The Saccharomyces cerevisiae COQ6 gene encodes a mitochondrial flavin-dependent monooxygenase required for coenzyme Q biosynthesis.

Coenzyme Q (Q) is a lipid that functions as an electron carrier in the mitochondrial respiratory chain in eukaryotes. There are eight complementation groups of Q-deficient Saccharomyces cerevisiae mutants, designated coq1-coq8. Here we have isolated the COQ6 gene by functional complementation and, in contrast to a previous report, find it is not an essential gene. coq6 mutants are unable to grow on nonfermentable carbon sources and do not synthesize Q but instead accumulate the Q biosynthetic intermediate 3-hexaprenyl-4-hydroxybenzoic acid. The Coq6 polypeptide is imported into the mitochondria in a membrane potential-dependent manner. Coq6p is a peripheral membrane protein that localizes to the matrix side of the inner mitochondrial membrane. Based on sequence homology to known proteins, we suggest that COQ6 encodes a flavin-dependent monooxygenase required for one or more steps in Q biosynthesis.

Coenzyme Q (ubiquinone or Q) 1 is a prenylated benzoquinone found in cell membranes that functions in redox chemistry as both an oxidant (Q) and reductant (QH 2 ) (1). Q is most commonly associated with its role in the respiratory chain where it transports electrons from either Complex I or Complex II to Complex III (2); however, Q serves multiple functions. Q has been demonstrated to play a role in stabilizing the bc 1 complex (3). Additionally, Q functions in the electron transport chains of lysosomal and plasma membranes (4,5), and QH 2 acts as a chain breaking antioxidant of lipid peroxyl radicals (6). In Escherichia coli, a high QH 2 :Q ratio is sensed by ArcB, a transmembrane sensor kinase, that phosphorylates ArcA, activating operons involved in fermentation and repressing those involved in respiration (7). In Caenorhabditis elegans, dietary Q produces a shortened life span (8). This phenomenon has been attributed to the generation of superoxide by Q . (the Q semiquinone radical) generated during respiratory electron transport (9). In humans, Q supplementation has been shown to be effective in treating patients with specific respiratory chain defects (10) and to slow the progression of Parkinson's disease symptoms (11).
Cells normally acquire Q through de novo synthesis, and the length of the prenyl tail varies among different organisms (12). Saccharomyces cerevisiae produce Q 6 , which has six isoprene units, whereas E. coli synthesize Q 8 , and humans synthesize Q 10 . Eight COQ genes have been identified to be required for Q synthesis in S. cerevisiae (13,14). Fig. 1 shows the pathway for Q biosynthesis in both prokaryotes and eukaryotes. In yeast, mutation in any of the eight COQ genes results in cells that cannot synthesize Q and fail to grow on nonfermentable carbon sources. Yeast coq3-coq8 mutants each accumulate the same early intermediate in Q biosynthesis, 3-hexaprenyl-4-hydroxybenzoic acid (Fig. 1, HHB or compound 1) (15).
In this work the yeast COQ6 gene has been isolated by functional complementation of a mutant from the G63 (coq6) complementation group. Transformation of coq6 mutant strains with a plasmid bearing wild type COQ6 restores Q biosynthesis and growth on nonfermentable carbon sources. In contrast to a previous report (16), we find COQ6 to be a nonessential gene. Here we show the Coq6 polypeptide is imported into mitochondria and is peripherally associated with the inner membrane on the matrix side. Based on sequence analysis and alignment with other known hydroxylases, the proposed function of Coq6 polypeptide is that of a flavin-dependent monooxygenase required for one or more steps in Q biosynthesis.
Cloning of the COQ6 Gene-The haploid strain SR128-3C containing the coq6-1 allele was obtained from mating C128 and FY250 (Table I). SR128-3C yeast were grown to early-log phase in YPD medium and transformed (20) with the YCp50 centromeric plasmid library of S. cerevisiae genomic DNA (21) containing the URA3 gene as a selectable marker. Transformants were selected on SDϪUra plates and after a 3-day incubation at 30°C were replica-plated onto YPG plates to test for respiratory growth. Putative Q prototrophic transformants, able to grow on medium containing a nonfermentable carbon source, were further purified and tested for co-segregation of uracil prototrophy and respiration competency following vegetative growth in rich medium. Such co-segregation was observed in two transformants, indicating that these traits were plasmid-linked. Yeast plasmid DNA was recovered from one transformant (494SR) and was amplified in DH5␣ E. coli (Invitrogen). The plasmid p494SR contained an insert of 4.1-kb and transformation of SR128-3C with p494SR restored growth on YPG medium. A similar cloning procedure was also performed with a multicopy expression library prepared from yeast genomic DNA in the vector YEp24 (22) and resulted in the isolation of pG63/T1, which was found to contain a 2.8-kb segment of DNA that overlapped with the insert present in p494SR (Fig. 2). Seven other rescuing yeast genomic DNA clones were similarly isolated from a recombinant pUV1-based plasmid library (generous gift of Junichi Nikawa and Michael Wigler, Cold Spring Harbor Laboratory). Southern analysis showed that all of the clones contained overlapping DNA fragments (data not shown).
Subcloning and Disruption of the COQ6 Gene-A 3.8-kb HindIII fragment was isolated from p494SR and was ligated to the HindIII site of pRS316 (23) to generate pSR1-1. The insert of pSR1-1 was sequenced and found to contain two ORFs located on a segment of chromosome VII. Only one ORF, YGR255C, was also present on pG63/T1, and was designated COQ6 (deposited in GenBank TM with accession number AF003698).
To construct a disruption allele of COQ6, a 2.9-kb HindIII fragment, FIG. 1. The Q biosynthetic pathway in eukaryotes and prokaryotes. Coq1p (S. cerevisiae) or IspB (E. coli) assemble the polyprenyldiphosphate tail with 6 and 8 isoprene units, respectively. After formation of 3-polyprenyl-4-hydroxybenzoic acid (compound 1) by the 4-hydroxybenzoic acid:polyprenyltransferase (Coq2p or UbiA), the proposed biosynthetic pathways for Q in eukaryotes and in prokaryotes is thought to diverge as shown. In yeast, n ϭ 6, and compound 1 is HHB. E. coli gene products are identified as Ubi (and also include IspB); S. cerevisiae gene products are identified as Coq.  (24). The resulting plasmid, pG63/ST3 was used to obtain a 4.0-kb EcoRI fragment with the disrupted gene. W303-1A, W303-1B, and SR128-3C were each transformed with 1 g of DNA (25). Most of the histidine prototrophic clones issued from the transformations were respiration-defective and were complemented by the °test strains JM6 or JM8 but not by the coq6-1 mutant strains, implying a genetic linkage between the coq6⌬::HIS3 and the coq6-1 alleles. Nuclear DNA from three independent transformants (one from each of the three separate parental strains) was digested with EcoRI and EcoRV, separated on 1% agarose, and transferred to a nitrocellulose membrane. The probe, a 1.1-kb EcoRI/EcoRV fragment within the COQ6 ORF, recognizes a 1.1-kb fragment in the genomic DNA of the parental strain. The genomic DNA of the mutant strains (W303⌬COQ6-1, ␣W303⌬COQ6-1, and SR⌬COQ6-1) contained a larger hybridizing species at ϳ3.15 kb, as expected for the disrupted allele. To verify the allelism between coq6⌬ and coq6-1 mutations, SR⌬COQ6-1 was mated to wild type strain W303-1A to form diploid cells, which were then sporulated to produce meiotic progeny. Spores from 14 tetrads showed 2:2 segregation for respiration competence and histidine auxotrophy. In all cases, the spores competent to grow on glycerol medium were histidine auxotrophs, whereas the spores unable to grow on glycerol were histidine prototrophs, confirming the allelism between the cloned COQ6 gene and the original coq6-1 mutation. Complete Deletion of the COQ6 ORF-A complete deletion of the COQ6 open reading frame was performed with a PCR-targeting strategy (26). A 1.53-kb fragment containing LEU2 was generated from the YEp13 vector (GenBank TM accession number U03498) using forward primer pPG6DLF 5Ј-ATAATTCTTAAAAGTGGAGCTAGTCTATTTCT-ATTTACATACCTCATTTTGTAATTTCGTGTCG-3Ј and reverse primer pPG6DLR 5Ј-TCAAATTGGTCTTTCAGTGAACCTTGTATCGATTGA-CACAGAGGCAGAGGTCGCCTGACGCATA-3Ј. The 5Ј end of the forward primer contained 45 nucleotides corresponding to Ϫ45 to Ϫ1 upstream of the COQ6 ORF and at the 3Ј end 19 nucleotides corresponding to nucleotides 5579 -5597 of YEp13. The reverse primer similarly corresponded to 45 nucleotides from ϩ1485 to ϩ1441 of the reverse strand of the COQ6 ORF followed by 19 residues from 7018 to 7000 of the reverse strand of YEp13. A W303-1AB diploid was created by mating W3031A with W3031B. W3031A, W3031B, CEN.PK2-1C, and W303-1AB cells were transformed with 1 g of the fragment using the lithium-acetate method as described (27). Transformed cells were grown on SDϪLeu plates at 30°C for 2 days. Gene disruption was verified by PCR using forward primer Coq6SeqF1 5Ј-ACCTTTGCAT-TACAAGTGCAACGCTCTACC-3Ј and reverse primer Coq6SeqR1 5Ј-GGTGACGCGTGTATCCGCCCGCTCTTTTGG-3Ј and produced a product of 1.84-kb for both the wild type and disrupted strains. The PCR product was then digested using the restriction enzyme EcoRI, which digests the wild type product and produces two products of 0.44 and 1.4 kb, whereas the EcoRI digestion fragments of the disrupted product are 1.2 and 0.64 kb. Tetrad analysis was performed as described above.
In Vivo Radiolabeling of Q 6 Intermediates, Lipid Extraction, and Analysis by HPLC-Yeast strains SR128-3C and W303⌬COQ6-1 were grown in 1 liter of SDC medium supplemented with 0.65 Ci of 4-[U- 14 C]hydroxybenzoic acid (365 Ci/mol), synthesized from L-[U-14 C]tyrosine by alkali heat fusion (28). Cells the were harvested after 3 days of incubation at 30°C with shaking (200 rpm, A 600 nm ϭ 10), and the lipids were extracted as described (29). The lipid extracts were concentrated, the volume was adjusted to 1 ml with hexane, and 0.10-ml aliquots (5.0 -12.0 ϫ 10 3 cpm) were analyzed by normal phase HPLC employing a cyanopropyl column (Zorbax, 5 m, 4.6 ϫ 250 mm; Mac-Mod Analytical, Chadds Ford, PA) as described (30). The column was equilibrated with a mobile phase composed of 98% solvent A (hexane) and 2% solvent B (isopropanol:hexane:water:methylene chloride, 52:41: 5:2) at a flow rate of 1 ml/min. Ten minutes after sample injection, the percentage of solvent B was increased linearly from 2 to 27% in 25 min (35 min from the start) and then from 27 to 100% in 20 min (55 min after the start). In the next 5 min, the percentage of solvent B was decreased linearly from 100 to 2% and remained at 2% for 30 min to equilibrate the column before next sample injection. The radioactivity of 1-ml fractions was measured by scintillation counting in 10 ml of BIOsafe nonaqueous scintillation mixture (Research Products International) with a Beckman scintillation counter (model LS-3133P). The efficiency of 14 C detection was 90%.
Determination of Q 6 Content in Lipid Extracts by HPLC and Electrochemical Detection-Lipid extraction and analysis were performed as described previously (31). Yeast strains W303-1A, SR128-3C, W303⌬COQ6-1, and W303⌬COQ6-1:pSR1-1 (the coq6-null strain harboring a COQ6 containing plasmid) were grown at 30°C with shaking (200 rpm) in 50 ml of YPGal to an A 600 nm of about 4 or alternatively were grown to saturation (2 days). 750 ng of Q 9 was added as an internal standard to a yeast cell pellet (100 mg of wet weight), and the cells were lysed by vortexing with 1 g of glass beads in 0.35 ml of water.
Mitochondrial Import Assay-An in vitro transcription template plasmid was constructed by inserting the COQ6 ORF into the pRS426 vector downstream of the T7 promoter (32). The COQ6 ORF with SalI and NotI linkers at 5Ј and 3Ј ends, respectively, was PCR-amplified with the template pSR1-1, 5Ј primer JF3 (5Ј-ACGCACGCGTCGACAT-GTTCTTTTCAAAAGTTATGC-3Ј), and 3Ј primer JF4 -1 (5Ј-ATAAGA-ATGCGGCCGCTCTCATTTCTCATTTCCTCC-3Ј) using Vent DNA polymerase (New England Biolabs). The PCR product was digested with SalI and NotI and inserted into the corresponding SalI and NotI sites in pRS426. The resulting product, pT7Q6, was linearized with XhoI and provided template DNA for in vitro transcription (Promega Ribomax large scale RNA production system). The resulting mRNA was then translated with Promega Flexi rabbit reticulocyte lysate system in the presence of [ 35  chondria (from D273-10B/A1) and import reaction conditions were performed according to Yaffe (33) as described previously (34). The cell cultures (1 liter) were grown in semisynthetic lactate medium to saturation density. The spheroplasts were prepared and lysed by Dounce homogenization with a tight fitting pestle as described (19). Purified mitochondria were isolated from a linear Nycodenz gradient as described (19). Each import reaction contained 6 l of radiolabeled in vitro translated product and isolated mitochondria (200 g of protein). Following the 30-min incubation at 30°C, the mitochondria were reisolated and washed once as described (33). Proteinase K treatment after the import was performed by adding proteinase K at final concentration of 50 g/ml to resuspended mitochondria. The proteolytic digest was allowed to proceed at 0°C for 20 min and was terminated by the addition of phenylmethylsulfonyl fluoride to a final concentration of 1 mM.
Plasmid Construction of HA-tagged COQ6 -Two yeast expression plasmids, one single copy and one multicopy, were constructed to express the Coq6 polypeptide containing a carboxyl-terminal peptide (MYPYDVPDYASLDGPMST) corresponding to the carboxyl terminus of the influenza HA viral protein, an epitope for the 12CA5 monoclonal antibody (35). The COQ6 ORF region without the stop codon was PCR-amplified using 5Ј primer JF3 (5Ј-ACGCACGCGTCGACATGT-TCTTTTCAAAAGTTATGC-3Ј) and 3Ј primer JF4 (5Ј-ATAAGAATGCG-GCCGCAGTTTCTCATTTCCTCCTAATGTG-3Ј). The PCR product was directionally cloned into the multicopy pADCL vector (SalI and NotI at 5Ј and 3Ј, respectively) (36), to generate pHA6. pSHA6, a single copy version of COQ6-HA construct, was generated by removing the 3.6-kb transcriptional cassette from pHA6 by BamHI partial digestion and insertion into BamHI site of the vector pRS316 (23). The yeast cells were transformed with pHA6, pSHA6, or pSR1-1 (25). Transformants were selected for the presence of either the LEU2 (pHA6) or URA3 (pSHA6) gene on SDϪLeu or SDϪUra plate media. Colonies obtained on the respective plate medium were subsequently replica-plated to YPG plate medium.
Generation of Antisera against Coq6p-The 1.44-kb COQ6 ORF was PCR-amplified with forward primer pPG6F 5Ј-GCGGATCCGATGT-TCTTTTCAAAAGTTATGCTT-3Ј and a reverse primer pPG6R 5Ј-GCG-GATCCTCATTTCTCATTTCCTCCTAATGTG-3Ј and Vent DNA polymerase. The product was then digested with BamHI and inserted into PET15b (Novagen) at the BamHI site to generate a fusion protein containing a His 6 tag at the amino terminus. The fusion protein was overexpressed in the E. coli BL21(DE3) under induction by 1 mM isopropyl-␤-D-thiogalactopyranoside and was purified over His-Bind resin (Novagen) and used to generate antiserum in rabbits (Cocalico).
Mitochondrial Localization of Coq6p-Yeast cultures (W303-1A) were grown in YPGal to an A 600 nm between 2 and 4, and mitochondria were isolated and purified as described above. Mitoplasts were generated by hypotonic treatment of mitochondria (19). Mitochondria (1 mg of protein) were suspended in five volumes of 20 mM HEPES-KOH, pH 7.4, and incubated on ice for 20 min. The mixture was then sedimented in a microcentrifuge for 10 min at 4°C to separate the intermembrane space components (supernatant) and the mitoplasts (pellet). The mitoplasts were then sonicated and centrifuged at 150,000 ϫ g for 60 min at 4°C to generate matrix (supernatant) and membrane (pellet) fractions. Alternatively, the mitoplasts were alkaline-extracted by incubating with 0.1 M Na 2 CO 3 , pH 11.5, for 30 min on ice, followed by centrifugation at 150,000 ϫ g for 60 min at 4°C to separate the integral membrane components (pellet) from the peripheral membrane and matrix components (supernatant) (37). Proteinase K protection experiments were carried out as described (38). The samples were analyzed by immunoblot.

Isolation and in Situ Disruption of the COQ6 Gene-
The original mutant in the coq6 complementation group of yeast mutants (G63) was identified as a Q-deficient nuclear petite strain (13). The respiratory deficiency was ascribed to a defect in Q because addition of exogenous Q 2 or Q 6 to isolated mitochondria rescued NADH-cytochrome c reductase activity (39). SR128-3C, one of the strains resulting from the cross between C128 (coq6-1) and FY251, had a very low reversion rate and provided the biological tool for cloning the COQ6 gene by screening yeast genomic DNA libraries as described under "Experimental Procedures." Nine respiration-competent transformants were obtained, and the representative plasmid p494SR was studied in detail. A 3.8-kb HindIII fragment was isolated from p494SR and ligated to the HindIII site of pRS316, a centromeric vector, to form pSR1-1 (Fig. 2). Transformation of SR128-3C with pSR1-1 restored growth on media containing glycerol. DNA sequence analysis revealed the insert to contain two complete ORFs, identified as YGR256W (the GND2 gene) and YGR255C. The plasmids pG63/T1 and pG63/ST2 containing the complete YGR255C ORF and only a portion of the GND2 gene, each restored the ability of SR128-3C to grow on YPG (Fig. 2). Expression of the YGR255C ORF as a carboxylterminal HA-tagged fusion protein from the alcohol dehydrogenase promoter also rescued YPG growth of SR128-3C.
COQ6 Is Not an Essential Gene-It has been reported that a complete disruption of the COQ6 ORF results in lethality because heterozygous knockouts failed to produce viable spores containing a coq6 disruption (16). However, all of the coq6 mutant strains used in this study were viable. A complete ORF disruption was created to address this apparent discrepancy. Three haploid strains (W303-1A, W3031B, and CEN.PK2-1C) were transformed with a PCR-generated disruption cassette, and all produced viable disruptants (W303⌬COQ6-2, W303⌬COQ6-2, and CEN⌬COQ6-2) as confirmed by PCR and restriction digest (data not shown). Similarly, a W303-1AB diploid was created by mating W303-1A with W303-1B and was transformed using the same cassette. The heterozygous COQ6/ coq6 diploid strain was subjected to sporulation and tetrad FIG. 3. Yeast coq6 mutants lack Q and accumulate a polar intermediate that corresponds to HHB. Lipid extracts were prepared from SR128-3C (coq6-1 allele) (a) or from W303⌬COQ6-1 (coq6 deletion mutant) (b) and separated by normal phase HPLC as described under "Experimental Procedures." The fractions were collected (1 ml), and 14 C radioactivity was determined by scintillation counting. A Q 6 standard eluted in fraction 6.

Coq6 Is a Flavin-dependent Monooxygenase in Q Biosynthesis
dissection. Each of the 10 tetrads analyzed produced four viable spores on YPD plate medium. The spores from each tetrad showed a 2:2 segregation for respiratory competence and leucine auxotrophy. These results indicate that the COQ6 gene is not essential for viability but is required for growth on nonfermentable carbon sources.
Yeast coq6-null Mutants Lack Q 6 and Accumulate HHB-Growth of coq6-1 mutants in the presence of 4-[U-14 C]hydroxybenzoic acid, the ring precursor in Q biosynthesis, showed this mutant lacked Q 6 and accumulated HHB (15). HHB is an early intermediate in the Q biosynthetic pathway and accumulates in mutant yeast strains harboring deletions or disruptions in any one of the COQ3, COQ4, COQ5, COQ7, or COQ8/ABC1 genes (15,29,31,40,41). To characterize the defect in Q biosynthesis in a coq6-null mutant, both SR128-3C (coq6-1) and W303⌬COQ6-1 (coq6⌬) were grown in the presence of 4-[U-14 C]hydroxybenzoic acid, and lipid extracts were analyzed by normal phase HPLC as described under "Experimental Procedures." Both strains lacked Q 6 and accumulate a radioactive intermediate that co-migrated with HHB in fraction 26 (Fig. 3). Neither SR128-3C nor W303⌬COQ6-1 was observed to produce Q 6 as analyzed by electrochemical detection, a method that can detect as little as 2 pmol of Q 6 /mg of wet weight of yeast. Q 6 levels of W303-1A and W303⌬COQ6-1:pSR1-1 were 187 and 148 pmol/mg of wet weight of yeast, respectively, when grown to log phase, and were 189 and 169 pmol/mg of wet weight of yeast, respectively, when grown to the stationary phase (data not shown).
Coq6p May Act as a Flavin-dependent Monooxygenase to Catalyze Quinone Formation-The predicted amino acid sequence of the COQ6 ORF revealed 21 and 24% sequence identity with the respective E. coli UbiH and UbiF polypeptides (Fig. 4). E. coli ubiH mutants lack Q 8 and accumulate compound 6, indicating that UbiH is required for the monooxygenase step that catalyzes quinone formation ( Fig. 1 and Ref. 42). The ubiH gene product was identified as a flavin-dependent monooxygenase (43). The E. coli ubiF gene product has also been identified as a flavin-dependent monooxygenase, with 31% amino acid sequence identity to UbiH (44). E. coli ubiF mutants lack Q 8 , accumulate compound 8, and both ubiH and ubiF mutants fail to grow on media containing succinate ( Fig.  1 and Ref. 45). Coq6p, UbiH, UbiF, and other eukaryotic homologs of Coq6 each contain three regions with amino acid sequence identities that are present in a large family of flavindependent monooxygenases (Fig. 5 and Ref. 46). Region 1 contains an ADP-binding fingerprint (47), Region 2 is implicated in the recognition of NAD(P)H and may also be involved indirectly in binding the pyrophosphate moiety of FAD (48), whereas Region 3 contains a consensus sequence for binding to the ribityl moiety of FAD (49). Based on the homology with UbiH, and UbiF and the presence of the conserved motifs found in other aromatic flavin-dependent monooxygenases, it seems likely that Coq6p functions in one or more hydroxylation steps in Q biosynthesis.

Coq6 Is a Flavin-dependent Monooxygenase in Q Biosynthesis
contains the sequence motif common in many mitochondrial matrix proteins that are proteolytically cleaved twice once imported into matrix, as characterized by an arginine at Ϫ10, a hydrophobic residue (phenylalanine, leucine, or isoleucine) at Ϫ8, and serine, threonine, or glycine at Ϫ5 relative to the amino-terminal residue of the proteolytically processed protein. To determine whether Coq6p is a mitochondrial protein, an in vitro mitochondrial import assay of the Coq6 polypeptide was performed. Upon incubation with mitochondria prepared from wild type yeast, the in vitro translated Coq6p was cleaved, and the resulting mature form of the protein was observed to have a mass of about 51 kDa, consistent with lysine 21 as the putative cleavage site (Fig. 6). This mature protein was protected and resistant to exogenous proteinase K treatment (Fig.  6, lane 3). Disruption of mitochondrial membrane with detergent exposed the imported and processed Coq6p, which became accessible to the proteinase K (Fig. 6, lane 4). The import of Coq6p also required a mitochondrial membrane potential, because the addition of valinomycin inhibited import of Coq6p (Fig. 6, lanes 5 and 6).
Mitochondrial Localization of Coq6p-Our initial investigation of the subcellular localization of Coq6p made use of a carboxyl-terminal fusion protein between Coq6p and the HA epitope peptide (see "Experimental Procedures"). The Coq6-HA epitope fusion protein retained activity as assayed by the ability of either the single-or multi-copy plasmid construct to rescue coq6-null mutant yeast strains for growth on medium containing a nonfermentable carbon source (YPG plates; data not shown). However, subsequent subcellular and submitochondrial fractionation analysis revealed that the Coq6-HA tagged construct was present in mitochondria as an insoluble aggregate because it remained in the pellet following treatment with 1% Triton X-100 and 1 M NaCl. For this reason, antibodies were generated to the Coq6 polypeptide.
A polyclonal antibody was generated in rabbit against a Coq6-His-tagged fusion protein. This antibody recognized a polypeptide of 51 kDa in wild type yeast cell extracts and also recognized the Coq6 HA-tagged polypeptide that migrates at about 57 kDa (Fig. 7a). No polypeptide at these molecular weights was detected in cell extracts of the coq6 deletion strain, indicating that the antibody is specific for Coq6p. Western blot analysis of subcellular yeast fractions revealed that Coq6p co-fractionated with the mitochondria along with Coq4p, Coq3p, and cytochrome b 2 (Fig. 7b). To determine whether Coq6p is a membrane-bound or soluble protein, purified mitochondria were osmotically shocked to disrupt the outer membrane and release components of the intermembrane space. Mitoplasts were then further disrupted by sonication to separate the remaining soluble and membrane components. Fig. 7c shows that Coq6p localizes to the membrane (pellet) fraction.
To determine whether Coq6p resides in the inner or outer membrane, mitochondria were treated with combinations of proteinase K, Triton X-100, and hypotonic swelling. Fig. 7d shows that Coq6p is degraded only when the detergent is present, as is the case for two matrix proteins, Hsp60p and Mas2p. Finally, to determine whether Coq6p is a peripheral or integral protein in the inner mitochondrial membrane, mitoplasts were treated with 0.1 M sodium carbonate buffer, pH 11.5. Fig. 7e shows that following this treatment Coq6p is released into the supernatant, as are the ␤ subunit of F 1 -ATPase and Coq4p (two peripheral membrane proteins), whereas cytochrome c 1 (an integral membrane protein) is not. FIG. 6. In vitro mitochondrial import and proteolytic processing of Coq6. Lane 1, 10% of the radiolabeled in vitro translation product used in each import reaction. Lane 2, in vitro translation product incubated with isolated mitochondria. Lane 3, same as lane 2 but including a post-import proteinase K treatment. Lane 4, same as lane 3 except Triton X-100 was added to disrupt the mitochondria before proteinase K treatment. Lanes 5 and 6, in vitro translation product was incubated with mitochondria in the presence of the uncoupler valinomycin (final concentration, 10 g/ml). After the incubation, the reisolated mitochondria were either treated with proteinase K as in lane 5 or were directly analyzed as in lane 6. The samples were mixed with sample buffer and heated at 100°C for 5 min prior to analysis by SDS-polyacrylamide electrophoresis (12% polyacrylamide) and autoradiography. P and M indicate the positions of the precursor form and the mature form of Coq6p, respectively.

FIG. 7.
Coq6p is peripherally associated with the inner membrane of mitochondria. a, the antiserum generated against Coq6p and its specificity was tested against the wild type strain W303-1A (WT), the coq6 deletion strain W303-1A⌬COQ6-1 (⌬), and the multicopy HA-tagged rescued strain W303-1A⌬COQ6-1:pHA6-1 (MC). b, whole cell yeast was homogenized and separated by differential centrifugation into nuclear (P1), crude mitochondrial (CM), Nycodenz gradient-purified mitochondrial (NM), and post-mitochondrial supernatant fractions (PS). c, Nycodenz-purified mitochondria (NM) were subjected to hypotonic swelling and centrifuged to separate intermembrane space proteins in the supernatant (IMS). The pellet was then sonicated and centrifuged to release a soluble matrix protein fraction in the supernatant (S) and membrane protein fraction pellet (P). d, intact mitochondria or mitoplasts generated by hypotonic swelling were treated with proteinase K (100 g/ml) for 30 min with or without 1% Triton X-100. e, mitoplasts were incubated with 0.1 M Na 2 CO 3 , pH 11.5, on ice 30 min. Centrifugation produced soluble (S) and insoluble (P) fractions, which were compared against Nycodenz-purified mitochondria (NM). All of the samples were separated by SDS-PAGE and analyzed by immunoblotting as described under "Experimental Procedures." The antiserum was used against Coq6p and compared with results of antisera against outer membrane protein OM45p, intermembrane space protein cytochrome b 2 (Cyt b 2 ), integral inner membrane protein cytochrome c 1 (Cyt c 1 ), peripheral inner membrane proteins Coq3p, Coq4p, and the ␤ subunit of F 1 -ATPase (F 1 ␤), and matrix proteins Hsp60 and Mas2.

DISCUSSION
This work characterizes yeast coq6 mutants and the isolation of the COQ6 gene. COQ6 is necessary for Q biosynthesis. In the original coq6 mutant and in both coq6 partial and complete deletion mutants, Q 6 is undetectable, and as a result, these cells are rendered respiratory incompetent. Based on the data presented here, COQ6 encodes a mitochondrial protein necessary for Q biosynthesis. Similar to Coq2p, Coq3p, Coq4p, Coq5p, Coq7p, and Coq8p/Abc1p, the Coq6 polypeptide is imported into mitochondria (14). In agreement with submitochondrial localization studies of Coq1p, Coq3p, Coq4p, Coq5p, and Coq7p, Coq6p is localized to the matrix side of the inner membrane (18,40,50). 2 S. cerevisiae is a facultative anaerobe capable of fermentation and aerobic respiration, and none of the other COQ genes have been identified as being essential. Complete deletions of the COQ6 gene were introduced into both haploid and diploid strains of W303 and also into CEN.PK2-1C, a haploid strain similar in genetic background to the diploid used by Fiori et al. (16). In contrast to their results, all of the yeast strains were viable when COQ6 was deleted, indicating that none of the eight COQ genes is essential. Q itself is nonessential for growth in this organism.
The function of Coq6p is still unknown, as are the functions of Coq4p and Coq8p. Based on sequence homology to known flavin dependent monooxygenases, it is likely that Coq6p acts similarly. Although Coq6p and UbiF share 24% sequence identity, it seems unlikely that Coq6p functions like UbiF in the hydroxylation of compound 8 in yeast, because previous studies have shown that yeast Coq7p is required for this hydroxylation step (29). The Coq7 polypeptide has been identified as a di-iron carboxylate protein, a member of a monooxygenase family distinct from UbiF (51). Although E. coli lack a homolog of Coq7p, homologs of yeast Coq7p from other prokaryotic species rescue the E. coli ubiF hydroxylase mutants, indicating that these two distinct types of monooxygenases each catalyze the same reaction in Q biosynthesis (51). Currently, two of the uncharacterized steps in Q biosynthesis are hydroxylations, and Coq6p may be responsible for either or both of these steps. Coq6p has 21% identity to the E. coli UbiH, which converts compound 6 to compound 7. Of course other hydroxylation substrates for Coq6p are also possible, including the hydroxylation of HHB to compound 4 (Fig. 1). Indeed, because coq6-1 and coq6-null mutants both accumulate HHB (Fig. 3), it seems attractive to postulate its action at this step. However, the accumulation of HHB is not diagnostic of the blocked step, because so many other coq mutants also accumulate HHB (15). Another possibility is that Coq6p catalyzes the oxidative decarboxylation of compound 5 to compound 7. It has been demonstrated that 4-hydroxybenzoate 1-hydroxylase oxidatively decarboxylates 4-hydroxybenzoic acid with the formation of hydroquinone (52). Salicylate hydroxylase similarly oxidatively decarboxylates salicylate to catechol (53). However, in yeast this is less likely given that previous work showed a Q-deficient yeast mutant accumulated compound 6 (54). That mutant was never sequenced, nor is it available. Identification of the role of Coq6p in the hydroxylation step(s) of Q biosynthesis will require development of in vitro assays for these hydroxylation steps.
Blast analysis of Coq6p against other higher eukaryotic genomes revealed that only one homolog existed for Homo sapiens, Mus musculus, and C. elegans, indicating a conserved evolution for this protein (Fig. 4). For Drosophila melanogaster, two matches were found, but only one had a high score and spanned the entire sequence. The homologs in each case re-tained the three conserved motifs depicted in Fig. 5.
Genetic data have suggested that a putative complex of these Coq polypeptides is responsible for Q biosynthesis because a deletion in any one of the eight COQ genes results in a profound decrease in steady state protein levels for Coq3p, Coq4p, and Coq7p (34). 2 Furthermore, null mutations in coq3, coq4, coq5, coq7, or coq8/abc1 result in the accumulation of HHB (15). Similarly, analyses of lipid extracts of coq6 mutants in the studies presented here show that these mutants also accumulate HHB. It has been observed that an E226K coq4 point mutant did allow for the steady state expression of Coq3p and Coq7p (40). In addition, a G104D coq7 point mutant allowed for the accumulation of 2-hexaprenyl-3-methyl-6-methoxy-1,4benzoquinone (Fig. 1, demethoxy-Q or compound 8) (29). It would appear that both the coq4 and coq7 amino acid substitution mutants allow for the expression of correctly folded but inactive Coq4p and Coq7p, respectively, and each may be interacting and stabilizing the steady state expression of the other Coq polypeptides in a complex. The accumulation of HHB in both the coq6 mutants further suggests that Coq6p similarly is required for stabilization of a complex. Additionally, the coq6-1 mutant has been found to be a nonsense mutation of the tyrosine at position 218. 3 This mutant would be unable to produce a stable polypeptide at full length and would explain why we see an accumulation of HHB.