NONPOLAR INTERACTIONS OF THROMBIN S' SUBSITES WITH ITS BIVALENT INHIBITOR - METHYL SCAN OF THE INHIBITOR LINKER

We have designed bivalent thrombin inhibitors, consisting of a nonsubs trate type active site blocking segment, a hirudin-based fibrinogen re cognition exosite blocking segment, and a linker connecting these segm ents. The inhibition provided by the bivalent inhibitors with various linker lengths revealed that a minimum of 15 atoms was required for si multaneous binding of the two blocking segments of the inhibitor to th rombin without significant distortion. The crystal structure of the in hibitors with a 16-atom linker showed some conformational flexibility in the linker portion which still lies deep in the groove joining the active site and the fibrinogen recognition exosite. Since the thrombin S' subsites are not well characterized, we designed a new strategy to search for possible nonpolar interactions between the linker and the thrombin S' subsites. This strategy, the ''methyl scan'', is based on the incorporation of a methyl side chain at each atom position of the linker by using sarcosine, D,L-alanine, D,L-3-aminoisobutyric acid, or N-methyl-beta-alanine. The methyl groups on the second and the eighth atom positions of the linker, which correspond to the side chains of the P1' and the P3' residues, respectively, improved the affinity of t he inhibitors significantly. Further study of the stereospecificity sh owed that L-Ala at the P1' residue and D-Ala at the P3' residue prefer ably improved the affinity of the inhibitors 20- and 25-fold, respecti vely. Molecular modeling calculations using a methyl probe were also c arried out to identify favorable nonpolar interacting sites on the thr ombin surface. Two sites were identified in the vicinity of the P1' an d the P3' residues, supporting the validity of the methyl scan method. Thus, this study has improved our understanding of the interactions t aking place in this groove. In particular, we have been able to show t hat some specific structural features, such as hydrophobic complementa rity between the linker and the thrombin S' subsites, could be exploit ed and make these inhibitors trivalent.