Among the most pharmacologically important proteins in human biology, G-protein-coupled receptors govern everything from pain perception and immune cell trafficking to metabolic regulation — yet they have resisted conventional protein-engineering approaches for decades. A new platform combining computational de novo design with high-throughput receptor-diversion microscopy now demonstrates that purpose-built miniproteins can achieve drug-like potency and selectivity across this historically intractable target class, potentially expanding the therapeutic toolkit well beyond small molecules and antibodies.
The work produced miniprotein agonists selective for receptors mediating itch and pain signaling, alongside antagonists targeting receptors implicated in cancer progression, diabetes, obesity, and migraine. Structural validation via cryo-electron microscopy confirmed that five receptor-bound miniprotein complexes matched their computational design models with high fidelity — a rare demonstration that predicted protein-receptor interfaces hold in atomic-resolution structures. Most notably, a de novo chemokine receptor antagonist mobilized hematopoietic stem and progenitor cells in living animals at levels comparable to plerixafor, a clinically approved mobilizing agent, while producing fewer adverse effects.
The significance here extends well beyond any individual target. GPCR-directed drugs constitute roughly 35% of all FDA-approved pharmaceuticals, yet nearly all are small molecules or peptides; biologics have largely failed in this space because standard antibodies cannot access the recessed, transmembrane-embedded binding pockets GPCRs present. Miniproteins — typically 40–80 residues — occupy a structural middle ground that small molecules cannot achieve and that antibodies overshoot. The receptor-diversion microscopy screening method is particularly notable as an enabling technology, allowing high-throughput discrimination of binding from functional activation. Key limitations to watch: all in vivo data derive from animal models, clinical translation requires extensive safety profiling, and manufacturability at scale for protein therapeutics remains a practical challenge. Still, this represents a potentially paradigm-shifting expansion of what computational protein design can deliver against membrane receptor biology.