Programmable C-to-G Transversions in Human Cells with CRISPR Base Editors

CRISPR-guided cytosine and adenine base editors (CBEs and ABEs) enable the efficient and programmable introduction of C-to-T and A-to-G changes, respectively. Although these editors have been widely adopted for many applications, they are limited in the scope of alterations they can create because they can only introduce transitions, and not transversions (i.e., purine-to-pyrimidine or vice versa) in DNA. To our knowledge, no enzyme that can directly catalyze DNA base transversions has been identified thus far, which makes it challenging to design a CRISPR-guided base editor capable of installing such alterations. However, we and others recently noted that an ABE could induce high purity C-to-G transversion edits at position 5, 6 or 7 of its CRISPR-targeted protospacer sequence in human cells5, 6. Based on this observation, we engineered two new base editor architectures that are capable of efficiently inducing targeted C-to-G DNA edits in human cells with a lower prevalence of C-to-W and indel byproducts. In its first iteration, the C-to-G base editor (CGBE1) harbors an RNA-guided Cas9 nickase, an E. coli-derived uracil DNA N-glycosylase (eUNG), and an APOBEC1 cytidine deaminase variant (R33A) which was shown to exhibit reduced off-target RNA and DNA editing activities. We found that using this variant - in the absence of uracil N-glycosylase inhibitors (UGIs) - also yielded enhanced C-to-G editing efficiency, especially in AT-rich sequence contexts. We also observed that removal of the eUNG yields a CGBE variant (miniCGBE1) that reduces indels at the cost of slightly reduced C-to-G editing efficiencies. These initial CGBE1 and miniCGBE1 variants provide an important proof-of-principle for inducing programmable C-to-G edits and a platform for engineering optimized versions of editors to fully unlock C-to-G transversions for research and therapeutic applications.