Supplemental Oxygen Alters the Airway Microbiome in Cystic Fibrosis

Abstract

Features of the airway microbiome in persons with cystic fibrosis (pwCF) are correlated with disease progression. Microbes have traditionally been classified for their ability to tolerate oxygen. It is unknown whether supplemental oxygen, a common medical intervention, affects the airway microbiome of pwCF. We hypothesized that hyperoxia significantly impacts the pulmonary microbiome in cystic fibrosis. In this study, we cultured spontaneously expectorated sputum from pwCF in artificial sputum medium under 21%, 50%, and 100% oxygen conditions using a previously validated model system that recapitulates microbial community composition in uncultured sputum. Culture aliquots taken at 24, 48, and 72 h, along with uncultured sputum, underwent shotgun metagenomic sequencing with absolute abundance values obtained with the use of spike-in bacteria. Raw sequencing files were processed using the bioBakery pipeline to determine changes in taxonomy, predicted function, antimicrobial resistance genes, and mobile genetic elements. Hyperoxia reduced absolute microbial load, species richness, and diversity. Hyperoxia reduced absolute abundance of specific microbes, including facultative anaerobes such as Rothia and some Streptococcus species, with minimal impact on canonical CF pathogens such as Pseudomonas aeruginosa and Staphylococcus aureus. The effect size of hyperoxia on predicted functional pathways was stronger than that on taxonomy. Large changes in microbial cooccurrence networks were noted. Hyperoxia exposure perturbs airway microbial communities in a manner well tolerated by key pathogens. Supplemental oxygen use may enable the growth of lung pathogens and should be further studied in the clinical setting. IMPORTANCE The airway microbiome in persons with cystic fibrosis (pwCF) is correlated with lung function and disease severity. Supplemental oxygen use is common in more advanced CF, yet its role in perturbing airway microbial communities is unknown. By culturing sputum samples from pwCF under normal and elevated oxygen conditions, we found that increased oxygen led to reduced total numbers and diversity of microbes, with relative sparing of common CF pathogens such as Pseudomonas aeruginosa and Staphylococcus aureus. Supplemental oxygen use may enable the growth of lung pathogens and should be further studied in the clinical setting.

Keywords: cystic fibrosis; hyperoxia; lung; microbiome; oxygen; persons with cystic fibrosis; perturbation; pwCF.

FIG 1

Overview of study design. Each patient sputum sample generated 10 samples (9 cultured and 1 uncultured) for metagenomic sequencing. Sputum was cultured in artificial sputum medium under 21%, 50%, and 100% oxygen atmospheres, with aliquots taken at 24, 48, and 72 h. One aliquot was processed uncultured as the “source” sputum.

FIG 2

Hyperoxia reduces microbial load and community diversity. (A) Distributions of microbial load and alpha diversity metrics stratified by culture condition. (B) Estimated effect size and 95% confidence intervals for the effect of oxygen and time on microbial load and alpha diversity from linear mixed-effects regressions. (C) Balloon plot with predicted values for microbial load and alpha diversity for each oxygen and time condition.

FIG 3

Per-participant sputum community taxonomic profiles. Leftmost “source” bar corresponds to uncultured sputum samples; the remainder represent the nine culture conditions. Colors correspond to the 12 most abundant genera, with the remainder grouped in gray as “other.” (A) Relative abundance community profiles, grouped by study participant. (B) Absolute abundance community profiles, grouped by study participant. Calculated using spike-in bacteria. The y axis differs between study participants due to differences in microbial load.

FIG 4

Per-species differential effects of hyperoxia. Left, phylogenetic tree of the 30 most prevalent microbial species. Names in bold represent species identified as significantly impacted by hyperoxia. Middle, taxon-normalized conditional abundance values for each species. The highest conditional mean for each taxon is set to 100%. Right, significance level of differential effect of oxygen. Solid blue stars indicate q values of <0.1 after Bonferroni-Hochberg multiple-hypothesis correction; white stars indicate raw P values of <0.05 but q values of >0.1.

FIG 5

Associations between predicted functional pathways and CF microbes. The y axis contains the 30 most prevalent microbial taxa, and the x axis contains a curated subset of predicted functional pathways, including fermentation, the electron transport chain, respiration, and metabolism pathways. Significance annotations indicate microbes/pathways that are significantly impacted by hyperoxia after Bonferroni-Hochberg multiple-hypothesis correction with a threshold q value of <0.1; blue boxes indicate q values of <0.1, gray boxes indicate q values of ≥0.1 The main heat map plots the Spearman rank correlation of each microbe against each of the selected pathways. Dendrograms relate microbes by functional pattern and functions by microbial pattern.

FIG 6

Hyperoxia alters microbial network topology. Comparison of microbial association networks from sputum cultured under 21% and 100% oxygen. Network associations were calculated based on Spearman’s correlation from absolute abundance data and sparsified via adaptive Bonferroni-Hochberg corrected t tests with a q value threshold of 0.1. Edge weights were based on Spearman’s correlation, with synergistic (positive) relationships indicated by green lines and mediating (negative) relationships indicated by magenta lines. Each node corresponds to a bacterial or fungal species, with node color determined by association cluster and node size determined by average absolute abundance scaled using a log₁₀ transformation.