Class A GPCR structures usually capture the kinetically preferred G protein without a universal geometric signature

Abstract Cryo-electron microscopy reveals how G protein-coupled receptors (GPCRs) engage their G proteins, and AI-based prediction now models many more such complexes. But does the captured partner indicate which signalling route a receptor prefers? Here I compare 207 curated receptor–G protein structural units from 164 class A receptors with published G-protein activation data. In 81% of the 94 units with kinetic data, the captured family was also the most rapidly activated. Structural scores, however, predicted activation of uncaptured families only modestly, and this did not replicate in independent assays. Receptor geometry recognised the captured family beyond close homologues (AUC 0.73 versus 0.59 for conserved sequence positions at 50% identity), but class-wide differences such as TM6 opening largely vanished within the same receptor. A stable complex therefore shows which partner a receptor can accommodate, not necessarily which it prefers. Resolving preference will require time-resolved kinetic experiments and simulations of the steps preceding complex formation.