Peroxiredoxin Chaperone Activity Is Critical for Protein Homeostasis in Zinc-deficient Yeast* ♦

Background: Zinc is required as a structural cofactor for the folding of many proteins. Results: The chaperone activity of the Tsa1 peroxiredoxin is essential for protein homeostasis and growth of zinc-deficient yeast. Conclusion: Zinc limitation disrupts protein homeostasis, and cells need Tsa1 for tolerance. Significance: Disrupted protein homeostasis is a major and previously unrecognized stress of zinc deficiency. Zinc is required for the folding and function of many proteins. In Saccharomyces cerevisiae, homeostatic and adaptive responses to zinc deficiency are regulated by the Zap1 transcription factor. One Zap1 target gene encodes the Tsa1 peroxiredoxin, a protein with both peroxidase and protein chaperone activities. Consistent with its regulation, Tsa1 is critical for growth under low zinc conditions. We previously showed that Tsa1's peroxidase function decreases the oxidative stress that occurs in zinc deficiency. In this report, we show that Tsa1 chaperone, and not peroxidase, activity is the more critical function in zinc-deficient cells. Mutations restoring growth to zinc-deficient tsa1 cells inactivated TRR1, encoding thioredoxin reductase. Because Trr1 is required for oxidative stress tolerance, this result implicated the Tsa1 chaperone function in tolerance to zinc deficiency. Consistent with this hypothesis, the tsa1Δ zinc requirement was complemented by a Tsa1 mutant allele that retained only chaperone function. Additionally, growth of tsa1Δ was also restored by overexpression of holdase chaperones Hsp26 and Hsp42, which lack peroxidase activity, and the Tsa1 paralog Tsa2 contributed to suppression by trr1Δ, even though trr1Δ inactivates Tsa2 peroxidase activity. The essentiality of the Tsa1 chaperone suggested that zinc-deficient cells experience a crisis of disrupted protein folding. Consistent with this model, assays of protein homeostasis suggested that zinc-limited tsa1Δ mutants accumulated unfolded proteins and induced a corresponding stress response. These observations demonstrate a clear physiological role for a peroxiredoxin chaperone and reveal a novel and unexpected role for protein homeostasis in tolerating metal deficiency.

Zinc is required for the folding and function of many proteins. In Saccharomyces cerevisiae, homeostatic and adaptive responses to zinc deficiency are regulated by the Zap1 transcription factor. One Zap1 target gene encodes the Tsa1 peroxiredoxin, a protein with both peroxidase and protein chaperone activities. Consistent with its regulation, Tsa1 is critical for growth under low zinc conditions. We previously showed that Tsa1's peroxidase function decreases the oxidative stress that occurs in zinc deficiency. In this report, we show that Tsa1 chaperone, and not peroxidase, activity is the more critical function in zinc-deficient cells. Mutations restoring growth to zinc-deficient tsa1 cells inactivated TRR1, encoding thioredoxin reductase. Because Trr1 is required for oxidative stress tolerance, this result implicated the Tsa1 chaperone function in tolerance to zinc deficiency. Consistent with this hypothesis, the tsa1⌬ zinc requirement was complemented by a Tsa1 mutant allele that retained only chaperone function. Additionally, growth of tsa1⌬ was also restored by overexpression of holdase chaperones Hsp26 and Hsp42, which lack peroxidase activity, and the Tsa1 paralog Tsa2 contributed to suppression by trr1⌬, even though trr1⌬ inactivates Tsa2 peroxidase activity. The essentiality of the Tsa1 chaperone suggested that zinc-deficient cells experience a crisis of disrupted protein folding. Consistent with this model, assays of protein homeostasis suggested that zinclimited tsa1⌬ mutants accumulated unfolded proteins and induced a corresponding stress response. These observations demonstrate a clear physiological role for a peroxiredoxin chaperone and reveal a novel and unexpected role for protein homeostasis in tolerating metal deficiency.
Zinc plays important roles in biology as a catalytic cofactor for many enzymes. In addition, zinc is a structural cofactor mediating the folding of many protein domains (1). Approximately 10% of proteins encoded by the human genome are estimated to require zinc for their folding and function (2). Critical zinc-containing proteins include zinc finger transcription factors such as TFIIIA (3). Another such regulator is the yeast Zap1 protein, which activates gene expression in response to zinc deficiency. Zap1 regulates expression of ϳ80 genes grouped into two broad categories as follows: those involved in maintaining zinc homeostasis (e.g. zinc transporters), and those that help the cell adapt to the stress of zinc deficiency (4 -6). Adaptive responses include altering phospholipid synthesis pathways to maintain their production (7) and suppressing sulfate assimilation to preserve NADPH for use in antioxidant pathways (8).
We previously reported that zinc-limited yeast show increased oxidative stress (9,10). Increased levels of reactive oxygen species (ROS) 2 have been observed in zinc-deficient cultured mammalian cells as well as whole animals (11,12). Thus, despite not being redox-active, zinc plays an antioxidant role. The source of the increased ROS in deficient cells is unknown. Some Zap1 target genes may be induced to counteract oxidative stress; for example, CTT1 encodes cytosolic catalase, and TSA1 is a Zap1-regulated gene that encodes a cytosolic peroxidase (9). Tsa1 is a member of the peroxiredoxin family that reduces hydrogen peroxide and organic hydroperoxides using electrons supplied by the thioredoxin/thioredoxin reductase pathway (13). Consistent with a role for Tsa1 in counteracting oxidative stress, a zinc-deficient tsa1⌬ mutant accumulated more ROS (9).
All peroxiredoxins have a conserved cysteine (Cys) at their N terminus (the "peroxidatic" cysteine) that reacts with hydrogen peroxide and is oxidized to the sulfenic acid (-SOH) form during the peroxidase catalytic cycle. Peroxiredoxins can be divided into subclasses depending on the number of cysteines that act in peroxide reduction (14). Tsa1 and its paralog in yeast Tsa2 are members of the 2-Cys subclass (15,16). 2-Cys peroxiredoxins have a second cysteine near their C terminus, the "resolving" cysteine, which reacts with the peroxidatic sulfenic acid intermediate to produce a disulfide-bonded dimer. The disulfide is then reduced to the active thiols by the thioredoxin/ thioredoxin reductase pathway to reactivate the peroxidase (17,18). Thus, thioredoxin reductase and thioredoxin are essential for Tsa1 and Tsa2 peroxidase activity.
In addition to their peroxidase activities, 2-Cys peroxiredoxins like Tsa1 and Tsa2 can also act as molecular chaperones (19,20), possessing a "holdase"-type activity that binds unfolded proteins and prevents their aggregation. Although forms of peroxiredoxin in the peroxidase catalytic cycle have low chaperone activity, hyperoxidation of the peroxidatic cysteine from sulfenic acid to either sulfinic (-SO 2 H) or sulfonic (-SO 3 H) acid results in a transition from lower order complexes (i.e. dimers and decamers) to a higher order "superchaperone" structure (21). The superchaperone form lacks peroxidase activity but shows a dramatic increase in chaperone activity (19,22). Although the chaperone activity of 2-Cys peroxiredoxins has been well characterized in vitro, and they may contribute to tolerance of heat stress (19), their physiological role is not yet clear.
With the goal of identifying the source of ROS in zinc-deficient cells, we isolated mutations that suppressed the growth defect of tsa1⌬ mutants in zinc-deficient conditions, believing such mutations might disrupt pathways responsible for ROS generation. Surprisingly, one such strain carried a loss-of-function mutation in the TRR1 gene, encoding thioredoxin reductase. Trr1 is an essential component of thioredoxin-dependent antioxidant pathways, and cells lacking Trr1 function are more sensitive to oxidative stress. For this reason, we re-examined the role of Tsa1 in zinc-deficient cells. Our analysis revealed that although Tsa1 peroxidase activity decreases oxidative stress in low zinc, the Tsa1 chaperone function is the more critical activity for growth under those conditions. Our obser-vations indicate that Tsa1 protects zinc-deficient cells from defective protein homeostasis.

EXPERIMENTAL PROCEDURES
Yeast Strains, Growth Media, and Standard Methods-All yeast strains used in this work are listed in Table 1. Yeast strains were routinely grown in rich or synthetic medium as described previously (23). For zinc-deficient conditions, synthetic low zinc medium (LZM) was prepared as described previously (24). LZM is zinc-limiting because it contains 1 mM EDTA and 20 mM citrate as metal buffers. In all experiments, LZM ϩ 1 M ZnCl 2 was used as the zinc-deficient condition, and LZM ϩ 1 mM ZnCl 2 was used as the replete condition. To aid growth of S288c-derived mutant strains with strong growth defects (e.g. trr1), LZM was supplemented with amino acids and inositol as described (25). Yeast transformation was performed using standard methods (26). ␤-Galactosidase activity was measured by the method of Guarente (27). Cells were harvested during exponential growth, and activity was calculated as follows: (A 415 ϫ 1000)/(min ϫ ml of culture used ϫ culture A 595 ). lacZ reporter genes with very low activity (e.g. pHSE-lacZ) were assayed using Beta-Glo (Promega).
Construction of Yeast Mutant Strains-The tsa1⌬::LEU2 allele was originally generated by transformation of CWY8 (tsa1⌬:: KanMX4) with a LEU2 marker swap plasmid (kanMX::LEU2) (28). The tsa1⌬::LEU2 marker was transferred to other strains via mating or by PCR amplification and transformation. The tsa2⌬::HphMX4 strains were generated by transformation with a PCR product generated by amplification of the HphMX4 gene from the pAG32 plasmid (29) using oligonucleotides designed to add 82 bases of homology to regions directly flanking the TSA2-coding sequence. The trr1⌬::KanMX4 marker was amplified from a diploid mutant (Invitrogen) and transferred to other strains by transformation. Plasmid Constructions-Plasmids used in this work are listed in Table 2. All plasmids constructed were assembled by gap repair in yeast (30). To construct pHA-TSA2, the TSA2 5Ј-intergenic region and the combined TSA2-coding sequence and 3Ј-intergenic region were amplified from CWY2 genomic DNA as separate PCR products. Primers were designed to include 30 bases of homology to a yeast vector (pFL38) at the 5Ј end of the TSA2 promoter fragment and at the 3Ј end of the coding DNA sequence intergenic fragment. The oligonucleotide used to amplify the 5Ј end of the coding sequence fragment included a region of homology to the 5Ј fragment, followed by an ATG start codon, and two repeats of the HA tag sequence fused 5Ј to the TSA2 coding DNA sequence lacking the native TSA2 start codon. Both fragments were combined with restriction-digested vector and used to transform a yeast strain (CWY2), selecting for URA3 clones. The intact recombinant vector was recovered from the resulting transformants. Two other versions of this plasmid were constructed using the same strategy. To generate pHA-TSA2 Tn , the 5Ј-intergenic fragment was amplified from genomic DNA of a strain carrying the TSA2 Tn allele. To generate pYRE⌬-HA-TSA2, a mutant version of the TSA2 promoter fragment lacking all three YRE sequences was amplified from the pTSA2mYRE1,2,3-lacZ plasmid. pHSP104-GFP was constructed by amplifying the HSP104-GFP fusion from genomic DNA of the EY0986/HSP104-GFP yeast strain (Invitrogen) (31), using primers that included homology to the pFL38 vector. Similarly, to construct pDR-HSP26 and pDR-HSP42, pDR195 was digested with XhoI and BamHI and cotransformed into yeast with ORF-containing PCR products to generate PMA1 promoter-driven alleles.
Isolating Transposon-linked tsa1⌬ Suppressor Mutants-Mutant tsa1⌬ strains in the S288c background (CWY8) or the W303 background (CWM48) were transformed with independently constructed pools of yeast genomic DNA fragments containing random transposon insertions marked with the LEU2 gene (32). Library DNA was digested with NotI before transformation. Insertion mutants were selected on plates lacking leucine and incubated until the appearance of colonies (2 days). Colonies were recovered in liquid SC-leucine medium for 4 h and then used to inoculate cultures in zinc-deficient medium (LZM ϩ 1 M zinc) at an initial A 595 of 0.1. Cultures were grown until A 595 reached ϳ1.0 and then rediluted to an A 595 of 0.02 and grown for 5 days. Transformant pools reaching a final density of Ͼ0.3 A 595 were diluted and plated on SC-leucine plates to isolate single clones. Growth of independent clones in zinc-deficient conditions was assayed to identify suppressed strains. Independent clones were backcrossed to tsa1⌬ mutant strains to follow segregation of the suppression trait and the LEU2-marked transposon in tetrads. Clones showing 2:2 segregation of a strong suppressor trait linked to LEU2 were selected for further analysis. The location of the transposon insertion in the genome of suppressor mutant strains was determined using an inverse PCR method. Briefly, genomic DNA was isolated from the mutant strains and digested with an enzyme (TaqI) that cuts at a known site close to the end of the transposon and at unknown sites nearby in flanking DNA. Genomic DNA fragments were recircularized with DNA ligase, and a fragment containing the insertion site was amplified by inverse PCR from divergent primer sites within the transposon sequence. The PCR fragment was then sequenced to identify the flanking sequence and insertion site.
Pooled Linkage Analysis and Whole Genome Sequencing-Several suppressor mutations segregated as single loci unlinked to the LEU2 transposon marker. These mutations were first classed as recessive or dominant by constructing tsa1⌬/tsa1⌬ sup/ϩ diploid strains and assaying their growth under zincdeficient conditions. Most of the suppressor mutations were recessive. To identify one recessive suppressor mutation (clone 23-2), we used pooled linkage analysis with whole genome sequencing (33). The original mutant strain was backcrossed twice to isogenic tsa1⌬ mutant strains to reduce the number of unrelated mutations. Fourteen tetrads from the second cross were dissected and genotyped to identify 28 haploid segregants each of the tsa1⌬ and tsa1⌬ sup double mutant genotypes. Segregants were grown separately in culture, and an equal number of cells from each class were pooled to give separate tsa1⌬ and tsa1⌬ sup pools. Genomic DNA was extracted from the pools and subjected to whole genome sequencing on an Illumina HiSeq2000 machine using 1ϫ 100-bp reads. Sequence image analysis and base calling were performed using the CASAVA 1.7.0 pipeline (Illumina). Primary analysis of the data (mapping and trimming) was performed with CLC Genomics Workbench 4.7.1 and mapped to the reference S288c sequence build (as of June 2011). Average sequence coverage was ϳ250 reads/ base for both pools. We identified 19 single nucleotide polymorphisms that differed from the reference sequence and were only present in the suppressed DNA pool. Examination of these suppressor-specific single nucleotide polymorphisms indicated that most were likely to be suppressor-specific due to low sequence coverage (10 or fewer reads) at that position. Only one suppressor-specific single nucleotide polymorphism (chromosome IV T1184128A, trr1 L277 *) showed sufficient sequence coverage (204 reads) to be a reliable candidate (data not shown).
Preparation of Protein Extracts-Yeast protein extracts were prepared using a TCA extraction protocol. Cells from 5-ml cultures were collected, washed with water, and resuspended in 1 ml of ice-cold 10% TCA. The cells were collected by centrifugation and either frozen at Ϫ80°C or immediately processed. Cell pellets were resuspended in 200 l of 10% TCA; a 100-l volume of glass beads was added to the tube and the suspension vortexed for 10 min at 4°C. The suspension was removed from the beads and crude protein collected by centrifugation at 16,000 ϫ g for 5 min at 4°C. TCA was removed and the pellet washed twice in 1 ml of acetone followed by centrifugation (16,000 ϫ g for 5 min at 4°C). Proteins were solubilized in 200 l of buffer A (100 mM Tris-Cl, pH 8, 1% SDS, 1 mM EDTA) containing 1ϫ complete protease inhibitors (Roche Applied Science) and 1 mM PMSF and incubated for 1 h at 37°C. Insoluble debris was removed by centrifugation (12,000 ϫ g/5 min) and supernatant transferred to a new tube. Protein concentration was determined using a Bio-Rad DC assay kit.
Maleimide-PEG Modification of Protein Extracts-Yeast protein extracts for methoxypolyethylene glycol maleimide M r 5000 (mPEG) modification (34) were prepared using the TCA extraction procedure detailed above, with the following modifications. The concentration of DTT in buffer A was increased to 50 mM to ensure complete reduction of disulfide bonds. After the last step, the supernatant containing the soluble protein was re-precipitated with an equal volume of 20% TCA. The protein was collected by centrifugation (12,000 ϫ g for 5 min at 4°C) and the pellet washed twice with acetone. After the last wash, the tube was centrifuged again and residual acetone removed. The pellet was immediately re-dissolved in 200 l of buffer A (without DTT) ϩ 2 mM mPEG (4 l of 100 mM mPEG stock, made fresh by dissolving 50 mg in 100 l of buffer A). Reactions were incubated for 30 min at 37°C in the dark to prevent inactivation of mPEG. Aggregates were dispersed with a pipette, and the reaction was incubated an additional 12 h. mPEG addition to specific target proteins was detected by immunoblotting.
Preparation of Soluble and Insoluble Proteins from Yeast Cells-Insoluble protein aggregates were separated from soluble proteins using a centrifugation method based on two previously published protocols (36,37). Thirty five-ml cultures were grown to a density of 4 ϫ 10 6 cells/ml (A 600 Ϸ0.4). Cultures were grown for sufficient time to allow at least three generations of growth before harvest. Strains with growth defects in zinc-deficient conditions (e.g. tsa1⌬) were inoculated to an initial A 600 of 0.2 and maintained below 0.6 A 600 by dilution with fresh medium. At harvest, the cells were transferred to a 50-ml conical tube and chilled on ice. Cells were collected by centrifugation (5 min/3000 ϫ g at 4°C) and resuspended in 1 ml of ice-cold Tris-Cl (50 mM, pH 8.5), transferred to 1.5-ml microcentrifuge tubes, and centrifuged again. Cell pellets were frozen in liquid nitrogen and stored at Ϫ80°C. Protein extracts were prepared under nondenaturing conditions. The frozen cell pellets were thawed on ice and resuspended in 50 l of Soluble Protein Buffer (SPB) (50 mM Tris-Cl, pH 8.5, 500 mM NaCl, 1 mM PMSF, 1ϫ Complete Mini EDTA-free protease inhibitor (Roche Applied Science)). For each sample, a 2-ml round-bottom microcentrifuge tube was racked under liquid N 2 , and a 7-mm stainless steel ball (Retsch 05.368.0035) was placed inside. Thawed cells were transferred to tubes half-filled with liquid N 2 to ensure the sample did not freeze the ball to the bottom of the tube. The tube was removed to a rack to allow all the N 2 to evaporate and then closed and placed back in liquid N 2 . Tubes were racked into a PTFE 2-ml tube adaptor for the Retsch Mixer Mill MM400 (Retsch 22.008.0005) and agitated for four times for 90 s at 30 Hz, returning the sample holder to chill in liquid N 2 between sessions. Tubes were then removed and placed on ice, and 950 l of ice-cold SPB was added to each tube. After the samples had thawed, the tubes were gently mixed by inversion, and the ball was removed with a magnet. Lysates were centrifuged at 3000 ϫ g for 30 s at 4°C to remove unbroken cells. The supernatant was carefully removed (avoiding the pellet) and transferred to a 1.5-ml polyallomer conical centrifuge tube (Beckman Coulter). Tubes were centrifuged at 100,000 ϫ g for 20 min; the supernatant (representing the soluble fraction) was removed to a 1.5-ml microcentrifuge tube, frozen in liquid N 2 , and stored at Ϫ80°C. After removal of any residual supernatant, 500 l of wash buffer (50 mM Tris-Cl, pH 8.5, 150 mM NaCl, 1 mM PMSF, 1ϫ Complete Mini EDTA-free protease inhibitor (Roche Applied Science)) was added to the tube, and the pellet was resuspended by repeated pipetting until homogeneous. One hundred l of 10% Nonidet P-40 was added and the sample gently rotated at 4°C for 30 min. Tubes were centrifuged at 100,000 ϫ g for 20 min, and the supernatant was discarded. One hundred l of insoluble protein buffer (IPB, 50 mM Tris-Cl, pH 8.5, 50 mM NaCl, 8 M urea, 2% SDS, 0.5 mM DTT, 1 mM PMSF, 1ϫ Complete Mini EDTA-free protease inhibitor (Roche Applied Science)) was added to the pellet, and the samples were boiled for 5 min. The pellet was resuspended with a pipette until homogeneous, and the sample was vortexed for 15 min and then boiled again for 5 min. Samples were centrifuged at 16,000 ϫ g for 10 min, and the supernatant (the insoluble fraction) was removed to a new 1.5-ml microcentrifuge tube, avoiding the pellet. Hsp104-GFP was detected in the soluble and insoluble fractions by immunoblotting.
Fluorescence Microscopy-GFP fluorescence was visualized using an epifluorescence microscope. Cells were harvested, washed once with PBS, fixed for 2 h with 1% paraformaldehyde, washed with PBS, and collected by centrifugation. Cells were resuspended in 50 l of PBS and transferred to a 0.2-mm thick pad of 1% agarose in PBS on a microscope slide. A coverslip was applied, and the edges were sealed with paraffin wax. High resolution Z-stacks were obtained with Volocity Version 6.11 and processed to Z-projections using ImageJ Version 10.2.

RESULTS
Genetic Screen for tsa1⌬ Suppressors-To identify processes that might produce ROS in zinc-deficient cells, we used a transposon mutagenesis strategy to isolate tsa1⌬ strains carrying suppressor mutations that improved growth in zinc-deficient conditions. One suppressor was the result of a transposon insertion into the promoter of the TSA2 gene 336 bp upstream of the start codon (TSA2 Tn ). The TSA2 Tn allele conferred improved growth on a tsa1⌬ mutant, as shown here using pooled haploid segregants from a tsa1⌬/tsa1⌬ TSA2 Tn (LEU2)/ϩ (leu2) diploid parent (Fig. 1A). Suppression by the TSA2 Tn allele was confirmed by cloning the wild-type and transposon insertion TSA2 alleles and testing their ability to suppress tsa1⌬ (Fig. 1B). Although an increased copy number of wild-type TSA2 improved growth to a small degree (p Ͻ 0.01), the transposon insertion allele conferred strong suppression. Tsa2 is a close ortholog of Tsa1, but its expression is much lower (38,39). The TSA2 Tn allele increased Tsa2 accumulation ϳ7-fold (Fig. 1C). These results indicated that elevated Tsa2 could substitute for Tsa1 function in zinc-deficient cells.
A TRR1 Loss-of-Function Mutation Suppresses tsa1⌬-Analysis of additional suppressed strains indicated that many carried single recessive mutations unlinked to an inserted transposon (data not shown). Consistent with this observation, tsa1⌬ is known to cause a higher frequency of spontaneous mutations (40,41). We identified one suppressor mutation by pooled linkage analysis and whole genome sequencing (33). This analysis suggested that suppression was caused by a nonsense mutation at codon 277 of the TRR1 gene (trr1 L277 *). In support of this identification, suppression by trr1 L277 * was complemented (i.e. eliminated) by the wild-type TRR1 gene ( Fig. 2A). In addition, complete deletion of TRR1 also suppressed the growth defect of tsa1⌬ under deficient conditions (see Fig. 5B), confirming that trr1 L277 * was a loss of function allele.
Trr1 is required for function of the thioredoxin antioxidant pathway and resistance to oxidative stress (17,42). Because previous evidence suggested that the growth defect of zinc-deficient tsa1⌬ cells resulted from increased ROS, we were surprised to find that trr1 loss-of-function alleles suppressed tsa1⌬. One possible explanation was that trr1 mutation somehow increased oxidant tolerance in a tsa1 background. How-ever, we confirmed that trr1⌬ strongly decreased resistance to hydrogen peroxide, either alone or in combination with tsa1⌬ (Fig. 2B).
Alternative Model for Tsa1 Function in Deficient Cells-Identifying trr1⌬ as a tsa1⌬ suppressor argued that the tsa1⌬ growth defect was not a consequence of elevated oxidative stress. Therefore, we re-examined the role of peroxiredoxins in zinc-deficient cells. In addition to peroxidase activity, Tsa1 has holdase-type chaperone activity that maintains solubility of unfolded proteins (19). We used a genetic strategy to examine the relative importance of these two activities. Tsa1 contains two cysteines, the Cys-48 peroxidatic and Cys-171 resolving cysteines, that are both critical for peroxidase function (17). Thus, Tsa1-C48S and C171S mutants lack peroxidase function (19), and neither allele complemented the hydrogen peroxide sensitivity of tsa1⌬ (Fig. 3A). In contrast, full chaperone function requires only Cys-48 because it is highly activated by hyperoxidation of Cys-48 to the sulfinic acid (-SO 2 H) or fur- . C, elevated Tsa2 accumulation in TSA2 Tn strains. Immunoblot of Tsa2 protein from a wild-type strain (CWY2) carrying pFL38 (Vec), wild-type HA-tagged TSA2, or transposon-modified HA-tagged TSA2 plasmids after growth in zinc-deficient conditions. All panels were from the same blot with equal exposure times. The average fold change in Tsa2 levels was calculated from five independent experiments and standard deviation (SD) is shown. Pgk1 was used as a loading control. OCTOBER 25, 2013 • VOLUME 288 • NUMBER 43 ther to the sulfonic acid (-SO 3 H) forms (22,43). Cys-171 is not required for chaperone function. Although tsa1⌬ cells expressing Tsa1 C48S (peroxidase Ϫ chaperone Ϫ ) grew poorly in low zinc, Tsa1 C171S (peroxidase Ϫ chaperone ϩ ) complemented nearly as effectively as wild-type Tsa1 (Fig. 3B). The ineffectiveness of Tsa1 C48S was not due to reduced stability, as both mutant proteins accumulated to similar levels (data not shown). Thus, zinc-deficient cells primarily require Tsa1 chaperone activity.

Tsa1 Chaperone Is Essential in Zinc Deficiency
Given these results, we suspected that other holdase-type chaperones might be capable of substituting for Tsa1 in zincdeficient cells. Hsp26 and Hsp42 are heat shock-inducible chaperones with holdase activity (35, 44 -47). Unlike Tsa1, neither protein is known to have peroxidase activity. Overexpression of these proteins from the strong PMA1 promoter suppressed the tsa1⌬ growth defect (Fig. 3C), further indicating that Tsa1's holdase chaperone activity alone is critical under zinc-deficient conditions.
trr1⌬ Activates Yap1 in tsa1⌬ Mutants-The above observations suggested a model to explain suppression by trr1 mutations. Previous work showed that trr1 mutations constitutively activate the Yap1 oxidative stress-responsive transcription fac-tor (42). Yap1 binds to the YRE promoter elements and is activated by oxidation and deactivated by thioredoxin-dependent reduction (48). In a trr1 mutant lacking reduced thioredoxin, Yap1 is constitutively active and up-regulates a number of genes, including TSA2 (42). We reasoned that in a trr1⌬ tsa1⌬ mutant, increased Yap1-driven expression of Tsa2 chaperone function might suppress the tsa1⌬ growth defect. Although Yap1 regulates expression of other protein chaperones (49), we had already observed that increased Tsa2 could substitute for Tsa1 (Fig. 1). Investigating the role of Tsa2 in suppression by trr1 also allowed us to further test the importance of peroxiredoxin chaperone activity in deficient cells, because in a trr1 strain only Tsa2 chaperone activity is functional.

Tsa1 Chaperone Is Essential in Zinc Deficiency
stantial induction of a TSA2-lacZ reporter was also observed in both trr1⌬ and trr1⌬ tsa1⌬ strains under deficient conditions (Fig. 4C). Immunoblotting confirmed a similar effect of trr1⌬ on Tsa2 protein accumulation in zinc-replete and -deficient cells (Fig. 4D). Thus, trr1 alleles constitutively activated Yap1 and induced its target genes, including TSA2.
We then tested whether Tsa2 activity was required for suppression in trr1⌬ tsa1⌬ mutants. Growth of strains carrying combinations of trr1⌬, tsa1⌬, and tsa2⌬ mutations was compared in zinc-replete and -deficient conditions. In zinc-replete conditions, the trr1⌬ single mutant displayed a growth defect, but all other strains grew well (Fig. 5A). In deficient conditions, however, growth of the trr1⌬ tsa1⌬ tsa2⌬ triple mutant was reduced ϳ50% relative to the trr1⌬ tsa1⌬ double mutant (Fig.  5B). Similar results were obtained with the trr1 L277 * allele (data not shown). These observations establish that in a zinc-deficient trr1⌬ tsa1⌬ mutant, TSA2 is both markedly induced and required for full suppression. The observation that the tsa2⌬ mutation did not completely eliminate suppression by trr1⌬ suggested that activation of other Yap1 target genes, or perhaps other trr1-associated changes in gene expression, also contributes to suppression.
If the induction of Yap1-regulated target genes was responsible for suppression in trr1⌬ tsa1⌬ mutants, we predicted that inactivating Yap1 would block this suppression. However, we could not isolate yap1 trr1 strains due to synthetic lethality. Instead, we determined whether Yap1-mediated induction of TSA2 was required for full suppression. A TSA2-lacZ reporter lacking YREs was previously shown to be unresponsive to Yap1 (39). We introduced wild-type and YRE⌬ mutant TSA2-lacZ reporters into wild-type, tsa1⌬, trr1⌬, and trr1⌬ tsa1⌬ strains. In zinc-deficient cells, the trr1⌬ mutation activated the wildtype TSA2-lacZ construct ϳ10-fold (comparing tsa1⌬ and trr1⌬ tsa1⌬ strains, Fig. 5C). In contrast, expression from the YRE⌬ mutant promoter increased only 2-fold. This effect of the YRE⌬ promoter on TSA2 expression was also observed at the protein level (Fig. 5D) confirming that Yap1 is needed for full trr1⌬-induced up-regulation of TSA2 expression.
We then determined the effect of the TSA2 YRE⌬ mutations on suppression by trr1⌬ (Fig. 5E). When TSA2 was expressed from its wild-type promoter, the trr1⌬ mutation conferred an 11-fold increase in growth yield under low zinc conditions. This contribution of TSA2 to suppression was greater than we observed when comparing deletion mutants (Fig. 5B), perhaps (YRE-lacZ) (B), or pTSA2-lacZ (TSA2-lacZ) (C) reporter genes were grown to log phase in zinc-deficient medium prior to ␤-galactosidase assays. Shown are the averages of at least three replicates, and the error bars denote Ϯ 1 S.D. MU, Miller units. Strains used in A were DY1457, CWM48, CWM115, and CWM113; strains used in B were CWY2, CWM20, CWM163, and CWM83, and strains used in C were CWY2, CWM20, CWM170, and CWM83. D, trr1⌬ mutation induces Tsa2 protein accumulation. Strains of the indicated genotypes were grown to log phase in zinc-replete and zinc-deficient medium before protein extraction. One representative immunoblot of three independent experiments is shown. Values below (Tsa2 signal) are the average fold change in Tsa2 band intensity relative to the corresponding wild-type/pHA-Tsa2 sample (S.D. ϭ standard deviation, three replicates). Pgk1 was used as a loading control. Vec, vector. OCTOBER 25, 2013 • VOLUME 288 • NUMBER 43 due to additional suppression effects from increased TSA2 copy number. When TSA2 was expressed from the Yap1-insensitive YRE⌬ mutant promoter, growth of the suppressed strain was reduced by ϳ30%. These results indicate that the Yap1-mediated up-regulation of Tsa2 contributes to a significant proportion of tsa1⌬ suppression in a trr1 mutant. The observation that the YRE⌬ mutations did not totally eliminate TSA2 induction (Fig. 5, C and D) or Tsa2's contribution to suppression (Fig. 5E) in trr1⌬ cells suggests that other regulatory factors may also contribute to TSA2 activation by trr1⌬ (e.g. Msn2/4) (39).

Tsa1 Chaperone Is Essential in Zinc Deficiency
Detection of the Hyperoxidized Peroxiredoxin Chaperone Form-These results indicated that the chaperone function of Tsa1 (and of Tsa2 in a trr1⌬ tsa1⌬ strain) is critical for low zinc growth. Because the activated chaperone form of Tsa1 is hyperoxidized on Cys-48, we predicted that hyperoxidized Tsa1 would be present in zinc-limited cells. To test this prediction, we first determined whether zinc-deficient cells accumulated Cys-48 sulfonic acid-Tsa1. Using a specific antibody (22), sulfonic acid-Tsa1 was readily detectable on immunoblots of protein from cells treated with H 2 O 2 (data not shown). However, it was not detected in zinc-deficient protein extracts, indicating zinc deficiency did not generate sufficient oxidative stress to cause terminal hyperoxidation of Tsa1.
To determine whether Tsa1 accumulated as the Cys-48 sulfinic acid form, we used a thiol-reactive modifying reagent (mPEG) (34). Treatment with mPEG adds ϳ25 kDa to the apparent molecular mass of a protein per modified cysteine residue. Because hyperoxidized cysteine sulfinic and sulfonic acids cannot be reduced by DTT, DTT treatment of proteins followed by mPEG modification can distinguish Tsa1 hyperoxidized at Cys-48 (single-modified) from the fully reduced, disulfide, or sulfenic acid forms (double-modified). Accordingly, mPEG treatment of samples from unstressed cells generated two slower migrating forms of Tsa1 (Fig. 6A). Control experiments indicated modification was specific to cysteines, as Tsa1 C48S and C171S mutant proteins were modified only once (data not shown). Moreover, we found that mPEG modification of cysteines that are not hyperoxidized is highly efficient, as ϳ98% of the single cysteine present in the Pgk1 protein was modified ( Fig. 6A and data not shown). In both zinc-replete and -deficient tsa1⌬ cells expressing Myc-tagged Tsa1, ϳ25% of Tsa1 was singly modified (Fig. 6, B and C), suggesting that the sulfinic acid chaperone form was abundant. As expected, hydrogen peroxide treatment of cells efficiently converted Tsa1 to the singly modified (Cys-48 hyperoxidized) form. Thus, our data suggest that zinc deficiency does not increase Tsa1 hyperoxidation but also show that hyperoxidized Tsa1 is abundant in both zinc-deficient and -replete cells.
The trr1⌬ suppression phenotype prompted us to determine its effect on Tsa1 and Tsa2 hyperoxidation. No trr1⌬-depen-  Suppression by trr1 mutations requires TSA2. Cells of the indicated genotypes were inoculated at an A 595 of 0.01 in zinc-replete (LZM ϩ 1 mM ZnCl 2 ) (A) or -deficient (B) media and incubated for 1 or 3 days, respectively, prior to measuring cell density. Strains used were CWY2, CWM20, CWM163, CWM84, CWM138, and CWM140. Shown are averages of three independent cultures, and the error bars denote Ϯ 1 S.D. C, strains of the indicated genotypes bearing a wild-type TSA2 promoter-lacZ fusion (pTSA2-lacZ) or a YRE mutant TSA2 promoter-lacZ fusion (pYRE⌬-TSA2-lacZ) were grown in deficient medium prior to ␤-galactosidase assay. Strains used were the same as A. Shown are the averages of three independent replicates, and error bars denote Ϯ 1 S.D. D, strains of the indicated genotype bearing the vector (Vec, pFL38), pHA-TSA2, or a derivative in which the YREs were deleted (pYRE⌬-HA-TSA2) were grown in zinc-deficient medium before protein extraction. One representative immunoblot of three independent experiments is shown. Values at bottom (Tsa2 signal) are the average fold changes in Tsa2 band intensity relative to the tsa2⌬/pHA-TSA2 strain (SD ϭ standard deviation, three replicates). Pgk1 was included as a loading control. E, CWM138 (tsa1⌬ tsa2⌬) and CWM140 (trr1⌬ tsa1⌬ tsa2⌬) were transformed with the vector (Vec, pFL38), pHA-TSA2, or pYRE⌬-HA-TSA2. Cultures were inoculated in zinc-deficient media at an A 595 of 0.01 and incubated for 3 days. Values are averages of three replicates and error bars denote Ϯ 1 S.D.

Tsa1 Chaperone Is Essential in Zinc Deficiency
dent change in Tsa1 hyperoxidation was seen in either zincreplete or -deficient cells; ϳ25% of the protein was in the hyperoxidized chaperone form in both conditions (Fig. 6C). Peroxide treatment of trr1⌬ cells had little effect on Tsa1 hyperoxidation, which is consistent with Tsa1 accumulating in vivo as the disulfide-bonded form. When we examined Tsa2, we also detected a basal level of hyperoxidation in replete cells that was increased by peroxide treatment (Fig. 6D, ZnR and ZnRϩH 2 O 2 ). Low expression in wild-type and tsa1⌬ strains prevented accurate determination of the ratio of reduced and hyperoxidized protein. However, the trr1⌬ mutation dramatically increased both total Tsa2 accumulation and the amount of the hyperoxidized form. Under zinc-deficient conditions, hyperoxidized Tsa2 was undetectable in wild-type and tsa1⌬ cells but very abundant in trr1⌬ strains. This observation supports our hypothesis that an increase in Tsa2 chaperone activity contributes to suppression by trr1⌬.
Zinc-deficient tsa1⌬ Mutants Exhibit Unfolded Protein Stress-The requirement for Tsa1 chaperone function in zincdeficient cells strongly suggested that they are challenged by the accumulation of unfolded proteins, and the Tsa1 chaperone was required to stabilize these proteins. In yeast, unfolded protein stress activates the Hsf1 and Msn2/4 transcription factors, which induce expression of chaperones to facilitate folding and prevent aggregation (50). We predicted that in zinc-deficient tsa1⌬ cells, accumulation of unfolded proteins would activate Hsf1 and/or Msn2/4, leading to accumulation of their target chaperones. To test this prediction, we measured the accumulation of the Hsp70-type foldase isozymes Ssa3 and Ssa4, the holdase proteins Hsp26 and Hsp42, and the Hsp104 disaggregase under zinc-replete and -deficient conditions (Fig. 7, A and  B). Zinc deficiency had little effect on chaperone accumulation in wild-type cells. In contrast, the tsa1⌬ mutation increased the accumulation of all these chaperones under zinc-deficient conditions, from ϳ2-fold for Hsp104 to more than 100-fold for Hsp26. This result suggested that the loss of Tsa1 activity caused induction of an unfolded protein stress response in low zinc.
Because oxidative stress caused by loss of Tsa1 peroxidase activity might also activate Hsf1 or Msn2/4 (37, 46, 51), we determined whether this response was due to the absence of the Tsa1 peroxidase or its chaperone activity. Ssa3/4 accumulation was compared in tsa1⌬ expressing no Tsa1 (vector only), wildtype Tsa1, Tsa1 C48S (peroxidase Ϫ chaperone Ϫ ), or Tsa1 C171S (peroxidase Ϫ chaperone ϩ ) (Fig. 7C). No difference in Ssa3/4 accumulation between these genotypes was observed under zinc-replete conditions. However, a zinc-deficient tsa1⌬ strain showed a 9-fold increase in Ssa3/4 accumulation compared with the same strain expressing wild-type Tsa1. A strain expressing Tsa1 C48S showed a substantial 5-fold increase in Ssa3/4, while the strain expressing Tsa1 C171S was similar to wild-type Tsa1. These observations argue that tsa1⌬-associated Ssa3/4 induction primarily occurs due to loss of Tsa1 chaperone activity and provides further evidence that zinc-deficient cells are challenged by unfolded protein stress.
The increased chaperone accumulation in tsa1⌬ could result from activation of the Hsf1 or Msn2/4 factors or from some post-transcriptional change, e.g. in chaperone stability. To determine whether transcription was activated in zinc-deficient cells, we compared the activity of a lacZ reporter gene derived from the SSA3 promoter in wild-type and tsa1⌬ strains. Approximately 5-fold more SSA3-lacZ expression was detected in zinc-deficient tsa1⌬ cells (Fig. 7D). The induction was not due to cells undergoing diauxic shift (which activates Msn2/4 and Hsf1), as they were maintained at low density throughout the growth period. Thus, Ssa3 accumulation in tsa1⌬ cells was (at least in part) due to the induction of the SSA3 promoter, indicating that the Hsf1 and/or Msn2/4 regulators are activated.
If the tsa1⌬ growth defect was related to unfolded protein accumulation, we predicted that trr1⌬ might reverse this effect FIGURE 6. Tsa1 chaperone form is abundant in zinc-deficient cells and Tsa2 chaperone levels increase in trr1 mutants. A, mPEG modification of Tsa1 cysteine residues. CWM20 (tsa1⌬) was transformed with plasmids for wild-type untagged Tsa1 (pTsa1) or Myc-tagged Tsa1 (pmyc-Tsa1). Strains were grown under zinc-replete conditions, and protein samples were extracted and treated as described under "Experimental Procedures." Some control samples were left untreated with DTT and/or mPEG as indicated. Samples were then analyzed by immunoblotting, and Tsa1 and Pgk1 were detected with anti-Myc and anti-Pgk1 antibodies, respectively. B, representative immunoblot of Tsa1 and Pgk1 detected in total yeast protein samples after treatment with DTT and mPEG. tsa1⌬ and trr1⌬ tsa1⌬ strains (CWM20 and CWM83, respectively) expressing epitope-tagged Tsa1 (pmyc-Tsa1) were grown in zinc-replete (R) or -deficient (D) media to log phase before analysis. Replicate zinc-replete cultures were treated with 5 mM H 2 O 2 for 20 min before harvest, DTT/mPEG treatment, and analysis by immunoblotting. C, proportion of hyperoxidized Tsa1 accumulated by cells was calculated from three independent replicates of the experiment shown in B. Band intensities of the modified Tsa1 forms (singly modified/hyperoxidized or doubly modified/ fully reduced) were quantified, and the hyperoxidized Tsa1 signal was expressed as percent of total Tsa1. Error bars denote Ϯ 1 S.D. D, effect of zinc supply and trr1⌬ on Tsa2 hyperoxidation. Yeast strains CWY2 (WT), CWM20 (tsa1⌬), CWM170 (trr1⌬), and CWM83 (trr1⌬ tsa1⌬) were transformed with empty vector (Vec, pFL38) or an epitope-tagged Tsa2 plasmid (pHA-Tsa2). Strains were grown to log phase in replete or deficient conditions or treated with peroxide as described for B. Total protein samples were modified with mPEG, and Tsa2 was detected by immunoblotting using an anti-HA antibody. One representative replicate of three is shown. OCTOBER 25, 2013 • VOLUME 288 • NUMBER 43 and lower chaperone accumulation. Surprisingly, however, the trr1⌬ mutation increased Ssa3/4 accumulation, particularly in zinc-deficient tsa1⌬ trr1⌬ cells (Fig. 7C, ZnD). This effect was likely mediated by activation of Hsf1, because an Hsf1-specific HSE-lacZ reporter gene was induced in trr1⌬ (Fig. 7E). Activation of Hsf1 in trr1⌬ is a novel observation, and the mechanism of this effect is unclear. Given that Hsf1 regulates many different chaperones, its activation in trr1⌬ may contribute to the observed Tsa2-independent component of tsa1⌬ suppression (Fig. 5B). The observation that the magnitude of trr1-mediated Hsf1 activation was similar to that produced by sustained heat shock (Fig. 7E) is consistent with this hypothesis.

Tsa1 Chaperone Is Essential in Zinc Deficiency
Tsa1 Protects Zinc-deficient Cells from Protein Aggregation-Our observations strongly suggested that zinc-deficient tsa1⌬ cells accumulate unfolded proteins, which are prone to aggregate. Without knowing the specific identity of the affected proteins, we could not test this hypothesis directly. However, indirect evidence for protein aggregation could be obtained by examining chaperones that associate with such aggregates. For example, Hsp104 binds to the surface of aggregates and disassembles their component proteins for refolding (52). Accordingly, both native (53) and GFP-tagged (54) versions of Hsp104 were reported to coalesce into foci after heat stress, as we also observed (data not shown). We therefore investigated if zinc deficiency and tsa1⌬ affected Hsp104-GFP behavior at optimal growth temperature. In most zinc-replete wild-type and tsa1⌬ cells, Hsp104-GFP was evenly distributed in the cytosol (Fig.  8A), but 2-3% of cells displayed small fluorescent foci, suggesting the rare presence of chaperone-associated protein aggregates (Fig. 8, A and B). Zinc deficiency had little effect on the frequency of wild-type cells showing foci, suggesting little change in protein solubility. In contrast, Hsp104-GFP foci were present in ϳ70% of zinc-deficient tsa1⌬ cells (Fig. 8B). A single aggregate was normally present in each cell, with greatly increased size and fluorescence intensity compared with the few observed in wild-type cells. Zinc deficiency had no effect on the distribution of GFP alone in tsa1⌬ cells (data not shown), indicating that the effect was a property of the Hsp104 chaperone. Foci formation was not likely to result simply from increased Hsp104 accumulation because we saw only a 2-fold increase in deficient tsa1⌬ cells (Fig. 7B). To determine whether the Hsp104-GFP aggregation phenotype was due to loss of Tsa1 chaperone or peroxidase activity, we examined cells expressing the cysteine mutants of Tsa1. Expressing Tsa1 C171S in tsa1⌬ almost completely complemented the Hsp104 aggregation phenotype, although expression of Tsa1 C48S had little effect (Fig.  8C), indicating a requirement for chaperone function. The trr1⌬ suppressor mutation also effectively suppressed the tsa1⌬ Hsp104-GFP aggregation phenotype (Fig. 8D), supporting a correlation between the tsa1⌬ growth defect and chaperone aggregation phenotype.  (Vec, pFL38), a wild-type Tsa1 plasmid (pmyc-Tsa1), or Tsa1 cysteine mutant plasmids. Also shown are trr1⌬ (CWM170) and trr1⌬ tsa1⌬ (t/t⌬, CWM83) strains transformed with vector alone. Cultures were grown to log phase in zinc-replete or -deficient conditions before protein extraction. Values below the blots are the average intensities of Ssa3/4 bands from three independent experiments (SD, standard deviation). D, SSA3-lacZ activity in wild-type (CWY2/pSSA3-lacZ) and tsa1⌬ (CWM20/pSSA3-lacZ) strains after growth in zinc-replete and -deficient conditions. E, HSE-lacZ activity was measured in wild-type (CWY2) and trr1⌬ (CWM170) strains under zinc-replete and -deficient conditions. Zinc-replete cultures of CWY2 were also subjected to heat shock (37°C/30 min) before harvest.

Tsa1 Chaperone Is Essential in Zinc Deficiency
The Hsp26 holdase chaperone also associates with protein aggregates after heat stress (55). Therefore, we examined the effect of zinc supply on the distribution of Hsp26-GFP in tsa1⌬ cells. In zinc-replete cells, no Hsp26-GFP fluorescence was detectable, consistent with the low expression of Hsp26 under these conditions (Fig. 7B). Approximately 18% of zinc-deficient wild-type cells displayed Hsp26-GFP fluorescence, predominantly as multiple small foci per cell (Fig. 8, E and F). In zinclimited tsa1⌬ cultures, however, the proportion of cells with Hsp26-GFP foci increased dramatically, and the protein aggregated further to form 1-2 large foci/cell. The change in Hsp104 and Hsp26 distribution in tsa1⌬ suggests that these proteins are trafficked to organized sites of misfolded protein deposition, which are thought to arise when protein homeostasis is disrupted (55,56).
tsa1⌬ Mutation Decreases Hsp104 Solubility-We suspected that the large Hsp104-GFP foci seen in zinc-deficient tsa1⌬ cells contained aggregated insoluble protein. If so, we expected that the solubility of Hsp104 would also be reduced. To test this prediction, we isolated soluble and insoluble protein fractions from wild-type, tsa1⌬, trr1⌬, and trr1⌬ tsa1⌬ double mutant strains grown under zinc-replete and -deficient conditions, and we examined the distribution of Hsp104-GFP in these fractions. Hsp104-GFP was ideal for this experiment because it showed high basal expression (Fig. 7B), allowing comparison of its solubility between unstressed and stressed cells. Control experiments confirmed that Hsp104-GFP solubility was dramatically reduced in wild-type cells exposed to heat stress (data not shown). Similarly, we observed a decrease in Hsp104-GFP solubility in tsa1⌬ strains after transfer to zinc-deficient condi-FIGURE 8. Effect of tsa1⌬ and trr1⌬ mutations on chaperone aggregation and solubility. A, wild-type and tsa1⌬ mutant cells expressing Hsp104-GFP (strains HSP104-GFP and CWM180, respectively) were grown in zinc-replete or -deficient media for at least four generations, maintaining cell density below 0.4 A 595 by dilution. Cells were fixed with paraformaldehyde, and images were captured using fluorescence (GFP) and differential interference contrast (DIC) microscopy. B, foci prevalence was measured in replicate cultures grown as in A. Cells containing 1-2 or Ͼ2 Hsp104-GFP foci were counted and are presented as the percentage of total cells. For all foci quantitations in this figure, the averages of three independent replicates are shown, and error bars denote Ϯ 1 S.D. C, tsa1⌬ mutant expressing Hsp104-GFP (CWM180) was transformed with vector alone (Vec, pFL38) or plasmids expressing wild-type, C48S, or C171S alleles of Tsa1. Cells were grown in zinc-deficient medium, and foci frequency was determined. D, wild-type (CWY2), tsa1⌬ (CWM20), trr1⌬ (CWM170), and trr1⌬ tsa1⌬ (CWM83) cells expressing Hsp104-GFP (pHSP104-GFP) were grown in deficient medium, and foci frequency was determined. E and F, wild-type (HSP26-GFP) and tsa1 (CWM188) strains expressing Hsp26-GFP were grown in zinc-deficient medium as described for A. Fluorescence images were captured and quantified as described for B. G, Hsp104-GFP solubility in zinc-replete and -deficient cultures of strains described in D. Cultures were grown for at least four generations, maintaining cell density below 0.4 A 595 by dilution. Lysates were prepared, separated into soluble and insoluble fractions, and analyzed by immunoblotting with anti-GFP antibody. Hsp104-GFP band density was quantified and normalized to protein loaded, and the ratio of insoluble/soluble Hsp104 was calculated. Shown are the averages of three independent replicates, and error bars denote Ϯ 1 S.D. OCTOBER 25, 2013 • VOLUME 288 • NUMBER 43 tions. Zinc deficiency induced a 3-fold change in wild-type cells but a much larger (50-fold) change in a tsa1⌬ strain (Fig. 8G). In contrast, a control protein (Pgk1) was almost undetectable in the insoluble fractions extracted from zinc-deficient cells of either wild-type or tsa1⌬ strains (data not shown). Finally, we examined the effect of the trr1⌬ suppressor mutation on Hsp104-GFP solubility. In isolation, trr1⌬ had little effect on solubility in replete cells but slightly increased Hsp104-GFP solubility in deficient cells. When combined with tsa1⌬, however, the effect of trr1⌬ was striking, completely restoring Hsp104-GFP solubility in deficient cells. Thus, trr1⌬ suppressed the major phenotypes of tsa1⌬ mutants under zincdeficient conditions, suggesting these phenotypes are a consequence of a common lesion. The nature of these phenotypes suggests that the lesion is the accumulation of excess unfolded proteins in the absence of stabilizing Tsa1 chaperone activity.

DISCUSSION
The mechanisms of protein homeostasis include protein folding, degradation, and sequestration. Protein chaperones play roles in all three by mediating folding (57), aiding in ubiquitination and proteasomal degradation (58), and controlling sequestration of protein aggregates (59). Some chaperones, e.g. the Hsp70/Hsp40 systems, are termed "foldases" and directly mediate folding. Other chaperones, e.g. Hsp26, Hsp42, and Tsa1-like peroxiredoxins, are "holdases" that bind to unfolded proteins to prevent their aggregation. In this study, we discovered that zinc-limited cells experience a severe disruption of protein homeostasis and that the holdase chaperone function of Tsa1 protects these cells from protein aggregation. Although peroxiredoxin chaperone activity has been well characterized in vitro (19,20,22), its relevance in vivo has been more elusive. Tsa1's chaperone function was shown to contribute to heat tolerance (19), and it was suggested that the Tsa1 chaperone protects ribosomal subunits from aggregation in DTT-treated cells (60), but this was not tested. It appears that Tsa1 plays a uniquely important role in zinc deficiency, because the other holdase chaperones Hsp26 and Hsp42 cannot substitute for Tsa1 unless highly overexpressed (61).
Our initial goal was to identify the source(s) of oxidative stress in zinc-limited cells by isolating mutations that suppressed the growth defect of a tsa1⌬ mutant. A key discovery was that loss-of-function mutations in the TRR1 thioredoxin reductase gene strongly suppressed the tsa1⌬ growth defect in zinc-limited cells. In low zinc, single trr1⌬ mutants grow much better than tsa1⌬ mutants (Fig. 5B). A trr1⌬ mutant lacks Tsa1 peroxidase activity, but retains the hyperoxidized (chaperone) form (Fig. 6). These observations alone suggest it is Tsa1's chaperone function that is vital for deficient cells. This hypothesis was confirmed by the observation that the Tsa1 C171S mutant, which lacks peroxidase function but retains chaperone activity, complemented the tsa1⌬ growth defect, although the Tsa1 C48S allele, which lacks peroxidase activity and has much less chaperone function, does not. Additional confirmation came from the fact that overexpression of Hsp26 and Hsp42 suppressed the tsa1⌬ growth defect. These are holdase-type chaperones like Tsa1 but have no known peroxidase function (35,44). A final argument for the essentiality of the chaperone came from the observation that suppression in trr1⌬ tsa1⌬ double mutants is partially dependent on the Tsa2 paralog of Tsa1. Because Tsa2 also requires Trr1 for its peroxidase function, increased ROS metabolism cannot explain its contribution to suppression.
The critical importance of Tsa1 chaperone activity to zinclimited growth indicates that zinc deficiency is associated with a major disruption in protein homeostasis. Several of our observations support this hypothesis. First, severe zinc deficiency caused by mutating ZAP1 induces activity of several Hsf1-responsive promoters (62), although we recognize that this effect may reflect decreased Tsa1 expression. Second, the behavior of chaperones Hsp104 and Hsp26 in zinc-deficient tsa1⌬ is consistent with the accumulation and aggregation of unfolded proteins. Hsp104 is a "disaggregase"-type chaperone that disassembles protein aggregates for folding by other chaperones (e.g. the Hsp70/Hsp40 systems) (63)(64)(65). In response to heat shock, Hsp104 associates with aggregated proteins in cytosolic foci to mediate their refolding or degradation (56,65). In zinc-limited tsa1⌬ cells, we observed an increase in Hsp104-GFP aggregation and a decrease in its solubility, suggesting it was associated with insoluble protein aggregates. Hsp26 also associates with protein aggregates (65,66) and showed similar aggregation in deficient tsa1⌬ cells. Two distinct quality control compartments, called the JUNQ and the IPOD, form in cells with defective protein homeostasis (56). Recent work argues that Hsp104-GFP primarily associates with the IPOD (54), whereas Hsp26 associates with diverse protein aggregates (55,67). Interestingly, we observed that Hsp104 primarily formed single foci in zinc-deficient tsa1⌬ cells (suggesting this compartment represents the IPOD), whereas Hsp26-GFP primarily formed two (Fig. 8). The decrease in Hsp104-GFP solubility that we observed would also be consistent with its location in the IPOD, as this compartment is thought to accumulate terminally misfolded and insoluble components (56). However, the identity of the tsa1⌬-associated chaperone foci and their relationship to the IPOD and JUNQ remain to be determined.
The switch from peroxidase to full chaperone function involves hyperoxidation of Cys-48 to the sulfinic (-SO 2 H) or sulfonic (-SO 3 H) acid forms (19,20,22). In zinc-replete cells, we estimate that ϳ25% of Tsa1 was in the sulfinic chaperone form. As about 400,000 Tsa1 molecules are normally present in unstressed cells (31), they therefore contain ϳ100,000 chaperone molecules. Zinc deficiency did not increase the proportion of Tsa1 in the chaperone form. However, given that total Tsa1 protein levels increase about 3-fold in zinc-limited cells by Zap1 transcriptional activation (9), we predict that ϳ300,000 Tsa1 molecules are present in the chaperone form in low zinc. By comparison, the Hsp42 and Hsp26 holdases are estimated to be much less abundant (31), and their expression is not strongly induced by zinc deficiency in wild-type cells (Fig. 7B).
A supply of 300,000 holdase molecules represents a substantial chaperone capacity for the protection of proteins from misfolding and irreversible aggregation. What stress to protein homeostasis in low zinc would require such a large capacity? One intriguing possibility is the failure to metallate zinc-dependent proteins. An estimated 10% of proteins in eukaryotes require zinc for folding, and many of these are very abundant in Tsa1 Chaperone Is Essential in Zinc Deficiency cells (2,68). For example, cytosolic superoxide dismutase and alcohol dehydrogenase normally accumulate to several hundred thousand molecules per cell (31). A failure to metallate even a fraction of zinc-binding proteins would result in a large burden of incompletely folded proteins. In support of this model, we have previously shown that at least some zinc sites are poorly metallated in cells grown under the same zinc-deficient conditions used here (10,69,70).
Although this hypothesis is intriguing, other potential mechanisms also warrant consideration. For example, the 26 S proteasome includes an essential zinc-dependent subunit, Rpn11 (71). In zinc-deficient cells, decreased Rpn11 activity might increase the accumulation of ubiquitinated proteasome substrates. Alternatively, the critical lesion in zinc-deficient cells could result from more general defects in protein folding. The Ydj1 protein is the major cytosolic DnaJ-type Hsp40 co-chaperone, which helps bind and present substrate proteins to Hsp70 for folding. Ydj1 contains two C4-type zinc binding domains, which may play a role in presenting some substrates to Hsp70 (72). Thus, under-metallation of Ydj1 could disrupt the efficient folding of many proteins.
Considerable effort has been directed toward understanding how potentially promiscuous and toxic metal ions are delivered to specific apoproteins to enable their correct folding and function. Our observations emphasize that metal-deficient cells must deal with another and just as critical an issue, i.e. how to maintain apoproteins in a folding-competent state in the absence of their essential metal cofactor. To our knowledge, the importance of adequate zinc to protein homeostasis has not previously been documented. Regardless of the cause of disrupted protein homeostasis, it is likely that zinc deficiency may have similar effects in vertebrate cells. This is especially true if the stress is due to unmetallated zinc apoproteins, because zinc is a common structural cofactor in all organisms. If zinc deficiency in humans also disrupts protein homeostasis, it may be an important environmental factor in the etiology of diseases of protein misfolding, such as Alzheimer, Parkinson, and Huntington diseases or prion diseases such as Creutzfeldt-Jakob. For example, failure to metallate Cu,Zn superoxide dismutase with zinc has been linked to the neurotoxicity associated with the familial form of amyotrophic lateral sclerosis (73). Moreover, given that the chaperone function of Tsa1 is conserved in human 2-Cys peroxiredoxins (20), it seems likely that these proteins play key roles in maintaining protein homeostasis in zinc-deficient cells of vertebrates. The World Health Organization estimates that more than 15% of the world's population is at risk for zinc deficiency. Our results suggest that this high prevalence of zinc deficiency may have a major unforeseen impact on human health.