Alveolar fibroblast lineage orchestrates lung inflammation and fibrosis

Abstract

Fibroblasts are present throughout the body and function to maintain tissue homeostasis. Recent studies have identified diverse fibroblast subsets in healthy and injured tissues^1,2, but the origins and functional roles of injury-induced fibroblast lineages remain unclear. Here we show that lung-specialized alveolar fibroblasts take on multiple molecular states with distinct roles in facilitating responses to fibrotic lung injury. We generate a genetic tool that uniquely targets alveolar fibroblasts to demonstrate their role in providing niches for alveolar stem cells in homeostasis and show that loss of this niche leads to exaggerated responses to acute lung injury. Lineage tracing identifies alveolar fibroblasts as the dominant origin for multiple emergent fibroblast subsets sequentially driven by inflammatory and pro-fibrotic signals after injury. We identify similar, but not completely identical, fibroblast lineages in human pulmonary fibrosis. TGFβ negatively regulates an inflammatory fibroblast subset that emerges early after injury and stimulates the differentiation into fibrotic fibroblasts to elicit intra-alveolar fibrosis. Blocking the induction of fibrotic fibroblasts in the alveolar fibroblast lineage abrogates fibrosis but exacerbates lung inflammation. These results demonstrate the multifaceted roles of the alveolar fibroblast lineage in maintaining normal alveolar homeostasis and orchestrating sequential responses to lung injury.

Extended Data Fig. 1.. Scube2-CreER specifically labels alveolar fibroblasts, which provide a niche to support AT2 cells

(a) UMAP plots of all lung cells from scRNA-seq data of Tsukui et al. 2020, for cell types (left) and Scube2 expression (right). (b) UMAP plots of Col1a1+ cells from scRNA-seq data of Tsukui et al. 2020, for cell types (left) and Scube2 expression (right). (c-f) Gating strategy to evaluate the specificity of tdTomato+ cells for lineage (CD31, CD45, EpCAM, Mcam, Ter119), Sca1, and CD9. (g) Gating strategy for alveolar fibroblasts. (h) Flow cytometric quantification for percent tdTomato+ of alveolar fibroblasts. n = 3 mice. (i) Pi16 staining of a lung section from Scube2-CreER/Rosa26-tdTomato Col-GFP mouse. tdTomato is shown in magenta. Col-GFP is shown in green. Pi16 is shown in grey. DAPI is shown in blue. aw, airway. (j) Three representative planes from z-stack images with z-positions are shown below the images. tdTomato is shown in red. proSP-C is shown in cyan. Asterisks indicate the same proSP-C+ cell. Arrows point to projections extending from alveolar fibroblasts. Scale bars, 200 μm (i) or 5 μm (j). Data are mean ± SEM. Data are representative of at least two independent experiments.

Extended Data Fig. 2.. Scube2-CreER-labeled alveolar fibroblasts provide a niche to support AT2 cells

(a, b) Gating strategy for alveolar fibroblast-ablation experiments with Scube2-CreER/Rosa26-DTA mice. (a) Gating strategy for EpCAM+ subpopulations. (b) Gating strategy for fibroblast subsets. Lineage markers include CD31, CD45, EpCAM, and Ter119. (c) Flow cytometric cell count for each population, normalized to means of vehicle groups. n = 4 (tamoxifen) or 5 (vehicle) mice. (d) Representative images of H&E staining of lung sections from alveolar fibroblast-ablation experiments. Scale bars, 200 μm. (e) Quantification of mean linear intercept of alveolar regions. n = 4 (tamoxifen) or 5 (vehicle) mice. (f) qPCR analysis of whole lung cells 6 days after bleomycin treatment. n = 5 mice. (g) Gating strategy for CD4 and γδ T cells. (h) Gating strategy for ILCs. (i) Flow cytometric quantification in bleomycin-treated lungs. n = 4 (Rosa26-DTA/DTA) or 7 (Rosa26-WT/WT) mice. (j) Gating strategy for IL-17a+ cells. Data are mean ± SEM. Data are representative of at least two experiments. Statistical analysis was performed using unpaired two-tailed t-test followed by Holm–Sidak’s multiple-comparisons adjustment (f) or two-way ANOVA followed by Sidak’s multiple comparison test (i).

Extended Data Fig. 3.. Longitudinal scRNA-seq reveals multiple fibroblast subsets that emerge after lung injury

(a) Gating strategy for purifying lineage (CD31, CD45, EpCAM, Ter119)− mesenchymal cells for scRNA-seq. (b) UMAP plots for cells obtained before (day 0) or at various time points after bleomycin treatment. (c) Dot plots showing the top differentially expressed genes for each cluster. (d) GO enrichment analysis by DAVID for differentially expressed genes of inflammatory fibroblasts. (e) GO enrichment analysis by DAVID for differentially expressed genes of stress-activated fibroblasts. (f) GO over-representation analysis with the Fisher test for all clusters.

Extended Data Fig. 4.. Lineage tracing by scRNA-seq reveals alveolar fibroblasts as the origin of multiple pathologic fibroblast subsets

(a) UMAP plot with tdTomato expression. (b) Violin plot for tdTomato shows peaks for tdTomato^low or tdTomato^high cells. The threshold for tdTomato+ cells was defined as an expression level > 3.5. (c) UMAP plots with tdTomato expression split by biological replicates. (d) Plot showing percent tdTomato+ of alveolar fibroblasts (x-axis) versus percent tdTomato+ of fibrotic fibroblasts (y-axis) for each biological replicate.

Extended Data Fig. 5.. Scube2-CreER-labeled alveolar fibroblasts differentiate into fibrotic or inflammatory fibroblasts after lung injury

(a) Maximum projection of whole lung imaging for untreated or bleomycin day 14 Scube2-CreER/Rosa26-tdTomato mice. (b) Representative optical sections from whole lung imaging. tdTomato is shown in magenta. Autofluorescence is shown in grey (a, b). (c) Flow cytometry plots showing the increase of CD9+ cells among Scube2-CreER-labeled (tdTomato+) cells on day 21 after bleomycin treatment. (d) Flow cytometric quantification of percent CD9+ of tdTomato+ cells. n = 4 mice. Statistical analysis was performed using unpaired two-tailed t-test. (e) qPCR analysis of sorted cells from Scube2-CreER/Rosa26-tdTomato mice. All lineage (CD31, CD45, EpCAM, Ter119)− tdTomato+ cells (untreated or bleomycin day 21) or lineage− tdTomato+ CD9+ cells (bleomycin day 21) were sorted. The y-axis is the relative expression level to the housekeeping gene Rps3. n = 3 mice. (f) Saa3 staining in sections from Scube2-CreER/Rosa26-tdTomato mice 7 days after bleomycin treatment. Arrows indicate tdTomato and Saa3 double-positive cells. (g) Quantification of percent Saa3+ of tdTomato+ cells. n = 3 mice. tdTomato is shown in red. Saa3 is shown in cyan. DAPI is shown in blue. Collagen 4 is shown in grey. Data are mean ± SEM. Data are representative of at least two independent experiments. Scale bars, 1mm (a, b), 20 μm (f).

Extended Data Fig. 6.. Alveolar fibroblasts up-regulated activation markers and formed silicotic nodules in the silicosis model

(a) Time course of tamoxifen and silica treatment. (b) Representative lung sections of saline or silica-treated mice. (c, d) Sirius red staining of silica-treated lung section imaged as bright field (c) or polarized light (d). (e) Fluorescence imaging of a sequential section of (c, d). (f) Representative images of silicotic nodules. (g) Histological quantification of % tdTomato+ cells of Pdgfra+ cells inside silicotic nodules. n = 5 mice. (h) Flow cytometric analysis of Pdgfra and CD9 on tdTomato+ cells from saline or silica-treated lungs. (i) Flow cytometric quantification of %CD9+ Pdgfra-low cells of tdTomato+ cells. n = 5 mice. (j, k) qPCR analysis of purified populations. The y-axis is the relative expression level to the housekeeping gene Rps3. n = 5 mice. Scale bars, 200 μm (b), 500 μm (c-e), 50 μm (f). Data are mean ± SEM. Data are representative of at least two independent experiments. Statistical analysis was performed using unpaired two-tailed t-test (i, k) or Tukey’s multiple comparisons test after one-way ANOVA (j).

Extended Data Fig. 7.. TGF-β1 antagonizes inflammatory marker expression induced by IL-1β and induces fibrotic markers

(a) Pseudotime analysis of tdTomato+ clusters suggests that both stress-activated fibroblasts and fibrotic fibroblasts can emerge from inflammatory fibroblasts. (b) UMAP plots of re-clustered alveolar, inflammatory, and fibrotic lineage split by days after bleomycin treatment. (c) Expression of selected markers on UMAP plots. (d) Schematic of sequential cytokine stimulations for primary alveolar fibroblasts. (e) qPCR analysis after sequential cytokine stimulations. Group names indicate (first stimulation) → (second stimulation). DMEM means medium-only control. The y-axis is the relative expression level to the housekeeping gene Rps3. n = 3 wells. Data are mean ± SEM. Data are representative of two independent experiments.

Extended Data Fig. 8.. Re-analysis of our publicly available scRNA-seq data from human pulmonary fibrosis reveals inflammatory and fibrotic clusters

(a) UMAP plots of re-clustered pathologic and alveolar clusters shown for individual patients or donors (control). (b) Dot plot showing top differentially expressed genes for each cluster. (c) GO enrichment analysis by DAVID for differentially expressed genes in inflammatory fibroblast 1 cluster. (d) GO enrichment analysis by DAVID for differentially expressed genes in inflammatory fibroblast 2 cluster. (e) Heat map showing Spearman’s correlation coefficients of average gene expression from mouse and human emergent clusters. (f) GO over-representation analysis with the Fisher test for all clusters.

Extended Data Fig. 9.. Combined analysis of publicly available scRNA-seq data sets from human pulmonary fibrosis from 3 groups reveals conserved inflammatory and fibrotic clusters

(a) UMAP plot of mesenchymal cells from Adams et al. (b) Expression levels of selected genes on UMAP plot of Adams et al. mesenchymal cells show alveolar and pathologic fibroblast clusters. (c) UMAP plot of mesenchymal cells from Habermann et al. (d) Expression levels of selected genes on UMAP plot of Habermann et al. mesenchymal cells show that “Myofibroblasts” cluster contains alveolar and pathologic fibroblasts. (e) UMAP plot after combining alveolar and pathologic fibroblasts from Adams et al., Habermann et al., and Tsukui et al. (f) UMAP plots of combined data split by original data set. (g) UMAP plots shown for each data set and colored by samples. (h) Expression levels of selected markers for fibrotic (COL1A1, CTHRC1), inflammatory1 (SFRP2, CXCL12), inflammatory2 (SFRP4, CXCL14) and alveolar fibroblasts (NPNT, TCF21) on UMAP plots.

Extended Data Fig. 10.. Cthrc1-CreER mouse demonstrates the pro-fibrotic function of Cthrc1+ fibroblasts

(a) Gating strategy for cell size, singlet, and live cells. (b) Flow cytometry plots show an increase in lineage− tdTomato+ cells on day 14 after bleomycin treatment. (c) Flow cytometric cell count of lineage (CD31, CD45, EpCAM, Ter119)− tdTomato+ cells on day 14. n = 3 (Saline-Vehicle) or 5 (Saline-Tamoxifen, Bleo-Tamoxifen) mice. (d) Flow cytometry plots show CD9 expression on tdTomato+ lineage− cells increases between day 14 and 21. (e) Flow cytometric quantification of percent CD9+ of tdTomato+ lineage− cells. n = 5 mice. (f) Mean fluorescence intensity (MFI) of CD9 on tdTomato+ lineage− cells. n = 5 mice. (g) Saa3 staining in sections from Cthrc1-CreER/Rosa26-tdTomato mice 14 days after bleomycin treatment with tamoxifen injected on days 8 –12. tdTomato is shown in red. Saa3 is shown in cyan. Collagen 4 is shown in grey. DAPI is shown in blue. (h) Schematic for localization of inflammatory and fibrotic fibroblasts. (i) Representative images of sequential lung sections from Cthrc1-CreER/Rosa26-tdTomato mice stained for collagen 1 or Pi16 (shown in grey). tdTomato is shown in red. DAPI is shown in blue. (j, k) Image quantifications of the mean distance to collagen 1 or Pi16 from tdTomato or DAPI on the sections. n = 4 mice. (l) Representative images of lung sections stained for collagen 1 (shown in grey) from ablation experiments using Cthrc1-CreER/Rosa26-DTA mice. DAPI is shown in blue. (m) Image quantification of % collagen 1+ area on the sections. n = 3 (saline), 8 (bleomycin, Rosa26-WT/WT), or 10 (bleomycin, Rosa26-DTA/DTA) mice. Data are mean ± SEM. Data are representative of at least two independent experiments. Scale bars, 20 μm (g), 1 mm (i, l). Statistical analysis was performed using unpaired two-tailed t-test (j, k) or two-way ANOVA followed by Sidak’s multiple comparison test (m).

Extended Data Fig. 11.. Tgfbr2 conditional knockout in alveolar fibroblasts abrogates fibrosis but exacerbates inflammation

(a) Representative images of whole sections stained for collagen 1 (magenta) and collagen 4 (green). Arrows indicate regions of intra-alveolar collagen 1. (b) Survival after bleomycin treatment. n = 15 mice. (c) qPCR of purified tdTomato+ cells from saline-treated mice showed no difference for fibrotic and inflammatory genes between control and Tgfbr2 cKO. (d) Flow cytometric counting of myeloid populations in BAL from saline-treated mice showed no difference between control and Tgfbr2 cKO. n = 5 mice (c, b). (e) Gating strategy for myeloid populations in BAL. (f) Representative images of sections from bleomycin-treated lungs stained for Saa3 (magenta) and CD68 (green). tdTomato is shown in blue. Magnified single-channel images of yellow rectangles are shown on the right. (g, h) Image quantification of Saa3 (g), or CD68 (h). n = 5 (control) or 6 (Tgfbr2 fl/fl) mice. Data are representative of at least two independent experiments. Data are mean ± SEM. Statistical analysis was performed using unpaired two-tailed t-test (g, h). Scale bars, 1 mm (a), 100 μm (f, wide), or 20 μm (f, magnified).

Fig. 1.. Scube2-CreER specifically labels alveolar fibroblasts and ablation of these cells leads to the loss of alveolar stem cell niches.

(a) Schematic of Scube2-CreER mouse generation and experiments. (b) Flow cytometric quantification of tdTomato+ cells in each fraction. Pdgfra+, CD9+, or Sca1+ fractions were pre-gated on lineage− cells. n = 3 mice. (c) Flow cytometric analysis of lineage− Sca1− cells. (d) Maximum projection of whole lung imaging. (e) Maximum projection of 32 z-stack images with step size 0.9 μm. aw, airway. bv, blood vessel. al, alveoli. cuff, cuff space. (f) 28 z-stack images with step size 0.5 μm shown as maximum projection (left), color-coded projection to the depth from the top (middle) or the bottom (right). (g) Time course of tamoxifen treatment. (h) Representative lung sections with proSP-C staining. (i) Histological quantification of proSP-C+ cells. n = 5 mice. (j, k) Flow cytometric counting of alveolar fibroblasts (j) or AT2 cells (k). (l) qPCR analysis of all lung cells. n = 4 (vehicle) or 5 (Tamoxifen) mice (j, k). (m) Time course of tamoxifen and bleomycin treatment. (n) IgM in BAL measured by ELISA. n = 4 (saline) or 5 (bleomycin) mice. (o) Flow cytometric quantification of neutrophils in bleomycin-treated lungs. (p) qPCR for Il17a in purified populations. (q) Flow cytometric quantification of IL-17a+ cells in bleomycin-treated lungs. (o-q) n = 4 (Rosa26-DTA/DTA) or 7 (Rosa26-WT/WT) mice. (r) Percent body weight change after bleomycin treatment. (s) Survival after bleomycin treatment. n = 10 mice (r, s). Scale bars, 1 mm (d), 100 μm (e), 10 μm (f), 50 μm (h). Data are mean ± SEM. Data are representative of at least two independent experiments. Statistical analysis was performed using unpaired two-tailed t-test (i, j, k, o), unpaired two-tailed t-test followed by Holm–Sidak’s multiple-comparisons adjustment (l), two-tailed Mann-Whitney test (n), or two-way analysis of variance (ANOVA) followed by Sidak’s multiple comparison test (p, q).

Fig. 2.. Lineage tracing by scRNA-seq reveals alveolar fibroblasts as the origin of multiple emergent fibroblast subsets

(a) Schematic of scRNA-seq experiment design. (b) UMAP plot of the scRNA-seq data. Clusters were shown with different colors. (c) Frequency of each cluster on different time points. (d) Dot plot showing representative markers for each cluster. (e) Dot plot showing markers for subsets that emerge after injury. (f) UMAP plot showing tdTomato+ and tdTomato-negative cells. (g) Percent tdTomato+ of each subset that was present in normal lungs. (h) Percent tdTomato+ of each subset that emerged after injury. n = 3 mice (g, h). Data are mean ± SEM.

Fig. 3.. Alveolar fibroblasts sequentially differentiate into inflammatory and fibrotic fibroblasts in mouse and human pulmonary fibrosis

(a) UMAP plot of scRNA-seq data subsetted into alveolar, inflammatory, and fibrotic fibroblasts. (b) UMAP plot overlaid with pseudotime. (c) Scaled expression of representative markers in pseudospace. (d) Heat map with cells arranged in pseudotemporal order showing changes in representative markers. Cluster annotations for each cell were shown above the heat map. (e, f) in vitro cytokine stimulation of primary alveolar fibroblasts. (e) Schematic of the experiment. (f) qPCR analysis for representative genes. n = 3 wells. Data are mean ± SEM. Data are representative of three experiments. (g) Schematic of re-analysis of our previous human scRNA-seq data. (h) UMAP plot after subsetting and re-clustering alveolar and pathologic clusters. (i) UMAP plots for cells from control (n = 3), scleroderma (n = 2) or IPF (n = 3) lungs. (j) Dot plot for representative markers for each subset. (k) UMAP plot overlaid with pseudotime. (l) Schematic of in situ hybridization experiment. (m) Dot plot showing ITGA8 expression can distinguish inflammatory fibroblasts from adventitial fibroblasts. (n) Joint density plots showing two markers that can highlight fibroblast subsets. (o) In situ hybridization on sequential sections from an IPF lung. Yellow squares in the left images are magnified in the right images. Arrows indicate cells expressing the markers for each subset. Yellow dashed circles indicate a fibroblastic focus. Scale bars, 50 μm (merge), 10 μm (magnified). Images are representative of 3 IPF patients. (p) Schematic of localization for fibrotic and two inflammatory fibroblast subsets.

Fig. 4.. Cthrc1-CreER mouse demonstrates the pro-fibrotic function of Cthrc1+ fibroblasts

(a) Schematic of Cthrc1-CreER mouse generation. (b) Time course of bleomycin and tamoxifen treatment for day 14 analysis. (c) Maximum projection of whole lung imaging. (d) qPCR analysis of all lung cells, lineage− cells, and tdTomato+ cells. n = 3 mice. (e) Collagen 1 and collagen 4 staining on day 14. Scube2-CreER mice were treated with tamoxifen as shown in Fig. 2a. Arrowheads indicate fibroblasts inside the basal lamina. Arrows indicate fibroblasts outside of the basal lamina. (f) Image quantification of mean distances between tdTomato and collagen 4. (g) Image quantification of tdTomato+ cells that are directly associated with intra-alveolar collagen 1. n= 4 mice (f, g). (h) Time course of bleomycin and tamoxifen treatment for day 28 analysis. (i, j) Ablation efficiency assessed by lineage− tdTomato+ cell number in left lobes (i) or by whole lung qPCR for Cthrc1 (j) from Cthrc1-CreER+/− Rosa26-tdTomato/WT (n = 11 mice) or Cthrc1-CreER+/− Rosa26-tdTomato/DTA (n = 12 mice). (k) Hydroxyproline assay on day 28 of Cthrc1-CreER+/− Rosa26-WT/WT (n = 16 mice for saline, n = 26 mice for bleomycin) or Cthrc1-CreER+/− Rosa26-DTA/DTA (n = 17 mice for saline, n = 31 mice for bleomycin). Scale bars, 1 mm (c), 20 μm (e). Data are representative of at least two independent experiments except (k), which is a pool from two independent experiments. Data are mean ± SEM. Statistical analysis was performed using Tukey’s multiple comparisons test after one-way ANOVA (f), unpaired two-tailed t-test (g), or two-tailed Mann-Whitney test (i, j, k).

Fig. 5.. Tgfbr2 conditional knockout in alveolar fibroblasts abrogates fibrosis but exacerbates inflammation

(a) Schematic of Tgfbr2 conditional knockout experiments. (b) Sections after bleomycin treatment were stained for collagen 1 and collagen 4. Arrows indicate intra-alveolar collagen 1. Scale bars, 100 μm. (c) Quantification of Collagen 1+ area of whole sections. n = 3 (saline) or 6 (bleomycin) mice. (d) Hydroxyproline assays of left lobes. n = 5 (saline), 15 (bleomycin, control), or 13 (bleomycin, Tgfbr2 fl/fl) mice. (e) Percent body weight change after bleomycin. n = 15 (bleomycin, control) or 13 (bleomycin, Tgfbr2 fl/fl) mice. (f) qPCR of purified tdTomato+ cells after bleomycin showing the relative expression to the control group. n = 13 (bleomycin, control) or 11 (bleomycin, Tgfbr2 fl/fl) mice. (g) ELISA of BAL for IgM. n = 5 (saline), 13 (bleomycin, control), or 11 (bleomycin, Tgfbr2 fl/fl) mice. (h) Flow cytometric counting of myeloid populations in BAL. n = 13 (bleomycin, control), or 10 (bleomycin, Tgfbr2 fl/fl) mice. (i) Schematic of sequential differentiation of alveolar fibroblast lineage after injury. Data are representative of at least two independent experiments. Data are mean ± SEM. Statistical analysis was performed using two-tailed Mann-Whitney test (c, d, g, h) or unpaired two-tailed t-test (f). P-values were adjusted using Holm–Sidak’s multiple-comparisons adjustment (f, h).