Response of Saccharomyces cerevisiae to severe osmotic stress: evidence for a novel activation mechanism of the HOG MAP kinase pathway


  • O. Van Wuytswinkel,

    1. Department of Biochemistry and Molecular Biology, Vrije Universiteit, De Boelelaan 1083, 1081 HV Amsterdam, The Netherlands.
    Search for more papers by this author
  • V. Reiser,

    1. Institute of Biochemistry and Molecular Cell Biology, and Ludwig Boltzmann-Forschungsstelle für Biochemie, University of Vienna, A-1030 Vienna, Austria.
    Search for more papers by this author
  • M. Siderius,

    1. Department of Biochemistry and Molecular Biology, Vrije Universiteit, De Boelelaan 1083, 1081 HV Amsterdam, The Netherlands.
    Search for more papers by this author
  • M. C. Kelders,

    1. Department of Biochemistry and Molecular Biology, Vrije Universiteit, De Boelelaan 1083, 1081 HV Amsterdam, The Netherlands.
    Search for more papers by this author
  • G. Ammerer,

    1. Institute of Biochemistry and Molecular Cell Biology, and Ludwig Boltzmann-Forschungsstelle für Biochemie, University of Vienna, A-1030 Vienna, Austria.
    Search for more papers by this author
  • H. Ruis,

    1. Institute of Biochemistry and Molecular Cell Biology, and Ludwig Boltzmann-Forschungsstelle für Biochemie, University of Vienna, A-1030 Vienna, Austria.
    Search for more papers by this author
  • W. H. Mager

    Corresponding author
    1. Department of Biochemistry and Molecular Biology, Vrije Universiteit, De Boelelaan 1083, 1081 HV Amsterdam, The Netherlands.
    • *For correspondence. E-mail; Tel. (31) 20 444 75 69; Fax (31) 20 444 75 53.

    Search for more papers by this author


The HOG/p38 MAP kinase route is an important stress-activated signal transduction pathway that is well conserved among eukaryotes. Here, we describe a novel mechanism of activation of the HOG pathway in budding yeast. This mechanism operates upon severe osmostress conditions (1.4 M NaCl) and is independent of the Sln1p and Sho1p osmosensors. The alternative input feeds into the HOG pathway MAPKK Pbs2p and requires activation of Pbs2p by phosphorylation. We show that, upon severe osmotic shock, Hog1p nuclear accumulation and phosphorylation is delayed compared with mild stress. Moreover, both events lost their transient pattern, presumably because of the absence of negative feedback mediated by Ptp2p tyrosine phosphatase, which we found to be localized in the nucleus. Under severe osmotic stress conditions, the delayed nuclear accumulation correlates with a delay in stress-responsive gene expression. Severe osmoshock leads to a situation in which active and nuclear-localized Hog1p is transiently unable to induce transcription of osmotic stress-responsive genes. It also appeared from our studies that the Sho1p osmosensor is less active under severe osmotic stress conditions, whereas the Sln1p/Ypd1p/Ssk1p sensor and signal transducer functions normally under these circumstances.


Adaptation to changes in the environment is of crucial importance for the survival of any living organism. These adaptation mechanisms can be divided into three phases: (i) perception of an external signal; (ii) intracellular transmission of this signal, leading to (iii) an adaptive response. A wide variety of signal transduction machineries have been characterized that enable yeast cells to trigger the appropriate responses to external stress stimuli (Hohmann and Mager, 1997).

When yeast cells are exposed to an increase in external osmolarity, immediate growth arrest occurs. Then, specific responses are triggered in order to repair molecular damage and induce adaptation to the new conditions, after which growth resumes. Hyperosmotic stress leads to altered transcription of stress-responsive genes and to an intracellular accumulation of glycerol, the yeast's osmolyte (Hohmann, 1997). These responses are mainly controlled by two signal transduction pathways: (i) the HOG MAP kinase pathway; and (ii) the operationally defined general stress response pathway respectively (Siderius and Mager, 1997).

The general stress response mediates the expression of a large number of genes in response to a variety of stress conditions (Hohmann and Mager, 1997). Upon challenge of yeast by increases in osmolarity, expression of general stress-responsive genes, such as the small heat shock protein gene HSP12 and the gene encoding catalase, CTT1, is induced. These general stress-responsive genes are characterized by the presence of STREs, general stress-responsive elements, in their promoters (Marchler et al., 1993; Moskvina et al., 1998). After stress exposure, two transcription factors, Msn2p and Msn4p, translocate from the cytoplasm to the nucleus and bind to the STRE sequences. Nuclear localization of the Msn2/4 proteins appears to be controlled by protein kinase A (PKA) activity (Martinez-Pastor et al., 1996; Gorner et al., 1998). PKA activity plays a dominant, yet elusive, role in the co-ordination of nutrient sensing, metabolism and growth in Saccharomyces cerevisiae and therefore probably plays a role in the integration of growth-related signal transduction and stress response (Siderius and Mager, 1997). Deletion of the MSN2 and MSN4 genes largely abolishes the transcriptional response under many stress conditions with one exception: hyperosmotic stress. Under osmoshock conditions, a residual increase in general stress-responsive mRNA levels is still apparent. This residual induction of stress-responsive gene expression is mainly caused by the activity of the HOG (high-osmolarity glycerol) MAP kinase pathway (Martinez-Pastor et al., 1996).

MAPK (mitogen-activated protein kinase) cascades play a central role in many signal transduction pathways in eukaryotic cells (Robinson and Cobb, 1997). They are involved in the regulation of the physiological responses in many organisms, such as plants, fungi, insects and mammals, in response to different kinds of stimuli, such as growth factors, mitogens, cytokines, plant hormones as well as environmental stress (Whitmarsh and Davis, 1998).

MAP kinase cascades are composed of a triplet of sequentially acting protein kinases, including a MAP kinase kinase kinase (MAPKKK), a MAP kinase kinase (MAPKK) and a MAP kinase responsible for the appropriate cellular response. MAP kinases are activated through a dual phosphorylation on serine/threonine and tyrosine residues (Robinson and Cobb, 1997; Banuett, 1998).

In the budding yeast S. cerevisiae, four complete MAPK pathways have been identified so far (reviewed by Herskowitz, 1995; Levin and Errede, 1995). The pheromone response pathway is involved in the preparation of the cells for mating (Leberer et al., 1997). The pseudohyphal/invasive growth pathway is triggered by environmental conditions leading to the formation of pseudohyphal cells (diploids) or to invasive growth (haploids) (Liu et al., 1993; Roberts and Fink, 1994). The PKC-regulated pathway responds to heat and hypo-osmotic stress (Davenport et al., 1995; Kamada et al., 1995). Finally, the HOG pathway responds to hyperosmotic stress (Brewster et al., 1993). A fifth MAPK pathway, not completely elucidated, regulates spore wall assembly (Krisak et al., 1994) and meiosis (Friesen et al., 1994).

The HOG MAPK pathway (see Fig. 1) was characterized by complementing osmosensitive mutants unable to accumulate intracellular glycerol levels upon an increase in extracellular osmolarity. Activation of the Hog1p MAPK depends on the activation of its MAPKK (Pbs2p). Pbs2p activity can be stimulated after a high-osmolarity challenge in two different ways. A putative osmosensing receptor Sho1p triggers Pbs2p activity through the stimulation of Ste11p, a MAPKKK also acting in the mating pheromone MAPK pathway in S. cerevisiae (Maeda et al., 1995; Posas and Saito, 1997). Recently, at least two other Ste proteins, Ste20p and Ste50p, have been shown to be implicated in the transfer of the signal from Sho1p to Ste11p, suggesting an elaborate interplay between the HOG pathway and other MAPK cascade components in yeast (O'Rourke and Herskowitz, 1998; Posas et al., 1998). In addition, a second putative osmosensor, the Sln1p histidine kinase receptor, feeds into a three-component system consisting of Ypd1p and Ssk1p as response regulators (Posas et al., 1996). Stimulation of the Sln1p kinase results in the activation of a pair of MAPKKKs, Ssk2p and Ssk22p, that can trigger Pbs2p activity. Phosphorylation of Hog1p by Pbs2p regulates its nuclear localization (Ferrigno et al., 1998; Reiser et al., 1999). It has been shown recently that intracellular distribution of Hog1p is regulated not only by its phosphorylation status but also by its binding to nuclear targets and by induction of its nuclear export mediated by its own kinase activity (Reiser et al., 1999). Once in the nucleus, Hog1p most probably exerts its further functions. Components of the HOG pathway have been studied extensively, and Hog1p function has been shown to be involved in diverse processes, such as increased transcription of genes, rearrangement of the actin cytoskeleton (Brewster and Gustin, 1994) and increased glycerol levels upon hyperosmotic shock (Hohmann, 1997). Yet, the actual molecular target(s) of the Hog1p MAPK have not been clearly identified. Among several studies, the bZIP repressor protein Sko1p has been shown to confer HOG-dependent osmotic regulation of expression to the ENA1 gene (Proft and Serrano, 1999). Another transcription factor, Hot1p, was recently described as a possible target of the HOG pathway (Rep et al., 1999a). In addition, the Msn2p and Msn4p general stress response factors were shown to play a role in the nuclear retention of Hog1p (Reiser et al., 1999). The HOG pathway is downregulated by the activity of two phosphatases, Ptp2p and Ptp3p, dephosphorylating Hog1p at Tyr-176 (Jacoby et al., 1997; Wurgler-Murphy et al., 1997). The Ptc1p phosphatase also participates in the downregulation of the HOG pathway (Maeda et al., 1993; Jiang et al., 1995).

Figure 1.

Model of the HOG pathway in S. cerevisiae. Two osmosensors, Sho1p and Sln1p, stimulate the HOG MAP kinase cascade by different mechanisms. Activation of this MAP kinase cascade leads to the phosphorylation of the Hog1p MAPK, which is consequently found in the nuclear compartment, where it exerts its function. The nuclear targets of Hog1p are not known, despite the description of several candidates (see text for more details). Hog1p may also have a cytoplasmic function.

In this work, the existence of a novel input into the HOG pathway in S. cerevisiae is described. Evidence for this third activating mechanism was obtained by using a sho1 ssk2 ssk22 triple mutant in the study of the HOG pathway activation under severe osmotic stress. This study also demonstrates that this alternative input into the HOG pathway feeds into Pbs2p. This result has prompted us to investigate the effect of severe osmotic shock on the mechanisms operating during the yeast response to such stress. A systematic study led us to several conclusions with respect to the functioning of the HOG pathway, especially regarding its osmosensor specificities and its role in the transcriptional control of osmostress-responsive genes.


Severe osmotic stress response indicates the existence of a novel HOG pathway input

In order to identify a putative differential role for the Sho1p- and Sln1p-mediated input into the HOG pathway, phosphorylation of Hog1p under different osmotic stress conditions was monitored in different HOG pathway mutants. The sho1 ssk2 ssk22 triple mutant displays a slightly less osmosensitive phenotype compared with the pbs2 or hog1 single mutants (Fig. 2). This may indicate the existence of an alternative mechanism leading to the activation of Hog1p, independently of the Sho1p and Sln1p branches of the HOG pathway.

Figure 2.

The osmosensitivity phenotype of S. cerevisiae HOG pathway mutants. Serial dilutions of several HOG pathway mutant strains (see Table 1) grown overnight in YPD medium were spotted on YPD plates supplemented with various concentrations of sorbitol. Cells were cultured for 3 days at 30°C. The TM141 wild-type strain was cultured and spotted as a growth control.

Osmotic stress leads to activation of the HOG pathway through a dual phosphorylation of the Thr-174 and Tyr-176 residues of the Hog1p MAPK by its dual specificity kinase Pbs2p. These phosphorylations are necessary and sufficient to allow translocation of Hog1p from the cytoplasm to the nucleus, most probably leading to the induction of expression of target genes (Ferrigno et al., 1998; Reiser et al., 1999). Phosphorylation of Hog1p can be monitored by Western blot analysis using a commercially available antibody specifically recognizing the phosphorylated forms of the p38-type kinase Hog1p Tyr-176 and Thr-174 residues.

Wild-type and sho1 ssk2 ssk22 mutant cells were osmostressed and then rapidly harvested for total protein extraction on a 60–240 min time scale. Total protein samples were separated on an SDS–PAGE gel, and Hog1p phosphorylation was analysed by Western blotting. As a gel loading control, the blots were stripped and incubated with an antibody raised against the C-terminus of Hog1p. It must be noticed that ECL detection (see Experimental procedures) was performed for all blots at the same time, allowing us directly to compare signals obtained for the different strains.

Under mild stress conditions (0.4 M NaCl), Hog1p dual phosphorylation is transient in a wild-type strain, appearing within 1 min and disappearing between 10 and 30 min after the onset of the stress (Fig. 3A). Dephosphorylation is supposed to result from the activity of the Ptp2/3p tyrosine phosphatases, which are activated by Hog1p activity (Wurgler-Murphy et al., 1997; Ferrigno et al., 1998). In the sho1 ssk2 ssk22 mutant, Hog1p phosphorylation is not detectable in response to these mild stress conditions.

Figure 3.

Evidence for an alternative input into the HOG pathway.

A. Hog1p MAPK phosphorylation is prolonged under severe osmoshock conditions and is still detected in a sho1 ssk2 ssk22 mutant. Wild-type and HOG pathway mutant strains were exposed to 0.4 M NaCl or 1.4 M NaCl in YPD medium for the time indicated. Cells were harvested, and total protein extracts were prepared as described in Experimental procedures. Protein samples (20 µg) were run on an SDS–PAGE gel and blotted on nitrocellulose for Western analysis. Hog1p Thr-174 and Tyr-176 phosphorylation status was measured using an anti-dually phosphorylated p38 antibody (α-phospho Hog1p). The blots were stripped, and the Hog1p loading control was measured using an anti-Hog1p C-ter (α-Hog1p). In order to compare the signal levels, the ECL detection was performed at the same time on all blots.

B. The alternative input into the HOG pathway is Pbs2p dependent. Another set of HOG pathway mutants was stressed with 1.4 M NaCl in YPD medium (sho1ssk2ssk22, ste11 ssk2 ssk22) or selective YNB medium (pbs2 and pbs2 + PBS2 S/A T/A). Total protein extraction and Western analysis was performed as in (A). The pbs2 + PBS2 S/A T/A strain is the result of the transformation of the pbs2 mutant by the pVR20 plasmid (see Table 1).

Under severe stress conditions (1.4 M NaCl), Hog1p dual phosphorylation starts to appear in a wild-type strain after 1 min, but is maintained over a very long period of time (3 h time scale), reaching a maximum between 30 and 45 min. When the sho1 ssk2 ssk22 triple mutant is subjected to this severe stress condition, phosphorylation of Hog1p is still manifest (Fig. 3A). Similar results were obtained using a sln1 sho1 ssk2 ssk22 mutant (result not shown). These data demonstrate that an alternative input, independent of the two known branches of the HOG pathway, is activated under severe osmotic stress conditions. Increased phosphorylation of Hog1p in the sho1 ssk2 ssk22 mutant was also characterized when the severe osmotic shock was performed using either KCl (1.4 M) or sorbitol (2.5 M) (data not shown), showing that this effect is osmospecific and is not caused by sodium toxicity.

In order to identify to which component of the HOG pathway the identified alternative branch is connected, Hog1p phosphorylation in response to 1.4 M NaCl was measured in the pbs2 mutant and in the ste11 ssk2 ssk22 triple mutant. Hog1p phosphorylation is still detectable in the ste11 ssk2 ssk22 mutant, as in the sho1 ssk2 ssk22 strain, but is completely absent in the pbs2 mutant (Fig. 3B). This result clearly shows that the alternative input feeds into the Pbs2p MAPKK. This may be consistent with the observation that the pbs2 mutant, just like the hog1 mutant, is more osmosensitive than the sho1 ssk2 ssk22 or ste11 ssk2 ssk22 mutants (Fig. 2). Under the same severe stress conditions, Hog1p phosphorylation was not detectable in the pbs2 mutant expressing a PBS2 non-phosphorylatable allele (Fig. 3C), indicating that the Pbs2p MAPKK needs to be activated by phosphorylation to give rise to Hog1p phosphorylation.

The observed Hog1p phosphorylation only occurred after the osmostress challenge. But as Pbs2p seems to play a central role in this process, it could not be excluded that the Hog1p phosphorylation observed in the sho1 ssk2 ssk22 mutant under severe osmostress is caused by Pbs2p basal kinase activity itself. In order to investigate this possibility, Hog1p phosphorylation in the sho1 ssk2 ssk22 mutant overexpressing PBS2 was examined in unstressed and stressed cells. No difference in Hog1p phosphorylation was observed in unstressed cells, and Hog1p was phosphorylated after exposure to 1.4 M NaCl in the strain overexpressing PBS2, as in the untransformed sho1 ssk2 ssk22 strain (data not shown). Based on the absence of Hog1p phosphorylation in unstressed sho1 ssk2 ssk22 strains overexpressing PBS2, we concluded that a mere increase in Pbs2p basal activity could not account for the observed Hog1p phosphorylation in the sho1 ssk2 ssk22 when exposed to severe osmotic stress.

In addition, it is interesting to notice that the signal of the Hog1p loading control detected on the wild-type blots seems to decrease when Hog1p is phosphorylated (Fig. 3A). This is not the case on the blots for the sho1 ssk2 ssk22 triple mutant, in which the signal is constant. This might reflect a difference in conformation (and hence a difference in protease sensitivity) between the normally and alternatively activated form of Hog1p or, rather, a change in protease or phosphatase activity in yeast cells exposed to severe osmostress. Further studies are needed to clarify this.

The phosphorylation status of Hog1p in response to mild and severe osmotic stress conditions was also measured in the sho1 mutant and the ssk2 ssk22 double mutant (Fig. 3A). The phosphorylation kinetics obtained using the sho1 mutant are similar to those obtained using the wild-type strain. However, in the ssk2 ssk22 double mutant, the mild stress conditions result in Hog1p phosphorylation kinetics similar to the wild type, whereas the severe osmoshock displays a response identical to that observed in the sho1 ssk2 ssk22 triple mutant. This result indicates that the Sho1p osmosensor is active under mild stress conditions, but is less responsive to severe osmostress.

It must also be noted that, under extreme stress conditions, Hog1p in both wild-type and mutant cells is not rapidly dephosphorylated.

Severe hyperosmotic stress results in a delayed induction of stress-responsive genes

In response to hyperosmotic shock, a wide set of genes, directly or indirectly involved in resistance to such stress, shows induced expression. This set of genes can be divided into different categories: genes under the control of the HOG pathway only or the general stress response pathway, or both pathways (Rep et al., 2000). In order to identify the effects of the prolonged phosphorylation of Hog1p observed under severe osmostress conditions, induction of expression of model genes was followed kinetically in response to osmotic shock of different strengths. Such a systematic study also allowed discrimination between early and late (i.e. after adaptation) responses elicited by the cells upon osmostress.

Wild-type yeast cultures in rich medium and at logarithmic growth phase were exposed to mild (0.4 M NaCl), moderate (0.8 M NaCl) and severe (1.4 M NaCl) osmotic stress. The growth rate of the stressed cultures was monitored by OD measurements, showing that the mildly stressed cells rapidly resumed growth. In contrast to low-stress conditions, the severely stressed cells did not resume growth during the experimental time scale of 5 h (data not shown). Induction of expression of the monitor genes HSP12 and/or GRE2 was followed by Northern analysis. HSP12 is mainly under the control of both the general stress response pathway and the HOG pathway. In contrast, in response to osmostress, GRE2 is only under the control of the HOG pathway (Rep et al., 2000). Mild stress leads to a rapid (within 15 min) and strong transcriptional induction of the tested stress genes. Under higher osmotic shock conditions, induction of transcription of the genes occurs later (visible 3 h after the 1.4 M NaCl stress) but to the same extent compared with the mild stress conditions (Fig. 4A). This delayed response is strain independent, concerns all the stress genes we have studied so far and is osmospecific, as osmostress performed using sorbitol or KCl gave rise to the same effect (data not shown). The observed delay in induction of transcription of the stress genes relative to the phosphorylation status of the Hog1p factor is notable and will be discussed later.

Figure 4.

Growth and gene expression response of S. cerevisiae to severe osmoshock conditions.

A. Overnight cultures of the wild-type (TM141) and mutant strains were diluted to OD660 = 0.1 and allowed to grow until an OD660 = 0.2 at 30°C. Subsequently, cells were exposed to an osmotic shock with 0.4 M or 1.4 M NaCl, and total RNA was isolated at the time points indicated. Actin, HSP12 and/or GRE2 mRNA levels were determined by Northern blotting.

B. Wild-type (TM141) and sho1 ssk2 ssk22 strains were used to analyse HSP12 and GRE2 mRNA levels in a Northern blot experiment as described in (A). Actin probe was used as a loading control. Samples were taken at later time points than in (A).

Induction of transcription of HSP12 and/or GRE2 was also followed in the different mutants used previously for the Hog1p phosphorylation study (Fig. 4A). It must be noticed that the sho1 mutant shows the same kinetic pattern compared with the wild type. In contrast, the sho1 ssk2 ssk22 triple mutant and the ssk2 ssk22 double mutant showed no induction of the GRE2 gene under the 1.4 M NaCl stress condition, despite the fact that Hog1p is phosphorylated through the putative alternative pathway under these conditions (Fig. 3A). On the other hand, transcriptional induction does occur in the ssk2 ssk22 double mutant under the 0.4 M NaCl stress condition. These data are consistent with our previous observation indicating that the Sho1p putative osmosensor is fully functional at 0.4 M NaCl, but not under severe osmoshock conditions.

The apparent absence of transcription of GRE2 and HSP12 in the triple mutant upon severe osmoshock on a 5 h time scale may reflect a further delay in transcriptional activation. In order to test this possibility, the experiment was repeated using the wild type and triple mutant, but this time samples were taken at later time points (Fig. 4B). It is clear that, indeed, transcriptional induction of HSP12 and GRE2 does occur in the triple mutant but at even later time points than observed in the wild-type strain challenged with severe osmostress. The levels of gene expression, however, are significantly lower than in wild-type cells. This is consistent with the inviability of the mutant cells after long exposure to severe osmotic stress, as also reflected by the strongly reduced actin mRNA levels.

Hog1p, phosphorylated upon severe osmotic stress, is enzymatically active

Hog1p is phosphorylated in sho1 ssk2 ssk22 mutant cells upon severe osmotic stress (see Fig. 3). Yet, no transcriptional induction of Hog1p target genes occurs, at least for 5 h after the onset of the stress (see Fig. 4A). This finding raises the question as to whether Hog1p phosphorylated under these circumstances is enzymatically active. Therefore, we measured Hog1p kinase activity using the in vitro kinase assay. Wild type and sho1 ssk2 ssk22 mutant were transformed with a haemagglutinin-tagged version of HOG1 (HOG1-HA), which appeared to be functional as it complemented the lethal effect of HOG1 deletion under the conditions of hyperosmotic stress (data not shown). Hog1p-HA was immunoprecipitated from cell extracts prepared from non-stressed or hyperosmotically (0.4 M or 1.4 M NaCl) stressed cells of wild-type or sho1 ssk2 ssk22 mutant strains and used for the kinase assay. No physiological target protein known to be phosphorylated by Hog1p has been found yet. Instead, we used myelin basic protein (MBP), which appears to be efficiently phosphorylated by Hog1p in the kinase assay (V. Reiser, unpublished results). The result of the kinase assay showed that Hog1p is activated by severe osmotic stress in both the wild type and the triple sho1 ssk2 ssk22 mutant, although less efficiently (Fig. 5). Hog1p kinase activity parallels the Hog1p phosphorylation profile under severe stress conditions. These data led us to conclude that the alternative mechanism of HOG pathway activation indeed produces enzymatically active Hog1p. Moreover, apparently the lack of transcriptional response in the sho1 ssk2 ssk22 mutant does not result from an absence of Hog1p kinase activity.

Figure 5.

Hog1p kinase activity under severe osmoshock conditions. Wild-type and sho1 ssk2 ssk22 strains transformed with the HOG1-HA plasmid (pVR53-U) were stressed with the indicated NaCl concentrations for the indicated times. The hog1 strain transformed with the HOG1 kinase-deficient plasmid (pVR53-K/R-L) was used as a control. Total protein extracts were prepared, and an immune complex kinase assay was performed as described in Experimental procedures, using MBP as a substrate. Immunoprecipitated extracts were analysed by Western blot with phospho-specific p38 MAP kinase antibody and anti-HA antibody.

Nuclear accumulation of Hog1p is perturbed upon severe osmotic stress

The delayed transcriptional response observed under severe osmotic stress conditions may reflect a transient defect in nucleo-cytoplasmic trafficking. Translocation of cytoplasmic factors from the cytoplasm to the nucleus is an important event during the response to osmotic stress, both for the HOG pathway and for the general stress response pathway (Ferrigno et al., 1998; Gorner et al., 1998; Reiser et al., 1999). As a kinetic discrepancy was observed between the Hog1p phosphorylation and induction of transcription of the stress genes upon severe osmotic shock, it is possible that, although phosphorylated, Hog1p is unable to enter the nucleus rapidly upon severe osmotic stress. This translocation problem could also affect the Msn2p general stress response pathway factor.

In order to test this hypothesis, nuclear translocation of Hog1p upon severe osmostress was monitored in vivo using a green fluorescent protein (GFP) fusion form of this protein expressed in a wild-type strain. Hog1p is translocated to the nucleus extremely rapidly (within 5 min) under mild 0.4 M NaCl stress conditions (Fig. 6A). Msn2p nuclear translocation occurs similarly rapidly (data not shown). Once in the nucleus, the GFP fluorescence remains in this compartment for about 10 min before re-entering the cytoplasm. Thus, dephosphorylation of Hog1p correlates with its exit from the nucleus, as reported previously (Ferrigno et al., 1998; Reiser et al., 1999). When cells expressing the GFP-tagged proteins were stressed with 1.4 M NaCl, two phenomena were observed. First, nuclear translocation of these factors takes place only around 45 min after application of the stress. Secondly, before translocation, the GFP fluorescence is detected in the cytoplasm, but forming ‘patches’. Return of the fluorescence back to the cytoplasm was not observed within the experimental time scale (Fig. 6A). It can be concluded that, upon severe osmotic shock, nuclear translocation of factors involved in the response is perturbed, but not abolished. A phosphorylated form of Hog1p is undoubtedly translocated to the nucleus, but with some delay. The kinetic discrepancy observed between nuclear translocation of Hog1p (45 min after the stress; Fig. 6A) and induction of transcription of GRE2 (taking place 3 h after the stress; Fig. 4A) seems to indicate that a transient block in nuclear translocation is not the only explanation for the delayed expression of the target genes upon severe osmoshock. Obviously, such stress conditions are leading to a transient state in which a nuclear and phosphorylated Hog1p is unable to play its signalling role in the stress response. As shown by the kinase activity assay performed on the wild-type strain (see Fig. 5), the prolonged Hog1p phosphorylation as well as the delayed expression of the HOG-controlled genes cannot be explained by a transient inhibition of Hog1p kinase activity upon severe osmostress. Inhibition of components involved in its dephosphorylation, such as phosphatases Ptp2p and Ptp3p (Wurgler-Murphy et al., 1997), must then be considered. This phenomenon could even be enhanced by a different compartmental localization of Hog1p and the phosphatases. In order to address this question, Ptp2p, a major phosphatase responsible for the tyrosine dephosphorylation of Hog1p, was fused to a GFP tag. Ptp2p-GFP turned out to be localized permanently in the nucleus, independently of the osmolarity of the extracellular environment (Fig. 6B). Thus, the delayed nuclear accumulation of Hog1p under severe stress (Fig. 6A) might prevent rapid activation of Ptp2p in the nucleus and, hence, might result in the increase in phosphorylated Hog1p in the cytoplasm.

Figure 6.

Nucleo-cytoplasmic distribution of Hog1p after mild or severe osmoshock.

A. Severe osmostress (1.4 M NaCl) induces a delayed nuclear accumulation of Hog1p. A wild-type (W303-1A) strain was transformed with the pVR65 (HOG1-GFP) plasmid. The logarithmically growing strain was examined by fluorescence under iso-osmotic (standard) or hyperosmotic growth conditions (0.4 M NaCl or 1.4 M NaCl) at the indicated times after stress. As the DAPI staining conditions used preferentially stain the mitochondrial DNA, nuclei are indicated by an arrow.

B. Nuclear localization of the Ptp2p phosphatase. A wild-type strain was transformed with the PTP2-GFP plasmid. Conditions were the same as in (A).

Hog1p dephosphorylation coincides with activation of the expression of the HOG pathway target genes

Severe osmotic stress leads to a temporary situation in which a nuclear, phosphorylated Hog1p is unable to induce transcription of the stress-responsive genes and cannot be dephosphorylated. The inhibition of dephosphorylation of Hog1p, upon severe osmotic stress, may be caused by different factors and is temporary. In order to determine whether this situation can be reversed, cells severely stressed during a period of time necessary to obtain a nuclear and phosphorylated Hog1p, were downshifted to a lower osmolarity. A wild-type cell culture was stressed for 1 h with 1.4 M NaCl and then downshocked to 0.4 M NaCl. Cells were collected for RNA extraction after the downshock on a 90 min time scale. As a control, cells continuously stressed at 1.4 M NaCl were used. Interestingly, a rapid reactivation of transcription of the stress genes was observed upon the downshock to mild or moderate stress conditions (Fig. 7A). The cells kept under severe stress conditions exhibited a (slight) increase in GRE2 expression starting at about 210 min after exposure to 1.4 M NaCl. This set of experiments indicates that the inhibition of gene expression can be readily overcome by lowering the stress conditions. Moreover, when the Hog1p phosphorylation status was monitored during a similar severe prestress/downshock experiment, it was evident that prolonged phosphorylation of Hog1p upon severe osmotic stress was rapidly disappearing as a result of the downshock (Fig. 7B). This suggests that Hog1p dephosphorylation by the Ptp2/3p phosphatases coincides with the induction of transcription of the genes controlled by the HOG pathway. In addition, the cellular localization of Hog1p was checked upon a similar severe prestress/downshock experiment. Hog1p is nuclear after the 1.4 M NaCl prestress and is still in the nucleus 15 min after the downshock (Fig. 7C), although it is dephosphorylated as shown in Fig. 7B. This indicates that Hog1p dephosphorylation is not the only factor controlling the return of Hog1p to the cytoplasm.

Figure 7.

Block in HOG pathway signalling after a severe osmoshock is overcome upon a downshock to decreased stress conditions. Wild-type (TM141) cells were prestressed with 1.4 M NaCl for 1 h and then downshocked to 0.4 M NaCl stress conditions as described in Experimental procedures. Cells were then harvested at the indicated times in order to isolate RNA for Northern analysis (A) or to isolate total protein extracts for Western analysis (B). A GRE2 probe was used for the Northern analysis, using actin as a loading control. Hog1p phosphorylation status was measured as described previously for the Western analysis on a gel loaded with 20 µg protein samples. Nucleo-cytoplasmic localization of Hog1p was followed on a wild-type strain (W303-1A) transformed with the HOG1-GFP plasmid (pVR65, see Table 1) under the same prestress/downshock conditions (C).


Evidence for an osmostress-induced alternative input into the HOG pathway

The occurrence of an alternative input into the HOG pathway, independent of its two known branches, has been suggested previously (O'Rourke and Herskowitz, 1998). This possibility is also demonstrated by the slight differences in osmosensitivity and cell phenotype observed between the hog1 and pbs2 single mutants and other combinations of mutations interrupting both the Sho1p- and Sln1p-dependent branches of the HOG pathway (Fig. 2). Moreover, the existence of alternative activation mechanisms of MAP kinase pathways, independent of their MAPKKK activity, was shown previously in Schizosaccharomyces pombe (Shieh et al., 1998; Shiozaki et al., 1998).

In this study, we show clear evidence for the existence of this alternative input in the response to osmotic stress. Indeed, activation of the phosphorylation of the Hog1p MAPK was observed in the sho1 ssk2 ssk22 mutant in response to a severe osmotic stress. This activation requires the Pbs2p MAPKK, more precisely its phosphorylatable form, but no other element of the HOG pathway described so far (Fig. 3).

The fact that the alternative pathway requires a phosphorylatable form of Pbs2p seems to indicate the presence of an upstream MAPKKK in this pathway. We showed that the Ste11p and Ssk2p/Ssk22p MAPKKK are not involved in the alternative pathway activation. The only remaining MAPKKK identified in yeast is Bck1p, the MAPKKK of the hypo-osmotic stress-induced PKC pathway. The possibility that Bck1p is one of the upstream elements of the alternative pathway is currently under investigation in our laboratory.

The level of phosphorylated Hog1p, as measured under severe stress conditions in the sho1 ssk2 ssk22 triple mutant, is lower compared with the wild type. Yet, the alternative pathway-dependent Hog1p phosphorylation, evidenced in the same triple mutant, leads to a transcriptional induction of HOG-dependent genes (such as GRE2 or HSP12; see Fig. 4B). This induction is weaker and occurs even later than that observed in wild-type cells challenged with severe osmostress. Under our conditions, Hog1p phosphorylated by the alternative pathway has a detectable kinase activity (see Fig. 5), which should be able to stimulate expression of the HOG pathway-dependent genes. These results suggest that HOG pathway signalling is defective under increased severity of stress. We cannot exclude the possibility that the alternative pathway, activated by severe osmotic stress, controls the expression of a separate set of genes. It has been shown previously that the expression of certain proteins (of unknown function) in S. cerevisiae was specifically induced under severe (1.4 M NaCl) salt stress conditions (Norbeck and Blomberg, 1996). It is not known as yet whether Hog1p activated by the alternative pathway accumulates in the nucleus or if it is localized to the cytoplasm. As Hog1p was shown to play a role in actin cytoskeleton rearrangements (Brewster and Gustin, 1994), our results might also reflect a cytoplasmic function for Hog1p, controlled by the alternative pathway.

Severe osmostress conditions lead to a prolonged Hog1p phosphorylation and to a retarded transcriptional response

The results discussed above prompted us to perform a study of the response of yeast cells to extreme osmostress. The major effect of a severe osmotic shock, apart from a growth defect, is a retarded signal transduction reflected by a delayed transcriptional response. This was emphasized by (i) the delayed Hog1p (and Msn2p) nuclear translocation (see Fig. 6A); (ii) the prolonged Hog1p phosphorylation (see Fig. 3A); and (iii) the delayed transcriptional induction of genes controlled by the HOG pathway (and general stress response pathway; see Fig. 4). This effect on the kinetics of induction of the stress-responsive genes is surprising because an earlier and stronger adaptive response would be expected to occur in severely stressed cells compared with mildly stressed cells. The delayed response is particularly surprising for genes such as GPD1 (data not shown; Rep et al., 1999b), involved in the production of the yeast osmolyte glycerol.

It was shown recently that phosphorylation of Hog1p at its dual Tyr-174/Thr-176 site is necessary and sufficient to allow its nuclear localization. Once in the nucleus, Hog1p kinase activity probably allows the phosphorylation of transcription factors and of the Ptp2/3p tyrosine phosphatases, which serves to inactivate Hog1p itself (Ferrigno et al., 1998; Reiser et al., 1999). In response to severe osmotic shock, we observed, a temporal delay between Hog1p phosphorylation, its nuclear translocation and, finally, activation of the transcription of a HOG-dependent gene. It seems that, under these conditions, each step of the HOG response is slowed down, nuclear translocation of Hog1p (and Msn2p) taking place 45 min after application of the stress (see Fig. 6A). We observed an apparent aggregation of Hog1p in the cytoplasmic compartment shortly after severe osmotic stress. Co-localization of Hog1p with actin patches has previously been suggested to occur (Ferrigno et al., 1998), and actin cytoskeleton appears to play a role in the mammalian stress-activated protein kinase pathway (Alberts and Treisman, 1998) or at least in the nuclear translocation apparatus (Smiths and Raikhel, 1998). The delayed Hog1p nuclear translocation observed upon severe osmostress might be linked to a modification in the cytoskeleton structure under these conditions. It is possible that – actin microfilaments being disassembled upon severe osmotic stress – translocation of Hog1p could just take place when the cytoskeleton is reformed.

Upon severe osmostress, nuclear Hog1p stays phosphorylated for a long period of time. This effect could be caused by several possible mechanisms: (i) the respective Ptp2/3p phosphatases are inactivated by the severe osmostress conditions or cannot recognize the phosphorylated form of Hog1p under these conditions; (ii) as the kinase activity of Hog1p is supposed to be required for its own tyrosine dephosphorylation (Wurgler-Murphy et al., 1997; Ferrigno et al., 1998) or for the modulation of its cellular distribution (Reiser et al., 1999), severe osmoshock might inhibit the Hog1p kinase activity. We have shown that prolonged phosphorylation of nuclear Hog1p is not caused by a difference in cellular localization between Hog1p and the Ptp2p phosphatase, as they are both present in the nucleus after 45 min of severe osmoshock (see Fig. 6A and B). Interestingly, the same prolonged phosphorylation occurs with the (KM)HOG1 allele, a mutant lacking kinase activity (Ferrigno et al., 1998). As the inhibition of Hog1p kinase activity by severe stress might explain both the prolonged Hog1p phosphorylation and the delayed induction of transcription observed under severe stress conditions, assuming that these two events are directly linked, we measured Hog1p kinase activity in these conditions. We showed that Hog1p kinase activity is not inhibited under severe stress conditions and follows the Hog1p phosphorylation pattern (see Fig. 5). In conclusion, our data seem to indicate that prolonged phosphorylation of Hog1p upon severe osmotic stress is linked to a dephosphorylation defect, either because the Ptp2p phosphatase is inactive or because the severe stress form of Hog1p is non-dephosphorylatable. In addition, the delayed induction of expression of the HOG-dependent genes upon severe osmotic stress probably results from a signalling defect downstream of Hog1p.

The prolonged phosphorylation status of Hog1p upon severe osmotic stress could also be linked to a specific function of Hog1p under these conditions. It was shown in mammalian cells that the kinetics of phosphorylation of the MAP kinases modulate specific responses. For example, a prolonged phosphorylation of the ERK2 MAPK in PC12 cells resulted in an adaptive response leading to differentiation of these cells in sympathetic neurons (Marshall, 1995). On the other hand, a transient phosphorylation of ERK2 leads to a proliferation signal. However, in such cells, the prolonged phosphorylation of ERK2 is required for its nuclear localization. This is not the case with Hog1p. Once phosphorylated (transiently or for a long period), Hog1p is translocated into the nucleus at a certain time.

Moreover, when the severely stressed cells were downshocked to a lower NaCl concentration, Hog1p dephosphorylation and reactivation of transcription occurred rapidly (see Fig. 7). Thus, it seems that the downshock overcomes the Hog1p dephosphorylation defect observed upon severe osmoshock as well as the signalling defect occurring downstream of Hog1p under the same conditions. As the response to a downshock in yeast is regulated via the PKC-dependent MAPK pathway (Davenport et al., 1995), it is possible that our results reflect a cross-talk between the HOG pathway and the PKC pathway. A further indication for this may be that the final MAPK of the PKC pathway, Mpk1p, is dephosphorylated by the Ptp2/3 phosphatases as well. Our results could also suggest that Hog1p dephosphorylation is the trigger for activation of the downstream elements of the HOG pathway as, in the prestress experiment, nuclear Hog1p dephosphorylation coincided with the induction of expression of the target genes, whereas Hog1p was still localized to the nucleus. This model could be considered as contradicting the data suggesting that Hog1p dephosphorylation regulates its back-translocation to the cytoplasm. However, it was shown in S. pombe that the nuclear accumulation of Sty1p, the S. pombe Hog1p homologue, is maintained in this compartment in part by the tethering of Sty1p to its target, Atf1p (Gaits et al., 1998). Nuclear retention of Hog1p by nuclear targets was also shown to be a molecular mechanism regulating its cellular distribution (Reiser et al., 1999). If this mechanism is involved in the regulation of the Hog1p-driven response, a dephosphorylated form of Hog1p could be maintained in the nucleus for a sufficient time to allow induction of the transcriptional response.

The Sho1p putative osmosensor is sensitive to severe osmostress

Another finding from this study is that, although the Sln1p osmosensor functions normally under severe osmotic stress conditions, the Sho1p osmosensor does not. This can be concluded from Figs 3A and 4A. In the Δssk2Δssk22 double mutant, Hog1p phosphorylation is induced at 0.4 M NaCl stress and is as strong as in the wild type. In contrast, in the same mutant but stressed with 1.4 M NaCl, Hog1p phosphorylation is only induced by the alternative pathway, as the signal is as strong as in the sho1 ssk2 ssk22 triple mutant stressed under the same conditions (see Fig. 3A). This correlates with the corresponding expression of stress-responsive genes (see Fig. 4A). The existence of two independent osmosensors leading to phosphorylation of Hog1p could be a reflection of a difference in sensitivity of these two sensors. Apparently, the Sln1p osmosensor is able to cope with a wide range of stress conditions. This is not the case for Sho1p.

Experimental procedures

Strains and growth conditions

Yeast strains used in this study are listed in Table 1. Yeasts were cultured in YP medium supplemented with glucose (2%) or YNB supplemented with glucose as well as the appropriate amino acids. Standard liquid culturing conditions were at 30°C on a rotational shaker (220 r.p.m.). Analysis of growth on solid medium was assayed by spotting 5 µl of serially diluted (1:10, starting from OD660 = 0.1) samples in YNB.

Table 1. Yeast strains and plasmids used.
Strain or plasmidGenotype
 TM141MATa leu2 ura3 trp1 his3
 TM233MATa leu2 ura3 his3 hog1::TRP1
 TM261MATa ura3 trp1 his3 pbs2::LEU2
 TM284MATa leu2 ura3 trp1 his3 sho1::TRP1
 TM295MATa leu2 ura3 trp1 his3 ssk2::URA3 ssk22::LEU2
 TM310MATa leu2 ura3 trp1 his3 sho1::TRP1 ssk2::URA3 ssk22::LEU2
 TM310 U > KMATa leu2 ura3 trp1 his3 sho1::TRP1 ssk2::KAN ssk22::LEU2
 TM295 U > KMATa leu2 ura3 trp1 his3 ssk2::KAN ssk22::LEU2
 KK311MATa leu2 ura3 trp1 his3 ssk2::KAN ssk22::LEU2 ste11::TRP1
 pVR20 TRP1 CEN4 ARS1 PBS2 S/A, T/A (PBS2 with S-514A and T-518A mutations)
 pVR53-K/R-LLEU2 CEN4 ARS1 HOG1K/R-HA (HOG1 with K-52R mutation)


The plasmids pVR65-WT (HOG1-GFP) and pVR20 (PBS2 with S-514A, T-518A substitutions) have been described by Reiser et al., 1999.

Plasmids pVR53-L and pVR53-U (a wild-type HOG1 gene fused to a HA-tag expressed from its native promoter from a low-copy plasmid YCp111 or YCp33 respectively) was generated by cloning a cassette encoding three tandem repeats of HA-tag flanked with Not1 restriction site into pVR50-NotI (Reiser et al., 1999). Plasmid pVR53-K/R-L (K-52 to R replacement) was constructed by replacing a 1.1 kb SalI–SalI fragment of pVR53-L with the corresponding fragments of the mutant genes described by Schüller et al. (1994).

To construct the GFP fusion with the PTP2 gene, the PTP2 coding sequence was amplified from a chromosomal DNA preparation of the wild-type W303 strain using TATTTTAACCATGGATCGCATAGCACAGC and ATTTTACACCATGGTAC-AAGGTAACGCGTTCTT oligonucleotides (NcoI sites underlined). The polymerase chain reaction (PCR) fragment was cloned into the NcoI site of plasmid ADH1::(NcoI)GFP (Gorner et al., 1998) using the NcoI restriction site.

High-osmolarity stress experiments and RNA isolation

Cells cultured overnight were diluted to an OD660 of 0.1, grown until the early exponential phase (OD660 of 0.2) and then subjected to high-osmolarity stress by the addition of a concentrated solution of NaCl, KCl or sorbitol. The resulting concentrations of external NaCl or sorbitol are indicated for each experiment. The severe prestress experiments were performed as follows. Cells were stressed with 1.4 M NaCl under the same conditions as described above. After 1 h at 1.4 M NaCl, cells were briefly centrifuged, and the cell pellet was resuspended in the same initial volume of YP medium containing 0.4 M NaCl. Resuspension of the cells in this medium forms the first time point of the downshock.

Samples were taken at the time points indicated in the figures, centrifuged briefly, and the resulting cell pellets were frozen in liquid nitrogen for storage at −80°C. After thawing the samples in cold water, RNA was isolated as described previously (Siderius et al., 1997).

Separation of RNA was performed as described by Varela et al. (1995). DNA probes were labelled using the random priming method (Prime-a-Gene labelling system; Promega). Signal quantification was performed by phosphorimaging (Phosphorimager 425; Molecular Dynamics).

Western blotting and antibody staining

Dual phosphorylation of Hog1p on Thr-174 and Tyr-176 was examined in a Western analysis using an anti-dually phosphorylated p38 antibody (New England Biolabs) that cross-reacts with its homologue in S. cerevisiae, Hog1p, when phosphorylated by Pbs2p (O'Rourke and Herskowitz, 1998; Reiser et al., 1999). Total protein samples were basically isolated as described by Siderius et al. (1997) and Davenport et al. (1995) with one exception. Cells (10 ml of culture; OD660 = 0.2–0.3) were pelleted at the indicated time points, supernatant was poured off and cells were dropped in liquid nitrogen using the remaining medium. Frozen drops of cells were then broken using a bead beater (2 min at 2000 r.p.m.), after which the resulting cell powder was taken up in lysis buffer containing the phosphatase inhibitors (α-complete mix; Boehringer Mannheim). After centrifugation (5 min at 14.000 r.p.m. in an Eppendorf centrifuge), the protein concentration of each extract was determined (Bio-Rad protein assay). Proteins (total protein quantity as indicated in the figures) were separated on 10% polyacrylamide gels and blotted onto nitrocellulose membrane. Dually phosphorylated Hog1p was detected using the anti-p38 antibody. Anti-Hog1p antibody was purchased from Santa Cruz (C-terminal anti-Hog1p). Antibody binding was visualized using ECL (Amersham) after binding of a horseradish peroxidase-conjugated secondary antibody.

Stripping of the blots was performed at 50°C for 30 min in the following buffer: 100 mM β-mercaptoethanol, 62.5 mM Tris-HCl, pH 6.7, 2% SDS.

Fluorescence microscopy

Cells were grown to logarithmic phase (OD600 ≈ 0.8–1.0) in synthetic complete medium omitting components used as selective markers. DAPI (final concentration 2 µg ml−1) was added as DNA dye 15 min before microscopy. The proportion of cells with nuclear signal accumulation was determined by counting cells with distinctly enhanced nuclear fluorescence. The number of cells with accumulated nuclear signal was expressed relative to the sum of cells counted showing fluorescence (total number of cells examined > 300) after growth under iso-osmotic conditions or after exposure to the indicated NaCl concentrations for the indicated times. For time course experiments, cells were scored in a similar manner using the camera pictures of samples taken at the time points indicated. Living cells were viewed with a Zeiss Axioplan 2 fluorescence microscope, and images were scanned with a Quantix charge-coupled device camera (Photometrics) using iplab software (Signal Analytics). Pictures were assembled with Adobe photoshop 4.0 software.

Immune complex kinase assay

Cells were grown in appropriate synthetic medium to OD600 ≈ 0.8–1.0. When they were hyperosmotically stressed, NaCl (final concentration 0.4 M or 1.4 M) was added at room temperature for the time period indicated. To prepare protein extracts, cultures were rapidly cooled down in ice-water and harvested by centrifugation. Cells were suspended in ice-cold lysis buffer [0.1 M MES, 10 mM EGTA, 1% dimethyl sulphoxide (DMSO), 1 mM dithiothreitol (DTT), pH 6.5] containing Complete protease inhibitor mix (Boehringer) and phosphatase inhibitors (0.1 mM sodium vanadate, 10 mM NaF). The cell pellet was then resuspended in ice-cold lysis buffer and vortexed with glass beads for 3 × 5 min. Samples were centrifuged at 14 000 r.p.m. for 2 × 10 min. All operations were performed at 4°C. Protein concentration was determined in the supernatants using a Bio-Rad protein assay kit. The crude extracts (≈ 500 µg) were used to immunoprecipitate Hog1p-HA with anti-HA antibody for 2 h and then incubated with GammaBind G Sepharose beads (Amersham Pharmacia) for 1 h. Beads were washed five times with the lysis buffer, twice with the kinase buffer (50 mM Tris-HCl, pH 7.5, 10 mM MgCl2, 1 mM DTT, 0.1 mM sodium vanadate and 10 mM NaF) and used for the kinase reaction (30 min at 30°C) after the addition of 0.5 µl of [γ-32-P]-ATP (3000 Ci mmol−1; Amersham Pharmacia) and 3 µl (1 mg ml−1) of MBP (Sigma). After the immunoprecipitation, an aliquot was analysed by Western blot analysis with phosphospecific p38 MAP kinase antibody and anti-HA antibody as described above. Probes for immunodetection were removed by incubation in stripping buffer according to the recommendations of the suppliers before applying another primary antibody to the same membrane.