RNF6 Promotes Colorectal Cancer by Activating the Wnt/β-Catenin Pathway via Ubiquitination of TLE3 Free

Colorectal cancer is the third most common cancer and a major cause of cancer mortality worldwide (1). Both the incidence and death rate of colorectal cancer are increasing rapidly and maintaining an upward trend in Asian countries (2). The genetic mechanisms in colorectal cancer initiation and progression are complex and heterogeneous, involving somatic gene mutations, abnormal gene fusion, deletion or amplification, and epigenetic alterations (3–5). Chromosomal instability (CIN), including somatic DNA copy number amplification, is a hallmark of cancer and is present in approximately 60% to 70% of sporadic colorectal cancer, and the CIN pathway is usually associated with a stepwise process of “adenoma to carcinoma sequence” (6–8). The somatic copy number alterations (SCNA)–high subtype, which represents nearly 60% of primary colorectal cancers, demonstrated significant overlap (62%) with Wnt and MYC activation (7). The Wnt/β-catenin signaling pathway plays a pivotal role in regulation of cell proliferation and cell migration in the intestinal tract and colorectal tumorigenesis (9). Despite the high prevalence of DNA copy number amplification, their role in the pathogenesis of colorectal cancer remains poorly understood. Gene amplification of oncogenes, leading to their overexpression, is one of the potential mechanisms in colorectal cancer (10). Thus, identification of novel driver genes aberrantly upregulated by copy number amplification may uncover oncogenic pathways underlying the initiation and progression of colorectal cancer and discover potential biomarkers for diagnosis, prognosis, and treatment of patients with colorectal cancer.

We recently performed whole-genome sequencing in paired colorectal cancer tissues and identified a novel amplification gene, Ring finger protein 6 (RNF6). RNF6 is located on chromosome 13q12.13, one of the most frequently amplified regions in colorectal cancer (11, 12). RNF6 is a RING-type E3 ubiquitin-protein ligase that belongs to the RNF family, which mediates the ubiquitination of its target proteins and tags them for proteasomal degradation (13, 14). Several RNF family members are involved in the regulation of cell proliferation and differentiation, and they function as oncogenes or tumor suppressors in cancer, depending on their target proteins and the cellular context (15–17). Some RNF proteins are potential prognostic biomarkers and novel therapeutic targets for cancer treatment (18–20). However, the molecular function of RNF6, its target protein substrates, its involvement in signal transduction in cancer, as well as its clinical significance in patients with colorectal cancer are still elusive.

In this study, we uncovered the oncogenic role of RNF6 on colorectal cancer cell functions and tumorigenicity. We examined the molecular mechanisms of RNF6 and revealed that RNF6 directly interacted with Groucho family member transducin-like enhancer of split 3 (TLE3), a critical transcriptional repressor of the β-catenin/TCF4 complex; RNF6 caused TLE3 ubiquitination and degradation, thereby promoting β-catenin recruitment to TCF4/LEF and activating the Wnt/β-catenin signaling cascade. We also unveiled the clinical significance of RNF6 in both the testing and validation cohorts of patients with colorectal cancer.

The colon cancer cell lines (DLD-1, HCT116, HT29, LOVO, SW480, SW620, and SW1116) were obtained from the U.S. ATCC between 2014 and 2015, and cell authentication was confirmed by short tandem repeat profiling. Cells were routinely cultured and maintained in DMEM supplemented with 10% FBS and antibiotics (Gibco BRL) according to the ATCC protocols, except HCT116 cells, which were cultured in complete McCoy's 5A medium (Gibco BRL). Routine Mycoplasma testing was performed by PCR. Cells were grown for no more than 25 passages in total for any experiment.

A total of 118 colorectal cancer tumor tissues (BeiJing cohort 1) and an additional 150 paired colorectal cancer tumor and adjacent nontumor tissue samples (BeiJing cohort 2) were obtained from Peking University Cancer Hospital, Beijing, China. Moreover, 50 paired primary colorectal cancer and adjacent normal colon tissues (HongKong cohort 1) and 160 primary colorectal cancer tumor tissues (HongKong cohort 2) were collected at Prince of Wales Hospital, The Chinese University of Hong Kong. Samples were confirmed histologically and staged according to the tumor–node–metastasis (TNM) staging system. Patients were followed up regularly, and the median follow-up duration since the time of diagnosis was 55.2 months (range 12.4 to 170 months). The clinicopathologic features of all the patients are provided in Supplementary Table S1. Colorectal cancer dataset was downloaded from The Cancer Genome Atlas (TCGA) colon and rectal cancer database (616 patients). The study was approved by the Clinical Research Ethics Committee of The Chinese University of Hong Kong; The Clinical Research Ethics Committee of Peking University Cancer Hospital and all subjects provided informed consent for obtaining the study specimens. This study was carried out in accordance with the Declaration of Helsinki of the World Medical Association.

Genomic DNA of colorectal cancer tissues was isolated using DNA Mini Kit (Qiagen) according to the manufacturer's protocol. RNF6-specific probe (Hs01093471_cn) was used to assess DNA copy number status in the colorectal cancer tissues. Real-time PCR was performed using TaqMan copy number master mixture (Thermo Fisher Scientific) on ViiA 7 real-time PCR instrument according to the protocol provided by company. Each sample was tested in triplicate. The results were analyzed in CopyCaller software to determine the copy number of the target gene.

For tumorigenicity assay, HCT116 cells (5 × 10⁶) stably expressing RNF6 or empty vector and DLD-1 cells stably transfected with shRNF6 and shControl were suspended in 0.1-mL PBS and were subcutaneously injected into both dorsal flanks of 4-week-old male athymic Balb/c nude mice (4/group). The tumor volume was measured by the Vernier caliper and calculated according to the formula tumor volume (mm³) = (longer diameter × shorter diameter²)/2. The mice were monitored at indicated time points for tumor growth and sacrificed when the tumors reached a maximal diameter of 1.0 cm. Tumors were excised and weighed. The excised tissues were rapidly frozen in liquid nitrogen until further protein extraction. Tumor sections from paraffin-embedded blocks of the harvested tissues were immunostained with the indicated antibodies. All animal experiments were performed in accordance with guidelines approved by the Animal Ethics Committee of The Chinese University of Hong Kong. All animal studies were approved by the government of the Hong Kong Special Administrative Region Department of Health.

RNF6- or vector-stable expression of HCT116 cells were injected at a final concentration of 5 × 10⁶ cells/100-μL PBS into the lateral tail veins of each 4-week-old male Balb/c nude mice (5/group). Mice were sacrificed after 28 days, and the lungs were dissected and fixed in 10% neutral-buffered formalin. Nude mice lung paraffin sections were prepared and analyzed by hematoxylin and eosin staining. Lung metastasis was quantified under microscopy. All animal experiments were performed in accordance with guidelines approved by the Animal Ethics Committee of The Chinese University of Hong Kong. All animal studies were approved by the government of the Hong Kong Special Administrative Region Department of Health.

SW480 cells or HCT116 cells were transfected with pcDNA3.1-RNF6–expressing vector and treated with 10 μmol/L MG132 (Cell Signaling Technology) for 10 hours; cell lysates were collected and immunoprecipitated with either 2 μg indicated or IgG antibody at 4°C overnight. The antibodies were then pulled down with 25 μL protein A/G magnetic beads (Thermo Fisher Scientific) for 4 hours at 4°C. After extensive washing, proteins were eluted with low-pH buffer and separated by SDS-PAGE gel, followed by silver staining. Candidate bands were subjected to mass spectrometry analysis for protein identification.

RNF6 stably transfected HT29 and HCT116 cells or RNF6-depleted SW480 cells and control cells were transfected with HA-ubiquitin plasmid. Two days after transfection, cells were treated with 10 μmol/L MG132 for 10 hours to block proteasomal degradation. Cell lysates were pulled down with either 2 μg HA-tag or IgG antibody. Eluates were separated by SDS-PAGE and immunoblotted with antibodies against TLE3, RNF6, and GAPDH, respectively.

Indicated colorectal cancer cells (1 × 10⁵ cells/well) were cotransfected with TOPflash luciferase reporter plasmid (200 ng/well), pRL-CMV vector (5 ng/well), and RNF6 plasmid or control empty vector for overexpression study, or shRNF6 and control shRNA for knockdown study using Lipofectamine 2000 transfection reagent (Life Technologies). TLE3 plasmid and siTLE3 and corresponding controls cotransfect with TOPflash luciferase reporter conduct in the rescue experiments. Cells were harvested at 48 hours posttransfection and luciferase activities were quantified by the Dual-Luciferase Reporter Assay System (Promega). The experiments were conducted three times in triplicates.

The results were expressed as mean ± SD. The Pearson correlation coefficient was used to evaluate the correlation between RNF6 amplification and its expression in the clinical samples. Crude relative risks (RR) of prognostic factors associated with RNF6 expression and other predictor variables were estimated by using a univariate Cox proportional hazards regression model and multivariate Cox model. Overall survival or recurrence rate in relation to RNF6 expression or copy number status was evaluated by the Kaplan–Meier survival analysis and the log-rank test. The RNF6 amplification rate between groups was tested byχ² test. A two-tailed, unpaired, or paired Student t test was performed to compare the variables of two groups, and ANOVA was used for multigroup comparisons. The difference in cell viability and tumor growth rate was determined by repeated measures ANOVA. P < 0.05 was regarded as statistically significant.

Other experimental methods are provided in the Supplementary Information. All primer sequences and the antibodies used and their dilutions are listed in the Supplementary Methods (Supplementary Tables S2 and S3).

To identify driver oncogenes deregulated by DNA copy number amplification in colorectal cancer, we performed whole-genome sequencing in paired colorectal cancer and adjacent normal samples, which revealed that RNF6 was one of the top outlier genes overexpressed in colorectal cancer, with a high copy ratio compared with baseline copy ratio (Fig. 1A). RNF6 mRNA expression was examined in two independent cohorts of paired colorectal cancer and adjacent normal tissues from Hong Kong (n = 50) and Beijing (n = 150) by real-time PCR. RNF6 mRNA expression was significantly higher in tumor tissues as compared with their adjacent normal tissues in both Hong Kong (37/50, 74%) and Beijing cohorts (110/150, 73.3%; 147/200, 73.5%, in total; Fig. 1B). Consistent with our data, RNF6 mRNA was overexpressed in the TCGA dataset consisting of 380 patients with colorectal cancer (Fig. 1B). In keeping with the increased RNF6 mRNA, RNF6 protein expression was significantly higher in colorectal cancer tumors as compared with adjacent normal tissues in 15 paired colorectal cancer samples by IHC (Fig. 1C).

A comparison of RNF6 gene expression in 30 cancer studies based on RNA sequencing results revealed that colorectal cancer is the top cancer associated with RNF6 upregulation (Supplementary Fig. S1A). Therefore, RNF6 can serve as a biomarker overexpressed in human colorectal cancer.

We first evaluated the clinicopathologic features and prognostic significance of RNF6 in 160 patients with colorectal cancer (HongKong cohort 1 and BeiJing cohort 1) with follow-up information. Kaplan–Meier analysis showed that colorectal cancer patients with high RNF6 expression had significantly poorer survival compared with those with low expression, especially in early stage (I/II) patients with colorectal cancer, but not in advanced stage (III/IV) patients (Fig. 1D). Colorectal cancer patients with high RNF6 expression also showed a higher recurrence rate compared with those with low expression (Fig. 1E). In univariate Cox regression analysis, high RNF6 expression was associated with poor outcomes of patients with colorectal cancer [HR, 1.801; 95% confidence interval (CI), 1.100–2.949; P = 0.019]. Multivariate Cox regression analysis adjusting for age, gender, and TNM stage demonstrated that RNF6 high expression is an independent prognostic factor for poor survival of patients with colorectal cancer (HR 1.646, 95% CI, 1.002–2.703; P = 0.046; Supplementary Table S4). The independent cohort included paired samples of adjacent normal colon and colorectal cancer from 150 patients (BeiJing cohort 2, two cases without survival data). Kaplan–Meier analysis showed that high expression of RNF6 was correlated with poor overall survival (Supplementary Fig. S1B). Furthermore, multivariate Cox regression analysis revealed RNF6 expression to be an independent prognostic factor for poor survival in patients with colorectal cancer (Supplementary Table S5). We analyzed RNF6 mRNA (HongKong cohort 1) and protein (HongKong cohort 2) expression according to tumor stage (Supplementary Fig. S1C). Both mRNA and protein expression of RNF6 was increased in late TNM stage (III/IV) compared with early TNM stage (I/II), suggesting that RNF6 levels increased during tumor progression. The correlation analysis showed that the protein expression of RNF6 by IHC was positively correlated with its mRNA expression by qPCR (Supplementary Fig. S1D). RNF6 protein expression was also evaluated in 160 patients with colorectal cancer (HongKong cohort 2) by IHC tissue microarray. Kaplan–Meier analysis revealed that high RNF6 expression in colorectal cancer was correlated with poor overall survival and greater recurrence risk (Fig. 1F). Both mRNA and protein expression of RNF6 indicated a significantly worse prognosis, suggesting that RNF6 could be regarded as a valuable new prognostic factor for colorectal cancer.

The RNF6 gene is located on chromosome 13q12.13, a recurrent amplification region in colorectal cancer (11, 12). RNF6 is rarely mutated (1%, 6/631 cases) in colorectal cancer, although analysis of the TCGA pan-cancer dataset revealed that the RNF6 gene is uniquely amplified in colorectal cancer, but not in other human cancers across 12 cancer types (Fig. 2A). Thus, we speculated that gene amplification may be a major mechanism through which RNF6 is activated in colorectal cancer. Notably, there is an increased RNF6 amplification rate in colorectal cancer patients with mutant adenomatous polyposis coli (APC) compared with those with wild-type APC (Fig. 2B). RNF6 amplification occurs in the early stages (I/II) of colorectal tumorigenesis and further increases during colorectal cancer progression (stage III/IV; Fig. 2B). In our BeiJing cohort 1, we observed RNF6 copy number amplification in 33.1% (39/118) of primary colorectal cancer tumors by genomic DNA quantitative PCR (Fig. 2C). RNF6 mRNA levels were significantly higher in colorectal cancer with copy number amplification, and RNF6 mRNA showed positive correlation with its DNA copy number amplification (R = 0.698; P < 0.0001; Fig. 2D). Analysis of the TCGA dataset found that RNF6 DNA copy number amplification is present in 66.3% (390/616) of primary colorectal cancer samples and confirmed a positive association between RNF6 copy number amplification and its mRNA expression (R = 0.842; P < 0.0001; Fig. 2E). Moreover, colorectal cancer patients with RNF6 copy number amplification had a higher recurrence risk compared with those without, especially in early-stage colorectal cancer patients (Fig. 2F). Multivariate Cox regression analysis showed that RNF6 copy number amplification was an independent risk factor associated with high recurrence risk in patients with colorectal cancer (Supplementary Table S6). These data collectively indicate that RNF6 copy number amplification mediates RNF6 overexpression in colorectal cancer and predicts a high risk of recurrence in patients with colorectal cancer.

As compared with normal colon tissues, RNF6 was overexpressed in five of seven colorectal cancer cell lines as determined by RT-PCR and Western blot analysis (Fig. 3A). The frequent overexpression of RNF6 in colorectal cancer tumors and cell lines prompted us to investigate its oncogenic role in colorectal cancer. We generated HCT116 and HT29 cells stably expressing RNF6 or control vector, whereas RNF6-targeting siRNA or shRNA and corresponding controls were used to knockdown RNF6 in DLD-1 and SW480 cells with high endogenous RNF6 expression (Supplementary Fig. S2A). Ectopic RNF6 expression in HCT116 and HT29 cells significantly promoted cell proliferation, as evidenced by colony formation and cell growth curve assays. Conversely, RNF6 knockdown by siRNA or shRNA in DLD-1 and SW480 cells suppressed cell proliferation (Fig. 3B and C). To assess the effect of RNF6 on tumorigenesis in vivo, we performed subcutaneous xenograft assay in nude mice using HCT116 cells stably expressing RNF6 or empty vector. RNF6 overexpression significantly increased tumor growth and tumor weight (Fig. 3D). Western blot analysis confirmed RNF6 overexpression in the xenograft tumors (Fig. 3E). Moreover, enhanced cell proliferation and reduced cell apoptosis were demonstrated in RNF6-overexpressing tumors using Ki67 and TUNEL staining, respectively (Fig. 3F). On the other hand, knockdown of RNF6 in DLD-1 cells by shRNA inhibited tumor growth, concomitant with reduced cell proliferation (Ki67), and increased cell apoptosis (TUNEL; Supplementary Fig. S2B). These results suggest that RNF6 plays an important oncogenic role in promoting colorectal cancer cell growth and tumorigenicity.

To investigate the mechanism by which RNF6 promoted cell growth, we examined the effect of RNF6 on cell-cycle progression by flow cytometry. Ectopic expression of RNF6 in HT29 and HCT116 cells decreased the G₀ to G₁ phase cell population, with a corresponding increase in S-phase population (Fig. 4A), whereas RNF6 knockdown in DLD-1 and SW480 cells induced G₀ to G₁ arrest (Supplementary Fig. S3A), indicating that RNF6 promotes G₁ to S transition. Consistently, Western blot analysis showed that RNF6 enhanced the expression of proliferating cell nuclear antigen (PCNA), the key factor in DNA replication and cell-cycle regulation (21), and cyclin D1 and cyclin-dependent kinase (CDK4), which promote G₁ phase progression, but simultaneously reduced the expression of G₁ gatekeepers, such as p53, p21^Cip1, and p27^Kip1 (Fig. 4B; refs. 22, 23). RNF6 knockdown in DLD-1 and SW480 cells had an opposite effect on the expression of these key cell-cycle regulators (Supplementary Fig. S3A).

We next determined the effect of RNF6 on the induction of apoptosis using Annexin V/7-aminoactinomycin D (7-AAD) staining and flow cytometry analysis. Ectopic RNF6 expression inhibited both early and late apoptosis in HT29 and HCT116 cells (Fig. 4C). On the contrary, RNF6 knockdown in DLD-1 and SW480 cells increased apoptosis induction (Supplementary Fig. S3B), consistent with an antiapoptotic function of RNF6. Suppression of apoptosis by RNF6 was further evidenced by the reduced protein expression of cleaved forms of caspase-8, caspase-9, caspase-3, caspase-7, and PARP in stably transfected HT29 and HCT116 cells as compared with controls (Fig. 4B), whereas RNF6 knockdown showed increased expression of these apoptosis factors (Supplementary Fig. S3B). Therefore, the oncogenic effect of RNF6 through suppression of cell apoptosis and promotion of cell-cycle progression in colorectal cancer cells.

We next investigated the effect of RNF6 on metastasis using cell migration and invasion assays in vitro and experimental lung metastasis assay in vivo. Wound healing and Matrigel invasion assays showed that both the migratory and invasive capabilities of HT29 and HCT116 cells were increased by RNF6 overexpression (Fig. 4D and E). In contrast, RNF6 knockdown markedly abolished cell migration and invasion in DLD-1 and SW480 cells (Supplementary Fig. S3C). In vivo tail vein metastasis assay with HCT116 cells stably expressing RNF6 or control vector demonstrated that RNF6 significantly enhanced the incidence and number of lung metastasis nodules (Fig. 4F).

In keeping with the effect of RNF6 on cell migration/invasion and metastasis, HT29 and HCT116 cells expressing RNF6 demonstrated features of epithelial-to-mesenchymal transition (EMT) including increased expression of N-cadherin and vimentin but decreased expression of E-cadherin and Claudin-1 (Fig. 4G; refs. 24, 25). On the other hand, RNF6 knockdown in DLD-1 and SW480 reversed the EMT phenotype (Supplementary Fig. S3C). We performed Western blot analysis to determine the expression of EMT-associated transcriptional factors. In HCT116 and HT29 cells, RNF6 overexpression induced protein expression of SNAI1, ZEB1, and ZEB2 (Fig. 4G).

We next sought to determine the molecular mechanisms underlying the oncogenic effect of RNF6. To probe RNF6-interacting partners, we performed coimmunoprecipitation with endogenous RNF6 antibody to pull down potential interacting proteins from SW480 cells, followed by mass spectrometry for protein identification (Fig. 5A). The transcriptional corepressor TLE3 was found to be a high scoring candidate that may interact with RNF6. Coimmunoprecipitation experiments in SW480 and HCT116 (with RNF6 overexpression) cells confirmed that RNF6 binds to TLE3, and reciprocal coimmunoprecipitation further ascertained the interaction between RNF6 and TLE3 using endogenous TLE3 and RNF6 antibodies (Fig. 5B).

In light of the interaction between RNF6 and TLE3, we tested whether RNF6 could alter TLE3 expression. RNF6 exhibited a predominantly nuclear localization (Fig. 5C). Nuclear TLE3 protein levels were significantly suppressed in RNF6-transfected HT29 and HCT116 cells, whereas depletion of RNF6 in DLD-1 and SW480 cells stabilized TLE3 levels in the nucleus (Fig. 5C). In either case, the mRNA level of TLE3 was not altered (Fig. 5D), indicating that RNF6 negatively regulates TLE3 at posttranscriptional level. Given that RNF6 is an E3 ubiquitin ligase, we hypothesized that RNF6 may function to promote TLE3 ubiquitination and degradation via the ubiquitin–proteasome system (UPS). In line with this possibility, TLE3 expression was restored after treatment with the proteasome inhibitor MG132, suggesting that RNF6-induced TLE3 degradation depends on the UPS (Fig. 5E). To confirm the role of RNF6 in UPS-mediated TLE3 degradation, we assessed TLE3 ubiquitin level. Consistent with our hypothesis, ectopic expression of RNF6 did indeed increase the ubiquitination of TLE3, whilst knockdown of RNF6 decreased the polyubiquitination of TLE3 (Fig. 5F). Taken together, these results indicate that RNF6 functions as an E3 ubiquitin ligase responsible for TLE3 degradation via the ubiquitin proteasome pathway in colorectal cancer cells.

It has been previously established that TLE family transcriptional repressors (TLE 1–3) binds to the TCF4/LEF complex with overlapping binding sites with β-catenin (26, 27). Given that RNF6 degrades TLE3, we next investigated how RNF6 might modulate the competition between TLE3 and β-catenin for binding to the TCF4/LEF complex. Congruently, dissociation of β-catenin from TCF4/LEF was found in RNF6-silenced cells with TLE3 enrichment (Fig. 5G). To determine the impact of RNF6 on the binding affinity of β-catenin to the TCF4/LEF complex, we pulled down TCF4 in control and RNF6-silenced SW480 cells. Knockdown of RNF6 significantly increased the binding between TLE3 and TCF4, but with a corresponding decrease in β-catenin/TCF4 interaction (Fig. 5H). RNF6-promoted degradation of TLE3 is thus required for recruitment of β-catenin to TCF4/LEF and stabilization of the β-catenin/TCF4/LEF complex. This connection was reinforced by the observation that nuclear β-catenin level was accumulated when RNF6 was overexpressed, as well as by the reduced nuclear β-catenin in RNF6 knockdown cells (Fig. 5C).

As RNF6-induced TLE3 degradation promotes binding of β-catenin to the TCF4/LEF complex, we speculated that RNF6 regulates Wnt/β-catenin signaling. Congruently, RNF6 accelerated TLE3 turnover when treated with the translational inhibitor cycloheximide (100 μg/mL), whereas β-catenin turnover was reduced, implying that RNF6 helps to stabilize β-catenin mainly by promoting TLE3 degradation (Fig. 6A). Hence, we determined Wnt/β-catenin pathway activity by luciferase reporter assays. As expected, RNF6 overexpression in HCT116 and HT29 cells enhanced activity of Wnt signaling, as demonstrated by TOPflash luciferase reporter assay, whereas knockdown of RNF6 in DLD-1 and SW480 cells dramatically abolished TOPflash activity (Fig. 6B). Consequently, we determined the expression of key downstream Wnt effectors associated with cell proliferation and EMT to illuminate the role of RNF6 in Wnt signaling activation. Consistent with the positive effect of RNF6 on Wnt signaling, Western blot analysis revealed that overexpression of RNF6 upregulated the levels of the Wnt effectors cyclin D1, C-myc, β-catenin, and active β-catenin (28, 29), whereas RNF6 knockdown diminished the expression of these key markers (Fig. 6C). To corroborate our in vitro data, we analyzed TCGA RNA sequencing results of colorectal cancer patients by gene-set enrichment analysis (GSEA). Among the 189 oncogenic signature gene sets, RNF6 was most strongly associated with Wnt/β-catenin, MYC, VEGF, and cell-cycle signaling pathways, whereas it was negatively correlated with apoptosis and cell adhesion signatures (Fig. 6D). RNF6 is therefore essential for Wnt signaling activation and its consequent cascades in colorectal cancer. Owing to its ability to recruit β-catenin to TCF4/LEF, RNF6 facilitates its transcriptional activity, as evidenced by increased the expression of β-catenin/TCF4 target genes, including c-Jun, Axin-2, matrix metalloproteinase-7 (MMP-7), CD44, and VEGF (Fig. 6E; refs. 30–34). Knockdown of RNF6 diminished the expression of β-catenin/TCF4 downstream targets that correlate with cell cycle and EMT process, which is in concordance with the results of the TOPflash assay. Thus, degradation of TLE3 by RNF6 is essential for activation and stabilization of β-catenin to facilitate downstream cell proliferation and EMT-related gene transcription to promote tumor growth and metastasis. In addition, mRNA expression of SNAI1/2, ZEB1, and TWIST1/2 was increased in HCT116 cells overexpressing RNF6, whereas silencing of RNF6 in SW480 decreased mRNA expression of SNAI1/2 and ZEB1 (TWIST1/2 was not detected; Fig. 6E).

Next, we asked whether RNF6 regulates Wnt/β-catenin signaling through directly modulating TLE3 levels. Restoration of TLE3 attenuated the upregulation of active β-catenin mediated by RNF6 (Fig. 6F). Likewise, knockdown of TLE3 abolished the active effect of RNF6 on the Wnt signaling pathway, as determined by TOPflash activity (Fig. 6G). As a further validation of our findings, silencing of TLE3 restored TOPflash activity caused by RNF6 depletion in SW480 cells (Fig. 6G). Notably, restoration of TLE3 abrogated the oncogenic effect of RNF6 in HCT116 cells, as determined by cell growth and colony formation assays. On the contrary, TLE3 knockdown rescued cell proliferation in SW480 cells with RNF6 loss (Fig. 6H). These results indicate that RNF6 exerts its oncogenic function via mediating TLE3 degradation. Collectively, our data demonstrated that RNF6 mediates its oncogenic effect by activating the Wnt signaling pathway, and this effect is partially dependent on TLE3 degradation.

In this study, we identified RNF6 as a novel oncogene that is frequently upregulated by gene amplification in primary colorectal cancer (73.5%) in two independent cohorts. RNF6 is located on chromosome 13q12.13, a genomic region with a higher frequency of amplification in colorectal cancer than in any other cancer type (7, 35). Noticeably, RNF6 amplification occurs in the early stage of colorectal cancer, and patients with early stage colorectal cancer diagnosed with RNF6 copy number amplification manifested a higher risk of recurrence. RNF6 amplification rate further increased in the late tumor stages, which may explain why RNF6 amplification is lower in our cohort (33.1% vs. 66.3%) compared with the TCGA colorectal cancer cohort, since our Beijing cohort 1 consisted of stage II and III colorectal cancer, without any stage IV patients. RNF6 amplification was positively associated with its mRNA expression in colorectal cancer from our cohort and the TCGA dataset, suggesting that gene amplification drives the overexpression RNF6, and RNF6 has a critical role in triggering colorectal cancer initiation and progression.

A series of in vitro and in vivo experiments established RNF6 as an oncogenic factor in colorectal cancer. RNF6 overexpression in colon cancer cells promoted cell proliferation in vitro and tumorigenicity in a xenograft model in mice. RNF6-overexpressing xenograft tumors had elevated cyclin D1 and active-β-catenin expression and a faster growth rate (Fig. 3E), supporting the relationships between RNF6 and Wnt activation in vivo. The oncogenic effect of RNF6 was mediated through accelerating cell-cycle progression and inhibition of apoptosis, involving the inhibition of both intrinsic and extrinsic apoptosis pathways (36). Tumor metastases are the cause of 90% of cancer-related deaths (37). Here, we provided first evidence that RNF6 promoted cell migration and invasion in colorectal cancer cells in vitro and lung metastasis in mice. RNF6 exerted its pro-metastatic effect by promoting EMT, as indicated by loss of the epithelial makers E-cadherin and Claudin-1, and increased expression of the mesenchymal markers N-cadherin, vimentin, SNAI1/2, TWIST1/2, and ZEB1/2, thereby favoring a mesenchymal phenotype that enables increased cell migration and invasion (24, 25, 38, 39). These findings are further verified in TCGA colorectal cancer transcriptome sequencing results, that is, RNF6 positively correlated with cell-cycle signature, whereas it was negatively correlated with apoptosis and cell adhesion signatures. The loss of intercellular adhesion makes it easy for tumor cells to migrate and invade, eventually leading to metastatic dissemination (40). Collectively, RNF6 functions as a versatile oncogene in colorectal cancer by promoting tumor growth and metastasis. RNF6 has also been shown to be a contextual oncogene or tumor suppressor gene in other cancer types, such as prostate cancer (14), leukemia (41), and esophageal squamous cell carcinoma (42, 43). In prostate cancer, RNF6 promoted androgen-independent cancer cell growth via ubiquitination of the androgen receptor (AR), which facilitates the recruitment of coactivators to induce AR pathway downstream gene transcription (14). However, RNF6 was believed to be a tumor suppressor in esophageal squamous cell carcinoma because of its chromosomal location and multiple somatic mutations (42, 43). Hence, the function of RNF6 is likely to be tissue- and cancer-type dependent. The unique amplification of RNF6 in patients with colorectal cancer, together with the compelling oncogenic effect of RNF6 in colorectal cancer cells and mouse models, suggests that RNF6 is particularly important for this specific cancer type.

RNF6 belongs to the RING finger E3 ligase family. Emerging evidence indicates that RING finger E3 ligases participate in numerous cellular processes primarily through facilitating the ubiquitin-dependent degradation of their target proteins (44). Unbiased screening of RNF6-binding proteins in colorectal cancer cells using immunoprecipitation–mass spectrometry unveiled TLE3 as a novel RNF6 interacting partner. Consistent with its function as an E3 ligase, we demonstrated that RNF6 mediated the polyubiquitination of TLE3 and its subsequent degradation in the proteasome in colorectal cancer cells. Strikingly, TLE3 is a powerful repressor of the β-catenin/TCF4 complex, and it negatively regulates Wnt/β-catenin signaling pathway (26, 27). Indeed, RNF6-mediated TLE3 degradation significantly suppressed the association of TLE3 with TCF4/LEF, which in turn, led to recruitment of β-catenin to TCF4/LEF, enhanced Wnt/β-catenin transcriptional activity, and the expression of its downstream factors associated with cell proliferation and EMT (depicted in Fig. 7). Concomitant with RNF6 expression, the clinical colorectal cancers enriched in the Wnt/β-catenin or Wnt-correlated pathways such as MYC and VEGF signatures by the GSEA analysis further supported our findings (29, 34). Furthermore, TLE3 gain- and loss-of-function experiments demonstrated that TLE3 degradation is essential for the oncogenic effect of RNF6. Thus, RNF6/TLE3/β-catenin form a functional axis that exerts progrowth and prometastatic behaviors in colorectal cancer progression.

Loss-of-function mutations of APC or gain-of-function mutations in β-catenin are well-known carcinogenic mechanisms to aberrantly activate canonical Wnt/β-catenin signaling in colorectal cancer (4, 11, 45). APC mutation is involved in aberrant crypts and early adenoma formation, and causes chromosomal instability (46). Concurrence with APC mutation and RNF6 copy number amplification is enhanced in patients with colorectal cancer patients. Mutations in the APC tumor suppressor have been shown to cause chromosomal instability by inducing spindle abnormalities (46). Mutant APC induces chromosome gain or loss randomly; however, it is reasonable to assume that RNF6 becomes specifically amplified through positive selection during tumorigenesis. Given the essentiality of Wnt/β-catenin activation in colorectal cancer development, the participation of RNF6 as a driver oncogene in this pivotal cascade is of particular significance. RNF6-mediated degradation of the Wnt suppressor TLE3 may therefore function as an enhancer to exacerbate Wnt/β-catenin signaling and drive colorectal carcinogenesis. Our results highlighted an underappreciated role of RNF6 gene amplification, which causes aberrant expression of this E3 ligase that specifically targets TLE3 for degradation, which in turn, promotes colorectal cancer development.

Our findings were further strengthened by the clinical impact of RNF6 amplification and overexpression in the survival of patients with colorectal cancer. Consistent with its oncogenic role, RNF6 mRNA expression was an independent prognostic factor that predicts poor survival of patients with colorectal cancer in two independent cohorts. Moreover, colorectal cancer patients with RNF6 copy number amplification had a higher risk of recurrence.

In summary, we demonstrate that RNF6 overexpression due to gene amplification is a common event in colorectal cancer. RNF6 plays a pivotal oncogenic role through TLE3 ubiquitin-mediated degradation, consequently activating Wnt/β-catenin pathway in colorectal carcinogenesis. Therapeutic intervention of RNF6 E3 ligase activity might be a novel strategy for blunting uncontrolled growth and metastasis in Wnt/β-catenin–addicted colorectal cancer. Moreover, RNF6 may serve as an independent prognosis marker for patients with colorectal cancer.

No potential conflicts of interest were disclosed.

Conception and design: L. Liu, J. Yu

Development of methodology: L. Liu, Y. Zhang, J. Zhang, J. Yu

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): L. Liu, Y. Zhang, Y. Dong, W. Kang, J. Yu

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): L. Liu, X. Li

Writing, review, and/or revision of the manuscript: L. Liu, Y. Zhang, C.C. Wong, J. Yu

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): L. Liu, F.K.L. Chan, J.J.Y. Sung, J. Yu

Study supervision: J. Yu

The authors would like to thank Professor Wei Jia (University of Hawaii Cancer Center) and Professor Xiaoyong Yang (Comparative Medicine and Cellular & Molecular Physiology, Yale University), who helped discuss the project and gave helpful suggestions.

This project was supported by RGC-GRF Hong Kong (14163817, 14106145, 14111216); HMRF Hong Kong (03140856); 973 Program China (2013CB531401); The National Key Technology R&D Program (2014BAI09B05); Science and Technology Program Grant, Shenzhen (JCYJ20170413161534162); and Shenzhen Virtual University Park Support Scheme to CUHK Shenzhen Research Institute and CUHK direct grant (to J. Yu).

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.