Researchers in Melbourne have developed an artificial intelligence-driven method to create synthetic protein inhibitors for CRISPR gene-editing systems. This AI-accelerated approach achieved results in 8 weeks that traditionally would require years of discovery research.
The study, published on January 26 in Nature Chemical Biology, details the design of AIcrs, which are AI-designed anti-CRISPRs. These inhibitors target the RNA-editing CRISPR-Cas13a system and exhibit nanomolar potency.
CRISPR technologies have advanced genetic medicine, but safety concerns persist due to the potential for the active enzyme to cause unintended damage to healthy genes. Anti-CRISPR proteins can mitigate this by regulating the gene-editing machinery. Natural anti-CRISPRs are rare; only 118 have been identified over 10 years of research.
Dr. Cyntia Taveneau, a protein designer at Monash University and lead author, stated that functional CRISPR inhibitors were rapidly produced using AI-accelerated protein design, and these inhibitors function in both bacterial and human cells.
The research team, led by Associate Professor Gavin Knott at the Monash University Biomedicine Discovery Institute and Dr. Rhys Grinter at the University of Melbourne’s Bio21 Institute, utilized RoseTTAFold Diffusion and ProteinMPNN. These tools generated 10,000 potential designs targeting the HEPN nuclease domain of LbuCas13a. From 96 filtered candidates, three lead inhibitors, named AIcrVIA1, AIcrVIA2, and AIcrVIA3, exhibited IC50 values of approximately 7 nanomolar, indicating high inhibitory potency.
Validation included X-ray crystallography and cryo-electron microscopy. The crystal structure of AIcrVIA1 at 1.9-angstrom resolution showed a close match between the actual protein and the computational design.
The inhibitors were effective in living systems. In bacterial cells, the expression of any of the three AIcrVIAs restored bacteriophage titers previously suppressed by CRISPR activity. In human HEK293T cells, the inhibitors restored fluorescent protein expression that had been reduced by Cas13a-mediated RNA cleavage.
Associate Professor Knott indicated that the ability to design bespoke inhibitors to regulate CRISPR will contribute to the development of CRISPR tools in research, medicine, agriculture, and microbiology.
Unlike natural anti-CRISPRs from phages, which offer limited mechanistic control, AI-designed inhibitors permit researchers to specify where and how they block CRISPR activity. This approach could be adapted to create inhibitors for other CRISPR systems, including TnpB, Fanzor, and CRISPR-guided DNA integrases.
Dr. Grinter noted that the discovery of natural inhibitors for clinically relevant targets remains challenging and time-consuming. He stated that this study implemented a rapid approach to anti-CRISPR design, using AI to create highly accurate and specific anti-CRISPRs.
