Yeast and Rat Coq3 and Escherichia coli UbiG Polypeptides Catalyze Both O-Methyltransferase Steps in Coenzyme Q Biosynthesis*

Ubiquinone (coenzyme Q or Q) is a lipid that functions in the electron transport chain in the inner mitochondrial membrane of eukaryotes and the plasma membrane of prokaryotes. Q-deficient mutants of Saccharomyces cerevisiae harbor defects in one of eight COQ genes (coq1–coq8) and are unable to grow on nonfermentable carbon sources. The biosynthesis of Q involves two separate O-methylation steps. In yeast, the first O-methylation utilizes 3,4-dihydroxy-5-hexaprenylbenzoic acid as a substrate and is thought to be catalyzed by Coq3p, a 32.7-kDa protein that is 40% identical to theEscherichia coli O-methyltransferase, UbiG. In this study, farnesylated analogs corresponding to the secondO-methylation step, demethyl-Q3 and Q3, have been chemically synthesized and used to study Q biosynthesis in yeast mitochondria in vitro. Both yeast and rat Coq3p recognize the demethyl-Q3 precursor as a substrate. In addition, E. coli UbiGp was purified and found to catalyze both O-methylation steps. Futhermore, antibodies to yeast Coq3p were used to determine that the Coq3 polypeptide is peripherally associated with the matrix-side of the inner membrane of yeast mitochondria. The results indicate that oneO-methyltransferase catalyzes both steps in Q biosynthesis in eukaryotes and prokaryotes and that Q biosynthesis is carried out within the matrix compartment of yeast mitochondria.

the isoprenoid tail functions to anchor Q in the membrane. In eukaryotes, Q functions to shuttle electrons from either Complex I or Complex II to Complex III/bc 1 complex. The transfer of electrons from Q to the bc 1 complex is coupled to proton-translocation via the Q cycle mechanism that was first proposed by Mitchell (2). A number of studies support such a mechanism (for a review, see Ref. 1) including the recently determined complete structure of the bc 1 complex (3).
The redox properties of Q also allow it to function as a lipid soluble antioxidant. Q functions by either directly scavenging lipid peroxyl radicals (4) or indirectly reducing ␣-tocopherol radicals to regenerate ␣-tocopherol (5,6). Additionally, Q protects cells from oxidative damage generated by the autoxidation of polyunsaturated fatty acids (7). Q is found in many eukaryotic intracellular membranes, including the plasma membrane, where, in conjunction with a plasma membrane electron transport system, it functions to scavenge ascorbate free radicals (8,9). In the plasma membrane of prokaryotes, Q participates in the maintenance of the enzymatic activity of DsbA/DsbB disulfide bond forming proteins (10), and Q-deficient Escherichia coli strains are hypersensitive to thiol exposure (11).
In both eukaryotes and prokaryotes, the first committed step in the biosynthesis of Q begins with the precursors p-hydroxybenzoic acid (pHB) and isoprenoid diphosphate, in which the isoprenoid is covalently attached to the aromatic ring. The pathway derives from the characterization of accumulating Q biosynthetic intermediates in studies with Saccharomyces cerevisiae (12) and E. coli (13) Q-deficient mutants. In yeast, Q mutant strains have been classified into eight complementation groups, and five COQ genes have been characterized. The COQ1 and COQ2 genes encode the polyprenyl diphosphate synthase and the pHB:polyprenyldiphosphate transferase, respectively (14,15). The COQ3 gene encodes the O-methyltransferase thought to catalyze the first O-methylation step (16,17), and the COQ5 gene encodes the C-methyltransferase in Q biosynthesis (18,19). Finally, the COQ7 gene encodes a protein that localizes to yeast mitochondria (20) and is required for the final monooxygenase step in Q biosynthesis (21), but has also been implicated in aging and development in C. elegans (22).
The Q biosynthetic pathway in E. coli has been carefully worked out by analyzing ubi mutant strains (23) for accumulating Q intermediates at the blocked metabolic steps, and many of the bacterial genes have been characterized (24). These include ubiC, ispB, and ubiA, which encode the chorismate pyruvate lyase (25), octaprenyl synthase (26), and the pHB:octaprenyltransferase (27), respectively. Genes encoding the hydroxylase (ubiH) (28) and the O-methyltransferase (ubiG) (29,30) have also been reported, and recently, the gene encoding the C-methyltransferase gene in E. coli was charac-terized (ubiE) (31). Although eukaryotes and prokaryotes share many similar steps in Q biosynthesis, the pathway diverges after the prenylation step (16,32,33). In prokaryotes, decarboxylation, hydroxylation, and methylation follow prenylation, whereas in eukaryotes, the sequence is hydroxylation, methylation, and then decarboxylation. Recent evidence suggests that the Q biosynthetic pathway in higher eukaryotes is similar to S. cerevisiae. Both rat and human COQ3 and COQ7 homologs can complement the corresponding defect in yeast (34 -36). 2 We have been examining the enzymes that catalyze the O-methylations in prokaryotic and eukaryotic Q biosynthesis. E. coli strains harboring null mutations in the ubiG gene are defective in the first O-methylation step (conversion of compound 1 to 2, Fig. 1) (30). Surprisingly, strains harboring leaky mutant alleles of ubiG accumulate demethyl-Q 8 , the last intermediate in Q biosynthesis (Fig. 1, compound 5), and are unable to carry out the last O-methylation step (37,38). The analysis of both null and leaky mutant alleles of ubiG suggested that the ubiG gene product was required for both of the O-methylations in Q biosynthesis (30). Unlike the E. coli ubi mutants, analysis of accumulating Q intermediates in yeast coq mutants has been less informative. Yeast strains harboring coq3, coq4, coq5, coq6, coq7, or coq8 mutant alleles all accumulate the same single predominant intermediate, 3-hexaprenyl-4-hydroxybenzoic acid (39,40). For this reason, it has often been instructive to compare the yeast COQ genes with the E. coli ubi gene counterparts. The encoded amino acid sequence of yeast COQ3 is 40% identical with the E. coli UbiG protein and both sequences contain the four motifs identified in a large family of S-adenosyl-L-methionine (AdoMet)-dependent methyltransferases (41). In this study, in vitro assays have been developed that facilitate the study the catalytic role of both the UbiG and Coq3 proteins in O-methylation reactions. These assays demonstrate that each enzyme is active at all three O-methylation steps shown in Fig. 1. Mitochondria subfractionation studies indicate that the Coq3 polypeptide is a peripherally associated inner membrane protein, located on the matrix side. The results presented suggest that both the first and last O-methylation steps in the yeast Q biosynthetic pathway occur within the mitochondria matrix compartment.

EXPERIMENTAL PROCEDURES
General Synthetic Procedures-All reagents were used as received from Aldrich Chemical Co. unless otherwise noted. Unless specified as dry, the solvents were of unpurified reagent grade. Diethyl ether was distilled from sodium using benzophenone as an indicator. All air-or water-sensitive reactions were carried out under positive pressure of argon. Reactions were followed by TLC using Whatman precoated plates of silica gel 60 with fluorescent indicator. Reactions forming quinones were followed by leucomethylene blue stain. Normal phase flash chromatography was performed on Davisil Grade 643 silica gel (230 -400 mesh). NMR spectra were measured on a Bruker ARX400 or ARX500 MHz spectrometer. Low and high resolution mass spectra were determined on a VG Autospec. Synthetic procedures used to generate farnesylated analogs of compounds 1, 2, 3, and 4 ( Fig. 1) were described previously (30,42,43).
3-Hydroxy-4,5-dimethoxy-2-acetyltoluene (8)-In a glove bag under N 2 , AlCl 3 (3.29 g, 24.7 mmol) was placed into a 100-ml round-bottomed flask. The flask was sealed and transferred to an argon atmosphere before anhydrous ether (15 ml) was slowly added, followed by 3,4,5trimethoxytoluene (7) (2.8 ml, 16.6 mmol) and acetyl chloride (1.5 ml, 17.3 mmol). The reaction mixture turned dark and murky and was stirred for 20 h at room temperature. Following the addition of water (10 ml) and concentrated HCl (1 ml), the mixture was extracted with ether (three times, 15 ml). The combined ether layers were extracted with 1 M NaOH (three times, 20 ml), and the resulting aqueous layers were acidified by dropwise addition of concentrated HCl and then cooled in an ice-bath for 1 h. The product crystallized and was filtered using a Buchner funnel with Whatman No. 50 paper to give 1.56 g (44.6% yield) of pale yellow solid 8. 1  2,3-Dihydroxy-4,5-dimethoxytoluene (9)-Compound 8 (180 mg, 0.86 mmol) was dissolved in a solution of sodium hydroxide (68 mg, 1.7 mmol) and water (4 ml). Hydrogen peroxide (0.12 ml, 30% in H 2 O) was added dropwise to the reaction mixture via an addition funnel over 10 min. The mixture was then heated at 45°C for 2 h. Five minutes after the heating was initiated, the solution darkened from a pale yellow to a deep violet. The reaction was quenched by the addition of 1 M HCl (15 ml) and then extracted with dichloromethane (3 ϫ 20 ml). The combined organic layers were dried over Na 2 SO 4 , filtered, and concentrated, by rotary evaporation. Flash chromatography using hexane:ethyl acetate (9:1) gave yellow solid 9 (80 mg, 51% yield). 1  2-Hydroxy-3-methoxy-6-methyl-1,4-benzoquinone or Fumigatin (10)-Compound 9 (73.4 mg, 0.40 mmol) was dissolved in a 1:1 mixture of dichloromethane:acetonitrile (4 ml). A solution of ammonium cerium (IV) nitrate (655.4 mg, 1.20 mmol) in dichloromethane:acetonitrile (1:1, 2 ml) was then added dropwise to the reaction mixture over 5 min. The solution color changed from yellow to a turbid maroon. Stirring was continued for 5 min before the reaction was quenched by the addition of 10 ml of water. The reaction mixture was extracted with dichloromethane (three times, 15 ml) and the combined organic layers were concentrated by rotary evaporation. The crude residue was redissolved in ether (20 ml) and then treated with 1 M NaHCO 3 (20 ml). The aqueous layer became a bright violet color. Following two washes with ether, the aqueous layer was slowly acidified using concentrated HCl until the solution color changed from deep violet to orange. The aqueous layer was then extracted three times with ether. The combined ether extracts were dried over MgSO 4 , filtered, and concentrated by rotary evaporation to give fumigatin (56.5 mg, 84.4% yield), a red crystalline solid. 1
Plasmid Construction-DNA constructions were performed as described (48). pTHG was constructed to express UbiG as a fusion protein with a 33-amino acid N-terminal extension, containing 6 His residues (His 6 -UbiG) to provide for metal affinity column purification. A DNA segment containing the complete ubiG ORF (851-1572) was generated by a polymerase chain reaction with Vent DNA Polymerase (New England Biolabs) using pRPB (29) as the DNA template with the primers, pGB, (5Ј-GCGGATCCGATGAATGCCGAAAAATCGCCGGTA-3Ј) and pCC4K (30). The resulting 723-base pair product was inserted after digestion with BamHI and KpnI into the similarly digested vector pTrcHisB (Invitrogen) to generate pTHG. The plasmids, pQM and pCHQ3, were described previously (30).
Purification of His 6 -UbiG Fusion Protein-Purification of His 6 UbiG was done with the TALON metal affinity resin (CLONTECH) as described by the manufacturer. The E. coli strain, DH5␣:pTHG, containing the His 6 -UbiG was grown in LBϩAmp (50 g/ml) and induced with isopropyl-1-thio-␤-D-galactopyranoside (final concentration, 0.4 mM), and cells were disrupted by the French press method. His 6 -UbiG was purified on a TALON column under native conditions. The resin was washed with 15 mM imidazole to remove nonspecifically bound proteins, His 6 -UbiG was eluted from the resin with 250 mM imidazole, and the imidazole was removed by dialysis against 0.05 M sodium phosphate, pH 7.0.
Generation of Yeast Coq3p Antibodies-A plasmid encoding a glutathione S-transferase-Coq3p fusion protein was constructed by subcloning the 1.7-kilobase EcoRI fragment of pRS12A (17) into the EcoRI site of pGEX-2T (Amersham Pharmacia Biotech). The fusion protein contained amino acids 64 -316 of yeast Coq3p as a C-terminal fusion to glutathione S-transferase and was produced in E. coli and the insoluble fraction was separated by preparative SDS-polyacrylamide gel electrophoresis. The 50-kDa fusion protein was visualized by copper staining (49) and eluted from the gel by diffusion (50). The protein was injected into rabbits, and antibodies were affinity purified according to standard techniques (51).
In Vitro Assays-Assays for O-methyltransferase activities were determined with the three synthetic methyl-acceptors, compounds 1, 3, and 5. Stocks of 1, 3, and 5 were stored undiluted at Ϫ20°C under argon. In assays with either 1 or 3, the substrates were redissolved into methanol, and each reaction mixture  (53). After incubation, the reaction was stopped by addition of glacial acetic acid (5 l), and the lipids were extracted with chloroform, concentrated and analyzed by high performance liquid chromatography as described (30). In vitro assays with compound 5 were the same as described above except that 5 was redissolved into hexane and NADH (3 mM) was included in the assays with yeast mitochondria in order to form the hydroquinone. In assays with purified His 6 -UbiG, 5 was reduced with 10% sodium dithionite, and prior to addition, the sodium dithionite was removed by centrifugation. Following incubation, reactions were terminated by the addition of excess ammonium cerium (IV) nitrate to oxidize the methylated product, and lipids were extracted with hexane (two times, 0.5 ml), concentrated, and analyzed by high performance liquid chromatography as described above.
Localization of Yeast Coq3p-Yeast (W3031A or CC3031B) was grown in YPGal media to saturation density (A 600 ϳ 10.0), and a crude mitochondria fraction was isolated and further purified over a linear Nycodenz gradient as described (52). Subfractionation of purified mitochondria was carried out by generating mitoplasts as described (20), and fractionation of mitoplasts was accomplished by either sonication (four 10-s pulses, 20% duty cycle, 2.5 output setting Sonifier W350, Branson Sonic Power Co.) or alkaline carbonate extraction (54,55). Protease protection experiments were carried out as described (56). Equal amounts of protein as determined by the BCA method (Pierce) were separated by electrophoresis on 12% polyacrylamide gels (57) and subsequently transferred to Hybond ECL nitrocellulose (Amersham Pharmacia Biotech). Western analysis with the ECL system was carried out as described by Amersham Pharmacia Biotech except that 10 mM Tris, pH 8.0, 154 mM NaCl, 0.1% Triton X-100 was used as the Western washing buffer. The primary antibodies were used at the following dilutions: ␣-Coq3, 1:1,000; Hsp 60, 1:10,000; F 1 ␤ATPase, 1:10,000; cytochrome c 1 , 1:400; cytochrome b 2 , 1:5000. Horseradish peroxidase-linked secondary antibodies to rabbit IgG (Amersham Pharmacia Biotech) were used in a 1:1000 dilution.

Coq3p Is Required for Both O-Methylation Steps in Ubiquinone Biosynthesis-Our previous O-methyltransferase in vitro
assays indicated that multiple steps may be catalyzed by the same enzyme (30). Specifically, in vitro assays with cell free extracts of E. coli showed that the ubiG gene was required for the methylation of both compounds 1 and 3. These results indicated that UbiG was involved in both O-methylation steps of Q biosynthesis, because Leppik et al. (38) showed that UbiG was required for the methylation of 5 to 6. By analogy, it seemed likely that the COQ3 gene product may also be required for both O-methylation steps in eukaryotic Q biosynthesis. To test this idea, in vitro O-methylation assays were performed with the synthetic Q-intermediate analog 5 (n ϭ 3) as substrate. The methyl donor was [methyl-3 H]AdoMet, and mitochondria were isolated from three yeast strains: 1) a wildtype respiratory competent strain (JM45), 2) the coq3 deletion mutant harboring the plasmid vector as a control (JM45⌬coq3: pQM), and 3) a rescued mutant with a multicopy plasmid encoding yeast COQ3 expressed from the CYC1 promoter (JM45⌬coq3:pCHQ3) (Fig. 3A). Mitochondria from respiratory competent yeast produced a radioactive product that co-migrated with the Q 3 standard (compound 6) on reverse-phase high performance liquid chromatography (Fig. 3A) (fraction 17). This activity (40.2 pmol/mg of protein/h) required the reducing agent, NADH, because omitting NADH resulted in no O-methyltransferase activity. No activity was detected in mitochondria isolated from a coq3 null mutant (JM45⌬coq3: pQM). However, transformation of this strain with the COQ3 gene (JM45⌬coq3:pCHQ3) restored activity (161 pmol/mg of protein/h). Thus, a functional Coq3 polypeptide is required for both the first (43) and second O-methylation steps in yeast Q biosynthesis.
Similar in vitro assays were carried out to determine whether Coq3p was required for the methylation of the farnesylated analog of the E. coli substrate (compound 1). As shown in Fig. 3B, mitochondria from wild-type yeast contained high activity (22.8 pmol/mg of protein/h) and produced a radiolabeled product that co-migrated with the farnesylated analog of 2. This activity was not detected in the coq3 null mutant (JM45⌬coq3:pQM), but the activity (16.3 pmol/mg of protein/h) was again restored when mitochondria from the rescued strain were examined (JM45⌬coq3:pCHQ3). These results suggest that the Coq3p O-methyltransferase is capable of methylating multiple Q precursor analogs.
Conservation of Function between Yeast and Rat O-Methyltransferase Activity-To examine whether the in vitro assays described above could be used to study Q biosynthetic steps in higher eukaryotes, the plasmid pAB2 (34), which contains the rat COQ3 cDNA, was transformed into JM45⌬coq3. Mitochondria were isolated from this strain and assayed for O-methylation activity with farnesylated analogs of 1 (Fig. 4A), 5 (Fig.  4B) or 3 (Fig. 4C). In each case, the radioactive methylated products were detected that eluted with chemically synthesized methylated products (2, 6, and 4, respectively). The activities were 174.2, 42.5, and 54.1 pmol/mg of protein/h, respectively. These assays demonstrate that farnesylated analogs of Q biosynthetic intermediates can be used to study Q biosynthesis in higher eukaryotes. Additionally, these results indicate that both O-methylation steps in rats also require Coq3p and that this O-methyltransferase has a wide substrate specificity.   (Fig. 5A), 5 (Fig. 5B) or 3 (Fig. 5C) were tested as methyl-acceptor substrates. In each case, radioactive methylated products were detected that co-eluted with chemically synthesized methylated products 2 (813 mol/mg of protein/h), 6 (275 mol/mg of protein/h), and 4 (2, 290 mol/mg of protein/h), respectively. Methylation of 5 required that it be reduced prior to addition (data not shown). Thus, the purified UbiG polypeptide is sufficient for the catalysis of both O-methylation steps in the biosynthesis of Q in E. coli, and is capable of methylating the eukaryotic substrate.
Subcellular Localization of the Coq3 Polypeptide-Previous work showed that the yeast Coq3p precursor was imported into mitochondria in vitro, and a mitochondrial membrane potential was required for processing to the mature (protease resistant) form (30). To determine the location of Coq3p in yeast, affinity purified polyclonal antibodies were prepared against Coq3p. Fractions of cytosol, crude mitochondria, and Nycodenz-puri-fied mitochondria were prepared from both the CC3031B null mutant strain and the wild-type parental strain, W3031A. Immunoblot analysis (Fig. 6A) of each fraction indicated that the 33-kDa polypeptide (corresponding to the mature Coq3p) was detected only in the mitochondrial fractions of wild-type yeast but not in the coq3 null mutant.
Submitochondrial Localization of the Coq3 Polypeptide-To determine the submitochondrial localization of Coq3p, yeast mitochondria were further fractionated (54). Purified mitochondria from W3031A were subjected to treatment with hypotonic buffer, which disrupts the outer membrane and releases soluble proteins of the intermembrane space while keeping the inner membrane intact. Western analysis of the soluble fraction indicated that Coq3p remained associated with the pellet (mitoplast fraction) and did not co-purify with the intermembrane space marker, cytochrome b 2 (data not shown). Mitoplasts were further fractionated either by sonication, which releases soluble matrix proteins into the supernatant following centrifugation, or by extraction with alkaline carbonate, which releases both soluble and peripherally bound membrane proteins into the supernatant (61). As shown in Fig. 6B, Coq3p was released by alkaline carbonate extraction, which was similar to the matrix marker, Hsp60 (62), and the peripheral inner membrane protein F 1 ␤ATPase (63). In contrast, these conditions did not release the integral membrane marker, cytochrome c 1 (64). However, sonication conditions that release Hsp60 into the supernatant fraction, did not re- FIG. 4. The rat COQ3 gene restores O-methyltransferase activity in coq3 null mutant yeast. Yeast crude mitochondrial extracts were prepared from JM45⌬coq3:pAB2 (rat COQ3 gene) and in vitro O-methylation assays were carried out as described in Fig. 5. Three different analogs of Q-intermediates were used as substrates: A, 2-farnesyl-6-hydroxyphenol (compound 1); B, demethyl-Q 3 (compound 5); C, 3,4-dihydroxy-5farnesylbenzoic acid (compound 3). In each assay, O-methyltranferase activity required the rat COQ3 gene because no activity was detected in its absence (see Fig. 3, A and B). The elution positions of methylated farnesylated standards (2, 6, and 4) are indicated. lease Coq3p or the peripheral membrane marker, F 1 ␤ATPase. These results indicate that Coq3p is a peripheral membrane protein similar to the F 1 ␤ATPase.
To determine whether Coq3p is associated with the matrixside or the outside of the inner membrane of yeast mitochondria, purified mitochondria or mitoplasts were subjected to increasing concentrations of proteinase K and then subjected to Western analysis (Fig. 6C). The results indicate that Coq3p was protected from protease treatment in both intact mitochondria and mitoplasts. This degree of protease protection is also a property of the inner membrane marker, F 1 ␤ATPase, and Hsp60, a matrix marker. However, cytochrome b 2 , an intermembrane space protein, was fully digested in mitoplasts as expected. Additionally, treatment of mitoplasts with 1% Triton X-100 detergent rendered all proteins protease-sensitive. These data indicate that the Coq3 polypeptide is peripherally associated with the matrix side of the inner membrane of mitochondria. DISCUSSION This study demonstrates that both O-methylation steps in Q biosynthesis are catalyzed by the same enzyme. The in vitro O-methylation assays employ farnesylated analogs of compounds 1, 3, and 5 as substrates, [methyl-3 H]AdoMet, and the detection of radiolabeled methylated products corresponding to compounds 2, 4, and 6. Such assays have been performed with isolated yeast mitochondria containing yeast Coq3p (Fig. 3) (43), yeast mitochondria containing rat Coq3p (Fig. 4), cell free extracts of E. coli (30), and with purified UbiG polypeptide (Fig.  5). In each case, the presence of either Coq3 or UbiG is required to observe in vitro O-methylation, and both Coq3p and UbiG methylate all three substrates.
These assays showed that methylation of 5 by yeast mitochondria required NADH. A similar requirement was observed for the O-methylation of 5 by E. coli extracts (38) and rat liver mitochondria (65). It is likely that NADH provides the reducing equivalents for the generation of the hydroquinone. Accordingly, the purified UbiG O-methyltransferase also requires 5 to be present in the reduced form (Fig. 5). All three compounds thus contain a similar catechol functional group.
The O-methylation of the farnesylated analogs of Q-intermediates by yeast and rat Coq3 and E. coli UbiG is interesting because the naturally occurring quinone species in each of these organisms is different. In yeast, the prenyl tail length (n) is 6; in E. coli, n ϭ 8; and in rats, n ϭ 9 or 10. Additionally, Q biosynthesis can be restored in coq3 null mutants by the human COQ3 homolog. 2 Therefore, it is likely that the human Coq3p recognizes the farnesylated species as well. Such promiscuity is not uncommon in Q biosynthesis because the pHB: polyprenyldiphosphate transferase from rats can recognize other aromatic precursors (66,67), and in yeast, it can utilize polyprenyl groups ranging from n ϭ 5 to n ϭ 10 (68). Also, the C-methyltransferase enzyme in E. coli carries out steps in both Q and menaquinone biosynthesis (31).
A low degree of substrate specificity is also seen for the enzyme, catechol-O-methyltransferase (COMT). COMT is known to methylate numerous neurotransmitters (dopamine, norepinephrine, and epinephrine), their hydroxylated derivatives, and other analogs (69). Both COMT and Coq3/UbiG enzymes require a divalent cation, but comparison of their primary amino acid sequences fails to reveal any homology aside from the AdoMet-dependent methyltransferase motifs. The recent structure of COMT from rat liver (70) provides insight into the mechanism for the O-methylation reaction. The O-methyltransferase in Q biosynthesis may rely on a similar mechanism as the one reported for COMT. Subcellular fractionation localizes Coq3p to the mitochondria. These data confirm and extend previous results that demonstrated import of the yeast Coq3p precursor into the mitochondria in vitro, and showed that such import required a membrane potential (30). The N terminus of the precursor Coq3p contains a putative mitochondrial leader sequence (71,72), which is proteolytically cleaved upon import to produce the mature form (30). The submitochondrial localization of Coq3p was also determined (Fig. 6). Mitochondrial fractionation and protease protection experiments coupled with Western analysis demonstrated that Coq3p was a peripherally associated protein of the inner mitochondrial membrane. This evidence localizes Coq3p and therefore the site for both O-methylation steps of Q biosynthesis within the mitochondrial matrix.
The intracellular site(s) for Q biosynthesis in eukaryotes is still not elucidated. Studies in yeast show that the hexaprenyldiphosphate synthase and the pHB:polyprenyldiphosphate  1, 3, and 5), were grown to saturation, and the cells were collected, lysed, and fractionated by standard methods. Crude mitochondria were then purified over Nycodenz gradients (see under "Experimental Procedures"). Samples (5 g of protein) from the cytoplasmic fractions (lanes 1 and 2), the crude mitochondrial fractions (crude mito) (lanes 3 and 4), and the Nycodenz purified mitochondrial fractions (pure mito) (lanes 5 and 6) were separated by SDS-polyacrylamide gel electrophoresis. The gel was transferred to nitrocellulose for Western analysis by chemiluminescence detection using antibodies to Coq3p(␣Coq3p) and the mitochondrial marker protein, cytochrome c 1 (64). B, mitoplasts from yeast strain W3031A were generated and either sonicated or treated with 0.1 M Na 2 CO 3 , pH 11.5, incubated on ice, and centrifuged as described (see under "Experimental Procedures"). 10 g of protein from each resultant supernatant (S) and pellet (P) fraction were analyzed by SDS-polyacrylamide gel electrophoresis, transferred for Western analysis, and probed via chemiluminescence detection using antibodies to Coq3p, cytochrome c 1 , Hsp 60, or F 1 ␤ATPase. C, intact mitochondria (left four lanes), mitoplasts, or mitoplasts containing 1% Triton X-100 were treated with increasing concentrations of proteinase K (0, 10, 25, 50, and 100 g/ml). Samples of each (1 g) were analyzed by SDS-polyacrylamide gel electrophoresis, transferred for Western analysis, and probed via chemiluminescence detection using antibodies to Coq3p, cytochrome b 2 , F 1 ␤ATPase, or Hsp 60. transferase activities reside in mitochondria (73), and both proteins contain typical mitochondrial leader sequences (13,14). Recently, the yeast COQ5 gene encoding the C-methyltransferase was localized to mitochondria (18,19). The Coq7 (Cat5/Clk-1) polypeptide, which is required in one or more hydroxylase steps in Q biosynthesis (21), was also found in the mitochondria (20). The COQ3 gene product from Arabidopsis was recently localized to the membrane fraction of mitochondria (74). Also, it was previously shown that the O-methyltransferase responsible for converting 5 to Q in rat liver was localized to the inner membrane of the mitochondria (65). However, studies with rat liver show Q biosynthesis occurring in the endoplasmic reticulum-Golgi system (75)(76)(77). These results conflict with earlier studies that indicate that Q is synthesized solely in the mitochondria (65,78,79). The ability of the rat Coq3p to rescue a yeast coq3 mutant (34) suggests that it must be present in the mitochondria of yeast and of rats as well. This conclusion is further supported by the rescue of a coq3⌬ mutant with the E. coli homolog, ubiG, on a single copy plasmid that required that UbiG contain a mitochondrial targeting sequence at the N terminus (30). Although redistribution of the mitochondrial targeted protein fumarase has been reported (80), this requires a cotranslational insertion mechanism that is not required for Coq3p.
UbiG can function as a soluble enzyme. Earlier studies showed that UbiG activity was associated with the E. coli plasma membrane, but it could be solubilized (30,38). This differs from yeast and higher eukaryotes, in which the corresponding homolog, Coq3, appears tightly associated with the inner mitochondrial membrane. Our attempts to solubilize Coq3p activity by sonication or detergent treatments have been unsuccessful. However, activity for the second O-methyltransferase in rat liver mitochondria was solubilized by treatment with Triton X-100 (65). The native molecular weight for the enzyme in those studies was not determined.
Unlike UbiG, which is readily purified as an active soluble enzyme, overexpression of Coq3p in E. coli produced no active enzyme and failed to rescue the ubiG growth defect in E. coli. These observations suggest that Coq3p may require additional polypeptides that 1) may function to keep it peripherally associated with the membrane, or 2) may function in a possible uncharacterized regulatory manner not present in prokaryotes. In either case, these additional polypeptides evidently are required for activity. The evidence for a possible complex in Q biosynthesis in eukaryotes is further supported by the lack of O-methyltransferase activity in other coq null mutants 3 that may lack the required "additional" proteins. In Nocardia lactamdurans, the biosynthesis of cephamycin C involves the interaction of two proteins, a hydroxylase and a methyltransferase, encoded by the genes cmcI and cmcJ, respectively, that are required for function (81). The sequence of hydroxylation and methylation in cephamycin C biosynthesis is similar to Q biosynthesis. The possibility of a protein complex involved in Q biosynthesis will require further study.