Enhanced Rice Blast Resistance by CRISPR/Cas9-Targeted Mutagenesis of the ERF Transcription Factor Gene OsERF922

Rice blast is one of the most destructive diseases affecting rice worldwide. The adoption of host resistance has proven to be the most economical and effective approach to control rice blast. In recent years, sequence-specific nucleases (SSNs) have been demonstrated to be powerful tools for the improvement of crops via gene-specific genome editing, and CRISPR/Cas9 is thought to be the most effective SSN. Here, we report the improvement of rice blast resistance by engineering a CRISPR/Cas9 SSN (C-ERF922) targeting the OsERF922 gene in rice. Twenty-one C-ERF922-induced mutant plants (42.0%) were identified from 50 T0 transgenic plants. Sanger sequencing revealed that these plants harbored various insertion or deletion (InDel) mutations at the target site. We showed that all of the C-ERF922-induced allele mutations were transmitted to subsequent generations. Mutant plants harboring the desired gene modification but not containing the transferred DNA were obtained by segregation in the T1 and T2 generations. Six T2 homozygous mutant lines were further examined for a blast resistance phenotype and agronomic traits, such as plant height, flag leaf length and width, number of productive panicles, panicle length, number of grains per panicle, seed setting percentage and thousand seed weight. The results revealed that the number of blast lesions formed following pathogen infection was significantly decreased in all 6 mutant lines compared with wild-type plants at both the seedling and tillering stages. Furthermore, there were no significant differences between any of the 6 T2 mutant lines and the wild-type plants with regard to the agronomic traits tested. We also simultaneously targeted multiple sites within OsERF922 by using Cas9/Multi-target-sgRNAs (C-ERF922S1S2 and C-ERF922S1S2S3) to obtain plants harboring mutations at two or three sites. Our results indicate that gene modification via CRISPR/Cas9 is a useful approach for enhancing blast resistance in rice.

Introduction technology, SSN-based genome editing can achieve complete knockout without incorporating exogenous DNA. To date, successful examples of ZFN-and TALEN-based improvements of agronomically important traits in major crops have been reported [47,[55][56][57][58][59][60]. Here, we report the improvement of rice blast resistance via CRISPR/Cas9-targeted knockout of the ERF transcription factor gene OsERF922 in Kuiku131, a japonica rice variety widely cultivated in northern China.

Plant growth
The japonica rice variety Kuiku131 and all transgenic plants were grown in a net house and greenhouse at 28-35°C in Beijing, or fields in the experimental station under normal growth conditions in Sanya. The experimental station is specialized for genetically modified crops planting permitted by Chinese Ministry of Agriculture. For blast inoculation at the seedling stage, rice seeds were grown in 60 × 30 × 5 cm plastic seedling-nursing trays supplemented with a mineral nutrient solution in a greenhouse maintained at approximately 28-35°C under natural sunlight [61]. Briefly, moistened seeds of wild-type rice and rice mutants (T 2 progeny) were sown in rows (15 seed per row) in triplicate trays.

Rice transformation
The Cas9/sgRNA-expressing binary vectors (pC-ERF922, pC-ERF922S1S2 and pC-ERF922S1S2S3) were transformed into an Agrobacterium tumefaciens strain EHA105 by electroporation. Agrobacterium-mediated transformation of the embryogenic calli derived from the japonica rice variety Kuiku131 was performed according to Hiei et al. [62]. Briefly, hygromycin-containing medium was used to select hygromycin-resistant calli, and then the hygromycin-resistant calli were transferred onto regeneration medium for the regeneration of OsERF922 contains a single exon indicated by gray rectangles. The translation initiation codon (ATG) and termination codon (TGA) are shown. The target site nucleotides are shown in capital letters, and the protospacer adjacent motif (PAM) site is underlined. (B) Schematic diagram of the pC-ERF922 construct for expressing the CRISPR/Cas9 protein C-ERF922. The positions and orientations of the primers Cas9p-F and Cas9p-R, which were used to screen Cas9-free mutants, are indicated by small arrows. The expression of Cas9 is driven by the maize ubiquitin promoter (Ubi); the expression of the sgRNA scaffold is driven by the rice U6a small nuclear RNA promoter (OsU6a); the expression of hygromycin (HPT) is driven by 2 CaMV35S promoters (2 × 35S). NLS: nuclear localization signal; Tnos: gene terminator; LB and RB: left border and right border, respectively. (C) Nucleotide sequences at the target site in the 7 T 0 mutant rice plants. The recovered mutated alleles are shown below the wild-type sequence. The target site nucleotides are indicated as black capital letters and black dashes. The PAM site is underlined. The red dashes indicate the deleted nucleotides. The red capital letters indicate the inserted nucleotides. The numbers on the right indicate the type of mutation and the number of nucleotides involved. "−" and "+" indicate the deletion and insertion of the indicated number of nucleotides, respectively; "−/+" indicates the simultaneous deletion and insertion of the indicated number of nucleotide.

Protoplast assay
Kuiku131 seeds were sterilized with 0.2% potassium permanganate solution for 24 h and then imbibed in water at 37°C for 24 h before germination. Seedlings were grown in cylindrical glass bottles lined with wet toilet paper under a regime of 12 h light (150-200 μmol m -2 s -1 )/12 h dark at 26°C in an incubator for 10-14 days before protoplast isolation. The protoplast isolation and transformation were performed following protocols published by Zhang et al. [63] and Shan et al. [64], respectively.
Healthy and fresh rice stems and sheaths from 50 rice plants were used. A bundle of rice plants were cut together into fine strips approximately 0.5 mm in length using sharp razors. The fine strips were immediately transferred into an enzyme solution (1.5% Cellulase RS, 0.75% Macerozyme R-10, 0.6 M mannitol, 10 mM MES at pH 5.7, 10 mM CaCl 2 and 0.1% BSA). After 5 h digestion with gentle shaking (60-80 rpm) in the dark, an equal volume of W5 solution (154 mM NaCl, 125 mM CaCl 2 , 5 mM KCl and 2 mM MES at pH 5.7) was added, followed by light shaking for 10 sec. The protoplasts were released by filtering through 40 μm nylon meshes into round bottom tubes and were washed 2-3 times using W5 solution. The pellets were collected by centrifugation at 250 g for 3 min. After washing, the pellets were resuspended in MMG solution (0.4 M mannitol, 15 mM MgCl 2 and 4 mM MES at pH 5.7) at a concentration of 2 × 10 6 cells mL -1 , as calculated using a hematocytometer.
Protoplast transformation was carried out in a poly-ethylene glycol (PEG) solution [40% (W/V) PEG 4000, 0.2 M mannitol and 0.1 M CaCl 2 ]. For one sample, 20 μg of plasmid DNA was mixed with 200 μL protoplasts (approximately 4 × 10 5 cells) and 220 μl freshly prepared PEG solution, and the mixture was incubated at room temperature for 20 min in the dark. After incubation, 880 μL of W5 solution was added slowly, and the protoplast cells were harvested by centrifugation at 250 g for 3 min. The protoplast cells were resuspended gently in 2 mL WI solution (0.5 M mannitol, 20 mM KCl and 4 mM MES at pH 5.7) and cultured in 6-well plates in darkness at room temperature for 48 h.
The genomic DNA was extracted from protoplast cells transformed with pC-ERF922 plasmid by the SDS method [65]. The protoplast genomic DNA was subjected to PCR with the gene-specific primer pair E922-KF/E922-KR (S1 Table) to amplify DNA fragments across the target site. Then, the PCR amplicons were cloned into the pEASY-Blunt vector (TransGen Biotech, Beijing, China), and a total of 48 randomly selected colonies were further characterized by Sanger DNA sequencing.

Pathogen inoculation
To evaluate the resistance to M. oryzae at the seedling stage, the inoculation of rice blast fungus M. oryzae was performed according to a method described by Wang et al. [20]. Briefly, [3][4] week-old wild-type and homozygous mutant plants were inoculated by spraying with conidial suspensions (2 × 10 5 conidia mL -1 , 0.02% Tween 20) of M. oryzae isolate 06-47-6. The inoculated plants were grown in an ENCONAIR phytotron at 26°C (95% humidity) in the dark for 24 h, then were grown under conditions of a 16 h/8 h light/dark cycle with 95% humidity. Disease severity was evaluated according to the method described by Fukuoka et al. [68] at 7 d post inoculation (dpi). The area of the lesions was determined for the third leaves of 10 plants of each line. The experiments were repeated three times.
To confirm the presence of disease, all seedlings of wild-type and homozygous mutant lines were transplanted to the field and injection-inoculated at the tillering stage according to a method described by Ma et al. [69]. Briefly, when rice plants grow to 5-6 tillers, conidial suspensions (2 × 10 4 conidia mL -1 ) of M. oryzae were injected into the rice stem with a syringe until the suspension emerged from the heart-leaf. The injection sites on the stems were approximately 10 cm from the top of the tillers. Five tillers were injected for each plant. Disease severity was evaluated according to method described by Kobayashi et al. [70] at 7 dpi. The length of the lesions was assessed on the inoculated leaves of five tillers for each line. The experiments were repeated three times.

Agronomic trait characterization
Wild-type and homozygous T 2 mutant lines were grown in the field under normal growth conditions in Beijing. Agronomic traits were characterized by measuring plant height, flag leaf length and width, the number of productive panicles, panicle length, the number of grains per panicle, the seed setting rate, and thousand seed weight after the rice had reached maturity. Five plants were investigated for each line.

CRISPR/Cas9 design and the assessment of gene-editing activity
To design a CRISPR/Cas9 (C-ERF922) targeting the OsERF922 gene in rice, a 20-bp nucleotide sequence containing the initiation codon of the open reading frame of OsERF922 was chosen as the target site (ERF922-S2) ( Fig 1A). The predicted Cas9 cleavage site in the coding region of the gene was seven base-pairs downstream from the ATG initiation codon. The binary plasmid pC-ERF922 (Fig 1B) was then constructed based on the CRISPR/Cas9 vector described by Ma et al. [38]. To test the gene-editing efficacy of C-ERF922, rice protoplasts were transformed with pC-ERF922, and genomic DNA was extracted to amplify the DNA fragment containing the target site. PCR amplicons generated with the primers E922-KF and E922-KR (S1 Table) were cloned into the pEASY-Blunt vector to isolate the colonies for sequencing. Three mutants (6.3%, S3 Fig) were recovered from 48 randomly selected colonies. Sequencing revealed that the mutation in colony C1 was a single nucleotide substitution; in colony C2 was a 5-bp deletion; and in colony C3 was a 30-bp insertion (S3 Fig). These observations showed that the C-ERF922-expression construct in pC-ERF922 exhibits gene-editing activity in rice protoplasts and can be used for creating mutant rice plants.

Recovery of rice plants with mutations in OsERF922
The pC-ERF922 construct was used to transform the rice variety Kuiku131 by Agrobacteriummediated transformation, with the goal of enhancing its blast resistance by gene-specific editing. We obtained 50 positive transgenic (T 0 ) plants and analyzed the target site in 21 of the plants (S4 Fig). Direct Sanger-sequencing of the target-containing amplicons followed by decoding via the DSD method [66,67] showed that among the 21 plants, there were 16 bi-allelic mutations, 3 homozygous mutations, 1 heterozygous mutation, and 1 chimeric mutation ( Table 1). Based on allele mutation types, more than half (64.3%, 27/42) of the mutations were nucleotide deletions, 23.8% (10/42) of the mutations were nucleotide insertions, and 11.9% (5/ 42) of the mutations were simultaneous nucleotide deletions and insertions ( Table 1)

Transmission of C-ERF922-induced mutations from the T 0 to the T 1 and T 2 generations
To determine whether and how the C-ERF922-induced mutations were transmitted to the next generation, 4 bi-allelic (KS2-45, 70, 75, 144), 1 chimeric (KS2-12), 1 homozygous (KS2-27) and 1 heterozygous (KS2-44) T 0 mutant plants ( Fig 1C) were self-pollinated, and their progenies were genotyped at the target site. A total of 120 T 1 plants derived from the T 0 mutant plants were genotyped by PCR and DNA sequencing ( Table 2). We found that all allelic mutations in the T 0 mutant plants were transmitted to the T 1 generation with a transmission rate of 100% (Table 2). In theory, allelic mutations in the bi-allelic T 0 mutant plants should segregate to T 1 plants following Mendelian genetic law (1xx:2xy:1yy). As expected, homozygous genotypes were detected in all T 1 populations derived from the T 0 mutant plants, even when the T 1 segregation pattern of progeny from the chimeric T 0 mutant plant KS2-12 was more diverse and less predictable. For all bi-allelic T 0 mutant plants with the exception of KS2-45, the segregation ratio of [homozygote (xx): bi-allele mutants (xy): homozygote (yy)] in the T 1 populations fit a 1:2:1 ratio shown to be statistically reliable in a chi-square test of the T 1 plants (Table 2). For example, the bi-allelic T 0 mutant plant KS2-70 harbors two mutations [a 23-bp deletion (-23) and a 1-bp insertion (+1)]; its T 1 progenies segregated in a ratio of [11(-23): 20(-23, +1): 10 (+1)], matching the (1xx: 2xy: 1yy) ratio well ( Table 2).

Selection of T-DNA-free mutant rice lines
To investigate the possibility of obtaining rice lines harboring the desired modifications in OsERF922 but without transferred DNA (T-DNA) of the construct pC-ERF922, we designed the Cas9 gene-specific PCR primers Cas9p-F and Cas9p-R (Fig 1B and S1 Table) and performed PCR assays of the T 1 and T 2 plants. All 120 T 1 plants were subjected to PCR assays, and 10 (8.3%) T 1 plants failed to generate a Cas9-specific 531-bp amplicon from the transferred pC-ERF922 construct (Fig 2 and Table 2). Similarly, the PCR assay also failed to detect the pC-ERF922 construct in 16 Table 2). Notably, all 30 T 2 plants derived from the T 1 mutant plant KS2-45-6 failed to generate the Cas9-specific amplicon  (

Resistance to M. oryzae was enhanced in C-ERF922-induced rice mutants
To characterize the blast resistance phenotype of the rice mutants, 6 homozygous mutant T 2 lines (Fig 3A) with different types of allelic mutations were inoculated with the fungal pathogen M. oryzae isolate 06-47-6 at the seedling stage. The leaves of wild-type plants nearly died due to pathogen infection, likely because the pathogenicity of isolate 06-47-6 was very strong, and the wild-type variety was highly susceptible (Fig 3B). Nevertheless, the lesion areas formed by pathogen infection were significantly decreased in all mutant rice lines compared with wildtype plants (Fig 3B). The differences were further evaluated by quantification of the lesion areas and significance analysis using Student's t-test (Fig 3C), which indicated that the mutant rice lines enhanced rice blast resistance. Similarly, lesion lengths formed by pathogen infection were also decreased in the mutant rice lines compared with the wild-type plants at the tillering stage (Fig 3D), and significant difference analysis of quantitative lesion length revealed that all mutant rice lines were significantly different from wild-type plants (Fig 3E). These results indicated that C-ERF922-induced frame shifts in the OsERF922 gene enhanced resistance to M. oryzae in the rice mutants because OsERF922 negatively regulates the blast resistance of rice [18].

The main agronomic traits were not altered in C-ERF922-induced rice mutants
To determine whether mutations in the OsERF922 gene affect agronomic traits, we characterized all 6 homozygous T 2 mutant lines (Fig 3A) by measuring their plant height, flag leaf length and width, the number of productive panicles, panicle length, the number of grains per panicle, seed setting rate, and thousand seed weight. Student's t-test revealed that none of the 6 T 2 mutant lines differed significantly from wild-type plants under normal growth conditions with regard to the agronomic traits investigated (Table 3). The mutagenic frequency and mutagenic frequency of homozygous plants were increased by targeting multiple sites within OsERF922 To examine whether the mutagenic frequency could be increased by targeting multiple sites within one gene, we designed Cas9/two-target-sgRNAs (C-ERF922S1S2) and Cas9/three-target-sgRNAs (C-ERF922S1S2S3) to target two sites (ERF922-S1 and ERF922-S2, Fig 1A and S1  Fig) and three sites (ERF922-S1, ERF922-S2 and ERF922-S3; Fig 1A and S1 Fig) (Table 4, S2 and S3  Tables). Direct Sanger-sequencing of the target-containing amplicons followed by decoding with the DSD method [66,67] showed that among the C-ERF922S1S2-induced mutant plants were 19 (63.3%) plants harboring mutations at both target sites (Table 4 and S2 Table). Furthermore, 47.6% (10/21, Table 5) of the mutants were homozygotes. In addition, among C-ERF922S1S2S3-induced mutant plants, all 27 (90.0%) plants harbored mutations at all three target sites (Table 4 and S3 Table), and 40.7% (11/27, Table 5) of the mutants were homozygotes. These results demonstrated that the mutagenic frequencies increased when targeting more sites within one gene, and Cas9/two-target-sgRNAs resulted in the highest mutagenic frequency in homozygotes (Tables 1 and 5).

Discussion
Genome editing using SSNs provides an opportunity for crop improvement. Thus far, SSNs have been used to improve a variety of important crops, such as rice [56,60], wheat [47], maize [55], soybean [58] and potato [59], by creating specific gene knockouts. However, examples of crop improvement via the creation of novel genotypes, agronomic traits or disease resistance remain limited. The first example was using ZFNs to target the maize IPK1 gene, which resulted in reduced levels of phytate-an anti-nutritional component of feed grains-and reduced phosphate pollution in waste streams from cattle-feeding operations [55]. In rice, the disease-susceptibility gene and the sucrose efflux transporter gene OsSWEET14, which aids in pathogen survival and virulence, was mutated by TALENs to produce disease-resistant rice with normal phenotypes [56]; moreover, using TALENs to target the OsBADH2 gene produced a generation of fragrant rice that contain 2-acetyl-1-pyrroline (2AP), a major fragrance compound [60]. In addition, knocking out all three MILDEW-RESISTANCE LOCUS (MLO) alleles in bread wheat using one pair of TALENs resulted in the creation of stable mutant lines exhibiting broad-spectrum resistance to powdery mildew [47]. In the present study, we used C-ERF922, C-ERF922S1S2 and C-ERF922S1S2S3 to knockout OsERF922 and achieved 42.0%, 70.0% and 90.0% recovery of C-ERF922-, C-ERF922S1S2-and C-ERF922S1S2S3-induced mutant plants, respectively, in T 0 transgenic plants; all of the allele mutations were transmitted to the T 1 and T 2 generations. We obtained more than 20 mutant plants that harbor the desired modification in OsERF922 but not containing the transgene, which was eliminated via segregation in the T 1 and T 2 generations. Inoculation with M. oryzae revealed that blast resistance in the T 2 homozygous mutant lines tested was significantly enhanced compared with that of wildtype plants at both the seedling and tillering stages. In addition, we showed that there was no significant difference between T 2 homozygous mutant lines and wild-type plants with respect to the agronomic traits, such as plant height, flag leaf length and width, the number of productive panicles, panicle length, the number of grains per panicle, seed setting rate, and thousand seed weight. This study provides a successful example of improving rice blast resistance using CRISPR/Cas9 technology. The C-ERF922-induced mutagenic frequency of T 0 plants in this study was 42.0%, similar to those previously reported for CRISPR/Cas9-induced mutations in rice [31,35,38,39]. In addition, the genotypes of these T 0 mutant plants were primarily bi-allelic (76.1%) and  (Table 1), which was also similar to previous observations [29-33, 35-38, 71]. This phenomenon is in stark contrast to that of TALEN-induced mutant genotypes, wherein the heterozygous mutants were more frequent than bi-allele mutants in T 0 plants [60,64,72,73]. This difference might be caused by the different target site cleavage efficiencies of CRISPR/Cas9 and TALENs. Multiple mutations were detected at the target site in the T 0 mutant plant KS2-12 ( Fig 1C). The presence of chimeric mutations in a single mutant plant may result from delayed cleavage in the primary embryogenic cell. This phenomenon has been reported in rice [29,34,35,38], Arabidopsis [74], wheat [47], tomato [49] and maize [42]. In addition, the segregation ratios observed for the T 1 plants derived from KS2-12 were not Mendelian, probably because the chimeric mutations were restricted to somatic cells that did not participate in the production of gametes. Furthermore, the segregation ratios found in T 1 plants obtained from the bi-allelic mutant plant KS2-45 did not conform to a Mendelian ratio, but the T 2 plants derived from the same bi-allelic mutant plant did ( Table 2), indicating that the number of T 1 plants was smaller or that homozygous mutation (11-bp deletion) may be induced through detrimental mutations caused by T-DNA insertion, which has less chance of survival compared with the other types of mutants. Notably, several novel mutations were detected in the T 2 offspring of KS2-44-1 (S5 Fig). This could be explained by the fact that KS2-44-1 was a heterozygote in which the C-ERF922 construct remained active and continually cleaved the target site in T 2 plants, resulting in new mutations. However, we did not detect new mutations in the T 1 offspring of KS2-44, probably due to the existence of only 3 T 1 plants (Table 2).
Both CRISPR/Cas9 and TALENs are effective tools for gene modification; however, each has specific advantages and limitations. Compared with TALENs, CRISPR/Cas9-expressing vectors are much easier to construct and can be competed in just two or three days [38,75], whereas the construction of TALENs typically requires over seven days [76][77][78]. Furthermore, CRISPR/Cas9 induces a much higher mutation rate than TALENs. For example, the frequency of CRISPR/ Cas9-targeted mutagenesis ranged from 21.1% to 66.7% (average 44.4%) for 11 rice genes [35]. Likewise, Ma and colleagues recently reported that the average mutation rate was 85.4% for CRISPR/Cas9-based editing of 46 target sites in rice [38]. However, the frequency of TALEN-targeted mutagenesis ranged from 0-30% overall [47,64,73]. Nevertheless, the requirement for a PAM (-NGG) sequence and the possibility of off-target effects are limitations of CRISPR/Cas9 system. For example, previous studies have demonstrated that off-target effects were common at the level of one nucleotide mismatch in plant species [37,50], fish [79] and human cells [80,81]. In contrast, off-target effects were extremely rare in the event of one nucleotide mismatch for the TALENs-editing system [82,83]. The present study indicates that the CRISPR/Cas9 system is indeed a powerful tool for crop improvement via site-specific genome editing.  Table. Primers used in this study. (PDF) S2 Table. Nucleotide sequences at the target sites in C-ERF922S1S2-induced T 0 mutant rice plants. (PDF) S3 Table. Nucleotide sequences at the target sites in C-ERF922S1S2S3-induced T 0 mutant rice plants. (PDF)