Characterization of New Virulence Factors Involved in the Intracellular Growth and Survival of Burkholderia pseudomallei

Abstract

Burkholderia pseudomallei, the causative agent of melioidosis, has complex and poorly understood extracellular and intracellular lifestyles. We used transposon-directed insertion site sequencing (TraDIS) to retrospectively analyze a transposon library that had previously been screened through a BALB/c mouse model to identify genes important for growth and survival in vivo. This allowed us to identify the insertion sites and phenotypes of negatively selected mutants that were previously overlooked due to technical constraints. All 23 unique genes identified in the original screen were confirmed by TraDIS, and an additional 105 mutants with various degrees of attenuation in vivo were identified. Five of the newly identified genes were chosen for further characterization, and clean, unmarked bpsl2248, tex, rpiR, bpsl1728, and bpss1528 deletion mutants were constructed from the wild-type strain K96243. Each of these mutants was tested in vitro and in vivo to confirm their attenuated phenotypes and investigate the nature of the attenuation. Our results confirm that we have identified new genes important to in vivo virulence with roles in different stages of B. pseudomallei pathogenesis, including extracellular and intracellular survival. Of particular interest, deletion of the transcription accessory protein Tex was shown to be highly attenuating, and the tex mutant was capable of providing protective immunity against challenge with wild-type B. pseudomallei, suggesting that the genes identified in our TraDIS screen have the potential to be investigated as live vaccine candidates.

INTRODUCTION

Burkholderia pseudomallei is a Gram-negative, motile saprophytic bacterium that is the causative agent of melioidosis. This emerging human pathogen is endemic to the soil and water of tropical areas, including Thailand, Singapore, and northern Australia, and can cause infection through contact with broken skin or through ingestion or inhalation of the bacterium (1). The resulting disease can manifest as a localized skin ulcer or can progress to a systemic infection that is associated with mortality rates as high as 50% in some regions of endemicity (2). There is currently no licensed vaccine against B. pseudomallei available, and it is highly resistant to most antibiotics, severely limiting treatment options (3). Due to the virulent nature of the pathogen, potential for aerosol transmission, and lack of therapeutic options, B. pseudomallei is listed as a Tier 1 bioterrorism threat by the Centers for Disease Control and Prevention (4).

B. pseudomallei is a facultative intracellular pathogen capable of invading and replicating within both epithelial cells and macrophages (5). While B. pseudomallei is capable of extracellular growth and survival and is highly resistant to complement-mediated killing in human sera, intracellular growth is essential for virulence (2, 6). When B. pseudomallei enters the host cell, either through phagocytosis or by inducing its own uptake into nonphagocytic cells, it is able to escape from the phagosome or endocytic vacuole into the cell cytoplasm (7). There, B. pseudomallei is able to exploit the host cell cytoskeleton by inducing actin polymerization at one pole of the bacterium, forming actin comet tails which propel the bacteria through the cytoplasm and forming membrane protrusions into adjacent cells, facilitating cell-to-cell spread (8). Unique among bacterial pathogens that polymerize actin for motility, B. pseudomallei is capable of inducing cell fusion upon contact with neighboring cells, resulting in the formation of multinucleated giant cells (MNGCs) that can contain up to hundreds of nuclei (9).

This complex intracellular lifestyle is regulated by a number of virulence factors encoded within the large, 7.25-megabase B. pseudomallei genome, including three type III secretion systems (T3SS), six type VI secretion systems (T6SS), multiple polysaccharide loci, and a number of secreted effectors (10). The B. pseudomallei polysaccharide capsule and lipopolysaccharide (LPS) help the bacteria survive extracellularly and resist complement deposition (2, 11, 12), while the Bsa T3SS has been implicated in helping B. pseudomallei induce uptake into nonphagocytic cells, escape the vacuole, and resist killing by autophagy (13, 14). In addition, actin polymerization has been shown to be mediated by the autotransporter BimA, which is expressed on one pole of the bacteria and stimulates the formation of new actin filaments (15, 16). Finally, the T6SS-1 is required for cell fusion and the formation of MNGCs (17, 18).

The identification and characterization of these important virulence factors have greatly improved our understanding of B. pseudomallei pathogenesis. However, much remains poorly understood, and the vast majority of B. pseudomallei virulence factors remain to be identified. One technique that has been highly successful at identifying genes that are required for the in vivo virulence of many bacterial species has been the application of large-scale forward genetic screens using libraries of bacterial transposon insertion mutants (19–24). We have previously successfully applied this strategy to the study of B. pseudomallei using an approach known as signature-tagged mutagenesis (STM), in which pools of mutants, each containing a unique tag, are used to infect an animal model (25, 26). By comparing the population of mutants present in infected animals (output pools) to the original pool of mutants used to infect the animals (input pools), it is possible to identify mutants that are unable to survive and grow in vivo. This method identified the B. pseudomallei capsule and the branched-chain amino acid synthase gene ilvE as essential for in vivo survival, which led to the development of an ilvE mutant as a live attenuated vaccine candidate. A number of additional virulence factors have also been identified by this method, the majority of which are predicted to be involved in metabolism and replication (25, 26). However, these studies were constrained by technical limitations regarding library size and lacked the sensitivity to distinguish mild attenuation phenotypes. While microarray technology was used to identify mutants negatively selected in the output pools, the insertion site of each mutant needed to be identified individually using a difficult and time-consuming PCR approach. As a result, only the most strongly attenuated mutants were followed up to determine the gene of interest.

More recently, the development of next-generation sequencing technology has resulted in the development of transposon library sequencing techniques known as transposon-directed insertion site sequencing (TraDIS) and transposon sequencing (Tn-seq), which allow entire libraries to be screened and all insertion sites identified quickly and easily. This technique has been applied to large bacterial libraries to identify every essential gene within the genome and to identify new in vivo virulence factors (27, 28). It can also be retrospectively applied to previously screened STM libraries to identify the insertion sites and phenotypes of mutants that were previously overlooked due to technical constraints, allowing the identification of new virulence factors without undertaking further animal experiments (29). Here we describe the retroactive sequencing of a B. pseudomallei K96243 STM library that we previously screened through an in vivo mouse model (26). Using this improved technique, we were able to identify many new potential virulence factors and overcome biases that had constrained the original screen without the requirement for further animal experiments. Moreover, we were able to identify mutants with intermediate phenotypes that would otherwise have been overlooked. We selected five of these newly identified mutants for additional characterization and created clean unmarked deletion mutants for each gene of interest. We then confirmed the in vivo growth and survival defect identified in our screen and examined the ability of each mutant to enter and replicate within epithelial cells and macrophages and complete the B. pseudomallei intracellular life cycle as well as survive extracellular serum killing. We found that each of these mutants was attenuated to various degrees, confirming that we have identified new genes with important roles in different stages of B. pseudomallei pathogenesis, increasing our understanding of this important human pathogen.

MATERIALS AND METHODS

Bacterial strains and culture conditions.

B. pseudomallei strain K96243, a clinical isolate from Thailand, was used for the construction of the STM library and for each of the individual mutants. Escherichia coli 19851 (pir⁺) was used for direct conjugation in the construction of the STM library, and E. coli MFD pir was used for conjugation in the construction of individual mutants (30). All experiments were performed in Luria-Bertani (LB) broth or agar at 37°C, and E. coli MFD pir cells were also supplemented with 0.3 mM diaminopimelic acid (DAP). When necessary, plates and cultures were supplemented with antibiotics at the following concentrations: 100 μg/ml Zeocin (Life Technologies), 400 μg/ml kanamycin, and 100 μg/ml ampicillin.

Genomic DNA extraction.

Ten milliliters of overnight shaken cultures was spun down at 4,000 rpm in a benchtop centrifuge and resuspended in 10 ml of lysis buffer (100 μg/ml proteinase K, 10 ml NaCl, 20 ml Tris HCl [pH 8], 1 mM EDTA, 0.5% SDS). Three milliliters of sodium perchlorate was added to the solution and incubated for 1 h at room temperature. Genomic DNA was isolated using a phenol-chloroform-isoamyl alcohol extraction (25:24:1), precipitated with ethanol, and spooled into deionized water.

Illumina sequencing.

Approximately 5 μg of genomic DNA from each of the input, lung, and spleen samples was fragmented to ∼300 bp by sonication in a BioRupter. The fragmented DNA was end repaired and A-tailed using the NEBNext DNA library preparation reagent kit for Illumina (NEB). Annealed adapters Ind_Ad_T (ACACTCTTTCCCTACACGACGCTCTTCCGATC*T, where * indicates phosphorothioate) and Ind_Ad_B (pGATCGGAAGAGCGGTTCAGCAGGAATGCCGAGACCGATCTC) were ligated onto the samples. PCR was performed using primers PE_PCR_V3.3 (CAAGCAGAAGACGGCATACGAGATCGGTACACTCTTTCCCTACACGACGCTCTTCCGATC) and MnTn5_P5_3pr_3 (AATGATACGGCGACCACCGAGATCTACACCTAGGCTGCGGCTGCACTTGTG), which include flow cell binding sites. The PCR program used was 2 min at 94°C, 22 cycles of 30 s at 94°C, 20 s at 65°C, and 30 s at 72°C, and 10 min at 72°C. Samples were then size selected to between 200 and 400 bp in a 2% agarose gel made up with 1× Tris-borate-EDTA (TBE) buffer, with purification with a Qiagen gel extraction kit. The final concentrations of the samples were checked with both a BioAnalyzer and quantitative PCR (qPCR). Preparation products were sequenced on an Illumina Hi-Seq 2000 as 36-bp single-end reads. Concentrations of the samples were established using qPCR with the primers Syb_FP5 (ATGATACGGCGACCACCGAG) and Syb_RP7 (CAAGCAGAAGACGGCATACGAG). Preparation products were sequenced on an Illumina Hi-Seq 2000 as 100-bp single-end reads.

Bioinformatic and statistical analysis.

Raw reads that passed quality control filters and contained the transposon were mapped onto the B. pseudomallei K96243 reference genome (version 6) using bowtie (version 2-1.0), allowing for zero mismatches and excluding nonuniquely mapped reads. The SAMtools toolkit (samtools.sourceforge.net) was applied to the alignment files to determine insertion sites and coverage. For differential expression analysis, the coverage values were variance stabilized using an arcsine-root transformation, and log₂ ratios between the input pools and the lung and spleen samples were calculated. Minimum starting values of 200 sequencing reads within the input pool were used to ensure sufficient starting quantities for negative-selection analysis and avoid background. To define negative selection, cutoffs of the lowest 2.5% of log₂ ratios within the spleen pool comparisons and the lowest 5% of ratios within the lung pool comparisons were set based on the mean distribution of the log₂ fold change.

Generation of clean deletion mutants.

Unmarked deletion mutants were constructed as has been previously described using the suicide vector pDM4 (31). Briefly, 600- to 1,000-bp regions flanking each gene of interest were amplified with an XbaI restriction site on the 5′ end and overlapping sequences on the 3′ end of the PCR product using Phusion high-fidelity PCR master mix (Thermo Scientific). The resulting products were then spliced together using splicing by overlapping extension PCR (SOE PCR) to generate a full-length product consisting of the upstream and downstream flanks lacking the target gene. This product was cloned into the intermediate plasmid pGEM-T and then subcloned into pDM4 using XbaI. The resulting mutagenesis construct was introduced into E. coli MFD pir cultured in LB medium containing 0.3 mM DAP and then transferred into B. pseudomallei K96243 by direct mating. Merodiploids containing the integrated plasmid were selected for on LB agar containing 30 μg/ml chloramphenicol and screened using primers designed against the gene of interest. Successful clones were then plated onto high-sucrose agar (10 g/liter tryptone, 5 g/liter yeast extract, 100 g/liter sucrose) and grown for 48 to 72 h at 24°C. Colonies were screened for sensitivity to chloramphenicol due to loss of the pDM4 cassette, as well as by PCR using primers designed against the gene of interest and across the deletion junction. The resulting mutants were confirmed by full-genome sequencing using an Illumina MiSeq sequencer to confirm the loss of pDM4 and the null mutation. The primers used for mutagenesis and screening the resulting clones are listed in Table S2 in the supplemental material.

Mouse infections.

Female BALB/c mice (Charles Rivers Laboratories International, Inc., Kent, United Kingdom) aged between 6 and 8 weeks were used. Mice were housed under specific-pathogen-free conditions with free access to food and water. All animal experiments were performed in accordance with the Animals (Scientific Procedures) Act of 1986 and the local Ethical Review Committee, under animal biohazard containment level 3 (CL3) conditions. For infections, aliquots of B. pseudomallei K96243 mutants were thawed from frozen stocks, diluted to the desired concentration in pyrogen-free saline (PFS), and administered via the intranasal (i.n.) route. A sample of the inoculum was diluted appropriately, plated out on tryptic soy agar (TSA), and incubated overnight at 37°C to confirm the actual inoculation dose. For each infection, mice were anesthetized intraperitoneally (i.p.) with a combination of ketamine (50 mg/kg; Ketalar, Pfizer Itd, Kent, United Kingdom) and xylazine (10 mg/kg; Rompun, Berkshire, United Kingdom) diluted in PFS. Each mouse was weighed, and the volume of anesthetic given was adjusted accordingly. Once mice were anesthetized, the inoculum was administered by slowly pipetting a total of 50 μl into both nostrils. Mice were then held upright for 30 s to ensure that the liquid had passed into the lungs and were monitored until they had fully recovered from the anesthetic. In all cases, mice were checked at least daily for signs of illness and were culled if determined to have reached the humane endpoint specified in the project license.

Tissue culture infections.

A549 human lung epithelial cells were grown in F12-K tissue culture medium supplemented with 10% fetal bovine serum (FBS), and J774 mouse macrophages were cultured in Dulbecco's modified Eagle medium (DMEM) supplemented with 10% FBS. For invasion and intracellular growth assays, 2 × 10⁵ cells were seeded into 24-well tissue culture dishes and allowed to adhere for 16 h. The cells were then washed with phosphate-buffered saline (PBS) and inoculated with 2 × 10⁶ CFU of wild-type B. pseudomallei or one of the TraDIS mutants in 1 ml of DMEM. The infection was allowed to proceed for 1 h, at which point the medium was removed, the cells were washed with PBS, and 1 ml fresh DMEM or F12-K medium containing 200 μg/ml of kanamycin was added to the wells. The cells were incubated with antibiotics for 2 h at 37°C. For invasion assays, the cells were then lysed immediately with 0.1% Triton X-100 and 10-fold dilutions were plated out onto LB agar to determine how many cells were internalized. For intracellular growth assays, the infections were allowed to proceed for 6 to 24 h, at which point the cells were lysed and CFU plated as described.

Immunofluorescence.

J774A mouse macrophages were seeded onto glass coverslips in 6-well tissue cultures plates at a concentration of 10⁵ CFU/ml and infected with Burkholderia strains as described above. At 24 h postinfection, the cells were washed twice with PBS and fixed with 4% paraformaldehyde overnight at 4°C. The fixed cells were then washed again with PBS, permeabilized with 0.5% Triton X-100, and blocked for 1 h at 37°C with 5% FBS. The coverslips were then incubated with a 1:1,000 dilution of monoclonal antibody (MAb) CC6 (47) for 1 h at 37°C, washed 3 times in PBS for 5 min each, and then incubated again with a 1:10,000 dilution of Alexa Fluor 488-conjugated anti-mouse secondary antibody (Molecular Probes) and Alexa Fluor 555-phalloidin conjugate solution (Molecular Probes) for 1 h at 37°C. The cells were then again washed 3 times for 5 min in PBS to remove unbound antibodies and stained with DAPI (4′,6′-diamidino-2-phenylindole) (Molecular Probes) according to the manufacturer's instructions before the coverslips were mounted onto glass slides using DPX mounting medium. Samples were analyzed using a charge-coupled-device (CCD) fluorescence microscope (Axioplan 2 upright microscope).

Serum survival assays.

Wild-type B. pseudomallei and the TraDIS mutant strains were incubated with 30% pooled normal human serum (NHS) or heat-inactivated (HI) serum in PBS at 37°C for 2 h. HI serum was prepared by incubating the NHS at 56°C for 1 h. Following serum exposure, the samples were serially diluted and plated onto LB agar to determine viable bacterial counts.

RESULTS

Identification of novel B. pseudomallei K96243 genes important for growth and survival in vivo.

We previously identified 39 B. pseudomallei mutants that were unable to grow and/or disseminate in an in vivo murine infection model using a signature-tagged mutagenesis (STM) screen (26). Pools of 96 mutants were used to infect BALB/c mice via the intranasal route, and mutants that were negatively selected in lungs and spleens were identified using microarrays directed against the unique tag on each mutant. However, due to the difficulty of identifying the transposon insertion site of each mutant with this method, only the most strongly attenuated mutants as visualized with microarrays were selected to determine the nature of the mutation and verify the attenuated phenotype. We hypothesized that by applying the recently developed TraDIS technique, we could quickly and easily extract additional information regarding B. pseudomallei pathogenesis from the archived bacterial genomic DNA samples from this STM screen without the requirement to undertake additional animal infections. We predicted that this method could identify additional mutations involved in pathogenesis, including those with more subtle effects acting at different stages of infection. To prepare TraDIS libraries, we pooled the archived genomic DNA samples from each input pool to create an input sample representing the entire library. As each original pool of 96 mutants was assayed through two mice, samples from the mice (one mouse from each pool) were combined to produce biological duplicate lung and spleen output pools. We then applied the TraDIS sequencing technique and compared the input and output pools using a fold change analysis. This improved method allowed us to gather information on every individual mutant within the library and determine whether they were negatively selected, positively selected, or unchanged between input and output pools.

To identify mutants that were negatively selected in the mouse lung and spleen samples, we used a previously described quantification method (29). The total number of sequencing reads matched to each gene in the library was converted using an arcsine-root transformation, and log₂ fold change values between input and lung and input and spleen pools were calculated to determine the fitness of each mutant in terms of its ability to colonize within lung tissue and disseminate to and colonize the spleen. To define attenuation, we set a cutoff of the 2.5% most attenuated mutant genes in the spleen and the 5% most attenuated mutant genes in the lung based on the mean distribution of the log₂ fold change. This resulted in a list of 129 mutants that were negatively selected in mouse spleen samples, representing approximately 10% of the library of 1,248 mutants screened (see Table S1 in the supplemental material). Nine of these mutants were also strongly negatively selected in the mouse lung despite being inoculated through an intranasal route, indicating an inability to survive in that tissue. None of the mutants screened in our experiment were positively selected by our statistical cutoffs. The original 39 mutations previously identified using STM mapped to 23 different genes, all of which were also identified as negatively selected by the TraDIS method. The majority of these genes were among the most strongly negatively selected, with 20 of the 23 genes found within our selected cutoff of the top 2.5% of log₂ ratios. The three remaining transposon mutations mapped either to intergenic regions or to sequences that matched more than one gene and thus were unable to be confirmed. Six of the genes identified by STM, wcbC, wcbJ, wcbN, gmhA, aroB, and vacJ, had previously been independently confirmed to be attenuated for growth and survival in individual intranasal infections of BALB/c mice (26; J. Lim et al., unpublished data). This confirms the ability of our screen to identify attenuated B. pseudomallei mutants and serves as further proof of principle of the TraDIS assay.

Confirmation of attenuated TraDIS mutant phenotypes with unmarked deletion mutants.

Five genes identified as negatively selected in the spleen output pools, some of which were also negatively selected in the lungs, were selected for further characterization based on the strength of the phenotype and predicted functional domains. bpsl1527, which encodes the transcription accessory protein Tex, was selected because this mutant demonstrated one of the strongest attenuated phenotypes in both lung and spleen output pools. Tex is required for toxin regulation in Bordetella pertussis and Clostridium perfringens and has been shown to play a role in virulence in Pseudomonas aeruginosa and Streptococcus pneumoniae (32–35). Moreover, the structure of the P. aeruginosa homolog has been solved and shown to bind DNA, suggesting that this gene is likely to function as a transcriptional regulator. Another strongly negatively selected putative transcriptional regulator, RpiR, is encoded by bpsl0629. RpiR has been demonstrated to regulate various virulence factors of Staphylococcus aureus, suggesting that this could be another conserved regulatory gene required for in vivo virulence (36). bpsl1728 and bpss1528 both encode predicted secreted proteins, with bpss1528 encoding a type III secretion system secreted protein and the bpsl1728 product showing homology to a secreted outer membrane porin from Bordetella pertussis. bpsl1728 was also of interest to us for technical reasons because it is present just above the predicted threshold of detection in our input pool, which allowed us to use this gene as an indicator of the sensitivity and accuracy of our TraDIS screen. bpss1528, which encodes the putative type III secretion system effector protein BapA, was selected because although the B. pseudomallei T3SS-3 is known to be required for virulence, in studies with this mutant in a hamster model it did not display any survival phenotype, suggesting that our TraDIS assay may be able to pick up moderately attenuated mutants that would be missed with other screening methods (37). Finally, bpsl2248 was selected for further characterization because it encodes a putative glycosyltransferase that is not associated with any of the previously characterized polysaccharide loci in B. pseudomallei. Our previous STM screen and a number of additional studies have indicated the importance of polysaccharides to B. pseudomallei virulence (26, 38).

To absolutely confirm the attenuated phenotype of each of these mutants and address the possibility of polar effects, clean unmarked deletion mutants were constructed for each gene as described previously (31). Briefly, a suicide plasmid containing a null allele consisting of the upstream and downstream flanking regions of the gene of interest was introduced via homologous recombination with chloramphenicol selection. A second recombination event was then selected for with sacB-mediated counterselection against sucrose sensitivity, and the resulting colonies were screened by PCR for loss of the wild-type allele. Each mutant was then verified by Illumina whole-genome sequencing to confirm the expected deletion of each gene of interest and to ascertain that no secondary mutations had occurred. Three of the genes selected for mutagenesis, bapA, rpiR, and bpss2248, are located within predicted operons, while tex and bspl1728 do not have any downstream genes located within the same reading frame (39). Due to the nature of our mutagenesis strategy, we did not expect to see polar effects from any of the mutants we constructed, including those with mutations in genes within operons. However, to be certain that transcription of downstream genes was not affected by mutagenesis, we performed reverse transcription-PCR (RT-PCR) analysis of each gene within the operons of our genes of interest and the nearest genes to tex and bspl1728, and we found that transcription was not affected for any of the genes tested (see Fig. S1 in the supplemental material).

The resulting deletion mutants, the Δtex, ΔrpiR, Δbspl1728 (Δ1728), Δbpss2248 (Δ2248), and ΔbapA mutants, were each used to infect BALB/c mice via an intranasal route alongside five mice infected with wild-type B. pseudomallei K96243. CFU were plated from the inocula to determine the exact infectious dose, and the infections were allowed to proceed for 48 h to match the time point of the original screen. At this point, lungs and spleens from each mouse were harvested, homogenized, and plated for CFU. Four of the mutants, the Δtex, ΔrpiR, Δ1728, and Δ2248 mutants, demonstrated significantly reduced numbers of CFU in mouse spleens compared to wild-type B. pseudomallei. The Δtex mutants displayed the strongest attenuation within the spleen and also displayed strong attenuation within the lung, consistent with the TraDIS screen predictions. The remaining mutant, the ΔbapA mutant, showed slightly reduced CFU compared to the wild type, but this decrease was not statistically significant (Fig. 1). These results showed that the TraDIS screen not only was able to identify genes important for growth and survival in a mouse model but was able to do so in a semiquantitative manner and predict the relative strength of the phenotype.

FIG 1.

TraDIS mutants show reduced bacterial burdens in infected BALB/c mice. BALB/c mice (n = 5) were infected intranasally with either B. pseudomallei K96243 or the individual deletion mutant indicated. At 48 h postinfection, spleens and lungs were harvested from the infected animals and bacterial loads were determined. Lines indicate the mean and standard error for each sample. Statistical significance was determined using the Mann-Whitney test, with P values indicated above each sample. ns, not significant. Mice were infected with 500 CFU of K96243 or the ΔrpiR mutant (spleen, P = 0.0079) (A), 500 CFU of K96243 or 800 CFU of the ΔbapA mutant (ns) (B), 1 × 10³ CFU of K96243 or the Δ1728 mutant (lungs, P = 0.0317; spleen, P = 0.0159) (C), 1 × 10³ CFU of K96243 or the Δtex mutant (lungs, P = 0.0079; spleen, P = 0.0079) (D), and 2 × 10³ CFU of K96243 or the Δ2248 mutant (lungs, ns; spleen, P = 0.0317) (E).

B. pseudomallei Δtex mutants are highly attenuated and protect against challenge with wild-type B. pseudomallei.

We next tested each TraDIS mutant in a survival assay to determine if the reduced CFU seen in lung and spleen tissues correlated with decreased virulence. Interestingly, the ΔrpiR, Δ1728, Δ2248, and ΔbapA mutants demonstrated survival phenotypes similar to that of wild-type B. pseudomallei at an infectious dose of approximately 10³ CFU despite significant reduction of bacterial CFU in the spleen. This suggests that the sensitivity of our TraDIS assay allowed the identification of mildly attenuated phenotypes below the threshold of attenuation that would lead to a decrease in virulence as defined by survival. Supporting this hypothesis, the mutant with the strongest TraDIS phenotype, the Δtex mutant, showed increased mouse survival compared to that with wild-type bacteria, with over 80% long-term survival (Fig. 2A). To determine if the surviving animals had completely cleared the infection with the Δtex mutant, we plated CFU from four of the remaining mice at 60 days postinfection. We found that all four mice retained Δtex CFU within the spleen, while only half of the mice had CFU above the level of detection within the lungs (Fig. 2B).

FIG 2.

Survival of BALB/c mice following infection with TraDIS mutants. (A) BALB/c mice (n = 5) were infected with 10³ CFU of K96243, the Δ1728 mutant, or the Δtex mutant. The median survival for K96243 was 2.5 days postinfection, while the median survival for the Δ1728 mutant was 3 days postinfection. Four out of 5 mice infected with the Δtex mutant were still alive when the experiment was terminated at 60 days postinfection. The survival of both mutants was statistically significantly different from that of the wild type as determined by the log rank (Mantle-Cox) test, with P values of 0.0449 and 0.0009, respectively. (B) The bacterial load in mice infected with the Δtex mutant was determined in surviving mice at 60 days postinfection. All four mice displayed detectable levels of B. pseudomallei Δtex in the spleen, while only two animals had detectable CFU in the lungs. (C) BALB/c mice were vaccinated with either 10³ CFU of the Δtex mutant or a saline control and challenged intranasally with 10³ CFU of wild-type B. pseudomallei at 5 weeks postvaccination. Survival up to 80 days postchallenge is shown, with the animals vaccinated with the Δtex mutant showing a statistically significantly different mean time to death of 31 days, compared to 6 days for saline-vaccinated animals (P < 0.001). (D) At 80 days postchallenge, surviving mice vaccinated with the Δtex mutant were sacrificed, and lungs and spleens were harvested and plated to determine if the bacteria had been cleared from the animals. All colonies isolated from both organs were determined to be wild-type B. pseudomallei by PCR screening.

Since the Δtex mutant proved to be attenuated in the acute model of infection, we sought to examine whether it is able to confer protection against subsequent challenge with virulent wild-type B. pseudomallei. Five weeks after intranasal challenge with either saline or the Δtex mutant, BALB/c mice were challenged with approximately 1,000 CFU of B. pseudomallei K96243 and survival was monitored. Our data indicate that the Δtex mutant is able to provide protection in the acute model of infection (Fig. 2C), resulting in a significantly increased time to death. Analysis of organ CFU from surviving mice revealed the retention of wild-type bacteria in both the lung and spleen (Fig. 2D) and splenomegaly in a minority of cases (data not shown). However, unlike in the challenge with the Δtex mutant, none of the surviving mice demonstrated retention of the Δtex mutant in the lung or spleen (data not shown).

The Δtex, ΔrpiR, Δ1728, and ΔbapA mutants display decreased intracellular survival but are able to complete the intracellular life cycle.

B. pseudomallei is considered a facultative intracellular pathogen but is highly resistant to killing by human sera and is able to survive and replicate extracellularly. We were interested in determining how large a role, if any, intracellular survival and replication played in the attenuated phenotypes of the TraDIS mutants. As B. pseudomallei is able to induce its own uptake into epithelial cells (9), we first analyzed invasion of A549 human lung epithelial cells by infecting a monolayer of cells with a multiplicity of infection (MOI) of 10 CFU of the Δtex, ΔrpiR, Δ1728, Δ2248, or ΔbapA mutant or wild-type B. pseudomallei K96243 per cell and allowed the infection to proceed for 1 h. The cells were then gently washed, and kanamycin was added to the medium to kill any remaining extracellular bacteria. At 2 h postinfection, the cells were lysed and plated to determine the number of intracellular CFU. Intracellular bacteria were present under every condition tested, and none of the mutants appeared to be internalized differently from wild-type bacteria, suggesting that they do not have defects related to adhesion or invasion of host cells (Fig. 3A).

FIG 3.

Internalization, growth, and survival of TraDIS mutants in cultured cells. (A) The B. pseudomallei TraDIS mutants are all able to induce their own uptake into A549 human lung epithelial cells. Cells were infected at an MOI of 10 for 1 h and then washed and overlaid with 400 μg/ml kanamycin. At 2 h postinfection, cells were lysed and the bacterial loads determined. None of the mutants displayed statistically significantly different numbers of bacterial CFU compared to the wild type as determined by analysis of variance (ANOVA). (B) The B. pseudomallei TraDIS mutants show variable growth and survival in A549 cells. Cells were infected at an MOI of 1, and the infection was allowed to proceed for 18 h. The ΔrpiR, ΔbapA, and Δtex mutants all showed statistically significantly decreased bacterial loads as determined by the Mann-Whitney test, with P values of 0.0005, <0.00001, and <0.00001, respectively. (C) The B. pseudomallei TraDIS mutants show variable growth and survival in J774 murine macrophages. Cells were infected at an MOI of 1, and the infection was allowed to proceed for 16 h. The ΔrpiR, ΔbapA, and Δtex mutants all had statistically significantly decreased bacterial loads as determined by the Mann-Whitney test, with P values of 0.0045, 0.0078, and 0.0002, respectively. Interestingly, in this cell line, the Δ1728 mutant also showed a reduced bacterial load compared to the wild type (P = 0.0019).

We next analyzed whether the TraDIS mutants were able to survive and replicate within A549 lung epithelial cells. We found that at 18 h postinfection, the Δ1728 and Δ2248 mutants replicated to levels similar to that of wild-type B. pseudomallei, while the ΔrpiR and ΔbapA mutants showed reduced intracellular CFU. The most highly attenuated mutant, the Δtex mutant, demonstrated significantly reduced levels of intracellular bacteria, suggesting that this mutant either is killed by intracellular immune responses such as autophagy or is not capable of completing the intracellular life cycle (Fig. 3B). As B. pseudomallei is also capable of replicating within professional phagocytes such as macrophages, we also analyzed intracellular survival within J774 mouse macrophage cells. We found that at 16 h postinfection, all of the mutants with decreased CFU within A549 cells were also attenuated within J774 cells. Interestingly, the Δ1728 mutant, which showed intracellular growth and survival comparable to those of wild-type B. pseudomallei in A549 cells, demonstrated a reduced bacterial load in J774 cells, suggesting susceptibility to innate immune killing mechanisms (Fig. 3C).

To determine whether the attenuated phenotypes of the TraDIS mutants are due to an impaired intracellular life cycle, we analyzed the ability of each mutant to escape from the phagocytic vacuole, polymerize actin to become motile within the host cell cytoplasm, and form multinucleated giant cells (MNGCs) by fusing the infected host cell with neighboring cells. At 6 h postinfection, we found that each of the TraDIS mutants was present in the host cell cytoplasm and could be seen to polymerize actin comet tails that allow the bacteria to extrude out of the host cell (Fig. 4). Moreover, despite the decreased levels of bacteria within the host cells, each mutant was also able to form MNGCs, showing that they are capable of spreading from cell to cell and inducing cell fusion (data not shown). This suggests that none of the TraDIS mutants are blocked at any stage of the intracellular life cycle but rather that they are less capable of surviving intracellularly and/or have a delayed life cycle.

FIG 4.

All B. pseudomallei TraDIS mutants can polymerize actin. J774 murine macrophages were infected with either B. pseudomallei K96243 (A) or the ΔrpiR (B), ΔbapA (C), Δ1728 (D), Δtex (E), or Δ2248 (F) mutant at an MOI of 10. After 4 h, cells were fixed and stained with the CC6 monoclonal antibody against B. pseudomallei LPS (green) and phalloidin (red), which stains actin filaments. Actin comet tails (blue arrows) were visible in all samples, indicating that the B. pseudomallei mutants are capable of entering cells and escaping into the cytoplasm, where they are able to polymerize actin to spread from cell to cell.

The Δ1728 and Δ2248 mutants are sensitive to killing by human sera.

We next tested whether the TraDIS mutants are resistant to killing by human serum. B. pseudomallei has been shown to be highly resistant to complement-mediated killing and complement deposition, and it is capable of surviving within human sera. We found that while the Δtex, ΔrpiR, and ΔbapA mutants are also fully resistant to human sera, both the Δ1728 and Δ2248 mutants show reduced survival in 30% pooled normal human serum (NHS) compared to PBS (Fig. 5). This suggests that extracellular survival may play a role in the attenuation of at least two of the TraDIS mutants and that our TraDIS screen is capable of identifying attenuated mutants with more than one phenotype.

FIG 5.

Sensitivity of B. pseudomallei TraDIS mutants to human serum. B. pseudomallei wild-type K96243 and the ΔrpiR, ΔbapA, Δ1728, Δtex, and Δ2248 mutants (10⁶ CFU) were incubated with either 30% natural human serum (NHS), 30% heat-inactivated NHS, or a PBS control for 2 h at 37°C. While the wild-type, ΔrpiR, ΔbapA, and Δtex strains were resistant to killing by human serum, as has been previously reported for B. pseudomallei K96243, the Δ1728 and Δ2248 strains were both sensitive to complement killing by human serum (P values of 0.0029 and 0.0000056, respectively).

DISCUSSION

TraDIS technology has previously been demonstrated to be useful for mining new data from archived experimental samples. While the microarray-based method used in STM screens relies on hybridization of fluorescent probes and is thus only semiquantitative, TraDIS can quantitate the number of sequencing reads that match to each gene in every pool, allowing a statistical comparison (see Table S1 in the supplemental material) (27, 29). We were able to reanalyze our archived STM samples using TraDIS and identify over 100 new attenuated mutants as well as provide fitness information for every mutant screened without the need for additional animal experiments. This demonstrated the sensitivity and value of the TraDIS technology over other screening techniques and identified novel virulence factors for future characterization. By comparing the TraDIS and STM data, we also noticed that the STM analysis was biased toward identifying genes which were heavily represented in the input pool, while the TraDIS analysis gave us information on every mutant regardless of how abundant it was in the library.

Our TraDIS screening method successfully identified the 23 genes previously determined to be attenuated in our STM screen, providing a proof of principle for the TraDIS method and validating our STM data. In both screens, the majority of the attenuation mutants were negatively selected only in mouse spleens, while a minority were attenuated in both spleens and lungs. This is most likely a consequence of the intranasal route of infection used for these experiments, as dissemination to other tissue types represents a more extreme selection than survival and replication within the tissue that was directly inoculated. In addition to those described in this article, a number of the mutants identified in both our STM and TraDIS screens have since been individually tested and confirmed to be attenuated following the initial screen, which further validates both screening methods. These include mutants with mutations in multiple genes within the bacterial capsule locus which have since been further characterized to clarify their role in capsule biosynthesis (26). In addition, both aroB (bpsl3168) and vacJ (bpsl3147) mutants have been independently confirmed to have delayed mean-time-to-death and decreased CFU phenotypes compared to wild-type B. pseudomallei (J. Lim et al., unpublished data).

Among the genes newly identified as negatively selected by TraDIS were multiple genes that had been previously demonstrated to be involved in virulence in B. pseudomallei and in other species of bacteria. These include the genes flgK and fliN, which are associated with flagellum biosynthesis and function, the tpx gene for thiol peroxidase, which mediates resistance to oxidative stress, and the shikimate dehydrogenase aroE (40-43). A number of metabolic genes and transcriptional regulator genes were also identified, suggesting that B. pseudomallei K96243 must adapt its metabolic functions in an in vivo environment in order to be a successful pathogen. Furthermore, many of the novel B. pseudomallei genes that were identified in our negative-selection screen have been shown to play a role in virulence in the closely related species P. aeruginosa, including the tryptophan synthesis genes trpB, trpE, and trpF as well as the methyltransferase gene hemK (44, 45). Many of the other genes identified were genes encoding hypothetical proteins or genes that have not yet been shown to play a role in bacterial virulence.

A number of the newly identified attenuated mutants have mutations in putative polysaccharide biosynthesis genes, including bpss2167, bpsl1444, and bpss2248. The B. pseudomallei genome carries four large polysaccharide loci, all of which have been demonstrated to play a role in virulence in vivo; these encode the type I O-PS capsule, the type II O-PS LPS, and two additional clusters defined as type III O-PS and type IV O-PS. However, the genes identified in our screen do not belong to any of these clusters, which suggests that the role of polysaccharides in B. pseudomallei infections is even more complex than has been previously described. bpss2167 and bpss2248 both encode predicted glycosyltransferases belonging to glycosyltransferase family 2, but their specific roles are unknown. bpsl1444 shows similarity to the glycotransferase gene waaG. This is notable because many of the other waa genes, which are involved in the biosynthesis and construction of the core sugar of the B. pseudomallei LPS, were identified as essential genes (28). It would be of interest in future experiments to determine if bpsl1444 plays a role in virulence due to being important to the structural integrity of the bacterium or if this phenotype is due to compromised LPS.

It is interesting to note that of the five mutants characterized in this study, all but one displayed some level of intracellular attenuation in at least one cell line. This is not unexpected, as B. pseudomallei is considered to be a facultative intracellular pathogen, but is notable because many of the best-studied B. pseudomallei virulence factors are genes associated with the capsule, LPS, and flagella, which all play a role in extracellular, rather than intracellular, survival (26, 38, 46). Moreover, the majority of B. pseudomallei genes that have been implicated in intracellular growth and survival, such as those for BimA and the Bsa type III secretion system, have been demonstrated to interfere with at least one stage of the intracellular life cycle (13, 16). This suggests that the mutants described here represent a class of virulence factors required for intracellular survival rather than subjugation of the host cell to complete the bacterial life cycle. A similar class of virulence factors was identified in an in vitro screen for B. pseudomallei mutants that failed to form plaques on cell monolayers by Pilatz et al., and it is interesting to note that one of the nine genes identified in their screen, purM, was also identified in our assay (see Table S1 in the supplemental material) (6). It is likely that the TraDIS screen was able to identify this class of mutants in an in vivo model because this technique is capable of following mild attenuation phenotypes that would otherwise be overlooked in screens that focus on animal survival and/or host cell death.

Of the mutants characterized in this work, only the Δtex mutant displayed a degree of attenuation both in vivo and in vitro that is comparable to that of the mutants identified in our original STM screen. It is likely that this mutant was missed in the STM screen only because it is less highly represented in the input pool than the capsule mutants, making the difference between input and output pools less obvious by microarray analysis (see Table S1 in the supplemental material). Tex has been shown to play an important role in virulence in both B. pertussis and S. pneumoniae, but the exact nature of this role appears to differ between species, as Tex regulates toxin expression in B. pertussis but not S. pneumoniae (32, 33). As Tex is predicted to be a transcription factor and has been shown to bind DNA in both S. pneumoniae and P. aeruginosa (33, 34), it will be interesting to determine the transcriptome of this gene in B. pseudomallei and to determine whether Tex regulates toxin expression or other known virulence factors. Moreover, since the protection provided by Δtex mutants is comparable to that provided by other B. pseudomallei mutants that have been investigated as live vaccine candidates, it will be interesting to further investigate the potential of Δtex vaccine candidates (25). Transposon mutant screens have historically been successful at identifying both major virulence factors and potential live vaccine candidates, and the identification of B. pseudomallei Tex demonstrates that TraDIS can be used to identify such genes that may have been missed in previous screening methods, as well as to identify mutants with mild virulence phenotypes that can provide new insight into aspects of bacterial pathogenesis that would otherwise be overlooked.

Funding Statement

This work was partially funded under the grant titled “Transposon mutagenesis and antibiotic development in Burkholderia pseudomallei” from the Defense Science and Technology Laboratories, United Kingdom.