Abstract
Base editors, composed of engineered deaminases fused with Cas proteins and a guide RNA, enable precise, programmable alteration of single nucleotides within the genome and transcriptome. This innovative technology holds promising therapeutic potential for correcting disease-causing point mutations. However, its clinical translation hinges on both high efficacy and accuracy. Non-specific unintended edits by base editors remain a critical challenge. Efforts to mitigate off-target activity have focused mostly on detecting recurrent RNA deaminations at specific sites. Complementarily, our methodology quantifies the total burden of RNA alterations, which is particularly effective for capturing stochastic off-target edits that evade conventional detection. Here, we applied the RNA editing index algorithm to quantify off-target levels across individual genes and identified 2,844 adenine base editors and 1,253 cytosine base editor hotspot genes susceptible to aberrant editing. Exon-level analysis revealed localized regions within genes that are particularly prone to off-target editing, including regions where edits introduce premature stop codons, a critical risk for therapeutic applications. By uncovering these previously unrecognized off-target landscapes, our study deepens our understanding of base editor specificity and provides a framework for optimizing their precision, accelerating the development of safer next-generation editing tools.