Transcriptome Analysis of Pseudomonas aeruginosa Biofilm Infection in an Ex Vivo Pig Model of the Cystic Fibrosis Lung

Abstract

Pseudomonas aeruginosa is the predominant cause of chronic biofilm infections that form in the lungs of people with cystic fibrosis (CF). These infections are highly resistant to antibiotics and persist for years in the respiratory tract. One of the main research challenges is that current laboratory models do not accurately replicate key aspects of a P. aeruginosa biofilm infection, highlighted by previous RNA-sequencing studies. We compared the P. aeruginosa PA14 transcriptome in an ex vivo pig lung (EVPL) model of CF and a well-studied synthetic cystic fibrosis sputum medium (SCFM). P. aeruginosa was grown in the EVPL model for 1, 2, and 7 days, and in vitro in SCFM for 1 and 2 days. The RNA was extracted and sequenced at each time point. Our findings demonstrate that expression of antimicrobial resistance genes was cued by growth in the EVPL model, highlighting the importance of growth environment in determining accurate resistance profiles. The EVPL model created two distinct growth environments: tissue-associated biofilm and the SCFM surrounding tissue, each cuing a transcriptome distinct from that seen in SCFM in vitro. The expression of quorum sensing associated genes in the EVPL tissue-associated biofilm at 48 h relative to in vitro SCFM was similar to CF sputum versus in vitro conditions. Hence, the EVPL model can replicate key aspects of in vivo biofilm infection that are missing from other current models. It provides a more accurate P. aeruginosa growth environment for determining antimicrobial resistance that quickly drives P. aeruginosa into a chronic-like infection phenotype. IMPORTANCE Pseudomonas aeruginosa lung infections that affect people with cystic fibrosis are resistant to most available antimicrobial treatments. The lack of a laboratory model that captures all key aspects of these infections hinders not only research progression but also clinical diagnostics. We used transcriptome analysis to demonstrate how a model using pig lungs can more accurately replicate key characteristics of P. aeruginosa lung infection, including mechanisms of antibiotic resistance and infection establishment. Therefore, this model may be used in the future to further understand infection dynamics to develop novel treatments and more accurate treatment plans. This could improve clinical outcomes as well as quality of life for individuals affected by these infections.

Keywords: Pseudomonas aeruginosa; RNA sequencing; RNAseq; antibiotic resistance; antimicrobial resistance; biofilm; biofilms; chronic infection; cystic fibrosis; ex vivo model; quorum sensing; transcriptome.

FIG 1

Initial investigation of the Pseudomonas aeruginosa PA14 transcriptome across 3 environments: in vitro synthetic cystic fibrosis sputum media (SCFM) and the two niches in the ex vivo pig lung model: lung tissue surface (lung) and the SCFM surrounding the lung tissue (surrounding SCFM), based on the whole genome (n = 5829 genes). (A) Principal-component analysis (PCA) considering all genes. Each environment is shown by a different color and each time point shown by different shaped points (see key). The 95% confidence ellipses are shown. Individual data points represent each RNA sample; three tissue pieces from each of two independent pig lungs were used for the EVPL model environments at each time point, and three replica in vitro SCFM cultures were sequenced per time point. (B) Heatmap showing hierarchical clustering analysis and the Pearson’s correlation coefficient value between each sample (all r > 0.9). The P. aeruginosa PA14 growth environment and infection time for each sample is shown by different combinations of colors (see key). (C) Venn diagrams of the number of significant differentially expressed P. aeruginosa PA14 genes (DEGs) from each contrast, using threshold values of P < 0.05 and log₂ fold change ≥ |1.5|. The shared DEGs are genes that are either underexpressed or overexpressed in both the lung and surrounding SCFM versus in vitro SCFM. Genes that were significant DEGs in both contrasts at each time point, but in opposite directions, are not considered to be shared between both contrasts. The full list of significant DEGs is provided in the data supplemental material.

FIG 2

Significantly enriched Pseudomonas aeruginosa PA14 biological processes gene ontology (GO) terms (P < 0.05, log₂ fold change ≥ |1.5|) from contrasts between growth from each of the two ex vivo pig lung model locations: the lung tissue (lung) and the synthetic cystic fibrosis sputum media (SCFM) surrounding the lung tissue (surrounding SCFM), and in vitro SCFM. The analysis was performed on samples from 24 h postinfection (A, B) and 48 h postinfection (C, D). Each graph shows the significantly enriched biological processes GO terms for the particular contrast on the y axis and fold change enrichment on the x axis. The fold change enrichment is the fold difference in expression of significant differentially expressed genes (DEGs) in the analysis associated with that GO term than expected by random chance. Green bars show terms where the associated significantly DEGs were either overexpressed or underexpressed, showing that the process was affected irrespective of the direction of expression. The red bars show terms where all associated significant DEGs in the contrast were underexpressed, and blue bars show terms where all associated significant DEGs were overexpressed.

FIG 3

The log₂ fold change (LFC) in expression of 42 Pseudomonas aeruginosa quorum sensing genes, controlled by the las regulon, conserved in human infection. The expression of P. aeruginosa PA14 grown on the lung tissue of the ex vivo pig lung tissue (Lung) versus in vitro synthetic cystic fibrosis media (SCFM) at 24 h (A) and 48 h (B) postinfection are shown by the purple bars. Each graph also includes expression of the gene set by P. aeruginosa from human cystic fibrosis sputum versus in vitro conditions taken from Cornforth et al. (8), shown by the black bars. The locus tags shown are for P. aeruginosa PA14 with gene names in bold where appropriate. Bars with the striped fill are not significantly differentially expressed for that contrast (P < 0.05, LFC ≥ |1.5|). The dashed lines represent the threshold LFC value for differential expression.

FIG 4

The log₂ fold change (LFC) of genes of interest associated with antibiotic resistance predicted by the Comprehensive Antibiotic Resistance Database (CARD) (26). Comparisons of gene expression were performed for the two environments of the ex vivo pig lung model: the lung tissue and surrounding synthetic cystic fibrosis media (SCFM) versus in vitro SCFM at two time points (24 h and 48 h postinfection). The dashed lines represent the threshold LFC value for a gene to be considered significantly differentially expressed (LFC ≥ |1.5|), and comparisons where this was statistically significant (P < 0.05) are denoted with a *. Each bar color represents a different comparison and time point (see keys). (A–C) The LFC values for each comparison of efflux pumps and any repressors where significant expression differences were found in at least one comparison. (D–E) The LFC values for each comparison of individual genes where significant expression differences were found in at least one comparison.