Stress-specific Activation Mechanisms for the “Cell Integrity” MAPK Pathway*

Many environmental stresses trigger cellular responses by activating mitogen-activated protein kinase (MAPK) pathways. Once activated, these highly conserved protein kinase cascades can elicit cellular responses such as transcriptional activation of response genes, cytoskeletal rearrangement, and cell cycle arrest. The mechanism of pathway activation by environmental stresses is in most cases unknown. We have analyzed the activation of the budding yeast “cell integrity” MAPK pathway by heat shock, hypoosmotic shock, and actin perturbation, and we report that different stresses regulate this pathway at different steps. In no case can MAPK activation be explained by the prevailing view that stresses simply induce GTP loading of the Rho1p GTPase at the “top” of the pathway. Instead, our findings suggest that the stresses can modulate at least three distinct kinases acting between Rho1p and the MAPK. These findings suggest that stresses provide “lateral” inputs into this regulatory pathway, rather than operating in a linear “top-down” manner.

Both extracellular signaling molecules and intracellular stresses frequently elicit their specific cellular responses through highly conserved signal transduction modules consisting of protein kinase cascades culminating in the activation of mitogen-activated protein kinases (MAPKs) 1 (1). MAPKs are a family of serine/threonine protein kinases that are activated by an unusual mechanism involving dual threonine and tyrosine phosphorylation, which is catalyzed by a family of MAPK kinases (MAPKKs). These in turn are phosphorylated and activated by a family of MAPKK kinases (MAPKKKs), and the three kinases of a particular MAPK module are frequently physically associated with each other (sometimes with the help of noncatalytic scaffold proteins) in multimeric complexes (2). In contrast to the highly conserved molecular architecture of the MAPK cascade, the mechanisms whereby various stimuli activate MAPK activity appear to be quite diverse.
Perhaps the most extensively investigated instance of MAPK activation involves signal transduction by receptor tyrosine kinases, which activate a MAPK cascade in response to extra-cellular growth factors (3). Activation of the receptor upon growth factor binding promotes the recruitment of the guanine nucleotide exchange factor Sos to the plasma membrane, where Sos encounters its target GTPase, Ras, leading to GTP loading of Ras. GTP-bound Ras then stimulates a MAPK cascade involving the kinases Raf or Raf-B (MAPKKK), MEK (MAPKK), and ERK (MAPK) (4).
In contrast to the detailed understanding of MAPK cascade activation by growth factors, much less is known about how MAPK pathways are stimulated by cellular stresses. One stress-responsive MAPK that has been extensively studied in Saccharomyces cerevisiae is the Slt2p/Mpk1p MAPK, which is activated by the related (and redundant) MAPKKs Mkk1p and Mkk2p, which are in turn activated by the MAPKKK Bck1p (5). This signaling module has been called the "cell integrity" MAPK cascade, as it is required for proper construction of the cell wall in order to prevent cell lysis (6,7). Mpk1p is activated in response to various stresses, and Mpk1p activation serves to protect yeast cells from stress by inducing transcription of genes that promote cell wall remodeling and by contributing to the cell cycle arrest triggered by the morphogenesis checkpoint, which delays mitosis until cells have successfully built a bud (5,8,9).
Genetic studies established that Pkc1p, the sole protein kinase C homologue in S. cerevisiae, is absolutely required for any activity in the cell integrity MAPK pathway, and acts upstream of the MAPKKK Bck1p (10,11). Several additional functions, independent of the MAPK pathway, have also been ascribed to Pkc1p (12). Pkc1p activity in turn requires the binding of GTP-loaded Rho1p (13,14), and GTP loading of Rho1p is mediated by the partially redundant guanine nucleotide exchange factors Rom1p and Rom2p (15). By analogy to the Ras-Raf-MEK-ERK pathway, it has been suggested that stresses activate the cell integrity pathway by stimulating GTP loading of Rho1p, thereby activating Pkc1p to phosphorylate Bck1p, initiating activation of the kinase cascade (5,9,16,17).
A multiplicity of stimuli have been shown to promote Mpk1p activation. These include heat shock, hypoosmotic shock, actin depolymerization, and treatment with chlorpromazine, caffeine, vanadate, zymolyase, Congo red, calcofluor, rapamycin, and mating pheromone (8, 10, 18 -21). A simplifying hypothesis to accommodate these observations is that there is a common stressful consequence of all of these treatments that causes cells to activate Mpk1p (10). A number of plasma membrane glycoproteins (Wsc1p-3p, Mid2p) that influence Mpk1p activity have been identified, and it is thought that these proteins might act as "stress sensors" that somehow detect cellular stress and transduce a signal to activate Rom1p/Rom2p (16,(22)(23)(24). However, these putative stress sensors appear to be dispensable for Mpk1p activation in response to actin depolymerization (8) or rapamycin (18), raising the possibility that different stresses might employ distinct pathways to activate Mpk1p. Precedent for stress-specific regulation of MAPK pathways is provided by findings on the Schizosaccharomyces pombe Sty1p pathway, in which it appears that although other stresses activate Sty1p through regulation of its upstream kinases, heat shock activates Sty1p primarily through regulation of Sty1p-directed phosphatases (25)(26)(27).
In this report, we have dissected the requirements for activation of the cell integrity pathway by different stresses. Our findings are inconsistent with the view that all stresses activate the pathway in a "top-down" manner by stimulating Rho1p GTP loading, resulting in the activation of a linear Pkc1p/Bck1p/Mkk1,2p/Mpk1p kinase cascade. We find that different stresses can activate the pathway by different mechanisms, in one case regulating the core kinases Mkk1,2p/Mpk1p themselves rather than Rho1p or Pkc1p. Thus, different stresses activate Mpk1p through signaling pathways that impact the MAPK cascade at different levels.

MATERIALS AND METHODS
Media, Growth Conditions, Stresses, and Drug Treatments-Yeast media (YEPD rich media, synthetic media lacking specific nutrients, and sporulation media) have been described (49). For heat shock cells were grown to mid-log phase in media supplemented with 1 M sorbitol at 24°C and then shifted to 39°C for 35 min. For Lat-B treatment cells were grown to mid-log phase in media supplemented with 1 M sorbitol at 30°C and then treated with Lat-B (BioMol Research Laboratories, Inc., Plymouth Meeting, PA) at 100 M for 2 h. For hypoosmotic shock, cells were grown to mid-log phase in media supplemented with 1 M sorbitol at 24°C, diluted 10-fold into water for ϳ1 min, and immediately pelleted and frozen at Ϫ80°C.
Strains, Plasmids, and PCR Manipulations-Standard media and methods were used for plasmid manipulations (50) and yeast genetic manipulations (49). The yeast strains used in this study are listed in Table I.
To express Rho1p Q68H in yeast, a 2-kb HindIII/SacI fragment containing RHO1 Q68H (29) was isolated from p68LEUe (gift from A. Myers) and ligated into the corresponding sites in pRS305 (51), yielding pDLB2257. Integration at the RHO1 genomic locus was targeted by digestion with MluI leaving the endogenous RHO1 gene intact.
To express Bck1-20p in yeast, a 6.5-kb XhoI/NotI fragment from pRS314BCK1-20 (11) was cloned into pRS306 (51), yielding pDLB1774. Integration at the BCK1 genomic locus was targeted by digestion with MluI. Because the MluI site is in the BCK1 promoter, this plasmid can also be integrated next to bck1⌬LEU2.
To analyze cells lacking Pkc1p activity, we employed strains in which the only copy of Pkc1p was expressed from the GAL1 promoter (8). These strains were grown in osmotically stabilized dextrose-containing media for at least 36 h to repress Pkc1p.
Biochemical Procedures-Procedures for cell lysis, SDS-PAGE, and Western blotting were as described (8).

RESULTS
Effect of RHO1 Q68H on Mpk1p Activity-As mentioned in the Introduction, the activity of Pkc1p and subsequent kinases in the cell integrity pathway requires GTP-bound Rho1p. Although it is thought that stresses trigger an increase in GTP loading of Rho1p and in Pkc1p activity, technical limitations have thus far precluded direct measurement of the amount of Rho1p-GTP or of the activity of Pkc1p in vivo. In the case of the mitogenic growth factors that promote GTP loading of Ras, one of the major findings that established the paradigm was that oncogenic mutants of Ras that lock the protein in its GTPbound state mimic constitutive signaling by the receptors, leading to constitutive ERK activation and uncontrolled proliferation (28). By analogy, we reasoned that expression of a GTP- locked mutant of RHO1 should lead to constitutive Mpk1p activation. Mpk1p activity can be monitored using a phosphospecific antibody that recognizes only the doubly phosphorylated (threonine and tyrosine) form of Mpk1p. Using this reagent, we compared Mpk1p activity in wild-type cells to that in cells that expressed the GTP-locked RHO1 Q68H allele (29), analogous to the GTP-locked oncogenic Ras mutant Ras Q61H (28).
Previous studies showed that overexpression of RHO1 Q68H from the strong GAL1 promoter triggered the accumulation of multiple transcripts known to be responsive to Mpk1p activity (30), indicating that Mpk1p had indeed been activated by excess GTP-Rho1p. However, GAL1-RHO1 Q68H also caused dramatic actin depolarization, through a pathway that required Pkc1p but not Mpk1p (31). This result raised the possibility that Mpk1p activation in GAL1-RHO1 Q68H strains was the result of actin stress induced by overexpression. To assess whether more physiological (and therefore less disruptive) activation of Rho1p would activate Mpk1p, we introduced a single integrated copy of RHO1 Q68H expressed from the RHO1 promoter into a wild-type strain (in addition to wild-type RHO1). Single-copy RHO1 Q68H caused detectable but much less severe actin depolarization than GAL1-RHO1 Q68H (data not shown), suggesting that limited hyperactivation of Pkc1p took place, but the cells were viable and proliferated normally. However, the activity of Mpk1p in cells with a single integrated copy of RHO1 Q68H was indistinguishable from that in wild-type cells (Fig. 1A). The simplest interpretation of this result is that additional GTP loading of Rho1p (without overexpression) does not automatically elevate Mpk1p activity. Alternatively, constitutive activation of Rho1p might provoke adaptive desensitization of the pathway, returning Mpk1p activation to basal levels. If these cells were desensitized, then we would expect that they would no longer activate Mpk1p in response to stress. However, activation of Mpk1p in response to stress from heat shock, hypoosmotic shock, or actin depolymerization was completely unaffected by RHO1 Q68H (Fig. 1B). This stress responsiveness was not because of Rho2p (which has some functional overlap with Rho1p (15, 32)), because similar results were obtained in cells lacking Rho2p (Fig. 1B). These results are inconsistent with the hypothesis that Mpk1p regulation occurs simply by modulating Rho1p GTP loading, and imply that significant regulation of the cell integrity pathway must occur downstream from Rho1p, presumably by regulating the activity of one or more of the downstream kinases.
Regulatory or Permissive Roles for the Cell Integrity Kinases?-For each of the kinases upstream of Mpk1p in the cell integrity pathway, one can ask whether regulation of that kinase is important for the activation of Mpk1p in response to a given stress. Previous studies demonstrated that these kinases are each absolutely required for Mpk1p activity (10,11,13,14,33), but that requirement does not discriminate whether a given kinase plays an active role in stress-mediated Mpk1p activation or a passive role providing a constitutive pathway input that is then modulated at a level downstream of that kinase by stress-responsive signaling pathways. In the latter scenario, it should be possible to dispense with a particular kinase entirely yet retain stress-responsive Mpk1p activity if an alternative source for the constitutive pathway input could be found. In the studies reported below, we have employed mutant alleles of BCK1 and MKK1 to provide a constitutive pathway input in the absence of the normally essential upstream kinases (Pkc1p and Bck1p, respectively).
To provide a constitutive pathway input in the absence of Pkc1p, we employed the previously described BCK1-20 allele (11). The lethality of the PKC1 deletion is suppressed by the dominant BCK1-20 allele, a result that we confirmed in our strain background ( Fig. 2A). Bck1p is a large 1478-amino acid kinase with a C-terminal catalytic domain (residues 1175 to 1440), and Bck1-20p contains a missense mutation immediately upstream of the kinase domain (Ala-1174 to Pro) (11). Bck1-20p is thought to encode a constitutively activated, Pkc1p-independent form of Bck1p that mimics Pkc1p-phosphorylated Bck1p, although the Pkc1p-targeted phosphorylation sites on Bck1p have not yet been mapped (11).
To provide a constitutive pathway input in the absence of Bck1p, we constructed the MKK1 DD allele. In other MAPK cascades, MAPKK activation by the MAPKKK involves the phosphorylation of two conserved serine residues on the MAPKK (34,35). Mutation of these residues to aspartic acid mimics phosphorylation, yielding constitutively active MAP-KKs (26,34). By analogy to these earlier studies we generated a mutant, MKK1 DD , predicted to encode an activated version of Mkk1p. Mkk1p DD was expressed at levels comparable with those of wild-type Mkk1p (data not shown), and it was functional by the criterion that it could rescue the caffeine sensitivity of mkk1⌬ mkk2⌬ cells (Fig. 2C) (yeast cells lacking components of the cell integrity pathway are unable to grow on medium containing 10 mM caffeine, presumably because of their weakened cell wall (36). Mkk1p DD , but not wild-type Mkk1p, also rescued the caffeine sensitivity of bck1⌬ cells (Fig.   FIG. 1. Rho1-GTP does not dominantly activate Mpk1p. A, wild type (DLY1) or RHO1 Q68H (DLY4509) cells were grown in YEPD at 30°C and activity of Mpk1p was monitored by Western blotting with anti-diphospho-p44/p42 MAPK. Different amounts of cell lysate were examined to aid the comparison. Loading control for this and all other experiments is the anti-PSTAIRE antibody that recognizes Cdc28p and Pho85p. B, WT (DLY1), RHO1 Q68H (DLY4509), RHO1 Q68H rho2⌬ (DLY6473), and rho2⌬ (DLY4497) cells were subjected to the indicated stresses and Mpk1p activity was monitored. For heat shock, cells were grown at 24°C in YEPD supplemented with 1 M sorbitol and shifted to 39°C for 35 min. For hypoosmotic shock, cells were grown at 24°C in YEPD supplemented with 1 M sorbitol then diluted 10-fold in water and immediately (ϳ1 min) pelleted and frozen at Ϫ80°C. For actin perturbation, cells were grown at 30°C in YEPD supplemented with 1 M sorbitol and treated with 100 M Lat-B for 2 h. WT, wild-type. 2C). Furthermore, cells expressing Mkk1p DD displayed a detectable basal (although lower than wild-type) level of Mpk1p activity, in the presence or absence of Bck1p. The basal level in cells grown with osmotic support was very low (Fig. 3D) and in some gels fell below the threshold for detection (Fig. 3E). These results suggest that Mkk1p DD is a Bck1p-independent form of Mkk1p, although it may not be as active as appropriately phosphorylated Mkk1p. With these mutants, we could now ask whether any specific stress could still regulate Mpk1p activity in the absence of Pkc1p (using the pkc1⌬ BCK1-20 strain) or Bck1p (using the bck1⌬ MKK1 DD strain).
Role of Pkc1p and Bck1p in the Regulation of Mpk1p by Stress-We examined whether cells lacking Pkc1p but express-ing Bck1-20p can regulate the activity of Mpk1p in response to cellular stresses. As shown in Fig. 3, B and C, neither actin perturbation by Lat-B nor hypoosmotic shock elicited Mpk1p activation in this strain, whereas both stimuli strongly activated Mpk1p in wild-type strains. In contrast, heat shock elicited Mpk1p activation even in the pkc1⌬ BCK1-20 strain (Fig.  3A), indicating that this stimulus regulates Mpk1p activity by a pathway that does not require modulation of Pkc1p.
Based on the Pkc1p-independent activity exhibited by Bck1-20p, we expected that cells containing Bck1-20p as the sole source of Bck1p would be unresponsive to signaling from Pkc1p. However, we found that actin depolymerization readily activated Mpk1p in BCK1-20 cells when Pkc1p was present, but not when Pkc1p was absent (Fig. 3C). This observation suggested that Bck1-20p was still responsive to Pkc1p even though it had Pkc1p-independent basal activity. Consistent with this hypothesis, expression of an activated allele of Pkc1p strongly induced Mpk1p in cells containing Bck1-20p as the sole source of Bck1p (Fig. 2B). We conclude that Bck1-20p can still respond to Pkc1p, and that stimulation of Mpk1p by Lat-B requires modulation of Pkc1p.
Intriguingly, we found that hypoosmotic shock was unable to robustly induce Mpk1p in cells containing both Pkc1p and Bck1-20p (Fig. 3B). This result suggests that even though the Bck1-20p allele allows signaling from Pkc1p, it is unresponsive to the signal triggered by hypoosmotic shock. The simplest interpretation of this result is that hypoosmotic shock acts by modulating Bck1p activity in a manner that is different from Bck1p control by Pkc1p (see "Discussion" for other possible interpretations).
We next examined whether any of the cellular stresses could regulate Mpk1p in the absence of Bck1p. Cells expressing Mkk1p DD as their only source of Mkk1p/Mkk2p displayed a basal level of Mpk1p activity that was not affected by the presence of Bck1p. The activity of Mpk1p was not efficiently stimulated by either hypoosmotic shock (Fig. 3E) or actin depolymerization (Fig. 3F), in the presence or absence of Bck1p. These results suggest that Mkk1p DD is no longer responsive either to Bck1p or to regulation by these two stresses, consistent with our conclusions (above) that both of these stresses act via Bck1p. In contrast to these results, heat shock robustly activated Mpk1p in cells carrying Mkk1p DD , either in the presence or absence of Bck1p (Fig. 3D). This result indicates that heat shock can promote pathway activation at the level of Mkk1/2p and/or Mpk1p.
One possible scenario is that heat shock inhibits the phosphatases that dephosphorylate Mpk1p (Msg5p, Sdp1p, Ptp2p, and Ptp3p), so that constitutive signaling through the kinase cascade is allowed to activate Mpk1p. Indeed, a similar scheme has been shown to apply for heat shock-induced activation of the MAPK Sty1p in S. pombe (27). Analyses of strains deleted for individual phosphatases have shown that deletion of the phosphatase caused elevation of both the basal and the heat shock-induced level of Mpk1p activity (19,(37)(38)(39)(40). However, there remained the possibility that the phosphatases were coordinately regulated by heat shock so that inhibition of the phosphatases still present could mediate the response. To address this issue we generated a panel of strains deleted for every combination of the four phosphatases. As shown in Fig.  4A, Mpk1p basal activity increased as progressively more phosphatases were deleted, reaching very high levels in the sdp1⌬ msg5⌬ ptp2⌬ ptp3⌬ quadruple mutant. Nevertheless, Mpk1p could still be stimulated further by heat shock treatment of these strains (Fig. 4B). These findings suggest that heat shock could still induce Mpk1p independent of phosphatase regulation, perhaps by activating the kinases Mkk1/2p that act on FIG. 2. Suppression of pkc1⌬ and bck1⌬ phenotypes by activated alleles of Bck1p and Mkk1p. A, wild type (DLY1), pkc1⌬ (DLY3962), and pkc1⌬ BCK1-20 (DLY456) strains were grown on YEPD plates supplemented with 1 M sorbitol for 2 days at 24°C and then replica plated to YEPD plates for 2 additional days of growth at 24°C. B, wild type (WT) (DLY1), bck1⌬ (DLY3994), and BCK1-20 (DLY455) cells carrying pDL242 (CEN URA3 GAL1-PKC1*) were grown in non-inducing sucrose media, Pkc1p* was induced by addition of galactose to the culture for the indicated time, and Mpk1p activity was monitored by Western blot. C, characterization of the dominant active MKK1 DD allele. Wild type (DLY1), mkk1⌬ mkk2⌬ (DLY4351), or bck1⌬ (DLY3994) cells carrying vector alone (pRS314), MKK1 (pDLB823), or MKK1 DD (pDLB824) plasmids as indicated were streaked onto Ϫtrp media (left plate) and also onto a YEPD plate containing 10 mM caffeine (right plate). Cells lacking activity of the Pkc1p/Mpk1p pathway cannot grow on this media (see text). In the absence of Bck1p, wild type Mkk1p is inactive and cannot support pathway activity. bck1⌬ cells carrying MKK1 DD , however, are viable on caffeine suggesting that Mkk1p DD is activated and does not require Bck1p for its function. Mpk1p (Fig. 5). In aggregate, these data provide strong evidence that regulation of the cell integrity pathway is mediated by distinct, stress-specific mechanisms.

DISCUSSION
The most striking conclusion from our results is that the prevailing top-down view of MAPK signaling, wherein regulatory signals act to stimulate guanine nucleotide exchange factor-mediated GTP loading of the G protein at the top of the cascade and activation is then transmitted automatically through sequential activation of the various kinases, cannot account for stress-induced regulation of the cell integrity pathway in yeast. Instead, our results suggest that various steps downstream of Rho1p GTP loading can be modulated by separate inputs that act in response to specific stresses (Fig. 5).
We found that expression of a constitutively GTP-loaded allele of Rho1p, Rho1p Q68H , from its own promoter was insufficient to stimulate Mpk1p activity. Moreover, cells containing this excess GTP-Rho1p were still responsive to the stresses we examined. This result indicates that cells must have "buffering" mechanisms to protect themselves from excess GTP-Rho1p. One likely layer of protection is provided by the multiple GTPase activating proteins (GAPs) that are thought to act on Rho1p-GTP including Bem2p, Sac7p, Lrg1p, and Bag7p (32,(41)(42)(43)(44). It has been suggested that the different GAPs may regulate distinct pools of Rho1p that control different downstream pathways (44,45). It seems plausible that the GAPs may sequester much of the Rho1p Q68H , and that stresses trigger release of the sequestered Rho1p Q68H from specific GAPs. In the context of wild-type Rho1p, such GAP down-regulation by stress would "open the gate" for signaling by GTP-Rho1p, and could act together with up-regulation of Rho1p GTP loading to achieve appropriate signaling.
Downstream of Rho1p, we found that different stresses exhibited different genetic requirements for signaling. Previous studies have shown that Mpk1p activity is absolutely dependent on the upstream kinases Pkc1p and Bck1p (10), but they did not distinguish whether those kinases were regulated in response to stress or were merely required to provide a basal pathway input that could be regulated downstream of those kinases. Our data on Mpk1p activation by heat shock strongly support the latter hypothesis. In particular, we found that strains lacking Pkc1p (and providing pathway activity through Bck1-20p) and strains lacking Bck1p (and providing pathway activity through Mkk1p DD ) were still able to activate Mpk1p in response to heat shock. The simplest interpretation of these results is that heat shock triggers a signaling pathway that regulates either Mkk1/2p or Mpk1p itself (Fig. 5).
In contrast to heat shock, actin depolymerization and hypoosmotic shock no longer promoted Mpk1p activation if Bck1p was absent, suggesting that these stresses regulate the pathway at or above the level of Bck1p. Interestingly, actin stress was capable of activating Mpk1p in cells containing Bck1-20p as the sole source of Bck1p activity, whereas hypoosmotic shock was not. This result indicates that the two stresses must have different effects on the cell integrity pathway. Elucidating the basis for that difference is complicated by the poorly understood nature of the Bck1-20p protein. Previous studies, confirmed here, indicated that Bck1-20p could suppress the need for Pkc1p in Mpk1p activation, and suggested that Bck1-20p might encode an activated, Pkc1p-independent form of Bck1p (11). However, although Bck1-20p certainly displays a Pkc1pindependent basal activity, we found that overexpression of an activated allele of Pkc1p dramatically activated Mpk1p in cells containing Bck1-20p. This result suggests that Bck1-20p is still responsive to Pkc1p, although it is no longer absolutely dependent on Pkc1p.
Actin stress could activate Mpk1p in a strain containing both Pkc1p and Bck1-20p, but could not activate Mpk1p in a strain containing Bck1-20p but lacking Pkc1p. This result suggests that actin stress-mediated signaling is transmitted through Pkc1p, and we suggest that this stress acts either by modulating the Rho1p GAPs (allowing Pkc1p activation by Rho1p-GTP) or through a separate input to modulate Pkc1p activity (Fig. 5). Pkc1p is subject to several modes of regulation including Rho1p binding, interaction with membrane lipids, localization, and phosphorylation (13,14,46,47). Whether any of these modes is responsible for Mpk1p regulation in response to actin perturbation is unknown.
Unexpectedly, we found that hypoosmotic shock was unable to activate Mpk1p in cells containing Bck1-20p as the only source of Bck1p, even if they contained Pkc1p. This defect was recessive, in that cells containing both wild-type Bck1p and Bck1-20p were able to activate Mpk1p following hypoosmotic shock (data not shown). These findings suggest that Bck1-20p is specifically unable to mediate signaling by hypoosmotic shock, perhaps suggesting that this stress normally acts on the pathway at the level of Bck1p (Fig. 5). Alternatively, the BCK1-20 mutation may affect scaffolding functions of Bck1p important for allowing effective input at other levels of the pathway. Given the marked difference in the time course of Mpk1p activation by actin stress (gradual sustained increase over 2 h) and hypoosmotic shock (rapid but transient induction peaking at ϳ1 min), it is also possible that both stresses act through Pkc1p, but that Bck1-20p may be a "slow" form of Bck1p, unable to respond to Pkc1p rapidly enough to activate Mpk1p following hypoosmotic shock.
Many MAPK pathways have been identified that respond to various forms of "stress" rather than to specific extracellular signaling molecules. These include the Sty1p pathway in S. pombe, the cell integrity and Hog1p pathways in S. cerevisiae, and the p38 and c-Jun N-terminal kinase pathways in animal cells. In the p38 and c-Jun N-terminal kinase pathways it is believed that regulation of upstream kinases underlies stress-mediated activation of the MAPKKK and hence the MAPK cascade (48). However, we are not aware of any comparable experiments to those performed here that would allow one to discriminate whether the signaling elements in those pathways are regulatory or simply necessary for basal pathway activity. It may be that the use of multiple "lateral" inputs into the different steps in these pathways, as described here for the cell integrity pathway, endows them with the flexibility to integrate information from several different stress sensors to determine the appropriate pathway output under different conditions.  Table I for strains listing) were grown to mid-log phase in YEPD supplemented with 1 M sorbitol at 30°C and analyzed for Mpk1p activity. Basal Mpk1p activity rose as more and more phosphatases were deleted. In this strain background, Msg5p appears to be the most potent phosphatase, as it can reduce basal Mpk1p activity to near wild-type levels even when the other three phosphatases are deleted. Basal levels of Mpk1p activity were very elevated upon deletion of all four phosphatases, even in media containing osmotic support to reduce pathway activity (although significantly more activity was detected when these cells were grown without osmotic support: data not shown). B, strains deleted for three phosphatases and containing Sdp1p (DLY6536), Msg5p (DLY6521), Ptp3p (DLY6508), or Ptp2p (DLY6523) as the only remaining phosphatase, as well as a strain lacking all four phosphatases (DLY6522, indicated by "none"), were tested for activation of Mpk1p following heat shock. Robust induction was observed in every case.