Silencing Mediator for Retinoid and Thyroid Hormone Receptor and Nuclear Receptor Corepressor Attenuate Transcriptional Activation by the β-Catenin-TCF4 Complex*

β-Catenin is a multifunctional mediator of cellular signaling and an oncogene. Nuclear β-catenin, when complexed with members of the T-cell factor (TCF)/leukocyte enhancer factor family of DNA-binding proteins, mediates transcriptional activation important for embryonic development and adult cell homeostasis. Deregulation of intracellular levels of β-catenin is an early event in the development of a variety of cancers. We observed that the proteins silencing mediator for retinoid and thyroid hormone receptor (SMRT) and the nuclear receptor corepressor (NCoR) are negative regulators of transcription induced by the β-catenin-TCF4 complex. Overexpression of SMRT and NCoR attenuated the transcription of β-catenin-TCF4-specific reporter gene and of CCND1, an endogenous β-catenin target gene. Knockdown of endogenous SMRT or NCoR by short interfering RNA augmented the β-catenin-TCF4-mediated reporter gene expression. Glutathione S-transferase pulldown experiments showed there was a direct physical association of SMRT and NCoR with both β-catenin and TCF4. DNA-protein interaction studies revealed that the interactions between either SMRT or NCoR and β-catenin or TCF4 occurred at the promoter regions of CCND1 and other target genes. These findings demonstrate an important role for corepressors SMRT and NCoR in the regulation of β-catenin-TCF4-mediated gene transcription.

␤-Catenin is a multifunctional mediator of cellular signaling and an oncogene. Nuclear ␤-catenin, when complexed with members of the T-cell factor (TCF)/leukocyte enhancer factor family of DNA-binding proteins, mediates transcriptional activation important for embryonic development and adult cell homeostasis. Deregulation of intracellular levels of ␤-catenin is an early event in the development of a variety of cancers. We observed that the proteins silencing mediator for retinoid and thyroid hormone receptor (SMRT) and the nuclear receptor corepressor (NCoR) are negative regulators of transcription induced by the ␤-catenin-TCF4 complex. Overexpression of SMRT and NCoR attenuated the transcription of ␤-catenin-TCF4-specific reporter gene and of CCND1, an endogenous ␤-catenin target gene. Knockdown of endogenous SMRT or NCoR by short interfering RNA augmented the ␤-catenin-TCF4-mediated reporter gene expression. Glutathione S-transferase pulldown experiments showed there was a direct physical association of SMRT and NCoR with both ␤-catenin and TCF4. DNA-protein interaction studies revealed that the interactions between either SMRT or NCoR and ␤-catenin or TCF4 occurred at the promoter regions of CCND1 and other target genes. These findings demonstrate an important role for corepressors SMRT and NCoR in the regulation of ␤-catenin-TCF4-mediated gene transcription.
␤-Catenin is an important nuclear effector of the Wnt signaling pathway that plays a critical role in cell fate determination, tissue homeostasis, and tumorigenesis (1). ␤-Catenin is sequestered in the cytoplasm and functions at both the cell membrane and the nucleus. At the cell membrane ␤-catenin binds to a complex that regulates E-cadherin-mediated epithelial cell adhesion (2). In the nucleus ␤-catenin binds to members of the T-cell factor (TCF) 2 /leukocyte enhancer factor (LEF) family of proteins to form a heterodimeric transcription complex that activates gene expression downstream from WNT signals (1). In the absence of WNT signals, the cytosolic pool of ␤-catenin is bound to axin and to the adenomatous polyposis coli protein that favor phosphorylation of ␤-catenin at N-terminal Ser/Thr residues by casein kinase I␣ and glycogen synthase kinase 3␤ leading to ubiquitination and degradation by the 26 S proteasome (3).
When nuclear ␤-catenin is of low abundance, TCF/LEF proteins that have DNA binding domains, but lack transcriptional activation function, act as transcriptional repressors, recruiting Groucho/transducin-like enhancer proteins, which in turn recruit histone deacetylase (HDAC) to repress the target gene promoters (4,5). Changes in chromatin structure are required to relieve the transcriptional repression mediated by these protein complexes. In response to WNT signals, ␤-catenin phosphorylation is attenuated, leading to increased ␤-catenin levels in the cytoplasm and the nucleus, where ␤-catenin binds to TCF/LEF, displaces the corepressor complexes containing HDACs, and enhances transcriptional activation of WNT-responsive target genes (6 -8). Transcriptional activation by ␤-catenin-TCF/LEF is mediated by recruitment of basal transcription machinery (9) and is enhanced by interaction with many other coactivators, such as the histone acetylase CBP/ p300 (10,11), the p160 coactivator GRIP1 (12), a component of the SWI/SNF chromatin-remodeling complex, Brg-1 (13), and the Legless-Pygopus complex (14).
Even though SMRT and NCoR were initially linked with transcriptional repression by nuclear receptors, there is evidence that they also play roles in transcriptional repression by other transcription factors, such as the leukemia chimeric fusion protein PLZF (34,35), Notch-binding protein CBF-1 (36), members of the AP (activator protein)-1 family (37-39), NF-B factors (38 -40), homeodomain-containing proteins (41)(42)(43), Eto (44,45), and the E-26-transforming specific domain proteins Tel (46,47) and MEF2c (16). The many associations of SMRT and NCoR suggest that the corepressors may play diverse roles in mediating active transcriptional repression during development. Indeed, deletion of the murine NCoR gene resulted in embryonic lethality and severe developmental defects in the lymphocytic and erythropoietic lineages and in the central nervous system (48).
Here we show that the SMRT and NCoR directly interact with ␤-catenin and TCF4, both in vitro and in vivo, and that overexpression of either SMRT or NCoR attenuates the transactivation of ␤-catenin-TCF4 target genes, such as CCND1 by a mechanism dependent on the TCF4-binding element (TBE). Knockdown of either endogenous SMRT or NCoR expression augments the transactivation of ␤-catenin-TCF4 reporter genes. Thus, our results suggest that both SMRT and NCoR function as corepressors of ␤-catenin-TCF-mediated signaling pathway by negatively regulating the expression of WNT target genes.

EXPERIMENTAL PROCEDURES
Plasmids-The Renilla null luciferase reporter was purchased from Promega Corp. (Madison, WI), and pFR-LUC was from Stratagene (La Jolla, CA). TCF4 plasmid and OT-Luc reporter vectors were provided by Kenneth Kinzler (The Johns Hopkins Oncology Center, Baltimore, MD). ␤-Catenin, GAL4/ ␤-catenin, GAL4/␤-catenin-ARM, GAL4/␤-catenin⌬N, and GAL4/␤-catenin⌬C have been described previously (49,50 constructs. The target sequences were 5Ј-AAGAAGGATC-CAGCATTCGGA-3Ј for NCoR and 5Ј-AAGCTGAAGAA-GAAGCAGCAA-3Ј for SMRT. Cell Culture and Transfection-Monolayer cultures of CV-1, 293T, and SW480 cells were grown as described previously (49,50). For transient transactivation assays, cells were plated onto 24-well plates at a density of 4 ϫ 10 4 /well, and cells were transfected for 16 h using Lipofectamine (Invitrogen) as recommended by the supplier. For each well of a 24-well plate, we used 100 ng of reporter (OT-Luc, FR-LUC or CCND1-Luc) and 10 ng of Renilla plus various combinations of other expression vectors. Equimolar amounts of expression vectors (pCMX) lacking NCoR or SMRT were included to keep the molarity of each vector constant, with the total transfected DNA brought to 300 ng/well with pBSK ϩ unless otherwise indicated. After transfection, the cells were cultured for an additional 24 h in complete medium and harvested in 1ϫ Passive Lysis Buffer (100 l/well; Promega). Cell lysates (30 l) were used to assay for luciferase activity using the Dual-Luciferase assay system from Promega according to the supplier. The data were then normalized for the cotransfected Renilla activity.
GST-Pulldown Assays-GST-pulldown assays were performed as described previously (49,50). Briefly, GST fusion proteins (GST-␤-catenin, GST-SMRT, and GST-NCoR) were expressed in Escherichia coli and isolated as described before. 35 S-Labeled ␤-catenin, VP16/SMRT.ID, VP16/NCoR.ID, and TCF4 mutants were produced in vitro using TNT-coupled/reticulocyte lysate system (Promega) according to the manufacturer's recommendations in the presence of [ 35 S]methionine (Amersham Biosciences). Sonicated bacterial lysates containing overexpressed GST or GST fusion proteins were linked to glutathione-Sepharose beads. Immobilized GST or GST fusion proteins were then incubated overnight at 4°C with either 35 Slabeled proteins or cell lysates. After extensive washing, the immobilized proteins were removed from the beads by heating at 90°C for 5 min in 40 l of 2ϫ SDS loading buffer. The proteins were then separated on 4 -20% SDS-polyacrylamide gels, and the bound proteins were visualized by either autoradiography for 35 S-labeled proteins or Western blotting for cell lysates.
In Re-ChIP experiments, sonicated and precleared samples were first immunoprecipitated with anti-TCF4 antibody as described above. The antibody-bound chromatin fragments were then eluted by incubation for 30 min at 37°C in 50 l of 10 mM dithiothreitol. After centrifugation, the supernatant was diluted 20 times with RIPA buffer and subjected again to the ChIP procedure with either normal IgG or antibodies against ␤-catenin, SMRT, or NCoR.
Cell Proliferation Assay-SW480 cells were plated onto 48-well plates (1.5 ϫ 10 4 per well) and transfected with 150 ng/well of control vector pSilencerH1 3.0 or pSilencer-NCoR, pSilencer-SMRT, or both. At 16 h after transfection, fresh complete medium containing 10% fetal bovine serum was added, and cells were cultured for 5 days. Cells were visualized by crystal violet staining and quantified by measurement of absorbance at A 570 .
Statistical Analysis-Unless otherwise noted, values shown represent mean Ϯ S.D. The differences between groups were analyzed for statistical significance by the two-tailed Student's t test using the program GraphPad Prism software version 4.02 (GraphPad Software, San Diego, CA). p Ͻ 0.05 was considered significant. In figures the following symbols are used to represent p values: *, p Ͻ 0.05; **, p Ͻ 0.01; ***, p Ͻ 0.001; and ****, p Ͻ 0.0001.

SMRT and NCoR Affect TCF4-mediated Reporter Gene
Expression-To study the effect of corepressors SMRT and NCoR on TCF4-mediated transcription, we first analyzed the effect s-SMRT and NCoR have on TCF4-dependent reporter activity in CV-1 cells. S-SMRT (originally called SMRT or TRAC-2, and in our paper we still use SMRT instead of s-SMRT) is the truncated form of the full-length SMRT (called SMRT␣ or SMRTe). S-SMRT lacks most of the N-terminal repression/cofactor docking sites found in SMRT␣ (53) (Fig. 1). CV-1 cells were cotransfected with the OT-Luc reporter together with ␤-catenin and increasing amounts of either NCoR or SMRT (0 -200 ng/well). Both SMRT and NCoR significantly inhibited TCF4-mediated transactivation of the reporter in a dose-dependent manner ( Fig. 2A). SMRT appeared to be more potent because maximal inhibition was achieved with 50 ng/well of the SMRT expression plasmid compared with 100 ng/well of the NCoR expression plasmid. The lesser inhibitory effect of NCoR on ␤-catenin/TCF4-mediated transactivation as compared with SMRT might have been due to a lower expression level of NCoR as compared with SMRT (Fig. 2B). A fixed amount of SMRT expression plasmid inhib-ited reporter expression driven by increasing amounts of ␤-catenin (Fig. 2C). SMRT also inhibited TCF4-mediated transcription from the Ϫ1745 CCND1 promoter, showing that the effect was seen with more than one reporter construct (Fig. 2D). In contrast NCoR expression had no effect on the CCND1-driven reporter construct. Again, this might have been due to the low expression level of transfected NCoR in this system.
Physical Interaction of SMRT and NCoR with ␤-Catenin-To evaluate whether there was a direct physical interaction between either SMRT or NCoR and ␤-catenin, we performed a modified mammalian twohybrid assay. CV-1 cells were transfected with Gal4/␤-catenin (Gal4/ ␤cat) constructs that contained full-length or truncated ␤-catenin, corepressors SMRT, NCoR (Fig. 1), or the corresponding empty vector as a control, and the reporter construct pFR-Luc. The Gal4 segment of the fusion protein binds to the FR-Luc reporter, and the ␤-catenin fragment initiates assembly of the transcriptional complex. The fulllength and truncated Gal4/␤-catenin constructs all were active at initiating transcription, and each was inhibited by either SMRT or NCoR (Fig. 3A). Because each of the Gal4/ ␤-catenin constructs contained the central ␤-catenin armadillo repeat region, which is important for interaction of ␤-catenin with TCF4, the data suggested that the armadillo repeat region may also be the domain for the association of ␤-catenin with either SMRT or NCoR. In this experiment we again observed that the apparent effect of SMRT was greater than that of NCoR because of the lower expression level of NCoR as compared with that of SMRT.
We performed a modified mammalian two-hybrid experiment with VP16/SMRT.ID or VP16/ NCoR.ID. VP16/SMRT.ID and VP16/NCoR.ID are the C-terminal regions of corepressors that contain the interaction domains (ID) responsible for the association with nuclear receptors (54,55). Gal4/␤catFL, Gal4/␤cat⌬N, and Gal4/  ␤cat⌬C-mediated reporter activities could be enhanced significantly by both VP16/SMRT.ID and VP16/NCoR.ID (Fig. 3B) even though the VP16/SMRT.ID induced a higher level of reporter activity because of a relatively higher expression level of the VP16/SMRT.ID fusion protein (Fig. 3B). The Gal4/␤catenin fusion proteins were expressed at comparable levels in this system (Fig. 3C). These results indicated that all these ␤-catenin constructs are able to interact with the C-terminal regions of corepressors SMRT and NCoR.
To confirm the associations of VP16/SMRT.ID and VP16/ NCoR.ID with ␤-catenin, we did a transient transactivation assay by cotransfection of cells with the CCND1 promoter reporter, together with increasing amount of either VP16/ SMRT.ID or VP16/NCoR.ID. In this assay, interaction between ␤-catenin and the VP16/SMRT.ID or VP16/NCoR.ID was reflected by increased reporter gene activity. We found that the reporter activities were enhanced by VP16/SMRT.ID and VP16/NCoR.ID (Fig. 3D). Because the expression level of VP16/NCoR.ID was lower than that of VP16/SMRT.ID (Fig.  3E), it is not surprising that the reporter activity induced by VP16/NCoR.ID was weaker than that by VP16/SMRT.ID. The reason that VP16/NCoR.ID did not induce a dose-dependent increase in the reporter activity in this experiment as VP16/ SMRT.ID did was unknown, most probably because of the relatively low level expression of the VP16/NCoR.ID fusion proteins despite adding the maximum amount we could in the transfections.
The interactions of either SMRT or NCoR and ␤-catenin were further examined by GST-pulldown assay using GST-␤cat constructs and in vitro translated SMRT or NCoR constructs. Either VP16/ SMRT.ID or VP16/NCoR.ID was pulled down by GST-␤-catenin but not by GST alone (Fig. 4A), confirming the results of the modified mammalian two-hybrid assay in Fig. 3D. Equal amounts of VP16/NCoR.ID and VP16/SMRT.ID gave rise to equal binding in this assay. The GST-pulldown assay indicates that the physical associations of NCoR and SMRT with ␤-catenin are comparable and are consistent with our interpretation that differences in the ␤-catenin/TCF4-mediated transactivation assay were due primarily to the lower expression of the NCoR vector as compared with SMRT. We also performed a GSTpulldown assay with GST-NCoR.ID and ␤-catenin from cell lysates. GST-NCoR.ID is a fusion protein between GST and the C terminus of NCoR (residues 1944 -2453). With this fusion protein, we were able to demonstrate the binding of NCoR to ␤-catenin from either LNCaP or 293T cells. The binding of ␤-catenin from LNCaP cells to GST-NCoR.ID was stronger than that from 293T cells reflecting the disparity in input levels of protein (Fig. 4B).
To determine the regions of ␤-catenin required for the association with corepressor IDs, GST fusion constructs of different ␤-catenin fragments (Fig. 4C) were used. GST-␤catenin-(120 -683) bound strongly to VP16/SMRT.ID and VP16/NCoR.ID compared with the fusion construct with full-length ␤-catenin (Fig. 4D). The autoradiography exposure time for Fig. 4D was much shorter than that for Fig. 4A, accounting for the apparent differences in full-length ␤-catenin binding between the two panels. Except for GST-␤-catenin-(120 -683) and full-length ␤-catenin, all other GST-␤-catenin fusion constructs did not bind to either SMRT or NCoR. GST-␤-catenin-(120 -683) contained the armadillo repeat units plus about 20 amino acids additionally at both the N and C termini. These data suggest that armadillo repeat regions are required for the association between ␤-catenin and the corepressors IDs.
Physical Interaction of SMRT and NCoR with TCF4-SMRT and NCoR may inhibit the ␤-catenin/TCF4-mediated transactivation by direct association with either ␤-catenin, with TCF4, or with both. To determine whether SMRT and NCoR might also be able to interact directly with TCF4, we performed  Fig. 5A demonstrated that both GST-SMRT.ID and GST-NCoR.ID bound TCF4. It is noteworthy that the binding of GST-SMRT.ID to TCF4 was much stronger than that of GST-NCoR.ID.
To determine the regions of TCF4 required for the association with the corepressors, different subgenomic fragments of TCF4 (Fig. 5B) were subjected to in vitro transcription/translation, and the products were used in pulldown experiments with GST-NCoR.ID. Full-length TCF4 and TCF4-(201-395) that included the TCF4 high mobility group domain bound GST-NCoR.ID. Neither the N-terminal region containing the ␤-catenin binding domain (BD) nor the C-terminal region of TCF4 bound GST-NCoR.ID (Fig. 5C).

Endogenous SMRT and NCoR Are Recruited to the TCF4binding Sites in Promoters of Endogenous Target
Genes-CCND1 is one of the well documented transcription targets of ␤-catenin. To obtain direct evidence that endogenous core-pressors SMRT and NCoR were recruited to TCF4-binding elements (TBE) in the CCND1 promoter, chromatin immunoprecipitation (ChIP) analysis was performed. The cross-linked and sheared nuclear fractions from SW480 cells were subjected to immunoprecipitation with antibodies directed against SMRT, NCoR, TCF4, and ␤-catenin, and the precipitated DNA was analyzed by PCR amplification using a primer set specific for DNA fragment encompassing the TBE sites of the CCND1 promoter. PCR amplification using the primer set produces a 402-bp product from the CCND1 promoter with the sheared genomic DNA as a template (Fig. 6A, input). TCF4 and ␤-catenin were preferentially recruited to the TBE sites in the CCND1 promoter. Similarly, corepressors SMRT and NCoR were also preferentially recruited to the same region in CCND1 promoter. Both mouse and goat IgG controls showed equally low background levels. To extend our findings, a different set of ChIP assays was performed with SW480 cells, and it was found that SMRT and NCoR could also be recruited to the TBE sites in the AXIN2 gene promoter, another well documented target gene of Wnt/␤-catenin signaling (Fig. 6A). To confirm that SMRT and NCoR were not binding to the genomic sequences in a nonspecific manner, we did ChIP assays with PCR primers specific for the upstream region of the GAPDH promoter sequence that does not contain any known TBE sites. There  were no PCR products with any antibodies tested when the GAPDH promoter was used as a target (Fig. 6A). These data indicate that corepressors SMRT and NCoR specifically bound to the CCND1 TBE regions in vivo.
Because our mammalian two-hybrid and GST-pulldown assays showed that corepressors SMRT and NCoR could physically associate with both TCF4 and ␤-catenin, it is reasonable to conclude that corepressors SMRT and NCoR bound indirectly to the TBE sites of the CCND1 promoter via TCF4 and/or ␤-catenin proteins. To demonstrate this binding relationship, ChIP re-ChIP assays were performed. Anti-TCF4 or anti-␤catenin antibodies were used for the primary immunoprecipitation. The eluted samples were then subjected to the secondary immunoprecipitation by antibodies against ␤-catenin, TCF4, SMRT, or NCoR. As shown in Fig. 6, B and C, TCF4 and ␤-catenin occupied the same complex on the TBE sites in the promoters of all target genes tested (CCND1, AXIN2, and DKK1). When primary immunoprecipitation was done with anti-TCF4 antibody followed by secondary immunoprecipitation with either SMRT or NCoR antibodies, SMRT and NCoR were found to be complexed with TCF4 in the promoter region of AXIN2. However, in the CCND1 promoter, no clear PCR product was seen when secondary immunoprecipitation was done with NCoR antibody (Fig. 6B). We believe that this result reflects the comparatively weak interaction between TCF4 and NCoR as shown by GST-pulldown assays (Fig. 5A) or that the association of TCF4-NCoR is promoter-specific. When SMRT and NCoR antibodies were used after primary immunoprecipitation with ␤-catenin antibody, specific PCR products were seen in both DKK1 and AXIN2 promoter regions (Fig. 6C), indicating that both SMRT and NCoR reside in the same complex with ␤-catenin in the TBE sites of these two target genes.
The above ChIP re-ChIP results demonstrating that both SMRT and NCoR reside in the same complex with TCF4 and/or ␤-catenin in the promoter regions of at least one tested target gene largely exclude the possibility that the corepressors SMRT and NCoR might be able to compete for DNA binding with TCF4/␤-catenin. To further examine this, we did a time course study to test the dynamic recruitment of TCF4, ␤-catenin, SMRT, and NCoR in 293T cells. Treatment of 293T cells with 20 mM LiCl induced a rapid stabilization and accumulation of nuclear ␤-catenin (Fig. 7A). Correlated with the increase in nuclear ␤-catenin, treatment of cells with LiCl caused a progressive increase in ␤-catenin binding to the TBE sites in the promoter regions of all target genes tested (CCDN1, c-MYC, DKK1, and AXIN2, Fig. 7, B-E). Not surprisingly, TCF4 could be constitutively recruited to the TBE sites. Interestingly, the recruitment of SMRT and NCoR to the TBE sites of all tested target genes was not influenced by the treatment with LiCl. This result indicated that the recruitment of SMRT and NCoR to the TBE sites may not be due primarily to ␤-catenin. Fig. 7 also indicated that corepressors SMRT and NCoR were not released from the TBE sites upon ␤-catenin binding, again suggesting that the binding of TCF4, ␤-catenin, and corepressors SMRT and NCoR was not mutually exclusive.
Overexpression of SMRT and NCoR Inhibited the Expression of Endogenous CCND1-We next wanted to demonstrate the effects of SMRT and NCoR on CCND1 expression in SW480 cells that have constitutively activated ␤-catenin. SW480 cells were transfected with increasing amounts of either SMRT or NCoR, and CCND1 mRNA and protein levels were determined by semi-quantitative RT-PCR and Western blot analysis. As we observed in CV-1 cells, the expression level of transfected SMRT in SW480 cells was again much higher than that of FIGURE 6. Recruitment of SMRT and NCoR to the endogenous TCF4-binding sites in promoters of target genes. A, ChIP analysis was performed with chromatin isolated from SW480 cells, and PCRs were performed using specific primers covering the TBE sites of the CCND1 and AXIN2 promoters or primers amplifying the promoter region of GAPDH, which does not contain any known TBE. PCR products were resolved by agarose gel and stained with ethidium bromide (left panel), and the percentage of the signal corresponding to each PCR product was determined by densitometric scanning as indicated on the right panel that is representative of three different experiments. B and C, ChIP Re-ChIP assays. Chromatin prepared from SW480 cells was first immunoprecipitated (1st IP) with TCF4 antibody (B) or ␤-catenin antibody (C). After extensive washing, immunoprecipitates were eluted by incubation with 10 mM dithiothreitol at 37°C for 30 min, diluted 20 times with RIPA buffer, followed by a second immunoprecipitation (ChIP Re-IP) with the indicated antibodies as described under "Experimental Procedures." Specific primer sets covering the TBE sites of the CCND1, AXIN2, and DKK1 promoters were used for PCR. NCoR (Fig. 8A). The expression level of endogenous CCND1 mRNA in SW480 cells was markedly decreased by both SMRT and NCoR, whereas the levels of ␤-actin mRNA were unaffected (Fig. 8B). Western blot analyses showed that expression of either SMRT or NCoR could significantly inhibit the expression of cyclin D1. To ask whether expression of exogenous SMRT and NCoR had any significant effect on the expression of ␤-catenin and TCF4, the same membranes were exposed to antibodies against ␤-catenin and TCF4. Expression of either exogenous SMRT or NCoR had no significant effect on the expression of endogenous ␤-catenin and TCF4 (Fig. 8C). These data suggested that the corepressors SMRT and NCoR modulated the ␤-catenin/ TCF4-mediated transcription because of direct interaction with TCF4 and ␤-catenin rather than via alterations in the expression levels of TCF4 or ␤-catenin.

Knockdown of Endogenous SMRT and NCoR Increased TCF4
Transcriptional Activity and Promoted Cell Growth-To address further the potential effect of endogenous corepressors SMRT and NCoR on TCF4 transcriptional activity, we knocked FIGURE 7. SMRT and NCoR constitutively occupy the Wnt/␤-catenin target gene promoters. A, LiCl induced ␤-catenin (␤-cat) stabilization and nuclear accumulation in 293T cells. Nuclear extracts were prepared from 293T cells after treatment with 20 mM LiCl for the indicated times, and Western blot analysis of ␤-catenin is shown. As a loading control, Western blot analysis of histone H1 was done. B-E, SMRT and NCoR constitutively occupied on Wnt/ ␤-catenin target gene promoters. 293T cells were treated with 20 mM LiCl for 0 -4 h. Chromatin was isolated and sonicated as described, and ChIP assays were performed. Specific primer sets covering TBE sites of CCND1 (B), c-MYC (C), DKK1 (D), and AXIN2 (E) were used for PCR. PCR products were resolved by agarose gel ethidium bromide staining (left panel), and the percentage of the signal corresponding to each PCR product was determined by densitometric scanning and is indicated on the right panel. Each data point is the average of duplicates from a representative experiment. Experiments were repeated twice.

FIGURE 8. SMRT and NCoR inhibit the expression of endogenous CCND1.
A, SMRT and NCoR were expressed in SW480 cells. Cells were plated onto 6-well plates and transfected with increasing amounts of SMRT or NCoR (0, 500, 1000, and 2000 ng/well). Western blot analysis was performed with anti-FLAG M2 and anti-SMRT antibodies to detect NCoR and SMRT, respectively. B, CCND1 mRNA expression is down-regulated by overexpression of SMRT or NCoR. SW480 cells were plated onto 6-well plates and transfected with increasing amounts of SMRT or NCoR. After 48 h of incubation, total RNA was isolated, and the expression level of CCND1 and ␤-actin mRNAs was determined by semiquantitative RT-PCR as described under "Experimental Procedures." PCR products were detected by ethidium bromide. Similar results were obtained from two additional experiments. C, CCND1 protein levels were decreased in SW480 cells overexpressing SMRT or NCoR. SW480 cells were transfected with increasing amounts of SMRT or NCoR as described in A and B. After 48 h of incubation, cell lysates were prepared, and the expression level of CCND1, ␤-catenin, TCF4, and ␤-actin was determined by SDS-PAGE followed by Western blotting as described under "Experimental Procedures." Similar results were obtained from one additional experiment. down the expression of endogenous SMRT and NCoR with siRNA in 293T cells, and the ␤-catenin/TCF4-mediated transcriptional activities were monitored by transactivation assays. The SMRT and NCoR siRNA constructs used in this experiments could knock down the expression of endogenous corepressors SMRT and NCoR (17) (Fig. 9A). In the transactivation assay, the reporter activity was very low if ␤-catenin was not overexpressed or activated (Fig. 9B, control). Therefore, we used LiCl, a commonly used inhibitor of glycogen synthase kinase-3␤ (56), to abrogate ␤-catenin phosphorylation. When cells were treated with 20 mM LiCl overnight, the OT-Luc reporter activity was induced dramatically (Fig. 9B, LiCl). In the presence of LiCl, the TCF4-mediated reporter activity was further augmented by knockdown of either SMRT or NCoR. A similar experiment was performed with the CCND1 promoter-driven reporter (Fig. 9C). Knockdown of either endogenous SMRT or NCoR increased the CCND1 promoter activity in the absence or presence of LiCl. Taken together, these results provided further evidence that endogenous corepressors SMRT and NCoR play a role in the regulation of ␤-catenin signaling.
Previous studies have shown that down-regulation of ␤-catenin by either siRNA or antisense oligonucleotides could inhibit SW480 cell growth (57,58). We next tested the effect of corepressor knockdown by siRNA on cell growth of SW480 cells. As shown in Fig. 9, D and E, knockdown of SMRT and NCoR individually could induce a slight but statistically significant increase in cell growth. Knockdown of SMRT and NCoR simultaneously led to a further increase in cell growth, suggesting that there is a functional redundancy of SMRT and NCoR.

DISCUSSION
Increasing evidence suggests that ␤-catenin signaling is able to undergo cross-talk with and integrate multiple intracellular signals to coordinate cellular functions, such as those involving forkhead box O transcription factors (59), transforming growth factor ␤ (60), RAS signaling (61), and AP-1 signalings (62). More recently, we and others have found that ␤-catenin signaling can undergo cross-talk with androgen receptor signaling (21,49,50,63,64). Androgen receptor is a member of steroid/nuclear receptor superfamily (65). It has been well documented that the mechanism of transcriptional activation by androgen receptor involves many associated proteins, including coactivators and corepressors (18). Interestingly, some of the coactivators, such as CBP and p300, also play pivotal roles in the ␤-catenin/TCF4-mediated transactivation (10,11), and the p160 family coactivator GRIP1 has been reported to enhance ␤-catenin/TCF4-mediated transactivation synergistically with ␤-catenin (12). FIGURE 9. Knockdown of SMRT or NCoR enhances the transcriptional activity of ␤-catenin/TCF4 and promotes cell proliferation. A, knockdown of endogenous SMRT and NCoR. 293T cells were transfected with 150 ng/well of empty vector pSilencer or vectors encoding specific siRNA against SMRT or NCoR. 48 h after transfection, total RNA was isolated, and RT-PCR was performed to determine the expression of NCoR and SMRT mRNA. B and C, luciferase reporter assays. Triplicate 293T cells were cotransfected with either OT-Luc (B) or CCND1-Luc (C), together with 150 ng/well of either SMRT siRNA, NCoR siRNA, or control vector pSilencer and Renilla (10 ng/well). Sixteen hours after transfection, cells were treated with or without 20 mM LiCl for 24 h, and cell lysates were prepared, and reporter activity was determined as described under "Experimental Procedures." Data represent mean Ϯ S.D. of three independent experiments, and p values are shown for comparisons versus empty vector controls. D and E, cell proliferation assays. SW480 cells were plated onto 48-well plates and transfected with indicated constructs (150 ng/will). Sixteen hours after transfection, fresh complete medium containing 10% fetal bovine serum was added, and cells were cultured for 5 days. Crystal violet staining of cell cultures (D) and corresponding measurements of OD570 (E) are shown. Data in D and E represent one of three independent experiments. RLU, relative luciferase units.
Our experiments show that SMRT and NCoR, two important corepressors for androgen receptor and other steroid/nuclear receptor-mediated transcriptional repression, play a role in modulation of ␤-catenin signaling. The ␤-catenin/TCF4mediated transcription of two different reporter constructs and of endogenous CCND1 was down-regulated by expression of exogenous SMRT or NCoR. Importantly, knockdown of endogenous SMRT or NCoR enhanced the activities of ␤-catenin/ TCF4-responsive reporters and promoted cell proliferation. These results indicated that corepressors SMRT and NCoR repressed the ␤-catenin/TCF4-mediated transactivation, and this repression occurred in the presence of native levels of endogenous corepressors. Interestingly, the attenuated ␤-catenin/TCF4-mediated transactivation and the decreased expression of CCND1 are not because of the decreased expression of ␤-catenin or TCF4. Thus the effect of SMRT and NCoR may occur at transcriptional levels by either directly inhibiting the binding of ␤-catenin, TCF4, and the TCF4 binding region of target genes or by inducing chromatin modification. The latter explanation is supported by the fact that corepressors SMRT and NCoR are essential for transcriptional repression by unliganded nuclear receptors and antagonist-bound steroid receptors by directly associating with various HDACs, thereby modulating the chromatin structure and silencing transcription (66). In addition, our results show that corepressors SMRT and NCoR directly interact with both TCF4 and ␤-catenin. The mammalian two-hybrid experiments and GST-pulldown assays indicate that the C-terminal nuclear receptor interaction domains are required for the association of corepressors with both ␤-catenin and TCF4. For ␤-catenin and TCF4, the armadillo repeat domains and high mobility group domains, respectively, are required for the interaction with SMRT and NCoR. These results demonstrate that the corepressors SMRT and NCoR repress ␤-catenin signaling by association with both ␤-catenin and with TCF4. ChIP analysis showed that endogenous corepressors SMRT and NCoR bind indirectly via ␤-cate-nin and/or TCF4 to the TCF4-binding sites of the CCND1, DKK1, AXIN2, and c-MYC promoters. Because SMRT, NCoR, ␤-catenin, and TCF4 are recruited at the same sites on the target gene promoters, we do not believe that SMRT and NCoR reduce the interaction of TCF4 with DNA by sequestering TCF4 and preventing its binding to DNA. The dynamic recruitments of SMRT/NCoR, TCF4, and ␤-catenin to the Wnt/␤-catenin target gene promoters do support our proposal that coactivators (CBP/ p300 and others) and the corepressors SMRT/NCoR could bind to the TBE of target genes simultaneously. Similar findings have been reported recently. For example, the nuclear receptor cofactor SHARP can interact with both the coactivator SRA and the SMRT-HDAC corepressor complex (67). Similar to our studies, it has been demonstrated previously that the homeobox protein heterodimer Hox-pbx, N-CoR, and CBP exist in a single complex and that cAMP-dependent protein kinase-mediated phosphorylation of Hox-pbx by cAMPdependent protein kinase permits this transcription factor to activate target gene transcription (42,43).
Because the WNT/␤-catenin signaling pathway plays a critical regulatory role in many biological processes, including embryonic development and stem cell maintenance (8), it is not surprising that deregulation of WNT/␤-catenin signaling is associated with multiple diseases, including various cancers (2,68). For example, abnormal activation of the WNT/␤-catenin signaling in colorectal cancer is caused by either loss-of-function mutations in APC or gain-of-function mutations in the ␤-catenin gene, leading to the accumulation of nuclear ␤-catenin (2,8,69,70). Recent data suggest that corepressors may play an important regulatory role in the development and/or progression of colorectal cancer. In colorectal cancers, IB kinase ␣ is aberrantly activated and recruited to chromatin concomitantly with the phosphorylation of SMRT and NCoR, leading to an increased affinity of the corepressors for the 14-3-3 adaptor proteins and to the cytoplasmic export of both SMRT and NCoR (71,72). The findings that aberrant cytoplasmic distribution of SMRT and/or NCoR is seen in colorectal cancer cells lead us to speculate that cytoplasmic redistribution of corepressors SMRT/NCoR further amplifies WNT/␤-catenin signaling by diminishing the nuclear pool of SMRT and/or NCoR.
In summary, we have shown that SMRT and NCoR inhibit ␤-catenin/TCF4-dependent gene expression, including CCND1 (Fig. 10). This is important for understanding the regulation of the Wnt/␤-catenin signaling pathway. Because corepressors play critical roles in steroid/nuclear receptor and other transcription factor-mediated transcriptional repression, the data of this study may provide an additional mechanism FIGURE 10. Model for negative regulation of Wnt signaling pathway by corepressors SMRT and NCoR. In cells with a low level of corepressors SMRT and NCoR or aberrant cytoplasmic localization (left), Wnt signals enable the accumulation of ␤-catenin in the nucleus, where it binds to TCF4, displaces the corepressor complexes containing HDACs, and enhances transcriptional activation of Wnt-responsive target genes by recruiting additional coactivators, such as the histone acetylase CBP/p300. In cells with high level of SMRT and NCoR (right), SMRT and NCoR can form complexes with both ␤-catenin and TCF4. In response to Wnt signals, TCF4 and ␤-catenin can still be recruited to the promoter region of Wnt signaling target genes. However, SMRT and NCoR are also indirectly recruited by both ␤-catenin and TCF4 to the same binding sites. Because SMRT and NCoR can associate with HDACs and form a corepressor complex with additional proteins, the transcriptional activity of Wnt signaling target genes is inhibited. responsible for the cross-talk between ␤-catenin/TCF4 and many other transcription factor-mediated signaling pathways.