Until recently it was assumed that the healthy mammalian lung did not harbour a microbiota, unlike other body sites. However, through the use of sequencing based technologies this has been shown to not be the case. Low biomass communities of microbes can be identified in the healthy lung and the lung microbiota in various diseases states has been shown to differ form these 'healthy' communities. The sheep respiratory microbiota is of interest from both an animal health perspective and due to the potential use of the sheep as a large animal model for studying the lung microbiota. In this thesis I seek to characterise the composition and variability of the sheep lung microbiota; the differences between the sheep upper and lower respiratory tract bacterial communities and to assess whether exhaled breath condensate collection can be used as a non-invasive lung microbiota sampling method. To study the bacterial communities present in samples I have used 16S rRNA gene sequencing and analysis. In Chapter 3 I examine the inter-individual and spatial variability present within the sheep lung microbiota. Protected specimen brushings were collected from three lung segments in six animals at three time-points. In a separate sheep a greater number of brushings was taken (n=16) in order to examine the amount of variability over a smaller spatial scale. I find that there can be large differences between the bacterial communities isolated from different locations within the lung, even over short distances. Samples also cluster by the sheep from which they were taken, indicating a host specific influence on the lung microbiota. In Chapter 4 I compare whole lung washes and oropharyngeal swabs from 40 lambs in order to examine the differences between the upper and lower respiratory tract microbiotas. I find that oropharyngeal swabs separate into rumen-like or upper respiratory tract-like bacterial communities. Despite the fact that in humans the upper and lower respiratory microbiotas have been shown to have similar compositions, the sheep lung microbiota samples in this study do not resemble either oropharyngeal samples or reagent only controls. In my first two results chapters, lung sampling methods were used which involved either anaesthesia combined with a bronchoscopic procedure (Chapter 3) or samples being taken from dead animals (Chapter 4). In Chapter 5 I assess whether there is a less invasive way of taking lung microbiota samples from a living individual, both to minimise the procedural stress on animals used as models and to increase the pool of potential volunteers for human lung microbiota studies. I compared samples taken via protected specimen brushings to samples taken via exhaled breath condensate collection, a less invasive sampling technique. I find that condensate samples contain less bacterial DNA and different bacteria than brushing samples, indicating that it is unlikely they could be used as a replacement for invasive sampling methods. In my final results chapter I compare the results across Chapters 3, 4 and 5 to identify bacteria which occur consistently in the sheep lung and could therefore potentially be described as core lung microbiota members. In conclusion, while I have found that there are large differences between the sheep lung microbiota and that which has previously been described in humans, the sheep can still be of use as a model in studies where these differences would not have a significant impact, such as in Chapter 5 of this thesis. I have identified several bacterial members of the core sheep lung microbiota which in future it would be interesting to better characterise and to assess whether they play a role in sheep health.

The bronchiectatic lung is a diseased state in which the airways are chronically damaged and dilated. This state is found in the clinical entities of cystic fibrosis and non-cystic fibrosis bronchiectasis. These are two highly relevant chronic suppurative lung diseases in which an understanding of the microbiology of these patients is considered key to appropriate management. This has traditionally been via the use of traditional culture techniques. However, with the development of molecular methodologies, the previously perceived wisdom is being challenged. In both cystic fibrosis and non-cystic fibrosis bronchiectasis, Pseudomonas aeruginosa is considered the most significant pathogen. In CF there has been considerable concern surrounding the risk of transmission of Pseudomonas aeruginosa between patients on the basis of a significant quantity of research into this matter. In contrast, there has been very little research performed into the equivalent risk in non-cystic fibrosis bronchiectasis. In this thesis we describe an extensive single-centre epidemiological review of Pseudomonas aeruginosa spanning both these diseases. Via this we have shown evidence of cross-infection within a non-cystic fibrosis bronchiectasis cohort. This epidemiological review has included multiple genotyping methods including multilocus sequence typing and whole genome sequencing, As an extension of the epidemiological review, we have performed an in silico prediction of hypermutator status from the whole genome sequencing data to provide greater understanding of the likelihood of cross-infection, and have also demonstrated a culture-independent adaption of multilocus sequence typing for potential screening for cross-infection. In addition to Pseudomonas aeruginosa, we have also looked at the wider bacterial community in the lungs of patients with these two conditions via culture-independent techniques. We have shown that whilst Pseudomonas aeruginosa is often an important component, these are clearly complex communities. We have primarily investigated the cohort with non-cystic fibrosis bronchiectasis, but we have demonstrated associations between clinically-relevant markers and complexity of the bacterial communities within the lungs of both these cohorts of patients. Whilst we have used the gold-standard technique of 16S rRNA sequencing, we have also shown the validity of a simple and potentially more feasible profiling technique for standard clinical care. In summary, through the application of culture-dependent and independent molecular techniques, this research has shed light on the epidemiology of Pseudomonas aeruginosa within our respiratory cohorts, and the complexity and clinical relevance of the wider microbial communities within these patients. Such studies are essential if we are to advance our understanding of the bronchiectatic lung and optimise strategies for patient management.

Le Virus Respiratoire Syncytial (VRS) est responsable de 70 % des cas de bronchiolite chez les enfants de moins de cinq ans. La survenue de bronchiolites sévères chez le nourrisson est un facteur de risque de développement d’asthme en grandissant. Aucun vaccin contre le VRS n’est disponible chez l’Homme. Le système immunitaire inné est la première ligne de défense de l’organisme contre les infections. De plus, en interaction avec la flore bactérienne commensale des poumons, l’immunité innée participe à la maturation de la réponse immunitaire adaptative qui confère à l’individu une protection sur le long terme vis-à-vis des pathogènes. Afin d’expliquer la susceptibilité néonatale au VRS, nous avons caractérisé un nouveau mécanisme de contrôle de la réponse innée antivirale lors de l’infection de souriceaux. Nous avons également testé une nouvelle approche de modulation de la réponse immunitaire au VRS par le microbiote pulmonaire. Ainsi, mieux comprendre les mécanismes immunologiques et virologiques responsables de bronchiolites sévères en période néonatale permettra de développer des moyens de lutte sûrs et efficaces contre l’infection par le VRS. / Respiratory Syncytial Virus (RSV) is responsible for 70 % of bronchiolitis in children under five years old. Severe bronchiolitis in infants is a risk factor for asthma development. No vaccine against RSV is available in humans. The innate immune system is the first line of defense against infections. Moreover, in interaction with lung microbiota, innate immunity shapes adaptive immune response responsible for long-term protection against pathogens. To explain the susceptibility of young children to RSV, we characterized a novel regulatory mechanism of the innate antiviral response during neonatal RSV infection in the murine model. We also tested a new approach for modulating immune responses to RSV by the pulmonary microbiota. Thus, a better understanding of immunological and virological mechanisms responsible for severe bronchiolitis during the neonatal period will allow the development of safe and effective therapeutic strategies against RSV infection.

L'établissement d'un répertoire exhaustif ainsi que son élargissement constituent les deux objectifs principaux de ce travail. Nous avons d'abord établi la toute première liste de bactéries identifiées par culture des voies respiratoires au travers de la littérature scientifique. Nous répertorions ici 756 espèces, ce qui représente 27,23% de l'ensemble des bactéries isolées chez l'homme lorsque comparé au répertoire établi récemment par Bilen et al. Parmi ces bactéries, 514 avaient déjà été isolées au moins une fois dans les poumons. Plus de la moitié (i.e., 65,5%) des bactéries isolées pour la première fois dans des échantillons de poumons, ont été identifiées après les années 2000, soulignant la nécessité de poursuivre les efforts pour cultiver des microbes à partir des échantillons de voies respiratoires. Nous pensons que la combinaison de méthodes de culture à grande échelle telles que la culturomique et la métagénomique aidera à mieux décrire le microbiote pulmonaire. Des études antérieures sur le microbiote digestif le démontre. Nous avons ensuite utilisé des approches culturomiques et métagénomiques pour explorer le microbiote respiratoire d'individus sains. Nous avons isolé 193 bactéries par culturomics. Parmi ceux-ci, nous avons ajouté 84 au répertoire du microbiote respiratoire, dont 14 nouvelles espèces. En utilisant des approches métagénomiques, 139 OTU identifiées au rang de l'espèce dont seulement 49 (17,3%) étaient également retrouvées par culturomique, confortant la complémentarité des deux approches. Enfin, nous avons utilisé la taxonogénomique, une nouvelle approche permettant la description de nouvelles espèces bactériennes, pour décrire 19 bactéries. / The establishment of a comprehensive directory and its expansion through the use of high-speed culture methods are the two main objectives of this thesis work. We first established the first list of bacteria identified by airway culture through the scientific literature. Here we list 756 species, representing 27.23% of all bacteria isolated from humans when compared to the recently established repertoire of Bilen et al. Of these bacteria, 514 had already been isolated at least once in the lungs. Considering bacteria isolated for the first time in lung samples, more than half (ie, 65.5%) were identified after the 2000s, highlighting the need for continued efforts to grow microbes from lane samples respiratory. We believe that the combination of large scale culture methods such as culturomics and metagenomics will help to better describe the pulmonary microbiota. Previous studies on the digestive microbiota have shown the complementarity of these two approaches. We then used culturomic and metagenomic approaches to explore the respiratory microbiota of healthy individuals. We isolated 193 bacteria by culturomics. Of these, we added 84 bacteria to the repertoire of the respiratory microbiota, including 14 new species discovered. Using metagenomic approaches, 139 OTUs identified with the rank of the species of which only 49 (17.3%) were also recovered by culturomics, reinforcing the complementarity of the two approaches. Finally, we used taxonogenomics, a new approach for describing new bacterial species by integrating genomic and proteomic data with those that are classically integrated. Using this approach, 19 bacteria were described as part of this work.