.
F U L L T E X T S O U R C E : Genome Research
Abstract
Background
Aging is a pleiotropic process affecting many aspects of organismal and cellular physiology. Mammalian organisms are composed of a constellation of distinct cell type and state identities residing within different tissue environments. Due to technological limitations, the study of aging has traditionally focused on changes within individual cell types, or the aggregate changes across cell types within a tissue. The influence of cell identity and tissue environment on the trajectory of aging therefore remains unclear.
Results
Here, we perform single cell RNA-seq on >50,000 individual cells across three tissues in young and aged mice. These molecular profiles allow for comparison of aging phenotypes across cell types and tissue environments. We find transcriptional features of aging common across many cell types, as well as features of aging unique to each type. Leveraging matrix factorization and optimal transport methods, we compute a trajectory and magnitude of aging for each cell type. We find that cell type exerts a larger influence on these measures than tissue environment.
Conclusion
In this study, we use single cell RNA-seq to dissect the influence of cell identity and tissue environment on the aging process. Single cell analysis reveals that cell identities age in unique ways, with some common features of aging shared across identities. We find that both cell identities and tissue environments exert influence on the trajectory and magnitude of aging, with cell identity influence predominating. These results suggest that aging manifests with unique directionality and magnitude across the diverse cell identities in mammals.
Introduction
Aging is a gradual process of functional and homeostatic decline in living systems. This decline results in increased mortality risk and disease prevalence, eventually resulting in death. Aging appears to be a conserved feature of eukaryotic biology, affecting organisms as phylogenetically diverse as the single celled S. cerevisiea, the eutelic nematode C. elegans, mice, and humans [45, 63, 91]. Despite the near universal nature of the aging process, the underlying causes of aging are poorly understood. Aging phenotypes have been observed and hypotheses have been proposed for more than a hundred years [97, 50, 37, 64], but we do not yet know the cellular and molecular players that cause aging or how they differ between biological contexts. Both the fundamental nature of aging and its negative effects provide motivation to enumerate these players and establish causal relationships among aging phenotypes.
Mammalian aging phenotypes manifest at the organismal, tissue, cellular, and molecular levels [100]. Extensive research has produced catalogs of aging phenotypes at the physiological level, providing functional and behavioral hallmarks of age related decline. Likewise, molecular profiling of nucleic acids, proteins, and metabolites has provided a phenotypic description of aging in individual tissues [88, 42, 4, 56, 12].
Additional lines of inquiry have worked to address a classical question of aging biology – do different tissues age in the same way? Transcriptomic analysis at the bulk tissue level have revealed common traits of aging, as well as tissue-specific features [82, 43, 12]. Proteomic analysis of brain and liver in young and old mice similarly suggests that most age related changes are tissue-specific [72]. However, the cellular origins of aging phenotypes within a tissue remain largely unknown [12, 70, 78].
Our current understanding of aging phenotypes at the cellular level is less complete than at the tissue level. Mammals contain a multitude of distinct cell identities, each exhibiting specialized functions. In the mouse alone, recent cell atlas efforts have revealed more than 100 cell types [80, 39]. These surveys have catalogued diverse murine cell identities, but the plasticity of these identities and their contributions to tissue and organism level pathology remain unknown.
At both the molecular and functional level, a host of aging phenotypes and associated mechanisms have been revealed in individual cell types [23, 18, 35, 46, 48, 84, 20, 62]. While some of these studies present unique features of aging within individual cell identities, it is difficult to compare them systematically due to differences in experimental conditions and assay methodology. Using traditional molecular biology assays, it is difficult to measure high-dimensional molecular phenotypes across multiple cell identities, making large scale comparisons of aging phenotypes across cell identities intractable. The recent development of single cell RNA-sequencing (scRNA-seq) has ameliorated this limitation, allowing for measurement of transcriptional features across all prevalent cell identities in a tissue in a single experiment.
Although the technology has only recently matured, scRNA-seq experiments in individual tissues have already revealed novel aspects of the aging process. In a pioneering single cell RNA-sequencing study of hematopoietic progenitors, the axis of aging was shown to be opposite the axis of differentiation [53]. Multiple investigations have reported that cell-cell heterogeneity [34, 4] and gene expression variance [67] increase with age. However, the specific influence of cell identity and tissue environment on the trajectory and magnitude of aging has yet to be resolved.
Here, we employ scRNA-seq to generate a set of molecular profiles in which we can compare aging phenotypes across cell identities. By profiling 50,000+ cells from three tissues in young and old mice, we identify common features of aging that span cell identities, as well as features unique to each identity. Using matrix factorization and optimal transport methods, we compute trajectories of aging for each cell identity and assess the influence of identity and environment on these trajectories.
Results
Single cell RNA-sequencing identifies a diversity of cell types and states in young and old mouse tissue
We collected transcriptional profiles of young and old cells of many identities by isolating single cells from the kidney, lung, and spleen of n = 4 young (7 months) and n =3 old (22-23 months) C57Bl/6 mice. All three tissues were collected from the same animals. Isolations were performed at the same time of day for each animal, limiting circadian variation which affects the expression of nearly half of all murine genes [101]. After single cell isolation, cells were immediately encapsulated and barcoded for library preparation using the 10X Genomics microfluidics system, followed by subsequent sequencing. Using standard techniques for the identification of cell containing microfluidic droplets, we recovered 55,293 individual cell transcriptomes (see Methods).
We determined cell type and state identity by leveraging annotations provided in the Tabula Muris compendium [80]. These annotations provide labels at the cell type level and follow the structured hierarchy of the “cell ontology” [8]. Some age-related changes may be unique to individual states within a cell type. Similarly, some changes at the level of cell types may be mediated by differences in cell state proportions (i.e. CD4 vs. CD8 T cells). To ensure that we can explicitly detect these cell state level changes, we manually annotated cell states within each cell type in the Tabula Muris (see Methods, Supp. Fig. 1). We use the term “cell identity” to refer to the combination of cell type and state labels, such that CD4 T cells and CD8 T cells are different cell identities (Fig. 1).
Figure I:
scRNA-seq reveals that non-immune cell type proportions are preserved with age.
(A) Schematic representation of the experimental design. Kidney, lung, and spleen tissue were isolated simultaneously from each young and old mouse. After generating single cell suspensions, cells were prepared for scRNA-seq using the 10X Chromium system. (B) UMAP embeddings of each tissue investigated in our data set. Colors represent cell type annotations. Cell types are derived using a deep neural network classifier trained on the Tabula Muris data set. © Matching UMAP embeddings depicting the age of each cell. (D) For each animal, we computed the proportion of cells in each state. The mean proportion of each cell state for each age across animals is presented as a bar graph. The underlying proportions observed in individual animals are overlaid as black dots.
We trained deep neural networks to classify cell types based on these annotations, then used these networks to predict cell types in our data (Fig. 1A, B; see Methods). We found that the networks transfer labels with high fidelity. We validated classifications by inspecting marker gene expression post-hoc (Supp. Fig. 2, 3) and computing correlations between cell identities in our data and the Tabula Muris (Supp. Fig. 4). From these cell identity annotations, we identify 19 unique cell types and 28 unique cell states across the three tissues. Comparing our cell types to the Tabula Muris, we recover all but one of the cell types identified (kidney loop of Henle epithelial cells, Supp. Fig. 3, 5). The cell type proportions we recover differ from the Tabula Muris – e.g. we recover comparatively more immune cells in the kidney and lung – but are not outside expected range based on previous comparisons between single cell RNA-seq datasets [80, 74].
Immune cells are more prevalent in old kidneys and lungs, while non-immune cell type proportions are preserved
One prospective way in which aging may influence tissue function is by altering the proportion of each cellular identity within the tissue. This change in cell identity compositions may occur at either the level of cell types, or by shifting the distribution of cell states within a cell type. To investigate the former possibility, we quantified the proportion of each cell type within each tissue across ages.
In both the kidney and lung, lymphocytes were significantly more abundant in old animals compared to young animals (t-test on additive log-ratio transformed proportions, q < 0.05)(Fig. 1D). In old kidneys, we found a roughly 2 fold increase in T cells, classical monocytes, and non-classical monocytes, and in old lungs a corresponding 2 fold increase in classical and non-classical monocytes and a roughly 1.3 fold increase in T cells. This may reveal increasing immune infiltration of the the non-lymphoid tissues with age, as suggested in previous studies of kidney, lung, and other non-lymphoid tissues [70, 78, 92, 6, 65]. However, we can not rule out that our recovery of specific cell types may be confounded by an interaction of aging with our isolation procedures.
Considering only non-immune cells in the kidney and lung, cell type proportions were not substantially altered by aging (Supp. Fig. 6B). Likewise in the spleen, we found minimal change in the proportion of cell types between young and old animals (Fig. 1D). While changes in non-immune cell type proportions are subtle, we cannot rule out that even these subtle changes may influence the aging process.
Shifting cell state proportions within a cell type may be an alternative mechanism by which aging phenotypes manifest. Examples of this phenomenon are present in the literature, such as the observed decrease in naive CD8 T cells relative to other T cell states [29, 38] and the shift from highly regenerative to less regenerative stem cell states in the blood and muscle [23, 14]. To address whether cell state proportions change with age, we quantify the proportion of cells in a given cell state for each cell type.
Investigating spleen T cells, we recapitulate the finding that CD8 T cells are less abundant in old animals. We observed a shift in the population of kidney collecting duct epithelial cells – old animals exhibit a decreased frequency of Cald1+ collecting duct cells relative to Slc12a3+ collecting duct cells (χ2 contingency table, q < 0.05). Aqp3+ and Slc12a3+ cells are likely principle cells of the collecting duct based on expression of Scnn1a. Cald1+ cells are also marked by Phgdh, which has a reported mosaic expression pattern in the proximal tubule, and Cryab, which was similarly identified to mark a distinct subpopulation in the Mouse Microwell Atlas [39] (Supp. Fig. 6C). However, we are not aware of a defined role for this cell population, making it difficult to speculate on the impact of this shift in cell state proportions. Other cell types with notable cell state substructure do not show shifts with age, such as lung stromal cells (Supp. Fig. 6A).
Cycling cells are similarly rare in young and old animals
Previous reports have suggested that cell cycle activity changes with age in multiple cell populations. In blood progenitors, cell cycle kinetics are accelerated with age [53], while in muscle progenitors the frequency of cycling cells increases with age [23], and in the intestinal crypt cycling cells become less frequent [69]. To investigate the possibility of changes in cell cycle frequency in our data, we evaluated cell cycle activity by scoring the expression of S-phase associated genes and G2M-associated genes [90] (see Methods). We observe only subtle changes in either of these cell cycle module scores with age across cell identities. The proportion of putatively cycling cells is also very small across cell identities (Supp. Fig. 7). These results suggest that the cell cycle rate in cell identities we observe is not dramatically changed with age. However, we cannot discount the possibility that the cell cycle scoring method we use is insufficient to detect differences.
The accumulation of non-cycling senescent cells in aging tissues has been reported in several previous studies [25]. The reported magnitude of senescent cell accumulation varies between tissues. In aging mouse kidney, the proportion of cells with senescence associated β-galactosidase activity increased from roughly 0.2% to 1.2%, whereas in epicardial cells the proportion increases from 2% to 10% (12 months to 18 months) [7]. Similar observations have been made using paired-end single cell RNA-seq in the human pancreas, where the proportion of cells expressing senescence marker gene CDKN2A increases from roughly 7% to 15+% between early- (21-22 years) and mid-adulthood (38-54 years) [34]. To investigate whether senescent cells are more prevalent in the old tissues we observe, we similarly measured expression of Cdkn2a. We find that the Cdkn2a locus (p16-Ink4a and p19-Arf) is not significantly upregulated with age in any of the cell identities we observe (Supp. Fig. 8). Due to the overlapping nature of the p16-Ink4a and p19-Arf reading frames, we note that we cannot distinguish transcripts from these two proteins using 3′-end RNA-seq alone [44]. We also scored the activity of a curated set of senescence-associated genes using the AUCell approach [1], but we do not find large differences in this score with age (Supp. Fig. 8).
Changes in cell-cell variation with age depend on cell identity
Single cell analysis allows us to measure not only the mean expression of each gene, but also the variation within a cell population. Previous work has suggested that both gene expression variance and cell-cell heterogeneity increase with age [67, 34, 5]. These two types of variation differ in subtle but important ways. Gene expression variance quantifies the mean dispersion across genes in the transcriptome, such that each gene contributes equally. Because genes are equally weighted, changes in gene expression variance are unlikely to be driven by a small number of genes. Likewise, multivariate differences in gene expression between cells are not resolved due to the focus on mean dispersion values. Increased gene expression variance may reflect a global change in transcriptional noise, perhaps due to loss of regulatory control, as suggested in previous aging studies [93].
By contrast, cell-cell heterogeneity measures the average distance in transcriptional space between cells in a population. These distances capture multivariate differences in gene expression, capturing variation in cell state that manifests across genes. They also account for gene expression level, such that a small number of more highly expressed genes can drive changes in cell-cell heterogeneity. Increased levels of cell-cell heterogeneity may reflect a diversification of cellular states within a population (Fig. 2A).
.../...
F O R T H E R E S T O F T H E S T U D Y, P L E A S E V I S I T T H E S O U R C E .
.
