The Activity of Yeast Hog1 MAPK Is Required during Endoplasmic Reticulum Stress Induced by Tunicamycin Exposure*

Accumulation of unfolded proteins in the endoplasmic reticulum (ER) triggers the so-called unfolded protein response (UPR), a conserved signaling pathway that drives the transcription of genes such as chaperones and folding enzymes. Nevertheless, the activity of the UPR accounts only for a part of the gene expression program activated upon ER stress. Moreover, the mechanism(s) for how cells adapt and survive to this stress are largely unknown. Here, we show that the yeast high osmolarity glycerol (HOG) pathway plays a role in ER stress resistance. Strains lacking the MAPK Hog1p displayed sensitivity to tunicamycin or β-mercaptoethanol, whereas hyperactivation of the pathway enhanced their resistance. However, these effects were not due to Hog1p-mediated regulation of the UPR. Northern blot analysis demonstrated that Hog1p controls the tunicamycin-induced transcriptional change of GPD1 and that wild-type cells exposed to the drug accumulated glycerol in a Hog1p-dependent manner. Consistent with this, deletion of genes involved in glycerol synthesis caused increased sensitivity to tunicamycin, whereas overexpression of GPD1 provided higher tolerance to both wild-type and hog1Δ mutant cells. Quite remarkably, these effects were mediated by the basal activity of the MAPK because tunicamycin exposure does not trigger the phosphorylation of Hog1p or its nuclear import. Hence, our results describe new aspects of the yeast response to ER stress and identify additional functions of glycerol and the Hog1p MAPK to provide stress resistance.

Cells have evolved sophisticated mechanisms to synthesize correctly folded proteins and to ensure that only fully functional molecules are released and targeted to their final destination. Stringent quality control is achieved through the action of sensors located at the endoplasmic reticulum (ER), 4 the cellular compartment where most secreted and plasma membrane proteins fold and mature. When environmental conditions increase the load of unfolded proteins (ER stress), these ER resident sensors induce adaptive responses, collectively termed the unfolded protein response, UPR (1,2). In mammalian cells, ER stress triggers at least three individual signaling pathways or branches of the UPR (3) initiated at the level of ER membraneanchored proteins, IRE1 (inositol-requiring protein-1), ATF6 (activating transcription factor-6), and PERK (protein kinase RNA-like ER kinase) respectively, that collaborate to activate subsets of UPR target genes (4). As a result, the synthesis of new proteins decreases and the folding capacity of ER increases. Upon persistent ER stress, the UPR also triggers apoptosis (5). Thus, the UPR combines short and long term responses that determine survival or cell death. Accumulating evidence also suggests that activation of some UPR branches is important in tumor development and growth (6). While the mechanisms by which UPR signals are transmitted from the ER to the nucleus are well characterized, less is known of how these different signals are selectively regulated and controlled and how their activity determines the response of the cell to ER stress.
In the yeast Saccharomyces cerevisiae, Ire1p appears to be the only element responsible for sensing the protein folding status of ER and transmitting this information to downstream effectors in the cytosol (7,8). Indeed, no ATF6 ortholog has yet been discovered in yeast, whereas Gcn2p, the putative functional counterpart to PERK, has not been directly implied in the control of UPR target genes. Activation of the transmembrane kinase/nuclease Ire1p (9,10), causes the endonucleolytic cleavage of the HAC1 mRNA (11,12), a transcription factor named Xbp-1 (X-box binding protein-1) mRNA in mammalians (13,14). Spliced Xbp-1/HAC1 mRNA translocates to the nucleus and drives the transcription of genes such as ER chaperones and folding enzymes (4,15), thereby increasing the protein folding capacity. Nevertheless, the activity of the Ire1p-Hac1p system appears to account only for a part of the gene expression program activated upon ER stress. For instance, Ͼ100 open read-* This work was supported by the Comisió n Interministerial de Ciencia y Tecnología project (AGL2007-65498-C02-01) from the Ministry of Science and Technology of Spain. This work was also supported in part by Instituto de Fisiología Celular, Universidad Nacional Autó noma de Mé xico CONACyT Grant 80343. □ S The on-line version of this article (available at http://www.jbc.org) contains supplemental Tables S1 and S2, Fig. S1, and additional references. 1 Supported by Predoctoral Fellowship 185969 within the "Becas Mixtas" program from CONACyT. 2 Supported by a predoctoral fellowship from the I3P Program from Consejo Superior de Investigaciones Científicas. 3 To whom correspondence should be addressed: Instituto  ing frames induced by the UPR and with a known function have no apparent connection with expected ER or secretory pathway functions (4). Hence, ER stress could be influenced by the activity of phosphatases and/or imply additional effectors or signaling routes traditionally associated with different stress situations. For example, mammalian IRE1 signaling activates the c-Jun N-terminal kinase (JNK) pathway (16). Likewise, the SLT2 MAPK pathway and the osmosensing HOG MAPK pathway have been identified as required for survival of yeast cells during ER stress (17). However, the exact function of these signaling pathways in the stress adaptation of ER has not been clarified.
The HOG pathway is one of the five MAPK pathways characterized so far in S. cerevisiae (18 -21). It has been implicated in activating different physiological functions and protective mechanisms in response to high osmolarity, although several studies have revealed additional functions for this MAPK route. Evidence indicates that the HOG pathway is essential for regulating adaptation to citric acid (22), heat stress (23), and low temperature (24). HOG basal activity is also involved in methylglyoxal resistance (25), distribution of proteins within the Golgi (26), and cell wall maintenance (27). The HOG pathway consists of two discrete signaling branches composed of putative osmosensors coupled to a MAPK cascade, some of which can lead to the phosphorylation and activation of the core MAPK Hog1p, the ortholog to mammalian p38 and fission yeast Sty1p SAPK (28). Osmostress-induced phosphorylation of Hog1p triggers its nuclear accumulation (29,30) and the later induction of many osmostress-responsive genes (31,32), a process that is regulated at different levels. Active Hog1p phosphorylates several transcription factors, associates at stress-responsive promoters, serves as a platform to recruit general transcription factors, and has a role in elongation (33). The activity of Hog1p is therefore essential in providing a rapid change on the cellular transcriptional capacity upon osmotic stress and consequently elucidating the function of the MAPK in response to other environmental cues is of major interest.
Here, we analyzed the activation status of Hog1p upon ER stress and characterized the molecular mechanisms mediating the functional role of this MAPK in the survival of yeast to unfolded protein accumulation. Our results have shown that in S. cerevisiae cells, Hog1p provides a necessary function to cope with enhanced levels of unfolded proteins. This function is apparently basal and independent of the canonical UPR signaling system through Ire1p-Hac1p. Our results also confirm the importance of the osmolyte glycerol in providing cell protection during unfolded protein stress.

EXPERIMENTAL PROCEDURES
Media and Culture Conditions-Yeast cells were cultured at 30°C in defined medium, YPD (1% yeast extract, 2% peptone, and 2% glucose) or synthetic defined medium (0.17% yeast nitrogen base without amino acids (Difco), 0.5% ammonium sulfate, and 2% glucose) supplemented with the appropriate dropout mixtures (complete supplement mixture, Formedium TM , England) to support a vigorous growth of cells and maintain selection for plasmids (34). Activation of the UPR was monitored in YPD or synthetic defined medium containing tunica-mycin or ␤-mercaptoethanol at the indicated concentration. S. cerevisiae transformants carrying the geneticin (kanMX4) or the nourseothricin (natMX4)-resistant module were selected on YPD agar plates containing 200 mg/liter of G-418 (Sigma) or 50 mg/liter of nourseothricin (clonNAT, WERNER Bioagents), respectively. E. coli was grown in LB medium supplemented with ampicillin (50 mg/liter). A stock solution of 25 mg/ml tunicamycin was prepared in DMSO, sampled in small volumes, and stored at Ϫ20°C until use. For each experiment, a fresh sample was thawed and diluted to the working concentration.
ER Stress Sensitivity Tests-For stress experiments, cells were grown to mid-exponential phase at 30°C, collected, and transferred to fresh medium containing the corresponding ER stress-inducer drug. Plate phenotype experiments were made by diluting the cultures to A 600 ϭ 0.5 and spotting (3 l) 10-fold serial dilutions. When tunicamycin was used, control assays containing DMSO alone were included to discard a specific effect of the dissolvent. Unless indicated, colony growth was inspected after 2-4 days of incubation at 30°C.
Northern Blot Analysis-Total RNA was extracted, purified, and separated as described previously (24). Then, samples were transferred to a nylon membrane and hybridized with nonradioactive digoxigenin-labeled probes containing sequences of KAR2 (ϩ151 to ϩ751), GPD1 (ϩ43 to ϩ835), HAC1 (ϩ18 to ϩ657), or ACT1 (ϩ10 to ϩ1075). DNA sequences were obtained from the MIPS (Munich information center for protein sequences) database. PCR labeling of DNA probes, membrane pre-hybridizations and hybridizations were performed with the PCR digoxigenin probe synthesis kit and Digoxigenin Easy Hyb solution of Roche (Roche Diagnostics GmbH, Mannheim, Germany). After stringency washes, the blots were subjected to immunological detection using anti-digoxigenin antibody conjugated to alkaline phosphatase (Roche Diagnostics), followed by CDP-Star detection (Roche Diagnostics). Images were captured with the Las-1000 Plus imaging system (Fuji, Kyoto, Japan). Spot intensities were quantified with Image Gauge software (version 3.12; Fuji). Values of spot intensity were corrected with respect to the ACT1 mRNA level and represented as the relative mRNA level. The highest relative mRNA for each gene, and the analyzed sample was set to 100.
Western Blot-Whole-cell extracts for Hog1p detection, protein separation by SDS-PAGE, and electroblotting were carried out as described (24). Dual phosphorylated Hog1p was detected by an antibody specific to phosphorylated p38 MAPK (reference no. 9215, Cell Signaling, Beverly, MA). A rabbit polyclonal antibody raised against a recombinant protein corresponding to the carboxyl terminus-(221-435) of S. cerevisiae Hog1p was used as a loading control (reference no. sc-9079, Santa Cruz Biotechnology, Santa Cruz, CA). The antisera were applied at 1:1000 (phosphorylated Hog1p) and 1:6000 (total Hog1p) dilutions. As secondary antibody, we used horseradish peroxidaseconjugated goat anti-rabbit (1:2000, reference no. 7074, Cell Signaling). Blots were developed using the ECL Western blotting detection kit from Pierce. Images were captured as described above.
Fluorescence Microscopy-Cells were cultured as described. One ml samples were taken at different times before and after 2 g/ml tunicamycin addition (30, 60, 90, and 180 min) or 5 min after exposure to 0.5 M NaCl (positive control). Samples were spun for 10 s, the supernatant was decanted, and cells were resuspended in the residual liquid. Aliquots of 10 l were directly observed in a 100ϫ microscopic field employing filters for GFP and mCherry fluorescence. The merge images were obtained with software from the manufacturer (Olympus Fluoview FV100, version 1.4.) Raw images were superposed, cropped, and adjusted to equivalent brightness and contrast by using Photoshop CS (Adobe Systems). At least 100 cells were visualized each time.
Glycerol Assay-To determine total glycerol content under different stress conditions, YPD grown cells (A 600 ϭ 0.5) were collected by centrifugation and resuspended in the same medium containing 2 g/ml tunicamycin, DMSO alone, or 1 M sorbitol. At different times, aliquots (1 ml) of the cultures were removed, boiled for 10 min, and centrifuged at 15,300 ϫ g for 10 min (4°C), and the supernatants were assayed for glycerol content. Glycerol was determined colorimetrically with a commercial kit (BEN Biochemical Enterprise, Code TG9940), following the manufacturer's instructions. The values obtained are expressed as g of glycerol per mg of yeast cells, dry weight. Growth throughout the experiment was estimated by measuring as the A 600 (A 600 ϭ 1 equals 0.3 mg cells dry weight/ml). For tunicamycin-treated samples, increase in glycerol content was estimated after subtraction of the values measured in the control assays containing DMSO alone. The values given represent the average of three independent experiments, each conducted in triplicate.
␤-Galactosidase Assay-Aliquots of yeast cultures (15 units of A 600 ) were harvested, washed with Z buffer (60 mM Na 2 HPO 4 , 40 mM NaH 2 PO 4 , 10 mM KCl, and 1 mM MgSO 4 ), and centrifuged at 3,000 ϫ g for 2 min (4°C), and the cell pellets were frozen at Ϫ20°C for further analysis. Cell extracts were prepared as described previously (44). Total protein was determined using the Bio-Rad Bradford assay kit and bovine serum albumin as standard protein. ␤-Galactosidase activity was determined at room temperature using the substrate ONPG as described previously (45). One unit is defined as the amount of enzyme that is able to convert 1 nmol of ONPG per min under the assay conditions. The values given represent the average of three independent experiments, each conducted in triplicate.

RESULTS
The MAP Kinases Pbs2p and Hog1p Are Important Determinants of Survival to ER Stress in S. cerevisiae-We first tried to confirm the implication of the HOG pathway in the adaptive response of yeast cells to ER stress. To do this, we examined growth in the presence of agents that induce unfolded protein stress using a strain lacking IRE1 as representative of a sensitivity phenotype. As expected, deletion of HOG1 in the W303-1A wild-type strain, resulted in sensitivity to tunicamycin (Fig. 1A), a natural inhibitor of N-linked glycosylation that is widely FIGURE 1. The HOG pathway plays a role in determining the survival of S. cerevisiae to ER stress. Cells of HOG pathway mutant strains were assayed for growth in the presence of ER stress-inducer agents, tunicamycin (Tn) and ␤-mercaptoethanol (␤-ME). A, growth was examined in cells of the W303-1A wild-type (wt), hog1⌬, pbs2⌬, sho1⌬, and ssk1⌬ strains exposed to tunicamycin at the indicated concentration. Cells of the UPR mutant ire1⌬ were used as indicative of a sensitive strain. B, wild-type cells of the BY4741 and corresponding hog1⌬ and pbs2⌬ mutant strains were assayed for sensitivity to tunicamycin and ␤-mercaptoethanol. In all cases, cultures were grown in YPD at 30°C until exponential phase and adjusted to A 600 ϭ 0.5. Then, serial dilutions (1 ϫ 10 Ϫ3 ) of the cultures were spotted (3 l) onto YPD plates lacking or containing the ER-stress inducer to be tested and incubated at 30°C for 2-4 days, respectively. A representative experiment is shown. In the case of tunicamycin, the presence of DMSO had no effect on cell growth (see Fig. 2 for a control experiment, YPD ϩ DMSO).
employed as an inducer of ER stress (46). Similar results were observed in a strain lacking Pbs2p, the MAPKK of the HOG pathway (Fig. 1A). These effects were not strain-dependent, as hog1⌬ and pbs2⌬ derivatives of the S. cerevisiae BY4741 wildtype strain showed the same phenotype (Fig. 1B), although at higher drug concentrations (0.5 g/ml). Furthermore, the same mutant strains of the HOG pathway were also sensitive to the presence of ␤-mercaptoethanol (Fig. 1B), a reducing agent that also induces ER stress. Hence, key MAPKs of the HOG pathway are essential to cope with chemicals agents that generate accumulation of unfolded or misfolded proteins in the ER.
Tunicamycin Resistance Depends on the Sln1 Branch of the HOG Pathway-Signaling through the HOG pathway is triggered by two independent mechanisms, composed by the membrane protein Sho1p and the Sln1p-Ypd1p-Ssk1p phosphorelay system. Therefore, we analyzed whether either or both of these branches could be involved in the ER stress response. Mutation of SLN1 causes lethality (47), so this sensor system can only be blocked by disruption of SSK1. As shown in Fig. 1A, the ssk1⌬ mutant displayed tunicamycin sensitivity, whereas no effect was found in the sho1⌬ mutant. Thus, only the Sln1p branch of the HOG pathway appears to be involved in mediating this response.
The Activity of the HOG Pathway Determines the Level of Sensitivity to ER Stress-The protein phosphatases Ptp2p and Ptp3p have been shown to down-regulate the HOG pathway by dephosphorylation of the MAPK (48,49) and regulating its subcellular localization (50). Consequently, deletion of PTP2 and PTP3 genes leads to an overactivated Hog1p. We analyzed the effect of tunicamycin on the growth of ptp2⌬, ptp3⌬, and ptp2⌬ptp3⌬ mutant cells. As can be seen in Fig. 2, the ptp2⌬ mutant strain showed enhanced resistance to tunicamycin as compared with the wild-type strain. On the contrary, deletion of PTP3 had no effect on sensitivity, although cells of the double disruption mutant ptp2⌬ptp3⌬ were more resistant than the single ptp2⌬ mutant (Fig. 2). Prior work has shown that Ptp3p plays a minor role in the phosphorylation level of Hog1p (49,50). Finally, a single deletion of PTP2 or combined PTP2 and PTP3 in the hog1⌬ mutant strain was unable to confer resistance to tunicamycin (Fig. 2). Hence, elevated tolerance to ER stress of ptp2⌬ and ptp2⌬ptp3⌬ mutant strains is fully dependent on the presence of functional Hog1p activity.
We next examined the tunicamycin sensitivity of hog1⌬ mutant cells expressing a mutated HOG1 allele (K52R, HOG1-KD) that lacks kinase activity (40). Mutant cells transformed with plasmids containing either a native HOG1 or a nonphosphorylatable HOG1 allele (T174A/Y176F, HOG1-nP), were also analyzed. As shown in Fig. 3, a HOG1-KD allele was unable to complement the sensitivity of hog1⌬ mutant cells in tunicamycin-containing plates, whereas transformants expressing a wild-type HOG1 showed a similar growth than the parental strain. Expression of the HOG1-nP allele improved the growth of hog1⌬ cells in the presence of tunicamycin (Fig. 3), although this was still lower than that displayed by the wild-type strain. We presume that mutating the two consensus Hog1p phosphorylation sites affects its kinase activity, making cells somewhat more susceptible to the drug. These results were not the consequence of differences in the level of Hog1p, as demonstrated a Western blot assay of protein extracts revealed with anti-Hog1p (data not shown). Hence, our results indicate that the kinase activity of Hog1p is required to deal with the N-glycosylation defects promoted by tunicamycin exposure.  Table S1) were made isogenic for auxotrophic markers by transforming with the integrative plasmids described by Gietz and Sugino (35), YIplac204 (TRP1), YIplac211 (URA3), and/or YIplac128 (LEU2). As a control, growth was inspected on YPD plates lacking tunicamycin or containing DMSO alone. Cells were pregrown, diluted, plated, and incubated as described in Fig. 1. A representative experiment is shown. FIGURE 3. Tunicamycin sensitivity of hog1⌬ mutant cells expressing mutated HOG1 alleles. hog1⌬ mutant cells were transformed with plasmids containing a native (HOG1), a nonphosphorylatable (HOG1-nP) or a kinase dead (HOG1-KD) HOG1 allele and examined for growth upon exposure to tunicamycin (Tn). Wild-type (wt) and ire1⌬ mutant cells containing an empty plasmid were used as control. Cells were pregrown, diluted, plated, and incubated as described in Fig. 1. A representative experiment is shown. JUNE 25, 2010 • VOLUME 285 • NUMBER 26

JOURNAL OF BIOLOGICAL CHEMISTRY 20091
Hog1p Phosphorylation and Nuclear Import-We examined the phosphorylation state of Hog1p in cells exposed to tunicamycin. Supplemental Fig. S1A shows the results of a Western blot analysis of protein extracts analyzed with anti-phospho p38 antibody, and polyclonal anti-Hog1p as a loading control. As can be seen, we could not detect phosphorylation of the MAPK during 120-min treatment of yeast cells with 2 g/ml tunicamycin. Total Hog1p protein was neither seen to change (supplemental Fig. S1A). Similar results were obtained when cells were exposed to tunicamycin for extended periods or to higher concentrations, i.e. 2-20 g/ml (data not shown). Only in a ptp2⌬ptp3⌬ mutant, the level of Hog1p phosphorylation appeared to show a very weak increase after tunicamycin treatment (supplemental Fig. S1B). Thus, overphosphorylation of the MAPK does not appear to be a good indicator of its activity in response to ER stress.
To corroborate this idea, we monitored the distribution of an Hog1p-tagged GFP fusion protein in wild-type yeast cells treated with tunicamycin or NaCl as control. As mentioned previously, once overphosphorylated, the level of nuclear Hog1p shows a transient increase (29,30). Therefore, colocalization of Hog1-GFP with Htb2-mCherry (38), a protein containing the nuclear histone Htb2p (51) fused with the red fluorophore mCherry, was used as evidence of this event. As shown in supplemental Fig. S1C, Hog1-GFP in tunicamycintreated cells was located in the cytosol. On the contrary, nuclear accumulation of the MAPK was detectable after 5 min of the onset of high osmolarity stress, as evidenced the appearance of overlapped fluorescence in the merged images (supplemental Fig. S1C).
The Activity of the HOG Pathway Does Not Alter UPR Signaling-We went into the mechanism(s) underlying the effect of the HOG pathway in the resistance of yeast cells to tunicamycin. First, we measured the activation of an UPRE::lacZ reporter, which contains the UPRE (CAGCGTG) fused to the E. coli lacZ gene (12), in mutant cells exposed to 2 g/ml tunicamycin (Fig. 4A). As expected, no significant ␤-galactosidase activity could be detected in S. cerevisiae cells lacking the kinase/endoribonuclease Ire1p. In contrast, ϳ35-fold inductions were observed in wild-type cells carrying multiple copies of the reporter construct after 90 min of exposure to the drug. Deletion of HOG1 or PBS2 MAPK genes had no effect on the induced level of ␤-galactosidase activity. Other deletion mutants, such as ssk1⌬ or sho1⌬, also showed activated UPREregulated expression similar to that observed in the W303-1A wild-type strain (Fig. 4A).
Next, we monitored HAC1 splicing by Northern blot analysis of total RNA from wild-type and hog1⌬ mutant cells exposed to tunicamycin. Activated Ire1p promotes splicing of HAC1 precursor mRNA (HAC1 u , for uninduced), producing the mature form (HAC1 i , for induced) of the transcription factor (11,53). According to this, HAC1 splicing was absent in ire1⌬ mutant cells exposed to tunicamycin (Fig. 4B). On the contrary, the splicing reaction was evident in cells of either hog1⌬ or wildtype strain (Fig. 4B).
Finally, we analyzed the epistatic effects on tunicamycin and NaCl sensitivity of HOG1 and IRE1. As can be seen in Fig. 4C, growth inhibition by tunicamycin exposure was more pro-nounced in the double hog1⌬ire1⌬ mutant than in either of the corresponding single hog1⌬ or ire1⌬ mutants. In addition, the combined effects appeared to be entirely additives. Furthermore, no effect on NaCl resistance was observed by absence of Ire1p (Fig. 4D), and the double mutant showed the same phenotype as the single hog1⌬ mutant. Altogether, our results suggest that these two proteins are involved in different processes in response to ER stress and that activation of the UPR signaling is independent of Hog1p.
Hog1p Plays a Role in the Transcriptional Response to ER Stress-The failure to detect a functional link between the HOG and the UPR pathways led us to investigate more in depth the transcriptional role of this MAPK in response to ER stress. For this, the mRNA levels of different marker genes were probed by Northern blot of RNA samples from wild-type, ire1⌬ and hog1⌬ mutant cells exposed to tunicamycin. Fig. 5 shows the results for KAR2, which encodes a heat shock protein of the ER lumen (54) and GPD1, the gene for glycerol-3-phosphate dehydrogenase, the main enzyme involved in the synthesis of glycerol (55). Expression of KAR2 upon ER stress depends on the Ire1p -Hac1p system (4, 56), FIGURE 4. Interaction between the UPR and the HOG pathways. A, the activity of a UPRE-lacZ reporter was assayed in cells of the W303-1A wild-type (wt), hog1⌬, pbs2⌬, sho1⌬, and ssk1⌬ strains exposed to 2 g/ml tunicamycin (Tn) for 90 min. Cells of the UPR mutant ire1⌬ were used as indicative of a strain unable to induce the expression of the reporter. Data represent the mean value Ϯ S.E. of three independent experiments. B, cells of the indicated strains were treated with tunicamycin and at the indicated times, and aliquots were withdrawn and processed for Northern blot analysis of HAC1 mRNA splicing. HAC1 u (uninduced) and HAC1 i (induced) forms of HAC1 mRNA are indicated. A representative experiment is shown. C, growth of the W303-1A wild-type strain and corresponding mutants hog1⌬, ire1⌬, and hog1⌬ire1⌬, was examined in the presence of tunicamycin at the indicated concentrations. D, the same strains were tested for growth under osmotic stress provided by NaCl. For growth experiments, cells were pregrown, spotted, and incubated as described in Fig. 1. whereas GPD1 has been reported to show little or no induction under these conditions (4).
As anticipated, KAR2 was early up-regulated in tunicamycintreated cells of the W303-1A wild-type strain. In addition, its regulation was strongly decreased by deletion of IRE1. On the contrary, lack of the MAPK Hog1p had no major effects. Only the basal level of transcript was found to be increased in the hog1⌬ mutant (Fig. 5). This result reinforces the idea that Hog1p plays a role in the basal (noninduced) expression of different yeast genes, as it has been reported previously (25,(57)(58)(59). Comparing with KAR2, the expression profile of GPD1 showed a slow induction that reached near maximum values after 120 min. We also noted that GPD1 expression was clearly diminished by deletion of HOG1 (Fig. 5). On the contrary, lack of a functional Hac1p-Ire1p pathway had no detrimental effects or even appeared to speed up the genetic response of GPD1 (Fig. 5, see 30-min lane). Similar results were observed for other UPR-dependent and -independent marker genes (data not shown).
Tunicamycin Exposure Triggers the Overproduction of Glycerol, which Provides Cell Protection against the Drug-We analyzed whether the exposure of yeast cells to tunicamycin triggers the production of glycerol. The synthesis of the osmolyte is a main feature of the yeast response to osmotic stress and requires Hog1p activity for maximal production (55). Consequently, cells exposed to 1 M sorbitol were analyzed as a positive control. As shown in Fig. 6A, when logarithmic growing cells of the BY4741 wild-type strain were exposed for 3 h to tunicamycin or sorbitol, the glycerol content increased in ϳ200 and 500 g/mg of cells, dry weight, respectively. No significant difference was observed between wild-type and hac1⌬ mutant cells, confirming the lack of functional interactions between the HOG and UPR pathway. On the contrary, the increase in glycerol content was lower in hog1⌬ mutant cells exposed to either tunicamycin or sorbitol (Fig. 6A). Thus, ER stress induces the accumulation of glycerol and this event is dependent, at least in part, of the activity of the MAPK Hog1p. Next, we went on to test whether this response might be important to cope with ER stress. As shown in Fig. 6B, a strain lacking GPD1 was only slightly more sensitive to tunicamycin than the corresponding wild-type strain. However, this effect was more evident when gpd1⌬ cells were also disrupted in GPD2, which encodes the second isoenzyme of glycerol 3-phosphate dehydrogenase in S. cerevisiae (60). Nevertheless, gpd1⌬gpd2⌬ mutant cells were still more resistant to tunicamycin than the strain without Hog1p. On the other hand, deletion of GPD1 in the hog1⌬ background increased the growth defect observed in the single MAPK mutant (Fig. 6B). This result suggests that the accumulation of glycerol still observed in hog1⌬ cells (Fig. 6A) was prevented by additional knock-out of GPD1.
We tried to further confirm the functional role of glycerol under ER  . Glycerol overaccumulates in tunicamycin-treated cells and is required to provide survival to ER stress. A, cell cultures of the wild-type BY4741 strain (wt) and corresponding hog1⌬ and hac1⌬ mutant strains were grown in YPD at 30°C (A 600 ϭ 0.4 -0.6) and then transferred to the same medium containing 1 M sorbitol or 2 g/ml tunicamycin. After 3 h, aliquots (1 ml) of the cultures were withdrawn and processed for glycerol content. Values represent the difference in total glycerol before and after the indicated treatment and are the means of at least three independent experiments. dw, dry weight. B, growth in the presence of tunicamycin (Tn) was tested for yeast cells of the BY4741 background lacking the genes encoding NAD-dependent glycerol-3-phosphate dehydrogenase activity, gpd1⌬, gpd2⌬, and gpd1⌬gpd2⌬. Cells of the wild-type strain (wt), hog1⌬, and double hog1⌬gpd1⌬ mutant were also included in the experiment. A representative experiment is shown. C, the effects of tunicamycin and NaCl on yeast growth were examined in YEplac195 (empty plasmid, URA3) or YEpGPD1 transformants (GPD1) of the W303-1A wild-type strain and hog1⌬ mutant. Cells were pregrown, diluted, plated, and incubated as described in Fig. 1. A representative experiment is shown. stress. Thus, we examined the protective effect conferred by overexpression of GPD1. As shown in Fig. 6C, a high copy number of this gene increased the tolerance of either wild-type or hog1⌬ mutant cells to tunicamycin. In a similar manner, overexpression of GPD1 suppressed the growth defect of the mutant strain under mild osmotic stress provided by 0.2 M NaCl (Fig. 6C). However, the protection conferred by GPD1 was unable to rescue the sensitivity phenotype of the MAPK mutant at the highest drug doses tested, 0.25 g/ml (Fig. 6C), or under severe osmotic conditions (data not shown).

DISCUSSION
Yeast cells are extremely sensitive to ER stress, and consequently, their ability to establish proper protective responses is critical for cell viability. In this work, we have studied the need of a functional HOG pathway to cope with folding stress. Indeed, absence of the MAPK Hog1p leads to increased sensitivity to ER stress induced by chemical agents such as tunicamycin, a strong N-glycosylation inhibitor, or the reducing agent ␤-mercaptoethanol, which prevents disulfide bond formation. This effect was dependent on Hog1p activity because cells lacking the upstream MAPKK Pbs2p or expressing a kinase dead (HOG1-KD) allele were unable to grow when challenged with tunicamycin, whereas hyperactivation of Hog1p by disruption of the protein phosphatases PTP2 and PTP3 enhanced the resistance to the drug. Hog1p governs the tunicamycin-induced expression of GPD1 and the accumulation of the compatible osmolyte glycerol, which confers cell protection under these conditions. Hence, our results describe new aspects of the response of S. cerevisiae to ER stress and identify additional functions of the Hog1p MAPK to provide stress resistance.
ER stress in S. cerevisiae triggers different signals and activates a variety of responses that help to prevent cell death. Out of the UPR, cells respond to defects in folding or assembly of proteins in the ER by activating the cell wall integrity MAPK cascade (17,61,62). This involves the protein Pkc1 and the MAPK of the Mpk1p-Slt2p pathway, which, in turn, stimulates the Cch1p-Mid1p Ca 2ϩ channel and calcineurin signaling (61,63). Consistent with this, mutants of the cell wall integrity pathway display sensitivity to agents that induce folding stress (64). Moreover, cell wall stress activates UPR signaling, and loss of UPR function causes cell wall defects (62). Hence, ER and cell wall stress responses are coordinated by cell wall integrity and UPR signaling pathways to protect cells against these stressors (65).
Unlike this coordinated regulation, our results rule out an interaction between the UPR and the HOG pathways in activating or modulating each other in response to ER stress. Thus, deletion of HOG1 had no effect on the UPR activity based on ␤-galactosidase reporters of UPR activation, appearance of spliced HAC1 mRNA, or expression by Northern blot of UPRdependent target genes, like KAR2, an ER resident chaperone. Furthermore, double mutants lacking Ire1p and Hog1p do not show epistatic defects with respect to ER stress. Finally, our results showed no effect on the tunicamycin-induced accumulation of glycerol by lack of a functional UPR pathway. Hence, these pathways appear to control different targets and different mechanisms that allow yeast cells to confront with increased unfolded proteins.
Exposure of yeast cells to tunicamycin did not trigger the phosphorylation of Hog1p. Even in the absence of Ptp2p and Ptp3p, there was no major change in the phosphorylation level of the MAPK. Furthermore, we were unable to show nuclear import of Hog1p in tunicamycin-treated cells. Therefore, there is no apparent signaling through the HOG pathway in response to ER stress. This finding was, in some way, not surprising. Exposure of yeast cells to heat shock (23), metalloid salts (40), or zymolyase (66), results in the phosphorylation of Hog1p, but a nuclear accumulation of the MAPK could not be detected. Oxidative stress triggers a subtle phosphorylation of Hog1p in a narrow range of H 2 O 2 concentrations (67), whereas activation of the MAPK in response to methylglyoxal was found to proceed without phosphorylation (25). Thus, different stressors generate different signal inputs that determine the transient response of Hog1p, especially its phosphorylation and enrichment in the nucleus. On the other hand, it seems clear that certain activity of the MAPK is independent of these transient changes and that this catalytic activity, referred as basal, might be the responsible of the role of Hog1p in several cellular processes (25)(26)(27).
In fact, our current knowledge suggests that Hog1p could have some nuclear function under normal growth conditions, including that of structural adaptor of transcriptional factors or enzymatic modulator of the general transcription machinery (59,68,69). This would explain why cells carrying a nonphosphorylatable HOG1 allele (40), exhibited a higher tunicamycin tolerance than the hog1⌬ mutant strain. Overphosphorylation of Hog1p has been reported to be an essential event previous to the nuclear enrichment of the MAPK upon osmotic stress (29). Moreover, several observations suggest the existence of an intrinsic HOG signaling activity even in the absence of stress (70). Remarkably, such signaling activity would be controlled by the Sln1 branch of the pathway, whereas the Sho1 branch would be an inducible nonbasal system (70). This model could then explain why growth on tunicamycin was impaired in the ssk1⌬ mutant strain, whereas disruption of SHO1 had no effect. Hence, Hog1p-mediated effects under ER stress appear to rely on the basal activity of the MAPK.
But how does basal activity of Hog1p contribute to cell growth during ER stress? As we showed for the first time, unfolded protein stress triggers the accumulation of glycerol, and this event was partially dependent of Hog1p. Moreover, the MAPK controls the transcriptional induction of GPD1, the gene for glycerol synthesis (55). In complete agreement with this, a gpd1⌬gpd2⌬ double mutant, showed enhanced sensitivity to ER stress than the wild-type, whereas overexpression of GPD1 provided higher tolerance to both wild-type and hog1⌬ mutant cells. It is worth noting that like in the case of osmotic stress (55), some glycerol overproduction was detected in the hog1⌬ mutant strain. This explains why the phenotype of sensitivity to tunicamycin was more severe in the double hog1⌬gpd1⌬ mutant than in the single hog1⌬ strain. In addition, the strain lacking Hog1p showed a phenotype more pronounced than the gpd1⌬gpd2⌬ mutant, suggesting that the MAPK regulates important targets, other than glycerol, in response to ER stress.
Synthesis of glycerol in response to unfolded protein stress might also be favored by the cytosolic activity of the MAPK Hog1p. Thus, a previous report by Winkler et al. (23) suggested that heat stress-activated Hog1p may exert its effect by phosphorylating cytoplasmic proteins. Following this argument, Westfall et al. (38) have proposed that the main physiological role of Hog1p would be cytosolic. In this, the MAPK would establish the metabolic conditions for elevated glycerol production, the most important feature of the osmoprotection mechanism in S. cerevisiae. Consistent with this, stimulation of the HOG-MAPK pathway by increased osmolarity leads to an activation of 6-phosphofructo-2-kinase, which, in turn, results in the activation of the upper part of glycolysis and increased glycerol production (71). Furthermore, a recent chemical genetic screening has identified four novel substrates of Hog1p kinase activity, including the glycolytic enzyme Tdh3p (72), the yeast glyceraldehyde-3-phosphate dehydrogenase. Further work is, however, needed to demonstrate whether or not the proposed cytosolic role of the MAPK can be extended to other stress situations such as ER stress.
Finally, our results raised the question of why accumulation of unfolded ER proteins triggers the synthesis of glycerol. Osmolytes are often referred to as "chemical chaperones" since they can increase thermodynamic stability of folded proteins without perturbing other cellular processes (73). Indeed, it has been shown that osmolytes can provide significant stability to protein by hiding the backbone into the core of folded proteins (52,74,75). Accordingly, the synthesis of glycerol in cells exposed to tunicamycin has physiological significance as part of the survival mechanisms to unfolded protein stress and represents a new instance where S. cerevisiae generates this compatible polyol.