G Protein Pathway Suppressor 2 (GPS2) Is a Transcriptional Corepressor Important for Estrogen Receptor α-mediated Transcriptional Regulation*

We have identified G protein suppressor 2 (GPS2) as a stable component of the SMRT corepressor complexes. GPS2 potently represses basal transcription, with the repression domain mapped to the N-terminal silencing mediator of retinoic acid and thyroid hormone receptor (SMRT)-interacting domain. Knockdown of GPS2 abrogates, whereas overexpression potentiates, SMRT-mediated repression activity. The SMRT complexes are involved in 4-hydroxyl-tamoxifen (4OHT)-mediated gene repression by estrogen receptor α (ERα). We show that 4OHT recruits SMRT and GPS2 to the promoter of pS2, an ERα target gene, in a dynamic manner. Unexpectedly, we also found that estradiol (E2) promotes promoter recruitment of the SMRT complexes. While knockdown of GPS2 compromised 4OHT-mediated repression, it enhanced E2-induced expression of a reporter gene and several endogenous ERα target genes, including pS2, cyclin D1 (CCND1), progesterone receptor (PR), and c-MYC. Finally, we show that depletion of GPS2 or SMRT by siRNA promotes cell proliferation in MCF-7 breast cancer cells. Thus, we concluded that GPS2 is an integral component of the SMRT complexes, important for ligand-dependent gene regulations by ERα and a suppressor for MCF-7 cell proliferation.

Transcriptional regulation by estrogen receptor ␣ (ER␣) 2 is mediated by transcriptional coactivators and corepressors. SMRT and N-CoR are two closely related transcriptional corepressors that mediate transcriptional repression by many nuclear hormone receptors including ER␣ (1)(2)(3)(4)(5). Both SMRT and N-CoR are large nuclear proteins that contain multiple functional domains, including four repression domains (RDs) (2,4,6,7). Transcriptional repression by SMRT or N-CoR is manifested through their recruitment of a histone deacetylase (HDAC) complex containing mSin3A and HDAC1 (6 -8). In addition, SMRT and N-CoR also interact with HDAC3 (9 -11) and with class II HDACs including HDAC4, -5, and -7 (12,13). TBL1 and TBRL1 are also common stable components of the SMRT and N-CoR complexes (9,14). Despite this long list, SMRT and N-CoR are large proteins, and other components of their complexes undoubtedly exist but have not yet been identified.
GPS2 (G protein suppressor 2) was first identified as a suppressor of G protein-activated MAPK signaling in both yeast and mammalian cells (16,17). GPS2 is also involved in transcription regulation through its association with the papillomavirus E2 and E6 proteins (18,19), the human T-cell lymphotrophic viral oncoprotein Tax (20), p53 and p300 (18,19,21), and regulatory factor X4 variant transcript 3 (RFX4_v3) (22). GPS2 was also found in the N-CoR corepressor complex and is capable of repressing basal transcription when forcibly bound to a promoter (23). Besides transcription regulation, previous studies suggested that GPS2 is involved in cell cycle regulation, DNA repair, cytoskeleton architecture, brain development, and metabolism (22, 24 -29). These data indicate that GPS2 is involved in multiple cellular pathways.
To better understand the mechanisms by which SMRT regulates transcription, we initiated yeast two-hybrid (Y2H) screens and identified GPS2 as a SMRT-interacting protein.
We dissected the role of GPS2 in transcriptional regulation by SMRT and ER␣. Our data suggest that GPS2 is required for optimal 4OHT-mediated repression of ER␣ target genes, while knockdown of GPS2 increases E2-induced ER␣-mediated reporter activity and some natural ER␣ target gene expression. We also show that GPS2-containing SMRT corepressor complexes are recruited to the promoter of the pS2 gene, a wellknown ER␣ target gene, in response to 4OHT treatment in a dynamic way. Knockdown of GPS2 dramatically changed this recruitment pattern. Unexpectedly, we also found that E2 induced recruitment of GPS2 and SMRT complexes to pS2 promoter. Together, our data suggested that GPS2 is an important component of SMRT corepressor complexes and functionally involved in agonist-or antagonist-regulated ER␣-mediated transcription.

EXPERIMENTAL PROCEDURES
Transient Transfection of siRNA-Transient transfection of siRNA was performed according to a protocol adapted from the manufacturer. Briefly, cells at 60% confluency were transfected with commercial siRNA at a final concentration of 100 nM using the DharmaFect no. 1 transfection reagent (Dharmacon) in serum reduced OPTI-MEM. The knockdown efficiency was confirmed by both RT-PCR and Western blot analyses.
Luciferase Reporter Assays-Reporter plasmids used in this report, MH100 (Gal4-tk-luc) and ERE-tk-Luc, have been previously described (30). For MH100 reporter assays, 70% confluent cells plated in 48-well plates were co-transfected with MH100, CMX-␤-gal (internal control), and corresponding expression plasmids with Lipofectamine 2000 following the manufacturer's protocol (Invitrogen). Empty CMX vector was used as supplement to assure equal amounts of total DNA were transfected. 36ϳ48 h post-transfection, cells were harvested and subjected to luciferase and ␤-galactosidase assays according to the manufacturer's instruction (Promega). For siRNA transfection followed by luciferase assays, the cells were transfected with siRNA in 10-cm plates 24 h prior to seeding into 48-well plates for reporter gene transfection. In ERE-tk-Luc reporter assays, MCF-7 cells were transfected with siRNA in normal Dulbecco's modified Eagle's medium for 24 h. Cells were subsequently transfected with ERE-tk-Luc reporter along with the CMX-␤-gal by Fugene 6 (Roche) according to the manufactur-er's protocol and cultured in charcoal-stripped medium free of phenol red for 48 h. Cells were then treated with 10 nM E2 or 100 nM tamoxifen (TAM) for 24 h before luciferase assays. Unless otherwise specified, all luciferase assays were normalized to ␤-galactosidase activity, and results were depicted as mean Ϯ S.E. of relative luciferase units from three independent experiments.
Chromatin Immunoprecipitation (ChIP) Assays-The ChIP assay protocol was modified from the Farnham Laboratories protocol (31). Briefly, for ChIP with endogenous proteins, MCF-7 cells were cultured in charcoal-stripped, phenol redfree Dulbecco's modified Eagle's medium for 48 h until 90% confluent. The cells were treated with 2.5 mM a-amanitin (Sigma) for 2 h to preclear the promoter region (32) prior to the addition of E2 hormone or 4OHT drug for up to 180 min. After treatment, cells were crosslinked with 1% formaldehyde for 10 min at room temperature and stopped by the addition of glycine to a final concentration of 125 mM. Cross-linked cells were collected in 1ϫ phosphate-buffered saline and lysed in 5 ml of cell lysis buffer (5 mM PIPES pH 8.0, 85 mM KCl, 0.5% Nonidet P-40, 0.5 mM EDTA) plus proteinase inhibitors mixture on ice for 10 min. The nuclei were pelleted by centrifugation and subsequently dissolved in 500 ml of nuclei lysis buffer (50 mM Tris-Cl, pH 8.1, 10 mM EDTA, 1% SDS) plus protease inhibitors at 4°C for 10 min. The chromatin was sheared to 200ϳ500-bp fragments by sonication (Fisher 550a) with the microtip probe installed at a power setting of 4 for 150 s total sonication time (10 s on and 50 s off on ice for 15 cycles). The sonicated samples were centrifuged at 14,000 rpm at 4°C for 15 min, and the sheared chromatin was collected in the supernatant. The chromatin was then diluted 10-fold with IP dilution buffer (0.01% SDS, 1.1% Triton X-100, 1.2 mM EDTA, 16.7 mM Tris-Cl, pH 8.1, 167 mM NaCl) and cleared by incubating with preblocked protein A slurry (blocked by incubating with 1 mg/ml bovine serum albumin and 1 mg/ml sheared herring DNA overnight) at 4°C for 2 h. The cleared chromatin was recovered from the supernatant after centrifugation, then aliquoted and mixed with a specific antibody for immunoprecipitation overnight at 4°C. The immune complexes were pulled down with preblocked protein A slurry and washed in IP washing buffer (100 mM Tris-Cl, pH 9.0 for polyclonal antibody or pH 8.0 for monoclonal antibody, 500 mM LiCl, 1% Nonidet P-40, 1% deoxycholic acid) 4ϳ6 times. The precipitated chromatin was eluted twice by incubation with 150 ml of elution buffer (50 mM NaHCO 3 , 1% SDS) for 15 min at room temperature. The eluted chromatin was reverse crosslinked by incubating with 300 mM NaCl and 1 mg/ml RNase at 68°C overnight. The chromatin was then incubated with 100 mg/ml proteinase K at 45°C for 3 h, followed by extraction first in phenol-chloroform-isoamyl alcohol mix (25:24:1) and then in chloroform-isoamyl alcohol (24:1). 30 ml of 5 M NaCl and 10 mg of glycogen (Roche) were added to each sample, and the DNA was precipitated with 2.5-fold volume of pure ethanol in Ϫ20°C overnight. The DNA was retrieved by centrifugation at 14,000 rpm, 4°C for 20 min, dissolved in 1ϫ TE buffer, and subjected to PCR analysis. The results were calculated from three independent real-time PCR experiments and presented as the mean Ϯ S.E. of the relative fold enrichment as the percentage of input signal after subtraction of the signal from the noantibody mock control as background.
For ChIP assays with a transfected luciferase reporter, Gal4-DBD or Gal4-SMRT was co-transfected with MH100 reporter into HeLa cells. 72-h post-transfection, the cells were harvested FIGURE 2. GPS2 interacts with SMRT in yeast and in vitro. A, SMRT contains two GPS2-interaction domains, GID I-(159 -316) and GID II-(1281-1831). Y2H assays were conducted as described under "Experimental Procedures" and the association between full-length GPS2 and the diagramed SMRT fragments are depicted as positive (ϩ) or negative (Ϫ). Notably, N-CoR fragments containing the equivalent of two conserved domains also interact with GPS2. B, GPS2 interacts with SMRT through the former N-terminal coiled-coil domain (CC). Y2H assays were conducted to map the interaction between SMRT and GPS2 fragments. C, SMRT fragments interact with GPS2 in pull-down assays. Immobilized GST-SMRT fusion proteins were used for pull-down assays with in vitro transcribed and translated FLAG-GPS2. Consistent with Y2H assays, the two GID regions of SMRT were able to interact with GPS2 (lanes 5 and 6). GID I-(159 -316) along with some N-terminal flanking sequence (102-316, lane 4) also interacted with GPS2, whereas amino acids 1060 -1470 (lane 3) did not. As controls, 20% input and GST-alone beads were also included (lanes 1 and 2). D, GST-SMRT fusion proteins interact with the N terminus of GPS2 in vitro. A set of deleted or truncated GPS2 expression constructs corresponding to the Y2H fragments ( Fig. S2) were ectopically expressed in 293 cells, and the cell lysates were used for pull-down assays with GST-SMRT-(102-316) (middle panel) and GST-SMRT-(1281-1833) (bottom panel). Both SMRT fragments were able to pull-down the CC domain-containing GPS2 fragments (lanes 1 and 2); but not others. The top panel is the 10% input to serve as positive control. and subjected to ChIP assays with ␣-Gal4, ␣-SMRT, or ␣-GPS2 antibodies. PCR products were analyzed by agarose gel electrophoresis (see supplemental Materials for details of other materials and methods used in this study).

Isolation of GPS2 as a SMRT Interaction
Protein-Yeast twohybrid (Y2H) screens were employed to isolate proteins interacting with repression domains III and IV (amino acids 1188 -1833) of SMRT. We isolated G-protein suppressor 2 (or GPS2) as a SMRT-interacting protein. We first examined whether SMRT and GPS2 form a stable complex in cells. We carried out immunoprecipitation experiments using HeLa nuclear extracts. Anti-SMRT and anti-GPS2 antibodies were employed to immunoprecipitate endogenous SMRT, HDAC3, and GPS2. As a control, normal serum was included. Fig. 1A shows that both anti-SMRT and anti-GPS2 antibodies coprecipitated GPS2, HDAC3, and SMRT. We further examined the association between GPS2 and SMRT by epitope-tagged overexpression and co-immunoprecipitation experiments. FLAG-GPS2 and HA-SMRT expression plasmids were co-transfected into HEK293 cells. As expected, FLAG-GPS2 was immunoprecipitated by anti-HA antibodies only in the presence of HA-SMRT (Fig. 1B). Furthermore, immunostaining studies indicated that SMRT and GPS2 co-localized in the nucleus of MCF-7 ( Fig.  1C), HeLa, and CV-1 cells (data not shown). Therefore, we conclude that SMRT and GPS2 form stable complexes in mammalian cells.
Mapping of Interaction Domains-Y2H assays were employed to determine the GPS2 interaction domains within SMRT. A panel of pGBT9-Gal4-SMRT deletion and truncation Y2H constructs were co-transformed with pGAD-GPS2, and interactions were detected with liquid LacZ reporter assays. As summarized in Fig. 2A, in addition to the fragment flanking amino acids 1281-1831 (the RD III and RD IV region used originally as bait), we found that GPS2 also interacts with a region at the N terminus of SMRT encompassing amino acids 159 -316 (RD I). This region is highly homologous to N-CoR with an overall 95% identity. As expected, we found that GPS2 also interacts with the N terminus of N-CoR (data not shown). A reciprocal experiment was conducted to map the minimal SMRT interaction domains within GPS2. We found that the first 120 amino acids of GPS2 are important for the SMRT association (Fig. 2B).
GPS2 Binds to SMRT in Vitro-To determine whether GPS2 binds to SMRT in vitro, we carried out GST pull-down assays. In vitro transcribed and translated (TNT) FLAG-GPS2 was analyzed for its ability to bind GST-SMRT fusion proteins. We found that SMRT interacts with GPS2 via two regions including amino acids 159 -316 and 1281-1831 (Fig. 2C).
We Also Mapped SMRT Interaction Domains in GPS2 in Vitro-A set of constructs (corresponding to Y2H clones, Fig.  2B) expressing HA-tagged GPS2 were transfected into HEK293 cells, and the lysates were used in GST pull-down assays with GST-SMRT fusion proteins (102-316 or 1281-1833) (Fig. 2D). Our results indicated that GPS2 interacts with SMRT through its first 120 amino acids. Taken together, our in vitro binding assays are in agreement with the Y2H data (Fig. 2B).
GPS2 Associates with Histone Deacetylases-Because SMRT associates with both class I and class II HDACs (6,9,10,12,13,33), we tested whether GPS2 was able to co-precipitate with histone deacetylase activity. HA-GPS2 was expressed in HEK293 cells and anti-HA immunocomplexes were subjected to histone deacetylase activity assays. As positive controls, HA-HDAC1 and HA-HDAC5 expression plasmids were also included. Fig. 3A shows that anti-HA antibodies immunopre- HA-HDAC1 or HA-HDAC3 and FLAG-GPS2 were singly or co-transfected into HEK293 cells as indicated. Co-immunoprecipitations were performed using anti-HA antibodies followed by Western blotting with ␣-HA or ␣-Flag antibodies (IP, lanes 4 -6). WCEs were included in lanes 1-3 as controls. C, GPS2 associate with HDAC5 but not HDAC7. The experiments were carried out as in B. Lanes 1-3: WCEs; lanes 4 -6: IP with HA antibody. D, HA-HDAC5 recruits endogenous GPS2 to the HDAC5-enriched nuclear speckles. An HA-HDAC5 expression plasmid was transiently transfected into HeLa cells followed by immunofluorescence microscopy using anti-HA and anti-GPS2 antibodies. a, DAPI-stained nuclei; b, endogenous GPS2; c, overexpressed HA-HDAC5; d and e, merged staining with DAPI; f, merged staining of endogenous GPS2 and overexpressed HA-HDAC5. Notice the HDAC5-enriched nuclear speckles pattern in transfected cells in panels c and e and the altered endogenous GPS2 location pattern in panels b and d. The endogenous GPS2 is largely co-localized with HA-HDAC5, indicated by merged yellow color in panel f.

GPS2 Essential for SMRT-mediated Transcriptional Repression
cipitated a significant amount of histone deacetylase activity from the GPS2-containing extracts.
To test whether GPS2 associates with HDACs in mammalian cells, we co-transfected FLAG-GPS2 and HA-tagged class I or class II HDAC expression plasmids and performed immunoprecipitation experiments using anti-HA antibodies. We found that GPS2 co-precipitated with class I HDACs, HDAC1, and HDAC3 (Fig. 3B), inconsistent with the previous study that FLAG-HDAC3 pulled down endogenous GPS2 (34). Furthermore, GPS2 co-precipitated with HDAC5, but not HDAC7 (Fig. 3C). In support, ectopic expression of HDAC5 in HeLa cells recruited endogenous GPS2 to HDAC5-enriched nuclear speckles (Fig. 3D).
GPS2 Is a Transcriptional Corepressor-We next examined whether GPS2 functions as a transcriptional corepressor in transient transfection assays. Increasing amounts of Gal4-GPS2 expression plasmids were co-transfected with MH100 luciferase reporter and ␤-gal internal control for luciferase assays as described under "Experimental Procedures." We found that full-length Gal4-GPS2 potently repressed basal tran-scription in both CV-1 (Fig. 4A) and HEK293 cells (data not shown) in a dose-dependent manner. To further analyze the minimal repression domain in GPS2, we generated deletion and truncation mutants of pGal4-GPS2 (Fig. 4B). Transient transfection assays were carried out to map the repression domain in GPS2. We found that the repression activity is mediated primarily by the first 120 amino acids of GPS2 (Fig. 4, C and D).
To investigate the role of GPS2 in SMRT-mediated repression, we determined whether endogenous GPS2 is recruited to the promoter of the reporter construct along with transfected Gal4-SMRT by ChIP assays (Fig. 5A). Gal4 alone (Gal4-DBD) or Gal4-SMRT expression plasmids were transiently transfected into HeLa cells followed by ChIP assays using anti-Gal4, anti-SMRT, and anti-GPS2 antibodies. Fig. 5A shows that increased recruitment of GPS2 was detected when Gal4-SMRT was transfected, indicating that ectopically expressed Gal4-SMRT recruits endogenous GPS2 to the Gal4 binding sites (compare lane 2 to lane 1). We next tested the effects of overexpression or knockdown of GPS2 on SMRT repression activity by transient transfection reporter assays. We found that overexpression of GPS2 enhanced Gal4-SMRT repression activity in a dose-dependent manner (Fig.  5B), whereas knockdown of GPS2 compromised SMRT repression activity (Fig. 5C). These results support a model in which GPS2 is a functional component of SMRT corepressor complex and that GPS2 is important for SMRT-mediated repression.
Because SMRT is implicated in TAM-mediated transcriptional repression by ER␣, we next examined whether GPS2 is important for TAM-mediated repression on estrogen response element (ERE) by knocking down endogenous GPS2 protein in MCF-7 cells (Fig. 5D). Our data showed that knocking down GPS2 significantly abrogated the ability of TAM to repress ERE reporter activity (Fig. 5E); however, unexpectedly knocking down GPS2 also enhanced estrogen (E2)-induced reporter activity (Fig. 5F). These results indicate that GPS2 is an integral component of the SMRT corepressor complex and suggest that GPS2 may also play a role in ER␣-mediated transcription.
Knockdown of GPS2 Altered the Expression of ER␣ Target Genes-We further determined the effects of GPS2 knockdown on the expression of ER␣ target genes (Fig. 6). We found that E2 activated the expression of the pS2, cyclin D1 (CCND1), pro- gesterone receptor (PR), and c-Myc genes, while TAM and 4OHT repressed the expression of these genes (Fig. 6, siCTRL  groups). Notably, knockdown of GPS2 increased E2-induced expression of the pS2, CCND1, PR, and c-Myc genes (Fig. 6, compare siGPS2 to siCTRL). This observation is consistent with the data shown in Fig. 3E where knocking down GPS2 increased E2-induced reporter gene transcription. Moreover, knockdown of GPS2 compromised the ability of TAM-and 4OHT to repress the expression of pS2, CCND1, PR, and c-Myc. These data indicated that GPS2 participates in both TAM-and E2-mediated transcriptional regulation by ER␣.
It has been well established that tamoxifen induces the association of the SMRT corepressor complex with the promoters of ER␣ target genes (35)(36)(37)(38)(39)(40). Consistent with this notion, we found that in response to tamoxifen treatment, ER␣, SMRT, and GPS2 were transiently recruited to the pS2 gene pro-moter (Fig. 7, A and B). Unexpectedly, we noticed that E2 also induced an association of SMRT and GPS2 with the pS2 promoter (Fig. 7A). Furthermore, we found that E2 treatment resulted in temporal association of ER␣, SMRT, and GPS2 with the pS2 promoter (Fig. 7C).
To further dissect the role of GPS2 in ER␣-mediated transcription in response to antagonist treatment, we carried out ChIP assays and determined the temporal pattern of the recruitment of components of SMRT complexes on the pS2 promoter (Fig. 8) with treatment of 4OHT. 4OHT is a more potent inhibitor than TAM, and therefore the results in Figs. 7 and 8 are not directly comparable. The data in Fig. 8 demonstrate that endogenous ER␣, SMRT, and HDAC3 are recruited to the pS2 promoter following 4OHT treatment with a sharp peak of occupancy at 20 -30 min (Fig. 8, A, B, and  D). N-CoR occupancy was also observed to peak at 30 min (Fig. 8E). By contrast, GPS2 recruitment to the pS2 promoter following 4OHT exposure had a broad pattern of recruitment with a peak at 60 min (Fig. 8C). Notably, all of the corepressors, as well as ER␣ dissociated from the promoter by 90 min. These data suggest that in response to 4OHT, the components of the SMRT complex are transiently recruited to the pS2 promoter and that GPS2 enters the complexes more slowly than other core components.

GPS2 and SMRT Suppress Breast Cancer Cell Proliferation-
Our data suggest that GPS2 is a functional component of SMRT corepressor complexes, and is intimately involved in both 4OHT-and E2-mediated transcription. Because ER␣-mediated cell growth is critical for breast cancer growth, we further determined whether GPS2 affects cell proliferation. To do so, siRNA against luciferase or GPS2 was transiently transfected into MCF-7 cells, and cell proliferation assays were performed. The cells were cultured in charcoal-stripped, phenol redfree medium, supplemented with ethanol (vehicle control), estrogen (E2), or 4OHT. As shown in Fig. 9, knockdown of GPS2 promoted cell proliferation in all experimental conditions. We further examined whether knockdown of SMRT affected MCF-7 cell proliferation under the same conditions (Fig. 10). As observed with GPS2 knockdown, knockdown of SMRT significantly promoted cell proliferation in all experi- Increasing amounts of a GPS2 expression plasmid was co-transfected with either Gal4 alone or Gal4-SMRT for MH100 luciferase reporter assays. The fold repression of reporter is depicted. C, knockdown of GPS2 impairs Gal4-SMRT-mediated repression of MH100 reporter activity. HeLa cells were transiently transfected with siRNA to knockdown GPS2 prior to transfection of gradual increasing amount of Gal4-SMRT as indicated. The fold repression of the reporter is presented relative to the Gal4 vector alone (Ϫ). The siCTRL is a non-targeting siRNA. D, knockdown of GSP2 protein expression by siRNA transient transfection. MCF-7 cells were transiently transfected with siRNA against luciferase (siCTRL) or GPS2 (siGPS2) followed by subcellular fractionation and Western blotting as described under "Experimental Procedures." HDAC1 and GAPDH were used as nuclear (N) and cytoplasmic (C) markers, respectively. The bands were quantified, and the relative densities are shown. E, knockdown of GPS2 abolishes the ability of tamoxifen to inhibit ERE-tk-Luc reporter activity in MCF-7 cells. TAM concentration increased in a gradient as indicated below the figure. Relative reporter activity was normalized to the activity of the same volume of an ethanol vehicle control (Ϫ). Tamoxifen effects were analyzed by one-way ANOVA within siCTRL or siGPS2 groups and the corresponding p values are presented. Notably, GPS2 knockdown cells showed no significant (NS) effects of tamoxifen as compared with controls (p ϭ 0.0059). F, knockdown of GPS2 enhances the ability of estrogen to activate ERE-tk-Luc reporter activity in MCF-7 cells. Reporter activity is presented relative to the ethanol vehicle control (Ϫ). mental conditions tested. Taken together, these data demonstrate that both GPS2 and SMRT are important to maintain normal proliferation rate in MCF-7 and that knockdown of corepressor complex promotes cell growth.

DISCUSSION
It has been previously shown that GPS2 is an integral subunit of N-CoR complex (23). However, the role of GPS2 in ER␣mediated transcriptional regulation and its biological functions in MCF-7 breast cancer cells remain elusive. In this report, we demonstrate that GPS2 is also a stable component of the SMRT corepressor complexes. Overexpression of GPS2 potentiated, whereas knockdown alleviated SMRT-mediated repression of reporter activities, suggesting that GPS2 is functional component of SMRT corepressor complex. Because SMRT complex is one of the key effectors in ER␣-mediated gene repression in the presence of 4OHT, we showed that GPS2 is important for optimal TAM-or 4OHT-mediated repression of ERE reporter activity or ER␣ target genes. We are also the first to characterize the temporal pattern of recruitment of GPS2-containing SMRT complex to the ER␣ target gene promoter in response to 4OHT treatment. These data provide mechanistic insights into how SMRT corepressor complexes mediate transcriptional repression of ER␣ target genes. Unexpectedly, our data also indicated that GPS2-containing SMRT complexes are involved in E2-induced ER␣ target gene expression. Finally, we found that GPS2 and SMRT are generally required to maintain the normal growth rate of MCF-7 breast cancer cells. Taken together, we conclude that GPS2 is a functional component of SMRT corepressor complex, which is important for ER␣-mediated transcriptional regulation. Our findings highlight mechanistic understanding of SMRT corepressor complexes in ER␣-mediated gene repression, adding knowledge to the biological functions of GPS2.
The Role of GPS2 in Transcriptional Regulation-When tethered to a promoter region, GPS2 potently represses basal transcription (Fig. 4). This is possibly due to its association with SMRT and/or HDACs (Figs. 1-3). Interestingly, overexpression of GPS2 potentiates SMRT repression, whereas knocking down GPS2 relieved SMRT repression activity, indicating that GPS2 is required for optimal SMRT-mediated repression activity (Fig. 5). GPS2 does not harbor a nuclear receptor-interacting motif (LXXLL) and probably does not bind ER␣ directly. This suggests that GPS2 was recruited to the ER␣ promoter indirectly through its association with the SMRT or N-CoR complexes. Two functional assays strongly suggest that GPS2 is an important regulator in ER␣-mediated transcriptional regulation. First, using transient transfection reporter assays, we demonstrated that knockdown of GPS2 compromised TAM-mediated repression and enhanced E2-induced activation of an ER␣ reporter activity (Fig. 5). Secondly, knockdown of GPS2 alleviated 4OHT-repressed and potentiated E2-induced expression of several ER␣ target genes (Fig. 6). These observations indicate that GPS2 is not merely a submissive component of the corepressor complexes, but also plays an active role in transcriptional regulation by ER␣, either in the presence of agonist or antagonist.
The Role of GPS2 in ER␣-mediated Transcription-We have previously shown that knockdown of SMRT blocks TAMmediated inhibition of the expression of CCND1 PR, and c-Myc, all of which are ER␣ target genes, and promotes cell growth in the presence of TAM in BT-474 breast cancer cells (41). Furthermore, knockdown of SMRT, or its related protein N-CoR, abrogates TAM-mediated inhibition of expression of some ER␣ target genes in MCF-7 cells (42). These data suggest the corepressor complexes are important mediators in TAM-mediated transcriptional repression by ER␣. In support of this hypothesis, knockdown of GPS2 alleviated the ability of TAM to inhibit ERE reporter activity (Fig. 5) and abolished the ability of 4OHT to repress ER␣ target gene expression (Fig. 6). In addition, GPS2 was recruited to an ER␣ target gene promoter along with ER␣ and other components of SMRT corepressor complex (Figs. 7-8). These data suggest that GPS2 is a critical component in SMRT corepres-  DECEMBER 25, 2009 • VOLUME 284 • NUMBER 52 JOURNAL OF BIOLOGICAL CHEMISTRY 36401 sor complexes and that reduced levels of GPS2 causes a tamoxifen resistance-like phenotype.

GPS2 Essential for SMRT-mediated Transcriptional Repression
Intriguingly, the components of the SMRT complexes were also recruited to the pS2 promoter in an E2-dependent manner (Fig. 7, A and  C), suggesting a role of the SMRT complexes in E2-induced gene activation, in addition to their role in gene repression. Indeed, knockdown of GPS2 further enhanced E2-induced reporter activity (Fig.  5F) and the expression of some ER␣ target genes in MCF-7 cells (Fig. 6). Although the mechanism is not clear, these observations suggest a direct involvement of GPS2 and maybe SMRT corepressor complex, in limiting E2-induced transcriptional activation. However, E2-and 4OHT-induced recruitment of the corepressor complex may possess distinct components or structural arrangements. Alternatively, the enhanced E2 induction is due to the de-repression of SMRT corepressor by knockdown of GPS2. Our finding that E2 induced the association of corepressors to an ER␣ targeted promoter raises the possibility that 4OHT may be capable of promoting the association of coactivators to ER␣-targeted promoters.
Our GST pull-down and Y2H assays indicated that GPS2 interacts with two independent SMRT domains (Fig. 2). One possibility is that GPS2 interacts with these two domains sequentially. This hypothesis predicts a conformational reconfiguration within the complex on the pS2 promoter. An alternate possibility is that there is more than one GPS2-containing SMRT complex. Our current data do not address this issue. Notably, GPS2 is not only stably associated with SMRT complex, but has been shown to associate with N-CoR (23). There is evidence indicating that SMRT and N-CoR possess distinct functions in ER␣-mediated transcription (43)(44)(45)(46), and knockdown of SMRT or N-CoR alone relieves repression of ER␣ target genes (42). These observations suggest that SMRT/GPS2 and N-CoR/  GPS2 complexes may function independently. How these complexes distinctively modulate ER␣ transcriptional regulation warrant further investigation.
Recruitment of Corepressors to ER␣ Target Gene Promoter-With E2 hormone treatment, ER␣ binds to pS2 promoter in a cyclic manner, thus directing a variety of transcription factors, mostly co-activators, to the promoter in a similar cyclic, dynamic patterns (32). We expanded this to 4OHT treatment condition and examined the recruitment patterns of SMRT corepressor complexes to the pS2 promoter. Our data suggest that 4OHT is capable of promoting the recruitment of corepressor complex in a temporal manner. We believe this is direct evidence for the role of GPS2 in SMRT-mediated ER␣ target gene repression.
The Role of Corepressor Complex in Breast Cancers-SMRT and N-CoR are important effectors in the treatment of ER␣positive breast cancer by tamoxifen. In tamoxifen-responsive breast cancer cells, tamoxifen-bound ER␣ recruits corepressor complexes to inhibit target gene expression. In addition, decreased levels of corepressors correlate with acquired tamoxifen resistance (39,42). Our findings suggest that GPS2 is required for the optimal antagonism of tamoxifen in the transcriptional repression of ER␣ target genes. Our ChIP data imply that abnormal recruitment of the corepressor complexes may contribute to the loss-of-antagonism of tamoxifen. In addition, knockdown of GPS2 promotes the agonism of E2 hormone on the reporter activity and several ER␣ target gene transcriptions. Based on other studies, we hypothesize that the efficacy of hormone or drug on the ER␣-mediated gene transcription largely depends on the balance between coactivators and corepressors (36,(47)(48)(49)(50). For example, high expression of the coactivators favors the agonist effect of tamoxifen (51,52). In contrast, overexpression of SMRT or N-CoR promotes the antagonist activity of tamoxifen (49). Knockdown of GPS2 may shift the balance toward coactivators, thus promoting E2-mediated ER␣ target gene transcription.
Whereas GPS2 and SMRT regulate ER␣-mediated transcriptional regulation, the mechanisms by which they suppress MCF-7 cell proliferation appear to be independent of ER␣ action. Knockdown of GPS2 or SMRT promoted MCF-7 cell proliferation, regardless of the presence of ER␣ ligands (Figs. 9 and 10). Several other transcription factors including NF-B and AP1 have been shown to mediate their repressive activity by interacting with SMRT complex (53). As such, SMRT complex may modulate transcription activity of NF-B and AP1 to control MCF-7 cell proliferation, even in the absence of ER␣ ligands. Alternatively but not exclusively, the SMRT complex may possess non-genomic activity to regulate cell cycle progression (for a review, see Ref. 15).
In conclusion, our studies demonstrated that GPS2 is an integral component of the SMRT corepressor complex that is important for the normal regulation of ER␣-mediated transcriptional control and a suppressor for MCF-7 breast cancer cell proliferation.