the steps were performed under nitrogen atmosphere using a degassed buffer. Quantitative estimate of the secondary structure contents was made by using the DICROPROT software package (17).
ESI MS/MS Analysis of Disulfide Pattern in HCox17-Recombinant human Cox17 3S-S (35 M) was reduced with 1 mM DTT in 20 mM ammonium acetate buffer, pH 7.5, for 1 min at 25°C, and the resultant Cox17 2S-S was alkylated with 5 mM iodoacetamide (1 h at 25°C in dark). Carboxamidomethylated Cox17 2S-S was desalted using HiTRAP TM desalting column (5 ml) (Amersham Biosciences) into 20 mM ammonium acetate buffer, pH 7.5, and two stable disulfides were reduced with 2 mM DTT at 55°C (incubation time, 120 min). Resultant Cox17 0S-S with two covalently attached carboxamidomethyl groups was digested with trypsin (using ratio 1:20) at 37°C for 30 min. The reaction products were separated on a reverse phase HPLC column Agilent Eclipse XDB-C18 (4.6 ϫ 150 mm; bead size, 5 m) by using a gradient from 5-40% of buffer B over 5 column volumes. Buffer A was 0.1% trifluoroacetic acid in water, and buffer B was 0.1% trifluoroacetic acid in 95% acetonitrile. The fraction containing Cox17 peptide of 1530.73 Da (KPLKPCCACPETK, residues 17-29 with two covalently attached carboxamidomethyl groups) was injected at 5 l/min into ESI-Q-TOF MS/MS instrument QSTAR Elite from Applied Biosystems (Foster City, CA) and analyzed in TOF MS and Q-TOF MS/MS mode. In MS/MS experiment a doubly protonated peak at 766.38 Da was selected for fragmentation, and collision energy between 40 and 60 CE was applied. Obtained MS/MS data were analyzed by program Bioanalyst 2.0 from Applied Biosystems. Cu(I) 1 HCox17 2S-S was produced by adding two equivalents of Cu(I)DTT complex to 40 M HCox17 2S-S , which was obtained by reduction of HCox17 3S-S as described above. Resultant Cu(I) 1 HCox17 2S-S was alkylated with 5 mM iodoacetamide after 120 min of incubation with copper ions. Carboxamidomethylated Cox17 2S-S was desalted, reduced, and trypsinolyzed as described above. Reaction products were separated by reverse phase HPLC as described above, and fraction containing Cox17 peptide of 3178.49-Da (PGLVD-SNPAPPESQEKKPLKPCCACPETK; residues 1-29 with two covalently attached carboxamidomethyl groups) was injected at 5 l/min into ESI-Q-TOF MS/MS instrument QSTAR Elite from Applied Biosystems and analyzed in TOF MS and Q-TOF MS/MS mode. In MS/MS experiment a triply protonated peak at 1060.55 Da was selected for fragmentation, and collision energy between 40 and 60 CE was applied. Obtained MS/MS data were analyzed as described above.
NMR Spectroscopy-All of the NMR experiments used for resonance assignment and structure calculations were performed on 0.5-1 mM 13 C, 15 N labeled HCox17 2S-S and HCox17 3S-S samples in 50 mM phosphate buffer, pH 7.2, containing 10% D 2 O (plus 1 mM DTT for HCox17 2S-S ) and are summarized in supplemental Table S1. All of the NMR spectra were collected at 298 K, processed using the standard Bruker software (XWINNMR), and analyzed through CARA program (18). The 1 H, 13  Structure calculations were performed with the software package ATNOS/CANDID/CYANA (20 -22), using as input the amino acid sequence, the chemical shift lists, and three 1 H, 1 H NOE experiments: two-dimensional NOESY, three-dimensional 13 C-resolved NOESY, and three-dimensional 15 Nresolved NOESY recorded at 800 and 900 MHz with a mixing time of 100 ms. The standard protocol with seven cycles of peak picking using ATNOS, NOE assignment with CANDID, and structure calculation with CYANA-2.1 (22) was applied. and dihedral angle constraints were derived from the chemical shift index (23) and TALOS analysis (24). In each ATNOS/ CANDID cycle, the angle constraints were combined with the updated NOE upper distance constraints in the input for the subsequent CYANA-2.1 structure calculation cycle. In the seventh ATNOS/CANDID/CYANA cycle, a total of 2366 or 2203 NOE cross-peaks were assigned from 2825 or 2555 peaks picked in the spectra of apoHCox17 2S-S and Cu 1 (I)HCox17 2S-S , respectively, which yielded 939 or 834 meaningful NOE upper distance limits. In addition, two disulfide bonds between Cys 35 and Cys 44 and between Cys 25 and Cys 54 were imposed, as resulted from their 13 C chemical shift analysis, by adding two lower and two upper distance constraints of 2.0 and 2.1 Å, respectively, between the S␥ atoms. The copper(I) ion was finally included in the calculations of the copper-loaded form by adding a new residue in the amino acid sequence. This residue is formed from a chain of dummy atoms with zero van der Waals' radii, so that they can freely penetrate into the protein, and by one atom with a radius of 1.4 Å, which mimics the cop-per ion. The sulfur atoms of Cys ligands were linked to the metal ion through upper distance limits of 2.3 Å, according to the yeast Cox17 S␥-Cu(I) distance (25). This approach does not impose any fixed orientation of the ligands with respect to the copper ion.
The 20 conformers with the lowest residual target function values were subjected to restrained energy minimization in explicit water with AMBER 8.0 (26). NOE and torsion angle constraints were applied with force constants of 50 kcal mol Ϫ1 Å Ϫ2 and 32 kcal mol Ϫ1 rad Ϫ2 , respectively. The force field parameters for the copper(I) ion and the ligands were adapted from those already reported for similar copper(I) sites in copper proteins (27,28). The quality of the structures was evaluated using the programs PROCHECK, PROCHECK-NMR (29), and WHATIF (30).
The root mean square deviation to the mean structure for the structured region of the protein (residues 24 -61) is 0.32 Ϯ 0.07 Å for the backbone and 0.74 Ϯ 0.07 Å for all heavy atoms for apoHCox17 2S-S and 0.31 Ϯ 0.09 Å for the backbone and 0.72 Ϯ 0.11 Å for all heavy atoms for Cu 1 (I)HCox17 2S-S . The conformational and energetic analysis of both structures are reported in supplemental Tables S5 and S6.
The atomic coordinates, structural restraints, and resonance assignments of apoHCox17 2S-S and Cu 1 (I)HCox17 2S-S have been deposited in the Protein Data Bank Resonance (PDB ID 2RN9 and 2RN8) and BioMagResBank (BRMB codes 11019 and 11020).
Relaxation experiments were performed on 15 N-labeled samples at 500 MHz. The 15 N backbone longitudinal (R 1 ) and transverse (R 2 ) relaxation rates as well as heteronuclear 15 N{ 1 H} NOEs were measured as previously described (31,32). 15 N relaxation parameters were then analyzed following the standard Tensor2 protocol (33).
Disulfide reduction of HCox17 2S-S and HCox17 3S-S forms was followed by NMR. To a 1:1 mixture of apo and Cu 1 (I)HCox17 2S-S or to HCox17 3S-S , both in 50 mM phosphate buffer, pH 7.2, containing 10% D 2 O, up to 20 mM or 1 mM DTT was added stepwise in anaerobic conditions, respectively, and two-dimensional 1 H-15 N HSQC spectra were acquired.

RESULTS
Human Cox17 in the fully oxidized and partially oxidized states, i.e. with three or two disulfide bonds, shows 1 H-15 N HSQC spectra with good signal spreading in both the apo and Cu(I) bound forms but also with a number of signals clustered in the central region of the spectrum, which is typical of unfolded polypeptides (amide proton resonances clustered between 8 and 8.5 ppm) (Fig. 1). This suggests that the protein contains both structured and unstructured regions. From the chemical shift index (23) and 3 J-coupling analysis, it appears indeed that all three protein forms have two helices (segments 26 -38 and 45-57), whereas the rest of the protein is not in any secondary structure element. The NH signals clustered in the central region of 1 H-15 N HSQC spectra, which belong to the first 17 amino acids of the protein, experience negative 15 N{ 1 H} NOE values indicating that they are highly flexible (see below). The partially oxidized form of HCox17 has features similar to those of the yeast homologue (9). Few NH signals of HCox17 2S-S exhibit significant spectral variations upon Cu(I) addition (Fig. 2), the most dramatic ones being for residues 20 -24, which comprise the metal binding motif Cys 22 -Cys 23 . These residues indeed experience, upon metal addition, either large chemical shift variations (Lys 20 and Ala 24 ) (Fig. 2) or the appearance of their NH signals (Cys 22 and Cys 23 ), which are not observed in the apo state (Fig. 1). A similar trend is observed going from the partially oxidized HCox17 2S-S state toward fully oxidized HCox17 3S-S one. Significant spectral changes are indeed observed only in the vicinity of the two Cys involved in the formation of the third disulfide bond, where either chemical shift variations (A24) (Fig. 2) or the appearance of the NH signal of Cys 23 are occurring (Fig. 1). These data suggest that both copper(I) binding to HCox17 2S-S as well as the formation of the third disulfide essentially determines local structural changes restricted to the Cys 22 -Cys 23 -Ala 24 region. However, at variance with the HCox17 2S-S state, in HCox17 3S-S some NHs (Gly 59 , Leu 58 , Met 55 , Lys 30 , Glu 27 , and Cys 25 ) located in the vicinity of the Cys 22 -Cys 23 -Ala 24 motif display two conformations as detected in the 1 H-15 N HSQC map (Fig. 1), indicating that the disulfide bond formation determines a structural heterogeneity in the surrounding of this region. Double conformations have been reported to occur for disulfide bonds as a result of their isomerization (34).
The structures of apoHCox17 2S-S , Cu 1 (I)HCox17 2S-S , and apoHCox17 3S-S have the coiled coil-helix-coiled coil-helix structural motif (CHCH) (Fig. 3) as observed in the yeast homologue (9,10). This structural motif is predicted to be common to several mitochondrial proteins like Cox19 (35) and Cox23 (36), which are, similar to Cox17, involved in copper ion insertion into CcO (3), Mia40, which is required for the import of Cox17  and Cox19 into the IMS (37), and to several other IMS proteins whose functions are unrelated to CcO assembly (38). Accordingly, all the above proteins have four cysteine residues organized in the twin CX 9 C motif, located at the N-and C-terminal ends of each helix of the predicted CHCH motif. These conserved cysteines are those that form two interhelical disulfide bonds (Fig. 3), thus forcing the two helices to get close each other in an antiparallel mode forming an ␣-hairpin. The global backbone root mean square deviation value (calculated on the structured region) between yeast and human Cox17 2S-S structures is low (1.4 Å). The only meaningful structural difference between them is found at the N-terminal part of the second helix, which is indeed shorter in HCox17 2S-S of one turn but, at variance with yeast Cox17 2S-S , is followed by a 3 10 helix (Fig. 4).
Copper(I) ion in Cu 1 (I)HCox17 2S-S is coordinated by the sulfurs of two adjacent Cys, forming a S-Cu-S angle of about 130° (  Fig. 3). The two copper-binding cysteines, which are conserved in all Cox17 proteins, are the ones adjacent within the Cys 22 -Cys 23 -Ala 24 -Cys 25 motif. No other protein atoms appear close enough (Ͻ2.5 Å) to be the third copper(I) ligand in HCox17 2S-S . From the structure and the 2 J NH coupling-based 1 H-15 N HSQC experiment, it results that all the three His, potential ligands of copper, are protonated on N⑀2, and no one is coordinated to the metal ion (39,40). The coordination sphere of Cu(I), which is extensively solvent exposed, could be completed by an exogenous molecule, such as DTT, which is present in the sample in 1 mM concentration. Such tricoordinated sulfur environment has been already suggested in several members (Atx1, Hah1, and CopZ) of a cytoplasmic metallochaperone family (41)(42)(43).
Cu(I) 1 HCox17 2S-S and apoHCox17 2S-S were also studied by ESI MS/MS analysis to characterize the metal-binding cysteines. For this purpose both Cu 1 HCox17 2S-S and apoHCox17 2S-S were treated with iodoacetamide, which is able to alkylate reduced cysteine, and these modified residues were identified by MS/MS analysis of trypsinolytic peptides containing CCAC fragment. Analysis of doubly carboxamidomethylated peptide fragment 17-29 (KPLKPCCACPETK), obtained from apoHCox17 2S-S , indicated that the first two adjacent Cys residues (Cys 22 and Cys 23 ) were carboxamidomethylated, whereas Cys 25 of the peptide was unmodified (supplemental Fig. S1). After alkylation and trypsinolytic treatment of Cu(I) 1 HCox17 2S-S , we identified by ESI-MS peptide 1-29 (PGLVDSNPAPPESQEKKPLKPCCACPETK) where two of the three contained Cys are covalently attached by carboxamidomethyl groups (3178.49 Da). Theoretical molecular mass of peptide 1-29 is 3064.54 Da, whereas theoretical molecular mass of peptide 1-29 with two carboxamidomethylated Cys residues is 3178.54 Da. From MS/MS spectra of peptide 1-29 we identified following masses: 473.25, 576.26, and 647.29 Da, which correspond to y 4 (PETK), y 5 (C 25 PETK), y 6 (AC 25 PETK) fragment, and 807.30 Da, which corresponds to fragment y 7 (C 23 AC 25 PETK) containing one carboxamidomethylated Cys residue. Overall, these results indicate that the two adjacent Cys residues (Cys 22 and Cys 23 ) are both carboxamidomethylated, whereas the remaining Cys 25 of the peptide is not alkylated.  Thus, in both apo and Cu(I) 1 HCox17 2S-S , the two adjacent Cys residues (Cys 22 and Cys 23 ) are in reduced state, in agreement with NMR results.
The heteronuclear relaxation data on both apoHCox17 2S-S and Cu 1 (I)HCox17 2S-S (supplemental Fig. S2) point at two protein regions with distinct motional regimes, one for the first 17 amino acids and the other for the 25-62 segment of the protein.
The N-terminal region is characterized by negative 15 N{ 1 H} NOE values that indicate the presence of motions faster than the overall molecular tumbling. Also, the S 2 values, estimated through a model-free approach with Tensor2 program (33), are quite low for the N-terminal segment (residues 2-17), being 0.40 Ϯ 0.09 in the apo form and 0.36 Ϯ 0.16 in the Cu(I) form. On the contrary, the majority of residues in the 25-62 segment are more rigid in both forms (supplemental Fig. S2). However, the 25-62 segment of apoHCox17 2S-S shows fast backbone NH motions with larger amplitude than Cu 1 (I)HCox17 2S-S , as resulted from its lower average S 2 parameter (S 2 (apo) ϭ 0.65 Ϯ 0.14 versus S 2 (Cu(I)) ϭ 0.86 Ϯ 0.13). At variance with Cu 1 (I)HCox17 2S-S , backbone NH motions of several residues belonging to the 25-62 segment of apoHCox17 2S-S are also characterized by exchange contributions (R ex ) to their relaxation. Backbone NHs of the five, non-proline, residues in between the unstructured N-terminal tail and the 25-62 segment, comprising the CCAC motif display conformational motions much more pronounced in apoHCox17 2S-S than in Cu(I) 1 HCox17 2S-S . NHs of the copper(I)-binding cysteines are indeed not detected in the apo form, likely as a consequence of exchange processes, whereas their NH cross-peaks appear upon copper(I) binding, indicating a more rigid backbone conformation in the latter form. However, their R 2 values in Cu(I) 1 HCox17 2S-S are higher than the average (calculated on the 25-61 segment) (supplemental Fig. S2), indicating that a certain degree of conformational mobility is still present. Copper(I) binding also reduces the backbone flexibility of Leu 19 and Lys 20 . Their 15 N{ 1 H} NOE negative in the apo form indicate motions faster than the overall molecular tumbling rate, whereas positive values in the copper(I) form indicate a decreased flexibility (supplemental Fig. S2). In conclusion, copper(I) binding drastically reduces the backbone motions in the metal-binding region of HCox17 2S-S .
Investigating the effect of reducing agents on HCox17 3S-S , 1 H-15 N HSQC data show that 1 mM DTT is able to easily reduce the disulfide bond formed within the CC motif in Cox17 3S-S , thus producing the partially oxidized Cox17 2S-S form. NMR titration of a 1:1 mixture of apo and Cu 1 (I)HCox17 2S-S shows that the addition of 15 mM DTT is necessary to completely remove copper(I) ion, whereas the two disulfides within the CHCH motif can be reduced only with further additions of DTT up to 20 mM. At the latter DTT concentration, all of the signals that are spread out in the folded region of the 1 H-15 N spectrum of apoHCox17 2S-S disappear, with the concomitant formation of new cross-peaks clustered in the spectral region typical of unstructured polypeptides (amide proton resonances between 8 and 8.5 ppm) (Fig. 5). The complete reduction of all disulfides of HCox17 therefore determines the formation of a state without a well defined tertiary structure. CD spectra of apoHCox17 2S-S were then measured in the presence of various concentrations of DTT (supplemental Fig. S3) to address secondary structural variations occurring from the partially oxidized to the fully reduced states. At 1 mM DTT, Cox17 2S-S exhibits the characteristic bands of ␣ helix conformation, with double minima at 222 and 206 nm as well as a positive maximum at 192 nm. The fitting of the CD spectrum indicates an ␣-helical content of 40%. After overnight incubation with 20 mM DTT, thus obtaining the fully reduced species, the ␣-helical content is still about 30%. This indicates that the polypeptide chain has a high propensity to adopt a helical conformation even in a completely reduced state. This behavior together with that observed through NMR indicates therefore that the fully reduced form is essentially in a molten globule state and is the same observed in the yeast Cox17 homologue (9).

DISCUSSION
Human Cox17 in both apo and copper(I) forms is a protein constituted by a CHCH motif of about 40 residues plus an unstructured, flexible N-terminal tail of about 15 residues. The structural and dynamical properties of the residues in between these two regions (residues 17-24), which comprise the metal binding motif, are the only ones significantly modulated by the binding of copper(I) ion. The latter region in apoHCox17 2S-S is indeed highly unstructured with a large degree of backbone flexibility, whereas upon copper(I) binding, it becomes more structured and less flexible. In particular, copper(I) binding determines local structural rearrangements around the two coordinating cysteines of the CC motif, determining the formation of a turn that positions the two consecutive Cys ligands close to each other in an optimal conformation for metal binding (Fig. 3). To our knowledge, this is the first example of a copper(I) ion coordinated by two consecutive Cys residues. Around the metal-binding region of Cu(I) 1 HCox17 2S-S , no other protein atom is at bond distance, but in the second coordination sphere (Ͻ6.0 Å), the copper(I) ion is surrounded by two conserved charged residues, Lys 20 and Lys 29 (Fig. 6). In particular, Lys 20 , located at the end of the CC turn above the metal ion, takes a well defined conformation very close to the copper(I) ion (mean distance in the family of conformers 4.5 Å); on the contrary, it is highly conformationally disordered in the copper-free form. The fast internal motions of Lys 20 are also highly reduced upon copper(I) binding. These features strongly resemble those of the copper-binding region of a well characterized cytosolic metallochaperone, Atx1 (44). It has been found that the eukaryotic chaperones possess a conserved lysine residue located adjacent to the copper-binding site (27). This lysine (Lys 65 in yeast Atx1) has been proposed to have a functional role in stabilizing copper binding (27) and modulating copper transfer (45). Therefore, we suggest that, similarly to yeast Atx1, the proximity of Lys 20 , which takes a defined conformation in the copper form, contributes to the stabilization of the overall negative charge resulting from binding of Cu(I) to two cysteinate anions. Its approach toward the copper site could represent a local rearrangement of the protein structure for optimizing the electrostatic interactions upon copper binding. The effect of a mutation on this position is also similar in the two proteins, because its change to an Ala residue leaves a still functional protein in both cases (46,47), suggesting that a neutral residue at this position appears well tolerated in both cases. Negative charge is, however, not well tolerated in this position, resulting indeed in a compromised function for the Atx1 chaperone (47).
Another important structural feature of the metal binding surroundings, helping to organize the copper ligands and Lys 20 in the appropriate orientation for metal binding, is determined by the hydrophobic contacts between the residues located at the end of the C-terminal helix (Met 55 , Leu 58 , and Phe 60 ) with those at the N terminus (Leu 19 , Pro 21 , and Ala 24 ) (Fig. 6). All of these residues, which are highly conserved in homologous sequences, form in Cu(I) 1 HCox17 2S-S a compact hydrophobic patch that orients the cysteine thiols, and Lys 20 , toward the protein surface and exposed to the solvent (Fig. 6), thus favoring an efficient metal transfer with the protein partners, according to the metallochaperone function of this protein in the IMS (Fig. 7). In the apoHCox17 2S-S form this hydrophobic patch is partially destabilized by the higher degree of backbone flexibility of the residues at the N terminus, determining less compact hydrophobic interactions (Fig. 6). This behavior is again similar to what was observed for the Atx1 metallochaperone where, in the metal-binding region, less compact hydrophobic interactions associated with an increase of backbone motions are observed upon copper(I) release, determining a reduction of the first turn at the N terminus of helix ␣1 in the metal-binding site (27). Overall, from the analysis of the structural and dynamical properties, we can therefore conclude that the copper(I) form of Cox17 2S-S has features specific to copper chaperones.  Copper transfer from Cu(I) 1 HCox17 2S-S to its protein partners, apoHSco1 and apoHCox11, occurring in the IMS is depicted. The solvent exposed metal-binding site of HCox17 2S-S can easily be accessible for coordination by a further thiol group of a cysteine residue belonging to apoHSco1 or apoHCox11. Cys residues involved in the two disulfide bonds within CHCH motif are not involved in the metal transfer mechanism as a consequence of their high stability toward reduction.
Concerning biological context, as human Cox17 displays several structural and dynamical properties that are typical of the cytoplasmic metallochaperone Atx1, it can therefore play a role within IMS for tightly controlling the copper concentration, similar to what it has been found in the cytoplasm where several systems of cellular copper transport, involving also Atx1, have been discovered (48). However, at variance with Atx1, the global fold of HCox17 is quite atypical with, indeed, a large unstructured segment. It is also completely unrelated to its protein partners Sco1 (49) and Cox11 (50), which in turn have different folds one from the other. Atx1 has, on the contrary, the same fold as its protein partner Ccc2a without the presence of unstructured regions (27,51). These structural differences of the Cox17/Sco1/Cox11 pattern versus the Atx1/ Ccc2a pattern can play an important role in modulating the metal transfer processes in different ways. Similar folds in the protein partners can be indeed important in determining a reversible copper transfer mechanism, as found in the Atx1/ Ccc2a interaction (52), whereas the different fold of the protein partners found in the Cox17/Sco1/Cox11 interactions can somehow have a role in the quantitative copper transfer observed for the Cox17/Sco1 pair (13), as well as in the recognition and binding to several different partners. In the case of the Cox17/Sco1 pairs, the quantitative copper transfer can be also driven by the higher number of Sco1 metal-binding ligands, with the copper(I) ion in Sco1 being indeed coordinated by two cysteines and one histidine far away in the sequence (49). The fold of HCox17 contains two CX 9 C motifs and is typical of systems involved in Cys redox reaction within the IMS (38). In particular, the protein partner Mia40 (37) also has the twin CX 9 C motifs, thus presumably having a similar fold that can be involved in the protein-protein recognition process occurring during HCox17 entrapment into the IMS. On the contrary, the KXCC motif is consistent with a chaperone function.
Human Cox17 3S-S protein has an easily reducible disulfide bond, corresponding to the one involved in copper(I) binding, and two disulfides in the CHCH motif that are, on the contrary, highly stable toward reduction. These results support a model where, once the protein is maturated into the IMS forming the Cox17 2S-S state through Mia40 assistance (12), it performs its function of copper(I) chaperone remaining in the latter redox state (Fig. 7), which can be thus considered the predominant one within the IMS, where the redox environment is likely more oxidative as compared with the cytosol (11,53). On the contrary, the rupture of the two disulfide bonds within the CX 9 C motifs in high reducing conditions (20 mM DTT), which reasonably mimic the cytosolic redox environment, determines the disruption of the ␣ hairpin structure, generating a fully Cysreduced molten globule state of HCox17 necessary for its import from the cytoplasm into the IMS (12).
It should be pointed out that replacement of the conserved Cys 57 with a Ser does not perturb Cox17 function in yeast, at variance with the substitution of Cys 26 with a Ser (54). Because the latter two cysteines are disulfide partners in the HCox17 2S-S structure and are not involved in copper binding, it is possible that these mutations affect the protein import and/or retention mechanism into IMS via Mia40 (12, 55), preventing the forma-tion of a functional Cox17 state only when Cys 26 is mutated. Accordingly, their amount within the IMS is largely reduced with respect to that of wild-type protein (54). On the contrary, mutations of the other two disulfide-linked Cys in HCox17 2S-S structure results in a functional protein in yeast (54), suggesting that they are not essentially perturbing Cox17 import. Cys 22 and Cys 23 are essential for CcO activity of yeast (54) because they are involved in Cu(I) 1 HCox17 2S-S structure in copper(I) binding, thus abolishing copper chaperone function. From this structural-functional analysis, we can suggest that the copper chaperone function of HCox17 relies essentially only in the availability of the KXCC motif, whereas the CHCH motif is important to allow HCox17 to be trapped in the IMS through the interaction with Mia40.
In conclusion, in this work we have elucidated the structural, dynamical, and redox properties of human apo and Cu(I)Cox17 2S-S and discussed them in relation to their involvement in copper(I) transfer toward cytochrome c oxidase copper sites as well as in the protein import into the IMS.