Macrophage-induced reduction of bacteriophage density limits the efficacy of in vivo pulmonary phage therapy

Abstract

The rise of antimicrobial resistance leads to renewed interest in evaluating phage therapy. In murine models highly effective treatment of acute pneumonia caused by Pseudomonas aeruginosa relies on the synergistic antibacterial activity of bacteriophages with neutrophils. Here, we show that depletion of alveolar macrophages (AM) shortens the survival of adult male mice without boosting the P. aeruginosa load in the lungs. Unexpectedly, upon bacteriophage treatment, pulmonary levels of P. aeruginosa are significantly lower in AM-depleted than in immunocompetent mice. To explore potential mechanisms underlying the benefit of AM-depletion in treated mice, we develop a mathematical model of bacteriophage, bacteria, and innate immune system dynamics. Simulations from the model fitted to data suggest that AM reduce bacteriophage density in the lungs. We experimentally confirm that the in vivo decay of bacteriophage is faster in immunocompetent compared to AM-depleted animals and that AM phagocytize therapeutic bacteriophage. These findings demonstrate the involvement of feedback between bacteriophage, bacteria, and the immune system in shaping the outcomes of phage therapy in clinical settings.

Fig. 1. Phage therapy in the macrophage depleted host.

A Flow cytometry analysis of bronco alveolar lavages (BAL) from mice that were given 96 h before clodronate liposomes (depleted) or empty liposomes (control) by intra-tracheal instillation (n = 3-4 each). Cell debris and doublets were first excluded based on CD45 staining. Among the CD45+ population Ly6G+ was used to gate PMN, in the Ly6G- population, the CD11b- correspond to dendritic cells (DC) and alveoli macrophage (AM). The DCs were gated as F4/80- and CD11c+ low while the AM as F4/80+ and CD11c+ high (p value from Mann-Whitney U test two tailed, are indicated). B Survival of acute respiratory infection caused by P. aeruginosa (5 × 10⁶ CFU of strain PAKlumi) of AM-depleted mice (clodronate, n = 8) and control mice (n = 5) was significantly different (p = 0.0015 Mantel-Cox log rank test) in absence of phage treatment (full lines). With phage PAK_P1 (5 × 10⁷ PFU) administered intranasally at 2 h p.i. (dashed lines) in AM-depleted (n = 5) and control (EL) (n = 6) mice, phage treatment significantly improved their survival (p = 0.0058 and 0.0015, Mantel-Cox log rank test, for AM-depleted and control mice, respectively). Data were assembled from 5 independent experiments. C The colonization kinetics of the bioluminescent strain PAKlumi in the lungs is plotted as mean radiance (p/s/cm²/sr) over time with error bars representing SD. Groups of mice were infected and treated as in panel (B) with control (n = 6), AM-depleted (n = 8), phage-treated control (n = 9) and phage treated AM-depleted (n = 13). Statistical analysis (two-tailed Mann–Whitney U test) is indicated for the time point 22 h p.i. LOQ, limit of quantification. LOD, limit of detection. Data were assembled from 6 independent experiments. D Bacterial counts (CFUs in log scale) from lung homogenates taken from mice (same animals as in panel C) 22 h post infection. LOD limit of detection (p values from two-tailed Mann–Whitney U test are displayed). Data were assembled from 5 independent experiments.

Fig. 2. Schematic model of bacteria-phage-macrophage-neutrophils interaction in the lungs.

An immune cell close to the arrow represents its impact on the corresponding process. Created in BioRender.com. A full list of equations and parameterization are found in Supplementary Methods.

Fig. 3. Model of bacteria-phage-macrophage-neutrophils interaction in the lungs.

Individual model predictions and data for 3 mice from each group ranked in columns with corresponding headers. Bacteria are in red, phage in green, alveolar macrophage (AM) in light blue, and lung neutrophils in dark blue. Lines are model-based individual dynamics, dots are bacterial densities, triangles are phage concentrations, crossed squares are lung neutrophils and squares are AM. The light red horizontal dashed lines represent the limit of quantification (LOQ) for bacterial density (LOQ = 5.6 log₁₀ CFUeq/mL), light red circles crossed by this line at their center are below the limit of quantification (BLQ) of bacterial density and light red lines are BLQ of bacterial load model-based predictions (i.e., the discrepancy between data and model cannot be evaluated from this plot in case of BLQ data).

Fig. 4. Median dynamics with various AM-mediated phage decay rate parameter values after infection and phage treatment.

Dynamics of bacteria (red), phage (green), alveolar macrophage (AM, light blue), and lung neutrophils (dark blue) based on the model and its median parameters (Fig. 2A, Supplementary Methods), assuming no interaction between macrophages and phage except AM-mediated decay. Each panel corresponds to the different mice groups according to their immune status: A immunocompetent control (i.e., empty liposome), B AM-depleted (clodronate). Line widths increase depending on the AM-mediated decay rate (multiplied by the macrophage density): 0,1,2,4,8 and 16 times the phage decay in absence of AM, from the thinnest to the widest line. This number is also indicated above the corresponding median bacteria and phage dynamics. The AM-mediated/natural decay rate value leading to the best model (i.e., 8) is represented by solid lines, while the others are represented by dotted lines. The violin and the dots represent the distribution and values of observed CFUs. The limit of detection (LOD) for CFUs is represented by the light red horizontal dashed line (LOD = 2.4 log₁₀ CFU/mL). For AM-depletion, all the observations were below the LOD.

Fig. 5. AM-associated phage decay and adsorption inhibition assumption.

A Uninfected mice were given clodronate (blue) or EL (red) by intra-tracheal instillation (n = 3–4 each) 4 days prior phage PAK_P1 (5 × 10⁷ PFU) at t = 0 h. Lungs were collected and homogenized at indicated time points. Data fitted using a monoexponential decay model with decay rate values indicated (p values, two-tailed Mann–Whitney U test, are indicated). B Uninfected mice received intranasally PBS (n = 2) or phage PAK_P1 (5 × 10⁷ PFU) (n = 6) at t = 0 h and broncho alveolar lavage was performed 4 h post-administration to sort and count AM and quantify phage DNA by qPCR. C Counts of phage-associated fluorescent signals in individual AM collected from uninfected mice and incubated without (control) or with phage PAK_P1 in absence or presence of cytochalasin B (N = number of AM observed for each condition). Boxplots display median (center line), first and third quartiles (box limits), and whiskers extending to 1.5×IQR from the quartiles (p values, two-tailed Mann–Whitney U test, are indicated). D Representative images from AM incubated with phage PAK_P1 in presence or absence of cytochalasin B. Median dynamics with various adsorption inhibition parameter values after infection and phage treatment for (E) immunocompetent control (i.e. EL) and F AM-depleted mice. Dynamics of bacteria (red), phage (green), alveolar macrophage (AM, light blue), and lung neutrophils (dark blue) based on the model and its median parameters (Fig. 2A, Supplementary Methods), assuming phagocytosis rate value from the inference of the data of Fig. 4A. Line widths increase depending on the maximal adsorption inhibition by AM: 0%, 15%, 30% and 60%, from the thinnest to the widest line. The maximal adsorption inhibition value leading to the best model (i.e., 30%) is represented by solid lines, while the others are represented by dotted lines. The violin and dots represent the distribution and values of observations: CFUs in red and PFUs in green. The limit of detection (LOD) for CFUs is represented by the light red horizontal dashed line (LOD = 2.4 log₁₀ CFU/mL) For AM-depletion, all the observations were below the LOD.

Fig. 6. Inter-individual variability in the dynamics of bacteria, phage, alveolar macrophage and lung neutrophils.

Based on the model (Fig. 2) and its median and variability parameters (Supplementary Methods), simulations for 1000 individuals in each group are shown by areas covering interquartile interval (Q1-Q3). Each panel corresponds to the different mice groups according to their immune status and phage treatment.