Structure of a non-peptide inhibitor complexed with HIV-1 protease. Developing a cycle of structure-based drug design.

A stable, non-peptide inhibitor of the protease from type 1 human immunodeficiency virus has been developed, and the stereochemistry of binding defined through crystallographic three-dimensional structure determination. The initial compound, haloperidol, was discovered through computational screening of the Cambridge Structural Database using a shape complementarity algorithm. The subsequent modification is a non-peptidic lateral lead, which belongs to a family of compounds with well characterized pharmacological properties. This thioketal derivative of haloperidol and a halide counterion are bound within the enzyme active site in a mode distinct from the observed for peptide-based inhibitors. A variant of the protease cocrystallized with this inhibitor shows binding in the manner predicted during the initial computer-based search. The structures provide the context for subsequent synthetic modifications of the inhibitor.

Structure-based design of nonpeptide inhibitors specific for the human immunodeficiency virus 1 protease.

By using a structure-based computer-assisted search, we have found a butyrophenone derivative that is a selective inhibitor of the human immunodeficiency virus 1 (HIV-1) protease. The computer program creates a negative image of the active site cavity using the crystal structure of the HIV-1 protease. This image was compared for steric complementarity with 10,000 molecules of the Cambridge Crystallographic Database. One of the most interesting candidates identified was bromperidol. Haloperidol, a closely related compound and known antipsychotic agent, was chosen for testing. Haloperidol inhibits the HIV-1 and HIV-2 proteases in a concentration-dependent fashion with a Ki of approximately 100 microM. It is highly selective, having little inhibitory effect on pepsin activity and no effect on renin at concentrations as high as 5 mM. The hydroxy derivative of haloperidol has a similar effect on HIV-1 protease but a lower potency against the HIV-2 enzyme. Both haloperidol and its hydroxy derivative showed activity against maturation of viral polypeptides in a cell assay system. Although this discovery holds promise for the generation of nonpeptide protease inhibitors, we caution that the serum concentrations of haloperidol in normal use as an antipsychotic agent are less than 10 ng/ml (0.03 microM). Thus, concentrations required to inhibit the HIV-1 protease are greater than 1000 times higher than the concentrations normally used. Haloperidol is highly toxic at elevated doses and can be life-threatening. Haloperidol is not useful as a treatment for AIDS but may be a useful lead compound for the development of an antiviral pharmaceutical.

Synthesis of 2,2-dimethyl-4-hydroxy-4-androstene-3,17-dione as an inhibitor of aromatase.

2,2-Dimethyl-4-hydroxy-4-androstene-3,17-dione (4) has been synthesized and has been shown to be a powerful competitive inhibitor of aromatase (Ki = 11.4 nM). However, compound 4 does not cause time-dependent loss of enzyme activity, in contrast to the unmethylated parent compound, 4-OHA.

Tritium release from [19-3H]-19,19-difluoroandrost-4-ene-3,17-dione during inactivation of aromatase.

Aromatase is a cytochrome P-450 enzyme involved in the conversion of androst-4-ene-3,17-dione to estrogen via sequential oxidations at the 19-methyl group. Previous studies from this laboratory showed that 19,19-difluoroandrost-4-ene-3,17-dione (5) is a mechanism-based inactivator of aromatase. The mechanism of inactivation was postulated to involve enzymic oxidation at, and hydrogen loss from, the 19-carbon. The deuteriated analogue 5b has now been synthesized and shown to inactivate aromatase at the same rate as the nondeuteriated parent (5). We conclude that C19-H bond cleavage is not the rate-limiting step in the overall inactivation process caused by 5. [19-3H]-19,19-Difluoroandrost-4-ene-3,17-dione (5b) with specific activity of 31 mCi/mmol was also synthesized to study the release of tritium into solution during the enzyme inactivation process. Incubation of [19-3H]19,19-difluoroandrost-4-ene-3,17-dione with human placental microsomal aromatase at differing protein concentrations resulted in time-dependent NADPH-dependent, and protein-dependent release of tritium. This tritium release is not observed in the presence of (19R)-10 beta-oxiranylestr-4-ene-3,17-dione, a powerful competitive inhibitor of aromatase. We conclude that aromatase attacks the 19-carbon of 19,19-difluoroandrost-4-ene-3,17-dione, as originally postulated.

Stereoselective inhibition of human placental aromatase.

We have synthesized the (19R)- and (19S)-isomers (2 and 3 respectively) of 10 beta-oxiranylestr-4-ene-3,17-dione. The configurations and conformations of these compounds were established by X-ray crystallographic analysis. Each of these compounds is a powerful competitive inhibitor of human placental microsomal aromatase, and stereoselectivity of inhibition was observed (Ki values for 2 and 3 were 7 and 75 nanomolar, respectively). Spectroscopic studies with purified aromatase indicate that the inhibition process involves reversible binding of oxirane oxygen to the heme iron of the enzyme. The (19R)- and (19S)-10 beta-thiiranes (6 and 7) corresponding to 2 and 3 have been synthesized from the oxiranes by a stereospecific process. The thiiranes are very effective competitive inhibitors of placental aromatase, and show even greater stereoselectivity in binding than the oxiranes (Ki values for 6 and 7 were 1 and 75 nanomolar, respectively). Spectroscopic studies with purified aromatase indicate that the inhibition process involves reversible binding of thiirane sulfur to heme iron.