ABSTRACT
Cas nucleases are popular tools for genome editing applications due to their ability to introduce DNA breaks at desired genomic locations. Such differential targeting is achieved through loading an RNA guide complimentary to the intended target sequence. As it turns out, sequences with only a partial match to the guide can also be cleaved. A large number of experiments have shed light on this off-targeting, outlining a number of rather peculiar empirical rules that detail the effect of mismatches at various positions and at various relative distances. We construct a kinetic model predicting on-target cleavage efficiency as well as off-target specificity. Our model explains a unified targeting rule for any target harboring mismatches, independent of their abundance and placing, and the observed decoupling between efficiency and specificity when protein-DNA interactions are weakened. We favorably compare our model to published experimental data from CRISPR-Cas9, CRISPR-Cpf1, CRISPR-Cascade, as well as to the human Argonaute 2 systems. Understanding the origins of off-targeting principles is important for the further development of CRISPR-Cas as a precise genome editing tool.
