Genome-Wide Association Study Identiﬁes a Rice Panicle Blast Resistance Gene Pb3 Encoding NLR Protein

. Abstract: Rice blast is a worldwide fungal disease that seriously affects the yield and quality of rice. Identiﬁcation of resistance genes against rice blast disease is one of the effective ways to control this disease. However, panicle blast resistance genes, which are useful in the ﬁelds, have rarely been studied due to the difﬁculty in phenotypic identiﬁcation and the environmental inﬂuences. Here, panicle blast resistance-3 ( Pb3 ) was identiﬁed by a genome-wide association study (GWAS) based on the panicle blast resistance phenotypes of 230 Rice Diversity Panel I (RDP-I) accessions with 700,000 single-nucleotide polymorphism (SNP) markers. A total of 16 panicle blast resistance loci (PBRLs) within three years including one repeated locus PBRL3 located in chromosome 11 were identiﬁed. In addition, 7 genes in PBRL3 were identiﬁed as candidate genes by haplotype analysis, which showed signiﬁcant differences between resistant and susceptible varieties. Among them, one nucleotide-binding domain and Leucine-rich Repeat (NLR) gene Pb3 was highly conserved in multiple resistant rice cultivars, and its expression was signiﬁcantly induced after rice blast inoculation. Evolutionary analysis showed that Pb3 was a typical disease resistance gene containing coiled-coil, NB-ARC, and LRR domains. T-DNA insertion mutants and CRISPR lines of Pb3 showed signiﬁcantly reduced panicle blast resistance. These results indicate that Pb3 is a panicle blast resistance gene and GWAS is a rapid method for identifying panicle blast resistance in rice.


Introduction
Rice blast, caused by the fungi Magnaporthe oryzae, is one of the most serious and recurrent diseases in rice-growing regions worldwide [1]. It is estimated that every year, the disease-destroyed rice could feed 60 million people [2,3]. Breeding new resistant varieties with resistance (R) genes are the most effective way to solve the problem [4]. To date, more than 100 rice blast resistance (R) genes have been identified and 35 of them have been cloned [5]. However, panicle blast resistance genes have rarely been studied due to the difficulty in phenotypic identification and the environmental influences [6]. Panicle blast resistance-1 (Pb1) was the first cloned panicle blast resistance gene encoding a coiled-coilnucleotide binding site-leucine rich repeat (CC-NBS-LRR) protein, which was isolated from the indica cultivar Modan and conferred durable and broad-spectrum resistance to rice blast [7]. Pi25 was derived from the indica rice variety Gumei 2 and provided both panicle blast and leaf blast resistance [8]. The Pigm contains two NBS-LRR proteins, PigmR and PigmS, which endow rice with blast resistance and yield balance [9]. Pb2 is an NLR gene gene PiPR1 was identified [36]. Pb2 encodes an NBS-LRR protein, identified in the GWAS of 230 cultivars of RDP-1, which enhanced the resistance of panicle blast and leaf blast without affecting agronomic traits and can be used for rice blast resistance breeding [10]. GWAS analysis was carried out by using broad-spectrum resistant rice variety Digu and 66 no broad-spectrum resistant accessions, and found that the natural variation of bsr-d1 promoter altered resistance to rice blast [37]. bsr-d1 is a C2H2-type transcription factor that is regulated by the MYB transcription factor and enhances blast resistance by inhibiting H 2 O 2 degradation, and has no obvious effect on rice yield and quality. Therefore, bsr-d1 has a broad application prospect in rice breeding for blast resistance.
In this study, we evaluated the panicle blast resistance of 230 cultivars derived from RDP-1 by field injection inoculation with the M. oryzae strain 2014-290-B27. By GWAS analysis of population phenotypes and 700K SNPs, 16 panicle blast resistance loci (PBRL) within three years including one repeated locus PBRL3 located in chromosome 11 were identified. In addition, 7 genes in PBRL3 were identified as candidate genes by haplotype analysis, which showed a significant difference between resistant and susceptible varieties. Among them, one Nucleotide-binding domain and Leucine-rich Repeat (NLR) gene Pb3 was highly conserved in multiple resistant rice cultivars, and its expression was significantly induced after rice blast inoculation. Compared with the wild-type, the T-DNA insertion mutant lines and CRISPR knockout transgenic lines showed significantly reduced resistance to leaf blast and panicle blast. In summary, this study identified a Nucleotide-binding domain and Leucine-rich Repeat (NLR) panicle blast resistance gene Pb3 by GWAS analysis, which can significantly improve rice panicle blast and leaf blast resistance.

Identification of PBRLs by Using GWAS
We performed GWAS for the three experiments with the Efficient Mixed-Model Association eXpedited (EMMAX), which accounts for both population structure and kinship [38]. The quantile-quantile plots showed that the model of the three environments could be used to identify association signals (Supplementary Figure S1). A total of 16 panicle blast resistance loci (PBRLs) mainly distributed on chromosomes 6 and 10 were identified by GWAS with high-quality genetic markers filtered from 700K SNPs ( Figure 3A-C). Among the 16 PBRLs, 9 loci were co-localized with previously mapped or cloned R genes or loci (Table 1). PBRL1 was located at the terminal of chromosome 2 and co-localized with the previously identified Pig(t) [39]. PBRL6 was located in the 20 Mb region of chromosome 4 and contained two reported resistance genes Pi-21 or PBRL-14 locus [10,15]. Among the 16 PBRLs, PBRL3 was the only repeated locus detected both in 2017 and 2019, and it was located in the Pik locus at the end of chromosome 11 [40].

Identification of PBRLs by Using GWAS
We performed GWAS for the three experiments with the Efficient Mixed-Model Association eXpedited (EMMAX), which accounts for both population structure and kinship [38]. The quantile-quantile plots showed that the model of the three environments could

Identification of PBRLs by Using GWAS
We performed GWAS for the three experiments with the Efficient Mixed-Model Association eXpedited (EMMAX), which accounts for both population structure and kinship [38]. The quantile-quantile plots showed that the model of the three environments could or loci (Table 1). PBRL1 was located at the terminal of chromosome 2 and co-localized with the previously identified Pig(t) [39]. PBRL6 was located in the 20 Mb region of chromosome 4 and contained two reported resistance genes Pi-21 or PBRL-14 locus [10,15]. Among the 16 PBRLs, PBRL3 was the only repeated locus detected both in 2017 and 2019, and it was located in the Pik locus at the end of chromosome 11 [40].

Analysis of Candidate Genes for Panicle Blast Resistance
To identify candidate resistance genes underlying the stable PBRL3 locus, we investigated the haplotype block structure based on locus within 500 kb intervals; however, it was difficult to find the LD block in the interval ( Figure 4A-C). According to genome-wide LD, decay rates of indica and japonica were estimated at ∼200 kb [41], and a 200 kb interval (Chr11:27.1 Mb-27.3 Mb) was selected to predict candidate genes. With reference to the Nipponbare genome sequence and gene annotation, 15 candidate genes with SNPs on the promoter or exons region were obtained. Haplotype analysis results showed that 7 candidate genes with different haplotypes had significant phenotypic differences in the natural population and were selected as potential candidate genes for further analysis ( Figure 5A). Among them, LOC_Os11g44990 and LOC_Os11g45090 encode CC-NBS-LRR proteins, LOC_Os11g44910 encodes a DEAD-box RNA helicase, LOC_Os11g44930 encodes pentatricopeptide repeat domain protein, LOC_Os11g44950 encodes a glycosyl hydrolase, and LOC_Os11g44890 and LOC_Os11g45030 encode expressed proteins ( Table 2). To investigate expression patterns of candidate genes after M.oryzae infection, we performed quantitative real-time PCR (qRT-PCR) analysis of resistant japonica landrace Bodao after seedling blast inoculation. The results showed that LOC_Os11g44910, LOC_Os11g449300, LOC_Os11g44950, LOC_Os11g44990, and LOC_Os11g45090 were significantly induced after inoculation. Among them, LOC_Os11g45090, which was most significantly induced by rice blast fungi and encodes NBS-LRR domain protein, was reported to be involved in rice blast resistance [42]. We concluded that LOC_Os11g45090 was the candidate gene in PBRL3 and then was named Pb3 ( Figure 5B).
To identify candidate resistance genes underlying the stable PBRL3 locus, we investigated the haplotype block structure based on locus within 500 kb intervals; however, it was difficult to find the LD block in the interval ( Figure 4A-C). According to genomewide LD, decay rates of indica and japonica were estimated at ∼200 kb [41], and a 200 kb interval (Chr11:27.1 Mb-27.3 Mb) was selected to predict candidate genes. With reference to the Nipponbare genome sequence and gene annotation, 15 candidate genes with SNPs on the promoter or exons region were obtained. Haplotype analysis results showed that 7 candidate genes with different haplotypes had significant phenotypic differences in the natural population and were selected as potential candidate genes for further analysis ( Figure 5A). Among them, LOC_Os11g44990 and LOC_Os11g45090 encode CC-NBS-LRR proteins, LOC_Os11g44910 encodes a DEAD-box RNA helicase, LOC_Os11g44930 encodes pentatricopeptide repeat domain protein, LOC_Os11g44950 encodes a glycosyl hydrolase, and LOC_Os11g44890 and LOC_Os11g45030 encode expressed proteins (Table 2). To investigate expression patterns of candidate genes after M.oryzae infection, we performed quantitative real-time PCR (qRT-PCR) analysis of resistant japonica landrace Bodao after seedling blast inoculation. The results showed that LOC_Os11g44910, LOC_Os11g449300, LOC_Os11g44950, LOC_Os11g44990, and LOC_Os11g45090 were significantly induced after inoculation. Among them, LOC_Os11g45090, which was most significantly induced by rice blast fungi and encodes NBS-LRR domain protein, was reported to be involved in rice blast resistance [42]. We concluded that LOC_Os11g45090 was the candidate gene in PBRL3 and then was named Pb3 ( Figure 5B).   Pb3 encodes a protein with a CC-NBS-LRR domain. In the high-quality 700K SNPs dataset, six SNPs causing amino acid changes were found in exons. In addition, SNP11-

Pb3 Is an R Gene That Regulates Panicle Blast Resistance
Pb3 encodes a protein with a CC-NBS-LRR domain. In the high-quality 700K SNPs dataset, six SNPs causing amino acid changes were found in exons. In addition, SNP11-27282578 was located on the coiled-coil domain, SNP11-27282896 was located on the predicted NB-ARC domain, and SNP11-27284006 and SNP11-27285004 were located on the LRR domains. Moreover, the variation in the SNP11-27285004 in natural populations (CAG to TAG) leads to premature termination of Pb3 protein translation, suggesting that the terminal LRR domain may be involved in resistance to rice blast. Haplotype analysis showed that there were 5 haplotypes of Pb3 in the natural population ( Figure 6A,B). The Hap.A haplotype was significantly more resistant to panicle blast than Hap.C and Hap.D ( Figure 6C,D). The resistance haplotype A mainly contains TEJ cultivars, while the susceptible haplotypes C and D include 20% indica and 60% japonica ( Figure 6E). To study the genetic diversity of Pb3 in natural populations, we calculated the nucleotide diversity (π value) of the chromosome 11 region of 230 varieties by VCF tools, and the results showed that the π value of the Pb3 locus was 0.0002, which was significantly lower than the average nucleotide diversity of the chromosome 11 (π value = 0.0003), illustrating the sequence conservation of Pb3 in natural populations ( Figure 6F). We carried out evolutionary analysis on the cloned rice blast R gene, and the results showed that Pb3 was a typical NLR gene, which was highly homologous to Pit and was predicted to play similar functions ( Figure 7A,B).
27282578 was located on the coiled-coil domain, SNP11-27282896 was located on the predicted NB-ARC domain, and SNP11-27284006 and SNP11-27285004 were located on the LRR domains. Moreover, the variation in the SNP11-27285004 in natural populations (CAG to TAG) leads to premature termination of Pb3 protein translation, suggesting that the terminal LRR domain may be involved in resistance to rice blast. Haplotype analysis showed that there were 5 haplotypes of Pb3 in the natural population ( Figure 6A,B). The Hap.A haplotype was significantly more resistant to panicle blast than Hap.C and Hap.D ( Figure 6C,D). The resistance haplotype A mainly contains TEJ cultivars, while the susceptible haplotypes C and D include 20% indica and 60% japonica ( Figure 6E). To study the genetic diversity of Pb3 in natural populations, we calculated the nucleotide diversity (π value) of the chromosome 11 region of 230 varieties by VCF tools, and the results showed that the π value of the Pb3 locus was 0.0002, which was significantly lower than the average nucleotide diversity of the chromosome 11 (π value = 0.0003), illustrating the sequence conservation of Pb3 in natural populations ( Figure 6F). We carried out evolutionary analysis on the cloned rice blast R gene, and the results showed that Pb3 was a typical NLR gene, which was highly homologous to Pit and was predicted to play similar functions ( Figure 7A,B). To investigate whether Pb3 was involved in rice blast resistance, we obtained T-DNA insertion lines (pb3) from Salk Institute Genomic Analysis Laboratory ( Figure 8A). PCR analysis confirmed T-DNA insertions in the exon region (Supplementary Figure S2). Then, we inoculated mutant and wild-type Dongjin at the booting and seedling stages, respectively. Blast vaccination results showed that the average lesion length and number of Pb3 were significantly higher than those of the wild type ( Figure 8B,C,E). Moreover, 5 mutants were more susceptible to the M. oryzae strain than WT by blast inoculation at the booting stage ( Figure 8D,F). To further verify the rice blast resistance of Pb3, we knocked out this gene in the Bodao background with the CRISPR/Cas9 gene editing system. A total of 3 homozygous CRISPR lines (type1, type2, and type3) were obtained ( Figure 9A,B). The mutation target was located on the LRR domain at the end of the gene. Inoculation of the booting stage results showed that 5 mutants of 2 lines were more susceptible than WT to To investigate whether Pb3 was involved in rice blast resistance, we obtained T-DNA insertion lines (pb3) from Salk Institute Genomic Analysis Laboratory ( Figure 8A). Figure S2). Then, we inoculated mutant and wild-type Dongjin at the booting and seedling stages, respectively. Blast vaccination results showed that the average lesion length and number of Pb3 were significantly higher than those of the wild type ( Figure 8B,C,E). Moreover, 5 mutants were more susceptible to the M. oryzae strain than WT by blast inoculation at the booting stage ( Figure 8D,F). To further verify the rice blast resistance of Pb3, we knocked out this gene in the Bodao background with the CRISPR/Cas9 gene editing system. A total of 3 homozygous CRISPR lines (type1, type2, and type3) were obtained ( Figure 9A,B). The mutation target was located on the LRR domain at the end of the gene. Inoculation of the booting stage results showed that 5 mutants of 2 lines were more susceptible than WT to the M. oryzae strain ( Figure 9C,D). It showed that the LRR domain confers resistance to rice blast in Pb3, which was consistent with the results of haplotype analysis. These results further validated that Pb3 was a new R gene in the Pik locus that regulates both leaf blast and panicle blast resistance in rice. the M. oryzae strain ( Figure 9C,D). It showed that the LRR domain confers resistance to rice blast in Pb3, which was consistent with the results of haplotype analysis. These results further validated that Pb3 was a new R gene in the Pik locus that regulates both leaf blast and panicle blast resistance in rice.

Discussion
Rice blast is one of the critical rice diseases, and leaf blast and panicle blast are the most common types of this disease. Compared with leaf blast, panicle blast is considered to be more destructive, which can cause direct yield losses up to 70% or even 100% in fields by affecting grain sterility, rotting the branch and neck, and even losing the entire panicle [43,44]. Isolating the panicle blast resistance gene and introducing it into the susceptible elite cultivars is the most economical and effective way to control rice blast disease. Blast resistance is generally classified into complete and partial resistance [45]. Complete resistance to blast is qualitative and race-specific and is usually controlled by a major gene [4]. However, this type of resistance is easily overcome by the variety of races in the field. Partial resistance controlled by minor polygenes or the quantitative trait locus (QTL) confers broad-spectrum and durable resistance to M. oryzae.
The Rice Diversity Panel (RDP-1) contains substantial genetic and phenotypic diversity and has been used for GWAS analysis of various traits in rice [46]. We selected 230 cultivars in RDP-1 for panicle blast resistance identification in the field. Three experimental field panicle blast inoculations were performed on these cultivars. The average diseased grains percentages were 23.33%, 12.53%, 50.37%, and 38.70% in IND, AUS, TEJ, and TRJ, respectively. These results are consistent with previous studies where the TRJ and IND subgroups of rice were more resistant than the TEJ subgroups [33]. We chose the EMMAX model for association analysis and identified a repetitively mapped locus PBRL3 within the Pik locus at the end of chromosome 11. Haplotype analysis and expression pattern analysis identified 5 candidate genes that were significantly induced by M. oryzae. Among them, LOC_Os11g44910 encodes a DEAD-box RNA helicase, which contains a conserved amino acid sequence Asp-Glu-Ala-Asp (DEAD). DEAD box RNA helicase family proteins can hydrolyze ATP to obtain double-stranded energy unwinding RNA, and are reported to be involved in plant leaf growth and the cold stress response [47,48]. LOC_Os11g44930 encodes a pentatricopeptide-repeat (PPR)-domain-containing protein. Plant PPR proteins are usually located in mitochondria or chloroplasts, which edit and modify the RNA transcripts to regulate plant growth and development [49,50]. LOC_Os11g44950 is a member of the glycosyl hydrolase family 3. Glycosyl hydrolase can effectively hydrolyze sugarbased substances and are involved in cell wall remodeling [51]. LOC_Os11g44990 and LOC_Os11g45090 are NBS-LRR-type genes that are widely reported to be involved in rice blast resistance [52]. Furthermore, after inoculation with M. oryzae, the expression of LOC_Os11g45090 significantly increased at 24 h by 4.07-fold. These results indicate that LOC_Os11g45090 (Pb3) is the most likely candidate gene in PBRL3.
In the 230 natural populations, five SNPs causing amino acid changes were detected in the coding region of Pb3. Among them, SNP11-27282578 fell on the CC domain, SNP11-27282896 was located in the NB-ARC domain, and SNP27284006 and SNP11-27285004 fell on the LRR domain. Combining the haplotype analysis results, we found that the difference between resistant haplotype Hap. A and susceptible haplotypes Hap. C and Hap. D contained these four SNPs that were located in the functional domain of Pb3. In NBS-LRR proteins, the CC domain mainly mediates the interaction with transcription factors and structural proteins [53]. NBS has the function of ATP or GTP binding and hydrolysis, which can resist the invasion of pathogens by gaining energy [11]. The easily altered structure of LRR may be related to its ability to specifically recognize pathogen-effector proteins [54]. To verify the function of these domains, we obtained the transgenic plants of Pb3. We first obtained a T-DNA insertion mutant of Pb3, and the insertion position was located in the double coiled-coil region. The susceptible phenotype of the mutant line verified that Pb3 was indeed involved in rice blast resistance. The genotype data revealed an SNP11-27285004 on the LRR domain in the end of Pb3 in natural populations, which can prematurely terminate the translation of the protein, resulting in the deletion of the LRR domain. Therefore, we specifically constructed a CRISPR-Cas9 mutation target located on the terminal LRR domain, and CRISPR showed that it was more susceptible than wild-type thin rice, indicating the importance of the LRR domain for Pb3 to play the role of panicle blast resistance.
Most of the cloned rice blast resistance genes encode NBS-LRR domain proteins. Therefore, we performed phylogenetic tree analysis on Pb3 and cloned R genes and found that Pit has the highest homology with PB3. Pit interacts with OsRac1 through its NB-ARC domain and activates OsRac1, which plays an important role in Pit-mediated immune responses [55]. Therefore, Pb3 may also mediate resistance by binding to a key immune regulatory factor. In the following studies, we can verify the interaction between Pb3 and OsRac1, and can also search for other interacting proteins in combination with yeast two-hybrid assays to research the specific resistance mechanism of Pb3. On the other hand, we verified the function of the LRR domain at the end of Pb3 through the CRISPR plants.
The deletion of the LRR domain may affect the recognition of effector proteins of rice blast fungi, further inhibit the activation of the rice defense response, and reduce rice resistance to rice blast. In summary, we identified a repeat site PBRL3 on chromosome 11 by GWAS. Through functional verification of candidate genes, we found that Pb3, which encodes NBS-LRR protein, is a causal gene in PBRL3, which can regulate the resistance of rice to leaf blast and panicle blast. In this work, we carried out GWAS analysis to quickly and accurately obtain the rice blast resistance R gene, which provides a theoretical basis for mining a large number of panicle blast resistance genes. With the continuous development of sequencing technology, it will be more accurate to locate panicle blast resistance genes through GWAS in the future. However, due to the influence of environmental factors, identification of resistance to panicle blast in the field is still a difficult problem, which is crucial to the results of GWAS analysis.

Plant Materials and Fungal Growth
RDP-1, which contains 413 O. sativa accessions from 82 countries, was provided by Cornell University [46]. According to the growth period of these accessions, we selected 230 accessions from the panel to ensure the growth period consistency of rice. The whole panel was divided into six subpopulations, including Indica, Aus, Tropical japonica, Temperate japonica, Aromatic, and Admixtures [46]. These germplasms were planted in an experimental field in Nanjing for panicle blast inoculation. The seedlings of these germplasms were cultivated in 96-hole plugs in the greenhouse for seedling leaf blast inoculation. The rice blast isolate 2014-290B27 was provided by the Jiangsu Academy of Agricultural Sciences. The blast fungi were grown on corn rice straw agar plates at 28 • C for seven days and then transferred in black light (20 W) in an incubator at 28 • C for sporulation and culture for 7-10 days to promote spore production. They were then washed with sterilized distilled water to a spore suspension with a concentration of 1 × 10 5 /mL [16,56].

Inoculation and Phenotypic Survey
In 2017 and 2018, natural population materials were planted in Nanjing, and inoculation was carried out at the booting stage. Each rice accession was inoculated with 9 rice panicles. The ear was sampled 15 days after inoculation, and the resistance survey was measured by calculating the percentage of the number of diseased grains in the whole panicle of rice (0-100%) [57]. For the leaf blast resistance identification of T-DNA insertion mutants and CRISPR plants of Pb3, the seedlings were planted in the greenhouse of Nanjing Agricultural University to the two-leaf and one-heart stage for spray inoculation [58]. The length and number of lesions were recorded after six days. Six plants for each line and three replicates were measured, and the average was calculated for the resistance phenotype. Suyunuo, a highly susceptible japonica rice variety, was planted around the tray as a control.

Population Structure Analysis and Genome-Wide Association Study
The phylogenetic tree of the 230 accessions was constructed using the MEGA7 program [59]. Multiple sequence alignment of 230 accessions was performed with the ClustalW program with the standard setting. The neighbor-joining (NJ) with maximum composite likelihood method was used to construct phylogenetic trees with a bootstrap value of 1000 replicates in MEGA 7. The beautification of the phylogenetic tree was conducted with iTOL online tools [60]. Based on the mixed linear model, GWAS was used to detect resistance-related loci [61]. The data of 400K SNPs (MAF ≥ 5%, missing rate < 25%) corresponding to the study accessions were filtered from the 700K high-density genetic markers of RDP-1 through the TASSEL 5.0 software. The MLM model was combined with the EMMAX online script (http://genetics.cs.ucla.edu/emmax/install.html, accessed on 20 December 2019) and R packages for analysis. Manhattan plots were drawn by using the R package qqman. p-value < 10 −4 was used as the significance threshold to obtain the loci associated with leaf blast.

Candidate Genes and Haplotype Analysis
The LD heatmap of the associated loci in the GWAS was constructed using the Haploview4.2 [62]. Haplotype blocks were constructed using the confidence interval (r2 > 0.6) [21]. Candidate genes were predicted according to the MSU Rice Genome Annotation Project (http://rice.plantbiology.msu.edu, accessed on 10 June 2020). We analyzed the polymorphisms in the candidate region that were located on the promoter and caused amino acid changes. Haplotype analysis of these candidate genes was performed to screen out significant difference genes by using the Haploview4.2.

Genomic Nucleotide Diversity and Phylogenetic Analysis
To clarify the Pb3 genetic differentiation of the natural population, the PLINK online script was used to extract SNPs on chromosome 11 in natural populations [63]. Then, nucleotide diversity (π) was calculated using VCFtools software [64]. Sliding 100 kb windows were used during the calculation with a 10 kb sliding step. The multiple sequences of R genes were aligned with MEGA7, then the phylogenetic trees were calculated and constructed using the neighbor-joining method within 1000 replicates. The conserved domain of the R gene was predicted by the Pfam online website (http://pfam.xfam.org/, accessed on 25 November 2021), and the gene structure was drawn by the IBS1.0.3 software [65].

RNA Extraction and qRT-PCR Analysis
Variety leaves infected with M. oryzae strains were collected at different time points after inoculation. The samples were immediately frozen with liquid nitrogen and stored in the refrigerator at −80 • C. The total RNA was extracted from leaves using TRIpure Reagent (Wuxi, China, Bio Teke Corporation, http://www.bioteke.com, accessed on 5 June 2021). First-strand cDNA synthesis was performed using HiScript II Q-RT Super Mix (Nanjing, China, Vazyme, http://www.vazyme.com, accessed on 6 June 2021). qRT-PCR was performed using the AceQ ® qPCR SYBR Green Master Mix. The primers in Supplemental Table  S1 were amplified for the genes' expression analysis. The rice actin gene (LOC_Os03g50885) was used as the internal control. Each experiment was performed with three biological samples and each sample was assayed with three technical replications.

Generation of Transgenic Plants
The CRISPR-Cas9 gene mutation system was obtained from the Chen Qijun Laboratory, College of Biology, China Agricultural University, Beijing China [66]. The target primers were designed using the online website (http://skl.scau.edu.cn/primerdesign/, accessed on 12 September 2021). The CRISPR vectors pBUE411-2gR containing the Pb3-specific silencing target were transformed into japonica rice Bodao by an agrobacterium-mediated transformation method [67]. We obtained six CRISPR mutants of Pb3 containing three editing types: type1:KO-1/ KO-2; type2: KO-3/KO-4/KO-5; type3: KO-6. The T-DNA insertion mutant of the Dongjin background of Pb3 was purchased from Salk Institute Genomic Analysis Laboratory (http://signal.salk.edu/index.html, accessed on 11 July 2021). Sequencing results showed that an 8275 bp T-DNA was inserted into the exon of Pb3 in the mutant. We obtained 5 homozygous mutant plants (pb3-1 to pb3-5) from 10 T 0 generation seeds of the mutant. The primers in Supplemental Table S1 were amplified for positive verification.