Role of Cyclooxygenase 2 in Protein Kinase C βII-mediated Colon Carcinogenesis*

Elevated expression of protein kinase C βII (PKCβII) is an early promotive event in colon carcinogenesis (Gokmen-Polar, Y., Murray, N. R., Velasco, M. A., Gatalica, Z., and Fields, A. P. (2001) Cancer Res. 61, 1375–1381). Expression of PKCβII in the colon of transgenic mice leads to hyperproliferation and increased susceptibility to colon carcinogenesis due, at least in part, to repression of transforming growth factor beta type II receptor (TGF-βRII) expression (Murray, N. R., Davidson, L. A., Chapkin, R. S., Gustafson, W. C., Schattenberg, D. G., and Fields, A. P. (1999)J. Cell Biol., 145, 699–711). Here we report that PKCβII induces the expression of cyclooxygenase type 2 (Cox-2) in rat intestinal epithelial (RIE) cells in vitro and in transgenic PKCβII mice in vivo. Cox-2 mRNA increases more than 10-fold with corresponding increases in Cox-2 protein and PGE2 production in RIE/PKCβII cells. PKCβII activates the Cox-2 promoter by 2- to 3-fold and stabilizes Cox-2 mRNA by at least 4-fold. The selective Cox-2 inhibitor Celecoxib restores expression of TGF-βRII both in vitro and in vivo and restores TGFβ-mediated transcription in RIE/PKCβII cells. Likewise, the ω-3 fatty acid eicosapentaenoic acid (EPA), which inhibits PKCβII activity and colon carcinogenesis, causes inhibition of Cox-2 protein expression, re-expression of TGF-βRII, and restoration of TGF-β1-mediated transcription in RIE/PKCβII cells. Our data demonstrate that PKCβII promotes colon cancer, at least in part, through induction of Cox-2, suppression of TGF-β signaling, and establishment of a TGF-β-resistant, hyperproliferative state in the colonic epithelium. Our data define a procarcinogenic PKCβII → Cox-2 → TGF-β signaling axis within the colonic epithelium, and provide a molecular mechanism by which dietary ω-3 fatty acids and nonsteroidal antiinflammatory agents such as Celecoxib suppress colon carcinogenesis.

(PKC) 1 is a family of ubiquitously expressed serine/threonine protein kinases whose members play central roles in cell proliferation, differentiation, and apoptosis (reviewed in Ref. 3). The discovery that PKC is a major cellular target for the tumor-promoting phorbol esters suggested a role for aberrant PKC signaling in tumor initiation and progression (4). However, the relative contribution of individual PKC isozymes to carcinogenesis is not well understood. Ultimately, the role of individual PKC isozymes in carcinogenesis will be understood through identification of downstream targets that participate in specific aspects of the transformed phenotype.
We have focused our recent efforts on deciphering the role of specific PKC isozymes in the development of colon cancer (5)(6)(7). We have shown that colon carcinogenesis is accompanied by changes in PKC isozyme expression, including a dramatic increase in the level of PKC␤II expression (5). PKC␤II protein levels are elevated in preneoplastic lesions in the colon, aberrant crypt foci, and are further elevated in colon tumors (5). To determine whether elevated PKC␤II levels contribute to colon carcinogenesis, we developed transgenic PKC␤II mice that express elevated PKC␤II in the colonic epithelium to levels comparable with those observed in carcinogen-induced colon tumors (6,7). These animals exhibit hyperproliferation of the colonic epithelium and enhanced susceptibility to carcinogeninduced carcinogenesis (6). We recently characterized the transforming growth factor ␤ receptor type II (TGF-␤RII) as a target for PKC␤II-mediated transcriptional repression in intestinal epithelial cells and in the colonic epithelium of transgenic PKC␤II mice (7). PKC␤II-induced inhibition of TGF-␤RII renders intestinal epithelial cells insensitive to growth inhibition by TGF-␤ and accounts, at least in part, for the colonic hyperproliferation and increased sensitivity to colon carcinogenesis characteristic of transgenic PKC␤II mice (6,7). Our studies to date indicate that PKC␤II plays a critical role in the early stages of colon carcinogenesis by inducing the loss of TGF-␤ responsiveness, thereby imposing a hyperproliferative phenotype, two prominent characteristics of colon cancer. We reasoned that the cellular phenotype induced by PKC␤II is mediated through changes in gene expression. Thus, we have initiated a genomic analysis to identify PKC␤II target genes in rat intestinal epithelial (RIE-1) cells. Among the gene targets induced by PKC␤II is the inducible form of cyclooxygenase, Cox-2. Cox-2 was originally cloned as a phorbol ester-inducible gene (8,9), and it has been implicated in the etiology of colon cancer in rodents and humans (10 -13). Our present data demonstrate that Cox-2 is a specific genomic target of PKC␤II and that PKC␤II-mediated repression of TGF-␤RII depends on Cox-2. Finally, we show that the chemopreventive -3 polyunsaturated fatty acid eicosapentaenoic acid (EPA), a known PKC␤II inhibitor in vivo and in vitro (7), inhibits Cox-2 expression, induces TGF-␤RII expression, and restores TGF-␤ responsiveness in RIE/PKC␤II cells.

EXPERIMENTAL PROCEDURES
Cell Culture and Cell Treatments-RIE-1 cells and derivatives were grown in 5% fetal bovine serum in Dulbecco's modified Eagle's medium as previously described (14). HEK293 cells were maintained in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum. Mid-log-phase cultures were used for all experiments unless otherwise specified. Construction of RIE/PKC␤II cells has been described elsewhere (7). RIE/PKC cells were produced by infection of RIE-1 cells with a retrovirus containing the full-length human PKC cDNA. RIE/H-Ras and RIE/Cox-2 cells were generous gifts of Drs. Hongmiao Sheng and Ray DuBois, Vanderbilt University (11). In some experiments, cells were incubated with the -3 fatty acid EPA (Cayman Biochemicals) at the concentrations and for the times indicated in the figure legends. In some cases, cells were incubated with 25 M Celecoxib (UTMB Pharmacy) and/or 120 pM TGF-␤1 (BD Biosciences) in the culture medium for the times indicated in the figure legends. EPA and Celecoxib were solubilized in dimethyl sulfoxide (Me 2 SO). A final Me 2 SO concentration of 0.1% was used for all treatments, and 0.1% Me 2 SO was used as a diluent control. The stability of the Cox-2 mRNA was determined in RIE-1 and RIE/PKC␤II cells by treatment of cells with 25 M 5,6-dichlorobenzimidazole riboside to inhibit RNA polymerase II. Total cellular RNA was isolated as described previously (14) at various times after dichlorobenzimidazole riboside exposure and subjected to real-time RT-PCR analysis for Cox-2 mRNA expression as described below.
Immunoblot Analysis-For immunoblot analysis, cells were washed twice with ice-cold phosphate-buffered saline and lysed in protein lysis buffer consisting of 50 mM Tris-HCl pH 7.4, 150 mM NaCl, 0.5% Nonidet P-40, 0.1% SDS, and protease inhibitor mixture (Sigma) for 30 min on ice. After removing particulate matter by centrifugation at 10,000 ϫ g for 20 min, aliqouts of total cellular protein (50 g) were electrophoresed in 10% acrylamide Tris-glycine gels (Invitrogen) and electrophoretically transferred to polyvinylidene difluoride membrane (Bio-Rad). The membranes were incubated with 5% nonfat dried milk in 10 mM Tris-HCl pH 7.4, 150 mM NaCl, and 0.05% Tween 20 (TBST) overnight at 4°C to block excess protein sites. Membranes were incubated with rabbit polyclonal antibodies against PKC␤I (Santa Cruz; 1:2,000 dilution), PKC␤II (Santa Cruz; 1:6,000 dilution), Cox-2 (Cayman; 1:1,000 dilution), TGF-␤RII (Santa Cruz; 1:1,000 dilution), or actin (Santa Cruz; 1:10,000 dilution) in TBST at room temperature for 3 h, after which the membranes were washed in TBST three times for 15 min each. The membranes were incubated with horseradish peroxidaseconjugated goat anti-rabbit antibody (Kirkegaard and Perry Laboratories,1:125,000 dilution) in TBST for 1 h at room temperature. The membranes were washed three times in TBST for 15 min each, and antigen-antibody complexes were detected using chemiluminescence (Amersham Biosciences) according to the manufacturer's instructions.
Treatment of Transgenic Mice with Celecoxib and Isolation of Colonic Epithelium-Transgenic PKC␤II mice expressing PKC␤II in the colonic epithelium were characterized previously (6,7). Transgenic PKC␤II mice and nontransgenic littermates were administered 6 mg/kg Celecoxib by oral gavage twice daily for 3 days. As controls, some mice were administered an equivalent volume of diluent (0.5% carboxylmethylcellulose). Mice were terminated on the morning of the fourth day. The colonic epithelium was isolated by scraping and subjected to immunoblot analysis for TGF-␤RII expression as described previously (7).
Northern Blot Analysis-Total RNA from RIE-1, RIE/PKC␤II, and RIE/Cox-2 cells was isolated by the guanidinium-thiocyanate-phenolchloroform method (15). Total RNA (10 g) from each cell line was electrophoresed in 1% agarose gels containing 0.66 M formaldehyde and electrophoretically transferred to nitrocellulose membranes (Intermountain Scientific Corp). After incubation at 80°C for 2 h, the membranes were prehybridized for 2h in a solution containing 50% formamide in 5ϫ Denhardt's solution, 0.1% SDS, 5ϫ SSPE (750 mM NaCl, 50 mM NaH 2 PO 4 pH 7.4, 5 mM EDTA), and 100 g/ml single-stranded sperm DNA and then hybridized overnight at 42°C with a radiolabeled cDNA probe to the rat Cox-2 gene consisting of 0.81 kb of the rat Cox-2 cDNA excised from PCB7-cox2. The probe was labeled with [␣-32 P]dCTP (800 Ci/mM, PerkinElmer) by the random priming method (Amersham Biosciences). After hybridization, membranes were washed three times in 0.1ϫ SSPE, 0.1% SDS for 20 min at 55°C and exposed to Eastman Kodak X-Omat AR film at 80°C. The intensity of the autoradiographic signal was quantified by a Lynx densitometer (Applied Imaging). To verify equivalency of RNA loading of individual samples, the blot was stripped of radioactivity and rehybridized with an 18S rRNA probe as described previously (14).
Determination of PGE2 Production-PGE2 analysis was conducted on equal numbers of RIE-1, RIE/PKC␤II, and RIE/Cox-2 cells cultured in 100-mm tissue culture plates in Dulbecco's modified Eagle's medium containing 5% fetal bovine serum for 2 days. Aliquots of culture medium (50 l) were collected and subjected to a specific enzyme-linked immunosorbent assay for PGE2 (Amersham Biosciences) following the manufacturer's instructions. Culture medium was diluted or concentrated appropriately to achieve values within the linear range of the assay (50 -6400 pg/ml). The range of the standard PGE2 curve was from 2.5 pg to 320 pg/well. Triplicate samples were analyzed in each experiment, and the results were expressed as pg of PGE2/ml of culture medium.
Analysis of Cox-2 Promoter and TGF-␤-dependent Transcriptional Activity-Cox-2 promoter activity and TGF-␤ transcriptional responses were assessed by transient transfection of RIE-1, RIE/PKC␤II, or RIE/ H-Ras cells with either a Cox-2 promoter or TGF-␤-responsive promoter construct linked to a luciferase reporter gene, respectively. Cox-2 promoter activity was determined using 4 kb of the mouse Cox-2 promoter cloned into the PGL3 reporter plasmid (16). TGF-␤ transcriptional responses were assessed using the 3TP-Luc reporter as described previously (7). The appropriate expression vector was cotransfected with TK-pRL (Renilla luciferase transcribed from the HSV TK promoter) into 70 -80% confluent RIE-1, RIE/PKC␤II, or RIE/H-ras cell lines in 6-well plates using Tfx-50 (Promega) at a DNA:liposome ratio of 1:3 as described previously (7). Fresh medium was added 3 h after transfection, and cells were harvested after 24 h. Total cell extracts were prepared for dual-luciferase assay according to the manufacturer's instructions (Promega) using a Monolight 2010 Luminometer (Analytical Luminescence Laboratory). The activity of Renilla luciferase was used as an internal control. Results are expressed as the mean of triplicate determinations Ϯ standard deviation.
Gene Microarray Analysis-Gene-profiling analysis was performed on RIE-1 and RIE/PKC␤II cells using RG-U34A Gene Chips microarrays (Affymetrix). Total RNA (25 g) was used for first-strand cDNA synthesis using a T7-(dT)24 oligomer (5Ј-GGCCAGTGAATTGTAATA-CGACTCACTATAGGGAGGCGG-dT24 -3Ј) and SuperScript II reverse transcriptase (Invitrogen). The cDNA was converted to double-stranded DNA by transcription in vitro. cRNAs were synthesized using bacteriophage T7 RNA polymerase in the presence of biotinylated nucleotides. Biotin-labeled target RNAs were fragmented to a mean size of 200 bases according to the manufacturer's protocol. Hybridization of the rat RG-U34A microarrays was performed at 45°C for 16 h in 0.1 M 2-morpholinoethanesulfonic acid (MES) pH 6.6, 1 M NaCl, 0.02 M EDTA, and 0.01% Tween 20. Microarrays were washed using both nonstringent (1 M NaCl, 25°C) and stringent (1 M NaCl, 50°C) conditions prior to staining with phycoerythrin-labeled streptavidin (10 g/ml final concentration). Data were collected using a Gene Array Scanner (Hewlett Packard) and analyzed using the Affymetrix Gene Chip Analysis Suite 5.0 software.
Real-time Reverse Transcriptase-Polymerase Chain Reaction Analysis of Gene Expression-Real-time reverse transcriptase-polymerase chain reaction (real time RT-PCR) assays were used to determine gene expression using TaqMan technology on an Applied Biosystems 7000 sequence detection system. Applied Biosystems Assays-By-Design containing a 20ϫ assay mix of primers and TaqMan MGB probes (FAMTM dye-labeled) were used for all target genes and the endogenous control, rat ␤-actin. These assays were designed using primers that span exonexon junctions so as not to detect genomic DNA. All primer and probe sequences were searched against the Celera data base to confirm specificity. The primer and probe sequences used were as follows: rat PAI1probe spanning exon8 CCAACAGAGACAATCC, forward primer ACC-GATCCTTTCTCTTTGTGGTT, reverse primer CATCAGCTGGCCCAT-GAAG; rat Cox-2-probe spanning exon8 CCCAGCAACCCGG, forward primer GAGTCATTCACCA-GACAGATTGCT, reverse primer GTACA-GCGATTGGAACATTCCTT; human PKC␤II-probe spanning the PKC-␤I/PKC␤II alternative splice junction TCGCCCACAAGCT, forward primer AAACTTGAACGCAAAGAGATCCA, reverse primer ATCGGTCGAAGTTTTCAGCATT.
The efficiency of target amplification was validated using a reference amplification reaction. Absolute values of the slope of log input RNA amount versus ⌬CT were Ͻ0.1 in all experiments. One-step RT-PCR reactions were performed on 20 ng of input RNA for both target genes and endogenous controls using the TaqMan one-step RT-PCR master mix reagent kit (Applied Biosystems). The cycling parameters were as follows: reverse transcription, 48°C for 30 min; AmpliTaq activation, 95°C for 10 min; denaturation, 95°C for 15 s; and annealing/extension, 60°C for 1 min for 40 cycles. Duplicate CT values were analyzed in Microsoft Excel using the comparative CT(⌬⌬CT) method as described by the manufacturer (Applied Biosystems).

RESULTS
Our recent studies demonstrated that elevated expression of PKC␤II in the colonic epithelium is an early, promotive event in colon carcinogenesis (5)(6)(7). To elucidate the molecular mechanisms by which PKC␤II mediates increased colon carcinogenesis, we established a cell model system in which to explore PKC␤II-mediated signaling. RIE-1 cells are immortalized but not transformed and consequently express abundant PKC␤I but little or no detectable PKCBII protein (Fig. 1A). This pattern of expression of PKC␤I and PKC␤II is consistent with that observed in the colonic epithelium in vivo (5). To assess the cellular and genomic effects of PKC␤II expression, we created the RIE/PKC␤II cell line that expresses abundant human PKC␤II (Fig. 1A). A real-time RT-PCR assay for human PKC␤II mRNA demonstrated that RIE/PKC␤II cells expressed human PKC␤II mRNA at levels somewhat lower than those observed in human HEK293 cells, which express abundant endogenous PKC␤II protein (Fig. 1B). We previously demonstrated that the growth rate of RIE/PKC␤II cells is indistinguishable from that of RIE-1 cells (7), indicating that overexpression of PKC␤II has no significant effect on proliferation or apoptosis of RIE-1 cells in culture. Similarly, no changes in gross cellular morphology were noted in RIE/PKC␤II cells, nor were these cells able to form colonies in soft agar, indicating that expression of PKC␤II is not sufficient to cause cellular transformation (data not shown).
Affymetrix Gene Chips were used to identify genes that are either induced or inhibited in RIE/PKC␤II cells when compared with RIE-1 cells. Significance analysis of microarrays (17) was used to compare expression profiles of RNA extracted from control RIE-1 cells (n ϭ 7) and RIE/PKC␤II cells (n ϭ 3). Among the probe sets whose expression changed by Ͼ2.0-fold in RIE/PKC␤II cells was that encoding the inducible form of cyclooxygenase, Cox-2. Because Cox-2 expression can be induced by phorbol esters (8,9), and because of the strong association between Cox-2 expression and colon cancer (10 -13), we focused our analysis on this gene. We confirmed our microarray analysis using a quantitative real-time RT-PCR assay specific for Cox-2 mRNA (Fig. 2A). The mean signal intensities obtained from Cox-2 probe sets in microarrays from RIE-1 (n ϭ 7) and RIE/PKC␤II (n ϭ 3) cells (gray bars) correlated well with the level of Cox-2 mRNA detected by real time RT-PCR (black bars), providing independent confirmation of elevated expression of Cox-2 RNA in RIE/PKC␤II cells. Northern blot analysis indicated that Cox-2 mRNA expression was increased Ͼ10-fold in RIE/PKC␤II cells when compared with RIE-1 cells (Fig. 2B), providing an independent confirmation of the microarray and real-time RT-PCR data. As a positive control, RIE/Cox2 cells, which express a Cox-2 transgene that is deleted of the 3Јuntranslated region (18), was used to compare the abundance of Cox-2 mRNA in RIE/PKC␤II cells expressing the endogenous, full-length Cox-2 transcript. The data in Fig. 2B indicate that Cox-2 mRNA is even more abundant in RIE/PKC␤II cells

FIG. 2. The Cox-2 gene is induced by PKC␤II.
A, total RNA from RIE-1 and RIE/PKC␤II cells was subjected to microarray analysis using Affymetrix Gene Chips as described under "Experimental Procedures." Rat Cox-2 mRNA abundance was independently measured by real-time RT-PCR as described under "Experimental Procedures." Results are plotted as signal intensity Ϯ S.D. B, Northern blot analysis for Cox-2 mRNA was performed on total RNA isolated from RIE-1, RIE/PKC␤II, and RIE/Cox2 cells as described under "Experimental Procedures." An 18S RNA probe was used to assess RNA loading on the gel. than in RIE/Cox2 cells. Real-time RT-PCR analysis confirmed that RIE/PKC␤II cells express 1.4-fold more PKC␤II mRNA than RIE/Cox-2 cells. Taken together, these data provide conclusive evidence that Cox-2 mRNA is induced by PKC␤II in RIE-1 cells.
Immunoblot analysis was used to assess the level of Cox-2 protein expression in RIE-1, RIE/PKC␤II, and RIE/Cox2 cells (Fig. 3A). Cox-2 protein was expressed at nearly undetectable levels in RIE-1 cells. In contrast, RIE/PKC␤II cells expressed abundant Cox-2 protein comparable with the amount of Cox-2 protein expressed in RIE/Cox2 cells. To assess whether induction of Cox-2 expression is specific for PKC␤II expression, we assayed Cox-2 protein levels in RIE/PKC cells, which were engineered to overexpress transgenic human PKC (Fig. 3B). RIE/PKC cells expressed very low levels of Cox-2 (which could be observed only after very long exposures of the immunoblots) comparable with those observed in RIE-1 cells. These results are consistent with our microarray analysis of RIE/PKC cells, which did not identify Cox-2 as a potential transcriptional target of PKC (data not shown). These results indicate that induction of Cox-2 is not a general response to the expression of any PKC isozyme, but rather is a specific response to PKC␤II expression.
To assess whether the Cox-2 protein expressed in RIE/ PKC␤II cells was functional, RIE-1, RIE/PKC␤II, and RIE/ Cox-2 cells were assayed for production of PGE2, a product of Cox-2 enzyme activity (Fig. 3C). Enzyme-linked immunosorbent assay analysis of culture supernatants demonstrated that RIE/PKC␤II cells, like RIE/Cox2 cells, secreted 3-to 4-fold higher levels of PGE2 than RIE-1 cells. These results demon-strate that PKC␤II induced the expression of active Cox-2 enzyme in RIE/PKC␤II cells.
We next wished to determine whether Cox-2 is also a target for PKC␤II regulation in the colonic epithelium in vivo. We have developed transgenic PCK␤II mice overexpressing PKC␤II in the colonic epithelium that exhibit an increased sensitivity to azoxymethane-mediated colon carcinogenesis (6,7). Immunoblot analysis of colonic epithelium from nontransgenic and transgenic PKC␤II mice demonstrate that transgenic PKC␤II mice expressed significantly more PKC␤II and Cox-2 protein than their nontransgenic littermates (Fig. 3D). These data demonstrate that Cox-2 is a significant genomic target of PKC␤II both in RIE-1 cells in culture and in the colonic epithelium in vivo.
We next assessed the mechanism by which PKC␤II leads to elevated Cox-2 mRNA levels in RIE-1 cells. For this purpose, a Cox-2 promoter/luciferase reporter gene was transiently cotransfected into RIE-1 cells along with a PKC␤II expression vector to assess the effect of PKC␤II on Cox-2 promoter activity (Fig. 4A). The activity of the Cox-2/luciferase reporter was increased by 2-to 3-fold in RIE-1 cells in which a PKC␤II expression vector was simultaneously introduced. These experiments utilized a human Cox-2 promoter construct consisting of 4 kb of 5Ј-flanking sequence, but consistent results were also obtained using promoters containing 7 kb or 1.4 kb of 5Јflanking sequence (data not shown). To independently assess the effect of PKC␤II expression on Cox-2 promoter activity, the Cox-2 promoter/luciferase reporter was transfected into RIE-1, RIE/PKC␤II, and RIE/Ras cells, and the activity of the reporter was assessed by luciferase assay (Fig. 4B). RIE/Ras cells express an activated H-Ras allele previously shown to activate Cox-2 transcription (11,19) and served as a positive control for activation of Cox-2 promoter activity. Consistent with the transient cotransfection data shown in Fig. 4A, the Cox-2 promoter was 2-3 times more active in RIE/PKC␤II cells than in RIE-1 cells. As expected, the Cox-2 promoter was also more active, by about 5-fold, in RIE/Ras cells, which are known to express high levels of Cox-2 (11). These data demonstrate that PKC␤II causes a 2-to 3-fold increase in Cox-2 promoter activity in RIE cells. Although this represents a significant and reproducible increase in Cox-2 promoter activity, the magnitude of the effect was clearly not sufficient to account for the dramatic increase in Cox-2 mRNA expression observed in RIE/PKC␤II cells. Taken together, these data indicate that an additional mechanism(s) may be responsible for the effects of PKC␤II on Cox-2 mRNA levels in RIE-1 cells.
The Cox-2 gene is known to be regulated not only at the transcriptional level but also at the level of mRNA stability (20 -22). Therefore, Cox-2 mRNA stability was compared in RIE-1 and RIE/PKC␤II cells by measuring mRNA abundance by quantitative real-time RT-PCR as a function of time after addition of the nonspecific RNA polymerase II inhibitor dichlorobenzimidazole riboside (Fig. 4C). Cox-2 mRNA in RIE-1 cells was relatively unstable with an estimated t 1 ր 2 of degradation of ϳ12-16 min. In contrast, the apparent t 1 ր 2 of degradation of Cox-2 mRNA in RIE/PKC␤II cells was greater than 50 min, indicating that PKC␤II expression results in significant stabilization of Cox-2 mRNA. Thus, PKC␤II-mediated elevation of Cox-2 mRNA levels result from a combined effect of PKC␤II on Cox-2 gene transcription and Cox-2 mRNA stability.
We recently showed that PKC␤II inhibits expression of TGF-␤RII both in the colonic epithelium of transgenic PKC␤II mice and in RIE/PKC␤II cells (7). Therefore, we assessed whether the effect of PKC␤II on TGF␤RII expression is mediated through Cox-2 activation (Fig. 5). RIE-1 cells expressed abundant TGF␤RII protein as assessed by immunoblot analysis

FIG. 3. Cox-2 protein and enzyme activity is induced in RIE/ PKC␤II cells and in the colons of transgenic PKC␤II mice.
A, total cell lysates from RIE-1, RIE/PKC␤II, and RIE/Cox2 cells were subjected to immunoblot analysis for PKC␤II, Cox-2, and actin as described under "Experimental Procedures." B, total cell lysates from RIE-1 and RIE/PKC cells were subjected to immunoblot analysis for Cox-2 and actin. C, PGE2 levels from culture supernatants from RIE-1, RIE/PKC␤II, and RIE/Cox2 cells were measured by enzyme-linked immunosorbent assay as described under "Experimental Procedures." Results are expressed as pg/ml PGE2 Ϯ S.E. for three independent measurements. p-values were determined by Student's t test. D, lysates from colonic epithelium from transgenic PKC␤II mice (6, 7) and nontransgenic littermates were subjected to immunoblot analysis for PKC␤II, Cox-2, and actin as described under "Experimental Procedures." (Fig. 5A, lane 1), whereas RIE/PKC␤II cells expressed significantly lower levels of TGF␤RII protein (Fig. 5A, lane 2). Treatment of RIE/PKC␤II cells with 25 M Celecoxib for 0, 24, or 48 h led to a time-dependent increase in TGF-␤RII protein expression (Fig. 5A, lanes 3-5). We have also observed that treatment of RIE-1 cells with Celecoxib leads to increased TGF-␤RII protein expression, indicating that Cox-2 exerts a tonic suppressive effect on TGF-␤RII expression in RIE-1 cells that can be reversed by inhibition of the enzyme(data not shown). RIE/PKC␤II cells exhibit a profound loss of TGF-␤-mediated transcriptional activity as a consequence of PKC␤II-mediated repression of TGF-␤RII expression (7). We therefore assessed the ability of Celecoxib to restore TGF-␤-mediated transcriptional activity in RIE/PKC␤II cells (Fig. 5, B and C). RIE/ PKC␤II cells were transiently transfected with a TGF-␤-reponsive luciferase reporter plasmid and then treated with either TGF-␤1, Celecoxib, or both (Fig. 5B). RIE/PKC␤II cells exhibited little or no transcriptional response to TGF-␤1, consistent with our previous results (7). However, when these cells were treated with Celecoxib prior to exposure to TGF-␤1, they exhibited a robust transcriptional response. Celecoxib treatment had no effect in the absence of TGF-␤1, demonstrating that the observed transcriptional effects of Celecoxib are TGF-␤dependent. Consistent with these results, the level of the mRNA for the endogenous TGF-␤1-responsive gene, plasminogen activator inihibitor-1 (PAI-1) (23), was dramatically induced when RIE/PKC␤II cells were treated with Celecoxib for 24 or 48 h prior to exposure to TGF-␤1 (Fig. 5C). To determine whether the PKC␤II-mediated repression of TGF-␤RII expression is Cox-2-dependent in vivo, we assessed the effect of treating transgenic PKC␤II mice with Celecoxib on TGF-␤RII expression in the colonic epithelium (Fig. 5D). Treatment of transgenic PKC␤II mice with Celecoxib led to reexpression of TGF-␤RII as assessed by immunoblot analysis. Therefore, PKC␤II-mediated repression of TGF-␤RII expression and TGF-␤-responsiveness are dependent on Cox-2 activity both in intestinal epithelial cells in vitro and in the colonic epithelium in vivo Chemopreventive dietary -3 fatty acids such as EPA block Cox-2 induction in azoxymethane-treated mice (24). We and others have shown that azoxymethane induces colonic PKC␤II expression (5,25). We have found that a diet high in -3 fatty acids inhibits colonic PKC␤II activity, induces TGF-␤RII expression, and blocks PKC␤II-mediated colon carcinogenesis in transgenic PKC␤II mice (7). These observations lead to the hypothesis that the chemopreventive effects of -3 fatty acids are mediated through inhibition of a PKC␤II 3 Cox-2 3 TGF-␤ signaling pathway. To test this hypothesis, we treated RIE-1 and RIE/PKC␤II cells with EPA, an -3 fatty acid found in fish oil, and measured TGF-␤RII and Cox-2 expression by immunoblot analysis (Fig. 6A). Treatment of RIE/PKC␤II cells with EPA led to a dose-dependent increase in TGF-␤RII expression and a concomitant decrease in Cox-2 expression (Fig. 6A, left  panel). This effect was dependent on PKC␤II expression because no significant changes in either TGF-␤RII or Cox-2 expression were observed in RIE-1 cells treated with EPA (Fig.   6A, right panel). Quantitative analysis of these data revealed that Cox-2 expression was inhibited by Ͼ50% in RIE/PKC␤II cells but was unaffected in RIE-1 cells (Fig. 6B). On the other hand, TGF-␤RII was induced in RIE/PKC␤II cells but not in RIE-1 cells treated with EPA (Fig. 6C). These data are consistent with our recent observation that a diet high in -3 fatty acids induces TGF-␤RII expression in the colonic epithelium of transgenic PKC␤II mice and blocks PKC␤II-mediated colon carcinogenesis (7). They also establish a direct link between cancer-preventive dietary -3 fatty acids, PKC␤II activity, Cox-2 expression, and TGF-␤ signaling in vitro and in vivo. DISCUSSION We have focused our recent attention on the role of individual PKC isozymes in the development of colon cancer (5-7). We have found that PKC␤II, which is expressed at very low levels in the proliferative zone of normal colonic epithelium, is rapidly induced in the colonic epithelium of azoxymethane-treated rodents and is abundantly expressed in both preneoplastic aberrant crypt foci and colon tumors that form following azoxymethane treatment (5). Transgenic PKC␤II mice overexpressing PKC␤II in the colonic epithelium exhibit hyperproliferation and increased susceptibility to azoxymethane-induced colon carcinogenesis (6,7), demonstrating that PKC␤II plays a critical promotive role in colon carcinogenesis.
We also recently demonstrated that PKC␤II is an important cellular target for the cancer-preventive activity of dietary -3 fatty acids (7). Diets high in -3 fatty acids have been shown to block colon carcinogenesis in rodent models (7, 26 -29), and epidemiologic studies indicate that -3 fatty acids have chemopreventive effects against colon cancer in humans (30 -33). We found that -3 fatty acids, which are abundant in dietary fish oils, inhibit colonic PKC␤II activity and suppress the hyperproliferative and cancer-prone phenotype of transgenic PKC␤II mice (7). In those studies, we also established that the TGF-␤RII gene is a target for repression by PKC␤II (7).
The present studies were initiated in an effort to further elucidate the molecular mechanism(s) by which PKC␤II promotes colon carcinogenesis. The data presented herein identify the Cox-2 gene as a prominent target for PKC␤II-mediated regulation in intestinal epithelial cells in vitro and the colonic epithelium in vivo. PKC␤II causes induction of Cox-2 expression by at least two distinct mechanisms. First, PKC␤II leads to a modest, 2-to 3-fold induction of Cox-2 gene transcription. However, the more dramatic effect of PKC␤II on Cox-2 gene expression appears to be through stabilization of Cox-2 mRNA. The half-life of Cox-2 mRNA in RIE/PKC␤II cells is more than 50 min. This represents a significant stabilization of the Cox-2 mRNA, which exhibits a half-life of 12-16 min in RIE-1 cells. Taken together, these two actions of PKC␤II provide a plausible mechanism by which PKC␤II leads to the dramatic increase in Cox-2 mRNA, protein, and activity levels observed in RIE/ PKC␤II cells.
There is abundant evidence that both Cox-2 and PKC␤II are involved in colon carcinogenesis, and several interesting parallels exist between these two genes. First, the Cox-2 gene was originally described as an immediate early gene that is induced by either serum or phorbol ester stimulation of quiescent cells (8,9), suggesting a connection between Cox-2 gene regulation and PKC signaling. Second, both Cox-2 and PKC␤II are elevated in the colonic epithelium of rodents exposed to the colon carcinogen azoxymethane and in aberrant cryptic foci and colon tumors that develop in azoxymethane-treated animals (5,34,35), indicating that these genes are involved in early events in the carcinogenic process.
The data presented here provide the first direct mechanistic link between elevated PKC␤II and Cox-2 expression during colon carcinogenesis. Based on our current and previously published data, we propose a model for how PKC␤II and Cox-2 promote colon cancer (Fig. 7). In this model, the tissueselective carcinogen azoxymethane induces PKC␤II expression in the colonic epithelium, an event that has been well documented by our group and others (5,25). Our current data show that PKC␤II induces Cox-2 expression in intestinal epithelial cells in vitro and in the colonic epithelium in vivo. We have also demonstrated that PKC␤II leads to repression of TGF-␤RII expression and TGF-␤1 signaling in RIE cells and to a loss of TGF-␤RII expression in the colonic epithelium of transgenic PKC␤II mice (7). In the present study, we demonstrate that PKC␤II-mediated repression of TGF-␤RII expression and signaling can be reversed by the Cox-2 inhibitor Celecoxib, indicating that PKC␤II mediates its effects on TGF-␤ signaling through Cox-2. These data are consistent with the recent association between elevated Cox-2 expression and loss of TGF-␤1 responsiveness and TGF-␤RII expression in RIE cell variants selected for loss of TGF-␤1 responses (36).
The effects of Cox-2 in intestinal epithelial cells and in colon tumors are thought to be mediated through production of prostaglandins, particularly PGE2 (37). PGE2 in turn signals through binding to specific PGE2 receptors, of which there are four well characterized members, termed EP 1-4 (reviewed in Ref. 38). The EPs are members of the G-protein-coupled receptor family whose downstream effectors include adenylate cyclase and phosphatidylinositol-phospholipase C␤ (38). Accumulating evidence demonstrates that EP1 plays a pivotal role in colon carcinogenesis (39 -41). Specifically, two different pharmacologic inhibitors selective for EP1 have been shown to inhibit formation of preneoplastic aberrant cryptic foci in azoxymethane-treated mice and of intestinal polyps in APC min mice (39 -41). Furthermore, mice that are nullizygous for EP1 exhibit suppressed colon carcinogenesis (39).
It is well documented that EP1 signals through activation of PI-PLC and generation of the second messengers diacylglycerol and inositol trisphosphate, which in turn lead to intracellular calcium mobilization and PKC activation (38). Based on these observations, and our present data, it is attractive to suggest that PGE2-mediated activation of EP1 leads to activation of PKC␤II, a classical PKC isozyme, generating an autocrine positive-feedback loop. In this model, activated PKC␤II in turn causes repression of TGF-␤RII function by an as-yet-unidentified mechanism. It should be noted, however, that the role of EPs in colon cancer development appears to be complex. In this regard, both EP2 (42) and EP4 (43) have also been implicated in colon carcinogenesis, indicating that PGE2 can activate multiple signaling pathways in the colonic epithelium. The complex role of EPs, as well as other possible Cox-2 products, in PKC␤II-mediated effects in intestinal epithelial cells will require further experimentation.
In recent studies, we demonstrated that the chemopreventive effects of dietary -3 fatty acids are mediated, at least in part, through inhibition of PKC␤II activity in the colonic epithelium (7). Furthermore, we demonstrated that -3 fatty acids can block the hyperproliferation and enhanced colon carcinogenesis exhibited by transgenic PKC␤II mice through inhibition of colonic PKC␤II activity (7). In this study we show that -3 fatty acids inhibit Cox-2 expression and induce TGF-␤RII by a mechanism that depends on PKC␤II expression. These data are consistent with the model proposed in Fig. 7, because there is abundant evidence that dietary -3 fatty acids prevent azoxymethane-induced elevation of Cox-2 expression in the colonic epithelium (24). -3 fatty acids such as EPA have been shown in some cell systems to inhibit Cox-2 activity (44,45). However, our data indicate that EPA does not induce TGF-␤RII expression in RIE/PKC␤II cells through inhibiton of Cox-2 for the following reasons. First, both RIE-1 and RIE/PKC␤II cells express active Cox-2 enzyme (Fig. 3C). Despite that fact, EPA induces TGF-␤RII expression in RIE/PKC␤II cells, but not in RIE-1 cells (Fig. 6). Treatment of RIE-1 cells with Celecoxib however, leads to elevated expression of TGF-␤RII in both RIE-1 and RIE/PKC␤II cells. If EPA were acting through inhibition of Cox-2 activity, it should induce TGF-␤RII expresion in both cell lines. Therefore, it is unlikely that the effects of EPA on TGF-␤RII expression in RIE/PKC␤II cells are caused by Cox-2 inhibition.
The accumulating evidence that PKC␤II and Cox-2 expression and activity can be regulated in a similar fashion by dietary components that modulate colon cancer risk further suggests a mechanistic link between these two cancer-promoting genes. Our model is attractive in that it reconciles many seemingly disparate observations in the literature regarding the role of PKC␤II, Cox-2, and TGF-␤ signaling in colon carcinogenesis. Furthermore, it provides a paradigm within which to understand the mechanism(s) by which dietary compounds can modulate colon cancer risk. We are currently using our transgenic cell and animal models to explore the mechanism(s) by which azoxymethane and dietary factors such as EPA regulate PKC␤II expression and activity and to assess whether PKC␤II expression is required for Cox-2 gene induction and colon carcinogenesis in vivo.
In summary, we have shown that PKC␤II induces Cox-2 expression both in vitro and in vivo. Cox-2 is intimately linked to the development of colon cancer, and our studies provide a molecular mechanism by which induction of PKC␤II expression during azoxymethane-induced carcinogenesis, or overexpression of PKC␤II in transgenic mice, predisposes mice to colon cancer. The elucidation of a PKC␤II 3 Cox-2 3 TGF-␤ signaling axis, which is operative in both intestinal epithelial cells in culture, and the colonic epithelium in vivo, provides an important mechanistic link that can explain how changes in PKC␤II expression promotes colon carcinogenesis and how dietary lipids and nonsteroidal antiinflammatory drugs can modulate colon cancer risk.