Genetic Analysis of Glutathione Peroxidase in Oxidative Stress Response of Saccharomyces cerevisiae *

Three glutathione peroxidase homologs (YKL026C, YBR244W, andYIR037W/HYR1) were found in theSaccharomyces Genome Database. We named themGPX1, GPX2, and GPX3, respectively, and we investigated the function of each gene product. Thegpx3Δ mutant was hypersensitive to peroxides, whereas null mutants of the GPX1 and GPX2 did not show any obvious phenotypes. Glutathione peroxidase activity decreased approximately 57 and 93% in the gpx3Δ andgpx1Δ/gpx2Δ/gpx3Δ mutants, respectively, compared with that of wild type. Expression of theGPX3 gene was not induced by any stresses tested, whereas that of the GPX1 gene was induced by glucose starvation. The GPX2 gene expression was induced by oxidative stress, which was dependent upon the Yap1p. The TSA1(thiol-specific antioxidant) gene encodes thioredoxin peroxidase that can reduce peroxides by using thioredoxin as a reducing power. Disruption of the TSA1 gene enhanced the basal expression level of the Yap1p target genes such as GSH1,GLR1, and GPX2 and that resulted in increases of total glutathione level and activities of glutathione reductase and glutathione peroxidase. However, expression of the TSA1gene did not increase in thegpx1Δ/gpx2Δ/gpx3Δ mutant. Therefore, de novo synthesis and recycling of glutathione were increased in the tsa1Δ mutant to maintain the catalytic cycle of glutathione peroxidase reaction efficiently as a backup system for thioredoxin peroxidase.

occurrence of the LOOHs in biological membranes may be one of the major oxidative damages to the cells.
Because reactive oxygen species are commonplace in aerobic organisms, they have enzymatic as well as non-enzymatic defense systems. For 1 Ascorbate can also work as a reductant for ascorbate peroxidase in plants (2). ␤-Carotene and tocopherol function as radical scavengers. Glutathione is also a major antioxidant in aerobic cells. However, it has been widely believed that microorganisms do not have peroxidases whose electron donor is glutathione. Microorganisms are believed to use cytochrome c as an electron donor for the peroxidase reaction (cytochrome c peroxidase). GPx has been thought to be evolutionarily acquired by mammals. However, we have demonstrated that yeasts have GPx and that glutathione plays a crucial role in the defense line against reactive oxygen species. For example, we have previously shown that catalase-deficient (ctt1⌬/cta1⌬) mutant of Saccharomyces cerevisiae showed almost the same sensitivity to H 2 O 2 compared with that of wild type, although the gsh1-deficient mutant was hypersensitive to H 2 O 2 and could not show an adaptive response to oxidative stress (3,4). The GSH1 gene encodes ␥-glutamylcysteine synthetase which is a rate-limiting enzyme for glutathione biosynthesis. Additionally, we purified GPx from the yeast Hansenula mrakii (5). The GPx of H. mrakii was found in both the cytoplasmic membrane and the inner membrane of mitochondria, which is an organelle where a large amount of reactive oxygen species are generated during oxygen respiration (6). We isolated H. mrakii as a LOOH-resistant yeast by screening (7), and the LOOH-sensitive mutants derived from H. mrakii could not induce GPx under the oxidative stress conditions (8). From these backgrounds, we have been persuaded that yeasts also must have the functional GPx (9). We searched the Saccharomyces Genome Data base for mammalian GPx homologs, and we found three hypothetical open reading frames (YKL026C, YBR244W, and YIR037W). One of them, YIR037W, has been referred to the HYR1 (hydrogen peroxide resistance) gene in the data base (GenBank TM accession number U22446), although its function has not yet been identified. Here we name them GPX1, GPX2, and GPX3, respectively, and disrupt each gene to investigate the functions through analysis of their phenotypes with respect to the oxidative stress response.
The TSA1 gene has been cloned as a thiol-specific antioxidant, which protects glutamine synthetase from oxidative in-activation in S. cerevisiae (10). Later, Tsa1p was found to have peroxidase activity in coupling with thioredoxin as a reducing power, i.e. thioredoxin peroxidase (TPx) (11)(12)(13). TPxs have been discovered widely from various types of cells ranging from those of yeasts to mammals, and they constitute a large peroxiredoxin family (14,15). TPx can reduce H 2 O 2 and LOOH in the presence of thioredoxin in reduced form, and oxidized form of thioredoxin is reduced by thioredoxin reductase with NADPH ( Fig. 1). This catalytic cycle is similar to that of GPx, i.e. GPx reduces H 2 O 2 and LOOH in the presence of reduced glutathione, and oxidized glutathione (glutathione disulfide) is reduced by glutathione reductase by using NADPH as a reducing power. In both reaction cycles, NADPH is oxidized to NADP ϩ , and NADP ϩ thus formed is reduced to NADPH by an action of glucose-6-phosphate dehydrogenase (Fig. 1). In this paper, we also investigate the correlation between the GPx system and the TPx system in oxidative stress response, and we demonstrate here that GPx is partially functioning as a backup system for TPx in S. cerevisiae.

MATERIALS AND METHODS
Strains-S. cerevisiae YPH250 (MATa trp1-⌬1 his3-⌬200 leu2-⌬1 lys2-801 ade2-101 ura3-52) was obtained from the Yeast Genetic Stock Center, University of California, Berkeley, and was used as a wild type strain. All disruptants in this study were constructed on the basis of the strain YPH250.
Construction of Disruptants-The GPX1 gene (YKL026C) was cloned by PCR using the primers as follows: GPHF-1, 5Ј-AGTTAATGTTGGA-TCCGCCCGGGGAGTTAA-3Ј, and GPHR-1, 5Ј-CGCCCTGAAGGTCG-ACCGTGTAAGAATCCT-3Ј. The GPHF-1 and GPHR-1 were designed to contain the BamHI site and SalI site (underlined), respectively. The PCR product containing the GPX1 gene (1924 bp) was treated with BamHI and Klenow and cloned to pUC19 which was digested by EcoRI and SphI and then treated by Klenow to yield pUGPH6. The plasmid (pUGPH6) was digested with EcoRI and SphI, treated with Klenow, and then the HIS3 gene was inserted to yield pGPX1⌬His3. The resultant plasmid was digested by SmaI (which was designed to contain in the PCR primer GPHF-1, italicized) and XbaI, and the gpx1⌬::HIS3 cassette was used to disrupt the GPX1 gene. The disruption allele was designated gpx1-⌬1::HIS3.
The GPX2 gene (YBR244W) was amplified by the following two primers: GPIF-1, 5Ј-GTTACCGTTCCCCGGGTTGGTCGACTTGAT-3Ј, and GPIR-1, 5Ј-GATCAAGCGTGTCGACATGCAACAAGAGGC-3Ј. Both primers were designed to have the SalI site (underlined). The PCR product containing the PGX2 gene (1873 bp) was digested with SalI, treated by Klenow, and then cloned to pUC19 which was treated by XbaI, HincII, and Klenow to give pUGI4. The plasmid (pUGI4) was digested with XbaI and StyI followed by treatment with Klenow, and the URA3 gene was inserted to give pGPX2⌬Ura3. The resultant plasmid was digested with BamHI and SphI, and the gpx2⌬::URA3 cassette was used to disrupt the GPX2 gene. The disruption allele was designated gpx2-⌬1::URA3.
The TSA1 gene was amplified by PCR using the following primers: TSA1F, 5Ј-AAATCGAGAGACCACGGGATTCGACACTCT-3Ј, and TSA1R, 5Ј-GGACGATTGATACTGGTGTTTAGCGGTGGT-3Ј. The PCR product containing TSA1 gene (2764 bp) was digested with EcoRI and PstI and then cloned to EcoRI/PstI site of pUC19 (pUTSA1). The pUTSA1 was digested with SalI and ApaI followed by treatment with Klenow, and the TRP1 gene was cloned to give pTSA1⌬Trp1. The resultant plasmid was digested with EcoRI and PstI, and the tsa1⌬::TRP1 cassette was used to disrupt the TSA1 gene. The disruption allele was designated tsa1-⌬1::TRP1.
Spot Assay-Cells were cultured in YPD medium until A 610 reached 0.5 (approximately 5 ϫ 10 6 cells/ml) and were diluted with sterilized 0.85% NaCl solution. Ten microliters of each sample was spotted onto YPD agar plate containing 4.0 mM H 2 O 2 or 1.5 mM t-BHP, and incubated at 28°C for 2 days.
Assay for Glutathione Peroxidase Activity-Cells were cultured in YPD medium until A 610 reached approximately 1.0, collected by centrifugation, and washed with 0.85% NaCl solution. Cells were resuspended in 10 mM potassium phosphate buffer (pH 7.0) and disrupted with glass beads. Cell homogenates were centrifuged at 14,000 rpm at 4°C for 10 min, and the resultant supernatants were used as cell extracts. GPx activity was measured in a reaction mixture (1.0 ml) in 50 mM potassium phosphate buffer (pH 7.0) containing 5.0 mM glutathione, 1.0 mM t-BHP, 0.16 mM NADPH, 1.0 mM NaN 3 , 0.24 unit/ml glutathione reductase (Oriental Yeast Co., Kyoto, Japan), and cell extracts. Before the addition of t-BHP, the reaction mixture was kept at 25°C for 5 min, and the reaction was started by the addition of t-BHP. Decrease of absorbance at 340 nm (A 340 -sample) was measured for 1 min. Three different blanks were taken for each assay for GPx activity. 1) To eliminate glutathione-independent NADPH-oxidizing activity, glutathione was omitted from the complete reaction mixture (A 340 -GSH). 2) To eliminate peroxide-independent NADPH-oxidizing activity, Me 2 SO (dimethyl sulfoxide) which was used to dilute t-BHP was added to the reaction mixture instead of t-BHP (A 340 -Me 2 SO). 3) To eliminate non-enzymatic NADPH consuming activity, cell extracts were omitted from the complete reaction mixture (A 340 -EXTR). To obtain the GPx activity (A 340 -GPx) in cell extracts, the following equation was used: A 340 -GPx ϭ (A 340 -sample) Ϫ ((A 340 -GSH) ϩ (A 340 -Me 2 SO) ϩ (A 340 -EXTR)). One unit of the activity was defined as the amount of enzyme oxidizing 1 mol of glutathione per min at 25°C. Millimolar absorption coefficient 6.22 mM Ϫ1 cm Ϫ1 for NADPH was used. Protein was determined by the method of Bradford (18).
Northern Blot Analysis-To see the effect of several environmental stresses on expression of the GPX genes, cells of S. cerevisiae YPH250 were cultured in YPD medium until A 610 reached approximately 1.0, and then concentrated solutions of various agents were added. In the case of osmotic stress experiment, solid NaCl was added to the culture. For heat shock experiment, the culture was transferred immediately to an incubator preheated at 37°C, and then the culture was continued. For glucose-starvation experiment, the cells were collected by centrifugation at 28°C, washed once with sterilized 0.85% NaCl solution, resuspended in fresh YPD medium without glucose (YP medium), and incubated at 28°C. After 30 min incubation under each stressed condition, total RNA was prepared according to the method of Schmitt et al. (19). Each GPX gene was labeled by [␣-32 P]dCTP, using a kit (Takara Co., Kyoto, Japan), and used as a probe. To see the effect of Yap1p on expression of the GPX gene, cells were cultured in SD minimum medium with appropriate amino acids and bases until A 610 reached ap-FIG. 1. Catalytic cycle of GPx and TPx reaction. Glutathione is synthesized by a sequential reaction of ␥-glutamylcysteine synthetase (GSH I) and glutathione synthetase (GSH II). GR, glutathione reductase; G6PDH, glucose-6-phosphate dehydrogenase; TR, thioredoxin reductase; GSH, glutathione (reduced form); GSSG, glutathione disulfide; Trx, thioredoxin; ROOH, peroxides. proximately 1.0, and the total RNA was prepared as described above.
Western Blot Analysis-Cells were cultured in YPD medium to log phase, and cell extracts were prepared as described above. Cellular proteins (100 g) were separated by SDS-polyacrylamide gel electrophoresis and were electrically transferred to polyvinylidene difluoride membrane (Immobilon; Millipore, Bedford, MA). Anti-Yap1p antiserum raised in rabbit was used as the primary antibody, and anti-rabbit IgG antibody conjugated with horseradish peroxidase (New England Biolabs, Inc., Beverly, MA) was used as the secondary antibody. Diaminobenzidine was used to visualize immunoreactive protein.
Measurement of Intracellular Oxidation Level-Intracellular oxidation level of yeast was measured using the oxidant-sensitive probe 2Ј,7Ј-dichlorofluorescin diacetate (Molecular Probes, Eugene, OR) (20). Cells growing exponentially in SD minimum medium were collected and resuspended in 10 mM potassium phosphate buffer (pH 7.0) containing 0.01 mM 2Ј,7Ј-dichlorofluorescin diacetate. After incubation at 28°C for 15 min, cells were washed, resuspended in distilled water, and disrupted with glass beads by vortex mixer for 3 min. Cell extracts (50 l) were mixed in 500 l of distilled water, and fluorescence was measured with EX ϭ 490 nm and EM ϭ 524 nm using a Hitachi F-3000 spectrofluorometer. The value of EM ϭ 524 nm was normalized by protein in the mixture.
Construction of GSH1-lacZ Fusion Gene and Assay of ␤-Galactosidase-The GSH1 promoter fragment (Ϫ800 to ϩ33, ϩ1 represents the adenine of translational initiation codon) containing the YRE was amplified by PCR using the primers 5Ј-TAATCTTATGAATCCCGGGGAT-TTTATCGG-3Ј and 5Ј-CTAGACTCAAACCCGGGCAAAGGCGTGCCC-3Ј. Both primers were designed to contain the SmaI site (underlined), and the DNA fragment amplified by PCR was digested with SmaI and cloned into the SmaI site of pMC1871 containing the coding region of the lacZ (21), to yield pMC-GSH1lacZ. As a result of this construction, the first 11 amino acid residues of Gsh1p were fused to ␤-galactosidase whose first 8 amino acids were deleted. The GSH1-lacZ fragment in the pMC-GSH1lacZ was isolated by digestion with SalI and cloned into the SalI site of YCp50 (URA3 marker) to yield pCGSH1-lacZ. The transformants carrying this plasmid were cultured in SD minimum medium supplemented with appropriate amino acids and bases without uracil at 28°C for 16 -20 h. A portion of the culture was transferred to a 200-ml flask containing 50 ml of YPD medium and cultured at 28°C with reciprocal shaking until A 610 reached approximately 1.0. In the case of stress experiments, H 2 O 2 was added to final concentration of 0.2 mM, and culture was continued for another 60 min. Cell extracts were prepared as described above, and ␤-galactosidase activity was measured as described by Miller (22). One unit of the activity was defined as the amount of enzyme increasing A 420 ϫ 1000 per min at 30°C.

S. cerevisiae Has Three GPx Homologs-We searched the
Saccharomyces Genome Data base for the homologs of mammalian GPx, and we found three candidates that share high sequence homology, i.e. YKL026C, YBR244W, and YIR037W.
We named them GPX1, GPX2, and GPX3, respectively. Sequence data of the YIR037W was deposited to the GenBank TM by Budde and Stahl in 1995 2 and was referred to as the HYR1 (hydrogen peroxide resistance) gene (GenBank TM accession number U22446), although functional analysis of the HYR1/ GPX3 gene product has not yet been reported thereafter. Fig. 2 shows the alignment of the amino acid sequence deduced from the DNA sequence of each GPx homolog gene from S. cerevisiae and that of human GPx-I (25). It has been known that mammalian GPxs have a selenocysteine in their active site, and the codon corresponding to the selenocysteine is TGA which is usually used as a stop codon (25). On the other hand, in the case of S. cerevisiae GPx homologs, cysteine was used instead of selenocysteine, and the cysteine residue was conserved in these three GPx homologs at the position of selenocysteine in human GPx-I. The amino acid sequence around the active site was conserved between mammalian GPx and yeast homologs, although identity in other regions was not so high compared with the region around the active site. Overall homology (identity) between mammalian GPx and yeast GPx homologs was approximately 36%. On the other hand, identity among yeast homologs was higher compared with mammalian GPx, and the values are 53% between Gpx1p and Gpx2p, 58% between Gpx1p and Gpx3p, and 74% between Gpx2p and Gpx3p.
The molecular weight of a subunit of GPxs from mammalian sources is approximately 22 kDa, and the enzymes consist of a homotetramer (26). The molecular weights of Gpx1p, Gpx2p, and Gpx3p in S. cerevisiae were calculated to be 19,483.68, 18,405.25, and 18,640.61, respectively, which was somehow smaller than those of mammalian enzymes. We have previously purified GPx from a yeast H. mrakii (5). The molecular mass of the enzyme was 28 kDa by SDS-polyacrylamide gel electrophoresis. Because the GPx of H. mrakii was bound to the cell membrane and detergents were used to solubilize it, we could not determine the molecular weight of the native enzyme by a gel filtration. Thus, the subunit structure of the yeast enzyme has not yet been identified.

FIG. 2. Alignment of the amino acid sequence of GPx homologs and human GPx.
Amino acid residues conserved in all species are shown by white letters against a black background, and conserved in two or three species are shown by white letters against gray background. Selenocysteine in human GPx-I is shown by the $ and is indicated by asterisk.
LOOH, and both are substrate for GPxs from mammals. Disruption of the GPX1 or GPX2 alone or double disruption of them did not affect the sensitivity to peroxides, although disruption of the GPX3 gene in these genetic backgrounds enhanced the susceptibility.
To assess whether these GPx homolog genes actually correspond to glutathione peroxidase, the enzyme activity was measured. So far, there are several reports describing the GPxlike activity in S. cerevisiae, if the yeasts were cultured in the presence of Cu 2ϩ or selenite (27,28). However, the GPx-like activity in such cases does not accurately reflect the GPx activity itself; contamination in the cell extracts with Cu 2ϩ and selenite, both of which are redox-active metals, can mimic the GPx activity. It has been well known that the selenium-containing compound, 2-phenyl-1,2-benzselenazol-3(2H)-one (so called ebselen), can display GPx-like activity (29,30). To measure the GPx activity correctly, we cultured the cells without the addition of any metals, and we carefully took three different blanks for each assay as described under "Materials and Methods." As shown in Fig. 4, the enzyme activity decreased approximately 57% in the gpx3⌬ mutant, and it was only 7% in the gpx1⌬/gpx2⌬/gpx3⌬ mutant compared with that of wild type strain. Together with the results of spot assay experiments (Fig. 3), the GPX3 gene product is thought to be the major GPx that scavenges peroxides in S. cerevisiae.
Stress Response of GPX Genes-The expression pattern of each GPX gene under several stress conditions was examined by Northern blot analysis. As shown in Fig. 5, the basal expression level of the GPX3 gene was constitutively high compared with those of other two GPX genes, although its expression was not induced by any environmental stresses tested. The expression level of the GPX1 gene was induced under the glucosestarved conditions. Menadione and 9,10-naphthoquinone, both are O 2 . -generating agents, and heat shock also slightly induced the expression of GPX1. Expression of the GPX2 gene was strongly induced by several oxidative stresses such as t-BHP, cumene hydroperoxide, and O 2 . -generating agents. Consequently the GPx activity was increased when the cells were exposed to oxidative stress (see Fig. 12B).
Yap1p is a critical transcription factor in the oxidative stress response in S. cerevisiae (31,32). Yap1p recognizes and binds to the specific DNA sequence termed YRE (Yap1p response element; optimum consensus sequence, 5Ј-TTA(C/G)TAA-3Ј) (33). The GPX2 gene has three YRE sequences in its 5Ј-upstream region (620-bp upstream from the translational initiation ATG codon, TTAGTAA; 410-bp upstream, TTACTAA; and 253-bp upstream, TTAGTAA). To assess whether expression of the GPX2 gene is regulated by Yap1p, basal expression level of the GPX2 gene was measured in the yap1⌬ mutant and in the YAP1-overexpressing strain. As shown in Fig. 6A, basal expression level of the GPX2 gene decreased in the yap1⌬ mutant, whereas it increased in the strain overexpressing the YAP1. Furthermore, no oxidative stress-induced expression of the GPX2 gene was observed in the yap1⌬ mutant (Fig. 6B). These results indicate that expression of the GPX2 gene is under the control of Yap1p. Expression of the GPX1 gene was slightly  6. Oxidative stress-induced expression of the GPX2 gene is controlled by Yap1p. A, total RNA was prepared from each strain cultured in SD minimum medium to log phase. YEp-YAP1 was constructed on the basis of a multicopy vector YEp13 (17). WT, wild type. B, the yap1⌬ mutant was cultured in YPD medium to log phase and treated with 0.2 mM H 2 O 2 or 0.1 mM t-BHP for 30 min.  (Fig. 6A). Additionally, no YRE-like sequence was found in the GPX1 promoter region; therefore, Yap1p is not likely to be involved in the regulatory mechanism of the GPX1 gene. Similarly, the expression level of the GPX3 gene did not change in the presence or absence of Yap1p in the cell (Fig. 6A). GPx Activity Increases in tsa1⌬ Mutant-The TSA1 gene (thiol-specific antioxidant) encodes a peroxidase whose electron donor is thioredoxin (TPx, thioredoxin peroxidase) (11)(12)(13). In order to analyze the correlation between GPx and TPx in the oxidative stress response in yeast, the TSA1 gene was disrupted and sensitivity to the peroxides was investigated. As shown in Fig. 7A, the tsa1⌬ mutant was sensitive to H 2 O 2 but not to t-BHP. Combination of tsa1⌬ mutation with gpx1⌬/ gpx2⌬/gpx3⌬ mutation was not synthetically lethal, although such a quadruplex mutant was sensitive to both H 2 O 2 and t-BHP. Fig. 7B shows the growth curves of tsa1⌬ and gpx1⌬/ gpx2⌬/gpx3⌬ mutants in SD minimum medium without any oxidants. The tsa1⌬ mutant showed slow growth, and such a phenotype was enhanced in the tsa1⌬/gpx1⌬/gpx2⌬/gpx3⌬ quadruplex mutant. This was also observed in the spot assay experiments (Fig. 7A).
To see the effect of disruption of the TSA1 gene on the GPx activity, the enzyme activity was measured. As shown in Fig.  8A, GPx activity increased approximately 3-fold in the tsa1⌬ mutant compared with that of wild type. Northern blot analysis was done to clarify which GPX gene expression was induced. The basal expression level of GPX2 was increased in the tsa1⌬ mutant, whereas those of other two GPX genes were not affected (Fig. 9A). Therefore, increase of the GPx activity in the tsa1⌬ mutant was found to be due to the increased level of GPX2 gene expression. We then examined whether TSA1 gene expression is increased in the gpx1⌬/gpx2⌬/gpx3⌬ mutant. As shown in Fig. 9B, disruption of the GPX gene did not affect the basal expression level of the TSA1 gene.
In order for the catalytic cycle of the GPx reaction to proceed, a supplement of reduced glutathione is required (Fig. 1). We measured the activity of glutathione reductase, which is a major enzyme catalyzing the reduction of glutathione disulfide to reduced glutathione. The enzyme activity was increased in the tsa1⌬ mutant (Fig. 8B). We also measured the glutathione synthesizing activity by using the GSH1-lacZ reporter gene. The GSH1 gene product (␥-glutamylcysteine synthetase) is a rate-limiting enzyme for the de novo synthesis of glutathione in S. cerevisiae (34). As shown in Fig. 8C, the basal expression level of the GSH1-lacZ reporter gene increased in the tsa1⌬ mutant. We also confirmed the increase of the GSH1 mRNA level by Northern blot analysis (data not shown). Consequently total glutathione level in the tsa1⌬ mutant was increased (wild type, 1.48 Ϯ 0.09 mol/g wet cell; tsa1⌬, 1.71 Ϯ 0.07 mol/g wet cell). These results suggest that GPx is partially functioning as a backup system for TPx, and de novo synthesis and recycling of glutathione were accelerated in the tsa1⌬ mutant to maintain the catalytic cycle of GPx reaction efficiently.
Effect of tsa1 Null Mutation on Expression of the Yap1p Target Genes-As shown in Figs. 8 and 9, disruption of the TSA1 gene affected the de novo synthesis and recycling of glutathione as well as expression of the GPX2 gene. Glutathione reductase is encoded by the GLR1 gene, and its expression is dependent upon Yap1p (35). The GSH1 gene is also a target gene for Yap1p (36), and here we demonstrated that the GPX2 gene is one of its targets (Fig. 6). These observations imply that disruption of the TSA1 gene affects the activity of Yap1p. It has been reported that Yap1p is distributed in both cytosol and nucleus if the cells are under the non-stressed conditions. Once the cells are exposed to oxidative stress, Yap1p is accumulated in the nucleus, and expression of its target genes is enhanced (23). The critical step in the oxidative stress-induced expres- FIG. 7. Effect of tsa1 null mutation on sensitivity to oxidative stress. Detailed conditions for experiments are described under "Materials and Methods." A, spot assay. B, growth curve. Each strain was cultured in SD minimum medium supplemented with appropriate amino acids and bases at 28°C with reciprocal shaking, and A 610 of the culture was monitored periodically. Numbers in the figure correspond to the slot number in A as follows: 1, wild type; 2, gpx1⌬/gpx2⌬/gpx3⌬; 3, tsa1⌬; and 4, tsa1⌬/gpx1⌬/gpx2⌬/gpx3⌬.

FIG. 8. Effect of tsa1 null mutation on glutathione metabolism.
A, glutathione peroxidase activity. B, glutathione reductase (GR) activity. One unit of the activity was defined as the amount of enzyme reducing 1 mol of glutathione disulfide per min at 25°C as described in our previous paper (61). C, ␤-galactosidase activity derived from the GSH1-lacZ reporter gene. Results indicate the average Ϯ S.D. of three independent experiments. The GPX2 gene expression was specifically increased in the tsa1⌬ mutant, because only the GPX2 gene was the target for the Yap1p among three GPX genes. The TRR1 gene is also the Yap1p target gene, and its basal expression level increased in the tsa1⌬ mutant. Disruption of the GPX genes did not affect the expressions of the TSA1 and TRR1. WT, wild type. sion of the Yap1p target genes is thought to be the nuclear localization of Yap1p (37), because neither Yap1p protein level nor DNA binding activity of Yap1p significantly increases during oxidative stress response (23,38,39). Regarding the increment of expression level and corresponding enzyme activity of the Yap1p target genes in the tsa1⌬ mutant, several possibilities can be considered. One possible explanation is that peroxides are accumulated in the cells by disruption of the TSA1 gene, causing endogenous oxidative stress, and consequently the distribution of Yap1p is changed to activate the transcription of its target genes. Indeed, the growth rate of the tsa1⌬ mutant is reduced compared with that of wild type cells even in the absence of oxidative stress (Fig. 7B). It might be presumed to be due to the accumulation of reactive oxygen species in the cells. To confirm this possibility, we measured intracellular oxidation level and surveyed the distribution of Yap1p in the cells by using GFP-Yap1p fusion protein. As shown in Fig. 10, however, intracellular oxidation levels in both the tsa1⌬ mutant and the gpx1⌬/gpx2⌬/gpx3⌬ mutant were not so much increased compared with that of wild type cell. Furthermore, predominant accumulation of GFP-Yap1p in the nucleus was not observed in the tsa1⌬ mutant as far as we investigated (Fig.  11). No distinct difference in the amount of Yap1p was observed between the wild type and tsa1⌬ mutant by Western blot analysis (data not shown). The basal expression level of the TRR1 gene, which encodes thioredoxin reductase and is a target for the Yap1p (12,40), was increased in the tsa1⌬ mutant (Fig. 9A) but not in the gpx1⌬/gpx2⌬/gpx3⌬ mutant (Fig. 9B). The mRNA levels of TRX2 (encoding thioredoxin) (41) and YCF1 (encoding ABC transporter on the vacuolar membrane) (42), which are also the targets for the Yap1p, are increased in the tsa1⌬ mutant (data not shown). Therefore, function of the Yap1p seems to be activated in the tsa1⌬ mutant; however, it is not likely to be dependent upon the predominant nuclear localization of Yap1p itself.
We next investigated whether expression of the Yap1p target gene is further increased or not in the tsa1⌬ mutant if the cells are exposed to oxidative stress. As shown in Fig. 12A, ␤-galactosidase activity derived from the GSH1-lacZ reporter gene was increased by the oxidative stress in the tsa1⌬ mutant as well as in the wild type cell. The GPx activity was also further increased in the tsa1⌬ mutant by oxidative stress (Fig. 12B). Therefore, the increase of the basal expression levels of the Yap1p target genes in the tsa1⌬ mutant is not saturated; i.e. the genes still remain to be regulated under the oxidative stress conditions.
We then constructed the tsa1⌬/yap1⌬ mutant to obtain direct evidence whether the increase of the basal expression levels of Yap1p target genes is regulated by Yap1p or not. Because GSH1 is a target for Yap1p, the basal expression level of the GSH1-lacZ reporter gene decreased in the yap1⌬ mutant; however, it did not increase in the tsa1⌬/yap1⌬ mutant compared with that in the yap1⌬ mutant (Fig. 12A). Additionally, ␤-galactosidase activity derived from the GSH1-lacZ reporter gene was no longer increased by treatment of the tsa1⌬/yap1⌬ mutant with H 2 O 2 . Similar results were obtained in the case of GPx activity (Fig. 12B). Therefore, we concluded that increase of the basal expression levels of the Yap1p target genes in the tsa1⌬ mutant was dependent upon the Yap1p, even though it was not predominantly accumulated in the nucleus (Fig. 11).

DISCUSSION
Yeast Has Functional GPxs-We have been studying the oxidative stress response in yeast and demonstrating that glutathione plays a crucial role in the resistance and acquisition of adaptation to oxidative stress in yeast (3). Nevertheless, it has been widely believed that microorganisms do not have GPx. In contrast, we purified and characterized GPx from the yeast H. mrakii (5,6,9). We then adopted the genetic approach to identify whether three GPx homologs found in the Saccharomyces Genome Data base actually correspond to glutathione peroxidase in S. cerevisiae.
Mammalian GPxs have been known to contain selenocys- teine in their active site (25), although this was not the case for the yeast homologs. Incorporation of selenocysteine into the polypeptide chain is a cotranslational event. The codon corresponding to selenocysteine is TGA which is usually used as a stop codon. In the case of yeast GPxs, cysteine was used instead of selenocysteine. If the TGA codon had been present in the GPx homolog gene, such a sequence has not been considered as the open reading frame by the genome sequence project. Many selenoproteins have been reported in both prokaryotes and eukaryotes (43,44). Many of them are redox enzymes, such as formate dehydrogenase in Escherichia coli (45), thioredoxin reductase (46), and GPx (25) in mammals, and one selenocysteine residue is incorporated at their active site. It has been proved that a stable stem-loop structure immediately 3Ј to the TGA codon is required for incorporation of selenocysteine into the polypeptide chain in E. coli (47,48). Requirement of a stable stem-loop structure is common between prokaryotes and eukaryotes, although it is located approximately 500 -2000 nucleotides apart from the TGA codon in the 3Ј-untranslated region in the mRNA of eukaryotes (49). We searched for a stem-loop structure within the open reading frame and in the 3Ј-untranslated region of each GPX gene, although no such a structure was found. Rocher et al. (50) reported that mammalian GPx whose selenocysteine was substituted to cysteine ([Cys]GPx) still possessed the enzyme activity, even though the specific activity was decreased. Judging from the amino acid sequence deduced from the DNA sequence of each GPX gene, yeast GPxs are thought to be a [Cys]GPx.
Previously we have purified GPx from the yeast H. mrakii (5). The GPx was mostly active toward alkyl hydroperoxides rather than H 2 O 2 . Although we do not have any biochemical data on the Gpx3p at the moment, the GPX3 gene product may be active toward both H 2 O 2 and LOOH because the gpx3⌬ mutant was hypersensitive to both of them (Fig. 3). The basal mRNA level of GPX3 was constitutively higher than those of other two GPX genes, and the gpx3⌬ mutant was hypersensitive to peroxides, whereas disruption of the GPX1 or GPX2 did not show any obvious phenotype with respect to the tolerance to oxidative stress; therefore, the GPX3 gene product may be a major GPx in S. cerevisiae. Nevertheless, less GPx activity was detected in the gpx1⌬/gpx2⌬/gpx3⌬ mutant than in the gpx3⌬ mutant (Fig. 4). Mammalian GPxs have been reported to consist of a homotetramer (26,51), although the subunit structure of the GPx from H. mrakii has not yet been identified (5). We cannot rule out a possibility that GPxs in S. cerevisiae may form a heterocomplex among these three molecular species in vivo.
Function of GPx in Oxidative Stress Response in Yeast-To obtain a clue to solve the differences of the roles of these three GPxs in yeast cells, expression of each GPX gene under several environmental stress conditions was investigated. Expression level of the GPX3 gene was constitutively higher compared with those of other two GPX genes, and it was not induced by any stresses as far as we tested (Fig. 5). Disruption of the GPX3 gene enhanced susceptibility to peroxides (Fig. 3); thus the Gpx3p is thought to be a major GPx functioning in scavenging peroxides generated during normal metabolism in S. cerevisiae. On the other hand, expression of the GPX2 gene was induced by oxidative stress in the Yap1p-dependent manner. Additionally, STRE (stress response element, 5Ј-AGGGG-3Ј or 5Ј-CCCCT-3Ј) was found in the GPX2 gene promoter (166-bp upstream from ATG codon, CCCCT). Some genes encoding heat shock protein (HSP) in S. cerevisiae, such as HSP104, HSP26 and HSP12, have STRE other than the heat shock element (52)(53)(54). Besides these HSP genes, many stress-induced genes also have been known to contain STRE in their promoter region, and two zinc-finger proteins, Msn2p and Msn4p, are involved in the transcriptional regulation of the genes (17,52,55). Expression of such genes is induced by a wide variety of stresses, such as osmotic stress and oxidative stress as well as heat shock, and these stress signals are thought to be focusing on the STRE (52). Expression of the GPX2 gene was mostly induced by oxidative stress, but other environmental stresses such as heat shock and osmotic stress did not significantly affect the expression of the GPX2 gene as much as the oxidative stress did (Fig. 5).
Expression of the GPX1 gene was induced if the cells were exposed to glucose-starved conditions (Fig. 5). Mig1p is involved in glucose repression (56 -58) and recognizes an AT-rich sequence followed by a G cluster (59). A potential Mig1p-binding site (5Ј-TTAAAGCCGGGG-3Ј) is located 182 bp upstream of the translational initiation codon of the GPX1 gene.
GPx Is Functioning as a Backup System for Thioredoxin Peroxidase-The TSA1 gene product has been known to be functioning as a TPx (Fig. 1), and TPx belongs to a large family of peroxiredoxin (15). TPx in S. cerevisiae has been thought to predominantly scavenge H 2 O 2 rather than alkyl peroxides (12). The tsa1⌬ mutant became hypersensitive to H 2 O 2 but not to t-BHP (Fig. 7A) which apparently supports such previous data. On the other hand, Jeong et al. (60) recently reported that purified Tsa1p (TPx) could reduce t-BHP (K m ϭ 10 M, V max ϭ 2.4 mol/min/mg) to almost the same efficiency as that for H 2 O 2 (K m ϭ 3 M, V max ϭ 4.8 mol/min/mg) in vitro. We demonstrated in this paper that basal expression level of the GPX2 gene increased in the tsa1⌬ mutant, which consequently resulted in increase of the GPx activity (Figs. 8 and 9A). GPx purified from H. mrakii was mostly active to alkyl hydroperoxides including phospholipid hydroperoxide and cholesterol 5␣hydroperoxide rather than H 2 O 2 (5). Therefore, the reason why the tsa1⌬ mutant did not become hypersensitive to t-BHP could be explained that the induction of GPx, which was due to the increase of GPX2 gene expression, partially suppressed the susceptibility to t-BHP.
In order to proceed with the catalytic cycle of GPx reaction, an ample supply of glutathione must be required. In the tsa1⌬ mutant, both the de novo synthesis and recycling of glutathione were enhanced (Fig. 8). Total glutathione level was also increased in the tsa1⌬ mutant. These metabolic changes seem to be reasonable for catalytic cycle of GPx reaction (Fig. 1). Taken together with induction of GPx in the tsa1⌬ mutant, GPx (especially Gpx2p) is working as a backup system for TPx in S. cerevisiae. We also measured catalase activity in the tsa1⌬, gpx1⌬/gpx2⌬/gpx3⌬, and tsa1⌬/gpx1⌬/gpx2⌬/gpx3⌬ mutants whether it can also serve as a backup system for these peroxidases, but no induction was observed (data not shown).
Function of Yap1p Is Activated in tsa1⌬ Mutant-An interesting phenotype of the tsa1⌬ mutant that should be noted is the increase of the basal expression levels of the Yap1p target genes (Figs. 8 and 9). Among three GPX genes, the GPX2 gene was only found to be regulated by Yap1p (Fig. 6), and therefore basal expression levels of it specifically increased in the tsa1⌬ mutant (Fig. 8). Recently, we found that the thioredoxin-deficient mutant (trx1⌬/trx2⌬ double mutant) caused constitutive localization of Yap1p in the nucleus that resulted in an increase of basal expression levels of the Yap1p target genes (62). In the thioredoxin null mutant, intracellular oxidation level increased approximately 2-fold; however, this was not the case for the tsa1⌬ mutant (Fig. 10). In the case of thioredoxin-deficient mutant, increased level of the GSH1-lacZ reporter gene expression was saturated; i.e. further induction was no longer observed if the trx1⌬/trx2⌬ mutant was exposed to oxidative stress (62). However, in the case of tsa1⌬ mutant, further oxidative stress-induced expression of the GSH1-lacZ reporter gene as well as increase of the GPx activity were observed (Fig.  12). It is clear that increase of the basal expression levels of Yap1p target genes in the tsa1⌬ mutant is dependent upon the Yap1p, because such an increase was not observed in the tsa1⌬/ yap1⌬ mutant (Fig. 12). From these observations, we speculate that properties of Yap1p, e.g. DNA binding activity to YRE or ability for transcriptional activation of its target genes, or both, seem to be enhanced in the tsa1⌬ mutant by an unknown mechanism besides alteration of its localization. Of course, several explanations can be possible, e.g. some factor(s) that can activate the function of Yap1p are negatively regulated by TPx (TSA1 gene product) or some factor(s) that repress the activity of Yap1p are activated by TPx.
The TPx reduces peroxides by using thioredoxin as a reducing power in vitro (Fig. 1); thus the TSA1 gene product is expected to interact directly with thioredoxin in vivo. Interestingly, we found that combination of null mutations of the Yap1p and thioredoxin (trx1⌬/trx2⌬) was synthetically lethal (62); however, the tsa1⌬/yap1⌬ mutant was viable. Therefore, even though deficiency of thioredoxin or TPx gives similar phenotype with respect to the basal expression levels of the Yap1p target genes, and both TPx and thioredoxin are involved in the same antioxidant system (Fig. 1), different regulatory mechanisms may underlie the control of the function of Yap1p.