Single nuclei profiling identifies cell specific markers of skeletal muscle aging, frailty, and senescence

Abstract

Aging is accompanied by a loss of muscle mass and function, termed sarcopenia, which causes numerous morbidities and economic burdens in human populations. Mechanisms implicated in age-related sarcopenia or frailty include inflammation, muscle stem cell depletion, mitochondrial dysfunction, and loss of motor neurons, but whether there are key drivers of sarcopenia are not yet known. To gain deeper insights into age-related muscle loss, we performed transcriptome profiling on lower limb muscle biopsies from 72 young, elderly, and frail human subjects using bulk RNA-seq (N = 72) and single-nuclei RNA-seq (N = 17). This combined approach revealed changes in gene expression that occur with age and frailty in multiple cell types comprising mature skeletal muscle. Notably, we found increased expression of the genes MYH8 and PDK4, and decreased expression of the gene IGFN1, in aged muscle. We validated several key genes changes in fixed human muscle tissue using digital spatial profiling. We also identified a small population of nuclei that express CDKN1A, present only in aged samples, consistent with p21^cip1-driven senescence in this subpopulation. Overall, our findings identify unique cellular subpopulations in aged and sarcopenic skeletal muscle, which will facilitate the development of new therapeutic strategies to combat age-related frailty.

Keywords: aging; muscle; sarcopenia; senescence; transcriptomics.

Figure 1

Bulk RNA-seq identifies major gene expression changes in muscle with age. (A) Principal component analysis (PCA) of bulk young, old and frail skeletal muscle. (left) Young (less than 20 years old) in blue, old (more than 65 years old) in red. (right) Young (blue), old (red), frail subjects (green). (B) Volcano plot of expression changes in old vs. young muscle. Labelled top 30 by abs (logFC) × -loglO (p-value). (C) Log (CPM) of MYH8, COL19Al, MTRNL8, CDKNlA, CDKN2B, AREG in young (green), old (blue) and frail subjects (red). Boxplot shows 25% percentile, 75% percentile and median. Stars were added when significant compared to young healthy (q < .01). (D) Number of DEGs per comparison. (E) Pathway analysis of dysregulated genes with age using KEGG, GO database (GSEA).

Figure 2

Single-nuclei sequencing reveals 7 clusters of unique cell types, and differential gene expression with age. (A) Uniform Manifold Approximation and Projection (UMAP) of 5′ single nuclei sequencing of human muscle. All samples are shown, after data normalization and Louvain clustering. (B) Top 20 differentially expressed genes (DEG), in old vs. young samples. All cells from all cell types are used in this test. Wilcoxon test, top 20 DEGs by logFC. (C) Expression of PDK4 and IGFNl in young and old samples.

Figure 3

Common and cell-type specific gene expression changes with age. Significant differentially expressed genes (DEG) in old versus young samples. A Wilcoxon test was performed for each gene in each cell type between samples, with a logFold-Change (logFC) threshold of .25, and False-Discovery Rate (FDR) <1%. Red is upregulated with age, blue is downregulated. (A) All DEGs are shown by cell type. (B) Top 20 DEGs are shown by cell type, ranked by absolute logFC.

Figure 4

mRNA translation, gamma interferon and complement cascade are upregulated in selective cell types with aging. Pathway analysis of top 100 up-regulated and top 100 down-regulated genes with age in each cell type. GO, KEGG, Reactome pathways were queried. Over-representation was assessed using a hyper-geometric test at FDR 1%. (A) Upregulated with age. (B) Downregulated with age.

Figure 5

Identification of a small population of senescent cells in the fast skeletal muscle. (A) Subtypes of fast-skeletal muscle cells (UMAP, all samples). (B) Cluster 5 is circled, with expression of LRRK2, CDKN1A, MYH8, COL19A1 and TNNT3. (C) Difference in proportions between young and old for all subtypes. Significance of the t-test between young and old is shown at the top of 5C.

Figure 6

Reorganization of muscle fibers with age revealed by spatial transcriptomics. (A) Young muscle fibers, several ROls are shown in yellow delineating individual sections of distinct fibers. Desmin (blue), Syto83 (green), aSMA (yellow), CD68 (red). (B) Differentially expressed genes in old versus young spatial profiled muscle. (C) log (counts) of top differentially expressed genes.

Figure 7

Validation of senescent markers in cultured human muscle cells. Quantitative PCR (qPCR) of CDKN1A, MYH8, COL19A1, LRRK2, EDA2R and PDK4 after 7 days of incubation in senescent vs. non-senescent cells. Senescence was induced using Doxorubicin in a cell line of Myogenic Progenitor cells (un-differentiated cells, left) and Myotubes (differentiated HSMMs, right). Expression is shown relative to Actin. 3 replicates in each condition/gene.