(Received for publication, February 26, 1997, and in revised form, May 5, 1997)
From the Department of Chemistry and Biochemistry, University of Colorado, Boulder, Colorado 80309-0215
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.
MAPK1 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. 1-5). 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. 7-10). 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 (11-14). 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 (22-24) 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.
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 µm HOG1, a multicopy
2-µm-based plasmid bearing HOG1 and TRP1,
kindly provided by M. Gustin (26). The strains bearing 2 µm 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.
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 ~8-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
5
end of the gene. A fragment corresponding to the 3
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+ 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 3
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+ to Trp
spore clones.
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, 5-GGAACCCTGCAGAAGAATGGACTAA-3
,
containing the underlined mutation was paired with a second
oligonucleotide upstream of this site, 5
-GGCACCTGCAGTTTCTGAAGCATC-3
.
To produce PTP2-C666A, where Cys-666 is mutated to Ala, the
oligonucleotide, 5
-CACCCTGCAGAAGCATGGACTAAT-3
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-µm-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, 5
-GGTTCT-GCAGGGGCTGGAAGAACAGG-3
bearing the
underlined mutations and a second oligonucleotide corresponding to a
region in the 3
-flanking sequence of the gene,
5
-CCCAAGCTTGATATCGCAAAAATAAAAC-3
, 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-C666A,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,
5-CAAAGTACCAGTCCTTCCACAACCTGCGGAAGCATGAACCAAAATGG-3
, which was paired with a second primer upstream of this site,
5
-GAGACGTATTTGAGTGCAGTC-3
. 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 ~2-kb PstI-XbaI
fragment together with a ~1.5-kb XbaI-EcoRI
fragment from the wild type PTP3 gene were simultaneously
ligated into YEplac 181, a 2-µm-based plasmid bearing the
LEU2 gene (40). Sequencing identified only two mutations
corresponding to those introduced by the mutagenic oligonucleotide.
Ptp2 and Ptp3 were expressed as fusions to GST using
pEG(KT), a 2-µm-based 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, 5-GCGGATCCATGGATCGCATAGC-ACAGC-3
(start
codon underlined) was paired with an oligonucleotide,
5
-GCCGA-TATCCTTAGCATTGG-3
, 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
5
-CGGGATCCATGAAGGACAGTGTAGACTGC-3
(start codon
underlined). This oligonucleotide was paired with
5
-GGCATGTTCGGTAAACGGCGGCC-3
, 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.
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 ~1.0 unit (A600 nm) were harvested and homogenized by glass beading in lysis buffer, 50 mM Tris-HCl (pH 7.5), 50 mM NaCl, 5 mM MgCl2, 0.2% Triton X-100, 0.1% 2-mercaptoethanol, with protease inhibitors (leupeptin, pepstatin A, antipain, aprotinin, and chymostatin each at 20 µg/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 µl of a 1:1 slurry of glutathione-Sepharose beads (Pharmacia Biotech Inc.), for 1.5 h at 4 °C. The beads were washed 3 × with lysis buffer, followed by 3 washes with lysis buffer containing 150 mM NaCl, and finally 3 × with lysis buffer containing 300 mM NaCl. Two proteins of ~112 kDa corresponding to GST-Ptp2 and ~125 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 ~25-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 AssayBinding 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 µM copper sulfate. Cells from 100 ml of culture grown to ~1 unit (A600 nm) were harvested and lysed by glass beading in lysis buffer. The lysates were centrifuged, and the supernatant was incubated with 45 µl 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 AnalysisHog1-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 µM 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 phosphatase-conjugated 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).
Yeast grown to ~1 unit, A600 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 32P-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 ~1.1-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 32P-labeled probes (39).
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+ 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.
To begin examining why two protein-tyrosine
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-µm-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.
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, ~400 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,
p-nitrophenyl 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
A410 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
A410 nm of 0.158 was observed for ~0.5
µg of GST-Ptp2 and a
A410 nm of 0.313 for
~0.5 µg 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-µm-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-µm-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-µm-based 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 µm 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 cells are
osmotically shocked Hog1-Tyr(P) levels increase rapidly, reaching
maximal levels by ~5 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.
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
~2-3-fold, whereas PTP3 transcript increased ~5-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, C4T, 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+, a gene encoding a PTP in S. pombe, is induced by an osmotic stress response pathway (15, 19,
20). Strains lacking pyp1+ or
pyp2+ 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 differences 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+ is induced. Also, deletion of PTP2 and PTP3 confers no obvious growth defect, whereas deletion of pyp1+ and pyp2+ is lethal. Since lethality can be suppressed by mutational inactivation of Sty1/Spc1, deletion of pyp1+ and pyp2+ 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.
We thank Alex Varshavsky in whose laboratory this work was begun; Lisa Freeman-Cook for designing the PTP3 deletion construct; Mike Gustin, Bob Sclafani, Jurgen Dohmen, Norbert Schnell, Jim Broach, Mark Winey, and Natalie Ahn for yeast strains and reagents; and Tim Lewis, Natalie Ahn, and Jim Goodrich for helpful discussions and comments on the manuscript.