Prostaglandin E2 Stimulates the beta-catenin/T cell factor-dependent transcription in colon cancer.

Cyclooxygenase and its derived prostaglandin E2 (PGE2) have been shown to stimulate the growth of cancer cells and promote tumor angiogenesis. Here, we show that PGE2 activated the beta-catenin/T cell factor-dependent transcription in colon cancer cells through the cAMP/protein kinase A pathway. The expression of cyclin D1 and vascular endothelial growth factor was induced by PGE2 in LS-174T cells. Moreover, PGE2 and mutated beta-catenin stimulated the transcription of cyclin D1 and vascular endothelial growth factor in a synergistic fashion. Mechanistically, PGE2 increased the phosphorylation of glycogen synthase kinase-3 and consequently accumulated beta-catenin. In addition, PGE2 induced the expression of T cell factor-4 transcription factor, which formed transcriptionally active complex with beta-catenin. In animal experiments, administration of 16,16-dimethyl PGE2 strongly increased the expression of cyclin D1 and vascular endothelial growth factor in APC(min/+) mouse polyps. Thus, our results provide a novel mechanism, suggesting that cyclooxygenase-2/PGE2 may exert pro-oncogenic actions through stimulating the beta-catenin/T cell factor-mediated transcription, which plays critical roles in colorectal carcinogenesis.

A large body of studies demonstrates a strong link between COX 1 -2/PGE 2 signaling and the APC/␤-catenin/TCF pathway in intestinal neoplasia. The most convincing evidence was provided by clinical trials conducted in familial adenomatous polyposis patients that result from germ line mutations of the APC gene. Administration of sulindac or celecoxib, which inhibit COX enzyme activity, significantly reduces the number and size of polyps in these patients (1,2). A murine model for familial adenomatous polyposis patients containing a germ line mutation in APC alleles is a primary animal model for investigation of pro-oncogenic actions of COX-2/PGE 2 (3)(4)(5). Disruption of COX-2 gene, which encodes the key enzyme for conversion of arachidonic acid to prostaglandins (PGs), results in a ϳ60% reduction of the number of polyps in APC knock-out mice (3). Disruption of the E type prostaglandin receptor EP 2 , which mediates PGE 2 signaling, significantly reduces the number and size of adenomas in APC ⌬716 mice as well (6). These data are supported by pharmacologic inhibition of the COX enzyme that achieves similar anti-neoplastic effects in APC-mutated mice (7). Greater than 95% of APC mutations in colorectal cancers are truncations of the C termini (8), which is required for degradation of ␤-catenin (9). Stabilized ␤-catenin binds to members of the TCF/LEF family of transcription factors and results in inappropriate oncogenic Wnt signaling (10). The regulation of a number of pro-oncogenic genes involves ␤-catenin/TCF-dependent transcription that include cyclin D1 (11), c-myc (12), vascular endothelia growth factor (VEGF) (13), and matrilysin (14).
Cumulative evidence indicates that COX-2-derived PGE 2 provides growth advantage to colorectal carcinomas through transactivation of the epidermal growth factor receptor (EGFR) signaling system (15)(16)(17). Further evidence demonstrates that COX-2/PGE 2 promotes the growth of colon cancers through enhancing tumor angiogenesis (18,19). Growth of solid tumors requires a blood supply that is achieved through neoangiogenesis, which is controlled by a number of growth factors, including VEGF. Knock-out of EP 2 receptor results in a decrease in the number and size of intestinal polyps in APC ⌬716 mice that is associated with a dramatic reduction of VEGF expression (6). Treatment with selective COX-2 inhibitors reduces levels of VEGF, particularly the membrane-bound form, in polyps of APC ⌬716 mice (4). On the other hand, PGE 2 exposure induces the expression of VEGF in colon cancer cells (20). These observations suggest a critical interaction between the COX-2/PGE 2 pro-oncogenic pathway and the oncogenic APC/␤-catenin/TCF pathway in colorectal neoplasia. In the present study, we demonstrated that PGE 2 transactivated the ␤-catenin/TCF-dependent transcription through inhibition of glycogen synthase kinase (GSK)-3 and induction of TCF-4 in colon cancer cells. In addition, PGE 2 and mutated ␤-catenin induced the transcription of TCF target genes cyclin D1 and VEGF in a synergistic manner. These results suggest that COX-2/PGE 2 may robustly enhance the oncogenic activity of the APC/␤-catenin/TCF pathway, which play key roles in colorectal carcinogenesis.

MATERIALS AND METHODS
Cell Culture and Chemicals-LS-174T cells were purchased from ATCC (Manassas, VA), and the HCA-7 cell line was a generous gift from Susan Kirkland (University of London). Human colon cancer cells were maintained in McCoy's 5A medium containing 10% fetal bovine serum. PGE 2 was purchased from Cayman Chemicals (Ann Arbor, MI). H-89, LY-294002, PD-98059, and PD-153035 were purchased from Calbiochem. For the GSK-3 inhibitors, LiCl was from Sigma, SB-216763 was from Biomol (Plymouth Meeting, PA), and BIO was from Calbiochem.
Celecoxib was a generous gift from Pharmacia (St. Louis, MO).
RNA Extraction and Northern Blot Analysis-Extraction of total cellular RNA was carried out as described previously (16). RNA samples (20 g/lane) were separated on formaldehyde-agarose gels and blotted onto nitrocellulose membranes. The blots were hybridized with cDNA probes labeled with [␣-32 P]dCTP by random primer extension (Stratagene, La Jolla, CA). After hybridization and washes, the blots were subjected to autoradiography.
Enzyme-linked Immunosorbent Assay-Levels of VEGF protein in cell culture media were quantified using an enzyme-linked immunosorbent assay kit (R & D System, Minneapolis, MN). Cells (2 ϫ 10 5 ) were seeded in 24-well plate, and serum was deprived for 48 h prior to PGE 2 treatment. Culture media were collected and stored at Ϫ80°C for assays.
Transient Transfection and Luciferase Assay-Assays to determine transcriptional activity were described previously (21). The TCF report plasmid, TOPflash, containing two sets (with the second set in the reverse orientation) of three copies of the TCF site (ATCAAAG) upstream of the thymidine kinase minimal promoter and luciferase open reading frame was purchased from Upstate (Lake Placid, NY). The FOPflash construct containing mutated TCF elements was obtained from Upstate also. The reporter construct for cyclin D1 promoter (Ϫ1745 to ϩ134 in PA3Luc vector) was kindly provided by Dr. Richard Pestell (22). The reporter construct for the VEGF promoter containing the 5Ј-flanking sequence of the human VEGF gene between Ϫ2279 and ϩ54 in pGL-2 vector was kindly provided by Dr. Hideo Kimura (23). Truncated ␤-catenin (⌬89) was provided by Dr. Paul Polakis (24). LEF-1 expression vector is a kind gift from Dr. Elaine Fuchs (25). Wild type and dominant negative TCF-4 expression vectors were purchased from Upstate. The active MAPK/ERK (MEK)-1 expression plasmid was purchased from Upstate as well.
For transient transfection, cells were co-transfected with 0.5 g of reporter plasmid and 0.3 g of the pRL-thymidine kinase plasmid, containing the Renilla luciferase gene (Promega), using the FuGENE 6 procedure (Roche Applied Science) as described in the manufacturer's protocol. Transfected cells were lysed at the indicated times for luciferase assay. Firefly and Renilla luciferase activities were measured using a dual-luciferase reporter assay system (Promega) and a luminometer. Firefly luciferase values were standardized to Renilla values.
Animal Study-APC min/ϩ mice were purchased from Jackson Laboratory (Bar Harbor, Maine) and were housed in the Animal Care Facility according to National Institutes of Health and institutional guidelines for laboratory animals. 16,16-Dimethyl PGE 2 (Cayman Chemicals, Ann Arbor, MI) was dissolved in 10% EtOH and was administered by intraperitoneal injection (100 g/kg of body weight). Intestinal polyps were collected for preparation of RNA and protein.
Data Analysis-All statistical analyses were performed on a personal computer with the StatView 5.0.1 software (SAS Institute Inc. Cary, NC). Analyses between multiple groups were determined by analysis of variance (ANOVA). Analyses between two groups were determined using the unpaired Student's t test. Differences with a p value of Ͻ0.05 were considered as statistically significant.

PGE 2 Induced TCF-dependent Transcription-
The Wnt signaling pathway stabilizes cytosolic ␤-catenin. Accumulated ␤-catenin binds and activates transcription factors of the TCF/ LEF family, which, in turn, regulates target genes by binding to TCF/LEF elements within the promoter of the gene (27). To determine whether PGE 2 was able to stimulate TCF transcription, LS-174T cells were transfected with a TCF reporter vector, TOPflash, which contains a combination of TCF binding elements. Exposure to PGE 2 increased the transcriptional activity of TOPflash ϳ10-fold (Fig. 1A). As a positive control, co-transfection with a LEF-1 expression vector increased the activity of TOPflash ϳ40-fold in LS-174T cells. In contrast, the reporter construct containing mutated TCF sites, FOPflash, had very low activity in LS-174T cells; treatment with PGE 2 did not significantly increase luciferase activity. Furthermore, PGE 2 induced TCF activity in a concentration-dependent fash- ion; as low as 1 nM, PGE 2 clearly activated TCF elements and, at 0.1 M, PGE 2 reached the maximum effect (Fig. 1B). To investigate the effect of endogenous PGE 2 on TCF activity, we treated HCA-7 cells with celecoxib at low concentrations. Previous studies have shown that HCA-7 cells express high levels of COX-2 and release PGE 2 into culture media (26). Treatment with 0.1 M celecoxib is able to completely block PGE 2 production but does not affect HCA-7 cell growth (28). TOPflash reporter vector was introduced into HCA-7 cells. 0.1 or 1 M celecoxib was added immediately after the transfection procedure. Inhibition of COX-2 activity reduced TCF activity by ϳ50% (Fig. 1C), suggesting that endogenous PGE 2 was partially responsible for the TCF-mediated transcription in HCA-7 cells. PGE 2 signals via activation of the cAMP/protein kinase A (PKA) pathway in LS-174T cells (21). As demonstrated in Fig.  1D, a specific PKA inhibitor, H-89, inhibited the PGE 2 -induced TCF activation in a concentration-dependent manner. Because PGE 2 may transactivate the epidermal growth factor receptor (EGFR) signaling system (17), we next tested whether PGE 2stimulated TCF activation involved EGFR-mediated signaling pathways. LS-174T cells were treated with PD-153035, an EGFR tyrosine kinase inhibitor, LY-294002, a phosphatidylinositol 3-kinase inhibitor, or PD-98059, a MAPK/ERK kinase (MEK) inhibitor, along with PGE 2 exposure. Inhibition of EGFR signaling did not attenuate the PGE 2 -induced TCF transactivation (Fig. 1E). Furthermore, inhibition of the phosphatidylinositol 3-kinase pathway or the MEK/ERK MAPK pathway did not reduce the PGE 2 -induced TCF activation as well.
PGE 2 Induced the Expression of Cyclin D1-Although PGE 2 greatly activated the artificial TCF reporter system; it was of importance to determine whether PGE 2 truly induced the expression of TCF target genes. Cyclin D1 has been determined as a TCF-dependent gene, which contains a number of TCF elements within its promoter region; the expression of mutated ␤-catenin increases the activity of cyclin D1 promoter and levels of cyclin D1 mRNA (11). We treated LS-174T cells with PGE 2 and found that levels of cyclin D1 protein were rapidly increased ( Fig. 2A). To determine whether transcriptional regulation was involved in PGE 2 -induced cyclin D1 expression, LS-174T cells were transfected with a cyclin D1 promoter driving reporter vector. Exposure to PGE 2 increased the luciferase activity ϳ4.5-fold (Fig. 2B), indicating that PGE 2 regulated cyclin D1 expression, at least partially, at a transcriptional level. In addition, activation of cAMP/PKA was required for PGE 2 -stimulated cyclin D1 transcription (Fig. 2C); treatment with H-89 completely attenuated the PGE 2 -induced activation of the cyclin D1 promoter. In contrast, PD-153035, LY-294002, and PD-98059 did not significantly alter the PGE 2 -induced cyclin D1 transcription. To further elucidate the potential interaction between the APC/␤-catenin/TCF pathway and PGE 2 signaling, we introduced wild type ␤-catenin or truncated ␤-catenin (⌬-89) into LS-174T cells. Although LS-174T cells express mutated ␤-catenin (29), expression of stabilized ␤-catenin (⌬-89) increased the activity of the cyclin D1 promoter ϳ2-fold (Fig. 2D). Interestingly, PGE 2 and mutated ␤-catenin (⌬-89) induced cyclin D1 transcription in a synergistic manner (Fig. 2D).
PGE 2 Induced the Expression of VEGF-Tumor angiogenesis plays critical roles in tumor growth, and VEGF is one of the major regulators for neoangiogenesis. The regulation of VEGF is thought to involve TCF activation, and there are seven TCF binding elements within the promoter of VEGF1 (13). Treatment with PGE 2 increased VEGF1 mRNA in LS-174T cells, as analyzed by Northern blot (Fig. 3A). To determine whether PGE 2 increased VEGF protein and its secretion in colon cancer cells, LS-174T cells were treated with either vehicle or PGE 2 . Levels of VEGF in cell culture medium were determined by enzyme-linked immunosorbent assay. The basal level of VEGF protein in LS-174T cell culture media was ϳ120 pg/ml. PGE 2 stimulation increased the levels of VEGF ϳ2-fold (Fig. 3B). To determine whether PGE 2 regulated VEGF1 expression at the transcriptional level, LS-174T cells were transfected with a VEGF1 promoter-driving reporter vector. As shown in Fig. 3C, PGE 2 treatment increased the transcription of VEGF1 ϳ5-fold (Fig. 3C). Although inhibition of PKA activity strongly blocked the PGE 2 -induced VEGF transcription, PD-153035 and PD-98059 also significantly reduced the PGE 2 -stimulated VEGF promoter activity. These results suggested that PGE 2 induced VEGF transcription predominantly through activation of the cAMP/PKA pathway; however, transactivation of the EGFR system also contributed to PGE 2 -induced VEGF expression. Furthermore, the ectopic expression of stabilized ␤-catenin (⌬89) increased the activity of the VEGF1 promoter ϳ3-fold, compared with the cells transfected with empty vector or wild type ␤-catenin (Fig. 3D). Interestingly, PGE 2 synergistically enhanced the action of ␤-catenin and increased the activity of the VEGF1 promoter ϳ12-fold. PGE 2 Stimulated GSK-3 Phosphorylation-PGE 2 increases the phosphorylation of GSK-3␣ in human embryonic kidney cells that ectopically express E-prostanoid receptors (30) and in human neuronal cells (31). The phosphorylation of GSK-3 inhibits its kinase activity, which is required for phosphorylation and degradation of ␤-catenin (32). To determine whether PGE 2 increased the phosphorylation of GSK-3 in colon cancer cells, LS-174T cells were treated with PGE 2 . Levels of phosphorylated GSK-3␣ were rapidly elevated (Fig. 4A). An increase in the level of ␤-catenin was detected as well. The function of inhibition of the GSK-3 was next investigated. LS-174T cells were treated with LiCl, which is an established inhibitor of both GSK-3␣ and GSK-3␤ (33). Treatment with LiCl at 10 -20 mM increased the transcriptional activity of TOPflash reporter, cyclin D1 promoter, and VEGF promoter 2-4-fold (Fig. 4B). Because LiCl is not a selective inhibitor of GSK-3 (34), we tested two more specific GSK-3 inhibitors for their ability to induce TCF-dependent transcription in LS-174T cells. SB-216763 and BIO are structurally distinct inhibitors of GSK-3␣ and GSK-3␤ (35,36). Both SB-216763 and BIO at relatively low concentrations (0.1-1 M) significantly induced the transcription of TOPflash reporter, cyclin D1 promoter, and VEGF promoter (Fig. 4C). PGE 2 Induced TCF-4 Expression-The stimulatory effect of PGE 2 on TCF-dependent transcription was significantly stronger than the effects of GSK-3 inhibitors, suggesting that PGE 2 transactivation of TCF-mediated transcription involved additional mechanisms. TCF-4 is a member of the TCF transcription factor family, which is expressed in intestinal epithelial cells (37). TCF-4 activates transcription of target genes when associated with ␤-catenin. Cyclin D1 and c-myc have been identified as targets of TCF-4 (11,12). To determine whether PGE 2 -stimulated TCF transcription involved the regulation of TCF-4 expression, levels of TCF-4 protein were examined in LS-174T cells. Exposure to PGE 2 strongly increased the levels of TCF-4 protein in LS-174T cells (Fig. 5A). Previous studies have shown that TCF-4 plays critical role in TCF-dependent transcription in LS-174T cells (29). We found that ectopic expression of wild type TCF-4 protein dramatically elevated the transcription of TOPflash; addition of PGE 2 further increased TCF transcription in a synergistic manner (Fig. 5B, left panel). Similar results were observed when LS-174T cells were cotransfected with the TCF-4 expression vector along with the cyclin D1 promoter reporter (Fig. 5B, middle panel). Expres-sion of TCF-4 increased cyclin D1 transcription ϳ5-fold; addition of PGE 2 further increased the TCF-4-induced cyclin D1 transcription in an additive manner. In contrast, expression of TCF-4 did not significantly alter VEGF transcription, and there were no collaborative actions between TCF-4 and PGE 2 on VEGF transcription as well (Fig. 5B, right panel). It has been demonstrated that expression of a dominant negative TCF-4 protein (dnTCF-4), which does not bind to ␤-catenin and acts as a potent inhibitor of the ␤-catenin-TCF complex, strongly inhibits TCF-dependent transcription in LS-174T cells (29). The expression of dnTCF-4 reduced the basal activity of TOPflash by ϳ70%. Interestingly, dnTCF-4 almost completely attenuated the PGE 2 -induced TCF activation in LS-174T cells (Fig. 5C, left panel). Moreover, dnTCF-4 decreased the basal activity of cyclin D1 promoter by ϳ50% but completely attenuated the PGE 2 -induced cyclin D1 transcription (Fig. 5C, middle panel). In contrast, expression of dnTCF-4 did not alter the basal transcriptional activity of VEGF and only slightly reduced the PGE 2 -induced VEGF transcription (Fig. 5C, right  panel).
Synergistic Action of ␤-Catenin and MEK-1 on VEGF Transcription-To further elucidate the molecular mechanism by which PGE 2 synergistically enhanced ␤-catenin-induced VEGF transcription, we investigated the interaction between ␤-catenin/TCF signaling and other signaling pathway. Because PGE 2 -induced VEGF transcription involved the transactivation of EGFR/MEK signaling (Fig. 3C), collaborative action between ␤-catenin and MEK-1 was evaluated. LS-174T cells were co-transfected with ␤-catenin (⌬-89) and active MEK-1 expression vectors along with the VEGF promoter reporter. Expression of either ␤-catenin (⌬-89) or MEK-1 increased VEGF transcription; however, in combination, they stimulated the activation of VEGF promoter in a synergistic manner (Fig. 6). PGE 2 Induced Cyclin D1 and VEGF in Vivo-Numerous studies demonstrate that inhibition of COX-2 enzyme results in decreased tumor cell proliferation in APC min/ϩ mouse polyps, suggesting that COX-2-derived PGs are critical for tumor cell proliferation in germ line APC mutation-induced neoplasia (3). To determine whether PGE 2 was able to activate ␤-catenin/ TCF signaling and stimulate the expression of TCF target genes in vivo, 16,16-dimethyl PGE 2 was administered to APCmin/ϩ mice. Protein was extracted from the polyps of an APCmin/ϩ mouse intestine. The expression of cyclin D1 protein was barely detected in vehicle-treated APC min/ϩ mouse tumors. In contrast, cyclin D1 was strongly expressed in all tumors collected from PGE 2 -treated mouse intestine (Fig. 7A). Disruption of COX-2 gene or inhibition of COX-2 enzyme decreases tumor angiogenesis and VEGF expression in APC ⌬716 mouse adenomas (6). To determine whether PGE 2 stimulated the expression of VEGF in vivo, 16,16-dimethyl PGE 2 was administered to APC min/ϩ mice. RNA and protein were extracted from polyps of APC min/ϩ mouse ileum. Levels of VEGF mRNA (Fig. 7B) and protein (Fig. 7C) were dramatically elevated in APC min/ϩ mouse tumors following administration of 16,16-dimethyl PGE 2 . DISCUSSION COX-2 is expressed at low levels in normal intestinal mucosa; its activity increases dramatically following mutations of the APC gene, suggesting that COX-2-generated prostaglandins are involved in the oncogenic activity of the APC/␤-catenin/TCF pathway. It has been shown that PGE 2 transactivates the ␤-catenin/TCF pathway in human embryonic kidney-293 cells, which ectopically express EP 2 or EP 4 receptors (30). In the present study, we found that PGE 2 activated TCF-dependent transcription in colon cancer cells at concentrations of 1-100 nM, which were similar to the levels of PGE 2 detected either in APC min/ϩ polyps or in colon cancer cells (5,26). In support with these results, inhibition of COX-2 enzyme activity significantly reduced TCF activity in HCA-7 cells, suggesting that COX-2/PGE 2 signaling may stimulate cell proliferation through transactivation of the TCF-dependent transcription. Mechanistically, we found that PGE 2 -stimulated TCF transcription involved phosphorylation of GSK-3 and induction of TCF-4 transcription factor, which formed a transcriptionally active complex with ␤-catenin. These results suggest a positive feedback loop between the APC/␤-catenin/TCF pathway and the COX-2/PGE 2 signaling system. APC mutation-activated TCF transcription increases the expression of COX-2 and production of PGE 2 (3,38), which then in turn enhances the ␤-catenin/TCF-mediated transcription through a cAMP/PKAdependent mechanism. Apparently, PGE 2 induced oncogenic activity is crucial for tumor development in APC mutationinitiated colorectal carcinogenesis, as inhibition of COX-2, a key enzyme for conversion of arachidonic acid to PGs, results in a dramatic reduction of tumor number and size in humans and mice with germ line APC mutations (2,3).
Numerous studies demonstrate that expression of COX-2 is associated with a growth advantage in colon cancer cells (16,17). The molecular mechanism by which COX-2 promotes proliferation and growth of cancer cells are complex. Recent studies suggest that PGE 2 may transactivate the EGFR signaling system through various mechanisms (17,21,39,40). In the present study, we showed that PGE 2 induced the transcription of cyclin D1, a TCF-4 target gene, through induction of both ␤-catenin and TCF-4. Progression through the mid to late G 1 phase of the mammalian cell cycle is dependent upon the cyclin D1-mediated activation of cyclin-dependent kinases (41). The activated cyclin D-dependent kinases phosphorylate and inactivate the retinoblastoma protein, thereby preventing its inhibition of transcription factors including the E2Fs that are essential for DNA synthesis. The expression of cyclin D1 is elevated in both the familial adenomatous polyposis patient tumor and APC min/ϩ mouse polyp and is thought to be critical for APC mutation-induced neoplasia (42). Given the critical roles of cyclin D1 in cell cycle progression and in colorectal carcinogenesis, our results provide an additional mechanism by which COX-2/PGE 2 promotes the proliferation and growth of colon cancer cells.
VEGF stimulates neoangiogenesis by inducing endothelial cell proliferation, migration, and tubular organization. Seven consensus TCF binding sites have been mapped in the promoter of the human VEGF-1 gene (13). Levels of VEGF-1 mRNA are significantly elevated in primary colon cancers, which contain a mutated APC gene in comparison with tumors that have a wild type APC gene, suggesting that VEGF-1 is an APC/␤-catenin/TCF target gene. On the other hand, COX-2/PGE 2 signaling plays critical roles in neoangiogenesis; homozygous deletion of the EP 2 receptor significantly reduces the number and size of intestinal polyps in APC ⌬716 mice, which is associated with a reduction of VEGF expression, suggesting that PGE 2 /EP 2 signaling is critical for increased levels of VEGF in intestinal neoplasm (6). In support with these observations, we found that administration of 16,16-dimethyl PGE 2 strongly increased levels of VEGF mRNA and protein in APC min/ϩ mouse polyps. Moreover, our data demonstrated that inhibition of GSK-3 or ectopic expression of mutated ␤-catenin significantly increased VEGF transcription, indicating that VEGF-1 is transcriptionally regulated by ␤-catenin/ was dissolved in 10% EtOH and administered by intraperitoneal injection (100 g/kg of body weight) to 4-month-old APC min/ϩ mice (lower panel). Control animals received 10% EtOH (V, upper panel). Each group included three mice and three polyps were randomly collected from each mouse 2 h after 16,16-dimethyl PGE 2 treatment. Protein was extracted from the polyps, and the levels of cyclin D1 were analyzed. Both Western blots (V and PGE2) were subjected to a completely identical procedure. Results shown are representative of two separate experiments. B, PGE 2 induction of VEGF mRNA. APC min/ϩ mice were treated with vehicle or 16,16-dimethyl PGE 2 for the indicated times (2 mice/time point). Levels of VEGF mRNA from pooled intestinal polyps were determined by Northern analyses. Results were similar in two independent experiments. C, PGE 2 induction of VEGF protein. APCmin/ϩ mice were treated with vehicle or 16,16-dimethyl PGE 2 . Each group included three mice, and three polyps were collected from each mouse 2 h after the treatment. The levels of VEGF protein were analyzed by Western blot. Results were similar in two separate experiments.
TCF. However, expression of TCF-4 did not increase VEGF promoter activity; additional experiments are required to identify which member(s) of the TCF/LEF family mediated the activation of VEGF transcription. It has been reported that PGE 2 induction of VEGF in HCT-116 colon cancer cells is mediated by the transcriptional activator hypoxia-inducible factor 1 (20). Our results showed that PGE 2 -increased VEGF transcription involved transactivation of the EGFR system and the MEK pathway synergistically enhanced the ␤-catenin-induced VEGF transcription. These findings suggest that regulation of VEGF expression by PGE 2 is mediated by a complex mechanism and the synergistic actions of PGE 2 on VEGF transcription involves both ␤-catenin/TCF-dependent and -independent mechanisms.
Cumulative evidence suggests that the COX-2/PGE 2 signaling pathway plays critical roles in neoplasia of a variety of organs (43,44); however, this pathway is not a putative oncogenic pathway. Understanding the molecular mechanism mediating COX-2/PGE 2 tumor-promoting effects will help to design novel strategies for prevention and treatment of colon cancers. In the present study, we demonstrated synergistic actions of PGE 2 signaling on the ␤-catenin/TCF-dependent transcription through regulating levels of both ␤-catenin and TCF-4. In addition, PGE 2 may synergistically enhance the transcriptional activity of the ␤-catenin/TCF pathway through transactivating other signaling systems. Previously, we have shown that PGE 2 and the EGFR signaling system synergistically stimulate the growth and migration of colon cancer cells (40). In addition, PGE 2 and the Ras/MAPK pathway synergistically induce the expression of an EGF-like growth factor, amphiregulin. Our studies provide an interesting hypothesis that PGE 2 exerts proneoplastic effects by stimulating critical oncogenic pathways, such as the ␤-catenin/TCF-4 pathway, the EGFR signaling system, and the Ras/MAPK pathway, in a collaborative fashion.
Fujino et al. (30) have reported that the EP 2 signaling pathway induces TCF activity through a cAMP/PKA-dependent mechanism, whereas EP 4 -mediated TCF activation is largely dependent on the phosphatidylinositol 3-kinase pathway. Phosphorylation of ␤-catenin by GSK-3 is crucial for its degradation. Both PKA and phosphatidylinositol 3-kinase/Akt can phosphorylate GSK-3 and inhibit its kinase activity and therefore activate the ␤-catenin/TCF pathway (45). Our previous studies have shown that LS-174T cells express both EP 2 and EP 4 receptors and that activation of the EP 4 receptor is critical for PGE 2 -induced cell migration (16). Additional studies are required to determine which EP receptor(s) is responsible for the PGE 2 -activated ␤-catenin/TCF-dependent transcription in colon cancer cells.
In summary, inactivation of GSK-3 by mutations of the APC gene, which stabilizes ␤-catenin and stimulates TCF-dependent transcription, is thought to be the earliest genetic event in colorectal carcinogenesis (46). COX-2-derived PGE 2 may enhance the APC/␤-catenin/TCF pathway via various mechanisms (Fig. 8). 1) PGE 2 activates the cAMP/PKA pathway, which phosphorylates GSK-3 and further increases levels of cytosolic ␤-catenin. 2) PGE 2 increases the expression of TCF-4 transcription factor, which forms transcriptionally active complex with ␤-catenin. 3) PGE 2 transactivates other signaling pathways, which enhances the ␤-catenin/TCF-dependent transcription in a collaborative fashion. These findings provide a novel mechanism for the understanding of COX-2/PGE 2 proneoplastic actions in colorectal neoplasia.