Measurable Metrics of Mesenchymal Stem Cell Aging

The aim of the study is to analyze the manifestation of selected cellular senescence markers on the models of replicative senescence, stress-induced senescence, and chronological aging of human mesenchymal stem cells and to study the feasibility of predictive models for assessing the age and duration in vitro cultivation based on the transcriptomic data and investigation of cell morphology.

In the study, the dynamics of expression of individual genes encoding key regulators of cellular aging across various models of cellular senescence, as well as telomere length were investigated by real-time PCR. The analysis of the high- throughput transcriptome sequencing datasets of mesenchymal stem cells from the donors of different ages has been performed. Using regression methods, predictive models based on transcriptomic data were developed to estimate chronological age and the duration of in vitro cultivation. Using microscopy methods and subsequent image analysis by machine-learning algorithms, morphological alterations associated with cellular senescence have been explored and segmentation neural network model has been created for extracting nuclear morphology parameters and classification of the cells based on the duration of cultivation in vitro.

CDKN1A, LMNB1, HMGB2 genes demonstrated reproducible similar dynamics on the models of replicative or stress-induced senescence and chronological aging of mesenchymal stem cells. The expression profile of the senescence-associated inflammatory phenotype components was variable in different models of cell aging. The analysis of mesenchymal stem cell transcriptomes from the donors of various ages revealed considerable donor-dependent heterogeneity of the cells, which complicates the development of precise transcriptome data-based predictive models. Investigation of the changes in the telomere length has demonstrated its applicability for assessing the dynamics of replicative senescence in vitro. The developed segmentation neural network model allowed for detecting senescence-associated dynamics of nuclear morphology alterations in the process of replicative aging.