Architecture and matrix assembly determinants of Bordetella pertussis biofilms on primary human airway epithelium

Abstract

Traditionally, whooping cough or pertussis caused by the obligate human pathogen Bordetella pertussis (Bp) is described as an acute disease with severe symptoms. However, many individuals who contract pertussis are either asymptomatic or show very mild symptoms and yet can serve as carriers and sources of bacterial transmission. Biofilms are an important survival mechanism for bacteria in human infections and disease. However, bacterial determinants that drive biofilm formation in humans are ill-defined. In the current study, we show that Bp infection of well-differentiated primary human bronchial epithelial cells leads to formation of bacterial aggregates, clusters, and highly structured biofilms which are colocalized with cilia. These findings mimic observations from pathological analyses of tissues from pertussis patients. Distinct arrangements (mono-, bi-, and tri-partite) of the polysaccharide Bps, extracellular DNA, and bacterial cells were visualized, suggesting complex heterogeneity in bacteria-matrix interactions. Analyses of mutant biofilms revealed positive roles in matrix production, cell cluster formation, and biofilm maturity for three critical Bp virulence factors: Bps, filamentous hemagglutinin, and adenylate cyclase toxin. Adherence assays identified Bps as a new Bp adhesin for primary human airway cells. Taken together, our results demonstrate the multi-factorial nature of the biofilm extracellular matrix and biofilm development process under conditions mimicking the human respiratory tract and highlight the importance of model systems resembling the natural host environment to investigate pathogenesis and potential therapeutic strategies.

Fig 1. Bps and FHA promote infection of HBE cells.

(A) CFUs of WT, ΔbpsA-D, and ΔfhaB strains after 1 and 24 h of infection. Mean values of four biological replicates in technical duplicate are presented with standard error of the mean. Statistical differences were assessed by two-way ANOVA. *, p < 0.05; **, p < 0.005; ***, p < 0.0005. (B and C) Bacterial strains were added to empty wells (without Transwells) of a 24-well plate in either SS, DMEM, or apical HBE washes. After 1 h (B) or 24 h (C), CFUs were enumerated on BG agar. Mean and standard error of the mean of three biological replicates in technical duplicate are presented. Statistical differences were assessed by two-way ANOVA. **, p < 0.005; n.s., not significant.

Fig 2. Bp induces mucin secretion in HBE cells.

HBE cells were infected as described in the ‘Materials and Methods’ at an MOI of 10, for 24 h with different Bp strains. Fixed cells were, sectioned and stained with (A) Hematoxylin & Eosin, or (B) PAS/AB. (C) Semi-quantification of mucin production from PAS/AB-stained images. Images are representative of two biological replicates. Statistical differences were assessed by One-way ANOVA. *, p≤0.05; n.s., not significant.

Fig 3. Bp forms aggregates and clusters on HBE cells.

(A) Representative live cell images of GFP-labeled bacteria on apical surface of HBE cells after 24 and 48 h (10X objective). (B) Immunofluorescent staining of α-tubulin (cilia) and DAPI (nuclei) on fixed HBE cells infected with GFP-labeled bacteria after 48 h (10X objective). Zoomed in image of HBE cells infected with the WT strain showing colocalization of bacterial aggregates with punctate acetylated-α-tubulin staining (arrow). (C-E) Scanning electron microscopy of Bp strains 24 h after infection of HBE cells. All images are representative of three biological replicates.

Fig 4. Visualization of Bp biofilms and ECM components on HBE cells.

HBE cells infected with GFP-expressing WT (A-D), ΔbpsA-D (E-H), or ΔfhaB (I-L) strains for 24 h (A-B, E-F, I-J) or 48 h (C-D, G-H, K-L) were fixed and stained for Bps (α-PNAG, red) and eDNA (α-dsDNA, blue), then imaged using an Olympus FV3000 confocal microscope (20X objective). B, D, F, H, J, and L are zoomed in images of A, C, E, G, I, and K, respectively. Z-stacks were acquired at 1-μm intervals. IMARIS software was used to produce a shading picture of the biofilms.

Fig 5. The role of adenylate cyclase toxin in Bp biofilm formation on HBE cells.

(A) Biofilm formation after 96 h was evaluated by GFP-labeled bacteria in microtiter plates by crystal violet method. Bars indicate the mean and standard error of the mean of two independent experiments performed by triplicate. Significant differences were calculated by using one-way ANOVA and Bonferroni post hoc. ***, p < 0.0005. (B-E) HBE cells were apically infected with bacterial strains in DMEM for 1 h at MOI 10. (B) CFUs were enumerated after 24 h on BG agar. Mean and standard error of the mean of two biological replicates in technical duplicate are presented. Statistical differences were assessed by unpaired Student’s t-test. n.s., not significant. (C-D) HBE cells infected with GFP-expressing WT (C) or ΔcyaA (D) strains for 48 h were fixed and imaged using an Olympus FV3000 confocal microscope (20X objective). Z-stacks were acquired at 1-μm intervals. IMARIS software was used to produce a shading picture of the biofilms. (E) Representative live cell images of GFP-labeled bacteria on apical surface of HBE cells after 48 h (10X objective). (F) Scanning electron microscopy of bacterial strains 24 h after infection of HBE cells. For C-F, images are representative of at least two biological replicates.