A complete RXFP1-relaxin interaction model unlocks the design of potent mini-protein modulators

Abstract

Relaxin family peptide receptor 1 (RXFP1) is a multi-domain GPCR with compelling therapeutic potential, yet uncertainty surrounding the mechanism of its activation by the hormone H2 relaxin has hindered the development of selective modulators. Here, we combine deep learning based structural modelling with de novo protein design to overcome this barrier. We generate a high-confidence structural model of the RXFP1-relaxin complex that is strongly supported by existing biochemical and functional evidence. This model reveals that relaxin engagement stabilises the RXFP1 extracellular linker, thereby triggering receptor activation. Guided by this model, we design mini-protein modulators that either block linker stabilisation or enforce it and induce an active receptor geometry. These molecules act as potent, selective RXFP1 antagonists or agonists, achieving low-nanomolar activity in both engineered and endogenously expressing cell lines despite adopting folds unrelated to relaxin. Together, these findings define the mechanistic basis of RXFP1 signalling, establish the first de novo agonists and antagonists of this receptor, and demonstrate how AI-enabled modelling and design can target structurally complex GPCRs previously inaccessible to structure-guided drug discovery.