Triple-negative breast cancer (TNBC) is an aggressive and highly heterogeneous cancer subtype characterized by the lack of hormone receptors and HER2 expression, which limits targeted therapeutic options and contributes to high relapse and mortality rates. A feature of these tumors is the resistance to chemotherapy, strongly linked to cancer stem cell (CSC) populations and the tumor microenvironment (TME), particularly cancer-associated fibroblasts (CAFs), which promote tumor progression, extracellular matrix remodeling, and impaired drug penetration. Novel therapeutic strategies addressing these mechanisms are urgently needed. Here, we present a TNBC drug discovery project that relies on physiologically relevant 3D tumor models, including multicellular spheroids (MCS) composed of CSCs, differentiated tumor cells, and cancer-associated fibroblasts (CAFs), as well as patient-derived xenograft organoids (PDxOs), which recapitulate tumor heterogeneity and clinical drug responses. As an unexplored source of therapeutic molecules, we selected a subset of 5,000 microbial natural products (NPs) extracts with diverse origin from MEDINA’s collection. Antitumoral activity was assessed in the 3D models using High-content imaging (HCI) and m3DinAI HCS Drug Quest, a pipeline fully developed in MEDINA that integrates machine learning algorithms to classify 3D phenotypes using multiparametric morphological features.
This integrated screening platform enabled the identification of eight extracts with anti-TNBC activity in complex TME-mimicking systems, supporting the discovery of novel candidates with effect on TNBC. Active extracts were subsequently fractionated to identify the bioactive constituents, resulting in the proposal of six candidate molecular formulae. Overall, this strategy provides a robust framework for the identification of new therapeutic leads and advances the development of more biologically relevant drug discovery approaches for TNBC.
