Minimizing Off-Target Mutations in DNA Editing by Re-engineering an Adenosine Deaminase for RNA
Background
Adenine base editors (ABEs) are tools that correct single bases without double-strand breaks, making them promising candidates for treating genetic diseases. Existing ABEs use evolved TadA deaminases derived from E. coli to convert adenine to inosine on DNA. A problem is bystander editing, where multiple adenines within the editing window are simultaneously converted. Even when targeting a single base, unintended A-to-G changes within a 4-8 nt range can alter amino acid sequences or disrupt splicing signals. In therapeutic applications for patient cells, bystander editing is a critical safety concern. Attempts have been made to narrow the editing window or screen TadA variants, but selectively reducing bystander editing while maintaining target efficiency has been challenging.
Key Findings
The research team took a different approach by re-engineering ADAR (adenosine deaminase acting on RNA), which normally acts on double-stranded RNA, for DNA editing. ADAR has high structural selectivity, recognizing double-stranded RNA structures and deaminating only specific adenosines. The researchers engineered the catalytic domain of ADAR to utilize this structural discrimination in the context of DNA-RNA heteroduplexes. They rearranged substrate recognition residues to ensure that only adenine at a specific location within the local double helix formed by the guide RNA and target DNA is efficiently deaminated, with minimal action on adjacent adenines.
As a result, bystander editing was significantly reduced compared to existing ABEs, while target editing efficiency was maintained. In sites with multiple adenines within the editing window, where the existing ABE8e edited two to three adenines simultaneously, the new editor demonstrated precision in selectively converting only the intended single adenine.
Significance and Implications
The reduction in bystander editing goes beyond simple precision improvement and increases the potential for clinical translation of base editing therapeutics. Reducing off-target mutations makes it easier to meet the safety evaluation criteria of regulatory agencies. Furthermore, it allows for the correction of desired locations even in coding regions with dense adenine sequences, expanding the range of treatable pathogenic variants that were previously difficult to address with ABEs. However, cell-type-specific delivery efficiency, the potential for off-target RNA editing, and reproducibility in large animal models remain as issues to be verified in subsequent studies.
Nature Biotechnology, Published online: 10 July 2026; doi:10.1038/s41587-026-03223-zBystander edits are minimized by engineering an adenosine deaminase acting on RNA for DNA base editing.
This technology can be directly applied to gene editing therapies for single-point mutation diseases such as sickle cell anemia and familial hypercholesterolemia. With existing ABEs, it was difficult to ensure the uniformity of the product because adenines around the target were also changed, but with bystander editing suppressed, the purity of the product increases in the manufacturing process, and batch-to-batch consistency is also ensured. From the perspective of cell therapy manufacturers, it becomes easier to meet quality control (QC) standards and simplify the non-clinical safety data package for IND applications. In the long term, it can also guarantee the precision of each locus independently in multi-locus simultaneous editing strategies, increasing the design freedom of complex gene editing therapies.