HIC-5 Is a Novel Repressor of Lymphoid Enhancer Factor/T-cell Factor-driven Transcription*

Activation of Wnt/β-catenin target genes is regulated by a heterodimer of β-catenin and the high mobility group box transcription factors of the lymphoid enhancer factor (LEF)/T-cell factor (TCF) family. In vertebrates, four LEF/TCF family members have been identified. They all contain a conserved β-catenin-binding motif at the N terminus and a highly conserved high mobility group box for DNA binding. The core sequence between these motifs is less conserved and contributes to the specific properties of the individual family members. To identify interacting proteins that allocate specific functions to the individual LEF/TCF transcription factors, we performed a yeast two-hybrid screen using the less conserved core sequence as bait. We isolated the murine LIM protein HIC-5 (hydrogen peroxide-induced clone 5; also termed ARA-55 (androgen receptor activator of 55 kDa)) and cloned the highly conserved Xenopus homolog. In addition, we report that the LIM domain-containing C-terminal half of HIC-5 binds to a conserved alternatively spliced exon in LEF/TCF transcription factors. Our functional analyses revealed that HIC-5 acts as negative regulator of a subset of LEF/TCF family members, which have been characterized as activators in reporter gene analyses and in the Xenopus axis induction assay. In addition, we observed a repressive interference of LEF/TCF family members with HIC-5-mediated activation of glucocorticoid-driven transcription, which again could be allocated to specific LEF/TCF subtypes. With the characterization of HIC-5 as a binding partner of the alternatively spliced exon in LEF/TCF transcription factors, we identified a novel molecular mechanism in the dialog of steroid and canonical Wnt signaling that is LEF/TCF subtype-dependent.

The four vertebrate lymphoid enhancer factor (LEF) 2 /T-cell factor (TCF) transcription factors TCF-1, TCF-3, TCF-4, and LEF-1 are the nuclear transducers of an activated Wnt/␤-catenin pathway. They all contain a highly conserved ␤-catenin-binding domain and an even more conserved DNA-binding site, the high mobility group (HMG) box. In general, they are activated by recruiting the coactivator ␤-catenin, which is thought to replace the repressor Groucho (available at www.stanford.edu/ϳrnusse/wntwindow.html) (1,2). Thereby, LEF/ TCF target genes that are repressed in the absence of Wnt/␤-catenin signaling become activated. Apart from a complex regulatory network in the cytoplasm that controls ␤-catenin stability and binding behavior, modulator proteins in the nucleus further decide the cell competence to respond to canonical Wnt signaling. In addition, the Wnt/␤-catenin pathway is influenced by cross-talk with other signaling cascades, including transforming growth factor-␤/SMAD (3), transforming growth factor-␤/Nemo-like kinase (4), and Delta/Notch (5), and by protein modulators such as ALY (6) and PIAS (7).
Although the general view of LEF/TCF action is well understood, the functional specificity of individual LEF/TCF proteins are often ignored. Most of the LEF/TCF-binding proteins mentioned above do not discriminate between the different family members or splice variants. Furthermore, proteins such as CtBP and ALY that bind selectively to individual LEF/TCF proteins cannot completely explain their functional differences. For example, CtBP binds to PLSL(T/V) motifs in the C termini of TCF-3 and TCF-4E, resulting in repression of target genes (8,9). But even after depletion of the CtBP-binding site, XTCF-3 still neither activates target genes nor induces a secondary axis in Xenopus embryos (10). Recently, we have shown that the core region between the ␤-catenin-binding domain and the HMG box confers specific properties to individual LEF/TCF transcriptions factors (10). This region has formerly been described as the interaction domain for Groucho (11,12) and ALY (6). It contains a highly conserved alternatively spliced exon, which is termed exon IVa in TCF-1, exon VI in LEF-1, and exon VIII in TCF-4. Interestingly, a Xenopus LEF-1 RNA containing this exon has not been reported so far. Two small repressive peptide motifs adjacent to this exon are alternatively expressed in TCF-4 (10,13). With the exception of this conserved exon, the sequence between the ␤-cateninbinding domain and the HMG box is less conserved and only poorly characterized.
In addition to the modulators of LEF/TCF activity mentioned above, steroid receptors not only bind ␤-catenin (14,15), but also interact directly with TCF proteins (16 -18). Steroid receptors belong to the family of ligand-activated zinc finger transcription factors and consist of an N-terminal transactivation domain, two zinc fingers, and a hormone-binding site. Upon ligand binding, the steroid receptor changes its conformation, which reduces the binding affinity of inhibitors (e.g. hsp90). After dimerization, it enters the nucleus and activates target genes. Interestingly, the cross-talk between Wnt/␤-catenin signaling and steroid response is bidirectional: glucocorticoids inhibit the transcriptional activity of LEF/TCF proteins, whereas TCF proteins modulate estrogen receptor activity. Depending on individual TCF family members, the latter can result in enhancement (TCF-1) or suppression (TCF-4) of the estrogen-driven promoter response (16,17).
Apart from the internal transactivation domain, several adaptor proteins that bind to steroid receptors and activate expression of target genes have been identified. Among them, ARA-55 (androgen receptor activator of 55 kDa; also named HIC-5 (hydrogen peroxide-induced clone 5)) was initially described as a component of the focal adhesion complex, in which it binds to the focal adhesion kinase (19,20). HIC-5/ ARA-5 (referred to below only as HIC-5) belongs to the paxillin family of LIM proteins because it shares common protein-protein interaction motifs with paxillin: HIC-5 possesses three LD domains in the N-terminal half and four LIM domains in the C-terminal half.
Recent studies confirmed a nuclear role of HIC-5 as coactivator of steroid receptors (21,22) and peroxisome proliferator-activated receptor ␥ (PPAR␥) (23) and as co-regulator of transcription factors SMAD3 and Sp1 (24,25). The pleiotropic functions of HIC-5 imply an important role as a regulatory protein, allowing the cell to integrate the input of different signaling cascades.
In this study, we identify HIC-5 as a novel binding partner of LEF/ TCF proteins. The HIC-5 C terminus containing the LIM domains binds to a conserved exon in LEF/TCF proteins. This interaction of HIC-5 and LEF/TCF proteins is conserved in vertebrates and results in a complex that represses both LEF/TCF target gene activation and HIC-5-induced steroid receptor activation. However, this does not present a general regulatory mechanism of LEF/TCF target gene activation because only those family members that contain the conserved exon and that act as activators (Xenopus (X) TCF-4C, murine (m) LEF-1, human (h) TCF-1, and hTCF-4) are repressed by HIC-5. Other family members that activate target genes but do not contain the exon (e.g. XLEF-1) are not regulated by HIC-5. However, transcription factors that do not activate target gene promoters (TCF-3 and XTCF-4A) are not regulated by HIC-5. Instead, they repress HIC-5-induced steroid receptor activation. Thus, HIC-5 is a novel LEF/TCF binding partner that mediates the TCF subtype-specific cross-talk between Wnt/␤catenin signaling and steroid receptor activation.
Yeast Two-hybrid Screen-A mouse embryonic day 10 library cloned in the pVP16 vector was kindly provided by Jürgen Behrens. L40 yeast cells were transformed with BTM116 constructs and used for screening ϳ1 ϫ 10 5 transformants of library-transformed cells. After the interacting clones of the bait vector were cured, the clones were tested for specific interaction with XTCF-3 and XTCF-4 by mating with AMR70 previously transformed with the respective bait construct (28). Positive clones were isolated, sequenced, and analyzed.
Pull-down Assays-Bacterially expressed GST-tagged HIC-5 protein was immobilized on glutathione-Sepharose beads for 2 h at 4°C in buffer A (10 mM Tris-Cl, pH 7.8, 150 mM NaCl, 1 mM MgCl 2 ⅐6H 2 O, 0.75 mM CaCl 2 ⅐2H 2 O, 2% Nonidet P-40, and protease inhibitors) and incubated with buffer A lysate from transfected human embryonic kidney (HEK) epithelial 293 cells. After binding for 2 h at 4°C, the samples were washed three times with buffer A, boiled in SDS sample buffer for 5 min, and subjected to 10% SDS-PAGE. Proteins were transferred onto nitrocellulose, probed with anti-Myc monoclonal antibody 9E10, and revealed by the chemiluminescence reaction (ECL, Amersham Biosciences).
Immobilized GST-HIC-5 constructs and in vitro translated 35 S-labeled LEF/TCF proteins or immobilized GST-LEF/TCF constructs and in vitro translated 35 S-labeled HIC-5 proteins were incubated in buffer A, washed, eluted and separated on a 10% SDS gel. After Coomassie Blue staining, the gel was dried and subjected to PhosphorImager (Raytest) analysis to visualize the bound protein.
Injection Experiments-mRNA was synthesized in vitro using the mMESSAGE mMACHINE kit (Ambion, Inc.). 500 pg of LEF-1 or HIC-5 mRNA or 70 pg of XWNT-8 mRNA were injected into both ventral blastomeres of Xenopus four-cell stage embryos. Embryos were kept as described previously (29) and analyzed for the appearance of secondary axis, dorso-anteriorization, and target gene expression.
Transfection and Reporter Gene Assays-HeLa and HEK293 cells were routinely grown in Dulbecco's modified Eagle's medium containing 10% fetal calf serum. HEK293 cells were transfected by calcium phosphate precipitation according to Gorman (30), and HeLa cells were transfected with MATra reagent (IBA GmbH, Göttingen, Germany) according to the manufacturer's recommendations. 48 h after transfection, cells were harvested. Reporter gene assays were performed as described (26). To analyze the glucocorticoid response, transfected cells were treated with 10 nM dexamethasone.
Isolation of Full-length XHIC-5-Using degenerated primers, we amplified a 500-bp fragment of the 5Ј-region of XHic-5 from stage 18 cDNA. 3 To obtain the entire open reading frame of XHic-5, we screened a Xenopus tailbud -ZAP cDNA library with this fragment. The open reading frame of XHic-5 (GenBank TM accession number AY971603) was subcloned into the NcoI/XhoI sites of pCS2-Myc.

XHIC-5 Discriminates between XTCF-3/4 and XLEF-1-To identify
new binding partners, we screened a mouse embryonic day 10 library using the core domain (between the ␤-catenin-binding site and the HMG box) of Xenopus TCF-3 and TCF-4 as bait. We identified HIC-5 as a candidate protein that binds to both XTCF-3 and XTCF-4 (data not shown). The full-length XHic-5 cDNA (GenBank TM accession number AY971603) was isolated by screening a Xenopus tailbud -ZAP library. The XHIC-5 protein is 40% identical to its human ortholog. The similarity in the conserved three LD and four LIM domains ranges from 55% (LIM domain 2) to 100% (LD domain 2) (Fig. 1A). The most obvious differences between the Xenopus and mammalian HIC proteins are the absence of a 33-amino acid proline-rich region flanking a highly conserved part and the presence of a 55-amino acid acidic and serine-rich region between LD domains 2 and 3 in the Xenopus protein. To confirm the physical interaction found in the yeast two-hybrid screen, we carried out GST pull-down assays using bacterially expressed GST-LEF/TCF fusion proteins and in vitro translated 35 S-labeled XHIC-5. Indeed, we found that XHIC-5 bound to XTCF-3 and XTCF-4, but not to the GST control (Fig. 1B). We further confirmed the physical interaction using transfected Myc-tagged XHIC-5 and bacterially expressed GST-LEF/ TCF fusion proteins (Fig. 1C). Interestingly, compared with XTCF-3 and XTCF-4, the binding of XHIC-5 to XLEF-1 was very weak (Fig. 1, B and C).
HIC-5 Binds to a Conserved Exon-The most obvious difference between XLEF-1 and XTCF-3/4 is a conserved exon that fails in XLEF-1. This exon corresponds to exon IVa in hTCF-1, exon VI in hLEF-1, and exon VIII in hTCF-4, all of which are known to be alternatively spliced. Alternative splicing of this exon has not been reported for the Xenopus homolog of LEF-1; instead, this exon is missing in all XLef-1 cDNAs reported so far (10,31). By constructing chimeric proteins, we have recently shown that this exon promotes target gene activation (10). To confirm that this conserved exon is the HIC-5-binding domain, we transfected Myc-tagged Xenopus LEF/TCF constructs and analyzed their binding to recombinantly expressed HIC-5 in GST pull-down assays. Indeed, wild-type XLEF-1 hardly bound HIC-5, whereas a chimeric XLEF-1 construct containing the conserved exon of XTCF-3 (XLEF-1ϩExon) was precipitated with immobilized GST-HIC-5 ( Fig.  2A). Furthermore, in vitro translated HIC-5 bound to the recombinantly expressed exon of XTCF-3 and XTCF-4 (Fig. 2B). Interestingly, HIC-5 did not discriminate among XTCF-4A, -4B, and -4C, which differ in the presence of two small peptide motifs flanking the conserved exon (Fig. 2B).
The C Terminus (but Not the N Terminus) of HIC-5 Binds to LEF/TCF Proteins-We next tried to map the binding domain in HIC-5. As HIC-5 contains two putative protein-protein interaction sites, the LD and LIM domains, we fused the LD domain-and LIM domain-containing parts separately to GST. Therefore, we cut the protein into two halves, the N-terminal half containing the three LD domains and the C-terminal half containing the four LIM domains (Fig. 3). GST pull-down assays with in vitro translated LEF/TCF proteins revealed that the LEF/TCFbinding site is the LIM domain-containing C terminus, but not the LD domains (Fig. 3). Thus, the conserved exon of LEF/TCF proteins interacts with the C terminus, most likely with the LIM domains, of HIC-5. Because the binding of HIC-5 to LEF/TCF proteins was observed for the Xenopus and murine proteins and also between the corresponding binding partners of different species, we conclude that the interaction is conserved in vertebrates.
HIC-5 Suppresses LEF/TCF-induced Target Gene Activation-Next, we asked whether the physical interaction between LEF/TCF proteins
This repressive function of HIC-5 is conserved among vertebrates because the results were similar and even more pronounced when we studied mHIC-5 in combination with human or murine LEF/TCF proteins. mLEF-1, hTCF-1, and h-TCF-4 activated the TOPFlash promoter by 3.1-3.5-fold. As shown for the Xenopus LEF/TCF proteins, cotransfection of HIC-5 inhibited mammalian LEF/TCF-induced promoter activation (Fig. 4C). Although TCF-3 contains the conserved exon, it did not activate the TOPFlash promoter and was not regulated by HIC-5.
The regulation of Wnt target genes by HIC-5 is not restricted to the artificial TOPFlash promoter. We observed a similar response when we used the Xenopus fibronectin promoter (Fig. 4D). Again, cotransfection of HIC-5 prevented promoter activation by XLEF-1ϩExon (1.7-fold versus 3.2-fold) and XTCF-4C (1.6-fold versus 2.3-fold), but not by XLEF-1. Thus, the binding of the conserved exon of LEF/TCF proteins to the C-terminal half of HIC-5 is conserved among vertebrates and prevents activation of Wnt/␤-catenin target gene promoters.

HIC-5 Suppresses Secondary Axis Formation-Ectopic activation of
the Wnt/␤-catenin cascade in the ventral hemisphere of Xenopus embryos results in the induction of a secondary Spemann organizer and subsequently in the appearance of a secondary body axis. The only LEF/ TCF family member that mimics an activated Wnt/␤-catenin cascade and induces a secondary body axis upon ventral injection is LEF-1. We showed recently that the frequency of secondary axis formation is higher following mLEF-1 injection than following XLEF-1 injection and that this difference is due to the presence of the conserved exon (10). If HIC-5 is indeed a general repressor that binds to the conserved exon, it should suppress mLEF-1 (but not XLEF-1)-induced secondary axis formation in Xenopus embryos. Therefore, we co-injected 500 pg of HIC-5 mRNA together with 500 pg of mLEF-1 or XLEF mRNA into both ventral blastomeres of Xenopus four-cell stage embryos and scored the appearance of a secondary axis. As expected, co-injected HIC-5 reduced the frequency of secondary axis formation induced by mLEF-1 from 33 to 15%, but had no effect on the frequency of XLEF-1-induced secondary axis formation (Fig. 5, A and B).
After injection of 70 pg of XWNT-8 mRNA, most of the embryos (94%, n ϭ 80) showed the most severe canonical Wnt phenotype, a complete dorso-anteriorization. This phenotype is best seen by a ringshaped cement gland (Fig. 5C). After co-injection of HIC-5, only 3% (n ϭ 104) of the injected embryos showed this complete dorso-anteriorization. Instead, 58% of the embryos now showed a partial rescue as seen by the appearance of a secondary axis, and 39% revealed a complete rescue as seen by the appearance of a single axis (Fig. 5C).
LEF/TCF Proteins Repress Steroid Receptor Target Genes in a HIC-5dependent Manner-We next investigated whether the physical interaction of HIC-5 and LEF/TCF proteins also affects the activity of other HIC-5-regulated target gene promoters. Because HIC-5 is known to be a coactivator of steroid receptors (21,22,32), we analyzed whether LEF/TCF proteins influence the function of HIC-5 as coactivator of the glucocorticoid receptor by reporter gene assays in HeLa cells. HeLa cells  endogenously express HIC-5 and contain substantial amounts of HIC-5 protein in the nucleus, where it co-localizes with LEF/TCF proteins (Fig.  6A). The MMTV long terminal repeat is a model promoter for studying steroid receptor activation because it directs target gene regulation by glucocorticoid, mineralocorticoid, progesterone, and androgen receptors.

HIC-5 Is a Novel Repressor of LEF/TCF-driven Transcription
Dexamethasone activated the MMTV promoter in a dose-dependent manner (Fig. 6B). Up to 1 nM dexamethasone activated the promoter by ϳ3-fold and 10 nM dexamethasone by ϳ9-fold, and further increases to 20 nM resulted in 15-fold activation. We also observed this dose dependence at the level of target gene transcription. Expression of p21 cip1 and p27 kip1 was almost undetectable at 1 nM dexamethasone, but robustly increased at higher concentrations (Fig. 6C).
Cotransfection of HIC-5 at moderate dexamethasone concentrations (10 nM) enhanced promoter activation from ϳ9 -20-fold (Fig. 6, D and  E). Overexpression of XLEF-1 and XLEF-1ϩExon did not influence the glucocorticoid response by itself; but in the presence of the conserved exon, activation mediated by cotransfected HIC-5 was decreased (Fig.  6D). XTCF-3 and XTCF-4 reduced the dexamethasone-induced activation and, more strikingly, turned coexpressed HIC-5 into a repressor. Thus, LEF/TCF proteins that contain the conserved exon repress HIC-5-mediated glucocorticoid receptor activation. Both ventral blastomeres of Xenopus four-cell stage embryos were injected with 500 pg or mLEF-1, XLEF-1, or mHIC-5 mRNA or co-injected with 500 pg of mHIC-5 mRNA ϩ 500 pg of mLEF-1 or XLEF-1 mRNA. Embryos were cultivated until the neurula stage and analyzed for secondary axis formation. A, the phenotypes of injected embryos are shown. Secondary axes induced by mLEF-1 and XLEF-1 are best seen by the appearance of a secondary neural tube. B, secondary axis formation was quantified. n, number of injected embryos. C, embryos injected with 70 pg of XWNT-8 RNA revealed complete dorso-anteriorization, which is best seen by the appearance of a ring-shaped cement gland. Co-injection of 500 pg of HIC-5 mRNA resulted in a partial rescue and the appearance of secondary axis or in a complete rescue and the appearance of a single axis. D, expression of the Wnt target genes siamois and Xnr-3 was analyzed by reverse transcription-PCR in injected and uninjected stage 10.5 Xenopus embryos. H4, amplification of the histone H4 housekeeping gene; ϪRT, control amplification without reverse transcription. We obtained similar results when we cotransfected the human androgen receptor and studied the response to dihydrotestosterone (data not shown) or when we used mammalian LEF/TCF proteins instead of the Xenopus homologs (Fig. 6E). Again, the promoter was not regulated by LEF-1, but LEF-1 slightly repressed activation via cotransfected HIC-5. TCF-3 and TCF-4 repressed the dexamethasone-induced activation and, like their Xenopus homologs, turned coexpressed HIC-5 into a repressor.

HIC-5 Is a Novel Repressor of LEF/TCF-driven Transcription
We found that expression of the glucocorticoid receptor target genes p21 cip1 and p27 kip1 was induced by HIC-5, but not by TCF-4, and that TCF-4 suppressed the HIC-5-induced activation (Fig. 6C), consistent with the reporter gene data. Thus, our data indicate that the physical interaction of HIC-5 and LEF/TCF proteins results in a repressive complex. We identified HIC-5 as a conserved mediator that regulates the cross-talk between LEF/TCF proteins and steroid receptors in a LEF/ TCF subtype-specific manner.

DISCUSSION
In this study, we have identified HIC-5 as a novel LEF/TCF binding partner. The binding of the LIM domain-containing C terminus of HIC-5 to LEF/TCF transcription factors results in a repressive complex that prevents LEF/TCF-induced target gene activation and axis induction. Up to now, only a few LEF/TCF-interacting proteins such as SMAD4 (3) and ALY (33) that promote target gene activation have been identified. All others, including Groucho (2) and CtBP (9), act as corepressors. Thus, the identification of HIC-5 as a new corepressor of LEF/ TCF proteins further supports the idea that repression by a LEF/ TCF⅐corepressor complex is the primary regulatory principle for Wnt/ ␤-catenin target genes.
Because the interaction between LEF/TCF proteins and HIC-5 was found to be conserved in different species, we suggest that HIC-5 is a general important regulator of canonical Wnt signaling in vertebrates. We mapped a conserved and alternatively spliced exon as the binding domain in LEF/TCF proteins. Notable, unlike Groucho, HIC-5 did not completely block LEF/TCF-driven target gene activation. Instead, HIC-5 regulated only a subset of LEF/TCF proteins. Thus, we identified a molecular mechanism through which LEF/TCF subtypes acquire individual properties.
The conserved LIM domains of HIC-5 are known to bind a multitude of proteins as well as DNA (23,24,34,35). In this study, we have demonstrated for the first time that the LIM domain-containing C terminus also binds to LEF/TCF transcription factors. The binding domain in LEF/TCF proteins overlaps with the Groucho-binding domain, which has been shown to be the region between the ␤-catenin-binding site and the HMG box (12,36). Nevertheless, the binding domains for Groucho and HIC-5 are not identical because XLEF-1 binds HIC-5 only weakly (this study), but still binds Groucho (13). We identified a highly conserved alternatively spliced exon as the interaction domain in LEF/TCF proteins, and this exon has 80% identity between LEF-1 and TCF-4 and 70% identity between TCF-3 and TCF-4. In Xenopus, this exon is present in the mRNA of XTCF-3 and XTCF-4, but not XLEF-1. The interaction of HIC-5 with the alternatively spliced exon is conserved in vertebrates because we found it in different vertebrate species and even across species. Interestingly, this exon is part of context-dependent activation domain B in hLEF-1 (37) and thus distinct from the ALY interaction site, which is located predominantly in context-dependent activation domain A (33). Using chimeric proteins, we recently identified this exon as an activating element (10). Although the mode of activation via the conserved exon is still unknown, the binding of the corepressor HIC-5 might provide an explanation for its absence in XLEF-1: lack of this exon prevents inappropriate repression by HIC-5. For different LEF/TCF family members, it has been shown that the HIC-5 interaction site is alternatively spliced (38,39), indicating that the regulation of Wnt/␤-catenin signaling by HIC-5 is restricted to distinct LEF/TCF isoforms. Thus, in addition to alternatively expressed C termini, which define whether the general repressor CtBP can bind to TCF-3 and TCF-4 or not (8,9), alternative splicing in the core domain between the ␤-catenin-binding site and the HMG box modulates the activity of LEF/ TCF transcription factors by recruiting HIC-5.
Although HIC-5 has been characterized as coactivator for Sp1 and steroid receptors, we have demonstrated here that the interaction with LEF/TCF proteins results in a repressive complex. This repression was not restricted to the regulation of the artificial LEF/TCF-responsive promoter TOPFlash or the promoter of the Wnt/␤-catenin target gene fibronectin, but was also observed in vivo in injected Xenopus embryos.
The pleiotropic functions of HIC-5 on several transcriptional regulators, including steroid receptors (21,22), PPAR␥ (23), Sp1 (24), and SMAD (25), imply that it might be a threshold protein that controls the communication of different signaling cascades. It has recently been shown that, as an interaction partner of PPAR␥, HIC-5 induces the epithelial program in the intestine (23). In contrast, TCF-4 has been shown to be essential for the maintenance of a stem cell population in the intestine (40) and for maturation of Paneth cells in intestinal crypts (41), and inappropriate Wnt signaling in the colon results in cancer formation (42). Although PPAR␥ is expressed in microvilli and TCF-4 expression is restricted to crypt cells, HIC-5 expression increases from the crypt to the microvillus tip (23,41). One might speculate that the interaction of HIC-5 with PPAR␥ (23) and LEF/TCF proteins (this study) is involved in the decision of whether the intestinal cells continue to proliferate (high TCF-4 and low HIC-5 and PPAR␥) or whether they start to differentiate to epithelial cells (low TCF-4 and high HIC-5 and PPAR␥).
The cross-talk between steroid receptor signaling and canonical Wnt signaling is regulated at several levels. We have provided evidence that, in addition to the direct interaction of steroid receptors with ␤-catenin (15) and LEF/TCF proteins (16, 17) a TCF⅐HIC-5 complex represses the glucocorticoid response. Interestingly, the different LEF/TCF proteins regulated the glucocorticoid response in different ways. Although the HIC-5-mediated glucocorticoid response was not altered by LEF-1 without the exon, it was repressed by the exon-containing chimeric protein and exon-containing mLEF-1. This repression was even more pronounced by cotransfected TCF-3 or TCF-4. Thus, HIC-5 mediates a LEF/TCF subtype-specific regulation of the glucocorticoid response, with LEF-1⅐HIC-5 acting as a weak repressor and TCF-3⅐HIC-5 and TCF-4⅐HIC-5 acting as robust repressors both of glucocorticoid-responsive promoters and of expression of the endogenous glucocorticoid receptor target genes p21 cip1 and p27 kip1 . Taken together, our data reveal that HIC-5 is a new LEF/TCF subtype-specific corepressor and that the interaction of the LIM domain-containing C-terminal half of HIC-5 with an alternatively spliced exon in LEF/TCF proteins defines a new level of the cross-talk between Wnt/␤-catenin signaling and steroid receptor activation.