p38 Isoforms Have Opposite Effects on AP-1-dependent Transcription through Regulation of c-Jun

p38 MAPK pathway signaling is known to participate in cell proliferation, apoptosis, and differentiation, in a manner dependent on the cellular context. The factors that determine the specific biological response in a given cell type, however, remain largely unknown. We report opposite effects of the p38 isoforms on regulation of AP-1-dependent activities by p38 activators MAPK kinase 6 (MKK6) and/or arsenite in human breast cancer cells. The p38β isoform increases the activation of AP-1 transcriptional activities by MKK6 and/or arsenite, whereas p38γ/p38δ inhibits or has no effect on the stimulation. The p38β does so by increasing the levels of phosphorylated c-Jun, whereas the p38γ and -δ isoforms may act by regulating the c-jun transcription. AP-1-dependent processes such as vitamin D receptor gene promoter activation and cellular proliferation were similarly activated by the p38β or inhibited by the p38γ and/or -δ isoforms. Whereas the human breast cancer cells express all four isoforms, mouse NIH 3T3 and EMT-6 cells express only some of the p38 family members, with p38β higher in 3T3 cells but p38δ only detected in the EMT-6 line. Consistent with the positive and negative roles of p38β and p38δ in AP-1 regulation, MKK6 stimulates AP-1-dependent transcription in NIH 3T3 but not EMT-6 cells. In support of a role of c-Jun regulation by p38 isoforms in determining AP-1 activity, the levels of endogenous c-Jun and its phosphorylated form on p38 activation are higher in NIH 3T3 cells. These results demonstrate the contrasting activities of the different p38 isoforms in transmitting the upstream signal to AP-1 and show that the expression profile of p38 isoforms determines whether the p38 signal pathway activates or inhibits AP-1-dependent processes.

Extracellular signals regulate cellular proliferation, differentiation, and death through activation of kinase cascades of the mitogen-activated protein kinases (MAPKs) 1 including ERK, JNK, and p38 (1)(2)(3)(4). The ERK MAPKs are most frequently activated by mitogenes, whereas the JNK and p38 MAPKs are strongly responsive to stress and inflammatory signals. The then individual MAPK activities often either collaborate or oppose each other in the regulation of biological responses to different stimuli. ERK activity, for example, has been shown to antagonize the p38 and JNK activities in regulation of apoptosis in PC-12 cells (5). At the level of a target gene expression, ERK kinase stimulates cyclin D1 transcription, which is suppressed by p38 MAPK (6). A similar opposing effect between ERKs and p38 MAPK is also observed in regulation of chondrogenesis of mesenchymes (7). Moreover, an antagonizing effect can occur between the JNK and p38 kinases, despite the fact that both pathways frequently respond to the same classes of stimuli (8,9). Hypertrophic agonists including endothelin-1 and phenylephrine, for example, stimulate p38 and JNK kinases in myocytes, in which p38 promotes but JNK suppresses the development of myocyte hypertrophy (10). Our previous work of analyzing Ras signal transduction pathways demonstrated that oncogenic Ras stimulates ERK, JNK, and p38 in NIH 3T3 cells, but the p38 activation in this process blocks the Ras signal through its inhibition of Ras-induced JNK downstream effects (11). In addition to their opposing activities, under certain circumstances ERK and p38 kinases cooperate in regulation of c-fos expression in response to UV light (12,13). It appears, therefore, that signaling cross-talks and integrations among the ERK, JNK, and p38 MAPK pathways are important steps to regulate the final signaling output from MAPK pathway activations. Since each of MAPKs has several subfamily members, it is possible that this signaling cross-talk and integration might also occur inside each type of MAPK among its isoforms to meet cellular requirements for various intricate and delicate biological processing of signals from stimuli.
AP-1 (activating proteins 1) are sequence-specific transcription factors composed of homodimers or heterodimers of the Jun family (c-Jun, JunD, and JunB) or heterodimers of the Jun family member with any of the Fos family members (c-Fos, FosB, Fra1, and Fra2) or other transcription factors such as activating transcription factor-2 (ATF2, cAMP-response element-binding protein, and NFAT (14 -16). AP-1 transcription factors are key regulatory molecules that play a central role in control of cell proliferation and transformation (14,16,17) by converting MAPK signals into expression of specific target genes (3,18). Members of the AP-1 family are regulated at both the transcriptional and posttranscriptional levels by MAPKs (14,16). c-fos expression, for example, is regulated by all of the MAPK pathways including the ERK (19), JNK (20), and p38 pathways (12). c-Jun, on the other hand, is phosphorylated at Ser-63 and Ser-73 by JNK (21,22). The activated c-Jun can then autoregulate its own expression through a c-Jun/AP-1 enhancer element in its promoter (23). p38 can also stimulate c-Jun expression by activation of the transcription factor ATF2 (24,25), which forms a heterodimer with c-Jun that acts on the AP-1 enhancer element in the c-jun gene (23). Moreover, p38 may activate transcription factors myocyte-enhancing factor 2C (MEF2C) and MEF2A, which transactivate c-Jun through a MEF-responsive element also present in the c-jun gene regulatory region (26 -29). Thus, AP-1 can be activated by all three MAPK pathways in a selective manner and serve as a common integrator of MAPK signaling to specific target gene expressions (14,16,30).
p38 activation is known to trigger pleiotropic biological effects, including cell death, differentiation, and proliferation, by mechanisms mostly unknown (1)(2)(3). The signal relay along the p38 pathway involves a kinase cascade, which generally consists of the upstream stimulators MAPK kinase 6 (MKK6) and 3 (MKK3), the p38 kinases (␣, ␤, ␥, and ␦), and the downstream effectors (3). p38 kinases downstream phosphorylate protein kinases including MAPK-activated protein kinase 2 and p38related/activated protein kinase as well as the nonkinase small heat shock protein 27 (Hsp27) and transcription factors such as AP-1 family proteins ATF2, c-fos, and MEF2 (3). p38␣, also called p38, was first cloned as a 38-kDa protein (31)(32)(33). Three other p38 isoforms (p38␤, -␥, and -␦) were isolated later and have more than 60% identity in sequence in comparison with p38␣. p38␣ and p38␤ are ubiquitously expressed, whereas the p38␥ and p38␦ products are only detected in certain cell types (3). The most important functions of the p38 pathway so far deduced were largely discovered by analyzing p38␣, despite the fact that each p38 family member may have distinct functions (3). p38␣, for example, is known to be essential for mammalian embryonic development as demonstrated by knock-out studies (34), is widely involved in cytokine production in response to inflammatory stimuli (32,33,35), and regulates cardiomyocyte differentiation/apoptosis (36,37). Whereas some of these functions have also recently been ascribed to p38␥ in some cell types (15,27,38), the contributions of other p38 family members remain mostly obscure. Since p38 family members are expressed differently in different cell types and tissues, and each of the family members may have distinct functions, we sought to test the hypothesis that the cell type-specific isoform expression profiles of p38 family members determine the biological effect of p38 pathway activation in that cell type. Our results demonstrated that p38␤ has an opposing effect to p38␥ and p38␦ in mediating MKK6 and sodium arsenite (ARS) signaling to activate AP-1 trans-activation of target genes through mechanisms involving c-Jun regulation. These results suggest that an upstream signal of the p38 pathway is interpreted by the p38 isoform expression pattern through differentially regulating AP-1 transcription factor activity, leading to a specific cellular outcome that is dependent on the spectrum of isoform expression but independent of the upstream pathway stimulator.

MATERIALS AND METHODS
cDNA Constructs and Expression Plasmids-pcDNA3-HA-MKK6 is a constitutively active mutant of MKK6 with Ser-207 and Thr-211 replaced by Glu (MKK6/2E) (39,40). This plasmid as well as the FLAGtagged wild type and dominant negative forms (AF or KM) of human p38␣, p38␤, p38␥, and p38␦ have been previously described (32, 39 -41). The AFs are p38 mutants that cannot be phosphorylated, since the TGY dual phosphorylation site has been changed to AGF, whereas the ki-nase-dead KM mutants were generated by a mutation of the ATPbinding site (Lys to Met (K to M)). Mutations were generated by PCRbased techniques using the QuikChange site-directed mutagenesis kit from Stratagene as described (40,41). A p38␦ double mutant (p38␦/AF) was created by substituting Thr 180 with Ala and Tyr 182 with Phe (underlined) using a PCR-based procedure (the primer sequence: GCA-GACGCCGAGATGGCTGGCTTCGTGGTGACCCGCTGG). pGEX for GST-ATF2-(1-109) was kindly provided by Dr. Roger Davis (25). pSV-GFP was purchased from Invitrogen. A full length of mouse c-Jun was cloned into a mammalian expressing vector pHM6-HA (Roche Molecular Biochemicals). An AP-1 luciferase construct (AP-1-Luc) was kindly provided by Craig Hauser, which was generated by cloning three AP-1 repeats into a luciferase reporter gene containing a minimal Fos promoter (42). The mouse VDR promoter (VDR-Luc, 0.5 kb) was cloned and inserted into a pGL2 basic vector (Promega) in front of the luciferase gene as previously described (43,44). A c-fos luciferase reporter (c-fos Luc, containing Ϫ356 to ϩ109 of the murine c-fos promoter) (45) and c-jun luciferase promoter (c-jun Luc, containing base pairs Ϫ225 to ϩ150 of the promoter) (26) have been previously described.
Other Reagents-Minimum essential medium, L-glutamine, and antibiotics were supplied by Invitrogen. Fetal bovine serum was obtained from BioWhittaker. DNA was prepared using an Endofree Kit from Qiagen. A DNA transfection kit (calcium phosphate) and a dual luciferase kit were purchased from Promega. Fugene 6 reagent for transfection was purchased from Roche Molecular Biochemicals. Glutathione-agarose beads were from Sigma. Proteinase inhibitors and other chemicals including ARS and anisomycin (ANI) are from Sigma. Protein G-Sepharose 4B and protein A-Sepharose 4B beads were purchased from Zymed. Rabbit antisera against p38␣, p38␤, p38␥, and p38␦ have been previously described (38). Anti-phospho-c-Jun (Ser-63) and anti-HA antibody were purchased from Cell signaling and Roche Molecular Biochemicals, respectively. Anti-FLAG M2 affinity gel and anti-FLAG mouse monoclonal antibody were from Sigma. [␥-32 P]ATP was from Amersham Biosciences. The GST-ATF2-(1-109) was prepared by affinity chromatography over GSH-agarose beads (Sigma).
Cell Culture, Transfection, and Cell Proliferation and Luciferase Assay-Human breast cancer cell line MCF-7 and MDA-MB-468 were obtained from ATCC and maintained in minimal essential medium containing 10% fetal bovine serum and antibiotics at 37°C with 5% CO 2 . Mouse fibroblasts NIH 3T3, the Ras-transformed counterpart, and mammary carcinoma cells EMT-6 have been previously described (11,46). The wild type and c-Jun knockout MEF were kindly provided by Ron Wisdom (47). For promoter analyses, the protocol of calcium phosphate-mediated transfection from Promega was followed. To increase transfection efficiency, Fugene 6 was used for cell proliferation, kinase assay, and MKK6-p38s binding. The luciferase construct (AP-1-Luc or VDR-Luc) or FLAG-tagged p38 (wild type or the AF or KM mutant p38␣, p38␤, p38␥, and p38␦) was expressed at a 1:1 ratio with vector or the active MKK6 (MKK6/E). For cell proliferation, MCF-7 cells were transfected with p38␤ or p38␥ together with a marker plasmid pSV-GFP and, after 48 h, were pulse-labeled with 5-bromo-2Ј-dexoyuridine (BrdUrd) (Roche Molecular Biochemicals) and fixed in 4% paraformaldehyde. The GFP-positive and BrdUrd-positive cells were counted under fluorescence microscope (Leica). For luciferase assay, cells were collected 48 h later in the lysis buffer, and the luciferase activity of the promoter was assayed with a dual luciferase kit from Promega by using pRL-TK (encoding Renilla luciferase) as a normalization control in a TD-20/20 Luminometer (Turner Designs). To assess ARS-induced AP-1 activity, cells were treated with 2 mM ARS for 30 min 24 h before the luciferase assay. The results from at least three separate experiments were analyzed with Student's t test for the statistically significant difference.
Western Blotting Analyses-For Western blot analyses, cells in good growth condition were lysed in modified radioimmune precipitation 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). Protein concentration was determined by a DC Protein Assay kit (Bio-Rad). Typically, 50 g of protein was separated on a SDS-PAGE, which was transferred to a nitrocellulose membrane for detection of the molecule of interest, using ECL (Amersham Biosciences) as previously described (48). For detection of endogenous or FLAG-tagged p38s, the experimental conditions of previous publications (38,49) were followed by using either p38 isoform-specific or anti-FLAG antibodies. The different dilutions of the polyclonal rabbit anti-p38 antibodies have been previously titrated and shown to yield similar binding affinities to their respective proteins (38). The antibody dilutions used in this work were as follows: anti-p38␣, 1:5,000; p38␤, 1:5,000; p38␥, 1:2,000; p38␦, 1:1,000. Under the these experimental conditions, the dilutions of the antibodies showed similar affinities toward their respective bacterially expressed recombinant p38 isoform proteins with little cross-reactions (data not shown).
Immunoprecipitation and Protein Kinase Assay-Following transfection, cells were collected by trypsinization, and pellets were quickfrozen in liquid nitrogen after washing and spin. For ARS-induced p38 activation, cells transiently expressing different FLAG-tagged p38s were treated with 2 mM ARS for 30 min, followed by cell collection in the lysis buffer. The cell pellet was either stored at Ϫ80°C or used immediately for protein preparation (resuspended in 100 l of the lysis buffer and incubated on ice for 30 min with vortexing for 30 s every 10 min). Protein (200 g) was incubated with 20 l of anti-FLAG M2 affinity gel (Sigma) for FLAG-tagged p38 at 4°C in a rotating plate overnight. The precipitates were washed two times with respective lysis buffer and two times with kinase binding buffer (20 mM HEPES, pH 7.6, 50 mM NaCl, 0.05% Triton X-100, 0.1 mM EDTA, 2.5 mM MgCl 2 ). The kinase reaction was carried out at 30°C for 30 min in 25 l of kinase reaction buffer (20 mM HEPES, pH 7.6, 20 mM MgCl 2 , 15 M ATP, 20 mM ␤-glycerolphosphate, 20 mM p-nitrophenyl phosphate, 0.5 mM Na 3 VO 4 , 2 mM dithiothreitol) as described previously (11,48). [␥-32 P]ATP (5 Ci) and 1 g of GST-ATF2-(1-109) were used for each sample. Following a 30-min reaction at 30°C, an equal volume of 2ϫ Laemmli buffer was added to stop the reaction. The phosphorylated protein was separated in a SDS-PAGE. The gels were dried, scanned, and quantitated in a Phosphor-Imager (Amersham Biosciences). Each kinase assay, starting at transfection and ending with the determination of the protein kinase activity, was repeated two or three times with similar results.

Expression of Endogenous p38s in Human Breast Cancer Cells and Their Contribution to MKK6-induced AP-1 Activation-p38
kinase is known to stimulate AP-1 via activation of one or more its components (12,25,26). To assess the contribution of each p38 family member to transduction of AP-1 stimulatory signaling, Western blot analyses was carried out to determine the expression of endogenous p38 family members in the two human breast cancer cell lines, MCF-7 and MDA-MB-468 (468). Cell lysates were prepared and examined for p38 isoform expressions using the isoform-specific antibodies as previously described (38). A similar ECL staining intensity was shown for these antibodies toward the same amount of the purified recombinant p38 isoform proteins under the antibody dilutions utilized here (see "Materials and Methods"). The results of Fig. 1 (upper panel) reveal that all four p38 isoforms are present in MCF-7 and 468 cells, which are migrated differently on SDS-PAGE due to their different molecular weights (3,38). p38␣ displayed a major band at 38 kDa in these two human breast cancer cell lines, followed by p38␦ (39 kDa), p38␥ (43 kDa), and p38␤ (40 kDa). The higher level of p38␣ than other isoforms in both cell lines is consistent with the previous literature, indicating that p38␣ is the major predominant form of the p38 family (3). However, a considerable amount of the p38␦ and p38␥ proteins is also present in these cells, which isoforms are known to be selectively expressed only in certain tissues or cell types (3). This is perhaps the first report showing that proteins of all p38 family members are expressed in the same cell lines.
Our previous work demonstrated that expression of the p38 stimulator MKK6 in these breast cancer cells increases AP-1 trans-activity (50). Expression of all four p38 family members promoted us to determine which of these family members contribute to transduction of the MKK6 signaling to AP-1. The dominant negative mutant form of each p38 member was coexpressed with or without MKK6, together with AP-1-Luc (a minimal c-fos promoter containing an additional three AP-1 enhancer elements regulating the luciferase gene transcription (42). Since the AF mutants do not have the dual phosphorylation residue TGY and the KM mutant is inactivated in the ATP-binding site, both should function as a kinase-dead form of p38 to block the upstream signals, as previously demonstrated (51). Expression of these kinase-dead forms of p38 alone has different effects on AP-1 reporter activity, with a slight stimulation by p38␣/AF and p38␤/AF but not by p38␥/AF and p38␦/KM (data not shown). All four mutant forms, however, are able to block MKK6-induced AP-1 activity (Fig. 1, bottom panel). The strongest inhibitions were observed for p38␥/AF (p Ͻ 0.01 versus MKK6 alone) and the p38␦/KM (p Ͻ 0.01), followed by a moderate suppression of MKK6-induced AP-1 activity by p38␤/〈F and the p38␣/AF (p Ͻ 0.05 in both cases). Application of p38␦/AF also achieved a significant inhibition of the AP-1 activation by MKK6, as in the case of p38␦/KM (p Ͻ 0.02, data not shown). Of interest, the inhibitory activity of these dominant negatives on MKK6-induced AP-1 activity does not correlate with the expression levels of their endogenous proteins. For example, p38␤ and p38␥ are expressed at a similar level, yet transfection of the same amounts of p38␥/AF and p38␤/AF shows that the p38␥/AF is much more efficient in inhibition of MKK6 stimulation of AP-1. These results suggest that each of these family isoforms may have different roles in mediating MKK6 signaling to AP-1.
p38␤ Potentiates but p38␥ and p38␦ Inhibit or Have No Effects on MKK6-and ARS-induced AP-1 Activation-MKK6 is a specific upstream kinase that activates all four family members of p38 (1,3,25,39). To further evaluate roles of each p38 family member in transduction of signaling to activate AP-1dependent transcription, each of the wild-type p38 family members were individually expressed with or without the active MKK6, and their effects were determined on the stimulated AP-1 reporter activity (Fig. 2). Transfection of either wild-type p38␣ or p38␤ but not p38␥ and p38␦ alone increases the basal AP-1-Luc activity about 5-fold (p Ͻ 0.05). To our surprise, whereas co-expression of wild-type p38␣ had no effect on the MKK6-induced AP-1 activity, compared with either MKK6 or with p38␣ alone, although its endogenous content is most predominant. Unexpectedly, transfection of the same amount of wild-type p38␤ synergistically increased the AP-1 activity by more than 30-fold (p Ͻ 0.05 versus MKK6). In contrast, both p38␥ and p38␦ significantly inhibited the stimulation by MKK6 (p Ͻ 0.01, both versus MKK6 alone in 468 cells), indicating an opposing effect of p38␤ with p38␥ and p38␦. The enhancing effect of p38␤ is kinase-dependent, since this effect is not seen with the phosphorylation-dead form p38␤/AF (Fig. 1), but the inhibition by p38␥ and p38␦ occurs regardless whether the mutants (Fig. 1) or the wild-type forms of the p38s (Fig. 2) are applied.
To confirm this result independently of MKK6 overexpression, ARS, a well known chemical p38 activator (11,(52)(53)(54), was utilized. Our previous work demonstrated that ARS stimulates AP-1 trans-activity and AP-1 binding in these breast cancer cells (50). Different p38 isoforms were expressed in 468 cells, which were then treated with 2 mM ARS for 30 min 24 h before the assessing of AP-1 luciferase activity. As for MKK6 activation, ARS treatment alone increased AP-1 activity of about 2-fold, but this stimulation was increased to 10-fold when p38␤ was also transfected into the cells (p Ͻ 0.01 versus ARS alone). Expression of other p38␥ and p38␦, however, failed to have a significant impact on the ARS-induced AP-1 activation, implying that these isoforms play a different role in transmitting MKK6 and ARS signaling to AP-1 (Fig. 2, upper panel). The results suggest that p38␤ is the only p38 family member to transmit both MKK6 and ARS signal to AP-1. To determine whether the opposing effect of p38␤ with p38␥ and p38␦ is cell type-specific, another human breast cancer cell line, MCF-7, which has a similar expression profile of p38 family members ( Fig. 1), was examined for effects of p38 isoform expression on MKK6 activation of AP-1 activity. As shown in the bottom panel of Fig. 2, expression of p38␤ again increased MKK6 stimulation of AP-1 from 3.2-to 16.5-fold (p Ͻ 0.01 versus MKK6 alone). Expression of p38␥ or p38␦, on the other hand, failed to inhibit the AP-1 stimulation by MKK6 (p Ͼ 0.05, both versus MKK6 alone). Together, these results demonstrate that at least in these two human breast cell lines in which all of the isoforms are expressed, p38␤ increases MKK6 and/or ARS signaling to stimulate AP-1-dependent transcription, whereas p38␥ and p38␦ are either inhibitory or have no effects on the AP-1 activation by MKK6 and/or ARS.
Opposing Effects of p38␤ with p38␥ and p38␦ on MKK6mediated VDR trans-Activation-AP-1 is a key transcription factor that converts MAPK signaling into target gene expression (14,16). Our previous work has established that MKK6 trans-activates VDR in these human breast cancer cells in a manner dependent on c-Jun/AP-1 activity (50). It is therefore critical to determine whether the opposing effects of p38␤ with p38␥ and p38␦ also occur in regulation of this p38/AP-1 target gene transcription. A reporter plasmid containing 0.5 kb of the mouse VDR promoter (43,44) in front of the luciferase gene was co-transfected with each isoform of the p38 family members in the presence or absence of the active MKK6, and the VDR-luciferase activity was assessed. As illustrated in Fig. 3 (upper panel) in 468 cells, the VDR promoter activity conferred by MKK6 expression was significantly increased from 2.2-to 11-fold by p38␤ (p Ͻ 0.05, versus either MKK6 or p38␤ alone). Expression of each of the other family members (p38␣, p38␤, and p38␥ with MKK6), on the other hand, reduced the MKK6 stimulation of the VDR gene promoter (p Ͻ 0.05, the p38 isoform and MKK6 combination versus MKK6 alone). These results further support an opposing role of p38␤ with p38␥ and p38␦ (probably also p38␣) in regulation of an AP-1-dependent VDR gene transcription by MKK6. When the same experiment was carried out in MCF-7 cells (Fig. 3, bottom panel), an inhibitory effect was similarly observed for p38␥, p38␦, and p38␣ (p Ͻ 0.05 versus MKK6 alone) but not for p38␤ on the MKK6 activation of the VDR promoter. The stimulation of AP-1 and VDR-Luc activities by expression of p38␣ or p38␤ alone observed in the estrogen receptor-negative 468 cells but not in estrogen receptor-positive MCF-7 cells (Figs. 2 and 3) implies either a cell line-specific phenomenon or estrogen receptorrelated effect. The enhancing effect of p38␤ on MKK6 activation of AP-1 (Fig. 2) but not VDR promoter (Fig. 3) activity in MCF-7 cells may suggest that p38␤ can mediate MKK6 signaling to VDR via AP-1-independent mechanisms in this cell line. Nevertheless, a clearly different effect is observed for p38␤mediated transduction on MKK6 stimulation of VDR transcription than from the rest of other three p38 isoforms in both human breast cancer cell lines.
Expression Profiles of Endogenous p38 Family Members Determine a Specific Response of a p38-activating Signal-Our results of the antagonizing activities of p38␤ with p38␥/␦ in MKK6 activation of AP-1/VDR suggest that effects of p38 activation on AP-1-dependent transcription are determined by the expression pattern of endogenous p38 isoforms. p38 activation would be stimulatory to the AP-1-dependent transcription if p38␤ is the major form, whereas it would be inhibitory in cells predominantly expressing p38␥ and/or p38␦. To directly test this possibility, a series of cell lines were screened for expression of p38 isoforms by Western blot analyses as described above. As shown in Fig. 4A, the p38␤ protein level is about 2 times higher in mouse NIH 3T3 cells compared with mouse mammary carcinoma EMT-6 cells, and p38␦, on the other hand, is only detected in EMT-6 but not in the other cell lines screened. Exactly as predicted, MKK6 expression in 3T3 cells significantly activates AP-1 (Fig. 4B) and VDR promoter activity (Fig. 4C, p Ͻ 0.05 in both cases), but the same MKK6 transfection in EMT-6 cells has no effects on the basal AP-1 activity (Fig. 4B) and inhibits the VDR promoter activity by more than 50% (p Ͻ 0.05, Fig. 4C). These results provide direct evidence that the endogenously expressed p38 isoforms may determine the final outcome of the p38 activation signaling on AP-1-dependent gene transcription.
Increase by p38␤ but Suppression by p38␥ of Cell Proliferation by MKK6 -In addition to regulation of target gene expression, AP-1 activity can be either growth-inhibitory or growthstimulatory by itself and/or as a result of integration of the effects on multiple cellular AP-1-dependent target genes (14). To examine whether the AP-1 activity in our analyses corresponds to a proliferative or growth-inhibitory signal, cell proliferation assays were performed in which p38␤ or p38␥ was co-expressed with MKK6 in MCF-7 cells, together with a marker plasmid pSV-GFP. Following transfection, cells were pulse-labeled with BrdUrd and examined for BrdUrd-positive cells in the transfected population (GFP-positive, about 10 -20%) under a fluorescence microscope. Expression of MKK6 or p38␤ or p38␥ alone had no significant effect on BrdUrd labeling over vector control (Fig. 5). Co-expression of MKK6 with p38␤, however, increased the labeling, and that with p38␥ decreased BrdUrd incorporation (p Ͻ 0.05 for MKK6 plus p38␤ or MKK6 plus p38␥ versus MKK6 alone). The slight inhibition of BrdUrd incorporation by MKK6 alone (26%, p Ͼ 0.05 versus control) may reflect the net effects of transfected MKK6 on the four p38 endogenous isoforms shown to be expressed in these cells. The lack of significant growth-regulatory effects of p38␤ or p38␥ by itself suggests that these activities require activation by upstream signals such as MKK6. The significant increase of growth promoted by co-transfection of MKK6 plus p38␤, and the growth inhibition by co-transfection of MKK6 plus p38␥ indicates that the positive or negative growth regulatory effects of the isoforms can be observed when these proteins are overproduced on a background in which all four isoforms are detected. Furthermore, in order to see an effect, the p38 pathway must be activated by an upstream activator such as MKK6. These growth-regulatory activities correlate with the changes of AP-1 transcriptional activity observed in these human breast cancer cells (Fig. 2). Although alterations of AP-1 activity may correspond to either cell proliferation, differentiation, or apoptosis, depending on cellular contents, our results suggest that the activity corresponds to cell proliferation under our experimental conditions.
Roles of the p38 Kinase Activity and MKK6-p38 Binding in the Opposing Effects of p38␤ and p38␥/␦-The Thr-Gly-Tyr (TGY) dual phosphorylation motif presents in all four members of the p38 family (3,41). So far, all regulatory and biological activities of p38s have been ascribed to the tyrosine and threonine phosphorylation on this motif (3). The p38 activator MKK6 has been reported to phosphorylate and activate all p38 family members (1,25,39). We wished to determine whether the opposing effects between p38␤ and either p38␥ or p38␦ are due to difference in isoform kinase activations by MKK6/ARS in these two human breast cancer cell lines. FLAG-tagged p38s were expressed with or without the active MKK6 in these breast cancer cells. Confirmation of their expressions was determined by immunoprecipitation and Western analysis using anti-FLAG antibody. Results in Fig. 6A showed that all transfected p38 isoforms are expressed in both 468 and MCF-7 cells. The similar levels of these four exogenous p38s in 468 cells and of transfected p38␣, -␤, and -␥ in MCF-7 cells demonstrated that the opposing effects of p38␤ with p38␥/␦ are not due to the differences in expression of these transfected isoforms. The higher level of p38 expression in MCF-7 cells, on the other hand, most likely results from higher transfection efficiency in these cells.
Activation of the transfected p38 isoforms by MKK6 and ARS was examined next. Cells were cotransfected with the active MKK6 and p38s, and 48 h later, the FLAG-p38s were immunoprecipitated with anti-FLAG antibody, and their in vitro kinase activity was determined using recombinant GST-ATF2 as the substrate. The kinase activity of p38␣ and p38␤ was most strongly activated by MKK6 in MCF-7 cells compared with their respective controls (Fig. 6B, top panel) in a manner similar to that observed in COS-7 cells (55), whereas MKK6 stimulated p38␦ and p38␥ activations to a lesser extent at similar expressed levels. Of interest, in 468 cells, p38␥ was most highly activated by MKK6, followed by p38␣, p38␤, and p38␦. Whereas previous in vitro data indicated that p38␥ can not phosphorylate ATF2 (49), the results in COS-7 (55) as well as in these 468 human breast cancer cells suggest that ATF2 can be phosphorylated by p38␥. Co-expression of the mutant forms of p38s (p38␣/AF, p38␤/AF, p38␥/AF, and p38␦/KM) with MKK6 did not reveal any kinase activity compared with vector controls (data not shown). These results demonstrated that MKK6 activates ATF2 phosphorylation by all four forms of p38s in both breast cancer cell lines, with p38␣ and p38␤ more strongly in MCF-7 cells but p38␥ more efficiently in 468 cells. In none of these cases, however, does the kinase activity of p38s stimulated by MKK6 offer an explanation for their effects on the AP-1-dependent transcription. Whereas p38␣ is mostly strongly activated by MKK6 in MCF-7 cells, for example, it has no obvious effect on the induction of AP-1 or VDR promoter activity. This result further consolidates our conclusion that the effect of p38s on transduction of MKK6 signaling to AP-1 is isoform-specific and does not correlate with the observed differences between their MKK6 activations as measured against the p38 substrate ATF2.
To further assess the role of the kinase activity of p38s in ARS-induced AP-1 activation, groups of cells transfected with FLAG-tagged p38 isoforms were also treated with ARS, the FLAG-tagged p38s were isolated by immunoprecipitation, and their in vitro kinase activity was determined as above. Results shown in Fig. 6B (bottom panel) demonstrated that ARS activates all p38s similarly in 468 cells, whereas the effect on p38␣ and p38␤ was greater than p38␥ and p38␦ in MCF-7 cells, similar to the activation by MKK6 (middle panel). Once again, none of these kinase activities explain the opposing effects of p38␤ with p38␥ and p38␦ in AP-1-dependent transcription. Since the kinase-dead form of p38␤ does not increase MKK6induced AP-1 transcription, whereas the dominant negatives of either p38␥ or p38␦ may inhibit the AP-1 stimulation by MKK6 (Fig. 1), these results suggest that p38␤ mediates MKK6 and/or ARS signaling to AP-1 dependent on its kinase activity, whereas p38␥ and p38␦ inhibit the AP-1-dependent transcription in a manner independent of kinase activity.
Enzyme binding to a substrate is required for MKK6 activation of the p38s (56,57). To examine whether the opposing effects may be due to different associations of the p38 isoforms with MKK6, an in vivo binding experiment was performed by co-expression of HA-MKK6 with wild-type and dominant negative FLAG-p38s in MCF-7 cells. Transfected p38s were isolated by immunoprecipitation with anti-FLAG antibody and examined for the presence of co-precipitated HA-MKK6. As in Fig. 7A, HA-MKK6 was detected in every group whether the wild-type or dominant negative p38 isoforms were transfected. If the FLAG-p38 isoform was normalized by co-precipitated HA-MKK6 in each of the transfections (p38/MKK6 ratio constant), there appeared to be more p38␣ and p38␤ binding to MKK6 with both wild-type and the dominant negative p38 mutant forms (Fig. 7A). These results may explain why the in vitro kinase activity of both p38␣ and p38␤ is relatively higher in these cells after co-expression with MKK6 (Fig. 6B), but once more this result does not correlate with their effects on AP-1dependent transcription.
Contributions of c-Jun Transcription and c-Jun Phosphorylation to the Opposing Activities-AP-1 activity most frequently consists of a heterodimer of c-Jun and c-Fos, both of which can Wild-type p38s were expressed alone, co-expressed with MKK6, or treated with ARS 24 h after the transfection, as described in the legend to Fig. 2. Transfected p38s were isolated by anti-FLAG immunoprecipitation and assessed for their in vitro kinase activity against ATF2. The results at the top represent the basal activity of transfected p38 (control) in comparison with p38 stimulated by MKK6 overexpression (middle) and by ARS (bottom). A similar result was obtained from a separate experiment.
be activated by p38 kinases (3,14,16). In order to dissect whether c-Fos or c-Jun is involved in p38 isoform-specific regulations of AP-1 activity by MKK6, the luciferase activity driven by either the c-fos (45) or c-jun (26) promoter was analyzed. Results in Fig. 7B showed that MKK6 alone selectively trans-activates c-Jun (left) but not c-Fos (right). These results are consistent with our published data that c-Jun but not c-Fos or ATF2 is the major component of the AP-1 activity induced by MKK6 in these breast cancer cells (50). Of interest, co-transfection of p38␥ or p38␦ inhibited the c-jun promoter activity induced by MKK6, whereas p38␤ appears to increase the stimulation. These results indicate that the opposing effects of p38␤ with p38␥ and p38␦ correlate with their different effects on MKK6-induced c-jun transcription.
Alterations in c-jun transcription by p38 isoforms in combination with MKK6 could contribute to the opposing effects on AP-1 activity. The p38 isoform-specific effect on c-jun transcription may be a result of their effects on regulation of other AP-1 components or regulators such as MEF2s (26,29). Alternatively, it could occur as a consequence of AP-1 regulation via the c-Jun/AP-1 enhancer element in c-jun promoter by its positive autoregulatory loop (23). Besides transcriptional regulation, c-Jun phosphorylation is the second major mechanism by which c-Jun/AP-1 is activated by MAPKs, especially by JNK (14,16). Whereas p38s were reported by several investigators not to phosphorylate c-Jun in vitro (32,41,58), recent studies suggest that immunoprecipitated and transfected human p38 (59) or murine p38␦ in mammalian cells (60) phosphorylates c-Jun in an in vitro kinase assay.
We have previously demonstrated that adenovirus-mediated MKK6 gene delivery in these breast cancer cells phosphorylates endogenous c-Jun (50). To determine whether p38 isoforms may differently regulate c-Jun phosphorylation by MKK6, HA-c-Jun was transfected with MKK6 in the absence or presence of wild-type or dominant negative p38s. Following immunoprecipitation with anti-HA antibody, c-Jun phosphorylation status was examined by Western blot using specific anti-phospho-c-Jun (Ser-63) antibody, and the same membrane was stripped off and reprobed with anti c-Jun antibody. As shown in Fig. 7C, MKK6 induced higher levels of both c-Jun and phosphorylated c-Jun, the later consistent with the observation with the adenovirus infection (50). Of great interest, wild-type p38␤ increased phosphorylated c-Jun, but c-Jun levels were equal (p38␤ plus MKK6 versus MKK6 alone). Both p38␥ and p38␦, on the other hand, inhibited the MKK6 induction of c-Jun as well as phosphorylated c-Jun levels. The decrease of c-Jun phosphorylation by either p38␥ or p38␦ in this case may represent a consequence of inhibition of the total c-Jun protein. These results indicate that p38␤ may increase the AP-1 activity predominantly by stimulation of the c-Jun phosphorylation, whereas p38␥ or p38␦ may inhibit primarily by suppression of the c-Jun trans-activation by MKK6. The moderate inhibitory effects on c-Jun phosphorylation and to a lesser extent c-Jun levels by each of the dominant negative p38 isoforms, on the other hand, may explain their moderate inhibitory activities on MKK6-induced AP-1 activation (Fig. 1). The different c-Jun phosphorylation is not due to the potential presence of active JNK or p38 in the anti-HA complex, as detected with phosphor-JNK or phosphor-p38 antibody (data not shown). Since c-Jun but not c-Fos or ATF2 is the major component of MKK6-induced AP-1 activity in these cells as shown by Western blotting and gel retardation (50), these results strongly suggest that the p38 isoform-specific regulation of c-Jun represents one mechanism for the opposing effects of p38␤ with p38␥ and p38␦ on AP-1-dependent transcription.

Correlations of Expression of Endogenous p38 Isoforms with Endogenous c-Jun Expression and Phosphorylations-
The c-Jun regulation by MKK6 and p38s may provide an important mechanism for the opposing effects of p38␤ with p38␥/␦ in AP-1-dependent transcription. Results obtained above, however, are based on experiments with co-transfection and overexpression and consequently need to be further confirmed in physiologically relevant conditions. Mouse NIH 3T3 and EMT-6 cells were applied here again to assess the total and phospho-c-Jun levels by p38 activation by virtue of their distinct expression pattern of endogenous p38 isoforms. NIH 3T3 cells contain higher concentrations of p38␤, and p38␦ is only detected in EMT-6 cells (Fig. 4). If these two p38 isoforms indeed contribute to regulations of the endogenous c-Jun expression and phosphorylations, the total and phosphorylated c-Jun should be higher in NIH 3T3 than in EMT-6 cells. Total and phospho-c-Jun levels were determined in NIH 3T3 and EMT-6 by Western analysis. Exactly as expected, the level of total c-Jun protein is about 3 times higher in NIH 3T3 cells over that in EMT-6 cells (p Ͻ 0.01, Fig. 8, A and B). These results thus consolidate our conclusion obtained above with the HA-MKK6 was cotransfected with each FLAG-tagged p38 in MCF-7 cells, and the transfected p38s were purified by anti-FLAG immunoprecipitation. The precipitates were examined for the presence of HA-MKK6 by Western analysis with anti-HA antibody (bottom) and FLAG-p38s (top). B, selective activation of c-jun transcription by MKK6 and its regulation by p38 isoforms. c-jun Luc containing residues Ϫ225 to ϩ150 of the c-jun promoter (left) and c-fos Luc containing residues Ϫ356 to ϩ109 of the c-fos promoter (right) were cotransfected with MKK6 in the presence or absence of p38 isoforms, and the luciferase activity was assessed as above. The open and filled columns indicate transfections without and with MKK6, respectively. Results of one representative experiment from two similar experiments are shown. C, increase of MKK6-induced c-Jun phosphorylation by p38␤ and its inhibition by p38␥ or p38␦. The active MKK6 was cotransfected with HA-c-Jun in the absence or presence of wild type or dominant negative p38 isoforms. Transfected c-Jun was isolated by anti-HA immunoprecipitation, and its phosphorylation status was determined by Western blot using antiphospho-c-Jun (Ser-63) antibody. The same membrane was stripped off and reprobed with anti-HA antibody for c-Jun expression (bottom). c-jun promoter analysis (Fig. 7B) that p38␦ is also suppressive to endogenous c-Jun expression. More importantly, after treatment with anisomycin (10 g/ml for 30 min), a well known p38 stimulus that activates all four p38 isoforms (41,55), phosphorylated c-Jun is only detected in NIH 3T3 cells in which p38␤ is higher and p38␦ is undetectable (Fig. 8C). ANI also increases the phospho-p38 but does not increase total c-Jun protein expression in both cell lines (data not shown). These results thus provide strong evidence supporting the notion that the expression profiles of p38 isoforms regulate the level of endogenous c-Jun expression and phosphorylated c-Jun. DISCUSSION The p38 MAPKs are universal signaling cascades that transmit extracellular signals into target gene expressions through their regulation of transcription factor activities. Whereas various mechanisms for signaling specificity such as complex formation and selective recognition of specific docking sites and the activation the T-loop of the p38s (56, 61) have been proposed, it remains unclear why the same p38 activation in different cellular contents triggers different biological response (1)(2)(3)(4). Here our results demonstrate that upstream stimulatory signals of the p38 pathway may induce different cellular outcomes, depending on the expression profile of p38 isoforms and in a manner independent of the upstream pathway stimulator. This conclusion is supported by our finding that p38␤ increases, but p38␥ and/or p38␦ inhibit, AP-1-dependent transcription and cell proliferation induced by MKK6 in human breast cancer cells (Figs. 2, 3, and 5). A similar effect of the p38 isoforms on regulation of AP-1 activity was also observed with activation of the pathway by ARS. The stimulatory activity of p38␤ and the inhibition effect of p38␦ were further confirmed in NIH 3T3 and EMT-6 cells in which endogenous p38␤ protein level is higher in 3T3 than that in EMT-6 cells, respectively, whereas p38␦ is expressed in EMT-6 but not NIH 3T3 cells. Moreover, evidence is presented indicating that the enhancing effect of p38␤ requires the kinase activity, whereas the inhibition by p38␥ and p38␦ is independent of p38 phosphorylation.
These effects were further demonstrated by both transfection and in endogenous systems, at least in part, to be due to the isoform-specific regulations of c-jun transcription/expression and phosphorylations. The results thus show that specific outcomes of p38 MAPK activation in regulation of AP-1-dependent transcription and proliferation in a given cell type depend on the expression profile of p38 family members in those cells.
Different activities of p38 isoforms have been previously suggested, but the opposing effects within the same family members have not been described (3). p38␣ is the first isolated family member and has been mostly studied and characterized (3). It is not clear why its expression has the least effect on AP-1 stimulation by the upstream stimulators MKK6 and/or ARS. Published studies suggest that p38␤ may be mitogenic and/or antiapoptotic, whereas p38␥ may be involved in stress response. In HeLa cells, for example, adenovirus-mediated p38␤ delivery was demonstrated to protect SB202190-induced apoptosis (62). Furthermore, p38␤ but not p38␣ was shown to protect mesangilal cells from tumor necrosis factor-␣-induced apoptosis (63). Our results, however, showed that expression of either p38␤ or p38␥ alone has no substantial effects on cell proliferation, and their opposing effects only became obvious when their upstream activator, MKK6, is also co-expressed (Fig. 5). The role of p38␥ in stress response was, on the other hand, suggested by the observation in which inhibition of p38␥ by p38␥/AF expression suppressed ␥-radiation-induced G 2 arrest in human osteosarcoma U2OS cells, whereas inhibition of other family members by the dominant negatives had no effect (38). There are also, however, exceptions to this distinction; both p38␣ and p38␥ are involved in hypoxia-induced downregulation of cyclin D1 in PC12 cells (64). Green tea polyphenol, on the other hand, selectively stimulates p38␦ phosphorylation, but this effect is not associated with inhibition, rather activation of AP-1 activity in human keratinocytes as a result of simultaneous stimulation of the Ras/MEKK pathways (65). These results are consistent with the concept that each p38 FIG. 9. An experimental model describing a determinant role of expression profiles of p38 family members in p38 signal specificity. p38-activating signal (MKK6 or ARS) stimulates AP-1-dependent transcription through c-Jun regulation via p38␤, but p38␥ or p38␦ inhibits this process. p38-activating signal would be stimulatory to AP-1-dependent activities in cells expressing high levels of p38␤, but it is inhibitory in cells predominantly expressing p38␥ and/or p38␦. The net response in cells expressing all p38 family members is determined by integrations of the positive (p38␤) and the negative (p38␥ and -␦) AP-1 regulatory signaling. family member has a distinct function. However, whether these effects are mitogenic or proapoptotic may depend on the individual cellular content.
p38s are serine/threonine protein kinases, and the biological effects of p38 MAPKs have been so far exclusively ascribed to the kinase activities (1)(2)(3). Our result that the increase in c-Jun phosphorylation and AP-1-dependent transcription is observed with wild-type but not kinase-dead mutant p38␤ is consistent with this notion. Paradoxically, the kinase activities may not be required for the inhibitory effects of p38␥ and p38␦ on MKK6induced c-Jun phosphorylation and AP-1 activation, although the wild-type p38␥ and p38␦ appear to be more effective (Figs.  1, 2, and 7C). However, since all dominant negatives of p38s inhibit c-Jun phosphorylation (Fig. 7C), which more or less correlates with their inhibition on AP-1-dependent transcription (Fig. 1, bottom), the inhibitory effects by these mutants do not appear to be isoform-specific. These results suggest that p38␤ increases the AP-1 by stimulation of c-Jun phosphorylation, whereas p38␥ and p38␦ inhibit this by mechanisms involving suppression of the c-jun transcription. Mechanisms operative in this process are unclear at present but may involve p38 isoform-specific protein-protein interactions in which the wild-type ␥ and ␦ isoforms act as dominant negatives. Efforts in this research direction will facilitate identification of novel functions of p38 MAPKs.
The opposing activity of p38␤ to p38␥ and p38␦ in AP-1 trans-activation of transcription has important implications for understanding p38 signaling specificity. p38 activation would correspond to an increase in AP-1 activity if p38␤ is the dominant form. However, if p38␥ and/or p38␦ are the major isoforms, p38 activation would correspond to an inhibition of AP-1-dependent gene expression. AP-1 activity induced by p38 activation in a cell type in which all four p38 family members are expressed, as in the case of 468 and MCF-7 cells, would represent the net output of the signal integration among its family members (Fig. 9). AP-1 plays a key role in regulation of many target genes whose expression levels control cell proliferation, cell differentiation, and apoptosis, including cyclin D1, c-Jun, p53, p16 ARF (14), and VDR (50). Since many cell lines only express some of p38 family members (Fig. 4) (3), the p38 isoform-specific effect on AP-1-dependent gene expression would provide an explanation as to why the same p38 activator could lead to various biological outcomes in different cell types.