Oxidation of a Eukaryotic 2-Cys Peroxiredoxin Is a Molecular Switch Controlling the Transcriptional Response to Increasing Levels of Hydrogen Peroxide*

Although activation of the AP-1-like transcription factor Pap1 in Schizosaccharomyces pombe is important for oxidative stress-induced gene expression, this activation is delayed at higher concentrations of peroxide. Here, we reveal that the 2-Cys peroxiredoxin (2-Cys Prx) Tpx1 is required for the peroxide-induced activation of Pap1. Tpx1, like other eukaryotic 2-Cys Prxs, is highly sensitive to oxidation, which inactivates its thioredoxin peroxidase activity. Our data suggest that the reduced thioredoxin peroxidase-active form of Tpx1 is required for the peroxide-induced oxidation and nuclear accumulation of Pap1. Indeed, in contrast to the previously described role for Tpx1 in the activation of the Sty1 stress-activated protein kinase by peroxide, we find that both catalytic cysteines of Tpx1 are required for Pap1 activation. Moreover, overexpression of the conserved sulfiredoxin Srx1, which interacts with and reduces Tpx1, allows rapid activation of Pap1 at higher concentrations of H2O2. Conversely, loss of Srx1 prevents the reduction of oxidized Tpx1 and prolongs the inhibition of Pap1 activation. Collectively, these data suggest that redox regulation of the thioredoxin peroxidase activity of Tpx1 acts as a molecular switch controlling the transcriptional response to H2O2. Furthermore, they reveal that a single eukaryotic 2-Cys Prx regulates peroxide signaling by multiple independent mechanisms.

Oxidative stress, the intracellular damage caused by reactive oxygen species, has been implicated in numerous diseases, including neurodegenerative disorders, diabetes, atherosclerosis, and cancer, and proposed as a molecular cause of aging (1)(2)(3). Reactive oxygen species, including hydrogen peroxide (H 2 O 2 ), superoxide anions (O ⅐ ), and hydroxyl radicals (OH ⅐ ), are generated as by-products of cellular metabolism and as a result of exposure to UV light and other environmental agents (3). Reactive oxygen species are also employed as intracellular signaling molecules by pathways involved in proliferation, stress responses, mating behavior, and apoptosis. Increased levels of reactive oxygen species activate signaling pathways as part of the oxidative stress response to restore homeostasis by increasing the expression of genes encoding antioxidant and repair proteins.
2-Cys peroxiredoxins (2-Cys Prxs) 1 are extremely abundant and highly conserved thioredoxin peroxidase enzymes with important roles in resistance to oxidative stress (4,5) and heat stress (6) and in tumorigenesis (7). The catalytic mechanism of these enzymes involves two conserved cysteine residues; the "peroxidatic" and the "resolving" cysteines. During the catalytic breakdown of peroxide, the peroxidatic cysteine thiol is oxidized by peroxide to a sulfenic acid. The reduction of the enzyme to its original active redox state occurs by a mechanism involving thioredoxin and the resolving cysteine. The sulfenic acid derivative of the peroxidatic cysteine can be further oxidized to an inactive sulfinic form that cannot be reduced by thioredoxin (8). Although eukaryotic 2-Cys Prxs are particularly susceptible to inactivation by overoxidation of the peroxidatic cysteine (8), enzyme activities involved in catalyzing the reduction of the sulfinic acid derivative of eukaryotic 2-Cys Prxs have been identified in budding yeast (9) and mammals (10). For example, in Saccharomyces cerevisiae the sulfiredoxin Srx1 forms a peroxide-induced disulfide with the 2-Cys Prx Tsa1 and reduces the sulfinic acid to a sulfenic acid form that can then be further reduced by thioredoxin (9). Orthologues of Srx1 are conserved in mammals and have been shown to be involved in reduction of the sulfinic acid form of 2-Cys Prx from humans (11,12).
It has been proposed that the susceptibility of the thioredoxin peroxidase activity of eukaryotic 2-Cys Prxs to inactivation allows increasing levels of H 2 O 2 to activate peroxide signaling (8). However, there is also evidence that 2-Cys Prxs have active roles in peroxide signaling in both budding yeast (13) and fission yeast (14). Indeed, our recent studies in the fission yeast Schizosaccharomyces pombe demonstrated that the 2-Cys Prx Tpx1 is required for the peroxide-induced activation of Sty1, the S. pombe homologue of the p38/c-Jun N-terminal kinase stress-activated protein kinase, through the formation of a peroxide-induced disulfide between the peroxidatic cysteine of Tpx1 and a cysteine residue of Sty1 (14). Interestingly, unlike the thioredoxin peroxidase activity of Tpx1, the regulation of Sty1 by Tpx1 does not require the resolving cysteine (14,15). Furthermore, recent studies have revealed that the S. cerevisiae 2-Cys Prx, Tsa1, has a peroxidase-independent chaperone activity that is important in protecting cells from heat stress (6). Together, these studies suggest that eukaryotic 2-Cys Prxs have thioredoxin peroxidase-independent roles in signal transduction and protection of cells from stress.
In S. pombe, different levels of H 2 O 2 induce distinct patterns of oxidative stress-induced gene expression (16,17). These different transcriptional responses are dependent on the AP-1like transcription factor Pap1 and the Atf1 transcription factor, which are related to the mammalian transcription factors Nrf2 and ATF2, respectively (16). The transcriptional response to low levels of H 2 O 2 is dependent on Pap1, whereas Atf1 is more important at higher levels ( Fig. 1) (16). In unstressed conditions Pap1 associates with the nuclear export factor Crm1, ensuring that Pap1 remains cytoplasmic (18). However, following stress Pap1 becomes oxidized, inhibiting its association with Crm1 and resulting in the nuclear accumulation of Pap1 (16, 18 -21). Atf1 is phosphorylated by the Sty1 stress-activated protein kinase (SAPK), although the role of this phosphorylation is unknown (22). Whereas Sty1 and Atf1 are increasingly activated by increasing levels of peroxide (16), in contrast Pap1 is oxidized and accumulates in the nucleus much more rapidly at low levels of peroxide than at higher levels (16,21). Consistent with these observations, the induction of Pap1-dependent genes is reduced, and Atf1-dependent transcription increases as peroxide levels increase (Fig. 1). The molecular mechanism(s) by which Pap1 oxidation and activation are inhibited at increased peroxide concentrations is unknown.
In the budding yeast S. cerevisiae the Gpx3 peroxidase is directly involved in the peroxide-induced oxidation and nuclear accumulation of the Yap1 transcription factor (23). In contrast, here we show that the single orthologue of Gpx3 in S. pombe, Gpx1, is not required for the peroxide-induced activation of Pap1. Instead, we reveal that the 2-Cys Prx Tpx1 is essential for the peroxide-induced oxidation and nuclear accumulation of Pap1. Furthermore, both catalytic cysteines of Tpx1 are required for the regulation of Pap1, in contrast to the Tpx1-dependent regulation of the peroxide-induced activation of the Sty1 SAPK (14). We propose that the inhibition of Pap1 activation by increasing peroxide concentrations is due to increased oxidation of Tpx1, inactivating the thioredoxin peroxidase activity. In support of this model, we demonstrate that the overexpression of srx1 ϩ , which encodes the S. pombe orthologue of the Srx1 sulfiredoxin in S. cerevisiae, reduces the peroxide-induced oxidation of Tpx1 and promotes the nuclear accumulation of Pap1 at higher levels of H 2 O 2 . Furthermore, deletion of the srx1 ϩ gene prevents reduction of oxidized Tpx1, inhibiting the peroxide-induced nuclear accumulation of Pap1 and Pap1-dependent gene expression at higher concentrations of H 2 O 2 . Thus, we propose that the levels of H 2 O 2 and the peroxide-induced expression of the Sty1-regulated gene srx1 ϩ together control the redox state of the peroxidatic cysteine of Tpx1 and thus act as the key determinants of the pattern of gene expression induced by different levels of oxidative stress.

EXPERIMENTAL PROCEDURES
Growth Media-The S. pombe strains used in this study are listed in Table I. All strains were grown in rich medium (YE5S) or synthetic minimal medium (EMM2) as described previously (24). All strains were grown at 30°C, and experiments were performed using mid-log phase cultures (A 595 ϭ 0.2-0.5). Transformed haploid cells were grown on EMM2 containing 75 g/ml adenine and 75 g/ml histidine.
Gene Tagging and Strain Construction-Strains were constructed that express either Pap1 (SB3) or Tpx1 (AD1) tagged at the C terminus with 3xPk epitopes from their normal genomic locus, respectively. ϳ500 bp of the C terminus of the coding region of Pap1 and the whole coding region of Tpx1 were amplified by PCR and ligated into PstI/BamH1digested or NdeI/BamH1-digested Rep42PkC derivatives, respectively. These plasmids were linearized and introduced into CHP429 cells to create SB3 and AD1, respectively. Strains containing Pk epitope-tagged Pap1 and either ⌬tpx1 (SB4) or ⌬sty1 (SB5) were generated from a cross between SB3 and either VX00 or JP140, respectively. The SB6 (⌬srx1) and SB13 (⌬gpx1) strains were generated using specific ura4 ϩ disruption cassettes created by PCR with oligonucleotide primers containing 90 bp of DNA upstream and downstream of the chosen open reading frame. The disruption cassettes were introduced into either CHP429 or CHP428 to create SB6 and SB13, respectively. An h ϩ ⌬srx1 strain (SB7) was obtained from a cross between SB6 and CHP428. SB8 (⌬srx1 Pap1-Pk) and SB9 (⌬srx1 Tpx1-Pk) were from crosses between SB7 and either SB3 or AD1, respectively. SB14 was from a cross between SB13 and SB3. The DNA sequences of oligonucleotide primers used in strain construction are available on request.
RNA Analysis-Cells were grown to mid-log phase in either EMM2 or YE5S medium and then harvested. RNA extraction (27) and Northern blotting with gene-specific probes were performed as described previously (28). The DNA sequence of oligonucleotide primers used to obtain specific probes are available on request.
Sensitivity Tests-The sensitivity of mutant strains to peroxide were compared by halo tests; ϳ2 ϫ 10 6 cells were pipetted along separate lines radially from the edge of the plate toward a disk soaked in 5 l of 15% (v/v) H 2 O 2 . Plates were examined after 3 days of incubation at 30°C. The small zone of growth inhibition of wild type cells around a peroxide-soaked filter gives only a small window for detecting increases in resistance. Therefore, spot tests were used to examine potential increases in the resistance of wild-type cells; 10-fold dilutions of cultures, grown to mid-log phase in EMM2, were spotted onto EMM2 agar plates containing the indicated concentrations of the more thermostable peroxide, tert-butyl hydroperoxide, using a 96-well plate replica plater (Sigma-Aldrich). Indirect Immunofluorescence-To examine the intracellular localization of Pap1, mid-log phase growing cells were harvested from cultures treated with H 2 O 2 or diethylmaleate (DEM), fixed with paraformaldehyde, and analyzed using the immunofluorescence protocol described previously (29). Cells were incubated overnight at room temperature with an an anti-Pk primary antibody (mouse anti-Pk-TAG; Serotec) and detected using a 1:200 dilution of Alexa Fluor 488 goat anti-mouse secondary antibody (Molecular Probes, Eugene, OR). Pk epitope-tagged Pap1 was visualized by excitation at 450 -490 nm using a Zeiss Axioscope fluorescence microscope with a 63ϫ oil immersion objective and the Axiovision digital imaging system.
Detecting the Redox State of Tpx1 and Pap1-To examine the oxidation state of Tpx1 or Pap1, exponentially growing cells expressing either Pk epitope-tagged Tpx1 or Pap1 were harvested, and lysates were then prepared by acid lysis with trichloroacetic acid (30). In the case of Pap1, these lysates were also treated with alkaline phosphatase (Roche Diagnostics). Proteins were separated by electrophoresis on non-reducing 15% (Tpx1) or 8% (Pap1) SDS-polyacrylamide gels and transferred to a Protran® nitrocellulose transfer membrane (Schleicher & Schuell Bioscience). Mouse anti-Pk primary antibodies (mouse anti-Pk-TAG; Serotec) and horseradish peroxidase-conjugated anti-mouse IgG secondary antibodies (Sigma) were used to detect epitope-tagged Pap1 or Tpx1, which were visualized using the ECL detection system (Amersham Biosciences).
Immunoprecipitation of Srx1-Exponentially growing cells were harvested and lysates were prepared as described previously (29). FLAG epitope-tagged Srx1 was partially purified from duplicate samples of lysates using anti-FLAG (M2) antibody-conjugated agarose (Sigma). Immunoprecipitated proteins were separated by electrophoresis on a non-reducing 15% SDS-polyacrylamide gel and then transferred to nitrocellulose (see above). FLAG-Srx1 was detected on one-half of the membrane using anti-FLAG (M2) primary antibodies (Sigma), and Pk epitope-tagged Tpx1 was detected on the other half using anti-Pk primary antibodies (mouse anti-Pk-TAG; Serotec). Both halves were then probed with horseradish peroxidase-conjugated anti-mouse IgG secondary antibodies (Sigma), and proteins were visualized as described above.

Tpx1
Regulates Pap1-dependent Gene Expression-It has recently been shown that increased expression of antioxidant enzymes with peroxidase activity, such as the glutathione peroxidase Gpx1, stimulates the peroxide-induced nuclear accumulation of Pap1 (21). Moreover, in S. cerevisiae an orthologue of Gpx1, the Gpx3 peroxidase, acts as a redox sensor and transducer for the peroxide-induced activation of Yap1 (23). Hence, to investigate whether Gpx1 is required for the peroxide-induced activation of Pap1, a ⌬gpx1 mutant was constructed. Examination of the peroxide-sensitivity revealed that the ⌬gpx1 mutant was much less sensitive to peroxide than a ⌬pap1 mutant ( Fig. 2A), suggesting that Gpx1 does not play a major role in the regulation of Pap1. Moreover, the peroxideinduced expression of the Pap1-dependent gene trr1 ϩ , which encodes thioredoxin reductase (18), was not inhibited by loss of Gpx1 expression (Fig. 2B). These data suggest that Pap1 is normally activated by a Gpx1-independent mechanism(s) in response to peroxide.
Recently, we demonstrated that the 2-Cys peroxiredoxin Tpx1 regulates peroxide-induced activation of Sty1 and Atf1 (14). Furthermore, the peroxide-induced expression of the ctt1 ϩ gene encoding catalase is dependent on Tpx1 (14). The expression of ctt1 ϩ has been shown to be regulated by both the Pap1 and Atf1 transcription factors ( Fig. 1) (16,18). Therefore, to investigate whether Tpx1 regulates Pap1 activity we examined the effect of loss of tpx1 ϩ on the peroxide-induced expression of trr1 ϩ , a gene whose expression is dependent on Pap1 but not on Atf1 ( Fig. 1 and data not shown) (18,21). Analyses of RNA isolated from wild-type, ⌬pap1, and ⌬tpx1 strains revealed that Tpx1, like Pap1, was required for the peroxide-induced expression of trr1 ϩ (Fig. 2B). It was unlikely that the complete ab-sence of any peroxide-induced increase in trr1 ϩ expression in ⌬tpx1 cells was due to the reduced peroxide-induced activation of Sty1 in these cells (14), as it has previously been shown that trr1 ϩ expression is induced by 0.2 mM H 2 O 2 even in a ⌬sty1 strain (18,21). Indeed, the similar effects of the loss of Pap1 or Tpx1 on peroxide-induced expression of trr1 ϩ (Fig. 2B) suggest that Tpx1 is a key regulator of Pap1 activity in S. pombe.
Tpx1 Is Specifically Required for the Peroxide-induced Nuclear Accumulation and Oxidation of Pap1-Previous data revealed that Pap1 is predominantly regulated through alteration of its cellular distribution in response to oxidative stress (18). Hence, we examined whether Tpx1 was required for the peroxide-induced nuclear accumulation of Pap1 by indirect immunofluorescent analyses using anti-Pk antibodies of tpx1 ϩ and ⌬tpx1 cells expressing Pk epitope-tagged Pap1 from its normal chromosomal locus. As expected, Pap1 was detected in the nucleus of tpx1 ϩ cells within 5 min following treatment with 0.2 mM H 2 O 2 and was excluded again from the nucleus by 60 min (Fig. 3A). In contrast, there was no detectable accumulation of Pap1 in the nuclei of ⌬tpx1 cells at any time point following treatment with 0.2 mM H 2 O 2 (Fig. 3A). Importantly, re-introduction of the wild-type tpx1 ϩ gene into ⌬tpx1 cells restored the peroxide-induced nuclear accumulation of Pap1 (see Fig. 9) Together, these data reveal that Tpx1 is required for the peroxide-induced nuclear accumulation of Pap1. Consistent with the absence of any effect of loss of gpx1 ϩ on Pap1dependent gene expression (Fig. 2B), the peroxide-induced nuclear accumulation of Pap1 was similar in both ⌬gpx1 cells and wild-type control cells (Fig. 3A).
In addition to H 2 O 2 , the nuclear accumulation of Pap1 is induced by other oxidizing agents such as DEM (19). However, in contrast to H 2 O 2 treatment, the DEM-induced nuclear accumulation of Pap1 occurred with similar kinetics in tpx1 ϩ and ⌬tpx1 cells (Fig. 3B), indicating that Tpx1 is specifically required for the peroxide-induced nuclear accumulation of Pap1.
Both DEM and H 2 O 2 cause oxidative changes in Pap1 that interfere with nuclear export by Crm1, resulting in the nuclear accumulation of Pap1 (20,21). However, whereas DEM forms a stable thioether bond with Cys-532 of Pap1, H 2 O 2 induces an intramolecular disulfide bond between Cys-278 and Cys-501 (20,21). To examine whether Tpx1 is required for the peroxideinduced formation of this intramolecular disulfide bond in Pap1, we examined the redox state of Pap1 in wild-type control ⌬gpx1 and ⌬tpx1 cells expressing Pk epitope-tagged Pap1 from its native chromosomal locus. To preserve in vivo redox states, we used methods similar to those described previously (21). Protein extracts were prepared by acid lysis from untreated and peroxide-treated wild-type control (SB3), ⌬gpx1, and ⌬tpx1 cells and then treated with iodoacetamide (IAA) to alkylate any free thiol groups (present on reduced cysteines) before separation by non-reducing SDS-PAGE and Western blotting analy- In contrast to Pap1, Atf1 becomes increasingly activated as H 2 O 2 levels increase. Examples of Pap1-dependent (trr1 ϩ encoding thioredoxin reductase), Atf1-dependent (gpx1 ϩ encoding glutathione peroxidase), and Pap1/Atf1-dependent (ctt1 ϩ encoding catalase) genes are indicated.
sis. This method allows detection of Pap1 oxidation as an increase in mobility (because of decreased reactivity with IAA) (21,30). Indeed, following treatment with 0.2 mM H 2 O 2 for 10 min, the pattern of Pap1-specific bands was altered in wildtype control cells; the predominant lower mobility reduced band of Pap1, present in unstressed cells, was replaced with two more mobile bands, consistent with the formation of oxidized forms of Pap1 (Fig. 3C). In contrast to the effects of loss of Gpx3 on the peroxide-induced oxidation of Yap1 in S. cerevisiae cells, a similar pattern of Pap1 oxidation was observed in gpx1 ϩ and ⌬gpx1 cells (Fig. 3C). However, oxidized forms of Pap1 were undetectable in the absence of Tpx1 (Fig. 3C), even when the film was overexposed (data not shown), suggesting that Tpx1 is essential for the peroxide-induced oxidation of Pap1 (Fig. 3C). Taken together, these data (Figs. 2 and 3) show that in contrast with S. cerevisiae, where Gpx3 plays a major role in the regulation of Yap1 (23, 31), Tpx1 (rather than Gpx1, the orthologue of Gpx3) is required for the normal peroxideinduced oxidation and nuclear accumulation of Pap1.
Oxidation of Tpx1 by H 2 O 2 Coincides with the Inhibition of Pap1 Nuclear Accumulation-Our data suggested that Tpx1 is a key regulator of the peroxide-induced activation of Pap1. Tpx1, like other eukaryotic 2-Cys Prxs, is susceptible to oxidation (15). Interestingly, the oxidation and nuclear accumulation of Pap1 are inhibited by similar levels of H 2 O 2 to those that have been reported to oxidize Tpx1, inhibiting its thioredoxin peroxidase activity (15,16,21). To further investigate this correlation, we compared the kinetics of Tpx1 and Pap1 oxidation in response to different levels of H 2 O 2 . Lysates were extracted by acid lysis from cells expressing Pk epitope-tagged Tpx1 following treatment with either 0.2 or 1 mM H 2 O 2 . These cell lysates were then treated with AMS, an agent that irreversibly alkylates reduced cysteines but not oxidized sulfinic acid derivatives, and analyzed by non-reducing SDS-PAGE and Western blotting with anti-Pk antibodies (9,30). The lower mobility band represents the AMS-reacted, reduced form of Tpx1, whereas, following exposure to 1 mM H 2 O 2 , a more mobile band appears to represent the AMS-resistant, oxidized form of Tpx1 (Fig. 4A). Very little transient oxidation of Tpx1 was detected following exposure to 0.2 mM H 2 O 2 (Fig. 4A). In contrast, both the extent and the duration of Tpx1 oxidation  A, Pk epitope-tagged Tpx1 was analyzed by non-reducing SDS-PAGE and Western blotting of AMS-modified proteins prepared by acid lysis from AD1 cells and exposed to either 0.2 or 1 mM H 2 O 2 for 0, 5, 10, 20, 60, or 120 min. AMS-resistant, oxidized Tpx1 (Tpx1 ox ) has a faster mobility than reduced Tpx1 (Tpx1 red ). Lysate isolated from CHP429 cells expressing untagged Tpx1 was included as a control. B, indirect immunofluorescence was used to visualize Pk epitope-tagged Pap1 after treatment of SB3 cells with 1 mM H 2 O 2 for 0, 10, or 60 min. In addition, iodoacetylated (IAA) proteins were extracted by acid lysis from SB3 cells, treated with alkaline phosphatase, and analyzed by non-reducing SDS-PAGE and Western blotting. IAA-resistant, oxidized Pap1 (Pap1 ox ) has a faster mobility than reduced Pap1 (Pap1 red ).
was significantly increased when cells were exposed to 1 mM H 2 O 2 (Fig. 4A). Significantly, this increased oxidation of Tpx1 correlates with the inhibition of the nuclear accumulation of Pap1 following treatment with increasing levels of H 2 O 2 (16,21) (Fig. 4B). For example, Pap1 only appears to accumulate in the nucleus 60 min after exposure of cells to 1 mM H 2 O 2 (Fig.  4B), which correlates with a corresponding decrease in the level of oxidized Tpx1 (Fig. 4A).
The timing of Pap1 oxidation corresponds with the nuclear accumulation of the protein (21). Therefore, we investigated the level and timing of Pap1 oxidation following treatment with 1 mM H 2 O 2 . Cells expressing Pk epitope-tagged Pap1 were exposed to 1 mM H 2 O 2 for 0, 10, or 60 min, and the oxidation state of Pap1 was examined as described previously (21). As expected, Pap1 only became oxidized at 60 min after treatment with 1 mM H 2 O 2 (Fig. 4B). The correspondence of the timing of the reappearance of reduced Tpx1 with the oxidation and nuclear accumulation of Pap1 suggests that the reduced, thioredoxin peroxidase-active form of Tpx1 is required for the peroxide-induced oxidation of Pap1. Furthermore, these data suggest that the oxidation of Tpx1 is the key event that prevents the activation of Pap1 at high levels of H 2 O 2 .
The Peroxide-induced Sulfiredoxin Srx1 Interacts with and Reduces Oxidized Tpx1-Our studies suggest that the peroxide-induced activation of Pap1 depends on the redox state of Tpx1. In this model, the pathway(s) that regulates the redox state and thioredoxin peroxidase activity of Tpx1 would be predicted to influence the regulation of Pap1. Recent studies have shown that the inactivation of 2-Cys Prxs by high levels of peroxide is reversible (9,32). For example, a sulfiredoxin, Srx1, was identified in S. cerevisiae that forms a complex with the 2-Cys Prx Tsa1 and can reduce overoxidized Tsa1 to the active form (9). Our data indicated that the oxidation of Tpx1 to AMS-unreactive forms could also be reversed following exposure to 1 mM H 2 O 2 (Fig. 4A, compare the 20-min and 60-min samples), suggesting the presence of an enzymatic activity that reverses this oxidation. Indeed, analysis of the S. pombe genome data base revealed an open reading frame, SPBC106.02c, encoding a potential protein with a high degree of homology to Srx1 in S. cerevisiae (11) (Fig. 5A). Hence, based on this homology we have named this open reading frame srx1 ϩ . Global expression studies in S. pombe had previously revealed that the expression of srx1 ϩ was induced ϳ34-fold by 0.5 mM H 2 O 2 in a Sty1-dependent manner, suggesting an important role for Srx1 in the oxidative stress response (17). Indeed, our RNA analysis confirmed that srx1 ϩ was induced in a Sty1-dependent manner following treatment of cells with 1 mM H 2 O 2 (Fig. 5B).
Based on the studies in S. cerevisiae (9), Tpx1 and Srx1 would be predicted to form a complex. To investigate this possibility, FLAG epitope-tagged Srx1 was immunopurified from cells (AD1) expressing Pk epitope-tagged Tpx1 and analyzed by Western blotting with anti-Pk antibodies (Fig. 5C). This analysis revealed that Tpx1 and Srx1 do indeed form a complex in vivo and, importantly, that this interaction is stimulated by oxidative stress (Fig. 5C).
Next, we examined whether Srx1 could reduce oxidized Tpx1. Oxidation of Tpx1 peaked at 5 min following exposure of control cells (AD1) to 1 mM H 2 O 2 (Fig. 5D). In contrast, there was significantly less oxidation of Tpx1 in cells overexpressing srx1 ϩ at any of the time points following treatment with 1 mM H 2 O 2 (Fig. 5D). This suggests that, like its S. cerevisiae counterpart, Srx1 acts to directly reduce Tpx1 to its thioredoxin peroxidase-active form. Indeed, consistent with these observations, overexpression of srx1 ϩ , similar to overexpression of tpx1 ϩ , increased the resistance of tpx1 ϩ cells (SB3) to peroxide (Fig. 6A). Moreover, overexpression of srx1 ϩ did not signifi-cantly increase the resistance of ⌬tpx1 cells (SB4) to peroxide, suggesting that the ability of Srx1 to protect against oxidative stress is dependent on the presence of Tpx1 (Fig. 6B).
Srx1 Stimulates the Peroxide-induced Nuclear Accumulation of Pap1 in a Tpx1-dependent Manner-The identification of the sulfiredoxin that reduces Tpx1 allowed us to test the hypothesis that the oxidation of Tpx1 following exposure to high levels of H 2 O 2 prevents the peroxide-induced nuclear accumulation of Pap1. First, we examined the effect of overexpression of srx1 ϩ , which inhibits the oxidation of Tpx1 at 1 mM H 2 O 2 (Fig. 5D), on the nuclear accumulation of Pap1 at 1 mM H 2 O 2 . As predicted by our model, we found that the overexpression of srx1 ϩ stimulated the peroxide-induced nuclear accumulation of Pap1 at this level of oxidative stress (Fig. 6C). Importantly, this effect of srx1 ϩ overexpression was Tpx1-dependent (Fig. 6C).
Based on our analysis of Srx1 function and our model of Tpx1-dependent regulation of Pap1, deletion of the srx1 ϩ gene would be predicted to prevent the reversal of oxidation of Tpx1 and, thus, the oxidation and nuclear accumulation of Pap1 at 60 min following treatment with 1 mM H 2 O 2 . To test these predictions, we constructed an ⌬srx1 mutant. Peroxide sensitivity tests revealed that ⌬srx1 cells are more sensitive than srx1 ϩ cells but much less sensitive than ⌬tpx1 cells (Fig. 7A). In addition, as expected the deletion of the srx1 ϩ gene prevented the reduction of oxidized Tpx1 to its AMS-sensitive form fol- Gene-specific probes were used to examine the levels of srx1 ϩ RNA relative to the leu1 ϩ loading control. C, cells (AD1) expressing Pk epitope-tagged Tpx1 from its native chromosomal locus and containing either Rep41FLAG or Rep41FLAGsrx1 ϩ were treated with 1 mM H 2 O 2 for 0 or 10 min. Anti-FLAG-agarose immunoprecipitates of cell lysates were then analyzed by non-reducing SDS-PAGE and Western blotting with anti-FLAG or anti-Pk antibodies. A nonspecific band is indicated by an asterisk. D, Pk epitope-tagged Tpx1 was analyzed by non-reducing SDS-PAGE and Western blotting of AMS-modified proteins prepared by acid lysis from AD1 cells containing either Rep1 (vector) or Rep1srx1 ϩ (ϩsrx1 ϩ ) and treated with 1 mM H 2 O 2 for 0, 5, 10, 20, 60, or 120 min. AMS-resistant, oxidized Tpx1 (Tpx1 ox ) has a faster mobility than reduced Tpx1 (Tpx1 red ). lowing exposure of cells to 1 mM H 2 O 2 ( Fig. 7B; compare the 60-min time point in the presence and absence of srx1 ϩ ). In further support of our model that the redox state of Tpx1 regulates Pap1, we found that, in addition to the increased oxidation of Tpx1, the oxidation and nuclear accumulation of Pap1 was also prevented in ⌬srx1 cells following treatment with 1 mM H 2 O 2 (Fig. 7, C and D). In agreement with these observations, RNA analysis revealed that Pap1-regulated gene expression was also inhibited in the ⌬srx1 mutant at 1 mM H 2 O 2 (Fig. 7E). In contrast to the effects of loss of Tpx1 activity, the regulation of trr1 ϩ expression was unaffected when ⌬srx1 cells were exposed to 0.2 mM H 2 O 2 (Fig. 7E). Consistent with these observations, we did not find any changes in the oxidation of Tpx1 or the normal peroxide-induced nuclear accumulation of Pap1 when srx1 ϩ and ⌬srx1 cells were exposed to 0.2 mM H 2 O 2 (data not shown).
Collectively, these data provide further evidence that the oxidation state of Tpx1 regulates the activation of Pap1. Significantly, the data also indicate that an important function of Srx1, following the exposure of cells to high levels of peroxide, is to promote the oxidation and nuclear accumulation of Pap1 by reducing Tpx1 to its thioredoxin peroxidase-active form. As srx1 ϩ expression is strongly induced by oxidative stress (17; Fig. 5B), this suggests that Srx1 has an important regulatory role in controlling the timing of Pap1-dependent transcription (see Fig. 11).
Tpx1 Stimulates Pap1 Activation Independently of Sty1-We previously established a thioredoxin peroxidase-independent role for Tpx1 in peroxide signaling by demonstrating that Tpx1 is required for the peroxide-induced activation of the Sty1 SAPK (14). At 1 mM H 2 O 2 , the peroxide-induced nuclear accumulation of Pap1 and the expression of ctt1 ϩ and trr1 ϩ are dependent on Sty1 (16,18,21,33) (data not shown). Therefore, although our data clearly show that Tpx1 is essential for the activation of Pap1 by lower concentrations of H 2 O 2 (0.2 mM) where Pap1 activation is less dependent on Sty1 (21), it was possible that the Tpx1 regulation of Pap1 was, at least in part, through its ability to stimulate the peroxide-induced activation of Sty1. Significantly, however, we found that overexpression of tpx1 ϩ (Rep1tpx1 ϩ ) restored inducible ctt1 ϩ expression in ⌬sty1 cells (Fig. 8A), although, as expected, in the absence of Sty1dependent activation of Atf1 the level of ctt1 ϩ expression was much lower than in sty1 ϩ cells (data not shown). As shown previously, Pap1 did not accumulate in the nucleus of ⌬sty1 mutant cells at 1 mM H 2 O 2 (16,18,21) (Fig. 8B). However, overexpression of Tpx1 or Srx1 stimulated the peroxide-induced nuclear accumulation of Pap1 in these cells at 1 mM H 2 O 2 (Fig. 8B). These data suggest that Tpx1 regulates Pap1 independently of the Sty1 pathway. The data also suggest that the loss of Srx1 expression in a ⌬sty1 (Fig. 5B) may be important for the loss of Pap1 activity at 1 mM H 2 O 2 (16,21).
As part of the 2-Cys Prx family, Tpx1 has a peroxidatic cysteine (Cys-48) and a resolving cysteine (Cys-169), both of which are needed for the thioredoxin peroxidase activity of these enzymes (for example, see Refs. 8 and 15). To assess the importance of these cysteines of Tpx1 in the regulation of Pap1, we examined the ability of plasmids expressing either wildtype Tpx1 or mutant versions of Tpx1 where the individual cysteine residues had been substituted with serine (Tpx1 C48S or Tpx1 C169S ) to stimulate the nuclear accumulation of Pap1. As expected, the expression of wild-type Tpx1 restored the peroxide-induced nuclear accumulation of Pap1 when these cells were exposed to 0.25 mM H 2 O 2 (Fig. 9). However, in contrast, the peroxide-induced nuclear accumulation of Pap1 was not restored by the expression of either Tpx1 C48S or Tpx1 C169S (Fig. 9). These data contrast with our previous studies showing that Tpx1 C169S could restore activation of StyI phosphorylation in ⌬tpx1 cells at similar time points and levels of peroxide (14).
Collectively, these data suggest that Tpx1 regulates Sty1 and Pap1 by distinct mechanisms; the thioredoxin peroxidase activity of Tpx1 is dispensable for the regulation of Sty1 but may also be important for the activation of Pap1. Examination of the ability of Tpx1 C169S to restore the transcriptional response of ⌬tpx1 cells to 0.2 and 1 mM H 2 O 2 revealed that ectopic expression of Tpx1 C169S was completely unable to restore the inducible expression of the Pap1-dependent, Atf1-independent trr1 ϩ gene (Fig. 10A). Consistent with previous data, loss of Tpx1 resulted in increased basal expression of the Atf1-dependent, Pap1-independent, gpx1 ϩ gene, but with no significant further increase in gpx1 ϩ mRNA levels following exposure to H 2 O 2 (Fig. 10B) (14). However, the loss of induction of the gpx1 ϩ gene in a ⌬tpx1 mutant (Fig. 10B) (14) was completely rescued by the ectopic expression of Tpx1 C169S , providing further strong evidence that Tpx1 is involved in separate mechanisms to regulate the activation of Pap1 and Sty1.
In summary, we find that a single 2-Cys Prx is a key redox sensor that directly regulates multiple peroxide signaling pathways by distinct thioredoxin peroxidase-dependent and -independent mechanisms (Fig. 11). Moreover, we also present the first evidence that the sensitivity of a eukaryotic 2-Cys Prx to oxidation and its regulation by a sulfiredoxin play a central role in both sensing and coordinating the cellular responses to oxidative stress. DISCUSSION The activation of the AP-1-like transcription factor Pap1 is critical for the adaptive response of S. pombe to oxidative stress (18). However, Pap1 is activated faster in response to low levels of H 2 O 2 ; as H 2 O 2 levels increase, Pap1 activation becomes inhibited. Here, we reveal that the 2-Cys peroxiredoxin Tpx1 and the sulfiredoxin Srx1 regulate peroxide signal transduction to Pap1 and present a model to explain the inefficient activation of Pap1 by increasing concentrations of H 2 O 2 . Im-FIG. 6. Overexpression of srx1 ؉ increases resistance to peroxide and stimulates the peroxide-induced nuclear accumulation of Pap1 in a Tpx1-dependent manner. A, 10-fold serial dilutions of mid-log phase cultures of wild-type cells (SB3) containing either Rep1 (vector), Rep1tpx1 ϩ (ϩtpx1 ϩ ), or Rep1srx1 ϩ (ϩsrx1 ϩ ) were plated on EMM2 media containing the indicated concentrations of tert-butyl hydroperoxide (t-BOOH) and incubated at 30°C. B, ⌬tpx1 mutant (SB4) cells containing either Rep1 (vector), Rep1tpx1 ϩ (ϩtpx1 ϩ ), or Rep1srx1 ϩ (ϩsrx1 ϩ ) were evenly distributed along a radial line from the edge of an agar plate toward a central disk soaked in H 2 O 2 . Plates were incubated for 3 days at 30°C. C, Pk epitope-tagged Pap1 was visualized by indirect immunofluorescence in tpx1 ϩ (SB3) and ⌬tpx1 (SB4) cells containing either Rep1 (vector) or Rep1srx1 ϩ (ϩsrx1 ϩ ) after treatment with 1 mM H 2 O 2 for 0, 5, 10, 20, or 60 min. portantly, the regulation of Pap1 by Tpx1 and Srx1 is in addition to and independent from the recently identified the thioredoxin peroxidase-independent role of Tpx1 in the regulation of the Sty1 SAPK (14). Thus, our data provide new insight into the role(s) of eukaryotic 2-Cys Prxs and sulfiredoxins in signal transduction and offer a mechanism for the regulation of distinct transcriptional responses to different levels of stress.
Previous studies revealed that the accumulation of Pap1 in the nucleus and the activation of Pap1-dependent gene expression are delayed at higher concentrations of H 2 O 2 (16). Our discovery that the reduced thioredoxin peroxidase-active form of Tpx1 is required for the peroxide-induced nuclear accumulation of Pap1 suggests that the oxidation of Tpx1 by high concentrations of H 2 O 2 prevents the efficient activation of Pap1.
Recent studies have identified enzymatic activities in eukaryotes capable of reducing the oxidized sulfinic acid form of the peroxidatic cysteine of 2-Cys Prxs (9,32). For example, the Srx1 sulfiredoxin was identified as possessing the enzyme ac-tivity responsible for the reduction of the oxidized sulfinic acid derivative of the 2-Cys Prx Tsa1 in S. cerevisiae (9). In mammals, the p53-regulated sestrins (10) and orthologues of Srx1 (11,12) have been shown to reduce the sulfinic acid derivative of 2-Cys Prx. Here, we have shown that the S. pombe orthologue of Srx1 interacts with and is required for the reduction of oxidized Tpx1, indicating that the function of Srx proteins in the reduction of 2-Cys Prxs is conserved in a wide range of eukaryotes.
The identification of proteins such as Srx1 that are capable of reversing the oxidation of eukaryotic 2-Cys Prxs led to the suggestion that these proteins play important roles in the regulation of peroxide signaling (9,10,32). Indeed, we show that the overexpression of Srx1 prevents oxidation of Tpx1 (Fig.  5D) and stimulates the peroxide-induced nuclear accumulation of Pap1 (Fig. 6C). Furthermore, deletion of the srx1 ϩ gene prolongs oxidation of Tpx1, inhibiting the peroxide-induced nuclear accumulation of Pap1 and the activation of Pap1-dependent gene expression (Fig. 7). Interestingly, our prelimi- FIG. 7. Srx1 is required for the reduction of oxidized Tpx1 and the activation of Pap1 by peroxide. A, wildtype control (CHP429), ⌬tpx1 (VX00), and ⌬srx1 (SB6) strains were evenly distributed along a radial line from the edge of an agar plate toward a central disk soaked in H 2 O 2 and incubated at 30°C. B, lysates were obtained by acid lysis from cells of the wild-type control (AD1) and ⌬srx1 (SB9) strains expressing Pk epitope-tagged Tpx1 following exposure to 1 mM H 2 O 2 for 0, 5, 10, 20, or 60 min. Lysates were treated with AMS and analyzed by non-reducing SDS-PAGE and Western blotting. Oxidized Tpx1 (Tpx1 ox ) and reduced Tpx1 (Tpx1 red ) are indicated. C, Pk epitope-tagged Pap1 was analyzed by non-reducing SDS-PAGE and Western blotting of iodoacetylated proteins prepared by acid lysis from cells of the wildtype control (SB3) and ⌬srx1 (SB8) strains treated with 1 mM H 2 O 2 for 0, 10, or 60 min. IAA-resistant oxidized Pap1 (Pap1 ox ) has a faster mobility than reduced Pap1 (Pap1 red ). D, Pk epitopetagged Pap1 was visualized by indirect immunofluorescence in srx1 ϩ (SB3) and ⌬srx1 (SB8) cells treated with 1 mM H 2 O 2 for 0, 10, or 60 min. E, Northern blot analysis of RNA isolated from cells of the wild-type control (AD1), ⌬tpx1 (VX00), and ⌬srx1 (SB6) strains exposed to either 0.2 or 1 mM H 2 O 2 for 0, 20, or 40 min. Gene-specific probes were used to examine the expression of the indicated genes. The level of trr1 ϩ RNA was quantified relative to the leu1 ϩ loading control. nary analysis suggests that Srx1 is not involved in the Tpx1dependent regulation of the peroxide-induced activation of Sty1 (data not shown). This possibility is consistent with our previous observations that the peroxide-induced activation of Sty1 remains partially dependent on Tpx1, even at the high concentrations of H 2 O 2 where Tpx1 is predominantly oxidized (14). Moreover, the formation of Tpx1-Sty1 disulfides is not inhibited at these higher concentrations (14) (data not shown). Thus, it appears that Srx1 is only involved in regulating the thioredoxin peroxidase-dependent activity of Tpx1 in the regulation of Pap1. srx1 ϩ expression is highly induced following treatment with H 2 O 2 in a Sty1-dependent manner (17) (see Fig. 5B). Hence, our data are consistent with a model whereby cells respond to low levels of H 2 O 2 by the Tpx1-dependent activation of Pap1. In response to high levels of H 2 O 2 , Tpx1 is initially oxidized at early time points, preventing activation of Pap1. However, at these levels of stress the oxidation of Tpx1 does not prevent the activation of the Sty1 pathway and the expression of Sty1-dependent genes such as srx1 ϩ (14) (Fig. 11). As Srx1 levels increase, the thioredoxin peroxidase activity of Tpx1 is restored, resulting in stimulation of the nuclear accumulation of Pap1 and Pap1-dependent gene expression (Fig. 11). In summary, these data suggest that waves of transcription factor activity regulate the cellular response to H 2 O 2 with the thiore-doxin peroxidase activity of Tpx1 and its regulation by Srx1, playing a central role in regulating these transcriptional responses.
Although the peroxide-induced nuclear accumulation of Pap1 is impaired in ⌬sty1 cells, Pap1 does not appear to be a substrate for Sty1, leading to the suggestion that Sty1 regulation of Pap1 is indirect (18,21). Vivancos et al. (21) recently showed that Pap1 oxidation is inhibited by high levels of peroxide and suggested that the reduced expression of genes encoding antioxidant enzymes such as catalase and glutathione peroxidase in ⌬sty1 cells results in prolonged exposure to high intracellular peroxide levels, preventing the oxidation and subsequent nuclear accumulation of Pap1. In general, our identification of the Tpx1 peroxidase as a critical redox sensor required for the oxidation of Pap1 supports their model. Indeed, we have extended their study to demonstrate that deletion of either the tpx1 ϩ or the srx1 ϩ gene, but not the gpx1 ϩ gene, prevents the normal activation of Pap1. However, in the light of our data showing that Tpx1 is oxidized and unable to stimulate FIG. 11. Tpx1 regulates the Sty1 pathway and the Pap1 transcription factor by independent mechanisms. Section 1, the thioredoxin peroxidase activity (the peroxidatic and resolving cysteine residues) of Tpx1 is required for the peroxide-induced activation of Pap1 via the formation of an intramolecular disulfide in Pap1. As H 2 O 2 levels increase, the thioredoxin peroxidase activity of Tpx1 is inactivated by oxidation preventing oxidation, and, thus, activation of Pap1. Srx1 activity reverses this oxidation of Tpx1 and therefore acts to prevent inhibition of Pap1 activation. Section 2, Tpx1 is required for the peroxide-induced activation of Sty1 by a thioredoxin peroxidase-independent mechanism (requiring the peroxidatic cysteine but not the resolving cysteine). This involves the formation of a peroxide-induced intermolecular disulfide between Tpx1 and Sty1. Active Sty1 is required for the peroxide-induced expression of srx1 ϩ , which regenerates active Tpx1 to allow the activation of Pap1. Pap1 oxidation at high levels of H 2 O 2 , we suggest that an important role of Sty1 in the peroxide-induced activation of Pap1 is to regulate the expression of srx1 ϩ (Fig. 5B), allowing the reduction of Tpx1 and thus restoring thioredoxin peroxidase activity and the ability to activate Pap1. This possibility is supported by our data showing that the ectopic expression of Srx1 restores peroxide-induced nuclear accumulation of Pap1 in ⌬sty1 cells (Fig. 8B).
The mechanism by which the peroxide-induced intramolecular disulfide bond forms in Pap1 is unknown (21). Studies in S. cerevisiae have revealed that Gpx3, a thiol peroxidase unrelated to 2-Cys Prxs, and a second protein, Ybp1, are directly involved in the peroxide-induced oxidation and nuclear accumulation of Yap1 (23,31). However, analyses of the effects of deleting the gpx1 ϩ gene, the single homologue of GPX3 in the S. pombe data base, on the regulation of Pap1 suggests that Pap1 and Yap1 are regulated by different mechanisms. Indeed, it has previously been demonstrated that an intramolecular disulfide in Pap1 forms more rapidly at low concentrations of H 2 O 2 , suggesting that a cysteine(s) in Pap1 and/or in a sensor protein is sensitive to overoxidation by high levels of H 2 O 2 (21). Our data showing that the timing of oxidation of Tpx1 by increasing concentrations of H 2 O 2 is linked with Pap1 activation provide evidence that Tpx1 acts as an H 2 O 2 sensor in the activation of Pap1 and that Tpx1 is required for the normal oxidation of Pap1. The 2-Cys Prx Tsa1 appears to be directly involved in the oxidation and activation of Yap1 in strains of S. cerevisiae expressing a defective form of Ybp1 (34). The precise molecular mechanism underlying the regulation of Pap1 by Tpx1 is under investigation. Although we have found that a small amount of Tpx1 is copurified with Pap1, suggesting that Tpx1 may directly regulate oxidation of Pap1, there is no change in the level of this interaction following peroxide treatment (data not shown). Moreover, we have been unable to detect the formation of an intermolecular disulfide between Tpx1 and Pap1 (data not shown). Thus. it remains a strong possibility that the regulation of Pap1 by Tpx1 is indirect.
Several thioredoxin peroxidase-independent activities of 2-Cys Prxs have recently been identified (6,14). Indeed, the thioredoxin peroxidase-independent chaperone activity of Tsa1 in S. cerevisiae is stimulated by peroxide-induced structural changes that inactivate its peroxidase activity (6). In addition, we have shown that Tpx1 in S. pombe regulates the Sty1 SAPK through the formation of an intermolecular disulfide complex with Sty1 in a thioredoxin peroxidase-independent manner (14). In contrast, the present study has demonstrated a Sty1independent role for Tpx1 in signal transduction that is dependent on the presence of both catalytic cysteines of Tpx1. Hence, together with previous studies our results suggest that eukaryotic 2-Cys Prxs are versatile peroxide sensors that employ multiple mechanisms to stimulate transcriptional responses appropriate to the level of stress. As abundant redoxsensitive modulators of peroxide levels, 2-Cys Prxs are well placed to serve as sensors during peroxide signaling and to co-ordinate cellular responses to oxidative stress. Significantly, the observation that Srx family members can reduce oxidized 2-Cys Prxs from yeast to mammals (9, 11, 12) (our study) and our finding that Srx1 is involved in regulating the cellular response to oxidative stress in S. pombe have important implications for the roles of sulfiredoxins in eukaryotes. In metazoans, recent studies have suggested that cell type-specific responses to oxidative stress are required (35,36) and, thus, it will be important to examine the potential role of 2-Cys Prxs and sulfiredoxins in such responses.