Hunhun Wng,Jing Ling,Like Chen,Bufng Deng,Dongfng Gu,Xioshung Liu,Shn Jin,Rongfng Xu,Ruiying Qin,Yitong Zhu,Lingxi Zho,Dourong Kou,Ynjun Chen,Yingli Jing,Jun Li,,Pengcheng Wei,

a College of Agronomy,Anhui Agricultural University,Hefei 230036,Anhui,China

b Key Laboratory of Rice Genetic Breeding of Anhui Province,Rice Research Institute,Anhui Academy of Agricultural Sciences,Hefei 230031,Anhui,China

Keywords: Cas12a Base editing Rice ABE CBE

ABSTRACT Base editors of the Cas9 system have been widely used for precise nucleotide substitution in crops.In this study,Cas12a was applied to construct plant cytosine base editors(CBEs).The main elements of Cas12a-CBEs were engineered and their efficiency was evaluated in stably transformed rice cells.An optimized ttCas12a-hyA3Bctd editor,consisting of a LbCas12a variant carrying catalytic inactive D832A and temperature-tolerance D156R double mutations,a truncated human APOBEC3B deaminase,a human RAD51 single-stranded DNA-binding domain,and double copies of UGI,outperformed other Cas12a-CBEs in base editing efficiency.In T0 transgenic rice plants,ttCas12a-hyA3Bctd edited an average of 42.01% and a maximum of 68.75% of lines at six genomic targets.A-to-G conversions were generated in rice by an adenine base editor with a similar architecture to the optimized CBE.Our results provide preliminary evidence for the feasibility of robust and efficient plant Cas12a base editing systems,which could be useful for precise crop breeding.

1.Introduction

CRISPR/Cas genome editing technology is revolutionizing plant research and crop breeding [1].Typical Cas9 and Cas12a CRISPR tools produce DNA double-strand breaks (DSBs) to edit targeted genes.In plants,DSBs are repaired mainly via the error-prone nonhomologous end joining(NHEJ)pathway and often produce knockout mutations by introducing nucleotide deletions or insertions(InDels).Although precise mutations can be introduced from a donor template via the homology-directed repair (HDR) pathway,the process suffers from low efficiency,owing mainly to weak HDR activity and donor delivery barriers in plant cells [2].Base editing is an alternative system for precisely engineering nucleotides without requiring DSB and donor templates.Base editors(BEs) share a similar structure,containing a catalytically impaired Cas endonuclease and a single-stranded DNA deaminase [3].Most plant BEs have been developed from RuvC domain-inactivated Cas9 mutants [4,5].These nickases activate a DNA mismatchrepair mechanism to excise the target strand complementary to the edited strand,thus increasing the installation frequency of desired base substitutions.Current BEs,including cytosine base editors (CBEs) and adenine base editors (ABEs),have been extensively optimized to achieve highly efficient conversion between C·G and T·A base pairs in the plant genome[4,5].Recently,glycosylase base editors(GBEs)have been reported[6-8]to confer C·Gto-G·C transversion in crops and tree cells,but further improvements in accurate editing efficiency are still needed.

Cas12a is a class of well-characterized CRISPR RNA (crRNA)-guided nucleases that have some unique features.Unlike conventional Cas9,which normally induces 1-3-bp InDels proximal to the G-preferred protospacer-adjacent motif (PAM),Cas12a tools generate larger deletions distal to Cas9-inaccessible T-rich regions[9].Because of its pre-crRNA processing capability,Cas12a has been exploited for simplified multiplexed genome modifications and RNA template-mediated HDR gene replacement in plants[10].Although there is a lack of nickase mutants to date,there have been some efforts to render Cas12a a specific base editing platform.Several CBEs were established from a catalytically inactive version of Cas12a [11-14].A set of Cas12a-CBEs,BEACONs,show editing efficiency comparable to that of SpCas9-BEs in mammalian cells and induce base editing in mouse embryos and offspring [14].In contrast,Cas12a CBE-mediated C-to-T conversions have not yet been reported in plants,although adenine base editing has been induced in rice and cotton at a few genomic targets with low frequencies [15,16].

In this study,CBEs with various architectures were engineered from Cas12a.Efficient programmable base conversions were obtained with high product purity in transgenic rice.Our results suggest that Cas12a could provide an alternative platform for developing precise genome editing tools in plants.

2.Materials and methods

2.1.Vector construction

To generate Cas12a-BE3,the D832A mutation was introduced into LbCas12a from pHUN611 [17].The rAPO1 and UGI components were amplified from the eBE3 vector for assembly with dCas12a through Gibson cloning [18].Anc689,evoFERNY,and A3Bctd were codon-optimized for rice expression (Supplementary sequences) and synthesized (Genscript,Suzhou,China).A3A from SpG-A3A was applied to generate the N57G mutation of eA3A[19].Human RAD51 single-stranded DNA binding domain (DBD)was amplified from Nm2-hyBE3 [20].In the hyperactive base editor(hyBEs)architecture,Cas12a-hyCBEs were built by sequentially assembling several deaminases,the DBD,and the dCas12a-UGI fragment.Gibson assembly was applied to fuse TadA8 from SpGABE8e and dCas12a [19].Cas12a-ABE8e was obtained by addition of triplet copies of SV40 NLS at the 3′terminus of dCas12a by direct PCR.Amplifications were performed with Q5 High-Fidelity DNA polymerase (NEB,Ipswich,USA).A site-directed mutagenesis kit was applied to precisely induce point mutations (Transgene,Beijing,China).An NEBuilder HiFi DNA Assembly kit was used for fragment assembly.The cloned,mutated and assembled sequences of the components or full-length BEs were confirmed by Sanger sequencing before further vector construction (Sangon Biotech,Shanghai,China).Cas12a-BEs were then separately inserted into the pHUC backbone viaPstI/SacI double digestion.To build an expression cassette of the crRNA,a CmYLCV promoter was cloned from pRN120 (Addgene #160696) [21].The tRNA and HDV ribozymes were separately designed in the long primers of a spectinomycin resistance(SpR)gene.Finally,the promoter,tRNA-SpR-HDV element and a poly T terminator were assembled as the crRNA expression component and inserted into the pHUC-Cas12a-BE vectors byHindIII digestion.To construct Cas12a-BEs for specific genome targets,the forward and reverse oligos of the 23-nt guide RNA sequence were annealed.Together with the 4n96-type crRNA scaffold,the crRNA was inserted into the binary vector to replace SpR via theBsaI-mediated GoldenGate cloning method.Following previously described procedures [22],crRNA-containing clones were selected positively with kanamycin and negatively with spectinomycin and then confirmed by Sanger sequencing.

2.2.Plant transformation

The vectors were introduced intoAgrobacteriumstrain EHA105 via the freeze-thaw method [23].The clones were identified by sequencing the regions of crRNA and base editor before plant transformation.

Rice transformation was performed following a previously established protocol with modifications [24].Mature rice seeds were dehulled and sterilized.The embryos were isolated to induce calli for 3 to 4 weeks.After infection withAgrobacterium,350-400 solid and yellowish calli per transformation were selected under 50 mg L-1hygromycin pressure for 4 weeks.Approximately 200 resistance events were selected for plant regeneration.Only one plant from an independent event was chosen for rooting and genotyping.All plant materials were grown at 28 °C with a 16 h light and 8 h dark lighting cycle.

2.3.Sampling and genotyping

To quantify the editing efficiency,rice calli were separately infected by three independentAgrobacteriumclones of each vector.After 10-14 days of growth on selection medium,approximately 200 newly emerged resistant calli of one transformation were collected as a biological replicate.The samples were ground in liquid nitrogen to extract genomic DNA with a Hi-DNAsecure Plant Kit(Tiangen,Beijing,China).The target was amplified with sitespecific primers and deep sequencing with an Illumina Hi-seq platform with a paired-end sequencing length of 150 nt (PE150) at Sangon Biotech Corporation (Shanghai,China).At least 50,000 clean reads were generated per sample.Amplicon nextgeneration sequencing (NGS) data have been assigned BioProject accession PRJNA858397.

In 48 randomly selected regenerated plants for each vector,leaves of one independent line were sampled from at least three different tillers.Genomic DNA was prepared following a modified hexadecyltrimethylammonium bromide (CTAB) method [25] and diluted to 1 μg μL-1as a template for PCR.Amplicon NGS-based genotyping was performed with the Hi-TOM assay protocol [26]with a 5% threshold.All primers used in the study are listed in Table S1.

3.Results

To construct the plant Cas12a-BE3,the rat cytidine deaminase APOBEC1 (rAPO1) and two copies of uracil glycosylase inhibitor(UGI)were fused to the N-and C-termini of aLachnospiraceae bacteriumCas12a D832A variant (nuclease dead version LbCas12a,dCas12a).To assess the editing capability,Cas12a-BE3 was expressed by maize ubiquitin promoter 1 in a pHUC binary vector backbone (Fig.1A).A 4n96-type scaffold,which increases Cas12a activity [27],was employed with 23 nt guide sequences.The crRNAs were expressed with the RNA polymerase II (Pol II) promoter CmYLCV.The tRNA and HDV ribozyme was added at the terminus of crRNA for accurate processingin vivo.Two sites harboring multiple cytosines in the nontarget strand were selected in theOsPDSgene as targets.AfterAgrobacterium-mediated transformation,transgenic rice (Oryza sativaL.spp.japonica,var.Nipponbare)calli were collected to identify the targeted mutations.Amplicon-NGS data showed that only respectively 1.66% and 3.59% of reads carried the desired nucleotide substitutions at PDS-T1 and PDST2 (Fig.1B).When stable transgenic plants were regenerated by retransformation,no edited rice was detected at either target(Fig.2A).These observations suggested that the activity of Cas12a-BE3 might be severely restricted in plants compared to BE with similar architecture in human cells [13].Integrating DBD domain in hyBEs of Cas9 increases the affinity of the deaminase to the single-stranded DNA substrate,thus increasing editing efficiency and extending the editing window[28].Inspired by successful applications of hyBEs in plants[29],Cas12a-BE3 was refined by inserting DBD between rAPO1 and dCas12a to obtain Cas12ahyBE3,and results showed that the editing efficiency of Cas12ahyBE3 was 2.66-fold higher than that of Cas12a-BE3 at PDS-T1(Fig.1B,P＜0.05).A T0line was edited by Cas12a-hyBE3 at PDST2 (Fig.2A),revealed the capability of Cas12a base editors in plants.

Fig.1.Base editing mediated by Cas12a-CBEs in rice calli.(A)Schematic illustration of the CBEs developed from LbCas12a.Two types of short peptide,SGGSGGSGGS linker(cyan)and a 32-aa linker(green),were employed to fuse the components of base editors.NLS,nuclear localization signal.(B)Editing activity analysis of Cas12a-CBEs by NGS at PDS-T1 and PDS-T2 in rice calli.For each sample,approximately 200 newly emerged independent calli after two-week selection were collected for amplicon NGS.The ratio of edited reads to total reads was calculated from three independent biological replicates.Significance was determined by a two-tailed t-test.*,P ＜0.05,**,P ＜0.01.(C)Heat map of cytosine base editing mediated by Cas12a-CBEs at PDS-T1 and PDS-T2.The editing frequency at each position was calculated from NGS data.PAM sequences are labeled in blue.Editable nucleotides in the protospacer region are labeled in orange.RS,reverse-complement sequence of the spacer.(D)Representative mutations induced by ttCas12a-hyA3Bctd at PDS-T1 and PDS-T2.The three frequently occurring types of targeted mutation in NGS assays are listed.The 23-bp protospacer region is labeled in blue,and the PAM is underlined.WT,wild type;edited bases are labeled with red lowercase letters.At the far right,the frequency of each mutation was averaged over three independent repetitions.

Fig.2.Base editing mediated by Cas12a-CBEs in T0 transgenic lines.(A) Editing of Cas12a-CBEs at PDS-T1 and PDS-T2 in regenerated T0 rice lines.For each vector,48 independent T0 transgenic plants were randomly selected and genotyped by Hi-TOM assay with a 5% threshold.The ratio of the zygosity is indicated.WT,wild type;He,heterozygous;Ch,chimeric;Ho/Bi,homozygous and biallelic mutants.(B) Sequencing chromatogram of a representative line edited by ttCas12a-hyA3Bctd at PDS-T1 and PDS-T2.C·G-to-T·A conversions are labeled with red arrows.Protospacer and PAM regions are underlined by solid and dashed lines,respectively.(C)Frequencies of targeted base conversions induced by ttCas12a-hyA3Bctd at PDS-T1 and PDS-T2 in rice plants.The ratio of the number of lines carrying the base conversion at the indicated site to the number of examined lines was calculated.

To test highly active Cas12a-derived CBEs in plants,Cas12ahyANC,-hyFERNY,-hyeA3A,and -hyA3Bctd were created from Cas12a-hyBE3 by replacing rAPO1 with the rice codon-optimized version of an ancestral edition of cytidine deaminase Anc689,an evolved variant evoFERNY,human APOBEC3A with the N57G mutation (eA3A),and truncated human APOBEC3B (A3Bctd),respectively.Generally,the Anc689 and evoFERNY base editors of SpCas9 variants have greater activity than that of rAPO1 [30,31].However,Cas12a-hyANC showed editing efficiencies similar to those of Cas12a-hyBE3 at PDS-T1 (3.62% vs.4.41%) and PDS-T2(4.97% vs.5.52%) in transgenic calli (Fig.1B,P＜0.05).Cas12ahyFERNY slightly increased the substitution efficiency compared with Cas12a-hyBE3 by 1.67-fold toward PDS-T1,whereas the efficiency was not increased by Cas12a-hyFERNY at PDS-T2.Thus,dCas12a may not be as compatible as the nCas9 variants with Anc689 and evoFERNY for plant cytosine base editing.Cas12ahyeA3A had analogous limited frequencies with Cas12a-hyBE3 at the targets (Fig.1B).Given that the eA3A-CBE-V01 of nSpCas9 showed low efficiencies in rice protoplasts as well[32],we speculated that eA3A would be less suitable for developing highly efficient plant CBEs.A3Bctd showed a similar editing profile but increased activity with respect to rAPO1 in animals and plants[33].The C-to-T conversion frequencies of Cas12a-hyA3Bctd were 2.86-and 6.05-fold higher than those of Cas12a-hyBE3 at PDS-T1 and PDS-T2,respectively.Cas12a-hyA3Bctd also outperformed Cas12a-hyANC,Cas12a-hyFERNY and Cas12a-hyeA3A at the sites(P＜ 0.05).Although the efficiency of C-to-T conversions was increased by Cas12a-hyA3Bctd,unintended edits,including InDels and undesired base substitutions,remained at a low level (＜0.42%),similar to that of the other editors.The editing of the hyCBEs of Cas12a was further evaluated in regenerated transgenic plants.The sequencing of independent lines showed that mutant frequencies induced by Cas12a-hyA3Bctd were markedly higher than those induced by other Cas12a-CBEs,including 27.08%versus 0% at PDS-T1 and 50.00% versus ＜8.33% at PDS-T2 (Fig.2A).In agreement with the increased activity of Cas12a-hyA3Bctd in calli,the results suggest that A3Bctd has high compatibility with Cas12a in plants.

Previously [34],a temperature-tolerant LbCas12a (ttCas12a)variant was engineered for efficient targeted mutagenesis in plants cultivated at low temperature.The temperature-tolerant D156R mutation of Cas12a was applied to increase Cas12a-A3Bctd activity.In the transformed cells,the editing efficiencies of ttCas12ahyA3Bctd reached 19.90% and 46.25%,representing 1.58-and 1.39-fold that of Cas12a-hyA3Bctd at PDS-T1 and PDS-T2,respectively (Fig.1B,P＜ 0.05).Most substitutions mediated by ttCas12a-hyA3Bctd occurred at positions 10 to 15 (counting the first nucleotide of crRNA distal to PAM as position 1),showing an editing window similar to those of the CBEs constructed in the present study (Fig.1C).Detailed analysis showed that multiple nucleotides in the window could be simultaneously edited(Fig.1D),confirming the high activity of ttCas12a-hyA3Bctd in plant genome modification.To further evaluate ttCas12ahyA3Bctd,six genomic sites,including PDS-T1 and PDS-T2,were chosen and examined in independent T0lines.It was found that 16.67%to 68.75%of transgenic plants(on average 42.01%of plants)were edited at the targets (Fig.2A;Table 1),all of which carried desired single-or multiple-base(s) C·G-to-T·A substitutions but not unwanted byproducts (Figs.2B,S1;Table 1),suggesting that the product purity of the base editor remained high along with the increase in deaminase activity.Homozygous and biallelic mutants were obtained at two of six sites (Fig.2A;Table 1).In agreement with the window range determined from amplicon NGS on calli,editing occurred at nucleotides at positions 10 to 15(Figs.2C,S2).The off-target effect of ttCas12a-hyA3Bctd was further examined in transgenic plants.The three most likely offtarget endogenous sites with 1-5 mismatches in the spacer were identified by Cas-OFFinder [35] and examined by site-specific sequencing,and potential off-target mutations were not detected in any plants (Table S2).We conclude that ttCas12a-hyA3Bctd is an ideal architecture that substantially increased the efficiencies of plant Cas12a-CBE.

Table 1Cas12a-BE-mediated base editing in T0 lines.

To test whether the improved architecture of CBE contributed to high ABE activity,ttCas12a-hyABE8e was constructed by adopting the TadA-8e,DBD and Cas12a D832A/D156R (ttdCas12a) variants (Fig.3A).Three endogenous sites were selected at theOsPDSandOsACC1genes to evaluate the activity of ttCas12a-hyABE8e.Sequencing of transgenic plants showed that respectively 54.17%,10.42%,and 14.58% of T0lines harbored A·T-to-G·C conversions at the PDS-T4,PDS-T5,and ACC-T2 sites (Figs.3B,S3).Among them,most edited lines were heterozygous (Table 1).Only single-base conversion was obtained at one nucleotide or two separate nucleotides between positions 12 and 16,suggesting a restricted editing range of ttCas12a-hyABE8e (Figs.3C,S4).

Fig.3.Adenine base editing mediated by ttCas12a-hyABE8e in rice.(A) Schematic illustration of ttCas12a-hyABE8e.TadA-8e (red column),DBD (yellow),and ttdCas12a(blue)were fused with 32 aa linkers(green).Single and triple copies of SV40 NLS(gray)were attached to the N and C termini of the fusion,respectively.(B)Representative edited plants induced by ttCas12a-hyA3Bctd at PDS-T4.Top,sequence alignment for the representative transgenic line;bottom,sequencing chromatogram of the edited line.The protospacer and PAM are labeled by a solid line and a dashed line,respectively.The red arrow indicates targeted nucleotide substitutions in the protospacer region.(C)Frequencies of targeted base conversions induced by ttCas12a-hyABE8e at PDS-T4.The ratio of the number of lines carrying the base conversion at the indicated site to the number of examined lines was calculated.

4.Discussion

Given that the CRISPR-Cas12a system is an editing tool broadly complementary to Cas9,it is highly desirable to develop precise genome editors with Cas12a.Unlike base editors of Cas9 nickase,BEs of Cas12a exploited its catalytic inactive variant.The absence of cleavage activity on the unedited strand thus may lead to lower editing efficiency of Cas12a-derived BEs than of Cas9 in mammalian cells [11,13].In plants,even though the CBEs and ABEs of Cas12a were improved in this study,their efficiencies generally remained low compared to those generated by deeply optimized plant nCas9-BEs with the same deaminase [29-32].During cytosine base editing,the uracil deaminated from targeted cytidine could be repaired to form C-to-T conversion [3].However,uracil can also be transformed into an apurinic/apyrimidinic (AP) site by uracil DNA glycosylase (UDG),which results in InDels by triggering NHEJ repair or unintended C-to-A or C-to-G substitutions through translesion synthesis over the AP site [3].These byproducts occur much more frequently in BEs with the nickase of Cas9 than in the dead version [36],possibly owing to the high risk of forming DSBs in the DNA repair process.A previous study [14] in mammalian cells showed that dCas12a-BEs induced barely detectable DSBs,triggering a minimal DNA damage response in cells.In this study,the byproducts of the dCas12a-BEs were fully avoided in transgenic plants(Table 1),confirming the specificity of the editing.The finding that the dCas9-derived BE2 showed limited base editing efficiency compared with the dCas12a-BEs [14] suggests the advantage of Cas12a for developing useful complements of conventional Cas9 base editing tools.

To optimize the editing activity of Cas12a-CBEs,various highly efficient cytosine deaminases were tested in the constructs.Screening of CBEs of the SpCas9 variant suggested that evoFERNY outperformed other deaminases overall,including A3A and an upgraded version of rAPO1 [31].However,the similar activities of Cas12a-hyFERNY and Cas12a-hyBE3 imply the lower compatibility of evoFERNY with the Cas12a system.A highly efficient Cas12a-derived base editor BEACON system was developed with A3A and conferred comparable activity to the widely used AncBE4-max in human cells[14].In the present study,we applied an eA3A variant with minimized bystander and off-target activities [37].Although eA3A-BE3 provides sufficient activity for editing,Cas12a-hyeA3A showed a negligible increase compared to Cas12a-hyBE3 even in its most favorable TC context.We also tried to use the human APOBEC3A/Y130F variant,which showed high editing efficiency in the Cas9-based CBE system in mammalian cells and rice[32,38].However,potential cell toxicity might change the construction,given that Cas12a-hyA3A/Y130F clones in replicated experiments carried random point mutations (data not shown).In rice protoplasts,A3Bctd-BE3 displayed comparable editing efficiency in a more limited window than A3A-BE3 [33].The finding that A3Bctd exhibited better performance than the other cytosine deaminases suggest the high compatibility of A3Bctd for developing Cas12a-CBEs.

Generally,Cas12a has less off-targeting activity than Cas9 [39-41].It has been reported [13,42] that Cas12a-BE has higher specificity than Cas9-derived BE in human cells.Our finding that no mutation was detected at closely matched genomic sites in transgenic plants of ttCas12a-hyA3Bctd (Table S2) confirms the low sgRNA-dependent off-target effects of the Cas12a-CBEs in plants.SgRNA-independent off-target mutations of CBEs have also been found in rice,mouse,and human cells [43-45],owing to the single-stranded DNA binding affinity of cytosine deaminase.Recently [46,47],several variants have been engineered from TadA-8e for highly efficient C·G-to-G·C/T·A conversion to form Cas9-CBEs with minimum levels of sgRNA-independent DNA and RNA off-target activity.Combining the TadA-8e variants and the experience obtained in this study may lead to more robust and specific Cas12a-base editing in the plant genome in the near future.

Various Cas12a orthologs have been engineered for plant genome editing with differing PAM preferences [48].In future,it would be worth developing Cas12a-BEs with active orthologs or evolved variants in our optimized architecture to increase the efficiency and expand the scope of editing.When this manuscript was in preparation,it was reported [49] that optimized Cas12a-ABEs could perform efficient adenine base editing in maize and wheat cells.Although several optimization elements similar to those in our study were described,specific strategies for improving editing were also presented,such as applying bipartite nuclear localization signals (BP NLS) and 6XGGGGS linkers.We suggest that editing efficacy could be further increased by stacking these strategies.This approach would give Cas12a platforms the potential to develop robust and high-performance plant base editors for precise genome modification.

CRediT authorship contribution statement

Huanhuan Wang:Investigation,Data curation,Formal analysis.Jing Liang:Data curation,Methodology,Writing -review &editing.Like Chen:Data curation,Formal analysis.Bufang Deng:Resources,Validation.Dongfang Gu:Software.Xiaoshuang Liu:Conceptualization,Validation.Shan Jin:Methodology.Rongfang Xu:Validation.Ruiying Qin:Methodology.Yitong Zhu:Methodology.Liangxia Zhao:Methodology.Dourong Kou:Methodology.Yanjun Chen:Formal analysis.Yingli Jiang:Project administration.Juan Li:Conceptualization,Writing -original draft.Pengcheng Wei:Supervision,Conceptualization,Writing -review &editing.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Acknowledgments

This work was funded by the National Natural Science Foundation of China(U19A2022 and 32000284),the Natural Science Foundation of Anhui Province (2208085Y11,2108085Y07,2008085QC101,and 2008085MC71),the University Synergy Innovation Program of Anhui Province (GXXT-2021-056),Open Research Fund Program of Anhui Province Key Laboratory of Rice Genetics and Breeding (SDKF-2021-01 and SDKF-2022-04),and Natural Science Research Project for Anhui Universities(KJ2021A0196).

Appendix A.Supplementary data

Supplementary data for this article can be found online at https://doi.org/10.1016/j.cj.2023.03.002.