Dissecting the Structural Rearrangement and Activation of αIIbβ3 Integrin with a Disulfide Bridge.

Platelet αIIbβ3 integrin undergoes extensive structural rearrangement upon activation. Crystallographic and high-resolution electron microscopic observation revealed that integrins could assume multiple conformers under physiological conditions. However, functional assignment of each conformer has not been clearly determined. It is postulated that transition from the bent to the extended conformer (switchblade-like movement) and widening of the angle between the βA and the β hybrid domains (swing-out of the hybrid domain) that accompanies the extension is the critical event for integrin activation. To examine the authenticity of this hypothesis, we created and characterized mutant αIIbβ3 of which conformation was fixed in a specific conformer with an artificially introduced disulfide bridge. A double cysteine mutation βV332C/βS674C was designed to ligate βA and βTD, thus it is expected to fix integrin in a bent conformer by stabilizing the βA/βTD interface. Another mutation αD319C/βV359C was designed to ligate α β-propeller and the β hybrid domains. This mutation is expected to prevent the swing-out of the hybrid domain and the possible α/β headpiece dissociation. When expressed in CHO cells, both mutants were unable to bind fibrinogen, unless the disulfide bridge had been disrupted by DTT treatment. The LIBS epitope expression in βV332C/βS674C was significantly impaired and did not respond to induction by RGD peptide. The binding of ligand-mimetic mAb OP-G2 was also significantly impaired. By contrast, αD319C/βV359C showed comparable LIBS epitope expression and OP-G2 binding as wild-type αIIbβ3. The RGD peptide induced LIBS epitope expression, albeit slightly weaker than wild-type. Consistently, the βG327C/βV419C mutation that was designed to ligate βA and β hybrid domains, thus fix integrin headpiece in closed conformer by preventing the swing-out of the hybrid domain without affecting the α/β head interface, specifically blocked fibrinogen binding without affecting the binding of anti-LIBS or OP-G2. These results indicate that βV332C/βS674C is absolutely inactive and incapable of undergoing structural rearrangements associated with activation, while the αD319C/βV359C and βG327C/βV419C are capable of doing so upon binding to ligand-mimetic RGD peptide. Disruption of the βA/βTD interface by introducing a bulky amino acid residue in the interface or by deletion of the CD loop in the βTD did not induce integrin activation. However, introducing a bulky N-glycan at βV332 or at βS674 in the βA/βTD interface significantly potentiated ligand binding. Notably, the effects of the two N-glycans were synergistic. In summary, these results suggest, 1) that the bent conformer represents an inactive conformer, 2) that the contribution of the putative βA/βTD interface interaction in restraining integrin in the bent conformer is minimal, 3) that increasing the distance between the βA and the βTD, i.e. decreasing the angle of the bend, makes integrin highly susceptible to activation, 4) integrin with closed headpiece represents low affinity conformer, but it is capable of undergoing structural rearrangement upon ligand binding. Taken together, our results are consistent with the switchblade model. Conformational change other than the swing-out of the hybrid domain may trigger the structural rearrangement from bent to extended conformer upon ligand binding.