Evolutionary origin and structural ligand mimicry by the inserted domain of alpha-integrin proteins

Heterodimeric integrin proteins transmit signals through conformational changes upon ligand binding between their alpha (α) and beta (β) subunits. Early in chordate evolution, some α subunits acquired an "inserted" (I) domain, which expanded their ligand binding capacity but simultaneously obstructed the ancestral ligand-binding pocket. While this would seemingly impede conventional ligand-mediated integrin activation, it was proposed that the I domain itself could serve both as a ligand replacement and an activation trigger. Here, we provide compelling evidence in support of this longstanding hypothesis using high-resolution cryo-electron microscopy structures of two distinct integrin complexes: the ligand-free and E-cadherin-bound states of the αEβ7 integrin with the I domain, as well as the α4β7 integrin lacking the I domain in both a ligand-free state and bound to MadCAM-1. We trace the evolutionary origin of the I domain to an ancestral collagen-collagen interaction domain. Our analyses illuminate how the I domain intrinsically mimics an extrinsic ligand, enabling integrins to undergo the canonical allosteric cascade of conformational activation and dramatically expanding the range of cellular communication mechanisms in vertebrates.