Inhibition of DNMTs and RNA editing increases immunogenic transposable element RNA to reduce tumor burden and prolong survival in a murine ovarian cancer model

Abstract

Abstract Novel therapies are urgently needed for ovarian cancer (OC), the fifth deadliest cancer in women. Preclinical data shows great promise for the treatment of OC with epigenetic therapy. The inhibition of DNMT activates transcription of double-stranded (ds)RNA, including transposable elements. These dsRNAs activate sensors in the cytoplasm and trigger the induction of type I interferon signaling. Adenosine deaminase 1 (ADAR1) is induced by interferon signaling and edits mammalian dsRNA with an A-to-I nucleotide change. These edited dsRNAs cannot be sensed by dsRNA sensors, and thus ADAR1 inhibits the type I interferon response. We show that transposable elements including Alus, LINEs, and ERVs, are edited by ADAR1 in human OC cell lines after DNMTi treatment. We hypothesized that increased ADAR1 editing reduces the DNMTi immune response and treated Adar1 knockdown cells with DNMT1 in vitro and in vivo. Combining ADAR1 knockdown with DNMTi significantly increased pro-inflammatory cytokine production compared to either perturbation alone. We treated an immunocompetent mouse model of OC with DNMTi after implanting them with Adar1 knockdown cells or control cells (ID8 Trp53−/−) and found that DNMTi treatment and Adar1 loss reduces tumor burden and extends survival in this model, and the combination of both is the most effective. Combining Adar1 loss and DNMTi transforms the immune microenvironment, including increased recruitment of natural killer cells and M1 macrophage polarization. This survival benefit is a result of type I interferon signaling and CD8+ T cells. Thus, epigenetically inducing transposable element transcription, followed by inhibition of RNA editing, is a novel therapeutic strategy to reverse immune evasion in OC. Research was supported by the National Cancer Institute R00CA204592 and R37CA251270 (to KBC), the DOD Ovarian Cancer Research Program (W81XWH2010273 to KBC), and by the Marlene and Michael Berman Endowed Fund for Ovarian Cancer Research. JIM was supported by an NCI NRSA Institutional Research Training Program grant (T32 CA 247756). SG was supported by a NRSA Predoctoral Fellowship (NIH/NCI 1F31CA254315-01). The authors would like to acknowledge the Institute for Biomedical Sciences at the George Washington University for graduate student support and training (SG, EEG, TK, VB).