scATAC-seq-based characterization of COPD-related changes in chromatin accessibility within murine lung tissue
A comprehensive approach was employed herein to characterize the mechanisms governing the pathogenesis of COPD in murine lungs. Initially, the chromatin accessibility landscape was profiled at the single-cell level via scATAC-seq, as in prior reports [40–41] (Fig. 2a). To mitigate gender- or individual-specific differences, an equal gender ratio was maintained during testing[42]. Lung tissues from four COPD model mice and two control mice were processed to obtain single-cell suspensions, from which nuclei were extracted and treated with Tn5 transposase. The quality of scATAC-seq data was assured by evaluating transcriptional start site (TSS) enrichment, mitochondrial DNA contamination, fragment size distribution, and doublet exclusion. In total, 22,038 nuclei that passed the quality filter were identified, corresponding to 12,158 cells from the COPD model group and 9,880 from the control group. These yielded a median TSS enrichment score of up to 26, a median of 4610 fragments, and 251,024 peaks that were called, with > 50% of the fragment fraction overlapping with these peaks. These data exhibited appropriate fragment size periodicity, with > 30% of fragments located at TSS and a high degree of correlation between the aggregate profiles within each group. The estimated doublet/multiple percentages in each group were 3.8% on average, and the corresponding data were removed from further analysis. Collectively, these results indicated consistent enrichment across open chromatin regions, providing a comprehensive dataset for further analysis.
To explore molecular and cellular heterogeneity related to COPD development more effectively, subsequent analyses were performed with the ArchR pipeline [43]. Dimensionality reduction was initially performed with TF-IDF, identifying 31 cell clusters at a resolution of 0.8. Distance-weighted accessibility models were then utilized to calculate a gene score matrix, converting the degree of chromatin accessibility into a corresponding measure of gene expression. Established clusters were then identified based on cell type-specific marker patterns using marker collections from sources including the CellMarker [44], Mouse Cell Atlas (MCA) [45], and PanglaoDB [46] databases as well as published analyses of murine tissues, particularly studies focused on murine lings. Ultimately, these 31 clusters were classified into 4 cell groups, including epithelial, endothelial, immune, and stromal cells, with 18 cell subtypes (Fig. 2b), in line with published profiles from mouse lungs [47–48]. AT1 and AT2 cells, alveolar macrophages, and J-chain + B cells were identified based on Ager, Sftpb, Mcemp1, and Jchain expression, respectively (Figure S1). Following these cell type determinations, differential peaks associated with genes and chromatin accessibility were compared between the control and COPD groups, revealing a marked increase in the accessibility of promoter and intronic regions in the latter group (Fig. 2c). COPD is known to be linked to the induction of an inflammatory response, reflected by the high levels of promoter accessibility in several immune cell populations in the COPD group. Analysis of peak regions and associated genes across 18 cell types (Supplementary Table 2) identified 1333 differential peaks in B cells, 6563 in endothelial cells, and 6188 in alveolar macrophages, with significant differences assessed by Wilcoxon test (false discovery rate [FDR] < 0.05, |log2FC| > 0.5) showing a predominance in immune cells, with about 40% of these peaks in promoter regions (Fig. 2d, Figure S2). This suggests a high degree of transcriptional activity associated with these regions, prompting further comparisons of total peaks with the genome to clarify trends in COPD-related accessibility changes.
A cis-regulatory interaction network, constructed using data from immune cell subsets, revealed consistent peak patterns across different cell types, with stronger signals in the COPD model group (Fig. 2e, Figures S3-S4). Notably, IL-10, an immunomodulatory cytokine with anti-inflammatory properties linked to various disease processes[49], showed increased fragment coverage and complex co-accessibility in the COPD group, especially in monocytes and B cells. In contrast, CD44 co-accessibility spanned a larger region and was relatively dispersed, although it exhibited similar between-group variance. Differential signals were also evident between COPD and control groups in the B cell and alveolar macrophage populations. Given that TFs are essential regulators of diverse cellular processes, motif identification was performed based on the differentially accessible peaks in these different groups, revealing multiple TFs that exhibited significant differences (Fig. 2f). These included the known inflammation-relate TFs Ilf3 and Nfkb2. These TFs showed higher binding signals in the COPD group, consistent with higher levels of chromosome accessibility and the activation of associated regulatory mechanisms.
Comparative analysis of cases and controls across all samples identified 28,139 DEGs, with 17,965 upregulated and 10,174 downregulated, predominantly enriched in immune-associated and universally essential pathways (Fig. 2g). Both the TNF and cytokine-cytokine receptor interaction pathways were enriched in most cell types, consistent with their broadly important roles in shaping diverse processes, including proliferation, apoptosis, differentiation, and immunoregulatory activity in response to corresponding activating stimuli and receptor-ligand binding events. Other pathways, including the Th17 cell differentiation and B cell receptor signaling pathways, were enriched in immune cell populations. Enriched GO biological process terms for these DEGs included protein binding, cytokine activity, and energy supplementation. Collectively, these findings from scATAC-seq analyses indicate enhanced cellular and intercellular activity in lung cells from the COPD model group.
scRNA-seq clarifies the pulmonary immune-related transcriptomic landscape associated with murine COPD
Lungs from eight C57BL/6 mice (six COPD models and two controls) were harvested and digested for scRNA-seq analyses to compare the two groups. Twenty-one cell types were identified based on molecular markers, including CD4 T cells, CD8 T cells, general T cells, B cells, J-chain+ B cells, proliferating lymphocytes, NK cells, monocytes, interstitial macrophages, alveolar macrophages, CD209+ DCs, CD103+/CCL17+ DCs, ciliated cells, Col13a1+ fibroblasts, Col14a1+ fibroblasts, smooth muscle cells, vascular endothelial cells, Vcam1+ endothelial cells, Pen cells, AT1 cells, and AT2 cells (Figs. 3a, S5). Relative to the control group, the COPD model group showed marked reductions in the relative abundance of AT2 cells, vascular endothelial cells, Vcam1+ endothelial cells, ciliated cells, and pen cells. In contrast, there were significant increases in the levels of B cells, monocytes, and NK cells, together with slight increases in CD4 T cell, CD8 T cell, and J-chain+ B cell abundance (Fig. 3b). Analyses of the proportions of differential genes expressed in these different cell types revealed many DEGs in B cells, T cells, monocytes, alveolar macrophages, and NK cells, the majority of which were upregulated (Fig. 3c). This was consistent with the overall cell type trends detected in this experiment. AT2, Col13a1+ fibroblast, and Col14a1+ fibroblast cells only showed upregulated DEGs, while ciliated cells, smooth muscle cells, and Vcam1+ endothelial cells exhibited fewer upregulated DEGs, and vascular endothelial cells exhibited more upregulated DEGs (Fig. 3c).
Enrichment analyses revealed the enrichment of DCs for neutrophil degranulation, regulation of cytokine production, and cellular responses to external stimuli signaling pathways. Monocytes were enriched for the positive regulation of cell death, neutrophil degranulation, regulation of cytokine production, inflammatory cell response, and cellular responses to external stimuli signal pathways. Interstitial and alveolar macrophages were enriched to regulate cytokine production, inflammatory response, and cellular responses to external stimuli pathways. CD4+ and CD8+ T cells were enriched for the positive regulation of cell death, neutrophil degranulation, cellular responses to external stimuli, phagosome, and apoptosis signaling pathways. B cells were enriched for the positive regulation of cell death, neutrophil degranulation, phagosome, and cellular responses to external stimuli signal pathways (Fig. 3d). Endothelial cells were enriched for the positive regulation of cell death, apoptosis, and cellular responses to external stimuli signaling pathways. Col13a1+ and Col14a1+ fibroblasts were enriched for the cellular response to stress and reactive oxygen species metabolic rate, Oxidative Stress, Redox, and cellular responses to external stimuli signaling pathways. Ciliated cells were only enriched for the cellular responses to external stimuli pathway.
Further differential TF expression analysis indicated downregulation of SOX7, HOXA5, KLF2, and FOXF1, linked to tracheal, alveolar, and lung mesenchymal development and differentiation in the COPD model group. In contrast, increases in the expression of FOXP1, EBF, and POU2AF1 associated with B cells and activated DCs were evident in COPD model mice (Fig. 3e). This suggests that alveolar and epithelial/endothelial cell development and differentiation are disrupted in COPD, whereas DCs and B cells are activated in this pathological setting. This may contribute to the induction of adaptive immunity, lymphatic follicular hyperplasia, and persistent lung tissue inflammation, resulting in progressive alveolar destruction and increasingly limited airflow.
Cell interaction analyses revealed that COPD was associated with increased interaction strengths between T cells, DCs, and other cell types, suggesting that both smoke and LPS exposure may stimulate DC activation (Fig. 3f, Figure S6). This activation may lead to the immune and inflammatory cell response, further implicating successive activation of AT2 cells and fibroblasts. These activated cells trigger the activation of damage-related molecular patterns associated with inflammatory responses, contributing to the release of assorted cytokines, chemokines, acute phase proteins, and antimicrobial peptides. The enhanced communication between AT1 cells, fibroblasts, and other cell types suggests that immune cell-regulated inflammatory signaling induced renewed AT1 and fibroblast differentiation and associated alveolar regeneration (Fig. 3f, Figure S6).
Integrated analyses of COPD-related scATAC-seq and scRNA-seq datasets
Integrated analyses of COPD-related scATAC-seq and scRNA-seq datasets provided a comprehensive view of transcriptional processes in matched lung tissue samples, confirming uniform data integration without batch effects using ArchR (Fig. 4a). Cell type classification was then performed using the integrated data (Fig. 4b), identifying 21 cell groups based on the expression of specific genes (Figure S3). Genetic analyses for these 21 cell groups revealed highly consistent trends with respect to the COPD-related changes evident in the scRNA-seq and scATAC-seq datasets (Fig. 4c). Color-based visualization of all samples in these two datasets confirmed excellent concordance between the scRNA-seq and scATAC-seq classification, with only relatively limited individual differences. Differential gene analysis and KEGG pathway enrichment indicated that the TNF signaling pathway might be associated with COPD incidence (Fig. 4d). TNFR1 upregulation in alveolar macrophages and active cellular communication was observed, along with scATAC-seq data showing increased accessibility of Bcl3, IL1b, Fos, and Csf1 loci, aligning with scRNA-seq findings of increased expression in COPD model cells compared to controls (Fig. 4e).