Researchers have utilized Google's AlphaFold AI to enhance the safety of gene-editing proteins by redesigning them to reduce off-target effects. Gene-editing therapies, while promising, face challenges with safety due to the possibility of editing unintended DNA sequences.

Off-target edits occur when gene-editing systems, like CRISPR-Cas9, mistakenly bind to and alter DNA sequences that are similar but not identical to the intended target. While efforts have been made to improve specificity, errors can still arise when editing a large number of cells.

The new approach involved using AlphaFold to analyze the protein structures of gene-editing systems. By comparing how the Cas9 protein interacts with both on-target and off-target DNA sequences, researchers identified specific amino acids in Cas9 that change their contact points with the RNA when mismatches occur. This analysis, termed "Contact Seek," pinpointed key regions within the protein.

Through targeted modifications of these identified amino acids, the team successfully created Cas9 variants. One such variant demonstrated a significant reduction in off-target activity, dropping from 28 percent to 5 percent, while maintaining similar on-target editing efficiency. This method offers a way to tailor gene-editing systems for greater safety and could be a breakthrough for therapeutic development.