Adrenal Mitochondrial Cytochrome P-450,,, CHOLESTEROL AND ADRENODOXIN INTERACTIONS AT EQUILIBRIUM AND DURING TURNOVER*

Purified cytochrome P-450,,, from bovine adrenal cortex mitochondria after treatment with BrCN yielded a core peptide which retains heme. The amino acid composition of this peptide was similar to that of the analogous peptide isolated from cytochrome P-450,of Pseudomonas putida. Adrenodoxin and cholesterol association with P450,, was analyzed in nonionic Tween 20 micelles where cholesterol appears to be fully in equilibrium with the cytochrome. Adrenodoxin binding to cholesterol-free P-450.,, was observed by a type I spectral shift in the cytochrome (Kd = 4 X lo” M). Binding to the cholesterol-P-450,, complex was over 10 times stronger (Kd = 3 X M). Binding of adrenodoxin to both free enzyme and cholesterol complex was unaffected by Tween 20, indicating a clear separation of the adrenodoxin binding site from the hydrophobic membrane binding domain. Adrenodoxin binding is driven by a large increase in entropy (AS = 30 e.u., A H = 0 kcal), while cholesterol activation of this binding is a consequence of a further increase in A S (from 30 to 53 e.u.) which more than offsets an increase in A H (5.7 kcal). The spin states of the complexes of cytochrome P-450.,, with both cholesterol and adrenodoxin and of the ternary complex were insensitive to changes of temperature (5-35”C), but the high spin content of the cholesterol complex in 4-(2-hydroxyethyl)-l-piperazineethanesulfonic acid buffer was raised by increased ionic strength (200 m~ KC1,83%) and by the binding of adrenodoxin (92%). The Kd for adrenodoxin-P-450,,, complex and the apparent K,,, for adrenodoxin in cholesterol side chain cleavage reaction both increased exponentially with ionic strength. In contrast, the analogous constants for cholesterol were insensitive to ionic strength. The initial velocity patterns with varied adrenodoxin and cholesterol intersected below the horizontal axis, the apparent K,,, for each reactant increasing with increasing concentrations of the second reactant. The true K,,, for each reactant was manyfold greater than the respective Kd for complex formation with P-450,,,.

The results, overall, are consistent with a random nonrapid equilibrium mechanism with positive cooperativity (synergism) for the binding of adrenodoxin and cholesterol to P-450,,, during turnover.
The mitochondrial cytochromes P-450 are in many respects similar to the Pseudomonas putida P-450,,, which functions in camphor hydroxylation (6). We have recently shown immunological cross-reactivity between cytochromes P-450,,,. and P-450,,, (7). BrCN cleavage of bacterial as well as microsomal cytochromes P-450 yields a heme-binding peptide containing 40-50 amino acids which probably retains certain structural features of the heme and substrate binding sites of the cytochrome (8)(9)(10). In this paper, we describe the isolation of a similar peptide from cytochrome P-450,,,, the first to be isolated from a mitochondrial cytochrome P-450.
Monooxygenation by mitochondrial P-450 hemeproteins requires both a flavoprotein (adrenodoxin reductase) and a ferredoxin-type iron-sulfur oxidation-reduction protein (adrenodoxin) which function in the transport of electrons from NADPH to P-450 (6,11). Also, in this respect, the mitochondrial monooxygenases resemble the camphor hydroxylation system of P. putida (6,11,12). However, in contrast to the bacterial enzymes which are soluble, the mitochondrial enzymes are membrane-bound adrenodoxin reductase and ADX appear to be peripheral membrane proteins and P-450,,, is an integral membrane protein (11,13,14). Recent evidence suggests that ADX transports electrons from adrenodoxin reductase to P-450 by shuttling between these two enzymes (14)(15)(16)(17). According to Kid0 and Kimura, ADX binding to cytochrome P-45OS,, requires both cholesterol and a detergent or a phospholipid (18,19), while in apparent contrast, Lambeth et al. (20) report that ADX can bind to cholesterol-free cytochrome P-450,, in phospholipid vesicles.
In this paper, we examine the interactions of ADX and cholesterol with cytochrome P-450,,, in the presence and absence of the detergent Tween 20.' This detergent provides optimal activity for cholesterol side chain cleavage (17, 21) and its nonionic nature permits the analysis of ion effects on the enzymes independent of interactions of ions and proteins with charged head groups of phospholipid vesicles. Our results indicate that Tween 20 micelles do not significantly affect the affinity of adrenodoxin to P-450,,, and provide an environment where cholesterol appears to be fully in equilibrium with the cytochrome. The binding of ADX to P-450,,, is shown to be much more sensitive to changes in metal ion concentrations than is cholesterol binding, and this difference is observed to be maintained in the catalytic constants. Steady state kinetics is used to examine the sequence of cholesterol and ADX binding to cytochrome P-450,,, during side chain cleavage, and the results are compared with the monooxygenase scheme for camphor hydroxylation by P-450,.,.

METHODS"
A n l l l n e Sepharore--An~l1ne-Sepharare w a s prepared by a modlflcatlon of the method of CUatreCdPas ( 2 2 ) . Washed Sepharose 48 (100 m l ) *ai d l l u t e d 1:l w I t h water. BrCN ( 2 . 5 ml, 1 glnl a c e t o n l t r l l e ) was added with r a p l d i t l r r r n g and t h e s l u r r y was m i n t a l n e d a t pH 11 : 0.5 by drapwlre l d d l t l o n of 5 N NdOH and a t ZO'C by dddltlon O f ice. A f t e r 20 mln.
Sepharole was rapidly washed with 1.5 I l t e r l 0.2 M NIHC03 (pH 9.5) and then added t o an a n l l l n e -b u f f e r mixture (IO m l f r e s h l y r e d l l t l l l e d a n l l l n e 1s *,red v l t h 125 m l water.
followed by A f t e r I t l r r l n g O v e r n l g h t a t

4'C.
the anlllne-Sepharore #ai washed I~L L e s s l~e l y w i t h 500 m1 O f 0.1 N sodium acetate (pH 4.0).
was suspended I n 50 nM K phosphate W f f e r , pH 7.9. and 0.02% I o d l m azlde.
2 N uvea. and 0.1 N NaHC03, each c a n t a l n l n g 0.5 N NaCI. F r n a l l y , the anlllne-Sepharare The r e s u l t a n t s u p e r n a t a n t was centrifuged a t 16.0W X 9 for 10 mln. The p e l l e t was suspended I n 1W n*l K phosphate.    the loury assay (27) and the reduced-C0 difference spectrum 91 ( 2 8 ) .
9lYLWOl. * I S diluted * i t h an equal volume Of H20 and placed ~n a 3. ml cuvette. *Ox, AR and WOPH *ere added I n that Order to both the reference and sample r w e t t e (final Spectra-All optical spectra were carried out on a DW-2 spectrophotometer (Aminco) operating either in the dual wavelength or split beam modes. ADX binding curves were determined from the AA (390-420 nm) responses upon addition of concentrated solutions of ADX in 10 mM Tris buffer (500 p~) to P-450,,. Cholesterol binding to P-450,, was determined by measuring the changes at 420 and 390 nm in direct spectra of solutions made up in the appropriate concentration of cholesterol. For ADX binding, it was necessary to calculate the concentrations of bound and free proteins. AAmaX and hence AE for complex formation was determined by measuring the response induced by saturating concentrations of ADX. The small direct contribution of the ADX absorption was measured independently and subtracted from the type I response. The concentration of P-450,, was quantitated by the reduced CO difference spectrum (28). ADX and adrenodoxin reductase concentrations were determined using E = 10 and 11 mM" cm" at 414 and 450 nm, respectively (32,33).
Analysis of Kinetic Data-The steady state kinetic nomenclature used in this paper is that of Cleland (34). The initial velocity pattern presented in Fig. 11 was analyzed by SEQUEN and PINGPONG programs in Fortran (35). These programs fit the data to the intersecting and parallel patterns described by sequential and ping-pong mechanisms, respectively, using a nonlinear least squares method. The lines in Fig. 10 are based on the SEQUEN fit. The lines in Fig.  8 were fit to the data using the Fortran program HYPER (35). The concentration of ADXf,, in Figs. 8 and 10 was calculated as previously described (17). The Kd values for the adrenodoxin reductase-ADX complex (2.5-40 X lo-@ M ) were taken from Fig. 2 of Lambetb et al. (16), and Kd values for the ADX P-450,,, complex (1.5-58 X
Other Procedures-Lipid content was measured by analysis of organic phosphate after Folch extraction (36,37) and cholesterol content by gas-liquid chromatography using [3H]cholesterol to quantitate recoveries after extraction by methanol and hot ethyl acetate. Unless otherwise specified, all protein determinations were done using the biuret procedure (38). Cholesterol side chain cleavage was assayed in 0.3% Tween 20 at 37°C. The preparation of Tween 20-cholesterol solution and the assays of [3H]cholesterol conversion to [3H]pregnenolone were carried out as previously described (39).

RESULTS
In the present studies, cytochrome P-450,, has been purified to high specific activity (13 nmol of P-450/mg of protein, A280,A393 = 1.2, > 10 preparations) using.a modification of the   (Table I)  Characterization of the "Hemepeptide"-A core peptide containing 20% of the heme of P-450,,, was readily generated from cytochrome P-450,,, by degradation with BrCN and resolved from other peptides (29). The amino acid composition of this peptide is shown in Table I. The Soret maximum (ernM = 62 in 5% acetic acid) at 360 nm (Fig. 1B) is clearly distinct from that of free heme (Fig. 1 0 . The 390 nm shoulder of the hemepeptide spectrum probably, in part, reflects the absorbance of dissociated heme, which under these conditions, has a Soret maximum at 393 nm. The reduced-CO complex of the hemepeptide has a Soret maximum at 406 nm while that of hemin is at 410 nm. The extinction coefficient of the complex is increased by decreased polarity in the solvent, but even in 5% acetic acid/acetone (1: l), the extinction coefficient of the heme-CO complex (ernM = 45) is substantially less than in the peptide (ernM = 68).
Binding of Adrenodoxin and Cholesterol to P-45OS,,-Detergent-free P-450,,,, as isolated, was mostly in a high spin form, probably due to complex formation with the endogenous cholesterol. In the presence of Tween 20, the P-450,,, preparation became fully low spin. Restoration of the high spin state by addition of cholesterol to Tween 20, indicated that the Tween 20-induced shift to low spin derived from dissociation of endogenously bound cholesterol into the detergent. ADX bound to detergent-free P-450,,, with high affinity and with a type I difference change (Fig. 2). ADX also bound to  (Fig. 2).
The affinity of cholesterol-free P-450 for ADX was not significantly affected by 0.3% Tween 20 (& = 0.4 p~ in both cases). The affinity for ADX was greatly enhanced by the presence of cholesterol (e.g. 6 times by 90 p~ cholesterol, Fig.   3). The Kd for ADX binding at saturation of cholesterol, calculated from the plot in Fig. 3 (inset)  over binding in absence of cholesterol. Similarly, 0.5 p~ ADX enhanced the affinity of P-450,,, for cholesterol by 6 to 7 times (data not shown). Affinity of ADX to P-450,,, in Tween 20 with 200 p~ cholesterol was 2 times stronger than to P-450,,, formed in a complex with endogenous cholesterol in absence of Tween 20 (the percentage of P-450,, formed in a complex with cholesterol was similar under the two conditions), suggesting a small stabilizing effect of Tween 20 on the ternary complex of ADX-cholesterol-P-450scc (Fig. 2). The spin states of the various P-450,,, complexes have been calculated from changes in the Soret absorption bands relative to the spectrum of a fully low spin cholesterol-free cytochrome ( Table 11). The proportions of high spin state induced at saturation with only ADX or only cholesterol were 20% and 6776, respectively; whereas at saturation of both, it was 92%, suggesting an additive effect of the two ligands on this parameter ( Table 11).
Effect of Temperature on Complex Formation-The binding of ADX to cytochrome P-450,,, in the absence of cholesterol was insensitive to temperature changes from 22-35°C. However, below 22"C, complex formation increased with decrease of temperature. The differences shown in Fig. 4 between complex formation at 6°C and 22°C were measured by changing the temperature on a single sample and were therefore directly observable. Spectral changes produced by wanning and cooling the sample in the range 6-35°C were fully reversible.

TABLE 11
Spin states of cytochrome P-450,, complexes   Fig. 4. The inset shows Van't Hoff plots of these data and for a parallel experiment with 0.5 /.LM ADX.
-7.4 kcal mol" and A S = -5.9 e.u. In the presence of ADX (0.5 p~) , Kd was less temperature-sensitive with A H = -1.7 kcal mol" and A S = 17 e.u. (Fig. 5). The increase in temperature induced only a slight decrease in the proportion of high spin state in the cholesterol-P-450,,, complex formed by saturating levels of cholesterol (73% at 5°C to 67% at 30°C). The spin state of the ternary cholesterol-ADX-P-450,,, complex was also relatively insensitive to temperature (Table 11).
Effect of Ions on Complex Formation-Increasing concentrations of NaCl increased the Kd for ADX both in the absence and presence of 45 p~ cholesterol (Fig. 6). Effects of NaCl on the optical change at saturation of ADX, both with and without cholesterol, were negligible (Fig. 6). In contrast, although NaCl progressively increased the proportion of high spin cytochrome at saturation with cholesterol (60-85%), it did not significantly change the Kd for cholesterol (Fig. 7). NaCl also increased the proportion of high spin cytochrome P-450 when added to the cholesterol complex in the absence of Tween 20. CaClz (1-5 mM) had no effect on either the affinity of the cytochrome for cholesterol or the spectrum of the complex.
Effect of Ionic Strength on Adrenodoxin and Cholesterol Dependence of Side Chain Cleavage Activity-When cholesterol side chain cleavage was reconstituted from purified adrenodoxin reductase, ADX, and cytochrome P-450,,, in the presence of 0.3% Tween 20, activity in 5 mM Hepes buffer was greatly enhanced by the addition of univalent ions. At 5 p~ ADX, we observed a similar bell-shaped dependence on the concentration of added NaCl to that previously reported by Takikawa et al. (21) with peak activity between 80 and 150 mM KC1. However, the fall-off in activity at high salt was caused by an increase in the K,,, for ADX which paralleled the increase in the Kd for ADX (Fig. 8, inset). Indeed, at 200 mM NaCl and 200 p~ cholesterol, the V,,, in terms of saturation with ADX was at the highest value of 30 nmol of pregnenolone/nmol P-450/min. The same dependence of activity on ionic strength was obtained when Hepes/NaCl was replaced by increasing concentrations of K phosphate buffer. In this buffer, salt had only a slight effect on the K,,, for cholesterol up to 42 mM K phosphate (ionic strength equivalent to 100 mM NaC1) (Fig. 9). However, at 82 mM K phosphate (equivalent to 200 mM NaCl), there was a decrease in the apparent K,,, for cholesterol. Thus, a 2-fold increase in ionic strength (100-200) produces a 15-fold increase in K , for ADX and, at most, a 2-fold decrease in K,,, for cholesterol.
Side Chain Cleavage Activity: Initial Velocity Patterns-In order to determine the kinetic mechanism of cholesterol and ADX additions during each cycle of cholesterol side chain cleavage, initial velocity pattern studies were carried out. Plots of v-l versus [ADXf,,,]" exhibited a decreasing slope with increasing cholesterol, intersecting at a single point (Fig.  10). The same data replotted as u-l uersus [cholesterol]" at various [ADXr,,,] also provides a set of intersecting lines (Fig.  10, inset). A notable feature of these plots is that the apparent K,,, increases in both cases as the concentration of the fmed component is increased.  mitochondria (6, 14, 43). P-450,,, is an integral membrane protein, whereas adrenodoxin reductase and ADX appear to be peripheral membrane proteins (14).
Cholesterol, the substrate, is very insoluble in water (CMC z 30 nm, Ref. 44) and more readily gains access to the cytochrome within detergent micelles or phospholipid bilayers ( 14).
The extensive studies on P. putida ferredoxin-cytochrome P-45OC,,-catalyzed camphor monooxygenation provide the basis for our current understanding of the catalytic cycle for this class of reactions (12). Many similarities are apparent between cytochromes P-450,,, and P-450,,,, including immunological cross-reactivity (7) and inter-relationship of substrate and ferredoxin binding to the cytochromes. Cyto- chromes P-450,,, and P-450,., have nearly identical sues while the amino acid composition of P-450,,, differs most noticeably from that of P-450,,, in the large increases in aromatic amino acids and lysine as compared to a substantial decrease in alanine ( Table I).
Hemepeptide-P-450,,, also shares with P-450,,, and P-4 5 0~~ a structural feature which permits release with BrCN of a very hydrophobic, heme-binding core peptide which in each case comprises about 10% of the total proteins (7,8, 29). The peculiar spacing of methionine residues in the surrounding sequences possibly makes this domain fortuitously accessible via BrCN cleavage. The purified hemepeptide of P-450,,, was found to contain 20% of the heme of the native hemeprotein. However, recoveries of up to 80% have been observed in hemepeptides released from P-450,, photocovalently labeled at a site close to the heme binding site by an azido derivative of the inhibitor, aminoglutethimide. For other P-450 cytochromes, a substrate-based photoaffinity probe also greatly increases this recovery, thus further relating the peptide to the functional center of the cytochrome (8,29).
Adrenodoxin and Cholesterol Binding to P-450,,,-There has been a conflict in previous reports (18)(19)(20) concerning the binding of ADX to substrate depleted P-45OS,,. This study clearly shows that ADX binds to P-450,,, metabolically depleted of endogenous cholesterol in the presence or absence of detergent Tween 20 (Fig. 2). The kinetics of cholesterol association with P-450,,, in Tween 20 (Fig. 5) indicate that in Tween 20 micelles cholesterol is fully in equilibrium with the cytochrome. The positive cooperativity we observe between the interactions of cholesterol and ADX with P-450,,, in Tween 20 is quantitatively similar to that reported by Lambeth et al. for P-450,,, in lecithin-cholesterol vesicles (20). In the present studies, we also show that Tween slightly activates rather than inhibits the binding of ADX to the P-450,,cholesterol complex and that it has no effect on binding to the substrate-free cytochrome. This insensitivity to Tween 20 implies that the hydrophobic domain which anchors the cytochrome to membranes is clearly separated from the ADX binding site. The small perturbation of the Soret band of the substrate-free cytochrome by Tween 20 does, however, suggest a communication between the membrane environment and the active center (Fig. 2, inset) ADX binding to P-450.,, is driven by a large increase in entropy. Consideration of the full cycle (Fig. 11) indicates that cholesterol activation of this binding is a consequence of a further increase in A S (from 30 to 53 e.u.) which more than offsets an increase in AH (5.7 kcal). The high negative charge on the surface of ADX and the observed competition between ions and ADX for the cytochrome suggest that the entropy increase may derive from a release of water and ions during the interaction of complementary charged domains on the two proteins. The spin state change from low to high spin is associated with an increase in entropy (14-30 e.u.) and a decrease in enthalpy (2.5-10 kcal) (46, 53).4 Subtraction of the contribution from the spin state change (60%) indicates that cholesterol binding is associated with large opposing changes in enthalpy and entropy. The transference of cholesterol from detergent to binding site therefore seems to be associated with restricted movement, possibly of both cholesterol and protein, which is partially relieved by prior binding of ADX.
The spin state of cytochrome P-450 is an indicator of the configuration of ligands around the heme (12). Increases in the proportion of the high spin state are also frequently associated with facilitated reduction of the cytochrome (45).
Both ADX and cholesterol shift the spin state equilibrium of P-450,,, to high spin form (17-20, 51). This provides a difference from P-450,,, where camphor stabilizes the high spin form but putidaredoxin causes a shift to the low spin form (12, 52).
The high spin state induced by a fxed concentration of cholesterol declines with rising temperature. In Tween 20, this change derives from an increased dissociation of cholesterol rather than from an intrinsic change in the spin state of the substrate complex, as has been reported for P-450,,, (46). The similar temperature-dependent change to low spin observed for the P-450,,, complex with endogenous cholesterol in absence of Tween 20 (51) presumably also derives from movement of cholesterol out of the substrate site. However, cholesterol remains bound to the protein under these conditions, suggesting that cholesterol moves to a secondary site on the protein associated with a low spin complex. ADX and ions increase the proportion of high spin state in the cholesterol-P-450,, complex by comparable amounts (Table IT)  reduction (16). Activation of ADX reduction also results in decreased levels of oxidized ADX which we have previously shown to inhibit monooxygenase activity (17, 47). Here, we show that the inactivation at high concentrations of univalent ions is only apparent and is due entirely to an over 50-fold increase in the K,,, for ADX which parallels the increase in the Kd for ADX (Fig. 8). Indeed, V,,, at 200 pM cholesterol continues to increase up to the highest salt concentration examined. In contrast, the apparent K , and the K d for cholesterol are both relatively insensitive to changes in salt concentration (Figs. 7 and 9). The shift in the apparent K , for cholesterol at the highest salt concentration examined is in the same direction as that caused by decreased [ADX] (Fig.  10). Thus, this shift is probably due to an increase in the K , for ADX (Fig. 8 ) (which would render the [ADX] used in this experiment less than saturating), rather than a true decrease in the K, for cholesterol.
Complex Formation during Enzyme Turnover-The catalytic cycle for camphor monooxygenation at P-450,,, has been presented as an ordered sequence of substrate binding followed by putidaredoxin binding and first electron transfer (12, 48). However, our results, as well as those of Lambeth et al. (20), clearly indicate that ADX can bind to P-450,,, that is free of substrate cholesterol. Since on the basis of these results there appears to be no obligatory order for the binding of ADX and cholesterol to P-450,,,, the sequence of additions of these two reactants would have to be considered as random rather than ordered (34, 49). Superficially, the intersecting initial velocity pattern and the linearity of the plots (Fig. 10) are consistent with a futed order of binding of cholesterol and ADX (Ref. 49, p. 564). However, apparent order can be reconciled with random binding if, at these relative concentrations of reactants under these experimental conditions, the addition of one of the reactants occurs first with much higher frequency. These reaction conditions are far removed from those found in the adrenal cortex mitochondria where high local concentrations of both ADX and P-450,,, on the surface of the inner membrane and a low steady state level of cholesterol may result in frequent binding of ADX to cholesterolfree P-450,,.
Given the substantial synergism in the binding of cholesterol and ADX to P-450,,, the observed initial velocity patterns in which cholesterol increases the K, for ADX and vice uersa (point of intersection below the abscissa, Fig. 10) are atypical for a random rapid equilibrium mechanism (Ref. 49, p. 274). Indeed, there is strong evidence that, unlike camphor hydroxylation (48), rapid equilibration of these reactants does not occur during cholesterol side chain cleavage. The true K , values for cholesterol and ADX are manyfold larger than the respective K d values for either binary or ternary complex formation. This is normally observed when the rate-limiting step is faster than the dissociation of the substrates by similar factors. Simulation of kinetic patterns corresponding to such nonrapid equilibrium conditions shows that synergism in the binding of reactants can be consistent with an intersection point below the abscissa (50).
The apparent slow dissociation of reduced ADX from P-450,,, raises the important question of whether oxidized ADX dissociates from the cytochrome during the catalytic cycle. Possibly, dissociation may be accelerated by either reduction of P-450,,, or oxidation of cholesterol. In view of the evidence that adrenodoxin reductase, ADX, and P-450,,, do not readily form a ternary complex (14-17), the dissociation of oxidized ADX from P-450,,, must be considered a potential rate-limiting factor in the cholesterol side chain cleavage mechanism.
Acknowledgments-We are grateful to Dr. W. W. Cleland for reviewing the initial velocity pattern analysis and for providing copies