Wnt ligands have pleiotropic and context-specific roles in embryogenesis and adult tissues. Among other effects, certain Wnts stabilize the β-catenin protein, leading to the ability of β-catenin to activate T-cell factor (TCF)-mediated transcription. Mutations resulting in constitutive β-catenin stabilization underlie development of several human cancers. Genetic studies in Drosophila highlighted the split ends (spen) gene as a positive regulator of Wnt-dependent signaling. We have assessed the role of SHARP, a human homologue of spen, in Wnt/β-catenin/TCF function in mammalian cells. We found that SHARP gene and protein expression is elevated in human colon and ovarian endometrioid adenocarcinomas and mouse colon adenomas and carcinomas carrying gene defects leading to β-catenin dysregulation. When ectopically expressed, the silencing mediator for retinoid and thyroid receptors/histone deacetylase 1-associated repressor protein (SHARP) protein potently enhanced β-catenin/TCF transcription of a model reporter gene and cellular target genes. Inhibition of endogenous SHARP function via RNA inhibitory (RNAi) approaches antagonized β-catenin/TCF-mediated activation of target genes. The effect of SHARP on β-catenin/TCF-regulated genes was mediated via a functional interaction between SHARP and TCF. β-Catenin–dependent neoplastic transformation of RK3E cells was enhanced by ectopic expression of SHARP, and RNAi-mediated inhibition of endogenous SHARP in colon cancer cells inhibited their transformed growth. In toto, our findings implicate SHARP as an important positive regulator of Wnt signaling in cancers with β-catenin dysregulation. [Cancer Res 2007;67(2):482–91]
The Wnt signaling pathway has a vital role in diverse cellular processes, such as cell proliferation, cell fate determination, and cell survival (1, 2). β-Catenin is an essential factor in the so-called “canonical” Wnt pathway, with its levels tightly controlled by activating Wnt ligands. In the absence of Wnts, β-catenin is phosphorylated at defined serine and threonine residues in its NH2 terminus by casein kinase I and glycogen synthase kinase 3β (GSK3β). The ability of GSK3β to phosphorylate β-catenin depends on interaction with the adenomatous polyposis coli (APC) and Axin1/Axin2 proteins (3–5). The NH2-terminally phosphorylated β-catenin is recognized by an ubiquitin ligase complex containing a cellular β-transducin repeat-containing protein (e.g., βTrCP1). Following polyubiquitination, β-catenin is degraded in the 26S proteasome (3, 4, 6, 7). Binding of an activating Wnt ligand to its cognate Frizzled-low density lipoprotein-related protein (LRP)-5/LRP-6 coreceptor complex inhibits the GSK3β and Axin proteins by mechanisms not yet fully clarified. The net consequence is stabilization of the “free pool” of β-catenin (8). In the nucleus, β-catenin can bind to members of the T-cell factor (TCF) family of transcription factors, leading to enhanced transcription of certain target genes with TCF-binding sites in their regulatory regions (9). Mutational defects interfering with β-catenin degradation contribute to the development of human cancers (10, 11). Such mutations include inactivation of the APC or Axin1 tumor suppressor proteins or oncogenic mutations in the NH2-terminal phosphorylation and degradation motif of β-catenin (10, 11). A chief result of these defects is constitutively activated β-catenin/TCF-mediated transcription of key target genes.
Many TCF/β-catenin–regulated genes have been uncovered, including c-MYC (12), cyclin D1 (13), matrilysin/MMP7 (14), Survivin (15), dickkopf-1 (DKK-1; refs. 16, 17), AXIN2 (18–20), and ITF-2 (21). Although knowledge of the identities of TCF/β-catenin–regulated genes has grown much, more limited progress has been made in clarifying the means by which the β-catenin/TCF complex and other proteins cooperate to modulate target gene transcription. The proteins BCL9/Legless and Pygopus play key roles in a β-catenin/TCF-mediated transcription possibly by recruiting chromatin remodeling factors to TCF-binding sites (22, 23) and/or by promoting nuclear import of β-catenin (24). The histone acetyltransferase cyclic AMP-responsive element binding protein–binding protein (CBP) and the related protein p300 (25, 26), as well as the potential ATP-dependent DNA helicase TIP49 (27), may alter local chromatin structure by modifying histone acetylation status in the vicinity of TCF-binding sites, perhaps increasing access of cofactors to target gene promoters. The SWI/SNF component BRG1 also apparently interacts with β-catenin to promote transcriptional activation (28). When β-catenin is absent from the nucleus, the Groucho/transducin-like enhancer of split proteins associate with TCFs and inhibit TCF target gene transcription, likely in part by recruitment of histone deacetylases (HDACs; refs. 29, 30). β-Catenin/TCF transcriptional activity can also be inhibited by disruption of the TCF-β-catenin interaction, such as by the proteins inhibitor of β-catenin and TCF (31) or Chibby (32). Interestingly, in Drosophila, CBP may inhibit β-catenin/TCF transcription of some target genes by acetylation of TCF (33).
A genetic study from Lin et al. (34) showed the Drosophila split ends (spen) gene was required for Wnt-dependent signaling in the eye, wing, and leg imaginal discs but not in the embryo. The Drosophila spen gene has also been found to have roles in neuronal cell fate specification, cell survival, axonal guidance (35, 36), and cell cycle regulation (37), perhaps via the ability of the 5,554-amino acid spen protein to regulate the Notch and/or epidermal growth factor receptor/Ras signaling pathways (35, 36). The human homologue of spen encodes a 3,664-amino acid protein known as silencing mediator for retinoid and thyroid receptors (SMRT)/HDAC1-associated repressor protein (SHARP). The mouse homologue encodes a 3,576-amino acid protein known as Msx2-interacting nuclear target protein (MINT; ref. 38). The spen, SHARP, and MINT proteins are characterized by NH2-terminal RNA recognition motifs (RRM) and a Spen paralogue and orthologue COOH-terminal (SPOC) domain, which may mediate interaction with the transcriptional corepressor SMRT (39).
Given the intriguing role of spen in Wnt-dependent signaling in Drosophila, we embarked on efforts to explore SHARP function in mammalian cells. We report on studies showing that SHARP is highly expressed in cancer cells with dysregulated β-catenin and that the SHARP protein functions to potently enhance β-catenin/TCF transcriptional activity. The findings implicate SHARP as an important factor to enhance the constitutively activated Wnt/β-catenin/TCF pathway in colon and other cancer cells.
Materials and Methods
Plasmids. Sequences for the full-length open reading frame of SHARP were obtained by PCR using random hexamer-primed cDNA from 293 cells as template. The sequences were assembled into a full-length SHARP cDNA in the pcDNA3 vector (Invitrogen, San Diego, CA). The construct comprising amino acids 1 to 567 of SHARP was obtained by deleting the coding region downstream of a 5′ PvuII site in the SHARP cDNA sequences. Vectors encoding COOH-terminal fragments of SHARP (SHARP amino acids 567–3,664) were generated by PCR-based approaches. All SHARP expression constructs encode for two Flag epitope tags at the NH2 terminus, with the Flag epitope tags inserted between the KpnI and BamHI sites of the pcDNA3 vector. The GAL4-S33Yβ-catenin plasmid was generated by inserting the S33Yβ-catenin open reading frame downstream of the GAL4 DNA-binding domain (amino acids 1–147), which has been first subcloned into pcDNA3. The full-length LacZ gene was PCR amplified and cloned into pcDNA3 vector to generate the LacZ/pcDNA3 plasmid. To generate the glutathione S-transferase (GST)-E-cadherin fusion construct, the COOH-terminal cytoplasmic domain of human E-cadherin was amplified by PCR using the primers 5′-CGGGATCCAGAGCGGTGGTCAAAGAGCCC-3′ (sense) and 5′-CGGAATTCGCCTCTCTCGAGTCCCCTA-3′ (antisense) from a full-length E-cadherin cDNA plasmid. The resulting product was inserted into pGEX-2T vector (Amersham Biosciences, Piscataway, NJ) in-frame downstream of the GST sequences using the BamHI and EcoRI sites. The Flag-tagged polypeptide derived from the GSK3β-interacting domain (GID) of human Axin1 was amplified by PCR and subcloned into pcDNA3 to generate the GID/pcDNA3 plasmid. All plasmid sequences were verified by automated DNA sequencing. The plasmids LEF-1ΔN-β-cat and LEF-1ΔN-VP16 were kindly provided by Dr. Andreas Hecht (University of Freiburg, Freiburg, Germany; ref. 40). Constructs with the Flag-tagged S33Yβ-catenin cDNA cloned into the pcDNA3 vector or the pBMN vector have been described previously (41). The reporter constructs pTOPFLASH and pFOPFLASH (kindly provided by Dr. Bert Vogelstein, Johns Hopkins University, Baltimore, MD) contain either three copies of an optimal TCF-binding motif or three copies of a mutant motif, respectively. The GAL4-responsive reporter construct G5E1BLUC (kindly provided by Dr. Richard Baer, Columbia University, New York, NY) contains luciferase driven by a GAL4-responsive promoter (42).
Cell lines and tissue culture. Amphotropic Phoenix, RK3E, 293, and HCT116 cells were grown in DMEM supplemented with 10% fetal bovine serum (FBS). LS174T cells were grown in MEM supplemented with 10% FBS. The polyclonal 293 lines stably expressing either S33Yβ-catenin or only the neomycin resistance gene (Neo) were obtained following infection with the pPGS-CMV-CITE-neo vector-based retrovirus encoding S33Yβ-catenin (S33Y) or the control retroviral vector (neo), respectively. The bulk cell population was then selected in G418, and the expression of S33Yβ-catenin was confirmed by Western blot analysis. The monoclonal RK3E lines stably expressing full-length SHARP were generated by transfection of the cells with the SHARP/pcDNA3 plasmid and subsequent selection of the transduced cells in 1 mg/mL G418. Single G418-resistant clones were then tested for expression of SHARP by Western blot analysis by using a monoclonal anti-Flag M2 antibody (Stratagene, La Jolla, CA). The control G418-resistant lines were prepared in parallel by transfection of RK3E cells with pcDNA3 empty vector. Four SHARP-expressing clones and three control clones were subjected to focus formation assays as described previously (41). To obtain polyclonal HCT116 and LS174T cell lines stably expressing SHARP-specific short hairpin RNAs (shRNA) or the control scrambled shRNA, cells were transduced with retroviruses containing a H1 promoter cassette driving the expression of the shRNA (pSUPERIOR-RETRO-PURO, OligoEngine, Inc., Seattle, WA). The following oligonucleotides were annealed and ligated into the BglII and HindIII sites according to the supplier: scrambled (Sc), 5′-GATCCCTTCTCCGAACGTGTCACGTTTCAAGAGAACGTGACACGTTCGGAGAATTTTTA-3′ (sense) and 5′-AGCTTAAAAATTCTCCGAACGTGTCACGTTCTCTTGAAACGTGACACGTTCGGAGAAGG-3′ (reverse); first SHARP shRNA (Sh1), 5′-GATCCCGCATGGATAGGTCCAGAAATTCAAGAGATTTCTGGACCTATCCATGCTTTTTA-3′ (sense) and 5′-AGCTTAAAAAGCATGGATAGGTCCAGAAATCTCTTGAATTTCTGGACCTATCCATGCGG-3′ (reverse); and second SHARP shRNA (Sh2), 5′-GATCCCGGTGGTGTTTGACTGCTTATTCAAGAGATAAGCGGTCAAACACCACCTTTTTA-3′ (sense) and 5′-AGCTTAAAAAGGTGGTGTTTGACCGCTTATCTCTTGAATAAGCAGTCAAACACCACCGG-3′ (reverse). Amphotropic retroviruses were produced using the Phoenix cell system (kindly provided by Garry Nolan, Stanford University, Stanford, CA) and used to infect HCT116 and LS174T colorectal carcinoma cells. A polyclonal population was selected for 6 days in 1 μg/mL puromycin (Sigma, St. Louis, MO).
Mouse colon adenoma and carcinoma studies. Colon adenomas and carcinomas from transgenic mice with conditional inactivation of the murine Apc gene in colon epithelium were studied.6
Colony formation in soft agar. Soft agar assays on HCT116 and LS174T cell lines were essentially done as described previously (41). Briefly, 5,000 cells were plated in DMEM with 17% FBS containing 0.3% soft agar on top of an underlayer containing 0.6% soft agar in 12-well plates. Plates were fixed with glutaraldehyde and stained with methylene blue (Sigma) after 3 weeks. Colony numbers were determined using the NIH ImageJ software package (44).
In situ hybridization and immunohistochemical analysis. A cDNA fragment spanning SHARP nucleotides 10,633 to 11,066 (between EcoRI and PstI sites) was subcloned into the pPST18 vector (Roche Applied Science, Indianapolis, IN). Digoxigenin-labeled riboprobes were prepared with T7 and SP6 RNA polymerase using the DIG RNA labeling kit (Roche Applied Science). In situ hybridization (ISH) assays on 14 human colon tumor samples and 10 primary ovarian endometrioid-type adenocarcinomas (OEA) were done as previously described by Zhai et al. (45). Human colon specimens were obtained from the University of Michigan Hospital (Ann Arbor, MI), and ovarian specimens were obtained from the Cooperative Human Tissue Network/Gynecologic Oncology Group Tissue Bank. Analysis of tissues from human subjects was approved by the Institutional Review Board of the University of Michigan. SHARP expression was scored as strong (++), moderate (+), or weak/absent (+/−) based on signal intensity in the epithelial cells. Fisher's exact test was used to determine statistical significance between groups with strong or moderate versus weak/absent expression of SHARP transcripts. Immunohistochemical analysis was carried out as described previously (46) with a mouse monoclonal anti-β-catenin antibody (C19220; Transduction Laboratories, Lexington, KY) at a dilution of 1:500 on human colon tumors and OEAs and at 1:200 on mouse tissues. For detection of mouse monoclonal anti-β-catenin antibody on mouse tissue, the Vector M.O.M. Immunodetection kit (Vector Laboratories, Inc., Burlingame, CA) was used as instructed by the manufacturer. An anti-SHARP rabbit polyclonal antibody was generated against an NH2-terminal fragment of SHARP (amino acids 2–202; Cocalico Biologicals, Reamstown, PA). The rabbit polyclonal anti-SHARP antibody was used for immunohistochemistry at a dilution of 1:400.
Luciferase reporter gene assays. Transient transfection and reporter gene assays were carried out essentially as described previously (41). In brief, 293 cells were transfected with 0.1 μg of pcDNA3/S33Yβ-catenin or pcDNA3/GID, 1.3 μg of the pcDNA3 expression vectors encoding wild-type (WT) SHARP or the empty vectors, 0.4 μg of pTOPFLASH or pFOPFLASH, and 0.2 μg of LacZ/pcDNA3. 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. For reporter gene assays with SHARP-specific short interfering RNAs (siRNA), 293 cells stably expressing the S33Yβ-catenin mutant were transfected with 0.5 μg of the reporter vectors and 0.5 μg of LacZ/pcDNA3 along with 4 μg of siRNA duplexes by using Lipofectamine 2000 (Invitrogen). Luciferase activities were measured 48 h after transfection. The SHARP siRNA and the control scrambled siRNA duplexes with two thymidine overhangs were purchased from Qiagen, Inc. (Valencia, CA). The target sequences for the SHARP siRNAs are the following: first SHARP siRNA (Si1), 5′-GCATGGATAGGTCCAGAAA-3′; second SHARP siRNA (Si2), 5′-GGTGGTGTTTGACCGCTTA-3′.
Northern blot analysis. Total RNA was extracted from cells using Trizol reagent (Invitrogen). Total RNA (10 μg) was separated on 1.2% formaldehyde-agarose gels and transferred to Zeta-Probe GT-membranes (Bio-Rad, Hercules, CA) by capillary action. Human SHARP, AXIN2, DKK-1, and glyceraldehyde-3-phosphate dehydrogenase (GAPDH) gene fragments were amplified by PCR and labeled with [32P]dCTP by random hexamer 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 phosphoimaging cassette (Molecular Dynamics, Sunnyvale, CA). The cassettes were then exposed in a phosphoimager (Molecular Dynamics), and subsequent quantitative analysis was done using ImageQuant software (Molecular Dynamics).
GST pull-down assay and Western blot analysis. To determine the free pool of β-catenin, cells were lysed in lysis buffer [20 mmol/L Tris (pH 7.5), 140 mmol/L NaCl, 1% Triton X-100, 10% glycerol] supplemented with proteinase inhibitors (Complete proteinase inhibitors, Roche Applied Science). The cell extracts were first precleared by rotation with unbound glutathione-Sepharose 4B beads (Amersham Biosciences) for 45 min at 4°C and then incubated at 4°C for 1 h with glutathione beads to which the GST-E-cadherin fusion protein has been bound. The beads were washed four times with the lysis buffer and then solubilized in loading buffer for SDS-PAGE and Western blot analysis. Either the monoclonal anti-Flag M2 antibody or an anti-β-catenin antibody was used to detect the transfected or endogenous β-catenin, respectively. For detection of endogenous SHARP protein in Western blot analysis, an anti-SHARP rabbit polyclonal antibody against the NH2-terminal 200 amino acids of SHARP was used. A monoclonal anti-β-galactosidase antibody (Promega Corp., Madison, WI) was used to detect the transfected β-galactosidase. All Western blot analyses were done as described previously (41).
Increased SHARP expression in cancers with β-catenin defects. As noted above, Lin et al. (34) found that Drosophila spen was required for Wnt-dependent signaling in eye, wing, and leg imaginal discs but not in the embryo. The authors' results also suggested that the nuclear spen protein functioned downstream of the stabilization of Armadillo, the Drosophila homologue of β-catenin. Given the importance of dysregulated β-catenin signaling in multiple different cancers, we set out to study whether SHARP could potentially affect β-catenin signal transduction in primary human cancers. We did ISH to detect SHARP transcripts in 14 human primary colorectal carcinomas and adjacent normal tissues. Consistent with prior studies showing that the majority of primary colorectal carcinomas carry genetic lesions leading to the stabilization of β-catenin (1, 47), 11 of the 14 colorectal carcinomas showed strong β-catenin cytoplasmic and nuclear staining, whereas 3 of the 14 colorectal carcinomas and adjacent normal mucosa in all 14 cases showed predominantly membrane β-catenin staining (Fig. 1; data not shown). Of the 11 colorectal carcinomas with strong cytoplasmic and nuclear β-catenin staining, 10 showed strong SHARP expression in the neoplastic cells compared with low to absent SHARP expression in adjacent normal colon epithelium (representative examples in Fig. 1). In normal tissues, lymphocytes in the stroma adjacent to the epithelium often showed SHARP expression (Fig. 1). In the three colorectal carcinomas with no apparent defects in β-catenin regulation, two showed no difference in SHARP expression between cancer and adjacent normal mucosa, whereas one showed strong SHARP expression in cancer cells. To further explore the relationship between β-catenin regulatory defects and SHARP expression in human cancer, we also assessed SHARP expression in primary OEAs because, although OEAs share comparable histologic features, only ∼30% of the lesions have mutational defects affecting β-catenin regulation (most commonly activating mutations in β-catenin; refs. 46, 48). Using ISH to detect SHARP transcripts, there was moderate to strong expression of SHARP in all five OEAs with strong β-catenin nuclear and cytoplasmic staining. Of five OEAs with no apparent β-catenin dysregulation, three cases showed weak or undetectable SHARP expression and two cases showed moderate to strong SHARP expression. Overall, the studies of human colorectal carcinomas and OEAs indicate that SHARP gene expression is increased in cancers with constitutively dysregulated β-catenin/TCF transcription (P = 0.006).
To assess SHARP protein expression in normal and neoplastic cells, a rabbit polyclonal antibody was raised against an NH2-terminal fragment of SHARP (amino acids 2–202). This anti-SHARP antibody specifically recognized endogenous SHARP protein of ∼400 kDa in human colon cancer–derived cell lines (Fig. 2A). Evidence for the specificity of the anti-SHARP antibody was shown by reduced levels of SHARP protein in 293 cells transiently transfected with a SHARP-specific siRNA (Fig. 2A). Consistent with our observations on the up-regulation of SHARP transcript levels in human cancer cells with defects in β-catenin regulation, elevated SHARP protein levels were seen in five independent RK3E cell lines neoplastically transformed by β-catenin oncogenic alleles (41) relative to the levels of SHARP expression in parental RK3E cells (Fig. 2B). To further assess SHARP protein expression in tumors with defects in β-catenin dysregulation, we used the anti-SHARP antibody in immunohistochemical studies on mouse colon adenomas and carcinomas. We studied 30 colon tumors from six CDX2P-NLS Cre; Apc+/loxP mice and found, in contrast to the membrane staining for β-catenin in adjacent normal epithelium, that the neoplastic cells showed strong cytoplasmic and nuclear staining for β-catenin (Fig. 2C,, top). In the mouse colon tumors, strong nuclear staining for SHARP was seen in the neoplastic cells compared with weak staining in adjacent normal epithelium (Fig. 2C , bottom) Attempts to use the anti-SHARP antibody in immunohistochemical studies of human primary colon tumors generally revealed stronger nuclear staining for SHARP in neoplastic cells compared with the patterns seen in adjacent normal mucosa, although variable cytoplasmic background staining was observed in several cases (data not shown). The higher apparent background staining for the anti-SHARP antibody in formalin-fixed and embedded human tissue specimens compared with mouse tissues may reflect cross-reactivity with unrelated proteins and/or technical problems relating to variations in tissue quality and fixation. Nevertheless, in aggregate, our data provide strong support for the view that SHARP gene and protein expression is often elevated in cancers with mutational defects leading to β-catenin dysregulation.
We considered the hypothesis that the SHARP gene might itself be a direct downstream target of β-catenin/TCF action in the nucleus. In support of this hypothesis, we observed modest to moderate inhibition of SHARP transcript levels in four of five colon cancer cell lines in which we ectopically expressed a dominant-negative TCF4 protein that inhibits expression of β-catenin/TCF-regulated genes (data not shown). We also observed a modest increase in SHARP transcript levels in HEK293 cells stably overexpressing a mutant stabilized form of β-catenin (data not shown). Further evidence for a direct role of β-catenin in the regulation of SHARP was either inconsistent (e.g., treatment of cells with selected inhibitors of GSK3β yielded some positive and some negative findings depending on the inhibitor used) or not supportive (e.g., studies with hormone-regulated variants of the β-catenin protein; data not shown). We therefore favor a model where β-catenin/TCF may play a more indirect role in regulating SHARP gene expression. However, even in the absence of a direct role for β-catenin in regulation of SHARP transcription, the elevated expression of SHARP in cancers with β-catenin dysregulation stimulated further studies of SHARP function in the cancer process.
SHARP is a positive regulator of Wnt/β-catenin/TCF signaling in vertebrates. We next assessed whether SHARP could affect the activity of the model TCF-dependent reporter gene construct TOPFLASH. Consistent with prior studies (41), ectopic expression of a mutant oncogenic form of β-catenin (S33Yβ-catenin) strongly activated TCF-dependent transcription in 293 cells (Fig. 3A,, left). Cotransfection of an expression construct for SHARP augmented S33Yβ-catenin–induced reporter gene activity by ∼3-fold (Fig. 3A,, left). In contrast to the effects of SHARP on TCF-dependent gene expression in cells where β-catenin levels were elevated, SHARP had negligible effect on basal TCF activity (Fig. 3A,, middle). To assess whether the enhancement of SHARP of TCF activity was due to an effect on β-catenin protein levels, we used Western blot assays to monitor total levels of expression of the transfected β-catenin protein as well as the levels of free pool of β-catenin. The free pool of β-catenin was assessed using a pull-down assay with a recombinant protein containing GST fused upstream of the cytoplasmic tail of E-cadherin (49). The levels of total and free S33Yβ-catenin protein were not changed by SHARP expression (Fig. 3A , right).
To determine if SHARP could also enhance TCF reporter gene activity in cells where endogenous WT β-catenin was stabilized, we stabilized β-catenin by expression of a polypeptide derived from the GID of Axin1. The GID polypeptide has previously been shown to robustly activate β-catenin/TCF-mediated transcription through inhibition of GSK3β activity (50). Expression of the GID protein significantly activated TCF-dependent transcription, and ectopic expression of SHARP further augmented the effect of GID on TCF transcription by ∼3-fold (Fig. 3B,, left). Expression of SHARP did not alter the total levels of GID-stabilized endogenous β-catenin or the levels of the free pool of β-catenin (Fig. 3B,, right). The ability of SHARP to enhance β-catenin/TCF-dependent transcription required full-length SHARP protein (Fig. 3C). Although they were expressed at levels similar to those of full-length SHARP (data not shown), NH2-terminal and COOH-terminal fragments of SHARP failed to augment the effects of β-catenin on TCF-dependent transcription (Fig. 3C). Consistent with the ability of SHARP to augment the effects of β-catenin on a TCF reporter gene, SHARP increased by ∼2-fold the β-catenin–mediated induction of the β-catenin/TCF-regulated genes AXIN2 (18–20) and DKK-1 (Fig. 3D; refs. 16, 17).
We next sought to address the role of endogenous SHARP in β-catenin/TCF-mediated transcription. We used siRNAs to reduce endogenous SHARP levels in a polyclonal population of 293 cells stably expressing the S33Yβ-catenin mutant protein (i.e., 293/S33Yβ-catenin). In 293/S33Yβ-catenin, transfection of SHARP-specific siRNAs (Si1 or Si2) reduced SHARP transcript levels and protein levels by >70% compared with cells transfected with a control (Sc) siRNA (Fig. 4A). In 293/S33Yβ-catenin, transfection of SHARP-specific siRNAs not only reduced TCF reporter gene activity by >70% (Fig. 4A) but also inhibited S33Yβ-catenin–induced expression of the β-catenin/TCF target gene AXIN2 by ∼50% compared with cells transfected with the control siRNA (Fig. 4B). To address the role of SHARP in human cancer cells with defective β-catenin regulation, we generated stable cell lines of HCT116 (β-catenin mutant) human colon cancer cells expressing shRNAs against SHARP (Sh1 or Sh2) or a control (Sc) shRNA. Expression of the SHARP-specific shRNA led to a reduction in SHARP transcript levels and SHARP protein levels (Fig. 4C) as well as a reduction in expression of the β-catenin/TCF target gene AXIN2 (Fig. 4C). As expected from our studies above indicating that SHARP acts downstream of stabilized β-catenin, we found that shRNA-mediated inhibition of SHARP expression had no effect on total cellular levels of β-catenin or the free pool of β-catenin (Fig. 4D). Taken together, the studies and data in Figs. 3 and 4 indicate that SHARP has a positive and significant role in regulating β-catenin/TCF transcription and acts downstream of stabilized β-catenin.
SHARP enhances β-catenin/TCF-mediated transcription in a TCF-dependent manner. To address the means by which SHARP exerts effects on β-catenin/TCF-mediated transcription, we pursued studies of the effects of SHARP on the activity of TCF proteins lacking the well-characterized β-catenin–binding region at the NH2 terminus. We used two chimeric proteins containing sequences of the TCF family member lymphoid enhancer factor-1 (LEF-1). The NH2-terminal β-catenin–binding domain of LEF-1 was deleted from both chimeric proteins (Fig. 5A). In one chimeric protein, full-length WT β-catenin sequences were fused in-frame downstream of the LEF-1 sequences (LEF-1ΔN-β-cat, Fig. 5A), whereas in the other fusion protein the herpes simplex virus VP16 transcriptional activation domain was fused in-frame downstream of the LEF-1 sequences (LEF-1ΔN-VP16, Fig. 5A). SHARP was able to augment the TCF-dependent reporter gene activity of both the LEF-1ΔN-β-cat and LEF-1ΔN-VP16 fusion proteins (Fig. 5B). To further establish that SHARP mediates its effects via functional interaction with TCF and not via interaction with β-catenin, we generated a construct encoding the yeast GAL4 DNA-binding domain fused to full-length mutant (S33Y) β-catenin sequences (GAL4-S33Yβ-cat, Fig. 5A). SHARP was able to enhance the effect of the GAL4-S33Yβ-cat fusion protein on a TCF-dependent reporter gene (Fig. 5C,, left). However, consistent with the notion that SHARP does not function via interaction with β-catenin, SHARP failed to enhance the activity of the GAL4-S33Yβ-cat protein on a GAL4-dependent reporter (Fig. 5C , right). Studies to establish that TCFs and SHARP proteins bind to one another in in vitro binding assays did not yield data to support a direct interaction between TCFs and SHARP (data not shown).
SHARP functions in both initiation and maintenance of β-catenin–dependent neoplastic transformation. To address the functional role of SHARP in human cancer cells with defective β-catenin regulation, we generated polyclonal populations of HCT116 and LS174T human colon cancer cells expressing shRNAs (Sh1 or Sh2) against SHARP or a control shRNA (Sc). Of note, both HCT116 and LS174T carry oncogenic (activating) mutations in β-catenin. Expression of the SHARP-specific shRNA led to a reduction in SHARP levels in both HCT116 and LS174T cells (Figs. 4C and 6A). Consistent with a role for SHARP in the maintenance of neoplastic growth properties of human colon cancer cells, anchorage-independent (soft agar) growth of both HCT116 and LS174T cells expressing SHARP-specific shRNAs was markedly inhibited (Fig. 6B).
Our prior work has shown that cancer-derived mutant forms of β-catenin, but not WT β-catenin, can readily generate foci of neoplastically transformed cells when introduced into the RK3E cell line, an adenovirus E1A-immortalized rat kidney epithelial line (41). As such, we sought to determine whether SHARP had a contributing role in β-catenin–induced neoplastic transformation of RK3E. Four clonal G418-resistant RK3E lines with stable ectopic expression of SHARP (e.g., Sharp3 and Sharp5) and three control clones carrying only the empty vector (e.g., vec1 and vec3) were generated. Focus formation assays with the cells were done using replication-defective retroviruses encoding the S33Y oncogenic mutant form of β-catenin. No foci were observed in any of the RK3E/SHARP and control RK3E/vec clones following infection with retroviruses encoding WT β-catenin or the empty retroviral vector (data not shown). However, when S33Yβ-catenin was transduced, there was a 2- to 7-fold increase in the number of foci for the four RK3E/SHARP lines when compared with the three control cell lines (Fig. 6C). These results, taken together with the results of our studies of the consequences of inhibiting SHARP expression in HCT116 and LS174T colon cancer cells, strongly implicate SHARP as an important factor in β-catenin–mediated neoplastic transformation.
Mutations altering essential factors in the canonical Wnt pathway, such as inactivation of the APC or Axin1 tumor suppressor proteins or activating (oncogenic) mutations in the phosphorylation and degradation motif at the NH2 terminus of β-catenin, have been conclusively implicated in cancer. Stabilization of β-catenin via Wnts or cancer-related mutations leads to β-catenin nuclear accumulation and enhanced binding of β-catenin to TCFs. In turn, β-catenin/TCF complexes affect expression of genes with effects on cell fate, proliferation, and other processes. Data implicating β-catenin/TCF complexes in regulation of specific target genes, including proto-oncogenes, such as c-MYC and cyclin D1, have been offered (12, 13). However, understanding of the role of β-catenin/TCF pathway defects in the pathogenesis and clinical behavior of colon and other cancers remains far from complete. In addition, more in-depth knowledge is needed of the identity and role of nuclear factors that cooperate with β-catenin and TCFs to regulate transcription of Wnt/β-catenin/TCF target genes in normal and cancer cells.
Here, we have provided data that advance knowledge of Wnt/β-catenin/TCF signaling on several fronts. We have shown that SHARP, a human homologue of Drosophila spen, functions as a significant positive regulator of β-catenin/TCF transcription and that the contribution of SHARP in this regard is dependent on its interaction with TCF but not with β-catenin. Moreover, we have provided evidence that the function of SHARP as a positive regulator in Wnt/β-catenin/TCF signaling is relevant in neoplastic transformation in several settings, including colon cancer.
Interestingly, we have found that SHARP is significantly overexpressed in human colon and OEA cancer cells with defects in β-catenin regulation and in mouse colon tumors resulting from Apc inactivation. As a positive regulator of β-catenin/TCF transcription in cancer cells with mutational defects leading to β-catenin stabilization, SHARP may serve to reinforce and maintain the altered phenotype through constitutively augmenting transcription of β-catenin/TCF target genes. This notion has been supported by our studies showing that SHARP plays an important role in β-catenin–mediated neoplastic transformation in RK3E and human colon cancer cells.
SHARP was initially uncovered as a transcriptional corepressor for steroid hormone receptors (51) and the DNA-binding protein RBP-Jk, which mediates Notch signaling (52). These prior studies also showed that SHARP exerts its inhibitory effects by recruiting a corepressor complex, including SMART/NCoR and HDACs, through its SPOC domain (39, 51, 52). In contrast to its role in transcriptional repression, the means by which SHARP might function in transcriptional activation are unclear. Prior to the data reported here for SHARP, the role of spen homologues in transcriptional activation was limited to data on the mouse homologue MINT. The MINT protein had been suggested to bind to DNA in a sequence-specific manner via its RRM domains (38). Interestingly, reminiscent of the case for the effects of SHARP on β-catenin/TCF target gene transcription described here, MINT did not have a strong stimulatory effect in its own right on the osteocalcin promoter. Rather, MINT augmented transcriptional activation of the promoter by Runx2 (38, 53). Based on the apparent localization of MINT in the nuclear matrix, Sierra et al. proposed that MINT might serve as a nuclear matrix-associated scaffolding factor that organizes the Runx2-dependent activation of the osteocalcin gene. A similar mechanism might be plausible for the effects of SHARP on β-catenin/TCF target genes in part because our data imply that β-catenin is not required for SHARP to augment transcriptional activation of TCF-regulated genes (e.g., the enhanced activity of the LEF-1ΔN-VP16 fusion protein by SHARP). However, weighing in against this proposal for the function of SHARP is the unlikely possibility that specific DNA-binding sites for SHARP would be present in the TOPFLASH construct. Another possible mechanism for SHARP in transcriptional activation via its functional interactions with TCFs and/or the transcriptional machinery is suggested by the fact that SHARP can support gene silencing by recruiting chromatin modifying/remodeling cofactors. As such, it is conceivable that in some contexts SHARP might recruit factors that alter the local chromatin structure surrounding the TCF sites or that help to sequester the components of a repression complex, such as Groucho and HDACs, away from the TCF sites. This in turn could improve the accessibility of transacting factors, such as β-catenin, to the promoter region of the target genes.
Although the effects of SHARP on TCF target gene transcription seem to be the most likely explanation for how SHARP affects expression of β-catenin/TCF target genes, the RRMs found in SHARP are also found in various mRNA splicing proteins. Indeed, both SHARP and OTT1, another member of Spen protein family, have been detected in the purified spliceosome (54). Additionally, the RRMs in SHARP are capable of binding the RNA steroid receptor RNA coactivator (51). Furthermore, a recent study by Hiriart et al. (55) showed that OTT3, a newly identified member of the Spen protein family, may be involved in mRNA splicing. In this recent study, it was also suggested that SHARP can interact with the EBV mRNA export factor EB2 (55), although the relevance of this interaction requires further investigation. These lines of evidence suggest that SHARP might be involved in splicing and nuclear export of RNAs. Thus, we cannot exclude the possibility that SHARP might modulate endogenous β-catenin target gene expression via regulation of RNA maturation or transport. However, for our reporter gene studies, we clearly show that a significant part of the effect of SHARP is dependent on the presence of TCF sites in the promoter. Moreover, the effect of SHARP is obvious even on a TCF reporter gene where no mRNA splicing is necessary for reporter activity. Future studies should yield more detailed insights into the means by which SHARP functions to augment β-catenin/TCF transcription in normal and neoplastic cells.
Grant support: NIH grants CA082223, CA085463, and CA094172.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
We thank Drs. Richard Baer, A. Heldman, and Bert Vogelstein for generously providing the plasmids used in this study.