Network Pharmacology and Machine-Learning in the Identification of Senotherapeutics

Abstract

Introduction: Aging is a complex biological process that involves the accumulation of damage in the organism influenced by internal and external factors, which has led to identify the so-called Hallmarks of Aging, among which cellular senescence stands out (López-Otín et al. 2013). Which is characterized by cell cycle arrest and secretion of inflammatory molecules (SASP), is a key process that is triggered by DNA damage and cellular dysfunctions, these cells accumulate as we age and contribute to the development of age-associated diseases (Huang et al. 2022a). Therapies known as senotherapies seek to address this process by eliminating senescent cells (senolytic) or reducing the negative effects of SASP (senomorphic). Examples include Metformin, Navitoclax and Quercetin-Dasatinib, the latter showing promising effects in preclinical trials. On the other hand, network pharmacology and machine learning (ML) are revolutionizing senolytic (Smer-Barreto et al. 2023) and other new drug discovery. Therefore, in this project we set the objective of identifying compounds with potential senolytic activity by means of network pharmacology and ML. Methods: To map by network pharmacology the proteins involved in cellular senescence and search the Comparative Toxicogenomics Database (CTD) using supervised machine-learning models (RFC, SVM, KNN) to identify new candidate compounds for sinotherapeutics, optimizing the drug discovery process by computational prediction based on chemical properties of the compounds; as well as validate human fibroblast and myofibroblast cell culture some candidate sinotherapeutic using a model of cell senescence induced by H₂O₂ at concentrations of 100 µM, 200 µM, 300 µM, 400 µM, 600 µM, 800 µM and 1000 µM. Results: We identified by network pharmacology and ML 270 candidate potencial senotherapeutics with desired druglikeness properties. Among which flavonoids, terpenes and azoles are the most abundant. In addition, a web tool was created for their dissemination, available free of charge at: (https://gcoixc-laboratorio0de0bioinform0tic-inger.shinyapps.io/Senotherapeutics_Shiny/). On the other hand, they tested different concentrations of H₂O₂ (100 µM, 200 µM, 300 µM, 400 µM, 600 µM, 800 µM and 1000 µM). It was identified that in a 3-hour exposure with H₂O₂ at a concentration of 600 µM the viability of fibroblasts dropped to 51%, while at a concentration of 600 µM the viability of myofibroblasts dropped to 63%. Conclusion: Network pharmacology and ML offers a comprehensive and robust approach for rational drug discovery, it allowed us to screen a set of candidate compounds for sinotherapeutics. Furthermore, preliminary functional validation in a model of H₂O₂-induced senescence provides a solid experimental basis for future testing.