Regulation of the Yeast TSA1 Peroxiredoxin by ZAP1 Is an Adaptive Response to the Oxidative Stress of Zinc Deficiency*

Zinc deficiency is a potential risk factor for disease in humans because it leads to increased oxidative stress and DNA damage. We show here that the yeast Saccharomyces cerevisiae also experiences oxidative stress when zinc-deficient, and we have identified one mechanism yeast cells use to defend themselves against this stress. The Zap1p transcription factor is a central player in the response of yeast to zinc deficiency. To identify genes important for growth in low zinc, DNA microarrays were used to identify genes directly regulated by Zap1p. We found that the TSA1 gene is one such Zap1p target whose expression is increased under zinc deficiency. TSA1 encodes a cytosolic thioredoxin-dependent peroxidase responsible for degrading hydrogen peroxide and organic hydroperoxides. Consistent with its regulation by Zap1p, we showed that tsa1Δ mutants have a growth defect in low zinc that can be suppressed by zinc but not by other metals. Anaerobic conditions also suppressed the tsa1Δ low zinc growth defect indicating that oxidative stress is the likely cause of the poor growth. Consistent with this hypothesis, we demonstrated that zinc deficiency causes increased reactive oxygen species in wild type cells and that this increase is further exacerbated in tsa1Δ mutants. The role of this regulation by Zap1p in limiting oxidative stress in low zinc was confirmed when the Zap1p-binding site was specifically mutated in the chromosomal TSA1 promoter. Thus, we conclude that TSA1 induction by Zap1p is an adaptive response to deal with the increased oxidative stress caused by zinc deficiency.

Zinc deficiency is a potential risk factor for disease in humans because it leads to increased oxidative stress and DNA damage. We show here that the yeast Saccharomyces cerevisiae also experiences oxidative stress when zinc-deficient, and we have identified one mechanism yeast cells use to defend themselves against this stress. The Zap1p transcription factor is a central player in the response of yeast to zinc deficiency. To identify genes important for growth in low zinc, DNA microarrays were used to identify genes directly regulated by Zap1p. We found that the TSA1 gene is one such Zap1p target whose expression is increased under zinc deficiency. TSA1 encodes a cytosolic thioredoxin-dependent peroxidase responsible for degrading hydrogen peroxide and organic hydroperoxides. Consistent with its regulation by Zap1p, we showed that tsa1⌬ mutants have a growth defect in low zinc that can be suppressed by zinc but not by other metals. Anaerobic conditions also suppressed the tsa1⌬ low zinc growth defect indicating that oxidative stress is the likely cause of the poor growth. Consistent with this hypothesis, we demonstrated that zinc deficiency causes increased reactive oxygen species in wild type cells and that this increase is further exacerbated in tsa1⌬ mutants. The role of this regulation by Zap1p in limiting oxidative stress in low zinc was confirmed when the Zap1p-binding site was specifically mutated in the chromosomal TSA1 promoter. Thus, we conclude that TSA1 induction by Zap1p is an adaptive response to deal with the increased oxidative stress caused by zinc deficiency.
Zinc is an essential nutrient for all organisms because it is required as a structural and/or catalytic cofactor by hundreds of proteins. It was recently estimated that almost 3000 different proteins encoded by the human genome, i.e. ϳ10% of all human proteins, bind zinc (1). Because of the many functions of zinc, nutritional deficiency of this metal perturbs a wide variety of processes. In mammals, zinc deficiency is associated with retarded growth, immune system dysfunction, and impaired reproduction (2,3). In addition, studies both in vitro and in vivo have established that zinc deficiency leads to increased oxidative stress in mammalian cells (for review see Ref. 4). Zinc deficiency is associated with increased levels of lipid and protein oxidation (5,6). In addition, the oxidative stress associated with zinc deficiency leads to increased levels of DNA damage (6 -8). For these reasons, zinc deficiency has been proposed to be an important risk factor for cancer and other human diseases (9,10).
Because of the important functions of zinc, organisms have evolved with regulatory circuits to maintain intracellular zinc at optimal levels. In the yeast Saccharomyces cerevisiae, zinc homeostasis is maintained through both transcriptional and post-transcriptional mechanisms (11). At the level of transcription, expression of many genes is induced in zinc-limited cells through the action of the Zap1p transcription factor (12). Under low zinc conditions, Zap1p induces the expression of genes involved in zinc uptake (ZRT1, ZRT2, and FET4) (13,14) and genes involved in generating and mobilizing a supply of zinc stored in the vacuole (ZRC1 and ZRT3) (15,16). In addition, Zap1p regulates the expression of a gene responsible for zinc transport into the secretory pathway (ZRG17) (17,18) and controls the level of lipid biosynthetic enzymes (PIS1 and EKI1) to maintain membrane phospholipid composition (19,20). Finally, Zap1p induces its own expression under zinc deficiency by positive autoregulation (13). These many target genes of Zap1p contain one or more 11-bp zinc-responsive elements (ZREs) 2 in their promoters to which Zap1p binds. The consensus ZRE sequence has been identified as 5Ј-ACCTTNAAG-GT-3Ј (17).
The Zap1p protein has three known functional domains. At its C terminus is a DNA binding domain consisting of five C 2 H 2 zinc finger motifs (21,22). Zap1p also contains two activation domains that are independently regulated by zinc (23). Zincregulated activation domain 1 (AD1) maps to residues 182-502 (24). The second Zap1p activation domain is designated AD2. The AD2 region, residues 611-641, is located within two additional C 2 H 2 zinc fingers (25). These activation domains are apparently regulated by zinc binding directly to Zap1p. Thus, mutation of known or potential zinc ligand residues in AD1 and AD2, such as in the "Zap1p Triple/C2Q2 " (referred to here as "Zap1p TC ") allele (24), inhibit the zinc responsiveness of Zap1p and render its activity constitutive.
Previous microarray studies indicated that Zap1p regulates the expression of more than 40 genes in the yeast genome (17).
These previous studies were performed under conditions of relatively moderate zinc deficiency, i.e. conditions that may have missed Zap1p targets only induced by more severe zinc limitation. In this study, we conducted new microarray experiments using conditions of severe zinc deficiency in an effort to identify Zap1p target genes not recognized previously. As a result, we describe here the identification of one such new Zap1p target gene, TSA1. TSA1 encodes a cytosolic thioredoxindependent peroxidase responsible for the breakdown of hydrogen peroxide and organic hydroperoxides (26). We show that Tsa1p is required for growth under zinc-limiting conditions and that induction by Zap1p is an adaptive response to the increased oxidative stress experienced by zinc-deficient cells.
With the identification of TSA1 as a Zap1p target gene, we have uncovered a new mechanism for how zinc-deficient cells adapt to and survive the stress of low zinc.

EXPERIMENTAL PROCEDURES
Growth Conditions and Strains-Yeast cells were grown in YPD (YP medium ϩ 2% glucose) and in synthetic defined SD medium with 2% glucose and any necessary auxotrophic requirements (27). YPD and SD are zinc-replete because they contain micromolar levels of zinc and lack strong zinc chelators. Yeasts were made zinc limited by culturing in low zinc medium (LZM) prepared as described previously (28). LZM is zinc limiting because it contains 1 mM EDTA and 20 mM citrate to buffer metal availability. Zinc was added as ZnCl 2 . Cells were grown anaerobically using the BBL TM GasPak TM system (Pharmingen). Strains used are listed in Table 1. CWY14 was constructed by first integrating the CORE cassette containing the Kluveromyces lactis URA3 gene and kanMX4 into the TSA1 promoter of CWY2 deleting bases Ϫ203 to Ϫ149 (numbered relative to the first base of the ATG start codon). The promoter fragment from pTSA1 m3ZRE1 -lacZ (see below) was then transformed into the CORE cassette-containing strain and selected for loss of the URA3 gene by selection on 5-fluoroorotic acid (29). Correct mutation of the chromosomal TSA1 promoter was confirmed by PCR and DNA sequencing.
RNA and Protein Analysis-The microarray analyses used cells grown under three different paired conditions, and each experiment was performed in duplicate with independent cultures. In experiment 1, wild type (DY1457) cells were grown in zinc-limiting (LZM ϩ 3 M ZnCl 2 ) and in zinc-replete (LZM ϩ 3000 M ZnCl 2 ) media for 14 -16 h and harvested at an optical density measured at 600 nm (OD 600 ) of ϳ0.7. In experiment 2, wild type (DY1457) and zap1⌬ mutant (ZHY6) cells were initially grown in zinc-replete LZM ϩ 1000 M ZnCl 2 to exponential phase, washed twice in LZM ϩ 1 M ZnCl 2 , and then inoc-ulated into zinc-limiting LZM ϩ 1 M ZnCl 2 and grown for 6 h prior to harvesting. The final OD 600 was ϳ0.7. In experiment 3, wild type DY1457 cells were transformed with the vector (pYef2) or a plasmid encoding a constitutive allele of Zap1p, pYef2-Zap1p TC , under the regulation of the galactose-inducible GAL1 promoter (24). These transformants were inoculated into zinc-replete SD medium ϩ 2% galactose and grown for 20 -24 h before harvesting at an OD 600 of ϳ0.8. Total RNA was extracted with hot phenol, and mRNA was isolated using the PolyATtract mRNA Isolation System IV kit (Promega). Microarray analyses were performed as described previously (30). S1 nuclease protection assays were performed with total RNA as described (31). For each reaction, 15 g of total RNA was hybridized to a 32 P-end-labeled DNA oligonucleotide probe before digestion with S1 nuclease and separation on a 10% polyacrylamide, 5 M urea polyacrylamide gel. Band intensities were quantitated by PhosphorImager analysis (PerkinElmer Life Sciences). Crude protein extracts were generated by lysis in trichloroacetic acid, and immunoblot analysis was performed as described previously (24). The primary antibodies used were anti-HA (Roche Applied Science) and anti-Pgk1 (Molecular Probes).
Promoter Motif Analysis-ZREs were identified using a position-specific probability matrix generated from the ZREs of the 46 potential Zap1p targets identified previously (17). Potential ZREs in the TSA1 promoter were then identified using this matrix and RSA-TOOLS.
Electrophoretic Mobility Shift Assays-The Zap1p DNA binding domain (Zap1p DBD , residues 687-880) was expressed in Escherichia coli as a fusion to glutathione S-transferase and purified, and the glutathione S-transferase tag was then removed as described previously (21). Electrophoretic mobility shift assays were performed as described previously using purified Zap1p DBD protein and radiolabeled ZRE oligonucleotides ( Table 2) (21). In brief, 50 pmol of 32 P-end-labeled oligonucleotides were purified using G-50 Quick Spin columns (Roche Applied Science). Double-stranded oligonucleotides were prepared by annealing 1 M complementary single-stranded oligonucleotides in 10 mM Tris-HCl (pH 7.5), 100 mM NaCl, and 1 mM EDTA. The annealing mixtures were incubated for 15 min at 85°C and then 55°C for 4 h. For electrophoretic mobility shift assays, 15-l reactions were prepared containing 0.5 pmol of radiolabeled DNA oligonucleotide (20,000 cpm/pmol), 10 mM Tris-HCl (pH 8.0), 10 mM MgCl 2 , 50 mM KCl, 1 mM dithiothreitol, 0.02 mg/ml poly(dI-dC)⅐poly(dI-dC), 0.2 mg/ml bovine serum albumin, 0.04% IGEPAL CA-630, 10% glycerol, and the indicated concentrations of purified Zap1p DBD . After incubation for 1 h at room temperature, the samples were resolved on 6% polyacrylamide gels. Gels were dried onto blotting paper, and the signals were visualized by autoradiography.
Plasmid Constructs-Reporter plasmid pTSA1-lacZ was constructed in YEp353 (32) by homologous recombination (33). PCR products were generated from genomic DNA that contained 1000 bp of TSA1 promoter sequence (bases Ϫ1000 to ϩ1) flanked by homology to the vector. This fragment was gelpurified and cotransformed with EcoRI-and BamHI-digested YEp353; transformants were selected for URA3 prototrophy. The mutant alleles of TSA1 ZRE1 and/or ZRE2 elements (pTSA1 m3ZRE1 -lacZ, pTSA1 mZRE2 -lacZ, and pTSA1 mZRE1/2 -lacZ) were constructed in a similar fashion after generation of the mutant promoter fragments by overlap PCR (34).
To generate pFL-TSA1, the TSA1 open reading frame plus 1 kb of upstream and 0.5 kb of downstream flanking sequence was amplified by PCR and inserted by homologous recombination into BamHI-and BstXI-digested pFL38. Plasmid pFL-TSA1 C47S was generated in the same way following site-directed mutagenesis of Tsa1p residue Cys-47 by overlap PCR. A single hemagglutinin antigen epitope tag (35) was inserted at the N terminus of the TSA1 open reading frame in these two plasmids, generating pTSA1-HA and pTSA1 C47S -HA, by overlap PCR and insertion into pFL38 by homologous recombination. Function of the TSA1 allele in pTSA1-HA was confirmed by complementation of the mutant defects of a tsa1⌬ strain. All plasmid constructs were confirmed by DNA sequencing. Plasmids pZRC1-lacZ (16), pYef2 (36), and pYef2-Zap1p TC (24) were described previously.
␤-Galactosidase Assays-Cells were grown for 15-20 h to exponential phase in the indicated media. ␤-Galactosidase activity was measured in permeabilized cells as described previously (37), and activity units were calculated as follows: (⌬A 420 ϫ 1000)/(min ϫ ml of culture ϫ the culture OD 600 ).
DCF Assays-Measurement of the production of the reactive oxygen species (ROS) used 2,7-dichlorodihydrofluoroscein diacetate (DCFH-DA) (Calbiochem). DCFH-DA is membranepermeable and is trapped intracellularly following deacetylation. The resulting compound, DCFH, reacts with ROS (primarily H 2 O 2 and hydroxyl radicals) to produce the oxidized fluorescent form 2,7-dichlorofluoroscein (DCF). ROS analysis using DCFH-DA was performed as follows. Yeast cells were treated with 10 M DCFH-DA in culture media for 1 h prior to harvesting. Cells were then washed twice in ice-cold phosphate-buffered saline, resuspended in phosphate-buffered saline, and disrupted by vortexing with glass beads. Following centrifugation at 13,200 ϫ g for 10 min at 4°C, the supernatant was collected, and protein concentration was measured by the Bradford method (38). DCF fluorescence intensity was measured at an excitation wavelength of 504 nm and an emission wavelength of 524 nm and then normalized to protein level.
Glutathione Assays-Glutathione levels were measured using the method described by Vandeputte et al. (39). Cells were grown to exponential phase (ϳ1 ϫ 10 7 cells/ml), washed twice with distilled deionized H 2 O, and resuspended in 250 l of cold 1% 5-sulfosalicylic acid. Cells were broken by vortexing with glass beads and incubated at 4°C for 15 min to precipitate protein prior to centrifugation for 10 min at 13,200 ϫ g. The supernatants were used to determine glutathione levels. Total glutathione was determined by adding 10 l of lysate to 150 l of assay mixture (0.1 M potassium phosphate, pH 7.0, 1 mM EDTA, 0.03 mg/ml 5,5Ј-dithiobis(2-nitrobenzoic acid), 0.12 unit of glutathione reductase). The samples were mixed and incubated for 5 min at room temperature, and 50 l of NADPH (0.16 mg/ml) was then added. The formation of thiobis(2-nitrobenzoic acid) was monitored spectrophotometrically at 420 nm over a 5-min period. The reaction rate was proportional to the concentration of glutathione. Standard curves were generated for each experiment using 0 -0.5 nmol of glutathione in 1% 5-sulfosalicylic acid.
To measure GSSG alone, 100-l lysate samples were derivatized by adding 2 l of 97% 2-vinylpyridine, and the pH was adjusted by adding 2 l of 25% triethanolamine. These samples were then incubated at room temperature for 60 min. The samples were then assayed as described above for total glutathione. GSSG standards (0 -0.1 nmol) were also treated with 2-vinylpyridine in an identical manner to the samples. Subtraction of the amount of GSSG in the lysate from the total glutathione concentration allowed a determination of GSH levels present in each sample.

TSA1 Is a Zap1p
Target Gene-We used DNA microarray analysis to find genes with expression patterns consistent with their regulation by Zap1p. First, we identified genes that were expressed at higher levels in zinc-limited versus zinc-replete wild type cells (experiment 1). Second, we identified genes that were expressed at higher levels in wild type cells versus a zap1⌬ mutant strain grown under zinc-limiting conditions (experiment 2). Third, we identified genes with increased expression when a wild type strain was transformed with a constitutive Zap1p mutant allele (Zap1p TC ) versus the vector and grown on zinc-replete media (experiment 3). Zap1p TC is not repressed by high zinc because of mutations in both AD1 and AD2 that disrupt zinc regulation (24). Each experiment was performed in duplicate with independent cultures. Several known Zap1p target genes were found to have increased expression in these experiments including ZRT1, ZRT2, ZRT3, ZRC1, and FET4 (Table 3). One additional gene that showed these effects was the TSA1 gene that encodes the major cytosolic peroxiredoxin isozyme. Peroxiredoxins are antioxidant proteins that reduce a TSA1 mZRE1 and mZRE2 are mutated such that each position in the potential ZRE was altered by a transversion mutation. b ZREs or regions mutated in each complementary oligonucleotide pair are indicated by the underline. The lowercase letters indicate EagI-and SalI-complementary overhangs included for cloning these fragments into a lacZ reporter plasmid for future studies.
hydrogen peroxide, peroxynitrite, and organic hydroperoxides using thioredoxin as the electron donor (26). This report focuses on TSA1, and the complete results of these microarray experiments will be addressed in another report.
The microarray results suggested that TSA1 is a Zap1p target gene. To confirm these observations, cells were grown as they were for the microarray experiments 1-3, and TSA1 mRNA levels were assayed by S1 nuclease protection assays. As shown in Fig. 1A, this analysis confirmed the microarray results; TSA1 gene expression increased in response to zinc deficiency (experiment 1); this response was dependent on Zap1p (experiment 2), and TSA1 expression was increased by the constitutive Zap1p TC allele (experiment 3). The CMD1 gene encoding calmodulin was used as a loading control for these assays. We also assayed TSA1 mRNA levels over a range of zinc concentrations to determine what severity of zinc deficiency was required for this induction. As shown in Fig. 1B, TSA1 expression was induced during growth in LZM medium with 3 M or lower added zinc. This corresponds to conditions of severe degree of zinc deficiency as determined in previous studies (40). No induction was observed in a zap1⌬ mutant consistent with the hypothesis that TSA1 is a Zap1p target. To confirm that changes in mRNA levels also alter Tsa1p protein levels, we inserted a hemagglutinin antigen (HA) epitope tag at the N terminus of the TSA1 open reading frame expressed from its own promoter. This epitope-tagged allele was found to fully complement tsa1⌬ mutant phenotypes (see below). We then assayed Tsa1p protein levels over a range of zinc concentrations by immunoblotting. Tsa1p protein levels increased greater than 2-fold in response to zinc deficiency with the same dose-response profile as was observed for the TSA1 mRNA (Fig. 1C). No change in level was observed for a control protein, 3-phosphoglycerate kinase (Pgk1). Thus, Tsa1p protein levels correlate with the observed changes in mRNA level.
An additional feature of Zap1p target genes is the presence of one or more ZREs in their promoters. To identify potential ZREs in the TSA1 promoter, a motif identification algorithm (RSA-TOOLS) was used to scan the promoters of TSA1 and several known Zap1p target genes for potential ZREs. Two candidate ZREs, designated ZRE1 and ZRE2, were identified in the TSA1 promoter located between 160 and 195 bp upstream of the open reading frame ( Table 4). The start site for TSA1 transcription has not yet been determined. The numerical RSAT scores, a representation of the match of the sequence to the consensus, of these putative ZREs were relatively low but still higher than the scores of some ZRE sequences from other pro-moters that have been shown experimentally to be functional in vivo (e.g. ZRE3 of ZRT2) (41).
To determine whether Zap1p can bind to these potential TSA1 ZREs, electrophoretic mobility shift assays were performed using the Zap1p DNA binding domain (Zap1p DBD ) purified from E. coli. Increasing concentrations of the Zap1p DBD protein incubated with ZRE1 TSA1 resulted in increased abundance of a protein-DNA complex ( Fig. 2A). Similar results were obtained with a ZRE2 TSA1 -containing oligonucleotide, although Zap1p binding to this sequence appeared to be less efficient. Binding of Zap1p DBD to the ZRT1 ZRE1 sequence, a strong element closely matching the ZRE consensus sequence, showed efficient complex formation at even lower levels of Zap1p DBD . The apparent weakness of Zap1p binding to the TSA1 ZREs is consistent with the induction of Tsa1p expression only under conditions of severe zinc deficiency where Zap1p levels are highest. When the TSA1 ZRE elements were mutagenized to disrupt their similarity to the ZRE consensus sequence (Table 2), no Zap1p DBD binding was observed (Fig. 2B). These results indicated that Zap1p binds to TSA1 ZRE1 and ZRE2 with sequence specificity.
To determine whether the TSA1 ZREs are functional in vivo, they were mutated either singly or together in the context of the full-length TSA1 promoter fused to the E. coli lacZ gene. As shown in Fig. 3A, the TSA1 promoter contains elements for regulation by two other factors, Yap1 and Skn7 (42). Yap1 and Skn7 increase expression of TSA1 in response to oxidative stress. Skn7 acts through OSRE1 and OSRE2 in the TSA1 promoter, and Yap1 binds to a YRE sequence located between the candidate ZRE1 and ZRE2 sites. OSRE1 overlaps with the ZRE1 sequence. To mutate ZRE1 without affecting Skn7 regulation, we introduced a 3-bp substitution (GGT to TTG) at the open reading frame-proximal end of the element such that OSRE1 was unaltered. Previous studies showed that these 3 bases are essential for sequence-specific Zap1p binding (22). Each position of the 11-bp sequence of ZRE2 was altered by a transversion mutation in the constructs in which this element was disrupted.
Expression of these TSA1-lacZ reporters was then assayed in zinc-replete and zinc-deficient cells. A control Zap1p-regulated reporter containing the ZRC1 promoter, pZRC1-lacZ, showed low expression in zinc-replete wild type cells and high expression in zinc-limited cells (Fig. 3B). Increased expression of ZRC1-lacZ in low zinc was eliminated in a zap1⌬ mutant strain. The wild type TSA1-lacZ reporter in wild type cells showed a small but reproducible increase in expression in low zinc. This induction was abolished by mutating three critical bases of ZRE1 (pTSA1 m3ZRE1 -lacZ). In fact, expression in low zinc was greatly reduced relative to the levels of expression observed in zinc-replete cells. These data indicate that expression of the reporter in low zinc conditions is largely dependent on Zap1p activity. This conclusion was confirmed when these reporters were assayed in the zap1⌬ mutant strain. In contrast to the effects of ZRE1 mutation, disruption of ZRE2 had little effect on expression of the TSA1-lacZ fusion either in the presence or absence of a functional ZRE1 sequence. These results  The position of the 1st base of each ZRE relative to the ATG start codon is indicated. b RSAT score was generated using the RSA-TOOLS algorithm. FIGURE 2. Zap1p binds specifically to candidate ZREs identified in the TSA1 promoter. A, electrophoretic mobility shift assay of TSA1 ZRE sequences. Samples of radiolabeled double-stranded oligonucleotides containing the ZRE sequences from the TSA1 promoter (ZRE1 TSA1 and ZRE2 TSA1 or from the ZRT1 promoter (ZRE1 ZRT1 ) were incubated with purified Zap1p DBD for 1 h prior to native gel electrophoresis. The experiments were performed with 0, 0.4, 0.8, and 1.2 g per reaction of Zap1p DBD for ZRE1 TSA1 and ZRE2 TSA1 , and with 0.05 and 0.2 g per reaction of Zap1p DBD for ZRE1 ZRT1 . FP denotes the free probe, and the arrowhead indicates the Zap1p DBD -ZRE complex. B, sequence specificity of Zap1p DBD binding. Radiolabeled oligonucleotides with the wild type TSA1 ZREs (ZRE1 TSA1 and ZRE2 TSA1 ) or mutated forms (mZRE1 TSA1 and mZRE2 TSA1 ) were analyzed as in A with (ϩ) or without (Ϫ) 0.4 g of purified Zap1p DBD . The data shown are representative of two independent experiments.
indicated that ZRE1 plays the major role for the up-regulation of TSA1 under low zinc conditions. It is noteworthy that expression of these reporters was unaltered in zinc-deficient ⌬skn7 or ⌬yap1 mutants (data not shown). This result indicates that these other factors are not required for TSA1 expression in low zinc.
The Importance of TSA1 to Zinc-limited Growth-Induction of TSA1 expression by Zap1p under zinc-limiting conditions suggested that the Tsa1p protein is important for growth of zinc-deficient cells. To test this hypothesis, wild type and tsa1⌬ deletion mutant cells were inoculated into zinc-replete or zinclimiting media, and cell growth, measured by the optical density of the culture, was monitored over time. Both wild type and tsa1⌬ mutant cells grew well when inoculated into zinc-replete medium (LZM ϩ 1000 M ZnCl 2 ) (Fig. 4A). Although growth of wild type cells was somewhat impaired in a zinc-limiting medium (LZM ϩ 1 M ZnCl 2 ), the tsa1⌬ mutant was severely defective for growth under these conditions. To determine the level of zinc required for optimal growth of the mutant, wild type and tsa1⌬ cells were inoculated into LZM supplemented over a range of zinc concentrations, and cell density was determined after 48 h. Although the poor growth of the tsa1⌬ mutant was again observed in LZM ϩ 1 M ZnCl 2 , as little as 10 M added zinc was sufficient to largely restore growth of this mutant strain (Fig. 4B). These results indicate that tsa1⌬ mutants are sensitive to severe zinc limitation but tolerate more mild deficiency. As shown in Fig. 4C, the growth defect was specific to zinc deficiency; supplementation with 100 M concentrations of several other metal ions did not restore growth to the tsa1⌬ mutant strain.
The Tsa1p protein has two functional activities. First, Tsa1p is a thioredoxin-dependent peroxidase. Second, Tsa1p has molecular chaperone activity and mediates the correct folding of misfolded proteins (43). To determine which of these activities was required for low zinc growth, the TSA1 gene on a low copy plasmid was mutated to substitute a serine residue for a critical cysteine residue at position 47 (C47S). This mutation completely disrupts peroxidase activity, yet the protein retains Ͼ90% of chaperone activity when assayed in vitro (43). A wild type plasmid copy of TSA1 complemented the low zinc growth defect of a tsa1⌬ mutant (Fig. 5A). In contrast, the Tsa1p C47S allele failed to complement the mutant growth defect. Immunoblots using the HA epitope tag confirmed that the Tsa1p C47S protein accumulated to wild type levels and was similarly zincregulated (Fig. 5B). These results suggest that the peroxidase activity is required for zinc-limited growth, and the Tsa1p chaperone activity may play a less important role.
A requirement for Tsa1p peroxidase activity suggested that tsa1⌬ mutant cells grown under low zinc conditions are experiencing increased oxidative stress. If so, we predicted that growth under anaerobic conditions, under which the generation of ROS is diminished, would suppress the tsa1⌬ mutant growth defect. As shown in Fig. 5C, this was indeed the case.
Although growth of wild type cells was not greatly altered by anaerobic growth, anaerobic tsa1⌬ mutants grew as well as wild type cells in low and high zinc. An alternative explanation for these results is that cells grown under anaerobic conditions may have a lower requirement for zinc than do aerobic cells. If this were the case, the improved growth of tsa1⌬ mutants under anaerobic conditions would be due to zinc repletion rather than decreased oxidative stress. However, a zrt1⌬ mutant defective  for zinc uptake grew poorly under low zinc conditions under both aerobic and anaerobic conditions (Fig. 5C), suggesting that zinc requirements were not greatly altered by anaerobiosis. In addition, wild type cells showed a similar dose response of growth to added zinc under aerobic and anaerobic conditions indicating that zinc requirement of yeast cells is not greatly altered by the absence of oxygen (data not shown).
These results suggested that zinc deficiency increases oxidative stress in yeast. To test this hypothesis, we determined the sensitivity of wild type and tsa1⌬ mutant yeast to hydrogen peroxide. We predicted that increased intracellular oxidative stress in zinc-limited cells would lead to greater sensitivity to an external oxidizing agent. Consistent with this prediction, wild type cells were more sensitive to H 2 O 2 treatment when grown under zinc-limiting conditions than were zinc-replete cells (Fig.  6A). Mutant tsa1⌬ cells showed sensitivity to H 2 O 2 at concentrations lower than those affecting wild type cells and showed even greater sensitivity when grown under low zinc conditions. The low zinc growth defect of the tsa1⌬ mutant is apparent in the cells not treated with H 2 O 2 . Addition of 0.1 mM H 2 O 2 further retarded growth of the zinc-limited tsa1⌬ mutant but had little effect on zinc-replete cells.
In addition, we assayed the accumulation of ROS in vivo in wild type and tsa1⌬ mutant cells grown under zinc-limiting and -replete conditions (Fig. 6B). This assay used a fluorescent probe, DCFH-DA, which accumulates in cells and increases in fluorescence intensity when oxidized by reactive oxygen. There was an ϳ3-fold increase in fluorescence in zinc-limited versus replete wild type cells indicating the accumulation of reactive oxygen species under zinc deficiency. In tsa1⌬ mutant cells, there was an even greater (ϳ10-fold) increase in DCF fluorescence in response to zinc deficiency. Treatment of zinc-limited cells with exogenous H 2 O 2 resulted in the same level of DCF fluorescence indicating that the differences observed between wild type and tsa1⌬ mutants were not because of differences in DCFH-DA loading or its responsiveness to ROS.
We also tested the correlation between increased ROS and the growth defect of tsa1⌬ mutants. When zinc-replete wild type and tsa1⌬ mutants were inoculated into zinc-limiting medium, they grew equally well for ϳ8 h, after which growth of the mutant was severely impaired (Fig. 6C). Growth over the first 8 h is likely maintained by the mobilization of intracellular zinc stores in the vacuole (15). When ROS levels were assayed in these cells using DCFH-DA, we found that the growth defect of the tsa1⌬ mutant correlated closely with a rise in ROS after 8 h of culturing. These data are consistent with the hypothesis that the growth defect of the tsa1⌬ mutant is because of increased ROS in zinc-deficient cells. ROS levels in the tsa1⌬ mutant at 22 h are reduced relative to earlier time points, perhaps due to the action of other antioxidant mechanisms, but are still significantly higher than wild type levels. As an additional test of the oxidative stress of zinc deficiency, we measured total glutathione, GSH and GSSG levels, and the GSH:GSSG ratio in wild type and tsa1⌬ mutants grown in high and low zinc. Oxidative stress is indicated by increased total glutathione production, increased GSH and GSSG levels, and a decreased GSH:GSSG ratio. By comparing wild type cells grown in high and low zinc, oxidative stress of zinc deficiency was indicated by all four indices (Fig. 7). Comparison of wild type versus the tsa1⌬ mutant grown in low zinc indicated that the oxidative stress of zinc deficiency is of much greater severity in the mutant; total glutathione, GSH, and GSSG were each elevated relative to wild type, and the GSH:GSSG levels were reduced in the mutant strain. Most remarkably, GSSG levels were elevated 20-fold, and the GSH:GSSG ratio was reduced 12-fold in zinc-limited tsa1⌬ mutants relative to zinc-limited wild type cells.
Taken together, the results presented in Figs. 4 -7 indicate that zinc deficiency is a major source of oxidative stress for yeast, and Tsa1p is required to eliminate that stress. These experiments used a tsa1⌬ mutant in which the entire open reading frame was deleted. To assess the role of the zinc-responsive regulation of TSA1 in oxidative stress resistance, we mutated ZRE1 in the chromosomal TSA1 gene to generate the tsa1-1 m3ZRE allele. To achieve this, the wild type ZRE1 sequence in the promoter was replaced by the nonfunctional m3ZRE1 mutation (Fig. 3). The TSA1 open reading frame is completely intact in the tsa1-1 m3ZRE allele; only its regulation by Zap1p is altered. This mutation disrupted induction of chromosomal TSA1 under low zinc conditions but did not affect expression in zinc-replete cells (Fig. 8A). These data confirmed the results obtained with the TSA1-lacZ reporters indicating that Zap1p contributes to TSA1 expression in zinc-limited cells. When assayed for growth in low zinc, we found that the tsa1-1 m3ZRE strain showed no obvious growth defect relative to wild type cells indicating that the Zap1p-independent expression of TSA1 is sufficient to largely maintain growth under these conditions (data not shown). However, when ROS levels FIGURE 8. Mutant tsa1-1 m3ZRE1 cells are defective for ROS elimination in low zinc. A, wild type (WT) (CWY2) and tsa1-1 m3ZRE1 (CWY14) cells were grown with the indicated level of zinc added to the medium. The cells were harvested, and RNA was isolated and assayed by S1 nuclease protection assay. The data shown are representative of three independent experiments. B, cells grown as in A were analyzed for ROS by DCF fluorescence. The data plotted are the means of three independent cultures, and the error bars indicate 1 S.D.  were assayed by DCF fluorescence in zinc-limited tsa1-1 m3ZRE cells, we found that oxidative stress was clearly increased relative to wild type cells (Fig. 8B). No differences were observed in zinc-replete cells. When assayed for GSH and GSSG levels, no differences were detected between wild type and tsa1-1 m3ZRE mutants grown in high zinc (Table 5). However, both GSH and GSSG levels were elevated in these strains grown in low zinc. In zinc-limited cells, the levels of these metabolites in the tsa1-1 m3ZRE mutant were 15-20% higher than in the wild type. These results further suggest that there is greater oxidative stress in the tsa1-1 m3ZRE mutant. GSH:GSSG ratios were reduced in zinc-limited relative to zinc-replete cells indicative of the oxidative stress of zinc deficiency, but no significant differences in this ratio were observed between wild type and mutant strains. These results suggest that the cellular redox environment is adequately buffered in the tsa1-1 m3ZRE mutant despite the increased ROS.

DISCUSSION
In this study, we present evidence that the TSA1 gene is a direct target of Zap1p transcriptional activation in zinc-limited cells. The Tsa1p protein is an important antioxidant molecule in cells. As a thioredoxin-dependent peroxidase, Tsa1p reduces H 2 O 2 and organic hydroperoxides during oxidative stress (26). In addition to conferring resistance to exogenous oxidizing agents, the antioxidant function of Tsa1p was found to prevent cellular damage caused by endogenous oxidative stress. For example, the Tsa1p protein prevents oxidative DNA damage and reduces the frequency of spontaneous mutations that arise during growth under nutrient-sufficient conditions (44,45). Tsa1p is also required to protect cells from endogenously generated oxidative stress when respiration is disrupted (46,47).
Our studies now show that Tsa1p function is very important for the growth of cells under zinc-limiting conditions. The peroxidase activity of Tsalp is likely to be of major importance to zinc-deficient cells because these cells experience increased endogenous oxidative stress. Oxidative stress caused by zinc deficiency has been observed in many studies of mammalian cells grown in culture as well as in studies of animals fed zincdeficient diets (4). Thus, our results indicate that oxidative stress under zinc deficiency is a problem encountered by a phylogenetically wide range of organisms. Yeast has apparently evolved to combat that increased oxidative stress by inducing an antioxidant protein directly in response to zinc deficiency.
The many studies indicating increased oxidative stress under zinc deficiency have led to the notion that zinc plays an antioxidant role despite the fact that this metal is not redox-active under physiological conditions (4,48). The source of the oxidative stress associated with zinc deficiency has not been determined, but several models have been proposed. First, Cu,Znsuperoxide dismutase is a zinc-dependent enzyme whose antioxidant activity may be compromised under zinc deficiency. Yeast expresses a Cu,Zn-superoxide dismutase protein, Sod1p, that plays important antioxidant roles, so its impairment is one possible source of oxidative stress in zinc-deficient yeast. Second, zinc-metallothionein has recently been shown to also have antioxidant activity through the oxidation of metalbound cysteine ligands in the protein (49). Expression of many metallothioneins is induced by zinc treatment possibly leading to an increased ability to eliminate reactive oxygen species. Although the yeast metallothioneins Cup1p and Crs5p are not induced by zinc, Cup1p can bind zinc in vitro (50). Therefore, loss of zinc-Cup1p/Crs5p complexes could potentially be a contributing factor in the oxidative stress experienced by zinclimited yeast cells. As a third mechanism, it has been proposed that zinc may compete with redox active metal ions like copper and iron for binding to sites on proteins and other cellular macromolecules. Such competition would inhibit the site-specific production of oxygen radicals through metal-catalyzed Fenton chemistry. Finally, zinc has been proposed to bind to and protect free sulfhydryl groups in proteins. This may be another way in which zinc normally plays an antioxidant role and inhibits production of reactive oxygen species. These various models are currently under investigation to determine the source of oxidative stress in zinc-deficient yeast. Through the analysis of yeast, we hope to ultimately determine the source(s) of oxidative stress under zinc deficiency in other organisms.
In addition to its peroxidase activity, Tsa1p can act as a molecular chaperone and facilitate protein folding (43). The chaperone function of Tsa1p is important for survival during heat shock and protects ribosomal proteins from aggregation during reductive stress (51). Although we believe that the peroxidase activity plays the primary role in zinc-limited cells, our studies do not eliminate some role for Tsa1p chaperone function in zinc deficiency. For example, Tsa1p may help maintain the apo-form of zinc metalloproteins in a state competent for zinc binding. The caveat in our studies designed to address the role of the chaperone function arises from the observation that although Tsa1p residue cysteine 47 is not required for chaperone function in vitro, it is required for the peroxide-induced transition from a low molecular weight dimeric chaperone form to an even more active multimeric "superchaperone" (43). Therefore, it is still possible that the superchaperone form is important for low zinc growth.
The level of Zap1p-mediated induction of TSA1 expression is relatively modest when compared with some other Zap1p targets. For example, whereas TSA1 is induced only about 2-3fold, ZRT1 is induced ϳ60-fold (Table 3). Nonetheless, the small level of TSA1 induction could have a large impact on the amount of Tsa1p protein in the cell. Tsa1p is an abundant protein that accumulates in zinc-replete cells to ϳ400,000 molecules per cell (52,53). A 3-fold increase in protein expression such as we have observed would add another 800,000 molecules of Tsa1p per cell, i.e. a major increase in the antioxidant capacity of the cell. That this increased level of Tsa1p is involved in reducing ROS levels in zinc-limited cells was demonstrated when we specifically mutated the Zap1p-dependent regulation of the TSA1 promoter ( Fig. 8 and Table 5).
Why do the ZRE-less tsa1-1 m3ZRE mutants not show a growth defect in low zinc as we observed for the complete tsa1⌬ deletion mutant? The simplest explanation is that the oxidative stress experienced in this partially active promoter mutant, although clearly elevated relative to wild type, is not enough to completely overwhelm the redox buffering systems of the cell. This is suggested by the stable GSH:GSSG ratios observed in wild type and tsa1-1 m3ZRE mutants (Table 5). Why then would TSA1 regulation evolve in this way? We believe that the higher oxidative stress observed in the tsa1-1 m3ZRE mutant in low zinc places an additional burden on these cells and puts them at a disadvantage relative to wild type cells. Oxidative damage is likely elevated in these promoter mutant cells, but the level of this damage may be too subtle to detect in our short term growth assays. However, a small increase in oxidative damage could still influence the long term evolution of these stressresponse regulatory circuits. Further evidence supporting this hypothesis comes from the observation that the ZRE1 sequence is evolutionarily conserved in the promoters of TSA1 orthologs in other yeast species such as Saccharomyces paradoxus and Saccharomyces mikatae (Saccharomyces Genome Data base) (data not shown).
Surprisingly, TSA1 was also identified previously as a gene induced by high zinc conditions (54). Those results were obtained using a reporter allele of TSA1 in which the E. coli lacZ gene was inserted within the open reading frame. We have been unable to detect increased expression in zinc-treated cells by S1 nuclease protection analysis of endogenous TSA1 expression (data not shown) suggesting that this previous result may have been an artifact of the lacZ reporter used in that study. Nonetheless, the possibility that TSA1 is induced by high zinc under some conditions does not challenge our hypothesis regarding the importance of this protein for low zinc growth.
The TSA1 gene is also regulated by the Yap1 and Skn7 transcription factors that up-regulate the gene in response to oxidative stress. We found that the increased expression of TSA1 in low zinc was not dependent on either Yap1 or Skn7. From these observations, we conclude that Yap1 and/or Skn7 are either less responsive to oxidative stress in zinc-limited cells or, alternatively, the level of oxidative stress induced by zinc deficiency is insufficient to activate the Yap1/Skn7 response. Direct regulation of TSA1 by Zap1p also has other advantages to the cell. If increased oxidative stress is an unavoidable consequence of zinclimited growth, a cell that induces TSA1 expression specifically in response to low zinc would have a fitness advantage over a cell that can only respond to the resulting oxidative stress. By using Zap1p, yeasts induce TSA1 expression prior to or concurrent with the increasing oxidative stress and therefore are better prepared to limit oxidative damage.
Induction of Tsa1p expression in zinc-deficient yeast is a useful strategy to deal with the elevated oxidative stress associated with zinc limitation. Peroxiredoxins related to Tsa1p are found at all phylogenetic levels, including humans and other mammals (26). It will be interesting to determine whether orthologs of Tsa1p in other organisms are also induced by low zinc. Given the effects of oxidative stress on aging and disease, this issue is of great potential importance to human health.