Inhibitors of HIV-1 aspartic protease, an enzyme which is essential for viral processing and maturation, represent an important new type of anti-AIDS drug. To be effective, these must be potent and selec tive, must present a suitable pharmacological profile with respect to drug uptake and clearance, and should also be synthesizeable in good yields. We have developed a design method based on computer aided modification of a reference compound for which the crystal structure of the complex with the enzyme is known. After relaxation of the new structures to an optimized geometry, the complexation energy is calculated, relative to that of the reference inhibitor, taking into account all aspects of the interaction, including solvation. This affords a numerical prediction of the putative effectiveness of the new structure as an inhibitor, relative to that of the reference structure, and thus allows us to rapidly evaluate modifications that could result in increased potency and reduce the number of compounds that it is actually necessary to synthesize so as to obtain useful lead compounds for drug development. In an initial study, we have determined the role of flanking residues in modulating inhibition, for hexapeptide mimetics with a central, reduced amide non-cleavable bond. The structure was based on that of the reference inhibitor MVT-101. We have found that by tuning the residues flanking the central bond, the complexation energy could be markedly improved, and have determined that the putatively optimal structure should contain an aromatic residue (Phe or Tyr) at positions p1 and p1', immediately flanking the central bond, a hydrophobic residue at P2, a glutamic acid residue at P2', and an aromatic residue (Phe, Tyr or Trp) at positions P3 and P3', most distant from the central bond. Synthesis of a series of inhibitors containing these modifications was carried out entirely in the solid phase, using Fmoc type chemistry, on a polyethylene glycol/polystyrene resin, and in vitro enzyme inhibition assays have confirmed the computer-based predictions. In particular, it was possible to obtain several inhibitors with potency in the low nanomolar range, an improvement of several orders of magnitude with respect to the parent compound.
