Clear benefits have been reported in epidemiological studies with non-steroidal anti-inflammatory drugs (NSAIDs) as chemopreventive agents against colon cancers, one of the most common malignancies in humans.¹ Chemically induced colon carcinogenesis in rodents is also suppressed by administration of NSAIDs.^2–4 Moreover, intestinal polyp formation in familial adenomatous polyposis coli patients is markedly reduced after application of agents such as sulindac or indomethacin.^5–8 The common mechanism of action of NSAIDs is inhibition of cyclooxygenase (COX) activity, two distinct isoforms of which have been reported: a constitutive enzyme, COX-1, and an inducible enzyme, COX-2.⁹ COX-1 and COX-2 are rate limiting enzymes in the synthesis of prostanoids which affect cell proliferation, tumour growth, apoptosis, and immune responsiveness, and both COX isoforms have been reported to be involved in colon carcinogenesis.^1,4,10

Prostanoids such as prostaglandin (PG)E₂, PGD₂, PGF_2α, PGI₂, and TXA₂ exert their biological actions through binding to nine specific receptors with seven transmembrane domains: the four subtypes EP₁–EP₄ for PGE₂, DP and CRTH2 for PGD₂, FP for PGF_2α, IP for PGI₂, and TP for TXA₂.^11,12 Several reports have demonstrated increased levels of PGE₂ in human colon cancer tissues compared with surrounding normal mucosa.^13–15 Signal transduction pathways of PGE₂ receptors have been studied by examining agonist induced changes in the levels of second messengers such as cAMP and free Ca²⁺ and by identifying G protein coupling by various methods.¹¹ The EP₁ receptor is known to mediate PGE₂ induced elevation of free Ca²⁺ concentration although the species of G protein to which EP₁ receptor is coupled remains unidentified. EP₂ and EP₄ receptors are coupled to Gs and stimulate cAMP production by adenylate cyclase. In contrast, the major signalling pathway for the EP₃ receptor is inhibition of adenylate cyclase via Gi. In addition, another function has been suggested for this receptor type in which cell phenotype is regulated through activation of Rho via G proteins other than Gi.¹⁶

Establishment of mice lacking the genes encoding prostanoid receptors has promoted understanding of the involvement of prostanoids¹¹ and their receptors in the development of colon cancer.^16–18 In previous studies, we demonstrated that deficiency of either EP₁ or EP₄ receptor decreases formation of azoxymethane (AOM) induced aberrant crypt foci (ACF), putative preneoplastic lesions in the colon.^17,18 Moreover, antagonists of EP₁ and EP₄ receptors suppress formation of AOM induced ACF in the colon of mice and intestinal polyp formation in Apc gene deficient Min mice.^17,18 Recently, it was also reported that homozygous deletion of the gene encoding the EP₂ receptor resulted in a decrease in intestinal polyp formation in Apc knockout mice.¹⁹ As already mentioned, EP₂ and EP₄ stimulate adenylate cyclase whereas EP₃ exerts an inhibitory influence, suggesting a possible suppressive role against colon carcinogenesis. However, deficiency of EP₃ did not affect AOM induced ACF formation in our previous study.¹⁷

In the present study, we hypothesised that EP₃ might act at a later stage in colon carcinogenesis. Examination of mRNA expression for EP₁, EP₂, EP₃, and EP₄ in colon carcinomas of mice, rats, and humans demonstrated that levels of EP₃ were markedly decreased compared with normal mucosa. An increase in colon carcinoma formation induced by AOM was also demonstrated in EP₃ receptor knockout mice. Furthermore, activation of the EP₃ receptor showed a suppressive effect on cell growth in a colon cancer cell line in which EP₃ was expressed. In most human colon cancer cell lines tested, EP₃ expression was not detected but treatment with 5-aza-2′-deoxycytidine (5-aza-dC) restored EP₃ expression in some cell lines. On the basis of the results obtained, the role of the EP₃ receptor in colon carcinogenesis is discussed.

MATERIALS AND METHODS

Animals

The mouse gene encoding the PGE₂ receptor EP₃ was disrupted by a gene knockout method using homologous recombination, as reported previously.¹⁷ The generated chimeric mice were backcrossed with C57BL/6Cr mice, and the resulting homozygous mutant mice of these F2 progeny were backcrossed into the C57BL/6Cr background for 10 generations. EP₃ receptor deficient male mice were used at six weeks of age. Genotypes of the knockout mice were confirmed by polymerase chain reaction (PCR) according to the method described previously.¹⁷ Animals were housed in plastic cages at 24±2°C and 55% relative humidity with a 12 h/12 h light/dark cycle. Water and basal diet (AIN-76A; Bio-Serv, Frenchtown, New Jersey, USA) were given ad libitum. Body weights and food intake were measured weekly.

Colon tumour samples and cell lines

Mouse colon tumours and normal colon mucosa tissues were obtained from C57BL/6J male mice treated with AOM, as previously reported.¹⁸ Rat colon tumours and normal colon mucosa tissues were obtained from eight F344 male rats treated with AOM, as previously reported.²⁰ Frozen samples of mouse and rat tissues were used for reverse transcription (RT)-PCR analyses, and formalin fixed, paraffin embedded rat tissue samples were employed for immunohistochemical staining.

Surgical specimens of human colon cancer and adjacent normal colon mucosa tissues were taken from eight Japanese patients who had undergone surgical operations for colorectal cancers at the National Cancer Center Hospital, Tokyo, and samples were immediately frozen in liquid nitrogen.

Eleven human colon cancer cell lines were subjected to RT-PCR analysis. HCA-7 colony 29, a human colon adenocarcinoma cell line, was kindly provided by Dr Susan Kirkland, Imperial College of Science, Technology, and Medicine (London, UK).²¹ HCA-7 cells were maintained in Dulbecco’s minimum essential medium supplemented with 5% heat inactivated fetal bovine serum (FBS) (Hyclone Laboratories, Inc., Logan, Utah, USA) and antibiotics (100 μg/ml of streptomycin and 100 units/ml of penicillin) at 37°C in 5% CO₂. Colo 201, DLD-1, HCT-116, SW48, SW480, SW620, WiDr (Dainippon Pharmaceutical Co., Ltd, Osaka, Japan), CACO-2, Colo 320, and CW-2 (Riken Cell Bank, Tsukuba, Japan) were purchased and cultured according to the manufacturer’s instructions.

Analysis of EP receptor expression in colon cancers by RT-PCR

Total RNA was extracted from tissues and cultured cells by direct homogenisation in Isogen (Nippon Gene Co., Tokyo, Japan), and spectrophotometry was used for quantification. Aliquots (3 μg) of total RNA were subjected to the RT reaction with oligo-dT primer using an Omniscript Reverse Transcriptase kit (Qiagen, Hilden, Germany). After reverse transcription, PCR was carried out with Hotstartaq (Qiagen), according to the manufacturer’s instructions. To test cDNA integrity, the β-actin gene was amplified for each sample. Primers were designed using the computer program OLIGO 4.0-s (National Biosciences, Maryland, USA) and were based on published sequences in Genbank. Primers were designed to cross an exon-exon boundary or insertion of intron to ensure that genomic DNA was not being amplified. BLAST searches confirmed that the primers were specific for the target gene. Primers for the β-actin and EP receptor genes are listed in table 1. PCR amplifications were performed in a thermocycler (Gene Amp PCR System 9600; Perkin-Elmer Applied Biosystems, Foster City, California, USA), with 18–40 cycles of 94°C for 20 seconds, 60°C for 30 seconds, and 72°C for one min using the specific primer sets. PCR products were then analysed by electrophoresis on 2% agarose gel.

Quantitative real time RT-PCR analysis

Quantitative real time RT-PCR analysis was performed using the Smart Cycler system with the Ex Taq R-PCR version 2 kit and SYBR Green (Takara Shuzo Co., Shiga, Japan) according to the manufacturer’s instructions. Primers for the β-actin and EP₃ genes, and cycle conditions for PCR, are listed in table 2. To assess the specificity of each primer set, amplicons generated from the PCR reaction were analysed by their melting point curves and additionally run on 2% agarose gels to confirm the correct sizes of the PCR products. Each PCR product was subcloned into the TA cloning plasmid vector pGEN-T easy vector (Promega Co., Madison, Wisconsin, USA) and used as a positive control for real time PCR analyses. The number of molecules of specific gene products in each sample was determined using a standard curve generated by amplification of 10²–10⁸ copies of the control plasmid. Each sample was analysed in triplicate.

Immunohistochemical staining

Immunohistochemical analyses of colon tumours and normal mucosa samples from F344 male rats treated with AOM were performed with the avidin-biotin complex immunoperoxidase technique, as previously reported.²⁰ As the primary antibody, a polyclonal rabbit anti-EP₃ antibody raised against rat EP₃ receptors was used at a 50× dilution.²² As the secondary antibody, biotinylated antirabbit IgG (H+L) raised in a goat, affinity purified, and absorbed with rat serum (Vector Laboratories, Inc., Burlingame, California, USA) was used at a 200× dilution. Staining was performed using avidin-biotin reagents (Vectastain ABC reagents; Vector Laboratories.), 3,3′-diaminobenzidine, and hydrogen peroxide. Sections were counterstained with haematoxylin. As a negative control, the primary antibody was preincubated with a 16-fold (molar ratio) excess amount of the fusion protein used as the immunogen for one hour at room temperature prior to incubation of the sections.²²

AOM induced colon tumour development in EP₃ receptor knockout mice

Male EP₃ receptor deficient homozygous mice (EP₃^−/−) and wild-type mice received AOM at a dose of 10 mg/kg body weight intraperitoneally once a week for six weeks. At 56 weeks of age, mice were sacrificed under ether euthanasia and complete autopsy was performed. After laparotomy, the entire intestines were resected and opened longitudinally, and the contents were flushed with normal saline. Using a dissection microscope, colon tumours were noted grossly for their location, number, and diameter, measured with callipers. All tumours from AOM treated mice were subjected to histological examination after routine processing and haematoxylin and eosin staining. The experimental protocol was according to the guidelines for Animal Experiments in the National Cancer Center.

Effects of ONO-AE-248 on growth of colon cancer cells

The EP₃ receptor selective agonist 16-(3-methoxymethyl)phenyl-ω-tetranor-3,7-dithiaPGE₁ (ONO-AE-248) was chemically synthesised at Ono Pharmaceutical Co. Ltd.²³ DLD-1 and HCA-7 cells were seeded in plastic 96 well plates at a density of 2×10³ cells per well, and grown for 24 hours with media containing 5% FBS. The EP₃ receptor selective agonist ONO-AE-248 was added daily on days 0–4, and then numbers of viable cells on day 1, 3, and 5 were measured by colorimetric assay using the cell proliferation assay reagent WST-1 (Wako Chemicals, Osaka, Japan) with a microplate reader (Bio Rad, Hercules, California, USA) at a reference wavelength of 655 nm and a test wavelength of 450 nm. Cell viability was determined as per cent of control values. Experiments were repeated three times and data were measured six times (n = 6).

5′-Aza-2′-deoxycytidine treatment

CACO-2, CW-2, DLD-1, HCA-7, and WiDr cells were seeded at a density of 5×10⁴ cells/10 cm dish on day 0 and treated with 1 and 2 μM 5-aza-dC (Sigma, St Louis, Missouri, USA) on days 1, 3, and 5. After each treatment, cells were placed in fresh media and harvested on day 6, and total cellular RNA was prepared using Isogen on day 7.

Statistical analysis

The significance of differences in the incidences of tumours was analysed using the χ² test and other differences using the Student’s t test. Differences were considered statistically significant at p<0.05.

RESULTS

Different expression of PGE₂ receptors EP₁, EP₂, EP₃, and EP₄ in normal colon mucosa and colon tumours

Expression of PGE₂ receptors EP₁, EP₂, EP₃, and EP₄ in normal colon mucosa and colon tumours of AOM treated mice and rats, and in human tissues, were examined by RT-PCR (figs 1, 2). In the three mouse colon adenocarcinomas tested, expression of EP₁ and EP₂ receptor mRNAs was increased compared with levels in normal mucosa. EP₄ mRNA was equally expressed in carcinomas and normal mucosa. In contrast, expression of EP₃ mRNA was markedly decreased in all carcinoma samples compared with normal colon mucosa (fig 1A). Expression patterns of EP₁, EP₂, EP₃, and EP₄ receptors in eight pairs of samples of adenocarcinoma and normal mucosa from AOM treated rats were similar to those in mice. Patterns for EP₁, EP₂, EP₃, and EP₄ receptors in four typical pairs of samples are shown in fig 1B. In the case of human colon tissues, EP₃ receptor mRNA was markedly decreased in seven of eight samples for adenocarcinomas compared with adjacent normal mucosa of the colon. Expression levels of EP₂ receptor mRNA were increased in seven of eight human colon adenocarcinomas compared with levels in normal mucosa, but expression of EP₁ receptor was not clearly increased in human colon carcinoma. EP₄ mRNA was equally expressed in carcinomas and normal mucosa in all cases. Figure 2A shows expression of EP₁, EP₂, EP₃, and EP₄ receptors of colon carcinoma and normal mucosa in four typical pairs of samples.

Furthermore, downregulation of EP₃ was confirmed by quantitative real time RT-PCR (figs 1C, 1D, 2B, 2C). Expression of EP₃ receptor mRNA was significantly downregulated in tumours, being 5% in mice (fig 1C), 9% in rats (fig 1D), and 28% in humans (fig 2C) of the average value of that in the respective normal colon mucosa.

Localisation of EP₃ receptor protein in rat colon tumours

Immunohistochemical analysis of paraffin embedded specimens of eight colon tumours and normal colon mucosa in rats treated with AOM was performed. Slight background staining was widely detected in both negative controls, those stained without antirat EP₃ receptor antibody (fig 3A, B) and those stained with anti-EP₃ receptor antibody preabsorbed with fusion EP₃ receptor protein (fig 3C, D). Moreover, slight non-specific staining was detected in red blood cells. In normal colon mucosa tissues, EP₃ receptor expression was prominent in epithelial cells (fig 3E), and the muscular coat was also positively stained. Similarly, positive staining of EP₃ receptors was observed in hyperplastic ACF of the colon (data not shown). In contrast, staining was very faint, minimal, or absent in epithelial cells of colon adenocarcinomas (fig 3F), being totally lacking in seven cases, sized 3–9 mm in diameter. Only one carcinoma sample was weakly stained, and its size was 2 mm.

Colon tumour development in EP₃ receptor knockout mice

To assess the role of EP₃ receptors in colon tumour development, EP₃ receptor knockout mice were used in an in vivo model. Data for the incidence (percentage of mice with tumours) and multiplicity (number of tumours per mouse) of colon tumours induced by AOM are summarised in table 3. Tumour incidence was increased to 78% in EP₃ receptor knockout mice compared with 57% in wild-type mice. Regarding tumour multiplicity, values were 2.17 (0.51) for EP₃ receptor knockout mice and 0.75 (0.15) for wild-type mice (p<0.05). Histopathological examination revealed 20 colon tumours to be adenocarcinomas in wild-type, and 50 colon tumours to be three adenomas and 47 adenocarcinomas in EP₃ receptor knockout mice. Figure 4 shows the size distribution, demonstrating a significant increase in tumours measuring ⩾2.0 mm in diameter in EP₃ receptor knockout mice (2.00 (0.48) v 0.50 (0.11); p<0.01) but not in those measuring <2.0 mm in diameter (0.17 (0.08) v 0.25 (0.11)).

Expressions of PGE₂ receptors in colon cancer cell lines, and effects of the EP₃ selective agonist on growth of colon cancer cells

Expression of PGE₂ receptors in 11 human colon cell lines was examined by RT-PCR. EP₁, EP₂, and EP₄ were widely detected in the human colon cancer cell lines (in 10 of 11 for EP₁, nine of 11 for EP₂, and nine of 11 for EP₄) but EP₃ was only detected in HCA-7 (fig 5A).

To evaluate the physiological functions of the EP₃ receptor, the effect of an EP₃ receptor selective agonist ONO-AE-248 on viable cell numbers of DLD-1 and HCA-7 in monolayer cultures was examined. In the HCA-7 human colon adenocarcinoma cell line, expression of the EP₃ receptor and other PGE₂ receptors (EP₁, EP₂, and EP₄) were detected by RT-PCR analysis (fig 5A). As shown in fig 5B, HCA-7 cell numbers were significantly decreased dose dependently by addition of ONO-AE-248, with 8%, 17%, and 30% decreases (p<0.05, p<0.01, and p<0.01) in the presence of 1, 3, and 5 μM ONO-AE-248 on day 5, respectively. On the other hand, treatment with ONO-AE-248 did not affect growth of DLD-1 cells which were not expressing EP₃ mRNA. The experiments were repeated three times and similar results were obtained.

Effect of 5-aza-dC on EP₃ expression

To determine whether silencing by DNA methylation could be involved in reduced expression of EP₃ receptor in colon tumours, we tested the effects of 5-aza-dC, a demethylating agent, on EP₃ receptor expression in colon cancer cell lines. Human colon cancer cell lines CACO-2, CW-2, DLD-1, HCA-7, and WiDr were treated with 5-aza-dC, and expression levels of EP₃ receptor were analysed by RT-PCR. Without 5-aza-dC treatment, expression of EP₃ receptor was detected in HCA-7, but not in CACO-2, CW-2, DLD-1, or WiDr (fig 5A). After 5-aza-dC treatment, expression was restored in CACO-2, CW-2, and DLD-1, but not in WiDr (fig 6).

DISCUSSION

In the present study, examination of mRNA expression levels for EP₁, EP₂, EP₃, and EP₄ receptors in colon tissues in mice, rats, and humans by RT-PCR and quantitative RT-PCR provided evidence of a marked reduction in EP₃ receptors in colon cancers, in clear contrast with the increase observed for EP₁ and EP₂. Additionally, results of mRNA expression of EP receptors in 11 human colon cancer cell lines support the above findings and further indicate the events may occur in colon cancer cells. Recently, we reported enhancement of AOM induced colon tumours with exogenous administration of PGE₂ in male F344 rats, and that colon tumours exhibited similar expression patterns in EP receptors as those observed in the present study.²⁴ Sonoshita et al reported that mRNA expression of EP₂ was strongly increased and EP₃ was weakly decreased in colon polyps compared with normal colon in APC^Δ716mice.¹⁹ These reports support our data that downregulation of EP₃ is a common feature in colon cancer of mice, rats, and humans. It has been reported that expression of the EP₃ receptor is widely distributed throughout the body, and its mRNA has been identified in almost all tissues in mice and rats, as well as in humans.^25–27 Northern blot analysis revealed that expression of EP₃ receptor mRNA was mainly localised in the muscle layer in the rat gastrointestinal tract,²⁷ and the present immunohistochemical analysis indicated that EP₃ receptors were detectable in rat normal colon epithelial cells and the muscular coat, but not in rat colon adenocarcinomas. In our previous study, we demonstrated that deficiency of EP₁ or EP₄ receptor reduced formation of AOM induced ACF while EP₃ receptors had no effect, using eight types of EP receptor knockout mice.^17,18 However, long term in vivo examination of AOM induced colon tumour development using EP₃ receptor knockout mice, conducted here in the present study, demonstrated enhancement of tumour incidence and multiplicity. Moreover, the size of the tumours was significantly increased. Thus based on our present and previous results, we suggest that the EP₃ receptor does not influence the early stage of colon carcinogenesis, including ACF formation, but its downregulation could be important to cancer development at a later stage.

In our present study, PCR primers of mouse, rat, and human EP₃ receptors targeted a common sequence in each species. PCR products would be expected to be derived from the entire range of splice variants (figs 1A–B, 2A, 5A). It is noteworthy that there are three splice variants of the EP₃ receptor in mice and rats, and nine in humans, coupled to different G protein signalling pathways.^28–33 These variants are different in the carboxy terminal tail, and the amino acid sequence has an important role in G protein coupling specificity.^30,31 Two of the three variants of the mouse EP₃ receptors are EP_3α and EP_3β, which are coupled to G_i and cause inhibition of adenylate cyclase.³⁰ The mouse EP_3γ receptor, in contrast, is coupled to G_s, in addition to G_i, and evokes pertussis toxin insensitive cAMP production.³¹ Preliminarily, we examined expression of three splice variants of mouse EP₃ receptors by RT-PCR using specific primers for each variant, and found EP_3α to be the major form in mouse normal mucosa (data not shown). These observations support the conclusion that the major splice variants of EP₃ receptors are coupled to G_i and act to inhibit adenylate cyclase in normal colon mucosa in mice. On the other hand, EP₂ and EP₄ receptors are coupled to G_s and stimulate cAMP production by this enzyme. Increased cAMP levels result in activation of cAMP dependent protein kinase (PKA) and transcriptional factors that bind to cAMP responsive elements to transactivate the transcription of specific primary response genes that initiate cell proliferation.³⁴ In our previous study,¹⁸ the EP₄ receptor selective agonist ONO-AE1-329 was shown to enhance colony formation by the HCA-7 human colon adenocarcinoma cell line. The EP₃ receptor selective agonist ONO-AE-248 was demonstrated to suppress cell growth in HCA-7 in the present study. It has been reported that ONO-AE-248 attenuates the rise in intracellular cAMP induced by forskolin, an activator of adenylate cyclase, in CHO cells transfected with EP_3α receptor.²³ Therefore, the EP₃ receptor pathway may play an important role in counteracting the effects of EP₂ and EP₄ receptors, and its downregulation in later stages of colon carcinogenesis may enhance cancer development. Additional studies are needed to investigate interactions between the EP₃ receptor signalling pathway and others linked to EP receptors.

Hypermethylation of CpG islands in promoter regions is known to cause silencing of genes in various human cancers,^35,36 and silencing of COX-2 and APC genes by hypermethylation has been reported in human colon cancer.^37,38 Although hypermethylation of the prostaglandin receptor gene has not been reported,^37,38 DNA sequences in the promoter region and exon 1 of the human EP₃ gene are GC rich (Genbank AL031429). Therefore, in the present study, we examined the effects of demethylation of DNA with 5-aza-dC on EP₃ expression in human colon cancer cell lines. Demethylation of five cell lines by 5-aza-dC treatment resulted in restoration of EP₃ receptor expression in three cell lines. These findings suggest that the DNA sequence of the EP₃ receptor may be methylated but further studies are needed to clarify whether hypermethylation of the EP₃ receptor gene occurs and regulates EP₃ expression in colon cancers.

In conclusion, data obtained in our present and previous studies suggest that the PGE₂ receptor subtype EP₃ plays an important role in suppression of cell growth and that its downregulation enhances colon carcinogenesis at a later stage. The underlying mechanisms clearly warrant further investigation.