Deciphering Cas9 specificity: Role of domain dynamics and RNA:DNA hybrid interactions revealed through machine learning and accelerated molecular simulations

CRISPR/Cas9 technology is widely used for gene editing, but off-targeting still remains a major concern in therapeutic applications. Although Cas9 variants with better mismatch discrimination have been developed, they have significantly lower rates of on-target DNA cleavage. This study compares the dynamics of the highly specific Cas9 from Francisella novicida (FnCas9) to the commonly used SpCas9. Using long-scale atomistic Gaussian accelerated molecular dynamic simulations and machine learning techniques, we deciphered the structural factors behind FnCas9's higher specificity in native and off-target forms. Our analysis revealed that Cas9's cleavage specificity relies more on its domain rearrangement than on RNA:DNA heteroduplex shape, with significant conformational variations in Cas9 domains among off-target forms, while the RNA:DNA hybrid showed minimal changes, especially in FnCas9 compared to SpCas9. REC1-REC3 domains contacts with the RNA:DNA hybrid in FnCas9 acted as critical discriminator of off-target effects playing a pivotal role in influencing specificity. In FnCas9, allosteric signal transmission involves the REC3 and HNH domain, bypassing REC2, leading to a superior efficiency in information transmission. This study offers a quantitative framework for understanding the structural basis of elevated specificity, paving the way for the rational design of Cas9 variants with improved precision and specificity in genome editing applications.