Using systems biology to explore temporal changes in human fibroblast cellular senescence.

Abstract

Cellular senescence is a complex phenotype characterised by permanent cell cycle arrest and a senescence-associated secretory phenotype, which includes growth factors and inflammatory cytokines. While primarily thought of as a tumour suppressive mechanism, senescence also plays roles in wound healing and embryogenesis. Senescent cells are normally transient but accumulate with age due to dysregulated immune clearance, contributing to low-grade chronic inflammation and age-related diseases. Characterising senescent cells is challenging due to phenotype heterogeneity, and the temporal dynamics of senescence remain poorly understood. This thesis employed an integrated approach to investigate human fibroblast senescence at transcriptomic and protein levels. A systematic review identified 119 transcriptomic datasets on human fibroblast senescence, forming the database SenOmic, publicly hosted online and allowing users to filter by variables of interest such as gene and timepoint. Computational modelling of key selected proteins in DNA damage-induced senescence (DDIS) and oncogene-induced senescence (OIS) was also performed, including knockdown interventions. Analysis of SenOmic reinforced the challenges of defining a universal geneset across cell lines and senescence types. However, 28 genes were significantly up- or downregulated across DDIS, OIS, replicative senescence, and bystander senescence compared to proliferating controls, with only one gene present in the KEGG senescence pathway. Distinct phenotypes were also observed, including significantly stronger p53 signalling in DDIS compared to OIS, clustering of samples by time, and significant upregulation of genes involved in protein secretion between days 5-7 in gene set enrichment analysis. Protein level modelling demonstrated the importance of multi-macrolevel analysis, highlighting post-translational modifications and network-wide effects of knockdowns. In conclusion, while unique universal senescence biomarkers remain challenging to identify, conventional senescence markers follow predictable profiles, distinct phenotypic differences exist across timepoints and senescence types, and further interrogation of resources like SenOmic with an established framework provides a valuable means to enhance our understanding of senescence.