GPX2, Encoding a Phospholipid Hydroperoxide Glutathione Peroxidase Homologue, Codes for an Atypical 2-Cys Peroxiredoxin in Saccharomyces cerevisiae*

We have previously reported that Saccharomyces cerevisiae has three glutathione peroxidase homologues (GPX1, GPX2, and GPX3) (Inoue, Y., Matsuda, T., Sugiyama, K., Izawa, S., and Kimura, A. (1999) J. Biol. Chem. 274, 27002–27009). Of these, the GPX2 gene product (Gpx2) shows the greatest similarity to phospholipid hydroperoxide glutathione peroxidase. Here we show that GPX2 encodes an atypical 2-Cys peroxiredoxin which uses thioredoxin as an electron donor. Gpx2 was essentially in a reduced form even in mutants defective in glutathione reductase or glutaredoxin under oxidative stressed conditions. On the other hand, Gpx2 was partially oxidized in a mutant defective in cytosolic thioredoxin (trx1Δtrx2Δ) under non-stressed conditions and completely oxidized in tert-butyl hydroperoxide-treated cells of trx1Δtrx2Δ and thioredoxin reductase-deficient mutant cells. Alanine scanning of cysteine residues of Gpx2 revealed that an intramolecular disulfide bond was formed between Cys37 and Cys83 in vivo. Gpx2 was purified to determine whether it functions as a peroxidase that uses thioredoxin as an electron donor in vitro. Gpx2 reduced H2O2 and tert-butyl hydroperoxide in the presence of thioredoxin, thioredoxin reductase, and NADPH (for H2O2, Km = 20 μm, kcat = 9.57 × 102 s-1; for tert-butyl hydroperoxide, Km = 62.5 μm, kcat = 3.68 × 102 s-1); however, it showed remarkably less activity toward these peroxides in the presence of glutathione, glutathione reductase, and NADPH. The sensitivity of yeast cells to tert-butyl hydroperoxide was found to be exacerbated by the co-existence of Ca2+, a tendency that was most obvious in gpx2Δ cells. Although the redox state of Gpx2 was not affected by Ca2+, the Gpx2 level was markedly increased in the presence of both tert-butyl hydroperoxide and Ca2+. Gpx2 is likely to play an important role in the protection of cells from oxidative stress in the presence of Ca2+.

Compared with anaerobic organisms, aerobic organisms produce energy (ATP) more efficiently because of the electron transfer system in mitochondria. At the final step in the electron transfer process, molecular oxygen (O 2 ) accepts four electrons and is reduced to H 2 O by cytochrome c oxidase. In the electron transfer system as well as normal metabolism in the cytoplasm, a portion of O 2 remains in an insufficient state in reduction, which in turn is referred to as reactive oxygen species. Organisms of all types possess antioxidant systems, and the diversity of the systems is of interest in terms of understanding the evolution of organisms. Superoxide dismutase catalyzes the disproportionation of superoxide anion radicals (O 2 . ) to O 2 2 is used by glutathione peroxidase (GPx) in mammals. We have cloned three GPx homologue genes (GPX1, GPX2, and GPX3) from the budding yeast Saccharomyces cerevisiae (1). Phylogenetic analysis of each gene product revealed that yeast GPxs are similar to mammalian phospholipid hydroperoxide glutathione peroxidase (PHGPx) (2). In mammalian GPxs, selenocysteine (SeCys) is conserved at the active site of the enzyme, whereas in such GPx homologues Cys is conserved corresponding to the position of SeCys in mammalian GPxs (1). In the GPx reaction, oxidized SeCys (SeCys-OH) is reduced by glutathione (GSH), and the glutathione disulfide (oxidized glutathione, GSSG) thus formed is reduced to GSH by glutathione reductase using NADPH as a reducing power. It has been reported that the conversion of SeCys to Cys in mammalian GPx reduces the enzyme activity drastically (3,4); therefore, the catalytic mechanism of such non-SeCys-type GPx homologues, including yeast GPxs, is of considerable interest. A series of peroxidases referred to as peroxiredoxins (Prxs) have been found from lower prokaryotes to higher eukaryotes, and from Archaea, and they constitute a large family (5). There are several common features of Prxs, i.e. these enzymes do not have redox cofactors such as metal and prosthetic group and contain conserved Cys residue(s). Depending on the number of Cys residues conserved, Prxs are basically divided into two groups, 1-Cys Prx and 2-Cys Prx. In both types of Prx, the N-terminal Cys is absolutely conserved, which is sometimes called the "peroxidatic" cysteine (Cys-S P H) (6). In the 1-Cys Prx reaction, Cys-S P H is oxidized to cysteine sulfenic acid (Cys-S P OH) through the reduction of H 2 O 2 to H 2 O, and the Cys-S P OH formed is reduced to Cys-S P H by an unknown reducing agent. In many 1-Cys Prxs, TRX serves as the endogenous reducing agent. By contrast to the 1-Cys Prx, the 2-Cys Prx has a second conserved Cys in the C terminus, which is sometimes called the "resolving" cysteine (Cys-S R H) (6). Because Cys-S P OH is unstable, it undergoes further oxidation to yield cysteine sulfinic acid (Cys-S P OOH), the formation of which turns 2-Cys Prx inactive as a peroxidase. Hence, sulfiredoxin recovers Cys-S P OOH to Cys-S P H.
Alternatively, the Cys-S P OH is attacked by the Cys-S R H of another monomer to form intermolecular disulfide bonds, and these disulfide bridges are reduced by TRX. Due to the reducing agent in the peroxidase reaction, Prx is also referred to as the thioredoxin peroxidase (TPx). S. cerevisiae has five Prx/TPx genes, i.e. TSA1, TSA2, AHP1, DOT5, and PRX1 (7). The first four genes code for 2-Cys Prx, and the last codes for 1-Cys Prx (6,8).
Delaunay et al. (9) have recently reported that Gpx3 also functions as a redox transducer of Yap1, a transcription factor critical for oxidative stress response in S. cerevisiae (10). They proposed that Cys 36 of Gpx3, which corresponds to the position of SeCys in mammalian GPx (1), is oxidized to Cys 36 -SOH by H 2 O 2 , and Cys 598 of Yap1, which is located in the C terminus cysteine-rich domain (c-CRD) overlapping the nuclear export signal of this b-ZIP transcription factor (11)(12)(13), attacks Cys 36 -SOH to form a mix disulfide between Gpx3 and Yap1. Subsequently, the intermolecular disulfide bridge is reduced by Cys 303 of Yap1, which is located in the n-CRD, and consequently, the intramolecular disulfide bond within Yap1 is formed, and Gpx3 is reverted to the reduced form. Additionally, if Yap1 is absent, the Cys 82 within the same monomer of Gpx3 attacks the Cys 36 -SOH to form an intramolecular disulfide bond, and this disulfide bond is reduced by TRX. As a result, a TRX-dependent peroxidase reaction is accomplished. This mode of disulfide bridge formation during a peroxidase reaction (intramolecular disulfide bond) is the atypical 2-Cys Prx; however, again, the Gpx3 of S. cerevisiae is a homologue of GPx but not Prx.
In the present study we demonstrate that Gpx2 is an atypical 2-Cys Prx. We show that the redox state of Gpx2 is maintained by a TRX-dependent system in vivo. We also show that the purified Gpx2 exhibits TRX-dependent peroxidase activity toward H 2 O 2 and t-BHP in the presence of thioredoxin reductase and NADPH in vitro but exhibits low peroxidase activity in the presence of GSH, glutathione reductase, and NADPH. The alanine scanning of cysteine residues of Gpx2 revealed that Cys 37 and Cys 83 seem to contribute to the formation of an intramolecular disulfide bond as a result of the peroxidase reaction. We also show that Gpx2 is important in the response to oxidative stress in the presence of calcium.
The construction of YEp13ϩTRX1, YEp13ϩTRX2, YEp13ϩGRX1, and YEp13ϩGRX2 was described previously (15). To clone the overexpression allele of GPX2, the PCR was done with the primers (GPX2-3 and GPIR-1) (1, 16) and the genomic DNA of wild-type cells. The PCR fragment was digested with SalI, and the resultant DNA fragment was cloned into the SalI site of pRS413 to yield pRSGPX2-3H.
Cys Modification with 4Ј-Acetamido-4Ј-maleimidylstilbene-2,2Ј-disulfonic acid (AMS)-Cells were cultured in SD minimal medium (2% glucose, 0.67% yeast nitrogen base w/o amino acids supplemented with the appropriate amino acids and bases) at 28°C. When the A 610 of the culture reached 1.0, peroxides (0.4 mM H 2 O 2 , 0.6 mM t-BHP, or 0.2 mM cumene hydroperoxide) were added. Cells were cultured for another hour at 28°C, and five A 610 units of the culture for the wild type, grx1⌬grx2⌬, glr1⌬, and trr2⌬ cells or two A 610 units of the culture for trx1⌬trx2⌬, trx1⌬trx2⌬trx3⌬, and trr1⌬ cells were collected by centrifugation. Cells were washed twice with a 20% trichloroacetic acid solution and once with the acetone. After removal of acetone, cells were treated with AMS (Molecular Probes) as described previously (17). Briefly, cells were suspended with 40 l of sample buffer (80 mM Tris-HCl (pH 6.8), 2% SDS, 6 M urea, 1 mM phenylmethylsulfonyl fluoride, and 0.05% bromphenol blue) containing 20 mM AMS. A small portion (ϳ4 l) of 1 M Tris-HCl (pH 8.0) was added to adjust the pH of the sample, and the mixture was boiled for 2 min. Twenty microliters of the sample was subjected to non-reducing SDS-PAGE followed by Western blotting using anti-Gpx2 antibody to determine the redox state of Gpx2.
Western Blotting-After SDS-PAGE, separated proteins were transferred to a polyvinylidene difluoride membrane (Immobilon, Millipore). Anti-Gpx2 antibody and anti-yeast TRX antibody raised in rabbits were described previously (15,18). Anti-rabbit IgG conjugated with horseradish peroxidase (Sigma) was used as the secondary antibody.
Thioredoxin Peroxidase Assay-The reaction mixture containing 100 mM Tris-HCl (pH 7.0), 0.3 mM NADPH, 4 M yeast thioredoxin II (Calbiochem), 2 M yeast thioredoxin reductase (Calbiochem), and 1 g of purified Gpx2 was incubated at room temperature for 1 min, and then 0.3 mM H 2 O 2 or 0.1 mM t-BHP was added (total volume, 120 l). One hundred microliters of the mixture was transferred to a micro cuvette, and the decrease in absorbance at 340 nm (A 340 ) was monitored for 10 min. For a blank, 10 mM Tris-HCl (pH 7.0) was used instead of the purified Gpx2. One unit of activity was defined as the amount of enzyme oxidizing 1 mol of NADPH/min using 6.22 mM Ϫ1 cm Ϫ1 as a millimolar extinction coefficient for NADPH.
Glutathione Peroxidase Assay-The reaction mixture containing 100 mM Tris-HCl (pH 7.0), 0.3 mM NADPH, 3 mM reduced glutathione, 8.3 units/ml yeast glutathione reductase (Oriental Yeast Co., Ltd.), and 1 g of purified Gpx2 was incubated at room temperature for 5 min, and then 0.3 mM H 2 O 2 or 0.8 mM t-BHP was added (total volume, 120 l). One hundred microliters of the mixture was transferred to a micro-cuvette, and the decrease in A 340 was monitored for 10 min. For a blank, 10 mM Tris-HCl (pH 7.0) was used instead of the purified Gpx2. One unit of activity was defined as the amount of enzyme oxidizing 1 mol of NADPH/min using 6.22 mM Ϫ1 cm Ϫ1 as a millimolar extinction coefficient for NADPH.
In Vitro Oxidation of Gpx2-In vitro oxidation of the purified Gpx2 was carried out essentially as described by Ursini et al. (19). Briefly, the purified Gpx2 was first reduced by incubating with 2-mercaptoethanol and passed through a gel filtration column. Hydrogen peroxide (final concentration, 0.1, 1.0, or 10 mM) was added to the enzyme solution, and the mixture was incubated at 37°C for 1 h. The sample was subjected to reducing and non-reducing SDS-PAGE, and the gels were stained for proteins with Coomassie Brilliant Blue.
Survival Test-Cells were cultured in YPD medium until the A 610 reached 0.5 and treated with 3 mM t-BHP, 200 mM CaCl 2 , or 3 mM t-BHP plus 200 mM CaCl 2 at 28°C. At the prescribed time, a portion of the cell suspension was withdrawn, diluted with a sterilized 0.85% NaCl solution, and spread onto YPD agar plates. Cells were cultured at 28°C for 2 days, and viable cells were counted.
Spot Assay-gpx3⌬ cells carrying pRS413 or pRSGPX2-3H (pRS413ϩGPX2-3), which carry the GPX2 overexpression allele, were cultured in SD minimal medium until the A 610 reached 0.5. Cells were diluted with a sterilized 0.85% NaCl solution and spotted on to SD agar plates containing H 2 O 2 or t-BHP.
Phylogenetic Analysis-The multiple alignment and phylogenetic analysis were done using DNASIS Version 3.5 (Hitachi Software Engineering).

Determination of Redox State of Gpx2 with AMS Modification Assay-
If the peroxidase reaction is carried out by Gpx2, the sulfhydryl group of the active site Cys would be initially oxidized to cysteine sulfenic acid. To accomplish the catalytic cycle of the peroxidase reaction in vivo, there are two pathways to regenerate the sulfhydryl group at the active site Cys, i.e. (i) directly via an endogenous reducing agent such as GSH or TRX, or (ii) by generating a disulfide bond with another sulfhydryl group intramolecularly or intermolecularly and, subsequently, resolving the disulfide bridge thus formed with a protein-disulfide oxidoreductase such as TRX or glutaredoxin. We determined the redox state of Gpx2 in yeast cells using the AMS modification assay. AMS modifies the sulfhydryl group of a protein irreversibly. Because the molecular weight of AMS is ϳ500, the apparent molecular weight of Gpx2 will increase by 500 with every modification of a Cys residue in Gpx2. Indeed, because Gpx2 has four Cys residues, the complete modification of all Cys residues of Gpx2 with AMS (which means Gpx2 is in the reduced form) results in an increase of 2000 in the apparent molecular weight.
To verify the validity of the AMS modification assay, we measured the mobility shift of purified Gpx2 after treatment with AMS in non-reducing SDS-PAGE. The molecular weight of Gpx2 deduced from the amino acid sequence is 18,406. As shown in Fig. 1A, the molecular mass of unmodified Gpx2 was estimated to be ϳ18 kDa. After treatment with AMS, the apparent molecular mass was estimated to be 20 kDa, suggesting that all Cys residues of Gpx2 were modified with AMS. We confirmed it by titration of remaining Cys residues with 5,5Ј-dithiobis(2-nitrobenzoic acid). Hence, the AMS-modified Gpx2, i.e. Gpx2 in reduced form (Gpx2 red ), migrated slowly in non-reducing SDS-PAGE.
Redox State of Gpx2 in Mutants Defective in Glutathione Recycling-In the GPx reaction GSH is used as an endogenous electron donor, and oxidized glutathione (GSSG) is reduced by glutathione reductase. Therefore, if the recycling of GSH is impaired, the catalytic cycle of GPx will be arrested. S. cerevisiae has a single glutathione reductase encoded by GLR1 (20), and therefore, the redox ratio of GSSG to total glutathione in glr1⌬ cells increased 3-fold compared with that of wild-type cells due to the impairment of the recycling of GSH (21). However, as shown in Fig. 1B, Gpx2 was essentially in the reduced form in wild-type as well as in glr1⌬ cells even under conditions of oxidative stress. The oxidized Gpx2 (Gpx2 ox ) in a trx1⌬trx2⌬ mutant was shown as a control (details are described in Fig. 1C and corresponding description). Glutaredoxin is one of the protein-disulfide oxidoreductases that catalyze the reduction of disulfide bonds of proteins using GSH as a reducing power. We determined the redox state of Gpx2 in a glutaredoxin-deficient mutant (grx1⌬grx2⌬) in the presence or absence of oxidative stress, although Gpx2 was virtually in the reduced form (Fig. 1B). These results suggest that the redox regulation of Gpx2 is not likely regulated by a GSH-dependent system in vivo.
Redox State of Gpx2 in Mutants Defective in Thioredoxin-Besides glutaredoxin, TRX is also a small enzyme that functions as a proteindisulfide oxidoreductase. In many Prxs, TRX serves as the endogenous reducing power. In addition, some kinds of GPx have been reported to use TRX as a reducing power (22,23). S. cerevisiae has three TRXs (TRX1, TRX2, and TRX3) (24,25). Both Trx1 and Trx2 are basically cytosolic residents (15), whereas Trx3 is located in mitochondria (25). We disrupted each of the three TRXs and determined the redox states of Gpx2 in the resultant mutant, although no distinct difference in the redox state was observed in a single mutant of each TRX even under conditions of oxidative stress (data not shown, result for trx3⌬ is shown in Fig. 1C). However, in a cytosolic TRX-deficient mutant (trx1⌬trx2⌬), a small proportion of Gpx2 was in the oxidized form under non-stressed conditions, and a large amount of Gpx2 ox appeared after the treatment with H 2 O 2 , t-BHP, and cumene hydroperoxide (Figs. 1, B and C).
We examined the effect of H 2 O 2 and t-BHP concentrations on the redox state of Gpx2 in trx1⌬trx2⌬ cells. As shown in Fig. 2A, Gpx2 was oxidized with 0.1 mM H 2 O 2 , and at Ͼ0.4 mM H 2 O 2 , the proportion of Gpx2 ox exceeded 40%. Regarding the redox of Gpx2 in t-BHP-treated cells, 0.2 mM t-BHP completely oxidized Gpx2 in trx1⌬trx2⌬ cells; however, Gpx2 was mostly in the reduced form in wild-type cells even if 1.0 mM t-BHP was present. At lower concentrations of t-BHP (Ͻ50 M), the proportion of Gpx2 ox in trx1⌬trx2⌬ cells was gradually increased in accordance with the concentrations of t-BHP ( Fig. 2A, inset). We also monitored the time course of Gpx2 oxidation in trx1⌬trx2⌬ cells after the treatment with 0.4 mM H 2 O 2 or 0.6 mM t-BHP. In both cases Gpx2 was oxidized within 5 min after the treatment with these peroxides (Fig. 2B).
Gpx2 was completely oxidized after the treatment with 0.2 mM t-BHP for 5 min in trx1⌬trx2⌬ cells; however, this was not the case in H 2 O 2treated cells even at higher concentrations and long-term exposure (Fig.  2). These results suggest that the prooxidant activity of t-BHP is higher than that of H 2 O 2 . In addition, because S. cerevisiae has two catalases (Ctt1 and Cta1), H 2 O 2 may be decomposed in vivo. Furthermore, Collinson et al. (26) have reported that glutaredoxins (Grx1 and Grx2) exhibit GPx activity in the presence of GSH recycling system, and overexpression of either GRX1 or GRX2 conferred resistance to H 2 O 2 but not to t-BHP (26). Taken together, H 2 O 2 may be consumed more efficiently than t-BHP in trx1⌬trx2⌬ cells, and therefore, complete oxidation of Gpx2 was not observed after H 2 O 2 treatment.
Next, we investigated the effect of mitochondrial TRX-deficiency on the redox state of Gpx2. As shown in Fig. 1C, the ratio of Gpx2 ox to total Gpx2 under non-stressed conditions slightly increased in a TRX-null mutant (trx1⌬trx2⌬trx3⌬) compared with that in the trx1⌬trx2⌬ mutant. Intriguingly, Gpx2 was completely oxidized in trx1⌬trx2⌬trx3⌬ cells after the treatment with 0.4 mM H 2 O 2 (Fig. 1C), under which conditions ϳ50% of Gpx2 was in the reduced form in trx1⌬trx2⌬ cells (Figs. 1, B and C). Gpx2 was mostly in the oxidized form in 0.6 mM t-BHP-treated trx1⌬trx2⌬trx3⌬ cells, as was the case on trx1⌬trx2⌬ cells (Figs. 1, B and C). These results suggest that the redox of Gpx2 is predominantly regulated by cytosolic TRX, although mitochondrial TRX also seems to contribute. It should be noted that the cytosolic TRX is a negative regulator of Yap1 (15), and GPX2 is one of the targets of this transcription factor (1, 18), therefore, the steady state level of Gpx2 increases in cells with the trx1⌬trx2⌬ as well as trr1⌬ background because of the constitutive activation of Yap1 (15,27).
Redox State of Gpx2 Is Associated with That of TRX-The oxidized TRX is reduced by thioredoxin reductase. S. cerevisiae has two genes coding for this enzyme; i.e. TRR1 for reduction of the cytosolic TRX (Trx1 and Trx2), and TRR2 for mitochondrial TRX (Trx3). The Gpx2 in a trr1⌬ mutant treated with 0.4 mM H 2 O 2 was mostly in the reduced form (Fig. 1C), and this was also the case in a trr2⌬ mutant (Fig. 1C). Meanwhile, Gpx2 was completely oxidized in trr1⌬ cells after the treat-FIGURE 1. Redox state of Gpx2 in vivo in different genetic backgrounds. A, AMS modification of the purified Gpx2. The purified Gpx2 was treated with AMS as described under "Experimental Procedures" and subjected to non-reducing SDS-PAGE followed by Western blotting. MW, molecular weight. B, effect of peroxides on the redox state of Gpx2. Cells were cultured in SD medium to log phase and exposed to peroxides as indicated in the figure for 1 h. Subsequently, Gpx2 was modified with AMS. Strains used in each lane are indicated in figure. CHP, cumene hydroperoxide. C, effect of peroxides on the redox state of Gpx2 in mutants defective in TRX and its recycling system. Cells were cultured and exposed to peroxides as described in B. Strains used in each lane are indicated in the figure. Because the expression of GPX2 is induced by oxidative stress in a Yap1-dependent manner (1,18), the Gpx2 level is increased in cells treated with H 2 O 2 or t-BHP. In addition, Yap1 is constitutively activated in trx1⌬trx2⌬ and trr1⌬ cells (15,26). Basal levels of Gpx2 in these mutants are higher than those of other strains. ment with 0.6 mM t-BHP; nevertheless, Gpx2 was in the reduced form in trr2⌬ cells under the same conditions (Fig. 1C). This suggests that the redox status of the cytosolic TRX may influence the redox state of Gpx2.
To explore the correlation between the redox state of TRX and that of Gpx2 more directly, we determined the redox state of TRX with an AMS modification assay. As shown in Fig. 3A, cytosolic TRX was virtu-  ally in the reduced form in wild-type cells in the presence or absence of peroxides. On the other hand, small amounts of cytosolic TRX were in the oxidized form (TRX ox ) in trr1⌬ cells under non-stressed conditions, and the proportion of TRX ox increased after the treatment with 0.4 mM H 2 O 2 (ϳ50%) and reached 100% with 0.6 mM t-BHP. These results suggest that the redox state of Gpx2 is linked with that of TRX, i.e. Gpx2 in trr1⌬ cells treated with H 2 O 2 was in the reduced form because onehalf of all the TRX was still TRX red under such conditions. Because the antibody against TRX we used is not able to distinguish between Trx1 and Trx2 (15), the TRX monitored here is the total cytosolic TRX.
Next, we expressed TRX1 or TRX2 with a multicopy plasmid in the trr1⌬ mutant. We confirmed that the plasmid-borne Trx1 and Trx2 were functional in terms of the redox regulation of Gpx2 by expressing them in trx1⌬trx2⌬ cells. As a result, the overexpression of TRX1 and TRX2 was not able to suppress the TRR1-deficiency with respect to the redox regulation of Gpx2 after the treatment with t-BHP (Fig. 3B). We also expressed GRX1 or GRX2 in the trx1⌬trx2⌬ and trr1⌬ mutants, although predictably, neither Grx1 nor Grx2 was able to compensate for the loss of TRX and its recycling system (Fig. 3C). Collectively, the redox state of Gpx2 in yeast cells is likely to be regulated in a TRX-dependent manner.
Intramolecular Disulfide Bond Formation between Cys 37 and Cys 83 of Gpx2-We could not detect the band shift corresponding to the dimer formation of Gpx2 in trx1⌬trx2⌬ and trr1⌬ cells after treatment with H 2 O 2 or t-BHP (data not shown); meanwhile, the apparent molecular weight of Gpx2 in the AMS modification assay after treatment of cells with peroxides was smaller than that in untreated cells. These observations suggest that an intramolecular disulfide bond is formed within a monomer of Gpx2. To determine which Cys residues correspond to the intramolecular disulfide bond, each Cys residue was substituted with Ala, and the resultant mutants (C10A, C37A, C65A, and C83A) were expressed in gpx2⌬trx1⌬trx2⌬ cells (Fig. 4). The reduced form of Gpx2 (Gpx2 red ) of the Cys-substituted Gpx2 mutants after the AMS modification migrated slightly faster than wild-type Gpx2 red , which means that the number of AMS molecules associated with each Gpx2 molecule is decreased (WT, 4 ϫ AMS; Cys mutants, 3 ϫ AMS). The band corresponding to Gpx2 ox was not observed in C37A and C83A mutants after the treatment with H 2 O 2 or t-BHP. This suggests that the intramolecular disulfide bond is formed between Cys 37 and Cys 83 . Consequently, the oxidized form of Gpx2 (Gpx2 ox ) of the C10A mutant after the AMS modification (Cys 37 -Cys 83 and Cys 65 -AMS) migrated slightly faster than wild-type Gpx2 ox (Cys 37 -Cys 83 ; Cys 10 -AMS and Cys 65 -AMS), which was also because the number of AMS molecules able to associate with each Gpx2 ox protein was decreased. However, intriguingly, the apparent molecular weight of the oxidized form of the C65A mutant (Cys 37 -Cys 83 ; Cys 10 -AMS) was almost the same as that of wild-type Gpx2 ox ; i.e. migration was slightly retarded. On the other hand, a faint band migrating slightly faster than the major reduced band appeared in the C83A mutant.
TRX-dependent Peroxidase Activity of Gpx2 in Vitro-We have demonstrated that the redox state of Gpx2 is regulated by a TRX-dependent system in vivo, which implies that Gpx2 behaves as a TPx. To address this possibility, we purified Gpx2 and measured the peroxidase activity with the TRX system (TRX, thioredoxin reductase, and NADPH) or with the GSH system (GSH, glutathione reductase, and NADPH).
His-tagged Gpx2 was purified from E. coli and the His tag was removed (Fig. 5A). The purified Gpx2 was subjected to a TPx assay. Gpx2 exhibited TPx activity toward H 2 O 2 and t-BHP. This activity was dependent on both Gpx2 and TRX. The kinetic parameters are summarized in TABLE TWO. The peroxidase activity of Gpx2 toward H 2 O 2 with the GSH system was ϳ10 times lower than that with the TRX system (TABLE TWO), and the catalytic efficiency of Gpx2 in the reduction of H 2 O 2 with the TRX system was 100 times that with the GSH system (k cat /K m value in the TRX system was 4.79 ϫ 10 7 M Ϫ1 s Ϫ1 , and that in the GSH system was 5.85 ϫ 10 5 M Ϫ1 s Ϫ1 ). On the other hand, the K m value for t-BHP with the GSH system was 5 times larger than that with the TRX system, and the k cat /K m value for t-BHP with the TRX system was ϳ10 times larger (5.89 ϫ 10 6 M Ϫ1 s Ϫ1 ) than that with the GSH system (3.48 ϫ 10 5 M Ϫ1 s Ϫ1 ).
To confirm whether an intramolecular disulfide bond is formed in vitro, the enzyme was incubated with H 2 O 2 without the TRX system. In this experiment, the H 2 O 2 -treated Gpx2 was directly subjected to nonreducing and reducing SDS-PAGE, respectively, and the gels were stained for proteins with Coomassie Brilliant Blue. As shown in Fig. 5B, the majority of Gpx2 turned into the oxidized form after the treatment with H 2 O 2 , which migrates faster in non-reducing SDS-PAGE. Only a small portion of Gpx2 seems to form a dimer through an intermolecular disulfide bond. These results together with the redox state regulation in vivo strongly suggest that Gpx2 functions as an atypical 2-Cys peroxiredoxin using TRX as the endogenous reducing power.
Roles of Gpx2 in Oxidative Stress Response in Yeast-Previously we showed that a gpx2⌬ mutant did not exhibit susceptibility to H 2 O 2 and t-BHP using a spot assay (1). This was confirmed in the present study (data not shown). We then determined the susceptibility to t-BHP by another method, i.e. cells were treated with t-BHP in the medium for a prescribed time, and viability was monitored by counting the colonies on YPD agar plates (survival test). As shown in Fig. 6A, the gpx2⌬ mutant was slightly sensitive to t-BHP compared with the isogenic wildtype strain. We have also reported that expression of GPX2 was induced in cells exposed to oxidative stress and CaCl 2 (1,16,18), and therefore, Gpx2 may have some physiological relevance to the oxidative stress response in the presence of Ca 2ϩ . We then determined the susceptibility to t-BHP in the presence of CaCl 2 . Interestingly, the susceptibility of wild-type cells to t-BHP was slightly increased when CaCl 2 was present, and this tendency was remarkable in gpx2⌬ cells (Fig. 6A). We monitored the redox state of Gpx2 in cells treated with CaCl 2 in the presence or absence of t-BHP, although CaCl 2 did not affect the redox balance of Gpx2 (Fig. 6B).
Next, to determine whether Gpx2 is able to substitute for Gpx3 as an antioxidant, we overexpressed Gpx2 in gpx3⌬ cells. Regarding the overexpression of GPX2, we have previously found that the deletion of a part of the GPX2 promoter (Ϫ709 to Ϫ306) enhanced the basal expression level (16), and therefore, the partially promoter-deleted GPX2 (designated GPX2-3) was introduced into the gpx3⌬ mutant. BecauseYap1 is not fully activated in gpx3⌬ cells, oxidative stress-induced expression of GPX2 was impaired in the gpx3⌬ mutant; nevertheless, we confirmed that the Gpx2 protein level was high in cells carrying the GPX2-3 allele (Fig. 6C, upper panel), and consequently, the susceptibility of gpx3⌬ cells to peroxides was partially suppressed (Fig. 6C, lower panel).

DISCUSSION
Roles of Gpx2 as an Antioxidant-S. cerevisiae has five TPX genes (TSA1, TSA2, AHP1, DOT5, and PRX1) (7), and the enzymatic proper-ties of some of them have been extensively studied. The activity of Tsa1 toward H 2 O 2 and t-BHP with the TRX system in vitro (V max values with H 2 O 2 as substrate were 4.8 mol/min/mg of protein, and for t-BHP, 2.4 mol/min/mg of protein) (28) was comparable with that of Gpx2 (TABLE TWO). However, because Tsa1 has higher affinity for peroxides (K m values for H 2 O 2 were 3 M and for t-BHP, were 10 M) than Gpx2 (TABLE TWO) and TSA1 is highly expressed throughout the growth phase of yeast (29), Tsa1 is most likely to be the major TPx/Prx in the cytoplasm. On the other hand, a gpx3⌬ mutant was hypersensitive to H 2 O 2 and t-BHP, whereas a gpx2⌬ mutant was not (1). Delaunay et al. (9) have reported that Gpx3 has TPx activity (9). Although the V max value of Gpx3 to H 2 O 2 (ϳ1.29 mol/min/mg of protein) (9) with the TRX system was lower than that of Gpx2 (2.6 mol/min/mg of protein), the GPX3 gene is constitutively expressed at higher levels in yeast cells (1,29). We then explored whether overexpression of GPX2 can compensate for the loss of GPX3 in vivo. As shown in Fig. 6C, susceptibility to peroxides of the gpx3⌬ mutant was partially suppressed by overexpression of GPX2.
We found that the viability of yeast cells was reduced when CaCl 2 and t-BHP coexisted, and this tendency was much more obvious in the gpx2⌬ mutant (Fig. 6A). As shown in Fig. 6B, CaCl 2 did not affect the redox state of Gpx2, whereas the amount of Gpx2 was further increased in cells treated with CaCl 2 and t-BHP concomitantly. We have previously reported that oxidative stress-induced expression of GPX2 is independent of the regulatory mechanism of Ca 2ϩ -mediated signaling to the GPX2 promoter (16), and therefore, an additive effect in terms of the induction of GPX2 in the presence of peroxide and CaCl 2 was observed in the GPX2-lacZ reporter assay (16). We confirmed that this is true at the protein level (Fig. 6B). Taken together, our results suggest that Gpx2 is important in the oxidative stress-induced response in the presence of Ca 2ϩ .
In mammalian cells excessive Ca 2ϩ accumulation in mitochondria induces opening of the mitochondrial permeability transition pore, which eventually leads to the release of cytochrome c from mitochondria, an early trigger in the apoptotic cascade (30). By contrast, in S. cerevisiae, Ca 2ϩ alone is not able to raise the permeability of the mitochondrial membrane, although the co-existence of t-BHP leads to the similar phenomenon of opening of mitochondrial permeability transition pore (31). In mammals, the opening of mitochondrial permeability transition pore is inhibited by many antioxidants (32)(33)(34)(35)(36), and the same thing can be observed in yeast mitochondria; i.e. Kowaltowski et al. (31) reported that Tsa1 and catalase protect mitochondria from permeabilization induced by Ca 2ϩ and t-BHP. We have not yet determined whether or not the cell death caused by t-BHP and Ca 2ϩ is apoptosis; determination of the physiological relevance as well as intracellular localization of Gpx2 in the response to oxidative stress in the presence of Ca 2ϩ is now underway.
Redox Regulation of Non-SeCys-type GPx-Rocher et al. (3) reported that the substitution of SeCys with Cys in murine cytosolic GPx (cGPx/ GPx1) decreased the enzyme activity 1000-fold. This was also the case for PHGPx (4). In the peroxidase reaction of Prxs, the conserved per- oxidatic cysteine (Cys-S P H) is oxidized to cysteine sulfenic acid (Cys-S P OH). Therefore, Cys at the active site of the SeCys-substituted GPx is presumed to be oxidized to Cys-SOH by H 2 O 2 . cGPx/GPx1 is a homotetramer, and PHGPx is a monomeric enzyme. If a disulfide bond is formed intermolecularly or intramolecularly as a consequence of the reduction of H 2 O 2 by the SeCys-substituted enzyme, a thiol-disulfide oxidoreductase may be necessary for recovery of the active site Cys. The same thing may be applied also to the peroxidase reaction carried out by the "GPx-like proteins" that do not contain SeCys at the active site. We have demonstrated here that an intramolecular disulfide bond between Cys 37 and Cys 83 seems to be formed within a monomer of Gpx2 in vivo (Fig. 4). Presumably, Cys 37 may function as a peroxidatic Cys, and Cys 83 may be a resolving Cys analogous to Gpx3. Regarding the reduction of Gpx2, TRX, but not glutaredoxin, is the preferable reductant to recover Cys 37 -SH and Cys 83 -SH in vivo, and GSH is also able to function as a reducing power in vitro even if the efficiency is lower than that by TRX. Delaunay et al. (9) have reported that the peroxidase activity of Gpx3 is strictly dependent on TRX, and GSH is not able to serve as the reducing power for Gpx3 in vitro. However, Avery and Avery (2) have reported that Gpx3 showed GSH-dependent peroxidase activity toward t-BHP and phospholipid hydroperoxide. Hence, the electron donor for Gpx3 is still controversial.  (18). To determine the susceptibility to peroxides, gpx3⌬ cells carrying pRS413 (vector) or pRSGPX2-3H (GPX2-3) were serially diluted and spotted onto SD agar plates containing H 2 O 2 or t-BHP.