Stimulation of cytochrome P450 reactions by apo-cytochrome b5: evidence against transfer of heme from cytochrome P450 3A4 to apo-cytochrome b5 or heme oxygenase.

Many cytochrome P450 (P450)-dependent reactions have been shown to be stimulated by another microsomal protein, cytochrome b(5) (b(5)). Two major explanations are (i) direct electron transfer from b(5) and (ii) a conformational effect in the absence of electron transfer. Some P450s (e.g. 3A4, 2C9, 17A, and 4A7) are stimulated by either b(5) or b(5) devoid of heme (apo-b(5)), indicating a lack of electron transfer, whereas other P450s (e.g. 2E1) are stimulated by b(5) but not by apo-b(5). Recently, a proposal has been made by Guryev et al. (Biochemistry 40, 5018-5031, 2001) that the stimulation by apo-b(5) can be explained only by transfer of heme from P450 preparations to apo-b(5), enabling electron transfer. We have repeated earlier findings of stimulation of catalytic activity of testosterone 6beta-hydroxylation activities with four P450 preparations, in which nearly all of the heme was accounted for as P450. Spectral analysis of mixtures indicated that only approximately 5% of the heme can be transferred to apo-b(5), which cannot account for the observed stimulation. The presence of the heme scavenger apomyoglobin did not inhibit the stimulation of P450 3A4-dependent testosterone or nifedipine oxidation activity. Further evidence against the presence of loosely bound P450 3A4 heme was provided in experiments with apo-heme oxygenase, in which only 3% of the P450 heme was converted to biliverdin. Finally, b(5) supported NADH-b(5) reductase/P450 3A4-dependent testosterone 6beta-hydroxylation, but apo-b(5) did not. Thus, apo-b(5) can stimulate P450 3A4 reactions as well as b(5) in the absence of electron transfer, and heme transfer from P450 3A4 to apo-b(5) cannot be used to explain the catalytic stimulation.

P450 1 enzymes are widespread in nature and well known for their catalytic versatility (2)(3)(4)(5). Interest in these mammalian enzymes is strong because of their critical roles in the metabolism of drugs, steroids, and carcinogens (2,3,5,6). The most common mode of catalysis is mixed-function oxidation, which involves the use of both electrons and O 2 as substrates.
Electron input from pyridine nucleotides into P450s usually follows one of two general modes. The first, which is common in some bacterial and mitochondrial P450s, involves transfer of electrons from a flavoprotein to an iron-sulfur protein to P450 heme. The other major reaction, which is common in the microsomal P450s, involves electron transfer from a 2-flavin flavoprotein (NADPH-P450 reductase) into the P450 heme. Some P450-catalyzed reactions are stimulated by b 5 , as first discovered in a series of studies in which NADH stimulated NADPHdependent reactions in liver microsomes (7). The role of b 5 in some reactions was later reinforced in reconstitution (8 -10) and immunochemical inhibition experiments (11). The occurrence and mechanism of b 5 involvement in P450-catalyzed reactions have been studied extensively in the past three decades (12)(13)(14)(15)(16)(17). Some P450 reactions involve b 5 , but others do not, even with the same P450 enzyme (10,18). An added complexity is that NADPH-P450 reductase reduces b 5 efficiently as well as P450s, rendering the path of electron transfer difficult to establish in most cases. In general, direct electron transfer to ferric P450 is inefficient because of the unfavorable difference in oxidation potentials (19,20). Until recently, the generally accepted mechanism for b 5 stimulation has been introduction of an electron into the P450 FeO 2 2ϩ complex, which is considered to have a much higher oxidation-reduction potential than the P450 Fe 3ϩ /Fe 2ϩ couple (19 -22). Early studies with liver microsomes showed changes in the redox state of b 5 (7,23). Other early experiments with b 5 substituted with redox-inactive metals were also used to argue in favor of an electron transfer mechanism. However, this issue has been difficult to address directly because of the kinetic complexities in mixing b 5 with P450 in the P450 FeO 2 2ϩ form (24 -26). In 1995, we reported that the reduction of ferric P450 3A4 required the presence of b 5 for maximum rates, as well as a substrate (testosterone) (27). The following year, in the course of doing what were intended to be controls for other experiments, we found that apo-b 5 was as effective as b 5 in stimulating P450 3A4-catalyzed testosterone 6␤-hydroxylation and nifedipine oxidation (22). We extended this work and reported that P450 2C9 reactions were stimulated by apo-b 5 as well as b 5 , but with P450 2E1, only b 5 was stimulatory (28). Other laboratories reported the stimulation of P450s 17A (29) and 4A7 (30, 31) by apo-b 5 . Recent work in our own groups has also shown stimulation of catalytic activities of P450s 2A6, 2B6, 2C8, 2C19, and 3A5 by both b 5 and apo-b 5 . 2 (Some P450s have not shown effects of b 5 or apo-b 5 , i.e. 1A1, 1A2, 1B1, and 2D6). 2 The conclusion regarding these experiments with apo-b 5 is that electron transfer (to P450) by b 5 is not involved in the stimulation (22,29,31). The effect of b 5 varies depending upon the particular P450 3A4 system. Enhancement of some (but not all (27,32)) catalytic activities is seen with either recombinant P450 3A4 or the enzyme purified from liver microsomes (33,34). The extent of stimulation does vary in different systems, and the reconstituted systems with higher activities appear to show less stimulation (22,35). However, the effect is usually 2-4-fold. Bacterial or baculovirus-based insect cell membranes in which P450 3A4 and NADPH-P450 reductase are simultaneously expressed have high catalytic activities (including testosterone 6␤-hydroxylation activity) in the absence of b 5 (25,36), although b 5 can be added to the membranes to further stimulate activity (37). The critical issue here is what the role of b 5 is in liver (or other tissues), and studies with anti-b 5 antibodies have shown a strong contribution of b 5 to catalytic activity in liver microsomes, regardless of the mechanism (33,35).
The conclusions reached with apo-b 5 may or may not be relevant to microsomes. However, these stimulatory effects have been observed with numerous purified P450 preparations (3A4, 2C9, 17A, and 4A7) (22, 28 -31), and conclusions about modulation of catalytic activities are not surprising in light of our current general understanding of protein-protein interactions. Some P450s have even shown catalytic stimulation by other P450s (28,38). However, very recently, Guryev et al. (39) have questioned the conclusions of the apo-b 5 studies (22, 29 -31, 37). They repeated the experiments cited above with P450s 3A4 and 17A and also found stimulation of catalytic activities by apo-b 5 . Some transfer of heme to apo-b 5 was also found; the issue of whether the source was the P450 itself (3A4 or 17A) or adventitiously bound heme was unresolved. These investigators also added heme to purified apo-b 5 and apomyoglobin and characterized the binding as rapid and tight. A key experiment was the subsequent use of apomyoglobin as a heme scavenger in blocking stimulation by apo-b 5 (39).
The results of Guryev et al. (39) are not in accord with the previously held general view about the lack of electron transfer in the stimulation of some reactions by apo-b 5 . Concerns involve the report of the presence of excess heme (30%?) in the preparations and the relatively low fraction of transfer of heme (ϳ 10%) (39). We address some of the issues raised in that work, those that bear directly on the issues of how much heme transfer contributes to the stimulation of P450 activities by apo-b 5 and b 5 . Other experiments presented (39) do not directly address the question and are not considered here (e.g. binding to P450, truncations, mutants and use of alternate metals, and reduction in the absence of substrate); differences between apo-b 5 , b 5 , and b 5 derivatives may seem subtle but are known to have cooperative structural effects (40) that may not be readily interpreted.
We considered a set of P450 3A4 preparations made since the earlier work (22,28) because our purification methodology has been changed. In these preparations we account for nearly all of the measured heme as P450, although some key experiments were done with a preparation of one of the stocks in which 30% of the P450 had converted to cytochrome P420. Our earlier reported stimulatory effects of apo-b 5 were fully repeatable, and we proceeded to focus on events that could occur in the time frame used to see the stimulations, e.g. reconstitution and assay. The transfer of 3-5% of the heme from P450 3A4 preparations does occur with apo-b 5 or the enzyme heme oxygenase, with conversion of heme to biliverdin in the latter case. This fraction is attributed to trace cytochrome P420 and adds evi-dence to the view that heme oxygenase, which is normally involved in the physiological degradation of free heme, does not act directly on P450 heme. The lack of an effect of highly purified apomyoglobin as a heme scavenger and the inability of apo-b 5 to support NADH-b 5 reductase/P450 3A4-catalyzed testosterone 6␤-hydroxylation provide more evidence against an obligatory role for heme transfer to apo-b 5 in the stimulations of P450s.
Spectroscopy and HPLC-UV-visible spectra were recorded using OLIS-Cary 14 and -Amino DW2 instruments (On-Line Instrument Systems, Bogart, GA). HPLC was done with a Spectra-Physics 8700 pumping system and ThermoSeparation 6000 rapid scanning monochromator or a JASCO PU-980 pumping system (Japan Spectroscopic Co., Tokyo, Japan) and Chromatopac C-R3A detector (Shimadzu, Kyoto, Japan).
E. coli-expressed human heme oxygenase (devoid of 23 residues at the C terminus) (51) was a gift of K. Auclair and P. R. Ortiz de Montellano (University of California, San Francisco, CA). The heme in the sample (36 nmol of enzyme) was converted to biliverdin by incubation with an equal concentration of NADPH-P450 reductase in the presence of 0.10 M potassium phosphate buffer (pH 7.4) and an NADPH-generating system (vide infra). Spectral analysis indicated that the reaction was complete in the first minute. The mixture (in 3 ml) was applied to a 1.2 ϫ 60-cm Sephadex G-10 column equilibrated with 50 mM potassium phosphate (pH 7.4) containing 0.10 mM EDTA and eluted with the same. The void volume fraction was detected by A 280 measurements; spectral analysis indicated the apparent presence of stable NADPH-P450 reductase flavin semiquinone. The sample was concentrated (to 1.9 ml) using an Amicon-Millipore ultrafiltration device equipped with a PM-10 membrane (Millipore Corp., Bedford, MA). Protein analysis indicated nearly complete recovery of the apo-heme oxygenase and NADPH-P450 reductase. This preparation of apo-heme oxygenase was used with hemin chloride, and a preliminary spectral analysis indicated that it was capable of forming biliverdin. NADPH-P450 reductase was added to the apo-heme oxygenase preparation in subsequent work with P450 3A4.

RESULTS AND DISCUSSION
Stimulation of P450 3A4 Catalytic Activity by Apo-b 5 -Since our first report of the stimulation of P450 3A4 activities by b 5 and apo-b 5 (22), the preparation of recombinant P450 3A4 has been changed in this laboratory. A C-terminal (His) 5 tag had been incorporated to facilitate purification (44). Purification of P450s is done in the absence of nonionic detergents because these compounds have been found to be substrates for P450s, particularly P450 3A4 (55). Four preparations were used in this work. All had been prepared during the time period 1998 -2001 and stored at Ϫ20°C. All were relatively low in P420, as indicated in some of the figures presented later (P420 indicates spectrally distinct denatured P450 or loosely bound heme). Analysis of these preparations for P450 (⌬⑀ 450 -490 ϭ 91 mM Ϫ1 cm Ϫ1 for Fe 2ϩ ⅐CO versus Fe 2ϩ ) and heme (pyridine hemochrome ⌬⑀ 557-575 ϭ 32.4 mM Ϫ1 cm Ϫ1 ) gave the following respective concentrations in the preparations: 21 (Fig. 1D). Thus, the heme appears to be present nearly completely as P450, assuming the accuracy of these extinction coefficients (52,53) and allowing for what is probably ϳ5% error in replicate measurements.
With all four of these P450 3A4 preparations, b 5 and apo-b 5 both stimulated testosterone 6␤-hydroxylation nearly equally (Fig. 1). The extent of stimulation was 3-4-fold, similar to that reported earlier with other preparations of these proteins (22) and in the report of Guryev et al. (39).
Maximum stimulation of testosterone 6␤-hydroxylation was achieved with a ratio of ϳ1:1 b 5 or apo-b 5 to P450, similar to that seen earlier with P450 3A4-catalyzed testosterone 6␤hydroxylation (22) and some catalytic activities of P450 2C9 (28). For unknown reasons, the stimulation of P450 3A4-catalyzed nifedipine (ring) oxidation is maximally stimulated by lower ratios of b 5 or apo-b 5 to P450 (22,28). It is conceivable that only a fraction of the P450 3A4 binds nifedipine and participates in the reaction (56), but we have no direct evidence to support this possibility.
Analysis of Movement of Heme into b 5 -The proposal of Gu-ryev et al. (39) regarding the transfer of heme from P450 3A4 or heme adventitiously bound to P450 3A4 into apo-b 5 was considered because that work showed that the incorporation of heme into apo-b 5 was a rapid process. However, consideration of the known three-dimensional structures of prokaryotic P450s and P450 2C5 indicates that the heme prosthetic group is embedded inside of the protein (57,58). Rapid loss of heme from P450s would appear to be highly unlikely, given the known thiolate axial ligation. A rapid exchange of P450 3A4 heme with its environment would make the purification of the holoenzyme nearly impossible.
If heme transfer to apo-b 5 is the explanation for the stimulation of P450 3A4 reactions by apo-b 5 (39), then the transfer must be rapid because the reconstitution/assay process is complete within 10 -20 min. We restricted most analyses to this time frame (Ͻ30 min). The spectra presented in Fig. 2 were recorded after incubations of all system components (except NADPH) under the normal reconstitution conditions. The spectra obtained with apo-b 5 (Fig. 2, C and D) match those obtained in the absence of b 5 (Fig. 2, A and B) and not those obtained with b 5 (Fig. 2, E and F), as judged by the absence of the peaks at 424 and 550 nm. It should also be noted that the P450 spectrum was not decreased in the presence of apo-b 5 (Fig. 2D).
However, the transfer of traces of heme into b 5 might not have been detected in the design used in Fig. 2. We also used a protocol employed by Guryev et al. (39) in which a spectrum of an S 2 O 4 2Ϫ -reduced mixture of P450 3A4/b 5 was recorded versus S 2 O 4 2Ϫ -reduced P450 3A4 (Fig. 3). In this assay, a small peak at 424 nm was detected when a mixture of apo-b 5 and P450 3A4 was reduced, possibly indicating b 5 formation. The absorbance of this peak is ϳ5% of that seen when a b 5 solution was used at the same concentration.
We conclude that a finite but small amount of b 5 might have been formed due to transfer of heme to apo-b 5 . In considering the results presented in Fig. 1, 5% incorporation of heme into apo-b 5 cannot explain the stimulatory effects of apo-b 5 because the amounts of b 5 and apo-b 5 needed to achieve a given amount of stimulation are nearly identical.
Lack of Function of Apo-b 5 in NADH-b 5 Reductase/P450 3A4-catalyzed Testosterone 6␤-Hydroxylation-Some P450 reactions can be supported with NADH-b 5 reductase and b 5 substituted for NADPH-P450 reductase, with slower rates (59). In these systems, b 5 is required because NADH-b 5 reductase cannot directly transfer electrons to P450s, including P450 3A4 (35). We previously showed that P450 3A4 can catalyze b 5 -dependent testosterone 6␤-hydroxylation in such a system (35) but that apo-b 5 did not substitute for b 5 (22).
This experiment was repeated with new proteins, and the results were identical (Fig. 4). If heme were transferred from P450 3A4 or anything in the P450 3A4 preparation to convert apo-b 5 to b 5 , the apo-b 5 should have been functional in this reaction. This is clearly not the case here, with the limit of ϳ5% contribution (Fig. 4).
Lack of Inhibition of P450 3A4-supported Reactions by Apomyoglobin-One of the critical arguments used by Guryev et al. (39) to support the view of the necessary transfer of heme from P450 3A4 to apo-b 5 was the inhibition of apo-b 5 stimulation of P450 3A4-catalyzed testosterone 6␤-hydroxylation by apomyoglobin. Apomyoglobin rapidly binds heme in solution and was used as a scavenger. Guryev et al. (39) reported that apomyoglobin, added at the concentration of b 5 or apo-b 5 , inhibited the stimulation of catalytic activity by apo-b 5 but not by b 5 .
We examined this phenomenon with both testosterone 6␤hydroxylation and nifedipine oxidation because of the possible significance of the report (Fig. 5). The two P450 3A4 preparations used were those from Fig. 1, but in Fig. 5, B and D, we used a stock in which the P420 content was 30% of the total P450 plus P420 to investigate the possibility that potentially looser binding of heme might facilitate the heme transfer process proposed by Guryev et al. (39).
The clear difference with the result reported by Guryev et al. (39) is unexplained. Clearly one of the most significant differences in the systems studied here and in that work is the matter of the extra heme in the preparation reported by those authors (39). As indicated above, one of the stocks was picked to be one in which P420 was increased (probably due to use with repeated thawing and refreezing), but similar results were obtained (Fig. 5, B and D). Sigma sequencing grade horse apomyoglobin was utilized in our work (this is a standard for optimization of amino acid sequenators). Our work was all done with rabbit b 5 and apo-b 5 instead of human apo-b 5 (39), although the stimulation of P450 activities by b 5 /apo-b 5 has been demonstrated with proteins from varying sources (22,29,31), and the difference is not expected to account for the varying results.
Interaction of Heme Oxygenase with Heme Bound to P450 3A4 -In the early literature involving the enzyme heme oxygenase (the enzyme that degrades free heme to biliverdin IX␣ and CO), the inference was often made that heme oxygenase degrades the heme in P450 (60,61). Metals such as Co 2ϩ induce heme oxygenase and decrease total P450 in rats, but the relationship of these events is not causal. We reported that partially purified hog spleen heme oxygenase has little if any catalytic activity toward purified (rat) P450 2B1 (62). Kutty et In A, C, and E, a baseline was recorded, and then solid Na 2 S 2 O 4 was added to the sample cuvette; spectra were recorded (Fe 2ϩ versus Fe 3ϩ ). In B, D, and F, CO gas was bubbled through the sample cuvette for 60 s, and a baseline was recorded. Na 2 S 2 O 4 was added to the sample cuvette, and spectra were recorded (Fe 2ϩ ⅐CO versus Fe 3ϩ ). The spectra did not change significantly for C and D after preincubation for 15-75 min before obtaining spectra. T, testosterone, T6␤OH, 6␤-hydroxy-testosterone (t R 3.45 min, distinct from the t R 3.33 min peak observed in A and B but obscured by T6␤OH in C). The reaction measured in C corresponds to a rate of 0.67 nmol product formed min Ϫ1 (nmol P450 3A4) Ϫ1 .
al. (63) reported that P450s 2B1 and 1A1 are substrates for rat heme oxygenase, with heme being converted to biliverdin. However, the P450 preparations contained P420, and the assays were done in the presence of detergents (Emulgen 911) in the absence of glycerol, conditions long known to convert P450 to P420 (64). Under these conditions, 35% of the added P450 2B1 was destroyed, and most (70%) was reported to be converted to biliverdin. However, the obligate conversion of P450 to P420 in this process could not be ruled out, as admitted by the authors (63). In the context of current knowledge of the crystal structures of P450 and the position of the heme, the direct access of P450-bound heme to heme oxygenase appears to be highly unlikely. As discussed earlier, the facile exchange of P450 heme with its medium also appears to be highly unlikely.
The analysis of heme loosely bound to heme proteins is not trivial, including P450s. Kutty et al. (63) reported that the Fe 2ϩ ⅐CO versus Fe 2ϩ difference spectra of free heme, heme albumin (bovine), heme added to P450 2B1, and P420 had discernable spectra with wavelength maxima at 409, 414, 415, and 420 nm, respectively. However, we detected wavelength maxima for free heme, methemalbumin, and heme added to P450 3A4 at 413, 413, and 422 nm, respectively (using the peak finder or derivative function of the OLIS software). As discussed some time ago (64), release of heme from its thiol ligation results in the loss of a P450 spectrum (450 nm for Fe 2ϩ ⅐CO) and the appearance of a band in the region of ϳ420 nm. The term P420 is rather operational, and spectra do not discern among P420 proteins with heme bound at various sites, inside and on the outside of proteins (see also Ref. 65 regarding the variablity of spectra of P420 reconstituted from apo-P450 2B4 and heme).
If P450 3A4 were able to freely exchange heme with the medium or had excess heme that could be accessible to apo-b 5 or apomyoglobin, then this should also be accessible to heme oxygenase, which would produce biliverdin in the presence of NADPH-P450 reductase and NADPH. We therefore decided to utilize heme oxygenase as a heme scavenger, similar to apo-myoglobin. However, interpretation of the catalytic inhibition with high concentrations of heme oxygenase would be complicated by the competition of heme oxygenase with P450 for NADPH-P450 reductase, and we restricted our analysis to the destruction of P450 heme and formation of biliverdin. To increase the sensitivity of the assays, we converted the heme bound to heme oxygenase into biliverdin in the presence of NADPH-P450 reductase and prepared apo-heme oxygenase us- ing gel filtration and ultrafiltration, yielding a preparation that had activity toward free heme.
Incubation of apo-heme oxygenase, NADPH-P450 reductase, and NADPH with P450 3A4 yielded little if any change in the P450 spectrum (Fig. 6). These results argue that the heme in P450 3A4 is not readily accessible to heme oxygenase.
The difficulty in discerning the nature of heme bound in P420 preparations has been discussed above. We incubated P450 3A4 with apo-heme oxygenase, NADPH-P450 reductase, and NADPH for 30 min at 37°C (similar to Kutty et al. (63), but in the absence of detergent) and analyzed extracts for the product biliverdin. The method (54) was very sensitive, and a few picomoles of biliverdin could be detected. A peak with the expected t R and the expected ratio of A 670 :A 405 was detected (Fig. 7). This peak was not seen in control experiments devoid of NADPH or apo-heme oxygenase and accounts for 3% of the heme in the P450 preparation (mean of duplicate experiments). This loss would, of course, not have been detectable in Fig. 6.
We conclude that, as expected, P450 3A4 is a poor substrate for heme oxygenase, and this is probably the case for all intact P450s. We also conclude that the small amount of biliverdin formed in the presence of a high concentration of heme oxygenase is further evidence that only trace amounts of loosely bound heme are associated with these P450 3A4 preparations. The biliverdin that is formed (ϳ3%) is probably derived from the small amount of P420 present (i.e. Fig. 6) and is consistent with the low amount of heme transferred to apo-b 5 (Figs. 2 and  3).
Conclusions-The literature contains reports of at least four P450s for which catalysis is stimulated by apo-b 5 as well as b 5 (22, 28 -31). However, some P450s (e.g. 2E1) are clearly stimulated by b 5 but not by apo-b 5 (28,66), consistent with a role for electron transfer. Recently Guryev et al. (39) have raised the possibility that the stimulations by apo-b 5 may be the result of transfer of heme from P450 preparations to apo-b 5 to form b 5 , which then functions in an electron transfer role. The facile equilibration of P450 heme with the medium is highly unlikely, but the possibility exists that adventitiously bound heme could be transferred to fill the role of the b 5 prosthetic group. The varying effects of apo-b 5 versus b 5 on different P450s (e.g. 2E1 versus 3A4, 4A7, 17A, and 2C9) might be attributable to differences in the content of loosely bound heme, if this hypothesis has merit. To investigate these possibilities, we re-examined several features of the stimulation of some P450 3A4 catalytic activities by apo-b 5 .
We confirm our previous results (22,28) on the stimulation of P450 activities by b 5 and apo-b 5 , using four different P450 3A4 preparations isolated with a different procedure. Apo-b 5 and b 5 are similar in their stimulation (Fig. 1), and all of the P450 3A4 preparations had nearly equivalent amounts of detectable P450 and total heme as judged by the application of the generally used extinction coefficients. Spectral studies indicate that some of these P450 3A4 preparations do contain trace heme that can be transferred to apo-b 5 (Fig. 3), but only to the extent of ϳ5%, which will not account for the observed concentration-dependent stimulation of P450 3A4 (Fig. 1). Highly purified apomyoglobin had no discernable effect on the stimulation of either testosterone 6␤-hydroxylation or nifedipine oxidation by b 5 or apo-b 5 , even with a P450 preparation with a considerable amount of P420. A reconstituted heme oxygenase system had little effect on the P450 spectrum after incubation with P450 3A4 (Fig. 6), and only ϳ3% of the heme was converted to biliverdin (Fig. 7), ruling out the presence of appreciable adventitiously bound heme and the use of intact P450 3A4 as a substrate by heme oxygenase. A further important argument against the necessity of heme transfer to apo-b 5 for stimulation is an experiment done previously (22) on the lack of substitution of apo-b 5 in NADH-b 5 reductase/P450 3A4dependent testosterone 6␤-hydroxylation, which was repeated here with the same result.
We do not fully understand the details of how apo-b 5 and, by extension, b 5 stimulate all P450 activities. This is not unusual in the context of modern biochemistry, where many protein interactions result in remarkable changes in biological activity but are not well understood. The complexity of the interactions of P450 3A4 with various ligands has been reviewed recently (67). Because P450 2E1 activities are stimulated by b 5 but not by apo-b 5 (66), a case can clearly be made there for electron transfer. With P450 3A4, both apo-b 5 and b 5 stimulated the reduction of ferric P450 in the presence of the substrate testosterone (22,27); this effect was not observed with P450 2C9 (28) and was observed to only a limited extent with 4A7 (30,31). The K m (testosterone) and oxidation-reduction potential for P450 3A4 were not changed (22). Other studies with P450s 3A4, 17A, and 4A7 (22,28,29,31) have not defined mechanisms of apo-b 5 stimulation to date, and these P450s may differ in the exact roles that b 5 and apo-b 5 play.