Regulation of the TAK1 signaling pathway by protein phosphatase 2C.

Protein phosphatase 2C (PP2C) is implicated in the negative regulation of stress-activated protein kinase cascades in yeast and mammalian cells. In this study, we determined the role of PP2Cbeta-1, a major isoform of mammalian PP2C, in the TAK1 signaling pathway, a stress-activated protein kinase cascade that is activated by interleukin-1, transforming growth factor-beta, or stress. Ectopic expression of PP2Cbeta-1 inhibited the TAK1-mediated mitogen-activated protein kinase kinase 4-c-Jun amino-terminal kinase and mitogen-activated protein kinase kinase 6-p38 signaling pathways. In vitro, PP2Cbeta-1 dephosphorylated and inactivated TAK1. Coimmunoprecipitation experiments indicated that PP2Cbeta-1 associates with the central region of TAK1. A phosphatase-negative mutant of PP2Cbeta-1, PP2Cbeta-1 (R/G), acted as a dominant negative mutant, inhibiting dephosphorylation of TAK1 by wild-type PP2Cbeta-1 in vitro. In addition, ectopic expression of PP2Cbeta-1(R/G) enhanced interleukin-1-induced activation of an AP-1 reporter gene. Collectively, these results indicate that PP2Cbeta negatively regulates the TAK1 signaling pathway by direct dephosphorylation of TAK1.

Stress-activated protein kinases (SAPKs) 1 are a subfamily of the mitogen-activated protein kinase (MAPK) superfamily and are highly conserved from yeast to mammalian cells. SAPKs relay signals in response to various extracellular stimuli, including environmental stress and inflammatory cytokines. In mammalian cells, two distinct classes of SAPKs have been identified: the c-Jun amino-terminal kinases (JNKs) (JNK1, JNK2, and JNK3) and the p38 MAPKs (p38␣, p38␤, p38␥, and p38␦) (1,2). Activation of SAPKs requires phosphorylation at conserved tyrosine and threonine residues in the catalytic domain. This phosphorylation is mediated by dual specificity protein kinases, which are the members of the MAPK kinase (MKK) family. Of these, MKK3 and MKK6 phosphorylate p38, MKK7 phosphorylates JNK, and MKK4 can phosphorylate either. These MKKs, in turn, are activated by phosphorylation of conserved serine and threonine residues (1,2). Recently, several MKK-activating MKK kinases (MKKKs) have been identified. Some of these MKKKs are also known to be activated by phosphorylation, but the details are unclear at present.
In the absence of signaling, SAPK cascades return to their inactive, dephosphorylated state, suggesting a possible role for phosphatases in SAPK regulation. In yeast cells, molecular genetic analysis has indicated that two distinct protein phosphatase groups, protein tyrosine phosphatase and protein serine/threonine phosphatase type 2C (PP2C), act as negative regulators of SAPK pathways (3). In Schizosaccharomyces pombe, tyrosine phosphatase Pyp2 and the yeast homolog of PP2C (Ptc1 and Ptc3) have been shown to dephosphorylate and inactivate Spc1, the yeast homolog of SAPK (4,5).
PP2C is one of four major protein serine/threonine phosphatases (PP1, PP2A, PP2B, and PP2C) in eukaryotes and is implicated in the regulation of various cellular functions. To date, at least six distinct PP2C gene products (2C␣, 2C␤, 2C␥, 2C␦, Wip1, and Ca 2ϩ /calmodulin-dependent protein kinase phosphatase) have been found in mammalian cells (6 -12). In addition, two distinct isoforms of the human PP2C␣ (␣-1 and -2) and five isoforms of the mouse PP2C␤ (␤-1, -2, -3, -4, and -5) have been identified (13)(14)(15)(16). These isoforms are generated in each species as splicing variants of a single pre-mRNA. We have recently reported that ectopic expression of mouse PP2C␣ or PP2C␤-1 inhibited the stress-activated MKK3/6-p38 and MKK4/7-JNK pathways but not the mitogen-activated MKK1-ERK1 pathway. Thus, negative regulation by PP2C␣ and PP2C␤-1 is selective for different SAPK pathways (17). Essentially the same results were obtained in studies of human PP2C␣-1 and -2 in mammalian cells (14). Currently, the in vivo target molecule(s) of PP2C is unknown, although MKK4, MKK6, and p38 have been proposed as substrates of PP2C␣-2 (14). TAK1 was originally identified as an MKKK that functions in the transforming growth factor-␤ signaling pathway (18). TAK1 can activate both the MKK4-JNK and MKK6-p38 pathways (18). Recent studies have indicated that TAK1 is also activated by various stimuli, including environmental stress and inflammatory cytokines, and that it plays critical roles in various cellular responses (19 -22). Studies on the regulation of TAK1 activity have revealed that a TAK1-binding protein, TAB1, functions as an activator promoting TAK1 autophosphorylation (21,23). However, the protein phosphatase(s) responsible for inactivation of TAK1 has not been identified. In this study, we provide evidence indicating that PP2C␤-1 selectively associates with TAK1 and inhibits the TAK1 signaling pathway by direct dephosphorylation.

EXPERIMENTAL PROCEDURES
Materials-The restriction enzymes and other modifying enzymes used for DNA manipulation were obtained from Takara (Kyoto, Japan). Anti-6xHis, anti-Myc, and anti-TAK1 antibodies (Abs) were obtained from Santa Cruz Biotechnology (Santa Cruz, CA). Anti-phospho-MKK4 and anti-phospho-MKK3/6 Abs were supplied by New England Biolabs (Beverly, MA). Anti-hemagglutinin (HA; 12CA5) and anti-Flag (M2) Abs were purchased from Roche Molecular Biochemicals and Kodak Scientific Imaging Systems, respectively. Anti-PP2C␤ Ab was raised in rabbit against an oligopeptide of mouse PP2C␤ (RILSAENIPNLPPGG-GLAGK). Human interleukin-1␤ (IL-1␤) was from Roche Molecular Biochemicals. All the other reagents used were from Wako Pure Chemical (Osaka, Japan).
Construction of Expression Plasmids-Expression plasmids were constructed by standard procedures. Plasmids that express PP2C, TAK1, TAB1, MAPKs, MKKs, and MKKKs in mammalian cells were constructed using cDNAs encoding these proteins (17,21) under the control of the CMV promoter. Epitope tags were added to the constructs using synthesized oligonucleotides. Mutated cDNAs were generated by polymerase chain reaction. For bacterial expression of proteins, cDNAs encoding the proteins were subcloned into pGEX (Amersham Pharmacia Biotech) to generate glutathione S-transferase (GST) fusion proteins or into pQE31 (Qiagen, Hilden, Germany) to generate hexahistidinetagged protein and affinity-purified by standard procedures. Other expression plasmids were as described elsewhere (23,24) Cell Culture and Transfection-COS7, 293, and 293IL-1RI (25) cells were grown in Dulbecco's modified Eagle's medium (Life Technologies, Inc.) supplemented with 10% (v/v) fetal bovine serum. At 50 -80% confluency the cells were transfected by the DEAE-dextran method or using LipofectAMINE (Life Technologies, Inc.). The total amount of DNA (0.5-2 g per 35-mm dish) was kept constant by supplementing with empty vector. The cells were cultured for 24 -48 h after transfection and then harvested.
Kinase and Phosphatase Assays-Immune complex kinase assays were performed as follows. The cells were lysed in a buffer containing 20 mM Tris-HCl, pH 7.5, 1% (v/v) Triton X-100, 150 mM NaCl, 1 mM EGTA, 1 mM sodium orthovanadate, 50 mM NaF, 1 mM dithiothreitol, and 1 mM phenylmethylsulfonyl fluoride, and the lysates were incubated with appropriate Abs for 1 h at 4°C. The resulting immune complexes were recovered with protein G-Sepharose (Amersham Pharmacia Biotech), washed twice with Tris-buffered saline (20 mM Tris-HCl, pH 7.5, 150 mM NaCl), twice with 20 mM Tris-HCl, pH 7.5, and then incubated with or without appropriate substrates in 25 l of kinase buffer (20 mM Tris-HCl, pH 7.5, 10 mM MgCl 2 , and 1 mM dithiothreitol) containing 0.5-3 Ci of [␥-32 P]ATP (NEG-002A, PerkinElmer Life Sciences) at 30°C for 10 -30 min. The reactions were stopped by adding SDS-sample buffer and boiled for 2 min. Protein phosphatase assays were carried out as follows. COS7 cells seeded onto 10-cm dishes were cotransfected with Flag-TAK1 and Myc-TAB1 expression plasmids. The Flag-TAK1-Myc-TAB1 complex was immunoprecipitated from cell extracts with anti-Flag Ab, and phosphorylation was carried out in kinase buffer containing [␥-32 P]ATP at 30°C for 30 min. After washing three times with 20 mM Tris-HCl, pH 7.5, the immune complex was then incubated with or without recombinant GST-PP2C␤ in kinase buffer at 30°C for the indicated times. Phosphorylated proteins were separated by SDSpolyacrylamide gel electrophoresis (SDS-PAGE), and the radioactivities incorporated into the proteins were detected with a BAS 2000 image analyzer (Fuji, Tokyo, Japan).
Western Blot Analysis-Proteins in the cell lysates and immunoprecipitates were separated by SDS-PAGE and electroblotted onto polyvinylidene difluoride membranes. The membranes were incubated with primary Abs at 4°C for 16 h and then incubated with horseradish peroxidase-conjugated secondary Ab at 25°C for 1 h. The chemiluminescence of each blot was detected with an enhanced chemiluminescence system (Amersham Pharmacia Biotech).
Coimmunoprecipitation Assay-Cells were lysed with a buffer containing 20 mM Tris-HCl, pH 7.5, 5 mM EDTA, 150 mM NaCl, 1% (v/v) Triton X-100, and 1 mM phenylmethylsulfonyl fluoride. The cell lysates were incubated with the indicated Abs for 1 h at 4°C. The immunoprecipitated proteins were washed three times with Tris-buffered saline and submitted to Western blot analysis.
Reporter Assay-Cells were transfected with the AP-1-luciferase reporter plasmid (26). After the transfection, cells were treated with IL-1␤ for 6 h. Luciferase activity was determined with the Luciferase Assay System (Promega). ␤-actin-␤-galactosidase reporter plasmid was cotransfected for normalizing transfection efficiencies.
We then tested whether PP2C␤-1 expression affects TAK1induced activation of JNK1 and p38␣. Both the JNK1 and p38 kinases expressed in COS7 cells were activated by the exogenous TAK1. However, these kinase activities were inhibited when PP2C␤-1 was coexpressed ( Fig. 1, C and D). In contrast, expression of PP2C␤-1(R/G), a phosphatase-defective mutant containing an Arg-179 to Gly mutation, had no inhibitory effect on TAK1-induced activation of JNK1 or p38. These results suggest that PP2C␤-1 inhibits the TAK1 signaling pathway at TAK1 or downstream of TAK1, e.g. MKKs and MAPKs.
PP2C␤ Acts upon TAK1-We have previously shown that TAK1, when coexpressed with TAB1, is activated by autophosphorylation (23). TAK1 autophosphorylation can be monitored by decreased mobility on SDS-PAGE, and this mobility shift was cancelled when Ser-192 of TAK1, which is the site of autophosphorylation, was mutated to alanine (Ref. 23; also shown in Fig. 2).
To investigate whether TAK1 is a substrate of PP2C, we examined the phosphorylation and kinase activity of TAK1 incubated with PP2C in vitro. Flag-TAK1 and TAB1 were coexpressed in COS7 cells, and Flag-TAK1 was immunoprecipitated from cell extracts with anti-Flag Ab. When the immunopurified TAK1 complex was incubated with [␥-32 P]ATP, TAK1 became autophosphorylated. This reaction mixture was next incubated with bacterially produced GST-PP2C␤-1 or GST-PP2C␤-1(R/G). TAK1 was found to be dephosphorylated by GST-PP2C␤-1, but not by GST-PP2C␤-1(R/G), in a dose-dependent manner (Fig. 3A). The PP2C␤-1-mediated dephosphorylation reaction was dependent on the presence of Mg 2ϩ (Fig.  3B).
PP2C␤ Does Not Dephosphorylate MKK6 in Vitro-Recent studies have indicated that one of the human PP2C isoforms, PP2C␣-2, dephosphorylates and inactivates MKK4, MKK6, and p38 in vitro (14). Therefore, we tested whether PP2C␤-1 could also dephosphorylate and inactivate MKK6 in vitro. Bacterially produced MKK6 is activated by autophosphorylation and is able to phosphorylate p38 in vitro (27). We used this system to determine the effect of recombinant PP2C␤-1 on MKK6 activity. We found that PP2C␤-1 treatment did not affect MKK6 kinase activity under conditions where it inactivates TAK1 (Fig. 3, A and C versus Fig. 4A).
Next, we determined the effect of PP2C␤-1 on stress-induced phosphorylation of MKK6. COS7 cells were transfected with Flag-MKK6 and subjected to hyperosmotic stress, and Flag-MKK6 was immunoprecipitated from the cell lysates with anti-Flag Ab. The immunoprecipitates were then incubated with GST-PP2C␤-1. Increasing concentrations of GST-PP2C␤-1 had no effect on the phosphorylation level of MKK6 (Fig. 4B). Taken together, these results indicate that PP2C␤-1 does not act upon MKK6.
The observation that the catalytically inactive PP2C␤ has a higher affinity for TAK1 than that of wild-type PP2C␤ suggested that PP2C␤ might preferentially bind phosphorylated TAK1. The TAK1(S/A) mutant, in which Ser-192 is replaced by Ala, is defective in both phosphorylation and activation (23). We coexpressed PP2C␤-1 and TAK1 or TAK1(S/A) in 293 cells and performed coimmunoprecipitation experiments. We found that TAK1(S/A) had an affinity for PP2C␤-1 similar to that of wild-type TAK1 (Fig. 5C), indicating that phosphorylation at Ser-192 is not required for association with PP2C␤-1.
We next examined whether endogenous PP2C␤-1 and TAK1, expressed at lower physiological levels, can also interact with one another. As shown in Fig. 5D, the 70-kDa endogenous TAK1 was specifically detected by anti-TAK1 Ab in the endogenous PP2C␤ immunoprecipitates from 293 cells, but not in the rabbit IgG immunoprecipitates.
It has recently been reported that IL-1 treatment of cells activates the JNK signaling pathway through activation of TAK1 (21). Therefore, we examined the ability of PP2C␤-1 to affect activation of TAK1 and AP-1 following IL-1 stimulation. We transfected 293IL-1RI cells with PP2C␤-1 and TAK1 and determined the effect of PP2C␤-1 expression on IL-1-induced mobility shift on SDS-PAGE and activation of TAK1. IL-1 treatment caused a slight mobility shift of TAK1 on SDS-PAGE, confirming our previous observation (23) (Fig. 8B, lower  panel). However, the coexpression of PP2C␤-1 totally reversed the mobility shift of TAK1. The expression of PP2C␤-1 also inhibited the IL-1-induced activation of TAK1 (Fig. 8B, upper  panel). Next, we transfected 293IL-1RI cells with PP2C␤-1 or/and PP2C␤-1(R/G) and assayed AP-1 activity using an AP-1-dependent luciferase reporter gene. PP2C␤-1 blocked IL-1induced AP-1 activity in a dose-dependent manner (Fig. 8C). However, inhibition of IL-1-induced AP-1 activation by PP2C␤-1 was reversed by cotransfection with PP2C␤-1(R/G) (Fig. 8D). Furthermore, ectopic expression of PP2C␤-1(R/G) enhanced IL-1-induced AP-1 activity in a dose-dependent manner (Fig. 8E). DISCUSSION MAPK cascades are intracellular signaling modules composed of three tiers of sequentially activating protein kinases: MKKK, MKK, and MAPK (1,2). Because phosphorylation of these components is essential for the activation of the MAPK cascades, protein phosphatases may be expected to play important roles in the regulation of these cascades. Indeed, we recently demonstrated that two major protein serine/threonine phosphatases, PP2C␣ and PP2C␤, inactivate the stress-activated JNK and p38 MAPK pathways (17). Furthermore, Takekawa et al. (14) showed that PP2C␣ inhibits the JNK and p38 cascades by dephosphorylating MKK4, MKK6, and p38. TAK1 is a member of the MKKK family and activates the JNK and p38 pathways. In this study we elucidated the role of PP2C␤ in TAK1-mediated signaling pathways.
We present several lines of evidence suggesting that PP2C␤ negatively regulates the TAK1 pathways by dephosphorylating and inactivating TAK1. First, ectopic expression of PP2C␤ inhibits the MKK4-JNK and MKK6-p38 pathways activated by TAK1 (Fig. 1). Second, it is known that the TAK1-binding protein TAB1 activates TAK1 by promoting its autophosphorylation (23). We found that PP2C␤ overexpression decreased TAB1-induced TAK1 autophosphorylation in vivo (Fig. 2). Third, PP2C␤ dephosphorylates and inactivates TAK1 in vitro (Fig. 3) but fails to dephosphorylate MKK6 (Fig. 4). Finally, PP2C␤ interacts with TAK1 but not with MEKK3, MKK4, MKK6, JNK, or p38 (Figs. 5 and 7). Collectively, these data are consistent with the idea that PP2C␤ suppresses TAK1-mediated signaling by associating with and dephosphorylating TAK1. Because TAK1 functions in various biological responses, including acting as a positive regulator of transforming growth factor-␤-and IL-1-induced signal transduction (18,21) and as a negative regulator in Wnt-induced signal transduction (22), it would be interesting to examine whether PP2C␤ contributes to the control of these physiological responses.
Coexpression of TAK1/TAB1 with PP2C␤-1 did not result in complete dephosphorylation of TAK1, as judged by the fact that the mobility of TAK1 is still slower than that of TAK1 ex- pressed by itself (Fig. 2). COS7 cells contain a substantial amount of free, endogenous TAB1. Therefore, we speculate that the reason for the incomplete dephosphorylation may be that the dephosphorylated TAK1 can be rephosphorylated, because both TAB1 and ATP are present in the cells. Alternatively, this may suggest that there are other phosphorylation sites in TAK1 that are not substrates for PP2C. TAK1 associates with PP2C␤ but not with PP2C␣ (Fig. 5). Thus, the interaction of TAK1 with PP2C␤ is rather specific. TAK1 is activated via autophosphorylation of Ser-192 in the activation loop between kinase domains VII and VIII. Mutation of TAK1 Ser-192 to Ala to create TAK1(S/A) abolishes both phosphorylation and activation of TAK1 (23). TAK1(S/A) has an affinity for PP2C␤ similar to that of wild-type TAK1 (Fig.  5C), indicating that phosphorylation of TAK1 is not required for its association with PP2C␤. This suggests that the association of TAK1 with PP2C␤ does not occur simply through affinity of the enzyme (PP2C␤) for its substrate (phosphorylated TAK1), but rather that PP2C␤ and TAK1 are stably associated. This may ensure appropriate localization of PP2C␤ and facilitate the specific and rapid deactivation of TAK1.
The central region of TAK1 is required for its association with PP2C␤ (Fig. 6). A similar region of TAK1 is involved in its association with TAB1 (20), which suggests that PP2C␤ might prevent the association of TAK1 with TAB1. However, this possibility is unlikely, because we did not observe any competition between TAB1 and PP2C␤ in their association with TAK1. 2 Consistent with this, endogenous TAK1 constitutively associates with TAB1 in the absence of ligand stimulation (23). Therefore, the minimum regions of TAK1 required for association with PP2C␤ and TAB1 must be different. It is still not clear whether PP2C␤ associates with TAK1 directly or indirectly. However, the observation that PP2C␤ fails to interact with TAB1 2 argues against the possibility that TAB1 mediates the association between PP2C␤ and TAK1.
To understand what role PP2C may play in regulating SAPK signaling pathways, it is important to determine how cellular PP2C activity is affected by extracellular stimuli. In fission yeast cells, the expression of Ptc1 is enhanced by hyperosmotic stress (29). In contrast, expression levels of PP2C␣ and PP2C␤-1 are not altered following stress treatment of cells (17). PP2C␣ has been shown to preferentially bind to the phosphorylated form of p38 and may function in the adaptive phase of the stimulation cycle to restore p38 to the inactive state following stimulation by stress (14). PP2C␤ may play an analogous role in maintaining TAK1 signaling. TAK1 mediates IL-1-induced JNK signaling (21), and ectopic expression of PP2C␤ blocks IL-1-induced AP-1 activation. PP2C␤(R/G), a catalytically inactive mutant, has a higher affinity for TAK1 than does wild-type PP2C␤ and acts as a dominant negative factor, antagonizing the inhibitory effect of wild-type PP2C␤ on IL-1induced AP-1 activation. Furthermore, ectopic expression of PP2C␤(R/G) enhances IL-1-stimulated AP-1 activation but does not cause constitutive activation of AP-1. 2 These results raise the possibility that PP2C␤ may down-regulate TAK1 activity after ligand stimulation. Because endogenous PP2C␤ constitutively associates with TAK1 (Fig. 5D), and ligand stimulation does not affect this association, 2 it is tempting to speculate that regulation of PP2C␤ enzymatic activity is involved in regulation of TAK1 signaling. Alternatively, PP2C␤ activity may be constitutive and serve to restore TAK1 to the inactive state following stimulation. Therefore it is important to determine whether the phosphatase activity of PP2C␤ is enhanced when cells are subjected to stress or treated with pro-inflammatory cytokines.
Takekawa et al. (14) recently reported that PP2C␣ dephosphorylates MKK4, MKK6, and p38 in vitro. In this study, we show that PP2C␤ dephosphorylates and inactivates TAK1. Thus, in mammalian cells, SAPK pathways are negatively regulated by multiple PP2C isoforms at different levels;  Fig. 3 were incubated with the indicated amounts of GST-PP2C␤-1(R/G) and/or GST-PP2C␤-1 for 30 min at 30°C. The proteins were separated by SDS-PAGE and analyzed by autoradiography. B, the expression plasmids of Flag-PP2C␤-1 and/or HA-TAK1 were transfected into 293 IL-1RI cells. After the treatment with IL-1 for 10 min, the cell lysates were subjected to immune complex kinase assay using MKK6 as the substrate (upper panel) or immunoblotted with anti-HA Ab (lower panel). C-E, 293IL-1RI cells were transfected with AP-1luciferase reporter plasmid with or without the indicated amounts of the expression plasmids for PP2C␤-1 (C), PP2C␤-1(R/G) and PP2C␤-1 (D), or PP2C␤-1(R/G) (E). After the treatment with IL-1 for 6 h, cells were lysed, and luciferase activities were determined and normalized on the basis of ␤-galactosidase expression. The data shown are the mean Ϯ S.D. (n ϭ 3).
inhibits the pathways at the TAK1 MKKK level, and PP2C␣ acts at the MKK and MAPK levels. In addition, two distinct groups of protein phosphatases other than PP2C also participate in the regulation of the SAPK pathways. The first group consists of the dual specificity phosphatases (also known as MAPK phosphatases) that inactivate MAPKs by dephosphorylating both tyrosine and threonine residues in the catalytic domains. Of the nine isolated MAPK phosphatases, M3/6 and MAPK phosphatase-5 have been shown to selectively dephosphorylate and inactivate p38 and JNK (30,31). The second group includes PP2A, which inactivates partially purified p38 kinase in vitro (32). Cells treated with the PP2A inhibitor okadaic acid show enhanced MKK6 activity in epithelial cells (27). These results suggest that PP2A may also negatively regulate SAPK pathways and raise the possibility that several different groups of protein phosphatases each negatively regulate distinct targets in SAPK pathways.