Targeting fidelity of adenine and cytosine base editors in mouse embryos

Base editing directly converts a target base pair into a different base pair in the genome of living cells without introducing double-stranded DNA breaks. While cytosine base editors (CBE) and adenine base editors (ABE) are used to install and correct point mutations in a wide range of organisms, the extent and distribution of off-target edits in mammalian embryos have not been studied in detail. We analyze on-target and proximal off-target editing at 13 loci by a variety of CBEs and ABE in more than 430 alleles generated from mouse zygotic injections using newly generated and published sequencing data. ABE predominantly generates anticipated A•T-to-G•C edits. Among CBEs, SaBE3 and BE4, result in the highest frequencies of anticipated C•G-to-T•A products relative to editing byproducts. Together, these findings highlight the remarkable fidelity of ABE in mouse embryos and identify preferred CBE variants when fidelity in vivo is critical.

1. The key data are presented in Table 1 and they are, frankly, very hard to digest. The authors need to come up with a way to represent these data such that one can derive some general conclusions about their findings by using a graphical display. For instance, one could take a generic (schematic) locus, and then, for each of the 6 base editors they analyse, graph out (perhaps as bars above the position of interest): -Desired point mutation at the target -Undesired point mutation at the target -Any point mutation in the region covered by the gRNA -All other point mutations in that haplotype

Response
We thank the reviewer for the suggestions, which were implemented in a new figure (figure 1) and the data are much easier to digest. The new graphs show (1) the percentage of on-target, bystander and proximal mutations as well as indels that are based on the locations of target sites, and (2) either desired (also called intended or anticipated) and undesired (also called unintended or unanticipated) mutations based on the editing purity.
2. Essentially this is the display in supplementary notes, except the nature of the alleles is, frankly, not important for the average user -the average user needs to know, what to expect from using a particular form of a base editor.

Response
We moved those data into Supplementary Figures 4. Line 66: should be "within the on-target"

Response
We corrected the sentence. intended to show the total number of founders as well as the number of analyzed alleles (founders can carry two or more mutant alleles), but not 3 the number of edited alleles at on-target in the text. We edited the text in the manuscript and added the numbers. 5. "Product purity" is an awkward and ambiguous phrase (putting it mildly). "The intended allele was obtained in x%" should be used -or, for the Table -"intended allele."

Response
We believe that "product purity" is the appropriate term and reflects "intended mutation" or "anticipated mutation", which was now added to the text.
Thank you for handling our manuscript entitled "Targeting fidelity of adenine and cytosine base editors in mouse embryos". We were pleased that the reviewers considered our work important and we thank them for their constructive comments.
Specifically, based on the suggestions by reviewer #2, we added graphs (now figure 1) to illustrate the results in a comprehensive and clear way.

Reviewer #1
The manuscript by Lee et al. reports on the fidelity of various CBE and ABE base editors in mouse zygotes, based on the analysis of 436 mutant alleles. This large dataset is a valuable resource and guide for the future production of targeted mouse mutants as the degree of on-target, bystander and proximal off-target editing differs among the editors.
The manuscript is well organized, easy to read and compares the author´s results to previous literature.

Reviewer #2
Lee and colleagues perform an important and timely analysis of specificity of base editing using the constellation of such tools engineered by David Liu's lab. Of particular broad interest is the fact that this analysis is done in mice (where HDR efficiency may be lower).
The key data are presented in Table 1 and they are, frankly, very hard to digest. The authors need to come up with a way to represent these data such that one can derive some general conclusions about their findings by using a graphical display. For instance, one could take a generic (schematic) locus, and then, for each of the 6 base editors they analyse, graph out ( 4. Line 66: should be "within the on-target"

Response
We corrected the sentence. We intended to show the total number of founders as well as the number of analyzed alleles (founders can carry two or more mutant alleles), but not the number of edited alleles at on-target in the text. We edited the text in the manuscript and added the numbers.