Extended Data Fig. 3: Ageing delays the molecular progression of LUAD in vivo and in vitro.
From: Ageing limits stemness and tumorigenesis by reprogramming iron homeostasis
(a) Uniform manifold approximation projection (UMAP) embedding of LUAD single-cell transcriptomes isolated from young and aged KP tumors labeled based on previously defined transcriptionally distinct subsets23, at 4, 12, and 17 weeks post-tumor initiation. The numbers indicate order of cell state progression23. WT AT2: wild-type AT2 cells, the cell of origin (“0”). (b) MetaCell neighborhoods70, groups of similar cells representing discrete cell states, of 4-week-old KP LUAD tumors projected onto the UMAP introduced in (a). Circle size corresponds to the number of cells, while color signifies the fraction of young and aged cells forming a MetaCell neighborhood (blue: enriched in young; red: enriched in aged). Note enrichment of young cancer cells at the high-plasticity cell state (HPCS, dashed blue oval). Validation of higher proportion of young cancer cells in HPCS by immunofluorescence for the HPCS marker integrin α2 (green) at 4 weeks post-tumor initiation. tdTomato (red) marks cancer cells. N = 54 aged and 125 young tumors. Scale bar: 50 µm. Arrowheads indicate integrin α2 positive cells. (c) Quantification of KRT8-high tdTomato+ LUAD cells at 4 weeks post-tumor induction (n = 99 aged and 91 young tumors). Scale bar: 50 µm. Arrowheads point to KRT8-high cells. (d) MetaCell analysis of KP LUAD tumors at 17 weeks post-tumor initiation. Note enrichment of young cancer cells at the endoderm-like state (dashed blue circle), which is validated by immunohistochemical staining for the endoderm-like state marker HNF4α. Inset shows nuclear HNF4α immunolabeling in neoplastic cells (brown) in the tumors. The proportion of tumors containing >5% HNF4α+ cells per the total numbers of tumors is shown. N = 10 young and 11 aged tumor-bearing mice. Scale bar: 100 µm. (e) Schematic summary of experiment evaluating molecular progression of KP LUAD cells in vitro. Briefly, AT2 cells were isolated from aged vs. young KP-Cas9 mice and transformed ex vivo by lentiviral Cre recombinase (P0, as shown in Fig. 1g). Bulk mRNA sequencing (RNA-seq) was performed over eight serial passages of tumor spheres (P1-P8). Non-KP alveolar organoids from both aged and young AT2 cells are also included. (f) Projection of bulk RNA-seq data from ex vivo transformed young and aged AT2 cells in diffusion pseudotime space. Upper panel: samples colored according to diffusion pseudotime using untransformed AT2 cells as the starting point (0). Lower panels: samples colored based on age (blue: young; aged: red). Note enrichment of young tumor spheres at later pseudotime points (blue dashed oval), whereas aged spheres are enriched at the midway point (dashed red circle). (g) Boxplots showing diffusion pseudotime distribution of young (blue) and aged (red) ex vivo transformed cells stratified according to passage (P). Individual samples are represented by data dots (p = 0.0014, Wilcoxon ranked sum test). N = 5 young and 5 aged biological ex vivo transformed tumor sphere cell lines. Maxima and minima are indicated by the whiskers; 25th and 75th percentiles are shown by the boundaries of the box; median is shown as the center line in the box. (h) Heatmap showing the mean expression of alveolar epithelial lineage markers [AT2 marker genes (Sftpb, Sftpc, Lyz1 and Lyz2) and AT1 marker (Ager)]in young and aged ex vivo transformed cells stratified according to passage. The colormap shows the log2-fold change compared to sample mean for each gene harmonized for young and aged samples. (i) Heatmap displaying the MAST DEG coefficient of the top 25 most consistently upregulated and downregulated genes across normal and tumor cell types. The top 25 upregulated and top 25 downregulated genes with aging were selected based on (i) significance in normal AT2 with FDR < 0.1; (ii) similar gene expression change trend across AT2 cells and all LUAD cell states; (iii) the sum of absolute DEG coefficient across AT2 cells and all LUAD cell states. The color bar indicates the MAST coefficient value. The genes are ordered from most downregulated (dark blue) to the most upregulated (dark red) in aged across the x-axis. The uniform manifold approximation and projection (UMAP) plot of the single-cell transcriptomics showing wildtype AT2 cells (cluster 0) and five LUAD cell states (AT2-like, AT1-like, high-plasticity cell state, HPCS, endoderm-like and ribosome-high, cluster 1-5) that molecularly define LUAD progression is shown in Extended Data Fig. 3a. Note that the expression of the DEGs in the wildtype AT2 is shown at the top row and the five LUAD cell clusters follow. The DEGs were ranked based on a two-part generalized linear statistical model where they were identified in all tumor cells with batch, sex, cell type, and cellular detection rate added as covariates. Subsequently, the aging-related DEGs in the normal AT2 cells and in all five LUAD cell states were identified separately with batch, sex, and cellular detection rate included in the model as covariates. Y: young, A: aged. Median and 25th and 75th percentiles are shown in (b-c). Mean with SD is shown in (d). Two-tailed Student’s t-test was used in (b-d).
