Reciprocal regulation between Slt2 MAPK and isoforms of Msg5 dual-specificity protein phosphatase modulates the yeast cell integrity pathway.

Dual-specificity protein phosphatases (DSPs) are involved in the negative regulation of mitogen-activated protein kinases (MAPKs) by dephosphorylating both threonine- and tyrosine-conserved residues located at the activation loop. Here we show that Msg5 DSP activity is essential for maintaining a low level of signaling through the cell integrity pathway in Saccharomyces cerevisiae. Consistent with a role of this phosphatase on cell wall physiology, cells lacking Msg5 displayed an increased sensitivity to the cell wall-interfering compound Congo Red. We have observed that the N-terminal non-catalytic region of this phosphatase was responsible for binding to the kinase domain of Slt2, the MAPK that operates in this pathway. In vivo and in vitro experiments revealed that both proteins act on each other. Msg5 bound and dephosphorylated activated Slt2. Reciprocally, Slt2 phosphorylated Msg5 as a consequence of the activation of the cell integrity pathway. In addition, alternative use of translation initiation sites at MSG5 resulted in two protein forms that are functional on Slt2 and became equally phosphorylated following activation of this MAPK. Under activating conditions, a decrease in the affinity between Msg5 and Slt2 was observed, leading us to suggest that the mechanism by which Slt2 controls the action of Msg5 was via the modulation of protein-protein interactions. Our results indicate the existence of posttranscriptional mechanisms of regulation of DSPs in yeast and provide new insights into the negative control of the cell integrity pathway.

Dual-specificity protein phosphatases (DSPs) are involved in the negative regulation of mitogen-activated protein kinases (MAPKs) by dephosphorylating both threonine-and tyrosine-conserved residues located at the activation loop. Here we show that Msg5 DSP activity is essential for maintaining a low level of signaling through the cell integrity pathway in Saccharomyces cerevisiae. Consistent with a role of this phosphatase on cell wall physiology, cells lacking Msg5 displayed an increased sensitivity to the cell wall-interfering compound Congo Red. We have observed that the N-terminal non-catalytic region of this phosphatase was responsible for binding to the kinase domain of Slt2, the MAPK that operates in this pathway. In vivo and in vitro experiments revealed that both proteins act on each other. Msg5 bound and dephosphorylated activated Slt2. Reciprocally, Slt2 phosphorylated Msg5 as a consequence of the activation of the cell integrity pathway. In addition, alternative use of translation initiation sites at MSG5 resulted in two protein forms that are functional on Slt2 and became equally phosphorylated following activation of this MAPK. Under activating conditions, a decrease in the affinity between Msg5 and Slt2 was observed, leading us to suggest that the mechanism by which Slt2 controls the action of Msg5 was via the modulation of protein-protein interactions. Our results indicate the existence of posttranscriptional mechanisms of regulation of DSPs in yeast and provide new insights into the negative control of the cell integrity pathway.
Signal transduction pathways that lead to the activation of mitogen-activated protein kinases (MAPKs) 1 are widely used by eukaryotic cells. Reversible protein phosphorylation is a key mechanism involved in the modulation of signaling through evolutionary conserved modules in which MAPKs are the final components. These proteins are activated through the phospho-rylation of conserved threonine and tyrosine residues in their phosphorylation lip by specific MAPK kinases (MAPKKs) (1).
Because both higher and lower eukaryotic cells share the components operating in these pathways, studies with model systems, particularly yeast, have been of great importance in elucidating the mechanisms through which MAPK-mediated signaling is regulated. The budding yeast Saccharomyces cerevisiae contains several MAPK cascades involved in the mediation of different physiological responses (for review see Ref. 2). Pheromones activate the Fus3 MAPK-mating module, which allows fusion of the mating partners. The filamentous growth pathway, which is activated by nutrient starvation, is mediated by the MAPK Kss1. In the high osmolarity glycerol (HOG) pathway, the MAPK Hog1 is activated to trigger the osmo-adaptive response. Different stimuli associated with cell surface alterations activate the Slt2 MAPK-mediated cell integrity pathway, which leads to an appropriate response oriented to maintaining cell wall stability (3,4). Recently, an additional pathway, which used the same MAPK module as the filamentous growth cascade, the STEvegetative growth pathway, has been proposed to be also involved in cell wall integrity (5).
Proteins and mechanisms that lead to MAPK activation have been extensively studied in the recent years. However, the duration and extent of MAPK activation not only relies on the activity of the upstream kinases, but also on protein phosphatases that dephosphorylate either or both of the Thr and Tyr residues in the TXY motif in MAPKs (6). The identity and regulation of the protein phosphatases that down-regulate these pathways are less known.
Frequently, the same protein phosphatase can act on different MAPKs. For example, the tyrosine phosphatases Ptp2 and Ptp3 have been shown to down-regulate the mating, the HOG, and the cell integrity pathways in S. cerevisiae by acting on Fus3, Hog1, and Slt2 MAPKs, respectively (7-10). Furthermore, the same protein phosphatase is able to regulate a pathway at different stages of the signaling process. This is again the case of Ptp2 and Ptp3, which seem to be responsible for the tyrosine dephosphorylation of these MAPKs to maintain a low basal activity and also to inactivate them following stimulation of the pathway, thereby playing a role in adaptation.
It is also remarkable that a single MAPK can be regulated by more than one class of protein phosphatases. Hog1 has also been shown to be inactivated by Ptc1, a type 2C Ser/Thr phosphatase, both in basal conditions and in adaptation (11). Although there is currently no evidence for direct dephosphorylation of Slt2 by Ptc1, a genetic interaction of PTC1 with the cell integrity pathway has been reported previously (12).
Ser/Thr phosphatases and Tyr phosphatases dephosphorylate either Thr or Tyr residues, which is sufficient for MAPK inactivation. However, to control the activity of MAPKs, eu-karyotic cells also use dual-specificity phosphatases (DSPs) that cleave phosphoester bonds in phospho-Tyr as well as in phospho-Thr and phospho-Ser residues. Although Ͼ10 DSPs have been identified in higher eukaryotes (13), only two DSPs have been shown to act on MAPKs in S. cerevisiae. One of them, Msg5, is a phosphatase that promotes adaptation to pheromone response by dephosphorylating Fus3 (10,14). Besides, the involvement of Msg5 in the maintenance of a low basal level of activity through the cell integrity pathway has been suggested because the disruption of MSG5 results in increased phospho-Slt2 levels under non-inducing conditions (15). The second one, Sdp1, has recently been shown to target Slt2, regulating the phosphorylation level of this MAPK in response to heat shock (16,17).
Thus, MAPK phosphatases are very versatile elements that must be perfectly regulated to ensure the appropriate inactivation of the different MAPKs and the correct moment for acting on each substrate. Some studies with cells from higher organisms have indicated the existence of posttranslational mechanisms that might regulate the action of MAPK-directed phosphatases (18 -20). However, to date no posttranslational modifications of these phosphatases have been described in S. cerevisiae, an organism that has been essential in obtaining current knowledge regarding MAPK signaling.
In an effort to gain insight into the mechanisms that modulate the action of the MAPKs phosphatases on their substrates, we have studied the role of the DSP Msg5 within the cell integrity pathway in S. cerevisiae. Here we report the existence of a reciprocal regulation between the MAPK Slt2 and various isoforms of Msg5 generated by alternative translational initiation and phosphorylation. These studies indicate the existence of different posttranscriptional mechanisms that operate on MAPK phosphatases in S. cerevisiae, reinforcing the importance of this yeast as a model for studying MAPK signaling in eukaryotic cells.

EXPERIMENTAL PROCEDURES
Strains and Yeast Genetic Methods-The S. cerevisiae strains used in this study are listed in Table I. Standard procedures were employed for yeast genetic manipulations (21).
DNA Manipulation and Plasmids-General DNA methods were performed using standard techniques (22). To obtain a C-terminal Myc 6tagged version of Slt2, plasmid pRS305SLT2m was constructed. The SLT2 open reading frame was amplified with primers 5Ј-CCCGGATC-CGTCGACTTGAGGAGAATT-3Ј and 5Ј-CCCGAATTCGCGGCCGCAA-AAAAAATATTTTCTATCTAATC-3Ј (BamHI and EcoRI sites are underlined) and subcloned into pGFP-C-FUS (23). It was digested with HindIII, and the resulting fragment was introduced into pRS305m to produce pRS305SLT2m. pRS305m was obtained by subcloning a Myc 6 epitope into the integrative LEU2 vector pRS305 (24). pRS305SLT2m expressed a functional Slt2-Myc 6 protein because it suppressed the lytic phenotype of slt2 mutants. To integrate SLT2-Myc 6 into the SLT2 locus, pRS305SLT2m was linearized at the unique PstI site.
To construct pRS305MSG5m, the MSG5 open reading frame was PCR-amplified with primers MSGU 5Ј-CCCCCCGGGATCCATGCAAT-TTCACTCAG-3Ј and MSGD 5Ј-CCCCCCGGGGATCCAGGAAGAAAC-ATCATCTG-3Ј (BamHI site is underlined), subcloned into the vector pGEM-T (Promega) to afford pGEMT-MSG5, and sequenced to confirm the absence of mutations. It was digested with BamHI, and the DNA fragment bearing MSG5 was introduced into pRS305m. To integrate MSG5-Myc 6 into the MSG5 locus, pRS305MSG5m was linearized at the unique SphI site.
To construct YCplac22MSG5m, PROMMSG5 5Ј-CCGGATCCGTAG-TGATGGATAATGTG-3Ј (BamHI site is underlined) and INT2MSG5 5Ј-CCGGCGATGGGAATTTTAC-3Ј were used to amplify a fragment spanning from 1 kb upstream to 140 bp downstream from the initiation codon of MSG5, cloned into pGEM-T to afford pGEMT-PMSG5, and sequenced to verify the absence of mutations. EcoRI fragment from pRS305MSG5m then was inserted into pGEMT-PMSG5 to generate MSG5 with promoter sequences. The BamHI-SacI DNA fragment from pRS305MSG5m carrying the Myc 6 epitope and the BamHI fragment bearing MSG5 were sequentially subcloned into YCplac22 (CEN4, TRP1) (25) to give YCplac22MSG5m.
To obtain the mutant allele MSG5-M1A, primers 5Ј-GCGATAAGT-GCACGCGCAATTTCACTCAGATAAGC-3 and 5Ј-GCTTATCTGAGTG-AAATTGCGCGTGCACTTATCGC-3Ј were used. To generate MSG5-M-45A, primers 5Ј-CGATGAGAATTCCGCTAATGGATGGAGTG-3 and 5Ј-CACTCCATCCATTAGCGGAATTCTCATCG-3Ј were used. In both cases, PCR amplification from plasmid YCplac22MSG5m was performed as described previously (26). msg5C319A was obtained by mutagenic PCR in two parts using the MSGU with the mutagenic oligonucleotide 5Ј-CTCCACACTGAGCGTGTACGAGTAT-3Ј and the mutagenic oligonucleotide 5Ј-ATACTCGTACACGCTCAGTGTGGAG-3Ј with MSGD. Products were combined, and a second amplification was performed using MSGU and MSGD. The product was subcloned into pRS305m to afford pRS305MSG5C319Am. YCplac22MSG5C319Am was obtained by replacing the StuI fragment of YCplac22MSG5 by the same fragment from pRS305MSG5C319Am.
␤-Galactosidase Assays-␤-Galactosidase activities were determined according to Guarente (31). Values are averages of at least three independent transformants assayed in duplicate.
In Vivo Binding Assays-Cells bearing pEGKG-derived plasmids were grown in SG medium lacking uracil and containing 2% galactose at the indicated temperature. Cells were collected and lysed as above in lysis buffer lacking SDS and Nonidet P-40. The lysates were clarified and incubated with glutathione-Sepharose beads (Amersham Biosciences) for 2 h. Beads were washed extensively with the same buffer and resuspended in SDS loading buffer, and proteins were analyzed by SDS-PAGE and immunoblotting.
Expression and Purification of Recombinant GST, GST-Msg5, and GST-Msg5C319A in Escherichia coli-E. coli DH5␣ cells transformed with the plasmid pGEX-KG, pGEX-MSG5, or pGEX-MSG5C319A were grown in LB with ampicillin to A 600 ϭ 0.5, and then isopropyl-1-thio-␤-D-galactopyranoside was added to a final concentration of 0.4 mM for 3 h. Protein was isolated from overexpressing cells as described previously (32).
Alkaline Phosphatase Treatment of Msg5-Myc 6 -Msg5-Myc 6 protein from cells incubated for 1 h at 39°C was immunoprecipitated with anti-Myc antibodies as described above. The immunoprecipitates were washed three times with IPP buffer and three times with calf intestine alkaline phosphatase buffer (Roche Diagnostics) and resuspended in this buffer with or without 1 mM sodium orthovanadate. Calf intestine alkaline phosphatase then was added, and the reaction was incubated at 37°C for 20 min and stopped by the addition of SDS-sample loading buffer.

Msg5
Phosphatase Activity Regulates the Activation of the Cell Integrity Pathway-We have previously shown that deletion of MSG5 resulted in increased levels of phospho-Slt2 as revealed by Western blotting using antibodies that detect dually phosphorylated and thereby activated Slt2 (15). To gain insight into the molecular mechanisms used by Msg5 to operate in the cell integrity pathway, we investigated whether the observed down-regulating effect of Msg5 on Slt2 phosphorylation was the result of the phosphatase activity of this protein.
To this end, the codon for the conserved cysteine residue at the catalytic site (Cys-319) of MSG5, essential for phosphatase activity (14), was mutated to encode alanine. As shown in Fig.  1A, cells bearing the msg5C319A allele in a centromeric plasmid, growing either at 24 or 39°C, displayed phospho-Slt2 levels similar to those of null msg5⌬ cells. Therefore, the negative regulation exerted by Msg5 on Slt2 must be because of its phosphatase activity. Consistent with our previous results (15), the impact of Msg5 on Slt2 phosphorylation is higher at 24°C than at 39°C (Fig. 1A), indicating that Msg5 plays a more important role in down-regulating Slt2 under basal conditions than following induction.
We next speculated that the Slt2 up-regulation promoted by the lack of Msg5 could have phenotypic consequences related to cell wall integrity. Fig. 1B shows that the absence of Msg5 leads to an increased sensitivity of cells to Congo Red, a compound that interferes with cell wall assembly (33). Consistent with a role for Msg5 in cell wall construction, we also observed that msg5⌬ cells transformed with a FKS2-lacZ reporter displayed double the ␤-galactosidase activity of the wild type cells (data not shown), reflecting the transcriptional induction of FKS2, a known output of the cell integrity pathway (34). These studies support the notion that Msg5 regulates cell wall physiology, and taken together, all of these findings indicate that this phosphatase is a key regulator of the cell integrity pathway.
Two-hybrid Analysis of the Msg5 and Slt2 Interaction-We next wished to know whether Slt2 and Msg5 interact in vivo. To this end, we used the two-hybrid system. Slt2 was fused to the Gal4 transcriptional activation domain, and its interaction with Msg5 fused to the Gal4 DNA-binding domain was tested. Cells containing both fusion proteins exhibited a reproducible 30-fold increase in ␤-galactosidase activity (Fig. 2). This activity was triple the magnitude of that observed for the interaction between Slt2 and its direct activator, the MAPKK Mkk1. Thus, Slt2 and Msg5 are capable of physically associating in vivo, suggesting that Msg5 might directly dephosphorylate Slt2.
We extended these studies by determining the regions of Slt2 and Msg5 that are essential for their interaction. In the case of Slt2, the protein kinase domain was located at the N terminus, whereas the catalytic active site of Msg5, similar to other DSPs, was localized within the C-terminal half of the protein.
The function of the Slt2 C-terminal and Msg5 N-terminal regions is currently unknown, although they are assumed to perform a regulatory function. As shown in Fig. 2, the Slt2 N-terminal region was sufficient for binding to Msg5 and the N-terminal domain of Msg5 was the one involved in the association with Slt2.
Msg5 Dephosphorylates Slt2 in Vitro-The results described above suggested that Msg5 inactivates Slt2 directly. To test this possibility, we studied whether recombinant Msg5 was able to dephosphorylate Slt2 in vitro. Wild type Msg5 and the catalytically inactive Msg5C319A forms were tagged with GST at the N terminus and expressed in E. coli. As expected, purified GST-Msg5 but not GST-Msg5C319A showed phosphatase activity toward the substrate p-nitrophenyl phosphate, indicating that the GST-Msg5 obtained was active (data not shown). Recombinant GST-Msg5 but not GST dephosphorylated GST-Slt2 that had been purified from heat-shocked yeast cells (Fig.  3). The mutant form GST-Msg5C319A retained some activity on Slt2 although much less pronounced than the wild type protein (Fig. 3). This result is similar to that previously obtained when the same mutant protein was assayed for its activity on the MAPK Fus3 (14). All of these findings indicate that phosphatase activity of Msg5 is responsible for the ob-served Slt2 dephosphorylation in vitro, suggesting that in vivo Msg5 would act as a negative regulator of the cell integrity pathway by directly inactivating Slt2.
Alternative Translational Initiation Sites Result in Two Different Forms of Msg5 That Are Active and Phosphorylated following Heat Shock-In contrast to that observed following stimulation of the pheromone pathway, MSG5 transcription does not seem to be increased under conditions of cell integrity pathway activation (14,16,35). Therefore, we analyzed whether Msg5 might be somehow regulated by posttranscriptional modifications. To facilitate the analysis of Msg5 protein, the chromosomal MSG5 gene was C-terminally tagged with a sequence encoding six copies of the Myc epitope. As shown in Fig. 4A, Western blotting analysis revealed that Msg5-Myc 6 from wild-type cells growing at 24°C migrated as two major bands of similar intensity. One of these bands corresponded to the expected mobility in SDS-PAGE for Msg5-Myc 6 with an apparent molecular mass of 91 kDa. The other predominant band had an apparent molecular mass of 85 kDa. Interestingly, the shifting of cells to 39°C resulted in a progressive disappearance of these two Msg5 bands and a concomitant accumulation of two forms of lower mobility (100 and 107 kDa) (Fig.  4A). An increase in the phospho-Slt2 level along the time, characteristic of the heat-mediated activation of the cell integrity pathway, was also observed (Fig. 4A). Similar results were found in different yeast strains (data not shown), indicating that these phenomena were not background-specific. These intriguing observations prompted us to study the reason for the Msg5 mobility shift and the nature of the Msg5 doublet.
First, we investigated whether the heat shock-dependent mobility shift of Msg5 could be attributed to phosphorylation. The treatment with calf intestine alkaline phosphatase, a nonspecific protein phosphatase, transformed the immunoprecipitated slower migrating forms of Msg5, which appeared in cells at 39°C, into the fastest migrating Msg5 species obtained from non-stressed cells (Fig. 4B). These data show that Msg5 is phosphorylated in vivo in response to heat shock, which causes an upward mobility shift of this protein in SDS-PAGE.
The two Msg5 forms observed in cells growing at 24°C displayed a mobility in SDS-PAGE corresponding to proteins that differed by Ϸ6 kDa, the expected molecular mass of the predicted peptide sequence between the two first methionines (Met 1 and Met 45 ) in the Msg5 primary sequence (Fig. 4C). Therefore, we addressed the question of whether these two Msg5 forms could be the consequence of alternative translational initiation sites, one of them corresponding to the expected initiation codon and the other corresponding to the second ATG present in the MSG5 coding sequence. As shown in  In contrast, when the residue Met 45 was mutated, the lowest band (85 kDa) was not observed. These findings indicate that, together with the predicted full-length Msg5, a Msg5 form lacking the first 44 amino acids is also present in the cells. They also strongly suggest that the two Msg5 forms, which henceforth we shall refer as Msg5 L (long) and Msg5 S (short), would be attributed to the existence of two different translational initiation sites. When the same cells were transferred to 39°C, Msg5 L migrated as a main band of a mobility of 107 kDa and Msg5 S migrated as a 100-kDa protein band (Fig. 4C), suggesting that the two Msg5 forms are similarly phosphorylated following heat shock.
Finally, we also studied the functionality of the two Msg5 forms with regard to their ability to negatively modulate Slt2. At 24°C, the absence of Msg5 led to a high level of Slt2 phosphorylation, whereas the presence of either of the two Msg5 forms reduced the amount of phospho-Slt2 to the wildtype level (Fig. 4C). At high temperature, both Msg5 forms were also similarly functional in dephosphorylating Slt2, again leading to wild-type phospho-Slt2 levels. Therefore, this finding indicates that both forms are active against Slt2 to a similar extent under the conditions tested.
Collectively, all of these findings indicate that Msg5 is not a single species. Two main forms of Msg5 exist. Both are functional in dephosphorylating Slt2, and both are modified by phosphorylation during heat stress.
Slt2 Phosphorylates Msg5 under Conditions of Cell Integrity Pathway Activation-We reasoned that if Msg5 phosphorylation was a result of activation of the cell integrity pathway, other stimuli that activate this pathway could also lead to this modification. As shown in Fig. 5A, treatment of cells expressing Msg5-Myc 6 with Calcufluor White, an agent that has been reported to activate the Slt2-mediated pathway (4), led to an increase in the proportion of the phosphorylated isoforms of Msg5-Myc 6 . In contrast, Msg5 mobility was not modified following hyperosmotic stress, a condition known to activate the HOG pathway (data not shown).
It has been established that Slt2 tyrosine phosphorylation was induced during bud formation and repressed around the time of mitosis (36). We tested Slt2 and Msg5 phosphorylation through the cell cycle by synchronizing a dbf2 ts mutant strain, which arrested in telophase upon being shifted to the nonpermissive temperature (37). After cells were allowed to reenter the cell cycle by incubating them at 24°C, dual Slt2 phosphorylation increased just before cells began to bud and it was maintained during bud emergence and growth and transiently decreased at mitosis as Clb2 levels rose (Fig. 5B). This pattern of Slt2 phosphorylation resembles that previously reported by Zarzov et al. (36) when synchronized cells were grown at 35°C, whereas these authors found that tyrosine phosphorylation is limited to a significantly short period at the beginning of bud formation at 30°C. This discrepancy could be explained in terms of differences in strain background and methodology to obtain and detect Slt2 protein and phosphorylation. Interestingly, as shown in Fig. 5B, the amount of phosphorylated Msg5 varied concomitantly with Slt2 phosphorylation. The Msg5 slower migrating bands were visible over a large part of the cell cycle, but they were greatly reduced around the time of mitosis. Taken together, all of these findings suggest that Msg5 is specifically phosphorylated in response to stimuli that trigger the cell integrity pathway and that this modification is mediated by this pathway.
These observations supported the idea that Slt2 could be the kinase that phosphorylates Msg5. To check this possibility, we analyzed Msg5 mobility at 24°C and after heat stress or Calcofluor White treatment in wild type, slt2⌬ mutant cells, and in these cells transformed with a centromeric vector bearing either wild-type SLT2 or the catalytically inactive allele slt2K54F (30). As shown in Fig. 5C, Msg5-Myc 6 in cells carrying wild-type SLT2 underwent the expected upwards mobility shift after stimula- FIG. 4. Msg5 is produced as two forms that are phosphorylated after heat shock treatment and show the same activity on Slt2. A, analysis of the Msg5-Myc 6 mobility in SDS-PAGE. The Msg5-Myc 6tagged YMF-1 strain was grown in YPD medium to mid-log phase at 24°C and then heat-shocked at 39°C. Aliquots of the same culture were withdrawn before (0) and at the indicated times following induction. Proteins from extracts were separated by SDS-PAGE and analyzed by Western blotting with both anti-Myc and anti-phospho-p42/44 antibodies. B, alkaline phosphatase treatment of Msg5-Myc 6 from heat-shocked cells. Immunoprecipitated Msg5-Myc 6 from the YMF1 strain grown at 24 or 39°C was treated with (ϩ) or without (Ϫ) alkaline phosphatase from calf intestine in the absence (Ϫ) or presence (ϩ) of sodium orthovanadate (Na 3 VO 4 ). Msg5 isoforms were detected by immunoblotting with anti-Myc antibody. C, ATG mutation constructs of MSG5 and Western blotting analysis of the generated Msg5-Myc 6 mutant forms. YCplac22m (Vector), YCplac22MSG5m (Msg5), YCplac22MSG5-M1 (Msg5M1A), and YCplac22MSG5-M45A (Msg5M45A) plasmids were transformed into the msg5⌬ (DD1-2D) strain. Exponential cultures growing in selective medium were incubated at 24°C and heat-shocked at 39°C for 2 h. Protein extracts were prepared, and anti-Myc and anti-phospho-Slt2 immunoblot analysis was performed. tion, whereas this phenomenon was not observed in cells lacking Slt2 or bearing the inactive MAPK. A very faint signal that could correspond to phosphorylated Msg5 can be hardly observed in slt2 mutants only after thermal stress. This could be the result of a residual Msg5 phosphorylation by other protein kinases in response to this stimulus, which is known to trigger different cellular responses. It is also interesting to note that the nonphosphorylated Msg5 short form almost disappeared in extracts from cells growing at 39°C, suggesting that this form is less stable than the long one under this stress. Furthermore, a decrease in the overall amount of the phosphatase occurred in both activating conditions. The results described above indicate that Msg5 phosphorylation following activation of the cell integrity pathway depends on a functional Slt2, suggesting that this protein directly phosphorylates Msg5. To confirm this finding, an in vitro kinase assay was developed using E. coli produced GST-Msg5C319A as substrate to prevent an autodephosphorylation reaction. As shown in Fig. 6, phosphorylation of recombinant GST-Msg5C319A occurred in the reaction with precipitated GST-Slt2 from heat-shocked yeast cells but not in similar reactions with precipitated GST or the kinase-dead form, GST-Slt2K54F. Furthermore, no phosphorylation was detected when GST was used as a substrate. Taken together with the previous results, these findings provide strong evidence that Slt2 is the protein kinase that phosphorylates Msg5 under conditions of cell integrity pathway activation.
Binding of Msg5 to Slt2 Is Reduced at 39°C-We wished to determine whether activation of the cell integrity pathway altered the affinity between Msg5 and Slt2. To test this possibility, we performed co-purification experiments from extracts of cells co-expressing GST-Msg5 and Slt2-Myc 6 . GST-Msg5 protein complexes were purified, and Slt2-Myc 6 in precipitates was probed using anti-Myc antibodies. As shown in Fig. 7A, GST-Msg5, but not GST, precipitated Slt2-Myc 6 . Interestingly, when cells had been previously incubated at 39°C, a much smaller amount of Slt2-Myc 6 was bound to GST-Msg5. These differences in binding were not due to different Slt2-Myc 6 levels in cell extracts because they were the same in both conditions (Fig. 7A). These results confirm the interaction data obtained by two-hybrid analysis and indicate that this interaction is reduced at high temperature.
To confirm and extend these results, we performed a complementary experiment by purifying GST-Slt2 and assaying for the presence of Msg5-Myc 6 . Results identical to the above were obtained in which Msg5-Myc 6 co-purified with GST-Slt2 but not with GST (Fig. 7B). When cells were grown at 24°C, the two main non-phosphorylated forms of Msg5 interacted with Slt2 to a similar extent. However, the association of the phosphorylated forms of Msg5 with Slt2 was strongly reduced in extracts from YCplac22MSG5m was transformed into the dbf2 MGY21b mutant strain. Cells growing exponentially at 24°C were shifted to 37°C for 3 h. Cells were checked for arresting in mitosis and then shifted back to 24°C for them to enter a synchronous division cycle. At the indicated time points, cells were collected, protein extracts were analyzed by immunoblotting with anti-Myc, anti-phospho-p42/44, anti-Slt2, and anti-Clb2, and the numbers of unbudded, small budded (with a bud smaller than half of the size of the mother cell), and large budded cells were counted. Because dbf2 ts mutant cells arrested in mitosis at 37°C do not separate once they resume growth at permissive temperature, initial doublets were considered as two independent cells. C, effect of the Slt2 activity on Msg5 phosphorylation. Exponentially growing cells of the wild-type 1783 (WT) and DL454 (slt2⌬) strains and the latter strain transformed with pHR0 (SLT2) or pHR3 (slt2K54F) plasmids at 24°C, heat-shocked at 39°C for 2 h (upper panel), and treated with Calcofluor White (50 g/ml) for 1 h (lower panel) were collected, protein extracts were prepared, and immunoblot analysis was performed with anti-Myc, anti-phospho-p42/44, and anti-actin antibodies. cells growing at 39°C. These results confirm that heat stress induces dissociation of the MAPK-phosphatase complex. DISCUSSION With a view to shedding light on the negative regulation exerted by protein phosphatases on MAPK-mediated signaling, we attempted to gain insight into the role of the DSP Msg5 on the cell integrity pathway that operates in S. cerevisiae. The data reported here show that Msg5 regulates this pathway by directly dephosphorylating the Slt2 MAPK. In all of the strains tested, the exploration of dual Slt2 phosphorylation before and during thermal treatment in a wild type and a msg5 strain revealed that Msg5 played a fundamental role in down-regulating the basal activity of Slt2. The absence of either Ptp2 or Ptp3 does not lead to a significant increase in the amount of basal dually phosphorylated Slt2 (9,16). Furthermore, the dual-specificity protein phosphatase Sdp1 does not seem to be involved in maintaining a low basal Slt2 phosphorylation (16). Therefore, so far, it seems that Msg5 is the most important protein phosphatase involved in maintaining Slt2 with a low activity in the absence of stimulation.
The role of Msg5 in modulating signaling through the cell integrity pathway was not only suggested by the above-described experiments but also by the induction of FKS2 transcription and the increased sensitivity to Congo red of msg5 mutant cells. This is a typical phenotype indicative of an affected cell wall, probably derived from an altered Slt2-mediated response. In fact, some cell wall-related mutants that display a constitutive high level of Slt2 phosphorylation, such as gas1, ecm33, itc1, or rot2, have been shown to be hypersensitive to Calcofluor White, a compound that interferes with cell wall assembly in a similar way to Congo Red (4,38). In agreement with a role for MSG5 in modulating cell wall-related cellular signaling, this gene has recently been isolated as a multicopy suppressor of the Calcofluor White sensitivity of cells overexpressing KIC1, a gene encoding a protein kinase involved in cell wall integrity (39).
The data presented here also shed some light onto the mechanisms that mediate the interaction between MAPKs and MAPK-protein phosphatases. The N-terminal non-catalytic region of DSPs has been shown to be responsible for binding to MAPKs (40). Most mammalian DSPs possess two motifs designated CH2 (Cdc25 homology regions) at the N-terminal half of the protein. Although the functional significance of these motifs remains unknown, it has been proposed that they could be involved in the association with MAPKs (41). In support of this hypothesis, the yeast tyrosine phosphatase Ptp3 has been shown to interact with the MAPK Fus3 through the CH2 domains (42). Recently, a basic motif-based docking domain, present in different types of MAPK-interacting proteins, has been shown to be involved in their association with MAPKs (43). This domain is composed of three motifs: a cluster of basic residues, a LXL element, and a cluster of hydrophobic residues, although the presence of two of these submotifs has been shown to be enough for binding to MAPKs (44). Msg5 does not present CH2 domains (10) but in silico analysis has revealed that the two forms of the phosphatase exhibited a domain of amino acid similarity to docking domains at the N-terminal region. This Msg5 domain presents a basic motif (Arg 96 -Arg 97 ) followed by a cluster of hydrophobic residues (IYTLPTSL, residues 102-109). The responsibility of the N-terminal part of this protein in binding to Slt2 suggests the involvement of this putative D-domain of Msg5 in the association to this MAPK and the dispensability of CH2 domains within the N-terminal non-catalytic regions of DSPs phosphatases for binding to MAPKs.
Adaptation to the stimulus is a typical feature of signal transduction systems. In most MAPK-mediated signaling pathways, several desensitization or adaptation mechanisms operate to progressively reduce the intensity of the response. Among them, a reduction of MAPK activation by DSPs seems to play an important role. Following activation of the signaling pathways, expression of DSPs is usually induced, indicating their involvement in negative feedback loops regulating MAPK activity (41). In yeast, consistent with its role in adaptation of the mating pathway, Msg5 transcription is up-regulated in response to pheromone treatment (14). Although the existence of adaptation in the cell integrity pathway has been reported in a particular background (9), in most of the strains the level of phosphorylated Slt2 remained high and does not decrease for several hours upon stim-  ulation (15,36). The absence of an increase in MSG5 transcription following heat shock is consistent with the lack of a Msg5mediated negative feedback mechanism in this signaling pathway. Moreover, it has been reported that even a transient decrease in MSG5 transcription occurs after thermal stress (16). In agreement with this finding, a decrease in the overall amount of Msg5 after stimulation of the cell integrity pathway can be observed in Figs. 4 and 5.
The absence of adaptation in the cell integrity pathway in most strains could be explained in terms of the need for cell wall remodeling as long as the cell surface alteration was present. This necessity of a high Slt2-sustained activation could be the reason for the decreased binding of Msg5 to Slt2 upon stimulation of the pathway. Phosphorylation of the mammalian tyrosine phosphatase HePTP by ERK2 has been shown to reduce the affinity of both proteins (18). Therefore, it is very tempting to speculate that the phosphorylation of Msg5 by activated Slt2 would be the molecular mechanism responsible for lowering the affinity of both proteins, increasing Slt2 activation and leading to amplification of the signal. The observation that ERK2 binding to the N-terminal regulatory region of the DSP MKP-3 results in the catalytic activation of the protein phosphatase to rapidly inactivate the bound MAPK (45) suggests that this could also be a conserved mechanism operating in DSPs. Therefore, the decrease in the affinity between Slt2 and Msg5 would not only reduce the amount of phosphatase able to act on Slt2 but might also prevent the activation of Msg5. Taken together, our findings support for the first time a model of reciprocal regulation between a yeast MAPK and its DSP in which the MAPK is modulating by phosphorylation the intensity of the DSP action on the MAPK.
Besides its action on Slt2, Msg5 has been shown to regulate signaling through the mating pathway by dephosphorylating Fus3. Therefore, it would be the first yeast DSP shown to act on different MAPKs. In addition, overexpression of MSG5 hinders the phosphorylation of Kss1 following hyperosmotic shock (46) and its deletion results in increased levels of phospho-Kss1, 2 suggesting that Msg5 could also regulate Kss1 activity. As in the case of other protein phosphatases, one of most puzzling questions concerning Msg5 is how the same protein can selectively participate in the regulation of different cellular processes governed by distinct MAPKs. The production of two Msg5 forms might lead to functionally distinct molecules that could act specifically on different MAPKs. As precedents, alternative initiation of translation has been shown to yield different isoforms of the transcription protein JunD or the human caspase-2 (47,48). Although no functional differences have been reported among these isoforms in higher eukaryotes, differences in protein localization have been observed between the two isoforms of the yeast tRNA modification enzyme Mod5 and the glucoamylase Sta2 (49,50). Furthermore, the modification of Msg5 forms by phosphorylation could also contribute to provide substrate specificity. For example, following pheromone stimulation, both Slt2 and Fus3 are activated and MSG5 transcription is up-regulated. As a consequence, Fus3 activation is reduced as part of the adaptation process (10,14). However, Slt2 phosphorylation is maintained (36), suggesting that Msg5 is only acting on Fus3 in such conditions. Therefore, how can Msg5 act selectively on Fus3 but not on Slt2? An exciting possibility is that the phosphorylation of Msg5 might prevent it from acting on Slt2, as stated above, but not on Fus3. Further investigation will determine how these biochemically distinguishable DSP species contribute to modulate signaling through the different MAPK pathways.