Bis tertiary amide inhibitors of the HIV-1 protease generated via protein structure-based iterative design.

A series of potent nonpeptide inhibitors of the HIV protease have been identified. Using the structure of compound 3 bound to the HIV protease, bis tertiary amide inhibitor 9 was designed and prepared. Compound 9 was found to be about 17 times more potent than 3, and the structure of the protein-ligand complex of 9 revealed the inhibitor binds in an inverted binding mode relative to 3. Examination of the protein-ligand complex of 9 suggested several modifications in the P1 and P1' pockets. Through these modifications it was possible to improve the activity of the inhibitors another 100-fold, highlighting the utility of crystallographic feedback in inhibitor design. These compounds were found to have good antiviral activity in cell culture, were selective for the HIV protease, and were orally available in three animal models.

Structure-based design of lipophilic quinazoline inhibitors of thymidylate synthase.

To develop novel lipophilic thymidylate synthase (TS) inhibitors, the X-ray structure of Escherichia coli TS in ternary complex with FdUMP and the inhibitor 10-propargyl-5,8-dideazafolic acid (CB3717) was used as a basis for structure-based design. A total of 31 novel lipophilic TS inhibitors, lacking a glutamate residue, were synthesized; 26 of them had in common a N-((3,4-dihydro-2-methyl-6-quinazolinyl)methyl)-N-prop-2-ynylaniline+ ++ structure in which the aniline was appropriately substituted with simple lipophilic substituents either in position 3 or 4, or in both. Compounds were tested for their inhibition of E. coli TS and human TS and also for their inhibition of the growth in tissue culture of a murine leukemia, a human leukemia, and a thymidine kinase-deficient human adenocarcinoma. The crystal structures of five inhibitors complexed with E. coli TS were determined. Five main conclusions are drawn from this study. (i) A 3-substituent such as CF(3), iodo, or ethynyl enhances binding by up to 1 order of magnitude and in the case of CF(3) was proven to fill a nearby pocket in the enzyme. (ii) A simple strongly electron-withdrawing substituent such as NO(2) or CF(3)SO(2) in the 4-position enhances binding by 2 orders of magnitude; it is hypothesized that the transannular dipole so induced interacts favorably with the protein. (iii) Attempts to combine the enhancements of i and ii in the same molecule were generally unsuccessful (iv) A 4-C(6)H(5)SO(2) substituent provided both electron withdrawal and a van der Waal's interaction of the phenyl group with a hydrophobic surface at the mouth of the active site. The inhibition (K(is) = 12 nM) of human TS by this compound, 7n, showed that C(6)H(5)SO(2) provided virtually as much binding affinity as the CO-glutamate which it had replaced. (v) The series of compounds were poorly water soluble, and also the potent TS inhibition shown by several of them did not translate into good cytotoxicity. Compounds with large cyclic groups linked to position 4 by an SO or SO(2) group did, however, have IC(50)'s in the range 1-5 microM. Of these, 4-(N-((3,4-dihydro-2-methyl-6-quinazolinyl)methyl)-N-prop-2-ynylamino )phenyl phenyl sulfone, 7n, had IC(50)'s of about 1 microM and was chosen for further elaboration.

Structure-based design of substituted diphenyl sulfones and sulfoxides as lipophilic inhibitors of thymidylate synthase.

Six new diphenyl sulfoxide and five new diphenyl sulfones were designed, synthesized, and tested for their inhibition of human and Escherichia coli thymidylate synthase (TS) and of the growth of cells in tissue culture. The best sulfoxide inhibitor of human TS was 3-chloro-N-((3,4-dihydro-2-methyl-4-oxo-6-quinazolinyl)methyl)-4- (phenylsulfinyl)-N-(prop-2-ynyl)-aniline (7c) that had a Ki of 27 nM. No sulfone improved on TS inhibition by the previously reported 4-(N-((3,4-dihydro-2-methyl-6-quinazolinyl)methyl)-N-prop-2- ynylamino)phenyl phenyl sulfone (Ki = 12 nM). Nevertheless, one sulfone, 4-((2-chlorophenyl)sulfonyl)-N-((3,4-dihydro-2-methyl-4-oxo-6- quinazolinyl)methyl)-N-(prop-2-ynyl)aniline, was selected, on the basis of its inhibition of both TS and cell growth, for antitumor testing; it gave a 61% increase in life span to mice bearing the thymidino kinase-deficient L5178Y (TK-) lymphoma. A crystal structure of N-((3,4-dihydro-2-methyl-4-oxo-6-quinazolinyl)methyl)-4-((2- methylphenyl)sulfinyl)-N-(prop-2-ynyl)aniline complexed with E. coli TS was solved and revealed selective binding of one sulfoxide enantiomer. AMBER calculations showed that the enantioselection was due to asymmetric electrostatic effects at the mouth of the active site. In contrast, a similar crystal structure of the sulfoxide 7c, along with AMBER calculations, indicated that both enantiomers bound, but with different affinities. The side chain of Phe176 shifted in order to structurally accommodate the chlorine of the more weakly bound enantiomer.

Protein structure-based design of potent orally bioavailable, nonpeptide inhibitors of human immunodeficiency virus protease.

A class of potent nonpeptidic inhibitors of human immunodeficiency virus protease has been designed by using the three-dimensional structure of the enzyme as a guide. By employing iterative protein cocrystal structure analysis, design, and synthesis the binding affinity of the lead compound was incrementally improved by over four orders of magnitude. An inversion in inhibitor binding mode was observed crystallographically, providing information critical for subsequent design and highlighting the utility of structural feedback in inhibitor optimization. These inhibitors are selective for the viral protease enzyme, possess good antiviral activity, and are orally available in three species.