The dual specificity mitogen-activated protein kinase phosphatase-1 and -2 are induced by the p42/p44MAPK cascade.

Mitogen-activated protein (MAP) kinase phosphatase-1 (MKP-1) and MKP-2 are two members of a recently described family of dual specificity phosphatases that are capable of dephosphorylating p42/p44MAPK. Overexpression of MKP-1 or MKP-2 inhibits MAP kinase-dependent intracellular signaling events and fibroblast proliferation. By using specific antibodies that recognize endogenous MKP-1 and MKP-2 in CCL39 cells, we show that MKP-1 and MKP-2 are not expressed in quiescent cells, but are rapidly induced following serum addition, with protein detectable as early as 30 min (MKP-1) or 60 min (MKP-2). Serum induction of MKP-1 and MKP-2 is sustained, with protein detectable up to 14 h after serum addition. Induction of MKP-1 and, to a lesser extent, MKP-2 temporally correlates with p42/p44MAPK inactivation. To analyze the contribution of the MAP kinase cascade to MKP-1 and MKP-2 induction, we examined CCL39 cells transformed with either v-ras or a constitutively active direct upstream activator of MAP kinase, mitogen-activated protein kinase kinase-1 (MKK-1; MKK-1(SD/SD) mutant). In both cell models, MKP-1 and MKP-2 are constitutively expressed, with MKP-2 being prevalent. In addition, in CCL39 cells expressing an estradiol-inducible ΔRaf-1::ER chimera, activation of Raf alone is sufficient to induce MKP-1 and MKP-2. The role of the MAP kinase cascade in MKP induction was highlighted by the MKK-1 inhibitor PD 098059, which blunted both the activation of p42/p44MAPK and the induction of MKP-1 and MKP-2. However, the MAP kinase cascade is not absolutely required for the induction of MKP-1, as this phosphatase, but not MKP-2, was induced to detectable levels by agents that stimulate protein kinases A and C. Thus, activation of the p42/p44MAPK cascade promotes the induction of MKP-1 and MKP-2, which may then attenuate p42/p44MAPK-dependent events in an inhibitory feedback loop.

A major characteristic of protein phosphorylation is its reversibility. In the cell, a dynamic balance exists between phosphorylation and dephosphorylation, resulting from interplay between protein phosphatases and protein kinases. As a consequence, a modification of either component is likely to have an equally important impact on signal transduction (1)(2)(3).
p42/p44 MAPK are highly homologous ternary members of a group of ubiquitously expressed serine/threonine kinases that, in fibroblasts, are activated in response to all mitogenic stimuli (24). Sustained activation of the MAP kinase cascade is an absolute requirement for fibroblasts to exit from the quiescent G 0 state and to pass the restriction point in G 1 . Thus, blockade of MAP kinase signaling with antisense MAP kinase constructs or dominant-negative MAP kinase molecules (25) or overexpression of MKP-1 (26) all prevent cell cycle reentry. The MAP kinase family is exemplified by three distinct subtypes: p42/ p44 MAPK (27), p38 MAPK (28), and p46/p54 JNK (29), which are predominately cytoplasmic proteins in quiescent cells. Upon stimulation, p42/p44 MAPK and p46/p54 JNK translocate to the nucleus, where they may phosphorylate nuclear transcription factors and thus regulate gene transcription (30 -32). Full enzymatic activity of MAP kinase requires a dual phosphorylation on a Thr-X-Tyr motif, which is performed in vivo by specific upstream activators termed MAP kinase kinase (MKK-1/2, MKK-3/6, and MKK-4 for the p42/p44 MAPK , p38 MAPK , and p46/ p54 JNK enzymes, respectively) (33)(34)(35). Consequently, each MAP kinase family member is a candidate for dual dephosphorylation by MKPs. This raises the question of specificity between cognate kinase-phosphatase partners. However, previous studies have shown that MKP-1, MKP-2, MKP-3, and PAC1 (restricted to hematopoietic cell lines) are all able to inactivate p42/p44 MAPK and that specificity is correspondingly low (8,15,19,36).
Several MKP family members were identified by virtue of their being encoded by immediate-early genes (10,37). With the exception of Pyst1 (23), MKPs are not expressed in quies-cent cells and may be induced upon stimulation with agonists that include mitogens, oxidative stress, heat shock, and UV irradiation (9). Specificity of MKPs for the various MAP kinase members may therefore depend on their presence during a particular cellular event. An understanding of the mechanisms controlling MKP induction may explain how MAP kinase signaling pathways are regulated.
We report that two dual specificity phosphatases (MKP-1 and MKP-2) may be induced in resting CCL39 fibroblasts following serum exposure. We have attempted to correlate their induction with the activity status of signal transduction pathways involving different MAP kinase family members. To this end, we investigated the level of MKP-1 and MKP-2 in cells expressing either constitutively active or inducible forms of each element of the Ras/Raf-1/MKK module. Our results suggest that activation of the p42/p44 MAPK cascade is sufficient to promote the expression of MKP-1 and MKP-2. We propose that p42/p44 MAPK down-regulation when cells progress through G 1 is dependent on a feedback loop that involves MKP family members.

Materials
[␥-32 P]ATP and the enhanced chemiluminescence (ECL) immunodetection system were obtained from Amersham Corp.; antiserum E1B, which specifically recognizes p42 and p44 MAPK on Western blots, and antiserum Kelly, which specifically immunoprecipitates p42/p44 MAPK activity, were as previously detailed (38). Antisera specifically immunoprecipitating p44 MAPK were a kind gift of Dr. Sylvain Meloche (39). Anti-cyclin D1 antisera were a kind gift of Dr. Véronique Baldin. The GST-Jun-(1-79) and GST-ATF-2-(1-149) expression vectors were kind gifts of Dr. M. Karin and Dr. R. Davis, respectively, and were as described (40). The Flag-p46 JNK expression vector was a kind gift of Dr. B. Dérijard (35). The specific p38 MAPK inhibitor SB 203580 was supplied by SmithKline Beecham. The specific MKK-1/2 inhibitor PD 098059 was purchased from New England Biolabs Inc. Bovine myelin basic protein and bovine serum albumin were purchased from Sigma. Triton X-100 and Nonidet P-40 were from Pierce.

DNA Constructs and Expression in HEK 293 Cells
The EcoRI fragment of the full-length cDNA of CL100 (human MKP-1) previously introduced into the pECE vector (Stratagene) was digested by SmaI/XbaI and cloned into the pcDNAneoI mammalian expression vector (EcoRV and XbaI sites). The EcoRI/EcoRI fragment of the full-length cDNA of hVH-2 (human MKP-2) in the pBS vector was digested by BamHI/XhoI and cloned into the pcDNAneoI expression vector (BamHI and XhoI sites) (Invitrogen). HEK 293 cells were seeded at a density of 0.5 ϫ 10 6 cells/well (six-well plate) and transfected the following day by the classical calcium phosphate coprecipitation technique with 10 g of the corresponding construct (43). 48 h after transfection, cells were lysed for Western blot analysis with Laemmli sample buffer (45).

Preparation of Polyclonal Antiserum against MKP-1
A synthetic peptide corresponding to the last 12 C-terminal amino acids of 3CH134 protein (mouse MKP-1) conjugated to keyhole limpet hemocyanin was used for rabbit immunization. This antibody is referred to here as Alb-1. The specificity of Alb-1 was tested by Western blotting using cell lysates from HEK 293 cells transfected with the pcDNAneo/MKP-1 construct, the pcDNAneo/MKP-2 construct, or an empty vector.

Expression of MAP Kinase and MAP Kinase Phosphatase Family
Members in CCL39 Cells CCL39 cells were seeded at a density of 0.8 ϫ 10 6 cells/dish (60-cm plate) (day 0) and transfected the following day (day 1) by the classical calcium phosphate coprecipitation technique with 2 g of the relevant HA or Flag epitope-tagged kinase construct associated either with 25 g of empty vector or with 25 g of phosphatase-expressing vector. The cells were then serum-starved overnight (day 2), and on day 3, cells were stimulated by serum or anisomycin for 30 min. Kinase assays were performed as described below. For HA-p44 MAPK and HA-p38 MAPK , immunoprecipitation was performed with monoclonal anti-HA antibodies. For Flag-p46 JNK , immunoprecipitation was performed with monoclonal anti-Flag antibodies. In experiments to determine the effect of MKP-1 and MKP-2 on CCL39 cell growth, CCL39 cells were seeded at a density of 3 ϫ 10 5 cells/well (six-well plate) and transfected with plasmid encoding the Na ϩ /H ϩ antiporter NHE-3 (2 g) either with pcDNAneo alone or with pcDNAneo encoding MKP-1 or MKP-2 (20 g). Cells were selected by the established proton-killing technique (25) performed over a 2-week period, after which colonies were stained with Giemsa blue (10% in PBS) and counted.

Northern Blot Analysis
CCL39 cells were seeded in 10-cm plates and, when confluent, serum-starved for 24 h. Cells were then stimulated with the appropriate agonist and, following a suitable time period, washed twice with cold PBS and lysed with Bioprobe RNA preparation (acid phenol and guadininium chloride). RNA was prepared by phenol/chloroform extraction and sodium acetate precipitation. Following RNA separation (30 g of total RNA) on formaldehyde-agarose gels, RNA was transferred to nitrocellulose and probed by the technique of Church and Gilbert (44). The MKP-1 probe was obtained following purification from a StuI digestion (780 base pairs) of pcDNAneo/CL100. The MKP-2 probe was purified as a StuI/SacII digest (1055 base pairs) of pcDNAneo/hVH-2.

Western Blot Analysis
Cells were washed twice with cold PBS and lysed in Triton X-100 lysis buffer (50 mM Tris-HCl, pH 7.5, 100 mM NaCl, 50 mM NaF, 5 mM EDTA, 40 mM ␤-glycerophosphate, 200 M sodium orthovanadate, 10 Ϫ4 M phenylmethylsulfonyl fluoride, 1 g/ml leupeptin, 1 M pepstatin A, and 1% Triton X-100) for 15 min at 4°C. Insoluble material was removed by centrifugation at 12,000 ϫ g for 2 min at 4°C. Proteins from cell lysates (100 g) were separated by 10% SDS-PAGE and electrophoretically transferred to Hybond-C Extra membranes (Amersham Corp.) in 25 mM Tris-HCl and 0.19 M glycine. Membranes were blocked in Tris-buffered saline (25 mM Tris-HCl, pH 7.5, and 137 mM NaCl) containing 5% nonfat dry milk. The blots were then incubated with antiserum Alb-1 (1:300) in blocking solution for 2 h at room temperature. After washing in Tris-buffered saline and 0.1% Triton X-100, blots were incubated with horseradish peroxidase-conjugated goat anti-rabbit IgG (1:3000) in blocking solution for 1 h and revealed with ECL. Where indicated, the activity status of p42/p44 MAPK was determined by a mobility shift assay in which, following cell lysis, proteins were separated by SDS-PAGE (12.5% acrylamide, 0.0625% bisacrylamide) and Western blotting was performed with antiserum E1B, which preferentially recognizes p42 MAPK .
Immune Complex Kinase Assays p42/p44 MAPK -CCL39 cells or their derivatives were seeded in 12well plates and rendered quiescent at confluence by serum starvation for 24 h. Cells were stimulated in DMEM with 10% serum (CCL39) or in phenol red-free DMEM with either estradiol (1 M) or serum (5%) (⌬Raf-1::ER) at 37°C for the times indicated, prior to being washed with ice-cold PBS and lysed with Triton X-100 lysis buffer as described for Western blot analysis. Proteins from lysates (200 g) were incubated with specific anti-p44 MAPK antibodies preadsorbed to protein A-Sepharose-coated beads for 2 h at 4°C. Immune complexes were washed three times with Triton X-100 lysis buffer and twice with kinase buffer (20 mM HEPES, pH 7.4, 20 mM MgCl 2 , 1 mM dithiothreitol, and 10 mM p-nitrophenyl phosphate). p42/p44 MAPK activity was assayed by resuspending the final pellet in 40 l of kinase buffer containing 50 M [␥-32 P]ATP (5000 cpm/pmol) and 0.25 mg/ml myelin basic protein. The reaction was incubated for 10 min at 30°C and stopped by addition of Laemmli sample buffer (45).
p46/p54 JNK -Cells were deprived of serum for 18 h in DMEM and stimulated with FCS (10%), IL-1␤ (10 ng/ml), sorbitol (300 mM), or sodium arsenite (200 M). Cells were washed and lysed with the same protocol as that described for p42/p44 MAPK . Cleared lysates (500 g of protein) were incubated with glutathione-Sepharose coupled with 6 g of GST-Jun-(1-79) for 3 h. Complexes were washed three times with 150 mM Tris, pH 7.5, and 1% Nonidet P-40; once with PBS and 0.5 M LiCl; and twice with kinase buffer prior to the kinase reaction.
p38 MAPK -Following immunoprecipitation of p38 MAPK from cell lysates (400 g of protein), immune complexes were treated exactly as described for p42/p44 MAPK , with myelin basic protein being substituted in the kinase assay for GST-ATF-2 (5 g).
In each case, samples were heated at 95°C for 2 min and separated by 10% SDS-PAGE. Gels were exposed, and the relevant bands were quantified using a Fuji phosphorimager.

Protein Determinations
Protein determinations were evaluated using the bicinchoninic acid protein assay kit (Pierce) with bovine serum albumin as standard.

Data Reproducibility
All experiments were performed at least three times. Where indicated, data shown are representative of at least two additional experiments that gave qualitatively identical results.

Characterization of Rabbit Polyclonal Antiserum Directed against MKP-1 and MKP-2 Proteins-Previous studies that
have examined the expression of MKP family members have principally employed Northern blot analysis. In serum-stimulated (2 h), but not quiescent, CCL39 cells, two RNA species that migrate with the size expected of MKP-1 and MKP-2 (ϳ2.4 kilobase pairs) were detected using probes that were designed to specifically identify either MKP-1 or MKP-2 mRNA species, respectively (Fig. 1A). A more detailed analysis revealed that the two mRNA species that were identified with MKP-1 and MKP-2 probes were induced by serum with a time course similar to that previously noted for MKP-1 and MKP-2 (data not shown).
To specifically identify MKP-1 and MKP-2 proteins, we raised rabbit polyclonal antiserum directed against a synthetic peptide corresponding to the last 12 C-terminal amino acids of mouse MKP-1 (3CH134 protein) (37). This sequence has only one amino acid change from human MKP-1 (Lys to Gln) (10) and one amino acid change in the same position in MKP-2 (Lys to His in both rat and human sequences) (15,46). In contrast, the C-terminal regions of the other MKPs bear little or no primary structure similarity. Thus, we expected that our anti-MKP antisera would be capable of recognizing both MKP-1 and MKP-2, but not other MKP family members. In Western blots of HEK 293 cells transiently transfected with expression vectors encoding MKP-1 or MKP-2, but not with empty vector, two single major bands were identified by anti-MKP-1/2 antiserum Alb-1 with apparent molecular masses of 40 and 42 kDa, in close agreement with the calculated molecular mass from the primary sequence of MKP-1 and MKP-2, respectively ( Fig. 1B).
We next performed Western blotting on cell lysates derived from serum-deprived or serum-stimulated CCL39 cells. Following serum stimulation (3.5 h) of quiescent CCL39 cells, Alb-1 antisera recognized two major bands with the expected molecular mass of MKP-1 and MKP-2 (Fig. 1B). In addition, these two bands comigrated with MKP-1 and MKP-2 transiently expressed in HEK 293 cells. When immunoblotting was performed in the presence of the specific peptide used for rabbit immunization, but not with an unrelated peptide sequence, the two bands with apparent molecular masses of 40 and 42 kDa were no longer detectable. Additionally, serum-mediated induction of MKP-1 and MKP-2 was dose-dependent in nature (Fig. 1C). These data strongly suggest that our anti-MKP-1/2 antiserum is capable of specifically detecting the hamster homologues of MKP-1 and MKP-2 in CCL39 cells.
MKP-1 and MKP-2 Are Rapidly but Differentially Induced in Response to Serum: Correlation with p42/p44 MAPK Inactivation Time Course-To analyze the induction of MKP-1 and MKP-2 in CCL39 cells, we performed Western blotting on cell lysates from serum-stimulated cells ( Fig. 2A). In quiescent CCL39 cells, both MKP-1 and MKP-2 are undetectable. Upon addition of serum, MKP-1 and MKP-2 are induced, with MKP-1 present as early as 1 h after addition of serum. Comparison of a range of experiments shows that MKP-1 is often detectable as early as 30 min after addition of serum (data not shown). In contrast, MKP-2 is induced later, with detectable protein present at 2-3 h after serum addition. Each phosphatase is present for at least 14 h after serum addition. After 10 -12 h of serum addition, the CCL39 cell populations had started to enter S phase, correlating with the expression of cyclin D1. Thus, MKP-1 and MKP-2 are induced and present during G 0 -G 1 transition and S phase entry in CCL39 cells. As we have previously reported (38), serum stimulation of p44 MAPK is rapid, being maximal after 5 min. p44 MAPK then declines slowly, until 1-2 h following serum addition, where there is a significant loss of activity (Fig. 2B, lower panel) (47). This loss of activity correlates temporally with the induction and presence of MKP-1 protein and, to a lesser extent, with the production of MKP-2 (Fig. 2, upper panel). As most members of the MKP family are encoded by immediate-early genes, addition of protein synthesis inhibitors to quiescent CCL39 cells should block their induction. To examine the contribution of MKP-1 and MKP-2 to the inactivation of p42/p44 MAPK , we examined the time course of MAP kinase activity in the presence or absence of cycloheximide (Fig. 2C). As detailed in Fig. 2B, serum-stimulated p42/p44 MAPK activity is maximal at 5 min and then declines such that ϳ30% of the maximal activity elicited by serum (5-min time point) is present 4 h after addition of serum. However, in the presence of cycloheximide, the long-term inactivation of p42/p44 MAPK is reduced such that 54% of the maximal activity elicited by serum (5-min time point) is present 4 h after addition of serum. Additional experiments revealed that cycloheximide inhibited the serum-mediated induction of both MKP-1 and MKP-2 in a dose-dependent manner that correlated with the extent of protein synthesis inhibition (data not shown). Thus, although a considerable fraction of the p42/ p44 MAPK inhibitory activity is independent of cycloheximide treatment, immediate-early genes and probably MKP-1 and MKP-2 play a role in setting the level of p42/p44 MAPK activity in CCL39 cells.
MKP-1 and MKP-2 Block the Activation of p44 MAPK , p46 JNK , and p38 MAPK and Inhibit CCL39 Cell Proliferation-To explore the substrate specificity of two MKP family members, MKP-1 and MKP-2 expressed by CCL39 cells, we performed cotransfection assays with each phosphatase together with epitopetagged p44 MAPK , p38 MAPK , or p46 JNK (Fig. 3A). In response to serum, HA-p44 MAPK may be activated and can phosphorylate its substrate, myelin basic protein. However, following cotransfection with either MKP-1 or MKP-2, p44 MAPK is no longer activable. In a similar manner, in response to specific stress stimulus, such as anisomycin, both HA-p38 MAPK and Flag-p46 JNK are able to phosphorylate their respective substrates in the absence, but not in the presence, of either MKP-1 or MKP-2. Thus, the transient expression of both MKP-1 and MKP-2 attenuates the activation in vivo of each of the three kinases examined: p44 MAPK , p38 MAPK , and p46 JNK .
We have previously shown that deregulated expression of MKP-1 in CCL39 fibroblasts prevents cell cycle reentry (26). As MKP-2 is also able to inhibit p42/p44 MAPK in vivo (Fig. 3A), we expected that MKP-2 would also exert a strong antiproliferative effect. To examine this hypothesis, we transfected CCL39 cells with either MKP-1 or MKP-2 and determined the colonyforming ability of transfected cells (Fig. 3B). In the presence of both MKP-1 and MKP-2, colony formation and hence cell division are blocked by up to 80%, demonstrating that both MKP-1 and MKP-2 inhibit the proliferation of CCL39 cells.
Inactivation of Stress Kinases Does Not Require MKP-1 or MKP-2-The transient expression studies described above (Fig. 3) show that overexpression of MKP-1 or MKP-2 can inhibit p38 MAPK and p46 JNK activity. However, this does not unequivocally demonstrate that endogenous MKP-1 and MKP-2 are required for inactivation of either kinase. Hence, we examined the time course of activation of p46/p54 JNK in response to a range of agonists, together with the possible induction of MKP-1 or MKP-2 (Fig. 4). Activation of p46/p54 JNK by osmotic shock (sorbitol), sodium arsenite (equivalent to heat shock (48)), or IL-1␤ was slow in comparison with agonistmediated activation of p42/p44 MAPK (Fig. 2), peaking at ϳ30 min after addition of either sorbitol or IL-1␤. In the case of sodium arsenite and sorbitol, this activity was sustained, lasting for at least 3 h in the continual presence of agonist. In contrast, stimulation of p46/p54 JNK by IL-1␤ was transient, having returned to basal values within 2-3 h after addition of agonist. None of the agonists tested were able to induce MKP-1 or MKP-2 to detectable levels, in contrast to control cells, which were stimulated with serum. Thus, inactivation of p46/p54 JNK

FIG. 2. Time course of FCS-stimulated MKP-1 and MKP-2 induction, p44 MAPK activity, and effect of cycloheximide in CCL39 cells.
A, quiescent CCL39 cells in 12-well plates were stimulated for the times indicated with FCS (20%). Cells were lysed, and proteins (100 g) were separated by 10% SDS-PAGE and Western-blotted with both anti-MKP-1/2 and anti-cyclin D1 (cyc D1) antisera as described in the legend to Fig.  1. B, quiescent CCL39 cells in 12-well plates were stimulated for the times indicated with FCS (20%). Cells were then lysed; and proteins (100 g) were separated by 10% SDS-PAGE and Western-blotted with anti-MKP-1/2 antisera (upper panel), or p44 MAPK activity was determined as described under "Experimental Procedures" (lower panel). The autoradiograph is shown. C, quiescent CCL39 cells in 12-well plates were stimulated for the times indicated with FCS (20%) in the presence (E) or absence (f) of cycloheximide (CHX; 30 g/ml), which was added 15 min prior to serum addition. Cells were then lysed, and p44 MAPK activity was measured as described under "Experimental Procedures." At least three experiments were performed that gave qualitatively identical results. NS, not stimulated; MBP, myelin basic protein.
proceeds in the absence of detectable MKP-1 and MKP-2.
Serum Induction of MKP-1 and MKP-2 Does Not Require p38 MAPK Activity-p38 MAPK is weakly, although significantly activated by serum in CCL39 cells (data not shown) and could therefore play a role in MKP induction. Pretreatment of CCL39 cells with the specific p38 MAPK inhibitor SB 203580 (49,50) completely blocks the activation of p38 MAPK (51). However, following inhibitor pretreatment, the ability of FCS both to stimulate p42/p44 MAPK and to induce MKP-1 and MKP-2 was unimpaired (Fig. 5). Hence, activation of the p38 MAPK signaling pathway is not a requirement for the serum-mediated induction of MKP-1 and MKP-2.
MKP-1 and MKP-2 Are Constitutively Expressed in CCL39 Cells Expressing v-ras or MKK-1(SD/SD)-As both MKP-1 and MKP-2 are able to inactivate p42/p44 MAPK and agents that activate MAP kinase also provoke the induction of MKP-1 and MKP-2, we hypothesized that p42/p44 MAPK might regulate the induction of its own inhibitor. Activation of p42/p44 MAPK requires the sequential activation of Ras, Raf-1, and MKK-1/2, with p42/p44 MAPK being the last step in the kinase cascade. To examine the contribution of the MAP kinase cascade to the induction of MKP-1 and MKP-2, we analyzed CCL39 cells expressing constitutively active members of this pathway. In CCL39 cells expressing v-ras, p42/p44 MAPK is constitutively active (Ref. 52 and data not shown), and MKP-2 is constitutively expressed, with MKP-1 present at a much lower level, but still detectable. Addition of serum results in the induction of MKP-1, with no appreciable change in MKP-2 (Fig. 6). In addition to the p42/p44 MAPK cascade, Ras is also known to control signaling pathways such as those linked to phosphati- . Cells were lysed, and p46/p54 JNK activity was determined as described under "Experimental Procedures." In a parallel experiment (lower panels), cells were lysed, and proteins (100 g) were separated by 10% SDS-PAGE and Western-blotted with anti-MKP-1/2 antiserum as described in the legend to Fig. 1. Data are from a single experiment that was performed three times. dylinositol 3-kinase or Rac/Rho proteins (53). Hence, to specifically activate only p42/p44 MAPK , we analyzed CCL39 cells transformed by a mutated, constitutively active MKK (MKK-1(SD/SD)) (clone SS3 (54)). As with CCL39 cells transformed by v-ras, p42/p44 MAPK is constitutively active (54), MKP-2 is constitutively expressed, and MKP-1 is barely detectable. Addition of serum provokes the induction of MKP-1 in SS3 cells, with MKP-2 levels not modified to a significant degree (Fig. 6). Therefore, it appears that sustained p42/p44 MAPK activity is sufficient to promote the expression of both MKP-1 and MKP-2, with MKP-2 being preferentially induced.
Activation of Raf-1 Is Sufficient to Induce MKP-1 and MKP-2-The transformation of fibroblasts by the ectopic expression of oncogenic proteins may result in the secretion of growth factors or hormones, which may then stimulate the cell population in an autocrine/paracrine manner. Indeed, conditioned serum-free medium from v-ras-transformed, but not wild-type, CCL39 cells can modify signal transduction pathways when added back to wild-type CCL39 cells (data not shown). Hence, the constitutive expression of MKP-1/2 in v-ras-or MKK-1(SD/ SD)-transformed CCL39 cells may not be due solely to the constitutive activation of the p42/p44 MAPK cascade, but may involve additional signaling pathways. To rigorously assess the role of the p42/p44 MAPK cascade alone in the regulation of MKP induction, we used CCL39 cells expressing an inducible member of the MAP kinase signaling cascade, ⌬Raf-1::ER (42,55). This chimeric protein has been shown to be conditionally activable by exposure to the estrogen analogue estradiol and to be able to specifically activate p45 MKK-1 and p42/p44 MAPK with no interference with stress MAP kinase pathways (55). In ⌬Raf-1::ER cells, estradiol addition potently activates p42/ p44 MAPK (Ref. 42 and data not shown) and promotes MKP-1 and MKP-2 expression (Fig. 7), whereas in parental cells, estradiol alone has no effect on p42/p44 MAPK activities and MKP-1/2 expression (data not shown). The ability of the ⌬Raf-1::ER construct to induce MKP-1 and more markedly MKP-2 is aug-mented in the presence of serum. Hence, activation of Raf-1, MKK-1, and, by consequence, p42/p44 MAPK is sufficient to induce expression of MKP-1 and MKP-2.
Serum-mediated Induction of MKP-1 and MKP-2 Requires MKK-1/2 Activity-The experiments above (Fig. 7) show that activation of Raf-1 is sufficient to induce MKP-1 and MKP-2. To determine whether activation of p42/p44 MAPK is necessary for the serum-mediated induction of MKP-1 and MKP-2, we employed a specific inhibitor of MKK-1/2, PD 098059 (56,57). In the presence of PD 098059 (10 M), the ability of serum to stimulate p42 MAPK is inhibited by 70%, and the induction of MKP-1 and MKP-2 is significantly, although not completely attenuated (Fig. 8). Higher concentrations of PD 098059 are able to completely block the activation of p42/p44 MAPK and the induction of MKP-1 and MKP-2, but also inhibit protein synthesis. 2 However, this inhibitor, when used at a concentration of 10 M, allows us to conclude that p42/p44 MAPK are required for full serum-mediated induction of MKP proteins in CCL39 cells.
MKP-1 and MKP-2 Are Differentially Induced by Protein Kinase C-and cAMP-elevating Agents-Addition of the tumorpromoting agent PMA to fibroblasts leads to the stimulation of protein kinase C and p42/p44 MAPK , but not p38 MAPK and p46/ p54 JNK . When CCL39 cells are treated with PMA, MKP-1, but not MKP-2, is transiently induced, with maximal induction evident 1 h after addition of agonist (Fig. 9A, upper panel). To determine whether protein kinase C pathways are also associated with the serum-dependent induction of MKP-1/2, we pretreated quiescent cells with a specific inhibitor of protein kinase C enzymes, GF 109203X (58), and followed the serummediated induction of MKP-1 and MKP-2. In addition to attenuating the ability of PMA to induce MKP-1 (Fig. 9B), the serum-mediated induction of MKP-1 and MKP-2 is also reduced by pretreatment with GF 109203X. In addition to an AP-1 element present in the MKP-1 promoter, two cyclic AMPresponsive elements (CRE) are evident. Sustained activation of protein kinase A in CCL39 cells leads to the induction of MKP-1, with MKP-2 undetectable over a 3-h time course. Maximal induction of MKP-1 is evident after 1 h of agonist addition. The level of induction of MKP-1 is significantly lower than that elicited by serum (Fig. 9A, lower panel). As activation of protein kinase A does not stimulate the p42/p44 MAPK pathway in CCL39 cells, this suggests that although full induction of MKP-1 and MKP-2 requires p42/p44 MAPK activation, p42/ p44 MAPK activity is not an absolute requirement.  5. Induction of MKP-1 and MKP-2 does not require p38 MAPK activity. CCL39 cells were seeded in 12-well plates and rendered quiescent when confluent. Cells were then pretreated or not with the specific p38 MAPK inhibitor SB 203580 for 18 h prior to addition or not of FCS (10%) for a further 3 h. Cells were then lysed; and p42/p44 MAPK activity was determined following immunoprecipitation as described under "Experimental Procedures" (upper panel), or proteins (100 g) were separated by 10% SDS-PAGE and Western-blotted with anti-MKP-1/2 antiserum as described in the legend to Fig. 1 (lower  panel). were seeded in 12-well plates and rendered quiescent when confluent. Cells were then stimulated with 5% FCS, 1 M estradiol (Est), or the two together for 3 or 5 h. Cells were lysed, and proteins (100 g) were separated by 10% SDS-PAGE and Western-blotted with anti-MKP-1/2 antiserum as described in the legend to Fig. 1. Data are from a single experiment that was performed two times.

DISCUSSION
Dual phosphorylated p42/p44 MAPK is an excellent substrate for the MKP family of dual specificity phosphatases in vitro (9), and all MKP family members tested have been shown to inactivate p42/p44 MAPK in vivo. At least eight MKPs are known to exist in mammalian systems (see the Introduction), which, together with the identification of additional MAP kinase family members, reflects signaling complexity (59).
MKPs, with the exception of Pyst1 (23), are the product of immediate-early genes and, based on Northern analysis, share overlapping tissue distributions (17,46). Purified MKPs are constitutively active (15,60,61). Although this does not rule out the possibility that post-translational regulation of MKP activity occurs, it suggests that the principal point of MKP regulation is at the level of transcription. Hence, the specificity of interaction between MAP kinase and MKP family members may depend on the specific induction of one or more MKP. A detailed analysis of the factors required for induction of MKPs would increase our understanding of their physiological role.
We show that MKP-1 and MKP-2 are expressed in the well established CCL39 fibroblast cell line (Fig. 1). Both phosphatases are induced by serum, albeit with a different time course of induction, suggesting that the mechanisms involved in their induction may not be identical (Fig. 2). Induction of MKP-1 and, to a lesser extent, MKP-2 correlates with an attenuation of p44 MAPK activity. It is therefore possible that p44 MAPK is specifically targeted by MKP-1 in CCL39 cells. However, it should be noted that in contrast to NIH3T3 fibroblasts (8) and Rat-1 cells (data not shown), where there is very little inactivation of p42/p44 MAPK in the presence of cycloheximide, in CCL39 cells, ϳ50% of the p42/p44 MAPK inactivating activity is insensitive to cycloheximide and hence does not involve the majority of MKP family members. Overexpression of either MKP-1 or MKP-2 completely blocks the activity of p44 MAPK and two members of the stress kinase family, p38 MAPK and p46 JNK . Thus, both MKP-1 and MKP-2 are potent inhibitors of CCL39 cell cycle reentry (Fig. 3). We cannot conclude that this block is entirely due to p42/p44 MAPK inhibition or is a result of inhibition of p42/p44 MAPK , stress kinases, and some of the more recently identified MAP kinase family members. However, it is important to note that overexpression of MKP-1 does not result in a complete loss of substrate specificity as we have previously shown that activation of p70 S6K occurs normally in cells that overexpress MKP-1 (42). Our results are not in total agreement with those reported by Chu et al. (62), who have demonstrated by a similar transient transfection technique of several cell types that while MKP-1 dephosphorylates p42 MAPK , p38 MAPK , and p54 JNK , MKP-2 is more discriminating and will dephosphorylate p42 MAPK and p54 JNK , but not p38 MAPK . This discrepancy may arise from differences in the expression levels of each phosphatase.
To define specificity between endogenous MAP kinases and MKPs, we examined the ability of a range of agonists known to stimulate either the mitogenic p42/p44 MAPK pathway or the stress kinase pathways to induce MKP-1 or MKP-2. In apparent contradiction to previous reports based on Northern blot analysis (63), we found that none of the stress agents tested, IL-1␤, osmotic shock, or sodium arsenite, were capable of inducing MKP-1 or MKP-2 to detectable levels. However, with the exception of IL-1␤, all of these agents are potent inhibitors of protein synthesis in CCL39 cells. Hence, it is not surprising that we were unable to detect protein expression of two immediate-early genes. In addition, protein synthesis inhibition has been shown to up-regulate immediate-early gene mRNA induction, most probably through an increase in mRNA stability (64). Thus, an analysis of MKP mRNA induction is response to stress agents may be difficult to interpret. Stimulation of CCL39 stress pathways with the above agonists, in particular IL-1␤, resulted in only a transient activation of both p46/ p54 JNK and p38 MAPK (data not shown). As we were unable to detect induction of MKP-1 or MKP-2 in response to any stress agonists tested, we conclude that neither phosphatase is involved in the inactivation of stress kinases following addition of these agonists alone (Fig. 4). Unfortunately, it is not possible to block production of MKP family members with protein synthesis inhibitors and then follow a time course of stress kinase activity in response to agonist stimulation, as an inhibition of protein synthesis by itself activates stress kinases (Ref. 65 and data not shown). The inhibition of p38 MAPK failed to modify serum-stimulated induction of MKP-1 and MKP-2, demonstrating that activation of this stress kinase is not a require- FIG. 8. Serum-mediated induction of MKP-1 and MKP-2 requires p42/p44 MAPK activity. CCL39 cells were seeded in 12-well plates and rendered quiescent when confluent. Cells were then stimulated or not with FCS (10%) in the presence or absence of the MKK-1 inhibitor PD 098059 (10 M), which was added 30 min prior to FCS addition. Cells were lysed; and p42 MAPK activity was determined by a mobility shift assay (upper panel), or proteins (100 g) were separated by 10% SDS-PAGE and Western-blotted with anti-MKP-1/2 antiserum as described in the legend to Fig. 1 (lower panel). Western blots representative of three performed that gave qualitatively identical results are shown. ment for MKP-1 and MKP-2 induction. Whether stress kinase inactivation in response to IL-1␤ stimulation requires the expression of other members of the MKP phosphatase family remains to be determined.
Stimulation of CCL39 cells results in the activation of a range of signaling pathways, one or more of which lead to the induction of MKP-1 and MKP-2. To fully analyze the contribution of the MAP kinase cascade to the induction of MKP-1 and MKP-2, we employed three different approaches: (i) CCL39 cells transformed by constitutively active components of the MAP kinase pathway; (ii) CCL39 cells expressing an inducible Raf-1 construct, ⌬Raf-1::ER; and finally, (iii) a chemical inhibitor of MKK-1, the direct upstream activator of p42/p44 MAPK . We thus identified the activation of MKK-1 as being sufficient to strongly induce MKP-2, with MKP-1 induced more weakly. A similar pattern of induction was obtained in cells transformed by v-ras (Fig. 6). These experiments suggest that activation of the MAP kinase pathway is more tightly linked to the induction of MKP-2 than of MKP-1. However, one should be careful in analyzing signal transduction pathways in transformed cell lines. Hence, in untransformed CCL39 cells expressing ⌬Raf-1::ER (Fig. 7), we show that activation of the Raf/MKK/ MAP kinase module is sufficient for the induction of MKP-1 and MKP-2. These results are similar to those of Krautwald et al. (66), who have shown that expression of v-raf in macrophages can lead to the induction of mRNA encoding MKP-1.
Although activation of the p42/p44 MAPK pathway induces MKP-1 and MKP-2, it seems that additional pathways may be implicated in their regulation. MKP-1, but not MKP-2, is detectably induced in CCL39 cells in response to protein kinase C-and protein kinase A-activating agents (Fig. 9). In addition, in cultured glomerular mesangial cells, atrial natriuretic peptide, which does not activate the p42/p44 MAPK cascade, is able to induce MKP-1 expression (67).
The promoter sequence of the genes encoding both MKP-1 and pac1 have recently been described (14,68,69), providing information regarding the intracellular events controlling their transcription. An analysis of the PAC1 gene in murine B-and T-lymphocytes demonstrates that up-regulation of PAC1 transcription is mediated via an AP-2-and an E-box-binding protein. Furthermore, induction of PAC1 in response to both v-ras and v-raf is attenuated following transfection of "dominantnegative" p42 MAPK , thus highlighting a role for the MAP kinase cascade (70). In addition to an AP-2 element and an E-box, the MKP-1 promoter contains an AP-1-binding site, which may respond to phorbol esters (71), and two CRE elements, which may bind and have their activity modified by CRE-binding protein family members including c-Jun, CREM, and ATF-2 (72). We have recently shown that activation of the Raf-1::ER chimera in CCL39 cells may lead to a p42/p44 MAPK -independent activation of p70 S6K (42). As activation of p70 S6K can lead to the activation of CREM-and CRE-dependent gene expression (73), it is possible that the induction of MKP-1 and MKP-2 by Raf-1::ER in CCL39 cells is a consequence of both p42/p44 MAPK and p70 S6K activation. However, this is unlikely as insulin, a potent stimulator of both p70 S6K and protein synthesis in CCL39 cells, fails to induce either MKP-1 or MKP-2 (data not shown).
In the yeast Schizosaccharomyces pombe, the MAP kinase homologue Spc1 is activated in response to osmotic stress and is inactivated by its cognate phosphatase, Pyp2 (74). Interestingly, Pyp2 is also transcriptionally induced by osmotic stress, indicating that a Spc1-Pyp2 negative feedback loop exists (41).
In agreement with this model, we suggest that p42/p44 MAPK down-regulation may depend on a similar negative feedback loop involving MKP family members.
Finally, we have shown that MKP-1 and MKP-2 are principally induced in CCL39 cells by the MAP kinase cascade, but maximal induction involves the interplay of at least one additional signaling pathway. However, we have been unable to address whether the endogenous level of expression of MKP-1 or MKP-2 is sufficient to dephosphorylate p42/p44 MAPK and whether p42/p44 MAPK is preferentially dephosphorylated by endogenous MKP-1 or MKP-2. In an attempt to address the latter question, we are adopting an antisense strategy to specifically target and prevent the induction of endogenous MKP-1 and MKP-2. Current research in this laboratory is aimed at developing cell models in which MKPs are maximally inducible in the absence of activation of the MAP kinase pathway, which may allow us to answer the former question.