Infection Is Not Required for Mucoinflammatory Lung Disease in CFTR-Knockout Ferrets

Abstract

Rationale: Classical interpretation of cystic fibrosis (CF) lung disease pathogenesis suggests that infection initiates disease progression, leading to an exuberant inflammatory response, excessive mucus, and ultimately bronchiectasis. Although symptomatic antibiotic treatment controls lung infections early in disease, lifelong bacterial residence typically ensues. Processes that control the establishment of persistent bacteria in the CF lung, and the contribution of noninfectious components to disease pathogenesis, are poorly understood.

Objectives: To evaluate whether continuous antibiotic therapy protects the CF lung from disease using a ferret model that rapidly acquires lethal bacterial lung infections in the absence of antibiotics.

Methods: CFTR (cystic fibrosis transmembrane conductance regulator)-knockout ferrets were treated with three antibiotics from birth to several years of age and lung disease was followed by quantitative computed tomography, BAL, and histopathology. Lung disease was compared with CFTR-knockout ferrets treated symptomatically with antibiotics.

Measurements and main results: Bronchiectasis was quantified from computed tomography images. BAL was evaluated for cellular differential and features of inflammatory cellular activation, bacteria, fungi, and quantitative proteomics. Semiquantitative histopathology was compared across experimental groups. We demonstrate that lifelong antibiotics can protect the CF ferret lung from infections for several years. Surprisingly, CF animals still developed hallmarks of structural bronchiectasis, neutrophil-mediated inflammation, and mucus accumulation, despite the lack of infection. Quantitative proteomics of BAL from CF and non-CF pairs demonstrated a mucoinflammatory signature in the CF lung dominated by Muc5B and neutrophil chemoattractants and products.

Conclusions: These findings implicate mucoinflammatory processes in the CF lung as pathogenic in the absence of clinically apparent bacterial and fungal infections.

Keywords: airway obstruction; cystic fibrosis; ferret; inflammation; mucus.

Figure 1.

Cystic fibrosis (CF) ferrets develop bronchiectasis despite lifelong maintenance of antibiotics. (A–H) Representative computed tomography images acquired from non-CF (A–D) and CF (E–H) ferrets maintained on three antibiotics for life (age range of images shown: 515–1,025 d). In each case, inset panels (B–D and F–H) show representative details of three core findings that are diagnostic of bronchiectasis: increased airway wall thickness (B and F) (marked by solid lines), increased ratio of airway to vessel diameter (C and G) (marked by solid lines), and distance of the cartilaginous airways to the pleural surface (D and H) with the most distal airway marked by arrows. All images quantified are from age-matched ferrets and were acquired under identical conditions. (I–K) Blinded quantification of the previously described parameters from computed tomography examinations performed while on antibiotics. Each data point represents one ferret (n = 7; P value by paired two-sided Wilcoxon rank sum test). (I) Wall thickness in CF versus non-CF control animals, P = 0.016. (J) Ratio of airway diameter to that of adjacent vessel, P = 0.031. (K) Distance from cartilaginous airways to pleural surface, P = 0.047. Graphs show mean ± SEM. *P < 0.05.

Figure 2.

Lifelong maintenance of antibiotics prevents microbial colonization of the cystic fibrosis (CF) ferret lung and prolongs survival. (A) Colony-forming units (CFU) per milliliter in BAL in non-CF and CF ferrets maintained on antibiotics (n = 17 for seven paired animals; mixed effects negative binomial modeling using R package glmmTMB; the ratio of non-CF to CF is 0.76) (95% confidence interval [CI], 0.19–3.04; P = 0.69). (B and C) Number of copies of bacterial genome, as assessed by analysis of 16S ribosomal DNA, in BAL supernatant (B) and pellet (C) (n = 17 for seven paired animals for both). Based on mixed effects negative binomial modeling, for BAL supernatant (B), the ratio of non-CF to CF is 0.7 (95% CI, 0.29–1.87; P = 0.48). In the BAL pellet (C), the ratio is 0.79 (95% CI, 0.35–1.78; P = 0.57). (D) Number of copies of fungal internal transcribed spacer-1 per milliliter BAL, as assessed by quantitative PCR (n = 6–7 animals per genotype; P value by two-sample Wilcoxon rank sum test, P = 0.5). (E) Bacterial CFU/mg protein of lung lysates from non-CF and CF ferrets treated with two different antibiotic regimens: group 1 (n = 9–10; P value by two-sample Wilcoxon rank sum test, P < 0.0001) was treated symptomatically based on weight changes, whereas group 2 (n = 6; P value by paired Wilcoxon rank sum test, P > 0.9999) was reared on continuous antibiotics from birth. (F) Kaplan-Meier survival curves of CF ferrets reared on the different antibiotic regimens as described in E, P value by log-rank test. Graphs in A–E show mean ± SEM. ITS1 = internal transcribed spacer-1.

Figure 3.

BAL from cystic fibrosis (CF) ferrets maintained on antibiotics exhibits a neutrophil-predominant mucoinflammatory signature despite normal levels of microbes in the lung. (A) Photographs of nonsoluble material in CF and non-CF BAL samples, as indicated. Arrows in main panels and arrowheads in insets indicate the top of the nonsoluble material pellet. Inset, when present, provides a contrasted image to emphasize the solid pellet. (B, E, H, and I) Differential of cell types in BAL from non-CF and CF ferrets (n = 13–14 for seven paired animals; P value by mixed effect β modeling [B, E, and I] or linear mixed effects model [H], *P < 0.05 for neutrophils). (B) Percentage of neutrophils in BAL had an estimated difference in proportion for non-CF versus CF of −0.84 (95% confidence interval [CI], −1.51 to −0.16). Active neutrophil elastase in the BAL supernatant (C) and cell pellet (D) (n = 7 animals for each genotype; P value by linear mixed effects model, P = 0.48 and 0.15, respectively). (E) Macrophage percent in BAL had an estimated difference in proportion for non-CF versus CF of 0.1 (95% CI, 0.00 to 0.21; P = 0.061). (F) Macrophage morphology seen in BAL cell pellets (by Diff-Qwik stain). Scale bar = 25 μm for both micrographs. (G) Percent of macrophages in BAL exhibiting activated phenotype in non-CF and CF samples (n = 7 animals for each genotype; P value by linear mixed effects model, **P = 0.0085). Estimated change from CF to non-CF was −33.1 (95% CI, −53.82 to −12.46). (H) Lymphocyte percent in BAL had an estimated difference in proportion for non-CF versus CF of 0.005 (95% CI, −0.004 to 0.015; P = 0.28). (I) Total cell counts for BAL in non-CF and CF BAL with an estimated ratio for non-CF versus CF of 0.78 (95% CI, 0.42 to 1.46; P = 0.43). Graphs show mean ± SEM.

Figure 4.

BAL from cystic fibrosis (CF) ferrets maintained on antibiotics exhibits a mucoinflammatory proteomic signature despite normal levels of microbes in the lung. (A) Volcano plot of more than 275 proteins whose levels were evaluated by quantitative proteomics of BAL from non-CF and CF ferrets (open circles, not statistically significant; solid circles, statistically significant; n = 13 samples from seven animals of each genotype; P value by Scaffold software t test). (B and C) Proteins from A with statistically significant P value (P < 0.05) and log₂ (fold change) CF/non-CF of ≥2 (B) or ≤−2 (C). Asterisk identifies neutrophil-associated proteins in BAL proteome. Heatmap depicts the log₂ (fold-change) in protein content of the BAL with red increased and blue decreased in CF. See Table E4A for all proteins identified, log₂ fold-change, and P value. Ingenuity pathway analysis was conducted on the proteome identified in A and arranged by disease (D) and canonical (E) pathways. Shown are selected top pathways ranked by increasing statistical significance. See Tables E4B and E4C for all pathways identified by ingenuity pathway analysis. COPD = chronic obstructive pulmonary disease; FC = fold change; ROS = reactive oxygen species.

Figure 5.

Cystic fibrosis (CF) ferrets maintained on antibiotics still develop mucus obstruction and inflammation in the lung despite lack of bacterial colonization. (A and B) Sections of large airways and adjacent alveoli from non-CF ferrets maintained on antibiotic therapy, stained with hematoxylin and eosin. (C–E) Sections from large airways of CF ferret stained with hematoxylin and eosin (C) or diastase-resistant periodic acid–Schiff (D and E). Mucus is marked by asterisk. (F and G) Sections of large (F) and small (G) airways from CF ferrets stained with hematoxylin and eosin, revealing inflammatory cells (arrows) and mucus (asterisk) within the lumen. Boxed regions in F and G are enlarged in panels H and I, respectively. Scale bars: A–F, 100 μm; G–I, 50 μm.

Figure 6.

Semiquantitative analysis of histopathology demonstrates that mucoinflammatory cystic fibrosis (CF) lung disease develops without bacterial infection. (A) Histopathologic sections that represent severity scores (indicated in left lower corner) assigned to features of inflammation and mucus accumulation (specific components are listed to the left of the panel). Scale bars in scores 0 and 1 are 500 μm, while scale bars in scores 2 and 3 are 200 μm. Inset scale bars are 20 μm in all panels. (B) Inflammation and mucus scores for primary cohort of six older CF ferrets and active non-CF control animals maintained on lifelong bacterial eradication therapy (CF continuous). One CF ferret had lungs that could not be scored for inflammation because of comorbid tumor (n = 5–6 ferrets per group; P value by two-sample Wilcoxon rank sum test, *P < 0.05). (C) Scores for comparison of younger animals that either received symptomatic antibiotic therapy (group 1) or continuous antibiotic regimens (group 2); non-CF ferrets are age-matched as best possible to provide comparison (sample size as indicated; P value by two-sample Wilcoxon rank sum test for the indicated groups, *P < 0.05 and **P < 0.01). (D) Scores from all CF and non-CF ferrets in B and C plotted against age in days. Linear regression was calculated for each group and displayed on the graphs. Note that points were slightly offset only to better display overlapping data. Graphs show mean ± SEM.