Molecular recognition with biological receptors: Structure-based design of thrombin inhibitors

Molecular recognition is at the center of biological function. Consequently, a profound understanding of the underlying nonbonding interactions is required to intervene in a rational way in biological processes. Such detailed knowledge can be gained in studies with designed artificial receptors or, more directly, with biological receptors such as the enzyme thrombin. X-ray structural information on this key enzyme in the blood coagulation cascade has guided the structure-based design of a class of active and selective non-peptidic, reversibly binding low molecular weight inhibitors. These compounds feature a conformationally rigid bi- or tricyclic core structure from which side chains diverge into the four major binding pockets (distal D, proximal P, recognition or selectivity S1, and oxyanion hole) at the thrombin active site. With their rigid central core, all inhibitors prefer similar modes of association to thrombin, and detailed information on the strength of individual intermolecular bonding interactions and their incremental contribution to the overall free enthalpy of complexation is generated in correlative binding studies. Phenylamidinium is the side chain of choice for the S1-pocket. Attempts to replace this group with less basic functional groups, which cannot undergo bidentate ionic H-bonding to the carboxylate of Asp189 at the bottom of this pocket, were unsuccessful. The P-pocket is occupied by an isopropyl group, in analogy to the natural substrate fibrinogen, which uses the side chain of a valine residue to fill this site. The large hydrophobic D-pocket was found to accommodate one and even two aromatic residues. Attempts to direct side chains bearing H-bond acceptor groups into the oxyanion hole are described. The most active inhibitor prepared in this investigation showed a K-i value for thrombin inhibition of 9 nM and a 800-fold selectivity for binding thrombin over trypsin.