Precise in vivo functional analysis of DNA variants with base editing using ACEofBASEs target prediction

Your paper is considered potentially of high interest to the field. However, there are several noted questions/concerns that are essential to be addressed:

1. F0 work – there are several key issues, including the substantive delay noted in the tnnt2a F0 embryos not noted in the ENU or insertional mutant alleles. Reconciling whether this is due to the specific allele they are making versus unanticipated negative consequences in the injected animals needs to be resolved.

We thank the referees and editors for pointing out the discrepancy between previously reported zebrafish mutant alleles with particular loss of cardiac contraction (silent heart) and the observed medaka “silent heart” phenotype, associated with a global developmental delay.

We have performed two sets of additional experiments to address this issue, validating the medaka tnnt2a silent-heart phenotype, first targeting tnnt2a at another site to generate additional alleles and second to establish stable homozygous mutants in F1.

1) Creating a different PTC (W201X) using the tnnt2a-W201 sgRNA with evoBE4max microinjections. With this independent sgRNAtargeting a different site of the protein, we obtained robust (91%) medaka silent-heart phenotypes fully resembling the initially observed phenotypes when introducing a PTC at position Q114 (Q114X) (Figure 5-supplement 2).

2) To validate the relevance of the F0 editant phenotype, we established, raised and in-crossed heterozygous F0 founder fish with germline transmission of the Q114X allele. The resulting F1 homozygous tnnt2a-Q114X mutant embryos displayed the same medaka silent heart phenotype initially observed (Figure 5f’) and were undistinguishable from the F0 editants.

While these results confirm our initial tnnt2a editant observations, they also highlight a distinct, broader function of tnnt2a in medaka compared to zebrafish, where the mutant phenotype is limited to the heart.

These findings are fully detailed in the revised version of the manuscript.

2. The presentation suggests that base editors will not result in double-stranded DNA breaks. That is not the case in other systems due to the integral single-stranded DNA 'nicking' in base editors. In addition, the rapidly dividing zebrafish embryo will convert a nick in DNA into a full double-stranded break upon replication. Either the language included in the paper needs to be substantively revised, or experimental data that measures this issue needs to be documented.

We thank the referees for raising this point. We have extended our analysis and have included Amplicon-seq to reveal and quantify alleles not immediately uncovered in the Sanger analysis. This NGS analysis revealed a potential nick-dependent indel formation in editants of 4.9% and 15.8%.

We analyzed all editors tested at the medaka oca2-Q333 locus and incorporated our findings in the new Figure 3, connected to Figure 2, as the samples for the NGS analysis were the same as used for Sanger sequencing.

Additionally, we performed NGS analysis for ABE8e GFP-C71 experiments (Figure 4g), evoBE4max tnnt2a-Q114 experiments (Figure 5e), ABE8e kcnh6a-R512 experiments (Figure 6f) and evoBE4max ube2b-R8 experiments (Figure 7-supplement 3).

We summarized our findings and directly compared the target C or A conversion efficiencies obtained by Sanger sequencing as opposed to the newly acquired Illumina data on the same samples in Table 5. Moreover, we have removed results obtained by TIDE analysis to avoid confusion from the main text and Tables 1-3, as well as the corresponding Materials and Method section.

We have taken care to quantify editing and indel formation in response to the editing attempts in the revised version of the manuscript.

Germline work

3. The authors present 8 different loci with good F0 success. However, there have been false-positives from F0 gene editing science in the field. In addition, the noted differences in the phenotype from tnnt2a and prior alleles is also suggesting some likely limitations on F0 work. The authors need to provide key comparison data from their gene editing via germline propagation for at least tnnt2a and one or two of the novel cardiac mutants that show modest penetrance in the F0 screening work.

We thank the referees for raising this point. We have extended our analysis and have included Amplicon-seq to reveal and quantify alleles not immediately uncovered in the Sanger analysis. This NGS analysis revealed a potential nick-dependent indel formation in editants of 4.9% and 15.8%.

We analyzed all editors tested at the medaka oca2-Q333 locus and incorporated our findings in the new Figure 3, connected to Figure 2, as the samples for the NGS analysis were the same as used for Sanger sequencing.

Additionally, we performed NGS analysis for ABE8e GFP-C71 experiments (Figure 4g), evoBE4max tnnt2a-Q114 experiments (Figure 5e), ABE8e kcnh6a-R512 experiments (Figure 6f) and evoBE4max ube2b-R8 experiments (Figure 7-supplement 3).

We summarized our findings and directly compared the target C or A conversion efficiencies obtained by Sanger sequencing as opposed to the newly acquired Illumina data on the same samples in Table 5. Moreover, we have removed results obtained by TIDE analysis to avoid confusion from the main text and Tables 1-3, as well as the corresponding Materials and Method section.

We have taken care to quantify editing and indel formation in response to the editing attempts in the revised version of the manuscript.