v- src Induces Prostaglandin Synthase 2 Gene Expression by Activation of the c-Jun N-terminal Kinase and the c-Jun Transcription Factor*

A consensus cyclic AMP response element (CRE) in the murine prostaglandin synthase-2 (PGS2) promoter is essential for pgs2 gene expression induced by pp60 v- src , the v- src oncogene product. In this study, we inves-tigate (i) the transcription factors active at the PGS2 “CRE site” in response to v- src activation and (ii) the signal transduction pathways by which pp60 v- src activates these transcription factors. Transient transfection assays with pgs2 promoter/luciferase reporter chimeric genes suggest that c-Jun mediates v- src -induced pgs2 gene expression. Antibody supershift experiments demonstrate that c-Jun can participate in a complex with the pgs2 promoter CRE site. Moreover, in vitro immunocomplex assays demonstrate that pp60 v- src expression strongly activates c-Jun N-terminal kinase (JNK1) en- zyme activity. Serines 63 and 73, the sites of c-Jun phosphorylation by JNK, are essential for v- src -induced, pgs2 promoter-mediated luciferase expression. Cotransfection studies with plasmids expressing wild-type JNK, dominant-negative JNK, and dominant-negative MEKK-1 confirm that activation of the Ras/MEKK-1/ JNK/c-Jun pathway is required for v- src -induced pgs2 gene expression. Overexpression of either wild-type ERK-1 or ERK-2 proteins also potentiate v- src -mediated luciferase expression driven by the pgs2 promoter, and expression of dominant-negative mutants of ERK-1,

The prostaglandins play key roles in a variety of biological processes, including cell division, blood pressure regulation, immune responses, ovulation, bone development, wound healing, and water balance. Altered prostaglandin production is associated with several pathophysiological states, including bone resorption, cardiovascular disease, acute inflammation, atherosclerosis, and colon cancer (1). Prostaglandin synthase (PGS), 1 also known as cyclooxygenase, is the key enzyme in the conversion of free arachidonic acid to PGH 2 , the common precursor to all the prostaglandins, prostacyclins, and thromboxanes (2). Several laboratories have recently described the cloning of a second, inducible form of PGS. This gene, now referred to as pgs2 or cox2, was identified in differential library screens in fibroblasts transformed by the v-src oncogene (3), in fibroblasts treated with the tumor promoter 12-O-tetradecanoylphorbol-13-acetate (4), and in fibroblasts treated with plateletderived growth factor (5). Subsequent studies have shown that the pgs2 gene can be induced in a variety of cells, including macrophages, mast cells, epithelial cells, endothelial cells, neurons, smooth muscle cells, and ovarian granulosa cells. A variety of stimuli, including agents that act via G proteinmediated mechanisms, as well as the protein kinase C-mediated pathway activated by 12-O-tetradecanoylphorbol-13-acetate and the tyrosine kinase-mediated pathways activated by both growth factor receptors and pp60 v-src , the product of the v-src oncogene, can stimulate the expression of the pgs2 gene (reviewed in Ref. 6). However, the signal transduction mechanisms, the transcription factors, and the regulatory regions of the pgs2 gene necessary for PGS2 induction are not well understood.
pp60 v-src encodes a non-receptor tyrosine kinase (7). Several different approaches have been taken to try to elucidate how v-src transforms cells. These include identifying the substrates of the pp60 v-src tyrosine kinase, isolating v-src inducible genes, and studying the transcriptional regulation of known genes whose mRNA levels are elevated in response to v-src expression. Because of (i) the wide range of cells that express PGS2 and the diverse stimuli that modulate pgs2 gene expression, (ii) the great interest in therapeutic modulation of prostaglandin synthesis, and (iii) the great interest in the mechanisms of v-src-mediated gene expression and transformation, we have been investigating the cis-acting DNA elements, the transcription factors, and the signal transduction pathways that mediate v-src activation of the pgs2 gene. By using promoter deletions and site-directed mutagenesis in pgs2-luciferase chimeric reporter genes, we show that a CRE site located between nucleotides Ϫ56 and Ϫ52 of the pgs2 gene is essential for v-src induction (8). A dominant-negative cyclic AMP binding protein (CREB) mutant, M1, blocked v-src-mediated expression, supporting the conclusion that this CRE site is necessary for pgs2 expression induced by pp60 v-src (8). Finally, cotransfection of a dominant-negative Ras mutant blocked v-src-mediated induction of the pgs2-luciferase chimeric gene, suggesting that the pp60 v-src -initiated pathway of PGS2 activation is mediated by Ras (8). In this report, we identify c-Jun as a transcription factor active at the PGS2 CRE site following v-src expression and determine the signal transduction pathway that mediates the activation of this factor.

EXPERIMENTAL PROCEDURES
Plasmids-We previously referred to PGS2 as TIS10 (4,9). Our promoter constructs thus historically have the TIS10 designation (8,9). The reporter constructs pTIS10 Ϫ80 LUC and pTIS10 Ϫ40 LUC, as well as the v-src expression vector pMV-src were described previously (8). The overlapping CRE and E-box sites of the PGS2 regulatory region are located at nucleotides Ϫ56 to Ϫ48. Thus, the pTIS10 Ϫ80 LUC contains the overlapping CRE and E-box cis-acting sites of the pgs2 promoter, while the pTIS10 Ϫ40 LUC construct has these sites deleted. The plasmid pGal4TIS10 Ϫ40 LUC was constructed by ligating five GAL4 DNA binding sites to the 5Ј-end of the pTIS10 Ϫ40 LUC reporter. Expression vectors of pRSV-CREB, GAL4-CREB (a fusion protein containing the GAL4 DNA binding domain fused to CREB), and GAL4-M1 (a fusion protein containing the GAL4 DNA binding domain fused to the dominant-negative M1 CREB mutant) (10) were the gifts of Dr. Marc Montminy (Salk Institute). The expression vectors pSR␣MSVtkNeo-c-Jun and pSR␣MSVtkNeo-c-Fos (11) were the gifts of Dr. Charles Sawyers (UCLA). The expression vectors for GAL4DB, GAL4-c-Jun, and GAL4c-JUN 63/73 (a fusion protein containing the GAL4 DNA binding domain fused to a c-Jun protein in which serines 63 and 73 are substituted by leucines; Ref. 12) were gifts from Dr. Andrew Kraft (University of Alabama, Birmingham). Plasmid pRSV-ATF3, an expression vector that encodes ATF-3 (13), was the gift of Dr. Kenneth Low (New England Biolabs). Expression vectors for ATF-2 and GAL4-ATF2 (14) were the gifts of Dr. Michael Green (University of Massachusetts). pSR␣MEK⌬(K432M), an expression vector that encodes a dominantnegative MEKK-1, was from Dr. Michael Karin (University of California, San Diego) (15). pEVX-3RatK375A, an expression vector that encodes a dominant-negative Raf-1 (16), was from Dr. Susan Macdonald (ONYX, CA). pCDNA-Flag-JNK1 and pCDNA-DN-JNK1, expression vectors for Flag-tagged JNK1 and kinase-defective JNK1, respectively, and GST-c-Jun (amino acids 1-79) (15) were the gifts of Dr. Roger Davis (University of Massachusetts). The expression vectors pCEP4Erk1, pCEP4Erk2, pCEP4Erk1 K71R, and pCEP4Erk2 K52R, encoding wildtype ERK1, wild-type ERK2, dominant-negative ERK1, and dominantnegative ERK2, respectively, were the gifts of Dr. Melanie Cobb (University of Texas, Southwestern).
Cells and Cell Transfection-NIH 3T3 cells were grown and transfected as described previously (8). 3 g of various reporter constructs and 1.5 g of the v-src expression vector pMV-src or the empty expression vector pEVX per 60-mm dish were used for all experiments. The amounts of other constructs are specified in each experiment. The luciferase assay was performed as described previously (8). Triplicate dishes were used for all transfections, and all experiments were repeated at least twice.
Immunocomplex JNK Assay-The JNK immunocomplex kinase assay was performed as described (15). 6 g of plasmid encoding Flagtagged JNK1 was transfected into NIH 3T3 cells in 100-mm dishes, using the calcium phosphate method (8), along with either 4 g of v-src expression vector pMV-src or empty vector pEVX. The cells were kept in 0.5% newborn calf serum for 24 h after transfection, then lysed in lysis buffer (25 mM HEPES, pH 7.5, 10% glycerol, 1% Triton X-100, 137 mM NaCl, 2 mM EDTA, 25 mM sodium glycerophosphate, 1 mM sodium orthovanadate, 10 g/ml leupeptin, 1 mM phenylmethylsulfonyl fluoride). Extracts were pre-absorbed with Sepharose-4B beads (Pharmacia Biotech Inc.) for 30 min before the M2 monoclonal antibody (Eastman Kodak Co.) against Flag was added. A small sample of cell extract from each cell lysate was taken for protein assay to standardize for gel loading after the kinase reaction. M2 antibody was mixed with Sepharose beads for 1.5 h in a cold room, and extra antibody was washed away using lysis buffer. Then, the antibody-conjugated beads were added to the preabsorbed cell lysate for another 1.5 h. The binding mixtures were washed three times with lysis buffer and twice with kinase assay buffer (15). The kinase assay was carried out in a 30-l reaction mixture containing 20 mM HEPES, 20 mM MgCl 2 , 25 mM sodium glycerophosphate, 100 M sodium orthovanadate, 2 mM dithiothreitol, 20 M ATP, 2 g of GST-c-Jun, 10 Ci of [␥-32 P]ATP at 30°C for 30 min. The reactions were terminated by adding SDS loading buffer and incubating in boiling water for 5 min. The reaction mixtures were subjected to electrophoresis on SDS-polyacrylamide gel electrophoresis gels, which were then dried and exposed to x-ray film.
Electrophoretic Mobility Gel Shift (EMGS) Assays-Nuclear extracts were prepared from UP1A1 cells, a 3T3 cell line carrying a temperature sensitive v-src gene (8). The nuclear extract used for gel shift in this report was prepared from cells grown at the permissive temperature, where v-src is active. The gel shift and supershift protocols were as described previously (8), with the exception that the running buffer used in this study is composed of 10 mM HEPES (pH 8.0), 10 mM Tris, 5 mM EDTA, instead of Tris-glycine buffer. The c-Jun and CREB binding protein antibodies were purchased from Santa Cruz Biotech.

Overexpression of ATF Proteins
Blocks v-src-mediated Induction from the pgs2 Promoter-Deletion and mutation analysis of the pgs2 promoter identified the CRE site CGTCA located at nucleotide Ϫ56 of the murine pgs2 gene as essential for v-srcinduced pgs2 expression (8). EMGS supershift experiments with anti-CREB antibody demonstrated that CREB binds to this PGS2 CRE site (8). CREB-M1, a dominant-negative CREB mutant that binds to CRE sequences but has an inactive transactivation domain (10), blocks v-src-mediated PGS2 activation (8). We proposed that CREB and/or another member of the ATF transcription factor family might mediate v-src induction from the CRE site of the pgs2 promoter (8).
We first tested whether increasing the levels of wild-type CREB in cells could enhance v-src-mediated induction from the pgs2 promoter of the pTIS10 Ϫ80 LUC chimeric reporter gene. This construct contains nucleotides Ϫ80 to ϩ3 of the pgs2 promoter (we previously referred to PGS2 as TIS10 (4,9); our promoter constructs thus historically have the TIS10 designation (8,9)). However, overexpression of CREB, like overexpression of the dominant-negative CREB-M1 mutant, inhibits v-src-induced luciferase expression from the PGS2 reporter ( Fig. 1). Because we were concerned that high levels of CREB might "squelch" (17) CREB-dependent expression from the pgs2 promoter, we also used lower concentrations of wild-type CREB expression vector (data not shown). Enhancement of luciferase expression from the pgs2 promoter by CREB was never observed. We conclude that some transcription factor other than CREB is activated by pp60 v-src expression and elevates pgs2 gene expression by transcriptional activation from the PGS2 CRE site.
We next tested whether overexpression of other ATF transcription factors might modulate v-src induction from the pgs2 FIG. 1. CREB overexpression blocks v-src-mediated induction from the pgs2 promoter. We previously referred to PGS2 as TIS10 (4). Our promoter constructs thus historically have the TIS10 designation (8). 3 g of the pgs2 luciferase reporter pTIS10 Ϫ80 LUC, which utilizes nucleotides from Ϫ80 to ϩ3 of the murine tis10/pgs2 promoter to drive them to the luciferase reporter gene (8), were transfected into NIH 3T3 cells in 60-mm culture dishes, along with 1.5 g of the v-src expression vector pMV-src (ϩ symbols in the v-src row) or corresponding empty vector pEVX (Ϫ symbols in the v-src row). As indicated in the figure, 1 g of dominant-negative CREB (DN-CREB) expression vector, 1 g of CREB expression vector, or 3 g of CREB expression vector were cotransfected with the v-src and luciferase reporter vectors. The total amount of DNA in each reaction was kept constant by using the corresponding empty expression vectors. Three plates of NIH 3T3 cells were used for each transfection condition. Data are expressed as averages Ϯ S.D.
promoter. Overexpression of both ATF-2 and ATF-3, like CREB or CREB-M1 overexpression, also inhibited, rather than augmented, v-src-induced expression from the PGS2 CRE site (data not shown). We conclude that the ATF family members CREB, ATF-2, and ATF-3 do not mediate v-src induction of the pgs2 gene. The data presented in this section demonstrating that occupation of the PGS2 CRE site by CREB, DN-CREB, ATF-2, and ATF-3 block v-src-mediated pgs2 gene expression suggest (i) that this site is necessary for v-src-mediated induction of the pgs2 gene and (ii) that some transcription factor(s) other than these molecules must mediate this induction.
c-Jun Binds to the CRE Site of the pgs2 Promoter-EMGS and antibody-supershift experiments previously demonstrated that other, unidentified nuclear protein(s) in addition to CREB can bind to the PGS2 CRE site (8). c-Jun, a member of the bZIP transcription factor superfamily, can form heterodimers with members of the ATF family and bind to CRE sequences (18). Moreover, c-Jun⅐c-Fos heterodimers can also bind to CRE sequences (19). We next investigated whether c-Jun might play a role in the v-src-mediated induction of the pgs2 gene. We used antibody to c-Jun in a supershift experiment to determine if c-Jun is present in a complex that binds to the PGS2 CRE site.
An E-box overlaps the CRE site of the pgs2 gene and acts as an alternative site for binding of nuclear proteins (8). By adding E-box oligonucleotide competitor, we were able to reduce binding of proteins in nuclear extracts to the E-box sequence of the pgs2 promoter and enhance the binding of CREB and the additional nuclear factor(s) that binds to the PGS2 CRE site (8). The gel shift experiment shown in Fig. 2 was performed in the presence of E-box competitor to reduce (but not eliminate) the E-box complexes. The faster moving CRE site complex was previously identified, by supershift with anti-CREB antibody, as a CREB-containing complex (8). c-Jun antibody can recognize and supershift one of the slower moving complexes of the PGS2 CRE site (Fig. 2). In contrast, antibody to another protein (CREB binding protein) does not cause a supershift. c-Jun antibody alone does not form any complex with the CRE probe. The data demonstrate that the c-Jun protein is able to participate in a binding complex at the PGS2 CRE site.
v-src Activates JNK-Although the ERK1 and ERK2 MAP kinases can phosphorylate c-Jun in vitro (20), it is now clear that the recently cloned (21) JNKs are the kinases that generally phosphorylate and activate c-Jun in vivo (21)(22)(23). JNK1 and JNK2 are strongly activated by tumor necrosis factor ␣ (24), environmental stress (24), and UV light (22). If c-Jun activation plays a role in v-src-mediated PGS2 induction, we would expect that v-src expression would activate JNK. Flag epitope-tagged JNK1 was immunoprecipitated, with an anti-Flag monoclonal antibody, from NIH 3T3 cells cotransfected either with a vector encoding Flag epitope-tagged JNK1 and either the v-src expression vector pMV-src or the empty vector pEVX. Immunocomplex kinase assays were then performed, using a GST-c-Jun recombinant protein as substrate (15). Expression of v-src activates JNK1 kinase activity (Fig. 3).
The MEKK-1/JNK Signal Transduction Pathway Mediates v-src Induction of pgs2 Gene Expression-If v-src-mediated induction of PGS2 requires JNK activation and phosphorylation of c-Jun, we would expect that modulation of JNK levels in cells would alter the response of the pgs2 promoter to v-src expression. This is the case; overexpression of wild-type JNK1 potentiates v-src-mediated luciferase expression from the pTIS10 Ϫ80 LUC luciferase reporter gene (Fig. 4, left panel, lane  3), while co-expression of a kinase-defective, dominant-negative JNK1 attenuates v-src-mediated expression of luciferase from this reporter (Fig. 4, left panel, lanes 4 and 5). These data suggest that JNK activation plays an obligate role in v-srcmediated induction of gene expression from the pgs2 promoter.
We previously demonstrated that v-src-induced pgs2 gene expression required mediation by Ras (8). Ras activates several signal transduction pathways, each pathway leading to phosphorylation of distinct subsets of transcription factors (25). The downstream effector of Ras leading to activation of JNK enzyme activity and phosphorylation of c-Jun is the MAP kinase kinase kinase MEKK-1 (15). In contrast, Ras activation of the Raf-1 MAP kinase kinase kinase leads to phosphorylation of transcription factors such as TCF/Elk-1 and c-Myc (25). Expression of kinase-defective, dominant-negative MEKK-1 blocks v-src-mediated expression of luciferase from the pTIS10 Ϫ80 LUC luciferase reporter gene (Fig. 4, right panel). We conclude that activated transcription from the pgs2 gene following expression of pp60 v-src requires Ras activation of MEKK-1 and JNK, leading to phosphorylation of c-Jun, and subsequent increased transcription mediated by the CRE site of the pgs2 promoter.
The c-Jun Activation Domain Is Required for v-src-mediated Luciferase Expression from a gal4-pgs2 Chimeric Promoter-We wished to more fully explore the nature of the regulated activation domain(s) required to drive luciferase expression by the transcription factor complex assembled at the pgs2 gene in response to pp60 v-src . We replaced the CRE sequence in the pgs2 luciferase reporter with yeast transcription factor GAL4 v-src Induction of Prostaglandin Synthase 2 DNA binding sites (Fig. 5, top) and investigated what transactivation domains in chimeric GAL4 DNA binding proteins can mediate v-src activation of the minimal pgs2 promoter. No endogenous transcription factor in 3T3 cells will bind to the GAL4 sites.
A luciferase reporter gene containing only the first 40 nucleotides of the pgs2 promoter, TIS10 Ϫ40 LUC, expresses only minimal luciferase activity. This plasmid does not contain the overlapping CRE and E-box sites of the pgs2 promoter and is not responsive to v-src expression (8). If five GAL4 binding sites are added to this promoter, this gal4-pgs2 luciferase construct similarly expresses only minimal luciferase activity (  2). However, when pp60 v-src is also expressed, along with the GAL4-JUN protein, expression from the GAL4-TIS10 Ϫ40 reporter vector is substantially enhanced (Fig. 5, middle panel, lane 6; lower panel, lane 3). We conclude that the c-Jun activation domain, if positioned upstream of the minimal pgs2 promoter, can drive v-src-induced gene expression. In contrast, if the activation domains of ATF-2 or CREB are positioned adjacent to the minimal pgs2 promoter by the GAL4 DNA binding domain of GAL4-DB chimeric proteins, these activation domains are unable to mediate either basal or v-src-induced luciferase expression from the pgs2 promoter (Fig. 5, lower panel, lanes 5 and 6).
Phosphorylation of two serine residues, Ser-63 and Ser-73, is required for c-Jun activation and transcription from AP-1 sites (26,27). To determine if phosphorylation of these two sites is essential for v-src-induced expression from the pgs2 reporter, we used an expression vector in which the serines at these two sites have been mutated to leucines. GAL4-JUN 63/73, in which these sites of phosphorylation have been altered, is un-able to mediate v-src activation of the gal4-pgs2 luciferase reporter gene (Fig. 5, lower panel, lane 4). An intact c-Jun activation domain, with sites of phosphorylation available, is necessary for v-src-mediated transcriptional activation of the pgs2 promoter.
ERK-1 and ERK-2 Activation Also Participate in PGS2 Induction by v-src-Like JNK, the MAP kinase enzymes ERK-1 and ERK-2 can also be activated by pp60 v-src expression (28,29). We therefore asked whether these kinases might also mediate PGS2 induction by v-src. Co-expression of ERK-1 or ERK-2 potentiates v-src-mediated luciferase expression from the pTIS10 Ϫ80 LUC luciferase reporter gene (Fig. 6, left panel). Moreover, co-expression of kinase-deficient, dominant-negative forms of ERK-1 or ERK-2 substantially attenuate v-src-mediated luciferase expression from this reporter.
Raf-1 is the MAP kinase kinase kinase that mediates activation of the ERK enzymes by Ras (25). If the Ras/Raf-1/MEK/ ERK pathway plays a role in v-src induction of gene expression from the pgs2 promoter, one would expect that a dominantnegative Raf-1 mutation should also block this induction. This is the case; increased inhibition of v-src-mediated luciferase induction from the pgs2 promoter is observed as increasing amounts of a dominant-negative Raf-1 protein are expressed v-src Induction of Prostaglandin Synthase 2 ( Fig. 6, right panel).
c-Jun/c-Fos Overexpression Potentiates v-src-mediated Induction from the pgs2 Promoter-If c-Jun mediates v-src-induced pgs2 gene expression, one might expect that overexpression of c-Jun would augment luciferase expression from the pTIS10 Ϫ80 LUC reporter in response to v-src in a cotransfection assay. This is not the case in NIH 3T3 cells (Fig. 7). However, in contrast to overexpression of ATF proteins, which block v-src induction from the pgs2 promoter (Fig. 7), c-Jun overexpression does not interfere with v-src induction of pgs2 gene expression. Our data suggest that, while c-Jun participates in v-src-mediated induction of pgs2 gene expression, the amount of c-Jun in NIH 3T3 cells is not limiting in this induction. c-Jun can form heterodimers with members of the ATF family (18) and bind to CRE sequences. Moreover, c-Jun⅐c-Fos heterodimers can also bind to CRE sequences (19). We therefore examined the ability of these two transcription factors, alone and in combination, to modulate v-src induction from the pgs2 promoter. Like overexpression of c-Jun alone, overexpression of c-Fos alone neither enhances or attenuates v-src-induced luciferase expression from pTIS10 Ϫ80 LUC. However, co-expression of both c-Jun and c-Fos together potentiates v-src-induced luciferase expression from the pgs2 reporter. In contrast, as discussed above, the ATF proteins CREB and ATF-3 block v-src-mediated induction from the pgs2 promoter (Figs. 1 and 7). These data suggest that both components of the AP-1 transcription complex may participate in the v-src activated transcription of the pgs2 promoter from the CRE site.

DISCUSSION
Elevated prostaglandin production is a characteristic of Rous sarcoma virus-transformed fibroblasts (30). Persistent elevation of pgs2 gene expression in v-src-transformed cells is responsible for this increased prostaglandin production (3,31). A CRE at nucleotides Ϫ56 to Ϫ52 of the pgs2 promoter is the critical element in v-src-mediated activation of the pgs2 gene (8). pp60 v-src has been reported to modulate immediate-early gene expression through the serum response element of the egr1/tis8 gene (32), at a dyad symmetry element and a Sisinducible factor responsive element in the c-fos gene (33), through the CCAAT and TATAA elements of the junB gene (34), and via a v-src-responsive element of the 9E3/CEF-4 gene (35). However, v-src modulation of gene expression via a CRE has only been observed for the pgs2 gene (8).
c-Jun Transcriptional Activation Plays a Major Role in v-src-induced pgs2 Gene Expression-We anticipated that the CREdependent v-src transactivation of the pgs2 gene might involve c-Jun activation, since oncogenic forms of both Ras and Src can activate transcription via c-Jun phosphorylation (26,27). Our demonstration that v-src-induced activation of JNK activity correlates with PGS2 induction is consistent with this hypothesis. The ability of dominant-negative inhibitory forms of Ras, MEKK, and JNK, the mediators of the c-Jun phosphorylation/ activation pathway, to block v-src-induced expression from the pgs2 promoter also supports this proposal. The antibody supershift demonstration that c-Jun is a part of a protein complex that can recognize the PGS2 CRE also supports this hypothesis. Based on these data, we conclude that v-src activates the Ras/MEKK1/JNKK(SEK1)/JNK(SAPK) signal transduction pathway to phosphorylate c-Jun and stimulate transcription from the CRE of the pgs2 gene. v-src-mediated activation of c-Jun and pgs2 gene transcription is diagramed in Fig. 8. The c-Jun transactivation domain, when fused to a GAL4 DNA binding domain, can drive expression from a pgs2 promoter in which the CRE is replaced by GAL4 DNA binding sites. In contrast, ATF transactivation domains placed at this site are unable to mediate pgs2 gene expression. Moreover, this response is eliminated when the sites of JNK phosphorylation are mutationally altered in the c-Jun transactivation domain of the GAL4-JUN chimeric transactivator, clearly demonstrating that phosphorylation of the c-Jun transactivation domain by JNK plays a critical role in v-src-induced transcription from the pgs2 gene.
c-Jun is also phosphorylated following exposure of cells to inflammatory mediators such as tumor necrosis factor ␣ (24) and interleukin 1 (21). Tumor necrosis factor ␣ and interleukin 1 are also potent activators of prostaglandin production and induce the expression of the pgs2 gene in rat mesangial cells (38). Thus, c-Jun activation and the CRE of the pgs2 promoter may also play a role in mediating PGS2 induction by these inflammatory cytokines. UV irradiation is among the strongest JNK activators (21). It will be of great interest to determine whether UV radiation can induce pgs2 gene expression.
ATF-2 Does Not Mediate v-src-induced pgs2 Gene Expres- v-src Induction of Prostaglandin Synthase 2 sion-We thought c-Jun⅐ATF-2 heterodimer would be a likely candidate for the v-src-activated transcription factor active at the PGS2 CRE, since c-Jun⅐ATF-2 heterodimer mediates both virus activation at the human interferon 1␤ gene CRE (39) and E1A activation of c-Jun transcription at the c-Jun CRE (40). Moreover, ATF-2 is also a JNK substrate (41). However, wildtype ATF-2 protein inhibits PGS2 induction by v-src, suggesting that ATF-2, either as a heterodimer with c-Jun or as a homodimer, does not play a role in v-src-mediated induction of pgs2 expression. We obtained similar results with ATF-3 transfection, demonstrating that this member of the ATF family of transcription factors also does not play a role in v-src-mediated induction of pgs2 expression. The inability of GAL4-ATF2 fusion protein to mediate v-src induction from the gal4-pgs2 chimeric promoter further substantiates the conclusion that v-src-activated transcription at the pgs2 promoter does not involve ATF-2.
It is possible that another member(s) of the ATF family may act as a heterodimer with c-Jun to mediate v-src-mediated induction of the pgs2 gene. Protein X in Fig. 8 might thus be an as yet untested member of the ATF transcription factor family, forming a heterodimer with activated c-Jun. This possibility could be explored by similar transfection experiments with other members of the ATF family. Antisense experiments to inhibit the expression of members of the ATF family could also be used to further explore this question, as described by Du et al. (39).
c-Fos May Participate in v-src-mediated pgs2 Gene Expression-c-Fos⅐c-Jun (AP-1) heterodimers are able to bind CRE sequences (19,42). Moreover, c-Fos⅐c-Jun heterodimers can modulate transcription from a CRE (43). c-Fos overexpression can cooperate with c-Jun overexpression to cause increased transactivation at the pgs2 promoter CRE in response to pp60 vsrc (Fig. 7), suggesting that c-Jun⅐c-Fos heterodimer may play a role in pgs2 gene activation. We have no evidence that v-src can cause activation of pre-existing c-Fos protein, as required for immediate-early gene expression (44). However, ligand and oncogene activation of a c-Fos kinase(s) distinct from either JNK or ERK kinases has been described (37,45). Moreover, c-Fos phosphorylation by this kinase activates c-Fos transcriptional activity (45). Thus, it is possible that c-Jun/c-Fos transcriptional activity at the PGS2 CRE may be directly activated by parallel protein kinase pathways activated by v-src. If this is the case, protein X in Fig. 8 would be a c-Fos protein, phosphorylated by one of the recently described c-Fos kinases. It will be necessary to examine the consequences of modulating Fos ki-nase activity on v-src-mediated induction of pgs2 gene expression, once this kinase(s) has been cloned, to determine if direct activation of c-Fos plays a role in this v-src-induced response. We suggest that v-src may be a powerful activator of immediate-early gene transcription because it can activate a variety of kinase-mediated signal transduction cascades, leading to phosphorylation and transcriptional activation of a number of preexisting transcription factors.
The Ras/Raf/ERK Pathway Plays a Role in v-src-induced pgs2 Expression-v-src-activated expression from the PGS2 CRE can be enhanced by expression of ERK proteins and partially blocked by expression of dominant interfering ERK or Raf-1 proteins. These results demonstrate that the Raf/MEK/ ERK(MAPK) pathway also plays a role in v-src activation of pgs2 gene expression at the PGS2 CRE. ERKs do not transcriptionally activate either c-Jun or c-Fos by direct phosphorylation (23,25). They do, however, increase AP-1 activity in cells by phosphorylating and activating the transcription factor TCF/ Elk-1 responsible for increased expression from the c-Fos gene (23,25). One might expect that the Raf/MEK/ERK pathway might play a major role in secondary response genes whose induction is mediated by AP-1, following the immediate-early gene mechanism that increases c-Fos and c-Jun proteins in cells. In this case, protein X expressed in response to MAPK activation of ELK/SRF in Fig. 8 would be the c-Fos protein. In this regard, it is of considerable interest to note that PGS2 expression following v-src activation is persistent; pgs2 expression remains constantly elevated in v-src-transformed cells (3,30). In contrast, PGS2 induction by growth factors, tumor promoters, or inflammatory cytokines is transient and returns to basal values even if ligand is present continuously (6). These data suggest that v-src may induce AP-1 expression by a Raf/ MEK/ERK pathway, activating an additional, transcription-dependent mechanism of increased pgs2 gene expression that is not shared with other inducers.