3.1 Multicenter and Large Sample Combination Analysis to Generate New Mechanistic Insights for Nintedanib for Pulmonary Fibrosis
In our study, we explored single-cell RNA-seq and bulk-seq datasets of mouse and human lung tissue during both healthy and pulmonary fibrosis states. Our comprehensive approach involved the systematic compilation of 7 independent single-cell RNA-seq datasets of human pulmonary fibrosis, together with 1 single-cell RNA-seq dataset of lung tissue from a bleomycin-treated mouse model of pulmonary fibrosis, 1 bulk-seq dataset of human pulmonary fibrosis containing 254 samples, and 1 bulk-seq dataset of nintedanib intervention in human fibroblasts (Supplementary Table 1). Our data screening strategy is outlined in Fig. 1.
For the analysis of human single-cell RNA-seq data, we integrated data from each study center and employed meta-analysis to evaluate the proportion of nintedanib-targeted gene-positive cells in each dataset, as shown in Fig. 2A-I and Supplementary Table 2. Additionally, we combined human pulmonary fibrosis bulk-seq, mouse bleomycin pulmonary fibrosis single-cell RNA-seq, and human fibroblast nintedanib intervention bulk-seq data to further identify nintedanib-sensitive and resistant cell populations.
3.2 Meta-analysis of human pulmonary fibrosis scRNA_seq sequencing data demonstrating the proportion of nintedanib target gene positive cells in different cell subpopulations of patients with different types of pulmonary fibrosis.
We sorted out fibroblast subpopulations from lung scRNA sequencing data in different kinds of ILD patients and counted the proportion of nintedanib target gene positive fibroblasts, and found that the proportion of FGFR2, PDGFRB, FGFR1, and FLT4 gene positive fibroblasts was significantly higher in IPF using meta subgroup analysis, and the proportion of FGFR1 gene positive fibroblasts was cHP were significantly elevated (Fig. 2A, Supplementary Table 2).
In smooth muscle cells, the proportion of FLT1 and FGFR1 gene-positive SMCs was also significantly elevated in pulmonary fibrosis, with the proportion of FLT1 + SMCs cells significantly elevated in cHP patients and the proportion of FGFR1 + SMCs cells significantly elevated in IPF and NSIP patients (Fig. 2B, Supplementary Table 2).
Among Ciliated cells, we found that the proportion of FLT1, FGFR2, KDR, PDGFRB and FGFR1 gene-positive Ciliated cells was also significantly elevated in pulmonary fibrosis, with the proportion of FLT1+, PDGFRB + and FGFR1 + ciliated cells significantly elevated in IPF patients, while the proportion of FGFR2 + ciliated The proportion of FLT1+, PDGFRB + and FGFR1 + ciliated cells was significantly higher in IPF patients, while the proportion of FGFR2 + ciliated cells was significantly higher in cHP patients (Fig. 2C, Supplementary Table 2).
In addition, among endothelial cells, the proportion of PDGFRA and PDGFRB gene-positive endothelial cells was also significantly elevated in pulmonary fibrosis. Subgroup analysis showed that the proportion of PDGFRA + ECs cells was significantly higher in patients with sarcoidosis, whereas the proportion of PDGFRB + ECs cells was significantly higher in patients with IPF (Fig. 2E, Supplementary Table 2).
For immune cells, we found that the proportion of nintedanib target gene positive cells was significantly elevated in T cells, B cells, Macrophages and Monocytes. Among them, In Macrophages, the proportion of FGFR1 + cells were significantly higher in IPF patients, while the proportion of FGFR2 and FGFR3 positive cells was significantly higher in SSc patients (Fig. 2F, Supplementary Table 2). in T cells, the proportion of FGFR2 + cells was significantly higher in NSIP patients, while the proportion of FLT4 + cells was significantly higher in cHP and NSIP patients (Fig. 2G, Supplementary Table 2); In Monocytes, the proportion of FGFR1 + cells were significantly higher in SSc patients (Fig. 2H, Supplementary Table 2).in B cells, the proportion of FLT1 + cells was significantly higher in NSIP patients (Fig. 2D, Supplementary Table 2).
In AT1 cells, we found that the proportion of FGFR3 and KDR gene-positive cells was also significantly higher in pulmonary fibrosis, with the proportion of FGFR3 + AT1 cells significantly higher in Sarcoidosis and NSIP patients, and the proportion of KDR + AT1 cells significantly higher in IPF patients (Fig. 2I, Supplementary Table 2).
Additionally, we combined the fibroblasts from the seven GSE data and divided the cell subpopulations into Mast cells, lymphatic ECs, AT1, SMCs, Ciliated cells, B cells, ECs, Fibroblasts, AT2, T cells, Monocytes, and Macrophages after removing the batch effect (Fig. 3A-B). We discovered that the proportion of cells positive for 15 nintedanib target genes was significantly higher in every cell subpopulation of IPF patients compared to those with other ILD types. The percentage of FGFR1-positive cells was found to be most significantly elevated, as demonstrated in Fig. 3C-F. Following our analysis, we further investigated nintedanib target gene expression in various cell subpopulations of lung tissue from human fibrotic patients and healthy controls. We found that FGFR1 expression was significantly elevated in fibroblasts, as demonstrated in Fig. 3G. Furthermore, we found that COL1A1 was significantly correlated with the distribution range of FGFR1 in cell subpopulations, as displayed in Fig. 3H. Our findings suggest that FGFR1 gene expression is significantly elevated in fibroblasts of both IPF and cHP, as shown in Fig. 3I.
3.3 Combined scRNA-seq and bulk-seq analysis identifies nintedanib -sensitive fibroblast subpopulations in IPF and cHP patients.
We combined the fibroblast data from the seven GSEs, removed batch effects, and classified the cell clusters into eight subgroups: Profibroblast, Mesothelial, Lipofibroblast, DGKG + Myofibroblast, NFIC + Fibroblast, ADIRF + Myofibroblast, Fibroblast, and SFRP2 + Fibroblast, as shown in Fig. 4A-B. Next, we analyzed the expression of differentially expressed genes (DEGs) in FGFR1 positive lung fibroblasts from IPF patients and found that 1299 genes were significantly upregulated, giving us new insights into gene expression changes in IPF (Fig. 4C, Supplementary Table 3). Additionally, we compared the DEGs after nintedanib intervention in fibroblasts and discovered that 128 genes were significantly downregulated, which suggested the effect of nintedanib on fibroblasts (Fig. 4D, Supplementary Table 4). We then compared the upregulated genes in lung fibroblasts from FGFR1-positive IPF patients with those downregulated after nintedanib intervention and identified 15 shared nintedanib-sensitive DEGs, which are shown in Fig. 4E. These findings may aid in the development of targeted therapies for fibrotic lung diseases.
We conducted an analysis of the 15 nintedanib-sensitive DEGs in different species of ILD (Fig. 4F) and observed the expression of gene sets in various cell subpopulations. As a result, we identified a subpopulation of nintedanib-sensitive fibroblasts (SFRP2 + Fibroblast), which is displayed in Fig. 4G-H and Supplementary Table 5. We found that this cell subpopulation had a significantly higher score in the IPF group relative to the normal control group (Fig. 4I). Additionally, we performed an analysis of lung bulk-seq data for various ILD patients and identified 7 genes that had negative associations with lung function, as shown in Fig. 4J. For cHP patients, we identified 30 nintedanib-sensitive genes (Fig. 4K, Supplementary Table 6) that were most highly expressed in the SFRP2-positive fibroblast subpopulation and were significantly higher in the cHP group than in the control group (Fig. 4M-N, Supplementary Table 7). We also used ILD lung bulk-seq data analysis to identify 11 genes that were negatively associated with lung function, which is represented in Fig. 4O. Moreover, we used GEPIA o analyze lung adenocarcinoma patients collected by TCGA and discovered that high expression of TPBG, a gene shared by cHP and IPF fibroblasts, is a predictor of lower survival rates (Fig. 4P). We also found that the expression of TPBG is positively correlated with the expression of lung adenocarcinoma mutant genes such as EGFR, BRCA, TP53, and KRAS (Fig. 4Q-T).
3.4 Discovery of a subpopulation of nintedanib -resistant genetic fibroblasts
In this study, we performed an analysis of the differentially expressed genes (DEGs) that were upregulated after nintedanib intervention in fibroblasts. We discovered that these genes were significantly enriched in fibronectin-related signaling pathways, as shown in Fig. 5A. To further investigate the expression of these signaling pathways in cell subpopulations, we conducted a clustering analysis, which identified a subpopulation of ADIRF myofibroblasts that were found to be resistant to nintedanib, as presented in Fig. 5B-C and Supplementary Table 8. We discovered that this cell subpopulation had significantly higher scores in the IPF and cHP groups when compared to normal controls, as shown in Fig. 5D-E.Our analysis of the association between drug resistance genes and clinical outcomes revealed that the ADGRG1 gene was negatively associated with lung function, as displayed in Fig. 5F. Interestingly, we found no significant difference in the expression of drug resistance genes in different pulmonary fibrosis fibroblasts, as shown in Fig. 5G.Additionally, we used GEPIA to analyze lung adenocarcinoma patients collected by TCGA and found that high expression of the drug resistance gene ADGRG1 was not significantly associated with survival, as displayed in Fig. 5H.
3.5 Combined human and mouse scRNA-seq analysis to identify nintedanib -sensitive and drug-resistant cell subpopulations.
We obtained lung scRNA data from the selected GEO database for nintedanib-treated bleomycin model mice. After quality control, we identified 11 different major cell subpopulations, with a total of 18,697 cells retained, as seen in Fig. 6A-E. Common marker genes were selected to label subpopulations, and we derived the cell proportions of single cell subpopulations in the three groups of mice. Furthermore, we extracted FGFR1-positive fibroblasts and found that the nintedanib-sensitive gene cluster scored significantly higher in the bleomycin group, while the score decreased significantly after nintedanib treatment, indicated by Fig. 6F. We also analyzed the expression of the drug-resistant gene set in lung fibroblasts of the three groups of mice and found that the gene set score increased after nintedanib treatment, with statistically significant differences, as seen in Fig. 6G.