Negative and Positive Regulation of MAPK Phosphatase 3 Controls Platelet-derived Growth Factor-induced Erk Activation*

MAPK phosphatases (MKPs) are dual specificity phosphatases that dephosphorylate and thereby inactivate MAPKs. In the present study, we provide evidence that platelet-derived growth factor BB (PDGF-BB) regulates MKP3 (DUSP6), which is considered to be a phosphatase highly selective for Erk. Intriguingly, we observed that Mek is positively regulated by MKP3, whereas Erk itself is negatively regulated. In addition, we found that activation of PDGF receptor α or β leads to a rapid proteasomal degradation of MKP3 in a manner that requires Mek activation; this feed-forward mechanism was found to be essential for efficient Erk phosphorylation. We could also demonstrate that PDGF-BB stimulation induces phosphorylation of MKP3 at Ser-174 and Ser-300; phosphorylation of Ser-174 is involved in PDGF-induced MKP3 degradation, since mutation of this site stabilized MKP3. Moreover, activated Erk induces mkp3 expression, leading to restoration of MKP3 levels after 1-2 h and a concomitant dephosphorylation of Erk in cells with activated PDGFRα. Reducing the MKP3 level by small interfering RNA leads to an increased Erk activation and mitogenic response to PDGF-BB. In conclusion, MKP3 is an important regulator of PDGF-induced Erk phosphorylation acting in both a rapid positive feed-forward and a later negative feed-back loop.

and -D chains bind to PDGFR␤. Ligand binding induces receptor homo-or heterodimerization and autophosphorylation. The phosphorylated tyrosine residues constitute recruitment sites for Src homology 2 domain-containing proteins, including the Grb2-Sos complex that activates Ras and the Erk MAPK pathway, the tyrosine kinase Src, the tyrosine phosphatase Shp2, phosphatidylinositol 3-kinase, and phospholipase C␥; in addition, the PDGFR␤ but not PDGFR␣ binds RasGAP, which inactivates Ras (1,3). These pathways mediate cell growth, survival, and migration.
The MAPK family is evolutionarily conserved, and mammalian cells possess several MAPK pathways (i.e. Erk1 and -2; p38␣, -␤, -␦, and -␥; and JNK1, -2, and -3) (4,5). The Erk pathway has been connected to cell proliferation and differentiation (6). In contrast, the p38 and JNK pathways have more established roles in apoptotic signaling but may under certain circumstances also contribute to cell proliferation and migration (7). Receptor tyrosine kinases often activate the Erk pathway by recruiting a complex between the adaptor protein Grb2 and the guanine exchange factor Sos to the plasma membrane, where it activates the small G-protein Ras, which then activates the three-tiered kinase module consisting of Raf, Mek, and Erk (6). A large portion of human tumors have deregulated Erk kinase activity, underscoring its importance in tumor formation (8). Unrestrained Erk kinase activity can arise from inappropriate activation of an upstream signaling component (e.g. oncogenic Ras) or loss of an inhibitory regulator, such as MAPK phosphatases.
The biological consequence of Erk activation is connected to the magnitude as well as temporal pattern of Erk phosphorylation. For example, in PC12 cells, sustained Erk activation induced by nerve growth factor leads to differentiation, whereas transient EGF-induced Erk phosphorylation results in proliferation (9). Other studies have demonstrated that sustained Erk activation is necessary for cell cycle progression (10 -12). Thus, mechanisms that regulate the kinetics of Erk activation have a major biological impact, although there is significant cell type specificity regarding the outcome.
MAPK phosphatases (MKPs) are dual specificity phosphatases that negatively regulate the activity of MAPKs by dephosphorylating the essential threonine and tyrosine residues in the activation loop (13,14). Treatment with growth factors, such as NGF, induces MKP expression (15)(16)(17), which can modulate both the intensity and duration of MAPK signaling. The different MKPs have distinct substrate specificities (i.e. selectively dephosphorylating Erk, p38, or JNK MAPKs), enabling the cell to specifically control different MAPK pathways (14). In the present study, we have explored the role of MKPs in PDGFinduced Erk activation.

EXPERIMENTAL PROCEDURES
Reagents-Recombinant human PDGF-BB was generously provided by Amgen (Thousand Oaks, CA). The inhibitor U0126 was from Calbiochem, MG132 and cycloheximide from Sigma and STI571 from Novartis Pharma AG (Basel, Switzerland). Monoclonal ␤-actin antibody was from Sigma (A5441), and antibodies against phosphorylated-Mek1/2 (number 9154), total Mek1/2 (number 9126), and phosphorylated Erk1/2 (number 9106) were purchased from Cell Signaling Technology (Beverly, MA). Anti-phosphotyrosine (PY99, sc-7029) and anti-ubiquitin (P4D1, sc-8017) antisera were from Santa Cruz Biotechnology, Inc. (Santa Cruz, CA). Antiserum against MKP3 was raised by immunizing a rabbit with the synthetic peptide CPSNQNVYQVDSLQST conjugated to keyhole limpet hemocyanin. The antibodies were purified by chromatography over a column of the corresponding peptide immobilized on Sepharose beads. Purified antibodies were stored in 0.15 M NaCl, 20 mM Hepes, pH 7.4, 50% glycerol at Ϫ70°C. A rabbit antiserum recognizing Erk was raised against a peptide corresponding to the carboxyl-terminal sequence EETARF-QPGYRS conjugated to keyhole limpet hemocyanin. A rabbit polyclonal antiserum against PDGFR␤ was raised against a glutathione S-transferase fusion protein containing the COOHterminal amino acid residues of PDGFR␤. A rabbit antiserum recognizing PDGFR␣ was generated using a synthetic peptide corresponding to amino acids 1066 -1084 in the human PDGFR␣. The MKP3 expression construct was a kind gift of Dr.
Stephen Keyse (University of Dundee), and the FLAG-ubiquitin expression construct was a kind gift from Dr Maréne Landström (Uppsala University). [ 32 P]orthophosphate (PBS43) was purchased from Amersham Biosciences.
Cell Culture-Porcine aortic endothelial (PAE) and 293T cells were cultured in Ham's F-12 medium with L-glutamine or Dulbecco's modified Eagle's medium, respectively, supplemented with 10% fetal bovine serum and 100 units/ml penicillin, 100 g/ml streptomycin. For serum starvation, cells were washed once and incubated in Ham's F-12 or Dulbecco's modified Eagle's medium containing 1% or 0.1% fetal bovine serum, respectively.
Transient Transfections-Transfections were performed with Lipofectamine 2000 according to the protocol supplied by the manufacturer (Invitrogen). Cells were used for experiments 48 h after transfection.
Immunoblotting-Subconfluent cells were starved and incubated with inhibitors or vehicle at the indicated concentrations and thereafter stimulated with 100 ng/ml PDGF-BB for the indicated periods of time. Cells were washed with ice-cold phosphate-buffered saline and lysed (1% Triton X-100, 150 mM NaCl, 10% glycerol, 5 mM EDTA, 20 mM Tris, pH 7.4, 1 mM Pefa Bloc, 1% Trasylol, 10 mM NaF, 1 mM sodium orthovanadate). Extracts were clarified by centrifugation, and protein concentration was determined by the BCA protein assay system (Pierce). Equal amounts of lysates were boiled with SDS sample buffer containing dithiothreitol and separated by SDS-PAGE. For Western blotting, samples were electrotransferred to polyvinylidene difluoride membranes (Immobilon P), which were blocked in 5% dry milk in Tris-buffered saline solution containing 0.1% Tween 20. Commercial primary antibodies were used at concentrations recommended by the suppliers, and antibodies produced in house were used at 2 g/ml and incubated overnight in the cold. After washing, the membranes were incubated with horseradish peroxidase-conjugated anti-mouse or anti-rabbit IgG antibodies (both from Amersham Biosciences), and proteins were visualized using ECL Western blotting detection systems from Roche Applied Science on a cooled chargecoupled device (CCD) camera (Fuji, Minami-Ashigata, Japan). For reprobing, membranes were stripped with 0.4 M NaOH for 10 min at room temperature, blocked, and incubated with the corresponding antibodies. Where quantifications of band intensities are shown, the number corresponds to the average value Ϯ S.E.
[ 32 P]Orthophosphate Labeling-Cells were starved for 16 h, preincubated for 3 h with 4 mCi/ml [ 32 P]orthophosphate in phosphate-free Ham's F-12 supplemented with 0.3% dialyzed fetal bovine serum, and then stimulated with PDGF-BB for 15 min. Cell lysates were subjected to immunoprecipitation with an antibody recognizing MKP3.
Real Time PCR-Total DNA-free cellular RNA was extracted from cells treated with PDGF for different periods of time with the RNeasy kit (Qiagen) and reverse-transcribed (SuperScript II RNase; Invitrogen) to create cDNA templates. The PCR was performed by the qPCR TM core kit for SYBR TM Green I (Bio-Rad) according to the manufacturer's instructions. Glyceraldehyde-3-phosphate dehydrogenase was used as an endogenous control for the relative quantification of the target message. Specific primers were as follows: for mkp3, GTTTTTCCCT-GAGGCCATTTC (forward) and TCACAGTGACTGAGCG-GCTAAT (reverse); for glyceraldehyde-3-phosphate dehydrogenase, CCCTTCATTGACCTCCACTACAT (forward) and GGGATTTCCATTGATGACAAG (reverse).
siRNA Knockdown-Down-regulation of MKP3 was performed by using specific siRNA purchased from Dharmacon Research. For every experiment performed, nontargeting siRNA was used as a control (target sequence 5Ј-CGT-ACGCGGAATACTTCGA-3Ј). Transfection of siRNA was done for 24 h with SilentFect from Bio-Rad. Levels of knockdown were tested after 40 h by measuring protein levels by immunoblotting.
DNA Synthesis Assay-The cells plated into 24-well plates were serum-starved and then incubated for 24 h with PDGF-BB in Ham's F-12 containing [ 3 H]thymidine (0.1 Ci/ml). Thymidine incorporation into acid-insoluble material was measured by a scintillation counter.

PDGFR␣ and PDGFR␤ Activate Erk with Different Kinetics-
PAE cells transfected with PDGFR␣ or PDGFR␤ (denoted PAE/PDGFR␣ and PAE/PDGFR␤, respectively) were used to investigate Erk activation downstream of the two PDGFR isoforms expressed at a comparable level in the same cellular background. Cells were stimulated with PDGF-BB, which binds with similar affinity to both PDGFR␣ and PDGFR␤, and the phosphorylation of Erk was analyzed at different points by immunoblotting using phospho-specific antibodies. We found that the kinetics of PDGF-BB-induced Erk phosphorylation was different, depending on which PDGFR isoform was expressed. In PAE/PDGFR␣ cells, a biphasic activation was observed with an initial peak of Erk phosphorylation after 15 min, followed by a sharp decrease and a later second increase after 4 h of stimulation (Fig. 1A, upper panels). In contrast, activation of PDGFR␤ resulted in Erk phosphorylation that peaked after ϳ30 min of stimulation and then remained relatively constant at a lower level up to 6 h of stimulation (Fig. 1A, lower panels). Next, we analyzed the phosphorylation status of the upstream kinase Mek. As can be seen in Fig. 1B, Mek was transiently activated downstream of PDGFR␣ and more sustained downstream of PDGFR␤, consistent with the initial kinetics of Erk activation. For reference, the kinetics of PDGFR␣ and -␤ phosphorylation was also included in Fig. 1C. PDGF-BB Regulates the Expression Level of MAPK Phosphatase 3-In order to investigate whether the initial rapid increase of Erk phosphorylation was accompanied by a decrease in MKP expression, we analyzed by Western blotting the protein levels of the different MKPs known to be able to dephosphorylate Erk. Although the levels of MKP1, MKP2, hVH3, and MKP4 were not influenced by PDGF-BB stimulation (supplemental Fig. 1, A and B, for PDGFR␣ and PDGFR␤, respectively), MKP3 underwent rapid degradation in response to PDGF-BB stimulation in both PDGFR␣and PDGFR␤-expressing cells; after 1-2 h of PDGFR␣ stimulation, MKP3 levels returned to initial levels, followed by a second phase of degradation ( Fig.  2A, upper panel). In contrast, activation of PDGFR␤ induced a slower degradation of MKP3 and a minor wave of MKP3 protein expression, followed by a second phase of MKP3 decrease ( Fig. 2A, lower panel). The biphasic activation of Erk downstream of PDGFR␣ was inversely correlated with biphasic kinetics of MKP3 protein expression (compare Figs. 1A and 2A). This suggests that PDGF-BB-induced changes in the intensity of Erk activity are related to the level of MKP3 protein. To further characterize the MKP3 degradation downstream of PDGFR␣, we measured the protein stability in the presence of the protein synthesis inhibitor cycloheximide in the presence or absence of PDGF-BB; the half-life of MKP3 was estimated for both the first and second waves of degradation (Fig. 2B). We found the half-life for the first wave of degradation to be ϳ20 min, compared with 25 min for the second wave. The high level of MKP3 expression after 1-2 h of PDGF-BB stimulation of PDGFR␣ may explain why the Erk phosphorylation is low despite significant Mek phosphorylation at these time points (compare Figs. 1, A and B, and 2A). Stimulation of PDGFR␤ induced a more sustained Erk phosphorylation, which is compatible with the observation that after the initial MKP3 degradation, the protein level remained low with only a minor transient increase after 2 h of PDGF-BB stimulation ( Fig.  2A, lower panel).
PDGF-BB-induced Erk Activation Requires Proteasomal Degradation of MKP3-To investigate whether the rapid degradation of MKP3 induced by PDGF-BB occurred through the proteasome, we analyzed the ability of PDGF-BB to induce ubiquitination of MKP3 and the effect of treatment with the proteasomal inhibitor MG132 on MKP3 stability. Fig. 2C shows that that MKP3 indeed becomes ubiquitinated in response to PDGF-BB stimulation. Furthermore, we found that MG132 treatment completely inhibited the degradation of MKP3 both in PDGFR␣and PDGFR␤-expressing cells (Fig. 2D). Furthermore, in the presence of proteasomal inhibitor, we observed that PDGF-BB was no longer able to induce a robust Erk activation (Fig. 2D).  FEBRUARY 13, 2009 • VOLUME 284 • NUMBER 7

JOURNAL OF BIOLOGICAL CHEMISTRY 4629
The blunted Erk phosphorylation may be attributed to stabilization of MKP3 and thus an increased dephosphorylation of Erk. It is possible that stabilization of other phosphatases also is involved. However, we were unable to detect any effects of MG132 on the levels of MKP1, MKP2, and hVH3 ( Fig. 2D and supplemental Fig. 1, C and D), arguing for a relatively specific role of MKP3 in this process.
In order to determine whether activation of the Erk pathway is important for the degradation of MKP3, we treated cells with the low molecular weight inhibitor U0126, which inhibits Mek and consequently Erk activation, and investigated the effect on PDGF-BB-induced MKP3 degradation. Treatment with U0126 resulted in a significant reduction in the rapid degradation of MKP3 after PDGF-BB stimulation, both in PDGFR␣and PDGFR␤-expressing cells (Fig. 2E; compare bands after 30 min of PDGF-BB stimulation in the absence or presence of U0126). Furthermore, we observed that the increased levels of MKP3 normally seen after 2 h of PDGF-BB stimulation did not occur in the presence of U0126, and in fact we found a decrease in MKP3 level (Fig. 2E). In Fig. 4B, we show that the mkp3 gene is activated by PDGF-BB in a manner requiring Erk activity. Thus, the decrease in MKP3 level after 2 h of PDGF-BB stimulation in the presence of U0126 may be caused by general protein turnover in combination with inhibited expression of the mkp3 gene. Another possibility is that U0126 has been partially metabolized, thereby allowing for weak PDGF-BB-induced MKP3 degradation in combination with partially inhibited mkp3 gene expression. Normally, MKP3 appears as multiple bands in immunoblotting. However, after U0126 treatment, the slowest migrating band disappeared, suggesting that it can represent a phosphorylated species of MKP3 (Fig. 2E). In support of this notion, PDGF-BB induced phosphorylation of MKP3 in a Mekdependent manner, as determined by metabolic [ 32 P]orthophosphate labeling of cells treated with PDGF-BB in the absence or presence of U0126 as indicated (Fig. 3A).
To identify the phosphorylation sites in MKP3 following PDGF-BB stimulation, we performed two-dimensional phosphopeptide mapping of the tryptic MKP3 digest (Fig.  3B). Two phosphopeptides (denoted peptide 1 and peptide 2) were recovered and subjected to Edman degradation and phosphoamino acid analysis (Fig. 3C). Matching the cycle in which radioactivity was released during Edman degradation and the result from the phosphoamino acid analysis with the sequences of all tryptic fragments of MKP3 revealed Ser-174 and Ser-300 as novel PDGF-induced phosphorylation sites in MKP3. In the case of Ser-174 (peptide 1), only one MKP3-derived tryptic peptide was consistent with the cycle where radioactivity was released, whereas both Ser-300 and Ser-328 could match the pattern obtained for peptide 2. However, through mutational analysis we could exclude Ser-328 (data not shown). To elucidate whether the PDGF-induced phosphorylation was impor- P]orthophosphate, treated with U0126 as indicated, and stimulated with PDGF-BB followed by lysis and MKP3 immunoprecipitation, SDS-PAGE, transfer to nitrocellulose filter, and visualization by autoradiography. The band corresponding to MKP3 was excised and digested in situ with trypsin. B, the resulting digests were separated on TLC plates at pH 1.9, followed by ascending chromatography. Radioactive peptides were detected by autoradiography. C, phosphopeptides 1 and 2 were eluted and subjected to Edman degradation using a gas phase sequencer as well as to phosphoamino acid analysis. The release of radioactivity in each cycle is indicated. The amino acid sequences of tryptic peptides derived from MKP3 containing a serine residue at position where radioactivity was released are presented. D, to analyze the effect of mutating Ser-159, Ser-174, Ser-197, and Ser-300 to alanine residues on MKP3 stability, 293T cells were transfected with these mutant forms of MKP3 together with PDGFR␣. After 30 min of PDGF-BB stimulation, total cell lysates were prepared, separated by SDS-PAGE, and subjected to MKP3 immunoblotting. All mutations were expressed in combination with C293A rendering MKP3 phosphatase inactive, thereby allowing overexpressed protein to be phosphorylated. E, the effect on the mutation of Ser-174 to alanine was investigated during extended PDGF-BB stimulation in the presence of 100 g/ml cycloheximide. Quantifications shown are the averages of three independent experiments Ϯ S.E. tant for MKP3 stability, we mutated Ser-174 and Ser-300 to Ala (denoted S174A and S300A) and investigated the impact on MKP3 stability. In addition, also the previously identified serum-induced MKP3 phosphorylation sites Ser-159 and  were mutated to Ala, denoted S159A and S197A; these sites may represent serum-specific phosphorylation sites in MKP3, since these were not identified as phosphorylated in our model system in response to PDGF-BB stimulation. To be able to express these MKP3 mutants in transient transfection experiments using 293T cells without abolishing Erk phosphorylation, we also had to mutate the active site Cys-293 to Ala (C293A). As can be seen in Fig. 3D, mutation of Ser-174 caused a marked stabilization of MKP3, suggesting an important role of this phosphorylation site in the regulation of MKP3 protein stability. Consistent with an earlier report (19), we were able to show that mutation of Ser-159 also stabilizes MKP3. Mutation of Ser-197 did not have any effect on MKP3 stability. When we mutated Ser-300, we were unable to obtain reproducible data on MKP3 stability; therefore, this mutant was not further studied in this regard (data not shown). To further investigate the effect on MKP3 stability, we stimulated cells transfected with wild-type MKP3 or with S174A mutant MKP3 with PDGF-BB in the presence of cycloheximide; as shown in Fig. 3E, mutation of Ser-174 substantially stabilized MKP3 compared with control. Ser-174 is located in a region of MKP3 that may influence its localization; however, using subcellular fractionation, we were not able to observe a change in localization for the S174A mutant compared with wild-type MKP3 (supplemental Fig.  2A). Another concern is whether mutation of Ser-174 or Ser-300 affects the catalytic activity of MKP3. To this end, we expressed these mutants in 293T cells and analyzed their impact on Erk dephosphorylation. As can be seen in supplemental Fig. 2B, both S174A and S300A mutant MKP3 effi-ciently abolished Erk phosphorylation. In conclusion, PDGF-BBinduced Mek activation leads to increased phosphorylation of MKP3, which correlates with a rapid degradation of MKP3.
PDGF-BB Induces mkp3 mRNA Synthesis in an Erk-dependent Manner-In order to investigate the mechanism(s) behind the induction of MKP3 1-2 h after PDGF-BB stimulation, we performed quantitative real time PCR analysis of mkp3 mRNA expression. PDGF-BB stimulation was found to cause about a 3-fold increase of mkp3 mRNA in PDGFR␣-expressing cells and about a 2-fold induction in cells expressing PDGFR␤; in both cases, mRNA induction occurred after 1 h of PDGF stimulation (Fig. 4A).
In order to explore whether activation of Erk is important for induction of the mkp3 gene, we analyzed the effect of treatment of cells with U0126. After pretreatment with U0126, we could not observe the increase in MKP3 protein levels usually detected after 2 h of PDGF-BB stimulation of PDGFR␣ and to a lesser extent PDGFR␤ (Fig. 2E). Instead, we found a gradual decrease in MKP3 protein levels, probably due to general protein turnover. In addition, siRNA-mediated silencing of Erk also abolished the increase in MKP3 protein levels after 2 h of PDGF-BB treatment (data not shown). This prompted us to investigate whether Erk pathway activation was necessary for the PDGF-BB-induced transcriptional activation of the mkp3 gene. Therefore, we stimulated cells with PDGF-BB for various periods of time in the absence or presence of U0126 and then prepared total RNA. Using quantitative real time PCR we could indeed demonstrate a requirement for Erk activation in the induction of mkp3 gene expression (Fig. 4B). However, we observed an escape from the U0126-mediated inhibition in mkp3 expression after 2 h of PDGF-BB treatment. Possible explanations for this include metabolism of the U0126 compound after prolonged incubation or activation of the mkp3 gene through an alternative mechanism that does not require the Erk pathway. Thus, the Erk pathway is involved both in a positive feed-forward mechanism to cause degradation of MKP3 and in a negative feed-back loop by inducing the expression of MKP3 at later time points.

MKP3 Promotes Persistent Mek Activation but Negatively Regulates PDGF-BB-induced Erk Activation and Proliferation-
To further elucidate the functional role of MKP3, we downregulated its expression using siRNA and analyzed the effects on PDGF-induced Mek and Erk phosphorylation as well as the mitogenic response. We found that down-regulation of MKP3 by ϳ70% enhanced the intensity of Erk phosphorylation in response to PDGF-BB, consistent with the role of MKP3 as a negative Erk regulator, but had no major effect on the kinetics of Erk phosphorylation (Fig. 5A). We observed that siRNAmediated down-regulation of MKP3 reduced Mek phosphorylation in response to PDGF-BB stimulation of the PDGFR␣ (Fig.  5B, upper panels). In concurrence, MKP3 overexpression led to enhanced Mek but reduced Erk activation (Fig. 5B, middle panels). In addition, stabilization of MKP3 through proteasomal inhibition led to a prolonged Mek activation (Fig. 5B, lower  panels). This may reflect loss of a negative feed-back control of Mek by Erk (20). Hence, MKP3 had opposing effects on Mek and Erk phosphorylation, but the dominant function was to act as a negative regulator of PDGF-mediated Erk activation.
In cells with enhanced Erk activation after silencing of MKP3, we observed a significant increase in the mitogenic response to PDGF-BB in both PAE/PDGFR␣ and PAE/ PDGFR␤ cells, in accordance with a critical role of Erk in mediating proliferative signals (Fig. 6).

DISCUSSION
In the present study, we have investigated the mechanisms by which PDGF-BB activates the Erk MAPK pathway. We provide  evidence that MKP3, a cytoplasmic Erk-selective phosphatase (21), has an important role in the regulation of Erk phosphorylation in response to PDGF-BB stimulation. In an early phase after PDGF-BB stimulation, MKP3 is ubiquitinated and targeted for proteasomal degradation in a Mek-dependent manner, requiring phosphorylation of Ser-174 or Ser-159 in MKP3. The MKP3 degradation is essential for activation of Erk. At a later phase, mkp3 is induced and MKP3 is synthesized, leading to an efficient dephosphorylation and deactivation of Erk. Interestingly, both the degradation and subsequent synthesis of MKP3 are dependent on activation of the Erk pathway, indicating that they represent positive feed-forward and negative feedback mechanisms, respectively (Fig. 7). In fact, it was reported by Lin and Yang (22) that Erk activation induced proteasomal degradation of MKP1, thus establishing a similar positive feedforward loop. Importantly, down-regulation of MKP3 expression leads to enhanced Erk phosphorylation and proliferation after treatment with PDGF-BB, illustrating the importance of this pathway for an appropriate response to PDGF-BB.
Stimulation of cells expressing PDGFR␣ or PDGFR␤ led to a transient increase in gene expression of the dual specificity phosphatases mkp1, mkp2, mkp3, and hvh3, but this did not translate into changes in protein levels for MKP1, MKP2, and hVH3. This finding may be explained by the presence of a translational block that possible could be released by a signal not present in our cellular model. In contrast, the increase of mkp3 mRNA induced by PDGFR␣ translated into efficient MKP3 protein synthesis, whereas PDGFR␤-mediated mkp3 gene expression resulted in only a minor increase in MKP3 protein levels. This control of protein translation, which differs downstream of PDGFR␣ and PDGFR␤, provides an additional level of regulation. The phosphatases MKP1, MKP2, MKP4, MKP6, and hVH3 are able to dephosphorylate Erk (14,23,24). Since the protein levels of those MKPs are not affected by PDGF-BB treatment, they may provide PDGF-independent phosphatase activity that inactivates Erk in the absence of a continued activating signal.
In addition, a possibility that remains to be elucidated is that the catalytic activity of the MKPs may be regulated by PDGF stimulation although their expression level remains constant.
We observed that PDGF-BB induced a slower migration of MKP3 in SDS-PAGE, and we were able to show that PDGF-BB induced phosphorylation of MKP3 in a Mek-dependent manner. Moreover, this modification was required for degradation of MKP3, which we speculate may occur through a phosphorylation-dependent recruitment of a ubiquitin ligase. In accordance with this hypothesis, it has been reported that Erk-mediated phosphorylation of MKP1 (Ser-296 and Ser-323) leads to recruitment of the ubiquitinligase SCF Skp2 (22). However, manipulation of SCF Skp2 levels did not influence the PDGF-BB-induced MKP3 degradation in our model system (data not shown). In contrast, phosphorylation of MKP1 (Ser-359 and Ser-364) by Erk has been associated with increased MKP1 stability (25). Thus, regulation of MKP1 protein stability is complex, and it appears that different phosphorylation sites may differentially control MKP1 degradation. It has been demonstrated that serum stimulation of fibroblasts leads to MKP3 degradation through Erk-and mTor-mediated phosphorylation of Ser-159 and Ser-197 (19,26). In our model system, we were able to identify the novel phosphorylation site Ser-174 as important in regulating PDGF-induced MKP3 protein degradation. When we expressed MKP3 with Ser-174 mutated to Ala (in combination with mutation of the active site Cys-293 to Ala in order to avoid complete Erk dephosphorylation) together with PDGFR␣ in 293T cells, we found that the mutant MKP3 form was stabilized compared with MKP3 only containing the C293A mutation. In fact, also in the absence of PDGF-BB stimulation, we observed a stabilization of the Ser-174 (and Ser-159) mutant forms of MKP3, implying that there is a constitutive rate of MKP3 degradation, which may be accelerated by PDGF-BB treatment. Consistently, also in the absence of PDGF-BB stimulation we observed MKP3 migrating as multiple bands suggesting basal phosphorylation when expressed in 293T cells. In addition, endogenous MKP3 was stabilized by MG132 both in the absence and presence of PDGF-BB stimulation, suggesting a constitutive proteasomal turnover. Although we were not able to detect PDGF-BB-induced phosphorylation of Ser-159 and Ser-197, which are phosphorylated after serum stimulation (19), we could confirm a role of Ser-159 in MKP3 stability by site-directed mutagenesis. This observation may be explained by phosphorylation of Ser-159 at low stoichiometry induced by the low amount of serum present in our starvation medium. An alternative explanation could be that mutation of Ser-159 could influence interaction(s) with protein(s) necessary for MKP3 degradation. We were not able to detect an effect of mutating Ser-197 to Ala on MKP3 degradation. A possible explanation for the lack of effect of the S197A mutation on MKP3 stability in our experiments, although such an effect has been previously reported (19,26), may be that Ser-197 interacts in a phosphorylation-dependent manner with a scaffolding protein not expressed in our model system.  FEBRUARY 13, 2009 • VOLUME 284 • NUMBER 7

MKP3 Regulates PDGF-induced Erk Activation
In addition, Erk activation is necessary for PDGF-mediated mkp3 gene expression, which ultimately halts Erk signaling after prolonged ligand stimulation. In concurrence, it was recently demonstrated that Erk via Ets factors is required for mkp3 gene expression in response to FGF (27). Moreover, also serum-induced MKP1 and MKP2 protein expression is sensitive to inhibition of the Erk pathway (28).
Our data suggest that MKP3 acts as a positive regulator of Mek. The reduced Mek phosphorylation observed after MKP3 down-regulation and, conversely, enhanced phosphorylation when MKP3 was overexpressed can be due to changes in Erk activity; Erk may, through direct phosphorylation of Mek, Raf, or Sos, negatively regulate Mek activation (20). Thus, by changing the level of MKP3, the ability of Erk to establish negative feed-back loops could be affected.
We observed that the difference in MKP3 protein expression downstream of PDGFR␣ and PDGFR␤ was correlated with a difference in temporal pattern of Erk phosphorylation. Since the kinetics of Erk phosphorylation is of importance for the end effect on cell growth versus differentiation (9), the difference in kinetics could contribute to the subtle differences in signaling via the two PDGF receptors. siRNAmediated down-regulation of MKP3 resulted in a marked increase in Erk phosphorylation in response to PDGF-BB, despite less persistent Mek activation. Consistent with reports suggesting that a sustained and robust Erk activation is necessary to induce cell proliferation (10 -12), we observed an increased mitogenic response to PDGF-BB when MKP3 was down-regulated. Our data indicate that MKP3 serves to dampen the PDGF-induced Erk activation and proliferative signaling. In summary, Erk exploits MKP3 to regulate its own activity in two signal transduction phases: first in the initial amplification of PDGF-BB-induced Erk activation through phosphorylation-induced MKP3 degradation and later, after prolonged PDGF-BB stimulation, in the reduction in Erk activity via resynthesis of MKP3.