All strains, plasmids, and oligonucleotides are summarized in Table 1. Luria broth (LB; Lennox, 1955) or trypticase soy broth (dialyzed and chelated) (TSBDC; Brett et al., 1997) were used to culture B. pseudomallei strains at 37°C with shaking. E. coli strain TOP10 was used for cloning purposes. Antibiotics were used at the following concentrations: kanamycin (Km: 25 μg/ml), streptomycin (Sm: 100 μg/ml), gentamicin (Gm: 20 μg/ml), and polymyxin B (Pm: 50 μg/ml).

The cloning site of pGSV3-lux was modified to allow for directional cloning of DNA fragments upstream of the luxCDABE operon. A 850-bp fragment was amplified by PCR from pGSV3-lux [pGSV4 ApaI(+)/luxCAatII(−)] and cloned into pCR4. This fragment was excised from pCR4 using EcoRI/AatII, and cloned into the same restriction sites of pGSV3-lux to yield pGSV4. Subsequently, two PCR fragments from the B. pseudomallei genome [5′ara EcoRI(+)/5′ara ApaI(−) and 3′ara NotI(+)/3′ara SpeI(−)] were cloned into pGSV4 as EcoRI/ApaI (812 bp) and NotI/SpeI (867 bp) fragments to yield pGSV7. The tolC promoter was PCR amplified from B. pseudomallei genomic DNA [PtolC(+)/PtolC(−)] and cloned into pGSV7 as an NheI/BamHI fragment. A 7.6-kb EcoRI/KpnI fragment was excised from pGSV7-PtolC and cloned into the same sites of pKAS46 to yield pKAS46-araPtolC-lux, which was transformed into the E. coli strain S17-1. S17-1/ pKAS46-araPtolC-lux was conjugated with B. pseudomallei strains for the insertion of the PtolC-lux construct into the ancestrally deleted arabinose utilization operon in an allelic exchange protocol described elsewhere (Warawa et al., 2009). The DD503::PtolC-lux strain was named JW280, and the capsule mutant JW270::PtolC-lux strain was named JW280Δwcb. Strains JW280 and JW280Δwcb were identified by their ability to produce light, their resistance to streptomycin, and sensitivity to kanamycin.

J774A.1 cells were maintained in DMEM supplemented with 10% heat inactivated fetal bovine serum (Invitrogen) and grown at 37°C with 5% CO₂. For infection studies, J774A.1 cells were seeded at 7.5 × 10⁴ cells per well (100 μl) in a white 96-well plate (Greiner Bio-One) 1 day before infection. B. pseudomallei strains were subcultured 1:25 (v/v) from an LB overnight culture into TSBDC and grown at 37°C with shaking for 3 h. The bacterial concentration was estimated from OD₆₀₀ measurements, bacterial suspensions were appropriately diluted into PBS, and a 5-μl aliquot of B. pseudomallei suspension was used to inoculate each well of J774A.1 cells. Four replicate plates were infected with B. pseudomallei strains JW280 and JW280Δwcb and treated with gentamicin (20 μg/ml final) 1 h after infection. Lux measurements were taken (Victor, Perkin Elmer) from one of the replicate plates at either 3, 5, 7, or 9 h post-infection, then immediately samples were processed for plate enumerations by removing the media, lysing the samples in 0.1% Triton X-100 for 5 min, serially diluting samples in PBS, and enumerating on LB plates.

Murine infection studies were conducted at BSL-3 confinement and were approved by the Rocky Mountain Laboratories Biosafety and Animal Care and Use Committees in accordance with National Institutes of Health guidelines. Intranasal (i.n.) infection of groups of six 8-week-old female BALB/c mice (Charles River Laboratories) was carried out as previously described (Warawa et al., 2009), using lux+ B. pseudomallei strains JW280 and JW280Δwcb. Animals were euthanized at the onset of terminal disease symptoms, thus 50% infectious dose (ID₅₀) values were calculated as opposed to death endpoint based lethal dose (LD₅₀) values. Statistical analysis of survival data was conducted using GraphPad Prism 5 to evaluate significant differences in survival curves by log-rank (Mantel–Cox) test. StatPlus 2008 was used to perform Probit Analysis to estimate ID₅₀ values for JW280 and JW280Δwcb in the SKH1 model.

Tissues were harvested from animals involved in the survival studies. Tissues were fixed in formalin, processed and stained with hematoxylin and eosin (H&E), and the histopathology of lung, liver, and spleen was scored as previously described (Warawa et al., 2009).

Groups of six BALB/c and SKH1 female 8-week-old mice (Charles River Laboratories) were infected i.n. with 30 μl of PBS bacterial suspension containing either 1 × 10⁴ CFU JW280 or 1 × 10⁶ CFU JW280Δwcb. BALB/c mice were shaved to reduce the amount of fur covering their dorsal thoracic cavities. Mice were anesthetized with 2.5% isoflurane and imaged at ∼12 h intervals, with 5 min exposures from the dorsal perspective using an IVIS Lumina (Caliper Life Sciences). Measurements of lung-specific lux activity were taken by selecting regions of interest (ROI; Living Image 3.0, Caliper Life Sciences) of the dorsally viewed chest cavity and area-normalized backgrounds were subtracted. Animals were euthanized at first presentation of terminal disease symptoms, and lungs were harvested for bacterial enumeration, as described previously (Warawa et al., 2009).

Unless otherwise stated, Graph Pad Prism 5 was used to plot data and conduct statistical analysis by unpaired Student’s t-test.

We sought to engineer a strain of B. pseudomallei which constitutively and stably produced light from the Photorhabdus luminescens luxCDABE operon. To ensure stability of the lux operon in the absence of antibiotic selection, the operon was introduced into the chromosome by allelic exchange. The insertion site chosen for insertion of the lux operon was the site of an ancestrally deleted arabinose utilization operon, known to be present in B. thailandensis while only gene fragments are present in the B. pseudomallei genome (Moore et al., 2004). Several promoters were evaluated for their ability to constitutively produce light under a variety of growth conditions. One such promoter, PtolC, was previously demonstrated to constitutively produce high level photon emission in Pseudomonas aeruginosa (Kadurugamuwa et al., 2003). Therefore, the PtolC-lux construct was introduced into the genome of wild type B. pseudomallei strain DD503 to generate a stable Lux-expressing strain JW280. Similarly, the previously characterized capsular polysaccharide operon mutant, JW270, was engineered to produce light resulting in strain JW280Δwcb.

To investigate whether the light production from JW280 and JW280Δwcb correlates with bacterial numbers, B. pseudomallei strains were examined in a cell culture model system. It has been reported that B. pseudomallei survive within cultured macrophages, escape from the phagosome in a type III secretion-dependent mechanism, and replicate in the cytoplasm (Stevens et al., 2002). The capsular polysaccharide was previously found to not be critical for intracellular survival of B. pseudomallei at early time points (Warawa et al., 2009), however capsule mutants may be required for prolonged survival within macrophages (Wikraiphat et al., 2009).In a gentamicin protection assay, light producing B. pseudomallei strains JW280 and JW280Δwcb were found to infect J774A.1 macrophages and proliferate as previously reported (Figure 1A). No significant difference in growth was observed between wild type and capsule mutant at 3, 5, or 7 h post-infection. However, at 9 h post-infection the capsule mutant had fewer organisms in infected J774A.1 macrophages than wild type. This was consistent with previous reports which suggested that the capsule mutant was not as fit as wild type bacteria during late stage infection of macrophages (Wikraiphat et al., 2009). Bioluminescence was measured by plate reader immediately before harvesting samples for bacterial enumeration (counts per second, cps). Changes in both bacterial number and emitted photons were plotted as a function of time for both wild type JW280 (Figure 1B) and the capsule mutant (Figure 1C). Both JW280 and JW280Δwcb produced light consistent with bacterial number from 3 to 7 h post-infection with an average cps/CFU ratio of 2.79 × 10⁻⁴ and 3.02 × 10⁻⁴, respectively. At 9 h post-infection, relative light production increased with JW280 and JW280Δwcb ratios calculated to be 3.07 × 10⁻⁴ and 4.49 × 10⁻⁴ cps/CFU, respectively. Thus, both JW280 and JW280Δwcb produce light constitutively during host–pathogen interaction studies, with the potential for variations in the cps/CFU ratio based on bacterial fitness during these interactions.

We have previously demonstrated that a B. pseudomallei capsule mutant is attenuated 10^1.8 fold when introduced by the i.n. route in the BALB/c mouse model (Warawa et al., 2009). Using both wild type and capsule mutant strains of B. pseudomallei, we examined whether the SKH1 mouse strain represents a viable model for the study of respiratory melioidosis, and whether Lux-producing strains of B. pseudomallei are attenuated. SKH1 mice are an immunocompetent hairless strain that potentially permit enhanced bioluminescent detection due to their lack of fur. A pairwise comparison of lux− and lux+ versions of both wild type (DD503 and JW280) and capsule mutant (JW270 and JW280Δwcb) strains were not significantly different in their virulence in BALB/c mice (Figure 2A, DD503 and JW270 survival curves reproduced as per Warawa et al., 2009). This finding indicates that introduction of the luxCDABE operon into the B. pseudomallei genome did not significantly impact its virulence in a characterized disease model. The Lux-producing wild type and capsule mutant strains were next used to compare the susceptibility of the SKH1 mouse line to melioidosis relative to the previously characterized BALB/c mouse line. Both SKH1 and BALB/c mice were similarly susceptible to respiratory melioidosis when infected with the wild type strain (Figure 2B, left panel). Interestingly, SKH1 mice were more resistant to infection with the capsule mutant than BALB/c mice (Figure 2B, right panel), suggesting that greater resolution may be observed in the SKH1 model when evaluating attenuation of bacterial strains or potentially when studying therapeutic interventions. A dose response evaluation of wild type and capsule mutant strains was next conducted in SKH1 mice to facilitate estimation of the 50% ID₅₀ values. Probit analysis of the survival curve data (Figure 2C) was used to estimate an ID₅₀ of 2.47 × 10³ CFU (95% CI: 3.5 × 10² to 1.72 × 10⁴ CFU) for JW280 and 5.23 × 10⁵ CFU (95% CI: 2.47 × 10⁵ to 1.11 × 10⁶ CFU) for JW280Δwcb. It is noteworthy that infection with 10⁷ CFU of JW280Δwcb produced rapid morbidity suggestive of endotoxicity rather than disease, where the ID₅₀ of JW280Δwcb may be higher than estimated. Thus, the capsule mutant is attenuated greater than 200-fold in the SKH1 model.

Infected SKH1 tissues were harvested for evaluation of histopathological damage. As previously reported in the BALB/c model (Warawa et al., 2009), both the wild type and capsule mutant strains produce significant pathology in the lung and liver at the moribund stage of disease (Figure 3). Extensive lesions are present throughout the lung associated with inflammatory cell recruitment, fibrin deposition, and cellular necrosis. The liver pathology was characterized by multifocal lesions, and as previously reported, the capsule mutant was found to induce fewer and smaller lesions. Pathology was also detected in the nasal cavity (rhinitis), middle ear (otitis media), and brain (meningitis; data not shown). Blind scoring of the tissue histopathology was conducted to confirm that the capsule mutant indeed produced significantly reduced hepatic damage (Figure 4). It was noted, however, that neither the wild type nor capsule mutant strains induced significant pathology in the spleen. This is in contrast to our previous finding of wild type, but not capsule mutant, pathology being detected in the spleen of BALB/c mice. This suggests that B. pseudomallei may not colonize the spleen of SKH1 to the same extent as in BALB/c mice, or that the response to B. pseudomallei in the spleen of SKH1 is altered from the response in BALB/c mice.

In order to investigate the potential for in vivo diagnostic imaging using bioluminescent B. pseudomallei, SKH1 and BALB/c mice were infected i.n. with 10⁴ CFU of JW280 or 10⁶ CFU of JW280Δwcb, and time course in vivo imaging was conducted. Mice were euthanized at the onset of moribund disease and lung, liver, and spleen tissues were collected to enumerate bacterial colonization at key sites of infection. Both the SKH1 and BALB/c mice in the moribund stage of melioidosis were found to be colonized at similar levels by both wild type and capsule mutant strains of B. pseudomallei, with the exception that there were significantly fewer JW280 detected in the late stage infected lung of BALB/c mice relative to SKH1 mice with a 2.5-fold difference in average bacterial colonization (Figure 5). It was noted that the spleen of SKH1 mice was colonized at a similar level as the spleen of BALB/c mice, suggesting that bacterial colonization does not account for the lack of splenic pathology in SKH1 mice (Figure 4). It was also determined that the lung is the most significantly colonized organ in SKH1 infected mice with an average 10^4.2 fold more B. pseudomallei in the lung than the liver, and 10^5.4 fold more B. pseudomallei in the lung than the spleen. In BALB/c mice, the average colonization ratios are 10^2.8 fold lung to liver and 10^4.7 fold lung to spleen.

Images of bioluminescence detection from in vivo diagnostic imaging were collected using an IVIS Lumina system. Figure 6 presents a representative collection of images demonstrating the potential to track developing pulmonary infection in mice over numerous time points in individual mice, where the last image of each panel represents moribund disease. In both mouse models and using both B. pseudomallei strains, in vivo detection of pulmonary infection was successfully achieved. Consistent with the bacterial enumeration data, less light emission is detected in the lung of JW280-infected BALB/c mice at moribund disease relative to JW280 infection of SKH1 mice.

Photon emission was quantified from both the URT and dorsal thoracic cavity in all imaged mice beginning 24 h post-infection. An analysis of the correlation between total emitted light (total flux) from the thoracic cavity and bacterial colonization of the lung was performed. Trend lines were constrained through the y-axis at the level of noise estimated from mock-infected animals. It was determined that BALB/c mice have a much higher level of noise (2.6 ± 0.7 × 10⁴ p/s) than SKH1 mice (3.1 ± 2.1 × 10³ p/s; **p = 0.0064), which was subsequently found to relate to the amount of residual fur present on BALB/c mice (data not shown). In both the BALB/c and SKH1 mouse models, the capsule mutant produced greater light per CFU than the wild type strain in the lung (Figure 7A) as previously observed in cell culture assays (Figure 1), suggesting that the capsule mutant may be experiencing a similar environment in vivo as that modeled in cultured murine macrophages. Using the calculated equations of the trend lines, it was estimated that the level of detection of wild type and capsule mutant bacteria in the lung of SKH1 mice is 5.7 × 10⁶ and 2.1 × 10⁶ CFU, respectively, at one standard deviation of noise above the average level of noise. Similarly, the BALB/c detection levels were estimated to be 2.6 × 10⁷ and 4.4 × 10⁶ CFU for the wild type and capsule mutant strains, respectively.

The total flux was also plotted as a function of time for wild type and capsule mutant infections of BALB/c (Figure 7B) and SKH1 (Figure 7C) mice, and the noise level indicated in each plot. The ability to sensitively detect photon emission from BALB/c mice was reduced relative to SKH1 mice, with temporal plots of total flux often overlapping the estimated noise level (Figure 7B, horizontal bar). In the BALB/c model, both the wild type and capsule mutants exhibited a biphasic expansion in the lung characterized by an initial increase in pulmonary bacterial burden until approximately 40 h post-infection, followed by a decrease in lung-associated bacteria and subsequent second phase expansion beginning at approximately 60 h post-infection (Figure 7B).

In the hairless SKH1 model, the low noise (Figure 7C, x-axis) allows for sensitive detection of both wild type and capsule mutant bacteria at 24 h post-infection, with the potential for detection at much earlier time points. Unlike the BALB/c model, a biphasic expansion is not observed, and instead both wild type and capsule mutant strains B. pseudomallei exhibit an initial lag phase prior to logarithmic expansion in the lung (Figure 7C). Of the four wild type-infected SKH1 mice which became morbidly infected, one mouse had two drops in bacterial colonization before the onset of late stage disease, indicating that individual-to-individual alterations in disease kinetics can be detected using this system. The wild type and capsule mutant strains are associated with an initial low rate expansion in the lung until a titer of approximately 3.2 × 10⁴ organisms is achieved, at which point rapid exponential increase is observed. Statistical analysis was conducted on parsed data sets representing early and late stage expansion for both strains, and slopes from logarithmically transformed data were significantly different when comparing early to late stage colonization rates for each strain, but no intra-strain differences were statistically significant (data not shown).

Mice that survived the pneumonia phase of infection (first 5 days) were also evaluated for fluctuations in total flux as a function of time (Figure 7D). These mice were characterized as having oscillatory changes in total flux throughout the imaging conducted to day 6. Wild type-infected total flux plots trended downward from 100 h post-infection, which was consistent with the inability to culture organisms from the lung at the termination of the experiment at 14 days post-infection. In contrast, the oscillating total flux of the capsule mutant survivors (Figure 7D, right panel) trended upward, and at 14 days post-infection, a bacterial burden of 9.5 × 10³ (±1.52 × 10³) CFU/lung was observed. This data suggests that the capsule mutant potentially establishes a chronic infection in SKH1 mice while survivors of wild type infection effectively clear B. pseudomallei.

A striking feature of the bioluminescence overlay images is the observation of significant colonization of the URT as a result of i.n. infection (Figure 6), as observed previously (Owen et al., 2009). An evaluation of the ratio of nasal cavity total flux was conducted as a function of the level of thoracic cavity total flux in order to estimate the fold increase in number of B. pseudomallei associated preferentially with the nasal cavity. This analysis made use of mice that develop lethal pneumonia and therefore achieve high titers of B. pseudomallei in the lung. At all measured time points, the ratio of nasal cavity flux to thoracic cavity flux was included to account for temporal fluctuations. In both the SKH1 and BALB/c models, the capsule mutant had a significantly higher proportion of bacteria associated with the URT with an average of 4.5-fold higher flux in SKH1 mice and 3.1-fold higher flux in BALB/c mice (Figure 8). These data indicate that the capsule mutant colonizes the nasal cavity at a significantly higher level than wild type B. pseudomallei.