The major human liver drug-metabolizing cytochrome P450 enzymes P450 3A4 and P450 3A5 share >85% amino acid sequence identity yet exhibit different regioselectivity toward aflatoxin B(1) (AFB(1)) biotransformation [Gillam et al. (1995) Arch. Biochem. Biophys. 317, 74-384]. P450 3A4 prefers AFB1 3alpha-hydroxylation, which detoxifies and subsequently eliminates the hepatotoxin, over AFB1 exo-8,9-oxidation. P450 3A5, on the other hand, is a relatively sluggish 3alpha-hydroxylase and converts AFB(1) predominantly to the genotoxic exo-8,9-epoxide. Using a combination of approaches (sequence alignment, homology modeling and site-directed mutagenesis), we have previously identified several divergent residues in four of the six putative substrate recognition sites (SRSs) of P450 3A4, which when replaced individually with the corresponding amino acid of P450 3A5, resulted in a significant switch of the characteristic P450 3A4 AFB(1) regioselectivity toward that of P450 3A5 [Wang et al. (1998) Biochemistry 37, 12536-12545]. In particular, residues N206 and L210 in SRS-2 were found to be critical for AFB(1) detoxification via 3alpha-hydroxylation, and the corresponding mutants N206S and L210F most closely mimicked P450 3A5, not only in its regioselectivity of AFB(1) metabolism but also in its overall functional capacity. We have now further explored the plausible reasons for such relative inactivity of the SRS-2 mutants by examining N206S and additional mutants (L210A, L211F, L211A, and N206E) and found that the dramatically lowered activities of the N206S mutant are accompanied by a loss of cooperativity of AFB(1) oxidation. Molecular dynamics analyses with an existing P450 3A4 homology model [Szklarz and Halpert (1997) J. Comput. Aided Mol. Des. 11, 265] suggested that N206 (helix F) interacts with E244 (helix G), creating a salt bridge that stabilizes the protein structure and/or defines the active site cavity. To examine this possibility, several E244 mutants (E244A, V, N, S) were tested, of which E244S was the most notable for its relatively greater impairment of P450 3A4-dependent AFB(1) 3alpha-hydroxylation. However, the results with these E244 mutants failed to validate the N206-E244 interaction predicted from these molecular dynamics analyses. Collectively, our findings to date have led us to reconsider our original interpretations and to reexamine them in the light of AFB(1) molecular modeling analyses with a newly refined P450 3A4 homology model. These analyses predicted that F304 in SRS-4 (I-helix) plays a pivotal role in AFB(1) binding at the active site in either orientation leading to 3alpha- or exo-8,9-oxidation. Consistent with this prediction, conversion of F304 to Ala abolished P450 3A4-dependent AFB(1) 3alpha-hydroxylation and exo-8,9-oxidation.