Dual Roles for the Phosphatase PPM1D in Regulating Progesterone Receptor Function*

Although protein phosphatase magnesium-dependent 1 δ (PPM1D) was initially characterized as a p53-regulated phosphatase responsible for inactivation of p38 MAPK and consequent inactivation of p53, its overexpression and amplification in human breast cancers led us to assess its role in steroid hormone action. We found that PPM1D stimulated the activity of several nuclear receptors including the progesterone receptor (PR) and estrogen receptor. Although p38 MAPK inhibited PR activity, PPM1D stimulation of PR activity was greater than that achieved by a chemical inhibitor of p38 MAPK, SB202190. This suggests an additional novel function for PPM1D. Consistent with this, the transcriptional activity of endogenous PR in MCF-7 breast cancer cells was preferentially inhibited by small interfering RNA for PPM1D; SB202190 failed to reverse the inhibition. Although PPM1D phosphatase activity was required for stimulation of transcriptional activity, the activity of a PR phosphorylation site null mutant was enhanced by PPM1D, indicating that PR is not the direct target. Additional studies revealed that PPM1D enhanced the intrinsic activity of p160 coactivators such as steroid receptor coactivator-1 and promoted the interaction between PR and steroid receptor coactivator-1 in a mammalian two-hybrid assay. Neither activity was induced by SB202190. Although PPM1D stimulated PR activity in part through inhibition of p38 MAPK, its primary action is novel and independent of p38 MAPK. Thus, we speculate that PPM1D promotes breast tumor growth both by inhibiting p53 activity and by enhancing steroid hormone receptor action.

Although protein phosphatase magnesium-dependent 1 ␦ (PPM1D) was initially characterized as a p53-regulated phosphatase responsible for inactivation of p38 MAPK and consequent inactivation of p53, its overexpression and amplification in human breast cancers led us to assess its role in steroid hormone action. We found that PPM1D stimulated the activity of several nuclear receptors including the progesterone receptor (PR) and estrogen receptor. Although p38 MAPK inhibited PR activity, PPM1D stimulation of PR activity was greater than that achieved by a chemical inhibitor of p38 MAPK, SB202190. This suggests an additional novel function for PPM1D. Consistent with this, the transcriptional activity of endogenous PR in MCF-7 breast cancer cells was preferentially inhibited by small interfering RNA for PPM1D; SB202190 failed to reverse the inhibition. Although PPM1D phosphatase activity was required for stimulation of transcriptional activity, the activity of a PR phosphorylation site null mutant was enhanced by PPM1D, indicating that PR is not the direct target. Additional studies revealed that PPM1D enhanced the intrinsic activity of p160 coactivators such as steroid receptor coactivator-1 and promoted the interaction between PR and steroid receptor coactivator-1 in a mammalian two-hybrid assay. Neither activity was induced by SB202190. Although PPM1D stimulated PR activity in part through inhibition of p38 MAPK, its primary action is novel and independent of p38 MAPK. Thus, we speculate that PPM1D promotes breast tumor growth both by inhibiting p53 activity and by enhancing steroid hormone receptor action.
Progesterone mediates its effects through the progesterone receptor (PR). 3 Encoded by a single gene, the two isoforms (PRA and PRB) are identical except that PRB contains an additional 164 amino acids (1). The forms are present in the same target cells in the human and for the most part are expressed at comparable levels; however, they exhibit distinct transcriptional activities (2)(3)(4) and play unique physiological roles (5,6). The actions of steroid receptors are regulated not only by the level of hormone but also by the levels/activity of coactivators that are recruited to target promoters to remodel chromatin and facilitate transcription (7). Steroid receptors and their coregulators are phosphoproteins whose activities are regulated by diverse cell signaling pathways (8 -12). Although several candidate kinases have been identified (13)(14)(15), less is known regarding the role of phosphatases in steroid receptor action. One candidate PR regulatory phosphatase that we investigated was protein phosphatase magnesium-dependent 1 ␦ (PPM1D), also known as Wip1 (wild type p53-induced phosphatase 1). PPM1D was discovered in a screen for ionizing radiation and p53-dependent target genes (16) and was originally characterized as a p53-induced target that acts to dephosphorylate and inactivate p38 mitogen-activated protein kinase (MAPK), leading to inactivation of p53 (17). PPM1D also abrogates cell cycle checkpoints by reducing p53 and Chk1 activities through direct dephosphorylation (18). PPM1D has been shown to be amplified and overexpressed in some human breast and prostate cancers, two steroid hormone-dependent cancers (19,20). PPM1D null mice are also less susceptible to mammary cancer (21). It has been shown recently that elimination of PPM1D in mice resulted in a variety of postnatal abnormalities in male mice such as reproductive organ atrophy and reduced fertility and longevity (22,23). The finding that PPM1D is overexpressed in breast cancer coupled with the observation that androgen receptor knock-out mice share a phenotype similar to PPM1D null mice with regard to disorganization of the seminiferous tubules (24) led us to ask whether PPM1D plays a role in the actions of steroid receptors. We show here that PPM1D stimulates the activity of steroid receptors and is required for optimal PR activity; this requirement is independent of its ability to inhibit p38 MAPK activity suggesting additional targets for PPM1D action. FIGURE 1. PPM1D stimulates the transcriptional activity of multiple nuclear receptors. A, HeLa cells were transfected with 1 ng of PRB, 1 ng of ER␣, or 0.5 ng of AR; 250 ng of the corresponding reporter GRE 2 -E1b-LUC (PR-and AR-responsive), ERE2-E1B-Luc (ER), or VDRE-tk-Luc (VDR); 50 ng CMV ␤-Gal; and 50 ng of vector or PPM1D plasmid. For measurement of endogenous VDR activity, cells were transfected with VDRE-tk-Luc, 50 ng of CMV ␤-Gal, and 50 ng of vector or PPM1D plasmid. After 24 h, cells were treated with ethanol vehicle, 10 nM R5020 (promegestone) (PR), 10 nM R1881 (methyltrienolone) (AR), 10 nM estradiol (ER), or 10 nM 1,25-dihydroxyvitamin D 3 (VDR). Luciferase activity was measured 48 h (Madison, WI). All competent cells were obtained from Promega. Pfu Turbo DNA polymerase was obtained from Stratagene. T4 DNA ligase was purchased from Invitrogen.
Cell Culture-HeLa and MCF-7 cells were obtained from the American Type Culture Collection (Manassas, VA) and were grown in Dulbecco's modified Eagle's medium (DMEM) in 10% fetal bovine serum. Cells were seeded in DMEM containing 10% charcoal-stripped fetal calf serum (Sigma) with penicillin and streptomycin (Invitrogen) at 200,000, 125,000, and 300,000 cells/well, respectively, in 6-well plates. All cells were maintained at 37°C with 5% CO 2 in a tissue culture incubator.
Site-directed Mutagenesis-Mutagenesis was performed following the protocol included with Pfu Turbo polymerase from Stratagene. To generate the S309A and S361A Cdc25B mutants the following mutagenic oligonucleotides were used: S309A, 5Ј-CGG CTC TTC CGC TCT  CCG GCC ATG CCC TGC AG-3Ј; S361A, 5Ј-GTC CTC CGC TCA  AAA GCA CTG TGT CAC GAT GAG-3Ј. Transient Transfection-Transient transfection of the cells using lysine-coupled inactivated adenovirus was carried out as described previously (38). Briefly for each well of a 6-well plate, the indicated amount of plasmid DNA was mixed with HBS (0.15 M NaCl, 0.02 M HEPES, pH 7.2) and then incubated with 10 8 virus particles. Thirty minutes later, additional poly-L-lysine (1.3 g of poly-L-lysine/g of DNA) was added to shrink the DNA on the viral surface. The virus-DNA complex was added to cells 30 min later and allowed to infect cells for 2 h in medium lacking serum. The medium was subsequently supplemented with charcoal-stripped serum to a final concentration of 5%. Twenty-four hours post-transfection, cells were treated with hormone (10 nM) for an additional 24 h and harvested as described below.
Transfection of Small Interfering RNA (siRNA)-MCF-7 cells plated at 200,000 cells/6-well plate were transfected with 100 pmol of control siRNA or siRNA against PPM1D. siRNAs were purchased from Ambion (Austin, TX), using predesigned siRNA against PPM1D (identification number 4785) and non-targeting control siRNA or SMART pool siRNA (M-004554-00) for PPM1D and Non-Specific Control SMARTpool reagent from Dharmacon (Lafayette, CO). Oligofectamine (Invitrogen) was incubated with DMEM at a 1:4 ratio at room temperature for 10 FIGURE 2. PPM1D is required for optimal PR activity. A, MCF-7 cells were transfected with 50 ng of vector control or wt PPM1D, 250 ng of GRE 2 -E1b-LUC, and 50 ng of CMV ␤-Gal and subsequently incubated with or without 10 nM R5020. Luciferase activity was measured 48 h post-transfection, and values were normalized using ␤-galactosidase. B, MCF-7 cells were transfected with 100 pmol of control siRNA or siRNA against PPM1D (Ambion) and 250 ng of GRE 2 -E1b-Luc. Cells were treated with 10 nM estradiol subsequent to transfection and 24 h later were treated with 10 nM R5020. Steady state levels were determined by Western blot, using equal amounts of protein determined by Bradford assay, probing for PPM1D, PR, and ␤-actin expression as control. C, 24 h later, luciferase activity was measured, and values were normalized using equal units of protein.
Each experiment was done a minimum of three times with three independent wells of cells for each variable in the luciferase assays. A representative experiment is shown with error bars indicating the S.D. Cntrl, control; RLU, relative luciferase units.
post-transfection, and values were normalized using ␤-galactosidase. B, PRA and PPM1D levels in extracts of the PRA samples from A were determined by Western blot, using equal amounts of ␤-galactosidase units, probing for PR, PPM1D, and ␤-actin expression as control. C, CMV Luc (10 ng), p53 (10 ng) with 250 ng of p53-LUC, or GRE 2 E1b-LUC (250 ng) were transfected in HeLa cells with 50 ng of vector or PPM1D and 50 ng of CMV ␤-Gal. Luciferase activity was measured and normalized using ␤-galactosidase. Each experiment was done a minimum of three times with three independent wells of cells for each variable in the luciferase assays. A representative experiment is shown with error bars indicating the S.D. Ctrl, control; D3, 1,25-dihydroxyvitamin D 3 ; E2, estradiol; RLU, relative luciferase units; WB, Western blot.
min. In a separate tube, the indicated amount of siRNA was incubated with 185 l/well DMEM. The Oligofectamine/DMEM mixture (15 l/well) was added to the siRNA mixture and incubated at room temperature for 15 min. Medium was aspirated off the MCF-7 cells, and 800 l of DMEM were added to each well with 200 l of the transfection mixture. Cells were incubated at 37°C for 4 h before 1 ml of DMEM containing 10% charcoal-stripped serum was added, and the cells were treated with 10 nM estradiol to induce PR expression. Twenty-four hours later, the progestin-responsive reporter GRE 2 -E1b-LUC was transfected into the cells, which were ϳ70% confluent. Per well, 3 l of TransIT-LT1 (Mirus, Madison, WI) was incubated with 100 l of warmed Opti-MEM medium (Invitrogen) for 10 min at room temperature. The GRE 2 -E1b-LUC reporter was added to the mixture (0.25 g/well) and incubated for 5-10 min at room temperature. The transfection mixture was then added directly to the cells (100 l). After 24 h, medium was replaced with 5% charcoal-stripped serum in DMEM containing 10 nM estradiol and 10 nM R5020. Twenty-four hours later, cells were harvested as described below, and luciferase assays were performed as well as Western blot analysis using equal amounts of protein as determined by Bradford assay.
Protein Extract Preparation and Reporter Gene Analysis-Luciferase assays were performed using reagents from Promega on a Monolight 2010 luminometer (Analytical Luminescence Lab, Ann Arbor, MI). The luciferase values were normalized to ␤-galactosidase levels unless otherwise noted. Cells were harvested in TEN (0.15 M NaCl, 0.01 M EDTA, 0.04 M Tris, pH. 8.0) at room temperature. The cells were pelleted at 12,500 rpm for 25 s. Cell pellets were resuspended in 250 mM Tris, pH 7.5 containing 0.4 M NaCl and subjected to three rounds of freezethaws. Protein levels were determined by Bradford assay (Bio-Rad), and ␤-galactosidase levels were determined using ortho-nitrophenyl-␤-Dgalactopyranoside (39). Reporter gene assays were performed a minimum of three times with each variable in triplicate, and a representative assay is shown.
Western Analysis-Volumes of cell extracts containing equal protein or ␤-galactosidase units were run on an SDS-PAGE gel and transferred to a nitrocellulose membrane using a semidry apparatus at 30 mA for 15 min. For the PR antibody 1294, the membrane was blocked in 1ϫ Trisbuffered saline ϩ 0.1% Tween 20 (TBST) ϩ 1% bovine serum albumin for 30 min and incubated with 0.5 g/ml antibody overnight at 4°C. The next day, the blots were washed three times with TBST and incubated with 1 g/5 ml rabbit anti-mouse IgG in TBST with 1% bovine serum albumin for 1 h at room temperature. Blots were washed three times with TBST and incubated with horseradish peroxidase-tagged antibody to the secondary antibody for 1 h. The blots were washed three times with TBST and once with Tris-buffered saline ϩ 0.3% Tween 20. The signals were detected by enhanced chemiluminescence (Amersham Biosciences). For ␤-actin and PPM1D antibodies, milk was substituted for bovine serum albumin during blocking. For primary blot incubation, a 1:30,000 dilution of ␤-actin antibody was used, and for PPM1D 1 g/ml antibody was used; both were in 1ϫ TBST with 1% milk. Blots were incubated for 1 h at room temperature, washed three times with TBST, then incubated with 1 g/10 ml donkey anti-rabbit horseradish peroxidase-tagged antibody for 1 h in TBST, and detected as described above.

PPM1D Enhances the Activity of Class One and Class Two Nuclear
Receptors-To determine whether PPM1D is capable of stimulating steroid receptor activity, expression vectors for the two PR isoforms, androgen receptor (AR), or estrogen receptor ␣ (ER␣) were cotransfected into HeLa cells with their corresponding luciferase reporter plasmids, and the effects of PPM1D on hormone-dependent induction of reporter activity were measured. The activity of endogenous vitamin D receptor (VDR) was measured by transfecting cells with a VDR-responsive reporter. Coexpression of PPM1D enhanced PRA and PRB activity 3-5-fold as well as the activities of AR, ER␣, and VDR (Fig. 1A). To exclude the possibility that the observed enhancement in activity was due to increased receptor expression, PRA levels in the samples from Fig. 1A were measured by Western blotting (Fig. 1B); no increase in PRA expression was observed. A similar enhancement of PR-dependent activity was seen in COS cells as well as on an MMTV-Luc reporter (data not shown) suggesting that PPM1D stimulation is not limited to a specific cell type or promoter.
Although PPM1D enhances the activity of all tested nuclear receptors, this is not a result of a general increase in transcription. As a control, the effect of PPM1D on the activity of a constitutively active CMV Luc construct was assessed, and no change in the transactivation was detected (Fig. 1C). As an additional control, we confirmed that PPM1D inhibits p53 activity as expected by transfecting HeLa cells with a p53 expression vector, a p53-responsive reporter, and PPM1D or vector control. As shown in Fig. 1C, PPM1D expression inhibited p53 activity ϳ50%. Although the basal activity of the GRE 2 -E1b-LUC reporter in the absence of added steroid receptor is quite low, as a final control we measured the effect of PPM1D on this activity but found essentially no effect (Fig. 1C).
Endogenous PPM1D Is Required for Optimal PR Activity-To determine whether PPM1D modulates activity of endogenously expressed PR, MCF-7 breast cancer cells (which express both PRA and PRB) were transfected with empty vector or PPM1D and the PR-responsive GRE 2 E1b-LUC reporter; a PPM1D-dependent increase in total luciferase activity was observed ( Fig. 2A). We next asked whether reducing the expression of endogenous PPM1D would affect PR function. MCF-7 cells were first transfected with 100 pmol of non-targeting control siRNA or 100 pmol of siRNA against PPM1D and treated 4 h posttransfection with 10 nM estradiol to induce PR expression. Twenty-four hours later, cells were transfected with a GRE 2 E1b-LUC reporter plasmid, and after 24 h, the medium was replaced and supplemented with 10 nM estradiol and 10 nM R5020. Twenty-four hours later cells were harvested and assayed for PPM1D protein expression (Fig. 2B) and PR activity (Fig. 2C). The specific PPM1D siRNA caused a significant decrease in PPM1D protein levels compared with the control siRNA  MARCH 17, 2006 • VOLUME 281 • NUMBER 11 without altering ␤-actin levels (Fig. 2B). Importantly PR activity measured using a PR-responsive reporter decreased almost to basal activity only in the presence of siRNA against PPM1D (Fig. 2C) indicating that PPM1D is required for optimal PR activity. Similar results were obtained with a second siRNA targeting PPM1D that was also used for studies in Fig. 10 (data not shown).
PPM1D Enhances the Activity of AF-1 and AF-2-Steroid hormone receptors have two common activation function domains, AF-1 and AF-2, utilized by receptor coregulators to bind to and coactivate the receptors (7,40). To define the domain or domains responsible for the increase in transcriptional activity of PR by PPM1D, constructs of PRB and PRA lacking the hormone binding domain but containing the remaining portions of PRA and PRB containing amino acids 164 -684 (NTA) and 1-684 (NTB), respectively, were made. Additionally a C/D/E construct coding for amino acids 545-933, containing just the DNA binding domain, hinge, and ligand binding domain, was tested. The truncated constructs were cotransfected with PPM1D or vector in HeLa cells and treated with hormone, and luciferase activity was measured. PPM1D increased the constitutive activity of the amino-terminal PRB fragment 4-fold (Fig. 3A) and the amino-terminal PRA fragment 2-fold (Fig. 3B). The addition of PPM1D to cells transfected with the C/D/E construct also caused a hormone-dependent increase in activity (Fig. 3C). These results suggest that PPM1D acts through both PR activation functions similar to some coactivators.
Phosphatase-dependent Modulation of PR by PPM1D-To determine whether stimulation of PR activity by PPM1D is dependent upon its phosphatase activity, a conserved aspartic acid residue at 307 in the catalytic domain of PPM1D was mutated to alanine to inhibit the enzymatic action of PPM1D (17). The phosphatase-deficient mutant was unable to enhance PRA activity indicating that the phosphatase activity of PPM1D is required (Fig. 4A).
To determine whether PPM1D acts by removing an inhibitory phosphorylation from PR, a PRA mutant (termed Ala 10) with Ala substitutions for Ser at all of the published PRA phosphorylation sites (13,41) as well as all of the additional (Ser/Thr)-Pro motifs (a total of 10 sites) was prepared (Fig. 4B), and the effect of PPM1D on its activity was assessed. The activities of the wild type and Ala 10 mutant were stimulated to similar extents (Fig. 4C) indicating that the primary stimulation is independent of phosphorylation of these sites. Furthermore PPM1D had no effect on the hormone-dependent mobility change of PRA on SDS gels that is also due to phosphorylation (Fig. 4C).
The PP2C group of phosphatases is distinguished from other phosphatases by its insensitivity to okadaic acid and a requirement for divalent cations (Mn 2ϩ or Mg 2ϩ ). PPM1D contains three distinct domains (23). The unique amino-terminal and carboxyl-terminal domains are not present in other eukaryotic PP2C proteins and do not share homology with other known proteins, suggesting an interface for proteinprotein interactions. The third domain consists of an evolutionarily conserved PP2C functional domain. Due to this sequence similarity, we sought to determine whether the function of PPM1D on PR was unique or due to the PP2C domain common among PP2C phosphatases. HeLa cells were transfected with PRA and PPM1D, PP2C␣, or PP1 (a phosphatase that does not contain a PP2C domain), and the resulting activity was measured. Whereas PPM1D stimulated the activity of PRA ϳ3-fold, neither PP2C nor PP1 altered PR activity (Fig. 4D).
Defining a Mechanism for PPM1D Action-Our studies suggest that PPM1D does not act by reducing PR phosphorylation. However, there are several additional possibilities. The primary target of PPM1D is p38 MAPK. If p38 MAPK inhibits PR activity, PPM1D might indirectly stimulate PR activity by inactivating p38 MAPK. Alternatively because PR activity can vary as a function of cell cycle (34), PPM1D may indirectly stimulate PR activity by counteracting the p38 MAPK-dependent inhibition of cell cycle progression. Finally and most novel, PPM1D may be able to act on coactivators or other components of the transcription complex independently of its effects on p38 MAPK.
MKK6-dependent Activation of p38 Inhibits PR Activity-To examine the role of the PPM1D target p38 MAPK in PR signaling, the specific upstream kinase MKK6 was used as it has been shown to activate all four family members of p38 MAPK (42). Transfection of increasing concentrations of constitutively active MKK6 (MKK6 EE) with PRA in the  presence of R5020 resulted in a dose-dependent decrease in receptor activity (Fig. 5A). In contrast, the dominant negative control MKK6 gene (MKK6 AA) modestly enhanced PR activity. Stimulation of p38 MAPK activity also decreased the activity of ER␣ greater than 50% (Fig.  5B) and AR greater than 60% (Fig. 5C) but had no effect on a constitutive CMV Luc reporter (Fig. 5D).
Because p38 MAPK inhibits PR activity, we asked whether a specific p38 MAPK inhibitor, SB202190, could mimic the stimulation of PR activity by PPM1D. Chemical inhibition of p38 MAPK was found to modestly stimulate receptor transactivation (1.5-2-fold) (Fig. 6, A and  B) similar to the dominant negative MKK6 AA (Fig. 5A). However, activation of PR by PPM1D appeared greater than the stimulation achieved by SB202190. To compare the two directly, the effects of SB202190 on PR activity were compared with the activity induced by increasing amounts of PPM1D. PPM1D was more efficacious than SB202190 in inducing PR activity (Fig. 6A) indicating that PPM1D may act both through the inhibition of p38 MAPK and through an additional, unidentified target. The specificity of the inhibition of PR activity by MKK6 EE was examined in Fig. 6B. In the presence of MKK6 EE, SB202190 restored PR activity to the level observed without MKK6 EE (Fig. 6B, lane 8 versus lane 2), but U0126 did not (lane 12). Consistent with its role as an inhibitor of p38 MAPK, PPM1D stimulation of PR activity was unaffected by cotransfection of constitutively active MKK6 EE (Fig. 6C). We next sought to better characterize the actions of p38 MAPK and asked whether another unrelated stimulator of PR activity, the coactivator SRC-1, could overcome the inhibition by MKK6 EE. As shown in Fig. 6D, the stimulation of PR activity by SRC-1 was strongly inhibited by MKK6 EE. Thus, the reversal by PPM1D is not a general property of stimulators of PR activity.
Next to determine whether inhibition of PR by p38 MAPK was due to PR phosphorylation, the PRA Ala 10 mutant (in which all candidate p38 MAPK phosphorylation sites had been mutated) was used. The activities of the wild type and mutant PR were comparably inhibited. Thus, the p38 MAPK target is not PR (Fig. 6E).
Rapid Inhibition of PR by p38 MAPK-Because elevated levels of p38 MAPK kinase activity can cause cell cycle arrest (20) and PR activity can vary with cell cycle (34), we next asked whether inhibition of PR activity occurs rapidly or with kinetics more consistent with cells undergoing cell cycle arrest. To rapidly and specifically activate p38 MAPK, cells were transfected with or without MKK6 EE, and some of the cells were simultaneously treated with SB202190 to prevent activation of p38 MAPK during the initial expression of the receptor and kinase plasmids. Twenty-four hours later cells were washed to remove SB202190, and medium was replenished. After 2 h in fresh medium (the time necessary for p38 MAPK to become fully active, not shown), the cells were treated with hormone and then harvested after an additional 6, 12, or 24 h. Cells containing PR and control vector that were pretreated with SB20190 exhibited a 4-fold hormone-dependent induction of reporter activity after 6 h of hormone treatment. The activity in the cells containing active p38 MAPK for 6 h displayed markedly reduced reporter activity (about 50% of cells without MKK6 EE) (Fig. 7A). Cells with active kinase for the entire time showed even lower activity. This rapid inhibition suggests a direct effect on PR signaling rather than actions secondary to cell cycle arrest. Control cells harvested at 12 and 24 h contained significantly higher levels of luciferase activity with minimal increases in cells with activated kinase (Fig. 7A). Analysis of PR expression by Western blot using equal units of protein showed no difference in PR expression nor was there a change in mobility of receptor (Fig. 7B).
Inhibition of PR by MKK6 Is Not Due to Inhibition of Cdc25B Activity-Because of the evidence that p38 MAPK inhibits receptor activity independently of cell cycle inhibition, we investigated other known targets of the kinase. Phosphorylation of Ser 309 by p38 MAPK can inhibit the activity of Cdc25B phosphatase, causing a cell cycle arrest in G 2 (42). Cdc25B is also a PR coactivator that functions independently of the cell cycle to enhance receptor function by recruiting chromatin-remodeling proteins (27). To determine whether p38 MAPK inhibited PR activity through inhibition of Cdc25B, we compared the ability of wt and mutant Cdc25B to stimulate PR activity in the presence of active p38 MAPK. When PR was cotransfected with MKK6 EE, PR activity decreased greater than 50% (Fig. 7C). When PR was cotransfected with wt Cdc25B, PR activity was enhanced as has been reported previously (27). The mutant Cdc25B, resistant to inhibition by p38 MAPK, behaves identically to wt Cdc25B and does not preferentially counteract the inhibition by MKK6. Therefore, the inhibition of PR activity by p38 MAPK is not due to its effect on Cdc25B.
A Role for PPM1D Independent of p38 MAPK-Our data suggest that PPM1D also enhances PR activity independently of its capacity to inhibit p38 MAPK. Because PR itself does not appear to be a direct target of PPM1D action, we next asked whether PPM1D influences the activity of PR coactivators. To determine whether PPM1D functions to activate PR by enhancing coactivator function, cells were transfected with PRA and PPM1D, SRC-1, or both. PPM1D increased PRA activity nearly 4-fold, whereas SRC-1 increased PRA activity about 3-fold under these conditions (Fig. 8A). Cotransfection of PPM1D and SRC-1 resulted in a 6-fold increase in PRA activity suggesting PPM1D might modulate SRC-1 activity. To test this, the intrinsic activity of p160 coactivators or the coregulator CBP was measured using Gal4 DNA binding domain coactivator fusion proteins. HeLa cells were transfected with Gal coactivator fusions, a Gal-responsive reporter, 17-mer Luc, with or without PPM1D. As shown in Fig. 8B, PPM1D increased the intrinsic activity of Gal-SRC-1 but had little effect on the Gal DNA binding domain (DBD) alone. Interestingly PPM1D also stimulated the intrinsic activity of SRC-2 (TIF2/Grip-1) and SRC-3 (Rac3/AIB1) but not CBP. In contrast, inhibition of p38 MAPK by SB202190 had no effect on the intrinsic activity of SRC-1 further supporting an additional mechanism of action for PPM1D (Fig. 8C). The finding that the stimulation of the intrinsic activity of SRC-1 is independent of the ability of PPM1D to inhibit p38 MAPK raised the question of whether the enzymatic activity of PPM1D was required. To test this, we compared the activation of Gal-SRC-1 by PPM1D and the catalytically inactive mutant D307A and found that only the wt PPM1D stimulated SRC-1 activity (Fig. 8D). Thus, PPM1D activity is required, although its target is unknown.

PPM1D Promotes Binding between SRC-1 and the PR Ligand Binding Domain (LBD) Measured Using a Mammalian Two-hybrid Assay-We
next asked whether PPM1D enhanced steroid receptor function through increased coactivator association with receptors. Mammalian two-hybrid analysis was performed using a Gal DBD fused to the LBD of PR, VP16 fused to SRC-1, and the 17-mer Luc reporter. In the presence of both SRC-1 and PR LBD, luciferase activity was induced in a liganddependent manner ϳ20-fold (Fig. 9A). PPM1D, but not the control vector, enhanced reporter activity 3-fold when PR LBD and SRC-1 were both expressed, whereas inhibition of p38 MAPK by SB202190 had no positive impact on the binding. As a negative control, PPM1D was transfected with Gal fused to VP16, and the results showed that PPM1D has no influence on its activity (Fig. 9B). Thus, the observed enhancement is not due to stimulation of the intrinsic activity of the VP16 activation domain. We next investigated whether PPM1D required its phosphatase activity to enhance the interactions between PR and SRC-1. As shown in Fig. 9C, the catalytically inactive PPM1D did not stimulate receptor-coactivator interactions suggesting that PPM1D requires its enzymatic activity and that there is a yet unidentified target of PPM1D phosphatase activity. Although these two-hybrid studies are consistent with a modest increase in the interaction between PR and SRC-1, we were unable to consistently detect enhanced interaction between full-length PR and SRC-1 in coimmunoprecipitation assays. The assay is sensitive enough to detect hormone dependence but may not be sensitive enough to detect the effects of PPM1D (20-fold for hormone and 3-fold for PPM1D). It is also possible that there is no difference in interaction between SRC-1 and full-length PR, which contains an additional amino-terminal interaction site for SRC-1 (43) that is lacking in the ligand binding domain construct.
PR Requires a Function of PPM1D Independent of p38 MAPK Inhibition-Shown previously in the siRNA experiments, PPM1D appears to be required for optimal PR function in MCF-7 breast cancer cells. One explanation for this may be that loss of PPM1D results in the constitutive activation of p38 MAPK. Indeed Bulavin et al. (44) found a greater level of p38 MAPK phosphorylation, indicative of its activation, in transformed mouse embryonic fibroblasts devoid of PPM1D versus transformed wild type cells. To determine whether the loss of PR activity in MCF-7 cells as a result of reduced PPM1D expression is dependent upon active p38 MAPK we asked whether PR activity could be restored in PPM1D-deficient MCF-7 cells by inhibiting p38 MAPK activity. Reduction of PPM1D expression by siRNA resulted in a greater than 80% loss of PR activity measured using a PR-responsive reporter (Fig.  10A). If this change in PR activity were strictly due to hyperactive p38 MAPK then inhibition of the kinase would restore PR activity; however, our data show that SB202190 was not able to block PR inhibition (Fig.  10A) when PPM1D expression was reduced (Fig. 10B). Thus, endogenous PPM1D plays a role in the activity of endogenous PR that is independent of its ability to suppress p38 MAPK activity.

DISCUSSION
Although there is substantial evidence for activating phosphorylations in the regulation of steroid receptor activity (10,11), the roles of phosphatases are less well defined. Treatment with okadaic acid, an inhibitor of PP1 and PP2a, stimulates PR activity (45). PP2a associates with ER inhibiting phosphorylation of Ser 118 , a site important for ligand- independent activation of ER (46). The Ser/Thr phosphatase PP5 inhibits glucocorticoid receptor activity (47). Although the phosphatase Cdc25B stimulates steroid receptor activity, the stimulation is independent of the phosphatase activity of Cdc25B (27). To our knowledge, this is the first example of a phosphatase whose activity is required for optimal activity of PR. Our study shows not only that overexpression of PPM1D stimulates PR activity, but reducing endogenous expression in MCF-7 breast cancer cells, a line that expresses substantial amounts of PPM1D, greatly reduces endogenous PR transcriptional activity.
The phosphatase activity of PPM1D is required for the transcriptional activation of PR. In searching for the mechanism by which PPM1D stimulates PR activity, we first asked whether its known function as an inhibitor of p38 MAPK activity is responsible for its effect on PR. We report that PR activity was inhibited by p38 MAPK but that the inhibition was independent of phosphorylation of candidate p38 MAPK sites in PR. Although PR exhibits cell cycle-dependent regulation of activity (34) and p38 MAPK causes a G 2 /M arrest (44,48), the inhibition does not appear to be second-ary to p38 MAPK-induced cell cycle arrest as the inhibition was too rapid to require cell cycle redistribution. Moreover substitution of a Cdc25B mutant that cannot be inhibited by p38 MAPK and therefore should promote cell cycle progression did not prevent the inhibition of PR function.
Although the target of p38 MAPK is unidentified, our comparison of the effects of PPM1D and a p38 MAPK inhibitor, SB202190, on PR function revealed that although PPM1D can inhibit p38 MAPK activity as reported previously, this inactivation was a minor component of its actions in stimulating PR activity. Our studies showed that SB202190 is unable to reverse the inhibition of PR activity elicited by reducing PPM1D expression in MCF-7 cells. The substrate specificity of PPM1D is not well defined. Yamaguchi et al. (49) have shown that PPM1D most efficiently dephosphorylates the threonine in a doubly phosphorylated peptide, pTXpY (where pT is phosphothreonine and pY is phosphotyrosine), as is found in p38 MAPK, but some of the sites in more recently identified substrates such as p53 and Chk1 contain phosphoserines without a nearby tyrosine (18). Thus, PPM1D has a broader range of substrates than originally reported. Although the phosphatase activity of PPM1D was required for stimulation of PR activity, we found that mutation of all known phosphorylation sites in PR had no effect on PPM1D-dependent stimulation. We were unable to detect a physical interaction between PR and PPM1D, suggesting that it is unlikely to be acting as a coactivator.
Our data suggest that SRC-1 and/or other p160 coactivators or their associated proteins may be the target of PPM1D action. The p160 coactivators form multiprotein complexes bringing numerous proteins that modify chromatin to target promoters enhancing transcription of target genes (50), and phosphorylation is a common means of regulating protein-protein interactions. The p160 coactivators are phosphoproteins (8,9,51), and SRC-1 is phosphorylated on numerous sites (51). There is increasing evidence that site-specific phosphorylation is a determining factor in which a transcription factor will recruit a coactivator from a limited pool of coactivators. We have shown recently that the activity of the cyclin-dependent kinase Cdk2 is required for PR transcriptional activity and for interaction between PR and SRC-1 (36). The interaction is independent of PR phosphorylation, but treatment of SRC-1 with phosphatase reduces its interaction with PR, and rephosphorylation of SRC-1 with cyclinA/Cdk2 restores the interaction (36). Phosphorylation of the coregulator p300 reduces its ability to interact with nuclear receptors but not with other classes of transcription factors (52). The stimulation of the intrinsic transcriptional activity of the p160 family members by PPM1D (Fig. 9) implies that PPM1D may dephosphorylate a site on SRC-1 and/or one of its interacting proteins enhancing the interaction of the proteins and consequent transcriptional activity. The PPM1D-dependent increase in activity in the mammalian two-hybrid assay using PR and SRC-1 suggests that there may also be a PPM1Dsensitive phosphorylation site on SRC-1 that reduces interaction between PR and SRC-1, although the difference was insufficient to detect under conditions of a coimmunoprecipitation assay.
The finding that PPM1D positively regulates the activity of estrogen, progesterone, and androgen receptors is intriguing in light of the reported overexpression of this phosphatase in breast and prostate cancer (19,20). Although the capacity to inhibit p53 function is likely a major factor in its actions in cancer cells, the ability to stimulate the activity of the steroid receptors that contribute to breast and prostate cancer growth is an additional previously unknown means by which PPM1D may contribute to tumor growth. Interestingly, Belova et al. (53) have just reported that a newly characterized chemical inhibitor of PPM1D inhibits MCF-7 cell growth in vitro; administration of the inhibitor in vivo also inhibits growth of MCF-7 xenografts. The finding that the actions of PPM1D on PR are predominantly independent of its ability to inhibit p38 MAPK raises the question of additional PPM1D targets. Lu et al. (25) reported that PPM1D can interact with and dephosphorylate uracil-DNA glycosylase suppressing base excision repair and more recently have reported that Chk1 and p53 are also PPM1D targets (18). Thus, PPM1D acts on multiple targets to regulate the growth of cells, and our studies reveal that steroid receptor signaling is one of its targets.