Single-cell transcriptomic profiling of the aging mouse brain

Abstract

The mammalian brain is complex, with multiple cell types performing a variety of diverse functions, but exactly how each cell type is affected with aging remains largely unknown. Here, we performed a single-cell transcriptomic analysis of young and old mouse brains. We provide comprehensive datasets of aging-related genes, pathways and ligand-receptor interactions in nearly all brain cell types. Our analysis identified gene signatures that vary in a coordinated manner across cell types and gene sets that are regulated in a cell-type specific manner, even at times in opposite directions. These data reveal that aging, rather than inducing a universal program, drives a distinct transcriptional course in each cell population, and highlight key molecular processes, including ribosome biogenesis, underlying brain aging. Overall, these largescale datasets provide an important resource for the neuroscience community (accessible online at https://portals.broadinstitute.org/single_cell/study/aging-mouse-brain) that will facilitate additional discoveries directed towards understanding and modifying the aging process.

Zonation-dependent single-endothelial cell transcriptomic changes in the aged brain

“…Several of the shared downregulated genes (e.g. Cst3, Pcsk1n and Sparc) were previously reported to be generally downregulated in EC and other aged brain cell types 40 (Fig. 1h).…”

“…Prompted by this observation, we further examined patterns of expression changes of AJ, TJ component-encoding genes, and genes that regulate BBB transcellular transport in different brain EC subtypes. The claudin 5-encoding gene Cldn5 has been reported to be downregulated in ECs in general 40 . In our data, a significant downregulation of Cldn5 occurred only in aged vcapEC ( Fig.…”

Detection of differentially abundant cell subpopulations discriminates biological states in scRNA-seq data

“…Importantly, it also allows us to uncover markers characteristic to a DA region that might not be detected by standard clustering-based approaches. In this section, we apply our method to four publicly available single cell datasets [14,31,36,39] (refer to the appendix for additional results based on simulated data). In datasets [14,31], most cells are not distributed in a clear structure with well-separated clusters.…”

“…The normalized expression matrix of scRNA-seq data from young and old mice was generated by Ximerakis et al [39]. Cell metadata -including ceype,cell sample labels (from young and old mice) -was also obtained from the original paper.…”

“…Cell metadata -including ceype,cell sample labels (from young and old mice) -was also obtained from the original paper. As described in [39], PCA was carried out after the identification of variable genes by the "mean variance plot" method from Seurat. The top 20 PCs were retained to calculate 2D embedding with t-SNE.…”