Coordinated activation of DNMT3a and TET2 in cancer stem cell-like cells initiates and sustains drug resistance in hepatocellular carcinoma

Background: Resistance to targeted therapies represents a major hurdle to successfully treat hepatocellular carcinoma (HCC). While epigenetic abnormalities are critical determinants of HCC relapse and therapeutic resistance, the underlying mechanisms are poorly understood. We aimed to address whether and how dysregulated epigenetic regulators have regulatory and functional communications in establishing and maintaining drug resistance. Methods: HCC resistant cells were characterized by CCK-8, IncuCyte Live-Cell analysis, flow cytometry and wound-healing assays. Target expression was assessed by qPCR and Western blotting. Global and promoter DNA methylation was measured by dotblotting, methylated-DNA immunoprecipitation and enzymatic digestion. Protein interaction, promoter binding and gene mutations of DNMT3a-TET2 was investigated by co-immunoprecipitation, ChIP-qPCR and Sanger sequencing. The regulatory and functional roles of DNMT3a and TET2 were studied by lentivirus infection and puromycin selection. The association of DNMT and TET expression with drug response and survival of HCC patients was assessed by public datasets, spearman correlation coefficients and online tools. Results: We identified a coordination of DNMT3a and TET2 as an actionable mechanism of drug resistance in HCC. The faster growth and migration of resistant HCC cells were attributed to DNMT3a and TET2 upregulation followed by increased 5mC and 5hmC production. HCC patients with higher DNMT3a and TET2 had a shorter survival time with less responding to sorafenib therapy than those with lower expression. Cancer stem cell-like cells (CSCs) displayed DNMT3a and TET2 overexpression, which were insensitive to sorafenib. Either genetic or pharmacological suppression of DNMT3a or/and TET2 impaired resistant cell growth and oncosphere formation, and restored sorafenib sensitivity. Mechanistically, DNMT3a did not establish a regulatory circuit with TET2, but formed a complex with TET2 and HDAC2. This complex bound the promoters of oncogenes (i.e., CDK1, CCNA2, RASEF), and upregulated them without involving promoter DNA methylation. In contrast, DNMT3a-TET2 crosstalk silences tumor suppressors (i.e., P15, SOCS2) through a corepressor complex with HDAC2 along with increased DNA methylation in their promoters. Conclusions: We demonstrate that DNMT3a and TET2 act coordinately to regulate HCC cell fate in DNA methylation-dependent and -independent manners, representing strong predictors for drug resistance and poor prognosis, and promising therapeutic targets for refractory HCC.