A cooperative role for Atf1 and Pap1 in the detoxification of the oxidative stress induced by glucose deprivation in Schizosaccharomyces pombe.

In Schizosaccharomyces pombe, glucose concentrations below a certain threshold trigger the stress-activated protein kinase (SAPK) signal transduction pathway and promote increased transcription of Atf1-dependent genes coding for the general stress response. Removal of glucose specifically induces the nuclear accumulation of green fluorescent protein-labeled Pap1 (GFP-Pap1) and the expression of genes dependent on this transcription factor. In contrast, depletion of the nitrogen source triggers the SAPK pathway but does not activate Pap1-dependent gene transcription, indicating that carbon stress rather than growth arrest leads to an endogenous oxidative condition that favors nuclear accumulation of Pap1. The reductant agents glutathione or N-acetylcysteine suppress the nuclear accumulation of GFP-Pap1 induced by glucose deprivation without inhibiting the activation of the MAPK Sty1. In addition, cells expressing a mutant GFP-Pap1 unable to accumulate into the nucleus upon hydrogen peroxide-mediated oxidative stress failed to show this protein into the nucleus in the absence of glucose. These results support the concept of a concerted action between the SAPK pathway and the Pap1 transcription factor during glucose exhaustion by which glucose limitation induces activation of the SAPK pathway prior to the oxidative stress caused by glucose deprivation. The ensuing induction of Atf1-dependent genes (catalase) decreases the level of hydroperoxides allowing Pap1 nuclear accumulation and function. Congruent with this interpretation, glucose-depleted cells show higher adaptive response to exogenous oxidative stress than those maintained in the presence of glucose.

Glucose is a powerful signaling molecule that promotes major metabolic changes in cells (1). Glucose metabolism produces compounds directly related to the detoxification of intracellular hydroperoxides formed as byproducts by ongoing metabolic processes (2). In human tumor cells, which typically show strong glycolysis and a reduced rate of respiration, deprivation of glucose causes a strong metabolic oxidative stress characterized by increased steady state levels of intracellular hydroperoxides and glutathione disulfide (3,4). Strong evidence indicates that the absence of glucose also triggers signaling cascades that activate transcription factors and the expression of stress-related genes attempting to redirect cellular functions (2). Remarkably, glucose metabolism in some fermenting yeasts is quite similar to that of tumor cells. In particular, the Crabtree-positive fission yeast S. pombe ferments glucose under aerobic conditions. Unlike Saccharomyces cerevisiae, this yeast lacks enzymes of the glyoxylate cycle that maintain diauxic growth after glucose depletion and utilizes very few nonsugar carbon sources (5,6). Thus, as soon as glucose disappears and respiration of the fermentation products is impaired, cultures of S. pombe may suffer nutritional stress.
The mitogen-activated protein kinase (MAPK) 1 signaling pathways are critical for the sensing and response of eukaryotic cells to changes in the external environment (7,8). These MAPK cascades are highly conserved through evolution and serve to transduce signals to the nucleus, resulting in new patterns of gene expression (9,10). The identification of a highly conserved stress-activated protein kinase (SAPK) pathway in S. pombe allows us to analyze the mechanisms by which SAPKs are activated in a system more amenable than higher eukaryotic cells (11)(12)(13)(14)(15). In this yeast, the central element of the SAPK cascade is the MAPK Sty1 (also known as Spc1 or Phh1), which is highly homologous to mammalian p38 kinase and becomes activated by a similar series of stresses (12,13,(15)(16)(17). MAPK Sty1 is directly phosphorylated by MAPK kinase Wis1; however, the transmission pathway of the stress signal to Wis1 is dual, and either MAPK kinase kinase Wak1 (also known as Wis4 or Wik1) or MAPK kinase kinase Win1 is responsible for Wis1 phosphorylation (18). A response regulator protein, Mcs4, associates with Wak1, and probably with Win1, to regulate MAPK kinase kinase activity in response to several stimuli (17,19). In S. pombe different transcription factors function downstream of the MAPK Sty1 cascade, among which Atf1 and Pap1 have been studied extensively. Atf1 (also known as Gad7 or Mts1) is a mammalian ATF-2 homologue b-ZIP protein that associates to and is phosphorylated by Sty1 following different stresses (20 -22). In fact, Sty1 is the only * This work was supported in part by Grant BMC 2002-01104 from the Ministerio de Ciencia y Tecnología (MCYT, Spain) (to J. C.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
§ A predoctoral fellow from the Ministerio de Educación, Cultura y Deporte (Spain).
ʈ To whom correspondence should be addressed: Dept. of Genetics and Microbiology, Facultad de Biología, University of Murcia, Campus Universitario de Espinardo, 30071 Murcia, Spain. Tel.: 34-968367132; Fax: 34-968363963; E-mail: maga@um.es. known kinase involved in Atf1 phosphorylation during stress. Transcription of a wide array of stress-response genes like gpx1 ϩ (coding for glutathione peroxidase), ntp1 ϩ (neutral trehalase), ctt1 ϩ (cytoplasmic catalase), fbp1 ϩ (fructose-1,6bisphosphatase), or ste11 ϩ (a high mobility group protein involved in the regulation of sexual differentiation) is controlled by Sty1 through Atf1 (21,22). Another transcription factor, Pap1, encoded by the pap1 ϩ gene, has been isolated as required for survival to oxidative stress and, like its S. cerevisiae homologue YAP1, shows high homology to mammalian c-Jun (23). S. pombe cells deleted in pap1 ϩ show high sensitivity to oxidative stress but not to osmotic stress or nutrient deprivation (24). Moreover, Pap1 activity is regulated at the level of cellular localization. In glucose-growing cells Pap1 localizes to the cytoplasm but accumulates in the cell nucleus upon oxidative stress with hydrogen peroxide or the glutathione-depleting agent diethyl maleate (DEM). Hydrogen peroxide reversibly oxidizes two cysteine residues in Pap1 (at positions 278 and 501), whereas DEM induces a non-reversible modification (25). As a result, modified Pap1 is unable to interact in both cases with the exportin Crm1 through the nuclear export signal located at the Pap1 carboxyl terminus (25). This favors its nuclear accumulation and the increased transcription of essential genes for defense against oxidative stress, like ctt1 ϩ (catalase), trr1 ϩ (thioredoxin reductase), or sod1 ϩ (superoxide dismutase) (24,26). In contrast to Atf1, Pap1 is neither phosphorylated nor a substrate for Sty1 upon stress conditions; however, Sty1 presence/function is needed for nuclear accumulation of Pap1 in response to high concentrations of hydrogen peroxide, but not at low concentrations of the prooxidant (27,28).
Several results have been reported previously that carbon starvation is an environmental stress able to activate the MAPK Sty1 in S. pombe (29,30). However, the contribution of the SAPK pathway to the cellular responses under glucose depletion after fermentative growth has not been investigated in detail. Transfer of cells from a glucose-rich to a glucose-free culture medium without an alternative carbon source may help to reveal characteristic responses normally masked under conditions of sugar availability or cell growth. Following such an approach we have analyzed the stress signal induced by glucose deprivation in cultures maintained under similar osmotic contexts to avoid the influence of stressing conditions unrelated to the absence of glucose. In this work we demonstrate that glucose limitation in S. pombe not only promotes activation of the SAPK signaling pathway that results in increased expression of Atf1-dependent stress-related genes but also induces an oxidative stress that favors the concerted expression of additional genes depending on the transcription factor Pap1. These results may help us to understand the mechanisms underlying the comparative resistance of glucose-depleted cells against external oxidative conditions.
Purification and Detection of Activated Sty1-Ha6H and Atf1-Ha6H Proteins following Glucose Deprivation-Yeast cells grown in EMM2 with 7% glucose to an A 600 of 0.5 (actual glucose concentration ϭ 6%, determined by the glucose oxidase method) were recovered by filtration and resuspended in the same medium devoid of glucose. EMM2 without glucose was osmotically equilibrated with 3% glycerol, 2.8% glycerol plus 0.5% glucose, 2.5% glycerol plus 1% glucose, 6% sorbitol, 5.5% sorbitol plus 0.5% glucose, or 5% sorbitol plus 1% glucose. At different times, the cells from 30 ml of culture were harvested by centrifugation at 4°C, and yeast pellets were immediately frozen in liquid nitrogen. Under these conditions, the previously described Sty1 phosphorylation resulting from centrifugation (30,32) was not observed in unstressed cells. To analyze Sty1, total cell homogenates were prepared under native conditions employing chilled acid-washed glass beads and lysis buffer (10% glycerol, 50 mM Tris-HCl, pH 7.5, 150 mM NaCl, 0.1% Nonidet P-40, supplemented with specific protease and phosphatase inhibitor cocktails (Sigma)). The lysates were removed and cleared by centrifugation at 13,000 ϫ g for 15 min. Ha6H-tagged Sty1 was purified by using nickel-nitrilotriacetic acid-agarose beads (Qiagen Inc.) as reported previously (32). The purified proteins were resolved in 10% SDS-polyacrylamide gels, transferred to nitrocellulose filters (Amersham Biosciences), and incubated with either mouse anti-Ha (Roche Applied Science, clone 12CA5) or mouse antiphospho-p38 (New England Biolabs) antibodies. The immunoreactive bands were revealed with an anti-mouse horseradish peroxidase-conjugated secondary antibody (Sigma) and the ECL system (Amersham Biosciences). For Atf1-Ha6H purification, the pelleted cells were lysed into denaturing lysis buffer (6 M guanidine HCl, 0.1 M sodium phosphate, 50 mM Tris HCl, pH 8.0), and the Atf1 protein was isolated by affinity precipitation on nickel-nitrilotriacetic acid-agarose beads as described previously (34). The purified proteins were resolved in 6% SDS-polyacrylamide gels, transferred to nitrocellulose filters (Amersham Biosciences), and incubated with a mouse anti-Ha antibody (12CA5).
RNA Isolation and Hybridization-Yeast cells grown in EMM2 with 7% glucose to an A 600 of 0.5 were recovered by filtration and resuspended in the same medium with 3% glycerol. Volumes of 50 ml of the cultures were recovered at different times, and total RNA preparations were obtained as described by Franco et al. (35) and resolved through 1.5% agarose-formaldehyde gels. Northern (RNA)-hybridization analyses were performed as described by Sambrook et al. (36). The probes employed were amplified by PCR and included the following: a 0.7-kbp fragment of the apt1 ϩ gene (23) that was amplified with the 5Ј-oligonucleotide CCCAGTATGTCTACC and the 3Ј-oligonucleotide AAGTCT-TACTTGCGG, a 0.4-kbp fragment of the gpx1 ϩ gene (37) amplified with the 5Ј-oligonucleotide TTCTACGACTTGGCT and the 3Ј-oligonucleotide ACACTCTCGATATCG, a 0.9-kbp fragment of the trr1 ϩ gene (38) amplified with the 5Ј-oligonucleotide GTGACTCACAACAAG and the 3Јoligonucleotide TAATCGGTATCTTCC, a 2.1-kbp fragment of the pyp2 ϩ gene (12) amplified with the 5Ј-oligonucleotide CCGAGAGCGTTTCT-TGGA and the 3Ј-oligonucleotide AAGGGCTTGGAAGCCTGG, and a 1-kbp fragment of the fbp1 ϩ gene (39) amplified with the 5Ј-oligonucleotide CTTCCAAGCCAAATACTG and the 3Ј-oligonucleotide GATCTCGACGAAATCGAC. Probes for ctt1 ϩ and leu1 ϩ were prepared as reported previously (32). To establish quantitative conclusions, the level of mRNAs was determined in a PhosphorImager (Amersham Biosciences) and compared with the internal control (leu1 ϩ mRNA).
Fluorescence Microscopy-To localize the GFP-Pap1 fusion, yeast cultures were grown in EMM2 with 7% glucose (with or without GSH and NAC) to an A 600 of 0.5, recovered by filtration, and resuspended in the same medium with 6% glucose but lacking the nitrogen source (EMM2-N), 3% glycerol, 6% sorbitol, or 0.01-0.1% glucose plus 3% glycerol. Treatment with DEM was performed by adding the compound to glucose-growing cells at a final concentration of 4 mM. Small aliquots (10 l) of the yeast cultures were loaded onto poly-L-lysine-coated slides, and the remaining suspension was withdrawn by aspiration. For nu-clear staining, 3 l of Hoechst in 50% glycerol was added. Fluorescence microscopy was performed on a Leica DM 4000B microscope with a ϫ100 objective. Images were captured with a cooled Leica DC 300F camera and IM50 software and then imported into Adobe PhotoShop 6.0 (Adobe Systems, Mountain View, CA).
Cell Viability Assays and Analytical Determinations-Yeast strains grown in YES medium (2% glucose and 0.6% yeast extract) with 7% glucose to an A 600 of 0.5 were recovered by filtration, resuspended for 1 h at 28°C in the same medium with either 6% glucose or 3% glycerol, and treated for 1 h with 80 mM H 2 O 2 . The samples were diluted and spread in triplicate onto plates containing YES solid medium, and cell viability was measured by their ability to form colonies on this medium after incubation at 28°C for 5 days. Results represent the mean values Ϯ S.D. from three different experiments. Glucose concentration in the growth media was assayed by the glucose oxidase method (40). Protein determination was performed according to Lowry et al. (41).

RESULTS
SAPK Activation during Glucose Depletion-We first analyzed the kinetics of Sty1 activation due to glucose exhaustion. An exponentially glucose-growing culture of strain JM1521, which harbors a genomic copy of sty1 ϩ tagged with two copies of the Ha epitope and six histidine residues (32), was shifted to a similar medium containing a non-assimilated carbon source (glycerol or sorbitol) instead of glucose. It should be noted that S. pombe does not feed on glycerol alone unless glucose is present for initial growth (42) 2 and that sorbitol is not a carbon source for this yeast (43) either. These compounds were used to prevent drastic osmotic changes in the non-glucose-containing medium and, hence, to avoid any potential disturbance of the SAPK pathway unrelated to glucose deprivation (44). Samples were collected at different times from the new medium. Sty1-HA6H protein was purified by affinity chromatography, and its activation was analyzed by Western immunoblotting using antiphospho-p38 antibodies. As shown in Fig. 1A and confirming previous reports (30), the absence of glucose provoked a clear peak of activation of Sty1, with maximal phosphorylation at 5 min, followed by a rather slight decrease. The occurrence of a rapid Sty1-mediated response during the transition from glucose to non-glucose conditions led us to assess the phosphorylation status of the downstream transcription factor Atf1. This was performed by employing strain JM1821, which carries a genomic copy of the atf1 ϩ gene tagged with two copies of the Ha epitope and six histidine residues (32). Earlier studies demon-FIG. 2. The SAPK pathway regulates the induction of different stress-responsive genes under glucose limitation. Strains TK003 (wild type (WT)) and NT146 (⌬atf1 ϩ ) were grown in EMM2 plus 7% glucose to early log phase and transferred to the same medium with 3% glycerol for the times indicated. Total RNA was extracted from each sample, and 20 g was resolved in 1.5% agarose-formaldehyde gels. Denatured RNAs were transferred to nylon membranes and hybridized with 32 P-labeled probes of fbp1 ϩ , pyp2 ϩ , gpx1 ϩ , ctt1 ϩ , and leu1 ϩ (loading control).
FIG. 1. SAPK pathway activation in S. pombe under glucose limitation. Strains JM1521 and JM1821 carrying Ha6H-tagged chromosomal versions of the sty1 ϩ or the atf1 ϩ gene, respectively, were grown in EMM2 plus 7% glucose to early log phase and transferred to the same medium with 3% glycerol (A), 2.5% glycerol plus 1% glucose (B), 2.8% glycerol plus 0.5% glucose (C), 6% sorbitol (D), 5% sorbitol plus 1% glucose (E), and 5.5% sorbitol plus 0.5% glucose (F). Aliquots were harvested at different times, and Sty1 or Atf1 was purified by affinity chromatography. Activated Sty1 was detected by immunoblotting with antiphospho-p38 antibodies. Total Sty1 and Atf1 (in both phosphorylated and unphosphorylated forms) were determined by immunoblotting with anti-Ha antibody. strated that Atf1 of unstressed cells migrates in gel as a single protein of ϳ85 kDa that undergoes a phosphorylation-dependent band shift under different stresses (32). As shown in Fig.  1A, glucose deprivation induced a Sty1-dependent band shift in the migration of Atf1 because of in vivo phosphorylation, whose initial kinetics matched closely that observed for Sty1 phosphorylation. Essentially identical results were obtained when sorbitol was used to balance osmolarity (Fig. 1D), confirming that the activation of the SAPK pathway is exclusively due to glucose limitation.
To clarify whether SAPK activation was triggered only after a complete exhaustion of glucose, we further analyzed the glucose deprivation-induced phosphorylation of Sty1/Atf1. As shown in Fig. 1, B and E, a switch from high glucose-containing medium to osmotically equilibrated medium (containing glycerol or sorbitol) with glucose concentrations of 1% (w/v) or higher (not shown) did not induce Sty1/Atf1 phosphorylation. However, lower glucose concentrations (0.5% or less) prompted a rapid and transient increase in both Sty1 and Atf1 phosphorylation ( Fig. 1, C and F) that was almost identical to that observed after complete glucose depletion (Fig. 1, A and D). Similar results were obtained when yeast cultures reached the early stationary phase of growth. 3 Hence, these data demonstrate that the activation of the SAPK pathway in S. pombe by a downshift in glucose concentration takes place even in the presence of a certain amount of this carbon source.
Different Stress-responsive Genes Are Induced through the SAPK Pathway by Glucose Depletion-Because a decreased glucose concentration fully activates the SAPK pathway in S. pombe, we determined the expression of a set of genes previously described as totally or partially dependent on Sty1 through its main downstream effector Atf1. Cells from glucosegrowing cultures were transferred to the same medium but containing glycerol or sorbitol as the sole carbon source. Samples were taken at different times, and total RNAs were hybridized with probes corresponding to several stress-responsive genes. The expression of the glucose-repressible fbp1 ϩ gene, coding for fructose-1,6-bisphosphatase, which is required to assimilate non-fermentable carbon sources (39), was absent in glucose-growing wild type cells but was sharply induced upon transfer to non-glucose-containing medium (Fig. 2). In agreement with previous studies (29), the SAPK pathway and Atf1 were essential for fbp1 ϩ derepression in response to glucose limitation (Fig. 2). On the other hand, transcription of pyp2 ϩ , which codes for a tyrosine phosphatase involved in the attenuation of Sty1 activity and whose expression has been shown to be regulated by the Atf1 transcription factor (21), was triggered upon glucose deprivation in an Atf1-dependent way (Fig. 2). Similarly, the expression of gpx1 ϩ , which encodes the enzyme glutathione peroxidase involved in the degradation of hydrogen peroxide (37), was induced after transfer of wild type cells to medium devoid of glucose (Fig. 2). This increase fully relied on the presence of Atf1 as reported previously in studies performed under stressing oxidative conditions (37). Finally, the expression of ctt1 ϩ gene, encoding catalase, which is strongly induced under oxidative stress (24), was clearly enhanced in glucose-free medium and virtually undetectable in the absence of Sty1 (data not shown). However, deletion of atf1 ϩ provoked only a partial attenuation in ctt1 ϩ expression by glucose deprivation (Fig. 2), suggesting that other factors are involved in its transcriptional activation. It is worth mentioning that the mRNA steady state levels of other stress-related genes, including gpd1 ϩ (glycerol-3-phosphate dehydrogenase) (15) and ntp1 ϩ (neutral trehalase) (45), greatly increased upon the pas-sage of cells to non-glucose-containing medium (not shown). Taken together, these results show that under conditions of glucose limitation, the SAPK pathway regulates the increased expression of a wide array of stress-responsive genes through Sty1 kinase and its main effector Atf1.
Pap1 Nuclear Accumulation during Transition to Glucose Starvation-In eukaryotic cells glutathione peroxidase and catalase, together with other enzymes like thioredoxin reductase, play a critical role in the removal of intracellular hydrogen peroxide (46). This powerful oxidant results from the enzymatic dismutation of the superoxide radical, which is converted to water and oxygen by catalases and glutathione (47). The observed increase in gpx1 ϩ and ctt1 ϩ expression described above might indicate that the lack of available glucose encompasses an intracellular oxidative stress in S. pombe. In this context, it has been described previously that the hydrogen peroxidemediated expression of the ctt1 ϩ gene relies on transcription factors Atf1 and Pap1 depending on the concentration of the oxidative stimulus (27). Because ctt1 ϩ gene expression under glucose depletion is only partially dependent on Atf1 activity, we considered the possibility that glucose starvation might affect also the function of Pap1, a transcription factor involved in the transcriptional activation of genes involved in the response of S. pombe against oxidative stress at low doses (27). Consequently, we analyzed the cellular location of Pap1 following a change from glucose-to non-glucose-containing medium in strain EHH14, which expresses a genomic copy of an aminoterminal GFP-tagged version of Pap1 under the control of a mid-strength thiamine-repressible promoter (25). As shown in Fig. 3A, GFP-Pap1 fusion was mainly cytoplasmic in S. pombe cells growing in medium with a high glucose content but 3 E. Hidalgo, unpublished results.

FIG. 3. Glucose deprivation induces nuclear accumulation of
Pap1. Strain EHH14 (⌬pap1 with an integrated version of the GFP-Pap1 fusion under the control of the nmt1 promoter) was grown in EMM2 with 7% glucose to an A 600 of 0.5 and recovered by filtration. The cells then either were resuspended in (A) the same medium with either 3% glycerol or 6% sorbitol for 30 min and then shifted back to the same medium plus 6% glucose or were resuspended in (B) EMM2 plus 6% glucose and then shifted to the same medium without a nitrogen source (EMM2-N). Samples were taken at different times, and the cellular distribution of the GFP-Pap1 protein was determined by fluorescence microscopy.
showed preferential accumulation to the cell nucleus soon after removing glucose from the medium. The fluorescence of the fused transcription factor co-localized with the nuclear position as observed by staining with Hoechst (not shown). Nuclear accumulation was clearly evident after 15 min in the absence of glucose and reached a maximum at 30 -60 min (Fig. 3A). Moreover, a change back to glucose-rich medium caused a rapid relocation of Pap1 to the cytoplasm (Fig. 3A). Hence, the transfer from glucose to non-glucose conditions induced a clearly reversible nuclear accumulation of GFP-Pap1 fusion in S. pombe cells. Notably, cells starved for a nitrogen source in the presence of glucose did not show the transient shift of Pap1 (Fig. 3B). This observation supports that the endogenous oxidative stress responsible for the Pap1 shift upon glucose deprivation does not result from cell stasis but from the lack of the fermentable carbon source.
The above results might suggest that GFP-Pap1 nuclear accumulation in S. pombe follows a kinetics similar to that of Sty1 activation under complete absence of glucose. However, SAPK activation takes place also in the presence of low glucose concentrations in osmotically stabilized medium (Fig. 1, C and  F). Thus, we wondered whether the nuclear accumulation of Pap1 was affected by the presence of various amounts of glucose. We examined GFP-Pap1 localization in strain EHH14 exponentially growing in EMM with 7% glucose and then transferred to the same isotonic medium but containing only 0.01, 0.05, or 0.1% glucose. The results shown in Fig. 4 demonstrate that increased amounts of glucose in the medium increase the time required for maximal nuclear accumulation of the GFP-Pap1 fusion. For example, the percentage of cells with nuclear accumulation of the GFP-Pap1 fusion in medium supplemented with 0.01% glucose reached a maximum after 2 h of incubation, whereas the presence in the same medium of 0.1% glucose extended such time to 3.5 h (Fig. 4). Moreover, direct measurement of glucose concentrations during the experiment revealed that GFP-Pap1 nuclear accumulation occurred only after complete exhaustion of this sugar (Fig. 4,  arrows). In contrast, the kinetics of Sty1 phosphorylation was virtually the same in the presence of 0.1% glucose compared with complete glucose deprivation (Fig. 1A). These results indicate a difference in threshold level response by these proteins, with activation of Sty1 preceding nuclear translocation of Pap1 during the transition of S. pombe to non-glucose conditions.

Oxidative Stress and Nuclear Accumulation of Pap1 after
Glucose Removal-Nuclear accumulation of Pap1 mediated by hydrogen peroxide is due to the reversible oxidation of two cysteine residues in the Pap1 amino acid sequence, which prevents interaction with exportin Crm1 (25). To gain insight into the nature of the stress triggered by the absence of glucose, we investigated the nuclear accumulation of GFP-Pap1 during the transition to medium without glucose in strain EHH14.C278A, which expresses a version of GFP-Pap1 with a substitution in cysteine 278 to alanine. This variant GFP-Pap1 cannot accumulate into the nucleus following an oxidative stress induced by external addition of hydrogen peroxide (25). Fig. 5A shows that contrary to control strain EHH14, the mutated version of GFP-Pap1 failed to accumulate into the nucleus in the absence of glucose even at longer incubation times. However, treatment of the same cells with DEM, which induces a non-reversible modification at cysteines 523 and 532 in the Pap1 primary structure (25), provoked a clear nuclear retention of the fusion protein (Fig. 5A). These results strongly suggest that the mechanism responsible for the nuclear accumulation of GFP-Pap1 observed upon removal of glucose is an oxidative stress mediated by hydrogen peroxide.
Glucose deprivation in tumor cells causes an intracellular oxidative stress that is responsible for the activation of the SAPK pathway, and this effect can be avoided when the cells are preincubated in the presence of scavengers for peroxide radicals such as GSH or NAC (2). With these precedents, cells from strain EHH14 were pregrown in minimal medium plus glucose in the presence of 0.16 mM GSH or 30 mM NAC and then transferred to isotonic medium without glucose plus the reducing agents. Under these conditions, GFP-Pap1 fusion protein failed to accumulate in the nucleus (Fig. 5B), indicating that both GSH and NAC are able to effectively inactivate the intracellular peroxide radicals arising from the shift to medium lacking glucose. However, Sty1 displayed a typical pattern of phosphorylation independently in the presence of GSH (Fig.  5C) or NAC (data not shown). This result indicates that in S. pombe the original cause of the SAPK activation observed during the transition from glucose to non-glucose conditions is the absence of glucose itself and not an internal oxidative stress.
Sty1 Regulates Pap1 Nuclear Accumulation in Glucose-deprived Cells-Several studies have shown that the nuclear accumulation of Pap1 induced by hydrogen peroxide is depend- FIG. 4. Nuclear accumulation of GFP-Pap1 occurs only after complete glucose exhaustion. Strain EHH14 (GFP-Pap1 fusion) was grown in EMM2 with 7% glucose to an A 600 of 0.5, recovered by filtration, and resuspended in glycerol-stabilized medium containing different concentrations of glucose. In these experiments, the cells were 10-fold concentrated to shorten the time for glucose disappearance. The time for complete glucose depletion in each experiment is marked by a solid arrow. Samples were taken at different times, and the percentage of cells with nuclear accumulation of the GFP-Pap1 protein was determined by fluorescence microscopy. Results correspond to a representative experiment.

Atf1-and Pap1-mediated Responses during Glucose Depletion
ent on MAPK Sty1 function at high oxidative doses, whereas at low doses (0.07-0.2 mM H 2 O 2 ) it occurs in the absence of Sty1 (27,28). We studied the cellular localization of a GFP-Pap1 fusion protein expressed in the ⌬sty1 strain TK107. As shown in Fig. 6, contrary to wild type cells, deletion of sty1 ϩ inhibited GFP-Pap1 nuclear accumulation after a shift to osmotically stabilized medium without glucose, even in cells maintained for longer incubation times. In contrast, GFP-Pap1 was found in the nucleus when the cells were treated by DEM. This result, together with those shown in Fig. 5, strongly suggest that during the transition to glucose deprivation S. pombe cells undergo a high peroxide stress. Congruent with this interpretation, ctt1 ϩ expression was completely absent in ⌬sty1 cells depleted from glucose (not shown).
Pap1-regulated Gene Expression after Glucose Depletion-In S. pombe cells subjected to mild oxidative conditions, the nuclear accumulation of Pap1 enhances the expression of a number of genes whose products are involved in the oxidative stress response (27). Because the above results indicated that glucose deprivation causes an intracellular stress by peroxide, we determined whether the response included a Pap1-dependent up-regulation of genes coding for enzymes involved in peroxide detoxification. Fig. 7 shows that glucose deprivation was accompanied by a moderate but reproducible increase in the expression of two genes, apt1 ϩ (p25) and trr1 ϩ (thioredoxin reductase), whose expression under oxidative stress specifically relies upon the transcription factor Pap1 (24). Maximal expression for both genes was observed 60 min after glucose starvation and, as expected, was not observed in the pap1 Ϫ mutant strain TP108-3c (Fig. 7), indicating that nuclear accumulation of Pap1 correlates with its transcriptional activity.
As mentioned earlier, the absence of Atf1 transcription factor under glucose deprivation prompted a partial attenuation in ctt1 ϩ expression (Fig. 2). We observed that the increase in ctt1 ϩ expression induced by the absence of glucose was also partially dependent on Pap1 (Fig. 7). Moreover, deletion of both transcription factors prevented ctt1 ϩ induction under glucose starvation (Fig. 7). Thus, in S. pombe both Atf1 and Pap1 transcription factors regulate the expression of genes involved in the response against the endogenous oxidative stress caused by glucose starvation.
Stress Adaptation Induced by Glucose Deprivation-The above results support that in S. pombe glucose deprivation induces an endogenous oxidative stress alleviated by the activ-FIG. 5. Glucose deprivation-induced nuclear accumulation of Pap1 is mediated by a peroxide stress. A, strain EHH14.C278A, which encodes a mutated version of GFP-Pap1 in the cysteine residue 278 of Pap1 (critical for protein oxidation mediated by hydrogen peroxide), was grown in EMM2 with 7% glucose to an A 600 of 0.5, recovered by filtration, and resuspended in the same medium with either 3% glycerol or 6% glucose plus 4 mM DEM. Samples were taken at different times, and GFP-Pap1 distribution was determined in cells by fluorescence microscopy. B, strain EHH14 was grown in EMM2 with 7% glucose to an A 600 of 0.5 in the presence of 0.16 mM GSH or 30 mM NAC and resuspended in the same medium with 3% glycerol. Cellular distribution of the GFP-Pap1 protein was determined by fluorescence microscopy at different times. C, strain JM1521 (Ha6H-tagged chromosomal version of the sty1 ϩ ) was grown in EMM2 plus 7% glucose in the presence of 0.16 mM GSH and transferred to the same medium with 3% glycerol. After purification, activated Sty1 was detected with antiphospho-p38 antibodies. Total Sty1 was detected by immunoblotting with anti-Ha antibody.
FIG. 6. Glucose deprivation-induced nuclear accumulation of Pap1 is dependent on Sty1 activity. Strain TK107 (⌬sty1) was transformed with plasmid p41GFT-Pap1, which expresses an aminoterminal GFP-fused version of Pap1 under the control of the nmt1 promoter (41X; Ref. 24). One transformant was grown in EMM2 plus 7% glucose in the absence of thiamine to an A 600 of 0.5. The cells were recovered by filtration and resuspended in the same medium with either 3% glycerol or 6% glucose plus 4 mM DEM. Cellular distribution of the GFP-Pap1 protein was determined by fluorescence microscopy. ity of both Atf1 and Pap1. When cells are exposed to a nonlethal oxidant dose, they adapt and subsequently tolerate higher levels of oxidant than unexposed cells (27,48). Hence, a logical corollary to these observations should be the demonstration of an increased resistance of glucose-deprived cells against an external oxidative stimulus as compared with non-preadapted cells. We examined cell viability after a treatment with 80 mM hydrogen peroxide for 1 h in wild type, sty1 Ϫ , atf1 Ϫ , pap1 Ϫ , and atf1 Ϫ pap1 Ϫ cultures exponentially growing on glucose and in cultures maintained for 1 h in osmotically stabilized medium without glucose. In the presence of glucose, atf1 Ϫ cells displayed strong sensitivity to the oxidant in comparison with wild type; pap1 Ϫ cells showed a less significant sensitivity, and atf1 Ϫ pap1 Ϫ cells exhibited a dramatic drop in viability (Fig. 8).
These results confirm previous studies indicating that Atf1 plays a mayor role in the defense against high dose oxidative stress in glucose-growing cells (27). However, maintenance in medium without glucose markedly increased the resistance of wild type cells against the oxidative stress (Fig. 8). Significantly, glucose-deprived atf1 Ϫ cells displayed only a moderate increase in viability, whereas pap1 Ϫ cells increased the resistance to hydrogen peroxide, although to a lower extent than wild type cells (Fig. 8). As expected, the atf1 Ϫ pap1 Ϫ strain practically did not increase its tolerance to the oxidative stress under conditions of glucose exhaustion. Taken together, these observations indicate that after glucose depletion, both Atf1 and Pap1 are responsible for the protection of S. pombe cells against an exogenous acute oxidative stress, with Atf1 as the main factor responsible for this adaptive response (Fig. 8).

DISCUSSION
One major finding of this work is the demonstration that the signal transduction pathway that responds to hypoglycemia in S. pombe becomes activated at low levels of glucose prior to the subsequent oxidative stress produced in the absence of this sugar. We have shown that Sty1 phosphorylation takes place with decreased concentrations of glucose even in the presence of thiol antioxidants, which rescue cells from the oxidative stress caused by glucose depletion. In response to glucose exhaustion there is also an increased induction of both Atf1-dependent (ctt1 ϩ , gpx1 ϩ , fbp1 ϩ , and pyp2 ϩ ) and Pap1-dependent (ctt1 ϩ , apt1 ϩ , and trr1 ϩ ) genes, indicating that some change in the induced oxidative stress must occur to explain the functional activity of these two transcription factors. In evaluating our results, we have taken into account that Atf1 primarily controls the transcriptional response to high concentrations of hydrogen peroxide, whereas Pap1 activates target genes in response to low levels of hydrogen peroxide (27). We also have taken into consideration that the expression of ctt1 ϩ is induced in an Atf1/Pap1-dependent manner under glucose depletion. Therefore, our observations support the interpretation that the response to glucose deprivation in S. pombe is accompanied by a sequential endogenous oxidative stress whose intensity signal decreases with time.
Glucose starvation arrests cells of S. pombe at G 2 phase, and they assume a quiescent state because of the lack of a carbon source (49). It was therefore important to determine whether the effects described were due to growth arrest rather than to glucose depletion. Cells grown in medium containing glucose and then shifted to medium without a nitrogen source did not accumulate Pap1 into the nucleus, which suggests that Pap1 translocation is not merely a direct consequence of arrested cell growth. Nuclear accumulation of Pap1 was transitory, but the time length for maximal nuclear accumulation correlated directly with the initial concentration of glucose into the media. These data complement the conclusion that glucose exhaustion is the triggering factor for Pap1 accumulation, which in turn is preceded by activation of the SAPK pathway. Moreover, S. pombe cells expressing a version of Pap1 mutated in a cysteine residue critical for protein oxidation mediated by hydrogen peroxide (25) failed to accumulate Pap1 in the absence of glucose. However, treatment with DEM, which oxidizes Pap1 in different cysteine residues, provoked its nuclear retention. As a whole, these data favor the scenario suggested previously to explain the regulation of Pap1 action during hydrogen peroxide-induced oxidative stress and the delay in the nuclear accumulation of Pap1 by increasing concentrations of oxidant Strains TK003 (wild type (WT)) and TP108-3c (⌬pap1) were grown in EMM2 plus 7% glucose to early log phase and transferred to the same medium without glucose (3% glycerol) for the times indicated. Total RNA was extracted from each sample, and 20 g was resolved in 1.5% agaroseformaldehyde gels. The denatured RNAs were transferred to nylon membranes and hybridized with 32 P-labeled probes of apt1 ϩ , trr1 ϩ , ctt1 ϩ , and leu1 ϩ (loading control). (27). Accordingly, removal of glucose likely promotes a relatively high concentration of hydrogen peroxide inside the cells so that Pap1 assumes an inactive oxidized conformation. Induction of the Sty1-Atf1 pathway and their gene response would be required to allow Pap1 to reach an active oxidized form that locates into the nucleus (27,28). Consistent with this model, both nuclear accumulation of Pap1 and expression of Pap1-dependent genes in response to glucose removal were impaired in sty1 Ϫ cells (Fig. 6). Thus the two signals, glucose exhaustion and oxidative stress, appear to use the same pathway to activate the same (catalase) or a different set of genes through Atf1 and Pap1 transcription factors. In S. cerevisiae, the transcription factor YAP1 (homologue to S. pombe Pap1) is critical for the adaptive response to oxidative stress and accumulates in the nucleus to induce transcription of genes that are important for survival in oxidative environments (33). Interestingly, this localization is similarly sensitive to carbon stress because YAP1 becomes enriched in the nucleus when glucosegrown cells are shifted to medium containing no carbon source (33).
Two critical issues need to be addressed to gain a better understanding of the regulation of the response of S. pombe to glucose deprivation. Firstly, the identity of the upstream elements in the glucose signaling pathway has to be determined, and secondly, the origin of the oxidative effect by glucose deprivation must be outlined. Yeasts appear endowed with fine tuning to extracellular conditions by multiple glucose sensor systems. In S. cerevisiae, for example, the changes induced by glucose limitation in natural diauxic shifts begin well before complete glucose exhaustion, allowing a differential gene response to different thresholds of glucose limitation (50). However, glucose detection systems and associated signal transduction pathways are still under active examination in both budding and fission yeasts (51,52), and the complex responsible for sensing glucose limitation in S. pombe is as yet unknown. In any case, the activation signal for Sty1 phosphorylation following glucose depletion appears to be transduced through the main elements of the SAPK pathway, as Sty1 activation was completely abolished in cells disrupted in the upstream response regulator Mcs4 or MAPK kinase Wis1 (30). 2 The causal factors behind the self-inflicted oxidation in glucose-deprived cells also remain elusive. S. pombe is able to ferment glucose in aerobic conditions, and the major pathways for glucose metabolism include glycolysis, which results in the formation of pyruvate, and the pentose phosphate cycle, which yields NADPH as cellular reducing power. Pyruvate, in addition to its role in energy metabolism, has been shown to scavenge hydrogen peroxide and other hydroperoxides (3). Also, NADPH has been demonstrated to participate in the metabolic decomposition of reactive oxygen species, as this cofactor is a source of reducing equivalents for the glutathione/glutathione peroxidase/glutathione reductase system. Exposure of yeast cells to hydrogen peroxide alters glucose metabolism to favor generation of NADPH, and mutant cells lacking key enzymes of the pentose phosphate pathway show increased sensitivity to oxidizing agents (53). Therefore, fermentative metabolism of glucose provides mechanisms for detoxification of intracellular hydroperoxides that may form under aerobic conditions, and glucose deprivation might lead thus to hydrogen peroxide-induced toxicity. In human cells, prooxidant production during glucose deprivation is most likely due to the metabolism of fatty acids and amino acids via the tricarboxylic acid cycle leading to NADH and FADH 2 as a source of electrons for mitochondrial peroxide formation during respiration (2). However, this is unlikely in S. pombe because mitochondrial functions are glucose-repressed and because cells are unable to carry out de novo synthesis of active mitochondria in the absence of a carbon source. A potential candidate for hydrogen peroxide production might be acyl oxidase involved in peroxisomal ␤-oxidation (54), but studies to support this claim are lacking in S. pombe.
In human carcinoma cells glucose deprivation causes metabolic oxidative stress with high relative levels of intracellular prooxidants (3). Similar independent observations have been made in yeast systems more closely related to S. pombe (55). Several markers for oxidative stress increase during storage of glucose-limited stationary yeast cultures (56), and adaptation of yeasts to glucose depletion has been associated with increased resistance to oxidative stress (57). Also, S. cerevisiae shows a phenotype resistant to oxidative stress after a diauxic shift induced by glucose limitation (50), strongly suggesting a functional connection between carbon stress and oxidative stress. In this context, our results show that cells from glucosedepleted cultures of S. pombe are several orders of magnitude more resistant to an acute oxidative stress than those maintained in the presence of glucose.
In summary, we report that S. pombe exhibits a mechanism induced by low glucose concentration that first activates the SAPK signal transduction pathway to prepare the way for the ensuing oxidative stress that accompanies complete glucose withdrawal. This situation appears to be clearly different from what happens in human tumor cells, where removal of hydroperoxides by NAC eliminates SAPK activation, apparently by suppressing the redox-sensitive activation of MAPK kinase kinase ASK1 and its downstream targets in the signal transduction pathway (34).