Functional analysis of phenylalanine residues in the active site of cytochrome P450 2C9

Abstract

The two published crystal structures of cytochrome P450 2C9, complexed with ( S)-warfarin or flurbiprofen, implicate a cluster of three active site phenylalanine residues (F100, F114, F476) in ligand binding. However, these three residues appear to interact differently with these two ligands based on the static crystal structures. To elucidate the importance of CYP2C9's active site phenylalanines on substrate binding, orientation, and catalytic turnover, a series of leucine and tryptophan mutants were constructed and their interactions with ( S)-warfarin and ( S)-flurbiprofen examined. The F100-->L mutation had minor effects on substrate binding and metabolism of each substrate. In contrast, the F114L and F476L mutants exhibited substantially reduced ( S)-warfarin metabolism and altered hydroxy metabolite profiles but only modestly decreased nonsteroidal antiinflammatory drug (NSAID) turnover while maintaining product regioselectivity. The F114-->W and F476-->W mutations also had opposing effects on ( S)-warfarin versus NSAID turnover. Notably, the F476W mutant increased the efficiency of ( S)-warfarin metabolism 5-fold, yet decreased the efficiency of ( S)-flurbiprofen turnover 20-fold. (1)H NMR T 1 relaxation studies suggested a slightly closer positioning of ( S)-warfarin to the heme in the F476W mutant relative to the wild-type enzyme, and stoichiometry studies indicated enhanced coupling of reducing equivalents to product formation for ( S)-warfarin, again in contrast to effects observed with ( S)-flurbiprofen. These data demonstrate that F114 and F476, but not F100, influence ( S)-warfarin's catalytic orientation. Differential interactions of F476 mutants with the two substrates suggest that their catalytically productive binding modes are not superimposable.

Figure 1

(S)-Warfarin and flurbiprofen bound in crystal structures 1OG5 (Panel A) and 1R9O (Panel B), respectively. View is looking down the access channel to the heme (red). Bound ligands are shown in green. For clarity, the F and G helices (and F–G loop) have been removed, but for reference should appear at the top of each picture, closest to the viewer, with the F–G loop on the left hand side. Phenylalanine side-chains are depicted for F100 in yellow, F114 in blue and F476 in purple. Distances from the sites of oxidation to the heme iron are shown by dotted lines.

Figure 2

CYP2C9 metabolic turnover at saturating concentrations of (A) flurbiprofen, (B) diclofenac, and (C) (S)-warfarin. In (C), bars represent sum of 4′-, 6-, 7-, and 8-hydroxywarfarin. Values are represented as the percentage of wild-type turnover (A: 4′-hydroxyflurbiprofen formation = 8.0 ± 0.2 pmol/min/pmol P450; B: 4′-hydroxydiclofenac formation = 16 ± 0.2 pmol/min/pmol P450; C: total warfarin formation = 0.39 ± 0.04 pmol/min/pmol P450). Error bars indicate standard deviation of triplicate incubations. W/W: F114W/F476W. L/L: F114L/F476L. Insets: chemical structures. Solid arrows indicate major sites of metabolism for wild-type CYP2C9; dotted arrows indicate minor sites.

Figure 3

Regioselectivity of (S)-warfarin hydroxylation by CYP2C9 mutants. (S)-warfarin (0.5 mM) was incubated with CYP2C9 variants as described in the Materials and Methods. All incubations were performed in triplicate, and error bars indicate standard deviation. (A) Variants with major metabolite on the coumarin nucleus. W/W: F114W/F476W. (B) Variants with major metabolite on the C9 phenyl ring. L/L: F114L/F476L.

Figure 4

Binding spectra of (A) (S)-flurbiprofen or (B) (S)-warfarin to wild-type CYP2C9. Top insert: Hyperbolic saturation curve of the observed difference between 390 and 420 nm versus ligand concentration as fitted to the one site saturation model. Bottom insert: Scatchard plot of the saturation curve.