Angiotensin II and Epidermal Growth Factor Induce Cyclooxygenase-2 Expression in Intestinal Epithelial Cells through Small GTPases Using Distinct Signaling Pathways*

Colorectal carcinogenesis is a multistep process involving genetic mutations and alterations in rigorously controlled signaling pathways and gene expression that control intestinal epithelial cell proliferation, differentiation, and apoptosis. Cyclooxygenase-2 (COX-2) is aberrantly expressed in premalignant adenomatous polyps and colorectal carcinomas and is associated with increased epithelial cell proliferation, decreased apoptosis, and increased cell invasiveness. Currently, knowledge of the regulation of expression of COX-2 by endogenous cell-surface receptors is inadequate. Recently, in a non-transformed rat intestinal epithelial cell line (IEC-18), we showed induction of cell proliferation and DNA synthesis by angiotensin II (Ang II) via the endogenous Ang II type 1 receptor (Chiu, T., Santiskulvong, C., and Rozengurt, E. (2003) Am. J. Physiol. 285, G1–G11). We report that Ang II potently stimulated expression of COX-2 mRNA and protein as an immediate-early gene response through the Ang II type 1 receptor, correlating with an increase in prostaglandin I2 production. Ang II induced Cdc42 activation and filopodial formation. COX-2 expression was induced by epidermal growth factor (EGF), which activated Rac with lamellipodial formation. Inhibition of small GTPases by Clostridium difficile toxin B blocked COX-2 expression by Ang II and EGF. Inhibition of ERK activation by U0126 or PD98059 significantly decreased EGF-dependent COX-2 expression, but did not affect Ang II-dependent COX-2 expression. Conversely, inhibition of p38MAPK by SB202190 or PD169316 inhibited COX-2 expression by Ang II, but did not block COX-2 induction by EGF. Ang II caused Ca2+ mobilization. Inhibition of Ca2+ signaling by 2-aminobiphenyl borate blocked Ang II-dependent COX-2 expression. EGF did not induce Ca2+ mobilization, and 2-aminobiphenyl borate did not inhibit EGF-dependent COX-2 expression. Inhibition of COX-2 expression correlated with inhibition of prostaglandin I2 production. Luciferase promoter assays showed that Ang II-dependent transcriptional activation of the COX-2 promoter was dependent on activation of small GTPases and p38MAPK and on Ca2+ signaling via the cAMP-responsive element/activating transcription factor cis-acting element.

Prostaglandins (PGs) 1 play a fundamental role in a broad range of physiological processes, including secretion, cytoprotection, pain transmission, epithelial and endothelial barrier function, and motility (1)(2)(3)(4)(5). Alterations in PG production are also implicated in pathological processes, including acute and chronic inflammation, angiogenesis, cancer, and allergic diseases (6 -9). PGs are produced from arachidonic acid by cyclooxygenases (COX) (10,11). Acute changes in PG levels in cells are controlled by COX-2 (prostaglandin-endoperoxide synthase, EC 1.14.99.1), which is rapidly induced as an immediateearly gene in response to pro-inflammatory cytokines, tumor promoters, and growth factors (12)(13)(14)(15). COX-2 is overexpressed in cancers of the colon (16,17), stomach (18), lung (19), and breast (20). Tissue-specific overexpression of COX-2 in a transgenic mouse model has been shown to be carcinogenic (21). Chronic inhibition of COX activity by nonsteroidal anti-inflammatory drugs has been associated with chemopreventative effects on colon cancer (4). Consequently, the identification of the pathways and regulatory elements that control COX-2 expression in general and in intestinal cells in particular is the subject of major interest.
The sequential proliferation, lineage-specific differentiation, migration, and cell death of epithelial cells of the intestinal mucosa are a tightly regulated process that is modulated by a broad spectrum of regulatory peptides (22)(23)(24), but the signal transduction pathways involved remain incompletely understood. Non-transformed IEC-18 cells, derived from rat small intestinal crypts (25), have provided an in vitro model to examine intestinal epithelial cell migration, differentiation, and proliferation (26 -28). These cells are undifferentiated as judged by morphologic and functional criteria; they resemble stem cells and thus may serve as a model to study the crypt stem cell (29). The importance of examining this compartment is highlighted by recent reports suggesting that the initial mutant cell in the pathogenesis of colorectal adenoma may arise from the crypt stem cell compartment involving aberrant COX-2 expression (30,31).
Neuropeptides and vasoactive peptides that signal through G-protein-coupled receptors act as potent cell growth factors for a variety of cell types (32)(33)(34)(35). In particular, G-protein-coupled receptors and their agonists play a crucial role in the regulation of multiple functions in the gastrointestinal tract, including cell proliferation, motility, and transport. Angiotensin II (Ang II) stimulates ion transport in both the small and large intestines through Ang II type 1 (AT 1 ) receptors located on epithelial cells (36,37). We recently demonstrated that Ang II, one of the most potent regulators of blood pressure, induces proliferation and DNA synthesis in IEC-18 cells (39). We showed that Ang II, acting through the G q -coupled AT 1 receptor, stimulates protein kinase C-dependent ERK, protein kinase D, and tyrosine phosphorylation of Pyk2 (38 -40). Ang II regulates circulating blood volume (41) and stimulates neovascularization (42) and cell proliferation (21). Renin, an enzyme that produces angiotensin, is found in cancer blood vessels (43), and long-term administration of angiotensin-converting enzyme (ACE) inhibitors, antihypertensive agents (44), reduces morbidity and mortality in cancer patients (45). A recent study reports that specific COX-2 inhibitors and ACE inhibitors synergistically reduce tumor growth in an in vivo mouse model for colon cancer (46). These data suggest that Ang II could have a significant role in colon carcinogenesis possibly by induction of COX-2 expression.
We report that Ang II potently induced COX-2 mRNA and protein in non-transformed rat intestinal epithelial IEC-18 cells via an endogenous AT 1 receptor. Multiple signaling cascades mediated by AT 1 receptor activation involving pertussis toxin-insensitive G q and possibly G 12 families of het-erotrimeric G-proteins combine to induce transient COX-2 expression. Furthermore, we show an absolute requirement for Ca 2ϩ signaling and activation of small GTPases for Ang II-dependent COX-2 expression and PGI 2 production in IEC-18 cells. These findings define signaling pathways coupled to an endogenous AT 1 receptor that regulate COX-2 expression as a potential mechanism for epigenetic changes that play a role in colorectal carcinogenesis.
Cell Culture-IEC-18 cells were purchased from American Type Culture Collection. Stock cultures of these cells were maintained in Dulbecco's modified Eagle's medium (Sigma) supplemented with 10% fetal bovine serum and 1% antibiotic/antimycotic (Invitrogen) in a humidified 10% CO 2 atmosphere at 37°C. For experiments, cells were plated onto 100-mm dishes (Nunc, Naperville, IL) at 10 5 cells/dish, allowed to growth to confluency (5-7 days), and then changed to serum-free Opti-MEM (Invitrogen) for 18 -24 h prior to experiments.
Expression Vectors-IEC-18 cells were grown in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum in a hu-FIG. 1. Time-and dose-dependent induction of COX-2 by Ang II through the AT 1 receptor. A, confluent IEC-18 cells were serum-starved (18 h) before stimulation with Ang II (100 nM). RNA was isolated at 0, 0.5, 1, 2, 4, and 8 h after addition of agonist. The relative levels of COX-2 mRNA compared with GAPDH mRNA were measured by real-time RT-PCR. Western blot analysis (inset) was performed to measure Ang II-induced COX-2 protein expression at 0, 1, 2, 4, 8, and 24 h using anti-COX-2 antibody. Equal amounts of protein were loaded in each lane, which was verified by screening with anti-ERK2 antibody (data not shown). B, confluent serum-starved IEC-18 cells were treated with increasing amounts of Ang II (1 pM to 100 nM) for 4 h. The relative levels of COX-2 mRNA were measured by real-time RT-PCR. Induction of COX-2 protein expression levels by increasing amounts of Ang II was measured by Western blot analysis (inset). C, confluent serum-starved IEC-18 cells were pretreated with the AT 1 receptor antagonist losartan (1 M) 1 h prior to stimulation with Ang II (100 nM). Relative COX-2 mRNA expression levels were measured. D, confluent serum-starved IEC-18 cells were either untreated or pretreated with pertussis toxin (PTX; 50 ng/ml, 18 h) prior to stimulation with Ang II (100 nM). Relative COX-2 mRNA expression levels were measured after 1 h. Western blotting to measure Ang II-induced COX-2 protein expression levels after 4 h in cell pretreated with pertussis toxin was performed (inset). midified atmosphere containing 10% CO 2 at 37°C. The COX-2 promoter-reporter plasmid pTIS-10s contains luciferase cDNA under the control of the murine COX-2 promoter (Ϫ371 to ϩ70). The other COX-2 promoter-reporter plasmids are described (47). phRG-TK (Promega, Madison, WI) contains Renilla luciferase cDNA under the control of a minimal thymidine kinase promoter.
Western Blot Analysis-IEC-18 cells were grown on 60-mm dishes. When confluent and quiescent, the cells were incubated in Opti-MEM (18 h) prior to stimulation with agonists. The cells were washed twice with phosphate-buffered saline prior to lysis with 2ϫ SDS-PAGE sample buffer (Bio-Rad). Equal volumes of protein extract were subjected to SDS-PAGE using 10% acrylamide gels and then electroblotted onto polyvinylidene difluoride paper (Bio-Rad). Membranes were blocked in 5% nonfat milk in Tris-buffered saline for 1 h. Membranes were incubated overnight with anti-COX-2 antibody (Cayman Chemical Co., Inc., Ann Arbor, MI) at a 1:1000 dilution in Tris-buffered saline and 0.05% Tween 20. The membrane was washed three times with Tris-buffered saline and 0.05% Tween 20 and then incubated with horseradish peroxidase-conjugated anti-rabbit IgG antibody (1:5000; Amersham Biosciences) for 1 h. Immunoreactive bands were detected by enhanced chemiluminescence (Amersham Biosciences). Antibodies specific for ERK1/2, phospho-ERK, p38 MAPK , phospho-p38 MAPK , and activating transcription factor (ATF)-2 were purchased from Cell Signaling (Beverly, MA), and the antibody against phospho-ATF-2 was from BIO-SOURCE (Camarillo, CA).
Isolation of RNA and TaqMan Analysis-Total RNA was harvested using TRIzol reagent (Invitrogen) according to the manufacturer's instructions. The relative levels of COX-2 mRNA to glyceraldehyde-3-phosphate dehydrogenase (GAPDH) mRNA were measured using the real-time reverse transcription (RT)-PCR TaqMan assay (Applied Biosystems, Inc., Foster City, CA). The COX-2 primers used were 5Ј-GGCACAAATATGATGTTCGCA-3Ј (forward) and 5Ј-CCTCGCT-TCTGATCTGTCTTGA-3Ј (reverse), and the probe used was carboxyfluorescein-5Ј-TCTTTGCCCAGCACTTCACTCATCAGTTT-3Ј-carboxytetramethylrhodamine. GAPDH mRNA was measured using the rodent GAPDH TaqMan primer set (Applied Biosystems, Inc.). Dual reaction mixtures were made using the TaqMan one-step RT-PCR master mixture (Applied Biosystems, Inc.), 25 ng of sample RNA, and either the COX-2 or GAPDH primer/probe set. Assays were run on an ABI 7700 system using settings of 30 min at 48°C; 10 min at 95°C; and forty cycles of 15 s at 95°C and 60 s at 60°C at the Human Genetics Core Facility of UCLA. The change in COX-2 mRNA relative to GAPDH mRNA was calculated from the measured C T value using the following formula: Measurement of COX-2 mRNA Stability-IEC-18 cells were serumstarved overnight prior to stimulation with Ang II (100 nM) or EGF (25 ng/ml). After induction for 1 h, cells were treated with actinomycin D (5 g/ml). Total RNA was purified at 0, 15, 30, 45, and 60 min. COX-2 and GAPDH mRNA expression levels were measured by TaqMan analysis.
Transfection and Luciferase Assay-Cells (5 ϫ 10 4 /well) were plated onto 6-well dishes (Nunc). The next day, a transfection mixture was prepared by adding 0.51 g of the COX-2 promoter-reporter vector, 0.1 g of phRG-TK, and a combination of pcDNA3 and other expression vectors (totaling 0.24 g) to 51 l of Opti-MEM. Following a 20-min incubation after addition of Superfect (5 l; QIAGEN Inc., Chatsworth, CA), the transfection mixture was combined with 0.6 ml of complete Dulbecco's modified Eagle's medium and placed on the cells in each well. After 2 days, the medium was replaced with Opti-MEM.
The next day, the cells were treated with agonists. After the indicated times, the cells were washed with phosphate-buffered saline and then lysed with passive lysis buffer (150 l; Promega). The relative light units from firefly and Renilla luciferases (20 l) were measured using the Dual-Luciferase activity assay kit (100 l; Promega) on a Turner TD20/20 luminometer. The ratio of firefly to Renilla luciferase activity was calculated. The results are reported as relative changes in the ratio, with a relative -fold increase of 1.0 signifying no change. The results are reported as the means Ϯ S.E. of three or more sets of transfections done in triplicate.
Measurement of [Ca 2ϩ ] i -IEC-18 cells were grown on glass coverslips and then loaded with Fura-2/AM (5 M) in Hanks' balanced salt solution (pH 7.2) supplemented with 35 mM NaHCO 3 , 1.3 mM CaCl 2 , 0.5 mM MgCl 2 , 0.4 mM MgSO 4 , and 1% bovine serum albumin (buffer A) for 30 min at room temperature. The coverslip was mounted in a chamber on the stage of a Zeiss 100TV inverted microscope with a ϫ40 objective (Fluar). Mounted cells were perfused with buffer A (ϳ1 ml/min) using a two-channel peristaltic pump (Rainin Instrument Co. Inc., Woburn, MA) at 37°C. [Ca 2ϩ ] i was measured with a fluorescent video microscopy system (RatioVision, Atto Instruments, Inc., Rockville, MD) using 334and 380-nm excitation filters as described previously (48). Calibration of the Fura-2 fluorescence ratios was accomplished in vitro with a series of buffered Ca 2ϩ standards containing Mg 2ϩ .
Small GTPase Activity Assay-Confluent serum-starved IEC-18 cells were untreated or treated with either Ang II (100 nM, 5 min) or EGF (25 ng/ml, 5 min) prior to harvesting. The levels of GTP-bound Rho, Rac, and Cdc42 were measured using the appropriate activation assay kit (Cytoskeleton, Inc., Denver, CO) according to the manufacturer's instructions.
Nuclear Run-on Assay-Nuclei from agonist-treated IEC-18 cells (10 7 ) were isolated and stored in liquid N 2 (49). In vitro run-on transcription was carried out using 150 Ci of [␣-32 P]UTP/agonist condition at 30°C for 30 min. The labeled RNA transcripts were isolated using TRIzol. Equal amounts of radioactivity (5 ϫ 10 6 cpm) for each assay condition were hybridized to filter-immobilized cDNA in 1 ml of Ultrahyb (Ambion Inc., Austin TX) for 48 h at 30°C. The filters were washed twice with 2ϫ SSC at 52°C for 1 h and then incubated with RNase A (5 g/ml) for 30 min at 37°C, followed by a final wash with 2ϫ SSC at 52°C for 1 h. The filters were exposed to PhosphorImager screens for 2-3 days, and the radioactivity was measured using Typhoon (Amersham Biosciences).
Prostaglandin Assay-Confluent IEC-18 cells grown on 24-well plates were incubated with Opti-MEM 18 h prior to stimulation. Opti-MEM was replaced, and the cells were stimulated with Ang II (100 nM) or EGF (25 ng/ml) for 6 h. The medium was collected, and the level of 6-keto-PGF 1␣ was measured using an enzyme immunoassay kit (Cayman Chemical Co., Inc.).
Statistical Analysis-Data were analyzed by Student's t distribution using average values and the associated standard errors. A comparative p value of Ͻ0.05 was considered significant.

FIG. 2. Time-and dose-dependent induction of COX-2 by EGF.
A, confluent IEC-18 cells were serum-starved (18 h) before stimulation with EGF (25 ng/ml). RNA was isolated at 0, 0.5, 1, 2, 4, and 8 h after addition of agonist. The relative levels of COX-2 mRNA compared with GAPDH mRNA were measured by real-time RT-PCR. Western blot analysis (inset) was performed to measure EGF-induced COX-2 protein expression at 0, 1, 2, 4, 8, and 24 h. B, confluent serum-starved IEC-18 cells were treated with increasing amounts of EGF (1-25 ng/ml) for 4 h. The relative levels of COX-2 mRNA were measured by real-time RT-PCR. 1 Receptor-To determine whether Ang II induces COX-2 expression in non-transformed intestinal epithelial IEC-18 cells, total RNA was isolated from cells that were treated with this agonist for various times (0.5-8 h). The levels of COX-2 mRNA relative to GAPDH mRNA were measured by real-time RT-PCR (Fig. 1A). Ang II rapidly induced COX-2 mRNA, which peaked 1 h after addition of agonist and then returned to basal levels by 4 h. Induction of COX-2 mRNA in these cells is an immediate-early gene response because inhibition of de novo protein synthesis by cycloheximide did not block induction by Ang II (data not shown).

Ang II Induces COX-2 Expression in IEC-18 Cells through the AT
We next examined whether the transient increase in the expression of COX-2 mRNA induced by Ang II leads to an increase in the levels of COX-2 protein. Proteins from extracts of cells treated with Ang II for up to 24 h were analyzed by SDS-PAGE, followed by Western blotting using anti-COX-2 antibody (Fig. 1A, inset). Confluent serum-starved IEC-18 cells expressed low relative levels of COX-2 protein. Stimulation of these cells with Ang II induced a marked increase in COX-2 protein, which was evident within 1 h after agonist treatment and reached a maximum at 8 h. By 24 h after Ang II treatment, the level of COX-2 protein had declined to near basal levels.
Ang II induced COX-2 expression in serum-starved IEC-18 cells in a dose-dependent fashion (Fig. 1B). COX-2 mRNA levels were measured in cells treated with increasing amounts of Ang II. An EC 50 value of 1.76 Ϯ 0.01 nM was calculated for Ang II from a sigmoidal dose-response curve that was modeled to the data. The level of COX-2 protein also increased in response to increasing doses of Ang II (Fig. 1B, inset).
Ang II is known to exert its biological effects through binding to two receptor subtypes, AT 1 and AT 2 receptors, which are members of the G-protein-coupled receptor superfamily (50). Previously, we showed that Ang II potently activates multiple kinase signaling pathways through the AT 1 receptor (38,39). To determine which receptor subtype mediates Ang II-dependent induction of COX-2, cells were pretreated with losartan, a specific AT 1 receptor antagonist (50), prior to treatment with Ang II (Fig. 1C). The increases in the levels of COX-2 mRNA in response to Ang II were significantly reduced in the presence of losartan. Pretreatment of cells with the AT 2 receptor antagonist PD123319 (51) did not inhibit COX-2 mRNA induction by Ang II (data not shown).
The AT 1 receptor can be coupled to several distinct families of heterotrimeric G-proteins (e.g. G q , G i , and G 12 ). To determine whether G i signaling is involved in Ang II-dependent COX-2 expression, IEC-18 cells were pretreated with pertussis toxin (52), followed by induction with Ang II (Fig. 1D). Treatment with pertussis toxin did not decrease the level of Ang II-dependent COX-2 mRNA expression. Western blot analysis of cells treated with pertussis toxin showed no decrease in Ang II-dependent COX-2 protein expression (Fig. 1D, inset).
EGF Induces COX-2 Expression in IEC-18 Cells-EGF and transforming growth factor-␣ are structurally related peptides that stimulate DNA synthesis and cell growth in various systems, including the gastrointestinal tract (53). Numerous studies have shown COX-2 expression induced by EGF in multiple cell types (49,54,55). To examine whether EGF induces COX-2 expression in IEC-18 cells, total RNA was isolated from con-

FIG. 3. COX-2 expression by Ang II involves transcriptional and post-transcriptional mechanisms.
A, nuclear run-on analysis was performed on IEC-18 cells treated with Ang II (100 mM) for 30 min prior to isolation of nuclei. Radiolabeled nascent RNA was hybridized to immobilized plasmids containing portions of cDNA encoding COX-2, c-fos, and GAPDH. B, IEC-18 cells were treated with either Ang II (100 mM) or EGF (25 ng/ml). After 1 h, de novo RNA transcription was blocked using actinomycin D. Total RNA was isolated at the indicated times, and the relative COX-2 mRNA levels were measured. C, IEC-18 cells were untreated or treated with the specific COX-2 inhibitor NS-398 (10 M) 1 h prior to incubation with Ang II (100 nM) or EGF (25 ng/ml) for 6 h. The medium was collected, and the stable metabolite of PGI 2 , 6-keto-PGF 1␣ , was measured by enzyme immunoassays. fluent serum-starved cells that had been treated with EGF for various times (0.5-8 h). The relative levels of COX-2 mRNA were measured by real-time RT-PCR ( Fig. 2A). EGF induced COX-2 mRNA, which peaked 1 h after exposure to agonist and then decreased to basal levels by 4 h. Western blot analysis showed a time-dependent increase in COX-2 protein levels in cells treated with EGF ( Fig. 2A). EGF-induced COX-2 protein levels were elevated after 1 h and decreased to unstimulated levels after 24 h ( Fig. 2A, inset).
EGF induced COX-2 expression in serum-starved IEC-18 cells in a dose-dependent fashion (Fig. 2B). COX-2 mRNA levels were measured in cells treated with increasing amounts of EGF. The EC 50 for induction of COX-2 mRNA in IEC-18 cells by EGF was calculated to be 5.4 Ϯ 1.5 ng/ml. Ang II and EGF were each able to induce COX-2 mRNA and protein in IEC-18 cells. However, stimulation of cells with Ang II consistently resulted in a higher level of COX-2 mRNA and protein compared with stimulation with EGF.
Ang II-dependent COX-2 Expression Involves Transcriptional and Post-transcriptional Mechanisms-COX-2 expression is highly regulated by both transcriptional and post-transcriptional mechanisms. To determine whether Ang II increases the transcription of the COX-2 gene in IEC-18 cells, nuclear run-on assays were used to measure the relative transcription rate of the COX-2 gene (Fig. 3A). Serum-starved cells showed an increase in the level of de novo transcribed COX-2 RNA 30 min after stimulation with Ang II. The relative transcription levels of c-fos (an immediate-early gene) were also elevated 30 min after stimulation with Ang II.
Post-transcriptional regulation of COX-2 expression includes changes in COX-2 mRNA stability. It has been demonstrated in multiple cell types that MAPK signaling (i.e. ERK and p38 MAPK ) plays a role in stabilization of COX-2 mRNA via recognition of the AUUUA motifs present in the 3Ј-untranslated region of the COX-2 mRNA (56 -58). To determine whether COX-2 mRNA stability is different in cells treated with Ang II compared with EGF, the half-life of COX-2 mRNA was measured (Fig. 3B). After 1 h of stimulation by agonists, the half-life of COX-2 mRNA in IEC-18 cells was determined to be 11 min for Ang II-treated cells compared with 7 min for EGF-treated cells.
COX-2-dependent PGI 2 Production-A functional consequence of COX-2 expression is an increase in prostanoid production. The specific PG synthases expressed in IEC-18 cells FIG. 4. Small GTPases are required for COX-2 expression by Ang II or EGF. A, confluent serum-starved IEC-18 cells were pretreated with C. difficile toxin B (100 ng/ml, 4 h) prior to stimulation with either Ang II (100 nM) or EGF (25 ng/ml). After 1 h, total RNA was isolated, and the relative levels of COX-2 were measured. B, cells were pretreated with toxin B (Tox B; 25 ng/ml, 18 h) prior to stimulation with either Ang II (100 nM) or EGF (25 ng/ml). COX-2 protein expression levels were measured by Western blot analysis after stimulation with agonist for 4 h. C, IEC-18 cells were pretreated with toxin B (25 ng/ml, 18 h) prior to stimulation with Ang II (100 mM) or EGF (25 ng/ml). After 6 h, the medium was collected, and secreted 6-keto-PGF 1␣ was measured. D, IEC-18 cells were treated with either Ang II (100 nM) or EGF (25 ng/ml) for 5 min prior to measuring the relative levels of GTP-bound Rho, Rac, and Cdc42. E, IEC-18 cells were grown on glass coverslips and then stimulated with either Ang II (100 nM) or EGF (25 ng/ml). After 10 min, the cells were fixed, and the actin cytoskeleton was stained with fluorescently labeled phalloidin. determine the type of prostanoids that are produced from COX-2 actions on arachidonic acid. We found that Ang II and EGF acutely induced PGI synthase mRNA expression in IEC-18 cells. 2 Consequently, PGI 2 production was measured. Production of the PGI 2 stable metabolite 6-keto-PGF 1␣ was increased by Ang II and EGF (Fig. 3C). Pretreatment of cells with the specific COX-2 inhibitor NS-398 blocked agonist-induced PGI 2 production. These results show that prostacyclin production, viz. PGI 2 , by Ang II and EGF is mediated by COX-2.
COX-2 Expression Is Dependent on Small GTPases-Our previous studies have characterized the transcriptional activation of the COX-2 promoter via activation of small GTPases in fibroblasts (47,59). Because of the important role that COX-2 plays in intestinal epithelial cells, we wanted to determine whether small GTPases regulate Ang II-dependent COX-2 expression in IEC-18 cells. Cells were incubated with C. difficile toxin B, a cell-permeable protein that inactivates the Rho family of GTPases by monoglucosylation (60,61), prior to stimulation with agonists. Toxin B significantly inhibited COX-2 mRNA induction by Ang II or EGF (Fig. 4A). This inhibition of COX-2 mRNA expression is reflected by inhibition of COX-2 protein expression by Ang II or EGF in cells pretreated with toxin B (Fig. 4B).
The effectiveness of Ang II or EGF in the activation of the Rho family of GTPases in IEC-18 cells was measured (Fig. 4D). Ang II and EGF caused a mild increase in the level of GTPbound Rho in IEC-18 cells. Conversely, EGF preferentially increased the level of GTP-bound Rac by 5-fold compared with a 1.3-fold increase by Ang II. The basal level of activated GTP-bound Cdc42 was very low, but was induced by Ͼ6-fold by Ang II. Cells treated with EGF showed a 2.4-fold increase in the level of GTP-bound Cdc42.
Small GTPases play a critical role in regulating the organization of the cytoskeleton. Rho, Rac, and Cdc42 activation induces the formation of actin stress fibers, membrane ruffles, and filopodia, respectively. In light of the results presented in Fig. 4D, we evaluated actin organization in IEC-18 cells stimulated with Ang II or EGF. As shown in Fig. 4E, both Ang II and EGF were able to remodel the actin cytoskeleton. Serumstarved cells contained both cortical actin and stress fibers, suggesting an elevated basal level of activated Rac and Rho. Treatment with Ang II resulted in microspikes/filopodia, indicating signaling by Cdc42. In contrast, EGF caused thickening of the cortical actin forming ruffles, which is controlled by Rac. Prior treatment with toxin B modified the actin cytoskeleton in serum-starved cells, lowering the overall level of stress fibers and cortical actin. Toxin B inhibited EGF-induced cortical actin/lamellipodia and Ang II-dependent formation of filopodia. Despite the demonstrated shared requirement for small GTPases for COX-2 expression by Ang II and EGF in IEC-18 cells, our data also show significant differences in signaling pathways used by Ang II and EGF. 2 L. W. Slice and E. Rozengurt, unpublished data.

FIG. 5. ERK phosphorylation is not required for Ang II-dependent COX-2 expression.
A, confluent serum-starved IEC-18 cells were stimulated with Ang II (100 mM) or EGF (25 ng/ml), and protein extracts were taken at the indicated times. Western blots were analyzed using antibodies specific for phospho-ERK and ERK2. Cells were pretreated with PD98059 (10 M) or U0126 (1 M) for 30 min prior to stimulation with Ang II (100 nM) or EGF (25 ng/ml). B, relative COX-2 mRNA expression levels were measured in cells after 1 h. C, COX-2 protein expression levels were measured in cells after 4 h. D, shown are the results from Western blot analysis of ERK phosphorylation in cells pretreated with either PD98059 or U0126 prior to stimulation with Ang II or EGF for 10 min.
Ang II-induced COX-2 Expression Is Dependent on p38 MAPK , but Not ERK-Numerous studies have shown that MAPKs are regulated by small GTPases. Rac and Cdc42 bind to and activate the serine/threonine protein kinase p21 PAK (62). Activated p21 PAK can phosphorylate MEK1 at Thr 292 and Ser 298 , which leads to enhanced association of MEK with ERK, resulting in ERK activation (63,64). Activation of ERK has been shown to mediate the induction of COX-2 expression in several cell types, including rat intestinal epithelial cells stimulated by bombesin (58,(65)(66)(67). To define the role played by ERK in COX-2 expression, IEC-18 cells were stimulated with Ang II or EGF. These stimuli promoted a marked increase in ERK phosphorylation but with different kinetics (Fig. 5A). ERK phosphorylation by Ang II reached a maximum after 10 min and then rapidly decreased. ERK phosphorylation by EGF also reached its peak after 10 min, but in contrast, it remained elevated up to 30 min after initial stimulation.
Inhibition of ERK activation by the specific MEK inhibitors PD98059 and U0126 decreased the EGF-dependent COX-2 mRNA induction by 52 and 66%, respectively (Fig. 5B). In striking contrast, there was no significant decrease in Ang II-dependent COX-2 mRNA expression in cells treated with either PD98059 or U0126.
Inhibition of ERK activation by PD98059 or U0126 did not block the induction of COX-2 protein by Ang II in IEC-18 cells (Fig. 5C). COX-2 expression in response to EGF was signifi-cantly reduced in IEC-18 cells treated with PD98059 or U0126. We verified the effectiveness of PD98059 and U0126 in blocking MEK-dependent activation of ERK by Ang II or EGF in parallel cultures of IEC-18 cells (Fig. 5D).
Activation of p21 PAK can also lead to activation of p38 MAPK (68,69). Earlier studies using rat cardiomyocytes showed Ang II-induced prostacyclin production via p38 MAPK cascades (70). It is possible that Ang II-dependent COX-2 expression is mediated by Cdc42/p21 PAK -dependent activation of p38 MAPK signaling. To determine this, serum-starved IEC-18 cells were stimulated with Ang II, and the level of phosphorylated p38 MAPK was measured (Fig. 6A). p38 MAPK was rapidly phosphorylated at Thr 180 /Tyr 182 , reaching a maximum after 5 min and then rapidly declining to unstimulated levels by 20 min. Activation of p38 MAPK leads to ATF-2 phosphorylation (71), which has been demonstrated to increase its transcriptional activity (72). Ang II stimulated ATF-2 phosphorylation in IEC-18 cells, reaching a maximum at 10 min and then rapidly decreasing to basal levels by 30 min. EGF did not produce a significant increase in p38 MAPK phosphorylation in IEC-18 cells (data not shown).
To determine the role that p38 MAPK plays in Ang II-dependent COX-2 expression, IEC-18 cells were pretreated with specific inhibitors of p38 MAPK activity, SB202190 (10 M) and PD169316 (10 M), prior to stimulation with Ang II (Fig. 6B). Maximal induction of COX-2 mRNA by Ang II was decreased by FIG. 6. Ang II-dependent COX-2 expression is mediated by p38 MAPK . A, confluent serum-starved IEC-18 cells were treated with Ang II (100 nM) for the indicated times. Western blot analysis was done using antibodies against phospho-p38 MAPK (pp38 MAPK ) and phospho-ATF-2 (ppATF-2). B, cells were pretreated (30 min) with the specific p38 MAPK inhibitors SB202190 (10 M) and PD169316 (10 M) prior to stimulation with Ang II (100 nM, 1 h). Relative COX-2 mRNA expression levels were measured. C, cells were pretreated with SB202190 or PD169319 and then stimulated with Ang II (100 nM) or EGF (25 ng/ml). After 4 h, COX-2 protein expression levels were measured. D, cells were pretreated with SB202190 (10 M, 1 h) prior to stimulation with Ang II (100 mM) or EGF (25 ng/ml). After 6 h, the medium was collected, and secreted 6-keto-PGF 1␣ was measured. 75 and 80% in cells pretreated with SB202190 and PD169316, respectively. Pretreatment of IEC-18 cells with SB202190 or PD169316 also reduced the level of COX-2 protein induced by Ang II (Fig. 6C). In contrast, EGF-induced COX-2 protein expression was largely unaffected by pretreatment with SB202190 or PD169316. There was a significant reduction in the level of Ang II-dependent PGI 2 produced from cells that were pretreated with SB202190 (Fig. 6D). In contrast, there was no significant decrease in EGF-dependent PGI 2 produced from cells that were pretreated with SB202190. These results establish that Ang II induces COX-2 expression in IEC-18 cells via a p38 MAPK -dependent, ERK-independent signaling pathway. Conversely, EGF induces COX-2 expression through an ERK-dependent signaling pathway.
Ca 2ϩ Signaling, but Not Protein Kinase C Activation, Mediates Ang II-dependent COX-2 Expression-Agonist binding of the G q -coupled AT 1 receptor activates phospholipase C␤, which results in hydrolysis of phosphatidylinositol 4,5-bisphosphate, producing diacylglycerol and inositol 1,4,5-trisphosphate (Ins(1,4,5)P 3 ) (73). Ins(1,4,5)P 3 leads to Ca 2ϩ mobilization, whereas diacylglycerol and Ca 2ϩ activate the classic isoforms of protein kinase C in IEC-18 cells (38). Activation of protein kinase C results in MEK-dependent phosphorylation of ERK1/2 in IEC-18 cells (39). Our results showing ERKindependent COX-2 expression by Ang II predict that protein kinase C signaling is not required for Ang II-dependent COX-2 expression. Pretreatment of cells with the specific protein kinase C inhibitor bisindolylmaleimide (74) did not significantly inhibit Ang II-dependent induction of COX-2 mRNA or protein in IEC-18 cells, whereas it did completely block COX-2 mRNA induction by direct activation of protein kinase C by phorbol esters (data not shown). This confirms that COX-2 expression by the AT 1 receptor is not mediated by protein kinase C.
Ins(1,4,5)P 3 -mediated Ca 2ϩ mobilization has been shown to induce immediate-early gene expression (75-77) and therefore could play a role in Ang II-dependent induction of COX-2. To test this, IEC-18 cells were treated with 2-APB prior to induc-FIG. 7. Calcium mobilization is required for Ang II-dependent COX-2 expression. Confluent serum-starved IEC-18 cells were either untreated or pretreated with 2-APB (75 M) 30 min prior to agonist stimulation. A, relative COX-2 mRNA expression levels were measured after stimulation with Ang II (100 nM) or EGF (25 ng/ml) for 1 h. B, COX-2 protein expression levels were measured after stimulation with Ang II or EGF for 4 h. C, cells were pretreated with 2-APB (75 M, 1 h) prior to stimulation with Ang II (100 mM) or EGF (25 ng/ml). After 6 h, the medium was collected, and secreted 6-keto-PGF 1␣ was measured. D, IEC-18 cells were grown on coverslips and loaded with Fura-2. Calcium mobilization was measured in cells stimulated with agonist (Ang II (100 nM) or EGF (25 ng/ml); arrows). Cells were either untreated or treated with 2-APB (75 M) prior to stimulation with Ang II. tion with either Ang II or EGF. 2-APB inhibits activation of Ins(1,4,5)P 3 receptors and store-operated calcium channels (78 -80). This attenuates receptor-mediated Ca 2ϩ release from internal stores and Ca 2ϩ influx through store-operated calcium channels. The maximal level of COX-2 mRNA induced by Ang II decreased by 83% in 2-APB-treated cells (Fig. 7A). This is significantly different from the 65% increase in maximal levels of COX-2 mRNA induced by EGF in 2-APB-treated IEC-18 cells. Pretreatment of cells with 2-APB also inhibited the induction of COX-2 protein by Ang II, but did not block COX-2 protein induction by EGF (Fig. 7B).
Inhibition of Ca 2ϩ signaling by 2-APB differentially affected PGI 2 production in response to Ang II and EGF (Fig. 7C). Ang II-dependent induction of PGI 2 was blocked in IEC-18 cells pretreated with 2-APB. EGF-dependent induction of PGI 2 was not decreased by 2-APB. The level of agonist-induced PGI 2 correlated with COX-2 expression.
We confirmed that stimulation of IEC-18 cells with Ang II resulted in a biphasic [Ca 2ϩ ] i response consisting of the release of intracellular Ca 2ϩ , followed by influx of extracellular Ca 2ϩ (Fig. 7D). Pretreatment of cells with 2-APB blocked the release of intracellular Ca 2ϩ by Ang II and the subsequent influx of extracellular Ca 2ϩ . Treatment of IEC-18 cells with EGF did not result in mobilization of Ca 2ϩ (Fig. 7D).

COX-2 Promoter Activation by Ang II Is Mediated by the ATF Element-Since stimulation of IEC-18 cells resulted in in-
creased transcription from the COX-2 gene, we wanted to map the COX-2 promoter elements that are responsive to Ang II. IEC-18 cells were transfected with luciferase reporter vectors containing successive 5Ј-deletions from the COX-2 promoter and then stimulated with Ang II or EGF (Fig. 8A). Transfected cells produced a 2.5 to Ͼ3-fold induction of luciferase in response to Ang II for reporter vectors containing the proximal promoter region from Ϫ963 to Ϫ80. Ang II-dependent induction of luciferase was lost in cells transfected with the reporter vector containing Ϫ40 to ϩ3 of the COX-2 promoter. This suggests that there is a critical region between Ϫ80 and Ϫ40 in the COX-2 promoter that is required for Ang II-dependent induction. This region contains a cAMP-responsive element (CRE)/ATF consensus sequence (Ϫ56 to Ϫ48) that has been shown to be a point of convergence for Ras and Rac signaling in NIH 3T3 cells (12)(13)(14)81). To test whether Ang II activates COX-2 transcription via the CRE/ATF promoter elements, cells were transfected with the COX-2 promoter-reporter vector containing mutations in the ATF-binding site (Ϫ371 to ϩ70 (ATF*)) (Fig. 8A). Treatment with Ang II did not result in luciferase induction. This suggests that signaling by Ang II converges onto the CRE/ATF element on the COX-2 promoter and that this cis-acting element is required for transcriptional activation of the COX-2 gene by Ang II.
We found that the most effective inhibitors of Ang II-induced COX-2 expression in IEC-18 cells targeted distinct signaling events. C. difficile toxin B inhibits Rho, Rac, and Cdc42 signaling; 2-APB blocks Ca 2ϩ mobilization and influx; and SB202190 inhibits p38 MAPK activity. To determine whether these inhibitors converge onto a common signaling pathway involving transcriptional regulation, we performed COX-2 promoter-reporter assays on cells that were either untreated or pretreated with 2-APB, toxin B, or SB202190 (Fig. 8B). Untreated cells showed a 2.4 Ϯ 0.1-fold increase in luciferase levels in response to Ang II. Pretreatment with 2-APB or SB202190 blocked Ang II-dependent luciferase induction by 1.00 Ϯ 0.05-and 1.2 Ϯ 0.2-fold, respectively. Toxin B significantly decreased Ang II-induced luciferase expression by 1.50 Ϯ 0.06-fold. These results suggest that signaling events that are regulated by Ca 2ϩ , small GTPases, and p38 MAPK have significant roles in Ang II-dependent transcriptional activation of the COX-2 promoter via the ATF cis-acting elements. DISCUSSION Epidemiological, experimental, and clinical studies have demonstrated that aspirin and other nonsteroidal anti-inflammatory drugs that inhibit cyclooxygenase activity reduce the risk of colorectal cancer (82). COX-2-derived prostanoids can contribute to tumor development through promotion of angiogenesis, inhibition of apoptosis, increased invasiveness/motility, and modulation of inflammation and immune responses (83). Therefore, elucidating the regulation of COX-2 expression by endogenous receptors in intestinal epithelial cells is of great interest. Using the IEC-18 cell line as a model of intestinal epithelial cells, we have shown that Ang II is a potent stimulator of COX-2 expression. Agonist binding of the endogenous AT 1 receptor initiates multiple signaling events that result in an increase in the rate of COX-2 gene transcription and changes in the stability of COX-2 mRNA. The specific family of heterotrimeric G-proteins that interact with the AT 1 receptor (other than G q ) has yet to be characterized in IEC-18 cells. However, it is unlikely that the G i/o family of heterotrimeric G-proteins plays a role in Ang II-dependent COX-2 expression because prior treatment of cells with pertussis toxin did not block COX-2 induction by Ang II. This leaves the G q and G 12/13 families of heterotrimeric G-proteins as the most likely candidates for mediating the signals from the AT 1 receptor.

FIG. 8. Transcriptional activation of the COX-2 promoter by
Ang II is through the ATF element. IEC-18 cells were transiently transfected with COX-2 promoter-luciferase reporter plasmids. Cells were either unstimulated or exposed to Ang II (100 nM, 5 h) prior to measuring luciferase expression relative to Renilla luciferase expression (control for transfection efficiency). A, a series of reporter plasmids containing successive 5Ј-deletions of the COX-2 promoter were used. Tyrosine kinase receptors induce COX-2 expression primarily through activation of Ras and Rac, with subsequent phosphorylation of Elk-1 by ERK1/2 and of c-Jun by JNK (12)(13)(14). Therefore, it was not surprising that EGF induced COX-2 expression in IEC-18 cells. An early event in colorectal cancer is the oncogenic mutation of K-ras (84). This results in up-regulation of EGF receptor signaling by increased expression of the EGF receptor and its associated ligands (85). The resulting autocrine loop promotes cell proliferation and resistance to apoptosis. The combined signaling by genetic mutations (i.e. K-Ras) and epigenetic events (Ang II) could significantly increase COX-2 expression in intestinal epithelial cells, resulting in increased potential of tumorigenesis.
Our previous studies have demonstrated a crucial role for small GTPases in COX-2 expression (47,59,86). Inhibition of small GTPases by C. difficile toxin B blocks both Ang II-and EGF-dependent induction of COX-2 mRNA and protein and PGI 2 production. In IEC-18 cells, Ang II preferentially activates Cdc42. This corresponds to increased filopodial formation by Ang II. EGF preferentially activates Rac, which is consistent with the prominent thickening of cortical actin/lamellipodia. Rac and Cdc42 share a number of common effector targets, including p21 PAK . This protein kinase can activate MAPKs. However, despite the shared requirement for small GTPase signaling, it is evident that COX-2 induction by Ang II and EGF is mediated by each receptor's activation of distinct signaling pathways.
The MAPKs are a family of highly evolutionarily conserved protein kinases (ERK, JNK, and p38 MAPK ) connecting cellsurface receptors to critical regulatory targets within cells. They regulate important cellular processes, including gene expression, cell proliferation, and cell motility (87). ERK, JNK, and p38 MAPK play a role in COX-2 expression in an agonistand cell-specific manner. Several intestinal tumor studies have suggested that MEK and its downstream ERK signaling are required for both increased transcription and stability of COX-2 mRNA and K-Ras-induced COX-2 expression (88, 89).
Activated ERK can directly phosphorylate transcription factors such as Elk-1 and Sap-1, inducing transcription from immediate-early genes (i.e. c-fos) via activator protein-1 (90,91). Activator protein-1 can bind to the CRE/ATF site on the COX-2 promoter (20,58). However, in IEC-18 cells, ERK/activator protein-1-mediated signaling does not seem to be the predominant pathway for Ang II-dependent COX-2 expression since inhibition of ERK activity did not significantly affect Ang IIinduced COX-2 expression. In contrast, inhibition of ERK activity significantly decreased COX-2 expression by EGF. EGF stimulates a sustained period of ERK activation compared with transient activation by Ang II. Extending the time that ERK is active in cells has been shown to increase the level of c-fos expression, leading to increased DNA synthesis and mitogenesis (92).
A significant difference in signaling between Ang II and EGF is activation of p38 MAPK . Ang II transiently induces the phosphorylation of p38 MAPK , followed by a transient increase in the phosphorylation of the transcription factor ATF-2. We did not detect EGF-dependent phosphorylation of p38 MAPK . Jun family transcription factors can form heterodimers with ATF-2 and thereby increase COX-2 transcription by binding to the CRE/ ATF promoter element (93). Inhibition of p38 MAPK activity significantly blocked Ang II-dependent increases in COX-2 mRNA and protein and PGI 2 production in IEC-18 cells. COX-2 promoter assays demonstrated a loss of induction by Ang II in cells pretreated with the specific p38 MAPK inhibitor SB202190. Mutation of the critical CRE/ATF element located in the proximal region of the COX-2 promoter also blocked Ang II-dependent induction of luciferase, suggesting a link between Ang II-dependent activation of p38 MAPK and COX-2 transcriptional activation through the CRE/ATF promoter element.
Another major difference in signaling by Ang II compared with EGF in IEC-18 cells involves Ca 2ϩ signaling. Agonist binding to the AT 1 receptor activates G q -dependent phospholipase C␤, generating the second messenger Ins(1,4,5)P 3 , resulting in Ca 2ϩ mobilization. Activation of Ca 2ϩ -dependent FIG. 9. Ang II and EGF use distinct signaling pathways to induce COX-2 expression in IEC-18 cells. This scheme summarizes our results demonstrating that Ang II and EGF initiate distinct pathways leading to COX-2 mRNA, protein, and activity in intestinal epithelial cells. Ang II binding to the AT 1 receptor (AT 1 R) stimulates G q -linked phospholipase C␤ (PLC␤), resulting in activation of protein kinase C (PKC) and Ca 2ϩ signaling. The AT 1 receptor also activates Cdc42 (possibly through G 12 ), leading to ATF-2 phosphorylation via PAK1 and p38 MAPK . Ca 2ϩ signaling results in phosphorylation of ATF-2 (or possibly CREbinding protein (CREB)). Activation of the EGF receptor (EGFR) by EGF results in Ras and Rac activation. Elk-1 phosphorylation by ERK is mediated by the Ras/ Raf/MEK kinase cascade. Rac activation leads to phosphorylation of c-Jun via the PAK1/JNK cascade. These activated transcription factors (ATF-2, c-Jun, and CREbinding protein) stimulate the COX-2 promoter, resulting in COX-2 expression and prostaglandin production. The scheme also shows the pharmacological tools (indicated by boxes) used to block (Ќ) specific steps in this study. The dashed arrows imply the existence of intermediary steps leading to the subsequent event, whereas solid arrows represent direct consequences of the previous effector. PI3K, phosphatidylinositol 3-kinase.
protein kinases can result in immediate-early gene expression (i.e. c-fos) (77) via cis-acting promoter elements such as CRE, activator protein-1, and the serum response element (58,(75)(76)(77). Inhibition of release of Ca 2ϩ stores by Ins(1,4,5)P 3 receptors and influx of Ca 2ϩ through store-operated calcium channels by 2-APB blocked COX-2 induction by Ang II. COX-2 promoter assays showed a complete block in Ang II-dependent transcriptional activation in cells treated with 2-APB. These results are reminiscent of our previous studies in fibroblasts showing that Ca 2ϩ signaling mediated either by G q or by treatment with thapsigargin leads to Rho-dependent transcriptional activation of the COX-2 promoter (59). However, in epithelial cells, Ca 2ϩ -dependent COX-2 expression by a G-protein-coupled receptor that is coupled to G q appears to be Cdc42-dependent. Additionally, Guo et al. (58) reported Ca 2ϩ -dependent COX-2 expression by bombesin in a rat epithelial cell line that was transfected with a bombesin receptor expression vector, but the role of small GTPases was not determined. EGF did not induce Ca 2ϩ mobilization in IEC-18 cells, and inhibition of Ca 2ϩ signaling by 2-APB did not prevent EGF-dependent COX-2 expression. These results demonstrate, for the first time, a specific requirement for Ca 2ϩ signaling in Ang II-dependent transcriptional activation of the COX-2 gene, which results in COX-2 expression and production of PGI 2 .
We have developed a model representing the distinct signaling pathways that are differentially initiated by Ang II and EGF and that result in COX-2 expression in IEC-18 cells (Fig.  9). Signaling by EGF results in Ras and Rac activation, leading to sustained activation of ERK. Ang II activates Cdc42 and Ca 2ϩ signaling, leading to p38 MAPK -mediated increases in COX-2 gene transcription and stabilization of COX-2 mRNA, resulting in significant induction of COX-2 expression. A salient feature of this model is the potential for synergistic COX-2 expression by the combined signaling of Ang II and EGF. In fact, Ang II and EGF synergistically induce COX-2 expression in IEC-18 cells. 2 We are currently investigating the molecular basis for this synergy.
Traditionally, Ang II is recognized for its role in the regulation of systemic blood pressure and body fluid homeostasis. There is also considerable evidence indicating that Ang II regulates ion transport in intestinal epithelial cells (36,37,94). In addition, the components of the renin/angiotensin system (angiotensinogen, renin, and ACE) mRNAs have been found in human colonic mucosa (94). Immunohistochemical examination of colonoscopic biopsies revealed renin and ACE to be localized in vessel walls, mesenchymal cells in the lamina propria, and parts of the surface epithelium (95). Therefore, it is likely that Ang II is normally produced in intestinal tissue and acts as a local mediator of intestinal functions. We have demonstrated previously that Ang II induces DNA synthesis and proliferation of IEC-18 cells (39). Our present finding that Ang II potently induces COX-2 expression in IEC-18 cells, together with previous reports of ACE inhibitors decreasing tumor development and angiogenesis (46,96), suggests the possibility that Ang II and EGF act as growth factors with carcinogenic potential through COX-2 expression and subsequent production of prostaglandins.