p38α Antagonizes p38γ Activity through c-Jun-dependent Ubiquitin-proteasome Pathways in Regulating Ras Transformation and Stress Response*

p38 MAPK family consists of four isoform proteins (α, β, γ, and δ) that are activated by the same stimuli, but the information about how these proteins act together to yield a biological response is missing. Here we show a feed-forward mechanism by which p38α may regulate Ras transformation and stress response through depleting its family member p38γ protein via c-Jun-dependent ubiquitin-proteasome pathways. Analyses of MAPK kinase 6 (MKK6)-p38 fusion proteins showed that constitutively active p38α (MKK6-p38α) and p38γ (MKK6-p38γ) stimulates and inhibits c-Jun phosphorylation respectively, leading to a distinct AP-1 regulation. Depending on cell type and/or stimuli, p38α phosphorylation results in either Ras-transformation inhibition or a cell-death escalation that invariably couples with a decrease in p38γ protein expression. p38γ, on the other hand, increases Ras-dependent growth or inhibits stress induced cell-death independent of phosphorylation. In cells expressing both proteins, p38α phosphorylation decreases p38γ protein expression, whereas its inhibition increases cellular p38γ concentrations, indicating an active role of p38α phosphorylation in negatively regulating p38γ protein expression. Mechanistic analyses show that p38α requires c-Jun activation to deplete p38γ proteins by ubiquitin-proteasome pathways. These results suggest that p38α may, upon phosphorylation, act as a gatekeeper of the p38 MAPK family to yield a coordinative biological response through disrupting its antagonistic p38γ family protein.

Mitogen-activated protein kinase (MAPK) 3 pathways consist of extracellular signal-regulated kinase (ERK), c-Jun N-termi-nal kinase (JNK), and p38 cascades (1,2). The ERK activity is generally required for cell proliferation and transformation (3,4), whereas JNK and p38 pathways are involved in stress response (5,6). Both antagonistic and cooperative activities among these three MAPK pathways have been reported in mediating various biological responses (5,(7)(8)(9), but mechanisms involved in these signaling integrations are mostly unknown. Moreover, each of MAPK pathways consists of several family members (1,2,10), and how these isoform proteins coordinate for a biological response is unclear. Demonstrating both specific and integrated activities of MAPK family proteins as well as underlying mechanisms is essential for understanding MAPK functions in regulating proliferation, transformation, and stress response.
The p38 family consists of four isoforms, ␣, ␤, ␥, and ␦ (10). The p38 upstream activators include MAPK kinase 6 (MKK6) and MKK3. Downstream effectors consist of kinases such as MAPK-activating protein kinase 2 and PRAK (p38-related/activated protein kinase) as well as transcription factors including activating transcription factor-2 (ATF2), myocyte enhancement factor 2, and c-Jun (10,11). p38␣ (also called p38 (10)) is the most abundant and ubiquitously expressed family protein and has a well established role in stress response and inflammation (10 -12). Another important function of p38␣ is to inhibit Ras oncogene activity (7,(13)(14)(15)(16). p38␥, on the other hand, is also expressed in many cancer cell lines, and its phosphorylation has also been involved in stress response (17)(18)(19). Our recent studies showed that p38␥ expression is induced by Ras oncogene, which in turn is required for Ras transforming and/or invasive activity independent of phosphorylation (20,21). These results together suggest that p38␣ and p38␥ may antagonize each other, but the direct evidence for this antagonism has been lacking. This information is critical for understanding how these two p38 family proteins coordinate for an orchestrated biological response in Ras transformation and stress response.
One major obstacle for demonstrating specific effects of p38␣ versus p38␥ phosphorylation is the lack of constitutively active kinases. All MAPKs are activated by dual phosphoryla-tion on threonine and tyrosine residues within a conservative Thr-Xaa-Tyr motif (22)(23)(24). Although fusing enzyme-substrate approaches have been used for studying JNK (25,26) and ERK (27) activation, there have been so far no similar studies for p38 MAPKs. In this report constitutively active p38␥ and p38␣ fusion constructs (MKK6-p38) were first utilized to demonstrate their opposite and/or distinct localizations and activities in regulating c-Jun phosphorylation, Ras transformation, and/or cell death. Further experiments showed that p38␣ phosphorylation triggers p38␥ protein depletion by c-Jun-dependent ubiquitin-proteasome pathways in inhibiting Ras-dependent growth and/or increasing stress-induced cell death. These results suggest a feed-forward mechanism by which p38␣ phosphorylation augments with resultant p38␥ protein depletion in regulating Ras transformation and stress response.
Transfection, Infection, Small Interfering RNA, Soft Agar, and Cell Death Assays-For reporter and promoter assays, AP-1 Luc and VDR-Luc were transiently transfected by calcium phosphate, and the luciferase activity was assayed 48 h later (39). Adenoviral infection to overexpress MKK6 and retroviral infection (pSUPER) to deplete p38␥ protein were performed as previously described (20,21). For protein expression and localization, cells were transiently transfected and analyzed 48 -72 h later. To assess the effects of fusion protein expression in IEC-6/K-Ras cells, MKK6-p38␥ and MKK6-p38␣ constructs (and their AGF mutants) were stably transfected through G418 selection (32), and early passages of these cells were used for analyses.
For soft-agar assays, the vector and fusion construct stably transfected cells were plated in growth medium containing 0.33% Sea-plaque-agarose. Formation of multicellular colonies was visualized and quantitated about 2 weeks later as previously described (20). To assess cell death, cells were treated with and without 20 M arsenite (ARS) or infected with adenoviruses expressing MKK6 or ␤-galactosidase as a control, and cell death induced was estimated by viability assays (trypan blue staining) and/or flow cytometry as previously described (13,39). All experiments were repeated at least three times and analyzed by Student's t test and/or analysis of variance for statistically significant difference.
Immunostaining, Immunoprecipitation, and Western Blot Assays-For immunostaining, cells were plated on coverslips and fixed in 3.7% formaldehyde. After permeabilized in a buffer containing 0.5% Triton X-100 and 0.5% Nonidet P-40, cells were blocked in 3% bovine serum albumin in phosphate-buffered saline. Cells were then double-stained for HA-tagged proteins using a mouse monoclonal anti-HA/anti-mouse fluorescein isothiocyanate and for phospho-p38 using a rabbit phospho-p38/anti-rabbit Cy3, as previously described (20,35). For in vivo ubiquitination and protein degradation assays, 293T cells were transiently transfected for 24 h followed by a 2-h pulse treatment with and without 20 M ARS 1 day later. After an additional 24 h incubation (ϩ10 M MG132 or 15 M lactacystin for the last 6 h), cells were lysed 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). Lysates were then subjected to immunoprecipitation with a HA antibody or FLAG antibody with a portion analyzed directly by Western blot as an input control. All the following procedures for Western blotting were the same as previously described (31,32).

Constitutively Active p38␣ Increases, Whereas Constitutively Active p38␥ Decreases c-Jun Phosphorylation That Couples with
Their Opposite Localizations-To study the specific effects of p38␥ versus p38␣ phosphorylation, p38␥/␣ cDNA (and the non-phosphorylatable mutant through changing the conservative phosphorylation motif TGY to AGF) were fused in-frame with its activator MKK6 through a decapeptide linker (Gly-Glu) 5 (Fig. 1A) as previously described (25)(26)(27). In addition, a HA tag was incorporated at the N-terminal end to facilitate detection of the fusion protein. Western analysis of transiently transfected 293T cells showed expression of the MKK6-p38␥ and MKK6-p38␥/AGF fusion proteins at about 80 kDa, but only MKK6-p38␥ is recognized by a p-p38 antibody (constitutively active) (Fig. 1B). The integrity and specificity of the fusion proteins were further demonstrated by immunoprecipitation with a HA antibody and Western blotting with MKK6, p38␥, or p-p38 antibody (Fig. 1C). Similar results were also obtained with analyses of the p38␣ fusion proteins ( Fig. 2C and data not shown).
p38␣ is known to be translocated into the cytoplasm upon phosphorylation by stress signaling (40). p38␥, on the other hand, was previously shown to be both in the nucleus and cytoplasm (21,28), but the relationship between p38␥ phosphorylation and its localization has not been studied. To determine whether p-p38␥ is localized differently than p-p38␣, fusion proteins were transiently expressed, and their localizations were examined by double-immunostaining against HA and p-p38. Consistent with the cytosolic p-p38␣ (40), MKK6-p38␣ is localized predominantly in cytoplasm, whereas its AGF mutant is both in the nucleus and cytoplasm (Fig. 1D). On the FIGURE 1. Constitutively active p38␣ is cytosolic, whereas constitutively active p38␥ is both in nucleus and cytoplasm. A, a diagram for MKK6-p38␥ and MKK6-p38␣ fusion proteins (and their AGF mutants). B and C, expression and phosphorylation of p38 fusion proteins. 293T cells were transfected with indicated constructs, and the protein expression was examined by direct Western (WB; B) and HA immunoprecipitations (C). Similar results were also obtained for MKK6-p38␣ fusion proteins (data not shown). D and E, localizations of p38␣ (MKK6-p38␣), p38␥ (MKK6-p38␥), and ERK2 (ERK2-MEK1) fusion proteins. IEC-6/K-Ras cells were transiently transfected with different constructs and double immuno-stained against HA and p-p38 (or Myc) as indicated. DAPI, 4Ј,6-diamidino-2-phenylindole.

p38␣ Depletes p38␥ Proteins by Ubiquitin-proteasome Pathways
contrary, MKK6-p38␥ is both in the cytoplasm and nucleus, whereas its non-phosphorylatable AGF is predominantly in the cytoplasm (Fig. 1D). These opposite and phosphorylation-dependent localizations of p38␣ and p38␥ fusion proteins are not because of the integrated MKK6, as every fusion protein contains the same molecule that was visible both in nucleus and cytoplasm (Fig. 1D, bottom). The cellular distributions of the active p38␥ are similar to those described for the active ERK2 fusion proteins (ERK2-MEK1-LA), as previously described ( Fig. 1E; (27). These results indicate that cellular p38␥ opposes p38␣ in cellular localizations by phosphorylationdependent mechanisms.
Signaling through the p38 pathway is known to stimulate AP-1-dependent transcription (10). To demonstrate if p38 fusion proteins are functionally active in regulating AP-1, 293T cells were transiently transfected with these constructs together with an AP-1 reporter (AP-1-Luc) or an AP-1-dependent VDR genomic promoter (VDR-Luc) (32,33), and luciferase activity was assayed. Results in Fig. 2A showed that MKK6-p38␣ significantly increases AP-1 activity, an effect more than that achieved by expressing MKK6 alone, whereas all other fusion proteins have no substantial effects. A similar activation was also observed with the VDR promoter, albeit both AGF mutants are suppressive in this case (Fig. 2B). These results suggest that one specific function of the constitutively active p38␣ may be stimulating AP-1-dependent transcription, whereas the p38␥, regardless of its phosphorylation status, may not play a significant role in this regulation.
MAPKs regulate AP-1 activity by activating/phosphorylating its components such as c-Jun, c-Fos, and ATF2 (41). To explore whether MKK6-p38␣ stimulation of AP-1 couples with its activity to phosphorylate c-Jun and/or ATF2, fusion constructs were co-transfected with either a c-Jun-or ATF2-expressing plasmid, and their effects on c-Jun/ATF2 phosphorylation were examined by direct Western. Results in Fig. 2C showed that the constitutively active p38␣ increases, whereas its AGF mutant decreases c-Jun phosphorylation. These effects are opposite to p38␥ fusion proteins, as c-Jun phosphorylation was decreased by MKK6-p38␥ but increased by MKK6-p38␥/AGF. Although ATF2 is a well established substrate of all p38 family proteins in vitro and in vivo (10,31), both p38␣ and p38␥ fusion proteins increase ATF2 phosphorylation independent of phosphorylation. These results indicate that these p38 fusion proteins are functionally active in regulating c-Jun/AP-1 activity and that only MKK6-p38␣, but not MKK6-p38␥, induced c-Jun (not ATF2) phosphorylation contributes positively to the AP-1 transcriptional activity.
p38␥ Increases Ras Soft-agar Growth and Inhibits Stress-induced Cell Death Independent of Phosphorylation-Our previous studies have shown that Ras increases p38␥ protein expression, and induced p38␥ in turn promotes Ras transformation independent of phosphorylation (20). Specific effects of p38␥ phosphorylation on Ras transformation, however, remain unknown. To address this question, we examined whether the constitutively active p38␥ has a distinct activity in regulating Ras-dependent growth as compared with its non-phosphorylatable mutant as well as p38␣ fusion proteins. In this case, these fusion proteins were stably expressed in K-Ras transformed rat intestinal epithelial cells (IEC-6/K-Ras), and their effects on Ras transformation were examined by colony formation assays (20). Expression of both MKK6-p38␥ and MKK6-p38␥/AGF proteins increases the soft-agar growth as compared with the vector control with a more substantial effect observed with the mutant protein, whereas neither MKK6-p38␣ nor its AGF mutant showed a substantial effect (Fig. 3, A-D). These results indicate that both the active and mutant p38␥ can increase Ras transformation, and the non-phosphorylatable FIGURE 2. Phospho-p38␣ stimulates c-Jun phosphorylation and AP1-dependent transcription, whereas the active phospho-p38␥ inhibits c-Jun phosphorylation without significant effects on AP-1 activity. A and B, constitutively active p38␣ but not p38␥ increases AP-1-dependent transcription. Human 293T cells were transfected with different constructs together with an AP-1 reporter (A) or a mouse VDR promoter (B) and analyzed for luciferase activity 48 h later. Results shown are mean of three to four experiments (ϮS.D.) with the asterisk indicating a statistically significant difference as compared with the vector control (p Ͻ 0.05, Student's t test). C, constitutively active p38␣ stimulates and constitutively active p38␥ inhibits c-Jun phosphorylations. Cells were transiently transfected with plasmids as indicated together with either a c-Jun or ATF2 expression construct, and protein expression/phosphorylations were analyzed by direct Western. OCTOBER 26, 2007 • VOLUME 282 • NUMBER 43 fusion protein has a greater potency in this regulation. This effect is consistent with the effector role of p38␥ in Ras transformation through Ras-induced expression/dephosphorylation as we previously proposed (20). The inability of the constitutively active MKK6-p38␣ in regulating Ras soft-agar growth, on the other hand, differs from the previously observed growth inhibition by adenovirus-mediated MKK6 overexpression/ p38␣ phosphorylation (20). These differences probably result from sustained (stable) versus transient p38␣ phosphorylation. These results further establish a distinct role of p38␥ versus p38␣ in regulating Ras transformation.

p38␣ Depletes p38␥ Proteins by Ubiquitin-proteasome Pathways
p38 MAPK proteins by nature are stress kinases, and we wished next to determine whether stably expressed p38␥ and p38␣ fusion proteins may have distinct roles in regulating stress response. Of interest, stably expressed MKK6-p38␥ or MKK6-p38␥/AGF blocks ARS-induced JNK/c-Jun phosphorylation, whereas transfected MKK6-p38␣ fusion proteins are without effect (Fig. 4, A and B). The inhibitory effects of p38␥ fusion proteins on ARS-induced JNK/c-Jun activation are different from their intrinsic activities on c-Jun phosphorylation (Fig.  2C), as the latter depends on p38␥ phosphorylation and occurs without the concomitant JNK regulation. Importantly, stably expressed p38␥ but not p38␣ fusion proteins suppress ARSinduced cell death, and both MKK6-p38␥ and MKK6-p38␥/ AGF showed a similar protective activity (Fig. 4, C and D). The cell death inhibitory activity of the p38␥ fusion proteins was further confirmed by an increased ARS-induced toxicity after small interfering RNA-mediated p38␥ protein depletion in IEC-6/K-Ras cells (data not shown); the viability decreased from 46.0 Ϯ 3% in the ARS-treated control to 19.8 Ϯ 5.0% in p38␥-depleted ARS group (p Ͻ 0.05) with the sub-G 1 population correspondingly increased from 27.3 to 46.8%. These results together reveal a stress inhibitory activity of p38␥ independent of phosphorylation.
A cell death inhibitory activity of p38␥ promotes us to further explore if there is an increased sensitivity to stress-induced cell death in p38␥ knock-out (p38␥ Ϫ/Ϫ ) MEFs (28). Treatment of p38␥ Ϫ/Ϫ and wild-type (p38␥ ϩ/ϩ ) MEFs with ARS induces similar cell death, which couples with an increased JNK but not p38␣ phosphorylation in both lines (Fig. 4, E and F). Because ARS induces p38␣ phosphorylation in IEC-6/K-Ras but not in MEF cells (Fig. 4E and 5A), these results indicate that p38␥ may only be anti-apoptotic when p38␣ is phosphorylated. Consistent with the JNK inhibition by p38␥ fusion proteins in IEC-6/ K-Ras cells, however, there was an increased JNK expression/ activation in p38␥ Ϫ/Ϫ cells (Fig. 4E). These results together indicate that in addition to the positive role in Ras transformation, p38␥ also suppresses the JNK stress pathway, albeit this activity only leads to a cell death inhibitory response in IEC-6/ K-Ras cells.
Stress Preferably Phosphorylates p38␣ over p38␥, and Phosphorylated p38␣ Primes p38␥ for a Down-regulation-p38␥ has been shown to be activated by several types of stresses (17,18,42,43). Most of these studies, however, were performed by overexpressing p38␥ proteins and/or through isolating activated p38␥ kinases through immunoprecipitation. As a result, a physiological role of endogenous p38␥ in regulating stress response remains un-established. We sought to address this question in IEC-6/K-Ras cells that express both p38␣ and p38␥ proteins (20) (Fig. 5A, top left). In response to either ARS or sorbitol, a single band around 39 kDa was induced that was reacted with a specific p-p38 antibody that recognizes all phosphorylated p38 isoform proteins (Fig. 5A, top left). Because p38␣ is 38 kDa in size, whereas p38␥ is about 45 kDa, these results suggest that it is p38␣ and not p38␥ that is phosphorylated by stress signaling in these cells. This speculation was further confirmed by Western analyses of p-p38 (Fig. 5A, top right) or p38␥ precipitates (Fig. 5A, middle left). Direct Western analyses of 293T cells in which all four p38 family proteins are expressed further showed that only p-p38␣ is induced by ARS or sorbitol (data not shown), and increased p-p38␥ protein is only detectable through examining p38␥ precipitates (Fig. 5A,  middle right). These results together indicate that p38␣ is preferably phosphorylated over p38␥ in cells expressing both proteins in stress response.
The phosphorylation-independent stress-inhibitory property of p38␥ promotes us to further explore mechanisms for its resistance to stress-induced phosphorylation. Because p38␣ and p38␥ oppose each other in regulating c-Jun phosphorylation, Ras transformation, and cell death and p38␣ is preferably phosphorylated, p38␣ phosphorylation may directly suppress p38␥ activation. To test this possibility, MKK6 was overexpressed through adenovirus infection to examine if increasing endogenous p38␣ phosphorylation directly antagonizes p38␥ activity. Although MKK6 phosphorylates co-transfected p38␣ and p38␥ (31), there was a decrease in endogenous p38␥ protein expression in response to ad-MKK6-induced p38␣ phosphorylation in IEC-6/K-Ras cells with and without p38␥ fusion pro- MKK6-p38␥ and its AGF mutant were stably expressed in IEC-6/K-Ras cells through G418 selection, and early passages of these cells were analyzed for protein expression and soft-agar growth as previously described (20). Results in B are means of colony numbers per field from four separate experiments with each from at least 17 fields (ϮS.D., p Ͻ 0.05, analyzed with analysis of variance). C and D, stable expression of p38␣ fusion proteins does not affect anchorage-independent growth of IEC-6/K-Ras cells. Cells were transfected, selected, and analyzed for protein expression and the soft-agar growth as above, with results in D from three separate experiments with each from at least 19 fields (ϮS.D., p Ͼ 0.05, analysis of variance).

p38␣ Depletes p38␥ Proteins by Ubiquitin-proteasome Pathways
tein expression (Fig. 5B). Consistent with the previous observation (20), MKK6 infection results in a more than 50% inhibition of the soft-agar growth as compared with the control infection in all three sublines (data not shown). Because p-p38␣ suppresses and p38␥ promotes Ras transformation, these inhibitions probably represent a combination effect of transient p38␣ phosphorylation and the resultant p38␥ depletion.
A down-regulation of p38␥ protein expression by p38␣ activation suggests a novel cross-talk between these two p38 family proteins. We sought next to examine if inhibiting p38␣ phos-phorylation reverses the p38␥ depletion. Indeed, treatment of IEC-6/K-Ras cells with SB203580 (SB) that inhibits p38␣ but not p38␥ activity (10) elevates cellular p38␥ proteins dose-dependently (Fig. 5A,  bottom left), which also couples with an increased soft-agar growth (data not shown). Although ectopically expressed p38␥, but not p38␤, inhibits endogenous p38␣ phosphorylation, there are no alterations in p38␣ protein expression (Fig. 5A,  bottom right), indicating a specific antagonism between p38␥ protein expression and p38␣ phosphorylation. Moreover, experiments in human breast cancer 231 cells showed that the p38␥ protein depletion event occurs specifically to p38␣ (and less JNK) phosphorylation by MKK6 but not to JNK (less p38␣) phosphorylation by taxol-independent of Tet-inducible estrogen receptor protein expression (Fig. 5C) (21). Different than the growth inhibition in IEC-6/K-Ras cells, p38␣ phosphorylation and p38␥ depletion by MKK6 lead to an increased cell death in 231 cells in the absence of estrogen receptor expression (viability, 84 Ϯ 9.0% in control versus 38 Ϯ 15% in MKK6 group, p Ͻ 0.05) as previously described (35). These results together suggest that p38␣, upon phosphorylation, negatively regulates p38␥ protein expression, which may be required for an orchestrated response in regulating Ras transformation and/or cell-death by a cell-type dependent mechanism.
p38␣ Phosphorylation Decreases p38␥ Protein Expression by Ubiquitin-proteasome Pathways-Protein ubiquitination is an important posttranscriptional regulation that typically marks modified proteins for proteasome-mediated degradation (44,45). To explore whether p38␣ phosphorylation decreases p38␥ protein expression by ubiquitin-proteasome pathways, 293T cells were transiently transfected with a FLAG-tagged p38␣ and/or p38␥-expressing plasmid together with a HA-tagged Ub expressing cDNA (36,37). To stimulate p38␣ phosphorylation, cells were either co-transfected with MKK6 or pulse-treated with ARS 24 h before their collection. Also, a group of transfected cells were subjected to a short period treatment with a proteasome inhibitor MG132 to determine

. p38␥ inhibits JNK activation in stress response that couples to a cell-death inhibition in IEC-6/K-Ras cells but not in mouse fibroblasts.
A-D, stable expression of p38␥ (but not p38␣) fusion proteins inhibits stress-induced JNK activation and cell death. IEC-6/K-Ras cells stably expressing p38 fusion proteins were treated with and without 20 M ARS and analyzed by Western 24 h (A and B) and for cell death 48 h later (C and D). CO, control. Results in A were from two separate analyses of the same lysates, which was confirmed by an independent experiment. Cell death was assessed by trypan-blue staining (viability) and further confirmed by fluorescence-activated cell sorter (asterisk, indicating sub-G 1 populations in the % on top of each column for ARS-treated groups, which remained similar for control cells; data not shown). The viability results from C and D (open bar, control; striped bar, ARS) are from four to six individual experiments, with p Ͻ 0.05 for MKK6-p38␥ or MKK6-p38␥/AGF group versus the vector control after ARS by Student's t test, but p Ͼ 0.05 for the differences among these three groups by analysis of variance. There was no statistically significant difference in D by either test. E and F, an increased JNK expression and/or activation in p38␥ knock-out MEFs. p38␥ wild-type (p38␥ ϩ/ϩ ) and knock-out (p38␥ Ϫ/Ϫ ) MEFs between passage three to five were treated with and without ARS and analyzed by Western and viability assays. Results in F are mean of three separate experiments (ϮS.D., open bar, control; striped bar, ARS). OCTOBER 26, 2007 • VOLUME 282 • NUMBER 43 whether p38␣ phosphorylation-induced p38␥ protein depletion is dependent on proteasome activity.

p38␣ Depletes p38␥ Proteins by Ubiquitin-proteasome Pathways
Direct Western analyses show that levels of expressed p38␥ proteins are decreased in response to either MKK6 or ARS, whereas those of transfected p38␣ proteins remain relatively constant under the same conditions (Fig. 6C, Input Control). Importantly, MG132 treatment blocks the down-regulation, resulting in an increased p38␥ protein expression in both MKK6 and ARS group without significant impacts on the p38␣ protein contents. Similar results were also obtained in a separate experiment in which transfected cells were incubated with another proteasome inhibitor, lactacystin (Fig. 6D). A less substantial reversal effect by lactacystin in this experiment in the ARS group may be due to a slight, instead of significant (as observed in Fig. 6C), decrease in p38␥ protein expression. A general consistent down-regulation of p38␥ proteins by two types of stress stimuli as well as its reversal by two distinct proteasome inhibitors, on the other hand, strongly indicates that one mechanism by which p-p38␣ decreases p38␥ protein expression may occur by proteasome-dependent pathways. Western analyses of HA precipitates further showed that both transfected p38␣ and p38␥ are constitutively ubiquitinated in the absence of stress, and a unique mono-ubiquitinated band was only detected for the p38␥ (about 57 K Da) but not for p38␣ protein (about 46 kDa) (Fig. 6A). An increase in Ub-p38␥ proteins in HA precipitates after MG132 in p38␥ plus MKK6 or plus ARS groups but not in p38␥ expression alone (Fig. 6A, top, seventh and eighth lanes versus third and fourth lanes from right for the combination the and sixth versus second lane for p38␥ alone) indicates that only stress-induced, but not constitutively, ubiquitinated p38␥ protein is degraded by the proteasome-dependent pathway. Analyses of FLAG precipitates further reveal that a decreased p38␥ protein expression by either ARS or MKK6 is reversed by MG132 treatment (Fig. 6B, middle, third and fourth lanes versus the second lane from the right and the seventh and eight lanes versus the sixth lane). Consistent with the results from the HA precipitation, levels of p38␥ proteins from FLAG precipitates in p38␥ expression alone were not increased by the MG treatment (Fig. 6B, middle, sixth lane versus second lane from the right), further indicating that the constitutively ubiquitinated p38␥ may not be sensitive to the proteasome inhibition. This conclusion is further supported by an ineffectiveness of two proteasome inhibitors in increasing total p38␥ protein concentrations in the absence of stress as observed from direct Western ( Fig. 6C and 6D, sixth versus second lane from the right). These results together indicate that stress-induced but not constitutively ubiquitinated p38␥ proteins are degraded by proteasome-dependent pathways.
Both p38␣ and p38␥ Are Ubiquitinated Independent of Phosphorylation but p38␣ Requires Phosphorylation to Activate c-Jun in Depleting p38␥ Protein-One possibility for proteasome-dependent p38␥ depletion is that p38␥ phosphorylation itself, although undetectable by direct Western under normal conditions, may be able to trigger its own ubiquitination/degradation. To investigate this possibility, fusion proteins were expressed with HA-Ub, and their in vivo ubiquitination was analyzed by Western blotting. Results in Fig. 7A showed that both constitutively active and AGF forms of p38␣ and p38␥ fusion proteins are ubiquitinated to a similar extent, indicating a phosphorylation-independent modification. Consistent with results from the c-Jun co-expression (Fig. 2C), only constitutively active p38␣ stimulates endogenous c-Jun phosphorylation (Fig. 7A), leading to its increased ubiquitination, as recently reported in response to osmotic stress signaling (46). This c-Jun phosphorylation and consequent ubiquitination effect of FIGURE 5. p38␣ is preferably phosphorylated over p38␥ by stress signaling, and phosphorylated p38␣ triggers a down-regulation of p38␥ protein expression. A, stress signaling preferably phosphorylates p38␣ over p38␥, and there is a specific antagonism between p38␣ phosphorylation and p38␥ protein expression. IEC-6/K-Ras cells were treated with ARS or sorbitol (SOR) for 30 min and analyzed for p38␣ versus p38␥ phosphorylation by direct Western (top left) and immunoprecipitations (IP) followed by Western blotting (top right and middle left). In 293T cells stress-induced-p38␥ phosphorylation was examined by Western analyses of p38␥ immunoprecipitates (middle right). To inhibit p38␣ phosphorylation, IEC-6/K-Ras cells were incubated with and without SB for 24 h and analyzed for p38␣ phosphorylation/p38␥ protein expression by Western (bottom left). To study the effects of p38␥ overexpression, NIH3T3 cells were transiently expressed with FLAG-tagged p38␥ or p38␤ proteins, and their effects on endogenous p38␣ phosphorylation/ expression were analyzed by Western (bottom right; the asterisk indicate endogenous p-p38␣). B, p38␣ phosphorylation leads to a depletion of endogenous p38␥ proteins in IEC-6/K-Ras cells with and without p38␥ fusion protein expression. Shown are IEC-6/K-Ras cells stably transfected with MKK6-p38␥, MKK6-p38␥/AGF, or a vector were infected with adenovirus (MKK6 with ␤-galactosidase (␤-Gal) as a control) and analyzed by Western 48 h later. C, p38␣ but not JNK phosphorylation depletes p38␥ protein expression in estrogen receptor (ER)-positive (ϩTet) and negative (ϪTet) human breast cancer cells. Tet-on estrogen receptor 231 cells (21) were infected with adenovirus (MKK6, ␤-galactosidase) or treated with taxol (50 M, 24 h) in the presence or absence of Tet and analyzed for protein phosphorylation and expression.

p38␣ Depletes p38␥ Proteins by Ubiquitin-proteasome Pathways
MKK6-p38␣ again is similar to that observed with MKK6-p38␥/AGF expression (Fig. 7A). These results indicate that p38␥ phosphorylation itself does not trigger its own ubiquitination, but p38␣ may require phosphorylation to activate a c-Jun-associated ubiquitination regulatory cascade.
Stress frequently phosphorylates both p38␣ and c-Jun (13,35,47). c-Jun activity, on the other hand, can integrate various phosphorylation and ubiquitination regulatory events of MAPK signaling (48 -51). It is, therefore, important to determine whether c-Jun is required for p-p38␣-induced p38␥ down-regulation. In this regard, 293T cells were transiently transfected with wild-type or dominant negative c-Jun constructs (35) together with a HA-Ub expressing cDNA, and their effects on ARS-induced p38␣ phosphorylation/p38␥ depletion were analyzed. Results in Fig.  7B, top, showed that expression of the dominant negative c-Jun reversed ARS-triggered p38␥ downregulation without affecting p38␣ phosphorylation, indicating an involvement of c-Jun phosphorylation in p38␥ depletion. Similar results were also obtained when experiments were performed without HA-Ub co-transfection (Fig. 7B,  middle). The observation that ARSinduced p38␣ phosphorylation couples to a decreased p38␥ protein expression only in c-Jun ϩ/ϩ but not in c-Jun Ϫ/Ϫ cells (Fig. 7B, bottom) further supports the requirement of c-Jun in p-p38␣-induced p38␥ depletion. Together with the sufficient role of MKK6-p38␣ in increasing c-Jun phosphorylation, these results indicate that p-p38␣ requires c-Jun activity to deplete p38␥ protein expression.

DISCUSSION
The p38 MAPK pathway plays an important role in regulating stress response (13,(52)(53)(54) and Ras activity (7,13,14,16,20). Although stress phosphorylates all p38 family proteins (18,31), and Ras, on the other hand, stimulates p38␣ phosphorylation and increases p38␥ expression (20), mechanisms by which activated p38 family proteins coordinate for an integrated biological response remain unknown. Here we show that p38␣ and p38␥ have antagonistic activities in Ras transformation and stress response, and p38␣ phosphorylation primes p38␥ protein for depletion by c-Jun-dependent ubiquitin-proteasome pathways (Fig. 7C). These results reveal a novel feed-forward mechanism FIGURE 6. p38␣ phosphorylation decreases p38␥ protein expression by proteasome-dependent mechanisms. Human 293T cells were transiently transfected with p38␥ and/or p38␣ together with a HA-Ub. To stimulate p38 phosphorylations and/or inhibit proteasome activity, cells were either co-transfected with MKK6 or treated with ARS 24 h before lysate collection in which 10 M MG132 or Me 2 SO (DMSO) was added for the last 6 h. Protein expression and ubiquitination were analyzed by direct Western (as an input control, C) as well as immunoprecipitations (IP) followed by Western blots (WB) as indicated (A and B). The asterisk in A indicates mono-ubiquitinated p38␥ in HA precipitates. In panel D 293T cells were similarly transfected but treated instead with 15 M of lactacystin (or Me 2 SO) for the last 6 h before being analyzed for protein expression by direct Western. GAPDH, glyceraldehyde-3-phosphate dehydrogenase.
by which p38␣ phosphorylation concerts with its resultant p38␥ protein depletion for a Ras inhibitory and/or pro-apoptotic activity.
This signaling integration mechanism was first suggested by opposite localizations and antagonistic regulations of c-Jun phosphorylation by constitutively active p38␥ versus p38␣ fusion proteins. The observation that p38␥ both increases Ras transformation and inhibits cell death whereas p-p38␣ is either Ras inhibitory or pro-apoptotic further supports their opposite functions. A greater enhancing effect of MKK6-p38␥/AGF over its constitutively active counterpart on Ras soft-agar growth (Fig. 3B) but not on celldeath protection (Fig. 4C) may indicate distinct mechanisms involved in p38␥ increasing Ras transformation and inhibiting stress response. The coupling of an increased p38␣ phosphorylation with a decreased p38␥ protein expression in every case, however, provides the direct evidence for their integrated activities. An increased Ras transformation by a p38␣ inhibitor SB that concurrently elevates cellular p38␥ proteins further consolidates the co-requirement of inhibiting p38␣ phosphorylation and increasing p38␥ protein expression for increased Ras activity. Additional analyses reveal that this interfamily cross-talk is triggered by p38␣ phosphorylation leading to a c-Jundependent p38␥ protein depletion through ubiquitin-proteasome pathways. p38␣, upon phosphorylation, may, therefore, act as a gatekeeper of the p38 family through depleting the antagonistic p38␥ protein via proteasome degradation pathways to amplify its Ras inhibitory and/or proapoptotic signal.
The demonstration of p38␣ phosphorylation triggering a depletion of p38␥ protein expression has important implications. Because p38␣ is the most abundant family protein expressed in all types of cells/tissues and is easily phosphorylated by various environmental stresses, these results may explain the general phenomenon of a lower level of p38␥ protein expression as compared with p38␣ (10). Moreover, the ability of phosphorylated p38␣ to down-regulate p38␥ protein expression may explain why p38␣ is preferably phosphorylated over p38␥ in response to stress in cells expressing both proteins (Fig. 5A). Because Ras both stimulates p38␣ phosphorylation and increases p38␥ protein expression (20), its transforming activity in a given system will be determined by an integrated signaling between anti-Ras FIGURE 7. p38␣ requires c-Jun to deplete p38␥ proteins through ubiquitin-proteasome pathways. A, p38␣ and p38␥ are both ubiquitinated independent of phosphorylation, but only the active p38␣ and the inactive p38␥ phosphorylate endogenous c-Jun proteins. Human 293T cells were transiently transfected with fusion constructs together with a HA-Ub-expressing cDNAs. Protein expression and phosphorylation were examined by a direct Western. B, p38␣ phosphorylation requires c-Jun activation to down-regulate p38␥ protein expression. 293T cells were transiently transfected with a wild-type (WT) or dominant negative (DT) c-Jun constructs in the presence (top panel) and the absence (middle panel) of HA-Ub and their effects on ARS-induced p38␥ depletion and/or p38␣ phosphorylation were analyzed by Western blot. In the bottom panel, wild type (c-Jun ϩ/ϩ ) and knock-out (c-Jun Ϫ/Ϫ ) c-Jun cells were treated with ARS and analyzed by Western. ARS-induced c-Jun phosphorylation and protein ubiquitinations were also coupled to p38␥ depletion/ p38␣ phosphorylation in IEC-6/K-Ras cells (data not shown). C, an experimental model showing a feed-forward mechanism by which p38␣ phosphorylation primes p38␥ protein for depletion by c-Jun-dependent ubiquitinproteasome pathways in regulating Ras transformation and stress response. p38␣ is most frequently phosphorylated in stress response, whereas p38␥ expression is typically induced by Ras oncogene, and p38␣ phosphorylation requires c-Jun activation to deplete p38␥ proteins by ubiquitin-proteasome pathways. Because p38␥ opposes p-p38␣ in regulating Ras transformation and stress response, our results suggest a feed-forward mechanism by which p-p38␣ cooperates with resultant p38␥ protein depletion to inhibit Ras transformation and/or to induce cell death. p38␣ Depletes p38␥ Proteins by Ubiquitin-proteasome Pathways p38␣ phosphorylation and pro-Ras p38␥ protein expression (Fig. 7C). In response to stress, on the other hand, increased cell death may only be envisioned when a pro-apoptotic p38␣ phosphorylation couples with a depletion of anti-apoptotic p38␥ protein expression.
One intriguing aspect of p38␣ and p38␥ signaling integration is that this cross-talk occurs between p38␣ phosphorylation and p38␥ protein expression. Although high levels of p38␥ protein expression inhibit endogenous p38␣ phosphorylation (possibly as a result of their competition for a common activator(s)), levels of p38␣ protein expression remain unaltered. Moreover, this inhibitory effect is not specific for p38␣, as p38␥ also blocks JNK phosphorylation. A transient regulating of p38␣ phosphorylation, on the other hand, consistently leads to an opposite alteration in cellular p38␥ protein concentrations. Because signals regulating p38␣ phosphorylation are more abundant and p38␣ is preferably phosphorylated, these twoway cross-talks will conceivably favor the feed-forward mechanism by which increased p38␣ phosphorylation augments with the resultant p38␥ protein depletion to inhibit Ras transformation and/or increase cell death. It should be pointed out that p38␣ phosphorylation does not inhibit p38␥ transcription, as demonstrated by analyses of a human p38␥ promoter, 4 but instead triggers proteasome-dependent p38␥ degradation. Because the active p38␣ induces c-Jun phosphorylation that is required for p38␥ protein depletion, p38␣ upon phosphorylation may cooperate with its resultant c-Jun activity to downregulate p38␥ protein expression. These results together illustrate an interesting scenario in which a phosphorylation event of one p38 family member initiates an ubiquitinated modification of another p38 family protein to counteract its antagonistic activity for a coordinative response.
Experiments with transient transfections show that both p38␥ and p38␣ are constitutively ubiquitinated in the absence of stress, which, however, is not regulated by the proteasome inhibition. Whether these constitutively ubiquitinated p38␣/␥ proteins, especially the unique mono-ubiquitinated p38␥, dictate their distinct cellular localizations as recently described for PTEN (55) remains for further explorations. In response to stress, however, both MG132 and lactacystin increased levels of the decreased p38␥ proteins. These results indicate that stressinduced but not constitutively ubiquitinated p38␥ proteins are degraded, at least in part, by proteasome-dependent pathways. Ubiquitylation is one major inducible and reversible protein modification that regulates stability and/localizations of many key signaling proteins including c-Jun (36,48). Of interest, we show that p38␣ phosphorylation alone is sufficient to activate c-Jun, and this activation is required for its p38␥-depleting activity. JNK has been shown to phosphorylate/activate E3 ligase family proteins in regulating protein ubiquitination (49,51). p38␣ may act through similar mechanisms to prime p38␥ for ubiquitination and proteasome-dependent degradation. Because c-Jun can form a complex with key ligases in these reactions (50), c-Jun may serve as a platform to facilitate this reaction. It would be of interest to explore further why the p38␣/c-Jun cascade, instead of the classical JNK/c-Jun pathway, is involved in regulating p38␥ ubiquitination and degradation.