A MarR family transcriptional regulator and subinhibitory antibiotics regulate type VI secretion gene clusters in Burkholderia pseudomallei

Abstract

Burkholderia pseudomallei, the aetiological agent of melioidosis, is an inhabitant of soil and water in many tropical and subtropical regions worldwide. It possesses six distinct type VI secretion systems (T6SS-1 to T6SS-6), but little is known about most of them, as they are poorly expressed in laboratory culture media. A genetic screen was devised to locate a putative repressor of the T6SS-2 gene cluster and a MarR family transcriptional regulator, termed TctR, was identified. The inactivation of tctR resulted in a 50-fold increase in the expression of an hcp2–lacZ transcriptional fusion, indicating that TctR is a negative regulator of the T6SS-2 gene cluster. Surprisingly, the tctR mutation resulted in a significant decrease in the expression of an hcp6–lacZ transcriptional fusion. B. pseudomallei K96243 and a tctR mutant were grown to logarithmic phase in rich culture medium and RNA was isolated and sequenced in order to identify other genes regulated by TctR. The results identified seven gene clusters that were repressed by TctR, including T6SS-2, and three gene clusters that were significantly activated. A small molecule library consisting of 1120 structurally defined compounds was screened to identify a putative ligand (or ligands) that might bind TctR and derepress transcription of the T6SS-2 gene cluster. Seven compounds, six fluoroquinolones and one quinolone, activated the expression of hcp2–lacZ. Subinhibitory ciprofloxacin also increased the expression of the T6SS-3, T6SS-4 and T6SS-6 gene clusters. This study highlights the complex layers of regulatory control that B. pseudomallei utilizes to ensure that T6SS expression only occurs under very defined environmental conditions.

Introduction

Burkholderia pseudomallei is a sapronotic pathogen that can be found in soil and water in many countries between the Tropic of Cancer and the Tropic of Capricorn [1–3]. It is a tier 1 select agent that is responsible for melioidosis, a potentially fatal infection of humans and animals [4, 5]. A recent study predicted that there are ~165 000 worldwide cases of melioidosis per year and that ~89 000 are fatal [3]. Individuals exposed to environmental sources may be infected by inhalation, ingestion or skin inoculation and develop a variety of clinical presentations, but sepsis, pneumonia and/or localized abscesses are the most common.

B. pseudomallei is a facultative intracellular parasite that can survive and replicate in host phagocytic and epithelial cells. It utilizes a large arsenal of virulence determinants to promote its survival in this specialized niche [6]. Several secretion systems are involved in this process, including the cluster 3 type III secretion system (T3SS-3), the type II secretion system (T2SS), the type V secretion system (T5SS) protein BimA and the cluster 1 type VI secretion system (T6SS-1) [6]. T3SS-3 is required for escape from the phagosome into the cytosolic space, where a T2SS effector, TssM deubiquitinase, inhibits the activation of the NF-kB and type I IFN pathways [7–9]. Movement of B. pseudomallei through the host cell cytosol is mediated by BimA, a protein located at one pole of the bacterium that polymerizes actin and leads to the formation of ‘comet tails’ [10, 11]. Finally, the T6SS-1 VgrG tail spike protein fuses the membranes of adjacent host cells and allows bacterial cell-to-cell spread and the formation of multinucleated giant cells (MNGC) [12, 13]. This complex intracellular cycle may shield the pathogen from the host immune response.

Type VI secretion systems (T6SS) are membrane-spanning nanomachines that resemble inverted contractile bacteriophage tails. T6SS inject effector proteins into eukaryotic host cells or bacterial competitors in a contact-dependent manner and provide Gram-negative bacteria with a survival advantage in the host or the environment, respectively [14, 15]. B. pseudomallei harbours six T6SS gene clusters, but only the T6SS-1 gene cluster has been extensively characterized [12, 13, 16]. Two distinct Burkholderia T6SS nomenclature schemes were proposed in 2010 by Burtnick et al. [17] and Schwartz et al. [18], and the former is used exclusively in this communication. T6SS are commonly under tight regulatory control and are only expressed, or activated, when they are needed [19]. Five of the B. pseudomallei T6SS are poorly expressed in rich media, which hinders their study [12]. The T6SS-2, T6SS-3, T6SS-4, T6SS-5 and T6SS-6 gene clusters are not required for virulence in animal models of infection and may be required for contact-dependent competition with other microbes in the environment [12, 18]. The T6SS-6 gene cluster is expressed in rich media and has been studied in Burkholderia thailandensis, a nonpathogenic soil and water saprophyte that is closely related to B. pseudomallei [18, 20–22]. This gene cluster was required for B. thailandensis growth on solid media in the presence of Pseudomonas putida, P. fluorescens and Serratia proteamaculans [18], suggesting that it confers protection against contact-dependent growth inhibition caused by specific environmental bacteria. The main function of this system, however, may be to restrict the growth of intraspecies quorum-sensing ‘cheater’ mutants [21]. The B. thailandensis T6SS-6 exports the amidase effector Tae2, which cleaves peptidoglycan crosslinks and results in bacterial lysis in the absence of the immunity protein Tai2 [22].

We are interested in determining the function of the B. pseudomallei T6SS-2, T6SS-3, T6SS-4 and T6SS-5 gene clusters, but this requires an understanding of how these secretion systems are regulated. Here we developed a genetic screen to identify a transcriptional repressor of the T6SS-2 gene cluster. Genetic inactivation of the MarR family regulator, TctR, allowed constitutive expression of this gene cluster. Expression of this gene cluster was further enhanced by nutrient limitation and subinhibitory antibiotics, including the clinically relevant drugs, ceftazidime and co-trimoxazole. This raises the possibility that improper antibiotic therapy or follow-up could activate bacterial genes with unknown consequences. Interestingly, some of the other T6SS were also differentially regulated by TctR and subinhibitory antibiotics.

Methods

Bacterial strains, plasmids and growth conditions

The bacterial strains and plasmids used in this study are described in Table S1 (available in the online version of this article). Escherichia coli and B. pseudomallei were grown at 37 °C on Lysogeny broth (LB) agar (Lennox formulation), LB broth (Lennox formulation), Davis minimal broth (Sigma-Aldrich), or M9 minimal salts (Difco) containing 2 mM MgSO₄, 0.5 mM CaCl₂ and 0.4 % glucose. When appropriate, antibiotics were added at the following concentrations: 100 µg of ampicillin (Ap), 100 µg of carbenicillin (Cb), 50 µg of trimethoprim, 25 µg of streptomycin (Sm) and 25 µg of kanamycin (Km) ml⁻¹ for E. coli and 25 µg of polymyxin B (Pm) and 500 µg of Km ml⁻¹ for B. pseudomallei. In transposon mutagenesis experiments with B. pseudomallei AI and DDS0518-2, 25 µg of gentamicin (Gm) and 32 µg Km ml⁻¹ were used for selection. For induction studies, isopropyl-ß-D-1- thiogalactopyranoside (IPTG) was added to a final concentration of 0.5 mM. A 20 mg ml⁻¹ stock solution of the chromogenic indicator 5-bromo- 4-chloro-3-indolyl-β-d-galactopyranoside (X-Gal) was prepared in N, N dimethylformamide, and 40 µl was spread onto the surface of the plate medium for blue/white screening in E. cloni 10G chemically competent cells. X-Gal was incorporated into agar media at a final concentration of 60 µg ml⁻¹ for qualitative and semi-quantitative viewing of B. pseudomallei β-galactosidase activity on solid media. All manipulations with B. pseudomallei were carried out in a class II microbiological safety cabinet located in a designated biosafety level 3 (BSL-3) laboratory.

DNA manipulation

Restriction enzymes (Roche Molecular Biochemicals and New England BioLabs), Antarctic phosphatase (New England BioLabs) and T4 DNA ligase (Roche Molecular Biochemicals) were used according to the manufacturer’s instructions. When necessary, the End-It DNA end repair kit (Epicentre) was used to convert 5′ or 3′ protruding ends to blunt-ended DNA. The DNA fragments used in the cloning procedures were excised from agarose gels and purified with a PureLink Quick Gel Extraction kit (Invitrogen). Bacterial genomic DNA was prepared from overnight LB broth cultures with the GenElute Bacterial Genomic DNA kit (Sigma-Aldrich). Plasmids were purified from overnight LB broth cultures by using the Wizard Plus SV Miniprep DNA purification system (Promega).

PCR amplifications

The PCR primers are shown in Table S1. The PCR products were sized and isolated using agarose gel electrophoresis, cloned using the pCR2.1-TOPO TA cloning kit (Life Technologies) and transformed into chemically competent E. cloni 10G (Lucigen). The PCR amplifications were performed in a final reaction volume of 50 or 100 µl containing 1× FailSafe PCR PreMix D (Epicentre), 1.25 U FailSafe PCR enzyme mix (Epicentre), 1 µM PCR primers and approximately 200 ng of genomic DNA. Colony PCR was utilized to screen for B. pseudomallei deletion mutants. Briefly, sucrose-resistant and Km-sensitive colonies were resuspended in 50 µl of water, and 5 µl was added to the PCR rather than purified genomic DNA. PCR cycling was performed using a Mastercycler pro S (Eppendorf) and heated to 97 °C for 5 min. This was followed by 30 cycles of a three-temperature cycling protocol (97 °C for 30 s, 55 °C for 30 s and 72 °C for 1 min) and 1 cycle at 72 °C for 10 min. For PCR products larger than 1 kb, an additional 1 min per kb was added to the extension time.

TnMod-OGm′ mutagenesis and plasmid conjugations

TnMod-OGm′ was delivered to AI and DDS0518-2 via conjugation with SM10 (pTnMod-OGm′) by using a membrane filter mating technique. Briefly, SM10 (pTnMod-OGm′) was inoculated into 3 ml of LB broth containing Km and Gm and grown at 37 °C for 18 to 20 h with shaking (250 r.p.m.). B. pseudomallei was also grown under these conditions, but without antibiotic selection. One hundred microlitres of each saturated culture was added to 3 ml of sterile 10 mM MgSO₄, mixed and filtered through a 0.45 µm pore-size nitrocellulose filter using a 25 mm Swinnex filter apparatus (Millipore). Filters were placed on LB plates supplemented with 10 mM MgSO₄ and incubated for 8 h in a 37 °C incubator. The filters were washed with 6 ml of sterile phosphate-buffered saline (PBS), and 100 µl aliquots were spread onto LB agar plates containing Gm and LB agar plates containing Gm and Km. Gm^r and Gm^r Km^r colonies were identified after 48 h incubation at 37 °C. Plasmids pMo130 and pBHR2 and their derivatives were likewise conjugated to B. pseudomallei by using E. coli S17-1 as the donor strain (Table S1).

Construction of B. pseudomallei mutants

Gene replacement experiments with B. pseudomallei were performed using the sacB-based vector pMo130, as previously described [7, 23, 24]. Recombinant derivatives of pMo130 (Table S1) were electroporated into E. coli S17-1 (12.25 kV cm⁻¹) and conjugated with B. pseudomallei K96243 (or K96243 derivatives) for 8 h. Pm was used to counterselect E. coli S17-1. The optimal conditions for the resolution of the sacB constructs were found to be LB agar lacking NaCl and containing 10 % (wt/vol) sucrose, with incubation at 25 °C for 4 days. B. pseudomallei deletion mutants were identified by colony PCR using the primers flanking the deleted regions of the targeted genes (Table S1). As expected, the PCR products generated from the mutant strains were smaller than those obtained from the wild-type strain.

Transferring the hcp2–lacZYA transcriptional fusion from recombinant pRS551 to bacteriophage λRS45

The promoter region immediately upstream of hcp2 was fused to lacZYA on plasmid pRS551 and the transcriptional fusion was subsequently transferred to bacteriophage λRS45 by homologous recombination in E. coli JM105 [25]. Briefly, JM105 (pRS551-0518) was grown overnight in LB broth with 0.2 % maltose, 10 mM MgSO₄, Km and Cb, and 100 µl of saturated culture was mixed with 100 µl of λRS45 [~200 plaque-forming units (p.f.u.)] and incubated for 20 min at room temperature. The mixture was transferred to 37 °C for 10 min, 4.8 ml of 0.7 % molten top agar was added and the contents were poured onto an LB agar plate. The plate was incubated overnight at 37 °C and three isolated plaques were transferred to 1 ml of SM buffer [5.8 g NaCl, 2.0 g MgSO₄·7 H₂O, 0.1 % gelatin, 50 ml 1 M Tris-HCl (pH 7.4) in 1 l H₂O] using a Pasteur pipette. The phage were eluted from the top agar by constant mixing at room temperature for 5 h and then stored at 4 °C overnight. The phage lysate was filtered with a 0.45 mm Millex HV filter to remove any residual JM105. Simons et al. [25] reported that phage lysates prepared in this manner typically contained recombinant phage at a frequency of >10⁻⁴/total phage.

Isolation of an E. coli lysogen harbouring a single-copy hcp2–lacZYA transcriptional fusion

E. coli JM109 was grown overnight in LB broth with 0.2 % maltose and 10 mM MgSO₄, and 500 µl of saturated culture was pelleted and resuspended in 500 µl of SM. Five hundred microlitres of the recombinant phage lysate (~10⁴ pfu) was added and the mixture was incubated at room temperature for 20 min followed by 37 °C for 10 min. Two millilitres of LB broth was added to the bacteria/phage mixture and it was incubated in a rotary shaker at 250 r.p.m. for 2 h at 37 °C. Aliquots of the culture were spread onto MacConkey plates containing Km and Cb and incubated at 37 °C for 2 d. Two lactose-fermenting colonies were identified and one was designated as DD2 (Table S1).

β-galactosidase assays

B. pseudomallei strains and E. coli DD2 were inoculated into 3 ml of LB broth in 14 ml polypropylene round-bottom tubes and incubated at 37 °C in a rotary shaker (250 r.p.m.). B. pseudomallei strains harbouring pBHR2 and its derivatives were grown in LB broth containing Km, and DD2 (pMLBAD) and DD2(pMLBAD-tctR) were grown in LB broth containing Km, Cb and Tp. Overnight cultures were diluted 1 : 100 in LB broth, and 1 ml aliquots were removed at 5 h and assayed for β-galactosidase activity as described previously [26].

RNA sequencing (RNA-seq) analysis

K96243 and DDL3431 were grown in 4 ml LB broth to logarithmic phase on four separate occasions and RNA was extracted using the Zymo Research ZR Fungal/Bacterial RNA Miniprep kit. RNA-seq was conducted by Ambry Genetics (Aliso Viejo, CA, USA). The raw RPKM data were normalized by removing the rRNA transcripts and genes with three or fewer reads in both strains. The genes with a log₂ ratio ≥2 were considered to be significantly differentially regulated (Table 1), but even genes with a log₂ ratio ≥1.2 exhibited a standard deviation of 2.

Table 1. RNA-seq analysis of differentially regulated gene clusters in K96243 and DDL3431 grown in LB broth*.

Locus tags	DDL3431/K96243 log2 difference†	Predicted function of gene cluster
BPSL0284–BPSL0285	3.0	Chromate transport
BPSL1030–BPSL1032	2.0	Histidine ABC transport system
BPSS0516–BPSS0533	2.4	Type VI secretion system cluster 2 (T6SS-2)
BPSS0654–BPSS0656	4.1	Xaa-Pro dipeptidyl-peptidase/serine carboxypeptidase/hypothetical exported protein
BPSS0975–BPSS0977	2.4	FAD-containing d-sorbitol dehydrogenase small subunit/FAD-dependent oxidoreductase/cytochrome c-like domain protein
BPSS1979–BPSS1980	3.0	Glycoside hydrolase/hypothetical protein
BPSS2091–BPSS2109	2.0	Type VI secretion system cluster 3 (T6SS-3)
BPSL1800–BPSL1801	−3.1	Fimbrial biogenesis outer membrane usher protein/putative type-1 fimbrial protein
BPSL3090–BPSL3091, BPSL3109, BPSL3116	−2.0	Type VI secretion system cluster 6 (T6SS-6)
BPSS0167–BPSS0185	−2.4	Type VI secretion system cluster 4 (T6SS-4)

*Strains were grown to logarithmic phase in LB broth on four separate occasions and RNA was purified and subjected to RNA-seq analysis.

†Values are the mean of all differentially regulated genes within each cluster. Negative values indicate that genes are up-regulated in K96243.

Biolog Phenotype MicroArrays (PMs)

Sets of PM plates 1–20 for K96243 and DDL3431 were prepared in accordance with the Biolog PM procedure for E. coli and other Gram-negative bacteria. Kinetic data were collected every 15 min at 37 °C for 48 h and subsequently analysed using version 1.20.02 of the File Management/Kinetic Plot and Parametric software.

The data generated from the 20 PM plates were evaluated statistically to compare and rank those substrates producing the greatest differences between K96243 and DDL3431. The area under the curve (AUC) parameter was selected, and four different statistical methods were used to analyse the data: the Biolog Phenotype Microarray analysis software and a two-way analysis of variance (ANOVA) model using data normalized to positive control scores, as implemented in SAS systems’ PROC GENMOD procedure. The results of these analyses were in general agreement on the trends suggested by a comparison of the variants’ responses. The analysis using the PM version 1.20.02 of the File Management/Kinetic Plot and Parametric software is presented.

For an analysis using the PM software, DDL3431 served as the test and K96243 served as the reference. Since there was a notable signal in the A1 wells of plates 1–8, the A1 zero function was employed. The difference in the AUC between the two strains was analysed to identify the substrates that elicited the greatest difference in response. In instances where K96243 had a greater response than DDL3431, a threshold of −7000 omnilog units (OU) was used to select the most discriminating substrates. Similarly, a threshold of 7000 omnilog units (OU) was used to discriminate the substrates for which DDL3431 strain had a greater response. These thresholds generated lists of a reasonable size to use in further exploration of functional significance.

Peptide desalting, liquid chromatography–tandem mass spectrometry (LC-MS/MS), data processing and library searching

B. pseudomallei DDL3431 and DDL3431 ∆hcp2 were grown in liquid LB and DM to early logarithmic stage. The samples were centrifuged and the supernatants were filter-sterilized for LC-MS/MS and bioinformatic analysis as described previously [7]. The numerical values in Table S4 are non-normalized, base 10 log-adjusted intensity measurements.

Screening the Prestwick Chemical Compound Library (PCCL) for putative TctR ligands

The PCCL is a collection of 1120 molecules comprising 100 % approved drugs (FDA, EMEA and other agencies) selected for their high chemical and pharmacological diversity. These off-patent drugs are arrayed in 14 96-well plates (80 compounds/plate) and are provided with a database containing the structural and physiochemical properties of all compounds. We screened this library for compounds that could turn K96243 Δhcp2–lacZ blue when grown on LB agar plates with X-gal. Briefly, K96243 Δhcp2–lacZ was grown on LB agar, resuspended in PBS to ~85 % turbidity and spread onto 14 150×15 mm polystyrene petri plates containing LB agar and 60 µg ml⁻¹ X-gal using polyester fibre-tipped applicators (CardinalHealth, Waukegan, IL, USA). The 2 mg ml⁻¹ PCCL was diluted 1 : 500 and arrayed into new 96-well plates prior to screening. The compounds were transferred from the 96-well plates to petri plates using a 96 Solid Pin Multi-Blot Replicator (V&P Scientific, Inc., San Diego, CA, USA). The plates were incubated at 37 °C for 48 h and screened for compounds that turned the bacteria blue.

Etests

B. pseudomallei strains were grown for 24 to 48 h at 37 °C on LB agar and two or three colonies were resuspended in sterile PBS to resemble an 85 % turbidity standard. Sterile polyester fibre-tipped applicators (CardinalHealth) were dipped into the inoculum suspension and excess fluid was removed by swabbing against the inside wall of the tube. The entire surface of an LB agar plate containing X-gal (60 µg ml⁻¹) was swabbed four times by rotating the plate 90 degrees and applying a full swab of inoculum. The lids were removed and the plates were allowed to dry in a class II biological safety cabinet for 5–10 min. Etest strips (bioMérieux) were applied to the inoculated agar surface and incubated for 24 h at 37 °C.

Hamster virulence studies

Female Syrian hamsters (five per group) were infected by the intraperitoneal route with 10¹, 10² and 10³ colony-forming units (c.f.u.) of K96243 and DDL3431. The animals were monitored daily for 17 days and all moribund hamsters were humanely euthanized. The surviving animals from each group were euthanized and Barnard’s exact test P-values were calculated for the equality of survival at 17 days.

Research was conducted under an IACUC-approved protocol in compliance with the Animal Welfare Act, PHS Policy and other Federal statutes and regulations relating to animals and experiments involving animals. The facility where this research was conducted is accredited by the Association for Assessment and Accreditation of Laboratory Animal Care, International and adheres to principles stated in the Guide for the Care and Use of Laboratory Animals, National Research Council, 2011.

Results

Development of a genetic screen to isolate a putative repressor of the T6SS-2 gene cluster

Since the T6SS-2 gene cluster is poorly expressed in LB broth [12], a positive selection strategy was devised to isolate a putative T6SS-2 repressor mutant on LB agar. T6SSs have an essential tail tube component called Hcp and we predicted that the expression of the hcp2 gene would be representative of the majority of genes in the T6SS-2 gene cluster (Fig. 1a). An internal portion of the hcp2 gene was removed and replaced with a promoterless kanamycin resistance (pKm^r) gene in the aminoglycoside-sensitive B. pseudomallei strain AI (Table S1). The resulting strain, DDS0518-2, remained relatively sensitive to Km on LB agar because of the low level of transcription occurring in the Δhcp2-pKm^r allele. The Km minimum inhibitory concentration (MIC), determined using Etest strips on LB agar, was 6–8 µg ml⁻¹ for AI and 8–12 µg ml⁻¹ for DDS0518-2. The plasposon TnMod-OGm′ [27], which harbours an outward-facing gentamicin resistance (Gm^r) gene, was utilized to mutagenize AI and DDS0518-2. The mutagenesis mixtures were spread onto LB agar with Gm or LB agar with Gm and Km. Approximately 37 000 Gm^r mutants were generated for each strain, but no Gm^r Km^r colonies were obtained from the AI mutagenesis mixture. By comparison, 47 Gm^r Km^r colonies were obtained in the DDS0518-2 mutagenesis mixture.

Fig. 1.

(a) Genetic map of the B. pseudomallei K96243 T6SS-2 gene cluster, locus tags BPSS0515–BPSS0533. The location and direction of transcription of genes are represented by arrows. The proteins encoded by genes with black arrows include the haemolysin co-regulated protein (Hcp2) and two valine–glycine repeat protein Gs (VgrG2a and Vgr2b). (b) Genetic map of B. pseudomallei K96243 tctR, locus tag BPSL3431. TnMod-OGm′ insertions in DDS0518-2 resulting in a Km^r phenotype are shown schematically by round-top push pins. A 100 bp scale is shown at the bottom.

The TnMod-OGm′ insertion sites were determined for 46/47 Gm^r Km^r DDS0518-2 mutants and all but 1 of the 46 mutants harboured unique insertion sites. Two insertions occurred immediately upstream of the ribosome-binding site of the pKm^r gene and an additional 33 insertions occurred within 3 812 bp upstream of the pKm^r gene. This region includes the three genes upstream of hcp2, BPSS0515, BPSS0516 and BPSS0517, and 553 bp of the intergenic region upstream of BPSS0515 (Fig. 1a). Eighty-six per cent of the insertions occurring upstream of pKm^r were oriented such that transcriptional read through of the Gm^r gene might result in expression of the pKm^r gene, suggesting that the Km^r phenotype in these mutants was artificially induced by a TnMod-OGm′ polar effect. Single TnMod-OGm′ insertions occurred in BPSS0661, encoding a putative integrase, and in BPSS0112, encoding a TssF-like protein in the T6SS-5 gene cluster [12]. Eight TnMod-OGm′ insertions occurred in BPSL3431, the last annotated gene on chromosome 1 of K96243 (Fig. 1b). This 642 bp gene encodes a putative multiple antibiotic resistance regulator (MarR) family transcription factor [28]. Given its role in repression of the T6SS-2 gene cluster (see below), we have designated this gene as tctR for type VI secretion system cluster two regulator (Fig. 1b).

TctR negatively regulates expression of the T6SS-2 gene cluster

With eight independent insertions in the genetic screen, tctR was the best candidate as a potential negative regulator of the T6SS-2 gene cluster. We deleted an internal 96 bp PstI fragment from tctR in K96243 and created a ΔtctR mutant termed DDL3431 (Table S1). To assess the influence of tctR on expression of the T6SS-2 gene cluster, an internal portion of the hcp2 gene was removed and replaced with a promoterless E. coli lacZ gene (Δhcp2–lacZ) and returned to the chromosomes of K96243 and DDL3431. These strains were grown to mid-logarithmic phase in LB broth and assayed for β-galactosidase activity. As expected, T6SS-2 is poorly expressed in LB broth and K96243 Δhcp2–lacZ only produced 12 Miller units of β-galactosidase activity (Fig. 2a). DDL3431 Δhcp2–lacZ, on the other hand, produced >50 fold more β-galactosidase activity (641 Miller units). This suggests that TctR is a transcriptional repressor of the hcp2 gene. We next attempted to complement this phenotype by cloning tctR into the broad-host-range plasmid pBHR2 under the control of a constitutive promoter. Fig. 2(b) shows that DDL3431 Δhcp2–lacZ harbouring the empty vector produced 133 Miller units, while the vector expressing the wild-type tctR gene produced no detectable β-galactosidase activity. Taken together, these results demonstrate that the tctR deletion mutation does not have a polar effect on a downstream gene (or genes) and that TctR is a negative transcriptional regulator of the hcp2 gene. The RNA sequencing studies presented below also support the notion that the tctR deletion is not polar, as there were no significant differences in the expression of the downstream genes BPSL3430, BPSL3429 and BPSL3428 in K96243 and DDL3431 (Table S2).

Fig. 2.

(a) β-galactosidase production by K96243 Δhcp2–lacZ and DDL3431 Δhcp2–lacZ during mid-logarithmic phase growth in LB broth. (b) β-galactosidase production by DDL3431 Δhcp2–lacZ (pBHR2) and DDL3431 Δhcp2–lacZ (pBHR2-tctR). β-galactosidase was measured spectrophotometrically and the Miller units were assessed as described [26]. Each numerical value for β-galactosidase production is the mean of experiments performed on at least three separate occasions ±standard deviation (bars). Strain comparisons were made using two-sample Student’s t-tests with unequal variances. The asterisks represent statistical significance based on Student’s t-tests (P<0.001).

TctR negatively regulates the expression of a single-copy hcp2–lacZYA transcriptional fusion in E. coli

The 143 bp intergenic region immediately upstream of hcp2 was fused to the promoterless lacZYA operon in E. coli using a recombinant λRS45 prophage [25]. DD2 is a lysogen harbouring the single-copy hcp2–lacZYA fusion on the E. coli chromosome (Table S1). We cloned the tctR gene downstream of the arabinose-inducible P_BAD promoter on the Burkholderia broad-host-range vector pMLBAD [29] and transformed it into DD2. Fig. 3 shows that DD2 with the empty pMLBAD vector produced ~250 Miller units of β-galactosidase activity in the absence of arabinose, indicating that the hcp2 promoter is functional in E. coli. The addition of 2 or 3 % arabinose to the DD2 (pMLBAD) culture media had no statistically significant effect on β-galactosidase activity (Fig. 3). By comparison, the β-galactosidase activity of DD2 (pMLBAD-tctR) was decreased by 34 % in the presence of 2 % arabinose (P=0.0031) and by 43 % in the presence of 3 % arabinose (P=0.0017). The β-galactosidase activity produced by this strain at 2 and 3 % arabinose was not statistically different (P=0.1493), however. The results demonstrate that TctR is able to repress the expression of the hcp2–lacZYA fusion in E. coli and suggests that TctR interacts directly with the promoter region upstream of hcp2.

Fig. 3.

β-galactosidase production by DD2 (pMLBAD) and DD2 (pMLBAD-tctR). The strains were grown in LB broth containing 0 % (left), 2 % (centre) and 3 % (right) l-(+)-arabinose for 5 h at 37 °C. Each numerical value for β-galactosidase production is the mean of experiments performed on at least three separate occasions ±standard deviation (bars). Comparisons between DD2 (pMLBAD-tctR) grown in 0 and 2% arabinose and 0 and 3% arabinose were made using two-sample Student's t-tests with unequal variances. The asterisk represents statistical significance based on Student’s t-tests (*, P<0.01).

Nutrient limitation enhances T6SS-2 expression

Ooi et al. examined the genome-wide transcriptional response of B. pseudomallei to >80 different environmental conditions and found that many genes on chromosome 2 exhibited condition-dependent expression [30]. The T6SS-2 gene cluster (Fig. 1a), BPSS0515–BPSS0533, was found to be optimally expressed at 37 °C in modified M63 minimal medium broth. We examined the expression of the Δhcp2–lacZ transcriptional fusion during B. pseudomallei growth at 37 °C on LB agar with X-gal (rich medium) and on M9 minimal salts agar with 0.4 % glucose and X-gal (minimal medium). Fig. S1 shows that K96243 Δhcp2–lacZ colonies were blue on minimal medium, but exhibited no blue colour on rich medium. As expected, the DDL3431 Δhcp2–lacZ colonies were blue on rich medium, but were darker blue on minimal medium. These strains were grown to mid-logarithmic phase in minimal medium and assayed for β-galactosidase activity. While K96243 Δhcp2–lacZ only produced 12 Miller units of β-galactosidase activity when grown in rich medium (Fig. 2a), it produced 168 Miller units of β-galactosidase activity when grown in minimal medium (Fig. S1c). DDL3431 Δhcp2–lacZ produced 641 Miller units of β-galactosidase activity in rich medium (Fig. 2a) and 683 Miller units of β-galactosidase activity in minimal medium (Fig. S1c). Taken together, these results provide further evidence that the T6SS-2 gene cluster is transcribed under conditions of nutrient limitation. It appears that in the absence of tctR, transcription becomes fully derepressed and results in higher T6SS-2 expression than is achieved during normal signalling and induction under nutrient-limited conditions.

TctR positively regulates expression of the T6SS-6 gene cluster

To assess the influence of tctR on the expression of the other five B. pseudomallei T6SS gene clusters, internal portions of the hcp genes (hcp1, hcp3, hcp4, hcp5 and hcp6) were removed and replaced with the promoterless E. coli lacZ gene and returned to the chromosomes of K96243 and DDL3431. These strains were grown to mid-logarithmic phase in LB broth and assayed for β-galactosidase activity. As expected, there was limited transcription of the hcp genes from the T6SS-1, T6SS-3, T6SS-4 and T6SS-5 gene clusters in LB broth and TctR had no significant influence on the expression of these gene clusters (Fig. 4a).

Fig. 4.

(a) β-galactosidase production by K96243 and DDL3431 harbouring Δhcp1–lacZ, Δhcp3–lacZ, Δhcp4–lacZ and Δhcp5–lacZ transcriptional fusions. (b) β-galactosidase production by K96243 Δhcp1–lacZ (pBHR2), K96243 Δhcp1–lacZ (pBHR2-virAG), K96243 Δhcp6–lacZ and DDL3431 Δhcp6–lacZ. Each numerical value for β-galactosidase production is the mean of experiments performed on at least three separate occasions ±standard deviation (bars). Strain comparisons were made using two-sample Student's t-tests with unequal variances. The asterisks represent statistical significance based on Student’s t-tests (*, P<0.01; **P<0.001).

It is known that hcp1 is activated by the VirAG two-component regulatory system and as an internal control we assessed the β-galactosidase activity of K96243 Δhcp1–lacZ harbouring pBHR2 and pBHR2-virAG [12, 16]. Fig. 4(b) shows that there was a >4000-fold increase in β-galactosidase activity when virAG was expressed in trans, demonstrating that the Δhcp1–lacZ transcriptional fusion is functional and can be activated under the appropriate conditions. We next examined the expression of the T6SS-6 gene cluster, which is the only T6SS in B. pseudomallei that is known to be constitutively expressed in rich medium [12]. K96243 Δhcp6–lacZ produced 2618 Miller units, while DDL3431 Δhcp6–lacZ only produced 1419 Miller units (Fig. 4b). These results were significantly different (P<0.01) and suggest that tctR positively influences transcription of the T6SS-6 gene cluster.

Transcriptional profiling reveals TctR-mediated global expression differences

RNA was isolated from K96243 and DDL3431 grown to logarithmic phase in LB broth in an attempt to identify TctR-modulated transcripts. RNA sequencing (RNA-seq) identified 10 gene clusters that exhibited transcriptional differences (log₂ ratio ≥2), with 7 gene clusters being up-regulated in DDL3431 (TctR repressed loci) and 3 gene clusters being down-regulated in DDL3431 (TctR activated loci). Numerous monocistronic genes were also differentially regulated in DDL3431 as compared to K96243 (Table S2). As expected, 18/19 of the T6SS-2 genes (Fig. 1a) were significantly up-regulated in the ΔtctR mutant. This confirms the data obtained with the Δhcp2–lacZ reporter construct (Fig. 2a) and unequivocally shows that the T6SS-2 gene cluster is repressed, either directly or indirectly, by TctR. Interestingly, 18 of the T6SS-3 genes were also up-regulated in DDL3431 (Table 1). The Δhcp3–lacZ reporter gene was expressed two-fold higher in DDL3431 than in K96243 (Fig. 4a), but this was not a statistically significant difference (P>0.1). Gene clusters encoding putative chromate and histidine transporters, proteases, FAD-dependent enzymes and a glycoside hydrolase were also under negative control by TctR (Table 1).

Several gene clusters, including T6SS-6, T6SS-4 and a type 1 fimbrial locus, were under positive control by TctR (Table 1). Four genes in the T6SS-6 gene cluster were significantly down-regulated in the ΔtctR mutant, which supports the data obtained with the Δhcp6–lacZ reporter construct (Fig. 4b). Eighteen genes in the T6SS-4 gene cluster were also down-regulated in DDL3431, suggesting that this gene cluster is activated by TctR. However, expression of the Δhcp4–lacZ reporter gene was not significantly different in the K96243 and DDL3431 backgrounds (Fig. 4a). Additional studies will be required to confirm the role of TctR as a putative positive regulator of the T6SS-4 gene cluster. Taken together, the RNA-seq studies confirmed the role of TctR as a negative regulator of T6SS-2 and a positive regulator of T6SS-6, and suggest a potential role in regulation of the T6SS-3 and T6SS-4 gene clusters.

Biolog Phenotype MicroArrays (PMs) identify 65 metabolic differences between K96243 and DDL3431

The PM system of Biolog consists of 20 microplates and includes a total of 1920 substrates/chemicals. PM plates are used to characterize strains in their ability to use different compounds as sources of carbon, nitrogen, phosphorus and sulfur, or in their sensitivity to stressful environmental conditions, such as pH extremes or high salt concentrations, and antimicrobial chemicals such as antibiotics, detergents, oxidizing agents and others. DDL3431 exhibited a greater respiration rate than K96243 for 15 of the 1920 substrates, suggesting that it more efficiently metabolized these specific substrates (Table S3). One of these compounds was sodium dichromate, an environmental contaminant commonly present in ground water, soil and industrial effluents [31]. Interestingly, two putative chromate transport genes (BPSL0284–BPSL0285) were identified in the RNA-seq analysis as being significantly up-regulated in DDL3431 (Table 1). The proteins encoded by these genes are predicted to reduce chromate accumulation and are essential for chromate resistance. DDL3431 also grew better than K96243 in the presence of several potentially toxic anions, including sodium metavanadate, sodium orthovanadate, sodium nitrate and sodium selenite. While the molecular mechanism(s) of resistance to these compounds is unknown, the data suggests that the absence of TctR facilitates the growth of B. pseudomallei in the presence of these chemicals.

Based on the Phenotype MicroArray and RNA-seq results above, we determined the MIC of sodium dichromate for K96243 and DDL3431 grown in cation-adjusted (Ca²⁺ and Mg²⁺) Mueller–Hinton broth. Surprisingly, the sodium dichromate MIC for both strains was 4 µg ml⁻¹. This result indicates that there is no correlation between DDL3431’s increased cellular respiration rate in the presence of sodium dichromate and resistance to sodium dichromate. Cell respiration can occur independently of cell growth [32] and the results suggest that DDL3431 is able to respire better than K96243 in the presence of inhibitory concentrations of sodium dichromate.

K96243 exhibited a greater respiration rate than DDL3431 in the presence of 50 substrates/chemicals (Table S3). Many of these compounds are antimicrobial compounds and it appears that tctR is required for optimal respiration in the presence of multiple antibiotics. TctR is a member of the multiple antibiotic resistance regulator family and, like E. coli MarR, it may regulate resistance to multiple antibiotics [33, 34]. Unlike MarR, however, enhanced growth in the presence of these antimicrobial compounds requires a functional TctR. This suggests that TctR may be functioning as an activator rather than a repressor for the B. pseudomallei multiple antibiotic resistance phenotypes associated with the PM.

Secretome analysis of DDL3431 and DDL3431 Δhcp2

Little is known about the function of the B. pseudomallei T6SS-2 and no secreted effectors associated with this system have been identified [12]. The T6SS-2 gene cluster is derepressed in DDL3431 and we attempted to exploit this to identify putative T6SS-2 effectors. Since Hcp proteins are critical for the function of T6SSs, we engineered an Δhcp2 mutation in DDL3431 to serve as a T6SS-2-negative control for these experiments. B. pseudomallei DDL3431 and DDL3431 ∆hcp2 were grown in liquid LB and Davis minimal broth (DM) to early logarithmic stage. The samples were centrifuged and the supernatants were filter-sterilized for liquid chromatography–tandem mass spectrometry (LC-MS/MS) and bioinformatic analysis.

A total of 516 proteins were detected exclusively in the LB and DM broth supernatants of DDL3431 (Table S4). Approximately 14 % of these proteins contained Sec or Tat signal sequences, which are not characteristic of T6SS effectors. Many of these exported proteins were annotated as periplasmic proteins, outer membrane proteins and/or proteins secreted by the T2SS [7]. Many putative housekeeping proteins, such as DNA polymerase I, RecR and histidine-tRNA ligase, were also present in the DDL3431-specific secretome (Table S4). These results indicate that numerous cytoplasmic, periplasmic and membrane proteins were present in the DDL3431 supernatants, but not in the DDL3431 Δhcp2 supernatants. The T6SS-2 proteins BPSS0530 and BPSS0532, putative components of T6SS-2 baseplate and membrane-anchoring complexes, respectively, were also found the DDL3431 supernatants. The baseplate and membrane-anchoring complex proteins are part of the T6SS machinery and are not exported from the bacterial cell in T6SS-dependent fashion [35]. The Hcp tail tube protein and the VgrG tail spike protein, on the other hand, are often exported by functionally active T6SSs [35]. We did not identify Hcp2, VgrG2a or VgrG2b in the DDL3431 secretome, which suggests that the T6SS-2 is not functionally active under the conditions we employed here. We also did not find Hcp2 in the cell pellet or supernatant of DDL3431 grown to exponential or logarithmic phase in LB broth by immunoblotting with Hcp2-specific antisera (data not shown). The results suggest that the majority, if not all, of the DDL3431-specific secretome proteins are exported or released independently of the T6SS-2.

Screening a chemical library for a putative TctR ligand

DNA binding by MarR family repressors is often attenuated in the presence of specific chemical ligands, resulting in the activation of gene expression [28]. We have shown that TctR represses transcription of the T6SS-2 gene cluster, but the ligand or signal that TctR responds to is unknown. The Prestwick Chemical Compound Library (PCCL) is a collection of 1120 molecules comprising 100 % approved drugs (FDA, EMEA and other agencies) selected for their high chemical and pharmacological diversity. We developed a high-throughput plate assay to screen the PCCL for compounds that could turn K96243 Δhcp2–lacZ blue when grown on LB agar with X-gal. Seven fluoroquinolone and quinolone antibiotics (ciprofloxacin, oxolinic acid, flumequine, norfloxacin, ofloxacin, lomefloxacin and enoxacin) turned K96243 Δhcp2–lacZ blue in this screening assay. These molecules produced a zone of bacterial killing that was immediately surrounded by a ring of blue colour (Fig. S2), suggesting that subinhibitory antibiotic concentrations activated transcription of the Δhcp2–lacZ reporter. Surprisingly, two quinolone antibiotics in the PCCL (cinoxacin and piromidic acid) did not turn K96243 Δhcp2–lacZ blue in this assay (Fig. S2). The results demonstrate that multiple fluoroquinolone and quinolone antibiotics activate transcription of the T6SS-2 gene cluster at subinhibitory concentrations and represent candidate ligands for TctR.

Subinhibitory fluoroquinolones activate transcription of four T6SS gene clusters

We next examined the effect of subinhibitory fluoroquinolones using Etest strips. K96243 Δhcp2–lacZ and DDL3431 Δhcp2–lacZ were spread onto LB agar plates containing X-gal and levofloxacin (LE), moxifloxacin (MX), gatifloxacin (GA) and ofloxacin (OF) Etest strips were applied to the inoculated agar surface. There was no noticeable difference in the MIC of the four fluoroquinolones against these two strains (Fig. 5a). As expected, K96243 Δhcp2–lacZ was only blue in the area immediately around the zone of killing for all four fluoroquinolone Etest strips (Fig. 5a). DDL3431 Δhcp2–lacZ, on the other hand, exhibited a blue hue across the entire plate due to the ΔtctR mutation. This strain also exhibited a dark blue colour immediately surrounding the Etest strips (Fig. 5a), indicating elevated expression of the Δhcp2–lacZ reporter construct in the areas of subinhibitory fluoroquinolones. The enhanced blue colour around the Etest strips in the absence of TctR indicates that inactivation of tctR is required for maximal expression of the Δhcp2–lacZ reporter construct in the presence of subinhibitory fluoroquinolones. It appears that in the absence of tctR, transcription becomes fully derepressed and results in higher T6SS-2 expression than is achieved during normal signalling and induction in the presence of subinhibitory fluoroquinolones.

Fig. 5.

(a) Levofloxacin (LE), moxifloxacin (MX), gatifloxacin (GA) and ofloxacin (OF) Etest strips were applied to the inoculated agar surface of LB agar plates containing X-gal. The plates were inoculated with K96243 Δhcp2–lacZ (left) and DDL3431 Δhcp2–lacZ (right) and incubated at 37 °C for 24 h. (b) Ciprofloxacin (CI) Etest strips were applied to the inoculated agar surface of LB agar plates containing X-gal. The plates were inoculated with K96243 Δhcp1–lacZ, K96243 Δhcp2–lacZ, K96243 Δhcp3–lacZ, K96243 Δhcp4–lacZ, K96243 Δhcp5–lacZ and K96243 Δhcp6–lacZ and incubated at 37 °C for 24 h. The Etest strips contain a range of antibiotic concentrations and the minimal inhibitory concentrations (MICs) are where the bottoms of the inhibition zones intersect with the Etest strip.

Ciprofloxacin (CI) Etest strips were utilized to examine whether subinhibitory fluoroquinolones also alter the expression of Δhcp1–lacZ, Δhcp3–lacZ, Δhcp4–lacZ, Δhcp5–lacZ and Δhcp6–lacZ. Similar to Δhcp2–lacZ, transcription of Δhcp3–lacZ, Δhcp4–lacZ and Δhcp6–lacZ was induced in K96243 by subinhibitory ciprofloxacin (Fig. 5b). The expression of Δhcp1–lacZ and Δhcp5–lacZ, on the other hand, was not influenced by subinhibitory ciprofloxacin (Fig. 5b). These findings indicate that subinhibitory fluoroquinolones modulate the expression of four of the six T6SSs in B. pseudomallei.

While subinhibitory ciprofloxacin modulated Δhcp2–lacZ transcription on solid media, we were unable to detect any difference in Δhcp2–lacZ expression in liquid media containing subinhibitory ciprofloxacin (data not shown). A similar outcome was reported for a Salmonella typhimurium reporter library harbouring clones whose expression was modulated on solid media, but not in liquid media [36]. The authors of that study emphasized that sessile and planktonic bacterial cultures have different generation times and growth characteristics and might be expected to respond differently to subinhibitory antibiotics.

Clinically relevant antibiotics from different classes activate the expression of Δhcp2–lacZ

As shown above, antibiotics from the quinolone class can induce the expression of the Δhcp2–lacZ reporter construct (Fig. 5a). We next examined whether antibiotics from other classes could also activate Δhcp2–lacZ transcription. K96243 Δhcp2–lacZ and DDL3431 Δhcp2–lacZ were spread onto LB agar plates containing X-gal and CI, imipenem (IP), ceftazidime (CZ), tetracycline (TC), chloramphenicol (CL), trimethoprim/sulfamethoxazole (TS) and kanamycin (KM) Etest strips were applied to the inoculated agar surface. CI Etest strips served as a control for the quinolone class of antibiotics on all plates. K96243 Δhcp2–lacZ was blue in the area around the zone of killing or growth inhibition for the CI, CZ and TS Etest strips, but no Δhcp2–lacZ expression was detected in the subinhibitory zones of the IP, TC, CL and KM Etest strips (Fig. 6). Two of the antibiotics that induced Δhcp2–lacZ expression, CZ (β-lactam class) and TS (sulfonamide class), are recommended for the treatment of melioidosis patients [37]. As with the fluoroquinolone Etest strips, DDL3431 Δhcp2–lacZ exhibited a darker blue colour than K96243 Δhcp2–lacZ immediately surrounding the CZ and TS Etest strips. Interestingly, subinhibitory IP and KM also activated the expression of Δhcp2–lacZ in DDL3431, but not in K96243 (Fig. 6). The results indicate that multiple classes of antibiotics activate transcription of the T6SS-2 gene cluster at subinhibitory concentrations and expression is accentuated in the absence of TctR.

Fig. 6.

Ciprofloxacin (CI), imipenem (IP), ceftazidime (CZ), tetracycline (TC), chloramphenicol (CL), trimethoprim/sulfamethoxazole (TS) and kanamycin (KM) Etest strips were applied to the inoculated agar surface of LB agar plates containing X-gal. The plates were inoculated with K96243 Δhcp2–lacZ (left) and DDL3431 Δhcp2–lacZ (right) and incubated at 37 °C for 24 h. The Etest strips contain a range of antibiotic concentrations and the minimal inhibitory concentrations (MICs) are where the bottoms of the inhibition zones intersect with the Etest strip.

Hamster virulence studies with K96243 and DDL3431

Syrian hamsters are strikingly sensitive to B. pseudomallei and the 50 % lethal dose is ≤10 c.f.u. [38]. As TctR influences the expression of numerous B. pseudomallei genes (Tables 1 and S2), we examined the relative virulence of K96243 and DDL3431 in hamsters to determine whether the tctR gene was required for virulence. A previous study demonstrated that T6SS-2 and T6SS-6, two TctR-regulated gene clusters, are not required for virulence in the hamster model of acute infection [12]. Groups of five animals were injected by the intraperitoneal route of infection with 10, 100 and 1000 c.f.u. and monitored for signs of morbidity and mortality for 17 days. There was a small, but significant (P<0.05), difference in survival in the animals that received 10 c.f.u. (Fig. S3a). None of the mice infected with 10 c.f.u. of DDL3431 survived past day 4, but 40 % of the K96243 mice survived for 17 days. By comparison, there was no significant difference in survival between K96243- and DDL3431-infected hamsters at 17 days post-infection in the 100 and 1 000 c.f.u. groups (Fig. S3b, c). The results indicate that the ΔtctR mutant may be slightly more virulent in hamsters at low infectious doses, but is indistinguishable from the wild-type at higher infectious doses.

Discussion

The goal of this work was to identify a transcriptional regulator (or regulators) of the B. pseudomallei T6SS-2 gene cluster in order to facilitate the study of this system in the laboratory. Transposon mutagenesis of a reporter strain allowed the identification of eight distinct mutations in a gene encoding a MarR family regulator named TctR (BPSL3431). MarR family proteins are winged helix-turn-helix DNA-binding proteins that commonly function as transcriptional repressors [28, 39]. In many instances, the MarR regulator and the target gene(s) under its control are divergently oriented. Transcription in both directions is repressed when the MarR regulator binds to the intergenic region of the divergently transcribed genes. DNA binding is attenuated when the MarR proteins bind small molecule ligands or when specific cysteines are oxidized, resulting in activated gene expression in response to changing environmental stimuli [28, 39]. The K96243 genome encodes 13 MarR family transcriptional regulators, 8 on chromosome 1 and 5 on chromosome 2 [40]. We are not aware of any studies on B. pseudomallei MarR family proteins, but research has been conducted on two B. thailandensis MarR family proteins [41–43]. The B. thailandensis MftR protein, a BPSL1752 orthologue, represses a divergently transcribed gene encoding a major facilitator transport protein [41, 43]. DNA binding by MftR is attenuated in the presence of urate, xanthine, hypoxanthine and elevated temperature. B. thailandensis BifR, a BPSL0626 orthologue, is a redox-sensitive MarR family protein that regulates biofilm formation [42].

A whole-genome transcriptomics study by Ooi et al. found that optimal expression of the T6SS-2 gene cluster occurred at 37 °C in modified M63 minimal medium broth [30]. We also found that expression of this gene cluster was increased on M9 minimal salts with 0.4 % glucose agar relative to LB agar (Fig. S1). This finding may have important environmental implications, as B. pseudomallei thrives in rice fields that are nutrient-depleted [44]. The increased expression of T6SS-2 in a polymicrobial community with depleted soil nutrients may impart a contact-dependent growth advantage for B. pseudomallei. Further work will be necessary to understand the molecular mechanism by which the T6SS-2 gene cluster is activated by depleted nutrients. We propose that optimal expression of T6SS-2 in the environment will require a specific TctR ligand, such as an exported microbial metabolite, and nutrient-depleted soil (Fig. 7).

Fig. 7.

Regulation of T6SS gene clusters in B. pseudomallei by TctR and environmental cues. Arrows indicate activation of gene expression and the line that ends with an inverted T indicates repression of gene expression. The ‘?’ indicates that the natural ligand (or ligands) for TctR is currently unknown.

As mentioned above, genes regulated by a MarR family regulators are often divergently transcribed from the marR gene itself [28, 39]. The start codon of the tctR gene is 513 bp from the origin of replication on chromosome 1 and no divergently transcribed ‘target’ genes are annotated in this region, suggesting that TctR may be an ‘orphan’ MarR family regulator. In fact, in our genetic screen we found that TctR negatively regulates the T6SS-2 gene cluster on chromosome 2. RNA-seq was conducted with B. pseudomallei wild-type and a ΔtctR mutant in order to identify other genes regulated by this transcription factor. A log₂ ratio ≥2 was used as a cut-off for differentially regulated genes and 57 genes were down-regulated by TctR, while 23 genes were up-regulated (Table S1). The TctR-regulated genes were present on both chromosomes and many were part of large gene clusters, including the T6SS-2 and T6SS-6 gene clusters (Table 1). Unlike the T6SS-2 gene cluster, the T6SS-6 gene cluster was activated by TctR (Table 1 and Fig. 4b). While most MarR family regulators are repressors, some also act as transcriptional activators [28, 39]. Two putative transcriptional regulators were also activated by TctR, a LysR family regulator (BPSL0327) and a CRP-like protein (BPSL0616). We demonstrated that TctR could significantly repress the expression of an hcp2–lacZYA fusion in E. coli (Fig. 3), suggesting that there is a direct interaction of TctR with the hcp2 promoter region. However, it is currently unknown whether other transcription factors also participate in TctR-mediated gene regulation in B. pseudomallei. The results clearly demonstrate that TctR is a global transcriptional regulator that can activate or repress the expression of multiple genes on both chromosomes and the expression differences between K96243 and DDL3431 may be directly responsible for the 65 metabolic differences that were detected using the Biolog PM system (Table S3).

We conducted a secretome analysis on the ΔtctR mutant, DDL3431, in anticipation that expression of the T6SS-2 gene cluster would promote the secretion of protein effectors by this system. Surprisingly, we did not identify Hcp2, VgrG2a or VgrG2b (Fig. 1a) in DDL3431 LB or DM supernatants, even though these genes were highly expressed in this strain (Tables 1 and S2). This finding was surprising because Hcp and VgrG proteins are often present in the supernatants of bacteria that express a functional T6SS [35]. For example, Hcp1 was found in the supernatant of B. pseudomallei when expression of the T6SS-1 gene cluster was activated by VirAG in trans [12]. The absence of Hcp2, VgrG2a or VgrG2b in the DDL3431 LB and DM supernatants suggests that the T6SS-2 is not functional under these conditions. It is known that in Pseudomonas aeruginosa the HSI-1 T6SS is under posttranscriptional and posttranslational control and it is possible that similar mechanisms are regulating the B. pseudomallei T6SS-2 [19]. It is also possible that a specific environmental signal (or signals) or cell-to-cell contact with a bacterial competitor (or competitors) might be required for activation of the T6SS-2 [35].

We identified 516 cytoplasmic, periplasmic and membrane proteins that were present in the DDL3431 secretome, but not in the DDL3431 Δhcp2 secretome (Table S4). However, these proteins are unlikely to be effectors of the T6SS-2 because this system appears to be inactive under these growth conditions (see above). The presence of these proteins exclusively in the secretome of DDL3431 could be the result of membrane leakage due to the overexpression of T6SS-2 proteins and the formation of multiple membrane-spanning T6SS-2 extended complexes that cannot be activated under the conditions employed here [35]. The extended complexes may not be stable, or may not form, in DDL3431 Δhcp2 due to the absence of Hcp2. In support of this notion, putative components of T6SS-2 baseplate and membrane anchoring complex were found in the DDL3431 secretome, but not the DDL3431 Δhcp2 secretome. Further work will be necessary to identify the optimal conditions required to promote the secretion of T6SS-2 effectors.

The PCCL is a collection of 1120 molecules comprising 100 % approved drugs (FDA, EMEA and other agencies) selected for their high chemical and pharmacological diversity. A high-throughput screen of the PCCL identified seven compounds that turned K96243 Δhcp2–lacZ blue immediately surrounding a zone of killing on LB agar plates with X-gal (Fig. S2). These compounds were members of the quinolone class of antibiotics and all induced the expression of the Δhcp2–lacZ reporter construct at subinhibitory concentrations. These results were confirmed, and extended, using a variety of fluoroquinolone Etest strips (Fig. 5). Interestingly, subinhibitory concentrations of ciprofloxacin also activated the expression of the Δhcp3–lacZ, Δhcp4–lacZ and Δhcp6–lacZ reporter constructs. Fig. 6 shows that antibiotics from other classes also activate the Δhcp2–lacZ reporter construct at subinhibitory concentrations, including ceftazidime (β-lactam class) and trimethoprim/sulfamethoxazole (sulfonamide class).

Small organic microbial metabolites, such as antibiotics, abound in the environment and are produced largely by bacteria and fungi [45, 46]. While antibiotics can be therapeutically beneficial when given to patients at high doses, the natural concentration of these secondary metabolites in the environment is relatively low. Multiple studies have shown that subinhibitory antibiotics from multiple classes can serve as signalling molecules and can activate or repress the transcription of a large number of bacterial genes [36, 45–47]. For example, ~5 % of the clones in a Salmonella typhimurium promoter probe library were modulated by subinhibitory erythromycin and rifampicin [36]. Subinhibitory trimethoprim induced the production of over 100 secondary metabolites in B. thailandensis [48] and subinhibitory kanamycin induced biofilm formation and the expression of the HIS-1 T6SS gene cluster in P. aeruginosa [47]. Similarly, we demonstrated in this study that subinhibitory antibiotics induced the expression of four of the six B. pseudomallei T6SS gene clusters. The T6SS-2 was induced by multiple fluoroquinolones and by two clinically relevant antibiotics, ceftazidime and trimethoprim/sulfamethoxazole. The therapeutic dose of these antibiotics could drop to subinhibitory levels in melioidosis patients that receive improper therapy or do not adhere strictly to long-term oral eradication therapy [49, 50], which could result in an unexpected global transcriptional response in the latent bacteria that are not eliminated by drug therapy. Further studies are warranted on the global B. pseudomallei transcriptional response to subinhibitory antibiotics and on the molecular mechanism(s) by which multiple classes of antibiotics regulate the transcription bacterial genes. Fig. 7 shows a summary of the results presented in this study and demonstrates that the regulation of the B. pseudomallei T6SS gene clusters, particularly T6SS-2, is complex and highly dependent on environmental cues.