Key facts
- Researchers used AlphaFold AI to identify and modify gene-editing proteins, aiming to reduce off-target effects.
- The study focused on the Cas9 protein, a key component in gene-editing systems, and its interaction with DNA and guide RNA.
- By analyzing structural changes in Cas9 using AlphaFold, researchers pinpointed amino acids responsible for accommodating mismatched DNA bases.
- A new variant of Cas9 was developed, showing a significant reduction in off-target editing while maintaining on-target activity.
- The method proved effective for Cas9 and was also shown to work with a similar protein, Cas12.
Gene-editing technologies, while promising for therapeutic applications, have been hampered by safety concerns related to off-target effects, where the editing machinery mistakenly alters unintended DNA sequences. Even rare off-target events can become significant when therapies require editing numerous cells.
Researchers have explored various strategies to enhance the specificity of gene-editing systems. One common approach involves designing guide RNAs that share minimal similarity with other genomic locations. Additionally, modifications to Cas proteins, such as Cas9, have been developed to reduce their propensity for off-target edits. These proteins interact with guide RNA and DNA, enforcing specificity by typically disengaging if there are too many mismatches in base pairing.
The latest research, published in Nature, leveraged Google's AlphaFold AI, a powerful protein-folding software, to address this safety challenge. The team adapted AlphaFold, which has been updated to handle interactions between proteins and nucleic acids, to analyze the structural nuances of gene-editing complexes. Their goal was to identify specific regions within the Cas9 protein that mediate problematic interactions with mismatched DNA sequences.
To achieve this, the researchers first created a library of off-target editing sites. They then used AlphaFold to model the interaction between the Cas9 protein, guide RNA, and target DNA. By comparing the structures generated by AlphaFold for perfectly matched on-target sites versus mismatched off-target sites, they observed that many off-target interactions caused Cas9 to adopt slightly different conformations. Crucially, they found that over 95% of these off-target sites altered which amino acids within Cas9 contacted the RNA, indicating flexibility in accommodating mismatches.
This analysis led to the development of a computational tool called 'ContactSeek,' which identifies amino acids in Cas9 that change their contact probabilities when binding to off-target sites. Focusing on clusters of these amino acids, the researchers identified key positions within the protein. They then systematically tested variants of Cas9 with different amino acids at these identified positions.
This experimental approach resulted in the creation of a modified Cas9 variant that maintained high activity at the intended target sequence but significantly reduced off-target activity from 28% to 5%. The researchers also demonstrated that this method is effective with different guide RNAs and can be applied to other Cas family proteins, such as Cas12. This AI-driven redesign offers a promising avenue for developing safer and more precise gene-editing therapies.