Hsp31 Is a Stress Response Chaperone That Intervenes in the Protein Misfolding Process*

Background: Hsp31 is a multifunctional cellular stress response protein. Results: Hsp31 is a stress-inducible chaperone that has substoichiometric activity on a wide spectrum of substrates, including prions and the non-physiological client, α-synuclein. Conclusion: Hsp31 protects cells from α-synuclein-mediated toxicity via chaperone activity and independently from enzymatic activity and autophagy. Significance: This is the first characterization of the early acting and wide ranging chaperone activity of Hsp31. The Saccharomyces cerevisiae heat shock protein Hsp31 is a stress-inducible homodimeric protein that is involved in diauxic shift reprogramming and has glyoxalase activity. We show that substoichiometric concentrations of Hsp31 can abrogate aggregation of a broad array of substrates in vitro. Hsp31 also modulates the aggregation of α-synuclein (αSyn), a target of the chaperone activity of human DJ-1, an Hsp31 homolog. We demonstrate that Hsp31 is able to suppress the in vitro fibrillization or aggregation of αSyn, citrate synthase and insulin. Chaperone activity was also observed in vivo because constitutive overexpression of Hsp31 reduced the incidence of αSyn cytoplasmic foci, and yeast cells were rescued from αSyn-generated proteotoxicity upon Hsp31 overexpression. Moreover, we showed that Hsp31 protein levels are increased by H2O2, in the diauxic phase of normal growth conditions, and in cells under αSyn-mediated proteotoxic stress. We show that Hsp31 chaperone activity and not the methylglyoxalase activity or the autophagy pathway drives the protective effects. We also demonstrate reduced aggregation of the Sup35 prion domain, PrD-Sup35, as visualized by fluorescent protein fusions. In addition, Hsp31 acts on its substrates prior to the formation of large aggregates because Hsp31 does not mutually localize with prion aggregates, and it prevents the formation of detectable in vitro αSyn fibrils. These studies establish that the protective role of Hsp31 against cellular stress is achieved by chaperone activity that intervenes early in the protein misfolding process and is effective on a wide spectrum of substrate proteins, including αSyn and prion proteins.

The Saccharomyces cerevisiae heat shock protein Hsp31 is a stress-inducible homodimeric protein that is involved in diauxic shift reprogramming and has glyoxalase activity. We show that substoichiometric concentrations of Hsp31 can abrogate aggregation of a broad array of substrates in vitro. Hsp31 also modulates the aggregation of ␣-synuclein (␣Syn), a target of the chaperone activity of human DJ-1, an Hsp31 homolog. We demonstrate that Hsp31 is able to suppress the in vitro fibrillization or aggregation of ␣Syn, citrate synthase and insulin. Chaperone activity was also observed in vivo because constitutive overexpression of Hsp31 reduced the incidence of ␣Syn cytoplasmic foci, and yeast cells were rescued from ␣Syn-generated proteotoxicity upon Hsp31 overexpression. Moreover, we showed that Hsp31 protein levels are increased by H 2 O 2 , in the diauxic phase of normal growth conditions, and in cells under ␣Syn-mediated proteotoxic stress. We show that Hsp31 chaperone activity and not the methylglyoxalase activity or the autophagy pathway drives the protective effects. We also demonstrate reduced aggregation of the Sup35 prion domain, PrD-Sup35, as visualized by fluorescent protein fusions. In addition, Hsp31 acts on its substrates prior to the formation of large aggregates because Hsp31 does not mutually localize with prion aggregates, and it prevents the formation of detectable in vitro ␣Syn fibrils. These studies establish that the protective role of Hsp31 against cellular stress is achieved by chaperone activity that intervenes early in the protein misfolding process and is effective on a wide spectrum of substrate proteins, including ␣Syn and prion proteins.
Heat shock proteins (HSPs) 5 protect living cells against environmental stress and promote cell survival. This function is evolutionarily conserved throughout different organisms. HSPs interact with misfolded proteins and process them using several mechanisms, including chaperone functions to prevent misfolded protein aggregation, unraveling protein aggregates, or carrying out protease functions to eliminate irreversibly damaged proteins. The effect of HSPs in maintaining cellular homeostasis throughout the protein folding process is essential to preventing diseases associated with protein misfolding and aggregation, such as Alzheimer disease, Parkinson disease (PD), and Huntington disease. HSPs involved in counteracting these protein inclusions are currently under intensive study (1).
Among the neurodegenerative diseases, PD is the second most common neurodegenerative disorder that occurs in elderly individuals 50 years or older, and no cure is currently known (2,3). High levels of reactive oxygen species (ROS), dysfunctional mitochondria, and the aggregation of ␣Syn are observed in PD (4 -8). DJ-1, a gene linked to familial forms of PD, has been found to be involved in oxidative insults, proteasome degradation, and protein folding and is the focus of extensive studies to understand its functions in regulating levels of ␣Syn aggregation (9 -14).
Human DJ-1, yeast Hsp31 (YDR533c), and Escherichia coli Hsp31 (hchA or yedU) are members of the DJ-1/PfpI/ThiJ superfamily (10,15,16). Crystal structures of Hsp31 show that it contains a cysteine catalytic triad that is conserved in the DJ-1/PfpI/ThiJ family (17,18). E. coli hchA is a close ortholog with yeast Hsp31, and they share superimposable active sites in each monomer, but their quaternary structures appear to be different (17,19). This protein family is of biomedical importance due to the role these proteins play in chaperone-like and cytoprotective activities (15,19). Despite the shared structural similarity, this family can be further subcategorized into three classes according to the structural and functional properties of the proteins or into five subclasses by protein sequence alignments (15,19). The cellular functions of DJ-1 and E. coli hchA have been characterized, and their multifunctional roles in mediating oxidative stress, chaperone-like activity, methylglyoxalase activity, and cytoprotection have been described (10,15,16,19). There has been limited characterization of the cellular functions of Hsp31, but several recent studies have demonstrated that it has glyoxylase activity (20), chaperone activity (21), and a role in autophagy (22).
Yeast PD models have been used to study the mechanism of the sporadic and familial forms of PD (23)(24)(25). We are extending the utility of these yeast models by investigating the biological activities of yeast Hsp31. Hsp31 appears to be a stressinducible 26-kDa protein based on several large scale studies indicating that yeast Hsp31 is up-regulated when cells are exposed to environmental stress (26 -29). For example, Hsp31 was implicated to be protective against ROS because an hsp31⌬ strain was more sensitive to oxidative stress (26). HchA has structural similarity to Hsp31 and is established as having chaperone activities, suggesting that Hsp31 could act similarly; however, this has yet to be confirmed. The evidence for a role in stress response, possible chaperone activity, methylglyoxalase activity, and cross-talk with the autophagy pathway led us to investigate these activities of Hsp31 in modulating ␣Syn cytoplasmic foci formation and prion aggregation. ␣Syn is not present in yeast and hence is a non-physiological substrate, but it mimics physiologically relevant activity based on its localization and toxicity, and we capitalized on these features. Therefore, we designed a set of biochemical, genetic, and cell biology experiments to demonstrate the role of Hsp31 chaperone activity in protecting cells from ␣Syn-mediated cytotoxicity. We also demonstrate that Hsp31 has methylglyoxalase activity that converts methylglyoxal (MGO) substrate to D-lactate, but this enzyme activity is not essential for protecting against ␣Synmediated cytotoxicity.

Yeast Cell Growth Conditions
Yeast media were prepared according to Amberg et al. (30). Liquid yeast extract/peptone dextrose medium contained Bacto yeast extract (1%; Fisher), Bacto peptone (2%; Fisher), and glucose (2%; Fisher). Synthetic dextrose (SD) minimum liquid medium was made of 0.17% Difco yeast nitrogen base (without amino acids) and 2% glucose and supplemented with necessary amino acids for auxotrophic strains needed at concentrations described previously (30). Solid medium plates were made with the same components of liquid medium plus 2% agar (Fisher). To express galactose-inducible proteins, 2% raffinose (Affymetrix, Cleveland, OH) and 2% galactose (Affymetrix) were used to replace glucose. Fractions of culture were obtained at designated times to monitor the cell fitness and protein levels by A 600 and immunoblotting, respectively.
All of the yeast strains were grown at 30°C in 2% glucose minimal medium overnight. Strains were diluted to A 600 of 0.2 and grown in minimal medium switching the carbon source with 2% raffinose for the overnight period. Cultures were normalized to A 600 of 0.8, and 5-fold serial dilutions were aliquoted onto the respected appropriate dropout plate containing 2% glucose or galactose. Plates were incubated at 30°C for 2-3 days.

Yeast Strain Construction
Yeast strains used in this study are listed in Table 1. The hsp31⌬ and atg8⌬ deletions were generated in W303-1A or in a strain expressing one or two copies ␣Syn fused to fluorescent proteins (CFP and YFP) using the primers listed in Table 2. The nat (nourseothricin N-acetyltransferase) gene flanked by Hsp31 homology regions was obtained by PCR with 70-nucleotide-long forward primer and reverse primer. The primers consisted of 20 nucleotides for amplifying the nat gene from pFA6a-natNT2 (Euroscarf) and 50 nucleotides immediately preceding the HSP31 or ATG8 start codon or after the stop codon. The amplified product was integrated into W303 ␣Synexpressing strains at ChrIV:1502160 to 1501447, as described previously (31). For double knock-out hsp31⌬ atg8⌬ strain, the kanMX4 resistance cassette was used to delete Hsp31, and then the natMX4 resistance cassette was used for deletion of ATG8.

Hsp31-9myc Tagging
We endogenously tagged HSP31 with the 9myc epitope using a PCR-based integration (12). The pYM20 plasmid was used as a template, and primers were used to obtain PCR product with HSP31 genomic flanking followed by transformation into W303 and W303 ␣Syn-CFP ϩ ␣Syn-YFP strains. The transformants were selected on media containing hygromycin B (300 mg/liter), and correct integration was verified by PCR using primers spanning the integration junctions and by DNA sequence analysis.

Antibodies and Immunoblotting
To determine protein expression, yeast cells were collected by centrifuging at 5,000 rpm, and pellets were resuspended in extraction buffer (50 mM Tris-HCl, pH 7.5, 1 mM EDTA, 4 mM MgCl 2 , 5 mM DTT). Glass beads (Sigma) were added into the mixture, which was vortexed 5 times for 10 s each, plus 1-min intervals on ice. Clear crude protein lysate was obtained by spinning down the cell debris at 1,000 ϫ g at 4°C for 10 min. The SDS-PAGE loading dye (4% SDS, 40% glycerol, 0.02% bromphenol blue, Tris-Cl, pH 6.8) was added to the supernatant, and samples were boiled, followed by SDS-PAGE and immunoblotting with antibodies. Monoclonal anti-Myc and anti-␤-actin antibody were purchased from Sigma-Aldrich. Monoclonal anti-␣Syn was obtained from BD Biosciences. Antibodies were used to detect expression of Hsp104 (ab69549, Abcam), Hsp70 (SPA-82, Stressgen); DsRed (Sc-33353, Santa Cruz Biotechnology, Inc.), and GFP (anti-GFP, Roche Applied Science).

Protein Purification
BY4741 harboring yeast Hsp31 expression plasmid from the yeast moveable ORF collection (33) (Thermo Fisher Scientific) was used to express and purify the protein (33). We did not see any evidence of co-purifying proteins on SDS-polyacrylamide gels and observed similar high activity from protein purified under high salt conditions, which minimizes contaminant copurifying proteins. Human DJ-1 was encoded in pGEX 6p-1 (GE Healthcare), hchA was encoded in pNT-hchA, and ␣Syn was expressed from the pT7 plasmid. The protein expression and purification was described previously (34 -36). Briefly, constructs were transformed into BL21 (DE3) cells. The transformants were grown to an A 600 of 0.4 -0.6 in LB medium supplemented with 100 g/ml ampicillin at 37°C, and isopropyl ␤-D-1-thiogalactopyranoside (1 mM final concentration) was added to induce protein expression. After 3 h of induction at 37°C, cells were harvested by centrifugation at 2,000 ϫ g and resuspended in lysis buffer (25 mM KP i , pH 7.0, 200 mM KCl) containing protease inhibitor mixture set IV (Calbiochem). Crude protein was prepared by sonicating cells and clarifying the lysate by centrifugation at 10,000 ϫ g for 10 min. GSTtagged DJ-1 was immobilized by glutathione-agarose resin (Thermo Scientific), and DJ-1 was eluted by cleaving from the GST tag with PreScission Protease (GE Healthcare) (2 units of protease for every 100 g of tagged DJ-1), leaving a linker amino acid sequence: NPAFLYKVVDVSRHHHGRIFYPY-DVPDYAGLEVLFQ.
Sedimentation velocity ultracentrifugation was performed with two different Hsp31 concentrations (34 and 17 M) and run at 50,000 rpm in a Beckman Coulter XLA centrifuge (Beckman Coulter, Fullerton, CA). The sedimentation coefficients and apparent molecular weights were calculated from size distribution analyses (c(s)) using SEDFIT (37). His 6 -hchA was immobilized by HIS-Select nickel affinity gel (Sigma-Aldrich) and eluted with elution buffer (Tris-Cl, pH 7.5, 250 M imidazole). The purity of proteins was determined by SDS-PAGE followed by Coomassie Blue staining, and concentrations were determined by a Micro BCA protein assay kit (Thermo Scientific, Waltham MA). The ␣Syn protein for in vitro fibrillization assays was purified by expressing pT7-7 plasmid containing human ␣Syn cDNA (provided by R. Jakes and M. Goedert, Medical Research Council, Cambridge, UK) with isopropyl ␤-D-1-thiogalactopyranoside in BL21 (DE3) for 4 h of incubation at 37°C. The cells were harvested and resuspended in 10 mM Tris-Cl, pH 8, 1 mM EDTA, 1 mM PMSF at 1 ⁄ 20 culture volume. The cell lysate was obtained by one round of freezethawing and French press treatment, and protein was partially purified and enriched by two rounds of ammonium sulfate precipitation (30% saturation first and then 50% saturation at 0°C). The pellets were resuspended and boiled in 10 mM Tris-HCl, pH 7.4, for 5 min. The insoluble protein was removed by centrifugation, and the supernatant was filtered using a 0.22-m  filter. The filtered fraction of the protein was applied to a DEAE-Sepharose FF column (GE Healthcare) with 10 mM Tris-HCl, pH 7.4, and eluted with 0 -500 mM NaCl over 100 min. The fractions containing ␣Syn were dialyzed against 10 mM NH 4 HCO 3 and lyophilized. The protein was ϳ95% pure based on Coomassie Blue staining with SDS-PAGE.

Chaperone Activity Aggregation Suppression Assays
Thermal Aggregation Assay of Citrate Synthase (CS)-CS aggregation suppression assay was performed according to the procedures described previously (38). Briefly, CS (porcine CS, Sigma), Hsp31, DJ-1, and hchA were buffer-exchanged into 40 mM Hepes-KOH (pH 7.5). Proteins were diluted to concentrations ranging from 0.1 to 0.6 M in a 500-l reaction. Samples were loaded into a quartz cuvette and placed into the Fluoro-Max-3 spectrofluorometer (Horiba, John Yobin) heated to 43°C. Aggregation of CS, with and without Hsp31, was measured by light scattering with the excitation and emission set to 360 nm and a slit width of 2 nm. BSA fraction V and hchA served as negative and positive controls. Measurements were taken every second for 2,000 s. Data were plotted using GraphPad Prism version 5.0.
Reducing Agent-induced Aggregation of Insulin (38,39)-Briefly, Hsp31 was buffer-exchanged into PBS (pH 7.2). Insulin was diluted with PBS to 26 M, and 4 or 8 M of Hsp31 was used in a 500-l reaction. Samples were loaded into a quartz cuvette and placed into the FluoroMax-3 spectrofluorometer. Aggregation of insulin induced by 20 mM DTT, with and without Hsp31, was measured by light scattering with the excitation and emission set to 360 nm and a slit width of 1 nm. Measurements were taken every second for 2,000 s. Data were plotted using GraphPad Prism version 5.0.
Spontaneously Fibrillizing ␣Syn Assay-The procedure for performing the Thioflavin T aggregation assay was described previously (40). The ␣Syn was prepared by centrifuging the recombinant protein through a 100 kDa spin filter to remove preformed oligomers (35). DJ-1 and Hsp31 purified from either E. coli or Saccharomyces cerevisiae, respectively, were incubated with ␣Syn (17.5 M) for 5 days at 37°C. Thioflavin T (20 M), a fluorescent dye that emits a signal only when binding to ␤-sheet-containing protein aggregates, was included in the reaction to detect the fibrillar state of ␣Syn by a Spectrafluor Plus microplate reader (Tecan, Uppsala, Sweden) at 440 nm for the excitation and 490 nm for the emission wavelength.

Dilution Growth Assays
Plasmids (pAG415-GPD-ccdB-DsRed and pAG415-GPD-Hsp31-DsRed) were transformed into W303 ␣Syn-expressing strains with or without a genomic copy of the HSP31 gene knocked out using the PEG/lithium acetate transformation method. Single colonies of transformants were grown in SDϪLeu overnight, and the cells were washed, normalized, and grown in synthetic complete medium without leucine (SCϪLeu) ϩ raffinose (2%) to switch the carbon source. After incubating in the raffinose medium overnight, the cell number was normalized, and dilutions were aliquoted on an SDϪLeu (suppressed expression) plate and SCϪLeu with 2% raffinose and 2% galactose (induced expression) with 5-fold serial dilutions. The plates were incubated at 30°C for 3 days.

Fluorescence Imaging of ␣Syn Localization
pESC-Leu myc-Hsp31 and control empty vector were transformed into two copies of ␣Syn-expressing strain. The ␣Syn expression was induced by growing cells in synthetic minimum medium plus 2% raffinose and 2% galactose for 8 h. The live cells were visualized using a Nikon TE2000-U inverted fluorescence microscope with Nikon Plan apochromat ϫ60 (numerical aperture 1.4) oil immersion objective and YFP filters plus ϫ300 magnification. In order to differentiate the localization of protein, ϳ50 individual cells were randomly counted in multiple regions of interest for each set of experiments. The ratio of membrane to cytoplasmic localization of ␣Syn fluorescent fusion protein was presented as the mean from three independent sets of experiments.

MGO Addition and Microscopy
␣Syn-expressing strains with or without the HSP31 gene deleted were grown in liquid yeast extract/peptone dextrose medium and transferred into medium containing 2% galactose and MGO (0.5-2 mM) for 6 and 12 h. Cells were washed and subjected to confocal microscopy to observe ␣Syn foci formation. The presence of MGO inhibited the growth of cells for the initial 12 h and resulted in an equal growth rate between strains regardless of the HSP31 gene deletion. Strains could not be grown in the presence of high concentrations of MGO (20 mM), so ␣Syn was induced for 10 h, and then MGO was added and incubated for another 2 h before confocal microscopy analysis.

Assessment of Intracellular ROS
Cells were harvested after the induction of ␣Syn expression for 12 h, and 5 ϫ 10 6 cells were prepared for staining with dihydroethidium. Cells were suspended in 250 l of 2.5 g/ml dihydroethidium in PBS and incubated in the dark for 10 min. Cells were washed in the PBS and were subjected to fluorescence microscopy (excitation at 485 nm and emission at 520 nm) and flow cytometry with the FL-2 channel. FlowJo software was used to calculate median fluorescence intensity.

Prion Expression Experiments
W303 wild type strain was co-transformed with pAG424-GAL-PrD-Sup35-EYFP and pAG415-GPD-HSP31-DsRed. Single transformations were also performed using the same constructs and their corresponding control vector. Cells were grown at 30°C in appropriate medium overnight (SDϪTrp for pAG424, SDϪLeu for pAG415, and SDϪTrpϪLeu for cotransformants), and protein expression was induced for 24 h in SC 2% raffinose and 2% galactose-containing medium at 30°C. After induction, cells were subjected to fluorescence microscopy using a Nikon A1 confocal microscope with a Nikon Plan apochromat ϫ60 (numerical aperture 1.4) oil immersion objective to acquire fluorescence and differential interference contrast images and were analyzed using ImageJ. Cultures identical to those used in microscopy were used to prepare samples for flow cytometry, and cells were washed and resuspended in PBS at a density of ϳ10 ϫ 10 6 cells/ml. Cells were filtered and ana-lyzed for EYFP fluorescence intensity using a flow cytometer with the FL-1 channel. A total of 10,000 events were acquired for each sample, and data were analyzed using FlowJo software to calculate median fluorescence intensity after three biological replicates.

Glutathione-independent Glyoxalase Biochemical Assay
Purified recombinant Hsp31 and Hsp31 C138D with the GST tag removed were used to assay methylglyoxalase activities. The reaction was initiated by adding the specified concentration of proteins (1 M Hsp31 and 5 M Hsp31C138D) in reaction buffer (100 mM HEPES, pH 7.5, 50 mM KCl, 2 mM DTT) to a 6 mM initial concentration of MGO (Sigma; 40% solution), followed by incubation at 30°C. The assay was performed by removing 50-l aliquots of the reaction at fixed time points (15,30,45, and 120 s) after the addition of protein. The amount of D-lactic acid produced by Hsp31 as a result of MGO consumption was measured. The initial rate obtained was divided by the amount of protein in the reaction mixture to calculate the specific activity. Samples were also subjected to gas chromatography-mass spectroscopy analysis (Agilent 5975C MSD) to identify the presence of D-lactic acid. Samples were derivatized with N,O-bis(trimethylsilyl) trifluoroacetamide) and trimethylchlorosilane and heated to 50 ºC for 10 min and run in the electron impact mode with scanning from 42 to 400 atomic mass units. A lactic acid standard displayed a peak at 6.09 min identical to the peak detected in the Hsp31treated sample, whereas Hsp31C138D did not display this peak. The spectral scan was visualized and displayed using the OpenChrom software (41). To obtain the enzymatic parameters, the reaction was initiated as described above by adding 1 M Hsp31 or 5 M Hsp31C138D protein into reaction buffer, and reactions were stopped after 1 min by heating at 85°C for 30 s. A range of MGO substrate from 50 M to 2 mM and the EnzyFluo D-lactate assay kit (Bioassay Systems) were used to determine the amount of D-lactic acid produced in each reaction. Each rate was determined in triplicate, and mean values were plotted to obtain enzymatic parameters (V max and K m ) by fitting to a Michaelis-Menten model using GraphPad Prism.

Semidenaturing Detergent-Agarose Gel Electrophoresis (SDD-AGE)
W303 cells harboring pAG424-GAL-cPrDSup35-EYFP and pAG415-GPD-HSP31-DsRed plasmids were used, and aggregates were produced by inducing for 24 h in SC 2% raffinose ϩ 2% galactose medium. Prion aggregates were analyzed using SDD-AGE (42). Briefly, to prepare lysates, cells were harvested by centrifugation at 3,000 ϫ g for 2 min and resuspended in spheroplast solution (1.2 M D-sorbitol, 0.5 mM MgCl 2 , 20 mM Tris (pH 7.5), 50 mM ␤-mercaptoethanol, and 0.5 mg/ml Zymolyase) and incubated at 30°C for 30 min with shaking. After spheroplast formation, the samples were centrifuged at 800 rpm for 5 min at room temperature, and supernatant was removed. The pelleted spheroplasts were resuspended into 100 l of lysis buffer (100 mM Tris, pH 7.5, 50 mM NaCl, 10 mM ␤-mercaptoethanol, and protease inhibitor) and vortexed at high speed for 2 min. The lysates were collected and mixed with 4ϫ sample buffer (2ϫ TAE, 20% glycerol, 4% SDS, 0.01% bro-mphenol blue). The samples were incubated at room temperature for 15 min and loaded onto a 1.8% agarose gel containing 1ϫ TAE and 0.1% SDS and run at 50 V, followed by transfer onto nitrocellulose membrane using the capillary transfer method. The nitrocellulose membrane was subjected to Western blot analysis using anti-GFP antibody (Roche Applied Science). A strain harboring p2HG-Hsp104 plasmid was included as a positive control.

Results
Hsp31 Has Chaperone Activity-The S. cerevisiae open reading frame (YDR533C) has been assigned the name HSP31; however, direct functional studies demonstrating chaperone activity have not been conducted on this protein to support its designation within the HSP family. Hsp31, human DJ-1, and hchA are members of the ThiJ/DJ-1 superfamily based on similarity in structure, so we directly compared the in vitro chaperone activity between these proteins. The influence of Hsp31 on the aggregation of substrate proteins was measured using three different model substrates: CS, insulin, and ␣Syn. CS has been used as a classical substrate for in vitro chaperone assays, and both DJ-1 and hchA have demonstrated an ability to suppress CS aggregation (9,43). To assess the chaperone activity of Hsp31, we purified the protein from yeast using the BG1805 plasmid-based system from the yeast movable ORF (MORF) collection (33). The C terminus of the protein retains a 4.8-kDa tag, as confirmed by mass spectrometry, and includes a His 6 and HA tag (see the amino acid sequence under "Experimental Procedures"). We also expressed Hsp31 with a GST tag fused to its N terminus using the pGEX-6p1 plasmid in E. coli. The tag was removed by 3C protease, and only 3 amino acids (GPL) were retained on the recombinant Hsp31 protein, which was confirmed by mass spectrometry (data not shown). We determined that the purified proteins were folded correctly because they behaved as homodimers based on gel filtration chromatography (data not shown) and by analytical ultracentrifugation with a sedimentation coefficient of 3.7 S (Fig. 1A). The Hsp31-MORF, purified from yeast, can completely suppress the heatinduced aggregation of CS when present at a 1:1 molar ratio (Fig. 1B). BSA, which was added at a 6-fold molar excess concentration of CS, had a slight effect on suppressing CS aggregation in agreement with a previous study that reported BSA as having a mild chaperone effect in the CS aggregation assay (44). E. coli hchA suppressed CS aggregation in a dose-dependent manner, but the high concentration of 0.6 M did not approach the considerably more efficient suppression of CS aggregation by Hsp31-MORF (Fig. 1B). In addition, recombinant yeast Hsp31 purified from E. coli did not have activity at a 1:1 molar ratio and was needed at high molar ratios similar to hchA (data not shown) (Fig. 1B). In conclusion, Hsp31 purified from yeast can suppress CS aggregation at a 1:1 molar ratio, whereas hchA is needed at a 6:1 molar ratio to generate an appreciable suppression of aggregation. These results are consistent with previous studies, where high molar ratios of hchA were required in CS assays (38,45).
To further characterize the chaperone activity of Hsp31, additional protein substrates were studied. Hsp31-MORF suppressed insulin aggregation induced by reducing agent, when using a 3-6-fold molar excess of insulin (Fig. 1C), further establishing its chaperone activity in another commonly used chaperone assay. The mechanism of ␣Syn aggregation is unclear, and DJ-1 has been reported to decrease ␣Syn aggregates or cytoplasmic foci in vitro and in vivo (9,14,46,47). Hsp31 completely suppressed ␣Syn fibrillization despite the presence of an 8-fold molar excess of ␣Syn and maintained chaperone activity when ␣Syn was present at a 35:1 molar ratio (Fig. 1D). In agreement with this result, we observed that DJ-1 suppressed ␣Syn fibrillization although not to the same extent as Hsp31. The difference may be due to the oxidation state of DJ-1, which has been shown to affect its activity (46). Taken together, these results demonstrate that Hsp31 has the ability to act as a chaperone in preventing several types of protein aggregation at substoichiometric concentrations.
Hsp31 Protects against ␣Syn Toxicity and Reduces ␣Syn Foci Formation and ROS Levels-Outeiro and Lindquist (23) demonstrated that two genomic copies of GAL-␣Syn in yeast decreased cellular fitness, which was correlated with ␣Syn cellular aggregates (46). DJ-1 has been reported to alleviate protein aggregation (9,46), and ␣Syn has been implicated as a factor in the progression of PD (48 -50). To determine the chaperone effect of Hsp31 with its anti-aggregation activity and effect on cellular fitness, we tested the rescue of yeast cells from ␣Synmediated toxicity in dilution growth assays. Overexpression of one copy of the ␣Syn gene from the GAL (galactose) promoter had negligible effects on cellular fitness, but GAL-driven overexpression of two ␣Syn gene copies resulted in severe toxicity, confirming previous findings of dose-dependent toxicity. The ␣Syn toxicity was markedly alleviated in the presence of constitutive expression of Hsp31 (pGPD-HSP31), as evidenced by the full rescue of cell viability ( Fig. 2A). In addition, we expressed one copy of GAL-␣Syn in an hsp31⌬ strain to determine whether there was a synthetic genetic interaction between these genes. Dilution growth assays demonstrated that expression of a single copy of ␣Syn or knocking-out HSP31 alone was phenotypically similar to the wild-type strain with respect to viability. However, cellular fitness was significantly reduced when expressing a single copy of ␣Syn in the hsp31⌬ strain (Fig.  2B). The synthetic lethal interaction of ␣Syn and deleted HSP31  OCTOBER 9, 2015 • VOLUME 290 • NUMBER 41 using this approach provided further evidence that Hsp31 is a molecular chaperone with the ability to protect cells from ␣Syn-mediated toxicity.

Hsp31 Is a Stress Response Chaperone
Previous studies demonstrated that the subcellular localization of ␣Syn changed from the plasma membrane to cytoplasmic foci with increased expression of ␣Syn in yeast cells (23). Using this relocalization property, we examined the effect of pGAL HSP31 overexpression in altering the distinctive subcellular localization profile of 2xGAL-␣Syn. The same strains used in the dilution growth assay were used to monitor the effect of Hsp31 expression on ␣Syn localization. We quantified the number of cells with specific localization profiles at 8 h postinduction and observed that 60% of cells exhibited ␣Syn foci when only expressing ␣Syn, but only 33% of cells co-expressing ␣Syn and Hsp31 contained foci (Fig. 3, A and B). We observed that the suppression of ␣Syn foci by pGAL HSP31 was transient because 24 h after induction, the percentage of cells with foci increased and was similar to the percentage observed for cells expressing ␣Syn alone. This result is consistent with our rescue experiments, which were only successful when HSP31 was expressed from the GPD promoter (Fig. 2, A and B) and not when expressed from the GAL promoter (data not shown).
The expression of ␣Syn has been demonstrated to increase the level of ROS in yeast (51). The in vivo suppression of ␣Syn foci by Hsp31 and increased toxicity of one copy ␣Syn in the hsp31⌬ strain prompted us to investigate the level of ROS in these different genetic contexts. We found that wild-type yeast did not have detectable superoxide radicals when treated with dihydroethidium using microscopy and a fluorescence level of 0.5 arbitrary units by flow cytometry (Fig. 3, C and D). ROS levels were detectable in a fraction of cells in the hsp31⌬ strain (6 arbitrary units), indicating that the presence of Hsp31 in the cell can decrease ROS levels of normal cells not expressing ␣Syn. The remaining strains had increased ROS levels in the following order: ␣Syn-YFP strain (23 arbitrary units), ␣Syn-YFP hsp31⌬ (41 arbitrary units), ␣Syn-YFP ␣Syn-CFP (64 arbitrary units), and ␣Syn-YFP ␣Syn-CFP hsp31⌬ (133 arbitrary units). The latter three strains with the highest ROS levels exhibit reduced viability when ␣Syn is expressed and form increased foci, demonstrating a correlation between these phenotypes.
Hsp31 Is an Integral Part of the Yeast Cellular Stress Response-Multiple studies have shown that environmental stresses, such as oxidative stress and increasing temperature, increase the expression of Hsp31 (26,52). A recent study demonstrated increased HSP31 mRNA levels upon treatment with H 2 O 2 , during the diauxic shift growth phase and stationary phase (22). However, the Hsp31 protein level was not characterized under these conditions. We generated a yeast strain with a genomically integrated 9myc epitope at the C terminus of the HSP31 locus to quantify protein expression. We examined Hsp31 protein expression levels during growth in rich media relative to optical density and demonstrated that Hsp31 expression could be divided into three phases. We found that Hsp31 expression decreased during log phase and then reached maximum expression during early stationary phase relative to the cell density of the culture and remained at approximately the same level in stationary phase when cell density was saturated after growth for 24 h (Fig. 4A). This distinctive expression pattern results in the initiation of elevated expression of Hsp31 corresponding to the diauxic shift, which occurs between log phase and stationary phase (26). To assess whether Hsp31 expression increases under other stress conditions, we exposed cells to H 2 O 2 and observed increased protein expression within 1 h of exposure (Fig. 4B). This increased expression in response to an ROS-inducing agent has been observed previously (22,26).
We also examined the expression profile of Hsp31 to a proteotoxic stress in the form of ␣Syn overexpression, and the results indicated that increased expression of ␣Syn decreases cell viability and concomitantly increases the expression of Hsp31 during log phase compared with a strain not expressing ␣Syn (Fig. 4, C-E). The characteristic decrease in expression during log phase, 6 -9 h postinoculation, is not observed when ␣Syn is expressed. Our data demonstrate that Hsp31 expression is rapidly elevated at the protein level in the event of FIGURE 2. Hsp31 rescues cells from ␣Syn mediated toxicity. A, Hsp31 coexpression in yeast reduced ␣Syn toxicity. A yeast strain containing one or two copies of genomically expressed GAL-␣Syn was used in a growth dilution assay, in the presence of pGPD-HSP31-DsRed or empty vector, in media containing galactose/raffinose (left) or glucose (right). The strain harboring two copies of the ␣Syn gene had better fitness when transformed with the Hsp31encoding construct compared with the empty vector. B, combination of hsp31⌬ with single copy ␣Syn expression is toxic for yeast. The expression of a single copy of ␣Syn in hsp31⌬ strain reduced cellular fitness in the presence of glucose (top) and raffinose/galactose (bottom). The overexpression of Hsp31 in the same strain was able to rescue viability on both types of media.
growth stress associated with diauxic shift, oxidative stress, and proteotoxicity.
Hsp31 Methylglyoxalase Activity Is Not Required for Rescue of ␣Syn-mediated Toxicity-Recent studies have shown that DJ-1 and Hsp31 are methylglyoxalases that convert MGO into D-lactate in a single step independent of GSH (53, 54). Hsp31, like other members of the DJ-1 superfamily, possesses the con-served Glu-Cys-His catalytic triad (18). The first two residues of this triad are critical for MGO activity of DJ-1 as well as Hsp31 in Schizosaccharomyces pombe (54). We determined the methylglyoxalase activity for recombinant Hsp31 (purified from E. coli with the GST tag removed to retain 3 amino acids, GPL, on the C terminus) and a catalytic triad mutant, Hsp31 C138D, using a commercially available D-lactate assay kit. As expected, Hsp31 could convert MGO into D-lactate with a calculated specific activity of 28.4 mol/min/mg enzyme (using a saturating substrate concentration of MGO at 6 mM). These enzymatic parameters indicate a more efficient enzyme activity than the previously reported Hsp31-specific activity of 10.5 D-lactate mol/min/mg (53). The Hsp31-MORF purified from yeast had similar enzymatic parameters as the recombinant Hsp31 (data not shown). Enzymatic parameters were calculated by measuring the time-dependent production of D-lactate with varying substrate concentrations and fitting to a Michaelis-Menten kinetics model, resulting in a V max of 0.0356 mol of D-lactic acid/min (S.E. ϭ 0.0016), and K m was 247.4 M (S.E. ϭ 31.5) (Fig. 5A). Specific activity for the mutant could not be determined because low levels of D-lactate were produced despite using 5 times the amount of protein (detection limit of 0.4 mol/min/mg). The reaction products of these assays were examined by GC-MS, and a peak identical to the lactic acid standard (6.09-min elution time) was identified, confirming the production of lactic acid by Hsp31 (Fig. 5B). Lactic acid was undetectable by GC-MS analysis in the Hsp31 C138D sample (data not shown).
We next determined whether methylglyoxalase activity is required for rescue of ␣Syn toxicity. One copy of ␣Syn-YFP decreases fitness in the hsp31⌬ strain, but GPD-driven expression of Hsp31 rescues that toxicity. The expression of GPDdriven Hsp31 C138D restored the ␣Syn-expressing strain to full viability comparable with expression of the wild-type Hsp31 (Fig. 5C). The lack of enzyme activity for this mutant and the Cells were normalized to A 600 ϭ 0.2 at the start, and aliquots were collected until stationary phase was reached ϳ12 h postinoculation and at 24 h when the culture was saturated. The expression of genomically tagged Hsp31-9myc expression was assessed via immunoblot analysis of cell lysates obtained at corresponding time points. The plot shows expression levels normalized to ␤-actin levels (•). B, H 2 O 2 treatment increases the expression of Hsp31. The Hsp31-9myc strain was exposed to 1 mM H 2 O 2 treatment for 30 and 60 min, and increased Hsp31 expression was determined by immunoblot. Samples were normalized for cellular density by A 600 , and ␤-actin was used as a loading control. These data are representative of three independent experiments. C, growth curves showing the inhibitory effect of GAL-␣Syn expression in yeast harboring two copies of ␣Syn. Cells were normalized to A 600 ϭ 0.2 at hour 0, representing the time of inoculation, and the OD was monitored for 24 h at the identical time points used in A. Cells containing two copies of genomically integrated ␣Syn (f) grew more slowly 9 h after inoculation compared with the same strain without ␣Syn (•). D, Hsp31-9myc expression increased in the presence of induced GAL-␣Syn. The immunoblot was probed with anti-Myc and anti-␤-actin antibodies. The level of Hsp31 at the 7.5 and 9 h time points did not decrease in the 2x␣Syn strain compared with the strain that did not express ␣Syn. E, quantitation of Hsp31 expression level reveals increased expression during log phase. The Hsp31-9myc-tagged cells with and without two copies of ␣Syn were harvested at 0, 3, 4, 5, 6, 7.5, 9, 12, and 24 h after inoculation. The levels of Hsp31 expression were quantified and normalized to the ␤-actin signal. Three independent experiments were performed, and all experiments resulted in similar trends as observed here. A.U., arbitrary units. ability to rescue ␣Syn toxicity indicated that the in vivo mechanism of rescue is not dependent on methylglyoxalase activity.
MGO is a toxic metabolite produced during glycolysis (53-55) that reacts with proteins to yield toxic advanced glycation end products (56,57) and has been associated with increased ␣Syn cross-linking and formation of intracellular foci formation in neuroblastoma cells (58). We examined whether increased levels of MGO could increase the ␣Syn foci or toxicity by treating cells with exogenous MGO and found no increased foci in wild-type yeast, indicating that their formation is probably not influenced by MGO. In addition, the hsp31⌬ strain also did not have increased foci, suggesting that lack of Hsp31 is not sufficient to increase foci formation due to MGO treatment (Fig. 6, A and B). We determined that MGO was toxic to wild-type yeast at 10 and 20 mM MGO, and viability in the hsp31⌬ strains was similar to wild type, suggesting that Hsp31 is not sufficient for detoxification. The concomitant expression of ␣Syn (one or two copies) and treatment with MGO did not have differential effects on viability (Fig. 6C). Consistent with these results is that cell lysates of the hsp31⌬ strain had levels of overall methylglyoxalase activity equivalent with those of wild type because other yeast methylgloxalase enzymes, Glo1 and Glo2, are present in yeast and have been shown to detoxify MGO (59, 60) (data not shown).

Hsp31 and Catalytic Triad Mutants Prevent the Transition of ␣Syn Monomers to SDS-resistant Oligomeric Species-
The demonstrated chaperone activity of Hsp31 prompted us to investigate at what stage Hsp31 prevents protein aggregation. We used the spontaneous in vitro fibrillization of ␣Syn assay as a method to investigate the different oligomeric species in the presence of Hsp31. The prolonged incubation of recombinant monomeric ␣Syn results in fibrillization and precipitation of the protein. Fractionation of the precipitate versus supernatant and subsequent electrophoresis on SDS-PAGE allowed the detection of different oligomeric species that appear to be SDSresistant (Fig. 7A). The ␣Syn higher order oligomers were reduced in the pellet fraction and nearly non-existent in the supernatant when Hsp31 wild-type and catalytic triad mutants were present (Fig. 7B). The greatly reduced presence of soluble but SDS-resistant oligomers indicates that Hsp31 may interact with the monomer or an early oligomeric species to prevent formation of later stage oligomers. The same reactions were monitored using the ThioT assay, and only a baseline level of fluorescence was observed in the presence of Hsp31, further supporting the hypothesis that Hsp31 interacts with an ␣Syn species that precedes larger oligomers detected by ThioT (Fig.  7C). The ability of mutants to reduce fibrillization is consistent with our observation that the mutants suppress ␣Syn toxicity in yeast (Fig. 5C).
Hsp31 Prevents Formation of Large Prion Aggregates-The demonstration that Hsp31 possesses chaperone activity in a variety of anti-aggregation assays prompted us to investigate whether Hsp31 could also modulate the oligomeric state of yeast prion proteins, such as the Sup35 prion. The aggregation state of prions in living cells can be monitored by fusion of fluorescent protein tags to prion-forming domains (PrD), so we overexpressed Hsp31 to determine whether it modulated the aggregation of Sup35 PrD. Consistent with previous findings, cells overexpressing PrD in the presence of endogenous Sup35 showed characteristic ribbon-like or punctate aggregates, which were localized around the vacuoles and/or adjacent to plasma membranes (Fig. 8A). The number of cells co-expressing pGPD-HSP31-DsRed and pGAL-PrD-Sup35-EYFP had ϳ3-fold fewer Sup35 fluorescent aggregates compared with cells harboring pGAL-PrD-Sup35-EYFP plus vector control, indicating that Hsp31 can inhibit the formation of microscopically visible Sup35 prion aggregates (Fig. 8, A and B). The intrinsic fluorescence of Sup35-EYFP appears to increase upon aggregation, which allowed us to utilize flow cytometry as an alternative method to quantitatively measure and show decreased Sup35 aggregation (Fig. 8C). This phenomenon has not been exploited widely but was noted in at least one previous

Hsp31 Is a Stress Response Chaperone
study investigating the aggregation of a huntingtin-GFP fusion protein (61). Cell cultures co-expressing Sup35 and Hsp31 used in the microscopy experiments were prepared for flow cytometry and analyzed for EYFP fluorescence intensity. The flow cytometry results were similar to the microscopy results because Hsp31-expressing cells consistently exhibited lower median Sup35-EYFP fluorescence intensity when compared with cells containing Sup35-EYFP expression plasmid and empty vector (Fig. 8C). Individual traces depicting the number of counting events for each yeast strain also show the decrease in cellular fluorescence intensity when Hsp31 is expressed (data not shown).
Expression of Hsp31 with the GPD promoter was diffuse and cytoplasmic and varied in intensity in individual cells across the total cell population. The cytoplasmic localization profile of Hsp31 was not altered when PrD-Sup35 was expressed (Fig.  8A). Interestingly, we observed overlapping localization in cells with Sup35 aggregates and relatively high levels of Hsp31, but the proteins do not appear to mutually co-localize (Fig. 9A). In some cells, it was evident that the DsRed signal was decreased in the area where there was a Sup35 aggregate, suggesting that Hsp31 is not incorporated in aggregates (Fig. 9A). The aggregate appears to occlude Hsp31, and the diffuse overlapping signal observed in some cells is consistent with the presence of Hsp31 throughout the cytoplasm surrounding the aggregate. We also confirmed that Hsp31 overexpression reduces the formation of in vivo aggregates based on assessing cell lysates by SDD-AGE. We found that SDS-resistant Sup35 aggregates were greatly reduced when Hsp31 was overexpressed. The positive control, Hsp104, completely eliminated detectable Sup35 aggregates (Fig. 9B). These results are consistent with the ability of Hsp31 to inhibit the formation of large protein aggregates.
We also ascertained whether other chaperones are induced upon overexpression of Hsp31. We demonstrated that Hsp70 and Hsp104 expression are not altered, and hence elevated expression of these chaperones is not responsible for the observed suppression of prion aggregates (Fig. 9C). This is in agreement with our in vitro aggregation studies establishing that purified and recombinant Hsp31 protein is solely sufficient FIGURE 7. Hsp31 and catalytic mutants prevent the formation of ␣Syn oligomers. A, flow chart depicting the fractionation of the ␣Syn fibrillization assay and detection of monomeric and oligomeric species. SDS-PAGE of the pellet and supernatant fractions was performed, and antibodies were used to detect ␣Syn followed by stripping of the membrane and reprobing with anti-HA tag antibody to detect Hsp31 (note that Hsp31 was purified using the MORF tag, but Protein A was removed proteolytically, and the HA tag is retained). B, Hsp31 suppresses the formation of SDS-resistant higher order oligomers detected with Western blotting of the pellet and supernatant fractions. C, Hsp31 catalytic triad mutants suppress ␣Syn fibrillization. All proteins were added at the same level of 2 M. A.U., arbitrary units. OCTOBER 9, 2015 • VOLUME 290 • NUMBER 41 to prevent protein aggregation, although we cannot exclude the possibility that Hsp31 could collaborate with other chaperones to prevent prion aggregation in the cell.

Hsp31 Is a Stress Response Chaperone
Hsp31 Rescue of ␣Syn-mediated Toxicity Is Independent of the Autophagy Pathway-Recently, in several studies, autophagy has been identified as one of the main mechanisms to clear and recycle misfolded proteins and aggregates in the cytosol (62,63). In addition to the ubiquitin proteasome system, autophagy is a major cellular function involved in clearance of misfolded and aggregated proteins, including ␣Syn aggregates (62). Deletion of autophagy genes, such as ATG5 or ATG7, leads to neurodegenerative disease in mice (64,65). In yeast, deletion of the ATG1 gene reduced the clearance of ␣Syn aggregates (66). Deletion of HSP31 was also demonstrated to impair autophagy under carbon starvation conditions (22). These studies raise the possibility that Hsp31 could be promoting autophagy of aggregated protein. In order to exclude the possibility that the Hsp31 rescue effect on ␣Syn toxicity depends on autophagy, we deleted the ATG8 gene in the ␣Syn strains and assessed the synthetic lethal effects of ␣Syn expression in these strains. As expected, expression of ␣Syn in the atg8⌬ strain resulted in decreased viability compared with the wild-type strain. The atg8⌬ hsp31⌬ strain had viability similar to that of the atg8⌬ strain when ␣Syn was expressed (Fig. 10A). These results confirm that autophagy is a protective mechanism against ␣Syn toxicity. The lack of synthetic lethal defect between ATG8 and HSP31 with and without ␣Syn expression is consistent with genes that are in the same pathway and hence do not buffer each other. Expression of Hsp31 and the C138D mutant in the autophagy strains, atg8⌬ ␣Syn or atg8⌬ hsp31⌬ ␣Syn, increased viability in rich YEP media compared with vector controls (Fig. 10B). We found that the rescue effect in synthetic media was not easily observed because expression of ␣Syn was not toxic in the atg8⌬ strain background, but it was toxic in the hsp31⌬, indicating that media conditions are an important factor in this complex relationship (Fig. 10B, mid- dle). These results demonstrate that Hsp31 can rescue ␣Syn toxicity in autophagy-deficient cells.

Discussion
The roles of chaperones and proteases in maintaining integrity and activity of folded, partially folded, and misfolded proteins are well established. Members of the DJ-1/ThiJ/PfpI superfamily exist in prokaryotes and eukaryotes, and structural and biochemical analyses have established that they have multiple activities, including chaperone, protease, and enzymatic functions. Because of the multiple activities associated with this superfamily, we sought to characterize the chaperone activity of Hsp31 and delineate its contribution in protecting against cellular stress.
The crystal structure of yeast Hsp31 has been solved, and structural evidence shows that Hsp31 and hchA share high similarity, in terms of the position of a hydrophobic patch and a catalytic triad. Yeast Hsp31, E. coli hchA, and human DJ-1 are in the same DJ-1/PfpI/ThiJ superfamily, and the literature suggests that both DJ-1 and hchA have chaperone-like effects against protein aggregation. Despite these similarities, the structures differ in important details, and it could have been argued that Hsp31 might not possess chaperone activity based on the absence of 45 amino acids present in hchA and the distinct dimerization. We investigated whether the chaperonelike activities are conserved in Hsp31 and found that the protein showed broad chaperone activity on three different substrate proteins, with aggregation induced by different mechanisms. CS aggregation is triggered by elevated temperature, and we demonstrated that Hsp31 suppresses CS aggregation 6-fold more efficiently then hchA and DJ-1. Hsp31 also suppresses insulin aggregation, demonstrating that Hsp31 can maintain chaperone activity in a reducing environment in contrast to the redox-dependent DJ-1, which does not have activity in this assay.
Hsp31 also exhibited considerable activity in an in vitro ␣Syn fibrillization assay and completely suppressed fibrillization at a molar ratio of 8.75:1 (␣Syn/Hsp31) and maintained considerable activity at a ratio of 35:1. DJ-1 also had activity at substoichiometric levels but required higher molar ratios compared with Hsp31. We consistently observed that the Hsp31-MORF purified from yeast was more active than recombinant Hsp31, DJ-1, or hchA (Fig. 1, B and D). This could be due to the fusion tag type or fusion orientation affecting protein activity, which has been noted previously for hchA (17,18). Further studies are needed to delineate the source of these activity differences, including the possibility that it is related to the purification procedure, oxidation state of the protein, oligomeric state, or post-translational modifications. The broad range of substrates and substoichiometric activity all indicate that Hsp31 is a chap-erone with robust activity compared with other members of this superfamily.
Despite the fact that its cellular function is incompletely understood, ␣Syn is known to aggregate and form inclusions. These protein inclusions are precursors of Lewy bodies, the pathological hallmark of PD. Elucidating mechanisms of ␣Syn aggregation is important to understand PD pathogenesis, and the heterologous expression of ␣Syn in yeast established a useful platform to investigate ␣Syn biology (9). We extended the ␣Syn yeast model by demonstrating the modulation of ␣Syn toxicity and subcellular localization by Hsp31. Expression of 2xGAL-␣Syn is toxic to yeast cells, and constitutive expression of Hsp31 from the GPD promoter rescued cells from ␣Synmediated toxicity (Fig. 2). In addition, expression of one copy of GAL-␣Syn is not toxic to wild-type cells but is toxic in the hsp31⌬ strain. Taken together, it appears that the chaperone effect of Hsp31 is necessary to maintain cellular fitness upon ␣Syn expression.
Hsp31 also altered the subcellular localization profile of ␣Syn. Expression of 2xGAL-␣Syn results in cytosolic accumulations (foci), whereas the expression of one copy of GAL-␣Syn results in primarily plasma membrane localization. The in vivo FIGURE 10. The autophagy pathway prevents ␣Syn toxicity but is not required for Hsp31-mediated rescue. A, yeast strains serially diluted on YEP agar plates (glucose or raffinose/galactose) demonstrating the toxicity of GAL-␣Syn in the atg8⌬, hsp31⌬, and atg8⌬ hsp31⌬ strains. B, overexpression of HSP31 using pAG415-GPD-HSP31-DsRed or the C138D mutant partially rescued toxicity of ␣Syn expression on YEP media for the atg8⌬, hsp31⌬, and atg8⌬ hsp31⌬ strains (right). Rescue by HSP31 overexpression could not be assessed in synthetic media for strains with the atg8⌬ genotype because GAL-␣Syn expression was not toxic in these strains (middle). Rescue of GAL-␣Syn expression in the hsp31⌬ strain was toxic and rescued by HSP31 overexpression in synthetic and YEP media.
cytoplasmic foci are associated with vesicular membranes (39), and analysis of cellular lysates indicates the presence of higher order oligomeric forms of ␣Syn, although the nature of this aggregation is unclear (23,67). Expression of GAL-HSP31 results in more than 2-fold suppression of foci and increases the localization of ␣Syn to the plasma membrane (Fig. 3A). This effect was transiently observed at 8 h postinduction, but a high rate of puncta formation returned after 24 h, possibly because Hsp31 expression was induced at the same time as ␣Syn. This is consistent with the result where GAL-HSP31 does not rescue the synthetic lethal effect of ␣Syn expression in the hsp31⌬ strain, whereas the constitutive GPD-HSP31 rescues hsp31⌬.
These results indicate that a higher threshold level of Hsp31 is needed in the cell prior to ␣Syn induction to prevent toxicity. The exact temporal relationship between Hsp31 expression and ␣Syn foci suppression could provide insight into the aggregation state of ␣Syn species targeted by Hsp31.
Several orthologous experiments suggest that Hsp31 interacts early in the misfolding process and prevents the formation of larger oligomeric species. The addition of Hsp31 to ␣Syn monomers in a fibrillization assay resulted in a baseline level of ThioT fluorescence signal, indicating that the formation of larger oligomeric species was prevented (Fig. 1D). Analysis of the SDS-resistant oligomeric species by SDS-PAGE demonstrated that the presence of Hsp31 prevented the formation of higher order oligomeric ␣Syn species (Fig. 7A). Our prion aggregation studies indicate that Hsp31 inhibits prion assembly before the formation of subcellular visible GFP-tagged PrD aggregates and those detected by SDD-AGE (Fig. 9B). Furthermore, the Sup35 aggregates occlude Hsp31, indicating that the anti-aggregation activity of Hsp31 commences prior to the formation of the visible aggregate and probably does not act to remove or disassemble any preformed aggregate (Fig. 9A). This is in contrast to large chaperones, Hsp104, Ssa1/2, Sis1, and Sse1, which mutually colocalize with prion aggregates (68). A previous study has demonstrated that small HSPs can inhibit the formation of Sup35 aggregates and found that Hsp26 and Hsp42 inhibit rare transient oligomers at distinctly different steps in the prion formation process (69). Further studies will be required to determine whether Hsp31 inhibits protein misfolding with a similar or distinct mechanism as Hsp26 or Hsp42. Overall, we demonstrate that Hsp31 can inhibit oligomerization or aggregation of many different proteins by intervening early in the process.
The role of Hsp31 in protecting cells from ␣Syn-mediated toxicity led us to investigate the expression of Hsp31 in response to proteotoxic stress and other types of stress. The expression pattern of Hsp31 in a genomically tagged strain decreased during log phase growth and increased at the diauxic shift and at saturation (Fig. 4). The increased expression of Hsp31 at the diauxic shift and in stationary phase could possibly be due to the stress of decreased nutrients or accumulation of metabolites associated with these growth phases and is part of a global transcriptional response program (70). An increase in Hsp31 expression in response to ROS has been observed previously (71), and we confirmed that our tagged strain behaved in a similar manner by demonstrating an increase in expression upon exposure to H 2 O 2 (Fig. 4B). This ROS response has been implicated to be dependent on the Yap1 stress response transcription factor, and a shortened Hsp31 promoter (Ϫ313 bp relative to the ATG), which removes the Yap1 binding site (Ϫ363 and Ϫ353 bp relative to the ATG), has decreased Hsp31 expression in response to oxidative stress (72). However, we note that the shortened promoter also removes an additional stress response element (a CCCCT site at Ϫ379 to Ϫ383 from the ATG) in proximity to the Yap1 binding site. Stress response elements have been demonstrated to be the binding site for Msn2/4, and numerous stress-related yeast genes have been documented to be transcriptionally activated by these transcription factors (26), including genes up-regulated after the diauxic shift in stationary phase cells (26). Hsp31 appears to be part of the global transcriptional stress response by Yap1 and Msn2/4 or other stress-related transcription factors, such as Hsf1 or Gis1.
Finally, we also examined the response of Hsp31 expression during ␣Syn-initiated proteotoxic stress and observed an increased expression in cells in log phase. We measured an increase in superoxide levels upon ␣Syn expression consistent with a previous report that indicated that ␣Syn expression is associated with caspase-mediated ROS generation (73) (Fig.  3D). Deletion of HSP31 also leads to increased superoxide levels with or without ␣Syn expression. These results establish that Hsp31 has a role in response to ROS, and its expression is induced in response to several types of stress.
We demonstrated the robust chaperone activity of Hsp31, but the in vivo rescue effects could have been mediated by the Hsp31 methylglyoxalase activity or interaction with other biological pathways, such as the metabolic or autophagy pathway. We confirmed that Hsp31 does have methylglyoxalase activity and demonstrated that mutation of the catalytic triad residue, C138D, results in greatly reduced methylglyoxalase activity (Fig. 5, A and B). However, this mutation did not inactivate the chaperone activity in the ␣Syn fibrillization assay (Fig. 7C), and expression of GPD-HSP31 C138D rescued cells from ␣Syn-mediated toxicity (Fig. 5C). Furthermore, we showed that that exogenous MGO did not differentially affect viability, suggesting that Hsp31 does not rescue by reducing accumulation of this toxic metabolite in ␣Syn-expressing cells (Fig. 6).
We also demonstrated that the autophagy pathway is an important mediator in the ␣Syn toxicity because deletion of ATG8 resulted in synthetic lethal interaction with ␣Syn expression (Fig. 10A). This is consistent with atg1⌬ or atg7⌬ strains having synthetic lethal interactions with ␣Syn expression and having defects in clearing ␣Syn foci (74). However, we show that the hsp31⌬ atg8⌬ strain does not have increased ␣Synmediated toxicity when compared with atg8⌬, consistent with these genes occurring within the same synthetic lethal pathway. A recent result indicates that hsp31⌬ strains are defective in autophagy (51), so the addition deletion of an autophagy gene should not increase toxicity. Interestingly, overexpression of Hsp31 in these autophagy-deficient strains can rescue cells from ␣Syn toxicity, indicating that the autophagy pathway is not essential for chaperone rescue (Fig. 10B). Autophagy may be beneficial for controlling ␣Syn toxicity, but it appears that chaperone activity of Hsp31 is also important. These experiments also are the first to delineate that enzymatic methylg-Hsp31 Is a Stress Response Chaperone OCTOBER 9, 2015 • VOLUME 290 • NUMBER 41 lyoxalase activity of Hsp31 is not essential to rescue ␣Syn toxicity.
Our results suggest that Hsp31 functions as a cytoplasmic chaperone against a variety of misfolded proteins. Recent studies have demonstrated the facility of yeast to investigate pathogenic mechanisms underlying ␣Syn toxicity, including the identification of novel biological pathways that impinge on ␣Syn biology and small molecule modulators of ␣Syn toxicity (66,69). In addition, the action of Hsp31 on ␣Syn may provide insight into the mode of toxicity of ␣Syn because recent evidence suggests that the ␣Syn toxic species is smaller than the visible aggregate (66). It should be noted that the prion expression studies done in this study were not completed with [PSIϩ] strains, but further studies using such strains would be of interest to determine the effect of Hsp31 on prion initiation and propagation.
We have established a framework that extends the yeast model for investigating mechanisms of ␣Syn toxicity in the context of the DJ-1/ThiJ/PfpI superfamily in yeast. Extension of this work will assist in elucidation of the chaperone-like mechanisms of Hsp31, and a comparison with DJ-1 may provide evolutionary insights into the activities of the DJ-1/PfpI/ThiJ superfamily. Our model may also be used to further delineate the nature of oligomeric species in the pathogenesis of PD and possibly what species of ␣Syn should be targeted therapeutically.