Report

Molecular Determinants for the Tissue Specificity of SERMs

See allHide authors and affiliations

Science  29 Mar 2002:
Vol. 295, Issue 5564, pp. 2465-2468
DOI: 10.1126/science.1068537

Abstract

Selective estrogen receptor modulators (SERMs) mimic estrogen action in certain tissues while opposing it in others. The therapeutic effectiveness of SERMs such as tamoxifen and raloxifene in breast cancer depends on their antiestrogenic activity. In the uterus, however, tamoxifen is estrogenic. Here, we show that both tamoxifen and raloxifene induce the recruitment of corepressors to target gene promoters in mammary cells. In endometrial cells, tamoxifen, but not raloxifene, acts like estrogen by stimulating the recruitment of coactivators to a subset of genes. The estrogen-like activity of tamoxifen in the uterus requires a high level of steroid receptor coactivator 1 (SRC-1) expression. Thus cell type– and promoter-specific differences in coregulator recruitment determine the cellular response to SERMs.

Tamoxifen and raloxifene are selective estrogen receptor modulators (SERMs) that bind the estrogen receptor (ER) and modulate ER-mediated gene transcription. Tamoxifen is an effective treatment for all stages of hormone-responsive breast cancer and can prevent breast cancer in high-risk women (1). However, tamoxifen has partial estrogenic activity in the uterus and is associated with an increased incidence of endometrial hyperplasia and cancer. Raloxifene, approved for the prevention and treatment of osteoporosis in postmenopausal women, also appears to prevent breast cancer, but it does not increase the incidence of endometrial cancer. The National Cancer Institute supported “Study of Tamoxifen and Raloxifene” (STAR Trial) is currently being conducted to compare the safety and effectiveness of these two agents for the prevention of breast cancer in postmenopausal women (2).

The molecular mechanism underlying the tissue-specificity of SERM action is not clear. The crystal structures of the liganded ER hormone-binding domain (HBD) indicate that both tamoxifen and raloxifene can act as ER antagonists by competing with estradiol (E2) for binding and by inducing conformational changes that block the interaction of ER with coactivator proteins (3,4). However, this does not explain how SERMs act as agonists or the differences in the spectrum of activity among various SERMs.

Estrogen receptor can regulate gene transcription either by binding directly to the promoter of target genes or by binding indirectly through a mechanism involving other transcription factors such as Sp1 and AP1. Genes regulated through direct ER binding, such asCATD (encoding cathepsin D) (5) andEBAG9 (encoding ER-binding fragment-associated antigen 9) (6, 7), typically harbor an estrogen responsive element (ERE) with a consensus sequence of 5′-GGTCAnnnTGACC-3′ in their promoters. Genes regulated by binding ER indirectly include c-Myc (8) andinsulin-like growth factor-I (IGF-I) (9), whose promoters do not contain a classical ERE.

We examined transcriptional responses to tamoxifen and raloxifene in the mammary carcinoma cell line MCF-7 and the endometrial carcinoma cell line Ishikawa. In both cell types, estradiol (E2) induced the expression of both the directly bound ER target genes CATDand EBAG9 and the indirectly bound target genesc-Myc and IGF-I (Fig. 1). Neither tamoxifen nor raloxifene stimulated the expression of CATDor EBAG9 in either MCF-7 or Ishikawa cells (Fig. 1). It is noteworthy, however, that in Ishikawa cells, but not in MCF-7 cells, tamoxifen, but not raloxifene, induced the expression c-Mycand IGF-I, whose promoters do not contain a classical ERE. Similar tissue-specific results were also obtained in another endometrial carcinoma cell line ECC-1 and another mammary carcinoma cell line T47-D (10). These observations suggest that promoter context is one of the determinants for tissue-specific activities of tamoxifen.

Figure 1

Stimulation of c-Myc and IGF-I expression by tamoxifen only in endometrial carcinoma cells. MCF-7 cells (A) or Ishikawa cells (B) were grown in phenol red–free Dulbecco's modified Eagle's medium (DMEM) supplemented with 5% charcoal-dextran–stripped fetal bovine serum (FBS) for at least 3 days and left untreated or treated with 100 nM of 17β-estradiol (E2), 1 μM of 4-hydroxytamoxifen (tamoxifen), or 1 μM of raloxifene for different times. Total RNAs were extracted using TRIzol Reagent (Invitrogen Corp., Carlsbad, CA), and the expression of c-Myc, IGF-I,EBAG9, or cathepsin D genes was measured by real-time reverse transcriptase (RT) polymerase chain reaction (PCR) using the ABI PRISM 7700 Sequence Detector and the TaqMan EZ RT-PCR kit (Applied Biosystems, Foster City, CA).

Estrogen receptor–mediated transcriptional activation is associated with the recruitment of coactivators, such as AIB1, GRIP1, SRC-1, CBP, p300, and pCAF, and subsequent histone acetylation (11–14). In contrast, antagonist-liganded ER is able to recruit corepressors (15–18). Previously, we showed in MCF-7 breast cancer cells that, when bound by tamoxifen, ER recruits the corepressors NCoR and SMRT and a subset of histone deacetylases (HDACs) to target promoters (18). Further examination of the recruitment of ER coregulators to target gene promoters by chromatin immunoprecipitation (ChIP) revealed that, in MCF-7 cells as well as in Ishikawa cells, both tamoxifen and raloxifene induce the recruitment of corepressors and HDACs to the CATD promoter (Fig. 2A, lower panels). In striking contrast, in Ishikawa cells, but not in MCF-7 cells, instead of inducing the recruitment of a corepressor complex, tamoxifen, but not raloxifene, induced the recruitment of a coactivator complex including SRC-1, AIB1, and CBP to the c-Myc promoter (Fig. 2B, upper panels, lanes 7, 15, 19). Tamoxifen-stimulated coactivator recruitment was accompanied by histone acetylation (Fig. 2B, upper panels, lane 23) consistent with the current model of gene activation by nuclear receptors. Tamoxifen-induced coactivator recruitment to thec-Myc promoter was also detected in ECC-1 cells and to theIGF-I promoter in both endometrial cancer cell lines (10).

Figure 2

Coregulator recruitment on ER target gene promoters. MCF-7 cells or Ishikawa cells were grown in phenol red–free DMEM supplemented with 5% charcoal-dextran–stripped FBS for at least 3 days and left untreated (C) or treated with 100 nM of E2 (E), 1 μM of 4-hydroxytamoxifen (T), or 1 μM of raloxifene (R) for 45 min. ChIP assays (18) were performed using specific antibodies against (A) NCoR, SMRT, and HDAC4; and HDAC2 (Santa Cruz Biotechnology, Santa Cruz, CA); and (B) ERα (Ab-10, NeoMarkers, Fremont, CA); SRC-1 (a mouse monoclonal); GRIP1 (rabbit polyclonal); AIB1 (affinity-purified rabbit polyclonal); CBP (mouse monoclonal AC26); and acetylated histones (AcH) (Upstate Biotechnology, Lake Placid, NY).

As ER regulates the rate of gene transcription through its association with coregulators, the overall balance of the relative expression levels of coactivators and corepressors may be an important determinant of the tissue-specificity of SERMs. Examination of the expression levels of ERα and a variety of coregulators indicated similar levels of expression in MCF-7 and Ishikawa cells with the exception of SRC-1 (Fig. 3A), whose expression is low in MCF-7 compared with that in Ishikawa cells. The high level of SRC-1 expression in endometrial cells as compared with mammary cells was confirmed in several different cell lines (10). To investigate whether this difference in the level of SRC-1 expression explained the ability of tamoxifen to stimulatec-Myc and IGF-I transcription, we first overexpressed SRC-1 in MCF-7 cells. Remarkably, expression of bothc–Myc and IGF-I was stimulated by tamoxifen in SRC-1–transfected MCF-7 cells but not in GRIP1- or AIB1-transfected cells (Fig. 3B). This finding supports our conclusion that a high level of SRC-1 expression is sufficient to support the agonist activity of tamoxifen.

Figure 3

(A) Comparison of SRC-1 expression levels in endometrial carcinoma cells and in mammary carcinoma cells. (A) Cells were grown in phenol red–free DMEM supplemented with 5% charcoal-dextran–stripped FBS. Total proteins were extracted, and Western blottings were performed using antibodies against ERα, CBP, AIB1, GRIP1, SRC-1, p300 (mouse monoclonal RW128), NCoR, SMRT, HDAC2, and HDAC4. (B) Stimulation of c-Myc expression by tamoxifen in MCF-7 cells overexpressing SRC-1. MCF-7 cells were seeded in phenol red–free DMEM supplemented with 5% charcoal-dextran-stripped FBS for 24 hours and were transfected with an expression construct for SRC-1, GRIP1, or AIB1 by using the Lipofectamine 2000 Reagent (Invitrogen Corp.). Forty-eight hours after transfection, cells were treated with 100 nM of 17β-estradiol (E2), 1 μM of 4-hydroxytamoxifen (tamoxifen), or 1 μM of raloxifene for different times. The TRIzol Reagent was used to extract total RNAs for measuring mRNA level by real-time RT-PCR.

To determine whether SRC-1 is required for tamoxifen agonism, we silenced its expression in Ishikawa cells by RNA interference using short interfering RNA (siRNA) molecules (10, 19). Reduction of SRC-1 levels in Ishikawa cells eliminated tamoxifen-stimulated expression of c-Mycand IGF-I (Fig. 4A). It was interesting that SRC-1 silencing had only minimal effects on the E2-stimulated expression of c-Myc and IGF-I. In contrast, silencing of AIB1 expression led to a modest decrease in both E2- and tamoxifen-stimulated expression of c-Myc andIGF-I (Fig. 4A). These results strongly suggest that, although AIB1 plays a role in the maximal activity of both estrogen and tamoxifen, SRC-1 is specifically necessary for the agonist activity of tamoxifen in endometrial cells. These observations also suggest that the specific coactivator requirements for estrogen- and tamoxifen-stimulated gene expression are distinct.

Figure 4

(A) The effect of SRC-1 silencing on tamoxifen-stimulated gene expression in Ishikawa cells. Ishikawa cells were seeded into 10-cm polystyrene cell-culture dishes (Becton Dickinson, Franklin Lakes, NJ) with phenol red–free DMEM supplemented with 5% charcoal-dextran-stripped FBS for 24 hours and transfected with 5 μg/dish of double-stranded, short interfering RNAs (siRNAs) for SRC-1, AIB1, or lamin A/C using the Oligofectamine Reagent (Invitrogen Corp.). Single-stranded RNAs were synthesized by Dharmacon Research, (Lafayette, CO). Before transfection, single-stranded RNAs were incubated at 90°C for 1 min, followed by annealing in annealing buffer (100 mM potassium acetate, 30 mM Hepes-KOH, pH 7.4, and 2 mM magnesium acetate) at 37°C for 2 hours. Forty-eight hours after transfection, cells were treated with 100 nM of 17β-estradiol (E2), 1 μM of 4-hydroxytamoxifen (Tamoxifen), or 1 μM of raloxifene. The TRIzol reagent was used to extract total RNAs for analyzing c-Myc and IGF-I mRNA by real-time RT-PCR. Transfection efficiency was monitored by cotransfection with an Escherichia coli lacZconstruct. (B) The effect of SRC-1 silencing on tamoxifen-stimulated cell-cycle entry. Ishikawa cells grown in phenol red–free DMEM supplemented with 5% charcoal-dextran–stripped FBS were cotransfected with 5 μg of SRC-1 siRNAs and a green fluorescent protein construct (pEGFP, Clontech) or cotransfected with lamin A siRNA and pEGFP. Forty-eight hours after transfection, cells were treated with 100 nM of 17β-estradiol (E2) or 1 μM of 4-hydroxytamoxifen (T) for another 16 hours. Cells were then collected and resuspended in PBS with 2% glucose and 3% paraformaldehyde. After permeabilization with ethanol, cells were stained with propidium iodide solution (69 μM propidium iodide and 38 mM sodium citrate). Cell-cycle data were collected with FACScan (Becton Dickinson Immunocytochemistry System) and analyzed with ModFit LT (Verity Software House, Topsham, ME).

To determine whether SRC-1 expression was required for the growth stimulatory effects of tamoxifen in endometrial cells, we examined the effects of SRC-1 silencing on tamoxifen-stimulated cell-cycle progression in Ishikawa cells (Fig. 4B). As was the case forc-Myc and IGF-I expression, SRC-1 silencing abolished tamoxifen-stimulated cell-cycle progression but had only minimal effects on E2-stimulated cell-cycle progression. These results indicate that SRC-1 is a necessary determinant for the estrogenic effect of tamoxifen in endometrial cells.

In summary, in the breast where tamoxifen and raloxifene are both antagonists, both SERMs induce the recruitment of corepressors and not coactivators to ER target promoters. In contrast, in the endometrium where tamoxifen acts as an agonist and raloxifene as an antagonist, tamoxifen recruits coactivators instead of corepressors to ER target genes that do not contain a classical ERE, such asc-Myc and IGF-I. Finally, SRC-1 is required for the estrogen-like properties of tamoxifen in the endometrium.

It is unclear how coactivators are recruited by tamoxifen-bound ER to promoters that do not contain an ERE. Whether the ER AF-1 domain implicated in the agonist activity of tamoxifen (20–23) or the reported in vitro interactions of SRC-1 with AF-1 (24, 25) are relevant to the recruitment of SRC-1 by tamoxifen-bound ER remains to be shown. It may be that the binding of coactivators to tamoxifen-liganded ER is blocked when ER is directly bound to DNA through a classical ERE, but that when interacting with promoters indirectly, tamoxifen-bound ER adopts a conformation that promotes SRC-1 binding.

These experiments are based on a limited number of ER target genes and coactivators. It remains to be determined if c-Mycand/or IGF-I are the critical genes involved in tamoxifen-stimulated endometrial growth or endometrial cancer. However,c-Myc has been implicated in cell growth, proliferation, apoptosis, and malignant transformation (26). In addition, overexpression of c-Myc and c-Myc gene amplification have been reported in a variety of malignancies including endometrial cancer (27, 28). Likewise, the roles of IGF-I in cell proliferation and survival have also been well established (29).

Finally, our results do not exclude the possibility that other as-yet-undetermined cell-specific factors may contribute to the spectrum of SERM action. Our findings, however, do establish that cell type– and promoter-specific differences in coregulator recruitment plays a critical role in determining SERM function in the breast and uterus and offers a paradigm for understanding SERM action in other important target organs such as the brain, skeleton, and cardiovascular system.

  • * To whom correspondence should be addressed. E-mail: myles_brown{at}dfci.harvard.edu

REFERENCES AND NOTES

View Abstract

Navigate This Article