Two Protein-tyrosine Phosphatases Inactivate the Osmotic Stress Response Pathway in Yeast by Targeting the Mitogen-activated Protein Kinase, Hog1*

Protein phosphatases inactivate mitogen-activated protein kinase (MAPK) signaling pathways by dephosphorylating components of the MAPK cascade. Two genes encoding protein-tyrosine phosphatases, PTP2, and a new phosphatase,PTP3, have been isolated in a genetic selection for negative regulators of an osmotic stress response pathway called HOG, for high osmolarity glycerol, in budding yeast. PTP2 and PTP3 were isolated as multicopy suppressors of a severe growth defect due to hyperactivation of the HOG pathway. Phosphatase activity is required for suppression since mutation of the catalytic Cys residue in Ptp2 and Ptp3, destroys suppressor function and biochemical activity. The substrate of these phosphatases is likely to be the MAPK, Hog1. Catalytically inactive Ptp2 and Ptp3 coprecipitate with Hog1 from yeast extracts. In addition, strains lacking PTP2 and PTP3 do not dephosphorylate Hog1-phosphotyrosine as well as wild type. The latter suggests that PTP2 and PTP3 play a role in adaptation. Consistent with this role, osmotic stress induces expression of PTP2 and PTP3 transcripts in a Hog1-dependent manner. Thus Ptp2 and Ptp3 likely act in a negative feedback loop to inactivate Hog1.

MAPK 1 signaling is ubiquitous among eukaryotes and regulates a variety of processes. In metazoans and in yeast, MAPK pathways regulate growth, development, and the response to stress (reviewed in Refs. [1][2][3][4][5]. In Drosophila, Caenorhabditis elegans, and vertebrates, receptor tyrosine kinases, acting through Ras and Raf, activate MEK and MAPK, regulating growth and development. In vertebrates, JNK/SAPK, and p38, kinases similar to MAPK, regulate stress responses. In Saccharomyces cerevisiae, six different MAPK signaling pathways regulate mating, pseudohyphal growth, invasiveness, cell wall biosynthesis, the response to osmotic stress, and spore wall formation. Common to all MAPK pathways are two sequentially acting kinases called MEK and MAPK, termed the MAPK module. MEK is activated by phosphorylation of Ser and Thr residues. MEK activates MAPK by phosphorylating a Thr and Tyr residue in a region called the phosphorylation lip. Although mechanisms that activate MAPK pathways have been well characterized, mechanisms that inactivate these pathways are not as well understood. Protein phosphatases negatively regulate MAPK pathways, but the identity of the physiologically relevant phosphatases and their targets is unclear. MEK is inactivated in vitro by the vertebrate Ser/Thr phosphatase, PP2A (6), but it is not established that this occurs in vivo. More is known about the inactivation of MAPK. Since MAPKs require phosphorylation of both a Thr and a Tyr residue for activity, dephosphorylation of either residue results in inactivation. Consistent with this, three different types of phosphatases including dual specificity phosphatases, Ser/Thr phosphatases, and protein-tyrosine phosphatases (PTPs) have been shown to inactivate MAPKs (reviewed in Refs. [7][8][9][10]. Dual specificity phosphatases in vertebrates and S. cerevisiae inactivate MAPKs in vitro and in vivo. The vertebrate MKP-1 inactivates ERK1 and ERK2 in vitro, but it is not certain this occurs in vivo (11)(12)(13)(14). The yeast Msg5 inactivates Fus3 in vitro and in vivo (15). The vertebrate Ser/Thr phosphatase PP2A inactivates MAPK in vitro (16), but it has not been established that this occurs in vivo. Cells treated with okadaic acid, an inhibitor of PP2A, activate MAPK in vivo (17). Since okadaic acid also activates MEK, it is not clear that PP2A inactivates MAPK in vivo (13,17,18). Two protein-tyrosine phosphatases, Pyp1 and Pyp2 in Schizosaccharomyces pombe, inactivate Sty1/ Spc1 in vitro and in vivo (19 -21). In S. cerevisiae, a putative protein-tyrosine phosphatase encoded by PTP2 (22)(23)(24) negatively regulates the osmotic stress response pathway, and indirect evidence suggests this occurs by dephosphorylation of Hog1-phosphotyrosine (Hog1-Tyr(P)) (25).
We sought to examine further the regulation of MAPK pathways by identifying and characterizing protein phosphatases that act on the HOG pathway in S. cerevisiae. This pathway allows yeast to grow in high osmolarity environments by inducing the expression of osmoprotectants via activation of the MAPK module, Pbs2-Hog1 ( Fig. 1) (26). Upstream of the MAPK module is a negative regulator, the "two-component system," comprised of three sequentially acting kinases including Sln1, a plasma membrane bound histidine/aspartyl kinase, Ypd1, a histidine kinase, and Ssk1, an aspartyl kinase (25,27,28). These kinases negatively regulate two MEKKs called Ssk2 and Ssk22 (29). There is also a positive regulator upstream of the MAPK module called Sho1 which activates Pbs2 directly (29). The model for activation of this pathway is as follows. Osmotic shock inactivates the two-component system, resulting in activation of the MEKKs that in turn activate the MAPK module. Osmotic shock activates Sho1 which directly activates Pbs2 via an SH3 domain.
Under standard growth conditions, mutational inactivation of the negative regulator, SLN1, constitutively activates the HOG pathway and is lethal on rich media and grows extremely poorly on less rich, synthetic media (27). Using this phenotype, we identified negative regulators as genes which, when overexpressed from a multicopy plasmid, restore viability. In addition to PTP2, we identified a second gene, PTP3, that inactivates this pathway by targeting Hog1. PTP2 and PTP3 are transcriptionally regulated; activation of the HOG pathway induces expression of PTP2 and PTP3, providing a mechanism for adaptation to signal.

MATERIALS AND METHODS
Strains, Media, and Genetic Techniques-To identify suppressors of the HOG pathway, a yeast genomic library was transformed into IMY101, a strain bearing a SLN1 deletion, sln1-⌬1:HIS3, and a low copy, CEN-based plasmid containing the wild type SLN1 gene and the URA3 gene (27). Yeast transformations were performed as described by Dohmen et al. (30). To follow the level of Hog1-Tyr(P), haploids derived from DF5 (31) were transformed with 2 m HOG1, a multicopy 2-mbased plasmid bearing HOG1 and TRP1, kindly provided by M. Gustin (26). The strains bearing 2 m HOG1 are BBY48, a wild type strain (32), IMY21a, a ptp2⌬ strain (22), HF2, a ptp3⌬ strain, and HFY6, a ptp2⌬ ptp3⌬ strain. For biochemical analysis of Ptp2 and Ptp3, glutathione S-transferase (GST)-PTP fusion proteins were expressed in RS334, kindly provided by R. Sclafani (33). To test whether the GST-PTP fusions were functional in vivo, the pGST-PTP plasmids (see below) were transformed into TJY1, which bears the SLN1 deletion, sln1-⌬1::HIS3, and pSLN1-URA3. TJY1 is derived from JD51, a galactose-inducible strain (34). A strain bearing a disruption of the PTC1/ TPD1 gene was produced by transformation of DF5 with the tpd1::LEU2-3 allele, kindly provided by J. Broach (35). The tpd1::LEU2-3/TPD1 diploids were sporulated and tetrads dissected, resulting in a 2:2 segregation of wild type and temperature-sensitive spore clones as reported (35). Media to culture yeast and bacteria were produced essentially as described by Sherman et al. (36). Standard rich media refers to YPD, and high osmolarity media refers to YPD containing 0.9 M sodium chloride unless otherwise noted.
Isolation and Deletion of PTP3-PTP3 was isolated as a negative regulator of the HOG pathway by selecting for plasmids from a yeast genomic library that suppressed lethality of the sln1⌬ strain, IMY101. This strain was transformed with a multicopy yeast genomic library based in the vector YEp13 (37) (American Type Culture Collection). Transformants capable of vigorous growth on 5-FOA were identified, and the plasmid DNA was isolated using standard methods. In addition to plasmids bearing PTP2, a different plasmid, pAF12, bearing an ϳ8-kb insert was isolated multiple times. Deletion of a SphI fragment or a HindIII fragment from pAF12 identified regions of the insert critical for suppression. Sequencing adjacent to the SphI site identified an open reading frame, Yer075p, sequenced by the genome project. We call this gene PTP3.
A strain bearing a deletion of the PTP3 gene was produced by transformation of a wild type diploid, DF5, with a PTP3 deletion construct. To produce the ptp3-⌬1::TRP1 allele, PCR was used to introduce a BamHI site 275 bp upstream of the start site and a SmaI site 118 bp downstream of the start codon, generating a 400-bp fragment corresponding to the 5Ј end of the gene. A fragment corresponding to the 3Ј end of the gene was produced using PCR to generate a SmaI site 2644 bp downstream of the start site and a EcoRI site 985 bp downstream of the stop codon. Both fragments were simultaneously ligated into pUC19 digested with BamHI and EcoRI to generate the plasmid pHF1. This plasmid was digested with SmaI, and a 850-bp blunt-ended EcoRI-BglII fragment of TRP1 (38) was ligated to generate pUC19-ptp3⌬::TRP1. This plasmid was digested with EcoRI and BamHI and transformed into DF5, and Trp ϩ transformants were selected. Southern analysis identified transformants bearing the deletion allele at the PTP3 locus (39). Briefly, genomic DNA was digested with XbaI and probed with the 1-kb SmaI-EcoRI fragment corresponding to the 3Ј end of the PTP3 gene. A 4-kb fragment corresponding to the ptp3⌬::TRP1 allele integrated at the correct locus, and a 5.1-kb fragment corresponding to the wild type PTP3 locus was detected in several transformants. Dissection of ptp3⌬::TRP1/PTP3 heterozygous diploid strains resulted in a 2:2 segregation of Trp ϩ to Trp Ϫ spore clones.
Mutagenesis of PTP2 and PTP3-PTP2 and PTP3 were mutagenized by PCR-based methods. To produce the mutant allele, PTP2-C666S, where Cys-666 is mutated to Ser, the oligonucleotide, 5Ј-GGAACCCT-GCAGAAGAATGGACTAA-3Ј, containing the underlined mutation was paired with a second oligonucleotide upstream of this site, 5Ј-GGCAC-CTGCAGTTTCTGAAGCATC-3Ј. To produce PTP2-C666A, where Cys-666 is mutated to Ala, the oligonucleotide, 5Ј-CACCCTGCAGAAGCAT-GGACTAAT-3Ј containing the underlined mutations was paired with the second oligonucleotide described above. In each case the 707-bp PCR product was digested with PstI at two naturally occurring sites and cloned into pUC19. This mutagenized fragment, when completely sequenced, identified only those mutations introduced by the mutagenic oligonucleotide. PTP2-C666S or PTP2-C666A was expressed in a low copy CEN-based plasmid bearing the TRP1 gene, by substituting the mutagenized PstI fragments for the wild type PstI fragment in the plasmid pHSe (22), generating pPTP2-C666S and pPTP2-C666A. PTP2-C666S and PTP2-C666A were expressed in a high copy 2-m-based plasmid, also bearing the TRP1 gene, by cloning PvuII fragments from these plasmids into YEplac112 (40). To produce the mutant allele PTP2-C670A the mutagenic oligonucleotide, 5Ј-GGTTCT-GCAGGGGC-TGGAAGAACAGG-3Ј bearing the underlined mutations and a second oligonucleotide corresponding to a region in the 3Ј-flanking sequence of the gene, 5Ј-CCCAAGCTTGATATCGCAAAAATAAAAC-3Ј, were used to produce a 450-bp PCR product that was ligated together with wild type fragments from PTP2 into YEplac112. To produce the PTP2-C-666A,C670A double mutant, fragments from each mutant were cloned into YEplac112.
The catalytic residue, Cys-804, in Ptp3, was mutated to Ala using similar methods. PTP3-C804A was produced by PCR using an oligonucleotide bearing two mutations, 5Ј-CAAAGTACCAGTCCTTCCACAAC-CTGCGGAAGCATGAACCAAAATGG-3Ј, which was paired with a second primer upstream of this site, 5Ј-GAGACGTATTTGAGTGCAGTC-3Ј. PCR generated a 1.4-kb PCR product that was digested with XbaI and BsrI at naturally occurring sites and, together with a 647-bp BsrI-PstI from the wild type PTP3 gene, cloned into pUC19. The ϳ2-kb PstI-XbaI fragment together with a ϳ1.5-kb XbaI-EcoRI fragment from the wild type PTP3 gene were simultaneously ligated into YEplac 181, a 2-m-based plasmid bearing the LEU2 gene (40). Sequencing identified only two mutations corresponding to those introduced by the mutagenic oligonucleotide.
Construction of PTP2, PTP3, and HOG1 Expression Plasmids-Ptp2 FIG. 1. The osmotic stress response pathway in S. cerevisiae. In response to osmotic stress, the dephosphorylated forms of Sln1, Ypd1, and Ssk1 activate the MEKKs, Ssk2 and Ssk22, the MEK, Pbs2, and the MAPK, Hog1. Activation of Hog1 requires phosphorylation of Thr-174 and Tyr-176 in the activation loop. PTC1 is a negative regulator of this pathway that encodes a type 2C Ser/Thr phosphatase. Its substrate in this pathway is not known. PTP2 and PTP3 encode protein-tyrosine phosphatases that inactivate the pathway by acting on Hog1. and Ptp3 were expressed as fusions to GST using pEG(KT), a 2-mbased vector bearing the URA3 gene and GST under regulation of the GAL1/10 promoter (41). The plasmid pGST-PTP2 was constructed by introducing a BamHI site just upstream of the PTP2 start codon using PCR. The oligonucleotide, 5Ј-GCGGATCCATGGATCGCATAGC-ACAGC-3Ј (start codon underlined) was paired with an oligonucleotide, 5Ј-GCCGA-TATCCTTAGCATTGG-3Ј, corresponding to an EcoRV site 320 downstream of the start codon. The fragment produced by PCR was digested with BamHI and EcoRV and, together with wild type PTP2 fragments from the plasmid pHS6.7 (22), cloned into pEG(KT). The plasmid pGST-PTP3 was constructed by introducing a BamHI site just upstream of the start codon using the oligonucleotide 5Ј-CGGGATC-CATGAAGGACAGTGTAGACTGC-3Ј (start codon underlined). This oligonucleotide was paired with 5Ј-GGCATGTTCGGTAAACGGCGGCC-3Ј, corresponding to a naturally occurring HindIII site 386 bp downstream of the start codon. The PCR product was digested with BamHI and HindIII and, together with fragments from the wild type PTP3 gene obtained from pAF12, cloned into pEG(KT). Mutants PTP2-C666S and PTP3-C804A were also fused to the carboxyl terminus of GST by similar methods. The plasmid pHOG1-HA, expresses Hog1 tagged at its carboxyl terminus with two repeats of the hemagglutinin epitope (Hog1-HA) under regulation of the CUP1 promoter in the vector YEp181. A HindIII site was engineered upstream of the HOG1 start codon, and a NotI site was substituted at the stop codon to generate the fusion to the HA epitope. A 400-bp BamHI-EcoRI fragment containing the CUP1 promoter was inserted upstream of the start codon. The Hog1-HA fusion protein is functional, since it complements the osmosensitivity of a hog1⌬ strain.
Assay for Phosphatase Activity-The phosphatase activity of Ptp2 and Ptp3 was tested as follows. RS334 carrying pGST-PTP2 or pGST-PTP3 was grown in synthetic media lacking uracil and containing 2% galactose. Cells from 250 ml of culture grown to ϳ1.0 unit (A 600 nm ) were harvested and homogenized by glass beading in lysis buffer, 50 mM Tris-HCl (pH 7.5), 50 mM NaCl, 5 mM MgCl 2 , 0.2% Triton X-100, 0.1% 2-mercaptoethanol, with protease inhibitors (leupeptin, pepstatin A, antipain, aprotinin, and chymostatin each at 20 g/ml), and 1 mM phenylmethylsulfonyl fluoride. The lysate was centrifuged at high speed for 10 min at 4°C, and the supernatant was incubated with 45 l of a 1:1 slurry of glutathione-Sepharose beads (Pharmacia Biotech Inc.), for 1.5 h at 4°C. The beads were washed 3 ϫ with lysis buffer, followed by 3 washes with lysis buffer containing 150 mM NaCl, and finally 3 ϫ with lysis buffer containing 300 mM NaCl. Two proteins of ϳ112 kDa corresponding to GST-Ptp2 and ϳ125 kDa corresponding to GST-Ptp3 were found in these preparations as detected by SDS-PAGE and immunoblotting with anti-GST antibody (Pharmacia) or by silver staining. These proteins were absent from RS334, which carries pEG(KT), and expresses a ϳ25-kDa protein corresponding to GST. The GST-Ptp2, GST-Ptp3, or GST proteins bound to glutathione-Sepharose beads were washed in phosphatase buffer, 50 mM imidazole HCl (pH 7.2), and 0.1% 2-mercaptoethanol and then incubated with phosphatase buffer containing 10 mM p-nitrophenyl phosphate (PNPP, Sigma) at 30°C. The hydrolysis of PNPP was monitored at 410 nm.
Coprecipitation Assay-Binding between PTPs and Hog1 was detected as follows. The plasmids pGST-PTP2-C666S, pGST-PTP3-C804A, or pEG(KT) were coexpressed with pHOG1-HA in RS334. Yeast were grown in synthetic media lacking uracil and leucine to select for plasmids and containing 2% galactose and 100 M copper sulfate. Cells from 100 ml of culture grown to ϳ1 unit (A 600 nm ) were harvested and lysed by glass beading in lysis buffer. The lysates were centrifuged, and the supernatant was incubated with 45 l of 1:1 slurry of glutathione-Sepharose (Pharmacia) and mixed for 1.5 h at 4°C. The beads were washed extensively, as described above (see PNPP assay), boiled in sample buffer, and analyzed by SDS-PAGE and immunoblotting.
Immunoblot Analysis-Hog1-Tyr(P) was detected by immunoblotting with antiphosphotyrosine antibody (PY20, ICN), and Hog1-HA was detected with anti-HA antibody (12CA5, Babco). Cells in exponential growth phase were osmotically shocked by the addition of an equal volume of media containing 0.8 M sodium chloride. Cells were harvested and homogenized by glass beading in lysis buffer containing 100 M sodium orthovanadate and 50 mM ␤-glycerophosphate. Lysates were boiled in sample buffer, separated in a 10% SDS-PAGE gel, and transferred to polyvinylidene difluoride (Millipore) in 20% methanol, 25 mM Tris, 0.2 M glycine, 0.01% SDS, using a Genie transfer apparatus (Idea Scientific). The blot was blocked in TNST, 20 mM Tris, 0.15 M NaCl, 0.01% Tween, containing 1% bovine serum albumin for 1 h and incubated with primary antibody (PY20, ICN) at a dilution of 1:1000 in TNST for 1 h. After washing, rabbit anti-mouse alkaline phosphataseconjugated antibody (Promega) was added at a dilution of 1:7500 in TNST for 30 min. The blot was washed with TNST, and the immunoreactivity was visualized using 5-bromo-4-chloro-3-indolyl phosphate and nitro blue tetrazolium (42) (Promega).
RNA Analysis-Yeast grown to ϳ1 unit, A 600 nm , in YPD were untreated or osmotically shocked for 10 min by the addition of YPD containing 0.8 M NaCl. The cells were harvested by centrifugation, and the RNA were prepared by freezing in phenol and SDS (43). Total RNA was electrophoresed in formaldehyde-containing agarose gels, and hybridization was performed by standard methods (39). The blot was hybridized with 32 P-labeled probes to PTP2, PTP3, GPD1, and TUB1. A 750-bp PstI fragment internal to the PTP2 open reading frame, a 1-kb ClaI fragment internal to the PTP3 open reading frame, a ϳ1.1-kb BglII-ClaI fragment from TUB1, and a 814-bp SalI fragment from GPD1, containing 352 bp of the GPD1 open reading frame and 462 bp of upstream sequence, were used to produce 32 P-labeled probes (39).

RESULTS AND DISCUSSION
Identification of PTP3 as a Negative Regulator of the HOG MAPK Pathway-Protein phosphatases that negatively regulate the S. cerevisiae HOG pathway were isolated by a genetic approach. A selection was devised using a sln1⌬ strain whose severe growth defects are due to constitutive activation of the HOG MAPK cascade (25,27). Since this growth defect can be alleviated by mutational inactivation of members of the MAPK cascade, overexpression of negative regulators should also suppress lethality. Such negative regulators should include protein phosphatases that inactivate the MAPK module. To identify such regulators, we used the strain, IMY101, lacking chromosomal copies of both SLN1 and URA3 and carrying both of these genes on a single plasmid, pSLN1-URA3 (27). Because this strain requires pSLN1-URA3 for viability, it is necessarily Ura ϩ and therefore unable to grow on media containing 5-FOA, which selects against URA3-expressing cells. Overexpression of negative regulators should allow survival of the sln1⌬ strain in the absence of pSLN1-URA3 and thus growth on 5-FOA. This selection yielded two protein-tyrosine phosphatases, PTP2, and a new gene, PTP3, which, when overexpressed, inactivate the HOG pathway.
PTP2 and PTP3 Show Differences in Their Ability to Inactivate the HOG Pathway-To begin examining why two proteintyrosine phosphatases regulate the HOG pathway, the effects of altering the level of expression of PTP2 and PTP3 were examined. Although both PTP2 and PTP3 suppress the severe growth defects of the sln1⌬ strain when expressed from a multicopy 2-m-based plasmid, only PTP2 suppressed the growth defect when expressed from a low copy CEN-based plasmid. Thus PTP2 may have a greater effect on the HOG pathway than PTP3. To test whether deletion of PTP2 would have a greater impact on the HOG pathway than deletion of PTP3, strains lacking either or both PTPs were constructed. No obvious differences were observed between ptp2⌬, ptp3⌬, ptp2⌬ ptp3⌬, and wild type strains grown under standard conditions or in high osmolarity media where the HOG pathway would be activated. Differences were observed between ptp2⌬ and ptp3⌬ strains, however, when combined with the sln1⌬ mutation. The sln1⌬ strain exhibits severe growth defects on synthetic media due to HOG pathway hyperactivation. If the role of these PTPs is to inactivate the HOG pathway, then their deletion should adversely affect the sln1⌬ strain. The sln1⌬ ptp2⌬ double mutant is lethal on synthetic media, whereas the sln1⌬ ptp3⌬ strain grows as well as the sln1⌬ strain on synthetic media (data not shown). These results suggest that the HOG pathway is more acutely affected by PTP2 than by PTP3.
Ptp2 and Ptp3 Are Phosphatases Whose Activity Is Important for Inactivating Hog1-Analysis of the primary sequence of Ptp2 and Ptp3 shows they are similar in structure, having a novel amino-terminal domain fused to a carboxyl-terminal protein-tyrosine phosphatase (PTP) domain. The amino-terminal domains of Ptp2 and Ptp3 show two small regions of similarity to each other and to Pyp1 and Pyp2, which regulate an osmotic stress response pathway in S. pombe. One of these regions is also similar to PAC1 (44, 45), a vertebrate dual specificity phosphatase, which shows greater activity toward the vertebrate homolog of Hog1, p38, and ERK than to JNK/SAPK (12). The PTP domains, ϳ400 amino acids in length in Ptp2 and Ptp3, are 32% identical to each other and significantly similar to other PTPs in yeast and vertebrates (Fig. 2) (8). The PTP domain of Ptp2 is most similar to Ptp3 and also shows strong similarities to the S. pombe Pyp2 and Pyp1.
To test whether PTP2 and PTP3 encode enzymes with phosphatase activity, they were expressed in yeast, and their activity was assayed in vitro with the phosphatase substrate, pnitrophenyl phosphate (PNPP). To facilitate purification, GST was fused to the amino terminus of Ptp2 or Ptp3 and placed under control of the GAL1 promoter to achieve high levels of expression. Fusion of GST to wild type Ptp2 or Ptp3 did not disrupt PTP activity, since each fusion retained the ability to suppress the severe growth defects of the sln1⌬ strain (data not shown). The fusion proteins were isolated by binding to glutathione-Sepharose, and after extensive washing, PNPP was added and its hydrolysis monitored at A 410 nm . These fusion proteins demonstrated activity toward PNPP that was not observed with GST alone. After 1 h of incubation at 30°C, a change in absorbance of ⌬A 410 nm of 0.158 was observed for ϳ0.5 g of GST-Ptp2 and a ⌬A 410 nm of 0.313 for ϳ0.5 g of GST-Ptp3. To determine whether the signature sequence found among all PTPs, (I/V)HCXAGXXR(S/T)G, is important for Ptp2 and Ptp3 activity, the Cys residue within this sequence was mutated to Ser or Ala. These mutations have been shown to inactivate other PTPs in vitro and in vivo (8,15,21). Furthermore, the sulfhydryl moiety has been shown to act as a nucleophile for phosphotyrosine hydrolysis (8). Cys-666 in Ptp2 was mutated to Ser or Ala, and Cys-804 in Ptp3 was mutated to Ala. These mutant fusion proteins expressed in yeast were unable to hydrolyze PNPP.
The requirement for Ptp2 and Ptp3 phosphatase activity was tested in vivo by determining whether the mutant PTPs could suppress the severe growth defect of the sln1⌬ strain. The mutants Ptp2-C666S (not shown) and Ptp2-C666A expressed from low copy CEN-based plasmids and the mutant Ptp3-C804A expressed from a multicopy 2-m-based plasmid were not able to suppress the severe growth defects of the sln1⌬ strain as well as their wild type counterparts (Fig. 3). Thus the phosphatase activity of Ptp2 and Ptp3 is required for inactivation of the HOG pathway. Interestingly, the Ptp2-C666S and Ptp2-C666A mutants retained some ability to suppress the growth defect of the sln1⌬ strain. Expression of Ptp2-C666A from a low copy plasmid allows the sln1⌬ strain to grow better than controls with empty vector (Fig. 3). Ptp2 and Ptp3 differ from all other PTPs in that they each contain two Cys residues within the PTP signature sequence (Fig. 2). To test whether the second Cys residue in Ptp2, Cys-670, could act as a nucleophile, it was mutated to Ala to generate Ptp2-C670A. This mutant suppressed the growth defect of the sln1⌬ strain as well as wild type PTP2 (Fig. 3). A single polypeptide bearing both mutations, Ptp2-C666A,C670A, suppressed as well as the single mutant Ptp2-C666A. Thus Cys-666 but not Cys-670 is important for PTP2 suppressor function. A third possibility was that Ptp2-C666S and Ptp2-C666A act as dominant negative mutants, inactivating the HOG pathway by binding and sequestering their substrate, preventing activation of downstream components, rather than by dephosphorylation of its substrate. This seems possible since active site mutants of dual specificity protein phosphatase and other PTPs show enhanced binding to their substrates in vitro (11,15,21). Consistent with this interpretation, Ptp2-C666S (not shown), Ptp2-C666A, or Ptp2-C666A,C670A when overexpressed from multicopy 2-m-based plasmids suppressed sln1⌬ growth defects as well as wild type PTP2 (Fig. 3). Ptp3-C804A mutants did not suppress sln1⌬ growth defects when overexpressed from a multicopy 2-mbased plasmid (Fig. 3). This was not surprising, however, since multicopy expression of wild type PTP3 is required for suppression of the sln1⌬ strain growth defect.
If Ptp2-C666S and Ptp2-C666A are dominant negative mutants, they should bind tightly to their substrate. Hog1 is the most likely target because it is the only component in this pathway known to require phosphorylation of a Tyr residue in addition to a Thr residue for activation (26). Therefore, we tested whether the catalytically inactive Ptp2 mutant, Ptp2-C666S, could bind Hog1 in vitro. A GST fusion to the catalytically inactive Ptp2 was coexpressed with epitope-tagged Hog1 (Hog1-HA) in yeast. Cells were osmotically shocked and the GST-Ptp2-C666S protein isolated by addition of glutathione-Sepharose beads. After extensive washing, SDS-PAGE and immunoblotting were performed to determine whether tyrosine-phosphorylated Hog1-HA (Hog1-HA-Tyr(P)) was bound. GST-Ptp2-C666S bound Hog1-HA-Tyr(P) as detected by immunoblotting with anti-phosphotyrosine or anti-HA antibody (Fig.  4). This binding interaction is due to Ptp2-C666S, since GST did not precipitate Hog1-HA from yeast extracts. Although the mutant, Ptp3-C804A did not act as a dominant negative mutant in vivo, this mutant also bound Hog1-HA-Tyr(P) in vitro, although less well than Ptp2-C666S (Fig. 4). The most likely explanation for the failure of the PTP3 mutant to suppress the HOG pathway in vivo is that at 2 m expression levels it cannot bind enough activated Hog1 to suppress growth defects. Thus both Ptp2 and Ptp3 are able to bind Hog1, and wild type Ptp2 and Ptp3 most likely inactivate the HOG pathway by dephosphorylating Hog1-Tyr(P).
Ptp2 and Ptp3 Regulate Hog1 in Vivo-If Ptp2 and Ptp3 dephosphorylate Hog1-Tyr(P) in vivo, strains lacking these phosphatases should show defects in their ability to dephosphorylate Hog1-Tyr(P). The following results indicate that Ptp2 and Ptp3 have two roles in the cell. One role is to maintain a low basal level of Hog1-Tyr(P) when cells are grown under standard conditions. A second role is in adaptation. The basal level of Hog1-Tyr(P) is highest in the ptp2⌬ ptp3⌬ double mutant and was also elevated in the ptp2⌬ mutant (Fig. 5). Low levels of Hog1-Tyr(P) were detected in the ptp3⌬ strain, and none was visible in wild type. These results are interesting because ptp null mutants express high levels of Hog1-Tyr(P) under standard growth conditions but do not exhibit the severe growth defects associated with Hog1 hyperactivation, as seen, for example, with mutations that inactivate SLN1 (25). 2 The most likely explanation for the lack of growth defects in ptp2⌬ and ptp2⌬ ptp3⌬ mutant strains is that Ptp2, and possibly Ptp3, specifically dephosphorylate phosphotyrosine but not phosphothreonine in the phosphorylation lip of Hog1. This might result if in ptp null strains, the level of Hog1-Tyr(P) is high, but the level of threonine phosphorylation is low, and as a result, Hog1 is not hyperactivated. Support for this hypothesis comes from the observation that deletion of a type 2C Ser/Thr phosphatase, PTC1, together with PTP2, produces a severe synthetic growth defect (47). PTC1 has been identified as a negative regulator of the HOG pathway, but its substrate is not known. If Ptc1 mediates dephosphorylation of phosphothreonine in Hog1, the synthetic growth defect of ptc1⌬ ptp2⌬ might be explained by hyperphosphorylation of Hog1 at both Thr and Tyr residues in the phosphorylation lip. To test this idea we deleted HOG1 or PBS2 in the ptc1 ptp2⌬ background. Both ptc1 ptp2⌬ hog1⌬ and ptc1 ptp2⌬ pbs2⌬ strains grew as well as the ptc1 mutant (data not shown), indicating that the severe synthetic growth defect of the ptc1 ptp2⌬ strain is due to hyperactivation of the HOG pathway. Ptc1 inactivation of Hog1 may be direct, via dephosphorylation of Hog1-Thr(P)-174, or indirect, via dephosphorylation of Pbs2 or other upstream activators. Similar tests performed with PTP3 demonstrated no interaction with PTC1; the double mutant ptc1⌬ ptp3⌬ showed no synthetic growth defect (data not shown). These results suggest that Ptp2 has a greater role than Ptp3 in maintaining the low basal level of Hog1-Tyr(P) under standard growth conditions.
A second role for these PTPs is in adaptation. When wild type FIG. 3. Ptp2 and Ptp3 suppress growth defects of the sln1⌬ strain that hyperactivates the HOG pathway. The ability of PTP2, PTP3, and mutant PTPs to suppress the nearly lethal phenotype of the sln1⌬ strain was tested on synthetic media containing 5-FOA. The sln1⌬ strain bearing the plasmids, pSLN1-URA3 (IMY101) and empty vector, is unable to grow on media containing 5-FOA. Introduction of PTP2, expressed from a low copy CEN-based plasmid, however, restores viability. Cys-666 but not Cys-670 within the signature sequence of Ptp2 is important for suppressor function. The mutants, Ptp2-C666A and Ptp2-C666A,C670A, do not suppress lethality when expressed from a low copy plasmid. Ptp2-C670A, however, suppresses as well as wild type PTP2. The Ptp2-C666A and Ptp2-C666A,C670A mutants, inactive at low copy, are able to suppress the sln1⌬ strain growth defect when expressed from a multicopy 2-m-based plasmid. Cys-804 in the signature sequence of Ptp3 is important for suppression. Ptp3-C804A is unable to suppress the sln1⌬ defect. cells are osmotically shocked Hog1-Tyr(P) levels increase rapidly, reaching maximal levels by ϳ5 min. This is followed by a rapid decrease in Hog1-Tyr(P) to nearly basal levels by 30 min (Fig. 5). If PTPs are responsible for this rapid decrease, strains lacking PTPs should dephosphorylate Hog1-Tyr(P) more slowly than wild type. Cultures of exponentially growing wild type and ptp2⌬, ptp3⌬, or ptp2⌬ ptp3⌬ strains were exposed to continuous osmotic stress, and at various times the level of Hog1-Tyr(P) was determined by SDS-PAGE and immunoblotting. Upon osmotic shock, Hog1-Tyr(P) increased in all strains, but the rates of Hog1-Tyr(P) dephosphorylation differ. The rate of dephosphorylation is most rapid in wild type and nearly as fast in the ptp3⌬ strain (Fig. 5). Dephosphorylation of Hog1-Tyr(P) is significantly slower in the ptp2⌬ strain and is dramatically slower in the ptp2⌬ ptp3⌬ double mutant (Fig. 5). Thus strains lacking PTP2, or PTP2 and PTP3, failed to dephosphorylate Hog1-Tyr(P) as well as wild type. One interesting feature of these data is that deletion of PTP3 had little effect on the rate of Hog1-Tyr(P) dephosphorylation, yet the ptp2⌬ ptp3⌬ had a synergistic effect, slowing Hog1-Tyr(P) dephosphorylation more than would be expected by the sum of each mutant alone. This effect could be explained if Ptp3 has other roles in addition to regulating Hog1 directly.
PTP2 and PTP3 Expression Are Induced by Osmotic Stress and Are Dependent upon Hog1-The results above indicate that Ptp2 and Ptp3 are involved in adaptation. Because the state of HOG pathway activation is sensitive to the level of PTP2 or PTP3 expression, one mechanism of adaptation might involve induction of PTP transcripts in response to osmotic stress. To test this idea, an exponential culture of a wild type strain was untreated or exposed to osmotic shock for 10 min at 30°C. Total RNA was examined by Northern analysis. The level of PTP2 transcript increased ϳ2-3-fold, whereas PTP3 transcript increased ϳ5-fold following osmotic shock (Fig. 6). GPD1, encoding glycerol 3-phosphate, increased substantially as described previously (48) and TUB1 decreased slightly. Total RNA prepared from a hog1⌬ strain that was untreated or exposed to osmotic stress showed no significant increases in the level of PTP2 or PTP3. The GPD1 transcript is still induced in the hog1⌬ strain but to a much lesser degree than in wild type (48). Thus the HOG pathway is required for induction of PTP2 and PTP3 transcripts in response to osmotic shock, suggesting that activation of the HOG pathway triggers a negative feedback loop to inactivate the pathway.
Induction of PTP2 is likely to involve promoter stress elements also present in other genes regulated by the HOG pathway (49). Although the PTP2 promoter contains two exact matches to the stress elements, C 4 T, PTP3 has none. Thus PTP3 induction may occur through other elements. Msn2 and Msn4, two zinc finger proteins, originally identified as suppressors of Snf1, a kinase involved in glucose sensing (50), have been shown to bind the stress elements in vitro, acting as transcriptional activators (51,52). Msn2 and Msn4, however, are unlikely to be the only mediators of Hog1-activated transcription since msn2⌬ msn4⌬ strains, unlike hog1⌬, grow as well as wild type on media containing 0.8 M NaCl (51). Thus expression of PTP2 may involve these activators as well as others. MAPK pathway-induced expression of protein phosphatases is likely to be a general mechanism of adaptation. Expression of MSG5, a gene encoding a dual specificity phosphatase in S. cerevisiae, is induced by activation of the pheromone response pathway, and pyp2 ϩ , a gene encoding a PTP in S. pombe, is induced by an osmotic stress response pathway (15,19,20). Strains lacking pyp1 ϩ or pyp2 ϩ show a decreased rate of Sty1/Spc1-Tyr(P) dephosphorylation (19 -21). Whether transcriptional activation of these genes is required for adaptation is not known.
In summary, we have described two protein-tyrosine phosphatases that regulate the HOG pathway in S. cerevisiae by targeting the Hog1 MAPK. Both PTP2 and PTP3 express phosphatase activity, and this activity is required in vivo for pathway inactivation. Mutant Ptp2 and Ptp3 bind Hog1 in vitro, and strains lacking PTP2 and PTP3 show elevated levels of Hog1-Tyr(P), strongly suggesting they inactivate this pathway by dephosphorylating Hog1-Tyr(P). Why two PTPs are needed to regulate Hog1 is not clear. We find, however, that the activity of the HOG pathway is more sensitive to PTP2 than to PTP3. Expression of PTP2 from a low copy plasmid is sufficient to suppress the growth defects of the sln1⌬ strain, whereas multicopy expression of PTP3 is required for suppression. This is corroborated by synthetic lethality of the ptp2⌬ sln1⌬ strain but not the ptp3⌬ sln1⌬ strain, the synthetic growth defect of the ptp2⌬ ptc1⌬ strain but not ptp3⌬ sln1⌬ strain, and differences in the level of Hog1-Tyr(P) in ptp2⌬ versus ptp3⌬ strains. The greater induction of PTP3 seen in response to osmotic shock compared with PTP2 does not compensate for the differ- The level of Hog1-Tyr(P) was determined in wild type, ptp2⌬, ptp3⌬, and ptp2⌬ ptp3⌬ strains in the absence of osmotic stress (0 min) or during continuous exposure to osmotic stress for the indicated times. Osmotic stress was initiated by the addition of sodium chloride to a final concentration of 0.4 M, and the cultures were incubated at 30°C. At various times, cells were harvested, boiled in sample buffer, and Hog1-Tyr(P), whose position is indicated by the filled circle, detected by immunoblotting with anti-phosphotyrosine antibody.
FIG. 6. PTP2 and PTP3 transcripts are induced in response to osmotic stress in a Hog1-dependent manner. The effect of osmotic stress on the expression of PTP2 and PTP3 was examined in ptp2⌬, ptp3⌬, hog1⌬, and wild type strains by Northern analysis. RNA was prepared from exponentially growing cells either untreated or exposed to osmotic stress by the addition of sodium chloride to a final concentration of 0.4 M for 10 min at 30°C. Total RNA was hybridized with 32 P-labeled DNA probes to PTP2, PTP3, GPD1, and TUB1. Ethidium staining indicated that samples were evenly loaded. ences between these PTPs since the ptp2⌬ strain dephosphorylates Hog1-Tyr(P) at a substantially slower rate compared with the ptp3⌬ strain. Further investigation of the regulation of these PTPs, possibly via their posttranslational modification or subcellular localization, should contribute to a better understanding of the roles of these PTPs in the HOG pathway.
Osmotic stress response pathways operate in S. cerevisiae, S. pombe, and mammals. Thus far, Ptp2, Ptp3, and the S. pombe Pyp1 and Pyp2 are the only PTPs known to regulate MAPK signaling pathways. These PTPs show sequence similarities not found in other PTPs, making it tempting to speculate that they confer specificity for osmoregulatory MAPKs. The PTP regulation of MAPKs differ between budding and fission yeast in two respects. Both PTP2 and PTP3 are induced upon osmotic shock while only pyp2 ϩ is induced. Also, deletion of PTP2 and PTP3 confers no obvious growth defect, whereas deletion of pyp1 ϩ and pyp2 ϩ is lethal. Since lethality can be suppressed by mutational inactivation of Sty1/Spc1, deletion of pyp1 ϩ and pyp2 ϩ is sufficient to activate Sty1/Spc1 (21). This difference could be explained if phosphorylation of the Thr residue in the phosphorylation lip of Sty1/Spc1 is regulated differently from Hog1. Thus variations occur in the negative feedback regulation of these pathways. The mouse p38, functionally analogous to the yeast Hog1, is inactivated by the vertebrate dual specificity phosphatases, MKP-1, PAC1, and M3/6 (12,46). Whether these phosphatases are physiologically relevant regulators of p38 and whether PTPs will be identified that regulate p38 remains to be established.