Metabolomics in drug discovery : cancer cells metabotyping to predict the mode of action of anticancer agents

Abstract

In the development of new anticancer drugs, the identification of the mode of action (MoA) remains a significant challenge. This thesis demonstrates the integration of metabolomics into the drug discovery pipeline to predict the MoAs of novel anti-proliferative drug candidates, specifically for human prostate cancer cells (PC-3). By studying 38 drugs known to affect 16 key processes of cancer metabolism, we profiled low molecular weight intermediates of the central carbon and cellular energy metabolism (CCEM) using LC-MS/MS. These metabolic patterns revealed distinct MoAs, enabling the accurate prediction of MoAs for novel agents through machine learning algorithms. The methodology was further validated by transferring MoA predictions to two other cancer cell models, breast cancer and Ewing’s sarcoma, confirming that correct MoA predictions across different cancer types are feasible, albeit with some reduction in prediction quality.