Correspondence: Saskia M. van der Vies, Department of Biochemistry and Molecular Biology, Faculty of Science, Vrije Universiteit, de Boelelaan 1083, 1081 HV Amsterdam, The Netherlands. Tel.: +31 20 5987548; fax: +31 20 5987555; e-mail: firstname.lastname@example.org
Exposure of Saccharomyces cerevisiae to high osmotic stress evokes a number of adaptive changes that are necessary for its survival. These adaptive responses are mediated via multiple mitogen-activated protein kinase pathways, of which the high-osmolarity glycerol (HOG) pathway has been studied most extensively. Yeast strains that bear the hsp82T22I or hsp82G81S mutant alleles are osmosensitive. Interestingly, the osmosensitive phenotype is not due to inappropriate functioning of the HOG pathway, as Hog1p phosphorylation and downstream responses including glycerol accumulation are not affected. Rather, the hsp82 mutants display features that are characteristic for cell-wall mutants, i.e. resistance to Zymolyase and sensitivity to Calcofluor White. The osmosensitivity of the hsp82T22I or hsp82G81S strains is suppressed by over-expression of the Hsp90 co-chaperone Cdc37p but not by other co-chaperones. Hsp90 is shown to be required for proper adaptation to high osmolarity via a novel signal transduction pathway that operates parallel to the HOG pathway and requires Cdc37p.
When exposed to increased osmolarity, yeast cells counteract the detrimental effects of this stress condition by eliciting a range of initial adaptive responses such as loss of turgor, cell shrinkage, disintegration of a polarized cytoskeleton and eventually growth arrest. Subsequent adaptation and survival requires cellular events such as gene expression, accumulation of the osmolyte glycerol, adjustment of cell-wall architecture and resumption of the cell cycle to occur in a well-orchestrated manner (Hohmann, 2002). These adaptive responses are mediated via multiple mitogen-activated protein (MAP) kinase pathways, of which the high-osmolarity glycerol (HOG) pathway has been studied most extensively (Fig. 1a, de Nadal et al., 2002); the sterile vegetative growth (SVG) pathway has also been suggested to be required for osmoadaptation (Lee & Elion, 1999; Alonso-Monge et al., 2001; Jung et al., 2002; Wojda et al., 2003). These signal transduction pathways contain two of the five MAP kinase modules that have been identified in vegetative Saccharomyces cerevisiae cells (Fig. 1a). MAP kinase modules are composed of three different kinases (MAPKKK, MAPKK and MAPK), some of which can function in different MAP kinase modules (Fig. 1a, Gustin et al., 1998). Remarkably little is known about the mechanism(s) that ensure specificity of signal transduction through the different MAP kinase pathways. One mechanism seems to be by insulating the pathways using scaffold proteins (Posas & Saito, 1997; Elion, 2001) or by using different receptors or upstream signalling proteins (Gustin et al., 1998), possibly in combination with different transcriptional regulators that mediate the transcription of specific genes (de Nadal et al., 2002; Hohmann, 2002; Zeitlinger et al., 2003). The activity of the individual MAP kinases might also be regulated in order to generate the required signalling. In addition to the MAP kinases, other types of kinases, such as Cdc28p and Swe1p, appear to play a role in osmoadaptation (Imai & Yahara, 2000; Mort-Bontemps-Soret et al., 2002; Tatebe & Shiozaki, 2003). Several kinases and transcription factors that function in cellular signalling require the ubiquitous molecular chaperone Hsp90 for their folding and activation (Csermely et al., 1998; Buchner, 1999). Hsp90 (Hsc82p and Hsp82p in S. cerevisiae) does not usually act alone but functions in association with co-chaperones such as Hop, p23 and p50 (Sti1p, Sba1p and Cdc37p in S. cerevisiae, respectively) that influence the interaction(s) with nucleotides and client proteins. The functional characterization of co-chaperones is still at an early stage, but some co-chaperones such as Aha1p appear to stimulate the Hsp90 ATPase activity, whereas others such as Hop (Sti1p) and p50 (Cdc37p) inhibit this activity (Chang et al., 1997; Liu et al., 1999). The co-chaperone p50 is required for the recruitment of protein kinases, such as p60v-src, Cdc28p and Raf1 to Hsp90 (Dey et al., 1996; Abbas-Terki et al., 2000; Farrell & Morgan, 2000; Bandhakavi et al., 2003), whereas p23 stimulates the release of substrates from the Hsp90 chaperone complex. The Hsp82 chaperone of S. cerevisiae and the co-chaperone Cdc37p have been shown to be required for the activation of Ste11p (yeast homologue of Raf1), a MAPKKK that is shared by several MAPK modules, including that in the HOG pathway (Fig. 1a, Louvion et al., 1998; Abbas-Terki et al., 2000). The observed interaction(s) of Hsp90, Cdc37p and Ste11p may be indicative of a mechanism that is used to attain specificity in MAPK signalling (Lee et al., 2004). This observation, plus the fact that other Hsp90 clients (Cdc28p, Swe1p) appear to play a role in osmoadaptation (Imai & Yahara, 2000; Mort-Bontemps-Soret et al., 2002; Tatebe & Shiozaki, 2003), prompted us to investigate the role of the Hsp90 molecular chaperone in the high osmotic stress response in S. cerevisae.
Materials and methods
Yeast strains, growth conditions and plasmids
The Saccharomyces cerevisiae strains that were used are all isogenic to W303-IA (Table 1). Yeast cells were either grown in yeast peptone medium plus glucose (2%) or yeast nitrogen base supplemented with glucose and appropriate amino acids at 24°C. Phenotype testing was performed by spotting 5 μL of 10-fold serial dilutions of exponentially growing yeast cells onto yeast peptone dextrose (YPD) agar (2%) plates with additions of Calcofluor White (CFW) or sorbitol. Yeast cells were transformed by standard technical procedures. Plasmids with the Slt1p-LacZ integration construct (YIP358R, de Nadal et al., 2003) and CNS1, SSF1 and HCH1 in pTV3 (Nathan et al., 1999) were kindly provided by Francesc Posas (University Pompeu Fabra, Barcelona, Spain) and Susan Lindquist (Whitehead Institute, Cambridge, MA), respectively.
Table 1. Yeast strains used in this study
W303 1a (wild-type)
MATa leu2-3/112 ura3-1 trp1-1 his3-11/15 ade2-1 can 1-100 GAL SUC mal0
Screen for multicopy suppressors of hsp82ts mutants
The hsp82T22I mutant was transformed with a multicopy (pFl44L) genomic library using the LiAc method (Gietz et al., 1992). Selection was carried out on YPD plates containing 1.6 M sorbitol. Amplification of plasmids was carried out in the DH5αEscherichia coli strain. CDC37 was subcloned to Yeplac195 as a KpnI fragment together with a part of the adjacent TAF25 gene after verification that TAF25 did not suppress the osmosensitive phenotype of the hsp82ts strains.
Hog1p phosphorylation was determined as described previously (de Nobel et al., 2000). Samples (20 μg of soluble protein fraction) were analysed by western blot analysis using an anti-phospho-p38 antibody (New England Biolabs, Beverly, MA) and anti-C-terminal-Hog1p antibody (Yc20, Santa Cruz Biotechnology, Santa Cruz, CA). Antibody binding was visualized with the electrochemiluminescence (ECL) detection system (Pierce Biotechnology, Rockford, IL).
STL1-LacZ integration and β-galactosidase activity assay
The STL1-LacZ reporter gene was integrated into the genome of the strains indicated, as described in de Nadal et al. (2003). The single-copy integrations were confirmed by Southern hybridization. Cells were inoculated from an overnight culture to an OD600 of 0.2 and continued to grow to an OD660 of 0.6. Samples were taken at the indicated time points before and after adding sorbitol to a final concentration of 1 M. Cell extracts were made as described under western blotting analysis. The β-galactosidase activity of the cell extracts was measured following standard procedures.
Measurement of intracellular glycerol concentration
Cells were grown in YPD medium to an OD660 of 0.5–0.6. Sorbitol was added to a final concentration of 1 M. Samples of 7 mL were taken at different time points and the cells were collected by centrifugation, washed in 1 mL of 0.1 M phosphate buffer (pH 5.9), re-suspended in 1 mL of 0.5 M Tris/HCl (pH 7.0) and lysed by boiling for 10 min. The amount of glycerol in the soluble protein fraction was determined using a Boehringer Mannheim kit (no.0148270; Mannheim, Germany).
Zymolyase sensitivity test
Yeast cells grown in YPD medium at 30°C overnight were diluted to an OD600 of 0.025. Aliquots of 200 μL were transferred to a flat-bottomed 96-well Greiner PS-microplate and different concentrations of Zymolyase 100 T (Seikagaku, Tokyo, Japan) were added. Plates were incubated at 30°C for 15 h without shaking and the OD was determined at 595 nm.
Certain hsp82ts mutant have lost the ability to adapt to high osmolarity
Nathan & Lindquist (1995) isolated a set of temperature-sensitive (ts) mutant strains that carry single point mutations in different regions of the HSP82 gene. Interestingly, some – but not all – of these mutants displayed additional phenotypes such as the inability to respond to pheromone (Louvion et al., 1998), allowing investigation into the role of the different Hsp82 domains (Louvion et al., 1998; Buchner, 1999). We analysed the osmosensitivity of the hsp82ts mutants by determining their ability to grow in the presence of 1.8 M sorbitol at a permissive temperature. Of the eight mutants analysed (Table 1), only hsp82T22I and hsp82G81S displayed an osmosensitive phenotype (Fig. 1b, and data not shown). Transformation of the hsp82T22I and hsp82G81S strains with a single-copy plasmid carrying the HSP82 gene yielded a strain with the ability to adapt to high osmolarity, comparable with the wild type (Fig. 1c, and data not shown). These results indicate that the Hsp82 molecular chaperone might be involved in mediating adaptive responses to high osmolarity in Saccharomyces cerevisiae, or might be required for the proper acquisition of functionality by a component of the osmoresponsive signalling pathway, or even for functionality of the targets of osmoadaptive signalling.
Effect of hsp82ts mutations on HOG pathway functionality
The Hsp82 T22I and G81S mutations that give the strain its osmosensitivity are present in the N-terminal ATPase domain of Hsp82p and, as mentioned above, may change the ability of the chaperone to activate a component of the osmoresponsive signalling pathway, such as the MAPKKK Ste11p, and hence admit signalling via the HOG pathway (Fig. 1a). Activation of the HOG pathway involves the phosphorylation of the Hog1p MAP kinase protein. Analysis showed that the kinetics of Hog1p phosphorylation in the hsp82G81S and hsp82T22I mutants was similar to that in the wild type (Fig. 2, and data not shown). However, even in the absence of a non-functional Ste11p, Hog1p might become phosphorylated by the Pbs2p MAPKK, which in turn becomes phosphorylated, not by Ste11p but by the MAPKKK Ssk2/Ssk22p (Fig. 1a). When both SSK2 and SSK22 were deleted in the hsp82G81S mutant strain the kinetics of Hog1p phosphorylation were not affected upon exposure to high osmolarity (Fig. 2a). Although this double-mutant strain did display an increased osmosensitivity compared with the single hsp82G81S mutant strain (data not shown), no significant difference in Hog1p-dependent gene expression was observed compared to the wild type, using the STL1-LacZ reporter gene assay (Fig. 2b). Hence, the observed osmosensitivity cannot simply be explained as the result of a defective Hog1p activation. The accumulation of intracellular glycerol to combat the loss of turgor is an important part of the adaptive responses to high osmolarity. Accumulation of glycerol is achieved in two ways: first by closure of the plasma membrane glycerol facilitator protein Fps1, to prevent the efflux of glycerol (Tamás et al., 1999), and second by enhanced expression of glycerol-3-phosphate dehydrogenase (Gpd1p), which catalyses the rate-limiting step of glycerol synthesis (Albertyn et al., 1994). When intracellular glycerol levels were measured in wild-type and hsp82ts mutant cells upon exposure to increased osmolarity, similar levels were observed (Fig. 2c). By contrast, as anticipated, the accumulation of intracellular glycerol was seriously impaired in the osmosensitive hog1 deletion strain. These results suggest that the proteins involved in high-osmolarity-induced accumulation of glycerol do not require the function of Hps82p for their folding or functioning. In addition, the seemingly lower level of unphosphorylated Hog1p in the hsp82G81S mutant strain (Fig. 2a) did not have an effect on the regulation of downstream transcription factors and the accumulation of glycerol.
The osmosensitive hsp82 strains display aberrant cell-wall properties
The results detailed above indicated that the Hsp82-mediated response to high osmolarity occurs independent of the HOG pathway, and hence other aspects of osmotic adaptation had to be considered. Under normal growth conditions both the cellular integrity [protein kinase C (PKC)] and the SVG MAP kinase pathways control the architecture of the yeast cell wall (Lee & Elion, 1999; Jung et al., 2002). However, under osmotic stress conditions the PKC pathway is inhibited and hence components of the SVG pathways are potential candidates for response and adaptation (de Nobel et al., 2000). Recently, the Ste11p kinase, a client protein of Hps90, has been implemented as a crucial component of the SVG pathway (Fig. 1a, Alonso-Monge et al., 2001; Wojda et al., 2003). Therefore, Hsp90 might influence osmoadaptation by regulating cell-wall integrity through activation of components of the SVG pathway. An indication for cell-wall phenotypes of the hsp82ts mutant strains came from the observation that the hsp82ts strains regained the ability to grow when sorbitol was added to the growth medium. Although the extent to which this phenotype was observed varied, it was possible to choose one concentration of sorbitol at which all the strains displayed this phenotype (Fig. 3a). This so-called osmoremedial effect is often indicative of altered cell-wall properties underlying the temperature sensitivity. Cell-wall mutant phenotypes are commonly analysed by determining the resistance to cell-wall-perturbing agents such as CFW and the β1-3 glucanase cocktail Zymolyase (Garcia-Rodriguez et al., 2000; Martin et al., 2000; Alonso-Monge et al., 2001; Kapteyn et al., 2001). Examination of the resistance of all of the hsp82ts mutant strains to Zymolyase revealed a sensitivity of the hsp82G313S mutant, whereas the other mutants, including the osmosensitive mutants hsp82T22I and hsp82G81S, displayed a similar resistance to this cell-wall-perturbing agent as the wild-type strain (data not shown). However, as shown in (Fig. 3b), the hsp82ts mutant strains displayed different levels of sensitivity to CFW. The osmosensitive hsp82ts mutants were found to be less sensitive to CFW than the wild type, but more sensitive than the hog1 deletion strain (Fig. 3b). The other hsp82ts strains again displayed a variable pattern of resistance (A41V, A587T) and sensitivity (G170D, G313S) to CFW, indicating that the hsp82 mutants indeed display cell-wall phenotypes.
Genetic relation between the hsp82ts alleles and the hog1 deletion
The hsp82T22I and hsp82G81S mutants displayed both osmosensitivity and cell-wall phenotypes distinct from those of HOG pathway mutants, suggesting that Hsp82 might be involved in osmoadaptation in a manner independent of the HOG pathway. To investigate this hypothesis, the hsp82ts alleles were combined with a deletion of the HOG1 gene, and the growth of these mutant strains in the presence of sorbitol was examined. As can be seen in Fig. 4(a, left panel), the hsp82T22IΔhog1 and hsp82G81SΔhog1 double mutants displayed increased osmosensitivity compared with the single-mutantΔhog1 strain. When compared with the single hsp82ts mutants (Fig. 1b), osmosensitivity of the double mutants was observed at a significantly lower sorbitol concentration (0.4 vs. 1.8 M). In addition, reduced sensitivity of the double mutants to CFW as compared with the single hsp82ts strains was observed (Fig. 4a, right panel). Therefore, combining the hsp82T22I and hsp82G81S alleles with a deletion of the HOG1 gene greatly reduced the hypersensitivity of the hog1 deletion strain and increased the sensitivity of the hsp82T22I and hsp82G81S mutants to Zymolyase (Fig. 4b). These ‘intermediate’ phenotypes indicate that Hsp82 functions in a pathway that operates parallel to the HOG signalling route.
Overexpression of CDC37 suppresses the osmosensitivity of hsp82 mutants
In order to identify additional proteins that are involved in the Hsp82-dependent response to high osmolarity, a genetic screen was conducted. In addition to the wild-type HSP82 and HSC82 alleles, the gene encoding the co-chaperone Cdc37p was isolated as a multicopy suppressor of the osmosensitive phenotype of both the hsp82T22I and the hsp82G81S mutants (Fig. 5a). Overexpression of other co-chaperone genes such as CNS1, SSF1 and HCH1, which had been identified as suppressors of the temperature-sensitive hsp82E381K mutant (Nathan et al., 1999), did not result in suppression of the osmosensitive phenotype (Fig. 5a). Additional support for the unique role of Cdc37p in osmoadaptation was derived from the observation that a cdc37ts mutant (cdc37-34, Fliss et al., 1997) displayed an osmosensitive phenotype similar to that observed in the hsp82T22I and hsp82G81S mutants (Fig. 5b). Deletion of other co-chaperone genes such as SBA1, STI1 and AHA1 did not render the yeast cells osmosensitive (data not shown). Because, in addition, overexpression of CDC37 in HOG pathway mutants did not result in improved osmotolerance (data not shown), the role of Cdc37p in osmoadaptation appears to be coupled to Hsp82 in a unique manner and is independent of the HOG pathway.
The Hsp90 molecular chaperone was originally identified in mammalian cells as a component of a regulatory protein complex that controls the activation of signalling proteins. Saccharomyces cerevisiae contains two homologues of Hsp90, encoded by HSP82 and HSC82. Both proteins are able to substitute Hsp90 functionality in the activation of heterologously expressed mammalian p60v-Src or the steroid hormone receptor (Picard et al., 1990; Xu & Lindquist, 1993). In addition, mammalian Hsp90 suppresses the lethal phenotype of the hsp82hsc82 deletion strain (Piper et al., 2003), suggesting that the Hsp90 system in yeast functions in a manner resembling that of the higher eukaryotic chaperones. One of the first identified client proteins of Hsp82 in yeast, the MAPKKK kinase Ste11p, was shown to require this molecular chaperone for activation in response to mating pheromone (Louvion et al., 1998). Ste11p is a remarkable MAPKKK in S. cerevisiae as it is a component of multiple MAP kinase pathways (Fig. 1a, Gustin et al., 1998; Alonso-Monge et al., 2001). Here we report that two hsp82ts mutants display an osmosensitive phenotype. Both mutations are located in the ATPase domain of Hsp82. The hsp82T22I allele encodes an Hps82 mutant protein with a constitutive high ATPase activity, which may cause premature release of the client protein from the Hsp90 complex (Prodromou et al., 2000). The properties of the hsp82G81S mutant protein have not been characterized, but one might expect enhancement of ATPase activity. Mutations in the C-terminal dimerization and middle domain, to which Cdc37p (Silverstein et al., 1998; Grammatikakis et al., 1999) and co-factors such as Aha1p, respectively, bind (Panaretou et al., 2002; Lotz et al., 2003), did not result in osmosensitive phenotypes. However, the interaction of Cdc37 with Hsp90 might be more complicated. A mutation in the N-terminus of Cdc37p, well outside the binding domain region, nevertheless produced a c. 50% decrease in Hsp90 binding (Shao et al., 2003). Cdc37p has been reported to suppress the ATP turnover by Hsp90, thereby keeping Hsp90 in a ‘relaxed’ ADP-bound conformation that is required for the loading of client proteins. We isolated the gene encoding the co-chaperone Cdc37p as a multicopy suppressor of the Hps90 osmosensitive mutants. This result might indicate that Cdc37p suppresses this phenotype by inhibiting the high ATPase activity of the hsp82G81S mutant protein and hence by allowing proper binding of client proteins. However, other phenotypic traits of the hsp82T22I and hsp82G81S mutants, such as the lack of response to mating pheromone and the ts phenotype, are not suppressed by overexpression of Cdc37p (our unpublished results), indicating that the suppression by Cdc37p is specific. In addition, overexpression of other genes, known to suppress hsp82 phenotypes, HCH1, CNS1, SSF1 (suppressors of the ts phenotype of the hsp82G381K mutant strain, Nathan et al., 1999) or SBA1 (a co-chaperone that facilitates substrate release), could not suppress the osmosensitivity of the hsp82T22I and hsp82G81S mutant. Furthermore, overexpression of the co-chaperones STI1 (Hsp70–Hsp90 operating protein) or AHA1 (stimulator of Hsp90 ATPase activity) did not suppress the osmosensitivity. Together, our data underline the specificity of Cdc37p to suppress the osmosensitivity of the hsp82T22I and hsp82G81S mutant strains.
In mammalian cells, Cdc37p functions tightly with Hsp90 to control MAP kinase signalling (Miyata et al., 2001). It was recently reported to be involved in osmoadaptive responses in Schizosaccharomyces pombe, possibly by stabilizing the Hog1p homologue Spc1p (Tatebe & Shiozaki, 2003). However, there was no indication for an involvement of Hsp90 in osmoadaptive responses in Sch. pombe. This suggests that the involvement of the Hsp90 chaperone system in osmoadaptive responses is different in these two eukaryotic model systems.
Here we have described that in the osmosensitive hsp82T22I and hsp82G81S strains, the integrity of the HOG pathway is not affected, as judged by: (1) Hog1p activation after high osmotic challenge, (2) Hog1p-regulated gene expression and (3) high-osmolarity-induced intracellular accumulation of glycerol, suggesting that Hsp90 mediates osmoadaptation in a different manner. The hsp82ts mutants displayed osmosensitivity when challenged with relatively high sorbitol concentrations. However, the double hsp82ts hog1 strains displayed aggravated osmosensitivity, even compared with the hog1 deletion strain. In addition, the mutants displayed altered cell-wall properties, as compared to a hog1 strain, indicating that the Hsp82 functions in an osmoadaptive pathway that operates parallel to the HOG pathway. Osmoadaptation of yeast appears to be dependent on cell-wall properties that can be influenced by overexpression of genes affecting the cell wall (LRE1, HLR1 and WSC3), or by growth at elevated temperature (Siderius et al., 2000; Alonso-Monge et al., 2001; Wojda et al., 2003). The aforementioned genes did not suppress the osmosensitive phenotype of the hsp82T22I and hsp82G81S mutant strains. It is noteworthy that overexpression of LRE1, which is known to induce signalling through the cell integrity MAPK pathway (Sekiya-Kawasaki et al., 2002), slightly improved tolerance of the hsp82T22I strain to growth at 37°C (our unpublished observation). This may be indicative for a role of the Hsp90 chaperone system in regulating the osmoadaptive mechanisms by mediating changes in cell-wall composition. In addition, the observed effects of Zymolyase and CFW on the hsp82T22I and hsp82G81S mutant strains is viewed as an indication that osmoadaptation may occur via a mechanism involving maintenance of cell-wall integrity (Levin, 2005). The Hsp90/Cdc37p chaperone system may exert the regulatory role mentioned above via the SVG or cell integrity MAP kinase signalling pathways and may therefore be considered as a potential component ensuring MAP kinase signalling specificity. Characterization of client proteins implicated in the osmoregulatory role of Hsp90 and Cdc37p will be a next step in elucidating the novel in vivo functionality of these molecular chaperones in the osmotic stress responsiveness of eukaryotic cells.
We thank F. Posas (University Pompeu Fabra, Barcelona, Spain), S. Lindquist (Whitehead Institute, Cambridge, MA), P. Piper (University of Sheffield, Sheffield, UK) and A. Caplan (Mount Sinai School of Medicine, New York) for sending plasmids and yeast strains.