Characterization of COX17 , a Yeast Gene Involved in Copper Metabolism and Assembly of Cytochrome Oxidase*

Mutations in the COX17 gene of Saccharomyces cerevi- siae cause a respiratory deficiency due to a block in the production of a functional cytochrome oxidase complex. Because cox17 mutants are able to express both the mitochondrially and nuclearly encoded subunits of cyto- chrome oxidase, the Cox17p most likely affects some late posttranslational step of the assembly pathway. A frag- ment of yeast nuclear DNA capable of complementing the mutation has been cloned by transformation of the cox17 mutant with a library of genomic DNA. Subcloning and sequencing of the COX17 gene revealed that it codes for a cysteine-rich protein with a molecular weight of 8,057. Unlike other previously described accessory fac- tors involved in cytochrome oxidase assembly, all of which are components of mitochondria, Cox17p is a cy- toplasmic protein. The cytoplasmic location of Cox17p suggested that it might have a function in delivery of a prosthetic group to the holoenzyme. A requirement of Cox17p in providing the copper prosthetic group of cy- tochrome oxidase is supported by the finding that a cox17 null mutant is rescued by the addition of copper to the growth medium. Evidence is presented indicating that Cox17p is not involved in general copper metabo- lism in yeast but rather has a more specific function in the delivery of copper to mitochondria. Cytochrome c oxidase, the terminal complex of mitochondrial and bacterial respiratory chains (1), makes use of two different types of electron carriers. One is the central iron of the heme A prosthetic groups of cytochromes a and a 3 . Each cytochrome has an associated copper center, composed either of one or two copper ions. The ligands for the two heme A groups are con-tributed

Cytochrome c oxidase, the terminal complex of mitochondrial and bacterial respiratory chains (1), makes use of two different types of electron carriers. One is the central iron of the heme A prosthetic groups of cytochromes a and a 3 . Each cytochrome has an associated copper center, composed either of one or two copper ions. The ligands for the two heme A groups are contributed exclusively by subunit 1 of the enzyme, and those for copper are shared by subunits 1 and 2 (2,3).
Because the synthesis of heme A from protoheme occurs in the mitochondrial inner membrane (4,5), this prosthetic group is made available at the site of its utilization during assembly of the complex. Copper, however, must be imported from the cytoplasm in amounts commensurate with its requirement in mitochondria; most of the metal is probably used for cytochrome oxidase. It is unlikely that intracellular copper, whether cytoplasmic or mitochondrial, exists as a free solute. Instead, copper and other trace heavy metals are more likely to be complexed to storage proteins, carriers, and metalloen-zymes. The routes used for the intracellular distribution of copper in Saccharomyces cerevisiae are only now beginning to be discerned. External copper is imported into yeast by an ATP-dependent pump encoded by CTR1 (6). One of the better understood routes of internalized copper uses the products of CCC2 (7) and FET3 (8). CCC2 has been shown to code for a cytoplasmic protein of yeast that transfers copper to a membrane-bound ceruloplasmin-like oxidase encoded by FET3 (7,8). Mature Fet3p is a constituent of the cytoplasmic membrane and is distinct from the high-affinity iron transporter but is essential for iron uptake in yeast (6 -8). Even though mutations in CCC2 produce a respiratory defective phenotype, it is not clear whether this is because Ccc2p also functions on the pathway of copper transport to mitochondria or whether this is secondary to an iron deficiency. That copper delivery by Cccp2 is selective, rather than general, is supported by the observation that the total concentration of copper is not significantly different in ccc2 mutants and in wild-type cells, and that such mutants have normal cytoplasmic copper-dependent superoxide dismutase (7).
In the present study, we have characterized a pet mutant (nuclear respiratory-deficient mutant of yeast) assigned previously to complementation group G74. The failure of the mutant to assemble functional cytochrome oxidase is corrected by high concentrations of exogenous copper, indicating that the lesion limits the availability of copper during assembly of the complex. The responsible gene, referred to as COX17, has been cloned and shown to code for a cysteine-rich cytoplasmic protein. Cox17p is not involved in copper uptake in yeast but rather appears to function in the pathway responsible for copper delivery to mitochondria.

MATERIALS AND METHODS
Yeast Strains and Media-The genotypes and sources of the strains of S. cerevisiae used in this study are listed in Table I. The media used for the growth of yeast have been described elsewhere (4).
Preparation of Yeast Mitochondria and Enzyme Assays-Wild-type and mutant yeast were grown to stationary phase in YPGal (2% galactose, 1% yeast extract, and 2% peptone), and mitochondria were prepared by the procedure of Faye et al. (12), except that Glusulase was replaced by Zymolyase 20,000 (ICN Biomedicals, Inc.) to prepare spheroplasts.
Mitochondrial translation products were labeled with [ 35 S]methionine in the presence of cycloheximide, as described previously (13). Cytochrome spectra of mitochondrial extracts were obtained at room temperature. Cytochrome oxidase activity was measured by following oxidation of ferrocytochrome c at 550 nm (13).
Cloning of the COX17 Gene-The wild-type COX17 gene was cloned by transformation of the cox17 mutant C129/U1 by the method of Schiestl and Gietz (14). The library used for the transformation was constructed from partial Sau3A1 fragments of nuclear DNA (averaging 7-10 kb 1 ) cloned into the BamHI site of the shuttle vector YEp24 (15). This library was kindly provided by Dr. Marian Carlson (Department of Genetics and Development, Columbia University). Approximately 5 ϫ * This work was supported by Research Grant GM50187 from the National Institutes of Health, United States Public Health Service. 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.
‡ Recipient of a Medical Research Council of Canada postdoctoral fellowship.
Construction of the COX17-BIO Fusion Gene-To create a gene expressing Cox17p with a biotinylation site at its carboxyl terminus (16), the termination codon of COX17 was destroyed and replaced by a PstI site. The two primers used to amplify the modified gene were each 24 nucleotides long and included the sequence from Ϫ99 to the termination codon. The amplified fragment was digested with a combination of HindIII (site upstream of the coding sequence) and PstI and cloned into YEp352-Bio7 (11). The resultant construct consisted of 99 nucleotides 5Ј to the initiation codon plus the entire COX17 coding sequence fused in-frame to the 270-nucleotide fragment coding for the in vivo biotinylation signal sequence. Since the modified gene containing only 99 nucleotides of the 5Ј sequence was not efficiently expressed, the construct was further modified by the addition of a 750-base pair HindIII-EcoRI fragment from pG74/ST4 (Fig. 4). The newer construct with 550 nucleotides of upstream sequence was also transferred to the integrative vector YIp352 (17).
Miscellaneous Procedures-Standard procedures were used for the preparation and ligation of DNA fragments and for transformation and recovery of plasmid DNA from Escherichia coli (18). The preparation of yeast nuclear DNA and the conditions for the Southern hybridizations were as described by Myers et al. (19), except that probes were labeled by random priming (20). DNA was sequenced by the method of Maxam and Gilbert (21). Proteins were separated by polyacrylamide gel electrophoresis in the buffer system of Laemmli (22). Immunodetection of proteins on Western blot was carried out with 125 I-labeled protein A (23). Protein concentrations were determined by the method of Lowry et al. (24).

Phenotype of cox17
Mutants-C129 is a single mutant of complementation group G74 of our pet mutant collection (10). The phenotype of this mutant suggests that the respiratory deficiency stems from a specific defect in cytochrome oxidase. The mutant shows a substantial decrease in cytochromes a and a 3 but not in cytochrome b (Fig. 1A). The cytochrome oxidase deficiency was also confirmed by direct measurements of enzyme activity (data not shown). The respiratory deficiency in C129 is complemented by a o (cytoplasmic petite mutant lacking mitochondrial DNA) tester strain indicating that the cytochrome oxidase defect is the result of a recessive mutation in a nuclear gene. This gene has been named COX17, in keeping with the nomenclature adopted previously in designating nuclear genes required for expression of yeast cytochrome oxidase (11).
The low level of cytochrome oxidase seen in the spectrum of C129 is a consequence of the C129 mutant allele. This is supported by the phenotype of a strain containing a null mutation in the COX17 gene, which lacks any detectable a-type cytochromes ( Fig. 1A) as well as cytochrome oxidase activity. In vivo labeling of the mitochondrially encoded subunits indicated that their synthesis is not affected in the mutants (data not shown). Western blot analysis of the subunits synthesized in the cytoplasm also precludes an involvement of the gene in expression or import of these proteins (Fig. 1B). These results indicated that Cox17p was likely to be involved in the addition/ synthesis of a prosthetic group or in the assembly of the subunits.
Cloning and Sequencing of COX17-To clone the gene carrying the mutation in C129, a derivative strain, C129/U1, with a ura3 mutation, was transformed with a yeast genomic library as described under "Materials and Methods." Three of the uracil-independent and respiratory-competent clones obtained from the transformation were found to have plasmid with related inserts of yeast nuclear DNA. The region responsible for complementation of C129/U1 was mapped to a 700-base pair   1. Phenotype of cox17 mutants. A, spectra of W303-1B (Wild-Type), the cox17 mutants C129 (cox17-1) and W303⌬COX17 (cox17::TRP1), and W303⌬COX17/ST16 (cox17::TRP1 ϩ COX17-BIO). Mitochondria from the wild-type and mutant cells were extracted with deoxycholate to solubilize all the cytochromes (13). Difference spectra of reduced versus oxidized extracts were recorded at room temperature. The absorption ␣-bands of cytochromes a, a 3 (aa 3 ), cytochrome b (b), and cytochromes c, c 1 (c,c 1 ) are indicated by the arrows. B, Western blot analysis of cytochrome oxidase subunits in mitochondria of W303-1B (Wild-Type) and the cox17 mutants C129 (cox17-1) and W303⌬COX17 (cox17::TRP1). Total mitochondrial proteins (20 g of protein) were separated on a 15% polyacrylamide gel containing 15% glycerol (11). After transfer to nitrocellulose, cytochrome oxidase subunits were detected either with subunit-specific antisera or with antiserum to the holoenzyme (11). Right, subunits 1-8 of cytochrome oxidase.
XbaI-NheI fragment (Fig. 2). The sequence of this region, obtained from both DNA strands, is shown in Fig. 3. A search in GenBank indicated that the cloned fragment overlaps with a sequence reported previously to code for an ATP-dependent RNA helicase (25). The sequence from NheI to the XbaI site ( Fig. 4) contained four open reading frames (orfs 1-4) of sufficient length to qualify as the complementing gene. The four frames overlap with one another and are located near the NheI site. Two of the frames (orfs 1 and 2) are encoded in the same strand as the helicase, while the other two (orfs 3 and 4) are in the opposite strand (Fig. 4). The candidacy of orf4 as the complementing gene was excluded by subclone pG74/ST2, which contained only this frame in its entirety but failed to complement. Neither pG74/ST2 nor any of the other subclones, however, discriminated among the other three possible genes. To distinguish which of the three was responsible for the complementation, two additional subclones were obtained by PCR amplification. Only pG74/ST8 complemented C129/U1. The failure of pG74/ST7 to complement the mutant eliminated orf1 as the gene.
The almost complete overlap of orf2 and orf3 made further subcloning impractical. To identify which of the two reading frames encodes the gene of interest, each frame was fused to a sequence containing the biotinylation site of a bacterial transcarboxylase (see "Materials and Methods"). Since this sequence is also recognized as a biotinylation signal by yeast, it was possible to determine whether both or only one of the two frames is expressed. Multicopy plasmids containing the fusion genes were used to transform a strain with a disrupted copy of orf2 and orf3 (W303⌬COX17; see below). Single transformants verified to harbor each fusion construct were grown up under derepressed conditions, and both the mitochondria and postmitochondrial supernatant fractions were assayed for the presence of a novel biotinylated protein (Fig. 5). The results of these analyses failed to show the presence of a new biotinylated protein in either fraction of the transformant containing orf2 fused to the biotinylation signal sequence. The transformant harboring the orf3 fusion, however, had an abundant novel biotinylated protein of M r 16,000. This product was found exclusively in the postmicrosomal supernatant fraction. The size of the new biotinylated protein is in agreement with the expected size of the protein expected to be translated from the orf3 fusion gene. The finding that only the orf3 fusion was capable of being expressed in vivo provided strong evidence for the identity of this frame as the COX17 gene.
COX17 encodes an acidic protein of 69 amino acids with a high content of cysteine (10%) residues (Fig. 3). The sequence is very hydrophilic, consistent with its properties. The mutation in C129 was determined by sequence analysis of two independent clones obtained by PCR amplification of nuclear DNA prepared from this mutant. A comparison of the PCR-generated sequences with that of the wild-type gene indicated a single mutation at codon 57 of the gene, resulting in a cysteine to tyrosine substitution.
In Situ Disruption of COX17-A null allele of COX17 was created by insertion of the yeast TRP1 gene at the EcoRI site internal to the coding sequence (Fig. 6). The disrupted allele was isolated on a linear fragment and was used to transform the respiratory-competent haploid strains W303-1A and W303-1B. Two respiratory deficient and tryptophan prototrophic clones (W303⌬COX17 and aW303⌬COX17) obtained from the transformation were verified by genomic Southern analysis to have acquired the disrupted cox17::TRP1 allele (Fig. 6). The respiratory-deficient phenotype of the two transformants was complemented by o strains but not by C129. W303⌬COX17 strains exhibit a specific cytochrome oxidase deficiency similar to C129 (Fig. 1).
Localization of a Biotinylated COX17 Fusion Protein-As indicated above, the biotinylated product was detected in the postmicrosomal supernatant but not in mitochondria, suggesting that Cox17p is a cytoplasmic protein. Since the localization of Cox17p was done in a transformant containing the fusion gene on a multicopy plasmid, the possibility existed that overproduction of the protein interfered with its import into mitochondria. This was excluded by transformation of a COX17 null mutant with the same construct in an integrative vector (pG74/ ST17). The presence in the transformant W303⌬COX17/ST17 of the COX17-BIO gene, in a single copy at the URA3 locus, restored normal growth on nonfermentable carbon sources, indicating that the presence of the carboxyl-terminal extension with biotin does not significantly affect the activity of the protein (see also the spectrum in Fig. 1A).
Analysis of the distribution of biotinylated Cox17p expressed from the chromosomally integrated COX17-BIO fusion also showed the protein to be present exclusively in the postmicrosomal supernatant fraction (Fig. 5). This fraction consists not only of the soluble cytoplasmic proteins of yeast but also contains proteins released from nuclei as a result of the fractionation procedure. A nuclear localization of the Cox17p was excluded since purified nuclei did not contain any biotinylated Cox17p (data not shown).

Rescue of cox17 Mutants by High Concentrations of Exogenous Copper-
The cytoplasmic localization of Cox17p precluded a chaperone-like function for this protein. The presence of near normal levels of the mature, cytoplasmically synthesized proteins in the mutant also strongly argue against a role of Cox17p in expression of this set of cytochrome oxidase subunit polypeptides. A cytoplasmic protein, however, could affect production of the functional complex if it were required for the synthesis or mitochondrial import of a prosthetic group or cofactor. The two known electron carriers of cytochrome oxidase are heme a and copper (1)(2)(3). Since heme a is synthesized in mitochondria (5), it is unlikely that Cox17p is involved in this process. The alternative possibility that Cox17p might be involved in copper metabolism, was tested by examining the effect of copper on growth of cox17 mutants on nonfermentable substrates as carbon sources. As shown in Fig. 7A, growth of the null strain W303⌬COX17 was restored on ethanol/glycerol when the medium was supplemented with 0.4% copper. This effect was also seen at lower copper concentrations (0.1%), but the cells grew more slowly. Higher concentrations of copper (0.8%) were lethal to both the mutant and the wild-type strain. Even though CTR1 (structural gene for the copper transporter) on a high copy plasmid does not suppress the respiratory defect of the cox17 null mutant, it does substantially lower the con-  Fig. 4) fused in-frame to a sequence encoding the bacterial biotinylation signal were introduced into the cox17 null mutant W303⌬COX17. Mitochondria and postribosomal supernatant fractions were isolated from W303⌬COX17/ST9 (transformant with frame 2 fusion) and from W303⌬COX17/ST10 (transformant with frame 3 fusion). Total proteins (20 g) of each fraction were separated on a 15% polyacrylamide gel, transferred to nitrocellulose, and biotin-containing proteins were visualized with avidin-conjugated peroxidase (11). EcoRI (E) sites are indicated on the map. W303⌬COX17 and aW303⌬COX17 are two tryptophan-independent clones obtained by transformation of W303-1B and W303-1A, respectively, with the linear HindIII fragment of DNA containing the disrupted gene (26). Chromosomal DNA purified from each tryptophan prototrophic transformant (lanes 1 and 3) and from the parental strains W303-1B (lane 2) and W303-1A (lane 4) were digested with HindIII, separated on a 1% agarose gel, and transferred to nitrocellulose. The blot was hybridized with the 1.4-kb HindIII fragment labeled with 32 P by random priming. The probe detects the expected single 1.4-kb HindIII fragment in the parental strains and two new fragments of approximately 1.6 and 1.2 kb in the mutants, consistent with the restriction map of the cox17::TRP1 disrupted allele. Upper right, the migration of known DNA standards. centration of copper (0.01%) needed to support growth on nonfermentable substrates (Fig. 7B). The ctr1 mutant and the cox17 mutant transformed with CTR1 on a high copy plasmid (W303⌬COX17/ST19) both acquire respiratory competency in the presence of 0.01% copper in the medium (Ref . 6; Fig. 7). The cox17 transformant, however, grows more slowly.
The ability of copper to rescue the deficiency of the cox17 mutation is not general to cytochrome oxidase mutants. Copper supplemented media failed to elicit growth of a large number of different cytochrome oxidase mutants, including the sco1 mutant shown in Fig. 7. The SCO1 gene has been shown to be required for a posttranslational step during cytochrome oxidase assembly (27).
Is Cox17p Required for Maturation of Cytoplasmic Superoxide Dismutase?-The ability of copper to reverse the cytochrome oxidase defect in cox17 mutants strongly supported the idea that the encoded protein played an important role in delivery of copper to mitochondria. It was not excluded, however, that Cox17p might have a more general function in making copper available to other copper-bearing proteins of yeast such as the cytoplasmic superoxide dismutase. The cytoplasmic enzyme is known to use copper as its prosthetic group (28). The superoxide dismutase activity was compared in wild-type and in several different pet strains, including the cox17 and ctr1 null mutants (Fig. 8). The ctr1 mutant, because of its impaired copper transport, fails to make active cytoplasmic superoxide dismutase (30). The absence of cytoplasmic superoxide dismutase in the ctr1 strain is confirmed by the results shown in Fig. 8. In contrast, the cox17 mutant shows the presence of superoxide dismutase activity, although it is somewhat lower than in the wild-type parent or in the mutant transformed with COX17. This partial decrease, also observed in another cytochrome oxidase deficient mutant (cox14), is probably a secondary effect of decreased cellular ATP in respiratory-defective mutants.

DISCUSSION
The synthesis of cytochrome oxidase in S. cerevisiae is a complex process requiring the expression not only of the nuclearly and mitochondrially encoded subunits of the enzyme but also numerous other nuclear genes that code for factors involved in the processing of mitochondrial cytochrome oxidasespecific pre-mRNAs (31,32), translation of the mitochondrial mRNAs for subunits 2 and 3 (33,34), and the assembly process itself, which to date remains poorly understood (11,27,35,36). The COX17 gene reported here does not appear to be important for either synthesis, import, or processing of the subunits but rather fits into the category of genes whose products intercede during the late stages of cytochrome oxidase assembly. There are currently half a dozen examples of such proteins (11,27,35,36). The COX17 gene described here is unusual because its FIG. 7. Restoration by copper of growth of cox17 mutants on nonfermentable substrates. A, the wild-type strain W303-1B (WT), respiratory-defective mutants W303⌬COX17 (⌬COX17) and W303⌬SCO1 (⌬SCO1), and the transformant W303⌬COX17/ST8 (⌬COX17/ST8) were streaked on YPD (glucose medium containing yeast extract and peptone). The YPD master was replicated on YEPG (glycerol plus ethanol medium containing yeast extract and peptone) and on YEPG supplemented with 0.4% copper sulfate. The growth of W303⌬COX17 on copper-supplemented YEPG required at least 4 days of incubation at 30°C. B, the wild-type strain W303-1B (WT), the mutants W303⌬COX17 (⌬COX17) and YSC5#2 (⌬CTR1), and the transformant W303⌬COX17/ST19 (⌬COX17/ST19) were replicated on YEPG and YEPG supplemented with 0.01% copper sulfate. Growth of YSC5#2 on the copper-supplemented medium was discerned after 1 day incubation at 30°C, whereas growth of the transformant required an additional 2-3 days of incubation. product is a cytoplasmic protein. In contrast, all of the other genes currently known to affect cytochrome oxidase assembly code for proteins that are located in the mitochondrial inner membrane, where assembly occurs.
The cytoplasmic localization of the COX17 product provided the initial clue that cox17 mutants might be deficient in mitochondrial copper. This is supported by the following observations. Copper-supplemented media can support growth of a cox17 null mutant on nonfermentable carbon sources. The concentration of copper required to rescue cox17, however, is at least one order of magnitude higher than that needed to promote growth of ctr1 mutants, which have lesions in the copper pump. Although the concentration of copper in the medium can be substantially reduced when the CTR1 gene is overexpressed in the cox17 mutant, the transformant nonetheless grows more slowly than ctr1 mutants under the same conditions. The differences in the copper-dependent growth of the two mutants suggests that the lesion in the cox17 strain is not related to copper uptake. The presence in cox17 mutants of cytoplasmic superoxide dismutase, the activity of which depends on copper (28), provides additional evidence that Cox17p is not essential for copper uptake. Since the cox17 mutation appears to preferentially affect a copper enzyme of mitochondria, we propose that Cox17p functions on the pathway responsible for the delivery of copper to mitochondria (Fig. 9). Externally high concentrations of copper probably compensate for the absence of Cox17p by allowing the metal to passively diffuse into mitochondria as a result of higher concentrations of the metal in the cytoplasm. The primary sequence of Cox17p suggests that it may complex copper targeted for mitochondria through its cysteine residues, which make up 10% of the protein. The partial loss of function as a result of the Cys57 3 Tyr substitution in C129 is consistent with this notion.
The importance of heme A and copper as obligatory electron carriers of cytochrome oxidase is well established (1-3). Less appreciated, however, are more recent findings indicating that the heme A and copper prosthetic groups probably also make a structural contribution by stabilizing the final assembled complex. In yeast, mutations interfering with heme A biosynthesis have a profound effect on cytochrome oxidase (4,5). The reduction in the steady-state levels of some subunits in such mutants suggests that heme A protects the holoenzyme, or partially assembled intermediates, against proteolytic degradation (4).
Recent studies indicate that mutations in the copper ligands of subunit 2 also destabilize the enzyme (37). This implies that mutations affecting the availability of copper for cytochrome oxidase synthesis should also increase turnover of the less stable subunits. Western analysis has revealed cox17 mutants to have severely reduced steady-state concentrations of subunits 1 and 2, a hallmark of mutants arrested in cytochrome oxidase assembly (11, 27, 34 -36). This further supports the notion that copper contributes to the structural stability of the holoenzyme.