Type VI secretion systems of plant‐pathogenic Burkholderia glumae BGR1 play a functionally distinct role in interspecies interactions and virulence

Abstract In the environment, bacteria show close association, such as interspecies interaction, with other bacteria as well as host organisms. The type VI secretion system (T6SS) in gram‐negative bacteria is involved in bacterial competition or virulence. The plant pathogen Burkholderia glumae BGR1, causing bacterial panicle blight in rice, has four T6SS gene clusters. The presence of at least one T6SS gene cluster in an organism indicates its distinct role, like in the bacterial and eukaryotic cell targeting system. In this study, deletion mutants targeting four tssD genes, which encode the main component of T6SS needle formation, were constructed to functionally dissect the four T6SSs in B. glumae BGR1. We found that both T6SS group_4 and group_5, belonging to the eukaryotic targeting system, act independently as bacterial virulence factors toward host plants. In contrast, T6SS group_1 is involved in bacterial competition by exerting antibacterial effects. The Δ tssD1 mutant lost the antibacterial effect of T6SS group_1. The ΔtssD1 mutant showed similar virulence as the wild‐type BGR1 in rice because the ΔtssD1 mutant, like the wild‐type BGR1, still has key virulence factors such as toxin production towards rice. However, metagenomic analysis showed different bacterial communities in rice infected with the ΔtssD1 mutant compared to wild‐type BGR1. In particular, the T6SS group_1 controls endophytic plant‐associated bacteria such as Luteibacter and Dyella in rice plants and may have an advantage in competing with endophytic plant‐associated bacteria for settlement inside rice plants in the environment. Thus, B. glumae BGR1 causes disease using T6SSs with functionally distinct roles.


| INTRODUC TI ON
Myriad bacteria demonstrate close associations with nearby bacteria, as well as with their host organisms (Freestone, 2013;Stubbendieck et al., 2016). Bacteria can persist and adapt in their ecological niche by continuously interacting with other organisms, and have developed many protective and infectious mechanisms against competitors and host organisms. Some of these, such as bacterial secretion systems, biofilms, and phytotoxins, provide survival benefits against bacterial and eukaryotic cells, and facilitate environmental adaptation (Backert and Meyer, 2006). In particular, bacterial secretion systems are considered the primary and effective means for communication and adaption to the surroundings (Russell et al., 2014).
Bacterial secretion systems are classified into nine groups (Tat system, types I-VII, type IX) according to their structures, functions, and specific effectors (Stanley et al., 2000;Whitney et al., 2013;Kneuper et al., 2014;Green and Mecsas, 2016). The type VI secretion system (T6SS) was identified as a bacterial secretion system in an opportunistic pathogen, Pseudomonas aeruginosa (Mougous et al., 2006). The T6SS is widespread in around 25% of gram-negative bacteria and is easily found in pathogenic or symbiotic plant-associated bacteria (Boyer et al., 2009;Bernal et al., 2018). The T6SS forms proteinaceous machinery that delivers substrates (effectors) to adjacent bacterial or eukaryotic cells in a contact-dependent manner or releases effectors tothe extracellular environment. The effectors secreted by these systems are used to destroy and manipulate eukaryotic cells and/or to fight other bacteria to gain dominant status in the same ecological niche (Ma and Mekalanos, 2010). At least one T6SS gene cluster is present in some bacteria. The presence of two T6SS gene clusters in one organism increases the possibility that each system has a different function depending on the circumstances (Bingle et al., 2008). More recently, functional T6SSs were discovered in a broad spectrum of bacterial species and were found to contribute to virulence in host organisms, improve bacterial robustness, and enhance adaptation to polymicrobial environments through bacterial competition (Russell et al., 2014). For example, P. aeruginosa has H1-T6SS, H2-T6SS, and H3-T6SS. H1-T6SS contributes only to interactions between bacteria by the delivery of toxins and effector molecules . H2-T6SS and H3-T6SS also exhibit antibacterial activity by secreting trans-kingdom effectors, but are involved in virulence against eukaryotes Jiang et al., 2014). T6SS-5 of Burkholderia thailandensis is also a major virulence factor in the murine model of acute melioidosis (Schwarz et al., 2010).
The T6SS is composed of a minimal set of 13 core components, TssA-TssM (type six secretion). Apart from these major components, several accessory proteins are encoded in the T6SS gene cluster, and work together with the T6SS effector to damage the inner membrane of prokaryotic target cells or participate in negative regulation of the T6SS (Miyata et al., 2013;Lin et al., 2018). These components are generally encoded by clustered genes; each cluster has a different gene organization despite the clusters being in a single organism.
Among the core components, TssD, also known as Hcp (hemolysin co-regulated protein), and VgrG (valine-glycine repeat protein G) have been considered the hallmark apparatus of T6SS as well as secreted effectors in all bacteria with functional T6SSs (Pukatzki et al., 2007).
B. glumae is a gram-negative bacterium known as a seedborne phytopathogenic bacterium that causes bacterial panicle blight as well as sheath and seedling rot in rice (Oryza sativa) (Goto, 1956;Goto et al., 1987). The major virulence factors of B. glumae are toxoflavin, lipase, type III effectors, and extracellular polysaccharides, along with cell motility (Iiyama et al., 1995;Suzuki et al., 2004;Kim et al., 2007;Kang et al., 2008;Lee et al., 2016;Jung et al., 2018). The T6SS was classified into six groups based on the distribution of 5 out of 13 T6SS components in 12 Burkholderia strains, including B. glumae (Seo et al., 2015). Of these six T6SS groups, four T6SS groups are found in B. glumae BGR1. However, their functions are still unexplored.
In this study, we report that T6SS group_1 has antibacterial effects but is dispensable for virulence, whereas T6SS group_4 and T6SS group_5 were found to directly affect the pathogenicity of B. glumae BGR1 targeting rice plants through phylogenetic analysis and a phenotypic assay. Furthermore, metagenomics analysis of the endophytic bacterial community inside plants identified different bacterial communities with the ΔtssD1 mutant compared to those with the wild-type BGR1. This indicates that T6SS group_1 inhibits or limits the growth of endophytic plant-associated bacteria, thereby helping pathogenic bacteria form a bacterial community easily. Our findings clearly describe the important role of T6SS in interspecies interactions, including pathogen-host interaction and interaction with other bacteria.

| Evolutionary analysis of type VI secretion systems in Burkholderia species shows functional differentiation based on environmental adaptation
The evolutionary analysis of Burkholderia T6SS in B. mallei,B. pseudomallei,B. thailandensis,and B. cepacia,B. gladioli,and B. glumae,was performed to identify the distinct roles of the four T6SS in B. glumae BGR1, using at least 10 major core components of T6SS. The T6SS gene clusters of B. glumae BGR1, B. plantarii ATCC43733, and B. gladioli BSR3 were annotated in our previous study (Seo et al., 2015). All Burkholderia species have at least one T6SS gene cluster. However, we found that not all T6SS gene clusters of Burkholderia species were completely composed of 13 major elements. In the resulting phylogeny, six rice-pathogenic Burkholderia species were highly conserved in at least one T6SS gene cluster, which includes T6SS group_1 of B. glumae, B. gladioli, and B. plantarii ( Figure S1). These T6SS gene clusters are closely clustered with B. thailandensis E264 T6SS-1, a well-known bacterial cell targeting system (Schwarz et al., 2010). In addition, T6SS group_4 and T6SS group_5 of B. glumae BGR1 were clustered with the T6SS gene clusters of the eukaryotic cell targeting system, which are B. thailandensis E264 T6SS-5 and B. pseudomallei K96243 T6SS-5, in one or two subtrees ( Figure S1) (Schwarz et al., 2010;Lennings et al., 2019). T6SS group_2 was clustered in one or two subtrees with T6SS-2 of B. pseudomallei K96243 and B. thailandensis E264. However, T6SS group_2 was not clustered with the bacterial and eukaryotic cell targeting systems despite the presence of a complete T6SS gene cluster. Therefore, we assumed that the T6SS of B. glumae BGR1 is involved in the bacterial cell targeting system and the eukaryotic cell targeting system through T6SS group_1 and both T6SS group_4 and 5, respectively.
To determine the functions of each T6SS, we constructed markerless deletion mutants targeting the four tssD genes via two homologous recombinations. After completion of the second recombination, tssD deletion mutant strains were confirmed by PCR ( Figure S2 and Table S3). Single deletion mutants called ΔtssD1, ΔtssD2, ΔtssD4, and ΔtssD5, corresponding to the genes tssD1, tssD2, tssD4, and tssD5, respectively, were generated. Double deletion mutants, ΔtssD12 and ΔtssD45, were generated. In addition, a quadruple deletion mutant called ΔtssD1245 was also constructed. In order to complement the phenotype of tssD deletion mutants ΔtssD1, ΔtssD4, ΔtssD5, and ΔtssD45, the respective complementation strains, Δtss1D-C, Δtss4D-C, ΔtssD5-C, and ΔtssD45-C were generated.

| Phenotypic characteristics such as motility and toxoflavin synthesis between the B. glumae BGR1 and the tssD deletion mutants are not different
There were no differences in the growth rate between wild-type B. glumae BGR1 and the single or quadruple tssD deletion mutants ( Figure S3). Several virulence-related phenotypic characters were tested. However, no apparent difference in swarming motility was observed in the mutant strains compared to that in wild-type BGR1, that is, comparable swarming motility was observed in all strains, as assessed by the formation of dendritic patterns ( Figure S4a).

| T6SS group_4 and T6SS group_5 contribute to virulence toward rice plants
To determine whether the T6SS is involved in the eukaryotic cell- Specifically, disease severity (0-5) was 4.42 ± 0.13 in wild-type BGR1, and decreased to 2.35 ± 0.11 in ΔtssD4, to 2.73 ± 0.07 in ΔtssD5, and to 3.15 ± 0.24 in ΔtssD45. Furthermore, the complemented strains, showed no synergistic effect via tssD4 and tssD5.

| T6SS group_1 is involved in antibacterial competition ability
The antibacterial competition ability was tested using apramycinresistant Escherichia coli cells as prey and cocultured B. glumae BGR1 as predators. The single E. coli cell cultures proliferated to approximately 9.33 × 10 7 cfu/ml, whereas E. coli cocultured with an equal number of wild-type BGR1 cells at a 1:1 ratio for 6 hr showed a decrease to 2.67 × 10 3 cfu/ml. As a result, the surviving number of E. coli cells grown in pure culture was more than 10 4 times higher compared to those cocultured with wild-type BGR1. To investigate whether the T6SS of B. glumae BGR1 is involved in this antibacterial effect, the numbers of E. coli cells surviving coculture with F I G U R E 1 Genetic organization of the four type VI secretion system (T6SS) clusters in Burkholderia glumae BGR1. B. glumae BGR1 has four T6SS gene clusters in its genome. The genes are indicated by the locus ID (e.g., bglu_1g03850) and are to scale. The COGs of 13 conserved T6SSs from the SecreT6 database (http://db-mml.sjtu.edu.cn/SecRe T6/) are marked with specific colours in the four T6SS clusters of B. glumae BGR1 (Li et al., 2015) ΔtssD1245 were compared with those of E. coli cells cultured alone.
The number of E. coli surviving coculture with ΔtssD1245, which proliferated to approximately 3.25 × 10 7 cfu/ml, was similar to that of E. coli cultured alone. Next, antibacterial competition assays with tssD1, tssD2, tssD4, and tssD5 single mutants as predators were conducted to determine which T6SS clusters were associated with the antibacterial effects ( Figure 3a). Only the effect of ΔtssD1 was found to be comparable with that of ΔtssD1245. The numbers of E. coli cells surviving coculture with ΔtssD1 proliferated to approximately 2.67 × 10 7 cfu/ml. The complemented strain ΔtssD1-C, with recovered function of tssD1, showed antibacterial activity almost completely restored to wild-type levels, as reflected by a sharp decrease in the surviving number of E. coli prey cells ( Figure 3b).
The BGR1 pBBR1MCS2 strain, as the negative control of the complementation strain, showed an antibacterial effect similar to wild-type BGR1 ( Figure S5d). The antibacterial effect of ΔtssD1 and ΔtssD1245 was also observed using green fluorescent protein (GFP)-labelled E. coli as prey ( Figure 3c). Therefore, only the tssD1 gene contributed to the antibacterial effect in in vitro interbacterial competition assays.

| The presence of prey increases the gene expression levels of tssD1 but not tssD2, tssD4, and tssD5
Quantitative reverse transcription PCR (RT-qPCR) was used to assess the expression levels of the four tssD genes in the presence of prey. In the presence of E. coli as prey, the relative gene expression level of only tssD1 was increased whereas the relative gene expression levels of the tssD2, tssD4, and tssD5 genes showed little difference regardless of the presence of prey ( Figure 4).

F I G U R E 2
In vivo pathogenicity assay of the vegetative and reproductive stage. (a) T6SS deletion mutants were inoculated with 10 8 cfu/ml to assess the virulence of the T6SS cluster in the xylem of rice stem (Oryza sativa) at the vegetative stage. The experiments were conducted by three replicates (n = 3). (b) Bacterial suspensions were inoculated into rice panicles at the reproductive stage to assess the virulence of ΔtssD4, ΔtssD5, ΔtssD45, ΔtssD4-C, ΔtssD5-C, and ΔtssD45-C. Representative of four replicates. (c) Disease severity on the rice panicles was calculated on a scale of 0 to 5 after inoculating the bacterial suspension. The data are presented as the mean ± SD of four replicates (n = 4). Mean values followed by the same letters are not significantly different according to Tukey's HSD test (*p < .05, **p < .01, ***p < .001). Disease symptoms at 8 days post-inoculation. Distilled water (D.W.) was used as the negative control F I G U R E 3 Interbacterial interaction between Escherichia coli DH5α and four tssD mutants of Burkholderia glumae BGR1. (a) and (b) Antibacterial effects of the four T6SS clusters in B. glumae BGR1. Survival of prey cell was strongly decreased by coculturing with wild-type BGR1, ΔtssD2, ΔtssD4, and ΔtssD5. The survival of prey cells by coculture with ΔtssD1 and ΔtssD1245 was similar to that of pure culture of prey without predators (sole). The ΔtssD1 complementation strains restored the wild-type antibacterial activity. The experiments were conducted by three replicates (n = 3). (c) Observation of living prey cells expressing green fluorescent protein (GFP) by spotting the prey and predator mixture. Except in the coculture of ΔtssD1 and ΔtssD1245, the intensity of GFP expressed in living prey cells was found to disappear after 6 hr. The experiments were conducted using three replicates (n = 3). This is representative of the results from independent experiments with three replicates showing the same pattern the observed operational taxonomic units (OTUs), the phylogenetic diversity (PD) whole tree, and the Simpson index ( Figure 5a).
Alpha diversity between rice infected with BGR1 and rice infected with ΔtssD1 showed that the observed OTUs (P observed OTUs = 0.003) and PD whole tree (P PD whole tree = 0.008) were significantly decreased in rice infected with BGR1, even though statistical testing using Chao1 (P Chao1 = 0.052) and the Simpson index (P Simpson = 0.408) showed no difference ( Figure 5a). All alpha diversity indices of noninfected rice were higher than the other groups.
The relative abundance of the top 20 genera showed that the genus Pantoea dominated the noninfected rice sample, followed by the genus Burkholderia. In contrast, the most abundant genera in rice infected with BGR1 and ΔtssD1 were Burkholderia followed by Achromobacter ( Figure S6) was used to find the strongest effects for group differentiation using imbalanced OTUs. Heatmap abundance showed differential abundance and frequency of the total bacterial groups (p < .05, LDA score > 2.5) among the three groups ( Figure 6). PCoA plots and heatmap abundance (p < .05, LDA score > 2.5) showed that the three groups constituted their own communities and presence or absence of antibacterial effects led to differences in the community of rice infected with BGR1 and ΔtssD1 at 8 dpi ( Figures 5 and 6

| D ISCUSS I ON
Type VI secretion systems (T6SSs)  forms an inner tube of stacked hexamers with helical symmetry on the assembly baseplate to transport effectors, is essential to form the T6SS needle to puncture adjacent cells (Pell et al., 2009;Cascales and Cambillau, 2012). The T6SS needle, with the VgrG-PAAR spike complex at its end, comprises an inner tube and a TssB-TssC ring surrounding the inner tube. Interestingly, Hcp (TssD) and VgrG proteins are mutually dependent when the surface is assembled (Pukatzki et al., 2007;Zheng and Leung, 2007;Hachani et al., 2011). Consequently, tssD deletion leads to loss of function of the T6SS in B. glumae BGR1 because the TssD protein is required to build an intact T6SS.
The T6SS is involved in bacterial virulence as a eukaryotic cell targeting system for plants or animals. The T6SSs as a eukaryotic cell targeting system of animals have been reported a to be essential elements for the virulence of bacterial pathogens (Keynan and Rubinstein, 2007;Suarez et al., 2010;Russell et al., 2013). Furthermore, the T6SSs, as a eukaryotic Thus, T6SS group_4 and T6SS group_5 contribute to bacterial virulence as a completely independent eukaryotic targeting system.
The T6SS, involved in the eukaryotic targeting system, is activated when certain conditions are met on entering the host. For example, T6SS-1 of B. pseudomallei, which is involved in virulence in mice and hamsters, is not expressed during culture in rich media but is expressed following uptake by the host cell (Shalom et al., 2007;Burtnick et al., 2011;Chen et al., 2011). Only T6SS group_5 of B. glumae was expressed and activated during infection of rice in Table S1 ( Kim et al., 2014). However, the exact mechanism for the involvement of T6SS group_4 and T6SS group_5 in the infection process and for the activation of T6SSs remains unclear. For a comprehensive understanding of the T6SS activation mechanisms, bioinformatics and comparative secretome approaches need to be employed in future studies.
In addition to engaging in bacterial virulence toward host organisms, the T6SS is involved in bacterial competition via antibacterial effects by T6SS-dependent antibacterial effectors like peptidoglycan hydrolases, phospholipases, nucleases, and F I G U R E 6 Identifying taxa with the most differences abundance from LEfSe analysis. LefSe analysis was performed with p < .05, LDA score > 2.5 with the relative abundance of bacterial OTUs. Blue represents lower abundance and red represents higher abundance. Relative abundances were normalized to z-score to indicate differences between samples. The side bar on the left of the heatmap indicates the bacterial phyla NAD(P) + -glycohydrolases (Koskiniemi et al., 2013;Russell et al., 2013;Whitney et al., 2013;Tang et al., 2018 (Ivanova et al., 2000;Delmotte et al., 2009). In contrast, in rice infected by B. glumae BGR1, the order Burkholderiales, which contains several pathogenic bacterial species, prevailed (Hauser et al., 2011;Voronina et al., 2015).

| In vivo pathogenicity assay at vegetative stage and reproductive stage
To examine the eukaryotic-targeting system of B. glumae BGR1 in rice plants at the vegetative stage and reproductive stage, cultured bacterial cells at the mid-logarithmic phase were harvested by centrifugation, washed, and resuspended with distilled water.
The optical density of bacterial suspensions was adjusted to OD 600 nm = 0.8. The rice plants used in this experiment were grown under greenhouse conditions (average 30°C in the day and 25°C at night). Next, the stems of rice plants in the vegetative stage were inoculated with bacterial suspensions using a syringe and grown for 8 days. The disease severity was observed at the inoculated area. Suspensions for each bacterial stain were prepared and adjusted to OD 600 nm = 0.5 to confirm the bacterial panicle blight disease at the reproductive stage of rice. At the flowering stage, the rice panicles were inoculated by dipping them into 50 ml of bacterial suspensions for 1 min. At 8 dpi, the disease severity in rice panicles was evaluated using the following scale: 0, healthy panicle; 1, 0%-20% diseased panicle; 2, 21%-40% diseased panicle; 3, 41%-60% diseased panicle; 4, 61%-80% diseased panicle; 5, 81%-100% diseased panicle. Disease severity was calculated using the following formula: disease severity = Σ(number of samples per rating × rating value)/total number of panicles. BGR1 pBBR1MCS2, which contains an empty vector, was used as a negative control of the complementation strain in the in vivo pathogenicity assay.

| In vitro interbacterial competition assay
B. glumae strains (Table S1)  Next, about 6 g of six rice stem segments were squeezed tightly in a 15-ml conical tube, and centrifuged at 4 °C and 1,800 × g for 7 min directly to harvest a pellet of the bacteria inside the stem segments.
The bacterial pellet acquired after centrifugation was immediately used to extract DNA following the classic cetyltrimethyl ammonium bromide-based protocol with modifications (Murray and Thompson, 1980). The extracted DNA was amplified to obtain paired-end se-

| Bacterial community analysis
The paired-end sequences with a p value of 0.3, obtained from Illumina MiSeq, were merged using meren/illumina-utils (https:// github.com/meren/ illum ina-utils) (Eren et al., 2013). The merged sequences were analysed using the QIIME (Quantitative Insights into Microbial Ecology) pipeline v. 1.9.1 (Caporaso et al., 2010b). The total merged sequences were clustered into de novo OTUs with a sequence identity of 97% using UCLUST (Edgar, 2010). The representative sequences were designated using the most abundant sequences within each OTU. The representative sequences were aligned using PyNAST aligner (Caporaso et al., 2010a), and FastTree (Price et al., 2009)

| Statistical analysis
All experiments except bacterial community analysis were conducted twice with at least three replicates. Bacterial community analysis was conducted three times with six replicates. Analysis of variance was conducted using the generalized linear model proce-  (Segata et al., 2011).

DATA AVA I L A B I L I T Y S TAT E M E N T
The sequencing data that support the findings of this study are openly available in the NCBI Bioproject database at https://www.ncbi. conducted in MEGA X. The eukaryotic targeting system is marked in red and the bacterial targeting system is marked in blue FIGURE S2 Confirmation of tssD mutants using PCR. (a) Deletion mutants were generated and PCR was conducted using primers targeting the 5′-upstream and 3′-downstream regions of each tssD region to verify the mutant strains. (b) Agarose gel electrophoresis of the PCR products was performed to distinguish the wild-type and mutant strains based on different fragment lengths. The PCR product size for each mutant strain was smaller than that obtained from wild-type BGR1

FIGURE S3
Growth curve of Burkholderia glumae BGR1 and tssD mutants. Wild-type BGR1, tssD single to quadruple deletion mutants, and tssD complementation strains were grown overnight at 37 °C.
Overnight bacterial cultures were diluted into fresh Luria broth and then incubated at 37 °C. Bacterial growth was monitored using samples withdrawn every 2 hr and measuring the optical density