Library thrombin inhibitors

For a whole decade a research group at Hofmann-LaRoche AG tried, without success, to find suitable thrombine inhibitors by the coventional methods. But only in 1995 Weber et al." discovered two such desired products, 23a and 23b (Scheme 1.9), when they used libraries of 4-CR products for their systematically planned search, which also included mathematically oriented methods. 

Linusson, a., Gotteries, )., Olsson, T., Oernskov, E., Folestad, S., Noeden, B., and Wold, S. Statistical molecular design, parallel synthesis, and biological evaluation of a library of thrombin inhibitors. /. Med. Chem. 2001, 44, 3424-3439. 

In another example of combinatorial parallel chemistry, we have recently used the Ugi three-component reactions (Ugi 3-CR) to construct a library of 16,840 protease inhibitors (25). It has been demonstrated previously that the Ugi-3CR reaction provides a useful chemical scaffold for the design of serine protease inhibitors N-substituted 2-substituted-glycine /V-ary 1/alky 1 -amidcs have been identified that are potent factor Xa, factor Vila, or thrombin inhibitors. The three variable substituents of this scaffold, provided by the amine, aldehyde, and isonitrile starting materials, span a favorable pyramidal pharma-cophoric scaffold that can fill the S1, S2, and S3 pockets of the respective protease. This library was screened against five proteases (factor Xa, trypsin, uro-... 

Figure 6. Amines that were used for the 15360member thrombin inhibitor library.

More unique are the xanthene and cubane cores (Figure 5) chosen by Rebek for the solution-phase synthesis of highly pooled combinatorial libraries in a single step. By iterative deconvolution, a thrombin inhibitor was identified from this complex collection. Similarly, workers at Isis reported the functionalization of a novel polyazacyclophane core, which led to the discovery of antibacterial leads. Clearly, in both these cases, the hits are structurally unrelated to common drug-discovery scaffolds. 

Pooling the combined research in the thrombin area provides munerous libraries of inhibitors with diverse structures and a range of affinities for thrombin spanning 9 orders of magnitude (millimolar to picomolar). Thrombin, in contrast to the other coagulation proteinases shows only a small scale of allosteric changes [97,98]. 

Thrombin inhibitors (127), where Bohm et al. used LUDI to dock and score computationally available primary amines and then score the virtual library generated from benzaldehydes with the top-scoring hit. 

A published combinatorial library of thrombin inhibitors [133] is examined by means of different computational methods in this section. The library is enumer-... 

Fig. 24 Schematic set-up of the virtual library of thrombin inhibitors, and the three sublibraries.

Three scaffolds (an amide, a sulfonamide, and a urea) are decorated with the same variable residues, R. The educt libraries include 26, 15, and 14 different reactants, respectively, which leads (combined with the three scaffolds) to a virtual library that comprises 16 380 compounds in total. In addition, 2779 high-potential thrombin inhibitors are identified (17%) by screening and virtual screening. 

Fig. 25 Some of the building blocks used for the thrombin inhibitor library. The building block libraries comprise 26, 15 and 14 variations in total for the variable residues R2, Rl, and R3.

Figure 12.9 Successful applications of structure-based library design, (a) Thrombin inhibitor,...

Byvatov and Schneider compared the SVM-based and the Kolmogorov-Smirnov feature selection methods to characterize ligand-receptor interactions in focused compound libraries.Three datasets were used to compare the feature selection algorithms 226 kinase inhibitors and 4479 noninhibitors 227 factor Xa inhibitors and 4478 noninhibitors and 227 factor Xa inhibitors and 195 thrombin inhibitors. SVM classifiers with a degree 5 polynomial kernel were used for all computations, and the molecular structure was encoded into 182 MOE descriptors and 225 topological pharmacophores. In one test, both feature selection algorithms produced comparable results, whereas in all other cases, SVM-based feature selection had better predictions. 

---