Expression of oestrogen receptors, ERα, ERβ, and ERβ variants, in endometrial cancers and evidence that prostaglandin F may play a role in regulating expression of ERα

Endometrial cancer is the most common gynaecological malignancy and accounts for 5% of cancers in women http://​info.​cancerresearchuk​.​org/​cancerstats/​. The majority of endometrial cancers occur in post-menopausal women and 80% of patients are diagnosed when the tumour is confined to the uterus (stage 1 disease). Many of the established risk factors for developing endometrial cancer are associated with excess exposure to oestrogen unopposed by progesterone. For example, several studies have reported that use of oestrogen-only hormone replacement therapy (HRT) increases the risk of developing both localized and widespread endometrial cancer [1, 2]. The menopausal transition (perimenopause), a time when oestrogens may be elevated and anovulatory cycles mean that progesterone levels are reduced, has been proposed as a possible 'window of risk' for the development of the disease [3]. A high body mass index (BMI) [4, 5] increases the risk of developing endometrial cancer and patients with a high BMI have a poorer prognosis [6]. Expression of enzymes involved in biosynthesis of oestrogens such as CYP19A1 and 17β HSD type 2 have been documented in endometrial carcinomas [7, 8] and concentrations of oestradiol (E2) in tumour tissues have been correlated positively with the clinical stage of disease and rate of tumour invasion in both pre- and post-menopausal women [9].

The impact of oestrogenic ligands on endometrial cells is mediated via oestrogen receptors that act as ligand-activated transcription factors. There are two oestrogen receptors, ERα [ESR1] and ERβ [ESR2], encoded by different genes. The human ERβ gene is alternatively spliced at its 3' end resulting in formation of mRNAs that encode not only a full-length protein (ERβ1) capable of binding to E2 but also truncated isoforms (ERβ2, ERβ5) lacking an intact binding pocket [10]. Expression of ERs in normal pre-menopausal endometrium has been well documented with immunoexpression of ERα being intense in both glands and stroma during the proliferative, oestrogen-dominant phase but reduced in the secretory phase following the post ovulatory rise in progesterone [11]. ERβ1 and ERβ2 are both expressed during the proliferative phase however following ovulation ERβ1 continues to be expressed, ERβ2 is selectively down-regulated in the glandular epithelium [12] and the pattern of expression of ERβ5 has not been described.

In normal endometrium expression of progesterone receptor (PR) is induced during the oestrogen-dominated proliferative phase and a number of response elements capable of activation by ERs have been described within the regulatory region of the PR gene [13]. During the secretory phase when circulating concentrations of progesterone are maximal activation of PR results in reduced proliferation and increased cellular differentiation. If progesterone biosynthesis is inadequate/absent as might occur during anovulatory cycles the endometrium can become hyperplastic. Notably, development of complex atypical hyperplasia carries a 25% risk of developing subsequent endometrial adenocarcinoma. Biochemical studies record lower concentrations of ER and PR in endometrial cancers from clinical stages III-IV than those from clinical stage I; in stage I samples higher concentrations of receptor were measured in the well and moderately differentiated samples [14]. In endometrial carcinomas mRNAs for several ERβ isoforms have been detected [15‐17] but detailed immunolocalisation studies comparing their expression have not been reported. It has been claimed that PR immunohistochemistry provides the most reliable means for predicting survival in endometrial adenocarcinoma [18], that detection of PR is associated with better disease free survival [19] and that administration of progestins is an effective treatment for pre-menopausal women with endometrial carcinomas or atypical hyperplasia [19].

In the reproductive tract, the predominant prostaglandins are the E- and F-series prostanoids [20]. These are synthesised from arachidonic acid by cyclooxygenase (COX) and prostaglandin synthase enzymes and act in an autocrine or paracrine manner by binding to specific G-protein coupled receptors (GPCR; reviewed in [21]). There is emerging evidence supporting a complex interplay between the production and action of oestrogens and prostaglandins within the microenvironment of tumours and endometrial pathologies such as endometriosis. For example, E2 can increase expression of COX enzymes [22, 23] and the existence of an oestrogen response element has been documented in the promoter of the gene encoding prostaglandin synthase enzymes [24]. There is convincing evidence that PGE2 stimulates biosynthesis of oestrogens by enhancing expression of the aromatase (CYP19A1) gene in endometriotic tissue [25] and expression of aromatase can be suppressed by COX-2 selective inhibitors [26].

In endometrial adenocarcinoma, expression of COX-2 but not COX-1 is upregulated compared with normal endometrium [27, 28]. Moreover, we have demonstrated a role for the F Prostanoid (FP) receptor (the receptor for prostaglandin PGF2α) in endometrial adenocarcinoma, with evidence that elevated PGF2α-FP receptor signalling results in an up regulation in expression of angiogenic and tumorigenic genes including COX-2 [29], FGF2 [30] and VEGF [31] as well as an increase in proliferation and migration of neoplastic epithelial cells [32]. In the present study we investigated whether expression of ERs, including ERβ variants, could be correlated with the degree of differentiation of grade 1 tumours and/or expression of PR and COX-2. We also investigated the impact of PGF2α on expression of ERα, ERβ and PR in cancer-derived endometrial epithelial cells.

A recent paper reported that women with variants of the aromatase (CYP19A1) gene that are associated with a 10-20% increase in circulating oestrogen levels after menopause have an increased risk of endometrial cancer [42], In the present study we have examined the patterns of expression of ERα, the full length ERβ receptor (ERβ1) and two ERβ splice variant isoforms (ERβ2, ERβ5) in well-characterised stage I endometrioid adenocarcinomas. This extends a preliminary study that discovered ERβ2 and ERβ5 mRNAs were more abundant than those of ERβ4 in human endometrium and Ishikawa cells [43].

In a fixed tissue set comprising 30 well characterised cancers (well, moderately and poorly differentiated) expression of ERα was reduced in the poorly differentiated tissues a finding that is in agreement with previous reports [14, 44]. Although studies in rodents have demonstrated that ERα-dependent gene activation plays a key role in endometrial epithelial cell proliferation [45] in our samples proliferative activity of endometrial adenomas (as determined by immunostaining for Ki67 or histone H3, unpublished observations) was highest in the poorly differentiated tumours even when they were ERα-negative (not shown). These results agree with a recent study documenting increased expression of Ki67 and other cell cycle regulators such as cyclin A during the progression from a normal to malignant endometrial phenotype [46] and higher expression of Ki67 in ERα-negative tumours with a more aggressive phenotype [47].

To date studies on the role(s) played by ERβ in disease progression, cell survival and proliferation have been dominated by studies on breast cancer tissues and breast cancer cell lines. In these samples over-expression of ERβ results in anti-proliferative and pro-apoptotic effects [48] and expression of ERβ2 correlates with favourable response to endocrine therapy and improved survival [49]. Other studies have reported no correlation between expression of ERβ2 mRNA and response to tamoxifen [50, 51]. A recent study used tissue microarrays to determine expression of ERβ1, β2 and β5 in a series of 880 cases of primary invasive breast carcinomas from patients with long term follow up. Expression of ERβ2 or ERβ5, but not ERβ1 significantly correlated with overall survival [39]. To date only two studies have examined expression of ERβ in endometrial cancers. In both studies samples were ERα-positive; one group reported detection of ERβ5 mRNA [16] the other reported finding no correlation between ERβ mRNA expression and PR labeling index, cell proliferation or histologic grade [15]. We believe this is the first paper demonstrating immunoexpression of ERβ5 protein in cell nuclei within stage 1 endometrial adenocarcinomas regardless of whether they were well or poorly differentiated. Expression of ERβ5 is not unique to tumour cells and we have immunolocalised the protein to multiple cell types in normal cycling endometrium, first trimester decidua and placenta (Fitzgerald, MacPherson and Saunders, unpublished observations). Molecular modelling of the ERβ5 protein suggests that it does not contain a functional ligand-binding pocket [10]. ERβ5 has been demonstrated to form a hetero-dimeric complex with ERα which negatively regulated transcriptional activity [52]: this may explain why ERβ5 expression was associated with a better prognosis in breast cancer [53]. Leung et al [10] detected increased activation of an ERE-luciferase reporter in HEK293 cells incubated with oestrogens including E2 when cells were co-transfected with ERβ1 and ERβ5 compared with those transfected with ERβ1 alone.

In the current study expression of PR in endometrial adenocarcinoma tissues broadly paralleled that of ERα with minimal expression of PR in the poorly differentiated cancers even though these tissues maintained expression of ERβ. In our ERα^pos/ERβ^pos Ishikawa (A) cells expression of PR mRNA and a luciferase gene driven by a consensus 3xERE promoter were both induced by E2 treatment. No activity was detected in the ERα^neg/ERβ^pos Ishikawa cells (line B) even though they were able to activate the ERE-luciferase when ERα was reintroduced into the cells suggesting the lack of response was not due to lack of transcriptional competence; both cell lines expressed similar concentrations of ERβ5 mRNA. Our results are in agreement with those of others [54] who reported that ERβ was unable to up-regulate expression of the PRB promoter in HeLa, BT-20 or Ishikawa cells although in SK-BR-3 cells both receptors were able to repress promoter activity. The potential that ERβ-dependent gene activation can occur in the endometrial cancers is supported by the results of studies using tamoxifen, a SERM that acts as a potent transcriptional activator of ERβ at AP-1 response elements [55]. Treatment with tamoxifen results in a more aggressive endometrial cancer phenotype and development of a distinctive 'tamoxifen-specific' gene profile [56, 57].

Expression of COX-2 but not COX-1 is up-regulated in endometrial adenocarcinoma compared with expression levels observed in normal endometrium [27, 28]. This is associated with increased biosynthesis of prostaglandins and increased expression of FP receptors resulting in a stimulation of FP-receptor dependent signalling and production of angiogenic factors. [29]. In addition there is evidence that PGE2 can up-regulate expression of steroidogenic genes including CYP19A1 and thereby contribute to increased local concentrations of oestrogenic ligands that could bind ERα and/or ERβ [58]. We believe the data in the present paper provide preliminary evidence for a link between signalling via the FP receptor and an apparent reduction in expression of ERα and PR. The human ERα gene is transcribed from at least seven promoters into multiple transcripts that vary in their 5' UTRs. Tissue specific expression of transcripts has been documented as having differential use of promoters in normal and cancerous breast tissue (reviewed in [59]). The signalling pathway responsible for down regulation in the amount of ERα mRNA after incubation of endometrial Ishikawa cells with prostaglandin F2α requires further investigation in order to determine whether the effects we observed are mediated by transcriptional or post transcriptional mechanism(s).

Our results shed new light on the interplay between PG and ER-dependent patterns of gene expression in endometrial cancers. First we would speculate that ligand-dependent or ligand-independent activation of ERβ isoforms could have an impact on progression of endometrial cancers especially those with a more aggressive phenotype that are ERα-negative and this merits further investigation. Second, although increased biosynthesis of prostaglandins is known to occur in endometrial cancers we believe our study provides the first evidence that down-regulation in expression of ERα, and the consequent reduction in expression of PR, may be one of the downstream consequences of F prostaglandin-dependent signalling.