Female Balb/c mice (Charles River, UK) were housed with access to food and water ad libitum at ACDP (Advisory Committee on Dangerous Pathogens) biological safety level (BSL)-3. Procedures were performed in accordance with the Scientific Procedures (Animals) Act 1986 and the Codes of Practice for the Housing and Care of Animals Used in Scientific Procedures, 1989. All mice were 6–8 weeks old and weighing 20–25 g upon commencement of aerosol exposures.

B. pseudomallei was obtained from the Defence Science and Technology Laboratory culture collection (Dstl Porton Down, Salisbury, Wiltshire, UK). Stock cultures were maintained by at −80°C in nutrient broth containing 10% (vol/vol) glycerol. Enumeration was routinely performed on Columbia Blood agar plates at 37°C for 24 h. For aerosol exposures, nutrient broth cultures were shaken at 120 revolutions min⁻¹ for 24 h at 37°C. The required dilutions were prepared in nutrient broth in 10 ml volumes immediately prior to challenge.

Groups of 10 mice were nose-only exposed for a period of 10 min to aerosols generated by the Collison nebuliser or flow-focusing aerosol generator (FFAG) according to the methodology previously described (Thomas et al., 2008, 2009). Aerosol samples from the challenges involving the Collison nebuliser were collected for 1 min into an all-glass impinger (AGI-30; Ace Glass Inc., NJ) containing 10 ml of phosphate buffered saline (PBS) at a flow rate of 12 l min⁻¹. The impinger samples were serially diluted and plated onto Columbia blood agar for enumeration. All exposures were performed within a rigid unpolymerized polyvinylchloride (PVCu) half-suit isolator at BSL-3. The mass median aerosol diameters (MMAD) of the particles were 1–3 μm and 12 μm, respectively, for the Collison nebuliser and the FFAG. The mean calculated inhaled dose for small particles was determined using the impinger data and Guyton's formula (Guyton, 1947). The actual mean inhaled dose for small or large particles was derived from physical washing or dissection and homogenization of the nasal passages, trachea, and lungs as described below.

At specific time-points, five mice were culled by intraperitoneal administration of 0.5 ml sodium pentobarbital (0 h) or halothane intoxication (remaining time-points) and lungs, mediastinal lymph node, Peyer's patches, gastrointestinal lymph nodes (mesenteric, jejunal, colonic), trachea, oesophagus, stomach, intestine, liver, and kidneys aseptically collected. Blood was collected by cardiac puncture after halothane anaesthetization into heparin coated plasma tubes (BD Biosciences, UK). Bacteria deposited in the nasal passages (nasal washings) were collected by inserting a catheter (outer and inner diameter 1.02 and 0.58 mm, respectively; Harvard Apparatus) into a tracheal incision and flushing with 1 ml of PBS containing final concentrations of 1 mM NaOH and 0.1% (v/v) Triton-X100. Tissue samples were homogenized in PBS (Gibco®, Invitrogen Ltd., UK). All samples were enumerated by serial dilution and plating of 100 μ l aliquots in triplicate onto Columbia blood agar plates.

Groups of 10 mice were exposed nose-only for 10 min to aerosolized B. pseudomallei K96243 at five different 10-fold dilutions generated by either the Collison nebuliser or the FFAG decreasing from the “neat” suspension of 10⁹ cfu ml⁻¹. Mice were monitored for 21 d post-exposure for signs of infection. Mice showing signs of infection were culled by cervical dislocation at defined humane end-points. The MLD was calculated by the 50% end-point method (Reed and Muench, 1938), based on the number of bacteria retained in the lungs or nasal passages of animals that received the neat suspension. Immediately after exposure, five animals per dose were sacrificed for necropsy with removal of the lungs, trachea, oesophagus, and stomach. The nasal passages were washed as previously described. The number of bacteria retained in the lungs or nasal passages was demonstrated to decrease by one logarithm with each logarithmic dilution of the spray suspension used in the dilution series.

The mice were dissected to expose the abdominal and thoracic cavities. The lungs were inflated by instillation of 1.5 mL 10% (v/v) neutral-buffered formalin (NBF) to provide a pressure of 25–30 cm³ H₂O. The whole mouse was fixed in 10% (v/v) NBF for 6 d with replacement with a fresh aliquot after 3 d. After a further 3 d fixation, samples from a range of tissues including lungs (apical, medium, and caudal lobes), lymph nodes (mediastinal, submandibular/cervical, inguinal, and mesenteric nodes), nasal cavity including NALT and turbinates, pharynx, trachea, thymus, thyroid, eyes, brain (including olfactory bulb), oesophagus, stomach, intestine, jejunum containing Peyer's patches, spleen, liver, kidneys, and sternal bone marrow were blocked and embedded in paraffin wax. Serial 4 μm thick sections were cut using a Leica RM2025 microtome (Leica Microsystems Ltd., UK) and dried onto charged slides at 37°C overnight. Sections were stained with haematoxylin and eosin or Gram Twort before examination using a Leica DM4000B microscope. Images were taken using a Leica DFC480 digital camera (Leica Microsystems Ltd., UK). Lesions were scored on the basis of the number of animals displaying specific pathological changes and the severity of these lesions according to the system previously described (Thomas et al., 2009).

Groups of 10 mice challenged with either small or large particle aerosols (10 or 100 MLDs) containing B. pseudomallei K96243 were administered doxycycline or co-trimoxazole orally for 14 days. Doxycycline oral suspension (Sigma-Aldrich, UK) was administered to mice daily at 1 mg/mouse (50 mg/kg). Co-trimoxazole oral suspension (Septrin®, Glaxo SmithKline Laboratories, UK) was administered to mice twice daily at 4.8 mg/mouse (240 mg/kg). Control groups of 10 infected, non-treated mice were given sterile deionized water orally for 14 days. All mice were observed twice daily for clinical signs of disease for up to 96 days post-infection and mortality recorded. Doses were derived from Food and Drugs Agency Guidance using inversed scaling based on body weight (FDA, 2005).

Data are expressed as the standard error (S.E) around the mean. Tissues containing bacterial loads below the limit of detection (<10 cfu per organ) were designated culture-negative. Two-sample t-test was used to test the significant differences in the deposition data. The significant differences in the antibiotic studies was analysed by ANOVAs followed by Tukey's honestly significant difference post-test. P values of <0.05 are considered significantly different.

The regional deposition profiles of B. pseudomallei were determined to evaluate where the organisms would predominantly deposit when inhaled within 1 or 12 μm particles (Figure 1). A culture containing 2.95 ± 0.35 × 10⁹ cfu ml⁻¹ of B. pseudomallei was aerosolized and the retained dose determined. When the 1 μm particle aerosol was generated, impinger samples indicated a retained bacterial dose of 2.79 × 10⁵ cfu. The actual retained dose determined at post-mortem was 1.40 ± 0.09 × 10⁵ cfu in lung tissues and 1.56 ± 0.77 × 10³ cfu in the nasal passages. Hence, 87.7% and 1.0% of the total deposited fraction were retained in the lungs and nasal passages, respectively; this was a significant difference (p = 0.0002). When a 12 μm particle aerosol was generated, 1.40 ± 0.22 × 10⁵ and 9.50 ± 5.2 × 10³ cfu were recovered at post-mortem from the nasal passages and lungs, respectively. This was a significant difference, equivalent to 87.6% and 6.0% of the total retained bacteria being present within the nasal passages and lungs respectively (p = 0.0006). The percentage of total deposited bacteria recovered from the stomach was much reduced at 9.9% and 3.8% for mice that inhaled the 1 or 12 μm particles, respectively. The numbers of bacteria present in the trachea and oesophagus was less than 2% of the total retained fraction irrespective of the size of particle inhaled. No significant differences were observed between 1 and 12 μm particle aerosols for the numbers of B. pseudomallei deposited in the trachea, oesophagus, and stomach (p > 0.09). Ten-fold serial dilutions of the initial challenge culture resulted in approximate 10-fold lower numbers deposited in the respiratory tracts of mice (data not shown).

The MLD of B. pseudomallei was determined to investigate the number of bacteria required to generate infection in the URT and LRT of mice challenged with 1 or 12 μm particles Mice with a retained dose of 10³ cfu of 1 μm particles displayed a MTD of 52.8 ± 2.57 h, whilst a retained dose of 10² cfu resulted in protracted infection with all of the mice succumbing by 22 d post-infection with a MTD of 146.9 ± 37.5 h post-infection. Mice with retained doses of 1 cfu resulted in 30% survival. When mice received a retained dose of 3 × 10³ cfu of B. pseudomallei in 12 μm particles all of the animals died with a MTD of 91.9 ± 17.5 h whilst a retained dose of 30 cfu resulted in protracted infection with 90% of the mice succumbing by 21 d post-infection with a MTD of 259.5 ± 33.6 h. Only 20% of mice that received a retained dose of 3 cfu succumbed to infection. The mice that survived until the end of the experiment had cleared the infection as evidenced by bacteriological analysis of lung and spleen tissues. Irrespective of particle size, none of the mice challenged with concentrations predicted to contain less than a single viable bacterium died during the course of the experiment. Calculation of the MLD produced values of 4 and 12 cfu, respectively for B. pseudomallei in 1 or 12 μm particles.

The temporal progression of inhalational melioidosis caused by deposition of B. pseudomallei within either the LRT or URT was characterized by bacteriological and histopathological analysis; summarized in Table 1. Inhalation of the aerosol produced by the Collison nebuliser (1 μm particles) resulted in the deposition of 2.93 ± 0.3 × 10² and 1.6 ± 0.1 × 10¹ cfu in the lungs and nasal passages, respectively. Conversely, the inhalation of the aerosol produced by the FFAG (12 μm particles) resulted in the deposition of 5.2 ± 4.1 × 10¹ and 3.0 ± 0.2 × 10² cfu in the lungs and nasal passages, respectively. Irrespective of the aerosol generator used, the resultant infection was characterized by primary pneumonic infection of the lung. Significant differences were observed in the bacterial load in the lungs at 0 and 24 h post-challenge (Figure 2; p = 0.034), however, by 96 h post-challenge the loads were similar irrespective of the size of the particles inhaled reaching 1.3–1.5 × 10⁹ cfu g⁻¹ (p = 0.96). Initially at 24 h post-challenge small inflammatory foci were observed with neutrophilic exudates into the alveoli and necrosis indicative of acute alveolitis that increased in size as the infection developed (Figure 3). The foci were present throughout the cranial, middle and caudal lobes. Inflammatory exudates were observed occupying bronchiolar airways by 72 h post-challenge (Figure 3C). In the majority of mice that inhaled both 1 μm (100%; 8/8) or 12 μm particles (82%; 14/17) the inflammatory foci merged into large areas of consolidation comprising predominantly neutrophils and fewer macrophages, bacterial colonies, and necrotic tissue (Figure 3D). Most mice succumbed to infection by 72–120 h, however, some survived for prolonged periods of up to 21 d post-challenge. Interestingly at this stage, the lungs were almost completely consolidated (Figures 3E,F), however, surprisingly there were no obvious signs of severe respiratory distress other than the mice demonstrating clinical signs of ruffled fur and a hunched posture.

However, differences were observed in the degree of URT and neurological involvement between the infections. Differences in the bacterial load within the nasal cavity were observed over the time-course of experimentation with significance at 24 h post-challenge (p = 0.009; Figure 2). Bacterial loads in the nasal cavity increased reaching 1.3 ± 0.5 × 10⁴ and 7.1 ± 4.1 × 10⁵ cfu ml⁻¹ at 72 h post-challenge in mice that inhaled 1 and 12 μm particles, respectively (p = 0.073). By 48 h post-challenge, significant pathological changes of epithelium of the nasal mucosa was observed in mice exposed to B. pseudomallei with 12 μm aerosol particles. The olfactory epithelium was preferentially targeted with 82% (14/17) of the mice demonstrating infection in this region compared to the respiratory mucosa of the nasal cavity (71%; 12/17). As early as 24 h post-challenge, neutrophil infiltration of the lamina propria and exudation in the nasal passages with associated bacteria can be observed (Figure 4A).

NALT was preferentially infected in mice that inhaled B. pseudomallei within 12 μm particles (41%; 7/17) compared to 1 μm particles (12.5%; 1/8) with lymphocytic degeneration and necrosis, neutrophilic infiltration and large numbers of bacteria present within the necrotic tissue (Figure 4B). Interestingly, those mice that survived past 120 h infection and subjected to histological analysis did not show evidence of NALT involvement (0%; 0/3). Irrespective of the size of the particles inhaled, the pharynx and trachea were infected similarly at 75% (6/8) and 80% (12/15) for mice that inhaled B. pseudomallei within 1 and 12 μm particle aerosols, respectively. In the pharynx, exudation into the luminal space of sloughed epithelial cells was observed in 6–12% of mice. An ulcerative and suppurative tracheitis occurred in 75–80% of mice with sloughing of epithelial cells across extensive areas, neutrophil infiltration, and exudation into the tracheal lumen (Figure 4D). The severity of pharyngeal and tracheal pathology appeared earlier and was more extensive in mice exposed to the 12 μm particle aerosol (Table 1). In mice that inhaled 12 μm particles, the cervical lymph nodes became infected in 71% of mice (12/17) by 24 h post-challenge. Pathological changes in the cervical lymph nodes included multifocal necrosis, neutrophilic infiltration, and granulomatic encapsulation by macrophages as the early stages of abscessation (Figures 4E,F). In contrast, only 38% of mice (3/8) that inhaled B. pseudomallei within 1 μm particles displayed similar histological changes in the cervical lymph nodes.

In mice that inhaled 12 μm particles, the inflammatory response in the nasal cavity progressed over time, with extensive neutrophilic infiltration and bacterial proliferation encompassing both the respiratory and olfactory epithelia observed at 96 h post-challenge (Figures 4C and 5A,B). Neutrophilic infiltration, bacterial infection, and inflammation of the olfactory nerves/tract (80%; 12/15) and olfactory bulb (60%; 9/15) were observed by 48–96 h post-challenge. Histological analysis indicated neutrophilic infiltration into the axonal bundle of the olfactory neurons followed by temporal passage into the olfactory bulb (Figures 5C–E). Generally the infection was acute enough for the mice to be humanely culled by 96–120 h post-challenge, however, some mice survived for prolonged periods up to 20 d post-challenge developing a chronic infection. In 33% of mice (2/6) that developed chronic infection (2/6) histological analysis revealed infection of the olfactory tracts extending through the ethmoidal foramina into the olfactory bulb of the brain. Microabscesses were also observed in the medulla and brainstem with the larger abscesses (~0.5 mm) located in the ventrorostral and mid-regions of the brain (caudate putamen, olfactory areas) (Figure 5F). Areas of suppurative encephalitis can be observed with neutrophilic infiltration into the brain tissue. Infection of the olfactory and respiratory epithelium was observed in mice that inhaled 1 μm particles containing B. pseudomallei, however, comparatively this occurred later during the infection by 96 h and to a lesser degree (37%; 3/8). Furthermore, only 12% of mice (1/8) that inhaled 1 μm particles demonstrated pathology of the olfactory bulb.

Bacteraemia was observed at similar levels at 48 h post-challenge in both inhalational infections, 0.57–1.4 × 10² cfu ml⁻¹ (p = 0.066) increasing to 1.3–2.2 × 10⁵ cfu ml⁻¹ by 96 h post-challenge (p = 0.19; Figure 2). Bacteraemic and/or lymphatic spread to visceral organs such as the spleen and liver was evident with hyperplasia in the red pulp, and multifocal microabscess formation observed in both tissues as early as 48 h post-challenge contributing at later time-points to spleno- and hepatomegaly. In the spleen, the microabscesses merged into larger abscesses at the later time-points (>96 h). The bacterial loads in the spleen and liver were highly variable, particularly at 96 h post-challenge, probably due to the degree of abscessation observed (Figure 2). For example at 96 h post-challenge, although the spleen bacterial loads in the mice exposed to the 1 μm particle aerosol was 5.85 ± 5.72 × 10⁶ cfu g⁻¹ and greater than that observed in mice exposed to 12 μm particle aerosols (4.79 ± 7.37 × 10⁵ cfu g⁻¹), this was not a significant difference due to the variance in loads observed between individual mice (p = 0.46). The thymus was infected similarly in both the infections caused by the inhalation of 1 and 12 μm particle aerosols at 50% (4/8) and 53% (9/17), respectively, with lymphocytic depletion and increased tangible body macrophages (Table 1). Pathology in the inguinal lymph nodes was only observed at 96 h post-challenge in 12% (2/17) of mice that inhaled 12 μm particle aerosols. The mesenteric lymph nodes only displayed pathological changes late in the infection in mice that inhaled 12 μm particles (29%; 5/17) displaying distension of the medulla and/or subcapsular sinuses and marked medullary histiocytosis, probably indicative of bacteraemic spread. No other gastrointestinal pathology was observed irrespective of the size of particles inhaled.

Little published data exist on the in vivo efficacy of antimicrobials against experimental melioidosis, particularly via inhalational challenge. In human cases of melioidosis, prolonged combinatorial antimicrobial treatment is required with ceftazidime, in conjunction with imipenem, co-trimoxazole, amoxicillin-clavulanate, doxycycline, or chloramphenicol. However, doxycycline (100 mg twice daily) or co-trimoxazole (960 mg twice daily) are the prophylactic antibiotics of choice in a mass-casualty setting resulting from an intentional release of B. pseudomallei and hence the focus for this study (Chaowagul et al., 2005; Gilad et al., 2007; HPA, 2008). A comparison was made between the efficacy of doxycycline and co-trimoxazole to protect mice against B. pseudomallei deposited in the respiratory tract within either 1 or 12 μm particle aerosols (Figure 6). The MICs of doxycycline and co-trimoxazole (trimethoprim/sulfamethoxazole) for B. pseudomallei K96243 are 0.5 and 32/16 μg ml⁻¹, respectively (Barnes, pers. comm.). Mice were challenged with approximately 100 MLDs of B. pseudomallei within either 1 or 12 μm particle aerosols. The actual retained doses were 5.7 ± 1.1 × 10² cfu and 8.7 ± 0.9 × 10² cfu for the 1 or 12 μm particle aerosols within the lungs and nasal passages, respectively. Irrespective of particle size, all of the untreated control mice succumbed to the challenge. The MTD for the untreated control mice were significantly different at 2.5 ± 0 and 16.0 ± 5.2 d for the 1 and 12 μm particle aerosol infections, respectively (p = 0.028).

Antimicrobial therapy was commenced, with a once or twice-daily regimen of doxycycline (50 mg kg⁻¹) or co-trimoxazole (240 mg kg⁻¹) at 12 h intervals for 14 d. Doxycycline was significantly more effective against the 12 μm particle infection compared to the 1 μm particle infection. Despite doxycycline treatment, 90% of mice that inhaled 1 μm particles succumbed by 26 d post-infection. Survival was increased to 69 d post-infection in doxycycline treated mice that inhaled B. pseudomallei within 12 μm particle aerosols. This was exemplified by significantly different MTDs for the 1 and 12 μm particle infections at 13.0 ± 5.1 and 61.4 ± 1.5 d, respectively (p = 0.0001). Administration of co-trimoxazole significantly prolonged survival with all of the animals surviving until day 58 irrespective of the particle size inhaled; however, clearance was not observed and the mice started developing signs of infection 2–3 weeks after antimicrobial dosing was completed and began succumbing to infection 6–7 weeks post-infection. The experiment was terminated 93 d post-infection with 0% and 20% survival observed in groups that inhaled B. pseudomallei within 1 and 12 μm particles respectively. The MTDs for the 1 and 12 μm particle infections were 68.6 ± 4.5 and 88.0 ± 6.4 d, respectively (p = 0.0005). In all of the antibiotic treated groups, bacteriological analysis of the lungs and spleens indicated that clearance had not occurred. Bacterial burdens in the lungs were significantly increased in mice treated with doxycycline (1.1 ± 0.11 × 10⁸ cfu g⁻¹) compared to co-trimoxazole treated mice (5.8 ± 2.4 × 10⁴ cfu g⁻¹; p = 0.007). No difference was observed in bacterial loads from spleens irrespective of the antibiotic treatment received with numbers ranging from 1.1 to 8.1 × 10⁷ cfu g⁻¹ (p = 0.14). Post-mortem analysis of mice in groups that succumbed to infection 6–7 weeks post-treatment indicated large abscesses present within the lungs, spleens and often the cervical lymph nodes.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.