Characterization of YpmQ, an accessory protein required for the expression of cytochrome c oxidase in Bacillus subtilis.

A search of the Bacillus subtilis genome identifies a potential homolog, ypmQ, of the inner mitochondrial membrane protein Sco1 from yeast. Sco1 has been found to aid the delivery of copper to cytochrome c oxidase. B. subtilis expresses two members of the cytochrome oxidase family, a cytochrome c oxidase that has two copper centers, Cu(A) and Cu(B), and a menaquinol oxidase that has only Cu(B). Deletion of ypmQ in B. subtilis depresses expression of cytochrome c oxidase but not menaquinol oxidase. Levels of cytochrome c oxidase recover when copper is added to the growth medium of the DeltaypmQ strain or when ypmQ is expressed from a plasmid. Neither treatment affects the amount or activity of menaquinol oxidase. YpmQ in which two conserved cysteines are replaced by serines and a conserved histidine is replaced by alanine do not complement the deletion of ypmQ even though these mutant forms are found in the membrane extract at a level similar to the wild type protein. We propose that the two cysteines and the histidine are critical for the function of YpmQ and suggest they are involved in copper exchange between YpmQ and the Cu(A) site of cytochrome c oxidase.

A search of the Bacillus subtilis genome identifies a potential homolog, ypmQ, of the inner mitochondrial membrane protein Sco1 from yeast. Sco1 has been found to aid the delivery of copper to cytochrome c oxidase. B. subtilis expresses two members of the cytochrome oxidase family, a cytochrome c oxidase that has two copper centers, Cu A and Cu B , and a menaquinol oxidase that has only Cu B . Deletion of ypmQ in B. subtilis depresses expression of cytochrome c oxidase but not menaquinol oxidase. Levels of cytochrome c oxidase recover when copper is added to the growth medium of the ⌬ypmQ strain or when ypmQ is expressed from a plasmid. Neither treatment affects the amount or activity of menaquinol oxidase. YpmQ in which two conserved cysteines are replaced by serines and a conserved histidine is replaced by alanine do not complement the deletion of ypmQ even though these mutant forms are found in the membrane extract at a level similar to the wild type protein. We propose that the two cysteines and the histidine are critical for the function of YpmQ and suggest they are involved in copper exchange between YpmQ and the Cu A site of cytochrome c oxidase.
Cytochrome oxidases are integral membrane protein complexes that catalyze the reduction of oxygen to water and capture some of the redox free energy of this reaction as a transmembrane electrochemical gradient. The key structural features that are shared among all members of this family of enzymes are found in its largest subunit, subunit I. Subunit I is an integral membrane protein with 12-14 membrane-spanning helical segments that provide binding sites for two heme A moieties, known as cytochrome a and cytochrome a 3 , and one copper center, Cu B . Cytochrome a 3 sits in close proximity to Cu B , and together they form a binuclear site that is responsible for binding oxygen and its partially reduced states, which arise transiently in the course of catalysis. Cytochrome a is a low spin heme that functions to deliver electrons to cytochrome a 3 -Cu B (1).
The nature of the electron input site reflects a division in the cytochrome oxidase family of enzymes into two groups. In cytochrome c oxidases, such as the enzyme found in the mitochondrial inner membrane of eukaryotes, reducing equivalents are delivered from the soluble protein ferrocytochrome c. The cytochrome c interaction site on the oxidase is predominantly defined by subunit II (2,3). Subunit II has two transmembrane helices that anchor a solvent-exposed domain, which provides the inner sphere ligands for the dinuclear Cu A center. Although there is no direct structural information on a cytochrome ccytochrome c oxidase complex, kinetic (4,5), mutagenic (6,7), and modeling (8) studies indicate that cytochrome c binds at a site near the Cu A center to allow for efficient electron transfer from cytochrome c to Cu A . Thus, electrons enter cytochrome c oxidase via Cu A and are transferred to the cytochrome a 3 -Cu B center through cytochrome a (9). The second group within the cytochrome oxidase family is the quinol oxidases, which receive reducing equivalents from a lipid-soluble quinol. The best known of this group is the ubiquinol oxidase from Escherichia coli (10). Even though the quinol oxidases do not oxidize cytochrome c they do have a subunit II that has overall homology with the subunit II of the cytochrome c oxidases. The major difference in subunit II of the quinol oxidase is the lack of the amino acid ligands for the Cu A center. The lack of Cu A accounts for the lack of reactivity of the quinol oxidases with cytochrome c.
Copper is an element that is used in proteins to fulfill specific catalytic and structural roles. However, copper is also a potential danger in biological systems due to its ability to catalyze oxidative damage of many cellular components. A new class of proteins, metallochaperones, have been identified that mediate the incorporation of copper into a variety of specific binding sites (11). As outlined above, the integral membrane enzyme cytochrome c oxidase has two biochemically and physically distinct copper centers. The Cu A center is composed of two copper ions that are held by a set of amino acid ligands such that the coppers are within bonding distance from one another (12). The second copper center of cytochrome c oxidases is known as Cu B and is physically associated with cytochrome a 3 . Cu A is contained in the extra-membranous domain of subunit II, whereas Cu B is found in the membrane-embedded domain of subunit I (13,14). There is much known about the role of these copper centers in the catalytic cycle of the enzyme, but relatively little is known about the mechanism of their assembly.
A number of protein factors have been proposed to have a role in the assembly of complex integral membrane proteins such as cytochrome c oxidase (15). In the last few years some of these assembly factors have been more specifically defined. A pathway for the import of copper into mitochondria and its assembly into the copper centers of cytochrome c oxidase has been proposed from studies in yeast (16). A pair of copper transporters, CTR1 (17) and CTR3 (18), have been found in the plasma membrane that are responsible for the specific, high affinity uptake of copper. Copper is then passed to a set of soluble binding proteins, one of which is Cox17. Cox17 binds copper in the cytoplasm and moves to the mitochondrial intermembrane space by an as yet unknown mechanism. Another protein Sco1, which is an integral component of the inner mitochondrial membrane, has been implicated in copper delivery to cytochrome c oxidase because yeast strains made defi-* 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. cient in Cox17 are rescued by overexpression of Sco1 (19). Whether Cox17 passes copper to cytochrome c oxidase, or to Sco1, is not known. Sco1 may play a direct role in the transfer of copper to cytochrome c oxidase perhaps acting to facilitate the transfer of copper from Cox17 to cytochrome c oxidase. Alternatively, the role of Sco1 may be indirect, perhaps serving as a means of copper accumulation in mitochondria (20).
In order to address questions concerning the specific requirements for the assembly of the Cu A and Cu B centers in cytochrome c oxidase, we chose to look at their assembly in the respiratory chain of the aerobic bacterium Bacillus subtilis. B. subtilis expresses a cytochrome c oxidase that, like the canonical mitochondrial enzyme, has both Cu A and Cu B centers within homologs of subunit II and subunit I, respectively. The cytochrome c oxidase of B. subtilis has the added feature of having a cytochrome c domain fused to its subunit II, and this places it among the cytochrome caa 3 subclass of the cytochrome oxidase family. In addition, B. subtilis expresses a menaquinol oxidase that contains cytochrome a, cytochrome a 3 , and Cu B within a subunit I that is homologous (47.6% identical, 76.9% similar) to subunit I of cytochrome caa 3 . The menaquinol oxidase does not, however, have a Cu A center, but subunit II of menaquinol oxidase is homologous (62% similar) to subunit II of cytochrome caa 3 . The particular amino acids used to form the Cu A site are absent in the sequence of B. subtilis menaquinol oxidase (21). The presence of these two related members of the cytochrome oxidase family with distinct copper requirements make B. subtilis a good organism for studying the requirements of copper center assembly.
A search of the B. subtilis chromosome reveals a potential homolog of the yeast mitochondrial protein Sco1 but does not identify a homolog of Cox17. Here, we show that disruption of the expression of ypmQ results in cytochrome c oxidase-deficient B. subtilis. In contrast, the expression of menaquinol oxidase is unaffected in the same strains. We show that the absence of ypmQ from the genome can be overcome by supplementing the growth medium with copper or by expressing ypmQ from a plasmid. We have made site-directed changes to two conserved cysteine residues and one histidine within the sequence of YpmQ that support a role for these residues as potential copper-binding ligands. It is proposed that YpmQ is involved in the assembly of the Cu A center but is not involved in the assembly of the Cu B center.

EXPERIMENTAL PROCEDURES
Bacterial Strains and Cultivation Conditions-B. subtilis and Escherichia coli strains used are shown in Table I. All E. coli strains were grown on LB medium at 37°C; LB agar also contained 1.5% agar (Difco). All B. subtilis strains were grown on super rich medium (22) or tryptose blood agar base (Difco) at 37°C. Antibiotic concentrations for phenotype testing and routine growth were as follows: 100 g/ml ampicillin, 5 g/ml chloramphenicol, 10 g/ml kanamycin, 10 g/ml tetracycline, 7 g/ml Nm, 1 100 g/ml spectinomycin, and 75 g/ml streptomycin. Antibiotics and other chemicals were from Sigma unless otherwise noted.
Cloning of ypmQ-The ypmQ gene was amplified by PCR from B. subtilis 168 genomic DNA template by using primers 1 and 2 (Table II),  (23) and was used to transform E. coli DH5␣ (Life Technologies, Inc.) to ampicillin resistance, creating pNM87. The ypmQ-containing fragment of pNM87 was restricted with MluI and SpeI (Fig. 1A), filled-in with T4 polymerase and dNTPs, and subcloned into shuttle plasmid pHP13 (24) that was restricted with SmaI. This produced pNM137 that was used to transform E. coli to chloramphenicol resistance. pNM137 was sequenced (Cortec, Queen's University, Kingston, Ontario, Canada) and found to be identical to the published chromosomal sequence (25) with the exception of a single bp change of guanine 254 to adenine in ypmQ, resulting in a change from glycine to glutamate at position 85 in the predicted amino acid sequence. This change has been confirmed by sequencing the amplicon directly. pNM137, along with other transformants tested, contained ypmQ in opposite orientation to the lac promoter in pHP13, even though both orientations should have resulted from the blunt-ended ligation. It has been our experience that plasmid constructions with genes that oppose the lac promoter in pHP13 are sometimes problematic in B. subtilis. In order to place ypmQ in the same orientation as lac, we used pNM137 in a triple ligation (26) to construct pNM171. pNM171 contained ypmQ in the same orientation as lac but could only be isolated in E. coli DH5␣ under glucose-induced catabolite repression or in E. coli strain TP611 (27) that is deficient in adenylate cyclase. These results suggest that the leakiness of the lac promoter allows for some expression of ypmQ, and this is toxic to E. coli. Chromosomal Deletion of ypmQ-To construct a chromosomal deletion of ypmQ, a 523-bp deletion of pNM87 was created within ypmQ by restriction with HpaI and EcoRV (Fig. 1A), followed by insertion of a Nm resistance cassette, which is terminated by the Lambda t o -terminator, from pNM46. Plasmid pNM209 was isolated with the Nm resistance cassette in the same orientation as ypmQ; this plasmid was linearized and used to transform B. subtilis 168 to Nm resistance, resulting in strain BH144. A double-reciprocal recombination that deleted ypmQ was confirmed by amplification of BH144 genomic DNA (data not shown).
Mutagenesis of ypmQ-pNM171 was used as template in an inverse PCR (28) that amplified ypmQ with an added carboxyl-terminal 6-histidine tag by using primers 3 and 4 (Table II) and Pfu polymerase. pNM175 was isolated and sequenced, which confirmed that it contained the codons for the histidine tag. Cysteines at positions 64 and 68 of YpmQ ( Fig. 1) were mutated to serine by using pNM175 as template with primers 5 and 6, 5 and 8, and 6 and 7 (Table II) in three separate inverse PCRs. pNM207, pNM315, and pNM324 were isolated, sequenced, and shown to be C64S/C68S, C68S, and C64S, respectively. Histidine 154 of YpmQ was changed to alanine ( Fig. 1) using primers 9 and 10 (Table II) in an inverse PCR to create plasmid pNM305, which was sequenced and shown to be H154A. BH144 was transformed with pNM171, pNM175, and pNM207, pNM305, pNM315, and pNM324 to chloramphenicol resistance and produced strains BH149, BH150, BH148, BH180, BH183, and BH193, respectively.
Chromosomal Deletion of the Qox Operon-The qox operon was cloned from pDIA5340 (gift from P. Glaser, Institut Pasteur, Paris, France) using EcoRV restriction sites into pNM34, which created pNM35. The qox operon, carried by pNM35, was isolated from E. coli dam-deficient strain GM2163 (29) and was restricted with XbaI and BamHI, and the spectinomycin resistance cassette from pIC156 (30) was inserted in the same orientation as qox, creating pNM42, which deleted almost all of the operon. pNM42 was linearized and used to transformed B. subtilis 168 to spectinomycin resistance. BH101 was isolated and shown to be ⌬qox by amplification of genomic DNA.
TMPD Plate Assay-This assay was performed according to the method described by Mueller and Tabor (31). The B. subtilis colony to be tested was grown on tryptose blood agar base. The cells were frozen at Ϫ20°C for 30 min and then returned to room temperature and fixed in place with hair spray. The cells were overlaid with 5.2 ml of a TMPD/ agar solution containing 0.6% (w/v) agar, 2% Triton X-100, 0.02 M potassium phosphate, pH 7.4, 1% sodium deoxycholate, 20% ethanol and 0.1% (w/v) TMPD. The TMPD was added just prior to applying the agar to the cell culture.
Western Blots with Anti-His Tag and Anti-caa 3 Antibodies-Membrane proteins of B. subtilis were extracted in Triton X-100 as described (22). Extracted proteins were separated by SDS-PAGE (32) and blotted to PVDF membrane using a mini trans-blot cell (Bio-Rad). The preparation and detection of the blotted membrane was done using a chemiluminescent method as described by the manufacturer (NEN Life Science Products). The secondary antibodies that were used for detection were either the tetra-His antibody (Qiagen, Mississauga, Ontario, Canada) or anti-caa 3 antibodies. Anti-caa 3 antibodies were obtained against the highly purified, two-subunit form of the protein. The purified protein after solubilization with SDS was used to immunize mice. Quantification of the bands on these gels was done using the program ScnImage (Scion Corp., MD).
Assays of Cytochrome c Oxidase and Menaquinol Oxidase Activities-Membrane extracts were assayed for cytochrome c oxidase activity using the substrates ascorbate plus TMPD. Activity was monitored spectrophotometrically by observing the level of reduction of cytochrome aa 3 at 444 -460 nm as a function of time. The assay was initiated by addition of the reductants to the air-saturated buffer containing the membrane extract. Aliquots of the same extracts were assayed for menaquinol oxidase activity using 50 M of the substrate analog 2,3-dimethyl-1,4-naphthaquinone (St. Sava-Bioanlytical & Medical Research, Kingston, Ontario, Canada). 2,3-Dimethyl-1,4-naphthaquinone reduction by 2 mM NADH was catalyzed by 2 mg/ml DTdiaphorase. Absorption spectra were monitored with a Hewlett-Packard diode array, and the temperature was maintained at 20°C.

RESULTS
A Blast2 (33) search of the protein data base derived from the B. subtilis genome (25), using the sequence of the cytochrome c oxidase assembly protein Sco1 from yeast, identifies a possible homolog, YpmQ (Fig. 1, A and B). The same approach does not identify a homolog of the yeast copper chaperone Cox17 in the B. subtilis genome. Analysis of the ypmQ gene finds two potential promoters, A and H (34). The H promoter is probably the most efficient as it has perfect consensus for promoters of this type. Also, H is produced late in growth, which correlates well with the appearance of cytochrome c oxidase (35). YpmQ is predicted to have a molecular mass of approximately 21.5 kDa and, from hydropathy analysis (36), to contain a single aminoterminal transmembrane-spanning helix, suggesting that it is a membrane-bound protein. Further examination of the predicted transmembrane helix shows that it also contains the sequence of amino acids that specify the covalent attachment of lipid (37). The alignment of the solvent exposed domains of YpmQ and Sco1 shows 27% identity and 48% similarity (Fig.  1B). This alignment identifies a CXXXC sequence and a histidine in YpmQ that are conserved elements compared with Sco1.
The ypmQ gene was cloned from the chromosome of B. subtilis using PCR to yield pNM87. A Nm antibiotic resistance cassette was used to replace most of the ypmQ gene in pNM87 to make pNM209. Since this construct was made using pBR322, which does not replicate independently in B. subtilis, transformation of B. subtilis with pNM209 requires homologous recombination with ypmQ on the chromosome to rescue the Nm cassette. Strain BH144 is Nm-resistant and ⌬ypmQ. A plate assay based on the production of the intensely colored Wurster's blue radical that results from TMPD oxidation by cytochrome c oxidase was used to test B. subtilis colonies for the presence of cytochrome c oxidase activity (31). It is impor- YpmQ, an Assembly Factor for Cytochrome c Oxidase tant to note that TMPD is a poor electron donor to menaquinol oxidase, the other heme A-based terminal oxidase present in membrane extracts of B. subtilis (38). Fig. 2 shows the color that develops in this assay with a number of B. subtilis strains. A strong purple color develops in wild type B. subtilis 168 and is completely absent in strain BH170 in which the expression of cytochrome c oxidase is disrupted directly by insertional inactivation of the cytochrome c oxidase operon ( Fig. 2A, lanes 1  and 2). The strain of B. subtilis in which ypmQ is deleted (i.e. BH144) is also unable to oxidize TMPD ( Fig. 2A, lane 3). When ypmQ is supplied to strain BH144 on plasmid pNM171 (i.e. BH149) the ability to oxidize TMPD is restored (Fig. 2A, lane  4). In order to demonstrate that the ability of BH149 to oxidize TMPD is not simply due to an overexpression of ypmQ, we expressed ypmQ in a background deficient in cytochrome c oxidase. This strain, BH191, is unable to oxidize TMPD ( Fig.  2A, lane 5) implying that, as expected, this activity is coupled to the functional expression of cytochrome c oxidase. Another manipulation illustrated in Fig. 2 is the effect of the addition of a 6-histidine tag on the functional expression of ypmQ (BH150). The histidine-tagged version of ypmQ is equal to the wild type protein in its ability to complement ⌬ypmQ (Fig. 2B, lanes 1  and 3).
Since ypmQ was identified by its sequence similarity to Sco1 from yeast, and Sco1 is implicated in copper delivery to cytochrome c oxidase, we decided to test whether the deficiency of cytochrome c oxidase that occurs when ypmQ is deleted can be overcome by supplementing the growth medium with copper. We prepared membrane extracts from wild type B. subtilis, BH144 (⌬ypmQ), in which ypmQ is deleted, and BH150 (⌬yp-mQ[ypmQ ϩ ]), in which the ypmQ deletion is complemented by expression of ypmQ from a plasmid. These three strains were grown under conditions of low copper and are compared with BH144 grown in medium containing high copper (Figs. 3 and 4  and Table III). Our super-rich medium used for routine culturing of wild type B. subtilis contains 0.8 M copper supplied in a micro-nutrient supplement. In the low copper growth condition copper was excluded from the micro-nutrient solution, and the medium was supplemented with 20 nM of the copper chelator bathocuproine disulfonate. In high copper conditions the low copper medium was supplemented with 50 M CuCl 2 , which in separate experiments was shown not to affect the growth of wild type B. subtilis or strain BH150. The steady-state activities for membrane extracts from wild type, BH144, and BH150 grown with low copper are compared with BH144 grown with high copper (Fig. 3). In this assay the steady-state absorption at 444 -460 nm versus time is a measure of both the enzyme activity and the amount of cytochromes a and a 3 . In each trace the oxidized level of the sample was measured prior to addition of the reductants, ascorbate and TMPD, which produce a steady-state level of reduction until the dissolved O 2 is completely consumed at which time the absorption increases rapidly to full reduction. We have shown previously that aerobic cultures of B. subtilis contain two components that are reduced in this assay, namely cytochrome c oxidase and menaquinol oxidase (38). Furthermore, more than 90% of the flux from ascorbate plus TMPD is carried by cytochrome c oxidase. The steady-state cytochrome c oxidase activity of the wild type extract is similar to that reported previously (38). The apparent activity of BH144 grown on low copper media is reduced 25-fold relative to the wild type extract (Table III). This residual activity of BH144 is presumably due to slow turnover of the FIG. 1. Sequence of ypmQ and its relationship to the yeast copper assembly protein Sco1. A, the ypmQ gene is 579 bp and produces a gene product of 193 amino acids. For reference, the 3Ј-end of the upstream gene ypmP is shown and is separated from its stop codon by a space. A potential Rho-independent transcription terminator may stop transcription from ypmR (Ͻ ϽϾ Ͼ). Two putative promoter elements were found upstream of ypmQ, A, and H (underlined). A Shine-Dalgarno sequence occurs eight nucleotides from the start codon of ypmQ (boldface). MluI and SpeI restriction sites, which were used to subclone ypmQ from pNM87 to pNM137, are indicated (double underline). The HpaI and EcoRV restriction sites that were used to delete most of the ypmQ gene in pNM209 and in strain BH144 are also indicated. Amino acids Val-2 to Gly-21 may produce a transmembrane helix (gray box). Cysteine 19 is a possible lipid attachment site. Cysteines 64 and 68 and histidine 154, which are possible ligands for copper, were mutated to serine and alanine (black box). B, a BLAST2 search of the B. subtilis 168 protein data base using Sco1 finds a conserved amino acid core with YpmQ. Cysteines 64 and 68 and histidine 154 are shown (black box). menaquinol oxidase. The activity of BH150 is about 60% of the wild type activity. The addition of extra copper to the growth medium of BH144 stimulates the activity more than 7-fold over the same strain grown with low copper, which brings it to a level about 25% that seen with the wild type preparation.
Table III also shows the total cytochrome aa 3 content for the four conditions mentioned above. In BH144 (⌬ypmQ) grown with low copper the cytochrome aa 3 content in the extract is reduced to about one-half that seen in the wild type extract. The remaining cytochrome aa 3 content is due to the presence of menaquinol oxidase. This change is in keeping with our previous estimate of the cytochrome c oxidase and menaquinol oxidase contents in wild type cells (38) and the loss of cytochrome c oxidase in BH144 (⌬ypmQ). When the growth medium is supplemented with copper the cytochrome caa 3 content of BH144 is increased to 21% that of wild type. In strain BH150 the level of cytochrome caa 3 indicates a recovery to 50% of the wild type level. These cytochrome levels are consistent with the apparent activities, which indicates that to a first approximation the cytochrome c oxidase assembled under these different conditions has close to the same molecular activity. We have also found that the menaquinol oxidase activity is the same in these membrane extracts implying that the lack of YpmQ does not affect its assembly.
In BH144 (⌬ypmQ) grown with low copper the spectral properties and activity of the cytochrome c oxidase are absent. Fig.  4 shows the results of an experiment in which antibodies to a highly purified two-subunit form of the cytochrome caa 3 complex were used to determine whether the subunits of the enzyme can be detected in this strain. Membrane extracts from wild type B. subtilis, strain BH144 (⌬ypmQ), and from BH150 (⌬ypmQ [ypmQ ϩ ]) were run on an SDS gel and transferred to PVDF for Western blotting. The two major subunits, I and II, are clearly observed in the wild type extracts and are both depressed in strain BH144 grown with low copper. Our antibody to cytochrome caa 3 has better reactivity with subunit II, and Table III reports the integrated intensity of the subunit II band corrected for the small variation in the amount of membrane protein loaded in each lane of Fig. 4. The level of subunit II is more than 10-fold lower than wild type in BH144 and recovers to more than 80% of wild type levels in BH150 grown with low copper. In BH144 grown with excess copper the subunit II level is 21% of the intensity observed with wild type. The values for subunit II levels are consistent with the cytochrome content and activity levels of these strains (Table III).
A key feature of ypmQ is a set of highly conserved residues that are proposed to function in binding copper. In YpmQ, the cysteines at positions 64 and 68 and histidine at 154 are conserved when compared with the sequence of Sco1 from yeast (39). We expressed ypmQ from a plasmid with amino acids Cys-64 and Cys-68 replaced by serine both individually and together and His-154 changed to alanine. These strains (i.e. BH193, BH183, BH148, and BH180) are not able to oxidize TMPD (Fig. 2B, lanes 4 -7) and have a phenotype equivalent to the cytochrome c oxidase disrupted strain (i.e. BH170, Fig. 2B  lane 2). Fig. 5 compares the cytochrome content of a B. subtilis strain expressing wild type YpmQ to a strain expressing an inactive mutant of YpmQ (BH180 and H154A). The strain expressing wild type YpmQ has an absorption maximum at 602 nm that is a composite of contributions from the two heme A-containing oxidases (Fig. 5A). In contrast, the strain expressing an inactive mutant of YpmQ has an absorption maximum FIG. 3. Steady-state activity profiles for membrane extracts of different strains of B. subtilis. Each assay contains 5 mM sodium ascorbate and 200 M TMPD and was performed as outlined under "Experimental Procedures." Traces are as follows: trace a is from wild type B. subtilis grown with low copper; trace b is from BH144 (⌬ypmQ) grown with low copper; trace c is from BH150 (⌬ypmQ [ypmQ ϩ ]) grown with low copper; and trace d is from BH144 (⌬ypmQ) grown with high copper.

FIG. 4. Western blot of membrane extracts for cytochrome caa 3 subunit content.
Membrane extracts were prepared, run on SDS-PAGE, electrophoresed to PVDF membrane, and probed with anticytochrome caa 3 antibodies. Lane 1 is a sample from wild type B. subtilis 168 (80.2 g of total protein); lane 2 is from BH144 (⌬ypmQ) grown with low copper (74 g of total protein); lane 3 is BH150 (⌬ypmQ [ypmQ ϩ ]) with low copper (74 g of total protein); and lane 4 is BH144 (⌬ypmQ) grown with high copper (88 g of total protein). at 600 nm, which is characteristic of the menaquinol oxidase. Furthermore, the difference spectrum between the membrane extract of the wild type strain minus the mutant strain shows a deficiency in cytochrome c oxidase (Fig. 5B). The double difference spectrum has a set of bands at 416, 520, and 550 nm due to the cytochrome c domain of subunit II and at 444 and 604 nm due to cytochromes a and a 3 , which are signatures of B. subtilis cytochrome c oxidase. There is also a difference in the level of cytochrome b in the two strains as evidenced by the peak at 560 nm in the double difference spectrum. The lack of functional cytochrome c oxidase in the strains expressing site-specific mutants of YpmQ (see Fig. 2B, lanes 4 -7) could be due to poor expression or improper folding of the mutant forms of YpmQ. To address this question, we have taken advantage of the histidine tag on each of these constructs and used an anti-histidine tag antibody to detect YpmQ. A Western blot of membrane extracts separated by SDS-PAGE shows that histidine-tagged, native YpmQ and His-tagged, mutant YpmQ proteins are present at similar levels in these different strains (Fig. 6). Since the mutants are expressed to a similar level as wild type and are present in the plasma membrane, it is unlikely that the cysteine to serine or the histidine to alanine changes have compromised the proper expression and folding of these mutant YpmQ proteins.
The effect of ypmQ's deletion on menaquinol oxidase function can also be tested by growth and antibiotic sensitivity. It has been shown that a deficiency in menaquinol oxidase leads to a small colony morphology and confers streptomycin resistance to B. subtilis (21,40). We have confirmed that a ⌬qox strain (BH101) in which the menaquinol oxidase operon is deleted is resistant to streptomycin, whereas the wild type strain and BH144 (⌬ypmQ) fail to grow in the presence of streptomycin. We have also observed that BH144 (⌬ypmQ) has a colony morphology that is similar to the wild type strain 168 when both are compared with the small colony phenotype observed for BH101 (⌬qox) (data not shown). Spectral evidence, kinetic activity, and growth phenotype support the conclusion that deletion of ypmQ does not affect the expression of menaquinol oxidase.

DISCUSSION
Study of the assembly of copper centers in a number of copper-containing proteins has resulted in the identification of specific accessory proteins that aid the assembly process. In some cases transfer of copper from the assembly protein to an apoprotein target has been demonstrated, and these proteins are identified as copper chaperones. This new class of proteins functions to deliver copper and avoids the potential deleterious side effects that could arise from redox reactions initiated by free copper. For example, the copper center in copper, zinc, superoxide dismutase in yeast and humans is assembled with the aid of the copper chaperone protein, CCS (41). In the case of mitochondrial cytochrome c oxidase it has been postulated that the protein Cox17 is responsible for delivering copper from the cytoplasm to the intermembrane space of mitochondria where the metal is incorporated into cytochrome c oxidase (20). Yeast that are deficient in cox17 are unable to grow on non-fermentable substrates but can be rescued by addition of copper to the growth medium (42) or by overexpression of the inner mitochondrial membrane protein, Sco1 (19). In contrast, yeast deficient in Sco1 cannot be rescued by addition of copper to the growth medium. Thus, the working model for copper delivery to cytochrome c oxidase has Cox17-binding copper in the cytoplasm, passing through the outer mitochondrial membrane and delivering copper to the intermembrane space (19). Whether copper is passed from Cox17 to cytochrome c oxidase directly, or with the intervention of Sco1, is not known.
In B. subtilis two homologous members of the cytochrome oxidase family are expressed that have different copper contents. The cytochrome c oxidase complex is most similar to the mitochondrial enzyme in having both Cu A and Cu B centers. The menaquinol oxidase does not have Cu A but does have the other redox active metals including Cu B . When the expression of the Sco1 homolog, YpmQ, is disrupted the spectral properties, major subunits, and activity of cytochrome caa 3 are lost without affecting the properties of the menaquinol oxidase. This disruption can be overcome partially when extra copper is added to the growth medium and more fully when native YpmQ is expressed from a plasmid. These data lead us to suggest that YpmQ is involved in the assembly of Cu A but is not involved with Cu B assembly. Furthermore, we suggest that our results support the proposal that the role of Cox17 in yeast is to deliver copper to the mitochondrial compartment but Sco1 mediates copper incorporation into cytochrome c oxidase. B. subtilis does not, therefore, require a homolog of Cox17 because of its simpler prokaryotic cell structure. However, further work is required to demonstrate the direct role of these proteins in copper delivery to cytochrome c oxidase, and we cannot exclude the presence of an analogous protein to yeast Cox17 being present in B. subtilis.
In our work here it is shown that the failure to produce YpmQ results in a loss of the cytochrome c oxidase complex. Previous work on the related quinol oxidase from E. coli shows that deprivation of copper from the growth medium yields an enzyme that has both the low and high spin heme centers but lacks Cu B (43). More recently Hosler and co-workers (44) demonstrate similar effects in cytochrome c oxidase of Rhodobacter sphaeroides when an assembly gene, cox11, is deleted. Deletion of cox11 results in the expression of a cytochrome c oxidase complex that has all the redox metal centers of the native enzyme, including Cu A , but lacks Cu B . Moreover, based on this work and that on the E. coli oxidase, it appears that it is possible to assemble a Cu B -less enzyme and implies that this step occurs late in the assembly of the oxidase complex. In contrast, in our work we do not find a cytochrome c oxidase complex assembled that simply lacks the Cu A center, and this implies that Cu A assembly occurs at an early stage and is required for subsequent steps or for the overall stability of the complex in vivo.
In a recent paper (45) on mutants of Sco1 from yeast, two conserved cysteine residues, which we have studied in YpmQ, were changed to alanine, and these mutants failed to restore respiratory competence. However, as the authors of this work acknowledge, they cannot exclude that the mutant forms of Sco1 were improperly processed in some way and that this may account for the inactivity. However, we demonstrate that our mutant and wild type versions of YpmQ are equally well expressed and are all found in the plasma membrane. We propose, therefore, that the two conserved cysteine residues and, in addition, the conserved histidine play a direct role in the function of YpmQ. If YpmQ is a homolog of the Sco proteins from yeast and human, then the cysteines and histidine are also critical to their function (30). We have generated histidinetagged, wild type YpmQ as well as versions in which each of the putative copper ligating residues has been inactivated. We will isolate each of these proteins and investigate their metal-binding properties to clarify further the role proposed above for these residues in copper binding.
The copper chaperone protein, CCS, has considerable homology to its target protein, superoxide dismutase (46). In fact, recent mutagenesis work shows that CCS can be converted to an active superoxide dismutase by a single amino acid change (47). The structural similarity between CCS and superoxide dismutase is proposed to mediate copper exchange via heterodimer complex formation. In the case of cytochrome c oxidase the copper chaperone Cox17 has been shown to bind two copper ions (48). This could be of significance given that one of its plausible targets, Cu A , is a dinuclear copper center, although the geometry of the two sites is different. Sco1 and YpmQ exhibit sequence homology with a small portion (i.e. 20 amino acids) of the Cu A domain of cytochrome c oxidases that is centered around the pair of cysteine residues of the copperbinding sites (19). Such limited structural similarities between the Cu A site and its possible chaperones, Cox17 and YpmQ or Sco1, do not support a chaperone/target recognition mechanism such as that proposed for the CCS/superoxide dismutase system.
Our results support the proposal that YpmQ is a homolog of the yeast protein Sco1 and that it plays a functional role in the assembly of cytochrome c oxidase in B. subtilis. Deletion of ypmQ disrupts the expression of cytochrome c oxidase but not of menaquinol oxidase. Therefore, we propose that the protein encoded by ypmQ is involved specifically in the assembly of Cu A but is not required for the assembly of Cu B . This may be a reflection of the different structural features of the Cu A and Cu B centers. For example, Cu A is located in an aqueous-exposed domain about 5 Å from the surface of the protein, whereas Cu B is bound to three histidine residues contained within a transmembrane helix of subunit I and is about 15 Å from the closest water-exposed surface of the protein (13,14).