Unraveling the diverse roles of cellular metabolites using advanced computational approaches

Abstract

Cellular metabolites, traditionally seen as intermediates in metabolic pathways, perform diverse roles that extend into cellular signaling, homeostasis, and disease progression. This study deciphers the multifaceted roles of metabolites as internal threats, signaling modulators, and regulators of complex cellular networks, employing a multidisciplinary approach that combines computational modeling, deep learning frameworks, and experimental validations. In one of the objectives, we developed a novel computational framework, Metabokiller, to predict the carcinogenic potential of metabolites by integrating biochemical properties such as oxidative stress induction, genomic instability, anti-apoptotic activity, and epigenetic alterations. By leveraging a robust ensemble learning approach, Metabokiller identified a range of endogenous metabolites with carcinogenic potential. Experimental validation of two predicted metabolites, 4-nitrocatechol and 3,4-dihydroxyphenylacetic acid, demonstrated their ability to induce genotoxicity, mutagenesis, and malignant transformation in yeast and human cell lines. These findings underscore the significant yet underexplored role of metabolites as internal carcinogenic threats, advancing the predictive capabilities of computational models for toxicology and drug safety. In another objective involving the field of olfactory and GPCR signaling, two advanced platforms, OdoriFy and EvOlf, were developed to decode ligand-receptor interactions with high precision. OdoriFyis an artificial intelligence-driven platform designed to decode the molecular interactions between human odorant receptors and their ligands. Through its four predictive engines, Odorant Predictor, Odor Finder, OR Finder, and Odorant-OR Pair Analysis, OdoriFy enables the systematic identification and characterization of odorant-receptor interactions, offering mechanistic insights through explainable artificial intelligence. Extending beyond odorant research, EvOlf, a unified deep learning framework, was developed to predict ligand interactions across a wide spectrum of GPCRs, including both odorant and non-odorant receptors. With 105,235 GPCR-ligand interaction data points spanning multiple species, EvOlf bridges evolutionary diversity and functional specificity, highlighting the role of GPCRs as key mediators of metabolite signaling and their potential as therapeutic targets. Further investigations into the regulatory roles of metabolites revealed their capacity to modulate GPCR signaling pathways. In Saccharomyces cerevisiae, the ability of endogenous metabolites, including zymosterol, lanosterol, and ubiquinone-6, to act as allosteric regulators of the pheromone receptor Ste2pwas explored. These metabolites were found to suppress pheromone-induced programmed cell death(PCD) by targeting the Ste2p-Gpa1p interface, a critical node in GPCR signaling. Experimental approaches, including growth kinetics, cell viability assays, shmoo formation, and MAPK activation consistently demonstrated the ability of these metabolites to attenuate Ste2p-mediated signaling. Site-directed mutagenesis of key residues at the Ste2p-Gpa1p interface further confirmed the mechanistic basis of metabolite-mediated regulation, underscoring their specificity and functional importance. This work takes a multi-faceted approach by integrating computational and experimental methodologies to uncover the diverse roles of metabolites in health and disease. It not only advances our understanding of how metabolites function as internal threats and signaling regulators but also highlights their therapeutic potential in targeting complex biological systems. By decoding the biochemical and molecular mechanisms underlying metabolite interactions, these works lay the groundwork for future studies aimed at exploiting metabolites as biomarkers, therapeutic agents, and tools for predictive toxicology, opening new avenues in cancer biology, chemosensory research, and GPCR-based drug discovery.