Variant-specific pharmacophoric shifts in glucagon-like peptide-1 receptor-orforglipron complexes revealed by Boltz-2 co-folding and membrane molecular dynamics.

BACKGROUND: Orforglipron (ORF), the first orally bioavailable non-peptide glucagon-like peptide-1 receptor (GLP-1R) agonist to complete Phase 3 clinical trials, binds a distinct extracellular vestibule pocket not engaged by peptide GLP-1RAs. How commonly GLP-1R single-nucleotide polymorphisms alter this unique binding mode at the molecular level remains uncharacterized. METHODS: We employed an integrative computational framework that combines Boltz-2 deep learning co-folding for variant complex generation, 500 ns all-atom molecular dynamics simulations in a 1-palmitoyl-2-oleoyl-sn-glycero-3-phosphocholine (POPC) lipid bilayer, MM/GBSA binding free-energy decomposition, and protein ligand hydrogen-bond occupancy analysis. Wild-type (WT) GLP-1R and four clinically annotated variants L260F3.53b (TM3), G168S (stalk), R131Q (ECD), and A316T5.46 (TM5) were characterized. RESULTS: All variants exhibited enhanced MM/GBSA binding (24-35% vs WT) despite conformational destabilization, as evidenced by elevated RMSD, altered free energy landscapes, and reorganized H-bond networks. H-bond occupancy trajectory analysis identified Lys1972.67 as the canonical pharmacophoric anchor in WT (79.8%), preserved in L260F (57.1%) and A316T (63.9%). However, Lys1972.67 was absent in G168S and R131Q, which substitute Tyr2052.75/Gln221 and Tyr202, respectively. G168S exhibited the strongest binding (-117.18 kcal/mol) through compensatory contacts, while displaying a diffuse free-energy landscape and conformational disorder. A316T exhibited bistable dynamics related to Y2423.45-T3165.46H-bond-mediated constitutive activation, while preserving canonical engagement. CONCLUSIONS: Binding energies and contact analysis describe variant-dependent ORF engagement. Binding enhancement alone is insufficient to characterize pharmacophoric quality. Lys1972.67 occupancy differentiates variants with canonical binding from those substituting it, offering a structure-based framework for pharmacogenomic associations. The loss of Lys1972.67 in G168S offers a testable structural hypothesis for the diminished glycemic response observed in carriers, which requires functional validation.