RESEARCHFree Access icon

Genetic Diversity and Distribution of Korean Isolates of Burkholderia glumae

    Affiliations
    Authors and Affiliations
    • Okhee Choi1
    • Seunghoe Kim2
    • Byeongsam Kang3
    • Yeyeong Lee2
    • Juyoung Bae2
    • Jinwoo Kim1 2 3
    1. 1Institute of Agriculture & Life Science, Gyeongsang National University, Jinju 52828, Republic of Korea
    2. 2Department of Plant Medicine, Gyeongsang National University, Jinju 52828, Republic of Korea
    3. 3Division of Applied Life Science, Gyeongsang National University, Jinju, Republic of Korea

    Abstract

    Burkholderia glumae causes panicle blight of rice (grain rot in Japan and Korea), and the severity of damage is increasing worldwide. During 2017 and 2018, 137 isolates of B. glumae were isolated from symptomatic grain rot of rice cultivated in paddy fields throughout South Korea. Genetic diversity of the isolates was determined using transposase-based PCR (Tnp-PCR) genomic fingerprinting. All 138 isolates, including the B. glumae BGR1 strain, produced toxoflavin in various amounts, and 17 isolates produced an unidentified purple or orange pigment on Luria-Bertani medium and casamino acid-peptone-glucose medium, respectively, at 28°C. Transposase-based PCR genomic fingerprinting was performed using a novel primer designed based on transposase (tnp) gene sequences located at the ends of the toxoflavin efflux transporter operon; this method provided reliable and reproducible results. Through Tnp-PCR genomic fingerprinting, the genetic groups of Korean B. glumae isolates were divided into 11 clusters and three divisions. The Korean B. glumae isolates were mainly grouped in division I (73%). Interestingly, most of the pigment-producing isolates were grouped in divisions II and III; of these, 10 were grouped in cluster VIII, which comprised 67% of this cluster. Results of a phylogenetic analysis based on tofI and hrpB gene sequences were consistent with classification by Tnp-PCR genomic fingerprinting. The BGR1 strain did not belong to any of the clusters, indicating that this strain does not exhibit the typical genetic representation of B. glumae. B. glumae isolates showed diversity in the use of carbon and nitrogen sources, but no correlation with genetic classification by PCR fingerprinting was found. This is the first study to analyze the geographical distribution and genetic diversity of Korean B. glumae isolates.

    Bacterial grain rot of rice caused by Burkholderia glumae is one of the most severe diseases in rice-growing countries, including Asia, South and Central America, and Africa (Karki et al. 2012; Kim et al. 2010; Zhou-qi et al. 2016). B. glumae has been reported to cause bacterial wilt in many field crops, and it is an important opportunistic human pathogen in chronic granulomatous disease patients (Jeong et al. 2003; Weinberg et al. 2007). B. glumae causes spikelet sterility and rot of young grains, resulting in large losses in yield (Fang et al. 2009; Tsushima et al. 1996). This bacterial disease occurs at the flowering stage and causes disease at high temperatures and during periods of frequent rainfall (Cha et al. 2001). In Korea, the incidence of this disease is expected to increase because of increased temperatures and humidity levels in the summer. Although some studies have reported the use of chemicals and the development of disease-resistant varieties, effective methods of control are still lacking (Karki et al. 2012; Koh et al. 2011; Zhou-qi et al. 2016).

    The detection and identification of B. glumae is based on traditional biochemical and molecular methods. Phenotypic descriptions including physiological characteristics, colony morphology, pathogenicity tests, biology, and fatty acid methylester analyses have been used for the identification of B. glumae (Luo et al. 2008). PCRs, real-time PCRs, and bio-PCRs using specific primers designed for gyrB and rpoD (Maeda et al. 2006), the 16S to 23S rDNA spacer region (Sayler et al. 2006), and the rhs gene family (Kim et al. 2012a) have been developed to detect bacteria. To analyze genetic similarities and differences in B. glumae, genomic fingerprints obtained using repetitive extragenic palindromic (REP) elements, enterobacterial repetitive intergenic consensus (ERIC) sequences, and BOX element sequence-based PCRs have been reported (Saylerp et al. 2006; Karki et al. 2012).

    Despite the occurrence of rice grain rot caused by B. glumae, the distribution and genetic diversity of Korean B. glumae isolates have not been studied. In this study, we collected 137 isolates of B. glumae from diverse geographical rice fields in Korea in 2017 and 2018. Because the previously described rep-PCR method was not successful with the Korean isolates, we developed a novel transposase-based PCR (Tnp-PCR) genomic fingerprinting method using newly designed primers. We found that all 137 isolates as well as the B. glumae BGR1 strain produced comparable amplified fragments, and the isolates were composed of 11 clusters and three divisions. Furthermore, the results of a Tnp-PCR genomic fingerprinting analysis were consistent with tofI and hrpB gene sequence analyses results. In this study, the division I group of B. glumae isolates comprised the predominant isolates distributed in Korea. The results of this study will be important in effectively managing rice grain rot in Korea.

    Materials and Methods

    Isolation and identification of the pathogen.

    Samples of panicle blight with grain rot were collected from rice fields at 137 different locations in seven Korean provinces (Gyeonggi-do, Chungcheongbuk-do, Chungcheongnam-do, Jeollabuk-do, Jeollanam-do, Gyeongsangbuk-do, and Gyeongsangnam-do) in 2017 and 2018. Bacterial isolation was performed as previously described (Kim et al. 2010). Colonies with numerous crystals of oxalic acid and toxoflavin on Luria-Bertani (LB) agar plates (10 g tryptone, 5 g yeast extract, and 5 g NaCl per liter) were observed under a microscope, isolated, and purified by streaking on new LB agar plates. Table 1 lists the 138 isolates and the type strain ATCC 33617 analyzed in this study. To confirm the genus and species of the isolates, the complete 1.5-kb fragment of the bacterial 16S rRNA gene region was amplified and sequenced. The primers and PCR reactions used to identity B. glumae were the same as those used previously (Jeong et al. 2003). To confirm the identity, all isolates were subjected to whole-cell proteome analyses based on whole-cell matrix-assisted laser desorption/ionization time-of-flight mass spectrometry (MALDI-TOF MS). A search using MALDI Biotyper software was performed according to the manufacturer’s instructions.

    Table 1. Burkholderia glumae isolates used in this study

    Plant assays.

    All isolates were subjected to hypersensitivity assays on tobacco leaves and pathogenicity tests on rice sheaths as described previously (Choi et al. 2012; Kim et al. 2012b).

    Toxoflavin production.

    The toxoflavin production of B. glumae isolates was determined by the previously described toxoflavin biosensor (Choi et al. 2013) and extraction method (Kim et al. 2004). The isolates, including the BGR1 strain, were cultured in 2 ml of LB broth at 37°C for 24 h with shaking. Cell-free supernatants were extracted with an equal volume of chloroform. After evaporation, the chloroform residue was dissolved in 200 μl of sterile distilled water (SDW), and the absorbance at 260 nm was measured immediately with a spectrophotometer (NanoDrop 2000c, Thermo Fisher Scientific, Waltham, MA). The concentration of toxoflavin produced by the isolates was measured by comparison with the standard curve of synthetic toxoflavin at different concentrations.

    Pigmentation determination.

    To assess pigmentation of B. glumae isolates, the isolates BGR1 and ATCC 33617 were streaked on LB or casamino acid-peptone-glucose (CPG) agar plates (1 g casein hydrolysate, 10 g peptone, 5 g glucose, and 15 g agar per liter) and incubated at 28°C for 7 days. Orange colonies on LB agar plates and purple colonies on CPG agar plates were selected, suspended in 1 ml of SDW, and extracted with 1 ml of 100% methanol. Absorbance values of the orange and purple pigments were 450 and 600 nm, respectively, as measured with a spectrophotometer (GENESYS 10S UV-VIS, Thermo Fisher Scientific).

    Tnp-PCR genomic fingerprinting.

    ERIC-PCR and Rep-PCR primers and reactions were the same as those used previously (Louws et al. 1994, 1999). To analyze the genetic diversity of Korean B. glumae isolates, we performed Tnp-PCR genomic fingerprinting using BG-tnp1 (5′-TCCTCTCGGTTCGCCGCCGGCGCA-3′) as an anchor primer and MarTDL2 (5′-GACACGGGCCTCGANGNNNCNTNGG-3′) as a touchdown primer (Xu et al. 2013). The total DNA content was extracted using the Phire Plant Direct PCR Kit (Thermo Fisher Scientific) according to the manufacturer’s instructions. The PCR amplification was performed using a T100 thermal cycler (Bio-Rad Laboratories, Hercules, CA, USA) as follows: 95°C for 3 minutes; 30 cycles of 95°C for 30 seconds, 60°C for 30 seconds and decreasing by 0.5°C per cycle, and 72°C for 30 seconds; 20 cycles of 95°C for 30 seconds, 45°C for 30 seconds, and 72°C for 30 seconds; and 72°C for 7 minutes. The amplified products were separated by electrophoresis on 1.5% (weight/volume) agarose gels. Individual DNA fragments detected reproducibly were scored as present or absent. Binary matrices consisting of 0 (absence of a given band) and 1 (presence of a given band) were analyzed to obtain simple matching coefficients among the isolates using NTSYS-pc (version 2.2, Exeter Biological Software, Setauket, NY). Then, the simple matching coefficients were clustered to construct similarity trees with the SHAN clustering program using the unweighted pair-group method with arithmetic average (UPGMA) algorithm in NTSYS-pc (Jeong et al. 2007). Three independent experiments were performed for each Tnp-PCR run.

    Phylotype analysis.

    To analyze the phylotype of B. glumae isolates, the hrpB and tofI genes of 11 representative cluster isolates (RCIs) (R1, R5, R12, R13, R50, R65, R84, R93, R95, R96, and R121) taken from each Tnp-PCR cluster (I to XI) were amplified with primer pairs hrpB-1 (5′-ATGTTTTCGATGCTCTAT-3′) and hrpB-2 (5′-TCAGCGCTCGAGCGTCTC-3′) and tofI-NdeI (5′-GGCCATATGCAAACTTTCGTTCAC-3′) and tofI-XhoI (5′-GGCCTCGAGGGCCGCTTCGGGTTGCGA-3′), respectively (Kang et al. 2008; Kim et al. 2004). Then, PCR amplification was performed as described for the 16S rRNA gene region. The PCR products were purified using Expin Gel SV (GeneAll Biotechnology, Seoul, Korea) according to the manufacturer’s instructions. Sequencing was performed at Macrogen (Daejeon, Korea) using the same primer pairs that were used for PCR amplification. Partial sequences were analyzed using the BLAST program of the National Center for Biotechnology Information reference database. A phylogenetic tree was constructed using the neighbor-joining method and Tajima-Nei distance model in MEGA X software (Kumar et al. 2018).

    Substrate utilization profiles.

    The substrate utilization profiles of 11 representative isolates taken from each Tnp-PCR cluster were analyzed using the Biolog GN microplate system (Biolog Inc., Hayward, CA) according to the manufacturer’s instructions. Bacterial cells were grown on universal growth agar medium for 24 hours at 37°C and suspended in SDW. The bacterial suspensions were dispensed onto microplates. The plates were incubated for 24 hours at 37°C, and the color change in each well was recorded. The isolates were also tested for their ability to oxidize dulcitol as described previously (Jeong et al. 2007). Then, the simple matching coefficients were clustered to construct similarity trees with the SHAN clustering program using the UPGMA algorithm in NTSYS-pc (Jeong et al. 2007).

    Results

    Identification and geographical distribution of B. glumae in Korea.

    Panicles infected with B. glumae were clearly distinguishable from healthy panicles in rice fields. As shown in Figure 1, the diseased, discolored panicles were erect because of spikelet sterility. B. glumae forms numerous crystals of oxalic acid and toxoflavin in colonies at 37°C; this bacterium is easily isolated from diseased panicles because of its production of toxoflavin, which has antimicrobial activity. We sampled symptomatic rice panicles with blight and grain rot during rice-growing seasons in 2017 and 2018 in South Korea. A total of 137 Korean B. glumae isolates (R1 to R138) were isolated (Table 1) and identified by a whole-cell proteome analysis (MALDI-TOF MS) and sequence analysis of the 16S rRNA gene region. The 16S rRNA gene region sequence analysis by BLAST revealed high identities with B. glumae (>99%), and the MALDI-TOF MS analysis provided species identification (≥2.1 MALDI-TOF MS score value). Both methods confirmed identification of the species; based on these results, a geographic distribution map was generated. B. glumae is not limited to a specific region; it is distributed throughout Korea, mainly in coastal and plain areas (Fig. 1; Table 1).

    Fig. 1.

    Fig. 1. Symptoms, colony morphological features, and geographic map showing locations where Burkholderia glumae isolates were collected. A, Typical panicle blight with grain rot of rice. B, A single colony (1 mm) of B. glumae filled with oxalic acid crystals and toxoflavin was observed under a microscope (×40 magnification). C, Locations in South Korea where B. glumae isolates were collected from symptomatic grain rot of rice or panicle blight. The B. glumae isolates were collected in 2017 (yellow circle) and 2018 (orange square). Blue symbols indicate the locations of pigment-producing isolates.

    Download as PowerPoint

    Toxoflavin production and plant assays.

    The key virulence factor of rice grain rot-causing B. glumae is toxoflavin (Jeong et al. 2003; Karki et al. 2012; Kim et al. 2004). Toxoflavin production was determined by the toxoflavin-biosensor strain and chloroform extraction; the BGR1 strain was used as a positive control. All of the isolates including the BGR1 strain produced toxoflavin (Table 1); the range of toxoflavin production was 18.7 to 230.5 µg/ml (Supplementary Fig. S1). All of the isolates including the BGR1 strain produced a hypersensitive response on tobacco leaves and sheath rot of rice plants (Table 1). The B. glumae type strain ATCC 33617, known as the tofR spontaneous mutant (Devescovi et al. 2007), did not produce toxoflavin and was nonpathogenic to rice (Table 1).

    Purple and orange pigmentations.

    Interestingly, some isolates produced purple or orange pigment at 28°C, and pigmented and nonpigmented isolates were characterized. Seventeen isolates (R2, R16, R21, R33, R43, R45, R47, R55, R90, R93, R98, R99, R101, R104, R116, R123, and R132) produced a nondiffusible purple pigment on CPG at 28°C (Fig. 2). The amount of pigment produced by each of the 17 isolates was variable (Fig. 3A). Methanol extracts exhibited antifungal activity against Colletotrichum orbiculare (Supplementary Fig. S2). These results are consistent with those of a previous study that showed that some strains of B. glumae produced a purple pigment on CPG medium, and that the pigment exhibited antifungal activity against C. orbiculare (Karki et al. 2012).

    Fig. 2.

    Fig. 2. Purple and orange pigments produced by Burkholderia glumae isolates. Production of orange and purple pigments on Luria-Bertani (LB) medium and casamino acid-peptone-glucose (CPG) medium, respectively, by B. glumae isolates. Photographs were obtained after 7 days at 28°C.

    Download as PowerPoint
    Fig. 3.

    Fig. 3. Purple and orange pigments and toxoflavins production by Burkholderia glumae isolates. A, Purple pigment produced by B. glumae isolates. B, Orange pigment produced by B. glumae isolates.

    Download as PowerPoint

    In addition, after incubation on LB at 28°C for 5 days, a clearly distinct nondiffusible orange pigment produced by isolates was observed with the naked eye (Fig. 2). The ultraviolet (UV) absorbance of methanol extracts of these isolates was measured at 450 nm. The amount of pigment produced by each of the 17 isolates was variable (Fig. 3B). Unlike the purple pigment, the orange pigment did not show antifungal activity against C. orbiculare (Supplementary Fig. S2). At present, we do not know the chemical compositions of the purple and orange pigments; ATCC 33617 and BGR1 strains do not produce pigment (Fig. 2). Toxoflavin production of the 17 isolates was also compared, and no association between toxoflavin production and pigmentation was found (Supplementary Fig. S1).

    Primer design for Tnp-PCR genomic fingerprinting.

    To analyze the genetic diversity of the 137 Korean rice isolates of B. glumae, Rep-PCR fingerprint using ERIC-PCR, BOX-PCR, and REP-PCR was first performed. ERIC-PCR amplified the fragments marked by band patterning on the agarose gel in only 70% of the isolates (data not shown). To conduct genetic diversity studies, the method used must be feasible in at least 90% of isolates. Therefore, we designed a new strategy (Fig. 4A). To design specific PCR primers, we focused on the structure of the toxoflavin-producing gene operons of B. glumae. We reported the presence of two truncated transposase genes (tnp) at the ends of the toxI and toxE genes (Kim et al. 2004). The presence of transposase suggests transfer of the toxoflavin gene operons, which makes it possible to conduct genetic diversity studies via analysis of the surrounding DNA sequences. We designed two anchor primers, BG-tnp1 and BG-tnp2, at both tnp ends. The touchdown primers MarTDL1 and MarTDL2 (Xu et al. 2013) were used as complementary primers. Figure 4B illustrates the Tnp-PCR genomic fingerprinting fragments amplified using the BG-tnp1 and MarTD2 primer pairs. The newly synthesized and complementary strands are marked in the same color. The size of fragments was dependent on the DNA sequence surrounding tnp, which was used to study the genetic diversity of B. glumae (Fig. 4B). The PCR results using different primer combinations revealed that the combination of BG-tnp1 and the complementary primer MarTD1 or MarTD2 resulted in a large number of amplified PCR fragments (Fig. 4C). Therefore, the primers BG-tnp1 and MarTD2 were used in subsequent Tnp-PCR genomic fingerprinting analysis.

    Fig. 4.

    Fig. 4. Primer design for Tnp-PCR genomic fingerprinting. A, Organization of the toxoflavin efflux transporter and biosynthetic gene operons and positions of the BG-tnp primers. Green and yellow arrows indicate toxoflavin transport and biosynthetic genes, respectively. Orange arrows indicate the transposase gene. B, Illustration of Tnp-PCR genomic fingerprinting fragments amplified using anchor and touchdown primer pairs. The newly synthesized strand and its complementary strand are marked in the same color. C, Tnp-PCR genomic fingerprinting of the BGR1 strain using two anchor primers, BG-tnp1 or BG-tnp2, at both tnp ends and complementary touchdown primers, MarTDL1 or MarTDL2. Green box indicates optimal primers.

    Download as PowerPoint

    Analysis of the genetic diversity of B. glumae isolates using Tnp-PCR genomic fingerprinting.

    Amplified DNA fragments ranged in size from 180 bp to 7.0 kb and exhibited 6 to 18 bands for the whole set of isolates; a total of 41 discrete bands were scored. The amplified DNA fragments of all isolates were consistently and reproducibly produced in three independent reactions. A dendrogram was used to separate the isolates into three major divisions. Division I, with a coefficient of 0.73, contained 102 B. glumae isolates and was further divided into seven clusters (I to VII). The BGR1 strain and four isolates (R19, R46, R94, and R99) did not belong to any of the clusters. Most of the division I isolates were nonpigment producers, and only four isolates (R16, R90, R99, and R101) produced pigment (Fig. 5). Division II, with a coefficient of 0.72, contained 23 isolates and was divided into two clusters (VIII and IX). In particular, 10 (R33, R98, R2, R47, R116, R21, R45, R123, R43, R123, R43, and R55) of the 23 isolates produced pigment and all belonged to cluster VIII (Fig. 5). Division III, with a coefficient of 0.61, contained 14 isolates and was divided into two clusters (X and XI). The band pattern of division III was more diverse than those of other divisions. Five isolates (R4, R37, R79, R117, and R138) did not belong to any of the clusters. Three isolates (R93, R104, and R132) produced pigment and all belonged to cluster XI (Fig. 5). Interestingly, most pigment-producing isolates were grouped into clusters VIII and XI of divisions II and III, respectively. There were 15 isolates grouped in cluster VIII; of these, 10 produced pigments and accounted for 67% of the cluster. These results indicate that the groups obtained by Tnp-PCR genomic fingerprinting accounted for the genetic diversity. This result shows that the phenotypic and genotypic characteristics are well-matched, and that the Tnp-PCR method used during this study should be useful in further analyses of genetic diversity. However, the group obtained by Tnp-PCR genomic fingerprinting failed to explain its geographic origin (Figs. 1 and 5). The most interesting aspect of the Tnp-PCR results is that the BGR1 strain did not exhibit the typical genetic representation of B. glumae. The ATCC 33617 was grouped in division I and cluster V (Fig. 5).

    Fig. 5.

    Fig. 5. Dendrogram obtained by a comparison of Tnp-PCR genomic fingerprinting patterns from Burkholderia glumae isolates. The data were clustered by the unweighted pair-group method with the arithmetic average to indicate correlations between Tnp-PCR genomic fingerprints of B. glumae isolates. Pigment-producing isolates are denoted by +.

    Download as PowerPoint

    Phylogenetic analysis of 16S rRNA, hrpB, and tofI.

    The complete 16S rRNA gene sequences of 11 RCIs (R1, R5, R12, R13, R50, R65, R84, R93, R95, R96, and R121) taken from each Tnp-PCR cluster and ATCC 33617 and BGR1 strains were analyzed. A comparison of 1455 nucleotides revealed that four of the 11 representative isolates showed a difference ranging from one to four nucleotides compared with the BGR1 strain; the remaining nucleotides were identical (Table 1, Fig. 6A). Isolate R13, which belongs to cluster VI and division I, and isolate R93, which belongs to cluster XI and division III, were distinct from the other isolates at nucleotides 814 (U) and 1221 (U), respectively. Isolate R84, which belongs to cluster IX and division II, was distinct at nucleotides 987 (C), 1097 (G), 1167 (C), and 1217 (G). Isolate R121, which belongs to cluster IV and division I, was distinct at nucleotides 442 (U), 1305 (U), 1337 (G), and 1345 (U) (Supplementary Table S1). The bootstrap value of the 16S rRNA gene sequences in 11 RCIs and ATCC 33671 and BGR1 strains ranged from 37 to 65% (Fig. 6A). A comparative analysis of the 16S rRNA gene sequence did not explain the division obtained by Tnp-PCR genomic fingerprinting (Fig. 6A).

    Fig. 6.

    Fig. 6. Phylogenetic trees based on 16S rRNA, hrpB, and tofI gene sequences of 11 cluster representative isolates of Burkholderia glumae. A, Dendrogram based on a comparison of 16S rRNA sequences constructed using the pair-group method, with the arithmetic average showing the phylogenetic relationships among B. glumae isolates. Numbers at branch points are percentages of bootstrap replicates in which clusters were found. Bars indicate the sequence differences among isolates. B, Phylogenetic tree based on a comparison of hrpB gene sequences. C, Phylogenetic tree based on a comparison of tofI gene sequences. The phylogenetic trees in B and C were generated using the neighbor-joining method. aData were grouped by Tnp-PCR genomic fingerprinting patterns. Values at branches indicate the percent bootstrap support for 1000 replicates. Bars indicate one nucleotide change per 100 nucleotide positions.

    Download as PowerPoint

    Partial hrpB gene sequences (1404 bases) covering approximately 97% of the intact gene from 11 RCIs and ATCC 33617 and BGR1 strains were compared. The phylogenetic tree was clearly separated into division I and divisions II and III based on Tnp-PCR genomic fingerprinting (Fig. 6B, Table 2). Partial tofI (acyl-HSL synthase) gene sequences (573 bases) covering approximately 94% of the entire gene were also compared. Consistent with the hrpB gene analysis, the dendrogram was also grouped into division I and divisions II and III (Fig. 6C, Table 2). The phylogenetic analysis of hrpB and tofI was consistent with Tnp-PCR genomic fingerprinting analysis.

    Table 2. Cluster representative isolates of Burkholderia glumae used in hrpB and tofI gene sequence analyses

    Substrate utilization.

    Substrate utilization of the 11 RCIs and the BGR1 strain was analyzed. In 97 substrate reactions, 11 RCIs and BGR1 showed differences in the utilization of 24 substrates. The analysis of 24 nutritional characteristics used two separate groups. The first group contained all division (I to III) isolates, and the second group comprised division I and II isolates (Supplementary Fig. S3). The nutrient utilization profiles showed diversity in carbon and nitrogen utilization patterns of the Korean isolates of B. glumae, but this diversity did not explain the geographical origin or genetic diversity.

    Discussion

    B. glumae is a bacterial pathogen that causes panicle blight, grain rot of rice, and wilt in a variety of field crops, resulting in significant losses in crop yield. This bacterium grows well and produces a major virulence factor, toxoflavin, at 37°C, which may become an issue in rice cultivation countries with subtropical climates, such as Korea, China, and Japan (Cha et al. 2001), because outbreak of this disease is closely related to temperature (Jeong et al. 2003). We have been monitoring rice grain rot since 2000 in South Korea (Jeong et al. 2003; Kim et al. 2010); rice grain rot occurred throughout South Korea in 2017 and 2018. The recent increase in the maximum temperature in major rice cultivation regions in South Korea and outbreaks of this disease since 2017 may be linked (Supplementary Fig. S4). In 2017 and 2018, we isolated 137 Korean B. glumae isolates from symptomatic grain rot or panicle blight of rice throughout South Korea and prepared a distribution map of B. glumae (Fig. 1). Prediction of rice grain rot by analyzing trends and correlations of maximum temperatures during the rice growing period is thought to be possible.

    Although the amount of toxoflavin produced by each isolate in vitro varied, a precise test to determine a correlation between the amount of toxoflavin produced and pathogenicity was not performed. Although toxoflavin is important, conditions for its production in vitro can differ significantly from those in vivo. Moreover, the amount of toxoflavin produced by each isolate may vary depending on the in vitro and in vivo conditions. Quantifying the amount of toxoflavin produced in vivo could be correlated with pathogenicity. Plant pathogenic bacteria have evolved to monitor environmental factors and have developed virulence in plants accordingly (Leonard et al. 2017). Environmental conditions are important for the survival of B. glumae.

    Seventeen isolates produced an unidentified, nondiffusible purple pigment on CPG and an orange pigment on LB at 28°C (Fig. 2). The purple pigment produced by B. glumae on CPG was previously characterized as a phenazine-like compound, not a melanin-type or shikimate pathway-dependent pigment, which have antifungal activity (Karki et al. 2012; Karki and Ham 2014). These characterizations were consistent with our results (Supplementary Fig. S2). The purple pigment has a role in the tolerance of B. glumae to UV light and has no effect on toxoflavin production (Karki and Ham 2014). It is unknown whether there is a link between purple pigment synthesis and orange pigment synthesis. The orange pigment looks similar to the bacterial carotenoid pigment; however, the structure and properties of this orange pigment require further study.

    Universal DNA fingerprinting systems such as REP-PCR, BOX-PCR, and ERIC-PCR are mainly performed to analyze genetic diversity in bacteria (Chen et al. 2011; Choi et al. 2016; Gent et al. 2005; Sayler et al. 2006). PCR-based fingerprinting is a simple and widely used typing method in many laboratories; however, this method is known to be sensitive to the quality of the material and equipment used and has low reproducibility (Isaeva et al. 2010; Johnson and Clabots 2000). The weaknesses of these PCR-based fingerprinting technologies were reproduced in our study, and a new primer was designed. Specific PCR primers were designed using the structures of toxoflavin production genes in B. glumae. Interestingly, two tnp genes exist at both ends of the toxoflavin efflux transporter and biosynthetic gene operons of B. glumae BGR1. The tnp gene is a good indicator of genetic diversity in prokaryotes and eukaryotes. The tnp genes located at the ends of the gene operons that synthesize and transport toxoflavin appear to be truncated and nonfunctional. B. glumae may have acquired these genes evolutionarily to induce disease in the host. Therefore, the DNA sequences surrounding the toxoflavin-producing genes may vary among different isolates. Tnp-PCR using the anchor primer BG-tnp1 and touchdown primer MarTDL2 generated a variety of amplicons suitable for pattern analysis of all isolates. Using the Tnp-PCR fingerprinting method, Korean B. glumae isolates were classified into 11 clusters and three divisions. Seventy-three percent of Korean B. glumae isolates were grouped in division I, which indicates that division I is the most common division among the Korean isolates. If the same lineage strains are isolated from various rice cultivars grown over the years in a particular region, then seed transmission could be considered an important route of infection for B. glumae (Azegami et al. 1988; Maeda et al. 2007; Tabei et al. 1989). According to Maeda et al. (2007), a specific lineage strain was distributed in the Iwate region of Japan in 2007. In the Iwate region, because multiple cultivars of rice seedlings were usually produced and supplied by one seedling-producing institute, it was possible that the bacteria were horizontally transferred to other varieties during seedling production through seed infection. However, unlike Iwate, various cultivars of rice seedlings have been supplied from several private nurseries in Korea; therefore, it is unlikely that certain lineage strains will intensively spread during seedling processes by seed transmission. As shown by our results, division I isolates are not concentrated in a specific region; they are distributed nationwide, suggesting that division I isolates have already settled in the Korean rice fields and are adapted to the agricultural environment of South Korea. Among the Korean isolates, 10 isolates including the BGR1 strain were not grouped in any cluster in Tnp-PCR fingerprinting. To explain these results, information regarding the varieties of the isolated origin and cultivation history is required; however, there is currently no such information available. However, because the BGR1 strain was isolated before 2001 and used as a Korean type strain, it is obvious that there was a large change in the genetic diversity of B. glumae in Korea. Interestingly, pigment-producing isolates were mainly grouped in divisions II and III; 10 pigment-producing isolates were grouped into cluster VIII (67%). These results support the use of Tnp-PCR to study the genetic diversity of B. glumae and confirm the presence of genetic diversity in Korean B. glumae isolates.

    Based on the 16S rRNA gene sequence analysis, we confirmed that the sequence similarity among 11 RCIs was relatively high. In contrast, the 16S rRNA gene sequences of Ralstonia solanacearum isolates collected from several hosts and countries showed various differences (Jeong et al. 2007). This result is thought to be attributable to the simplicity of the host. The phylogenetic analysis based on hrpB and tofI loci was consistent with grouping by Tnp-PCR fingerprinting. However, classification based on an analysis of Tnp-PCR fingerprinting and the phylogenetic tree did not reflect the geographical origin. Substrate utilization profiling has been traditionally used to classify bacteria (Marques et al. 2008). In addition to genetic classifications, phenotypic descriptions have been used to classify various bacteria such as Pseudomonas, Ralstonia, and Xanthomonas (Gent et al. 2005; Jeong et al. 2007; Marques et al. 2008). A substrate utilization analysis revealed that 11 RCIs of B. glumae showed various levels of carbon and nitrogen utilization, but no significant correlation was found with genetic diversity based on Tnp-PCR fingerprinting and phylogenetic analysis.

    Collectively, 137 Korean B. glumae isolates were collected from rice in Korea. These isolates were classified into 11 clusters and three divisions based on Tnp-PCR-fingerprinting using specific tnp primers. Genetically, the Korean B. glumae isolates were predominantly distributed in division I. In particular, the Tnp-PCR fingerprinting method produced stable and reproducible results. Because the existence of tnp in the genome is indicative of genetic diversity, the Tnp-based PCR fingerprinting performed during this study will be useful for other bacteria as well as eukaryotic organisms.

    The author(s) declare no conflict of interest.

    Literature Cited

    Okhee Choi and Seunghoe Kim contributed equally to this work.

    Funding: This research was supported by Basic Science Research Program through the National Research Foundation of Korea (NRF) funded by the Ministry of Education of the Republic of Korea (2015R1A6A1A03031413).

    The author(s) declare no conflict of interest.