Lack of electron transfer from cytochrome b5 in stimulation of catalytic activities of cytochrome P450 3A4. Characterization of a reconstituted cytochrome P450 3A4/NADPH-cytochrome P450 reductase system and studies with apo-cytochrome b5.

Many catalytic activities of cytochrome P450 (P450) 3A4, the major human liver P450 enzyme, require cytochrome b5 (b5) for optimal rates. The stimulatory effect of b5 on P450 reactions has generally been thought to be due to transfer of electrons from ferrous b5 to the ferrous P450-O2-substrate complex. We found that apo-b5, devoid of heme, could substitute for b5 in stimulating two prototypic activities, testosterone 6β hydroxylation and nifedipine oxidation. The stimulatory effect was not seen with albumin, hemoglobin, catalase, or cytochrome c. Apo-b5 could not substitute for b5 in a testosterone 6β hydroxylation system composed of NADH-b5 reductase and P450 3A4. Rates of electron transfer from NADPH-P450 reductase to ferric P450 3A4 were too slow (<2 min−1) to support testosterone 6β hydroxylation (∼14 min−1) unless b5 or apo-b5 was present, when rates of ∼700 min−1 were measured. The oxidation-reduction potential (Em,7) of the ferric/ferrous couple of P450 3A4 was unchanged (∼−310 mV) under different conditions in which the kinetics of reduction were altered by the addition of substrate and/or apo-b5. Rapid reduction of P450 3A4 required substrate and a preformed complex of P450 3A4, NADPH-P450 reductase, and b5; the rates of binding of the proteins to each other were 2-3 orders of magnitude less than reduction rates. We conclude that b5 can facilitate some P450 3A4-catalyzed oxidations by complexing with P450 3A4 and enhancing its reduction by NADPH-P450 reductase, without directly transferring electrons to P450.

More than 40 P450 1 enzymes are found in a single mammalian species (2). The proteins constitute a superfamily and collectively contribute extensively to the oxidation of xenobiotic chemicals (e.g. drugs, carcinogens, pesticides, alkaloids, and other natural products) and also endobiotics (e.g. steroids, eicosanoids, fat-soluble vitamins, fatty acids) (3)(4)(5)(6). The contributions of these P450 enzymes to metabolism in humans are well recognized, particularly regarding issues of drug clearance (7)(8)(9). There is general agreement that, in most humans, P450 3A4 is the most abundant of the P450s in both liver and small intestine (8,9); it can constitute up to 60% of the total P450 in the liver (10). The intestinal enzyme has been implicated in variation in the bioavailability of many orally administered drugs (11).
P450 3A4 has a very broad range of substrates, with more than 60 drugs having been already identified (9). These vary widely in structure, and one of the questions about this enzyme has been the molecular basis of its broad catalytic specificity (12,13). Other mechanistic questions involve the basis of the sigmoidal plots of enzyme velocity versus substrate seen with some compounds (14 -16) and the stimulation of activity by chemicals other than the substrate (14,17,18). The purified enzyme, along with other P450 3A subfamily enzymes, is much more sensitive to its reconstitution environment than are most other P450s (19,20). A variety of components have been reported to stimulate catalytic activity, including long chain unsaturated phosphatidylcholines (21), phosphatidylserine (20,22), ionic detergents (21,22), GSH (23), divalent cations (24,25), and b 5 (19,21). Not all of these components are directly relevant to the membrane-bound enzyme, but Mg 2ϩ has been shown to stimulate activity of the enzyme in microsomes (24,25) and antibodies raised against b 5 can inhibit some catalytic activities of P450 3A4 in microsomes (19,25). Somewhat surprisingly, certain catalytic activities of P450 3A4 are quite refractory to alterations in lipids and b 5 (24,26).
In order to better understand this complex but important system, we initiated a systematic investigation of some of the system components on individual steps in the catalytic cycle of purified recombinant P450 3A4 (14,19,20,(23)(24)(25). A general conclusion about the role of b 5 in modulating P450 reactions has been that electron transfer from b 5 to P450 occurs in step 4 of Scheme 1 (27). Recently we found qualitative evidence that b 5 could also stimulate the reduction of P450 3A4 by the flavoprotein NADPH-P450 reductase (24). We now report that apo-b 5 (devoid of heme) can replace b 5 in the efficient oxidation of the prototypic P450 substrates testosterone and nifedipine and that apo-b 5 can also replace b 5 in the facilitating electron flow from NADPH-P450 reductase to P450 3A4, in the absence of electron transfer from b 5 or modulation of the E m,7 of P450 3A4. * This research was supported in part by United States Public Health Service Grants R35 CA44353 and P30 ES00267 and grants from the Ministry of Education, Science, and Culture of Japan and the Ministry of Health and Welfare of Japan. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
¶ Supported in part by United States Public Health Service Postdoctoral Fellowship F32 ES05663.

EXPERIMENTAL PROCEDURES
Chemicals-Nifedipine was purchased from Sigma and recrystallized from aqueous C 2 H 5 OH in amber glass (28). Proflavin sulfate and benzyl viologen were purchased from ICN Pharmaceuticals (Plainview, NY), and safranine T was from Fluka (Ronkonkoma, NY).
Apo-b 5 was prepared from rabbit liver b 5 by acid-acetone treatment (34). b 5 (735 nmol, in 1.2 ml of H 2 O) was added dropwise to 25 ml of cold acetone containing 0.2% HCl (Ϫ20°C), and the mixture was stirred for 15 min at 0°C. The white precipitate recovered by centrifugation (10 4 ϫ g, 15 min, 4°C) was dried under an N 2 stream and homogenized in 4.0 ml of 0.20 M potassium phosphate buffer (pH 7.25) containing 0.4% sodium cholate using a hand-held Teflon/glass device. The soluble material recovered from centrifugation (as above) was dialyzed three times versus 100 ml of 0.10 M potassium phosphate buffer containing 0.1 mM EDTA over a period of 2 days, at 4°C. UV-visible spectra and gel electrophoretograms of b 5 and apo-b 5 are shown in Fig. 1.
Spectroscopy-UV-visible spectra were recorded using a Cary 14/ OLIS spectrophotometer (On-Line Instrument Systems, Bogart, GA). Pre-steady-state absorbance measurements were made using an Applied Photosystems SX-17MV instrument (Applied Photophysics, Leatherhead, UK) operating at 37°C. In most cases 90 l of an enzyme mixture were mixed with 10 l of a second solution (e.g. NADPH or O 2 -saturated buffer). In other situations, equal volumes (50 l) from the two syringes were mixed. Data were collected using the Applied Photophysics software system and fitted to single-and bi-exponential plots, with analysis of residuals, using a Marquardt-Levenberg algorithm for nonlinear regression analysis of traces to analytical equations. Results shown are usually the average of three to eight individual reactions.
Anaerobic Methods-The basic method is as described elsewhere (35,36), with Linde research grade argon (Specialty Gases, Louisville, KY; Ͻ1 ppm O 2 ) passed through a tower of BASF catalyst (160°C, catalyst R3-11, Chemical Dynamics Corp., South Plainfield, NJ) and then an Alltech Oxytrap (designed to reduce Ar or CO to Ͻ1 ppm O 2 , Alltech Associates, Deerfield, IL) prior to the manifold. A separate line contained CO gas passed through an Alltech Oxytrap into the manifold. All-glass tonometers (37) or cuvettes were attached to the manifold and alternated between 10 and 12 cycles of vacuum (from a Sargent Welch 1402R pump, Sargent-Welch, Chicago, IL) and argon. In experiments with P450 3A4 and b 5 reduction, 10 cycles were done with argon and finally three with CO (Linde, CP Grade). In these reduction experiments or in determinations of E m,7 , enzyme mixtures included glucose oxidase (15 units ml Ϫ1 ) and catalase (0.18 M); D-glucose (mutarotated) was added to 0.12 M after five cycles of vacuum/argon (without exposure to the atmosphere). In some cases this O 2 scavenging system was replaced by a mixture of 0.7 M protocatechuate dioxygenase and 80 M protocatechuate, with similar results.
An anaerobic solution of photoreduced safranine T, in Tris-EDTA buffer, was used to deplete O 2 from the syringes and lines of the instrument prior to introduction of anaerobic protein samples, utilizing a lamp with the dye stored in the drive syringes.
Measurement of E m,7 -The general procedure involved equilibration with dyes of known potential and is described elsewhere (35,37). In  7 . In a trial experiment with recombinant human P450 1A2 (38), values of n ϭ 0.82 and E m,7 ϭ Ϫ313 mV were obtained (cf. Ϫ299 and Ϫ357 mV for rat and rabbit P450 1A2, respectively) (35,39).
Steady-state Kinetic Experiments-With many of the reactions catalyzed by P450 3A4, the system is sensitive to the concentrations of individual components and even the order of mixing. The following order of mixing was found to be optimal and was used here (all concentrations are for the final incubation mix): P450 3A4 (40 nM), NADPH-P450 reductase (120 nM), b 5 (80 nM), sodium cholate (to 0.5 mM, diluted from a 25 mM stock), and a phospholipid mixture (to 20 g ml Ϫ1 , diluted from a stock of 0.5 mg ml Ϫ1 and consisting of a 1:1:1 mixture (by mass) of L-␣-dioleoyl-sn-glycero-3-phosphocholine (DLPC) and bovine brain phosphatidylserine (sonicated under argon); these components were mixed and allowed to stand 10 min at room temperature, followed by the addition of potassium HEPES (50 mM, pH 7.7), substrate (200 M testosterone or nifedipine), GSH (3.0 mM), and MgCl 2 (30 mM). These components were mixed and incubated for 3 min at 37°C; reactions were initiated by the addition of an NADPH-generating system consisting of (final concentrations) 10 mM glucose 6-phosphate, 0.5 mM NADP ϩ , and 1 IU yeast glucose 6-phosphate dehydrogenase ml Ϫ1 .
Incubations with the substrates nifedipine and testosterone were generally done for 10 min at 37°C. Reactions were terminated and the products were analyzed by HPLC using methods described elsewhere (12,19).
Reduction Kinetics-In general, the components were mixed in the same order as described for steady-state reactions, except that the concentrations of P450 3A4, NADPH-P450 reductase, and b 5 were 0.35, 0.80, and 0.35 M, respectively, unless indicated otherwise. All experiments were done in the stopped-flow apparatus at 37°C under a CO atmosphere (positive pressure), with data collected at 447 nm (Fe 2ϩ ⅐CO complex). In some cases the disappearance of the oxidized P450 (Fe 3ϩ ) was monitored at 390 nm. Reduction of b 5 was monitored by the increase in A 424 or decrease in A 409 . An alternate system involved enzymatic reduction of NADP ϩ by reduced P450. The cuvette was made up with 0.25 M NADPH-P450 reductase, plus all of the other usual components, 0.15 M NH 4 Cl, and 6.7 mM ␣-ketoglutarate. After the system was rendered anaerobic, NADPH (33 M) was tipped from a side arm into the main compartment and the course of reduction of the P450 3A4 was monitored (⌬A 395 , ⌬A 418 ) in the Cary 14 spectrophotometer (ϳ30 min at 23°C). L-Glutamate dehydrogenase was then tipped into the main compartment (final concentration, 0.02 IU ml Ϫ1 ) from a second side arm to remove residual NADPH. This process was monitored by the decrease in A 340 . The rate of oxidation of ferrous P450, which occurred subsequently, was also monitored (increase in A 395 and decrease in A 418 ). Under these conditions, the rate of oxidation of photochemically reduced P450 3A4 was negligible.

Steady-state Assays: Stimulation of Catalytic Activities by
Apo-b 5 -Apo-b 5 failed to substitute for b 5 in a testosterone 6␤-hydroxylation system consisting of NADH-b 5 reductase, b 5 , and recombinant P450 3A4 (Fig. 2), because NADH-b 5 reductase cannot directly transfer electrons to P450 3A4 (25). However, apo-b 5 was as effective as b 5 in stimulating the testosterone 6␤ hydroxylation activity of recombinant P450 3A4 in a system containing NADPH-P450 reductase ( Table I). The need for Mg 2ϩ in this system was seen, as reported earlier (24). When P450 3A4 purified from human liver was substituted for recombinant P450 3A4, the same patterns of dependence upon assay components were seen. Apo-b 5 was nearly as effective as b 5 , on a concentration basis, in stimulating testosterone 6␤ hydroxylation in a system containing NADPH-P450 reductase and P450 3A4 (Fig. 3A). Similar enhancement was seen with another P450 3A4 reaction, nifedipine oxidation (Fig. 3B). It is also of interest to note that maximal stimulation was observed at a ratio of 0.25-0.5 b 5 or apo-b 5 per P450.
Neither bovine nor human serum albumin nor several hemoproteins (cytochrome c, hemoglobin, catalase) could substitute for the stimulatory effect of b 5 or apo-b 5 (Table II). The effect of both b 5 and apo-b 5 was to decrease the K m (for testosterone) and to increase V max (Table III).
Kinetics of P450 3A4 Reduction: Reconstituted System-Previous work had qualitatively demonstrated the ability of b 5 to enhance rates of P450 3A4 reduction by NADPH-P450 reductase (24,25). These results were confirmed and extended here. P450 3A4 reduction was rapid and appeared to be first-order under what seem to be optimal conditions, identical to those used in steady-state assays, with a rate of ϳ750 min Ϫ1 at 37°C   systems containing P450 3A4 A standard reaction mixture (0.25 ml) consisting of P450 3A4 (10 pmol), NADPH-P450 reductase (20 pmol), b 5 or another indicated protein (20 pmol), the lipid mixture (20 g ml Ϫ1 ), sodium cholate (0.5 mM), MgCl 2 (30 mM), and GSH (3.0 mM) in 50 mM potassium HEPES buffer (pH 7.4) was used for the assay of testosterone 6␤ hydroxylation activities. Other proteins (10 pmol) were added to the preincubation mixture, and GSH was added after the preincubation. Reactions were initiated by adding 1 mM NADPH and terminated after 10 min with CH 2 Cl 2 .  (Fig. 4A). A similar rate was measured in the presence of apo-b 5 (Fig. 4B). As noted earlier (24), reduction was extremely slow in the absence of b 5 or apo-b 5 (Fig. 4C). The results of these and similar experiments are summarized in Table IV. Most of the reductions could be fitted well to single-exponential equations (e.g. Fig. 4), in contrast to our previous work (24,25) and much of the existing literature on P450 reduction (40 -43), where reactions are biphasic. Rapid reduction (ϳ700 min Ϫ1 ) required the presence of substrate (testosterone or aflatoxin B 1 ) and either b 5 or apo-b 5 (Table IV). As noted earlier (24), the presence of ethylmorphine (a P450 3A4 substrate for which b 5 does not stimulate oxidation) enhanced reduction in the absence of b 5 . In some cases b 5 reduction was also examined (⌬A 424 ). b 5 reduction was rapid, with a rate on the same order as that of P450 3A4.
Reduction of NADPH-P450 Reductase by Reduced P450 3A4 -We noted in preliminary experiments with human P450 1A2 that ferrous P450 appeared to be slowly oxidized in the presence of NADP ϩ , apparently via the reverse reaction of NADPH-P450 reductase. To our knowledge this back reaction has not been identified previously, although it should be thermodynamically favorable when a reduced, low-potential P450 is mixed with NADP ϩ . In preliminary experiments, we found that L-glutamate dehydrogenase could oxidize NADPH in a mixture of ␣-ketoglutarate and NH 4 ϩ with a t1 ⁄2 of 40 s at 23°C (k ϳ 1.5 min Ϫ1 ). A mixture of P450 3A4 (1.65 M), NADPH-P450 reductase (0.44 M), ␣-ketoglutarate, NH 4 ϩ , and all typical reconstitution components (including 100 M testosterone) was made anaerobic and then reduced by the addition of NADPH, a process that was monitored by the absorbance spectra. L-Glutamate dehydrogenase was then added and the oxidation of the NADPH (⌬A 340 ) and P450 3A4 (⌬A 395 , ⌬A 418 ) were monitored spectrophotometrically. The t1 ⁄2 for P450 3A4 oxidation was 13 min, corresponding to k ϭ 0.05 min Ϫ1 , which should represent the rate of electron flow to NADP ϩ (a similar rate was estimated with recombinant human P450 1A2).
Estimation of E m,7 of P450 3A4 -Several different experimental designs were utilized in measuring the E m,7 of P450 3A4, including the addition of dyes of varying potential at low concentrations (Ͻ1 M, with safranine T reduction monitored by fluorescence with excitation at 520 nm and emission at 580 nm) or high concentrations (with safranine T and benzyl viologen reduction both measured by visible spectroscopy). The results shown in Fig. 5 are representative and presented to demonstrate the proximity of the E m,7 of P450 3A4 to that of safranine T. In this particular experiment, done in the presence of testosterone and apo-b 5 , the E m,7 was Ϫ312 mV (n ϭ 0.8) (mean of three experiments). For comparison, the estimated E m,7 for a similar system devoid of apo-b 5 and testosterone was Ϫ316 (Ϯ7) mV (n ϭ 1.2) and Ϫ320 mV (n ϭ 0.9) for a system devoid of apo-b 5 but with ethylmorphine substituted for testosterone.
Kinetics of Interaction of P450 3A4 with NADPH-P450 Reductase and b 5 -The reduction experiments mentioned previously had all been done with preformed complexes of P450 3A4 with other enzymes, mixing with a solution of NADPH from the second reaction syringe. When NADPH-P450 reductase was moved to the NADPH syringe (and P450 3A4 and b 5 were in the other syringe), reduction was very slow (Fig. 6A, k ϭ 1.2 min Ϫ1 ). The experiment was repeated with P450 3A4 and NADPH-P450 reductase in one syringe and with b 5 and NADPH in the other (all other components in both). Again, reduction was very slow (Fig. 6B, k ϭ 0.2 min Ϫ1 ).
The results indicate that the three protein components associate rather slowly. If this is true, then one might expect to observe "burst" kinetics for the reduction of P450 3A4 and b 5 in a system in which the reductase is the limiting component. This is indeed the case, as shown by the results presented in Fig. 7. The reduction of P450 3A4 showed a rapid burst (Fig.  7A), followed by a lag and then a single-exponential reduction of the rest of the P450. The rates of the two phases were 920 and 0.7 min Ϫ1 . With b 5 , part is reduced at 830 min Ϫ1 and the rest at 3.8 min Ϫ1 . The concentrations of P450 3A4, b 5 , and reductase were 1.0, 0.7, and 0.4 M, but the amplitudes in the different phases do not match the concentrations of the components. This phenomenon could be due to the lack of 1:1 stoichiometry in clusters of protein partners that participate in the initial fast burst phase of reduction. DISCUSSION Evidence for the involvement of b 5 in a P450 catalytic system was first published in 1971 (44) and the literature now contains   numerous reports of P450 reactions in which there is either (i) enhancement, (ii) no effect, or (iii) inhibition by b 5 (45)(46)(47)(48)(49)(50)(51)(52). Evidence for the role of b 5 in the transfer of an electron to the P450 ferrous-O 2 complex has been presented (44,53), and this is almost universally accepted to be the basis for enhancement of catalysis by b 5 (Scheme 1). In several studies, derivatives of b 5 that were devoid of electron transfer capability were not effective in stimulating reactions catalyzed by other P450 s (21,47). We have previously presented qualitative evidence for the stimulation of the reduction of ferric P450 3A4 by b 5 (24). The possibility exists that this phenomenon could occur via transfer of electrons from b 5 to P450 3A4, although the thermodynamics are unfavorable (i.e. the E m,7 of b 5 is ϳϩ20 mV and that of most P450s is ϳ Ϫ300 mV) (35,54). Rates of steady-state reactions and ferric P450 3A4 reduction in systems composed of NADH-b 5 reductase and b 5 are too slow to be compatible with the rates of NADPH-P450 reductase-dependent reactions (25). We now provide evidence that apo-b 5 , devoid of electron transfer capability, can stimulate steady-state catalytic activity and the reduction of P450 3A4 (Tables I-III, Figs. 3 and 4). This effect on reduction rate may contribute to the enhancement of P450 3A4-catalyzed testosterone 6␤-hydroxylation. These effects were also seen with the substrates nifedipine and aflatoxin B 1 (Fig. 3, Table IV), reactions that require b 5 (14,19).
With ethylmorphine, a substrate whose oxidation does not require b 5 (24,26), the substrate enhanced P450 3A4 reduction to the rate seen with a combination of testosterone plus b 5 (Table IV).
The enhancement of the rate of P450 3A4 reduction is apparently not linked to a change in E m,7 of the ferric/ferrous couple (Fig. 5). Although the spin-state, kinetics, and thermodynamics of P450 reduction are often suggested to be linked, based upon work with bacterial P450 101 (55), this is not generally the case with microsomal P450 s (35), and it is not the case here. Neither testosterone, ethylmorphine, nor apo-b 5 appeared to change the E m, 7 , even though about one-half of the P450 3A4 was converted from the low spin to high spin iron configuration by the addition of testosterone, and it appears possible to have two orders of magnitude difference in reduction rates in the absence of a change in E m,7 . Other literature argues for lack of linkage of substrate binding, spin state, E m,7 , and rate of reduction (35,39,41), and it would appear this dependence is not very common among P450s.
Previously evidence has been presented for slow association and dissociation of P450s and NADPH-P450 reductase (40,41). We also found that these were relatively slow processes with P450 3A4 for both NADPH-P450 reductase and b 5 (Fig. 6). However, the kinetics vary among P450s. For instance, the rate measured for association of P450 3A4 and NADPH-P450 reductase was 1.2 min Ϫ1 with P450 3A4 in the experiment of Fig. 6A but 30 min Ϫ1 with human P450 1A2 in a separate experiment (results not shown, reduction rate of 720 min Ϫ1 for preformed complex). Such differences may have relevance when mixtures of P450s are considered in terms of competition for limiting reductase (56,57). When limiting NADPH-P450 reductase was mixed with excess P450 3A4 and b 5 and NADPH was added from a second syringe, a rapid burst of reduction was seen for each hemoprotein, followed by slow reduction dominated by association/dissociation kinetics (Fig. 6). It should also be pointed out that many of the kinetic traces of the reduction of preformed P450 or P450⅐b 5 complexes could be fit to first order (single exponential) equations (e.g. Fig. 5). The first accounts of the reduction kinetics of P450s described biphasic kinetics (58,59). Successive reports also present biphasic kinetics even with purified P450 s (24, 25, 40 -42, 49, 52, 60), although one of these reports indicates that 90% of the reduction was done in the fast phase (60). Most of these studies were done at 25°C or less, and part of the reason we saw primarily single exponential reduction kinetics may be due to the 37°C temperature we used (for comparison to steady-state results). Lipid phase transitions and related phenomena may be operative at lower temperatures. However, in systems where the amount of reductase is limiting, diffusion is a likely explanation (42) rather than inherent properties of the P450(s) or reductase.
The b 5 stimulation of ferric P450 reduction does not seem to be general. We have not seen this effect in other studies with recombinant human P450 1A2 or 2E1. Reduction of P450 2E1 is slower in the presence of b 5   NADPH-P450 reductase-catalyzed reduction of b 5 . The postulate of "simultaneous" 2-electron reduction of ferric P450 and b 5 presented in that article (i.e. sequential rapid reduction of ferric P450, transfer of an electron from ferrous P450 to ferric b 5 , and reduction of ferric P450) is of interest in light of the similarity of the rates of reduction of P450 3A4 and b 5 seen in some of the experiments described in Table IV. However, this hypothesis does not seem to generally apply here. CO traps ferrous P450 but b 5 has a similar rate of reduction as P450 in the presence of CO. 2 Also, in other studies (not presented) with a recombinant P450 3A4:NADPH-P450 reductase fusion protein (62), reduction of b 5 was faster than P450.
In conclusion, b 5 has been shown to stimulate P450 3A4catalyzed reactions without obligate electron transfer. In these cases (testosterone 6␤ hydroxylation, aflatoxin B 1 oxidation) the mechanism of stimulation appears to involve enhancement of rates of ferric P450 reduction, a process which seems not to depend on the oxidation state of or transfer of electrons from b 5 . A mechanistic proposal has been presented previously (24) and still appears to be valid as a working model. As suggested by Lipka and Waskell (48), different substrates may preferentially bind to a P450, P450 3A4 in this case, to induce b 5 binding. Alternatively, b 5 may bind to the P450 in the absence of the substrate but the mode of interaction may be altered in the ternary P450-substrate-b 5 complex. This binding evidently influences the transfer of electrons from NADPH-P450 reductase to the ferric enzyme and probably the ferrous-O 2 complex.
However, some substrates such as ethylmorphine appear to be able to bind to P450 3A4 and induce rapid reduction in the absence of b 5 (Table IV).