|Home | About | Journals | Submit | Contact Us | Français|
B-cell Translocation Gene 2 (BTG2/TIS21/PC3) is an anti-proliferative tumor suppressor gene whose expression is significantly reduced in breast carcinomas, and in MCF-7 and T-47D breast cancer cell lines treated with estradiol (E2). In this study the mechanisms involved in E2 down regulation of BTG2 gene expression were examined. Depletion of ERα by siRNA indicated that the receptor is required for E2 down regulation of BTG2 mRNA levels, and cycloheximide experiments indicated that the effect of E2 on BTG2 expression was independent of intermediary protein synthesis. Chromatin immunoprecipitation analyses revealed that ERα interacts with the BTG2 promoter in a ligand-independent fashion whereas transfection experiments indicated that ERα’s DNA and ligand binding domains are required for E2 repression of BTG promoter activity. Surprisingly, histone deacetylase (HDACs) activity is essential for basal expression as evidenced by trichostatin A inhibition of BTG2 mRNA levels. Estradiol treatment did not alter histone H3 acetylation although it did induce displacement of RNA polymerase II from the BTG2 gene. Depletion of the ER specific corepressor REA (Repressor of Estrogen Receptor Activity) significantly abrogated E2-mediated BTG2 repression. Taken together, our results reveal a requirement of HDAC activity for basal BTG2 expression and the ERα-REA interaction for estrogen repression of the BTG2 gene. The ability of E2-bound ERα and REA to suppress BTG2 expression indicates a positive role for this corepressor in regulation of breast cancer cell proliferation.
B-cell translocation gene 2 (BTG2) is an antiproliferative (ARPO) tumor suppressor protein, because its overexpression leads to blockade of the cells at the G1 phase of the cell cycle.1,2 It was originally identified in the rat pheochromocytoma PC-12 cell line undergoing neuronal differentiation induced by nerve growth factor, where it was classified as an immediate early gene whose expression could be controlled by mitogenic, differentiative and antiproliferative factors.3,4 Later, it was demonstrated that BTG2 expression is induced by cytotoxic and genotoxic stress through a p53-dependent mechanism.2,5 Binding sites for p53 and NF-κB are located within the BTG2 promoter, and the latter correlates well to the regulation of BTG2 expression by NF-κB activators such as tumor necrosis factor-α (TNFα) and some members of the transforming growth factor-β (TGFβ) family.6,7 Induction of BTG2 expression inhibits S-phase entry via either retinoblastoma (Rb)-dependent8 or independent pathways,9,10 and this is attributed, at least in part, to transcriptional suppression of the cyclin D1 gene whose product is required for the G1 to S phase transition.8 A loss of nuclear BTG2 expression is observed in ERα positive human breast tumor samples,11 and a correlation between breast tumor size and BTG2 expression has also been demonstrated.12 In normal mammary gland, BTG2 expression is highly regulated by estrogen and progestin.7 Levels of both RNA and protein are suppressed during lactation, but return to prepregnancy levels rapidly once lactation is terminated. Consistent with in vivo studies, a decrease in BTG2 mRNA has been observed for E2 and progestin treated MCF-7 and T-47D breast cancer cell lines, respectively.7 Moreover, microarray analyses reveal that BTG2 mRNA levels are reduced in E2-treated MCF-7 cells.13 Taken together, these studies suggest that E2 dependent inhibition of BTG2 expression may help to promote cell cycle progression.
The 2 estrogen receptor genes (ERα and ERβ) belong to the steroid receptor superfamily and encode transcription factors that regulate diverse physiological processes in mammary gland, bone, brain and cardiovascular tissues.14 In the presence of its cognate ligand 17β-estradiol (E2), ERβ undergoes a series of sequential events that include changes in conformation followed by dimerization and recruitment to the regulatory regions of genes by either interacting directly with an estrogen response element (ERE) or indirectly via tethering to other DNA binding proteins such as AP-1, Sp1 or NF-κB. In the course of this process, ERα recruits coregulatory proteins that remodel chromatin structure and ultimately determine the fate of gene transcription. In general, ERα exerts its function via 2 distinct regions; the N-terminally located activation function 1 (AF1) domain is constitutive and the C-terminal activation function domain (AF2) is ligand dependent. Moreover, the contribution of AF1 and AF2 to ERα activity is both cell type and promoter specific.15,16 Depending on the nature of the ligand, particularly whether it is an agonist, pure antagonist or selective estrogen receptor modulator (SERM), binding to ERα induces its ligand binding domain to undergo a structural reorganization that dictates the ability of the receptor to interact with different coregulators (coactivators or corepressors) and therefore regulate gene expression.17
Although considerable progress in our understanding of the molecular mechanism of estrogen action has been achieved in recent years, the specific roles of estrogen receptors in transcriptional events related to abnormal mammary cell proliferation and carcinogenesis are poorly understood. The advent of microarray technology made possible the large scale identification of mRNAs sensitive to estradiol treatment.13,18-21 Although some of the up-regulated genes identified in these studies are important for cell cycle progression (e.g., c-Myc, cyclin D1), a number of them (e.g., GATA1, pS2, Cathepsin D, NRIP1) appear to have little relationship with regulation of cell proliferation or survival. Moreover, although most investigations of estrogen regulated gene expression in breast cancer have focused on defining mechanisms of transcriptional activation, recent microarray analyses of estradiol-treated MCF-7 cells revealed a significant number of down-regulated genes.13 Interestingly, a number of these encode proapoptotic proteins (e.g., IEX-1, caspase 9) consistent with the ability of estrogens to promote cell survival.22 Likewise, others are negative cell cycle regulatory genes (e.g., cyclin G2, p21), whose inhibition enables cells to progress through the cell cycle.23 More recently chromatin immunoprecipitation (ChIP) based methods (i.e., ChIP cloning or ChIP-chip) applied to MCF-7 cells have identified ERα genomic binding sites and further defined possible ERα negative regulatory regions.24-28 In one ChIP-chip study examining ERα and RNA polymerase II (Pol II) interactions with gene promoter regions, it was found that E2 treatment caused a reduction in Pol II binding to 16 of the 663 ERα target genes examined,26 and this suggested that E2-ERα can play a direct role in suppressing gene expression. On the basis of the proportion of genes downregulated versus upregulated by E2 in MCF-7 cells (36-70% to 64-30%, respectively, reviewed in ref. 29), it would appear that E2 repression of transcription may be as necessary as gene activation for appropriate regulation of cell proliferation and survival. This highlights the importance of understanding the mechanism(s) employed by agonist-bound ERα to repress gene expression.
The list of well-characterized genes for which mechanisms of negative regulation by agonist-bound ER have been determined is relatively small and our understanding of the mechanism(s) involved in such repressive processes is undoubtedly incomplete. In some instances, E2-bound ERα repression of gene expression is associated with interactions between ERα and transcription factors such as p53 and Nrf2.30,31 It also has been shown that E2-dependent recruitment of the NCoR corepressor is associated with many of these negatively regulated genes.32-35 Somewhat surprisingly, the SRC-2/GRIP1 coactivator has been found to be a key factor for repression of some genes36 and a repression domain has been mapped within this coregulator indicating that distinct surfaces of the SRC-2/GRIP1 coregulator are utilized in activation and repression contexts.37 In addition, the involvement of the CBP/p300 coactivators in negative regulation of c-Myc in quiescent cells recently has been reported,38,39 and this further suggests that a complex array of mechanisms employing coactivators and corepressors can be utilized to repress gene expression.
On the basis of the microarray evidence that E2 treatment reduces steady state levels of BTG2 mRNA and the relative paucity of information on mechanisms of E2-ERα suppression of gene expression, we investigated the molecular mechanisms employed by agonist occupied ER to regulate the expression of the BTG2 gene. Our studies demonstrate an unexpected requirement of histone deacetylase activity for basal BTG2 mRNA expression and reveal a novel mechanism in which agonist-bound ERα in conjunction with the REA (Repressor of Estrogen receptor Activity) corepressor actively represses expression of the BTG2 gene.
17β-Estradiol and the partial antiestrogen 4-hydroxytamoxifen (4HT) were purchased from Sigma (St. Louis, MO), whereas the pure antiestrogen ICI 182,780 (ICI) was obtained from Tocris (Ellisville, MO). Trichostatin A (TSA) was obtained from Millipore (Millipore Corp, Bedford, MA) and cycloheximide (CHX) was from Sigma.
The plasmids that express human full length ERα (pCR3.1-hERα), ERβ (pCXN2-hERβ) and also the expression plasmids for ERα lacking the A/B domain (pCR3.1-hERα-179C, amino acids 1-178 deleted), ERα with 3 mutations (K362D/V376D/L539A) in the ligand binding domain (pCR3.1-hERα-KVL), ERα with a double mutation (E203A/G204A) within the DBD (pCR3.1-hERα-DBD) and ERα containing the A/B and DNA binding domains (pCR3.1-hERα-N282G, amino acids 1-282) have been described previously.40,41 Luciferase constructs p2658 and p266 (nucleotides -1 to -2658 and -1 to -266 upstream of translational start site of the BTG2 gene cloned in pGL3) were kindly provided by Dr. Alain Puisieux.6 All siRNAs were chemically synthesized by Ambion (Austin, TX) as oligonucleotide duplexes. Target siRNA sequences for ERα,41 SMRT,42 NCoR,42 SRC-1,43 SRC-244 and SRC-343 were described previously. Predesigned siRNA for CBP (ID no. 146409) was obtained from Ambion. The siRNA against REA was obtained as a SMART pool from Dharmacon Research (Lafayette, CO). Ambion’s Negative control siRNA no. 2 was used as nonspecific siRNA control.
The MCF-7 human breast cancer cell line was maintained in Dulbecco’s modified Eagles’s medium (DMEM) supplemented with 10% fetal bovine serum (FBS). For treatments or siRNA transfections, 3×105 cells MCF-7 cells were plated in each well of a 6-well plate and grown overnight in phenol-red-free DMEM containing 10% FBS. For treatments with TSA, CHX or ERα ligands, 24 hr after plating cells were switched to phenol red-free DMEM containing 10% stripped serum and treated for the indicated time periods. For transfection experiments, one day after plating, cells were transfected with 20-40 pmol/well siRNA using Oligofectamine (Invitrogen, Carlsbad, CA) following the manufacturer’s instructions. Four hours after transfection, medium was switched to phenol-red-free DMEM containing 10% sFBS and 24-48 hr later cells were treated with vehicle or hormone for 4 hr. Concentrations of siRNA, hormone and other treatments and their duration are indicated in the figure legends.
The HeLa human cervical cancer cell line was maintained in DMEM supplemented with 10% FBS. For ERα trans-activation assays, 3×105 Hela cells were plated in each well of a 6-well plate and grown overnight in phenol-red free DMEM containing 5% sFBS, followed by transfection with the indicated plasmids using Lipofectamine (Invitrogen). Thereafter cells were treated with vehicle or 10 nM E2, and 16 hr later were harvested for luciferase activity assays using a luciferase assay kit (Promega) and a Luminoskan Ascent Thermo Labsystems apparatus (Thermo Electron, Milford, MA). Relative luciferase units were normalized to total protein content of the cell lysate measured by Bio-Rad protein assay (Bio-Rad, Hercules, CA).
For mRNA measurements, total RNA was isolated by TRIzol® reagent (Invitrogen) following the manufacturer’s instructions and analyzed by RT-qPCR using an ABI Prism 7500 sequence analyzer (PE Applied Biosystems, Foster City CA). Predeveloped 20× primer and probe mix was purchased from Applied Biosystems (BTG2: Cat no. Hs00198887_m1 and 18s: Cat no. 4310893E). Primers for cyclin G2 mRNA34 were obtained from Invitrogen. BTG2 and 18S was measured by employing TaqMan chemistry, whereas the SYBR method was used to measure cyclin G2.
Western blot analyses were performed essentially as described previously.42 Briefly, equal amounts of protein from cell lysates were resolved by SDS-PAGE, electro-transferred to nitrocellulose membrane and probed with primary antibodies (see below) and the appropriate HRP-conjugated secondary antibody. Signals were detected by chemiluminescence using ECL Plus (GE healthcare) with XO-1 blue film (Kodak; Branchburg, NJ). The primary antibodies are as follows: anti-ERα (HC-20, Santa Cruz Biotechnology), anti-SRC-1 (612378, BD Transduction Laboratories), anti-SRC-2 (610985, BD Transduction Laboratories), anti-SRC-3 (611095, BD Transduction Laboratories), anti-CBP (557021, BD Transduction Laboratories), anti-NCoR (06-892, Upstate), anti-SMRT (611386, BD Transduction Laboratories), anti-REA (A300-658A, Bethyl Laboratories, Montgomery, TX) and anti-actin (MAB1501R, Chemicon).
Chromatin immunoprecipitation (ChIP) assays were performed as described earlier.42 Briefly, MCF-7 cells were grown in phenol red-free DMEM containing 5% sFBS for 48 hr and treated with 2.5 μM α-amanitin for 90 min. After removal of α-amanitin, cells were washed twice with PBS and then treated with vehicle or 10 nM E2 for 45 min before cell fixation with formaldehyde. Chromatin was precipitated with antibodies against ERα (HC-20 & H-184), SRC-3 (C-20), NCoR (N-19 & C-20), RNA polymerase II (N-20 & H-224) or normal rabbit IgG; all antibodies were obtained from Santa Cruz Biotechnology. Antibodies to REA (A300-656A & A300-658A) were from Bethyl Laboratories and acetyl-histone H3 antibody (06-599) was from Upstate (Upstate Biotechnology, Lake Placid, NY). Purified, precipitated chromatin was quantified by qPCR using SYBR green chemistry and normalized against the input chromatin. Primer sequences used in the qPCR are provided in Table I.
A time course microarray analysis of estradiol (E2)-treated MCF-7 cells identified BTG2 as an early estrogen down-regulated gene.13 Levels of BTG2 mRNA were reduced to 40% of their starting levels within 4 hr of E2 exposure and remained suppressed for up to 48 hr after hormone treatment. Our first objective was to verify that E2 treatment suppresses BTG2 mRNA expression in MCF-7 breast cancer cells and to determine the ability of partial and pure ER antagonists to affect BTG2 gene expression. As shown in Figure 1, 4 hr treatment of MCF-7 cells with 10 nM E2 resulted in a (65 ± 6)% reduction in BTG2 mRNA levels, consistent with previous microarray results.13 In contrast, RT-qPCR analyses demonstrated that BTG2 mRNA expression was unaffected by 4 hr exposure to either 100 nM 4HT or ICI under similar culture conditions, and this suggests that unliganded ERα does not significantly affect BTG2 gene expression. To test whether estrogen’s suppressive effect on BTG2 mRNA expression is mediated by an estrogen-regulated intermediate, MCF-7 cells were exposed to the translation inhibitor cycloheximide for 4 hr before treatment with vehicle or E2 for 4 hr. Quantitation of BTG2 mRNA levels revealed that cycloheximide was unable to block E2 downregulation (Fig. 1b), indicating that E2 suppression of BTG2 mRNA levels does not require the synthesis of an intermediary protein.
To assess the involvement of ERα in E2-directed BTG2 mRNA repression, cellular ERα levels in MCF-7 cells were depleted by specific siRNA and then cells were treated with vehicle or estrogen. Depletion of ERα completely abrogated the E2-mediated repression of BTG2 mRNA expression (Fig. 2a), indicating that this receptor is required for E2 to inhibit BTG2 gene expression. The efficacy of the siRNA for inhibition of ERα expression was demonstrated by Western blot analysis (Fig. 2b). Next, experiments were initiated to determine if the proximal BTG2 promoter was responsive to regulation by ERα and E2. With the exception of the identification of functional p53 and retinoic acid receptor (RAR) response elements located upstreams of the translational start site,6,45 little is known about transcription factor interactions with the BTG2 promoter. Therefore, 2 luciferase reporter constructs consisting of nucleotides -2658 → -1 (p2658) or -266 → -1 (p266) of the BTG2 promoter cloned into the pGL3 luciferase expression vector were tested to determine their respective responsiveness to signaling by E2 and ERα in ER-negative Hela cells. These constructs revealed that an active region of the BTG2 promoter lies between 266 and 2658 nucleotides upstream of the translational start site, as an ~5 fold increase in luciferase activity was obtained with the p2658 compared to the p266 reporter plasmid (Fig. 2c). Transient transfection of ERα along with the p2658 reporter and subsequent treatment with E2 significantly decreased luciferase gene expression by (47 ± 4)% compared to the vehicle-treated control (Fig. 2d). The extent to which estrogen suppresses the expression of the p2658 BTG2 reporter is consistent with the results for ERα-positive MCF-7 cells obtained above in which E2 treatment significantly decreased BTG2 mRNA levels by 51-65% in comparison to vehicle controls (Figs. (Figs.1a1a and and2a).2a). Furthermore, treatment with the ERα antagonists 4HT and ICI prevented the E2-mediated downregulation of the p2658 reporter indicating a clear role for E2-ERα in the repression of BTG2 promoter activity (data not shown). It was also noted that ectopic expression of ERβ in HeLa cells supported E2 repression of the p2658 reporter gene (Fig. 2e), and this indicates that either isoform of ER is sufficient to downregulate BTG2 gene expression in these E2 treated cells.
Estrogen receptors can interact with target genes directly or indirectly via tethering to certain other transcription factors [e.g., AP-1,46,47 Sp1,48,49 p5331,50 and NF-κB51], and to determine whether potential interaction sites for ERs were present within 3000 nucleotide upstream of the BTG2 translation start site, MaT-Inspector software (http://www.genomatix.de) was used to inspect for estrogen response elements (EREs) and binding sites for other transcription factors known to interact with ERs. This computational analysis revealed six half ERE sites and potential binding sites for AP-1, Sp1, NF-κB and p53 transcription factors (Fig. 3a). In addition, a recent genome-wide ERα ChIP-on-chip analysis suggested the presence of an ERα binding site ~2000 bp upstream of the BTG2 translational start site24; thus, a number of potential ERα interaction sites are located within the BTG2 promoter region. Therefore, a ChIP-PCR scanning approach was employed to identify ERα binding regions within the endogenous BTG2 gene in cells that permitted its E2 downregulation; this avoids the potentially confounding influence of transiently transfected reporter templates and overexpressed exogenous proteins. Sheared chromatin with an average size of 500 nucleotides was prepared from MCF-7 cells treated with vehicle or E2 for 45 min and immunoprecipitated with either ERα specific antibodies or normal rabbit IgG to assess nonspecific interactions. Purified DNA was quantitated by qPCR using sets of PCR primers designed to amplify regions of the promoter ~500-1000 bp apart, thereby covering an area from -3000 to +2000 nt with respect to the BTG2 translational start site (Fig. 3a). One set of primers was intentionally designed to reside within the region implicated as an ER binding site in prior ChIP-on-chip analyses.24 A control primer set far upstream of the BTG2 transcriptional start site (located 7958 bp upstream of the translational start site), was used for PCR amplification as a negative control. Quantitative PCR of the immunoprecipitated chromatin samples with the P1-P6 primer sets (Table I and Fig. 3a) demonstrated E2-independent binding of ERα across ~3000 nt of the BTG2 promoter and to a lesser extent downstream of the translational start site (Fig. 3b). The signal obtained for ERα is ~6-fold higher than for the IgG negative controls upstream of the translational start site, and declines downstream of this position. In contrast, the signal obtained for ERα with the Pup primers at the -7958 location was indistinguishable from the signal obtained for normal IgG indicating no specific receptor interaction at that site. As a positive control, the same chromatin samples were used to determine the occupancy of ERα at the known estrogen target gene, pS2, using primers adjacent to the ERE within the pS2 promoter.52 In contrast to the BTG2 results, E2 treatment induced an approximately 10-fold induction of ERα recruitment to the pS2 promoter (data not shown), and this clearly distinguishes the hormone-independent nature of ERα’s interaction with the BTG2 promoter from its E2-stimulated interaction with the pS2 gene. Moreover, as estrogen repression of BTG2 mRNA expression is not reflected in hormone-dependent recruitment of ERα to the gene, these results suggest that other E2-dependent events are required for BTG2 down regulation.
Deacetylation of histones is an important regulatory event frequently associated with gene repression. To assess the participation of HDACs in E2-dependent repression of BTG2 expression, cells were pretreated with the general HDAC inhibitor trichostatin A (TSA) for 16 hr followed by 4 hr hormone treatment, and BTG2 mRNA expression was measured thereafter by reverse transcription and qPCR. Surprisingly, TSA alone decreased BTG2 mRNA expression to an extent similar to that obtained for E2 treatment (Fig. 4a), indicating that HDAC activity is required for basal BTG2 expression. In TSA treated cells, E2 had no further effect on BTG2 mRNA levels and this suggests that either TSA blocks the activity of HDACs required for E2 repression of BTG2 expression or possibly TSA blocks the activity and/or expression of other factors required for basal BTG2 mRNA expression and the low levels of the BTG2 mRNA remaining after the TSA treatment cannot be further reduced by E2-bound ERα. These results contrast with those obtained for a previously characterized, estrogen down-regulated gene, cyclin G2.34 Pretreatment with TSA blocks downregulation of cyclin G2 mRNA levels by E2 but does not affect basal expression of this gene (Fig. 4b).
To take a more global approach to determining whether HDAC activity is associated with E2 repression of BTG2 mRNA expression, ChIP assays using an antibody against pan-acetylated histone H3 were used to investigate whether E2 treatment produced a deacetylation of histones at the BTG2 promoter. Chromatin was prepared for MCF-7 cells treated with either vehicle or E2 for 45 min and immunoprecipitated DNA was quantitated using the P3-P6 primers upstream and downstream of the translational start site (Fig. 5a). As expected, high levels of acetylated histone H3 were observed for primer locations near the transcriptional start site in comparison to the IgG controls, with the highest levels obtained for the primer sets located immediately upstream and downstream of this region (Fig. 5b). However, the extent of total histone H3 acetylation was unaffected by E2 treatment suggesting that deacetylation of this histone in this region does not contribute to repression of BTG2 gene expression. In contrast, ChIP assays employing antibody to RNA polymerase II (Pol II) revealed that Pol II binding, which is highest near the BTG2 translational start site, is reduced in E2 versus vehicle-treated cells (Fig. 5c). The levels of acetylated histone H3 and Pol II were similar to the IgG negative control at the P1 and P2 regions defined as in Figure 3. Thus, the collective results indicate that E2 repression of the BTG2 gene is associated with ERα interaction with a broad region of the BTG2 promoter and displacement of Pol II from the BTG2 gene.
To define the structural features of ERα that are required for repression of BTG2 gene expression a series of ERα mutants encompassing either deletions or point mutations that affect known receptor functions (e.g., DNA binding, ligand binding, coactivator interaction or transcriptional activation) were tested for their ability to inhibit the activity of the p2658 BTG2 promoter-luciferase reporter gene. A schematic representation of the mutants used in these experiments is shown in Figure 6a. Hela cells were transfected with expression plasmids for wild type or mutant forms of ERα, along with the p2658 reporter construct followed by hormone treatment. In the case of the AF1 deletion mutant (ERα-179C), repression of the activity of the BTG2 promoter was similar to that obtained for wild type ERα, indicating that BTG2 repression by E2 does not require the ligand-independent AF1 domain of ERα (Fig. 6b). However, disruption of the DBD by a mutation within the P-box that blocks DNA binding ability effectively prevents this ERα mutant from repressing BTG2 promoter activity. Next, the N282G ERα mutant which lacks the ligand binding domain (LBD) and consequently AF-2 was tested and found unable to repress BTG2 gene expression suggesting that the LBD and/or AF2 is required for E2-dependent down regulation of BTG2 gene expression by ERα. To further ascertain if an intact AF-2 domain within the LBD is necessary for repression of BTG2, a specific LBD mutant ERα-KVL which possesses 3 mutations that disrupt the integrity of the AF-2 coactivator binding groove was tested in the BTG2 trans-repression assay. The inability of ERα-KVL to repress the BTG2-luciferase reporter in E2-treated cells indicates that coregulators that interact with receptor via the coactivator binding groove are required for ERα to repress BTG2 gene expression. Moreover, it also was noted that the activity of the p2658 reporter was greater in cells co-transfected with ERα-KVL than wild type ERα suggesting that this ERα mutant, perhaps through a basal interaction at the BTG2 promoter, blocks repressive events that negatively affect the gene under basal conditions. Taken together, these experiments indicate that E2-induced BTG2 gene repression requires the intact DBD and LBD of ERα, and suggest that coregulators that interact with ERα via its coactivator binding groove play an important role in the downregulation of this gene.
Previous studies examining the inhibition of ERα target gene expression by unliganded or antagonist-bound receptors have implicated the corepressors NCoR and SMRT with transcriptional repression,53-55 and NCoR is recruited to the promoters of several E2-downregulated genes including cyclin G2, VEGFR2, BMP7, ABCG2 and BCL-3.33-35 However, the p160 family coactivator SRC-2 (also known as TIF2 or GRIP1) has also been shown to be involved in E2-mediated repression of the TNFα gene,36 and both coactivators and corepressors were therefore tested for their ability to mediate E2-ERα repression of the BTG2 gene, focusing on the p160 family of coactivators (SRC-1, SRC-2 and SRC-3), the classical nuclear receptor corepressors NCoR and SMRT and the ER-specific corepressor REA. In addition, CBP/p300 which has been associated with repression of cMyc was tested.38,39 MCF-7 cells were transfected with specific siRNAs for each coregulator and cells subsequently were treated with hormone for 4 hr, followed by RT-qPCR for BTG2 mRNA expression. Depletion of endogenous proteins by siRNA was verified by Western blotting (Fig. 7c). The effect of coregulator depletion was first examined with respect to effects on BTG2 mRNA expression under basal conditions. Out of all the cofactors tested, only SRC-3 depletion induced a significant increase in basal BTG2 mRNA expression, whereas depletion of NCoR, SRC-1 and SRC-2 produced modest but insignificant decreases in basal BTG2 mRNA levels (Fig. 7a). The expected E2-dependent repression of BTG2 mRNA expression in control siRNA treated cells, and the loss of this repression in cells depleted of ERα is demonstrated in Figure 7b. Estradiol-dependent repression of BTG2 mRNA levels was not affected by depletion of any of the tested nuclear receptor coactivators (i.e., SRC-1, SRC-2, SRC-3 or CBP). However, depletion of REA effectively abrogated E2-mediated BTG2 downregulation, whereas inhibition of SMRT or NCoR expression modestly attenuated the ability of E2 to repress BTG2 mRNA levels. Taken together, these data indicate that REA exerts a major effect in E2 repression of BTG2 gene expression, with minor contributions by SMRT and NCoR.
To extend these results, ChIP assays were performed to examine the interaction of REA, NCoR and SRC-3 with the BTG2 gene at positions corresponding to the P1 to P4 amplicons. Although low levels of NCoR and SRC-3 were detected at each of these positions, consistent with their respective abilities to exert effects on BTG2 gene expression, there was no change in their interaction with the BTG2 promoter following E2 treatment (data not shown). To verify that the ChIP assays were working properly, the same immunoprecipitated chromatin samples were examined with primers for the pS2 gene which revealed the expected E2-dependent recruitment of SRC-3 and release of NCoR.43,56 Surprisingly, REA was undetectable at any of the tested locations (P1-P4) of the BTG2 gene. Although this result suggests that REA does not interact with the BTG2 promoter, it is also possible that the available REA antibody is not suitable for ChIP assay or that the epitopes recognized by the REA antibody are masked by other factors located at the BTG2 gene.
It is now appreciated that control of cell proliferation by estrogen receptors is achieved not only by activating the expression of cell cycle regulatory and anti-apoptotic genes but also by down regulating genes that encode proteins with proapoptotic properties, or those which directly or indirectly inhibit cell cycle progression. Although considerable attention has been focused on delineating the mechanisms of agonist-dependent gene activation and antagonist-mediated gene repression by ERα, relatively little is known about how E2 represses gene expression. Our results demonstrate that E2 acts via estrogen receptors to directly repress the BTG2 gene. This negative regulation is dependent on the ER specific corepressor REA, with lesser contributions from the general corepressors, NCoR and SMRT; it also is accompanied by displacement of Pol II from the BTG2 promoter. Our data also demonstrate that HDAC activity, typically associated with gene repression, is required for maximal BTG2 gene expression, and this indicates that HDACs can exert positive effects on gene expression in addition to their well established repressive effects.
Our ChIP experiments indicate that ERα interacts strongly with a broad region of the BTG2 promoter region, and can be found at lower levels downstream of the presumed translational start site. However, there was not any change in the level of ERα bound to the BTG2 promoter associated with E2 treatment. Although this was unexpected, recent work by the Kraus laboratory recently analyzed estrogen-regulated promoters based on ERα and Pol II binding. Our work on BTG2 indicates that it falls within the Class IVB promoter group that is characterized by reduced Pol II binding with no change in recruitment of ERα.26 The E2-ERα repressed gene, E-cadherin, also shows no E2-dependent change in ERα occupancy at its promoter57 suggesting that it may fall into the same category of E2 repressed genes as BTG2. An in silico search of the BTG2 promoter region from -3000 to +1 by the Estrogen Responsive Genes Database (ERGDB) revealed no EREs; however, several ERE half-sites were detected as were binding sites for transcription factors (e.g., AP-1, Sp1, p53) known to mediate indirect binding of ERα to genes. It has been demonstrated that the nature of ERα’s interaction with target genes can influence whether it activates or represses gene expression. Specifically, ERα interaction with the initiator and flanking sequences of the folate receptor (FR)-α core promoter is sufficient for E2 repression of this gene by E2.32 In contrast, when stronger ER binding sties are inserted into the context of the core promoter, E2 stimulates gene expression. Thus our data adds to the cumulative evidence demonstrating that nuclear receptors can employ diverse modes of interaction with target genes to inhibit their expression.
Repression of the BTG2 gene by E2 does not require the ligand-independent AF1 domain of ERα, and this is similar to the finding for several other E2-ERα down regulated genes, such as FRα and vascular endothelial growth factor receptor (VEGFR)-2.32,33 However, in the case of cyclin G2, another E2-ERα inhibited gene, the N-terminal domain of ERα was found to be indispensable for repression, whereas the ligand-dependent AF2 domain was not required.34 The ERα AF2 domain is essential for E2-dependent repression of BTG2 gene expression, and together, these data suggest that ERα can employ different mechanisms to facilitate coregulator interactions with repressed genes. Interestingly, a requirement of the ERα DNA binding domain is common to a number of E2 repressed genes including BTG2.32,33,34 As noted above, there are several half-ERE sites within the BTG2 promoter, and it is possible that the DBD enables ERα to bind directly to these elements. However, other investigators have shown that the DBD is essential for indirect interactions of ERα with DNA via tethering of the receptor to Sp1 and AP-1 transcription factors.58,59 Thus, we are unable to conclude that E2-ERα repression of BTG2 requires direct interaction of the receptor with DNA, as it is possible that the DBD is required for maintaining the structural organization of the receptor and/or its ability to interact with other transcription factors and coregulators essential for BTG2 gene repression.
The importance of ERα interactions with coregulators for BTG2 repression was first suggested by the ability of the K362D/V376D/L539A mutations to abrogate E2-dependent repression of BTG2. Depletion of SMRT or NCoR by siRNA had a modest impact on E2 repression of BTG2 mRNA expression, and this distinguishes BTG2 from other E2 down regulated genes (FRα, VEGFR2, cyclin G2) that required NCoR for this purpose.32-34 In contrast, E2 repression of BTG2 was significantly abrogated by depletion of the ER-specific corepressor, REA. Prior work on REA has focused on its ability to attenuate E2 induction of gene expression60-62 and its ability to negatively regulate mammary gland proliferation and differentiation in mice.63,64 Biochemical studies reveal that ERα interactions with REA are ligand-dependent and can be induced by either E2 or the antagonist, 4HT.62 In this regard, it is somewhat surprising that the tested antagonists, 4HT and ICI, did not repress BTG2 gene expression, and this suggests that factors in addition to ERα may be required to facilitate REA repression of BTG2 gene expression. Unlike other ERα corepressors such as SMRT and NCoR, REA does not employ NR (Nuclear Receptor) or CoRNR (Corepressor-Nuclear Receptor) box motifs for its interaction with ERα or its repressive activity.60 The repressive effects of REA on E2-stimulated ERα-dependent gene expression are achieved via competition between REA and the SRC-1 coactivator for binding to ERα, and consequently reductions in receptor-coactivator interaction.60,62 Knock-down of SRC-1 does not block E2 repression of the BTG2 gene and it therefore appears that REA repression in the context of E2-inhibited transcription differs mechanistically from its inhibitory effects on E2-induced genes.
In general, repression of gene expression is associated with recruitment of corepressors and their associated HDACs and consequently deacetylation of histones, compaction of chromatin structure and ultimately inhibition of transcription, and it has been shown that REA can interact with HDACs.65 Interestingly, in the case of BTG2, treatment with the HDAC inhibitor TSA significantly reduced basal BTG2 gene expression, and blocked further reduction by E2 treatment. Although it is possible that TSA treatment blocks E2-ERα suppression of BTG2 gene expression either at the level of ERα-REA-HDAC or through an alternative repressive process, it also is possible that TSA inhibition of basal BTG2 expression leaves little transcription to be inhibited by E2 treatment. The TSA inhibition of basal BTG2 mRNA expression was unexpected, but other instances of a requirement of HDAC activity for gene transcription have been noted. For instance, transcription of selected interferon mediated immediate early genes that are regulated by signal transducer and activator of transcription-1 (STAT-1) and STAT-2 transcription factors requires HDAC activity,66,67 and HDAC1 is important for glucocorticoid receptor induction of reporter gene activation.68
In addition to a requirement of HDAC activity for maximal expression of the BTG2 gene, our depletion experiments also revealed that depletion of the SRC-3 coactivator increased basal BTG2 expression. Interestingly, SRC-3 possesses intrinsic histone acetyltransferase activity,69 and this further suggests that acetylation may reduce BTG2 mRNA levels in MCF-7 cells, consistent with our observation of a requirement of deacetylation for basal BTG2 expression. It is, however, also possible that the stimulatory effect of SRC-3 depletion represents an indirect effect via alterations in the expression of other cofactors that regulate BTG2 mRNA production. Nonetheless, it is interesting to note that increased BTG2 expression in SRC-3 depleted MCF-7 cells may be physiologically relevant as we have recently demonstrated that depletion of SRC-3 inhibits cell cycle progression and cellular proliferation (Karmakar and Smith, unpublished communication) which is anticipated for cells that express increased levels of the antiproliferative BTG2 protein.
The ability of E2-bound ERα to stimulate cyclin D1 expression in breast cancer cells is well studied, and it is likely that estrogen stimulation of cellular proliferation represents the combined effects of direct ERα-mediated BTG2 inhibition and cyclin D1 stimulation. The ability of ectopic BTG2 expression to reduce the growth of MCF-7 breast cancer cells on soft agar and to increase the percentage of cells in the G1 phase of the cell cycle has demonstrated the importance of BTG2 to control of breast cancer cell proliferation.7 Indeed, the importance of one transcription factor, ERα, being able to coordinately regulate the expression of these proteins is easily appreciated when one considers that exogenous expression of cyclin D1 can overcome cell cycle arrest induced by overexpression of BTG2.8 Conversely, in the absence of estrogens, low ERα activity results in low cyclin D1 expression and an increased level of BTG2 which further inhibits cyclin D1 mRNA expression.8
In summary, this study provides evidence for E2-bound ERα and REA dependent down regulation of the BTG2 gene. One of the questions that remain in this and other studies examining ERα repression of gene expression is what determines whether agonist-bound receptor binds to coactivator or corepressor. It appears logical that the identity of the target gene must play a major role and it is possible that target genes impart structural alterations in the receptor or alter the spatial relationship between receptors within an ERα homodimer and thereby influence receptor interactions with corepressors. This would be analogous to prior investigations in which it was demonstrated that the nature of the ERα binding site within a target gene influences receptor structure and interactions with coactivators.70-72 Thus the specificity of ERα down regulation of gene expression may also provide an avenue to specifically block this mode of gene regulation. Because the ability of E2-ERα to suppress BTG2 expression may be a major pathway for estrogen stimulation of ERα-positive breast cancer cell proliferation, we postulate that inhibition of ERα-REA interactions may have significant potential as a targeted therapeutic strategy for breast cancer.
The technical support of Judy Roscoe and Cheryl Parker for cell culture is gratefully acknowledged. SK was supported by a post-doctoral fellowship award (PDF 0707868) from the Susan G. Komen for the Cure foundation. This work was supported by Public Health Service grants to CLS.
Grant sponsor: Public Health Service; Grant numbers: DK53002, DK64308.