Investigating pulmonary fibrosis using a single-cell transcriptomics approach

Abstract

Fibrosis is the final common endpoint of chronic inflammation and aberrant tissue repair. There are several conditions that cause pervasive fibrosis in the human lung, the commonest of which is Idiopathic Pulmonary Fibrosis (IPF). IPF is characterised by progressive breathlessness ultimately leading to respiratory failure and death within 3 to 5 years of diagnosis. Recent advances in therapeutics have seen the introduction of specific antifibrotic therapies in IPF. However, these agents are poorly tolerated and only act to slow the decline of lung function. There is a significant need to identify new diagnostic tests to enable early detection and new therapies to effectively treat disease.

Single-cell RNA sequencing (scRNA-seq) is a technique that facilitates the unbiased profiling of the transcriptomes of individual cells. Studying transcriptomic information at this resolution allows for the interrogation of complex cellular networks and the identification of rare sub-populations in single-cell suspensions derived from tissues. A variation of this technique, termed single-nuclei RNA sequencing (snRNA-seq), utilises a similar methodology but relies upon the sequencing of nuclei isolated from individual cells, enabling its application to archival frozen samples.

I hypothesised that single-cell transcriptomic methods can be used to identify disease-specific subpopulations within the pulmonary fibrotic niche, and that these populations interact to drive fibrogenesis. I used a murine model of bleomycin-induced pulmonary fibrosis, optimised tissue digestion and lineage enrichment protocols, to successfully perform scRNA-seq of mesenchymal and endothelial cells at two time points - healthy and 14-days following bleomycin injury. This analysis identified two disease-specific populations of fibroblasts in the mesenchymal dataset: one novel (activated peribronchial fibroblasts) and one described recently in the literature (Cthrc1+ fibroblasts). To my knowledge, this is also the first detailed single-cell transcriptomic dataset of endothelial cells from a model of pulmonary fibrosis. In this lineage I identified differentially expressed genes at day-14 post-bleomycin, outlining broad endothelial responses in this model. I combined these two datasets with an existing scRNA-seq dataset of myeloid cells generated in our group to create a multi-lineage single-cell transcriptomic atlas of non-epithelial cells in this model. I also performed snRNA-seq on frozen samples of mouse lung from this model, producing a dataset rich in epithelial cells as no enzymatic dissociation was required for this technique. Using this atlas I identified the emergence of disease-specific cellular changes and populations (such as transitioning epithelial cells), corroborating existing evidence and identifying molecular targets for further investigation.

My second aim was to characterise the topography, dynamics and function of disease-specific subpopulations, with a specific focus on the endothelial lineage. I identified the increased expression of Ntrk2, which encodes for TrkB, a receptor tyrosine kinase, in endothelial cells at day-14 following bleomycin injury. I utilised immunofluorescence to validate this finding in mouse lung tissue sections, confirming the increased expression of TrkB along the course of fibrogenesis, peaking at day-14. Using combined immunofluorescence and RNA fluorescence in situ hybridisation, I demonstrated that TrkB expression was associated with collagen accumulation, but not restricted to areas of active collagen deposition. TrkB expression was associated with HIF-1α expression, suggesting that spatial expression of TrkB could be indicative of tissue hypoxia. I also identified that the expression of BDNF, a key TrkB ligand, was predominantly in ATI cells but also in disease-specific mesenchymal subpopulations (Cthrc1+ fibroblasts and activated peribronchial fibroblasts) with BDNF expression levels (determined by immunofluorescence), decreasing following bleomycin injury. I investigated the function of endothelial TrkB expression in this model by administering a TrkB agonist - 7-8-dihydroxyflavone – demonstrating that TrkB agonism does not impact collagen deposition, morbidity or mortality.

My final aim in this project was to identify transcriptomic signals associated with disease progression by performing snRNA-seq on archived bronchoalveolar-lavage (BAL) samples from patients with IPF. I proved that this was technically feasible and produced a dataset comprising macrophages and T-cells, reflecting the cellular composition of the alveolar space. My analysis of this dataset demonstrated heterogeneity in the macrophage populations sequenced, partly explained by macrophage maturity and pro-fibrotic features.

Collectively my findings in this project demonstrate the utility of single-cell transcriptomics to identify disease-specific subpopulations in mouse and human samples of pulmonary fibrosis. I have validated my findings particularly in the endothelial compartment, and have also generated rich datasets for further downstream analysis. To my knowledge, this is also the first demonstration of the potential of snRNA-seq on frozen BAL samples, potentially vastly expanding the utility of this technique on archived samples to improve our understanding of IPF and other respiratory diseases.