AlphaFold3 Breakthrough Could Revolutionize Gene Therapy with Safer Editing Tools

AlphaFold3, the AI successor to Google DeepMind's Nobel Prize-winning protein-folding tool, is being used to redesign gene-editing proteins and improve their safety. Researchers have developed a framework called ContactSeek that uses AlphaFold3's structural predictions to redesign base editors, which are the molecular scissors used in gene therapy. The result is gene-editing proteins that are more accurate than existing high-fidelity alternatives.

ContactSeek works by identifying contact probabilities between the editor protein and DNA, then swapping out specific amino acid residues to make the tool more discerning about what it grabs onto. The researchers focused on adenine base editors built from the Cas9-TadA system and were able to create a variant with just two mutations that vastly outperformed established high-fidelity editors in both precision and activity.

The method is also modular, allowing researchers to adapt ContactSeek for different classes of editing tools. The team successfully adapted it for LbCas12a-based cytosine base editors, another type of editing tool. All data and code for ContactSeek v1.0.0 have been made publicly available on Zenodo and GitHub.