TIP49 Regulates β-Catenin-Mediated Neoplastic Transformation and T-Cell Factor Target Gene Induction via Effects on Chromatin Remodeling

The Wnt signaling pathway regulates many important processes, such as polarity of cell division, cell proliferation, and cell fate determination (1, 2). Deregulation of Wnt signaling has been implicated in the pathogenesis of many different human cancers (reviewed in Refs. 3 and 4). β-Catenin has a central role in the canonical Wnt signaling pathway, and tight control of β-catenin protein levels and its subcellular localization play a crucial role in regulating the signaling activity of β-catenin. In the absence of Wnt stimulation, glycogen synthase kinase 3β (GSK3β) appears to phosphorylate several serine and/or threonine residues in β-catenin’s NH₂-terminal region, leading to recognition of phosphorylated β-catenin by the β-TrCP1 ubiquitin ligase, polyubiquitination, and subsequent degradation of cytosolic β-catenin by the 26S proteasome (5, 6, 7, 8, 9, 10). Complex formation of GSK3β with β-catenin, adenomatous polyposis coli, and Axin (or its homolog, known as Conductin, Axil, or Axin2) facilitates β-catenin phosphorylation (11). Binding of a Wnt protein to its cognate Frizzled receptor complex at the cell surface leads to inhibition of the kinase activity of GSK3β, thus allowing unphosphorylated β-catenin protein to escape degradation and to accumulate and enter the nucleus (12). There, β-catenin can bind to members of the T-cell factor (TCF) family to promote target gene transcription (13, 14). In cancers of various types, β-catenin degradation is often inhibited as a result of inactivating mutations in adenomatous polyposis coli or Axin1 or gain of function mutations affecting the key phosphorylation sites in the NH₂ terminus of β-catenin (4). As a consequence, β-catenin is stabilized, and complex formation with TCF is enhanced, leading to activation of a variety of downstream genes, including c-Myc (15), Cyclin D1 (16, 17), matrilysin/MMP7 (18), and ITF-2 (19).⁶

Several proteins have been implicated in modulating the activity of the β-catenin/TCF complex and/or expression of TCF-regulated genes. For instance, the Groucho/transducin-like enhancer of split family of proteins have been shown to bind to TCFs in the absence of β-catenin, and the Groucho/transducin-like enhancer of split proteins repress TCF-dependent gene expression (20). Groucho interacts with histones (21) and the histone deacetylase Rpd3 (22), presumably forming a chromatin structure incompatible with transcription. Other mechanisms of inhibition of β-catenin/TCF transcription may include acetylation of TCF by cAMP-responsive element binding protein (CREB)-binding protein (CBP) in Drosophila (23) and binding of β-catenin by the small polypeptide inhibitor ICAT (inhibitor of β-cat and TCF) (24) or the proteins Duplin (25) or Chibby (26).

At present, the mechanisms by which the β-catenin/TCF complex activates expression of key target genes are not well understood. Previous attempts to search for critical β-catenin cofactors identified TATA box-binding protein (TBP; Ref. 27) and p300/CBP (28, 29, 30). Both proteins may function to link β-catenin to the basal transcription machinery. In addition, p300/CBP may help to loosen local chromatin structure via its histone acetyltransferase (HAT) activity, allowing access of other cofactors to target gene promoters. However, the interactions with TBP and p300/CBP do not appear to fully explain how β-catenin stimulates transcription. For example, some transactivating elements at the β-catenin COOH-terminal region can stimulate gene expression without interacting with TBP or p300 (27, 31). In addition, although TBP and p300/CBP are ubiquitously present, nuclear localization of β-catenin and its interaction with TCFs are not always sufficient for robust activation of TCF transcriptional activity, as is the case in normal T lymphocytes (32). Finally, the activity of p300 is evident only for a subset of β-catenin target gene promoters (28). β-Catenin, therefore, may use different cofactors to fulfill its function on particular target genes, and these cofactors may include components of the chromatin remodeling complexes or the basal transcription machinery. As an example, recent studies by Barker et al. (33) showed that the chromatin remodeling factor Brg-1 can interact with β-catenin to promote activation of certain TCF-target genes. Additionally, genetic approaches in Drosophila have identified the protein pygopus as a cofactor of β-catenin/TCF-mediated transcription (34, 35, 36). The presence of a PHD domain in pygopus implicates pygopus in chromatin remodeling (37, 38), but the exact mechanism underlying its function is not clear.

We have chosen to explore further the interaction of β-catenin with TIP49 (TBP-interacting protein 49; Ref. 39). The TIP49 protein [also known as TIP49a (40), Pontin52 (41), NMP238 (42), RUVBL1 (43), Rvbl (44), TAP54α (45), and TIH1p (46)] and TIP49’s close relative, TIP48 (39) [also known as TIP49b (47), Reptin52 (48), RVBL2 (43), Rvb2 (44), TAP54b (45), and TIH2p (49)], are members of a highly conserved protein family with sequence similarity to bacterial RuvB, an ATP-dependent DNA helicase that catalyzes branch migration in the Holiday junction. The TIP49 and TIP48 proteins have been suggested to possess single-strand DNA-stimulated ATPase activities and DNA helicase activities of opposite polarity (47, 50), consistent with a possible role for the proteins in chromatin metabolism and transcriptional regulation. Indeed, TIP49 and TIP48 have been found in the yeast INO80 chromatin remodeling complex, which is involved in transcription and DNA repair (44), as well as in the human TIP60 (human immunodeficiency virus-1 Tat interacting protein 60) HAT complex, which plays a role in DNA repair and apoptosis, possibly through histone modifications (45). Additionally, Jonsson et al. (51) showed that Rvb1 and Rvb2, the yeast homologs of human TIP49 and TIP48, are associated with a complex that shows ATP-dependent chromatin remodeling activity in vitro and that mutation of either protein affects transcription of over 5% of yeast genes. A more direct role of TIP49 and TIP48 in transcription is suggested by recent data on their ability to modestly affect β-catenin/TCF transcription of a model reporter gene (48), but the significance and role of TIP49 and TIP48 in β-catenin-mediated regulation of endogenous cellular genes have not been explored. Of some potential interest, in prior studies, TIP49 was found to interact with c-Myc and to affect c-Myc-mediated oncogenic transformation (39). However, no evidence was offered that TIP49 affects c-Myc-mediated transcription of reporter genes (39).

Here, we report data implicating TIP49 as an essential cofactor in β-catenin-mediated neoplastic transformation. Consistent with this observation, stable expression of an ATPase-deficient TIP49 mutant blocked expression of endogenous β-catenin/TCF target genes, and small interfering RNAs (siRNAs) directed against endogenous TIP49 had similar effects on β-catenin/TCF target gene activation. Inhibition of β-catenin/TCF target gene expression was linked to inhibition of histone acetylation of β-catenin target gene sequences, suggesting that TIP49 mediates effects on β-catenin/TCF target transcription through chromatin remodeling mechanisms.

The wild-type allele of TIP49 was cloned by PCR using random hexamer-primed cDNA from 293 cells as template. The wild-type TIP49 cDNA was then used as a template in a PCR-based approach to generate TIP49D302N. Both wild-type TIP49 and TIP49D302N PCR-generated cDNAs contain a FLAG epitope tag at the COOH terminus. Both PCR products were subcloned into either the pcDNA3 vector (Invitrogen, San Diego, CA) or the retroviral vector pPGS-CMV-CITE-neo (52). All plasmid sequences were verified by automated DNA sequencing. The pBabe-S33Yβ-catenin-puro expression vector encoding the S33Yβ-catenin mutant protein was generated by cloning the S33Yβ-catenin cDNA into the retroviral vector pBabe-puro (53). The plasmid carrying a mutant K-ras cDNA (a valine to cysteine mutation at codon 12) in the retroviral vector pBMN and the constructs with the FLAG-tagged S33Yβ-catenin cDNA cloned into the pcDNA3 vector or the pBMN vector have been described previously (52). The reporter constructs pTOPFLASH and pFOPFLASH (kindly provided by B. Vogelstein) contain either three copies of an optimal TCF-binding motif or three copies of a mutant motif, respectively (54). Plasmid pCH110 (Pharmacia, Piscataway, NJ) contains a functional LacZ gene downstream of a cytomegalovirus early-region promoter-enhancer element and was used as control for transfection efficiency in reporter assays.

Phoenix, RK3E, 293, LoVo, SW480, and HCT116 cells were grown in DMEM supplemented with 10% fetal bovine serum. β-Catenin-transformed cell lines RK3E/S33Y-B, RK3E/S33Y-C, and RK3E/ΔN47 have been described previously by Kolligs et al. (52). The polyclonal RK3E-derived cell lines RK3E/Neo, RK3E/TIP49wt, and RK3E/TIP49D302N were obtained after infection with the respective pPGS-CMV-CITE-neo vector-based retrovirus and subsequent selection of the bulk cell population in G418 at a concentration of 1 mg/ml. After 1 week, the G418 concentration was reduced to 250 μg/ml, and the expression of the FLAG-tagged TIP49 protein was confirmed by Western blot analysis. Focus formation assays in these stable RK3E/Neo, RK3E/TIP49wt, and RK3E/TIP49D302N lines were carried out as described previously (52). Polyclonal G418-resistant β-catenin-transformed RK3E, SW480, HCT116, or LoVo cell lines stably expressing wild-type TIP49, the TIP49D302N mutant, or only the neomycin resistance gene were generated in the same manner as that for RK3E-derived cells. Assays of colony formation in soft agar were performed as described previously (52). To assess the effect of TIP49D302N on histone acetylation of β-catenin target gene promoters, RK3E/Neo, RK3E/TIP49wt,and RK3E/TIP49D302N cells were infected with retroviral supernatant from amphotrophic Phoenix cells transfected with pBabe-S33Yβ-catenin-puro. After 1 week of drug selection of the transduced cells in 1.0 μg/ml puromycin (Sigma, St. Louis, MO), cells were subjected to analysis with chromatin immunoprecipitation (ChIP) assays.

Transient transfection and reporter gene assays were carried out essentially as described previously (52). In brief, RK3E cells were transfected with 0.2 μg of pcDNA3/S33Yβ-catenin, 1.2 μg of the pcDNA3 expression vectors encoding either wild-type TIP49 or TIP49D302N, 0.4 μg of pTOPFLASH or pFOPFLASH, and 0.2 μg of pCH110. The total mass of DNA for each transfection was kept constant by adding empty pcDNA3 vector. Two days after transfection, luciferase activities were measured in a luminometer and normalized to β-galactosidase expression.

Total RNA was extracted from cells using Trizol reagent (Invitrogen). Ten μg of total RNA were separated on 1.2% formaldehyde-agarose gels and transferred to Zeta-Probe GT-membranes (Bio-Rad, Hercules, CA) by capillary action. Human ITF-2 and Axin 2 and rat ITF-2, Axil, and glyceraldehyde-3-phosphate dehydrogenase (GAPDH) gene fragments were amplified by PCR and labeled with [³²P]dCTP by random priming. Northern blot hybridization to ³²P-labeled probes was carried out by standard methods. Signals were detected by exposure to BioMax-MS film (Kodak, Rochester, NY) at −80°C with an intensifying screen. Just after the autoradiograph was obtained for each gene, the membrane was placed in a phosphorimaging cassette (Molecular Dynamics, Sunnyvale, CA). The cassettes were then exposed in a PhosphorImager (Molecular Dynamics), and subsequent quantitative analysis was performed using Imagequant software (Molecular Dynamics).

Western blot analysis was performed as described previously (52). For coimmunoprecipitation, LoVo cells stably expressing wild-type TIP49, the TIP49D302N mutant, or only the neomycin resistance gene were lysed in lysis buffer [10 mm Tris (pH 7.05), 50 mm NaCl, and 1% Triton X-100] supplemented with proteinase inhibitors (complete proteinase inhibitors; Roche Applied Science, Indianapolis, IN). To determine the level of protein expression, whole-cell lysates were analyzed by Western blotting with anti-β-catenin monoclonal antibody (Transduction Laboratories, Lexington, KY) or anti-FLAG M2 antibody (Sigma) for detection of TIP49. Immunoprecipitations (IPs) were carried out in coimmunoprecipitation buffer [50 mm Tris (pH 7.5), 100 mm NaCl, 0.1% Triton X-100, 15 mm EGTA, and proteinase inhibitors] with anti-FLAG polyclonal antibody (Sigma) followed by incubation with protein A-agarose (Pierce, Rocklord, IL). Precipitates were washed four times with coimmunoprecipitation buffer and subjected to Western blot analysis with anti-β-catenin monoclonal antibody and anti-FLAG M2 antibody. For immunoprecipitation with HCT116 and SW480 lines stably expressing FLAG-tagged TIP49, anti-FLAG M2 affinity gel (Sigma) was used, and the assay was performed according to the manufacturer’s suggestions. Polyclonal antibody against BAF53 was kindly provided by Gerald Crabtree. For assessment of interaction of β-catenin with transactivation/transformation domain-associated protein (TRRAP), 293 cells were lysed in lysis buffer [20 mm Tris (pH 7.5), 140 mm NaCl, 1% Triton X-100, and 10% glycerol] supplemented with complete proteinase inhibitors and immunoprecipitated with rabbit polyclonal antibody against TRRAP (Santa Cruz Biotechnology, Santa Cruz, CA) or polyclonal anti-polyhistidine antibody (Santa Cruz Biotechnology) as control. To assess protein expression in RK3E cells, whole-cell extracts were prepared with radioimmunoprecipitation assay lysis buffer (Tris-buffered saline, 0.5% deoxcholate, 0.1% SDS, and 1% NP40) supplemented with complete proteinase inhibitors. Lysates were then analyzed by Western blotting as described previously (52) with anti-β-catenin monoclonal antibody and anti-FLAG M2 antibody. Expression of β-actin was used to control for loading of the lanes and detected with anti-actin polyclonal antibody (Sigma).

The TIP49 siRNA and the control scrambled siRNA duplexes with two thymidine overhangs were purchased from Dharmacon Research (Lafayette, CO). The target sequence for the TIP49 siRNA is 5′-AATGGCTGGAAGAGCTGTCTT-3′. For transfection, 293 cells were cotransfected with 1.68 μg of siRNA duplxes, together with 4 μg of pcDNA3/S33Yβ-catenin or 4 μg of pcDNA3/S33Yβ-catenin plus 4 μg of pcDNA3/TIP49D302N (in 60-mm culture dish) by Lipofectamine 2000 (Invitrogen). The total mass of plasmid DNA for each transfection was kept constant by adding empty pcDNA3 vector. Two days after transfection, cells were lysed and subjected to Western blot using an anti-TIP49 rabbit polyclonal antibody generated against a TIP49 peptide corresponding to amino acids 437–456 (Alpha Diagnostic, San Antonio, TX) for endogenous TIP49 expression and an anti-FLAG M2 antibody for detection of S33Yβ-catenin expression. For detection of the transfected FLAG-tagged TIP49D302N, the anti-FLAG M2 antibody was used. Expression of β-actin was used for loading control. In parallel, total RNA from the transfected cells was prepared and subjected to Northern blot analysis using human ITF-2, AXIN2, TIP49, and GAPDH cDNA probes.

The ChIP assay was performed according to the manufacturer’s recommendation (Upstate Biotechnology, Lake Placid, NY). In brief, 1 × 10⁶ to 4 × 10⁶ cells were treated with 1% formaldehyde to cross-link proteins to the DNA. The chromatin was then prepared, sonicated, and immunoprecipitated with rabbit polyclonal antibodies against acetylated histone H4 (Upstate Biotechnology), TIP49, TRRAP (Santa Cruz Biotechnology), and TIP60 (Upstate Biotechnology). The recovered DNA was analyzed by PCR using a pair of primers that flank the TCF-binding site at −565 in the rat ITF-2 promoter or the primers flanking the TCF-binding site at −1874 in the human ITF-2 promoter. As negative controls, distal sequences were amplified with primers corresponding to the region 1 kb upstream of the TCF-binding site or the downstream coding region in the rat ITF-2 gene and primers located at the downstream coding region in human ITF-2. The sequences for these primers are as follows: (a) the TCF site in rat ITF-2, 5′-GTAAGGAGGTGGTCAATACTCTAG-3′ (forward primer) and 5′-GTAATCCTGAGCTTGAAGAGTGG-3′ (reverse primer); (b) upstream for rat ITF-2, 5′-CGACTCGATCATAGGCTACCC-3′ (forward primer) and 5′-CTGAGAAAGCACAGTCAGGTTC-3′ (reverse primer); (c) downstream for rat ITF-2, 5′-GAGCCGAATTGAAGACCGTTTAG-3′ (forward primer) and 5′-ATGAGTGAATGTCTGTTGGCTGAG-3′ (reverse primer); (d) the TCF site in human ITF-2, 5′-GGAGAGGACCCACATCCCTC-3′ (forward primer) and 5′-GAGCAGGCGACCATAGAGTGG-3′ (reverse primer); and (e) downstream for human ITF-2, 5′-AAGCCGAATTGAAGATCGTTTAG-3′ (forward primer) and 5′-ATGAGTGAATGTCTGTTGGCTGAG-3′ (reverse primer). The input sample contains 0.1% of the total input chromatin as PCR template.

Perhaps in part because it may be a component of several different nuclear protein complexes, TIP49 has been implicated in diverse pathways including neoplastic transformation (39), DNA repair, and apoptosis (44, 45). Functions in chromatin metabolism and transcription have been suggested for TIP49, but its specific role in these processes is rather obscure. TIP49 is homologous to the bacterial RuvB protein, and it possesses intrinsic ATPase activity and contains the Walker A and Walker B motifs (Fig. 1,A), which are found in proteins that bind and hydrolyze ATP (55, 56). A single missense mutation in the Walker B box (DEVH→NEVH) of bacterial RuvB is sufficient to its abolish ATPase activities, creating a dominant negative allele (57). A previous study by Wood et al. (39) showed that an analogous mutation in the Walker B box of TIP49 (D302→N; Fig. 1,A) was able to interfere with c-Myc-mediated transformation. However, it was also reported in the Wood et al. (39) study that the TIP49D302N mutant did not show any effect on the ability of c-Myc to activate transcription of c-Myc-dependent reporter genes. In light of the fact that TIP49 interacts with β-catenin and has been suggested to positively modulate β-catenin/TCF transcription (41, 48), we first tested whether the TIP49D302N mutant could affect activation of the model TCF reporter construct pTOPFLASH, which contains three optimal TCF-binding sites upstream of a minimal promoter element and the luciferase gene (54). Ectopic expression of the S33Y oncogenic form of β-catenin (S33Yβ-catenin; Ref. 52) strongly activated TCF-dependent transcription in the E1A-immortalized RK3E rat epithelial cell line (Fig. 1,B). However, whereas expression of wild-type TIP49 did not enhance the ability of β-catenin to activate TCF reporter gene activity, expression of the TIP49D302N mutant had a modest ability to inhibit the effect of β-catenin on TCF-dependent reporter gene activity (Fig. 1 B).

Our previous work has shown that cancer-derived mutant forms of β-catenin can readily induce morphologically transformed foci when introduced into the RK3E cell line, an adenovirus E1A-immortalized rat epithelial line (52). Although c-Myc has been suggested to be a key downstream target of Wnt/β-catenin/TCF action in cancer cells (15), in the RK3E system, it appears that stabilized oncogenic forms of β-catenin mediate transformation through c-Myc-independent mechanisms (52). Based on the interaction of TIP49 with β-catenin, we sought to determine whether TIP49 plays a role in β-catenin-induced transformation of RK3E. Polyclonal G418-resistant RK3E lines with stable expression of either wild-type TIP49 (TIP49wt) or TIP49D302N were generated, and focus formation assays with these cells were carried out using replication-defective retroviruses encoding the S33Y oncogenic mutant form of β-catenin. Consistent with prior studies (52), S33Yβ-catenin induced 500–600 foci in the control RK3E/Neo cells in 4 weeks (Fig. 2). Although S33Yβ-catenin generated roughly equivalent numbers of foci in the RK3E/TIP49wt and control RK3E/Neo lines, the number of foci generated in the RK3E/TIP49D302N line was 70–80% less than that in the control line (Fig. 2, A and B). Expression of the TIP49wt and TIP49D302N proteins in the RK3E lines is shown in Fig. 2 C.

To assess the specificity of the effect of TIP49D302N on β-catenin-induced neoplastic transformation versus cell growth and neoplastic transformation in general, the growth and transformation susceptibility of the various RK3E lines were studied further. No clear-cut effects on in vitro proliferation rates were observed when the RK3E lines expressing wild-type or D302N forms of TIP49 were compared with one another or with RK3E/Neo control cells (data not shown). In addition, the ability of mutant K-ras to induce anchorage-independent growth in the control RK3E/Neo and RK3E/TIP49D302N lines was assessed. As shown in Fig. 3, mutant K-ras showed similar transforming activity in the RK3E/TIP49D302N and RK3E/Neo lines, indicating that the requirement for TIP49 in β-catenin-mediated transformation of RK3E cells does not simply reflect a role for TIP49 in oncogenic transformation in general.

To assess the role of TIP49 in human cancer cells with defective β-catenin regulation, we generated stable cell lines of LoVo (adenomatous polyposis coli mutant) human colon cancer cells expressing TIP49wt or TIP49D302N, as well as control G418-resistant cells. A marked reduction in colony formation in soft agar was seen in LoVo cells expressing TIP49D302N relative to the other two lines (Fig. 4). Taken together, the results in RK3E and LoVo cells imply that TIP49 plays a role in β-catenin-mediated neoplastic transformation and that its ATPase activity is likely required for its function.

The oncogenic activity of β-catenin appears to be closely related to its transcriptional activity (52). Our finding that TIP49D302N strongly inhibited β-catenin-mediated neoplastic transformation contrasted sharply with the modest inhibitory effects of TIP49D302N in the transient reporter gene assay with the pTOPFLASH model reporter gene construct. Considering the possible function of TIP49 in chromatin remodeling, we hypothesized that the effect of TIP49 on β-catenin/TCF transcription might not be readily apparent with DNA templates not organized into higher order chromatin complexes. Rather, the function of TIP49 in β-catenin/TCF transcription might be more obvious in studies of endogenous β-catenin/TCF-regulated genes. To test this idea, we performed Northern blot analyses to assess expression of two endogenous β-catenin/TCF target genes, namely, ITF-2 (19) and Axil (58, 59, 60, 61). As shown in Fig. 5,A, Northern blot analysis of the time course of ITF-2 and Axil gene expression after infection of the control RK3E/Neo cells with the retrovirus encoding the S33Yβ-catenin mutant protein or the control retroviral vector revealed that both ITF-2 and Axil were induced by β-catenin at 2 days, with much stronger induction at day 6, largely paralleling the time course of exogenous S33Yβ-catenin protein accumulation in the cells after retrovirus infection (Fig. 5,B). Although no obvious difference in induction of ITF-2 and Axil by β-catenin was observed at day 2 for RK3E cells stably expressing TIP49D302N versus the control RK3E/neo cells, the ability of the stabilized S33Yβ-catenin protein to mediate further elevation of ITF-2 and Axil expression at day 6 was substantially blunted in RK3E/TIP49D302N cells (Fig. 5, A and C). The minimal induction of ITF-2 and Axil by β-catenin seen in RK3E/TIP49D302N cells may be due to the incomplete disruption of normal TIP49 function by the TIP49D302N mutant.

To further address the role of TIP49 in β-catenin/TCF transcription, we also pursued Northern blot studies of ITF-2 and Axil expression in β-catenin-transformed lines stably expressing the TIP49D302N mutant. Consistent with our previous results (19, 58), ITF-2 and Axil were highly expressed in the three β-catenin-transformed RK3E lines compared with the parental RK3E cells (Fig. 5, D and F). Stable ectopic expression of the TIP49D302N protein in all three β-catenin-transformed RK3E lines reduced ITF-2 and Axil expression, with ITF-2 expression decreased by 30–40% and Axil expression decreased by 40–50%, when compared with the levels of expression seen in the control RK3E line (Neo) or the line stably expressing wild-type TIP49 (Fig. 5, D and F). Expression of TIP49D302N did not affect β-catenin expression (Fig. 5 E).

To strengthen the case that endogenous TIP49 plays a role in regulating the transcription of chromosomal β-catenin/TCF target genes, we used siRNA approaches to reduce endogenous TIP49 levels in human 293 cells and then used Northern blot approaches to study effects on expression of the ITF-2 and AXIN2 genes. As shown in Fig. 6,A, endogenous TIP49 levels were moderately reduced in 293 cells transfected with the TIP49-specific siRNAs (Fig. 6,A, Lanes 2 and 4), but not the control scrambled siRNAs (Lanes 1 and 3). As expected, in the cotransfection experiments, the siRNAs had no effect on ectopic S33Yβ-catenin expression (Fig. 6,A, Lanes 3 and 4). In cells transduced with the TIP49-specific siRNAs, modest inhibition of basal expression of ITF-2 and AXIN2 was seen, and more significant inhibition of S33Yβ-catenin-induced ITF-2 and AXIN2 expression was observed (Fig. 6, B and C). In fact, the inhibitory effects on β-catenin/TCF target gene expression seen with the TIP49 siRNAs were a bit stronger than those seen with the TIP49D302N mutant (Fig. 6, B and C). Based on these results and the finding that TIP49 is required for β-catenin-mediated transformation, we propose that TIP49 is a critical cofactor for neoplastic transformation by β-catenin via its role in regulating β-catenin/TCF transcription of endogenous chromosomal target genes.

As noted above, TIP49 has previously been found to be complexed in the nucleus with β-catenin and several other proteins, including c-Myc, TRRAP, a c-Myc-interacting protein that is part of several distinct HAT complexes, and BAF53, another c-Myc interacting cofactor that may have independent associations with chromatin remodeling factors (39, 62). Other work has indicated that TIP49 may be present in chromatin modifying and remodeling complexes, such as the yeast INO80 remodeling complex (44) and the human TIP60 HAT complex (45). To address the means by which TIP49 exerts its effects on β-catenin/TCF target gene expression, we sought to show that both wild-type TIP49 and the TIP49D302N proteins did, in fact, interact with β-catenin in cells. As shown in Fig. 7,A, IP of the FLAG-tagged TIP49 and TIP49D302N proteins from LoVo cells stably expressing the FLAG-tagged proteins yielded similar amounts of immunoprecipitated β-catenin protein. To determine whether BAF53 was complexed with TIP49 and β-catenin in cells, we generated HCT116 and SW480 colon cancer cell lines expressing a FLAG-epitope-tagged form of wild-type TIP49 as well as control drug-resistant lines. After IP of the lysates with anti-FLAG antibodies, we analyzed the immunoprecipitates for the presence of β-catenin and BAF53 by immunoblotting. As shown in Fig. 7,B, endogenous β-catenin and BAF53 proteins were specifically present in the anti-FLAG immunoprecipitates from HCT116 and SW480 cells expressing the FLAG-tagged TIP49 proteins. Efforts to establish using combined IP and immunoblot approaches that endogenous TIP49 associated with endogenous β-catenin and BAF53 proteins were hindered by the fact that the TIP49 and BAF53 proteins essentially comigrate with the strongly reactive immunoglobulin heavy chain band in combined IP-immunoblot experiments (data not shown). Nevertheless, we were able to show that endogenous TRRAP protein is complexed with β-catenin in 293 cells in which endogenous β-catenin was transiently stabilized by treatment with the GSK3β inhibitor lithium chloride (Fig. 7 C).

Based on the evidence that TIP49 binds to β-catenin in protein complexes that likely include the TRRAP and BAF53 proteins and our prior results implicating TIP49 in the regulation of expression of β-catenin/TCF target genes, we sought to determine whether we could find evidence that TIP49 and associated proteins were specifically tethered to TCF sites in regulatory regions of β-catenin/TCF-regulated genes. We carried out ChIP studies assessing the ability of antibodies against TIP49, TRRAP, TCF-4, and the TIP60 HAT protein to immunoprecipitate sequences containing a previously implicated TCF site in the human ITF-2 promoter region (Fig. 7,D). As shown in Fig. 7,E, in ChIP studies of formaldehyde cross-linked chromatin preparations isolated from 293 cells in which endogenous β-catenin was stabilized by treatment with the GSK3β inhibitor lithium chloride, the control IgG antibody yielded no specific PCR product for the region containing the TCF site. In contrast, DNA was specifically recovered in the ChIPs with antibodies against TIP49 as well as with antibodies against TCF-4, TIP60, and TRRAP. Further evidence of the specificity of the ChIP of TIP49 and the TIP49-associated proteins with the TCF site in the ITF-2 promoter region was obtained by showing that DNA sequences located at the downstream coding region in the ITF-2 gene were not specifically recovered in ChIPs (Fig. 7 E).

The findings above suggest that TIP49 and associated proteins, such as TRRAP and the TIP60 HAT factor, may facilitate β-catenin/TCF-mediated transcription via effects on chromatin structure and activity. As such, we sought to determine, using ChIP assays, whether the TIP49D302N mutant protein could affect chromatin modification when recruited by β-catenin to TCF target gene promoters, such as the TCF site in the proximal promoter region of the rat ITF-2 gene. Chromatin was prepared from RK3E/Neo, RK3E/TIP49, and RK3E/TIP49D302N cells, all of which had been stably transduced to express S33Yβ-catenin protein ectopically. Evidence was obtained that TIP49 specifically interacted in ChIP assays with the TCF site in the proximal promoter region of the rat ITF-2 gene, but not with nonspecific upstream or downstream sequences at the ITF-2 locus (Fig. 8,B). In ChIPs using antibodies against acetylated histone H4, we found that histone acetylation of the ITF-2 promoter region covering the TCF-binding site at −565 was greatly reduced in RK3E/TIP49D302N cells, when compared with the acetylation state of the ITF-2 promoter region in control RK3E cells and cells stably expressing wild-type TIP49 (Fig. 8,C). Expression of the TIP49D302N protein also seemed to moderately reduce histone acetylation in other regions of the ITF-2 gene (Fig. 8 C, upstream and downstream). The results suggest that TIP49, when recruited by β-catenin to promoters of TCF-target genes, affects histone acetylation of nucleosomes nearby to where it is recruited. Notably, histone acetylation is a modification that is generally correlated with increased transcriptional activity (63, 64, 65). TIP49 may exert effects on histone acetylation through its ATP-dependent chromatin remodeling activity and/or, perhaps more likely, by its ability to recruit other cofactors that directly alter the acetylation state of histones.

The findings presented here implicate TIP49 as an important cofactor in activation of β-catenin/TCF-regulated genes and β-catenin-mediated neoplastic transformation. We found that expression of a presumptive dominant negative allele of TIP49 (TIP49D302N) inhibited β-catenin-induced neoplastic transformation of RK3E cells and suppressed the anchorage-independent growth properties of a human colon carcinoma line with defective β-catenin regulation. Additionally, the TIP49D302N mutant protein inhibited activation of endogenous β-catenin/TCF target genes. Using siRNA-based methods to inhibit endogenous TIP49 function, we obtained further support for the role of TIP49 in regulating β-catenin/TCF target gene expression. Using IP-Western blot approaches, TIP49 was found in complexes with chromatin remodeling and histone modifying factors and cofactors, including the TIP60 histone acetylase-associated proteins TRRAP and BAF53. In ChIP assays, the TIP49, TIP60, and TRRAP proteins were found to interact with sequences in the regulatory region of the gene for ITF-2, a TCF-dependent cellular gene. Moreover, the ability of TIP49D302N to inhibit ITF-2 gene expression was linked to decreased acetylation of histones in the vicinity of the TCF-binding sites in the ITF-2 promoter region. Based on these data, we suggest that TIP49 regulates β-catenin-mediated TCF-target gene expression via effects on chromatin structure and activity.

Although TIP49 was initially identified in large part because of its binding to various nuclear factors, the functional significance of the interaction between TIP49 and many of the presumptive TIP49 partner proteins has not been well established. For instance, the results of Wood et al. (39) suggested that TIP49 played a crucial role in c-Myc-mediated oncogenic transformation. However, in contrast to the data on the function of TIP49 in activation of β-catenin/TCF-regulated target genes, no direct role for TIP49 in transcriptional activation of c-Myc reporter genes or endogenous c-Myc-regulated target genes has yet been provided in the literature. At this point, it is not clear whether the inability to demonstrate a role for TIP49 in the regulation of c-Myc-dependent gene expression may be largely due to the transient reporter gene assays performed in the initial studies (39). As we have found, the effects of TIP49 on β-catenin/TCF transcription are very modest in transient reporter assays, where model reporter gene constructs are not likely to be present in the higher order chromatin structures characteristic of endogenous cellular genes. Nevertheless, in light of the prior observations on the role of TIP49 in c-Myc-mediated neoplastic transformation and the fact that the c-MYC gene has been identified as a downstream target of the Wnt/β-catenin/TCF pathway in human colorectal cancer (15), it could be hypothesized that TIP49 functions in β-catenin-mediated transformation by regulating c-Myc expression and/or activity. We have largely excluded this possibility. First, the TIP49D302N mutant protein had no effect on c-Myc expression in RK3E cells (data not shown). Second, our previous studies have indicated that β-catenin-induced neoplastic transformation of RK3E cells is independent of c-Myc expression and activity (52). Moreover, in contrast to the rather unresolved situation regarding the effects of TIP49 on the function of c-MYC in gene regulation, the ability of β-catenin/TCF to activate transcription of endogenous target genes, such as Axil/AXIN2 and ITF-2, appears to be critically dependent on TIP49 function.

In our work, we have used a TIP49D302N mutant protein to assess the functional role of TIP49 in β-catenin-mediated transformation and transcriptional activation. As described above, TIP49D302N contains a missense mutation in the Walker B motif (DEVH→NEVH), which is expected to abolish the ATPase activity of TIP49, as has been shown for the bacterial RuvB protein (57). The mutant TIP49 protein can still form a complex with β-catenin and TIP48 (39), but it cannot bind and hydrolyze ATP. Thus, TIP49D302N likely functions in a dominant negative manner, and this same mutant form of TIP49 is the one used to demonstrate a role for TIP49 in c-Myc-mediated neoplastic transformation (39). The involvement of ATPase activity in the function of RuvB-like proteins is also supported by recent study showing that the liebeskummer mutation in the zebrafish homologue of Reptin (also known as TIP48) increases Reptin/TIP48 ATPase activity and enhances its repressive effect on β-catenin-mediated transcription. The mutant allele has been shown to promote cardiac hyperplasia during zebrafish development (66), and, in the same study, it was shown that reduction in TIP49 levels during zebrafish development essentially phenocopies the effects of the liebeskummer mutation in Reptin. More definitive insights into why the TIP48/reptin protein appears to antagonize β-catenin/TCF transcription and the TIP49 protein appears to stimulate β-catenin/TCF transcription await the results of additional studies.

Interestingly, the TIP49D302N mutant in our study had only very modest inhibitory effects on β-catenin-induced activation of a transiently transfected TCF-responsive reporter gene. Perhaps consistent with this result, as noted above, TIP49D302N also showed no effect on c-Myc-mediated transcription of several different c-Myc response reporter constructs (39). The difference we observed in the magnitude of the effect of TIP49D302N on transcription of a transiently introduced reporter template versus chromatinized templates might be attributable to the requirement of TIP49 DNA remodeling activity for transcription. As an ATP-dependent DNA helicase, TIP49 has been identified in several chromatin remodeling complexes. Transiently introduced reporter gene templates may not be assembled into a phased array of nucleosomes. As such, significant alterations in chromatin structure may not be required to allow access of trans-acting factors to the promoter region of the reporter gene template. It should be noted that significant differences in the transcriptional activity of episomal versus chromosomal templates have been observed previously in the case of glucocorticoid receptor-mediated transcription. In particular, in contrast to stable chromatin templates, a transiently introduced mouse mammary tumor virus reporter is activated by glucocorticoid receptor in a manner that is independent of the hBrg-1/BAF chromatin remodeling complex (67).

In light of these results with glucocorticoid receptor-mediated transcription and the results presented here, we would propose that in the absence of Wnt signaling or other β-catenin stabilizing factors, TCF target genes are in a “closed” chromatin structure that is imposed by Groucho corepressor proteins and their associated histone deacetylases. To activate TCF-regulated target genes, β-catenin may act to relax chromatin structure by recruiting chromatin modifying and remodeling cofactors. In support of this notion, previous studies have shown that β-catenin interacts with the chromatin remodeling factor Brg-1 (33) and HAT p300/CBP (28, 29, 30), and β-catenin appears to require their activities for activation of certain β-catenin/TCF target genes. Indeed, recent studies with the Cyclin D2 gene, a presumptive Wnt/β-cateninTCF pathway downstream gene in pituitary development, showed that specific coactivators, including p300/CBP, TIP60, and TRRAP, were recruited to the Cyclin D2 promoter in a temporally specific manner (68). Also of interest with regard to mechanisms β-catenin may use to activate gene expression are findings on the recently discovered protein called pygopus. Pygopus is a PHD domain-containing protein that has been shown to complex with β- catenin and regulate activation of β-catenin/TCF target genes (34, 35, 36). Because many PHD domain-containing proteins also have other domains whose function is linked to chromatin (37, 38), the presence of a PHD domain in pygopus might predict a role for pygopus in chromatin-related functions. Undoubtedly, future studies will illuminate the complex interplay of interactions between β-catenin, TCFs, TIP49, and the various other chromatin remodeling proteins and associated cofactors that have been implicated in regulation of β-catenin/TCF target gene transcription. Although perhaps an obvious point, the studies and their likely results have major implications for our understanding of the role of Wnt signaling in normal and cancer cells.

We thank Drs. Bert Vogelstein, Gerald Crabtree, and Gary Nabel for generously providing reagents and Ira Winer and David Van Mater for providing comments on the manuscript.