Bailey McIntosh, Yen Chin Koay

Current Opinion in Chemical Biology

DOI: 10.1016/j.cbpa.2026.102726

Abstract

Natural product (NP)-inspired scaffolds occupy a privileged position in interface-centric drug discovery, as their preorganised three-dimensional architectures, multivalent contact surfaces, and covalent anchoring potential enable them to engage extended, topographically complex protein-protein and protein-nucleic acid interfaces. This review argues that modern platforms are now making these interactions systematically accessible, shifting NP drug discovery from occupancy-based inhibition toward interface stabilisation and induced proximity at targets previously considered undruggable. We evaluate how trillion-scale DNA-encoded libraries, RaPID-based macrocycle selection, and covalent chemoproteomics interrogate vast chemical spaces to identify compounds with non-classical binding modes relevant to proximity-driven therapeutic discovery, assessing both their demonstrated strengths and current limitations in the context of interface-centric therapeutics. We further discuss how AI-based tools support network prioritisation, interface hotspot prediction, and ternary complex modelling, while noting that computational predictions require independent experimental validation and do not yet replace biochemical evidence. We propose that systematic integration of NP-inspired scaffolds with modern discovery platforms and computational tools, represents the most productive current framework for targeting disease-relevant complexes previously considered inaccessible, expanding the boundary of druggability across oncology, cardiovascular, and metabolic medicine.