TRIM67 Activates p53 to Suppress Colorectal Cancer Initiation and Progression

Colorectal cancer is one of the most common malignancy and a leading cause of cancer-related deaths worldwide (1). Moreover, an increasing number of young people are dying from colorectal cancer (2). Most colorectal cancers develop from the sequential inactivation of tumor suppressor genes including APC, TP53, and SMAD4, as well as activation of oncogenes (e.g., KRAS; ref. 1). Recent studies have defined the “driver” mutations involved in colorectal cancer. In one study, the introduction of five key mutations, APC, TP53, SMAD4, KRAS, and PIK3CA, in normal human intestinal organoid induces transformation and tumorigenesis (3). On the other hand, mouse organoids with only three driver mutations (Apc, p53, and Kras) were able to recapitulate all stages of colorectal cancer from adenoma to metastasis (4).

p53 is a classical tumor suppressor frequently mutated across multiple cancer types. As a pivotal gatekeeper in tumorigenesis, p53 is activated in response to oncogenic or other cellular stress and it regulates a large number of genes involved in cell-cycle arrest, DNA repair, senescence, and apoptosis (5). In colorectal cancer, a high frequency of p53 mutations was observed in both the proximal (34%) and distal (45%) tumors, and p53 mutational status was associated with disease outcomes (6). Besides, successful response to chemotherapies such as 5-fluorouracil (5-FU) is dependent on the normal functioning of p53 (7). Apart from genetic mutations, emerging data also indicate that the protein level or activity of p53 is inhibited in a substantial proportion of patients with cancer with wild-type p53 through epigenetic or posttranslational mechanisms (5). Given a pivotal role of p53 inactivation in colorectal tumorigenesis and therapeutic response, it is critical to clarify the molecular mechanism by which p53 is dysregulated in cancers.

The TRIM family is defined by the presence of a common domain structure composed of a RING finger followed by a B-box and coiled coil domain (8). The presence of a RING finger indicates that TRIM proteins can function as E3 ligases. Indeed, certain TRIM proteins act as E3 ubiquitin ligases (9), small ubiquitin-like modifier (SUMO) or IFN-stimulated gene 15 (ISG15) ligases (10, 11) and they mediate specific protein degradation. TRIM family proteins have been reported to play an important role in diverse biological processes and their dysregulation is associated with carcinogenesis either as oncogene or tumor suppressors in a context-dependent manner (8, 12, 13). To unravel novel tumor-suppressive TRIM proteins, we interrogated the promoter DNA methylation status of all TRIM family members in The Cancer Genome Atlas (TCGA) database, leading to the discovery of a novel TRIM member, TRIM67, which is epigenetically silenced in colorectal cancer. However, the biological function, molecular mechanism, and clinical significance of TRIM67 in colorectal cancer are elusive. In this study, we examined the causality between the silencing of TRIM67 and colorectal tumorigenesis using colon-specific Trim67 knockout mice, and investigated whether TRIM67 can serve as a prognostic marker or therapeutic target for cancer therapy.

A total of 138 patients with histologically confirmed colorectal cancer who underwent surgery at Prince of Wales Hospital, the Chinese University of Hong Kong (Hong Kong), were enrolled in the study. Biopsy samples from primary colorectal cancer tumor and adjacent normal were obtained from patients with colorectal cancer at the time of operation before any therapeutic intervention. This cohort included 83 men and 55 women. The median age of patients was 58.8 years (range, 29–81 years). In addition, 20 age-matched normal colon mucosae from healthy subjects were collected as normal control. All patients provided informed consent for collecting the specimens for study. The study protocols have been approved by the Clinical Research Ethics Committee of Prince of Wales Hospital and the Chinese University of Hong Kong (Hong Kong). All patients provided written informed consent for obtaining the study specimens. This study was carried out in accordance with the Declaration of Helsinki of the World Medical Association.

LoxP flanked Trim67 transgenic mice were generated by Beijing Biocytogen Co., Ltd. Exons 4 and 5 of Trim67 gene were flanked by LoxP sites and were deleted by cross-mating with CMV-cre mice (#006054, The Jackson Laboratory; Supplementary Table S1). All of the transgenic mice were housed in a pathogen-free barrier environment for the duration of the study at the Department of Laboratory Animal Science at the Army Medical University in Chongqing, China. Trim67^+/− Apc^Min/+ mice were generated by intercrossing female Trim67^+/− with male Apc^Min/+ (strain: T001457; Nanjing Biomedical Research Institute of Nanjing University, Jiangsu, China). Trim67^+/− Apc^Min/+ mice were back-mated with Trim67^−/− mice to generate Trim67^−/− Apc^Min/+ mice. All experimental procedures were approved by the Animal Ethics Committee of the Army Medical University (Chongqing, China).

CDX2-CreER^T2 transgenic mice were purchased from The Jackson Laboratory (#022390). Cre-ER^T2 can only gain access to the nuclear compartment after exposure to tamoxifen (#T5648, Sigma-Aldrich). Trim67^fl/+ CDX2P-CreER^T2 and Trim67^fl/fl CDX2P-CreER^T2 mice were generated by a series of intercrossing. To induce intestinal epithelium–specific deletion of Trim67, 6- to 7-week-old mice with respective genotypes were daily intraperitoneally injected with tamoxifen dissolved in corn oil, with a dose of 100 mg/kg for four consecutive days. Two weeks after first dosing of tamoxifen, mice were injected intraperitoneally with 10 mg/kg azoxymethane (AOM; #A5486, Sigma-Aldrich) once a week for 6 consecutive weeks (14). Mice were monitored once a week and sacrificed 25 weeks after the first dosing of AOM. Intestine examination and histology inspection were performed. All experimental procedures were approved by the Animal Ethics Committee of the Army Medical University (Chongqing, China).

The isolation of mouse embryonic fibroblasts (MEF) was performed as described previously (15). Trim67^−/+ and Trim67^−/− MEFs were isolated from embryonic day 13∼14 fetuses that were derived from intercrosses of Trim67^−/+ mice. WT MEFs were obtained from intercrosses of WT mice. Pregnant mice were sacrificed at 13∼14-day postcoitum and each embryo was separated. All experimental procedures were approved by the Animal Ethics Committee of the Army Medical University (Chongqing, China).

A total of 1 × 10⁶ stable TRIM67-expressing or control HCT116 cells were injected subcutaneously into the left dorsal flank of 4-week-old male Balb/c nude mice. For drug treatment assays, 6 days after subcutaneous inoculation, mice injected with TRIM67-expressing or control cells were randomly divided into two groups, respectively. In 5-FU–treated group, 5-FU (50 mg/kg) was given by multipoint intratumoral injection, twice per week. All experimental procedures were approved by the Animal Ethics Committee of the Chinese University of Hong Kong (Hong Kong).

Data are presented as mean ± SD. The independent Student t test was used to compare the difference between two preselected groups. The difference in tumor growth rate between the two groups of mice was determined by repeated-measures ANOVA. The univariate and multivariate Cox regression analysis was performed to assess the prognostic value of TRIM67 protein expression status. Overall survival in relation to TRIM67 expression status was evaluated by the Kaplan–Meier 5-year survival curve and the log-rank test. Value of P < 0.05 was taken as statistical significance.

We systematically analyzed the core promoter regions (200bp upstream of transcription start site, TSS) of all human TRIM family members using Infinium Human Methylation 450K array dataset from TCGA database. Among TRIM proteins, TRIM67 and TRIM71 were frequently methylated in colorectal cancer compared with adjacent nontumor tissues (Fig. 1A, left). In particular, TRIM67 was hypermethylated at two loci (cg21178978 and cg27504802) in colorectal cancer tissues compared with adjacent nontumor tissues (Fig. 1A, right), suggesting an altered epigenetic regulation of this TRIM member in colorectal cancer.

We next evaluated TRIM67 mRNA expression and promoter methylation in our primary colorectal cancer cohort. TRIM67 mRNA expression was downregulated in 79% (109/138) of colorectal cancer tissues compared with their adjacent nontumor tissues (P < 0.0001; Fig. 1B); whilst promoter methylation of TRIM67 was significantly increased in colorectal cancer compared with their paired adjacent normal tissues by bisulfite genomic sequencing ((P < 0.0001; Fig. 1C; Supplementary Fig. S1A and S1B). TRIM67 mRNA expression was negatively correlated with its promoter methylation in colorectal cancer (P < 0.0001; Fig. 1D). Consistent with tissue samples, TRIM67 was silenced in all 7 colorectal cancer cell lines examined, but was expressed in normal colon tissues (Supplementary Fig. S1C). Dense promoter methylation was observed in colorectal cancer cell lines, but not in normal colon tissues (Supplementary Fig. S1D). Demethylation treatment with 5-aza-2′-deoxycytidine restored TRIM67 mRNA expression in colorectal cancer cells (Supplementary Fig. S1E), implying that its transcriptional silence was mediated by DNA methylation.

Consistently, TRIM67 protein expression was significantly reduced in colorectal cancer tumor tissues compared with adjacent nontumor tissues by IHC (P < 0.0001; Fig. 1E). We evaluated the protein expression of TRIM67 by IHC in tissue microarrays (TMA) consisting of 141 patients with colorectal cancer. Low TRIM67 protein expression was identified in 77.3% (109/141) of patients with colorectal cancer (Fig. 1F; Supplementary Fig. S1F). Kaplan–Meier survival analysis demonstrated that patients with colorectal cancer with low TRIM67 protein expression had significantly shorter survival (P = 0.031, log-rank test; Fig. 1F). Multivariate Cox regression analysis showed that low TRIM67 expression was an independent predictor of poor survival of patients with colorectal cancer (RR, 4.195; 95% confidence interval, 1.594–11.043; P = 0.004; Supplementary Table S2).

To determine whether Trim67 deficiency might predispose colorectal carcinogenesis, we generated constitutive Trim67 knockout mice and then crossed them to Apc^Min/+ mice (Fig. 2A; Supplementary Fig. S2A and S2B). At 3 months, Trim67^−/+ and Trim67^−/− mice showed normal colon histology (Supplementary Fig. S2C). Less than half of the Apc^Min/+ mice (9/20, 45%) developed tumors in the colon, whereas we observed a 100% tumor incidence in the colon of age-matched Trim67^−/+Apc^Min/+ and Trim67^−/−Apc^Min/+ mice (Fig. 2B and C; Supplementary Table S3). Moreover, 41% (9/20) and 44% (7/16) of Trim67^−/+Apc^Min/+ and Trim67^−/−Apc^Min/+ mice developed multiple tumors, respectively (Fig. 2B). In contrast, only 10% of Apc^Min/+ mice (2/20) developed multiple tumors (Fig. 2B). As a consequence, tumor multiplicity was significantly higher in heterozygous or homozygous Trim67-knockout Apc^Min/+ mice (Fig. 2D). In terms of overall tumor burden (sum of the total tumor size/animal; ref. 14), 4- and 5-fold increase was observed in Apc^Min/+Trim67^± and Apc^Min/+Trim67^−/− compared with Apc^Min/+ mice, respectively (Fig. 2D). Trim67 knockout was confirmed by qPCR (Supplementary Fig. S2D). Collectively, our results indicate that Trim67 functions as a tumor suppressor and its knockout promotes genetically (Apc^Min/+)-induced colorectal tumorigenesis in mice.

We next investigated the effect of colon-specific Trim67 deletion in a chemically AOM-induced model of colorectal carcinogenesis. Colon-specific, conditional Trim67 knockout mice were generated by crossing Trim67^fl/fl mice to Cdx2-CreER^T2 mice (Supplementary Fig. S2A and S2B). Trim67 was deleted specifically in the colon after exposure to tamoxifen at 2 months of age, and colorectal tumorigenesis was subsequently induced by AOM injection (Fig. 2E; ref. 14). As shown in Fig. 2F, the body weights of colon-specific Trim67^+/− (Trim67^fl/+Cdx2-CreER^T2) and Trim67^−/− (Trim67^fl/flCdx2-CreER^T2) were significantly decreased compared with wild-type mice. At the end of the study, 50% (5/10) of wild-type mice developed colon tumors, whilst 73% (8/11) of Trim67^−/+ mice and all of Trim67^−/− mice (8/8) developed colon tumors (Fig. 2G and H; Supplementary Table S3). Moreover, Trim67^−/− mice showed a significant increase in tumor multiplicity (4.9 vs. 0.8; P < 0.01) and burden (11.66 mm vs. 1.49 mm; P < 0.01) compared with wild-type mice (Fig. 2F). Trim67 expression was evidently reduced in tumors from colon-specific Trim67 knockout mice relative to that from control mice (Supplementary Fig. S2D). These findings indicated that Trim67 deficiency plays a causal role in promoting chemically induced colorectal tumorigenesis in mice.

We next determined the biological function of TRIM67 in human colorectal cancer cell lines. Ectopic expression of TRIM67 suppressed cell proliferation, as evidenced by cell viability assay and colony formation assay in HCT116, RKO, and LoVo cells (Fig. 3A; Supplementary Fig. S3A). Conversely, knockdown of TRIM67 in normal human colon mucosal epithelial cell line NCM460 and normal gastric epithelium cell line GES1 cells led to an increased ability to proliferate as compared with control cells (Fig. 3B). TRIM67 overexpression led to a significant increase in apoptosis (Supplementary Fig. S3B). Apoptosis induction was further evidenced by the cleavage of PARP and caspase-3, -7, -8, and -9 in HCT116, RKO, and LoVo cells overexpressing TRIM67 (Fig. 3C). TRIM67 also suppressed G₁–S cell-cycle phase transition in these colorectal cancer cell lines (Fig. 3A).

We next examined effect of TRIM67 on colorectal cancer tumor growth in vivo by subcutaneously injecting TRIM67-transfected and empty vector–transfected HCT116 cells in nude mice. Consistent with in vitro data, ectopic TRIM67 expression significantly suppressed tumor growth (P < 0.0001; Fig. 3D). Tumor weight at the end point was also significantly reduced in TRIM67-transfected cells compared with control cells (P < 0.001; Supplementary Fig. S3C). Overexpression of TRIM67 was validated by Western blot (Supplementary Fig. S3C). TRIM67 reduced tumor growth in vivo through inhibiting cell proliferation and promoting apoptosis, as indicated by Ki-67 and terminal deoxynucleotidyl transferase dUTP nick end labeling (TUNEL) assays, respectively (both P < 0.01; Fig. 3D).

To gain novel insights into molecular mechanism(s) by which TRIM67 inhibits colorectal tumorigenesis, we performed RNA sequencing (RNA-seq) on HCT116 cells expressing TRIM67 or control vector (Fig. 3E; Supplementary Fig. S4A; Supplementary Table S4). Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis of differentially expressed genes (DEG) revealed that DEGs upregulated by TRIM67 were most significantly enriched in the p53 signaling pathway (Fig. 3E). These included p53 downstream target genes involved in cell-cycle arrest [SFN, CCND1, and CDKN1A (p21)], apoptosis (p21, BAX, NOXA, FAS, OSGIN1, and PUMA) and other cellular functions such as angiogenesis and autophagy (SESN2, THBS1, PPM1D, SERPINE1, MDM2, ISG20L1, and GADD45A; Fig. 3E). The induction of p53 downstream target genes by TRIM67 was abolished in HCT116 p53^−/− knockout cells (Supplementary Fig. S4B; ref. 16), indicating that their induction was dependent on a functional p53.

In HCT116, RKO, and LoVo cells expressing wild-type p53, TRIM67 increased protein expression of p53 and p21, concomitant with enhanced transcription of p53 downstream targets (MDM2, p21, NOXA, GADD45A, and OSGIN1; Fig. 3F; Supplementary Fig. S4C and S4D). However, p53 mRNA level was not changed (Fig. 3F; Supplementary Fig. S4C), implying a posttranscriptional mechanism. We therefore examined whether TRIM67 affects p53 protein stability. We first treated TRIM67-overexpressing and control HCT116 cells with cycloheximide, a protein synthesis blocker, and compared the kinetics of p53 degradation. As shown in Fig. 3G, p53 degradation was reduced in TRIM67-expressing HCT116 cells compared with control cells. Immunofluorescence with anti-p53 antibody showed that the percentage of cells with strong nuclear p53 expression was dramatically increased in TRIM67-overexpressing HCT116 cells (P < 0.001; Supplementary Fig. S4E). In line with these observations, analysis of cytoplasmic and nucleus fractions revealed that TRIM67 overexpression led to accumulation of stabilized p53 in the nucleus (Fig. 3H). Taken together, our results indicate that TRIM67 mediates its tumor-suppressive effect by preventing p53 degradation, thereby activating p53 signaling pathway.

Posttranscriptional regulation of p53 is primarily mediated by MDM2, which is not only a transcriptional target of p53 but also an E3 ubiquitin ligase that mediates degradation of p53 via the ubiquitin–proteasome pathway (17). TRIM67 overexpression increased MDM2 at mRNA level, but had no effect on MDM2 protein expression (Fig. 3E and F). Notably, overexpression of TRIM67 induced MDM2 cleavage in the nucleus, generating a 60-kDa fragment and leading to the loss of C-terminal RING domain responsible for p53 ubiquitination activity (Fig. 3F and H; refs. 18, 19).

To ask whether TRIM67 could directly interfere with p53–MDM2 axis, coimmunoprecipitation (co-IP) was performed to probe the interaction between TRIM67 and endogenous p53 and MDM2. Pulldown of Flag-tagged TRIM67-associated protein complex identified both p53 and MDM2, inferring an association of TRIM67 with p53 and MDM2 (Fig. 4A). In addition, interaction between TRIM67 and ectopically expressed p53 was observed in co-IP using cells cotransfected with Flag-tagged TRIM67 and HA-tagged p53 plasmids (Fig. 4A). To verify the specificity of the interaction between TRIM67, p53, and MDM2 in cell-free conditions, we performed in vitro protein binding assays using purified recombinant GST-tagged p53, GST-tagged MDM2, and Flag-tagged TRIM67 proteins (Fig. 4B). Purified TRIM67 interacted with p53 but not MDM2 (Fig. 4B), indicating that p53 is a direct binding partner of TRIM67. Interaction between TRIM67 and p53 was readily detected after only 30-minute incubation co-IP (Fig. 4C), implying a strong interaction between these two proteins. Co-IP further showed that TRIM67 disrupted p53–MDM2 interaction (Fig. 4D). Since MDM2 catalyzed p53 polyubiquitination (20), we examined whether TRIM67 could block MDM2-mediated p53 ubiquitination. Consistent with our hypothesis, p53 polyubiquitination was dramatically inhibited in TRIM67-overexpressing cells (Fig. 4E). We next sought to identify p53 domain(s) responsible for binding to TRIM67. p53 possesses a N-terminal transcriptional activation region (TAD domain), followed by a polyproline region (PPR), a central DNA-binding domain (DBD), and a C-terminal domain (21). We generated three p53 deletion mutants lacking TAD-PPR (N-D2), C-terminal (C-D3), or TAD plus C-terminal (N-D3) domains and tested their interaction with TRIM67 (Fig. 4F). Co-IP demonstrated that p53 N-D2, but not C-D1 or N-D3, binds to TRIM67 (Fig. 4G), implying that p53 C terminus was responsible for its interaction with TRIM67. The p53 C terminus is subjected to multiple posttranslational modifications and serves as recruitment sites for cofactors (21). Although the p53–MDM2 interaction relies primarily on the TAD domain, MDM2 also binds to C terminus where it attaches the ubiquitin molecules, a prerequisite step for p53 degradation (21). Our results suggest that binding of TRIM67 to C-terminal region of p53 leads to disruption of the interaction between p53 and MDM2, thereby repressing MDM2-mediated p53 ubiquitination and stabilizing p53 protein levels.

We next tried to determine which functional domain(s) in TRIM67 are responsible for its effect on p53 activation. TRIM67 possesses four major domains: a RING finger domain, a coiled coil domain (CC), a COS domain, and a SPRY domain (Supplementary Fig. S5A). We constructed eight TRIM67 deletion mutants (Supplementary Fig. S5A). COS-SPRY deletion completely abolished nuclear translocation of RING-CC mutant; while other mutants were expressed in both cytosol and nucleus (Supplementary Fig. S5B). As the nuclear localization of TRIM67 is essential for its capacity to protect p53, we fused RING-CC to a nuclear localization signal (NLS) to promote nuclear localization (Supplementary Fig. S5A and S5B). We next tested effect of TRIM67 deletion mutants on expression of p53 downstream targets (p21, MDM2, and FAS). TRIM67 deletion mutants lacking either N-terminal RING finger domain or C-terminal COS and SPRY domains failed to induce expression of these genes, whereas TRIM67 deletion mutant with the partial deletion of CC domain (ΔC mutant) induced p53 target genes comparably with that of full-length TRIM67 (Supplementary Fig. S5C). Correspondingly, Western blot analysis confirmed that only the ΔC deletion mutant retained the ability to promote p53 protein expression and induce cleavage of MDM2 (Supplementary Fig. S5D). As the CC domain is a secondary structure formed by intertwining of multiple alpha-helix repeats and is believed to mediate TRIM dimerization and homo-oligomerization (22), we reasoned that CC domain may be dispensable for TRIM67–p53 interaction. Nevertheless, the complete deletion of CC domain (ΔCC) might impact tertiary structure, which compromised protein activity (Supplementary Fig. S5C and S5D). We thus conclude that the integrity of RING finger, COS, and SPRY domains are essential for the effect of TRIM67 on p53 signaling.

Given that TRIM67 functions as a key tumor suppressor in colorectal cancer, we next investigated the underlying mechanism regulating its transcription. Normal colon epithelial cell line NCM460 and normal gastric epithelium cell line GES1 were used as only these two cell lines, but not colorectal cancer cell lines, express endogenous TRIM67 with partial methylation in TRIM67 promoter (Supplementary Fig. S1D). We observed that DNA-damaging chemotherapeutic drugs doxorubicin and actinomycin D dramatically induced TRIM67 expression in a dose-dependent manner in NCM460 and GES1 cells (Fig. 5A; Supplementary Fig. S6A). As both of these drugs could trigger a p53-mediated DNA damage response, we hypothesized that TRIM67 might be a target of p53 signaling. In response to cellular stress, p53 binds to the consensus sequence motif, which is two decamers with a spacer (RRRCWWGYYY-N_0-13bp- RRRCWWGYYY, where R is A/G, W is A/T, Y is C/T, N is any base; ref. 23). Recent studies showed that a single decamer (half-site) can act as a p53 minimal essential binding unit to drive transcription under stress conditions (23, 24). We thus searched for putative p53 binding motifs within TRIM67 promoter region (−2kb to +1kb of TSS). While the consensus sequence motif could not be found, we identified two potential p53 half-sites: −86bp to −77bp (half-site 1) and +339bp to +348bp (half-site 2; Fig. 5B). We thus examined the occupancy of p53 on TRIM67 promoter in NCM460 cells exposed to doxorubicin by chromatin immunoprecipitation (ChIP)-PCR using six pairs of primers spanning TRIM67 promoter region (Fig. 5B). Among the regions targeted by these primers, P4 (−162bp to −81bp) and P5 (+232bp to +378bp) overlapped with half-site 1 and 2, respectively. P1 and P6 that, respectively, target approximately 1kb upstream and downstream of the TRIM67 TSS served as negative controls, together with a positive control targeting the p21 promoter. ChIP-PCR demonstrated that P2-P4, but not P5, exhibited significant p53 binding upon doxorubicin treatment (Fig. 5C), inferring that p53 interacts with TRIM67 promoter through half-site 1. Moreover, ectopic p53 expression induced the significant enrichment of p53 binding in the regions targeted by P2-P4 along with the induction of TRIM67 mRNA (Fig. 5D and E; Supplementary Fig. S6B). Induction of TRIM67 mRNA expression by doxorubicin was significantly reduced in p53 knockdown cells (Fig. 5F; Supplementary Fig. S6C), indicating a p53-dependent transcriptional induction of TRIM67 by DNA-damaging drugs. To confirm that p53 activates TRIM67 expression via half-site 1, we generated luciferase reporters containing TRIM67 promoter region ranging from −1200 to +100 (Fig. 5G). Mutation of half-site 1 completely abolished induction of TRIM67 luciferase reporter activity by p53 (Fig. 5G). Thus, this p53-responsive element in the TRIM67 promoter is essential for p53-dependent regulation of TRIM67 transcription. We also analyzed p53 ChIP-seq peaks at the TRIM67 promoter in the Cistrome database (http://cistrome.org/db/#/). Interestingly, we identified p53 ChIP-seq peak signals around the TSS of TRIM67 gene (−527bp to +172bp) in two untransformed human fetal fibroblast lines GM00011 and GM06170 after treatment with doxorubicin for 12 hours (Fig. 5H; ref. 25). Importantly, half-site 1 was located in close proximity to the summit of the p53 ChIP-seq peak (Fig. 5H). RNA-seq data from the same dataset also revealed that doxorubicin upregulated TRIM67 mRNA as well as other p53 targets MDM2, p21, PMAIP1, BAX, SFN, and FAS (Fig. 5H; ref. 25). In addition, TRIM67 remained silenced upon ectopic p53 expression in HCT116 and RKO cells, in which TRIM67 promoter was densely methylated (Fig. 5I), indicating that the induction of TRIM67 by p53 is abrogated because of promoter hypermethylation in colorectal cancer cells. Our results indicate that p53 binds to the TRIM67 promoter and directly upregulates its expression during periods of cellular stress.

Because p53 activation drives TRIM67 expression, and TRIM67 directly stabilizes p53 protein, we reasoned that TRIM67 is engaged in a positive regulatory loop to trigger p53-induced proliferative arrest and apoptosis induction in response to cellular stress, such as DNA damage. To test this possibility, we treated NCM460 and GES1 cells with doxorubicin with or without knockdown of TRIM67. In control cells with TRIM67 expression, doxorubicin addition completely abolished cell proliferation (Fig. 6A). On the contrary, cells expressing TRIM67-shRNA continued to proliferate even in the presence of doxorubicin (Fig. 6A). Consistently, the silencing of TRIM67 significantly reduced doxorubicin -induced apoptosis and G₂–M arrest in NCM460 and GES1 cells, as determined by flow cytometry (Fig. 6A; Supplementary Fig. S7A) and Western blot analysis (Fig. 6B). TRIM67 silencing–associated doxorubicin resistance was likely a consequence of dampened p53 activation, as demonstrated by a significant reduction in doxorubicin -induced p53 and p21 protein expression (Fig. 6B). We also detected reduced phosphorylation of p53 at Ser15 in TRIM67 knockdown cells (Fig. 6B). Phosphorylation of p53 (Ser15) is known to disrupt p53–MDM2 interaction and promote p53 stability and activity (26). Besides, doxorubicin -induced upregulation of other DNA damage checkpoint proteins, such as phospho-CHK1, phospho-CHK2, and phospho-BRCA1, were attenuated by TRIM67 knockdown (Fig. 6B). In line with these results, TRIM67 knockdown partially abolished doxorubicin -induced mRNA expression of p53 downstream target genes, including p21, MDM2, NOXA, BAX, and GADD45A (Fig. 6C). Taken together, TRIM67 is an essential component of the p53-mediated DNA damage response by amplifying p53 response via a positive feedback loop.

We next dissected the role of TRIM67 on p53 activity in Trim67 knockout mice. We first evaluated the effect of Trim67 loss on p53 pathway in MEFs derived from wild-type, Trim67^+/−, and Trim67^−/− mice (Fig. 6D). In these MEFs, basal p53 activity was not significant different between wild-type or Trim67 knockout cells (Fig. 6E). Upon doxorubicin treatment, however, Trim67-deficient MEFs displayed a marked reduction in the protein expression of p53, p21, phospho-p53 (Ser15), and the DNA damage response indicator γ-H2A.X (Fig. 6E). Accordingly, doxorubicin -induced mRNA expression of p53 downstream targets (p21, Mdm2, Fas, Noxa, Sfn, and Isg20l1) was repressed in Trim67 knockout MEFs (Supplementary Fig. S7B). In light of these data, we examined mRNA expression of p53 and its target genes p21, Fas, Sfn, Mdm2, Isg20l1 in colon tumors and their normal colon mucosa from 5 Trim67^−/−Apc^Min/+ mice and 4 Apc^Min/+ mice, respectively. Compared with adjacent normal colon tissues, p53 mRNA was prominently induced in tumors from both Trim67^−/−Apc^Min/+ mice and Apc^Min/+ mice (Fig. 6F). However, p21, Fas, and Sfn expression were significantly reduced in Trim67^−/−Apc^Min/+ tumors compared with Apc^Min/+ tumors (Fig. 6F), indicating that Trim67 knockout inhibited p53 signaling pathway during tumor progression in transgenic mouse model. These results further confirmed the important role of Trim67 in p53 activation in response to DNA damage in colorectal tumorigenesis in vivo.

Approximately 40%–50% of sporadic colorectal cancer carry p53 mutation (27), a majority of which are missense mutations (28, 29). Mutation hotspots include two subtypes: DNA contact mutants (e.g., R273H) that disrupt DBD domain; or confirmational mutants that alter DBD conformation (e.g., R175H; ref. 29). To clarify the function of TRIM67 in colorectal cancer cells expressing mutant p53, we selected three colorectal cancer cell lines: HT-29 (R273H), SW480 (R273H/P309S), and DLD-1 (S241F). In contrast to p53 wild-type colorectal cancer cells, TRIM67 overexpression in these three cell lines failed to alter p53 or p21 protein (Supplementary Fig. S8A), indicating that TRIM67 could not stabilize mutant p53 protein. Accordingly, overexpression of TRIM67 had no effects on cell growth and apoptosis in mutant p53 colorectal cancer cells, as indicated by cell viability assay and cleaved PARP expression, respectively (Supplementary Fig. S8A and S8B). Mutant p53 also possesses distinct transcriptional signatures (30). Indeed, TRIM67 mRNA expression could not be activated by overexpression of mutant p53 (R175H, R273H, and R248W) in NCM460 and GES1 cells (Supplementary Fig. S8C). Thus, the TRIM67-p53 positive feedback loop was abrogated in colorectal cancer cells harboring mutant p53.

Because many chemotherapeutic agents exert their anticancer properties by inducing DNA damage (31), TRIM67 might enhance the sensitivity of colorectal cancer cells expressing functional p53 signaling to these genotoxic drugs. To test this hypothesis, we treated p53 wild-type and null isogenic (p53^−/−) HCT116 cells ectopically expressing TRIM67 with doxorubicin, 5-FU, or oxaliplatin (Supplementary Fig. S9). Compared with control cells, TRIM67 overexpression in wild-type HCT116 cells enhanced cell growth inhibition (Fig. 7A), apoptosis (Fig. 7B), and p53 protein levels (Fig. 7B) after drug treatment. In contrast, such an effect could not be replicated in p53^−/− HCT116 cells (Fig. 7A and C), suggesting that the sensitizing effect of TRIM67 is dependent on a functional p53. To determine whether the enhanced sensitivity to chemotherapy drugs is observed in vivo, we treated mice harboring p53 wild-type HCT116 xenografts expressing TRIM67 or empty vector with vehicle or 5-FU (50 mg/kg, twice/week) for 17 days. As shown in Fig. 7D, TRIM67 overexpression plus 5-FU synergistically inhibited the growth of HCT116 xenografts compared with vehicle or single treatment as early as day 9 posttreatment, which was maintained until the end of the treatment [average tumor size, 165.8 mm³ vs. 381.9 mm³ (5-FU only), P < 0.01]. Hence, TRIM67 improved the anticancer activity of chemotherapeutic drugs in p53 wild-type colorectal cancer.

p53 signaling is a major gatekeeper that prevents tumorigenesis by conserving genomic stability upon detection of DNA damage. In this study, we first identified TRIM67 as an essential component of p53-mediated DNA damage response. TRIM67 is epigenetically silenced in approximately 80% of colorectal cancer, and its silencing correlates with poor prognosis. Consistent with its role as a functional tumor suppressor, Trim67 knockout exacerbated Apc^Min/+ and chemically induced colorectal cancer in mice. Mechanistically, we demonstrated that TRIM67 participates in a positive feedback circuitry in which activated p53 drives the expression of TRIM67, which reinforces p53 stability via disruption of p53–MDM2 interaction. Our work suggests that the TRIM67–p53 axis is critical to arrest colorectal carcinogenesis and its inactivation is a pivotal event in colorectal cancer development.

A series of in vitro and in vivo studies clearly demonstrated that TRIM67 possesses an antitumorigenic function in colorectal cancer. Using mutant Apc and AOM-induced colorectal cancer models, we demonstrated that either heterozygous or homozygous deletion of Trim67-predisposed mice to colorectal tumorigenesis. In particular, the loss of both Trim67 alleles greatly accelerated tumorigenesis, both in terms of tumor multiplicity and burden. In agreement with a tumor suppressor role of TRIM67, TRIM67 knockdown promoted cell proliferation of normal colon and gastric epithelial cell lines, whereas ectopic expression of TRIM67 in established colorectal cancer cell lines suppressed cell proliferation in vitro and xenograft growth in nude mice. Our findings therefore provide evidence that TRIM67 functions as a tumor suppressor in colorectal cancer.

TP53, a bona fide tumor suppressor, is essential for maintaining genomic stability and fidelity. Approximately, half of patients with colorectal cancer harbor inactivating mutations in TP53 gene (6). However, p53 is frequently silenced in patients with wild-type p53 via transcriptional or posttranscriptional regulation, thereby contributing to p53 loss of function (5). TRIM67 is a member of the TRIM family that possesses a degenerate RING domain with defective E3 ligase activity. Here, we demonstrated that TRIM67 participates in the posttranscriptional regulation of p53 by directly interacting with p53 and protecting it from MDM2-mediated ubiquitination. MDM2 is the major E3 ligase of p53. MDM2 interacts with p53 at N-terminal TAD domain, central DNA-binding domain, and C terminus, and each of these interactions is required for optimal negative regulation of p53 by MDM2 (21, 32). We found that TRIM67 inhibits MDM2 action by occupying C terminus of p53, which destabilizes p53–MDM2 complex and blocks MDM2-mediated p53 ubiquitination and proteasome degradation. All four domains (RING, CC, COS, SPRY) of TRIM67 are indispensable for p53 activation. As a result, the ectopic expression of TRIM67 in colorectal cancer cell lines stabilized p53 protein expression and activated p53 signaling. On the contrary, the knockout of TRIM67 in cell lines and transgenic mouse models led to a marked reduction in p53 protein expression and activity. Apart from TRIM67, other TRIM proteins have been reported to induce (TRIM6/TRIM13/TRIM19) or inhibit (TRIM28/TRIM65) p53 signaling (13, 33), underscoring the regulatory role TRIM proteins in p53 activation.

Our work further identified TRIM67 as a pivotal component of a positive feedback loop that amplifies p53 signaling under stress conditions. We revealed that TRIM67 promoter region contains binding sites for p53 and p53 directly regulates TRIM67. Indeed, treatment with DNA-damaging drugs dramatically upregulated p53-induced expression of TRIM67, which in turn, stabilized p53 and promoted p53-mediated downstream gene expression. Basal TRIM67 expression levels are normally kept low in normal cells, and the positive feedback regulation might explain the dramatic upregulation of TRIM67 expression by several tens to thousands of fold upon the induction of DNA damage. Loss of TRIM67 had thus a significant impact on p53-orchestrated DNA damage response. It is noteworthy that normal functioning of the p53 pathway is vital to the effectiveness of a range of chemotherapeutic agents that targets DNA (34). In response to chemotherapy, p53 activates DNA damage response and triggers various antitumor programs, including apoptosis and cell-cycle arrest (34). The dysfunction of p53 pathway plays a critical role in resistance to chemotherapy (35). As TRIM67 loss compromises p53 activity and its downstream signaling, its restoration might constitute a therapeutic strategy for p53 wild-type colorectal cancer. Supporting our idea, reintroduction of TRIM67 enhanced the antiproliferative effect of chemotherapy drugs doxorubicin, 5-FU, or oxaliplatin in p53 wild-type colorectal cancer cells. Corroborating these findings, TRIM67 low expression predicted poor survival in patients with colorectal cancer. Several DNA-demethylating drugs are clinically available or have been entering clinical trials (36). Given that TRIM67 expression can be restored by demethylation drugs, combining these agents with chemotherapy might be a promising approach to improve p53 response and chemotherapy efficacy in p53 wild-type patients with colorectal cancer with low TRIM67 expression.

In summary, our work establishes TRIM67 as a novel tumor suppressor in colorectal cancer; and its loss is sufficient to drive colorectal tumorigenesis in cell lines and transgenic mice models. We also identify a novel TRIM67-dependent positive regulatory loop that amplifies p53 signaling under cellular stress. Finally, our work highlights the potential of TRIM67 as a biomarker and a therapeutic target in colorectal cancer.

No potential conflicts of interest were disclosed.

Conception and design: J. Yu, S. Wang, Y. Zhang

Development of methodology: S. Wang, J. Huang, J. Zhai, C. Li, L. Zhao

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): S. Wang, Y. Zhang, J. Huang, J. Zhai, G. Wang, H. Wei

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): S. Wang, Y. Zhang, G. Wei

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

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): G. Wang, H. Wei, Z. Zhao

Study supervision: J. Yu

This study was supported by RGC-GRF Hong Kong (14111216, 14106415, 14163817); National Natural Science Foundation of China (NSFC; No.81572758); Science and Technology Program Grant Shenzhen (JCYJ20170413161534162); National Key Research and Development Program Fund China (2016YFC1303200); Grant from Faculty of Medicine CUHK on Microbiota Research, Vice-Chancellor's Discretionary Fund CUHK, CUHK direct grant, Shenzhen Virtual University Park Support Scheme to CUHK Shenzhen Research Institute.

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.