Background
For the 35 million people living with human immunodeficiency virus (PLWH), antiretroviral therapy (ART) has been a lifesaving measure. As a result, AIDS-related morbidity and mortality have decreased markedly; however, aging with HIV has brought other challenges [
1]. For instance, PLWH are more likely to develop chronic obstructive pulmonary disease (COPD) [
2] and are also more likely to suffer from severe respiratory symptom burdens even after adjustment for smoking habits [
3]. Although the pathogenesis of accelerated COPD in PLWH is poorly understood, the unique risk for pulmonary infections in this setting suggests that shifts in the lung microbiome might account for this phenomenon.
Investigations into the HIV lung microbiome have yielded interesting insights but no clear consensus. Lozupone et al. found that the abundance of
Tropheryma whipplei was significantly increased in bronchoalveolar lavage (BAL) samples of PLWH compared with HIV- control subjects [
4], while a second study by Beck et al. showed no differences between the two groups [
5]. A third study also evaluating BAL demonstrated that PLWH who had advanced disease (CD4 cell counts < 500 cells/mm
3) had significantly reduced microbiome diversity when compared to HIV- controls, with diversity starting to return to normal levels once ART was initiated [
6]. While these studies have offered the first insights into the HIV lung microbiome, the reliance on BAL fluid may fail to identify important changes at the specific initial site of injury in the pathogenesis of COPD, namely the small airway [
7]. The small airway epithelium (SAE) is the first line of defense against toxins such as cigarette smoke and microbial pathogens. In COPD, remodeling of this layer with squamous metaplasia, goblet cell hyperplasia, and breakdown of the epithelial barrier junction are critical to injury development [
8]. Moreover, evidence that endotoxins produced by
Staphylococcus aureus and
Haemophilus influenza can damage epithelial barrier function suggests that an important relationship between the microbiome, epithelial injury, and COPD may exist [
9]. Previous work by our group identified that within PLWH, the absence of Pasteurellaceae and
Brachybacterium and the presence of
Yersinia species in the SAE could help identify those with COPD [
10]. Our study explores whether significant differences exist between the SAE microbiomes of PLWH and uninfected controls.
Discussion
In this first comparison of the HIV and non-HIV SAE microbiome, we discovered that PLWH, had significantly lower bacterial loads, microbial diversity, and species richness compared to HIV- controls. While this observation may seem contrary to what one would expect in a disease associated with immunosuppression and frequent pulmonary infections, it is in fact consistent with numerous microbiome studies comparing disease and non-disease states. Specifically, regardless of the organ examined, disease states are often associated with lower microbial diversity, suggesting that a certain degree of diversity is a hallmark of healthy tissue [
24‐
27]. In the lung, for instance, chronic respiratory conditions such as cystic fibrosis [
28,
29] and COPD [
30] are associated with lower alpha diversity in sputum and lung samples. Disease severity in the lung also appears to track inversely with diversity [
30‐
32]. In similar fashion, HIV and/or the repeated antibiotic exposures these patients may have experienced due to frequent infections may lower SAE diversity. Supporting this theory are the recent findings by Twigg et al. that patients with uncontrolled HIV have significantly decreased BAL diversity compared with HIV-uninfected controls [
6].
The connection between decreased diversity and dysbiosis to COPD pathogenesis is likely multifactorial. In fact, this relationship may be different between HIV+ and HIV- groups, with only the latter showing a correlation between decreased diversity and reduced FEV1/FVC. We were unable to demonstrate significant microbiome differences between patients with and without COPD, regardless of their HIV status. One possible explanation may be that our study was underpowered to detect any difference. Larger studies examining the relationship between airflow obstruction, HIV, and the microbiome are warranted. Targeted brushings in areas of advanced emphysema compared with brushings taken from normal lung within the same PLWH may also help us understand what role the microbiome may play in COPD pathogenesis. Linking the microbiome with metabolomic, transcriptomic, and epigenetic modifications will provide additional clues as to how dysbiosis can set the stage for progressive airflow obstruction. In a previous study by our group looking specifically at the SAE microbiome and its associated transcriptome in HIV [
10], we demonstrated that the abundance of Firmicutes was negatively correlated with the expression of cilia-related genes and positively correlated with the expression of immune response genes.
Haemophilus species were also negatively correlated with the expression of cilia-related genes. These genetic pathways could be critical in the pathogenesis of accelerated COPD in HIV; however, direct comparisons of these relationships to those observed in HIV- subjects are further required.
Interestingly, the SAE phyla distribution in PLWH was markedly different from that in HIV- controls in a pattern that was reminiscent of the differences previously noted between COPD and non-COPD lungs in an HIV- cohort. We found that PLWH had an increase in Proteobacteria and decreases in both Bacteroidetes and Firmicutes compared with HIV- controls. Similarly, Sze et al. found that lung tissue from GOLD Stage 4 COPD patients had increased Proteobacteria and decreased Bacteroidetes and Firmicutes compared to control lungs [
30]. For PLWH, changes in the abundance of these three phyla may represent an early stage in the COPD development. Longitudinal studies evaluating the progression of phyla distribution from healthy to diseased lungs may help to clarify this association.
A novel aspect of this study was the use of SAE cells to investigate the unique lung microbiome in HIV. Previous studies have largely focused on BAL fluid, a useful compartment with which to identify generalized inflammation in the lung but not one that necessarily provides specific information on the pathogenesis of COPD. Profound structural changes occur in the SAE in COPD, including squamous metaplasia, ciliary dysfunction, mucous cell hyperplasia, and the breakdown of apical junctional barriers [
8]. Even prior to the onset of overt COPD, smoking-related changes in the airway epithelium can be observed. These include senescent signatures determined by telomere length and growth differential factor 15 production [
33,
34] and gene expression alterations along immunity and oxidative stress pathways [
35]. SAE changes in HIV have not yet been fully characterized, although two studies have recently shed greater light. In one study, the presence of X4 tropic HIV increased both epithelial cell layer permeability and the expression of pro-inflammatory cytokines [
36]. In another study, HIV was found to bind to airway epithelial basal cells, resulting in a tissue-destructive phenotype [
37]. Whether or not the distinct microbiome of the HIV SAE plays a role in these processes is certainly a question worth pursuing in future experiments.
There are a number of limitations noted in our study. First, contamination by oropharyngeal and environmental elements is always a concern in a lung microbiome study in which specimens are obtained via bronchoscopy. This is particularly true for organs with relatively low microbial biomass such as the lung. Ideally, reagent samples and oral and bronchoscope channel washes prior to the procedure would have helped to identify potential contaminants in bronchial brushings [
38]. Nonetheless, all bronchoscopies were performed with no suction used upon insertion of the bronchoscope to avoid oral and large airway contamination. Second, patients enrolled had other pulmonary concerns, including lung masses, nodules, and pneumonia. While these could conceivably pose as confounders, we did not find significant diversity differences between those with and without these conditions. This was likely due to the fact that sample acquisition took place specifically in lobes of the lung away from clinically important lesions. Further studies evaluating the microbiome in asymptomatic PLWH will be necessary moving forward to confirm our findings.
Acknowledgements
The authors would like to acknowledge Fernando Studart for his assistance in editing the manuscript.
An abstract version of this manuscript was presented at the American Thoracic Society Conference in San Francisco, CA, in May of 2016.