A Novel Estradiol/Estrogen Receptor α-dependent Transcriptional Mechanism Controls Expression of the Human Prolactin Receptor*

Prolactin exerts diverse functions in target tissues through its membrane receptors, and is a potent mitogen in normal and neoplastic breast cells. Estradiol (E2) induces human prolactin receptor (hPRLR) gene expression through stimulation of its generic promoter (PIII). This study identifies a novel E2-regulated non-estrogen responsive element-dependent transcriptional mechanism that mediates E2-induced hPRLR expression. E2 stimulated transcriptional activity in MCF7A2 cells transfected with PIII lacking an estrogen responsive element, and increased hPRLR mRNA and protein. The abolition of the E2 effect by mutation of Sp1 or C/EBP elements that bind Sp1/Sp3 and C/EBPβ within PIII indicated the cooperation of these transfactors in E2-induced transcription of the hPRLR. DNA affinity protein assay showed that E2 induced estrogen receptor α (ERα) binding to Sp1/Sp3 and C/EBPβ DNA-protein complexes. The ligand-binding domain of ERα was essential for its physical interaction with C/EBPβ, and E2 promoted this association, and its DNA binding domain was required for transactivation of PIII. Co-immunoprecipitation studies revealed tethering of C/EBPβ to Sp1 by E2-activated ERα. Chromatin immunoprecipitation analysis showed that E2 induced recruitment of C/EBPβ, ERα, SRC1, p300, pCAF, TFIIB, and Pol II, with no change in Sp1/Sp3. E2 also induced promoter-associated acetylation of H3 and H4. These findings demonstrate that an E2/ERα, Sp1, and C/EBPβ complex with recruitment of coactivators and TFIIB and Pol II are required for E2-activated transcriptional expression of the hPRLR through PIII. Estradiol produced in breast stroma and adipose tissue, which are major sources of estrogen in post-menopausal women, could up-regulate hPRLR gene expression and stimulate breast tumor growth.

moter in rodents (1,2) and has been implicated in the development of breast cancer. The hPRLR has several forms, including long form (stimulatory) and short form(s) (inhibitory) (3,4), which are expressed in normal and tumoral breast tissue and in most human breast cancer cells (5)(6)(7), and prolactin exhibits synergistic actions with estrogens on the proliferation of these cells (8). There is local production of PRL in mammary epithelial cells, and increased expression of the PRLR long form occurs in a significant number of human mammary tumors (9 -12). A lower ratio of short (inhibitory forms)/long (activating form) receptors reported in breast tumor tissues could cause unopposed prolactin-mediated stimulatory actions of the long form and may contribute to breast tumor development and progression (7). Moreover, the PRL antagonist G129R was reported to cause apoptosis in breast cancer cells (8,13). These findings, and the correlation between serum PRL and the incidence and progression of breast tumors (11,12,14), indicate that PRL has a role in human breast cancer. Stromal and adipose tissue are the major sources of estrogen in post-menopausal women, and could exert paracrine control of prolactin and prolactin receptor expression in adjacent mammary epithelial cells.
Our previous studies on the hPRLR gene revealed its complex 5Ј genomic structure, with multiple (six) alternative non-coding exons 1 and promoter utilization (15,16). These include the preferentially utilized, generic promoter 1/exon-1 (PIII/hE1 3 ), which is also present in rat and mouse, and five human specific exon-1/promoters (hE1 N1-5 ) (15)(16)(17). These forms were found to be expressed in breast cancer cells, and variably in other tissues (16). Quantitative competitive reverse transcriptase-PCR analysis showed that E 2 induced increases of PRLR non-coding exon-1 hE1 3 (generic) mRNA transcripts directed by promoter III (hPIII) in breast cancer cells. Also, in transfection studies E 2 activated the hPIII promoter (18). This promoter contains functional Sp1 and C/EBP sites that bind Sp1/Sp3 and C/EBP␤, respectively (16,19). The lack of a formal ERE in the hPIII promoter suggested that the effect of estradiol is mediated through association of the activated ER with relevant DNA binding transfactors. Thus, although there is not a classical ERE within the hPRLR promoter/5Ј-flanking region of hPIII, our initial studies demonstrated an estrogen regulatory role in hPRLR expression. In this study we have investigated control mechanism(s) underlying human PRLR gene expression, and characterized a novel non-classical ERE-independent mechanism by which estrogen regulates hPRLR gene expression.

Reporter Gene Constructs and Expression Vectors, Expression, and Purification of Glutathione S-Transferase (GST) C/EBP␤ Fusion Protein-
All plasmids were constructed by standard recombinant DNA techniques. The hPRLR PIII reporter pGL2 gene constructs have been previously * This work was supported by the Intramural Research Program of the NICHD, National Institutes of Health. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. 1 To whom correspondence should be addressed: ERRB, NICHD, National Institutes of Health, Bldg. 49 described (16). These include constructs of the pGL2 reporter gene plasmid (Promega, Madison, WI) with insertion of DNA fragment (Ϫ931/Ϫ112 bp) containing the 5Ј-flanking region/promoter/exon 1 of hPRLR hPIII or the hPIII promoter/exon 1 (Ϫ480/Ϫ112 bp), either wild-type or harboring mutations in Sp1 and/or C/EBP (within the hPIII) or the ERE half-site (upstream to the promoter) (Fig. 1A). The constructs were numbered relative to the translation initiation codon (ATG ϩ1) in exon 3 (E3) (Fig. 1A). All plasmid constructs were restriction mapped and sequenced. The hER␣/pcDNA 3.1 expression constructs containing specific deletions, and C-terminal fusion of a V5 tag (Fig. 6A) with inclusion of Kozak sequence were generated by conventional PCR. ER␣ and its truncated forms were synthesized in vitro using the TNT T7 Quick-coupled transcription/translation system (Promega). One-tenth of the individual reactions (5 l) were assessed as input of the pulldown analysis (45 l). The GST C/EBP␤ fusion expression construct was prepared by inserting full-length C/EBP␤ cDNA into the pET41a vector (Novagen, Madison WI) at EcoRI and XhoI sites in-frame with 5Ј GST.
Bacterial BL21a (DE3) pLysS strains transformed with pET41a or the pET41a/C/EBP␤-GST fusion construct were cultured at 37°C for 16 h (A 600 of 0.6 to 0.7). Cells were then incubated with 0.2 mM isopropyl-␤-D-thiogalactopyranoside (Invitrogen) for 1 h at 37°C. Cells harvested and lysed by sonication in B-PER bacterial lysis buffer were subjected to affinity purification using immobilized glutathione resin included in the B-PER GST Spin Purification Kit (Pierce).
GST Pulldown Assays-TNT-expressed ER␣/V5 and truncated/V5 forms (45 l of reaction mixture) were subjected to pulldown assays by incubation with GST (control) or GST-C/EBP␤ (attached to glutathione-Sepharose beads) for 16 h at 4°C in the presence or absence of 100 nM E 2 in a total volume of 500 l of the binding buffer (50 mM NaCl, 50 mM Tris, pH 8.0, 0.5 mM dithiothreitol, 0.05% Nonidet P-40 and proteases inhibitors mixture). Beads were collected by centrifugation and washed five times in 500 l of washing buffer (50 mM Tris, pH 8.0, 100 mM NaCl, 0.1 Nonidet P-40 and protease inhibitors mixture). Washed beads suspended in 20 l of sample buffer (Tris glycine, 2ϫ SDS, Invitrogen) were denatured at 100°C for 5 min. The soluble fraction was recovered by centrifugation, and resolved in 4 -20% Tris glycine gels (Invitrogen). Interactions were evaluated by Western blot analysis using V5 antibody and quantitated by densitometry. -Fold increases induced by E 2 were derived from bound values (percent of input). GST was used as a control. All experiments were performed at least three times. Total cell and nuclear extracts from HCC1806 breast cancer cells transfected with wild-type ER␣ or deletion constructs with V5 tag were assessed by Western blots for expression of constructs.
Cell Culture and Transient Transfection-MCF-7A2 cells (MCF-7) (20) were a kind gift from Dr. Erica Berleth, C. Roswell Park Cancer Institute, Buffalo, NY. HCC 1806 breast cancer cells (ER negative expression) were obtained from American Type Culture Collection (Manassas, VA). Both cell lines were maintained in RPMI 1640 medium supplemented with 10% fetal bovine serum (Invitrogen). Cells were cultured in six-well plates under very low serum and steroid-free conditions in phenol red-free media without supplemental growth factors with 5 and 1% dextran charcoal-treated fetal bovine serum for 3 days at each serum concentration. MCF-7 cells were transiently co-transfected with 0.8 g of wild type, various mutant constructs of PIII, and the 5Ј-flanking region of hPRLR with the reporter PGL2 gene or empty vector (basal control). HCC 1806 cells plated were cotransfected with wild-type PIII (Ϫ480/Ϫ112 bp) reporter pGL2 construct (0.5 g), wildtype ER␣ and deletion constructs (1 g), and 0.2 g of pCMVSport/␤galactosidase construct (for normalization) using the Lipofectamine Plus reagents following the procedures recommended by the manufac-turer (Invitrogen). For Western analysis of total and nuclear expression of wild-type and deleted ER␣ constructs, 10-cm culture plates were used for transfection and all plasmid additions were increased 5-fold. After incubation for 5 h, cells were treated with 17␤-estradiol (Sigma) (100 nM in 0.01% ethanol), and the ER antagonist ICI 182,780 (ICI), 5 M in 0.1% Me 2 SO (Fig. 1), or the highly selective ER␣ antagonist, MPP dihydrochloride (1,3-bis(4-hydroxyphenyl)-4-methyl-5,[4-(2 piperidinylethoxy)phenol]-1H-pyrazole dihydrochloride) (1 M) (Tocris, Ellisville, Mo), and 0.01% ethanol with or without 0.1% Me 2 SO as vehicle control in fresh medium. Cell lysates were prepared after 24 h treatment, and luciferase assays were performed using luciferase protocol (Promega) and transfection efficiency was normalized by ␤-galactosidase activity. The results are expressed as mean Ϯ S.E. from at least three separate experiments in triplicate.
Real-time PCR Quantitation of hPRLR mRNA-Total RNA was isolated from MCF-7 cells treated with 17␤-estradiol (0 -100 nM) for the times indicated, using the RNA isolation kit (Stratagene). Prior to reverse transcription reaction, total RNA was treated with DNase I to remove any possible copurified DNA. 2 g of DNA was reversed transcribed using a SuperScript II kit (Invitrogen) containing a mixture of oligo(dT) 20 and random hexamer primers. The first-strand DNA from 100 ng of RNA was used as a template in real-time PCR with SYBR Green Master Mix and an ABI 7500 sequence detection system (Applied Biosynthesis). The cycling program was set as follows: denature at 95°C for 10 min, followed by 40 cycles of 95°C for 15 s and 60°C for 1 min. The primers utilized for PCR to detect hPRLR mRNA transcribed from promoter hPIII were a specific forward primer derived from non-coding exon 1 (hE1 3 ) (Ϫ135/Ϫ114); 5Ј-TGAT-GTGGCAGACTTTGCTCCC-3Ј, and a reverse primer 5Ј-CCATTCA-GAAGGCAGGTGTTGAG-3Ј from sequences located in common hE3 (ϩ49/ϩ71). The primers for detecting ␤-actin were 5Ј-CTGGCACCA-CACCTTCTACAAT-3Ј and 5Ј-AATGTCACGCACGATTTCCCGC-3Ј. The specificity of the PCR products was verified by melting curve analyses at the end of the PCR. The standard curves were created by a 10-fold serial dilution of the pcDNA/PRLR vector. Results presented are from two individual experiments, each sample was assayed in triplicate, and normalized to the level of ␤-actin mRNA and expressed as -fold change from controls.
Preparation of Nuclear Extracts and Western Blot Analysis-Nuclear extracts from MCF-7 and HCC1806 cells treated with 100 nM 17␤-estradiol alone and/or 1 M MPP or 5 M ICI 182,780, or vehicle control for 18 h, were prepared as described previously (21,22). 30 g of nuclear protein was separated on 10% NuPAGE (MCF-7) or 4 -20% Tris glycine gels (HCC1806) (Invitrogen) and transferred to nitrocellulose membranes that were then incubated with primary antibodies to Sp1, Sp3, C/EBP␤, ER␣, SRC-1, p300, pCAF, RNA Pol II and TFIIB (Santa Cruz Biotechnology, Santa Cruz, CA), anti-AcH 3 and AcH 4 (Upstate Biotechnology). PRLR protein in cell extracts was assessed by Western blot using a PRLR monoclonal antibody (Affinity BioReagents Inc., Golden, CO). After adding the appropriate secondary antibodies, the immunolabeled bands were detected by enhanced chemiluminescence (Pierce) and the signals were recorded on x-ray films. Histone deacetylase-1 was used to evaluate the purity of the nuclear preparation and found to be unchanged by estradiol treatment.
Preparation of Nuclear Extracts Immunodepleted of ER␣-Antibody-conjugated agarose beads were used to deplete ER␣ from nuclear extracts of MCF-7 cells as previously described (22) with modifications. Briefly, 40 g of ER␣ polyclonal antibody (Santa Cruz) was incubated with 200 l of protein A-agarose beads in 1 ml of binding buffer (0.14 M NaCl, 0.008 m sodium phosphate, 0.002 M potassium phosphate, and 0.01 m KCl) for 2 h, and subsequently washed three times with PBST (PBS buffer, 0.02% Tween 20) buffer and twice with PBS buffer. The beads were then incubated with 100 g of nuclear protein for 2 h at 4°C with rotation. The suspension was centrifuged, and the supernatant was subjected to one more round of depletion by re-incubation for 1 h with 200 l of 50% slurry protein A-agarose bead antibody. The ER␣-depleted nuclear extracts were analyzed by Western blots to confirm the removal of ER␣ from the nuclear protein extracts.
DNA Affinity Protein Assay (DAPA)-DAPA were performed essentially as previously described (23). 5Ј-Biotin end-labeled sense and antisense oligonucleotides corresponding to the wild-type Sp1 binding site ( Ϫ375 CACTGACTCCTCCTCTCATGA Ϫ355 ) and its mutant (5Ј-CACTGAaTaCTaaTCTCATGA-3Ј, and to the C/EBP wild-type binding site ( Ϫ388 ATAAATGTTGCAACACTGACT Ϫ368 ) and its mutant (5Ј-ATAAATaccatAtaACTGACT-3Ј) of the hPRLR promoter III were custom made by GeneProbe Technology, Inc. (Gaithersburg, MD). The oligomers were annealed and gel purified by 12% polyacrylamide gel electrophoresis. 50 g of nuclear extracts were preincubated with 40 l of streptavidin-agarose (50% slurry from Invitrogen) for 1 h at 4°C with rotation. The supernatant collected by centrifugation was incubated with 0.2 g of wild or mutant biotin-labeled probe in binding buffer (60 mM KCl, 12 mM HEPES, pH 7.9, 4 mM Tris-HCl, 5% glycerol, 0.5 mM EDTA, 1 mM dithiothreitol, and 1ϫ protease inhibitor mixture) overnight at 4°C with rotation. DNA-protein complexes were then incubated with 40 l of 50% slurry of streptavidin-agarose (pre-equilibrated in the binding buffer for 1 h) overnight at 4°C with gentle rotation. DNA-protein complexes were washed five times with the binding buffer, the pellet was resuspended in 25 l of 2ϫ protein sample buffer (Invitrogen) and then boiled for 5 min to dissociate the complexes. The proteins were resolved by polyacrylamide gel electrophoresis, followed by Western blot detection with specific antibodies.
Chromatin Immunoprecipitation (ChIP)-ChIP assays were performed using the Chromatin Immunoprecipitation Assay Kit (Upstate Biotechnology, Lake Placid, NY) according to the manufacturers specifications and following previously described methodology (23). Briefly 5 ϫ 10 6 MCF-7 cells under different treatments for the times indicated in the corresponding experiments were used for each ChIP reaction. Cells were cross-linked with 1% formaldehyde and lysed in lysis buffer. Soluble chromatin was prepared by sonication following by immunoclearing. Immunoprecipitation was carried out at 4°C overnight utilizing specific antibodies to transcription factors (Sp1, Sp3, and C/EBP␤), coactivators (SRC-1, p300, and pCAF), acetylated histones H3 and H4, TFIIB, Pol II, and ER␣ and ER␤. Following immunoprecipitation, protein A-Sepharose beads and salmon sperm DNA were added. After several rounds of washing the precipitates were extracted three times with 1% SDS, 0.1% NaHCO 3 . After reversal of cross-linking by heating the samples at 65°C for 6 h, these were treated with proteinase K, followed by phenol extraction and ethanol precipitation. The purified DNA was analyzed by conventional PCR and real-time PCR for the presence of the human prolactin receptor PIII fragment Ϫ497/Ϫ321 bp using primers 5Ј-CTTCGCAGGATTCCAGCTCCCCCAAC-3Ј (forward), 5Ј-GAAGCTCAACTCGGTGCACTTGTTC-3Ј (reverse), and for the fragment Ϫ881 to Ϫ703 bp upstream to the promoter, primers 5Ј-TATGGCAGGAGAATAAACAC-3Ј (forward) and 5Ј-CATTGAA-CAGCAGCTATAT-3Ј (reverse) (Fig. 7). Titration of PCR cycles was performed to ensure that measurements were made in the linear range of amplification. PCR amplification of 1% of the soluble chromatin prior to immunoprecipitation was used as input control. The ChIP-precipitated DNA and input DNA were subjected to real-time PCR analyses using SYBR Green Master mixture in an ABI 7500 sequence detection system, and samples from two individual ChIP assays were analyzed in triplicate. The samples normalized by input were expressed as -fold increase over control untreated cells in basal conditions (C) designated as 1.
ChIP Re-immunoprecipitation-Complexes were eluted from the primary immunoprecipitation by incubation with 5 mM dithiothreitol at 37°C for 20 min and diluted with buffer (1% Triton X-100, 2 mM EDTA, 150 mM NaCl, 20 mM Tris-HCl at pH 8.1) followed by re-immunoprecipitation with a different relevant antibody. Subsequent steps of ChIP reimmunoprecipitations were as for the initial immunoprecipitations.
Co-immunoprecipitation (Co-IP)-100 g of nuclear protein prepared from MCF7 cells were initially subjected to preclearing by incubation with 40 l of protein A-agarose (50% of slurry) and 2 g of normal rabbit or mouse immunoglobulin G (IgG) in the immunoprecipitation assay buffer (1ϫ PBS, 0.5% Nonidet P-40, 0.5% sodium deoxycholate, 0.1% SDS) for 30 min at 4°C with gentle agitation. The recovered supernatant was incubated with 2 g of an antibody to a member of the complex for 2 h at 4°C in the presence of 1ϫ protease inhibitor mixture. Then, 50 l of protein A-agarose in 50% slurry was added, and the incubation was continued for overnight. Protein A-precipitated protein complex was recovered by brief centrifugation, followed by three times washes with immunoprecipitation assay buffer. The harvested beads resuspended in 25 l of 2ϫ protein sample buffer containing 5% of ␤-mercaptoethanol were boiled for 5 min to release the bound protein. The samples were then analyzed by Western blot with a specific antibody to another member of the complex. For Co-IP of ER␣ with C/EBP␤ we utilized ER␣ antibody cross-linked to protein G-agarose beads to immunoprecipitate ER␣ employing the Size X Protein G Kit (Pierce). This avoided masking of the C/EBP␤ band (45 kDa) by the immunoglobulin heavy chain dissociated from the immunoprecipitates.

Identification of Sites Critical for E 2 -induced Transcriptional Activation of the hPRLR Gene-Previous
studies demonstrated that the hPIII promoter (Ϫ480/Ϫ112) contains C/EBP and Sp1 functional elements that bind C/EBP␤ and Sp1/Sp3 and that both contribute to basal transcriptional activity (16). Furthermore, 5Ј-flanking regions to the promoter did not influence basal transcriptional activity in T-47D and MCF-7 cells (16). In addition E 2 was found to activate PIII in T-47 D cells transfected with the hPIII-luciferase construct (18). In this study we initially explored the functional domains within the hPIII promoter and 5Ј-flanking region involved in the E 2 activation of transcriptional activity in MCF-7 cells (Fig. 1). E 2 stimulated the hPRLR Ϫ480/Ϫ112 hPIII promoter/luciferase construct 6 -8-fold, and neither the addition of 5Ј-flanking sequences (Ϫ481/Ϫ931) nor the mutation of ER one-half element at Ϫ802 GGTCA Ϫ798 had effect in the E 2 activation of transcriptional activity (Fig. 1B). The E 2 activation was inhibited to basic construct levels by the addition of the E 2 receptor antagonist ICI 182,780 to the cultures (Fig. 1C). Mutation of either Sp1 or C/EBP sites inhibited E 2 activation to near basic control values in both the Ϫ931/Ϫ112 construct or in the Ϫ480/Ϫ112 promoter construct (Fig. 1, B and C). This indicated a requirement of cooperative effects of transfactors C/EBP␤ and Sp1/Sp3 in E 2 activation of hPRLR transcription through hPIII.
Estrogen Activated the Transcription of hPRLR and Caused Increases in mRNA and Receptor Protein Expression-The hPRLR promoter activity was dose dependently increased by E 2 treatment of MCF-7 cells for 24 h (Fig. 2A). Induction of promoter activity was observed with 1 nM E 2 (2-3-fold), and a 6 -8-fold increase was observed with 100 nM E 2 ( Fig.  2A). In temporal studies, 100 nM E 2 -induced increases in promoter activity were initially observed at 6 h (1-fold), and continue to increase at 12 (3-4-fold) and 24 h (6 -8-fold) (Fig. 2B). In parallel studies, we evaluated whether the endogenous expression of the hPRL gene governed by its natural promoter could be regulated by E 2 . Real-time PCR analysis of RNA from E 2 -treated and -untreated MCF-7 cells demonstrated a dose-dependent increase of mRNA levels (Fig. 2C). These results paralleled those obtained for the activation of promoter activity (Fig. 1A). A significant activation of endogenous hPRLR exon 1 (E1 3 ) gene transcripts directed by the hPIII promoter was observed following 6 h incubation of the cells with 100 nM E 2 (2-fold), and further increases were observed at 12 and 24 h (Fig. 2D). Furthermore, PRLR protein doserelated increases were induced by E 2 (Fig. 2E) and these were observed at 12 and 24 h of treatment with 100 nM estradiol (Fig. 2D). Our results showed clear correlation between the transcriptional responses of the hPRLR gene to E 2 and demonstrated that hPRLR transcription markedly induced by E 2 caused significant activation of hPRLR gene expression in MCF-7 cells.
Interaction of ER␣ with C/EBP␤ and Sp1/Sp3 in the Activation of hPRLR Gene Transcription-Our results from transient transfection studies indicated that DNA-bound Sp1 and C/EBP␤ have a central role in E 2 -induced hPRLR gene expression through the hPIII promoter by interacting with ER protein (Fig. 1). To determine the nature of their participation we conducted DAPA using biotin-labeled C/EBP and Sp1 wild-type and mutant double-stranded sequences of hPIII PRLR as probe and nuclear extracts from control and cells treated with 100 nM E 2 , in the presence or absence of the estrogen receptor antagonist ICI or ICI alone. Western blots of nuclear extracts utilized showed no changes for Sp1/Sp3 and C/EBP␤ nuclear protein with the various treatments (Fig. 3A, left). In contrast, a major specific increase of ER␣ by the E 2 treatment, which was prevented by co-treatment with the E 2 receptor antagonist, was observed. ICI inhibits the actions of endogenous and exogenous estradiol and consequently reduced ER␣ in the nuclear preparations.  (16). Exons 3-10 are utilized for translation of the long and intermediate form of PRLR; in addition, exon 11 is utilized for translation of the short forms of the receptor S1a and S1b. Exon 10 sequences are not present in the S1b short form (4). B and C, effect of E 2 on transcription of the hPRLR gene directed through the hPIII promoter. Shown are representations of the DNA fragments (Ϫ931/Ϫ112 bp) and (Ϫ480/Ϫ112 bp) used in the preparation of expression constructs. These include the wild-type construct with sequences of promoter hPIII (Ϫ480/Ϫ358 bp) and relevant cis-elements, indicated by symbols, and most of non-coding exon-1, hE1 3 (Ϫ357/Ϫ112 bp). 5Ј-Flanking sequences (Ϫ931/Ϫ481bp) to the promoter are included in the constructs in B. Constructs of wild-type hPRLR in pGL2 vector and mutants (mutated elements indicated by X symbols) were transiently transfected/expressed in MCF-7 cells. Also, cells were transfected with promoterless vector pGL2 (Basic). After transfection MCF-7 cells were treated with and without 100 nM E 2 for 24 h in the presence or absence of 5 M ICI, or ICI alone. Relative luciferase activities were normalized by the activity of co-transfected ␤-galactosidase and expressed as -fold over control (untreated cells) for the constructs indicated. Mutated sequences are indicated below the sequences.
DAPA utilizing these nuclear protein preparations were used to determine whether bound Sp1/Sp3 and CEBP␤ targeted ER␣ to the complex (Fig. 3A, middle and right). The expected binding of Sp1, Sp3, and C/EBP␤ to the wild-type but not the mutated hPIII promoter sequences was revealed by their respective antibodies and their absence when their specific DNA sequences were mutated. This is consistent with findings from electrophoretic mobility shift analysis that showed Sp1/Sp3 and CEBP␤ binding to their cognate sequences of PIII transfected in T47D and MCF-7 cells (16). Moreover, nuclear protein from cells treated with E 2 revealed association of ER␣ but not ER␤ to the Sp1/Sp3 and C/EBP␤ transfactors bound to their respective binding elements. No differences from control were, however, observed when nuclear extracts of cells treated with ICI (not shown) or ICI/E 2 were employed. These results demonstrated association of ER␣ to Sp1/Sp3 and C/EBP␤ bound to their cognate binding sites through non-ER dependent complex formation. Furthermore, C/EBP␤ was found to associate to Sp1/Sp3 bound DNA in the absence of its binding element (Fig. 3A, middle) and conversely Sp1/Sp3 associated with C/EBP␤ bound in the absence of its element (Fig. 3A, right). Such association, however, required that one of the transfactors (Sp1/S3 or C/EBP␤) is bound to its binding element to be detected in DAPA. This association was revealed with the addition of nuclear extracts of cells treated with E 2 but was not present/or minimally present in controls and was markedly reduced by ICI. These findings indicate that in addition of the mutual recruitment of ER␣ to bound Sp1/Sp3 and C/EBP␤, ER␣ could link both transfactors when only one of them is bound to DNA. Thus, it is proposed that E 2 -induced transcription of hPRLR results from E 2 -activated ER␣ recruited to Sp1 and C/EBP␤. Also, the possible inclusion in the complex of coactivators known to associate with ER␣ was investigated. Western blots (Fig. 3B, left) demonstrated that E 2 did not affect the protein levels of coactivators (SRC1, p300/CBP, and pCAF). However, an increased association of coactivators presumably to ER␣ recruited through the Sp1 or C/EBP␤ bound to their respective elements was observed (Fig. 3B, right). Such interactions were specific because they were only observed when utilizing nuclear preparations of cells treated with E 2 that contain high concentration of ER␣ (Fig. 3A, left), and only small changes over control were found in nuclear preparations from cells treated with the ICI, which contain only trace amounts of ER␣.
We subsequently performed co-immunoprecipitation assays to further characterize the protein-protein interaction within the complex (Fig. 4), C/EBP␤ and ER␣ were both co-precipitated by the specific antibody against Sp1 (left), similarly Sp1/Sp3 and ER␣ were co-precipitated by C/EBP␤ antibody (middle), and Sp1/Sp3 and C/EBP␤ by the ER␣ antibody (right). In all cases interactions were observed when using nuclear extracts of cells treated with E 2 , whereas minor or no interaction was present when utilizing control or nuclei preparations of cells treated with ICI. This confirms our initial proposal about the nature of the core interacting members of the complex and the requirement of ER␣ in complex formation.
Subsequent studies were directed to determine whether ER␣ per se or E 2 -activated ER␣ was required for the formation of the complex. For these studies we performed DAPA analysis and co-IP using MCF-7 nuclear cell extracts immunodepleted of endogenous ER␣ and incubated with a constant amount of exogenously added recombinant ER␣ in the presence or absence of E 2 . DAPA demonstrated that association of ER␣ to SP1 and C/EBP␤ bound to their respective elements was highly dependent on its activation by E 2 (Fig. 5A). This interaction was also demonstrated by co-IP where ER␣ in the presence of E 2 was coprecipitated by the Sp1 or C/EBP␤ antibody, whereas only a minor band was observed in the absence of the hormone (Fig. 5B). These results demonstrated that an activated ER␣ complex was required for its interaction with Sp1 and C/EBP␤.
A requirement of ER␣ to link Sp1 and C/EBP␤ within the complex was indicated in DAPA analyses utilizing nuclear extracts from cells treated with E 2 (Fig. 3A, middle and right). Furthermore, studies showed that hor- FIGURE 4. Determination of protein-protein interaction within the complex formed at the hPRLR promoter. Nuclear extracts prepared from MCF-7 cells cultured with or without 100 nM E 2 in the presence or absence or ER antagonist for 24 h were subjected to co-IP with Sp1 or C/EBP␤ antibody (left) or ER␣ antibody (right) followed by immunoblot using Sp1 rabbit polyclonal antibody and C/EBP␤ or ER␣ mouse monoclonal antibodies. The arrowheads indicate the immunosignals of transcription factors analyzed. Left, interaction of Sp1 with ER␣ and C/EBP␤ is markedly increased in the presence of E 2 . with E 2 (100 nM) or without E 2 (C) for 24 h in the presence or absence of ICI 182,780 or ICI alone. DAPA was performed using 5Ј-biotin-labeled wild-type Sp1 probe (Sp1) or mutant oligomer devoid of Sp1/Sp3 binding activity (Sp1 X) (A, middle) or using 5Ј-biotinylated labeled C/EBP (C/EBP) or mutant devoid of C/EBP␤ binding (C/EBPX) (A, right); or both wild-type probes (B, right). The probes were incubated with nuclear extracts prepared from MCF-7 cells treated with or without 100 nM E 2 for 24 h in the presence or absence of ICI 182,780. The avidin-precipitated protein complexes were analyzed by Western blots with immunodetection of bound transcription factors using antibodies against Sp1, Sp3, C/EBP␤, ER␣, ER␤ (A) and for coactivators SRC-1, p300, and pCAF (B). Wild-type and mutated probe sequences are shown in A. mone-activated ER␣ induced complex formation, because C/EBP␤ was recruited to the Sp1 bound DNA and Sp1 was associated to the C/EBP␤ in the presence of E 2 (Fig. 5C). Thus, E 2 /ER␣ through its binding to both C/EBP␤ and Sp1 bound DNA connects these essential transfactors for effective E 2 induction of transcription of the hPRLR through the hPIII promoter.
In further studies we explored the domain of ER␣ that is involved in the interaction with C/EBP␤. Seven deletion constructs of ER␣ with C-terminal fusion of a V5 tag were generated to analyze the proteinprotein interaction by pulldown assay with GST-C/EBP␤ (Fig. 6). One construct contained the N-terminal region of ER␣ that harbors the AF-1 domain but lacks the DNA-binding domain (DBD), hinge region (H), and C-terminal ligand-binding (LBD)/AF-2 domains (M1). Another deletion construct included most of the sequence of ER␣ (DNA-BD, H, and LBD/AF-2) with a truncation of the AF-1 domain (M2). An additional construct excluded the AF-1 and DBD and included the H-LBD/AF-2 (M3), another contained the LBD/AF2 domain (M4) and one contained only the LBD (M5) (Fig. 6A). Two additional constructs comprising the AF1, DNA-BD, H (M6) or AF1, DNA-BD (M7) domains were also employed. The AF1 construct (M1), and constructs containing DNA-BD, H (M6), or DNA-BD domains (M7), showed no interaction (Fig. 6C, right). In contrast, other constructs lacking either the AF1 domain, alone (M2) or with the DBD (M3), as well as the hinge region (M4) or all of these and also the AF2 domain, only containing the LBD domain (M5), showed basal interaction with C/EBP␤ that was markedly increased by E 2 (Fig. 6C, left). Furthermore, the observed differences of the ER␣/C/EBP␤ interaction or E 2 effect on the interaction were not related to variations in protein expression of the construct (Fig. 6B). Thus, these studies indicated that the LBD of ER␣ was responsible for its interaction with C/EBP␤.
Definition of the functional region(s) of ER␣ necessary for hPIII transcriptional activation through the complex C/EBP␤/Sp1 was subsequently pursued in cotransfection studies with ER␣ deletion constructs and the hPIII promoter/reporter gene construct in HCC 1806 breast cancer cells, which lack ER␣ expression. All constructs were found to have total cellular and also nuclear expression (Fig. 6, D and E). The nuclear expression of ER␣ constructs (M2-M5), which contained the LBD, was increased by E 2 treatment of cells. In contrast, the expression of contructs lacking the LBD (M1, M6, and M7) were unchanged by E2 treatment. All constructs with the exception of M1 (AF1 domain) contained the nuclear localization sequence(s) (24). These competently entered the nucleus. The relatively small size of M1 permitted its entry to the nucleus, probably by diffusion. In cells cotransfected with wildtype ER␣, E 2 caused a 3-4-fold increase in transcriptional activity. Expression of a construct lacking the DBD but bearing the AF1, LBD/ AF2, LBD, and hinge (H) regions did not transactivate the promoter. Only the M2 ER␣ deletion construct containing the DBD, the LBD interacting domain, as well as inactive functional regions (AF2 and H), caused E 2 -induced transcriptional activation of hPIII that was equivalent to the wild-type construct (Fig. 6F). Based on these findings, we conclude that sequences within the DNA-binding domain are also required for functional transactivation of the hPIII promoter by E 2 .
Endogenous Recruitment of Individual Components of the Complex on the PIII hPRLR Promoter-Our previous studies demonstrated that the hPRLR hPIII promoter binds Sp1 and C/EBP␤ through their elements (16). In this study, we provide evidence for the existence of a complex anchored by these two elements with activated ER␣ acting as a connector between these transfactors. Subsequently, the impact of E 2 /ER␣ on the endogenous recruitment of core transfactors to the complex and other factors/cofactors associated were investigated (Fig. 7). ChIP assays showed no apparent differences in the recruitment of Sp1/ Sp3 to the hPIII hPRLR promoter in the presence or absence of E 2 with or without addition of the ER␣-specific antagonist MPP (1 M) or MPP alone. However, recruitment of C/EBP␤ and ER␣ was highly induced by E 2 but was prevented by MPP (Fig. 7, A and C). The association of these transfactors in the complex was further shown in re-ChIP assays by the subsequent use of C/EBP␤ and ER␣ antibodies and in reverse order (Fig.  7D). No association of ER␤ was observed (not shown). The association FIGURE 5. A, effect of E 2 on the recruitment of ER␣ by Sp1 and C/EBP␤. B, effect of E 2 on ER␣ induced complex formation. C, linking of C/EBP␤ and Sp1 to the complex by E 2 -activated ER␣. A and C, DAPAs were carried out with incubation of MCF-7 nuclear extracts, depleted of ER␣ with 5Ј biotin-labeled wild-type Sp1 or C/EBP in the presence or absence of added recombinant ER␣ and E 2 . The avidin-precipitated complexes were subjected to Western blot analyses for immunodetection of ER␣ (A) C/EBP␤ or Sp1 (C ). Nuclear extracts depleted of ER␣, in the presence or absence of added ER␣ with or without E 2 were subjected to a co-immunoprecipitation assay with C/EBP␤ or Sp1 antibody followed by immunoblotting using ER␣ antiserum (B). Arrowheads indicate the ER␣ immunosignal and IgG (heavy chain). Pulldown assays of GST with ER␣ and its mutant proteins were used as negative controls (not shown). The complexes precipitated by immobilized glutathione resin were extracted and subjected to Western blot analyses for immunodetection with V5 antibody to reveal ER␣ and fragments associated with C/EBP␤. Estrogen effect on the interaction between C/EBP␤ and ER␣ wild-type and deletion constructs. -Fold increase induced by estradiol derived from bound values (percent of input). B and C illustrate one of three independent experiments. D-F, co-transfection studies of the wild-type hPRLR gene promoter/reporter gene construct (Ϫ480/Ϫ112 bp) with the wildtype and ER␣ mutant constructs in HCC 1806 breast cancer cells. D, Western blot analyses of total cell expression of co-transfected ER␣ constructs. E, Western blot analyses of the nuclear expression of cotransfected ER␣ constructs. F, luciferase activities were normalized by ␤-galactosidase activity and expressed as mean Ϯ S.E. of three independent experiments, each performed in triplicate.
of E 2 -activated ER␣ with the complex within the promoter region (Ϫ497/Ϫ321 bp) was clearly established. In contrast, no association of ER␣, transfactors, AcH3/H4, TFIIB, and Pol II with the Ϫ881/Ϫ773-bp DNA fragment containing the ER␣ half-site non-functional element was observed (Fig. 7B). This element and sequences 5Ј to the promoter were shown (Fig. 1) not to participate in E 2 -induced hPIII transcription (Fig. 1B). SRC-1, p300, and pCAF coactivators were recruited to the complex by E 2 /ER␣. E 2 treatment likely caused the marked increase of both acetylated H3 and H4 within the promoter domain, and such association was specific for this region (Fig. 7, A and C). The epigenetic changes induced by E 2 /ER␣ localization could provide a more accessible promoter environment for recruitment of components of general transcriptional machinery. Specific recruitment of TFIIB and RNA polymerase II to the hPRLR gene promoter hPIII was observed upon E 2 treatment .

DISCUSSION
These studies on the molecular basis of transcriptional regulation of the prolactin receptor by estradiol have demonstrated that estradiol increases reporter activity up to 8-fold in MCF-7 cells transfected with the PRLR PIII promoter containing Sp1 and C/EBP elements that bind transfactors Sp1/Sp3 and C/EBP␤, respectively. Also, that both of these functional elements are essential for E 2 action. Consistent with these findings, estradiol increased PRLR transcription directed by its PIII promoter in a dose-dependent manner. In the absence of an ERE, indirect effects of E 2 /ER␣ through interaction with Sp1 and C/EBP␤ bound to their cognate elements caused transcriptional activation of the hPRLR hPIII promoter in breast cancer cells. The assembly of an E 2 /ER␣, Sp1, and C/EBP␤ complex was required for transcriptional expression of the hPRLR through the PIII promoter in MCF-7 cells. E 2 /ER␣ increased recruitment of C/EBP␤ to the hPRLR promoter-linked C/EBP␤ with SP1 within the complex. Estradiol induced the association of ER␣ with C/EBP␤ through the LBD of ER␣, as well as recruitment of coactivators p300, SRC-1, and pCAF to the complex, with consequent region-specific changes in histone acetylation. These hormone/receptor-induced associations and chromatin changes favored TFIIB and RNA polymerase recruitment and the activation of hPIII-directed hPRLR transcription.
The actions of estradiol that regulate transcription of the PRLR gene through its PIII promoter do not involve direct binding of the agonistactivated ER to an ERE. This promoter is devoid of a classical ERE, and the ERE half-site located upstream of the promoter is not functional (Fig. 1). ER␣ and ER␤ are known to, respectively, stimulate or inhibit the expression of several genes indirectly without DNA association. The ER can alter transcription at specific sites by binding to transfactors AP1 and Sp1 (predominantly stimulatory) bound to their cognate elements in the DNA (25)(26)(27)(28). In addition, the ER can inhibit gene expression by abolishing the DNA interaction with or activity of NFB (29 -34). However, the ER effect is complex and depends on the ER subtype (ER␣ versus ER␤), the nature of the ligand (hormone, agonist/antagonist), the structure of the promoter including placement of the transfactor(s)-binding elements within the promoter for association, and the cell type. Previous studies have demonstrated E 2 -and antiestrogen-dependent activation of transcription via ER␣/Sp1 in breast cancer cells using GC-rich constructs (three tandem of Sp1) or E 2 -responsive GCrich promoters from the retinoic acid receptor or adenosine deaminase (28).
In our study, activation of the PRLR gene promoter PIII by E 2 through ER␣ is caused indirectly by its binding of the Sp1 and C/EBP␤ transcription factors. All E 2 effects on PIII, including binding to relevant transcriptional factors, transactivation, and recruitment to chromatin, were prevented by the addition of the ER antagonist ICI 182,780 or the highly selective ER␣ receptor antagonist, MPP dihydrochloride (35). DAPA studies demonstrated that the ER␤ present in nuclear extracts prepared from MCF-7 cells treated with E 2 , in contrast to ER␣, does not associate with Sp1 or C/EBP␤ bound to their respective DNA probes (Fig. 3). Also, and more importantly, E 2 did not cause recruitment of ER␤ to chromatin (Fig. 7A). Taken together, our results demonstrate an exclusive role of ER␣ in the effects of E 2 on the PIII PRLR promoter. It is also of interest that E 2 treatment of cells caused major increases in ER␣ protein present in nuclear extracts, from near-undetectable to high levels. Such changes were not found for ER␤, constant levels of which were present under all experimental conditions. In our previous study (16), we demonstrated by mutation analysis of Sp1 and C/EBP␤-binding elements that these sites contribute 80 and 50%, respectively, to the basal promoter activity of hPRLR. In this study, we have shown that both transfactors are essential for E 2 /ER␣ receptor transactivation, because mutation of either element (Sp1 or C/EBP) prevented E 2 /ER␣-induced activation of PRLR PIII. Our studies utilizing DAPA and co-immunoprecipitation with nuclear extracts of E 2 -treated cells provided strong evidence for an interaction of ER␣ with Sp1 and C/EBP␤ in the formation of a complex, and the association of coactivators with the complex (Figs. 3 and 4). Subsequently, we defined the mode of association of the proteins within the complex and the participation of E 2 therein. Using ER␣-depleted nuclear extracts of MCF-7 cells in DAPA and co-IP analyses with addition of equivalent amounts of exogenous ER␣, we defined a direct role of E 2 /ER␣ in the association of ER with Sp1 and C/EBP␤. This study has also provided evidence for the lack of direct interaction of Sp1 with C/EBP␤, although their two respective elements are separated only by 5 base pairs. In this regard, previous studies that demonstrated functional cooperation between C/EBP␣ or C/EBP␤ with Sp1 in the activation of the basal transcription of several genes including lactoferrin, CDC11c integrin, CYP2D5 P-450, and the human insulin receptor genes (36 -39) showed no evidence for direct interaction of these transfactors. The leucine zipper and activation domain of C/EBP␤ are essential for transactivation of the CYP2D5-P450 promoter (40). Repression of the interleukin-6 promoter by the estrogen receptor was shown to be mediated by NFB and C/EBP␤ and the functional interaction of the Rel elements of NFB and bZIP region of C/EBP␤ with the hinge region of ER (32). Nuclear receptors were shown to modulate the interaction of Sp1 and GC-rich DNA and thyroid hormone receptor up-regulated an Sp1-driven reporter in a ligand-dependent manner (41). Physical and functional interactions between the ER␣ and Sp1 protein linked to specific GC promoters have been shown to mediate transcriptional activation of several E 2 -responsive genes including c-fos, cathepsin D, retinoic acid receptor 1␣, adenosine deaminase, E 2 F1, bcl-2, and insulin growth factor-binding protein (28).
The LBD domain of ER␣ was identified in the present study as the interacting surface for its association with C/EBP␤. This association was present basally and was highly magnified by E 2 exposure, further indicating an E 2 /ER␣ interaction with C/EBP␤ (Fig. 6C). The ER␣ interacting domain for C/EBP␤ found in this study is distinct from the functional AF-1 domain site that is involved in its functional interaction with Sp1 (28). In this study, we found that the functional activation of the C/EBP␤⅐Sp1 complex by ER␣ to increase transcription of hPIII required the presence of the DBD and that this activation was only observed in the presence of E 2 . It is likely that ER␣ links the two transcriptional factors through their association at two distinct sites. The demonstration of a unique essential requirement for the cooperation of two DNA binding transfactors, tethered by a nuclear hormone receptor in its liganded configuration, provides novel evidence for a tandem-induced mode of activation of the hPRLR promoter.
The process of gene activation in eukaryotic cells is exceedingly complex. However, ChIP is a powerful technique that allows detection of endogenous transcription factors bound to gene promoters and 5Ј-flanking regions in vivo, and permits analysis of their regulated association. In addition to monitoring the direct interaction of DNA binding transfactors with the promoter, ChIP detects proteins that are not bound directly to DNA and depend on other proteins for their associa-tion with the promoter. Our data from ChIP analysis showed binding of Sp1/Sp3 to the PRLR promoter that was not influenced by E 2 (Fig. 7, A  and D), which is consistent with our findings derived from DAPA (Fig.  3). This indicates that Sp1/Sp3 constitutively associated with the hPRLR promoter contribute to basal levels of hPRL transcription. In contrast, E 2 induced the association of ER␣ and C/EBP␤ association with the promoter. E 2 /ER␣ presumably associates with C/EBP␤, bridging this transfactor to Sp1 and favoring interaction of C/EBP␤ with its cognate element. Formation of this complex promotes the recruitment of coactivators and epigenetic changes reflected in increases of acetylated histones H3 and H4. This presumably induces changes in chromatin organization that favor the association of TFIIB with component(s) of the complex and recruitment of Pol II to the PRLR promoter (Fig. 8).
Current evidence indicates that TFIIB serves as a bridge between the basal machinery and specific transactivators, and may also have a major role in transactivation. Interactions of TFIIB with transcription factors, including nuclear hormone receptors, have been reported (42)(43)(44)(45)(46)(47)(48)(49)(50). The ER was found to associate with various components of the preinitiation complex, including the TATA-binding protein, TFIIB, and its associated factor TAF II 30 (42,49), and to have direct interaction with TFIIB through its DBD (42,50). Recently, binding of TFIIB with Sp1 at the proximal Sp1 site of the TATA-less promoter of the luteinizing hormone receptor was found to be essential for transcriptional activity. However, this interaction appears to involve an adaptor protein (22). In addition p300, which we found to associate with the E 2 -activated ER␣/ C/EBP␤/Sp1, was shown to interact with the cyclin E-Cdk2 homology domain of TFIIB (51). Thus, it is conceivable that in the case of the PRLR gene, TFIIB interacts directly or indirectly with one or more members of the complex at the PRLR promoter. The E 2 -induced recruitment of TFIIB to the PRLR core promoter, in conjunction with components of the activating complex, ER␣ and C/EBP␤ and Sp1/Sp3 and recruited coactivators, indicates that TFIIB has an important role in the hormone-induced activation of PRLR transcription. In conclusion, these studies have demonstrated that Sp1/Sp3 and CEBP␤ are essential participants in the transcriptional activation of the PRLR by E 2 , revealing novel functions for these transcription factors, and that E 2 /ER␣ is the key effector in integrating the core activation mechanism in PRLR transcription.