Equilibrium and Non-Equilibrium Molecular Dynamics Simulations Provide Potential Mechanisms for the Loss of Ligand Binding to αIIbβ3 Mutants Affecting the Ligand-Induced Metal Binding Site (LIMBS).

Although the role of the β3 MIDAS metal ion in ligand binding to αIIbβ3 is well established, serving as the site of interaction of the ligand Asp residue, the role of the nearby LIMBS metal ion is less well defined. Previous studies suggested a role for the LIMBS in ligand binding. We confirmed this by showing that HEK293 cells expressing normal αIIbβ3 adhered to both immobilized fibrinogen and the RGD-containing venom echistatin in the presence of either Mg++/Ca++ or Mn++, whereas two different αIIbβ3 LIMBS mutants (β3 N215A and D217A) failed to adhere to either protein. In addition, we found that both mutations also increased the binding of mAb AP5, which recognizes a ligand-induced binding site (LIBS) in the β3 PSI domain (normal 7±4% vs N215A 46±12% and D217A 41±20% of mAb anti-αIIb (HIP8) binding; mean±SD, n=6, p<0.05 for both), indicating that the mutations caused allosteric changes in the conformation of the receptor. To define the mechanism(s) by which the LIMBS mutants affect ligand binding, we carried out equilibrium and non-equilibrium (steered) molecular dynamics (MD) simulations of the cyclic peptide ligand eptifibatide in complex with either the fully hydrated normal αIIbβ3 integrin headpiece (PDB 1TY6) or the equivalent β3 D217A mutant, with and without the LIMBS metal ion. Simulations were carried out using the GROMACS package with the OPLS all-atom force-field. During the simulation, the hybrid domain of the D217A mutant demonstrated greater structural fluctuations than the normal αIIbβ3. Although Craig et al. have reported the appearance of a new contact between the RGD peptide ligand Asp carboxyl and the LIMBS metal ion in αVβ3 after 10 ps of a 1 ns simulation, we did not observe the appearance of such an interaction between the eptifibatide carboxyl and the normal αIIbβ3 LIMBS metal ion even after 20 ns. We did, however, observe such an interaction with the LIMBS metal ion in the D217A mutant. This interaction was facilitated by the movement of the LIMBS ~ 2 Å closer to the MIDAS, and was accompanied by rearrangements of the LIMBS coordinating residues D158 and N215. When the D217A mutant simulation was performed in the absence of the LIMBS metal ion, changes in the orientation of E220 were also observed. The D217A mutant demonstrated increased fluctuations in the C177–C184 specificity-determining loop (SDL), which has been implicated in ligand binding, and decreased fluctuations in K209. Steered MD were used to investigate the pulling forces required to unbind eptifibatide from its binding site. Notably, although the unbinding force decreased modestly when the LIMBS metal ion was removed, it required removal of both the LIMBS and MIDAS metal ions to effect a marked reduction in unbinding force. The binding free energies of the association of the αIIb and β3 subunits were also calculated, and the D217A mutant in the presence of the LIMBS metal ion demonstrated much tighter binding than normal integrin αIIbβ3 (ΔGb −162±6 vs −119±6 Kcal/mol; mean±SD; n=500). We conclude that the LIMBS plays a crucial role in ligand binding to αIIbβ3, perhaps by virtue of its effects on the coordination of the MIDAS, the accentuated mobility of specific domains (e.g., the SDL and the hybrid domains), and/or the number and strength of contacts between αIIb and β3.