Transcriptional Regulation of the Cyclooxygenase-2 Gene in Activated Mast Cells*

Activation of mast cells by aggregation of their IgE receptors induces rapid and transient synthesis of cy-clooxygenase-2 (COX-2). In this study we investigated (i) the cis-acting response elements and transcription factors active at the COX-2 promoter and (ii) the signal transduction pathways mediating COX-2 induction following aggregation of mast cell IgE receptors. Transient transfection assays with COX-2 promoter/luciferase constructs suggest that a consensus cyclic AMP response element is essential for induced COX-2 expression. Cotransfection studies with plasmids expressing c-Jun, dominant negative Ras, dominant negative c-Jun NH 2 - terminal kinase, and dominant negative MEKK1 demonstrate that activation of the Ras/MEKK1/c-Jun NH 2 -ter- minal kinase/c-Jun pathway is required for COX-2 promoter-mediated luciferase expression. Attenuation of COX-2 promoter activity by dominant negative constructs for Raf-1, ERK1, and ERK2 suggests that the Ras/Raf-1/extracellular signal-regulated kinase pathway is also necessary for COX-2 induction. Although mutating the two NF-IL6 sites individually did not affect COX-2 promoter activity, mutating both NF-IL6 sites substantially inhibits COX-2 promoter activity. More-over, overexpression of wild type CCAAT/enhancer-binding protein- b (C/EBP b ) augments COX-2 promoter activity in activated mast cells and cotransfection of a dominant negative C/EBP b construct completely blocks COX-2 promoter/luciferase expression. Our data suggest that in activated mast cells, a Ras/MEKK1/c-Jun expression vector. DNA concentrations for the transfections were held constant by adding appropriate empty vector DNA. 18 h later, the transfected cells were treated with IgE (1 h) and a -IgE (4 h) to activate the mast cells. Cells were then harvested, and extracts were assayed for luciferase activity and total protein. Renilla luciferase plasmid (0.1 m g) was used for transfection efficiency. Data are expressed as the average 6 S.D. The results were similar in three separate experiments. In this and in all other transfection experiments, the effect of dominant negative and other modifying expression plasmids had less than a 10% effect on the levels of Renilla luciferase activity.

eases (1,2). Cyclooxygenase (COX), 1 also known as prostaglandin synthase, is the key enzyme in prostaglandin, prostacyclin, and thromboxane synthesis from arachidonic acid (1). COX converts arachidonic acid, released from membrane phospholipid stores by phospholipases, to prostaglandin H 2 , the common precursor of all prostanoids. Two COX isoforms have been described (3). COX-1 is constitutively expressed in nearly all cells. The second COX isoform, COX-2, is induced by a wide range of ligands in many distinct cell types (4 -6) and is involved in stimulus-induced prostaglandin synthesis.
Several consensus sequences, including those for NF-B, NF-IL6, ATF/CRE, and an E-box found in the 5Ј region of the COX-2 gene, have been identified as regulatory sequences involved in COX-2 induction in response to a variety of stimuli in different species and cell types (7)(8)(9)(10)(11). In murine fibroblasts, the cyclic AMP response element, or CRE, located between nucleotides Ϫ56 and Ϫ52 of the murine COX-2 gene is necessary for the induction of COX-2 transcription mediated by v-src, serum, and PDGF (9,12). COX-2 induction via the CRE by v-src, serum, and PDGF in these fibroblast cells is mediated through both the Ras/MEKK1/JNK/c-Jun-and Ras/Raf-1/MAP kinase kinase/ERK-signaling pathways (9,12,13). IL-1␤ induction of COX-2 expression in both NIH3T3 cells and primary rat renal mesangial cells involves the activation of both JNK/ stress-activated protein kinase and p38 MAP kinase pathways (14) and the c-Jun transcription factor (12,14). The C/EBP family of transcription factors plays an important role in COX-2 induction by lipopolysaccharide and phorbol ester in human vascular endothelial cells (7), by tumor necrosis factor-␣ in murine MC3T3-E1 osteoblastic cells (10), and in mouse skin carcinoma cells (11). Transcription factor NF-B has been reported to mediate COX-2 induction by lipopolysaccharide in differentiated U937 monocytic cells (15) and by tumor necrosis factor-␣ in the MC3T3-E1 cell line (10). Thus, transcriptional mechanisms of COX-2 induction seem to be agonist-and cell type-specific and appear to involve context-specific interactions among several cis-acting regulatory elements, transcription factors, and signal transduction pathways.
Mast cells, distributed throughout vascularized epithelial tissue, play a critical role both in immune responses and in allergic disease. Mast cells, activated either by aggregation of their high affinity IgE receptors or by other effectors, release stored inflammatory mediators such as histamine and serotonin. Aggregation of mast cell IgE receptors also mediates the induced synthesis and release of inflammatory mediators such as leukotrienes and prostaglandin D 2 (PGD 2 ) (16). Unlike nearly all other cell types, prostaglandin production in activated mast cells occurs in two distinct phases, an immediate, activation-induced PGD 2 release completed within 10 -15 min, and a delayed phase of PGD 2 synthesis and secretion that peaks at 4 -6 h after activation (6). The immediate phase of PGD 2 synthesis in activated mast cells is due to conversion of arachidonic acid to prostaglandin by preexisting COX-1. In contrast, the delayed phase of PGD 2 synthesis and secretion following IgE receptor aggregation requires activation-induced transcription of COX-2 mRNA and production of functional protein (6,17).
The signal transduction pathways, transcription factors, and COX-2 promoter elements participating in stimulus-induced COX-2 expression have not been described for mast cells. In this report, we identify the cis-acting elements of the COX-2 promoter, the transcription factors, and the signal transduction pathways necessary for COX-2 induction in activated MMC-34 murine mast cells.

EXPERIMENTAL PROCEDURES
Plasmids-A wild type COX-2 promoter fragment from Ϫ724 to ϩ7 was PCR-amplified with Pfu polymerase (Stratagene) using the murine COX-2 reporter plasmid pT10L (9) as template. Mutant COX-2 promoter fragments were constructed in a two-stage PCR procedure. 2 Mutant constructs were made using PCR-amplified promoter fragments from pT10L, which were then cut with HindIII and XhoI, polyacrylamide gel-purified, and cloned into the HindIII and XhoI sites of the luciferase reporter plasmid pXP2. The CREB expression vector pRSV-CREB (18) was a gift from Marc Montminy (Harvard University). The c-Jun expression vector pSR␣MSVtkNeo-c-Jun (19) was provided by Charles Sawyers (UCLA). pSR␣MEKOE(K432M), an expression vector encoding a dominant negative MEKK1, was provided by Michael Karin (University of California, San Diego) (20). pEVX-3RatK375A, an expression vector that encodes a dominant negative Raf-1 (21), was from Susan MacDonald (ONYX, Richmond, CA). The expression vector for a kinase-defective JNK1 (pCDNA-DN-JNK1) (20) was provided by Roger Davis (University of Massachusetts). The expression vectors for pCEP4Erk1 K71R and pCEP4Erk2 K52R, encoding dominant negative Erk1 and dominant negative Erk2, respectively, were gifts from Melanie Cobb (University of Texas, Southwestern). The expression vectors for wild type C/EBP␤ and dominant C/EBP␤ were kindly provided by Stephen Smale (UCLA) and Robert Modlin (UCLA). pZIPM17 (22), an expression vector dominant negative Ha-Ras was the gift of Geoffrey Cooper (Harvard).
Cells and Transfections-Murine MMC-34 mast cells were cultured as described previously (6). Transient transfections were performed using Superfect reagent (Qiagen, Chatsworth, CA) according to the manufacturer's protocol for suspension cells, with slight modifications. MMC-34 mast cells were plated in 3 ml of regular medium at a density of 2 ϫ 10 6 cells/ml. Ten g of plasmid DNA was prepared in 150 l of serum-free, antibiotic-free medium and incubated with 30 l of Superfect reagent prepared separately in 150 l of serum-free, antibiotic-free medium. After 15 min, the DNA superfect complexes were added to cells and incubated for 2 h. Cells were then washed in phosphate-buffered saline, resuspended in medium supplemented with 0.5% serum, plated into 6-well dishes (one 6-well dish/10-cm dish), and incubated over night at 37°C. In cotransfection experiments, appropriate empty vector DNA was used to ensure similar DNA concentrations in all conditions. In all transfections, 0.1 g of Renilla luciferase plasmid (Promega, Madison, WI) was included to control for transfection efficiency. Protein concentrations were determined by Bradford assay.
Mast Cell Activation-The day after transfection (approximately 18 h), MMC-34 mast cells were activated as described previously (6). Briefly, MMC-34 cells were treated with 1 g/ml mouse IgE (PharMingen, San Diego, CA) for 1 h, washed, and further treated with 1 g/ml anti-IgE (PharMingen, San Diego, CA) for 4 h control cells received only medium after IgE treatment. After incubation, cells were washed with phosphate-buffered saline, lysed in passive lysis buffer provided in the dual luciferase kit (Promega), and assayed for luciferase activity ac-cording to the manufacturer's protocol, using a LUMAT LB9501 luminometer (Wallac inc., Gaithersberg, MD).

RESULTS
The CRE Site at Nucleotide Ϫ56 of the COX-2 5Ј-Flanking Sequence Is Essential for both Basal and Induced COX-2 Gene Expression in Activated Mast Cells-In NIH3T3 cells, the CRE element located between nucleotides Ϫ56 and Ϫ52 of the COX-2 gene is necessary for both basal COX-2 transcription and for induction mediated by v-src, serum, and PDGF (9,12). For the NIH3T3 studies, COX-2 promoter constructs that contained either Ϫ80 nucleotides (12) or Ϫ371 nucleotides (13) upstream of the transcription start site of COX-2 gene were used. The Ϫ80 construct containing the overlapping CRE and E-box elements was sufficient for COX-2 induction by v-src, and the Ϫ371 construct containing two additional NF-IL6 sites was sufficient for COX-2 induction by serum and PDGF. However, both of these promoter construct sets only had mutations in either the CRE or the E-box sites. More recently, several laboratories have reported regulation of COX-2 gene by NF-IL6 sites (7,10) and the NF-B site (10,15). To test the roles of the NF-IL6 and NF-B sites as well as the E-box and CRE site in the regulation of the COX-2 gene in mast cells, we generated a new set of COX-2 promoter/luciferase constructs. The wild type promoter, [COX-2 Ϫ724 ][Luc], includes 724 nucleotides upstream of the transcription start site and extends to position ϩ7 (Fig. 1). We utilized site-directed mutagenesis to generate mutant [COX-2 Ϫ724 ][Luc] constructs with specific mutations in the CRE, E-box, each of the two NF-IL6 sites, and the NF-B site in the COX-2 promoter. We also created an additional mutant COX-2 reporter in which both NF-IL6 sites were mutated ( Fig. 1).
MMC-34 murine mast cells were transfected with each of the COX-2 reporter mutants, and luciferase expression was examined 4 h after activation by aggregation of IgE receptors (Fig.  2). Mutation of the CRE element of the COX-2 gene reduced luciferase expression by more than 90% in both control and activated MMC-34 mast cells. In contrast, mutations in either the E-box or NF-B elements have no inhibitory effect on luciferase activity, either in control or in activated MMC-34 mast cells. Mutant constructs in which only one of the NF-IL6 sites was mutated did not differ from the wild type COX-2 expression vector in luciferase expression ( Fig. 2 and data not shown). However, luciferase expression in a mutant construct in which both NF-IL6 sites were mutated was reduced substan- A wild type (WT) promoter fragment between nucleotides Ϫ724 and ϩ7 was PCR-amplified from a COX-2 genomic fragment, and its sequence was verified. Site-directed mutant (m) constructs were prepared by PCR and verified by sequence analysis. Wild type and mutant COX-2 promoter fragments, from Ϫ724 to ϩ7, were cloned into HindIII-XhoI sites of pXP2, a promoter-less luciferase (firefly) plasmid. tially in activated MMC-34 cells. In contrast to the CRE mutation, the double NF-IL6 mutation did not cause a significant change for basal luciferase expression (Fig. 2). Our results suggest that the CRE site of the COX-2 regulatory region is essential for basal expression and for optimal induction of the COX-2 gene in activated MMC-34 mast cells. Although both NF-IL6 sites are not essential, regulation by this cis-acting region of the COX-2 promoter appears to also be important for COX-2 expression in activated mast cells.

Ras Is Required for [COX-2 Ϫ724 ][Luc] Reporter Induction in Activated Mast
Cells-Signal transduction pathways stemming from Ras are necessary for COX-2 induction by v-src, serum, and PDGF in NIH3T3 cells (12,13). Furthermore, aggregation of IgE receptors resulted in the activation of Ras and Ras effector pathways in mast cells (23). Moreover, transcription factors like Elk-1, an immediate early gene regulator, and the nuclear factor for activated T-cells are identified as Ras targets in mast cells activated by cross-linking of IgE receptors (23). We examined whether Ras activation is also required for  (Fig. 3), suggesting that Ras activation is required for COX-2 induction following IgE receptor aggregation on mast cells.

Induction of COX-2 Promoter in Activated Mast Cells Is Mediated by the RAS/MEKK1/JNK Signal Transduction
Pathway-Ras activates several signal transduction pathways. Each pathway leads to a coordinated phosphorylation of distinct subsets of transcription factors (24). In NIH3T3 cells, at least two distinct pathways stemming from Ras, MEKK1/JNK and Raf/MEK, are required for COX-2 induction (12, 13). We first examined whether the Ras-mediated MEKK1/JNK path-way is involved in COX-2 induction in activated mast cells. Expression of either kinase-defective dominant negative MEKK1 or kinase-defective dominant negative JNK1 blocks luciferase expression from the [COX-2 Ϫ724 ][Luc] in activated MMC-34 cells (Fig. 4).
The Raf-1/ERK Pathway Is Also Necessary for Induction of the COX-2 Promoter in Activated Mast Cells-Like the MEKK1/JNK pathway, the Raf/MEK/ERK pathway is necessary for COX-2 induction in NIH3T3 cells (12,13). The MAP kinase pathway enzyme MEK and the ERKs are Raf-dependent targets of Ras activation following aggregation of IgE receptors in mast cells, and activation of the Ras/MEK/ERK pathway results in the activation of transcription factors like Elk in activated mast cells (23). We next examined whether the Ras/MEK/ERK pathway plays a role in COX-2 induction in activated MMC-34 mast cells. Kinase-defective, dominant negative expression plasmids for Raf-1, ERK1, and ERK2 were cotransfected into MMC-34 cells along with [COX-2 Ϫ724 ][Luc]. All three dominant negative expression plasmids block COX-2 induction in activated MMC-34 cells (Fig. 5), suggesting that (i) the Ras/MEK/ERK pathway is also necessary for COX-2 induction in mast cells and (ii) the signaling transduction pathways mediating the induction of COX-2 promoter in activated mast cells are very similar to those identified in NIH3T3 cells in response to v-src, PDGF, and serum (12,13).
Overexpression of the CREB Transcription Factor Blocks COX-2 Promoter Activation in Mast Cells-Mutational analysis of the COX-2 promoter identified the CRE site CGTCA, located at nucleotide Ϫ56 of the murine COX-2 gene, as essential for COX-2 expression in activated MMC-34 mast cells (Fig. 2). We have previously shown, using gel shift assays in NIH3T3 cells, that CREB can bind to the CRE element of the murine COX-2 promoter (9). To identify the transcription factors that might reporter plasmids in 10-cm dishes, as described under "Experimental Procedures." Cells were then washed in phosphate-buffered saline, resuspended in medium supplemented with 0.5% serum, plated into 6-well dishes (one 6-well dish/10-cm dish), and incubated overnight at 37°C. Each transfected 10-cm dish was used to plate one 6-well dish in order to have triplicates for each condition, with and without activation. The day after transfection (approximately 18 h), MMC-34 cells were activated as described previously (6). Briefly, cells were treated with 1 g/ml mouse IgE for 1 h, washed, and further treated with 1 g/ml anti-IgE for 4 h. Control cells received medium alone after IgE treatment. After incubation, cells were washed with phosphate-buffered saline, lysed in passive lysis buffer provided in the Promega dual luciferase kit and assayed for luciferase activity. Renilla luciferase plasmid (0. play a role at the CRE, we first tested whether the classic CRE binding transcription factor, CREB, can mediate COX-2 induction in activated mast cells. If CREB is involved in COX-2 induction, we would expect wild type CREB protein to augment the induction from the COX-2 promoter in activated mast cells. However, when MMC-34 mast cells were transfected with an expression plasmid for CREB, luciferase expression induced from the COX-2 promoter by IgE receptor aggregation was completely blocked (Fig. 6). We conclude that some transcription factor other than CREB is involved in COX-2 transcriptional regulation at the CRE site in activated mast cells.
Overexpression of the c-Jun Transcription Factor Augments Expression from the COX-2 Promoter in Control and Activated Mast Cells-The c-Jun transcription factor can bind to the CRE site in the murine COX-2 gene (12). c-Jun mediates v-src, PDGF, and serum induction of COX-2 expression in NIH3T3 cells (12,13). c-Jun also mediates IL-1␤ induced COX-2 expression in rat mesangial cells (14). To determine whether c-Jun is also involved in the induction of COX-2 expression in activated mast cells, we transfected MMC-34 cells with the [COX-2 Ϫ724 ][Luc] reporter and a c-Jun expression vector. Overexpression of c-Jun augments the induced COX-2 expression in activated MMC-34 mast cells by more than 7-fold (Fig. 7). c-Jun overexpression is also able to activate the COX-2 promoter activity in unstimulated MMC-34 cells (Fig. 7). Our data suggest that c-Jun plays a critical role in COX-2 gene expression in mast cells.

Transcription Factor C/EBP␤ Augments Induction of the COX-2 Promoter in Control and Activated Mast Cells-Sirois
and Richards (25) report that C/EBP␤ may play a role in luteinizing hormone/follicle-stimulating hormone-mediated COX-2 induction in rat granulosa cells (25). In the mouse MC3T3-E1 osteoblastic cell line, tumor necrosis factor-␣-induced expression of COX-2 was mediated by two positive regulatory regions (Ϫ186 to Ϫ131 and Ϫ512 to Ϫ385) of the COX-2 promoter. The first element included a putative NF-IL6 element (C/EBP␤), and the second has an NF-B motif. Both of these elements were shown to be important in COX-2 regulation in MC3T3-E1 cells (10). The NF-IL6 and CRE sites are also involved in the transcriptional regulation of the human COX-2 gene by lipopolysaccharide and by phorbol ester in vascular endothelial cells (7). More recently, transcriptional regulation of the COX-2 gene in mouse skin carcinoma cells was shown to be mediated by the C/EBP family of proteins (11). COX-2 expression was substantially inhibited in activated MMC-34 mast cells when both NF-IL6 sites were mutated (Fig. 2). To directly test whether C/EBP␤ plays a role in COX-2 induction in activated mast cells, we examined the effect of C/EBP␤ overexpression on luciferase expression from the [COX-2 Ϫ724 ][luc] reporter. Expression of C/EBP␤ augmented COX-2 induction by more than 5-fold in activated MMC-34 cells (Fig.  7). Like c-Jun overexpression, C/EBP␤ overexpression also augmented the basal transcription from the COX-2 promoter. We conclude from these experiments that C/EBP transcription factors play an important role in the induction of the COX-2 gene in activated mast cells.
Dominant Negative C/EBP␤ Blocks both the Basal and Induced COX-2 Promoter Activity in Mast Cells-We next examined whether blocking C/EBP␤ function has any effect on COX-2 gene expression in activated mast cells. The C/EBP␤ mRNA encodes two different proteins from alternate translation start sites (26). The active form of C/EBP␤, containing the transactivation domain, DNA binding domain, and proteinprotein interaction domain was originally isolated as a "liveractivating protein," or LAP (27). A second C/EBP isoform without the transactivation domain was generated from a second translation start site. This C/EBP isoform was originally isolated as a "liver inhibitory protein" or LIP (26). LIP inhibits C/EBP␤ (liver-activating protein (LAP)) activity in a dominant negative fashion (26). An expression plasmid containing LIP (i.e. DN-C/EBP␤) was transfected into MMC-34 mast cells along with [COX-2 Ϫ724 ][Luc]. After activation by aggregation of IgE receptors, cells were harvested and analyzed for luciferase expression. Expression of DN-C/EBP␤ (LIP) completely blocked both the basal and induced expression from the COX-2 promoter (Fig. 8A).

Dominant Negative C/EBP␤ Can Inhibit COX-2 Gene Expression in Activated Mast Cells in a NF-IL6 Site-independent
Fashion-The most likely mechanism by which C/EBP transcription factors exert their regulatory roles on COX-2 gene expression in activated mast cells would be through the NF-IL6 sites. To test whether the NF-IL6 sites are necessary for the inhibition of COX-2 expression by LIP/DN-C/EBP␤, we repeated the LIP transfection experiment using a shorter COX-2/luciferase reporter construct, ([COX-2 Ϫ80 ][Luc]) (14). Luciferase activity was induced from [COX-2 Ϫ80 ][Luc] by receptor aggregation in mast cells, albeit at a reduced level when compared with the longer [COX-2 Ϫ724 ][Luc] promoter construct (Fig. 8B). Once again, co-expression of LIP (DN-C/EBP␤) blocked both the basal and induced expression from the COX-2 promoter in MMC-34 mast cells, even in the absence of the two NF-IL6 sites (Fig. 8B). Moreover, wild type C/EBP␤ also enhanced luciferase expression in activated MMC-34 mast cells transfected with [COX-2 Ϫ80 ][Luc] (data not shown). Our results suggest that C/EBP␤ may contribute to COX-2 expression both through activation at the NF-IL6 sites and via a mechanism not requiring interaction with the NF-IL6 sites.

DISCUSSION
The molecular mechanisms by which COX-2 gene expression is elevated in mast cells following aggregation of their high affinity IgE receptors have not previously been addressed. In this report, we use deletion and mutation constructs of the COX-2 promoter as well as wild type and dominant negative constructs for a number of signaling proteins to identify (i) the cis-acting response element(s) responsible for the induction of COX-2 expression in activated MMC-34 mast cells and (ii) signal transduction pathways that mediate COX-2 induction and (iii) the transcription factors involved in the induction of COX-2 promoter activity.
Cis-acting Elements of the COX-2 Promoter That Mediate COX-2 Expression in Activated Mast Cells-Previous studies in our laboratory have shown that, in murine NIH3T3 cells, the CRE element located between nucleotides Ϫ56 and Ϫ52 of COX-2 gene is necessary for the induction of COX-2 transcription-mediated by v-src, serum, and PDGF (9,12). Several reports suggest a role for NF-IL6 and NF-B as well as E-box sites in the transcriptional regulation of COX-2 in other cell types (7,10,25). To facilitate characterization of transcriptional regulation of the COX-2 gene, we created new constructs that contain 724 nucleotides of the COX-2 promoter. This set of constructs contains all the response elements that have been implicated thus far in COX-2 regulation. Mutations were introduced in the CRE, E-box, NF-IL6, and NF-B sites (Fig. 1). Similar to our previous observations in NIH3T3 cells (9), induction from the COX-2 promoter in activated MMC-34 mast cells also requires an intact CRE response element (Fig. 2).
There was no effect on luciferase expression when we used COX-2 promoter constructs harboring mutations in either the E-box or the NF-B elements. The E-box of the COX-2 gene was suggested to play a critical role in the regulation of COX-2 expression in rat ovarian granulosa cells (8). However, unlike the murine and human COX-2 promoters, the rat COX-2 promoter does not contain the CGTCA CRE element at nucleotide Ϫ56 (7). Kim and Fischer (11) report that the E-box of the murine COX-2 gene plays a prominent role in COX-2 transcriptional regulation in mouse skin carcinoma cells. However, the "E-box mutation" on which they base their conclusion changes two of the five critical nucleotides of the overlapping CRE of the COX-2 gene. We conclude from our data with murine mast cells and fibroblasts and data from other laboratories studying the human gene (7) that the COX-2 CRE plays a pivotal role in COX-2 gene expression in a wide range of cells, including mast cells, in response to a wide variety of stimuli.
Mutating either the CRE or NF-IL6 site, in human vascular endothelial cells, reduces lipopolysaccharide/12-O-tetradecanoylphorbol-13-acetate-induced COX-2 promoter activity by 40% and 10% respectively. Mutating both the CRE and NF-IL6 sites results in the maximum inhibition of activity, Ͼ75% (7). Using deletion constructs of the COX-2 promoter, Inoue et al. (7) conclude that transcriptional regulation of COX-2 in vascular endothelial cells is regulated through a combination of the NF-IL6 and CRE sites (7). Our results indicate that the regulation of the COX-2 gene might share similar characteristics in mast cells. The murine COX-2 promoter has two consensus NF-IL6 sites. We constructed vectors harboring mutations in either of the NF-IL6 sites or in both NF-IL6 sites. Although neither of the single site mutants has any effect on COX-2 promoter activity, we observe a significant (albeit not complete) inhibition when we use the construct harboring mutations in both NF-IL6 sites (Fig. 2). An interaction of these two sites thus appears to play a role in regulation of the COX-2 gene in activated mast cells. The human COX-2 promoter has only one putative NF-IL6 site. Subtle species differences in regulation of the COX-2 gene may exist as a consequence of these differences in promoter structure. In addition, at least one transcription factor that binds the NF-IL6 consensus sequences also appears to influence COX-2 gene expression in mast cells, albeit at least in part in a fashion independent of these sites (see below).
Signal Transduction Pathways That Mediate COX-2 Expression in Activated Mast Cells-Expression of a dominant negative Ras protein completely blocks luciferase induction from the COX-2 promoter in activated MMC-34 mast cells (Fig. 3). Ras also mediates oncogene and growth factor-induced transcriptional regulation of COX-2 in NIH3T3 cells (12). A potential link between high affinity IgE receptors and the Ras/mitogenactivated protein kinase-signaling pathway through SOS and Grb2 in mast cells has been reported (29). Moreover, distinct downstream Ras effector pathways have also been reported to be involved in the regulation of gene expression following aggregation of the high affinity receptors on mast cells (23). Cotransfection experiments utilizing dominant negative constructs for the several pathways downstream of Ras demonstrate that regulation of COX-2 expression in activated MMC-34 mast cells is mediated both by Ras/MEKK/JNK and Ras/Raf/ERK pathways. In this regard, COX-2 induction in activated mast cells and mitogen-induced fibroblasts share common features.
A number of recent reports describe a role for Ras activation of p38 MAP kinase signaling in induction of the COX-2 gene in several cell types, in response to a variety of ligands (14). Although we have not investigated p38 MAP kinase, it seems likely that this Ras-activated pathway may also play a role in induced COX-2 gene expression in activated mast cells.
Transcription Factors That Mediate COX-2 Expression in Activated Mast Cells-Previous studies in our laboratory demonstrated that c-Jun mediates v-src, PDGF, and serum induc-tion of COX-2 expression (12,13). c-Jun has also been implicated in COX-2 induction in response to IL-1␤ in rat renal mesangial cells (14). We also find that the CRE plays a major role in the induction of COX-2 gene expression in murine osteoblasts, in response to a variety of inducers, and that c-Jun plays the major role in transcriptional modulation in these cells (31). c-Jun, acting at the murine COX-2 promoter, also plays the major transcriptional role in mediating endotoxin induction of COX-2 expression in macrophages. 2 In our experiments with MMC-34 mast cells, overexpression of c-Jun augments COX-2 expression even more than it does in NIH3T3 cells (12,13). Although enhancement of gene expression by overexpression of a transcription factor does not conclusively demonstrate that this same transcription factor mediates the expression of the gene in question in vivo, our results are consistent with the suggestion that c-Jun plays a major role in COX-2 induction in mast cells.
Our observation that wild type CREB blocks expression from the COX-2 promoter in activated mast cells demonstrates that this classic transcriptional CRE activation factor does not mediate COX-2 gene expression in mast cells. In fibroblasts, we demonstrated by the use of chimeric transcription factors and an altered DNA binding site that the activation domain of c-Jun is responsible for induced COX-2 gene expression at the position of the CRE in the COX-2 promoter and that the activation domain of CREB is unable to elevate COX-2 gene expression (13). CREB also blocks COX-2 activation in osteoblasts (31) and in macrophages. 2 The inability of CREB to activate COX-2 gene expression in mast cells is consistent with an alternate transcription factor, c-Jun, playing a major role in COX-2 gene expression in activated mast cells.
The trans-acting factors that bind to the NF-IL6 site have many isoforms, including C/EBP␣, C/EBP␤, and C/EBP␦ (32)(33)(34). All of the C/EBP isoforms have a leucine zipper motif for dimer formation, thus allowing substantial cross-talk with other transcription factors (30). Moreover, phosphorylation of Thr-235 of C/EBP␤ protein by a Ras-dependent mitogen-activated protein kinase cascade is essential for C/EBP␤ activation (28), making C/EBP␤ a good candidate as a transcription factor required for COX-2 induction that is activated via the Raf/ERK pathway. Sirois and Richards (25) report that C/EBP␤ may play a key role in regulating COX-2 induction in rat granulosa cells. C/EBP␤ overexpression enhances and DN-C/EBP␤ inhibits COX-2 promoter activity of the [COX-2 Ϫ724 ][Luc] reporter gene in activated mast cells, suggesting that C/EBP␤ plays a role in COX-2 gene expression following mast cell activation. We also found that, in mast cells transfected with [COX-2 Ϫ80 ][Luc], (i) overexpressing C/EBP␤ can enhance luciferase induction, and (ii) expression of DN-C/EBP␤ can block luciferase induction. [COX-2 Ϫ80 ][Luc] does not contain either NF-IL6 site of the COX-2 promoter (data not shown). Without a deletion and/or mutational analysis of the region between Ϫ80 and the transcription start site of the COX-2 gene, we cannot formally rule out the possibility that C/EBP can also modulate COX-2 expression by interacting with an alternative binding site in this region. However, no conventional C/EBP binding sites are present in this region of the COX-2 gene, suggesting that C/EBP␤ can modulate COX-2 gene expression by proteinprotein interactions that are independent of DNA binding. Since the only known regulatory element found in [COX-2 Ϫ80 ] is the CRE, C/EBP proteins may modulate COX-2 gene expression in activated mast cells both by direct interactions with NF-IL6 binding sites and by modulation of the transcription factor binding and/or activation at the COX-2 CRE.
In summary, the transcriptional regulation of the COX-2 gene in activated mast cells is mediated (i) by both the CRE element present between Ϫ52 and Ϫ58 nucleotides on the COX-2 promoter and by the two NF-IL6 sites on the COX-2 promoter, (ii) by at least two Ras-dependent signaling pathways, Ras/MEKK/JNK and Ras/Raf/ERK, and (iii) by the transcription factors c-Jun and C/EBP␤.