Positive regulation of β-catenin–PROX1 signaling axis by DBC1 in colon cancer progression

Abstract

Aberrant activation of Wnt/β-catenin pathway contributes to colorectal cancer (CRC) progression. However, little is known about regulatory mechanisms of the β-catenin activity in cancer progression. Here we investigated the role of DBC1, which was recently reported as a negative regulator of SIRT1 and a transcriptional coactivator, in the regulation of Wnt/β-catenin signaling. We identified the genome-wide targets of DBC1 and found that loss of DBC1 inhibits the expression of β-catenin target genes including PROX1, a transcription factor linked to CRC progression. Mechanistically, DBC1 stabilizes LEF1–β-catenin interaction by inhibiting SIRT1-mediated β-catenin deacetylation, thereby enhancing LEF1–β-catenin complex formation and long-range chromatin looping at the PROX1 locus. Furthermore, DBC1 is also required for the transcriptional activity of PROX1, suggesting that DBC1 has a dual function in regulating β-catenin–PROX1 signaling axis: as a coactivator for both β-catenin and PROX1. Importantly, loss of DBC1 inhibited growth and tumorigenic potential of colon cancer cells, and DBC1 expression correlated with shorter relapse-free survival in patients with advanced CRC. Our results firmly establish DBC1 as a critical positive regulator of β-catenin–PROX1 signaling axis and a key factor in β-catenin–PROX1-mediated CRC progression.

INTRODUCTION

Wnt signaling promotes expression of a large spectrum of genes involved in various developmental and oncogenic processes.^1,2 β-catenin is a key factor of the canonical Wnt signaling pathway, and its aberrant activation is associated with the onset and progression of a variety of human cancers, including colorectal cancer (CRC). Upon Wnt signaling activation, β-catenin is stabilized by escaping from phosphorylation-dependent ubiquitination and degradation and translocates to nucleus to regulate Wnt target gene expression through interactions with transcription factors of the LEF/TCF family.^1–3 Although it is well established that the formation of nuclear β-catenin/LEF1 complexes has a crucial role in the activation of Wnt target genes, the details of the regulatory mechanisms of β-catenin activity and its interaction with LEF1 are still under investigation. Also, although various β-catenin coregulators have been identified,^1,3–7 the pathophysiological relevance of these coregulators to colorectal tumorigenesis remains elusive. PROX1, a Wnt/β-catenin-inducible transcription factor, induces colon cancer progression by promoting the transition from benign adenoma to carcinoma and the expansion of colon cancer stem cell population, and its expression is associated with poor outcome of colon cancer patients.^8–11 However, how the expression and activity of PROX1 are regulated in colon cancer cells is poorly understood.

Deleted in breast cancer (DBC1; also known as CCAR2) is a transcriptional coactivator for nuclear receptors and other transcription factors, as well as a negative regulator of deacetylases such as SIRT1 and HDAC3.^12–15 Recently, we have shown that DBC1 is overexpressed in breast cancer and that DBC1 functions as a coactivator for estrogen receptor (ER) and PEA3 in ER-positive and -negative breast cancer cells, respectively.^12,13 DBC1 inhibits SIRT1-mediated deacetylation of ER and PEA3 and consequently enhances their DNA-binding and transcriptional activities. In addition to breast cancer, DBC1 has been reported to be frequently overexpressed in various types of human cancer including CRC,^15–17 although the exact mechanism underlying DBC1 function is largely unknown. Here, we report a novel role of DBC1 as an integral component of the Wnt/β-catenin signaling pathway and as a key factor in β-catenin–PROX1 axis-mediated colon cancer progression.

RESULTS

DBC1 functions as a coactivator for β-catenin-mediated transcription

Given that CCAR1, a paralog of DBC1/CCAR2, has been shown to associate and cooperate with β-catenin,⁵ we tested the possibility that DBC1 may also interact with β-catenin. Endogenous interaction between DBC1 and β-catenin was detected in SW480, SW620 and HT-29 colon cancer cells (Figure 1a). Similarly, DBC1 was co-immunoprecipitated specifically with β-catenin from extracts of transiently transfected 293T cells (Supplementary Figure S1a). In vitro glutathione S-transferase pull-down assays confirmed their interaction (Figure 1b) and showed that the N-terminal domain of DBC1 binds to the C-terminal region of β-catenin (Supplementary Figures S1b–d), suggesting that DBC1 interacts directly with β-catenin. We next tested whether DBC1 can function as a coactivator for LEF1–β-catenin-mediated transcription. In reporter gene assays using the pGL3OT reporter containing LEF-responsive elements, β-catenin enhanced LEF1 activity, and coexpression of β-catenin and DBC1 further enhanced LEF1 function in a synergistic manner (Figure 1c). However, DBC1 alone had little effect on LEF1 activity. As DBC1 does not interact directly with LEF1 (Figure 1b), these results suggest that DBC1 functions as a secondary coactivator for LEF1-mediated transcriptional activation. To extend these observations, functional interaction between DBC1 and β-catenin was assessed in a mammalian one-hybrid system using a reporter plasmid containing Gal4-responsive elements and β-catenin tethered to Gal4 DNA-binding domain. Gal4-β-catenin strongly activated the reporter activity, as reported previously,^5,6 and this activity was further enhanced by DBC1 in a dose-dependent manner (Supplementary Figure S1e). When DBC1 protein levels were specifically reduced in 293T cells by small interfering RNA transfection (Supplementary Figure S1f), both the coactivator activity of β-catenin for LEF1 and the transcriptional activity of Gal4-β-catenin were repressed (Supplementary Figures S1g and h), suggesting that DBC1 is required for the coactivator function of and transcriptional activation by β-catenin.

Figure 1.

Coactivator function of DBC1 in LEF1–β-catenin-mediated transcription and identification of DBC1 target genes by microarray analysis. (a) Endogenous interaction between DBC1 and β-catenin. Whole-cell lysates of SW480, SW620 and HT-29 were immunoprecipitated with normal IgG or anti-DBC1 antibody. The immunoprecipitates were analyzed by immunoblot with the indicated antibodies. (b) In vitro-translated HA-tagged DBC1 was incubated with glutathione S-transferase (GST) fusion proteins as indicated. Bound proteins were analyzed by immunoblot with an anti-HA antibody. (c) CV1 cells were transfected with pGL3OT reporter and expression vectors as indicated and harvested for luciferase assays. Data are means ± s.d. (n = 3). (d) DBC1 protein levels in WT and DBC1 KO SW480 cells were monitored by immunoblot. (e) Pie graph shows that 1433 genes are differentially expressed in WT versus DBC1 KO SW480 cells. (f) GSEA profile showing that the Wnt/β-catenin gene set is highly enriched among genes downregulated by DBC1 KO in SW480 cells. (g) Validation of DBC1-regulated β-catenin target genes. Total RNAs were examined by qRT–PCR analysis with primers specific for the indicated mRNAs. Data are means ± s.d. (n = 3).

DBC1 is required for efficient expression of Wnt/β-catenin target genes

To assess the global effect of DBC1 on the expression of Wnt/β-catenin-responsive genes in an unbiased manner and identify DBC1 target genes in colon cancer cells, we performed genome-wide gene expression analysis in control (WT) and DBC1 knockout (KO) SW480 cells. The DBC1/CCAR2 gene was knocked out using transcription activator-like effector nucleases (TALENs) technology (Supplementary Figure S2), and TALEN-mediated gene targeting eliminated DBC1 expression in SW480 cells (Figure 1d). Microarray analysis revealed that 1433 genes were differentially expressed (> 1.5 fold change, P < 0.05) between WT and DBC1 KO cells, with 961 downregulated and 472 upregulated genes (Figure 1e). Gene set enrichment analysis (GSEA) using the Wnt/β-catenin target gene signature revealed a significant reduction in the expression of Wnt/β-catenin target genes in DBC1 KO cells (Figure 1f and Supplementary Figures S3a and b), suggesting a positive role of DBC1 in Wnt/β-catenin-mediated transcription. Fifteen well-characterized β-catenin target genes among the 91 differentially expressed β-catenin target genes were selected for validation by quantitative reverse transcriptase-PCR (qRT–PCR). The validated data were highly consistent with the microarray data (Figure 1g and Supplementary Figure S4a), and similar results were also observed in DBC1-depleted SW480 and SW620 cells (Supplementary Figures S4b and c) indicating that DBC1 is required for the expression of a subset of Wnt/β-catenin target genes in colon cancer cells.

DBC1 is required for LEF1–β-catenin transcription complex assembly on Wnt-responsive enhancers (WREs)

To investigate the direct involvement of DBC1 in LEF1–β-catenin-mediated transcription, we performed chromatin immunoprecipitation (ChIP) assays in SW480 cells. LEF1, β-catenin, p300 and RNA polymerase II (Pol II) were recruited to the well-characterized WREs of LEF1–β-catenin target genes, PROX1, MMP9, MYCN, DKK1 and FGF20 (Figure 2a and Supplementary Figure S5a), but not to a region lacking WRE. DBC1 was also specifically recruited to the WREs, indicating that DBC1 is directly involved in the transcriptional regulation of endogenous LEF1–β-catenin target genes. In addition, reciprocal Re-ChIP experiments demonstrated that β-catenin and DBC1 exist in the same transcription complex on the WREs of PROX1, MMP9 and MYCN genes (Figure 2b and Supplementary Figure S5b).

Figure 2.

DBC1 is required for optimal association of LEF1 and β-catenin with target enhancers and chromatin looping between the PROX1 enhancer and promoter. (a) ChIP assays. Cross-linked, sheared chromatin from SW480 cells was immunoprecipitated with the indicated antibodies. qPCR analyses were performed using primers specific for the indicated regions of the PROX1, MMP9 and MYCN genes. A 28-kb downstream region of PROX1 gene was used as a negative control region. The results are shown as the percentage of input and are means ± s.d. (n = 3). (b) Reciprocal Re-ChIP analysis. Soluble chromatin was immunoprecipitated (1° ChIP) with antibodies to β-catenin and DBC1, respectively. Immunocomplexes were eluted and reimmunoprecipitated (2° ChIP) with reciprocal antibodies. qPCR analyses were performed as described in panel a. Asterisks indicate statistically significant differences (*P < 0.05 and **P < 0.01; Student’s t-test). (c, d) Protein levels in WT and DBC1 KO SW480 cells were monitored by immunoblot using the indicated antibodies (c). ChIP assays using the indicated antibodies were performed as described in panel a (d). (e) 3C assays were performed using cross-linked, BstYI-digested chromatin from SW480 cells. Primers flanking − 40-kb WRE and − 500-bp promoter regions were used to amplify 3C DNA after ligation. (f) 3C assays were performed using chromatin from WT and DBC1 KO SW480 cells as described in panel e.

Recently, we have reported that DBC1 has an important role in transcription complex assembly and recruitment of Pol II to target enhancers.^12,13 To examine the role of DBC1 in LEF1–β-catenin transcription complex assembly on the WREs, we repeated ChIP assays in SW480 WT and DBC1 KO cells. DBC1 KO had no measurable effect on the cellular levels of LEF1, β-catenin, p300 and Pol II (Figure 2c). However, loss of DBC1 almost completely eliminated occupancy of LEF1, β-catenin and DBC1 on the WREs of PROX1, MMP9, MYCN and DKK1 genes, and the recruitment of p300 and Pol II and histone H3 acetylation levels were also greatly reduced by loss of DBC1 (Figure 2d and Supplementary Figure S5c). These results suggest that DBC1 is required for efficient binding of LEF1 to its regulatory regions and/or for stable association of LEF1 with β-catenin, which strongly enhances the binding of LEF1 to chromatin,¹⁸ and thereby facilitates the occupancy of p300 and Pol II on the regulatory regions of LEF1–β-catenin target genes.

DBC1 is required for spatial communication between the PROX1 WRE and promoter

Based on our initial results, we focused further on a Wnt-inducible transcription factor PROX1, which has been shown to have a critical role in colon cancer progression and function as a stem cell regulator in colon cancer.^8–10 The PROX1 WRE is located approximately 40-kb upstream of the PROX1 promoter (Figure 2e),⁸ but the mechanism by which these two elements communicate is not known. Given our finding that Pol II is associated with the PROX1 WRE, we hypothesized that the WRE may be in close proximity to the PROX1 promoter. To this end, we performed chromosome conformation capture (3C) analysis, a technique used to detect long-range chromatin interactions. The cross-linked chromatin from SW480, SW620 or HT-29 cells was digested with BstYI and ligated. After reverse cross-linking, PCR was performed with one primer in the PROX1 enhancer and another in the PROX1 promoter (primer E+/P−) using equal amounts of 3C DNA. As shown in Figure 2e and Supplementary Figure S6a, the specific PCR product is formed in a ligase-dependent manner, and control PCR (primer E+/E−) for the level of input chromatin was equivalent under all conditions. The 448-bp PCR-amplified product was sequenced to confirm the interaction between the PROX1 WRE and promoter (Supplementary Figure S7). These results suggest that − 40-kb WRE specifically interacts with the PROX1 promoter region by forming a chromatin loop. As DBC1 is required for LEF1–β-catenin complex assembly on the PROX1 WRE, we next asked whether DBC1 can affect the PROX1 WRE-promoter loop formation. Loss or depletion of DBC1 severely reduced the interaction between the PROX1 WRE and promoter (Figure 2f and Supplementary Figure S6b), suggesting the involvement of DBC1 in bridging between the WRE and promoter of the PROX1 gene.

DBC1 inhibits SIRT1-mediated deacetylation of β-catenin and repression of β-catenin activity

β-catenin is acetylated by p300/CBP and deacetylated by SIRT1.^4,19 Acetylation of β-catenin enhances its interaction with LEF1 and its transactivation activity, and conversely, deacetylation by SIRT1 represses β-catenin transcriptional activity. Recent studies reported that DBC1 binds to SIRT1 and inhibits its deacetylase activity.^12–14 To further investigate the mechanism by which DBC1 promotes β-catenin-dependent transcription, we first determined the acetylation levels of endogenous β-catenin in WT and DBC1 KO SW480 cells. As shown in Figure 3a, acetylation levels of endogenous β-catenin were markedly reduced in DBC1 KO SW480 cells. We next examined the effect of DBC1 expression on β-catenin deacetylation by SIRT1 in vivo. 293T cells were transfected with plasmids expressing p300, SIRT1 and DBC1, and the acetylated levels of endogenous β-catenin were determined. As reported previously,^4,19 β-catenin was acetylated and deacetylated by p300 and SIRT1, respectively, and importantly, the deacetylation of β-catenin by SIRT1 was reversed by DBC1 expression (Figure 3b). To confirm these results in vitro, we generated K49 or K345 (two major acetylation sites) acetylated β-catenin recombinant proteins using a recently developed method that allows the site-specific acetylation of recombinant proteins in bacteria²⁰ (Supplementary Figure S8a). As reported previously,⁴ acetylation of β-catenin increased its binding affinity for LEF1 (Supplementary information, Figure S8b). In in vitro deacetylation assays using purified recombinant SIRT1, DBC1 and acetylated β-catenin proteins, DBC1 inhibited the deacetylase activity of SIRT1 for β-catenin (Figure 3c). Furthermore, the interaction of β-catenin with SIRT1, but not with p300, was decreased when DBC1 was coexpressed in coimmunoprecipitation experiments (Figure 3d and Supplementary Figure S8c), and similar results were observed in in vitro competitive binding assays (Supplementary Figure S8d). In line with these results, endogenous interaction between β-catenin and SIRT1 was increased in DBC1 KO SW480 cells (Figure 3e). In reporter gene assays, SIRT1 repressed the transcriptional activity of β-catenin (Supplementary Figure S9), and DBC1 blocked the inhibitory effect of SIRT1 on the coactivator function of β-catenin for LEF1 and transcriptional activity of β-catenin stimulated by p300 (Figures 3f and g). Consistent with these results, DBC1 increased the interaction of β-catenin with LEF1 in vivo and in vitro (Figures 3h and i). Furthermore, endogenous interaction between β-catenin and LEF1 was decreased in DBC1 KO SW480 cells (Figure 3j), suggesting that DBC1 stabilizes the interaction between β-catenin and LEF1 by interacting with β-catenin. Together, these results suggest that DBC1 functions as a positive regulator of β-catenin by inhibiting SIRT1-mediated deacetylation and repression of β-catenin through blocking the interaction between β-catenin and SIRT1, thereby enhancing the association of β-catenin with LEF1 and promoting β-catenin-mediated transcription (Figure 3k).

Figure 3.

DBC1 inhibits SIRT1-mediated deacetylation and repression of β-catenin. (a) Acetylation (Ac) levels of endogenous β-catenin in WT and DBC1 KO SW480 cells were determined by immunoblot using anti-Ac-β-catenin antibody (Ac-K49). (b) 293T cells were transfected with expression vectors as indicated. Cell extracts were immunoprecipitated with anti-β-catenin antibody, and the levels of Ac-β-catenin were determined by immunoblot using anti-Ac-β-catenin antibody (Ac-K49). Input and immunoprecipitated proteins were analyzed by immunoblot with the indicated antibodies. (c) Site specifically acetylated β-catenin proteins were incubated with glutathione S-transferase (GST)-SIRT1 and GST-DBC1 in the presence of NAD as indicated. Immunoblots were performed as described in panel b. (d) 293T cells were transfected as indicated, and FLAG-SIRT1 immunoprecipitations were analyzed by immunoblot with the indicated antibodies. (e) Endogenous interaction between β-catenin and SIRT1 in SW480 WT and DBC1 KO cells. Whole-cell lysates were immunoprecipitated with anti-SIRT1 antibody. The immunoprecipitates were analyzed by immunoblot with the indicated antibodies. (f, g) DBC1 reverses SIRT1-mediated repression of β-catenin. CV1 cells were transfected with pGL3OT (f) or pG5-LUC (g) reporter and expression vectors as indicated and harvested for luciferase assays. Data are means ± s.d. (n = 3). (h) 293T cells were transfected as indicated, and His-β-catenin immunoprecipitations were analyzed by immunoblot with the indicated antibodies. (i) In vitro-translated LEF1 and DBC1 were incubated with His-β-catenin bound to beads as indicated. Bound proteins were analyzed by immunoblot with indicated antibodies. (j) Endogenous interaction between β-catenin and LEF1 in SW480 WT and DBC1 KO cells. Whole-cell lysates were immunoprecipitated with anti-LEF1 antibody. The immunoprecipitates were analyzed by immunoblot with the indicated antibodies. (k) The role of DBC1 as a β-catenin coactivator. DBC1 inhibits SIRT1-mediated β-catenin deacetylation, thereby increasing the interaction of β-catenin with LEF1 and their transcriptional activity.

DBC1 functions as a PROX1 coactivator

Given previous studies showing that PROX1 contributes to metastasis by regulating the transcriptional network in CRC^8–10 and our findings that DBC1 is an essential coactivator for PROX1 expression, we investigated the global effect of loss of DBC1 on the expression of PROX1 target genes. In line with the downregulation of PROX1 expression upon loss of DBC1 (Figure 1g and Supplementary Figure S10a), GSEA using the PROX1 gene signature revealed significant deregulation of PROX1 target gene expression in WT versus DBC1 KO SW480 cells (Figure 4a and Supplementary Figures S3c and d). Expression changes of 14 selected genes were validated by qRT–PCR (Figure 4b and Supplementary Figure S10b). Strikingly, restoration of PROX1 expression in DBC1 KO SW480 cells via Tet-inducible system failed to rescue the expression of PROX1 target genes (Figure 4c), suggesting a requirement of DBC1 for PROX1-mediated transcription. Indeed, coexpression of DBC1 showed rescue effect on PROX1 target gene expression (Figure 4c). We therefore investigated whether DBC1 also can function as a coactivator of PROX1. In coimmunoprecipitation experiments, endogenous DBC1 was bound to PROX1 in SW620 cells (Figure 4d). Similar results were observed in coimmunoprecipitation experiments with ectopically expressed proteins and glutathione S-transferase pull-down assays using recombinant proteins (Supplementary Figures S11a and b), suggesting that DBC1 interacts directly with PROX1. In reporter gene assays using reporters containing six copies of the PROX1-binding sites, DBC1 increased reporter activity in a PROX1-dependent manner in PROX1-negative HCT-116 cells (Figure 4e), and loss of DBC1 decreased reporter activity in SW480 cells (Supplementary Figure S11c). In addition, ChIP and Re-ChIP assays showed that DBC1 is recruited to and coexists with PROX1 on the PROX1-binding sites associated with PROX1 target genes (Figures 4f and g). Collectively, these results show that DBC1 is not just required for PROX1 expression but also for PROX1-mediated transcription as a coregulator for PROX1.

Figure 4.

DBC1 functions as a coregulator for PROX1. (a) GSEA profile showing that the PROX1 target gene signature is highly enriched among genes differentially regulated by DBC1 KO in SW480 cells. (b) Validation of DBC1-regulated PROX1 target genes. Total RNAs were examined by qRT–PCR analysis with primers specific for the indicated mRNAs. (c) SW480 DBC1 KO/rtTA-PROX1 (Tet-P1) and DBC1 KO/rtTA-PROX1-DBC1 (Tet-P1&D1) cells were treated with doxycycline (Dox) for 48 h, as indicated. Protein levels were monitored by immunoblot using the indicated antibodies (left). Total RNA was examined by real-time qRT–PCR analysis with primers specific for the indicated mRNAs (right). (d) Whole-cell lysates of SW620 cells were immunoprecipitated with normal IgG or anti-PROX1 antibody. The immunoprecipitates were analyzed by immunoblot with the indicated antibodies. (e) HCT-116 cells were transfected with 6xPROX1-binding site (PBS)-LUC reporter and expression vectors as indicated and harvested for luciferase assays. Data are means ± s.d. (n = 3). (f, g) ChIP (f) and Re-ChIP (g) assays using the indicated antibodies were performed as described in Figures 2a and b. qPCR analyses were performed using primers specific for the indicated regions of the LUM and DPP4 genes. The results are shown as the percentage of input and are means ± s.d. (n = 3).

DBC1 is required for tumorigenic potential of colon cancer cells, and its overexpression correlates with poor outcome in advanced-stage colon cancer

As upregulation of the Wnt/β-catenin–PROX1 signaling axis is an important factor in colon cancer growth and progression,⁸ we examined the effect of DBC1 KO or depletion on tumorigenic properties of colon cancer cells. DBC1 KO attenuated SW480 cell proliferation, and, similarly, DBC1 depletion by short hairpin RNA decreased cell proliferation in SW480, SW620 and HT-29 cells (Figure 5a). In addition, DBC1 KO inhibited the clonogenic survival, migration and invasion of SW480 cells (Figures 5b–d), suggesting that DBC1 is required for the tumorigenic and metastatic properties of colon cancer cells. Colon cancer cells form spherical colonies (colonospheres) when cultured on non-adherent surfaces, and these colonospheres are enriched in cells with functional characteristics of cancer stem cells.²¹ β-catenin and PROX1 have critical roles in growth and maintenance of colonospheres.^9,10,21 In colonosphere formation assays using SW620 cells, colonosphere size and number were greatly reduced by DBC1 depletion (Figure 5e), indicating that DBC1 is also required for the self-renewal activity needed to form colonospheres. In agreement, loss of DBC1 reduced cancer stem cell marker expression (LGR5, CD24, DPP4/CD26, PROM1/CD133 and ALCAM/CD166) in colon cancer cells²² (Supplementary Figures S3 and S4). To further examine the role of DBC1 in promoting colon tumorigenesis, we assessed the effect of DBC1 KO on the growth of SW480 xenograft tumors in nude mice injected with WT or DBC1 KO SW480 cells expressing luciferase (SW480-LUC). The expression and activity of luciferase in SW480-LUC cells were not affected by DBC1 KO (Supplementary Figures S12a and b). Although SW480 cells grew fast with a rapid increase in tumor volume, DBC1 KO SW480 cells grew slowly with a significant reduction in tumor volume (Figures 5f and g). Similar results were also obtained in DBC1-depleted SW480 cells (Supplementary Figures S12c–f), suggesting that DBC1 has a critical role in tumorigenic growth of colon cancer cells.

Figure 5.

DBC1 is required for the tumorigenic properties of colon cancer cells. (a) Cell proliferation analysis of DBC1 KO or -depleted colon cancer cells. Data are means ± s.d. (n = 6). *P < 0.01. (b, c, d) Quantitative analysis of colony formation (b), migration (c) and invasion (d) assays using SW480 WT and DBC1 KO cells. For the colony formation assay, viable colonies were stained with crystal violet; the dye was extracted and quantified by spectrophotometry. Migration and invasion assays were performed using Transwell chambers coated without or with Matrigel, respectively. Data are means ± s.d. (n = 3). *P < 0.05. (e) Sphere formation analysis of SW620 cells infected with lentiviruses expressing shNS or shDBC1. *P < 0.01. (f, g) Mouse xenograft tumors of SW480-LUC (DBC1 WT or KO) cells. Representative bioluminescence images of tumor-bearing mice and their tumors are shown (left), and the average signal intensity (n = 6, ± s.d.) of regions of interest is shown (right) (f). Tumor growth curves are shown (g). *P < 0.01. (h) Kaplan–Meier analysis of relapse-free survival (RFS) of advanced-stage CRC patients with low or high DBC1 expression (left) and with different DBC1 and β-catenin expression levels (right). *P < 0.001, ^†P = 0.032 and ^‡P = 0.064. (i) The proposed dual role of DBC1 in CRC promoting β-catenin–PROX1 signaling axis. DBC1 promotes PROX1 expression by acting as a coactivator of β-catenin and also contributes to PROX1-mediated transcription by functioning as a coregulator for PROX1 in colon cancer cells.

Finally, to investigate the clinical significance and prognostic value of DBC1 expression in CRC, we analyzed DBC1 protein expression by immunohistochemistry in tissue microarrays containing surgical specimens from 213 CRC patients. DBC1 was mainly stained in the nucleus, and tumors were divided into two groups according to DBC1 expression (Supplementary Figure S13a). When patients were divided into two TNM (tumor-node-metastasis) stage groups, early (I/II) and advanced (III/IV) stages, a significant positive correlation was observed between high levels of DBC1 protein and worse relapse-free survival in advanced-stage patients (Figure 5h). However, there is no significant association between survival and DBC1 levels in early-stage patients (Supplementary Figure S13b). Importantly, nuclear β-catenin/DBC1 double-positive patients had worse relapse-free survival compared with double-negative patients (Figure 5h). In addition, there was a trend that, among nuclear β-catenin-positive CRC, DBC1-positive staining correlated with worse prognosis. Together, our results suggest that increased expression of DBC1 is associated with tumor progression and poor prognosis in advanced-stage colon cancer.

DISCUSSION

Uncontrolled Wnt/β-catenin signaling has been linked to various cancers, and accumulating studies strongly suggest that targeting of Wnt/β-catenin signaling is a promising strategy for cancer therapy.²³ In this regard, unraveling of the molecular mechanisms involved in the aberrant activation of β-catenin and its target genes has been of great interest. Here, we extend the current understanding of Wnt/β-catenin signaling by identifying DBC1 as a key regulator of β-catenin–PROX1 signaling axis in colon cancer progression. DBC1 cooperated synergistically with β-catenin to enhance LEF1-mediated transcription (Figure 1c). Loss of DBC1 caused reduction in the expression of β-catenin target genes (Figure 1f and Supplementary Figure S3), markedly reduced recruitment and assembly of LEF1–β-catenin complex at the WREs associated with LEF1–β-catenin target genes (Figure 2d), and attenuated metastatic potential and in vitro and in vivo growth of colon cancer cells (Figure 5). Furthermore, DBC1 expression was significantly associated with shorter relapse-free survival of colon cancer patients (Figure 5h). These results firmly establish DBC1 as a key factor in β-catenin-mediated colon cancer progression.

Acetylation of β-catenin by p300 increases its coactivator function for LEF1 and is reversed by SIRT1.^4,19 A recent study demonstrated that DBC1 is required for β-catenin-mediated transcription.²⁴ However, the detailed mechanism of how DBC1 contributes to β-catenin-mediated transcription has not been fully elucidated. Here we showed several lines of mechanistic evidence that DBC1 promotes β-catenin activity by negatively regulating SIRT1 activity (Figure 3): SIRT1-mediated deacetylation of β-catenin was inhibited by DBC1; DBC1 blocks the interaction of SIRT1 with β-catenin; and SIRT1-mediated repression of β-catenin activity was reversed by DBC1 expression. Consistent with the model in which β-catenin acetylation enhances its binding to LEF1 and leads to increased chromatin occupancy of LEF1 by stabilizing LEF1–β-catenin complex,^4,18 our results further showed that DBC1 enhances the interaction of β-catenin with LEF1 (Figures 3h and i) and is required for efficient recruitment of LEF1 and β-catenin to their target chromatin regions (Figure 2d), probably through protecting β-catenin from SIRT1-mediated deacetylation (Figure 3k).

Among β-catenin target genes, PROX1 has emerged as an important transcription factor promoting colon cancer progression from an early to advanced stage.⁸ Increased PROX1 expression by activated Wnt/β-catenin is associated with self-renewal of cancer stem cells, cancer progression and poor prognosis in colon cancer.^9–11 Here we showed that DBC1 contributes to upregulation of PROX1 expression by facilitating chromatin looping and LEF1–β-catenin transcription complex assembly on the PROX1 WRE (Figures 2d and f). In addition to its role in promoting PROX1 expression via cooperation with β-catenin, DBC1 also contributes to PROX1 transcriptional function and stem cell marker expression (Figure 4 and Supplementary Figures S3 and S4). Thus, DBC1 exerts a dual function as a coactivator in β-catenin–PROX1 signaling axis by enhancing LEF1–β-catenin-mediated transcription and by activating the transcriptional activity of β-catenin-induced transcription factor PROX1 (Figure 5i). In addition, by global gene expression analysis, we found that DBC1 regulates positively or negatively the transcriptional efficiency of specific subsets of β-catenin and PROX1 target genes (Supplementary Figure S3). In most cases, their positive and negative target genes were downregulated and upregulated, respectively, by loss of DBC1, suggesting a critical coregulator (coactivator and corepressor) role for DBC1 in β-catenin and PROX1-mediated transcription.

The role of DBC1 in human cancer has been controversial because it has been reported to suppress or promote cancer cell growth in different studies.^12–15 A recent paper suggested that TP53 mutation status in cancer cells may be the cause of these conflicting results.²⁵ DBC1 stabilizes TP53 and functions as a tumor suppressor in WT TP53 expressing cells. However, DBC1 also stabilizes oncogenic TP53 mutants in cancer cells and increases cancer cell proliferation. TP53 mutations are associated with the conversion from colorectal adenoma to carcinoma.²⁶ Interestingly, PROX1 has been identified as a TP53-repressed gene, and its upregulation after loss of the TP53 tumor-suppressor function promotes CRC progression, rather than initiation, by promoting cell invasion.²⁷ We proposed that DBC1 expression in the context of coexisting activated Wnt/β-catenin signaling, TP53 mutations and PROX1 upregulation may be associated with colon cancer progression and poor prognosis. In support of this model, we found a positive correlation between nuclear β-catenin/DBC1 expression and prognosis in advanced colon cancer (Figure 5h), but not in early stage of the disease (Supplementary Figure S13b), suggesting that DBC1 can be used as a prognostic marker in patients with advanced colon cancer.

MATERIALS AND METHODS

Cell culture and transient transfection

SW480, SW620, HT-29, CV1 and 293T cells were cultured in Dulbecco’s modified Eagle’s medium with 10% fetal bovine serum. HCT-116 and DLD1 cells were cultured in RPMI-1640 with 10% fetal bovine serum. All cell lines used in this study were obtained from American Type Culture Collection (ATCC, Manassas, VA, USA) or from Korean Cell Line Bank (KCLB, Seoul, Korea), tested for mycoplasma contamination and authenticated by STR DNA profiling (Samsung Biomedical Research Institute, Seoul, Korea). Transient transfections and reporter gene assays were performed as described previously.^5–7 Each experiment was repeated independently at least three times.

TALEN-mediated KO of DBC1 gene in SW480 cells

To establish DBC1 KO SW480 cell lines, a pair of TALEN constructs (Hs-H70503) targeting DBC1 gene and a surrogate reporter pRGS2 were purchased from ToolGen (Seoul, Korea). SW480 cells were transfected with the TALEN constructs along with pRGS2 reporter for cell sorting. After 48 h transfection, single cells expressing both green fluorescent protein and red fluorescent protein were sorted using a BD Biosciences (San Jose, CA, USA) FACS Aria III, and clonal selection was carried out by limiting dilution in 96-well plates. Individual cell clones were screened for DBC1 expression by immunoblot analysis. The targeted exon was PCR-amplified from genomic DNA isolated from selected DBC1 null clones (three independent clones), and analyzed for frameshift mutations in the DBC1 gene by DNA sequencing. SW480 DBC1 KO #16 cell clone was chosen for further experiments.

Gene expression array analysis and GSEA

Total RNAs were isolated from SW480 WT and SW480 DBC1 KO cells using the RNeasy mini kit (Qiagen, Valencia, CA, USA) for gene expression array analysis. Gene expression profiling was performed with three biological replicates from three independent experiments following the Affymetrix (Santa Clara, CA, USA) standard protocol as described previously.^13,28 Using normalized log₂ intensities, we identified DBC1-dependent genes by comparing the differential expression (cut-off values41.5 fold change and P < 0.05) between SW480 WT and SW480 DBC1 KO samples. This DBC1 target gene signature consists of 961 unique DBC1 KO downregulated genes (that is, positively regulated by DBC1) and 472 unique DBC1 KO upregulated genes (that is, negatively regulated by DBC1). The data have been deposited in the Gene Expression Omnibus (GEO) database under accession number GSE65841. GSEA was performed using GSEA software v2.0.14 (http://www.broadinstitute.org/gsea/) with the DBC1 target gene signature (1433 genes) and Wnt/β-catenin target gene set (537 genes) obtained from the Wnt/β-catenin database (www.stanford.edu/~rnusse/wntwindow.html) or PROX1 target gene sets (449 genes) obtained from the studies of Ragusa et al.⁹ and Petrova et al.⁸

RNA interference and real-time qRT–PCR

The depletion of DBC1 by small interfering RNAs or short hairpin RNAs was performed according to previously described protocol.^12,13 qRT–PCR was performed with total RNA and Brilliant SYBR Green QRT–PCR Master Mix 1-Step (Stratagene, Santa Clara, CA, USA). Data were normalized to β-actin or glyceraldehyde 3-phosphate dehydrogenase mRNA levels. The primers used are listed in Supplementary Information.

ChIP assays

ChIP and Re-ChIP experiments were performed according to the procedure described previously.^12,13,29,30 The immunoprecipitated DNAs were amplified by qPCR. The primers used are listed in Supplementary Information.

3C assays

3C assays were performed according to the procedure described previously.²⁸ Briefly, cross-linked chromatin was digested overnight with BstYI, and the digested chromatin was diluted 50-fold in ligase buffer and ligated overnight with T4 DNA ligase. After reverse cross-linking, the purified 3C DNA was amplified by PCR. The primers used are listed in Supplementary Information.

Expression and purification of site specifically acetylated β-catenin

Site specifically acetylated β-catenin was generated using a strategy described previously.^13,20 Briefly, BL21(DE3) cells transformed with pAcKRS-3 and pCDF PylT-1-β-catenin with amber codons at K49 or K345 were grown in Luria broth supplemented with 50 µg/ml kanamycin and 50 µg/ml spectinomycin. Before induction with 0.5 mm isopropyl-β-thiogalactopyranoside, cells were supplemented with 20 mm nicotinamide and 10 mm acetyl-lysine. Purification of acetylated proteins was carried out as described previously.¹³

Deacetylation assays

Deacetylation assays were performed as described previously.^12,13 Briefly, 293T cells were transfected with expression plasmids for p300, SIRT1 and DBC1 as indicated in the figure legends. After 36 h of transfection, the cells were treated with 0.5 µM trichostatin A (Sigma, St Louis, MO, USA) for 12 h and then lysed in FLAG lysis buffer supplemented with 0.5 µM trichostatin A and protease inhibitor cocktail (Roche, Indianapolis, IN, USA). Endogenous β-catenin was immunoprecipitated with anti-β-catenin antibody, and acetylation levels were determined by immunoblot with anti-acetyl β-catenin (Lys49) or anti-acetyl-lysine antibody. For in vitro deacetylation assays, site specifically acetylated β-catenin was incubated with purified recombinant glutathione S-transferase-SIRT1 and glutathione S-transferase-DBC1 and 1 mm NAD+ in SIRT1 reaction buffer for 3 h at 30 °C, as described previously.^12,13

Generation of Tet-inducible stable cell lines

SW480 cells inducibly expressing PROX1 were generated using Lenti-X Tet-On Advanced Inducible Expression System (Clontech, Moutain View, CA, USA). Briefly, SW480 DBC1 KO cells were infected with lentiviruses expressing reverse-Tet-controlled transactivator (rtTA-Advanced). Selected G418-resistant clones (SW480 DBC1 KO/rtTA) were analyzed for doxycycline inducibility using pLVX-Tight-Puro-LUC. SW480 DBC1 KO/rtTA cells were then transduced with lentiviral particles packaged with pLVX-Tight-Puro-PROX1 and pLVX-Tight-Puro-DBC1. Selected G418/puromycin-resistant cells (SW480 DBC1 KO/rtTA-PROX1 or SW480 DBC1 KO/rtTA-PROX1-DBC1) were analyzed for DBC1 or PROX1 expression by immunoblot and qRT–PCR with and without doxycycline induction.

Xenograft experiments

Mouse xenograft experiments were performed as described previously.^13,28,29 The detailed method is provided in Supplementary Information. Animal experiments were conducted with the approval of the Institutional Animal Care and Use Committee of Laboratory Animal Research Center at Samsung Biomedical Research Institute.

Tissue microarray and immunohistochemistry

Two hundred thirteen tumor tissues from CRC patients who underwent surgical resection at the Samsung Medical Center, Cancer Hospital from January 2000 to December 2000 were selected for this study. Tissue microarray construction and immunohistochemical staining and scoring of DBC1 were performed as described previously.^13,17,31 The detailed methods are provided in Supplementary Information. This study was approved by institutional review board of Samsung Medical Center and conducted in accordance with the 1996 Declaration of Helsinki.

A full description of the methods and primers used in this study can be found in Supplementary Information.