Tissue Selective Estrogen Complexes (TSECs) Differentially Modulate Markers of Proliferation and Differentiation in Endometrial

Abstract

Selective estrogen receptor modulators (SERMs) have tissue-specific estrogen receptor (ER) modulating properties. Combining an SERM with one or more estrogens to form a tissue selective estrogen complex (TSEC) can provide an improved blend of tissue-specific ER agonist and antagonist effects. While both estrogens and SERMs affect the uterine endometrium, not all TSECs reverse the endometrial effects of estrogens preventing endometrial proliferation and hyperplasia. Their action in uterine cells is not completely understood. HOXA 10, leukemia inhibitory factor (LIF), progesterone receptor (PR), and EMX2 are genes known to regulate endometrial proliferation and differentiation. The expression of these genes was used to assess endometrial effects of SERMs and TSECs. We evaluated the effects of raloxifene (RAL), tamoxifen (TAM), lasofoxifene (LAS), bazedoxifene acetate (BZA), and progesterone (P) alone and in combination with estradiol (E2) in Ishikawa cells. Increased HOXA10, LIF, PR, and EMX2 messenger RNA (mRNA) expression was noted in E2-treated cells compared with vehicle-treated controls. All TSECs maintained E2-induced PR expression and all except TAM prevented estrogen-induced LIF expression. The TSEC containing BZA uniquely decreased HOXA10 expression and increased EMX2 expression. The TSECs alter endometrial cell proliferation by selective modulation of estrogen responsive genes, maintaining the antiproliferative effects mediated by PR and inhibiting LIF. The differential effect of TSECs on endometrial gene expression suggests a mechanism by which they manifest differential effects on endometrial safety against the risk of estrogen-induced endometrial hyperplasia.

Introduction

As the population of the United States ages, the number of postmenopausal women is increasing.¹ When estrogen levels decline at menopause, many women experience vasomotor instability and vaginal atrophy; further, they are at increased risk for developing osteoporosis.^2–4

Hormone therapy can address many of the consequences of reduced ovarian sex steroid production, however not all menopausal women are inclined to use hormone therapy due to the widely publicized controversies surrounding its benefits and risks.^5–7

In reality, conjugated estrogens (CEs) have been well studied and have a favorable benefit /risk profile.⁸ Treatment with CE was recently shown to have minimal risks, especially in younger menopausal women.^8,9 However, in patients with a uterus, estrogens must be administered concomitantly with a progestin to prevent endometrial hyperplasia.¹⁰ The results of the Women’s Health Initiative (WHI) demonstrate few risks with the use of CE, especially in younger postmenopausal women,⁸ however some of the benefits were attenuated by the addition of a progestin.¹¹

In the WHI, CE treatment reduced net cardiovascular events in the 50- to 59-year-old age group, however no such beneficial effect was observed with combined conjugated equine estrogens (CEE) and medroxyprogesterone acetate (MPA) therapy.¹¹ Similarly, CEE reduced breast cancer incidence in the WHI, an effect that became statistically significant in the postintervention phase.^{8 ,9} In contrast, a small increase in breast cancer and breast cancer mortality was seen in women using CEE/MPA,¹¹ that did not persist in the postintervention phase.¹² It is likely that the risk of breast cancer associated with hormone therapy is predominantly attributable to the progestin. However, in a woman with a uterus, use of estrogens without progestins can lead to endometrial hyperplasia and therefore estrogens cannot be used alone.

Progestins are used in combination with estrogens solely for endometrial protection.^{10 ,11} Clearly, alternatives to progestins that counteract estrogen effects in the uterus are needed. Tissue selective estrogen complexes (TSECs) are a new class of compounds that pair a selective estrogen receptor modulator (SERM) with 1 or more estrogens in order to attain unique ER actions that are tissue specific.¹³ An ideal TSEC should allow estrogen action where it is beneficial (ie, central nervous system to reduce vasomotor symptoms and bone) while counteracting the effects of estrogens in the uterus, eliminating the risk of endometrial hyperplasia. While 1 TSEC consisting of raloxifene and estradiol (RAL/E2) has been implicated in endometrial hyperplasia, another containing bazodoxifene acetate and conjugated estrogens (BZA/CE) has demonstrated endometrial safety.^{13 ,14}

Classically, estrogens function via binding to 1 of 2 ERs (ERα and ERβ).¹⁵ The molecular basis of estrogen activity involves liganded ER, undergoing conformational changes that facilitate interactions with coactivator or corepressor proteins. These complexes subsequently enhance or suppress transcription of target genes.¹⁶ Tissue response to estrogens is dependent on the individual ligand or combination of ligands, levels of ERα and ERβ expressed as well as the availability of coactivators and/or corepressors.¹⁷ Each ligand or a blend of ligands can therefore induce a distinct set of responses that are also tissue specific.

Selective estrogen receptor modulators comprise a group of compounds that are capable of ER agonist and antagonist effects in a tissue-specific manner.¹³ The activity of SERM is intrinsic to each compound that accomplishes a differential tissue-specific profile by formation of distinct transcriptional complexes with unique combinations of coactivators and corepressors. The cell-type-specific action of this class of compounds is clinically useful; however each SERM’s pharmacology is different and needs to be evaluated individually when considering its use clinically.

Since the development of tamoxifen (TAM), other triphenylethylene SERMs have been studied for breast cancer prevention, including droloxifene, idoxifene, toremifene, and ospemifene.¹⁸ Recently other SERMs have entered clinical development including benzothiophenes (raloxifene and arzoxifene), benzopyrans (ormeloxifene, levormeloxifene, and EM-800), lasofoxifene (LAS), pipendoxifene, bazedoxifene, HMR-3339, and fulvestrant, an antiestrogen which is approved for breast cancer treatment.^{19 ,20} Differences in SERM activity suggest that any clinical end point must be evaluated individually, and conclusions about any particular SERM may vary by tissue. The mechanisms responsible for differential SERM activity are poorly understood but known to be cell type and gene promoter dependent.¹⁸ The effects of TSECs on target tissues are not well characterized and large clinical trials have thus far been conducted on only a single TSEC containing BZA.²¹ This TSEC relieves vasomotor symptoms and improves bone mineral density while avoiding endometrial hyperplasia.^{21 ,22}

An essential prerequisite for the clinical use of a TSEC is the lack of endometrial stimulation and hyperplasia. While administration of estrogens alone leads to endometrial hyperplasia, use of the TSEC containing BZA/CE did not result in hyperplasia in clinical trials.¹³ While all estrogen, SERM, and TSEC functions are mediated through ER, not all result in the same pattern of gene activation.^{16 ,23} We hypothesized that the mechanism of TSEC action varies depending on the individual TSECs ability to differentially activate downstream target genes. Currently there are no studies that identify endometrial target genes preferentially activated or repressed by a TSEC. Some of those target genes regulate pathways essential for endometrial proliferation and therefore have a propensity to promote hyperplasia. The aim of this study was to compare the effect of several TSECs on well-characterized markers of proliferation and differentiation in endometrial cells.

Methods

Results

Preclinical studies have shown that SERMs can antagonize 17-β E2-mediated stimulation of breast cancer cell proliferation.^{16 ,18,26} Recently, BZA was shown to effectively antagonize the E2-stimulated proliferation of MCF-7 cells.²⁷ However, SERMs have not been evaluated for their antagonistic effect on E2-treated endometrial epithelial cells. Therefore, we used Ishikawa cells to study the effect of E2, RAL, TAM, LAS, and BZA alone and in combination with E2 forming a TSEC. Estradiol and each SERM were used at 0.1 μmol/L concentration. While the potency of each SERM may vary with the gene studied, we initially screened for differences in target gene expression when each SERM and E2 were used at equimolar concentrations. The expression of 4 genes that are well-characterized markers of endometrial proliferation and differentiation were used to evaluate the effect of each SERM/TSEC in Ishikawa cells. As shown in Figures 1 to 4, HOXA10, LIF, PR, and EMX2 were each regulated in Ishikawa endometrial cells by E2: P = .03, P = .06, P = .001, and P = .001, respectively. Gene expression in response to E2, SERMs, and TSECs compared to vehicle control is described subsequent paragraphs.

HOXA10 was used as a well-characterized marker of estrogen response in endometrial cells.^28–31 HOXA10 mediates both endometrial cell proliferation and differentiation.^28,32–34 Compared to treatment with vehicle control, E2 treatment significantly increased HOXA10 messenger RNA (mRNA) expression, as previously reported.^30,31,35 HOXA 10 gene expression was not significantly changed by treatment with any of the SERMs in Ishikawa cells (Figure 1A). Estradiol was used in combination with each SERM and the resultant HOXA10 expression regulated by each TSEC was evaluated. While each TSEC appeared to counter the effect of E2 (BZA + E2 > LAS + E2 > TAM + E2> RAL + E2.), BZA was the only SERM that significantly blocked the E2 effect on HOXA10 gene expression. (P = .03; Figure 1B).

Leukemia inhibitory factor was used as a second marker of estrogen effect on endometrial cells. Leukemia inhibitory factor is known to lead to increased proliferation of endometrial cells.^36,37 There were no significant changes in LIF expression induced by RAL, TAM, or LAS. Treatment with BZA led to a statistically significant decrease in LIF expression as compared to control (P = .02; Figure 2A). In response to concomitant administration of E2, all TSECs significantly prevented estrogen-induced LIF expression except for the TSEC that included TAM (Figure 2B).

Estradiol-induced expression of PR is required to mediate the differentiating effects of progesterone.³⁸ Here, as expected, E2 treatment led to increased PR mRNA expression. Both TAM and LAS were also able to induce PR gene expression (P = .0004 and P = .01) respectively, whereas no change in PR expression was seen after RAL or BZA treatment (Figure 3A). All TSECs maintained the E2 effect on PR. A 3- to 15-fold increase in PR was seen in TSEC-treated cells compared with SERM treatment alone: RAL + E2 versus RAL (P = .005); TAM + E2 versus TAM (P = .0002); LAS + E2 versus LAS (P = .0008); BZA + E2 versus BZA (P < .00001; Figure 3B). Addition of an SERM did not lead to antagonism of estrogen action on PR expression.

EMX2 has been demonstrated to inhibit the proliferation of endometrial cells.¹⁷ Expression of EMX2 gene in cells treated with TAM and BZA was significantly increased compared to control (P < .001; Figure 4A). Estradiol-treated cells showed low EMX2 gene expression; its repression by E2 is consistent with the need to repress EMX2-mediated suppression of endometrial cell proliferation (P < .001). The combination BZA + E2 significantly increased EMX2 gene expression in comparison to E2 alone (Figure 4B; P = .03). The TSEC containing RAL did not increase EMX2 to a significant degree, consistent with the risk of endometrial hyperplasia reported in women using this combination of estrogen and SERM.³⁹

Discussion

As a class of compounds, SERMS act on ERs, demonstrating mixed function depending on the particular SERM as well as the target gene and tissue.¹⁸ The SERMs when used alone typically prevent uterine cell proliferation, however one (TAM) may have a stimulatory effect.^27,40 Their ability to counteract the effect of E2 on endometrial cells has not been previously tested. In this study, we have shown that proliferation and differentiation markers can be affected differentially by the combination of E2 and SERMS. The TSECs vary depending on the specific SERM used. Therefore, SERMs that antagonize estrogen effects in the endometrium may be better suited for inclusion in TSECs.

HOXA10 encodes an evolutionarily conserved transcription factor that is essential to endometrial development and endometrial receptivity.^28,29 HOXA10 expression is apparent in endometrial stroma and glands, where it is regulated by sex steroid hormones.^28,30 We have previously shown that HOXA10 expression is significantly upregulated in response to E2 in Ishikawa cells³⁰ as also demonstrated in this study. All TSECs showed some degree of attenuation of this E2-mediated increase in HOXA10 in Ishikawa cells, however the only combination that led to a statistically significant decrease was BZA + E2. The BZA displayed a significant ER antagonist effect on the expression of this gene. While the data reported here were determined in an endometrial cell line, the findings may help to explain the clinical endometrial effects of the BZA/E2 combination in postmenopausal patients.^{13 ,41} A difference in HOXA10 expression alters cell proliferation and differentiation, likely contributing to the differential effect of BZA compared to other SERMs in a TSEC. The CE/BZA does not induce endometrial hyperplasia while other TSECs have been demonstrated to do so.^{13 ,21}

Leukemia inhibitory factor is an essential cytokine in the reproductive tract, regulating epithelial development, proliferation, and gland formation.^42,43 It is expressed in the glandular epithelium of the endometrium where it is essential for implantation in humans,^44,45 The LIF is a mediator of endometrial cell proliferation.^44,46 Tamoxifen was the only SERM that did not block E2 action on LIF expression. The data suggest that TAM is a poor choice for inclusion in a TSEC. Consistent with this conclusion, multiple studies have reported the development of endometrial cancer in TAM-treated breast cancer patients.^{24 ,25} Several investigators prospectively followed patients on TAM and detected a variable incidence of endometrial pathologies. The incidence of hyperplasia was reportedly up to 20%, while carcinoma developed in 0% to 8%,^47–49 showing that TAM has a partial agonistic action on the endometrium.

Progesterone receptor is a target gene for estrogen action⁵⁰ and is of particular physiological importance in the uterus. Progesterone receptor mediates the response to progesterone that occurs during the luteal phase of the ovarian cycle and is essential for endometrial differentiation that leads to endometrial receptivity and embryo implantation. Here, as expected, E2 treatment led to increased PR expression. The addition of an SERM did not change this effect, showing that the potential antiproliferative mechanisms mediated by PR are maintained when TSECs are used.⁵¹ The effect on PR may be an important mediator of differentiation and contribute to the prevention of an estrogenic effect when the appropriate TSEC is used. However, all TSECs maintained PR expression, suggesting that this effect is not helpful in selection of SERMs for inclusion in clinically useful TSECs. Additionally the expression of PR may only become clinically relevant in the presence of a progestin, making this pathway of unlikely relevance in postmenopausal women administered a TSEC.

The EMX2 is expressed in Ishikawa cells and adult uterine endometrium.⁵² The EMX2 suppresses endometrial epithelial proliferation^53,54 and is a candidate tumor suppressor gene based on its location within a consensus region of allelic loss in uterine endometrial adenocarcinoma.⁵⁵ BZA/E2 was the only TSEC that significantly increased EMX2 gene expression when comparison to E2 alone, thus demonstrating an antiproliferative effect of this TSEC. The effects on EMX2 suggest that, among those tested, BZA may be the optimal SERM for use in a TSEC, inducing the antiproliferative and tumor-suppressing effects of EMX2.

The combination of an SERM and an estrogen produced differential effects on endometrial cell gene expression. The specific response depends on the SERM used as well as the target gene studied. This study showed that distinct molecular signals underlie the action of TSECs in the regulation of gene expression in Ishikawa cells. The differential gene expression induced by each TSEC indicates that these combinations may show differential clinical activity in the uteri of postmenopausal women using TSECs. Future studies will determine whether these findings in Ishikawa cells accurately model the uterus of postmenopausal women as well as explore the effects of differential dose and timing of TSEC administration.

Bazedoxifene is a novel SERM undergoing clinical development for the prevention and treatment of postmenopausal osteoporosis.^56,57 Results of preclinical studies using rodent models suggest that BZA alone has little to no endometrial stimulation and potently antagonizes CE-induced stimulation of the endometrium when the 2 are coadministered.^40,58 Similarly in women, CE/BZA treatment does not induce endometrial hyperplasia.^{13 ,41,59} The effects of BZA on markers of endometrial proliferation and differentiation demonstrated here suggest mechanisms by which BZA prevents endometrial proliferation leading to hyperplasia.

In conclusion, we demonstrate that the 4 SERMs analyzed here induced significantly different patterns of gene expression in an endometrial cancer cell line and showed differences in their abilities to antagonize E2-induced gene expression. Our data identify that BZA is more potent than RAL, LAS, and TAM in inhibiting the E2-mediated expression of proliferative genes (LIF and HOXA10) and at the same time maintains the expression of genes that have antiproliferative implications (PR and EMX2) in endometrial cells. Based on these data, a TSEC containing an estrogen and BZA offers optimal potential for endometrial protection, antagonizing the effects of E2.

Footnotes

References