Identification of Interacting Hot Spots in the αIIb Extracellular Stalk By Computational Alanine Scanning

Binding of macromolecular ligands like fibrinogen to the active conformation of the integrin αIIbβ3 mediates platelet aggregation. Because platelets circulate in a fibrinogen-rich milieu, the αIIbβ3 on circulating platelets is constrained in an inactive conformation by intramolecular interactions involving the cytoplasmic, transmembrane, and stalk domains of αIIb and β3 to prevent the spontaneous formation of platelet aggregates. Following platelet stimulation, αIIbβ3 undergoes a global rearrangement during which the αIIb and β3 cytoplasmic, transmembrane, and stalk domains separate and the αIIbβ3 headpiece opens to expose its ligand binding site. Protein-protein interfaces, such as those of the αIIbβ3 stalks, are usually large complementary surfaces in which specific side chains, representing energetic "hot spots", contribute disproportionately to the binding free energy. Previously, we reported the use of a computational method, comprehensive interface design, as a way to control the direction of protein interactions by introducing changes in protein structure that destabilize undesired interactions. This enabled us to identify energetic hot spots in the β3 stalk, mutation of which caused constitutive αIIbβ3 activation. Here, we have extended our design strategy to the αIIb stalk. We used the Rosetta alanine scanning algorithm and the available crystal structure of the αIIbβ3 heterodimer to identify alanine substitutions in the αIIb region extending from residues 602-959 that would be predicted to destabilize the resting αIIbβ3 stalk interface. The functional consequences of the predicted mutations were confirmed by introducing them into wild type αIIb, expressing the mutant αIIb together with wild type β3 in Chinese hamster ovary (CHO) cells, and measuring fibrinogen binding to αIIbβ3 in the absence and presence of dithiothreitol by flow cytometry. We identified 5 alanine substitutions predicted to destabilize the αIIbβ3 stalk interface by >1.0 kcal/mol, 5 alanine substitutions predicted to destabilize the interface by 0.3-1.0 kcal/mol, and 2 neutral alanine substitutions having no effect on the interface as negative controls. Four of the 5 substitutions predicted to be the most destabilizing were located in the αIIb calf-2, rather than the calf-1, domain, and involved residues R751, N753, F755, and E785. It is noteworthy in this regard that the 5 destabilizing β3 substitutions we previously reported were located in the distal β3 stalk. To confirm the predictions of the computational alanine scanning, we introduced the four most destabilizing calf-2 substitutions and the two neutral controls into αIIb and co-expressed them with wild type β3 in CHO cells. There was little to no expression of αIIb_R751Aβ3, implying that the presence of R751 is important for correct αIIb folding and/or αIIbβ3 assembly. By contrast, each of the other predicted destabilizing substitutions expressed and caused constitutive αIIbβ3 ligand binding activity. Moreover, as predicted, αIIbβ3 containing each of the neutral αIIb substitutions was inactive until the CHO cells were incubated with dithiothreitol. Thus, these results confirm the utility of our computational approach for identifying functionally significant regions in αIIbβ3. Further, they demonstrate that the interface between the distal αIIb stalk and the distal β3 stalk, much like the interface between the αIIb and β3 transmembrane domains, plays an important role in regulating the equilibrium between the inactive and active states of αIIbβ3.