Demethoxy-Q, An Intermediate of Coenzyme Q Biosynthesis, Fails to Support Respiration in Saccharomyces cerevisiae and Lacks Antioxidant Activity*

Caenorhabditis elegans clk-1 mutants cannot produce coenzyme Q9 and instead accumulate demethoxy-Q9 (DMQ9). DMQ9 has been proposed to be responsible for the extended lifespan of clk-1 mutants, theoretically through its enhanced antioxidant properties and its decreased function in respiratory chain electron transport. In the present study, we assess the functional roles of DMQ6 in the yeast Saccharomyces cerevisiae. Three mutations designed to mirror the clk-1 mutations of C. elegans were introduced into COQ7, the yeast homologue of clk-1: E233K, predicted to disrupt the di-iron carboxylate site considered essential for hydroxylase activity; L237Stop, a deletion of 36 amino acid residues from the carboxyl terminus; and P175Stop, a deletion of the carboxyl-terminal half of Coq7p. Growth on glycerol, quinone content, respiratory function, and response to oxidative stress were analyzed in each of the coq7 mutant strains. Yeast strains lacking Q6 and producing solely DMQ were respiratory deficient and unable to support 6either NADH-cytochrome c reductase or succinate-cytochrome c reductase activities. DMQ6 failed to protect cells against oxidative stress generated by H2O2 or linolenic acid. Thus, in the yeast model system, DMQ does not support respiratory activity and fails to act as an effective antioxidant. These results suggest that the life span extension observed in the C. elegans clk-1 mutants cannot be attributed to the presence of DMQ per se.

in the tail of Q varies depending on the species; humans, Caenorhabditis elegans, and Saccharomyces cerevisiae produce Q 10 , Q 9 , and Q 6 , respectively. Q is reversibly reduced (Q hydroquinone) and oxidized (Q) as it transports electrons from complex I or II to the cytochrome bc 1 complex in the inner mitochondrial membrane. Q hydroquinone is synthesized in the mitochondria, where it acts not only in respiration but also in fatty acid ␤-oxidation (2) and uridine synthesis (3). Reports point to a role for Q in the regulation of the permeability transition pore as well (4). Q is transported to other intracellular membranes and to the plasma membrane. It is thought to act in these nonmitochondrial sites as a lipid-soluble antioxidant (5,6). Plasma membrane Q is known to be involved in a trans-plasma-membrane electron transport system in which external impermeable oxidants are reduced and intracellular NADH serves as an electron donor (7,8).
The clk-1 gene of C. elegans is an orthologue of yeast COQ7 (15). Similar to the coq7/clk-1 orthologues from rat (16) and human (17), C. elegans clk-1 rescues the Q-biosynthetic defect of a yeast coq7-null mutant, indicating functional conservation. Escherichia coli lacks a COQ7 homologue and instead relies on the ubiF gene product, a flavin-dependent monooxygenase unrelated to COQ7 that is essential for the hydroxylation of DMQ 8 (18). Expression of the Pseudomonas aeruginosa orthologue of Coq7 restored Q biosynthesis in an E. coli ubiF mutant (19). Expression of the C. elegans clk-1 homologue also rescues Q biosynthesis in an E. coli ubiF mutant (20). The Coq7/Clk-1 polypeptide was identified as a member of the di-iron carboxylate family of proteins, including methane monooxygenase, reinforcing the idea that Coq7/Clk-1 polypeptides catalyze the hydroxylation of DMQ, the penultimate step of Q biosynthesis (19).
The free-living nematode C. elegans has provided a powerful model system for the study of development and aging. Mutations in clk-1 lead to delayed embryonic and postembryonic development, slowing of adult behaviors, resistance to UVinduced stress, and a long life span (21)(22)(23). The clk-1 animals fail to synthesize Q 9 but instead produce significant amounts of the Q biosynthetic intermediate DMQ 9 (24), rhodoquinone-9 (RQ 9 ), and they rely on Q 8 , obtained from their diet of E. coli, for growth and fertility (25)(26)(27). The long life span and delayed embryogenesis (28) of clk-1 mutants observed on the standard E. coli diet has been attributed to the presence of DMQ 9 , because it is believed to have fewer pro-oxidant properties than Q and is thought to function less efficiently in respiratory chain electron transport (29). However, Q 8 and RQ 9 are other components that may influence life span in C. elegans (30) and are confounding factors in determining the functional role of DMQ 9 in respiration and as an antioxidant.
In this study, we used the simpler model organism S. cerevisiae to study the effects of clk-1 mutations on growth, respiratory function, and quinone levels. The yeast model enables growth in the absence of dietary or exogenous Q and in the absence of RQ, which is known to function in anaerobic respiration and could possibly function in aerobic respiration as well (31). The mutations introduced into the COQ7 gene were designed to recapitulate the C. elegans clk-1 mutations and included: E233K (e2519), predicted to disrupt the di-iron carboxylate site thought to be required for hydroxylase activity (19); L237Stop (qm30), deleting 36 amino acid residues from the carboxyl terminus; and P175Stop (qm51), deleting the carboxyl-terminal half of Coq7p. This study shows that yeast strains producing only DMQ 6 are respiration-deficient and unable to support either NADH-cytochrome c reductase or succinatecytochrome c reductase activities. Steady-state levels of cytochromes c 1 and c are dramatically decreased in yeast containing DMQ 6 and lacking Q 6 . Furthermore, the presence of DMQ 6 does not protect cells against oxidative stress generated by H 2 O 2 or linolenic acid. The results presented here indicate that DMQ fails to support respiratory activity and is a less effective antioxidant than Q and suggest that the life span extension effects observed in the C. elegans clk-1 mutants cannot be attributed solely to the presence of DMQ.
Plasmid Construction-The plasmids used in this study are listed in Table I. pNMQ71 contains the wild-type COQ7 gene in the pRS316 low copy vector (10,36). PCHQ71 was created by digesting pNMQ7 (10) with ClaI and HindIII and ligating the fragment with pRS316 that had been similarly treated. Primer mediated mutagenesis of PCR fragments was performed with the plasmid pNMQ71. To create pE233K, a G-to-A substitution was made in the primer sequence that corresponded to position 697 in the yeast open reading frame. The forward inside primer pQ7XBAR (5Ј-GGATGACAAGCTAGAGCATCTAGACACCG-3Ј) was used with the vector primer pQ7HIN (5Ј-GACGGTATCGATAAGCTT-TCTTTTAATTAC-3Ј), whereas the inside reverse primer pQ7XBA (5Ј-GCGGTGTCTAGATGCTCTAGCTTGTCATCC-3Ј) was used with a primer corresponding to positions 218 -249 of the open reading frame, pQ7SPH (5Ј-AACCTGAGCATGCTCCCAAGTGTCAGAATT-3Ј). This results in a Glu-to-Lys point mutation at amino acid 233 in the encoded yeast polypeptide. To generate pL237Stop, primers contained an insertion of the TAA stop codon at positions 709 -711, thus creating a stop codon at amino acid 237. The forward inside primer pQ7MUT1 (5Ј-G-ACGAGCTAGAGCATTAAGACACCGCTATC-3Ј) was used with pQ7-HIN, whereas the reverse inside primer pQ7MUT2 (5Ј-GATAGCGGT-GTCTTAATGCTCTAGCTCGTC-3Ј) was used with pQ7SPH. In C. elegans, the qm51 allele is a mutation in the splice-site acceptor, resulting in the translation of intron 2, which contains a stop codon after seven amino acids. To create the ochre mutation of pP175Stop, the primer design included an insertion of the first eight amino acids of C. elegans intron 2, ending in a TAA stop codon at aa 175 of the yeast sequence. The forward inside primer pQ7MUT3-2 (5Ј-GGGAAGAAGTTTGTGG-ATTTGCTGTAAGCTGGTACC-3Ј) was used with pQ7HIN, whereas FIG. 1. Coenzyme Q 6 biosynthesis in yeast. The first compound in this scheme is 3-hexaprenyl-4-hydroxybenzoic acid (HHB), which accumulates in each of the coq-null mutants coq3-coq8. HHB is converted to 3,4-dihydroxy-5-hexaprenylbenzoic acid (DHHB) by an unidentified enzyme. The O-methyltransferase encoded by COQ3 catalyzes the next step to produce 3-methoxy-4-hydroxy-5-hexaprenylbenzoic acid. Successive modifications to the head group results in 5-demethoxyubiquinone (DMQ 6 ). Coq7p, a monooxygenase, converts DMQ 6 to 5-demethylubiquinone, which is methylated by the Coq3 O-methyltransferase to produce Q 6 . the reverse inside primer pQ7MUT4 -2 (5Ј-TTACAGCAAATCCACAA-ACTTCTTCCCCATTGCAAA-3Ј) was used with pQ7SPH.
For each construct, the first round of PCR using pNMQ71 as a template resulted in two PCR products that overlap and contain the same introduced mutation. These overlapping products underwent another round of PCR, with each product using the other as template (37). The final PCR product was digested with SphI and HindIII and ligated into pNMQ71 that had been similarly treated. Each of the mutations was also transferred to multicopy vectors by excising the ClaI and HindIII segment and ligating the inserts into pRS426 (36). These were designated pmE233K, pmL237Stop, and pmP175Stop. All constructs were transformed into competent DH5␣ and amplified.
A truncation representing a true L237Stop C-terminal deletion, ptdL237Stop, was created. The first PCR generated two overlapping fragments that were subsequently used in a second round of PCR to create the final product. This process removed the fragment from position 709 to the penultimate codon, maintaining the original STOP codon. The first fragment was obtained with the forward primer pQ7-d30 -1F (5Ј-AGCTCGGAATTAACCCTCACTAAAGGGA-3Ј) and the reverse primer pQ7d30 -1R (5Ј-TCTGGTGGTTATAGATGCTCTAGCTC-GTCA-3Ј). The second fragment was obtained with the forward primer pQ7d30 -2F (5Ј-GAGCATCTATAACCACCAGAAAGTGGCATA-3Ј) and the reverse primer pQ7d30 -2R (5Ј-ACGTTGTAAAACGACGGCCAGT-GAA-3Ј). pQ7d30 -1F and pQ7d30 -2R were used as primers for the second PCR reaction. Each of the constructs described (Table I) was sequenced and found to be identical to wild type except for the sitedirected changes specified by the primers.
Rescue Assays for Growth on a Nonfermentable Carbon Source-The constructs pE233K, pL237Stop, pP175Stop, or their multicopy counterparts were each placed into CEN.MP3-1A, a yeast strain harboring a deletion in the COQ7 gene. Transformants were selected on SD-Ura plates. Samples of each were streaked onto YPG and grown for 4 days at 30°C. For liquid growth, yeast strains were grown overnight in 5 ml of SD-Ura. The cultures were then diluted into 50 ml of YPG (A 600 ϭ 0.1). Growth was monitored by A 600 measurements.
Quinone Isolation and Identification by HPLC/ECD and Mass Spectrometry-CEN.MP3-1A harboring each of the mutant constructs were grown in SD-Ura and harvested at either log phase or stationary phase. Crude mitochondria were isolated as described previously (38). Lipid extractions of whole cells and quinone quantification were performed as described previously (40). Quinones were quantified in isolated mitochondria after SDS solubilization and lipid extraction as described previously (39). In both cases, Q 9 was used as an internal standard; for whole-cell extractions, 1 g of Q 9 was added per gram (wet weight) of cell pellets, and 500 ng of Q 9 was added per milligram of mitochondrial protein. Lipid components were separated by a Beckmann 166-126 HPLC system equipped with a 15-cm Kromasil C-18 column in a column oven set to 40°C, with a flow rate of 1 ml/min and a mobile phase containing 88:24:10 methanol/ethanol/2-propanol and 13.4 mM lithium perchlorate. Quinones were quantified with an ESA Coulochem III electrochemical detector (ECD) and a 5010 analytical cell (E1, Ϫ500 mV; E2, 500 mV). Hydroquinones present in samples were oxidized with a pre-column cell set in oxidizing mode (E, ϩ500 mV). Q 6 and Q 9 were quantified with external standards. The Q 6 external standard was used to quantify DMQ 6 , assuming a similar redox response in the ECD.
For mass spectrometry analysis, 1 liter of SD-Ura-His cultures were inoculated and harvested after 2 days of growth. A 600 ranged from 8.3-8.8 A 600 /ml. Mitochondria were isolated and lipid extracts were resolved over the HPLC/ECD system. The peaks corresponding to Q 6 and DMQ 6 were collected and dried down under nitrogen gas. The samples were analyzed with a PerkinElmerSciex Instruments API III triple quadropole mass spectrometer fitted with a heated nebulizer at 450°C. The solvent was 0.33% H 2 O in acetonitrile at a flow rate of 0.1 ml/min. Instrument conditions for multiple reaction monitoring analysis were 0.3 step size; dwell, 900 ms; scan speed, 4.0 s; and oriface at 50 V; cgt, 220. These samples were compared with Q 6 obtained from Sigma and DMQ 8  Mitochondrial Respiratory Chain Activities-Freshly obtained mitochondria were assayed for NADH-cytochrome c reductase and succinate-cytochrome c reductase activities. For assays of other respiratory chain activities, mitochondrial samples were subjected to one freezethaw cycle. All assays were performed in triplicate.
NADH-DCIP reductase activity was measured as described with minor modifications (42). The reaction was performed at 30°C with stirring. Mitochondrial protein (10 -50 g) was incubated in reaction buffer (20 mM MES-KOH, pH 6.5, 1 mM EDTA, and 200 mM KCl) for 3 min. The reaction was initiated with 100 M DCIP, without NADH; after 1 min, 200 M NADH was added. Specific NADH-DCIP reductase activity was determined by subtracting the initial linear rate before NADH addition from the rate after NADH addition. The specific activity of DCIP reduction was determined at 600 nm with an extinction Succinate-decylubiquinone reductase activity sensitive to malonate was used to measure complex II activity (43). Activity was measured in an assay volume of 1 ml with 40 mM sodium phosphate, pH 7.4, 40 mM sodium succinate, 100 M decylubiquinone, 250 M potassium cyanide, and 2 g/ml antimycin A. Samples of mitochondria (20 -50 g of protein) were incubated in the assay buffer for 3 min. The reaction was initiated with the addition of decylubiquinone; after 1 min, succinate was added. As a control, some samples were pre-incubated with 4 mM sodium malonate and this rate was subtracted as background. Reactions were monitored via spectrophotometric measurements of absorbance at 280 nm minus 340 nm. The specific activity was determined with an extinction coefficient of 14.5 mM Ϫ1 ⅐cm Ϫ1 .
NADH-cytochrome c reductase activity was measured in an assay volume of 1 ml in 40 mM sodium phosphate, pH 7.4, 0.2 mM NADH, 50 M cytochrome c, and 250 M potassium cyanide. Samples of mitochondria (10 -50 g of protein) were incubated in the assay buffer for 3 min. The reaction was initiated by addition of cytochrome c; after 1 min, NADH was added. The reaction was monitored via spectrophotometric measurements of absorbance at 550 nm minus 540 nm. The specific activity was determined with an extinction coefficient of 18.5 mM Ϫ1 ⅐ cm Ϫ1 . All assays were performed in triplicate. In some experiments, the activity was measured with the addition of exogenous 10 M Q 3 and 10 M DMQ 3 after a 10-min incubation. Given that both compounds were prepared in ethanol, controls with the solvent were performed. To check the nature of cytochrome c reduction, 300 units/ml of bovine superoxide dismutase was added as control. Succinate-cytochrome c reductase and cytochrome c oxidase activities were measured as described previously (44).
Superoxide production was determined in mitochondrial samples by the reduction of acetylated cytochrome c according to a method described previously (45). The reaction was measured in final volume of 1 ml in 40 mM sodium phosphate, pH 7.4, 0.2 mM NADH, 40 mM sodium succinate, 10 M acetylated cytochrome c, 250 M potassium cyanide, and mitochondria samples (50 g of protein). The reaction was initiated by addition of cytochrome c and after 1 min, NADH and succinate were added. The reaction was monitored in a manner similar to that described for NADH-cytochrome c reductase. Bovine superoxide dismutase (300 units/ml) was added in control experiments. All assays were performed in triplicate.
Analysis of Oxidative Stress Sensitivity by Colony Counting-Yeast cells were cultured in SDC or SD-Ura media, harvested at exponential growth phase (A 600 nm ϭ 1) or early stationary growth phase (A 600 nm ϭ 4 -5), washed and resuspended in sterile 100 mM sodium phosphate buffer, pH 6.2, with 0.2% glucose, at 180 ϫ 10 6 cells/ml. Cell suspensions were incubated with different concentrations of H 2 O 2 and linolenic acid for 4 h. Samples from each treatment were diluted and spread onto YPD plates. After 2 days at 30°C, colonies were counted. For each treatment and strain, three plates were counted.
Measurement of Superoxide Anion Generation by Flow Cytometry-Yeast cells were cultured until stationary phase and treated to produce oxidative stress as described above. Cells treated for 4 h with H 2 O 2 (5 and 10 mM) or linolenic acid (750 M) were harvested, washed, and resuspended at 2 ϫ 10 6 cells/ml in phosphate-buffered saline. To measure cellular superoxide levels, cells were incubated with 4 M hydroethidine (HE) for 30 min. HE is oxidized by superoxide, resulting in a product that has fluorescent properties similar to those of ethidium and that binds to DNA (46 -48). Cellular levels of this oxidized product were measured by flow cytometry after excitation at 488 nm with an argon laser and detection of emitted red fluorescence (Ͼ650 nm), with a Coulter Epics XL flow cytometer. The sample fluid pressure was adjusted to acquire ϳ200 events per second.

RESULTS
Growth Characteristics of Yeast coq7 Mutant Strains in Glycerol Media-Mutations were introduced into the yeast COQ7 gene to mirror each of three C. elegans clk-1 mutant alleles described previously(e2519, qm30, qm51; Table I) (15). Primermediated mutagenesis was used to generate the constructs in both low and multicopy plasmids. These plasmids were transformed into the coq7⌬ strain, CEN.MP3-1A. The transformants were tested for their ability to grow in liquid cultures containing glycerol as the sole carbon source (Fig. 2). The CEN.MP3-1A yeast strain harboring pL237Stop (encoding a truncated Coq7p lacking 36 residues of the carboxyl terminus) was capable of growing on glycerol after a long lag period (Fig.  2). CEN.MP3-1A yeast harboring pE233K (predicted to disrupt the di-iron site) and pP175Stop (lacking the carboxyl-terminal half of Coq7p) failed to grow ( Fig. 2A). These two transformants were also incapable of growth on media containing ethanol as the sole nonfermentable carbon source (data not shown). Therefore, the lack of growth on glycerol is not a specific consequence of a lack of Q 6 as an electron acceptor for glycerolphosphate dehydrogenase (GUT2) (49) but is caused by a general lack of mitochondrial electron transport mediated by Q 6. We further investigated the yeast strain CEN.MP3-1A: pL237Stop, which displayed growth after a lag period. When samples of this strain were reinoculated into fresh glycerol media, they did not exhibit a lag and grew like wild type (data not shown). The effect of overexpressing the L237Stop mutation (from the multicopy plasmid pmL237Stop) was dramatic in that this plasmid conferred almost wild-type levels of growth (Fig. 2B). It seemed likely that read-through of the introduced stop codon in pL237Stop was responsible for the observed growth adaptation. To investigate this possibility, a C-terminal deletion was prepared by truncating the C-terminal coding region at position 237, and this construct was designated ptdL237Stop. CEN.MP3-1A was transformed with ptdL237-Stop and tested for YPG growth. No rescue of growth was observed on this nonfermentable carbon source (Fig. 2). Therefore, read-through of the introduced stop codon was deemed the most likely explanation for the adaptation phenomenon. Such informational suppression is well documented (50 -52) and also accounts for the reproducibility of the lag period observed in response to selective pressure (Fig. 2) (53).
Identification and Quantification of DMQ 6 and Q 6 in coq7 Mutants-To analyze the relationship between growth in YPG and Q synthesis, lipid extracts were prepared from yeast bearing the coq7 mutant plasmids, and quinones were separated by HPLC and quantified by ECD. Similar trends were observed for quinone content in both whole cells (data not shown) and isolated mitochondria (Table II). Mitochondria isolated from CEN.MP3-1A yeast harboring the wild-type COQ7 gene and grown in glucose-containing media produced predominantly Q 6 , (1950 ng of Q 6 /mg of protein), although a significant amount of DMQ 6 (432 ng/mg of protein) was also present. The coq7 null mutant strain harboring empty vector (CEN.MP3-1A:pRS316), or harboring pP175Stop failed to synthesize either Q 6 or DMQ 6 . Q 6 was not detectable in CEN.MP3-1A:pE233K; however this strain accumulated DMQ 6 , 861 ng/mg of protein.
CEN.MP3-1A yeast bearing pL237Stop produced exclusively DMQ 6 (406 ng/mg of protein) when grown in media containing glucose. The content of Q 6 in some of the mutants was altered in response to growth in media containing glycerol. Mitochondria isolated from CEN.MP3-1A harboring either pL237Stop or pmL237Stop after culture in YPG accumulated significant amounts of Q 6 , ranging from 310 to 423 ng of Q 6 /mg of protein, although DMQ 6 was still the predominant quinone isolated from these strains. No other quinone containing intermediates were detected.
To verify the identities of the compounds quantified by HPLC/ECD, these fractions were collected, purified, and analyzed by mass spectrometry (Experimental Procedures). The expected daughter tropylium ions were present (a 197.2 ion for Q 6 and 167.2 for DMQ 6 ), verifying their identities.
Analysis of Respiratory Chain Activities in Yeast Mutants Producing Only DMQ 6 -To analyze the role of DMQ 6 in respiration, several segments of the mitochondrial respiratory chain were assayed for activity (Table II). The CEN.MP3-1A: pNMQ71 strain was used as a positive control, and these values were set at 100% as a comparison. Yeast that were incapable of synthesizing either Q 6 or DMQ 6 , CEN.MP3-1A: pRS316, and CEN.MP3-1A:pP175Stop displayed a decrease in all respiratory chain activities, ranging from 3 to 16% of the wild-type activities. The most dramatic effect was seen in the assays of succinate-cytochrome c reductase and NADH-cytochrome c reductase. Both of these assays rely on endogenous Q 6 present in isolated mitochondria as the electron carrier.
The CEN.MP3-1A strain harboring the plasmids pE233K or ptdL237Stop produces DMQ 6 exclusively and shows a decrease in all mitochondrial respiratory chain activities. This was also the finding for the CEN.MP3-1A:pL237Stop mutant cultured in glucose. However, in each case, NADH-DCIP and succinatedecylubiquinone reductase activities were significantly higher than in the CEN.MP3-1A:pRS316 strain, ranging from 30 to  The results of Table II suggest that the NADH cytochrome c reductase activity is similarly defective in mitochondria isolated from either the coq7 null or the DMQ 6 -producing yeast strains. Because supplementation of the growth media with exogenous Q 6 restores this activity, it was of interest to test whether in vitro addition of Q or DMQ analogs to isolated mitochondria would also act to restore activity. As shown in Table III, addition of either Q 3 or DMQ 3 to the assay produced only a very modest increase in NADH cytochrome c reductase activity. Most of this increase was inhibited by the addition of superoxide dismutase, suggesting that when either Q 3 or DMQ 3 are added to the assay, most of the cytochrome c reduction is mediated by superoxide and not via the bc 1 complex.
Supplementation of growth media with 2 M exogenous Q 6 restored NADH-DCIP reductase activity, and enhanced the other respiratory chain activities in the CEN.MP3-1A:pRS316 and in the CEN.MP3-1A:pL237Stop strains. Those yeast strains capable of producing significant amounts of Q 6 , in response to YPG culture media, namely CEN.MP3-1A: pL237Stop and CEN.MP3-1A:pmL237Stop, showed respiratory enzyme activities much higher than those strains exclusively producing DMQ 6 .
The Effect of DMQ 6 on Steady-state Levels of Mitochondrial Polypeptides-Our previous studies with coq7 null mutant yeast showed that Q 6 is required for stability of the cytochrome c 1 polypeptide (39). To determine whether the exclusive presence of DMQ 6 affected steady-state levels compared with Q 6producing strains, Western blot analysis of mitochondrial polypeptides, including porin, F 1 ␤-ATPase, cytochrome c 1 , and cytochrome c, were performed (Fig. 3A). Densitometric analysis showed that steady-state levels of cytochrome c, a substrate of complexes III and IV, and cytochrome c 1 , a subunit of the bc 1 complex, were dramatically decreased compared with porin and the F 1 ␤ subunit of complex V (Fig. 3B). Levels of cytochromes c and c 1 were fully restored when the culture media was supplemented with exogenous Q 6 . These results indicate that DMQ 6 is unable to provide for stable steady-state levels of the c-type cytochromes.
Sensitivity of coq7 Yeast Mutants to Oxidative Stress-It has been shown that the yeast coq null mutant strains are hypersensitive to oxidative stress caused by treatment with linolenic acid (11,55). These results indicate that Q 6 functions as an important cellular antioxidant, required to protect against the toxicity induced by polyunsaturated lipid autoxidation. To test the efficacy of DMQ 6 as an antioxidant, yeast strains producing Q 6 , DMQ 6 or no quinones were subjected to oxidative stress caused by treatment with either H 2 O 2 or linolenic acid (Fig.  4A-D). Cell viability was analyzed during both log and stationary phase. All strains showed a decrease in cell viability with increasing concentrations of H 2 O 2 (Fig. 4, A and B). Although the wild-type strain was significantly more resistant than the different mutant strains, the CEN.MP3-1A strain harboring pL237Stop was slightly more resistant than the other mutants. Treatment with linolenic acid produced a different effect on the cell viability (Fig. 4, C and D). Wild-type yeast were not affected by this treatment, whereas all coq7 mutants were sensitive to linolenic acid at both log and stationary phase, although the cells at log phase showed greatly increased sensitivity. It was interesting that CEN.MP3-1A:pE233K, which produces only DMQ 6 , was more sensitive to linolenic acid treatment at stationary phase than was the coq7 null mutant CEN.MP3-1A, which fails to produce Q 6 or DMQ 6 . These data indicate that DMQ 6 fails to act as an effective antioxidant. It was intriguing that the CEN.MP3-1A:pL237Stop mutant was significantly more resistant than the other mutants tested; the effect was most pronounced for cells treated with linolenic acid (Fig. 4, C and D).
Given that the CEN.MP3-1A:pL237Stop mutant strain adapts to growth under selective conditions (YPG) and synthesizes Q 6 (as a result of the stop codon read-through), we investigated whether linolenic acid treatment also selects for Q 6 production. CEN.MP3-1A cells harboring pL237Stop, ptdL237-Stop, or pE233K were cultured in SD-Ura and were treated with 750 M linolenic acid in phosphate buffer for 4 h. Analysis of lipid extracts prepared from these cells revealed that CEN.MP3-1A harboring pE233K or ptdL237Stop produced only DMQ 6 (data not shown). However, treatment of the CEN.MP3-1A:pL237Stop yeast strain with linolenic acid led to a distinct accumulation of Q6 (Fig. 5), that correlated well with the increased stress resistance (Fig. 4). It is likely that the stress produced by linolenic acid treatment selects for stop codon read-through, similar to the selection imposed by culture in media containing a nonfermentable carbon source.  (14) a Data shown are the average of three injections Ϯ S.D. from two independent extractions. ND, not detected (detection limit is 0.2 ng of Q 6 per injection); NM, not measured.
b Results are expressed as the average of three assays Ϯ S.D. from two separate experiments. In parentheses are percentages compared with the wild-type construct. NM, not measured. 6 -To analyze the superoxide anion generation by the mitochondrial respiratory chain, the reduction of acetylated cytochrome c was measured in mitochondrial samples using NADH and sodium succinate as electron donors. As with cytochrome c, acetylated cytochrome c can be reduced by superoxide anion; however, the multiple acetyl groups covalently linked to acetylated cytochrome c prevent its interaction with respiratory enzyme complexes (54). Mitochondria isolated from CEN.MP3-1A:pNMQ71 produced the highest amounts of superoxide anion (16.3 Ϯ 1.4 nmol/mg of protein/min). Superoxide production was significantly lower for the coq7 null mutant CEN MP3-1A:pRS316 (5.4 Ϯ 0.5 nmol/mg of protein/min) and for the DMQ 6 -producing strain CEN MP3-1A:pL237Stop (5 Ϯ 0.2 nmol/mg of protein/min). These results argue against the idea that partially assembled respiratory complexes in the Q-less yeast contribute to increased levels of superoxide Cellular superoxide generation during oxidative stress treatment was analyzed with flow cytometry and by monitoring the fluorogenic oxidation of HE, as described under "Experimental Procedures." Conditions that enhance superoxide production in yeast were used to validate this assay. The strain EG110 harbors a sod2-null mutation (56) and hence lacks the mitochondrial manganese superoxide dismutase. This strain accumulated high levels of superoxide compared with the isogenic wild-type strain EG103 (Fig. 6A). In addition, yeast cells were also incubated with antimycin A, a compound that produces superoxide by blocking the high potential cytochrome b in bc 1 complex (57). Wild-type yeast showed low levels of superoxide, which increased upon incubation with antimycin A (Fig. 6A). Antimycin A treatment did not affect superoxide production in the coq7 null strain. The levels of superoxide produced in CEN.MP3-1A mutants harboring pE233K or pL237Stop were significantly higher than wild type, although treatment with  3 All strains were cultured in S.D. selective media until stationary phase. Results are expressed as the average of three assays Ϯ S.D. from two separate experiments. Samples not incubated with Q 3 or DMQ 3 were incubated with the same volume of ethanol. Q 3 or DMQ 3 were incubated with the samples for 5 min at RT. SOD was present during the incubation.  The levels of expression in the wild-type strain were set at 100%. The same membrane was imaged after four exposure times (10, 30, 60, and 120 s). All plates were recorded using GPL mode (Good Laboratory Practice) of the Quantity One (Bio-Rad) software, which does not allow image modifications that change the raw image data. The best relationship between time and intensity was selected.

Superoxide Anion Generation in Yeast coq7 Mutants Producing DMQ
antimycin A did not further increase superoxide levels in these strains. Antimycin A requires functional mitochondria to produce superoxide, providing additional evidence that DMQ 6 fails to support mitochondrial respiratory activity.
Once the HE assay of superoxide was validated, yeast grown to stationary phase was subjected to oxidative stress by treatment with H 2 O 2 (5 or 10 mM) or with linolenic acid (750 M) (Fig. 6B), and HE oxidation was measured. Treatment with H 2 O 2 did not produce a significant increase in superoxide levels in any strain tested. In contrast, linolenic acid treatment produced high levels of HE oxidation in the DMQ 6 -producing strains but not in wild-type or null mutants. The amount of HE oxidation detected in these strains was double the amount detected in the sod2⌬ strain. However, HE oxidation dropped to wild-type levels when the CEN.MP3-1A:pL237Stop mutant was cultured in YPG. Thus, under conditions in which Q 6 is produced, superoxide is produced at minimal levels, even if DMQ 6 is the predominant quinone present. However, after linolenic acid treatment, and in the absence of Q 6 , DMQ 6 seems to have significant pro-oxidant activity.

DISCUSSION
Analysis of clk-1 nematodes has introduced some intriguing questions regarding the functional roles of Q and DMQ in development and aging. In particular, the extended lifespan of the C. elegans clk-1 mutants has been hypothesized to result from the presence of DMQ, because this intermediate is considered to have less pro-oxidant activity than Q and is also thought to be less efficient in mediating electron transport through the respiratory chain (29). However, hypotheses regarding these attributes of DMQ have been difficult to test, because the nematode system is complicated by the contributions of dietary Q 8 from the standard E. coli diet, and by endogenously produced rhodoquinone (RQ 9 ), which accumulates in the clk-1 mutants and is required for anaerobic respiratory metabolism (26). Both Q 8 and RQ 9 constitute other parameters that have the potential to influence lifespan in C. elegans (30). To dissect the functionality of DMQ we turned to a simpler model and employed S. cerevisiae and its clk-1 homologue, COQ7.
Three mutations were introduced into the yeast COQ7 gene to generate analogs of C. elegans clk-1 mutant alleles: E233K, with a disrupted di-iron binding site; L237Stop, missing 36 carboxyl-terminal amino acids; and P175Stop, missing the carboxyl-terminal half of the polypeptide. The E233K substitution for the putative bridging carboxylate ligand of the yeast Coq7 di-iron site is likely to inactivate enzyme activity, in that mutation of the corresponding glutamate ligand to alanine (E273A) in the alternative oxidase from Arabidopsis thaliana resulted in loss of the di-iron center (58). By expressing these coq7 mutations in a coq7 null mutant, yeast strains were generated that lacked Q 6 but produced significant quantities of DMQ 6 . Yeast mutants that produced only DMQ 6 failed to grow on media containing glycerol, a nonfermentable carbon source. Mitochondria isolated from these yeast strains lacked Q 6 , contained DMQ 6 , and failed to support NADH-cytochrome c reductase or succinate cytochrome c reductase activities. The DMQ 6 -producing yeast mutant strains also showed decreased steady-state levels of cytochrome c 1 , a polypeptide component of bc 1 complex, and cytochrome c, a substrate of the bc 1 complex. These data indicate that DMQ 6 fails to support respiratory electron transport in yeast.
To investigate the potential antioxidant properties of DMQ 6 , yeast were subjected to treatment with hydrogen peroxide and linolenic acid. When treated with H 2 O 2 , DMQ 6 -producing yeast mutant strains were found to be as sensitive as the coq7 null mutant. In previous work, yeast coq null mutants have been shown to be exquisitely sensitive to treatment with linolenic acid (11,55). In data presented here, we show that DMQ 6producing yeast mutant strains were even more sensitive to linolenic acid treatment than the coq7 null mutant. The fluorogenic oxidation of HE was used as an indicator of superoxide. This assay showed that superoxide levels were profoundly increased when DMQ 6 -producing yeast were treated with linolenic acid. In contrast, yeast strains producing both DMQ 6 and Q 6 showed enhanced resistance to linolenic acid treatment and decreased levels of superoxide. These results show that the pro-oxidant attributes of DMQ 6 are negated when Q 6 is also present, presumably because the Q 6 /Q 6 H 2 redox couple is capable of being regenerated by the electron transport chain. It is also possible that Q 6 acts to reduce and oxidize DMQ 6 via respiratory complex transhydrogenation reactions (59). These data indicate that DMQ 6 on its own fails to act as an effective antioxidant and, under certain stress conditions, acts as a pro-oxidant and is associated with increased levels of superoxide production.
The apparent pro-oxidant action of DMQ 6 in response to linolenic acid treatment is curious. It is possible that unstable or improperly assembled respiratory complexes are responsible for generating high levels of superoxide. However, mitochondria isolated from either the null coq7 mutant or the DMQ 6producing strains produced similar and very low levels of superoxide, as measured by the reduction of acetylated cytochrome c. Similar results have been described for null coq7 mutants with regard to hydrogen peroxide production in mitochondria (60). These results indicate that sources of oxidative stress in mitochondria are similar for the coq7 null and the DMQ 6 -producing mutants. In yeast, the bc1 complex is believed to generate the highest levels of superoxide, yet the steadystate levels of cytochromes c 1 and c are profoundly decreased in both strains, regardless of the presence or absence of DMQ 6 . This suggests that the bc 1 complexes in these two strains are similarly defective, an idea that is supported by the lack of NADH-cytochrome c oxidoreductase and succinate cytochrome c oxidoreductase activities in both the coq7-null and DMQ 6producing mutants (Table II) and by the inability to recover NADH-cytochrome c reductase activity in response to addition of either Q 3 or DMQ 3 to isolated mitochondria (Table III). However, it is likely that when these strains are treated with linolenic acid, the DMQ 6 -producing strain will uniquely have accumulated DMQ 6 H 2 , because of significant segment I and complex II activities (30 and 46% of wild-type) (Table II). Lipid autoxidation could trigger the oxidation of DMQ 6 H 2 producing the semiquinone radical, which in turn would contribute to superoxide formation and enhance the oxidation of hydroethidium.
In yeast, data suggest that Coq7p is required for a multisubunit Q biosynthetic complex, and mutations in Coq7p are considered to produce the partial or complete disruption of complex integrity and activity. CEN.MP3-1A mutants expressing the Coq7 polypeptide bearing the E233K mutation were able to produce the full-length form of Coq7p (data not shown). Because the polypeptide is stable, it is likely to maintain contacts with other Coq polypeptides or respiratory complexes. This allows for production of DMQ 6 , whereas coq7 null yeast mutants produce only the early intermediate HHB (10).
Is the Coq7/Clk-1 polypeptide part of a multisubunit Q biosynthetic complex in other eukaryotes? Although the answer to this question will depend on its physical characterization, there is evidence that C. elegans CLK-1 can function independently. An E. coli ubiF mutant known to have a defect in the hydroxylation of DMQ 8 was shown to be rescued by expression of the C. elegans clk-1 gene, indicating that the CLK-1 polypeptide is active when expressed on its own in E. coli (20). Mice and C. elegans mutants with deletions in the clk-1 gene continue to produce DMQ 9 although no CLK-1 polypeptide or RNA can be detected (61)(62)(63). This is distinct from the phenotype of coq7null yeast mutants that produce only HHB. These findings indicate that in the mouse and nematode, progression of Q biosynthesis from HHB to DMQ proceeds independently of the CLK-1 polypeptide.
Our finding that DMQ 6 cannot function in yeast mitochondrial electron transport counters previous studies that claimed DMQ 9 supported mitochondrial respiratory chain activities in C. elegans clk-1 mutants (24) and in mouse clk-1Ϫ/Ϫ embryonic stem cells (62). In the nematode model, the activity attributed to DMQ 9 , particularly for complexes II and III, could very well have been influenced by the presence of either dietary Q 8 or RQ 9 . It is noteworthy that in the mouse clk-1Ϫ/Ϫ mutant cells, levels of complexes II and III were drastically reduced, whereas levels of complex I and III activity were only mildly decreased. This is consistent with previous assays of membranes isolated from E. coli ubiF mutants; such mutants produce only DMQ 8 , and partial function of complex I was observed, but DMQ 8 was inactive in complex II activity (64). Given this, it is surprising that DMQ 6 -producing yeast failed to support activity of NADHcytochrome c oxidoreductase activity. Because NADH-DCIP reductase is only partially decreased in DMQ 6 -producing strains, the lack of activity detected for NADH-cytochrome c oxidoreductase assays is likely to result from the inability of DMQ 6 to stabilize bc 1 complex, as judged by the low steadystate levels of cytochromes c and c 1 . Addition of exogenous Q 6 to the growth media restores steady-state c 1 polypeptide levels in coq7 null mutants (39) and, as shown here, in DMQ 6 -producing yeast strains, but neither Q 3 nor DMQ 3 could restore any bc 1 complex activity (Table III) when added in the assay directly. The low levels of complex IV activities detected in DMQ 6producing yeast strains also supports this idea, in that bc 1 complex and complex IV are coordinately regulated. For example, the near elimination of bc 1 complex activity is accompanied by a reduction in complex IV activity to about 15% of wild type (65). Taken together, these data indicate that DMQ 6 -producing strains lack a functional and well-assembled bc 1 complex. However, it is not yet possible to discriminate whether DMQ 6 is a functional substrate for bc 1 complex.
It is clear that DMQ 6 cannot functionally replace Q 6 in the yeast system; in this aspect, our data are in agreement with previous studies in both the mouse and nematode models. Homozygous mclk-1 mouse mutants display embryonic lethality (62,63). C. elegans clk-1 mutants fed a Q-less diet contain significant amounts of DMQ 9 yet are developmentally arrested if deprived of a Q-replete diet as hatchlings or are sterile if deprived of Q during post-dauer development (26,66). Uptake of Q 8 from E. coli and transport of dietary Q 8 to the mitochondria are needed to prevent the arrest and sterility phenotypes (27). These data reveal an interesting dual nature of Q; it is required to sustain larval growth and germline development, yet dietary Q 8 has been shown to shorten life span in adult nematodes. The developmental and reproductive phenotypes of the clk-1 mutants have recently been shown to be unlinked to the longevity phenotype (67). We speculate that the clk-1 mutant nematodes fed Q 8 -replete diets, or rescued by the maternal mitochondrial contribution of Q 9 and the CLK-1 polypeptide, may have parallels to the yeast strains characterized here that produce both Q 6 and DMQ 6 . In this scenario, the small amount of Q 8 (or in the case of maternal rescue, Q 9 ) in the nematode, may act to stabilize respiratory electron transport complexes, support a low level of redox activity (or involve DMQ directly via transhydrogenation reactions (59)), and act to mask the pro-oxidant activity of DMQ 9 . The increased lifespan of the clk-1 nematode would then reflect the lowered respiratory chain activity, stemming either from the decreased level of aerobic respiration (because of less Q) and/or from the increase in anaerobic respiration (hypothetically caused by increased RQ 9 ). This model is consistent with increased reliance on anaerobic metabolism observed in the long lived dauer larvae of C. elegans (68) , the life span extensions observed in the nematode model after RNA interference-mediated down-regulation of mitochondrial respiratory chain components (69,70), and the RNA interference silencing of other COQ genes (71).