Combinatorial design of nonsymmetrical cyclic urea inhibitors of aspartic protease of HIV-1

The aspartic protease (PR) of the human immunodeficiency virus type 1 (HIV-1) is an important target for the design of specific antiviral agents dedicated to treatment of HIV-1 infection. We have employed computer-assisted combinatorial chemistry methods to design a small focused virtual library of nonsymmetrically substituted cyclic urea inhibitors of the PR. Nonsymmetrical compounds with decreased peptidic character were namely found to inhibit the PR with comparable inhibition potencies as their C2-pseudosymmetric counterparts and to possess superior pharmacokinetic properties. To generate the virtual library of fully nonsymmetrical cyclic urea analogs, diverse reagents were selected from databases of available chemicals with characteristics similar to those of the building blocks of known potent PR inhibitors. The X-ray structure of the protease–inhibitor complex PR–XV-638 was used as the receptor model in the structure-based focusing and in silico screening of the virtual library. A target-specific LUDI-type scoring function, parameterized for a QSAR training set of known cyclic urea inhibitors and validated on a set of compounds not included into the training set, was used to predict the inhibition constants (Ki) of the generated analogs toward the HIV-1 PR. The fragments most frequently occurring in the analogs with the highest predicted inhibition potencies (Ki*<10pM) were then selected to constitute a highly focused library subset containing novel nonsymmetrical cyclic ureas with predicted Ki*s 1 order of magnitude lower than the most potent known cyclic urea inhibitors. ADME properties calculated for the most promising analogs suggested that the cyclic ureas are endowed with a wide range of favorable pharmacokinetic properties, which may favor the discovery of a potent orally administrable antiviral drug.

We have designed a focused virtual library of fully nonsymmetrical cyclic urea inhibitors of aspartic protease of HIV-1. Target-specific scoring function, parameterized for a QSAR training set of known inhibitors, was used to predict the inhibition constants. The library contains virtual hits with predicted Kis in low picomolar range endowed with a wide range of pharmacokinetic properties that may allow discovery of a potent bioavailable antiviral drug.Figure optionsDownload as PowerPoint slide