Oxidant-specific Folding of Yap1p Regulates Both Transcriptional Activation and Nuclear Localization*

The yeast transcriptional regulator Yap1p is a key determinant in oxidative stress resistance. This protein is found in the cytoplasm under non-stressed conditions but rapidly accumulates in the nucleus following oxidant exposure. There it activates transcription of genes encoding antioxidants that return the redox balance of the cell to an acceptable range. Yap1p localization to the nucleus requires the oxidant-specific formation of disulfide bonds in the N-terminal cysteine-rich domain (N-CRD) and/or the C-terminal cysteine-rich domain (C-CRD). H2O2 exposure triggers the formation of two interdomain disulfide bonds between the N-and C-CRDs. This dually disulfide-bonded structure has been argued to mask the nuclear export signal in the C-CRD that would otherwise prevent Yap1p nuclear accumulation. The C-CRD is required for wild-type H2O2 tolerance but dispensable for resistance to diamide. The Saccharomyces cerevisiae TRX2 gene, encoding a thioredoxin protein, cannot be induced by H2O2 in the presence of various mutant forms of Yap1p lacking the normally functioning C-CRD. In this work, we demonstrate that the proper folding of Yap1p in the presence of H2O2 is required for recruitment of the mediator component Rox3p to the TRX2 promoter in addition to the nuclear accumulation of Yap1p during stress by this oxidant. These data demonstrate that the dually disulfide-bonded Yap1p N- and C-CRDs form a bifunctional protein domain controlling both nuclear localization and transcriptional activation.

the promoter of target genes (7)(8)(9). In the absence of oxidative challenge, Yap1p resides in a cytoplasmic location (10). However, upon stress by addition of oxidants like diamide and H 2 O 2 , nuclear export of Yap1p is inhibited (11,12), and the factor rapidly accumulates in the nucleus, where it induces the expression of genes that often encode antioxidant proteins (10,13). This oxidant-regulated nuclear localization has been the focus of much research, and important details have emerged illuminating the molecular mechanism. Yap1p contains two clusters of cysteine residues called cysteine-rich domains located at the N terminus (N-CRD) and the C terminus (C-CRD). These N-and C-CRDs provide all the cysteine residues found in the Yap1p sequence, and control of their redox status is crucial for control of Yap1p activity by oxidative stress. Previous work established that deletion of the C-CRD leads to Yap1p being constitutively located in the nucleus with transcription of an artificial Yap1p-responsive reporter plasmid and diamide tolerance being constitutively elevated (10,14). Interestingly, although elimination of the C-CRD leads to diamide hyper-resistance, H 2 O 2 tolerance is lowered relative to wild-type Yap1p (14).
Important characterization of this oxidant-selective behavior of Yap1p was provided by the biochemical analysis of Yap1p after diamide and H 2 O 2 exposure. Diamide challenge produces disulfide bonds that appear to form in either the N-or C-CRD (15), whereas H 2 O 2 stress induces the formation of a disulfide bond between the N-and C-CRDs (16). Mutational analysis of cysteine residues in the N-and C-CRDs indicated that Cys 303 and Cys 598 are required for both normal H 2 O 2 resistance and disulfide bond formation, but these same residues are dispensable for diamide tolerance (13,16). More recent experiments have demonstrated that two disulfide bonds are formed between residues in the N-and C-CRDs (17) in response to H 2 O 2 but not diamide. The oxidant-specific defects of Yap1p mutants have been linked to specific defects in the program of Yap1p-dependent gene activation (13,16).
The thioredoxin-encoding TRX2 gene is a Yap1p regulatory target that is required for normal H 2 O 2 resistance (7). Yap1p mutants that cannot form the interdomain disulfide bond between the N-and C-CRDs are unable to elevate TRX2 expression in the presence of H 2 O 2 (13,16). Experiments using fragments of the TRX2 promoter placed upstream of a CYC1-lacZ fusion gene indicated that TRX2 promoter sequences from positions Ϫ141 to Ϫ61 prevent overexpression of ␤-galactosidase activity in response to the presence of a constitutively nuclear Yap1p mutant (13). We have determined that the element blocking high-level CYC1-lacZ activation in this TRX2-CYC1-lacZ reporter corresponds to the TRX2 TATA box. Chromatin immunoprecipitation studies have indicated that the mediator component Rox3p is preferentially recruited to the TRX2 promoter during H 2 O 2 stress. Yap1p mutants unable to form the N-and C-CRD interdomain disulfide bonds fail either to recruit Rox3p to TRX2 or to activate expression of this gene in the face of H 2 O 2 challenge. This problem is seen even in a C-CRD mutant that is constitutively located in the nucleus. Together, these data argue that formation of properly disulfide-bonded Yap1p during H 2 O 2 stress is required to generate a new bifunctional protein domain that allows Yap1p to accumulate in the nucleus as well as to recruit Rox3p to TRX2.
Plasmids-Low-copy-number URA3-containing plasmids carrying the wild-type or C629A YAP1 gene have been described (13,23). LEU2 variants of wild-type and C629A YAP1 were generated by moving a SacI/HindIII fragment from pSM58wt or pC629A into SacI/HindIII-cut pRS315 (pSMS51 and pRS315-C629A respectively). The TRX2 Ϫ (13) and GSH1-lacZ (8) reporter plasmids have been described previously. A low-copy-number plasmid (pSC-T1) bearing the wild-type TRX2 coding region containing a BamHI site introduced after the translation start site was generated by PCR. Wild-type or mutant YRE forms of the TRX2 promoter were generated by PCR and fused to the TRX2 open reading frame (pSC-T1) or lacZ reporter plasmids (pSEYC102) (24). The primer used to alter the Ϫ218 YRE (GTT TAT ACT Cga gGT AAA GGA TGC TCC) introduced an XhoI site to facilitate identification. (The nucleotides changed are indicated by lowercase letters, and the restriction enzyme site introduced is underlined) The primer used to introduce changes into the Ϫ181 YRE (CTG AAC GCG tcT Aga AAG AAA AGA GCC) incorporated an XbaI site to ease identification. The mutant promoters were moved into the context of pSC-T1 or pSEYC102 as EcoRI/ BamHI fragments.
The 3Ј-deletion series was created by PCR amplification of a portion of the TRX2 promoter and were then placed upstream of the CYC1 promoter lacking its upstream activating sequence (UAS) and fused to lacZ. The forward primer for all plasmids in this series anneals at 237 bp upstream of the translation start site. Its sequence is GCG GAT CCA TGT GTA ATT GTT TAT ACT C. For the Ϫ120 deletion mutant, the reverse primer was CCG AGA TCT ATG GCT TTC TTA TAT ACT GAT. The Ϫ104 deletion mutant was created with the reverse primer CCG AGA TCT CTT TTC ATC CCC CGA ATG GCT T. For the Ϫ90 deletion mutant, the reverse primer was CCG AGA TCT TTA TTC TCT TGT CAG CTT TTC. These fragments were cloned into pCR2.1-TOPO (Invitrogen) and digested with EcoRI and BglII, and the TRX2 segments were inserted into EcoRI/BamHI-cut p314-CLZ as described (13). The Ϫ155 and Ϫ61 deletion mutants have been described (13).
A 240-bp fragment of the TRX2 promoter containing a mutant TATA box (pSAR28) was constructed by site-directed mutagenesis and cloned into the pCR2.1-TOPO cloning vector. The primer used to mutate the site was GAA AGA GGG ATA TCA GtC GAC AAG AAA GCC ATC GG to create pSAR28, which was then cleaved with EcoRI and BamHI; and the 240-bp fragment was moved into the pSEYC102 plasmid to create a lacZ fusion (pSAR29). A second 240-bp fragment was generated that would replace the TRX2 TATA box with the TRP5 TATA box. This fragment was ligated into pSEYC102 to create a lacZ fusion (pSAR31). The primer used to produce this mutant was GAA AGA GGG ATA TCA GTA TAA GAA AAA GCC ATT CGG. All PCR products were sequenced to ensure that no errors had occurred during amplification.
DNase I Footprinting-PCR products of the TRX2 promoter containing either wild-type or mutant YREs were amplified using primers Ϫ296-TRX2-For (GAT GGA TCC AAG ATC AGC ATA ACT TG) and TRX2-BglII-Rev (CCG AGA TCT TAT TGA TGT GTT ATT TAA AG) and cloned into pCR2.1-TOPO. These plasmids were digested with BamHI, treated with calf intestinal alkaline phosphatase, phosphorylated with T4 polynucleotide kinase and [␥-32 P]ATP, and finally cleaved with EcoRI. The appropriate labeled fragments were isolated from an agarose gel using a Qiagen mini-elute column. DNase I protection was performed using bacterially expressed Yap1p as reported previously (25).
Electrophoretic Mobility Shift Assay-For the TATA box-binding protein Tbp1p, a 237-bp segment of the wild-type TRX2 promoter was amplified using primers Ϫ236-TRX2-For (TGCGGATTCATGTGTA-ATTGTTTATAC) and TRX2-BglII-Rev. The resulting PCR fragment was cloned into pCR2.1-TOPO. A 5Ј-end-labeled fragment was prepared as described above for the DNase I probes. The labeled fragment was isolated and incubated with bacterially produced Tbp1p (provided by Dr. Anthony Weil, Vanderbilt University). In some cases, unlabeled competitor DNAs were included in the binding reaction to verify binding specificity. Electrophoretic mobility shift assays were performed basically as described (26). Binding was carried out in 4% glycerol, 20 mM Tris-HCl (pH 8.0), 60 mM KCl, 5 mM MgCl 2 , 100 g/ml bovine serum albumin, 1 mM dithiothreitol, and 20 g/ml poly(dI-dC) for 45 min at 30°C. DNA and protein⅐DNA complexes were resolved by electrophoresis on a 4% native acrylamide gel at 4°C using 1ϫ buffer containing 45 mM Tris (pH 8.0), 45 mM boric acid, and 1 mM EDTA. The gel was dried and visualized by autoradiography.
Primer Extension Mapping of the Transcription Start Site-RNA was isolated from yeast strains grown in YPD medium. Primer CAC TGT CGT ATT CAG AAG CG was first end-labeled with 32 P and then mixed with 50 g of total RNA. This mixture was ethanol-precipitated, and nucleic acids were brought up in 20 l of annealing buffer (250 mM KCl, 10 mM Tris-Cl (pH 8.2), and 1 mM EDTA (pH 8.0)) and heated to 65°C for 45 min. This mixture was cooled to and allowed to anneal at 37°C for 6 h. Five units of avian myeloblastosis virus reverse transcriptase, 5 g of actinomycin D, 50 mM Tris-HCl (pH 8.3), 50 mM KCl, 10 mM MgCl 2 , 10 mM dithiothreitol, 0.5 mM spermidine, 5 mM each dNTP, and 40 units of RNasin were mixed and brought to a final volume of 100 l. Reverse transcription took place at 37°C for 1 h. The DNA⅐RNA complexes were treated with 30 g of RNase A for 30 min at 37°C. The DNA was then precipitated, resuspended in stop buffer (95% formamide, 20 mM EDTA, 0.05% bromphenol blue, and 0.05% xylene cyanol), heated to 90°C, chilled briefly, and then run at 20 watts on a 6% denaturing acrylamide gel. Sequencing reactions were performed with the Sequenase DNA sequencing kit (U. S. Biochemical Corp.) according to the manufacturer's instructions.
Chromatin Immunoprecipitation-Cells were grown in liquid medium until an A 600 nm of 0.8 -1.0 was reached. Formaldehyde was directly added to the culture to a final concentration of 2% to form cross-linked protein⅐DNA complexes. Cells were incubated at room temperature for 15 min with occasional swirling. Formaldehyde crosslinks were quenched by addition of 2.5 M glycine to a final concentration of 250 mM. Cells were then lysed in chromatin immunoprecipitation (ChIP) lysis buffer (50 mM HEPES (pH 7.5), 140 mM NaCl, 1% Triton X-100, 0.1% sodium deoxycholate, and protease inhibitors), and the chromatin was sheared by sonication (Fisher Model 550 sonic dismembrator). The sheared chromatin was then immunoprecipitated using anti-tandem affinity purification (TAP) antibody (Open Biosystems) and protein A-agarose beads (Santa Cruz Biotechnology, Inc.). The beads were then washed with ChIP wash buffer (10 mM Tris (pH 8.0), 250 mM LiCl, 0.5% Nonidet P-40, 0.5% sodium deoxycholate, and 1 mM EDTA), and precipitates were eluted. The eluted precipitates were incubated overnight at 65°C to reverse cross-links. DNA was then precipitated and subjected to PCR analysis.
Co-immunoprecipitation-All immunoprecipitation assays were performed using lysed spheroplasts. In brief, 100 ml of cells were harvested at A 600 nm ϭ 0.8 -1.0 and resuspended in Tris-HCl (pH 8) and 10 mM dithiothreitol. Cells were incubated at room temperature for 10 min, centrifuged, and resuspended in 10 ml of spheroplast buffer (1 M sorbitol in 0.5ϫ YPD medium and 50 mM KPO 4 ) with oxylyticase and incubated at 30°C for 30 min on a shaker. After chilling on ice, cell suspensions were overlaid on a sucrose cushion (20 mM HEPES, 1.2 M sucrose, and 0.02% sodium azide) and pelleted by centrifugation at 3000 ϫ g for 25 min at 4°C. The resulting spheroplasts (treated and untreated with 1 mM hydrogen peroxide at 30°C for 40 min) were suspended in intracellular (lysis) buffer (100 mM potassium acetate, 50 mM KCl, 20 mM PIPES, and 200 mM sorbitol (pH 6.8)). Protein lysates were prepared by grinding the spheroplasts in the presence of glass beads, followed by centrifugation at 10,000 ϫ g with recovery of the supernatant. These lysates were precleared by rotation with protein A-agarose mixture (100 l/ml of lysate) at 4°C for 1 h, followed by centrifugation. Precleared lysates were rotated with 4 g of anti-TAP antibody for 3 h at 4°C. For immunoprecipitation, 50 l of protein A-agarose were added to samples and further rotated for 1 h at 4°C. The beads were then spun down at 10,000 ϫ g for 2 min and washed twice with intracellular buffer. The immunoprecipitated proteins were recovered by adding Twirl buffer (8 M urea, 5% SDS, 10% glycerol, and 50 mM Tris (pH 6.8)). These precipitates were then loaded onto 12% polyacrylamide gel and probed with anti-Yap1p antibody (14) to detect Yap1p or with peroxidase-conjugated anti-peroxidase antibody to detect TAP-tagged Rox3p as described (27). Aliquots of the protein lysates prior to immunoprecipitation were also electrophoresed and blotted for Yap1p to ensure that equivalent levels of Yap1p were used in the co-immunoprecipitation assays.

Differential Role of the Yap1p C Terminus in Oxidative Stress
Induction-The importance of the C terminus of Yap1p in the oxidantregulated nuclear localization of this factor has been well documented (10,11,13,16). Previously, we demonstrated that normal function of the C-terminal region of Yap1p is crucial for H 2 O 2 -induced activation of the thioredoxin-encoding TRX2 gene but not an artificial Yap1p-responsive reporter gene (13). A C-terminal mutant form of Yap1p lacking a cysteine residue in the C-CRD (C629A Yap1p) is constitutively localized to the nucleus and fails to normally induce TRX2 expression upon H 2 O 2 challenge (13). We compared the ability of this mutant Yap1p derivative to control expression of the oxidant-inducible GSH1 gene (28), encoding ␥-glutamylcysteine synthetase, the first step in glutathione biosynthesis (29). GSH1 transcription has been previously shown to be oxidant-responsive in a Yap1p-dependent fashion (8,28). A yap1⌬ mutant strain was transformed with low-copy-number plasmids expressing either wild-type or C629A Yap1p along with lacZ gene fusions to either TRX2 or GSH1. Appropriate transformants were then tested for ␤-galactosidase expression in the presence or absence of oxidative stress (Fig. 1).
GSH1-lacZ expression was elevated by ϳ2and 3.5-fold upon oxidative stress induced by H 2 O 2 or diamide, respectively, in the presence of wild-type Yap1p, consistent with the findings of other groups (28,30). Expression of C629A Yap1p increased GSH1 expression by 2-fold in the absence of oxidant, and this expression failed to increase in the face of H 2 O 2 exposure and was only modestly enhanced during diamide stress. Notably, expression of GSH1 during H 2 O 2 treatment was identical in the presence of either wild-type or C629A Yap1p. This behavior contrasts with a TRX2-lacZ fusion gene that showed a striking expression defect in the presence of C629A Yap1p during H 2 O 2 stress. C629A Yap1p was unable to significantly induce TRX2 upon H 2 O 2 challenge. We believe that this selective inability to drive TRX2 expression in the presence of H 2 O 2 is at the heart of the failure of C-terminal mutant forms of Yap1p to support normal resistance to this oxidant (13, 14, 16, 31).

FIGURE 1. Yap1p C-CRD function is differentially required at authentic target genes.
A strain containing a yap1⌬ allele was transformed with plasmids containing lacZ fusions to the wild-type GSH1 (upper panel) or TRX2 (lower panel) promoter region and ATG codons along with a vector plasmid or this same clone expressing wild-type or C629A Yap1p. These lacZ reporter plasmids have been described previously (21). All plasmids were maintained at low-copy number by use of a CEN/ARS replicon. Transformants were grown to mid-log phase, split into three equal aliquots, and then grown for additional time in the same medium (Ϫ) or in the presence of H 2 O 2 or diamide. Cultures were assayed for ␤-galactosidase activity as described (21).
This analysis demonstrated that the wild-type Yap1p C-CRD is required for significant induction of TRX2 but not GSH1 during H 2 O 2induced oxidative stress, indicating that TRX2 places a unique requirement on the C terminus of Yap1p that is not shared by all Yap1p target genes. To explore this unique demand on Yap1p at the TRX2 promoter, we characterized the structure of this promoter.
YREs in TRX2-Previous work suggested the presence of two YREs in the 5Ј-noncoding region of TRX2 located at positions Ϫ218 and Ϫ181 upstream of the ATG codon (7). A double mutant promoter lacking both of these YREs was unable to respond to Yap1p. To determine the individual contribution of these two putative YREs to Yap1p regulation of TRX2, we constructed a set of mutant TRX2 promoters lacking either one or both of the TRX2 YREs. We carried out DNase I protection analysis using bacterially produced Yap1p to determine whether both YREs were bound by Yap1p in vitro and to ensure that the mutant sites were no longer recognized by the recombinant protein (Fig. 2).
The YRE at position Ϫ181 was strongly protected from DNase I cleavage by Yap1p. Only very weak interaction was detected at the Ϫ218 YRE, and the mutant YREs failed to detectably bind this factor in vitro.
To assess the functional contribution of these YREs to TRX2 expression, two analyses were carried out. First, each mutant promoter was fused to lacZ to facilitate measurement of gene expression. Second, each mutant promoter was placed in the context of the wild-type TRX2 gene to evaluate the effect of removal of the YREs on the ability of the resulting constructs to complement the H 2 O 2 sensitivity of a trx2⌬ strain. Both the lacZ fusions and the TRX2 clones were returned to yeast on lowcopy-number plasmids, and appropriate assays were carried out (Fig. 3).
Each TRX2-lacZ fusion plasmid was introduced into either wild-type or yap1⌬ cells, and ␤-galactosidase activity was determined in the absence (no stress) or presence of H 2 O 2 . Loss of the single YRE located at position Ϫ181 prevented the Yap1p-dependent activation of TRX2 upon H 2 O 2 challenge. Elimination of the YRE at Ϫ218 had no effect on H 2 O 2 induction of TRX2 transcription, consistent with the lack of in vitro binding. The wild-type TRX2 promoter was induced by 300%, whereas a double mutant lacking both YREs was insensitive to H 2 O 2 exposure.
Similarly, loss of the Ϫ181 YRE from the wild-type TRX2 gene failed to normally correct the H 2 O 2 -sensitive defect of a trx2⌬ strain. However, an identical clone either containing both YREs or lacking only the Ϫ218 YRE restored normal H 2 O 2 resistance to this strain. Together, these data argue that only the Ϫ181 YRE is required for Yap1p-mediated activation of TRX2 by and resistance to H 2 O 2 . Having localized the site for Yap1p action at TRX2, we mapped the region of this promoter that requires normal Yap1p C-terminal function to permit wild-type induction by H 2 O 2 .
A Short Region in the TRX2 Promoter Blocks Downstream Promoter Activation by a C-terminal Form of Yap1p-Our previous experiments defined a region of TRX2 from positions Ϫ141 to Ϫ61 that, when present between the upstream region of TRX2 and a heterologous downstream reporter gene (CYC1-lacZ), prevents C629A Yap1p from inducing ␤-galactosidase expression in response to H 2 O 2 induction (13). Wild-type Yap1p is still able to normally elevate expression during H 2 O 2 exposure. To more precisely localize the segment of TRX2 responsible for this effect, we produced several new deletion mutants and placed them upstream of the CYC1-lacZ reporter. These constructs all had a common 5Ј-end (position Ϫ237) and extended to variable positions closer to the TRX2 ATG codon. Each deletion construct was FIGURE 2. Yap1p can bind to both YREs in vitro. Yap1p was produced in bacterial cells as described previously (25). TRX2 templates were radiolabeled with 32 P, incubated with 10 or 20 l of recombinant protein as indicated, and briefly digested with DNase I. Deproteinized DNA fragments were resolved by electrophoresis on urea-polyacrylamide gels and detected by autoradiography. Either the wild-type TRX2 promoter or a double mutant form lacking both YREs (dm TRX2) was used. The location of each YRE is indicated by the boxes to the right. Protein extract from a bacterial cell carrying the empty expression vector (Vector) was used as a negative control for Yap1p binding activity.  mYRE). B, the same series of TRX2 promoter mutations was placed upstream of an otherwise wild-type version of the TRX2 gene carried on a low-copy-number plasmid. These clones were transformed into a trx2⌬ strain and placed in YPD medium containing a gradient of H 2 O 2 as indicated. Transformants were grown to mid-log phase, and 1000 cells of each were placed as 5-l spots along the gradient and allowed to dry. The plate was then incubated at 30°C and photographed when the spots developed.
introduced into cells expressing either wild-type or C629A Yap1p, and appropriate transformants were then tested for their ability to respond to H 2 O 2 stress (Fig. 4).
Deletion of region Ϫ120 to Ϫ141 produced a reporter gene that was strongly elevated by the presence of C629A Yap1p irrespective of the presence of H 2 O 2 . Clones that contained 3Ј-ends at position Ϫ120, Ϫ104, or Ϫ90 were not responsive to the presence of C629A Yap1p. All of these clones could be induced by wild-type Yap1p upon H 2 O 2 challenge.
Inspection of TRX2 promoter sequence Ϫ141 to Ϫ120 indicated the presence of an element (TATAA) resembling a typical TATA box (32). We carried out several experiments to determine whether this putative TATA box might actually play this role at TRX2. First, the transcription start site was mapped by primer extension mapping to ensure that the position of this element was located upstream of the TRX2 mRNA start site. Second, we tested the ability of Tbp1p to bind to this element. Third, we assessed the function of this putative TATA box both in the wild-type TRX2 context as well as upstream of CYC1-lacZ.
To map the TRX2 start site, total RNA was prepared from trx1⌬ and trx2⌬ cells. A TRX2-specific DNA primer was radiolabeled with 32 P and then annealed to equal amounts of these RNAs. The resulting DNA/ RNA duplexes were extended by reverse transcription. Extension products were then electrophoresed on urea-polyacrylamide gel next to a DNA sequencing ladder from a TRX2 template using the same primer and visualized by autoradiography (Fig. 5).
Two major start sites were detected for TRX2 mRNA at positions Ϫ47 and Ϫ43 upstream of the ATG codon for the open reading frame. Identical primer extension reactions performed with RNA from trx2⌬ cells or with yeast tRNA failed to produce detectable extension products. This analysis suggested that the site for TRX2 transcription initiation is ϳ80 bp downstream from the putative TATA element described above and is in a reasonable location based on studies of other S. cerevisiae promoters (reviewed in Ref. 33).
To provide direct evidence that Tbp1p can interact with the TATA element at position Ϫ120, we carried out an electrophoretic mobility shift assay using recombinant Tbp1p purified from Escherichia coli. The TRX2 promoter was radiolabeled and incubated with Tbp1p. The resulting protein⅐DNA complexes were then separated in a nondenaturing gel system and visualized by autoradiography (Fig. 5). 10-Fold molar excesses of unlabeled competitor DNAs were included to assess the specificity of the protein⅐DNA complexes that were observed to form.
Tbp1p was able to form a protein⅐DNA complex with the wild-type TRX2 promoter region. Formation of this complex was prevented by including an excess of either the unlabeled TRX2 fragment or the same fragment containing the TRP5 TATA element. However, the presence of the TRX2 fragment containing the mutant TATA (mTATA) element failed to prevent formation of the Tbp1p⅐TRX2 complex. The results from these analyses are consistent with the TRX2 sequence at position Ϫ120 or its TRP5 cognate serving as a binding site for Tbp1p. The mTATA element failed to compete with the wild-type TRX2 TATA site even when present at a 10-fold molar excess. This finding supports the view that Tbp1p is unable to interact with the mTATA element.
TRX2 TATA Element Mediates Interaction with the Yap1p C Terminus-To directly evaluate whether the TRX2 TATA element serves to regulate the ability of Yap1p to activate downstream CYC1-lacZ expression, we removed this Tbp1p-binding site from a TRX2-CYC1-lacZ fusion gene by introducing the site-directed TATA element described above. The TRX2 TATA element was removed from a TRX2 DNA fragment extending from positions Ϫ237 to Ϫ90 and cloned upstream of CYC1-lacZ to produce a fusion gene referred to as mTATA-TRX2-CYC1-lacZ. This plasmid is identical to the Ϫ237/Ϫ90 TRX2-CYC1-lacZ clone used in the analysis described for Fig. 4, except that the TRX2 TATA element has been removed. We also replaced the TRX2 TATA element with the analogous sequence from the TRP5 promoter to form mTATA TRP5 -TRX2-CYC1-lacZ to determine whether the precise nucleotide sequence around the TRX2 TATA element is required for its influence on Yap1p. We first analyzed the function of the TRX2 mTATA and mTATA TRP5 replacements in the context of the wild-type TRX2 gene. These two TATA mutants were produced by site-directed mutagenesis in an otherwise wild-type TRX2-lacZ fusion. The desired mutants were introduced along with control plasmids into a yap1⌬ strain along with a low-copy-number plasmid expressing wildtype or C629A Yap1p. Appropriate transformants were assayed for ␤-galactosidase activity under non-stressed or H 2 O 2 -challenged conditions (Fig. 6B).
The mTATA element inactivated the TRX2 promoter as measured by ␤-galactosidase production. This further supports our assignment of the element at position Ϫ120 as the TATA box for TRX2. Insertion of the TRP5 TATA box in place of the normal TRX2 sequence fully  (54), and the binding site for the TATA box-binding protein Tbp1p. The lines on the left indicate the extent of the TRX2 promoter present in each clone. The boxes labeled CYC1-lacZ denote fusions between a CYC1 promoter lacking its normal UASs and lacZ. Each TRX2-CYC1-lacZ fusion plasmid was introduced into yap1⌬ cells along with a low-copy-number plasmid expressing either wild-type (wt) or C629A Yap1p. Transformants were assayed for ␤-galactosidase activity in the presence or absence of H 2 O 2 as described in the legend to Fig. 1. restored both expression and regulation of the TRX2-lacZ fusion. Notably, the mTATA TRP5 -TRX2-lacZ fusion gene was still unable to respond to C629A Yap1p, indicating that the requirement for the intact C terminus was maintained by this altered TRX2 TATA element.
Each mTATA element was then tested for its effect when placed between the TRX2 UASs and the CYC1-lacZ reporter gene. The mTATA-TRX2-CYC1-lacZ and mTATA TRP5 -TRX2-CYC1-lacZ plasmids were transformed along with the expression plasmid for wild-type or C629A Yap1p as described above. The CYC1-lacZ reporter plasmid lacking any inserted UAS elements was also introduced as a control. Transformants were tested for their ability to respond to H 2 O 2 challenge as measured by ␤-galactosidase activity (Fig. 6C).
Removal of the TATA element from fragment Ϫ237 to Ϫ90 of the TRX2 promoter eliminated the impediment to C629A Yap1p activation of CYC1-lacZ that was seen if the TATA element was present in this assay. Similar to the behavior of intact region Ϫ237 to Ϫ90, mTATA TRP5 -TRX2-CYC1-lacZ was unable to respond to C629A Yap1p but was normally induced in the presence of Yap1p. The fact that the TRP5 TATA element recapitulates the necessity of a wild-type Yap1p C terminus for H 2 O 2 inducibility suggests that the precise sequence of the TATA element is not essential, but the presence of a functional TATA motif is required.
These data support the view that the TRX2 TATA element is involved in placing a special demand on the C terminus of Yap1p for activation of TRX2 expression that is not required for induction of transcription of other Yap1p target genes like GSH1. To explore the nature of this unique requirement of TRX2 for C-terminally intact Yap1p, we examined the ability of Yap1p to recruit mediator components to TRX2.
Mediator Components Selectively Influence Yap1p Target Gene Expression-Genetic and biochemical experiments in yeast and mammalian cells have provided a wealth of information in terms of identifying components required for activator-dependent gene transcription. A key participant in permitting communication between DNA-bound activators and the general transcription machinery is the multiprotein mediator complex (recently reviewed in Ref. 34). Genetic ablation of mediator subunits is often lethal in yeast, consistent with the important role of this complex in gene transcription (reviewed in Ref. 35). Because we believe that interaction between YRE-bound Yap1p and the transcription machinery assembling on the TRX2 TATA box is a critical determinant in supporting H 2 O 2 induction of this gene, we tested the role of mediator components in Yap1p control of TRX2 transcription. We focused this study on mediator subunits corresponding to components that define each of the three putative mediator modules and that are nonessential for growth of cells. Strains lacking GAL11 (36), MED2 (37), MED1 (38), MED9 (39), and ROX3 (40) were analyzed for their role in H 2 O 2 tolerance by assessing their ability to grow in the presence of this oxidant. Isogenic strains were grown to mid-log phase, and equal numbers of cells were placed on solid medium containing a gradient of H 2 O 2 or cycloheximide. Cycloheximide resistance was used as a control for the general resistance of each strain, as Yap1p is not required for resistance to this toxin in wild-type cells (22).
Loss of ROX3 produced a dramatic decrease in H 2 O 2 tolerance (Fig.  7), whereas mutants lacking GAL11 or MED2 were not altered in their ability to tolerate this oxidant. Mutants lacking YAP1 were more sensitive to H 2 O 2 than were isogenic rox3⌬ strains, suggesting that some residual activation via Yap1p could occur even in the absence of this mediator component. Only gal11⌬ strains were found to be sensitive to cycloheximide exposure, arguing against the interpretation that the oxidant sensitivity of rox3⌬ cells represents a generalized stress defect. Mutants lacking the mediator component MED1 or MED9 did not exhibit significant reductions in resistance to either compound, further supporting the idea that the H 2 O 2 sensitivity of rox3⌬ strains represents a specific consequence of loss of this mediator subunit (data not shown).
To determine whether rox3⌬ cells are affected in their ability to induce TRX2, we introduced a TRX2-lacZ fusion gene into isogenic wild-type, rox3⌬, gal11⌬, and med2⌬ cells and assayed expression of this construct under normal or diamide-or H 2 O 2 -stressed conditions. Loss of ROX3 blocked induction by H 2 O 2 and reduced the level of TRX2 expression seen in non-stressed cells. Diamide inducibility was reduced but could still be detected. Deletion of the other mediator components did not cause a significant effect on TRX2 expression. This analysis indicated that Rox3p plays a uniquely important role in TRX2 gene activation. A, primer extension mapping of the TRX2 transcription start site. Total RNA was prepared from the indicated yeast strains lacking either TRX1 or TRX2 along with yeast tRNA as a negative control. A 5Ј-32 Plabeled oligonucleotide primer corresponding to TRX2 sequences within the coding sequence was annealed to each sample. Reverse transcriptase and dNTPs were added, and the reaction was held at 37°C for 1 h. RNase was added, and the samples were deproteinized and electrophoresed on urea-polyacrylamide gel. The same TRX2 primer was used to generate a sequencing ladder from a TRX2 plasmid template using the Sequenase DNA sequencing kit. The two major start sites detected are shown on the right along with the corresponding TRX2 DNA sequence. B, Tbp1p binding to TRX2. Electrophoretic mobility shift assays were carried out as described (26). A BglII/EcoRI fragment corresponding to the wild-type TRX2 promoter (wt) was labeled with 32 P at the BglII site using T4 polynucleotide kinase. Purified Tbp1p prepared from an E. coli strain expressing polyhistidine-tagged Tbp1p was included in the indicated reactions along with the radiolabeled TRX2 probe. 10-Fold molar excesses of unlabeled competitor DNAs were included to establish binding specificity.

Only Properly Oxidatively Folded Yap1p Can Recruit Rox3p during H 2 O 2 Stress-
The identification of Rox3p as a key mediator subunit involved in H 2 O 2 activation of TRX2 provides us with a new tool to compare the ability of different Yap1p mutants to bring Rox3p-containing mediator subunits to the TRX2 promoter. We hypothesized that the H 2 O 2 -induced folding of Yap1p is required both to allow nuclear accumulation of the factor and to permit this form of Yap1p to induce TRX2 expression. Because Rox3p appears to be a critical mediator component required for H 2 O 2 induction of TRX2 transcription, we used ChIP to compare the ability of different forms of Yap1p to recruit Rox3p to TRX2. Since H 2 O 2 -induced formation of correctly folded Yap1p requires disulfide bonds to form between cysteine residues at positions 303 and 598 and positions 310 and 629 (16,17), we used C303A and C629A Yap1p to prevent normal H 2 O 2 -triggered folding. Wild-type and mutant forms of Yap1p were expressed from low-copy-number plasmids, and transformants were challenged with H 2 O 2 or left untreated.
These plasmids were transformed into a strain expressing TAP-tagged Rox3p and lacking the YAP1 gene. Rox3p recruitment to the TRX2 promoter was then evaluated by ChIP (Fig. 8). Rox3p association with the PDR5 promoter was also compared as a specifity control because rox3⌬ strains exhibit normal cycloheximide resistance, a phenotype requiring expression of PDR5 (41). We also examined Rox3p association with the ATR1 gene, a known Yap1p regulatory target (42). ATR1 expression was stimulated by the presence of C629A Yap1p but was not H 2 O 2 -inducible. 5 The association of Rox3p with TRX2 was elevated in H 2 O 2 -challenged compared with non-stressed cells in the presence of wild-type Yap1p (Fig. 8). This H 2 O 2 -dependent stimulation of Rox3p recruitment was not seen if either C303A or C629A Yap1p was expressed. Rox3p was 5 K. Gulshan and W. Scott Moye-Rowley, unpublished data. Low-copy-number plasmids expressing either wild-type or C629A Yap1p were also transformed with each lacZ reporter, and co-transformants were assayed for ␤-galactosidase activity as described in the legend to Fig. 1. C, a DNA fragment corresponding to region Ϫ237 to Ϫ90 from the TRX2 promoter containing either the mTATA element or the TRP5 TATA box was placed upstream of a CYC1-lacZ fusion gene lacking its normal UASs. These TRX2-CYC1-lacZ fusion plasmids were introduced into yap1⌬ cells along with plasmids expressing wild-type or C629A Yap1p and assayed for ␤-galactosidase expression as described in the legend to Fig. 1. The CYC1-lacZ fusion plasmid with no inserted TRX2 fragment (No UAS) was assayed as a control.
found to poorly associate with PDR5 under all these conditions, consistent with the lack of an effect on cycloheximide resistance seen in the rox3⌬ strain. ATR1 association with Rox3p was stimulated by the presence of C629A Yap1p, but this stimulation was not responsive to H 2 O 2 . This is consistent with previous data demonstrating that C629A Yap1p drives higher expression of ATR1 with a concomitant increase in Atr1pmediated drug resistance (42). These data demonstrate that the Yap1pdependent recruitment of Rox3p is promoter-specific and that only normally oxidatively folded Yap1p can stimulate Rox3p association with TRX2.

Requirement of Oxidative Protein Folding for Yap1p/Rox3p
Interaction-Although the ChIP data demonstrated that Rox3p is recruited in a Yap1p-dependent fashion to the TRX2 promoter during H 2 O 2 stress, further interpretation of this finding is not straightforward. For example, Rox3p may be recruited to TRX2 via some other transcription factor (like Skn7p) in a fashion parallel to Yap1p activity. Alternatively, Rox3p may arrive at TRX2 owing to the presence of this mediator component in the large Srb4p module (43). To explore the nature of the Yap1p/Rox3p interaction, co-immunoprecipitation reactions were performed. A ROX3-TAP fusion gene was obtained (44) and introduced  Only the wild-type protein (WT) can form the dually disulfide-bonded N-and C-CRD structures upon H 2 O 2 stress, whereas neither the C-CRD mutant (C629A) nor the N-CRD mutant (C303A) is capable of normally forming this interdomain structure. Transformants were grown to mid-log phase, and equal cell numbers were then left unstressed (Ϫ) or were challenged by addition of H 2 O 2 (ϩ). ChIP was performed as described previously (55), but IgG-Sepharose beads coupled to anti-TAP antibody were used to recover the Rox3p-TAP-containing complexes. After precipitation and washing, the precipitated chromatin was analyzed by PCR using primers that detect the TRX2, PDR5, and ATR1 promoters. Input represents 1% of the total, whereas the immunoprecipitated (IP) samples represent 20% of the precipitated chromatin.
into a yap1⌬ strain. Low-copy-number plasmids expressing different forms of Yap1p were then transformed into this background. Appropriate transformants were grown to early log phase, and spheroplasts were produced. Spheroplasts were incubated at 25°C in the presence or absence of 1 mM H 2 O 2 and then lysed with glass beads. Aliquots of this initial protein lysate were withdrawn to serve as controls for the levels of Yap1p in each sample. The remaining lysate was incubated with protein A-agarose coupled to anti-TAP antibody. Immunoprecipitates were recovered by centrifugation and eluted from the agarose beads by resuspension in Twirl buffer. Equal aliquots of the immunoprecipitates and input fractions were electrophoresed on SDS-polyacrylamide gel and transferred to nitrocellulose membranes. Membranes were then probed with antiserum against Yap1p or the TAP epitope (Fig. 9).
Yap1p was efficiently immunoprecipitated by Rox3p-TAP when wild-type cells were challenged with H 2 O 2 but not in the absence of oxidative stress. C303A Yap1p was not found in a complex with Rox3p-TAP under either condition, whereas C629A Yap1p associated with Rox3p-TAP under both control and stressed conditions. The association of C629A Yap1p and Rox3p-TAP did not increase in response to H 2 O 2 exposure. We believe that this oxidant-independent association of C629A Yap1p and Rox3p-TAP occurs due to the constitutive nuclear localization of this mutant form of Yap1p and its previously demonstrated ability to form oxidized Yap1p, albeit at a reduced level (11,12). C303A Yap1p appears to lack this ability to produce any detectable oxidized form of Yap1p.
These data support the idea that Yap1p must be properly oxidatively folded if it is to associate with and recruit Rox3p to the TRX2 promoter in response to H 2 O 2 challenge. Additionally, the ability to co-immunoprecipitate Yap1p with Rox3p suggests that the interaction between these proteins represents formation of a stable complex involved in transactivation of TRX2 and likely other target genes. Formation of this complex containing Rox3p is critical for the Yap1p-dependent response to H 2 O 2 but not diamide-induced oxidative stress.

DISCUSSION
A key feature regulating the function of Yap1p during oxidative stress is the control of the subcellular localization of this factor. Experiments from a number of laboratories have provided evidence that activation of Yap1p function by oxidants like diamide and H 2 O 2 occurs via different mechanisms (13)(14)(15)(16)45). Significantly, disulfide bond formation between a pair of cysteine residues located in the N-and C-CRDs is required for normal H 2 O 2 -induced movement to the nucleus (16,17). The data reported here provide a second essential role for these disulfide-bonded N-and C-CRDs: recruitment of the mediator component Rox3p to the TRX2 promoter. Strains that cannot properly assemble the disulfide-bonded N-and C-CRDs or that lack Rox3p also fail to induce TRX2 upon H 2 O 2 challenge.
These findings argue that the dually disulfide-bonded domain in Yap1p is a bifunctional modification to the factor. First, this newly formed domain of the protein masks the nuclear export signal of Yap1p such that it is prevented from interacting with the Crm1p exportin that would otherwise remove this transcription factor from the nucleus (11,12). Second, formation of this domain allows Yap1p to induce TRX2 expression, an ability required for normal H 2 O 2 tolerance (13,16). Rox3p has been previously implicated in the heat shock and osmotic stress induction of CYC7 gene expression (46). The data reported here extend the role of Rox3p to the oxidative stress response.
Although Rox3p recruitment is used as a measure of TRX2 gene activation by Yap1p, we do not know if this is a direct interaction or if Rox3p is brought to TRX2 via association in a complex with another mediator factor that associates directly with Yap1p. Rox3p has been associated with the Srb4p mediator module by biochemical and genetic criteria (43,47). ChIP experiments using a Gal11p-TAP construct indicate that, although this mediator subunit can be found at TRX2, the degree of Gal11p association with this promoter is altered neither by the form of Yap1p present or by oxidative stress (data not shown). This observation is supported by the wild-type H 2 O 2 resistance phenotype and TRX2 inducibility seen in gal11⌬ cells (Fig. 7). These data argue that the selective Srb4p/Rox3p mediator module requirement at TRX2 explains the observed oxidant-specific behavior of Yap1p mutants described previously (13,16). The intact Yap1p N-and C-CRDs must be present to permit activation of TRX2 during H 2 O 2 challenge. Successful recruitment of this mediator module is a necessary precursor to H 2 O 2induced TRX2 transcription.
The use of TRX2-CYC1 promoter fusion plasmids provides insight into the mechanism of TRX2 activation during H 2 O 2 stress. When the TRX2 (or the closely related TRP5) TATA box is interposed between the TRX2 UAS and the CYC1 TATA element, only wild-type Yap1p can trigger high-level expression of the downstream CYC1-lacZ fusion gene. However, removal of the TRX2 TATA segment allows C629A Yap1p to drive high levels of CYC1-lacZ expression irrespective of the presence of H 2 O 2 . We interpret this to argue for the presence of a negatively acting complex that requires the presence of the TRX2 TATA box for its action. This putative negative regulation does not appear to occur at GSH1 because this gene can respond to the presence of C629A Yap1p and may be due to the different upstream factors and/or TATA sequence present at GSH1. In mutant forms of Yap1p that cannot assemble the correctly oxidatively folded N-and C-CRDs, this negatively acting complex cannot be displaced, and TRX2 induction is compromised. A number of negative regulators that act via the TATA element have been described, including NC2/Bur6p/Ydr1p (48 -50) and Mot1p (51,52). Further experiments will be required to determine the identity of this negative regulator controlling H 2 O 2 induction of TRX2 expression.
H 2 O 2 activation of TRX2 expression represents a relatively complex situation of oxidative stress regulation in which Yap1p participates. Even if Yap1p and its folding machinery are genetically intact, loss of the Skn7p transcription factor prevents normal H 2 O 2 induction of TRX2 (53). This requirement for Skn7p cannot be bypassed by mutants like C629A Yap1p (13) even though this same mutant can drive high-level expression of GSH1 (Fig. 1). The thioredoxin reductase gene TRR1 has FIGURE 9. Yap1p and Rox3p associate in vivo. A yap1⌬ strain containing an integrated ROX3-TAP fusion gene was transformed with a low-copy-number plasmid (pRS315) or the same plasmid carrying the wild-type YAP1 gene. A strain lacking any TAP tag (No Tap) was used as a control. Transformants were grown to mid-log phase, and spheroplasts were prepared by oxylyticase digestion. Equal numbers of spheroplasts were incubated at 30°C in the presence or absence of H 2 O 2 for 45 min. Protein extracts were prepared by glass bead lysis and clarified by centrifugation. Three % of these total extracts (Input) were blotted for the presence of Yap1p using a rabbit polyclonal antibody recognizing Yap1p or Rox3p-TAP. Immunoprecipitations were then carried out using using anti-TAP antibody to recover Rox3p-associated complexes. Immunoprecipitated proteins were analyzed by Western blotting for the presence of Yap1p using anti-Yap1p antiserum. CO-IP, co-immunoprecipitated. also been reported to exhibit the dual Skn7p/Yap1p dependence for its induction by H 2 O 2 (53). Skn7p cooperation with correctly oxidatively folded Yap1p appears to be a central feature in H 2 O 2 activation of these genes involved in thioredoxin homeostasis.
Conversely,althoughGSH1expressionisinduciblebyH 2 O 2 inaYap1pdependent fashion (28), this gene is highly responsive to C629A Yap1p. This constitutive expression of GSH1 in the presence of C629A Yap1p resembles the response of an artificial Yap1p-responsive TRP5-lacZ fusion gene that we have described previously (13,14). This reporter gene consists of three SV40 AP-1 recognition elements placed upstream of a TRP5-lacZ fusion gene that lacks the normal UAS elements of TRP5. The common responses of GSH1 and this AP-1 recognition element-TRP5-lacZ reporter to C629A Yap1p suggest that Yap1p alone can activate expression of GSH1 upon H 2 O 2 stress. Constitutively nuclear C629A Yap1p produced the same levels of GSH1 expression as did H 2 O 2 treatment of wild-type cells, consistent with the notion that nuclear accumulation of Yap1p is adequate for full induction of GSH1. This is contrary to the situation at TRX2, in which Yap1p must both accumulate in the nucleus and be properly folded for normal H 2 O 2 activation to occur.