Estrogen response element sequence impacts the conformation and transcriptional activity of estrogen receptor α¹

Abstract

Estrogens play a critical role in mammary gland development, bone homeostasis, reproduction, and the pathogenesis of breast cancer by activating estrogen receptors (ERs) α and β. Ligand-activated ER stimulates the expression of target proteins by interacting with specific DNA sequences: estrogen response elements (EREs). We have demonstrated that the ERE sequence and the nucleotide sequences flanking the ERE impact ERα binding affinity and transcriptional activation. Here, we examined whether the sequence of the ERE modulates ERα conformation by measuring changes in sensitivity to protease digestion. ERα, occupied by estradiol (E₂) or 4-hydroxytamoxifen (4-OHT), was incubated with select EREs and digested by chymotrypsin followed by a Western analysis with antibodies to ERα. ERE binding increased the sensitivity of ERα to chymotrypsin digestion. We found both ligand-specific and ERE-specific differences in ERα sensitivity to chymotrypsin digestion. The ERE-mediated increase in ERα sensitivity to chymotrypsin digestion correlates with E₂-stimulated transcriptional activity from the same EREs in transiently transfected cells. Transcriptional activity also correlates with the affinity of ERα-ERE binding in vitro. Our results support the hypothesis that the ERE sequence acts as an allosteric effector, altering ER conformation. We speculate that ERE-induced alterations in ERα conformation modulate interaction with co-regulatory proteins.

Introduction

Estrogen receptor (ER) is a member of the steroid/thyroid superfamily of proteins that act as hormone-inducible transcription factors (Mangelsdorf et al., 1995). The products of two ER genes, i.e. ERα and ERβ, mediate the actions of estrogens in target tissues, thus regulating estrogenic effects on reproduction, bone homeostasis, and mammary gland structure and function (Katzenellenbogen and Korach, 1997). Estradiol (E₂) and other ligands bind in the ligand-binding domain (LBD) of ERα and induce conformational changes leading to dimerization and high-affinity ERα binding to estrogen response elements (EREs) (Klinge, 2000). Once bound to an ERE, interaction between specific motifs within E₂-ERα and coactivator proteins, e.g. SRC-1 (Onate et al., 1995), or the TATA binding complex, e.g. TFIIB (Ing et al., 1992) and TATA-box binding protein (TBP) (Sadovsky et al., 1995), result in increased gene transcription (reviewed in McKenna et al., 1999, Klinge, 2000).

Occupation of ERα by selective estrogen receptor modulators (SERMs), e.g. tamoxifen (TAM), 4-hydroxy-TAM (4-OHT), or raloxifene (RAL), changes the interaction of ERα with coactivators. Whereas E₂-occupied ERα interacts with specific coactivators, e.g. SRC-1, occupation of ERα by TAM or 4-OHT prevents interaction with these proteins (reviewed in McKenna et al., 1999, Klinge, 2000). Although antagonist-liganded ERα interacts with corepressors NCoR and SMRT (Smith et al., 1997, Zhang et al., 1998), there is no evidence that these corepressors are required for the antagonist activity of TAM. TAM-occupied ERα also interacts with the ER-selective coregulator REA that potentiates the antagonist activity of TAM and suppresses E₂-induced transcriptional activity of ERα and ERβ (Montano et al., 1999). Similarly, the nuclear matrix protein HET/SEF-B shows a preferential interaction with TAM-occupied ERα and inhibits both E₂- and TAM- stimulated transcription in HepG2 cells in which TAM is an agonist (Oesterreich et al., 2000). Although crystal structure studies show a different conformation of the LBD occupied by 4-OHT versus E₂ or the synthetic estrogen diethylstilbestrol (DES) (Brzozowski et al., 1997, Shiau et al., 1998), specific details of the mechanisms of antagonist activity of 4-OHT remain to be elucidated.

In addition to ligand effects, Lefstin and Yamamoto proposed that response elements recognized by nuclear transcription factors, including members of the steroid/nuclear receptor superfamily, contain information that is interpreted by bound regulator factors (Lefstin and Yamamoto, 1998). Based on Lefstin and Yamamoto's model, we postulate that DNA acts as an allosteric ligand whose binding alters ERs' affinity for other ligands, such as coactivators or corepressors, and predict that the DBD of the ERα harbors activities in addition to DNA recognition. Indeed, the ERα DBD constitutes surfaces for both intramolecular and intermolecular protein–protein contacts (Chen et al., 1999). Thus, ERα-ERE binding should produce changes in ERα conformation that alter ERα interaction with other proteins.

To assess alterations in ERα conformation in the presence of various ERE sequences, we analyzed the sensitivity of ERα to α-chymotrypsin digestion. An earlier study showed that chymotrypsin or trypsin digestion of MCF-7 cell extracts resulted in different sizes of ERα fragments in cells treated with E₂ versus tamoxifen (TAM) (Hutchens et al., 1987). This result was an early indication that the conformation of E₂-occupied ERα is different from TAM-occupied ERα, a finding confirmed by X-ray crystal structure studies of purified ERα LBD occupied by 4-OHT versus E₂ or DES (Brzozowski et al., 1997, Shiau et al., 1998). A more recent report showed that specific amino acid (aa) mutations that constitutively activate ERα result in a proteolytic digestion pattern similar to that of the E₂-liganded wild-type (wt) ERα (Lazennec et al., 1997). These reports indicate that alterations in ERα conformation correlate with ERα agonist activity, whether induced by E₂ binding or by alterations in aa sequence. Because ER binding to an ERE activates gene expression from that element, it is important to investigate if there is a similar correlation between ERE-driven transcription and altered ER conformation.

Two reports conflict on the effect of DNA binding on ERα conformation as measured by sensitivity of ERα to trypsin digestion (Fritsch et al., 1992, Wood et al., 1998). One report showed that rat uterine ERα conformation is unaffected by DNA binding (Fritsch et al., 1992). The other report showed that yeast-expressed, recombinant human (rh) ERα is more sensitive to trypsin or chymotrypsin digestion when bound to the Xenopus vitellogenin A2 ERE than when ERα was incubated with the ERE from the human pS2 gene, containing a single nucleotide change in the 3’ half-site (Wood et al., 1998). The latter is important because, although many investigations of ERα activity utilize the perfectly palindromic Xenopus vitellogenin A2 ERE, the promoters of most natural estrogen responsive genes contain one or more imperfect EREs and/or ERE half-sites (Anolik et al., 1995). No one has examined how the ligand affects ERα conformation in the presence of perfect versus imperfect EREs.

In the present study, we assessed changes in the conformation of ERα in the presence or absence of a consensus ERE, naturally occurring non-palindromic EREs, or indirect and direct repeats (IR or DR, respectively) of the ERE half-site. We report that ERE binding enhances the sensitivity of ERα to proteolysis by chymotrypsin. In contrast, incubation of ERα with DNA sequences that do not bind ERα does not alter ERα sensitivity to proteolysis. We detected differences in the sensitivity of ERα to chymotrypsin digestion in the presence of different EREs and some changes in the sizes of proteolytic fragments with E₂ or 4-OHT. Although the affinity of ERα for the different EREs does not correlate with transcriptional activity in transiently transfected cells, the sensitivity to chymotrypsin digestion correlates with reporter gene expression. These results are consistent with a role for the sequence of the ERE as an allosteric regulator of ERα activity.