Student Posters 51st Lorne Proteins Conference 2026

AI-Guided De Novo Protein Binder Design Targeting RXFP1: A Computational-to-Experimental Pipeline (#111)

Janik JC Clement 1 , Bradley BH Hoare 2 , Tim TL Lkhagvajargal 2 , Daniel DF Fox 1 3 4 , Ross RADB Bathgate 1 2 , Rhys RG Grinter 1 3 4
  1. University of Melbourne, Brunswick, VICTORIA, Australia
  2. Neuropeptide Receptor Group, The Florey, Melbourne, VIC, Australia
  3. Department of Microbiology, Biomedicine Discovery Institute,, Monash University, Melbourne, VIC, Australia
  4. Centre for Electron Microscopy of Membrane Proteins, Monash University, Melbourne, VIC, Australia

We developed and validated a comprehensive computational pipeline for designing high-affinity protein binders against the relaxin family peptide receptor 1 (RXFP1), a therapeutically relevant G protein-coupled receptor (GPCR) implicated in cardiovascular disease and fibrotic disorders. Using state-of-the-art AI algorithms, including RFdiffusion, BindCraft, and ProteinMPNN, we computationally generated 378 and 318 candidate binders, respectively (696 total). Through proprietary selection scripts implementing AlphaFold2-based metrics, structural compatibility analysis, and glycosylation site avoidance algorithms, we systematically ranked and filtered candidates to identify the top 48 BindCraft and 96 RFdiffusion designs for experimental testing.
Using optimized batch expression and His-tag purification protocols in E. coli, we successfully identified eight high-affinity antagonist candidates that demonstrated potent binding in bio-layer interferometry (BLI) assays and nanomolar-range competition in functional inhibition studies. In parallel, we designed agonistic binders for RXFP1 using BindCraft, and from over 518 in silico designs, we expressed 48 binders in vitro chosen based on the best Alphafold2 binding prediction metrics. We were able to show that 66% of agonist binder designs were interacting with the target in BLI screening assays and chose to assess five lead candidates for agonist activity assessment in competition and functional assays. We could confirm agonist activity for four of these binders in functional assays and strong nanomolar-range agonist activity for two of these binders, indicating that the pipeline is capable not only of blocking RXFP1 signaling but also of selectively stabilizing its active conformational states.
This work establishes a validated computational framework for rationally designing both antagonist and agonist biologics against challenging GPCR targets, with immediate therapeutic relevance for RXFP1-mediated cardiovascular repair and fibrosis modulation, and broader implications for LRR-containing GPCR drug discovery.