Bacterial panicle blight (BPB) induced by the bacterial pathogen Burkholderia glumae was first identified in Japan in 1955 (Mizobuchi et al., 2020). Since then, the disease has been reported in multiple rice growing countries such as the United States (Nandakumar et al., 2009), East and Southeast Asia (Ashfaq et al., 2017; Jeong et al., 2003; Mondal et al., 2015), Latin America (Nandakumar et al., 2007), and South Africa (Zhou, 2014). Climate change causing the rise of global temperatures favored the proliferation of bacterial diseases. In this context, the BPB emerged as a new plant disease threatening rice productivity. The Burkholderia glumae is responsible for the BPB and bacterial seedling rot in rice (Oryza sativa L.), which are economically important diseases, affecting the global rice production (Mizobuchi et al., 2016). The reported optimal temperature range for the growth of Burkholderia glumae (30-35°C) is relatively high, which explains its predominance in tropical and subtropical regions. Global warming as a result of climate change could faster the development of the panicle grain rot disease caused by Burkholderia glumae to reach severe epidemics (Ham et al., 2011; Mizobuchi et al., 2023). Therefore, considering its high pathogenicity, Burkholderia glumae should be regarded as a threat to the global rice production (Shew et al., 2019; Sreenayana et al., 2024). The pathogen Burkholderia glumae uses various strategies for infection into rice plants and proliferation of the disease. One of them is the production of a phytotoxin known as toxoflavin, which mediate the induction of oxidative damage by accumulating more hydrogen peroxide (H2O2), an essential molecule for the pathogenicity of Burkholderia glumae. Typical symptoms include chlorotic symptoms on rice panicles blight (Lee et al., 2016; Naughton et al., 2016). Common symptoms are discoloration of rice grains from the base, turning to pale yellow-whitish color, and gradually expanding to the entire panicle and grains (Jeong et al., 2003; Ura et al., 2006). In the field, early infected panicles exhibit a pale reddish color, stand erect without drooping, and grains show impaired or halted development. However, late infected grains are smaller and develop pinkish stripes (Cha, 1995). This disease results in reduced grain quality and poor swelling, significantly affecting rice yield. The most effective control methods for Burkholderia glumae include seed treatments such as saline selection and iron coating to prevent occurrence from infected seeds, timing adjustment for heading date, and chemical control methods such as applying fungicides and other crop protectants. An assessment of current progress in investigating Burkholderia glumae resistance genes shows that the limited research aiming to develop effective control methods, coupled with the emergence of drug-resistant Burkholderia glumae strains, remain major concerns necessitating a close attention for the development of novel disease control strategies (Kim et al., 2023; Lee et al., 2010; Mizobuchi et al., 2023).
Therefore, the cultivation of Burkholderia glumae resistant varieties is regarded as a cost-effective and most effective method for controlling this disease. Previous studies have conducted extensive field studies to determine the regulation of BPB resistance using spray inoculation at the heading stage, or syringe inoculation at booting (Pinson et al., 2010; Shahjahan et al., 2000). Due to the significant influence of environmental conditions such as humidity and temperature, it is challenging to evaluate resistance in varieties with different heading dates using field inoculations (Tsushima, 1996). Field experiments evaluating rice BPB resistance have shown a broad spectrum of resistance responses, from high susceptibility to moderate resistance, but not highly resistance or immune (Mizobuchi et al., 2013, 2016; Pinson et al., 2010; Shew et al., 2019). This range of phenotypes indicate that resistance to BPB is quantitative and associated with multiple genes.
The use of molecular breeding techniques and the advent of genome sequencing technologies, couple with the emergence of bioinformatics approaches used for gene discovery, such as the genome-wide association studies (GWAS) have boosted the efficiency of plant breeding and accelerated variety development. In addition, the Genome Association and Prediction Integrated Tool (GAPIT) package in R software, with multiple statistical models boosted the power and accuracy of genomic prediction, and has proven useful and efficient in detecting reliable genetic loci associated with major agronomic traits in plant crops (Kabange et al., 2023). Furthermore, the use of high-density single nucleotide polymorphism (SNP) marker system gained momentum in plant bioscience, especially in plant molecular breeding program (Mammadov et al., 2012). Likewise, SNP genotyping and DNA chip technology emerged as essential genomic tools, as they offer useful information for generation advancement and crops improvement for desired traits (Thomson, 2014).
Several quantitative trait loci (QTLs) associated with the control of BPB have been identified using recombinant lines derived from crosses between BPB susceptible variety “Lemont” and moderately resistant variety “TeQing” identified QTLs linked to BPB resistance on chromosomes 1 (qBPB-1-1, qBPB-1-2, qBPB-1-3), 2 (qBPB-2-1, qBPB-2-2), 3 (qBPB-3-1, qBPB-3-2), 7 (qBPB-7-1), 8 (qBPB-8-1, qBPB-8-2), 10 (qBPB-10-1), and 11 (qBPB-11-1) (Pinson et al., 2010). Furthermore, QTLs linked to BPB resistance were identified from recombinant inbred lines progeny derived from crosses between varieties “Huazhan” and “Nekken 2” on chromosomes 1 (qBPB-1), 3 (qBPB-3), 4 (qBPB-4), 5 (qBPB-5), 7 (qBPB-7.1, qBPB-7.2), 8 (qBPB-8), 9 (qBPB-9), and 10 (qBPB-10) (Fang et al., 2023).
This study aimed to investigate novel genetic loci controlling in rice. We applied a GWAS strategy using high-density SNP Chip DNA markers on 307 rice cultivars. This work identified new BPB resistance candidate genes, here reported for the first time on chromosome 6 of rice. These QTLs assumed useful for improving BPB resistance in rice and developing resistant cultivars, based on the predicted annotation of candidate genes found within those QTL regions.
Materials and Methods
Plant materials and growth condition
A total of 307 Korean rice cultivars, of which 263 belonged to Oryza sativa L. ssp. Japonica as well as 44 Tongil-type varieties, were used to investigate novel genetic loci controlling BPB resistance. Seeds were seeded after induction of germination for 3 days, and grown in germination trays for 4 weeks after sowing. Then, on May 30, 2023, 4-week-old seedlings were transplanted to our experimental paddy field at the Department of Southern Area Crop Science of the National Institute of Crop Science, Miryang, Korea (latitude 35.5°N, longitude 128.8°E). Plants were grown under field conditions with regular fertilization (90 kg/ha N, 45 kg/ha P2O5, 57 kg/ha K2O) and regular water management regimes until inoculation time at heading stage.
BGR1 inoculum preparation and panicle inoculation
The bacterial pathogen Burkholderia glumae BGR1 virulent strain was obtained from the Department of Microbiology, Pusan National University. The bacterial cells were cultured and maintained following the method described by Kim et al. (2023). Briefly, bacterial cells were grown on Luria-Bertani (LB) agar in petri dishes supplemented with rifampicin (100 μg/mL), followed by incubation at 37 °C for 16 h. Single colonies were then preincubated in LB broth supplemented with rifampicin (100 μg/mL), followed by subculturing 1% under the same conditions until the bacterial culture reached the OD600 of 0.8 (equivalent to OD600 = 2.4 × 108 cells/mL) used for panicle inoculation. To measure the concentration of the culture, a spectrophotometer was used. Rice panicles were inoculated using the dipping method, consisting of submerging panicles for 2 min before covering panicles with a commercial polyvinyl bag for two days (Supplementary Fig. 1).
BPB inoculation
Prior to inoculating rice panicles with Burkholderia glumae, we assessed the heading date of all rice cultivars used in the study. The criteria for selecting panicles for inoculation with the BGR1 strain of Burkholderia glumae were that the panicle neck should be visible, and more than two-thirds of the stamens should have emerged from the panicle. Inoculation was performed between 11 am and 2 pm on sunny days, and no inoculation was done on rainy days. The evaluation of BPB resistance was conducted over 42 days period based on the heading dates of the rice varieties used in the experiment (July 20, 2023 to August 30, 2023). This implies that all tested rice cultivars were not inoculated at the same time, rather, a progressive inoculation method was employed following the heading date of each rice cultivar or group of rice cultivars reaching the heading stage the same time. During this period, the average daily temperature ranged from 24.9°C to 30.8°C, while the maximum temperature varied between 26.2°C to 37.8°C (Supplementary Fig. 2A). Changes in the relative humidity and precipitation are given in Supplementary Fig. 2B and 2C, respectively. A healthy seed was defined as a grain showing no visible symptoms of bacterial panicle blight, such as discoloration, sterility, or shriveling. For each plant, three panicles were harvested at maturity, and the proportion of healthy seeds per panicle was calculated as (Number of healthy seeds/Total seeds per panicle) × 100. The average value across the three panicles was used as the healthy seed percentage for each plant.
Genotyping analysis using 580K chip set
The genotyping data (Axiom_Oryza_580K_chipset) was provided by the Crop Breeding Division of the National Institute of Crop Science, based on the, consists of 542,333 SNP Chip DNA markers from 307 rice varieties and germplasm. Before performing a GWAS, the genotype data were filtered with TASSEL5 to remove markers with a minor allele frequency (MAF) less than 0.01. Then, 139,907 SNP markers were finally used for population structure analysis and GWAS.
Genetic structure and linkage disequilibrium analysis
The population structure of the 307 rice accessions was examined using ADMIXTURE version 1.3.0, where subgroup classification was determined based on the delta K value (Alexander et al., 2009). To validate the results, cross-validation (CV) analysis was conducted. Visualization of population structure was carried out using the Structure Plot V2.0 web application (http://omicsspeaks.com/strplot2/, accessed on 1 January 2025). Additionally, principal component analysis (PCA) was performed using the GAPIT (Lipka et al., 2012). The PCA results were further visualized using R software (R Foundation for Statistical Computing, Vienna, Austria). Phylogenetic trees were constructed using the Neighbor-Joining (NJ) method in MEGA X with 1,000 bootstrap replicates, and the resulting Newick format files were subsequently visualized and customized using the Interactive Tree of Life (iTOL) platform (https://itol.embl.de/, accessed on 15 January 2025) (Kumar et al., 2018; Letunic and Bork, 2011). To explore linkage disequilibrium (LD) decay patterns and identify potential candidate regions, LD decay analysis was conducted using PopLDdecay version 3.27 (Zhang et al., 2019).
Genome-wide association study analysis
GWAS was conducted to identify genotype-phenotype associations using the GAPIT package version 3.0 in R. The association analysis was performed based on the fixed and random model Circulating Probability Unification (FarmCPU) approach, implemented in the GAPIT package (Lipka et al., 2012; Liu et al., 2016). To control for false positives, the genome-wide significance threshold was adjusted using the Bonferroni correction (Armstrong, 2014). PCA and kinship matrices were incorporated into the model to account for population stratification. Additionally, significant SNPs were identified by filtering the genotype data using PLINK software (Purcell et al., 2007). Non-informative SNPs were excluded, leaving approximately 200,000 effective and independent SNPs for further analysis.
Haplotype analysis of candidate genes
For haplotype analysis of the 157 Korean rice cultivars, whole-genome sequencing was performed using the Illumina HiSeq 2500 platform (Illumina Inc., San Diego, CA, USA), producing an average sequencing depth of approximately 8×. Raw sequence reads were mapped to the rice reference genome (IRGSP 1.0) to generate high-confidence genotype calls. The haplotype analysis was conducted using 157 rice accessions from the GWAS panel of 307 accessions for which whole-genome resequencing data were available. Due to the limited availability of resequencing data, only these accessions were included in the haplotype-based evaluation. To ensure the quality of the genotype dataset, filtering criteria were applied using PLINK software, excluding variants with missing data >1%, MAF <5%, or heterozygosity >5%. These thresholds were selected to reduce noise and improve the reliability of haplotype detection in downstream analyses. Based on the phenotype data, the average trait values and the number of varieties corresponding to each haplotype were calculated. Haplotypes significantly associated with the target phenotype were identified through comparative analysis. The gene structure and positions of SNPs were visualized using the Gene Structure Display Server 2.0 (GSDS 2.0) (Hu et al., 2015), enabling intuitive understanding of SNP distribution across the gene region.
Results
Phenotypic evaluation for BPB resistance
To evaluate resistance to BPB, a total of 307 Korean rice cultivars, including 263 Japonica-type and 44 Tongil-type varieties, were evaluated based on the percentage of healthy seeds per panicle following artificial inoculation with Burkholderia glumae. The healthy seeds percentage showed a continuous distribution ranging from 0.41% to 84.75%, indicating substantial phenotypic variation among the tested cultivars. A large number of cultivars exhibited less than 30% healthy seeds per panicle, suggesting that a majority of the population is highly susceptible to BPB. In contrast, only a small subset of cultivars showed over 60% healthy seeds per panicle and were categorized as highly resistant. Representative cultivars from both extremes of the distribution were ‘Jinsumi’ (highly susceptible) and ‘Odae1’ (highly resistant), which exhibited striking phenotypic differences under both negative control and BGR1-inoculated conditions (Fig. 1). Among the cultivars with high resistance (≥60% healthy seeds per panicle), ‘Odae1’ was the only one that exhibited over 80% survival, followed by ‘Dunnae’, which also showed a relatively high level of resistance. In addition, ‘Nampung’ and ‘Namcheon’ exhibited healthy seeds percentages ranging from 60% to 70%, representing the highest levels of resistance among the Tongil-type cultivars (Table 1). Comparison of healthy seeds per panicle percentages between Japonica and Tongil ecotypes revealed no statistically significant difference, indicating that resistance to BPB was not strongly associated with ecotype (Fig. 1C).
Population structure and LD decay analysis
Population structure of the 307 Korean rice cultivars was assessed using CV, PCA, ADMIXTURE analysis, and an NJ phylogenetic tree. The CV error reached its lowest point at K = 2, suggesting that dividing the population into two major genetic clusters was optimal for this panel (Fig. 2A). PCA based on SNP genotype data revealed a clear separation of the cultivars into two groups. The first two principal components, PC1 and PC2, explained 63.57% and 2.53% of the total genetic variation, respectively, and distinguished cultivars primarily according to their ecotype—Japonica and Tongil (Fig. 2B, Supplementary Fig. 3). The NJ tree, constructed using pairwise genetic distance, exhibited a similar clustering pattern, with most cultivars grouping into two distinct clades corresponding to the two ecotypes (Fig. 2C). ADMIXTURE analysis at K = 2 also supported this grouping, with most cultivars showing a high proportion of ancestry from either of the two clusters (Fig. 2D). Only a small number of individuals showed admixture between clusters, indicating limited gene flow between the Japonica and Tongil types in this population.
Genome-wide association study for BPB resistance
The GWAS analysis for BPB resistance was performed using the GAPIT package in R. The Manhattan plot illustrates SNPs that are significantly associated with BPB resistance. A genome-wide significance threshold of −log10(P) ≥ 6.44 was determined using Bonferroni multiple-test correction at a significance level of P = 0.05, to minimize false-positive signals. Three lead SNPs significantly associated with BPB resistance were identified on chromosome 6 (8,361,583 bp [logarithm of the odds (LOD): 8.42], 8,362,639 bp [LOD: 8.64], and 8,373,932 bp [LOD: 8.31]), suggesting that this locus may contain candidate genes contributing to the genetic mechanism underlying resistance to BPB in rice (Fig. 3). Since our LD decay analysis was based on approximately 100K SNP chip dataset, the resulting r2 decay pattern may be less precise compared to analyses using whole-genome resequencing data. Therefore, we referred to previously reported LD decay estimates from large-scale studies utilizing diverse indica and japonica accessions, both of which reported an LD decay distance of approximately 200 kb. Based on this evidence, we used a ±200 kb window around each lead SNP to define the candidate genomic region (Huang et al., 2010; Yang et al., 2025).
Haplotype analysis of candidate genes associated with BPB resistance
Haplotype analysis was conducted to evaluate genetic variations associated with BPB resistance in the 28 candidate genes, among which only two genes showed statistically significant associations. In candidate gene Os06g0255900, two SNPs located within exon 1 were detected. Based on these SNP variations, cultivars were divided into three haplotypes (Hap1-Hap3). Hap1 exhibited the highest mean percentage of healthy seeds per panicle (30.71%), indicative of increased resistance, whereas Hap3 displayed the lowest mean (20.15%), suggesting susceptibility. The distribution of cultivars showed that Hap3 was the most frequent haplotype, followed by Hap2 and Hap1. Each haplotype predominantly consisted of Japonica-type cultivars, with Tongil-type cultivars represented primarily in Hap1 and Hap2. Statistical analysis using ANOVA confirmed significant differences among haplotypes (F = 8.08, P < 0.001) (Fig. 4).
Haplotype analysis was also performed on Os06g0259850, another candidate gene associated with BPB resistance. Three polymorphic loci, including one SNP located in exon 1 and two loci (one InDel and one SNP) in the 5′ untranslated region, were identified. Based on these polymorphisms, three distinct haplotypes (Hap1-Hap3) were defined. Hap3 displayed the highest mean percentage of healthy seeds per panicle (32.00%), suggesting enhanced resistance, whereas Hap1 had the lowest mean (20.01%), indicative of susceptibility. The distribution analysis indicated that Hap1 was most prevalent, followed by Hap2 and Hap3. Most cultivars were Japonica-type, but Hap3 exhibited the highest proportion of Tongil-type cultivars. ANOVA analysis revealed significant differences among haplotypes (F = 10.90, P < 0.001) (Fig. 5). Among 157 accessions with available whole-genome resequencing data, five cultivars exhibiting high resistance to BPB (≥50% healthy seeds per panicle) were analyzed for their haplotype composition (Table 2). All five cultivars, including ‘Nampung’, ‘Saeodae’, ‘Namcheon’, ‘Cheongbaekchal’, and ‘Mogyang’, consistently carried Hap1 at Os06g0255900 and Hap3 at Os06g0259850. This uniform haplotype combination among highly resistant cultivars suggests a potential association between Hap1-Hap3 configuration and enhanced resistance to BPB.
Discussion
In this study, a GWAS was conducted using 307 Korean rice cultivars to identify genetic loci associated with resistance to BPB. The analysis revealed a major QTL, qBG6.1, on chromosome 6, in which a lead SNP surpassed the genome-wide significance threshold. Within this QTL region, only two genes—Os06g0255900 and Os06g0259850—were found to be significantly associated with the trait based on haplotype analysis. These genes represent promising candidates for further investigation into the genetic basis of BPB resistance in rice.
Os06g0255900 (EXO70F5), encodes a member of the Exo70 subfamily within the exocyst complex. The exocyst is essential for polarized secretion and vesicle tethering at the plasma membrane, particularly during cell growth and pathogen defense (Holden et al., 2022; Munson and Novick, 2006; Wang et al., 2020). In plants, the Exo70 family has undergone significant expansion, with at least 47 genes in rice (Fujisaki et al., 2015; Xu et al., 2025; Žárský, 2022), some of which have been implicated in immune responses. EXO70F5 may participate in targeted secretion of antimicrobial molecules, suggesting a potential role in vesicle-mediated defense mechanisms.
The second gene, Os06g0259850, was annotated as “similar to TMV response-related protein,” implying possible involvement in plant-pathogen interactions. Although it has not been functionally characterized in rice, its sequence similarity to defense-related proteins indicates that it may contribute to biotic stress responses. Further investigation is required to determine its precise function.
Although the GWAS was conducted with a panel of 307 accessions, haplotype analysis was limited to 157 cultivars for which whole-genome resequencing data were available. Consequently, the haplotype results reflect only this subset and may not fully represent the entire genetic diversity of the panel. Expanding resequencing efforts to cover more accessions would increase the resolution and generalizability of haplotype-based associations.
While haplotype analysis offered valuable insight into candidate genes, their functional roles in BPB resistance remain to be verified. Functional validation, such as transgenic overexpression, gene knockout using CRISPR/Cas9, or expression profiling under pathogen infection, will be necessary to confirm their involvement in resistance mechanisms.
Collectively, our findings provide new insight into genetic resources and molecular markers associated with BPB resistance. These results may contribute to marker-assisted selection and the development of resistant rice cultivars, ultimately supporting more stable rice production under disease pressure.
PDF Links
PubReader
ePub Link
Full text via DOI
Full text via PMC
Download Citation
Supplement1
Print
