Multiple sgRNAs facilitate base editing-mediated i-stop to induce complete and precise gene disruption

Gene editing is a process to introduce desired changes into targeted loci of genomic DNA. Recently, type II clustered regularly interspaced short palindromic repeats-associated Cas9 endonuclease (CRISPR/Cas9) system has been demonstrated as a versatile tool for engineering eukaryote genome (Hsu et al., 2014), such as in mice (Zuo et al., 2017). CRISPR/Cas9-mediated genome editing is achieved by the error-prone DNA repair of non-homologous end joining (NHEJ) after double strand DNA cleavage. However, the editing results are unreliable due to uncontrolled random indels. Moreover, it was also occasionally reported that Cas9 may induce troublesome off-target effects (Hsu et al., 2013; Pattanayak et al., 2013; Cho et al., 2014). Scientists are making continuous efforts to modify and optimize the gene editing tools. In a landmark study, Komor et al. developed a ‘DNA’ base editor (BE), a novel genome editing tool which is applicable to change C/G base pairs to A/T without introducing DNA double strand breaks. Thereafter, various modifications have been created to the base editor system to improve its editing efficiency. BE3 can introduce C-to-T nucleotide substitution at the window of position 4–8 bases of the non-binding strand of the sgRNA (Komor et al., 2016). BE4max increased efficiency in a variety of mammalian cell types (Koblan et al., 2018). Interestingly, BE3 has been used to introduce early stop codon (TGA, TAG, TAA) from codons (CAA, CAG, CGA, TGG) to terminate gene expression (Kuscu et al., 2017), and provides a safer and much more precise knockout strategy than Cas9-mediated NHEJ (Kim et al., 2017; Komor et al., 2017). In this study, we used BE3 and BE4max to edit mouse genome by introducing stop codon (i-stop) in coding region of specific genes. We tested a multiple sgRNAs strategy and the results indicated that multiple sgRNAs dramatically increase the efficiencies of BE3-mediated and BE4maxmediated editing in mouse embryos and successfully generated DKO (double knockout) mice by BE3-mediated i-stop targeting Tyr and Pdcd1. First, mouse-derived Neuro-2a (N2a) cells were used as testing system. We designed and screened 14 sgRNAs targeting Tyr and Pdcd1 (7 sgRNAs for each gene), respectively. TYR gene encodes the tyrosinase enzyme, and its mutations result in impaired tyrosinase production leading to albinism (Witkop, 1979). PDCD1 is an immune checkpoint gene which guards against autoimmunity and regulatory T cells (Fife and Pauken, 2011). To screen the candidate sgRNA, BE3 plasmid and individual sgRNA were co-transfected into N2a cells. The results of chromatograms showed, 3 out of 7 sgRNAs for Tyr (Tyrsg1, Tyr-sg2 and Tyr-sg7 targeting exon1) and 3 out of 7 sgRNAs for Pdcd1 (Pdcd1-sg1, Pdcd1-sg2 and Pdcd1-sg3, targeting exon 1, exon 2 and exon 3, respectively) worked well in modifying the genome in coding regions (Fig. S1). We then test the editing efficiency of BE4max with these six sgRNAs (Fig. S3A). As expected, subsequent TA clone sequencing confirmed that all six sgRNAs introduced stop codon at the predicted sites with BE3 (Fig. S1D) or BE4max (Fig. S3B). For BE3, Tyr-sg1 and Tyr-sg2, Tyr-sg7 generated stop codon (Q48stop, W272stop, W12stop) at the frequencies of 13.3%, 22.2% and 14.3% respectively. On the other hand, Pdcd1-sg1, Pdcd1-sg2 and Pdcd1-sg3 generated stop codon (Q79stop, Q167stop and W12stop) at the frequencies of 10%, 37.5% and 20%. For BE4max, Tyr-sg1, Tyr-sg2 and Tyr-sg7 generated stop codon at the frequencies of 50%, 33.3% and 33.3%, respectively, while Pdcd1-sg1, Pdcd1sg2 and Pdcd1-sg3 generated stop codon at the frequencies of 30%, 50% and 33.2%. Thus, Tyr-sg1, 2, 7 and Pdcd1-sg1, 2, 3 were selected for the further study. We then attempted to test the efficiency of i-stop conversion in mouse embryos. To test multiple sgRNAs strategies, different combinations of sgRNAs and BE mRNA or BE4max mRNA were co-injected into zygotes (50 ng/μL BE mRNA and 25 ng/μL sgRNAs) (Fig. 1B). We successfully introduced stop codon (i-stop) with BE3. For Tyr, 2 out of 10 (20%) blastocysts (#5 and #9) for Tyr-sg1 harbored genomic modification of synonymous mutation (G47G), while 5 out of 8 (62.5%) blastocysts (#1–3, #6 and #7) harbored i-stop mutation for Tyr-sg1 combined with Tyrsg2, indicating that multiple sgRNAs can enhance i-stop introduction in embryos (Figs. 1C and S4A). For Pdcd1, 2 out of 10 (20%) (#5 and #6), 2 out of 8 (25%) (#1 and #7) and 5


Dear Editor,
Gene editing is a process to introduce desired changes into targeted loci of genomic DNA. Recently, type II clustered regularly interspaced short palindromic repeats-associated Cas9 endonuclease (CRISPR/Cas9) system has been demonstrated as a versatile tool for engineering eukaryote genome (Hsu et al., 2014), such as in mice (Zuo et al., 2017). CRISPR/Cas9-mediated genome editing is achieved by the error-prone DNA repair of non-homologous end joining (NHEJ) after double strand DNA cleavage. However, the editing results are unreliable due to uncontrolled random indels. Moreover, it was also occasionally reported that Cas9 may induce troublesome off-target effects (Hsu et al., 2013;Pattanayak et al., 2013;Cho et al., 2014).
Scientists are making continuous efforts to modify and optimize the gene editing tools. In a landmark study, Komor et al. developed a 'DNA' base editor (BE), a novel genome editing tool which is applicable to change C/G base pairs to A/T without introducing DNA double strand breaks. Thereafter, various modifications have been created to the base editor system to improve its editing efficiency. BE3 can introduce C-to-T nucleotide substitution at the window of position 4-8 bases of the non-binding strand of the sgRNA (Komor et al., 2016). BE4max increased efficiency in a variety of mammalian cell types (Koblan et al., 2018). Interestingly, BE3 has been used to introduce early stop codon (TGA, TAG, TAA) from codons (CAA, CAG, CGA, TGG) to terminate gene expression (Kuscu et al., 2017), and provides a safer and much more precise knockout strategy than Cas9-mediated NHEJ Komor et al., 2017).
In this study, we used BE3 and BE4max to edit mouse genome by introducing stop codon (i-stop) in coding region of specific genes. We tested a multiple sgRNAs strategy and the results indicated that multiple sgRNAs dramatically increase the efficiencies of BE3-mediated and BE4maxmediated editing in mouse embryos and successfully generated DKO (double knockout) mice by BE3-mediated i-stop targeting Tyr and Pdcd1.
First, mouse-derived Neuro-2a (N2a) cells were used as testing system. We designed and screened 14 sgRNAs targeting Tyr and Pdcd1 (7 sgRNAs for each gene), respectively. TYR gene encodes the tyrosinase enzyme, and its mutations result in impaired tyrosinase production leading to albinism (Witkop, 1979). PDCD1 is an immune checkpoint gene which guards against autoimmunity and regulatory T cells (Fife and Pauken, 2011).
We then attempted to test the efficiency of i-stop conversion in mouse embryos. To test multiple sgRNAs strategies, different combinations of sgRNAs and BE mRNA or BE4max mRNA were co-injected into zygotes (50 ng/μL BE mRNA and 25 ng/μL sgRNAs) (Fig. 1B).
Although founder #8 and #9 both harbored W272stop conversion at Tyr locus, none of the nine newborns (Experiment 1) displayed the albino phenotype of white skin indicating incomplete gene disruption (Fig. 1E). For those pups from BE3 combined with Tyr-sg1+2+7 and Pdcd1-sg1+2+3, 8 out of 11 newborns (Founder #1, #2, #4, #5, #15-18) displayed Tyr-deficient mice phenotype indicating the increased Tyr gene disruption efficiency mediated by multiple sgRNA i-stop strategy. Interestingly, all of the phenotypic mice show white skin over whole body instead of black-andwhite skin, the mosaic phenotype which displayed by mice harboring mosaic gene disruption of Tyr as previous reported (Zuo et al., 2017). These data further suggest multiple sgRNAs facilitated i-stop conversion and gene disruption, which allows phenotype analysis of founder animals.
We evaluated the on-/off-target effect though PCR-based deep sequencing. Using online tool (http://www.rgenome. net/cas-offinder/), we first selected 5 off-target sites for each sgRNA (Table S5). Off-target and on-target sites of six sgRNAs in this study were sequenced using tail DNA from four mice (#4, #16, #17 and #18), and 2 or 3 mice were analyzed for every sgRNA. Based on the sequencing results, no base substitution was detected at any off-target sites (30 sites in total) (Fig. 2C). To further explore the precision of BE3-mediated base editing, a WGS was performed using genomic DNA from two mutant mice (#16 and #18) and a wild-type mouse as the control at depth of about 24×. We analyzed a total of 7,234 sites, including 1 on-target site and 1,069, 2,414, 1,919, 175, 961, 690 off-target sites (with up to 3-nucleatide mismatch) on Tyr-sg1, Tyr-sg2, Tyr-sg7, Pdcd1-sg1, Pdcd1-sg2 and Pdcd1-sg3, respectively (Figs. 2E and S5G). Only the C-to-T substitution within the target window was observed (Fig. 2F).
Continuous modifications are being made to base editor ever since its discovery to improve the efficiency or precision. From BE1 to current BE4, different components were engineered in base editor system, resulting in steady improvement of editing efficiency. Unlike those biostructural modification, a new strategy was utilized in our study to improve editing efficiency. In summary, we utilized multiple sgRNAs to facilitate i-stop generation of two endogenous genes mediated by different base editor (BE3 and BE4max), resulting in efficient and multiple gene disruption. High throughout sequencing analysis showed multiple sgRNAs strategy facilitated editing is precise with minimal off-target effect and indel. Although the concentration of injected materials was not test, typical concentrations of base editor (50 ng/μL) and sgRNA (25 ng/μL) already achieved decent performance in zygote microinjection. Further studies could be carried out to titrate the injection concentrations and refine the protocol. Taken together, multiple sgRNAs is a universal strategy to achieve efficient gene knockout for phenotype analysis.