The H2O2 Stimulon in Saccharomyces cerevisiae *

The changes in gene expression underlying the yeast adaptive stress response to H2O2were analyzed by comparative two-dimensional gel electrophoresis of total cell proteins. The synthesis of at least 115 proteins is stimulated by H2O2, whereas 52 other proteins are repressed by this treatment. We have identified 71 of the stimulated and 44 of the repressed targets. The kinetics and dose-response parameters of the H2O2 genomic response were also analyzed. Identification of these proteins and their mapping into specific cellular processes give a distinct picture of the way in which yeast cells adapt to oxidative stress. As expected, H2O2-responsive targets include an important number of heat shock proteins and proteins with reactive oxygen intermediate scavenging activities. Exposure to H2O2 also results in a slowdown of protein biosynthetic processes and a stimulation of protein degradation pathways. Finally, the most remarkable result inferred from this study is the resetting of carbohydrate metabolism minutes after the exposure to H2O2. Carbohydrate fluxes are redirected to the regeneration of NADPH at the expense of glycolysis. This study represents the first genome-wide characterization of a H2O2-inducible stimulon in a eukaryote.

. (9,10). Yeast has the same defense mechanisms as higher eukaryotes (for review, see Refs. 11 and 12) and offers the power of genome-wide experimental approaches owing to the availability of the complete sequence of its genome. It therefore represents an ideal eukaryotic model in which to study the cellular redox control and ROI metabolism. We recently established a general method to identify yeast proteins based on two-dimensional gel electrophoresis (13). We used this genome-wide experimental approach to characterize proteins whose expression is altered upon exposure to low doses of H 2 O 2 . Such an oxidative stress challenge results in a dramatic genomic response involving at least 167 proteins. Identification of these proteins and their mapping into cellular processes give a global view of the ubiquitous cellular changes elicited by H 2 O 2 and provides the framework for understanding the mechanisms of cellular redox homeostasis and H 2 O 2 metabolism.
Identification of Protein Spots on Two-dimensional Gels-All the 32 new protein identifications were performed in the S288C strain background. Of these, three were identified by a peptide mass mapping approach using matrix-assisted laser desorption isonization-time of flight/mass spectrometry (16). The remaining 29 were identified by the method described in Maillet et al. (13). This method is based on the determination of the amino acid composition of a given protein by a double amino acid labeling technique with 35 S and 3 H as radioactive markers. The isotopic ratio determined for several pairs of amino acids together with the mass and the pI of a given spot is informative enough to find the corresponding open reading frame in a yeast data base containing 2700 protein sequences of codon bias index Ͼ0.1 with the help of a specific algorithm program. Eight different double labelings were performed here: 35 S-Met/ 3 H-Leu, 35 S-Met/ 3 H-Lys, 35 S-Met/ 3 H-Phe, 35 S-Met/ 3 H-Tyr, 35 S-Met/ 3 H-Trp, 35 S-Met/ 3 H-His, 35 S-Cys/ 3 H-Leu and 35 S-Cys/ 3 H-His. We analyzed 260 spots out of which 124 already identified spots were used as internal standards for the establishment of calibration curves. Six of these 124 proteins devoid of Cys (Ssc1p, Pgi1p, Hsp12p, Hsp82p, Hsc82p, and Cys3p) were used to estimate the metabolic interconversion of Cys to Met which is about 15%. Reciprocally, two proteins devoid of Met (Tpi1p and Tsa1p) were used to estimate the interconversion from Met to Cys which is about 33%. 3 H-Labeled amino acids were chosen among those which are not metabolized into other amino acids in our culture conditions (13). The isotopic ratio obtained for the 124 reference proteins were plotted against their known amino acid ratios taking into account Met and Cys interconversion. Experimental values were in good agreement with the theorical amino acid ratios. The standard deviation of this analysis ranged from 5 to 18%, depending on the double labeling. These calibration curves were used to determine the 8 amino acid ratios for the other 136 proteins analyzed. Mixtures were then extracted and processed for two-dimensional gel electrophoresis as described previously (13). Gels were then stained with Coomassie Brilliant Blue R-250, dried, and exposed to autoradiography. The more abundant protein spots were numbered, identified on the dried gel, and extracted for counting 35 S and 3 H emission in a scintillation counter (WALLAC 1409). The 35 S/ 3 H ratio of each spot normalized to the 35 S/ 3 H Act1p ratio defines the protein synthesis rate index. The ratio of induced to control synthesis rate indexes defines the stimulation (or repression) index. The accuracy of this procedure was demonstrated by comparing two identical uninduced cultures which yielded almost identical results for 400 spots analyzed (S.D. ϭ 0.10). The levels of 111 proteins which differed by a factor of 1.5 or more between uninduced and induced conditions are reported under "Results." The 233 other analyzed protein spots did not significantly change their expression level upon H 2 O 2 treatment. Among them, 110 proteins were identified on our two-dimensional maps: ABP1, ACO 2 , ACT1, ACT2, ADE1, ADH3, ADK1, APA1, ASN2,  ATP14, VMA2, BMH1, BMH2, CCT5, CCT6, CLC1, CMD1, COF1,  COR1, CPR1, CPR6, CPR7, CYS4, DAL1, DYS1, EGD1, EGD2, ERG1,  FBA1, FRS1, FUM1, GLN1, GND1, GRS1, GSP1, HOM2, HOM6,  HSP60, HXK1, HXK2, IDH2, IPP1, KAR2, KRS1, LEU1, LEU2, LPA13,  MAS1, MET25, MRP8, NPL3, NTF2, PFK1, PFK2, PFY1, PGI1, PGK1,  PRE2, PRE6, PRS4, PSA1, PUB1, PYK1, RBK1, RIB4, RPL1, RPLA3,  RPS25A, SBP1, SEC14, SER1, SRP1, SSA4, SSC1, SSE1, SSE2,  SUP45, TFP1, TIF11, TIF45, TOM40, TPI1, TPM1, TRP5, TUB2 Transcripts Quantification by Northern Analysis and RT-PCR-Exponentially growing cells (A 600 , 0.3) were exposed to H 2 O 2 (0.2 mM) at t ϭ 0 and collected for mRNA preparation at t ϭ 0, 15, and 60 min. Cells were disrupted with an Eaton press, and mRNA were extracted with Trizol according to the manufacturer's instructions (Life Technologies, Inc.). mRNA levels were also quantified by a mutiplex RT-PCR strategy based on coamplification of specific internal fragments from selected genes using a combination of primers with similar T m and designed to produce different fragment lengths to be resolved on a 4% acrylamide gel. Poly(A) ϩ mRNA were prepared with the poly(A) tract kit (Amer-sham Promega Biotech). Double-stranded cDNA libraries were generated from poly(A) ϩ messages with the Life Technologies, Inc. Superscript kit. These cDNA libraries were used in an RT-PCR reaction with appropriate primers and [ 33 P]dCTP. The labeled RT-PCR products were then resolved by electrophoresis. Gels were dried, exposed to autoradiography, and quantified by phosphor technology (PhosphorImager, Molecular Dynamics). ACT1 RT-PCR products were used as internal standard. All these PCR reactions were in the linear range as attested by comparison to preliminary calibration experiments.

RESULTS
The S. cerevisiae Genomic Response to H 2 O 2 -The exposure of exponentially growing cells to low doses of H 2 O 2 results in dramatic changes in protein synthesis. To characterize this oxidative stress genomic response, exponentially growing cells were treated with 0.4 mM H 2 O 2 for 15 min and pulse-labeled with [ 35 S]methionine. Labeled extracts from control untreated and treated cells were then subjected to comparative two-dimensional gel electrophoresis (Fig. 1). Changes in the intensity of a number of spots could be recognized by simple visual inspection. We sought to precisely measure these genomic changes. Usual methods rely on the comparative quantification of corresponding 35 S-labeled protein spots with a PhosphorImager. However, this method is imprecise at least in part due to differences in the yields of extraction of each protein. To improve this method, we defined an internal protein concentration standard for each spot by adding an equal aliquot of 3 H-labeled cells to the 35 S-labeled cultures. This allows one to express any change in the rate of protein synthesis as the ratio between the 35 S/ 3 H ratios of corresponding spots from two different gels and hence to correct for any variation not related to the H 2 O 2 treatment (see "Materials and Methods"). Preliminary experiments comparing two-dimensional gels from two identical cultures showed that the variations observed in the protein quantitation were below a factor of 1.3 for 96% of the 400 control spots analyzed and never exceeded 1.5 (Fig. 2). Therefore, differences by a factor greater than 1.5 between treated and control cultures were considered to be significant. Accordingly, 115 proteins were specifically induced by H 2 O 2 with a stimulation index ranging from 1.5 to 20 (see Table I). Conversely, 52 other proteins were repressed with a repression index ranging from 0.65 to 0.15 (Table I). Almost identical results were observed with two related yeast strains S288C and YPH98.
Proteins Induced by H 2 O 2 -The identity of 71 of the 115 proteins induced by H 2 O 2 is given in Table I along with their stimulation index. 39 of them were previously identified on two-dimensional maps (for review, see Ref. 17). The 32 other spots were identified in this work by amino acid analysis as described by Maillet et al. (13) or by mass spectrometry (16). H 2 O 2 -responsive proteins were sorted into seven different functional classes (see Table I). (i) Proteins directly related to the cellular antioxidant defense: This class shows a high stimulation index which ranges from 3 to 20 depending on the protein. It comprises the major oxidant scavenging enzymes cytochrome c peroxidase (Ccp1p), cytosolic catalase (Ctt1p), Cu/Zn and Mn superoxide dismutases (Sod1p and Sod2p), thioperoxidase (Tsa1p), thioredoxin (Trx1p or Trx2p), NADPH-dependent thioredoxin reductase (Trr1p), and glutathione reductase (Glr1p). Four newly identified proteins were included in this functional class on the basis of their homology to known oxidant scavenging enzymes and/or suspected antioxidant defense properties. YDR453Cp is one of the two other AhpC/TSA family members identified in the yeast genome. YCL035Cp is 86% similar and 68% identical to TTR1-encoded glutaredoxin (thioltransferase). YLR109Wp and YOL151Wp were classified here on the basis of their role in the tolerance to tert-butyl hydroperoxide and diamide, respectively (see "Discussion"). (ii) Heat shock proteins: This class also shows a strong stimulation index for several of its members. (iii) Proteases and proteasome subunits: The stimulation index is significant but not as high as in the two previous classes. Cdc37p has been shown to be a chaperone acting with Hsp90p and other chaperones to promote the folding/activity of a series of kinases (19).
Proteins Repressed by H 2 O 2 -44 out of these 52 proteins had been previously identified on two-dimensional maps (17). Each was assigned a repression factor and sorted into functional classes ( Table I). Most of these proteins are translational apparatus components and metabolic enzymes. They include the translation initiation factors eIF4A (Tif1p) and eIF5A (Tif51Ap) and the translation elongation factors EF1-␤ (Efb1p), EF-2 (Eft1p), and EF-3 (Yef3p), which are dramatically repressed. Rpa0p, Rpa2p, Rpa4p, and Rps5p are acidic ribosomal proteins that act both at the initiation and the elongation steps (20). Ssb1p and Ssb2p are heat shock proteins of the 70-kDa superfamily that are ribosomal-associated and have a role in the folding of nascent polypeptides emanating from the ribosome (21,22). Metabolic enzymes repressed by H 2 O 2 include enzymes involved in glycolysis, the Krebs cycle, purine and amino acid biosynthesis, sulfur metabolism, S-adenosylmethionine, and polyamine biosynthesis. Ilv2p(a) (acetolactate synthase) appears repressed by a factor of 0.35. Interestingly, we could identify a H 2 O 2 -responsive spot of pI 6.7 and M r 67,000 with a very good match with Ilv2p with regard to its amino acid composition, and therefore it is indicated as Ilv2p(b) in Fig. 1  (13 electrophoresis was then performed to determine the relative rate of synthesis of several proteins. Except for a few targets, the H 2 O 2 response was very rapid and transient. We could define three kinetic classes (Fig. 3A). Proteins of class A responded as early as 2 min after induction with a peak at approximately 15 min and a complete return to the base line after 1 h. Proteins of class B had a very similar kinetic profile but initiated their response with a lag period of at least 4 min. Class c proteins had a somewhat different kinetic profile with a relatively delayed response and a peak at 45 min or even at 1 h for Uba1p. Repression by H 2 O 2 was similarly transient with a nadir at approximately 15 min after stress imposition.
We also tested the H 2 O 2 dose-genomic response profile (Fig.  4). Synthesis rates were determined 15 min after exposure to 0.2, 0.4, or 0.8 mM H 2 O 2 for 61 H 2 O 2 -responsive targets. Trr1p and several other proteins were equally induced by each of the three H 2 O 2 doses tested (Fig. 4). Interestingly, Ccp1p and several other proteins were maximally induced by 0.2 mM H 2 O 2 . In contrast, most of the heat shock proteins exhibited their maximal response at 0.8 mM. These kinetics and doseresponse differences may be related to distinct regulatory mechanisms.
Alterations  Fig. 5. Levels of all the transcripts analyzed were increased by 5-37-fold at 15 min after H 2 O 2 exposure and had returned close to their basal levels at 60 min. We also analyzed by Northern blot the kinetic profile of the TRR1 message levels at several points after H 2 O 2 treatment (Fig. 3B). The kinetics of the message levels and protein synthesis rates are strikingly parallel after H 2 O 2 treatment. These data corroborate those obtained from the two-dimensional gel analysis and strongly suggest that the dramatic genomic response to H 2 O 2 involves, at least in part, a transcriptional control. DISCUSSION ROI are obligate by-products of aerobic life which can inflict structural damage to a wide variety of cell components, thus leading to oxidative stress and cell death. Stress-inducible defense or adaptive response mechanisms act to protect cells from these oxidative threats (2,4,5). For instance, the exposure of bacteria or yeast to low levels of H 2 O 2 -or O 2 . -generating drugs switches on within minutes a resistance to toxic doses of these oxidants. These adaptative stress responses are produced by the induction of distinct batteries of genes or stimulons. However, the genes which constitute these stimulons are, for the most part, not yet identified. We have attempted here a systematic identification of the gene products of the S. cerevisiae   redox center. YOL151W is similar to NADPH dihydroflavonoid reductases involved in the plant synthesis of isoflavonoid phytoalexins. The antioxidant defense properties of these reductases were demonstrated by the isolation of an isoflavonoid reductase gene from an Arabidopsis thaliana cDNA library in a search for activities able to rescue the diamide hypersensitivity phenotype of a yeast strain deleted for the oxidative stress response regulator YAP1 (33). YBR149W is related to aldo/keto reductases and may act as an NADPH-dependent aldehyde reductase to scavenge lipid peroxidation-derived toxic aldehydes by their reduction into alcohols (34,35). However, YBR149W could also act as an NADP ϩ -dependent glycerol dehydrogenase (see below). Although its product could not be detected in our two-dimensional gels maps, GSH1 (␥-glutamylcysteine synthase) mRNA levels are dramatically increased by H 2 O 2 (data not shown) and is therefore a part of the H 2 O 2 stimulon.
Heat Shock Proteins, Proteases, and the Translation Apparatus-Twelve heat shock proteins (HSP) as well as proline isomerase (Cpr3p) and protein disulfide isomerase subunits of the proteasome along with enzymes of the ubiquitin pathway, mitochondrial and lysosomal proteases is also consistent with an important proteolytic activity during the oxidative stress response. Induction of the ubiquitin pathway by oxidative stress and the specific degradation of oxidized proteins by the proteasome has been recently demonstrated (38,39). Most of the chaperones and proteases have essential roles under nonstress conditions by assisting protein biogenesis, oligomer assembly, traffic between cellular organelles, and selective protein degradation (40,41). Hence, in addition to their protective functions, they may help to reorchestrate the cell metabolism to the needs of the oxidative stress response. Associated with these changes, the repression of two translation initiation and four translation elongation indicates a global and nonspecific slowdown of protein translation. We could indeed demonstrate a 2.5-fold decrease of translation in response to 0.3 mM H 2 O 2 by [ 14 C]leucine labeling (data not shown). Taken together, the response of HSPs, proteases, and the translational apparatus to H 2 O 2 is probably important for switching the cellular activity from biosynthetic toward protective functions.
Carbohydrate Metabolism and NADPH Regeneration-Twenty-five H 2 O 2 -responsive targets were identified as metabolic enzymes. Although not exhaustive, this identification provides an indication of the metabolic fluxes redistribution occurring in response to H 2 O 2 . These changes dramatically affect carbohydrate metabolism which appears to be diverted to the generation of NADPH, the most important cellular reducing power (Fig. 6). (i) The hexose monophosphate pool: Repression of phosphomannomutase (Sec53p), stimulation of phosphoglucomutase (Pgm2p), and exclusion of glucose from glycolysis (see below) seem to redirect the hexose phosphate pool to the pentose phosphate pathway and the trehalose syn-thesis. (ii) Induction of trehalose synthesis: Trehalose synthesis is stimulated by oxidative stress as indicated by the induction of phosphoglucomutase (Pgm2p), UDP-glucose pyrophosphorylase (Ugp1p), and trehalose-6-phosphate synthase (Tps1p). Trehalose synthesis is also stimulated by heat shock and osmotic stress (42)(43)(44), and its accumulation correlates with thermotolerance (45). Parrou et al. (46) also observed the induction of TPS1 by H 2 O 2 but curiously without any trehalose accumulation. We also could not detect any change in trehalose steady state and synthesis rate levels by [ 14 C]glucose labeling (data not shown). These data suggest the existence of an enhanced recycling of this disaccharide (46,47). (iii) Induction of the pentose phosphate pathway: Three enzymes of the pentose phosphate pathway are induced by H 2 O 2 . Glucose-6-phosphate dehydrogenase (Zwf1p) regulates the carbon flow through this pathway by catalyzing its first step, leading to ribulose 5-phosphate, the precursor of purine biosynthesis (48). Then, pentose phosphates are interconverted to glyceraldehyde 3-phosphate or fructose 6-phosphate by transketolases (Tkl1p and Tkl2p) and transaldolase (Tal1p). Glyceraldehyde 3-phosphate can enter glycolysis and fructose 6-phosphate is converted to glucose 6-phosphate (49). However, repression of glycolysis (see below) and purine biosynthesis pathway suggest that most of the pentose phosphates are recycled to the hexose phosphate pool for NADPH production. (iv) Repression of glycolysis: H 2 O 2 treatment results in a slowdown of glycolysis as manifested by repression of Tdh2p and Tdh3p, and the isozymes of both enolase and pyruvate decarboxylase. (v) Repression of the tricarboxylic acid cycle: The decreased expression of pyruvate decarboxylase and pyruvate dehydrogenase, the two enzymes which catalyze the alternative entries into the trichloroacetic acid cycle suggests a further slowdown of the trichloroacetic acid cycle, which is already subject to catabolite repression. Repression of malate dehydrogenase (Mdh1p) is also consistent with this notion. (vi) Alteration of glycerol metabolism: Glycerol synthesis must be increased in response to H 2 O 2 as suggested by the induction of glycerol phosphate dehydrogenase (Gpd1p) and glycerol phosphate phosphatase (Gpp2p) ( Table I). In addition, Dak1p, a dihydroxyacetone kinase and YBR149Wp, a putative glycerol dehydrogenase, which have been assigned to a new salt-induced glycerol dissimilation pathway (50) are also induced by H 2 O 2 . This glycerol cycle composed of Gpd1p, Gpp2p, YBR149Wp, and Dak1p (Fig. 6) may function as a transhydrogenase activity to convert NADH to NADPH at the expense of one ATP (50).
In conclusion, the carbohydrate metabolism alteration seems to principally concur to the regeneration of NADPH. NADPH is important in the oxidative stress response as a cofactor for glutathione reductase and thioredoxin reductase, two critical activities in the cellular thiol redox control and antioxidant defense (28,29,52). The critical role of NADPH is also suggested by the H 2 O 2 hypersensitivity of strains mutated for any of the six enzymes of the pentose phosphate pathway (51,53) and by the capacity of TKL1 overexpression to suppress the oxygen sensitivity of a sod1 null mutant (54). Studies with cells carrying a genetic ablation of the pentose phosphate pathway have also suggested that other cellular mechanisms of NADPH production must exist (51,53,55). The glycerol dissimilation pathway might be such a mechanism of NADPH regeneration.
Large Scale Analysis of Gene Expression-We have analyzed the global changes in protein expression underlying the yeast adaptive stress response to H 2 O 2 by a two-dimensional gel approach. Although very informative, this analysis is limited by the fact that only soluble and abundant proteins are seen on two-dimensional gels. Membrane-bound proteins or those with a very low or very high molecular weight or with a pI higher than 7.5 are not observed. Based on the expression of about 4500 -5000 genes under normal growth conditions (56) and on the detection of about 1000 two-dimensional gel spots, we estimate to have covered about 20% of all expressed proteins. Brown and colleagues (57) have described a new method for the monitoring of gene expression on a genomic scale with a DNA microarray technique which covers the entire yeast genome. This approach monitors changes in mRNA levels for the entire yeast genome, but cannot reveal translational or post-translational control mechanisms. Furthermore, highly homologous isogenes are more readily differentiated by their protein products than by nucleic acid hybridization pattern. Although we observed the concordant alteration in the expression of several isogenes (Sod1p/Sod2p, Tsa1p/YDR453, Ssb1p/Ssb2p, Sam1p/ Sam2p, we also observed that several other isogenes were differentially expressed during the oxidative stress response (Tkl1p/Tkl2p, Ald5p/Ald6p, Ssa1p/Ssa2p, Hsc82p/Hsp82p, Gpp1p/Gpp2p, Tfs1p/YLR179C). Such differential expression of specific isozymes has also been observed with a two-dimensional gel approach by Norbeck and Blomberg (50), but its physiological meaning is not understood. In summary, we think that the power of the DNA microarray technique in the genomic scale analysis of gene expression should be complemented by the more limited but more biochemical two-dimensional gel approach that allows visualization of post-transcriptional modifications.
Perspectives-The rapid and widespread genomic response to H 2 O 2 seen here must involve several inducible control mechanisms. These mechanisms could be, at least in part, transcriptional as suggested by the good correlation between mRNA and protein levels for several of the targets analyzed. Our experimental system provides us a unique opportunity to identify the genetic circuitry that regulates and executes the adaptive re-sponse to H 2 O 2 . Several regulators suspected to play a role in the yeast H 2 O 2 response include Yap1p and Skn7p (30, 58 -60), Msn2p/4p (61), Hsf1p (62), and Gcn4p (63). The assignment of their respective targets to each of these regulators will help to define the different regulons involved in the oxidative stress response.