Carcinogenesis vol.29 no.12 pp.23852393, 2008 doi:10.

1093/carcin/bgn207 Advance Access publication September 4, 2008

Bile acid induces expression of COX-2 through the homeodomain transcription factor CDX1 and orphan nuclear receptor SHP in human gastric cancer cells
Min Jung Park, Kook Hwan Kim, Hye Young Kim, KyeongJin Kim and JaeHun Cheong
Department of Molecular Biology, College of Natural Sciences, Pusan National University, Busan 609-735, Korea
To whom correspondence should be addressed. Tel: 82 51 510 2277; Fax: 82 51 513 9258; Email: molecule85@pusan.ac.kr

The caudal-related homeobox gene, CDX1, encodes for an intestinal-specic transcription factor and is involved in the induction of intestinal metaplasia (IM) of the stomach in gastric cancer. Gastric IM induced by bile reux is a precancerous gastric adenocarcinomal lesion and has been associated with the induction of cyclooxygenase-2 (COX-2). In this study, we demonstrate the molecular mechanisms underlying the transcriptional regulation of COX-2 by bile acid in gastric cells. We noted that the ectopic expression of CDX1 enhanced COX-2 gene expression and that bile acid was associated with the induction of CDX1 expression. Furthermore, the induction of CDX1 by bile acid was mediated by the orphan nuclear receptor, small heterodimer partner (SHP). Finally, it was veried that the expression of COX-2, CDX1, SHP and CCAAT element-binding protein beta messenger RNA in human IM lesions were signicantly higher than in lesions associated with gastritis. Collectively, these results reveal that bile acid induces an increase in the gene expression of COX-2 via the sequential transcriptional induction of SHP and CDX1 in precancerous lesions of human gastric cancer.

of IM of the stomach. Little is currently known regarding the molecular mechanisms of action of CDX1 and CDX2 in the progression from normal gastric epithelium to IM. Cyclooxygenase-2 (COX-2) catalyzes a pivotal step in the formation of arachidonic acid to prostaglandins and is a typical molecular target of the non-steroidal anti-inammatory drugs (15,16). COX-2 is not normally expressed in the majority of normal tissues and can be induced readily by a variety of agents, including growth factors, tumor promoters and inammatory stimuli, but is upregulated in several cancers, most notably gastric cancer (1,1518). Furthermore, increased COX-2 expression has been detected in intestinal-type gastric carcinoma and dysplasia precursor lesions, thereby suggesting that COX-2 may perform a function in the early events associated with gastric carcinogenesis (17). The CCAAT element-binding protein beta (C/EBPb) is a member of the C/EBP family of transcriptional factors and is a crucial regulator of biological processes, including proliferation, differentiation and tumor invasiveness (19,20). Recently, the consensus sequence of C/EBPb has been described in the COX-2 promoter and C/EBPb has been shown to regulate COX-2 expression in gastric cancer cells (21). In accordance with these ndings, C/EBPb overexpression is recognized as a marker in IM and performs an active role in gastric carcinogenesis. However, the molecular relationship of C/EBPb in relation to CDX1-mediated gene regulation during IM remains to be elucidated. In this study, we demonstrated that bile acid-mediated increases in CDX1 expression are required for the induction of COX-2 expression. Moreover, our results demonstrated that the molecular mechanism by which CDX1 expression is mediated involves transcriptional regulation by the orphan nuclear receptor (NR), small heterodimer partner (SHP). Materials and methods

Downloaded from carcin.oxfordjournals.org at University of DundeeGardyne Road Library on June 20, 2011

Introduction Gastric cancer is one of the most frequently experienced malignancies and causes of cancer-related mortality in the world (1). The pathogenesis of gastric cancer is complex, and unlike some other solid tumors, the molecular mechanisms inherent to the development of stomach cancer have yet to be clearly elucidated. Gastric cancer has been classied previously into two histological groupsdiffuse and intestinal type, respectively (2,3). Intestinal metaplasia (IM) is a premalignant condition of the stomach and is associated with chronic atrophic gastritis and gastric carcinoma (3,4). Gastric pathogenesis is believed to result from a combination of environmental factors, including Helicobacter pylori infection, excessive salt intake, low vegetable and fruit intake and the accumulation of specic genetic alterations (3,5). Furthermore, associations between bile reux and IM have been reported, and bile reux has been generally considered to be a risk factor in the production of IM in the stomach and also is believed to function as an initiator of gastric carcinogenesis (6,7). However, the precise mechanism by which bile reux promotes gastric tumor formation remains to be characterized. CDX1 and CDX2, members of the caudal-related homeobox family, are known to be crucial to the control of normal intestinal differentiation and the maintenance of the intestinal phenotype (810). CDX1 and CDX2 are not expressed in the normal stomach, but it has been demonstrated that aberrant expression of these factors induces IM of the stomach, as well as gastric carcinoma (3,1114). Hence, CDX1 and CDX2 are most probably involved in the induction
Abbreviations: COX-2, cyclooxygenase-2; CDCA, chenodeoxycholic acid; C/EBPb, CCAAT element-binding protein beta; DCA, deoxycholic acid; EDTA, ethylenediaminetetraacetic acid; IM, intestinal metaplasia; mRNA, messenger RNA; NF-jB, nuclear factor-kappa B; NR, nuclear receptor; PCR, polymerase chain reaction; RT, reverse transcription; SDS, sodium dodecyl sulfate; SHP, small heterodimer partner; siRNA, small-interference RNA. Construction of reporter plasmids and expression vectors pGL2/COX-2 promoter (327 to 59) construct was generously provided by Dr D.S.Min (Pusan National University, Korea). Two nested 5#-end deletion constructs of the COX-2 promoter were generated by polymerase chain reaction (PCR) using the plasmid pGL2/COX-2 (327 to 59) as the template. Two nested 5#-end deletion constructs were generated using the restriction enzyme MulI. A BglII site was added to the 5# end of COX-2 antisense. Two constructs generated by PCR were amplied with the same 3# oligonucleotide (COX-2 antisense) and the corresponding specic 5# oligonucleotide: COX-2 80 (5#-CGACGCGTAGGTTCTCTCGGTTA-3#), COX-2 20 (5#-CGACGCGTAGGCGGAAAGAAACA-3#) and COX-2 antisense (5#-GGAAGATCTCGCTGCTGAGGAGTTC-3#). The PCR product was double digested with MulI and BglII and cloned into pGL2 luciferase reporter (Promega, Madison, WI). pcDNA3/CDX1 and pGL3B/hCDX1 promoter (1630 to 12) construct were a kind gift from M.A.Yoo (Pusan National University, Korea). pcDNA3/ HA/CDX1 was constructed by inserting PCR fragments of open reading frame into EcoRI- and XhoI-digested pcDNA3/HA expression vector. All plasmids were conrmed by automatic sequencing analysis. Cell culture and bile acid treatment Human gastric carcinoma cell lines MKN45 (a gift from B.J.Park, Pusan National University, Korea) and AGS (a gift from Y.H.Choi, Dongeui University, Korea) were maintained in RPMI 1640 supplemented with 10% heatinactivated fetal bovine serum (Gibco BRL, Grand Island, NY) and 1% (vol/ vol) penicillinstreptomycin (Gibco BRL) at 37C in a humidied atmosphere containing 5% CO2 and fed every 2 days. Deoxycholic acid (DCA) and chenodeoxycholic acid (CDCA) were from Sigma (St Louis, MO). Treatments (DCA and CDCA) were conducted in medium containing 0% fetal bovine serum and 0.2% lactoalbumin hydrolysate for 24 h to avoid the confounding variable of serum-induced signaling. RNA preparation and reverse transcriptionPCR Total RNA from MKN45 and AGS cells was prepared using Trizol reagent (Invitrogen, Carlsbad, CA) according to the manufacturers recommendation.

The Author 2008. Published by Oxford University Press. All rights reserved. For Permissions, please email: journals.permissions@oxfordjournals.org

2385

M.J.Park et al.

The complementary DNA was synthesized from 2 lg of total RNA with avian myeloblastosis virus Reverse Transcriptase (Promega) using a random hexamer (Bioneer, Daejeon, Korea) at 42C for 1 h. The PCR primer for CDX1 gene amplication were 5#-AGGACAAGTACCGCGTGGTCTA-3# (sense) and 5#-CTATGGCAGAAACTCCTCTTTC-3# (antisense); for SHP gene amplication, 5#-AGCTATGTGCACCTCATCGCACCTGC-3# (sense) and 5#-CAAGCAGGCTGGTCGGAATGGACTTG-3# (antisense); for COX-2 gene amplication, 5#-CCTTCCTCCTGTGCCTGATG-3# (sense) and 5#-CTGGCCTCGCTTATGATCT-3# (antisense); for MUC2 gene amplication, 5#-CCATTCTCAACGACAACCCCTACCCC-3# (sense) and 5#-TCCAATGGGAACATCAGGATACATGGTGGC-3# (antisense) and GAPDH gene amplication, 5#-GTGGTCTCCTCTGACTTCAAC-3# (sense) and 5#-TCTCTTCCTCTTGTGCTCTTG-3# (antisense). The complementary DNAs were amplied by PCR under the following conditions: 32 cycles of denaturation at 95C for 40 s, annealing at 57C for 40 s and extension at 72C for 50 s in a thermal cycle. The PCR products were examined by electrophoresis on a 1.2% agarose gel. Transient transfection assay and luciferase assay MKN45 and AGS cells were seeded in a 24-well culture plate and transfected with reporter vector and b-galactosidase expression plasmid, along with each indicated expression plasmids using Jetpei (Polyplus-transfection, Illkirch, France). The pcDNA3.1/HisC plasmid DNA was added to achieve the same total amount of plasmid DNA per transfection. After 48 h of transfection, the cells were rinsed with ice-cold phosphate-buffered saline and lysed with 1 cell culture lysis buffer (Promega). Luciferase activity was determined using an analytical luminescence luminometer according to the manufacturers instructions. Luciferase activity was normalized for transfection efciency using the corresponding b-galactosidase activity and all assays were performed at least in triplicate. The construction of small-interference RNA (siRNA) targeted to CDX1 and SHP was carried out according to the instruction manual provided by the manufacturer (Bioneer). For the construction of CDX1 siRNA (the sense, 5#-ACAAGGAUUCGCCCGUGUA-3# and the antisense, 5#-UACACGGGCGAAUCCUUGU-3#) and SHP siRNA (the sense, 5#-UCAGGAACAUCCUU-3# and the antisense, 5#-AAGGAUGGCAGGUUCCUGA-3#), oligonucleotides were synthesized. The transfection of MKN45 cells was carried out by HiPerFect according to the instruction manual provided by the manufacturer (QIAGEN, Hilden, Germany). Chromatin immunoprecipitation assay MKN45 cells were collected and cross-linked with 1% formaldehyde at 37C for 10 min and then rinsed with ice-cold phosphate-buffered saline twice and centrifuged for 5 min at 2000g. Cells were then resuspended in 0.2 ml lysis buffer [1% sodium dodecyl sulfate (SDS), 10 mM ethylenediaminetetraacetic acid (EDTA), 50 mM TrisCl, pH 8.1, 1 mM phenylmethylsulfonyl uoride and protease inhibitor cocktail (Roche, Rotkreuz, Switzerland)] and sonicated nine times for 10 s each, followed by centrifugation for 10 min. Supernatants were collected and diluted in buffer (1.1% Triton X-100, 0.01% SDS, 1.2 mM EDTA, 167 mM NaCl and 16.7 mM TrisHCl, pH 8.1) followed by immunoclearing with 60 ll protein G-agarose and 2 lg sheared salmon sperm DNA for 30 min at 4C. Immunoprecipitation was performed overnight at 4C with the anti-HA antibody (Roche). After immunoprecipitation, 60 ll protein G-agarose with 2 lg sheared salmon sperm DNA was added, and the incubation was continued for another hour. Precipitates were washed sequentially in the following three washing buffers for 5 min each: a low-salt immune complex washing buffer (0.1% , 1% Triton X-100, 2 mM EDTA, 20 mM TrisHCl, pH 8.1, and 150 mM NaCl), a high-salt immune complex washing buffer (0.1% SDS, 1% Triton X-100, 2 mM EDTA, 20 mM TrisHCl, pH 8.1, and 500 mM NaCl) and a LiCl immune complex washing buffer (250 mM LiCl, 1% NP-40, 1% deoxycholate, 1 mM EDTA and 10 mM TrisHCl, pH 8.1). Precipitates were washed two times with TrisEDTA buffer and then used in PCR assays to determine the binding sequences. For PCR assays, precipitates were extracted three times with 1% SDS and 0.1 M NaHCO3. Elutions were pooled and heated at 65C for at least 6 h to reverse the formaldehyde cross-linking. DNA fragments were puried; 1 ll from 30 ll of DNA extract was reserved for 32 cycles of PCR amplication. The primers for COX-2 promoter were 5#-GGCGGAAAGAAACAGTCATTTCGTC-3# (sense primer) and 5#-CGCTGCTGAGGAGTTCCTGGA-3# (antisense primer). SDSpolyacrylamide gel electrophoresis and western blots Cells were suspended in a lysis buffer (150 mM NaCl, 1% NP-40, 1 mM EDTA and 50 mM Tris, pH 7.5) containing protease inhibitors and quickly sonicated on ice. Protein concentrations were measured using a commercial Protein Assay (Bio-Rad). Whole-cell lysates were subjected to SDSpolyacrylamide gel electrophoresis and transferred to a polyvinylidene diuoride membrane (Millipore, Billerica, MA) by semidry electroblotting. The membranes were

then incubated for 2 h at room temperature with anti-CDX1 (AB4122, Chemicon, Temecula, CA), anti-C/EBPb (sc-150, Santa Cruz Biotechnology, Santa Cruz, CA), anti-COX-2 (sc-1745, Santa Cruz Biotechnology), anti-HA (Roche, Mannheim, Germany) and anti-actin (A 2066, Sigma, Steinheim, Germany) in tris-buffered saline containing 0.05% tween 20 (TBST) supplemented with 1% non-fat dry milk. After washing three times with cold TBST, the blotted membranes were incubated with peroxidase-conjugated secondary antibody (Santa Cruz Biotechnology) for 30 min at room temperature. After washing three times with cold TBST, the bands were detected using an enhanced chemiluminescence system (Amersham, Piscataway, NJ).

Results DCA induces COX-2 expression Different bile acids exert different biological effects on cell growth, cell transformation and gene expression (22,23). In particular, DCA and CDCA perform a function in tumor promotion, and higher levels of bile acid have been reported in patients suffering from atrophic chronic gastritis and IM in gastric carcinoma (6,7). Although bile acids are the principal etiologic agents associated with IM and neoplasia in the gastrointestinal tract and alterations in COX-2 expression are observed during gastric carcinogenesis, the molecular mechanisms underlying the bile acid-mediated overexpression of COX-2 remain to be determined. In an effort to assess the effects of different bile acids on the modulation of COX-2 expression, two gastric cancer cell linesMKN45 and AGS cellswere transiently transfected with human COX-2 promoter (327 to 59). As is shown in Figure 1A, DCA treatment induced an increase in the activity of the COX-2 promoter in a dose-dependent manner, with the maximum effect occurring at 200 lM in the MKN45 and AGS cells. However, CDCA did not induce COX-2 expression at the transcriptional level. As compared with the untreated cells, DCA also induced an increase in the levels of steady-state COX-2 messenger RNA (mRNA) and protein in a dose-dependent fashion in MKN45 and AGS cells (Figure 1B), and this induction began within the rst 6 h, reaching a maximum intensity 24 h after stimulation (Figure 1C). These results demonstrated that COX-2 expression is stimulated by DCA treatment in a dose- and time-dependent manner in gastric cells. CDX1 increases COX-2 expression via transcriptional regulation In an effort to understand the mechanism underlying the regulation of the COX-2 gene, the promoter region of the COX-2 gene was identied. From the database analysis of the promoter region for known consensus sequences in the COX-2 promoter, several putative CDX and C/EBPb response elements were detected in the COX-2 promoter. Other previous reports have indicated that C/EBPb regulates the activity of the COX-2 promoter (21). In order to determine whether CDX1, CDX2 and C/EBPb regulate the expression of the COX-2 gene, MKN45 cells were transfected with the COX-2 promoterluciferase reporter plasmid, along with expression vectors encoding for CDX1, CDX2 and C/EBPb, respectively. CDX1 and C/EBPb expression resulted in a gradual and dose-dependent increase in the promoter activity of COX-2 (Figure 2A). Consistent with the results of the reporter assay, the levels of COX-2 mRNA and protein were signicantly increased in the presence of CDX1 overexpression (Figure 2B). These results indicate that CDX1 and C/EBPb expression, but not CDX2 expression, upregulates COX-2 expression. Direct analysis of the COX-2 promoter sequence (327 to 59) revealed two highly conserved potential CDX1-binding sites. In order to search the CDX1-responsive site of the COX-2 promoter, a series of sequentially deleted constructs of the COX-2 promoter were prepared and subjected to reporter assays involving the cotransfection of the CDX1 expression plasmid into MKN45 cells (Figure 2C). Based on the cotransfection of the CDX1 expression plasmid and a series of COX-2-deleted promoter constructs into MKN45 cells, the COX2 promoter region (80 to 59), including just one CDX1 site, was considered a CDX1-responsive site on the transactivation assay (Figure 2D). In order to determine whether an increase in the transactivation of the COX-2 promoter as the result of CDX1 expression is

Downloaded from carcin.oxfordjournals.org at University of DundeeGardyne Road Library on June 20, 2011

2386

Expression of COX-2 in human gastric cancer cells

Downloaded from carcin.oxfordjournals.org at University of DundeeGardyne Road Library on June 20, 2011

Fig. 1. DCA induces COX-2 expression. (A) DCA induces COX-2 promoter activity in MKN45 and AGS cells. Cells were transfected with 0.1 lg of human COX-2-luciferase reporter vector (327 to 59). After 24 h of transfection, the cells were treated for 24 h with DCA and CDCA at the indicated concentrations. Forty-eight hours after transfection, the cells were harvested and the cell lysates were assayed for luciferase activity, and the results were normalized to the b-galactosidase activity. Bars indicate the mean value of at least three independent experiments. The standard deviation is indicated by error bars. (B) DCA increases COX-2 mRNA (upper panel) and protein (lower panel) in a dose-dependent manner. Cells were exposed for 24 h to the indicated concentration of DCA. (C) DCA increases COX-2 mRNA (upper panel) and protein (lower panel) in a time-dependent manner. The cells were treated with DCA (100 lM) for each indicated time period. Total RNA was prepared from the cells, and then the COX-2 mRNA levels were detected via RTPCR with GAPDH as a loading control (B and C, upper panel). Cell extracts (40 lg) from each sample were separated by SDSpolyacrylamide gel electrophoresis and analyzed via western blotting using specic antibodies for COX-2 and actin as a loading control (B and C, lower panel).

the result of direct DNA-binding activity, a chromatin immunoprecipitation assay was conducted. Isolated DNA was subjected to PCR both prior to (input) and after chromatin immunoprecipitation using primer sets designed to amplify the region with the COX-2 promoter harboring the CDX1-binding site. The results of the chromatin immunoprecipitation assay showed that CDX1 interacted directly with the COX-2 promoter (80 to 59) (Figure 2E). However, no amplied PCR products were detected in the lane of pcDNA/HA-transfected cells. This result indicates that CDX1 induces an increase in COX-2 expression at the transcriptional level via direct promoter binding. DCA increases COX-2 expression via the induction of CDX1 Next, in order to determine whether DCA increases COX-2 expression via the transcriptional regulation of CDX1, MKN45 and AGS cells were transfected with the human CDX1 promoter and then maintained either under control conditions or in the presence of different concentrations of DCA for 24 h. CDX1 promoter activity was increased in a DCA dose-dependent manner in two gastric cell lines (Figure 3A, upper panel). Furthermore, DCA (100 lM) induced an increase in the level of CDX1 mRNA in both cells as compared with the untreated cells (Figure 3A, lower panel). In Figure 3B, MKN45 and AGS cells were transfected with the indicated amounts of the expression vectors encoding for CDX1 along with the human COX-2 promoter in the absence or presence of different concentrations of DCA. The results demonstrated that CDX1 and DCA cooperatively increased COX-2 promoter activity in a dose-dependent fashion. In order to further conrm the induction of the COX-2 gene by CDX1 and DCA, MKN45 and AGS cells were transiently transfected with CDX1 expression vector or with empty vector as a control, then

treated with 100 lM of DCA and subjected to reverse transcription (RT)PCR and western blot assay. As is shown in Figure 3C, the CDX1 expression and DCA treatment signicantly stimulated COX2 mRNA and protein levels. These results show that DCA stimulates COX-2 expression via the induction of CDX1 in gastric cells. Next, in order to obtain additional insight into the function of CDX1 in the context of COX-2 expression, loss of CDX1 expression studies were conducted using a siRNA approach. The knockdown of CDX1 attenuated the transcriptional activity of the COX-2 promoter in the presence of DCA (Figure 3D). In order to verify the level of COX-2 expression along with siRNA directed against CDX1, RTPCR and western blot assays were conducted. The knockdown of CDX1 resulted in a reduction in the levels of COX-2 mRNA and protein expression (Figure 3E). These results showed that CDX1 performs a critical function in DCA-induced COX-2 expression. SHP is required for DCA-induced CDX1 expression NRs are ligand-activated transcription factors, which are involved in the transcriptional regulation of several aspects of mammalian physiology, and have been identied to regulate a variety of genes by ligands (24,25). SHP is a member of the NR-related proteins and is an unique orphan NR that lacks a conserved DNA-binding domain and is expressed specically in the liver and a limited number of other tissues (26,27). In the murine macrophage RAW 264.7 cell line, SHP interacts directly with the nuclear factor-kappa B (NF-jB) p65 subunit and functions as a positive transcriptional coregulator of NF-jB (28). In order to characterize the mechanism of CDX1 gene expression by bile acid, the promoter region of the CDX1 gene was

2387

M.J.Park et al.

Downloaded from carcin.oxfordjournals.org at University of DundeeGardyne Road Library on June 20, 2011

Fig. 2. CDX1 increases COX-2 expression. (A) MKN45 cells were transfected with the indicated amounts of the expression vectors encoding for CDX1, CDX2 and C/EBPb along with the human COX-2 promoterluciferase reporter vector. After 48 h of transfection, promoter activities were measured as luciferase activities normalized to b-galactosidase activities. Bars indicate the mean value of at least three independent experiments. The standard deviation is indicated by error bars. (B) MKN45 cells transiently transfected with the indicated amounts of expression vector encoding for CDX1. Total RNA was prepared from the cells, and then the CDX1 and COX-2 mRNA levels were detected by RTPCR with GAPDH as a loading control (upper panel) and protein expression was tested with western blot analysis using specic antibodies for HA, COX-2 and actin (lower panel). (C) A schematic representation of the deletion constructs of the hCOX-2 promoterreporter gene used in transient transfection assays. (D) Luciferase activity of the promoter-deletion constructs in MKN45 cells. The MKN45 cells were transfected with the indicated promoter-deletion constructs of the COX-2 gene along with the expression vector encoding for CDX1 or empty vector. (E) The chromatin immunoprecipitation assay. Empty vector pcDNA3/HA and pcDNA3/HA/CDX1 expression vector were transfected into MKN45 cells. Following formaldehyde cross-linking, soluble chromatin was prepared. After immunoprecipitation with HA antibody, precipitated DNAs were used in PCR analysis. The input lane shows the starting chromatin extracts.

identied. Via a search of the GeneBank database, the full-length CDX1 promoter harbored responsive sites for activator protein 1, acute myelogenous leukemia1/runt-related transcription factor1, NF-jB, sterol regulatory element binding protein and CDX (29,30). It was recently reported that bile acid increases the activation of NF-jB signaling and the active NF-jB subunit (p65) exerts a direct effect on CDX1 expression (30). On the basis of these ndings, in order to assess the role of SHP in DCA-induced CDX1 expression, the effects of SHP on the regulation of human CDX1 promoter activity were assessed. MKN45 and AGS cells were transiently transfected with an expression vector encoding for SHP and the CDX1 promoterreporter plasmid. The results provided in Figure 4A illustrate that SHP expression increased the luciferase activity derived from the CDX1 promoter. The protein and mRNA induction of COX-2 were also enhanced by SHP expression (Figure 4B). These results indicated that increased levels of CDX1 by SHP were associated with an increase in COX-2 expression. Next, in order to determine whether DCA-induced CDX1 expression could be mediated by SHP, MKN45 and AGS cells were transfected with the indicated amounts of the expression vectors encoding for SHP, then cultured in the presence or absence of DCA. As is shown in Figure 4C, DCA alone induced the expression of SHP. Furthermore, the cooperation of SHP and DCA induced a substantially increased expression of CDX1 mRNA (Figure 4C). In order to further conrm the role of SHP in CDX1 expression in gastric cells, the siRNA of SHP was transfected into both MKN45 and AGS cells. The knock-

down of SHP resulted in a reduction in the transcription of the CDX1 promoter in the presence of DCA (Figure 4D). The level of CDX1 expression along with siRNA directed against SHP was assessed via RTPCR. Consistent with the results of the examination of CDX1 promoter activity, the loss of SHP expression caused a reduction in the levels of CDX1 mRNA (Figure 4E). These results showed that SHP is required for DCA-induced CDX1 expression. Thus, DCA induces COX-2 expression in gastric cells via the induction of CDX1 expression through the transcriptional regulation of SHP. Cooperation of CDX1 and C/EBPb increases COX-2 promoter activity and protein expression The transcription factor, C/EBPb, has been associated previously with tumorigenesis and is also recognized to function as a transcription factor for COX-2 in cases of gastric cancer (1921). In order to determine whether C/EBPb performs a function in the CDX1-dependent transactivation of the COX-2 promoter, we cotransfected cells with the COX-2 luciferase reporter plasmid, along with expression vectors encoding for CDX1 and C/EBPb. The coexpression of CDX1 and C/ EBPb resulted in a cooperative transcription activity of the COX-2 promoter (Figure 5A). As had been predicted, the combined activity of CDX1 and C/EBPb gave rise to increased COX-2 protein expression (Figure 5B). These results raise the possibility that CDX1 and C/EBPb may be markers of an early event in IM and may also perform an active function in the regulation of COX-2 induction.

2388

Expression of COX-2 in human gastric cancer cells

Downloaded from carcin.oxfordjournals.org at University of DundeeGardyne Road Library on June 20, 2011

Fig. 3. DCA increases COX-2 expression by inducing CDX1. (A) MKN45 and AGS cells were transfected with 0.1 lg of human CDX1 promoterreporter vector. After 24 h of transfection, cells were treated for 24 h with DCA. Cell lysates were assayed for luciferase activity and the results were normalized to the b-galactosidase activity. Bars indicate the mean value of at least three independent experiments. The standard deviation is indicated by error bars (upper panel). The CDX1 mRNA levels were detected by RTPCR with GAPDH as a loading control (lower panel). (B) MKN45 and AGS cells were transfected with the indicated amounts of the expression vectors encoding for CDX1 with the human COX-2 promoterreporter vector in the presence or absence of different concentrations of DCA for 24 h. Cell lysates were assayed for luciferase activity. (C) MKN45 and AGS cells were transiently transfected with expression vector pcDNA3/HA alone or containing CDX1 expression vectors. The CDX1 and COX-2 mRNA levels were detected via RTPCR (left panel) and protein expressions were tested with western blot analysis using antibodies specic for HA, COX-2 and actin (right panel). (D) MKN45 cells transiently transfected with siRNA oligonucleotide targeted to CDX1 (CDX1 siRNA) or a control siRNA with the human COX-2 luciferase reporter vector and treated for 24 h with or without 100 lM DCA. Promoter activities were measured as luciferase activities. (E) The COX-2 mRNA levels were detected by RTPCR with GAPDH as a loading control (left panel). Western blot analysis using antibodies specic for CDX1, COX-2 and actin as indicated (right panel).

2389

M.J.Park et al.

Downloaded from carcin.oxfordjournals.org at University of DundeeGardyne Road Library on June 20, 2011

Fig. 4. SHP is required for DCA-induced CDX1 expression. (A) MKN45 and AGS cells were transfected with the indicated amounts of the expression vectors encoding for SHP along with the human CDX1 promoterreporter vector. After 48 h of transfection, promoter activities were measured as luciferase activities normalized to b-galactosidase activities. (B) MKN45 and AGS cells were transiently transfected with the indicated amounts of expression vector encoding for SHP or empty vector. Total RNA was prepared from the cells, and then the SHP, CDX1 and COX-2 mRNA levels were detected via RTPCR with GAPDH as a loading control (upper panel) and protein expressions were tested via western blot analysis using specic antibodies for HA, CDX1, COX-2, and actin (lower panel). (C) MKN45 and AGS cells were transiently transfected with expression vector encoding for SHP. After 24 h of transfection, the cells were treated for 24 h with 100 lM DCA. SHP and CDX1 mRNA levels were detected via RTPCR with GAPDH as a loading control. The values are expressed as the means SDs (n 5 6); P , 0.0005, compared with mock transfectants; P , 0.01, compared with SHP transfectants. (D) MKN45 and AGS cells were transfected with the siRNA oligonucleotides targeted to SHP or control siRNA along with the human CDX1 promoterreporter and treated for 24 h with 100 lM DCA. Promoter activities were measured as luciferase activities. (E) The SHP, CDX1 and COX-2 mRNA levels were detected via RTPCR using GAPDH as a loading control.

Fig. 5. Cooperation of CDX1 and C/EBPb increases COX-2 expression. (A) MKN45 cells were transiently transfected with COX-2 promoterreporter vector alone or with reporter plasmid plus the indicated expression vectors encoding for CDX1 and C/EBPb. Promoter activities were measured as luciferase activities normalized to b-galactosidase activities. Values were expressed as means SDs (n 5 5); P , 0.001, compared with mock transfectants; P , 0.005, compared with CDX1 transfectants. (B) After the transient transfection of HA/CDX1 and/or C/EBPb into MKN45 cells, the cell lysates were analyzed via western blot analysis using anti-HA, C/EBPb and COX-2. Actin was utilized as a loading control.

Relative expression levels of COX-2, CDX1, C/EBPb, SHP and MUC2 mRNA in gastritis patients with and without IM In order to compare the in vivo expression of IM markers and the transcription factors described in this study, human gastritis tissues were collected via endoscopic resection from each of 11 patients after they had provided their informed consent. The patients were dened by their possession of an aberrant high level of bile acid in the bottom of the stomach. The IM region and gastritis mucosa were extracted via

endoscopy and the pathological identities were veried via tissue examination, and RTPCR was conducted to assess the expression of COX-2, CDX1, SHP, C/EBPb and MUC2 mRNA in mucosal tissues with and without IM. Figure 6A shows that the mRNA expression of COX-2, CDX1, SHP and C/EBPb was signicantly higher in the IM than in gastritis. As is shown in Figure 6B and C, COX-2 and SHP were expressed abundantly in the IM region. MUC2 expression, which is detected in goblet cells and considered to be an intestinal differentiation marker, was evaluated in cells possessing IM. The

2390

Expression of COX-2 in human gastric cancer cells

Downloaded from carcin.oxfordjournals.org at University of DundeeGardyne Road Library on June 20, 2011

Fig. 6. Relative expression levels of COX-2, CDX1, C/EBPb, SHP and MUC2 mRNA in gastritis patients with and without IM. (A) mRNA expression pattern of COX-2, CDX1, SHP and C/EBPb in 11 patients. Tissue samples for molecular analysis were collected from patients undergoing gastritis surgery or endoscopy for the diagnosis of gastric disease. Numbers represent patient numbers evidencing the same intensity of mRNA expression. (B) Magnied images of surface pattern of IM and gastritis mucosa with hematoxylin and eosin staining. Total RNA was prepared from two gastric mucosa representing gastritis and IM, and then the COX-2, CDX1, SHP, C/EBPb and MUC2 mRNA levels were detected by RTPCR with GAPDH as a loading control. (C) Densitometry analysis of mRNA expression. The ratio of mRNA expression in IM is compared with its expression of gastritis as a control.

results indicate that the concomitant expression of CDX1, SHP and C/ EBPb may be diagnostic markers of IM and may induce IM via a cascade of transcriptional regulation for precancerous target genes. Discussion Metaplasia is a potentially reversible change from one cell type to another, a process referred to as transdifferentiation. Metaplastic cells are more susceptible to neoplasia. Understanding the steps leading to metaplasia is crucial for the identication of early prognostic or diagnostic markers of tumor development (5,31). Chronic gastritis and associated IM are fairly common precancerous conditions for gastric carcinoma (3). IM of the stomach is a risk factor in the development of intestinal-type gastric cancer. The most commonly recognized metaplasia of the gut is characterized by the transdifferentiation of gastric epithelial cells to an intestinal phenotype. The phenotype of IM is characterized by mucin-secreting goblet cells and absorptive enterocytes with villin that express intestinal plasma membrane-specic enzymes and may also include the presence of peptide hormonesecreting enteroendocrine cells and antimicrobial-secreting Paneth cells (13,32). This phenotype is also veried by demonstrating intestine-specic gene expression, including MUC2, villin, sucrase isomaltase, carbonic anhydrase I, intestinal trefoil factor and intestinal fatty acid-binding protein within the gastric mucosa (3,33). Microsatellite instability, mutations in p53, upregulation of oncogenes, downregulation of tumor suppressor genes and COX-2 expression have all been associated with the development of IM (34). Several transcription factors, including CDX1, CDX2, PDX1 and OCT1 have been recognized in the IM and are believed to control IM, but the exact molecular mechanisms underlying these phenomena remain to be clearly elucidated (3,13,35,36). It was suggested previously that there are two types of remnant gastritishistological gastritis caused by H.pylori infection and endoscopic gastritis induced by bile reux (37). Similar results have been reported elsewhere (38), but some researchers have reported the synergistic effects of bile reux and H.pylori infection on the development of chronic atrophic gastritis and IM. However, the mech-

anism underlying bile-induced gastric mucosal damage has yet to be fully elucidated. It has been reported previously that the impairment of energy metabolism (39), excessive nitric oxide production (40) and other factors may perform a crucial function in the development of gastric mucosal injury. Bile acids activate the farnesoid X receptor and increase the expression of the SHP gene, thereby inducing the activation of a variety of signaling cascades and promoting cell proliferation. It would be valuable to conduct continued molecular analyses of the effects of bile on the gastrointestinal mucosa. The pathogenesis of IM may be reversible via the treatment of H.pylori infection, as well as by dietary changes. Bile reux into the stomach is associated signicantly with the presence of IM. Furthermore, bile reux directly induces IM and progression to neoplasia in the gastrointestinal tract in animal models (7,41,42). On the basis of these studies, bile reux is considered to be the most important causative factor and an initiator of gastric carcinogenesis. In this study, we rst demonstrate that the caudal-related homeobox gene, CDX1, an intestine-specic transcription factor, participates in the control of COX-2 expression by bile acids in gastric cells. It has been shown that the interaction of bile acid with the M3 muscarinic receptor subtype expressed in chief cells may participate in mucosal damage, manifesting as active inammation, IMs, glandular atrophy and focal hyperplasias and other pathophysiological consequences of bile reux. Reuxed bile may induce direct injuries of mucous epithelial cells and the tight junctions between these cells. It could also provoke inammatory responses in the entirety of the mucous and submucous layers. Furthermore, by changing the chemical environment of the mucous surface, bile acid may inuence the pathogenic patterns of other damage-causing factors, including gastric acid and H.pylori, and may also potentiate the actions of these factors. Consequently, bile reux exerts harmful effects on gastric mucosa throughout the entire stomach. The transdifferentiation mechanism that characterizes IM might be explained by the switching of the expression of transcription factors such as homeobox genes. Several studies have reported previously that the expression of CDX1 and CDX2 plays a pivotal role in the development of IM and directs programs to induce IM with aberrant

2391

M.J.Park et al.

differentiation, structures and proliferation activities (1113). Villin, sucraseisomaltase, alkaline phosphatase, MUC2 and trefoil family factor 3 have all been investigated in conjunction with IM, leading to gastric cancer (4244). Recently, the intestine-specic homeobox genes, CDX1 and CDX2, were shown to be associated closely with MUC2 and trefoil family factor 3 expression, and it was suggested that CDX1 and/or CDX2 perform crucial roles in the aberrant intestinal differentiation program and gastric carcinogenesis, due in part to the regulation of MUC2 and trefoil family factor 3 at the transcriptional level (44). These markers have been useful for the identication of metaplastic tissue. However, it remains unclear as to whether the expression of these factors is functionally signicant. Interestingly, CDX1, but not CDX2, exerts a signicant effect on the COX-2 promoter in MKN45 cells, although the two genes harbor the same CDXbinding sites (TTTAT). These observations indicate that CDX1 may cooperate with other transcription factors, including C/EBPb, and that these factors are involved in the regulation of COX-2 expression during gastric carcinogenesis. Recently, the expression of the nuclear bile acid receptor, farnesoid X receptor, has been reported in association with gastric cancer. Bile acids activate this receptor and induce an increase in SHP gene expression, thereby causing the activation of a variety of signaling cascades and promoting cell proliferation (5,45). In this study, increased SHP expression has been demonstrated to potentiate the induction of CDX1 protein in gastric cells. CDX1 performs a critical role in increasing COX-2 production with another transcription factor, C/EBPb. The molecular mechanisms underlying the cooperation of C/EBPb with CDX1 for the enhancement of COX-2 expression remain elusive. It is possible that (i) the functional direct interaction of CDX1 and C/EBPb on the COX-2 promoter increases its gene transcription; (ii) since the CDX1 promoter has a few binding sites for C/EBPb, CDX1 expression is augmented by C/EBPb or (iii) CDX1 and C/EBPb may be able to form a transcriptional active complex on the COX-2 promoter via a coactivators. In conclusion, this study demonstrates the signaling mechanisms by which bile reux induces COX-2 expression to generate IM. Specifically, the results of this study demonstrate that bile acid activates SHP expression, which induces the expression of the intestinal-specic transcription factor, CDX1, in gastric carcinoma cells. In turn, CDX1 is capable of inducing COX-2. Moreover, the present observations suggest that the identication of the factors required for IM and the clarication of the molecular mechanisms responsible for an early phenotypic change in gastric carcinogenesis may facilitate the development of therapeutic interventions. Funding Korea Research Foundation Grant funded by the Korean Government (KRF-C00467(I03011)); Pusan National University in program. PostDoc. 2008. Acknowledgements
We thank Dr Ung Suk Yang (Bongseng Hospital) and Dr Won Il Park (Bongseng Hospital) for tissue samples used in this work. Conict of Interest Statement: None declared.

References
1. Ristimaki,A. et al. (1997) Expression of cyclooxygenase-2 in human gas tric carcinoma. Cancer Res., 57, 12761280. 2. Vauhkonen,M. et al. (2006) Pathology and molecular biology of gastric cancer. Best Pract. Res. Clin. Gastroenterol., 20, 651674. 3. Yuasa,Y. (2003) Control of gut differentiation and intestinal-type gastric carcinogenesis. Nat. Rev. Cancer, 3, 592600. 4. Tsukamoto,T. et al. (2006) Gastric-and-intestinal mixed-type intestinal metaplasia: aberrant expression of transcription factors and stem cell intestinalization. Gastric Cancer, 9, 156166. 5. Walker,M.M. (2003) Is intestinal metaplasia of the stomach reversible? Gut, 52, 14.

6. Dixon,M.F. et al. (2002) Bile reux gastritis and intestinal metaplasia at the cardia. Gut, 51, 351355. 7. Sobala,G.M. et al. (1993) Bile reux and intestinal metaplasia in gastric mucosa. J. Clin. Pathol., 46, 235240. 8. Beck,F. (2004) The role of Cdx genes in the mammalian gut. Gut, 53, 13941396. 9. Keller,M.S. et al. (2004) Cdx1 or Cdx2 expression activates E-cadherinmediated cell-cell adhesion and compaction in human COLO 205 cells. Am. J. Physiol. Gastrointest. Liver Physiol., 287, G104G114. 10. Silberg,D.G. et al. (2000) Cdx1 and cdx2 expression during intestinal development. Gastroenterology, 119, 961971. 11. Mutoh,H. et al. (2005) Cdx2 species the differentiation of morphological as well as functional absorptive enterocytes of the small intestine. Int. J. Dev. Biol., 49, 867871. 12. Tsukamoto,T. et al. (2004) Down-regulation of a gastric transcription factor, Sox2, and ectopic expression of intestinal homeobox genes, Cdx1 and Cdx2: inverse correlation during progression from gastric/intestinal-mixed to complete intestinal metaplasia. J. Cancer Res. Clin. Oncol., 130, 135145. 13. Mutoh,H. et al. (2004) Cdx1 induced intestinal metaplasia in the transgenic mouse stomach: comparative study with Cdx2 transgenic mice. Gut, 53, 14161423. 14. Silberg,D.G. et al. (2002) Cdx2 ectopic expression induces gastric intestinal metaplasia in transgenic mice. Gastroenterology, 122, 689696. 15. Vane,J.R. et al. (1998) Cyclooxygenases 1 and 2. Annu. Rev. Pharmacol. Toxicol., 38, 97120. 16. Dai,Y. (2006) Non-steroidal anti-inammatory drugs in prevention of gastric cancer. World J. Gastroenterol., 12, 28842889. 17. Saukkonen,K. et al. (2003) Cyclooxygenase-2 and gastric carcinogenesis. APMIS, 111, 915925. 18. Mao,X.Y. et al. (2007) COX-2 expression in gastric cancer and its relationship with angiogenesis using tissue microarray. World J. Gastroenterol., 13, 34663471. 19. Sankpal,N.V. et al. (2006) Overexpression of CEBPbeta correlates with decreased TFF1 in gastric cancer. Oncogene, 25, 643649. 20. Milne,A.N. et al. (2006) Early-onset gastric cancers have a different molecular expression prole than conventional gastric cancers. Mod. Pathol., 19, 564572. 21. Regalo,G. et al. (2006) C/EBPbeta is over-expressed in gastric carcinogenesis and is associated with COX-2 expression. J. Pathol., 210, 398404. 22. Martinez,J.D. et al. (1998) Different bile acids exhibit distinct biological effects. Nutr. Cancer, 31, 111118. 23. Shiraki,K. et al. (2005) Different effects of bile acids, ursodeoxycholic acid and deoxycholic acid, on cell growth and cell death in human colonic adenocarcinoma cells. Int. J. Mol. Med., 16, 729733. 24. Chiang,J.Y. (2002) Bile acid regulation of gene expression: roles of nuclear hormone receptors. Endocr. Rev., 23, 443463. 25. Moore,J.T. et al. (2002) Nuclear receptor regulation of genes involved in bile acid metabolism. Crit. Rev. Eukaryot. Gene Expr., 12, 119135. 26. Bavner,A. et al. (2005) Transcriptional corepression by SHP: molecular mechanisms and physiological consequences. Trends Endocrinol. Metab., 16, 478488. 27. Bookout,A.L. et al. (2006) Anatomical proling of nuclear receptor expression reveals a hierarchical transcriptional network. Cell, 126, 789799. 28. Kim,Y.S. et al. (2001) The orphan nuclear receptor small heterodimer partner as a novel coregulator of nuclear factor-kappa b in oxidized low density lipoprotein-treated macrophage cell line RAW 264.7. J. Biol. Chem., 276, 3373633740. 29. Beland,M. et al. (2004) Cdx1 autoregulation is governed by a novel Cdx1LEF1 transcription complex. Mol. Cell. Biol., 24, 50285038. 30. Wong,N.A. et al. (2005) CDX1 is an important molecular mediator of Barretts metaplasia. Proc. Natl Acad. Sci. USA, 102, 75657570. 31. Quinlan,J.M. et al. (2007) Epithelial metaplasia and the development of cancer. Biochim. Biophys. Acta, 1776, 1021. 32. Mesquita,P. et al. (2006) Metaplasia-a transdifferentiation process that facilitates cancer development: the model of gastric intestinal metaplasia. Crit. Rev. Oncog., 12, 123126. 33. Tatematsu,M. et al. (2003) Stem cells and gastric cancer: role of gastric and intestinal mixed intestinal metaplasia. Cancer Sci., 94, 135141. 34. Giraud,A.S. (2007) Metaplasia as a premalignant pathology in the stomach. Gastroenterology, 132, 20532056. 35. Sakai,H. et al. (2004) PDX1 homeobox protein expression in pseudopyloric glands and gastric carcinomas. Gut, 53, 323330. 36. Almeida,R. et al. (2005) OCT-1 is over-expressed in intestinal metaplasia and intestinal gastric carcinomas and binds to, but does not transactivate, CDX2 in gastric cells. J. Pathol., 207, 396401.

Downloaded from carcin.oxfordjournals.org at University of DundeeGardyne Road Library on June 20, 2011

2392

Expression of COX-2 in human gastric cancer cells

37. Abe,H. et al. (2005) Inuence of bile reux and Helicobacter pylori infection on gastritis in the remnant gastric mucosa after distal gastrectomy. J. Gastroenterol., 40, 563569. 38. Lee,Y. et al. (2004) Inammation of the gastric remnant after gastrectomy: muscosal erythema is associated with bile reux and inammatory cellular inltration is associated with Helicobacter pylori infection. J. Gastroenterol., 39, 520526. 39. Nunex,D. et al. (1989) Effects of bile salts on energy metabolism and acid secretion by the isolated toad gastric mucosa. Gastroenterology, 96, 130138. 40. Wallance,J.L. et al. (2000) Nitric oxide in mucosal defense: a little goes a long way. Gastroenterology, 119, 512520. 41. Kazumori,H. et al. (2006) Bile acids directly augment caudal related homeobox gene Cdx2 expression in oesophageal keratinocytes in Barretts epithelium. Gut, 55, 1625.

42. Newberne,P.M. et al. (1987) Gastric carcinogenesis: a model for the identication of risk factors. Cancer Lett., 38, 149163. 43. Matsuda,K. et al. (2007) Quantitative analysis of the effect of Helicobacter pylori on the expressions of SOX2, CDX2, MUC2, MUC5AC, MUC6, TFF1, TFF2, and TFF3 mRNAs in human gastric carcinoma cells. Scand. J. Gastroenterol., 27, 19. 44. Mesquita,P. et al. (2003) Human MUC2 mucin gene is transcriptionally regulated by Cdx homeodomain proteins in gastrointestinal carcinoma cell lines. J. Biol. Chem., 278, 5154951556. 45. Kuwahara,A. et al. (1989) Bile acids promote carcinogenesis in the remnant stomach of rats. J. Cancer Res. Clin. Oncol., 115, 423428. Received January 25, 2008; revised August 25, 2008; accepted August 25, 2008

Downloaded from carcin.oxfordjournals.org at University of DundeeGardyne Road Library on June 20, 2011

2393