Bacterial biogeography of adult airways in atopic asthma

Abstract

Background: Perturbations to the composition and function of bronchial bacterial communities appear to contribute to the pathophysiology of asthma. Unraveling the nature and mechanisms of these complex associations will require large longitudinal studies, for which bronchoscopy is poorly suited. Studies of samples obtained by sputum induction and nasopharyngeal brushing or lavage have also reported asthma-associated microbiota characteristics. It remains unknown, however, whether the microbiota detected in these less-invasive sample types reflect the composition of bronchial microbiota in asthma.

Results: Bacterial microbiota in paired protected bronchial brushings (BB; n = 45), induced sputum (IS; n = 45), oral wash (OW; n = 45), and nasal brushings (NB; n = 27) from adults with mild atopic asthma (AA), atopy without asthma (ANA), and healthy controls (HC) were profiled using 16S rRNA gene sequencing. Though microbiota composition varied with sample type (p < 0.001), compositional similarity was greatest for BB-IS, particularly in AAs and ANAs. The abundance of genera detected in BB correlated with those detected in IS and OW (r median [IQR] 0.869 [0.748-0.942] and 0.822 [0.687-0.909] respectively), but not with those in NB (r = 0.004 [- 0.003-0.011]). The number of taxa shared between IS-BB and NB-BB was greater in AAs than in HCs (p < 0.05) and included taxa previously associated with asthma. Of the genera abundant in NB, only Moraxella correlated positively with abundance in BB; specific members of this genus were shared between the two compartments only in AAs. Relative abundance of Moraxella in NB of AAs correlated negatively with that of Corynebacterium but positively with markers of eosinophilic inflammation in the blood and BAL fluid. The genus, Corynebacterium, trended to dominate all NB samples of HCs but only half of AAs (p = 0.07), in whom abundance of this genus was negatively associated with markers of eosinophilic inflammation.

Conclusions: Induced sputum is superior to nasal brush or oral wash for assessing bronchial microbiota composition in asthmatic adults. Although compositionally similar to the bronchial microbiota, the microbiota in induced sputum are distinct, reflecting enrichment of oral bacteria. Specific bacterial genera are shared between the nasal and the bronchial mucosa which are associated with markers of systemic and bronchial inflammation.

Keywords: Adult asthma; Atopy; Bronchial microbiota; Corynebacterium; Eosinophilic inflammation; Induced sputum microbiota; Lower airways; Moraxella; Nasal microbiota; Oral microbiota; Upper airways.

Fig. 1

Alpha diversity in the microbiota of different specimen types demonstrating that the upper airway harbors significantly sparser bacterial communities than the lower airways or the oral cavity. a Bacterial richness as indicated by the total number of taxa detected in each sample type. b Shannon index of bacterial diversity in each sample type. c Phylogenetic Faith’s index of bacterial diversity in each sample type. d Pielou’s index of community evenness in each sample type. Statistical significance was determined using Wilcoxon matched-pairs signed rank test

Fig. 2

Compositionally lower airway microbiota is more similar to the oral than nasal cavity. a Principal coordinate analysis (unweighted UniFrac) shows compositional dissimilarity between paired samples (linear mixed-effects model, p < 0.0001). b Mean intra-subject paired distance to BB shows that the NB microbiota are most distinct from BB (Wilcoxon matched-pairs signed rank test; Whiskers extend to 95% confidence interval). c Shorter mean intra-subject paired distance between IS and BB compared to OW and BB, suggests that the IS microbiota are more representative of BB than OW is of BB (Wilcoxon matched-pairs signed rank test; Whiskers extend to 95% confidence interval). d Shorter mean intra-subject paired distance is observed between BB and IS compared to OW in AAs and ANAs but not HCs (Paired t test). e Shorter mean intra-subject paired distance between IS and BB in AAs compared to HCs (Welch’s corrected t test; Whiskers extend to 95% confidence interval) suggests that IS microbiota are more representative of BB in asthmatic subjects. f Summary of the relative abundance of taxa identified in paired samples (n = 27) shows NB as being the most compositionally distinct. Bacterial taxonomic classification is shown at a family (genus) level

Fig. 3

Distribution of the relative abundance for most prevalent genera (present at ≥ 3% in any one of the samples) in the four paired (n = 27) specimen types, illustrates the dissimilarity in bacterial composition of NB compared to other samples. Significance was determined using Wilcoxon matched-pairs signed rank test (p > 0.05 not shown)

Fig. 4

A greater number of operational taxonomic units (OTUs) was exclusively shared between IS-BB in AAs compared to HC subjects and included taxa previously associated with asthma. a Number of OTUs associated with specific sample type in AAs, ANAs, and HC subjects. Values are shown as a median (IQR). b A greater number of OTUs was shared exclusively (in at least 20% of subjects) between IS and BB in AAs and ANAs compared to HCs, which was greater than those shared in BB-OW (Welch’s corrected t test and ^&Wilcoxon matched-pairs signed rank test; p > 0.2 not shown). c Faith’s Phylogenetic diversity (PD) of taxa shared between paired IS-BB samples trended to be higher in AAs and ANAs compared to HCs (Mann Whitney test; p > 0.2 not shown). d Frequency distribution of specific genera (present in at least 20% of participants for each group) for OTUs shared exclusively between IS-BB was distinct in AAs and HCs and included known asthma-associated taxa (Chi-square test; p > 0.05 not shown)

Fig. 5

A proportion of all the taxa detected in the BB is shared with paired NB and OW samples, with each compartment contributing distinct taxa to the bronchial community. a The number of taxa shared with BB is higher for OW compared to NB (Wilcoxon matched-pairs signed rank test). b Richness of taxa shared represents a larger proportion of the overall richness detected in BB sample for paired OW compared to NB (Wilcoxon matched-pairs signed rank). c Richness of the shared taxa between paired BB and NB was greater in AAs compared to HCs (^&Mann-Whitney test). d Distribution of specific taxa shared with paired BB is distinct between NB and OW samples in AAs. (Genera identified in at least 20% of subjects in the same sample type; Statistical significance between OW and NB genera was determined using Fisher’s exact test where ****p ≤ 0.0001, ***p ≤ 0.001, **p ≤ 0.01, *p ≤ 0.05)

Fig. 6

Nasal microbiota in asthmatic patients are associated with markers of systemic and bronchial inflammation. a Frequency distribution of Corynebacterium dominant communities in NB samples was lower in AAs (n = 18) compared to HC (n = 6) subjects (Fisher’s exact test). b The relative abundance of Corynebacterium in NB samples in AAs correlated negatively with a number of inflammatory markers of atopic asthma (Spearman correlation). c The relative abundance of Moraxella in NB samples in AAs correlated positively with a number of inflammatory markers of atopic asthma (Spearman correlation)