Highly efficient A-to-G base editing by ABE8.17 in rabbits

Adenine base editors (ABEs), composed of an evolved adenine deaminase fused to the Cas9 nickase, enable efficient and precise A-to-G conversion in various organisms. However, the base editing of some challenging loci with the ABE7.10 system in rabbits was inefficient in our previous study. Here, we show that ABE8.17 and SpRY-ABE8.17 can efficiently induce base editing in mouse and rabbit embryos. In addition, this strategy can be used to precisely mimic clinical point mutations in rabbits. Furthermore, by eliminating the linker in ABE8.17, we created ABE8.17-NL, which achieved efficient base editing within a narrowed window (2–4 nts) in human HEK293FT cells. Collectively, these findings show that ABE8.17 systems can efficiently induce efficient A-to-G base editing at desired sites and that the ABE7.10 system is inefficient, thus providing an efficient way to generate ideal disease models in rabbits.


INTRODUCTION
Adenine base editors (ABEs), containing an evolved Escherichia coli tRNA-specific adenosine deaminase (TadA) and a catalytically impaired Cas9 protein (nCas9), can efficiently convert A to G with little accompanying double-stranded DNA breaks (DSB) generation. 1 The ABE7.10 system has been applied to efficiently introduce single nucleotide modifications in various plants 2,3 and animals. [4][5][6][7][8] Recently, evolved TadA8 variants (TadA-8e and TadA-8s) were developed by introducing additional mutations into TadA-7.10. 9,10 Compared with TadA-7.10 monomers, TadA-8e and TadA-8s monomers support more efficient A-to-G base conversion with SpCas9 (ABE8e and ABE8s). Our previous study demonstrated that the ABE7.10 system provides a simple and highly efficient method for inducing single-nucleotide substitutions in rabbits. 8 However, some challenging loci were unable to be effectively base edited by ABE7.10 in the previous study. 8 To efficiently edit these challenging loci, ABE8 systems were used in this study. Our results show that ABE8.17 enables efficient A-to-G base editing at desired loci in rabbits. In addition, we found that ABE8.17 can edit multiple A bases in a broad window within the protospacer in rabbit embryos. Thus, we created ABE8.17-NL (no linker) by deleting the linker to narrow the window of activity, improving the precision of ABE8. 17.
Overall, our findings demonstrate that ABE8.17 provides a highly efficient system for the desired single-nucleotide substitutions in rabbits. By eliminating the linker sequence, we generated ABE8.17-NL, an optimized ABE, narrowing the editing window to 2-4 nts.
To expand the targeting scope of ABE8.17, we investigated whether SpRYCas9 12 (a Cas9 variant, PAM: NRN > NYN) could enable efficient A-to-G conversion in human HEK293FT cells by fusing it to TadA-8.17 ( Figure S1). As shown in Figure 1C, efficient base editing (3%-48%) was achieved at 8 sites by using SpRY-ABE8.17 (Table S3). In addition, a significant increase in base editing was achieved by using Nme2-ABE8.17 relative to Nme2-ABEmax, which was shown to be an inefficient base-editing system in our previous study (Figures 1D, and S1; Table S3). 13 Highly efficient A-to-G base editing by ABE8.17 in mouse and rabbit embryos First, 3 target loci (Tyr, Dmd and Lmna) were used to test the efficiency of ABE8.17 in mouse embryos (Figure 2A). Base editing was conducted in mouse embryos via the microinjection of ABE8.17-encoding mRNA or ABE7.10-encoding mRNA, and single-guide RNAs (sgRNAs). Sanger sequencing results showed that 12 of the 13 desired edits in Tyr exhibited efficiencies ranging from 10% to 80% ( Figure 2B; Table S1). Four of the 11 desired edits in Dmd showed efficiencies ranging from 8% to 79% when using ABE8.17 but no desired base-editing activity at these two loci when using ABE7.10 ( Figure 2C; Table  S1). A significantly increased base-editing efficiency was also achieved at the A6 and A9 sites of the Lmna gene by ABE8.17 relative to ABE7.10 ( Figure 2D; Table S1).
Then ABE8.17 was used to target the base editing of 3 rabbit loci, which was inefficient when using the ABE7.10 system in our previous study 8 ( Figure 2E). Sanger sequencing results showed that 3 of the 9 desired edits in Tyr-1 presented efficiencies ranging from 10% to 18% and that 4 of the 5 desired edits in Tyr-2 presented efficiencies ranging from 41% to 72% when using ABE8.17 ( Figure 2F and 2G; Table S2). However, none of the desired mutations in these two loci were detected when using ABE7.10 ( Figures 2F and 2G; Table S2). Similar results were obtained for the Lmna gene ( Figure 2H and Table S2), suggesting that ABE8.17 was more efficient than ABE7.10. Then, the SpRY-ABE8.17 with a relaxed PAM was used to target the desired A5 mutations at the Lmna locus. Sanger sequencing results showed that 5 of the 9 desired edits of A5 were achieved using SpRY-ABE8.17 with efficiencies ranging from 22% to 40% ( Figure 2I; Table S2).
These results indicated that the ABE8.17 and SpRY-ABE8.17 systems were efficient at the desired loci than ABE7.10 in mouse and rabbit embryos, suggesting the potential use of ABE8.17 to develop animal models for human genetic diseases in rabbits.

ABE8.17 can induce efficient A-to-G conversion in rabbits
The majority of known human genetic diseases are caused by point mutations, with C-to-T or A-to-G mutations accounting for  Table S3. **p < 0.01, two-tailed unpaired t-test.  approximately half of all known pathogenic SNPs. 1 Notably, rabbits are considered better animal models than mice in recapitulating some human diseases because of the higher similarity of their physiology, anatomy and genetics to those of humans. 14 Here, two mutation sites in Tyr (p.T325A) and Lmna (p.L530P) were used for A-G base editing by ABE8.17 in rabbits. The Tyr mutation (p.T325A) is the major causal genetic mutation responsible for human ocular albinism (OA) and oculocutaneous albinism (OCA). 15,16 OCA1 is an autosomal recessive disorder characterized by reduced or absent melanin pigment in the skin, hair, and eyes due to deficient tyrosinase catalytic activity. 17 As shown in Figures 3A and 3B; Table S5, 3 of the 5 desired Tyr (p.T325A) mutations were identified in rabbit pups by using Sanger sequencing, and the results of targeted deep sequencing showed 66.72%, 13.09%, and 40.27% Tyr mutation efficiencies in pups #1, #2, and #5, respectively ( Figure 3C).
A point mutation (p.L530P) in the Lmna gene results in the development of Emery-Dreifuss muscular dystrophy (EDMD), which is characterized by early elbow and Achilles tendon contracture, slow and progressive muscle atrophy and weakness, and cardiomyopathy with conduction block. 18 As shown in Figures 3D and 3E; Table S5, 1 of the 7 desired Lmna mutations (p.L530P) mutations was identified in rabbit pups (#6) by using Sanger sequencing, and the results of targeted deep sequencing showed a 45.41% desired Lmna mutation efficiency in rabbit pup #6 ( Figure 3F). In addition, decreased in lamin A/C protein expression relative to the WT rabbits was identified in rabbit pup #6 (p.L530P) by western blotting ( Figure 3G). In addition, the retention of 21 bp of intron 9 of Lmna was observed by using quantitative real-time polymerase chain reaction (PCR) with specific primers targeting the exon and intron ( Figures 3H, 3I, and S2), which was also confirmed by Sanger sequencing ( Figure 3J). Furthermore, no undesirable base change or sgRNA sequence-dependent off-target mutations were observed by deep sequencing in these rabbits, indicating the efficiency and precision of ABE8.17 in rabbits ( Figures S3 and S4; Data S1). These results show that ABE8.17 and SpRY-ABE8.17 can induce efficient A-to-G conversion and show great promise as base editing tools for the generation of rabbit disease models.
Efficient base editing with a narrow window using ABE8.17-NL Although ABE8.17 can induce efficient A-to-G conversion, the large window within the protospacer is a substantial limitation when multiple A bases are located in the sgRNA. Here, the ABE8.17-NL (no linker) system was used to narrow the editing window in HEK293FT cells ( Figure 4A).
We initially investigated the effects of base-editing precision and efficiency when six polyA sites were targeted by using the deleted linker sequence between TadA-8.17 and nCas9. As shown in Figures 4B-4G, the base-editing window was significantly narrowed by using ABE8.17-NL relative to ABE8.17, which contained the (SGGS) 2 -XTEN-(SGGS) 2 linker (32 amino acids initially included in the ABE7.10 construct to improve editing efficiency). 1 Then, we tested the effects of several sites containing A bases within the window on the base editing precision and efficiency by using ABE8.17-NL (Figure 4H). The results showed that the window from A3 to A6 (4 nts) was narrowed by using ABE8.17-NL relative to the large editing window from A3 to A8 (6 nts) and that high activity at A4 and A5 was maintained using ABE8.17( Figure 4I). Taken together, these results demonstrate that ABE8.17-NL can induce efficient base editing within a narrow window (2-4 nts) in human HEK293FT cells.

DISCUSSION
In this study, we confirmed that the ABE8.17 system enables robust A-to-G base editing at loci that are difficult to edit by using ABE7.10 in rabbits. First, we demonstrated that the ABE8 variants showed robust A-to-G base-editing in HEK293FT cells. Second, TadA8.17 was demonstrated to be compatible with SpRYCas9, providing an opportunity to eliminate the constraints imposed by PAM availability. Furthermore, this strategy can be used to precisely mimic human pathologies by efficiently inducing point mutations in rabbits. These results demonstrated the high efficiency of A-to-G base editing by ABE8.17 in rabbits.
There was no significant off-target effect of predicted gRNA-dependent sites in the base-editing rabbits in this study. The gRNA-independent sites also need to be checked. However, previous studies demonstrated that base editors may cause genome-wide off-target effects on DNA and RNA, 19,20 and sgRNA-independent off-target effects are mainly induced by the deaminase domain rather than the Cas9 domain. In addition, there are many gaps in the published rabbit genome, which is a major limitation of whole-genome sequencing. Thus, we focused on the sgRNA dependent off-target effects in this study.Deep sequencing results demonstrated that ABE8.17 and SpRY-ABE8.17 could induce site-specific, single-base substitutions with efficiency rates of 13.09-66.72% in F0 rabbit pups, which is induced by mosaicism mutations after the microinjection of the embryos, causing unequal genome editing in individual blastomeres. 21 These results confirmed that the mosaicism generated by CRISPR/ Cas9 was commonly detected in F0 gene-edited pups. 8,22,23 High-precision base editor systems represent essential tools for the future application of DNA editing, especially in gene therapy. However, ABE8.17 showed an expanded window within the protospacer when multiple A bases were located in the sgRNA, which is also a major limitation of the future application of ABE8.17. Therefore, we generated ABE8.17-NL, which reduced the width of the editing window  (2-4 nts) in human HEK293FT cells. However, compromised editing efficiencies at 4 out of 7 sites tested were also determined by using ABE8.17-NL. We attempted to improve efficiency by introducing the point mutations (N127K, Q154R) 24 inTadA-8.17, which showed superior editing activity to NG-ABEmax. Our results showed that the editing efficiency was improved by 4% at A8 and decreased 6% at A5 and A6 by introducing N127K and Q154R (data not shown). Thus, we hypothesize that there is a balance betweenthe editing efficiency and editing window size, as reported in previous studies. [24][25][26] In summary, the ABE8.17 system can be used to efficiently induce base mutations at target sites that are difficult to edit using ABE7.10, and provide an efficient method for mimicking clinical disease mutations in rabbits.

Ethics statement
New Zealand white rabbits, Lianshan black rabbits and ICR mice were obtained from the Laboratory Animal Center of Jilin University (Changchun, China). All animal studies were conducted according to experimental practices and standards approved by the Animal Welfare and Research Ethics Committee of Jilin University (IACUC number: pzpx20200102027).

Cell culture and transfection
Human kidney epithelial cells (HEK293FT) were cultured in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% fetal bovine serum (HyClone), 2 mM GlutaMAX (Life Technologies), 100 U/mL penicillin, and 100 mg/mL streptomycin and incubated at 37 C in an atmosphere containing 5% CO 2 . The cells were seeded into 6-well plates at a density of 120,000 cells per well and transfected using Hieff Trans Liposome nucleic acid transfection reagent (Yeasen Biotechnology, Shanghai, China). The sequences of the sgRNAs are listed in Table S3. The isolated DNA was amplified by PCR with the TIANamp Genomic DNA Kit (TIANGEN, Beijing, China) according to the manufacturer's instructions. The primer sequences are listed in Table S4.

Estimation of editing frequency
EditR (https://moriaritylab.shinyapps.io/editr_v10/) online software 27 and the TIDE web tool (https://tide.deskgen.com/) 28 were applied to estimate the editing frequency of Sanger sequencing. mRNA and gRNA preparation ABE8.17 and SpRY-ABE8.17 were linearized with NotI and transcribed in vitro using the HiScribe T7 ARCA mRNA Kit (NEB). The sgRNAs were then amplified and transcribed in vitro using the MAXIscript T7 Kit (Ambion). mRNA was purified using the RNeasy Mini Kit (Qiagen) according to the manufacturer's protocol.

Microinjection and embryo transfer
The protocol used for the microinjection of pronuclear-stage embryos was described in detail in our previously published study. 29 Briefly, ABE mRNA (100 ng/mL) and sgRNA (50 ng/mL) were coinjected into the cytoplasm of pronuclear-stage embryos. The injected embryos were cultured for 30-60 min, after which approximately 30-50 injected embryos were transferred into the oviduct of the recipient mother.

Single-embryo PCR amplification and rabbit genotyping
The injected embryos were collected at the blastocyst stage. Genomic DNA was extracted from each embryo using lysis buffer (1% N-P40) at 56 C for 60 min and 95 C for 10 min in a Bio-Rad PCR amplifier. The genomic region surrounding the target site was PCR amplified and subjected to Sanger sequencing.
Genomic DNA was extracted from the skin tissue collected by ear clipping in newborn rabbits. The genomic regions surrounding the target site were PCR amplified and then subjected to Sanger sequencing and deep sequencing. Targeted sites were amplified from genomic DNA using Phusion polymerase (Thermo Fisher Scientific). The paired-end sequencing of PCR amplicons was performed by Sangon Biotech (Shanghai) using an Illumina MiSeq instrument. Sequences for all the primers used for genotyping are listed in Table S4.

Real-time quantitative PCR and western blotting
Total RNA was isolated with TRIzol-A+ reagent (TIANGEN, Beijing, China) according to the manufacturer's instructions. cDNA was synthesized with DNase I (Fermentas)-treated total RNA using the BioRT cDNA First Stand Synthesis Kit (Bioer Technology, Hangzhou, China). Quantitative real-time PCR was performed using the BioEasy SYBR Green I Real-Time PCR Kit (Bioer Technology, Hangzhou, China) with the Bio-Rad Iq5 Multicolour Real-Time PCR Detection System. Relative gene expression normalized to GAPDH expression was determined by the 2 ÀDDCT method. All gene expression data were obtained at least three times.
For western blotting, the ear tissues of pup #6 and WT rabbits were homogenized in 150 mL lysis buffer. The protein concentrations were measured by the Braford method (Bio-Rad). An anti-Lamin A/C polyclonal antibody (1:2,000; Proteintech) and an anti-b-actin monoclonal antibody (1:2,000; Proteintech) were used in this experiment.

Statistical analyses
All data are expressed as the mean ± SD, with at least three individual determinations carried out in all experiments. The data were analyzed with a two-tailed unpaired t test (both populations having the same SD) or an unpaired t-test with Welch's correction (no assuming equal SDs) using GraphPad Prism software 8.0. p < 0.05 indicated statistical significance (*p < 0.05, **p < 0.01, ***p < 0.001).

Availability of data and materials
Deep sequencing data from this work have been deposited in the Sequence Read Archive under accession code PRJNA768986 (https://www.ncbi.nlm.nih.gov/sra/PRJNA768986).