HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

This Article Abstract Full Text (PDF) Alert me when this article is cited Alert me if a correction is posted Services Similar articles in this journal Similar articles in ISI Web of Science Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via ISI Web of Science (37) Citing Articles via Google Scholar Google Scholar Articles by Klinge, C M Articles by Zhang, J Search for Related Content PubMed PubMed Citation Articles by Klinge, C M Articles by Zhang, J

Department of Biochemistry and Molecular Biology, Center for Genetics and Molecular Medicine, University of Louisville School of Medicine, Louisville, Kentucky 40292, USA

(Requests for offprints should be addressed to C M Klinge; Email: carolyn.klinge{at}louisville.edu)

	Abstract

Top Abstract Introduction Materials and methods Results Discussion References

 
One mechanism by which ligand-activated estrogen receptors and ß (ER and ERß) stimulate gene transcription is through direct ER interaction with specific DNA sequences, estrogen response elements (EREs). ERE-bound ER recruits coactivators that stimulate gene transcription. Binding of ER to natural and synthetic EREs with different nucleotide sequences alters ER binding affinity, conformation, and transcriptional activity, indicating that the ERE sequence is an allosteric effector of ER action. Here we tested the hypothesis that alterations in ER conformation induced by binding to different ERE sequences modulates ER interaction with coactivators and corepressors. CHO-K1 cells transfected with ER or ERß show ERE sequence-dependent differences in the functional interaction of ER and ERß with coactivators steroid receptor coativator 1 (SRC-1), SRC-2 (glucocorticoid receptor interacting protein 1 (GRIP1)), SRC-3 amplified in breast cancer 1 (AIB1) and ACTR, cyclic AMP binding protein (CBP), and steroid receptor RNA activator (SRA), corepressors nuclear receptor co-repressor (NCoR) and silencing mediator for retinoid and thyroid hormone recpetors (SMRT), and secondary coactivators coactivator associated arginine methyltransferase 1 (CARM1) and protein arginine methyltransferase 1 (PRMT1). We note both ligand-independent as well estradiol- and 4-hydroxytamoxifen-dependent differences in ER-coregulator activity. In vitro ER-ERE binding assays using receptor interaction domains of these coregulators failed to recapitulate the cell-based results, substantiating the importance of the full-length proteins in regulating ER activity. These data demonstrated that the ERE sequence impacts estradiol-and 4-hydroxytamoxifen-occupied ER and ERß interaction with coregulators as measured by transcriptional activity in mammalian cells.

	Introduction

Top Abstract Introduction Materials and methods Results Discussion References

 
Estrogens exert a wide variety of effects on growth, development, and differentiation primarily through binding to a specific intranuclear estrogen receptor protein (ER) encoded by two genes: and ß (ER and ERß) (Couse & Korach 1999). Stimulation of target gene expression in response to 17ß-estradiol (E₂), or other agonists, is thought to be mediated by two mechanisms: (1) direct binding of E₂-liganded ER (E₂-ER) to a specific sequence called an estrogen response element (ERE) and (2) ‘tethering’, in which ER interacts with another DNA-bound transcription factor, e.g. Sp1 (Safe 2001) or AP-1 (Webb et al. 1995), in a way that stabilizes the DNA binding of that transcription factor in the absence of direct ER-DNA binding. Both processes result in recruitment of coactivators and components of the RNA polymerase II transcription initiation complex that enhances target gene transcription (Klinge 2000).

We recently reviewed studies on ER interaction with 38 estrogen-responsive genes whose promoters or 3' untranslated regions contain functional EREs (Klinge 2001). Most estrogen-regulated genes contain imperfect, non-palindromic EREs. One conclusion from our review is that the more nucleotide changes there are from the consensus within a half-site of the ERE palindrome, the lower the ER binding affinity and the lower the transcriptional activity (Klinge 2001). Further, EREs in which nucleotides are altered in both arms of the palindrome show lower transcriptional activity than those containing alterations in only one ERE half-site (Klinge 2001). These results have been confirmed experimentally (Driscoll et al. 1998, Tyulmenkov & Klinge 2000a, 2001a, Tyulmenkov & Klinge b, Tyulmenkov et al. 2000, Kulakosky et al. 2002).

Lefstin & Yamamoto (1998) 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. Our data fit the model of Lefstin and Yamamoto by showing that different EREs bound ER with different affinity and resulted in different amounts of transactivation of a reporter plasmid in transfected cells (Tyulmenkov et al. 2000, Klinge et al. 2001, Tyulmenkov & Klinge 2001b, Kulakosky et al. 2002). We and other investigators have proposed that DNA acts as an allosteric effector whose binding alters ER conformation such that different EREs would be hypothesized to regulate ER interaction with other proteins, e.g. coactivators or corepressors (reviewed in Klinge et al. 2001). In concordance with this hypothesis, we observed differences in ER conformation, assessed by -chymotrypsin or trypsin digestion, upon interaction with a palindromic ERE or non-palindromic EREs from the human pS2, progesterone receptor (PR), and c-fos genes and a direct repeat of the ERE half-site separated by 5 bp (Klinge et al. 2001, Ramsey & Klinge 2001). The ERE-mediated increase in ER sensitivity to protease digestion correlated with E₂-stimulated transcription from the same EREs in transiently transfected cells and with the affinity of ER-ERE binding in vitro (Klinge et al. 2001, Ramsey & Klinge 2001, Tyulmenkov & Klinge 2001a,b, Kulakosky et al. 2002).

Similarly, other investigators have reported differential recognition of ER by select antibodies when ER was bound to the Xenopus laevis vitellogenin A2 (vit A2) ERE versus the non-palindromic ERE from the human pS2 gene in electrophoretic mobility shift assay (EMSA) experiments (Wood et al. 1998). Limited protease digestion of ER bound to vit A2 ERE or the non-palindromic pS2, vitellogenin B1, and oxytocin EREs revealed different sized [³²P]DNA fragments in EMSAs (Wood et al. 2001). These data were interpreted as indicating that ER binding to different EREs changes ER conformation (Wood et al. 1998, 2001). Likewise, experiments using phage display revealed conformational differences in ER and ERß when bound to vit A2, pS2, lactoferrin, and complement 3 (C3) EREs in vitro (Hall et al. 2002).

Only a few investigators have examined how these ERE-induced alterations in ER conformation impact ER-coactivator interaction. One study reported that GRIP1 showed less interaction when ER bound to the pS2 ERE versus the vit A2, vitellogenin B1, or oxytocin EREs, indicating that the ERE sequence impacts ER-GRIP1 interaction in vitro (Wood et al. 2001). Similar conclusions based on EMSA and GST-pulldown assays were recently reported by another group of investigators (Yi et al. 2002). However, the functional significance of these observations of ERE-dependent differences in ER conformation and recognition by coactivator receptor interaction domains (RIDs) has not been tested by cell-based transcription assays.

Over the past 6 years, at least 28 different ER coactivator proteins have been identified (McKenna et al. 1999, Klinge 2000, McKenna & O’Malley 2002). Less information is available regarding ERß-coactivator interaction. In brief, coactivators function to (1) acetylate the N-terminal tails of lysine residues in histones H3 and H4 leading to ‘relaxed’ chromatin structure (reviewed in Struhl 1998), (2) acetylate other transcription factors and coactivators (Jiang et al. 2000), (3) recruit secondary coactivators including coactivator associated arginine methyltransferase 1 (CARM1) and protein arginine methyltransferase 1 (PRMT1) that methylate histones (Koh et al. 2001), (4) interact with components of various ATP-dependent chromatin-remodeling complexes (Hebbar & Archer 2003), and (5) direct interaction with and stabilization of basal transcription factor binding (reviewed in Hager et al. 1998). Once the transcription initiation complex is complete, RNA polymerase II is recruited to the transcription start site and begins transcription.

Most of the cell-based assays examining the effects of coactivators on ER-mediated transcription have used reporter genes containing two or three tandem copies of the vit A2 ERE (Norris et al. 1998, Lee et al. 1999, Mak et al. 1999, Delage-Mourroux et al. 2000). The non-palindromic EREs that have been used in transfection experiments are from the complement C3, transforming growth factor-ß3, and lactoferrin gene promoters that were used to examine the functional interaction of ER with the receptor of estrogen receptor activity (Delage-Mourroux et al. 2000) and SRC-2/GRIP1 (Norris et al. 1998).

Selective estrogen receptor modulators (SERMS) in vitro and in transfected cells (Metivier et al. 2002). Treatment of cells with the histone deacetylase (HDAC) inhibitor trichostatin A has been shown to enhance ligand-independent as well as E₂-, tamoxifen-, and raloxifene-stimulated ER transcription of an ERE-reporter gene, indicating that repression of ER transcriptional activity is mediated by a corepressor-HDAC complex (Webb et al. 2003). To address the functional consequences of the ERE sequence-dependent alterations on ER- or ERß-coactivator interaction, we have performed transient transfection experiments and quantitated the impact of representative coactivators on E₂-induced reporter gene expression from single copies of the vit A2, pS2, PR, and c-fos EREs. We report that different ERE sequences alter coactivator enhancement of E₂-induced transcriptional activity by ER and ERß. Further, differences in coactivator interaction between ER and ERß were also apparent. These data demonstrated that alterations in ERE sequence functionally impact coregulatory protein interaction with ER and ERß and that ER and ERß show subtype-dependent coactivator interactions.

	Materials and methods

Top Abstract Introduction Materials and methods Results Discussion References

 
ER preparation and ER-ERE K_d determinations

Human ER and ERß1 were expressed from baculoviruses in Sf21 cells as described (Kulakosky et al. 2002). EMSAs were used to determine the affinity of E₂-occupied ER and ERß for the following EREs: EREc13: 5'-CCGGTCAGAGTGACCAG-3'; EREc38 (which is identical to the commonly used Xenopus vit A2 ERE (Peale et al. 1989)): 5'-CCAGGTCAGAGTGACCTGAGCTAAAATAACACATT-3'; PR1148: 5'-AGCCCTCCCTCCTGCGAGGTCACCAGCTCTTGGTGCCTGTTT-3'; pS2: 5'-CTTCCCCCTGCAAGGTCAGCGTGGCCACCCCGTGAGCCACT-3'; and Fos-1211: 5'-AGCTTGGGCTGAGCCGGCAGCGTGACCCCGCATG-3'.

The underlined nucleotides correspond to the minimal core consensus ERE. The nucleotides in bold indicate an alteration in the consensus ERE. EREc38, PR1148, pS2, and Fos-1211 were cloned into the pGL3-promoter luciferase reporter vector (Promega, Madison, WI, USA) as previously described (Klinge et al. 1997). Details of the EMSA experiments to determine ER-ERE binding affinity are provided in Kulakosky et al.(2002), Tyulmenkov et al.(2000), and Tyulmenkov & Klinge (2001a,b).

Cell culture, transient transfection, and reporter assays

Chinese Hamster Ovary cells (CHO-K1) were purchased from ATCC (Manasas, VA, USA) and maintained in Iscove’s modified Dulbecco’s medium (IMDM) (Gibco/Invitrogen, Grand Island, NY, USA) supplemented with heat-treated, charcoal-stripped 10% fetal bovine serum (Atlanta Biologicals, Norcross, GA, USA). All other cell culture reagents were purchased from Gibco-BRL (Grand Island, NY, USA). For transient transfection, CHO-K1 cells were plated into 24-well plates at 1x10⁵ cells/well with IMDM (without phenol red) supplemented with 10% charcoal-stripped fetal bovine serum. The cells were transfected when 80% confluent with 0.25 µg reporter construct containing the ERE, 5 ng pRL-CMV (RL=Renilla luciferase from Promega, Madison, WI, USA), 10 ng pCMV-human ER or pCMV-rat ERß, graciously provided by Drs B S Katzenellenbogen and J-A Gustafsson respectively. The expression plasmid for NCoR was provided by Dr M Bagchi (Zhang et al. 1998). The expression plasmid for SMRT was provided by Dr B G Rowan (Coleman et al. 2003). The expression plasmids for SRC-1 and SRA were provided by Dr B W O’Malley (Lanz et al. 1999). The expression plasmid for CBP was provided by Dr M Hadzopoulou-Cladaras (Dell & Hadzopoulou-Cladaras 1999). The expression plasmids for GRIP1, ACTR, CARM1, and PRMT1 were provided by Dr M Stallcup (Koh et al. 2001). The amounts of each of the coregulator plasmids used in transfection are indicated in the Figures and their legends. In experiments comparing the activity of a single coactivator with that achieved by co-transfecting two coactivators, coactivators were run alone or in combination within the same transfection to obviate differences between transient transfections.

The transient transfections were performed using Transfast (Promega) as previously described (Klinge et al. 1999, 2000, Tyulmenkov et al. 2000). The total amount of DNA transfected was kept constant using pcDNA3 (Invitrogen, Carlsbad, CA, USA) as ‘filler DNA’. Cells were treated, in triplicate, 24 h later with ethanol (EtOH, vehicle), 10 nM E₂ (Sigma, St Louis, MO, USA), 100 nM 4-hydroxytamoxifen (4-OHT) (Research Biochemicals International, Natick, MA, USA), or both E₂ and 4-OHT. Cells were harvested 30 h later and luciferase and Renilla-luciferase (RL-luc) activities assayed using the Promega dual luciferase reporter assay (Klinge et al. 1997, 1999, Tyulmenkov et al. 2000). All data for transient transfections were normalized by RL-luc to account for transfection efficiency.

Statistical analyses

Statistical analyses were performed on fold-induction data values from multiple experiments (minimally three) in which each treatment was run in triplicate using either two-tailed Student’s t-test or one-way ANOVA followed by Dunn’s multiple comparison or Dunnett’s post-hoc test using GraphPad Prism (GraphPad Software, Inc., San Diego, CA, USA).

CHO-K1 cells were transfected with expression vectors for coactivators, corepressors, ER, ERß, and reporter vectors as described above. Whole cell extracts were prepared as described previously (Klinge et al., 2000). Protein concentrations were determined by BioRad’s DCC (Lowry) assay (BioRad, Hercules, CA, USA) using BSA as a standard. Equal amounts of protein (75 µg/lane) were separated by electrophoresis on 10% SDS-PAGE gels. Proteins were transferred to polyvinylidene difluoride membranes (Pall Corporation, Ann Arbor, MI, USA). The blots were probed with primary antibodies: SRC-1 (GeneTex, San Antonio, TX, USA); GRIP1 and ACTR (both from US Biological, Swampscott, MA, USA); ß-actin (Amersham, Arlington Heights, IL, USA); glyceraldehyde-3-phosphate dehydrogenase (GAPDH; Research Diagnostics, Inc., Flanders, NJ, USA). Immunoreactive complexes were visualized using SuperSignal West Pico Chemiluminescent substrate from Pierce (Rockford, IL, USA) on Eastman Kodak (Rochester, NY, USA) BioMax ML film. Data were quantitated from scanned films using Un-Scan-It software (Klinge et al. 2001).

GST-RID fusion protein preparation

The plasmids pGEX-2-TK-SRC-1 (219–399), pGEX-2TK-TIF2 (623–986), and pGEX-2TK-AIB1 (522–82) were graciously provided by Drs M Muylan and R Hilf of the University of Rochester (Yi et al. 2002). The plasmid pGEX-4T-1-NCoR was graciously provided by Dr A Hollenberg of Harvard University (Cohen et al. 2000). BL21 E. coli were used for expression of the GST-RID fusion proteins which were purified by GSH-Sepharose (Amersham Biosciences, Piscataway, NJ, USA) affinity chromatography as described previously (Klinge et al. 1997).

EMSA

Protein-DNA binding was measured by EMSA as previously reported (Klinge et al. 2001). Identical molar amounts of baculovirus-expressed human ER or ERß1, based on hydroxyapatite (HAP) assay results, were incubated with 150–500 fmol ³²P-labeled 50 bp oligomers EREc38, EREc13, pS2, PR1148, and Fos-1211. Binding reactions included 40 mM Tris–HCl (pH 7.5), 10% glycerol, 0.75 µg/µl BSA, and 0.02 µg/µl poly d(I–C) (Midland Certified Reagents, Midland, TX, USA), 111 mM KCl, 1 mM EDTA, 1 mM dithiothreitol, and 0.5 mM phenylmethylsulfonyl fluoride. ER-and ERß-specific antibodies (G-20 and Y19 from Santa Cruz Biotechnology, Santa Cruz, CA, USA) were included in selected reactions. Dried EMSA gels were analyzed using a Packard Instruments InstantImager (Meridian, CT, USA) and associated software, Packard Imager for Windows v2.04 (Tyulmenkov & Klinge 2001b).

	Results

Top Abstract Introduction Materials and methods Results Discussion References

 
ERE sequence impacts ER- and ERß-mediated transcriptional activity

Transient cotransfection assays were performed in CHO-K1 cells to determine how the nucleotide sequence of the ERE affects the enhancer activity of ER and ERß. This cell line was selected because it requires exogenous ER or ERß to activate ERE-driven reporter gene expression (McInerney et al. 1996) and thus allows evaluation of the transcriptional response of each ER subtype in isolation with each ERE. We selected five different ERE-driven luciferase reporter constructs to examine possible differences in transactivation potential between ER and ERß based on DNA sequence (see Materials and methods for sequences). EREc38 is a consensus ERE that includes the commonly used Xenopus vit A2 ERE palindrome (Klinge et al. 1992). EREc13 is the minimal consensus 13 bp palindromic ERE (Klein-Hitpass et al. 1988). The natural, non-consensus, imperfectly palindromic EREs were taken from the human pS2, c-fos, and PR genes (Kulakosky et al. 2002). The cells were treated with EtOH vehicle control, 10 nM E₂, 100 nM 4-OHT, or both, and luciferase and RL-luc activities were assayed (Fig. 1). For each of the four EREs tested, E₂ increased luciferase expression above the basal level for both ER and ERß. Luciferase activity detected in the presence of 4-OHT was either at or below basal expression. Co-treatment with E₂ and 4-OHT inhibited E₂-induced reporter activity, indicating that the E₂-induced activation is mediated specifically by ER and ERß. ER showed a greater E₂-induced fold-induction of reporter gene activity with the EREc38, EREc13, and pS2 EREs compared with ERß. However, the fold-induction of luciferase by ER and ERß was similar on the Fos-1211 and PR1148 EREs. These results indicated that the ERE sequence differentially impacts the E₂-dependent transcriptional activities of ER and ERß.

View larger version (38K): [in this window] [in a new window]   Figure 1 ERE sequence impacts ER and ERß transcriptional activity. CHO-K1 cells were co-transfected with (A) ER or (B) ERß plus the indicated pGL3-ERE-luciferase reporter and pRL-CMV as described in Materials and methods. Twenty-four hours after transfection, the cells were treated with EtOH, 10 nM E2, 100 nM 4-OHT, or 10 nM E2 plus 100 nM 4-OHT for 30 h. Cell extracts were prepared and assayed as described in Materials and methods. Data are displayed as luciferase activity divided by the RL-luc activity in each well and normalized by EtOH control (which is set to 1). Within each experiment, each treatment was performed in triplicate. The data shown are the means±S.E.M. from at least 11 separate experiments. *P<0.05, E2 values that are statistically different from the EtOH control value.

We have previously reported that the ERE nucleotide sequence altered ER conformation as assessed by protease digestion studies (Klinge et al. 2001, Tyulmenkov & Klinge 2001a). Here we tested the hypothesis that the observed ERE sequence-induced alterations in ER conformation may alter ER-coregulator interactions in vivo. This is the first study to directly compare coregulator interactions with ER and ERß in response to single copy EREs in the context of their natural flanking sequence in a functional assay. Further, although coactivators have been shown to impact ER and ERß activity in transfected cell lines using different ERE reporters, most prominently multiple tandem copies of the vit A2 ERE, few investigators have separated the effect of coregulators on the basal activity of ER on the reporter gene assayed versus that stimulated by E₂ (reviewed in Klinge 2003).

First, we examined the effect of coregulators on basal, ligand-independent ER activity by measuring luciferase expression from each ERE (Fig. 2). CHO-K1 contains endogenous SRC-1, GRIP1, ACTR, CBP, and SMRT (Western data not shown). To account for endogenous expression of coregulators, each ER-ERE combination tested was treated with ethanol in the absence of co-transfected coregulators. This control level of reporter activity was set to 1. SRC-1 and SRA increased basal ER activity in a concentration-dependent manner from EREc38 (Fig. 2A). CBP significantly stimulated basal activity only at 100 ng. PRMT1, CARM1, and NCoR stimulated basal ER transcriptional activity from EREc38 whereas SMRT repressed the basal ER activity in a concentration-dependent manner (Fig. 2B). Noteworthy is the fact that none of the coregulators affected the basal activity of ER or ERß on the parental pGL3-pro-luciferase vector without EREs (Fig. 2C and data not shown). None of the coregulators with the exception of CBP had any effect on either firefly luciferase activity from pGL3–1 EREc38 or the other reporter vectors or RL-luc activity. Since CBP did not stimulate luciferase activity from the pGL3-pro plasmid (Fig. 2C), despite the presence of one cyclic AMP response element binding protein (CREB) binding site in the luciferase gene and another in the ampicillain resistance gene (Promega; personal communication), the stimulation of luciferase activity from pGL3–1 EREc38 by CBP must be mediated by the inclusion of EREc38 which contains a cyclic AMP response element (CRE). Nonetheless, because our data are normalized for CBP stimulation on the ERE reporters in the absence of ER, the stimulation of ligand-independent ER basal activity by SRC-1, SRA, CBP, PRMT1, CARM1, and NCoR (Fig. 2A and B) is mediated by ERE-ER interaction. Based on the work of other investigators (Webb et al. 2003), we suggest that these effects of coactivators on unliganded ER may be mediated through the N-terminal activation domain (AF-1) which is ligand independent.

View larger version (61K): [in this window] [in a new window]   Figure 2 Coregulators influence ligand-independent ER activity. In (A and B) CHO-K1 cells were transfected with ER, EREc38-luciferase reporter, and the amounts of the indicated coactivators. In (C) CHO-K1 cells were transfected with ER, pGL3-pro-luciferase reporter (no EREs), and the amounts of the indicated coactivators. In (D–H) CHO-K1 cells were transfected with the indicated ERE-luciferase reporter, ER or ERß, and 250 ng of the indicated coregulators. All cells were co-transfected with RL-luc reporter control. Cells were treated with EtOH for 30 h and processed as described in Fig. 1 and in Materials and methods. Values are the fold-induction over EtOH activity in the absence of added coregulator and are the means±S.E.M. of three to eight different experiments. *P<0.05, values that are statistically different from the control value.

To address the role of the ERE nucleotide sequence in the effect of coregulators on ligand-independent ER and ERß, cells were transfected with EREc13, pS2, PR1148, and Fos-1211 EREs and a fixed amount of each coregulator (Fig. 2E–H). Each ERE showed both ER subtype- and coactivator-specific differences in transcriptional response. The most striking ligand-independent stimulation of ER and ERß activity was with SRA and SRC-1 on Fos-1211. ACTR did not stimulate either ER or ERß on any of the EREs. ER and ERß showed different responses to NCoR on EREc13 and pS2 and to PRMT1 and CARM1 on pS2.

To address the role of E₂ in mediating functional interaction of ER and ERß with coregulators, in all subsequent experiments cells were transfected with each coregulator and treated with EtOH and 10 nM E₂; the luciferase activity detected with E₂ was normalized by that for EtOH. Thus, any effect of a coregulator on basal, ligand-independent ER transcription was excluded from subsequent analysis. This allows comparison of the effect of a coregulator on ligand-activated ER transcription independent of effect of that coregulator on ligand-independent (basal) transcription from the reporter.

ERE sequence impacts coactivator-mediated transactivation by E₂-ER and E₂-ERß

Next, we compared how different ERE sequences affected the ability of coregulators to affect E₂-stimulated activity with ER or ERß. Figure 3A shows that of the tested coactivators, only SRC-1 stimulated E₂-induced ER activity on the vit A2 ERE (EREc38). Singly, none of the coactivators stimulated ERß activity. The combination of CBP and SRC-1 synergistically activated E₂-induced ERß but not ER activity. In contrast, both CARM1 and PRMT1 inhibited ER- and ERß-induced transcriptional activity (Fig. 3B). This result contrasts to the stimulation of basal activity of ER and ERß by CARM1 and PRMT1. Cotransfection of SRC-1 relieved PRMT1-mediated repression of ER, but not ERß activity. This result agrees with the ER-selective stimulation by SRC-1 shown in Fig. 3A. ACTR relieved PRMT1-mediated repression of both ER and ERß. GRIP1 did not relieve the PRMT1-mediated repression of either ER and ERß.

View larger version (53K): [in this window] [in a new window]   Figure 3 Transcriptional response of ER and ERß to coregulators on EREc38. CHO-K1 cells were transfected with ER or ERß and EREc38-luciferase reporter, pRL-CMV, and 250 ng of the indicated coregulator as described in Materials and methods. Cells were treated with EtOH, 10 nM E2, 100 nM 4-OHT, or both E2 and 4-OHT, as indicated, for 30 h and processed as described in Fig. 1 and in Materials and methods. The maximal activity with 10 nM E2 was set at 100 (absolute fold values are shown in Fig. 1). Values are the means±S.E.M. of three to eight different experiments. (A, B, D, and E) *P<0.05, values that are statistically different from the E2-induced value for that receptor subtype in the absence of coregulator. (C) P<0.05, values that are significantly different (Student’s t-test) between the 4-OHT activity without and with SMRT or NCoR.

Recently, the corepressors NCoR and SMRT were shown to interact directly with ACTR and facilitate thyroid receptor-ACTR binding in vitro and in a mammalian two-hybrid assay (Li et al. 2002). We tested the hypothesis that addition of SMRT or NCoR to cells transfected with ER or ERß and treated with E₂ would result in increased reporter gene expression. NCoR increased E₂-induced activity by ER only with the addition of 100 ng. In contrast, ERß activity was increased with 100 ng and higher amounts of NCoR (Fig. 3D). SMRT decreased E₂-induced activity by ER or ERß in a concentration-dependent manner that appeared to saturate for ERß at 100 ng (Fig. 3E). These data indicated that under the assay conditions used NCoR acts as a coactivator whereas SMRT acts as a corepressor of ER and ERß.

EREc13, the minimal consensus ERE (Klein-Hitpass et al. 1988), binds ER and ERß with significantly lower affinity than ERE palindromes containing 15 bp or more, i.e. K_d 1.1 nM and 1.7 nM versus 0.11 nM and 0.13 nM for ER and ERß respectively (Table 1 and (Kulakosky et al. 2002)). GRIP1, ACTR, and SRA increased E₂-ER activity on EREc13 (Fig. 4). In contrast, none of the coactivators stimulated E₂-ERß activity. These results differ from stimulation of unliganded ERß by SRA (Fig. 2E). However, the combination of SRC-1 and CBP stimulated E₂-induced ERß activity. NCoR stimulated E₂-induced ERß but not ER activity. These data indicated that ER and ERß show functional differences in coactivator interaction on a minimal ERE reporter in E₂-treated cells, a result concordant with differential interaction of small peptides mimicking coregulator interaction sequences with ER and ERß in vitro (Hall et al. 2000). Furthermore, comparison of the coactivator stimulation of ER and ERß on EREc38 versus EREc13 (Table 1) indicates that the length of the ERE palindrome differentially impacts ER and ERß interaction with coregulators, a result predicated by work using phage display to examine conformational differences between ER and ERß bound to different EREs (Hall et al. 2002).

View this table: [in this window] [in a new window]   Table 1 Effect of ERE sequence on E2-induced transcriptional activity by ER and ERß. Kd values are the means±S.E.M. of four to six separate EMSA determinations as described (Tyulmenkov et al. 2000, Klinge et al. 2001, Tyulmenkov & Klinge 2001b, Kulakosky et al. 2002). The effect of the indicated coactivator on E2-stimulated ER or ERß transcriptional activity is a summary of the relevant data from the transient transfection assays performed in CHO-K1 cells as shown in Figs 3–6

View larger version (28K): [in this window] [in a new window]   Figure 4 Transcriptional response of ER and ERß to coregulators on EREc13. CHO-K1 cells were transfected with ER or ERß, EREc13-luciferase, pRL-CMV, and 250 ng of the indicated coregulators as described in Materials and methods and Figs 1 and 3. Values are the means±S.E.M. of three to nine different experiments. *P<0.05, values that are statistically different from the E2-induced value for that receptor subtype in the absence of coregulator.

View larger version (45K): [in this window] [in a new window]   Figure 5 Natural variations on ERE palindrome sequence impact the effect of coactivators and corepressors on E2-induced ER and ERß transcription. CHO-K1 cells were transfected with ER or ERß and (A) pS2, (B) PR1148, or (C) Fos-1211 ERE-luciferase reporters plus pRL-CMV and 250 ng of each of the indicated coregulators as described in Materials and methods. (D) The impact of the addition of either NCoR or SMRT on E2-induced ER or ERß activity from the indicated ERE-luciferase reporter. Cells were treated with 10 nM E2, 100 nM 4-OHT, or both (as indicated) for 30 h and processed as described in Fig. 1 and in Materials and methods. The maximal activity with 10 nM E2 was set at 100% and the actual fold-induction values are shown in Fig. 1A and 1B for E2-ER and ERß respectively. Values are the means±S.E.M. of three to nine different experiments. *P<0.05, values that are statistically different from the E2-induced value for that receptor subtype in the absence of coregulator.

We also tested the effect of addition of NCoR or SMRT on E₂-induced reporter activity from the non-palindromic EREs (Fig. 5D). As with the data on coactivators, NCoR and SMRT differed in their effect on transcription in an ER subtype- and ERE sequence-dependent manner. NCoR decreased E₂-ER activity from pS2 and Fos-1211, but not PR1148. NCoR also decreased E₂-ERß activity from Fos-1211. NCoR enhanced E₂-ERß activity with PR1148. SMRT inhibited E₂-induced ER activity from PR1148 and ERß activity from Fos-1211 and PR1148. We concluded that SMRT and NCoR differentially impact E₂-induced transcriptional activity in an ER subtype- and ERE sequence-dependent manner.

A possible explanation for the differences detected between the effects of the coactivators on ER or ERß transcriptional activity is altered expression of the coactivators under different experimental conditions. To determine whether the expression of the coregulators was equal in transfected cells, Western blots were performed (Fig. 6). Densitometric analysis of the SRC-1, GRIP1, or ACTR/ß-actin ratio in repeated transient transfection experiments revealed no statistical differences in the amounts of coactivator expressed in cells transfected with either ER or ERß and that there was no effect of ERE sequence on the expression of transfected SRC-1, GRIP1, or ACTR in CHO-K1 cells. This obviates differences in coregulator expression as an explanation for the observed differences in transcriptional activity seen with the different ERE reporters and between ER or ERß.

View larger version (38K): [in this window] [in a new window]   Figure 6 Coactivator expression in transiently transfected CHO-K1 cells. CHO-K1 cells were transfected with the indicated amount of the expression vector, ER subtype, and ERE. Western blots were performed on cell extracts using antibodies against (A) SRC-1 and ß-actin, (B) GRIP1 and GAPDH, and (C) ACTR and ß-actin. These blots are representative of experiments that were repeated twice.

Synergism of combinations of coactivators with ER and ERß varies with ERE sequence

Coactivators including CBP, SRA, CARM1, and PRMT1 are recruited to target gene promoters through their interaction with p160 family coactivators interacting directly with ER and other nuclear receptors (NR) (Teyssier et al. 2002). To test whether SRA synergized with p160 coactivators to stimulate E₂-dependent transcription in an ERE sequence-dependent manner, CHO-K1 cells were transfected with either ER or ERß, plus SRA, alone or in combination with SRC-1, GRIP1, or ACTR with different ERE reporters (Fig. 7). SRA showed no synergism with ACTR (compare with SRA and ACTR alone in Figs 3A, 4, and 5). GRIP1 synergized with SRA for ER only on the Fos-1211 ERE (compare with GRIP1 and SRA alone in Fig. 5C). In contrast, SRA synergized with GRIP1 for ERß on EREc38, EREc13, and Fos-1211 (compare with data in Figs 3A, 4, and 5C). Lastly, SRA synergized with SRC-1 for ERß on EREc13 (compare with SRA and SRC-1 alone in Fig. 4).

View larger version (56K): [in this window] [in a new window]   Figure 7 Synergism between SRA and p160 coregulators depends on ER subtype and ERE sequence. CHO-K1 cells were transfected with ER or ERß and the indicated ERE-luciferase reporters plus pRL-CMV and 250 ng SRA plus each of the indicated p160 coactivators as described in Materials and methods and Fig. 1. The maximal activity with 10 nM E2 was set at 100. Values are the means±S.E.M. of three different experiments. *P<0.05, values that are statistically different from both the E2-induced value for that receptor subtype/ERE combination with either SRA or the indicated coactivator alone. The inset shows the relative activity of each coactivator alone with E2-ER or E2-ERß on the indicate ERE. These data are taken from Figs 3, 4, and 5.

Since coregulators showed differential effects on ER and ERß transcriptional activity in an ERE sequence-dependent manner in transfected cells, we tested whether the RIDs of selected p160 coactivators showed ERE sequence-dependent interaction with ER and ERß in EMSAs. GST-fusion proteins containing the RIDs of SRC-1, GRIP1, ACTR, or NCoR were incubated with E₂-occupied ER or ERß and different EREs in vitro. The sequence of the GST-RIDs and their location within the coregulator is shown in Table 2. As a negative control, each GST-RID was incubated with each ERE in the absence of ER. None of the GST-RIDs interacted with the EREs (Fig. 8 and data not shown). Figure 8 shows representative EMSAs for ER and ERß binding to the pS2 ERE with or without added GST-ACTR or GST-GRIP1.

View larger version (107K): [in this window] [in a new window]   Figure 8 The RID of ACTR, but not GRIP1, interacts with ER and ERß bound to the pS2 ERE in vitro. (A) Human ER was incubated with [32P]pS2-ERE as described in Materials and methods. GST-ACTR or GST-GRIP1 (0.5, 1, or 2 µg) were added to ER and [32P]pS2-ERE in lanes 2–4 and 6–7 respectively, as indicated. GST-ACTR and GST-GRIP alone did not bind pS2-ERE (lanes 5 and 9). Lane 10 included 1 µl ER-specific antibody G20. (B) Rat ERß was incubated with [32P]pS2-ERE as described in Materials and methods. GST-ACTR or GST-GRIP1 (0.5, 1, or 2 µg) were added to ERß and [32P]pS2-ERE in lanes 2–4 and 6–7 respectively, as indicated. GST-ACTR and GST-GRIP alone did not bind pS2-ERE (lanes 5 and 9). Lane 10 included 1 µl ERß-specific antibody Y-19. NS indicates non-specific binding of baculovirus proteins to the ERE; SS indicates the super-shifted complex formed between the ER antibody and the ER-pS2 complex. This autoradiograph is representative of two independent EMSA experiments for both ER and ERß showing similar results. (C and D) EMSA data were quantitated as described in Materials and methods. Values are the % of (C) ER or (D) ERß binding in the absence of added coregulator or antibody. %BD=ER-ERE bound; %SS=ER-ERE supershifted with the added GST fusion protein or antibody. Values are the average of two separate experiments±S.D.

View larger version (90K): [in this window] [in a new window]   Figure 9 Effect of GST-coregulators on ER-ERE binding in vitro. EMSA data were quantitated as described in Materials and methods. Values are the % of (A and C) ER or (B and D) ERß binding to (A and B) EREc38 or (C and D) EREc13 in the absence of added coregulator or antibody. %BD=ER-ERE bound; %SS=ER-ERE supershifted with the added GST fusion protein or antibody. Values are the average of two to three separate experiments±S.E.M.

View this table: [in this window] [in a new window]   Table 3 Comparison of the effect of the indicated coregulator on the transcriptional activity E2-ER or -ERß with the effect of the GST-RID of that coregulator on ER or ERß interaction with that of ERE in vitro in EMSA. Data are the summary of transient transfection assays shown in Figs 3–5 and three separate EMSA gels, examples of which are shown in Fig. 8, for each combination of ER-ERE and GST-RID

	Discussion

Top Abstract Introduction Materials and methods Results Discussion References

To separate the effects of ERE sequence on ER-coactivator interaction from the interaction of that coactivator with other transcription factors in a complex promoter, we used a simple model system in which each ERE, plus a relatively short stretch (6–20 bp) of nucleotides identical to those flanking that ERE in the natural human genes for pS2, c-Fos, and PR, was inserted in the same location in the multicloning region of a luciferase reporter gene. This obviates difficulties in data interpretation due to different distances between the ERE and transcription start site (Sathya et al. 1997, Nordeen et al. 1998) and focuses our study on the role of the ERE along with its natural flanking sequence in ER-coregulator interaction. Importantly, we evaluated ER-coregulator activity at a single copy of each ERE rather than multiple tandem copies of the vit A2 ERE. By using CHO-K1 cells that do not express ER or ERß (McInerney et al. 1996), we separated effects of coactivators and EREs on ER versus ERß. The importance of our study is that it is the first to examine ERß-ERE interactions in a functional assay in direct comparison with ER activity on the same single copy EREs in the context of their natural flanking sequence.

Our experiments have demonstrated that the sequence of the ERE impacts the functional interaction of ER and ERß with coactivators SRC-1, GRIP1, ACTR, CBP, and SRA, the coregulators CARM1 and PRMT1, and corepressors SMRT and NCoR in transiently transfected CHO-K1 cells. Notably, unliganded ER interacts with coregulators in an ERE sequence-dependent manner. Similar findings for CBP and SRC-1 were recently reported by others (Dutertre & Smith 2003). A brief summary model highlighting some of the ERE-specific differences between unliganded and E₂-occupied ER interaction with coregulators detected in our experiments is presented in Fig. 10. Although the affinity of ER or ERß binding to an ERE correlates with E₂-induced transcriptional activity (Klinge 2001), there is not a correlation between ERE occupancy (K_d) and the transcriptional response to the tested coactivators, coregulators, and corepressors in transfected CHO-K1 cells. Thus, the effects of coregulatory proteins on E₂-activated transcription is not governed solely by the affinity of ER-ERE interaction, but we suggest reflects ERE-mediated alterations in ER conformation (Klinge et al. 2001, Ramsey & Klinge 2001, Yi et al. 2002) that we propose impact ER-coactivator interaction. Hence, although we are unable to compose a simple rule or set of rules that will predict how an ERE sequence impacts ER-coactivator interaction in transfected cells, our data clearly demonstrated that variations within the core ERE sequence regulate coactivator interaction with liganded ER. An important difference between our study and most previous studies (reviewed in Klinge 2003) is that we subtracted the effect of each coregulator on the ligand-independent (basal) activity of ER or ERß for each ERE. Thus, the findings in Table 1 summarize how coregulators impact E₂-dependent ER and ERß transcriptional activity from different EREs. We suggest that small changes with individual coactivators may be physiologically significant when combined with the other coregulators in the cell. For example, a 50–100% increase in transcription of the transcription factors Fos or PR could effectively double the amount of Fos and PR proteins. Doubling PR expression subsequently could have a synergistic effect on downstream gene targets.

View larger version (29K): [in this window] [in a new window]   Figure 10 Model for ERE sequence-, ER subtype-, coregulator- and ligand state-dependent transcription by ER. Coactivators and corepressors regulate ER and ERß transcriptional activity depending on different conformations of the receptors in their unliganded versus E2-occupied as they bind to EREs from different genes. Please see Discussion for further details.

The pS2 ERE showed the most ligand-dependent ER responsiveness to the coactivators tested. Based on a report showing low interaction of ER-pS2 with SRC-1 and ACTR-RID peptides in vitro (Hall et al. 2002), we predicted that SRC-1 and ACTR would have minimal impact on E₂-ER activity on the pS2-ERE reporter. Indeed, SRC-1 enhanced E₂-ERß, but not ER transcription from the pS2-ERE. On the other hand, ACTR stimulated E₂-ER, but not ERß transcription from the pS2 ERE, despite the interaction of GST-ACTR with both ER and ERß bound to the pS2 ERE in vitro (Fig. 8). This difference from the aforementioned predication (Hall et al. 2002) suggests that amino acids in addition to the RID in ACTR impact transactivation by ER. This suggestion agrees with data showing that the three NR boxes from SRC-1, GRIP1, and ACTR have different affinities for E₂-occupied ER and ERß in vitro (Wong et al. 2001). Our cotransfection data are also in agreement with the observation that less GRIP1 was retained by ER bound to the pS2 ERE versus the vit A2 ERE in vitro (Wood et al. 2001).

As anticipated from the discussion above, data from our EMSA experiments using GST-RIDs did not fully correlate with results from the transient transfection assays. These divergent results are most likely due to the use of a single RID domain in the EMSA experiments and the use of full-length proteins in the transient transfection assays. Our EMSA data are in agreement with time-resolved fluorescence resonance energy transfer (FRET) experiments showing no interaction between ER and the first RID of SRC-1 and minimal recruitment to ERß in vitro (Bramlett et al. 2001). Similarly, our EMSAs and the FRET studies (Bramlett et al. 2001) indicate that the first RID of ACTR interacts with both ER and ERß. Further experiments are necessary to determine how full-length coactivators influence ER-ERE binding affinity.

Results from scintillation proximity and mammalian two-hybrid assays showed that the ligand binding domain (LBD) of ERß interacts with the RIDs of p160 coactivators in the following order of affinity (high–low): GRIP1>ACTR>SRC-1>>>p300/CBP (Northrop et al. 2000). These data suggest that E₂-ERß transcription would be expected to be stimulated most by GRIP1, less by ACTR and SRC-1, and the least with CBP. On the other hand, intact ERß interacted with comparable affinity with the immobilized NR boxes of SRC-1, GRIP1, and ACTR in vitro (Wong et al. 2001). Hence, the length of the ERß used influences coactivator interaction in vitro. In contrast to these in vitro results, we observed that GRIP1 stimulated E₂-ERß transcription only from the pS2 ERE. One possible explanation of these results is that GRIP1 requires a cellular cofactor(s) for the other EREs that is not expressed in CHO-K1 cells. However, GRIP1 did not enhance E₂-ERß activity on either EREc38 or pS2-ERE in HEC-1A human endome-trial cancer cells (C M Klinge, unpublished data), indicating that CHO-K1 are not unique in this response. Cells that express endogenous ERß should be tested to further evaluate the impact of GRIP1 on ERß activity.

ER interacts functionally with p300 and CBP (Shibata et al. 1997) which are ‘cointegrators’ because they form complexes with TBP and a variety of activator proteins (McKenna et al. 1999). CBP stimulated unliganded ER and ERß transcription at all EREs except PR1148. The finding that CBP enhanced E₂-ER transcription only from the PR1148 ERE is consistent with experiments showing that CBP shows weaker interaction with the ER LBD than SRC-1 in vitro (Heery et al. 2001). In contrast to our ER findings, CBP enhanced E₂-ERß transcription from pS2 and PR1148 EREs (Table 1). Ours is the first examination of the impact of CBP on ligand-dependent ERß-induced transcription. We conclude that CBP interacts differently with ER and ERß and that the ERE sequence impacts this interaction.

SRA is a unique ER coactivator because it functions as an RNA molecule in vivo and it enhances ER transactivation through AF-1 (Lanz et al. 1999). Ours is the first study to compare the activity of SRA with ER and ERß on natural EREs. Although a recent study reported that SRA enhanced the activities of E₂-ER and -ERß on a two-tandem vit A2 ERE-luciferase reporter in COS-1 or HEK-293 cells without affecting basal activity (Deblois & Giguere 2003), we observed that SRA stimulated basal, unliganded ER and ERß activity on all EREs with the exception of ERß on the pS2 ERE. SRA showed the greatest activity with unliganded ER and ERß on the Fos-1211 ERE. This finding warrants further investigation in future studies. Data showing that SRA interacts with p72 and p68 coactivators, which bind GRIP1 but do not interact with or activate ERß (Watanabe et al. 2001), imply that SRA may not stimulate ERß transcription. Our data are in agreement with this prediction: SRA did not enhance E₂-ERß transcription with the exception of the pS2 ERE, indicating that the ERE sequence impacts SRA-ERß interaction. Since SRA forms a complex with SRC-1 (Lanz et al. 1999), we evaluated the combined effect of the p160 coactivators and SRA on E₂-induced ER and ERß activity on different EREs. Our results indicated that both ER subtype and DNA sequence impact SRA activity.

CARM1 and PRMT1 interact with p160 family coactivators and stimulate NR transcription (Chen et al. 2000, Koh et al. 2001). PRMT1 stimulated unliganded ER and ERß activity on all EREs. CARM1 differed from PRMT1 in that it did not stimulate unliganded ER and ERß on the PR1148 and Fos-1211 EREs. The only E₂-dependent stimulatory activity of CARM1 was with E₂-ERß on the pS2 ERE. It will be important to determine if the ERE sequence impacts the methyltransferase activity of PRMT1 and CARM1.

In GST pull-down assays, corepressors NCoR and SMRT interact with ER in a ligand-independent manner (Smith et al. 1997, Lavinsky et al. 1998). Chromatin immunoprecipitation experiments revealed that NCoR interacts with ER in the promoters of the c-Myc and cathepsin D genes in MCF-7 breast cancer cells treated with tamoxifen or raloxifene (Shang & Brown 2002), indicating that these corepressors play a role in transcriptional silencing. Here we noted that NCoR increased unliganded ER and ERß activity from all but the pS2 and PR1148 EREs. Differences between native chromatin and partial chromatin structure on transfected reporter constructs may be responsible for this difference. The mechanism of stimulation of ER basal activity by NCoR is unclear; however, NCoR and SMRT interacted directly with ACTR, SRC-1, and GRIP1 in vitro and in transfected cells (Li et al. 2002). Further, NCoR enhanced unliganded TRß-ACTR interaction (Li et al. 2002). Whether NCoR facilitates ER-endogenous ACTR interaction in the absence of ligand is unknown. Further studies are needed to address the role of NCoR in fine-tuning transcriptional repression versus activation by ER and ERß.

In mammalian two-hybrid assays, NCoR had no effect on E₂-activated ER or ERß reporter expression (Zhang et al. 2000). In contrast, NCoR inhibited E₂-induced ER transcription only from the pS2 and Fos-1211 EREs in our cell-based assays. Therefore, the ERE sequence impacts NCoR-mediated transcriptional repression from a transfected plasmid. SMRT reduced basal and 4-OHT-stimulated reporter gene activity in HepG2 cells transfected with ER, but had no effect on E₂-activated reporter activity (Smith et al. 1997). Similarly, we found that SMRT inhibited basal transcription by ER or ERß, but SMRT inhibited E₂-induced transcription for both ER and ERß, although the effect was more