Host Response to the Lung Microbiome in Chronic Obstructive Pulmonary Disease

Abstract

Rationale: The relatively sparse but diverse microbiome in human lungs may become less diverse in chronic obstructive pulmonary disease (COPD). This article examines the relationship of this microbiome to emphysematous tissue destruction, number of terminal bronchioles, infiltrating inflammatory cells, and host gene expression.

Methods: Culture-independent pyrosequencing microbiome analysis was used to examine the V3-V5 regions of bacterial 16S ribosomal DNA in 40 samples of lung from 5 patients with COPD (Global Initiative for Chronic Obstructive Lung Disease [GOLD] stage 4) and 28 samples from 4 donors (controls). A second protocol based on the V1-V3 regions was used to verify the bacterial microbiome results. Within lung tissue samples the microbiome was compared with results of micro-computed tomography, infiltrating inflammatory cells measured by quantitative histology, and host gene expression.

Measurements and main results: Ten operational taxonomic units (OTUs) was found sufficient to discriminate between control and GOLD stage 4 lung tissue, which included known pathogens such as Haemophilus influenzae. We also observed a decline in microbial diversity that was associated with emphysematous destruction, remodeling of the bronchiolar and alveolar tissue, and the infiltration of the tissue by CD4(+) T cells. Specific OTUs were also associated with neutrophils, eosinophils, and B-cell infiltration (P < 0.05). The expression profiles of 859 genes and 235 genes were associated with either enrichment or reductions of Firmicutes and Proteobacteria, respectively, at a false discovery rate cutoff of less than 0.1.

Conclusions: These results support the hypothesis that there is a host immune response to microorganisms within the lung microbiome that appears to contribute to the pathogenesis of COPD.

Keywords: COPD; bacteria; inflammation; microbiome.

Figure 1.

(A) Operational taxonomic unit (OTU) richness as a function of alveolar surface area. (B) Ordination of samples based on Bray–Curtis dissimilarity of microbiomes. GOLD = Global Initiative for Chronic Obstructive Lung Disease; PC1 and PC2 = principal components 1 and 2, respectively. (C) Shannon diversity versus lung height between control and GOLD stage 4. (A) Alveolar surface area values and OTU richness were determined from spatially adjacent cores (R² = 0.27, P < 0.05). (B) Dissimilarity was calculated by the same approach as in C. The two groups were significantly different (permutational multivariate analysis of variance; pseudo-F = 6.58; P = 0.001). (C) Lower lung height values represent lung tissue taken closer to the apex; higher lung height values represent lung tissue taken closer to the base. There was a significant difference between control and GOLD stage 4 (P < 0.05) at the relative middle of the lung.

Figure 2.

(A) Phylum relative percent abundances in control and GOLD (Global Initiative for Chronic Obstructive Lung Disease) stage 4 lung tissue. (B) Most important species for discriminating control and GOLD stage 4 microbiomes, using random forest analysis with Boruta feature selection. (C) Heat map of the important bacterial species and relative abundance in each tissue sample. (A) The distribution of phyla was significantly different between control and GOLD stage 4 (P < 0.05), and this was driven by Proteobacteria (P < 0.05). (B) The average (±SD) 10-fold cross-validated error rate was 17 ± 2% with a per-class error rate of 34 ± 4% for the controls and 6 ± 3% for the GOLD stage 4 group. (C) Samples were clustered by similarity, and the three color-coded bars at the top represent control (red) or chronic obstructive pulmonary disease GOLD stage 4 group (blue), patient (nine distinct colors), and lung height (shades of pink, with dark pink being closer to the base and light pink closer to the apex). Dark red represents high abundance, pale blue represents low abundance, and dark blue represents undetectable values of the respective operational taxonomic unit.