Research article

Respiratory-induced coenzyme Q biosynthesis is regulated by a phosphorylation cycle of Cat5p/Coq7p

Alejandro Martín-Montalvo, Isabel González-Mariscal, Sergio Padilla, Manuel Ballesteros, David L. Brautigan, Plácido Navas, Carlos Santos-Ocaña


CoQ6 (coenzyme Q6) biosynthesis in yeast is a well-regulated process that requires the final conversion of the late intermediate DMQ6 (demethoxy-CoQ6) into CoQ6 in order to support respiratory metabolism in yeast. The gene CAT5/COQ7 encodes the Cat5/Coq7 protein that catalyses the hydroxylation step of DMQ6 conversion into CoQ6. In the present study, we demonstrated that yeast Coq7 recombinant protein purified in bacteria can be phosphorylated in vitro using commercial PKA (protein kinase A) or PKC (protein kinase C) at the predicted amino acids Ser20, Ser28 and Thr32. The total absence of phosphorylation in a Coq7p version containing alanine instead of these phospho-amino acids, the high extent of phosphorylation produced and the saturated conditions maintained in the phosphorylation assay indicate that probably no other putative amino acids are phosphorylated in Coq7p. Results from in vitro assays have been corroborated using phosphorylation assays performed in purified mitochondria without external or commercial kinases. Coq7p remains phosphorylated in fermentative conditions and becomes dephosphorylated when respiratory metabolism is induced. The substitution of phosphorylated residues to alanine dramatically increases CoQ6 levels (256%). Conversely, substitution with negatively charged residues decreases CoQ6 content (57%). These modifications produced in Coq7p also alter the ratio between DMQ6 and CoQ6 itself, indicating that the Coq7p phosphorylation state is a regulatory mechanism for CoQ6 synthesis.

  • coenzyme Q
  • mitochondrion
  • phosphorylation
  • respiration
  • yeast


Coenzyme Q (CoQ or Q) is an essential isoprenylated benzoquinone component of mitochondria. CoQ resides in the phospholipid bilayer of mitochondrial membranes [1] where it acts as an electron carrier from complex I or II to complex III at the inner membrane [2]. CoQ also works as an antioxidant agent quenching free radicals [3], and is required for other cellular functions that have been extensively reviewed [4].

Down-regulation of CoQ biosynthesis in humans causes a specific type of mitochondrial disease known as CoQ deficiency syndrome [5]. These patients show neuromuscular defects caused by very low CoQ levels. CoQ supplementation usually improves their clinical condition [6]. However, CoQ administration sometimes fails to alleviate the symptoms of the patients [7], probably due to the low and irregular distribution of exogenous CoQ in organs and tissues [8]. An alternative strategy to relieve CoQ deficiency is through the activation of its de novo synthesis. In order to up-regulate CoQ biosynthesis, a better understanding of CoQ biosynthesis and its regulation in eukaryotic cells is required.

In Saccharomyces cerevisiae, at least ten genes (COQ110) are involved in CoQ biosynthesis. Although most of the Coq proteins catalyse known reactions [9], other Coq proteins, such as Coq4p and Coq9p, do not seem to be directly involved and must participate in the regulation of CoQ6 biosynthesis [9]. Several lines of evidence suggest the existence of a CoQ biosynthetic complex. In general, null mutants of COQ genes accumulate the same early precursor known as HHB (hexaprenyl hydroxybenzoate) [10]. These data suggest that CoQ biosynthesis is produced by a protein complex that catalyses the modification of the benzenic ring [11]. Some COQ genes, such as CAT5/COQ7, display a different response when point mutations are introduced. CAT5 is the initial name given for this gene, but the name COQ7 is also accepted because it has been demonstrated that the requirement for gluconeogenesis corresponds to the inability to respire due to the lack of CoQ biosynthesis [12]. Previous studies have demonstrated that Coq7p is a mono-oxygenase that catalyses the hydroxylation of DMQ6 (demethoxy-CoQ6) to demethyl-CoQ6 (Figure 1) [12], but some point mutants of COQ7, such as e2519 [13], accumulate the precursor DMQ6, whereas more severe truncated alleles, such as qm51, only accumulate the early precursor HHB. Since DMQ6 does not support either the electron transport respiratory chain or prevent oxidative stress [13], it is considered that DMQ6 accumulation in wild-type strains acts as a reservoir. These results support the idea that Coq7p is a key player in CoQ6 biosynthesis, since DMQ6 is accumulated during fermentative growth in glucose, but is quickly modified to CoQ6 in two steps following the respiratory induction or when cells are subjected to oxidative stress [14]. Furthermore, this hypothesis is supported by the action of Coq8p. This protein belongs to a family of putative protein kinases known as ADCKs (aarF-domain-containing kinases) [15], although its alleged kinase activity remains to be confirmed. Interestingly, overexpression of Coq8p in yeast lacking Coq7p allows the production of the intermediate DMQ6, indicating a regulatory role for Coq8p in CoQ6 biosynthesis complex assembly [14]. COQ5 encodes a C-methyltransferase that converts 2-hexaprenyl-6-methoxy-1,4-benzoquinone into DMQ6 [16,17]. Some COQ5 point mutations (Coq5-2 and Coq5-5) fail to support C-methyltransferase activity, but show a steady-state expression similar to wild-type Coq5. These data demonstrate that Coq5p carries out two different functions: first, in modifying the benzenic ring, and secondly, as a structural component of the CoQ synthesis complex [18]. Previously, physical evidence has been reported of the existence of a CoQ complex by size-exclusion chromatography and Blue-native electrophoresis [1922] involving most Coq proteins. Coq5p and Coq7p are also involved in the up-regulation of CoQ6 biosynthesis induced by oleic acid [23], glycerol [14,23], oxidative stress [14] and the retrograde response [24].

Figure 1 Reaction catalysed by Coq7p

Coq7p is hydroxylase-linked extrinsically to the mitochondrial inner membrane (MIM) converting DMQ6 into demethyl-CoQ6.

The role of phosphorylation in mitochondrial signal transduction was initially poorly understood, but knowledge of it has increased in the last few years [25,26]. Phosphorylation by PKA (protein kinase A) has been related to mitochondrial biogenesis [27] and modulation of growth during the postdiauxic shift in yeast [28], although many other kinases may also be involved in mitochondrial metabolism [25]. The Coq7p amino acid sequence shows several putative phospho-sites suggestive of post-translational modification to regulate its activity. The regulation of mitochondrial complex activity by phosphorylation is a known process, as described for the PDH (pyruvate dehydrogenase) complex [29]. Phosphorylation is likely to be involved in the control of many other metabolic processes within the mitochondria, as depicted by the number of phospho-proteins detected in the yeast mitoproteome [30]. In the case of CoQ biosynthesis, previous studies have reported some indirect evidence for the phosphorylation of several catalytical Coq proteins by Coq8p [19,31], although the kinase activity of Coq8p has not to date been demonstrated.

In the present study we show that Coq7p is phosphorylated both in vitro and in vivo. The Coq7p phosphorylated state is induced during fermentative metabolism and becomes inhibited under respiratory conditions. Substitution of phospho-amino acid residues in Coq7p to non-phosphorylatable ones induces an increase in CoQ6 levels, but the presence of negative charges at these residues, mimicking a Coq7p phosphorylated state, produces a decrease in CoQ6 levels. These results provide direct evidence that Coq7p phosphorylation is a regulatory mechanism of CoQ6 biosynthesis in yeast.


Yeast strains and growth medium

Yeast strains used in the present study are listed in Supplementary Table S1 (at Growth media for yeast and bacteria were prepared as described previously [32]. Yeasts were grown at 30°C with shaking (200 rev./min).

Plasmid construction

Plasmid pNMQ7 was created by digesting the low copy number plasmid pRS316 with HindIII and XhoI and subsequent ligation of the fragment carrying COQ7 plus 1 kb both upstream and downstream in order to maintain the natural promoter and terminator sequences. The pL series (loss-of-function series) was created by digesting the plasmid with Kpn2I and HindIII and ligating the digested COQ7 versions obtained by mega-primer PCR mutagenesis [33]. The pG series (gain-of-function series) was obtained using the QuikChange® II XL site-directed mutagenesis kit (Stratagene) using the pNMQ7 plasmid as a template. The primers used are shown in Supplementary Table S2 (at GST (glutathione transferase) recombinant proteins were obtained by amplifying the sequences by PCR following ligation in the pGEX bacterial expression vector. PCR was performed using Fw-GEX and Rw-GEX primers. DNA sequencing was carried out by the MWG-Biotech AG Sequencing Service (Ebersberg, Germany).

Quinone identification and quantification

Total CoQ6 quantification in mitochondrial membrane samples was performed by HPLC-ECD (electrochemical detection). A description of the equipment and methods used has been published previously [34].

Protein purification and in vitro phosphorylation assay

Escherichia coli strain BL21 pLys was used for expression of Coq7p in the pGEX4T1 plasmid (GE Healthcare) using GST–Sepharose beads to purify Coq7p recombinant protein according to the manufacturer's instructions. A protein phosphorylation assay was performed [35,36] with recombinant active PKA (1 μg) (Sigma) or PKC (protein kinase C; 1 μg) (Sigma) from rat brain incubated with 0.1μCi of [γ-32P]ATP (PerkinElmer), 200 μM ATP (Sigma) and 2 μg of recombinant Coq7p protein in 50 μl of kinase buffer [50 mM Tris/HCl (pH 7.8), 25 mM 2-glycerophosphate, 5 mM MgCl2, 5 mM MnCl2 and 1 mM DTT (dithiothreitol)] and Sigma phosphatase inhibitor cocktail 1 [1:100 (v/v)]. Reaction mixtures were incubated at 30°C for 2 h. In some experiments, the GST moiety was released after thrombin digestion with 0.5 units of thrombin (Sigma) at 4°C overnight. Proteins were separated by SDS/PAGE (12% gel), and results were visualized using Pro-Q Diamond gel stain (Molecular Probes), autoradiography, silver staining and anti-GST tag immunodetection.

Phosphorylation assay in purified mitochondria

Fresh mitochondria from cells expressing V5-tagged Coq7p (100 μg of protein) were permeabilized in 50 μl of kinase buffer [50 mM Tris/HCl (pH 7.8), 25 mM 2-glycerophosphate, 5 mM MgCl2, 1 mM MnCl2, 1 mM DTT and 0.3 g of 106 μm glass beads] plus 1:100 (w/v) digitonin and Sigma phosphatase inhibitor cocktail 1 [1:100 (v/v)]. Samples were subjected to vortex mixing (15 min at 4°C). After centrifugation (12000 g for 5 min at 4°C) the supernatant was incubated with 0.1 μCi of [γ-32P]ATP (PerkinElmer) and 200 μM ATP at 30°C for 2 h. Proteins were separated by SDS/PAGE (12% gel), and results were visualized using Pro-Q Diamond gel stain (Molecular Probes), autoradiography, SYPRO Ruby Red (Sigma) and anti-V5 tag immunodetection.

Immunoprecipitation and 2-DE (two-dimensional electrophoresis)

Mitochondrial samples from cells expressing V5-tagged Coq7p were permeabilized and phosphorylated as indicated above with non-radioactive ATP. Immunoprecipitation was performed with anti-V5 agarose gel-affinity beads (Sigma) according to the manufacturer's procedure. Proteins were subjected to 2-DE as described below. Agarose beads were resuspended in rehydration buffer (DeStreak, GE Healthcare). Supernatants were applied to Immobiline DryStrips (7 cm, pH 3-11 NL, GE Healthcare) and incubated overnight at room temperature (23°C). IEF (isoelectric focussing) was performed in the Bio-Rad Protean IEF Cell in four steps (1 h constant at 200 V, 1 h gradient up to 500 V, 6 h gradient up to 10000 V and a further 0.4 h constant at 10000 V) at 20°C with 50 μA per strip. Strips were equilibrated for 15 min in SDS buffer [375 mM Tris/HCl (pH 8.8), 6 M urea, 20% glycerol and 2% SDS] containing 2% (w/v) DTT and then for 15 min in SDS buffer with 2.5% (w/v) iodoacetamide. For the second dimension, the strips were applied to 12% SDS gels and sealed with 0.5% agarose in running buffer and 0.002% Bromophenol Blue.

Other methods

Mitochondrial DNA integrity was checked in all strains using JM6 and JM8 strains, two ρ0 strains. Steady-state levels of specific peptides were measured by densitometric analysis, carried out with a GS800 densitometer (Bio-Rad) with ImageJ (version 1.41o) for analysis of software. All results are expressed as the means±S.D. Statistical analyses were carried out using the SigmaStat 3.0 (SPSS) statistical package (t test). Phosphorylation analysis in immobilized GST-tagged proteins was performed with a Beckman–Coulter scintillation counter using Ready Protein Plus scintillator cocktail.


Recombinant Coq7p is phosphorylated at three predicted phospho-amino acids

In silico analysis of the Coq7p amino acid sequence performed using different algorithms (PPSearch and NetPhos 2.0) [37,38] predicts that three amino acid residues can be phosphorylated by protein kinases [39]. These putative phospho-sites are located at Ser20, Ser28 and Thr32 (Figure 2A). In the COQ7 gene sequence, codons encoding the predicted phospho-amino acids were changed to alanine by recombinant PCR to obtain a non-phosphorylatable version of Coq7p (AAA) (Figure 2A). Wild-type and non-phosphorylatable versions of Coq7p GST-tagged proteins were purified by size-exclusion chromatography (Supplementary Figure S1 at and were phosphorylated with commercial protein kinases (PKA and PKC) plus [γ-32P]ATP while immobilized on glutathione–Sepharose beads. Phosphorylated proteins were released from beads by thrombin digestion and analysed by SDS/PAGE (Figure 2B). 32P-labelling was detected only in samples corresponding to wild-type Coq7p (Coq7p-SST–GST), but not in the Coq7p-AAA–GST version (S20A, S28A and T32A). Recombinant protein was subjected to 2-DE and SDS/PAGE (see the Experimental section). Spots corresponding to these proteins were digested (ProGest, Genomic Solutions) and peptides were analysed by MS-MS/MS (tandem MS) [MALDI (matrix-assisted laser-desorption ionization)–TOF (time of flight)/TOF] in a MALDI–TOF/TOF mass spectrometer (4700 Proteomics Analyzer, Applied Biosystems) at the proteomic service of the University of Córdoba (Supplementary Figure S2 at These spots can be produced by different extents of phosphorylation. Data analysis performed using SEQUEST search engine software on several protein databases confirmed both proteins as Coq7p from Saccharomyces cerevisiae. The false discovery rate was 1%. The coverage for Coq7p-SST for the spot analysed was between 60 and 63% with a score between 128.52 and 134.76.

Figure 2 In vitro phosphorylation of different versions of Coq7p

(A) The three predicted phospho-amino acids in the Coq7p sequence have been modified by site-directed mutagenesis. SST corresponds to the wild-type Coq7p, AAA is a Coq7p version displaying three alanine residues instead of phospho-amino acids Ser20, Ser28 and Thr32, and DED corresponds to a Coq7p version where the phospho-amino acids were changed into two aspartic acid and one glutamic acid. Coq7p-AAA is permanently dephosphorylated and Coq7p-DED is permanently negatively charged. (B) GST-free proteins were obtained after digestion of Coq7p-SST–GST and Coq7p-AAA–GST with thrombin, and these were used as substrates for in vitro phosphorylation with [γ-32P]ATP and PKA. Phosphorylated proteins were analysed by SDS/PAGE followed by silver staining, anti-GST immunostaining and 32P detection using a radioactivity scanner (Typhoon 9410). GST, free GST added to include a positive phosphorylation control as well as the presence of GST linked to Coq7 proteins; AAA, Coq7p-AAA; SST, Coq7p-SST. The analysis was performed with purified proteins from fractions 6, 7 and 8 from a previous size-exclusion chromatography (Supplementary Figure S1 at (C) Previously phosphorylated Coq7p-SST–GST (4 μg), Coq7p-AAA–GST (4 μg) and free GST (2 μg) were bound to glutathione–Sepharose beads. Each sample was divided into two aliquots and one was treated with λ-phosphatase (5 units) after removal from the original buffer. The bound radiolabelled phosphate was monitored in treated and non-treated samples in a scintillation counter. Data correspond to a representative experiment of a set of two and are represented as the means±S.D. **P<0.01, significant differences compared with the Coq7p-SST–GST protein.

To calculate the stoichiometry of Coq7p phosphorylation, 4 μg of Coq7p-SST–GST, 4 μg of Coq7p-AAA–GST and 2 μg of free GST were phosphorylated as previously indicated with [γ-32P]ATP and 200 μM non-radiolabelled ATP. The amount of GST included was half of the amount of Coq7p–GST protein in order to maintain the same number of molecules in the assay, since the molecular mass of GST is 26 kDa, whereas Coq7p–GST is 52 kDa. All proteins were immobilized on GST–Sepharose beads and washed several times. The samples were divided into two aliquots and one was treated with λ-phosphatase. Finally, both aliquots were analysed in a scintillation counter (Figure 2C). Coq7p-SST–GST showed the highest amount of bound phosphate compared with Coq7p-AAA–GST and free GST. Treatment with 5 units of λ-phosphatase completely removed the bound phosphate in all proteins analysed. GST is a protein that can be phosphorylated by PKC at Ser93 [35]. Phosphorylation is similar in free GST and Coq7p-AAA–GST proteins, indicating that the phosphorylation detected in Coq7p-AAA–GST is produced in the tag and not in the Coq7p component. In contrast, Coq7p-SST–GST is significantly phosphorylated in the Coq7p component, supporting the previous data obtained by SDS/PAGE and 32P detection.

In the case of Coq7p-SST–GST the net phosphate (total phosphorylation minus free GST phosphorylation) detected was 1951 c.p.m. When 96% efficiency of the scintillation cocktail and a concentration of non-radiolabelled ATP 1000-fold higher than [γ-32P]ATP were taken into account, the number of phosphate molecules bound to Coq7p-SST–GST was approximately 5.56×1013. In the assay there was 2 μg of Coq7p-SST–GST included, which contains approximately 2.31×1013 molecules based on the molecular mass of the recombinant protein (52 kDa). The ratio of 2.41 molecules of phosphate per molecule of protein fits with the number of phospho-amino acids expected, but it is not possible to assure that in vitro all phospho-amino acids were phosphorylated at the same proportion. Interestingly, we did not obtain any net phosphorylation in the Coq7p-AAA–GST version. The phosphorylation of the wild-type Coq7p (Coq7p-SST–GST) was also supported by the saturated conditions of the assay; ATP and the commercial kinases are present at an excess in the assay. In this sense, the absence of phosphorylation in Coq7p-AAA–GST is produced by the lack of phospho-sites accessible for commercial kinases, although it is not possible to dismiss other phospho-sites for mitochondrial kinases. At least these two, and probably three, putative phospho-amino acids are phospho-amino acids in vitro.

In vitro Coq7p phosphorylation is performed by mitochondrial kinases

Coq7p phosphorylation was corroborated in purified mitochondria (Figure 3A). Wild-type (Coq7p-SST V5) and non-phosphorylatable Coq7p (Coq7p-AAA V5) epitope-tagged versions of Coq7p were expressed in a null COQ7 strain cultured in 2% galactose. Coq7 versions were detected by Western blot analysis (with an anti-V5 antibody). A slight shift in migration was determined in wild-type compared with Coq7p-AAA. Autoradiography clearly detected 32P-labelling in wild-type Coq7p-SST. However, only a faint signal was detected in the Coq7p-AAA version. Pro-Q Diamond staining, a fluorochrome with affinity for phospho-amino acids, showed an intense band for Coq7p-SST and only a weak band in the Coq7p-AAA version, indicating that wild-type Coq7p was phosphorylated under these conditions.

Figure 3 Phosphorylation of Coq7p in isolated purified mitochondria

(A) Yeast COQ7 mutant strains expressing either Coq7p-SST or Coq7p-AAA proteins fused to a V5 epitope were cultured in 2% galactose for 16 h with shaking at 30°C. SST, wild-type Coq7p; AAA, dephosphorylated Coq7p. Mitochondria were permeabilized, incubated with [γ-32P]ATP and analysed by SDS/PAGE, followed by SYPRO Ruby Red staining, anti-V5 antibody immunostaining, 32P detection using a radioactivity scanner (Typhoon 9410) and also Pro-Q Diamond staining to detect phosphorylated proteins. (B) Similar amounts of mitochondria were subjected to immunoprecipitation with anti-V5 antibody and 2-DE. 2-DE gels were visualized by silver staining and analysed by Pro-Q Diamond staining to detect phosphorylated proteins and anti-V5 antibody immunostaining. Data correspond to a representative experiment of a set of two. M, molecular mass markers.

As further evidence of the differential phosphorylation state of Coq7p-SST compared with Coq7p-AAA protein, a similar amount of mitochondria was subjected to immunoprecipitation and 2-DE (Figure 3B). Both proteins migrated at 26 kDa, consistent with the predicted size. Spots were analysed by MS (see the Experimental section). The false discovery rate was 1%. The coverage for Coq7p-SST for the spot analysed was between 78 and 82% with a score between 178.45 and 191.34. In Coq7p-AAA, several spots stained with silver and anti-V5 antibody were observed, but none of them were stained with Pro-Q Diamond, consistent with a lack of phosphorylation in this version. However, these data do not exclude that other phospho-amino acids, different from the ones that we analysed, were phosphorylated in these conditions, but the phosphorylation could not be detected by Pro-Q Diamond. On the other hand, the Coq7p-SST showed at least two additional forms separated by IEF at the same size that were stained with silver and anti-V5 antibody, as well as Pro-Q Diamond, consistent with multiple protein phosphorylation.

Coq7p phosphorylation is modulated by the metabolic state of the cell

Respiratory growth is established during the postdiauxic shift phase. When yeast change their metabolism from glucose fermentation to ethanol respiration, the metabolic shift requires CoQ6 biosynthesis as an essential component of the mitochondrial respiratory chain. Coq7p is considered a protein that regulates the final steps of the biosynthesis (Figure 1) from the accumulated DMQ6.

V5-tagged Coq7p-SST was expressed in a null COQ7 mutant strain. Cells were transferred to several culture media with different concentrations of fermentable (0.5, 2 and 10% glucose) and non-fermentable (ethanol and glycerol) carbon sources. Mitochondria were permeabilized and analysed by SDS/PAGE (Figure 4A). Pro-Q Diamond showed differential staining of Coq7–V5 protein, depending on the carbon source included in the culture. The ratio of phosphorylated Coq7p/total Coq7p–V5 (Figure 4B) normalized with the total protein indicated that higher glucose concentrations (2 and 10%) yielded a higher level of phosphorylation in Coq7p compared with cells grown with 0.5% glucose. More dramatic changes were observed in cells grown in non-fermentable carbon sources, such as ethanol or glycerol (33 and 37% respectively when compared with growth in 10% glucose). The levels of CoQ were determined in order to relate changes in Coq7p phosphorylation state to the regulation of CoQ synthesis in mitochondrial samples used previously (Figure 4C). Samples incubated for 2 h in non-fermentable carbon sources or in 0.5% glucose produced a small, but significant, increase in the CoQ6 level compared with yeast cultured in 2% glucose. This increase was also accompanied by a parallel decrease in DMQ6 content. The modification produced in 10% glucose is the opposite, but it is not significant when compared with 2% glucose. The change of carbon source modifies the phosphorylation extent of Coq7p as well as the CoQ6/DMQ6 ratio, although it does not demonstrate a direct relationship between these two observations.

Figure 4 Coq7p phosphorylation is dependent on the carbon source

Wild-type COQ7 fused to V5 (Coq7p-SST–V5) was grown in 2% galactose for 48 h. Yeast were harvested, washed and resuspended in different culture media containing several carbon sources (glucose at 0.5%, 2% or 10%, YPG or YPE) and incubated for 2 h. Control, washed cells without incubation. Purified mitochondria from these cultures were permeabilized and analysed by SDS/PAGE followed by (A) Pro-Q Diamond staining to detect phosphorylated proteins and anti-V5 antibody immunostaining. M, molecular mass markers. (B) Densitometry analysis of phosphorylated Coq7p-SST–V5/total Coq7p-SST–V5 ratio (ImageJ version 1.41o). Data correspond to a representative experiment repeated twice. (C) Mitochondrial samples used in previous experiments were subjected to lipid extraction and quinone quantification. Samples were analysed in triplicate and the experiment shown is representative of a set of two. *P<0.05 or **P<0.01, significant differences compared with control.

Removal of Coq7p phospho-sites increases CoQ6 levels

COQ7 site-directed mutagenesis was performed in a single-copy cloning vector (pRS316) that maintains COQ7 expression under the control of its natural promoter (pNMQ7) [39]. Modifications from phosphorylatable amino acids (SST) to alanine (A) are named pL since it is considered a loss-of-function (L). We have obtained several pL alleles by a combination of single, double and triple mutations. The expression of pL alleles supports growth in liquid and on glycerol plates (Supplementary Figures S3 and S4 at indicating the ability to maintain respiratory metabolism. Yeast harbouring pL mutant versions of Coq7p contained significantly higher amounts of CoQ6 compared with the wild-type control, from 157 to 256% (Figure 5, and Supplementary Figure S5 and Table S3 at The amount of CoQ detected in yeast expressing the triple pL mutant (pL-AAA) was similar to the amount produced by a strain expressing the wild-type COQ7 in a multi-copy plasmid (pmQ7). Although versions of mutated Coq7p lacking one or more phospho-sites produced a CoQ6 increase, data suggest that the removal of the second and third phospho-sites results in a higher CoQ6 accumulation (Supplementary Table S3).

Figure 5 CoQ6 determination in mitochondria from different versions of Coq7p

Cells from a COQ7 mutant strain harbouring the indicated plasmids were cultured in 2% glucose for 72 h and subjected to mitochondrial purification, lipid extraction and quinone quantification. Samples were analysed in triplicate and the experiment was repeated at least three times. pRS316 is an empty single-copy vector that was used to clone the wild-type version of COQ7 controlled by its natural promoter and terminator (pNMQ7) and both mutated versions of COQ7 obtained by site-directed mutagenesis from pNMQ7 as a template, Coq7p-AAA (pL-AAA) and Coq7p-DED (pG-DED). pRS426 is an empty multi-copy vector that was used to clone the wild-type version of COQ7 controlled by its natural promoter and terminator (pmQ7). Data correspond to the BY4741 background. ND, non-detectable. The amount of CoQ6 in other Coq7p versions is shown in Supplementary Table S3 (at *P<0.05, **P<0.01, significant differences compared with the pNMQ7 control.

Phosphomimetic changes in Coq7p phospho-sites decrease CoQ6 levels

We introduced negatively charged amino acids to replace serine and threonine phospho-sites in Coq7p. Both glutamic and aspartic acids have been reported as useful modifications to mimic the phosphorylated state of proteins [40]. For example, the change of serine to alanine abolishes the ability of CKII (casein kinase II) to activate p53, but the change of alanine to aspartic acid partially restores this activity [41]. Ser20 was changed to aspartic acid by the mutation TC58GA, Ser28 was changed to glutamic acid by the mutation TC82GA, and Thr32 was changed to aspartic acid by the mutation AC94GA (Figure 2A). These plasmids were named pG to denote a gain-of-function, in contrast with the pL loss-of-function plasmids. We prepared different pG alleles with a combination of single, double and triple mutations. The expression of pG alleles supported the growth of yeast in liquid and glycerol plates (Supplementary Figures S3 and S4). Mitochondrial CoQ6 levels were significantly lower compared with the positive control, from 57 to 72%, and also compared with pL Coq7p mutants (Figure 5, and Supplementary Figure S5 and Table S3). As mentioned above, the most severe changes were obtained in mutants including the second and third phospho-sites.

Evolution of the CoQ6 and DMQ6 ratio is modified in phosphomimetic and permanent dephosphorylated Coq7p versions in glucose-based medium

We have previously demonstrated that DMQ6 levels were increased in wild-type cells during growth in glucose [14]. Initially, the DMQ6 concentration was increased, but quickly decreased when the culture reached the postdiauxic growth phase. CoQ6 displays the opposite pattern. These changes support the hypothesis of Coq7p participation in the biosynthetic regulation of CoQ6, and it was also the reason why we performed the same study in yeast expressing both modified Coq7p versions (Figure 6). The expression of wild-type Coq7p (pNMQ7) produces a similar profile of Q6/DMQ6 ratio that has been described previously in whole-cell lysates from wild-type yeast [14]. In the first phase there is an initial increase of DMQ6 followed by a subsequent decrease accompanied by increased levels of Q6. The same experiment was performed with the non-phosphorylatable version of Coq7p (pL-AAA). The amount of DMQ6 was significantly lower compared with CoQ6, even when the total amount of quinones was dramatically higher compared with wild-type cells. In the phosphomimetic version of Coq7p (pG-DED), there are no severe changes between CoQ6 and DMQ6 during growth in glucose. These results support the hypothesis in which Coq7p regulates CoQ6 biosynthesis, since DMQ6 and CoQ6 content and the DMQ6/CoQ6 ratio depend on the phosphorylation state of Coq7p.

Figure 6 Evolution of the CoQ6 and DMQ6 ratio in different versions of Coq7p in glucose-based medium

Cells from a COQ7 mutant strain harbouring several versions of Coq7p were cultured in 2% glucose for 24 h and subjected to mitochondrial purification, lipid extraction and quinone quantification. Samples were analysed in triplicate and the experiment shown is representative of a set of two. *P<0.05 or **P<0.01, significant differences compared with the previous bar. (Top panel) pNMQ7, wild-type version; (middle panel) Coq7p-AAA (pL-AAA); and (bottom panel) Coq7p-DED (pG-DED).


After postdiauxic shift, the transcription factor HAP4 [42] controls the expression of several mitochondrial genes [43] and appears to be involved in the expression of COQ2 and COQ7/CAT5 genes [44]. The assembly of Coq polypeptides into a complex leading to CoQ6 biosynthesis is another level of regulation [19,22]. Some studies have demonstrated that either CoQ6 addition or Coq8p overexpression produces DMQ6 accumulation by promoting full complex assembly in a COQ7-null mutant [14]. These data suggest that Coq7p could be a control point for the regulation of CoQ6 biosynthesis in yeast [14]. Given that Coq7p contains three putative phospho-sites, we have studied whether Coq7p activity is regulated by phosphorylation. In the present study we show that Coq7p is indeed a phospho-protein: in vitro phosphorylation performed with commercial PKA or PKC is not found after the removal of three potential phospho-amino acids. Recombinant Coq7-SST is phosphorylated by different methods which demonstrate the presence of two, and probably three, phospho-amino acids in Coq7p. These data make difficult the idea that phosphorylation takes place in any other Coq7p amino acids, but the present study is inconclusive for this point since other possible kinases present in yeast mitochondria are not analysed; multiple phospho-sites are possible given the large number of isoelectric points found by MS in Coq7p recombinant protein in the in vitro assay.

We also determined that Coq7p is phosphorylated within the mitochondria, supporting the in vitro experiments. These data involve kinase(s) naturally expressed in mitochondria that phosphorylate wild-type Coq7p, but not the non-phosphorylatable Coq7p version, and it also supports that these putative phospho-sites are biologically active. Obviously, these phosphorylation conditions in the assay are not the conditions in vivo, but still demonstrate that natural kinase(s) perform this reaction.

Moreover, we have shown evidence that phosphorylation of Coq7p is physiologically relevant since it is affected by changes in carbon source that are known to alter the metabolism of yeast. Fermentable carbon sources, such as high glucose, increase Coq7p phosphorylation, whereas non-fermentable carbon sources, such as glycerol or ethanol, or even lower glucose levels (0.5%), reduce Coq7p phosphorylation. Given that non-fermentable carbon sources require the activation of mitochondrial respiration, these conditions are similar to the postdiauxic shift changes occurring when glucose is exhausted in glucose-based culture media [45]. In that sense, it has been reported that glycerol growth induces the expression of COQ genes and Coq proteins [23]. It is also known that low-glucose culture media promote CoQ6 accumulation [46], and finally, during stationary growth phase, a high amount of CoQ6 is produced when compared with the exponential growth phase [47]. Our results support the hypothesis that the Coq7p phosphorylated state correlates with fermentation, whereas the non-phosphorylated state is associated with respiration. We have shown that changes in Coq7p phosphorylation induce modifications in the amount of CoQ6 and DMQ6 produced, supporting the idea that the non-phosphorylated state of Coq7p leads to higher CoQ6 levels in non-fermentable carbon sources as demonstrated by the analysis of CoQ6 levels in the phosphomimetic Coq7p version (Coq7-pG-DED) and the non-phosphorylatable version (Coq7-pL-AAA).

Phosphorylation is a well-known process that regulates the activity of other mitochondrial enzymes; for instance, phosphorylation leads to lower activity, whereas dephosphorylation activates the PDH complex [29,48]. We developed site-directed mutations in Coq7p in order to produce either a permanent dephosphorylated state of Coq7p (alanine substitutions of serine and threonine) or the phosphorylated state of this protein (aspartic acid or glutamic acid substitution of serine and threonine) [41]. These forms of Coq7p provided us with a tool to determine the effects of Coq7p phosphorylation on CoQ6 biosynthesis. Our results demonstrate that dephosphorylated Coq7p causes a significant increase in CoQ6 that is similar to the increase produced by the COQ7 gene overexpression with a multi-copy vector, supporting the hypothesis that the dephosphorylated form of Coq7p is the more active form. Conversely, the phosphomimetic form of Coq7p produces a decrease in CoQ6 biosynthesis. The effects of both Coq7p versions include a shift of the CoQ6/DMQ6 ratio during growth in glucose, as described previously [14]. This suggests that the phosphorylated state of the phospho-sites in Coq7p reduces its activity. It is also important to note that even if we did not determine exactly the number and location of the in vivo phospho-sites in Coq7p, it does not compromise the fact that the phosphorylation state of this protein is biologically relevant in yeast.

Phosphorylation is a poorly known post-translational modification process in mitochondria since only 15 kinases and five phosphatases are present in yeast mitochondria [49,50]. It has been suggested that most of the kinase phosphorylation processes take place in the cytosol [cAMP-dependent, MAP (mitogen-activated protein), TOR (target of rapamycin), cyclin and Snf1p kinases] that finally affect mitochondrial metabolism through the phosphorylation of mitochondrial or mitochondria-associated proteins. This may explain the low number of kinases/phosphatases observed in mitochondria [51]. In order to find the kinase responsible for Coq7 phosphorylation, we could speculate a general regulation process based on a general kinase or kinases that send a repression signal into the mitochondria. On the other hand, the existence of specific mitochondrial kinases devoted to regulate CoQ biosynthesis is also possible. In the first case, a lot of regulatory pathways that detect the nutrient availability have been described, such as TOR1, SCH9, RAS1/2 and SNF1 [52,53], which are related to energy metabolism. In contrast there are examples within the mitochondria, such as PDH complex regulation. In this case, PDH activity is specifically regulated by two pairs of kinase/phosphatases (PKP1/2 and PTC5/6) [54]. Coq8p is a protein required for CoQ biosynthesis [55] with unknown functions, although it belongs to a family of putative kinases (ADCKs) [15]. Some studies relate Coq8p to the phosphorylation of several Coq proteins, such as Coq3p [19], which has been previously indicated to be a target of phosphorylation [46]. Coq5p and Coq7p have also been predicted to be phosphorylatable by indirect methods [31].

We propose a novel model of Coq7p-based regulation of CoQ biosynthesis where participation of at least two new enzymes are required. A mitochondrial kinase and a mitochondrial phosphatase modify Coq7p activity through phosphorylation. Identifying these enzymes and linking their activity to the metabolic state of the cell is the new challenge in better understanding the regulation of CoQ6 biosynthesis.


Alejandro Martín-Montalvo performed most of the experiments; Isabel González-Mariscal was responsible for phosphorylation assays using Sepharose beads; Sergio Padilla obtained some modified versions of the COQ7 gene; Manuel Ballesteros performed the two-dimensional electrophoresis analysis; David Brautigan produced recombinant proteins and carried out in vitro kinase assays; Plácido Navas obtained the funding for the study and helped to write the paper; and Carlos Santos-Ocaña wrote the paper, was responsible for the study and was the Ph.D. advisor of Alejandro Martín-Montalvo.


This work was supported by the Spanish Ministerio de Ciencia y Tecnología [grant number FIS PI080500]; the Junta de Andalucía [grant number P08-CTS-03988]; and by a grant from the International Q10 Association entitled “Phosphorylation-based regulation of coenzyme Q biosynthesis in yeast”. A.M.-M. received a predoctoral fellowship from the Consejería de Innovación Ciencia y Empresa, Junta de Andalucía (Spain). I.G.-M. received a predoctoral fellowship from the Plan Propio of the Universidad Pablo de Olavide de Sevilla.


We thank Todd Prickett (University of Virginia, Charlottesville, VA, U.S.A.) for help with the protein purification and in vitro phosphorylation assay. We also thank Bethany Carboneau and Shakeela Faulkner for critiquing the paper prior to submission.

Abbreviations: ADCK, aarF-domain-containing kinase; CoQ, coenzyme Q; 2-DE, two-dimensional electrophoresis; DMQ6, demethoxy-CoQ6; DTT, dithiothreitol; GST, glutathione transferase; HHB, hexaprenyl hydroxybenzoate; IEF, isoelectric focussing; PDH, pyruvate dehydrogenase; PKA, protein kinase A; PKC, protein kinase C


View Abstract