Investigation of Quorum Sensing-Dependent Gene Expression in Burkholderia gladioli BSR3 through RNA-seq Analyses

Abstract

The plant pathogen Burkholderia gladioli, which has a broad host range that includes rice and onion, causes bacterial panicle blight and sheath rot. Based on the complete genome sequence of B. gladioli BSR3 isolated from infected rice sheaths, the genome of B. gladioli BSR3 contains the luxI/luxR family of genes. Members of this family encode N-acyl-homoserine lactone (AHL) quorum sensing (QS) signal synthase and the LuxR-family AHL signal receptor, which are similar to B. glumae BGR1. In B. glumae, QS has been shown to play pivotal roles in many bacterial behaviors. In this study, we compared the QS-dependent gene expression between B. gladioli BSR3 and a QS-defective B. gladioli BSR3 mutant in two different culture states (10 and 24 h after incubation, corresponding to an exponential phase and a stationary phase) using RNA sequencing (RNA-seq). RNA-seq analyses including gene ontology and pathway enrichment revealed that the B. gladioli BSR3 QS system regulates genes related to motility, toxin production, and oxalogenesis, which were previously reported in B. glumae. Moreover, the uncharacterized polyketide biosynthesis is activated by QS, which was not detected in B. glumae. Thus, we observed not only common QS-dependent genes between B. glumae BGR1 and B. gladioli BSR3, but also unique QS-dependent genes in B. gladioli BSR3.

Introduction

Since the bacterial quorum sensing (QS) system through signal molecules was first discovered in Vibrio fischeri [10], there have been numerous investigations of the mechanisms of QS in bacteria [15,38,63]. In general, proteobacteria use N-acyl-homoserine lactone (AHL) families as a signal molecule, and contain gene pairs encoding members of the LuxR-LuxI family, which are AHL signal synthase and LuxR-family AHL signal receptors, respectively. When signal molecules reach threshold levels due to accumulation of the cell population, QS causes bacteria to alter the expression of genes involved in multiple biological processes, such as biofilm formation, motility, toxin production, and changes in metabolic processes [25,36,46,48,58].

Burkholderia is a genus of proteobacteria with members that inhabit diverse ecological niches, such as soil, water, plants, and hospitals. This genus includes some members capable of causing diseases in animals, humans and plants [5]. Owing to their clinical importance, many studies of this genus have focused on human pathogens, including B. cenocepacia complex (Bcc), B. pseudomallei, and B. mallei, which cause lung infections in immunocompromised patients, melioidosis, and glanders, respectively [61]. Previous studies have shown that the AHL QS circuitry is widely present in members of the Burkholderia genus and plays a pivotal role in the pathogenesis and several phenotypes, including motility, biofilm formation, production of siderophore, and proteolytic activities, of some Burkholderia species [9].

Burkholderia gladioli was formally classified as Pseudomonas gladioli, but was later transferred to the Burkholderia genus by Yabuuchi et al. [67]. Since it was first identified in rotten Gladiolus corm in 1913 [54], B. gladioli has been isolated as a major or opportunistic pathogen from other diseased plants, fungi, and humans [56]. This pathogen can attack Gladiolus, onions, irises, and rice, causing partial or whole decay of plants [21,54,66,69]. Yield loss due to diseases caused by biotic stress is one of the major concerns in rice production. B. gladioli has been reported to cause bacterial panicle blight (BPB) and sheath rot in rice, and in many cases, it is isolated from symptomatic rice together with B. glumae [42,43,60]. These phylogenetically close species can produce toxoflavin, a yellow pigment phytotoxin considered to be one of the major virulence factors associated with the occurrence of disease on rice; however, the severity of disease caused by B. glumae is more aggressive [3,13]. Although B. gladioli-related symptoms in rice may have an economic impact due to decreased rice quality [13], the molecular basis for the increased pathogenesis of B. gladioli relative to B. glumae is not well understood.

As described above, these two phytopathogenic Burkholderia contain the AHL QS system [3]. In B. glumae, the AHL QS system is related to toxin production, motility, protection against visible light, lipase secretion, and oxalate biosynthesis [4,7,17,23,24]. Based on the complete genome sequence of rice isolate B. gladioli BSR3, it has two AHL QS systems with similar topology to that of B. glumae BGR1 isolated from rice [3,32,53]. However, the influences of those AHL QS systems on B. gladioli are not yet fully understood.

In this study, we carried out transcriptional profiling of wild-type B. gladioli BSR3 and QS-defective mutant B. gladioli BSR3 through RNA sequencing (RNA-seq) analysis to examine the relationship between AHL QS and physiological traits of B. gladioli. By analyzing two different culture states (incubation for 10 and 24 h in liquid culture) that represent the exponential phase and stationary phase, respectively, we obtained a better understanding of QS-dependent genes. Among 7,411 annotated genes in B. gladioli, 448 genes were identified as differentially expressed genes (DEGs) in the exponential phase, while 1,585 DEGs were observed in the stationary phase. Our analyses, including gene ontology (GO) enrichment and the Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment of these DEGs, revealed that B. gladioli BSR3 exhibits QS-dependent physiological features such as flagella motility and toxin production, which are consistent with well-known QS regulated bacterial behaviors of other Burkholderia, including B. glumae BGR1. Variations in the metabolic pathway that includes inositol degradation and changes in unique secondary metabolites such as uncharacterized polyketide were observed in B. gladioli BSR3, but not in B. glumae BGR1. Overall, our findings will help elucidate the mechanism for regulation by the QS system in B. gladioli BSR3.

 

Materials and Methods

Bacterial Strain, Plasmid, and Culture Conditions

All bacterial strains and plasmids used in this study are listed in Table S1. The B. gladioli strains used included wild-type strain BSR3 isolated from rice sheath [53] and a QS-defective BSR3 mutant. B. gladioli was grown at 28℃ in Luria-Bertani (LB) medium. Escherichia coli DH5α λpir [29] was used to carry the suicide vector, pVIK112 [20], and E. coli S17-1 λpir was used as a pVIK112 vector donor for conjugation experiments [20,55]. All E. coli strains were grown at 37℃ in LB medium, whereas Chromobacterium violaceum CV026 [37] was grown at 28℃ in LB medium. Antibiotics (50 or 100 µg/ml kanamycin, 50 or 100 µg/ml rifampicin, and 10 or 20 µg/ml tetracycline) were added when necessary.

Construction of QS-Defective Mutant and Complementation Strain

General and standard techniques for the construction of recombinant DNA were used in this study [51]. To generate the QS-defective B. gladioli BSR3 mutant, bgla_2g11050 that encodes a LuxI family AHL synthase was disrupted by insertion mutagenesis, as previously described, using a single crossover recombination of recombinant suicide vector pVIK112 [20]. The internal region of bgla_2g11050 was amplified by polymerase chain reaction (PCR) using the corresponding primers (Table S2), and then ligated into the pGEM-T Easy Vector (Promega). After sequencing to confirm ligation, the cloned vector was cleaved by EcoRI and KpnI, ligated into pVIK112, and cleaved by the identical restriction enzymes. Recombinant pVIK112 was transformed into DH5α λpir competent cells, which were then cultured overnight in LB medium with kanamycin (50 µg/ml). After that, the properly cloned pVIK112 plasmids were introduced to S17-1 λpir. Transformed S17-1 λpir was transferred to B. gladioli BSR3 via biparental mating as follows. Donors were washed twice during the mid-logarithmic phase, while recipient cells were washed twice with fresh LB. Next, the donors were mixed at a 1:1 ratio (500 µl), after which samples were centrifuged. The supernatant was then discarded and the pellet was resuspended in 30 µl of LB broth. Next, samples were spotted onto LB plates and incubated at 28℃ overnight. Colonies were then removed and resuspended into 1 ml of LB broth, after which 100 µl of the suspension was plated on LB agar containing rifampicin (100 µg/ml) and kanamycin (100 µg/ml). Colonies that grew on the selective plate were collected and confirmed as the COK94 (BSR3 bgla_2g11050::pCOK94) strain by PCR using an internal sequence of the pVIIK112 vector and upstream sequence of the target gene as primers.

To generate a construct for complementation of the COK94 strain, the upstream region and coding sequences of bgla_2g11050 were cloned into the pRK415 vector [22]. To accomplish this, DNA fragments were amplified from chromosomal DNA using a pair of primers, one of which included the KpnI site (5’-AAGGTACCC CGCTTCGCATTTCAAACGA-3’) and another that contained the EcoRI site (5’-AAAAAGAATTCGTGCAGAACCTCGACCTGAT-3’). After purification, the resultant construct was excised by restriction enzyme cleavage and ligated with cleaved pRK415 (KpnI/EcoRI), generating pSLRK01. The constructed plasmid pSLRK01 was transferred by conjugation from E. coli S17-1 λpir to COK94 as described above, and then plated on LB agar containing kanamycin (100 µg/ml) and tetracycline (20 µg/ml). Colonies that grew on the selective plates were collected and analyzed by colony PCR using primers pRK415_R and a2g_11050_comp_R.

Complementation Analysis

To monitor the capacity of AHL production by the constructed strains, an AHL biosensor system based on C. violaceum CV026 was used. Briefly, 1 ml of overnight culture of C. violaceum CV026 that grew in LB broth at 28℃ was embedded and mixed thoroughly in 50 ml of liquefied LB agar before pouring on plates. Next, 1 ml of supernatants from stationary phase (OD600 > 3.0) cultures of B. gladioli BSR3, COK94, and COK94comp were used to extract autoinducers. The equal volume of ethyl acetate was utilized, and removed by centrifugal evaporation. The residue was reconstituted in 10 µl of dimethyl sulfoxide (DMSO), after which 5 µl of suspension containing extracted autoinducers from each strain was dropped on the plate and incubated at 28℃ for 24 h to detect a purple pigment.

RNA Isolation and Library Preparation for RNA-seq

Total RNA was isolated from the cells collected from the cultures of bacteria. To prepare the samples, the wild-type B. gladioli BSR3 and QS-defective mutant were grown in 2 ml of LB broth at 28℃ with continuous shaking for overnight, after which they were subcultured for 10 or 24 h under the same conditions. Total RNA was then extracted from 4 ml of each bacterial culture using an RNeasy Midi Kit (Qiagen, USA). Residual genomic DNA contamination was subsequently removed using an RNase-Free DNase kit (Qiagen) according to the manufacturer’s protocols. After DNase I treatment, a MICROBExpress bacterial mRNA enrichment kit (Ambion, USA) was used to remove bacterial rRNA from the total RNA sample.

RNA-seq libraries were prepared using an Illumina TruSeq RNA sample prep kit (Illumina, USA) according to the standard protocol. Sequencing was performed on an Illumina HiSeq2000 instrument using the reagents provided in the Illumina TruSeq PE Cluster kits. The data discussed in this publication have been deposited in NCBI’s Gene Expression Omnibus (GEO) and are accessible through GEO Series under the accession number GSE60490.

Analysis of RNA-seq Data

After RNA sequencing, raw reads were mapped to the downloaded reference genome sequences of B. gladioli BSR3 (NCBI BioProject ID PRJNA66301) using the BWA program. SAM files generated by mapping were converted to BAM files of binary format, and then sorted by chromosomal coordinates using the SAMtools program. Mapped reads per annotated gene (total 7,411 genes) were counted by Bam2readcount. The relative transcript abundance was measured in reads per kilobase of exon per million mapped sequence reads (RPKM) [39]. Differentially expressed genes were then identified using the DEGseq package in R language [62].

To analyze DEGs, we conducted a GO enrichment assay using the UniProt (http://www.uniprot.org/) database, and then calculated the p-value. A p-value < 0.01 was considered to indicate significance for the enriched data. For subsequent analysis of the QS-dependent pathway, DEGs were mapped to the KEGG pathway database using the WGET program. The mapped pathway was enriched using a p-value < 0.01.

Quantitative Real-Time PCR Analysis

To confirm the gene expression patterns analyzed from the RNA-seq data, quantitative real-time PCR (qPCR) analysis was performed. The same total RNA samples used to construct the RNA-seq library were used as the template for synthesis of cDNA samples. To generate cDNA, a SuperScript III First-Strand Synthesis System (Invitrogen, USA) was used with 1 µg of each total RNA sample according to the manufacturer’s instructions. The cDNA was then diluted with 190 µl of distilled water and subjected to qPCR using Rotor-gene Q (Qiagen). The reactions were conducted by subjecting the samples to initial heating for 10 min at 95℃, then 40 cycles of 95℃ for 10 sec, 60℃ for 15 sec, and 72℃ for 20 sec. All samples were analyzed in triplicate in 25 µl reaction mixtures containing 5 µl of diluted cDNA, 2× Rotor-gene SYBR Green PCR Master Mix (Qiagen), and primers (Table S3). The fold changes were determined by the 2-∆∆Ct method with the constitutively expressed 16S ribosomal RNA gene as a control for normalization.

Phenotypic Analysis

QS-dependent phenotypes in B. gladioli BSR3 were investigated with wild-type and QS-defective mutant strains. Swarming assays were performed at 28℃ on LB plates containing 0.5% Bacto Agar (BD, Franklin Lakes, NJ, USA). Each strain was inoculated into 2 ml of LB for 24 h with shaking. Next, 1 ml of stationary-phase culture of each strain was centrifuged, washed twice with fresh LB broth, and resuspended in 100 µl of LB broth. Each cell suspension (5 µl) was subsequently dropped onto swarming assay plates and incubated for 48 h. To determine the production of toxoflavin by each strain, bacterial cells were smeared on LB agar plates and incubated for 36 h at 28℃. The production of toxoflavin by each strain was indicated by the yellow pigment of the culture growing on the plates.

 

Results and Discussion

Construction of QS-Defective Mutant of B. gladioli BSR3

To establish a QS-defective B. gladioli BSR3 mutant strain, we disrupted the bgla_2g11050, which encodes an AHL synthase. In B. gladioli BSR3, bgla_2g11050 is a putative ortholog of tofI, a well-known gene encoding AHL synthase of B. glumae BGR1 in the Burkholderia Genome Database, whereas BSR3 has an additional paired AHL synthase (bgla_1p1740) in the genome of the plasmid residing in the polyketide synthesis operon (http://www.burkholderia.com/) [64]. A QS-defective mutant strain, COK94, was constructed, and proper disruption of the target gene was confirmed by PCR verification (data not shown). In addition, the COK94 strain was very defective in AHL production as shown Fig. S1. Taken together, these findings indicate that the QS-defective mutant strain of B. gladioli BSR3 was successfully prepared.

Transcriptional Profiling of B. gladioli BSR3 and QSDefective Mutant

To gain insight into the genome-scale gene expression patterns regulated by the AHL QS system in B. gladioli BSR3, we conducted RNA-seq for the wild-type and QS-defective mutant strains. We acquired the transcriptome of each strain at two different time points (10 and 24 h), representing the exponential phase (OD600 < 1.0; E phase) and stationary phase (OD600 > 3.0; S phase), respectively.

RNA-seq was performed using the Illumina Hiseq 2000 system, and four libraries of B. gladioli BSR3 including wild-type and mutant strains yielded 35,767,138 to 81,669,770 total reads (Table S4). To profile the expression of the genes, we mapped the total reads of the four libraries to the reference genome of B. gladioli BSR3 by the BWA program. Overall, 15,590,730 to 25,635,166 reads were able to be mapped (Table S4).

The expression level of genes was quantified based on the RPKM [39] value. To identify QS-dependent regulated genes, we used the DEGseq package in R language [62], with a fold change (FC) calculated from the value of the wild-type RPKM / mutant RPKM. Consequently, 448 and 1,585 genes were identified as DEGs in the E phase and S phase, respectively. An overview and comparison of these DEGs are provided in Fig. S2.

Validation of RNA-seq Data by qPCR Analysis

To validate the RNA-seq results, qPCR was carried out using six randomly selected genes (Table S3). The relative expression levels of each gene in four RNA samples used for construction of the RNA-seq library were obtained. For comparison with the RNA-seq data, expression levels from qPCR were calculated as log2 fold changes in wild-type relative to mutant samples. As shown in Fig. 1, the results of each technique were compared by scatter plotting, and correlation coefficients of 0.9982 (Fig. 1A) and 0.9893 (Fig. 1C) were obtained, indicating that our analysis of the RNA-seq data was reliable.

Fig. 1.Validation of RNA-seq results using qPCR. Each log2 ratio of fold changes calculated from qPCR was compared with the log2 FC of the RNA-seq data. (A) Correlation of the fold change between RNA-seq (x-axis) and qPCR (y-axis) in the E phase. (B) Log2 fold changes in the E phase measured by qPCR. (C) Correlation of fold changes between RNA-seq (x-axis) and qPCR (y-axis) in the S phase. (D) Log2fold changes in the S phase measured by qPCR. Error bars represent the standard deviation.

Gene Ontology and Pathway Enrichment Analysis of QS-Dependent Genes

To obtain a comprehensive understanding of genes regulated in QS-dependent manners, we classified DEGs by GO enrichment analysis. Overall, 448 DEGs in E phase were assigned to 86 GO biological process terms and enriched in two terms, 158 GO molecular function terms were enriched in six terms, and 15 GO cellular components were enriched in two terms based on a p-value < 0.01 (Fig. 2A). In the S phase, 1,585 DEGs were assigned to 254 GO biological process terms and enriched in six terms, 358 GO molecular function terms and enriched in 10 terms, and 36 cellular components enriched in two terms with the same criteria (p-value < 0.01) (Fig. 2B). Among these, the terms associated with nitrogen metabolism (nitrate reductase activity; nitrate reductase complex) were identified for both conditions. Moreover, the terms connected to the electron transport system were identified under both conditions in a down-regulated manner, such as cellular respiration in the E phase (100% down-regulation), ATP synthesis coupled electron transport (100% down-regulation), NADH dehydrogenase (ubiquinone) activity (54.55% down-regulation), and quinone binding (37.50% down-regulation) in the S phase (Fig. 2). Conversely, terms signifying histidine degradation (histidine catabolic process to glutamate and formamide/formate) and bacterial motility (cilium or flagellum-dependent cell motility; motor activity; bacterialtype flagellum) were only enriched in the S phase with biased regulation (down- or up-regulation) of the genes.

Fig. 2.Gene ontology enrichment grouping of QS-dependent genes. Each bar represents the ratio of the number of DEGs (red, up-regulated DEGs; blue, down-regulated DEGs) belonging to the GO terms shown in the y-axis. (A) Results of analysis with DEGs in the E phase. (B) Results of analysis with DEGs in the S phase.

We also mapped DEGs into the KEGG pathway database and identified significant pathways using a cut-off threshold of p-value < 0.01. The distribution of enriched pathways under both conditions is shown in Fig. 3. In the E phase, a total of nine pathways were observed, among which two were up-regulated and the rest were down-regulated (Fig. 3A). In the S phase, seven up-regulated and four down-regulated pathways were observed. Overall, the results indicated that flagella assembly, starch and sucrose metabolism, and inositol phosphate metabolism genes were up-regulated, whereas histidine metabolism genes were down-regulated (Fig. 3B).

Fig. 3.Enriched QS-dependent pathways. The pathway name based on the KEGG database is shown in the y-axis, and the ratio of the number of DEGs in each pathway is shown in the x-axis. (A) Enriched pathway in the E phase. (B) Enriched pathway in the S phase.

QS Controls Biosynthesis of Toxoflavin by B. gladioli

Toxoflavin, which is synthesized by different bacteria, including B. glumae and B. gladioli, is toxic to various plants, fungi, animals, and bacteria owing to its ability to act as an effective electron carrier [12]. The molecular bases of toxoflavin biosynthesis and transport are well characterized in B. glumae [24,59]. In previous studies, the toxoflavin biosynthesis genes (toxABCDE) and four genes (toxFGHI) related to toxoflavin transport were reported to be regulated by the AHL QS system via activation of toxR and toxJ genes in B. glumae BGR1 [24]. In this study, we observed the differential expression of putative toxABCDE ortholog genes and the putative ortholog gene of the transcriptional regulator toxJ in B. gladioli BSR3 by the AHL QS system (Fig. 4A). In the E phase, these genes were up-regulated 2.77- to 15.41-fold, while they were up-regulated 4.40- to 13.61-fold in the S phase. Moreover, toxoflavin production was not observable in the QS-defective strain compared with wild-type or complemented strains (Fig. 4B). Conversely, putative toxFGHI ortholog genes were only slightly upregulated in the E phase (1.47- to 5.38-fold) and not identified as DEGs, except for a putative toxF ortholog that encodes an additional protein for typical RND-type threecomponent transport systems [24].

Fig. 4.Organization and expression patterns of putative toxoflavin biosynthesis and transport genes in B. gladioli BSR3. (A) Genetic organization of toxABCDE, toxFGHI, and related regulator toxJ and tofR in B. gladioli BSR3. All genes are putative orthologs of B. glumae identified in the Burkholderia genome database (http://www.burkholderia.com/). Gene expression levels are shown as color variation from yellow to red, corresponding to a range of 0.0 (minimum) to 4.0 (maximum). The genes that did not meet the criteria of p < 0.01 are denoted as non-DEG. (B) The toxoflavin production ability of B. gladioli BSR3 (wild-type), COK94 (mutant), and COK94comp (complemented strain). Yellow pigment was observed on the LB agar plate for all samples except for COK94. The photo was taken 36 h after incubation.

Taken together, our findings suggest that the biosynthesis of toxoflavin in B. gladioli BSR3 is regulated by the AHL QS system in both the E and S phases. However, further investigations are needed to identify the regulator genes induced by autoinducers, even at low cell densities.

QS Controls Bacterial Flagella Motility of B. gladioli

GO and KEGG pathway enrichment assays revealed that QS-induced genes belonging to the flagella assembly are highly enriched in the S phase (Figs. 2 and 3). We visualized gene expression patterns of the flagella assembly in both QS onset states using a heatmap (Fig. 5), and the regulation by QS appeared quite different. While all 51 genes were identified as QS-independent or down-regulated (log2 FC = -2.40 to 0.82) in the E phase, over half of the genes were differentially up-regulated in the S phase. These findings are consistent with the regulatory mechanism of the AHL QS system, which is cell density dependent.

Fig. 5.Alteration of flagella motility by QS in B. gladioli BSR3. (A) The gene expression patterns of the flagella assembly that involved genes in the E phase and the S phase were visualized as a heat map using MeV (Multiple Experiment Viewer), which can be downloaded from the public website (http://www.tm4.org/). Each column of the heat map represents the log2 FC value of two conditions with a green-black-red scheme, in which the lower limit is -0.5 and the upper limit is 2.5. (B) The swarming motility test of the B. gladioli BSR3 (wild-type), COK94 (mutant), and COK94comp (complemented strain). COK94 showed loss of swarming ability. The photo was taken 24 h after incubation.

In 2007, Kim et al. [23] reported that the AHL QS system in B. glumae BGR1 induces expression of an IclR-type transcriptional regulator, qsmR, which subsequently activates transcription of the flhDC genes, encoding a master flagella regulator [23]. However, in our analysis, the homolog of qsmR, bgla_1g11630 (identities = 88%), or the flhDC genes in B. gladioli BSR3 were QS-independently expressed. Interestingly, the flhBAFG and fliA genes were all upregulated (log2 FC = 1.04 to 2.11) in a QS-dependent manner during the S phase (Fig. 5). These genes are known to be cotranscribed in B. glumae BGR1 [19], and to have similar organization in both species (Fig. S3). Among these genes, flhF, which encodes a signal recognition particle-like GTPase, is involved in the regulation of flagella formation and motility in many bacteria [6,40,44,50]. Moreover, the regulation of flhF by a QS master regulator in Vibrio vulnificus was recently reported [27].

Based on the results of previous studies and our finding of abolished swarming motility in the QS-defective mutant strain (Fig. 5), the AHL QS system of B. gladioli BSR3 regulates swarming motility by activating the flhBAFG and fliA gene cluster using well-known regulators, such as FlhDC.

QS Controls myo-Inositol Catabolism of B. gladioli

myo-Inositol is the most common stereoisomer form of inositol, a polyol that is characterized as a hydroxylated six-carbon ring and essential for plants as a building block and signaling molecule [28,30,33]. The catabolism of myo-inosoitol has been studied in various bacteria, especially in the gram-positive bacterium Bacillus subtilis, in which its catabolic pathway and regulation are well understood [28]. In B. subtilis, the iolABCDEFGHIJ and iolRS operons are known to be involved in inositol utilization [68].

In the present study, KEGG pathway analysis revealed that QS-activated genes belonging to the inositol phosphate metabolism pathway were highly enriched (43.75%, p-value = 5.25E-05) during the S phase. Most of these genes are associated with myo-inositol catabolism, which yields acetyl-coenzyme A (CoA) (Fig. 6). In B. gladioli BSR3, nine genes are predicted as iol genes (iolA, iolB, iolC, iolD, iolE, and iolG), two of which are predicted as iolG genes and three that are predicted as iolA (mmsA) genes. These putative iol genes show a tendency for QS-independent expression in the E phase, but a switched pattern in the S phase (Fig. 6). Among the three putative iolA genes, only bgla_1g12490 follows the trend of other iol genes. This may imply that blga_1g12490 is more closely related to the myo-inositol catabolic pathway than the other two putative iolA genes.

Fig. 6.QS-dependent inositol catabolic pathway in B. gladioli BSR3. The myo-inositol degradation pathway in B. gladioli BSR3, based on the KEGG database (http://www.kegg.jp/), with minor modifications. The log2 FC of each gene, of which encoded product involved at a specific step of the pathway is visualized as color variation (yellow to red) of arrows with values from a minimum of 0.0 to a maximum of 3.0. E and S represent Log2 FC of E phase and the Log2 FC of S phase, respectively. Abbreviations: a, myo-inositol; b, 2-keto-myo-inositol; c, 3-D-(3,4/5)-trihydroxycyclohexane-1,2-dione; d, 5-deoxy glucuronic acid; e, 2-deoxy-5-keto-D-gluconic acid; f, 2-deoxy-5-keto-D-gluconic acid 6-phosphate; g, dihydroxyacetone phosphate; h, malonate semialdehyde; i, acetyl coenzyme A.

Based on the results of this study, there are two putative reasons for activation of myo-inositol catabolism by the QS system of B. gladioli BSR3. During plant pathogen infections, carbon metabolism in infected tissues changes [2,8,16,26]. Thus, one can be the utilization of an alternative carbon source when there is high cell density, and the other can be to gain the upper hand in competition with host plants by taking up myo-inositol when B. gladioli BSR3 proliferates. However, the utilization of myo-inositol as the sole carbon source and the mechanism of myo-inositol uptake in B. gladioli BSR3 remain unclear.

QS Controls Polyketide Synthesis of B. gladioli

Polyketides are a large family of secondary metabolites that have remarkably diverse structures and functions of medical importance [49]. In bacteria, many species produce polyketides using polyketide synthases (PKSs), which are separated into three types [49].

In the Burkholderia genus, there are several polyketide metabolites, including malleilactone [1], burkholderic acid [14], enacyloxin [35], thailandamide lactone [18], and rhizoxin [47]. To date, there have been several reports of a relationship between the AHL QS system and polyketides, such as upregulation of pyrrolnitrin biosynthesis in Bcc species [52], the requirement for QS during secretion of enacyloxin by B. ambifaria [35], and induced production of thailandamide lactone in a QS mutant strain of B. thailandensis [18].

Although the genes encoding polyketide biosynthesis proteins have not been fully characterized in B. gladioli BSR3, the genome of B. gladioli BSR3 contains a gene cluster (bgla_2g02250 and bgla_2g02270-300), which appears to generate an unusual polyketide in B. mallei and B. pseudomallei [45]. Furthermore, including this operon, the majority of putative polyketide biosynthesis genes are homologous with pks genes that encode constituent subunits of bacillaene synthase in Bacullus subtilis [57]. Therefore, we found a number of pks-related genes that are up-regulated by a QS-dependent manner in B. gladioli BSR3 (Table 1). The regulation is tighter in the E phase than in the S phase, indicating that the requirement for this metabolite is important during the E phase. Moreover, the results of this study suggest that these pks genes located on chromosome 2 have actual functions, because their transcript abundance is dramatically altered and they are well conserved in other pathogenic Burkholderia species, excluding B. glumae BGR1 (data not shown). Taken together, these findings indicate that the biosynthesis of this unknown polyketide, which may resemble bacillaene, is regulated by QS in B. gladioli BSR3, but not in B. glumae BGR1, due to a lack of this operon.

Table 1.aSome gene names are estimated by BLAST match results. bE, E phase; S, S phase.

QS Controls the Glyoxylate Cycle of B. gladioli

The glyoxylate cycle, a variation of the tricarboxylic acid (TCA) cycle, permits the synthesis of anaplerotic and gluconeogenic compounds from acetyl-CoA [11]. In this pathway, the isocitrate lyase and malate synthase, which are encoded in aceA and aceB, respectively, are required [65]. With these two enzymes, the glyoxylate cycle bypasses two oxidative steps in the TCA cycle that evolve CO2 , enabling the net accumulation of C4-dicarboxylic acids (e.g., succinate, malate, and oxaloacetate), which is not possible with the TCA cycle alone [65].

Based on our results, bgla_1g25970 (aceA) is up-regulated under both conditions, but more tightly during the S phase (log2 FC = 1.71; 4.09), while bgla_1g26010 (aceB) is only upregulated in the S phase (log2 FC = 2.91). The finding that QS activates aceAB genes is consistent with previous results of transcriptome analysis of QS regulation in Yersinia pestis [31]. Because microorganisms utilize the glyoxylate cycle for growth when primary carbon sources are not available [34], the regulation of the glyoxylate cycle by QS may be a strategy for proliferation in glucose-depleted environments due to increased cell populations.

This up-regulation may also be connected to QS-dependent oxalate production being an inevitable event in order to counteract base toxicity during the stationary growth phase of three Burkholderia species (B. glumae, B. mallei, and B. thailandensis) [17]. Among these species, oxalogenesis in B. glumae is conducted by two QS-induced genes encoding ObcA and ObcB, using oxaloacetate and acetyl-CoA as substrates [17,41]. Given that the putative ortholog genes of obcAB (bgla_1g20440-50) in B. gladioli BSR3 are highly up-regulated by QS (log2 FC = 6.73; 6.83 in the E phase, and 6.16; 6.23 in the S phase, respectively), QS-activated oxalogenesis using increased oxaloacetate via the glyoxylate cycle could be common in Burkholderia species.