Scholarly article on topic 'Differential Regulation of Cyclooxygenase-2 in Nontransformed, Ras-Transformed Intestinal Epithelial Cells'

Differential Regulation of Cyclooxygenase-2 in Nontransformed, Ras-Transformed Intestinal Epithelial Cells Academic research paper on "Biological sciences"

CC BY-NC-ND
0
0
Share paper
Academic journal
Neoplasia
OECD Field of science
Keywords
{Ras / RhoA / "transforming growth factor p" / COX-2 / "epidermal growth factor"}

Abstract of research paper on Biological sciences, author of scientific article — Jianguo Du, Bo Jiang, John Barnard

Abstract To determine signaling pathways responsible for modulation of COX-2 expression in nontransformed, transformed epithelial cells, we studied a rat intestinal epithelial (RIE) cell line expressing constitutively active Ras, RhoA. Expression of COX-2 protein was higher in RIE-RhoA(63L) (four-fold), RIE-Ras(12V) (seven-fold) cells than in parental cells. Prior work suggests that Ras hyperactivity induces the expression of transforming growth factor (TGF) β, increases epidermal growth factor (EGF)-related peptide signaling-possible mechanisms for increased COX-2 expression. Expression of COX-2 was stimulated by TGFβ, TGFα in RIE, RIE-Rho(63L) cells, but not further stimulated in RIE-Ras(12V) cells. PD153035, an inhibitor of EGF receptor tyrosine kinase, PD98059, an inhibitor of Erk, attenuated COX-2 expression in RIE, RIE-RhoA(63L). However, the high levels of COX-2 expression in RIE-Ras(12V) cells were not inhibited by either compound. Titration with a panneutralizing anti-TGFβ antibody did not decrease COX-2 in RIE-Ras(12V) cells, even with concurrent EGFR inhibition. Thus, stimulation of the EGF receptor is important in the modulation of COX-2 expression in nontransformed RIE, RIE-RhoA(63L) cells. In Rastransformed cells, signaling by additional Ras effector pathways, perhaps the RhoA pathway, must be invoked. Identification of these pathways is critical for therapeutic manipulation of COX-2 expression.

Academic research paper on topic "Differential Regulation of Cyclooxygenase-2 in Nontransformed, Ras-Transformed Intestinal Epithelial Cells"

RESEARCH ARTICLE

Neoplasia • Vol. 7, No. 8, August 2005, pp. 761 -770 761

www.neoplasia.com

Differential Regulation of Cyclooxygenase-2 in Nontransformed and Ras-Transformed Intestinal Epithelial Cells1

Jianguo Du, Bo Jiang and John Barnard

Department of Pediatrics, Center for Cell and Vascular Biology, Columbus Children's Research Institute and The Ohio State College of Medicine and Public Health, Columbus, OH, USA

Abstract

To determine signaling pathways responsible for modulation of COX-2 expression in nontransformed and transformed epithelial cells, we studied a rat intestinal epithelial (RIE) cell line expressing constitutively active Ras and RhoA. Expression of COX-2 protein was higher in RIE-RhoA(63L) (four-fold) and RIE-Ras(12V) (seven-fold) cells than in parental cells. Prior work suggests that Ras hyperactivity induces the expression of transforming growth factor (TGF) B and increases epidermal growth factor (EGF)-related peptide signaling—possible mechanisms for increased COX-2 expression. Expression of COX-2 was stimulated by TGFB and TGF« in RIE and RIE-Rho(63L) cells, but not further stimulated in RIE-Ras(12V) cells. PD153035, an inhibitor of EGF receptor tyrosine kinase, and PD98059, an inhibitor of Erk, attenuated COX-2 expression in RIE and RIE-RhoA(63L). However, the high levels of COX-2 expression in RIE-Ras(12V) cells were not inhibited by either compound. Titration with a pan-neutralizing anti-TGFB antibody did not decrease COX-2 in RIE-Ras(12V) cells, even with concurrent EGFR inhibition. Thus, stimulation of the EGF receptor is important in the modulation of COX-2 expression in nontransformed RIE and RIE-RhoA(63L) cells. In Ras-transformed cells, signaling by additional Ras effector pathways, perhaps the RhoA pathway, must be invoked. Identification of these pathways is critical for therapeutic manipulation of COX-2 expression. Neoplasia (2005) 7, 761-770

Keywords: Ras, RhoA, transforming growth factor p, COX-2, epidermal growth factor.

observed in human [2] and rodent intestinal tumors [3,4] when compared with normal adjacent mucosa, whereas COX-1 expression is not differentially expressed. Among the consequences of elevated COX-2 expression and increased prostaglandin production are stimulation of angiogenesis, inhibition of apoptosis, and stimulation of cellular proliferation, all of which favor tumorigenesis. Inhibitors of COX-2 activity such as aspirin, nonsteroidal anti-inflammatory agents (NSAIDs), and specific COX-2 antagonists significantly reduce the occurrence of colorectal carcinoma in both animal and human studies (reviewed in Ref. [5]). In addition, experimental animal and human data show that individuals with colon adenomas occurring in the context of a defect in the adenomatous polyposis coli (APC) gene product show a significant reduction in the number of adenomas following treatment with NSAIDs or more specific COX-2 inhibitors [6,7]. Collectively, these observations indicate that understanding the regulation of COX-2 activity is linked to unraveling the pathogenesis of colorectal cancer and to identifying strategies for its treatment and prevention.

In cell culture systems, a wide variety of growth factors and cytokines induce COX-2 expression, including epidermal growth factor (EGF)-related peptides [8], keratinocyte growth factor [9], basic fibroblast growth factors [10], transforming growth factor p (TGFp) [11], hepatocyte growth factor [12], platelet-derived growth factor [13], interferon g [14], and gastrin [15], among many others. Inasmuch as many of these growth factors are overexpressed in colorectal polyps and cancers, it is perhaps not unexpected that these same tumors frequently demonstrate markedly elevated levels of COX-2 and prostaglandins. Many growth factors stimulate cell surface membrane-associated tyrosine kinase receptors that activate cellular Ras and its associated effector pathways. Activating mutations in the K-ras protooncogene are found in approximately 50% of colorectal cancers, and multiple cell culture

Introduction

Cyclooxygenases (COXs), or prostaglandin-endoperoxide synthetases, catalyze the rate-limiting step in the synthesis of prostaglandins and other eicosanoids from arachidonic acid [1]. COX-1 and COX-2 are recognized as constitutive and inducible COX isoenzymes, respectively. Increased expression of COX-2 and prostaglandins has been widely

Address all correspondence to: John Barnard, Columbus Children's Research Institute, W502 700 Children's Drive, Columbus, OH 43205. E-mail: barnardj@pediatrics.ohio-state.edu 1This work was supported by R01 DK57128 (J.B.). Received 5 October 2004; Revised 29 April 2005; Accepted 4 May 2005.

Copyright © 2005 Neoplasia Press, Inc. All rights reserved 1522-8002/05/$25.00 DOI 10.1593/neo.04652

studies have reported the overexpression of COX-2 in the context of the mutant activation of Ras [16-18]. In some cell culture models, this induction appears to occur through increased COX-2 transcription [16] and an increase in the half-life of the COX-2 mRNA by activation of protein kinase B1 [19]. In addition to the activation of Ras, activation of other small GTPases such as RhoA may also increase COX-2 transcription by a Ras-independent pathway [17,20].

Clarification of the signaling pathways by which oncogenic Ras upregulates COX-2 may be important in understanding the pathogenesis of aberrant growth-regulatory pathways in the intestine. In its activated, GTP-bound state, Ras stimulates a complex network of downstream pathways potentially involved in COX-2 upregulation [21], including stimulation of EGF-related peptide synthesis, activation of EGFRs [22], and activation of TGF(3 transcription [23]. Activation of EGF signaling by Ras is especially interesting in light of recent studies highlighting synergy between the inhibition of COX-2 activity and the blockade of the EGFR signaling in models of colorectal neoplasia [24].

Herein, we examine the differential regulation of COX-2 expression in transformed and nontransformed intestinal epithelial cell model systems. Emphasis is given to the importance of epidermal growth factor receptor (EGFR) intrinsic tyrosine kinase activity and TGFb signaling in the regulation of COX-2 expression by Ras and Rho GTPases pathways in RIE-1 intestinal epithelial cells.

Materials and Methods

Cell Lines and Reagents

RIE-1 rat intestinal epithelial cells were obtained from Ken Brown (Cambridge, UK) and were maintained in Dulbecco's modified Eagle's medium (DMEM) supplemented with 5% fetal calf serum [25]. Dr. Robert Coffey (Vanderbilt University) kindly supplied RIE-Ras(12V) and RIE-Rho cells, which have been described in more detail elsewhere [26]. Briefly, RIE-Ras(12V) cells were stably transfected with pSV2-H-ras(12V) containing human sequences encoding the transforming H-Ras(12V) protein. RIE-RhoA(63L) cells were established by stable transfection with mutant active RhoA(63L) (pZIP-RhoA-63L). Unless otherwise noted, the cells used in our studies were propagated from pools of multiple G418-resistant clones (> 50). Ras hyperactivity in the RIE-Ras(12V) cells was confirmed based on binding of GTP-Ras to immobilized Raf-1, as previously reported by our laboratory [27]. In selected experiments, double stable transfectants were used. For these studies, RIE-Ras(12V) cells were transfected with dominant-negative mutant RhoA(N19) (kindly supplied by Dr. Gary Bokoch). RhoA(N19) was subcloned into pcDNA3.1/Zeo and transfected into RIE-Ras(12V) cells using the Lipofectamine 2000 protocol supplied by Invitrogen Life Technologies (Carlsbad, CA). Stable clones were double-selected in Zeocin and G418. Expression of RhoA(N19) was verified by Western blot analysis using both anti-HA and anti-RhoA antibodies. Multiple clones were selected for testing.

EGF was obtained from Calbiochem-Novabiochem Corp. (San Diego, CA). TGF(31 was obtained from R&D Systems (Minneapolis, MN). The following signaling inhibitors were obtained from Calbiochem-Novabiochem Corp.: PD153035 (EGFR tyrosine kinase inhibitor), PD98059 (MEK inhibitor), and Y-27632 (p160ROCK inhibitor). U0126 was obtained from Promega Corporation (Madison, WI). EKI-785 was a kind gift from Dr. Robert Coffey. A rabbit polyclonal antihuman COX-1 antibody and a goat polyclonal anti-human COX-2 antibody were obtained from Santa Cruz Biotechnology (Santa Cruz, CA). A mouse monoclonal anti-cyclin D1 antibody was obtained from Santa Cruz Biotechnology.

Cell Proliferation Assays

3H thymidine incorporation assays were carried out in 24-well tissue culture plates as described previously [28]. Cells were seeded at a density of 20,000 cells/well, allowed to attach for at least 24 hours, then treated as described in the figure legend. A 3H thymidine (Amersham Life Sciences, Arlington Heights, IL) pulse (1 mCi/well) was provided between the 18th and 21st hours of treatment. Radioactivity incorporated into trichloroacetic acid-insoluble material was determined by scintillation counting and results are presented as the mean ± SEM for triplicate or quadruplicate measurements. Experiments were repeated at least three times.

Immunoprecipitation and Western Blot Analysis

Cells in 12-well plates or 100-mm culture dishes were solubilized in lysis buffer (20 mM Tris-HCl, pH 7.4, 120 mM NaCl, 100 mM NaF, 200 mM Na3VO4, 4 mM PMSF, 10 mg/ml leupeptin, 10 mg/ml aprotinin, 0.5% NP-40, and 2 mM ben-zamidine) for 30 minutes at 4°C. The lysate was clarified by centrifugation and the supernatant was incubated with primary antibody overnight at 4°C. Immunoprecipitates were incubated with protein A-agarose for 1.5 hours and washed repeatedly in phosphate-buffered saline containing 0.05% NP-40. The immune complexes were eluted in SDS sample buffer and transferred onto PVDF membranes in 25 mM Tris, 192 mM glycine, and 20% methanol buffer at 30 V overnight. Membranes were then blocked, probed with the diluted antibody of interest, and then incubated with horseradish peroxidase-conjugated anti-IgG. The ECL Plus detection system (Amersham Life Sciences) was used to detect antigen -antibody complexes.

Prostaglandin E2 (PGE2) Assay

Cells were plated in 24-well plates and grown in serum-containing medium for 24 to 28 hours. The medium was replaced with serum-free DMEM for 24 hours, which was then collected for measurements of basal levels of secreted PGE2 by enzyme-linked immunoabsorbent assay (ELISA; Cayman Chemical, Ann Arbor, MI).

Northern Analysis

Total cellular RNA was extracted using Trizol. RNA samples (20 mg/lane) were loaded into 1% agarose/formaldehyde gels and separated by electrophoresis and blotted onto nitrocellulose membranes. COX-1 and COX-2 cDNA probes

were labeled by random primer extension using Redivue with a-32P dCTP and the Rediprime DNA labeling system from Amersham Life Sciences. After hybrizidation and washing, the membranes were subjected to autoradiography. The integrity of the RNA was assessed by 18 rRNA signals.

Results

Expression and Activity of COXs in RIE, RIE-Ras(12V), and RIE-Rho(63L) Cells

COX-2 expression is increased in cells transformed by on-cogenic activation of Ras [16,17,29,30]. Induction of COX-2 by RhoA activation has recently been reported, but is less well studied [17]. As previously shown, COX-2 expression in RIE-1 cells stably transfected and transformed with activated Ras showed a marked increase by immunoblot analysis (Figure 1). The level of COX-2 expression was also examined in cells that overexpress mutant-activated RhoA. These cells, which are nontransformed, exhibit four-fold greater COX-2 expression than parental RIE-1 cells, but less than that observed for transformed RIE-Ras(12V) cells. Expression of COX-1, a constitutive COX, is equivalent in these three cell lines. To supplement these observations, the level of PGE2, a product of COX activity, was measured to determine if the level of COX-2 protein observed in the RIE-Ras(12V) and RIE-RhoA cells lines is associated with an increase in prostaglandin synthesis. The level of PGE2 in RIE-Ras(12V) cells was 70-fold greater than the parental RIE-1 cell line, whereas levels in RIE-RhoA(63L) cells were two-fold greater. Although

PGE2 has previously shown to be the predominant prostaglandin synthesized in the context of Ras transformation, similar detailed studies have not been performed in cells over-expressing RhoA and it is plausible that other prostaglandins are synthesized in greater abundance. Notwithstanding, these results confirm increased expression of COX-2 protein and activity in both RIE-Ras(12V) and RIE-RhoA(63L) cells, when compared with the parental RIE-1 cell line.

In nontransformed cells, proliferation is regulated, in part, by cell density and confluence, whereas in transformed cells, contact inhibition is lost. Expression of EGF-related peptides, EGFRs, and EGF-signaling proteins is also regulated by epithelial cell density [31,32]. Therefore, we examined the expression of COX-2 protein as a function of RIE-1, RIE-Ras(12V), and RIE-RhoA(63L) cell confluence (Figure 2). In the nontransformed RIE-1 and RIE-RhoA(63L) cells, expression of COX-2 was decreased by approximately 60% when measured 48 hours after cells reached confluence. By contrast, the transformed RIE-Ras(12V) cells showed little difference in their high level of COX-2 expression, even at 72 hours—a time when the cells have become densely confluent. Thus, the high levels of COX-2 expressed in RIE-Ras(12V) cells occur independent of cellular confluence, whereas contact inhibition in nontransformed cells is associated with decreased COX-2 expression.

Regulation of COX-2 Expression by EGF-Related Peptides and TGFp

It is well established that COX-2 levels are increased by EGF-related peptides and TGFb in a variety of cell lines,

I 7 5 6

§3 û 2 О 1

□ RIE-1

■ Ras(12V) Ш RhoA(63L)

□ RIE-1 ■ RIE-Ras(12V) H RIE-RhoA(63L)

Figure 1. Expression of COX-2 protein and determination of PGE2 levels in intestinal epithelial cell lines. Total cellular lysates from RIE-1, RIE-Ras(12V), and RIE-RhoA(63L) cells growing in subconfluent cultures were isolated, and the basal expression of COX-2 and COX-1 was examined by Western blot analysis. The bar graph shows the relative expression of COX-2 in each cell line determined by densitometry. The graph on the right shows PGE2 levels in serum-free cell culture medium collected from each cell line. PGE2 levels were measured by ELISA as described in the Materials and Methods section.

RIE-1 Ras(12V) RhoA(63L)

24h 48h 24h 48h 72h 24 h 48h

co * to 1

a. & 08

□ RIE-1

■ RIE-Ras(12V) E RIE-RhoA(63L)

24 48 24 48 72 24 48

Figure 2. Regulation of COX-2 expression as a function of cell density. Cultures of RIE-1, RIE-Ras(12V), and RIE-RhoA(63L) cells were grown to the point at which they just reached confluence. This time point was considered time 0. Continued postconfluent growth of cells was allowed to occur and, at 24-hour intervals thereafter, cell lysates were isolated, separated by sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) in a single gel, and subjected to Western blot analysis for COX-2 expression. Because RIE-Ras(12V) cells did not show downregulation of COX-2 expression 48 hours after reaching confluence, an additional 24 hours of postconfluence growth was permitted. Quantified results, obtained by densitometry, are shown in the bar graph.

including the RIE-1 line used herein. To investigate the relative induction of COX-2 in RIE-1 cells overexpressing activated Ras(12V) and RhoA(63L), we compared the induction of COX-2 in response to treatment with TGFa and TGF(3. Consistent with prior reports [8], COX-2 was rapidly and markedly induced in RIE cells following treatment with TGFa, approximately 13-fold after 8 hours of treatment (Figure 3). Similarly, COX-2 was induced in the nontrans-formed RIE-RhoA(63L) cell line, albeit less than the parental cell line. The RIE-Ras(12V) cell line, which already has a markedly increased level of COX-2, showed no further induction in response to treatment with TGFa.

Although the level of COX-2 induction by TGF(3 was less prominent than for TGFa, qualitatively, the observations are similar [i.e., COX-2 is rapidly induced in the nontrans-formed cell lines, but not further induced in the transformed RIE-Ras(12V) cell line] (Figure 3). These results indicate that increased levels of COX-2 expression characteristic of RIE-Ras(12V) cells cannot be further induced in response to autocrine growth-regulatory factors that are commonly overexpressed in transformed intestinal epithelial cells. The data suggest that overexpression of EGF-related peptides, such as TGFa or TGF(3, is possibly a major driving force for COX-2 overexpression in RIE-Ras(12V) cells, and that

RhoA(63L) Ras(12V)

12 10 8 6 4 2 0

□ RIE-1 I RIE-Ras(12V) i RIE-RhoA(63L)

0 2h 4h 8h

0 2h 4h 8h

Figure 3. Induction of COX-2 protein expression by TGFa andTGF/3. Rapidly growing cultures of RIE-1, RIE-Ras(12V), and RIE-RhoA(63L) cells were treated with 2 ng/ml TGFa or 2 ng/ml TGFp 1 for the indicated time period. Total cellular lysates were collected, and COX-2 protein expression was examined by Western analysis using a COX-2-specific antibody. These experiments were performed three separate times and representative results are shown. As shown in Figures 1 and 2, the relative level of COX-2 expression is different for each cell line. In this figure, autoradiographic exposures for each cell line were varied to optimize exposure levels and results were quantified by densitometry.

120 100

б 80 a.

60 40 20 0

- RIE-Ras

AG9 0.05

PD153035, цМ PD153035 цМ

0 .01 .02 .05 1 25

Anti-pan TGFP COX-2 P-actin

PD153035 ЦМ EKI758 цМ

Ras(12V)

Figure 4. EGFR tyrosine kinase inhibitors decrease 3H thymidine incorporation but not COX-2 levels in Ras-transformed cells. RIE-Ras(12V) cells were rapidly treated with the indicated concentration of the tyrosine kinase inhibitor, PD153035, in serum-free medium. Eighteen hours later, a 3-hour pulse of 3H thymidine (1 pCi/well)) was provided, followed by measurements of thymidine incorporation into TCA-insoluble materials, as described in the Materials and Methods section. AG9, an inactive structural analogue ofAG1478, was used as a control in these experiments, and identical experiments with AG1478 were performed (not shown). Results, shown in the top panel, are expressed as percent of 3H thymidine incorporation in cells treated with dimethyl sulfoxide alone. Rapidly growing cultures of RIE-1, RIE-Ras(12V), and RIE-RhoA(63L) cells were treated for 24 hours with the highly specific tyrosine kinase inhibitor, PD153035, at concentrations between 0.01 and 25 mg/ml. Note that the inhibitor concentrations varied for each cell line. Total cellular lysates were collected and COX-2 protein expression was examined by Western blot analysis using a COX-2-specific antibody. These experiments were performed three separate times and representative results are shown. As shown in Figures 1 and 2, the relative level of COX-2 expression is different for each cell line. In this figure, autoradiographic exposures for each cell line were varied to optimize exposure levels. Because of the apparent insensitivity of RIE-Ras(12V) cells to PD153035, its activity as a tyrosine kinase inhibitor was checked in separate experiments, under identical conditions, using an alternate EGFR tyrosine kinase inhibitor, EKI-758. Again, reduced levels of COX-2 are not observed (bottom right). The inset shows that at a concentration of 1 pg/ml, PD153035 markedly inhibited EGFR phosphorylation, but not COX-2 expression, as shown by the arrow. In the experiment shown in the right panel, cells were grown to near confluence and treated with a pan-neutralizing anti-TGF/3 antibody for 24 hours. COX-2 protein expression in each cell line was measured by Western blot analysis.

addition of these peptides does not further increase the expression of COX-2.

Characterization of EGFR Tyrosine Kinase Inhibition in the RIE-Ras(12V) Cell Line

In our studies, the RIE-1 intestinal epithelial cell line is utilized as a model for normal intestinal epithelial cells, whereas RIE-Ras(12V) cells are used as a model for transformed cells. Prior studies have shown that parental RIE-1 cells are morphologically nontransformed and do not form colonies in soft agar, nor do they form tumors in nude mice [26]. In contrast, RIE cells that overexpress mutant-activated Ras(12V) assume a transformed morphology, form colonies in soft agar, and rapidly form tumors in nude mice [26]. These characteristics have been attributed, in part, to production of EGF-related peptides. To determine the effect

of EGFR kinase inhibition on the growth and morphology of RIE-Ras(12V) cells, PD153035 (AG1517) was used in 3H thymidine incorporation assays. As shown in Figure 4, DNA synthesis was decreased by concentrations of the tyro-sine kinase inhibitor between 0.05 and 2 mM, whereas 2 mM AG9, an inactive tyrphostin analogue, had no significant effect. Similar results were observed with the tyrosine kinase inhibitor, tyrphostin (AG1478). No effect was observed on the morphology of cells. Based on the near equivalence of tyrphostin and PD153035, further experiments described in this manuscript were performed using the latter compound. The inset in Figure 4 shows a marked decreased in the tyrosine phosphorylation of EGFR in RIE-Ras(12V) cells at a PD153035 concentration of 1 mM. Collectively, these concentrations are consistent with effective tyrosine kinase inhibitor concentrations reported

previously [33-35], and substantiate the design of subsequent experiments on COX-2 expression reported herein. The results also show the critical dependence of increased EGFR tyrosine kinase activity on cellular proliferation in RIE-Ras(12V) cells.

Effect of EGFR Tyrosine Kinase Inhibition and TGF/3 Inhibition on COX-2 Expression

To test the hypothesis that COX-2 overexpression in RIE-Ras(12V) cells is due to the high level of expression of EGF-related peptides characteristic of these cells, RIE-1, RIE-Ras(12V), and RIE-RhoA(63L) cells were treated for 24 hours with PD153035 at concentrations known to decrease the phosphorylation of EGFR (Refs. [33-35]; Figure 4, inset) and inhibit DNA synthesis RIE-Ras(12V) cell lines (Figure 4, top panel). The results show that COX-2 expression in parental RIE-1 cells is highly sensitive to the EGFR tyrosine kinase inhibitor, with half-maximal inhibition occurring between 0.02 and 0.05 mM PD153035 (Figure 4). RIE-Rho cells are somewhat less sensitive, with a half-maximal inhibition between 0.05 and 0.5 mM. By contrast, COX-2 levels are resistant to EGFR tyrosine kinase inhibition up to 5 mM in transformed cells overexpressing mutant-activated Ras. This relative (100-fold) resistance in RIE-Ras(12V) cells was confirmed by treatment with EKI-758, a structurally distinct EGFR tyrosine kinase inhibitor (Figure 4, bottom panel). Collectively, these results indicate that the basal levels of COX-2 in RIE-1 - and RIE-RhoA(63L)-expressing cells are maintained, in part, by autocrine stimulation of the

EGFR tyrosine kinase activity. Despite the high levels of production of EGF-related peptides in RIE cells that stably overexpress activated Ras, stimulation of the EGFR by these autocrine growth factors is not independently responsible for the maintenance of high levels of COX-2 expression characteristic of this cell line.

COX-2 expression is also stimulated by TGFb in many cell types, potentially explaining the increased COX-2 expression observed in Ras-transformed cells, as Ras activates the transcription of the TGFb gene [23]. To confirm this in RIE cells, TGFb levels in 72-hour conditioned medium were determined by ELISA and were approximately increased 10-fold in RIE-Ras(12V) cells compared with the parental RIE cell line (RIE = 21.1 pg/ml per 104 cells versus Ras(12V) = 191.9 pg/ml per 104 cells; RIE-RhoA(63L) cells = not performed). A pan-neutralizing anti-TGFb antibody that recognizes all three isoforms of mammalian TGFb was added to the cell culture medium in each of the three cell lines under study in an attempt to interrupt autocrine TGFb-stimulated COX-2 expression. Addition of the pan-neutralizing antibody (1 mg/ml) for 24 hours did not diminish COX-2 expression in any of the cell lines (Figure 4), although this was sufficient to reduce basal levels of Smad2 phosphorylation (not shown). Simultaneous treatment of RIE-Ras(12V) cells with pan-neutralizing anti-TGFb antibody and PD153035 also did not reduce COX-2 expression. This result indicates that autocrine production of TGFb does not contribute to the basal levels of COX-2 expression in nontransformed or transformed cell lines.

PD153035 1 y.M

PD98059 25 jiM

PD153035

PD98059

| □ RIE-1 ■ RIE-Ras(12V) B ME-RhoA(63L) |

Figure 5. Effect of Erk pathway inhibition on COX-2 expression. Rapidly growing, subconfluent RIE-1, RIE-Ras(12V), RIE-RhoA(63L), and RIE-Ras(12V)/ RhoA(N19) cells were treated for 24 hours with the indicated concentration of the EGFR tyrosine kinase inhibitor, PD153035, and the MEK inhibitor, PD98059 (left panel), or with U0126, a more potent inhibitor of MEK (right panel). Total cellular lysates were isolated from each cell line and loaded onto SDS-PAGE gels. Western blot analysis was performed for COX-2 and cyclin D1 expression. This experiment was repeated three times with similar results.

Y-27632 COX-2 cyclin D1 P-actin

Figure 6. Effect of Rho inhibition on COX-2 expression. Rapidly growing, subconfluent RIE-1, RIE-Ras(12V), and RIE-RhoA(63L) cells were treated for 24 hours with the indicated concentration of Y-27632, an inhibitor of the p160ROCK limb of the RhoA pathway. Total cellular lysates were isolated from each cell line and loaded onto SDS-PAGE gels. Western blot analysis was performed for COX-2 and cyclin D1 expression. This experiment was repeated three times with similar results.

cells and promote the acquisition of growth characteristics highly reminiscent of Ras-transformed cells [37]. Additionally, we found that dominant-negative expression of RhoA in RIE-Ras(12V) cells markedly inhibits elevated COX-2 expression and increased PGE2 levels characteristic of cells expressing hyperactive Ras, but did not affect levels of other proteins relevant to cellular proliferation, such as cyclin D1 [37]. To determine if the p160ROCK Rho effector pathway is involved in the stimulation of COX-2 expression, we treated RIE-1, RIE-Ras(12V), and RIE-RhoA(63L) cells with an inhibitor of p160ROCK, a downstream effector Rho-kinase. Although 10 mM Y-27632 reduced basal COX-2 expression in RIE cells by approximately 70%, significant inhibition in RIE-RhoA(63L) and RIE-Ras(12V) cells was not observed, and no alterations in the expression of cyclin D1 were observed, as shown in Figure 6. These results suggest that other RhoA effector pathways, independent of p160ROCK, are involved in the regulation of COX-2 expression in Ras-transformed intestinal epithelial cells.

Effect of MEK and p160ROCK Inhibition on COX-2 Expression

Activation of Ras stimulates a variety of diverse intracellular signaling pathways, the most extensively studied of which is the canonic Raf/MEK/Erk pathway. Inhibitors of MEK were used to determine the involvement of this pathway in the stimulation of COX-2 expression in intestinal epithelial cell lines. Preliminary experiments (not shown) confirmed that 25 mM PD98059 significantly inhibited MEK activity in the cells used in this study, consistent with published reports. Similarly, PD98059 inhibited basal COX-2 expression in parental RIE-1 cells, as well as inhibited the intermediate level of COX-2 expression in cells expressing activated RhoA (Figure 5). However, inhibition of COX-2 by PD98059 did not occur in cells overexpressing activated Ras. This observation was further explored by using U0126, an alternate, more potent, and selective inhibitor of MEK activity [36]. Using U0126 at a concentration of 10 mM, inhibition of MEK was virtually complete as shown in the inset in Figure 5, but inhibition of COX-2 expression was modest and did not decrease to the lower level of expression observed in parental RIE or RIE-RhoA(63L) cells. Thus, despite the expression of oncogenic Ras at a level sufficient to cause transformation, activation of Erk was not fully and independently responsible for the maintenance of high levels of COX-2 expression characteristic of this cell line.

As shown in Figure 5, transfection of RIE-Ras(12V) cells with a dominant-negative RhoA(N19) construct markedly decreased COX-2 levels, as we have reported previously [37]. The selectively of dominant-negative Rho(A) on expression was apparent by the lack of reduced cyclin D1 in dually transfected cells. When the dual transfectant was treated with U0126, decreased MEK activity further reduced COX-2 expression, suggesting that Rho and MEK jointly regulate COX-2 expression.

We previously found that activated RhoA and activated Raf function cooperatively to transform intestinal epithelial

Discussion

COX-2 is believed to play an important role in the progression of colorectal cancer. Its expression is markedly induced in human colorectal tumors and is not detectable in adjacent normal mucosa [2,38]. Although increased COX-2 expression is not sufficient for neoplastic transformation of cultured intestinal epithelial cells [39], reduction of COX-2 activity in cell culture models, experimental animals, and human research subjects decreases transformed cell behavior and tumor growth. For example, the growth of subcutaneously implanted lung carcinoma tumor cells in COX-2_/_ mice is attenuated [40] when compared with COX-2-expressing littermates. The growth of small intestinal tumors is significantly diminished in COX-2_/_ mice that are also deficient in the APC tumor suppressor gene product [41]. Similarly, the use of the COX-2 inhibitor, celocoxib, in persons with FAP diminishes the number of existing adenomas by about 25% [6], although standard doses of sulindac do not prevent the development of adenomas [42]. Thus, there is a connection between COX-2 activity and transformed cell behavior, which is explained, in part at least, by the effect of COX-2 -derived prostaglandins on cellular proliferation, inhibition of apoptosis, increased invasiveness, and stimulation of angio-genesis. Thus, understanding how COX-2 levels are regulated and how its activity can be modified to therapeutic advantage has become a major area of investigation in the past decade.

In quiescent G0 cells, COX-2 expression is low or not detectable; its levels are induced by growth factors, such as EGF-related peptides and TGF(3, and are super-induced by the protein synthesis inhibitor, cyclohexamide [8]. Thus, COX-2 is regulated as an immediate early gene in non-transformed intestinal epithelial cells and other nontrans-formed cells [8]. Diverse intracellular signal transduction pathways tightly regulate COX-2 expression in normal epithelial cells by the regulation of gene transcription, mRNA stability, and protein translation [43]. In the setting of cellular

transformation, COX-2 is constitutively overexpressed due to the activation of diverse intracellular signaling pathways. This is particularly true in the case of cells overexpres-sing mutant-activated Ras. For example, Ras upregulates COX-2 expression by activation of the transcription and stabilization of COX-2 mRNA by ERK, JNK, p38, Rho, and Akt signaling (reviewed in Ref. [43]). Activation of COX-2 expression by other small GTPases, such as RhoGTPases, also occurs [17] in some cells by a p160ROCK- and NF-nB-dependent mechanism, although there is a great deal of cell type specificity in this response [20]. Notwithstanding this growing body of information on intracellular signaling pathways involved in the regulation of COX-2 expression, little significant attention has been given to growth factor pathways that specifically activate COX-2 expression in Ras-transformed cells.

Because overexpression of EGF-related peptides and TGFb contributes to the Ras-transformed phenotype [22, 44,45], in the present work, we examined the relative importance of these growth factors in the regulation of COX-2 expression in the nontransformed rat intestinal epithelial cell line RIE-1, and in RIE-1 cells transformed by stable trans-fection of oncogenic Ras. In addition, regulation of COX-2 in RIE cells overexpressing mutant-activated RhoA was also examined because of emerging reports suggesting that RhoA transcriptionally activates COX-2 expression and our own work showing that simultaneous expression of activated RhoA and activated Raf in RIE cells mimics many of the characteristics of Ras-transformed cells [37].

EGFR activation by autocrine EGF-related peptides is a potentially attractive mechanism for the upregulation of COX-2 expression in Ras-transformed intestinal epithelial cells because upregulation of these peptides commonly occurs in a variety of Ras-transformed epithelial cell types, including intestinal epithelial cells [22] and keratinocytes [45]. In fact, a number of characteristics of Ras-transformed cells can be specifically attributed to signaling through by the EGFR. For example, Ras-mediated induction of vascular endothelial growth factor and tumor angiogenesis [46], growth in soft agar and morphologic transformation [22,26], and sensitivity to Ras inhibition by farnesyltransferase inhibitors [44] can be linked to the production of EGF-related peptides such as transforming growth factor a, heparin-binding EGF-like growth factor, and amphiregulin in RIE-1 cells. In the present work, we investigated the potential role of EGFR activation in contributing to the high levels of COX-2 in RIE-Ras(12V) cells. Treatment of nontransformed RIE-1 and RIE-RhoA cells with EGFR tyrosine kinase inhibitors reduced COX-2 expression in a dose-dependent fashion, indicating the stimulation of EGFR tyrosine kinase, presumably by autocrine production of TGFa, amphiregulin, betacellulin, and HB-EGF, which we have previously found in RIE-1 cells [47]. However, unlike the observations cited above, we found that COX-2 induction by stable Ras overexpression in RIE-Ras(12V) cells cannot be solely attributed to EGFR activation, despite a demonstrated inhibitory effect of EGFR tyrosine kinase inhibitors on Ras-transformed cell growth and EGFR phos-phorylation. Clearly, an additional pathway for the regulation of COx-2 expression must be invoked.

TGFb increases COX-2 expression in RIE-1 cells. In our studies, TGFb did not further increase the already elevated levels of COX-2 expression in Ras-transformed cells. We also found that inhibition of TGFb signaling by anti-TGFb-neutralizing antibodies did not reduce COX-2 levels in Ras-transformed cells. Our findings contrast with the observations of Sheng et al. [29], who rapidly induced Ras expression in RIE cells using an isopropyl-1-thio-b-D-galactopyranoside (IPTG)-inducible construct. In this system, TGFb further induced COX-2 expression and anti-TGFb-neutralizing antibodies reduced, but did not eliminate, IPTG-induced COX-2 expression. The divergence between our study and the Sheng et al. study is likely due to technical differences in the systems under study. Inducible systems are excellent for the analysis of the immediate signaling effects of oncogenic Ras overexpression, are but less attractive for understanding cellular consequences of the sustained overexpression of oncogenic Ras that typifies neo-plasia. Conversely, because induction of genomic instability by oncogenic Ras is a well-known phenomenon and, indeed, all in vitro cell culture systems are susceptible to genomic instability [48], it is feasible that other genetic alterations contribute to variations in the regulation of COX-2 expression reported in various published studies, especially when rodent cell lines are used.

Emerging data suggest the regulation of COX-2 expression by the small GTPase RhoA [17,20]. Recently, we found that dominant-negative expression of RhoA in RIE-Ras(12V) cells markedly and specifically inhibits elevated COX-2 expression and increases PGE2 levels characteristic of cells expressing hyperactive Ras [37]. This suggests that RhoA activation is necessary for increased COX-2 expression in the presence of mutant-activated Ras. In support of this, herein, we report that levels of COX-2 protein expression in RIE cells that stably overexpress mutant-activated RhoA are increased relative to parental RIE-1 cells, are but less than RIE-Ras cells. In addition, stable expression of a dominantnegative RhoA construct in RIE-Ras(12V) reduces COX-2 expression to relatively low levels seen in parental cells. Further, this residual COX-2 expression is reduced by a pharmacologic inhibitor of Erk activation, suggesting that the canonic Ras/Raf/MEK,Erk pathway and Rho jointly regulate COX-2 activation. Identification of the specific RhoA effector pathway(s) involved in the regulation of COX-2 expression will be problematic as more than 30 potential effectors of Rho-GTPases have been identified and, in most instances, downstream targets of these various Rho effectors are not yet known [49]. Also not investigated in our system is the potential role of other Rho proteins such as RhoB, RhoC, and Rac1, which share overlapping effector systems with RhoA [50]. Candidate effectors include mDIA, rhotekin, rhophilin, and PRK, among others. In the current study, because of the facile availability of a highly specific pharmacologic inhibitor (Y-27632), we studied p160ROCK, which has recently been identified as a RhoA effector pathway involved in a variety of functions such as actin stress fiber assembly [51] and tight junction formation [52]. Our findings indicate that ROCK does not mediate the induction of COX-2

in Ras-transformed cells. However, further support for the involvement of the RhoA pathway in the regulation of COX-2 expression is suggested by our work, indicating that additional investigation of this complex pathway is needed to fully understand the role of RhoA in the regulation of COX-2 expression in normal and transformed cells.

In summary, stimulation of the EGFR is an important inducer of basal COX-2 expression in nontransformed culture intestinal epithelial cells and in cells overexpressing activated RIE-RhoA(63L). By contrast, in cells transformed by overexpression of mutant-activated Ras, increased EGFR signaling and overproduction of TGFp do not account for increased COX-2 expression. Emerging data suggest that RhoA activation in Ras-transformed cells is a critical key regulator of COX-2 expression, although the precise mechanism by which this occurs requires further investigation. Identification of these complex pathways will be important in the design of therapeutic strategies to manipulate COX-2 expression in the gastrointestinal tract.

References

[1] FitzGerald G and Patrono C (2001). The coxibs, selective inhibitors of cyclooxygenase-2. N Engl J Med 345, 433-442.

[2] Eberhart CE, Coffey RJ, Radhik A, Giardiello FM, Ferrenbach S, and DuBois RN (1994). Up-regulation of cyclooxygenase 2 gene expression in human colorectal adenomas and adenocarcinomas. Gastroenterology 107, 1183-1188.

[3] Williams CS, Luongo C, Radhika A, Zhang T, Lamps LW, Nanney LB, Beauchamp RD, and DuBois RN (1996). Elevated cyclooxygenase-2 levels in Min mouse adenomas. Gastroenterology 111, 1134-1140.

[4] Shao J, Sheng H, Aramandla R, Pereira MA, Lubet RA, Hawk E, Grogan L, Kirsch IR, Washington MK, Beauchamp RD, et al. (1999). Coordinate regulation of cyclooxygenase-2 and TGF-beta1 in replication error-positive colon cancer and azoxymethane-induced rat colonic tumors. Carcinogenesis 20, 185-191.

[5] DuBois RN (2003). Cyclooxygenase-2 and colorectal cancer. Prog Exp Tumor Res 37, 124 -137.

[6] Steinbach G, Lynch PM, Phillips RK, Wallace MH, Hawk E, Gordon GB, Wakabayashi N, Saunders B, Shen Y, Fujimura T, et al. (2000). The effect of celecoxib, a cyclooxygenase-2 inhibitor, in familial adenoma-tous polyposis. N Engl J Med 342, 1946-1952.

[7] Jacoby RF, Cole CE, Tutsch K, Newton MA, Kelloff G, Hawk ET, and Lubet RA (2000). Chemopreventive efficacy of combined piroxicam and difluoromethylornithine treatment of APC mutant Min mouse adenomas, and selective toxicity against APC mutant embryos. Cancer Res 60, 1864-1870.

[8] DuBois RN, Awad J, Morrow JD, Roberts LJ, and Bishop PR (1994). Regulation of eicosanoid production and mitogenesis in rat intestinal epithelial cells by transforming growth factor alpha and phorbol ester. J Clin Invest 93, 493-498.

[9] Han DS, Li F, Holt L, Connolly K, Hubert M, Miceli R, Okoye Z, Santiago G, Windle K, Wong E, et al. (2000). Keratinocyte growth factor-2 (FGF-10) promotes healing of experimental small intestinal ulceration in rats. Am J Physiol Gastrointest Liver Physiol 279, G1011 -G1022.

[10] Tessner TG, Muhale F, Schloemann S, Cohn SM, Morrison A, and Stenson WF (2003). Basic fibroblast growth factor upregulates cyclo-oxygenase-2 in I407 cells through p38 MAP kinase. Am J Physiol Gas-trointest Liver Physiol 284, G269-G279.

[11] Sheng H, Shao J, Hooten EB, Tsujii M, DuBois RN, and Beauchamp RD (1997). Cyclooxygenase-2 induction and transforming growth factor beta growth inhibition in rat intestinal epithelial cells. Cell Growth Differ 8, 463-470.

[12] Jones MK, Sasaki E, Halter F, Pai R, Nakamura T, Arakawa T, Kuroki T, and Tarnawski AS (1999). HGF triggers activation of the COX-2 gene in rat gastric epithelial cells: activation mediated through the ERK2 signaling pathway. FASEB J13, 2186-2194.

[13] Xie W and Herschman HR (1996). Transcriptional regulation of prosta-

glandin synthase 2 gene expression by platelet-derived growth factor and serum. J Biol Chem 271, 31742-31748.

[14] Matsuura H, Sakaue M, Subbaramaiah K, Kamitani H, Eling TE, Dannenberg AJ, Tanabe T, Inoue H, Arata J, and Jetten AM (1999). Regulation of cyclooxygenase-2 by interferon gamma and transforming growth factor alpha in normal human epidermal keratinocytes and squamous carcinoma cells. J Biol Chem 274, 29138-29148.

[15] Guo YS, Cheng JZ, Jin GF, Gutkind JS, Hellmich MR, and Townsend CM Jr (2002). Gastrin stimulates cyclooxygenase-2 expression in intestinal epithelial cells through multiple signaling pathways. Evidence for involvement of ERK5 kinase and transactivation of the epidermal growth factor receptor. J Biol Chem 277, 48755-48763.

[16] Sheng H, Williams CS, Shao J, Liang P, DuBois RN, and Beauchamp RD (1998). Induction of cyclooxygenase-2 by activated Ha-Ras oncogene in Rat-1 fibroblasts and the role of mitogen-activated protein kinase pathway. J Biol Chem 273, 22120-22127.

[17] Slice LL, Bui L, Mak C, and Walsh JH (2000). Differential regulation of COX-2 transcription by Ras- and Rho-family of GTPases. Biochem Biophys Res Commun 276, 406-410.

[18] Fujita M, Fukui H, Kusaka T, Morita K, Fujii S, Ueda Y, Chiba T, Sakamoto C, Kawamata H, and Fujimori T (2000). Relationship between cyclooxygenase-2 expression and K-ras gene mutation in colorectal adenomas. J Gastroenterol Hepatol 15, 1277-1281.

[19] Sheng H, Shao J, and DuBois RN (2001). K-ras-mediated increase in cyclooxygenase 2 mRNA stability involves activation of the protein ki-nase B1. Cancer Res 61, 2670-2675.

[20] Benitah SA, Valeron PF, and Lacal JC (2003). ROCK and nuclear factor kappaB-dependent activation of cyclooxygenase-2 by RHO GTPases: effects on tumor growth and therapeutic consequences. Mol Biol Cell 14, 3041 -3054.

[21] Khosravi-Far R, Campbell S, Rossman KL, and Der CJ (1998). Increasing complexity of Ras signal transduction, involvement of Rho family proteins. Adv Cancer Res 72, 57-107.

[22] Gangarosa LM, Sizemore N, Graves-Deal R, Oldham SM, Der CJ, and Coffey RJ (1997). A Raf-independent epidermal growth factor receptor autocrine loop is necessary for Ras transformation of rat intestinal epithelial cells. J Biol Chem 272, 18926-18931.

[23] Geiser AG, Kim SJ, Roberts AB, and Sporn MB (1991). Characterization of the mouse transforming growth factor-beta 1 promoter and activation by the Ha-Ras oncogene. Mol Cell Biol 11, 84-92.

[24] Torrance C, Jackson P, Montgomery E, Kinzler KW, Vogelstein B, Wissner A, Nunes M, Frost P, and Discafani C (2000). Combinatorial chemoprevention of intestinal neoplasia. Nat Med 6, 1024-1028.

[25] Blay J and Brown KD (1985). Epidermal growth factor promotes the chemotactic migration of cultured rat intestinal epithelial cells. J Cell Physiol 124, 107-112.

[26] Oldham SM, Clark GJ, Gangarosa LM, Coffey JRJ, and Der CJ (1996). Activation of the Raf-1/MAP kinase cascade is not sufficient for Ras transformation of RIE-1 epithelial cells. Proc Natl Acad Sci USA 93, 6924-6928.

[27] Jiang B, Zhang JS, Du J, Urrutia R, and Barnard J (2003). Growth inhibitory signalling by TGFbeta is blocked in Ras-transformed intestinal epithelial cells at a post-receptor locus. Cell Signal 15, 699-708.

[28] Shipley GD, Childs CB, Volkenant ME, and Moses HL (1984). Differential effects of epidermal growth factor, transforming growth factor, and insulin on DNA and protein synthesis and morphology in serumfree cultures of AKR-2B cells. Cancer Res 44, 710-716.

[29] Sheng HJS, Dixon D, Williams CS, Prescott SM, DuBois RN, and Beauchamp RD (2000). Transforming growth factor-beta1 enhances Ha-Ras-induced expression of cyclooxygenase-2 in intestinal epithelial cells via stabilization of mRNA. J Biol Chem 275, 6628-6635.

[30] Subbaramaiah K, Telang N, Ramonetti JT, Araki R, De Vito B, Weksler BB, and Dannenberg AJ (1996). Transcription of cyclooxygenase-2 is enhanced in transformed mammary epithelial cells. Cancer Res 56, 4424-4429.

[31] Polk DB, McCollum GW, and Carpenter G (1995). Cell density-dependent regulation of PLC gamma 1 tyrosine phosphorylation and catalytic activity in an intestinal cell line (IEC-6). J Cell Physiol 162, 427-433.

[32] Pittelkow MR, Cook PW, Shipley GD, Derynck R, and Coffey RJ Jr (1993). Autonomous growth of human keratinocytes requires epidermal growth factor receptor occupancy. Cell Growth Differ 4, 513-521.

[33] Polk DB (1998). Epidermal growth factor receptor-stimulated intestinal epithelial cell migration requires phospholipase C activity. Gastroenter-ology 114, 493-502.

[34] Eguchi S, Numaguchi K, Iwasaki H, Matsumoto T, Yamakawa T, Utsunomiya H, Motley ED, Kawakatsu H, Owada KM, Hirata Y, et al.

(1998). Calcium-dependent epidermal growth factor receptor transacti-vation mediates the angiotensin II-induced mitogen-activated protein kinase activation in vascular smooth muscle cells. J Biol Chem 273, 8890-8896.

[35] Bos M, Mendelsohn J, Kim YM, Albanell J, Fry DW, and Baselga J (1997). PD153035, a tyrosine kinase inhibitor, prevents epidermal growth factor receptor activation and inhibits growth of cancer cells in a receptor number-dependent manner. Clin Cancer Res 3, 2099-2106.

[36] Favata MF, Horiuchi KY, Manos EJ, Daulerio AJ, Stradley DA, Feeser WS, Van Dyk DE, Pitts WJ, Earl RA, Hobbs F, et al. (1998). Identification of a novel inhibitor of mitogen-activated protein kinase kinase. J Biol Chem 273, 18623-18632.

[37] Du J, Jiang B, Coffey RJ, and Barnard J (2004). Raf and RhoA cooperate to transform intestinal epithelial cells and induce growth resistance to transforming growth factor beta. Mol Cancer Res 2, 233-241.

[38] Kargman SL, O'Neill GP, Vickers PJ, Evans JF, Mancini JA, and Jothy S (1995). Expression of prostaglandin G/H synthase-1 and -2 protein in human colon cancer. Cancer Res 55, 2556-2559.

[39] Tsujii M and DuBois RN (1995). Alterations in cellular adhesion and apoptosis in epithelial cells overexpressing prostaglandin endoperoxide synthase 2. Cell 83, 493-501.

[40] Williams CS, Tsujii M, Reese J, Dey SK, and DuBois RN (2000). Host cyclooxygenase-2 modulates carcinoma cell growth. J Clin Invest 105, 1589-1594.

[41] Oshima M, DinchukJE, Kargman SL, Oshima H, Hancock B, Kwong E, Trazaskos JM, Evans JF, and Taketo MM (1996). Suppression of intestinal polyposis in APC delta716 knockout mice by inhibition of cyclo-oxygenase-2. Cell 87, 803-809.

[42] Giardello F, Yang VW, Hylind LM, Krush AJ, Petersen GM, Trimbath JD, Piantadosi S, Garrett G, Geiman DE, Hubbard W, et al. (2002). Primary chemoprevention of familial adenomatous polyposis with sulin-dac. N Engl J Med 346, 1085-1087.

[43] Dixon DA (2003). Regulation of COX-2 expression in human cancers. Prog Exp Tumor Res 37, 52-71.

[44] Sizemore N, Cox AD, Barnard JA, Oldham SM, Reynolds ER, Der CJ, and Coffey RJ (1999). Pharmacological inhibition of Ras-transformed epithelial cell growth is linked to down-regulation of epidermal growth factor-related peptides. Gastroenterology 117, 567-576.

[45] Dlugosz AA, Cheng C, Williams EK, Darwiche N, Dempsey PJ, Mann B, Dunn AR, Coffey RJ Jr, and Yuspa SH (1995). Autocrine transforming growth factor alpha is dispensible for v-rasHa-induced epidermal neoplasia: potential involvement of alternate epidermal growth factor receptor ligands. Cancer Res 55, 1883-1893.

[46] Casanova ML, Larcher F, Casanova B, Murillas R, Fernandez-Acenero MJ, Villanueva C, Martinez-Palacio J, Ullrich A, Conti CJ, and Jorcano JL (2002). A critical role for ras-mediated, epidermal growth factor receptor-dependent angiogenesis in mouse skin carcinogenesis. Cancer Res 62, 3402-3407.

[47] Barnard JA, Graves-Deal R, Pittelkow MR, Dubois RN, Cook P, Ramsey GW, Bishop PR, Damstrup L, and Coffey RJ (1994). Auto- and cross-induction within the mammalian epidermal growth factor-related peptides and their relevance to gastrointestinal pathophysiology. Gastroenterology 108, 564-580.

[48] Woo RA and Poon RYC (2004). Activated oncogenes promote and cooperate with chromosomal instability for neoplastic transformation. Genes Dev 18, 1317-1330.

[49] Bishop AL and Hall A (2000). Rho GTPases and their effector proteins. Biochem J 348 (Part 2), 241 -255.

[50] Wheeler AP and Ridley AJ (2004). Why three Rho proteins? RhoA, RhoB, RhoC, and cell motility. Exp Cell Res 301, 43-49.

[51] Sinnett-Smith J, Lunn JA, Leopoldt D, and Rozengurt E (2001). Y-27632, an inhibitor of Rho-associated kinases, prevents tyrosine phosphoryla-tion of focal adhesion kinase and paxillin induced by bombesin: dissociation from tyrosine phosphorylation of p130(CAS). Exp Cell Res 266, 292-302.

[52] Walsh SV, Hopkins AM, Chen J, Narumiya S, Parkos CA, and Nusrat A (2001). Rho kinase regulates tight junction function and is necessary for tight junction assembly in polarized intestinal epithelia. Gastroenterology 121, 566-579.