Mitochondrial NAD kinase Pos5 is needed for making CoQ in yeast

Coenzyme Q (CoQ), also known as ubiquinone, is an essential component of the respiratory chain required for energy production. CoQ cycles between reduced [CoQH₂] and oxidized [CoQ] forms [1]. This is a redox property important for electron transfer during respiration and for functioning as an antioxidant. Eukaryotes and bacteria belonging to the phylum Pseudomonadota synthesize CoQ endogenously, with species-specific variations in side chain length; for example, Homo sapiens and Schizosaccharomyces pombe produce CoQ₁₀, where the number of isoprene units is ten, Saccharomyces cerevisiae produces CoQ₆, and Escherichia coli produces CoQ₈ [2,3]. CoQ biosynthesis comprises mainly three stages: benzoquinone ring formation, isoprene side chain synthesis, and modification of the prenylated quinone ring [4]. The precursor of the side chain is synthesized from isopentenyl diphosphate and farnesyl diphosphate by polyprenyl diphosphate synthase [5]. Then, it is transferred to p-hydroxybenzoic acid (PHB) or p-aminobenzoic acid by p-hydroxybenzoate–polyprenyl diphosphate transferase (Coq2 or Ppt1) [6,7]. In eukaryotes, PHB is derived from tyrosine or other amino acids. The quinone ring of prenylated PHB then undergoes modifications, including methylations (Coq3 and Coq5), decarboxylation (Coq4), and hydroxylations (Coq6 and Coq7), to generate mature CoQ [2,8,9]. These reaction enzymes are encoded by nine genes in S. cerevisiae (COQ1-COQ9) and ten genes in S. pombe (dps1, dlp1, ppt1, and coq3-coq9) [10–14]. Those genes were utilized for CoQ₁₀ bioproduction in S. pombe [15]. In addition, benzoic acid inhibits the synthesis of CoQ [11], protein kinase A (Pka1) controls the level of CoQ [16], and regulatory factors such as Coq11 and Coq12 have recently been identified in S. pombe, suggesting that further more unknown factors are involved in regulating CoQ biosynthesis [17].

Among the deletion mutants that showed a lower CoQ₁₀ level in S. pombe, we selected the pos5 mutants for further analysis. Although Pos5 has been extensively studied in S. cerevisiae, very little is known about its function in S. pombe. In S. cerevisiae, Pos5 is a unique mitochondrial nicotinamide adenine dinucleotide NAD(H) kinase that generates NADPH or NADP⁺ from NADH or NAD⁺ [18–22]. Because of its polarity, mitochondrial NADP(H) is synthesized from NAD(H) via mitochondrial NAD(H) kinase, as no mitochondrial transporter has been identified in yeast. In mitochondria, NADP⁺ is essential for several processes, including the TCA cycle, amino acid biosynthesis, glutathione reduction, and Fe-S cluster biogenesis [19–22]. However, the relevance of NAD(H) kinase activity in CoQ biosynthesis has not been documented in any organism. Therefore, in this study, we focused on elucidating the role of Pos5 involves in CoQ synthesis.

Yeasts and E. coli strains used in this study are listed in Table 1. Yeast standard media and genetic manipulation methods have been described previously [23]. S. pombe strains were grown in complete YES medium (0.5% yeast extract (OXOID), 3% glucose, supplemented with 225 mg/mL adenine sulfate, 225 mg/mL leucine, 225 mg/mL uracil, 225 mg/mL histidine, and 225 mg/mL lysine hydrochloride). A non-fermentable carbon source medium, YEGES, containing 0.5% yeast extract, 2% glycerol, 1% ethanol, supplemented with 225 mg/mL adenine sulfate, 225 mg/mL leucine, 225 mg/mL uracil, 225 mg/mL histidine, and 225 mg/mL lysine hydrochloride, was used. PM medium comprised 0.3% potassium hydrogen phthalate, 0.56% sodium phosphate, 0.5% ammonium chloride, 2% glucose, and standard vitamins, minerals, and salts. PMGALU medium contained 0.375% glutamate as the nitrogen source instead of ammonium chloride and was supplemented with adenine sulfate, leucine, and uracil in PM. S. cerevisiae strains were grown in YPD medium (1% yeast extract, 2% peptone, and 2% glucose). Synthetic defined (SD) medium (2% glucose and 6.7 g/L yeast nitrogen base without amino acids (BD Biosciences), containing 19 mg/L adenine sulfate; 76 mg/L each of arginine, histidine, lysine hydrochloride, methionine, uracil, and tryptophan; and 395 mg/L leucine). SD without glucose with glycerol (SD-C+glycerol) medium contained 3% glycerol and 6.7 g/L yeast nitrogen base without amino acids and the same amount of above amino acids, uracil, and adenine sulfate. SC medium consisted of 2% glucose and 6.7 g/L yeast nitrogen base without amino acids, supplemented with 19 mg/L adenine sulfate; 76 mg/L each of alanine, arginine, asparagine, aspartate, cysteine, glutamine, glutamate, glycine, isoleucine, histidine, L-inositol, lysine, methionine, phenylalanine, proline, serine, threonine, tryptophan, tyrosine, uracil, valine; 7.6 mg/L p-aminobenzoic acid; and 395 mg/L leucine).

The oligonucleotide primers used in this study are listed in S1 Table. S. pombe pos5 on the chromosome was disrupted by replacing pos5 with a selectable marker as previously described [24]. The 1.6-kb kanMX6 module was amplified using flanking sequences corresponding to the 5’ and 3’ ends of pos5. Resistant colonies were selected on YES plates containing 100 mg/L G418, and pos5 disruption was verified using colony PCR. DNA fragments of 500–600 bp corresponding to the 5’ or 3’ regions of the gene were amplified by PCR using pos5del-A and pos5del-B or pos5del-C and pos5del-D primer pairs (S1 Table). The amplicons were fused to the ends of the kanMX6 module using PCR. The PR110 strain was transformed with the resulting pos5::kanMX6 fragments to obtain the pos5 disruptant. The chromosomal deletion of pos5 was confirmed by PCR using the nb2 and pos5del-check primers. The obtained strain was designated as RYP7 (Δpos5). Pos5-GFP-tagged strain was constructed using the recombinant PCR approach described in a previous study [24]. The pFA6a-GFP(S65T)-kanMX6 plasmid [24] was used as the template DNA, and the resulting PCR products carried the GFP-kanMX6 cassette in the 3’ region downstream of pos5. The oligonucleotides pos5-TAGW, pos5-TAGX, pos5-TAGY, and pos5-TAGZ were used to construct the pos5-GFP-kanMX6 strain. The resulting pos5-GFP-kanMX6 cassette was introduced into the PR109 strain, and the transformants carrying the GFP-fused pos5 were verified by colony PCR [25]. The S. cerevisiae Δpos5 strain (ΔScpos5; MK1601) was provided by Shigeyuki Kawai (Ishikawa Prefectural University).

The plasmids used in this study were constructed by a method described previously (S2 Table) [10]. Each gene encoding NAD⁺/NADH kinase was PCR amplified using the S. pombe PR110 genome and the S. cerevisiae BY4741 genome as templates, with primers containing restriction sites. The amplified fragments were digested using restriction endonucleases and then inserted into the appropriate sites of the pREP41, pJK148-P_nmt1 or pJK148-P41_nmt1 vector by ligation. pREP41-pos5 was constructed by inserting the PCR product amplified using pos5(SalI)-F and pos5(BamHI)-R primers into the SalI and BamHI sites of pREP41. pREP41-coq6 was constructed by inserting the fragment digested from pREP1-coq6 by SalI and SmaI into the same sites of pREP41 [10]. Further, the other plasmids pREP41-ScPOS5 and UTR1 were also constructed similarly. To construct mitochondrial NAD kinase, mitochondrial-targeting sequence of coq3 from S. pombe was fused to the UTR1 sequence from S. cerevisiae. The mitochondrial transit peptide in S. cerevisiae Pos5p was 62 amino acids from the N-terminus, and its homologous position is 83 amino acids in S. pombe Pos5. Thus, the primers were designed to anneal at 298 bp from the 5’-terminus of Pos5. pJK148-Pnmt1 was constructed from pJK148 and pREP3X. The Pnmt1-MCS-Tnmt1 region was amplified and inserted into KpnI and SacI sites of pJK148. pJK148-P41nmt1 was constructed from pJK148 and pREP41X. The P41nmt1-MCS-Tnmt1 region was amplified and inserted into KpnI and SacI sites of pJK148. pJK148-P41nmt1-pos5 was constructed by inserting the pos5 insert fragment digested from pREP41-pos5 into the SalI and BamHI sites of pJK148-P41nmt1. The other plasmids pJK148-P41nmt1-ScPOS5, UTR1, Spcoq3MTS36UTR1, ΔMTS83pos5, and pJK148-Pnmt1-coq6 were constructed similarly. To examine the cellular localization of Pos5, GFP fusion was generated by inserting pos5 into the pSLF272L-GFP(S65A) vector [26,27]. pSLF272L-pos5-GFP(S65A) was constructed by inserting the PCR product amplified using the pos5-GFP(XhoI)-F and pos5-GFP(NotI)-R3 primers into XhoI and NotI sites of pSLF272L-GFP(S65A). The genes amplified by PCR were verified using DNA sequencing.

Yeast precultures were inoculated into large-volume media and incubated for the indicated times. Unless otherwise specified, strains were grown at 30°C in 55 mL of liquid media (with or without specific supplements), starting from an initial density of 1 × 10⁵ cells/mL, and cultured for 48 or 72 hours. Cell numbers were counted using a Sysmex CDA-1000B (Sysmex, Tokyo, Japan), and the OD₆₀₀ was measured using a Shimadzu UVmini-1240 spectrophotometer (Shimadzu, Kyoto, Japan). Cells were harvested, and CoQ was extracted using the autoclave method as described previously [10]. Prior to extraction, 5 µg of CoQ₆ was added to each sample as an internal standard. Crude CoQ samples were separated by normal-phase thin-layer chromatography using a Kieselgel 60 F₂₅₄ plate (Merck Millipore, MA, USA). The TLC was developed with benzene as the solvent. After development, the TLC plate was visualized under UV illumination, and the bands corresponding to CoQ₆ and CoQ₁₀ were excised and extracted with hexane/isopropanol (1:1, v/v). The sample solvents were evaporated, and the dried solids were dissolved in ethanol. Purified CoQ samples were analyzed using high-performance liquid chromatography on a Shimadzu HPLC Class VP series instrument (Shimadzu). A reversed-phase YMC-Pack ODS-A column (A-312–3 AA12S03-1506PT, 150 × 6 mm, 3-μm particle size, 120 Å, YMC, Kyoto, Japan) was used. The mobile phase consisted of ethanol at a flow rate of 1.0 mL/min. CoQ₆, and CoQ₁₀ were detected by UV absorption at 275 nm.

Yeast cells were pre-cultured for 24 hours in 100 mL of YES medium and then inoculated into 3 L of YES medium. After incubation for 16–20 hours, cells were harvested at OD₆₀₀ = 1. Mitochondria were isolated according to a previously described method [27] with slight modifications. In the current experiment, we incubated the pellet with 100 mM Tris-SO₄ and 10 mM DTT for 30 minutes at 30°C. To improve the yield of mitochondria, the pellet obtained by the initial homogenization and centrifugation was resuspended in a buffer containing 0.6 M mannitol, 20 mM HEPES-KOH, 0.5 mM EDTA, and 1 mM PMSF, and further homogenized 15 times.

The concentrations of NADP⁺ and NADPH in isolated yeast mitochondria extracts were determined using an enzymatic cycling assay according to a method reported previously [21,28,29]. Briefly, 50 µL of each sample was mixed with an equal volume of either 0.1 N HCl (for NADP⁺ measurement) or 0.1 N KOH (for NADPH measurement), followed by incubation at 85°C for 3 min. Subsequently, the treated extracts and the corresponding NADP⁺ or NADPH standards were added to a reaction mixture to a final volume of 200 µL containing 100 mM HEPES-KOH (pH 8.0), 0.5 mM EDTA, 2.5 mM glucose-6-phosphate (G6P), 1.66 mM phenazine ethosulfate, and 0.42 mM MTT (3-(4,5-dimethyl-2-thiazolyl)-2,5-diphenyl-2H-tetrazolium bromide). The reaction was initiated by the addition of 0.5 U of G6P dehydrogenase, and absorbance at 570 nm (A₅₇₀) was measured using a Corona SH-9000Lab microplate reader (Hitachi, Tokyo, Japan).

Mitochondria were stained using the MitoTracker Red FM dye (Invitrogen, Thermo Fisher Scientific, Inc). Cells were suspended in PMU medium and incubated with 50 nM MitoTracker Red FM at room temperature for 1 hour. Imaging was performed at 1000x magnification using a BX2-FL-2 fluorescence microscope (Olympus). GFP(S65A) fluorescence was observed at an excitation wavelength of 485 nm. Fluorescent images were obtained using a DP74-SET-A digital camera (Olympus) connected to the microscope and processed using cellSens ver.2.2 (Olympus).

Data from control and experimental samples were compared using the two-sample t-tests in Microsoft Excel (WA, USA). p-values <0.05 were considered statistically significant. Data from control and experimental samples were compared using one-way ANOVA with a post hoc test (Dunnett’s test) performed with EZR (Jichi Medical University, Tochigi, Japan) [30]. EZR is a graphical user interface for R (The R Foundation for Statistical Computing, Vienna, Austria). More precisely, it is a modified version of R commander designed to add statistical functions frequently used in biostatistics.

We have previously investigated the genes involved in CoQ biosynthesis of S. pombe using a Bioneer gene-deletion library and obtained approximately 40 individual gene-deleted strains with a CoQ₁₀ content lower than that of the wild-type strain [17]. In this study, we selected a Δpos5 strain from these strains for further analysis because it exhibited respiration deficiency, similar to CoQ-deficient strains, in addition to low CoQ₁₀ production. We independently constructed a Δpos5 strain to ensure that the phenotype observed in the Δpos5 strain from Bioneer Corp. is the same as our construct. CoQ levels in the Δpos5 strain of our construct were decreased to 0.2-fold of those in the wild-type strain (Fig 1A and 1B) as in the originally screened Bioneer Δpos5 strain (Fig 1C and 1D). Subsequently, we examined the phenotypes previously observed in CoQ-deficient strains of S. pombe, which exhibit respiratory deficiency, growth delay in minimal media, H₂O₂ sensitivity, and enhanced H₂S production [13,31]. The Δpos5 strain failed to grow on YEGES medium containing glycerol and ethanol as non-fermentable carbon sources and showed retarded growth on YES containing hydrogen peroxide as well as on minimal medium (Fig 1E). Supplementation with arginine partially restored Δpos5 growth (S1 Fig) as observed previously [32]. This is due to the requirement of NADPH for Arg11-catalyzed reaction in arginine biosynthesis. When grown on YES containing CuSO₄, Δpos5 colonies also developed a brown coloration, similar to the Δdps1 strain, which is completely defective in CoQ₁₀ synthesis (Fig 1E). In addition, Δpos5 cells showed a round morphology, which is often seen in the mutants related to sexual differentiation [33]. The phenotype was reverted to the normal rod shape upon expression of pos5 or mitochondrially targeted UTR1, which encodes a cytosolic NADK responsible for NADP(H) synthesis in S. cerevisiae, resembling the morphology of wild-type cells (S2 Fig). To determine whether the reduced CoQ level in the Δpos5 strain is simply a consequence of defective respiration, we next compared the CoQ content of the Δpos5 strain with that of a cytochrome c-deficient respiration mutant (Δcyc1) (Fig 1F and 1G). The Δcyc1 strain did not show a marked decrease in CoQ levels, suggesting that respiratory deficiency alone does not account for the low CoQ level in the Δpos5 strain. These results indicate that Pos5 is specifically important in CoQ biosynthesis in S. pombe.

S. pombe pos5 gene is predicted to encode a mitochondrial NAD(H) kinase because the Pos5 protein shares 37% identity with S. cerevisiae Pos5, a well-characterized mitochondrial NADH kinase (Fig 2A) [20,34]. To verify the functional similarity of SpPos5 and ScPos5, we constructed the Δpos5 + pJK148-P41nmt1-ScPOS5 strain (NSP7), in which ScPos5 was integrated at the chromosomal leu1 locus of the S. pombe Δpos5 strain, and measured its CoQ content. CoQ levels in the NSP7 strain were recovered to 0.9-fold of that observed in the Δpos5 + pos5 strain (NSP11) (Fig 2B and 2C), indicating functional similarity of these two proteins.

In the S. cerevisiae Δpos5 strain, mitochondrial NADP(H) levels are decreased [21]. To determine whether the S. pombe Δpos5 strain also influences mitochondrial NADP(H), we isolated mitochondria from the S. pombe Δpos5 strain and quantified NADP(H) content as described in materials and methods. Mitochondrial NADP⁺ level and total NADP(H) level in Δpos5 were decreased to 0.6- and 0.8-fold, respectively, of those in the wild-type strain (Fig 3A and 3B), supporting that S. pombe pos5 encodes NADP(H) kinase.

The S. cerevisiae Δpos5 (ΔScpos5) strain has previously been reported to exhibit respiratory deficiency, hydrogen peroxide sensitivity, and arginine auxotrophy [19]. We confirmed these phenotypes (Fig 4A). However, since the role of ScPos5 in CoQ biosynthesis has never been documented, we quantified CoQ levels in a ΔScpos5 strain. CoQ₆ levels in ΔScpos5 were decreased to 0.2-fold of those in the wild-type strain (Fig 4B and 4C), which is similar to the CoQ deficiency observed in the S. pombe Δpos5 mutant (Fig 1). Thus, ScPos5 is also involved in CoQ biosynthesis in S. cerevisiae to a similar extent as observed in the S. pombe Δpos5 strain.

We next investigated the localization of the Pos5-GFP strain, in which Pos5-GFP was expressed from the pos5 locus. The Pos5-GFP signal did not show the expected mitochondrial localization pattern (S3A Fig). In addition, the Pos5-GFP strain failed to maintain normal CoQ production and showed a CoQ level similar to the Δpos5 strain (S3B and S3C Fig), indicating a loss of Pos5 function by tagging GFP which probably interfered Pos5 function. Therefore, we constructed a pSLF272L-Pos5-GFP(S65A) plasmid to express Pos5-GFP exogenously and examined Pos5 localization by introducing it into the wild-type strain. The Pos5-GFP fluorescence overlapped with the MitoTracker Red FM signal (Fig 5), indicating that Pos5 localizes to mitochondria as a mitochondrial NAD(H) kinase. It also indicates Pos5-GFP retains partial functionality, because multicopy Pos5-GFP but not a single copy of that is functional.

Given the phenotypes of the pos5 strain are specific for mitochondrial function, the mitochondrial localization of Pos5 NAD(H) kinase is thought to be essential for CoQ synthesis as shown in S. cerevisiae [20]. To directly examine the significance of mitochondrial localization of NAD(H) kinase in CoQ biosynthesis, we constructed pJK148-P41nmt1-ΔMTS83pos5, pJK148-P41nmt1-UTR1, and pJK148-P41nmt1-MTS36UTR1 plasmids. The ΔMTS83pos5 construct is designed to express a Pos5 protein lacking the N-terminal 83 amino acids. The UTR1 construct expresses a cytosolic NAD(H) kinase from S. cerevisiae. The MTS36UTR1 construct is designed to express a fusion protein comprising the N-terminal 36 amino acids of SpCoq3 fused to S. cerevisiae Utr1p. These constructs were introduced into the Δpos5 strains to generate Δpos5 + ΔMTS83pos5 (NSP15), Δpos5 + UTR1 (NSP13), and Δpos5 + MTS36UTR1 (NSP12) strains. CoQ quantification showed that only MTS36UTR1 restored CoQ production, whereas neither ΔMTS83pos5 nor UTR1 could recover CoQ levels in the Δpos5 strain (Fig 6A and 6B). Consistently, a mitochondrial-targeted Utr1 restored CoQ levels in the Δpos5 strain (S4 Fig) and Utr1-GFP fusion (mito-UTR1-GFP) localized correctly to mitochondria and restored CoQ levels in such a strain (S5 Fig). Thus, these results demonstrate that cytosolic NAD(H) kinase from S. cerevisiae can replace the function of mitochondrial NAD(H) kinase when it is expressed in mitochondria, indicating that the localization of NAD(H) kinase to mitochondria is critical for CoQ biosynthesis.

Among the reactions in CoQ biosynthesis, Coq6 catalyzes C5-hydroxylation of the quinone precursor and requires reducing equivalents from NAD(P)H, through ferredoxin and ferredoxin reductase [35–37]. Ferredoxin is reduced by ferredoxin reductase utilizing NAD(P)H to provide electrons to Coq6 reaction in S. cerevisiae. In S. pombe, the ferredoxin reductase Arh1 utilizes both NADPH and NADH [38]. Based on these observations, we hypothesized Coq6 activity may be impaired in the Δpos5 strain. To test this, we overexpressed coq6 in the Δpos5 strain and measured CoQ content. However, CoQ levels in the Δpos5 strain overexpressing coq6 were comparable to that in the Δpos5 strain integrating a vector (Fig 7A and 7B).

In contrast, the Δcoq6 strain expressing coq6 on the chromosome clearly restored the CoQ level (S6 Fig). Because exogenous vanillic acid (VA) is known to restore CoQ levels in the Δcoq6 strain (Fig 8A) [17], we added VA to the Δpos5 strain. VA clearly increased CoQ levels in the Δcoq6 strain (Fig 8B and 8C). Although statistical difference was not observed, the addition of VA tended to increase CoQ levels in the Δpos5 strain (Fig 8D and 8E). These results suggest that Coq6 is not fully functional in the Δpos5 strain, but that impaired Coq6 activity is not the sole reason for the decreased CoQ content.

We next examined the effect of p-hydroxybenzoate (PHB), a quinone precursor, on CoQ production in the Δpos5 strain. PHB is condensed with decaprenyl diphosphate by Ppt1 to synthesize decaprenyl-PHB, which subsequently undergoes modifications to generate CoQ₁₀ [39]. Supplementation of 0.5 mM PHB partially increased CoQ levels in the Δpos5 strain compared to the untreated condition (Fig 8F and 8G), suggesting that NADP(H) availability affects a reaction upstream of CoQ biosynthesis.

We then tested overexpression of the atd1 gene, which encodes a potential enzyme that converts p-hydroxybenzaldehyde to PHB, in the Δpos5 strain to see any effect on CoQ biosynthesis. The result showed slight increased CoQ levels in such a strain comparing with the one without the atd1 expression (S7 Fig), but the difference was not statistically significant.