High expression of TRIM25 predicts recurrence in patients with CRC
We analyzed publicly available CRC mRNA expression profiles (GSE20842) obtained from the NCBI and found that the mRNA expression of TRIM25 is elevated in CRC tissues compared with that in normal tissues (Fig. 1a). To confirm our in silico observations, we detected TRIM25 protein levels in CRC samples. IHC staining showed that TRIM25 is present in both the cytoplasmic and nuclear regions of CRC cells. Further analysis demonstrated that the TRIM25 level in CRC tissues was significantly higher than that in adjacent normal tissues (Fig. 1b-c). To explore the potential role of TRIM25 in CRC therapy, we initially evaluated the TRIM25 level in 26 primary tumor tissues from patients with stage III CRC that were treated with OXA-based chemotherapy. The results showed that the TRIM25 level in OXA-resistant patients who developed recurrence during the follow-up period was significantly higher than that of patients who had no recurrence (Fig. 1d and Supplementary Figure S1a). We further analyzed TRIM25 expression in 223 paraffin-embedded human CRC specimens from patients who received Xelox or FOLFOX treatment after surgery (the patient characteristics are summarized in Supplementary Table 1). Representative IHC staining confirmed that TRIM25 levels were markedly increased in patients with CRC with tumor relapse (Fig. 1e). Patients with high TRIM25 levels demonstrated an observably higher recurrence rate than those with low TRIM25 levels (28.9% vs. 15.0%, P = 0.012, Fig. 1f and Table 1). Furthermore, compared with low TRIM25 levels, high TRIM25 levels were associated significantly with worse overall survival and disease-free survival (P = 0.006, Fig. 1g and Supplementary Figure S1b). Collectively, the above observations suggest that elevated TRIM25 levels contribute to the progression of CRC and are associated with the failure of OXA-based chemotherapy.
TRIM25 confers OXA resistance in CRC cells in vitro
We then examined whether TRIM25 was associated with resistance to OXA-based therapy in preclinical models. First, we established stable TRIM25 knockdown and overexpressing cells from the SW48 and SW480 CRC cell lines (Fig. 2a). A CCK8 assay showed that the IC50 values for OXA were decreased in the TRIM25-knockdown cells and increased in the TRIM25-overexpressing cells (Fig. 2b-c). Consistently, in the presence of OXA, TRIM25 knockdown dramatically inhibited, while TRIM25 overexpression enhanced, the colony-formation ability of SW48 and SW480 cells (Fig. 2d). Moreover, compared with the control cells, knockdown of TRIM25 resulted in significantly increased OXA-induced apoptosis of CRC cells, while overexpression of TRIM25 reduced OXA-induced apoptosis (Fig. 2e). Measurement of cleaved caspase 3 and cleaved PARP further confirmed that knockdown of TRIM25 increased OXA sensitivity and upregulation of TRIM25 conferred OXA resistance (Fig. 2f). Taken together, these results revealed that TRIM25 confers OXA resistance in CRC cells.
TRIM25 promotes stem cell properties of CRC cells
As reported previously, stemness is believed to be responsible for chemotherapy resistance, thus we hypothesized that TRIM25 is involved in regulating CRC stemness and performed experiments to test our hypothesis. We conducted the sphere formation assay, and found a decrease in sphere numbers and sizes in TRIM25 knockdown cells compared with the corresponding control cells, while overexpression of TRIM25 enhanced the sphere formation ability of CRC cells (Fig. 3a). In addition, the expression of stem cell related molecules, such as EpCAM, SOX2, CD133, and CD44, in SW48 and SW480 cells was markedly reduced after TRIM25 inhibition, while their expression increased after TRIM25 overexpression (Fig. 3b and Supplementary Figure S2). Furthermore, limiting dilution analysis (LDA) in vivo confirmed the markedly reduced stem cell frequency in TRIM25 knockdown SW480 cells (Fig. 3c), with the formation of smaller and lighter tumors than those formed by the control SW480 cells (Fig. 3d-e). In addition, no visible tumors could be formed in nude mice when 1×104 TRIM25-knockdown SW480 cells were inoculated. These findings indicate the crucial role of TRIM25 in promoting the stem cell properties of CRC cells.
TRIM25 regulates EZH2 stability in CRC cells
As the catalytic subunit of PRC2, EZH2 plays an essential role in tumor progression. Previous studies revealed that targeting EZH2 inhibits CSC self-renewal and enhances the sensitivity of CRC to OXA [2, 15]. In the present study, we found that knockdown of TRIM25 decreased the protein level of EZH2 (Fig. 4a). However, there was no significant effect on EZH2 mRNA levels in both SW48 and SW480 cells (Fig. 4b), suggesting that TRIM25 might affect the stability of EZH2. To substantiate this assumption, we treated TRIM25-knockdown or control CRC cells with the protein synthesis inhibitor cycloheximide (CHX) and the proteasome inhibitor MG132. The results showed that knockdown of TRIM25 shortened the half-life of endogenous EZH2 protein in CRC cells after CHX treatment (Fig. 4c-d). The level of EZH2 was modestly increased in TRIM25-knockdown SW48 cells treated with MG132, and the same results were obtained in SW480 cells (Fig. 4e), implying that the ubiquitin–proteasome pathway might be involved in TRIM25-mediated stability of EZH2. Finally, we detected the EZH2 levels in the same cohort of CRC samples used for TRIM25 analysis using IHC staining and found that high EZH2 levels correlated significantly and positively with high TRIM25 levels (Fig. 4f). CRC samples with high TRIM25 levels showed a higher proportion of high EZH2 levels, whereas samples with low TRIM25 levels exhibited a lower proportion of high EZH2 levels (63.9% vs. 45.7%, P = 0.009, Fig. 4g). Patients with CRC with high levels of TRIM25 and EZH2 had the shortest overall survival (P < 0.001, Supplementary Figure S1c) and disease-free survival (P < 0.001, Fig. 4h) compared with patients with low TRIM25 or low EZH2 levels. Taken together, these observations suggest that TRIM25 regulates EZH2 levels in CRC cells by reducing the degradation of EZH2.
TRIM25 blocks TRAF6-mediated ubiquitination of EZH2
TRIM25 is an E3 ligase, therefore, we wondered if TRIM25 modulates EZH2 stability through the ubiquitin–proteasome pathway. First, we analyzed the interaction of TRIM25 with EZH2 in CRC cells. Using co-immunoprecipitation and western blotting, we verified the interaction between TRIM25 and EZH2 at both the exogenous and endogenous protein levels (Fig. 5a-b). Further immunofluorescence double staining demonstrated the co-localization of TRIM25 and EZH2 in the nuclei of SW480 cells (Supplementary Figure S3a). Then, we performed in vivo ubiquitination assays in HEK293T cells transfected with siTRIM25, Flag-EZH2, His-tagged ubiquitin wild-type (WT), or mutation plasmids (K48 or K63 mutants). As shown in Fig. 5c, knockdown of TRIM25 in HEK293T cells increased the poly-ubiquitination of EZH2, indicating that EZH2 is not a substrate of TRIM25 E3 ligase. Interestingly, the enhanced EZH2 polyubiquitination by knockdown of TRIM25 was mainly extended through the K63-linkage instead of the K48-linkage. These findings were further validated in SW48 and SW480 cells (Supplementary Figure S3b).
TNF receptor associated factor 6 (TRAF6) is a member of the TNF receptor associated factor (TRAF) protein family, and functions as an E3 ubiquitin ligase and a scaffold protein. TRAF6 mediates the K63-linked ubiquitination of EZH2 in prostate cancer [27]. Thus, we speculated whether TRIM25 is involved in TRAF6-mediated ubiquitination of EZH2. First, we confirmed the co-localization of endogenous EZH2 and TRAF6 in the nuclei of SW480 cells (Supplementary Figure S3c). Then, we transfected siNC or siTRIM25 into SW480 cells. Western blotting analysis of the whole-cell extracts (WCE) showed that knockdown of TRIM25 had no significant effect on TRAF6 but decreased the level of EZH2 in SW480 cells, which was consistent with the results shown in Fig. 4a. A further immunoprecipitation assay showed that TRAF6 protein could be detected in EZH2 immunoprecipitate and knockdown of TRIM25 enhanced the interaction between EZH2 and TRAF6, implying that TRIM25 might interfere with TRAF6-mediated EZH2 ubiquitination (Fig. 5d). To confirm the role of TRIM25 in TRAF6-mediated EZH2 ubiquitination, we blocked TRAF6 in TRIM25-knockdown SW480 cells. Further co-immunoprecipitation and immunoblotting analysis showed that knockdown of TRIM25-induced EZH2 ubiquitination and degradation could be rescued using siTRAF6 (Fig. 5e-f). Taken together, these results suggest that TRIM25 stabilizes EZH2 by preventing TRAF6 binding to EZH2.
EZH2 is required for TRIM25-induced OXA-resistance in CRC both in vitro and in vivo
To further assess whether TRIM25 mediates OXA-resistance via EZH2, we blocked EZH2 in TRIM25-overexpressing cells using shRNA or an EZH2 inhibitor (UNC1999). Colony formation, CCK8 and Annexin V/PI apoptosis assays showed that inhibition of EZH2 significantly rescued the effect of TRIM25 overexpression on both the growth and anti-apoptotic capacity of CRC cells treated with OXA (Fig. 6a-c). Besides, the sphere formation ability of SW48 and SW480 cells overexpressing TRIM25 was markedly suppressed by EZH2 inhibition (Fig. 6d).
Moreover, a nude mouse xenograft model was used to evaluate the effect of EZH2 inhibition on TRIM25-induced OXA-resistance in vivo. Consistent with the in vitro findings, inhibition of EZH2 using shEZH2 or UNC1999 resulted in a significant reduction in tumor volume and weight when treated with OXA (Fig. 6e-f). Further TUNEL staining showed that inhibition of EZH2 resulted in more OXA-induced apoptosis compared with that in TRIM25-overespressing cells in vivo, as demonstrated by a higher proportion of TUNEL positively stained cells after OXA treatment (Fig. 6g-h). Overall, EZH2 is essential for TRIM25-induced OXA-resistance in CRC, and inhibition of EZH2 is expected to overcome TRIM25-induced OXA-resistance in clinical practice.