Generation of a high-efficiency adenine base editor with TadA8e for developing wheat dinitroaniline-resistant germplasm

Hunn Hn, Ziwen Wu, Ling Zheng, Jingyi Hn, Yi Zhng, Jihu Li, Shujun Zhng, Genying Li,Chngle M,*, Pingping Wng,*

a College of Life Sciences, Shandong Normal University, Wenhua East Road 88, Jinan 250014, Shandong, China

b Crop Research Institute, Shandong Academy of Agricultural Sciences, Jinan 250100, Shandong, China

Keywords:CRISPR Base editing Dinitroaniline Herbicide-resistant Wheat breeding

ABSTRACT Base editing using CRISPR technologies is an invaluable tool for crop breeding.One of the major base editors, the adenine base editor (ABE), has been successfully used in both model plants and many crops.However, owing to limited editing efficiency, the ABE has been difficult to apply in polyploid crops such as allohexaploid bread wheat that often require simultaneous mutation of multiple alleles for fast breeding.We have designed a wheat high-efficiency ABE(WhieABE), using the newly developed high-activity adenosine deaminase TadA8e. In vivo and in vitro analysis demonstrated the improved applicability of TadA8e over the commonly used TadA7.10.Dinitroaniline is a widely used herbicide with high effectiveness and low toxicity to animals.However,wheat cultivars with tolerance to dinitroaniline are rare,limiting the application of dinitroaniline in wheat planting. Using A-to-G editing with WhieABE, we found that a Met-to-Thr mutation in wheat tubulin alleles located on chromosomes 1A, 1B, 1D, 4A, and 4D increased the resistance of wheat to dinitroaniline, revealing a dosage effect of edited tubulins in resistance.The WhieABE promises to be a valuable editing tool for accelerating crop improvement and developing herbicide-resistant wheat germplasm.

1. Introduction

As a technological innovation, CRISPR/Cas9 has been widely applied in crop breeding, displaying greater efficiency than conventional cross-breeding or transgenics [1,2]. However, as most targets of herbicides are essential housekeeping genes,it is impossible to generate herbicide-resistant germplasm by gene knockout mediated by CRISPR/Cas9. Instead, base editors, which consist of deaminase fused with nicked or dead Cas9, were developed for inducing precise point mutations such as A to G(adenine base editor, ABE) or C to T (cytosine base editor,CBE), and have been used for creating herbicide-tolerant crops[3–5].In the wheat protoplast,mutants in many wheat genes such as TaDEP, TaGW2, and TaEPSP have been successfully edited by ABE[6].However,as bread wheat is an allohexaploid species, it is difficult to edit most target alleles in one generation using existing ABEs,resulting in genetic instability of mutant plants and time-consuming breeding. Recently, a new type of adenosine deaminase, TadA8e, generated by phageassisted evolution, has shown higher editing efficiency than the previous version, TadA7.10, in human and rice cells [7,8].

The cultivation of bread wheat, one of the three major grain crops, is generally impaired by weeds, necessitating the use of chemical herbicides [9,10]. To reduce the harmful side effects of herbicides on crops, it is desirable to develop herbicide-resistant germplasm [3,11]. In bread wheat, seedlings with mutations in TaALS display increased tolerance to sulfonylurea, imidazolinone,and aryloxyphenoxy propionate-type herbicides[5].Another common herbicide, dinitroaniline, targets tubulin genes (Tub) and disrupts microtubule polymerization during cell division and elongation, leading to toxicity to many weeds [12,13]. Dinitroaniline is thus usually used to inhibit the early growth of grasses after crops are planted, but its residue in the soil exerts long-lasting,negative effects on wheat germination[14,15].Because wheat cultivars with tolerance to dinitroaniline are rare, it is desirable to develop resistant wheat cultivars to counter the adverse side effects associated with dinitroaniline use. In goosegrass, a Met-268-Thr mutation in the α-tubulin gene has been found to increase tolerance to dinitroaniline [16], an effect further confirmed in rice[5]. Mutagenizing wheat tubulin genes appears a promising approach to developing dinitroaniline-resistant germplasm.

The objective of the present study was to generate a wheat high-efficiency ABE using codon and editor structure optimization.Comparison of the editing efficiency of TadA8e and TadA7.10 in this framework showed increased editing efficiency of ABE8e relative to ABE7.10 in wheat. We identified wheat tubulin genes and edited their key sites with the new ABE,resulting in increased tolerance of wheat seedlings to dinitroaniline. This study has generated both a useful editor and germplasm for practical use in wheat breeding.

2. Materials and methods

2.1. Plasmid construction

The key components in this study, including SpCas9n (D10A),TadA*7.10,and TadA*8e,were codon-optimized with an online tool(http://www.detaibio.com/sms2/rev_trans.html) and synthesized by GENEWIZ(Suzhou,Jiangsu,China).SpCas9,fused with TadA*7.10(WhieABE7.1) or TadA*8e (WhieABE8e) at its 5′terminal, was assembled downstream of the ZmUbi promoter between StuI and SacI restriction sites in the pLGY plasmid using the infusion method (In-Fusion HD Cloning Kits, CloneTech, CA, USA). Protospacers were assembled upstream of the gRNA scaffold by the Golden Gate method[17].The GFP-TAG-mCherry reporter module or the BlpR resistance screening module, together with the CaMV35S promoter, were assembled downstream of the ABE between SacI and KpnI restriction sites by the infusion method.The primers used in plasmid construction are listed in Table S1,and plasmids generated in this study are described in Table S2.

2.2. Protoplast isolation and transformation

The spring wheat strain YM20 (allohexaploid) was germinated and cultivated at 22 ℃under long-day conditions (16 h day/8 h night). Leaves of one-leaf-stage seedlings were selected for protoplast isolation and transformation, following Shan et al. [18]. High concentrations of plasmids used for protoplast transformation were extracted from Escherichia coli (DH5α) with an endotoxinfree plasmid extraction kit (NucleoBond Xtra Midi EF, M.N., NRW,Germany). Protoplasts cultured for 24–48 h were counted by flow cytometry(LSR Fortessa,Becton,Dickinson and Company,NJ,USA),or used for DNA extraction, PCR amplification, and sequencing.

2.3. Wheat transformation and characterization

Plasmids containing ABE7.1 or ABE8e,sgRNA and the herbicideresistance gene BlRp (for Basta) were introduced into hexaploid Fielder wheat by Agrobacterium tumefaciens-mediated transformation. T0plants were cultivated at 23 ℃, under long-day conditions(16 h day/8 h night).High-positioned leaf blades of one-month-old plants were selected for DNA extraction. Transgenic lines were screened by PCR using gDNA as templates and primers designed for Cas9. Target-adjacent fragments were detected by PCR with gDNA of transgenic lines and the presence of editing was identified by Sanger sequencing(TsingKe,Shandong,China).All primers used are listed in Table S1.

2.4. High-throughput sequencing

The target adjacent fragments were amplified from gDNA of positively edited plants (identified by Sanger sequencing) using common primers for TaTUB homologs.A unique six-nucleotide barcode was designed for each positively edited plant and added to the PCR products. The mixture of PCR products was used to construct a library and sequenced by Novogene (Tianjin, China). The reads were sorted by barcode and the TaTUB homologs were distinguished by unique SNPs. The editing ratios of each homolog in every plant were recorded. For gDNA isolated from protoplasts(for assessment of editing in TaGW2 and TaDEP1),a similar method was applied.

The extent of editing of TaALS and TaLOX2 in protoplasts and tubulin genes in the T1mutants was determined by the Hi-TOM method [19],in which the target sites were amplified with tagged primers and detected by an immobilization process (http://www.hi-tom.net/hi-tom/). All primers used are listed in Table S1.

2.5. Dinitroaniline treatment of wheat seedlings

Two commercial herbicides, pendimethalin and trifluralin(CAS:40487-42-1 & CAS:1582-09-8, Shanghai yuanye Bio-Technology, Shanghai, China), were selected for treatment of T1edited seeds. Seeds were soaked in herbicide solution (2.5 mg L-1pendimethalin and 5 mg L-1trifluralin, solubilized in 1/1000 DMSO)or H2O with 1/1000 DMSO(as a vehicle control)for germination(21 ℃,long days).The seedlings were treated for a week and the growth phenotype was observed. The treated plants were transplanted into cultivation medium and grown to maturity under normal conditions (21 ℃, long days). Plant height, spikes per plant (number of effective tillers), spikelets per spike, kernels per spike, and grain weight were recorded.

3. Results

3.1. Construction of WhieABE containing TadA8e

As an allohexaploid, bread wheat requires novel, efficient editing tools to allow fast breeding.To develop a wheat high-efficiency ABE (WhieABE), we made four modifications to existing editors(Fig. 1A). First, a newly developed deaminase TadA8e, which displays higher editing activity than TadA7.10 in mammalian cells[7], was used. Second, instead of using a TadA*-TadA module, we fused a single TadA* with SpCas9n(D10A), which has previously shown high editing efficiency in rice [20]. Third, both TadA* and SpCas9n were codon-optimized based on the wheat transcriptome to increase their expression in wheat.Finally,to permit the nuclear localization of the translated editor, a bpNLS (Bipartite-SV40-Nuc lear-Localization-Signal) was added at the N-terminus of TadA*and another bpNLS followed by a nucleoplasmin NLS (npNLS) at the C-terminus of SpCas9n.

3.2. Assessment of editing efficiency of WhieABE in wheat protoplasts

As TadA8e has not been previously used in wheat editing, we compared the activity of TadA8e with that of the commonly used deaminase TadA7.10 in WhieABE (WhieABE8e and WhieABE7.1,respectively). A reporter system in wheat protoplasts was designed,which contained a stop codon(TAG,as the editing target of ABE)between GFP and mCherry genes driven by a 35S promoter in the WhieABE vectors (Fig. 1A). In transformed protoplasts,if no editing occurred, only GFP would be observed, whereas successful editing of the TAG stop codon to TGG would allow translation of both GFP and mCherry(Fig.1B).Thus,editing efficiencies were calculated as the ratio of mCherry to GFP cells (the ‘‘editing ratio”).Flow cytometry separation revealed an editing efficiency of 3%for WhieABE7.1 and 24.5%for WhieABE8e at 36 h after transformation (Fig. 1C, D), indicating increased editing efficiency with WhieABE8e in wheat cells.

Fig.1. Construction of WhieABE and assessment of editing efficiency with a reporter system in wheat protoplasts.(A)Structure of the WhieABE and GFP-mCherry reporter system.(B)Editing efficiency testing using a GFP-mCherry reporter system in wheat protoplasts.Thirty-six hours after transformation,transformed cells translated both GFP and mCherry if edited,but only GFP if unedited.Scale bars,50 μm.(C)Fluorescence of transformed cells by flow cytometry.Three biological repeats with at least 20,000 cells per sample were tested. (D) Quantification of the editing ratio (# mCherry-cells /# GFP-cells) by flow cytometry. **, P ＜0.01.

To compare the ability of WhieABE8e and WhieABE7.1 to edit wheat endogenous genes, TaGW2 and TaDEP1 [6] were used as endogenous markers in transformed wheat protoplasts (Fig. 2).As GFP was co-expressed with the WhieABE and sgRNA,the transformation rates of protoplasts were measured by flow cytometry as the proportion of cells expressing GFP. The editing sites of these two endogenous genes in protoplasts were detected by deep amplicon sequencing. After standardization by the transformation rate, cells edited by WhieABE8e showed a 3–4 or 2–4 fold higher editing efficiency than WhieABE7.1 on TaGW2 and TaDEP1,respectively. The editing window features of WhieABE8e on two genes were similar,ranging from A4 to A10 and exhibiting a higher editing ratio at A8. The editing efficiency of two other endogenous wheat genes, TaALS and TaLOX2, was assessed by the Hi-TOM method in protoplasts. In comparison with WhieABE7.1, editing with WhieABE8e showed an increased mutant proportion in the targets of both genes (Table S3). These results showed a marked increase in editing activity caused by TadA8e compared to TadA7.10 in wheat cells. The off-target editing of the two endogenous genes (TaALS and TaLOX2) was assessed by the Hi-TOM method. Both WhieABE7.1 and WhieABE8e showed low off-target editing ratios (Table S3).

3.3. Identification of candidate dinitroaniline-resistant sites in wheat tubulins

Mutation of a conserved methionine(Met)to threonine(Thr)in tubulin genes has been shown to be associated with dinitroaniline resistance in goosegrass and rice [5,16]. By alignment based on amino acid sequence, we identified six wheat tubulin genes with the highest similarity to OsTubA2, located on chromosomes 1A,1B, 1D, 4A, 4B, and 4D (Fig. 3A). All six contained a conserved Met residue with the neighboring sequence resembling that of Met-268 in OsTubA2 (Fig. 3). Thus, a Met-to-Thr mutation in these tubulins might influence dinitroaniline resistance of wheat seedlings. Transformation from Met (ATG) to Thr (ACG) could be achieved by base editing with ABE, allowing a test of the application of our WhieABE in vivo.

Fig.2. Editing of TaGW2 and TaDEP1 in wheat protoplasts.(A)WhieABE vectors with a GFP marker and sgRNAs for TaGW2 or TaDEP1.(B–C)Editing ratios of adenines in the edit windows of TaGW2(B)and TaDEP1(C).**,P ＜0.01.Ratios were standardized by transformation rates based on GFP labeling(counted by flow cytometry in 10,000 cells per sample).

Fig.3. Identification of OsTubA2 homologs in bread wheat and identification of the candidate dinitroaniline-tolerance site.(A)Alignment of wheat tubulins with OsTubA2.(B)Identification of the dinitroaniline-tolerance site in wheat tubulin genes based on amino acid sequence alignment.The sequence matching the protospacer is in blue and the NGG PAM sites and the edited sites in red. The amino acid sequence of the target locus is shown at bottom.

3.4. Editing efficiency of tubulins by WhieABE in wheat

Fig.4. Characteristics of editing of tubulins with WhieABE in wheat plants.(A)WhieABE vectors with a resistance marker BlpR and a sgRNA for editing wheat tubulin genes.(B)Quantification of positively transformed wheat plants by Sanger sequencing.Prospective edited bases are marked with red frame.Green peaks represent unedited‘‘A”and black peaks edited ‘‘G”. Four groups with different editing intensities were defined according to the heights of green and black peaks. (C) Analysis of edited plants by deep amplicon sequencing. Scatter plots show editing ratios of edited reads for six wheat tubulin alleles in 60 positively edited lines (x-axis). Red line, 50% proportion of edited reads.Plant lines above the red line represent statistically significant editing events(SSEEs).Yellow spots,WhieABE7.1 SSEEs;green spots,WhieABE8e SSEEs.(D)Number of plants with SSEEs at each tubulin locus (edited by WhieABE7.1 or WhieABE8e). Note that owing to the single-base mismatch of the sgRNA with the tubulin locus on chromatin 4B, editing of this locus is considered off-target. (E) Off-target analysis in the wheat genome by Sanger sequencing. The target sequence matched by the protospacer is marked in blue. NGG PAM and edited sites are marked in red. Mismatch bases are marked in green.

Fixed by the NGG PAM, a protospacer was designed to match the anti-sense strand of the candidate resistance site in the six wheat tubulin genes. To make the desired Met-to-Thr mutation,A-to-G editing at the +7 position (counted from the 5′terminal of the protospacer) was needed (Fig. 4A). Our sgRNA containing this protospacer perfectly recognized the five tubulin genes located on chromosomes 1A,1B,1D,4A,and 4D,but had a single-base mismatch with the 4B allele(Fig.3).To compare the editing efficiency of TadA8e and TadA7.10 in wheat plants, WhieABE8e and WhieABE7.1 vectors containing the tubulin-target sgRNA and the basta-resistance gene BlRp were transformed into spring wheat Fielder with Agrobacterium (Fig. 4A). Following selection by phosphinothricin treatment and PCR for SpCas9, respectively 46 and 48 positive lines were successfully transformed with WhieABE8e and WhieABE7.1. Sanger sequencing revealed that 78% of the WhieABE8e and 50% of the WhieABE7.1 lines were edited in the desired position (Fig. 4B). For more detailed analysis (whether or not an allele was mutated)than editing success,the editing intensity was calculated from the ratio of the heights of the edited to those of the wild-type(WT) peaks in the Sanger sequencing trace,which shows the extent to which the corresponding locus was mutated (Fig. 4B). These editing events were divided into four groups with an editing intensity from weak to strong: I, stronger WT peak; II, similar WT and edited peaks; III, stronger edit peak;IV,only edit peak.Whereas WhieABE8e and WhieABE7.1 produced similar proportions of intensity I plants, WhieABE8e-edited plants showed higher proportions of intensities II, III, and IV. This result was further confirmed by deep amplicon sequencing of the target sites of the 60(24 WhieABE7.1 and 36 WhieAbe8e)correctly edited plants (Fig. 4C; Table S4). The number of statistically significant editing events(SSEEs),defined as plants with more than 50%edited reads of the five perfect-match tubulin alleles, was determined(Fig. 4D). For all five perfectly matched tubulin alleles, a greater proportion of SSEEs were generated by WhieABE8e than WhieABE7.1, confirming that TadA8e was more efficient than TadA7.10 in editing wheat genes in vivo. WhieABE8e not only edited more plants, but also induced higher editing intensity than WhieABE7.1, indicating increased activity of TadA8e relative to TadA7.10 in wheat base editing.

For the single-base mismatch(compared with our sgRNA)tubulin allele on chromosome 4B,0–8.4%or 0–12.7%reads were edited with WhieABE7.1 or WhieABE8e, respectively, showing slight offtarget effects (Fig. 4C; Table S4). Five candidate sites in the wheat genome, which had one- or two-base mismatches with our tubulin-target sgRNA,were checked for off-target effects by Sanger sequencing (Fig. 4E). Among the 40 plants with the highest ontarget editing intensity (half WhieABE7.1, half WhieABE8e), only a few(at most 3/20)off-target events below intensity II were found in three one-base-mismatch loci.Thus,both TadA8e and TadA7.10 showed low off-target editing of all tested endogenous genes.

3.5. Edited wheat tubulins conferred resistance to dinitroaniline

To test how the Met-to-Thr mutation of wheat tubulins affected dinitroaniline tolerance, tubulin-edited T1plants were exposed to two commercial dinitroaniline herbicides, pendimethalin and trifluralin, and their growth was observed (Fig. 5). The T1seeds,which carried 1 to 10 edited tubulin loci and came from T0parental lines (Fig. 5A; Table S5), were placed in water overnight and then germinated in one of the two dinitroaniline solutions. The edited and WT seedlings displayed similar growth phenotypes under normal cultivation(Fig.5B).WT seeds treated for one week with either pendimethalin(2.5 mg L-1)or trifluralin(5 mg L-1)showed inhibited germination with extremely stubby buds and roots. In contrast, the edited T1seedlings germinated and had longer leaves and roots.The seedlings were then transplanted into culture medium and cultivated under normal conditions to analyze the effects of dinitroaniline on reproduction. All untreated WT and edited seedlings thrived and eared well, showing similar plant height and effective tiller numbers (Figs.5C, S1). After transplantation,dinitroaniline-treated WT plants showed impaired growth and eventually died.However,treated edited plants recovered and finished their life cycles (Fig. 5D, E). Moreover, the edited quintuple tubulin mutant plants (from parental line 32), but not quadruple mutant plants (from parental line 10), displayed nearly normal growth.No significant difference in spikelet number,grain number per spike,or kernel size between tubulin mutant and WT untreated plants was observed, and only a slight decrease (P ＞0.05) in the treated quintuple mutant lines relative to untreated WT plants was observed (Figs. 5F, G, S1). Dinitroaniline resistance increased with numbers of edited tubulin alleles (quadruple vs. quintuple mutants), suggesting a dosage effect.

Fig.5. Dinitroaniline tolerance of T1 plants.(A)Editing analysis of T1 seedlings by Hi-TOM sequencing.HM,homozygous;HZ,heterozygous.(B)Germination phenotype of T1 seedlings. Seeds were germinated and treated with 1‰ dimethyl sulfoxide (control), pendimethalin (2.5 mg L-1), or trifluralin (5 mg L-1) for one week. Scale bars, 0.5 cm.(C–E) Reproduction following dinitroaniline treatment. Reproduction phenotype of T1 plants transplanted from control(C),pendimethalin (D) and trifluralin (E)conditions.Scale bars, 5 cm. (F) Spikes of treated and untreated plants. Scale bars, 5 cm. (G) Kernels (50 per group) of treated and untreated plants. Scale bars, 1 cm.

4. Discussion

4.1. WhieABE8e promoted rapid base editing in bread wheat

Because of its limited editing efficiency, gene editing in crop breeding is often hindered by interference from mosaicism and heterozygosity, which reduces the proportion of edited offspring or prolongs the breeding period, respectively [21,22]. Generating a homozygote with six edited alleles in wheat, a classical allohexaploid,is especially difficult to achieve in a short period of time.Even ABEs, which have generally shown good editing efficiency in many plants, have thus far not been sufficiently effective in bread wheat. Thus, development of an efficient ABE for wheat was needed. In addition to codon optimization and improving the nuclear targeting efficiency of Cas9 and deaminase [23], use of a better adenine deaminase variant would be necessary to increase the editing efficiency of ABE. The newly developed TadA8e shows higher affinity for DNA and increased deaminase activity,resulting in increased editing efficiency relative to that shown by TadA7.10 in mammalian cells [7]. Although TadA8e has been used in recent plant studies [8,24,25], no comparison of the editing efficiency of the new ABE8e with that of the current ABE7.10 in bread wheat had been undertaken. In our study, a wheatoptimized WhieABE was developed and the editing activity caused by TadA8e and TadA7.10 was compared in detail using protoplasts and transgenic plants. WhieABE8e showed higher efficiency than WhieABE7.1 and low off-target editing (at least for our edited locus), indicating the improved applicability of TadA8e relative to TadA7.10 for gene editing in bread wheat.The higher editing ratios of the T0plants edited by WhieABE8e at tubulin gene loci indicates the superior efficiency of WhieABE8e for generating multiple mutations (Fig. 4C; Table S4). These mutations were robustly transmitted to their offsprings(Table S5). In combination with our improved ABE framework and the more efficient editor TadA8e, WhieABE8e promises to be a high-efficiency editing tool for wheat breeding, which could shorten the usually long process of generating germplasms due to inefficient gene editing.

4.2. Met-to-Thr mutation in wheat tubulins increased dinitroaniline resistance via a dosage effect

Dinitroaniline, which interferes with microtubule polymerization during cell division and elongation, is an effective broadspectrum herbicide against which weeds rarely develop resistance[12]. Although the Met-268-Thr mutation in α-tubulins confers dinitroaniline resistance on goosegrass [16], heterozygous rice mutant lines displayed weaker resistance to dinitroaniline than homozygous mutant lines [5], indicating that Met-268-Thr was not a simple gain-of-function mutation. Five wheat α-tubulin genes were edited with WhieABE(Fig.4)and many wheat mutants with 1 to 10 edited tubulin alleles were generated(Table S5).Dinitroaniline resistance gradually increased with the number of mutated tubulin alleles (Fig. 5B, D, E). Thus, there appeared to be a dosage effect in which more mutant tubulins conferred greater dinitroaniline resistance.This putative dosage effect has two practical implications. First, as many common weeds in crop fields are polyploid[26,27],mutations in only a few tubulin homologs would not lead to high resistance to dinitroaniline,potentially explaining the infrequent emergence of weed resistance against dinitroaniline. Second, as allohexaploids, wheat cultivars must also mutate multiple tubulins to obtain high resistance to dinitroaniline.Although this was previously technically challenging to accomplish by conventional breeding methods, it can now be rapidly achieved by editing with WhieABE8e.

CRediT authorship contribution statement

Huanan Han:Conceptualization,Data curation,Funding acquisition, Project administration, Visualization, Writing - original draft.Ziwen Wu:Conceptualization,Data curation,Funding acquisition, Project administration, Visualization, Writing - original draft.Ling Zheng:Data curation, Formal analysis, Investigation,Resources.Jingyi Han:Investigation, Validation.Yi Zhang:Methodology.Jihu Li:Methodology, Software.Shujuan Zhang:Methodology.Genying Li:Methodology, Resources.Changle Ma:Conceptualization,Funding acquisition,Writing-review&editing,Supervision.Pingping Wang:Conceptualization, Writing - review& editing, Project administration, Supervision.

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 Agricultural Variety Improvement Project of Shandong Province (2019LZGC015) and the National Natural Science Foundation of China (31901432).

Appendix A. Supplementary data

Supplementary data for this article can be found online at https://doi.org/10.1016/j.cj.2021.08.006.