Essential Role of p38γ in K-Ras Transformation Independent of Phosphorylation*

MAPK cascades play the critical role in regulating Ras oncogene activity by phosphorylation-dependent mechanisms. Whereas the ERK MAPK pathway is required for Ras transformation, our previous works established that the p38 activity is inhibitory to Ras signaling in both experimental and ras-mutated cancer cells (Chen, G., Hitomi, M., Han, J., and Stacey, D. W. (2000) J. Biol. Chem. 275, 38973-38980; Qi, X., Tang, J., Pramanik, R., Schultz, R. M., Shirasawa, S., Sasazuki, T., Han, J., and Chen, G. (2004) J. Biol. Chem., 279, 22138-22144). Here we report that K-Ras activated p38γ, a p38 MAPK family member, by inducing its expression without increasing its phosphorylation and that depletion of induced p38γ suppressed Ras transformation in rat intestinal epithelial cells. This p38γ activity contrasts with that of its family member, p38α, which is activated by Ras through phosphorylation, leading to an inhibition of Ras transformation. Mechanistic analyses showed that unphosphorylated p38γ may promote Ras transformation through an increased complex formation with ERK proteins. Significantly, functional p38γ protein was expressed only in K-ras-mutated human colon cancer cells, and p38γ transcripts were ubiquitously increased in a set of primary human colon cancer tissues. These studies thus demonstrate the essential role of p38γ in K-Ras transformation independent of phosphorylation, and elevated p38γ may serve as a novel diagnostic marker and therapeutic target for human colon cancer.

p38 mitogen-activated protein kinase (MAPK) 1 was first identified in studies of endotoxin-induced cytokine expression (1,2). So far, four p38 isoforms have been cloned and characterized, including p38␣, p38␤, p38␥, and p38␦ (3). The p38 upstream activators include MAPK kinase (MKK) 3 and MKK6 (4). Its downstream effectors consist of kinases such as MAPKactivating protein kinase 2 and p38-related/activated protein kinase and transcription factors including activating transcrip-tion factor-2 and myocyte enhancement factor 2 (3). In addition to these effectors, p38 can also signal through cross-talk with c-Jun NH 2 -terminal kinase (JNK) (5,6) and extracellular signal-regulated kinase (ERK) pathways (7,8). p38 MAPK is mostly responsive to cytokines and inflammatory stress and plays an important role in regulating inflammation and immunoresponse (3,9). p38 activation, however, also triggers other biological effects, such as cell death, differentiation, and proliferation, by a cell type-specific mechanism (3,10). To date, biological functions of p38 MAPK have mostly been demonstrated by analyzing p38␣, and studies of other p38 isoform proteins will contribute to understanding pleiotropic activities of p38 activation.
All MAPKs are activated by dual phosphorylation on threonine and tyrosine residues within a Thr-Xaa-Tyr motif without significant alterations of their expression (11)(12)(13). p38␥ MAPK (also called SAPK3 or ERK6), which shares about 60% identity with p38␣ and p38␤ (14,15), has several unique properties. First, in contrast to the ubiquitously expressed p38␣, p38␥ mRNA is only detectable in normal skeletal muscle (14 -16). Recent studies, however, demonstrate that p38␥ protein is highly expressed in several human malignant cell lines (17)(18)(19)(20)(21)(22), indicating its possible role in tumorigenesis. Secondly, expression levels of p38␥ mRNA and/or protein are increased by a differentiation-associated process, an effect distinct from all other MAPKs (14,16,23). Furthermore, elevation of p38␥ concentration by transfection, in the absence of upstream activators, induces C2C12 cell differentiation (14). These results together suggest that in addition to phosphorylation-dependent kinase activity (17,20), an increased expression of p38␥ may regulate life-important biological processes such as malignant transformation and cell differentiation.
The Ras family of proteins consists of three isoforms (H-, K-, and N-Ras) that play a critical role in controlling normal and malignant cell growth (24). K-ras mutation is one of the most common abnormal genetic events in human cancer, with the highest incidence in pancreatic carcinomas (90%) and colorectal tumors (50%) (25,26). MAPKs (ERK, JNK, and p38) are the best-characterized signal pathways in transduction and regulation of Ras activity (12,27). ERK/MAPK activation has been shown to be both necessary and sufficient in transforming experimental NIH 3T3 cells (28,29), whereas the JNK pathway is also required for Ras transformation (30,31). On the contrary, activation of p38 MAPK is antagonistic to Ras activity, including inhibition of Ras-induced proliferation in NIH 3T3 cells (6), suppression of Ras transformation in RIE cells (30), and induction of K-Ras-dependent cell death in human colon cancer cells (32). In this study, we sought to test the hypothesis that endogenous p38 family members may regulate Ras activity by an isoform-specific mechanism. Our results showed that in contrast to p38␣, K-Ras activates p38␥ by inducing its expression without increasing its phosphorylation, and induced p38␥ is required for K-Ras transformation by a mechanism possibly involving a complex formation with ERK proteins.

MATERIALS AND METHODS
Reagents, Cell Lines, and cDNA Constructs-Cell culture materials were supplied by Invitrogen and chemicals purchased from Sigma. Fetal bovine serum was obtained from BioWhittaker. Protein Sepharose G and protein A-Sepharose 4B beads were purchased from Zymed Laboratories Inc.. p38-specific antibodies were described previously (17,18). Their specificity was further confirmed by a rabbit polyclonal antibody from Clontech. ERK1 and ERK2 antibodies were from Santa Cruz. p-p38 and p-ERK antibodies were from Cell Signaling. Rat intestinal epithelial IEC-6, mouse fibroblast NIH 3T3, human colon cancer HCT116, and HT-29 cells were purchased from American Type Culture Collection and maintained in modified Eagle's medium containing 10% fetal bovine serum and antibiotics at 37°C, 5% CO 2 . FLAG-tagged p38 isoform cDNAs in pcDNA3 vector and their dominant negative AGF counterparts were previously described (17,18). The adenovirus vector containing HA-tagged constitutively active MKK6 was used as previously described (17,32). HA-tagged K-Ras and H-Ras cDNAs (both with G12V mutation) were provided by Guthrie cDNA Resource Center and subcloned into retroviral vector LZRS (33). Retroviral vector pLHCX was used to express the wild-type and non-phosphorylated p38␥ cDNA as previously described (34). After transfection into Phoenix-Ampo retrovirus packaging cells (American Type Culture Collection), supernatants were used to infect IEC-6 and NIH 3T3 cells. To establish stable Ras-transformed cell lines, transduced cells were selected with puromycin for about 2 weeks.
Assays for Ras Transformation and Cell Proliferation-Morphological transformation following retroviral infection was examined under the phase-contrast microscope. For anchorage-independent growth, IEC6/K-Ras cells were either treated with different inhibitors for 24 h or infected with adenovirus (vector and Ad-MKK6) for 5 h or pSUPER siRNA overnight. 2 ϫ 10 4 cells were plated in growth medium containing 0.33% Sea-plaque-agarose. Formation of multicellular colonies was visualized and quantitated about 2 weeks later (30). The colonies formed on an entire 60-mm plate were photographed and manually counted, and the number of colonies/field shown came from two to three plates of one experiment and were analyzed for statistical significance using Student's t test. Similar results were obtained from at least two additional experiments. To analyze effects of p38␥ overexpression on cell proliferation, normal IEC-6 cells were infected with the retrovirus pLHCX, pLHCX-p38␥, or pLHCX-p38␥/AGF. The protein expression was assessed by Western blot 48 h, and [ 3 H]thymidine incorporation for DNA synthesis was determined 72 h later, as previously described (35). To assess effects of depleting endogenous p38␥ protein on human colon cancer cell proliferation, HCT116 cells were infected with pSR or pSR-siRNA for 72 h. Cells were then pulse-incubated with [ 3 H]thymidine, and DNA synthesis was measured (35).
Tansfection, Immunoprecipitation, Immunostaining, and Western Blot-Cells were transfected with FLAG-tagged p38 isoforms, with and without K-Ras, and collected in modified radioimmune precipitation assay buffer (50 mM Tris, pH 7.5, 150 mM NaCl, 1% Nonidet P-40, 0.25% sodium deoxycholate, 1 mM EGTA, 10 mM NaF, 1 mM Na 3 VO 4 , 1 mM phenylmethylsulfonyl fluoride, and 1 g/ml aprotinin, leupeptin, and pepstatin). The FLAG precipitates were analyzed for p38 isoform protein expression using a FLAG antibody, for their phosphorylation using a specific p-p38 antibody, and for bound endogenous ERK/p-ERK proteins by Western blot. The endogenous phosphorylated p38 proteins were isolated with a mouse monoclonal p-p38-specific antibody, and precipitates were analyzed for the presence of p38␣ and p38␥ using specific antibodies (17,18). To isolate endogenous p38␥ and p38␣ proteins, lysates were incubated with respective specific antibodies, and precipitates were examined for the presence of ERK/p-ERK protein by Western blot. To detect cellular distributions of ERK and p38␥, normal IEC-6 cells were transfected with HA-ERK1 in the presence and absence of FLAG-p38␥/AGF. Thereafter, cells were fixed and co-stained with mouse anti-FLAG antibody plus anti-mouse IgG Cy3 to detect transfected FLAG-p38␥/AGF and with anti-HA-IgG-fluorescein isothiocyanate conjugate to detect transfected HA-ERK1, as previously described (36). For Western blot analyses, cells were directly lysed in 1ϫ loading buffer and separated on SDS-PAGE. All of the following procedures were the same as those described previously (18,35).
Experiments with siRNA to Inhibit p38␥ Expression-To deplete endogenous p38␥ protein, retroviral vector pSR (VEC-pRT-0002; Oligo-Engine) was used as previously described (37). The target sequence (5Ј-AAGGAGATCATGAAGGTGACG-3Ј) was cloned into the pSR vector, which was transfected into the packaging cells to produce the virus-containing supernatants for infection, as we previously described (32). Among four sequences analyzed, this siRNA yielded the most substantial effect on p38␥ depletion in several cell types and was consequently used for all analyses. Typically, cells were analyzed for p38␥ protein depletion 72 h after infection.
Northern Blot and Gene Expression Array-For Northern blot, total RNA was prepared by using the TRIzol kit. Human p38␥ cDNA was used to generate a 520-bp fragment by PCR (primers: forward, 5Ј-GGCTTTTACCGCCAGGAG-3Ј; reverse, 5Ј-GTCATCTCACTGTCTGC-CTGCCT-3Ј). The probe was labeled with [ 32 P]dCTP using the high Prime Kit and purified with Quick Spin Columns (35). A matched tumor/normal expression array kit was purchased from Clontech (catalog number 7840-1). The membrane containing the matched cDNA samples was incubated with the [ 32 P]dCTP-labeled p38␥ probe according to the manufacturer's instructions. The specific radioactivity was measured with a PhosphorImager (Amersham Biosciences). Results were measured with Scion Image software and normalized by ubiquitin.

K-Ras Induces p38␥ Protein Expression without Increasing
Its Phosphorylation in IEC-6 Epithelial Cells-To analyze signaling interactions between Ras and the p38 family, HA-tagged activated K-Ras and H-Ras cDNAs were subcloned into retroviral vector LZRS (33). After transfection into Phoenix packaging cells, supernatants were used to infect rat intestinal epithelial IEC-6 and mouse fibroblast NIH 3T3 cells, followed by selection with puromycin. We used Western blot to examine these transfected cells for p38 family protein expression and phosphorylation. The specificity of p38 isoform-specific antibodies (17) has been established in our previous publication (18). Results in Fig. 1A show that the predominant form of p38 in vector-transfected IEC-6 cells is p38␣, with p38␥ barely detectable, whereas p38␣, p38␤, and p38␥ (but not p38␦; data not shown) proteins are expressed in NIH 3T3 cells. Consistent with our previous finding (6), increased p38 phosphorylation was observed in both cell lines transfected with either K-Ras or H-Ras oncogene, together with increased phosphorylated ERK proteins. Levels of total p38␣ and ERK1/2 proteins, however, remained relatively consistent with and without Ras transfections ( Fig. 1A; data not shown). Surprisingly, p38␥ protein expression was specifically induced by K-Ras in IEC-6 cells but not in NIH 3T3 cells, suggesting its potential role in K-Ras tumorigenesis in epithelial cells. Furthermore, levels of both p38␥ protein (Fig. 1A) and p38␥ RNA (Fig. 1B) were increased in K-Ras-as well as H-Ras-transfected IEC-6 cells, indicating a Ras isoform-independent p38␥ induction.
The p-p38 antibody used in the previous analysis reacts with all p38 family members dual-phosphorylated at the Thr and Tyr residues. A single phosphorylated p38 band around 39 kDa in IEC-6 cells (Fig. 1A) suggested that this p-protein may be p38␣ because only p38␣ and p38␥ proteins are expressed in these cells, and p38␥ (about 45 kDa) migrates more slowly than p-p38. To further confirm this speculation, an equal amount of lysates from the vector-and K-Ras-transfected IEC-6 cells was incubated with a mouse p-p38 antibody, and the precipitates were examined for their reactivity with a rabbit antibody against p38␣, p38␥, and p-p38 by Western blot. Results in Fig.  1C show that the recovery of p38␣ was greater in K-Rastransfected cells than that in mock-transfected cells, in which no change in p38␥ was observed. These results thus demonstrate that K-Ras selectively induces p38␣, not p38␥, phosphorylation in IEC-6 cells.
So far, almost all studies of MAPK signaling have focused on regulation of MAPK activity by phosphorylation (12, 38). To understand mechanisms by which Ras-induced p38␥ protein becomes unphosphorylated, a transient transfection experiment was performed. Normal IEC-6 cells in this case were transfected with FLAG-tagged p38 isoforms, with and without the K-Ras-expressing plasmid. The anti-FLAG precipitates were analyzed by Western blot for phosphorylation of transfected p38 proteins using a p-specific p38 antibody (Fig. 1D). Comparison of the FLAG-p38 (top panel) to the p-p38 band (bottom panel) shows that phosphorylated p38␣ and p38␤ signals were increased 1.7-and 3.2-fold by K-Ras, respectively. Surprisingly, K-Ras did not increase p38␥ protein phosphorylation, but instead decreased its level by 60% (Fig. 1D, bottom  panel). These experiments clearly demonstrated that K-Ras selectively phosphorylates p38␣ and p38␤ but dephosphorylates p38␥ in these epithelial cells. Inhibition of p38␥ phosphorylation by Ras may be due to limited amounts of endogenous upstream kinases to phosphorylate p38␥ upon Ras transfection and/or Ras activation of p38␥-specific phosphatases, leading to its dephosphorylation. Stimulation of p38␣ phosphorylation and inhibition of p38␥ phosphorylation by transient K-Ras expression provide an explanation for increased p38␣ phosphorylation and elevated unphosphorylated p38␥ protein expression in stably transfected IEC6/K-Ras cells.
p38␥ Is Selectively Induced by Ras Oncogene, and Its Overexpression Does Not Lead to Cell Proliferation or Malignant Transformation-That Ras induces p38␥ protein expression without increasing its phosphorylation is a novel observation. To explore whether this induction is specific to Ras, normal IEC-6 cells were treated with mitogens (serum and TPA) and stresses (arsenite and tumor necrosis factor-␣) and examined for p38␥ protein expression by Western blotting. As shown in Fig. 2A, p38␥ protein level was not increased by these stimuli, although ERK and/or p38/JNK phosphorylation was strongly increased under the same conditions (there was also no substantial p38␥ protein increase 24 h later; data not shown). Interestingly, transient Ras infection led to a substantial p38␥ protein elevation (Fig. 2B). Thus, p38␥ protein expression is selectively induced by Ras oncogene and not by other mitogens, pointing to its potential role in Ras malignant transformation.
Activated Ras induces cell transformation through its constitutively proliferative signaling. To explore whether p38␥ alone is mitogenic or oncogenic by a phosphorylation-dependent mechanism, p38␥ and its non-phosphorylated mutant p38␥/ AGF were overexpressed in normal IEC-6 cells by retroviral vector pLHCX (34), and their effects on cell proliferation were assessed by thymidine incorporation. Results in Fig. 2, C and D, show that higher levels of p38␥ proteins have no significant effects on DNA synthesis (thymidine incorporation). Furthermore, p38␥ overexpression did not lead to soft agar growth (data not shown). Consistent with this observation, both p38␥ and p38␥/AGF did not increase ERK phosphorylation, which is strongly induced by TPA and serum (Fig. 2, A and C). These studies thus reveal that p38␥ per se is not mitogenic or oncogenic.
Regulations of K-Ras Transformation by PD and SB Suggest the Required Role of p38␥ in Ras Transformation-Previous studies by ourselves (6,40) and others (30,39) have established the required role of ERK phosphorylation and the inhibitory role of p38␣ (also called p38) phosphorylation in Ras proliferative and transforming activity. Because Ras stimulates ERK/ p38␣ phosphorylation and induces p38␥ expression, we sought  1. K-Ras selectively induces unphosphorylated p38␥ protein expression in epithelial cells. A, K-Ras selectively induces p38␥ protein expression in IEC-6 cells. The vector and Ras stably transfected IEC-6 and NIH 3T3 cells were analyzed by Western blot for protein expression and phosphorylation. Similar results were obtained from additional two experiments. B, Ras oncogene induces p38␥ RNA expression in IEC-6 cells. C, the phosphorylated p38 in IEC6/K-Ras cells is p38␣ and not p38␥. Cell lysates were immunoprecipitated with a mouse p-p38 antibody, and the precipitates were examined for the presence of p38␣ and p38␥ proteins by Western blot. D, transient K-Ras expression phosphorylates p38␣/␤ but dephosphorylates p38␥. IEC-6 cells were transiently transfected with FLAG-tagged p38 isoforms with and without K-Ras, and the FLAG precipitates were examined for p38 expression and phosphorylation. The p-p38 ratio was calculated by dividing each p-p38 band of the K-Ras-transfected group by that of the corresponding vector group after normalization to the respective FLAG-p38 band.
to explore whether p38␥ is involved in their Ras regulatory activities.
To explore potential roles of p38␥ in these regulations, stable K-Ras-transformed IEC-6 cells (IEC6/K-Ras) were incubated with a specific ERK inhibitor, PD, or a p38␣/␤ inhibitor, SB, and their effects on Ras transformation and p38␥ protein expression were examined. PD treatment almost completely reversed the transformed morphology as compared with normal IEC-6 cells, whereas SB increased the transformation (cells became more refractile) (Fig. 3A). The morphological alterations were further confirmed by an increased soft agar formation by SB and a decreased anchorage-independent growth of IEC6/K-Ras cells by PD (Fig. 3B). Of great interest, inhibition of ERK phosphorylation by PD was coupled with a substantial decrease in p38␥ protein expression (Fig. 3C). These results suggest that the ERK kinase activity, as least as demonstrated with PD, may promote Ras transformation through increasing p38␥ expression. In the case of SB, on the other hand, it increases ERK phosphorylation and stimulates p38␥ expression (Fig. 3C). In addition to suppressing p38␣/␤ activity, SB can also activate c-Raf in cellular systems (41), which may contribute to the increased ERK phosphorylation. Because PD decreases p38␥ protein expression as well as Ras transformation, whereas SB has the opposite effect, results from these correlative analyses support the notion that p38␥ may be required for Ras malignant transformation.

Depletion of p38␥ Protein Demonstrates Its Essential Role in K-Ras Transformation in IEC-6 Cells and in K-Ras-dependent Proliferation in Human Colon
Cancer Cells-To directly prove requirements of p38␥ protein in Ras activity, endogenous p38␥ protein was depleted from IEC6/K-Ras cells, and the effects on Ras transformation were next determined. To silence p38␥ protein expression, a pSR siRNA retroviral vector was used (37). As shown in Fig. 4A, the p38␥ protein decreased by about 80% in comparison to the vector control 72 h after viral infection. The siRNA-mediated p38␥ depletion was specific because it had no effect on p38␣ or ERK protein expression. More importantly, p38␥ depletion led to a reversion of the morphological transformation, which was reflected by a reduction in cell density and a loss of spindly appearance (Fig. 4B). To further confirm the morphological reversion, cells were infected and plated on soft agar for anchorage-independent growth. Results in Fig. 4C reveal that the colony-forming activity of IEC6/K-Ras cells was substantially inhibited by p38␥ protein depletion, which correlates with a decreased DNA synthesis (data not shown). These results therefore directly demonstrate the essential role of p38␥ in K-Ras transformation.
Human colon cancer cells were further utilized to explore the roles of p38␥ in natural K-ras mutation-induced tumorigenesis. Consistent with our results in rat IEC-6 cells, there are higher levels of p38␥ protein (Fig. 4D) and RNA (Fig. 4E) in K-rasmutated HCT116 cells than in wild-type K-ras-containing HT-29 human colon cancer cells. Furthermore, depletion of p38␥ protein in HCT116 cells significantly inhibited DNA synthesis (Fig. 4, F and G), indicating that p38␥ protein is not only expressed in K-ras-mutated human colon cancer cells but also required for K-Ras-dependent malignant proliferation. Because the K-ras gene disruption inhibits HCT116 tumor growth in vitro and in vivo (42), these results suggest that p38␥ may be required for K-ras-dependent malignant proliferation in human colon cancers.
p38␥ May Promote Ras Transformation through Its Complex Formation with ERK Proteins-Previous studies have shown a physical interaction between p38␦ and ERK proteins, which may play a role in regulating keratinocyte differentiation (43,44). Because ERK activity is essential for Ras transformation

FIG. 4. Depletion of p38␥ protein suppresses K-Ras-induced transformation in IEC-6 cells and inhibits K-Ras-dependent proliferation in human colon cancer cells. A, depletion of p38␥ protein expression by siRNA retroviral infection in IEC6/K-Ras cells.
Cells were infected with retroviral vector pSR or pSR-ip38␥ (siRNA) and analyzed for protein expression/phosphorylation by Western blot. Results shown are representative of three separate experiments. B and C, p38␥ protein depletion reverses the transformed morphology (B) and inhibits IEC6/K-Ras cell growth on soft agar (C). Cells were infected overnight and plated for soft agar colony formation (C). For morphological observation, the picture was taken 72 h after infection (B). Results shown in C are the mean colony number/field (ϮS.D., n ϭ 30 from six plates in three separate experiments, p Ͻ 0.01). D and E, levels of p38␥ protein (D) and p38␥ mRNA (E) are higher in K-ras-mutated HCT116 cells than normal K-ras-containing HT29 human colon cancer cells. F and G, p38␥ depletion inhibits DNA synthesis in HCT116 cells. HCT116 cells were infected with pSR or pSR-ip38␥, and protein expression was examined 72 h later by Western blot (F). DNA synthesis was measured by thymidine incorporation (G). Results are the means of triplicate infections (ϮS.D., p Ͻ 0.01). (Fig. 3), p38 may interact with ERK proteins and thereby regulate Ras transforming activity. To this end, endogenous p38␥ and p38␣ from IEC6/K-Ras cells were isolated with respective antibodies, and the precipitates were examined for the presence of ERK proteins. Results in Fig. 5A show that both p38␥ and p38␣ have the ability to bind ERK proteins. To demonstrate whether these bindings are involved in regulation of Ras transformation, IEC-6/K-Ras cells were transiently overexpressed with MKK6, a p38 activator, using an adenovirus-mediated gene delivery (35,36). Consistent with previous reports (6,30), MKK6 overexpression completely abolishes the soft agar growth of the Ras-transformed cells (Fig. 5, D and E). Of great interest, the ERK and/or p-ERK protein in the p38␥ complex was significantly diminished, whereas it remained unchanged (p-ERK undetected; data not shown) in the p38␣ precipitates in response to MKK6 infection (Fig. 5, B and C). Because MKK6 induces p38␣ but not p38␥ phosphorylation and inhibits Ras transformation (Fig. 5, DϪF; data not shown), these results further established an inhibitory role of phosphorylated p38␣. Moreover, MKK6 overexpression does not alter levels of ERK/p-ERK and p38␥ proteins but decreases the p38␥bound ERK/p-ERK protein (Fig. 5, B and F), pointing to a critical role of ERK-p38␥ complex formation in Ras transformation. Because the positive correlation has been established between p38␥ protein levels and Ras transforming activity (as demonstrated with PD, SB, and siRNA), cellular p38␥ protein may promote Ras transformation through its ERK/p-ERK binding activity. Thus, either depleting p38␥ protein or disrupting p38␥-ERK binding can inhibit Ras transforming activity. Cellular co-localization is a strong indication for proteinprotein interaction. ERK proteins are known to be predominantly cytoplasmic in unstimulated cells, which are translocated into the nucleus following activation (45,46). p38␥, on the other hand, has been shown to be both cytoplasmic and nuclear in PC12 cells (47). To explore whether p38␥ is localized similarly to ERK protein in IEC-6 cells, HA-tagged ERK1 and FLAG-tagged p38␥/AGF were co-transfected, and their expres-sion was detected by fluorescence microscopy as previously described (36). p38␥/AGF overexpression was used to mimic higher levels of induced non-phosphorylated p38␥ proteins in IEC6/K-Ras cells. Results in Fig. 6 showed that transfected ERK proteins are mostly in the cytoplasm (top left panel). Although the signal is still strong in cytoplasm after p38␥ co-expression, a substantial portion became nuclear and was located around the nuclear membrane (Fig. 6, top right panel). Moreover, transfected p38␥/AGF and HA-ERK1 exhibited a similar distribution pattern. These results thus further confirm a physical interaction between p38␥ and ERK proteins and indicate that p38␥ may act as a Ras effector through increasing nuclear ERK accumulation.
Our results in Fig. 1 show that in IEC6/K-Ras cells, endogenous p38␣ is phosphorylated, whereas p38␥ remains in an unphosphorylated form. If the p38␥-ERK complex plays a role in Ras transformation, the ERK binding activity of p38␥ should be increased when it becomes unphosphorylated. To explore this possibility, normal IEC-6 cells were transiently transfected with FLAG-tagged wild-type p38␥ and its non-phosphorylated AGF mutant in the absence and presence of K-Ras by including p38␣ for comparison. Anti-FLAG precipitates were examined for the presence of endogenous ERK proteins. Results in Fig. 7 show that transfected p38␥ and p38␣ bind to endogenous ERK proteins independent of their phosphorylation status and independent of K-Ras transfection. Strikingly, the p38␥/AGF binds much more ERK/p-ERK protein following Ras transfection (Fig. 7, lane 4, left panel), an event not observed with wild-type p38␥ transfection. This effect is opposite to p38␣ because ERK proteins are only phosphorylated in the wild-type p38␣ complex and in the absence of K-Ras, indicating a possibility that p-ERK protein may be relocated from the p38␣ to the unphosphorylated p38␥ complex in response to Ras activation. These results together suggest a scenario in which p38␥ expression is induced, whereas its phosphorylation is concomitantly inhibited by Ras oncogene, and induced unphosphorylated p38␥ may promote Ras transformation through ERK binding, leading to increased/sustained ERK phosphorylation.
p38␥ Transcripts Are Elevated in Primary Human Colon Cancer Tissues-To further explore the role of p38␥ in human cancer development, we used a matched tumor/normal expression array to examine p38␥ gene expression. In this array, cDNA samples representing 11 different tissues of 68 cancer patients were immobilized to a nylon membrane. From each patient, a pair of cDNA samples was derived from the tumor FIG. 5. Roles of p38␥-ERK complex formation in K-Ras transformation. A, both p38␥ and p38␣ bind to ERK proteins. Lysates of IEC6/K-Ras cells were incubated with p38␥ and p38␣ antibody, and ERK protein in the precipitates was examined by Western blot. B and C, the ERK and/or p-ERK protein in p38␥ but not p38␣ complex is diminished by MKK6. Cells were infected with Ad-Vect or Ad-MKK6 for 48 h, and an equal amount of lysates was immunoprecipitated with a p38␥ or p38␣ antibody, followed by Western blot for the presence of ERK/p-ERK protein in the precipitates. Similar results were obtained from one additional experiment. D and E, inhibition of soft agar growth of IEC6/K-Ras cells by MKK6. Cells were infected and plated in growth medium containing Sea-plaque-agarose for anchorage-independent growth. Pictures shown in D were taken about 2 weeks later. The numbers in E are the mean of colony numbers from 15 different fields of two separate experiments, and there was no single colony formation in MKK6 group. F, infection with Ad-MKK6 phosphorylates p38␣ without affecting p38␥ protein expression. Cells were infected with Ad-Vect or Ad-MKK6 and analyzed for protein expression/phosphorylation by Western blot. Results are representative from two separate experiments.

FIG. 6. Localization of ERK and p38␥ proteins in IEC-6 cells.
Normal IEC-6 cells were transfected with HA-ERK1 in the presence and absence of FLAG-p38␥/AGF. Cells were fixed and co-immunostained for transfected ERK1 using anti-HA antibody and transfected p38␥ using anti-FLAG antibody as described under "Materials and Methods." and corresponding normal tissue. After hybridization with a radiolabeled p38␥ probe, the membrane was scanned with a PhosphorImager for specific binding. As shown in Fig. 8A, p38␥ transcript was detected in every tissue examined. Most strikingly, 100% of colon cancer patients (total, 11) showed higher levels of p38␥ signal in tumor tissues than in matched normal tissues (Fig. 8A, underlined). Furthermore, levels of p38␥ mRNA were also higher in most breast cancer tissues (8 of 9 specimens), a result similar to that previously reported about the hyper-expressed ERK MAPK in human breast cancer tissues (48). Expression levels of a housekeeping gene, ubiquitin, however, were similar between normal and tumor tissues (Fig.  8B). The p38␥ fold increase in human colon cancers varied from 1.3 to 4.0 after normalization to ubiquitin, with an average of 2.21 Ϯ 0.79-fold (p Ͻ 0.01) (Fig. 8C). Because K-ras mutation is frequently observed in human colon cancers (25), and p38␥ expression is increased in K-ras-mutated HCT116 cells (Fig. 4), these results strongly suggest that elevated p38␥ is likely to play a role in K-Ras-induced human colon tumorigenesis. DISCUSSION Demonstrating the essential role of p38␥ in K-Ras transformation independent of phosphorylation will greatly impact our understanding of MAPK signaling. These results suggest that in addition to phosphorylation-dependent effects, stress p38␥ MAPK can execute phosphorylation-independent functions as a Ras effector. A phosphorylation-dependent activity of p38␥ may be primarily regulating stress response through transient activation and inactivation (17,20). In response to Ras oncogene, however, p38␥ expression is induced, whereas its phosphorylation is inhibited, and elevated p38␥ proteins consequently play an essential role in maintaining Ras-transformed phenotype (Fig. 9). Although increased p38␥ RNA in K-Rastransformed and -mutated cells suggests a trans-activation mechanism, we cannot rule out the involvement of Ras-induced p38␥ dephosphorylation in elevating its protein concentration. The phosphorylation-independent activity of p38␥ is consistent with our previous observation that overexpression of both wildtype and non-phosphorylatable p38␥ showed a similar regulatory effect on gene expression (18). Thus, p38␥ MAPK has dual activity: it serves as a kinase to regulate stress response by a phosphorylation-dependent mechanism and acts as a Ras effector to promote Ras transformation through increased expression without phosphorylation.
Previous studies have shown that p38 family members can either collaborate or oppose each other in regulating gene expression in response to various signaling. Hypoxia, for example, induces p38␣/␥ phosphorylation in PC12 cells, and both p38␣ and p38␥ inhibit cyclin D1 expression (20). Our previous work further showed that MKK6 and arsenite stimulate phosphorylations of all four p38 family members in human breast cancer cells, in which p38␤ increases AP-1-dependent gene expression, but p38␥/␦ inhibits it or has no effect on it (18). A similar opposing effect of p38 isoforms was also demonstrated in stress regulating hemo-oxygenase-1 expression (49). All of these analyses, however, were carried out by p38 overexpressions, and physiological relevance of these observations has not consequently been established. Our present studies, on the other hand, demonstrate that K-Ras stimulates endogenous p38␣ phosphorylation while inducing endogenous p38␥ expression. Furthermore, experiments with SB to inhibit and MKK6 to stimulate p38␣ phosphorylation reveal the phosphorylationdependent Ras inhibitory role of p38␣, whereas the p38␥ depletion analyses show its requirement for Ras transformation independent of phosphorylation (Fig. 9). These results indicate that the transforming activity of Ras oncogene in a given system will be determined by the signaling integration among endogenous p38 family members. A higher ratio of unphosphorylated p38␥ proteins over phosphorylated p38␣ proteins would favor Ras transforming activity and vice versa.
It is unlikely that p38␥ promotes Ras transformation through its intrinsic mitogenic activity. This is because p38␥ expression is not induced by other mitogens, and p38␥ overexpression does not increase DNA synthesis or lead to transformation. Our results do suggest, on the other hand, that formation of a complex between p38␥ and ERK proteins may play a role in p38␥ maintaining the Ras-transformed phenotype (Fig.  9). This is indicated by reduced ERK/p-ERK proteins in the p38␥ but not p38␣ complex following MKK6-induced inhibition of Ras-dependent growth and by increased ERK/p-ERK binding through p38␥/AGF overexpression in response to Ras signaling. This conclusion, however, remains to be further proven by directly demonstrating the Ras transformation inhibitory activity of ERK binding-deficient p38␥ mutants. Because both p38␥ and p38␣ bind to ERK proteins, and only p38␥/AGF has an increased affinity to ERKs in the presence of Ras, future analyses should focus on the structural differences between p38␣ and p38␥ proteins as well as their relationships with phosphorylation on the kinase subdomain.
p38␥ contains a PDZ domain-binding motif (ETPL) in its C terminus, which is absent in p38␣ protein (50). This motif has been shown to be important for its subcellular localizations and/or its interactions with other PDZ domain-containing proteins to maintain certain structures of cytoskeleton, a process that is important for Ras transformation (51,52). Moreover, these interactions are also regulated by protein phosphorylations and dephosphorylations (51,53). Because ERK protein requires complex formation with other proteins for its activities such as nuclear localizations and induced epithelial morphogenesis (54,55), p38␥ may promote Ras transformation through its C terminus-mediated ERK binding by a scaffoldlike mechanism.
Human colorectal cancer is the second leading cause of cancer death in the United States, to which K-ras mutation is the most established contributing factor (56). Whereas various approaches have been tested to inhibit Ras oncogene activity (57,58), effective therapeutic agents to selectively inhibit activated ras oncogene in human cancer remain to be established (59). This slow progress is mainly due to lack of specific Ras oncogene effectors because many of those are shared by normal cellular Ras proteins in response to mitogenic signaling (26). p38␥, however, appears to be a specific Ras oncogene effector because it is not expressed in normal cells/tissues (3) and is induced selectively by Ras oncogene and not by other mitogens. FIG. 7. Increased ERK binding by non-phosphorylated p38␥ in response to transient K-Ras expression. Normal IEC-6 cells were transiently transfected with FLAG-tagged wild-type or the mutant p38s (AGF) in the absence or presence of K-Ras as indicated. FLAG precipitates were analyzed for the presence of endogenous ERK/p-ERK proteins. Similar results were obtained from a separate experiment.
Most significantly, our results further show that p38␥ gene expression is elevated ubiquitously in a set of primary human colon cancer tissues over matched normal tissues. Thus, p38␥ may serve as a novel diagnostic marker and therapeutic target for human colon cancer.  9. An experimental model shows a requirement of p38␥ for K-Ras transformation as opposed to the inhibitory activity of phosphorylated p38␣. K-Ras activates p38␥ by increasing its expression without phosphorylation but stimulates p38␣ by phosphorylation. Increased non-phosphorylated p38␥ protein promotes Ras transformation, whereas induced phosphorylated p38␣ inhibits Ras transforming activity. Experiments with several agents suggest the critical role of p38␥-ERK complex formation in p38␥ promoting Ras transformation, although this mechanism remains to be further established (?). This model suggests that Ras transforming activity in a given system will be determined by the signaling integration of p38 family members.