Cytochrome b 5 Augments the 17,20-Lyase Activity of Human P450c17 without Direct Electron Transfer*

In the biosynthesis of steroid hormones, P450c17 is the single enzyme that catalyzes both the 17 a -hydroxylation of 21-carbon steroids and the 17,20-lyase activity that cleaves the C 17 -C 20 bond to produce C 19 sex steroids. Cytochrome b 5 augments the 17,20-lyase activity of cy- tochrome P450c17 in vitro , but this has not been demonstrated in membranes, and the mechanism of this action is unknown. We expressed human P450c17, human P450-oxidoreductase (OR), and/or human cytochrome b 5 in Saccharomyces cerevisiae and analyzed the 17 a -hydrox-ylase and 17,20-lyase activities of the resulting yeast microsomes. Yeast expressing only P450c17 have 17 a hydroxylase and trace 17,20-lyase activities toward both D 4 and D 5 steroids. Coexpression of human OR with P450c17 increases the V max of both the 17 a -hydroxylase and 17,20-lyase reactions 5-fold; coexpression of human b 5 with P450c17 also increases the V max of the 17,20-lyase reactions but not of the 17 a -hydroxylase reactions. Si-multaneous expression of human b 5 with P450c17 and OR, or addition of purified human b 5 to microsomes from yeast coexpressing human P450c17 and OR, further increases the V max of the 17,20-lyase reaction with- out altering 17 a -hydroxylase activity. Genetically engineered Engineered yeast strains, W(B), W(hR), and W(B generated by targeted disrup- tion of the yeast CPR1 or YCY b 5 loci (11–14) and the yeast expression vectors V10 and V60 (11) were generous gifts of Dr. Denis Pompon (CNRS, Gif-sur-Yvette, France). Human P450c17 cDNA (15) was PCR amplified with Pfu polymerase La CA) using primers

Among the many chemical transformations catalyzed by cytochrome P450 enzymes, steroid hormone hydroxylations, and cleavages are of particular interest because of their mechanistic complexities and essential roles in physiology (1). P450c17 catalyzes both 17␣-hydroxylase and 17,20-lyase activities (2) (for review see Ref. 3) and also has a modest degree of 16␣hydroxylase activity (4). In human beings, the 17␣-hydroxylase reaction leads to the glucocorticoid, cortisol, and the subsequent 17,20-lyase reaction leads to precursors of sex steroids. As the sole pathway leading to biosynthesis of circulating sex steroids, the regulation of this 17,20-lyase activity is central to understanding the developmental regulation of dehydroepiandrosterone sulfate (DHEA) 1 with adrenarche and aging, and to the pathogenesis of the polycystic ovary syndrome (3). The 17,20-lyase activity, involving the oxidative cleavage of a carbon-carbon bond, is regulated in a tissue-specific and developmentally programmed manner by factors such as the abundance of the electron donor flavoprotein P450-oxidoreductase (OR) (5,6), the co-existence of 3␤-hydroxysteroid dehydrogenase and P450c21 (7), and post-translational modification of P450c17 (8).
To perform catalysis, P450c17, like all other microsomal P450 oxygenases, must receive two electrons from NADPH via OR. Cytochrome b 5 has also been implicated as a component of the 17,20-lyase reaction, as b 5 augments 17,20-lyase activity and occasionally 17␣-hydroxylase activity of P450c17 in reconstituted systems (9,10); however, our laboratory could not confirm this effect in transfected monkey kidney COS-1 cells (5). Inconsistencies in the animal species of P450c17, OR, and b 5 used in previous studies preclude extrapolation of the available biochemical data to human adrenal and gonadal physiology; furthermore, the mechanism(s) of these reported b 5 -mediated increases in 17,20-lyase activity remain unknown.
Among the various systems developed to study mammalian cytochromes P450, transfection of genetically modified yeast cells provides the opportunity to study the activities of a cytochrome P450 in the presence of various combinations of electron transfer proteins in the native microsomal environment (11). To clarify the function of cytochrome b 5 in 17,20-lyase activity, we systematically varied the abundance of putative electron transfer proteins in yeast microsomes containing human P450c17. We find that human, but not yeast cytochrome b 5 can selectively augment the rate of the 17,20-lyase reaction by more than 10-fold. However, this augmentation requires OR and occurs without electron transfer to or from cytochrome b 5 .

EXPERIMENTAL PROCEDURES
Yeast Strains and Expression Vectors-Wild type yeast strains W303A (Y150WT) (leu2-3, 112; his3-11, 15; trp1-1; ade2-1; ura3-1; mat a) and W303B (JC104) (trp1-1; ura3-1; ade2-1; can1-100; mat ␣) were generous gifts of Drs. Gregory Petsko and Ira Herskowitz. Engineered yeast strains, W(B), W(hR), and W(B⌬), generated by targeted disruption of the yeast CPR1 or YCY b 5 loci (11)(12)(13)(14) and the yeast expression vectors V10 and V60 (11) were generous gifts of Dr. Denis Pompon (CNRS, Gif-sur-Yvette, France). Human P450c17 cDNA (15) was PCR amplified with Pfu polymerase (Stratagene, La Jolla, CA) using primers c17-S-1 and c17-AS-1 (Table I) and pECE-c17 (16) as template. The resulting PCR product was digested with BamHI and EcoRI, facilitating directional cloning into complementary ends of BglII-EcoRI digested V10 vector, destroying the BglII site and placing the P450c17 cDNA under the control of the constitutive pgk promoter, producing vector V10-c17. V60 was modified by disruption of the ura3 gene by digestion with NcoI, blunting of the staggered ends, and religation, yielding vector pYeSF1. Using the unique BamHI site at the 5Ј end of the pgk promoter in V10, a BamHI-EcoRI fragment of the V10-c17 plasmid was cloned into BamHI-EcoRI digested pYeSF1 to provide ade2 complementation to the pgk-regulated P450c17 (vector pYeSF2-c17). Vector cDE2, used to generate doubly transformed yeast with either V10-c17 or pYeSF2-c17, was a generous gift from Dr. Ira Herskowitz (17). Human P450-oxidoreductase cDNA was PCR amplified from pECE-OR (5) using primers OR-S-1 and OR-AS-1 (Table I). The 5Ј-PCR primer contained two silent base pair changes from the wild type sequence to remove hairpin structures surrounding the translation initiation codon (12) which can inhibit transcription and translation in yeast (18). Cytochrome b 5 cDNA was generated by reverse transcription of total human testicular RNA using random hexamers followed by PCR amplification with primers b 5 -S-1 and b 5 -AS-1 (Table I) based on the human b 5 cDNA sequence (19). The human b 5 and OR cDNAs were then cloned into the EcoRI site of cDE2 vector under control of the constitutive adc1 promoter with a trp1 selectable marker. The accuracy and orientation of all constructions were confirmed by DNA sequencing.
Yeast Transformation and Growth-Yeast were transformed using 700 l of 40% polyethylene glycol 3350, 0.1 M lithium acetate, 10 mM Tris-HCl (pH 8), 1 mM EDTA to transform 10 6 yeast in 100 l of 0.1 M lithium acetate, 10 mM Tris-HCl (pH 8), 1 mM EDTA with 1-2 g of plasmid DNA and 50 g of denatured herring sperm carrier DNA (20). Cells were washed in 100 l of 1 M sorbitol before final resuspension in 100 l of 10 mM Tris-HCl (pH 8), 1 mM EDTA and plating onto selective media. All transformations introduced two plasmids simultaneously: the first, expressing P450c17, was either V10-c17 or pYeSF2-c17; the second was cDE2, containing no cDNA insert or the cDNA for either human OR or b 5 . Thus, P450c17 expression was always under the control of the constitutive pgk promotor, and all yeast producing different combinations of electron donor proteins were grown in the same culture medium. For microsome preparations, transformed yeast were cultured in minimal SD media containing 20 g/liter D-glucose or Dgalactose, 1.7 g/liter yeast nitrogen base without amino acids or ammonium sulfate (Difco, Detroit, MI), 5 g/liter ammonium sulfate, and supplemented with the requisite combination of 10 mg/liter leucine, 15 mg/liter adenine, and 10 mg/liter histidine (11).
Microsome Preparation and Characterization-Yeast cells harvested at a density of 4.5-6 ϫ 10 7 cells/ml were disrupted by manual breakage with glass beads (450 -600 micron) for 5 min (11). The breakage was stopped at 1-min intervals, and cells were iced for 30 s; 3 l of 0.5 M ethanolic phenylmethylsulfonyl fluoride was added after the first minute of breakage. For a typical 300-ml culture, crude extracts and beads were washed twice with 5-7 ml of 50 mM Tris-HCl (pH 8), 1 mM EDTA, 0.4 M sorbitol, and the cellular debris was collected by centrifugation at 4°C twice for 10 min at 14,000 ϫ g. Microsomes were pelleted by centrifugation at 4°C for 45 min at 100,000 ϫ g and were resuspended in 50 mM Tris-HCl (pH 8), 1 mM EDTA, 20% glycerol at 5-20 g/l total protein. Preparations were homogenized by shearing microsomes through a 27-gauge needle 10 times and were kept frozen at Ϫ70°C until needed. Human adrenal microsomes were prepared from excess surgical tissue as described (8).
Microsomal proteins were quantitated colorimetrically. Immunoblotting on polyvinylidene difluoride membranes (Millipore, Bedford, MA) was performed with rabbit antiserum to human P450c17 (5) or to human OR (generously provided by Prof. C. Roland Wolf, Imperial Cancer Institute, Dundee, United Kingdom) using secondary antibodyperoxidase conjugate and ECL reagents (Amersham, Arlington Heights, IL) and with goat antiserum to human b 5 (Oxford Biomedical, Rochester Hills, MI) using secondary antibody-peroxidase conjugate (Santa Cruz Biotechnology, Santa Cruz, CA) and ECL reagents. Microsomal P450 and cytochrome b 5 contents were measured spectroscopically (21) using either a Cary 3E or a Shimazdu UV160U spectrophotometer. P450 oxidoreductase activity was measured as described (22).
P450c17 Enzyme Assay-Microsomes were assayed under initial rate kinetics by preincubation in 50 mM potassium phosphate buffer (pH 7. , concentrated under nitrogen, separated by thin layer chromatography (Whatman PE SIL G/UV silica gel plates, Maidstone, Kent, UK) using 3:1 chloroform/ ethyl acetate, and quantitated as described (23). Purified recombinant human cytochrome b 5 (Pan Vera, Madison, WI), apo-human cytochrome b 5 , or horse heart cytochrome c (Sigma) were included in incubations as indicated. Apo-cytochrome b 5 was prepared from the Pan Vera holocytochrome b 5 as described (24), and absent electron transfer properties of the resulting material was confirmed by difference spectroscopy. Kinetic behavior was approximated as a Michaelis-Menten system for data analysis, and all error bars shown represent standard deviations.

RESULTS
Yeast Transfection and Microsome Characterization-The capacity of b 5 to increase the 17,20-lyase activity of P450c17 has been shown by several laboratories using purified, reconstituted protein systems (9,25), but this phenomenon has not been observed in intact cell and microsome preparations (5). The development of "humanized" yeast strains (12) that express both P450c17 and selected electron donor proteins has enabled us to dissect this problem without detergent solubilization of individual components.
To study the effects of human OR and b 5 on human P450c17 activities in yeast microsomes, parental yeast strain W303B was doubly transfected with vector V10 expressing human P450c17 and with vector cDE2 expressing either the cDNA for human OR or b 5 (or empty vector). Microsomes from these transfectants were characterized and used for kinetic studies; microsomes were also prepared from transfections using the same vectors with the cDNA inserts exchanged. Both sets of transfections were also performed using yeast strain W303A; abbreviated kinetic experiments using these W303A-derived microsome preparations yielded qualitatively similar results as did microsomes and spheroplasts prepared from the W303B transfectants; thus, W303B doubly transfected with V10-c17 plus cDE2 (for expressing an electron donor) was used in all subsequent experiments.
P450 content, total OR (cytochrome c reductase activity), and total cytochrome b 5 content were similar among the three microsome preparations from co-transfected W303B yeast (Table  II). Essentially all of the P450 is from P450c17, whereas cytochrome c reductase activity and total cytochrome b 5 content were similar in all transfectants, indicating that endogenous yeast OR and b 5 are the predominant electron transfer proteins in these microsomes. Comparable expression of P450c17 was demonstrated in all samples by Western blotting, and human OR and b 5 were detected only in samples from yeast containing their respective cDNAs, as expected ( Fig. 1).
Kinetics-To determine how the presence of human OR and/or b 5 alters the activities of human P450c17 in yeast microsomes, we measured apparent K m and V max values for both 17␣-hydroxylase and 17,20-lyase reactions for ⌬ 5 and ⌬ 4 substrates (Table III). Lineweaver-Burk plots (Fig. 2) show that yeast transfected with human P450c17 alone perform both 17␣-hydroxylase and 17,20-lyase reactions despite the absence of human electron transfer proteins, indicating that the endogenous yeast OR can couple with human P450c17, as has been shown for bovine P450c17 (26,27). In this system, however, 17,20-lyase activity is very low but not absent, as reported for bovine P450c17 (26). Co-expression of human OR substantially increases both activities. Using the ⌬ 5 steroid substrates pregnenolone and 17␣-hydroxypregnenolone, co-expression of human OR raises the V max for both the 17␣-hydroxylase and 17,20-lyase reactions 5-fold. In yeast microsomes, the apparent K m for both pregnenolone and 17␣-hydroxypregnenolone is about 1 M, and co-expression of human OR lowers the K m to below 0.5 M, suggesting that the association of human P450c17 and OR may increase the affinity of P450c17 for ⌬ 5 substrates. The presence of OR also alters the orientation of 1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine in the substratebinding pocket of P450 2D6 (28), suggesting that OR may participate in substrate discrimination by P450 enzymes. DHEA formed by microsomes containing both human P450c17 and OR is metabolized further to a more polar compound, possibly 16␣-hydroxy-DHEA, the major DHEA metabolite of a bovine P450c17/rat OR fusion protein (29). Human OR markedly stimulated both activities without a significant change in total cytochrome c reductase activity, indicating that yeast OR is an inefficient electron donor for P450c17 and does not significantly interfere with catalysis in the presence of human OR.
When ⌬ 4 substrates are used, human P450c17 efficiently catalyzed the conversion of progesterone to 17␣-hydroxyprogesterone, but the conversion of 17␣-hydroxyprogesterone to androstenedione was much less efficient than the corresponding conversion of 17␣-hydroxypregnenolone to DHEA (Table III). The slow turnover of 17␣-hydroxyprogesterone by human P450c17 explains why circulating androgens in humans derive principally from the isomerization and reduction of DHEA rather than by cleavage of 17␣-hydroxyprogesterone to androstenedione, the predominant pathway in rodents. Guinea pig P450c17, for example, preferentially converts progesterone to FIG. 1. Immunoblot of yeast microsomal proteins. Denatured microsomal proteins (2-5 g) from yeast transfected with the various vectors as labeled at the top of the figure were electrophoresed through a 10% SDS-polyacrylamide gel, blotted to polyvinylidene difluoride, and probed with antisera to P450c17 and OR. Expression of human P450c17 (c17) is comparable in all samples, but human OR is expressed only in yeast containing the OR expression vector. Immunoblotting with antisera to rat or rabbit b 5 showed aggregation and cross-reactivity (not shown). Dash (Ϫ) indicates use of empty vector.
FIG. 2. Lineweaver-Burk plots of 17␣-hydroxylase and 17,20-lyase activities. Lines were derived from least-squares fit to data points (r 2 Ͼ 0.92 for all lines). The apparent K m and V max values obtained from these data are shown in Table III. Microsomes were prepared from W303B yeast co-transfected with V10-c17 and the empty cDE2 vector (squares), V10-c17 and cDE2-OR (circles), or V10-c17 and cDE2-b 5 (triangles). androstenedione, some of which is sequentially metabolized without dissociation of the intermediate 17␣-hydroxyprogesterone from the active site (30,31). Co-expression of human OR similarly increases the V max of both activities toward ⌬ 4 steroids but without a significant change in apparent K m values (Table III). A second, more polar product, presumably 16␣hydroxyprogesterone (4,5,32), constitutes ϳ20 -25% of the products when ⌬ 4 -progesterone is the substrate with all microsomes tested. Co-expression of human b 5 with human P450c17 increases V max 10-fold for the 17,20-lyase reaction but not for the 17␣hydroxylase reaction with both ⌬ 5 and ⌬ 4 substrates and does not change the apparent K m for any substrate tested (Table  III). Although human OR improves the catalytic efficiency of P450c17 in yeast microsomes, both by lowering the K m of ⌬ 5 substrates and increasing the V max for all reactions, the sole effect of human b 5 is to augment the V max for the 17,20-lyase reactions. Our results generally agree with those obtained with reconstituted recombinant human P450c17 and rat OR, except that rat b 5 approximately doubles the rate of hydroxylation of ⌬ 5 -pregnenolone but not of ⌬ 4 -progesterone (25). Differences in species of origin of the OR and b 5 used may explain some differences in the results obtained in the two systems, as well as subtle differences in the activities of microsomal and detergent-solubilized proteins.
Activities in the Absence of Yeast OR or Yeast Cytochrome b 5 -The experiments described above were performed in the presence of endogenous yeast OR and b 5 in the microsome preparations. To determine whether the yeast electron donors influence human P450c17 activities, we expressed human P450c17, with human OR or b 5 , in engineered yeast strains lacking the endogenous yeast OR or b 5 genes. When human P450c17 and OR were coexpressed in yeast strain W(B⌬), which lacks the yeast homolog of the human b 5 gene (13), the resulting microsomes contained 85% of the 17␣-hydroxylase activity and 73% of the 17,20-lyase activity of microsomes from W303B yeast (Fig. 3A). The 17,20-lyase activity was minimally affected by the absence of yeast b 5 , demonstrating that OR is both necessary and sufficient to confer both 17␣-hydroxylase and 17,20-lyase activity to human P450c17 in yeast microsomes.
To confirm that OR was required for catalysis, we expressed human P450c17 in strain W(B), in which the endogenous yeast OR locus is replaced by the human b 5 cDNA under the control of the inducible Gal10/Cyc1 promoter (12). No 17␣-hydroxylase or 17,20-lyase activity is present in microsomes prepared from W(B) yeast transfected with human P450c17 and empty cDE2 vector, but both activities are restored by co-transfection of human OR (Fig. 3B). When W(B) yeast, transfected with both human P450c17 and OR, were grown in galactose to induce expression of human b 5 as well, the presence of human b 5 increased the V max of the 17,20-lyase reaction using 17␣-hydroxypregnenolone from 0.14 min Ϫ1 to 1.1 min Ϫ1 but did not change the apparent K m (0.3 M) (Fig. 3, B and C). This induc-  1 h incubation, lanes 1-4) or 17,20-lyase activity (4 h incubation, lanes 5-8), respectively. Panel C, Lineweaver-Burk plot of 17,20-lyase activity in microsomes from W(B) yeast co-transfected with pYeSF2-c17 and cDE2-OR. Microsomes were prepared from the same yeast clone expressing trace human b 5 (grown in glucose, squares), or expressing high b 5 (grown in galactose, circles). Apparent K m and V max values were derived from least-squares fits to the data. How Cytochrome b 5 Stimulates 17,20-Lyase Activity tion of human b 5 did not significantly change 17␣-hydroxylase activity, reflected by comparable pregnenolone consumption (Fig. 3B, lanes 3 and 4), but the 17␣-hydroxypregnenolone formed in the presence of high amounts of human b 5 was rapidly converted to DHEA, so that little 17␣-hydroxypregnenolone accumulated (Fig. 3B, lane 4). Effect of Exogenous Soluble b 5 on 17␣-Hydroxylase and 17,20-Lyase-Exogenously added soluble b 5 can influence other P450 reactions in yeast microsomes (11); therefore, we added purified human b 5 to yeast microsomes containing human P450c17. Although 17␣-hydroxylase activity against ⌬ 5 -pregnenolone or ⌬ 4 -progesterone was not changed (Fig. 4, A and C), 17,20-lyase activity against 17␣-hydroxypregnenolone was increased up to 10-fold in microsomes that did not already contain human b 5 (Fig. 4B). Purified b 5 also increased 17,20-lyase activity toward 17␣-hydroxyprogesterone, but only about 2-fold (Fig. 4D). These data demonstrate that yeast b 5 can neither support nor stimulate human P450c17 activities, as found for other human P450s (12). Furthermore, our results show that the only effect of human b 5 , either added in solution or coexpressed into microsomes, is to increase the rate of the 17,20lyase reactions, and that this action of b 5 requires the presence of yeast or human OR.
The results described above do not exclude a contribution of human b 5 as the donor of the second of the two electrons in the P450 catalytic cycle, as has been suggested (33,34). If b 5 functions as the donor of the second electron, b 5 should support catalysis by transporting electrons either from a reducing agent (sodium dithionite) or from NADPH-reduced OR to microsomes containing P450c17 that has already been reduced with the first electron. Dithionite, which can provide one elec-tron to either P450c17 or b 5 , does not support catalysis in microsomes containing both human P450c17 and b 5 , but dithionite does not abolish catalysis when the second electron is provided to P450c17 from NADPH via OR (Fig. 5A). To confirm this observation, we attempted to reconstitute 17,20-lyase activity by transferring electrons from NADPH to one pool of microsomes containing human OR (and no P450c17), then to soluble human b 5 as an electron conduit, and finally to human P450c17 in another pool of microsomes lacking OR. Soluble b 5 was first reduced with NADPH by microsomes containing OR (35), and then added to microsomes lacking yeast OR (strain W(B)) but containing human P450c17 alone (lane 1), human P450c17 and OR (lane 2), or human P450c17 and b 5 (but no OR, lane 3), all of which had been preincubated with 17␣-hydroxypregnenolone and dithionite to provide the first electron to P450c17. Microsomes lacking human OR converted only a trace of 17␣-hydroxypregnenolone to DHEA under these conditions, but microsomes containing both human P450c17 and OR could use the added NADPH to convert substrate to DHEA (Fig. 5B). These results confirm that b 5 , reduced either by dithionite or OR, cannot provide sufficient electron transfer to P450c17 to support significant 17,20-lyase activity. Therefore, these data suggest that the mechanism by which b 5 enhances 17,20-lyase activity does not involve electron transfer.
How Does Human Cytochrome b 5 Augment 17,20-Lyase Activity?-To explore the mechanism by which b 5 increases 17,20lyase activity, we assayed the 17,20-lyase activity of microsomes containing constant, high amounts of P450c17 and OR and varying amounts of b 5 . A sharp increase in 17,20-lyase activity was observed when the molar ratio of b 5 to P450c17 approached 1:1 (Fig. 6A). Activity reached a maximum at ratios of b 5 to P450c17 between 10:1 and 30:1; however, further addition of human b 5 progressively inhibited 17,20-lyase activity in both yeast and human adrenal microsomes. If human b 5 was acting as the preferred electron donor, 17,20-lyase activity should saturate and remain constant rather than fall at high b 5 /P450c17 ratios. Similarly, when we examined the influence of b 5 on the 17,20-lyase activity of microsomes containing very small amounts of human OR and no yeast OR (strain W(hR) transfected with P450c17 grown to high density in glucose), maximal stimulation occurred at a b 5 /P450c17 ratio between 1:1 and 3:1, and higher ratios were again inhibitory (Fig. 6A). Thus, the influence of human b 5 changes dramatically as the abundance of human OR and the b 5 /P450c17 ratio are varied. These data suggest that b 5 does not function as an electron donor, but instead exerts some other action, perhaps facilitating electron transfer from OR to P450c17 or improving coupling efficiency, as has been suggested for other P450 reactions stimulated by b 5 (24,36).
Inhibition of enzymatic activity at high b 5 /P450c17 ratios, a phenomenon also observed in guinea pig adrenal microsomes (10), could result from a second, inhibitory b 5 -binding site on P450c17 or from a competition between b 5 and P450c17 for electrons from limiting amounts of OR. Cytochrome c, which is also a substrate for reduction by OR (37), also inhibits 17,20lyase activity at molar ratios above 10:1, the same molar ratios at which b 5 becomes inhibitory (Fig. 6C). Inhibition by equivalent molar ratios of cytochrome c to P450c17 is consistent with b 5 competing with P450c17 for reduction when OR is limiting, but "reverse" electron transfer from P450c17 to b 5 (38) may also contribute to the inhibition observed at higher b 5 /P450c17 ratios. These data suggest that electron transfer from OR to b 5 is actually detrimental to 17,20-lyase activity. Therefore, we determined whether human apo-b 5 , which lacks the heme and hence cannot participate in electron transfer, modulates 17,20lyase activity differently than human holo-b 5 . In microsomes containing either low or high amounts of human OR, molar ratios of apo-b 5 to P450c17 between 1:1 and 10:1 augment 17,20-lyase activity (Fig. 6B). Unlike the data with holo-b 5 , the stimulatory effect of apo-b 5 remains constant rather than falling at higher b 5 /P450c17 ratios. These results exclude direct electron transfer from b 5 as the principal means by which b 5 augments 17,20-lyase activity and suggest that b 5 exerts a saturable, allosteric effect on the P450c17⅐OR complex.

DISCUSSION
The 17,20-lyase/17␣-hydroxylase ratio in the human adrenal rises dramatically with the onset of adrenarche at age 8 -10, reaches maximal values at age 25-35, and then falls progressively with aging (39); as these phenomena occur only in human beings and great apes (40), their study is difficult. These selective, physiologic, developmentally programmed changes in human adrenal 17,20-lyase activity imply regulatory mechanisms beyond transcription of P450c17 or OR (3,8). Most P450 enzymes catalyze multiple reactions, but the ratio of their activities remains fixed. The developmentally and possibly hor- Activity is expressed as the percent of conversion by the microsomes alone. The yeast microsomes were prepared from strain W(hR) either co-transfected with pYeSF2-c17 plus cDE2-OR and grown in galactose, yielding the microsomes with high amounts of human OR, or co-transfected with pYeSF2-c17 plus empty cDE2 vector and grown in glucose, yielding the microsomes with low amounts of human OR (cytochrome c reductase activities of 223 and 12 nmol/min/mg protein, respectively). monally programmed changes in the ratio of 17,20-lyase to 17␣-hydroxylase activities of human P450c17 provide a unique system for studying the differential regulation of two reactions catalyzed by a single P450 enzyme.
An augmentation of the 17,20-lyase activity of P450c17 by b 5 has been observed in vitro (9, 25) but was not seen in transfected COS-1 cells (5), possibly because the endogenous b 5 in those cells was sufficient to stimulate 17,20-lyase activity maximally. Thus, it has not been clear how or if b 5 regulates human P450c17 activities in vivo. The use of microsomes from yeast engineered to express human P450c17, OR, or b 5 from inducible promoters permits the quantitative manipulation of each protein in a membrane environment that should simulate events in vivo. This permits greater experimental flexibility than the use of bicistronic plasmids (41), fusion proteins (29), or viral vectors (42), and obviates concerns about the relevance of data from detergent-solubilized systems to in vivo systems.
Titration experiments with purified human holo-b 5 , apo-b 5 , and cytochrome c showed that the stimulatory effect of b 5 on 17,20-lyase activity is not mediated by electron transfer from b 5 and suggest that b 5 exerts an allosteric effect on the P450c17⅐OR complex. This proposed mechanism could explain three observations from other laboratories. First, b 5 facilitates electron transfer from OR to P450 3A4 only when all three proteins are premixed before adding NADPH and substrate, but not when b 5 is premixed with P450 3A4 and added to OR, NADPH, and substrate in stop-flow experiments (24). These data suggested that the stimulatory action of b 5 on testosterone 6␤-hydroxylation by P450 3A4 was an allosteric effect and was not mediated by an action of b 5 as an alternate electron donor (24). Second, b 5 is a more potent stimulator of 17,20-lyase activity when the abundance of OR is low, and this stimulation is quite sensitive to small changes in these low amounts of OR (10). Our results corroborate these studies and suggest that b 5 interacts primarily with the P450c17⅐OR complex and not with P450c17 alone. Third, the redox-active core 1 segment of porcine b 5 alone cannot augment the 17,20-lyase activity of human P450c17 (43), consistent with our findings that electron transfer from human b 5 is not required to stimulate 17,20-lyase activity.
Three conclusions about human physiology emerge from our analysis of the kinetics of human P450c17. First, human androgen biosynthesis proceeds predominantly through the pathway 17␣-hydroxypregnenolone 3 DHEA 3 androstenedione, rather than through the pathway 17␣-hydroxypregnenolone 3 17␣-hydroxyprogesterone 3 androstenedione. The pathway via DHEA predominates because the apparent K m for ⌬ 4 17␣-hydroxyprogesterone is about 10-fold higher and its V max is one-tenth as fast as the corresponding values for ⌬ 5 17␣-hydroxypregnenolone. Thus, the catalytic efficiency V max /K m for the 17,20-lyase reaction is nearly 100-fold greater for ⌬ 5 17␣hydroxypregnenolone than for ⌬ 4 17␣-hydroxyprogesterone. Second, significant androgen biosynthesis via the ⌬ 4 pathway can only occur in the presence of very high ⌬ 4 17␣-hydroxyprogesterone concentrations, as found in untreated patients with 21-hydroxylase deficiency (44). Third, considerable microsomal 17,20-lyase activity is found even in the complete absence of b 5 ; therefore, b 5 deficiency cannot cause a syndrome of complete 17,20-lyase deficiency (23) as has been suggested (45).
The structural nature of the interaction of P450c17 with OR is not known, but the x-ray crystal structures of rat OR (46) and P450-BMP (47) provide useful clues. The redox-partner binding site for P450-BMP, a Type II (microsomal) P450, comprises the surface surrounding a depression in the "proximal" face of the protein that extends down to the face of the heme opposite the substrate-binding pocket (47,48). This crevasse is lined on one side with positively charged residues from the JЈ and K helices (in P450-BMP, lysines 325, 328, and 331) which appear to participate in electrostatic pairing with negatively charged residues in OR. Molecular modeling shows that human P450c17 has a similarly located crevasse of positively charged residues that interact with redox partners (23). The electron-donating FMN moiety of rat OR also lies at the base of a concave cleft formed by the butterfly-shaped apposition of the FMN and FAD domains (46). However, the FMN domain joins the remainder of the protein via a disordered, flexible hinge that must flex about 90°for the FMN moiety to extend out from the concave cleft of OR (46) to approach the concave redox-partner binding site of P450c17 (23).
Because b 5 normally participates in redox reactions such as methemoglobin reduction (49) and stearyl-CoA desaturation (50) and can serve as an alternate electron donor in some other P450 reactions (51), our demonstration that b 5 serves a role as an allosteric facilitator of electron transfer from OR to P450c17 was unexpected. The binding of redox partners to P450c17 must transmit subtle changes to the substrate-binding pocket, as evidenced by the lower K m values for ⌬ 5 substrates in the presence of human OR (Table III) and by analogy to the altered regiospecificity of 1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine oxygenation by P450 2D6 in the presence of OR (28). Furthermore, the oxidative scission of the C 17 -C 20 bond of 17␣-hydroxypregnenolone appears to impose much more stringent constraints on the active-site topology of P450c17 than does the 17␣-hydroxylase reaction. Therefore, we propose that FIG. 7. Proposed function of cytochrome b 5 . I, NADPH donates two electrons to the FAD domain of OR (touching the microsomal membrane), which then pass to the FMN moiety. II, the FMN domain of OR, which is connected to the FAD domain by a connecting domain and a hinge region (H) must rotate about 90°( counterclockwise in the figure) to dock with the redox-partner binding site of P450c17. The interaction of P450c17 and OR is adequate to support 17␣-hydroxyation, but this complex rarely adopts the geometry required to catalyze the 17,20lyase reaction. III, the presence of either holo-b 5 or apo-b 5 favors the interaction of OR and P450c17 in an orientation that satisfies the more stringent conformational restrictions required by the 17,20lyase reaction, facilitating productive electron transfer from OR to P450c17 and subsequent catalysis. The precise site(s) of action of b 5 remain unknown. b 5 optimizes the geometry of the P450c17⅐OR complex for the more sensitive 17,20-lyase reaction perhaps by forming a ternary complex (Fig. 7). The structural core 2 domain of b 5 may be the region that stimulates 17,20-lyase activity, as core 2 adopts a similar conformation in the NMR structures of both isolated holo-b 5 (52) and apo-b 5 (53), whereas the heme-binding core 1 domain is disordered in apo-b 5 (53). Furthermore, core 2 retains its overall topology during molecular dynamics simulations of apo-b 5 , while core 1 loses secondary structure and exhibits conformational mobility (54).
A role for b 5 as an allosteric effector protein is consistent with our observation that serine phosphorylation of P450c17 selectively increases 17,20-lyase activity (8) and that mutations of arginine residues in the redox-partner binding site of human P450c17 cause isolated 17,20-lyase deficiency (23). The precise orientation of OR in the electron-donor docking region of P450c17 required to assemble the active oxygenating complex for the 17,20-lyase reaction is impaired by mutation of this surface and enhanced by b 5 or apo-b 5 . Phosphorylation of P450c17 probably favors assembly of productive complexes so that electron transfer is more rapid and coupling efficiency is higher; however, the exact mechanism by which phosphorylated serine residues enhance 17,20-lyase activity is not yet known. A more detailed understanding of these complexes is essential for understanding the regulation of 17,20-lyase activity; this in turn may permit development of agents to inhibit this activity, which will aid in the treatment of sex steroid-dependent malignancies and disorders of androgen excess.