Artificial intelligence (AI)-designed proteins can now expand the CRISPR toolbox beyond what evolution has produced. In a new study, researchers report the design of synthetic RNA-guided nucleases that match or exceed the activity of natural enzymes, revealing how structure-guided design can generate functional genome-editing proteins with substantially different sequences. CRISPR-Cas technologies have transformed genetic engineering by allowing scientists to target and modify DNA and RNA precisely. Researchers are now exploring whether protein design methods can create new RNA-guided nucleases with novel properties beyond those found in nature. However, this remains challenging because multi-domain proteins are complex and small changes can easily disrupt enzyme activity. Although AI models have shown success in the design of new nucleases, they often produce versions that closely resemble the reference sequences used to train the models themselves. Here, Petr Skopintsev and colleagues present a protein design strategy that combines the ESM Inverse Folding (ESM-IF1) model with evolution-informed residue constraints to generate new variants of TnpB, a minimal CRISPR-Cas12-like nuclease, termed SynTnpBs. Skopintsev et al. then screened these AI-designed proteins for activity in bacterial, plant, and human cells, and used cryo-electron microscopy (cryo-EM) to determine the structure of the most divergent variants. The authors found that many AI-designed nucleases retained or surpassed the activity of the natural enzyme across multiple cell types. Cryo-EM revealed that the engineered proteins formed new stabilizing interactions at the RNA-DNA interface across different conformations, providing the first experimentally determined structures of AI-designed RNA-guided nucleases. Together, these findings demonstrate how the design strategy can produce active, non-natural genome-editing enzymes.
Science
Structure and evolution-guided design of minimal RNA-guided nucleases
16-Jul-2026