Globally, most inhaled particulate matter derives from the domestic combustion of biomass fuels such as wood, charcoal, crop residue and dung [1, 2]. This Household Air Pollution (HAP) is associated with 4.3 million deaths per year from respiratory disease, including 900,000 childhood deaths from pneumonia [3–6]. Alterations in microbial populations in the lung caused by particle exposure could explain increased rates of respiratory infection in subjects exposed to HAP. Inhaled particulates are known to drive inflammation in the lung, to alter microbial binding to respiratory epithelium, and to act as vehicles delivering microbial molecules to the distal airways.

In healthy lungs, previously thought to be sterile environments, communities of bacteria, together with fungi and viruses form the microbiome. Variations in this microbiome may either reflect or drive mucosal inflammation and immune function [7]. Extensive sequencing of bacterial 16S rDNA from the lungs of healthy individuals has revealed the common presence of phyla such as Proteobacteria, Bacteroidetes, Actinobacteria and Firmicutes [8].

Study of the alterations in lung microbiota resulting from environmental exposures is a nascent field with the potential to explain the pathophysiological mechanisms of lung disease. Cigarette smoking does not appear to be associated with significant changes in the lung microbiome in a US cohort [9]. However, the effects of exposure to other environmental sources of particulates, particularly in low income countries, have not yet been described.

We hypothesised that, in healthy people, HAP would be associated with alterations in the lung microbiome in terms of diversity, richness and the relative abundance of various microbial taxa. Furthermore, prior comparative analysis of geographical differences in microbiota prevalence identified Petrobacter in more than a third of a group sampled in Malawi, but none in a US cohort. Petrobacter is a gram negative, aerobic bacterium identified in 2004 from oil reservoir samples. Given this niche, we hypothesised that the prevalence in a Malawi group would be due to inhalation of smoke from biomass fuel use: we present evidence of this association in this paper [10, 11].

Our study found that high particulate exposure as defined by alveolar macrophage carbon content was associated with altered relative abundance of bacteria within the lungs of healthy Malawian adults. Specifically, we report higher proportions of Neisseria and Streptococcus, and lower proportions of Tropheryma. However, lung microbiome diversity and the relative abundance at the phyla level were not significantly different based on our current sample size. Petrobacter was more abundant in individuals who used biomass fuels for cooking or lighting than in those who did not. Finally, particulate matter in alveolar macrophages did not directly correlate with biomass exposure as determined by questionnaire, suggesting in this small study that other factors contributed to alveolar macrophage particulate ingestion. These likely include ingress of pollution from neighbouring homes, traffic pollution when commuting to work and occupational exposures.

This study provides a novel analysis of the lung microbiome from individuals in a low income country with exposure to high biomass fuels. We characterised high particulate exposures according to the particulate burden in the cells of the distal airways, which reflects the cumulative exposure to respirable-size particulates. This is likely to be more relevant to changes in the microbiome than air sampling methods [23]. We did not find any significant differences in the lung microbiome as a whole between high and low particulate groups. However, both richness and abundance of the lower respiratory microbiome may be significantly altered by sampling techniques which vary in their potential for introducing “carryover” contamination from the upper airways [24]. Individual lower respiratory tract microbiome demonstrates less similarity to that in other individuals than to the upper respiratory tract of the same person [9, 25, 26]. Nevertheless, samples obtained directly from explanted lungs and those taken indirectly by bronchoscopy demonstrate similar patterns [24]. Our study did not include sampling of the upper airway. Therefore, one cannot determine conclusively whether our BAL findings reflect true differences in the lung microbiome or differences in upper airway carriage between subjects with low and high exposure to particulates. However, since micro-aspiration is common in humans [17], it is likely that a true lung microbiome will contain many of the same taxa found in the oral cavity. Our study was performed on stored samples: not all of the potential environmental controls were available. However, since all subjects underwent the same bronchoscopy protocol, findings between high and low biomass fuel groups are highly likely to represent non-artefactual differences. Sterile saline controls from the bronchoscope using the same kit as this study have been reported, and contain a very low number of reads [27].

Reduced diversity of the lung microbiome has been seen in disease states, such as COPD, [24] and with medical treatments, such as inhaled corticosteroids [28]. In our HIV-negative, healthy volunteers, however, we found no significant differences in diversity or richness between high and low particulate groups, mirroring findings of comparisons between smokers and non-smokers in the US [9]. Our study is compatible with prior descriptions of high diversity in the lung microbiome, with Streptococcus and Prevotella most frequently represented [9, 24].

Despite the absence of differences in the global lung microbiome between high and low particulate exposed subjects, interesting differences in specific taxa were observed. Streptococcus was more abundant in high particulate exposed participants after adjustment of potential confounders. In humans, ambient air pollution increases the risk of pneumonia in adults and children, [29, 30] and is commonly caused by Streptococcus pneumoniae. Analagous changes occur in cigarette smokers [31]. Bronchial epithelium exposed to urban particulates demonstrates increase expression of platelet activating factor and S. pneumoniae binding [32]. Nasopharyngeal carriage of S. pneumoniae is high in Malawi, occurs early, and pathogen specific mucosal T-cell regulation may contribute to prolonged carriage [33]. High levels of Streptococcus have also been demonstrated in the upper airway microbiome in a case-control study of infants in Ecuador [34]. Similarities with the Malawian study are low levels of pneumococcal immunisation and antibiotic use, and significant rural poverty. Taken together, these studies would provide a mechanistic link between our findings and the epidemiological associations of particulate concentration and pneumonia incidence.

Neisseria was also more prominent in high particulate group. This organism is not usually considered in the context of the lower airways, although serogroup Y has been associated with pneumonia in the elderly and Army recruits [35]. Nasopharyngeal carriage of N. meningitidis is increased amongst those exposed to cigarette smoke, [36] and ex vivo human epithelial cell models demonstrate bacterial binding is also increased [37]. The role of other ambient particulates is not well defined.

Tropheryma showed the inverse relationship, in that it was more frequently represented in the low particulate group. The pathogenic significance of this is uncertain: while T. whipplei is the aetiological agent of Whipple’s disease, it is commonly found in health in other studies of lung microbiome [9]. It occurs in especially high levels in microbiome studies of HIV-infected individuals from the US [38]. However, the lower levels in our particulate exposed healthy volunteers is unexplained.

While we hypothesised that particulate exposure could alter the microbiome, it is also possible that difference in the microbiome could affect particulate uptake in macrophages through effects on immune activation and clearance responses [39].

There are significant differences in the composition of the lung microbiome of Malawians with differing levels of particulate exposure as determined by macrophage carbon content. These differences might contribute to the excess respiratory infections associated with particulate exposure. We have further demonstrated that the finding of Petrobacter in the lung is associated with household biomass fuel exposure. Interventions to improve air quality have the potential to alter the microbiome of the lower respiratory tract and ultimately improve the lung health of Malawians.