Versatile genome editing using Type I-E CRISPR-Cas3 in rice

Abstract

The Type I-E CRISPR-Cas3 derived from Escherichia coli (Eco CRISPR-Cas3) can introduce large deletions in target sites and is available for mammalian genome editing. The use of Eco CRISPR-Cas3 in plants is challenging because seven CRISPR-Cas3 components (six Cas proteins and CRISPR RNA) must be expressed simultaneously in plant cells. To date, application has been limited to maize protoplasts, and no mutant plants have been produced. In this study, we developed a genome editing system in rice using Eco CRISPR-Cas3 via Agrobacterium-mediated transformation. Deletions in the target gene were detected in 39%–71% of transformed calli by polymerase chain reaction (PCR) analysis, and the frequency of alleles lacking a region 7.0 kb upstream of the protospacer adjacent motif sequence was estimated as 21%–61% by quantifying copy number by droplet digital PCR, suggesting that mutant plants could be obtained with reasonably high frequency. Deletions were determined in plants regenerated from transformed calli, and stably inherited to the progenies. Sequencing analysis showed that deletions of 0.1–7.2 kb were obtained, as reported previously in mammals. Interestingly, deletions separated by intervening fragments or with short insertion and inversion were also determined, suggesting the creation of novel alleles. Moreover, we demonstrated C to T base editing based on Type I-E CRISPR-Cas3 in rice, whereas base editing based on Type I-C and Type I-F2 CRISPR-Cas3 has been reported previously only in human cells. Overall, Eco CRISPR-Cas3 could be a promising genome editing tool for gene knockout, gene deletion, base editing, and genome rearrangement in plants.

Introduction

The sequence-specific nucleases of the CRISPR-Cas system are useful tools for genome editing in both basic research and molecular breeding in many organisms, including plants. Targeted mutagenesis using CRISPR-Cas9—the most widely used system—enables gene knockout and modification of gene expression because CRISPR-Cas9 introduces short insertions and deletions (Gilbertson et al. 2025). Larger deletions are required for the removal of targeted promoter regions, transposons, and gene clusters. For example, a 3.8-kb region of the rice Tos17 retrotransposon that has a long terminal repeat sequence at both ends was deleted completely by use of a single guide RNA (gRNA) targeting the long terminal repeat sequence (Saika et al. 2019). The frequency of transformed calli with polymerase chain reaction (PCR)-confirmed deletions was estimated at 10%–90% (Saika et al. 2019). If the region to be deleted is not flanked by repeated sequences, simultaneous DNA cleavage at both ends of the region to be removed using paired gRNAs can be applied. In rice, Zhou et al. (2014) reported deletion of a 245-kb region harboring 10 genes involved in diterpenoid synthesis using paired gRNAs. However, a previous study in our laboratory using low-performance gRNA(s) found no deletions between two target sites (Mikami et al. 2016). Thus, paired gRNAs that enable the induction of efficient breaks are required for successful targeted deletions.

CRISPR-Cas is divided into two classes: Class 2 CRISPR-Cas consists of a single Cas effector such as Cas9 and Cas12, whereas Class 1 CRISPR-Cas consists of multiple effector proteins. Class 1 Type I CRISPR-Cas induces DNA digestion in two steps: target sequence recognition by the complex of CRISPR-associated complex for antiviral defense (Cascade) with CRISPR RNA (crRNA), and subsequent DNA cleavage by Cas3 nuclease recruited by the Cascade complex. Class 1 CRISPR-Cas has been thought to be hard to use as a genome editing tool because of the requirement to express multiple CRISPR-Cas components in cells, although it is the major system in bacteria and archaea when compared with Class 2 CRISPR-Cas (Yoshimi and Mashimo 2022). Class 1 Type I CRISPR-Cas is divided into seven subtypes: I-A to I-G. Recently, the application of some Class 1 Type I CRISPR-Cas systems to genome editing in eukaryotes has been reported. For example, genome editing in human cells was reported using Type I-A (Hu et al. 2022), I-B and I-C (Tan et al. 2022), I-D (Osakabe et al. 2021), and I-E (Dolan et al. 2019, Morisaka et al. 2019). Many Class 1 CRISPR-Cas systems enable the introduction of large deletions (several kb), although their positions and sizes are not strictly determined. Class 1 CRISPR-Cas could be used to eliminate unnecessary genes, while both Cas9 and Cas12 enable the introduction of small insertions/deletions (indels) (Liu et al. 2022). In addition, Class 1 CRISPR-Cas has the potential to decrease off-target mutations due to longer spacers (32–37 nt) compared with Cas9 and Cas12 (Yoshimi and Mashimo 2022). Thus, Class 1 CRISPR-Cas has unique characteristics, and is expected to be an attractive genome editing tool also in plants. However, successful examples of genome editing using CRISPR-Cas3 in plants are limited to rice (Type I-C, Li et al. (2023)), maize (Types I-C and E, Li et al. (2023)), and tomato (Type I-D named as TiD, Osakabe et al. (2020)). Osakabe et al. (2020) succeeded in genome editing in tomato using TiD—the Type I-D CRISPR-Cas derived from Microcystis aeruginosa—and showed that these mutations were inherited to the next generation. The mutation patterns introduced by TiD were unique: both small indels (1–75 bp) and 2.5–7.3 kb of bidirectional and large deletions. In the TiD system, Cas10d has nuclease activity; Cas3d has no nuclease activity because it lacks a nuclease domain (Osakabe et al. 2020). The dual mutation patterns could be due to the characteristics of Cas10d nuclease (Wada et al. 2022). Li et al. (2023) applied Dvu I-C—a Type I-C CRISPR-Cas derived from Desulfovibrio vulgaris str Hildenborough—to genome editing in rice and maize, and showed the genetic inheritance of deletions introduced by Dvu I-C. Type I-C Cascade complex is a relatively compact system and enables the introduction of unidirectional large deletions (Yoshimi and Mashimo 2022). To overcome the unpredictability of the size of deletions induced by Class 1 CRISPR-Cas, Li et al. (2023) established a system with the introduction of targeted large deletions using paired crRNAs; they also reported that the transient expression of Eco CRISPR-Cas3—a Type I-E CRISPR-Cas derived from Escherichia coli—induced large deletions by using paired crRNAs in maize protoplasts.

Type I-E CRISPR-Cas consists of six effector proteins: Cas5, Cas6, Cas7, Cas8, and Cas11 (which form the Cascade complex), and Cas3 nuclease/helicase. With Eco CRISPR-Cas3, AAG, TAG, GAG, AGG, and ATG are recognized as a protospacer adjacent motif (PAM) sequence (Morisaka et al. 2019). After scanning the PAM sequence by EcoCas8 in the Cascade complex, an R-loop structure is formed by hybridizing the target DNA strand with crRNA. EcoCas3 is recruited at the R-loop structure and induces a nick in the non-target DNA strand. Double-stranded DNA is then unwound upstream of the PAM site by the helicase activity of EcoCas3. Target DNA is degraded by a combination of cleavage of the target strand in trans by collateral activity of EcoCas3 and repetitive cleavage of the non-target DNA strand in cis. As a result, long (0.5–80 kb), unidirectional deletions starting 0–400 bp upstream of the PAM sequence can be introduced (Yoshimi and Mashimo 2022). Successful applications of Eco CRISPR-Cas3 to genome editing have been reported in animal systems such as human cells, pig fibroblasts, and mice and rat zygotes (Morisaka et al. 2019, Miyagawa et al. 2022, Yoshimi et al. 2024). In mice and rat zygotes, DNA-free genome editing using mRNA and a ribonucleoprotein (RNP) complex of Eco CRISPR-Cas3 was reported (Yoshimi et al. 2024). In plants, however, to date there is only one report of Eco CRISPR-Cas3-mediated genome editing, which was in maize protoplasts, as described above (Li et al. 2023). As yet, there are no reports of the production of mutated plants by Type I-E CRISPR-Cas.

CRISPR-Cas9 has also been applied to precise modification, such as base editing (Komor et al. 2016, Nishida et al. 2016, Gaudelli et al. 2017) and prime editing (Anzalone et al. 2019). Besides targeted mutagenesis, Class 1 CRISPR-Cas has also been applied to precise modification. For example, base editing using cytidine deaminase and adenosine deaminase fused to Cas5, 7, 8, and 11 derived from Dvu I-C was analyzed in human cells (Li et al. 2024b). Adenosine deaminase fused to the N-terminus of Cas8 and a combination of apolipoprotein B mRNA editing enzyme catalytic subunit 1 (APOBEC) fused to the N-terminus of Cas8 and uracil DNA glycosylase inhibitor (UGI) fused to the C-terminus of Cas7 were found to be functional as base editors in several fusion proteins (Li et al. 2024b). In addition, base editing using adenosine deaminase fused to Cas5 or Cas7 of Type I-F2 CRISPR-Cas3 has also been reported in human cells (Guo et al. 2024). As yet, there are no reports of base editing based on the Class 1 CRISPR-Cas Cascade in plants.

Here, we succeeded in producing regenerated plants with large deletions in a target gene using Eco CRISPR-Cas3 in rice by Agrobacterium-mediated transformation, and demonstrated that deletions confirmed in regenerated plants were inherited stably to the next generation. In addition, we determined the deletion frequency introduced by Eco CRISPR-Cas3 accurately using droplet digital PCR (ddPCR). The frequency of alleles lacking a region at 2.6 kb and 7.0 kb upstream of the PAM sequence in transformed calli was estimated at 21.1%–60.8% and 39.5%–74.1%, respectively, suggesting a reasonably high frequency of mutant regenerated plant production. Furthermore, we demonstrated successful C to T substitution in rice calli using cytidine deaminase fused to EcoCas8. These results showed the effectiveness of Eco CRISPR-Cas3-mediated genome editing for the introduction of heritable large deletions and base substitutions in plants.

Results and Discussion

Targeted mutagenesis by Eco CRISPR-Cas3 in rice calli

First, to investigate whether Eco CRISPR-Cas3 can be used for targeted mutagenesis in rice, we constructed binary vectors harboring expression cassettes of six proteins and crRNA (Fig. 1a). To achieve high and simultaneous expression in rice cells, the following strategy was employed: (1) codon optimization of Cas effectors with bipartite nuclear localization signals (bpNLS) for rice. The type of NLS and position of the NLS–Cas9 fusion is known to affect genome editing frequency; for example, it was reported that bpNLS fused to both N- and C-termini of Cas9 greatly improved genome editing frequency in Arabidopsis (Develtere et al. 2024). (2) Strong expression in rice calli using the maize ubiquitin promoter. (3) Minimization of T-DNA length by separation of six Cas gene expression cassettes into two binary vectors: Cas537 (expressing EcoCas3, EcoCas5, and EcoCas7) and Cas6811 (expressing EcoCas6, EcoCas8, and EcoCas11) to facilitate vector construction; (4) higher DNA cleavage activity using crRNA consisting of pre-crRNA harboring an 86-nt leader sequence, and two 29-nt repeat sequences with an intervening 32-nt spacer sequence following a previous study in human cells (pre-crRNA (LRSR) in Morisaka et al. (2019)). To facilitate detection of genome editing events, OsPDS was selected as a target gene because knockout of OsPDS results in albino phenotype and makes it easier to distinguish genome-edited cells (Fig. 1b). In addition, a 32-nt spacer with 5′-AAG-3′ PAM was selected because AAG PAM showed the highest DNA cleavage activity in human cells (Morisaka et al. 2019). Rice calli were infected with a mixture of Agrobacterium harboring binary vectors Cas537 and Cas6811 + OsPDS, and selected with hygromycin and G418 for 1.5 months. Some pieces of albino calli were found in transformed calli, suggesting that biallelic mutations had occurred in the OsPDS (Fig. 1c). To confirm whether deletions were introduced in OsPDS in transformed calli, PCR analysis was performed using primers designed to be longer upstream of the PAM because EcoCas3 introduces unidirectional deletion (Morisaka et al. 2019). Multiple fragments shorter than 8.7 kb were amplified in transformed calli, although there were no bands other than 8.7 kb corresponding to wild type in non-transformants (Fig. 1d). The frequency of transformed calli in which shorter amplicons were found was estimated as 39.4%–48.2% (Supplementary Table S1). These results suggested that large deletion occurred specifically in OsPDS by Eco CRISPR-Cas3 with reasonably high frequency. Similar results were obtained in experiments using a different target designed in DROOPING LEAF (OsDL, Nagasawa et al. (2003)) (Supplementary Fig. S1A–C).

Figure 1.

Eco CRISPR-Cas3-mediated targeted mutagenesis in rice calli. (a) Eco CRISPR-Cas3 vector targeting OsPDS. Top, EcoCas3 expression cassette. Grey and black shading show the 5′-untranslated region of rice alcohol dehydrogenase and nucleus localization signal (NLS), respectively. Middle, binary vector Cas537; bottom, binary vector Cas6811; ZmUbi, maize ubiquitin-1 promoter; Tpea3A, pea rbcs terminator; hpt, hygromycin phosphotransferase; nptII, neomycin phosphotransferase II; LB, left border; RB, right border. (b) Structure of OsPDS locus. Black and white boxes show coding regions and untranslated regions of OsPDS, respectively. Orange arrowheads and purple bar show PCR primers and the probe for Southern blot analysis, respectively. Green text, PAM; blue text, target sequence for Eco CRISPR-Cas3. (c) Rice calli transformed with binary vectors harboring Eco CRISPR-Cas3 with the crRNAs for OsPDS shown in (a). (d) PCR analysis of transformed calli using primers shown in (b). Red arrowhead, fragment size amplified in wild type. M, size marker; NC, negative control (no template); NT, non-transformant.

Inheritance to progenies of mutations introduced by Eco CRISPR-Cas3

Hygromycin- and G418-resistant calli transformed with Cas537 and Cas6811 + OsPDS were transferred to the regeneration medium. A total of 24 independent regenerated plantlets were obtained successfully from transformed calli and 12 plants showed albino phenotype (Supplementary Fig. S2A). Of these, five independent regenerated plants, including three lines of albino plants, were selected randomly and analyzed. No fragments and shorter fragments were found in albino plants #2 and #3, respectively, although there were no bands other than 8.7 kb corresponding to wild type in green plants #4 and #5 and non-transformants (Supplementary Fig. S2B). In addition, Sanger sequencing of PCR fragment #1 confirmed a 391-bp deletion showing that plants in which deletions were introduced in OsPDS had been obtained successfully. To confirm the genetic inheritance of mutations found in regenerated plants, we focused on OsWx encoding granule-bound starch synthase for amylose synthesis in endosperm, and designed targeted mutagenesis using Eco CRISPR-Cas3 (Fig. 2a). A total of seven independent regenerated plants were obtained from rice calli transformed with Cas537 and Cas6811 + OsWx. A large deletion was found in one line by PCR analysis (Fig. 2b). A total of 12 T1 seeds were obtained from this regenerated plant. Although non-transformed seeds were stained blue, none of the T1 seeds showed iodine staining, suggesting knockout of OsWx (Fig. 2c). PCR analysis in 11 T1 seedlings (except for #8, which did not germinate) showed that, compared with wild type, shorter fragments (6.8 kb) were amplified in six lines and no amplicons were amplified in five lines (Fig. 2d). Sanger sequencing of shorter amplicons confirmed a 2.8 kb deletion in the OsWx locus (Fig. 2e). These results showed that large deletions introduced by Eco CRISPR-Cas3 were inherited successfully to the next generation, as with TiD and Dvu I-C (Osakabe et al. 2020, Li et al. 2023). Considering that all T1 seeds showed waxy phenotype (Fig. 2c and d), it was assumed that biallelic mutations (2.8 kb deletion and no amplification) were introduced in T0 plants. Thus, it is estimated that biallelic mutants were obtained at 14% (1/7 regenerated plants) using Eco CRISPR-Cas3, suggesting that genome editing by Eco CRISPR-Cas3 is efficient enough to produce mutant rice plants. We showed previously that targeted mutagenesis frequency in regenerated plants was improved by extension of the culture period of CRISPR-Cas9-transformed calli (Mikami et al. 2015). Longer cultivation of transformed calli helps obtain CRISPR-Cas3-mediated mutant regenerated plants more efficiently, although it could lead to an increased risk of somaclonal mutations.

Figure 2.

Inheritance in rice plants of mutations introduced by Eco CRISPR-Cas3. (a) Structure of OsWx locus. Details as in Fig. 1b. (b) PCR analysis of T0 plants transformed with Eco CRISPR-Cas3 for OsWx. Details as in Fig. 1d. Red arrowhead, fragment size amplified in wild type. (c) T1 seeds of #2 stained with iodine; NT, non-transformant. (d) PCR analysis of progenies in T1 generation. Red arrowhead, fragment size amplified in wild type. (e) Representative sequence chromatogram of #2 T1 plants.

All-in-one vector for Eco CRISPR-Cas3

Here, seven gene expression cassettes for six Cas effector proteins and crRNA of Eco CRISPR-Cas3 were prepared and carried separately on two binary vectors (Fig. 1a and Supplementary Fig. S1A). It was expected that all components could be expressed strongly using this strategy. However, the T-DNAs were large (15.7 kb and 14.2 kb in Cas537 and Cas6811 + OsPDS, respectively), and both binary vectors need to be co-transformed into rice cells. This might represent a bottleneck to expanding Eco CRISPR-Cas3-mediated genome editing to plants in which it is harder to obtain transformed and genome-edited cells than in rice. In a previous report on induced pluripotent stem cells, an expression vector in which Eco CRISPR-Cas3 proteins were combined with 2A self-cleavage peptides was used to introduce deletions (Morisaka et al. 2019). In addition, vectors to express multiple Cas effector proteins of TiD and Dvu I-C combined with 2A peptides can be used for genome editing in plants (Osakabe et al. 2020, Li et al. 2023). Thus, an all-in-one vector carrying four expression cassettes, Cas758 (EcoCas7, EcoCas5, and EcoCas8 combined with 2A peptides), Cas1163 (EcoCas11, EcoCas6, and EcoCas3 combined with 2A peptides), crRNA, and hygromycin phosphotransferase was constructed (Supplementary Fig. S3). Moreover, the crRNA expression cassette was inserted 400 bp away from the right border of T-DNA in the all-in-one vector, whereas it was inserted 90 bp from the right border of T-DNA in co-transformation vectors. An all-in-one vector harbors 15.9 kb of T-DNA. PCR analysis performed in rice calli transformed with all-in-one vectors targeting for OsPDS and OsDL estimated the frequency of transformed calli in which shorter amplicons were found at 55.3%–70.8% (Supplementary Table S1). These frequencies were considered high enough to obtain plantlets regenerated from calli transformed with all-in-one vectors.

Thus, in addition to co-transformation vectors carrying seven expression cassettes of Eco CRISPR-Cas3, a compact all-in-one vector can also be applied to genome editing in rice. The deletion frequency seemed to be slightly higher than co-transformation with two binary vectors (Supplementary Table S1). This might be attributed to a combination of the following points. First, the crRNA expression cassette was protected in an all-in-one vector because regions adjacent to the border sequences in T-DNA are often lacking. Second, hygromycin and G418 were used for selection of cells co-transformed with Cas537 and Cas6811 vectors, whereas only hygromycin was used with the all-in-one vector. The number of cell divisions and cell activity are thought to differ between calli harboring these vectors, although the calli propagation period was the same (1.5 months). The propagation of hygromycin- and G418-resistant calli seemed to be slower than that of hygromycin-resistant calli. Speed of propagation might affect mutation frequency by Eco CRISPR-Cas3.

Detailed analysis of mutation patterns induced by Eco CRISPR-Cas3

To characterize Eco CRISPR-Cas3-mediated deletions at the target site, the amplicons shown in Fig. 1d were cloned and sequenced. Sanger sequencing of clones selected randomly confirmed eight types of mutations (a–h) (Fig. 3a). Deletions with a start point in the region between 278 bp upstream and 53 bp downstream of the PAM sequence occurred, and deletions were large (136–7218 bp). In previous reports on Type I-E CRISPR-Cas3 derived from Thermobifida fusca (Tfu CRISPR-Cas3) in human cells (Dolan et al. 2019) and Dvu I-C in maize (Li et al. 2023), mutation patterns were categorized into four groups: (Group I) one seamless junction, (Group II) one junction with insertion or partial inversion, (Group III) one junction with mutation(s), and (Group IV) two junctions. In this study, three and four types of mutations in Groups I and IV, respectively, were found. Types (a), (b), and (g) were single large deletions (6617 bp, 7218 bp, and 4383 bp, respectively). On the other hand, mutations in types (c), (d), (e), and (f) consisted of two large deletions separated by intervening fragments of 8–361 bp. CRISPR-Cas3 induces DNA double-strand digestion by repetitive unwinding of target DNA and cleavage of DNA strands in cis and trans (Yoshimi et al. 2022). Cas3 can attack target sequences many times because, unlike Cas9, Type I-E Cas3-mediated deletions often occur upstream of the PAM sequences (Dolan et al. 2019, Morisaka et al. 2019, Yoshimi and Mashimo 2022). This suggested the possibility that these mutations resulted from some rounds of EcoCas3-mediated deletion events. Surprisingly, type (h) consisted of four large deletions separated by intervening fragments of 44 bp, 103 bp, and 124 bp that have not been found in previous reports (Dolan et al. 2019, Li et al. 2023). In types (d) and (e), a single large deletion concomitant with 85 bp and 4 bp of insertion, respectively, was observed. In type (e), inversion of a 156-bp fragment was observed. These unusual mutation patterns were also observed in Tfu CRISPR-Cas3-mediated genome editing in human cells (Dolan et al. 2019). Interestingly, a BLAST search showed that the 85 bp fragment in type (d) matched sequences located 23.8 kb upstream of the PAM sequence, suggesting the possibility that the genomic region including an 85 bp fragment 23.8 kb distant from PAM sequence accesses the target region on Eco CRISPR-Cas3 DNA digestion, although the detailed molecular mechanism is unclear. There were 2–6 bp microhomology sequences at the junctions in 7 of 15 patterns of deletion detected in this study, suggesting that microhomology-mediated end joining is involved in DNA double-strand repair in CRISPR-Cas3-mediated deletion as reported in Types I-C and I-D CRISPR-Cas3-mediated deletion in plants (Osakabe et al. 2020, Li et al. 2023).

Figure 3.

Deletions introduced by Eco CRISPR-Cas3 targeting OsPDS in rice calli. (a) Sequence analysis in transformed calli of #2. PCR fragments were sequenced with the primer shown (black arrowhead). Grey boxes and black solid lines indicate regions in which sequences were confirmed and not confirmed, respectively. A yellow box indicates an inverted region. Black numbers above grey and yellow boxes show their lengths. Green dotted lines/numbers and red arrowheads/numbers indicate the regions/lengths of deletion and insertion, respectively. Blue numbers indicate the distance from the PAM at the site where the deletion started. Purple asterisks and letters indicate microhomology sequences at the junctions. (b) Southern blot analysis of SacI-digested calli transformed Eco CRISPR-Cas3 with crRNAs for OsPDS using the probes shown in Fig. 1b and Supplementary Fig. S1B. Arrowhead, band corresponding to wild type. M, size marker; NT, non-transformant. (c) Estimation of deletion size by ddPCR. The y-axis shows the ratio of amplicon concentrations at sites located 33.0 kb, 7.0 kb, and 2.6 kb upstream, and 6.0 kb downstream of the PAM sequence to those at the control site (OsAct1). Bars, technical error (maximum and minimum Poisson distribution for the 95% confidence interval).

The length of sequence deleted by CRISPR-Cas3 is uncontrollable. Approaches to introduce precise large deletions with CRISPR-Cas3 using paired crRNA and inactivated Cas9, thus preventing continuous deletion by Cas3, have been reported (Li et al. 2023, Li et al. 2024a). However, it can also be desirable to produce mutant series harboring various sizes of deletion. It has been discussed that Dvu I-C tends to introduce ‘simple’ large deletions compared with Eco CRISPR-Cas3 (Li et al. 2023). Also in the case of TiD, single large deletion and single large deletion with small insertion were determined in tomato and human cells, respectively, although bidirectional deletions were introduced, unlike Types I-C and I-E (Osakabe et al. 2020, Osakabe et al. 2021). In this study, sequence analysis determined the presence of deletions concomitant with genome rearrangements such as short insertion and inversion in addition to simple deletions (Fig. 3a). This suggests that Eco CRISPR-Cas3 enables the production of both simple deletion and various kinds of alleles based on large deletions, although the frequency might be low. Rice calli stably transformed with Eco CRISPR-Cas3 were cultured for 1.5 months and analyzed in this study. On the other hand, CRISPR-Cas3 worked transiently in previous studies because expression vectors for Dvu I-C were transformed into maize protoplasts (Li et al. 2023) and RNP of Tfu CRISPR-Cas3-infected human cells (Dolan et al. 2019). This suggests that continuous and repetitive Eco CRISPR-Cas3 attack on target regions in stably transformed calli results in longer and more complex mutation patterns. Optimization of callus culture conditions and period could potentially generate various mutation patterns. Promoter engineering by genome editing in agronomical important genes has been reported in crops such as tomato (Rodríguez-Leal et al. 2017) and rice (Cui et al. 2023). Dozens of gRNAs were prepared to create promoter deletion alleles in these reports. Eco CRISPR-Cas3 could also be used to create promoter alleles harboring large deletions with/without short insertion and inversion by use of a single gRNA in rice (Fig. 3a). Eco CRISPR-Cas3 could possibly induce more dynamic rearrangements because the deletion size and position are flexible compared with Class 2 CRISPR-Cas. In this scenario, technologies to introduce shorter deletions, i.e. dozens to hundreds of base pairs, might be needed to expand the application of Eco CRISPR-Cas3-mediated genome editing.

Estimation of deletion frequency and length induced by Eco CRISPR-Cas3

Calli transformed with Eco CRISPR-Cas3 vectors are mosaics of various mutant cells. Electrophoresis of PCR products is a simple and easy way to find deletion events due to Eco CRISPR-Cas3. However, PCR bias makes it difficult to accurately calculate deletion frequencies or determine deletion length. Here, to estimate deletion frequency and length more precisely, the following analyses were performed. First, Southern blot analysis performed to visualize deletion events more clearly showed that shorter, and a few bands longer, than 14.1 kb corresponding to wild type were detected using a probe for OsPDS but not for OsDL as a control (Fig. 3b), supporting the view that deletions were introduced in this locus. Similar results were obtained in experiments using a different target designed in OsDL (Supplementary Fig. S1D). Second, the copy numbers of specific short regions around the target sequence were quantified using the droplet digital PCR (ddPCR) method, and the ratios to an internal control (OsAct1) were calculated (Fig. 3c). The ddPCR method was used to detect whether the target region was present or absent in experiments using Dvu I-C (Li et al. 2023). ddPCR can be used to estimate deletion length by quantification of presence or absence in multiple regions. Here, three regions at 33.0 kb, 7.0 kb, and 2.6 kb upstream of the PAM sequence and one region at 6.0 kb downstream of the PAM sequence were analyzed. The ratios of copy number of a region 2.6 kb upstream of the PAM sequence to those in OsAct1 were between 25.9% and 60.5% in all 12 samples analyzed, suggesting that the frequency of alleles lacking a region 2.6 kb upstream of the PAM sequence was 39.5%–74.1%. Similarly, the frequency of alleles lacking a region at 7.0 kb upstream of the PAM sequence was 21.1%–60.8%, although that frequency was 1.5%–25.2% in a region 6.0 kb downstream of the PAM sequence. This result is consistent with previous reports that Type I-E CRISPR-Cas3 introduces unidirectional deletion in human cells (Dolan et al. 2019, Morisaka et al. 2019).

Off-target mutations by Eco CRISPR-Cas3

The lower potential for off-target mutations due to the longer target sequence in Class 1 CRISPR-Cas compared to Class 2 CRISPR-Cas is considered one of the merits of Class 1 CRISPR-Cas (Yoshimi and Mashimo 2022). To investigate this, potential off-target sites of the 27-nt sequences used in this study were searched in the rice genome using the program GGGenome (https://gggenome.dbcls.jp/). This search focused on 27-nt sequences because every 6th, 12th, 18th, 24th, and 30th base from the 5′ end of spacer sequences are not involved in target recognition, although the length of spacer sequence target is 32-nt in Eco CRISPR-Cas3. Sequences with mismatches of fewer than 6 nt (more than 20% mismatches) and potential PAM sequences (AAG, TAG, GAG, AGG, and ATG) were picked up (supplementary Table S2). As a result, 1 and 3 sites with 5-nt mismatches to spacer sequences in OsPDS and OsDL were identified, respectively. However, there are no potential PAM sequences at the 5′ end of spacer sequences. Moreover, 14, 36, and 1 sites with 6 nt mismatches to spacer sequences in OsPDS, OsDL, and OsWx were identified. Among them, 1 and 2 sites with 6-nt mismatches to spacer sequences harboring potential PAM sequences in OsPDS and OsDL were identified. PCR analysis was performed to confirm whether or not deletions were introduced in these off-target candidates. As expected, shorter bands than wild type were not detected, indicating that large deletions were not introduced by Eco CRISPR-Cas3 (Fig. 4 and Supplementary Fig. S4). Type I-E CRISPR-Cas has been reported to introduce no small indels (Dolan et al. 2019, Morisaka et al. 2019, Li et al. 2023, Yoshimi et al. 2024). Thus, this result suggested that off-target mutations were not introduced in these sites, although detailed analysis such as whole genome sequence may be needed.

Figure 4.

Off-target analysis of Eco CRISPR-Cas3-mediated targeted mutagenesis for OsPDS in rice calli. (a) Off-target locus of crRNA for OsPDS on chromosome 7. Mismatches and bases at 6, 12, 18, 24, and 30 nt from PAM that do not involve target recognition are highlighted in light blue and yellow, respectively. (b) PCR analysis of transformed calli using the primer set shown as orange arrowheads in A.

C to T base substitution by target-AID fused to EcoCas8

Young et al. (2019) applied a Cascade derived from Type I-E CRISPR-Cas to an artificial transcriptional activator for the recognition of a targeted sequence in maize. They showed that the transcriptional activation domain fused to Cas8 enhanced target gene activation compared with activation domain fusions to Cas5 or Cas6 (Young et al. 2019). In addition, successful examples of base editing using cytidine deaminase and adenosine deaminase fused to Dvu I-C Cas8 have been reported in human cells (Li et al. 2024b). These results suggest that Cas8 can accept fusion of additional domains without impairing the target sequence recognition of Cascade. To investigate whether a base editor using Eco Cascade can function in plant cells, an expression cassette comprising a deaminase fused to EcoCas8 was constructed (Fig. 5a). Here, target-AID—a C to T base editor consisting of cytidine deaminase derived from sea lamprey (PmCDA) and UGI—was selected because both PmCDA and UGI can be fused to the C-terminal of Cas9 nickase (Nishida et al. 2016, Shimatani et al. 2017). Co-transformation vectors Cas57 (EcoCas5 and EcoCas7) and Cas68CBE11 (EcoCas6, target-AID fused to EcoCas8, and EcoCas11) with a crRNA expression cassette for OsPDS or OsDL were constructed (Fig. 5a) and used to transform rice calli. To confirm whether C to T substitutions were introduced in target sites in transformed calli, PCR fragments were sequenced directly using the Sanger method. The sequence chromatograms were checked visually, and those in which the peaks of other bases overlapped more than half the height of peaks of the original base in the target sequence were categorized as positive lines in which substitutions had been introduced in the target. Substitution frequency and pattern are summarized in Supplementary Table S3. In this analysis, C to T substations were confirmed at locations 4 bp, 5 bp, and 7 bp downstream of the PAM. This result was similar to a previous report of a base editor based on the Dvu I-C cascade, although APOBEC and UGI were fused separately to Cas8 and Cas7, respectively (Li et al. 2024b). The substitution sites found in this study were reasonable as Cas8 functions to recognize the PAM sequence. Besides C to T substitutions in the target sequences, G to A substitutions outside of the target sequence were found. These were thought to be due to C to T substitutions on the opposite strand, as reported in a base editor based on the Dvu I-C cascade (Li et al. 2024b). The overall frequency of transformed calli in which base substitutions were found was estimated as 22.9%–38.0%, suggesting that substitution occurred with reasonably high frequency.

Figure 5.

Base editing using target-AID based on Eco CRISPR-Cas3 Cascade in rice calli. (a) Target-AID vector based on Eco CRISPR-Cas3 Cascade targeting OsPDS. Top, target-AID fused to EcoCas8 expression cassette. Horizontal stripes show the linker between Cas8 and PmCDA-UGI. Middle, binary vector Cas57; bottom, binary vector Cas68CBE11 + OsPDS; T17.3, rice heat shock 17.3 protein terminator. Details as in Fig. 1a. (b) A representative image of a sequence chromatogram. The red arrowhead shows the position of the C to T substitution. Green text, PAM; blue text, target sequence for Eco CRISPR-Cas3.

Conclusion

In this study, we demonstrated targeted mutagenesis by Eco CRISPR-Cas3 in rice (Fig. 1 and Supplementary Fig. S1), and showed inheritance of the introduced deletion to the progenies (Fig. 2 and Supplementary Fig. S2). Off-target mutations due to CRISPR-Cas3 can be kept at a low frequency because CRISPR-Cas3 recognizes longer target sequences compared with Class 2 CRISPR-Cas. Off-target mutations, large deletions in this case, were not detected in rice (Fig. 4). As reported previously in human cells (Dolan et al. 2019, Morisaka et al. 2019), several kb of unidirectional deletion can be introduced into the target region by Eco CRISPR-Cas3 in rice (Fig. 3a and c). Moreover, deletions separated by intervening fragments and deletions with short insertion and inversion were also determined (Fig. 3a). Furthermore, we also demonstrated C to T substitution in rice using the target-AID system based on Eco Cascade (Fig. 5 and Supplementary Table S3). Our results suggest that targeted mutagenesis and base editing using Eco CRISPR-Cas3 can create novel gene alleles in addition to gene knockouts and deletions.

Materials and Methods

Oligonucleotides

Primers used in this study are listed in Supplementary Table S4.

Vector construction

Co-transformation vectors

To construct vectors for Eco CRISPR-Cas expression cassettes, rice codon optimized EcoCas3, EcoCas5, EcoCas6, EcoCas7, EcoCas8, and EcoCas11-coding sequences including bpNLS were synthesized by GeneArt Gene Synthesis (Thermo Fisher Scientific). These vectors inserted into SpeI/SacI-digested pE(R4-R3) ZmUbi_OsFnCpfI_Tpea3A (Endo et al. 2016) by an In-Fusion reaction (Takara), yielding vectors for Cas expression driven by the maize ubiquitin promoter. PCR fragment amplified fragments (using primer sets, attB1 f/attB4 r, attP4 f/attP3 r, and attB3 f/attB2 r) with these vectors as templates were cloned into entry clones, pDONR221 P1-P4, pDONR221 P4r-P3r, and pDONR221 P13-P2, respectively, by BP reaction (Invitrogen), yielding pDONR221(L1-L4)ZmUbiCas5Tpea3A, pDONR221(L1-L4)ZmUbiCas6Tpea3A, pDONR221(R4-R3)ZmUbiCas3Tpea3A, pDONR221(R4-R3)ZmUbiCas8Tpea3A, pDONR221(L3-L2)ZmUbiCas7Tpea3A, and pDONR221(L3-L2)ZmUbiCas11Tpea3A (Supplementary Sequence). To construct a destination vector harboring an nptII-expression cassette, an nptII fragment amplified with primer set nptII f2/nptII r2 was inserted into XhoI-digested pZD202Hyg vector (Kwon et al. 2012) by an In-Fusion reaction, yielding pZD202Km. Entry clones were cloned into pZD202Hyg and pZD202Km using LR clonase II plus, yielding Cas537 and Cas6811 (pZD202Hyg harboring Cas5, Cas3, and Cas7 expression cassettes and pZD202Km harboring Cas6, Cas8, and Cas11 expression cassettes, respectively). To construct a crRNA expression vector, a pOsU6Cas3 gRNA vector harboring OsU6-2 promoter, 86-nt leader sequence, and two 29-nt repeat sequences were synthesized by GeneArt Gene Synthesis. Annealed oligonucleotides in Supplementary Table S4 were ligated into BsaI-digested pOsU6Cas3 gRNA (Supplementary Sequence). The crRNA expression cassettes amplified with a primer set, gRNA PmeI f/gRNA PmeI r were inserted into PmeI-digested Cas537 and Cas6811, yielding Cas537 + OsDL, Cas6811 + OsPDS, and Cas6811 + OsWx.

All-in-one vector

A pE(L3-L2)P2X35S::I-SceI::Thsp vector (Kwon et al. 2012) was digested with AscI and PacI, and inserted into AscI-/PacI-digested pDONR221(L1-L4) vector, yielding pDONR221(L1-L4)2x35S-I-SceIos-T17.3. Four PCR fragments amplified using primer sets P35S-omega f1/RcCAT r2, bpNLS f1/T2A-NLS r1, T2A-NLS f1/T2A-NLS r2, T2A-NLS f2/T17.3 r1 with pDONR221(L1-L4)2x35S-I-SceIos-T17.3, pDONR221(L3-L2)ZmUbiCas7Tpea3A, pDONR221(L1-L4)ZmUbiCas5Tpea3A, and pDONR221(R4-R3)ZmUbiCas8Tpea3A, respectively, as templates were cloned into XbaI-/SacI-digested pDONR221(L1-L4)2x35S-I-SceIos-T17.3 by an In-Fusion reaction, yielding pDONR221(L1-L4) 2x35SΩint Cas758 T17.3 (Supplementary Sequence). Three PCR fragments amplified using primer sets ADHUTR-NLS f1/T2A-NLS r3, T2A-NLS f2/r1, T2A-NLS f1/Tpea3A r5 with pDONR221(L3-L2)ZmUbiCas11Tpea3A, pDONR221(L1-L4)ZmUbiCas6Tpea3A, and pDONR221(R4-R3)ZmUbiCas3Tpea3A, respectively, as templates were cloned into SpeI-/SacI-digested pDONR221(R4-R3)ZmUbiCas8Tpea3A by an In-Fusion reaction, yielding pDONR221(R4-R3)ZmUbiCas1163Tpea3A (Supplementary Sequence). Three entry clones, pDONR221(L1-L4) 2x35SΩint Cas758 T17.3, pDONR221(R4-R3)ZmUbiCas1163Tpea3A, and pDONR221(L3-L2)NoshptT35Snos harboring hpt driven by the Nos promoter and 35S/Nos double terminator were cloned into pZD202 (Kwon et al. 2012) using a LR clonase II plus, yielding pZD202 Cas758 + Cas1163 + hpt. The crRNA expression cassettes amplified with a primer set gRNA SpeI f/gRNA SpeI r were inserted into SpeI-digested pZD202 Cas758 + Cas1163 + hpt, yielding all-in-one + OsDL, all-in-one + OsPDS.

Target-AID vector

Two PCR fragments amplified using primer sets Cas8 f4/Cas8 r5 and deadSH f2/linkerNLS r2 with pDONR221(R4-R3)ZmUbiCas8Tpea3A and target-AID vector, a derivative of nSpCas9-NGv1-AID-UGI (Endo et al. 2019), respectively, as templates were cloned into AatII-/SacI-digested pDONR221(R4-R3)ZmUbiCas8Tpea3A by an In-Fusion reaction, yielding pDONR221(R4-R3)ZmUbiCas8CBETpea3A (Supplementary Sequence). Three entry clones, pDONR221(L1-L4)ZmUbiCas5Tpea3A, pDONR221(R4-R3)T17.3 (harboring rice heat shock 17.3 protein terminator), and pDONR221(L3-L2)ZmUbiCas7Tpea3A were cloned into pZD202Hyg using a LR clonase II plus, yielding Cas57. Three entry clones, pDONR221(L1-L4)ZmUbiCas5Tpea3A, pDONR221(R4-R3) ZmUbiCas8CBETpea3A, and pDONR221(L3-L2)ZmUbiCas11Tpea3A were cloned into pZD202Km using an LR clonase II plus, yielding pZD202Hyg and Cas68CBE11.

Transformation

For rice transformation, binary vectors were transformed into Agrobacterium tumefaciens strain EHA105 (Hood et al. 1993) by the electroporation method. Agrobacterium-mediated transformation of rice (Oryza sativa L. cv. Nipponbare) followed our previous reports (Toki 1997, Toki et al. 2006). Briefly, 4-week-old secondary calli derived from mature seeds were co-cultivated with Agrobacterium (mixed Agrobacterium harboring a different binary vector in case of co-transformation) for 3 days. Calli were washed to eliminate Agrobacterium and cultured on an N6D selection medium containing 50 mg/L hygromycin, 35 mg/L G418, and 25 mg/L meropenem in the case of co-transformation, or 50 mg/L hygromycin and 25 mg/L meropenem in the case of an all-in-one vector for 6 weeks. Calli were transferred to fresh selection medium every 2 weeks. Calli growing vigorously on the selection medium were cultured on regeneration medium (ReIII) containing 25 mg/L meropenem, and shoots arising from calli were transferred to hormone-free medium containing 25 mg/L meropenem.

DNA extraction and PCR analysis

Genomic DNA was extracted from transformed calli and leaves of rice using Agencourt chloropure (Beckman Coulter) or Nucleon Phytopure extraction kit (Cytiva) according to the manufacturer’s protocol. PCR analysis was performed with KOD One PCR master mix (TOYOBO) using the primer sets listed in Supplementary Table S4.

Southern blot analysis

Southern blot analysis was performed by following a conventional protocol. SacI-digested genomic DNA (5 μg) from calli was electrophoresed on a 0.7% gel at around 50 V and transferred to positively charged nylon membranes (Roche Diagnostics). Specific DNA probes were prepared using a PCR digoxigenin (DIG) probe synthesis kit (Roche Diagnostics) according to the manufacturer’s protocol using the primer sets listed in Supplementary Table S4. The probe-hybridized membranes were washed using DIG Wash and Block Buffer Set (Roche Diagnostics). Chemical luminescence on the membrane with CDP-Star treatment was detected with a ChemiDoc Touch (Bio-Rad).

d‌dPCR

For ddPCR, 22 μl of reaction mixture containing 2.5 ng of genomic DNA digested with PacI and SacI, ddPCR Supermix for Probes (no dUTP) (Bio-Rad), and PrimePCR Probe Assay (Bio-Rad) were prepared. In this study, PrimePCR Probe Assay: ACT1, rice labeled with HEX, PrimePCR Probe Assay: OS03G0183500 ^*, rice labeled with FAM (−33.0 kb), PrimePCR Probe Assay: OS03G0183900 ^*, rice labeled with FAM (−7.0 kb), PrimePCR Probe Assay: PDS, rice labeled with FAM (−2.6 kb), and PrimePCR Probe Assay: OS03G0184100 ^*, rice labeled with FAM (+6.0 kb) were used (Bio-Rad). Droplet generation using an Automated Droplet Generator (Bio-Rad), PCR using droplets, droplet analysis using a QX200 Droplet Reader (Bio-Rad), and data analysis with QuantaSoft software (Bio-Rad) followed our previous report (Nishizawa-Yokoi et al. 2021).

Sequencing analysis

For Sanger sequence analysis, fragments including target sequences were amplified. To analyze Eco CRISPR-Cas3-mediated deletions at the target sites, PCR products were cloned into pCR-Blunt II-TOPO (Invitrogen) and transformed into E. coli. Colony PCR was performed with KOD One PCR master mix (TOYOBO) using M13 forward and reverse primers. The resultant PCR products were used as a template for sequencing reaction using a BigDye Terminator v3.1 Cycle Sequencing Kit (Thermofisher Scientific) and subjected to sequence analysis using an ABI3130 sequencer (Thermofisher Scientific). Sequence data were analyzed using SnapGene (GSL Biotech LLC).

Author Contributions

H.S. and S.Y. designed the experiments. H.S. and H.N. conducted the experiments and analyzed the data. T. Muranaka, K.Y., T. Mashimo, and S.T. supervised the research. H.S. wrote the manuscript. All authors read and approved of the manuscript.

Conflict of Interest

K.Y. and T. Mashimo are cofounders of C4U Corporation. T. Mashimo is an outside board member of C4U and K.Y. is a scientific advisor to C4U. The remaining authors declare no conflict of interest.

Funding

This work was supported by the Ministry of Agriculture, Fisheries and Food (MAFF) Commissioned project study on ‘Development of new varieties and breeding materials in crops by genome editing’ [Grant Number JPJ008000] and Cross-ministerial Strategic Innovation Promotion Program (SIP), ‘Building a Resilient and Nourishing Food Supply Chain Management for a Sustainable Future’ (funding agency: Bio-oriented Technology Research Advancement Institution) [Grant Number JPJ012287].

Data Availability

The data supporting this study are available in the article and supplementary data.