Two Sources of Mitochondrial NADPH in the Yeast Saccharomyces cerevisiae*

Cells of the yeast Saccharomyces cerevisiae contain three NAD kinases; namely, cytosolic Utr1p, cytosolic Yef1p, and mitochondrial Pos5p. Previously, the NADH kinase reaction catalyzed by Pos5p, rather than the NAD kinase reaction followed by the NADP+-dependent dehydrogenase reaction, had been regarded as a critical source of mitochondrial NADPH, which plays vital roles in various mitochondrial functions. This study demonstrates that the mitochondrial NADH kinase reaction is dispensable as a source of mitochondrial NADPH and emphasizes the importance of the NAD kinase reaction, followed by the mitochondrial NADP+-dependent dehydrogenase reaction. Of the potential dehydrogenases (malic enzyme, Mae1p; isocitrate dehydrogenase, Idp1p; and acetaldehyde dehydrogenases, Ald4/5p), evidence is presented that acetaldehyde dehydrogenases, and in particular Ald4p, play a prominent role in generating mitochondrial NADPH in the absence of the NADH kinase reaction. The physiological significance of the mitochondrial NADH kinase reaction in the absence of Ald4p is also demonstrated. In addition, Pos5p is confirmed to have a considerably higher NADH kinase activity than NAD kinase activity. Taking these results together, it is proposed that there are two sources of mitochondrial NADPH in yeast: one is the mitochondrial Pos5p-NADH kinase reaction and the other is the mitochondrial Pos5p-NAD kinase reaction followed by the mitochondrial NADP+-dependent acetaldehyde dehydrogenase reaction.

NADPH plays vital roles in reactions that protect against oxidative stress as well as participating in a large number of biosynthetic reactions (1). It is generated by the NAD kinase (EC 2.7.1.23) reaction followed by the NADP ϩ -dependent dehydrogenase reaction. NAD kinase catalyzes the phosphorylation of NAD ϩ to give NADP ϩ , and NADP ϩ -dependent dehydrogenase reduces the NADP ϩ to yield NADPH. NADPH is also synthesized by the activity of NADH kinase (EC 2.7.1.86) or pyridine nucleotide transhydrogenase (EC 1.6.1.1) (1). NADH kinase catalyzes the phosphorylation of NADH to give NADPH, whereas pyridine nucleotide transhydrogenase transports protons across the membrane in concert with hydride exchange between NADH and NADP ϩ or NAD ϩ and NADPH, resulting in the formation of NADPH from NADP ϩ (1).
In contrast to the cytosol, the major source of mitochondrial NADPH in S. cerevisiae is the mitochondrial NAD kinase Pos5p (6) (see Fig. 1). pos5 exhibits several phenotypes, which either directly or indirectly result from decreased mitochondrial NADPH. The phenotypes include Arg Ϫ and sensitivity to oxidative stresses (paraquat, hyperoxia, and H 2 O 2 ), slow growth on non-fermentable carbon sources, defective biosynthesis of enzymes containing the Fe-S cluster, up-regulated transcription of the genes for iron uptake, abnormal accumulation of iron in the mitochondria, and accumulation of mutations in mitochondrial DNA (6,9,10).
In S. cerevisiae, mitochondrial NADP ϩ -dependent dehydrogenase appears to be nonessential as a source of mitochondrial NADPH. In mammals, mitochondrial NADP ϩ -dependent isocitrate dehydrogenase is a source of mitochondrial NADPH (11). Decreased expression of isocitrate dehydrogenase causes severe phenotypes, including elevation of reactive oxygen species (ROS) 2 production, lipid peroxidation, and mitochondrial * This work was supported by Grant-in-Aid 19780057 from the Ministry of Education, Culture, Sports, Science and Technology of Japan (to S. K.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. □ S The on-line version of this article (available at http://www.jbc.org) contains supplemental Tables S1 and S2. 1 To whom correspondence should be addressed. damage (11). In the case of S. cerevisiae, however, the lack of a mitochondrial NADP ϩ -dependent isocitrate dehydrogenase gene (IDP1) results in no detectable phenotype in the presence of 0.5 mM H 2 O 2 (12). The higher NADH kinase activity of Pos5p compared with its NAD kinase activity led Otten and Culotta (6,13) to speculate that the NADH kinase reaction catalyzed by Pos5p, rather than the NAD kinase reaction, is critical to the supply of mitochondrial NADPH. However, these authors presented no evidence to corroborate this speculation, and neither were the kinetic values of Pos5p determined. Only two studies have demonstrated that Pos5p uses NADH in preference to NAD ϩ (6,9). Moreover, in S. cerevisiae, there are other known mitochondrial NADP ϩ -dependent dehydrogenases such as malic enzyme (Mae1p) and two acetaldehyde dehydrogenases (Ald4/Ald5p) (Fig. 1), although there is no pyridine nucleotide transhydrogenase in the mitochondria (14 -16). The possibility remains that, in addition to the NAD kinase reaction catalyzed by Pos5p, any one of these other dehydrogenases may contribute physiologically to the supply of mitochondrial NADPH.
This study focuses on the reactions responsible for the generation of mitochondrial NADPH in S. cerevisiae (BY4742 background). Initially, we demonstrate that the NAD kinase triple mutant (utr1yef1pos5) is lethal, and using this triple mutant we reveal the dispensability of the mitochondrial NADH kinase reaction, thus emphasizing that the NAD kinase reaction followed by the NADP ϩ -dependent dehydrogenase reaction is able to produce mitochondrial NADPH. We also demonstrate that mitochondrial NADP ϩ -dependent acetaldehyde dehydrogenases (Ald4/Ald5p), and in particular Ald4p, are critical as a source of mitochondrial NADPH. The physiological significance of the NADH kinase reaction in the absence of Ald4p is also demonstrated.

EXPERIMENTAL PROCEDURES
Plasmids-The plasmids and primers used in this study are listed in Table 1 and supplemental Table S1, respectively. POS5 and its 406-bp upstream region (P POS5 ) were amplified by PCR using the genomic DNA from S. cerevisiae BY4742 and then inserted into pRS415, yielding pMK1643 (P POS5 ϩPOS5). Similarly, a BamHI fragment consisting of UTR1 plus its 503-bp upstream region (P UTR1 ) from YCp-UTR1 was inserted into pRS415, yielding pMK1702 (P UTR1 ϩUTR1). Following disruption of the endogenous NcoI site in the POS5 insert of pMK1643 to give pMK1645, NcoI sites were again introduced at positions ϩ1 and ϩ185 of the POS5 insert of pMK1645 to yield pMK1646 and pMK1647, respectively. UTR1 from pSK49 and yfjB from pSK65 were then inserted into the NcoI/BamHI sites of pMK1646 and pMK1647, resulting in pMK1700 (P POS5 ϩUTR1), pMK1701 (P POS5 ϩyfjB), pMK1722 (P POS5 ϩ 62MTSϩUTR1), and pMK1723 (P POS5 ϩ62MTSϩyfjB). 62MTS encodes a Pos5p-specific N-terminal additional sequence (62 amino acid residues) containing a putative mitochondrial targeting sequence (MTS) (Fig. 2) (9). Although a deficiency in nucleotide A at the ϩ957 position of POS5 in pMK1643 was later found, P POS5 plus the correct POS5 was again inserted into pRS415, yielding pMK2127 (P POS5 ϩPOS5). An NcoI site was again introduced at the ϩ1 position of POS5 in pMK2127, giving pMK2147 (P POS5 ϩPOS5 (NcoI at ϩ1)). MTS was removed from POS5 in pMK2127, giving pMK2145 (P POS5 ϩPOS5⌬MTS) encoding Pos5⌬MTSp (Fig. 2), using the primers pos5f-17 and pos5r-17p normal (supplemental Table  S1) and pMK2127 as a template. An NdeI site was then introduced at the ϩ1 position of POS5⌬MTS using the primers pos5f-17 and pos5rNdeI-17p (supplemental Table S1) and pMK2127 as a template, giving pMK2148. The NdeI/BamHI fragment (POS5⌬MTS) from pMK2148 was then inserted into pET-28b (Novagen, Darmstadt, Germany), yielding pMK2159. Accurate synthesis of all the constructed plasmids was confirmed by DNA sequencing.
Subcellular Fractionation-Subcellular fractionation into a mitochondrial fraction and a post-mitochondrial supernatant (PMS) fraction was conducted as described previously (21). The solution containing mitochondria was sonicated to disrupt the mitochondria and centrifuged at 4°C and 20,000 ϫ g for 10 min, and the supernatant was used as the mitochondrial fraction.
Assays-Protein concentrations were determined as described previously using bovine serum albumin as a standard (22). All activities were assayed at 30°C. NAD kinase activity was assayed in a 1.0-ml reaction mixture containing 5.0 mM NAD ϩ , 5.0 mM ATP, 5.0 mM MgCl 2 , and 100 mM Tris-HCl, pH 8.0. The reaction was terminated by heating the mixture in boiling water for 5 min. To assay the NAD kinase activity of the mitochondrial or PMS fraction from yeast, the NADP ϩ formed was determined by a cycling assay in a 1.0-ml reaction mixture containing 5.0 mM glucose 6-phosphate, 0.5 unit of glucose-6phosphate dehydrogenase, 0.2 mg/ml thiazolyl blue tetrazolium bromide, and 0.03 mg/ml 1-methoxy-5-methylphenazinium methyl sulfate (23). To assay the NAD kinase activity of purified Pos5⌬MTSp (to determine the optimum pH), the NADP ϩ formed was determined without either thiazolyl blue tetrazolium bromide or 1-methoxy-5-methylphenazinium methyl sulfate (5), and the assay was conducted in the presence of 2.0 mM NAD ϩ . The NAD kinase activity of purified Pos5⌬MTSp or purified Escherichia coli NAD kinase (YfjB) was also continuously assayed in a 1.0-ml reaction mixture contain-ing 5.0 mM NAD ϩ , 5.0 mM ATP, 5.0 mM MgCl 2 , 5.0 mM glucose 6-phosphate, 0.5 unit of glucose-6-phosphate dehydrogenase, and 100 mM Tris-HCl, pH 8.0 (5). The NADH kinase activity was assayed as described previously (5) in a 1.0-ml reaction mixture containing 2.0 mM NADH, 5.0 mM ATP, 5.0 mM MgCl 2 , and 100 mM Tris-HCl, pH 8.0. The reaction was terminated by the addition of 0.1 ml of 1.0 M NaOH followed by immediate immersion of the test tube in boiling water for 1.5 min. We confirmed that this boiling treatment under alkaline conditions completely degrades at least 0.1 mM NADP ϩ (data not shown). The mixture was neutralized by the addition of 0.3 ml of neutralization solution (0.5 M triethanolamine-HCl, 0.4 M Tris-HCl, 25 mM NH 4 Cl, 25 mM ␣-ketoglutarate, pH 7.8). The NADH and NADPH thus formed were enzymatically oxidized to NAD ϩ and NADP ϩ , respectively, by the addition of 12.5 units of glutamate dehydrogenase followed by incubation at 30°C for 10 min. Oxidation was monitored by observing the decrease in A 340 . After the oxidation reaction was terminated by immersing the test tube in boiling water for 5 min, the amount of NADP ϩ was determined. For the assay of the NADH kinase activity of purified Pos5⌬MTSp or purified YfjB, the amount of NADP ϩ was determined as described above without either thiazolyl blue tetrazolium bromide or 1-methoxy-5methylphenazinium methyl sulfate. For the assay of the NADH kinase activity of the mitochondrial or PMS fraction from yeast, the amount of NADP ϩ was determined by the cycling assay as described above. Glucose-6-phosphate dehydrogenase activity itself was continuously assayed in a 1.0-ml reaction mixture containing 1.0 mM glucose 6-phosphate, 0.5 mM NADP ϩ , 10 mM MgCl 2 , and 100 mM Tris-HCl, pH 8.0 (24). Cytochrome c oxidase activity was assayed using a cytochrome c oxidase assay kit (CYTOCOX1, Sigma). One unit of enzyme activity was defined as 1.0 mol of product formed in 1 min at 30°C, and specific activity was expressed in units/mg of protein.
Expression and Purification-Pos5⌬MTSp was expressed in E. coli RosettaBlue (Novagen) carrying pMK2159 (MK2162) as POS5⌬MTS in pET-28b, from NcoI/BamHI fragments of pMK2148 and pET-28b This study a POS5 in pMK1643 lacks nucleotide A at ϩ957. b NcoI in POS5 was disrupted by changing ϩ226 CCATGG ϩ231 to ϩ226 CCTTGG ϩ231 but had no effect on the encoding of amino acids. c NcoI was introduced into ϩ1 of POS5 in pMK1645 by changing Ϫ2 AAATGT ϩ4 to Ϫ2 CCATGG ϩ4, resulting in a change of encoded residues from 1 MF 2 to 1 MV 2 . d NcoI was introduced into a site at ϩ185 of POS5 in pMK1645 by changing ϩ185 TCTGGC ϩ190 to ϩ185 CCATGG ϩ190, resulting in a change in the encoded residues from 62 IWQ 64 to 62 TME 64 . e MTS (48 bp: ϩ4 to ϩ51), encoding 16 amino acid residues, was removed from POS5. f NdeI was introduced into ϩ1 of POS5⌬MTS by changing Ϫ3 AAAATG ϩ3 to Ϫ3 CATATG ϩ3, giving no change of residues. described previously (9), except that expression was induced at 16°C for 3 days after the addition of isopropyl thio-␤-galactoside to a final concentration of 0.2 mM. Purification was conducted as described previously (9). Purified Pos5⌬MTSp (7.6 mg) was obtained from an approximate 350-ml culture. The Pos5⌬MTSp expressed from MK2162 contains a tag (MGSSH-HHHHHSSGLVPRGSH) at its N terminus, but no tag at its C terminus. YfjB was expressed and purified as described previously (25) and was stored at Ϫ30°C.
Phylogenic Tree-A BLAST search (26) was conducted using the primary structure of Pos5p as the query and the Kyoto Encyclopedia of Genes and Genomes (KEGG) as the data base. Using the primary structure of the 50 proteins that show the highest homology with Pos5p, as well as that of Arabidopsis thaliana NADK3 (27) exhibiting only about a 750th homology with Pos5p, a ClustalW search (28) was performed. An unrooted NJ tree was then constructed. These searches were performed using the KEGG web site. Localization was predicted using TargetP (29).

RESULTS
Lethality of Utr1yef1pos5-We previously reported that the S. cerevisiae NAD kinase triple mutant utr1yef1pos5 (MK1219: BY4742 utr1⌬::kanMX4 yef1⌬::HIS3 pos5⌬::CgLEU2) was viable (5). However, during further analysis of MK1219, we found that, although POS5 was certainly replaced by CgLEU2, native POS5 was still unexpectedly detected by PCR in the genomic DNA of MK1219. Therefore, another triple mutant was constructed carrying YCp-UTR1 (MK1598 (utr1⌬::kanMX4 yef1⌬::HIS3 pos5⌬::hphMX4 YCp-UTR1)). It was confirmed that POS5 was replaced by hphMX4, because no native POS5 was detectable by PCR in the genomic DNA of MK1598. MK1598 was Arg Ϫ and was unable to grow on medium containing 5-fluoroorotic acid (data not shown), indicating that the triple mutant is lethal. Furthermore, it was subsequently reported that the utr1pos5 double mutant in S288C and SEY6210.5 backgrounds is lethal, and hence the utr1yef1pos5 triple mutant in a SEY6210.5 background is also lethal (7,10). In addition, re-examination of the genotype of the double mutant utr1pos5 (MK803: BY4742 utr1⌬::kanMX4 pos5⌬::HIS3) (5) revealed that UTR1 was not correctly replaced by kanMX4 in MK803. Thus, we concluded that the NAD kinase triple mutant is lethal and realized that MK1219 and MK803 were not the true triple and utr1pos5 double mutants, respectively.
Dispensability of the Mitochondrial NADH Kinase Reaction for the Generation of Mitochondrial NADPH-YfjB has previously been demonstrated to exhibit no NADH kinase activity (8,25). In the present study, we re-confirmed that purified YfjB (0.26 g), which has detectable NAD kinase activity (⌬A 340 of 0.56 for a 10-min reaction), exhibits no NADH kinase activity after a 10-min reaction. We further confirmed that purified excess YfjB (5.2 g and 130 g) exhibits no NADH kinase activity after a 10-min reaction, which is consistent with our previous results (8).
In Pos5p, the N-terminal amino acid sequence consisting of 17 residues is predicted to be the MTS required for transfer of Pos5p to the mitochondria (Fig. 2) (9). Alignment of the primary structures of Pos5p and YfjB shows that, although they are very similar to each other, Pos5p has an additional sequence (62MTS) consisting of 62 amino acid residues (Fig. 2). This suggested that Utr1p and YfjB might be delivered to the mitochondria if they are fused to the 62MTS.
MK1598 (utr1yef1pos5 carrying YCp-UTR1) cells were transformed with pRS415 (LEU2) or pRS415-based plasmids as shown in Fig. 3. The viability of each transformant was examined on solid medium containing 5-fluoroorotic acid. Although the transformant with pRS415 alone was not viable, the other transformants were at least viable in the presence of Arg and were Ura Ϫ (data not shown), indicating that utr1yef1pos5 survives due to the help of the NAD kinase gene supplied from the pRS415-based plasmid and that the NADH kinase reaction is dispensable for cellular viability.
The mutants were Arg ϩ and exhibited resistance to H 2 O 2 (H 2 O 2 r ) in the presence of 62MTS, whereas they were Arg Ϫ and exhibited sensitivity to H 2 O 2 (H 2 O 2 s ) in the absence of 62MTS or MTS (Fig. 3). This result also implied that 62MTS-tagged enzymes (Utr1p and YfjB) are delivered to the mitochondria, whereas untagged-enzymes or Pos5⌬MTSp fail to enter the mitochondria. If this is the case, it can be concluded that the mitochondrial NADH kinase reaction is dispensable for the supply of mitochondrial NADPH.
Delivery of 62MTS-tagged Enzymes to the Mitochondria-To demonstrate possible delivery of 62MTS-tagged enzymes to the mitochondria, triple mutants carrying POS5 as well as UTR1 and yfjB with or without 62MTS were biochemically fractionated into mitochondrial and PMS fractions (Fig. 4). The NAD kinase activity in each fraction was then assayed. The activities of cytochrome c oxidase and glucose-6-phosphate dehydrogenase, which are markers for the mitochondrial and PMS fractions (30), respectively, ensured correct fractionation (supplemental Table S2). Alignment was conducted using ClustalW (28). The Pos5p-specific additional N-terminal amino acid sequence consisting of 62 amino acid residues (62MTS) is underlined. The putative MTS is double-underlined, and the putative cleavage site is denoted by an arrow (9). Pos5⌬MTSp has an N-terminal sequence, MSTLDSHS, lacking 16 residues (2nd to 17th residue) from Pos5p. Identical residues are denoted by an asterisk (*), strongly conserved residues by a colon (:) and weakly conserved residues by a period (.).
NAD kinase activity was detected in the mitochondrial fraction from the mutants carrying POS5 or UTR1 and yfjB with 62MTS, but not in the PMS fraction. Conversely, activity was detected in the PMS fraction, but not in the mitochondrial fraction, in the absence of 62MTS (Fig. 4A), demonstrating that the 62MTS-tagged enzymes (Utr1p and YfjB) as well as Pos5p are delivered to the mitochondria, whereas untagged enzymes are not. Furthermore, NADH kinase activity was detected in the mitochondrial fraction from the mutants carrying POS5, but not in the PMS fraction (Fig. 4B). In the mito-chondrial fraction from the mutants carrying UTR1 and yfjB with 62MTS, no NADH kinase activity was detected (Fig. 4B). We attribute this lack of NADH kinase activity, at least that of the 62MTS-tagged Utr1p, to the sensitivity of the assay system, because purified Utr1p shows lower NADH kinase activity than NAD kinase activity (Table 2) (7). Collectively, these results suggest that the mitochondrial NADH kinase reaction is dispensable for the supply of mitochondrial NADPH.
The Mitochondrial NADP ϩ -dependent Dehydrogenase Reaction as a Source of Mitochondrial NADPH-The dispensability of the NADH kinase reaction emphasizes the significance of the NAD kinase reaction as well as the mitochondrial NADP ϩ -dependent dehydrogenase for the generation of mitochondrial NADPH. Pyridine nucleotide transhydrogenase activity is not detected in the mitochondria of S. cerevisiae (16). The known mitochondrial NADP ϩ -dependent dehydrogenases of S. cerevisiae are isocitrate dehydrogenase (Idp1p) (31), malic enzyme (Mae1p) (14), and acetaldehyde dehydrogenases (Ald4p and Ald5p) (15).
To evaluate the contribution of each dehydrogenase, the phenotypes of pos5, idp1, mae1, ald4, and ald5 carrying only pRS415 were examined.  (Fig. 5A). However, idp1 mutants carrying POS5 (pMK2127) or 62MTSϩyfjB (pMK1723) were also H 2 O 2 r (Fig. 5B). Moreover, pos5idp1 cells carrying 62MTSϩyfjB were resistant to 2.0 mM H 2 O 2 (Fig. 5C), indicating that Idp1p is not required for the generation of mitochondrial NADPH, even in the absence of the NADH kinase reaction.
pos5mae1 cells carrying 62MTSϩyfjB were also H 2 O 2 r (Fig. 5C) (Fig. 5C). pos5ald5 carrying 62MTSϩyfjB was slightly H 2 O 2 s (Fig.  5C). The possibility that 62MTSϩyfjB was not functionally expressed in pos5ald4 and pos5ald5 can be excluded, because both pos5ald4 and pos5ald5 carrying 62MTSϩyfjB were Arg ϩ (Fig. 5D), whereas both pos5ald4 and pos5ald5 carrying pRS415 alone were Arg Ϫ (data not shown). This indicates that the NAD kinase reaction catalyzed by the expressed 62MTSϩyfjB followed by the dehydrogenase reaction can supply the NADPH needed for Arg biosynthesis, but cannot supply that required to protect cells against 2.0 mM H 2 O 2 .
The H 2 O 2 s phenotype of pos5ald4 carrying 62MTSϩyfjB indicated the essentiality of the mitochondrial acetaldehyde dehydrogenase reaction catalyzed by FIGURE 3. Growth phenotypes of the NAD kinase triple mutants (utr1yef1pos5) having POS5, UTR1, and yfjB with or without 62MTS. The triple mutants carrying the indicated plasmids containing a denoted insert were spotted onto SD solid media with Arg (ϩArg) or without Arg (ϪArg), and SD solid medium containing both Arg and 2 mM H 2 O 2 (ϩArg ϩH 2 O 2 ). As a control, wild-type (BY4742) cells carrying pRS415 alone (marked by asterisk) were also spotted. The triple mutant carrying POS5 (NcoI site at ϩ1) (pMK2147) was Arg ϩ and H 2 O 2 r , indicating that Arg Ϫ and H 2 O 2 s in the absence of 62MTS were not attributed to the NcoI site that was introduced at ϩ1 of UTR1 and yfjB in pMK1700 and pMK1701, respectively. The mutant carrying P UTR1 ϩUTR1 (pMK1702) was also Arg Ϫ and H 2 O 2 s and grew at approximately the same rate as that carrying P POS5 ϩUTR1 (pMK1700) (data not shown), indicating that replacement of P UTR1 by P POS5 had no effect.  Table S2). The NAD kinase (A) and NADH kinase (B) activities in each fraction were assayed. NADH kinase activity in the fractions from mutants without 62MTS was not determined. NAD kinase activity was assayed in the presence of mitochondrial fractions from mutants carrying pMK1700 (36 and 72 g), pMK1722 (16,32,34,63,68, and 126 g), pMK1701 (28 and 56 g), pMK1723 (20,22,34,50,68, and 100 g), or pMK2127 (20 and 40 g), and in the presence of PMS fractions from mutants carrying pMK1700 (17 and 34 g), pMK1722 (39, 42, 60, and 120 g), pMK1701 (12 and 24 g), pMK1723 (27,42,49, and 98 g), or pMK2127 (28 and 43 g) for 30-and 60-min reactions. NADH kinase activity was assayed in the presence of mitochondrial fractions from mutants carrying pMK1722 (16,32,52, and 208 g), pMK1723 (18, 36, 52, and 208 g), or pMK2127 (17, 20, 34, and 40 g), and in the presence of PMS fractions from mutants carrying pMK1722 (39 and 128 g), pMK1723 (45 and 128 g), or pMK2127 (47 and 55 g) for 10-and 20-min reactions. NADH kinase activity of mitochondrial fractions from mutants carrying pMK1722 (16,32, and 52 g) or pMK1723 (52 g) was not detected even after 60-, 120-, 240-, and 600-min reactions. Means Ϯ S.D. values of NAD kinase and NADH kinase activities are presented (details are described in the legend of supplemental Table S2). ND, not detected; Mit, mitochondrial.
Ald4p in the absence of the mitochondrial NADH kinase reaction. The slight H 2 O 2 s phenotype of pos5ald5 carrying 62MTSϩyfjB also demonstrated that Ald5p partially contributes to the supply of mitochondrial NADPH. Simultaneously, the H 2 O 2 r phenotypes of pos5ald4 carrying POS5 and 62MTSϩUTR1 demonstrated that the mitochondrial NADH kinase reaction is indispensable in the absence of Ald4p, i.e. that the mitochondrial NADH kinase reaction is physiologically significant.
NAD Kinase and NADH Kinase Activity of Purified Pos5p (Pos5⌬MTSp)-Despite the significance of Pos5p, its biochemical properties have not been definitively determined (6,9). By removing MTS, soluble Pos5p (Pos5⌬MTSp) was expressed in E. coli, as reported previously (9), and purified (Fig. 6A). The purified Pos5p exhibited considerably higher NADH kinase than NAD kinase activity within a range of pH 5 to pH 11 (Fig.  6B). The optimum pH of the activities of NAD kinase and NADH kinase were pH 8.0 and pH 9.5, respectively. The kinetic values of Pos5p were determined and compared with those of Utr1p, Yef1p, and NADK3 from A. thaliana (Tables 2 and 3). NADK3 is also an NAD kinase exhibiting high NADH kinase activity (27). Comparison of the values of Pos5p with those of Utr1p and Yef1p clearly indicated that Pos5p exhibits a preference for NADH over NAD ϩ , whereas Utr1p exhibits a preference for NAD ϩ over NADH, and also that the activities of Yef1p toward both NAD ϩ and NADH were low. The high K m (5.3 mM) for the NAD ϩ of Pos5p could explain the low NAD kinase activity (shown in Fig. 6B), which was assayed in the presence of 2.0 mM NAD ϩ . It should be noted that NAD kinase activity in the mitochondrial fraction from the utr1yef1pos5 triple mutant carrying POS5 was assayed in the presence of 5.0 mM NAD ϩ (Fig. 4A). In addition, Pos5p exhibited a kinetic behavior that was different to that of NADK3 (Tables 2 and 3). In particular, the K m of NADK3 for NADH is lower than that of Pos5p.
BLAST analysis using the primary structure of Pos5p indicated that Pos5p shows highest homology to the NAD kinase homologs from fungi, plants, and other higher eukaryotes. Pos5p is not, however, homologous to Arabidopsis NADK3, which is only approximately the 750th homologous protein. Using the Pos5p-homologous proteins (50 proteins) plus NADK3, a phylogenic tree was constructed (Fig. 7). The tree indicates that Pos5p belongs to the fungal NAD kinase homolog group, which is distinguishable from another fungal group containing Utr1p and Yef1p. The proteins, which were predicted (using TargetP) to reside in the mitochondria, were concentrated in the group to which Pos5p belongs. The tree suggests that the proteins in the Pos5p fungal group are mitochondrial NAD kinases displaying high NADH kinase activity. NADK3 is located distant from Pos5p in the tree.

DISCUSSION
This study has demonstrated that the mitochondrial NAD kinase reaction followed by the NADP ϩ -dependent acetaldehyde dehydrogenase reaction catalyzed by Ald4/Ald5p, and in particular Ald4p, serves as a source of mitochondrial NADPH in S. cerevisiae. Although the NAD kinase activity of Pos5p was demonstrated to be lower than its NADH kinase activity (Tables 2 and 3 and Fig. 4), we assume that the former should be functional in mitochondria. It has been proposed that, when human cells are exposed to oxidative challenges, they rely on the existing NADP ϩ . Mutants carrying pRS415 grew as well as the WT carrying pRS415 on SD solid media containing Arg (data not shown). In the single mutants, each gene is replaced by kanMX4. In the double mutants, POS5 is replaced by HIS3MX6 and each dehydrogenase gene by kanMX4.  Values were determined by varying the concentrations of NAD ϩ or NADH in the presence of ATP at 5.0 mM (Pos5p, Utr1p, and Yef1p) and 4.0 mM (NADK3), except for the V max of Utr1p and Yef1p, which was the specific activity assayed in the presence of 1.0 mM ATP and 1.0 mM NAD ϩ or 1.0 mM NADH (7). pool and enhance the capacity to retain it in a reduced state by increasing the activity of NADP ϩ -dependent dehydrogenase (32). A single S. cerevisiae S288C cell is estimated to contain 22,200 molecules of Ald4p and 23,300 molecules of Ald5p, a substantial excess compared with the 4,650 molecules of Pos5p (33). Furthermore, a published microarray data base found in the Saccharomyces Genome Database reveals that ALD4 is up-regulated by H 2 O 2 treatment, whereas MAE1 is strongly down-regulated, ALD5 is slightly down-regulated, and IDP1 is slightly up-regulated. The different responses of ALD4 and ALD5 to H 2 O 2 may explain the reason why ALD4 is more important than ALD5 (Fig. 5C). We assume that the two successive reactions (the NAD kinase reaction followed by the acetaldehyde dehydrogenase reaction) can produce NADPH even when the amount of mitochondrial NADH is insufficient, e.g. when cell growth is solely dependant on fermentation,andevenwhentheNADP ϩ poolislimitedduetothelowerNAD kinase activity of Pos5p. Moreover, taken together with the significance of Ald6p as a cytosolic NADPH source (3), the present results confirm the importance of the acetaldehyde dehydrogenase reaction as a source of both cytosolic and mitochondrial NADPH.
In the case of mammals and humans, NAD kinase is located in the cytosol, not the mitochondria, and human NAD kinase