Integrating image-based phenotyping and QTL mapping to enhance genetic resistance and accelerate breeding for bacterial grain rot resistance in rice

Abstract

Bacterial grain rot (BGR), caused by Burkholderia glumae, is a major disease that reduces the yield of rice (Oryza sativa L.), thereby threatening food security. Conventional phenotypic analysis methods face limitations in objectively evaluating disease resistance and understanding the genetic basis. In this study, we integrated image-based phenotypic analysis with QTL mapping to screen for QTLs and candidate genes associated with B. glumae resistance. B. glumae was inoculated into 189 recombinant inbred lines (RILs) derived from Kele (resistant) and IS592BB (susceptible), followed by visualization and quantitative analysis using DAB staining. Phenotypic parameters, including the field resistance score, ratio of diseased spikelets (%), DAB staining intensity, and ratio of diseased area (%), were measured and used for QTL mapping. On chromosome 1, within Chr01_24592710-Chr01_37274755, four QTLs—qFRS1 [LOD: 5.98, phenotype variation explained (PVE): 15.41%], qRDS1 (LOD: 5.29, PVE: 18.56%), qQDS1 (LOD: 9.58, PVE: 22.02%), and qRDA1 (LOD: 8.44, PVE: 31.51%)—were identified as overlapping. After fine-mapping we narrow down Chr01_33472174-Chr01_33838140 and a total of 16 candidate genes were screened this region. Among which OsBGq1 was found to encode a nucleotide-binding LRR receptor (NLR) domain. OsBGq1 expression increased significantly upon B. glumae infection. Additionally, RILs Kele type of Chr01_33472174-Chr01_33838140 presented increased ROS-scavenging enzyme activity and phytoalexin accumulation upon B. glumae infection, contributing to increased resistance. The integration of DAB-based quantitative phenotyping with QTL mapping is proposed to provide a more objective indicator for identifying genes associated with resistance to BGR.

Supplementary Information

The online version contains supplementary material available at 10.1007/s00122-026-05216-7.

Introduction

Rice is a staple crop that accounts for a significant portion of the global caloric intake and is recognized as one of the most important food crops worldwide (Bin Rahman and Zhang 2023). However, the recent rapid rise in temperature due to climate change has resulted in increased pathogen activity, leading to a higher incidence of bacterial grain rot (BGR) (Echeverri-Rico et al. 2021). BGR, caused by Burkholderia glumae, is one of the most devastating diseases affecting rice. In the USA, BGR has led to a 40% yield loss in rice production in Louisiana (Ham et al. 2011). Notably, the optimal temperature for B. glumae activity is relatively high, ranging from 30 to 35 °C (Tsushima 1996). According to a report by the World Meteorological Organization (WMO), the global average temperature in 2024 was approximately 1.54 °C higher than that in the preindustrial period (1850–1900). This value surpasses the previous record of 1.45 °C set in 2023, highlighting the trend of record-breaking high temperatures being observed annually, which is indicative of the accelerations of global warming (Amnuaylojaroen 2023).

With continued global warming, the incidence of BGR is expected to increase further, underscoring the need for proactive measures to mitigate its impact (Ham et al. 2011). Primary method for controlling the BGR currently relies on chemical control through seed disinfection with oxolinic acid (Hikichi 1993). However, chemical control can lead to the emergence of new resistant strains and contribute to environmental pollution, including water and soil contamination (Sun et al. 2018). Previous research has focused on effective methods for controlling major rice diseases and pests, such as bacterial leaf blight, rice blast, and brown planthopper infestation, which cause severe yield losses and negatively impact grain quality. However, the most economical and effective approach involves the identification of resistance-related quantitative trait loci (QTLs) and genes, followed by their incorporation into new cultivars through breeding programs to develop resistant cultivars (Chukwu et al. 2019). To date, no effective bactericide has been developed for BGR. Owing to environmental changes, the incidence of BGR has increased rapidly in recent years. BGR, by infecting the panicle and spikelets, directly reduces the yield, so the development of BGR-resistant rice varieties is urgently needed (Mizobuchi et al. 2016).

Since QTL mapping involves analysis of the interaction between environmental and genetic factors, the results are highly dependent on phenotypic evaluation, making accurate phenotypic assessment a critical factor in the analysis (Cobb et al. 2013). Therefore, an objective and precise evaluation of phenotypic traits is essential for QTL mapping (Costa et al. 2019). Traditional phenotypic analysis has been performed using subjective methods based on the experience of breeders or researchers (Cooper et al. 2009). While this approach has the advantage of being based on knowledge and experience accumulated over time, it has been criticized for its high variability and subjectivity, as results can differ depending on the breeder’s level of expertise and experience (Yin et al. 2003).

Recently, a scanner or drones that provides high-resolution images quickly, has been proposed as a method to improve the accuracy of phenotypic assessment (Araus and Cairns 2014). Image analysis can be performed to eliminate external factors that may interfere with evaluation, ensure consistent assessment across samples, and detect subtle phenotypic differences that are difficult to observe with the naked eye (Gill et al. 2022). Furthermore, data obtained from high-resolution imaging devices can be analyzed using computer algorithms, enabling quantitative and objective assessments. Compared with traditional subjective evaluations, this approach offers greater reproducibility and reliability (Xiao et al. 2022). The 3,3ʹ-diaminobenzidine (DAB) staining method is a histochemical technique used to detect hydrogen peroxide (H₂O₂) accumulation. It is widely utilized to precisely visualize various plant responses to environmental stress (Šnyrychová et al. 2009). Notably, it enables the effective assessment of stress responses that are difficult to distinguish with the naked eye, making it a reliable tool for accurate phenotypic analysis (Bach-Pages and Preston 2017).

In this study, recombinant inbred lines (RILs) derived from a cross between the bacterial grain rot-resistant cultivar Kele (indica) and the susceptible cultivar IS592BB (japonica) were used to investigate the genetic basis of resistance to B. glumae. The genetic divergence between the indica and japonica ecotypes provides high polymorphism, facilitating efficient QTL detection for resistance-related traits. Phenotypic evaluation was conducted using image-based analysis, complemented by DAB staining to capture oxidative stress-related responses. QTL analysis revealed a major genomic region on chromosome 1 associated with BGR resistance. Subsequent genomic and expression analyses suggested a nucleotide-binding leucine-rich repeat (NLR)-encoding gene, designated OsBGq1, as a promising candidate. These findings provide insight into the genetic architecture of BGR resistance and suggest a potential role for immune-related genes in defense against B. glumae.

Materials and methods

Plant materials and field design

The F₈ KIRILs, consisting of 189 lines, were derived from an F₁ cross between Kele (Oryza sativa spp. indica cv. Kele) and IS592BB (Oryza sativa spp. japonica cv. IS592BB). The generation was advanced in the field of the National Institute of Crop Science (NICS) (36°6ʹ41.54″ N, 128°38′26.17″ E). Kele and IS592BB were obtained from Crop Breeding Division, NICS, Rural Development Administration (RDA). Kele is a representative rice cultivar with strong resistance to BGR, whereas IS592BB is a susceptible rice cultivar (Fig. 1). Therefore, Kele and IS592BB were selected to derive a population for QTL mapping related to BGR resistance. The KIRIL F₈ population was used for QTL mapping and the identification of candidate genes.

Fig. 1.

Histogram representation of KIRIL phenotyping for B. glumae resistance and the phenotyping methodology. A, B. glumae was inoculated into KIRILs, and histograms were constructed for the resistance score, ratio of diseased spikelets, DAB staining intensity, and ratio of diseased area. B, According to the investigated parameters, Kele was more resistant to B. glumae than IS592BB. Histograms of the resistance score and DAB staining intensity revealed a normal distribution, whereas the disease lesion area based on DAB staining followed a bimodal distribution, and the ratio of diseased spikelets showed a right-skewed distribution. Kele is resistant to B. glumae, whereas IS592BB is a susceptible cultivar. However, upon B. glumae infection, the black hull color of Kele made it challenging to distinguish disease symptoms visually. Although the hull color of IS592BB is brown, distinguishing disease symptoms remained challenging. However, DAB staining enhanced accuracy compared with that of field-based resistance evaluation, as resistant regions were decolorized, whereas diseased regions appeared as brown spots. ImageJ software was used to analyze the disease lesion area. To clearly distinguish the diseased regions, panicle images were converted to grayscale, and the total panicle area was measured (marked in red). Within the total panicle area, diseased regions are highlighted in black. To investigate disease severity, pixel values were measured across different regions of the panicle. Areas with intense DAB staining presented lower pixel values, whereas unstained white regions presented higher pixel values

To transplant the mapping population in the NICS field, seeds from each line were treated with Kiman Plus pesticide (25% Prochloraz, HANKOOKSAMGONG, Seoul, Republic of Korea) and incubated at 33 °C in darkness for three days. Sowing was performed on May 2, 2023, and May 4, 2024. After 30 days, the seedlings were transplanted in the NICS field. The plot area for each plant was set at 30 × 15 cm, and plants of each line were transplanted in two rows, with 25 plants per row. In the field, N–P₂O₅–K₂O fertilizer was applied at a rate of 9–4.5–5.7 kg per 10 ha. During this study, plant cultivation and breeding were performed according to standard methods provided by the RDA of Korea and with strict adherence to international regulations and guidelines. In addition, to ensure that cultivation was performed in compliance with local practices, the guidelines of the Convention on the Trade in Endangered Species of Wild Fauna and Flora (CITES) [https://www.cites.org/(accessed on 21 April 2023)] were followed. All the cultivars used in this study, including Kele, IS592BB were provided by Crop Breeding Division, NICS, RDA. And Hitomebore, Satojiman, and Akitakomachi, were provided by National Agrobiodiversity Center, National Institute of Agricultural Sciences, RDA.

Selection of the B. glumae strain and optimization of the inoculation concentration

For B. glumae inoculation, each line was classified on the basis of its heading date and transplanted in the field accordingly. The earliest heading date was August 15, whereas the latest heading date was September 13. During this period, 15 separate inoculations were conducted. Both Kele and IS592BB were inoculated on each inoculation date to allow comparative analysis of disease resistance on the basis of inoculation timing. In this study, panicles were inoculated by direct spraying three days after the heading date in the NICS field. Resistance was evaluated five days after inoculation. The B. glumae strain used for BGR induction was GR20004 (KACC 22751), a strain with stable virulence. GR20004 was obtained from the Crop Foundation Division, NICS, RDA. The obtained GR20004 strain was incubated in LB broth at 28 °C with shaking at 150 RPM for two days. On the day of inoculation, the GR20004 culture was adjusted to a concentration of 0.5 (10⁸ cfu/ml), and 2 ml of the suspension was used to inoculate the panicles of each line by spraying.

Imaging data collection for disease resistance evaluation

Disease symptoms were evaluated five days after inoculation. A 0–10 scale was applied, where a score of 0 was assigned if no spikelets in the panicle were infected (0%) and a score of 10 was assigned if all spikelets were infected (100%). To assess the impact of B. glumae during the grain-filling stage, the grain sterility rate was evaluated 30 days after inoculation. Each spikelet was individually pressed to determine whether the ovary was filled. Spikelets that were found to be completely empty upon pressing were classified as infertile. Finally, spikelet fertility was calculated as the proportion of filled spikelets relative to the total number of spikelets in each panicle. To analyze the lesion area from the images, an EPSON GT-X scanner was used to scan the inoculated panicles at 900 dpi resolution. The diseased area and normal area of each panicle were calibrated using distinguishable colors. The pixel counts of the diseased area and normal area were measured to analyze the relative disease severity. The relative disease severity was calculated by dividing the pixel count of the disease area by the total pixel count and multiplying by 100.

Construction of genetic map and QTL mapping

Polymorphism analysis between the parental lines Kele and IS592BB was performed using 2,565 SNP markers, of which 726 were polymorphic. After removing 20% missing data and 65 overlapping SNPs, 516 informative SNPs were selected for map construction. The genetic map was developed using IciMapping v4.0(Meng et al. 2015) and MapChart v2.0(Voorrips 2002), with genetic distances calculated based on the Kosambi mapping function. The selected SNPs were evenly distributed across all 12 rice chromosomes. QTL analysis was performed using phenotypic data on Field resistant score, Ratio of disease spikelets (%), Quantitative of DAB staining and Ratio of disease area. SNP clustering and genetic distance calculations were conducted using the ‘By LOD’ and ‘By Input’ functions in the ICIM program. The LOD threshold was set at 3.0, determined by 1000 permutations at a significance level of p < 0.05. Identified QTLs were named following the nomenclature system proposed by (McCouch et al. 2016).

DNA extraction and PCR

To extract genomic DNA, the plant samples were ground to powder using a tissue lyser (Qiagen, Cat. 85,300, Hilden, Germany) while being rapidly frozen with liquid nitrogen. The extraction was then performed following with the manufacturer’s instructions for the BioSprint 96 DNA Plant Kit (INDICAL BIOSCIENCE, Cat. SP947057; Leipzig, Germany). The concentration of the extracted genomic DNA was determined using a NanoDrop ND-1000 spectrophotometer (Thermo Fisher, Cat. ND 1000, Waltham, MA, USA), and the DNA was then diluted with ddH₂O to a final concentration of 20 ng/μL. The PCR mixture for amplifying the target region from the diluted DNA included 100 ng of DNA template, 10 pmol each of the forward and reverse primers, 2.5 mM dNTP mixture, 2.0 μL of 10 × Ex Taq buffer (50 mM KCl, 20 mM Tris–HCl (pH 8.0), 2.0 mM MgCl₂), and 1.0 U of Ex Taq polymerase (Inclone Biotech Co., IN5001). Information on the forward and reverse primers is provided in the supplementary data. The total reaction volume was adjusted to 20 µL using nuclease-free water (QIAGEN, Cat. No. 129114). The PCR conditions were as follows: a predenaturation step at 94 °C for 3 min, followed by 30 cycles of denaturation at 94 °C for 30s, annealing at 58 °C for 30s, and extension at 72 °C for 30s and a final extension at 72 °C for 3 min. The PCR product was detected by mixing with 6 × loading dye (SIGMA, Cat. G2526-5ML, Saint Louis, MO, USA) followed by electrophoresis at 220 V on a 1.2% agarose gel containing EtBr (SIGMA, Cat. E1510; Saint Louis, MO, USA) and 1 × Tris/borate/EDTA (TBE) buffer. The amplification of the target region was then confirmed using a UV transilluminator (Bio-Rad, Cat. 170-8070, Hercules, CA, USA).

Functional characterization of candidate genes in QTL regions through gene ontology (GO) enrichment analysis

Gene ontology (GO) analysis was performed to identify the functional roles of candidate genes screened through QTL mapping. Candidate genes within the QTL region were screened using the rice genome assembly (Oryza PanGenome Release 7.0, https://www.gramene.org), and annotation information was assigned to each candidate gene. GO terms were subsequently assigned to each candidate gene and classified into the cellular component, molecular function, or biological process category. GO enrichment analysis was performed using the AgriGO v2.0 platform (https://systemsbiology.cau.edu.cn/agriGOv2/). A list of candidate genes distributed within the QTL region was used as input data for the AgriGO v2.0 platform to identify overrepresented functional categories. To enhance the reliability of the analysis, Fisher’s exact test and Yekutieli’s method for multiple test correction were applied. Additionally, REVIGO (http://revigo.irb.hr/) was used to remove redundant GO terms and generate plots for visual representation of significantly enriched functional categories.

RNA extraction and transcript level analysis

Panicles from Kele, IS592BB, and KIRILs were sampled at 0, 12, 24, 48, 72, 96, and 120h after B. glumae inoculation. The sampled panicles were used for total RNA extraction following the manufacturer’s instructions for the RNeasy Micro Kit (QIAGEN, Cat. 74,004, Hilden, Germany). One microgram of the extracted RNA was used for cDNA synthesis following the manufacturer’s instructions for the qPCRBIO cDNA Synthesis Kit (PCRBIOSYSTEMS, USA). The synthesized cDNA was then diluted 1:10 (v/v) in nuclease-free water (QIAGEN, Cat. No. 129114). The reaction mixture for analyzing gene expression levels was prepared according to the manufacturer’s instructions for qRT‒PCR Master Mix (BIOFACT, Cat. RQ351-50h, Seoul, Korea). Gene expression analysis was performed using the Eco™ Real-Time PCR System (Illumina, Cat. EC-900-1001, CA, USA). To normalize the cycle threshold values of each sample, the housekeeping gene OsACTIN was used as a standard, and candidate genes were amplified using the corresponding primer set presented in Supplementary Table S1. The relative expression levels in each sample were calculated using the 2^‒ΔΔCt method.

Haplotype analysis of OsBGq1 and variation

Variations in OsBGq1 were identified from 4726 rice accession sequences available in RiceVarMap v2.0 (https://ricevarmap.ncpgr.cn/). Haplotype analysis was performed using 13 SNP variations identified within the 1 kb region upstream of OsBGq1 (vg0133816358, vg0133816465, vg0133816958, vg0133817050, vg0133817242, vg0133817697, vg0133817698, vg0133817729, vg0133817982, vg0133817983, vg0133818572, vg0133818578, vg0133820555). As a result, five distinct haplotypes with different genotypes were identified.

Detection and quantification of H₂O₂

A DAB staining method was used to detect H₂O₂ accumulation induced by B. glumae in rice panicles. Panicles inoculated with B. glumae were sampled five days after inoculation. To ensure proper penetration of the DAB solution into the cells, the sampled panicles were immediately washed with distilled water (DW) to remove any surface contaminants. The panicles were then completely submerged in 1 mg/mL DAB solution (pH 3.8). The staining reaction was carried out by incubating the samples at 25 °C with shaking at 130 RPM in darkness for 16h. After 16h, the DAB solution was removed, and the panicles were decolorized with 100% ethanol. The areas where H₂O₂ was produced due to B. glumae infection were identified as brown spots on the panicles. To quantify and compare H₂O₂ levels, ImageJ software (https://imagej.nih.gov/ij/) was used. Ten panicles from different plants of Kele, IS592BB, and KIRILs were sampled and subjected to DAB staining. Representative images are presented in the figures. In addition, the modified method proposed by Rao et al. (2000) was followed for extraction and analysis of the H₂O₂ present in each panicle (Rao et al. 2000). The H₂O₂ content in the panicles was analyzed following the manufacturer’s instructions for the Quantitative Peroxide Assay Kit (Thermo Fisher Scientific, Cat. 23,280, MA, USA). The absorbance values of the extracts were measured at OD₅₆₀ after calibration against a standard curve constructed using known H₂O₂ concentrations.

Analysis of the MDA, CAT, POD, SOD, and Proline contents

The proline content in panicles was analyzed following the method proposed by Bates et al. (1973) (Bates et al. 1973). Panicles inoculated with B. glumae were ground to obtain 5 g of powder. The powdered samples were then homogenized with 3% sulfosalicylic acid (w/v, SIGMA, Cat. S2130, SL, USA) at 100 °C for 10 min. The samples were centrifuged at 4500 RPM for 30 min, and then, 2 mL of the supernatant was transferred to a new Eppendorf tube. An equal volume of glacial acetic acid and acid ninhydrin was added to the supernatant in the Eppendorf tube. The mixture was then heated at 100 °C for 40 min and subsequently cooled on ice. Finally, the proline level in the presence of B. glumae was determined by measuring the absorbance at OD₅₂₀.

The malondialdehyde (MDA) content was analyzed following the method proposed by Heath and Packer (1968) (Heath and Packer 1968). Panicle samples were ground to obtain 1 g powder, and 10 mL of 10% trichloroacetic acid (v/v) was added. After centrifugation at 4500 RPM for 10 min, 2 mL of the supernatant was transferred to a new Eppendorf tube. An equal volume (2 mL) of thiobarbituric acid was added to the supernatant, and the mixture was incubated at 100 °C for 15 min. After incubation, the samples were chilled on ice. The MDA content was determined by measuring the OD₄₅₀, OD₅₃₂, and OD₆₀₀ values and applying them in the formula proposed by Shi et al. (2012): MDA content (μmol l − 1) = 6.45(OD₅₃₂ − OD₆₀₀) − 0.56 OD₄₅₀ (Shi et al. 2012).

Catalase (CAT) activity was analyzed by modifying and applying the method proposed by Beers and Sizer (1952) (Beers and Sizer 1952). The reaction mixture consisted of 2 mL of sodium phosphate buffer (50 mM, pH 7.0), 0.5 mL of H₂O₂ (40 mM), and 0.5 mL of enzyme extract, with a total volume of 3.0 mL. Finally, the decomposition of H₂O₂ was analyzed by measuring the absorbance at OD₂₄₀.

Superoxide dismutase (SOD) activity was analyzed using the method proposed by Duan et al. (2012) (Duan et al. 2012). SOD activity was defined as the minimum amount of enzyme required to inhibit the photochemical reduction of NBT chloride, measured as the OD₅₆₀, by 50%.

Peroxidase (POD) activity was determined using the guaiacol oxidation method proposed by Weydert and Cullen (2010) (Weydert and Cullen 2010). The reaction mixture consisted of 2 mL of sodium acetate buffer (50 mM, pH 5.0), 0.5 mL of guaiacol (20 mM), 0.4 mL of H₂O₂ (40 mM), and 0.1 mL of enzyme extract, with a total reaction volume of 3.0 mL. The OD₂₄₀ of the POD reaction mixture was measured at 1-min intervals within the first 3 min after the reaction was initiated.

Quantification of phytoalexins in rice through LC‒MS/MS analysis

To quantify phytoalexins produced in rice panicles, panicles were sampled five days after B. glumae inoculation. The samples were ground in liquid nitrogen to obtain 100 mg of powdered tissue. Four milliliters of 80% methanol was thoroughly mixed with the 100 mg of powdered tissue, and the mixture was incubated overnight in a shaking incubator set at 4 °C and 150 RPM. One milliliter of the extract was transferred to a new Eppendorf tube and centrifuged at 13,000 RPM for 10 min at 4 °C. After centrifugation, 5 μL of the supernatant, free of cell debris, was used for LC‒MS/MS analysis. Independent triplicate experiments were conducted. The monitoring transitions were set as follows: phytocassanes A and E, m/z 317 → 299; phytocassane B, m/z 335 → 317; phytocassane C, m/z 319 → 301; momilactone A, m/z 315 → 271; and momilactone B, m/z 331 → 269.

Scanning electron microscopy analysis of B. glumae in rice hulls

For SEM analysis of rice hulls, hulls were cut with a scalpel (SIGMA, Cat. S2646, Saint Louis, MO, USA) to separate the inner and outer surfaces. Each sample was coated with gold under vacuum using an ion sputtering device (JEOL, Cat. JFC-1100E, Tokyo, Japan). The morphology of the rice hull after B. glumae inoculation was observed using a scanning electron microscope (JSM-6390LV, JEOL) with an accelerating voltage of 10 kV and a spot size of 30 nm. The experiment was conducted in independent triplicates, and the most representative images are presented in the figures.

Statistical analysis

B. glumae infection-related phenotype data were subjected to statistical analysis using R (version 4.1.3, The R Foundation for Statistical Computing). To obtain the phenotypic data, B. glumae was used to inoculate five plants, from which three plants were randomly selected. From each selected plant, five panicles were randomly sampled. All the phenotypic data were investigated for five independent replicates. Statistical analysis of the measured traits, including mean and standard deviation comparisons, was conducted using Duncan’s multiple range test (DMRT) implemented in the agricolae package in R. The significance obtained from DMRT, one-way analysis of variance (ANOVA), and t tests was evaluated at P < 0.05.

Results

Evaluation of B. glumae resistance in the mapped population

To perform QTL mapping for resistance to B. glumae, the following parameters were investigated in infected panicles: field resistance score, ratio of diseased area (%), DAB staining intensity, and ratio of diseased spikelets (%) (Fig. 1A, B). Kele presented lower values than IS592BB did for all the investigated parameters, including the field resistance score (Kele 3.1 ± 0.8, IS592BB 8.5 ± 1.3), ratio of diseased area (35.27 ± 1.8%, 78.85 ± 2.5%), DAB staining intensity (311.22 ± 3.5, 358.65 ± 2.8), and ratio of diseased spikelets (33.2 ± 2.5%, 95.9 ± 3.8%). Additionally, all the phenotypic parameters investigated in the KIRILs exhibited a wide range of variation, encompassing the values observed for both Kele and IS592BB, with evidence of transgressive segregation. The histograms of the resistance score and DAB staining intensity revealed a normal distribution, whereas the ratio of the diseased area showed a bimodal distribution, and the ratio of diseased spikelets followed a right-skewed distribution. Correlation analysis of the investigated parameters revealed positive correlations of 0.5 or higher among all the parameters (Supplementary Figure S1). Notably, the highest correlation was observed between the field resistance score and the ratio of diseased spikelets (r = 0.96). Strong correlations were also detected between the DAB value and resistance score (r = 0.91), as well as between the DAB value and ratio of diseased spikelets (r = 0.89).

Identification of B. glumae resistance-related QTLs in KIRILs

Four distinct QTLs (qFRS1, qRDS1, qQDS1, and qRDA1) were identified on chromosome 1, within the Chr01_24592710-Chr01_37274755 region, for the parameters investigated in B. glumae-damaged rice panicles, including the field resistance score, ratio of diseased area (%), DAB staining intensity, and ratio of diseased spikelets (%) (Supplementary Table S2). The identified QTLs within the Chr01_24592710-Chr01_37274755 region were qFRS1 [LOD: 5.98, phenotype variation explained (PVE): 15.41%] and qRDS1 (LOD: 5.29, PVE: 18.56%). Additionally, the QTLs identified within the Chr01_24592710-Chr01_37274755 region were qQDS1 (LOD: 9.58, PVE: 22.02%) and qRDA1 (LOD: 8.44, PVE: 31.59%). Therefore, the Chr01_24592710-Chr01_37274755 region on chromosome 1 represents an overlapping QTL, identified through phenotyping of B. glumae-infected panicles, with distinct LOD scores and PVE values for each trait.

Fine mapping and screening of candidate genes associated with B. glumae resistance

Following B. glumae infection in KIRILs, the field resistance score, ratio of diseased area (%), DAB staining intensity, and ratio of diseased spikelets (%) were measured. A resistance-associated QTL was identified within the Chr01_24592710-Chr01_37274755 on chromosome 1, and the KIRILs were classified on the basis of their respective genotypes (Fig. 2). Genotypic analysis of six markers (Chr01_33340511, Chr01_33472174, Chr01_33838140, Chr01_34249454, Chr01_34327765, Chr01_37052448) within the QTL region was conducted in Kele, IS592BB, and 189 KIRILs. As a result, five major informative recombinants (KIRIL1, KIRIL2, KIRIL3, KIRIL4, and KIRIL5) were identified. The field resistance score, ratio of diseased area (%), DAB staining intensity, and ratio of diseased spikelets (%) in KIRIL1, KIRIL2, and KIRIL4 were similar to those of Kele. Consequently, the genomic interval of the QTL region was narrowed down between Chr01_33472174 and Chr01_33838140. According to the Rice Annotation Project Database, 16 candidate genes are located within the Chr01_33472174 and Chr01_33838140 marker interval: Os01g0794400, Os01g0794500, Os01g0795000, Os01g0796400, Os01g0797600, Os01g0797800, Os01g0798900, Os01g0799000 (OsBGq1), Os01g0799500, Os01g0799900, Os01g0800300, Os01g0800400, Os01g0800500, Os01g0801000, Os01g0801500, and Os01g0801600. Descriptions of these candidate genes are provided in Supplementary Table S3.

Fig. 2.

Fine mapping of the Chr01_24592710-Chr01_37274755 region on chromosome 1 in KIRILs to screen for candidate genes associated with B. glumae resistance. The black bars represent chromosome segments derived from Kele (resistant rice cultivar), whereas the white bars indicate segments inherited from IS592BB (susceptible rice cultivar). Six additional markers (Chr01_33340511, Chr01_33472174, Chr01_33838140, Chr01_34249454, Chr01_34327765, and Chr01_37052448) were incorporated within the Chr01_24592710– Chr01_37274755 region. In KIRILs, these markers identified five distinct genotypes. KIRIL1, KIRIL2, KIRIL3, KIRIL4, and KIRIL5 are representative recombinant inbred lines derived from the Kele and IS592BB cross, each carrying a distinct recombined genotype within the KIRIL population. Following B. glumae infection, phenotype parameters (RDA, DS, RS, and RDS) were measured and are summarized to the right of the genotype information. This figure illustrates the distribution of 16 candidate genes located within the Chr01_33472174– Chr01_33838140 region. The size of the blue boxes represents the relative length of the genomic DNA sequence, whereas the direction of the arrows indicates the transcriptional orientation of the candidate genes. The data are shown as the means ± SDs from five independent biological replicates per line. Means sharing the same letter are not significantly different (P < 0.05), as determined by Duncan’s multiple range test

Gene ontology enrichment analysis of candidate genes for functional characterization

GO enrichment analysis of candidate genes screened within the Chr01_33472174 and Chr01_33838140 region was performed using AgriGO v2.0, with significance thresholds set at P < 0.01 and FDR < 0.05. The identified GO terms were classified into three main categories, namely, biological process, cellular component, and molecular function, each of which was further subdivided (Fig. 3). Five GO terms were identified in the biological process category (Fig. 3A), with the most enriched terms exhibiting the following order: GO:0043086 (negative regulation of catalytic activity), GO:0044092 (negative regulation of molecular function), GO:0050790 (regulation of catalytic activity), and GO:0065009 (regulation of molecular function). Sixteen GO terms were identified in the cellular component category (Fig. 3B), with the most enriched terms appearing in the following order: GO:0009536 (plastid), GO:0016023 (cytoplasmic membrane-bounded vesicle), GO:0031410 (cytoplasmic vesicle), GO:0043229 (intracellular membrane-bounded organelle), GO:0031988 (membrane-bounded vesicle), GO:0044444 (cytoplasmic part), GO:0043229 (intracellular organelle), GO:0043227 (membrane-bounded organelle), GO:0031982 (vesicle), GO:0005737 (cytoplasm), GO:0043226 (organelle), GO:0044424 (intracellular part), GO:005622 (intracellular), GO:0044464 (cell part), and GO:0005623 (cell). Twelve GO terms were identified in the molecular function category (Fig. 3C), with the most enriched terms appearing in the following order: GO:0004252 (serine-type endopeptidase activity), GO:0042802 (identical protein binding), GO:0008236 (serine-type peptidase activity), GO:0017171 (serine hydrolase activity), GO:0070011 (peptidase activity acting on L-amino acids), and GO:0008233 (peptidase activity).

Fig. 3.

GO terms and pathway annotations of candidate genes associated with B. glumae resistance screened within the Chr01_33472174– Chr01_33838140 region. Singular enrichment analysis (SEA) of candidate genes was performed using AgriGO, and the genes were classified into three categories: molecular function, biological process, and cellular component. Each GO term and its corresponding ID are displayed within boxes, and significantly enriched GO terms (P < 0.01, FDR < 0.05) are highlighted in distinct colors to indicate statistical significance. The biological process A category includes six subcategories, the cellular component B category contains 16 subcategories, and the molecular function C category comprises 12 subcategories. The gradient shading in the pathway visualization represents the degree of GO term enrichment, with color intensity positively correlated with enrichment levels and indicative of the corresponding p value

Analysis of the expression levels of candidate genes by qRT‒PCR

To identify the effects of the candidate genes, their expression levels were analyzed via qRT‒PCR in B. glumae-damaged plants, including Kele, IS592BB, KIRIL1, KIRIL2, KIRIL3, KIRIL4, KIRIL5, Hitomebore, Satojiman, and Akitakomachi plants (Fig. 4). Kele is a B. glumae-resistant rice cultivar. KIRIL1, KIRIL2, KIRIL3, KIRIL4, and KIRIL5 are RILs derived from a cross between Kele and IS592BB, representing distinct genotypes within the KIRIL population. KIRIL1, KIRIL2, and KIRIL4 share the same genotype as Kele within the Chr01_33472174 and Chr01_33838140, whereas KIRIL3 and KIRIL5 possess the IS592BB genotype in this region. Kele is a highly B. glumae-resistant rice cultivar, whereas IS592BB, Hitomebore, Satojiman, and Akitakomachi are highly susceptible to B. glumae. The relative expression levels of 16 candidate genes identified through fine mapping [Os01g0794400, Os01g0794500, Os01g0795000, Os01g0796400, Os01g0797600, Os01g0797800, Os01g0798900, Os01g0799000 (OsBGq1), Os01g0799500, Os01g0799900, Os01g0800300, Os01g0800400, Os01g0800500, Os01g0801000, Os01g0801500, and Os01g0801600] were analyzed in panicles sampled at 0, 6, 12, 24, 48, 72, 96, and 120h after B. glumae inoculation. The analysis was conducted in Kele, IS592BB, KIRIL1, KIRIL2, KIRIL3, KIRIL4, KIRIL5, Hitomebore, Satojiman, and Akitakomachi. The expression levels of Os01g0794400, Os01g0796400, Os01g0797600, Os01g0799900, and Os01g0800400 were upregulated in response to B. glumae infection but remained consistent across Kele, KIRIL1, KIRIL2, and KIRIL4 (which Kele type of the Chr01_33472174– Chr01_33838140 region), as well as KIRIL3 and KIRIL5 (which IS592BB type of the Chr01_33472174– Chr01_33838140 region) and the susceptible rice cultivars IS592BB, Hitomebore, Satojiman, and Akitakomachi. The expression levels of Os01g0794500, Os01g0799500, and Os01g0801500 remained higher in Kele, KIRIL1, KIRIL2, and KIRIL4 than in KIRIL3, KIRIL5, IS592BB, Hitomebore, Satojiman, and Akitakomachi for up to 48h after B. glumae inoculation. However, by 72h postinoculation, the expression levels of these genes had equalized across all the genotypes, including Kele, IS592BB, KIRIL1, KIRIL2, KIRIL3, KIRIL4, KIRIL5, Hitomebore, Satojiman, and Akitakomachi. The expression levels of Os01g0800500, Os01g0795000, Os01g0799000 remained higher in Kele, KIRIL1, KIRIL2, and KIRIL4 than in KIRIL3, KIRIL5, IS592BB, Hitomebore, Satojiman, and Akitakomachi for up to 12h after B. glumae inoculation. However, by 24h postinoculation, the expression levels of these genes had equalized across all the genotypes. The expression levels of Os01g0799000 (OsBGq1) remained higher in Kele, KIRIL1, KIRIL2, and KIRIL4 than in KIRIL3, KIRIL5, IS592BB, Hitomebore, Satojiman, and Akitakomachi for up to 120h after B. glumae inoculation. In contrast, the expression levels of Os01g0797800, Os01g0800300, Os01g0801000, and Os01g0801600 remained consistently low throughout the 120-h period postinoculation, with no differences observed among Kele, IS592BB, KIRIL1, KIRIL2, KIRIL3, KIRIL4, KIRIL5, Hitomebore, Satojiman, and Akitakomachi.

Fig. 4.

Expression level analysis of potential candidate genes associated with B. glumae resistance. The expression levels of 16 candidate genes screened within the Chr01_33472174–Chr01_33838140 region on chromosome 1 were analyzed in Kele, IS592BB, KIRILs, Hitomebore, Satojiman, and Akitakomachi. Panicles from each line were sampled at 0h, 6h, 12h, 24h, 48h, 72h, 96h, and 120h after B. glumae treatment, along with control samples, to compare the relative expression levels of the screened candidate genes. OsBGq1 expression levels increased sharply in KIRIL1, KIRIL2, and KIRIL4 upon B. glumae infection. The expression level continued to increase until 24h post infection, after which it tended to decrease. However, in Kele, KIRIL1, KIRIL2, and KIRIL4, OsBGq1 expression consistently remained higher than that in IS592BB, Hitomebore, Satojiman, and Akitakomachi. ns, Not significantly different. *Significantly different at the 0.05 level. **Significantly different at the 0.01 level

Sequence variation of Os01g0799000 (OsBGq1) between indica rice and japonica rice

To identify the impact of OsBGq1 variations on resistance to B. glumae, a total of 13 sequence variations were identified in the upstream, exon, intron, and downstream regions using the RiceVarMap v2.0 database (Fig. 5). These variations included SNP1 (G > T), SNP2 (A > T), SNP3 (G > C), SNP4 (C > T), SNP5 (G > C), SNP6 (A > C), SNP7 (G > A), SNP8 (A > T), SNP9 (A > G), SNP10 (G > T), SNP11 (G > T), SNP12 (T > C), and SNP13 (C > T). On the basis of these variations, 4,713 rice accessions were classified into five distinct haplotypes (Hap1, Hap2, Hap3, Hap4, and Hap5). Hap1 was composed of 41.9% Indica, 8.0% Aus, and 45.5% Japonica. In contrast, Hap2 consisted of 97.0% Indica, 0.5% Aus, and 1.1% Japonica, whereas Hap3 comprised 98.3% Indica and 0.0% Aus and Japonica. Hap4 lacked all the subgroups (0.0%), and Hap5 was exclusively found in Indica (100.0%). In the cultivar Kele, the OsBGq1 sequence corresponded to the Hap2 haplotype. Hap2 variations hold significant potential for breeding programs aimed at improving B. glumae resistance in Japonica rice.

Fig. 5.

Haplotype classification of the OsBGq1 DNA sequence. A total of 13 SNPs were detected in the upstream, downstream, exon, and intron regions of OsBGq1. Among them, 10 SNPs were located in exons, one SNP in an intron, and the remaining two SNPs in the upstream region. Hap1 was found at similar frequencies in both Indica (42.0%) and Japonica (45.6%), whereas Hap2, Hap3, Hap4, and Hap5 were present predominantly in Indica

Modulation of oxidative stress indicators induced by B. glumae

B. glumae infection induces changes in proline, H₂O₂, MDA, and phenolic levels, all of which serve as indicators of oxidative stress (Fig. 6A). Before infection (pretreatment), there were no differences in the levels of proline, H₂O₂, MDA, and phenolics among Kele, IS592BB, KIRIL1, KIRIL2, KIRIL3, KIRIL4, KIRIL5, Hitomebore, Satojiman, and Akitakomachi, with all cultivars exhibiting similar baseline levels. However, upon B. glumae infection, the levels of proline, H₂O₂, MDA, and phenolics increased in all investigated lines compared with those under normal conditions. Notably, Kele, KIRIL1, KIRIL2, and KIRIL4 presented higher levels of proline and phenolics but lower levels of H₂O₂ and MDA than did KIRIL3, KIRIL5, IS592BB, Hitomebore, Satojiman, and Akitakomachi. These results suggest that the Kele-type of the Chr01_33472174 and Chr01_33838140 region modulates oxidative stress indicators, thereby attenuating the oxidative stress response and enhancing resistance to B. glumae.

Fig. 6.

KIRILs with the introduced Chr01_33472174– Chr01_33838140 region efficiently removed ROS and enhanced resistance to B. glumae. A, In all B. glumae-infected rice lines, including Kele, IS592BB, KIRILs, Hitomebore, Satojiman, and Akitakomachi, the accumulation of proline, H₂O₂, MDA, and phenolics was observed, along with changes in the activities of ROS-scavenging enzymes such as SOD, POD, and CAT and in DPPH scavenging. Kele, KIRIL1, KIRIL2, and KIRIL4 presented lower accumulation of proline, H₂O₂, MDA, and phenolics than did IS592BB, KIRIL3, KIRIL5, Hitomebore, Satojiman, and Akitakomachi due to the increased activities of SOD, POD, and CAT and DPPH scavenging. As a result, Kele, KIRIL1, KIRIL2, and KIRIL4 exhibited improved resistance to B. glumae. B, The formation of H₂O₂ induced by B. glumae in Kele, IS592BB, KIRILs, Hitomebore, Satojiman, and Akitakomachi was identified as brown spots through DAB staining. When DAB staining was performed, Kele, KIRIL1, KIRIL2, KIRIL3, and KIRIL4 presented fewer brown spots than did IS592BB, Hitomebore, Satojiman, and Akitakomachi. Among them, KIRIL1, KIRIL2, and KIRIL4 presented levels of brown spot formation similar to those of Kele. The panicles presented in the images represent the most representative results from five independent biological replicates for each line. The bar graphs represent the mean ± standard deviation, and means labeled with the same letter are not significantly different according to Duncan’s multiple range test (P < 0.05)

Additionally, before treatment, the activities of the antioxidant enzymes SOD, POD, and CAT were similar across all investigated lines, including Kele, IS592BB, KIRIL1, KIRIL2, KIRIL3, KIRIL4, KIRIL5, Hitomebore, Satojiman, and Akitakomachi, with no significant differences (Fig. 6A). However, upon B. glumae infection, the activities of SOD, POD, and CAT increased in all the lines compared with the preinfection levels. However, compared with those in KIRIL3, KIRIL5, IS592BB, Hitomebore, Satojiman, and Akitakomachi, the activities of SOD, POD, and CAT increased significantly in Kele, KIRIL1, KIRIL2, and KIRIL4. Additionally, DPPH scavenging activity was greater in Kele, KIRIL1, KIRIL2, and KIRIL4 than in KIRIL3, KIRIL5, IS592BB, Hitomebore, Satojiman, and Akitakomachi. These results suggest that the Chr01_33472174 and Chr01_33838140 region derived from Kele enhances resistance by increasing the activities of the antioxidant enzymes SOD, POD, and CAT, as well as improving DPPH scavenging activity, thereby regulating ROS levels (Fig. 6B).

Accumulation of phytoalexins enhances resistance to B. glumae

The accumulation of phytoalexins, including momilactone A, momilactone B, phytocassane B, phytocassane C, phytocassane E, and sakuranetin, in B. glumae-infected panicles was quantified and compared among rice lines. The results of the experiment showed Fig. 7A, B. glumae infection induced the accumulation of these phytoalexins in rice panicles. In all the investigated lines, including Kele, IS592BB, KIRIL1, KIRIL2, KIRIL3, KIRIL4, KIRIL5, Hitomebore, Satojiman, and Akitakomachi, the accumulation of momilactone A, momilactone B, phytocassane B, phytocassane C, phytocassane E, and sakuranetin increased following B. glumae infection compared with pretreatment levels. However, the accumulation of momilactone A, momilactone B, phytocassane B, phytocassane C, phytocassane E, and sakuranetin was significantly greater in Kele, KIRIL1, KIRIL2, and KIRIL4 than in KIRIL3, KIRIL5, IS592BB, Hitomebore, Satojiman, and Akitakomachi. These results suggest that the increased accumulation of these phytoalexins in Kele, KIRIL1, KIRIL2, and KIRIL4 contributes to improved resistance against B. glumae.

Fig. 7.

Accumulation of phytoalexins (omilactone A, momilactone B, phytocassane B, phytocassane C, phytocassane E, and sakuranetin) in rice panicles infected with B. glumae. A, The accumulation levels of momilactone A, momilactone B, phytocassane B, phytocassane C, phytocassane E, and sakuranetin were compared between rice panicles under normal conditions and those at 5 days (150h) after B. glumae infection. Under normal conditions, the accumulation levels of all investigated phytoalexins were low across Kele, IS592BB, KIRILs, Hitomebore, Satojiman, and Akitakomachi. However, at 5 days after B. glumae infection, all phytoalexins accumulated at significantly greater levels in Kele, KIRIL1, KIRIL2, and KIRIL4. In contrast, IS592BB, Hitomebore, Satojiman, and Akitakomachi presented relatively low levels of phytoalexin accumulation. B, Scanning electron microscopy (SEM) was used to analyze the presence of B. glumae on the surface and inner tissues of infected rice panicle spikelets. Red arrows indicate B. glumae. In Kele, KIRIL1, KIRIL2, and KIRIL4, B. glumae were detected at very low levels on both the surface and inner tissues. In contrast, IS592BB, Hitomebore, Satojiman, and Akitakomachi presented extensive B. glumae coverage on the spikelet surface, with substantial colonization also observed in the inner tissues. The bars indicate the mean values ± standard deviation. Means with the same letter are not significantly different (P < 0.05) according to Duncan’s multiple range test. K; Kele, I; IS592BB, K1; KIRIL1, K2; KIRIL2, K3; KIRIL3, K4; KIRIL4, K5; KIRIL5, H; Hitomebore, S; Satojiman, A; Akitakomachi

Five days after B. glumae inoculation, SEM was used to observe the spikelets in the panicles of Kele, IS592BB, KIRIL1, KIRIL2, KIRIL3, KIRIL4, KIRIL5, Hitomebore, Satojiman, and Akitakomachi. Although B. glumae spread was observed in all the lines, the extent of B. glumae spreading varied among them. In KIRIL5, IS592BB, Hitomebore, Satojiman, and Akitakomachi, B. glumae had fully spread across the spikelets, covering most of the surface, and KIRIL3 also presented a high level of B. glumae spread (Fig. 7B). In contrast, B. glumae were either not observed or present at very low levels on the spikelet surfaces of Kele, KIRIL1, KIRIL2, and KIRIL4. This effect was more pronounced on the outer surface of the spikelets than on the inner parts.

Discussion

BGR, caused by B. glumae during the heading stage of rice, leads to the formation of brown spots on each spikelet, directly causing spikelet sterility (Zhou-qi et al. 2016). This infection disrupts the rice maturation process, ultimately resulting in yield loss (Benchlih et al. 2025). To date, two QTLs associated with resistance to B. glumae have been reported: RBG2 (Mizobuchi et al. 2015) on chromosome 1 and qBPB-3-2 (Pinson et al. 2010) on chromosome 3. However, no additional QTLs have been identified, and neither of these QTLs has been finely mapped to a specific gene region. The development of B. glumae-induced disease is strongly influenced by environmental conditions at the time of inoculation, making field inoculation particularly challenging (Seo et al. 2015). Therefore, the previously reported RBG2 was identified on the basis of in vitro studies (Mizobuchi et al. 2013). Since breeding is strongly influenced by environmental factors, the results obtained from in vitro studies may not always be applicable to actual field conditions (Teressa et al. 2021). In particular, RBG2 was previously reported as a broad QTL region spanning approximately 502 kb on chromosome 1, without further refinement or systematic screening of candidate genes (Mizobuchi et al. 2015). In contrast, by applying image-based phenotyping, the present study was able to narrow the QTL interval and identify 16 candidate genes potentially associated with BGR resistance, among which OsBGq1 was selected as a potential resistance-associated candidate gene.

In this study, genetically distinct ecotypes, Kele (indica) and IS592BB (japonica), were employed to identify candidate genes associated with resistance to BGR. The indica ecotype harbors unique allelic variations shaped by adaptation to biotic stresses and has been widely used for identifying resistance genes against rice blast (Khan et al. 2018) and bacterial leaf blight (Suh et al. 2013). In contrast, japonica represents a genetically distinct group with agronomically important traits. The substantial genetic distance between indica and japonica increases polymorphism density and improves the resolution of genetic map construction (Wang et al. 2012), thereby enhancing the power to precisely identify QTLs associated with BGR resistance.

Accurate phenotypic evaluation is essential for successful QTL mapping (Houle et al. 2010). Errors in phenotypic evaluation can lead to the detection of QTLs unrelated to the target trait, thereby compromising the reliability of the study (Yang et al. 2020). Subjective or ambiguous phenotypic evaluation methods can weaken the association between genotype and phenotype, thereby reducing the power of QTL detection (Großkinsky et al. 2015). Kele is known to exhibit high resistance to B. glumae. However, Kele’s hull color is black, and in the Kele/IS592BB RIL population (KIRILs), hull color varied, exhibiting shades of black and brown. This variation made it particularly challenging to distinguish disease symptoms caused by B. glumae infection visually. Therefore, to assess damage caused by B. glumae infection more precisely, the DAB staining method was used in this study, with scanner-based image data collection used for quantitative analysis. This approach enabled the visualization and quantification of ROS accumulation within tissues, allowing more systematic assessment of stress responses in infected plants (Hu et al. 2017). Compared with conventional visual assessments, this approach enables more precise phenotyping and is expected to provide critical insights into the mechanisms underlying the response of plants to B. glumae infection (Araus and Cairns 2014). Transient increases in ROS are important for activating plant defense signaling; however, excessive or prolonged ROS accumulation can cause oxidative damage to host tissues and promote disease progression (Torres et al. 2006). In this study, the reduced ROS levels observed in association with increased OsBGq1 expression likely reflect improved ROS homeostasis during disease development, which may contribute to enhanced resistance to BGR.

Compared with traditional evaluation methods, image analysis allows more refined assessment and facilitates the advancement of precision agriculture (Zhang and Kovacs 2012). A comparison between traditional evaluation methods and image analysis revealed a strong correlation, demonstrating the reliability of the approach. Additionally, image analysis enables the assessment of previously unexamined aspects, further enhancing phenotypic evaluation (Li et al. 2019). Image analysis has improved the precision and efficiency of agriculture, and its application in breeding has recently been explored as a strategy for precision breeding (Zhang and Kovacs 2012). Bai et al. (2023) utilized UAV imagery to analyze disease severity to detect QTLs associated with resistance to bacterial leaf blight (Bai et al. 2023). By applying the collected data to QTL mapping, they identified qBB2R11a and qBB2R11b within the R11D100–RM254 region of chromosome 11, as well as qBB2R11c and qBB3R11 within the RM254–RM2064 region (Bai et al. 2023). This region is highly enriched with genes previously identified as major contributors to bacterial leaf blight resistance, including Xa4 (Hur et al. 2016), Xa30 (Yang et al. 2022), Xa35 (Fiyaz et al. 2022), and Xa36 (Yang et al. 2022). In this study, for the first time, DAB staining was applied to the Kele/IS592BB RIL population to quantify disease severity and analyze stress responses. Image analysis was used to assess lesion area and disease severity. Ultimately, the region on chromosome 1 with Chr01_24592710-Chr01_37274755 was successfully narrowed down to the Chr01_33472174-Chr01_33838140 interval. This region is near to RBG2 [28], so integration of DAB-based quantitative phenotyping with traditional QTL mapping approaches is expected to provide objective indicator for breeding programs aimed at improving disease resistance.

Previous in vitro studies have been limited in accurately reflecting actual field conditions (Khaled et al. 2018). In contrast, this study enhanced the practicality and effectiveness of this method by evaluating resistance expression under natural conditions through inoculation with B. glumae in the field via direct spraying. Additionally, image analysis was applied to improve the accuracy of resistance assessment. As a result, a QTL associated with B. glumae resistance was identified within the Chr01_33472174-Chr01_33838140 on chromosome 1. Within this region, Os01g0799000 (OsBGq1) was screened as a potential candidate gene involved in B. glumae resistance. We propose that OsBGq1 enhances B. glumae resistance by rapidly degrading the ROS accumulated due to infection and promoting the accumulation of phytoalexins. These findings suggest that the activity of OsBGq1 may contribute to restricting fungal hyphal expansion in rice panicles by increasing phytoalexin accumulation.

OsBGq1 encodes an NLR domain; such domains play a crucial role in plant immune responses by recognizing pathogens and transmitting defense signals (Jones and Dangl 2006). FLS2 and Xa21 both encode LRR domains, recognizing bacterial flagellin and Xanthomonas pathogens, respectively, thereby activating defense responses (Di Gaspero and Cipriani 2003). The LRR domain functions as a transcription factor that regulates the expression of antimicrobial genes (Goff and Ramonell 2007). Pathogen invasion induces the activation of the salicylic acid (SA), jasmonic acid (JA), and ethylene (ET) signaling pathways, which are closely associated with PR genes and antimicrobial proteins (Heil and Bostock 2002). Arabidopsis RPS2 (Bent et al. 1994) and rice Piz-t (Jeung et al. 2007) also encode LRR domains, which regulate phytoalexin synthesis and the production of antimicrobial secondary metabolites, thereby increasing pathogen resistance (Elmore et al. 2012).

Given that the rice phytoalexin sakuranetin has been reported to exhibit antifungal activity against B. glumae, the increased resistance observed in KIRIL1, KIRIL2, and KIRIL4, which carry the Kele-derived Chr01_33472174-Chr01_33838140 genotype, suggests a possible mechanism. We propose that this resistance is at least partially driven by the upregulation of OsBGq1, leading to increased phytoalexin accumulation and increased ROS degradation. Thus, although OsBGq1 does not appear to directly synthesize antimicrobial compounds, our data suggest that it may be involved in the regulation of plant defense signaling and disease resistance-associated protein networks that contribute to enhanced resistance against B. glumae. In this study, OsBGq1, together with other LRR domain-encoding genes, is proposed as a potential regulatory component of plant immune responses and antimicrobial secondary metabolite-related pathways. However, direct functional validation using transgenic or gene-editing approaches will be required to confirm the precise role of OsBGq1 in pathogen resistance.

This study proposes that OsBGq1, a member of the NLR family, may positively regulate resistance to B. glumae. Additionally, other LRR receptor members in rice have been implicated in various stress responses, contributing to improved resistance against environmental stresses (Liu et al. 2014). Genes containing the LRR receptor domain are widely recognized as major contributors to pathogen resistance. Notably, several LRR receptor domain-containing genes play crucial roles in resistance to Xanthomonas oryzae pv. oryzae, which causes bacterial leaf blight. For example, Xa4 enhances resistance by strengthening the cell wall to prevent pathogen invasion, Xa10 induces programmed cell death to confer resistance (Gu et al. 2008), and Xa13 inhibits pathogen proliferation (Chu et al. 2004). Additionally, Pi-b (Chen et al. 2018), Pi-kh (Wu et al. 2013), and Pi-9 (Zhou et al. 2020), which contain LRR receptor domains, play crucial roles in signaling pathways involved in interactions with different races of Magnaporthe grisea, the causal agent of rice blast. These genes activate defense responses, thereby increasing resistance to rice blast. These findings suggest that the NLR domain functions as a multifunctional regulator capable of integrating defense systems against a wide range of pathogens.

Candidate genes screened within the Chr01_33472174–Chr01_33838140 region presented diverse functional characteristics based on the enriched GO terms in the biological process, cellular component, and molecular function categories. Notably, GO:0050790 (regulation of catalytic activity) and GO:0044092 (negative regulation of molecular function) were significantly enriched, suggesting that the associated genes may play crucial roles in regulating specific enzyme activities and signal transduction pathways (Ashburner et al. 2000). Analysis of cellular component-related factors revealed the inclusion of various genes associated with GO:0005737 (cytoplasm) and GO:0043226 (organelle), indicating their potential involvement in intracellular signal transduction and stress-related responses. From a molecular function perspective, GO:0008233 (peptidase activity) and GO:0017171 (serine hydrolase activity) were prominently enriched, suggesting their possible roles in protein degradation and immune responses under stress conditions. OsBGq1, screened within the Chr01_33472174-Chr01_33838140 region, encodes an NLR domain, suggesting its potential involvement in ROS regulation and antimicrobial activity under stress conditions. The NLR domain plays a critical role in plant immune responses by recognizing pathogens, modulating ROS scavenging, and regulating signal transduction pathways (Grant and Loake 2000). Therefore, OsBGq1, identified in this study, likely contributes to rice immune responses and adaptation to environmental stress. Future studies should focus on functional validation of OsBGq1 through the development of genome-edited plants using CRISPR/Cas9 or transgenic plants with overexpression vectors to further elucidate its role (Liu et al. 2017). Notably, among the 16 candidate genes identified within the QTL interval, OsBGq1 exhibited the highest and most consistently maintained expression level under field conditions following B. glumae inoculation. When B. glumae was applied in the field, the expression of OsBGq1 remained at a substantially higher level than that of other resistance-related candidate genes, whereas many of the other candidates showed transient induction or pronounced variability in their expression patterns. The stable and elevated expression of OsBGq1 under actual infection conditions suggests that this gene is not merely responsive to pathogen challenge but is highly likely to function effectively in conferring resistance in planta. Based on this superior expression performance under field inoculation, OsBGq1 was therefore selected as the most promising candidate gene for bacterial grain rot resistance. Nevertheless, several additional genes within the same QTL interval may also contribute to BGR resistance. These include genes involved in pathogen-responsive transcriptional regulation (AP2/ERF) (Gutterson and Reuber 2004), redox homeostasis and ROS-mediated defense (thioredoxin-like proteins and apurinic endonuclease) (Nadarajah 2020), programmed cell death (caspase-like domain proteins) (Desaki et al. 2006), and cell wall degradation (glycoside hydrolase family 17) (Yang et al. 2021). Collectively, these findings suggest that resistance to B. glumae in this genomic region is likely governed by a coordinated defense network involving pathogen recognition, immune signaling, oxidative stress regulation, and localized cell death, rather than by the action of a single gene alone. From this perspective, the application of image analysis demonstrates its effectiveness in efficiently collecting data for identifying resistance-related QTLs and assessing disease symptoms that are difficult to distinguish through visual inspection, thereby increasing the precision of resistance evaluation (Mahlein 2016). With the rapid rise in global temperatures accelerating race differentiation and increasing disease incidence, the application of image analysis for data collection offers a precise solution for identifying resistance-related QTLs. This approach provides breeders with a valuable tool for improving strategies for resistance breeding. The application of image data provided results with reliability comparable to that of results obtained under controlled conditions while enabling a more precise narrowing down of the previously identified QTL.

In this study, we propose that OsBGq1 may function as a regulatory factor contributing to resistance against BGR. However, because this QTL and candidate gene were identified using a single bi-parental population evaluated under a specific geographical and environmental condition, their effectiveness across diverse rice cultivars with different genetic backgrounds and agro-ecological zones remains to be verified. Therefore, further validation using multiple genetic backgrounds and environments will be required to assess the broader applicability of OsBGq1-associated resistance.

Moreover, the precise molecular function of OsBGq1 and its role in rice physiological responses during B. glumae infection remain unclear. Functional validation through the development of CRISPR/Cas9-based genome-edited lines or transgenic plants will be essential to elucidate the underlying molecular mechanisms and to directly determine whether manipulation of OsBGq1 enhances BGR resistance. Such validation will be a critical next step toward evaluating the practical value of OsBGq1 and its potential deployment in rice breeding programs aimed at improving disease resistance and developing cultivars resilient to future climate-related challenges.

Conclusions

The importance of breeding to mitigate the effects of rising global temperatures and the consequent increase in disease incidence has long been a topic of discussion. This study applied image analysis to assess B. glumae-induced symptoms, enabling precise quantification and data collection. As a result, a QTL associated with B. glumae resistance was successfully narrowed down to region within Chr01_33472174-Chr01_33838140 on chromosome 1. KIRIL1, KIRIL2, and KIRIL4, which Kele type of the Chr01_33472174-Chr01_33838140 region, presented increased resistance to B. glumae infection, which was associated with increased ROS degradation and phytoalexin accumulation. Within this region, OsBGq1 was identified as a potential candidate gene linked to B. glumae resistance. OsBGq1 exhibited higher expression levels in resistant rice cultivars than in susceptible cultivars, allowing clear distinction between the two groups. OsBGq1 encodes an NLR domain, a protein region also found in genes associated with resistance against bacterial leaf blight and rice blast. OsBGq1 was classified into five haplotypes on the basis of 13 SNPs, with Hap2 being an Indica-specific haplotype that included Kele. This study offers an efficient and precise breeding strategy for screening QTLs and candidate genes associated with B. glumae resistance through image data collection and analysis. Future research will focus on elucidating the molecular mechanism underlying the physiological effects of OsBGq1 through the development of transgenic and genome-edited lines.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

JRP and GL conceptualized the experiments. JRP, HSL, and JS developed the methodology and research plan. SYL, SKH and KK performed the formal analysis. MJ, JPS, YMC, and BCL conducted the experiments. JRP was responsible for writing the original draft as well as reviewing and editing the manuscript. HSP was responsible for project administration. All the authors contributed to the article and approved the submitted version.

Funding

None.

Data availability

The datasets generated during and/or analyzed during the current research are available from the corresponding authors upon reasonable request. Also, the gene sequence data for this study (OsBGq1) can be found in the National Center for Biotechnology Information (https://www.ncbi.nlm.nih.gov/) under the accession numbers, PV236018.

Declarations

Conflict of interest

The authors declare that they have no conflicts of interest.

Ethical approval

The authors declare that no ethical standards have been violated during the course of the study.