Mitogen-activated protein kinase (MAPK) cascades are frequently used signal transduction mechanisms in eukaryotes. Of the five MAPK cascades in Saccharomyces cerevisiae, the high-osmolarity glycerol response (HOG) pathway functions to sense and respond to hypertonic stress. We utilized a partial loss-of-function mutant in the HOG pathway, pbs2-3, in a high-copy suppressor screen to identify proteins that modulate growth on high-osmolarity media. Three high-copy suppressors of pbs2-3 osmosensitivity were identified: MSG5, CAK1, and TRX1. Msg5p is a dual-specificity phosphatase that was previously demonstrated to dephosphorylate MAPKs in yeast. Deletions of the putative MAPK targets of Msg5p revealed that kss1Δ could suppress the osmosensitivity of pbs2-3. Kss1p is phosphorylated in response to hyperosmotic shock in a pbs2-3 strain, but not in a wild-type strain nor in a pbs2-3 strain overexpressing MSG5. Both TEC1 and FRE::lacZ expressions are activated in strains lacking a functional HOG pathway during osmotic stress in a filamentation/invasion-pathway-dependent manner. Additionally, the cellular projections formed by a pbs2-3 mutant on high osmolarity are absent in strains lacking KSS1 or STE7. These data suggest that the loss of filamentation/invasion pathway repression contributes to the HOG mutant phenotype.
YEAST cells, like many other eukaryotes, utilizemitogen-activated protein kinase (MAPK) cascades to transmit signals from plasma-membrane-associated sensory complexes to the nucleus, where transcriptional responses are elicited (for review see Banuett 1998; Gustinet al. 1998). MAPK cascades are composed of three conserved families of protein kinases: the MAPK, the MAPK and ERK kinase (MEK), and the MEK kinase (MEKK; for review see Davis 1993). The MEKK receives a signal from an upstream pathway component and phosphorylates a conserved threonine and serine residue within the MEK's activation domain (Kyriakiset al. 1992; Lange-Carteret al. 1993). Once phosphorylated, the activated MEK phosphorylates a threonine and a tyrosine residue within the activation domain in the MAPK (Crews and Erikson 1992). The phosphorylated MAPK is then able to phosphorylate the targets of the MAPK cascade, including transcription factors (Gilleet al. 1992; Sethet al. 1992) and other regulatory proteins (Cooket al. 1996).
The budding yeast Saccharomyces cerevisiae contains five MAPK cascades, each with its own unique MAPK: Fus3p in the pheromone response pathway, Kss1p in the filamentation/invasion pathway, Hog1p in the high-osmolarity-growth (HOG) pathway, Slt2p in the cell integrity pathway (Figure 1), and Smk1p in the spore wall assembly pathway. Although these pathways each have a separate activating signal and physiological response (for review see Gustinet al. 1998), it is becoming increasingly clear that the MAPK pathways interact with each other. For example, the HOG pathway MAPK Hog1p is rapidly dephosphorylated in response to decreases in extracellular osmolarity in a Slt2p-dependent manner (Davenportet al. 1995), suggesting that the HOG pathway is negatively regulated by the cell integrity pathway. The phosphorylation of the pheromone response pathway MAPK Fus3p in response to hypertonic stress is enhanced by the deletion of HOG1 or the HOG pathway MEK gene PBS2 (Hallet al. 1996). The observation that the pheromone response pathway is activated in response to hypertonic stress in the absence of Hog1p is further supported by evidence that a transcriptional target of the pheromone response pathway, FUS1, is also induced by osmotic stress in the absence of catalytically active Hog1p (Hallet al. 1996; O'Rourke and Herskowitz 1998). It has been suggested that these data indicate that Hog1p prevents the inappropriate activation of the pheromone response pathway by osmotic stress.
These cross-pathway interactions may provide a mechanism for establishing and maintaining signaling specificity. The question is whether these interactions are physiologically significant or merely introduced by the genetic manipulations of the organism. For example, the activation of Fus3p by high osmolarity in the absence of Hog1p (Hallet al. 1996) may indicate an important role for Hog1p in the maintenance of signal specificity, or it may be an artifact of the experimental overexpression of FUS3. Other examples, such as the activity of Fus3p in the repression of the filamentation/invasion pathway and the maintenance of pheromone response pathway signal specificity (Madhaniet al. 1997), seem to indicate that some interactions are physiologically significant. However, in general, the prevalence and physiological significance of these cross-pathway interactions are not yet well known.
The HOG pathway mutants hog1 and pbs2 were first isolated in a screen for yeast unable to grow or produce glycerol in high-osmolarity media (Brewsteret al. 1993). Additionally, budding and growth defects have also been described for these mutants (Brewster and Gustin 1994). In high-osmolarity media, hog1Δ or pbs2Δ cells will abandon a small bud and grow a new bud. This double-budded phenotype may indicate a defect in cell cycle regulation in HOG pathway mutants that are exposed to high-osmolarity media (Brewster and Gustin 1994). A second growth defect observed in HOG pathway mutants grown in high-osmolarity media is the production of long cellular projections (Brewster and Gustin 1994), indicating stimulated or unregulated polarized growth in HOG mutants exposed to high osmolarity. However, the precise cause of these morphological defects is not known.
In this article, we provide evidence that part of the HOG pathway mutant phenotype is a consequence of the loss of HOG-pathway-dependent inhibition of a second MAPK pathway, the filamentation/invasion pathway. Kss1p is shown here to be phosphorylated in response to hyperosmotic shock in a HOG pathway MEK mutant, pbs2-3. Prevention of Kss1p phosphorylation by the overexpression of the MAPK phosphatase MSG5, the deletion of KSS1, or the deletion of upstream filamentation/invasion pathway MAPK cascade genes is sufficient to suppress not only the osmosensitivity of pbs2-3, but also one of the morphological phenotypes associated with HOG pathway mutants on high-osmolarity media: the formation of long projections. These data indicate that the filamentation/invasion pathway is inappropriately activated by hyperosmotic stress in cells lacking a functional HOG pathway.
MATERIALS AND METHODS
Strains, media, and general methods: The yeast strains and plasmids used in this study are listed in Table 1. The bacterial strain DH5α was used for all plasmid amplifications and isolations. Growth media (YEPD, supplemented SD, and LB) were prepared as described (Kaiseret al. 1994). Common procedures (DNA manipulations, bacterial propagation, etc.) were followed as described (Sambrooket al. 1989). Techniques used for genetic crosses, sporulation, dissection, and propagation of S. cerevisiae are described elsewhere (Kaiseret al. 1994). Yeast were transformed by the one-step method (Chenet al. 1992).
Isolation, cloning, and sequencing of pbs2-3: Wild-type strain MG159B was mutagenized with ethyl methanesulfonate (EMS) and screened for colonies that failed to grow on YEPD supplemented with 900 mm NaCl or 1.5 m sorbitol and that showed a reduced production of glycerol in high-osmolarity media. Four complementation groups of recessive mutations were identified and designated hog1–hog4 (Brewsteret al. 1993). The mutants in the hog4 complementation group (alleles of PBS2; Brewsteret al. 1993) were then screened for the ability to grow at an intermediate osmolarity of 400 mm NaCl (YEPD plus 400 mm NaCl). Only one mutant, hog4-3 (renamed pbs2-3), was able to grow at this osmolarity. Thus, the phenotype of pbs2-3 was intermediate between a pbs2Δ strain and the wild-type strain. pbs2-3 was backcrossed to W303-1A four times to produce strain KDY1. Genomic DNA was extracted (Hoffman and Winston 1987) from strain KDY1 and used as a template to amplify the PBS2 locus by PCR. PCR products from three independent reactions were cloned using the TA cloning kit (Invitrogen, San Diego) to produce pKD10, pKD11, and pKD12. One clone, pKD10, was sequenced (Seqwrite) and compared to the Saccharomyces Genome Database and previously published sequences (Boguslawski and Polazzi 1987). Discrepancies between the pbs2-3 sequence and published sequences of PBS2 were verified by sequencing portions of pKD11 and pKD12 to ensure that the mutations identified in pKD10 were not introduced by PCR. Two mutations were identified, which are predicted to code for the following substitutions in the polypeptide sequence: proline for serine at position 168 (S168P) and aspartate for glycine at position 509 (G509D). Restriction fragments from pbs2-3 containing either one or both of the identified mutations were used to replace the corresponding restriction fragments of a plasmid containing PBS2 to produce pKD13, pKD14, and pKD15. A strain deleted for PBS2 (KDY9) was transformed with these plasmids and assayed for growth on high-osmolarity media (YEPD plus 400 or 900 mm NaCl) to identify which of the mutations were responsible for the pbs2-3 growth phenotype.
High-copy suppressor screen: A pbs2-3 strain (KDY1) was transformed with two high-copy plasmid yeast genomic DNA libraries (M. F. Hoekstra, unpublished results; C. J. Connelly and P. Hieter, unpublished results). The transformants (~30,000 per library) were screened for the ability to grow on YEPD supplemented with either 1 m KCl or 1 m sorbitol. Plasmid DNA was isolated from osmoresistant colonies, amplified in bacteria, and transformed into KDY1 to test for suppression of pbs2-3 osmosensitivity. Twelve colonies that demonstrated plasmid-dependent osmotic resistance were isolated. Plasmids were isolated and divided into five classes by restriction mapping. The largest class (five clones) consisted of plasmids containing PBS2. Two plasmids contained HOG1. The remaining three classes of high-copy suppressors contained MSG5 (two clones), CAK1 (two clones), or TRX1 (one clone), as determined by the comparison of the plasmid DNA sequence (Seqwrite) to the Saccharomyces Genome Database.
Plasmids: pKD2, pKD3, and pKD5–pKD7 were constructed by first amplifying the appropriate open reading frame (TRX2, TRX1, or CAK1) and 1-kb flanking DNA from wild-type (W303-1A) genomic DNA using primers designed to add XhoI sites to the ends of the PCR product. The PCR products were digested with XhoI and ligated to SalI-digested pRS426 (2μ URA3) or pRS423 (2μ HIS3; Christiansonet al. 1992). Plasmids pKD1 and pKD4 were constructed by subcloning an ~2.2-kb XhoI-NotI fragment of the library isolate containing MSG5 to XhoI-NotI-digested pRS426 or pRS423, respectively.
Primers designed to add an XhoI site on both ends of the resulting PCR product were used to amplify a fragment containing the KSS1 open reading frame and 500 bp of upstream- and downstream-flanking DNA from wild-type (W303-1A) genomic DNA. An XhoI-digested PCR fragment containing the KSS1-coding sequence and flanking DNA was ligated to XhoI-digested pGEM-7ZF(+) (Promega, Madison, WI) to produce pKD8. pKD8 was used as a template to amplify the 500-bp regions flanking the KSS1 open reading frame and the vector, but not the KSS1-coding sequence, using primers that add a NotI site to the ends of the resulting PCR product. Primers containing NotI sites were also used to amplify the URA3 gene from YDp-U (Berbenet al. 1991). These two PCR products were digested with NotI and ligated to produce pKD9. pKD9 was used as a template to amplify the URA3 and flanking DNA by PCR, and the product was used to replace the genomic copy of KSS1 in yeast. The deletion of KSS1 was confirmed in uracil prototrophs by PCR and by sterility in combination with fus3Δ.
The plasmids used in the disruption of STE11, STE7, and STE12 were a gift from George Sprague. pYEE98 (pUC119 fus3-6::LEU2) was a gift from the Elion lab. Lee Bardwell provided YEpT-KSS1 and YEpT-kss1. YEp352-CAK1, provided by Ed Winter, was used to further test whether overexpressed CAK1 can suppress pbs2-3 osmosensitivity.
Growth analysis: Overnight cultures of the desired strains were diluted to OD600 ~0.1 in YEPD and grown for 2–3 hr at 30°. Cell densities were again adjusted to OD600 = 0.1 in YEPD and used to make serial dilutions in a 96-well plate. Equal volumes of these cultures were transferred to YEPD plates containing various solutes by use of a 48-prong replicating tool. Plates were incubated at 30° for various lengths of time, as indicated in the figure legends.
Immunoblots: Protein extracts were prepared from treated and untreated cells, as described previously (Brewsteret al. 1993). A total of 20 μg of each protein sample was separated by SDS-PAGE and transferred to Protran nitrocellulose filters (Schleicher & Schuell, Keene, NH). Molecular weight markers (10 kD; GIBCO BRL, Gaithersburg, MD) were visualized by staining the membrane with Ponceau S (Sigma, St. Louis) and marking the location with a pencil. The membrane was rinsed and incubated with blocking buffer [3% BSA in TBS-T (6.056 g/liter Tris, 8.766 g/liter NaCl, 0.05% Tween 20, pH 8.0)]. Monoclonal antibodies [antiphosphotyrosine (Upstate Biotechnology, Lake Placid, NY) or anti-phosphoERK (Sigma)] were diluted 1:1000 in blocking buffer, added to the membrane, and incubated at room temperature for at least 90 min. Following incubation, the membranes were washed three times for 10 min in TBS-T. Secondary antibody (HRP-conjugated anti-mouse; Boehringer Mannheim, Indianapolis) was diluted 1:5000 in blocking buffer and incubated with the membranes for at least 40 min. The membranes were then washed four times for 10 min in TBS-T prior to the addition of ECL reagent (Amersham, Arlington Heights, IL) and exposure to Hyperfilm ECL (Amersham).
Northern analysis: Cells were grown overnight, diluted to OD600 ~0.4 in YEPD, and grown for an additional 2 hr prior to the addition of solute (NaCl, KCl, or sorbitol) to produce a hyperosmotic shock. Cells were pelleted and then lysed by vortexing with glass beads in the presence of phenol and RNA lysis buffer (0.5 m NaCl, 10 mm EDTA, 50 mm Tris, pH 8.0). The RNA in the aqueous phase was then precipitated, dried briefly, and then resuspended in 50 μl diethyl pyrocarbonate-treated water. A total of 2 μl 10× NBC (0.5 m boric acid, 10 mm sodium citrate, 50 mm NaOH, pH 7.5), 3 μl formaldehyde, 10 μl formamide, 2 μl 10× loading buffer (15% Ficoll, 0.1 m EDTA, 0.25% bromophenol blue), and 1 μl ethidium bromide (1 mg/ml) were added to 4 μg RNA. Following electrophoretic separation, the RNA was transferred to a Hybond N+ membrane (Amersham), UV cross-linked, and incubated for at least 1 hr at 65° with Church buffer (7% SDS, 250 mm Na2PO4, pH 7.5; Shifman and Stein 1995). Radiolabeled probes were prepared from gel-purified PCR products using a randomprimed labeling kit (Ambion, Austin, TX). The membrane and probe were then incubated overnight at 65°, washed, dried briefly, and exposed to a phosphoimager plate. The phosphoimaging plate was then processed using a Fujix BAS100 phosphoimager.
β-Galactosidase assays: Cells were grown to saturation overnight, diluted to OD600 ~0.4 in YEPD, and grown for an additional 2 hr prior to the addition of solutes for the desired stress. After the desired time had elapsed, cells were pelleted, transferred in 500 μl of ice-cold Z buffer (16.1 g/liter Na2 HPO4·7H2O, 5.5 g/liter NaH2PO4·H2O, 0.75 g/liter KCl, 0.246 g/liter MgSO4·7H2O, pH 7.0) to a 2-ml tube, pelleted, and frozen on dry ice. To isolate the protein, 250 μl Z buffer and 12.5 μl 40 mm PMSF were added to the pellet prior to four freeze/thaw cycles (10-sec incubation in liquid nitrogen and 90-sec incubation in a 37° water bath). Cell debris was removed by centrifugation, and the protein content of the supernatant was determined (Bio-Rad, Richmond, CA). Supernatant (100 μl) was added to 900 μl Z buffer with 200 μl 4 mg/ml o-nitrophenyl-β-d-galactopyranoside (Sigma). After a 5- to 20-min incubation, 1 ml 1 m Na2CO3 was added to stop the reaction, and β-galactosidase activity was determined by measuring the OD420.
Microscopy: Cells were grown to log phase, stressed with the addition of solute (1 m KCl, 1 m sorbitol, or 900 mm NaCl), and fixed by addition of formaldehyde to a final concentration of 3.7%. After incubation for at least 1 hr, cells were pelleted, washed once with PBS (8 g/liter NaCl, 0.2 g/liter KCl, 1.44 g/liter Na2HPO4, 0.24 g/liter KH2PO4, pH 7.4), and resuspended in PBS for storage. Cells were then diluted, sonicated briefly to disrupt cell clumps, and spotted onto Superfrost Plus slides (Fisher Scientific, Pittsburgh). After allowing the cells to adhere to the slide, the remaining liquid was removed by aspiration. Five microliters of mounting media (Vectashield; Vector Laboratories, Burlingame, CA) was added directly to the slide, covered with a coverslip, sealed using nail polish, and stored at 4°. The yeast cells were visualized using an Axioskop (Zeiss, Thornwood, NY) set for differential interference contrast (DIC) microscopy.
Isolation of a partially functional allele of PBS2: In a screen for osmosensitive mutants defective in HOG pathway function (see materials and methods), a partially functional allele of PBS2, pbs2-3, was isolated. pbs2-3 exhibits an intermediate osmosensitivity compared to wild-type and pbs2Δ strains. A pbs2-3 strain, unlike a pbs2Δ strain, can grow on media supplemented with 400 mm NaCl (Figure 2A). However, a pbs2-3 strain cannot grow on media supplemented with 900 mm NaCl, though a wild-type strain can grow under these conditions (Figure 2A).
Hypertonic-stress-induced changes in Hog1p phosphorylation and GPD1 transcription were analyzed to determine if the pbs2-3 mutation affected signal transduction through the HOG pathway. As a MAPK, Hog1p is activated by phosphorylation on a conserved tyrosine and threonine residue, allowing detection of activated Hog1p using commercially available antiphosphotyrosine antibodies. There is increased phosphorylation of Hog1p and subsequent induction of GPD1 mRNA accumulation in wild-type strains following osmotic stress (Figure 2, B and C), consistent with previous observations (Brewsteret al. 1993; Albertynet al. 1994). These responses to increases in osmolarity are absent in pbs2Δ strains. In a pbs2-3 strain, the levels of Hog1p phosphorylation and GPD1 mRNA following osmotic stress were intermediate between the wild-type and deletion strains. These data show that the partial osmosensitivity of the pbs2-3 strain is correlated with a partial loss of signaling through the HOG pathway.
To identify the mutations responsible for the pbs2-3 phenotype, the pbs2-3 locus was cloned and sequenced. Two mutations were identified by comparison of the pbs2-3 sequence to published PBS2 sequences. The pbs2-3 mutations are predicted to result in a substitution of proline for serine at position 168 (S168P) and a substitution of aspartate for glycine at position 509 (G509D) in the polypeptide chain of this MEK. Plasmids were constructed in which only one of the two mutations was present in an otherwise wild-type PBS2 gene. When the single mutant plasmids were introduced into pbs2Δ strains, the plasmid containing the S168P PBS2 mutation appeared to complement fully while the plasmid coding for the G509D PBS2 mutation gave a phenotype intermediate between wild type and pbs2Δ, similar to that seen in a pbs2-3 strain. Thus, the G509D substitution appears to be responsible for the phenotype of a pbs2-3 strain. The G509D substitution occurs near the residues (S514 and T518) predicted to be phosphorylated by the MEKKs of the HOG pathway (Maedaet al. 1995). The substitution of alanine for either S514 or T518 in Pbs2p prevents signaling through the HOG pathway in response to osmotic stress (Maedaet al. 1995). Although the exact effect of the G509D substitution on Pbs2p activity is unknown, it may be that the perturbation of the activation site by the G509D substitution in pbs2-3p could interfere with the efficient activation of pbs2-3p. The G509D substitution is also within the kinase domain of Pbs2p and, therefore, could also affect the catalytic activity of Pbs2p independently of any effect on activation of the MEK.
High-copy suppressor screen: The partially osmosensitive pbs2-3 strain (KDY1) was used in a screen to identify proteins that affect growth on high-osmolarity media. pbs2-3 was transformed with two high-copy yeast genomic libraries and screened for plasmid-dependent growth on high-osmolarity media (YEPD plus 900 mm NaCl). The HOG pathway genes PBS2 and HOG1 both suppressed the osmosensitivity of the pbs2-3 strain (data not shown). Three additional genes were also identified as high-copy extragenic suppressors of pbs2-3: CAK1, TRX1, and MSG5 (Figure 2A). Cak1p is the cyclin-dependent, kinase-activating kinase in yeast (Espinozaet al. 1996; Kaldiset al. 1996; Thuretet al. 1996). TRX1 is one of the two thioredoxin genes in yeast (Gan 1991). After obtaining TRX1 in the suppressor screen, a high-copy plasmid containing TRX2 was constructed and that plasmid also suppresses the osmosensitivity of pbs2-3 strains (data not shown). Msg5p is a dual-specificity (tyrosine and threonine) phosphoprotein phosphatase that has been implicated previously in the downregulation of the pheromone response pathway and cell integrity pathway MAPKs, Fus3p, and Slt2p, respectively (Doiet al. 1994; Watanabeet al. 1995). The suppression of a MAPK cascade mutant by the overexpression of a MAPK phosphatase was intriguing, so MSG5 was selected for further study.
Suppression of pbs2-3 osmosensitivity by high-copy MSG5 is not due to increased signaling through the HOG pathway: One possible explanation for the suppression of pbs2-3 by high-copy MSG5 is that the overexpression of Msg5p increases the signaling efficiency of the impaired HOG pathway. However, neither Hog1p tyrosine phosphorylation nor GPD1 mRNA levels are altered by MSG5 overexpression in a pbs2-3 strain (Figure 2, B and C). Additionally, high-copy MSG5 also weakly suppresses the osmosensitivity of a pbs2Δ strain (data not shown). One hypothesis to explain these data is that the activity of one or more Msg5p-regulated MAPK pathways is/are deleterious to growth at high osmolarity.
Kss1p is the relevant target of Msg5p in the suppression of pbs2-3 osmosensitivity: Msg5p is a phosphoprotein phosphatase that catalyzes the dephosphorylation of tyrosine and threonine residues of a subset of activated MAPKs, shifting them to a catalytically inactive state. Previous work identified Msg5p as a negative regulator of Fus3p and Slt2p, MAPKs in the pheromone response and cell integrity pathways, respectively, but not as a negative regulator of the MAPK Hog1p (Doiet al. 1994; Watanabeet al. 1995). Interestingly, both Slt2p and Fus3p have a threonine-glutamate-tyrosine (TEY) activation sequence similar to the mammalian Erk1p (Robinson and Cobb 1997), while Hog1p has a threonine-glycine-tyrosine (TGY) activation sequence similar to mammalian stress-activated protein kinases such as p38 (Kyriakis and Avruch 1996). Of the two other MAPKs in yeast, Kss1p also has a TEY activation sequence (Courchesneet al. 1989) and is a possible target of Msg5p. Smk1p, however, has a threonine-glutamine-tyrosine (TNY) activation sequence (Krisaket al. 1994), is expressed only during sporulation (Pierceet al. 1998), and is therefore an unlikely candidate. To identify which of the MAPKs is the relevant target of Msg5p with regard to the suppression of pbs2-3 osmosensitivity, deletions of FUS3, SLT2, and KSS1 were constructed in pbs2-3 and PBS2 backgrounds and tested for growth at high osmolarity.
pbs2-3 strains lacking SLT2 did not grow noticeably different than pbs2-3 alone on high-osmolarity media (Figure 3). In addition, the osmosensitivity of a pbs2-3 slt2Δ strain is still strongly suppressed by MSG5 overexpression (data not shown). These data suggest that Slt2p is not part of the putative MAPK pathway predicted to be downregulated by Msg5p as part of the suppression of pbs2-3 osmosensitivity.
pbs2-3 strains lacking FUS3 grew very poorly on the high-osmolarity media tested, though growth on YEPD alone was unaffected. This indicates that Fus3p is not the critical target of Msg5p in the suppression of pbs2-3 osmosensitivity. Surprisingly, the deletion of FUS3 not only failed to suppress the osmosensitivity of a pbs2-3 strain, it enhanced the osmosensitivity of this stain relative to pbs2-3 alone (Figure 3). Fus3p mediates mating responses (Elionet al. 1991; Gartneret al. 1992), and it appears to negatively regulate the filamentation/invasion pathway (Madhani and Fink 1997). For example, fus3Δ strains have significantly higher levels of induction of the filamentation/invasion pathway reporter gene FRE::lacZ, and they show enhanced invasion. Thus, the enhancement of pbs2-3 osmosensitivity by fus3Δ could be due to either the loss of a pheromone response pathway function or an increased activity of the filamentation/invasion pathway. This latter possibility is further supported by the following results.
In contrast to the results with slt2Δ and fus3Δ, deletion of KSS1 suppressed the osmosensitivity of a pbs2-3 strain on media containing 1 m sorbitol or 1 m KCl (Figure 3). Although interaction between Msg5p and Kss1p had not been described previously, the growth of the pbs2-3 strain lacking KSS1 on high-osmolarity media suggests that Kss1p is a target of Msg5p. This assertion is supported by the observation that there is no additional suppression of osmosensitivity by overexpressing MSG5 in a pbs2-3 kss1Δ strain (data not shown). These data indicate that activated Kss1p and, by extension, activated filamentation/invasion pathway may be deleterious for growth at high-osmolarity media in a strain with a defective HOG pathway.
Double and triple MAPK deletions were also made in pbs2-3 in an effort to unmask additional interactions between the various pathways analyzed. A consistent pattern of suppression by strains lacking KSS1 and sensitivity in strains with a KSS1 was observed regardless of the other deletions in the pbs2-3 strain. These data again support a model where Kss1p is the only relevant MAPK downregulated by Msg5p in the suppression of pbs2-3 osmosensitivity.
It has been demonstrated recently that Kss1p in its nonphosphorylated (nonactivated) state is an inhibitor of the filamentation/invasion pathway (Cooket al. 1997; Madhaniet al. 1997; Bardwellet al. 1998). When a Kss1p mutant lacking the threonine and tyrosine phosphorylation sites in the activation loop of the enzyme was expressed in a pbs2-3 strain, the suppression of the pbs2-3 osmosensitivity was even stronger than that seen in a pbs2-3 kss1Δ strain (data not shown). This further supports the assertion that activated Kss1p contributes to the growth defect of a pbs2-3 strain on high osmolarity.
Kss1p is phosphorylated in response to osmotic shock in the absence of a functional HOG pathway: If Kss1p dephosphorylation by overexpressed Msg5p suppresses the osmosensitivity of a pbs2-3 strain, then the level of Kss1p phosphorylation may be an important determinant for growth at high osmolarity in HOG pathway mutants. To investigate the phosphorylation state of Kss1p in pbs2-3 strains exposed to high osmolarity, immunoblots of proteins extracted from various strains before and after hyperosmotic shock were probed with a commercially available monoclonal antibody raised against a phosphorylated ERK1 activation loop. ERK1 is a mammalian MAPK with the same TEY activation sequence as the yeast MAPKs Kss1p, Fus3p, and Slt2p. Due to the sequence similarity in this region, the antiphosphoERK antibody also recognizes the phosphorylated forms of Kss1p, Fus3p, and Slt2p, albeit with different sensitivities (data not shown).
An antiphosphoERK immunoreactive band at the expected mobility for Kss1p (~43 kD) appears in the lane corresponding to pbs2-3 cells 1 hr following hypertonic shock, but not in wild-type cells nor in cells overexpressing MSG5 under the same conditions (Figure 4A). This 43-kD band is absent in extracts from cells lacking a KSS1 gene and is much darker in the lane corresponding to cells overexpressing KSS1 than in control cells (Figure 4B). In contrast, there was no evidence of increased phosphorylation of Fus3p or Mpk1p following osmotic stress (Figure 4, A and B). This evidence strongly suggests that Kss1p at its normal physiological concentration within the cell is phosphorylated in pbs2-3 cells following osmotic stress.
Downstream transcriptional targets of the filamentation/invasion pathway are activated in response to hypertonic stress in pbs2-3 mutants: The increased phosphorylation of Kss1p observed in a pbs2-3 strain suggests the possibility that the filamentation/invasion pathway is activated under hyperosmotic conditions. We further investigated this possibility by measuring the induction of downstream gene targets of the filamentation/invasion pathway. Normally, when the filamentation/invasion pathway phosphorylates Kss1p, Kss1p activates the transcription factors Ste12p and Tec1p, which in turn activate transcription from the filamentation/invasion response elements (FREs) present in the promoters of genes required for filamentous growth (Madhani and Fink 1997).
Two methods have been used to analyze the transcription of filamentation/invasion-pathway-responsive genes under a variety of conditions. The promoter for TEC1 has a FRE, providing a positive feedback loop for the filamentation/invasion pathway and making the accumulation of TEC1 mRNA a sensitive, endogenous indicator of pathway activation (Madhani and Fink 1997). However, a recent publication has indicated that TEC1 is not under the sole control of the filamentation/invasion pathway, as it is also activated by the pheromone response pathway (Oehlen 1998), a result independently confirmed by our lab (data not shown). An alternative assay for filamentous pathway activation is the use of a FRE::lacZ construct (Madhani and Fink 1997). This construct contains a β-galactosidase gene under the control of the filamentous pathway responsive elements. This construct appears to be very specific for filamentation/invasion pathway signals (Madhani and Fink 1997). We have analyzed both TEC1 and FRE::lacZ expression to assay the activation of the filamentation/invasion pathway transcriptional targets. Wild-type cells had a low basal level of TEC1 mRNA that did not increase appreciably following hyperosmotic shock (Figure 5A). In a pbs2-3 strain, there was a high basal level of TEC1 expression (~1.6-fold over wild type). Exposure of the pbs2-3 strain to hyperosmotic stress induced a two- to threefold increase in the TEC1 mRNA expression over the elevated basal levels and remained high. In contrast, pbs2-3 strains overexpressing MSG5 had a slightly elevated basal level of TEC1 mRNA (~1.3-fold over wild type) that increased moderately (1.4-fold over basal levels) 1 hr after hyperosmotic shock and decreased steadily thereafter.
Similar results were obtained using an FRE::lacZ reporter construct to assay filamentation/invasion pathway activity (Figure 5B). There is an increased basal level of FRE::lacZ expression in a pbs2-3 strain (1.6-fold) compared to a wild-type strain. In the pbs2-3 strain, a two- to threefold increase in FRE::lacZ was observed following osmotic stress, while there is a modest 1.6-fold increase in the wild-type strain with the same treatment. These data suggest that the filamentation/invasion-pathway-responsive genes are activated following osmotic stress in a pbs3-2 strain.
The requirement for KSS1 in the activation of FRE::lacZ was investigated. Interestingly, the deletion of KSS1 results in a reduced basal level of FRE::lacZ expression in the pbs2-3 strain back to the level observed in wild type. This low basal level does not increase following osmotic stress, consistent with the requirement for Kss1p in the osmotic induction of the filamentation/invasion pathway in strains lacking a functional HOG pathway. These data suggest that the transcriptional targets of the filamentation/invasion pathway are activated in response to osmotic stress in strains lacking a fully functional HOG pathway in a Kss1p-dependent manner.
Ste7p and Ste11p are required for growth inhibition and transcriptional activation of filamentation/invasion-pathway-responsive genes: To determine if other components of the filamentation/invasion pathway are required for the inappropriate activation of Kss1p in HOG pathway mutants following osmotic shock, deletions of the upstream components of the filamentation/invasion pathway were constructed in the pbs2-3 background. These strains were assayed for growth (Figure 6A), FRE::lacZ-dependent β-galactosidase activity (Figure 6B), and expression of TEC1 (data not shown) following osmotic stress.
The deletion of filamentation/invasion pathway MAPK cascade components (STE11, STE7, or KSS1) suppresses the osmosensitivity of a pbs2-3 strain (Figure 6A). The extent of suppression by these is roughly equivalent to that seen in a pbs2-3 strain overexpressing MSG5 (Figure 6A). Interestingly, the deletion of STE7 has a greater suppressive effect on a pbs2-3 strain than the deletion of either KSS1 or STE11.
The deletions of filamentation/invasion pathway MAPK signaling cascade components also inhibit the expression of FRE::lacZ activity in pbs2-3 cells (Figure 6B). The basal levels of FRE::lacZ expression are considerably diminished in pbs2-3 strains lacking the upstream kinases of the filamentation/invasion pathway. Also, strains lacking a functional filamentation/invasion pathway MAPK cascade no longer have an increase in FRE::lacZ-dependent β-galactosidase expression following osmotic stress. Similar results were obtained when assaying for TEC1 expression following hypertonic stress in pbs2-3 strains deleted for filamentation/invasion pathway MAPK cascade components (data not shown). The loss of filamentous-pathway-responsive gene induction following osmotic stress in strains with deletions in STE11, STE7, or KSS1 indicates that the entire MAPK module must be present for efficient activation of the filamentation/invasion pathway in strains lacking a fully functional HOG pathway.
Deletions of filamentation/invasion pathway components suppress the morphological defects of pbs2-3 on high osmolarity: HOG pathway mutants exhibit severe morphological defects following prolonged exposure to high-osmolarity conditions (Brewster and Gustin 1994). One such defect is the presence of multiple buds, a possible indication of a cell cycle defect in HOG pathway mutants on high osmolarity. A second form of aberrant morphology is the production of long projections reminiscent of the hyphae of pathogenic yeast during virulent growth phase (Loet al. 1997) and the pseudohyphae of S. cerevisiae (Gimenoet al. 1992). To determine if these morphological defects are dependent on the filamentation/invasion pathway, wild-type, pbs2-3, and pbs2-3 strains deleted for STE7 and KSS1 were grown in YEPD or YEPD supplemented with 1 m KCl and analyzed by DIC microscopy (Figure 7).
pbs2-3 strains exposed to high osmolarity produce long projections (Figure 7) similar to those observed in hog1Δ and pbs2Δ strains on high osmolarity (Brewsteret al. 1993). The projections frequently have striations that are strikingly similar to those produced by the incomplete septation of pseudohyphae (Gimenoet al. 1992). The long projections were not seen in pbs2-3 strains lacking KSS1 or STE7. However, several cells with multiple buds were observed in pbs2-3 strains lacking KSS1 or STE7 on high osmolarity (white arrows in Figure 7). These data indicate that the aberrant polarized growth morphology of HOG pathway mutants requires the filamentation/invasion pathway even though the multiple-bud phenotype does not.
In this article, we describe a previously unknown consequence of disrupting the HOG pathway—the activation of the filamentation/invasion pathway by hyperosmotic shock. One high-copy suppressor of the HOG pathway MEK mutant pbs2-3 identified in this study was MSG5 (Figure 2A), a MAPK phosphatase. The critical target of Msg5p in the suppression of pbs2-3 osmosensitivity, Kss1p, was identified on the basis of suppression of pbs2-3 by kss1Δ (Figure 3) and the lack of effect of high-copy MSG5 in a pbs2-3 kss1Δ strain (data not shown). In strains lacking a functional HOG pathway, Kss1p is phosphorylated (Figure 4) and filamentation/invasion pathway gene targets are transcriptionally activated (Figure 5) following osmotic stress. The deletion of the filamentation/invasion pathway MAPK module components STE7 or STE11 inhibited hyperosmolarity-induced filamentation/invasion pathway activation (Figure 6B) and restored high-osmolarity growth in pbs2-3 strains (Figure 6A). Finally, one of the morphological defects of HOG pathway mutants on high osmolarity, the presence of long projections, was demonstrated to be dependent on the presence of an intact filamentation/invasion pathway (Figure 7).
The HOG pathway negatively regulates the filamentation/invasion pathway: Our data demonstrate that the filamentation/invasion pathway is activated in response to osmotic stress in strains with a compromised HOG pathway, but that it is not activated in wild-type strains. Kss1p phosphorylation (Figure 4), TEC1 mRNA levels (Figure 5A), and FRE::lacZ expression (Figure 5B) are all increased following osmotic stress in strains lacking a fully functional HOG pathway. In strains with an intact HOG pathway, there is no detectable change in Kss1p phosphorylation (Figure 4) or TEC1 mRNA (Figure 5A) levels following an increase in osmolarity. Though a small increase in FRE::lacZ expression could be detected in wild-type strains following osmotic stress, the magnitude of the increase was lower than in HOG mutants (Figure 5B). These data are consistent with either the inappropriate activation or the loss of repression of the filamentation/invasion pathway in HOG mutants during osmotic stress. However, the increased basal levels of TEC1 mRNA (Figure 5A) and FRE::lacZ expression (Figure 5B) in pbs2-3, hog1Δ, or pbs2Δ strains are more consistent with the latter model. Together, these data support a role for the HOG pathway in negatively regulating the filamentation/invasion pathway.
Other results are consistent with a model in which the HOG pathway negatively regulates the filamentation/invasion pathway. Diploid hog1Δ/hog1Δ yeast of the R strain background show more active pseudohyphal development on low-nitrogen media than does a wild-type R strain, indicating an increased filamentation/invasion pathway activity (Madhaniet al. 1997; O'Rourke and Herskowitz 1998). Our data may provide a biochemical explanation for the hyperpseudohyphal phenotype of hog1Δ cells, as we have demonstrated that the basal filamentation/invasion pathway activity, as measured by TEC1 mRNA and FRE::lacZ activity, is increased in HOG pathway mutants. We observed that the elevated basal levels of TEC1 and FRE::lacZ could be eliminated by the deletion of the components of the filamentation/invasion pathway MAPK cascade. Together, these data suggest that it is the loss of filamentation/invasion pathway regulation by the HOG pathway that causes the hyperpseudohyphal phenotype of HOG pathway mutants in the R background.
The activation of the filamentation/invasion pathway is correlated with growth inhibition on high-osmolarity media: The activation of the filamentation/invasion pathway has a physiological effect on yeast—growth inhibition on high osmolarity. The level of FRE-dependent transcription and the amount of pbs2-3 strain growth on high-osmolarity media appear to be inversely correlated with each other. pbs2-3 strains have increased FRE::lacZ and TEC1 expression, and they have a growth defect on high-osmolarity media (Figures 5, B and A, and 2A, respectively). A pbs2-3 ste7Δ strain has a much lower level of FRE::lacZ expression than does a pbs2-3 strain, and it grows well on high-osmolarity media (Figures 6, B and A, respectively). The growth inhibition of a pbs2-3 kss1Δ strain is intermediate between that of a pbs2-3 strain and a pbs2-3 ste7Δ strain (Figure 6A), correlated with the intermediate levels of TEC1 mRNA accumulation (data not shown) and FRE::lacZ expression (Figure 6B). These data suggest that an activated filamentation/invasion pathway contributes to the decreased growth of a pbs2-3 strain on high-osmolarity media.
A pbs2-3 strain with a deletion in STE11 does not seem to fit such a model, though the complex roles of Ste11p may explain the discrepancy. A pbs2-3 ste11Δ strain and a pbs2-3 ste7Δ strain have equally low levels of TEC1 (data not shown) and FRE::lacZ expression following osmotic stress (Figure 6B). However, the deletion of STE11 does not suppress the osmosensitivity of pbs2-3 as well as the deletion of STE7 does (Figure 6A). The loss of Ste11p results in both a loss of filamentation/invasion pathway activation (increased osmotolerance) and the loss of one upstream branch of the HOG pathway (possibly decreased osmotolerance). Though the loss of the Sho1p branch of the HOG pathway did not affect high-osmolarity growth in an otherwise wild-type strain (Posas and Saito 1997), the increased sensitivity of the pbs2-3 strain to perturbations in signaling may account for the reduced growth on high-osmolarity in a ste11Δ strain relative to a ste7Δ strain. The correlation of filamentation/invasion pathway activity and decreased growth on high osmolarity is found not only in pbs2-3 cells, but also in other HOG pathway mutants. On high-osmolarity media, a pbs2Δ strain has higher levels of TEC1 mRNA and FRE::lacZ expression than does a pbs2-3 strain under the same conditions (data not shown). The higher activity of the filamentation/invasion pathway in pbs2Δ relative to pbs2-3 is correlated with greater growth inhibition of pbs2Δ relative to pbs2-3 (Figure 2A). Likewise, a hog1Δ strain has a greater accumulation of TEC1 mRNA following hyperosmotic stress than does a pbs2-3 strain, and it is also less osmotolerant than a pbs2-3 strain (data not shown).
The deletion of filamentation/invasion pathway genes failed to fully complement a pbs2-3 strain (Figure 6A) or cells with a deletion of either HOG1 or PBS2 (data not shown). These data indicate that the deactivation of the filamentation/invasion pathway is not the sole determinant of growth on high-osmolarity media. This is not unexpected, as previous work demonstrated the importance of GPD1 expression in the growth of yeast on high-osmolarity media (Albertynet al. 1994). The expression of GPD1 in a pbs2-3 strain is not altered by the overexpression of MSG5 (Figure 2C) nor by the deletion of filamentation/invasion pathway genes (data not shown). These data indicate that the HOG pathway coordinates responses to osmotic stress that affect growth on high-osmolarity media independently of the repression of the filamentation/invasion pathway.
Mechanism of filamentation/invasion pathway activation after osmotic stress: The research presented here demonstrates that the filamentation/invasion pathway is activated following hypertonic stress in HOG pathway mutants. It is perhaps significant that two pathways share a MEKK, Ste11p. It is therefore possible that the activation of the filamentation/invasion pathway following hypertonic stress occurs because of a leakage of signal through this common component. It is unclear whether Ste11p freely diffuses between pathways or is bound tightly in a signaling complex within the individual pathways. In the case of the pheromone response pathway, it appears that Ste11p is tightly bound to the scaffold protein Ste5p. Though a role for Pbs2p as a similar scaffold has been suggested (Posas and Saito 1997), it is unclear how tightly the signaling components of the HOG pathway are bound. Thus, the activation of the filamentation/invasion pathway may be an unfortunate consequence of the sharing of Ste11p between the two pathways.
Alternatively, the simultaneous activation of the filamentation/invasion pathway by osmotic stress and the repression of the filamentation/invasion pathway by the HOG pathway may be part of a mechanism providing the most advantageous response to a given stimuli. For example, if cells encounter both increased osmolarity and nutrient loss, it may be more advantageous for the yeast cell to move the yeast (or its progeny) to an environment more conducive to growth via filamentation, e.g., the inside of a grape vs. the skin. This response could be selected at the level of intracellular signaling if the filamentation/invasion pathway activation by both osmolarity and nutrient deprivation overcomes the HOG-pathway-mediated repression. However, if the cell is simply being dehydrated but has a nutrient-rich environment, the HOG pathway repression of the filamentation pathway will prevent the expenditure of cellular energy on filamentation and focus its energy on overcoming the osmotic challenge.
Cross-talk between MAPK cascades in yeast: One possible mechanism for cross-pathway regulation is the activation by one MAPK pathway of a phosphatase with a specificity for components of a second MAPK pathway. In HeLa cells, the MAPK ERK2 induces the expression of the tyrosine/threonine phosphatase MKP1-1, which is more active against the SAPK and p38 than ERK2 itself (Franklin and Kraft 1997). Thus, a mechanism exists for the downregulation of p38 and SAPK by ERK2 through the transcription of a phosphatase. MAPKs may also directly activate MAPK phosphatases, as is the case with ERK2 and the phosphatase MPK3-1 (Campset al. 1998). However, it is not yet known if the direct activation of phosphatases by MAPKs is part of the cross-pathway regulation mechanisms.
There are a number of phosphatases in yeast that could act in an analogous manner to the mammalian phosphatases described above. Ptp2p is transcriptionally regulated by the HOG pathway (Jacobyet al. 1997; Wurgler-Murphyet al. 1997), but is more specific for Hog1p than for the ERK-like MAPKs (Zhanet al. 1997). Ptp3p is perhaps a more likely candidate, as it is activated by the HOG pathway (Jacobyet al. 1997; Wurgler-Murphyet al. 1997) and has a higher specificity for the ERK-like MAPKs than for Hog1p (Zhanet al. 1997). However, neither the overexpression of Ptp2p nor Ptp3p suppressed the osmosensitivity of a pbs2-3 strain (data not shown), though a dependence on a wild-type HOG pathway for full phosphatase activity could account for the lack of suppression. Msg5p is specific for the ERK family of MAPKs, including Slt2p (Watanabeet al. 1995), Fus3p (Doiet al. 1994), and Kss1p (this study) in yeast, and suppresses pbs2-3 osmosensitivity when overexpressed. However, the transcription of Msg5p is not Hog1p dependent (data not shown), though post-transcriptional activation of Msg5p by the HOG pathway is still possible.
MAPK phosphatase specificity: MSG5 had been identified previously as a high-copy suppressor of constitutively active mutants in pheromone response (Doiet al. 1994) and cell integrity pathways (Watanabeet al. 1995), presumably by downregulating the MAPKs of these pathways, Fus3p and Slt2p. Here, we have provided evidence that is consistent with the hypothesis that Msg5p acts on the filamentation/invasion pathway MAPK Kss1p. It is perhaps significant that the three MAPKs that are targets of Msg5p have identical TEY activation sequences, while Hog1p and Smk1p have TGY and TQY sequences, respectively (Robinson and Cobb 1997; Gustinet al. 1998). This may indicate that Msg5p exhibits specificity for the ERK1 family of MAPKs. However, additional regions of the MAPK are believed to affect its interactions with upstream MAPK cascade components (Brunet and Pouyssegur 1996) and may affect the specificities of MAPK phosphatases as well.
The multiple responsibilities of the HOG pathway in the response to hyperosmotic shock are complicated: Hyperosmotic stress is a serious challenge to a cell's survival. As such, hyperosmotic stress response requires the simultaneous regulation of several components. In addition to producing a proper response to the stimulus of osmotic stress, the HOG pathway has an important role in preventing inappropriate signaling through other MAPK pathways. The mechanism of HOG-mediated regulation of other MAPK pathways in yeast is as yet unknown, but it will prove to be an exciting area of study in the future.
We thank Ed Winter, Elaine Elion, Jeremy Thorner, and Gerald Fink for their generous donation of several plasmids used in this study. United States Biologicals supplied many chemicals used in this research. We also thank Jacobus Albertyn, Matt Alexander, and Sue Gibson for their critique of this manuscript. Research was supported by the National Science Foundation (grant MCB 9506987), the American Cancer Society (grant BE-224), and the National Aeronautics and Space Administration (grant NAGS-4072). K.D.D. was supported by the National Institutes of Health (grant GMO-8362-7) and by Quality Bioresources, Inc.
Communicating editor: M. Johnston
- Received January 7, 1999.
- Accepted July 6, 1999.
- Copyright © 1999 by the Genetics Society of America