A Proteome Analysis of the Cadmium Response in Saccharomyces cerevisiae *

Cadmium is very toxic at low concentrations, but the basis for its toxicity is not clearly understood. We analyzed the proteomic response of yeast cells to acute cadmium stress and identified 54 induced and 43 repressed proteins. A striking result is the strong induction of 9 enzymes of the sulfur amino acid biosynthetic pathway. Accordingly, we observed that glutathione synthesis is strongly increased in response to cadmium treatment. Several proteins with antioxidant properties were also induced. The induction of nine proteins is dependent upon the transactivator Yap1p, consistent with the cadmium hypersensitive phenotype of the YAP1-disrupted strain. Most of these proteins are also overexpressed in a strain overexpressing Yap1p, a result that correlates with the cadmium hyper-resistant phenotype of this strain. Two of these Yap1p-dependent proteins, thioredoxin and thioredoxin reductase, play an important role in cadmium tolerance because strains lacking the corresponding genes are hypersensitive to this metal. Altogether, our data indicate that the two cellular thiol redox systems, glutathione and thioredoxin, are essential for cellular defense against cadmium.

Heavy metals represent major environmental hazards to human health. In particular, cadmium is very toxic and probably carcinogenic at low concentrations. However, the biological effects of this metal and the mechanism of its toxicity are not yet clearly understood. It has been proposed that Cd 2ϩ ions might displace Zn 2ϩ and Fe 2ϩ in proteins (1), resulting in their inactivation and in the release of free iron, which might generate highly reactive hydroxyl radicals (OH ⅐ ) (2). In support of this hypothesis, a major effect of cadmium is oxidative stress (3), particularly lipid peroxidation (1). However, it is not known whether these effects are responsible for the extreme toxicity of the metal.
Living organisms use several mechanisms to counter cadmium toxicity. In bacteria, efflux pumps are able to export toxic ions outside the cell (4). In higher eukaryotes, Cd 2ϩ is sequestered by metallothioneins through their high cysteine content (5). Cadmium can also be detoxified by chelation to GSH or to phytochelatin, a glutathione polymer of general structure (␥-Glu-Cys) n -Gly synthesized from GSH in plants and in the yeast Schizosaccharomyces pombe. Cd 2ϩ -phytochelatin and Cd 2ϩ ⅐ (GSH) 2 complexes are transported into the vacuole by ATPbinding cassette transporters (6 -8).
Yap1p and Skn7p are yeast transcription factors that regulate the adaptive response to oxidative stress (9 -11). Strains lacking either transcription factor are sensitive to H 2 O 2 and are defective in the induction by H 2 O 2 of several enzymes with antioxidant properties (9). Yap1p is also important in cadmium tolerance because yap1-deleted strains are very sensitive to cadmium, and strains overexpressing YAP1 are hyper-resistant to this toxic metal (12). The contribution of Skn7p to the cadmium response is more complex, because skn7-deleted strains are hyper-resistant to cadmium (9), suggesting that Skn7p is not only dispensable for this response but might also repress genes that are important for cadmium tolerance.
Using two-dimensional gel electrophoresis, we have identified several proteins induced by cadmium in Saccharomyces cerevisiae, providing a framework to the mechanisms of Cd 2ϩ toxicity and cellular protection against this toxic metal. In particular, our results highlight the importance of both glutathione and of the thioredoxin/thioredoxin reductase system in the cellular defense against cadmium.
Survival Assays-Aliquots from an exponential growth culture (A 600 , 0.5) were serially diluted in water (10-fold at each step). Ten l were spotted onto rich broth (YPD) plates containing different cadmium sulfate concentrations: 0, 25, 50, 75, 100, 150 and 200 M. Plates were incubated at 30°C for 2 to 5 days, and the colony-forming units were counted. 35 S Labeling-Ten-ml cultures were inoculated with a colony and cultivated overnight aerobically at 30°C with shaking. Two-ml aliquots of mid-log phase culture (A 600 , 0.4) were withdrawn and treated with 0, 0.5, 1, or 1.5 mM cadmium sulfate. 15   For analysis, the metabolites were oxidized by adding equal volume of supernatant and performic acid (16). This treatment converts both oxidized and reduced forms of glutathione to glutathione-sulfonic acid. Samples (0.1 l) were then applied on cellulose thin layer chromatography in the following solvent system: butanol-1/acetic acid/water (90: 15:33). The 35 S-labeled metabolites were quantified by phosphor technology (Phosphor-Imager, Molecular Dynamics).
Analysis of Protein Expression-Protein extraction and two-dimensional gel electrophoresis were performed as previously described (17) with a Millipore Investigator apparatus. The radioactive gels were recorded by PhosphorImager and were analyzed with two-dimensional gel analysis software (Melanie II, Bio-Rad). The spot intensities were obtained in pixel units and normalized to the total radioactivity of the gel. The cadmium stimulation index was calculated as the ratio of spot intensity between cadmium and standard conditions. Proteins with a stimulation index of higher than 1.8 are reported in Table I. Proteins with a stimulation index lower than 0.60 (repressed proteins) are: ACS2, ADH1, ADH2, ADK1, ALD6, ARG1, BEL1, COF1, CPA2, EFB1, EFT1, EGD1, EGD2, FBA1, GDH1, GLN1, GPP1, HIS4, HOM2, ILV2, ILV3, ILV5, KRS1, LYS9, LYS20, PDC1, PUB1, PRO2, RPA0, RPA2, RPL45, RPS5, SSB1, SSB2, TDH2, TIF1, TIF45, TIF51A, TPI1, VMA1, YEF3, YJL200C, and YKL056C. The other analyzed proteins with stimulation indices lower than 1.8 and higher than 0.60 are not reported in Table I (with the exception of CTT1, TRR1, and SOD1). Mutant and wild-type strains were compared by their ratio of spot intensities under different cadmium conditions. The experiments were performed at least twice with similar results, and the mean value was reported here. The S.D. of the analysis ranged from 20 to 25%.
Identification of Protein Spots on Two-dimensional Gels-Proteins induced by cadmium were all identified by matching two-dimensional maps of cadmium-treated cell extracts with a reference gel containing more than 450 previously identified proteins (17,18,19) with the exception of 3 new spots. These previously uncharacterized proteins were identified by a peptide mass mapping approach using matrixassisted laser desorption ionization-time of flight mass spectrometry (20).
Gene Disruptions-Standard protocols and buffers were used (21). Strains gre2⌬::TRP1 and gpx3⌬::TRP1 were prepared by a one step PCR 2 amplification protocol that replaced the entire GRE2 or GPX3 open-reading frame by the TRP1 gene (22). The mutant yap1⌬ was created by a one-step gene disruption technique (23), which removed the YAP1 coding sequence from the BamHI site (ϩ186) to the KpnI site (ϩ1650) relative to the ATG and replaced it with the TRP1 gene.
The ahp1⌬::TRP1 strain used in the present work was generated by one back-cross of the previously described ahp1⌬::TRP1 (14) with a wild-type strain.
RNA Analysis-Quantitative reverse transcriptase-PCR analysis was performed essentially as previously described (19). ACT1 reverse transcriptase-PCR products were used as internal standards.

RESULTS
The Yeast Genomic Response to Cadmium-We analyzed the proteomic response to cadmium to identify activities, which could be potentially important for the resistance to this toxic compound. Exponentially growing cells were untreated or treated with 1 mM cadmium for 60 min, pulse-labeled with [ 35 S]methionine, and analyzed by comparative two-dimensional gel electrophoresis (Fig. 1). The changes in spot intensity between untreated and treated cells were quantified by phosphorimager and software analysis (see "Materials and Methods"). More than 50 proteins were induced by a factor greater than 2 after cadmium treatment (Table I). Concomitantly, about 40 other proteins were significantly repressed. A lower dose of cadmium (0.5 mM) or shorter period of treatment (15 min) gave a similar pattern of protein expression, but with lower induction levels. Proteins induced by cadmium included enzymes with antioxidant properties, heat shock proteins, proteases, enzymes of the sulfur amino acid biosynthesis pathway, carbohydrate metabolism enzymes, and other unclassified proteins or with unknown function. Conversely, proteins repressed by cadmium were mainly components of the translational apparatus and metabolic enzymes.
Induction of Glutathione Synthesis-The particularly high induction level of enzymes of the sulfur amino acid pathway ( Fig. 2) suggested an increased synthesis of cysteine and perhaps of GSH, which is essential in the cellular detoxification of cadmium. To test this hypothesis, we directly measured GSH synthesis rate by TLC after [ 35 S]methionine pulse labeling. After cadmium treatment, the synthesis rate of GSH was increased more than 4-fold (Fig. 3A, lanes 1 and 2). As expected, in extracts from cells lacking GSH1 (encoding the rate-limiting enzyme of GSH biosynthesis), no GSH was produced (Fig. 3A,  lanes 3 and 4). The production of another 35 S-labeled compound was also augmented in response to cadmium treatment, but we could not identify this molecule.
The stimulation of GSH biosynthesis correlated with a more than 10-fold increase of GSH1 mRNA level after 1 h of cadmium treatment (not shown and Ref. 24). As metallothioneins are cysteine-rich and potentially important for cadmium detoxification, we also analyzed CUP1 mRNA levels and found no increased expression of this gene in response to cadmium, but rather a slight repression. These results strongly suggest that the cysteine biosynthesis pathway was induced to allow enhanced glutathione synthesis.
The Cadmium Response Is Altered in Regulatory Mutants-Strains lacking the transcriptional activator Yap1p are very sensitive to cadmium and conversely, strains overexpressing this regulator are hyper-resistant (12). However, strains lacking the transcriptional activator Skn7p are hyper-resistant to 2 The abbreviations used are: PCR, polymerase chain reaction, WT, wild-type.
FIG. 1. Two-dimensional gel electrophoresis analysis of total yeast proteins expressed in response to cadmium. Autoradiograms of two-dimensional gel electrophoresis of total yeast extracts from [ 35 S]methionine-labeled wild-type cells. 35 S labeling and two-dimensional gel electrophoresis were performed as described under " Materials and Methods." Extracts were prepared from control untreated cells (A) or cells treated with 0.5 mM cadmium for 15 min (B) or for 1 h (C) or treated with 1.5 mM cadmium for 1 h (D). A central region of the autoradiograms was blown up, and the induced proteins are indicated with black arrows.
this toxic metal (9). We therefore analyzed the cadmium proteomic response in strains lacking either YAP1 (yap1⌬) or SKN7 (skn7⌬) or overexpressing YAP1 (mcYAP1) (Fig. 4). The comparative measure of the proteomic response between these mutants and the wild-type cells is given in Table I. The yap1⌬ strain had a weaker induction of 9 of the 57 responsive proteins, by a factor of 2 or more. These included four antioxidant defense proteins (the superoxide dismutase Sod2p, the alkyl hydroperoxidase Ahp1p, the thioredoxin Trx2p, and the thioperoxidase Tsa1p) and five other proteins (Sam1p, Oye3p, Gre2p, YNL134C, and YNL274C). Except for Sam1p, all these proteins were significantly overexpressed in response to cadmium in mcYAP1. Additionally, most of the heat shock proteins and proteasome subunits were also overexpressed in mcYAP1 (Table I). In skn7⌬, only two proteins (Cdc48p and Shm2p) were not properly induced by cadmium. However, surprisingly, a The stimulation index is the ratio of protein expression in cadmium versus standard conditions. b Results for Ctt1p, Sod1p, and Trr1p were reported though these enzymes were not induced by cadmium. c Proteins identified in this work by mass spectrometry. d Hsp26p does not contain methionine or cysteine residues, and the induction was qualitatively observed by Coomassie Blue staining. e Gsh1p has not been identified on two-dimensional gels, and the induction was measured at the mRNA level by RT-PCR analysis. f nd, not determined.
Ctt1p, Ccp1p, Sod1p, Oye3p, Gre2p, YNL134C, and several heat shock proteins and proteasome subunits were overexpressed in response to cadmium. Given the importance of GSH in cellular cadmium protection, we also measured the biosynthesis of GSH in these mutants as we did in the wild-type strain. The GSH biosynthesis rate was significantly decreased in yap1⌬ and augmented in mcYAP1 (Fig. 3B, lanes 5-7). However, GSH biosynthesis rate was not modified in skn7⌬.
The Thioredoxin/Thioredoxin Reductase System Is Essential for Cadmium Tolerance-Proteins with both a defective cadmium induction in yap1⌬ and an increased expression in mcYAP1 could be important for Cd 2ϩ protection. 5 of 9 proteins matching this criteria were therefore evaluated for their role in Cd 2ϩ tolerance (Fig. 5A). Deletion of TRR1 encoding the thioredoxin reductase or of both TRX1 and TRX2 resulted in a significant decreased tolerance to cadmium (Fig. 5). However, the deletion of either thioredoxin does not affect the tolerance to cadmium, showing that these two proteins can substitute for each other in this function. In contrast, strains carrying a deletion of either GRE2, TSA1, or AHP1 had a wild-type tolerance to cadmium (data not shown). Note that the ahp1⌬ strain has a wild-type tolerance contradicting an earlier report (14) that demonstrates ahp1⌬ is sensitive to this metal. However, the cadmium phenotype of this strain was shown related to an additional unidentified mutation (data not shown).
One of the main Cd 2ϩ toxic effects is believed to be the generation of toxic lipid peroxides. It was therefore surprising to find that the strain deleted of AHP1, the gene encoding an important alkyl hydroperoxide reductase, is not defective in cadmium tolerance. Because another enzyme, the glutathione peroxidase Gpx3p, is also active against organic peroxides (25), we analyzed the cadmium resistance of the GPX3-disrupted strain. Although this mutant is clearly sensitive to tert-butyl hydroperoxide (t-BOOH, Ref. 25 and data not shown), it showed a normal cadmium resistance (Fig. 5). DISCUSSION To gain insight into the biological effects of Cd 2ϩ and to identify activities relevant to its detoxification, we have analyzed the proteomic response to this toxic metal. Fifty-seven proteins were found to be induced by Cd 2ϩ , including enzymes of the cysteine and glutathione biosynthesis pathway and proteins with antioxidant properties. In addition, the induction of several of these proteins is controlled by the transcription factor Yap1p.
Cadmium Stimulates the Biosynthesis of Cysteine and Glutathione-The proteomic response to cadmium has revealed a strong induction of eight enzymes of the sulfur amino acid and GSH biosynthesis pathways, particularly the last enzyme of the cysteine biosynthesis pathway, Cys3p, and the rate-limiting enzyme of glutathione biosynthesis, GSH1. This increased gene expression was correlated with a strong stimulation of GSH biosynthesis.
These results are consistent with the notion that GSH acts as a first line of defense against cadmium toxicity by chelation and sequestration of the toxic metal. In yeast, cadmium is sequestered in the vacuole upon transport of Cd 2ϩ ⅐(GSH) 2 complexes by the membrane ATP-binding cassette Ycf1 (6). The importance of this detoxification system is demonstrated by the hypersensitive phenotype resulting from the deletion of either YCF1 (26) or GSH1 (27). The cysteine biosynthetic pathway is also important as demonstrated by the cadmium hypersensitive phenotype of strains deleted for CYS4 (data not shown). This pathway is probably essential for the increased production of GSH under these stress conditions.
The coordinated regulation of GSH1 expression with enzymes of the sulfur amino acids pathway may suggest that a common transcriptional mechanism is involved in these inductions. Consistent with this idea, a recent work (28) has shown that the GSH1 induction by cadmium is dependent upon the transcription factors Met4p, Met31p, Met32p, and Cbf1p, which belong to the transcriptional complex of MET genes. It was also found that the GSH1 promotor contains functional elements typical of MET genes (28).
Interestingly, the sulfur amino acid pathway is not induced but rather repressed under most of the other stress conditions examined, including the oxidative stress (19), the osmotic stress (29), and the heat shock response (30). Furthermore, it is remarkable that among the enzymes involved in amino acid metabolism, only those of the sulfur amino acid biosynthetic pathway are induced, which is consistent with a very specific control mechanism.
Does Cadmium Cause Oxidative Stress?-Cd 2ϩ is not a redox active metal ion. However, it could cause oxidative stress and lipid peroxidation, perhaps by displacing protein-bound Fe 2ϩ , allowing this ion to become available for the Fenton reaction. This hypothesis is based on previous studies showing an increase in lipid peroxidation products after exposure to Cd 2ϩ (1,(31)(32)(33)(34). In addition, yeast strains deleted for cytosolic copper and zinc and mitochondrial manganese superoxide dismutases (SOD1 and SOD2) are hypersensitive to cadmium (3). The observation that several antioxidants such as Ahp1p, Ccp1p, Tsa1p, and Sod2p are induced by cadmium is consistent with this hypothesis. We also found that strains deleted for AHP1 or GPX3, which encode two main organic peroxide-scavenging activities, are not sensitive to cadmium, which does not support the idea that cadmium toxicity is related to the cellular generation of lipid peroxidation products. However, we cannot rigorously rule out the possibility that yeast tolerance to cadmiumgenerated lipid peroxidation products involves specific activities other than Ahp1p or Gpx3p, which have not yet been elucidated.
It is also possible that cadmium indirectly contributes to oxidative stress by affecting the cellular thiol redox balance. We indeed found that Trx2p is significantly induced by cadmium and that strains deleted for both TRX1 and TRX2 or for TRR1 are hypersensitive to this metal. In support of such a mechanism, a recent work has shown that cadmium inactivates thioredoxins, thioredoxin reductases, and glutathione reductases (35). Therefore, given the essential nature of these systems (36), thiol transferase inactivation could be the primary deleterious effect of cadmium. In addition, the inactivation of the thiol transferase pathway could also result in an increase in cellular lipid peroxidation products as a consequence of their unavailability of the main electron donors for the thiol and GSH peroxidases.
Yap1p Target Genes-The observation that YAP1-deleted strains were hypersensitive to cadmium (9, 12) and strains overexpressing YAP1 were hyper-resistant (27), suggested that this transcriptional activator might control genes important for FIG. 4. Comparative analysis of cadmium response in regulatory mutant strains. A, autoradiograms of two-dimensional gel electrophoresis of total yeast proteins from [ 35 S]methionine-labeled strains treated with 1 mM cadmium for 1 h. Extracts were prepared from wild-type (A1), yap1⌬ (A2), skn7⌬ (A3), and YAP1 multicopy strains (A4). A region of the autoradiograms was blown up, and the proteins that were no longer induced in the yap1⌬ strain are indicated by black arrows. LEU2 is the marker of the plasmid that overexpresses YAP1. B, histogram representation of noninduced (white bar) and Cd 2ϩ induced (black bar) synthesis rate indexes calculated in yap1⌬ (bar 1), wild-type (bar 2), YAP1-overexpressing strain (bar 3) and skn7⌬ (bar 4) cells as described under "Materials and Methods." For each protein spot, values were normalized to the wild-type noninduced level that was arbitrarily given the value of 1. The protein names are indicated above the histograms.
FIG . 5. Cadmium survival assay. Strains gpx3⌬, trr1⌬, trx2⌬, trx1⌬, trx1⌬trx2⌬, and the wild-type strain were tested for cadmium resistance as described under "Materials and Methods." The experiments were performed at least twice with similar results; a typical error bar is represented at the bottom left corner of the figure. cadmium tolerance. YCF1 and GSH1 are two of them (27,8). Given the importance of the Ycf1p-GSH detoxification system, these two genes are probably the primary targets by which Yap1p exerts a control of cadmium tolerance. In accord with this notion, we found that cadmium-induced GSH synthesis is controlled by Yap1p. We also discovered that nine other genes required the presence of Yap1p for their proper induction by cadmium. These mostly include antioxidant defense genes also induced by H 2 O 2 in a Yap1-dependent manner (9). Several other proteins of the Yap1p H 2 O 2 -inducible regulon (Ccp1p, Ssa1p, Hsp78p, Hsp82p, Mpr1p, Uba1p, Cys3p, Tal1p, Zwf1p, Dak1p, and Tps1p) were still induced by Cd 2ϩ in yap1⌬ (see Table I), suggesting the existence of other control pathways for the cadmium response. This pathway does not involve the Yap1p homologous factor Yap2p, because a strain deleted for both YAP1 and YAP2 had the same proteomic response to cadmium than the yap1⌬ strain (not shown).
Interestingly, most of the cadmium-inducible Yap1p-dependent proteins were also overexpressed in a YAP1 multicopy strain after cadmium treatment. This YAP1 gene dosage effect, which was not observed in the H 2 O 2 response 3 correlate well with the hyper-resistance of the YAP1-overexpressing strain toward Cd 2ϩ but not toward H 2 O 2 .
Skn7p Acts As a Repressor in Cadmium Response-Skn7p is an important regulator of the H 2 O 2 response that cooperates with Yap1p to activate the expression of several genes in response to H 2 O 2 (9). However, although Yap1 is also important for the cadmium response, the function of Skn7p in this response is more complex because skn7⌬ is hyper-resistant to this metal (9). This resistance phenotype could not be explained by a defective Cd 2ϩ intracellular transport or alternatively, by the overproduction of metallothioneins (data not shown) or glutathione (this work). However, this could be explained by the observation that several proteins of the cadmium stimulon were superinduced in the skn7⌬ mutant, supporting the idea that Skn7p may act to repress these genes upon cadmium treatment. This was particularly striking for heat shock proteins and for some Yap1p-dependent proteins of unknown function (YNL134C, Gre2p, or Oye3p). Interestingly, the YAP1 multicopy strain overexpresses nearly the same set of proteins, suggesting a correlation between the cadmium hyper-resistance phenotype and the overexpression of these proteins. Some of these proteins, yet to be studied, might be important for cadmium resistance.