Heightened Expression of Cyclooxygenase-2 and Peroxisome Proliferator-Activated Receptor-δ in Human Endometrial Adenocarcinoma

Abstract

Epidemiological studies indicate that nonsteroidal anti-inflammatory drugs (NSAIDs) significantly reduce the risk and mortality from colorectal cancer, in part by inhibiting prostaglandin (PG) synthesis. Cyclooxygenase (COX), the rate-limiting enzyme in PG biosynthesis, exists in two isoforms, COX-1 and COX-2. Genetic and pharmacological evidences suggest that COX-2 is involved in the development of colorectal cancer. We have previously shown that COX-2-derived prostacyclin participates in blastocyst implantation through activation of peroxisome proliferator activated receptor δ (PPARδ), a member of the nuclear hormone receptor family. Furthermore, our recent studies suggest that a similar pathway is operative during colorectal carcinogenesis. These observations prompted us to examine whether the COX-2-PPARδ signaling pathway is also involved during development of uterine adenocarcinoma. Here we describe for the first time the heightened expression of COX-2 and PPARδ, but not COX-1, in uterine endometrial adenocarcinoma.

Introduction

Cancer of the uterine corpus is the most common gynecologic malignancy in the United States making it the fourth most common cancer in women [1,2]. The etiology of adenocarcinomas arising from the uterine endometrium is not completely understood, although estrogen therapy unopposed by progesterone has been shown to increase the incidence of endometrial cancer [3–6]. Women with advanced or recurrent endometrial cancer generally die with progressive disease, accounting for 6000 cancer deaths annually [1]. If we could identify effective targets for prevention of this disease, it will have a major impact on women's health.

Cyclooxygenase (COX) is the rate-limiting enzyme in the conversion of arachidonic acid to prostanoids [7]. Two isoforms of the COX gene have been identified, COX-1 and COX-2 [8]. COX-1 is constitutively expressed and is involved in the cytoprotection of the stomach, renal vasodilatation, and platelet aggregation. COX-2 is an inducible enzyme stimulated by cytokines, growth factors, and tumor promoters, and is involved in inflammation, cell proliferation, and differentiation [9,10]. Mouse and human studies have shown the importance of COX in the female reproductive systems. In the mouse uterus, COX-2 is expressed in the luminal epithelium and underlying stroma surrounding the blastocyst at the time of embryo implantation, suggesting its role in increasing vascular permeability and angiogenesis [11]. In the human endometrium, COX-2 is expressed during the menstrual cycle and is upregulated by gonadotropins at the time of decidua formation [12].

COX has also been implicated in the promotion of carcinogenesis. One association between COX and carcinogenesis is evident from epidemiological studies showing lower incidence of gastrointestinal and other cancers in individuals chronically consuming nonsteroidal anti-inflammatory drugs (NSAIDs) [13,14]. Subsequent studies have confirmed that NSAIDs cause regression of colorectal adenomatous polyps in individuals with familial adenomatous polyposis [15]. Furthermore, COX-2-deficient mice when bred with APC⁷¹⁶ mice produce offspring with a seven-fold reduction in intestinal polyps [16]. In humans, a correlation of COX-2 expression with the size and stage of colorectal carcinomas has also been reported [17]. Other carcinomas that show increased expression of COX-2 include squamous cell carcinoma of the head and neck, adenocarcinoma of the lung, gastric carcinoma, pancreatic carcinoma, esophageal carcinoma, and hepatocellular carcinoma [18–25]. Although increased expression is correlated with tumor size and invasion in colon carcinoma, tumors such as adenocarcinoma of the lung and hepatocellular carcinoma show increased expression in the well-differentiated tumors or portion of tumors [17,19,25].

COX-1 and COX-2 are expressed in both the endoplasmic reticular membrane and the nuclear envelope, suggesting that prostaglandins (PGs) could function through two different classes of receptors [26]. PGs generated in the endoplasmic reticular membrane can exit the cell and bind to G-protein-coupled cell surface receptors that are linked to intracellular signaling pathways [27]. In contrast, PGs produced with nuclear COX can exert their effects directly within the nucleus through peroxisome proliferator-activated receptors (PPARs), which belong to the nuclear hormone receptor superfamily [28]. The PPAR family consists of PPARα, PPARδ, and PPARγ. These isoforms exhibit different expression patterns and ligand dependency [29–31]. PPARα is highly expressed in the liver and implicated in lipid homeostasis. It is activated by hypolipidemic drugs, fatty acids, and by PGI₂ agonists [32]. Fatty acids and PGI₂ agonists can also activate PPARδ. PPARγ is mainly expressed in white adipose tissues and implicated in adipocyte differentiation. The ligands for PPARγ include antidiabetic thiazolidinediones and a metabolite of PGJ₂ [33]. The activation of PPARγ also terminally differentiates tumor cells suggesting a role in cell cycle regulation [34,35]. PPARs modulate transcription by heterodimerization with retinoic acid X receptors (RXRs). Recent work has shown that COX-2-derived PGI₂ is involved in activating PPARδ during the process of implantation and decidualization [36]. This was the first evidence for a biologic role for PPARδ. Because decidualization is considered a pseudomalignant process, we speculated that PPARδ could be involved in tumorigenesis in the endometrium and other tissues. Indeed, the same signaling pathway has recently been implicated in colorectal cancer [37,38].

Because COX-2 is expressed in the human endometrium, we postulated that the COX-2/PPARδ signaling pathway could be involved in the genesis and progression of uterine adenocarcinoma. The present investigation compared the expression pattern of COX-1, COX-2, and PPARδ in normal human endometrium with those of adenocarcinoma specimens. The results show that COX-2 and PPARδ are expressed at higher levels in endometrial adenocarcinomas.

Materials and Methods

Patient Samples

Samples of 11 endometrial adenocarcinomas and three controls were obtained from surgical pathology specimens. The samples were immediately frozen in liquid nitrogen and stored at -70 C until analyzed. All endometrial tumors were adenocarcinomas. One was a papillary serous adenocarcinoma, the others were of squamous, endometrioid or villoglandular differentiation. The nuclear grade and stage of each tumor along with brief clinical histories of each patient were recorded.

RNA Isolation and Northern Blot Analysis

Total RNA was extracted from the tissue specimens by a modified guanidine thiocyanate procedure [39]. Total RNA (20 µg) was denatured, separated by formaldehyde-agarose gel electrophoresis, and transferred to nylon membranes. RNA was cross-linked to the membranes by UV irradiation (Spectrolinker, XL-1500: Spectronics, Westbury, NY) and the blots were prehybridized, hybridized, and washed as described previously [40]. For Northern hybridization, antisense ³²P-labeled cRNA probes for mouse COX-1, human COX-2, and human β-actin were generated. After hybridization, the blots were washed under stringent conditions and the hybrids detected by autoradiography [41]. The stripping of the hybridized probe for subsequent rehybridization was achieved as previously described [40]. Each blot was hybridized with the COX-1, COX-2, and β-actin probes, sequentially.

In Situ Hybridization

In situ hybridization followed the protocol previously described [41]. Tissue specimens were obtained immediately after surgery and pieces of tissues obtained from the pathologist were flash frozen in liquid Histo-freeze (Fisher). Frozen sections (10 µM) from each tissue specimen were mounted onto poly-l-lysine-coated slides and stored desiccated at -80 C until used. Sections were brought to room temperature, fixed in cold 4% paraformaldehyde solution in PBS, acetylated, and hybridized at 45°C for 4 hours in 50% formamide buffer containing ³⁵S-labeled antisense or sense cRNA probes specific to mouse cDNA to COX-1 and human specific cDNAs to COX-2 and PPARδ. After hybridization and washing, the slides were incubated with RNAse A (20 µg/ml) at 37°C for 20 min, and RNAse A-resistant hybrids were detected by autoradiography using Kodak NTB-2 liquid emulsion. Parallel sections hybridized with the sense probes served as negative controls. Slides were poststained with hematoxylin and eosin.

Immunohistochemistry

Immunolocalization of COX-2 and PPARδ were performed in paraformaldehyde-fixed frozen sections using a Zymed-Histostain SP kit (Zymed) as described previously [42]. Rabbit antipeptide antibody to mouse COX-2 was produced using the peptide, NASASHSRLDDINPT, corresponding to amino acids 563–577 of the COX-2 protein, as immunogen. Goat antipeptide antibody to mouse PPARδ was purchased from Santa Cruz Biotechnology (Santa Cruz, CA). These antibodies were found to cross-react with human tissues. The specificity of these antibodies has previously been characterized [11,36]. Red deposits indicate the site of immunoreactive protein.

Results

Analysis of Northern Hybridization

In a representative sample examined, Northern blot hybridization detected COX-2 mRNA in adenocarcinoma tissue, but not in a normal endometrial tissue. In contrast, COX-1 mRNA levels were similar with respect to β-actin mRNA labels in the control and adenocarcinoma tissue (Figure 1). Northern blot hybridization was performed to confirm the specificity of the probes and transcript sizes before proceeding to in situ hybridization experiments. As reported previously [42], a 2.8-kb transcript was detected for COX-1 mRNA whereas a 4.7-kb transcript was detected for COX-2 mRNA. Because uterine tissue is comprised of heterogeneous cell types, the detection of mRNA by Northern hybridization underestimates the expression levels due to the dilution effects and does not provide information on cell-specific gene expression. Therefore, we next examined the cell-specific expression of these genes by in situ hybridization of tissue sections.

Figure 1.

Northern blot hybridization of COX-1 and COX-2 mRNAs in normal uterine tissue (N) and endometrial adenocarcinoma (T). Total RNA (20 µg) was separated by agarose gel electrophoresis and hybridized to cRNA probes specific for COX-1, COX-2, and β-actin.

Analysis of In Situ Hybridization

In situ hybridization revealed accumulation of COX-2 mRNA within the neoplastic cells of 8 of 11 adenocarcinomas. Representative photomicrographs of in situ hybridization of normal (control) and selected adenocarcinomas are shown in Figure 2. Of the three adenocarcinomas in which COX-2 mRNA was undetectable, one patient had been on prolonged progesterone therapy for regulation of her menstrual periods, one was on tamoxifen therapy for the treatment of breast cancer, and the third patient had a papillary serous adenocarcinoma, which is a distinct subtype of adenocarcinoma, discussed below. Interestingly, although COX-2 mRNA was not expressed in the neoplastic tissue of the papillary serous adenocarcinoma, it was present in blood vessel endothelium within the tissue (Figure 2).

Figure 2.

In situ hybridization of COX-2 mRNA in normal tissue (N) and uterine endometrial adenocarcinomas (T). The middle panel shows COX-2 mRNA in a representative section of endometrial endometrioid adenocarcinoma, whereas the bottom panel is a papillary serous adenocarcinoma showing a unique distribution of COX-2 mRNA localized in blood vessels. Photomicrographs in the top, middle, and bottom panels are shown at x 100, x 40, and x 200, respectively, ge, glandular epithelium; s, stroma; tu, tumor; bv, blood vessels.

In situ hybridization detected COX-1 mRNA accumulation in three of seven of the adenocarcinoma tissues (Figure 3). On comparison of the localization of COX-1 and COX-2 mRNAs on sections of the same tumor tissue, COX-1 mRNA appears to be expressed in areas distinct from those expressing COX-2 mRNA.

Figure 3.

In situ hybridization of COX-1 mRNA in normal tissue (N) and uterine endometrial adenocarcinomas (T). Brightfield and darkfield photomicrographs showing COX-1 mRNA distribution at x 100.

The accumulation of PPARδ mRNA was noted in localized areas in 8 of 10 adenocarcinoma samples examined (Figure 4). However, the expression was much more widespread compared with COX-2 expression, which was focal in nature (compare Figure 2 vs. Figure 4). Using serial sections, we observed that many cells within tumor tissues showed coordinate expression of both COX-2 and PPARδ; in others the expression was disparate (see middle panels of Figures 2 and 4). These results suggest an autocrine and/or paracrine interaction between COX-2-derived PGs and PPARδ in tumor tissues. In situ hybridization did not show accumulation of COX-1, COX-2 or PPARδ mRNA in normal tissues. Sections hybridized with the sense cRNA probes were negative in all tissues (data not shown).

Figure 4.

In situ hybridization of PPARδ mRNA in normal tissue (N) and uterine endometrial adenocarcinomas (T). The middle and bottom panels show PPARδ mRNA in representative sections of endometrial endometrioid adenocarcinoma. Compare PPARδ mRNA distribution in the middle panel with that of COX-2 mRNA expression in a serial section in the middle panel of Figure 2. Brightfield and darkfield photomicrographs are shown at x 100.

Analysis of Immunohistochemistry

Immunohistochemistry was performed to examine whether the changes in mRNA levels reflected changes in protein accumulation. The tissues with elevated COX-2 and PPARδ mRNA levels were also positive for immunoreactive COX-2 and PPARδ proteins. As shown by representative photomicrographs in Figure 5, COX-2 and PPARδ proteins were highly expressed in tumor samples compared with controls. As previously described [36], PPARδ was localized in the nucleus, whereas COX-2 was localized in both the perinuclear membrane and cytoplasm. These results suggest that COX-2 and PPARδ mRNAs expressed in tumor cells are also translated into their respective proteins. Additionally, in normal tissue, immunoreactive PPARδ was localized in the surface epithelium. This is similar to what has been observed for normal colonic epithelium [37].

Figure 5.

Immunohistochemical staining of COX-2 and PPARδ in normal (N) and endometrial adenocarcinoma (T) tissues. Red deposits indicate positive immunostaining. Representative photomicrographs are shown at x 100. Note both perinuclear and cytoplasmic localization of COX-2 and nuclear localization of PPARδ.

Discussion

The present investigation describes for the first time the expression of COX-2 and PPARδ in endometrial adenocarcinoma. It is interesting to note that the accumulation of COX-2 mRNA was observed within neoplastic cells of endometrioid adenocarcinomas, whereas the expression was confined to blood vessels in a papillary serous adenocarcinoma subtype. However, this observation needs to be confirmed using a larger sample size. Similarly, two other endometrioid adenocarcinoma samples from patients undergoing treatment with progesterone or tamoxifen had undetectable levels of COX-2 mRNA by in situ hybridization. One patient had been on long-term medroxyprogesterone therapy to control irregular menstrual cycles, whereas the other was on tamoxifen for treatment of breast cancer. Whether these treatments had any effects on the expression of this gene is currently unknown. However, there is evidence that progesterone and tamoxifen can influence tumor progression in endometrial adenocarcinoma [3,43,44]. COX-2 has been shown to alter the metastatic potential of colorectal cancer cells, increasing their invasive properties by six-fold [45]. These findings suggest a role for COX-2 in tumor progression and raise questions for further research in this area regarding hormone responsiveness of COX-2 in endometrial cancers. The expression of COX-2 in tumor cells is specific, because this isoform is not expressed in the stroma surrounding the tumor or in normal endometrial tissues.

Papillary serous tumors of the uterine endometrium are a distinct subtype of endometrial adenocarcinoma that is highly aggressive, generally with extensive extrauterine metastasis and poor survival [46]. Although only one sample was analyzed, the absence of COX-2 mRNA in papillary serous adenocarcinoma is an interesting observation, because clinical characteristics of these tumors are distinct from those of other types of adenocarcinomas. In this respect, poorly differentiated areas of lung and hepatocellular carcinoma also failed to show expression of COX-2 [17,18,25]. The endometrial tumors examined were of grades 1 to 3 and stages 1A to 4A. However, our study did not reveal a correlation between grades of endometrioid adenocarcinoma and expression of COX-2. The findings in the papillary serous tumor also raise an interesting question of the role of COX-2 in tumor angiogenesis, because COX-2 was expressed only in blood vessels in this sample. However, the expression of PPARδ was not detected in the blood vessels or in the neoplastic cells of this sample. Further studies with angiogenic markers would be required to confirm this finding using a larger sample size. COX-2 has been shown to modulate production of angiogenic factors by colon cancer cells, indicating a role for COX-2 in tumor-induced angiogenesis [47].

The coexpression of COX-2 and PPARδ in endometrial cancer tissues suggests that COX-2-derived PGs, perhaps prostacyclin, could be involved in tumorigenesis through activation of PPARδ in an autocrine and/or paracrine manner. Indeed, COX-2-derived prostacyclin has been shown to activate PPARδ in colorectal cancer cells and an involvement of PPARδ in colon cancer has recently been implicated [37,38]. Although a role for COX-2-PPARδ signaling in endometrial tumorigenesis is suspected, the cause and effect relationship cannot be established from this type of study. However, regression of tumors with NSAIDs or COX-2 selective inhibitors suggests a strong involvement of COX-2 in tumorigenesis of colorectal cancers [37].

In conclusion, we observed heightened expression of COX-2 and PPARδ in uterine endometrial adenocarcinomas. The role of these genes in endometrial carcinogenesis or tumor activity remains to be elucidated. Further investigation is warranted to evaluate the usefulness of COX-2 as a prognostic indicator in endometrial adenocarcinoma and the possibility that selective COX-2 inhibitors could play a role in therapy for COX-2 positive endometrial carcinomas needs to be considered.