Generation of targeted mutant rice using a CRISPR ‐Cpf1 system

1The CRISPR‐Cas9 system has emerged as a precise and versatile technique for genome editing, and mounting research demonstrates that efficient editing could be induced by this system in various plant organisms, such as Arabidopsis, rice, wheat, soya bean, maize, tobacco and tomato (Belhaj et al.,; Bortesi and Fischer,; Liu et al., ). Furthermore, the development and customization of plant CRISPR/Cas9 toolboxes or platforms result in more powerful molecular tools for fundamental research and crop breeding (Lowder et al.,; Ma et al.,; Mao et al.,; Tang et al.,; Wang et al.,; Xie et al.,; Xing et al., ). Rice is the most frequently used crop species for testing and applying CRISPR‐Cas9 tools. Several agronomic traits of rice, including yield, fertility, architecture, biotic and abiotic stress response, and herbicide tolerance, have been successfully modified by targeted mutations in one or multiple major genes (Ikeda et al.,; Li et al.,,,; Osakabe et al.,; Sun et al.,; Wang et al.,; Xu et al.,,; Zhou et al., ).

2The recently identified type V CRISPR‐Cas system CRISPR‐Cpf1 was found to mediate targeted genome modification in mammalian cells (Hur et al.,; Kim et al.,; Zetsche et al., ). Cpf1 is an RNA‐guided endonuclease. Both crRNA and tracrRNA are required for Cas9‐mediated targeting. In contrast, a single ~44‐nucleotide (nt) crRNA with a 5′‐located direct repeat sequence and a spacer sequence that complements the target are sufficient to guide the DNA cleavage of Cpf1 (Zetsche et al., ). Moreover, Cpf1 recognizes the thymidine‐rich protospacer‐adjacent motif (PAM) and makes staggered cuts distal to the PAM site (Fonfara et al., ). Here, we demonstrated that efficient targeted mutagenesis can be achieved through the stable expression of certain types of crRNA and Cpf1 in rice (Oryza sativa), a major crop.

3First, the coding sequence of a Cpf1 ortholog from Lachnospiraceae bacterium ND2006 (LbCpf1; NCBI accession number: NZ_JNKS01000011, gene locus_tag: T521_RS08385) was codon‐optimized for expression in rice, and a nuclear‐localization signal was attached to the 3′ ends (Data S1). To test whether crRNA‐Cpf1 could induce genome mutagenesis in plants, we selected a target sequence in an exon of the rice phytoene desaturase (OsPDS) gene as a spacer to design crRNAs through the fusion of direct repeat sequences and constructed it into an engineered binary vector (Figures 1a and b and S1). After the transgenic plants were generated, the target region was examined using the T7 endonuclease (T7E1) assay (Figure S2a) and Sanger sequencing (Figure S3), and the results revealed that targeted mutagenesis could be achieved (Figure 1c and Table 1). Two putative null mutants were identified (Figure 1c) and, as expected, both exhibited an albino phenotype (Figure 1d). Another target located in the rice Bentazon‐sensitive‐lethal (OsBEL) gene was used to further confirm the genome editing of Cpf1 (Figures S2b and 4). Consistent with the OsPDS target, the mutants were efficiently induced through the transformation of LbCpf1 and its cognate crRNAs (Table 1). Cpf1 processes precursor crRNAs (pre‐crRNAs) into mature crRNAs to guide the endonuclease (Fonfara et al., ). Most CRISPR‐Cpf1 studies in mammalian cells have taken advantage of mature (processed) crRNAs (Hur et al.,; Kim et al.,; Zetsche et al., ). However, no mutant was identified from the transgenic plants generated by targeting either OsPDS or OsBEL (Table 1) directly using the mature crRNA sequence, suggesting that the mutagenesis efficiencies of mature crRNA‐Cpf1 complexes were low in the plants. By contrast, mutants were more frequently identified in the lines transformed with pre‐crRNAs (Table 1). We found that 13.6% and 20% of the transgenic plants targeting OsPDS and OsBEL, respectively, were mutated using pre‐crRNAs with the full‐length repeat‐spacer sequence (pre‐crRNA type I). Moreover, mutation frequencies of 21.4% and 41.2% were generated at the OsPDS and OsBEL targets, respectively, using longer pre‐crRNAs with the full‐length repeat‐spacer‐repeat sequence (pre‐crRNA type II). The higher mutation frequency of pre‐crRNAs in rice is consistent with the in vitro observation that the pre‐crRNA‐Cpf1 complexes exhibit stronger binding activity and increased nuclease activity than do mature crRNA‐Cpf1 (Fonfara et al., ).

4We identified several differences between the behaviour of Cpf1‐mediated mutagenesis and that of the previously reported Cas9 system. Compared with the common short indels (1~2 base pairs, bp) generated by Cas9 in rice (Zhang et al., ), most of the mutations generated by Cpf1 were relatively long (Figures 1c and S4), which might be caused by nonhomologous end‐joining repair on the 4~5‐nt 5′ overhangs resulting from the staggered cutting of Cpf1. In addition, several reports have indicated that homozygous mutations can be easily generated by CRISPR‐Cas9 in T0 plants (Shan et al.,; Zhang et al., ), which likely resulted from the Cas9‐induced DNA cleavage in the first cell. However, homozygous mutated lines generated by Cpf1 were not observed; instead, the mutated lines were either chimeric or heterozygous (Table 2). Because a heterozygous genotype could be generated by mutation in only a portion of cells, the results suggest that the Cpf1‐induced mutants might be largely somatic. Similar mosaicism has been frequently observed in Cpf1‐mutated mice (Kim et al., ), suggesting that Cpf1 should induce DNA cleavage after the first division of the embryogenic cell and might require a longer acting period than Cas9 in vivo.

5The off‐target effect of CRISPR‐Cpf1 might be minimal compared with that of CRISPR‐Cas9 in mammalian cells (Hur et al.,; Kim et al.,; Kleinstiver et al., ). To determine the off‐target effect of Cpf1 in plants, we selected genome sites with a high sequence similarity to the OsPDS or OsBEL targets (Table S1). For each target, five transgenic plants with on‐target mutations were separately analysed by direct Sanger sequencing of the potential off‐target sites. After examining the homologous sites with fewer than 7‐nt mismatches, we did not detect off‐target mutations. These results further support the finding that the CRISPR‐Cpf1 system, with careful target selection, is highly specific in vivo.

6To determine whether the Cpf1‐generated mutation could be transmitted through the germ‐line, we analysed the T1 generation populations of several mutated T0 lines and found that the majority of the progeny were still mutants (Table 3). Furthermore, most of the T0 mutation types were identified in T1 plants, suggesting that CRISPR‐Cpf1‐mediated genome mutagenesis is heritable. Fragment‐specific PCR showed that the T‐DNA insertion could be segregated out in certain T1 mutants (Fig. S5), suggesting that similar to the CRISPR‐Cas9 system, CRISPR‐Cpf1 could rapidly generate transgene‐free, genome‐modified crops (Zhang et al., ).

7The mouse was the first organism to be mutated with the CRISPR‐Cpf1 system. The mutants were obtained by transiently introducing a crRNA‐Cpf1 protein complex or a crRNA‐Cpf1 mRNA mixture (Hur et al.,; Kim et al., ). The data obtained in this study show that the expression of CRISPR‐Cpf1 through stable transformation of the engineered vector can generate heritable targeted modifications in the rice genome. We believe that in plant species with reliable genetic transformation systems, the stable expression of CRISPR‐Cpf1 should efficiently produce mutants more rapidly and with a lower cost compared with the transient strategy and that this technique constitutes a robust and ready‐to‐use approach for plant genome mutagenesis. Due to the substantial differences between Cpf1 and Cas9 in terms of target recognition and DNA cleavage, CRISPR‐Cpf1 not only provides a new and alternative method for plant targeted mutagenesis but also greatly enhances the scope and precision of crop genome editing.

8The coding sequence of LbCpf1 was codon‐optimized for rice and synthesized by GeneWiz (Suzhou). The CRISPR‐Cpf1 binary vector was developed from the pHSN400 plasmid (Xing et al., ). The Ubi promoter, rice codon‐optimized Cpf1 and 35S terminator (35‐T) were used instead of the 35S promoter, Cas9 and Nos terminator in pHSN400 through HindIII/NotI, NotI/SacI and SacI/EcoRI double digestions, respectively. A cassette for OsU3‐driven crRNA expression was synthesized (Data S1) and inserted into the plasmid by HindIII digestion to construct the CRISPR‐Cpf1 binary vector.

9To target the specific genome sites of OsPDS and OsBEL, we used 25‐ to 31‐nt sequences with the PAM sequence at their 5′‐ends as spacers. The spacers were fused with different repeat sequences to construct crRNAs (Figure S1). In accordance with a previous protocol (Xing et al., ), the double‐stranded crRNAs were assembled from two complementary oligos (Table S2) and inserted into the vector by BsaI digestion. The colonies were positively selected using kanamycin and negatively selected using spectinomycin and then further confirmed by sequencing.

10The vectors were introduced into a pSOUP‐containing agrobacterium strain EH105 (Hellens et al., ). For each vector, 600 (for the mature crRNA vector) or 300 (for the pre‐crRNA vector) embryonic calluses of rice (Oryza sativa L. ssp. japonica cv. Nipponbare) were infected with the agrobacterium according to a previously described protocol (Duan et al., ). The transgenic plants were selected and regenerated under selection with 50 mg/L hygromycin. After two to three weeks of rooting, the T0 plants were transplanted into soil and grown in a glasshouse at 30 °C. The progeny were generated through strict self‐pollination of each individual T0 event.

11The genomic DNA from the leaves of rice plants was isolated using the CTAB method. The genomic region containing the target site was amplified in the corresponding transgenic plants using site‐specific primers (Table S2) and the high‐fidelity HiFi PCR mixture (TransGen Biotech, Beijing). To detect the mutation, we mixed the PCR product of each transgenic plant with the product generated from the wild‐type plant and then analysed it using the T7E1 assay following a standard protocol (New England Biolabs, MA, USA). Moreover, the PCR products were sequenced directly using the respective forward primer or after cloning into a pEASY‐T vector (TransGen Biotech, Beijing). For each putative mutant event, at least 10 colonies were Sanger sequenced. Specific primers for the LbCpf1, crRNAs and hygromycin phosphotransferase (HPT) cassette were also designed (Table S2) to detect the presence of the T‐DNA fragment.

12The potential off‐target sites were predicted using the online tool Cas‐OFFinder (Bae et al., ). The sites of each target in five randomly selected mutant lines were examined by site‐specific PCR and direct Sanger sequencing. The sequences of the sites and related PCR primers are listed in Tables S1 and S3.