Overexpression of the Coq8 Kinase in Saccharomyces cerevisiae coq Null Mutants Allows for Accumulation of Diagnostic Intermediates of the Coenzyme Q6 Biosynthetic Pathway*

Background: Several steps of eukaryotic coenzyme Q biosynthesis are still in question. Results: Yeast coq null mutants overexpressing the Coq8 kinase have stable Coq polypeptides and accumulate new Q intermediates that help diagnose the blocked step. Conclusion: New functions for Coq polypeptides are proposed. Significance: Identification of the blocked step allows for the use of alternate ring precursors that rescue Q biosynthesis in some mutants. Most of the Coq proteins involved in coenzyme Q (ubiquinone or Q) biosynthesis are interdependent within a multiprotein complex in the yeast Saccharomyces cerevisiae. Lack of only one Coq polypeptide, as in Δcoq strains, results in the degradation of several Coq proteins. Consequently, Δcoq strains accumulate the same early intermediate of the Q6 biosynthetic pathway; this intermediate is therefore not informative about the deficient biosynthetic step in a particular Δcoq strain. In this work, we report that the overexpression of the protein Coq8 in Δcoq strains restores steady state levels of the unstable Coq proteins. Coq8 has been proposed to be a kinase, and we provide evidence that the kinase activity is essential for the stabilizing effect of Coq8 in the Δcoq strains. This stabilization results in the accumulation of several novel Q6 biosynthetic intermediates. These Q intermediates identify chemical steps impaired in cells lacking Coq4 and Coq9 polypeptides, for which no function has been established to date. Several of the new intermediates contain a C4-amine and provide information on the deamination reaction that takes place when para-aminobenzoic acid is used as a ring precursor of Q6. Finally, we used synthetic analogues of 4-hydroxybenzoic acid to bypass deficient biosynthetic steps, and we show here that 2,4-dihydroxybenzoic acid is able to restore Q6 biosynthesis and respiratory growth in a Δcoq7 strain overexpressing Coq8. The overexpression of Coq8 and the use of 4-hydroxybenzoic acid analogues represent innovative tools to elucidate the Q biosynthetic pathway.

biosynthetic complex in S. cerevisiae (4,5). The absence of a single Coq polypeptide from the complex causes a drastic diminution of the steady state levels of some Coq proteins. For example, the steady state levels of Coq4, Coq6, Coq7, and Coq9 are decreased in each of the ⌬coq1-⌬coq9 null strains (6). As a result, the same early intermediate 3-hexaprenyl-4-hydroxybenzoic acid (HHB; Fig. 1, path 1) accumulates in each of the ⌬coq3-⌬coq9 strains. Certain point mutations resulting in amino acid substitutions seem to have less impact on the integrity of the Q biosynthetic complex than a null mutation; expression of the inactive Coq7-E194K polypeptide in a ⌬coq7 strain caused accumulation of the expected intermediate demethoxy-Q 6 (DMQ 6 , Fig. 1) (7,8). Apart from this example, the absence of accumulation of biosynthetic intermediates downstream of HHB in ⌬coq strains has hindered our understanding of the Q biosynthetic pathway. Therefore, the precise order of certain biosynthetic steps is still elusive, and the function of Coq4 and Coq9 is not defined.
The yeast COQ8 gene was formerly called ABC1 and was thought to be essential for complex III function (9,10). However, it was later shown that COQ8 was required for Q 6 biosynthesis, and as such, its deletion only affected complex III activity indirectly (11). Coq8 is a matrix protein peripherally associated with the mitochondrial inner membrane (12) and belongs to the "atypical kinases" subgroup of the proteinkinase-like superfamily (13). Mutations in ADCK3, the human orthologue of COQ8, were shown to cause Q 10 deficiency and cerebellar ataxia (13,14). In yeast, Coq8 is essential for phosphorylation of Coq3 and for its association with the Q biosynthetic complex (12,15). In addition, several phosphorylated forms of Coq5 and Coq7 disappear in a yeast strain expressing the G130D mutant form of Coq8 (12), which mimics the pathogenic G272D mutation found in ADCK3 (14). Therefore, Coq8 appears to be a kinase essential for the phosphorylation of several conserved Coq polypeptides and some of these phosphorylated forms likely play a role in the assembly or maintenance of the Q 6 biosynthetic complex. Recent studies indicate that overexpression of COQ8 can have profound effects on Q 6 biosynthesis. Indeed, the overexpression of Coq8 (from now on referred to as Coq8 OE) in a ⌬coq7 strain promoted the accumulation of DMQ 6 (16), implying that all Coq proteins acting upstream of Coq7 in the biosynthetic pathway were stable and active. The effect of Coq8 OE is likely post-transcriptional because COQ4 mRNA levels in the ⌬coq7 strain were not dependent on the level of Coq8 OE (16). Recently, the low steady state level of Coq4 encountered in ⌬coq2, ⌬coq3, ⌬coq5, and ⌬coq7 strains was shown to be restored to wild-type levels by Coq8 OE (17). In the case of a ⌬coq6 strain, Coq8 OE allowed the specific accumulation of 3-hexaprenyl-4-hydroxyphenol (4-HP) (Fig. 1), which led us to identify Coq6 as the monooxygenase responsible for the C5-hydroxylation step (18). This example demonstrates that the accumulation of Q 6 biosynthetic intermediates in ⌬coq strains and the identification of their chemical structure are important for understanding the function of Coq proteins.
In addition to the classic Q biosynthetic pathway emanating from 4-hydroxybenzoic acid (4-HB), S. cerevisiae also makes use of para-aminobenzoic acid (pABA) as a ring precursor for Q 6 biosynthesis ( Fig. 1, path 2) (3,19). Coq2 is able to catalyze the prenylation of 4-HB to yield HHB, as well as the prenylation of pABA to yield 3-hexaprenyl-4-aminobenzoic acid (HAB). We have hypothesized that Coq3-Coq9 enzymes modify both HAB and HHB and that the C4-amino group must be removed from the HAB-derived intermediates to produce Q 6 (3,19). The deamination reaction likely occurs before the C6-hydroxylation step catalyzed by Coq7 because 4-imino-DMQ 6 (IDMQ 6 ) was proposed to be a precursor of DMQ 6 (19). Ring precursors other than 4-HB and pABA can also be used in vivo by S. cerevisiae to synthesize Q 6 . Indeed, 3,4-dihydroxybenzoic acid and vanillic acid bypass a deficiency in the Coq6-mediated C5-hydroxylation reaction and restore Q 6 biosynthesis in coq6 or yah1 mutant strains (18).
In this study, we show that Coq8 OE restores the steady state levels of the Coq proteins in most ⌬coq strains. The stabilization of the Coq polypeptides leads to the accumulation of Q 6 biosynthetic intermediates that allow the diagnosis of the impaired step. We have used this property to demonstrate that several biosynthetic steps are impaired in the ⌬coq4 and ⌬coq9 strains and to gain insights into the deamination reaction. Finally, the use of alternate ring precursors promoted the restoration of Q 6 biosynthesis and respiratory growth for a ⌬coq7 strain.

EXPERIMENTAL PROCEDURES
Yeast Strains and Culture Conditions-S. cerevisiae strains used in this study are listed in Table 1. S. cerevisiae strains were transformed with lithium acetate as described (20,21). YNB without pABA and folate (ϪpABAϪfolate) was purchased from MP Biomedicals. Rich YP medium was prepared as described (22). Dextrose or lactate-glycerol was used at 2%. In preparation for analyses by HPLC-ECD, yeast cells were cultured as described (3) and grown for 18 h at 30°C. Stock solutions of 4-HB analogues at 100 mM were prepared by slowly titrating NaOH (care was taken not to exceed pH 9) until complete dissolution. The solutions were then filtersterilized, and aliquots were kept at Ϫ20°C for several months. The 4-HB analogues were added to ϪpABAϪfolate growth medium at the indicated concentrations. Alternatively, in preparation for analyses by HPLC-MS/MS, yeast cells were cultured in Drop Out Galactose medium (DOGAL) (19) and labeled as described. Briefly, 100 A 600 cells were collected from overnight culture and transferred to fresh medium in the presence of various aromatic ring precursors for 2-4 h at 30°C. Cells were then collected by centrifugation and subject to LC-MS/MS analysis.
Plasmids-Plasmids used in this study are listed in Table 2. COQ7 ORF with its own promoter and terminator was cloned into pRS425 using HindIII and XhoI. Sequencing was used to confirm cloning products.
Lipid Extraction and Detection of Electroactive Compounds by HPLC-ECD-Cellular lipid extraction after addition of the Q 4 standard and detection of electroactive compounds by FIGURE 1. S. cerevisiae Q 6 biosynthetic pathway. Accumulation of Q 6 biosynthetic intermediates is caused by the overexpression of Coq8 in ⌬coq strains. The classic Q biosynthetic pathway is shown in path 1 emanating from 4-HB. Coq1 (not shown) synthesizes the hexaprenyl-diphosphate tail that is transferred by Coq2 to 4-HB to form HHB. R represents the hexaprenyl tail present in all intermediates from HHB to Q 6 . Alternatively, in path 2, pABA is prenylated by Coq2 to form HAB. Both HHB and HAB are early Q 6 intermediates, readily detected in each of the coq null strains (⌬coq3-⌬coq9). The numbering of the aromatic carbon atoms used throughout this study is shown on the reduced form of Q 6 , Q 6 H 2 . Coq8 OE in certain ⌬coq strains leads to the accumulation of the following compounds: 4-AP, 3-hexaprenyl-4-aminophenol; 4-HP, 3-hexaprenyl-4-hydroxyphenol; HHAB, 3-hexaprenyl-4amino-5-hydroxybenzoic acid; HMAB, 3-hexaprenyl-4-amino-5-methoxybenzoic acid; DDMQ 6 H 2 , is the reduced form of demethyl-demethoxy-Q 6 ; IDMQ 6 , 4-imino-demethoxy-Q 6 ; IDMQ 6 H 2 , 4-amino-demethoxy-Q 6 H 2 ; DMQ 6 , demethoxy-Q 6 ; DMQ 6 H 2 , demethoxy-Q 6 H 2 . IDDMQ 6 H 2 , 2-demethyl-4-amino-demethoxy-Q 6 H 2 , and DHHB, 3-hexaprenyl-4,5-dihydroxybenzoic acid, are shown in parentheses to indicate that they have not been detected in this study. Black dotted arrows (from path 2 to path 1) designate the replacement of the C4-amine with a C4-hydroxyl and correspond to the C4-deamination/ deimination reaction. A putative mechanism to replace the C4-imino group with the C4-hydroxy group is shown in blue brackets for IDMQ 6 but could also occur on 2-demethyl-4-amino-demethoxy-Q 6 (IDDMQ 6 ) (not shown). 4-AP and 4-HP, which are formed upon inhibition of the C5-hydroxylation catalyzed by Coq6, are shown in red. Analogues of 4-HB and pABA allowing the bypass of certain steps in Q biosynthesis are indicated in green. Steps defective in the ⌬coq9 strain are designated with a red asterisk. HPLC-ECD with a 5011A analytical cell (E1, Ϫ550 mV; E2, ϩ550 mV) were conducted as described (3). Hydroquinones present in samples were oxidized with a precolumn 5020 guard cell set in oxidizing mode (E, ϩ650 mV). Lipid Extraction and RP-HPLC-MS/MS-Lipid extractions of cells made use of Q 4 as internal standard, and all LC-MS/MS analysis were performed as described previously (19). Briefly, a 4000 QTRAP linear MS/MS spectrometer from Applied Biosystems (Foster City, CA) was used. Analyst version 1.4.2 software (Applied Biosystems) was used for data acquisition and processing. A binary HPLC solvent delivery system was used with a phenyl-hexyl column (Luna 5u, 100 ϫ 4.60 mm, 5-m, Phenomenex) for yeast extracts. The mobile phase consisted of Solvent A (methanol/isopropyl alcohol, 95:5, with 2.5 mM ammonium formate) and Solvent B (isopropyl alcohol, 2.5 mM ammonium formate). The percentage of Solvent B was increased linearly from 0 to 5% over 6 min, and the flow rate was increased from 600 to 800 l/min. The flow rate and mobile phase were changed back to initial condition linearly by 7 min. All samples were analyzed in multiple reaction monitoring mode (MRM).
Purification of Q Biosynthetic Intermediates and High Resolution Mass Spectrometry Measurements-The compounds were purified from yeast cells as described previously (3). Samples in methanol were diluted with 90% acetonitrile and 0.2% formic acid and were infused into the nanospray source of a discovery ORBITRAP instrument (Thermo Fischer Scientific) at a flow rate of 0.5 l/min for high resolution mass spectrometry analyses (3).

Overexpression of Coq8 Restores Steady State Levels of Coq
Proteins in Mitochondria of ⌬coq Strains-Immunodetection on whole cell lysates revealed that Coq4, Coq7, and Coq9 steady state levels were increased by Coq8 OE in ⌬coq3 and ⌬coq5 strains (Fig. 2, A and B). Similarly, Coq8 OE increased steady state levels of Coq9 and Coq7 in the ⌬coq4 strain and Coq9 and Coq4 in the ⌬coq7 strain ( Fig. 2A). Even though Coq8 OE resulted in comparable levels of Coq8 in all ⌬coq strains tested, it failed to increase the steady state levels of Coq7 and Coq9 in ⌬coq1 and ⌬coq2 (Fig. 2B). The inefficiency of Coq8 OE in these two strains likely results from the absence of synthesis of prenylated Q 6 intermediates that have been hypothesized to be important for the stability of Coq4 and Coq6 (23). In the ⌬coq9 strain, Coq8 OE did not restore the steady state levels of either Coq4 or Coq7 ( Fig. 2A).
Because the yeast Coq polypeptides localize to mitochondria (2), we verified that the proteins stabilized by Coq8 OE were present in this organelle. The Coq4, Coq7, and Coq9 proteins were readily detected in mitochondria prepared from ⌬coq5 and ⌬coq6 cells with Coq8 OE (Fig. 3A). Although previous studies indicated that the Coq3 polypeptide was labile in ⌬coq strains (6), subsequent experiments performed in the presence of both protease and phosphatase inhibitors preserved steady state levels of Coq3 (12,15). In agreement, Coq3 levels were unaffected in ⌬coq5 or ⌬coq6 cells (Fig. 3A). The Coq6 antibody reacts with many nonspecific polypeptides in yeast cell extracts, and therefore immunoblotting for Coq6 is best carried out on isolated mitochondria. The steady state level of Coq6 was also increased by Coq8 OE in ⌬coq3, ⌬coq4, and ⌬coq7, although it did not reach wild-type levels (Fig. 3B). Steady state levels of the Coq1 polypeptide provide a loading control, because Coq1 is unaffected by null mutations in ⌬coq2-⌬coq9 (6).
Moderate Overexpression of Kinase-active Coq8 Is Sufficient for the Stabilization of Coq Polypeptides-The yeast W222 mutant (24) has a mutated COQ8 gene that encodes a Coq8-G130D polypeptide that mimics the human G272D pathogenic mutation found in ADCK3 (14). W222 does not synthesize Q 6 , has low levels of Coq4, Coq7, and Coq9, and has no detectable phosphorylated form of Coq3, suggesting that the G130D mutation completely abolishes the kinase activity of Coq8 (12). We tested the importance of the kinase activity of Coq8 for the stabilization of Coq proteins in ⌬coq strains by overexpressing Coq8-G130D. The levels of Coq7 and Coq9 were restored in ⌬coq6 cells with Coq8 OE but remained undetectable with Coq8-G130D OE (Fig. 4). Similarly, Coq9 steady state levels were restored in ⌬coq7 cells with Coq8 OE but not with Coq8-G130D OE. Coq8-G130D steady state level was lower than that of wild-type Coq8 overexpressed from the same high copy vec-tor (Fig. 4). Transformation of ⌬coq6 and ⌬coq7 cells with a centromeric plasmid containing the COQ8 gene (lcCoq8) resulted in moderate expression of Coq8, which was sufficient to restore steady state levels of the Coq7 and Coq9 polypeptides (Fig. 4). Because the Coq8-G130D polypeptide also accumulates to a moderate level upon overexpression, we conclude that the kinase activity of Coq8 is necessary to cause the stabilization of the Coq polypeptides.
Overexpression of Coq8 Promotes the Accumulation of Diagnostic Intermediates in ⌬coq5 and ⌬coq7 Strains-Based on our observation that Coq8 OE stabilizes most Coq proteins in the collection of the ⌬coq strains, we reasoned that these mutants may now accumulate Q 6 biosynthetic intermediates downstream of HHB. In fact, it has been shown that a ⌬coq7 strain known to accumulate HHB can be induced to proceed further and to accumulate DMQ 6 upon Coq8 OE (16). In this study, we have used HPLC separation coupled to either ECD or mass spectrometry (MS) to detect Q 6 intermediates in lipid extracts of the different strains tested. In our HPLC-ECD system, a precolumn electrode oxidizes the reduced hydroquinone intermediates present in the lipid extracts into their oxidized quinone form, accounting for the presence of only oxidized products in the HPLC column eluate. In the HPLC-MS/MS detection, the redox state of the intermediates is detected and affects the elution position. MRM allows the detection of the molecular ion of Q 6 intermediates in conjunction with the corresponding product ion base peak. A ⌬coq7 strain with Coq8 OE revealed that DMQ 6 was produced in a synthetic medium supplemented with either pABA or 4-HB, although no electroactive compound was detected in the absence of Coq8 OE ( Fig. 5A and data not shown). The synthesis of DMQ 6 from pABA is consistent with our previous observations that DMQ 6 , but not its imino-counterpart IDMQ 6 , was the main intermediate synthesized when strains expressing either a partially inactive form of Coq7 (3) or a coq7 point mutant (19) were grown in pABAcontaining medium.
An electro-active compound that eluted earlier than DMQ 6 was accumulated in a ⌬coq5 strain only upon Coq8 OE ( Ϫ0.2 ppm), the expected substrate of Coq5. The identification of DDMQ 6 is further supported by the trap scan spectrum that showed a tropylium-like ion at m/z 153 characteristic of Q 6 -related compounds (supplemental Fig. S1A). DDMQ 6 was synthesized by the ⌬coq5 strain with Coq8 OE when either 4-HB or pABA was used as ring precursors (Fig. 5B). This result was further confirmed by detecting 13 C 6 -DDMQ 6 in ⌬coq5 with Coq8 OE grown either in the presence of 13 C 6 -pABA or 13 C 6 -4-HB ( Fig. 5C and supplemental Fig. S1B). As expected, no Q 6 biosynthetic intermediates besides HHB or HAB were detected in either ⌬coq5 or ⌬coq7 strains with Coq8-G130D OE (data not shown). Our results show that Coq8 OE in ⌬coq5 and ⌬coq7 strains leads to the biosynthesis of the expected Q 6 intermediates, DDMQ 6 and DMQ 6 , respectively.
Bypass of the Q Biosynthetic Deficiency in ⌬coq7 Cells by 4-HB Analogue-Based on our recent demonstration of the bypass of the deficient C5-hydroxylation reaction in a ⌬coq6  Whole cell lysates were prepared as described in Fig. 2 from W303-1B wild-type yeast (WT) or from ⌬coq6 or ⌬coq7 strains (W303 genetic background) harboring either no plasmid; pFL44, a multicopy plasmid expressing Coq8 (hcCoq8); G130D, a multicopy plasmid with Coq8 containing the G130D mutation (G130D); or p3HN4, a low copy plasmid expressing Coq8 (lcCoq8). Yeast cells were cultured in SDϪUra medium to mid-log (2-3 A 600 nm ) phase and harvested, and aliquots of 20 A 600 nm were lysed and subjected to immunoblotting analyses with antisera to the designated Coq polypeptides. Pda1 serves as a loading control.

Novel Q Intermediates in Coq Null Yeast Overexpressing Coq8
JULY 6, 2012 • VOLUME 287 • NUMBER 28 strain by 3,4-dihydroxybenzoic acid and vanillic acid (VA) (Fig.  1) (18), we grew ⌬coq7 cells with Coq8 OE in the presence of 2,4-dihydroxybenzoic acid (2,. In the presence of 2,4-diHB, cells contained a significant amount of Q 6 , whereas growth in the presence of 4-HB only produced DMQ 6 (Fig. 6A). The product eluting at 780 s in 2,4-diHB-treated cells is not DMQ 6 because its UV-visible spectrum has a maximum at 265 nm different from that of DMQ 6 at 271 nm (data not shown). Q 6 -deficient strains have a growth defect on respiratory carbon sources because the mitochondrial electron transport chain is interrupted. The quantity of Q 6 synthesized in the presence of 2,4-diHB was sufficient to allow respiratory growth on lactateglycerol (Fig. 6B). DMQ 6 was incompetent at re-establishing the flow of electrons in the respiratory chain as shown by the absence of growth on respiratory medium containing 4-HB (Fig. 6B). Likewise, Coq8-G130D OE failed to rescue the respiratory growth defect of the ⌬coq7 strain in the presence of 2,4-diHB. These results show that it is possible to bypass the deficient C6-hydroxylation reaction in a ⌬coq7 strain with 2,4-diHB provided that Coq polypeptides are stabilized by Coq8 OE.
In a ⌬coq4 strain grown in the presence of 13 C 6 -pABA, Coq8 OE caused the accumulation of a product whose mass is consistent with the 13 C 6 -labeled form of 3-hexaprenyl-4-amino-5hydroxybenzoic acid (HHAB; Fig. 7B). HHAB was also detected in the coq4-1 point mutant (Fig. 7B), and trap scan spectra confirmed both the unlabeled and 13 C 6 -labeled forms of HHAB (supplemental Fig. S2, A and B). The HHAB intermediate is not observed in pABA-labeled wild-type yeast (data not shown). Upon culture with 13 C 6 -4HB, the corresponding anticipated intermediate, 13 C 6 -DHHB, was not detected in Coq4-deficient strains; only 13 C 6 -HHB was detected (data not shown). We also failed to detect DHHB in the ⌬coq4 strain with Coq8 OE even when the medium was supplemented with 3,4-dihydroxybenzoic acid (100 g/ml; data not shown). The accumulation of HHAB suggested a deficient O5-methyltransferase activity, even though the Coq3 polypeptide is stable in the ⌬coq4 strain. Addition of VA to the ⌬coq4 strain with Coq8 OE failed to generate either DMQ 6 (Fig. 7A) or the predicted prenylation product, 3-hexaprenyl-4-hydroxy-5-methoxybenzoic acid (data not shown). Addition of 3-methoxy-4-aminobenzoic acid to the growth medium of the coq4-1 point mutant led to the accumulation of 3-hexaprenyl-4-amino-5-methoxybenzoic acid (HMAB) (Fig. 7C and supplemental Fig. S3). HMAB is expected to accumulate in cells deficient for the C1-decarboxylation reaction. However, it is important to note that HMAB is also observed in wild-type yeast incubated in the presence of 3-methoxy-4-aminobenzoic acid (Fig. 7C). Therefore, a possible defect in the C1-decarboxylation reaction in the coq4-1 point mutant cannot be probed by using 3-methoxy-4-aminobenzoic acid. Collectively, our data show that only early amino intermediates of the Q pathway accumulate in coq4 yeast mutants and that the use of analogues of 4-HB or pABA does not cause the accumulation of downstream intermediates contrary to what we obtained in ⌬coq3 (Fig. 7A) or ⌬coq7 (Fig. 6A) mutants. Therefore, in addition to the O5-methylation that we proved to be deficient in a coq4 mutant, at least one other downstream biosynthetic step of the Q 6 pathway is impaired. This situation is consistent with a more general functional/structural role for Coq4 in the Q 6 biosynthetic complex (5).
Yeast ⌬coq9 Cells Overexpressing Coq8 Accumulate Intermediates Diagnostic of a Deficiency in C5-and C6-Hydroxylation Reactions-Coq9 is essential for Q 6 biosynthesis but its molecular function is unknown (25). Upon Coq8 OE, we observed the accumulation of distinct electroactive compounds depending on whether pABA or 4-HB was added to the growth medium of a ⌬coq9 strain (Fig. 8A). In the presence of 4-HB, two compounds were detected with UV-visible spectra (not shown) and retention times characteristic of DMQ 6 and of the oxidized form of 3-hexaprenyl-4-hydroxyphenol (4-HP). The identity of these intermediates was further confirmed by labeling with 13 C 6 -4-HB and by comparing the lipid extracts of ⌬coq9 cells with those of ⌬coq6 cells that are known to contain 4-HP (18) (525.4 to 129 transition at 2.66 min, see Fig. 8B) and to those of ⌬coq7 cells that contain DMQ 6 (567.6 to 173 transition at 4.86 min, see Fig. 8C). In the presence of pABA (Fig. 8A), the compound eluting at 600 s corresponds to the oxidized form of 3-hexaprenyl-4-aminophenol (4-AP). The compound eluting at 860 s (Fig. 8A) S4A), which are shifted with M ϩ 6 (m/z) upon labeling with 13 C 6 -pABA (supplemental Fig. S4B). The 13 C 6 -labeled form of 4-AP (524.4 to 128 transition at 2.86 min, Fig. 8D; trap scan spectra supplemental Fig. S5) was detected in ⌬coq6 and ⌬coq9 cells, and 13 C 6 -IDMQ 6 (566.6 to 172 transition at 4.76 min, Fig.  8E) was present in ⌬coq7 and ⌬coq9 cells. The ion with a 525.4 to 129 transition at 2.86 min in ⌬coq6 cells with Coq8 OE (Fig.  8B) is attributed to the ϩ1 isotope of 13 C 6 -4-AP (compare Fig.  8, B and D). Likewise, the signal at 4.76 min (Fig. 8C) actually corresponds to the ϩ1 isotope of 13 C 6 -IDMQ 6 (566.6 to 172 transition; compare Fig. 8, C and E). Our experiments show that ⌬coq9 cells with Coq8 OE accumulate 4-AP/4-HP that are diagnostic of a deficiency in the C5-hydroxylation catalyzed by Coq6 (18) and DMQ 6 /IDMQ 6 , which are formed consequently to a defect in the C6-hydroxylation reaction catalyzed by Coq7. Moreover, the C4-deamination (black dotted arrows in Fig. 1) of Q 6 biosynthetic intermediates originating from pABA is not efficiently catalyzed in the absence of Coq9. In conclusion, deletion of coq9 impacts on the activity of multiple Coq proteins Coq6, Coq7, and the putative deaminase.
We next attempted to bypass the biosynthetic defects of ⌬coq9 cells. Supplementation of the growth medium with VA caused the accumulation of DMQ 6 (supplemental Fig. S6), showing that even though the Coq6 deficiency could be bypassed, a strong block subsists at the level of Coq7. With 2,4-diOH, no Q 6 could be detected; instead a new electroactive compound at 730 s appeared (supplemental Fig. S6). The absence of Q 6 biosynthesis may be caused by the poor stability of several Coq polypeptides even with Coq8 OE (Fig. 2A). Indeed, the quantity of intermediates accumulated in the ⌬coq9 strain is lower than that obtained in ⌬coq6 and ⌬coq7 strains (Fig. 8, B-D), which have higher levels of Coq polypeptides compared with the ⌬coq9 strain ( Fig. 2A). Finally, 2,3,4-trihydroxybenzoic acid, which could in theory bypass both C5-and C6-hydroxylations, was found to inhibit Q 6 biosynthesis in WT cells and was therefore not tested further (data not shown). transformed with an episomal vector coding for Coq8 (pFL44) were grown in YNBϪpABAϪfolate 2% dextrose containing 1 mM 2,4-dihydroxybenzoic acid (2, or not (Ϫ). Lipid extracts of 2 mg of cells were analyzed by HPLC-ECD. B, WT W303 cells or ⌬coq7 cells transformed either with an empty vector (vec), an episomal vector (pFL44) encoding Coq8 or Coq8-G130D, or with an episomal vector encoding Coq7 were grown in YNBϪpABAϪfolate 2% dextrose for 24 h, and serial dilutions were spotted onto agar plates. The plates contained either YP 2% dextrose (Glu) or synthetic mediumϪpABAϪfolate supplemented with 2% lactate-2% glycerol (LG) containing either 4-HB or 2,4-dihydroxybenzoic acid (2,4-diHB) at 1 mM. The plates were incubated for 2 days (Glu) or 4 days (LG) at 30°C.

DISCUSSION
Most of the Coq polypeptides involved in Q 6 biosynthesis in S. cerevisiae are part of a multiprotein complex. Among Coq proteins, only Coq2 is predicted to possess transmembrane domains. Organization of the Coq polypeptides in a Q 6 biosynthetic complex associated with the mitochondrial inner membrane helps rationalize how these proteins can gain access to their substrates, which are hexaprenylated lipophilic compounds likely embedded in the membrane (26). In the absence of any one of the Coq polypeptides (⌬coq1-⌬coq9 strains), assembly of the Q 6 biosynthetic complex is impaired, which results in the degradation of Coq4, Coq6, Coq7, and Coq9. This situation is not unique for multiprotein complexes in S. cerevisiae and is also encountered for example with cytochrome c oxidase. Indeed, most Cox proteins are degraded in the absence of any of the three mitochondrially encoded core subunits Cox1-Cox3 (27). Our present study establishes that Coq8 OE stabilizes the levels of Coq4, Coq6, Coq7, and Coq9 in ⌬coq3-⌬coq7 strains and therefore generalizes to most Coq polypeptides this stabilizing effect that had been described previously for Coq4 (17). The stabilized proteins are active because they allowed accumulation of novel Q 6 biosynthetic intermediates in several ⌬coq strains, suggesting that the Q 6 biosynthetic complex is at least partially assembled.
Coq8 has been proposed to possess a kinase activity (11) and has been shown to be necessary for the existence of phosphorylated forms of Coq3, Coq5, and Coq7 in vivo (12,15). A G130D mutation in Coq8 abolishes the phosphorylated form of Coq3 suggesting that this mutation impairs the kinase activity (12). Our data with the G130D mutant indicate that the kinase activity of Coq8 is also important for the stabilization of Coq6 and Coq7 (Fig. 4) and probably also of Coq4 and Coq9. How does Coq8 OE stabilize Coq polypeptides that are otherwise degraded? A direct phosphorylation by Coq8 of the polypeptides could occur, especially in the case of Coq7 for which Coq8-dependent phosphorylated forms have been detected (12). However, a direct phosphorylation seems unlikely in the case of Coq4 because no phosphorylated forms of this protein could be detected (12). In addition, we do not currently know whether Coq6 and Coq9 are phosphorylated in vivo (12). So, to explain the effect of Coq8, we favor an hypothesis in which Coq8 OE increases the phosphorylation state of a particular Coq protein that modulates the assembly and/or stability of the Q 6 biosynthetic complex. An obvious candidate is Coq3, whose association with the Q 6 biosynthetic complex was shown to be dependent on Coq8 (15). However, stable Coq proteins are detected in a ⌬coq3 strain with Coq8 OE (Fig. 2A), establishing that phosphorylated Coq3 is not the only factor promoting the stabilization of the Coq polypeptides. Coq8-dependent phosphorylation(s) may be considered a positive regulator of Q 6 biosynthesis because it seems to favor assembly of the Q 6 biosynthetic complex that should lead to increased Q 6 biosynthesis. On the contrary, a regulatory phosphorylation with a negative impact on Q 6 biosynthesis has recently been described for Coq7 (28). In that study, an increase in the phosphorylated forms of Coq7 correlated with a decreased level of Q 6 and an increased level of DMQ 6 , the substrate of Coq7. It was therefore concluded that phosphorylated Coq7 has diminished activity (28). It remains to be established which kinase is responsible for the phosphorylation of the regulatory sites of Coq7.
We exploited the stabilizing effect of Coq8 OE to understand more precisely the Q 6 biosynthetic pathway, especially the role played by Coq4 and Coq9. We observed the accumulation of several intermediates in strains lacking Coq9 or Coq4 contrary to strains lacking Coq5 or Coq7, which accumulate a single late-stage intermediate, the substrate of the missing enzyme. Consequently, the impairment of several biosynthetic steps in ⌬coq9 and ⌬coq4 cells points to a role of these proteins in the general function or organization of the Q 6 biosynthetic complex rather than to a role in the catalysis of a particular biosynthetic step. In ⌬coq9 cells with Coq8 OE, the accumulation of 4-AP/4-HP establishes that the C5-hydroxylation step catalyzed by Coq6 is limiting, but this C5-hydroxylation occurs to some extent as demonstrated by the accumulation of IDMQ 6 / DMQ 6 . These later intermediates reveal that the C6-hydroxylation catalyzed by Coq7 is also deficient. In fact, the C6-hydroxylation is completely impaired because addition of VA, which bypasses the C5-hydroxylation, resulted in increased accumulation of DMQ 6 without any production of detectable Q 6 or demethyl-Q 6 (supplemental Fig. S6). The role of Coq9 in the C5-and C6-hydroxylation reactions is not clear, but Coq9 appears to be important for the stability of Coq7 because steady state levels of Coq7 are not restored in ⌬coq9 cells with Coq8 OE. Nevertheless, Coq9 is not absolutely required for the stability of the Q 6 complex to which it has been shown to belong  (6), because the accumulation of DMQ 6 implies that the Coq enzymes implicated in the Q 6 pathway upstream of Coq7 are stable and active, at least partially.

Novel Q Intermediates in Coq Null Yeast Overexpressing Coq8
The ⌬coq4 strain was diagnosed to be impaired in multiple biosynthetic steps; Coq3, which catalyzes both the O5-and O6-methylation steps, does not function in the ⌬coq4 strain as established by the accumulation of HHAB. Furthermore, addition of VA did not generate DMQ 6 in the ⌬coq4 strain, contrary to what we observed in the ⌬coq3 strain (Fig. 7A), revealing that at least another biosynthetic step downstream of the O5-methylation is impaired. In agreement with these results, Coq4 was hypothesized to serve as an anchor for the Q 6 biosynthetic complex (5) and was recently proposed to bind the polyisoprenyl tail of Q 6 intermediates, therefore allowing sequential modification of the aromatic head group (29). It is of interest to note that a diploid yeast carrying a deletion of one allele of COQ4 showed a diminished Q 6 content demonstrating that wild-type level of Coq4 is crucial for the function of the Q 6 biosynthetic complex (30). Human Coq4 may play an analogous role in organizing Q 10 biosynthesis because COQ4 haploinsufficiency was recently shown to cause Q 10 deficiency in a patient (30).
It is interesting that Q 6 intermediates containing a catechol moiety (DHHB and demethyl-Q 6 ) could not be detected in this study despite the fact that we expected their synthesis in some of the strains tested. Indeed, we anticipated the formation of DHHB in a ⌬coq4 strain grown in the presence of 4-HB because this strain produced HHAB from pABA. Also, the ⌬coq3 strain synthesized DMQ 6 but not demethyl-Q 6 from VA. In consequence, the catechol-containing Q 6 intermediates may be unstable and hence cannot accumulate in detectable amounts in S. cerevisiae.
pABA is a precursor of Q 6 in S. cerevisiae, and the C4-amine must be replaced by a C4-hydroxyl along the Q 6 pathway (3,19). Our data reveal several new elements regarding this C4-deamination (or deimination) reaction. First, ⌬coq6 and ⌬coq9 cells with Coq8 OE synthesized C4-aminated intermediates (4-AP and IDMQ 6 ) in the presence of pABA but synthesized the C4-hydroxylated intermediates (4-HP and DMQ 6 ) in the presence of 4-HB. This result supports the notion that HHB and HAB follow the same biosynthetic steps up to DMQ 6 / IDMQ 6 and are modified by the same Coq enzymes, which thus accommodate both C4-hydroxylated and C4-aminated Q 6 biosynthetic intermediates (path 1 and path 2, Fig. 1). Second, the deamination reaction does not proceed via nonenzymatic hydrolysis because in this case, IDMQ 6 , which was previously established to be a precursor of DMQ 6 , should be converted into DMQ 6. Third, the C4-deamination/deimination reaction occurs efficiently in some mutants (⌬coq5 and ⌬coq7) but not in others (⌬coq6, ⌬coq9 and possibly ⌬coq4). The deamination reaction may implicate Coq6 or Coq9 or an uncharacterized protein that may be inactive or unstable in ⌬coq6 and ⌬coq9 cells. In this latter case, a differential proteomic analysis of ⌬coq5 and ⌬coq7 cells on the one hand and ⌬coq6 and ⌬coq9 cells on the other hand may reveal the identity of the C4-deaminase. Fourth, the deamination reaction can occur prior to the C2-methylation catalyzed by Coq5 as supported by the accumulation of DDMQ 6 and not 2-demethyl-4-amino-demethoxy-Q 6 by the ⌬coq5 strain cultured in the presence of pABA.
The step at which the deamination reaction occurs in a WT strain is still in question. Indeed, it could take place prior to the C2-methylation or prior to the C6-hydroxylation catalyzed by Coq7 as suggested by the results establishing that IDMQ 6 is a precursor of DMQ 6 (19) . In that study, the coq7-1 point mutant and also wild-type yeast grown in the presence of pABA accumulated lower quantities of IDMQ 6 compared with DMQ 6 (19), in agreement with results presented in this study. The real question now concerns the reasons of the synthesis of the minute amounts of IDMQ 6 detected in WT cells. Is it a "normal" intermediate that only accumulates in small quantities because it is efficiently converted into DMQ 6 or does IDMQ 6 represent a by-product formed when the deamination reaction that may occur prior to Coq5 is rate-limiting? Further experiments will address this question.
The accumulation of Q 6 intermediates in ⌬coq4 and ⌬coq9 cells with Coq8 OE shed some light on the biosynthetic defects in these strains and is a significant advance in our understanding of Q 6 biosynthesis. Furthermore, the use of the 2,4-diHB aromatic ring precursor together with Coq8 OE was shown to restore Q 6 biosynthesis in yeast ⌬coq7 cells. Finally, the accumulation of C4-aminated intermediates in some ⌬coq strains provided new information regarding the deamination reaction. Overall, our study describes that Coq8 OE and the use of 4-HB analogues represent valuable tools to advance our comprehension of the Q 6 biosynthetic pathway by allowing for the unprecedented molecular dissection of the defect of particular ⌬coq strains.