Comments
Abstract
The lung microbiome is associated with host immune response and health outcomes in experimental models and patient cohorts. Lung microbiome research is increasing in volume and scope; however, there are no established guidelines for study design, conduct, and reporting of lung microbiome studies. Standardized approaches to yield reliable and reproducible data that can be synthesized across studies will ultimately improve the scientific rigor and impact of published work and greatly benefit microbiome research. In this review, we identify and address several key elements of microbiome research: conceptual modeling and hypothesis framing; study design; experimental methodology and pitfalls; data analysis; and reporting considerations. Finally, we explore possible future directions and research opportunities. Our goal is to aid investigators who are interested in this burgeoning research area and hopefully provide the foundation for formulating consensus approaches in lung microbiome research.
Microbiome research represents a new medical science frontier. It is becoming increasingly apparent that the human microbiome influences health and disease to a comparable degree as genetics or environmental exposures (1). Microbiome research is inherently a “multi- ‘omics” endeavor generating “big data,” which can be used for future re- or meta-analyses. However, as with every nascent and evolving research field, there are still no established guidelines or widely adopted best practices for research methodology, resulting in significant problems of reproducibility and data synthesis across studies. Conversely, we have a tremendous opportunity to establish early consensus on salient methods for microbiome research, thus setting the foundation for significant future scientific returns. This review crystallizes our joint experience in lung microbiome research and aims to serve as a primer for interested scientists who want to engage more deeply in this field. Here, we address conceptual modeling and hypothesis framing, study design, experimental methodology and pitfalls, analysis, and reporting considerations. Finally, we explore possible future directions and research opportunities. We consciously focus the scope of this discussion on the lower airway microbiome; this should not detract from the important consideration of the upper airway (nasal and oral) microbiome and its influence on lung health or disease. This work is not meant to be prescriptive or proscriptive; rather, it provides the foundational platform upon which the lung research community can build a consensus approach, which will greatly enhance the quality and utility of published work.
The discovery of the lung microbiome—detectable in health, altered in disease, and correlated with inflammation—has broadened our conceptual model of respiratory disease pathogenesis. In addition to understanding direct effects of injurious exposures on lung biology, we must now consider dynamic and diverse microbial communities from the pharynx to the alveoli. This has created a web of interactions, making it difficult to determine cause and consequence behind an observed correlation (Figure 1A). Cross-sectional, observational studies cannot, alone, determine whether the lung microbiome plays a causal role in lung disease, or if it is an unimportant bystander, yet these interactions may be untangled via the articulation of three core hypotheses of lung microbiome studies (Figures 1B–1D).
Figure 1. A conceptual framework for the lung microbiome’s role in respiratory disease. (A) The discovery of the lung microbiome—and its disruption in respiratory disease—has broadened our model of pathogenesis, creating a web of potential associations with undetermined causal relationships. This complexity can be reduced to three key testable hypotheses that permit experimental validation. (B) Hypothesis 1: some exposures directly alter lung microbiota, mediating inflammation and injury. Likely examples include aspiration, sepsis (via gut–lung translocation), and inhaled oxygen. (C) Hypothesis 2: lung inflammation and injury alter the respiratory ecosystem, selectively favoring the outgrowth of some bacteria. Likely examples include toxic inhalations, viral infections, and inhaled medications. (D) Hypothesis 3: once dysbiosis and inflammation and/or injury are established, they perpetuate each other, prolonging respiratory disease even after the providing injury is gone. Likely examples include the persistent dysbiosis after the onset of an exacerbation of chronic airway disease. Reprinted by permission from Reference 163.
[More] [Minimize]The first hypothesis is that exposures (e.g., antibiotics [2–4]), disease processes, such as sepsis (via translocation of gut bacteria [5, 6]) or gastroesophageal reflux and aspiration (7, 8), directly alter the lung microbiome, thereby mediating inflammation and injury (Figure 1B). Animal work has demonstrated that the lung microbiome can be influenced by environmental conditions (2, 9), suggesting a potential novel pathway by which environmental exposures mediate lung disease.
The second hypothesis (Figure 1C) reflects the converse relationship: lung disease alters lung microbiota via perturbation of the respiratory ecosystem, which is normally inhospitable to bacterial reproduction (10). In disease, microbial growth conditions in the lungs are radically altered by the influx of nutrient-rich mucus and edema, the establishment of oxygen gradients, an increase in bacterial growth–promoting inflammatory molecules (11, 12), and impairment of local host defenses (13, 14). Even in the absence of specific exposures, genetic polymorphisms associated with lung disease (e.g., MUC5B [mucin 5B], CFTR [cystic fibrosis (CF) transmembrane conductance regulator]) likely alter the respiratory microenvironment (e.g., via increased mucus abundance), thus altering lung microbiota. In this hypothesis, the altered microbiome of diseased lungs is therefore a consequence, not a cause, of respiratory pathology.
Finally, a third hypothesis (Figure 1D) posits that, once both lung dysbiosis and lung disease are established, they perpetuate each other in a positive-feedback loop (14, 15). Altered microbiota provoke airway and alveolar inflammation via interactions between pathogen-associated molecular patterns and pattern recognition receptors, resulting in an inflammatory cascade that, in turn, changes the pulmonary ecosystem. This host–microbiome interaction may be subtle and remain undetected until the organism is challenged. This hypothesis could explain clinical phenomena, such as the inflammatory features of acute lung disease, or exacerbations of chronic airway disease outlasting their instigating triggers (e.g., influenza-induced acute respiratory distress syndrome [ARDS] and virus-provoked chronic obstructive pulmonary disease [COPD] exacerbations that persist long after the instigating virus is no longer detectable). This hypothesis may also explain why similar patients with identical exposures exhibit such wide variation in severity and duration of their lung injury.
In designing experiments and analyses of the microbiome to test these core hypotheses, it is crucial to consider whether the exposure in question is acting in isolation, or among other confounding influences. Table 1 lists common, important sources of unintentional confounding in lung microbiome studies.
| Potential Source of Confounding | Comment | |
|---|---|---|
| Human cohort studies | Severity of disease | Dysbiosis of lung microbiota is often correlated with severity of disease; microbiota may be a cause, effect, or bystander to disease progression or inflammation |
| Antibiotic exposure | Antibiotics alter lung microbiota in humans and animals (2–4); a common confounder in observational human studies | |
| Genetic factors | Host genetic variability (e.g., polymorphisms in host defense or mucociliary elevator genes like TOLLIP, MUC5B, CFTR, or alpha-1 antitrypsin) may impact the microbiome | |
| Immunosuppression | Immunosuppression likely alters lung microbiota in humans (45); a common confounder in observational human studies | |
| Demographic and geographic factors | Location, diet, sex/sex, and seasonality may alter the microbiome | |
| Extrapulmonary microbiota | Gut microbiota likely influence lung immunity (164–166) but paired specimen studies (lung and gut) are lacking | |
| Animal studies | Cohousing | Cohoused mice demonstrate a time-dependent convergence in their lung microbiota (2); if mice are housed by experimental group, “cage effect” may result in false-positive findings |
| Sex | Animal sex likely influences lung and gut microbiota (167); rarely considered in murine studies owing to cohousing concerns | |
| Age | Age likely influences the lung microbiome but to date has not been studied | |
| Behavior | Animal behavior (e.g., coprophagy or fasting) likely influences lung and gut microbiota | |
| Geographic factors | Location, including factors like temperature, humidity, and altitude | |
| Procedural factors | Sequencing contamination | Bacterial DNA is present in laboratory reagents and can contaminate low-biomass microbiome studies (25) |
| Batch effect | Use of distinct DNA extraction kits or sequencing runs can introduce false clustering of specimens (25) | |
| Extraction bias | The cell walls of select taxa (e.g., mycobacteria and fungi) are resistant to DNA extraction (168) | |
| Amplification bias | The 16S rRNA gene is present in highly variable numbers (1–14) among bacterial species (169) |
Once a core hypothesis is articulated, it is important to identify the purported ecologic mechanism of microbiome perturbation. Figure 2 illustrates a conceptual model of the ecologic determinants of the lung microbiome, which is determined by the balance of three ecologic factors: immigration (what moves in); elimination (what moves out or expires); and relative reproduction rate of community members, which is determined by environmental growth conditions. In the lungs, the primary mechanism of immigration in health is subclinical microaspiration of pharyngeal contents (10). Microbial elimination is a function of cough, mucociliary clearance, and host defenses. Finally, environmental growth conditions include temperature, pH, and oxygen tension, which differ both spatially within the lungs and across disease states (16).
Figure 2. The ecologic determinants of the lung microbiome. Any change in lung microbiota must be attributable to some combination of microbial immigration, elimination, or changes in microbial reproduction due to environmental growth conditions. In health, minimal reproduction of lung microbiota occurs, and the lung community is defined by the dynamic balance of immigration (via microaspiration) and elimination. In severe lung disease, alterations in environmental growth conditions (e.g., via increased nutrient availability) result in disease- and patient-specific communities, optimized for the ecologic conditions of injured airways. Reprinted by permission from Reference 15.
[More] [Minimize]Thus, a well-crystallized lung microbiome hypothesis will reflect both a core hypothesis (Figures 1B–1D) and an ecologic mechanism of change (Figure 2). Explicit articulation of core hypotheses and proposed ecologic mechanism of dysbiosis will ensure appropriate study design, types of analyses, and, eventually, clarity in interpretation and scientific rigor in lung microbiome studies.
No strict, prescriptive guidelines for microbiome study design exist, as each study design offers different advantages and information for answering the question at hand. Nonetheless, individual study designs should be considered in the context of well-established principles of experimental methodology, feasibility, and sample acquisition/processing requirements.
To capture study designs of the lung microbiome in recently published literature (2015–2018), we performed a systematic literature review (see the data supplement) and generated a quantitative summary of available studies (evidence map, Figure 3). We identified 221 lung microbiome studies reporting primary data. Human subject studies constitute the majority of the literature (80%), and among them, observational study design was most often used (95%), typically in cross-sectional comparisons with clinical variables (74%) as opposed to longitudinal assessment of outcomes (26%). The vast majority (92%) of studies used 16S rRNA gene sequencing technology to describe lung microbiota taxonomy. Sputum was the most commonly used biospecimen in adults, and BAL fluid (BALF) in children.
Figure 3. Evidence map of published lung microbiome studies in the years 2015–2018. (A) Breakdown of animal (20%) versus human studies (80%). (B) The majority of human studies were observational in design (95%), and of those observational studies, 74% were cross-sectional. (C) A total of 41% of animal studies were experimental in design modeling a human disease, whereas 49% were observational studies of animal respiratory diseases. (D) Bar graph of available studies in adult subjects by disease, indicating that chronic obstructive pulmonary disease (COPD) and cystic fibrosis (CF) are the most commonly studied diseases. Inserts: amplicon sequencing (16S or fungal genome targets) was the most commonly used sequencing technology (91%); sputum and BAL fluid (BALF) were the most commonly used sample types. (E) Bar graph of available studies in pediatric subjects by disease, indicating that CF and pneumonia are the most commonly studied diseases. Inserts: 16S rRNA gene sequencing was the most commonly used sequencing technology (93%) and BALF was the most commonly used sample type. BPD = bronchopulmonary dysplasia; IPF = idiopathic pulmonary fibrosis.
[More] [Minimize]The evidence map discloses a mainly descriptive body of literature reporting cross-sectional, genus-level, taxonomic comparisons of lung microbial communities between different disease states and/or healthy controls. Such work is important for discovering global microbial signatures in specific diseases and laying the foundation for hypothesis generation. However, future lung microbiome research should also address mechanistic questions based on well-articulated core hypotheses (Figure 1). In designing such studies, scientists should consider advantages and disadvantages of different study design types (Table 2), and the following set of specific elements:
Timing and frequency of sampling: The natural history of each disease of interest should inform how frequently and at what specific disease milestone (e.g., during an acute exacerbation [1–3]) lung microbiota should be sampled for assessing dynamic longitudinal changes. Prospective studies with serial sampling will advance our understanding of the temporal impact of microbial community shifts on human physiology and outcomes.
Optimal type of samples: Given that no universal standard biospecimen for lung microbiome studies can be proposed, choice of samples should be informed by the pathogenesis of the disease of interest and the feasibility/tolerability of sampling. For example, sputum samples may be preferable and, indeed, desirable in CF or chronic bronchitis, whereas more distal specimens obtained bronchoscopically may be better suited in studies of ARDS.
Observational versus interventional studies: In humans, causal effects of interventions on lung microbiota can only be confirmed in a randomized clinical trial. Thus, we need microbiome-targeted interventions in new clinical trials or in ancillary clinical trial studies with available biospecimens to examine microbiome effects of host-directed interventions. These studies could include elimination (e.g., through antibiotics), addition (e.g., probiotics), or modification (e.g., through prebiotics, microbial transplant, probiotics, etc.) of existing microbial taxa. Well-designed prospective observational studies will continue to be important for hypothesis generation as well as for assessing predictive utility of microbiome-based indices.
Controlling for confounders: After careful consideration of potential confounders of microbiome–outcome associations (Table 1), study protocols should define a priori plans for detailed recording of confounding variables, with explicit statistical analyses for modeling their effects.
Animal models: Consideration should be given to what animal experiments and exposures (including gnotobiotic models) will help answer the question at hand. The controlled conditions/exposures in animal experiments do not obviate the need for confounding assessments (2).
Sample size calculations: Lung microbiome studies are by definition multidimensional ‘omics experiments at risk for both type I and II errors relating to small sample sizes. Although certain approaches have been proposed for power analysis in microbiome research (17, 18), the potential effects of different diversity metrics (19), sparsity (20), and compositionality (21) of microbiome data in sample size estimation remain unclear. Thus, no generally accepted methodologies can be recommended. We therefore encourage researchers to be explicit in reporting how they derived the analyzed sample size, so that future meta-epidemiological research in the field may offer empirical insights into the impact of different sample size estimation approaches.
Functional studies: Analysis of functional associations requires acquisition and appropriate processing/storage of necessary biospecimens (e.g., rapid freezing of BALF samples for RNA studies, or appropriate storage media for metabolomics), which need to be considered up front.
Contamination risks: Of particular relevance to lung microbiome studies of low biomass samples are the unavoidable contamination risks, which are discussed in detail subsequently here.
Logistics: Pragmatically, study design will always need to strike a balance between the ideal desirable measurements, outcomes, and real-life concerns, such as resources, costs, and feasibility. We recommend that scientists be explicit about reasons leading to a specific study design in the methodology description.
| Advantages | Disadvantages | |
|---|---|---|
| Study Subjects | ||
| Animal studies | Genetic homogeneity | Unknown or questionable generalizability of findings to human disease |
| Controlled environmental conditions | Sampling challenges in small rodents (e.g., difficulties in obtaining BAL fluid samples) | |
| Gnotobiotic models (including germ-free and “humanized” animals) allowing for reductionist, causal modeling of lung microbiota roles | Cost and availability of gnotobiotic facilities | |
| Experimental manipulation of lung microbiota with targeted interventions | Several diseases lack high-fidelity animal models (e.g., cystic fibrosis or idiopathic pulmonary fibrosis) | |
| Ability to perform repeated sampling in nonlethal models of disease | ||
| Human studies | Clinical relevance of biospecimens for diagnostic and/or epidemiologic purposes | Inability to account for the effects of premorbid or prestudy entrance factors on the lung microbiota |
| Ability to deeply phenotype patients with different subtypes of certain diseases and understand patient-level heterogeneity | Limited ability to perform serial sampling, especially with invasive samples | |
| Requirement for large sample sizes to detect biological signals | ||
| Experimental design | ||
| Observational | Ability to leverage existing and ongoing cohorts or clinical trial population biospecimens | Confounding of associations on multiple levels |
| More feasible and less costly compared to clinical trials | Inability to examine causality and directionality of effects between host outcomes and lung microbiota, especially in cross-sectional examinations | |
| Ability to establish incidence and perform time-to-event analysis with temporal associations to lung microbiota in longitudinal studies | Generated evidence is often circumstantial and hypothesis generating | |
| Less ethical issues | ||
| Interventional | Randomized clinical trials are the only study design in humans that can confirm and reliably quantify the causal effects of treatment | Feasibility and/or cost considerations |
| Ability to obtain effect size estimates of specific interventions on lung microbiota communities free from confounding | Safety and tolerability of proposed microbiome-targeted intervention | |
| Complicated regulatory landscape for conducting microbiome replacement studies (e.g., fecal microbiota transplant) | ||
| Sample types | ||
| Invasive (BAL fluid, bronchoscopic protected brushes, or excised lung tissue) | Established best practice recommendations for bronchoscopic samples in microbiome research | Invasiveness of procedure (bronchoscopy and most importantly surgery) make repeated sampling difficult or practically impossible |
| BAL fluid samples obtain material from large areas of the bronchial tree and alveolar space with widely reproducible microbial signal | Patient risk and/or discomfort | |
| Bronchoscopic brushes can provide detailed characterization of airway resident microbiota | Preprocedural confounders (e.g., prophylactic antibiotics before surgery) | |
| Lung tissue samples (both human and animal) can provide more direct examination of microbiota in lung parenchyma | ||
| Noninvasive (sputum or tracheal aspirate) | Easily accessible and repeatable samples in certain conditions | Many patients are unable to produce sputum or cease to produce sputum after treatment |
| Higher microbial load than corresponding BAL fluid samples | Methods for optimal processing of sputa not defined (e.g., preprocessing with lysozymes or homogenization) | |
| Unknown origin of microbiota (proximal airways vs. lower respiratory tract or parenchyma) | ||
| Risk of oropharyngeal contamination |
In summary, when designing a study, scientists should adapt their methodology to the testable hypothesis, and the proposed mechanisms that are thought to contribute to the examined phenotype. Well-designed observational studies will remain important for hypothesis generation, but functional studies are increasingly needed for more mechanistic insights. Statistical analysis and power calculation are areas in which further development would be very helpful for microbiome research.
Respiratory tract specimens require special consideration when sampling, processing, and analyzing. Studies of the lower respiratory tract, a low–microbial biomass system compared with stool specimens (22), are implicitly at risk of false-positive results (type II error). This error is introduced by bacterial signals from the oropharynx during sampling, by technical contamination, and stochasticity during sequencing, and can impact interpretation and conclusions of lung microbiome studies. In this section, we discuss how these error sources can be identified and mitigated.
Sample contamination from the upper respiratory tract is a major concern. Sampling of the lower airway microbiome (excluding surgically excised tissue) requires the passage of the specimen or sampling instrument through the pharynx, which contains several-fold-more bacteria than the distal airway (23). In sputum specimens, contamination from the pharynx is inherent, although the extent varies depending on disease state (24) and sputum constituents.
Technical procedures during sample handling can also introduce contamination. Bacterial DNA is ubiquitous, not eradicated by standard sterilization techniques, and can even be found in laboratory reagents used in DNA isolation (25). Well-to-well contamination during DNA extraction and library preparation has recently been demonstrated (26). Finally, specimens with a low absolute abundance of bacterial DNA are subject to some degree of stochastic results, named sequencing stochasticity (27). Interrogating whether contamination or sequencing stochasticity is responsible for the error is possible. Contamination from reagents/instruments will result in replicates with similar detectable taxa, whereas sequencing stochasticity will demonstrate replicates that share minimal or no taxa (28).
Several authors have proposed methods to account for this error and bioinformatic pipelines for detecting potential contaminants (29). Currently, there are no detailed guidelines for the control of contamination in the isolation and analysis of DNA for lung microbiome studies. Droplet digital PCR may have advantages compared with qPCR in reducing noise when studying overall bacterial burden (30). At a minimum, we strongly suggest inclusion of negative and positive sequencing control specimens (see subsequent text here), ultraviolet-irradiating plasticware, and reagents to cross-link bacterial DNA, as well as open and transparent reporting of “background” bacterial taxa, because, given the compositional nature of microbiome data, removing one taxon affects the relative composition of the remaining taxa. Other important considerations include (28, 31):
• Use of as large a sample as possible to maximize input biomass.
• Incorporation of batch effect in statistical analysis.
• Use of alternative methods to validate results, including cultivation.
• Employing different DNA isolation kits and use kits that are certified as DNA free.
• “Biological plausibility” through examination of appropriate reference literature.
In summary, contamination is the main concern in lung microbiome research, due to the low biomass of the experimental samples. The methods described in this section can help minimize the confounding effect of contamination, but, ultimately, every study will need to address the tradeoff between sensitivity and specificity of applied methods.
Here, we focus on current sampling techniques for culture-independent analysis of lower respiratory tract microbiota for human and animal studies. Approaches include bronchoscopy specimens (BALF and protected specimen brushings), sputum, breath condensate, and surgically excised lung tissue. The most common approach is bronchoscopy-acquired BALF. However, there is no gold standard approach in the field, and no guidelines currently include recommendations for microbiome sampling.
Sputum is an extracellular gel, including water, heavily glycosylated mucins, inhaled toxins and particulate matter, host cells, and bacteria and their associated products (32), and is the result of a mucociliary escalator lining the airway mucosa. Sampling of sputum is noninvasive, and can be performed serially, which renders it advantageous compared with invasive bronchoscopy (33). Sputum may be induced through inhalation of saline solutions. It is unclear how well microbial communities in spontaneous and induced sputum correlate (34, 35). In general, a consistent approach is required when studying sputum samples, which have been used extensively to examine immunologic and microbial characteristics in chronic airway diseases. Given the potential for salivary contamination, measures should be taken (and have been routinely applied in large study networks [36, 37]) to assess this and to determine appropriate use and interpretation of subsequent data. Detailed protocols can be found in References 38–40.
Sputum inherently represents a variable mixture of upper and lower airway microbiota. Although unsurprisingly there is overlap and great similarity in the microbiota profiles of sputum and oral wash samples (24, 41), similarities between sputum and bronchoscopically obtained samples also have been described, and may be more apparent in disease (24, 42). Diseases with high microbial burden in the lower airways, such as CF, are more likely to yield sputum samples reflective of the lower airway microbiota. Conversely, upper airway disease, such as chronic rhinosinusitis, which may coexist with asthma or CF, could introduce additional variable elements to the analysis of sputum. Lastly, lung units that are functionally excluded from ventilation (e.g., distal lung in severe emphysema, or fibrotic diseases) may be underrepresented in the sputum analysis. Such factors should be taken into account when deciding the utility of sputum as a source for microbiome analysis.
Bronchoscopic sampling of the lower airway is safe, and serial studies can be performed for clinical and research indications. Detailed information on bronchoscopy studies is available elsewhere (28). Studies have used both whole lavage fluid and acellular fluid for lung microbiome studies, although bacterial signal and diversity are less robust in the latter approach, possibly due to the absence of cell-associated bacteria (43), and thus whole-lung lavage fluid may be preferable. For this reason, information about the processing of BALF should be included in the reports. Protected specimen brushings and endobronchial biopsies sample the bronchial mucosa, and analysis results from these may not be identical to those from BALF. Nevertheless, similarities have been reported in disease relationships to clinical outcomes or biomarkers of interest with both specimen types (42, 44, 45). It should be noted that spatial and anatomical variation in lung microbiota composition is understudied in disease. Studies of healthy subjects report that intrasubject variation is less then intersubject variation (46), suggesting that (at least in health) studies of multiple lung segments may not be required.
Several studies have addressed the question of potential contamination of bronchoscopy-derived specimens. The bronchoscope channel does not appear to be a considerable contamination source. Pharyngeal contamination of bronchoscope-derived specimens is a theoretical concern, but empirical studies have revealed minimal influence of pharyngeal microbiota on bronchoscopic specimens (10, 46). Carefully planned control specimens should be considered at every bronchoscopy and include scope rinse preinsertion with sterile saline and unused protected brushes. Controls can include oral and nasal rinses, but, because organisms are shared between upper and lower respiratory tracts, interpretation of such controls should acknowledge this. Other controls should include scope rinse (before use) and “sterile” specimens of the sampling materials (e.g., unused normal saline solution or protected specimen brushes) that are run through the entire purification and analysis pipeline. It will likely prove impossible to remove all possible contamination, as there is always a tradeoff with a concomitant reduction in detection sensitivity, especially in low-biomass samples. Therefore, thresholding decisions will be study dependent, and should be reported.
Lung tissue is usually acquired through invasive techniques, including surgical excision and bronchoscopy. Surgical excision by thoracotomy, although conceptually negating the effect of pharyngeal contamination, is associated with significant cost and perioperative risks that limit utility. Surgically excised lung tissue commonly represents a small relative volume of alveolar and distal bronchial airway, and is often fragmented. These factors suggest that the bacterial signal may be limited and not representative, particularly in diseases such as fibrosis, where sampled tissue may contain minimal alveolar or airway surface area due to remodeling (47). In addition, patients undergoing lung excision almost always receive preprocedural antibiotics (48, 49), further confounding sampling and microbiome assessment. Explanted lung has been used to improve our understanding of spatial variation of lung microbial communities, but has limitations in subsampling requirement and the end-stage nature of the disease process present (47, 50).
Exhaled breath condensate (EBC) is a biological matrix consisting of aerosolized particles from the airway lining fluid, distilled water (the water-saturated exhalant results in water condensation), and water-soluble volatiles (51). The bacterial mass of EBC is only a small fraction of what we know is an already low-biomass specimen. In large-animal models, EBC bacterial DNA was not representative of the lower respiratory tract communities and demonstrated significantly less bacterial burden (52). No study to date supports use of human EBC for microbiome studies, although one study of patients with asthma demonstrated some utility in the identification of fungal airway colonization (53).
Endotracheal intubation in mechanically ventilated patients allows for direct sampling through the endotracheal tube, bypassing the oropharynx. Targeted suctioning is a relatively safe and inexpensive procedure for sampling the lung microbiome. Aspirates likely represent both lower and upper airway microbiota as a result of mucociliary clearance and aspiration of pharyngeal secretions. Endotracheal aspirates have been leveraged in critically ill patients, and have provided informative results (6, 54–56). Further studies of endotracheal aspirates in mechanically ventilated patients with serial control specimens from other sources, microbiological data for tube biofilm formation, and optimal strategies for sampling are required.
In summary, BAL is theoretically the method that may more accurately represent the lower airway microbiome; however, alternative methods, such as (primarily) induced sputum or lung tissue, have significant applications depending on the disease studies and hypothesis tested. Clear and consistent methodology should be applied to minimize operator-dependent variability.
Studies of lung microbiota in large- and small-animal models, including mice (2, 57), sheep (58), pigs (59), horses (10), calves (60), and monkeys (61, 62), are feasible. Microbial ecology and immunology differences across species should be a consideration in any comparative study using animal models and drawing parallels to human disease. Recent efforts have used humanized mice, with adoptive transfer of both human immune and microbiome elements into immune-deficient recipient mice (63), to overcome some of these limitations. Such experiments are resource intensive and difficult to replicate, due to the variability in donor materials. Of course, even these experiments do not remove the possible effect of anatomical and histological differences between murine and human lungs.
Murine modeling has predominated as the model format in the published literature, although biological and anatomical lung structure between mice and humans differ significantly (64, 65). However, the murine lung microbiome has been categorically defined, and studied in health and disease models by leveraging technology, rigorous study design and methodology, a priori hypothesis-driven research questions, and the incorporation of multiple negative controls (5, 57, 66–72). Importantly, young mice are more susceptible to environmentally induced lung dysbiosis, whereas adult mice display greater resilience (73). Studies have specifically demonstrated the confounding influence of cage and cohousing on lung microbiota in particular, as mice are coprophagic, and microbial communities of the lung and gut may be predominantly influenced by cohoused mice (2). Several authors have proposed approaches to manage cage effects and the standardization of microbiota in animal models (e.g., by careful control of housing conditions, fecal transplants, and litter swaps) (74). In addition, experimental design that includes comparative cohoused treatment and control groups can be used to demonstrate that a change in the microbiota occurs, despite the cage effect (75). Lung microbiota sampling in murine models is most reliably achieved by the utilization of whole-lung homogenate over partial lung tissue or murine BALF (2). Microbial community composition differs in BALF compared with paired lung tissue in murine models, with lower numbers of unique taxa identified in BALF (67).
In summary, animal models assist the study of lung dysbiosis in the pathogenesis of acute and chronic lung disease. Local alveolar immune responses have been correlated with lung microbiota in several animal models (57, 66, 70), and the ability to use antibiotic depletion (76) and germ-free (GF) models (66) allows for further interrogation of potential mechanisms. Further work will promote the move to causal studies using advanced GF or antibiotic depletion models with both transgenic host platforms and modified taxa (77).
Bacterial culture is selective, identifying only those organisms capable of growth on the particular media and incubation conditions used. This is inherently biased when considering all microorganisms present in a sample. It is possible to apply more intensive culture approaches and to access a wider range of organisms, but this is laborious for large numbers of samples (78–81). Culture-independent methods allow all organisms present to be detected (82). These commonly depend on extracting all the DNA from a sample and use a variety of methods to specifically detect the DNA of microorganisms present in this mixed DNA extract. The DNA extraction method can be a source of bias (e.g., if microorganisms are not lysed appropriately, their DNA will not be detectable). Therefore, consideration of methodology is important (83).
All bacteria contain ribosomes, which consist of a 50S and a 30S subunit. The smaller 30S subunit is composed of 21S proteins and 16S ribosomal RNA. Given its essential functional role, the gene encoding the 16S rRNA subunit is highly conserved, and PCR primers have been designed that are capable of amplifying this gene across all bacteria. In addition, the gene has nine variable regions, and variation in these regions can be used as a molecular fingerprint to identify the organism (84, 85). DNA sequencing of the PCR products amplified from a mixed DNA extract reveals present microorganisms, and the counts of each sequencing read of each type reveal their relative abundance. This technique has been used for decades, but only in the last 5–10 years in a respiratory setting. It gives a profile of all the bacteria present and their relative abundance, allows comparison between samples over time, and, in a case–control setting, can reveal the impact of clinical disease, treatment, or infection on all bacteria present. Limitations include resolution (only genus-level taxonomy can be called accurately), lack of functional information (no pathway data as is possible with shotgun metagenomics), sensitivity to differences in methodology (DNA extraction, PCR approach, and PCR primers), and subsequent analysis (86–88). Despite this, this type of sequencing has been employed in a wide range of different respiratory disease settings, revealing a complex and dynamic community of microorganisms (89).
In metagenomics, instead of amplifying a single gene, the entire mixed DNA extract from a sample is sequenced, thereby removing the PCR steps of the 16S rRNA approach and reducing bias. Moreover, all functional genes of each organism in the sample can be accessed, potentially allowing the study of antimicrobial resistance or virulence genes or the metabolism of the organisms present. Finally, metagenomics can also be used to improve the resolution of bacterial identification in a sample, allowing strain-level typing and tracking of infectious agents during outbreaks (90–93). As bacterial genomes are roughly 4,000 times larger than bacterial genes, metagenomics requires increased sequencing depth to fully characterize a microbial community. Without the selection of a single bacterial gene, as with 16S rRNA gene sequencing, human DNA is also sequenced. In low microbial biomass respiratory samples, the vast majority of sequenced reads thus originates from human genomic DNA (93–95). This can make the approach expensive. Although a number of in-house (96, 97) and commercial methods have been developed for the depletion of human genomic DNA from clinical samples, these methods can create additional challenges due to autolysis of microorganisms and the introduction of reagent contaminants. Targeted capture methods have been successfully applied for the enrichment and sequencing of specific microbial pathogens (98, 99). However, this approach requires prior knowledge of the pathogen of interest. Given the costs and analytical requirements to resolve a complex mix of DNA sequences from a metagenome, there are fewer examples of respiratory studies (93, 95, 98–101). Future respiratory metagenomics work will likely focus on the improvement of methods for the depletion of the abundance of human genomic DNA from samples of interest.
It is possible to isolate microorganisms and sequence them individually; the first fully sequenced genome of a cellular organism was, in fact, the respiratory pathogen, Hemophilus influenzae (102). Whole-genome sequencing has been employed in a wide range of pathogens, with Streptococcus pneumoniae having 73 complete genomes and over 10,000 scaffolds or sets of sequenced contigs available at the National Center for Biotechnology Information genome browser (accessed June 11, 2019) (103). Whole-genome sequencing has been used to track organisms in outbreaks of Mycobacterium tuberculosis (104), demonstrating person-to-person transfer of nontuberculous mycobacteria (105). Importantly, advances in sequencing are starting to make this approach faster than traditional identification methods after isolation (106).
Although each of the techniques outlined above has been considered separately here, in reality they are complementary tools for understanding respiratory microbiota. Isolation of microorganisms allows them to be used in models and for their physiology to be understood, and their genome to be sequenced to reveal virulence genes and achieve strain level identification. This, in turn, supports their identification and tracking in metagenomic studies and functional annotation of mixed DNA extracts, with 16S profiling placing this in the context of the wider community of organisms present.
As mentioned in the previous section, microbiome data can reflect different aspects of a bacterial community: 16S rRNA gene sequencing provides information on composition (resolution generally limited to the genus level), whereas shotgun metagenomics (DNA sequencing) and metatranscriptomics (RNA sequencing) provide both composition (species or strain level) and functional potential/activity. Regardless of platform and sequencing approach, raw sequence data are initially processed to generate microbiota profiles (i.e., which bacteria are present and their relative abundance) before downstream analyses. Several well-established bioinformatics pipelines tackle these preprocessing steps, including mothur and QIIME (Quantitative Insights Into Microbial Ecology) (107, 108). Although the open-source nature of these tools facilitates analysis, researchers should be aware of the nuances in parameter choices that can impact results, such as similarity threshold, choice of reference database, and exact algorithm for operational taxonomic unit (OTU) or amplicon sequence variant (ASV). For example, closed-reference methods can only identify OTUs already existing in the reference database, but the OTUs generated can be combined with those from other studies (provided that the data processing methods were similar). On the other hand, de novo algorithms can identify novel taxa, but must be applied to the totality of the data in a single execution (i.e., OTUs identified in different studies are not directly comparable unless the raw sequencing data of both studies have been combined before OTU clustering). A more comprehensive description of analysis approaches and issues that are relevant to all microbiome research were covered in a recent review (109).
An important consideration in lung microbiome studies is how to control for the high risk of contamination due to low biomass in samples (25), which was discussed in more detail previously here. It is also important to consider that simply removing sequences found in negative controls can also remove “true” sequences in the samples, and so filtering out data should always be performed while trying to maintain a balance between type I and type II error rates.
As with all data analyses, curation of metadata (demographics, clinical information, etc.) is critical to correctly test the testable hypothesis. Importantly, this can be performed even before microbiome data have been generated, allows early identification of problems, and facilitates later analyses. In our experience, this process should aim at producing metadata that are complete (minimal missing values), valid (values constrained depending on the variables measured [e.g., height should be captured as an integer or real]), and consistent (same entries for same values [e.g., height always expressed in centimeters and or missing values always identified using the same text strings]). Importantly, incorporating metadata on potential confounders, such as antibiotics, comorbidities, or medications, will strengthen the robustness of conclusions. Finally, experimental metadata may help explain patterns in the data not associated with biological or clinical variables. For example, timing of sample collection, experimental protocols (DNA extraction kits, primers, sequencing instrument, and batch), or freeze/thaw cycles, among others, are important variables that should be regularly collected (110).
Downstream microbiome data analysis is generally performed at two different levels: community-wide measures of diversity, and association of bacterial features (taxa, functions) with metadata. Diversity is generally measured as α (i.e., total amount of diversity in a sample or group of samples) or β (distance between any two samples based on shared and unique taxa in each sample). β diversity distances can be used to perform ordination (i.e., dimensionality reduction) and visualization of samples using principal coordinate analysis (not to be confused with principal component analysis), or nonmetric multidimensional scaling (111), which can identify clustering patterns or gradients in the data. Association of bacterial features with metadata is generally performed using differential enrichment analysis (for discrete variables) or correlation analysis (continuous variables). In both cases, and because of the large number of tests being performed, it is fundamental to correct for multiple hypothesis testing to reduce false positives.
The above considerations are generally applicable to all microbiome data. Shotgun metagenomics and metatranscriptomics are more recent techniques that, although potentially providing deeper insights, are also less standardized. Bacterial identification from metagenomic data, for instance, relies heavily on the software tools being used and whether they are reference based (computationally more efficient, lower sequencing depth required) or assembly based (assemble bacterial genomes de novo; potentially more sensitive, but require higher sequencing depth). Metatranscriptomic analysis is further complicated by the fact that few established pipelines exist, and their validation has not yet been performed in a comprehensive manner.
In conclusion, analysis methodologies change rapidly, and staying up to date with new developments is not trivial, even for experts in the field. We therefore strongly encourage researchers to maintain detailed documentation of how analysis is performed (including tool versions and step-by-step instructions for analysis) to facilitate robust and reproducible results, as will be discussed in the next section.
Reproducibility, generally understood as the ability to reproduce experimental results from a study in a different study or cohort, is a fundamental requirement for the advancement of the scientific knowledge. However, when dealing with high-throughput sequencing data, internal reproducibility (i.e., the ability to replicate reported results using the same data and analysis) is a significant challenge that must be met even before considering external validation (112). Although we suggest guidelines on how to facilitate reproducibility of lung microbiome studies (Table 3), there are multiple approaches to design, execute, or analyze the data of any given study.
| Item No. | Recommendation | |
|---|---|---|
| Raw Data | 1 | Raw sequencing data should be deposited in one of the publicly available databases (ENA, EMBL, Genbank, DDJB, etc.) |
| Analytical methods | ||
| Preprocessing | 2 | Describe preprocessing filters used |
| Taxonomic assignment | 3 | Describe reference set and build used for assignment |
| Analysis | 4 | Describe analytical bioinformatics software used; provide coded analysis data (e.g., in Github https://github.com/ or Bitbucket https://bitbucket.org/ repositories, or as Jupyter notebooks) |
| Post-processing | 5 | Describe data transformations (e.g., logarithmic) and methodology for removal or exclusion of data from analysis |
| Processed data | 6 | Consider depositing processed data (e.g., at Figshare, https://figshare.com/ or Biostudies, https://www.ebi.ac.uk/biostudies/, etc.) |
| Metadata | ||
| Human | ||
| Sociodemographics | 7 | Age, sex/sex, diet, date, and location of sampling |
| Disease specific | Disease severity or stage, and antibiotic and other medication use | |
| Animal | 8 | |
| Housing conditions | 9 | Describe housing, diet, frequency of cage changes, and cohousing |
| Demographics | Sex, age, provenance (if commercially procured), and litter association | |
| Both | 10 | |
| Sample processing | 11 | Sample handling and DNA/RNA isolation methodology |
| Controls | 12 | Describe contamination filtering methods used |
| Genetics | 13 | Strain information (for mice) or genotyping (to the extent available, for humans) |
To be able to reproduce published results, we must have access to the raw sequencing data and the exact specifications of the analytical approaches applied to the data. Over the last few years, an increasing number of journals has started to require authors to provide access to data as a precondition for publication. The use of statements, such as “data available upon request,” is no longer acceptable (113), so as to reduce the hurdles of obtaining data from published articles, which can take several months (113). Thus, we strongly believe that data should be publicly available upon publication, and preferably at the time of peer review. There are multiple publicly available databases in which genomic datasets can be uploaded for dissemination, including GenBank, the European Nucleotide Archive (ENA)—European Bioinformatics Institute (EMBL-EBI), or DNA Data Bank of Japan (DDJB).
Access to data alone is insufficient to ensure reproducibility of results, and a complete description of the analysis is generally required. Although we have summarized the major steps involved in the analysis of microbiome data in the relevant section, we will highlight here some aspects that we consider of particular importance. Preprocessing of raw data, as the first step of the analyses, can have a major impact in downstream results. For example, the choice of different parameters for quality filtering can significantly affect bacterial estimates of diversity (114). In addition, taxonomic assignment can often be complicated by the fact that microbial reference databases are under continuous development. SILVA (115) has grown from 300,000 sequences (v89) to over 2 million sequences (v132). This can lead to inconsistent results as the reference sets are updated, and even preclude the reproducibility of published findings. For example, Tropheryma whipplei, which may colonize the lung of patients infected with HIV (116), cannot be identified using some versions of the Greengenes reference set (117). It is therefore critical to specify what taxonomy/version is being used in the analysis report. Clustering of sequences into OTUs or ASVs can also vastly differ depending on the algorithm used and its implementation (118–120). Reporting the exact parameter values and version of the tools used is therefore fundamental. Final post-processing of the data, such as removal of low-abundance taxa or data transformations (e.g., log transforms), should also be described in detail to guarantee reproducibility of the results.
Importantly, the amount of effort required to reproduce analysis and results can be significant even for groups with computational expertise (121), stressing the need for detailed, complete descriptions of analyses. The complexity and time required to reproduce results from sequencing data has resulted in an important dilemma during the peer-review process. Although many journals now provide reviewers with the opportunity to validate results by giving them early access to the data, in practice it is unfeasible to verify reproducibility using this approach. Alternative, less-burdensome approaches are therefore required to ensure a minimum level of validation at submission and/or publication time.
Based on these considerations, we strongly recommend that both raw data and analysis specification are provided at the time of publication as a minimum requirement. Some form of partially processed data (e.g., OTU/ASV tables) should be provided in addition to the raw data, but it should never serve as a substitute for it. Analysis can be provided in the form of code stored under a software repository (Github or Bitbucket are popular options in academic settings) or as Jupyter notebooks (http://jupyter.org). These recommendations would enhance transparency and reproducibility of studies in the lung microbiome, as well as promote scientific advancement by allowing more researchers to test alternative hypotheses on publicly available datasets.
To date, research on microbiome–lung interactions has largely been descriptive (Figure 3). Although it has uncovered key differences separating COPD (50, 122–125), asthma (89, 126–130), idiopathic pulmonary fibrosis (47, 131, 132), CF (42, 133–136), and HIV (116, 137–139) from healthy controls, microbial descriptions alone provide insufficient insight into mechanisms or potential therapeutic applications. Here, we explore knowledge gaps and research directions that may link microbiome with disease pathogenesis. Key to these efforts is utilization of well-described specimens, deep characterization of the specific host response to microbial composition, and experimental models in which host–microbiome interactions are specifically modulated. We also highlight ways to apply the information from microbiome sequencing to patients for risk stratification, diagnosis, outcome prediction, and novel therapeutics. Together, these strategies will translate microbiome research to the practical realities of patient care.
Assuming a role for the microbiome in lung disease pathogenesis, the microbiome must induce a host response, leading to homeostasis or disease. One strategy to decipher host–microbiome interactions is to correlate key microbiome features (e.g., relative OTU abundance) with deep phenotyping of the host (e.g., cellular or molecular disease markers) (140). Multiple ‘omics outputs add further dimensionality to microbiome–host interactions. For example, paired lung samples run concurrently for 16S sequencing and transcriptome profiling have identified genes and pathways implicated in microbiome shifts (141). Further integrating the microbiome with DNA methylation, proteomics and metabolomics (142, 143), and metagenomic analyses encompassing the larger multikingdom array of genomic material in the lung (143–147) will further extend our understanding of the microbiome role in lung biology. However, they also highlight the need for thoughtful planning and resource allocation when designing translational research protocols.
Integrated ‘omics analyses, despite their depth, do not establish causality. In vitro and animal models that mimic microbiome disruption can accomplish this, with the caveat that they are only approximations of the complexity of the human body. Three dimensional modeling, from air–liquid interface cultures to lung-on-a-chip microsystems, provide even more sophisticated representations for host–microbiome studies (148). These biomimetic systems may serve as next-step in vitro models to validate and extend the findings of descriptive microbiome studies.
Animal models in which microbiota are engrafted, transplanted, or manipulated provide experimental conditions for the investigation of causal relationships between microbiome and host biology. GF mice, bred in specialized conditions to prevent contamination with microbes, are a blank canvas on which microbiota can be applied and their effects observed (149). Nonetheless, costs and feasibility of GF mouse work can be prohibitive for many laboratories. Moreover, accounting for lung microbiome variability, which appears to be a very dynamic process (2), will inevitably pose a challenge and require further validation. Finally, the sometimes significant differences between human and experimental animal immunology and bacteriology should be acknowledged when attempting to apply animal model insights into human disease.
Microbiome research is still in its formative stage, and many questions remain unanswered. The following (not exhaustive) list highlights the knowledge gaps:
• Host–microbiome interaction mechanisms may vary: 1) metabolic (either by providing metabolites or by metabolizing substrates from eukaryotic cells); 2) immune activation (directly or indirectly); 3) local (in the lung) or at a distance (gut–lung axis, upper airway) effects; 4) extracellular (interactions with cell receptors) or intracellular (intracellular parasites or viruses that highjack cellular processes, viral sequences embedded in the mammalian genome); etc.
• Are there developmental stages for the lung microbiome, and is it ever set? Recent research suggests that lung microbiome is remarkably mutable (2).
• Is there seasonal, geographic, or sex variability in the microbiome or the host response to it (150–152)?
• What are immune mechanisms that dictate microbiome consistency and constitution (153)? How does genetic host variability influence microbiome effects?
• How do environmental exposures influence the microbiome?
• How can we differentiate homeostatic from disease-inducing effects of the microbiome?
• Is it time to include microbiome sampling in our clinical research design (like DNA sampling)?
The ultimate goal of lung microbiome research is to discover key diagnostic or therapeutic features that impact clinical outcomes. The lung microbiome may contain prognostic and diagnostic information, and thus serve as a useful clinical biomarker. For example, sputum microbiome characteristics during COPD exacerbations (presence of Staphylococcus, absence of Veillonella, and reduced α diversity) can predict future mortality risk (154). Furthermore, microbiome analysis may aid in the diagnosis of infection and the differentiation between infectious and noninfectious causes of lung injury and ARDS (155–158).
It is still unclear whether the microbiome itself can be of therapeutic benefit in lung disease. The gut microbiome may be modifiable through diet to alter the gut–lung axis. This may be relevant for diseases like asthma, which has been strongly associated with gut microbiome alterations as early as perinatally (159). Probiotics restoring a healthy gut microbiome may reduce Th2 cytokine responses (160); unfortunately, randomized, controlled trials have not yet demonstrated a reduction in asthma incidence (161) or clinical improvement in children with asthma (162). Ultimately, achieving a personalized therapeutic approach based on the microbiome will require proper characterization of the microbiome of specific disease endotypes, identification of endotype-relevant gene targets mediated by the microbiome, and development of novel methods to influence the microbiome. Until then, the lung microbiome will remain a promising future target of translational lung research.
| 1. | Young VB. The role of the microbiome in human health and disease: an introduction for clinicians. BMJ 2017;356:j831. |
| 2. | Dickson RP, Erb-Downward JR, Falkowski NR, Hunter EM, Ashley SL, Huffnagle GB. The lung microbiota of healthy mice are highly variable, cluster by environment, and reflect variation in baseline lung innate immunity. Am J Respir Crit Care Med 2018;198:497–508. |
| 3. | Segal LN, Clemente JC, Wu BG, Wikoff WR, Gao Z, Li Y, et al. Randomised, double-blind, placebo-controlled trial with azithromycin selects for anti-inflammatory microbial metabolites in the emphysematous lung. Thorax 2017;72:13–22. |
| 4. | Slater M, Rivett DW, Williams L, Martin M, Harrison T, Sayers I, et al. The impact of azithromycin therapy on the airway microbiota in asthma. Thorax 2014;69:673–674. |
| 5. | Dickson RP, Singer BH, Newstead MW, Falkowski NR, Erb-Downward JR, Standiford TJ, et al. Enrichment of the lung microbiome with gut bacteria in sepsis and the acute respiratory distress syndrome. Nat Microbiol 2016;1:16113. |
| 6. | Panzer AR, Lynch SV, Langelier C, Christie JD, McCauley K, Nelson M, et al. Lung microbiota is related to smoking status and to development of acute respiratory distress syndrome in critically ill trauma patients. Am J Respir Crit Care Med 2018;197:621–631. |
| 7. | Finn SMB, Scheuermann U, Holzknecht ZE, Parker W, Granek JA, Lin SS, et al. Effect of gastric fluid aspiration on the lung microbiota of laboratory rats. Exp Lung Res 2018;44:201–210. |
| 8. | Rosen R, Hu L, Amirault J, Khatwa U, Ward DV, Onderdonk A. 16S community profiling identifies proton pump inhibitor related differences in gastric, lung, and oropharyngeal microflora. J Pediatr 2015;166:917–923. |
| 9. | Fillion-Bertrand G, Dickson RP, Boivin R, Lavoie JP, Huffnagle GB, Leclere M. Lung microbiome is influenced by the environment and asthmatic status in an equine model of asthma. Am J Respir Cell Mol Biol 2019;60:189–197. |
| 10. | Dickson RP, Erb-Downward JR, Freeman CM, McCloskey L, Falkowski NR, Huffnagle GB, et al. Bacterial topography of the healthy human lower respiratory tract. MBio 2017;8:e02287-16. |
| 11. | Freestone PP, Hirst RA, Sandrini SM, Sharaff F, Fry H, Hyman S, et al. Pseudomonas aeruginosa–catecholamine inotrope interactions: a contributory factor in the development of ventilator-associated pneumonia? Chest 2012;142:1200–1210. |
| 12. | Dickson RP, Erb-Downward JR, Prescott HC, Martinez FJ, Curtis JL, Lama VN, et al. Intraalveolar catecholamines and the human lung microbiome. Am J Respir Crit Care Med 2015;192:257–259. |
| 13. | Dickson RP. The microbiome and critical illness. Lancet Respir Med 2016;4:59–72. |
| 14. | Dickson RP, Erb-Downward JR, Huffnagle GB. Homeostasis and its disruption in the lung microbiome. Am J Physiol Lung Cell Mol Physiol 2015;309:L1047–L1055. |
| 15. | Dickson RP, Martinez FJ, Huffnagle GB. The role of the microbiome in exacerbations of chronic lung diseases. Lancet 2014;384:691–702. |
| 16. | Dickson RP, Erb-Downward JR, Huffnagle GB. Towards an ecology of the lung: new conceptual models of pulmonary microbiology and pneumonia pathogenesis. Lancet Respir Med 2014;2:238–246. |
| 17. | Mattiello F, Verbist B, Faust K, Raes J, Shannon WD, Bijnens L, et al. A web application for sample size and power calculation in case-control microbiome studies. Bioinformatics 2016;32:2038–2040. |
| 18. | Kelly BJ, Gross R, Bittinger K, Sherrill-Mix S, Lewis JD, Collman RG, et al. Power and sample-size estimation for microbiome studies using pairwise distances and PERMANOVA. Bioinformatics 2015;31:2461–2468. |
| 19. | Kuczynski J, Liu Z, Lozupone C, McDonald D, Fierer N, Knight R. Microbial community resemblance methods differ in their ability to detect biologically relevant patterns. Nat Methods 2010;7:813–819. |
| 20. | Paulson JN, Stine OC, Bravo HC, Pop M. Differential abundance analysis for microbial marker-gene surveys. Nat Methods 2013;10:1200–1202. |
| 21. | Lovell D, Pawlowsky-Glahn V, Egozcue JJ, Marguerat S, Bähler J. Proportionality: a valid alternative to correlation for relative data. PLoS Comput Biol 2015;11:e1004075. |
| 22. | Dickson RP, Erb-Downward JR, Martinez FJ, Huffnagle GB. The microbiome and the respiratory tract. Annu Rev Physiol 2016;78:481–504. |
| 23. | Bassis CM, Erb-Downward JR, Dickson RP, Freeman CM, Schmidt TM, Young VB, et al. Analysis of the upper respiratory tract microbiotas as the source of the lung and gastric microbiotas in healthy individuals. MBio 2015;6:e00037. |
| 24. | Durack J, Huang YJ, Nariya S, Christian LS, Ansel KM, Beigelman A, et al.; National Heart, Lung and Blood Institute’s “AsthmaNet”. Bacterial biogeography of adult airways in atopic asthma. Microbiome 2018;6:104. |
| 25. | Salter SJ, Cox MJ, Turek EM, Calus ST, Cookson WO, Moffatt MF, et al. Reagent and laboratory contamination can critically impact sequence-based microbiome analyses. BMC Biol 2014;12:87. |
| 26. | Minich JJ, Sanders JG, Amir A, Humphrey G, Gilbert JA, Knight R. Quantifying and understanding well-to-well contamination in microbiome research. mSystems 2019;4:e00186-19. |
| 27. | Charlson ES, Bittinger K, Haas AR, Fitzgerald AS, Frank I, Yadav A, et al. Topographical continuity of bacterial populations in the healthy human respiratory tract. Am J Respir Crit Care Med 2011;184:957–963. |
| 28. | Dickson RP, Cox MJ. Sampling the lung microbiome. In: The lung Microbiome. Lausanne, Switzerland: European Respiratory Society; 2019. |
| 29. | Davis NM, Proctor DM, Holmes SP, Relman DA, Callahan BJ. Simple statistical identification and removal of contaminant sequences in marker-gene and metagenomics data. Microbiome 2018;6:226. |
| 30. | Sze MA, Abbasi M, Hogg JC, Sin DD. A comparison between droplet digital and quantitative PCR in the analysis of bacterial 16S load in lung tissue samples from control and COPD GOLD 2. PLoS One 2014;9:e110351. |
| 31. | de Goffau MC, Lager S, Salter SJ, Wagner J, Kronbichler A, Charnock-Jones DS, et al. Recognizing the reagent microbiome. Nat Microbiol 2018;3:851–853. |
| 32. | Fahy JV, Dickey BF. Airway mucus function and dysfunction. N Engl J Med 2010;363:2233–2247. |
| 33. | Jones FL Jr. The relative efficacy of spontaneous sputa, aerosol-induced sputa, and gastric aspirates in the bacteriologic diagnosis of pulmonary tuberculosis. Dis Chest 1966;50:403–408. |
| 34. | An SQ, Warris A, Turner S. Microbiome characteristics of induced sputum compared to bronchial fluid and upper airway samples. Pediatr Pulmonol 2018;53:921–928. |
| 35. | Zemanick ET, Wagner BD, Robertson CE, Stevens MJ, Szefler SJ, Accurso FJ, et al. Assessment of airway microbiota and inflammation in cystic fibrosis using multiple sampling methods. Ann Am Thorac Soc 2015;12:221–229. |
| 36. | Martin RJ, Kephart DK, Dyer AM, Fahy J, Kraft M; Asthma Clinical Trials Network. Quality control within the asthma clinical research network. Control Clin Trials 2001;22(Suppl):207S–221S. |
| 37. | Freeman CM, Crudgington S, Stolberg VR, Brown JP, Sonstein J, Alexis NE, et al. Design of a multi-center immunophenotyping analysis of peripheral blood, sputum and bronchoalveolar lavage fluid in the Subpopulations and Intermediate Outcome Measures in COPD Study (SPIROMICS). J Transl Med 2015;13:19. |
| 38. | Alexis N, Soukup J, Ghio A, Becker S. Sputum phagocytes from healthy individuals are functional and activated: a flow cytometric comparison with cells in bronchoalveolar lavage and peripheral blood. Clin Immunol 2000;97:21–32. |
| 39. | Gershman NH, Wong HH, Liu JT, Mahlmeister MJ, Fahy JV. Comparison of two methods of collecting induced sputum in asthmatic subjects. Eur Respir J 1996;9:2448–2453. |
| 40. | Fahy JV, Boushey HA, Lazarus SC, Mauger EA, Cherniack RM, Chinchilli VM, et al.; NHLBI Asthma Clinical Research Network. Safety and reproducibility of sputum induction in asthmatic subjects in a multicenter study. Am J Respir Crit Care Med 2001;163:1470–1475. |
| 41. | Sulaiman I, Wu BG, Li Y, Scott AS, Malecha P, Scaglione B, et al. Evaluation of the airway microbiome in nontuberculous mycobacteria disease. Eur Respir J 2018;52:1800810. |
| 42. | Hogan DA, Willger SD, Dolben EL, Hampton TH, Stanton BA, Morrison HG, et al. Analysis of lung microbiota in bronchoalveolar lavage, protected brush and sputum samples from subjects with mild-to-moderate cystic fibrosis lung disease. PLoS One 2016;11:e0149998. |
| 43. | Dickson RP, Erb-Downward JR, Prescott HC, Martinez FJ, Curtis JL, Lama VN, et al. Cell-associated bacteria in the human lung microbiome. Microbiome 2014;2:28. |
| 44. | Denner DR, Sangwan N, Becker JB, Hogarth DK, Oldham J, Castillo J, et al. Corticosteroid therapy and airflow obstruction influence the bronchial microbiome, which is distinct from that of bronchoalveolar lavage in asthmatic airways. J Allergy Clin Immunol 2016;137:1398–1405.e3. |
| 45. | Sharma A, Laxman B, Naureckas ET, Hogarth DK, Sperling AI, Solway J, et al. Associations between fungal and bacterial microbiota of airways and asthma endotypes. J Allergy Clin Immunol 2019;144:1214–1227.e7. |
| 46. | Dickson RP, Erb-Downward JR, Freeman CM, McCloskey L, Beck JM, Huffnagle GB, et al. Spatial variation in the healthy human lung microbiome and the adapted island model of lung biogeography. Ann Am Thorac Soc 2015;12:821–830. |
| 47. | Kitsios GD, Rojas M, Kass DJ, Fitch A, Sembrat JC, Qin S, et al. Microbiome in lung explants of idiopathic pulmonary fibrosis: a case–control study in patients with end-stage fibrosis. Thorax 2018;73:481–484. |
| 48. | Chang SH, Krupnick AS. Perioperative antibiotics in thoracic surgery. Thorac Surg Clin 2012;22:35–45, vi. |
| 49. | D’Andrilli A, Rendina EA. Enhanced recovery after surgery (ERAS) and fast-track in video-assisted thoracic surgery (VATS) lobectomy: preoperative optimisation and care-plans. J Vis Surg 2018;4:4. |
| 50. | Sze MA, Dimitriu PA, Hayashi S, Elliott WM, McDonough JE, Gosselink JV, et al. The lung tissue microbiome in chronic obstructive pulmonary disease. Am J Respir Crit Care Med 2012;185:1073–1080. |
| 51. | Davis MD, Montpetit AJ. Exhaled breath condensate: an update. Immunol Allergy Clin North Am 2018;38:667–678. |
| 52. | Glendinning L, Wright S, Tennant P, Gill AC, Collie D, McLachlan G. Microbiota in exhaled breath condensate and the lung. Appl Environ Microbiol 2017;83:e00515-17. |
| 53. | Carpagnano GE, Malerba M, Lacedonia D, Susca A, Logrieco A, Carone M, et al. Analysis of the fungal microbiome in exhaled breath condensate of patients with asthma. Allergy Asthma Proc 2016;37:41–46. |
| 54. | Kalantar KL, Moazed F, Christenson SC, Wilson J, Deiss T, Belzer A, et al. Metagenomic comparison of tracheal aspirate and mini-bronchial alveolar lavage for assessment of respiratory microbiota. Am J Physiol Lung Cell Mol Physiol 2019;316:L578–L584. |
| 55. | Kitsios GD, Fitch A, Manatakis DV, Rapport SF, Li K, Qin S, et al. Respiratory microbiome profiling for etiologic diagnosis of pneumonia in mechanically ventilated patients. Front Microbiol 2018;9:1413. |
| 56. | Zakharkina T, Martin-Loeches I, Matamoros S, Povoa P, Torres A, Kastelijn JB, et al. The dynamics of the pulmonary microbiome during mechanical ventilation in the intensive care unit and the association with occurrence of pneumonia. Thorax 2017;72:803–810. |
| 57. | O’Dwyer DN, Zhou X, Wilke CA, Xia M, Falkowski NR, Norman KC, et al. Lung dysbiosis, inflammation, and injury in hematopoietic cell transplantation. Am J Respir Crit Care Med 2018;198:1312–1321. |
| 58. | Glendinning L, Collie D, Wright S, Rutherford KMD, McLachlan G. Comparing microbiotas in the upper aerodigestive and lower respiratory tracts of lambs. Microbiome 2017;5:145. |
| 59. | Pezzulo AA, Kelly PH, Nassar BS, Rutland CJ, Gansemer ND, Dohrn CL, et al. Abundant DNase I–sensitive bacterial DNA in healthy porcine lungs and its implications for the lung microbiome. Appl Environ Microbiol 2013;79:5936–5941. |
| 60. | Nicola I, Cerutti F, Grego E, Bertone I, Gianella P, D’Angelo A, et al. Characterization of the upper and lower respiratory tract microbiota in Piedmontese calves. Microbiome 2017;5:152. |
| 61. | Morris A, Paulson JN, Talukder H, Tipton L, Kling H, Cui L, et al. Longitudinal analysis of the lung microbiota of cynomolgous macaques during long-term SHIV infection. Microbiome 2016;4:38. |
| 62. | Cadena AM, Ma Y, Ding T, Bryant M, Maiello P, Geber A, et al. Profiling the airway in the macaque model of tuberculosis reveals variable microbial dysbiosis and alteration of community structure. Microbiome 2018;6:180. |
| 63. | Daharsh L, Zhang J, Ramer-Tait A, Li Q. A double humanized BLT-mice model featuring a stable human-like gut microbiome and human immune system. J Vis Exp 2019;(150):e59773. |
| 64. | Knust J, Ochs M, Gundersen HJ, Nyengaard JR. Stereological estimates of alveolar number and size and capillary length and surface area in mice lungs. Anat Rec (Hoboken) 2009;292:113–122. |
| 65. | Ochs M, Nyengaard JR, Jung A, Knudsen L, Voigt M, Wahlers T, et al. The number of alveoli in the human lung. Am J Respir Crit Care Med 2004;169:120–124. |
| 66. | O’Dwyer DN, Ashley SL, Gurczynski SJ, Xia M, Wilke C, Falkowski NR, et al. Lung microbiota contribute to pulmonary inflammation and disease progression in pulmonary fibrosis. Am J Respir Crit Care Med 2019;199:1127–1138. |
| 67. | Barfod KK, Roggenbuck M, Hansen LH, Schjørring S, Larsen ST, Sørensen SJ, et al. The murine lung microbiome in relation to the intestinal and vaginal bacterial communities. BMC Microbiol 2013;13:303. |
| 68. | Yun Y, Srinivas G, Kuenzel S, Linnenbrink M, Alnahas S, Bruce KD, et al. Environmentally determined differences in the murine lung microbiota and their relation to alveolar architecture. PLoS One 2014;9:e113466. |
| 69. | Poroyko V, Meng F, Meliton A, Afonyushkin T, Ulanov A, Semenyuk E, et al. Alterations of lung microbiota in a mouse model of LPS-induced lung injury. Am J Physiol Lung Cell Mol Physiol 2015;309:L76–L83. |
| 70. | Yadava K, Pattaroni C, Sichelstiel AK, Trompette A, Gollwitzer ES, Salami O, et al. Microbiota promotes chronic pulmonary inflammation by enhancing IL-17a and autoantibodies. Am J Respir Crit Care Med 2016;193:975–987. |
| 71. | Scheiermann J, Klinman DM. Three distinct pneumotypes characterize the microbiome of the lung in BALB/cJ mice. PLoS One 2017;12:e0180561. |
| 72. | Singh N, Vats A, Sharma A, Arora A, Kumar A. The development of lower respiratory tract microbiome in mice. Microbiome 2017;5:61. |
| 73. | Kostric M, Milger K, Krauss-Etschmann S, Engel M, Vestergaard G, Schloter M, et al. Development of a stable lung microbiome in healthy neonatal mice. Microb Ecol 2018;75:529–542. |
| 74. | McCoy KD, Geuking MB, Ronchi F. Gut microbiome standardization in control and experimental mice. Curr Protoc Immunol 2017;117:23.1.1–23.1.13. |
| 75. | Groves HT, Cuthbertson L, James P, Moffatt MF, Cox MJ, Tregoning JS. Respiratory disease following viral lung infection alters the murine gut microbiota. Front Immunol 2018;9:182. |
| 76. | Yang D, Chen X, Wang J, Lou Q, Lou Y, Li L, et al. Dysregulated lung commensal bacteria drive interleukin-17B production to promote pulmonary fibrosis through their outer membrane vesicles. Immunity 2019;50:692–706.e7. |
| 77. | Fischbach MA. Microbiome: focus on causation and mechanism. Cell 2018;174:785–790. |
| 78. | Overmann J, Abt B, Sikorski J. Present and future of culturing bacteria. Annu Rev Microbiol 2017;71:711–730. |
| 79. | Browne HP, Forster SC, Anonye BO, Kumar N, Neville BA, Stares MD, et al. Culturing of ‘unculturable’ human microbiota reveals novel taxa and extensive sporulation. Nature 2016;533:543–546. |
| 80. | Sibley CD, Grinwis ME, Field TR, Eshaghurshan CS, Faria MM, Dowd SE, et al. Culture enriched molecular profiling of the cystic fibrosis airway microbiome. PLoS One 2011;6:e22702. |
| 81. | Tunney MM, Field TR, Moriarty TF, Patrick S, Doering G, Muhlebach MS, et al. Detection of anaerobic bacteria in high numbers in sputum from patients with cystic fibrosis. Am J Respir Crit Care Med 2008;177:995–1001. |
| 82. | Amann RI, Ludwig W, Schleifer KH. Phylogenetic identification and in situ detection of individual microbial cells without cultivation. Microbiol Rev 1995;59:143–169. |
| 83. | Griffiths RI, Whiteley AS, O’Donnell AG, Bailey MJ. Rapid method for coextraction of DNA and RNA from natural environments for analysis of ribosomal DNA- and rRNA-based microbial community composition. Appl Environ Microbiol 2000;66:5488–5491. |
| 84. | Lane DJ, Pace B, Olsen GJ, Stahl DA, Sogin ML, Pace NR. Rapid determination of 16S ribosomal RNA sequences for phylogenetic analyses. Proc Natl Acad Sci USA 1985;82:6955–6959. |
| 85. | Lane DJ. 16S/23S rRNA sequencing. In: Stackebrandt, E, Goodfellow, M, editors. Nucleic acid tecnniques in bacterial systems. New York: John Wiley and Sons; 1991. |
| 86. | Schloss PD, Gevers D, Westcott SL. Reducing the effects of PCR amplification and sequencing artifacts on 16S rRNA-based studies. PLoS One 2011;6:e27310. |
| 87. | Schloss PD. Evaluating different approaches that test whether microbial communities have the same structure. ISME J 2008;2:265–275. |
| 88. | Suzuki MT, Giovannoni SJ. Bias caused by template annealing in the amplification of mixtures of 16S rRNA genes by PCR. Appl Environ Microbiol 1996;62:625–630. |
| 89. | Hilty M, Burke C, Pedro H, Cardenas P, Bush A, Bossley C, et al. Disordered microbial communities in asthmatic airways. PLoS One 2010;5:e8578. |
| 90. | Loman NJ, Constantinidou C, Christner M, Rohde H, Chan JZ, Quick J, et al. A culture-independent sequence-based metagenomics approach to the investigation of an outbreak of Shiga-toxigenic Escherichia coli O104:H4. JAMA 2013;309:1502–1510. |
| 91. | Bengtsson-Palme J, Angelin M, Huss M, Kjellqvist S, Kristiansson E, Palmgren H, et al. The human gut microbiome as a transporter of antibiotic resistance genes between continents. Antimicrob Agents Chemother 2015;59:6551–6560. |
| 92. | Nordahl Petersen T, Rasmussen S, Hasman H, Carøe C, Bælum J, Schultz AC, et al. Meta-genomic analysis of toilet waste from long distance flights; a step towards global surveillance of infectious diseases and antimicrobial resistance. Sci Rep 2015;5:11444. |
| 93. | Pendleton KM, Erb-Downward JR, Bao Y, Branton WR, Falkowski NR, Newton DW, et al. Rapid pathogen identification in bacterial pneumonia using real-time metagenomics. Am J Respir Crit Care Med 2017;196:1610–1612. |
| 94. | Marotz CA, Sanders JG, Zuniga C, Zaramela LS, Knight R, Zengler K. Improving saliva shotgun metagenomics by chemical host DNA depletion. Microbiome 2018;6:42. |
| 95. | Charalampous TKG, O’Grady J. Applying clinical metagenomics for the detection and characterisation of respiratory infections. In: The lung microbiome. Lausanne, Switzerland: European Respiratory Society; 2019. pp. 35–49. |
| 96. | Charalampous T, Richardson H, Kay GL, Baldan R, Jeanes C, Rae D, et al. Rapid diagnostics of lower respiratory infection using nanopore-based clinical metagenomics [preprint]. bioRxiv; 2018 [accessed 2020 Jan 22]. Available from: https://www.biorxiv.org/content/10.1101/387548v1. |
| 97. | Hasan MR, Rawat A, Tang P, Jithesh PV, Thomas E, Tan R, et al. Depletion of human DNA in spiked clinical specimens for improvement of sensitivity of pathogen detection by next-generation sequencing. J Clin Microbiol 2016;54:919–927. |
| 98. | Brown AC, Bryant JM, Einer-Jensen K, Holdstock J, Houniet DT, Chan JZ, et al. Rapid whole-genome sequencing of Mycobacterium tuberculosis isolates directly from clinical samples. J Clin Microbiol 2015;53:2230–2237. |
| 99. | Doyle RM, Burgess C, Williams R, Gorton R, Booth H, Brown J, et al. Direct whole-genome sequencing of sputum accurately identifies drug-resistant Mycobacterium tuberculosis faster than MGIT culture sequencing. J Clin Microbiol 2018;56:e00666-18. |
| 100. | Willner D, Furlan M, Haynes M, Schmieder R, Angly FE, Silva J, et al. Metagenomic analysis of respiratory tract DNA viral communities in cystic fibrosis and non–cystic fibrosis individuals. PLoS One 2009;4:e7370. |
| 101. | Willner D, Haynes MR, Furlan M, Hanson N, Kirby B, Lim YW, et al. Case studies of the spatial heterogeneity of DNA viruses in the cystic fibrosis lung. Am J Respir Cell Mol Biol 2012;46:127–131. |
| 102. | Fleischmann RD, Adams MD, White O, Clayton RA, Kirkness EF, Kerlavage AR, et al. Whole-genome random sequencing and assembly of Haemophilus influenzae Rd. Science 1995;269:496–512. |
| 103. | Benson DA, Karsch-Mizrachi I, Lipman DJ, Ostell J, Wheeler DL. GenBank. Nucleic Acids Res 2003;31:23–27. |
| 104. | Cheng JM, Hiscoe L, Pollock SL, Hasselback P, Gardy JL, Parker R. A clonal outbreak of tuberculosis in a homeless population in the interior of British Columbia, Canada, 2008–2015. Epidemiol Infect 2015;143:3220–3226. |
| 105. | Bryant JM, Grogono DM, Rodriguez-Rincon D, Everall I, Brown KP, Moreno P, et al. Emergence and spread of a human-transmissible multidrug-resistant nontuberculous mycobacterium. Science 2016;354:751–757. |
| 106. | Bernardy EE, Petit RA III, Moller AG, Blumenthal JA, McAdam AJ, et al. Whole-genome sequences of Staphylococcus aureus isolates from cystic fibrosis lung infections. Microbiol Resour Announc 2019;8:e01564-18. |
| 107. | Schloss PD, Westcott SL, Ryabin T, Hall JR, Hartmann M, Hollister EB, et al. Introducing mothur: open-source, platform-independent, community-supported software for describing and comparing microbial communities. Appl Environ Microbiol 2009;75:7537–7541. |
| 108. | Caporaso JG, Kuczynski J, Stombaugh J, Bittinger K, Bushman FD, Costello EK, et al. QIIME allows analysis of high-throughput community sequencing data. Nat Methods 2010;7:335–336. |
| 109. | Knight R, Vrbanac A, Taylor BC, Aksenov A, Callewaert C, Debelius J, et al. Best practices for analysing microbiomes. Nat Rev Microbiol 2018;16:410–422. |
| 110. | Sinha R, Abu-Ali G, Vogtmann E, Fodor AA, Ren B, Amir A, et al.; Microbiome Quality Control Project Consortium. Assessment of variation in microbial community amplicon sequencing by the Microbiome Quality Control (MBQC) project consortium. Nat Biotechnol 2017;35:1077–1086. |
| 111. | Navas-Molina JA, Peralta-Sánchez JM, González A, McMurdie PJ, Vázquez-Baeza Y, Xu Z, et al. Advancing our understanding of the human microbiome using QIIME. Methods Enzymol 2013;531:371–444. |
| 112. | Schloss PD. Identifying and overcoming threats to reproducibility, replicability, robustness, and generalizability in microbiome research. MBio 2018;9:e00525-18. |
| 113. | Langille MGI, Ravel J, Fricke WF. “Available upon request”: not good enough for microbiome data! Microbiome 2018;6:8. |
| 114. | Bokulich NA, Subramanian S, Faith JJ, Gevers D, Gordon JI, Knight R, et al. Quality-filtering vastly improves diversity estimates from Illumina amplicon sequencing. Nat Methods 2013;10:57–59. |
| 115. | Quast C, Pruesse E, Yilmaz P, Gerken J, Schweer T, Yarza P, et al. The SILVA ribosomal RNA gene database project: improved data processing and web-based tools. Nucleic Acids Res 2013;41:D590–D596. |
| 116. | Lozupone C, Cota-Gomez A, Palmer BE, Linderman DJ, Charlson ES, Sodergren E, et al.; Lung HIV Microbiome Project. Widespread colonization of the lung by Tropheryma whipplei in HIV infection. Am J Respir Crit Care Med 2013;187:1110–1117. |
| 117. | McDonald D, Price MN, Goodrich J, Nawrocki EP, DeSantis TZ, Probst A, et al. An improved Greengenes taxonomy with explicit ranks for ecological and evolutionary analyses of bacteria and archaea. ISME J 2012;6:610–618. |
| 118. | Rideout JR, He Y, Navas-Molina JA, Walters WA, Ursell LK, Gibbons SM, et al. Subsampled open-reference clustering creates consistent, comprehensive OTU definitions and scales to billions of sequences. PeerJ 2014;2:e545. |
| 119. | Edgar RC. Accuracy of microbial community diversity estimated by closed- and open-reference OTUs. PeerJ 2017;5:e3889. |
| 120. | Callahan BJ, McMurdie PJ, Rosen MJ, Han AW, Johnson AJ, Holmes SP. DADA2: high-resolution sample inference from Illumina amplicon data. Nat Methods 2016;13:581–583. |
| 121. | Garijo D, Kinnings S, Xie L, Xie L, Zhang Y, Bourne PE, et al. Quantifying reproducibility in computational biology: the case of the tuberculosis drugome. PLoS One 2013;8:e80278. |
| 122. | Erb-Downward JR, Thompson DL, Han MK, Freeman CM, McCloskey L, Schmidt LA, et al. Analysis of the lung microbiome in the “healthy” smoker and in COPD. PLoS One 2011;6:e16384. |
| 123. | Pragman AA, Kim HB, Reilly CS, Wendt C, Isaacson RE. The lung microbiome in moderate and severe chronic obstructive pulmonary disease. PLoS One 2012;7:e47305. |
| 124. | Engel M, Endesfelder D, Schloter-Hai B, Kublik S, Granitsiotis MS, Boschetto P, et al. Influence of lung CT changes in chronic obstructive pulmonary disease (COPD) on the human lung microbiome. PLoS One 2017;12:e0180859. |
| 125. | Pragman AA, Lyu T, Baller JA, Gould TJ, Kelly RF, Reilly CS, et al. The lung tissue microbiota of mild and moderate chronic obstructive pulmonary disease. Microbiome 2018;6:7. |
| 126. | Durack J, Lynch SV, Nariya S, Bhakta NR, Beigelman A, Castro M, et al.; National Heart, Lung and Blood Institute’s “AsthmaNet”. Features of the bronchial bacterial microbiome associated with atopy, asthma, and responsiveness to inhaled corticosteroid treatment. J Allergy Clin Immunol 2017;140:63–75. |
| 127. | Depner M, Ege MJ, Cox MJ, Dwyer S, Walker AW, Birzele LT, et al. Bacterial microbiota of the upper respiratory tract and childhood asthma. J Allergy Clin Immunol 2017;139:826–834.e13. |
| 128. | Zhang Q, Cox M, Liang Z, Brinkmann F, Cardenas PA, Duff R, et al. Airway microbiota in severe asthma and relationship to asthma severity and phenotypes. PLoS One 2016;11:e0152724. |
| 129. | Huang YJ, Nariya S, Harris JM, Lynch SV, Choy DF, Arron JR, et al. The airway microbiome in patients with severe asthma: associations with disease features and severity. J Allergy Clin Immunol 2015;136:874–884. |
| 130. | Huang YJ, Nelson CE, Brodie EL, Desantis TZ, Baek MS, Liu J, et al. Airway microbiota and bronchial hyperresponsiveness in patients with suboptimally controlled asthma. J Allergy Clin Immunol 2011;127:372–381.e1–3. |
| 131. | Molyneaux PL, Cox MJ, Willis-Owen SA, Mallia P, Russell KE, Russell AM, et al. The role of bacteria in the pathogenesis and progression of idiopathic pulmonary fibrosis. Am J Respir Crit Care Med 2014;190:906–913. |
| 132. | Molyneaux PL, Willis-Owen SAG, Cox MJ, James P, Cowman S, Loebinger M, et al. Host-microbial interactions in idiopathic pulmonary fibrosis. Am J Respir Crit Care Med 2017;195:1640–1650. |
| 133. | Zemanick ET, Wagner BD, Robertson CE, Ahrens RC, Chmiel JF, Clancy JP, et al. Airway microbiota across age and disease spectrum in cystic fibrosis. Eur Respir J 2017;50:1700832. |
| 134. | Acosta N, Whelan FJ, Somayaji R, Poonja A, Surette MG, Rabin HR, et al. The evolving cystic fibrosis microbiome: a comparative cohort study spanning 16 Years. Ann Am Thorac Soc 2017;14:1288–1297. |
| 135. | Laguna TA, Wagner BD, Williams CB, Stevens MJ, Robertson CE, Welchlin CW, et al. Airway microbiota in bronchoalveolar lavage fluid from clinically well infants with cystic fibrosis. PLoS One 2016;11:e0167649. |
| 136. | Coburn B, Wang PW, Diaz Caballero J, Clark ST, Brahma V, Donaldson S, et al. Lung microbiota across age and disease stage in cystic fibrosis. Sci Rep 2015;5:10241. |
| 137. | Xu S, Tsai A, Sze MA, Vucic EA, Shaipanich T, Harris M, et al. Decreased microbiome diversity in the HIV small airway epithelium. Respir Res 2018;19:140. |
| 138. | Twigg HL III, Knox KS, Zhou J, Crothers KA, Nelson DE, Toh E, et al. Effect of advanced HIV infection on the respiratory microbiome. Am J Respir Crit Care Med 2016;194:226–235. |
| 139. | Beck JM, Schloss PD, Venkataraman A, Twigg H III, Jablonski KA, Bushman FD, et al.; Lung HIV Microbiome Project. Multicenter comparison of lung and oral microbiomes of HIV-infected and HIV-uninfected individuals. Am J Respir Crit Care Med 2015;192:1335–1344. |
| 140. | Sze MA, Dimitriu PA, Suzuki M, McDonough JE, Campbell JD, Brothers JF, et al. Host response to the lung microbiome in chronic obstructive pulmonary disease. Am J Respir Crit Care Med 2015;192:438–445. |
| 141. | Tsay JJ, Wu BG, Badri MH, Clemente JC, Shen N, Meyn P, et al. Airway microbiota is associated with upregulation of the PI3K pathway in lung cancer. Am J Respir Crit Care Med 2018;198:1188–1198. |
| 142. | Garg N, Wang M, Hyde E, da Silva RR, Melnik AV, Protsyuk I, et al. Three-dimensional microbiome and metabolome cartography of a diseased human lung. Cell Host Microbe 2017;22:705–716, e4. |
| 143. | Lal CV, Kandasamy J, Dolma K, Ramani M, Kumar R, Wilson L, et al. Early airway microbial metagenomic and metabolomic signatures are associated with development of severe bronchopulmonary dysplasia. Am J Physiol Lung Cell Mol Physiol 2018;315:L810–L815. |
| 144. | Bacci G, Mengoni A, Fiscarelli E, Segata N, Taccetti G, Dolce D, et al. A different microbiome gene repertoire in the airways of cystic fibrosis patients with severe lung disease. Int J Mol Sci 2017;18:E1654. |
| 145. | Cameron SJS, Lewis KE, Huws SA, Hegarty MJ, Lewis PD, Pachebat JA, et al. A pilot study using metagenomic sequencing of the sputum microbiome suggests potential bacterial biomarkers for lung cancer. PLoS One 2017;12:e0177062. |
| 146. | Millares L, Pérez-Brocal V, Ferrari R, Gallego M, Pomares X, García-Núñez M, et al. Functional metagenomics of the bronchial microbiome in COPD. PLoS One 2015;10:e0144448. |
| 147. | Lim YW, Schmieder R, Haynes M, Willner D, Furlan M, Youle M, et al. Metagenomics and metatranscriptomics: windows on CF-associated viral and microbial communities. J Cyst Fibros 2013;12:154–164. |
| 148. | Huh D, Matthews BD, Mammoto A, Montoya-Zavala M, Hsin HY, Ingber DE. Reconstituting organ-level lung functions on a chip. Science 2010;328:1662–1668. |
| 149. | Jin C, Lagoudas GK, Zhao C, Bullman S, Bhutkar A, Hu B, et al. Commensal microbiota promote lung cancer development via γδ T cells. Cell 2019;176:998–1013.e16. |
| 150. | Cho Y, Abu-Ali G, Tashiro H, Brown TA, Osgood RS, Kasahara DI, et al. Sex differences in pulmonary responses to ozone in mice: role of the microbiome. Am J Respir Cell Mol Biol 2019;60:198–208. |
| 151. | Gao X, Zhang M, Xue J, Huang J, Zhuang R, Zhou X, et al. Body mass index differences in the gut microbiota are gender specific. Front Microbiol 2018;9:1250. |
| 152. | Mueller S, Saunier K, Hanisch C, Norin E, Alm L, Midtvedt T, et al. Differences in fecal microbiota in different European study populations in relation to age, gender, and country: a cross-sectional study. Appl Environ Microbiol 2006;72:1027–1033. |
| 153. | Fulde M, Sommer F, Chassaing B, van Vorst K, Dupont A, Hensel M, et al. Neonatal selection by Toll-like receptor 5 influences long-term gut microbiota composition. Nature 2018;560:489–493. |
| 154. | Leitao Filho FS, Alotaibi NM, Ngan D, Tam S, Yang J, Hollander Z, et al. Sputum microbiome is associated with 1-year mortality following chronic obstructive pulmonary disease hospitalizations. Am J Respir Crit Care Med 2019;199:1205–1213. |
| 155. | Chanderraj R, Dickson RP. Rethinking pneumonia: a paradigm shift with practical utility. Proc Natl Acad Sci USA 2018;115:13148–13150. |
| 156. | Pendleton KM, Erb-Downward JR, Bao Y, Branton WR, Falkowski NR, Newton DW, et al. Reply: clinical metagenomics for the diagnosis of hospital-acquired infections: promises and hurdles. Am J Respir Crit Care Med 2017;196:1618–1619. |
| 157. | Kitsios GD. Translating lung microbiome profiles into the next-generation diagnostic gold standard for pneumonia: a clinical investigator’s perspective. mSystems 2018;3:e00153-17. |
| 158. | Cookson WOCM, Cox MJ, Moffatt MF. New opportunities for managing acute and chronic lung infections. Nat Rev Microbiol 2018;16:111–120. |
| 159. | van Nimwegen FA, Penders J, Stobberingh EE, Postma DS, Koppelman GH, Kerkhof M, et al. Mode and place of delivery, gastrointestinal microbiota, and their influence on asthma and atopy. J Allergy Clin Immunol 2011;128:948–955.e1-3. |
| 160. | van de Pol MA, Lutter R, Smids BS, Weersink EJ, van der Zee JS. Synbiotics reduce allergen-induced T-helper 2 response and improve peak expiratory flow in allergic asthmatics. Allergy 2011;66:39–47. |
| 161. | Wei X, Jiang P, Liu J, Sun R, Zhu L. Association between probiotic supplementation and asthma incidence in infants: a meta-analysis of randomized controlled trials. J Asthma 2019;18:1–12. |
| 162. | Lin J, Zhang Y, He C, Dai J. Probiotics supplementation in children with asthma: a systematic review and meta-analysis. J Paediatr Child Health 2018;54:953–961. |
| 163. | Dickson RP. The lung microbiome and ARDS: it’s time to broaden the model. Am J Respir Crit Care Med 2018;197:549–551. |
| 164. | Dickson RP, Cox MJ. Gut microbiota and protection from pneumococcal pneumonia. Gut 2017;66:384. |
| 165. | Marsland BJ, Trompette A, Gollwitzer ES. The gut–lung axis in respiratory disease. Ann Am Thorac Soc 2015;12:S150–S156. |
| 166. | Schuijt TJ, Lankelma JM, Scicluna BP, de Sousa e Melo F, Roelofs JJ, de Boer JD, et al. The gut microbiota plays a protective role in the host defence against pneumococcal pneumonia. Gut 2016;65:575–583. |
| 167. | Barfod KK, Vrankx K, Mirsepasi-Lauridsen HC, Hansen JS, Hougaard KS, Larsen ST, et al. The murine lung microbiome changes during lung inflammation and intranasal vancomycin treatment. Open Microbiol J 2015;9:167–179. |
| 168. | Sulaiman I, Wu BG, Li Y, Scott AS, Malecha P, Scaglione B, et al. Evaluation of the airway microbiome in non-tuberculous mycobacteria. Eur Respir J 2018;52:1800810. |
| 169. | Větrovský T, Baldrian P. The variability of the 16S rRNA gene in bacterial genomes and its consequences for bacterial community analyses. PLoS One 2013;8:e57923. |
Supported, in part, by the Division of Intramural Research, National Institute of Environmental Health Sciences.
Author Contributions: Conception, design, and critical revision of the manuscript for important intellectual content and approval of the final version to be published: S.M.C., J.C.C., M.J.C., R.P.D., Y.J.H., G.D.K., K.M.K., J.M.L., T.D.L., P.L.M., B.B.M., D.N.O'D., L.N.S., and S.G.
This article has a data supplement, which is accessible from this issue’s table of contents at www.atsjournals.org.
Originally Published in Press as DOI: 10.1165/rcmb.2019-0273TR on October 29, 2019
Author disclosures are available with the text of this article at www.atsjournals.org.
