Pan-cancer RNA editing activity reveals complex editing functions and cancer immunotherapy biomarkers

Pervasive RNA editing in human diversifies the transcriptome and proteome. However, biological functions of most RNA editing sites remain to be uncovered. Here, we developed a computational framework (iPEAPR) to investigate editing functions in various pathways and regulatory elements, by using only summary information of millions of editing sites, in >5,000 cancer samples and in >2,500 normal samples, and large numbers of manually curated functional annotations. We observed heterogeneous editing activities across features. Surprisingly, enhancers were among features showing the highest editing activities. Editing of epithelial-mesenchymal transition was the most significantly associated with patient survival. Moreover, editing associations with cancer stemness, DNA repair deficiency, and tumor immune infiltrations uncovered known and potential regulators of tumor features. We constructed an editing-mediated regulatory network, which revealed new editing modulators, including experimentally validated TNRC6A. Lastly, we demonstrated that EIs can act as biomarkers of both anti-PD1 immunotherapy and other cancer drugs, such as MEK and BRAF inhibitors. Collectively, iPEAPR enabled depicting RNA editing-dependent abnormal functions in cancer and can be applied to more diseases.