The One-electron Autoxidation of Human Cytochrome P450 3A4*

Monomeric cytochrome P450 3A4 (CYP3A4), the most prevalent cytochrome P450 in human liver, can simultaneously bind one, two, or three molecules of substrates and effectors. The difference in the functional properties of such binding intermediates gives rise to homotropic and heterotropic cooperative kinetics of this enzyme. To understand the overall kinetic processes operating in CYP3A4, we documented the kinetics of autoxidation of the oxy-ferrous intermediate of CYP3A4 as a function of testosterone concentration. The rate of autoxidation in the presence of testosterone was significantly lower than that observed with no substrate present. Stability of the oxy-ferrous complex in CYP3A4 and the amplitude of the geminate CO rebinding increased significantly as a result of binding of just one testosterone molecule. In contrast, the slow phase in the kinetics of cyanide binding to the ferric CYP3A4 correlated with a shift of the heme iron spin state, which is only caused by the association of a second molecule of testosterone. Our results show that the first substrate binding event prevents the escape of diatomic ligands from the distal heme binding pocket, stabilizes the oxy-ferrous complex, and thus serves as an important modulator of the uncoupling channel in the cytochromes P450.

Monomeric cytochrome P450 3A4 (CYP3A4), the most prevalent cytochrome P450 in human liver, can simultaneously bind one, two, or three molecules of substrates and effectors. The difference in the functional properties of such binding intermediates gives rise to homotropic and heterotropic cooperative kinetics of this enzyme. To understand the overall kinetic processes operating in CYP3A4, we documented the kinetics of autoxidation of the oxy-ferrous intermediate of CYP3A4 as a function of testosterone concentration. The rate of autoxidation in the presence of testosterone was significantly lower than that observed with no substrate present. Stability of the oxy-ferrous complex in CYP3A4 and the amplitude of the geminate CO rebinding increased significantly as a result of binding of just one testosterone molecule. In contrast, the slow phase in the kinetics of cyanide binding to the ferric CYP3A4 correlated with a shift of the heme iron spin state, which is only caused by the association of a second molecule of testosterone. Our results show that the first substrate binding event prevents the escape of diatomic ligands from the distal heme binding pocket, stabilizes the oxy-ferrous complex, and thus serves as an important modulator of the uncoupling channel in the cytochromes P450.
Cytochrome P450 3A4 (CYP3A4) 2 is the most prevalent cytochrome P450 in human liver and is responsible for the breakdown of numerous xenobiotics, including more than 50% of the currently marketed pharmaceuticals (1,2). Most cytochromes P450 operate via a catalytic cycle wherein atmospheric dioxygen is ligated to the ferrous heme of the P450 protein and is reduced through electron input from pyridine nucleotide oxidation as provided by an associated electron transfer chain. In the case of CYP3A4 this is an 80-kDa reductase molecule containing both FAD and FMN prosthetic groups. A central intermediate in this reaction cycle is the ferrous-dioxygen adduct generated through ferric-ferrous reduction of the heme center and dioxygen binding. This intermediate, the "oxy complex" can either decay via release of superoxide and regeneration of the ferric heme (3) or be reduced with a second redox equiva-lent resulting in the cleavage of the oxygen-oxygen bond and generation of a higher valence metal-oxo complex that can initiate radical chemistry and subsequent substrate oxygenation. The former autoxidation pathway represents the first branching point between productive and unproductive pathways in P450 catalysis wherein reducing equivalents leak into the production of unwanted reduced oxygen species (4 -7). Subsequent uncoupling channels include the release of the two electron-reduced dioxygen as peroxide and a further reduction of the higher valence metal-oxo complex to produce a second water molecule.
The oxy-ferrous intermediate in various cytochromes P450 was documented many decades ago (3, 8 -11). For the soluble bacterial systems, the autoxidation process was shown to be first order (3,8,9) but for the membrane-associated isozymes was often described by multiphasic processes (11) presumably due to the heterogeneous aggregated state of the protein preparations when placed in aqueous solution. The Nanodisc system (12,13), which consists of ϳ10-nm-diameter discoidal phospholipid bilayers rendered soluble in aqueous solution via an encompassing membrane scaffold protein, provides an ideal system to provide a homogenous, soluble, and monomeric CYP3A4 for detailed kinetic investigations.
Recently we reported the influence of substrate binding on the stability of this oxy-ferrous complex in human CYP3A4 using the Nanodisc system to provide a homogeneous and soluble preparation of this integral membrane protein (14). We have shown that the autoxidation rate of the substrate-free CYP3A4 is extremely rapid, notably much faster than the steady-state NADPH consumption rates that are measured under the same turnover conditions (14,15). Saturation with various substrates was found to reduce the rate of autoxidation by almost 2 orders of magnitude, thus inhibiting the unproductive decomposition of the oxy-ferrous intermediate and production of reduced oxygen species. This work provided important information in terms of the overall efficiency of redox equivalent consumption or coupling ratio determined as the ratio of the steady-state rates of product formation and NADPH consumption. Analysis of the functional properties of the stoichiometric complex of CYP3A4 with the cytochrome P450 reductase in these steady-state experiments of testosterone hydroxylation revealed the distinct fractional contributions of CYP3A4 with different numbers of testosterone molecules bound to metabolism and all uncoupling pathways (15). In particular, the rates of NADPH consumption and product formation were shown to be very different for the CYP3A4 with one and two testosterone (TS) molecules bound.
The rapid rate of autoxidation of the ferrous-oxy complex suggested that this pathway was a predominate means for uncoupling in this human hepatic enzyme. To better understand the mechanism of the allosteric regulation of CYP3A4 catalysis and the significance of simultaneous binding of multiple substrate molecules, we studied the autoxidation of CYP3A4 as a function of substrate concentration. This allowed determination of the fundamental one-electron autoxidation rate for each of the functional states of the enzyme with one, two, and three molecules of substrate bound. These results demonstrate that although the state with a single testosterone bound is not able to effectively catalyze substrate hydroxylation it is responsible for a dramatic slowing of the one-electron autoxidation rate, thus improving the overall coupling of pyridine nucleotide reducing equivalents. This suggests that the first substrate binding event induces a critical conformational change in the protein that stabilizes the ferrous-oxy complex and could block a superoxide exit pathway from the heme center. To explore the possibility of a steric block to ligand egress, we also measured the amplitude of CO geminate rebinding from the carbonmonoxy-ferrous complex in CYP3A4, which is a measure of ligand escape after photolysis, as well as the kinetics of cyanide binding to ferric CYP3A4. These experiments provide additional information on the role of the first and second substrate molecules in the modulation of diatomic ligand interactions with the heme iron and provide a consistent picture for the role of substrate binding in the efficient channeling of reducing equivalents to monooxygenation chemistry versus unproductive autoxidation.

MATERIALS AND METHODS
Expression and Purification of CYP3A4-Cytochrome P450 3A4 was expressed from the NF-14 construct in the PCWoriϩ vector with a C-terminal pentahistidine tag generously provided by Dr. F. P. Guengerich as described previously (16). The presence of the histidine tag has been shown to not perturb the measured turnover parameters of CYP3A4 (17). Heterologous expression and purification from Escherichia coli was carried out using a modified procedure (18) as described in the supplemental data contained in Ref. 14.
CYP3A4 in Nanodiscs-The application of the Nanodisc system for solubilization of integral membrane proteins incorporated into nanoscale bilayers has been described in detail in several publications (18 -23). Assembly of human CYP3A4 in Nanodiscs was accomplished using the scaffold protein MSP1D1 with the poly(histidine) tag (24) removed as described previously (18). Briefly purified CYP3A4 from the E. coli expression system was solubilized by 0.1% Emulgen 913 and mixed with the disc reconstitution mixture containing MSP1D1, POPC, and sodium cholate. A molar mixing ratio of 0.1:1:65:130 (CYP3A4:MSP1D1:POPC:cholate) was chosen to favor formation of monomeric CYP3A4 incorporated into Nanodiscs with the proper stoichiometry of scaffold protein and lipid. Detergents were removed by treatment with Biobeads (Bio-Rad), which initiates self-assembly. Purification of the fraction of Nanodiscs with incorporated CYP3A4 was achieved using a nickel-nitrilotriacetic acid affinity column followed by size exclusion chromatography as described previ-ously (14). The result of this self-assembly reaction is a monomer of CYP3A4 contained in a discoidal POPC bilayer ϳ10 nm in diameter stabilized by the encircling amphipathic membrane scaffold protein belt. CYP3A4 Nanodiscs were prepared in substrate-free form and kept at 4°C.
Substrate Binding-Formation and decay of the oxy-ferrous complex, geminate rebinding of carbon monoxide, and binding of cyanide was studied with CYP3A4 in the presence of different testosterone concentrations as well as in the substrate-free form. Testosterone was added using stock solutions in methanol with the final concentration of methanol always less than 1%. Spectral titration data were analyzed using the Hill equation F ϭ A⅐S n /(S D n ϩ S n ) where A is the spectroscopic amplitude, S is the substrate concentration, and S D is the spectral dissociation constant. For autoxidation and geminate rebinding experiments solutions of CYP3A4 in Nanodiscs were deoxygenated under the flow of argon gas and reduced by addition of a small excess of anaerobically prepared dithionite solution with the known concentration determined using a molar absorption coefficient of ⑀ 315 ϭ 8.05 mM Ϫ1 cm Ϫ1 (25).
Stopped-flow Experiments-All kinetic experiments utilized an Applied Photophysics SX.18MV stopped-flow spectrophotometer using typical mixing volumes of 150 l in each syringe with a dead time of 1.5 ms at the temperature 6°C. For autoxidation studies, anaerobic solutions of reduced CYP3A4 Nanodiscs (with or without substrate) in one syringe were mixed in a 1:1 ratio with buffer saturated with pure oxygen gas containing the same concentration of substrate if present. The resulting high oxygen concentration (ϳ690 M after mixing) was chosen to accelerate formation of the oxy complex. Kinetics of cyanide binding was studied by mixing the solution of ferric CYP3A4 in Nanodiscs with a solution of KCN (final concentration after mixing, 20 mM) with the same concentrations of testosterone present in both syringes. A high concentration of KCN was chosen based on the results of equilibrium titration experiments to ensure the pseudo-first order binding. In each kinetic experiment from 400 to 800 spectra were collected on a logarithmic time scale for the slow reactions (total collection time, 50 s) or a linear time scale for faster processes (collection time, up to 2.3 s).
Data Analysis-The data on autoxidation kinetics were analyzed according to the following reaction scheme.

SCHEME 1
Here Fe 2ϩ and Fe 3ϩ denote the reduced (ferrous) and oxidized (ferric) state of CYP3A4 heme iron, respectively; k 1 is the apparent pseudo-first order rate constant of oxygen binding to ferrous cytochrome P450; and k 2 is the rate constant for autoxidation, i.e. formation of ferric CYP3A4 and release of superoxide. Spectra collected in each kinetic experiment were arranged into a matrix with each column vector representing the spectra and rows representing the absorption at each wavelength as a function of time. The number of independent spectrally distinguishable components and their time-dependent concentrations were determined using singular value decomposition (26). The data analysis for substrate-free and substrate-saturated CYP3A4 was accomplished as described previously (14). For intermediate concentrations of TS the method was modified due to the presence of two spectral components originating from the high spin and low spin fractions of the final oxidized CYP3A4. Because of this, two spectral processes were observed simultaneously, namely the transition of the oxy-ferrous CYP3A4 to the high spin CYP3A4 and the same oxy-ferrous CYP3A4 to the low spin CYP3A4, each with the characteristic difference spectrum and total amplitude corresponding to the fractions of high spin and low spin states under experimental conditions and at the given TS concentration. After fast formation within 10 -20 ms, the oxy-ferrous CYP3A4 in the presence of intermediate TS concentrations autoxidized through a biphasic process with the typical apparent first-order rates of ϳ20 and ϳ0.5 s Ϫ1 . In most cases the slower phase could be better represented as a biexponential decay. Because of the large difference between the rates of the fast and slow phases, these two processes could be analyzed separately with the goal to minimize the number of fitted parameters and improve the precision. Calculations of the total fraction of CYP3A4 autoxidized at each kinetic phase was done based on the absolute difference spectra obtained by subtraction of the spectra of pure low spin and high spin CYP3A4 from the corresponding spectra of oxy-ferrous CYP3A4 in the absence of substrate and in the presence of 200 M TS (14). Amplitudes of fast and slow phases of cyanide binding to CYP3A4 at different TS concentrations were analyzed in a similar way using molar absorption spectra of substrate-free, substrate-bound, and cyanide-bound CYP3A4 measured under the same conditions. For the analysis of the fast phase with the rate 200 s Ϫ1 , the first 16 spectra collected at 2-24 ms were used to calculate the fraction of CYP3A4 that binds cyanide within this time frame. The slow phase amplitude was obtained from the spectra collected between 25 ms and 6 s. The total concentration of CYP3A4 was measured from the end point of each run using the known spectrum of cyanide complex of CYP3A4. The fraction of the slow phase was determined as the ratio of the total amplitude of the slow phase and the total CYP3A4 concentration. Carbon Monoxide Photolysis and Rebinding-Flash photolysis of the CO-bound form of CYP3A4 in Nanodiscs was accomplished using the system described previously (27) with the indicated modifications. Photolysis was initiated by a 200-mJ pulse of 532 nm light from a tripled neodymium-doped yttrium aluminium garnet (Nd:YAG) laser (Spectra Physics model GCR 150). The probe beam was from a Cermax 500-watt xenon light source (ILC Technologies) that was heat-filtered. For these experiments a monochromator was placed in front of the sample to produce a monochromatic probe light source at the probe wavelength indicated in the text. A second monochromator between the sample and the photomultiplier tube reduced the noise due to scattered laser light. Signals from the Oriel model 77341 photomultiplier were amplified using a Stanford Research Systems SR445 fast preamplifier (300-MHz bandwidth) and digitized using a LeCroy model 9350 oscilloscope (1 gigasample/s). Signals were continuously averaged at a flash repetition rate of 0.5 Hz until an acceptable signal to noise was obtained, usually 60 -75 transients. Samples of 1-2 M ferrous CYP3A4 were prepared in 0.8 ml of deoxygenated buffer containing 200 M sodium dithionite. The ferrous-CO complex was generated by the addition of 0.1 ml of CO-saturated buffer. Testosterone was added from stock solutions made in methanol. Absorbance values were typically measured at 450 nm and normalized to the expected change in absorbance obtained from the CO difference spectra from an individually run ferrous-CO minus ferrous CYP3A4 spectrum. The normalization results in a plot of fraction dissociated versus time. Samples were maintained at 6°C for all experiments, and the observed transients were fit to a sum of two exponentials. The resultant geminate yields were obtained from the value of the fit at times Ͼ4 s. The dependence of geminate yield on substrate concentration was fit to the following quadratic equation (28).
and we assume the following.
Here Y is geminate yield, Y 0 is geminate yield of substrate-free CYP3A4, Y max is the geminate yield of substrate saturated enzyme, ES is concentration of enzyme substrate complex, E t is the total enzyme concentration, and S t is the total substrate concentration.

RESULTS
Testosterone Binding-Testosterone binding to CYP3A4 in Nanodiscs was monitored at a temperature of 279 K (6°C) using absorption spectroscopy (300 -700 nm). The resulting optical spectra of CYP3A4 in the presence of different testosterone concentrations are shown in Fig. 1 and indicate the gradual shift of the heme iron spin state from predominantly low spin in the absence of the substrate to ϳ90% high spin at the highest TS concentration, 270 M. Analysis of the fraction of the spin shift as a function of TS concentration using the Hill equation returned a spectral dissociation constant of S D ϭ 56 M and a Hill coefficient of n ϭ 1.9 consistent with earlier reported data (18). The high apparent cooperativity of the spin shift is manifested through the relatively minor spectral changes initiated by the first TS binding event while exhibiting an almost complete spin shift following binding of the second TS molecule as was discussed previously (15,18).
Autoxidation Rates-Our previous results (14) indicate that the rate of autoxidation of CYP3A4 strongly depends on the presence of substrates. For substrate-free CYP3A4 the autoxidation is very fast with an apparent first order rate of ϳ20 s Ϫ1 at 278 K. At a near saturating concentration of testosterone, the autoxidation rate is slowed to ϳ0.37 s Ϫ1 , i.e. ϳ50 times lower (14). To better understand the substrate-dependent stability of the oxy-ferrous intermediate in the CYP3A4 catalytic cycle we measured the autoxidation rate at different TS concentrations at low temperatures. Rapid mixing of an anaerobic solution of the reduced CYP3A4 Nanodiscs with oxygen-saturated buffer with identical concentrations of TS present in both syringes resulted in fast (within several milliseconds) formation of oxyferrous complex of CYP3A4 with a broad absorption band at 424 nm as described previously (14). The kinetics of subsequent decomposition of this complex was monitored using UV-visible absorption spectroscopy at 279 K.
Experimental results showing the autoxidation kinetics in the presence of different TS concentrations are shown in Fig. 2. Even in the presence of 7 M TS, the kinetics is clearly biphasic with an apparent first order rate of the slow phase of 1.5 s Ϫ1 and an amplitude of ϳ20%. A further increase of TS concentration resulted in gradual elimination of the fast phase, which even at lower TS concentrations (60 M) was resolved only as a minor component (Ͻ5%) and was not observed at all at higher concentrations of this substrate. These data are summarized and compared with the equilibrium spectral titration of the ferric CYP3A4 with testosterone under identical conditions at 279 K in Fig. 3.
CYP3A4 can bind up to three molecules of testosterone, and the first binding event produces only a minor spin shift (15). Thus, the total high spin fraction of CYP3A4 is due to the contributions from the enzyme with two and three molecules bound. This is reflected in the spectral dissociation constant of S D ϭ 56 M obtained by the Hill analysis. However, from the data in Fig. 3 it is clear that the rate of autoxidation becomes significantly slower even at low TS concentrations. Thus, the presence of the first TS molecule bound to the enzyme has a profound effect on the stability of the oxy-ferrous complex, dramatically decreasing the rate of its irreversible decomposition into the ferric CYP3A4 and superoxide. The effect of sub-strate binding on the autoxidation of heme enzymes has been observed previously (Ref. 14 and references therein) and can be attributed to the structural and/or dynamic restrictions on the escape of superoxide from the distal binding pocket of the protein caused by the presence of the substrate and/or an increased stability due to conformational change or alteration of the local dielectric around the heme. In some cases this effect is correlated with the spin shift and redox potential of the heme iron as described previously (30,31); this effect can also be rationalized through a destabilization of the water ligand to the ferric heme iron (32)(33)(34). However, in the case of CYP3A4 binding testosterone, such a significant effect could not be predicted from the substrate-dependent spin shift and reveals the presence of a distinct mechanism controlling the overall stability of the oxyferrous complex. The origin of this increased stability could be due to the aforementioned alteration of the local dielectric or a direct occlusion of the superoxide escape channel, thereby   Table 2. increasing the "geminate" rebinding of superoxide. Either of these effects could be manifested through a substrate-induced conformational change of the enzyme that does not alter the water molecule bound as an axial heme ligand.
Diatomic Ligand Binding-To explore other measures of active site accessibility and obtain more detailed information we measured the access and escape of other diatomic ligands to CYP3A4. In a first set of experiments to probe the structure of the ferrous enzyme, we studied the amplitude of CO geminate rebinding from the ferrous-carbonmonoxy adduct. Extension to the ferric state of CYP3A4 was realized by quantitating the on-rate for cyanide binding. In both cases, measurements were made as a function of TS concentration to probe the reactivity profiles of each state with stoichiometric substrates bound.
Cyanide Association-Results of the studies of equilibrium binding of cyanide to CYP3A4 at 6°C in the presence of different testosterone concentrations are shown in Fig. 4. With no substrate present, CYP3A4 in Nanodiscs bound cyanide with a dissociation constant of K d ϭ 0.4 mM, whereas in the presence of 200 M TS cyanide the overall affinity was slightly weaker with a dissociation constant of 1.4 mM. Thus, binding of testosterone does not dramatically hinder cyanide binding to CYP3A4, certainly much less than that observed with either camphor binding to CYP101 (35) or androstenedione ligation to aromatase (36). The kinetics of cyanide binding to the substrate-free CYP3A4 is well described by a single exponential process with an observed pseudo-first order rate of k obs ϭ 200 s Ϫ1 at a cyanide ion concentration of 20 mM. The second order rate of cyanide binding, k on , as well as the dissociation rate from the cyanide complex of the substrate-free CYP3A4, k off , can be calculated using k obs ϭ k on [CN Ϫ ] ϩ k off ; K d ϭ k off /k on . The binding of cyanide to the substrate-free CYP3A4 is well described by the single exponential process with a spectral amplitude of more than 90% and the apparent first order rate of ϳ200 s Ϫ1 , corresponding to a second order binding rate of 10 mM Ϫ1 s Ϫ1 . A minor slow component with amplitude between 5 and 8% of the major peak and a rate ϳ80-fold slower was also usually detected. Both pseudo-first order rates were proportional to the concentration of cyanide, as shown in separate control experiments, confirming the true second order binding mechanism. From this analysis k on ϭ 10 mM Ϫ1 s Ϫ1 and k off ϭ 4 s Ϫ1 for CYP3A4 in Nanodiscs with no substrate present in the active site. These data are shown in Table 1 together with the rates of cyanide binding to different heme proteins including several other cytochromes P450.
Addition of substrate resulted in the gradual decrease in amplitude of the fast phase and an increase in the contribution of the slow process to the overall cyanide binding kinetics. The large difference between the rates of the fast and slow phases made possible the separate analysis of the data obtained within the first 25 ms following stopped-flow mixing of cyanide with the ferric enzyme. The fast phase process was completed during this time period, and at times between 0.03 and 10 s only a slow phase was observed. Separate singular value decomposition analysis of the time-resolved spectra obtained within the indicated time intervals revealed sufficient spectral differences between the fast and slow phases. The difference spectra of the fast phase correspond exclusively to cyanide binding to the low spin fraction of CYP3A4, whereas the spectral changes during the slow phase represent the transition from the predominantly high spin CYP3A4 with addition of a small fraction of the low spin CYP3A4 to the cyanide complex. This distinction between the fast and slow processes could be also monitored at the isosbestic points for the corresponding difference spectra shown in Fig. 4C, i.e. 420 and 426 nm for the cyanide binding to high spin CYP3A4 and low spin CYP3A4, respectively. Results of the analysis of the kinetic traces extracted at these wavelengths were in complete agreement with the results of singular value decomposition analysis of the full spectral matrix. At 420 nm, where the signal is due exclusively to the cyanide binding to a low spin fraction of CYP3A4, the kinetics is described by a single fast process (apparent first order rate, ϳ200 s Ϫ1 ) and almost no absorption changes at time longer than 30 ms. In contrast, at 426 nm the fast process cannot be seen at all, and the kinetics is described by a biexponential decay with rate constants of 3 and 0.6 s Ϫ1 .
The kinetics of cyanide binding to the ferric CYP3A4 in the presence of different TS concentrations monitored at 446 nm is depicted in Fig. 5 and Table 2. The amplitude of the slow phase was determined from biexponential fits of single wavelength kinetic traces extracted at the isosbestic point of the difference spectra shown in Fig. 4C as well as from the fits of the time-de-  pendent spectra collected from 370 to 500 nm for the difference basis spectra. The latter fits were used to estimate independently the spectra of the fast and slow fractions of CYP3A4 and to confirm that the fast phase is represented by exclusively low spin CYP3A4. Total concentrations of CYP3A4 were determined from the final spectra of cyanide-saturated enzyme in the end of each run, and the fraction of the slow phase was obtained as the ratio of the slow phase amplitude to the total concentration. The results of kinetic experiments of cyanide binding to CYP3A4 at different TS concentrations are summarized in Fig. 6. Comparison with the fraction of the high spin fraction of CYP3A4 revealed the similar dependence of the slow phase kinetic amplitude on TS concentration with only minor differences at low TS concentrations. This means that the kinetics of cyanide binding is determined by the spin state of the heme iron, whereas the diffusional constraints induced by the binding of the first TS molecule to CYP3A4 has almost no effect on the rates of formation of cyanoferric heme complex. The rate of cyanide binding, however, was much lower for the high spin CYP3A4. The same was true for the dissociation rate of cyanide, which can be calculated from the equilibrium dissociation constant and the binding rate. This result can be attributed to the steric constraints caused by the presence of the TS molecules in the immediate vicinity of the heme iron. As indicated by only a 3-fold difference between the dissociation constants for cyanide binding to CYP3A4 in the absence of substrate and in the presence of 200 M testosterone, the TS molecules bound at the active center of ferric CYP3A4 are sufficiently mobile and do not perturb significantly the stability of the distal cyanide ligand. At the same time, for the high spin fraction of CYP3A4 where the substrate destabilizes the distal water ligand, the binding and dissociation kinetics of the stronger ligand cyanide was significantly slower. Geminate Rebinding of Carbon Monoxide-Importantly the processes of cyanide binding to the ferric enzyme and superoxide escape from a ferrous-oxy intermediate may not be directly comparable. It is possible that the ferric-ferrous transition of the heme iron could be linked to a major structural rearrangement in CYP3A4. To provide an additional probe of the access of small molecules to the ferrous state of CYP3A4 the geminate yields were measured following laser flash photolysis of the ferrous-CO state, and the results are shown in Fig. 7. Previously we have reported on the usefulness of CO geminate recombination measurements to monitor the changes of the active site conformation and the effect of substrate on ligand access (27,37). Here we extend the concept to estimate the dissociation constant of the spectrally silent TS molecule. In Fig. 7 can be clearly seen that the geminate yield increased as a function of TS concentrations and displayed saturating behavior. The increased yields indicate that the presence of substrate occludes the exit pathway either directly or through a structural rearrangement FIGURE 5. Kinetics of cyanide binding to CYP3A4 at different testosterone concentrations as monitored at the isosbestic point of the difference spectra shown in Fig. 4C (446 nm). Closed circles are the experimental data normalized to the fraction of the slow phase using the differential extinction of 47 mM Ϫ1 cm Ϫ1 , whereas the lines illustrate the corresponding biexponential fits (parameters in Table 2   of the active site. More interesting is the fact that when fit to a simple binding isotherm the apparent dissociation constant is 7.5 M, nearly an order of magnitude smaller than the spectral dissociation constant, S D ϭ 56 M, once again indicating that the binding of the first TS molecule affects the escape of diatomic ligands but does not contribute significantly to the observed spin state change.

DISCUSSION
Efficient metabolism by cytochrome P450 systems requires tight coupling between the provision of needed reducing equivalents via pyridine nucleotide oxidation and the utilization of atmospheric dioxygen together with product formation. Alternate non-productive pathways include autoxidation of the oneelectron oxygen-bound ferrous protein to produce superoxide, which is subsequently converted to hydrogen peroxide, as well as the direct release of hydrogen peroxide from a two electronreduced peroxo intermediate. Previous investigations have suggested that, in the case of CYP3A4, a major pathway for uncoupling is from the oxy-ferrous state (14).
To better understand the regulation of stability of oxygenated CYP3A4 by substrate binding, we studied the autoxidation kinetics of this human cytochrome P450 solubilized in monomeric form in Nanodiscs as a function of testosterone concentration. As a result, we discovered that autoxidation is dramatically slowed, even at low testosterone concentrations, below that required for complete saturation of the enzyme. This is the first experimentally observed indication of the important functional changes of CYP3A4 caused by binding of one testosterone molecule. In addition we measured the amplitude of CO geminate rebinding following flash dissociation from the ferrous heme as well as the association kinetics of cyanide binding to ferric CYP3A4 to compare the changes caused by TS binding with the results obtained in the autoxidation studies. Both methods are commonly used as sensitive probes of the structural and dynamic effects caused by mutations and by interactions with substrates and cofactors in various heme enzymes (35,. As we described recently (15,60), the dependence of overall functional properties of CYP3A4 on testosterone concentra-tion can be represented as a sum of contributions of binding intermediates, i.e. the populations of binary, ternary, and quaternary complexes of CYP3A4 with testosterone, [CYP3A4-TS], [CYP3A4-(TS) 2 ], and [CYP3A4-(TS) 3 ]. We have found that binding of the first TS molecule leads to a relatively small spin shift and a 4-fold acceleration of NADPH consumption but no product formation, the latter result explaining the relatively high overall cooperativity of substrate turnover in this system. The binding of the second TS molecule is crucial for the formation of fully active enzyme-substrate complex, whereas the third binding event significantly improves efficiency of NADPH consumption with no further increase of the rate of substrate turnover. Clearly multiple TS molecules play substrate and effector roles at the same time, modulating the functional properties of ternary and quaternary enzyme-substrate complexes. It is important to document which reaction steps in the catalytic cycle are affected by sequential substrate binding events and what chemical mechanism is responsible for these functional differences between the binding intermediates.
The autoxidation of CYP3A4 is very fast compared with the overall steady-state product-forming kinetics (14) and thus may be a major branching point for the productive and unproductive paths in the catalytic cycle of this cytochrome P450. We describe in this work the kinetics of autoxidation as a function of TS concentration and show that the oxy-ferrous intermediate in CYP3A4 was significantly stabilized by the binding of the first substrate molecule. The magnitude of this effect clearly does not correlate with the minor high spin fraction measured under the same conditions but rather reflects an increased stability of the oxy-ferrous complex due to alteration of the local dielectric environment, formation of a significant steric and/or dynamic restriction for the escape of superoxide anion from the heme distal pocket caused by the bound TS molecule, or a protein conformational change linked to either of these processes. The formation of a steric block following the binding of a single TS molecule was confirmed by measurement of the amplitude of CO geminate rebinding, which was also dramatically increased at low TS concentrations. Both results indicate that not only the equilibrium properties of CYP3A4 but also the kinetics of interaction with diatomic ligands are modulated by the presence of TS in a complicated and non-trivial way.
The goal of the current study was to probe the kinetic properties of CYP3A4 at different redox and ligation states in the presence of different concentrations of testosterone. The changes of autoxidation kinetics provide information about the influence of substrate binding to CYP3A4 on the stability of the oxy-ferrous complex and the rate of superoxide escape from the distal pocket in the resultant oxidized ferric protein. Clearly the first TS binding event stabilizes the oxy-ferrous complex. Through water displacement upon substrate binding, a less polar environment near the heme iron may hinder the separation of charge occurring in the autoxidation reaction. In addition, substrate binding may increase the effective "geminate yield" of superoxide release by blocking an escape pathway. To probe the active site accessibility in the ferric state, the kinetics of cyanide binding was also measured as a function of TS concentration and compared with the geminate rebinding of carbon monoxide as a characteristic of the ferrous CYP3A4 conformation. We note a strong similarity between the changes of the fraction of the slow phase in autoxidation kinetics and the amplitude of geminate CO rebinding. A sharp increase of the geminate rebinding amplitude and of the fraction of the slow phase in autoxidation kinetics at low TS concentrations suggests that the escape of diatomic heme ligands, superoxide and carbon monoxide, is significantly hindered by the first testosterone molecule bound to CYP3A4. This conclusion is based on the large difference between these properties and the high spin fraction measured under the same conditions (Figs. 6 and 7). As has been documented, the shift of CYP3A4 spin state is caused by binding of the second TS molecule (15), and hence this is an effect occurring with only a single TS molecule bound at the active site.
The substrate access to and escape from the active center of cytochromes P450 has been studied by analysis of available x-ray crystal structures and molecular dynamics as reviewed in Refs. 61 and 62. Molecular dynamics has also been used to probe the pathways to and from the heme distal pocket of cytochromes P450 that can be accessed by small molecules, water for several human enzymes (63) and CO for CYP101 (29). All of these studies identify several pathways that can be either found in the x-ray structures or open and close transiently. Thus, the ability of the first TS molecule to hinder the escape of superoxide and carbon monoxide from CYP3A4 is in part unexpected because it is difficult to visualize a single substrate molecule blocking multiple paths at the same time in the absence of a major conformational change in the enzyme induced by the first TS binding event. Nevertheless it is still possible that one pathway may dominate for the escape of CO and superoxide and that this is blocked by the first TS bound. If the escape rates of these ligands through other pathways are much lower, the overall observed kinetics may mostly depend on the positioning of the substrate molecule at the main escape pathway. In addition, the dynamics of the whole CYP3A4 molecule may change as a result of the first TS binding, and the escape through other pathways, even if not blocked directly, may become much slower in the presence of substrate. A third possibility, which can be eliminated, is an unusual linkage between the sequential binding of the three testosterone molecules and the redox state of the CYP3A4 heme. Recent results in our laboratory suggest that testosterone binding significantly alters the observed redox potential of CYP3A4 and that this shift is directly correlated with the ferric spin state of the heme iron. 3 Hence the difference observed in the effect of testosterone concentration on the autoxidation of the oxy-ferrous intermediate as compared with the modulation of the ferric spin state equilibrium suggests that the tightest binding TS molecule can alter the flux through this uncoupling pathway through direct steric effects or a change in the global conformational dynamics of the enzyme.