Structure‐Based Design of Nonpeptidic Thrombin Inhibitors: Exploring the D‐Pocket and the Oxyanion Hole

Structure-activity relationships for new members of a class of nonpeptidic, low-molecular-weight inhibitors of thrombin, a key serine protease in the blood coagulation cascade, are described. These compounds, which originate from X-ray-structure-based design, feature a conformationally rigid, bi- or tricyclic core from which side chains diverge into the four major binding pockets (distal D, proximal P, recognition or specificity S1, and oxyanion hole O) at the thrombin active site (Fig. 1). Phenylamidinium is the side chain of choice for the S1 pocket, while the most active inhibitors orient an i-Pr group into the P-pocket (Table 1). The key step in the synthesis of the inhibitors is the construction of the central bi- or tricyclic scaffold by 1,3-dipolar cycloaddition of an in situ prepared azomethine ylide and an N-substituted maleimide (Schemes 1–3, and 8–10). One series of compounds was designed to explore the binding features of the large hydrophobic D pocket. This pocket provides space for lipophilic residues as bulky as benzhydryl groups. A new strategy was developed, allowing introduction of these sterically demanding substituents very late in the synthesis (Schemes 5 and 6). Benzhydryl derivative (±)-2 was found to be the most selective member (Ki (trypsin)/Ki (thrombin)=1200) of this class of nonpeptidic thrombin inhibitors, while the ‘dipiperonyl' analog (±)-3 (Ki=9 nM, 7.60-fold selectivity) displays the highest potency of all compounds prepared so far (Table 1). A second series of inhibitors features side chains designed to orient into the oxyanion hole and to undergo H-bonding with the backbone NH groups lining the catalytic site of the enzyme. Unfortunately, neither activity nor selectivity could be substantially improved by introduction of these substituents (Table 2). Presumably, the high degree of pre-organization and the rigidity of the tightly bound scaffolds prevents the new substituents from assuming a position that would allow favorable interactions in the oxyanion hole. However, the oxyanion hole and the S1′ pocket next to it were found to be capable of accommodating quite large groups, which leaves much room for further exploration.