Spatial phenotyping of human bronchial airways in obstructive lung disease

Abstract

Chronic respiratory diseases such as asthma and COPD involve interactions between multiple resident and immune cell types within bronchial airways, resulting in structural and functional changes. Thus cellular heterogeneity, arrangements and associated neighborhoods as well as interactions between cells and matrices represent intriguing yet challenging areas of study. Spatial phenotypic profiling facilitates exploration of these issues of the cellular microenvironment and identification of context-dependent cell-cell interactions. Utilizing spatial phenotyping, we interrogated the features and cellular landscape of lungs from non-asthmatics, asthmatics, and COPD in FFPE samples by developing a 10-plex antibody panel for the Akoya PhenoCycler^®-Fusion system, focused on immune cells (CD45, CD3, CD4, CD8), proliferative cells (Ki67, PCNA), angiogenesis (CD34), epithelium (E-cadherin), smooth muscle (SMA) and extracellular matrix (collagen). We performed cell segmentation on multiplex immunofluorescence images and quantified marker intensity in each cell. Phenotypes were manually identified after normalization, integration, and clustering cells across samples. The composition, cell profiling, and distribution varied significantly between asthmatics and COPD compared to non-asthmatics emphasizing disease heterogeneity. Spatially agnostic analysis revealed that the matrix cluster was more abundant in COPD compared to non-asthmatics and asthmatics, consistent with a greater role for fibrosis. However, asthmatic patients had a higher proportion of unclassified and CD8 + clusters highlighting immune responses. Co-localization analysis showed near random distribution in non-asthmatics. But strong spatial interaction between T cells and other immune or matrix cells in asthma, and a higher avoidance of smooth muscle and immune cells, and of proliferative markers in both asthmatic and COPD. Niche analysis demonstrated different recurrent cell-cell interactions in asthmatic and COPD cohorts. In COPD, the matrix cell-enriched niche was more abundant, while in asthmatics, the unclassified cell-enriched niche was more prevalent compared to non-asthmatics. These findings provide insights into differential spatial organization of cells and tissues in asthma and COPD, with immune and epithelial mechanisms suggesting active inflammation and remodeling in asthma, but fibrotic processes in COPD, and potential role for vascular processes in both conditions.

Keywords: Airway epithelium; Asthma; CODEX; COPD; Cell-cell interactions; Cell-matrix interactions; Functional phenotyping; Immune cell; Mesenchyme; Neighborhoods; Spatial biology.

Fig. 1

Representative (tissue microarray (TMA) core) imaging of key cell phenotypic markers at single cell resolution using the Akoya Phenocycler Fusion (non-asthmatics, asthmatics) and CODEX (COPD) to demonstrate topography of human lung tissues under different conditions. (A) Non-asthmatics, (B) Asthmatics, (C) COPD. Multiplex immunofluorescence in TMA images stained with smooth muscle actin (SMA; Green), E-Cadherin (Yellow), PCNA (Red), Ki67 (Pink), CD34 (Orange), CD8 (Teal), CD4 (white), CD3 (Yellow), CD45 (Magenta) are shown for illustration

Fig. 2

Cell phenotyping in human lung sections. (A) Clustering assignment heatmaps for non-asthmatics, asthmatics, and COPD where the heatmap shows a clustering dendrogram with cell types (columns) corresponding to the legend (far right). The heatmaps show cellular phenotyping markers (CD3, CD8, CD4, CD45, SMA, E-cadherin, Collagen IV) where the level of expression is shown in blue as low level and others colors indicate high levels. Each colored bar segment corresponds to a specific cell phenotype, such as Matrix, Cytotoxic T cells (CD8+), Epithelium, Helper T cells (CD4+), other immune cells, smooth muscle cells and unclassified cells. (B) Cell abundance analysis in non-asthmatics, asthmatics and COPD per ROI displaying the percentage distribution of different cell types in the 7 clusters. (C) Quantification of the 7 cellular phenotype clusters enable statistical comparisons of cell type abundances between the non-asthmatic, asthmatic, and COPD groups. For each cell type, p-values from statistical tests are shown above the bars, indicating the statistical significance of differences between the cohorts. The asterisks represent different significance levels: * (p < 0.05), *** (p < 0.01), and ns (not significant)

Fig. 3

Functional phenotyping in human lung sections: Abundance graph (stacked bar chart) displaying the percentage distribution of functional phenotypes across the different clusters in non-asthmatics, asthmatics, and COPD. (A) Abundance plots for Ki67, PCNA, and CD34 in the different clusters of non-asthmatics, asthmatics, and COPD comparing the presence and absence of the different functional phenotypes. (B) Statistical analysis of percentage of cells (% to the parent cell type) in the different clusters of non-asthmatics, asthmatics, and COPD. p-values from statistical tests are shown above the bars, indicating the statistical significance of differences between the cohorts. The asterisks represent different significance levels: * (p < 0.05), *** (p < 0.01), and ns (not significant)

Fig. 4

Cellular colocalization of pairs of cell types in spatial proximity: (A) Example of spatial visualization map of various cell types and their functions within lung tissue (non-asthmatic). The legend lists different cell types. Each colored dot on the map represents an individual cell detected and its localization. (B) Cellular co-occurrence in cellular phenotyping of pairs of cell types in spatial proximity with self-pairs in non-asthmatic, asthmatic, and COPD. (C) Co-occurrence in functional phenotyping of pairs of cell types in spatial proximity with self-pairs in non-asthmatic, asthmatic, and COPD. Colocalization permutations through SciMap with a radius of 96px/~48 μm permutation = 1200)

Fig. 5

Functional phenotyping of asthmatic and COPD samples reveals additional degrees of heterogeneity (Niche analysis): (A) Tissue map of cellular niche analysis for non-asthmatic, asthmatic, and COPD. (B) Spatial organization of communities reveals cell-cell interaction within specific spatial niches. Identification of the 7 distinct communities on the 75 cells of spatial characteristics and their respective frequencies (enrichment score) within each cellular neighborhood (data pooled from all cohorts) (C) Niche frequency per cohort. Each bar represents the mean of cellular neighborhood frequency, the bottom graph showing the total number of cells found in each community, and the bar charts showing the relative sizes of the neighborhoods, which can be related to the overall cellular composition and structure of the tissue. For each group, p-values from statistical tests are shown above the bars, indicating the statistical significance of differences between the cohorts. The asterisks represent different significance levels: * (p < 0.05), *** (p < 0.01), and ns (not significant)