Glucocorticoid Receptor-induced MAPK Phosphatase-1 (MPK-1) Expression Inhibits Paclitaxel-associated MAPK Activation and Contributes to Breast Cancer Cell Survival*

Glucocorticoid receptor (GR) activation has recently been shown to inhibit apoptosis in breast epithelial cells. We have previously described a group of genes that is rapidly up-regulated in these cells following dexamethasone (Dex) treatment. In an effort to dissect the mechanisms of GR-mediated breast epithelial cell survival, we now examine the molecular events downstream of GR activation. Here we show that GR activation leads to both the rapid induction of MAPK phosphatase-1 ( MKP-1 ) mRNA and its sustained expression. Induction of the MKP-1 protein in the MCF10A-Myc and MDA-MB-231 breast epithelial cell lines was also seen. Paclitaxel treatment resulted in MAPK activation and apoptosis of MDA-MB-231 breast cancer cells, and both processes were inhibited by Dex pretreatment. Furthermore, induction of MKP-1 correlated with the inhibition of extracellular signal-regulated kinase (ERK1/2) and c-Jun N-terminal kinase (JNK) activity, whereas p38 activity was minimally affected. Blocking Dex-induced MKP-1 induction using small interfering RNA increased ERK1/2 and JNK phosphorylation and decreased cell survival. ERK1/2 and JNK inactivation was associated with Ets-like transcription factor-1 (ELK-1) dephosphorylation. To explore the gene expression changes that occur downstream of ELK-1 dephosphorylation, we used a combination In Kinase Assay— ERK, p38, JNK kinase assays (cid:1) (cid:3) M (cid:3) g ELK-1 fusion for ERK1/2 activity (cid:3) ATF-2 fusion for p38 the reac- tions terminated SDS sample The min, 2 min. SDS-polyacrylamide electrophoresis to a nitrocellulose The blots incubated with the corresponding phospho-specific against ELK-1, ATF-2, or c-Jun, respectively. Band intensities were using a GS-710 densitometer (Bio-Rad), and the relative fold change at each time following paclitaxel or Dex/paclitaxel with the untreated baseline

Glucocorticoids are well known for their anti-inflammatory and immunosuppressive properties, as well as for their essential role in embryonic development and tissue homeostasis (1). Most glucocorticoid-mediated effects are thought to be a consequence of the ability of the activated glucocorticoid receptor (GR) 1 to act as a transcription factor, either through a DNA binding-dependent mechanism or through cross-talk and/or interference with other transcription factors such as activator protein-1 (2,3), signal transducers and activators of transcription-5 (4), and nuclear factor-B (NF-B) (5,6). The accumulated evidence shows that glucocorticoids induce apoptosis in lymphocytes, leukemic cells, lymphoma cells, and multiple myeloma cells (7,8). In contrast, it has been reported that glucocorticoids inhibit tumor necrosis factor ␣-induced apoptosis in subcutaneous adipocytes (9), mediate cell survival in primary cultures of human and rat hepatocytes (10), and protect against growth factor withdrawal-induced apoptosis in mammary epithelial cells (11)(12)(13). More recently, we and others observed that glucocorticoids can inhibit chemotherapy-induced apoptosis in vitro (14) and in vivo (15). These opposite outcomes are likely due to cell type-specific differences affecting GR-mediated transcriptional regulation (16), although nongenomic mechanisms may play a role as well (17).
Mitogen-activated protein kinases (MAPKs) constitute a superfamily of related kinases that are activated by a diverse array of extracellular stimuli (18). These include extracellular signal-regulated kinases (ERK1/2), c-Jun NH 2 -terminal kinase (JNK), and the p38 protein kinase (19). MAPKs are activated by reversible Thr and Tyr phosphorylation of a conserved Thr-X-Tyr motif (20). A growing family of dual Thr-and Tyr-specific phosphatases (DUSPs), termed MAPK phosphatases (MKPs), has also been identified. The prototypic member of this family, MKP-1, also called DUSP-1, CL100, 3CH13, or ERP, was originally identified as an ERK-specific phosphatase (21,22). However, MKP-1 can also dephosphorylate and inactivate both the stress-activated protein kinase (SAPK)/JNK and p38 (23)(24)(25). The substrate specificity of MKP-1 can vary significantly depending on cell type and context. MKP-1 is considered an immediate early response gene, and its expression can be induced by multiple stimuli such as growth factors (21,22), cellular stress (26,27), and retinoids (28). While these studies were ongoing, dexamethasone (Dex) was shown to induce MKP-1 in several cell types including the RBL-2H3 mast and NIH 3T3 fibroblast (29), the HeLa cervical epithelial (25), the MBA-15.4 bone marrow stromal, and the MG-63 preosteoblast cell lines (30). Both the rapid induction of MKP-1 by Dex and the presence of at least three putative glucocorticoid response elements in its promoter region suggest that MKP-1 is transcriptionally regulated by the GR (29). MKP-1 has also been shown to be overexpressed in breast (31), gastric (32), ovarian (33), and pancreatic tumors (34). Additional studies have shown that overexpression of MKP-1 can protect against apoptosis in prostate (35) and breast (14,36) cancer cell lines, Chinese hamster ovary cells (37), the mesangial cell line SM43 (28,38), and the CESS B cell line (39).
In previous studies, Dex-mediated inhibition of paclitaxelinduced apoptosis has been demonstrated (14,40). Here we explored the mechanism of Dex-mediated cell survival and found that the induction of MKP-1 following Dex treatment results in the specific inhibition of paclitaxel-induced ERK1/2 and JNK activity. Furthermore, MKP-1 small interfering RNA (siRNA) inhibits GR-mediated ERK and JNK dephosphorylation and cell survival. In an effort to further elucidate the role of ERK and JNK inactivation in inhibiting paclitaxel-induced apoptosis, potential downstream targets of the Ets-like transcription factor-1 (ELK-1) were examined. Using a combination of time course microarray and promoter element analyses, we identified 12 genes that were down-regulated by Dex and contain ELK-1 binding sites. Human tissue plasminogen activator (tPA) was chosen for further analysis, and Dex pretreatment was observed to inhibit paclitaxel-induced tPA promoter activation through an ELK-1 binding motif. These results suggest that GR-mediated MKP-1 induction and the subsequent inactivation of ERK and JNK signaling contribute to inhibition of paclitaxel-induced apoptosis. A set of ELK-1-dependent target genes was identified whose members are down-regulated 2-24 h following GR activation, suggesting that glucocorticoid signaling mediates both early (direct) and later (indirect) gene expression changes leading to cell survival.
For siRNA knockdown, MDA-MB-231 cells were seeded at 5 ϫ 10 5 cells per 6-cm plate and allowed to adhere overnight. The next day, pSilencer 1.0-U6-MKP-1 siRNA or pSilencer 1.0-U6-control siRNA plasmid DNA (14) was transfected into cells using Polyfect. After 40 h, transfected cells were split equally into three 6-cm plates and allowed to adhere overnight. Paclitaxel (10 Ϫ7 M) was added for various time periods (8, 24, or 30 h) with or without Dex (10 Ϫ6 M) pretreatment and scored for apoptosis. Parallel experimental cell lysates were collected for Western analysis.
Microarray Experiment and Data Analysis-The procedure for cRNA preparation and hybridization to high density oligonucleotide arrays was described previously (14). Briefly, MCF10A-Myc cells were subjected to 72 h of growth factor withdrawal followed by treatment with either vehicle (ethanol), Dex (10 Ϫ6 M), or concomitant Dex/RU486 (10 Ϫ7 M) for 0.5, 2, 4, or 24 h. Total RNA from each sample was extracted using a Qiagen RNeasy kit. The preparation of biotinylated cRNA and hybridization to oligonucleotide arrays (Affymetrix HG-U133A) were performed at the University of Chicago Microarray Core Facility. Gene chips were scanned and analyzed using Affymetrix Microarray Suite 5.0. To identify genes significantly regulated by GR activation, the first variation taken into account was the Affymetrix software description of a gene's expression, namely "absent" (gene intensity below an Affymetrix calculated threshold), "present," or "marginal." Genes deemed absent at all four time points in both Dex-treated and control samples were excluded from further analysis, because these gene intensity values were below the threshold and, therefore, considered to be unreliable. The second step was to exclude genes with intensities Յ20 in both Dex-treated and control samples at all four time points. This is because false positives are expected to be more frequent in genes with very low levels of expression. The third step was to set a cutoff of Ն1.5-fold (Dex-treated versus vehicle-treated) for "induction" and Յ0.5-fold (Dextreated versus vehicle-treated) for "repression" at a given time point.
Our main goal was to identify the transcriptional targets of ELK-1. A time lag is predicted to exist between the initial induction of MKP-1 gene expression and the subsequent repression of ELK-1 transcriptional targets (due to MKP-1 protein synthesis and dephosphorylation of ERK1/2 and ELK-1). Therefore, we first obtained a set of genes that have an ELK-1 binding motif, GGAANNTTAA, in their promoter regions using the Data Base of Transcriptional Start Sites (dbtss.hgc.jp). This software searches for promoter sequences with a specific transcription factor binding motif between Ϫ1000 nucleotide and ϩ200 nucleotide. We subsequently identified putative ELK-1 targets by searching for a set of genes that both contained an ELK-1 binding motif and were down-regulated at either 2, 4, or 24 h following GR activation.
Northern Blot Analysis-Total RNA (20 g) was isolated and fractionated on a 1% agarose-formaldehyde gel and transferred to nylon membranes. The membranes were hybridized overnight to [ 32 P]dCTPlabeled cDNA probes encoding MKP-1 (a gift from Dr. Christele Debros-Monbers, Institut National de la Santé et de la Recherche Médicale (INSERM), Paris, France) or rat GAPDH (13). Membranes were then washed and exposed to film. Band intensity was quantified using a GS-710 densitometer (Bio-Rad), and the relative ratio of MKP-1 signal to GAPDH signal was determined. The fold change was calculated by normalizing the MKP-1 signal from Dex or Dex/Ru486-treated cells to the vehicle-treated cell signal.
In Vitro Kinase Assay-Non-radioactive ERK, p38, and JNK kinase assays were performed according to the manufacturer's protocol (Cell Signaling). Briefly, 20 l of resuspended immobilized phospho-ERK1/2, phospho-p38 MAPK monoclonal antibody, or c-Jun fusion protein beads were incubated with 200 g of total protein lysate with gentle rocking overnight at 4°C. The next day, samples were microcentrifuged for 30 s at 4°C. Pellets were washed twice with 500 l of 1ϫ lysis buffer and twice with 500 l of 1ϫ kinase buffer. Pellets were suspended in 50 l of 1ϫ kinase buffer supplemented with 100 M ATP and 2 g of ELK-1 fusion protein for ERK1/2 activity or 2 g of ATF-2 fusion protein for p38 activity. The samples were incubated for 30 min at 30°C, and the reactions were terminated with 25 l of 3ϫ SDS sample buffer. The samples were then boiled for 5 min, vortexed, and microcentrifuged for 2 min. Samples were then loaded (30 l) into a 12% SDS-polyacrylamide gel for electrophoresis and transferred to a nitrocellulose membrane. The blots were incubated with the corresponding phospho-specific antibodies against ELK-1, ATF-2, or c-Jun, respectively. Band intensities were quantified using a GS-710 densitometer (Bio-Rad), and the relative fold change at each time point following paclitaxel or Dex/paclitaxel treatment was compared with the untreated baseline signals.
Apoptosis Assays-After transfection, MDA-MB-231 cells were treated with either paclitaxel (10 Ϫ7 M) alone or Dex (10 Ϫ6 M)/paclitaxel for 8, 24, or 30 h. Cells were then fixed immediately by adding 500 l of 37% formaldehyde to each well in a dropwise fashion and incubated at room temperature for 30 min. The fixative solution was subsequently aspirated, and cells were allowed to dry overnight. To score for apoptosis, cells were stained with a 1 M 4,6-diamidino-2-phenylindole/phosphate-buffered saline solution as described previously (12). A Nikon Eclipse E800 microscope with UV illumination at 600ϫ magnification was used to count at least 200 4Ј,6-diamidino-3-phenylindole-stained cells to determine the   percentage of apoptotic cells per experimental condition. All apoptosis assays were performed independently three times to calculate the average percentage of apoptosis and the S.E. Statistical significance between two conditions in the apoptosis assays was determined by a one-sided t test. p Ͻ 0.05 was considered significant. Luciferase Reporter Assay-A 709-bp fragment of the human tPA gene (NM_000930) promoter containing an ELK-1 binding consensus site (GGAANNTTAA) (41) (ϩ76 nucleotide/ϩ85 nucleotide) was amplified from genomic DNA by PCR using a Kpn-1 restriction site (5Ј-GGGGTACCAGAGAAAGATTCTCCCTAAAATTAC-3Ј) and a BglII restriction site primer (5Ј-GAAGATCTCCCCAAGGTACAGAAACCCG-3Ј) and cloned into the pGL3 vector (a kind gift from Dr. Barbara Kee, University of Chicago) with Kpn-1 and BglII restriction enzymes. The fragment of the human tPA gene promoter was then mutated to intro-duce a single G to C substitution in the core of the ELK-1 binding consensus site (GCAANNTTAA) by site-directed mutagenesis (Quik-Change site-directed mutagenesis kit; Stratagene) using the primers 5Ј-CCCCACCCCCTGCCTGCAAACTTAAAGGAGGCCGG-3Ј (forward) and 5Ј-CCGGCCTCCTTTAAGTTTGCAGGCAGGGGGTGGGG-3Ј (reverse). was added to the plates, which were then shaken at room temperature for 15 min. Luminescence was measured using a Wallace luminescence counter (Beckman), and ␤-galactosidase activity was measured at 420 nm. The experiments were done three times. The relative luciferase activity in each condition was normalized to ␤-galactosidase activity and reported as an average Ϯ S.E.

Induction of MKP-1 following GR Activation in Mammary
Epithelial Cells-As reported in many cell types (25,29,30), MKP-1 is up-regulated immediately following Dex treatment in MCF10A-Myc cells (14). Gene array analysis of MCF10A-Myc cells showed that MKP-1 gene expression increases initially at 30 min and is up-regulated ϳ3-fold from 2 to 24 h (Fig. 1A, solid  line). Microarray data also demonstrated that the induction of MKP-1 mRNA by Dex is reversed by the partial GR-agonist/ antagonist RU486 (Fig. 1A, dashed line). To validate the microarray data, we performed a Northern blot analysis of MKP-1 mRNA expression from MCF10A-Myc cells treated with Dex. As shown in Fig. 1B, MKP-1 mRNA expression was up-regulated 3-fold relative to vehicle-treated cells at 30 min following Dex (10 Ϫ6 M) treatment (solid line) and remained elevated for 24 h. This induction was also reversed by concomitant treatment with 10 Ϫ7 M RU486 (dashed line). Fig. 1C shows that in MCF10A-Myc cells, MKP-1 protein levels also increased as early as 2 h following GR activation. Protein induction was sustained up to 36 h, and the increase in protein levels was inhibited by concomitant treatment with RU486.
Because MKP-1 is only one of several DUSPs that might be induced following GR activation, we examined other DUSP family members. Table I shows microarray analysis of mRNA expression levels for all DUSPs present on the HG-U133A chip. Only DUSP1 (MKP-1), DUSP3 (vaccinia virus phosphatase VH1-related, VHR) and DUSP9 (MKP-4) showed significant (Ͼ1.5-fold) up-regulation following Dex treatment. However, Q-RT-PCR showed that only MKP-1 was significantly up-regulated in comparison with the other DUSP family members (Fig. 1D). Taken together, these data suggest that both MKP-1 mRNA and the MKP-1 protein are significantly up-regulated following GR activation in mammary epithelial cells.

Dex-induced MKP-1 Expression Is Associated with the Dephosphorylation of MAPKs-We demonstrated previously that
Dex can inhibit MDA-MB-231 breast cancer cells from chemotherapy-induced apoptosis and, conversely, that inhibition of MKP-1 induction by siRNA decreases Dex-induced protection from apoptosis (14). Examination of MKP-1 mRNA expression in MDA-MB-231 cells by Q-RT-PCR showed that MKP-1 mRNA levels were elevated 4-fold at 30 min following Dex treatment, and the levels were sustained for 24 h (Fig. 1D). We next examined MKP-1 protein expression following either paclitaxel treatment alone or pretreatment with Dex. As Fig. 2A shows, the MKP-1 protein was induced at 30 min and sustained over 24 h following Dex pretreatment, but it was not induced with paclitaxel treatment alone.
The mechanism of MKP-1-induced cell survival signaling is not known. Depending on cell type, MKP-1 might dephosphorylate and inactivate any of the three components of the MAPK pathway (42). We therefore examined the molecular consequences of MKP-1 up-regulation. To determine the effect of GR activation and MKP-1 induction on known MKP-1 substrates, phosphorylation of ERK1/2, JNK, and p38 were examined by Western blot analysis using phospho-specific antibodies. Fig.  2B shows that phosphorylation of ERK1/2 occurs 2 h following paclitaxel treatment (left); however, ERK1/2 phosphorylation was inhibited as a result of Dex pretreatment (right). Similarly, phosphorylation of JNK was inhibited by Dex pretreatment. p38 phosphorylation was least affected by Dex. ERK1/2, JNK, and p38 total protein levels remained unchanged and are shown below each phospho-specific blot. Quantification of the individual phosphorylated MAPKs shows that phospho-ERK1/2 is most significantly inhibited by Dex, whereas JNK1/2 is somewhat inhibited. Phospho-p38 levels are not significantly affected by Dex pretreatment (Fig. 2C).

MKP-1 siRNA Inhibits both GR-induced ERK1/2 and JNK Dephosphorylation and Cell Survival-To determine whether MKP-1 induction is required for GR-mediated MAPK inactivation, we examined MAPK phosphorylation under conditions
where MKP-1 induction was specifically blocked by siRNA. MDA-MB-231 breast cancer cells were transiently transfected with a plasmid encoding either a control (scrambled sequence) siRNA or an MKP-1-specific siRNA. Forty hours after transfection, cells were treated with Dex followed by paclitaxel. MKP-1 steady-state levels were evaluated by Western blot analysis. As shown in Fig. 4A, MKP-1 protein expression increased in control siRNA-expressing cells following Dex treatment, but MKP-1 induction was blocked by the expression of siRNA. MKP-1 siRNA-expressing cells retained ERK1/2 and JNK phosphorylation (Fig. 4A) after Dex/paclitaxel treatment, and control siRNA-expressing cells showed the characteristic GR-mediated inhibition of ERK1/2 and JNK phosphorylation. Consistent with previous results showing that p38 is unaffected by Dex pretreatment, the level of phospho-p38 did not show a significant difference in control versus MKP-1 siRNA expressing cells. Cells from the same experiment were also evaluated for apoptosis (Fig. 4B). MKP-1 siRNA-expressing cells were more susceptible to paclitaxel-induced apoptosis after Dex treatment than cells transfected with control siRNA (p Ͻ 0.05) (Fig. 4B). These data suggest that MKP-1 induction contributes to Dex-mediated cell survival and is specifically required for ERK1/2 and JNK dephosphorylation following GR activation.
GR Activation Results in the Down-regulation of ELK-1 Transcriptional Target Genes-Because GR-induced MKP-1 inhibited MAPK signaling, we examined downstream targets of MAPK inactivation for their possible role in cell survival. Among MAPK substrates, ELK-1 is a transcription factor that is inactivated following dephosphorylation (43). We used a combination of microarray data and promoter element analyses to identify a set of putative transcriptional targets of ELK-1 (see "Materials and Methods"). As shown previously, time course microarray and Q-RT-PCR experiments showed a significant induction of MKP-1 following GR activation in MCF10A-Myc cells (Fig. 5A). We hypothesized that because of the time required for up-regulation of MKP-1 protein and subsequent ERK1/2 and ELK-1 dephosphorylation, a time lag of at least 1.5 h should exist between the increase of MKP-1 mRNA steady-state levels and the subsequent down-regulation of ELK-1 target genes. We therefore identified 910 genes that  ). B, schematic showing that either the wild type tPA promoter construct containing a consensus ELK-1 binding site (5Ј-GGAAACTTAA-3Ј) (top) or a consensus sequence with a single nucleotide mutation (mut) (5Ј-GCAAACTTAA-3Ј) (mut tPA) was cloned in-frame into a pGL3 luciferase expression vector. C, wild type (wt) tPA promoter activity (white bar) versus mut-tPA promoter activity (gray bar) following paclitaxel alone (10 Ϫ7 M) or Dex (10 Ϫ6 M)/paclitaxel treatment. Activity was normalized to ␤-galactosidase expression to calculate the relative luciferase activity. Data are shown as the mean of three independent experiments Ϯ S.E.
were down-regulated at least 50% following GR-activation at either 2, 4, or 24 h. Using the Data Base of Transcriptional Start Sites program, 366 genes in the human genome data base were found to have an ELK-1 binding motif between Ϫ1000 and ϩ200 nucleotides. An intersection of the set of GR down-regulated genes and the set of genes in the data base that have an ELK-1 binding motif gave rise to 12 putative targets of GR-induced ELK-1 inactivation (Table II). None of these 12 genes had been previously identified as transcriptional targets of ELK-1.
We chose to examine human tPA as a putative ELK-1 target gene based on published studies suggesting that tPA expression can be pro-apoptotic (44,45). Although tPA had been previously identified as a GR-regulated gene (46), the mechanism of its regulation is not completely understood. Time course microarray data demonstrated that tPA was down-regulated 0.58-fold at 4 h and 0.38-fold at 24 h (Fig. 5A). The results from Q-RT-PCR confirmed the repression of tPA steadystate levels following MKP-1 induction. An examination of the promoter region of tPA suggested an ELK-1 consensus binding motif (GGAANNTTAA) at the ϩ76 nucleotide/ϩ85 nucleotide location. To test whether tPA promoter activity is indeed sensitive to glucocorticoid treatment in an ELK-1 specific manner, we made the pGL3 tPA promoter luciferase construct containing either the putative wild type ELK-1 binding element or a single nucleotide mutation in the ELK-1 binding site (Fig. 5B). The pGL3-wt tPA (wild type) promoter or the pGL3-mut tPA (mutant) promoter construct was then co-transfected with a pCMV-␤-galactosidase reporter into MDA-MB-231 cells. Transfected cells were pretreated with either vehicle or Dex (10 Ϫ6 M) for 1 h prior to paclitaxel (10 Ϫ7 M) treatment. tPA promoter activity was measured by luciferase assay and normalized to ␤-galactosidase activity 6 h following paclitaxel treatment. Fig.  5C shows that paclitaxel treatment resulted in 5-fold higher activation of the wild type tPA promoter (white bar) compared with the mutant tPA promoter (gray bar) (p Ͻ 0.05). This suggests that ELK-1 activity (i.e. phosphorylation) is required for tPA promoter activation following paclitaxel treatment. However, when cells were pre-treated with Dex, tPA promoter activity was significantly inhibited, suggesting that Dex treatment resulted in inhibition of ELK-1 activity. It is likely that additional transcription factors play a role in tPA promoter regulation, because promoter activity was not abolished by Dex pretreatment and the consequent ELK-1 dephosphorylation. The results suggest, however, that ELK-1 activity contributes to tPA promoter activation. A model summarizing some of the downstream events following GR activation that have been uncovered by these studies is illustrated in Fig. 6.

DISCUSSION
Glucocorticoids exhibit diverse functions that regulate many cellular activities including metabolism, lymphocyte apoptosis, and epithelial cell survival in the setting of environmental stress (12,13,15,47,48). Using large scale microarray analyses of mRNA from Dex-treated mammary epithelial cells, our laboratory has identified a set of genes that are regulated following GR activation. One of the consistently up-regulated genes that appears to contribute to the mechanism of GRmediated inhibition of apoptosis is MKP-1 (14). We show here that MKP-1 induction results in the inhibition of ERK1/2 and JNK activity and contributes to cell survival.
While this manuscript was in preparation, Pettersson et al. (49) observed that sustained activation of ERK1/2 and p38, as well as a weaker activation of JNK, contributes to the combination of retinoid and PKC inhibitor-induced apoptosis in MDA-MB-231 cells. To identify the potential downstream targets of GR-mediated ERK1/2 and JNK inactivation, we identified 12 GR down-regulated genes with promoters containing ELK-1 consensus binding sites (Table II). All of these putative ELK-1 targets were down-regulated by at least 50% at one or FIG. 6. Working model of a GR-mediated cell survival signaling pathway. Both microarray data and traditional assessments of phosphorylated proteins were used to predict a putative signaling pathway. GR-activation up-regulates MKP-1 expression, which is associated with ERK1/2 and JNK dephosphorylation and followed by ELK-1 inactivation. Gene array analysis at later time points suggests that tPA expression is inversely related to MKP-1 induction. Therefore, a functional relationship between MKP-1 up-regulation, ELK-1 inactivation, and tPA down-regulation is proposed. more time points following GR activation. Among the 12 putative ELK-1 target genes, tPA and GATA-6 have been previously shown to be GR-regulated. Glucocorticoids down-regulate tPA in rat mesangial cells (46) and up-regulate GATA-6 in mouse intestine epithelial cells (50). The mechanism of transcriptional regulation of these genes by either GR activation and/or MAPK activation is likely to be cell type-and context-dependent. We found previously that ectopic expression of MKP-1 only partially substitutes for the anti-apoptotic effects of glucocorticoid (14). This observation suggests that additional mechanisms are likely to play important roles in GR-mediated cell survival. NF-B activity is known to be down-regulated by GR activation in many cell types (51). Dex inhibition of NF-B activity in multiple myeloma cells has been observed to contribute to apoptosis (52). However, Dex inhibition of NF-B activity can also result in chemotherapy resistance in certain solid tumor cell types (40,53). Our time course microarray data following Dex treatment of MCF10A-Myc cells suggest that some known NF-B-induced anti-apoptotic genes (e.g. cIAP-2, ␥GCS, Bcl-XL, A20, and Cyclin D2) (54) are up-regulated following GR activation. On the other hand, some known pro-apoptotic NF-B target genes (54) are down-regulated following GR activation (e.g. Fas, FasL, and p53). These results suggest that Dex treatment affects NF-B-mediated transcription in a complex manner that favors gene expression leading to cell survival.
We and others have found that, similarly as in MKP-1 induction, the expression of serum and glucocorticoid-induced protein kinase-1 (SGK-1) is also rapidly increased following GR activation (13,55). SGK-1 is an important downstream effector of phosphatidylinositol 3-kinase signaling (56). Whether GRmediated activation of phosphatidylinositol 3-kinase signaling and inhibition of the MAPK pathway converge in a common anti-apoptotic pathway remains to be determined. Recently, SGK-1 protein expression following DNA damage-mediated cell stress was shown to require both p53 and ERK1/2 activity in mouse embryo fibroblasts (57). This observation may provide a connection between GR-mediated MAPK inhibition (which is initiated ϳ2 h after GR activation) and the decrease in SGK-1 protein expression seen at later time points following GR activation (14). It is likely that GR activation regulates several complementary signaling pathways leading to cell survival. Identifying the detailed components of these pathways may uncover mechanisms of inappropriate tumor cell survival and resistance to chemotherapy-induced cell death.
In summary, we have demonstrated that MKP-1 up-regulation following GR activation plays an important role in inhibiting chemotherapy-induced MAPK activation and apoptosis. Our data suggest that inhibition of ERK and JNK phosphorylation, in turn, decreases ELK-1 phosphorylation and the steady-state expression levels of several putative ELK-1-dependent target genes. Analysis of time course microarray data from both early and later time points following GR activation is likely to identify additional pathways that can be evaluated for their role in cell survival signaling.