A single-cell spatial chart of the airway wall reveals proinflammatory cellular ecosystems and their interactions in health and asthma

Abstract

Determining spatial location of cells within tissues gives vital insight into the interactions between resident and inflammatory cells and is a critical factor for uncoupling the mechanisms driving disease. Here, we apply single-cell spatial transcriptomics to reveal the airway wall landscape in health and during asthma. We identified proinflammatory cellular ecosystems that exist within discrete spatial niches in healthy and asthma samples. These cellular hubs are characterized by a high level of chemokine and alarmin expression, along with unique combinations of stromal cells. Mechanistically, we demonstrated that receptors, such as ACKR1, retain immune mediators locally, while amphiregulin-expressing mast cells are prominent within these proinflammatory hubs. Despite anti-inflammatory treatments, the asthma airway mucosa exhibited a distinct remodeling program within these cellular ecosystems, marked by increased proximity between key cell types. This study provides an unprecedented view of the topography of the airway wall, revealing distinct, specific ecosystems within spatial niches to target for therapeutic intervention.

Fig. 1. Single-cell spatial transcriptomics to analyze lung wall heterogeneity and cellular interaction.

a, Schematic showing the site of lung endobronchial biopsies and Xenium and GeoMX slide placement. b, Overview of tissue sections on a Xenium slide and zoom-in images of one section (white box) with corresponding H&E staining, Xenium cell segmentation, total transcripts detected and superimposed DAPI, cell segmentation and transcripts in one area (data are representative of two independent experiments). c, UMAP projection of all 60,567 cells captured using Xenium with the indicated cluster names identified. d, Cell fraction per donor (cohort 1; see Supplementary Table 1 and Supplementary Data 1 for participant information). e, Images of endobronchial biopsies stained for DNA (Syto83, green), vimentin (purple), CD45 (blue) and α-SMA (yellow) showing the selected ROIs (boxed regions, representative of four independent experiments); scale bars 100 µm. f, UMAP of all 43 ROIs according to their localization and pathology (n = 4 donors in each group, cohort 3; see Supplemental Table 1 for participant information). g, Cytokines, chemokines and proliferation genes shown as a transcript density map. h, Chemokines, cytokines and proliferation genes in different ROIs and pathology status (n = 15 (epithelium), 19 (subepithelium) and 13 (smooth muscle) ROIs from n = 4 donors per group, cohort 3); data are shown as mean ± s.e.m. i, Schematic depicting nearest neighbor analysis and a Xenium image showing the nearest endothelial cell (yellow dots) from each basal cell (green dots). Lines represent shortest distance (data are representative of two independent experiments). j, Repartition of cells and distance to BC2 for each cluster and average neighboring cells within 50 µm; SMCs, smooth muscle cells; Epi, epithelium; Muc, mucous glands. Data in h were analyzed by one-way analysis of variance (ANOVA), followed by a Tukey’s post hoc test. Illustrations were created with Adobe Illustrator and Biorender.com.

Fig. 2. Lung wall is characterized by discrete proinflammatory hubs.

a, Representative H&E-stained section and matching Xenium image of an asthma biopsy. Transcripts are visualized as a density map and regions used for epithelial and mucous gland niche analysis. All regions are shown in Extended Data Fig. 2. Data are representative of two independent experiments. b,c, UMAP of cells captured in epithelial (b) and mucous gland (c) niches with the indicated names of clusters (n = 8 donors, cohort 1). d, Representative Xenium image showing inflammatory hubs with transcripts for chemokines (red dots) and alarmins (orange; data are representative of two independent experiments). e,f, Gene dot plot for alarmins (that is, TSLP and IL33) and chemokine expression in stromal cell clusters for epithelial (e) and mucous gland (f) niches (n = 4 donors per group, cohort 1). g, Heat map projection of alarmin expression in epithelial and subepithelial ROIs (four to nine ROIs per group from n = 4 donors per group, cohort 3). For the gene dot plot, the size of the dot reflects the frequency of cells expressing at least one transcript, and color indicates the level of expression. Columns alternate between healthy (H) and asthma (A) samples. Scales are provided for each plot. Red boxes indicate genes with a P value of <0.05 compared to other clusters, and black boxes indicate genes with a P value of <0.05 between healthy tissue and tissue from individuals with asthma. Data were analyzed by two-sided Mann–Whitney t-test (e and f) and two-way ANOVA, followed by a Dunnett’s post hoc test (g, healthy vs. mild P value = 0.0002, healthy vs. severe P value = l0.0047).

Fig. 3. Basal cells, goblet cells and endothelial cells are key regulators of the epithelial niche.

a, Representative H&E image showing the analyzed area (left) and Xenium images of chemokines (right; red dots) and IL33 (orange) transcript expression by basal cells (light green, BC2; dark green, BC1), goblet cells (light blue) and endothelial cells (dark brown, EnC1; purple, EnC2); data are representative of two independent experiments. b, Gene dot plot for the top 20 expressed genes in BC1s/BC2s, goblet cells and EnC1s/EnC2s (n = 4 donors per group, cohort 1). c, Repartition of neighbors along distance for BC1s (top), goblet cells (middle) or EnC2s (bottom; n = 4 donors per group, cohort 1). d, Shortest cell-to-cell distance between BC1s (top), goblet cells (middle) or EnC2s (bottom) and other clusters (n = 4 donors per group, cohort 1). e, BC1, goblet cell and EnC2 enrichment scores (epithelium ROIs n = 4 (healthy), 5 (mild) and 7 (severe); subepithelium ROIs n = 6 (healthy), 4 (mild) and 5 (severe); smooth muscle ROIs n = 4 (healthy), 4 (mild) and 9 (severe) from n = 4 donors per group, cohort 3). Columns alternate between healthy (H) and asthma (A) samples. Scales are provided for each plot. Red boxes indicate genes with a P value of <0.05 compared to other clusters, and black boxes indicate genes with a P value of <0.05 between healthy and asthma samples. Data were analyzed by two-sided Mann–Whitney t-test (b and d) and two-way ANOVA, followed by a Dunnett’s post hoc test (e); AU, arbitrary units.

Fig. 4. Mucous glands show specialized populations of cells with dysregulated activity in asthma.

a, Representative H&E and Xenium images showing the area analyzed and chemokine (red dots) and IL33 (orange) transcript expression by fibroblasts (dark red) and endothelial cells (light brown, EnC1s; purple, EnC2s). Data are representative of two independent experiments. b, Gene dot plot for the top 20 expressed genes in EnC1s/EnC2s (top) and fibroblasts (bottom; n = 4 donors per group, cohort 1). c, Repartition of neighbors along distance for EnC2s and fibroblasts (n = 4 donors per group, cohort 1). d, H&E and Xenium images of the area analyzed and goblet cell gene (blue dots), serous cells gene (gray dots), MUC5B (red dots) and endothelial cell gene (brown) transcript expression. Data are representative of two independent experiments; scale bar, 100 μm (H&E images). e, Gene dot plot of the 20 most expressed genes by goblet cells from mucous glands (n = 4 donors per group, cohort 1). f, MUC5B and MUC5AC gene counts (n = 9 (healthy), 8 (mild) and 15 (severe) ROIs per group from n = 4 donors per group, cohort 3). Data are shown as mean ± s.e.m. g, Repartition of neighbors along distance for goblet cells (n = 4 donors per group, cohort 1). h, Gene dot plot of the 20 most expressed genes by serous cells from mucous glands (n = 4 donors per group, cohort 1). i, Repartition of neighbors along distance for serous cells (n = 4 donors per group, cohort 1). Columns alternate between samples from healthy individuals and those with asthma. Scales are provided for each plot. Red boxes indicate genes with a P value of <0.05 compared to other clusters, and black boxes indicate genes with a P value of <0.05 between healthy and asthma samples. Data were analyzed by two-sided Mann–Whitney t-test (b, e and h) and one-way ANOVA followed by a Tukey’s post hoc test (f).

Fig. 5. Mast cells are regulators of proinflammatory ecosystems.

a, Representative precision-cut lung slices (PCLSs) of human parenchyma showing widespread distribution of mast cells (MC). b, Number of mast cells normalized to the total volume of the image per donor (n = 8 (donors 1, 2 and 3), 11 (donor 4), 9 (donor 5) and 12 (donor 6) regions analyzed from n = 6 donors). Data are shown as mean ± s.e.m. c, Representative Xenium image of chemokine genes (red dots), ACKR1 (purple), IL33 (orange) and mast cell genes (cyan dots). Data are representative of two independent experiments. d, Gene dot plot of the 14 and 15 most expressed genes by mast cells (n = 4 donors per group, cohort 1). e, Repartition of neighbors along distance for mast cells (n = 4 donors per group, cohort 1). f, Shortest cell-to-cell distance between mast cells and other clusters (n = 4 donors per group, cohort 1). Columns alternate between healthy and asthma samples. Scales are provided for each plot. Red boxes indicate genes with a P value of <0.05 compared to other clusters, and black boxes indicate genes with a P value of <0.05 between healthy individuals and those with asthma. Data were analyzed by two-sided Mann–Whitney t-test.

Fig. 6. Imatinib treatment disrupts cell transcriptional activity and localization in individuals with asthma.

a, Clinical trial information. b, Representative H&E image (left), DAPI image (middle) and cell segmentation (right). Data are representative of two independent experiments. c, UMAP of 36,591 cells captured with the indicated names of clusters. The image on the right shows data from donors treated with placebo (gray dots) or imatinib (red dots; cohort 2); Mono, monocytes; DC, dendritic cells; Macro, macrophages. d, Gene dot plot for alarmin and chemokine expression in stromal cell clusters (n = 3–4 donors per group, cohort 2). e, Representative images showing the area analyzed and chemokine (red dots) and IL33 (orange) transcript expression by basal cells (green cells) and endothelial cells (light brown, EnC1s; purple, EnC2s). Data are representative of two independent experiments; scale bars, 50 μm (H&E). f, Gene dot plot for the top 15 expressed genes in basal cells (left) and EnC1s/EnC2s (right; n = 3–4 donors per group, cohort 2). g, Repartition of neighbors along distance for basal cells or EnC2s (n = 3–4 donors per group, cohort 2). h, Representative image of goblet cell (green dots), CXCL6 (red dots) and MUC5B (yellow dots) transcript expression in goblet cells (blue cells). Data are representative of two independent experiments. i, Gene dot plot of the 15 most expressed genes by goblet cells (n = 3–4 donors per group, cohort 2). j, Repartition of neighbors along distance for goblet cells (n = 3–4 donors per group, cohort 2). k, Representative image of mast cells (dark brown) and endothelial cells (light brown, EnC1s; purple, EnC2s) and transcripts for mast cells (brown dots), CPA3 (red dots) and EnC (purple dots) genes. Data are representative of two independent experiments. l, Gene dot plot of the ten most expressed genes by mast cells (n = 3–4 donors per group, cohort 2). m, Repartition of neighbors along distance for mast cells (n = 3–4 donors per group, cohort 2). Columns alternate between samples from donors treated with placebo (P) or imatinib (I). Scales are provided for each plot. Red boxes indicate genes with a P value of <0.05 compared to other clusters, and black boxes indicate genes with a P value of <0.05 between placebo and imatinib. Data were analyzed by two-sided Mann–Whitney t-test.

Fig. 7. Spatial Drug2Cell analysis in lung biopsies.

a, Schematic depicting the principle of the analysis combining Xenium data and data from the ChEMBL database. b, The top five drugs per cluster with the strongest interaction and frequency of cells impacted in the epithelium and mucous gland areas of individuals with asthma (cohort 1). c, The top five drugs per cluster of placebo- and imatinib-treated epithelial ROIs (cohort 2). d, Spatial representation of cellular targets for tisotumab vedotin and hydrocortisone in two asthma biopsies. The color scale indicates strength of interaction. e, Spatial visualization of cellular targets for imatinib-, gefitinib- and hydrocortisone-treated biopsies. The color scale indicates interaction strength. For drug dot plots, the size of the dot reflects the frequency of cells within the cluster impacted by the drug, and color indicates the interaction strength. Illustrations created with Adobe Illustrator and Biorender.com.

Extended Data Fig. 1. Identification of clusters generated by Xenium and projection on lung cell atlas.

a, Gene dot plot showing the top 2 genes expressed per cluster across all donors, size and color indicate level of expression (cohort #1). b, UMAP projection of lung cell atlas and expression level of the top 5 genes expressed by each cluster. Data available at (https://5locationslung.cellgeni.sanger.ac.uk/all). c, UMAP projection showing each donor (healthy and asthma). d, Corresponding H&E staining and Xenium identified clusters at the epithelium showing the cellular diversity identified and spatial organisation. Scale bars, 10 µm (representative of 2 independent experiments).

Extended Data Fig. 2. Region of interest selection across all healthy and asthma biopsies.

For each biopsy, H&E and Xenium images are shown, and regions used for epithelial and mucous gland downstream analysis are represented (cohort #1). Scale bars, 100 µm (representative of 1 independent experiment).

Extended Data Fig. 3. Epithelial and mucous gland niches chemokine, proliferation, extracellular matrix, goblet cell.

a, Gene dot plot for chemokine expression in immune and undefined cell clusters (cohort #1). b, Heat map projection of chemokines expression in epithelial and sub-epithelial ROIs (4-9 ROIs per group from n = 4 donors per group, cohort #3). c, Gene dot plot for proliferation markers expression in stroma cell clusters (cohort #1). d, Gene dot plot for extracellular matrix gene expression in stroma cell clusters (cohort #1).

Extended Data Fig. 4. Spatial parameters of epithelial niche.

a, Cell proportion for each cluster of the epithelial niche in healthy (H) and asthma (A) samples (n = 4 donors per group, cohort #1). b, Average neighboring cell number between basal cell 1, goblet cell and endothelial cell 2 and each cluster (n = 4 donors per group, cohort #1). Mean ± SEM.

Extended Data Fig. 5. Mucous glands area spatial profile.

a, Cell proportion for each cluster of the epithelial niche in healthy (H) and asthma (A) samples (n = 4 donors per group, cohort #1). Mean ± SEM. b, APOD count in sub-epithelial ROIs (n = 5 (healthy), 4 (mild) and 8 (severe) ROIs per group from n = 4 donors per group, cohort #3). Mean ± SEM. c, Shortest distance between endothelial cell 2 and fibroblast to each cluster (pooled cells n = 4 donors per group, cohort #1). d, Average neighbouring cell number between endothelial cell 2 and fibroblast to each cluster (n = 4 donors per group, cohort #1). Mean ± SEM e, Shortest distance between goblet cell to each cluster (pooled cells n = 4 donors per group, cohort #1). f, Average neighbouring cell number between goblet cell to each cluster (n = 4 donors per group, cohort #1). Mean ± SEM. g, Shortest distance between serous cells to each cluster (pooled cells n = 4 donors per group, cohort #1). h, Average neighbouring cell number between serous cell to each cluster (n = 4 donors per group, cohort #1). Mean ± SEM. i, Serous cell enrichment score (epithelium ROIs n = 4 (healthy), 5 (mild) and 7 (severe); sub-epithelial ROIs n = 6 (healthy), 4 (mild) and 5 (severe); smooth muscle ROIs n = 4 (healthy), 4 (mild) and 9 (severe) from n = 4 donors per group, cohort #3). Two-sided Mann-Whitney t test.

Extended Data Fig. 6. Macrophage spatial profile and mast cell neighbourhood.

a, Gene dot plot for top 20 expressed genes in macrophage cells in the epithelial and mucous gland niche (pooled cells from n = 4 donors per group, cohort #1). b, Repartition of macrophage neighbours in the epithelial and mucous gland niche (pooled cells from n = 4 donors per group, cohort #1). c, Shortest distance between macrophage to each cluster in the epithelial and mucous gland niche (pooled cells n = 4 donors per group, cohort #1). d, Average neighbouring cell number between macrophage to each cluster in the epithelial and mucous gland niche (n = 4 donors per group, cohort #1). Mean ± SEM. e, Average neighbouring cell around mast cells in the epithelial and mucous gland niche (n = 4 donors per group, cohort #1). Mean ± SEM. Columns alternate between healthy (H), and asthma (A) samples. Scales are provided for each plot, red boxes indicate genes with p value < 0.05 compared to other clusters, black boxes indicate genes with p value < 0.05 between healthy and asthma. Two-sided Mann-Whitney t test.

Extended Data Fig. 7. Identification of clusters in placebo and imatinib treated samples and projection on lung cell atlas.

a, Gene dot plot showing the top 2 genes expressed per cluster across all placebo and imatinib samples, size and colour indicate level of expression (cohort #2). b, UMAP projection of lung cell atlas and expression level of the top 5 genes expressed by each cluster. c, UMAP projection showing each donor (placebo and imatinib). d, Cell fraction of each cluster per donor.

Extended Data Fig. 8. Region of interest selection in placebo and imatinib treated samples.

For each biopsy, H&E and Xenium images are shown, and regions used for epithelial downstream analysis are represented (cohort #2). Scale bars, 50 µm (representative of 1 independent experiment).

Extended Data Fig. 9. Epithelial niche of placebo and imatinib treated donors for alarmins, chemokines, proliferation, extracellular matrix, immune regulation and spatial organisation.

a, Gene dot plot for chemokine expression in immune cell clusters (cohort #2). b, Gene dot plot for proliferation markers expression in all clusters. c, Gene dot plot for extracellular matrix gene expression in stroma cell clusters. d, Gene dot plot for immune regulation gene expression in immune cell clusters. e, Shortest distance between basal cell and endothelial cell 2 to each cluster in the epithelial niche (pooled cells n = 3/4 donors per group, cohort #2). f, Average neighbouring cell number between basal cell and endothelial cell 2 to each cluster in the epithelial niche (n = 4 (placebo) and 3 (imatinib) donors per group, cohort #2). Mean ± SEM. g, Shortest distance and average neighbouring cell between goblet cell to each cluster in the epithelial niche (n = 4 (placebo) and 3 (imatinib) donors per group, cohort #2). Mean ± SEM. h, Shortest distance and average neighbouring cell between mast cell to each cluster in the epithelial niche (n = 4 (placebo) and 3 (imatinib) donors per group, cohort #2). Mean ± SEM. Columns alternate between placebo (P), and imatinib (I) samples. Scales are provided for each plot, red boxes indicate genes with p value < 0.05 compared to other clusters, black boxes indicate genes with p value < 0.05 between placebo and imatinib. Two-sided Mann-Whitney t test.

Extended Data Fig. 10. Spatial drug2cell analysis of lung biopsies in health and asthma with respiratory drugs.

a-b, Top 5 drugs per cluster with strongest interaction and frequency of cells impacted in the epithelium (a) and mucous glands (b) areas of healthy donors (cohort #1). c-d, Respiratory approved drugs interaction with cell clusters in the epithelium or mucous glands areas of healthy and asthma biopsies (c, cohort #1) or epithelium of asthma patients treated with placebo and imatinib (d, cohort #2). For drug dot plot, size of the dot reflects frequency of cell within the cluster impacted by the drug and colour indicates the interaction strength.