Precision genome editing in plants via gene targeting and subsequent break-induced single-strand annealing.

Genome editing via artificial nucleases such as CRISPR/Cas9 has become popular in plants now. However, small insertions or deletions are major mutations and nucleotide substitutions rarely occur when DNA cleavage is induced. To induce nucleotide substitutions, a base editor utilizing dead or nickase-type Cas9 fused with deaminase have been developed. However, the direction and position of practical substitution are still limited. In this context, homologous recombination (HR)-mediated gene targeting (GT) has advantages because any mutations existing on the donor DNA are copied and passed onto the endogenous DNA. As HR-mediated GT is extremely rare in higher plants, positive-negative selection has been used to isolate cells in which GT has occurred. After successful selection, positive selection marker is no longer needed and should ideally be eliminated. In a previous study, we reported a seamless piggyBac-transposon-mediated marker elimination system. Precision marker elimination efficiency in this system is very high. The piggyBac transposon integrates into the host genome at TTAA elements and excises without leaving a footprint at the excised site, so a TTAA sequence is necessary at the location of a positive selection marker. To compensate for this limitation, we have developed a novel marker elimination system using an I-SceI break and subsequent single-strand annealing (SSA)-mediated DNA repair system.

Introduction

A positive–negative selection system is a strategy for enriching transgenic cells carrying a targeted gene replacement introduced via homologous recombination (HR) from among a large number of cells with non‐homologous end‐joining (NHEJ)‐mediated random integrations (Shimatani et al., 2014). To eliminate the positive marker from the GT locus, site‐specific recombination systems such as Cre/loxP and FLP/FRT systems have been utilized in mammals and plants (Dang et al., 2013; Terada et al., 2010; van der Weyden et al., 2002). However, site‐specific recombination systems leave dispensable sequences, for example, the 34‐bp recognition sequences of site‐specific recombinase (a single loxP and FRT site), at the excised site. To overcome this problem, in our previous study in rice, we utilized the piggyBac transposon, which does not leave a footprint after transposition, as a marker elimination device. Our system, using a piggyBac transposon derived from the cabbage looper moth, afforded precise genome modification via GT and subsequent marker excision in mammalian cells (Morioka et al., 2014; Sun and Zhao, 2014; Yusa et al., 2011). We confirmed that the piggyBac transposon is capable of accurate and effective transposase‐mediated transposition also in plant cells (Nishizawa‐Yokoi et al., 2014). In addition, we have demonstrated piggyBac transposition‐mediated excision of a selectable marker from a GT locus at a high frequency without concomitant reintegration of the transposon (Nishizawa‐Yokoi et al., 2015). However, the requirement for a TTAA sequence at the location of the positive selection marker is a factor limiting the universality in the piggyBac system; piggyBac transposon originally integrates into the host genome at TTAA elements and excises without leaving a footprint at the excised site (Cary et al., 1989). To overcome this inconvenience, we have developed a different precise marker elimination system.

Cleavage of both ends of a positive selection marker cassette is the easiest way to eliminate a positive selection marker. To establish a universal marker elimination system, the meganuclease I‐SceI, which has a specific 18‐bp recognition site, was used to eliminate the positive selection marker. By overlapping 30‐bp on 5′ and 3′ homology arms, short tandem microhomologies are generated flanking the selection marker, and single‐strand annealing (SSA)‐mediated DNA repair enable seamless marker elimination.

As a target gene for modification, we selected the gene encoding rice phytoene desaturase (PDS) since the amino acid sequence of PDS is widely conserved in plants, and mutations in PDS conferring herbicide inhibition have been reported in Synechococcus (Chamovitz et al., 1991) and aquatic weeds (Arias et al., 2005; Michel et al., 2004). PDS converts phytoene to ζ‐carotene. Carotenoids are essential components of the photosynthetic apparatus and participate in light harvesting and protection of chloroplasts from the harmful effect of singlet oxygen formed during photosynthesis (Sandmann and Böger, 1997). PDS‐inhibiting herbicides, such as fluridone, prevent the formation of carotenoids, resulting in the degradation of chlorophyll and the destruction of chloroplast membranes, which is characterized by the photobleaching of green tissues (Böger and Sandmann, 1998). Mutations of the cyanobacterium Synechococcus PDS have resulted in herbicide‐resistant microbial enzymes (Chamovitz et al., 1991) and naturally occurring mutations at amino acid 304 of PDS from Arg (CGT) to Ser (AGT), Cys (TGT) or His (CAT) in the aquatic weed Hydrilla verticillate (L.F. Royle) have been reported to impart herbicide resistance (Michel et al., 2004). Mutant Synechococcus PDS has conferred herbicide resistance when expressed in transgenic tobacco (Nicotiana tabacum; Wagner et al., 2002) and transgenic Arabidopsis (Arias et al., 2006) plants. In the latter Arabidopsis study, the authors tested H. verticillate PDS with all 19 other natural amino acids at amino acid 304 against the PDS‐inhibiting herbicide fluridone and showed that a high level of resistance to fluridone when tested in vitro was obtained with R304S. Because PDS‐inhibiting herbicide‐resistant rice are not isolated from natural sources, we tried to induced mutations corresponding to R304S in H. verticillate PDS in the endogenous rice PDS gene in order to confer herbicide resistance.

Results

Introduction of point mutations into the PDS gene via GT using a positive–negative selection system

The alignment of PDS protein sequence from Hydrilla (Hydrilla verticillata) (AY639658) and rice (Oryza sativa L.) (Os03t0184000) showed conserved Arg codons corresponding to position 304 of PDS in Hydrilla (Figure 1). So we decided to induce nucleotide substitutions that convert Arg to Ser at position 304 of the PDS gene in the rice genome. In rice, a strong positive–negative selection system using the hpt gene conferring resistance to hygromycin B as a positive selection marker and diphtheria toxin A subunit (DT‐A) gene as a negative selection marker has been developed for the selection of GT‐positive cells (Terada et al., 2002). Our GT vector carries DT‐A gene expression cassettes at both sides and a 2.6‐kb fragment containing a PDS coding region with mutations that alter a PacI site (TTAATTAAC) to a HpaI site (TTGGTTAAC) as well as the R304S mutation CGA (R) –> AGT (S) (Figure 2a). A positive selection maker, neomycin phosphotransferase II (NPTII) expression cassette with I‐SceI recognition sequences at both ends was inserted in the two homology arms. I‐SceI expression results in release of the positive selection marker. Simple ligation of DNA ends results in gaining extra sequence or loss of endogenous sequence. To inhibit simple ligation of DNA ends after I‐SceI‐mediated positive selection marker excision, I‐SceI recognition sequences at both ends of the positive selection marker are reverse orientated (I‐SceI site in the 5′‐homology arm, 5′‐TAGGGATAACAGGGTAAT‐3′; I‐SceI site in the 3′‐homology arm, 5′‐ATTACCCTGTTATCCCA‐3′) because the 18‐bp I‐SceI recognition sequence is not palindromic and non‐compatible cohesive ends may not be so easy to ligate. For seamless marker elimination, SSA is effective. However, homology‐directed repair (HDR) occurred without DNA double strand break (DSB) induction to some extent when the 557 bp repeat was present in rice (Kwon et al., 2012), and such unintended HDR during positive–negative selection loses precious GT cells. Thus, we decided to shorten the overlap sequences to 30‐bp (for details, see Discussion) and expected the occurrence of SSA after I‐SceI expression.

Figure 1

Amino acid sequence alignment of Hydrilla verticillata PDS and Oryza sativa PDS. Red arrowhead indicates the conserved R304. Identity and similarity of H. verticillata PDS and O. sativa PDS are 76.6% and 85.3%, respectively.

Figure 2

GT and marker elimination strategy. (a) Schematic representation of GT vector and process of positive–negative selection. The GT vector pGT_PDS_R304S contains the 0.5‐kb PDS 5′ UTR and 2‐kb PDS coding region as a 5′ homology arm, the NPTII expression cassette for G418 resistance, the 2.7‐kb PDS coding region and 0.3‐kb 3′UTR as a 3′ homology arm. The DT‐A genes encoding the diphtheria toxin A fragments, were placed as lethal negative‐selection markers at both ends of T‐DNA in order to efficiently eliminate transformants carrying entire T‐DNA segments. The black rectangles with RB and LB indicate the right and left borders of T‐DNA, respectively. The positive selection marker cassette shown by white arrow is located within 30 bp overlapped homologous arms and reverse orientated I‐SceI recognition sequences are nested between the selection marker and engineered microhomologies. Synthesis dependent strand annealing (SDSA) between endogenous PDS locus and the homology arm on pGT_PDS_R304S led to the integration of NPTII into the endogenous PDS sequence. White boxes, 5′ and 3′ UTR of the PDS gene; gray boxes, exons; black arrows, negative selection marker expression construct; red star, R304S mutation. Primer sequences are shown in the Table S1. (b) Representation of positive selection marker elimination step. After selecting GT cells, pZH_I‐SceI harbouring β‐estradiol inducible I‐SceI expression cassette and HPT expression cassette is integrated into the genome as a second transformation. Positive selection marker was eliminated by I‐SceI expression and SSA using 30‐bp overlapped sequence enables scarless marker elimination. Small arrowheads represent primers. (c,d) Screening of calli with GT. M, size marker; WT, wild‐type Nipponbare. (c) PCR using PDS‐F1and Posi‐R1 as primers. Red arrowheads show the expected size of PCR products. Light white line between calli number 19 and 20 is a trace of scotch‐tape. (d) PCR using Posi‐F1 and PDS‐R1 as primers. Red arrowheads show the expected size of PCR products. (e) Diagram of sequencing. PCR fragments amplified with Posi‐F1 and PDS‐R1 were sequenced by PDS‐R4.

Rice calli derived from Nipponbare mature seeds were inoculated with Agrobacterium harbouring the GT vector pGT_PDS_R304S (Figure 2a) for 3 days. After washing out Agrobacterium, transgenic calli were selected on medium containing geneticin 418 (G418) for 4 weeks. In total, 30 independent G418‐resistant calli were obtained from 1032 pieces (approximately 12 g) of calli (Table 1) and were subjected to polymerase chain reaction (PCR) analysis with the primer sets, PDS‐F1 and Posi‐R1, Posi‐F1 and PDS‐R1 shown in Figure 2b to detect GT events in the endogenous PDS gene. Seven calli (PDS‐GT#11, 20, 22, 27, 30, 31, 33) were PCR positive when primers PDS‐F1 and Posi‐R1 were used (Figure 2c) but no PCR product was detected in PDS‐GT #20 when Posi‐F1 and PDS‐R1 were used as primers (Figure 2d). These results indicate that the positive selection marker gene was introduced into the PDS locus by HR between the GT vector and the target locus in six lines (PDS‐GT #11, 22, 27, 30, 31 and 33).

Table 1

Summary of GT experiment

GT vector	TF calli	G418 resistant calli	5′ PCR‐positive calli †	3′ PCR‐positive calli ‡	Calli with HR‐mediated GT	GT Calli with R304S substitution
pGT_PDS_R304S	1032	30	6 (PDS‐GT #11, 22, 27, 30, 31, 33)	7 (PDS‐GT #11, 20, 22, 27, 30, 31, 33)	6 (PDS‐GT #11, 22, 27, 30, 31, 33)	2 (PDS‐GT #11, 22)

5′ PCR uses primer PDS‐F1 and Posi‐R.

3′ PCR uses primer Posi‐F and PDS‐R1.

Sequence analysis of fragments PCR‐amplified with Posi‐F1 and PDS‐R1 by PDS‐R4 revealed the existence of R304S at the targeted PDS locus in two callus lines (PDS‐GT #11, 22) and a lack of R304S mutations in four callus lines (PDS‐GT #27, 30, 31, 33; Figure 2e). In the 5′ homology arm of the GT vector, two point mutations exist—AA to GG—that change a PacI site to a HpaI site. In both PDS‐GT #11 and #22, the first G was present but the second G was not (data not shown). Such mosaic states of base change have been reported in other positive‐negative selection‐based GT (Johzuka‐Hisatomi et al., 2008; Nishizawa‐Yokoi et al., 2014) and in GT based on target gene specific selection (Endo et al., 2007). A similar result was reported in mouse embryonic stem cells (Steeg et al., 1990), and involvement of DNA mismatch repair was proposed. We expected that, during synthesis dependent strand annealing (SDSA), a heteroduplex of AA and GG was formed and the occurrence of independent mismatch repair resulted in the chimeric state of point mutations.

Precise marker excision from the GT locus by I‐SceI expression

To eliminate the positive selection marker precisely from PDS‐GT #11, 22 calli were infected with Agrobacterium harbouring inducible I‐SceI expression vector, pZH_I‐SceI (Figure 2b). Transgenic calli of pZH_I‐SceI were selected on N6D medium with hygromycin and meropenem. Because the β‐estradiol inducible I‐SceI expression construct was located on pZH_I‐SceI, the calli were treated with β‐estradiol to induce I‐SceI expression for 3 weeks, then transferred to regeneration medium. A total of 48 and 44 T0 plants regenerated from PDS‐GT #11 and #22, respectively, were subjected to marker excision analysis by PCR. Out of 48 pZH_I‐SceI‐transformed PDS‐GT #11 T0 plants, 9 showed no PCR amplification using primers PDS‐F1 and Posi‐R1 (PDS‐GT #11/I‐SceI #7, #13, #19, #20, #30, #33, #38, #43, #45 in Figure 3a, shown in red). On the other hand, all 44 T0 regenerated pZH_I‐SceI‐transformed PDS‐GT #22 plants showed PCR amplification, meaning that no elimination of the selection marker expression cassette had occurred (data not shown). To select regenerated plants carrying the R304S mutation with precise SSA‐mediated marker elimination, fragments PCR amplified using PDS‐F2 and PDS‐R3 (Figure 2b) were sequenced by PDS‐R6. The R304S mutation was detected in four regenerated plants, PDS‐GT #11/I‐SceI #7, #20, #33 and #45 (Figure 3b). Sequencing of the selection marker insertion site revealed seamless marker elimination in PDS‐GT #11/I‐SceI #20 and #45 (Figure 3c). In PDS‐GT #11/I‐SceI #7, an 8‐bp insertion existed in an overlapping region of the PDS gene. In PDS‐GT #11/I‐SceI #33, a 30‐bp overlapping region and partial I‐SceI recognition sequence remained (Figure 3c). Southern blot analysis supported elimination of the NPTII expression cassette in PDS‐GT #11/I‐SceI #7, #20 and #45 (Figure 4, #35 is a negative control). The results of positive selection marker elimination are summarized in Table 2.

Figure 3

Proof of scarless positive selection marker elimination. (a) PCR using PDS‐F1and Posi‐R1 as primers. No amplification of this PCR means lost positive selection marker. Red arrowheads show the expected size of PCR products when positive selection marker remains. Regenerated plants shown in red are the candidates in which I‐SceI and SSA‐mediated marker elimination occurs. M, size marker; WT, wild‐type Nipponbare. (b,c) Direct sequence of PCR products to know the scarless marker elimination. Genomic DNAs of the regenerated plants shown in red in (a) are used as templates. (b) PCR using PDS‐F2 and PDS‐R3 as primers amplify both non‐modified PDS allele and modified PDS alleles with positive selection marker elimination. Direct sequencing using PDS‐R6 as primer showed double peaks coding R (CGA) and S (AGT) at amino acid 304 in PDS‐GT #11/I‐SceI#7, #20, #33, #45 but single peak coding R (CGA) in PDS‐GT #11/I‐SceI#13, #19, #30, #38, #43. Left panel, example data of PDS‐GT #11/I‐SceI#20. Right panel, example data of PDS‐GT #11/I‐SceI#13. (c) Detailed sequences of the overlapping region and two I‐SceI recognition sequences at the both ends of positive selection marker cassette. Top, sequence on the GT vector. The 30‐bp overlapped sequences present in the 5′ and 3′ homology arms are shown in red. 18‐bp I‐SceI recognition sequences adjoin the positive selection marker cassettes. I‐SceI recognition sequence is not palindromic and two I‐SceI sequences existing as reverse orientations. Slashes show the cleavage points in the I‐SceI recognition sequences. In PDS‐GT #11/I‐SceI #7, partial I‐SceI sequence derived from the I‐SceI sequence next to the positive selection marker cassette remains. In PDS‐GT #11/I‐SceI #20 and #45, I‐SceI sequences disappeared and the overlap sequences combined into one. In PDS‐GT #11/I‐SceI #33, the positive selection marker cassette was eliminated but overlap region and partial I‐SceI sequence still remained. PDS, phytoene desaturase.

Figure 4

Southern blot analysis using PDS probe and NPTII probes. Genomic DNAs digested with SacI were hybridized with PDS (a) or NPTII (b) probes. (a) The 14,148‐bp band indicates a non‐modified or modified PDS locus with positive selection marker cassette elimination. Existence of the 8,508‐bp band means HR‐mediated GT occurs in the PDS locus but with no positive selection marker elimination. PDS‐GT #11/I‐SceI #35 is the negative control of positive selection marker elimination. WT, wild‐type Nipponbare. (b) NPTII probe hybridized with PDS‐GT#11, regenerated plants without pZH_I‐SceI transformation and PDS‐GT#11/I‐SceI #35 in which positive selection marker remained. PDS, phytoene desaturase; WT, wild‐type Nipponbare.

Table 2

Summary of marker elimination

GT calli	Regenerated plants with I‐SceI expression construct	Positive selection marker elimination	R304S remains	SSA‐mediated marker elimination
PDS‐GT #11	48	9 (#7, 13, 19, 20, 30, 33, 38, 43, 45)	4 (#7, 20, 33, 45)	2 (#20, 45)
PDS‐GT #22	44	0	0	0

Analysis of PDS mRNA in GT plants with/without selection marker

Because the NPTII expression cassette is located in an intron, the possibility exists that mature mRNA transcribed from the modified PDS allele with positive selection marker encodes functional PDS with the R304S mutation. We analysed the PDS cDNA sequence in PDS‐GT #11 with/without precise marker elimination using primers PDS‐F3 and PDS‐R4 (Figure 2b). In regenerated plants of PDS‐GT #11, an intact NPTII expression cassette existed in the 6th intron; the heterozygote status and large PCR product indicating that marker gene insertion were detected in genomic DNA (Figure 5a, PDS‐GT #11). On the other hand, no extra band appeared in PDS‐GT #11 cDNA. If the NPTII expression cassette is spliced out during the process of RNA splicing, the R304S mutation must be detected in the PCR product derived from cDNA. However, the sequence pattern was wild‐type (WT) and neither the R304S mutation nor splicing variants were detected in the PCR product using PDS‐GT #11 cDNA as a template (Figure 5b). This result might suggest that insertion of a 4.2‐kb fragment into the 6th intron disturbed splicing itself or changed the splicing pattern, and only the WT PDS allele was amplified by PCR using cDNA as template.

Figure 5

Analysis of PDS transcript. (a) PCR using PDS‐F3 and PDS‐R4 located on the 5th and 8th exon as primers. WT, wild‐type Nipponbare. (b) Direct sequence of PCR products derived from cDNAs. PDS‐F3 was used for sequencing. CGA single peak coding R304 was detected in PDS‐GT#11 and CGA and AGT double peak coding R304 and S304 respectively were detected in PDS‐GT#11/I‐SceI #7, #20, #45. 6th intron sequence where positive selection marker existed was not detected in all sequenced samples.

In PDS‐GT #11/I‐SceI #7, in which marker elimination succeeded but an 8‐bp partial I‐SceI fragment remained (Figure 3c), and in PDS‐GT #11/I‐SceI #20 and #45, in which no modification without R304S substitution remained, the 6th intron was spliced correctly, and CGA coding R and AGT coding S were detected in cDNA (Figure 5b). Splicing of the 6th intron in PDS‐GT#11, PDS‐GT #11/I‐SceI #7, #20 and #45 was same as that in WT (Figure 5b).

Analysis of positive selection marker‐free T2 progeny harbouring a modified PDS gene

To confirm that the R304S mutation induced via GT does not change the expression level of PDS but produces a PDS‐inhibiting herbicide‐tolerance, quantitative reverse transcription‐polymerase chain reaction (RT‐PCR) and a herbicide‐sensitivity test were conducted. T1 progeny plants obtained from self‐pollinating positive selection marker‐free T0 plants containing the R304S mutation in the PDS locus as a homozygote were selected, and T2 progenies derived from homozygote R304S plants were subjected to further analysis.

As controls, over‐expression plants of WT and mutated PDS with R304S were prepared. Quantitative RT‐PCR revealed that transcription levels of the PDS gene in WT and R304S mutated PDS over‐expressing plants were sixfold and sevenfold higher than in WT, respectively (Figure 6a). On the other hand, there was no significant difference in PDS gene transcription level between WT and homozygote plants carrying a PDS gene with the R304S mutation (PDS‐GT #11/I‐SceI #20‐1, Figure 6a). Plants over‐expressing the PDS gene with R304S mutation showed the highest resistance to the PDS‐inhibiting herbicides fluridone and norflurazon (PDS‐R304S Ox, Figure 6b,c) and plants over‐expressing WT PDS did not show resistance to either herbicide (PDS‐WT Ox, Figure 6b,c). Compared with WT, GT plants with the R304S mutation remained green but the herbicide resistance level of GT plants was much less than that of PDS‐R304S over‐expressing plants.

Figure 6

PDS‐inhibiting herbicide‐sensitivity test. (a) Transcription levels of PDS mRNA. WT, wild‐type Nipponbare; PDS‐WT Ox, over‐expression plants of WT PDS; PDS‐R304S Ox, over‐expression plants of PDS with R304S mutation; PDS‐GT#11/I‐SceI #20‐1, homozygote plants of R304S mutation in endogenous PDS gene. Data shown are means ± SD of three biological replicates. (b) Fluridone‐ and Norflurazon‐treated plants. From left to right, are WT, PDS‐WT Ox, PDS‐R304S Ox, PDS‐GT #11/ I‐SceI #20‐1. (c) Chlorophyll content in aerial part of the PDS‐inhibiting herbicide‐treated plants shown in (b). Data are given as means calculated from three replicates and error bars represent SD.

Discussion

Applying positive–negative selection, we obtained 30 G418 resistant calli from 1032 treated calli and, in 6 of these 30, HR‐mediated GT events were detected. In our previous study, GT efficiency was calculated as the ratio of calli with HR‐mediated GT per callus selected by positive–negative selection. From this point of view, GT efficiency (GT‐positive calli per G418‐resistant calli: 20%, 6/30, Table 1) in this study was higher than previous studies (0.053%–5.3%) (Dang et al., 2013; Moritoh et al., 2012; Nishizawa‐Yokoi et al., 2014; Ono et al., 2012; Terada et al., 2002, 2007; Yamauchi et al., 2009). However, four out of six GT calli did not contain the R304S mutation, and the AA to GG transition altering a PacI site to HpaI were partially induced in all six GT calli. Such partial induction of point mutations has been reported in other GT studies (Endo et al., 2007; Johzuka‐Hisatomi et al., 2008; Nishizawa‐Yokoi et al., 2014) and can be explained by the occurrence of mismatch correction of the heteroduplex molecules formed between the genomic DNA and targeting vector by DNA mismatch repair, or dissolution of D‐loop structures between the positive selection marker cassette and the R304S mutation. Details of the possible mechanism are shown in Johzuka‐Hisatomi et al. (2008).

To eliminate the positive selection marker cassette precisely and efficiently, we implemented several strategies: (i) I‐SceI recognition sequences were placed at both ends of the NPTII expression cassette for marker excision; (ii) two I‐SceI sites were reverse orientated to avoid simple ligation; and (iii) PDS sequence bordering the I‐SceI sites had a 30‐bp overlap to allow controllable break‐induced SSA repair. Following transformation with the I‐SceI expression construct, 19% of regenerated plants (9/48) showed excision of the positive selection marker (Figure 3a, Table 2). However, in five out of these nine plants, the R304S mutation was not detected, and finally two regenerated plants with precise marker elimination were obtained. One speculation to explain the disappearance of R304S is that, after positive selection marker elimination by I‐SceI expression, DNA broken ends were repaired via HR using the intact PDS allele as a template DNA. In this study, we performed GT and positive selection marker elimination in succession. However, extended cell culture reduces regeneration efficiency, so regenerate GT plants (T0) and use calli induced from T1 seeds for positive selection marker elimination seems to be the steady way for efficient positive selection marker elimination.

In this study, we applied a combination of I‐SceI‐mediated marker elimination and SSA‐mediated DNA repair for seamless restoration at the break point. A similar marker elimination system was reported in mammals as follows: In induced pluripotent stem cells (iPSCs), unique CRISPR‐Cas9 protospacer sequences nested between the selection marker and engineered microhomologies are cleaved after gene targeting, engaging microhomology‐mediated end‐joining and scarless excision (Kim et al., 2018). If we can design sgRNAs to cleave the junction between the plant genome and exogenous positive selection marker cassettes, engineered microhomologies may not be needed because simple ligation of DNA ends often occurs.

From the viewpoint of marker elimination efficiency, the piggyBac‐mediated marker elimination system, in which precision marker elimination occurred in more than 90% of regenerated plants possessing the hyPBase transgene (Nishizawa‐Yokoi et al., 2015) was far superior to the I‐SceI and SSA‐mediated system because precision marker elimination occurred in only 2 regenerated plants out of 92 regenerated plants harbouring I‐SceI expression constructs (2%). However, we need only a few GT plants, since the targeted gene can be modified precisely as expected, and the I‐SceI and SSA‐mediated marker elimination system has no limitation on the location of the positive selection marker. In general, the positive selection marker should be located near the desired mutations, which we want to introduce in an endogenous genome to avoid crossover between the positive selection marker and the desired mutations. In this respect, our I‐SceI and SSA‐mediated marker elimination system has an advantage because marker cassettes can be located anywhere, unlike the piggyBac system, which requires a TTAA sequence at the location of the positive selection inserted. To save time, we completed positive–negative selection‐mediated GT and I‐SceI‐mediated positive selection marker elimination in one generation. However, consecutive transformation of two vectors extends the culture period and reduces the regeneration efficiency of GT plants with precise positive selection maker elimination. Considering positive selection marker elimination efficiency, regenerating GT plants only once and then using calli induced from T1 seeds for positive selection marker elimination, or crossing T1 plants with stable transgenic plants of I‐SceI expression construct, would be a better approach.

In Saccharomyces cerevisiae, 30‐bp of a homologous sequence at each end of a DNA fragment is sufficient to integrate the fragment into a linearized plasmid (Hua et al., 1997). In this study, we demonstrated that 30‐bp of a homologous sequence at either end of an I‐SceI cleavage site is enough for successful scarless marker elimination in rice. It is known that long DNA repeats makes it easy to recombine but long, repeated DNA often induces homology‐mediated recombination even without artificial DNA cleavage (Figure S1 and S2), thus increasing the risk of positive selection marker deletion before completing selection of GT cells.

Recently, novel genome editing tools named prime editors were developed to generate precise editing without the requirement for DSBs or donor DNA (Anzalone et al., 2019). The major component of prime editors is a fusion of a reverse transcriptase and a Cas9 nickase (nCas9). A prime editing guide RNA (pegRNA) was designed to mediate site‐specific nicking by nCas9, and then served as a template for reverse transcriptase to install desired mutations. Prime editors could efficiently produce all possible base conversions and small insertions in a wider targeting range with limited byproducts in human cells (Anzalone et al., 2019) and plants (Li et al., 2020; Lin et al., 2020; Tang et al., 2020; Xu et al., 2020a; Xu et al., 2020b). Prime editing is a powerful tool for precise genome editing but still has limitations in large insertions and deletions. So, selection of an appropriate strategy is needed.

In a previous study, we revealed that rice plants homozygous for the acetolactate synthase (ALS) gene with W548L and S627I mutations created via GT showed hyper‐tolerance to bispyripac‐sodium (BS) compared to plants over‐expressing an ALS gene with the same mutations (Endo et al., 2007). In contrast to ALS, resistance levels of rice plants with a modified endogenous PDS gene as homozygote toward PDS‐inhibiting herbicides were somewhat lower than those of plants over‐expressing the modified PDS gene (Figure 6b,c). These results imply that not only the ratio, but also the amount of herbicide‐insensitive PDS enzyme significantly influences herbicide resistance levels. Rodríguez‐Leal et al. (2017) proved that CRISPR/Cas9‐mediated genome editing of promoters generates diverse cis‐regulatory alleles that provide beneficial quantitative variation for tomato breeding. Modification of both cis‐regulatory elements and coding regions makes it possible to over express herbicide‐resistant PDS.

The finding that not only amino acid substitution but also increased expression level is needed to produce enough resistance to PDS‐inhibiting herbicide may be one of the reasons to explain the difference in the emergence rates of weeds resistant to ALS‐ and PDS‐inhibiting herbicides. Resistance to ALS‐inhibiting herbicides due to a point mutation in the ALS gene has been reported in many weed species (Powles and Yu, 2010). The high‐level resistance conferred by an amino acid substitution in ALS allows plants with a heterozygous mutation to survive and propagate under herbicide stress. Compared to this, the fact that neither point mutations nor over‐expression confer marked resistance may keep heterozygous mutants unselected under PDS‐inhibiting herbicide selection. This may partly explain the very rare occurrence of weed resistance to PDS‐inhibiting herbicides, although other factors, especially a large difference in areas where ALS and PDS‐inhibiting herbicides are applied (Busi et al., 2017), would play more important roles.

Materials and methods

Vector construction

The GT vector, pGT_PDS_R304S was constructed using the binary vector pKO4 (Nishizawa‐Yokoi et al., 2014) with two DT‐A gene expression cassettes (maize polyubiquitin 1 promoter + DT‐A + rice heat shock protein (hsp) 16.9a terminator and rice elongation factor‐1a promoter + DT‐A + rice hsp 16.9b terminator), attL1, attL2 and ccdB sequence. The T‐DNA region of pKOD4 can be shown as follows: LB‐Pubi::DT‐A::Thsp16.9a‐attL1‐ccdB‐attL2‐Pef::DT‐A::Thsp16.9b‐RB. A 2.7‐kb PDS gene fragment was amplified using primers Asc‐PDS‐F and Pac‐I‐SceI‐PDS‐R as 5′ homology arm and cloned into pENTR (R1‐R4), yielding pE (R1‐R4) PDS5′. The PacI site in the 5′ homology arm was changed to a HpaI site using a QuickChange Site‐Directed Mutagenesis Kit (Agilent, CA, USA, https://www.agilent.com). Another 2.7‐kb PDS gene fragment was amplified using primers Asc‐I‐SceI‐PDS‐F and Pac‐PDS‐R as the 3′ homology arm and cloned into pENTR (R3‐R2), yielding pE (R3‐R2) PDS3′. A positive selection marker cassette, 4.1‐kb fragment containing the rice (Oryza sativa) actin terminator::CaMV35S promoter::NPTII::rice heat shock protein 17.3 terminator was cloned in pENTR (L4–L3), yielding pE (L4‐L3) NPTII. pE (R1‐R4) PDS5′, pE (L4‐L3) NPTII, pE (R3‐R2) PDS3′ were combined and cloned into the pKOD4 vector, yielding pGT_PDS_R304S.

The I‐SceI expression vector, pZH_I‐SceI was based on the XVE vector pER8 (Zuo et al., 2000). The I‐SceI coding sequence codon optimized for rice was inserted into the target expression cassette of pER8 to make pER‐I‐SceI. The region between the right and left border was inserted into pPZP202 to create pZH_I‐SceI.

Plant materials and Agrobacterium‐mediated transformation

Agrobacterium‐mediated transformation of rice was performed as described previously (Toki et al., 2006). Calli transformed with Agrobacterium harbouring pGT_PDS_R304S were selected on callus induction (N6D) medium solidified with 0.8% agar (FUJIFILM Wako Pure Chemical Corporation, Osaka, Japan, http://www.wako‐chem.co.jp/) containing 35 mg/L G418 (Nacalai tesque, https://www.nacalai.co.jp) and 25 mg/L meropenem (Wako Pure Chemical Industries) for 1 month. GT candidate calli were transferred to N6D medium without G418 and meropenem and cultured for another 1 month. For marker excision, GT candidate calli were transformed with Agrobacterium to introduce an expression vector, pZH_I‐SceI encoding inducible I‐SceI, and were selected on N6D medium with 50 mg/L hygromycin (Wako Pure Chemical Industries), 25 mg/L meropenem and 5 µm β‐estradiol for 1 month. For regeneration, transgenic calli were transferred to regeneration medium with 25 mg/L meropenem, and shoots arising from callus were transferred to Murashige and Skoog (MS) medium without phytohormones.

PCR screening of GT candidate calli and sequencing analysis for detecting mutations in the PDS gene

After a 1‐month selection period, genomic DNA was extracted from small pieces of G418‐resistant calli transformed with Agrobacterium harbouring a GT vector using Agencourt Chloropure (Beckman Coulter, CA, USA) according to the manufacturer’s protocol. PCR amplifications were performed with PrimeSTAR GXL (Takara Bio Inc. Shiga, Japan) using primer sets as follows: PDS‐F1 and Posi‐R1 for amplifying the 5′ homology region, Posi‐F1 and PDS‐R1 for amplifying the 3′ homology region. PCR products of the5′ homology region were sequenced by PDS‐F2 to detect the PacI ‐> HpaI mutation, and PCR products of the 3′ homology region were sequenced by PDS‐R4 to detect the R304S mutation.

PCR screening of positive selection marker‐free regenerated plants and sequencing analysis to confirm precise SSA‐mediated DSB repair

Genomic DNA was extracted from regenerated plants transformed with Agrobacterium harbouring an I‐SceI expression vector using Agencourt Chloropure (Bechman Coulter) according to the manufacturer’s protocol. PCR amplifications were performed with PrimeSTAR GXL (Takara) using primer sets PDS‐F1 and Posi‐R1, and plants without PCR amplification were considered as positive selection‐marker‐free GT plants. For selecting regenerated plants with the R304S mutation by SSA‐mediated marker elimination, PCR products amplified using PDS‐F2 and PDS‐R3 were sequenced.

Southern blot analysis

Genomic DNA was extracted from leaves of seedlings using the Nucleon Phytopure extraction kit (GE Healthcare, IL, USA, http://www3.gehealthcare.com/) according to the manufacturer’s protocol; 3 μg genomic DNA was digested with SacI and fractionated in a 1.0% agarose gel. Southern blot analysis was performed according to the digoxigenin (DIG) Application Manual (Roche Diagnostics, Tokyo, Japan, http://www.roche.com/). Specific DNA probes for PDS and NPTII were synthesized with a PCR DIG probe synthesis kit (Roche Diagnostics) according to the manufacturer’s protocol, using the primers PDS‐F5 and PDS‐R2, nptII + 23Fw and nptII + 527Rv, respectively.

Preparing plants over‐expressing the rice PDS gene with/without R304S mutation

The coding region of the PDS gene, with 3‐kb promoter region and 1‐kb terminator region, was amplified using Nipponbare genomic DNA as a template and cloned into pUCAP (van Engelen et al., 1995) using SacI and SalI sites. The R304S mutation was induced using a QuikChange II Site‐Directed Mutagenesis Kit (Agilent, CA, USA). Cloned PCR products with/without the R304S mutation were excised with AscI and PacI and cloned into the binary vector pZK2 harbouring an NPTII expression cassette (Kuroda et al., 2010) to prepare PDS‐WT Ox and PDS‐R304S Ox vectors. Agrobacterium‐mediated transformation of rice was performed as described previously (Toki et al., 2006) and transgenic calli were selected on G418‐containing medium. T2 plants that were confirmed to possess the transgene by PCR were used for quantitative RT‐PCR and PDS‐inhibiting herbicide‐susceptibility tests.

RNA extraction and reverse transcriptase‐polymerase chain reaction analysis

Total RNA was extracted from rice plants using an RNeasy Plant Mini Kit (Qiagen, Venlo, Netherlands, http://www.qiagen.com/). First‐strand cDNA was synthesized using ReverTra Ace (TOYOBO, Osaka, Japan, http://www.toyobo‐global.com/) with oligo (dT20) primer. The cDNA encoding PDS was amplified from the first‐strand cDNA using primers PDS‐F3 and PDS‐R4 and sequenced to detect the R304S mutation. For quantitative RT‐PCR, primer pairs PDS‐F4/PDS‐R5 and Act‐F2/Act‐R2 were used for amplifying PDS and Actin genes. PDS gene expression in each RNA sample was normalized to expression of the internal control gene, actin.

PDS‐inhibiting herbicide‐susceptibility test

Sterilized seeds were sown on MS medium containing 0, 0.2, 0.5 and 1 μM fluridone (FUJIFILM Wako Pure Chemical Corporation, Osaka, Japan, http://ffwk.fujifilm.co.jp/en/index.html) or 0.2, 0.5 and 1 μM Norflurazon (FUJIFILM Wako Pure Chemical Corporation, Osaka, Japan) and grown in a growth chamber at 29 °C under continuous light. Photographs were taken 1 week after herbicide treatment.

Chlorophyll extraction and measurement

Aerial parts (50 mg) of the seedling were harvested and cut into 1–2 mm pieces. These plant materials were soaked in 1 mL of N,N′‐dimethyl formamide and kept for 3 days in complete darkness. Supernatants were stored in another sampling tube. Absorbance of the extracts was measured at 663.8 and 646.8 nm with a spectrophotometer (Ultrospec 6300 pro; GE Healthcare, https://www.gehealthcare.com). Concentrations of chlorophyll a plus b in the extracts were calculated from these values through the following equations, as reported by Porra et al. (1989).

Conflict of interest

The authors declare that they have no conflict of interest.

Author contributions

M. Endo and S. Toki designed experiments. M. Endo performed gene targeting experiments. S. Iwakami selected target gene and mutations introduced by gene targeting and performed herbicide‐sensitivity assay.

Figure S1 Reporter constructs of intrachromosomal homologous recombination.

Figure S2 Comparison of intrachromosomal homologous recombination efficiency between different length of overlapping and direction of I‐SceI recognition site.

Table S1 Primers used in this study.

Click here for additional data file.