Single-cell transcriptomic analysis uncovers diverse and dynamic senescent cell populations

Abstract

Senescence is a state of enduring growth arrest triggered by sublethal cell damage. Given that senescent cells actively secrete proinflammatory and matrix-remodeling proteins, their accumulation in tissues of older persons has been linked to many diseases of aging. Despite intense interest in identifying robust markers of senescence, the highly heterogeneous and dynamic nature of the senescent phenotype has made this task difficult. Here, we set out to comprehensively analyze the senescent transcriptome of human diploid fibroblasts at the individualcell scale by performing single-cell RNA-sequencing analysis through two approaches. First, we characterized the different cell states in cultures undergoing senescence triggered by different stresses, and found distinct cell subpopulations that expressed mRNAs encoding proteins with roles in growth arrest, survival, and the secretory phenotype. Second, we characterized the dynamic changes in the transcriptomes of cells as they developed etoposide-induced senescence; by tracking cell transitions across this process, we found two different senescence programs that developed divergently, one in which cells expressed traditional senescence markers such as p16 (CDKN2A) mRNA, and another in which cells expressed long noncoding RNAs and splicing was dysregulated. Finally, we obtained evidence that the proliferation status at the time of senescence initiation affected the path of senescence, as determined based on the expressed RNAs. We propose that a deeper understanding of the transcriptomes during the progression of different senescent cell phenotypes will help develop more effective interventions directed at this detrimental cell population.

Subcellular structure, heterogeneity, and plasticity of senescent cells

“…Different senescence‐inducing factors can also lead to SnCs with varying transcriptomes (Hernandez‐Segura et al., 2017; Tang et al., 2019; Wechter et al., 2023), characterizing the inter‐inducer SH. After irradiation‐induced DNA damage, for instance, a single cluster of SnCs considering their transcriptome was observed (Tang et al., 2019), while two subpopulations of SnCs were found in fibroblasts undergoing senescence induced by replicative stress or etoposide‐induced DNA damage (Tang et al., 2019; Wechter et al., 2023; Wiley et al., 2017).…”

“…Different senescence‐inducing factors can also lead to SnCs with varying transcriptomes (Hernandez‐Segura et al., 2017 ; Tang et al., 2019 ; Wechter et al., 2023 ), characterizing the inter‐inducer SH. After irradiation‐induced DNA damage, for instance, a single cluster of SnCs considering their transcriptome was observed (Tang et al., 2019 ), while two subpopulations of SnCs were found in fibroblasts undergoing senescence induced by replicative stress or etoposide‐induced DNA damage (Tang et al., 2019 ; Wechter et al., 2023 ; Wiley et al., 2017 ). Indeed, the degree of heterogeneity seems to depend on the senescence‐inducing stress, since the transcriptomes of cells induced to senescence by different damage‐inducing agents, although different, are more similar among them compared to RS (Wechter et al., 2023 ).…”

“…After irradiation‐induced DNA damage, for instance, a single cluster of SnCs considering their transcriptome was observed (Tang et al., 2019 ), while two subpopulations of SnCs were found in fibroblasts undergoing senescence induced by replicative stress or etoposide‐induced DNA damage (Tang et al., 2019 ; Wechter et al., 2023 ; Wiley et al., 2017 ). Indeed, the degree of heterogeneity seems to depend on the senescence‐inducing stress, since the transcriptomes of cells induced to senescence by different damage‐inducing agents, although different, are more similar among them compared to RS (Wechter et al., 2023 ). Different senescence inducers can also lead the same cell type to produce a SASP with different compositions and functions.…”

“…Adding complexity to the biology of SnCs, they also vary their transcriptome over time, suggesting a dynamic state (González‐Gualda et al., 2021 ; Hernandez‐Segura et al., 2017 ). The intra‐popSH increases from Days 0 to 10 after DNA damage induced by etoposide, leading to two different senescence programs, one characterized by classical senescence markers like p16, and the other by RNA splicing and long non‐coding RNAs (Wechter et al., 2023 ). Likewise, during RS or DDIS, cells display a greater degree of variability in their transcriptome compared to the quiescent state (Hernandez‐Segura et al., 2017 ; Wiley et al., 2017 ).…”

“…Directly or indirectly, these pathways modulate the activity of Rb‐E2F complex, thus controlling the cell fate (Fujimaki & Yao, 2020 ). Other players controlling the cell decision between proliferation, quiescence, or senescence are p21 levels (Hsu et al., 2019 ) and proliferation status (i.e., if the cell is cycling or not) when senescence initiates (Wechter et al., 2023 ).…”