A panoramic view of cell population dynamics in mammalian aging

Abstract

To elucidate aging-associated cellular population dynamics, we present PanSci, a single-cell transcriptome atlas profiling >20 million cells from 623 mouse tissues across different life stages, sexes, and genotypes. This comprehensive dataset reveals >3000 different cellular states and >200 aging-associated cell populations. Our panoramic analysis uncovered organ-, lineage-, and sex-specific shifts in cellular dynamics during life-span progression. Moreover, we identify both systematic and organ-specific alterations in immune cell populations associated with aging. We further explored the regulatory roles of the immune system on aging and pinpointed specific age-related cell population expansions that are lymphocyte dependent. Our "cell-omics" strategy enhances comprehension of cellular aging and lays the groundwork for exploring the complex cellular regulatory networks in aging and aging-associated diseases.

Fig. 1.. Overview of experimental design and main cell type annotation across mammalian organs.

(A) (Top) Schematic representation of the sample collection process detailing the various ages, sexes, and genotypes (including wild-type and immunodeficient mice) used in the study. The approximate human age equivalents for different life stages of mice are referenced from the Jackson Laboratory website (95). (Bottom) Flowchart illustrating the experimental procedures of single-cell RNA sequencing by combinatorial indexing through EasySci. PCR, polymerase chain reaction. (B) Logarithmic scale bar plot depicting the number of high-quality cells profiled from each organ or tissue, after quality filtering. (C) UMAP plots displaying the cellular heterogeneity of each organ or tissue, with cells color coded by identified main cell types. Brain cell types were retrieved from (12). An aggregated UMAP plot of the entire dataset (comprising only wild-type cells, without batch correction) is also shown (bottom-right corner), with cells distinguished by organ or tissue origin and lineage.

Fig. 2.. Identification of aging-associated cell population shifts across organs and tissues.

(A) Dot plots illustrating cell type–specific fractional changes (log-transformed fold change) between ages 6 and 23 months. Main cell types are represented by triangles and subclusters by dots, with key gene markers labeled for select subclusters. The dendrogram is derived from hierarchical clustering of gene expression correlations among main cell types. AM, alveolar macrophages; IM, interstitial macrophages; DC, dendritic cells; ICB, type B intercalated cells; DCT, distal convoluted tubule cells; TAL, thick ascending limb of LOH cells; Sis, Sis positive cells; Uro, urothelial cells; VEC, vascular endothelial cells; Podo, podocytes; LEC, lymphatic endothelial cells; Meso, mesothelial cells; Type II, type II myonuclei; NJM, neuromuscular junction myonuclei. (B to E) Correlation scatterplots (using Spearman correlation) comparing fractional changes in main cell types [(B) and (D)] and subclusters [(C) and (E)] between female and male mice during two age intervals: 6 months versus 23 months [(B) and (C)] and 3 months versus 16 months [(D) and (E)], with a linear regression line. For all scatterplots, aging-associated cell types that are significantly changed in both age intervals are colored by the direction of changes.

Fig. 3.. The temporal dynamics, tissue distribution, and molecular signatures of aging-associated cell populations.

(A) Heatmap illustrating the fractional changes of aging-associated main cell types across five life stages. (B) Box plots depicting the fractional changes of normalized Mirg⁺ cell counts in muscle tissue from female (left) and male (right) mice across various age points. Each box plot represents the distribution of normalized cell numbers at specific ages, with linear regression lines indicating the trend of cell depletion over time. The slope, P value, and coefficient of determination (R²) value are marked on the plot. (C) Schematic of the Dlk1-Dio3 locus, highlighting Mirg⁺ cell marker genes. (D) Dot plot displaying marker gene expression in the PanSci muscle dataset, with color indicating average expression and dot size showing the percentage of cells expressing each marker. (E) Heatmap displaying the scaled proportions of aging-associated subclusters for each replicate within the specified age groups, normalized by total cells per tissue per animal. The top bars categorize the cell types by lineage and the organ or tissue type. Depletion and expansion dynamics are identified with hierarchical clustering of aggregated cell proportion of each time point for each aging-associated subcluster. (F) Stacked bar plots representing the proportions of aging-associated subclusters from different lineages and organs or tissues in each dynamic change. (G) Line plot showing normalized cell proportion changes in each aging dynamic, with Loess regression lines centered at the initial age point. (H) UMAP visualization of 634,185 wild-type cells from aging-associated subclusters, colored by organ or tissue. (I to N) Density plots showing the distribution of aging-associated subclusters from all depletion dynamics (I), all expansion dynamics (J), first depletion dynamics spanning 3 to 6 months (K), second depletion dynamics extending to 12 months (L), first expansion dynamics starting from 12 months (M), and second expansion dynamics from 16 months (N). Lymphoid subtypes are highlighted in blue, and cells from nonlymphoid lineage are annotated with enriched genes.

Fig. 4.. Identifying aging-associated lymphocytes across organs and tissues.

(A) UMAP visualization of 957,975 T cells and ILCs across various organs and tissues, colored by cluster ID. (B) Dot plot illustrating marker gene expression for T cell and ILC subtypes. The color denotes average expression values, and the dot size indicates the percentage of cells expressing these markers. (C) Heatmap displaying the normalized and scaled distribution of each T cell and ILC subtype across different organs and tissues. Aggregated cell numbers from all five age groups are normalized within each tissue and later calculated as proportions across tissues. Asterisks mark significant enrichment [false discovery rate (FDR) of 0.05, permutation test] in a given tissue relative to the remaining tissues. (D) Density plot highlighting the distribution of significantly depleted (left) and expanded (right) T cell and ILC subclusters in aging, with their respective marker genes. (E) Stacked bar plot depicting the proportion of CD4⁺ naïve T cells (left) and CD8⁺Gzmk⁺ cytotoxic T cells (right) within each organ or tissue in wild-type cells, normalized by organ and age group. (F) UMAP visualization of 1,072,614 B cells and plasma cells across organs and tissues, colored according to cluster ID. (G) Dot plot showing expression of marker genes for B cell and plasma cell subtypes, with color indicating average expression and dot size reflecting cell expression percentage. (H) Heatmap illustrating the normalized and scaled distribution of each B cell and plasma cell subtype across organs and tissues. Aggregated cell numbers from all five age groups are normalized within each tissue and later calculated as proportions across tissues. Asterisks mark significant enrichment (FDR of 0.05, permutation test) in a given tissue relative to the remaining tissues. (I) Density plot revealing the distribution of aging-associated B cell and plasma cell subclusters with significant expansion in aging, annotated with distinct marker genes. (J) Stacked bar plot indicating the expansion of IgM⁺ plasma cells (left) and Petpip2⁺ aging-associated B cells (right) in each wild-type organ or tissue, normalized by organ and age group. (K) UMAP visualization demonstrating the widespread expression of Sox5 and Cdk14 in expanded B cell and plasma cell populations.

Fig. 5.. Characterizing lymphocyte-dependent cell population dynamics in aging.

(A) Scatterplots comparing the proportion changes of main cell types between C57BL/6 wild-type mice and Rag1 (left) or Prkdc (right) mutants. Immune cell lineages are highlighted with black circles, with significant alterations labeled. (B) Box plots illustrating the fraction changes of Mfge8⁺ cells in the duodenum (top) and jejunum (bottom) across life stages in both wild-type and mutant mice. Each dot represents a biological replicate. Box plots display the median (middle line), quartiles (box edges), and 1.5× interquartile range (whiskers). (C) Dot plot showcasing the expression of Cxcl13 and its receptor Cxcr5 in the PanSci duodenum dataset, colored by average gene expression and sized by the percentage of cells expressing these markers. (D) Heatmap visualizing fraction changes of aging-associated subclusters (identified in Fig. 3) between 3 and 16 months in C57BL/6 wild-type and immunodeficiency mutants, with hierarchical clustering revealing four distinct dynamic patterns. (E to H) Stacked bar plots presenting the proportions of aging-associated subclusters from different lineages and organs or tissues in each dynamic pattern. (I and J) Case study of kidney principal cells: UMAP visualization of 39,286 kidney principal cells [(I), top] and density plot depicting the distribution and marker genes of aging-depleted principal cells [(I), bottom]; box plot detailing population shifts in aging-depleted principal cells across different life stages in wild-type and mutant mice (J). (K and L) Case study of lung fibroblasts: UMAP visualization of 85,625 lung fibroblasts [(K), top] and density plot depicting the distribution and marker genes of aging-expanded lung fibroblasts [(K), bottom]; box plot detailing population shifts in aging-expanded lung fibroblasts across different life stages in wild-type and mutant mice (L). (M and N) Case study of kidney connecting tubule cells: UMAP visualization of 57,619 kidney connecting tubule cells (CNT) [(M), top] and density plot showing the distribution and marker genes of aging-expanded CNT [(M), bottom]; box plot detailing population shifts in aging-expanded CNT across different life stages in wild-type and mutant mice (N). (O and P) Case study of kidney urothelial cells: UMAP visualization of 7670 kidney urothelial cells [(O), top] and density plot showing the distribution and marker genes of aging-expanded urothelial cells [(O), bottom]; box plot detailing population shifts in aging-expanded urothelial cells across different life stages in wild-type and mutant mice (P). (Q and R) Case study of lung interstitial macrophages: UMAP visualization of 18,418 lung interstitial macrophages [(Q), top] and density plot showing the distribution and marker genes of aging-expanded interstitial macrophages [(Q), bottom]; box plot detailing population shifts in aging-expanded interstitial macrophages across different life stages in wild-type and mutant mice (R).