Report

Differential Ligand Activation of Estrogen Receptors ERα and ERβ at AP1 Sites

See allHide authors and affiliations

Science  05 Sep 1997:
Vol. 277, Issue 5331, pp. 1508-1510
DOI: 10.1126/science.277.5331.1508

Abstract

The transactivation properties of the two estrogen receptors, ERα and ERβ, were examined with different ligands in the context of an estrogen response element and an AP1 element. ERα and ERβ were shown to signal in opposite ways when complexed with the natural hormone estradiol from an AP1 site: with ERα, 17β-estradiol activated transcription, whereas with ERβ, 17β-estradiol inhibited transcription. Moreover, the antiestrogens tamoxifen, raloxifene, and Imperial Chemical Industries 164384 were potent transcriptional activators with ERβ at an AP1 site. Thus, the two ERs signal in different ways depending on ligand and response element. This suggests that ERα and ERβ may play different roles in gene regulation.

Antiestrogens are therapeutic agents for the treatment and possible prevention of breast cancer. Tamoxifen (Fig. 1A) is an antiestrogen that is used in breast cancer chemotherapy and is believed to function as an antitumor agent by inhibiting the action of the estrogen receptor (ER) in breast tissue (1). Paradoxically, tamoxifen appears to function as an estrogen-like ligand in uterine tissue, and this tissue-specific estrogenic effect may explain the increased risk of uterine cancer that is observed with prolonged tamoxifen therapy (2). The related benzothiophene analog raloxifene (Fig. 1A) has been reported to retain the antiestrogen properties of tamoxifen in breast tissue and to show minimal estrogen effects in the uterus; in addition, it has potentially beneficial estrogen-like effects in nonreproductive tissue such as bone and cardiovascular tissue (3-7). One explanation for these tissue-specific actions of antiestrogens is that the ligand-bound ER may have different transactivation properties when bound to different types of DNA enhancer elements. The classical estrogen response element (ERE) is composed of two inverted hexanucleotide repeats, and ligand-bound ER binds to the ERE as a homodimer (Fig. 1B). The ER also mediates gene transcription from an AP1 enhancer element that requires ligand and the AP1 transcription factors Fos and Jun for transcriptional activation (Fig. 1B) (8). In transactivation experiments, tamoxifen inhibits the transcription of genes that are regulated by a classical ERE, but like the natural estrogen hormone 17β-estradiol [E2 (Fig. 1A)], tamoxifen activates the transcription of genes that are under the control of an AP1 element (9).

Figure 1

(A) Structures of ER ligands. The estrogens E2 and DES and the antiestrogens tamoxifen (Tam), raloxifene (Ral), and ICI 164384 (ICI) are shown. Bu, butyl; Me, methyl. (B) Models of ER action at a classical ERE and an ER-dependent AP1 response element. The filled circles represent the ligand bound to the ER. The AP1 proteins Jun and Fos are labeled J and F, respectively.

At the end of 1995, a second ER (ERβ) was cloned from a rat prostate cDNA library (10), and, subsequently, the human (11) and mouse (12) homologs were cloned. The first identified ER has been renamed ERα (10). The existence of two ERs presents another potential source of tissue-specific estrogen regulation. Here we compared the transactivation properties of ERα and ERβ with a panel of five ER ligands with the use of a reporter gene under the control of either a classical ERE or an AP1 element (13). Our results show that ERα and ERβ respond differently to certain ligands at an AP1 element. These results suggest different regulatory functions for the two ER subtypes.

We examined the transactivation properties of ERα (14) and ERβ (15) at a classical ERE in response to the estrogens E2 and diethylstilbestrol (DES) and the antiestrogens Imperial Chemical Industries (ICI) 164384, tamoxifen, and raloxifene (16). We conducted these experiments by transfecting HeLa cells with either an ERα or ERβ expression plasmid along with a reporter plasmid that contained a luciferase gene under the transcriptional control of an ERE (17). Both ERα (18) and ERβ (Fig.2) showed the same transactivation profiles with the panel of ligands. E2 and DES stimulated luciferase production 10-fold over ICI 164384, raloxifene, tamoxifen, and the control (no ligand added). The antiestrogens blocked E2 stimulation in ligand competition experiments (18).

Figure 2

ERβ action at an ERE. HeLa cells were transfected with an ERE-regulated luciferase reporter plasmid and an expression vector for rat ERβ (15). Transfected cells were treated with the five ligands (E2, 0.1 μM; DES, 1 μM; Ral, 1 μM; Tam, 5 μM; and ICI, 1 μM) or an ethyl alcohol (EtOH) vehicle (control) (17). Error bars show deviations between wells from a single representative transfection.

We next examined the ligand-induced transactivation behavior of ERα and ERβ at an AP1 site. With ERα, all five ligands stimulated luciferase transcription, including the antiestrogens ICI 164384, tamoxifen, and raloxifene (Fig.3). This stimulation was dependent on transfected ER, as cells transfected with only the reporter plasmid showed no induction of reporter transcription (18). Of the five ligands, raloxifene induced transcription the least, showing twofold induction compared with the sixfold inductions typically seen with E2 and tamoxifen. The raloxifene-induced transactivation was dose-dependent with a concentration value required for one-half maximal activation (EC50) of about 1 nM (18). In addition, raloxifene reduced the activation caused by E2 in a dose-dependent manner to the amount observed with raloxifene alone (18), demonstrating that raloxifene induction is weaker than induction by E2 and that raloxifene-induced transactivation results from binding to ERα. If E2 is classified as a full activator of ERα at an AP1 element (ERα-AP1), then raloxifene functions as a partial activator and tamoxifen functions as a full activator.

Figure 3

ERα action at an AP1 element. HeLa cells were transfected with an AP1 reporter plasmid and an ERα (14) expression plasmid and treated with the five ligands (17). Ligand concentrations were E2, 0.1 μM; DES, 1 μM; Ral, 1 μM; Tam, 5 μM; and ICI, 1 μM.

In contrast to the results seen with ERα-AP1, we observed a difference in the ligand activation profile of ERβ at an AP1 element (ERβ-AP1). In cells transfected with ERβ, treatment with the estrogens E2 and DES did not increase luciferase transcription over the control (no ligand added), whereas treatment with the antiestrogens ICI 164384, raloxifene, and tamoxifen increased luciferase transcription (Fig. 4A). This transcription activation required transfected ERβ, as cells that were transfected with only the reporter plasmid did not show transcriptional activation by the antiestrogens (18). The transcriptional activation caused by raloxifene was dose-dependent with an EC50 value of about 50 nM (Fig. 4B). In ligand competition experiments, both E2 and DES were able to block the raloxifene induction, and both estrogen ligands were able to reduce raloxifene induction to the basal level of transcription in a dose-dependent manner with concentration values required for one-half maximal inhibition of 1 to 10 nM (Fig. 4C). In a different ligand competition experiment, the inhibitory effect on transcription resulting from E2 treatment could be overcome by higher concentrations of raloxifene in a dose-dependent manner (Fig. 4D). Thus, it appears that the pharmacology of ER ligands is reversed at an AP1 element with ERβ: with ERβ-AP1, the antiestrogens act as transcription activators, and the estrogens act as transcription inhibitors.

Figure 4

(A) ERβ action at an AP1 response element. HeLa cells were transfected with an AP1 reporter plasmid and a rat ERβ expression plasmid (15). Transfected cells were treated with the following ligand concentrations: E2, 0.1 μM; DES, 1 μM; Ral, 1 μM; Tam, 5 μM; and ICI, 1 μM (17). (B) Dose response of raloxifene induction with ERβ at an AP1 element. HeLa cells transfected as described for (A) were treated with the indicated range of raloxifene concentrations. (C) Competitive inhibition of raloxifene induction by E2 and DES. HeLa cells were transfected as described for (A) and treated with ligands. The left panel shows transactivation induction by raloxifene (1 μM), the lack of induction by E2 (0.1 μM) and DES (1 μM), and the ability of E2 (0.1 μM) and DES (1 μM) to inhibit competitively raloxifene (1 μM) induction to the amount observed with the control (no ligand added). The right panel shows the dose dependence of inhibition of raloxifene (1 μM) induction by DES (solid line) and E2 (dashed line). (D) Raloxifene overriding E2 inhibition. HeLa cells were transfected as described for (A) and treated with ligands. The left panel shows the transcription induction resulting from the vehicle control (EtOH), Ral (10 μM) plus E2 (10 nM), and E2 (10 nM) alone. The right panel shows the dose dependence of raloxifene induction in the presence of E2 (10 nM).

We next asked whether the action of ERβ-AP1 could be observed in cell lines derived from estrogen target tissues such as the uterus and breast. We performed transactivation assays for ERβ-AP1 in Ishikawa cells (a human uterine cell line) (Fig.5A) and in MCF7 (Fig. 5B) and MDA453 (Fig. 5C) human breast cancer cells (19). In each of these cell lines, the ligands acted the same as they did in the HeLa cells; the three antiestrogens activated and the estrogens inhibited ERβ-dependent transcription from an AP1 site (Fig. 5). No induction was seen with cells that were not transfected with the ERβ expression plasmid, indicating that the antiestrogen induction required ERβ (18). Antiestrogen induction in the breast cell lines was higher than that observed in HeLa cells. Transfected MCF7 cells treated with raloxifene gave a 20- to 80-fold transactivation response over the control (no ligand added). In addition, raloxifene and ICI 164384 induced transcription more than tamoxifen in the breast cell lines (Fig. 5, B and C). MCF7 cells do not appear to contain high concentrations of endogenous ERβ mRNA (20); however, our results suggest that the additional transactivation machinery required for ERβ-AP1 function is present in these cells. With two of these target tissue cell lines, E2 treatment reduced the amount of transcription to less than that seen with the control (no ligand added). In MDA453 (Fig. 5C) and Ishikawa cells (Fig. 5A), E2 treatment resulted in a consistent 40 to 75% reduction of reporter transcription levels compared with the control. This effect was also observed in ligand competition experiments (Fig. 5, A and C); E2 and DES blocked raloxifene induction and reduced the amount of transcription to less than that seen for the control. Thus, when ERβ is bound by the estrogen hormone E2 or the synthetic estrogen DES, it may function as a negative regulator of genes controlled by an ER-dependent AP1 element.

Figure 5

(A) Ligand-dependent ERβ action at an AP1 element in Ishikawa cells. Ishikawa cells were transfected with an AP1-regulated luciferase reporter plasmid and an ERβ expression plasmid (19). Transfected cells were treated with one or two ligands as indicated (E2, 0.1 μM; DES, 1 μM; Ral, 1 μM; Tam, 5 μM; and ICI, 1 μM) or an EtOH vehicle (control) (17). (B) Ligand-dependent ERβ action at an AP1 element in MCF7 cells. MCF7 cells were treated and analyzed as described for (A). (C) Ligand-dependent ERβ action at an AP1 element in MDA453 cells. MDA453 cells were treated and analyzed as described for (A).

The ER is the only known member of the steroidal subfamily of nuclear receptors that has different subtypes (21,22). Nuclear receptors that respond to nonsteroidal hormones that have different known subtypes include the thyroid receptor (TRα and TRβ), the retinoic acid receptor (RARα, RARβ, and RARγ), and the retinoid X receptor (RXRα, RXRβ, and RXRγ) (23). Our results demonstrate that two nuclear receptor subtypes can respond in opposite regulatory modes to the natural hormone from the same DNA response element. Moreover, the ligand-induced responses with ERβ at an AP1 site provide an example of negative transcriptional regulation by the natural hormone and strong positive regulation by synthetic antiestrogens (24).

If signaling from ER-dependent AP1 elements occurs in estrogen target tissues, our finding that ERα and ERβ respond differently to ligands at AP1 sites reveals a potential control mechanism for transcriptional regulation of estrogen-responsive genes and adds a layer of complexity in analyzing the pharmacology of antiestrogen therapeutics. The role of E2 complexed to ERβ would be to turn off the transcription of these genes, whereas the antiestrogens raloxifene, tamoxifen, and ICI 164384 could overide this blockade and activate gene transcription. Thus, it may be helpful to search for genes in estrogen target tissues that are transcriptionally regulated by ERβ at an AP1 site and to characterize the phenotype of cells in which these genes are activated.

REFERENCES AND NOTES

View Abstract

Navigate This Article