Deletion of the Dual Specific Phosphatase-4 (DUSP-4) Gene Reveals an Essential Non-redundant Role for MAP Kinase Phosphatase-2 (MKP-2) in Proliferation and Cell Survival*

Mitogen-activated protein kinase phosphatase-2 (MKP-2) is a type 1 nuclear dual specific phosphatase (DUSP) implicated in a number of cancers. We examined the role of MKP-2 in the regulation of MAP kinase phosphorylation, cell proliferation, and survival responses in mouse embryonic fibroblasts (MEFs) derived from a novel MKP-2 (DUSP-4) deletion mouse. We show that serum and PDGF induced ERK-dependent MKP-2 expression in wild type MEFs but not in MKP-2−/− MEFs. PDGF stimulation of sustained ERK phosphorylation was enhanced in MKP-2−/− MEFs, whereas anisomycin-induced JNK was only marginally increased. However, marked effects upon cell growth parameters were observed. Cellular proliferation rates were significantly reduced in MKP-2−/− MEFs and associated with a significant increase in cell doubling time. Infection with adenoviral MKP-2 reversed the decrease in proliferation. Cell cycle analysis revealed a block in G2/M phase transition associated with cyclin B accumulation and enhanced cdc2 phosphorylation. MEFs from MKP-2−/− mice also showed enhanced apoptosis when stimulated with anisomycin correlated with increased caspase-3 cleavage and γH2AX phosphorylation. Increased apoptosis was reversed by adenoviral MKP-2 infection and correlated with selective inhibition of JNK signaling. Collectively, these data demonstrate for the first time a critical non-redundant role for MKP-2 in regulating cell cycle progression and apoptosis.

The amplitude and duration of MAP kinase signaling within a specific subcellular compartment are key features in the integration of extracellular stimuli and their effects on cellular (1). Three main MAP kinase groups, the ERKs, JNK, and p38 MAP kinases, are involved in regulating functions such as proliferation, apoptosis, and differentiation in response to growth factors, peptide hormones, stress, and infection (2). Perturbations in MAP kinase signaling are features of several different types of diseases including several types of cancers (3), diabetes (4), atherosclerosis (5,6), and immune disorders.
The kinetics of MAP kinase activation are strictly controlled principally by the mitogen-activated protein kinase phosphatases (MKPs), 3 a family of at least 10 dual specific phosphatases (DUSPs) that function to terminate MAP kinase signaling within a defined subcellular location (7). They share a common C-terminal catalytic domain and an N-terminal non-catalytic domain containing the MAP kinase interaction motif (8). Each isoform has unique yet overlapping features including substrate specificity, subcellular distribution, and factors regulating induction. For example, MKP-1 is a nuclear DUSP of the type 1 class and selective for all three major MAP kinases in vitro, whereas MKP-3, a type II DUSP, is a cytosolic phosphatase selective solely for ERK over the other kinases (7). Due to effects upon MAP kinase signaling, pertubations in the MKPs have been implicated principally in cancer (9). However, more recently, a role has been established in inflammation (10) and some cardiovascular disorders (11).
One poorly studied MKP is MKP-2 (12). This DUSP (DUSP-4) is a member of the type 1 family and has been shown to be induced in response to a number of stimuli including phorbol esters and growth hormones (12)(13)(14). Nuclear targeting is regulated by two distinct nuclear targeting sequences (15). Substrate specificity for ERK and JNK was originally demonstrated in vitro (16); however, selective inhibition of JNK has been implicated in cellular studies (17,18). Although MKP-2 has been recently regarded as a surrogate for the more well described MKP-1, recent cellular studies demonstrate a role in protection against apoptosis (17) and in senescence (19). However, there is still a lack of information describing the function of DUSP-4 in different cell types, in particular regarding substrate selectivity in vivo.
We have recently developed a DUSP-4 deletion mouse model and demonstrated a novel immunological phenotype in vivo (20). Using embryonic fibroblasts from DUSP-4 deletion mice, we now examine the effect of deletion upon MAP kinase signaling and growth parameters. We find that despite very moderate increases in ERK and JNK activity, MKP-2 deletion has profound effects upon cellular proliferation. In particular, we identify a role for MKP-2 in G 2 /M phase transition and demonstrate that MKP-2 plays a role in cell survival in response to the apo-* This work was supported by a studentship from the Nigerian Government and the University of Strathclyde (to A. L.) and a scholarship from government of Saudi Arabia (to S. A.). 1  ptotic stimulus anisomycin. These data demonstrate that MKP-2 is a non-redundant DUSP and has overlapping but distinct functions relative to the prototypic DUSP, MKP-1.

EXPERIMENTAL PROCEDURES
DUSP-4 deletion mice were generated in collaboration with Geno-way and have been genetically characterized previously (20). Adenoviral MKP-2 was generated by Vector Biolabs and used previously (21). All reagents were from Sigma (Poole, Dorset, UK) unless otherwise stated. Antibodies were purchased as follows. Rabbit polyclonal anti-JNK-1 (S-18, full length), anti-ERK-2 (K-23), and cyclin B1 antibodies were all purchased from Santa Cruz Biotechnology (Santa Cruz, CA). Phospho-ERK, anti-phospho-JNK1/2, p-cdc-2, and caspase-3 were purchased from Cell Signaling Technology. Anti-phospho-p38 was purchased from BIOSOURCE (Nivelles, Belgium). Anti-phospho-␥H2AX was purchased from Millipore. HRP-conjugated antirabbit antibody was purchased from Amersham Biosciences (Little Chalfont, UK). HRP-conjugated anti-mouse and conjugated anti-rabbit antibodies were purchased from Jackson ImmunoResearch Laboratories (Luton, Bedfordshire, UK). Phycoerythrin-annexin V apoptosis detection kit was purchased from BD Biosciences. All other materials used were of the highest commercial grade available and were obtained from Sigma and Invitrogen (Paisley, UK).
Cell Culture-Mouse embryonic fibroblasts (MEFs) were isolated from MKP-2 Ϫ/Ϫ or MKP-2 ϩ/ϩ mice (backcross 5) at 13.5-day post-coitum mice in compliance with British Home Office regulations. The cells were cultured in Dulbecco's modified Eagle's medium (DMEM) containing 10% FCS, L-glutamine, penicillin, and streptomycin, at 37°C, 5% CO 2 . All experiments were performed within three passages. Cardiac fibroblasts (CFs) were prepared from the ventricles of adult mice (8 -10 weeks old) from C57Bl/6 MKP-2 ϩ/ϩ and MKP-2 Ϫ/Ϫ (weighing between 20 and 30 g) as described previously (22). After 4 -5 days, subconfluent cultures were passaged by trypsinization and replated into 12-well plates. Passage 1 cells were used in all experiments. In all experiments, DMEM containing 20% FBS was washed out, and the cells were equilibrated in serum-free DMEM overnight before agonist stimulation. The purity of fibroblast cultures was established by immunofluorescence staining using anti-smooth muscle ␣ actin antibody. Bone marrow mouse macrophages were cultured as outlined previously (20). G 1 Synchronization-MEFs were grown in 3.5-cm dishes in DMEM containing 10% FCS, L-glutamine, penicillin, and streptomycin to 40% confluency. Thymidine was added to a final concentration of 2 mM in the medium. The cells were incubated at 37°C for 19 h. After the incubation, cells were washed with PBS three times, and fresh media were added without thymidine and incubated for 9 h at 37°C. Then thymidine was added again to a final of 2 mM and incubated for another 16 h. Cells were washed with PBS three times, and fresh media were added. Cells were harvested at different time points, and the cell cycle profile was analyzed by Western blotting.
Cell Cycle Analysis-Cell cycle profiles were analyzed by staining intracellular DNA with propidium iodide followed by flow cytometry using the BD FACSDiva software, BD Biosci-FIGURE 1. Serum mediated induction of MKP-2 in wild type but not in MKP-2 ؊/؊ fibroblasts. Confluent MEFs were rendered quiescent for 24 h in serum-free medium. In A, cells were incubated with 10% FCS for the times indicated. Cell extracts were analyzed by Western blotting using anti-MKP-2 and ERK1/2 antibodies as described under "Experimental Procedures." Each blot is representative of five others. In B, gels were quantified using densitometry, and each value represents the mean Ϯ S.E. of at least four experiments. **, p Ͻ 0.01 and ***, p Ͻ 0.001 versus 0 hrs. Fold stim., -fold stimulation. FIGURE 2. ERK mediates induction of MKP-2 in wild type MEFs. Confluent MEFs were quiescent for 24 h in serum-free medium, and vehicle (Ϫ) or MAPK inhibitors (ϩ) were added 1 h prior to stimulation with 10% FCS for the times indicated. In A, cell extracts were analyzed by Western blotting using anti-MKP-2 and p38 antibodies (loading control) as described under "Experimental Procedures." Each blot is representative of three others. In B-D, gels were quantified using densitometry, and each value represents the mean Ϯ S.E. of at least four experiments. **, p Ͻ 0.01 when compared with vehicle control. ences, Oxford, UK. MEFs were trypsinized and washed with PBS and prepared at 1 ϫ 10 6 in Eppendorf tubes. Cells were fixed in ice-cold 70% ethanol (dropwise while vortexing to ensure proper fixation of cells and prevent clumping) at 4°C overnight. Cells were washed with PBS and centrifuged at 2000 ϫ g for 10 min, and then RNase A (50 g/ml) was added to ensure only DNA staining. Finally, cells were stained with propidium iodide at 50 g/ml.
Proliferation Assays-Confluent MEFs or CFs were detached with trypsin-EDTA, seeded on a coverslip into 24-well plates (5000 cells/well) in 10% FCS-DMEM, and allowed to attach for 24 h. Cells were starved in serum-free medium for 24 h and then stimulated for 24, 48, or 72 h with 10% FCS. Cultures were washed with PBS and stained with hematoxylin. The number of cells was determined by counting in 10 random fields per each coverslip.
Infection of MEFs with MKP-2 Adenovirus-MKP-2 adenovirus was generated as outlined previously (21) Cells were seeded on coverslips into 12-or 24-well plates and grown to ϳ50% confluency. The cell number was determined using a hemocytometer. MKP-2 adenovirus (100 -300 pfu/cell) was added to the cells and incubated for 24 h in normal growth medium, and then serum-starved for 24 h before stimulation for 24, 48, or 72 h with 10% FCS.
Flow Cytometry Assay of Apoptosis-Cells were infected for 24 h and then stimulated for a further 24 h prior to analysis. Cells were trypsinized and then pelleted at 1000 rpm for 2 min. The pellet was then resuspended in 500 l of 1ϫ annexin binding buffer (10 mM Hepes/NaOH, pH 7.4, 140 mM NaCl, and 2.5 mM CaCl 2 ). Phycoerythrin-annexin V and 7-aminoactinomycin D were added to the cells according to the manufacturer's instructions, and the samples were read in the FACScan flow cytometer using the FACS Diva software (FACScan, BD Biosciences). The data were analyzed using the FACSDiva (BD Biosciences) and RCS Express (De Novo Software) software. A total of 10,000 events were measured per sample. Gating was determined using phycoerythrin-annexin V FL-2 and 7-amino- Cells were incubated with 10 ng/ml PDGF for up to 120 min (A) or up to 24 h (B). Cell extracts were analyzed by Western blotting using phospho-ERK, -ERK1/2, and -MKP-2 antibodies as described under "Experimental Procedures." Each blot is representative of three others. In C and D, gels were quantified using densitometry, and each value represents the mean Ϯ S.E. of at least four experiments. *, p Ͻ 0.05 when compared with wild type MEFs. Fold stim., -fold stimulation.
actinomycin D FL-3 standards attached to beads (BD Biosciences), and preliminary experiments were conducted using paraformaldehyde and serum deprivation to define apoptotic and necrotic populations as outlined by the manufacturer's instructions.
Data Analysis-Each figure represents one of at least four separate experiments. Western blots were scanned on an Epson perfection 1640SU scanner using Adobe Photoshop 5.0.2 software. For gels, densitometry measurement was performed using the Scion Image program. Data were expressed as mean Ϯ S.E. Statistical analysis was performed by one-way analysis of variance with Dunnett's post test (*, p Ͻ 0.05, **, p Ͻ 0.01, ***, p Ͻ 0.001).

RESULTS
Mice deficient for DUSP-4 have been genetically characterized elsewhere (20). However, to confirm MKP-2 deletion, quiescent mouse embryonic fibroblasts from wild type or DUSP-4 deletion mice were stimulated with FCS over a period of 24 h (Fig. 1). In wild type MEFs, induction of MKP-2 was marked, reaching a peak between 2 and 4 h before returning toward basal values by 24 h (Fig. 1, A and B). In contrast, there was no increase in MKP-2 in stimulated MEFs from the DUSP-4 KO mice (Fig. 1, A and B). A number of other agents also increased MKP-2 expression including PDGF and phorbol 12-myristate 13-acetate in wild type but not in DUSP-4 deletion mice. 4 To ensure that MKP-2 protein expression was regulated in the normal manner in MKP-2 ϩ/ϩ fibroblasts, we preincubated cells with different MAP kinase inhibitors prior to stimulation (Fig. 2). Specifically, preincubation with the MEK1/2 inhibitor, UO126, substantially and significantly reduced MKP-2 expression (Fig. 2, A and B). However, the JNK and p38 inhibitors, SP600125 (Fig. 2, A and C) and SB203580 (Fig. 2, A and D), did not decrease MKP-2 induction. These data suggest that, by and large, an ERK-dependent mechanism of induction for MKP-2 exists in the fibroblast model.
We next sought to determine whether the kinetics of ERK activation could therefore be modified in the absence of MKP-2 (Fig. 3). PDGF stimulated a rapid and sustained increase in ERK activation, which was maximal between 5 and 10 min and continued for up to 24 h albeit at lower levels of phosphorylation. We found that over the early time period, ERK activation was not altered in MKP-2 Ϫ/Ϫ MEFs (Fig. 3, A and C). However, an extended time course revealed a small but significant potentiation between 2 and 24 h (Fig. 3, B and D).
Although there was a clear increase in ERK induction in the absence of MKP-2, changes in SAP kinase signaling were very much less marked. Anisomycin induced a strong and sustained phosphorylation of both JNK and p38 MAP kinase. MKP-2 Ϫ/Ϫ deletion gave a very small but significant increase in the magnitude of JNK phosphorylation but had no effect upon the duration of the time course (Fig. 4A). Similarly, p38 MAP kinase phosphorylation was slightly but significantly increased at a single time point (Fig. 4B).
Nevertheless, despite only very small differences in kinase signaling, there were substantial changes in growth parameters following MKP-2 knock-out. Fig. 5 shows the effect of DUSP-4 deletion on the proliferative capacity of MEFs. In the absence of MKP-2 protein, serum-stimulated proliferation assessed over 72 h was significantly slowed when compared with wild type (Fig. 5, A and B). This was also manifest over a shorter time frame (Fig. 5C) and is reflected in an extended lag phase prior to 4 A. Lawan, S. Currie, and R. Plevin, unpublished observations.  (Fig. 5D) gave ϳ95% infection. Under these conditions, the inhibition of proliferative responses in MKP-2 Ϫ/Ϫ MEFs was effectively reversed, and in fact, rates of proliferation increased above growth rates for wild type MEFs (Fig. 5, E and F).
To explain these differences in proliferation, we examined a number of cellular growth parameters linked to cell survival (Fig. 6). Using propidium iodide staining, we observed a  APRIL  decrease in the number of cells in G 1 phase and an increase in the number of cell that accumulated at G 2 /M phase (Fig. 6A). In cells blocked in G 1 using thymidine, we found an increased number in sub-G 1 , indicative of increased apoptotic cells (Fig.  6B). When we examined both cyclin B1 expression and the related cdc-2 phosphorylation, both these parameters were found to increase prematurely in MKP-2 Ϫ/Ϫ MEFs in comparison with MKP-2 ϩ/ϩ cells (Fig. 6C), indicative of retention in G 2 /M phase. Furthermore, when cells were incubated with topotecan to cause G 2 /M phase block (23) and then subse-quently washed, p-cdc-2 levels remained higher over 24 and 48 h, again indicative of greater accumulation in G 2 /M phase for MKP-2 Ϫ/Ϫ MEFs (Fig. 6D).

MKP-2 and Cellular Proliferation
To confirm that the effect of DUSP-4 gene deletion had a universal effect upon proliferation, we assessed proliferation in two other cell types including adult CFs and mouse bone marrow macrophages (Fig. 7). For CFs, the deficit in proliferation rates was similar to that observed for MEFs, and the cell number was over 50% less when compared with wild type cells (Fig.  7A). Cells in G 2 /M phase were also increased in MKP-2 Ϫ/Ϫ MEFs but to a lesser extent than MEFs (Fig. 7B). A similar difference was observed for continuously growing macrophages; however, when these cells where rendered quiescent prior to restimulation with FCS, as carried out for both MEFs and CFs, the cells failed to grow at all and subsequently died (Fig. 7C). For these cells, only a marginal stage-specific effect was observed (Fig. 7D).

DISCUSSION
The role of MKP-2 in cellular function has been hampered by the lack of a DUSP-4 deletion mouse model. Using a model such as that recently described (20), we make important new observations regarding the functional role of MKP-2 in cells. We identify a critical role for MKP-2 in the regulation of cellular proliferation and survival. We demonstrate a specific effect upon accrual of cells in G 2 /M phase and a role in attenuating stress-induced apoptosis, which can be reversed by inhibiting JNK activity.
In wild type MEFs, MKP-2 induction was observed 1-2 h following serum stimulation, a time frame in keeping with previous studies in rodent systems (13). In addition, induction was specifically regulated by ERK signaling, again in keeping previous studies implicating MKP-2 as a part of a negative feedback system for ERK signaling (24). Indeed, MKP-2 deletion revealed enhanced PDGF-mediated ERK phosphorylation, particularly over longer time frames, correlating with the time course of MKP-2 induction. This suggests that ERK regulation is a bona fide component of MKP-2 function. However, in other cell types such as macrophages, we have observed no increase in ERK signaling following MKP-2 deletion (20), suggesting a cell type-specific difference in the function of MKP-2. We also observed only minor increases in JNK signaling in response to anisomycin, finding consistent with our own studies in macrophages (20) but at variance with other studies using siRNA rundown (25). These results are not easily explained as in wild type MEFs, anisomycin did not cause any increase in MKP-2 expression, suggesting no potential for regulation of the corresponding kinase pathways. Deletion of MKP-2 also did not effect p38 MAP kinase activation, which was not unexpected as MKP-2 does not dephosphorylate this protein in vitro. This strongly suggests that the function of MKP-2 in terms of substrate specificity in vivo may be cell type-specific and agonist-dependent rather than being defined by dephosphorylation studies performed in vitro.
The small differences in MAP kinase signaling following DUSP-4 deletion were nevertheless relevant to cellular proliferation. Deletion of MKP-2 resulted in a significant decrease in proliferation rate, an effect that was reversed by Adv-MKP-2. This is similar to a recent study using MKP-1 KO fibroblasts (26) but not earlier studies, which demonstrate no difference in cell growth parameters. Furthermore, MKP-2 deletion has been shown to be protective in pro B cells (27), suggesting a cell type-dependent role in the regulation of cell survival. We pinpoint this defect in proliferation to an effect upon G 2 /M phase transition, an effect not observed in MKP-1 deletion mice. A similar result has been obtained following MKP-2 knockdown in mice tumor cells (28), and this is associated with a complete loss in cyclin B expression and cdc2 phosphorylation. To our surprise, however, we did not find any reduction in these parameters; in fact, a small increase in cyclin B1 expression and cdc-2 phosphorylation was observed, which was also apparent when cells were arrested in G 2 /M phase with topotecan and then allowed to recover. This suggests a propensity for cells to accumulate within G 2 /M phase in MKP-2 Ϫ/Ϫ MEFs. Usually, under conditions of cell cycle arrest, cyclin B is reduced due to p53-mediated transcriptional repression of the cyclin B1 promoter (29). However, it has been shown that following chemical treatment or ionizing radiation, cyclin B1 can accumulate, particularly if p53 is non-functional (30), and this possibility is under investigation in our laboratory. Recently, MKP-2 has been identified as a target gene for p53 (31), and MKP-1 has also been shown to interact with p53, to regulate G 1 /S phase progression (32). However, no study to date has demonstrated an effect of any DUSP within G 2 /M phase of the cell cycle. This further highlights the different functional role of MKP-2 versus the prototypic MKP-1 in cell cycle regulation.
Analysis of apoptosis demonstrated an essential role for MKP-2 in protecting cellular integrity. Both anisomycinstimulated and UVC-stimulated apoptosis were significantly enhanced in MKP-2 Ϫ/Ϫ cells and, for anisomycin stimulation, this was associated with increased caspase-3 cleavage and ␥H2AX phosphorylation. Caspase-3 is a well recognized apoptotic mediator protein, whereas H2AX, a histone protein (33), mediates apoptotic DNA fragmentation and cooperates with the caspase-3/CAD pathway to mediate apoptosis (34). This aspect of MKP-2 function is essentially similar to that observed for MKP-1 where deletion resulted in increased apoptosis in response to hydrogen peroxide or anisomycin (26,31). In other studies, MKP-2 is implicated in apoptosis, suggesting potentially species-or cell type-specific differences (35).
A key issue, however, was in the identification of the kinase mediating the effect of MKP-2 deletion. To do this, we utilized adenoviral MKP-2 in gain of function studies. We have previously demonstrated the selective inhibitory effect of MKP-2 overexpression on stress-induced JNK signaling using either stable or inducible MKP-2-expressing cell lines (17,18) or adenoviral infection in primary endothelial cell cultures (21). Our present study confirms this JNK sensitivity for anisomycin signaling (Fig. 8). Although caspase-3 activation is well recognized to be regulated by JNK, only recently have studies also identified H2AX as a target (34,37,38). Overexpression of Ad-MKP-2 reversed enhanced anisomycin-mediated apoptosis in MKP-2 Ϫ/Ϫ MEFs and also the increased caspase-3 degradation and ␥H2AX phosphorylation. This was also associated with a specific dephosphorylation of JNK. Thus, despite little apparent change in JNK signaling in MKP-2 Ϫ/Ϫ MEFs per se, these findings link MKP-2 deletion to enhanced JNK-dependent signaling, DNA damage, and resultant apoptosis. Preliminary studies using the JNK inhibitor SP600125 confirm this interpretation. 4 Interestingly, a previous study demonstrated that pharmacological inhibition of p38 and not JNK was able to rescue MKP-1-deficient fibroblasts from enhanced cell death (26). Therefore, although the effects of MKP-2 and 1 deletion are similar in terms of apoptosis, the kinase mediating the effect of MKP deletion is clearly different.
Our findings also indirectly give an insight into the role of nuclear JNK in regulating apoptosis. Given that MKP-2 is a nuclear located enzyme and unlike MKP-1, which has been found in the mitochondria (39), there is little evidence for discrete pools within other compartments. This suggests that JNK-mediated cell death is mediated in part by nuclear located JNK. Indeed JNK phosphorylation was found to be more sensitive to Adv-MKP-2 in the MKP-2 Ϫ/Ϫ MEFs, suggesting that JNK activation occurs within this compartment. Thus, it is surprising that in addition to increases in the phosphorylation of the nuclear substrate ␥H2AX, the cleavage of caspase-3, a cytosolic protein, is also enhanced. Cleaved caspase-3, although identified in the nucleus (40), requires processing in the cytosol, and although it is accepted that JNK has both nuclear and cytosolic substrates (41), our studies suggest that JNK could be translocated to the cytosolic compartments, such as the mitochondria following activation within the nucleus. Alternatively, long term modulation of JNK signaling could have profound effects on initiation of apoptosis, presumably by changing expression of pro-and anti-apoptotic genes that regulate initiator caspases such as c-FLIP (42,43), XIAP, and A20 (44).
In conclusion, we demonstrate a critical non-redundant role of MKP-2 in cell survival. This might suggest that overexpression of MKP-2 may be a factor in the development of cancer. Indeed in other cellular studies, we have consistently identified overexpression of MKP-2 as mediating resistance to apoptosis (21). Furthermore, overexpression of DUSP-4 is observed in cancer cell lines (21), and high levels are associated with oesophagogastric rib metastasis and pancreatic and liver tumors (28). However, in other cancers, DUSP-4 deletion is associated with tumor formation in lung (45) and breast, and overexpression in breast cancer cells substantially reduces tumor formation in mice (35). MKP-2 may also function to prevent ovarian metastatic spread (36). Thus, it is possible that DUSP-4 plays opposing roles in different types of cancer depending on which kinase, ERK or JNK, is being preferentially regulated in that cell type.