Single-cell transcriptomics of peripheral blood in the aging mouse

scRNA-seq of young and old blood identifies different cell types

We performed scRNA-seq using 10x Chromium on peripheral blood obtained from 2 young and 2 old mice. We removed cells with more than 10% mitochondrial reads, less than 500 genes, or more than 4000 genes. 14588 cells that passed this filter (4642 cells from old mouse 1, 2187 cells from old mouse 2, 3902 cells from young mouse 1, and 3857 cells from young mouse 2) were subsequently processed using Seurat/2.3.0 and 17 clusters were obtained (Figure 1A, 1B). Cell-type specific marker genes identified the cell type of each cluster, including Cd3e for T cells, Cd79a and Ms4a2 for B cells, Nkg7 for NK cells, Ly6c2 and Cx3cr1 for monocytes or dendritic cells (DC), Ppbp, Pf4 and Gng11 for megakaryocytes, C1qa, C1qb and C1qc for macrophage, Fcer1a and Cd200r3 for basophil and Hba-a1 for erythrocytes (RBC) (Figure 1C). We identified 5 clusters of T cells, 4 clusters of B cells, 1 cluster of proliferating B or T cells, NK cells, monocytes, dendritic cells, megakaryocytes, macrophage, basophil, and (RBC), respectively (Figure 1C).

The naive T cells cluster is identified by two markers, Sell+ and Cd44-, the memory T cells cluster is identified by Klrb1c+, Id2^high, and Cd44+ and cytotoxic T cells are identified by high expression of Gzma (Figure 1C). We also identified four clusters of B cells, with one cluster expressing Fcer2a, Sell, and Ccr7 marker genes, which we herein refer to as Fcer2a B cells, and one cluster highly expressing Vpreb3 and Spib genes that we refer to as Vpreb3 B cells, one cluster with Crip1 and S100a6 high expression (Crip1 B cells) and lastly, one cluster of B cells highly expressing Zcwpw1, Lgals1 and Adm (Zcwpw1 B cells) (Supplementary Figure 1B). Classical monocytes are identified by a combination of markers, Ly6c2+, Ccr2+, Sell+ and Csf1r+ whereas monocyte-derived DC are Ly6c2+, Ccr2+, Cd209+ and Syngr2+ (Figure 1C).

Transcriptomic changes of the same cell type with age

We used Seurat to identify differential gene expression within clusters with age. Only genes expressed in at least 10% of cells in each cluster and age group were analyzed. We observed a general trend of the upregulation of genes involved in “Antigen processing and presentation” and “Chemokine signaling” pathways with age in naïve Cd4 and Cd8 T cells and Vpreb3 B cells. NK cells from old mice also exhibit increased expression of genes in the “Chemokine signaling” pathway (Figure 2A). In addition, we also observed an upregulation of genes in “Antigen processing and presentation” in macrophages, Crip1 B cells, and Fcer2aB cells. Crip1 B cells and monocytes also showed an upregulation of genes in the “Oxidative phosphorylation” pathway. In contrast, we observed a trend of downregulated genes involved in “Cytoplasmic Ribosomal Proteins” and “Ribosome” pathways in several cell types, naïve Cd4 and Cd8 T cells, Crip1 B cells, and monocytes, with age (Figure 2A).

Cell type composition with age

B cells comprised the largest proportion of immune cells isolated from the peripheral blood (57.4% in young mice and 70.9% in old mice) (Figure 3A). In young mice, 30.38% of the isolated cells are T cells, 4.66% are monocytes, 4.16% are NK cells, 0.6% are DC, 0.39% are basophils, 0.27% are macrophages, and 0.01% are Mk cells. In old mice, 8.41% of the cells are T cells, 10.84% are monocytes, 4.03% are Mk cells, 1.34% are NK cells, 1.09% are DC, 0.44% are macrophages, and 0.26% are basophils. Although red blood cell lysis was performed on the samples, we still retrieved 2.06% of RBC from the young mice and 2.68% from old mice. Using the same technique in extracting the cells from young and old mice, we observed a few changes to the composition of cell types with age. First, we detected a significantly lower percentage of T cells and NK cells and higher percentage of B cells and Mk cells in old mice compared to young mice (Figure 3B). Second, within the subsets of B cells, we observed a decrease in the percentage of Fcer2a B cells and increased Crip1 B cells in old mice (Figure 3C). We did not detect any significant changes to the subsets of T cells (Figure 3D).

Immunosenescence

We identified one cluster that consists of only cells from old mice (labeled as “Old T” in Figure 1B), and these cells expressed Cd3e, the marker of T cells. In addition, this cluster also expressed Cd40lg and Tnfsf8 compared to other T cell clusters, suggesting that it consists of activated T cells (Supplementary Figure 2). Pathway analysis of genes significantly upregulated between this cluster and other T cells clusters showed enrichment of “Antigen processing and presentation” and “NF-kappa B signaling pathway” (Figure 4A). In addition, this cluster expressed significantly higher Bcl2 expression (p=3.9x10^-15) (Figure 4B) than other T cells clusters. No other pro-apoptotic gene (Bax, Bak1, Bad, Bid, Bik) or anti-apoptotic gene (Bcl2, BclxL, Bcl2l1, Nfkb1, Cflar) was significantly differentially expressed. Notably, it also showed significantly higher expression of Cdkn2a and lower expression of Cd28 than all other clusters (p=5.38x10^-40 and p=6.26x10^-8, respectively) (Figure 4C and Supplementary Figure 2).

Immunosenescence has been widely implicated in aging. Cellular senescence, an irreversible cell cycle arrest biological process, was first discovered in fibroblast, and senescent cells can be identified by a collection of markers, which include, for example, upregulation of Cdkn1a and Cdkn2a. We observed a significant increase in the percentage of old blood cells (6.3%) expressing Cdkn1a and/or Cdkn2a compared to young cells (Fisher’s Exact Test p<2.2x10^-16) (Figure 5A). However, we did not detect any cell type with significant differences in the expression of these two genes between ages. Still, we did observe a moderate (p=0.066) increase in the number of Cdkn1a expressing cells in naïve Cd4 T cells (Figure 5B).

We also detected a cluster of old and young proliferating cells that consists of T and B cells (Figure 5C and Supplementary Table 1). The cells from old mice in this cluster are enriched in B cells but not T cell markers, whereas we saw a similar enrichment of T and B cells in young mice. This suggests a decreased number of proliferating T cells in the old mice.

Use of animals

C57BL/6 female 4- and 24 month mice were obtained from the National Institute of Aging (NIA). They were fed ad libitum and kept in standard housing conditions, and all procedures were approved by the Brown University Institutional Animal Care and Use Committees (IACUC) committee.

Isolation of immune cells from the peripheral blood

Blood samples were drawn from the heart of four young and four old mice. Samples from two mice within each age group were pooled, resulting in two pooled young and two pooled old samples. Pooling allowed us to increase the number of mice used in this study and simultaneously reduce the variance due to potential outliers. One limitation is that each individual sample does not represent an individual mouse. Subsequently, the pooled samples were diluted 1:1 with PBS+2% FBS and loaded into SepMate-15™ tubes with 4.5 mL of Lymphoprep™ with a density of 1.077g/mL (StemCell Technologies). Cells were centrifuged at 1200x g for 20min at 4° C. The top layer was collected and centrifuged at 300xg for 8 min. Supernatant was removed, and cells were resuspended in PBS + 2% FBS, and this step was repeated. Subsequently, red blood cell lysis was performed using the Red Blood Cell Lysis Solution (Miltenyi Biotec) following the manufacturer’s protocol, that includes the washing step.

Single-cell library construction

Single-cell RNA-seq protocol was performed using the Chromium™ Single Cell 3’ reagent kit v2 chemistry, and cells were loaded on a GemCode Single Cell Instrument (10x Genomics). Approximately 5000 single cells were targeted from each sample. Libraries were sequenced on Illumina HiSeq 2500 with the custom configuration of read 1 (26bp) and read 2 (98bp), i7 index (8bp), and i5 index (0bp).

Single-cell RNA-seq alignment, UMI counting, and analysis

The Cell Ranger Single Cell Software Suite 2.1.0 was used to perform single cells demultiplexing and UMI counting (https://support.10xgenomics.com/single-cell-gene-expression/software/overview/welcome). The transcriptome reference used was mm10. Subsequently, the duplicates from young and old mice, respectively, were aggregated using Cellranger aggr. Cells with more than 500 genes detected and less than 10% of mitochondrial reads were included in the downstream analysis using Seurat 2.3.0. Young and old UMI counts were merged using the MergeSeurat function. Default parameters of Seurat were used in the analysis unless otherwise stated. 1080 highly variable genes were used as an input for PCA. T-SNE projection and clustering analysis (dims=1:30 and resolution=0.4) were performed using Seurat. Markers genes for each cluster were found using the FindConservedMarkers function, and the changes with age within each cell type were identified using the FindMarkers function.

Number and percent of cells in each cell type cluster for each age: Basophil: N = (30 young, 18 old), percent = (0.4% young, 0.3% old); Crip1 B: N = (453 young, 2281 old), percent = (5.8% young, 33.4% old); Cytotoxic T: N = (82 young, 118 old), percent = (1.1% young, 1.7% old); DC: N = (46 young, 74 old), percent = (0.6% young, 1.1% old); Fcer2a B: N = (3647 young, 1914 old), percent = (47.0% young, 28.0% old); Macrophage: N = (21 young, 30 old), percent = (0.3% young, 0.4% old); Memory T: N = (247 young, 16 old), percent = (3.2% young, 0.2% old); Mk: N = (1 young, 274 old), percent = (0.0% young, 4.0% old); Mono: N = (360 young, 737 old), percent = (4.6% young, 10.8% old); NK: N = (321 young, 91 old), percent = (4.1% young, 1.3% old); Naive Cd4 T: N = (1261 young, 193 old), percent = (16.3% young, 2.8% old); Naive Cd8 T: N = (755 young, 179 old), percent = (9.7% young, 2.6% old); Old T: N = (0 young, 66 old), percent = (0.0% young, 1.0% old); RBC: N = (159 young, 182 old), percent = (2.0% young, 2.7% old); T and B: N = (39 young, 31 old), percent = (0.5% young, 0.5% old); Vpreb3 B: N = (335 young, 249 old), percent = (4.3% young, 3.6% old); Zcwpw1 B: N = (2 young, 376 old), percent = (0.0% young, 5.5% old).

Statistical analysis

All Student’s t-tests were performed in R, and p-values were adjusted using the Benjamini-Hochberg method. Barplots are represented as means with SEM. The top and bottom bounds of boxplots correspond to the 75 and 25th percentile, respectively. At the single-cell level, the statistical analyses were carried out with the cells as replicates. At the pooled samples level, two pooled samples are available for each age group; hence there are two replicates per group.

Data availability

Single-cell RNA-seq of old and young peripheral blood duplicates are accessible through GEO with the accession number GSE120505.