Biochemical experiments, using the well-defined human liver CYP2C9 expressed in yeast, and molecular modeling techniques were used to derive a predictive model for substrates of CYP2C9. The ability of 10 2-aroylthiophenes related to tienilic acid to act as substrates for CYP2C9 was studied. Four of them were original compounds that were synthesized and completely characterized by several spectroscopic techniques. In these 10 compounds various chemical functions, such as ester, amide, alcohol, phenol, ether or tetrazole functions, replaced the OCH2COOH function of tienilic acid. Among them, only the derivatives containing an acidic function (carboxylic acids, phenol, and tetrazole whose pKaS are 4.8, 6.3, and 3.8, respectively) underwent a 5-hydroxylation of their thiophene ring like tienilic acid. Despite their close structural analogy with tienilic acid, all of the other compounds not only did not undergo any 5-hydroxylation of their thiophene ring but also failed to act as inhibitors of CYP2C9. These results strongly suggested that the presence, at pH 7.4, of a negative charge on the substrate is a very important feature in its recognition by CYP2C9. In fact, the four new substrates of CYP2C9 described in this study, a carboxylic acid, phenol, and tetrazole derivative, each of which is related to tienilic acid, and the antiinflammatory drug, suprofen (with Km between 12 and 130 microM and kcat between 0.2 and 1.3 min-1), as well as almost all CYP2C9 substrates reported in the literature, exhibit a pKa below 7 (except phenytoin whose pKa is 8.1). They mainly exist as anions at physiological pH. By using molecular modeling techniques, 12 CYP2C9 substrates were superimposed with respect to their hydroxylation site and fitted onto templates, which were rigid molecules such as (S)-warfarin and phenytoin. It was thus possible to arrange them in order that all their anionic sites were at a distance around 4 A from a common point (a putative cationic site of the protein) in space. These results provide a model of the substrate binding site of CYP2C9, in which substrates interact through their anionic site A- with a cationic residue of the CYP2C9 protein C+. In that model, the distance between the hydroxylation site (Hy) and the anionic site (A-) is 7.8 +/- 1.6 A, and the <HyA-C+ angle is 82 +/- 15 degrees.