key: cord-0429450-7ahoiaxs
authors: Adams, Taylor S.; Schupp, Jonas C.; Poli, Sergio; Ayaub, Ehab A.; Neumark, Nir; Ahangari, Farida; Chu, Sarah G.; Raby, Benjamin A.; DeIuliis, Giuseppe; Januszyk, Michael; Duan, Qiaonan; Arnett, Heather A.; Siddiqui, Asim; Washko, George R.; Homer, Robert; Yan, Xiting; Rosas, Ivan O.; Kaminski, Naftali
title: Single Cell RNA-seq reveals ectopic and aberrant lung resident cell populations in Idiopathic Pulmonary Fibrosis
date: 2019-09-09
journal: bioRxiv
DOI: 10.1101/759902
sha: e3db6d266f45d0a8b091e95e053ff3ffbac34caf
doc_id: 429450
cord_uid: 7ahoiaxs

We provide a single cell atlas of Idiopathic Pulmonary Fibrosis (IPF), a fatal interstitial lung disease, focusing on resident lung cell populations. By profiling 312,928 cells from 32 IPF, 29 healthy control and 18 chronic obstructive pulmonary disease (COPD) lungs, we demonstrate that IPF is characterized by changes in discrete subpopulations of cells in the three major parenchymal compartments: the epithelium, endothelium and stroma. Among epithelial cells, we identify a novel population of IPF enriched aberrant basaloid cells that co-express basal epithelial markers, mesenchymal markers, senescence markers, developmental transcription factors and are located at the edge of myofibroblast foci in the IPF lung. Among vascular endothelial cells in the in IPF lung parenchyma we identify an expanded cell population transcriptomically identical to vascular endothelial cells normally restricted to the bronchial circulation. We confirm the presence of both populations by immunohistochemistry and independent datasets. Among stromal cells we identify fibroblasts and myofibroblasts in both control and IPF lungs and leverage manifold-based algorithms diffusion maps and diffusion pseudotime to infer the origins of the activated IPF myofibroblast. Our work provides a comprehensive catalogue of the aberrant cellular transcriptional programs in IPF, demonstrates a new framework for analyzing complex disease with scRNAseq, and provides the largest lung disease single-cell atlas to date.

We provide a single cell atlas of Idiopathic Pulmonary Fibrosis (IPF), a fatal interstitial lung disease, focusing on resident lung cell populations. By profiling 312,928 cells from 32 IPF, 29 healthy control and 18 chronic obstructive pulmonary disease (COPD) lungs, we demonstrate 25 that IPF is characterized by changes in discrete subpopulations of cells in the three major parenchymal compartments: the epithelium, endothelium and stroma. Among epithelial cells, we identify a novel population of IPF enriched aberrant basaloid cells that co-express basal epithelial markers, mesenchymal markers, senescence markers, developmental transcription factors and are located at the edge of myofibroblast foci in the IPF lung. Among vascular endothelial cells in the 30 in IPF lung parenchyma we identify an expanded cell population transcriptomically identical to vascular endothelial cells normally restricted to the bronchial circulation. We confirm the presence of both populations by immunohistochemistry and independent datasets. Among stromal cells we identify fibroblasts and myofibroblasts in both control and IPF lungs and leverage manifold-based algorithms diffusion maps and diffusion pseudotime to infer the origins of the activated IPF 35 myofibroblast. Our work provides a comprehensive catalogue of the aberrant cellular transcriptional programs in IPF, demonstrates a new framework for analyzing complex disease with scRNAseq, and provides the largest lung disease single-cell atlas to date.

Idiopathic Pulmonary Fibrosis (IPF) is a progressive lung disease characterized by irreversible scarring of the distal lung, leading to respiratory failure and death (1, 2). Despite significant progress in our understanding of pulmonary fibrosis in laboratory animals, we have a limited perspective of the cellular and molecular processes that determine the IPF lung phenotype. In fact, 45 IPF is still best described by its histopathological pattern of usual interstitial pneumonia (UIP), that includes presence of fibroblast foci; hyperplastic alveolar epithelial cells that localize adjacent to fibroblastic foci; a distortion of airway architecture combined with an accumulation of microscopic airway-epithelial lined cysts known as "honeycombs" in the distal parenchyma and the lack of evidence for other conditions(1-3). 50 Evidence for molecular aberrations in the IPF lung have mostly been obtained by following hypotheses derived from animal models of disease, from discovery of genetic associations in humans, or from genes differentially expressed in transcriptomic studies of bulk IPF tissue with limited cellular resolution (4, 5) . Recent studies have demonstrated the value of single cell RNA sequencing (scRNAseq) by identifying profibrotic macrophages in lungs of human and mice with 55 pulmonary fibrosis (6, 7). Here we harness the cell-level resolution afforded by scRNAseq to provide an atlas of the extent of complexity and diversity of aberrant cellular populations in the three major parenchymal compartments of the IPF lung: the epithelium, endothelium and stroma.

We profiled 312,928 cells from distal lung parenchyma samples obtained from 32 IPF, 18 COPD and 29 control donor lungs (Table S1 and Fig. 1A) , identifying 38 discrete cell types (Fig. 1B) based on distinct markers ( Figure 1C , Data S1-S4). Manually curated cell classifications are consistent with automated annotations drawn from several independent databases (Fig. S4) . The detailed cellular repertoires of epithelial, endothelial and mesenchymal cells are provided below. 65 Our data will be available on GEO (GSE136831) and the results can be explored through the IPF Cell Atlas Data Mining Site (8).

The epithelial cell repertoire of the fibrotic lung is dramatically changed and contains disease associated aberrant basaloid cells. In non-diseased tissue we identified all known lung epithelial cells populations including, alveolar type 1 and type 2 cells, ciliated cells, basal cells, goblet cells, (19, 20) . No aberrant basaloid cells could be detected in control lungs ( Fig. 2A-B) .

Of the 483 aberrant basaloid cells identified, 448 were from IPF lungs compared to just 33 cells from COPD lungs. We confirmed the presence and localized these cells in the IPF lung by immunohistochemistry (IHC) using p63, KRT17, HMGA2, COX2 and p21 as markers (Fig. 2D ).

In IPF lungs, these cells consistently localize to the epithelial layer covering myofibroblast foci. 3D ). In IPF lungs, COL15A1+ CD31+ VE cells are observed in the distal lung, typically at the parenchymal edge of fibroblastic foci and in areas of bronchiolization ( Fig 3D) . Re-analysis of a recently published scRNAseq dataset that contained normal airway and lung parenchyma samples 110 (23) confirmed that genes specific to the pVE were not observed in distal lung VE (Fig. 3E ).

Collectively, these observations indicate that COL15A1+ VE cells, represent an ectopic pVE population in the distal lung in IPF.

The IPF lung exhibits disease-specific archetypes among fibroblasts and myofibroblast. To 5A ). The topology of the IPF GRN exhibited higher density and modularity (Fig. 5B ). Comparing the array of cellular contributions to communities that comprise each GRN, we found that control GRN communities show a relatively diverse array of cellular contributions both within 140 communities and across them -consistent with organ function under homeostatic conditions. In IPF GRN, the major epithelial-associated communities remain largely isolated from the rest of the network, whose community with the highest density is predominantly driven by aberrant basaloid cells (Fig. 5C ). Using PageRank (28) centrality as a proxy for a gene's influence on the network, we identified the top 300 influencer genes ranked by PageRank differential compared to the control 145 GRN (Fig. 5D) . Gene set enrichment of these genes returned cellular aging, senescence, response to TGFB1, epithelial tube formation and smooth muscle cell differentiation (Figure5E). These findings underscore the heterotypic cellular contributions to the many key processes of molecular aberrancies in IPF.

In this study, we provide a single cell atlas of the IPF lung, with a focus on aberrant epithelial, well as molecules associated with IPF, and genes that are typically expressed in other organs. 175 Importantly, they express SOX9, a transcription factor critical to distal airway development, repair, also indicated in oncogenesis (19, 20) . Anatomically, these cells localize to a highly enigmatic region in the IPF lung, at the active edge of the myofibroblast foci. Thus, the identification of these cells may provide a critical answer to the question of the monolayer of cuboidal epithelial cells found on the surface of IPF myofibroblastic foci. Sometimes referred to as hyperplastic alveolar 180 epithelial cells (29, 30) , the nature of this cells been a source for controversy, because they could not be fully characterized with previous techniques (30) (31) (32) . Taken together with previous observations demonstrating the presence of p63+ cells undergoing partial EMT (33, 34), our results imply that this population is indeed distinct from resident alveolar epithelial cell populations and perhaps derived from a rare progenitor niche with the potential to serve as secondary progenitor 185 for depleted AT1 and AT2 cells in normal human lungs, much like broncho-alveolar stem cells are known to do in the murine lung (35, 36) . In IPF, repeated damage and potential genetic predisposition to replicative exhaustion could perhaps lead to the conflicting state of proliferation, differentiation and senescence apparent in these aberrant basaloid cells. Speculation aside, the revelation that these cells are fundamentally basaloid in nature, as well as the expansion of regular 190 airway basal cells in the IPF lung serves to couple the two most distinct histopathological features of IPF -fibroblastic foci and honeycomb cysts -to a singular commonality: the migration of airway basal cells from their natural airway niche into the distal lung in a failed attempt to repair the lung.

The absence of discriminating markers for lung vascular endothelial cells, and the difficulty in culturing these cells has limited investigations into vascular remodeling in IPF. Our scRNAseq 195 analysis led to the unexpected discovery of an expanded VE cell population that expresses COL15A1 in the IPF lung. These peribronchial VE cells are transcriptomically indistinguishable from systemically supplied bronchial VE cells detected in control lungs, but they localize to the vessels underneath fibroblastic foci and honeycomb cysts in IPF. While lacking support in more recent observations, it is possible that our discovery provides the cellular molecular correlate of 200 Turner-Warwick's 1963 observation that the bronchial vascular network is expanded throughout the IPF lung (37) . While it is impossible at this stage to tell whether the ectopic presence of peribronchial VE cells is involved in the pathogenesis of IPF, this novel finding fits a larger pattern in the IPF lung, where cellular population normally relegated to the airways are found in affected regions in the distal parenchyma. 205 While it is well recognized that lung fibroblast populations demonstrate considerable plasticity (35, (38) (39) (40) (41) (42) , cell-types are traditionally defined by the presence of a singular molecular feature. One relevant example being the use of ACTA2 to define the IPF myofibroblast that comprise the disease's lesions. The cellular source of this pathological cell population remains a matter of considerable debate, as cells that satisfy this absolute definition are not present in the 210 normal adult lung parenchyma. scRNAseq analysis provides a more comprehensive view of cellular identity, where global transcriptomic features -rather than singular cell markers -are exploited to define cells and assess the extent of population variance. Our analysis suggests that pathological, ACTA2-expressing IPF myofibroblast are not a discrete cell-type, but rather one extreme pole of a continuum connected to a quiescent ACTA2-negative stromal population 215 represented in control lungs. We find the cells that belong to this continuum can collectively be distinguished from other lung fibroblast by a panel of genes which include several myosin and contractile-associated features, suggesting that the pathological phenotype observed in IPF is a latent feature of this population, rather than the result of trans-differentiation from a discrete cellular population. More explicitly, we did not observe any evidence suggesting that resident lung 220 fibroblasts or activated fibroblasts were the source for IPF myofibroblasts.

Our results provide a comprehensive portrait of the fibrotic niche in IPF: where aberrant basaloid cells interface with aberrantly activated myofibroblasts, forming a lesion vascularized by ectopic bronchial vessels in the presence of profibrotic monocyte-derived macrophage. The 225 identification and detailed description of aberrant cell populations in the IPF lung may lead to identification of novel, cell-type specific therapies and biomarkers. Lastly, our lung cell atlas (8) provides an exhaustive reference of pulmonary cellular diversity in both healthy and diseased lung. GRNs with the top 300 nodes ranked by differential PageRank centrality between IPF and control 305 highlighted in red. Nodes sizes correspond to PageRank centralities. (E) Selected results from GO gene set enrichment of the top 300 differential PageRank nodes between IPF and controls, with all nodes used as a reference.

was deposited to the Gene Expression Omnibus under accession number GSE136831. A userfriendly and accessible online tool for our data is available at www.IPFCellAtlas.com.

Idiopathic Pulmonary Fibrosis

Idiopathic pulmonary fibrosis

Diagnosis of Idiopathic Pulmonary Fibrosis. An Official ATS/ERS/JRS/ALAT Clinical Practice Guideline

Fibrosis: Lessons from OMICS analyses of the human lung

Translational research in pulmonary fibrosis

Single-Cell Transcriptomic Analysis of Human Lung Provides Insights into the Pathobiology of Pulmonary Fibrosis

Reference-based analysis of lung single-cell sequencing reveals a transitional profibrotic macrophage

IPF Cell Atlas

Biomarkers for epithelial-mesenchymal transitions

Binding of alpha v beta 1 and alpha v beta 6 integrins to tenascin-C induces epithelial-mesenchymal transition-like change of breast cancer cells

HMGA2 Maintains Oncogenic RAS-Induced Epithelial-Mesenchymal Transition in Human Pancreatic Cancer Cells

The miR-200 family and miR-205 regulate epithelial to mesenchymal 360 transition by targeting ZEB1 and SIP1

Growth differentiation factor 15 (GDF15): A survival protein with therapeutic potential in metabolic diseases

Cellular Senescence as a Mechanism and Target in Chronic Lung Diseases

Relevance of the p53-MDM2 axis to aging

Gene expression analysis reveals matrilysin as a key regulator of pulmonary fibrosis in mice and humans

The integrin alpha v beta 6 binds and activates latent TGF beta 1: A 370 mechanism for regulating pulmonary inflammation and fibrosis

ADAM10-mediated ephrin-B2 shedding promotes myofibroblast activation and organ fibrosis

Hippo signaling promotes lung epithelial lineage commitment by curbing Fgf10 and beta-catenin signaling

Human lung branching morphogenesis is orchestrated by the spatiotemporal distribution of ACTA2, SOX2, and SOX9

Pathognomonic remodeling of blood and lymphatic capillaries in idiopathic pulmonary fibrosis

Tissue-based map of the human proteome

A cellular census of human lungs identifies novel cell states in health and in asthma

Diversity of Interstitial Lung Fibroblasts Is Regulated by Platelet-Derived Growth Factor Receptor alpha Kinase Activity

CXCL14 is a candidate biomarker for Hedgehog signalling in idiopathic pulmonary fibrosis

PAGA: graph abstraction reconciles clustering with trajectory inference through a topology preserving map of single cells

bigSCale: an analytical framework for big-scale single-cell data

Role of epithelial cells in idiopathic pulmonary fibrosis: from 395 innocent targets to serial killers

Idiopathic pulmonary fibrosis: immunohistochemical analysis provides fresh insights into lung tissue remodelling with implications for novel prognostic markers

Induction of epithelial-mesenchymal transition in alveolar epithelial cells by transforming growth factor-beta1: potential role in idiopathic pulmonary fibrosis

Epithelial-mesenchymal transition in human lungs with usual interstitial pneumonia: Quantitative immunohistochemistry

Abnormal re-epithelialization and lung remodeling in idiopathic pulmonary fibrosis: the role of deltaN-p63

Aberrant Wnt/beta-catenin pathway activation in idiopathic pulmonary fibrosis

Multiple stromal populations contribute to pulmonary fibrosis without 410 evidence for epithelial to mesenchymal transition

Identification of bronchioalveolar stem cells in normal lung and lung cancer

Precapillary Systemic-Pulmonary Anastomoses

Fibrotic extracellular matrix activates a profibrotic positive feedback loop

Fibroblast senescence in the pathology of idiopathic pulmonary fibrosis

PD-L1 on invasive fibroblasts drives fibrosis in a humanized model of idiopathic pulmonary fibrosis

The Toll-like receptor 3 L412F polymorphism and disease progression in idiopathic pulmonary fibrosis

Single-Cell Deconvolution of Fibroblast Heterogeneity in Mouse Pulmonary Fibrosis

We are indebted to all patients and control subjects who participated in this 310 study. The sequencing was conducted by Mei Zhong at Yale Stem Cell Center Genomics Core facility. We thank Amos Brooks and the staff of Yale Pathology Tissue Services for tissue processing. We also thank Martijn C. Nawijn for providing access to his group's data. Funding:This work was supported by NIH grants R01HL127349, U01HL145567, U01HL122626, and Theravance, LifeMax, Three Lake Partners, Optikira over the last 3 years and received nonfinancial support from MiRagen. Data and materials availability: All raw count expression data