A comparative study of in vitro air-liquid interface culture models of the human airway epithelium evaluating cellular heterogeneity and gene expression at single cell resolution

Abstract

The airway epithelium is composed of diverse cell types with specialized functions that mediate homeostasis and protect against respiratory pathogens. Human airway epithelial cultures at air-liquid interface (HAE) are a physiologically relevant in vitro model of this heterogeneous tissue, enabling numerous studies of airway disease ^1â€"7 . HAE cultures are classically derived from primary epithelial cells, the relatively limited passage capacity of which can limit experimental methods and study designs. BCi-NS1.1, a previously described and widely used basal cell line engineered to express hTERT, exhibits extended passage lifespan while retaining capacity for differentiation to HAE ⁵ . However, gene expression and innate immune function in HAE derived from BCi-NS1.1 versus primary cells have not been fully characterized. Here, combining single cell RNA-Seq (scRNA-Seq), immunohistochemistry, and functional experimentation, we confirm at high resolution that BCi-NS1.1 and primary HAE cultures are largely similar in morphology, cell type composition, and overall transcriptional patterns. While we observed cell-type specific expression differences of several interferon stimulated genes in BCi-NS1.1 HAE cultures, we did not observe significant differences in susceptibility to infection with influenza A virus and Staphylococcus aureus . Taken together, our results further support BCi-NS1.1-derived HAE cultures as a valuable tool for the study of airway infectious disease.

Figure 1:. Analysis of epithelial structure and cell types produced by BCi-NS1.1- and primary-derived HAE cultures.

a. Diagrammatical representation of experimental design. HAE cultures were generated from either primary cells or the hTERT-expressing precursor cell line BCi-NS1.1. Cells were differentiated on Transwells for three weeks at air-liquid interface to generate three replicates of BCi-NS1.1-HAE cultures (derived from one donor), and three HAE cultures from three separate primary donors. b. Hematoxylin & Eosin (H&E) and Periodic Acid-Schiff (PAS)-Alcian blue staining of formalin-fixed paraffin-embedded transections of HAE cultures from either BCi-NS1.1 or primary precursor cells. All cross-sectional images are oriented with the basolateral surface of the culture on the bottom of the image and the apical surface on the top of the image. Scale bars = 150μm c. Cell type-specific immunohistochemical staining of transections of BCi-NS1.1- or primary-derived HAE cultures. DAPI (nuclei) stained in blue. Cytokeratin 5 (KRT5, basal cells), villin-1 (ciliated cells), CC10 (encoded by SCGB1A1) (secretory cells) and MUC5B (secretory cells) stained in green. MUC5AC (secretory cells) stained in red. Scale bars = 150μm. d-g. Quantification using Imaris software of cells stained positive for each cell-type specific marker, KRT5 (c), villin-1 (d), CC10 (e), and MUC (f) (sum of MUC5B and MUC5AC positive cells), represented as a percentage of the total phalloidin (actin)-stained area of the HAE cultures. h. Quantification using Imaris software of cells stained positive for MUC5B, MUC5AC, or both, represented as a percentage of total MUC+ cells.

Figure 2:. Cell population annotation and pseudotime analyses of scRNA-Seq data from BCi-NS1.1 and primary –derived HAE cultures.

a. A UMAP dimensionality reduction plot of 12,801 single cell transcriptomes, colored by cell population assignment. Semi-supervised annotation of scRNA-Seq data resolved 12 HAE cell populations: basal cells, proliferating cells, suprabasal cells, intermediate cells, three populations of secretory cells differentiated by their MUC5AC/B expression, ciliated cells, and rare deuterosomal cells, tuft-like cells, ionocytes, and neuroendocrine cells. B. Cell proportions by biological replicate. Cell populations with altered frequencies are noted in the legend indicating the change in BCi-NS1.1 relative to primary HAE cultures along with the significance level (* P < 0.05, ** P < 0.01, *** P < 0.001). c. A dot plot summarizing the scaled expression (dot color) and level of expression (dot size) of signature marker genes for each cell population, broken out by biological replicate. d. Pseudotime trajectory inference identified a single lineage spanning basal-to-secretory cells (both HAE culture types superimposed). Proliferating cells and low frequency cell types (grey) were excluded from trajectory analysis. e. Cell densities along the basal-to-secretory trajectory reveal differential cell progression between the BCi-NS1.1- and primary cultures (Kolmogorov-Smirnov Goodness of Fit Test, p = 2.2e-16). f. A boxplot of cell pseudotime values by cell type and scRNA-Seq replicate. Hypothesis testing for differential median pseudotimes for basal, intermediate and secretory I cells all reached the minimal p value of 0.1 via a Wilcoxon Rank Sum Test with 3 replicates per group (‡) but do not clear a 95% significance level.

Figure 3:. Gene expression analyses of BCi-NS1.1 and primary –derived HAE cultures.

a. Triangular heatmap of Spearman correlation coefficients for pseudobulk transcriptional profiles aggregated by cell population (color) and culture group (shape) scRNA-Seq replicate. Row and column clustering was determined by Ward’s linkage method and shown as a marginal dendrogram. b. Principal component analysis (PCA) of pseudobulk transcriptional profiles, showing point plots for principal component (PC) 1 vs. PC2 and PC3 vs PC4 for the top 3,000 variable genes. The percentage of total variance assigned to each component is indicated in the axis title. c-e. Differential gene expression analyses contrasting BCi.NS1.1 and primary –derived HAE cultures by cell population for basal, suprabasal, intermediate, secretory I, and secretory II cells. c. Volcano plots summarizing the 1,052 differentially expressed genes (DEGs) calculated from pseudobulk profile contrasts, defining significance as an adjusted p-value (Benjamini-Hochberg) < 0.05, log2FC > 1, and per-cell population expression in > 10% of cells. TERT and ISGs IFIT1, ISG15, and MX1 are annotated where significantly differentially expressed, along with the number of DEGs for each cell population d. An “Upset” bar plot displaying the intersection of DEGs by cell population. The majority DEGs are unique to a particular cell population, but a core set of 30 genes (outlined in gold) were similarly differentially expressed across all groups. e. Gene set enrichment testing using CAMERA for the top 30 C2 canonical pathways gene sets by FDR, ordered by directional p-value across cell populations. Sets which were not significantly different across conditions are shown in dark grey. A positive directional p-value represents sets which are positively differentially expressed in BCi-NS1.1- vs primary-derived HAE cultures and a negative value represents negatively differentially expressed sets. Secretory II cells in particular display an elevated interferon response signature, highlighted in red: “WP Type II Interferon Signaling”, “Reactome Interferon Alpha Beta Signaling”, “WP Immune Response to Tuberculosis”, “WP Type I Interferon Induction And Signaling During SARS-CoV-2 Infection”.

Fig. 4:. Infection of BCi-NS1.1- and primary-derived HAE cultures with IAV and S. aureus.

a, c, e. RNA levels of MX1, IFN-b, IFIT1 and ISG15 were measured using RT-qPCR at steady-state (a), after IAV infection (c) and after IFN- b treatment. b, d, f. Representative western blots for MX1 and phosphorylated STAT1 (p-STAT1), along with GAPDH controls for mock treated (b), influenza A virus (IAV)-infected (d) and IFN-b-treated (f) cultures, and quantification of western blot band intensity from three individual blots for each condition. g. S. aureus colony-forming units (CFU) after 18h of infection h. IAV plaque-forming units (PFU) after 24h of infection i. influenza A virus-infected cultures were fixed and stained with an antibody against influenza virus-NP, then infection was quantified as total NP positive area. j. LDH release was quantified after influenza A virus or S. aureus infection and plotted as a fold change of LDH release over mock infected cultures. k. Top-down images of representative IAV-infected HAE cultures, showing DAPI (blue) and influenza virus-NP (green). Scale bars = 1mm. l. Top-down images of representative S. aureus and IAV-infected HAE cultures, showing Phalloidin (red) and DAPI (blue). Scale bars = 1mm.