The Wnt/β-catenin pathway has been implicated in human cancers. Here, we show that TC1 (C8orf4), a small protein present in vertebrates, functions as a positive regulator of the pathway. TC1 interacts with Chibby (Cby) and thereby enhances the signaling pathway by relieving the antagonistic function of Cby on the β-catenin–mediated transcription. Upon coexpression in mammalian cells, TC1 redistributes from nucleolus to nuclear speckles, where it colocalizes with Cby. TC1 up-regulates the expression of β-catenin target genes that are implicated in invasiveness and aggressive behavior of cancers, such as metalloproteinases, laminin γ2, and others. Our data indicate that TC1 is a novel upstream regulator of the Wnt/β-catenin pathway that enhances aggressive behavior of cancers. (Cancer Res 2006; 66(2): 723-8)
TC1 (C8orf4) was one of the up-regulated genes in high-grade gastric cancers in our previous expression profiling study (1). In the study, gastric cancers of various types were analyzed in comparison with their own normal mucosa controls. Among poorly differentiated carcinomas, including diffuse type, which consisted of 25% of the series, TC1 expression was up-regulated 1.67 times that of normal controls in average. It was significantly higher than the ratio of 0.94 in well-differentiated carcinomas (P < 0.04, t test), suggesting that it might be implicated in poor differentiation and/or aggressive biological behavior of cancers. We, then, decided to investigate the biological function of TC1 in detail.
TC1 was first described as one of the genes elevated in expression in thyroid cancers (2). It is present only in vertebrates and encodes a protein of 106 amino acids without identified functional domain (2, 3). It has been implicated in cancer and signal transduction (4, 5). However, the clinical relevance and precise biological functions have not heretofore been elucidated.
Here, we show that TC1 is a novel positive regulator of the Wnt/β-catenin signaling pathway, which has been widely implicated as regulating cell proliferation and differentiation in cancers and in development (6–10). TC1 interacts with Chibby (Cby), which regulates the β-catenin–mediated transcription antagonistically (11) and thereby enhances the signaling pathway through relieving the suppression by Cby. TC1 up-regulates β-catenin target genes that are implicated in invasiveness and aggressive behavior of cancers. The biological implication of TC1 regulation in cancer is discussed.
Materials and Methods
Cells and plasmids. HEK293, 293T, HeLa, NIH3T3, and AGS were grown in DMEM supplemented with 10% fetal bovine serum and 100 units/mL streptomycin/penicillin at 37°C in a humidified atmosphere of 5% CO2. Gastric cancer cell lines KATO-III and MKN45 were grown in RPMI 1640 with 10% fetal bovine serum similarly.
In-frame fusions of HA-TC1, HA-TC1ΔN (amino acids 51-106), and HA-TC1ΔC (amino acids 1-60) were constructed using pcDNA3 vector at EcoRI and XhoI sites for the mammalian expression (Invitrogen, Carlsbad, CA). For the bacterial expression, pET-TC1, pGEX-TC1, pGEX-TC1ΔN (amino acids 51-106), and pGEX-TC1ΔC (amino acids 1-60) were similarly cloned. All clones were confirmed by DNA sequencing. Cby-Flag, β-catenin-myc, pGEX-CbyΔN, pGEX-CbyΔC, pMALC2-Cby, and pET-β-catenin were described previously (11, 12). To generate point mutants, we used QuickChange II Site-Directed Mutagenesis kit (Stratagene, La Jolla, CA.) using complementary oligonucleotides: 5′-GCCATAGATCAAGATGAGGAGGAGAAAACGCGTG-3′ and 5′-CACGCGTTTTCTCCTCCTCATCTTGATCTATGGC-3′ for Val74Glu; 5′-CTGATGGCCTTGAAGGAGAGGACAAAAGACAAG-3′ and 5′-CTTGTCTTTTGTCCTCTCCTTCAAGGCCATCAG-3′ for Lys86Gln; 5′-GGCCTTGAAGAAGAGGAAAAAAGACAAGCTTTTCCAG-3′ and 5′-CTGGAAAAGCTTGTCTTTTTTCCTCTTCTTCAAGGCC-3′ for Thr88Lys; and, 5′-CAAGCTTTTCCAGTTTCGGAAACTGCGGAAATATTCC-3′ and 5′-GGAATATTTCCGCAGTTTCCGAAACTGGAAAAGCTTG-3′ for Leu96Arg.
Northern blot analysis. An adult human multi-tissue Northern blot (Clontech, Palo Alto, CA) was hybridized with a 32P-labeled full-length TC1 cDNA probe. The membrane was washed twice at 65°C in 0.5× SSC containing 5% SDS and twice in 1× SSC containing 0.125% SDS followed by autoradiograph.
Yeast two-hybrid screening. The yeast two-hybrid screening was done using a human fetal brain cDNA library and BD Matchmaker Two-Hybrid System3 (Clontech), as described previously (13). A bait plasmid expressing GAL4BD-TC1 fusion was constructed by inserting a 331-bp EcoRI/PstI fragment of full-length TC1 into a pGBKT7 vector. Colonies grown on the selection media were selected for LacZ gene transactivation by α-galactosidase activity in X-α-gal assay. Prey plasmids were recovered from the positive colonies and were subjected to sequence analysis. Reciprocal experiments were done exchanging the bait and prey vectors.
Protein interaction and binding site analysis. HEK293T cells were transiently transfected with Cby-Flag and wild-type or mutant HA-TC1, using a LipofectAMINE 2000 transfection kit (Invitrogen). Cells were harvested after culture for 48 hours with or without MG-132. For the triple interaction study, HA-TC1 (10 μg), Flag-Cby (0.5 μg), and various amounts of Myc-β-catenin were transfected in 5 × 106 HEK293T cells. Harvested cells were lysed in the sample buffer [20 mmol/L Tris (pH 8), 75 mmol/L NaCl, 1.5 mmol/L MgCl2, 1 mmol/L EGTA, 1 mmol/L phenylmethylsulfonyl fluoride, 0.5% NP40, 10% glycerol, 10 μg/mL leupeptin, and 10 μg/mL aprotinin] at 4°C. After the centrifugation at 10,000 × g for 30 minutes, expressed proteins were pulled down from the supernatant using either anti-hemagglutinin (anti-HA; Roche, Mannheim, Germany) or anti-Flag affinity matrix (Sigma, St. Louis, MO), according to the manufacturer's instruction. Beads were washed five times in the reaction buffer, and pulled-down proteins were separated on 15% SDS-PAGE. Proteins were blotted onto nitrocellulose membrane and probed with anti-HA (Roche), anti-Flag (Sigma), and/or anti-Myc antibodies (Santa Cruz Biotechnology, Santa Cruz, CA) followed by goat anti-mouse and anti-rat second antibodies (Amersham Biosciences, Piscataway, NJ) and visualized using enhanced chemiluminescence method (Amersham Biosciences). For in vitro binding analysis, His-tagged TC1 was incubated overnight with same amount of either GST-CbyΔN or GST-CbyΔC in the lysis buffer at 4°C. Alternatively, MBP-tagged Cby was incubated with either GST-TC1ΔN or GST-TC1ΔC. For the negative control, glutathione S-transferase (GST) protein alone was applied. After GST pulldown, proteins were analyzed using goat anti-GST (Amersham Biosciences) and/or mouse anti-T7·Tag antibody (Novagen, Darmstadt, Germany). For endogenous Cby in HEK293T, AGS, and KATO-III cells, 50 μg total proteins were loaded in each lane and analyzed by Western blotting using rabbit anti-Cby antiserum (11).
Immunofluorescence microscopy. HEK293T, HeLa, and MKN45 cells were transiently transfected using HA-TC1, HA-TC1 mutants, and/or Cby-Flag. Coverslips were immunostained using rat monoclonal anti-HA, rabbit anti-Flag, and/or rabbit anti-TC1 antiserum as primary antibodies. For double staining, rabbit anti-nucleolin antiserum (Santa Cruz Biotechnology) was applied together with monoclonal anti-HA antibody. After washing with PBS, Texas red–labeled anti-mouse and FITC-labeled anti-rabbit second antibodies (Jackson Immunoresearch, West Grove, PA) were applied. After DNA staining with 4′,6-diamidino-2-phenylindole (Sigma), cells were viewed using Olympus BX51 fluorescence microscope. For controls, primary antibodies were replaced with normal rabbit and/or mouse sera.
TOP/FOPFLASH analysis. At 24 hours after transfection of the luciferase reporters 8× TOP/FOPFLASH with various combinations of DNA, HEK293T cells were analyzed using a Luciferase reporter assay kit (Promega, Madison, WI) and Luminometer (ThermoLabsystems, Helsinki, Finland). Experiments were repeated in quadruplicate, and the fold changes were calculated using values normalized on the β-galactosidase expression. Alternatively, TC1 was knocked down in HEK293 cells using either one of two synthetic TC1-siRNAs (Stealth, Invitrogen), 5′-ACACAGACCAAGAATCACTAGAAAG-3′ or 5′-TCATCATGTCCACGTCGCTACGAGT-3′. For controls, the same amount of control RNA (Stealth RNAi Negative Control Medium GC, Invitrogen) was applied instead. The efficiency of TC1 knockdown was analyzed using real-time PCR at 48 hours following transfection. For real-time PCR analysis, Taqman Gene Expression Assay (Hs00535539_s1) was applied using Taqman MGB probe (6-FAM dye labeled), 5′-CCACTTCGACACAGCCTCTCGTAAG-3′ (Applied Biosystems, Foster City, CA). Measurements were done in triplicate using human β-actin gene as endogenous control. For the statistical analyses, ANOVA and/or Wilcoxon signed rank test was applied using the SPSS software (SPSS, Inc., Chicago, IL).
Downstream gene analysis. For the semiquantitative reverse transcription-PCR (RT-PCR) of target gene expression, HEK293T and/or AGS cells were transiently transfected with TC1-HA and harvested at 6 and 24 hours after transfection. Alternatively, the target gene expression was analyzed using KATO-III and HeLa cells at 24 and 48 hours after TC1-siRNA transfection. Total RNA was extracted using Trizol agent (Invitrogen), and cDNA was synthesized using M-MLV reverse transcriptase (Promega). PCR was done for 25 to 33 cycles as described previously (1). Primers for TC1: 5′-CAAGCCATCATCATGTCCAC-3′ and 5′-GTTGCCCACGGCTTTCTTAC-3′; β-catenin, 5′-TGGATACCTCCCAAGTCCTG-3′ and 5′-GGGATGAGCAGCATCAAACT-3′; Cby, 5′-AAGTCGGCATCTCTCTCCAA-3′ and 5′-GGCCATTTTCAAACTTCAGG-3′; c-Myc, 5′-TGCTCCATGAGGAGACACC-3′ and 5′-CTCTGACCTTTTGCCAGGAG-3′; cyclin D1, 5′-TCCTCTCCAAAATGCCAGAG-3′ and 5′-GGCGGATTGGAAATGAACT-3′; c-MET, 5′-GCATTTTTACGGACCCAATC-3′ and 5′-GCTGCAAAGCTGTGGTAAAC-3′; CD44, 5′-AAGACACATTCCACCCCAGT-3′ and 5′-CAGAGGTTGTGTTTGCTCCA-3′; matrix metalloproteinase-7 (MMP-7), 5′-TGCTCACTTCGATGAGGATG-3′ and 5′-ACCCAAAGAATGGCCAAGTT-3′; MMP-14, 5′-CCCCGTTGTCTCCTGCTC-3′ and 5′-GGGAGGCAGGTAGCCATATT-3′; and LAMC2, 5′-CAGAACCGAGTTCGGGATAC-3′ and 5′-TGGTCTGAGGCAGGAATGTT-3′. AGS cells were also analyzed by Western blotting at 6 and 24 hours after TC1 transfection using anti-LAMC2 and anti-c-Myc antibodies. The expression was analyzed by densitometer and normalized on β-actin loading control.
TC1 expression in normal tissue and cultured cells. We first examined the expression of TC1 in human tissues. Upon Northern blotting, a single TC1 band of about 1.4 kb was present in most adult human tissues, notably in liver, heart, placenta, lung, spleen, and kidney (Fig. 1A), and in the gastric cancer cell line MKN45. However, TC1 protein was not readily detected by Western blotting in whole-cell preparations (Fig. 1B). Transiently expressed TC1 in cultured cells also did not result in protein accumulation, suggesting a post-translational regulation. Levels of TC1 increased greatly when HEK293T cells were treated with a proteasome inhibitor MG-132 up to a concentration of 500 nmol/L (Fig. 1B), suggesting regulation of TC1 by the proteasome degradation pathway (14). No other positive band was noted upon Western blots using anti-HA or affinity-purified anti-TC1 antiserum.
Specific interaction of TC1 with Cby. We next asked how TC1 might function in the development and biology of cancers. TC1 itself was not tumorigenic on the transformation colony assay using NIH3T3 cells (data not shown). We then screened for binding partners of TC1 using a yeast two-hybrid system. The full-length TC1 bait did not show intrinsic activity on the reporters. From 1.27 × 106 independent transformants, 92 colonies were obtained in the selection media. Upon DNA sequencing, two colonies had full-length cDNA encoding Cby. This would be a very interesting interaction partner for TC1 as it would provide a direct link to the β-catenin pathway. Cby is a small protein of 126 amino acids that acts as an antagonistic regulator of β-catenin mediated transcription (Fig. 2A).
The interaction between TC1 and Cby was analyzed using HEK293T cells cotransfected with HA-TC1 and Flag-Cby. Upon TC1 pull-down using an anti-HA affinity matrix, Cby was coprecipitated, whereas it was not detected in negative control transfected with Flag-Cby alone, confirming the in vivo interaction (Fig. 2B). For the binding site analysis, we analyzed interactions of recombinant deletion mutants with full-length counterparts (Fig. 2A). Upon GST pulldown, MBP-Cby protein was coprecipitated with GST-TC1ΔN but not with GST-TC1ΔC, showing that the binding site of TC1 was in the COOH terminus (Fig. 2C,, left). Similarly, His-TC1 was coprecipitated with GST-CbyΔN but not with GST-CbyΔC, indicating that the binding site of Cby was also in the COOH terminus (Fig. 2C,, right). Both TC1 and Cby have amino acid sequences compatible with leucine zippers at the COOH termini (Fig. 2A). TC1 also has a putative nuclear localization signal (NLS) at amino acids 85 to 101 and a sequence suggestive of a potential nuclear export signal at the extreme COOH terminus.
To analyze the binding site further, four mutants of TC1 were produced having single amino acid replacement: TC1-Val74Glu, TC1-Lys86Gln, TC1-Thr88Lys, and TC1-Leu96Arg (Fig. 2A). HEK293T cells were cotransfected with Flag-Cby, and each mutant was tagged with HA. Wild-type and mutant TC1 expressed similarly (Fig. 2D,, bottom). Upon the HA-immunomatrix pulldown, Cby coprecipitated with TC1-Val74Glu and TC1-Leu96Arg mutants, whereas it did not with TC1-Lys86Gln or TC1-Thr88Lys (Fig. 2D , top), suggesting that the binding site of TC1 was not in the leucine zipper-like motif but in the putative NLS; thus, the interaction could influence intracellular localizations. Nuclear translocation of β-catenin is implicated as a critical step in the Wnt/β-catenin pathway regulation.
Intracellular localization of TC1. We then analyzed the intracellular localization of wild-type and mutant TC1 in HeLa, HEK293T, and MKN45 cells. Wild-type TC1 was mostly expressed in nucleoli and also in the nucleoplasm and cytoplasm (Fig. 3, row 1). It colocalized with nucleolin, a marker of nucleolus (Fig. 3, row 2). TC1-Val74Glu mutant was expressed similarly to the wild type (data not shown). TC1-Lys86Gln (Fig. 3, row 3) and TC1-Thr88Lys (data not shown) mutants were visualized as irregular perinuclear cytoplasmic aggregates with minimal nuclear expression, suggesting that the putative NLS indeed facilitated nuclear transport of TC1. The mutant TC1-Leu96Arg accumulated at the periphery of nucleoli being consistent with the granular component (15): no evident cytoplasmic staining was present, suggesting interrupted nuclear export (Fig. 3, row 4). The intracellular distribution of TC1 changed dramatically upon cotransfection with Cby: TC1 was not at the nucleoli any more but at small nuclear speckles (Fig. 3, row 5), coinciding with Cby (Fig. 3, row 6). The distribution was compatible with that of Cby alone (Fig. 3, row 7; ref. 11), suggesting that Cby and TC1 are preferential binding partners of each other, and Cby guides the intracellular localization of TC1 in vivo when they are present in abundance. Fine granular immunostaining of TC1 and Cby was also present in the cytoplasm of some cells.
TC1 and β-catenin compete for interaction with Cby. Cby also interacts with β-catenin through its COOH terminus (11). Thus, it is possible that TC1 and β-catenin compete for binding to Cby. To test this possibility, HEK293T cells were cotransfected using Flag-Cby and HA-TC1 along with various amounts of Myc-β-catenin. In the total cell lysate, β-catenin elevated proportionally as the transfected DNA amount increased, whereas Cby and TC1 were consistent regardless of β-catenin (Fig. 4). Upon the pulldown of Cby using Flag-immunomatrix, coprecipitated TC1 decreased in proportion with the increase of coprecipitated β-catenin (Fig. 4), suggesting that TC1 competes with β-catenin for the interaction with Cby.
TOP(FOP)FLASH analysis for β-catenin transcriptional activity. We then did TOP/FOPFLASH reporter assay to analyze the effects of TC1 expression on the Wnt/β-catenin pathway (16). HEK293T and/or HEK293 cells were used because they have been shown to yield most reproducible TOPFLASH results. Experiments were repeated in quadruplicate, and results were normalized on the β-galactosidase expression. As described previously (11), overexpression of Cby in HEK293T cells repressed the transcriptional activation of the β-catenin-dependent Tcf (T-cell factor) reporter TOPFLASH in a dose-dependent manner (P < 0.0001, ANOVA test), showing minimal effects on the mutant reporter FOPFLASH (Fig. 5A , lanes 3-5). Overexpression of TC1 alone did not show significant effect on the reporter (lanes 6-8). As TC1 was cotransfected with β-catenin and Cby, TC1 relieved the Cby-induced suppression of the β-catenin transcriptional activity in a dose-dependent manner with two different dosages of Cby DNA (lanes 9-11 and 12-14: P < 0.001 and P < 0.008, ANOVA test). The fold elevation in luciferase activity in lane 9 of this representative experiment was significantly higher than in lane 3, which did not have TC1 but otherwise same DNA composition (P < 0.029, Wilcoxon signed rank test).
Although gain-of-function reveals the potential activity of a protein, loss-of-function is more relevant to defining the normal requirements of a protein. For the loss-of-function analysis, TC1 was knocked down in HEK293 cells, which express endogenous TC1 more than HEK293T cells (data not shown). TC1 mRNA expression was knocked down efficiently using either one of two small interfering RNAs (siRNA) compared with controls using same amount of mock RNA (see Supplementary Fig. S1). Upon TOPFLASH analysis, Cby-induced suppression of reporter activity was augmented significantly in HEK293 cells as TC1-siRNA was increased to 10 nmol/L (P < 0.004, ANOVA test), showing the effect opposite to TC1 overexpression (Fig. 5B). The difference was not as much as the gain-of-function assay, probably reflecting the relatively low expression of endogenous TC1.
Target gene regulation by TC1 in cultured cells. We further investigated the regulation of the Wnt/β-catenin pathway target genes that have been implicated in cancer proliferation and invasion. The involvement of c-Myc and cyclin D1 in cancer growth has been characterized well (17, 18). MMP-7 (19, 20), MMP-14 (21, 22), CD44 (23, 24), c-Met (25, 26), and LAMC2 (27, 28) are β-catenin target genes that are associated with invasiveness of cancers. HEK293T and AGS cells, a gastric cancer cell line, were transiently transfected with TC1, and target gene expressions were analyzed using semiquantitative RT-PCR. As TC1 increased, MMP-7 and MMP-14 were up-regulated markedly (Fig. 6A,, left). Cyclin D1 and c-Met were also up-regulated significantly, although the expression was relatively low. LAMC2, c-Myc, and CD44 were up-regulated in AGS cells but not in 293T cells significantly (Fig. 6A,, left), suggesting that the extent of regulation might vary depending on cell types. Upon Western blotting, LAMC2 and c-Myc proteins were up-regulated considerably compared with the β-actin loading control (Fig. 6B). LAMC2 was shown as two bands of 140 and 155 kDa, as described previously (29). Upon Western blotting, endogenous Cby was expressed similarly in HEK293T, AGS, and KATO-III cells (Fig. 6C). In HEK293T cells, a 15-kDa band was shown in addition to the 17-kDa protein. The biological significance is not clear.
The loss-of-function effect on target gene expression was analyzed using KATO-III cell line, which was originated from a high-grade diffuse-type gastric carcinoma (30) and expressed TC1 more than the cells used in the gain-of-function study (data not shown). TC1 mRNA expression was knocked down efficiently using either one of two siRNAs compared with controls using same amount of mock RNA (see Supplementary Fig. S1). Following TC1-siRNA transfection, LAMC2, c-Myc, c-Met, and CD44 were down-regulated considerably in 48 hours (Fig. 6A , right). MMP-7 and MMP-14 were also mildly down-regulated, whereas cyclin D1 did not change significantly. Both TC1-siRNAs showed similar results (see Supplementary Fig. S2). Together with the TOPFLASH analysis, our loss-of-function and gain-of-function studies indicate that TC1 regulates target genes of the Wnt/β-catenin pathway in mammalian cells. Given the low level of endogenous TC1 protein, the extent of downstream regulation of cancer-related genes is quite remarkable as shown by the loss-of-function study, suggesting that TC1 is a major regulator of the pathway in vertebrates.
We have shown that TC1 is a novel positive regulator of the Wnt/β-catenin signaling pathway. TC1 interacts with Cby in competition with β-catenin. Upon coexpression in mammalian cells, TC1 relocates from the nucleolus to small nuclear speckles coinciding with Cby, indicating the biological relevance of the interaction. Our TOPFLASH analysis data showed that TC1 did not up-regulate the reporter directly but relieved the suppression by Cby in a dose-dependent manner, being consistent with the competitive interaction data. The TC1 regulation of the reporter activity was reproduced using two different dosages of Cby DNA and confirmed in knockdown experiments.
The function of Cby as an antagonistic regulator of the Wnt/β-catenin pathway suggests that it could be a tumor suppressor gene (11, 31, 32). However, there have heretofore been no direct data linking Cby to any cancers (33). Our data suggest that TC1 regulation of Cby would be of considerable biological significance in the Wnt/β-catenin pathway regulation in cancer. The regulation of Cby by TC1 needs to be analyzed in various cancers.
TC1 seems to be regulated tightly through the proteasome degradation pathway as many other regulators are. Despite the relatively low TC1 protein level, the extent of β-catenin target gene regulation was remarkable as shown by the loss-of-function experiments, suggesting that TC1 is a major regulator of the Wnt/β-catenin pathway arising in the evolution of vertebrates.
TC1 up-regulates the β-catenin target genes that are implicated in cancer invasion and/or proliferation, including LAMC2, MMP-7, MMP-14, c-Met, CD44, c-Myc, and cyclin D1, as shown by the gain-of-function and loss-of-function analyses at the mRNA and/or protein levels in cancer cells. As an upstream regulator of the genes, TC1 may be implicated in the regulation of biological behaviors of cancers through coordinated activation of the molecular network involved in cancer invasion and proliferation. Taken together, our data suggest that TC1 is an upstream regulator of the Wnt/β-catenin pathway that promotes aggressive biological behavior of cancers and could potentially be a therapeutic target of cancers.
Note: Supplementary data for this article are available at Cancer Research Online (http://cancerres.aacrjournals.org/).
Y. Jung and S. Bang contributed equally to this work.
R.T. Moon is an investigator of the Howard Hughes Medical Institute.
Grant support: Korean Ministry of Science and Technology's National Research Laboratory Project grant M10400000305-05J0000-30510 and Molecular and Cellular Function Discovery Project grant M10401000003-05N0100-00310.
We thank Hyunseok Lee for statistical analyses.