Phage Therapy of Pseudomonas aeruginosa Infection in a Mouse Burn Wound Model

Bacterial strains, bacteriophages, and culture conditions.PAO1^Rif is a rifampin-resistant derivative of virulent P. aeruginosa strain PAO1 (12), which was kindly provided by Abdul Hamood (11) and which was grown in Luria-Bertani (LB) medium supplemented with rifampin (80 μg/ml), 1 mM MgSO₄, and 1 mM CaCl₂ in a gyratory shaker at 250 rpm at 37°C.

The P. aeruginosa phages were plaque-purified subcultures of phages that had been purchased from the American Type Culture Collection (Catalogue of bacteria and bacteriophages, 18th ed., 1992; ATCC, Manassas, VA). Monophage preparations were propagated on their respective hosts growing in LB medium at 37°C in a gyratory shaking water bath at 250 rpm. Phage lysates were centrifuged (5,000 × g for 15 min) to remove cellular debris, filter sterilized (pore size, 0.22 μm; Millipore), and stored over a drop of chloroform at 4°C in amber bottles. The phage preparations to be used therapeutically were passed through a column containing Detoxi-endotoxin removing gel (Pierce, Rockford, IL), as recommended by the manufacturer, and eluted with pyrogen-free water. The eluted phages were diluted to the appropriate titer with filter-sterilized phosphate-buffered saline (PBS; pH 7.2) prior to administration to the mice. Phage titers were determined by serial dilution and plaque assays by the soft overlay technique (1). This entailed the addition of 100 μl of the different phage dilutions to 100 μl of an overnight culture of their host strain. The mixture was then allowed to stand at room temperature for 5 min for phage adsorption, after which 3 ml of soft agar (0.7% LB agar maintained at 48°C) was added and the mixture was poured over an LB agar plate. The soft agar overlay was allowed to solidify, the plates were incubated overnight at 37°C, and the plaques were counted to determine the phage titer.

Selection of therapeutic phages.To select for the phages to be used in our phage cocktail, we used two criteria: virulence and host range (i.e., utilization of different phage receptors). Of our 13 ATCC phages, 7 were able to grow on PAO1^Rif. Of those, two formed hazy plaques, suggesting that they may be lysogenic. To determine the relative virulence of the remaining five phages, we developed an in vitro virulence test in which we determined the MIC required to clear (lyse) an exponentially growing culture of PAO1^Rif (∼10⁷/ml) over a given period of time (5 h). Those with the lowest MICs were considered the most virulent. We realize, however, that we are measuring only one virulence parameter and that in vivo and in vitro virulence may not be equivalent and may be determined by other phenotypic traits (6, 21).

Resistance to phage often arises through bacterial mutations that alter receptors on the bacterial surface to which the phage binds (phage receptors). In an attempt to decrease the likelihood of the emergence of phage-resistant PAO1^Rif strains, we selected phages that utilized different phage receptors for infection. To identify such phages we isolated PAO1^Rif mutants resistant to each of the five different phages being analyzed and determined the sensitivities of these mutants to the other phages being analyzed. We assumed that a phage to which a PAO1^Rif phage-resistant mutant was sensitive did not share a common receptor with the phage to which the mutant was resistant.

Based on our virulence and host range (receptor usage) tests, we selected three phages for use in this study. The phage cocktails used in this study contained approximately 10⁸ PFU/100 μl inoculum of each of the following phages: Pa1 (ATCC 12175-B1); Pa2 (ATCC 14203-B1), and Pa11 (ATCC 14205-B1) (ATCC catalogue of bacteria and bacteriophages, 18th ed., 1992).

Animals.Adult female ND4 Swiss Webster mice (weight, 20 to 24 g) were used for this study. The animals were anesthetized with 0.4 ml of 5% sodium pentobarbital by intraperitoneal (i.p.) injection. The mice were housed in the Texas Tech University Health Sciences Center Vivarium. The animals were treated in accordance with protocol no. 96020-06, approved by the Animal Care and Use Committee at Texas Tech University Health Sciences Center in Lubbock.

Thermal injury model.The mouse model of thermal injury of Stieritz and Holder (28), as modified by Hamood (11), was used in this study. Briefly, the hair was clipped from the backs of anesthetized mice, and the area was denuded with a commercially available hair remover. The mice were securely placed into a template with an opening of 4.5 cm by 1.8 cm to expose their shaved backs. A nonlethal full-thickness thermal injury to the skin was induced by placing the exposed back area to 90°C water for 10 s. Fluid replacement therapy consisting of a subcutaneous (s.c.) injection of 0.8 ml of a 9% NaCl solution was administered immediately following the burn. The mice were challenged by s.c. injection of 100 μl of the PAO1^Rif inoculum (2 × 10² to 3 × 10² CFU) directly under the anterior end of the burn. The phage cocktail was administered immediately after the P. aeruginosa challenge.

During recovery from anesthesia, the mice were kept under warming lights and observation. The cumulative mortality among the treatment groups was recorded at 48 and 72 h postinfection (hpi). Lethargic animals were monitored every hour. The surviving animals were killed at 96 hpi. Liver and spleen tissue samples were removed from the animals, weighed, suspended in 2 ml of filter-sterilized PBS (pH 7.2), and homogenized with sterile mortars and motor-driven Teflon pestles. The numbers of PAO1^Rif cells (CFU/g tissue) were determined by serial dilution and plating on LB agar containing 80 mg/ml of rifampin. The titers (PFU/g tissue) of the phages recovered from the tissues of mice were determined by the soft agar overlay technique described above, following the addition of a few drops of chloroform (per ml).

Pharmacokinetic studies.A phage cocktail inoculum (∼3 × 10⁸/100 μl per inoculum) was administered to unwounded, noninfected mice i.p., intramuscularly (i.m.). or s.c. Three animals from each treatment group were killed at 0.5, 12, 24, 36, and 48 h following phage injection. Briefly, blood, obtained by cardiac puncture from anesthetized animals, was collected in tubes containing EDTA. After blood collection, the anesthetized mice were euthanized by injection of Fatal Plus (sodium pentobarbital), followed by cervical dislocation. Liver and spleen tissue samples were removed from the animals. The tissues were weighed, suspended in 2 ml of filter-sterilized PBS, and homogenized with sterile mortars and motor-driven Teflon pestles. The phages in the tissues were enumerated as described above and are expressed as PFU/gram of tissue.

Statistical analysis.Fisher's exact test (Statview for Windows; SAS Institute, Cary, NC) was used to determine the significance of the differences in survival among the treatment groups and the differences in the distributions of the phages to the tissues.

Protection studies.The ability of the P. aeruginosa-specific phages to prevent P. aeruginosa infections was examined by use of a modified mouse model of thermal injury, as described above. A phage cocktail containing 1 × 10⁸ PFU of each of three different phages (3.0 × 10⁸ PFU total) was administered i.p., i.m., or s.c. to infected and uninfected wounded animals. As a control for the virulence of the PAO1^Rif inoculum, another group of mice was injected with the bacterial inoculum only (no phage). To examine the toxicity of the phages in compromised animals, the wounded but noninfected groups were injected with the phage cocktail (but not P. aeruginosa). Animal deaths were recorded at 48 and 72 hpi.

The results of these experiments are presented in Table 1. All of the thermally injured mice that were not infected with PAO1^Rif but administered the phage cocktail survived, indicating that the phage cocktail was not toxic to traumatized mice. In the absence of phage there was a 94% rate of mortality in the wounded infected mice in the first 72 hpi. When the phages were administered i.m. or s.c., the rates of mortality were reduced to 72% and 78%, respectively; and by sharp contrast the rate of mortality was reduced to 12% when the phages were delivered i.p. These results demonstrate that the parenterally administered phages significantly increased survival in infected and wounded mice and that the relative protection afforded by the different routes of phage administration was i.p. > i.m. or s.c.

Of the wounded and infected animals receiving the phages i.p., only 2 of the 17 animals had died by 72 hpi (at 53 hpi and 64 hpi, respectively). The surviving animals were killed at 96 hpi, and the numbers of PAO1^Rif organisms detected from the tissues of surviving animals was compared to the numbers from the tissues of animals that died from the P. aeruginosa infection. The mean bacterial counts in the tissues of animals that died were 1.53 × 10⁹ CFU/g liver and 6.68 × 10⁷ CFU/g spleen. The mean bacterial counts in the tissues following successful i.p. phage therapy were 5.26 × 10² CFU/gram liver and 2.93 × 10² CFU/gram spleen. These results suggest that the cause of death was the result of systemic P. aeruginosa infection and that, as one might expect, successful phage therapy correlates with a reduction in the PAO1^Rif burden.

To determine if phage-resistant derivatives of PAO1^Rif emerged from the infected mice, we analyzed the PAO1^Rif isolates from the tissues of those mice that had died by 48 hpi for their phage sensitivities. All isolates tested (>100) were sensitive to each of the phages in the cocktail (data not shown). Hence, the death of these animals was not the result of the emergence of a phage-resistant derivative of the PAO1^Rif strain. The numbers of phages in the liver and spleen of animals that succumbed to P. aeruginosa infection in the first 48 hpi were also enumerated to determine if the phages had multiplied. Considering the average weight of these organs (4.6 g and 2.27 g of liver and spleen, respectively) and assuming 100% phage recovery, each mouse harbored a minimum of 7.5 × 10⁹ to 10 × 10⁹ phages. This represents an increase of approximately 20-fold over that administered to these mice (∼3 × 10⁸/mouse), indicating that the phages multiplied in vivo, although obviously not to levels that were enough to save the animal.

Pharmacokinetic studies.In an attempt to determine why delivery of the phages by the i.p. route was more efficacious than delivery of the phages by the i.m. or the s.c. route for the treatment of infected animals, we examined the pharmacokinetics of the phages introduced by the i.m., s.c., or i.p. route in uninjured, uninfected animals. Three animals each from groups receiving the phage cocktails i.p., i.m., or s.c. were killed at 0.5, 12, 24, 36, and 48 hpi. The numbers of phages detected per gram of liver and spleen and per milliliter of blood are shown in Fig. 1. In each tissue examined, a consistent pattern of the relative PFU levels after the administration of the phages by the different routes was observed: i.p. > i.m. > s.c.

• Open in new tab
• Download powerpoint

FIG. 1.

Pharmacokinetics of the phages in noninfected mice. A cocktail containing ∼3 × 10⁸ phage was injected i.p., i.m., or s.c. into uninjured, uninfected mice. Groups of mice (n = 3) were killed at 0.5 h hpi, 12 hpi, 24 hpi, 36 hpi, or 48 hpi. Tissues were removed and the numbers of PFU were determined by serial dilution and plating on PAO1^Rif. (A) PFU/gram liver; (B) PFU/gram spleen; (C) PFU/ml blood. Values are means ± standard errors of the means.