p38γ MAPK Cooperates with c-Jun in trans-Activating Matrix Metalloproteinase 9*

Mitogen-activated protein kinases (MAPKs) regulate gene expression through transcription factors. However, the precise mechanisms in this critical signal event are largely unknown. Here, we show that the transcription factor c-Jun is activated by p38γ MAPK, and the activated c-Jun then recruits p38γ as a cofactor into the matrix metalloproteinase 9 (MMP9) promoter to induce its trans-activation and cell invasion. This signaling event was initiated by hyperexpressed p38γ that led to increased c-Jun synthesis, MMP9 transcription, and MMP9-dependent invasion through p38γ interacting with c-Jun. p38γ requires phosphorylation and its C terminus to bind c-Jun, whereas both c-Jun and p38γ are required for the trans-activation of MMP9. The active p38γ/c-Jun/MMP9 pathway also exists in human colon cancer, and there is a coupling of increased p38γ and MMP9 expression in the primary tissues. These results reveal a new paradigm in which a MAPK acts both as an activator and a cofactor of a transcription factor to regulate gene expression leading to an invasive response.

MAPKs 3 (including extracellular signal-regulated kinases (ERKs), c-Jun N-terminal kinases (JNKs), and p38s) are critical signaling cascades that convert upstream signals into biological responses such as cell proliferation, invasion, and transformation (1). MAPKs are believed to do so by phosphorylating and activating a group of transcription factors, which through binding regulatory DNA elements lead to altered gene transcription. c-Jun is a major component of the AP-1 transcription factor downstream of MAPKs, whereas AP-1 is composed of homodimers of the Jun family or its heterodimers with another transcription factor such as c-Fos to bind the consensus DNA elements TGAg/cTCA (2). c-Jun is activated by JNK through phosphorylation at Ser-63, Ser-73, Thr-91, and Thr-93, and by ERK and p38 via increased gene expression. Activated c-Jun/ AP-1 leads to a cell type-specific biological response through integrated gene expression (1). However, the exact mechanism by which c-Jun converts a MAPK activity into a target gene expression remains mostly unknown.
p38 MAPKs consist of four family members (␣, ␤, ␥, and ␦) in which p38␣ is ubiquitously present, whereas p38␥ is highly expressed in certain cancers (3). In addition to well established regulatory effects in cytokine signaling and stress response, substantial evidence suggests that the p38␣ pathway functions as a tumor suppressor (4 -8). p38␥, on the other hand, is a 43-kDa protein with an unique C-terminal motif, KETXL, that can dock with the PDZ (PSD-95/Dlg/ZO-1 homology) domain of other proteins (9,10). In contrast to p38␣, our recent studies showed that p38␥ is induced by Ras and required for Ras transformation and invasion (11,12), indicating its oncogenic activity. The underlying mechanisms for p38␥ involvement in Ras tumorigenesis, however, have not been established. In this report, we show that p38␥ acts both as an activator and a cofactor for c-Jun in trans-activating MMP9, a critical matrix metalloproteinase involved in cancer invasion and metastasis (13,14). These results reveal a novel paradigm by which p38␥ increases c-Jun synthesis and activated c-Jun then recruits p38␥ as a cofactor onto a target gene promoter through AP-1 recognition leading to an increased gene expression and invasion.

EXPERIMENTAL PROCEDURES
Reagents, Cell Culture, and cDNA Constructs-Cell culture materials were supplied by Invitrogen and chemicals by Sigma. p38 isoform-specific antibodies were purchased from RD Systems. Glyceraldehyde-3-phosphate dehydrogenase, c-Jun, MMP9, and MMP2 antibodies were from Santa Cruz Biotechnology. Phosphorylated p38 (p-p38) and p-c-Jun (Ser-63/73) antibodies were from Cell Signaling. Mouse monoclonal antibodies against FLAG (M2) were from Sigma. IEC-6 cells as well as the procedure for establishing the Ras-transformed subline (IEC-6/K-Ras) were described previously (11). Human colon cancer cell lines were purchased from the American Type Culture Collection. p38␥ϩ/ϩ and p38␥Ϫ/Ϫ mouse embryonic fibroblasts (MEFs) have been described previously (10), and early passages of these cells were immortalized by infection with retroviruses expressing H-Ras and E1A. c-Junϩ/ϩ and c-JunϪ/Ϫ cells were provided by R. Wisdom (15) and have been used previously in our laboratory (16). All cell cultures were maintained in minimum Eagle's medium or Dulbecco's modified Eagle's medium containing 10% serum and antibiotics at 37°C, 5% CO 2 .
An AP-1 luciferase reporter (AP-1 Luc, 3 AP-1 repeats fused to a lucifease reporter gene containing a minimal Fos promoter) was described previously (4,17), whereas the wild-type (WT) and mutant human MMP-9 promoter (MMP9-Luc) were reported before (18). The Tet-on inducible expression system (T-Rex) was purchased from Invitrogen. A full-length human p38␥, its AGF mutant (p38␥/AGF), and the c-Jun luciferase promoter (c-Jun Luc, containing Ϫ225 to ϩ150 of the promoter) were provided by J. Han (19,20) and used previously in our laboratory (21). To generate a Tet-on system, a full-length human p38␥ or p38␥/AGF cDNA was cloned into pcDNA4 vector, which was then cotransfected with pcDNA6/TR into IEC-6 cells and selected/maintained as we described previously (12). Expression of p38␥ or p38␥/AGF was induced by addition of 1 g/ml of tetracycline that alone has been shown to have no effects on cell invasion or endogenous p38␥ protein expression (data not shown). The C-terminal truncated p38␥ mutants (p38␥⌬4 and p38␥⌬13) were generated by PCR and cloned into FLAG-tagged pcDNA3 vector as described (22).
Invasion/Migration Assays-Invasion assays were carried out using the BioCoat Matrigel Invasion Chamber (BD Biosciences, Bedford, MA) by using 20% fetal bovine serum as a chemoattractant according to the manufacturer's instruction, as we described previously (12). Invaded cells on the low surface of membrane were then fixed, stained, and counted. On the other hand, wound assays were used to assess the ability of the cell to migrate into a scratched area in serum-free medium.
Immunohistochemistry Studies of Primary Human Colon Cancer Tissues-Immunohistochemistry analyses were conducted in accordance with Institutional Review Board approval from the Medical College of Wisconsin and were performed as described previously (22,23). A rabbit anti-p38␥ (1:1200; R&D catalog no. AF1644) and a goat anti-MMP9 (1:150; Santa Cruz Biotechnology) were used as primary antibodies. Staining results were scored by two observers (22), and a consensus score was assigned to each case, which were then analyzed for the relationship between p38␥ and MMP9 by a double-blind procedure.
Chromatin Immunoprecipitation (ChIP) Assays-The ChIP assay was performed essentially as described previously (16). Briefly, cells were fixed in 1% formaldehyde solution to crosslink DNA with associated proteins, which were then sonicated and incubated with a specific antibody or IgG. For ChIP-re-ChIP assays, the first precipitates were washed and incubated with second antibody as described (24). Precipitated DNA was then extracted and used as a template for PCR. DNA was phenol-chloroform-extracted, ethanol-precipitated, and used as a template for PCR with primers that cover the AP-1 site of the rat MMP-9 promoter within nucleotides Ϫ547 to Ϫ327 (5Ј-ATCCTGCTTCAAAGAGCCTG-3Ј (sense) and 5Ј-GTCT-GAAGGCCCTGAGTGGT-3Ј (antisense). Total chromatin also was prepared in parallel and subjected to PCR as an input control.
To produce virus, lentiviral constructs were transfected into packaging cells, and supernatants were collected and filtered 48 h later. To deplete p38␥ protein expression, human colon cancer cells were double-infected with the viruses at a 2-h interval, which were processed for invasion assays and qRT-PCR/ Western blot (WB) analyses 48 and 72 h later, respectively. For reporter and promoter assays, AP-1 Luc, c-Jun-Luc, or MMP9-Luc was transiently coexpressed with various constructs, and lysates were prepared for the luciferase activity assays 48 h later using a dual luciferase kit from Promega (12). The procedures for immunoprecipitation and WB have been described previously (11).
Statistical Analyses-Results of multiple variables were analyzed by two-way analysis of variance followed by the Bonferroni post-test. Two variables were analyzed by Student's t test. The immunohistochemistry results were analyzed by a 2 test, and the linear regression analysis was used for assessing the relationship between the normalized p38␥ content with the c-Jun or MMP9. A statistically significant difference is reached when a p value is less than 0.05.

p38␥ Requires Phosphorylation and C Terminus to Stimulate
Invasion-To investigate signaling mechanisms for p38␥ oncogenic activity, we focused on how forced p38␥ expression leads to an increased invasion. Because normal cells express little p38␥ (11), a tetracycline-inducible (Tet-on) system was used to express WT p38␥ and its nonphosphorable mutant p38␥/AGF (by changing the dual phosphorylation motif TGY to AGF) in rat intestinal epithelial IEC-6 cells. Following an overnight incubation with and without Tet, cells were seeded in Matrigelcoated invasion chambers and analyzed for invasion (12). Results in Fig. 1, A and B, show that Tet-induced p38␥ stimulates invasion whereas its AGF mutant is without effect, which couples with its activity to increase c-Jun but not c-Fos protein expression. The invasionstimulatory activity of p38␥ is further confirmed by an increased migration in wound assays (Fig. 1C).
These results indicate that p38␥ stimulates cell invasion and/or migration by phosphorylation-dependent mechanisms.
Recent studies indicate that the C-terminal PDZ motif is required for the invasive activity of c-Src (26) and TAZ (tafazzin) proteins (27), and we then explored whether the C-terminal PDZ sequence of p38␥ plays a similar role. To remove the PDZ motif (-ETPL), FLAG-tagged C-terminal truncated p38␥⌬4 and p38␥⌬13 that lack the last four and 13 amino acids, respectively, were generated by PCR and stably expressed in IEC-6 cells by including FLAG-p38␥ and FLAG-p38␥/AGF for comparison. To explore their potential roles in Ras tumorigenesis, resistant clones were pooled and infected with retrovirus expressing K-Ras and their invasive activity then compared. Consistent with the results in Tet-on cells, stable expression of p38␥ significantly increases invasion over the vector control, whereas its AGF mutant has much less effect (Fig. 1, D and E). Interestingly, the invasive activity also was decreased in cells expressing both p38␥⌬4 and p38␥⌬13 (Fig. 1, D and E). These results together indicate that p38␥ requires both phosphorylation and its C terminus to stimulate invasion.
p38␥ Increases MMP9 Transcription by AP-1-dependent Mechanisms-MMPs consist of at least 23 family members and have long been associated with cancer invasion and metastases because of their role in breaking down the extracellular matrix (28). Among these family proteins, MMP9 is one of the best characterized AP-1 target genes involved in cancer invasion (13). Because p38 MAPKs were previously shown to regulate the AP-1 target gene expression (29,30), an AP-1 reporter (4) and a 670-bp human MMP9 promoter (18) were transiently expressed in IEC-6 cells to assess whether p38␥ increases their luciferase activity compared with p38␣. Results in Fig. 2, A and B, show that both p38␥ and p38␣ increase AP-1, but only p38␥ p38␥ Stimulates c-Jun and MMP9 Transcription MAY 14, 2010 • VOLUME 285 • NUMBER 20 stimulates the MMP9 promoter activity. These results are different from those previously published (31), most likely as a result of the different cell line used. To determine whether p38␥ stimulates MMP9 via AP-1, Tet-on cells were expressed with a WT or AP-1 or NF-B site-mutated MMP9 promoter construct (18), and luciferase activity was determined. Fig. 2C results show that Tet-p38␥ has a similar MMP9 stimulatory activity toward the WT and NF-B mutated promoter, which, however, was abolished by the mutations on two AP-1 sites, indicating a required role of the AP-1 in p38␥ trans-activating MMP9.
To demonstrate the MMP9 stimulatory effects further, RNA was prepared and analyzed by real-time qRT-PCR for MMP9 mRNA expression compared with MMP2. Results in Fig. 2D demonstrate that p38␥ significantly increases MMP9 but not MMP2 expression; this is likely as a result of the lack of AP-1 site in the MMP2 promoter. Importantly, analyses of medium collected from cultured Tet-on p38␥ cells revealed that there is an increased secreted MMP9 protein expression by WB and an elevated MMP9 gelatin activity by zymography (Fig. 2E), indicating the transcribed protein being functionally active. Consistent with these results, analysis of IEC-6/K-Ras cells stably transfected with p38␥s also show that p38␥ increases MMP9 RNA expression whereas all of its mutants have much less effects (Fig. 2F), which more or less correlates with their regulatory effects on AP-1 and/ or MMP9 transcriptional activity (supplemental Fig. S1, A and B,  respectively). These results together indicate that p38␥ requires both phosphorylation and the C terminus to stimulate AP-1-dependent MMP9 transcription.
p38␥ Binds the MMP9 Promoter through a Complex Formation with c-Jun-Both human (18) and rat (32) MMP9 promoters contain two AP-1 sites, and ChIP assays were next performed to explore whether p38␥ binds the endogenous MMP9 promoter around this region using primers that span the functional distal AP-1 site (see Fig. 6C). Following formaldehyde-induced DNA cross-linking with associated proteins, stably transfected p38␥ in IEC-6/K-Ras cells was isolated with a FLAG antibody, and the precipitates were subjected to PCR analysis, as described previously (16). As a control, a set of plates were processed for FLAG immunoprecipitation and WB analyses to explore whether p38␥ may be recruited into the MMP9 promoter through interaction with c-Jun and/or c-Fos proteins. Results in Fig. 3A (top) show that among these sublines, only precipitates from WT p38␥-expressed cells contain the MMP9 promoter. Of interest, the immunoprecipitation/WB analyses (Fig. 3A, bottom) revealed that p38␥, but not its mutants, increases c-Jun protein expression that couples with its c-Jun binding activity, whereas c-Fos remains undetectable. Because the MMP9 promoter binding couples with the p38␥ activity to increase c-Jun expression and bind c-Jun protein, these results suggest a scenario in FIGURE 2. p38␥ stimulates MMP9 via AP-1. A-C, p38␥ increases MMP9 transcription. Cells were transiently expressed with the indicated plasmids, and luciferase activities were determined 48 h later (*, p Ͻ 0.05 versus vector or no Tet control where Tet was present for total 72 h in the Tet group, 24 h before the transfection, and 48 h thereafter). D and E, p38␥ activates MMP9. Cells were incubated with and without Tet for 48 h, RNAs were prepared for qRT-PCR, and levels of MMP9 and MMP2 RNAs were normalized to the ␤-actin and expressed as relative over no Tet control (D, *, p Ͻ 0.05 versus no Tet). To assess the MMP9 activity, Tet-on cells were changed to a serum-free medium for the last 24 h, and concentrated medium was assayed for MMP9 activity by zymography and MMP9 protein expression by WB (E, top) in which cell lysates were also analyzed for protein expression (E, bottom). F, stable transfected p38␥ increases MMP9 RNA expression. RNA was prepared and subjected to qRT-PCR, and the ratio of MMP9/␤-actin was expressed as a fold change over the vector alone (*, p Ͻ 0.05 versus vector; **, p Ͻ 0.05 versus p38␥).

p38␥ Stimulates c-Jun and MMP9 Transcription
which p38␥ first activates c-Jun by increasing its expression, and activated c-Jun then recruits p38␥ onto the MMP9 promoter via a complex formation.
To demonstrate the c-Jun-mediated p38␥-MMP9 promoter binding further, Tet-on IEC-6 cells were next analyzed by ChIP and WB in which endogenous c-Jun was also purified and included as a positive control. Results in Fig. 3B show that in response to Tet addition, both p38␥ and c-Jun precipitates contain the MMP9 promoter by ChIP and its partner by WB, and there is an increased c-Jun protein expression, further indicating that p38␥ occupies the MMP9 promoter through interacting with c-Jun. Comparative analyses of c-Jun precipitates by ChIP and WB reveal another interesting phenomenon.
Although c-Jun precipitates contain more c-Jun proteins in the absence of Tet, c-Jun alone, in the absence of Tet-induced p38␥ expression, fails to bind the MMP9 promoter (Fig.  3B, top, lanes 6 and 7 versus bottom,  lanes 1 and 2). In response to Tet, however, c-Jun binds both the p38␥ protein and the MMP9 promoter, indicating a required role of p38␥ in c-Jun binding to the MMP9 promoter. Moreover, the interdependent role of p38␥ and c-Jun in the MMP9 promoter binding was further demonstrated by the ChIP-re-ChIP assay, and Tet-p38␥ only activates MMP9 without increasing another AP-1 target gene vitamin D receptor expression (29) and without stimulating c-Jun and ATF-2 phosphorylation (Figs. 2 and 3, B-D). Our previous studies have demonstrated an increased p38␥ protein expression by a p38␣/p38␤ inhibitor SB203580 (SB) in K-Rastransformed IEC-6 cells as a result of an antagonistic activity of p38␣ against p38␥ (33). We explored next whether endogenous p38␥ binds endogenous c-Jun and the MMP9 promoter in response to the SB treatment. Results in Fig. 3E show that there is a complex formation between p38␥ and c-Jun, which is increased by SB treatment. Importantly, both endogenous c-Jun and p38␥ bind to the MMP9 promoter, and the c-Jun binding activity is increased by the SB-induced p38␥/ c-Jun up-regulation, further supporting the role of p38␥ in c-Jun expression and its MMP9 binding activity. The SB treatment, however, failed to increase the p38␥-MMP9 binding, which may result from a general inhibitory activity of SB on MMP9 expression/ activity (34,35) and its effects on multiple AP-1 family members (36) and/or Raf pathways (37). These results together indicate a two-stage mechanism by which MMP9 is trans-activated by p38␥ MAPK; p38␥ first activates c-Jun by increasing its expression, and activated c-Jun then recruits p38␥ into the MMP9 promoter through a complex formation.
p38␥ MAPK Both Stimulates de Novo c-Jun Synthesis and Acts as a Cofactor for c-Jun-trans-Activating MMP9-Although p38 MAPKs are known to stimulate c-Jun promoter activity (30), so far there have been no reports about the p38induced increase in c-Jun protein expression. c-Jun promoter contains AP-1 and myocyte-enhancing factor 2 sites that are FIGURE 3. p38␥ is recruited into the MMP9 promoter through interaction with c-Jun. A, p38␥ depends on its phosphorylation and C terminus to bind c-Jun protein and MMP9 promoter. Stably transfected p38␥s from IEC-6/K-Ras cells were isolated and precipitates subjected to PCR analyses for their activity in binding MMP9 promoter (top) with a set of cells in parallel analyzed by immunoprecipitation (IP)/Western blotting for p38␥ interacting with c-Jun proteins (bottom). B and C, p38␥ and c-Jun bind the MMP9 promoter through a complex formation. Tet-on p38␥ IEC-6 cells were incubated with and without Tet and p38␥/c-Jun proteins isolated by their specific antibodies, and precipitates subjected to PCR (B, top), Western blotting (B, bottom) or a second IP/PCR for ChIP-re-ChIP (C). D, p38␥ does not induce c-Jun/ATF2 phosphorylation or vitamin D receptor (VDR) expression in IEC-6 cells. Cells were incubated with and without a p38 inhibitor SB203580 (SB) for 24 h and then pulse-treated with a p38 activator arsenite (ARS) for 2 h, and analyzed by Western blotting (see supplemental Fig. S2A for the relationship between c-Jun expression and c-Jun phosphorylation in response to Tet addition). E, endogenous p38␥ forms a complex with c-Jun on the MMP9 promoter in K-Ras-transformed IEC-6 cells. Cells were treated with SB or solvent control (Co) as previously described in these cells (33), and p38␥ and c-Jun proteins were isolated and assessed for their MMP9 promoter binding activity by ChIP and their complex formation by Western blotting.

p38␥ Stimulates c-Jun and MMP9 Transcription
involved in its positive autoregulation by c-Jun/AP-1 (38) and/or its stimulation by p38 MAPKs (30). Experiments were then performed to examine whether p38␥ up-regulates c-Jun transcription. In this regard, a c-Jun luciferase promoter (39) was transiently cotransfected with the indicated constructs in IEC-6 cells, and the luciferase activity was assessed. Results in Fig. 4A show that p38␥, but not its mutants or p38␣, increases c-Jun promoter activity. Moreover, analyses of stable transfected IEC-6/K-Ras cells further showed that only WT p38␥ increases c-Jun RNA expression (Fig. 4B). These results, together with the increased c-Jun protein expression (Fig. 3A), indicate that p38␥ stimulates de novo c-Jun synthesis by phos-phorylation and C terminus-dependent mechanisms as observed for the AP-1/MMP9 trans-activation (supplemental Fig. S1). To examine whether p38␥ may additionally stabilize c-Jun proteins, Tet-p38␥ cells were treated with a protein synthesis inhibitor cycloheximide and analyzed for protein expression. As illustrated in supplemental Fig. S2A, levels of expressed c-Jun proteins are decreased similarly after cycloheximide with and without Tet and become undetectable at 9 h despite the presence of p38␥ protein. These results indicate that p38␥ activates c-Jun primarily by stimulating its expression and not by increasing its protein stability and/or phosphorylation (a slight increase in p-c-Jun at 3 h is likely due to the stress insult from cycloheximide). The critical role of p38␥ in c-Jun and MMP9 expression was further demonstrated in p38␥ϩ/ϩ and p38␥Ϫ/Ϫ MEFs (10) in which p38␥ knockout decreases c-Jun and MMP9 expression that was restored by reexpressing WT p38␥ but not its mutants (Fig. 4, C-E). These results indicate a required role of p38␥ in c-Jun synthesis that may be critical for its MMP9 trans-activating activity.
c-Jun has a well established AP-1 binding activity through dimerization with itself or another transcription factor, but no studies have reported thus far that a MAPK can bind the AP-1 element to regulate gene expression. Because c-Jun depends on p38␥ to bind the MMP9 promoter (Fig. 3B), p38␥ may act as a cofactor for c-Jun in binding the AP-1 element in the regulation of gene transcription. If this is the case, c-Jun activation alone, in the absence of p38␥, should not be active in transcription. To test this possibility, c-Jun was transiently transfected into p38␥ϩ/ϩ and p38␥Ϫ/Ϫ MEFs together with the AP-1 reporter or the MMP9 promoter, and the luciferase activity was analyzed. Results in supplemental Fig. S2, B and C, show that c-Jun only stimulates AP-1/MMP9 transcription in p38␥ϩ/ϩ, but not in p38␥Ϫ/Ϫ cells. Conversely, p38␥ transfection only increases the AP-1/MMP9 in c-Junϩ/ϩ but not in its knock-out counterparts (15) despite similar levels of endogenous p38␥ protein expression in both lines (33) (supplemental Fig. S2, D and E). These results together indicate that p38␥ and c-Jun are both required for AP-1/MMP9 transcription, which thereby further p38␥ Stimulates c-Jun and MMP9 Transcription reinforces our conclusion that p38␥ is an essential cofactor of c-Jun in stimulating MMP9 expression.
p38␥ Controls Endogenous c-Jun and MMP9 Expression in Human Colon Cancer and Stimulates MMP9-dependent Invasion-To assess whether MMP9 activity is required for p38␥-induced invasion, Tet-on cells were incubated with and without MMP9 inhibitor ilomastat and SB-3CT, and their effects on invasion as well as MMP9 activity were assessed. Ilomastat suppresses both MMP2 and MMP9 activity (40), whereas SB-2CT is a specific MMP9 inhibitor (41). Results in Fig. 5, A and B, show that a 24-h incubation of p38␥-expressed IEC-6 cells with both inhibitors significantly reduces the MMP9 activity and almost completely abolishes p38␥-induced invasion. SB-3CT (41) and ilomastat (25) were previously shown to decrease the MMP9 activity, which was speculated to occur via a positive feedback regulation that links MMP9 activity to its transcription (41). Although these compounds may affect additional targets, a coupling of decreased MMP9 activity with reduced invasion by both inhibitors strongly suggests a role of MMP9 in p38␥ invasive activity.
To demonstrate whether endogenous p38␥/c-Jun/MMP9 pathways intrinsically exist in human colon cancer, a group of colon cancer cell lines with known Ras status were examined for their protein expression. Results in Fig. 5C show that p38␥ protein levels were increased in all three cell lines harboring Ras activations/mutations (HCT116, SW480, and LH147T) compared with those without (Caco-2, HT-29, and T48), indicating that activated Ras positively controls endogenous p38␥ protein expression. Of great interest, p38␥ protein expression levels were also significantly correlated with both c-Jun protein and MMP9 RNA expression in these cells (Fig. 5, C and D, and supplemental Fig. S3), suggesting an intrinsic p38␥/c-Jun/ MMP9 pathway in human colon cancer. To demonstrate whether this pathway is functionally active, p38␥ proteins were depleted by shRNA in HCT116 and SW480 cells, and its effects on c-Jun/MMP-9 expression as well as invasion were analyzed. Results in Fig. 5, E-H, show that silencing p38␥ by two separate shRNAs reduces c-Jun and MMP9 expression/activity that couples with a decreased invasion in both lines in which an invasion-suppressive effect was also achieved by the pharmacolog- p38␥ Stimulates c-Jun and MMP9 Transcription MAY 14, 2010 • VOLUME 285 • NUMBER 20 ical MMP9 inhibition. These results together reveal an intrinsic p38␥/c-Jun/MMP9 pathway that is functionally active in stimulating colon cancer invasion.
We previously showed an increased p38␥ expression in primary colon cancer tissues (11,22). To demonstrate further whether hyperexpressed p38␥ couples with an increased MMP9 expression in primary colon cancer tissues, a group of specimens were analyzed by immunohistochemistry for p38␥ as well as MMP9 protein expression. As previously observed (42), the MMP9 immunoreactivity was detected in both normal and colon cancer tissues in contrast to the tumor-associated p38␥ overexpression (brown color, Fig. 6A, top four  panels). Interestingly, a substantial portion of strong MMP9 signals was inside lumens of the malignant but not normal tissues as previously described (42) (Fig. 6A, bottom two panels, indicated by filled and open arrows, respectively). Importantly, a high intensity of such MMP9 signals correlates with an increased p38␥ protein expression in the malignant tissues (Fig. 6B). These results indicate that p38␥ in the primary malignant tissues may also increase MMP9 expression, which may play a critical role in colon cancer progression.

DISCUSSION
MAPKs function through transcription factors to convert transient regulatory signals into a sustained gene expression leading to a biological outcome. Despite extensive research, mechanisms by which a transcription factor translates a MAPK activity into gene expression remain unknown (43). Our studies show that p38␥ activates c-Jun by increasing its expression, and the resulting c-Jun, in turn, acts as a vehicle to recruit p38␥ into the MMP9 promoter, leading to increased MMP9 expression and enhanced invasion. c-Jun alone, in the absence of p38␥, is not active in the MMP9 promoter binding, and both c-Jun and p38␥ are required for MMP9 trans-activation, whereas p38␥ depends on its c-Jun binding activity to bind the MMP9 promoter, to increase MMP9 transcription, and stimulate MMP9-dependent invasion. The active p38␥/c-Jun/MMP9 invasion pathway also was FIGURE 6. Role of the p38␥-MMP9 pathway in human colon cancer. A, p38␥ protein expression is increased in primary human colon cancer tissues, whereas MMP9 is expressed in both benign and malignant tissues, but its luminal expression is only detectable in the malignant tissues. Please note that the brown MMP9 staining signals are detectable in the matrix of normal as well as colon cancer tissues, but significant portions of the MMP9 immunoreactivity are present inside lumens of the malignant (filled arrow) but not normal (open arrow) tissues (bottom two panels, from a separate set of specimens). H&E, hematoxylin and eosin. B, increased p38␥ protein expression significantly correlates with higher levels of MMP9 expression/secretion in primary colon cancer tissues. p38␥ signals in a group of invasive colon adenocarcinomas were subtracted from those of the matched normal tissues, which were compared with MMP9 intensity within the malignant lumens from the same group by the 2 test. C, an experimental model shows that p38␥ both acts as a c-Jun activator and a c-Jun cofactor in trans-activating MMP9 to stimulate colon cancer invasion. p38␥ was shown to depend on its phosphorylation and C terminus to activate c-Jun by increasing its synthesis. Activated c-Jun then recruits p38␥ as a cofactor into the AP-1 site of the MMP9 promoter through a complex formation, thereby stimulating MMP9 transcription and colon cancer invasion. Our model suggests that c-Jun alone, in the absence of p38␥, is not active in the MMP9 promoter binding, and therefore, p38␥ acts as a critical activator and an essential cofactor for c-Jun trans-activating MMP9.

p38␥ Stimulates c-Jun and MMP9 Transcription
demonstrated in Ras-activated human colon cancer, whereas hyperexpressed p38␥ was shown to couple with an increased luminal MMP9 expression in primary cancer tissues. These results together reveal a novel mechanism by which p38␥ MAPK acts both as a c-Jun activator and a c-Jun cofactor in the stimulation of MMP9 transcription leading to increased invasion (Fig. 6C). Because Ras-activated colon cancers are more metastatic (44) and MMP9 is a target antimetastatic therapy (14), p38␥ may promote colon cancer progression through transduction of Ras signaling to MMP9 via the c-Jun-mediated promoter binding.
Previous studies showed that Hog1 in yeast (corresponding to mammalian p38␣) is recruited to both promoter and coding regions of osmotic stress genes in stress response (45,46). In mammalian cells, activated p38␣ similarly was shown to interact with chromatin during cell differentiation (47). In all of these studies, however, mechanisms for p38␥-DNA bindings as well as biological consequences remain unknown. Here, we show that the p38␥-MMP9 promoter binding first requires p38␥-induced c-Jun expression and then its interaction with c-Jun protein. The signaling specificity of this regulation is suggested by the fact that MMP9 is stimulated by p38␥ but not by p38␣, which is mediated by c-Jun (but not c-Fos or ATF2) via the AP-1 (but not NF-B) site without significant effects on MMP2 or another AP-1 target gene vitamin D receptor expression. Although MMP9 may be one of the most documented AP-1 targets in cancer invasion and metastasis, it remains to be determined whether additional AP-1 target genes are involved in the p38␥ invasive phenotype through the promoter binding.
It is important that p38␥ was shown to activate c-Jun by increasing its expression, as up-regulated c-Jun protein expression has been observed in primary human colon cancer tissues (48) where it may be involved in colon cancer invasion (49). Because c-Jun is positively autoregulated by c-Jun/AP-1 (38) and p38␥ increases c-Jun/AP-1/MMP9 transcription through its c-Jun binding activity, the same c-Jun-mediated promoter binding may likely operate in p38␥ stimulating c-Jun as well as MMP9 transcription. With regard to the c-Jun-p38␥ binding, there may be two mechanisms involved. c-Jun lacks a PDZ domain that is required for a direct interaction with a PDZ motif-containing protein such as p38␥. Therefore, the requirement of p38␥ C terminus for its c-Jun binding suggests that both proteins may interact indirectly inside cells through additional PDZ proteins. On the other hand, phosphorylated p38␥ is known to be localized predominantly in the nucleus (33), which may be critical for its interaction with nuclear c-Jun as a cofactor. Although these different scenarios require further study, the required role of the c-Jun binding activity in p38␥ stimulating c-Jun synthesis, MMP9 transcription, and invasion highlights an essential role of this protein-complex in triggering the invasion cascade. c-Jun was previously shown to bind the MMP9 promoter in cardiac (50) but not in neuronal cells (51), and similarly we showed that c-Jun only binds the MMP9 in the presence of p38␥, indicating a determinant role of c-Jun cofactor abundances in its trans-activation of MMP9. Our results suggest that p38␥ may be one of these critical cofactors to determine c-Jun transcriptional activity by increasing its expression and facilitating its target gene promoter binding.