Collagen-producing lung cell atlas identifies multiple subsets with distinct localization and relevance to fibrosis

Abstract

Collagen-producing cells maintain the complex architecture of the lung and drive pathologic scarring in pulmonary fibrosis. Here we perform single-cell RNA-sequencing to identify all collagen-producing cells in normal and fibrotic lungs. We characterize multiple collagen-producing subpopulations with distinct anatomical localizations in different compartments of murine lungs. One subpopulation, characterized by expression of Cthrc1 (collagen triple helix repeat containing 1), emerges in fibrotic lungs and expresses the highest levels of collagens. Single-cell RNA-sequencing of human lungs, including those from idiopathic pulmonary fibrosis and scleroderma patients, demonstrate similar heterogeneity and CTHRC1-expressing fibroblasts present uniquely in fibrotic lungs. Immunostaining and in situ hybridization show that these cells are concentrated within fibroblastic foci. We purify collagen-producing subpopulations and find disease-relevant phenotypes of Cthrc1-expressing fibroblasts in in vitro and adoptive transfer experiments. Our atlas of collagen-producing cells provides a roadmap for studying the roles of these unique populations in homeostasis and pathologic fibrosis.

Fig. 1. scRNA-seq of murine lung cells in normal and fibrotic lungs.

a Schematic of scRNA-seq sample preparation. b Uniform manifold approximation and projection (UMAP) plot of all cells colored by GFP⁺ and GFP⁻ samples. c Col1a1 expression on UMAP plot of all cells. See Supplementary Fig. 1a for identifying the lineages. NK, natural killer cell; Neut, neutrophil; Mac, macrophage; DC, dendritic cell; Mono, monocyte. d–f UMAP plots of Col1a1⁺ cells. d Unsupervised clustering identifies 12 clusters. e Cells from bleomycin-treated lung are shown in red. Cells from untreated lungs are shown in blue. f, g Gene expression levels for each gene.

Fig. 2. Identification of alveolar, adventitial, and peribronchial fibroblasts in untreated lungs.

a Violin plots showing the expression levels in each cluster of representative marker genes. b, c Proximity ligation in situ hybridization (PLISH) images for Npnt (white) and Ces1d (magenta) (b), or for Pi16 (white) and Adh7 (magenta) (c). Magnified images of the white squares are shown in right panels. Arrows indicate co-localization of PLISH signals in GFP⁺ cells. d PLISH images for Ccl11 (white) and Adh7 (magenta). e PLISH images for Hhip (white) and Aspn (magenta). Magnified images of the white square are shown in right panels. Arrows indicate co-localization of PLISH signals in GFP⁺ cells. b–e Col-GFP is shown in green. DAPI signal is shown in blue. Scale bars, 50 μm. aw, airway; bv, blood vessel; cuff, cuff space. Images are representative of three experiments (n ≥ 2).

Fig. 3. Characterization of alveolar, adventitial, and peribronchial fibroblasts.

a–d Cleared thick sections of untreated lungs were imaged by confocal microscopy. Col-GFP is shown in green. DAPI is shown in blue. aw, airway; bv, blood vessel; cuff, cuff space. Images are representative from two experiments (n ≥ 2). a, b α-SMA staining is shown in magenta. Arrows indicate peribronchial fibroblasts. a Magnified image of the white square is shown in the right panel. Scale bars, 50 μm (left panel), 20 μm (right panel). b Series of z-stack images were 3D-reconstructed. See also Supplementary Movie 1. c Collagen 4 staining is shown in magenta. Magnified images of the white squares are shown in the right panels. Arrowheads indicate basement membranes. Scale bars, 50 μm (left panel), 20 μm (center and right panels). See also Supplementary Movie 2. d Representative image of Shh-Cre/Rosa26-lox-stop-lox-tdTomato/Col-GFP mice. tdTomato signal is shown in magenta. Arrows indicate the close localization of alveolar fibroblasts and AEC2s. Scale bars, 50 μm. See also Supplementary Movie 3. e Collagen-producing subpopulations identified are shown on UMAP plot of Col1a1⁺ cells. f Schematic showing the distinct localization of fibroblast subpopulations.

Fig. 4. Cthrc1⁺ cells express pathologic ECM genes in fibrotic lungs.

a Violin plots showing markers of cluster 8. b Expression levels of Col1a1 and Cthrc1 on UMAP plots. Cells from untreated lungs, bleomycin-treated lungs, or cluster 8 are shown separately in each column. c PLISH images for Cthrc1 (white) or Col1a1 (magenta) 14 days after bleomycin treatment. Col-GFP is shown in green. DAPI signal is shown in blue. H&E staining of serial section is also shown. Arrowheads indicate co-localization of Cthrc1 and Col1a1 in GFP⁺ cells. Scale bar, 50 μm. Images are representative from three experiments (n ≥ 2). d RNA velocity analysis overlaid on UMAP plot. Direction indicates transition towards the estimated future state of a cell. e UMAP plot showing only clusters 0, 1, 2, and 8. f Pseudotime trajectory analysis by Monocle 2. g Expression levels of Pdgfra and Col1a1 were analyzed along the pseudotime and fitted cubic smoothing spline curves are shown. The y-axis is scaled gene expression level. h Pdgfra and Col1a1 expression levels on UMAP plots.

Fig. 5. scRNA-seq identifies CTHRC1⁺ pathologic fibroblasts in human fibrotic lungs.

a UMAP plot of all cells acquired by scRNA-seq of human samples. Cells were obtained from three normal lungs (green), three IPF lungs (red), and two scleroderma lungs (blue). b COL1A1 expression on UMAP plot. See Supplementary Fig. 5a for lineage identification. DC, dendritic cells; Mac, macrophage. c UMAP plot of COL1A1⁺ cells. d Unsupervised clustering identified 7 clusters. e Expression levels of representative genes are shown on UMAP plots. f Expression of COL1A1, CTHRC1, and ACTA2 are shown on UMAP plots. Cells from normal, IPF, and scleroderma lungs are shown separately in each column. g Expression levels of selected ECM genes on UMAP plots.

Fig. 6. CTHRC1⁺ pathologic fibroblasts are localized within fibroblastic foci in IPF.

a Representative images of in situ hybridization (ISH) and immunohistochemistry (IHC) for CTHRC1 in the sections from three IPF patients. Scale bars, 100 μm. b CTHRC1 and α-SMA antibody staining in sequential sections of IPF lungs. Fibroblastic foci are outlined by dotted lines. Areas inside blue squares are magnified in lower panels. Arrows indicate α-SMA⁺ cells outside CTHRC1⁺ areas. Images are representative from four IPF patients. Scale bars, 100 μm.

Fig. 7. FACS purification of fibroblast subpopulations and their ability to colonize fibrosing lungs in adoptive transfer.

a Expression levels of Ly6a and Cd9 on UMAP plots from murine data. b Violin plots showing expressions of surface markers on collagen-producing populations. c Gating strategy to purify Mcam⁺ cells, adventitial fibroblasts, alveolar fibroblasts, and peribronchial fibroblasts from untreated lungs. d qPCR analysis of purified cells from untreated lungs. n = 5 mice. e Schematic for the adoptive transfer experiment. f The number of GFP⁺ cells in host lungs were analyzed. n = 5 mice. g qPCR analysis of purified GFP⁺ cells from the host lungs. Cells purified from untreated lungs were used as control. n = 5 mice. c, d, f, g Data are representative from two experiments. Source data are provided as a Source Data file. d, f, g Data are means ± SEM.

Fig. 8. Purified Cthrc1+ pathologic fibroblasts showed high migration and invasion capacity.

a Gating strategy for purifying collagen-producing populations from bleomycin-treated lungs. b qPCR analysis of purified cells from bleomycin-treated lungs (day 10). n = 4 mice. c Representative images from gap migration assay with cells purified from bleomycin-treated lungs (day 10). Broken lines show initial cell-free zones. Col-GFP is shown in green. Scale bars, 500 μm. d Quantification of migration assay. n = 4 mice. ***p < 0.001, two-way analysis of variance followed by the Tukey–Kramer post-test. e The number of GFP⁺ cells from the host lungs, which received purified cells from bleomycin-treated lungs, were analyzed. n = 5 (adventitial) or 6 (cluster 8 and alveolar) mice. **p < 0.01, ***p < 0.001, one-way analysis of variance followed by the Tukey–Kramer post-test. f Whole lung imaging of host lungs. Col-GFP (donor cells) is shown in green. Autofluorescence in RFP channel is shown in white to visualize host lung cells. Images were maximum projection of z-stack images. Scale bars, 1 mm. b–f Data are representative from two experiments. b, d, e Data are means ± SEM. Source data are provided as a Source Data file.