Yersinia pestis strain CO92 was cultured on blood agar base original (BAB) supplemented with 0.02% haemin or in BAB broth (15 g/L proteose peptone, 2.5 g/L liver digest, 5.0 g/L yeast extract, 5.0 g/L sodium chloride, all from Oxoid, UK). This strain has a minimum inhibitory concentration of <0.063 mg/L for ciprofloxacin (Russell et al., 1998). Bacteria were prepared for the animal challenge inoculum by incubating several colonies (grown for 48 h at 28°C on agar plates) in broth for 40 h at 28°C, with shaking at 180 rpm. Actual inoculum concentration was determined by enumeration of cultured bacteria on agar plates. All experiments with Y. pestis were carried out in a Class III microbiological safety cabinet complying with British Standard 5726.

All experiments with mice were carried out in accordance with the UK Animal (Scientific Procedures) Act (1986). Female BALB/cAnNCrl (BALB/c) mice, 8–10 weeks old and 20 g (±4 g), were obtained from Charles River Laboratories (UK) where they were implanted with a sub-cutaneous Pico transponder (Uno BV, Netherlands), enabling individual animals to be traced throughout the study. Mice were caged in groups of 5 in polypropylene cages with a stainless steel mesh cover, integral water bottle holder and diet hopper in an ACDP (UK) level 3 rigid wall isolator, complying with British Standard 5726. Corn cob grade 10/14 (International Product Supplies [IPS], irradiated) was used as a nesting material and a range of environmental enrichment was used throughout the study including cellulose dome home and aspen wood wool (IPS). Husbandry conditions consisted of a 12-h light-dark cycle (350 to 400 Lux during the day, 10 Lux during the night), 19 to 23°C and 45 to 65% relative humidity. Irradiated Labdiet rodent diet 5002 (IPS) and water were freely available during the study. Mice were allowed to acclimatize to their home cage environment for five days before challenge.

Ciprofloxacin (Ciproxin^®, Bayer), liposomal ciprofloxacin for inhalation (CFI) (Aradigm, USA) and dual release ciprofloxacin for inhalation (DRCFI) (Aradigm, USA) were used in this study. Ciproxin, 2 mg/ml solution for infusion, and CFI (ARD-3100, Lipoquin^®), 50 mg/ml liposome-encapsulated ciprofloxacin (expressed as hydrochloride) were used as supplied. DRCFI (ARD-3150, Pulmaquin^®), 35 mg/ml ciprofloxacin hydrochloride, was produced by mixing equal volumes of CFI and ciprofloxacin hydrochloride solution for inhalation (20 mg/ml) (Aradigm, USA) immediately prior to use.

Mice were administered either ciprofloxacin, CFI, or DRCFI. Aerosolized CFI and DRCFI were administered using the inhalational therapy system (ITS) (Hamblin et al., 2014a), with 6 ml of the drug formulation placed in the nebulizer. Mice were exposed to the CFI (n = 30) or DRCFI (n = 30) aerosols for 20 or 15 min, respectively. Not all of drug in the nebulizer was aerosolized, for example following a 20 min exposure approximately 4 ml of CFI was aerosolized with 2 ml remaining in the nebulizer. The majority of the drug passed through the mouse exposure chamber and was collected in the impinger placed after the exposure chamber (the air is moving through the exposure chamber at 6 L/min whereas mouse respiration is 20 ml/min).

A single dose of ciprofloxacin (30 mg/kg) was administered to mice (n = 30) via the intraperitoneal (i.p.) route. The mice (n = 3 /time point/group) were culled at 1, 10, 20, or 30 min and 1, 1.5, 2, 4, 8, 12 h following ciprofloxacin administration and 1, 15, or 30 min and 1, 2, 4, 6, 10, 18, or 24 h following DRCFI or CFI administration. Blood sampling points were chosen based upon the short half-life of ciprofloxacin and longer half-life of CFI (Hamblin et al., 2014b). Blood and lungs (whole organ) were collected post mortem for analysis. The lung doses following CFI or DRCFI administration were calculated using the concentration of ciprofloxacin in the lung samples at 1 min post-administration.

The concentration-time profile of ciprofloxacin in blood and lung homogenate was determined using liquid chromatography mass spectrometry (LC-MS) as previously described (Hamblin et al., 2014a; Norville et al., 2014). Non-compartmental pharmacokinetic (PK) analysis of the mean concentration–time profiles of ciprofloxacin in the mouse lung and plasma was performed using WinNonlin v.6.1 (Certara, USA). One-compartmental PK analysis of the 15 min DRCFI exposure was also completed to enable simulation of the longer, 30 min, DRCFI exposure.

For comparison purposes, human lung doses were estimated from in vitro measurements of droplet size distribution and emitted dose from the jet nebulizer used clinically (Cipolla et al., 2010).

Two separate animal experiments were conducted; the first experiment determined the efficacy of antibiotics administered at 24 h post-challenge (the post-exposure prophylaxis study) and the second experiment determined the impact of delaying the initiation of therapy until 42 h post-challenge (the treatment study). In these studies, the Y. pestis aerosol was generated using a Collison nebulizer containing 20 ml of Y. pestis strain CO92 at a concentration of approximately 2 × 10⁹ CFU/ml and conditioned using a modified Henderson apparatus (Druett, 1969). Mice were exposed to the aerosol, in groups of 20, for 10 min via a head-only exposure chamber, with aerosol sampling of the aerosol chamber performed using an all-glass impinger (AGI-30; Ace Glass, USA) containing PBS. Each mouse received a retained dose of approximately 1 × 10⁴ CFU in the treatment study or 3 × 10⁴ CFU in the prophylaxis study (determined though enumeration of the bacterial concentration in the aerosol with a calculation (Harper and Morton, 1953) using Guyton’s formula (Guyton, 1947)). The aerosol exposure system has previously been validated for Y. pestis (Thomas et al., 2009). Exposing animals to an aerosol of 1.52 × 10⁹ CFU of Y. pestis, resulted in a lung dose of 1.76 × 10⁴ ± 0.085 × 10⁴ CFU (experimentally determined) compared to an estimated retained dose of 1.9 × 10⁴ CFU (Thomas et al., 2009).

To determine the bacterial burden of mice administered therapy as prophylaxis or treatment, groups of 5 mice received a single dose of i.p. ciprofloxacin, i.p. PBS, aerosolized empty liposomes, aerosolized CFI or aerosolized DRCFI (administered as described above, but with a 30 min exposure for DRCFI) at 24 or 42 h post-challenge. A 30 mg/kg dose of i.p. ciprofloxacin administered produces a drug exposure which is similar to the drug exposure observed in humans following a 500 mg dose of oral ciprofloxacin. The empty liposome control was administered using the ITS (Hamblin et al., 2014a) with 6 ml placed in the nebulizer and mice exposed to the aerosol for 20 min. Mice received approximately 2.8 mg/kg (lung dose) of lipid which matches the lipid lung dose of CFI and DRCFI (2.7 and 3.2 mg/kg). All mice were culled at the onset of clinical signs in the control group of mice administered PBS (at approximately 60 h post-challenge). Spleens and lungs were harvested and processed to determine bacterial load. The limit of detection of bacteria in these organs was 50 CFU/g in the lung and 200 CFU/g in the spleen.

To determine survival following the Y. pestis challenge, groups of 10 mice were administered therapy from 24 or 42 h post-challenge. The groups received a single dose of antibiotic or 3 days of therapy; the mice receiving 3 days of therapy were dosed with ciprofloxacin twice daily whereas CFI, DRCFI and empty liposome control were administered once daily to model the human dosing regimen (Serisier, 2012; Serisier et al., 2013). Mice were observed for 35 days post-challenge and clinical signs were recorded twice daily using a scoring system describing the extent of piloerection, hunching, eye problems and locomotion with scores of 0, 1, or 2 specified for each category. Humane end-points were used to identify moribund mice and minimize suffering.

Graphs were constructed with PRISM (v6.0, GraphPad), statistical analysis was conducted with SPSS (v21.0, IBM) and power calculations were conducted with SPSS SamplePower (v3.0 IBM). In previous Y. pestis studies using similar doses of bacteria, mice began to succumb approximately 2.5 days post infection with a high hazard rate (probability that each mouse will succumb on a particular day) of approximately 1 (Thomas et al., 2009). We estimated that groups of 10 mice would be sufficient to provide a power of ∼80% with reduction in hazard rate to 0.25 (i.e., 25% of the normal virulence). No equivalent CFI/DRCFI treatment survival data was available for power calculation. For assessing differences in bacterial load, we estimated that the group size of 5 would be sufficient to show differences of 1.5 log10s with a ∼80% power as previous data had suggested that the standard deviation of Y. pestis colonization data (transformed to the logarithm of 10) would be around 0.5 (Thomas et al., 2009). The high frequency of animals where no bacteria could be recovered negated the possibility for using parametric analysis and Mood’s median tests with multiple comparisons used to compare bacterial colonization of groups. Where no bacteria could be isolated a number of 0 was assigned. Survival data were compared using Logrank tests. Differences of p < 0.05 were considered statistically significant.

The AUC/MIC (area under the curve/minimum inhibitory concentration) measurement has been shown to be the PK parameter most strongly associated with the efficacy of ciprofloxacin in animal models (Craig, 1998), although this has not been validated for local treatment of lung infections. In this study, administration of a single 30 mg/kg dose of ciprofloxacin, via the intraperitoneal route, to Balb/c mice resulted in a drug exposure which is similar to the drug exposure observed in humans following a 500 mg dose of oral ciprofloxacin (see AUC in Table 1).

Inhalational drug scaling between humans and animals relies upon attempting to deliver the same amount of drug to the lung per kg of body weight. In Phase II human clinical trials with DRCFI, 6 ml (210 mg) was nebulized, the corresponding dose delivered to the lung is approximately 0.8 mg/kg, assuming a 50-kg patient (Aradigm Corporation, unpublished data) which is similar to the dose of CFI delivered to mice (see Table 1). In this study, exposure to DRCFI resulted in a lower inhaled ciprofloxacin dose (see Table 1). Therefore, in order to obtain a similar dose of CFI and DRCFI, the length of exposure to DRCFI was increased. Table 1 shows the PK parameters of the shorter DRCFI exposure and the PK of the longer exposure simulated by the one-compartment models.

The PK parameters of CFI and DRCFI demonstrate the different distributions of ciprofloxacin following administration of these therapies compared to ciprofloxacin administered by the intraperitoneal route. As previously demonstrated in humans (Bruinenberg et al., 2009, 2010; Serisier et al., 2013), in the Balb/c mouse administration of CFI or DRCFI via the inhalational route results in low systemic concentrations of ciprofloxacin compared to ciprofloxacin delivered by the intraperitoneal route (see Figure 1) with a greater than 10-fold decrease in the plasma total exposure (as demonstrated by the AUC). Conversely, and as may be expected, inhalational administration of CFI or DRCFI produces a higher C_max and extended half-life in the lung (see Figure 1) resulting in a 20-fold higher overall exposure in the mouse lung compared to ciprofloxacin delivered by the intraperitoneal route.

First, an early initiation of therapy was investigated, with therapy administered from 24 h post-challenge. Using the Balb/c mouse model of aerosolized Y. pestis infection, bacteria could be detected (∼2 × 10³ CFU/ml) in the lung of all animals at this time, but no bacteria could be detected in the spleen or blood. The control mice (treated with PBS or empty liposomes) succumbed to the infection by day 4 post-challenge whereas a single dose of ciprofloxacin resulted in 60% of mice surviving the lethal challenge. However, a single dose of CFI afforded significantly better protection than ciprofloxacin with all mice surviving until the end of the experiment at day 35 (P = 0.029). There was no significant difference between DRCFI and ciprofloxacin or CFI, with a single dose of DRCFI treatment resulted in 80% survival. When therapy was administered for 3 days, all therapies provided full protection (see Figure 2).

To further differentiate between CFI, DRCFI, and ciprofloxacin, the bacterial burden in lungs and spleens of infected mice was determined at the onset of clinical signs in the control animals (those treated with PBS or empty liposomes). The bacterial load was high in both lungs and spleens in the control groups whereas all three antibiotics had profound effects on bacterial load (see Figure 3). No bacteria were isolated from the spleen of any antibiotic treated animal and, when compared to controls, this represents significant reductions in spleen bacterial loads (P = 0.023). CFI and DRCFI treatment also reduced the lung bacterial loads to lower than the limit of detection, significantly different to the empty liposome control (P = 0.023). Intraperitoneal ciprofloxacin did reduce the lung bacterial load compared to controls treated with intraperitoneal PBS (P = 0.023) (see Figure 3), however, bacteria could still be detected in 2 out of 5 mouse lungs. There were no significant differences in bacterial loads among the different therapies.

The efficacy of CFI and DRCFI against systemic Y. pestis infection was investigated. Specifically, administration of therapy was delayed until 42 h. At this time point, there was a high bacterial burden in the lungs (∼1 × 10⁶ CFU/organ) and bacteria could be detected in the blood and spleen of 3 out of 5 mice with an average bacterial burden of ∼3 × 10⁴ CFU/ml and ∼3 × 10⁵ CFU/organ, respectively. All control mice succumbed to infection by day 4 except for 1 mouse dosed with empty liposomes which survived until day 11 post-challenge. A single dose of all three therapies did increase time to death compared to the controls (P < 0.001) but did not offer a high level of protection. A single dose of ciprofloxacin resulted in 60% survival, and a single dose of CFI or DRCFI resulted in 20 and 10% survival, respectively. Here, ciprofloxacin offered significantly better protection than DRCFI (P = 0.016) but not when compared to CFI (P = 0.060) (see Figure 4). However, impressively, three days of antibiotic therapy offered high levels of protection with 100, 90, and 90% survival afforded by ciprofloxacin, CFI, and DRCFI, respectively (see Figure 4).

The bacterial burden at the onset of clinical signs in control animals challenged with Y. pestis was examined when therapy was delayed until 42 h post-challenge. High bacterial loads were present in the lungs and spleens of the control animals. In comparison, bacterial numbers were reduced to levels below a measureable threshold in the lungs and spleen of all the mice treated with CFI and 3 of the 5 mice treated with DRCFI. A comparison of bacterial numbers in the spleen and lung indicated that the bacterial load had been reduced by all three therapies when compared to their relevant control groups (P = 0.023). Furthermore, both CFI and DRCFI treatment significantly reduced bacterial burden in the lung compared to ciprofloxacin treated mice (P = 0.023). CFI treatment also significantly reduced the spleen bacterial burden compared to ciprofloxacin treatment (P = 0.023) (see Figure 5). However, there was no significant difference in spleen bacterial burden following DRCFI or ciprofloxacin therapy. In addition, there was no significant difference in the lung or spleen bacterial burden of CFI and DRCFI treated mice.