Advertisement
Browse Subject Areas
?

Click through the PLOS taxonomy to find articles in your field.

For more information about PLOS Subject Areas, click here.

  • Loading metrics

Genome-wide association analysis identifies resistance loci for bacterial blight in a diverse collection of indica rice germplasm

  • Fan Zhang ,

    Contributed equally to this work with: Fan Zhang, Zhi-Chao Wu

    Affiliation Institute of Crop Sciences/National Key Facility for Crop Gene Resources and Genetic Improvement, Chinese Academy of Agricultural Sciences, Beijing, China

  • Zhi-Chao Wu ,

    Contributed equally to this work with: Fan Zhang, Zhi-Chao Wu

    Affiliations Institute of Crop Sciences/National Key Facility for Crop Gene Resources and Genetic Improvement, Chinese Academy of Agricultural Sciences, Beijing, China, Graduate School of Chinese Academy of Agricultural Sciences, Chinese Academy of Agricultural Sciences, Beijing, China

  • Ming-Ming Wang,

    Affiliations Institute of Crop Sciences/National Key Facility for Crop Gene Resources and Genetic Improvement, Chinese Academy of Agricultural Sciences, Beijing, China, Graduate School of Chinese Academy of Agricultural Sciences, Chinese Academy of Agricultural Sciences, Beijing, China

  • Fan Zhang,

    Affiliations Institute of Crop Sciences/National Key Facility for Crop Gene Resources and Genetic Improvement, Chinese Academy of Agricultural Sciences, Beijing, China, Graduate School of Chinese Academy of Agricultural Sciences, Chinese Academy of Agricultural Sciences, Beijing, China

  • Michael Dingkuhn,

    Affiliations Crop and Environmental Sciences Division, International Rice Research Institute, Los Baños, Laguna, Philippines, Centre de Coopération Internationale en Recherche Agronomique pour le Développement (CIRAD), UMR AGAP, Montpellier, France

  • Jian-Long Xu ,

    zhouylcaas@126.com (YZ); xujianlong@caas.cn (JX)

    Affiliations Institute of Crop Sciences/National Key Facility for Crop Gene Resources and Genetic Improvement, Chinese Academy of Agricultural Sciences, Beijing, China, Shenzhen Institute of Breeding and Innovation, Chinese Academy of Agricultural Sciences, Shenzhen, Guangdong, China

  • Yong-Li Zhou ,

    zhouylcaas@126.com (YZ); xujianlong@caas.cn (JX)

    Affiliations Institute of Crop Sciences/National Key Facility for Crop Gene Resources and Genetic Improvement, Chinese Academy of Agricultural Sciences, Beijing, China, Shenzhen Institute of Breeding and Innovation, Chinese Academy of Agricultural Sciences, Shenzhen, Guangdong, China

  • Zhi-Kang Li

    Affiliations Institute of Crop Sciences/National Key Facility for Crop Gene Resources and Genetic Improvement, Chinese Academy of Agricultural Sciences, Beijing, China, Shenzhen Institute of Breeding and Innovation, Chinese Academy of Agricultural Sciences, Shenzhen, Guangdong, China

Genome-wide association analysis identifies resistance loci for bacterial blight in a diverse collection of indica rice germplasm

  • Fan Zhang, 
  • Zhi-Chao Wu, 
  • Ming-Ming Wang, 
  • Fan Zhang, 
  • Michael Dingkuhn, 
  • Jian-Long Xu, 
  • Yong-Li Zhou, 
  • Zhi-Kang Li
PLOS
x

Abstract

Bacterial blight, which is caused by Xanthomonas oryzae pv. oryzae (Xoo), is one of the most devastating rice diseases worldwide. The development and use of disease-resistant cultivars have been the most effective strategy to control bacterial blight. Identifying the genes mediating bacterial blight resistance is a prerequisite for breeding cultivars with broad-spectrum and durable resistance. We herein describe a genome-wide association study involving 172 diverse Oryza sativa ssp. indica accessions to identify loci influencing the resistance to representative strains of six Xoo races. Twelve resistance loci containing 121 significantly associated signals were identified using 317,894 single nucleotide polymorphisms, which explained 13.3–59.9% of the variability in lesion length caused by Xoo races P1, P6, and P9a. Two hotspot regions (L11 and L12) were located within or nearby two cloned R genes (xa25 and Xa26) and one fine-mapped R gene (Xa4). Our results confirmed the relatively high resolution of genome-wide association studies. Moreover, we detected novel significant associations on chromosomes 2, 3, and 6–10. Haplotype analyses of xa25, the Xa26 paralog (MRKc; LOC_Os11g47290), and a Xa4 candidate gene (LOC_11g46870) revealed differences in bacterial blight resistance among indica subgroups. These differences were responsible for the observed variations in lesion lengths resulting from infections by Xoo races P1 and P9a. Our findings may be relevant for future studies involving bacterial blight resistance gene cloning, and provide insights into the genetic basis for bacterial blight resistance in indica rice, which may be useful for knowledge-based crop improvement.

Introduction

Bacterial blight, which is caused by Xanthomonas oryzae pv. oryzae (Xoo), is one of the most devastating diseases of cultivated rice (Oryza sativa L.) in tropical and temperate regions worldwide [1]. This disease is frequently prevalent in southern China and southeast Asia, resulting in heavy rice yield losses [2, 3]. Developing and deploying resistant cultivars carrying major resistance (R) genes has been the most effective approach for managing bacterial blight [4].

Bacterial blight is characterized by a high degree of race–cultivar specificity. Several sets of isotype races/pathotypes have been identified since the 1980s in the Philippines, China, and other countries in rice cultivars differing in susceptibility to bacterial blight [49]. Based on analyses of phenotypic responses to Xoo races and molecular mapping results for identified genes, 41 R genes (i.e., 29 dominant and 12 recessive genes) conferring resistance to bacterial blight have been registered in the Oryzabase database (http://www.shigen.nig.ac.jp/rice/oryzabase/gene/list). Most of these genes are derived from Oryza sativa ssp. indica cultivars, but seven genes are from six related wild species [1012]. Additionally, some R genes or alleles have been generated by mutating cultivated rice lines [10, 13, 14], including the following nine isolated genes: Xa1, xa5, xa13, Xa21, Xa23, xa25, Xa26/Xa3, Xa27, and xa41 [1523]. Another nine genes have been fine-mapped (i.e., Xa2, Xa4, Xa7, Xa22, Xa30, Xa33, Xa38, Xa39, and Xa40) (http://www.shigen.nig.ac.jp/rice/oryzabase/gene/list). Moreover, quantitative trait loci (QTLs) for resistance to bacterial blight caused by different Xoo isolates have also been reported [24, 25].

Novel R genes and QTLs associated with bacterial blight resistance have been identified and used in breeding programs. However, the rapid loss of bacterial blight resistance in rice varieties carrying a single R gene remains a problem for breeders. This has been in part because the mechanisms mediating the pathogenesis of Xoo and the genetic basis for bacterial blight resistance in rice have not been fully characterized. To date, most of the studies on bacterial blight resistance have been based on a single resistant parent or bi-parental genetic mapping populations. Additionally, the genetic variability of bacterial blight resistance in rice accessions has yet to be addressed because only one or two parental lines have been used in previous studies. A genome-wide association study (GWAS) involving high-density single nucleotide polymorphisms (SNPs) based on next-generation sequencing may be useful for detecting genetic variants that can be directly applied to improve rice cultivars [26]. In this study, 172 global indica accessions were inoculated with representative strains of six Xoo races from China and the Philippines to evaluate their resistance reactions. We also conducted a GWAS of bacterial blight resistance based on the genotyping of 700,000 SNPs with a high-density rice array [27]. Our study objectives were to elucidate the genetic basis of bacterial blight resistance and identify loci related to bacterial blight resistance in indica rice lines, which may provide useful information for improved rice production.

Materials and methods

Plant materials and bacterial inoculations

We examined 172 indica rice accessions from 26 countries (S1 Table). To evaluate bacterial blight resistance, the seeds of all plant materials were sown in a seedling nursery, and 30-day-old seedlings were transplanted to the experimental farm at the Institute of Crop Sciences, Chinese Academy of Agricultural Sciences, Beijing, China. There were nine plants in each row (20 × 17 cm). The following six representative Xoo strains were used to artificially inoculate plants: GD1358 (race C5) and V (race GV) from China, and PXO61 (race P1), PXO340 (race P3c), PXO339 (race P9a), and PXO99 (race P6) from the Philippines. The bacterial races were grown on peptone sucrose agar medium at 30°C for 2 days, and each inoculum was prepared by suspending the bacterial mass in sterile water at a concentration of 108 cells ml−1. Five central plants for each line were inoculated with each Xoo race (with two replicates) at the tillering stage (plant age: 65 days), and four or five of the uppermost leaves of each plant were inoculated with each Xoo race using the leaf-clipping method [28]. The lesion lengths (LLs) were measured on all inoculated leaves 3 weeks post-inoculation when lesions were obvious and stable. The average LL of one accession was calculated based on 15 of the longest lesions from five individual plants (i.e., three lesions per plant) for each replicate. The average LL of two replicates for each accession was used as the phenotype for the GWAS.

Genotyping and population structure analysis

All 172 accessions were genotyped for 700,000 SNPs using a high-density rice array [27]. To avoid the influence of linked SNPs during the population structure analysis, we used the linkage disequilibrium (LD) pruning tool of PLINK [29] to obtain a subset of 45,311 independent SNPs with a minor allele frequency (MAF) > 5% and a missing data ratio (MDR) < 0.1 according to ‘indep-pairwise 50 10 0.5’. We used PHYLIP (version 3.6) [30] to construct an unrooted neighbor-joining tree with 100 bootstrap replicates. The genetic structure of the whole population was predicted with the ADMIXTURE program [31]. EIGENSOFT [32] was used to conduct principal components analysis (PCA) for estimating the number of subpopulations in the GWAS panel.

Association mapping

A total of 279,266 SNPs with a MAF > 5% and MDR < 0.2 were filtered for association analyses of the whole panel. The GWAS was completed using a linear mixed effects model to determine the associations between each SNP and the phenotype (i.e., mean lesion length caused by Xoo). We applied the efficient mixed model analysis (EMMA) feature of the EMMA eXpedited (EMMAX) software [33]. We used the Balding–Nichols matrix based on genome-wide SNP data to develop the kinship matrix, which measured the genetic similarity between individuals. The first seven principal components explaining 80% of the genetic variation were used as covariates.

Based on 1000 permutation tests, the genome-wide significance thresholds at a significance level of 0.05 were P = 3.49 × 10−7, 1.82 × 10−7, 2.58 × 10−6, 3.15 × 10−7, 3.50 × 10−7, and 1.34 × 10−7 for Xoo races P1, P3c, P6, P9a, C5, and GV, respectively. The Manhattan and quantile-quantile plots for GWAS results were created using the R package qqman [34]. We obtained independent association signals using a published method [35]. Briefly, multiple SNPs exceeding the threshold in a 5-Mb region were clustered using an LD r2 value ≥ 0.25, and SNPs with the minimum p value in a cluster were considered as the lead SNPs. The pairwise LD r2 values were calculated with PLINK [29], and the R package LDheatmap [36] was used to draw the heatmap of pairwise LD. The SNPs spanning a region from 2 kb beyond the 3′ end of a gene to the 5′ end were concatenated as the haplotype. We excluded SNPs with no functional effect according to the haplotype analysis, and only used the haplotypes shared by at least five accessions. Phenotypic variations explained by multiple lead SNPs of association loci were estimated by stepwise regression using SAS [37].

Analysis and annotation of significant signals

Synonymous and nonsynonymous SNPs and SNPs with large-effect changes were annotated based on the gene models of the annotated version of the Nipponbare reference genome (IRGSP 1.0) [38] using snpEff software (version 4.0) [39]. Enriched gene ontology terms were identified using the agriGO online tool [40]. We listed all significant SNPs located within genes and annotation information for the Nipponbare reference genome (IRGSP 1.0) [38].

Results

Population structure of rice accessions

We genotyped a diverse global collection of 172 O. sativa ssp. indica accessions (S1 Table) using a high-density rice array [27], resulting in 700,000 high-quality SNPs. A total of 45,311 independent SNPs with MAF > 5% and MDR < 0.1 were used for genetic structure analyses. The neighbor-joining tree developed using PHYLIP version 3.6 [30] with 100 bootstrap replicates and the ADMIXTURE results [31] revealed that the 172 accessions were grouped into three clusters (S1A and S1C Fig). A similar result was observed using PCA, in which most of the genetic variation in the accessions was explained by the first two principal components. When we plotted the first two components against each other, most accessions clustered in three groups with a few admixed accessions (S1B Fig). We then divided the 172 indica accessions into three subgroups, namely ind-I (78 accessions), ind-II (36 accessions), and ind-III (58 accessions) (S1 Table).

Reactions of accessions to different Xoo races

All accessions were inoculated with six Xoo races at the primary tillering stage (S1 Table). Based on the LL values calculated using the unweighted pair-group method with arithmetic mean, the six bacterial races were divided into three groups. One group consisted of race P9a, another group comprised races P1 and P6, and the third group included the remaining three races, which were more pathogenic than the races in the other groups (Fig 1A). Additionally, the accessions were clustered into three major groups (i.e., indicated in green, red, and blue in Fig 1A) based on the LL values for the six Xoo races. The accessions in the red group mainly consisted of indica-I accessions, and were equally resistant to race P1. The resistance level of each accession was based on the average LL for ten plants (two replicates) as follows: LL < 5 cm, 5 cm ≤ LL < 10 cm, 10 cm ≤ LL < 15 cm, and LL ≥ 15 cm represented resistant, moderately resistant, moderately susceptible, and susceptible levels, respectively. A considerable proportion (i.e., from 94.8% for P6 to 98.2% for P3c) of accessions were moderately susceptible or susceptible to Xoo races P6, P3c, C5, and GV, respectively (Fig 1B). However, there were similarly high proportions of accessions that were resistant and susceptible to both P1 and P9a, suggesting the R genes with large effect on P1 and P9a resistance were carried by only some accessions. The wide range of LLs observed in different rice germplasm accessions indicated there was substantial genotypic variability associated with resistance to Xoo races, especially for P9a and P1 (Fig 1C). For all Xoo races (except P9a), the average LL of subgroup indica-I was significantly smaller than that of the other two subgroups. For P9a, the average LL of subgroup indica-II was significantly higher than that of subgroups indica-I and indica-III (p < 0.001 for both; Fig 1C).

thumbnail
Fig 1. Susceptible and resistant reactions of a whole population and six subpopulations of 172 indica accessions inoculated with representative strains of six Xanthomonas oryzae pv. oryzae (Xoo) races from China and the Philippines.

(A) Hierarchical cluster of accessions and races based on lesion length (LL). (B) Number of accessions in the following reactions to six Xoo races: resistant (LL < 5 cm), moderately resistant (5 cm ≤ LL < 10 cm), moderately susceptible (10 cm ≤ LL < 15 cm), and susceptible (LL ≥ 15 cm). (C) Boxplots for LLs following infections by six Xoo races in three indica subgroups divided by population structure analysis results. Box edges represent the 0.25 and 0.75 quantiles with median values indicated by bold lines. ‘***’ refers to a significant difference in the average LLs among the indica subgroups (p < 0.001). We used the following Xoo races: C5 (strain GD1358) and GV (strain V) from China, and P1 (strain PXO61), P3c (strain PXO340), P9a (strain PXO339), and P6 (strain PXO99) from the Philippines.

https://doi.org/10.1371/journal.pone.0174598.g001

Genome-wide association analysis for Xoo resistance in indica rice

We used association mapping to identify genome-wide associated loci underlying the resistance to six Xoo races. A total of 279,266 SNPs with a MAF > 0.05 and MDR < 0.2 in the whole population and subpopulation (i.e., whole population without the admixed accessions highlighted in S1B Fig) were used for the association analyses with a linear mixed effects model in the EMMAX program [33]. We constructed the empirical kinship (K) matrix using the BN matrix with 211,943 SNPs (MAF > 0.01 and MDR < 0.05), and covariates (Q) were generated based on the PCA of the population structure. The similarities in the Manhattan plots and quantile-quantile plots for the whole population (Fig 2) and the subpopulation (S2 Fig) suggested the Q + K model effectively controlled the P-value inflation and the influence of the population structure on our GWAS. Furthermore, we focused on the GWAS results for the whole population, and detected 121 significantly associated SNPs on chromosomes 2, 3, and 6–12. There were 51, 68, and two SNPs associated with resistance to Xoo races P1, P9a, and P6 (Fig 2 and S2 Table), explaining 55.3%, 59.9%, and 13.3% of the variation in LL, respectively. Significantly associated SNPs were not identified for bacterial races P3c, C5, and GV. The 121 detected SNPs were distributed in 96 annotated genes as follows: 60 in intergenic regions, 14 in introns, one in a promoter, 13 in missense variants, 27 in synonymous variants, two in 5′ untranslated regions, two in 3′ untranslated regions, and two in stop codons (S2 Table). Of the 121 significant signals, only SNP rs11_ 28436056 was detected for both P1 and P6. This SNP was located in a Xa26 family gene (LOC_Os11g47290), which encodes a receptor kinase. Gene ontology analysis indicated that these 96 genes were significantly enriched for defense responses, programmed cell death biological processes, and purine ribonucleotide-binding–related functions (S3 Table).

thumbnail
Fig 2. Manhattan and quantile-quantile plots for a genome-wide association study of bacterial blight resistance in indica rice.

(A, G) P1 (strain PXO61). (B, H) P6 (strain PXO99). (C, I) C5 (strain GD1358). (D, J) P3c (strain PXO340). (E, K) P9a (strain PXO339). (F, L) GV (strain V). The strength of the associations for the lesion lengths caused by six Xanthomonas oryzae pv. oryzae (Xoo) strains is indicated as the negative logarithm of the p value for the linear mixed effects model. Based on 1000 permutation tests, the genome-wide significance thresholds (i.e., horizontal red lines in all Manhattan plots) at a significance level of 0.05 were P = 3.49 × 10−7, 1.82 × 10−7, 2.58 × 10−6, 3.15 × 10−7, 3.50 × 10−7, and 1.34 × 10−7 for P1, P3c, P6, P9a, C5, and GV, respectively.

https://doi.org/10.1371/journal.pone.0174598.g002

Regions strongly associated with bacterial blight resistance

The physical distance between neighboring SNPs was 1–131,763 bp (average: 1349 bp) in our GWAS panel. Thus, we combined adjacent significantly associated SNPs within an LD block (i.e., multiple SNPs exceeding the significance threshold in a 5-Mb region were clustered using an LD r2 value ≥ 0.25) as a resistance region. Additionally, the SNPs with the lowest p values in the LD block were considered the lead SNPs. We detected 12 strongly associated regions containing 120 significant SNPs (including one common SNP for P1 and P6) for three of six Xoo races (Table 1). Among these, one hotspot region was associated with resistance to P9a, and spanned an approximately 1.0-Mb interval (i.e., 16,502,066–17,531,046 bp) on chromosome 12, which included 58 (85.3%) of 68 significantly associated SNPs (Fig 3A). Nine significant SNPs (lead SNP rs12_17304839, P = 1.8 × 10−12) in this region overlapped with xa25 (LOC_Os12g29220) (S2 Table), which is a recessive R gene that is responsible for race-specific resistance to P9a in rice [17]. Another hotspot region (i.e., L11), which included 48 significant SNPs associated with resistance to P1, was located at the end of chromosome 11. We determined that 43 (89.6%) of the 48 SNPs in L11 were located in an approximately 1.45-Mb interval (i.e., 27,252,984–28,704,769 bp) (Fig 4A). The corresponding region contained 33 genes that encoded proteins mainly involved in defense responses (e.g., NB-ARC domain-containing protein and protein kinase) according to the Nipponbare reference genome. This region carried the well-known R genes Xa4 and Xa26 [16, 41]. The lead SNP (i.e., rs11_28142810, P = 2.2 × 10−15) of region L11 was located in the coding sequence (CDS) of LOC_Os11g46870, which encodes a protein kinase. This location corresponded to the Xa4 gene region based on the physical position of the BAC clone 3H8 that carries Xa4 [41]. The second leading SNP (i.e., rs11_28437434, P = 1.3 × 10−14) of region L11 was located in the CDS of LOC_Os11g47290 (i.e., encoding a receptor kinase), which is the Xa26 paralog MRKc [42] (S2 Table).

thumbnail
Fig 3. Hotspot region for the resistance to Xanthomonas oryzae pv. oryzae race P9a and haplotype analysis of the peak associated with the gene on chromosome 12.

(A) Local Manhattan plot (top) (16.5–17.5 Mb) and linkage disequilibrium heatmap (bottom) (17.2–17.5 Mb) surrounding the hotspot region on chromosome 12. The arrow indicates the position of the peak single nucleotide polymorphism (SNP) located in xa25 (LOC_Os12g29220). Dashed lines indicate the xa25 region. (B) Gene structure and haplotype analysis of xa25 in 140 accessions based on five significant SNPs in xa25. Haplotypes with fewer than five accessions are not shown. (C) Lesion lengths caused by P9a infections of accessions in three haplotypes of xa25 in different indica subgroups. Box edges represent the 0.25 and 0.75 quantiles with median values indicated by bold lines. Whiskers extend to data no more than 1.5-times the interquartile range, and the remaining data are represented by dots. ‘***’ refers to a significant difference based on Duncan’s multiple comparison tests (p < 0.001).

https://doi.org/10.1371/journal.pone.0174598.g003

thumbnail
Fig 4. Hotspot region for the resistance to Xanthomonas oryzae pv. oryzae race P1 and haplotype analysis of the peak associated with the gene on chromosome 11.

(A) Local Manhattan plot (top) (23–29 Mb) and linkage disequilibrium heatmap (bottom) (28.0–28.8 Mb) surrounding the hotspot region on chromosome 11. Red arrows and points indicate the positions of the peak single nucleotide polymorphisms located in the Xa4 candidate gene (i.e., LOC_Os11g46870) and Xa26 paralog (i.e., LOC_Os11g47290), respectively. Dashed lines indicate the xa25 region. (B) Gene structure and haplotype analysis of the Xa4 candidate gene (i.e., LOC_Os11g46870). (C) Gene structure and haplotype analysis of the Xa26 paralog (i.e., LOC_Os11g47290). Lesion lengths caused by P1 infections of accessions in three haplotypes of LOC_Os11g46870 (D) and LOC_Os11g47290 (E) in different indica subgroups. ‘***’ refers to a significant difference based on Duncan’s multiple comparison tests (p < 0.001).

https://doi.org/10.1371/journal.pone.0174598.g004

thumbnail
Table 1. Twelve regions with signals significantly associated with bacterial blight resistance based on a genome-wide association study involving indica rice.

https://doi.org/10.1371/journal.pone.0174598.t001

The other 10 regions (i.e., L1–10) on chromosomes 2, 3, and 6–10 were not associated with previously characterized bacterial blight resistance loci (Table 1, Fig 2 and S2 Table), and represented potentially novel bacterial blight resistance loci. Interestingly, the lead SNPs of six regions (i.e., L2–4 and L8–10) were located in the CDS or intragenic region of genes encoding retrotransposon proteins (i.e., LOC_Os03g33000, LOC_Os03g33010, LOC_Os03g45050, LOC_Os03g45070, LOC_Os06g28980, LOC_Os09g07130, LOC_Os09g16860, and LOC_Os10g19064) (S2 Table and Table 1). Another distinct peak comprising four SNPs (lead SNP rs8_3317231, P = 3.3 × 10−12) in region L6 located on chromosome 8 was significantly associated with resistance to P9a. This region contained the EARLY FLOWERING 7 gene (LOC_Os08g06070) (Table 1, Fig 2 and S2 Table). Similarly, we detected one distinct P1 association locus (i.e., L8) on chromosome 8 with a lead SNP (rs8_9671192, P = 4.2 × 10−8) in the CDS region of the NBS-LRR (Nucleotide binding site-leucine-rich repeats) disease resistance protein gene (LOC_Os08g15880). However, we did not identify bacterial blight resistance loci in this region.

Haplotype analysis for the known and putative novel resistance genes in the indica panel

Haplotype analysis for the P9a-specific resistance gene xa25.

The xa25 haplotypes were built based on two SNPs in a 2-kb region upstream of the LOC_Os12g29220 promoter, two significantly associated SNPs in the intron region, and one SNP in the 3′ untranslated region (Fig 3B). We detected three haplotypes shared by at least five accessions in 140 of 172 accessions (Fig 3B and S1 Table). Additionally, 45.5% of the variation in LL caused by P9a was explained by xa25 haplotypes. Multiple comparison tests of the lesions induced by P9a revealed that the average LL of Hap3 was significantly shorter than those of Hap1 and Hap2 (Fig 3B and 3C). The xa25 gene is a recessive R gene that provides rice with race-specific resistance to P9a [17]. It encodes a sucrose transporter (OsSWEET13). Additionally, variations in the OsSWEET13 promoter sequence results in cryptic recessive resistance to PthXo2-dependent Xoo in japonica rice [43]. PthXo2-containing strains induce OsSWEET13 expression in indica rice line IR24 because of the presence of an uncharacterized effector-binding site absent in the alleles of the japonica rice varieties Nipponbare and Kitaake [43]. We detected Hap3 in the Nipponbare reference genome. Among the 140 accessions with detected haplotypes, 42 (87.5%) of 48 accessions with resistant and moderately resistant reactions to P9a belonged to the Hap3 group (S1 Table). Only one of the 42 Hap1 accessions (i.e., accession ZAO SHOU 691–11) and one of the 27 Hap2 accessions (i.e., accession H 15-23-DA) were resistant to P9a (S1 Table). Additionally, 92.6% of Hap2 accessions belonged to the indica-I subgroup, while 64.3% of Hap1 accessions belonged to the indica-II subgroup. Meanwhile, 45.1% and 47.9% of the Hap3 accessions belonged to the indica-I and indica-III subgroups, respectively. These results explained the phenotypic variability among the three indica subgroups in response to Xoo race P9a (Fig 1C).

Haplotype analysis for the Xa26 paralog MRKc.

The R gene Xa26 provides resistance to Xoo race P1 [16]. The top hit for Xa26 following a blastp search of the Nipponbare reference genome (IRGSP 1.0) was LOC_Os11g47210, which encodes a receptor kinase-like protein. We did not detect any significant association signals or a resistant haplotype associated with LOC_Os11g47210 (data not shown). However, we identified one strong association signal for P1 located at around 28,437,434 bp (i.e., rs11_28437434, P = 1.3 × 10−14) on chromosome 11. There were also five significant SNPs for P1 and one significant SNP for P6 located in the CDS region of LOC_Os11g47290 (encoding a receptor kinase), which is a Xa26 paralog (i.e., MRKc) [42] (Fig 4A and S2 Table). We mainly analyzed the LOC_Os11g47290 haplotype, and identified three major haplotypes shared by at least five accessions in 132 accessions based on three SNPs with strong association signals in the gene coding region (Fig 4C and S1 Table). These haplotypes explained 45.3% and 32.2% of the phenotypic variations in the lesions induced by P1 and P6, respectively. The lesions induced by P1 in Hap1 accessions were significantly shorter than those of the Hap2 and Hap3 accessions (p < 0.001 for both; Fig 4C and 4E). Additionally, 29 (72.5%) of 40 accessions with resistant reactions to P1 belonged to the Hap1 group (Fig 4C and S1 Table). Only two of the 20 Hap2 accessions (i.e., INIAP 415 and HAO HOM) and one of the 71 Hap3 accessions (i.e., TAICHUNG NATIVE 1) were resistant to P1 (S1 Table). Meanwhile, 52.5%, 86.2%, and 76.2% of the Hap1, Hap2, and Hap3 accessions belonged to indica-I, indica-III, and indica-III subgroups, respectively. These results explained the phenotypic variability among the three indica subgroups in response to Xoo race P1 (Fig 1C).

Haplotype analysis for a Xa4 candidate gene.

The Xa4 gene has been reported to provide resistance against Xoo race P1 [1]. A comparison between the physical positions of BAC clones 3H8 and X4-88 carrying Xa4 [41] and the significantly associated SNPs in this study suggested a protein kinase gene (i.e., LOC_Os11g46870) might be a Xa4 candidate, in which the strongest association (i.e., rs11_28142810, P = 2.2 × 10−15) for P1 resistance was detected in the coding region (Fig 2 and S2 Table). Three major haplotypes shared by at least five accessions were detected in 146 accessions based on two SNPs in the coding region of LOC_Os11g46870 (Fig 4A). We determined that 54.5% and 30.5% of the phenotypic variability in lesions induced by P1 and P6 were explained by LOC_Os11g46870 haplotypes, respectively. A comparison of the LLs for the three haplotypes following an infection by Xoo race P1 revealed that Hap1 accessions had significantly shorter LLs than the other two haplotypes (p < 0.001 for both; Fig 4B and 4D). Furthermore, 33 (82.5%) of 40 accessions with a resistant reaction to P1 infection belonged to the Hap1 group, and 26 accessions belonged to the indica-I subgroup (Fig 4B and S1 Table). In contrast, only one of the 88 Hap2 accessions (i.e., INIAP 415) and 10 of the Hap3 accessions were resistant to P1 (S1 Table). Additionally, 78.2% and 70.0% of the Hap1 and Hap3 accessions belonged to the indica-I and indica-III subgroups, respectively, whereas the Hap2 accessions were relatively evenly distributed among the indica-I, indica-II, and indica-III subgroups. These results explained the phenotypic variations observed among the three indica subgroups in response to Xoo race P1 (Fig 1C).

Discussion

Distinct differences in bacterial blight resistance among indica rice subgroups

We conducted a GWAS involving 172 global indica accessions infected by representative strains of six Xoo races from China and the Philippines to analyze the genetic basis of bacterial blight resistance in rice, and detected two hotspot regions associated with Xoo resistance. One hotspot region consisted of an approximately 1.45-Mb interval (i.e., 27,252,984–28,704,769 bp) on chromosome 11. This region included 89.6% (i.e., 43 out of the 48) of the significant SNPs associated with resistance to P1. The other identified hotspot region spanned an approximately 1.0-Mb interval (i.e., 16,502,066–17,531,046 bp) on chromosome 12. This region contained 85.3% (i.e., 58 out of 68) of the significant SNPs associated with resistance to P9a (Fig 2 and S2 Table). Our results suggest that chromosomes 11 and 12 were important for the evolution of rice disease resistance. At the genome level, loci associated with bacterial blight resistance exhibited race specificity, similar to some reported R genes and QTLs [1, 24, 25]. In this study, only two SNPs with weak but significant signals associated to resistance for P6 were detected (Table 1 and Fig 2), no any SNPs significantly associated resistance for Xoo races P3c, C5, and GV was identified in the tested rice accessions due to limited phenotypic variation (Fig 1), suggesting lack of R gene/QTL with large effect on resistance for these races in the association panel. It is generally accepted that GWAS has low power to find associations for rare alleles [44]. To clarify whether low allele frequency resulted in false positive associations in GWAS, further studies will be needed to ascertain the functions of resistance candidate genes by expression profiling or genetic transformation.

Our GWAS and haplotype results provided some interesting information on the differentiation and accumulation of resistance loci for Xoo in rice. Bacterial blight is prevalent in both tropical and temperate areas, but outbreaks and epidemics of this disease frequently occur in the tropics because of a favorable climate for the survival, propagation, and infection of Xoo [1, 45]. During the co-evolution of plant hosts and pathogens, the strong selection pressure resulting from diverse pathogen populations might have induced indica rice lines to generate novel disease resistance specificities or evolve novel effectors to overcome the effects of genes associated with disease susceptibility. Considerable selection pressures from various Xoo races have contributed to the diversification of disease resistance loci in indica. This has resulted in the differentiation of bacterial blight resistance among the indica subgroups. In this study, 45.1% and 47.9% of the xa25 haplotype accessions resistant to P9a belonged to indica-I and indica-III subgroups, respectively (Fig 3B and S1 Table). However, 92.6% of the susceptible xa25 haplotype accessions (i.e., Hap2) belonged to the indica-I subgroup, in which 13 accessions resistant to P1 carried LOC_Os11g46870 or LOC_Os11g47290 resistance haplotypes (Fig 4 and S1 Table). Nipponbare reference genome indicated that this japonica rice variety carries xa25 [43] as well as the associated disease resistance haplotype. These results suggested that the effects of the xa25 alleles in indica accessions were overcome by P9a during the co-evolution between the host and pathogen.

Breeding for bacterial blight resistance has been one of the most important objectives among rice breeders in Asia since the 1970s. Rice germplasm carrying Xa4 and Xa3/Xa26 exhibit a broad resistance to Chinese Xoo races, and have been obtained from the International Rice Research Institute and used as the bacterial blight resistance donor in rice breeding programs [3]. The results in study suggest that breeding for bacterial blight resistance involved the end of the long arm of chromosome 11 (Fig 2 and Table 1), which carries several bacterial blight resistance genes (e.g., Xa4, and Xa26). The haplotype analysis indicated that accessions with P1 resistance haplotypes related to the Xa4 candidate gene (i.e., LOC_Os11g46870), Xa26 paralog gene (i.e., LOC_Os11g47290), and a P9a-specific susceptible haplotype of the xa25 gene (i.e., LOC_Os12g29220) were associated with differential disease susceptibilities among the indica subgroups (Fig 3 and Fig 4). The P1 resistance haplotypes associated with the Xa4 candidate gene (i.e., LOC_Os11g46870) and Xa26 paralog gene (i.e., LOC_Os11g47290) were more frequently detected in the indica-I subgroup than in the indica-II and indica-III subgroups (Fig 4B and 4C). This implies that conventional breeding has applied a strong selection pressure for certain R genes, and the breeding process for disease resistance of certain varieties has contributed to the accumulation of BB resistance loci.

Putative roles of transposable elements affecting bacterial blight resistance in rice

We identified several retrotransposons and transposons among the candidate genes anchored by the SNPs associated with bacterial blight resistance, including significant SNPs (S2 Table). Transposable elements have been suggested to contribute to the evolution of genes by providing cis-regulatory elements leading to changes in expression patterns [4649]. Stress-induced changes to retrotransposons may play a role in generating host genetic plasticity in response to environmental stresses [5052]. In rice, Xa21 encodes a receptor-like kinase and confers resistance to Xoo. This gene is a member of a multi-gene family in which 17 transposon-like elements have been identified in the 5′ and 3′ flanking regions and introns [49, 53, 54]. Researchers have attempted to diversify the rice bacterial blight resistance genes. Efforts have involved point mutations [55], deletion and duplication of intragenic DNA repeats encoding blocks of leucine-rich elements subject to adaptive selection [56], intragenic and extragenic recombinations [57], and insertion of transposable elements [49]. Our results suggest that transposable elements have highly enriched the diversity of bacterial blight resistance genes during evolution. Additionally, a large-scale attempt to identify and annotate loci provides useful insights into the genetic control of indica rice traits.

Candidates for fine-mapped genes and novel loci

We identified 12 bacterial blight resistance loci containing 121 significantly associated signals using 317,894 SNPs. Two hotspot regions (i.e., L11 and L12) were located within or nearby cloned R genes (i.e., xa25 and Xa26) and a fine-mapped R gene (i.e., Xa4). These observations revealed the relatively high resolution of a GWAS involving a relatively large population, which increased our ability to investigate genetic diversity and a high-density SNP map. Additionally, our haplotype analysis results suggest that the Xa4 candidate gene (i.e., LOC_Os11g46870) and the Xa26 paralog (i.e., LOC_Os11g47290) likely confer resistance to P1. Moreover, two novel loci on chromosomes 8 [i.e., L6 (rs8_3317231, P = 3.3 × 10−12) and L7 (rs8_9671192, P = 4.2 × 10−8)] were significantly associated with resistance to P9a and P1, respectively. These results provide valuable information for future studies involving bacterial blight resistance gene cloning. Moreover, the identified diverse rice accessions carrying fine-mapped and novel loci from nearly 20 countries will provide more available donors in breeding programs aiming at developing bacterial blight resistance in different rice growing regions.

Conclusions

Our study provides new insights into the genetic basis of the evolution of bacterial blight resistance in rice. The findings reported herein may be useful for knowledge-based crop improvement. Future research will focus on validating the effects of these candidate genes and their functional variants. We will use genetic transformations and DNA insertion mutant screens to verify that these genes confer bacterial blight resistance to rice.

Supporting information

S1 Fig. Population structure of the 172 indica rice accessions.

(A) Neighbor-joining tree of 172 accessions. (B) Principal component analysis plots for the first two components of 172 accessions. (C) Distribution of the estimated subpopulation components for each accession as determined by ADMIXTURE.

https://doi.org/10.1371/journal.pone.0174598.s001

(PDF)

S2 Fig. Manhattan and quantile-quantile plots of a genome-wide association study of bacterial blight resistance in indica rice subpopulations without admixed accessions based on principal component analysis.

(A, G) P1 (strain PXO61). (B, H) P6 (strain PXO99). (C, I) C5 (strain GD1358). (D, J) P3c (strain PXO340). (E, K) P9a (strain PXO339). (F, L) GV (strain V). The strength of the associations for the lesion lengths caused by representative strains of six Xanthomonas oryzae pv. oryzae (Xoo) races is indicated as the negative logarithm of the p value for the linear mixed effects model.

https://doi.org/10.1371/journal.pone.0174598.s002

(PDF)

S1 Table. Information regarding the 172 analyzed indica accessions.

https://doi.org/10.1371/journal.pone.0174598.s003

(XLSX)

S2 Table. All 121 significant association signals for three Xanthomonas oryzae pv. oryzae races.

https://doi.org/10.1371/journal.pone.0174598.s004

(XLSX)

S3 Table. Gene ontology enrichment results for the annotated genes carrying significant single nucleotide polymorphisms associated with bacterial blight resistance based on a genome-wide association study.

https://doi.org/10.1371/journal.pone.0174598.s005

(XLSX)

Author Contributions

  1. Conceptualization: YZ JX.
  2. Formal analysis: FZ1 ZW.
  3. Funding acquisition: YZ ZL.
  4. Investigation: MW FZ1,5 MD.
  5. Project administration: YZ JX.
  6. Resources: MD.
  7. Software: ZW FZ1.
  8. Supervision: YZ JX.
  9. Visualization: FZ1 ZW.
  10. Writing – original draft: FZ1 YZ.
  11. Writing – review & editing: YZ ZL.

References

  1. 1. Nino-Liu DO, Ronald PC, Bogdanove AJ. Xanthomonas oryzae pathovars: model pathogens of a model crop. Mol Plant Pathol. 2006;7(5):303–24. pmid:20507449
  2. 2. Zhang Q. Genetic evaluation and utilization of resistance to rice bacterial blight in China. Scientia Agricultura Sinica. 1991;24(2):26–36.
  3. 3. Khush GS, Mackill DJ, Sidhu GS. Breeding rice for resistance to bacterial blight. In: Banta SJ, editor. Bacterial Blight of Rice. Los Banos, Philippines: International Rice Research Institute; 1989. p. 207–17.
  4. 4. Ogawa T. Methods and strategy for monitoring race distribution and identification of resistance genes to bacterial leaf blight (Xanthomonas campetris pv. oryzae) in rice. JARQ-Japan Agricultural Research Quarterly. 1993;27:71–80.
  5. 5. Mew TW. Current status and future prospects of research on bacterial blight of rice. Annual Review of Phytopathology. 1987;25:359–82.
  6. 6. Fang CD, Xu ZG, Guo CJ, Yin SZ, Xu SZ, Xu XM, et al. Studies on pathotypes of Xanthomonas campestris pv. oryzae in China. Acta Phytopathologica Sinica. 1990;20(2):81–8.
  7. 7. Noda T, Yamamoto T, Kaku H, Horino O. Geographical Distribution of Pathogenic Races of Xanthomonas oryzae pv. oryzae in Japan in 1991 and 1993. Japanese Journal of Phytopathology. 1996;62(6):549–53.
  8. 8. Adhikari TB, Mew TW, Leach JE. Genotypic and Pathotypic Diversity in Xanthomonas oryzae pv. oryzae in Nepal. Phytopathology. 1999;89(8):687–94. pmid:18944682
  9. 9. Liu H, Yang W, Hu B, Liu F. Virulence Analysis and Race Classification of Xanthomonas oryzae pv. oryzae in China. Journal of Phytopathology. 2007;155(3):129–35.
  10. 10. Lee KS, Rasabandith S, Angeles ER, Khush GS. Inheritance of resistance to bacterial blight in 21 cultivars of rice. Phytopathology. 2003;93(2):147–52. pmid:18943128
  11. 11. Cheema KK, Grewal NK, Vikal Y, Sharma R, Lore JS, Das A, et al. A novel bacterial blight resistance gene from Oryza nivara mapped to 38 kb region on chromosome 4L and transferred to Oryza sativa L. Genet Res (Camb). 2008;90(5):397–407.
  12. 12. Brar DS, Khush GS. Alien introgression in rice. Plant Mol Biol. 1997;35(1–2):35–47. pmid:9291958
  13. 13. Gao DY, Xu ZG, Chen ZY, Sun LH, Sun QM, Lu F, et al. [Identification of a resistance gene to bacterial blight (Xanthomonas oryzae pv. oryzae) in a somaclonal mutant HX-3 of indica rice]. Yi Chuan Xue Bao. 2002;29(2):138–43. pmid:11901997
  14. 14. Nakai H, Nakamura K, Kuwahara S, Saito M. Genetic studies of an induced rice mutant resistant to multiple races of bacterial leaf blight. Rice Genetics Newsletter. 1988;5:101–3.
  15. 15. Song WY, Wang GL, Chen LL, Kim HS, Pi LY, Holsten T, et al. A receptor kinase-like protein encoded by the rice disease resistance gene, Xa21. Science. 1995;270(5243):1804–6. pmid:8525370
  16. 16. Sun X, Cao Y, Yang Z, Xu C, Li X, Wang S, et al. Xa26, a gene conferring resistance to Xanthomonas oryzae pv. oryzae in rice, encodes an LRR receptor kinase-like protein. Plant J. 2004;37(4):517–27. pmid:14756760
  17. 17. Liu Q, Yuan M, Zhou Y, Li X, Xiao J, Wang S. A paralog of the MtN3/saliva family recessively confers race-specific resistance to Xanthomonas oryzae in rice. Plant Cell Environ. 2011;34(11):1958–69. pmid:21726237
  18. 18. Yoshimura S, Yamanouchi U, Katayose Y, Toki S, Wang ZX, Kono I, et al. Expression of Xa1, a bacterial blight-resistance gene in rice, is induced by bacterial inoculation. Proc Natl Acad Sci U S A. 1998;95(4):1663–8. PubMed Central PMCID: PMCPMC19140. pmid:9465073
  19. 19. Iyer AS, McCouch SR. The rice bacterial blight resistance gene xa5 encodes a novel form of disease resistance. Mol Plant Microbe Interact. 2004;17(12):1348–54. pmid:15597740
  20. 20. Chu Z, Yuan M, Yao J, Ge X, Yuan B, Xu C, et al. Promoter mutations of an essential gene for pollen development result in disease resistance in rice. Genes & development. 2006;20(10):1250–5. PubMed Central PMCID: PMCPMC1472899.
  21. 21. Wang C, Zhang X, Fan Y, Gao Y, Zhu Q, Zheng C, et al. XA23 is an executor R protein and confers broad-spectrum disease resistance in rice. Mol Plant. 2015;8(2):290–302. pmid:25616388
  22. 22. Gu K, Yang B, Tian D, Wu L, Wang D, Sreekala C, et al. R gene expression induced by a type-III effector triggers disease resistance in rice. Nature. 2005;435(7045):1122–5. pmid:15973413
  23. 23. Hutin M, Sabot F, Ghesquiere A, Koebnik R, Szurek B. A knowledge-based molecular screen uncovers a broad-spectrum OsSWEET14 resistance allele to bacterial blight from wild rice. Plant J. 2015;84(4):694–703. pmid:26426417
  24. 24. Zhang F, Xie XW, Xu MR, Wang WS, Xu JL, Zhou YL, et al. Detecting major QTL associated with resistance to bacterial blight using a set of rice reciprocal introgression lines with high density SNP markers. Plant Breeding. 2015;134:286–92.
  25. 25. Li ZK, Arif M, Zhong DB, Fu BY, Xu JL, Domingo-Rey J, et al. Complex genetic networks underlying the defensive system of rice (Oryza sativa L.) to Xanthomonas oryzae pv. oryzae. Proc Natl Acad Sci U S A. 2006;103(21):7994–9. PubMed Central PMCID: PMCPMC1461406. pmid:16702555
  26. 26. Huang X, Zhao Y, Wei X, Li C, Wang A, Zhao Q, et al. Genome-wide association study of flowering time and grain yield traits in a worldwide collection of rice germplasm. Nat Genet. 2012;44(1):32–9.
  27. 27. McCouch SR, Wright MH, Tung CW, Maron LG, McNally KL, Fitzgerald M, et al. Open access resources for genome-wide association mapping in rice. Nat Commun. 2016;7:10532. PubMed Central PMCID: PMCPMC4742900. pmid:26842267
  28. 28. Kauffman HE, Reddy APK, Hsieh SPY, Merca SD. A improved technique for evaluation of resistance of rice varieties to Xanthomonas oryzea. Plant Dis Rep. 1973;57:537–41.
  29. 29. Purcell S, Neale B, Todd-Brown K, Thomas L, Ferreira MA, Bender D, et al. PLINK: a tool set for whole-genome association and population-based linkage analyses. Am J Hum Genet. 2007;81(3):559–75. PubMed Central PMCID: PMCPMC1950838. pmid:17701901
  30. 30. Felsenstein J. PHYLIP—Phylogeny Inference Package (Version 3.2). Cladistics. 1989;5:164–6.
  31. 31. Alexander DH, Novembre J, Lange K. Fast model-based estimation of ancestry in unrelated individuals. Genome Res. 2009;19(9):1655–64. PubMed Central PMCID: PMCPMC2752134. pmid:19648217
  32. 32. Patterson N, Price AL, Reich D. Population structure and eigenanalysis. PLoS Genet. 2006;2(12):e190. PubMed Central PMCID: PMCPMC1713260. pmid:17194218
  33. 33. Kang HM, Sul JH, Service SK, Zaitlen NA, Kong SY, Freimer NB, et al. Variance component model to account for sample structure in genome-wide association studies. Nat Genet. 2010;42(4):348–54. PubMed Central PMCID: PMCPMC3092069. pmid:20208533
  34. 34. Turner SD. qqman: an R package for visualizing GWAS results using Q-Q and manhattan plots. bioRxiv. 2014.
  35. 35. Wang Q, Xie W, Xing H, Yan J, Meng X, Li X, et al. Genetic Architecture of Natural Variation in Rice Chlorophyll Content Revealed by a Genome-Wide Association Study. Mol Plant. 2015;8(6):946–57. pmid:25747843
  36. 36. Shin J-H, Blay S, McNeney B, Graham J. LDheatmap: An R Function for Graphical Display of Pairwise Linkage Disequilibria Between Single Nucleotide Polymorphisms. 2006. 2006;16(Code Snippet 3):9. Epub 2006-04-26.
  37. 37. SAS Institute Inc. SAS/STAT 9.1 User' s Guide. Cary: SAS Institute Inc; 2004.
  38. 38. Kawahara Y, de la Bastide M, Hamilton JP, Kanamori H, McCombie WR, Ouyang S, et al. Improvement of the Oryza sativa Nipponbare reference genome using next generation sequence and optical map data. Rice (N Y). 2013;6(1):4.
  39. 39. Cingolani P, Platts A, Wang le L, Coon M, Nguyen T, Wang L, et al. A program for annotating and predicting the effects of single nucleotide polymorphisms, SnpEff: SNPs in the genome of Drosophila melanogaster strain w1118; iso-2; iso-3. Fly (Austin). 2012;6(2):80–92. PubMed Central PMCID: PMCPMC3679285.
  40. 40. Du Z, Zhou X, Ling Y, Zhang Z, Su Z. agriGO: a GO analysis toolkit for the agricultural community. Nucleic Acids Res. 2010;38(Web Server issue):W64–70. PubMed Central PMCID: PMCPMC2896167. pmid:20435677
  41. 41. Sun X, Yang Z, Wang S, Zhang Q. Identification of a 47-kb DNA fragment containing Xa4, a locus for bacterial blight resistance in rice. Theor Appl Genet. 2003;106(4):683–7. pmid:12595998
  42. 42. Li H, Li X, Xiao J, Wing RA, Wang S. Ortholog Alleles at Xa3/Xa26 Locus Confer Conserved Race-Specific Resistance against Xanthomonas oryzae in Rice. Molecular Plant. 2012;5:281–90. pmid:21930802
  43. 43. Zhou J, Peng Z, Long J, Sosso D, Liu B, Eom JS, et al. Gene targeting by the TAL effector PthXo2 reveals cryptic resistance gene for bacterial blight of rice. Plant J. 2015;82(4):632–43. pmid:25824104
  44. 44. Han B, Huang X. Sequencing-based genome-wide association study in rice. Curr Opin Plant Biol. 2013;16(2):133–8. pmid:23562592
  45. 45. Mew TW, Alvarez AM, Leach JE, Swings J. Focus on bacterial blight of rice. Plant Disease. 1993;77:5–12.
  46. 46. Woodrow P, Pontecorvo G, Ciarmiello LF, Fuggi A, Carillo P. Ttd1a promoter is involved in DNA-protein binding by salt and light stresses. Mol Biol Rep. 2011;38(6):3787–94. pmid:21104438
  47. 47. Johal GS, Briggs SP. Reductase activity encoded by the HM1 disease resistance gene in maize. Science. 1992;258(5084):985–7. pmid:1359642
  48. 48. Shalev G, Levy AA. The maize transposable element Ac induces recombination between the donor site and an homologous ectopic sequence. Genetics. 1997;146(3):1143–51. PubMed Central PMCID: PMCPMC1208042. pmid:9215915
  49. 49. Song WY, Pi LY, Bureau TE, Ronald PC. Identification and characterization of 14 transposon-like elements in the noncoding regions of members of the Xa21 family of disease resistance genes in rice. Mol Gen Genet. 1998;258(5):449–56. pmid:9669326
  50. 50. McClintock B. The significance of responses of the genome to challenge. Science. 1984;226(4676):792–801. pmid:15739260
  51. 51. Pouteau S, Grandbastien M, Boccara M. Microbial elicitors of plant defence responses activate transcription of a retrotransposon The Plant Journal. 1994;5(4):535–42.
  52. 52. Wessler SR, Bureau TE, White SE. LTR-retrotransposons and MITEs: important players in the evolution of plant genomes. Curr Opin Genet Dev. 1995;5(6):814–21. pmid:8745082
  53. 53. Bureau TE, Ronald PC, Wessler SR. A computer-based systematic survey reveals the predominance of small inverted-repeat elements in wild-type rice genes. Proc Natl Acad Sci U S A. 1996;93(16):8524–9. PubMed Central PMCID: PMCPMC38705. pmid:8710903
  54. 54. Song WY, Pi LY, Wang GL, Gardner J, Holsten T, Ronald PC. Evolution of the rice Xa21 disease resistance gene family. Plant Cell. 1997;9(8):1279–87. PubMed Central PMCID: PMCPMC156997. pmid:9286106
  55. 55. Chen S, Liu X, Zeng L, Ouyang D, Yang J, Zhu X. Genetic analysis and molecular mapping of a novel recessive gene xa34(t) for resistance against Xanthomonas oryzae pv. oryzae. Theor Appl Genet. 2011;122(7):1331–8. pmid:21274511
  56. 56. Wang GL, Ruan DL, Song WY, Sideris S, Chen L, Pi LY, et al. Xa21D encodes a receptor-like molecule with a leucine-rich repeat domain that determines race-specific recognition and is subject to adaptive evolution. Plant Cell. 1998;10(5):765–79. PubMed Central PMCID: PMCPMC144027. pmid:9596635
  57. 57. Zhang F, Zhuo DL, Zhang F, Huang LY, Wang WS, Xu JL, et al. Xa39, a novel dominant gene conferring broad-spectrum resistance to Xanthomonas oryzae pv. oryzae in rice. Plant Pathology. 2015;64(3):568–75.