Gram-negative bacteria act as a reservoir for aminoglycoside antibiotics that interact with host factors to enhance bacterial killing in a mouse model of pneumonia

Abstract

In vitro exposure of multiple Gram-negative bacteria to an aminoglycoside (AG) antibiotic has previously been demonstrated to result in bacterial alterations that interact with host factors to suppress Gram-negative pneumonia. However, the mechanisms resulting in suppression are not known. Here, the hypothesis that Gram-negative bacteria bind and retain AGs, which are introduced into the lung and interact with host defenses to affect bacterial killing, was tested. Following in vitro exposure of one of several, pathogenic Gram-negative bacteria to the AG antibiotics kanamycin or gentamicin, AGs were detected in bacterial cell pellets (up to 208 μg/mL). Using inhibitors of AG binding and internalization, the bacterial outer membrane was implicated as the predominant kanamycin and gentamicin reservoir. Following intranasal administration of gentamicin-bound bacteria or gentamicin solution at the time of infection with live, AG-naïve bacteria, gentamicin was detected in the lungs of infected mice (up to 8 μg/g). Co-inoculation with gentamicin-bound bacteria resulted in killing of AG-naïve bacteria by up to 3-log₁₀, mirroring the effects of intranasal gentamicin treatment. In vitro killing of AG-naïve bacteria mediated by kanamycin-bound bacteria required the presence of detergents or pulmonary surfactant, suggesting that increased bacterial killing inside the murine lung is facilitated by the detergent component of pulmonary surfactant. These findings demonstrate that Gram-negative bacteria bind and retain AGs that can interact with host-derived pulmonary surfactant to enhance bacterial killing in the lung. This may help explain why AGs appear to have unique efficacy in the lung and might expand their clinical utility.

Keywords: Gram-negative; aminoglycosides; antibiotics; bacterial pneumonia; host–microbe interactions; pulmonary surfactant.

Figure 1.

Gram-negative bacteria bind and retain AG antibiotics, which can interact with host factors in the lung to affect bacterial killing. (aandb), The concentrations of kanamycin in cell pellets of chemically killed, kanamycin-resistant (a) A. baumannii 17978/pMU368 (Km MIC: 104.0 mg/L; Km^R), E. coli DH5⍺/pCR2.1 (Km MIC: >256 mg/L; Km^R), P. aeruginosa PAO1/pME260 (Km MIC: >256 mg/L; Km^R), and K. pneumoniae 43816/pCR2.1 (Km MIC: 131.5 mg/L; Km^R) or kanamycin-susceptible (b) A. baumannii 17978 (Km MIC: 0.9 mg/L; Km^S), E. coli DH5⍺ (Km MIC: 1.25 mg/L; Km^S), P. aeruginosa PAO1 (Km MIC: 10 mg/L; Km^S), K. pneumoniae 43816 (Km MIC: ND; Km^S), and S. aureus USA300 LAC (Km MIC: ND; Km^S) exposed to media alone (LB) or media supplemented with kanamycin are shown as quantified by ELISA. (c), The concentrations of kanamycin in cell pellets of kanamycin-resistant A. baumannii/pMU368 and E. coli/pCR2.1, and kanamycin-susceptible A. baumannii, E. coli, and S. aureus exposed to media alone (LB) or media supplemented with kanamycin are shown as quantified by LC-MS. (d), Mid-exponential-phase WT A. baumannii (Km MIC: 0.9 mg/L; Km^S) grown in media without antibiotics (LB) was co-incubated with chemically killed WT A. baumannii (Km^S), A. baumannii/pMU368 (Km MIC: 104.0 mg/L; Km^R), or P. aeruginosa/pME260 (Km MIC: >256 mg/L; Km^R) exposed to media alone (LB) or media supplemented with kanamycin as indicated. Viability of AG-naïve, WT A. baumannii was monitored over time. (e), Mice were infected with mid-exponential-phase WT A. baumannii (Km MIC: 0.9 mg/L; Km^S) grown in media without antibiotics (LB) and co-inoculated with chemically killed A. baumannii/pMU368 (Km MIC: 104.0 mg/L; Km^R), P. aeruginosa/pME260 (Km MIC: >256 mg/L; Km^R), K. pneumoniae/pCR2.1 (Km MIC: 131.5 mg/L; Km^R), or E. coli/pCR2.1 (Km MIC: >256 mg/L; Km^R) exposed to media alone (LB) or media supplemented with kanamycin as indicated. Bacterial burdens in the lungs of infected mice were determined at 36 h.p.i. (a-c), N = 3 biological replicates per group, per experiment. Columns depict the mean and error bars show standard deviation (a and b) or standard error (c) of the mean. (d), N = 3 biological replicates per group, per experiment. Symbols depict the mean and error bars show standard deviation of the mean. (e), Circles represent individual animals, columns depict the mean, and error bars show standard deviation of the mean. Means were compared using a one-way ANOVA adjusted for multiple comparisons. ^****: P < 0.0001; ns: not significant. Ab: Acinetobacter baumannii; Ec: Escherichia coli; Pa: Pseudomonas aeruginosa; Kp: Klebsiella pneumoniae; Sa: Staphylococcus aureus; Km: kanamycin; ND: not determined..

Figure 2.

Co-inoculation of mice with AG-bound bacteria may be as effective as treatment of mice with inhaled AGs. (a and b), The concentrations of gentamicin in cell pellets of chemically killed A. baumannii 17978 ∆hcp::gm (Gm MIC: >256 mg/L; Gm^R), K. pneumoniae 43816 (Gm MIC: 1.5 mg/L; Gm^S), P. aeruginosa PAO1 (Gm MIC: 0.46 mg/L; Gm^S), E. coli DH5⍺ (Gm MIC: 1.25 mg/L; Gm^S), and S. aureus USA 300 LAC (Gm MIC: 1.5 mg/L; Gm^S) exposed to media with or without gentamicin are shown as quantified by ELISA (a) or LC-MS (b). (c), Bacterial burdens in the lungs of mice infected with mid-exponential-phase, WT A. baumannii 17978 (Gm MIC: 0.38 mg/L; Gm^S) exposed to media without antibiotics (LB); co-inoculated with A. baumannii Δhcp::gm (Gm MIC: >256 mg/L; Gm^R) exposed to LB ± gentamicin as indicated; and treated intranasally with PBS or PBS supplemented with gentamicin (64 μg/mL) are depicted. Bacterial burdens in the lungs of infected mice were determined at the indicated times post-infection. (d), concentrations of gentamicin detected in lung homogenates of infected mice using a competitive ELISA are shown. (a and b), N = 3–4 biological replicates per group, per experiment. Columns depict the mean and error bars show standard deviation of the mean. (c), symbols represent individual animals, center bars depict the mean, and error bars show standard deviation of the mean. (d), Columns depict the mean and error bars show standard deviation of the mean. (c and d), For each time point, means were compared to all other means using a one-way ANOVA adjusted for multiple comparisons. *: P< 0.05; **: P< 0.01; ***: P < 0.001; ^****: P < 0.0001; ns: not significant. Ab: Acinetobacter baumannii; Kp: Klebsiella pneumoniae; Pa: Pseudomonas aeruginosa; Ec: Escherichia coli; Sa: Staphylococcus aureus; Gm: gentamicin; h.p.i.: hours post-infection; μg/g: μg per gram of lung tissue.

Figure 3.

The Gram-negative outer membrane serves as a reservoir for AG antibiotics. (a), The concentration of gentamicin in cell pellets of chemically killed LOS-sufficient and LOS-insufficient A. baumannii 17978 (Gm MIC: 0.38 mg/L; Gm^S) exposed to media with gentamicin is shown as quantified by LC-MS (b), Viability of E. coli DH5⍺ (Gm MIC: 1.25 mg/L; Gm^S) exposed to PBS or gentamicin ± CCCP or MgSO₄in vitro before and after exposure is depicted. (c), The concentrations of gentamicin in cell pellets of chemically killed E. coli DH5⍺ (Gm MIC: 1.25 mg/L; Gm^S) exposed to PBS or gentamicin ± CCCP or MgSO₄in vitro are shown as quantified by ELISA. (d), Viability of E. coli DH5⍺ (Km MIC: 1.25 mg/L; Km^S) exposed to kanamycin ± CCCP or MgSO₄in vitro before and after exposure is depicted. (e), Bacterial burdens in the lungs of mice infected with mid-exponential phase WT A. baumannii 17978 (Km MIC: 0.9 mg/L; Km^S) grown in media alone (LB) and co-inoculated with chemically killed E. coli DH5⍺ (Km MIC: 1.25 mg/L; Km^S) exposed to kanamycin ± CCCP or MgSO₄in vitro prior to infection are shown. Bacterial burdens were determined at 36 h.p.i. (f), Bacterial burdens in the lungs of mice infected with mid-exponential phase, WT A. baumannii 17978 (Km MIC: 0.9 mg/L; Km^S) grown in media alone (LB) and co-inoculated with chemically killed A. baumannii Tn5A7 (Km MIC: 128 mg/L; Km^R) or WT A. baumannii 17978 (Km MIC: 0.9 mg/L; Km^S) exposed to varying concentrations of kanamycin as indicated. Bacterial burdens were determined at 36 h.p.i. (a), N = 4–5 replicates per group, per experiment. Columns depict the mean and error bars show standard deviation of the mean. Means were compared using a Welch's t-test. (b and d), N = 4 (b) or N = 5 replicates (d) per group, per experiment. Symbols depict the mean and error bars show standard deviation of the mean. Means were compared to the mean bacterial viability of the untreated group (PBS) (b) or to the group treated with kanamycin alone (Km) (d) using a one-way ANOVA adjusted for multiple comparisons. (c), N = 3–4 biological replicates per group, per experiment. Columns depict the mean and error bars show standard deviation of the mean. Means were compared to all other means using a one-way ANOVA adjusted for multiple comparisons. (e and f), Circles represent individual animals, columns depict the mean, and error bars show standard deviation of the mean. Means were compared to all other means (e) or to the mean of the first column (f) using a one-way ANOVA adjusted for multiple comparisons. *: P< 0.05; **: P< 0.01; ***: P < 0.001; ^****: P < 0.0001; ns: not significant. Km: kanamycin; Gm: gentamicin.

Figure 4.

AG-bound bacteria interact with pulmonary surfactant to affect AG-mediated killing of co-infecting bacteria in the mouse lung. (a), Viability of WT A. baumannii 17978 (Km MIC: 0.9 mg/L; Km^S) exposed to media alone (LB) co-incubated with killed, unexposed WT A. baumannii 17978 (Km MIC: 0.9 mg/L; Km^S) or killed, kanamycin-bound A. baumannii Tn5A7 (Km MIC: 128 mg/L; Km^R) in the presence of 50% porcine surfactant BALF is depicted. Bacterial viability was determined immediately prior to and after incubation in porcine surfactant BALF. (b–d), Viability of WT A. baumannii 17978 (Km MIC: 0.9 mg/L; Km^S) exposed to media alone (LB) co-incubated with killed, WT A. baumannii 17978 (Km MIC: 0.9 mg/L; Km^S) grown in media alone (LB) or killed, kanamycin-bound A. baumannii Tn5A7 (Km MIC: 128 mg/L; Km^R) in the presence of 5 μg/mL SP-B (b), 25 μg/mL SP-D (c), 5 μg/mL SP-B and 25 μg/mL SP-D (d), or PBS (no SPs) is depicted. Bacterial viability was measured over time. (eand f), Viability of WT A. baumannii 17978 (Km MIC: 0.9 mg/L; Km^S), grown in media alone (LB) co-incubated with or without killed A. baumannii or varying concentrations of kanamycin as indicated is depicted. Where indicated, WT A. baumannii was co-incubated with killed, WT A. baumannii 17978 (Km MIC: 0.9 mg/L; Km^S) grown in media alone (LB) or killed, kanamycin-bound A. baumannii Tn5A7 (Km MIC: 128 mg/L; Km^R). Bacterial suspensions were pelleted and resuspended in PBS supplemented with 0.1% Triton X-100 and bacterial viability was monitored over time. (g), The concentration of gentamicin in cell pellets and soluble lysates of killed, gentamicin-exposed WT A. baumannii 17978 (Gm MIC: 0.38 mg/L; Gm^S) and A. baumannii 17978 ∆hcp::gm (Gm MIC: >256 mg/L; Gm^R) incubated with PBS alone or PBS supplemented with deoxycholic acid (10 mg/mL) as measured by LC-MS is shown. (a-g), N = 3–4 biological replicates per group, per experiment. Graphs depict average (a-d) or representative (e-g) data from at least two independent experiments. Symbols (a-f) or columns (g) depict the mean, and error bars show standard deviation of the mean. Means were compared using a Welch's t-test (a) or a one-way ANOVA adjusted for multiple comparisons, for the 1.5h time point (a), for the 24h time point (b-d), or for each time point (e and f). (g), means were compared using a one-way ANOVA adjusted for multiple comparisons. *: P< 0.05; **: P< 0.01; ***: P < 0.001; ^****: P < P < 0.0001; ns: not significant. Km: kanamycin; Gm: gentamicin; DA: deoxycholic acid.

Figure 5.

Working model of killing of co-infecting bacteria inside the murine lung mediated by AG-bound bacteria. Prior to intranasal challenge of mice, bacteria are grown in media alone (LB) or media supplemented with kanamycin (Km), washed, and diluted in PBS to 1×10¹⁰ cfu/mL. For co-infections and co-inoculations, bacterial suspensions (at 1×10¹⁰ cfu/mL) are mixed in a 1:1 ratio. During AG exposure, Gram-negative bacteria bind bioactive AG molecules to their OM which are retained despite multiple washes. Inside the mouse lung, AG-bound bacteria interact with pulmonary surfactant to affect killing of susceptible, co-infecting bacteria.