Differential Up-regulation of MAP Kinase Phosphatases MKP3/DUSP6 and DUSP5 by Ets2 and c-Jun Converge in the Control of the Growth Arrest Versus Proliferation Response of MCF-7 Breast Cancer Cells to Phorbol Ester*

Different levels of regulation account for the inactivation of MAP kinases by MAPK phosphatases (MKPs), in a cell type- and stimuli-dependent manner. MCF-7 human breast carcinoma cells treated with the phorbol 12-myristate 13-acetate (PMA) suffer growth arrest and show morphological alterations, which depend on the activation of the ERK1/2 MAP kinases. MKP3/DUSP6 and DUSP5 MAP kinase phosphatases, two negative regulators of ERK1/2, were specifically up-regulated in MCF-7 and SKBR3 cells in response to PMA. MKP3 and DUSP5 up-regulation required the prolonged activation of the ERK1/2 pathway, and correlated with the shutdown of this route. MKP3 induction relied on the activation of the Ets2 transcription factor, whereas DUSP5 induction depended on the activation of c-Jun. Diminishing the expression of MKP3 and DUSP5 raised the activation of ERK1/2, and accelerated growth arrest of PMA-treated MCF-7 cells. Conversely, MCF-7 cell lines expressing high levels of MKP3 or DUSP5 did not undergo PMA-triggered growth arrest, displayed a migratory phenotype, and formed colonies in soft agar. We propose that the differential up-regulation of MKP3 by Ets2 and of DUSP5 by c-Jun may converge in similar functional roles for these MAP kinase phosphatases in the growth arrest versus proliferation decisions of breast cancer cells.

Interestingly, based on its anti-proliferative and/or pro-apoptotic properties on acute myeloid leukemia cells, which correlate with sustained activation of ERK1/2, PMA is under clinical trial for hematological malignancies (9,10).
The magnitude and duration of MAPKs activation during the cellular response to external cues can be decisive for cell fate. This makes MAPKs essential players in cellular processes such as proliferation, growth, division, survival, and differentiation. Regulation of MAPKs signaling is important in breast cancer, and MAPKs have been proposed as molecular targets for anticancer therapy (11,12). From the three major MAPK pathways that exist in human tissues, ERK1/2, JNKs, and p38s, the ERK1/2 pathway seems to be the most relevant in breast cancer (13,14). Sustained activation of ERK1/2 in distinct cell lines, including the breast carcinoma cell line MCF-7, favors senescence and cell cycle arrest, whereas ERK1/2 transient activation favors proliferation (4,(15)(16)(17)(18)(19). However, the cell type-specific molecular mechanisms that control the transient versus sustained activation of MAPK are poorly known. ERK1/2 are activated upon phosphorylation of the Thr and Tyr residues at their kinase activation loop by their specific MAPK kinases, MEK1/2. MAP kinase phosphatases (MKPs) constitute a family of dualspecificity phosphatases that inactivate the MAPKs by dephosphorylating both the phospho-Thr and phospho-Tyr regulatory residues. Ten different MKPs exist in mammalian cells that show different substrate specificity toward ERK1/2, JNKs and p38s, and which can inactivate these MAPKs both in the cytosol and nucleus (20). MKPs have been related with the development of breast cancer and with the outcome to different chemotherapeutic treatments (21). For example, MKP1/DUSP1 and MKP2/DUSP4 have been found to be overexpressed in breast cancer cell lines as well as in malignant breast cancer samples, in comparison with non-malignant samples (22). On the other hand, DNA microarray analysis of gene expression profiling of breast carcinomas has also delineated reduced expression of DUSP5 in malignant transformation (23). MKP1 mediates resistance to various breast cancer chemotherapeutic agents, including mechlorethamine, cisplatin, doxorubicin, paclitaxel, and bortezomib (24 -29), and elevated expression of MKP3/DUSP6 is sufficient to confer tamoxifen resistance in breast cancer cell lines (30). Intriguingly, the prolonged activation of ERK1/2 by constitutively active upstream kinases Raf-1 and MEK1 has also been linked with anti-estrogen resistance in MCF-7 cells (31,32), and a MEK1-independent sustained activation of ERK1/2 has been proposed to account for such resistance (33). Finally, increased levels of PAC1/DUSP2 are important in apoptotic signaling in breast cancer cells treated with the chemotherapeutic agent N-4-hydroxyphenylretinamide (34). Together, these findings suggest an active role for MKPs in both breast cancer and in the resistance to current breast cancer therapies.
In this study, we have performed a comprehensive analysis of MKP expression in human breast carcinoma cells treated with PMA. We have found the up-regulation of two ERK1/2-specific phosphatases, MKP3 and DUSP5, in MCF-7 and SKBR3 cell lines, which depended on the differential expression of Ets2 and c-Jun transcription factors. The study of the role of MKP3 and DUSP5 on cellular functions controlled by ERK1/2 revealed similar functional patterns for these two enzymes. Our findings indicate that up-regulation of MKP3 and DUSP5 by distinct transcription factors converge in the adaptative decisions of proliferation and migration versus growth arrest of MCF-7 cells exposed to PMA.
Microarray Analysis-Total RNA was extracted from two different empty vector MCF-7/Tet-On cells treated or not with PMA for 4 days. RNA was quantified by spectrometry (NanoDrop) and quality confirmed by a RNA 6000 Nano Bioanalyzer (Agilent Technologies) assay. 600 ng of total RNA were used to produce Cyanine 3-CTP-labeled cRNA using the Low RNA Input Linear Amplification Kit, PLUS, One-Color (Agilent) according to the manufacturer's instructions. Following One-Color Microarray-based Gene Expression Analysis protocol version 5.5 (Agilent), 3 g of labeled cRNA was hybridized with the Whole Human Genome Oligo Microarray Kit (Agilent) containing 41,000ϩ unique human genes and transcripts. Arrays were scanned in an Agilent Microarray Scanner according to the manufacturer's protocol and data were extracted using Agilent Feature Extraction Software 9.5.3. MKP gene expression was obtained from normalized data, and calculation of fold-change was performed dividing values from PMA treatment to control.
Semi-quantitative and Quantitative Real Time PCR-Semiquantitative RT-PCR was performed using poly(A) mRNA from MDA-MB-231, MDA-MB-468, BT474, SKBR3, and MCF-7 cells treated or not with PMA for 4 days, using the illustra TM QuickPrep Micro-mRNA Purification Kit (GE Healthcare). 10 ng of poly(A) RNA were reverse transcribed using RevertAid TM reverse transcriptase, oligo(dT) 18 primers, and RiboLock and RNase inhibitor (all from Fermentas). To assess the expression of the different MKP mRNAs, PCRs were performed on the synthesized cDNA samples (100 ng/reaction) using sets of specific MKP or ␤-actin (as a control) primers, as described (40), and GC-rich DNA polymerase (Fermentas). Quantification of the ethidium bromide-stained bands from agarose gels was done using the ImageQuant (TL version 2005) software package (Amersham Biosciences, GE Healthcare). Quantitative RT-PCR was performed using total RNA from MCF-7 cells treated or not with PMA for 1, 2, 3, and 4 days, and for 3 days in MDA-MB-231, MDA-MB-468, BT474, and SKBR3 cells, using illustra RNAspin mini purification kit (GE Healthcare). 1 g of total RNA was subjected to reverse transcription as indicated above. Quantitative real time PCRs were performed using Lightcycler 480 (Roche Applied Science) with the corresponding SYBR Green I Master (Roche Applied Science), and validated primer sets (Qiagen) specific for the MKPs, ER␣, p21 Waf1/Cip1 , Ets2, c-Jun, and the housekeeping gene hypoxanthine-guanine phosphoribosyltransferase. All quantifications were normalized to the hypoxanthine-guanine phosphoribosyltransferase data. Relative quantification was performed using the comparative ⌬⌬C t method according to the manufacturer's instructions.
Anchorage-independent Growth Assay in Soft Agar-Cells with or without 24 h of pre-treatment with PMA were plated at a density of 2500 cells per well (12-well plates) in 0.5 ml of complete medium plus doxycycline and 0.35% cell culturetested agar (Sigma), onto the solidified bottom layer of 0.5 ml of complete medium plus doxycycline and 0.4% agar. Colo-nies were stained after 2-3 weeks with 0.05% crystal violet, and photographed at ϫ4 and ϫ40 magnification.
Wound Healing Migration Assay-Cells were plated at a density of 1 ϫ 10 6 cells per well (6-well plates) and grown in complete medium plus doxycycline for 48 h. Subsequently, the cell monolayers were scratched with a sterile micropipette tip and incubated for another 24 h with or without PMA. For each sample, six defined areas were monitored during this period. The photographs were taken at the beginning of the assay (t ϭ 0 h) and 24 h later (t ϭ 24 h) at ϫ100 magnification, and distance between the wounded areas was measured.

Expression of MKPs in Growth-arrested MCF-7 Cells upon
PMA Treatment-The phorbol ester PMA triggered a strong and sustained activation of ERK1/2 in human breast carcinoma MCF-7 cells, whereas a weaker activation was observed for p38 and JNK MAPKs (Fig. 1). To investigate the regulation of MAPKs in MCF-7 cells upon PMA treatment, we performed a comprehensive analysis of the expression of the 10 active members of the MKP family of MAPK phosphatases, which are potential MAPK negative regulators (20). MKP mRNA expression was monitored using an unbiased gene array analysis ( Fig.  2A), as well as semi-quantitative (Fig. 2B) and real-time quantitative RT-PCR (qRT-PCR) (Fig. 2C). Together, our experiments demonstrate a significant increase in the mRNA levels of both MKP3/DUSP6 and DUSP5, which peaked after 72 h of FIGURE 1. Activation of MAPKs on human breast carcinoma MCF-7 cells treated with PMA. Cells were incubated in the presence of PMA (50 ng/ml) during the indicated times, and the activation status of ERK1/2, p38, and JNK determined using phospho-specific anti-ERK1/2 (␣-pERK1/2), anti-p38 (␣-pp38), or anti-JNK (␣-pJNK) antibodies. The total amount of ERK1/2, p38, and JNK is also shown. GAPDH expression is included as a loading control. The arrows in the ␣-pJNK and ␣-JNK panels indicate the migration of JNK isoforms. A representative experiment is shown of at least three different experiments. PMA treatment. An increase in the mRNA levels of MKP2/ DUSP4, MKP5/DUSP10, and B59/DUSP7 was also detected in some of our experimental approaches, although in some others these changes were not observed, and a slight decrease of PAC1/DUSP2 mRNA content was noticed. MKP1/DUSP1 and hVH5/DUSP8 mRNA levels remained unchanged after PMA stimulation, whereas MKP4/DUSP9 and MKP7/DUSP16 mRNAs were not detected in MCF-7 cells (Fig. 2, A-C). Thus, we focused our studies on MKP3 and DUSP5. As illustrated in Fig. 2D, the kinetics of up-regulation of MKP3 and DUSP5 mRNAs upon PMA treatment of MCF-7 cells was delayed with respect to the kinetics of activation of ERK1/2, suggesting that MKP3 and DUSP5 mRNA up-regulation could be a consequence of ERK1/2 activation (see below). MKP3 and DUSP5 protein content was also increased in MCF-7 cells treated with PMA (Fig. 2E), although the increase in MKP3 protein content was weak, likely a result of MKP3 degradation triggered by activation of the ERK1/2 pathway (41)  MKP3 and DUSP5 Up-regulation in PMA-treated MCF-7 Cells Depends on Sustained ERK1/2 Activation and Involves Different Transcription Factors-Using specific inhibitors, we tested the involvement of distinct signaling pathways, including ERK1/2, p38, JNK, PI3K/Akt, and PKC pathways, in the up-regulation of MKP3 and DUSP5 in MCF-7 cells treated with PMA. We also monitored, using activation-specific anti-phosphoantibodies, the activation status of the ERK1/2 and PI3K/Akt pathways under inhibition conditions (Fig. 3A). As shown, the PKC inhibitor GF109203X strongly diminished the PMA-triggered up-regulation of MKP3, DUSP5, and p21 Waf1/Cip1 , in agreement with previous reports that identified PKC as the major upstream target of PMA in MCF-7 cells (4,7,42,43). The ERK1/2 pathway-specific inhibitor PD98059 partially inhibited up-regulation of MKP3 and DUSP5, at both the mRNA and protein levels. On the contrary, the p38-specific SB203580 inhibitor increased MKP3 and DUSP5 up-regulation, whereas, the JNK-specific SP600125 inhibitor, or the PI3K-specific wortmannin inhibitor, showed no effect (Fig. 3A). A good positive correlation was observed between the degree of activation of ERK1/2 in distinct conditions and the extent of MKP3 and DUSP5 mRNA and protein up-regulation, demonstrating that ERK1/2 activation is responsible, at least in part, for such an up-regulation. Remarkably, the SB203580 inhibitor increased the pERK1/2 content, which was paralleled with increased MKP3 and DUSP5 mRNA and protein content (Fig. 3A). We also observed a partial inhibition of PMA-induced p21 Waf1/Cip1 up-regulation by inhibition of MEK/ERK and PKC pathways, consistent with previous reports (7,43,44).
Next, we tested whether sustained activation of ERK1/2 was necessary for accumulation of MKP3 and DUSP5 mRNAs and proteins. To this end, the PD98059 inhibitor was added to the cultures after 3 and 24 h of PMA cell treatment, and MKP3 and DUSP5 mRNA and protein expression was measured after a total of 48 h in the presence of PMA. As shown, inhibition of the ERK1/2 pathway during PMA treatment inhibited a further increase in the expression of MKP3 and DUSP5 mRNAs (Fig.  3B, upper panel), which was accompanied by an inhibition of MKP3 and DUSP5 protein accumulation (Fig. 3B, lower panel). This correlates with the inhibitory effect of PD98059 on ERK1/2 activation (Fig. 3B, lower panel). These results indicate that the sustained activation of ERK1/2 is required to maintain the increased expression of these two MKPs in MCF-7 cells in response to PMA.
Members of the Ets (E26) family of transcription factors are activated downstream of ERK1/2 and have been proposed to be involved in breast tumor progression (45,46). Ets1/2 has been shown to regulate MKP3 transcription in non-breast cancer cell lines (47)(48)(49). Ets2 mRNA, but not Ets1 mRNA, was detected in MCF-7 cells by semi-quantitative RT-PCR (Fig. 4A, upper left panel), and the gene array experiments showed up-regulation of Ets2 mRNA, but not of Ets1, in MCF-7 cells upon PMA treatment (Fig. 4A, lower left panel). qRT-PCR measurements and immunoblot experiments confirmed this finding and showed a sustained accumulation of Ets2 mRNA and protein in both MCF-7 and SKBR3 cells cultivated in the presence of PMA (Fig. 4A, middle and right panels). Remarkably, accumulation of Ets2 protein was accompanied by its activation, as reflected by its phospho-active content (Fig.  4A, right panels). Both, MKP3 and DUSP5 promoters display putative Ets2 binding sites (Fig. 4B). To test the involvement of Ets2 in PMA-induced transcription of MKP3 and DUSP5 in MCF-7 cells, Ets2 mRNA and protein were down-regulated by specific RNA silencing (Fig. 4C, left panel) and MKP3 and DUSP5mRNAlevelswerequantified (Fig.4C,rightpanel).PMAdependent up-regulation of both MKP3 and p21 Waf1/Cip1 mRNAs was prevented upon silencing Ets2 expression, whereas up-regulation of DUSP5 mRNA was unaffected (Fig. 4C, right  panel). c-Jun can also be up-regulated by PMA through ERK1/2 (8,50,51), and putative binding sites for the AP-1 transcription factor are found upstream from the translation initiation sites of MKP3 and DUSP5 genes (TFsitescan (52), not shown). Thus, we tested the involvement of c-Jun in up-regulation of MKP3 and DUSP5 mRNAs in MCF-7 cells. c-Jun protein expression and activation were augmented in MCF-7 cells upon PMA stimulation (Fig. 4D, left panel). Remarkably, MKP3 mRNA levels were not affected upon silencing of c-Jun, whereas the PMAinduced DUSP5 mRNA up-regulation was abrogated (Fig. 4D, right panel), in correlation with an increase in pERK1/2 content (Fig. 4D, left panel). Together, these results demonstrate a differential regulation of MKP3 and DUSP5 gene transcription on PMA-treated MCF-7-cells: MKP3 transcription mostly depends on expression of active Ets2, whereas DUSP5 transcription depends on expression of active c-Jun.

Silencing of MKP3 and DUSP5 Favors Filopodia Formation and Growth Arrest of MCF-7 Cells in Response to PMA-To
investigate the effects of MKP3 and DUSP5 up-regulation in the phenotype of MCF-7 cells treated with PMA, specific RNA silencing experiments were performed targeting these two MKPs. The efficiency of silencing was about 56% for MKP3 and 70% for DUSP5 at the mRNA level (Fig. 5A). Upon PMA treatment, cell morphological changes, quantified as the percentage of cells with filopodia, were displayed earlier and were more intense in MKP3-or DUSP5-silenced cells (Fig. 5, B and C), concomitantly with higher levels of activation of ERK1/2 and diminished levels of expression of ER␣, when compared with PMA-treated, control-silenced cells (Fig. 5D). Basal p21 Waf1/Cip1 levels in silenced MKP3 or DUSP5 cells were increased, and after 3 h of PMA treatment the induced levels were higher than in the control-silenced cells (Fig. 5D). Double silencing of MKP3 and DUSP5 did not result in a significant additive effect on the response of mcF-7 cells to PMA, suggesting that the effects of MKP3 and DUSP5 in ERK1/2-mediated functions are not cooperative (data not shown). These results demonstrate that endogenous MKP3 and DUSP5 exert on MCF-7 cells a negative regulation on ERK1/2-mediated responses to PMA.

Ectopic Expression of MKP3 and DUSP5 in MCF-7 Blocks Cell Growth Arrest, and Favors Soft Agar Colony Formation and
Migration in Response to PMA-The functional consequences of MKP3 and DUSP5 overexpression on the response of MCF-7 cells to PMA were also investigated. Stable MCF-7 clones were generated that overexpressed MKP3 or DUSP5 wild type or catalytically inactive forms (MKP3 C293S, DUSP5 C263S) in a doxycycline-dependent inducible manner. Overexpression of ectopic MKP3 and DUSP5 was efficiently induced upon doxycycline treatment, although some leakage expression was observed in the absence of doxycycline, especially in the case of  In the middle panel, Ets2 mRNA up-regulation was measured by qRT-PCR in MCF-7 and SKBR3 cells upon PMA (50 ng/ml) cell treatment during several periods of time, and relative expression values are shown. In the right panel, Ets2 protein up-regulation and activation (phospho-Ets2, pEts2) in MCF-7 and SKBR3 cells treated with PMA was monitored by immunoblot. Note that Ets2 up-regulation in SKBR3 cells followed a faster kinetics. B, alignment of MKP3 and DUSP5 promoters showed putative Ets2 transcription factor binding sites. Aligned upstream MKP3 genomic sequence, NC_000012.11:c89746400 -89746297; MKP3 genomic sequence, NC_000012.11:c89746296 -89741837; aligned upstream DUSP5 genomic sequence, NC_000010.10:c112257572-112257624; DUSP5 genomic sequence, NC_000010.10:c112257625-112271302. Alignments were made using EMBOSS pairwise alignment algorithms (72). C, transcription factor Ets2 is responsible for up-regulation of MKP3 and p21 Waf1/Cip1 in MCF-7 cells in the response to PMA. In the left panel, Ets2 protein expression and activation were monitored by immunoblot, from untreated (Ϫ) or PMA-treated (24 h) (ϩ) MCF-7 cells. Transfection with negative control (siNS, nonspecific silencer) or with Ets1-or Ets2-specific (siEts1, siEts2) silencing oligonucleotides was performed 48 h before PMA stimulation. The phospho-Ets2, Ets2, pERK1/2, and ERK1/2 contents under the silencing conditions are shown. In the right panel, mRNA levels of MKP3, DUSP5, and p21 Waf1/Cip1 , from silenced control or Ets2-silenced MCF-7 cells, grown in the absence (Ϫ) or presence of PMA (3 h) (ϩ), were measured by qRT-PCR, and relative expression values are shown. Results represent the mean Ϯ S.D. (*, p Ͻ 0,005). D, transcription factor c-Jun is responsible for up-regulation of DUSP5 in MCF-7 cells in the response to PMA. Cells were silenced with nonspecific (siNS) or with c-Jun-specific (sic-Jun) oligonucleotides, as in C, and phospho-c-Jun (pc-Jun), c-Jun, pERK1/2, and ERK1/2 content was monitored by immunoblot (left panel). In the right panel, MKP3, DUSP5, and p21 Waf1/Cip1 mRNAs were measured by qRT-PCR, as in C. Results represent the mean Ϯ S.D. (*, p Ͻ 0,005). Data are representative of more than 3 different experiments. In all the panels, a representative immunoblot is shown of at least three different experiments. The arrows indicate cells with filopodia. Cultures silenced for MKP3 of DUSP5 showed more cells with filopodia. C, quantification of filopodia from silenced MCF-7 cells. Cells were treated with PMA for 3 or 24 h and filopodia were counted. Data are shown as the mean Ϯ S.D. D, silencing of MKP3 and DUSP5 increased the activation of ERK1/2, caused a rapid decay of ER␣, and a faster up-regulation of p21 Waf1/Cip1 upon PMA treatment. 48 h after transfection with siRNA negative control (NS), siMKP3, or siDUSP5, cells were kept untreated or treated with PMA for 3 and 24 h, before harvesting. pERK1/2, ERK1/2, ER␣, and p21 Waf1/Cip1 levels were analyzed by immunoblot. Actin expression is included as a loading control. A representative immunoblot is shown of at least three different experiments.
DUSP5 was mainly nuclear (data not shown). In the absence of PMA, no significant differences were observed in the appearance of control cells when compared with cells overexpressing MKP3 or DUSP5. In the presence of PMA, the cells from clones expressing wild type MKP3 or DUSP5 abandoned their mosaic-like structure and dispersed on the plate surface more rapidly than cells from the control clones or from the clones expressing catalytically inactive MKP3 or DUSP5 (Fig. 6B). Also, MCF-7 cells that overexpressed wild type MKP3 or DUSP5, but not those overexpressing their catalytically inac- In B, empty vector-control cells, and MKP3-and DUSP5-inducible cell lines, were kept untreated (Ϫ) or were treated (ϩ) with PMA for 4 days, and pictures of the cultures were taken. In C, the distinct cell lines were grown in the presence of PMA from 1 to 4 days, and cell proliferation was measured by the 3-[4,5-dimethylthiazol-2-yl]-2,5-dephenyltetrazolium bromide assay. Note the significant increase in cell growth in the presence of PMA of the MKP3 and DUSP5 cell lines. D, p21 Waf1/Cip1 protein levels on stable MCF-7 cell lines expressing MKP3 and DUSP5. Cells were incubated in the presence of PMA during 24 h, and levels of p21 Waf1Cip1 and GAPDH were determined by immunoblot. E, stable MCF-7 cell lines expressing MKP3 and DUSP5 do not display prolonged activation of ERK1/2 upon PMA treatment. Cells were incubated in the presence of PMA during the indicated times, and levels of MKP3, DUSP5, pERK1/2, and ERK1/2 were determined by immunoblot. In all experiments, results from one clone are shown, which are representative of results from at least two different clones. In A, D, and E, a representative immunoblot is shown of at least three different experiments.
tive forms, proliferated more in the presence of PMA than the empty vector control cells, as measured by the 3-[4,5-dimethylthiazol-2-yl]-2,5-dephenyltetrazolium bromide cell proliferation assay (Fig. 6C), indicating that the phosphatase activity of MKP3 and DUSP5 overcomes the cell growth arrest triggered by PMA. In agreement with the above results, a modest decrease in the protein levels of p21 Waf1/Cip1 was observed in PMA-treated MKP3-or DUSP5-MCF-7 cells, as compared with control cells (Fig. 6D). As expected, the clones expressing wild type MKP3 or DUSP5 showed an impairment in the sustained activation of ERK1/2 caused by PMA stimulation (Fig.  6E). Nonetheless, the inhibition on ERK1/2 activation in the DUSP5 clones was consistently higher than inhibition on the MKP3 clones. This correlated with the inhibition of the PMAinduced cell growth arrest, suggesting that MKP3 and DUSP5, by restraining ERK1/2 sustained activation, facilitate the proliferation of MCF-7 cells upon PMA treatment.
Next, soft agar colony formation and migration of MCF-7 cells ectopically expressing MKP3 or DUSP5 were investigated. Cells from the parental MCF-7 cell line or the control (empty vector) MCF-7 clones form small colonies in soft agar, and PMA treatment abrogates this capacity (Fig. 7A). The soft agar colony formation in the absence of PMA was not affected in clones expressing MKP3 or DUSP5 (wild type or catalytically inactive mutant). Also, the inhibition of colony formation upon PMA treatment was unaffected in clones expressing catalytically inactive MKP3 or DUSP5. However, the clones expressing wild type MKP3 or DUSP5 formed colonies in soft agar in the presence of PMA (Fig. 7A), indicating that persistent inactivation of ERK1/2 by these MKPs prevents the inhibition of MCF-7 cell colony growth caused by the PMA.
The migration properties of the MKP3-and DUSP5-MCF-7 clones were also assessed by wound repair assays (Fig. 7B). In the absence of PMA, cells did not migrate into the platewounded area (data not shown). In the presence of PMA, cells from the control clones or clones expressing inactive MKP3 or DUSP5 migrated poorly into the wounded area. However, cells from clones expressing wild type MKP3 or DUSP5 displayed a high capacity of migration into the wounded area upon PMA treatment (Fig. 7B). This indicates that activity of MKP3 or DUSP5 promotes a migrating phenotype on MCF-7 cells treated with PMA. Together, our experiments demonstrate a positive role for MKP3 and DUSP5 in proliferation, soft agar colony formation, and migration of MCF-7 cells exposed to PMA.

DISCUSSION
The MKP family of dual-specificity phosphatases (MKPs) include 10 active enzymes in mammals, which share structural and functional properties, the most relevant being the presence of a conserved protein-tyrosine phosphatase domain that specifically dephosphorylates the regulatory Thr(P) and Tyr(P) residues of MAPKs, inactivating these enzymes. MKPs differ in their expression patterns, subcellular locations, and physiological functions, and play tissue-and/or developmentally specific roles directly related with the regulation of MAPK pathways (53). In this work, we have investigated the regulation of the expression and function of MKPs in the MCF-7 human breast cancer cell line grown in the presence of PMA, a stimulus that induces growth arrest in these cells by activation of PKC and ERK1/2 (Fig. 1). We have detected eight MKP mRNAs in MCF-7 cells, including MKP1/DUSP1, MKP2/DUSP4, MKP3/ DUSP6, MKP5/DUSP10, DUSP5, PAC1/DUSP2, DUSP8, and DUSP7. MKP4/DUSP9 and MKP7/DUSP16 mRNAs were mostly undetected in MCF-7 cells. A slight down-regulation of PAC1 mRNA levels was observed in MCF-7 cells upon PMA treatment. PAC1 is a transcription target of p53, and functions in vivo as a positive regulator of immune response and inflammation (54,55). Down-regulation of PAC1 expression has been found in human acute leukemias; and in breast cancer cells, PAC1 is the transcription target of E2F-1, and PAC1 overexpression inhibits tumorigenicity (34,54,56). On the other hand, PMA treatment of MCF-7 cells triggered mRNA up-regulation of MKP2, MKP3, MKP5, DUSP5, and DUSP7, which was accompanied by a sustained ERK1/2 activation. Because MKP3 and DUSP5 displayed the more robust increase in mRNA levels, we focused our studies on these two phosphatases, which were also found to be up-regulated in SKBR3 human breast carcinoma cells upon PMA treatment. Remarkably, both MKP3 and DUSP5 specifically inactivate ERK1/2, and display cytoplasmic and nuclear localizations, respectively (38,57). The kinetics of mRNA up-regulation of MKP3 and DUSP5 was delayed with respect to the activation of ERK1/2, which was optimal after 30 min and kept sustained along 48 h. This is in accordance with MKP3 and DUSP5 being up-regulated as a consequence of ERK1/2 activation, as it has been reported for MKP3 in the developing chick somite and in human breast and colon cancer cells (58 -60). Thus, inhibition of the ERK1/2 pathway by PD98059, at the beginning or during the course of PMA treatment, diminished the increase of MKP3 and DUSP5 levels, whereas inhibition of JNK or PI3K/Akt pathways by SP600125 or wortmannin, respectively, did not affect MKP3 and DUSP5 levels. Intriguingly, the SP600125 inhibitor consistently blocked the basal-and PMA-induced activation of Akt in MCF-7 cells, as measured by its phosphoactive content (Fig. 3A), suggesting a cross-talk between the JNK and PI3K pathways in this cell system. The ERK1/2 pathway inhibitor PD98059 partially abrogated MKP3 and DUSP5 mRNAs increase. We attribute this partial inhibitory effect to an incomplete inhibitory effect of PD98059 in the activation of the ERK1/2 pathway under our experimental conditions. In fact, a good correlation between ERK1/2 activation and up-regulation of MKP3 and DUSP5 mRNA and protein is observed when the effect of the distinct inhibitors is compared. Interestingly, inhibition of the p38 pathway by the SB203580 inhibitor resulted in a further increase in mRNA and protein levels of MKP3 and DUSP5 upon PMA treatment, suggesting a negative role for p38 MAPKs in the expression of these two MKPs in MCF-7 cells.
Our studies indicate that up-regulation of MKP3 in MCF-7 cells treated with PMA relies, at least in part, on the Ets2 transcription factor. Ets2 mRNA levels were increased in MCF-7 cells after PMA treatment, as well as in SKBR3 cells, and silencing Ets2 prevented MKP3 up-regulation. These findings are in accordance with the positive role of Ets2 in MKP3 transcription in other cell types (47,48), and with the finding that ERK1/2 phosphorylate Ets2 in MCF-7 cells treated with PMA (61). Although both MKP3 and DUSP5 possess putative Ets2 binding sites, DUSP5 mRNA levels were not affected in MCF-7 cells by Ets2 silencing. By contrast, silencing of c-Jun inhibited PMA-dependent DUSP5 mRNA upregulation, but not MKP3 mRNA up-regulation. These findings suggest a differential transcriptional control of MKP3 and DUSP5 by Ets2 and c-Jun, respectively, upon PMA stimulation of MCF-7 cells. Remarkably, both Ets2 and c-Jun activities have been associated with human breast cancer (62)(63)(64). In addition, the steady-state levels of MKP3 and DUSP5 proteins also seem to be differentially regulated in MCF-7 cells treated with PMA, the DUSP5 protein being more stable and/or more efficiently translated than the MKP3 protein. At this regard, MKP3 phosphatase activity, but not DUSP5 phosphatase activity, is activated upon binding to ERK1/2 (38,65). Also, MKP3 is a cytoplasmic protein, whereas DUSP5 is nuclear, making possible that the existence of different regulatory mechanisms of the transcription, translation, and/or activation of these two phosphatases reflects the different timing and extent of dephosphorylation of cytoplasmic and nuclear ERK1/2 after the initial activating stimulus. In line with this hypothesis, DUSP5 mRNA up-regulation in PMA-treated MCF-7 cells followed a kinetic slightly slower that MKP3 mRNA up-regulation (Fig. 2C), and the DUSP5 inhibitory effect on ERK1/2 activation was consistently stronger than the MKP3 effect (Fig. 6E). This is unlikely to be due to a weak intrinsic catalytic activity of MKP3 toward ERK1/2, because MKP3 caused a robust ERK2 dephosphorylation in transient co-expression experiments in HEK293 cells (data not shown). Remarkably, in the T89G human glioblastoma cell line, both MKP3 and DUSP5 mRNA were also up-regulated upon PDGF-induced ERK1/2 activation (66), and it has been reported that DUSP5 is a direct transcriptional target of p53 in U373MG glioblastoma-astrocytoma cells (67). However, we have been unable to detect decreased expression of DUSP5 mRNA or protein upon silencing of p53 in PMA-treated MCF-7 cells (data not shown), whereas c-Jun silencing abrogated DUSP5 mRNA up-regulation. Thus, upon activation of the ERK1/2 pathway, MKP3 and DUSP5 seem to be up-regulated in coordination in a transcription factor-cell type-specific dependent manner.
Our functional results support a positive role for MKP3 and DUSP5 in the acquisition of a proliferative and migratory phenotype by MCF-7 cells subjected to PMA stimulation. First, silencing of MKP3 or DUSP5 accelerated p21 Waf1/Cip1 up-regulation in MCF-7 cells upon PMA treatment. And second, ectopic expression of wild type MKP3 or DUSP5 prevented cell growth arrest and the inhibition of colony growth in soft agar of MCF-7 cells grown in the presence of PMA, and stimulated migration of MCF-7 cells, which is reminiscent of the acquisition of invasive properties (68,69). We hypothesize that MKP3 and DUSP5, by specific local and/or temporal inhibition of ERK1/2 in the cytoplasm and nucleus, block ERK1/2-mediated cell growth arrest, without affecting the additional proliferative signals triggered by PMA. We propose a model in which the activity of Ets2 and c-Jun transcription factors increases the levels of MKP3 and DUSP5 through parallel but distinct pathways. These two pathways converge, at the functional level, in the impairment of the sustained activation of ERK1/2 and the favoring of a proliferative/migratory phenotype in MCF-7 cells treated with PMA (Fig. 8). This may be relevant in breast cancer, in which growth, apoptosis, and differentiation cyclic signals are unbalanced toward proliferation and migration, eventually leading to cell transformation and metastasis (14,46). Furthermore, the Ets2 and c-Jun pathways could be relevant for acquisition of resistance to chemotherapy of breast cancer  cells, which has been related with MKP3 tumorigenic and/or anti-apoptotic activity (30,70,71).