Improved Base Editor for Efficiently Inducing Genetic Variations in Rice with CRISPR/Cas9-Guided Hyperactive hAID Mutant

Single-nucleotide polymorphism (SNP) is an abundant form of genetic variation among individuals in plant species, which has been believed to have effects on the gene function responsible for phenotypic diversity. Utilization of functional SNPs associated with diverse important agronomic traits is of great value in crop improvements. Recently, a simple base editing system, which introduces a single-nucleotide change in genes of interest for generating gain-of-function mutants, has been developed in plants (Li et al., 2017Li J. Sun Y. Du J. Zhao Y. Xia L. Generation of targeted point mutations in rice by a modified CRISPR/Cas9 system.Mol. Plant. 2017; 10: 526-529Abstract Full Text Full Text PDF PubMed Scopus (201) Google Scholar, Lu and Zhu, 2017Lu Y. Zhu J.K. Precise editing of a target base in the rice genome using a modified CRISPR/Cas9 system.Mol. Plant. 2017; 10: 523-525Abstract Full Text Full Text PDF PubMed Scopus (274) Google Scholar, Ren et al., 2017Ren B. Yan F. Kuang Y. Li N. Zhang D. Lin H. Zhou H. A CRISPR/Cas9 toolkit for efficient targeted base editing to induce genetic variations in rice.Sci. China Life Sci. 2017; 33: 516-519Crossref Scopus (68) Google Scholar), in which Cas9(D10A) nickase is engineered with the rat cytosine deaminase (APOBEC1) and the uracil glycosylase inhibitor (UGI). Under the guidance of sgRNA, the Cas9n fusion protein is directed to the target locus in the genome, deaminating cytidines (C) in the editing window (13–19 bp before the PAM sequence) into uridines (U) and enabling the introduction of thymines (T), adenines (A) or guanines (G) during DNA replication (Li et al., 2017Li J. Sun Y. Du J. Zhao Y. Xia L. Generation of targeted point mutations in rice by a modified CRISPR/Cas9 system.Mol. Plant. 2017; 10: 526-529Abstract Full Text Full Text PDF PubMed Scopus (201) Google Scholar, Lu and Zhu, 2017Lu Y. Zhu J.K. Precise editing of a target base in the rice genome using a modified CRISPR/Cas9 system.Mol. Plant. 2017; 10: 523-525Abstract Full Text Full Text PDF PubMed Scopus (274) Google Scholar, Ren et al., 2017Ren B. Yan F. Kuang Y. Li N. Zhang D. Lin H. Zhou H. A CRISPR/Cas9 toolkit for efficient targeted base editing to induce genetic variations in rice.Sci. China Life Sci. 2017; 33: 516-519Crossref Scopus (68) Google Scholar). However, consistent with the observation that the AID/APOBEC family members have differential sensitivities to the 5′ nucleotide preceding the target C (Beale et al., 2004Beale R.C.L. Petersen-Mahrt S.K. Watt I.N. Harris R.S. Rada C. Neuberger M.S. Comparison of the differential context-dependence of DNA deamination by APOBEC enzymes: correlation with mutation spectra in vivo.J. Mol. Biol. 2004; 337: 585-596Crossref PubMed Scopus (281) Google Scholar, Komor et al., 2016Komor A.C. Kim Y.B. Packer M.S. Zuris J.A. Liu D.R. Programmable editing of a target base in genomic DNA without double-stranded DNA cleavage.Nature. 2016; 533: 420-424Crossref PubMed Scopus (2430) Google Scholar, Komor et al., 2017Komor A.C. Zhao K.T. Packer M.S. Gaudelli N.M. Waterbury A.L. Koblan L.W. Kim Y.B. Badran A.H. Liu D.R. Improved base excision repair inhibition and bacteriophage Mu Gam protein yields C:G-to-T:A base editors with higher efficiency and product purity.Sci. Adv. 2017; 3: eaao4774Crossref PubMed Scopus (399) Google Scholar), APOBEC1 shows a striking preference for TC, and the editing efficiency of GC, in general, is quite low. To broaden the application of base editing technology in rice, which is characterized by high GC content, here we further optimize our rBE system with human AID (hAID), another type of cytosine deaminase with a priority on GC and AC (Beale et al., 2004Beale R.C.L. Petersen-Mahrt S.K. Watt I.N. Harris R.S. Rada C. Neuberger M.S. Comparison of the differential context-dependence of DNA deamination by APOBEC enzymes: correlation with mutation spectra in vivo.J. Mol. Biol. 2004; 337: 585-596Crossref PubMed Scopus (281) Google Scholar), to introduce desired point mutations in the target region for gene correction or modification as a proof of concept. The full-length hAID and Petromyzon marinus CDA1 (PmCDA1) were identified with nearly no cytidine deaminase activity detected in vitro (Komor et al., 2016Komor A.C. Kim Y.B. Packer M.S. Zuris J.A. Liu D.R. Programmable editing of a target base in genomic DNA without double-stranded DNA cleavage.Nature. 2016; 533: 420-424Crossref PubMed Scopus (2430) Google Scholar), and with less efficiency of base editing in rice (Shimatani et al., 2017Shimatani Z. Kashojiya S. Takayama M. Terada R. Arazoe T. Ishii H. Teramura H. Yamamoto T. Komatsu H. Miura K. et al.Targeted base editing in rice and tomato using a CRISPR-Cas9 cytidine deaminase fusion.Nat. Biotechnol. 2017; 35: 441-443Crossref PubMed Scopus (432) Google Scholar). In this case, the K10E/T82I/E156G hAID variant lacking the nuclear export signal sequence (termed hAID*Δ) was chosen for its improved catalytic activity (Wang et al., 2009Wang M. Yang Z.Z. Rada C. Neuberger M.S. AID upmutants isolated using a high-throughput screen highlight the immunity/cancer balance limiting DNA deaminase activity.Nat. Struct. Mol. Biol. 2009; 16: 769-776Crossref PubMed Scopus (62) Google Scholar, Hess et al., 2016Hess G.T. Fresard L. Han K. Lee C.H. Li A. Cimprich K.A. Montgomery S.B. Bassik M.C. Directed evolution using dCas9-targeted somatic hypermutation in mammalian cells.Nat. Methods. 2016; 13: 1036-1042Crossref PubMed Scopus (292) Google Scholar). The gene fragment of hAID*Δ was codon optimized for expression in rice (Supplemental Figure 1) and attached to the 5′ terminal end of Cas9n-NLS using XTEN linker (Figure 1A and 1B ). Therefore, it is likely that the editing window of hAID*Δ would be restricted to the sgRNA-targeting site similar to the case of APOBEC1. UGI was excluded at this time since more types of nucleotide conversions, produced in uracil N-glycosylase (UDG)-initiated base excision repair (BER), were anticipated. The hAID*Δ-XTEN-Cas9n-NLS chimeric gene (termed rBE5) was first tested in rice leaf sheath protoplast, with the sgRNAs targeting a GCAC-containing ApaLI restriction site in OsRLCK185 and a TCC-containing BamHI restriction site in OsCERK1. Sequencing data showed that distinct mutations with high frequency of C > T conversions occurred in the editing window (Supplemental Figure 2A–2H), suggesting rBE5 functions on GC, AC, TC as well as CC in rice cells. As a follow-up to our previous study on the development of the rBE3 system (Ren et al., 2017Ren B. Yan F. Kuang Y. Li N. Zhang D. Lin H. Zhou H. A CRISPR/Cas9 toolkit for efficient targeted base editing to induce genetic variations in rice.Sci. China Life Sci. 2017; 33: 516-519Crossref Scopus (68) Google Scholar), in this study the rBE3 gene in the binary vector pUbi:rBE3 was replaced by the rBE5 gene, namely pUbi:rBE5, and then further tested for efficiency in stable transgenic rice plants through Agrobacterium-mediated transformation. Pi-d2, an agriculturally important rice blast R gene in which a GC-containing region was resistant to rBE3-mediated editing in our earlier experiment, was first chosen for targeting (Figure 1C). As reported previously, a single amino acid substitution at position 441 in the recessive allele of Pi-d2 gene results in loss of resistance to Magnaporthe oryzae (Chen et al., 2006Chen X. Shang J. Chen D. Lei C. Zou Y. Zhai W. Liu G. Xu J. Ling Z. Cao G. et al.A B-lectin receptor kinase gene conferring rice blast resistance.Plant J. 2006; 46: 794-804Crossref PubMed Scopus (397) Google Scholar). Therefore, we assumed that introducing a G > A mutation (M441I) in endogenous pi-d2kit using rBE5 could recover its biological function. The same protospacer for rBE3 was used in this experiment; 26 independent transgenic lines of Kitaake were obtained after transformation and genotyped subsequently by Sanger sequencing. Eight heterozygous lines (30.8% efficiency) carrying a desired single G to A conversion at position −17 upstream of PAM were identified (Figure 1D and 1E; Supplemental Figure 3A), suggesting that rBE5 is much more efficient on target C that immediately follow a G than rBE3. We assumed that the successful gene correction of pi-d2 would be attributable to both the specificity and the improved activity of hAID*Δ. Remarkably, pi-d2 alleles with Indel mutations were also detected in five lines (Supplemental Figure 3B and 3C). In light of robust biochemical activity of hAID*Δ, we speculated that rBE5 might induce various point mutations of C2843 in OsFLS2, the rice homolog of the Arabidopsis flagellin-sensing receptor kinase gene FLS2, possibly leading to replacement of phosphorylation-blocking residue Ala948 by phosphorylation-mimic residues Asp or Glu (Figure 1F), and rendering rice plants more defensive to bacterial pathogens. As shown in Figure 1G–1I and Supplemental Figure 4A and 4B, of 114 independent transgenic lines confirmed by Sanger sequencing, 65 homozygous or heterozygous lines (57.0% efficiency) were identified with nucleotide changes in the OsFLS2 kinase domain; most of genomic variations were located in the editing window (position −18 to −14) upstream of PAM. rBE5 converted C to the other three bases, but more notably to T. In addition, a low level of G > C conversion in the editing window and C > A conversion near the target region were also found in our analysis. Notably, cytidine conversions occurred in the sequence context characterized by GC as well as CC. Finally, one heterozygous line in which the OsFLS2 allele carried C2834A, C2843A, and C2847T conversions (Figure 1G), encoding phosphorylation-mimic residues A945E, A948D, and unchanged G949, respectively, was identified as expected. To our surprise, 82 lines (71.9% frequency) were identified with mono- or bi-allelic Indel mutations, suggesting rBE5 is more active in inducing Indel mutations in rice than rBE3 and rBE4. This would not be problematic because Indels and nucleotide conversions resided on separate chromosomes in most cases (Figure 1I and Supplemental Figure 5), and will segregate during gametes production. Further sequence data analysis of 196 OsFLS2 alleles with genetic variations showed that the majority of Indel mutations occurred in the editing window, likely resulting from the unfaithful BER of the uracil through nonhomologous end joining (NHEJ). Therefore, cytidine deamination activity of rBE5 might be responsible for the genomic variations that occurred in 78.6% of OsFLS2 alleles. UGI from Bacillus subtilis bacteriophage PBS1 is capable of blocking UDG activity. It potentially subverts BER at the site of base editing (Komor et al., 2016Komor A.C. Kim Y.B. Packer M.S. Zuris J.A. Liu D.R. Programmable editing of a target base in genomic DNA without double-stranded DNA cleavage.Nature. 2016; 533: 420-424Crossref PubMed Scopus (2430) Google Scholar). Thereby, we fused UGI gene to the 3′ terminal of rBE5, resulting in rBE9, in order to further improve the base editing efficiency and outcome fidelity. As shown in Figure 1K, unwanted Indel mutations were drastically inhibited in the presence of UGI. However, Indel occurrence was positively correlated with the base editing efficiency, presumably due to the great performance of hAID*Δ at certain loci. rBE3 was also tested side by side with rBE9 to evaluate its effects on base editing in different sequence contexts at four different chromosomal sites (OsAOS1, OsJAR1, OsJAR2, and OsCOI2) in rice transgenic calli. Detailed sequence analysis (Figure 1K and 1L; Supplemental Figures 6–9) indicated that the base editing efficiencies of rBE9 were comparable with or more enhanced than that of rBE3. Notably, rBE9 functioned more robustly on GC and more efficiently on multiple target C in the editing window of sgRNA than rBE3, resulting in more genetic variation at the target loci. Next, we genotyped the T1 progeny plants to investigate whether the nucleotide changes introduced by rBE5 could be transmitted to the next generation of the selfed T0 plants. Direct sequencing analysis of pi-d2 or OsFLS2 in 170 individual T1 progeny plants showed that homozygous plants with either edited nucleotides or Indels were all identified (Supplemental Figure 10). Furthermore, transgene-free plants harboring stable DNA changes were also isolated through detecting the presence of T-DNA (Cas9, hptII and sgRNA genes) by PCR amplification (Supplemental Figure 10). We recently reported a CRISPR/Cas9 toolkit containing pUbi:rBE3 and pUbi:rBE4 for efficient targeted base editing to induce genetic variations in rice (Ren et al., 2017Ren B. Yan F. Kuang Y. Li N. Zhang D. Lin H. Zhou H. A CRISPR/Cas9 toolkit for efficient targeted base editing to induce genetic variations in rice.Sci. China Life Sci. 2017; 33: 516-519Crossref Scopus (68) Google Scholar). In this study, we have further expanded the toolkit with pUbi:rBE5 and pUbi:rBE9, which harbor an engineered hyperactive hAID mutant version. Distinct from the rat APOBEC1-containing rBE3 and rBE4, which exhibit high deaminase activity on TC, the hAID*Δ-based rBE5 and rBE9 proteins favor GC, and function on AC, TC, and CC as well. Therefore, the pUbi:rBE5 and pUbi:rBE9 vector would be much more useful for targeted point mutation in rice in terms of the high GC content of the rice genome. In addition, compared with the undetected in vitro deaminase activity of full-length hAID (Komor et al., 2016Komor A.C. Kim Y.B. Packer M.S. Zuris J.A. Liu D.R. Programmable editing of a target base in genomic DNA without double-stranded DNA cleavage.Nature. 2016; 533: 420-424Crossref PubMed Scopus (2430) Google Scholar) and poor performance of full-length PmCDA1 in rice (Shimatani et al., 2017Shimatani Z. Kashojiya S. Takayama M. Terada R. Arazoe T. Ishii H. Teramura H. Yamamoto T. Komatsu H. Miura K. et al.Targeted base editing in rice and tomato using a CRISPR-Cas9 cytidine deaminase fusion.Nat. Biotechnol. 2017; 35: 441-443Crossref PubMed Scopus (432) Google Scholar), AID*Δ shows highly enhanced base editing efficiency in generating both gain-of-function and loss-of-function mutants of rice. Therefore, we assume AID*Δ could be suitable for targeted base editing in other monocots and dicots as well. Overall, this base editor toolkit is powerful and has potential to be widely applied in both biological studies and crop molecular breeding in the future. B.R., F.Y., Y.K., D.Z., X.Z., H.L., and H.Z. designed the experiments; B.R., F.Y., Y.K., N.L., X.Z., and H.Z. performed experiments; B.R., F.Y., D.Z., X.Z., H.L., and H.Z. wrote the paper. We thank Dr. Li Yuan for proofreading. This study was supported by grants from the National Key R&D Program of China (2017YFD0200900), the Agricultural Science and Technology Innovation Program of The Chinese Academy of Agricultural Sciences to H.Z., and a grant from the National Natural Science Foundation of China (31701780) to F.Y. The authors have filed a patent application based on the results reported in this study. Download .pdf (2.48 MB) Help with pdf files Document S1. Supplemental Methods, Supplemental Figures 1–10, and Supplemental Tables 1 and 2