Transcriptional analysis of cystic fibrosis airways at single-cell resolution reveals altered epithelial cell states and composition - PubMed (original) (raw)

Multicenter Study

. 2021 May;27(5):806-814.

doi: 10.1038/s41591-021-01332-7. Epub 2021 May 6.

Justin Langerman # 2, Shan Sabri 2, Zareeb Lorenzana 3 4, Arunima Purkayastha 5, Guangzhu Zhang 1, Bindu Konda 1, Cody J Aros 5 6 7, Ben A Calvert 3, Aleks Szymaniak 8, Emily Wilson 8, Michael Mulligan 8, Priyanka Bhatt 8, Junjie Lu 8, Preethi Vijayaraj 5, Changfu Yao 1, David W Shia 5 6 7, Andrew J Lund 5 6, Edo Israely 1, Tammy M Rickabaugh 5, Jason Ernst 2 9 10, Martin Mense 8, Scott H Randell 11, Eszter K Vladar 12, Amy L Ryan 3 4, Kathrin Plath 13 14 15, John E Mahoney 16, Barry R Stripp 17, Brigitte N Gomperts 18 19 20 21

Affiliations

Multicenter Study

Transcriptional analysis of cystic fibrosis airways at single-cell resolution reveals altered epithelial cell states and composition

Gianni Carraro et al. Nat Med. 2021 May.

Abstract

Cystic fibrosis (CF) is a lethal autosomal recessive disorder that afflicts more than 70,000 people. People with CF experience multi-organ dysfunction resulting from aberrant electrolyte transport across polarized epithelia due to mutations in the cystic fibrosis transmembrane conductance regulator (CFTR) gene. CF-related lung disease is by far the most important determinant of morbidity and mortality. Here we report results from a multi-institute consortium in which single-cell transcriptomics were applied to define disease-related changes by comparing the proximal airway of CF donors (n = 19) undergoing transplantation for end-stage lung disease with that of previously healthy lung donors (n = 19). Disease-dependent differences observed include an overabundance of epithelial cells transitioning to specialized ciliated and secretory cell subsets coupled with an unexpected decrease in cycling basal cells. Our study yields a molecular atlas of the proximal airway epithelium that will provide insights for the development of new targeted therapies for CF airway disease.

PubMed Disclaimer

Conflict of interest statement

The authors have declared that no conflict of interest exists.

Figures

Extended Data Fig. 1.

Extended Data Fig. 1.. Cell subsets identified across institutions

(a) Visualization of the distribution of cells from the three institutions in the integrated embedding, showed by institution and (b) by samples of origin, visualized by UMAP. (c-f) Network distributions with differences between institutions, visualized by UMAP. (g) Major cell types identified using previously described markers, visualized by UMAP. (h) Ionocyte and NE cell subsets analyzed independently of other cell types, visualized by UMAP. (i) CO and CF sample contribution to cell populations and subsets, visualized by a stacked column chart. The ‘s’ indicates submucosal gland samples derived from matching ‘*’ CO and CF lungs. (j) Signatures of major cell types in 10706 ALI cells, created using previously published ALI gene lists, shown by violin plots. Overlaid are boxplots showing the quartiles, whiskers showing 1.5 times interquartile range, and dots showing outliers. (k) Distribution of major cell type proportions in freshly isolated and ALI datasets, for 38 and 5 independent biological samples respectively. Error bars show the standard error of the mean. (l) CFTR expression level per subtype, scaled over all cells.

Extended Data Fig. 2.

Extended Data Fig. 2.. Secretory cell networks.

(a) Heatmap showing the percent of normalized expression of the seven secretory networks across the secretory subset groups, divided by CO and CF. Each cell shows the average expression of all cells in that category, normalized by row. (b) Heatmap showing the percent of normalized expression within the secretory subset groups for the top five genes selected from each secretory network based on their pan-institutional identity as either the most Up or Down in CF within the given network. Up/Down and Network classification is shown by annotation to left of heatmap and in key at right. Note for Net S7, only three genes qualified as pan-institutional. (c) Bar plots showing the average expression of all genes in the remaining individual secretory networks per secretory subset group, in CO or CF cells.

Extended Data Fig. 3.

Extended Data Fig. 3.. Ciliated cell networks.

(a) Heatmap showing the percent of normalized expression of all ten ciliated networks across the ciliated subset groups, divided by CO and CF. Each cell shows the average expression of all cells in that category, normalized by row. (b) Heatmap showing the percent of normalized expression within the ciliated subset groups for the top five genes selected from each ciliated network based on their pan-institutional identity as either the most Up or Down in CF within the given network. Up/Down and Network classification is shown by annotation to left of heatmap and in key at right. (c) Bar plots showing the average expression of all genes in the remaining individual ciliated networks per ciliated subset group, in CO or CF cells.

Extended Data Fig. 4.

Extended Data Fig. 4.. Changes in CO and CF cilia biogenesis.

(a-j) For distinct categories of genes related to cilia biogenesis, the expansion of cilia gene expression is shown by violin plots and UMAP, indicating the changes in CO and CF for each cell subset. Overlaid are boxplots showing the quartiles, whiskers showing 1.5 times interquartile range, and dots showing outliers. Each Pair of CO and CF show the associated P value (Wilcox test).

Extended Data Fig. 5.

Extended Data Fig. 5.. Surface markers of basal cell subsets.

(a) Scaled expression of the top differentially expressed CD marker genes that inform specific basal cell subsets, visualized by heatmap. (b) FACS plots showing segregation of total basal cells (CD326+, CD271+, CD45−, CD31−) into basal subsets based on their preferential expression of CD66 and CD266, in freshly isolated CO (upper panel) and primary hBE culture (lower panel).

Extended Data Fig. 6.

Extended Data Fig. 6.. Basal cell networks.

(a) Heatmap showing the percent of normalized expression of the ten basal networks across the basal subset groups, divided by CO and CF. Each cell shows the average expression of all cells in that category, normalized by row. (b) Heatmap showing the percent of normalized expression within the basal subset groups for the top five genes selected from each basal network based on their pan-institutional identity as either the most Up or Down in CF within the given network. Up/Down and Network classification is shown by annotation to left of heatmap and in key at right (c) Bar plots showing the average expression of all genes in the remaining individual basal networks per basal subset group, in CO or CF cells.

Extended Data Fig. 7.

Extended Data Fig. 7.. Proliferative basal cells in CO and CF.

(a) Scoring of the proliferative state (generated using a gene signature from Basal2 subset, supp Table2), of primary hBE from CO and CF, visualized by UMAP. (b) Same scoring showed as violin plots with pairwise t-test comparison of CO and CF, *: p< 2.22e-16 (Wilcox test). Overlaid are boxplots showing the quartiles, whiskers showing 1.5 times interquartile range, and dots showing outliers. 3 clones were sampled for each condition.

Extended Data Fig. 8.

Extended Data Fig. 8.. Counting proliferative basal cell in CO and CF.

(a) Representative IF images of airways showing KRT5 (green) and PCNA (cyan), all nuclei are counterstained with DAPI (blue) in the merged image. Scale bar shows 75 μm. (b) Representative examples of watershed segmentation for isolated KRT5 and PCNA staining. (c) Representative images indicating counting of KRT5 (green) and PCNA (cyan) expressing cells in the segmented images. Scale bar shows 75 μm. Red and yellow boxes highlight areas that provide 4x zoomed images. (d) Segmentation data assumes a normal distribution. Each data point represents a possible cell and its corresponding area. Red line represents the mean area of the data and black line represents two standard deviations above the mean area. Representative tiles scan regions taken at 20x magnification for non-CF (e) and CF (f) subjects stained for KRT5 (green), PCNA (cyan) and nuclei are counterstained with DAPI (blue). Dimensions of the airways are indicated by the white lines. In all cases, images are representative of 14 CF and 17 CO fields of view.

Extended Data Fig. 9.

Extended Data Fig. 9.. FACs isolation of airway epithelial cells.

Representative FACS plots for isolation of epithelial cells to use in scRNAseq with 10X Genomics. Cell debris were excluded on the basis of FSC-A versus SSC-A, then doublets were removed using Trigger Pulse Width versus FSC-A (Influx). Dead cells were identified and excluded on the base of staining with DAPI. Negative gating for CD45, CD31, and CD235a, combined with positive gating for EPCAM (CD326) were used to identify epithelial cells.

Figure 1.

Figure 1.. Single cell transcriptome atlas of the epithelium lining proximal airways of control donors and donors with end-stage CF lung disease

(a) Locations of cell procurement for scRNAseq. (b) Methodology used for cell isolation by each institution. (c) Dimensional reduction of data generated from freshly isolated control and CF airway epithelium, visualized by UMAP, with cells colored by subsets as shown in key. (d) Distribution of cell subsets by institution. Error bars show standard error of the mean. N for UCLA=17, CSMS=16, and CFF=5 biologically independent samples. (e) Scaled expression of the top differentially expressed genes that inform specific cell subsets, for k-groups of control and CF cells further separated by subset, visualized by heatmap. (f) Dimensional reduction of data generated from air-liquid interface cultures (ALI) derived from samples shown above. Cells are colored by ALI-specific subsets, shown in key at right. (g) Heatmap of the scaled expression of the same fresh tissue subset genes from (e) but shown for groups of ALI- control and CF cells split by subset. (h-j) Comparison of subset-specific gene expression among fresh tissue subsets and cultured cells. (k) Distribution of the average proportion of cell subsets per sample, comparing CO and CF cells. Error bars show standard error of the mean. N is 19 CF and 19 CO samples. (l-p) CFTR expression in subset groups, key at right. (l) CFTR expression across all subsets, shown on the UMAP projection and as a boxplot of CO/CF versus expression level (m) Proportion of CFTR expressing cells per each subset. (n) Proportion of CFTR expressing cells and (o) CFTR expression, for _CFTR_+ cells only, visualized by stacked column charts. (p) Distribution of CFTR expression in all subsets, for _CFTR_+ cells only, divided by CO and CF status. P values (Wilcox test) shown at right indicate the significance of distribution differences between CO and CF per subset, bolded if p value < 0.05 . Whiskers show 1.5 times the interquartile range.

Figure 2:

Figure 2:. Expansion of secretory function, including mucus secretion and antimicrobial activity, in CF secretory cells

(a-e) Validation of secretory cell subsets in sections from CO lung tissue. Lower panels are magnifications of outlined dashed white box in the upper panels. (a, b) Immunostaining for SCGB1A1 (white), mucins 5B (green) and 5AC (red), identify secretory subsets 1, 2, and 4. SMG: submucosal gland. (c) In situ hybridization for Scgb1a1 (green), Muc16 (red), and Foxj1 (white), identify secretory subset 3. (d) Immunostaining for lactoferrin (LTF)(white), mucins 5B (green) and 5AC (red), identify secretory subset 5. (e) Dot plot indicating the expression of level and frequency of genes from panel a to d. Scale bars: e, h = 50 μm; f = 100 μm; g = 20 μm. (f) Dot plot indicating the expression level and frequency of differentially expressed genes from each secretory subset, across all subsets in CO and CF cells. Genes are expressed higher in either CO or CF, as indicated by label at top. (g) For gene networks preferentially located in secretory cells, shown is a gene ontology heatmap of the top 3 associated terms for each network with the term enrichment −log(p-value) colored as displayed in key. Networks with no associated ontology terms are blank (Net S6/S7). (h) For each cell, the average mean expression of the genes in a given network is shown, visualized on a UMAP. Cells are split by Secretory or non-Secretory, and CO or CF classification (i) Bar plots showing the average expression of all genes in individual secretory networks per secretory subset, in CO or CF cells.

Figure 3:

Figure 3:. Cilia related gene expression is vastly expanded outside of the main cilia subgroups in CF

(a) Dot plot indicating the expression level and frequency of differentially expressed genes in each ciliated subset, for CO or CF cells. (b) For gene networks preferentially expressed in ciliated cells, shown is a gene ontology heatmap of the top 3 associated terms for each network with the term enrichment −log(p-value) colored as displayed in key. (c) For each cell, the average mean expression of the genes in a given network is shown, visualized on a UMAP. Cells are split by Ciliated or non-Ciliated, and CO or CF classification (d) Bar plots showing the average expression of all genes in individual ciliated networks per ciliated subset group, in CO or CF cells. (e) For distinct categories of genes related to cilia biogenesis, the expansion of cilia gene expression is shown by a heatmap indicating the proportional percent change in amount of cells in each subset expressing each category above a threshold, towards CF(+%) versus CO(−%) cells. The percent change number between CF and CO samples is given in each heatmap cell and colored as indicated in key at right. (f, g) Validation of the basal to ciliated cell transition in sections from CO and CF lung tissue. Lower panels are magnifications of outlined dashed white box in the upper panels. (f) In situ hybridization for Krt5 (green) and Lrrc6 (red). Arrowhead indicates _Krt5_+ basal cell in suprabasal position showing co-expression for Lrrc6. Quantification of _Krt5_+ Lrrc6 + basal cells in CO and CF airways is shown by scatterplot. *: p=0.0119 (Wilcox test). (g) Immunostaining for KRT5 (red) and FOXJ1 (green). Arrowhead indicates KRT5+ basal cell in suprabasal position showing co-expression for FOXJ1. Quantification of KRT5+ FOXJ1+ basal cells in CO and CF airway is shown by scatterplot. *: p=0.0486 (Wilcox test). The red arrow indicates a CO sample that showed levels of colocalization similar to CF. The bar shows the mean and n=3 (f) or 4 (g) for each sample.

Figure 4:

Figure 4:. Depletion of metabolic stability, basal epithelial function, and cellular division is widespread in CF lung basal cells

(a) Dot plot indicating the expression level and frequency of differentially expressed genes in each basal subset, for CO or CF. (b) For gene networks highly expressed in basal cells, shown is a gene ontology heatmap of the top 3 associated terms for each network with the term enrichment −log(p-value) colored as displayed in key. (c) For each cell, the average mean expression of the genes in a given network is shown, visualized on a UMAP. Cells are split by Basal or non-Basal, and CO or CF classification (d) Bar plots showing the average expression of all genes in individual basal networks per basal subset group, in CO or CF cells. (e) Immunostaining for KRT5 (green) and PCNA (red) in sections from CF and CO lung tissue. Nuclei are stained with DAPI. Arrow indicate points of interest, while insets show magnification of the basal cell layer. Scale bar shows 50 μm. (f) Quantification of KRT5+ PCNA+ basal cells in CO and CF. *: p= 0.0034 (Wilcox test). Error bars show standard error of the mean, and n=6 for each sample. (g) Expression distributions of cell cycle genes in CO and CF cells, in the proliferating Basal2 subset.

Similar articles

Cited by

References

    1. Carraro G, et al. Single-Cell Reconstruction of Human Basal Cell Diversity in Normal and Idiopathic Pulmonary Fibrosis Lungs. Am J Respir Crit Care Med 202, 1540–1550 (2020). - PMC - PubMed
    1. Plasschaert LW et al. A single-cell atlas of the airway epithelium reveals the CFTR-rich pulmonary ionocyte. Nature 560, 377–381 (2018). - PMC - PubMed
    1. Montoro DT, et al. A revised airway epithelial hierarchy includes CFTR-expressing ionocytes. Nature 560, 319–324 (2018). - PMC - PubMed
    1. Pan S, Iannotti MJ & Sifers RN Analysis of serpin secretion, misfolding, and surveillance in the endoplasmic reticulum. Methods Enzymol 499, 1–16 (2011). - PubMed
    1. Rokicki W, Rokicki M, Wojtacha J & Dzeljijli A The role and importance of club cells (Clara cells) in the pathogenesis of some respiratory diseases. Kardiochir Torakochirurgia Pol 13, 26–30 (2016). - PMC - PubMed

Publication types

MeSH terms

Substances

Grants and funding

LinkOut - more resources