Molecular modeling of human P450c17 (17alpha-hydroxylase/17,20-lyase): insights into reaction mechanisms and effects of mutations.

P450c17 (17alpha-hydroxylase/17,20-lyase) catalyzes steroid 17alpha-hydroxylase and 17,20-lyase activities in the biosynthesis of androgens and estrogens. These two activities are differentially regulated in a tissue-specific and developmentally programmed manner. To visualize the active site topology of human P450c17 and to study the structural basis of its substrate specificity and catalytic selectivity, we constructed a second-generation computer-graphic model of human P450c17. The energetics of the model are comparable to those of the principal template of the model, P450BMP, as determined from its crystallographic coordinates. The protein structure analysis programs PROCHECK, WHATIF, and SurVol indicate that the predicted P450c17 structure is reasonable. The hydrophobic active site accommodates both delta4 and delta5 steroid substrates in a catalytically favorable orientation. The predicted contributions of positively charged residues to the redox-partner binding site were confirmed by site-directed mutagenesis. Molecular dynamic simulations with pregnenolone, 17-OH-pregnenolone, progesterone, and 17-OH-progesterone docked into the substrate-binding pocket demonstrated that regioselectivity of the hydroxylation reactions is determined both by proximity of hydrogens to the iron-oxo complex and by the stability of the carbon radicals generated after hydrogen abstraction. The model explains the activities of all known naturally occurring and synthetic human P450c17 mutants. The model predicted that mutation of lysine 89 would disrupt 17,20-lyase activity to a greater extent than 17alpha-hydroxylase activity; expression of a test mutant, K89N, in yeast confirmed this prediction. Hydrogen peroxide did not support catalysis of the 17,20-lyase reaction, as would be predicted by mechanisms involving a ferryl peroxide. Our present model and biochemical data suggest that both the hydroxylase and lyase activities proceed from a common steroid-binding geometry by an iron oxene mechanism. This model will facilitate studies of sex steroid synthesis and its disorders and the design of specific inhibitors useful in chemotherapy of sex steroid-dependent cancers.