Truncated Estrogen Receptor Product-1 Stimulates Estrogen Receptor α Transcriptional Activity by Titration of Repressor Proteins*

The truncated estrogen receptor product-1 (TERP-1, or TERP) is a pituitary-specific isoform of estrogen receptor α (ERα), and its expression is regulated by estrogen. TERP modulates the transcriptional activity of ERα but has no independent effect on transcription of estrogen-response element-containing promoters. At low concentrations, TERP stimulates ERα transcriptional activity in transient transfection assays. At TERP concentrations equal to or greater than full-length ERα, TERP forms dimers with ERα and reduces both ligand-dependent and -independent transcription. A dimerization mutant of TERP, TERP L509R, stimulated ERα transcription at all concentrations. We hypothesized that TERP stimulates ERα transcriptional activity by titrating suppressors of ERα activity. We found that repressor of estrogen receptor activity (REA), originally isolated from human breast cancer cells, is present in mouse pituitary gonadotrope cell lines. Levels of REA vary slightly throughout the rat reproductive cycle, but TERP mRNA and protein vary much more dramatically. In transfection experiments, REA suppressed ERα transcriptional activity, and TERP L509R was able to alleviate transcriptional suppression by REA. In glutathione S-transferase pull-down assays, TERP bound to REA more efficiently than did ERα at equivalent concentrations, suggesting that REA will preferentially bind to TERP. Our findings suggest that the stimulation of pituitary ERα activity by low concentrations of TERP can occur by titration of corepressors such as REA.

The truncated estrogen receptor product-1 (TERP-1, or TERP) is a pituitary-specific isoform of estrogen receptor ␣ (ER␣), and its expression is regulated by estrogen. TERP modulates the transcriptional activity of ER␣ but has no independent effect on transcription of estrogen-response element-containing promoters. At low concentrations, TERP stimulates ER␣ transcriptional activity in transient transfection assays. At TERP concentrations equal to or greater than full-length ER␣, TERP forms dimers with ER␣ and reduces both liganddependent and -independent transcription. A dimerization mutant of TERP, TERP L509R, stimulated ER␣ transcription at all concentrations. We hypothesized that TERP stimulates ER␣ transcriptional activity by titrating suppressors of ER␣ activity. We found that repressor of estrogen receptor activity (REA), originally isolated from human breast cancer cells, is present in mouse pituitary gonadotrope cell lines. Levels of REA vary slightly throughout the rat reproductive cycle, but TERP mRNA and protein vary much more dramatically. In transfection experiments, REA suppressed ER␣ transcriptional activity, and TERP L509R was able to alleviate transcriptional suppression by REA. In glutathione S-transferase pull-down assays, TERP bound to REA more efficiently than did ER␣ at equivalent concentrations, suggesting that REA will preferentially bind to TERP. Our findings suggest that the stimulation of pituitary ER␣ activity by low concentrations of TERP can occur by titration of corepressors such as REA.
The ligand-bound estrogen receptor (ER) 1 activates transcription and exerts its biological actions in a number of re-sponsive tissues such as uterus, mammary glands, brain, and pituitary (1)(2)(3). Like other nuclear receptors, the ER has five conserved structural domains. The N-terminal A/B region contains the ligand-independent activation function-1 (AF-1) and is the most variable in length and in sequence. It is thought to contribute to the ligand-independent activation of the receptor and is involved in promoter-and cell-specific activity (4 -6). The C-terminal ligand-binding domain (LBD) consists of 12 conserved helices. It contains the ligand-dependent activation function-2 (AF-2) and is also important in heat-shock protein binding, nuclear localization, and dimerization of the receptor (7). Both AF-1 and AF-2 domains contribute to the transcription activation of ER. The most conserved domain is the DNAbinding domain (DBD), which has two zinc fingers and recognizes specific DNA sequences, or estrogen-response elements (EREs), on hormone-responsive genes. The hinge region links the DNA-binding and ligand-binding domains and allows rotation of the DBD (7). Activation of the receptor by ligand requires a conformational change in the LBD in which helix 12 moves toward helix 3 and helix 4, forming a hydrophobic pocket and a surface that allows interaction with coactivators and other regulatory proteins (5,8). There are two subtypes of the estrogen receptor, ER␣ and ER␤, which are encoded by separate genes (2). The DBD is nearly identical and the LBD is similar, but the N-terminal regions of the receptor subtypes are different both in structure and in function. The AF-1 domain of ER␣ is very active in stimulation of transcription in different cell types, whereas the AF-1 activity of ER␤ is negligible under the same conditions (9). In the pituitary, ER␣ is the predominant isoform, and the AF-2 domain is the most important region for transcriptional activation (10,11).
Nuclear receptor activity is regulated by binding of ligands and coregulator proteins and the formation of heterodimers with other receptors. Coregulatory proteins may either enhance or suppress ER-dependent activity. Coactivators such as steroid receptor coactivator-1 (SRC-1) interact with nuclear receptors in an agonist-and AF-2-dependent manner (8,12) through their conserved LXXLL receptor interaction motif (12). These coactivators have been shown to activate nuclear receptors by chromatin remodeling via histone acetyltransferase activities or through direct interaction with the basal transcription machinery (12)(13)(14).
Corepressors negatively regulate transcriptional activity via histone deacetylase-dependent or -independent pathways (12,15). The nuclear receptor corepressor (NCoR) and silencing mediator for retinoid X receptor and thyroid hormone receptor (SMRT) suppress the transcriptional activity of unliganded nuclear receptors and ER bound to antagonists or selective estrogen receptor modulators through recruitment of histone deacetylase-containing protein complexes (13,16), contributing to chromatin compaction. A recently identified ER␣-associated protein, template-activating factor-1␤ (TAF-1␤), binds to the DBD and has been shown to inhibit acetylation of histones and ER␣, resulting in the suppression of transcription (17). Examples of repressor molecules that do not act directly through histone deacetylases include repressor of tamoxifen transcriptional activity (18), the LIM/homeodomain protein islet-1 (19), DEAD box RNA helicase DP97 (20), receptor-interacting protein 140 (RIP140) (21,22), and repressor of estrogen receptor activity (REA) (23). REA is the only ER-specific repressor protein known to date and was identified in an MCF-7 human breast cancer cell cDNA library by yeast two-hybrid screening. REA interacts with the ER␣ or ER␤ LBD in the presence of both agonists and antagonists (23,24). It does not have intrinsic transcription-repression activity and is thought to compete with SRC-1 binding to ER via the LXXLL motif at its N terminus (24).
ER transcriptional activity can also be modulated by the formation of heterodimers. For example, ER␤ influences overall ER activity in two ways. It may competitively bind to EREs and prevent ER␣ binding, or it may form ER␣/ER␤ heterodimers with lower transcriptional activity than ER␣ homodimers, possibly because of the less active N-terminal domain of ER␤ (25). An ER␣ isoform cloned in our laboratory, the truncated estrogen receptor product-1 (TERP-1, hereafter called TERP), has also been shown to form heterodimers with and modulate ER␣ and ER␤ activity (26,27). This pituitaryspecific isoform of ER␣ contains a unique untranslated exon 1 fused to exon 5 through exon 8 of ER␣ (26). The resulting protein lacks the DBD and does not bind ligand efficiently. TERP has no independent transcriptional activity on EREcontaining promoters but modulates ER transcriptional activity biphasically (27,28). In transient transfection assays, TERP stimulates ER␣ activity at low concentrations, where the ratio of transfected TERP to ER␣ is less than 1:1, but inhibits ER␣ activity at high concentrations, where the ratio of TERP to ER␣ is greater than 1:1 (28).
Levels of TERP mRNA change dramatically in response to estrogen (10,29,30), and it seems likely that TERP may have biological importance in the pituitary (31). TERP is transcribed from a promoter located in the intron between exons 4 and 5 of the rat ER␣ gene (32). The TERP promoter contains EREs, suggesting that ER may directly regulate TERP expression. The ratio of TERP/ER␣ mRNA varies during the reproductive cycle from undetectable during metestrus to 4-fold higher than ER␣ during proestrus, when estrogen levels are highest.
Previously, we and others demonstrated that the suppressive effects of TERP on ER␣ transcriptional activity, observed with TERP:ER␣ ratios greater than 1:1, is due to the formation of TERP/ER␣ heterodimers that inefficiently bind to an ERE (27,32). We hypothesized that the stimulatory effects of TERP observed at TERP:ER␣ ratios Ͻ 1:1 must occur through a separate mechanism. Here we show that low concentrations of TERP bind efficiently to the repressor molecule REA. We suggest that by the titration and sequestration of repressor proteins, TERP can stimulate ER transcriptional activity.

EXPERIMENTAL PROCEDURES
RNA Extraction-Cos-1, ␣T3, and L␤T2 cells were collected with guanidinium (4 M GdnSCN, 25 mM sodium citrate, 0.5% sarkosyl, and 0.1 M ␤-mercaptoethanol) and passed through an 18-gauge needle. The homogenate was layered on CsCl cushions (5.7 M CsCl and 0.01 M EDTA, pH 7) and centrifuged at 35,000 rpm at 20°C for 18 h (31). The pellet was removed and resuspended in 0.3 M sodium acetate, pH 5.5, and precipitated in ethanol overnight at Ϫ70°C. The precipitate was washed with 70% ethanol, dried, and rehydrated in sterile water.
Plasmids-ER␣, and TERP cDNA were generated by PCR and subcloned into a cytomegalovirus promoter containing pcDNA3.1 expression vector as described previously (27). PCR products of ER␣ and TERP were subcloned into the pGEX2T vector downstream of the glutathione S-transferase gene as described (27). Single point mutations of TERP at the dimerization domain (L509R) or at helix 12, the cofactor-binding site (E547K), were generated by site-directed mutagenesis as described (27). Mouse REA cDNA was generated by RT-PCR amplification from ␣T3 total RNA using REA-specific N-terminal and C-terminal primers (5Ј-ATGGCCCAGAACTTGAAGG-3Ј and 5Ј-CCTCATCAAGGGTAAGAAATGA-3Ј, respectively). The 899-bp PCR product was ligated to the PCR 2.1 vector (Invitrogen) and sequenced in both directions from three independent clones and then compared with the sequence for human REA isolated from human breast cancer MCF7 cells (GenBank TM accession number NM007273). REA was subsequently subcloned into the pcDNA3.1 expression vector and the pGEX-2T vector. The reporter vector for transfection experiments contains two copies of the vitellogenin consensus ERE fused to a 105-bp thymidine kinase promoter, (ERE)2-vit-TK-Luc.
RT-PCR-Reverse transcription from rat pituitary RNA was performed as described (31). PCR conditions and primers for ER␣, TERP, and actin were described (10), and the cycle number for their PCR reactions are 29, 39, and 26, respectively. PCR conditions for REA consisted of a single cycle of 3 min at 94°C followed by 34 cycles of 1 min at 94°C, 1 min at 56°C, and 2 min at 72°C and ending with a 10-min extension step at 72°C. The optimal cycle number for quantification of REA was determined based on PCR with 200 ng of female rat pituitary RNA over a range of 25-40 cycles. For each RNA sample, the linear range of the amplification curve was verified independently. For quantification, a minimum of two separate experiments and amplifications were performed for each RNA sample, and mRNA values from densitometric quantification were normalized for ␤-actin mRNA from the same samples and cDNA amplifications.
Cell Culture and Transfections-Mouse ␣T3 gonadotrope cells, originally obtained from Dr. Pamela Mellon (University of California, San Diego, CA) (33), were maintained in Dulbecco's minimal essential medium (DMEM) containing 10% fetal bovine serum, 100 units/ml penicillin, and 100 g/ml streptomycin. For transfection experiments, ␣T3 cells were plated in 30-mm wells at a density of 500,000 cells/well in phenol red-free DMEM with 5% charcoal-stripped newborn calf serum. The ER-negative monkey kidney Cos-1 cells were maintained in DMEM with 10% newborn calf serum. Cos-1 cells were plated in 30-mm wells at a density of 200,000 cells/well in phenol red-free DMEM with 5% stripped newborn calf serum in transfection experiments. 24 h after plating, cells were transfected with 1 g/well luciferase reporter plasmid, various expression constructs, and pcDNA3.1 vector (for standardization of total DNA) using calcium phosphate methods. Approximately 16 -18 h later, cells were washed with phosphate-buffered saline and incubated with or without 10 nM 17␤-estradiol (E 2 ) for 24 h before collection with 1ϫ cell lysis buffer (Promega Corp., Madison, WI). Luciferase activity was measured using a Turner 20e luminometer (Sunnyvale, CA) and normalized for total lysate protein assessed with Bio-Rad protein dye. Data were pooled from at least three independent experiments of duplicate or triplicate samples and are presented as arbitrary light units and normalized for protein levels.
GST Pull-down Assays-PGEX2T (GST alone), GST-ER␣LBD, GST-TERP, and GST-REA constructs were transformed into BL21 bacteria cells and grown in Luria broth containing 50 g/ml ampicillin in an orbital shaker at 37°C for 5 h. After induction with 0.1 mM isopropyl ␤-thiogalactopyranoside, the bacteria were grown overnight at 30°C. After centrifugation at 3,000 ϫ g for 30 min at 4°C, the bacterial pellets were resuspended and incubated on ice for 15 min in buffer containing 50 mM Tris, pH 7.5, 0.5 mM EDTA, 300 mM NaCl, 10 mg/ml lysozyme, and 1 mM dithiothreitol. After 20 l/ml of 10% Nonidet P-40 was added, the bacterial lysate was incubated on ice for 10 min and frozen at Ϫ70°C in an ethanol bath for 1 h. After thawing at room temperature, the lysate was resuspended and incubated on ice in buffer containing 1.5 M NaCl, 12 mM MgCl 2 , DNase I (5 g), 10 g/ml leupeptin, 1 g/ml pepstatin A, and 0.1 mM phenylmethylsulfonyl fluoride. Lysates were homogenized by passing through a 20-gauge needle and centrifuged at 10,000 ϫ g, 4°C for 30 min. Soluble lysate was conjugated with glutathione-agarose beads at 4°C overnight and assessed for protein concentration on 12% polyacrylamide denaturing gels by Coomassie Blue stain and immunoblotting (GST-ER␣LBD and GST-TERP) with ER␣ C1355 antibody (31). For pull-down experiments, ϳ0.5 g of GST alone or GST fusion proteins were incubated in a 150-l mixture containing 10 nM E 2 , 20 g/ml bovine serum albumin, [ 35 S]methionine-labeled (1,175 Ci/ mmol; 0.04 mCi/50-l reaction) in vitro translated proteins (TNT rabbit reticulocyte transcription/translation kit; Promega) and a GST wash buffer containing 10 mM MgCl 2 , 150 mM KCl, 20 mM HEPES, 10% glycerol, and 0.12% Nonidet P-40. After 1.5 h of incubation at 4°C, the beads were washed in GST wash buffer four times and resuspended in SDS loading buffer. The samples were boiled for 2 min prior to electrophoresis on SDS-containing 12% polyacrylamide gels at 145 V. Gels were dried and exposed to film overnight at Ϫ70°C and visualized by autoradiography. The protein band intensity was quantified by densitometry. Trichloroacetic acid precipitation was performed to estimate [ 35 S]methionine incorporation in the in vitro translated proteins. Briefly, 2 l of in vitro translated protein was incubated for 10 min with 98 l of 1 M NaOH at 37°C followed by a 30-min incubation with 25% trichloroacetic acid/2% tryptone on ice. The precipitate was washed on a GF/C filter with cold 5% trichloroacetic acid and 70% ethanol and assayed in a scintillation counter. For competition experiments, the molar ratio of incubated protein was calculated based on the [ 35  Quantification of Protein Expression-To quantify ER␣, TERP, and REA mRNA through the estrous cycle, intact female CD-1 rats (200 -225 g of body weight, Charles River Laboratories) were maintained on a 14:10-h light:dark cycle, and estrous cycle stage was determined by vaginal lavage. All animals were required to have two consecutive normal cycles prior to use in the study. Animals were euthanized at 9 a.m. and 5 p.m. of estrous and proestrus and at 9 a.m. on metestrus and diestrus (12-16 animals/group). Total pituitary RNA was prepared from pooled tissue (2-3/sample), and total protein was obtained by direct homogenization of individual pituitaries (three individual samples for each cycle day and time) in 50 mM Tris (pH 7.6) solution containing 2% SDS. Detection and quantitation of ER␣ and TERP protein were as described previously (11,31). Protein samples (75 g) were denatured by boiling with 2% ␤-mercaptoethanol and separated by electrophoresis on 12% polyacrylamide gels containing 1% SDS and then transferred to nitrocellulose membranes. Proteins were detected by the ER␣ C-terminal specific antibody C1355 (1:7,500) (31), peroxidase-labeled donkey anti-rabbit IgG, and the SuperSignal West Pico chemiluminescent detection system (Pierce). Full-length ER␣ and TERP protein were distinguished on the basis of size, quantitated by densitometric analysis, and normalized for ␤-actin protein in the same samples on the same blot as described previously (10). Data were analyzed from two separate experiments, with three samples/group. Animals were obtained and used in accordance with the guidelines established by the Animal Care and Use Committee of the University of Virginia.

Dimerization Mutant of TERP Stimulates ER␣-mediated
Transcription-Our laboratory has observed previously that TERP modulated ER␣ transcriptional activity biphasically (28). High concentrations of TERP were suppressive for ER activity through inactive dimer formation with ER␣ that disrupted ER␣ binding to an ERE. We speculated that if TERP could no longer form dimers, its stimulatory effect would be more apparent. A point mutation at leucine 509 of TERP disrupts its dimerization with ER␣. Although the protein of this mutant, TERP L509R, is made at the same level as the wild type, its ability to inhibit ER␣ is severely compromised (27). Cotransfection of ER␣ with increasing amounts of TERP L509R in ER-negative Cos-1 cells consistently stimulates ER␣ transcriptional activity on an ERE-containing promoter in a dose-dependent manner (Fig. 1A), whereas the same amount of wild type TERP is invariably suppressive (Fig. 1B). Similar results were observed in ER-positive mouse pituitary gonadotrope ␣T3 cells (Fig. 1C), a cell line in which TERP is normally expressed (10). We postulated that TERP L509R might be interacting with suppressive molecules and thus sequestering them from ER␣.
REA Is Expressed in Mouse Pituitary Gonadotrope Cell Lines and Suppresses ER-mediated Transcriptional Activity-We next examined the potential role of the ER-specific repressor protein, REA, in TERP regulation of ER␣ transcription. REA interacts with ER␣ at the LBD, most of which is contained in TERP. We first examined the expression of REA in gonadotrope cells. RT-PCR assays were performed on RNA extracted from mouse pituitary gonadotrope cell lines, ␣T3 and L␤T2, as well as monkey kidney Cos-1 cells, using primers designed from the published sequence of human REA (23). As shown in FIG. 1. TERP L509R stimulates ER-mediated transcription. In A, all Cos-1 cells were transfected with 1 g of (ERE)2-vit-TK-Luc luciferase reporter plasmid and 0.5 g of ER␣ expression vector in the absence or presence of TERP L509R at 0.5, 1, and 4 g. Cells were maintained in the absence (control, shown as Con) or presence of 10 nM E 2 for 24 h and assayed for luciferase reporter activity. rALU, relative arbitrary light units. In B, Cos-1 cells were transfected as in A with (ERE)2-vit-TK-Luc, 1 g of ER␣ expression plasmid, and 4 g of TERP or TERP L509R as indicated. Treatments were performed as in A, and luciferase activity was measured. Luciferase activity is expressed as arbitrary light units (rALU) normalized as in "Experimental Procedures" and is depicted relative to untreated controls set equal to 1. Data shown are the mean Ϯ S.E. from three (A) or five (B) independent experiments performed with duplicate or triplicate wells. In C, ␣T3 cells express ER␣ endogenously and were transfected with 1 g of (ERE)2vit-TK-Luc, with or without 4 g of TERP or 4 g of TERP L509R. After transfections, cells were treated and assayed for luciferase activity as in Cos-1 cells. Data shown are the mean Ϯ S.E. from four independent experiments performed with duplicate or triplicate wells for each treatment. *, p Ͻ 0.05 and ϩ, p Ͻ 0.01 versus E 2 -stimulated control. Fig. 2A, REA mRNA is expressed in both pituitary cell lines tested. Cos-1 cells also express REA (data not shown). Multiple clones of mouse REA were sequenced, and a BLAST sequence alignment with human REA indicated 91% similarity at the nucleic acid level and 100% at amino acid level. To examine the function of REA in pituitary cells, mouse REA was cloned and transfected into ␣T3 cells with an (ERE)2-vit-TK-Luciferase reporter construct. As shown in Fig. 2B, REA suppresses ER␣mediated transcriptional activity in ␣T3 cells. Similar results are observed in Cos-1 cells when REA is cotransfected with ER␣ (Fig. 2C). In addition, REA suppression is alleviated by cotransfection of TERP L509R, both in Cos-1 cells and ␣T3 cells (Fig. 2, B and C). The results suggest that TERP L509R stimulates ER␣-mediated transcriptional activity through interaction with and titration of repressor proteins such as REA.
TERP Binds to REA and Competes with ER␣ in Binding to REA-In order for TERP to influence ER␣ activity through REA, TERP must bind REA. Fig. 3A shows that GST-REA binds to in vitro translated [ 35 S]methionine-labeled TERP, but not to progesterone receptor, showing specificity of interaction. In vitro translated TERP is detected as a pair of doublets at 22-24 kDa, which likely represent translation from two methionines, at amino acids 393 and 408 in exon 5 (28,32). The larger protein is detected preferentially in transfected cells and in normal pituitaries (11,28,29,31,32). The dimerization mutants of ER␣ and TERP, ER␣ L509R and TERP L509R, also interact with GST-REA in a similar fashion (Fig. 3, B and C). These data indicate that monomers of TERP and ER␣ can also bind REA. A helix 12 point mutant of TERP, TERP E547K, is also capable of interacting with GST-REA (Fig. 3C), suggesting FIG. 2. Expression and functional analysis of REA. In A, RT-PCR was performed on RNA samples extracted from MCF-7 human breast cancer cells and ␣T3 and L␤T2 mouse pituitary gonadotrope cells, using REA-specific N-terminal and C-terminal primers. The PCR products were electrophoresed on a 1% agarose gel stained with ethidium bromide. Mrk, DNA size marker. In B, ␣T3 cells were transfected with 1 g of (ERE)2-vit-TK-Luc, 2 g of REA, and 2 g of TERP L509R as indicated. After transfection (16 h), cells were treated with E 2 for 24 h, and luciferase activity was assayed. In C, Cos-1 cells were transfected with 1 g of (ERE)2-vit-TK-Luc and 0.5 g of ER␣ expression plasmid in the absence or presence 1-2 g of REA and 0.5 and 1 g of TERP L509R as indicated. Data shown are the mean Ϯ S.E. from three independent experiments performed with triplicate wells for each treatment. *, p Ͻ 0.05 versus E 2 -stimulated wild type ER control. ϩ, p Ͻ 0.05 comparing TERP L509R stimulated with REA suppressed samples. that glutamic acid 547 in the region important for coactivator binding is not required for REA/ER␣ interaction in the rat receptor. If TERP sequesters REA physiologically, we expect that TERP would bind REA more strongly than ER␣. In GST pull-down assays, increasing amounts of TERP L509R decrease binding of ER␣ L509R to GST-REA (Fig. 4). The L509R mutants of both TERP and ER␣ bind to GST-REA similarly to the wild type TERP and ER␣. Thus, the dimerization mutants were used in competition pull-down experiments to prevent secondary binding so that TERP does not associate with REA through ER␣ or vice versa. Analysis of the GST pull-down data shows that at TERP L509R:ER␣ L509R ratios at 0.5:1 or greater, TERP L509R binds preferentially to REA and can displace ER␣ L509R from REA. Similar results were seen in three independent experiments, and studies performed with wild-type ER␣ or TERP also gave similar results (data not shown). The data suggest that TERP effectively competes with ER␣ in binding to REA and that titration of REA by TERP at low TERP concentrations can result in a stimulatory effect on ER␣-mediated transcription.
Expression of REA Compared with TERP throughout the Female Rat Estrous Cycle-Because the relative ratios of TERP, REA, and ER␣ will determine the overall ER␣ transcriptional activity in pituitary cells, we investigated whether REA expression changes in the female rat pituitary during the estrous cycle and how this relates to TERP and ER␣ expression during this time period. RT-PCR was performed on female rat pituitary RNA taken from different time points in the estrous cycle and normalized for ␤-actin levels. Protein levels of TERP and ER␣ were measured by immunoblotting and were also normalized for ␤-actin. As shown in Fig. 5, REA mRNA changes only slightly throughout the estrous cycle, peaking in the afternoon of proestrus and decreasing in the afternoon of estrus. The normalized level of REA mRNA is similar to that of ER␣ (data not shown), which also changes little during the cycle. ER␣ protein levels correlate tightly with mRNA levels and also change little throughout the cycle. In contrast, TERP mRNA and protein expression vary dramatically throughout the estrous cycle (Fig. 5B). TERP mRNA expression correlates with E 2 levels and is undetectable in the morning of metestrus and peaks by the morning of proestrus. TERP protein expression is greatest in the afternoon of proestrus. Overall, the TERP:ER protein ratio changes from 0.1 in the morning of metestrus to 4.3 by the morning of estrus. Because REA and ER␣ expression are relatively constant, the dramatic changes in TERP expression may be the more important determinant in the interaction among ER␣, TERP, and REA. DISCUSSION Our previous studies showed that the TERP:ER␣ ratio influences ER␣ transcriptional activity in a biphasic manner in transient transfection experiments (28). For TERP suppressive effects on ERE-containing promoters, TERP dimerization with ER␣ or ER␤ is required (27,34). We show in this study that the dimerization mutant of TERP (L509R) is stimulatory at all concentrations and mimics the effects of wild-type TERP at low concentrations. Because TERP L509R does not bind to DNA and cannot interact with ER␣, these data suggest that TERP is interacting with regulatory molecules involved in ER activation. These molecules can be saturated by TERP titration because the stimulatory effect on transcription plateaus at high concentrations of TERP L509R (Fig. 1A). We suggest that both TERP and TERP L509R recruit repressor molecules and make them less available to ER␣.
Cellular levels of coactivators and corepressors help to determine the ER␣ transcriptional activity and the tissue specificity of selective estrogen receptor modulators (35,36). Whereas the corepressors NCoR and SMRT bind ER only in the presence of antagonist, the recently identified repressor proteins may provide a more general way to influence ER␣ signaling because they bind to steroid receptors in the presence of either agonist or antagonist. Receptor-interacting protein 140 (RIP140) interacts with ER AF-2 in the presence of E 2 (21,22) and suppresses nuclear receptor-mediated transcription by competition with coactivator binding (37,38). Increasing expression of SRC-1 is sufficient to reverse the suppressive effect of RIP 140 (37,38). REA is similar to RIP140 in that it does not have intrinsic transcription-repression activity and suppresses ER activity through competition with SRC-1 binding via the LXXLL motif at its N terminus (24,39).
In this study, we show that cellular availability of the repressor REA, which is found in multiple cell types and tissues (23,40,41) including pituitary cells (Figs. 1 and 5), may be modulated by binding to or sequestration by TERP. Our protein binding data agree with previous observations that REA binding is ER-specific and that the LBD domain of ER␣ is the most important region for REA binding (23,39). Neither an intact dimerization domain nor an active helix 12 region in ER␣ and TERP is required for binding to REA. Most importantly, low levels of TERP and TERP L509R effectively compete with ER␣ and ER␣ L509R in binding to REA (Fig. 4). Differential binding efficiencies may be due to three-dimensional structure differences between ER␣ and TERP. It has been suggested that the C-terminal F domain of ER␣ prevents recruitment of REA to the unliganded receptor LBD and that deletion of the F domain increases REA binding (24). TERP lacks the N-terminal region of the LBD, and this structural modification may also contribute to enhanced TERP-REA binding relative to ER␣. The ability of TERP and TERP L509R to bind REA has functional consequences on ER␣ transcriptional activity. Increasing TERP L509R expression stimulates ER␣ transcription in both Cos-1 and ␣T3 pituitary gonadotrope cells and reverses the suppression by REA (Fig. 2, B and C). Stimulation by TERP L509R can also be reversed by REA (data not shown). TERP may bind other repressors as well. For example, the ER␣ repressor LIM/homeodomain protein islet-1 (ISL1) also binds to TERP, but the functional consequences are unknown (19).
TERP may be considered an orphan receptor, in that it is structurally similar to a known nuclear receptor (ER␣) but is inefficient or unable to bind ligand (26,28). At least two orphan nuclear receptors, short heterodimer partner (SHP) and DAX-1, have also been shown to modulate ER␣ activity (42,43). SHP is expressed in multiple cell types and contains a putative ligand-binding domain but lacks a DNA-binding domain. SHP does not interfere with ER␣ homodimerization but interacts with ER␣ through its central LXXLL-related motif and competes for coactivator binding to suppress ER␣ and ER␤ activities (44,45). DAX-1 is expressed in reproductive tissues and consists of a conserved LBD and a unique DNA/RNA-binding domain containing three leucine-rich repeats, one of which is the LXXLL motif. DAX-1 binds to both ER␣ and ER␤ via the LXXLL motif and is thought to suppress ER␣ action by preventing coactivator binding and recruiting the corepressor NCoR (43).
TERP is similar to SHP and DAX-1 in that it also forms heterodimers with ER␣, but through the dimerization helices 10/11 rather than through LXXLL motifs (27). Both SHP and TERP can interact with the coactivator SRC-1 (27,44). Whereas titration of SRC-1 is the predominant mechanism of SHP suppression of ER, TERP suppression is primarily via formation of heterodimers. This is because SRC-1 binding to liganded ER␣ is much greater than SRC-1 binding to TERP (27). In contrast, REA binds to TERP better than to liganded ER␣. It is not known whether SHP and DAX-1 bind to REA, although REA binding appears to be specific for ER␣ and ER␤ (23). Only TERP exhibits biphasic modulation of ER␣ activity, whereas SHP and DAX-1 are always suppressive (42,43,46). Physiological regulation of SHP and DAX-1 has not been shown, whereas levels of TERP expression vary with E 2 (10, 31) (Fig. 5).
Modulation of REA levels by hormones, as has been reported for SRC-1 and SMRT (47), could also contribute to changes in ER activity. However, we found that REA mRNA expression varies less than 2-fold throughout the estrous cycle and is similar to that of ER␣ (data not shown). In contrast, TERP expression changes dramatically in response to E 2 (26,31), during pregnancy (29), and throughout the estrous cycle (Fig.  5B). Thus, TERP can provide rapid and efficient regulation of ER activity in the pituitary. We propose a model in which ER␣ transcriptional activity is regulated by the differential expression of its tissue-specific isoform, TERP (Fig. 6). As TERP is expressed at levels lower than ER␣, TERP titrates the repressor protein REA from ER␣, allowing SRC-1 to bind and ER␣ activity to be stimulated. Higher concentrations of TERP will FIG. 6. Model for TERP modulation of ER␣ activity. The transcriptional activity of estrogen-bound ER␣ is dependent on its interaction with coregulator proteins such as SRC-1 and REA. SRC-1 binding increases ER␣ activity, whereas REA competes with SRC-1 for binding and will decrease ER␣ transcription. At low TERP levels (shown as TP) where the TERP:ER␣ ratio is lower than 1:1, TERP competes with ER␣ for available repressor protein REA, resulting in increased ER␣ transcription. At higher levels, TERP will still bind to REA but will also suppress ER␣ activity directly by forming inactive heterodimers that fail to bind DNA. also bind to and suppress ER␣ activity by formation of inactive heterodimers that inhibit DNA binding. Because the ratios of TERP:ER␣ vary with estrogen levels, both of these mechanisms could contribute to the physiological modulation of ER␣ activity.