Role of subunit interactions in P450 oligomers in the loss of homotropic cooperativity in the cytochrome P450 3A4 mutant L211F/D214E/F304W

Abstract

The contribution of conformational heterogeneity to cooperativity in cytochrome P450 3A4 was investigated using the mutant L211F/D214E/F304W. Initial spectral studies revealed a loss of cooperativity of the 1-pyrenebutanol (1-PB) induced spin shift (S₅₀ = 5.4 μM, n = 1.0) but retained cooperativity of α-naphthoflavone binding. Continuous variation (Job’s titration) experiments showed the existence of two pools of enzyme with different 1-PB binding characteristics. Monitoring of 1-PB binding by fluorescence resonance energy transfer from the substrate to the heme confirmed that the high affinity site (K_D = 0.3 μM) is retained in at least some fraction of the enzyme, although cooperativity is masked. Removal of apoprotein on a second column increased the high spin content and restored cooperativity of 1-PB binding and of progesterone and testosterone 6β-hydroxylation. The loss of cooperativity in the mutant is, therefore, mediated by the interaction of holo- and apo-P450 in mixed oligomers.

Introduction

In recent years extensive studies have shown that several mammalian cytochromes P450 (1A2, 2C9, 2B6, and 3A4) exhibit homo- and heterotropic cooperativity toward a number of substrates [1–7]. The most prominent examples of these phenomena are observed with P450 3A4 (CYP3A4), and studies of cooperativity of this enzyme are of considerable importance in elucidating the mechanisms of drug metabolism and drug-drug interactions in humans [7]. The proposed models for the cooperativity in cytochromes P450 involve multiple substrate molecules in one large binding pocket [8–10]. Direct physical evidence for the presence of two ketoconazole molecules in the CYP3A4 active site was provided very recently by x-ray crystallography [11]. Inherent in most of the models is the assumption that a loose fit of a single substrate molecule requires the binding of a second ligand for efficient binding and/or catalysis [9, 10, 12], although recent evidence for effector-induced conformational rearrangements has emerged from our laboratory and others [13–15]. Strong support for an important conformational transition in P450 function is provided by several recent x-ray crystal structures of mammalian P450 enzymes including CYP3A4, which have revealed considerable flexibility [11, 16–18].

Elucidating the mechanisms of CYP3A4 cooperativity is complicated by the persistent conformational heterogeneity observed, whereby the enzyme is thought to be represented by at least two conformers with different ligand-binding properties [19–24]. This functional heterogeneity appears to be linked closely to the oligomeric state of the enzyme [19, 20]. Microsomal cytochromes P450 are known to form oligomers both in solution [25–27] and in membranes [28–33], and direct association between P450s has been recognized as an important determinant of function [34–37]. In particular, functional heterogeneity of a single P450 species caused by oligomerization has been revealed in interactions with substrates [19, 38, 39].

Although most analysis of P450 cooperativity has been based on steady-state kinetics of substrate oxidation, more recent attention has focused on investigating the substrate binding step in the CYP3A4 catalytic cycle [20, 40–49]. In general, cooperativity in substrate binding is revealed by a sigmoidal curve of the substrate-induced displacement of the spin equilibrium of the heme iron, which is known to be an important determinant of the catalytic efficiency and coupling of cytochromes P450 [50–52]. Understanding the structural relationship between the binding of each substrate and/or effector molecule and the spin equilibrium of the heme iron is essential in unveiling the mechanism of P450 cooperativity.

Recent use of fluorescence resonance energy transfer (FRET) from the fluorescent substrate 1-pyrenebutanol (1-PB) to the heme of cytochrome P450eryF (P450 107A1) [14, 15, 40] or CYP3A4 [48] has allowed us to determine the dissociation constant of the first binding event directly. The results suggest that both enzymes possess two binding sites with different affinities for 1-PB. Binding of the substrate to the higher affinity binding site does not modulate the spin equilibrium but rather triggers a conformational transition that facilitates the spin shift elicited by substrate binding at a second, lower affinity site. This inference is also confirmed by the studies of the testosterone-induced spin shift in CYP3A4 by a combination of absorbance and EPR spectroscopy [43], and the interactions of CYP3A4 with testosterone or α-naphthoflavone (ANF) monitored by changes in the tryptophan fluorescence [53]. Work by Baas and co-authors deduced a similar conclusion from studies of the testosterone-induced spin shift in monomeric CYP3A4 incorporated into a nanoscale lipid bilayer (Nanodiscs) [42]. Recent work by Isin and Guengerich proposed a three-step binding model in which the first step does not perturb the heme spectrum and involves binding at a site peripheral to the active site [44]. This hypothesis is consistent with the observation of a peripheral progesterone binding site in a CYP3A4 X-ray crystal structure [54].

One means of elucidating mechanisms of cooperativity is through the use of site-directed mutants [12, 55, 56]. In the present study, we applied our recent spectroscopic approaches to the CYP3A4 mutant L211F/D214E/F304W (FEW), which displays no homotropic cooperativity of progesterone hydroxylation and abolished heterotropic activation by ANF [55]. Interestingly, two of the residues that are altered in this mutant, L211 and D214, reside in the proposed peripheral progesterone binding site [54], whereas in the ketoconazole-bound structure, L211 and F304 contact the proximal ligand molecule [11]. To gain insight into the basis for lost cooperativity in CYP3A4 FEW, we studied the interactions of the enzyme with 1-PB, bromocriptine (BCT), and ANF monitored by absorbance spectroscopy. In the case of 1-PB this was also complemented by direct monitoring of substrate binding by FRET to the heme. We also studied the effect of removal of apoprotein from CYP3A4 FEW on its interactions with substrates. The results suggest that the loss of cooperativity in the mutant is mediated by the interaction of holo- and apo-P450 in mixed oligomers, and that the conformational heterogeneity of CYP3A4 in the oligomers is involved in the mechanisms of cooperativity.

Experimental Procedures

Materials

1-PB was from Molecular Probes (Eugene, OR), ANF was from Indofine Chemical Company (Hillsbrough, NJ). BCT mesylate and Igepal CO-630 was from Sigma Chemicals (St. Louis, MO). All other chemicals were of the highest grade available from commercial sources and were used without further purification.

Expression and purification of the CYP3A4 mutant L211F/D214E/F304W

The enzyme was expressed as the His-tagged protein in Escherichia coli TOPP3, and the first purification was carried out using Ni-NTA resin (QIAGEN) under conditions described previously [12, 55, 56]. The protein was stored at −80 °C in 100 mM N-2-hydroxyethylpiperazine-N′-[2-ethanesulfonic acid] (HEPES) buffer (pH 7.4), containing 10% glycerol (v/v), 1 mM, dithiothreitol (DTT), and 1 mM ethylenediaminetetraacetic acid (EDTA) (buffer A). Some of the experiments (when indicated) were carried out with highly purified preparations of CYP3A4 FEW obtained through a combination of the above single-column procedure and a second ion-exchange purification step performed as described by Tsalkova et al. [57] with some modifications. Herein, small aliquots (60 nmol) of the protein in 30 mM HEPES (pH 7.4), containing 16% glycerol (v/v), 1 mM DTT, 1 mM EDTA, and 0.2% Igepal CO-630 were loaded onto a CM Sepharose CL-6B (Amersham) column (0.5 ml) equilibrated with the same buffer. The column was washed with 20–30 bed volumes of buffer A and with 10 bed volumes of buffer A containing 50 mM KCl. The protein was eluted with buffer A containing 200 mM KCl and stored at −80 °C. The total protein content was determined with a Bio-Rad protein assay (catalog # 500-0006) with bovine serum albumin as a standard, and the P450 content was determined from the difference spectra of the reduced CO-complex.

Experimental

The absorbance spectra were measured with a MC2000-2 multi-channel CCD rapid scanning spectrometer (Ocean Optics, Inc., Dunedin, FL, USA) using an L7893 UV-VIS fiber-optics light source (Hamamatsu Photonics K. K., Hamamatsu City, Shizuoka, Japan). Fluorescence measurements were performed with a Hitachi F-2000 fluorescence spectrometer equipped with a custom-designed thermostated cell holder. Here the excitation bandwidth was set to 10 nm. The spectra of emission were recorded in the 360 – 570 nm range and corrected for the changes in the intensity of the excitation light during the experiment as described [48].

The variable path length continuous variation (Job’s titration) experiments were carried out with a S2000 spectrometer (Ocean Optics Inc.) equipped with a custom-designed fiber optic adapter for a 10-cm long cylindrical cell (Cell Type-521, NSG Precision Cells, Farmingdale, NY) as described previously [48, 49]. All experiments were carried out at 25 °C in 100 mM HEPES buffer (pH 7.4), containing 1 mM DTT and 1 mM EDTA. An 8 – 15 mM stock solution of 1-PB in acetone or acetone/methanol (1:1) was used in all experiments.

Scanning stop-flow kinetic experiments were performed using a SF-MiniMixer Stopped-Flow apparatus (KinTek Corporation Austin, TX) with a 5 mm path length quartz cell combined with a rapid scanning MC2000-2 CCD-spectrometer (Ocean Optics, Inc., Dunedin, FL) as described previously [19]. All experiments were performed in 100 mM Na-HEPES buffer, pH 7.4, containing 1 mM dithiothreitol and 1 mM EDTA. The temperature was 25 °C, and the concentration of dithionite in the stop-flow cell was 12.5 mM. Both cytochrome P450 and sodium dithionite solutions were prepared using buffer saturated with CO by bubbling for 10 minutes. The spectral baseline was recorded with the buffer in the sample cell.

Data Processing

The series of absorbance and fluorescence spectra obtained in titration experiments were analyzed using principal component analysis (PCA) as described previously [14, 40, 58, 59]. To interpret the spectral transitions in terms of the concentration of P450 species we used a least-squares fitting of the spectra of principal components to the set of spectral standards of pure low-spin, high-spin and P420 species of the hemeprotein [49, 58, 59]. The series of spectra obtained in continuous variation with variable optical path length were normalized to the path length prior to further analysis [48, 49]. All data treatment procedures and curve fitting were performed using our SPECTRALAB software package [58].

We also used PCA in the analysis of the series of spectra obtained in rapid kinetic experiments [19]. This analysis gave us the curves representing time-dependent changes in the concentrations of the ferric low-spin, ferric high-spin, and the ferrous carbonyl complexes of P450 and P420 states of the heme protein. These kinetic curves were fitted to a multiexponential equation:

$$C_t=C_{\infty }-(C_{\infty }-C_0)·\sum _{i=1}^n{F}_i{e}^{-k_{i}t}$$

where C₀, C_t, and C_∞ are the concentrations of the compound at the beginning of the reaction, at time t and at an infinite time respectively; F_i and k_i are the fraction and the rate constant of i-th exponential phase. The number of exponents (n) varied from 1 to 3 depending on the case.

Analytical ultracentrifugation

Sedimentation velocity experiments were carried out using a Beckman-Coulter XL-A analytical ultracentrifuge with absorbance optics and an An60-Ti rotor at 20 °C and 60,000 rpm and monitoring at 419 nm. The analysis was carried out in 50 mM HEPES (pH 7.4) buffer containing 0.5 mM DTT, 0.5 mM EDTA, and 10% glycerol. Scans were analyzed with the continuous sedimentation coefficient (c(s)) distribution methods [60, 61] using the Sedfit program (www.analyticalultracentifugegation.com). For this analysis we used the value of the partial specific volume of 0.742 cm³/g.

Progesterone/testosterone hydroxylase assays and kinetic analysis of data

Purified FEW (5 pmol) were reconstituted by preincubation with 20 pmol of rat NADPH-cytochrome P450 reductase, 10 pmol of rat cytochrome b₅, 0.4% CHAPS (3-[(3-cholamidopropyl)-dimethylammonio]-1-propanesulfonic acid), and 100 mg/ml DOPC (dioleoylphosphatidylcholine). Reactions proceeded for 8 min at 37 ºC in 100 μl of 15 mM MgCl₂, 50 mM HEPES buffer (pH=7.6) in the presence of 100 mg/ml DOPC, and 0.04% CHAPS. 4-¹⁴C labeled testosterone (GE HealthCAre Bio-Sci, NJ) or progesterone (PerkinElmer Life Sciences) stock solutions in methanol with or without ANF were used as substrate at the concentrations indicated. Reactions were initiated by the addition of 1 mM NADPH (Sigma Chemical CO) and stopped after 8 min by the addition of 50 μl of tetrahydrofuran. Quantification of the metabolites by thin layer chromatography, autoradiography, and liquid scintillation counting was performed as described previously [62]. SPECTRALAB software package [58] was used to fit the data sets to the Hill or Michaelis-Menten equation.

Results

Interactions of CYP3A4 FEW with 1-PB, ANF, and BCT monitored by absorbance spectroscopy

Interactions of the substrates 1-PB, ANF, or BCT with the mutant CYP3A4 FEW purified as described previously [55] (peak maximum of oxidized spectrum at 419 nm) were monitored by substrate-induced changes in the Soret region absorbance spectra. All three substrates produced a Type I spectral change indicative of a low-to-high spin transition in the heme protein. Titrations with 1-PB are shown in Fig. 1a, and the % of high spin enzyme as a function of substrate concentration is shown in Fig. 1b. Fitting to the Hill equation and averaging the results of 3 individual experiments (Table 1) gives an S₅₀ value and Hill coefficient (n) of 5.4 ± 0.9 μM and 1.05 ± 0.08, respectively, compared with S₅₀ = 13.2 ± 2.9 μM and n = 1.6 ± 0.2 for CYP3A4 WT [57]. Thus, there is a clear loss of homotropic cooperativity of 1-PB binding in the mutant.

Figure 1.

Interactions of CYP3A4 FEW with 1-PB monitored by the substrate-induced spin shift. (a) A series of absorbance spectra obtained at no substrate present and at 1.4, 2.9, 4.3, 5.7, 8.6, 11, 19, and 29 μM 1-PB. (b) The same data shown as the plot of the percent high spin P450 versus the concentration of the substrate. The line shows the approximation of this data set by the Hill equation with S₅₀ = 6.31 μM, n = 0.97, and the maximal amplitude of the spin shift of 67%. The reaction mixture contained 0.50 μM CYP3A4 in 0.1 M Na-HEPES buffer (pH 7.4), 1 mM DTT, and 1 mM EDTA and was kept at 25 °C.

Table 1.

Parameters for the interactions of CYP3A4 WT and CYP3A4 FEW with 1-PB, ANF, and BCT^*

Protein	Parameter	Substrate
		1-PB	ANF	BCT
CYP3A4 WT	S50, or KD, μM N ΔHs,%b	13.2 ± 2.9 1.6 ± 0.2 40 ± 8	4.8 ± 2.2 1.9 ± 0.3 36 ± 8	0.65 ± 0.33 N/A 40 ± 15
One-column purified CYP3A4 FEW	S50, or KD, μM N ΔHs,%b	5.4 ± 0.9 1.05 ±0.08 42 ± 25	2.5 ± 0.2 2.0 ± 0.2 44 ± 12	0.30 ± 0.10 N/A 50 ± 18
Re-purified (2 columns) CYP3A4 FEW	S50, or KD, μM N ΔHs,%b	8.0 ± 3.7 1.6 ± 0.2 12 ± 4	3.2 ± 0.1 1.7 ± 0.1 13 ± 5	0.46 ± 0.18 N/A 19 ± 7

^*The values given in the table were obtained by averaging the results of three individual measurements, and the ± values show the confidence interval calculated for p=0.05.

^aIn the absence of substrate the content of the high-spin state in WT, FEW (1 column), and FEW (2 columns) are 16 ± 4%, 20 ± 6% and 49 ± 5% respectively.

^bTaken from Tsalkova et. al, [57], ΔH_s,%-Substrate-induced spin shift%.

In contrast, homotropic cooperativity of ANF binding to CYP3A4 FEW is retained (Table 1), in agreement with previous results [55]. It should be noted, however, that the values of S₅₀ for ANF obtained in our studies are lower than those reported earlier [55]. This distinction is attributable to the differences in experimental conditions, as Domanski and co-authors examined ANF binding in the presence of 0.1 mg/ml DOPC and 0.05% CHAPS. Under the current conditions, the S₅₀ value for wild type CYP3A4 is also lower than reported previously [57].

As with ANF, the spectrophotometric titration experiments revealed no considerable difference between CYP3A4 FEW and the wild type enzyme in their interactions with BCT (Table 1). Similar to the wild type enzyme [44, 48, 57], CYP3A4 FEW showed very high affinity, a substantial spin shift, and no cooperativity with this substrate (Table 1).

The stoichiometry of CYP3A4 FEW interactions with 1-PB with the continuous variation (Job’s titration) approach

The method of continuous variation, also known as the Job’s titration technique, is widely used for determination of the stoichiometry of enzyme-substrate interactions [63]. This technique is based on the variation of the molar ratio of two interacting species while keeping their total concentration constant [64]. Under conditions where the basic predicates of the method (homogeneity of the interacting species, absence of aggregation processes, etc.) are met, the position of the maximum of the bell-shaped titration curve obtained yields the stoichiometry of the interactions [63]. Applying our novel implementation of the Job’s method based on the variable path length titration technique [49] we recently demonstrated that the interaction of CYP3A4 with BCT obeys a stoichiometry of 1:1, while 1-PB binding is given by an asymmetric bell-shaped curve consistent with a sequential binding mechanism with 1:2 stoichiometry [48]. The titration curve obtained by this approach for the interactions of CYP3A4 FEW with 1-PB is shown in Figure 2. The shape of this asymmetric curve suggests a complex mechanism of interaction, which breaches the basic predicates of the Job’s titration theory [63]. The left branch of the curve (at excess enzyme) reaches the maximal level at a molar fraction of the substrate (F) of 0.5, which is consistent with 1:1 stoichiometry (Fig. 2, dashed line). However, the shape of the right branch (at excess substrate) deviates sharply from the mirror image of the left branch that would be expected for a simple 1:1 interaction mechanism. The maximal level reached at F=0.5 is retained until F=0.7, after which point the concentration of the complex decreases sharply. The descending branch of the bell (F≥0.7) obeys the shape expected for 1:2 stoichiometry (Fig. 2, solid line). This atypical behavior of the Job’s titration curve reveals a deviation from the simple bimolecular association mechanism that is not detectable in traditional titration experiments (Fig. 1b). The results indicate conformational heterogeneity of the enzyme, so that the total pool of CYP3A4 FEW in solution diverges into two persistent subpopulations, which differ in the stoichiometry of their interactions with 1-PB.

Figure 2.

Job’s titration of CYP3A4 FEW with 1-PB. The experiments were carried out at 25 °C in a 10-cm vertical cell with the optical path length increasing with the dilution of the sample. CYP3A4 FEW (500 μl) in 0.1 M Na-HEPES buffer (pH 7.4), 1 mM DTT, and 1 mM EDTA was placed into the optical cell. The substrate solution of the same concentration as the initial protein was added gradually until a volume of 5.1 ml, when the cell was completely filled. The sum of the concentration of the enzyme and the substrate was therefore kept constant during the experiment, and was equal to 11.6 μM. The solid line represents the approximation of the data set with the sequential binding mechanism [48] assuming a K_D1 equal to 0.3 μM, as determined from FRET experiments and the dashed line represent the approximation of the data set with the equation of bimolecular association.

Interactions of CYP3A4 with 1-PB monitored by FRET to the heme of the enzyme

In our previous studies we applied FRET from 1-PB to the P450 heme, in which fluorescencedue to determine the dissociation constant of the high-affinity binding site for this substrate in P450eryF and in CYP3A4 wild type [14, 40, 48]. The implementation of this approach used in the present work is based on a titration-by-dilution technique, where the concentration of the 1:1 mixture of the enzyme and substrate is changed gradually while keeping their molar ratio constant. The intensity of fluorescence normalizied on the concentration of the substrate is used to assess the concentration of the enzyme-substrate complex, in which the fluorescence of 1-PB is quenched due to FRET [48]. Figure 3a shows the changes in the specific fluorescence of 1-PB upon dilution of a 1:1 mixture with CYP3A4 FEW. As shown in Figure 3b, the dependence of the relative intensity of 1-PB fluorescence on the concentration of the enzyme-substrate mixture obeys the equation for the equilibrium of bimolecular association [65]. The value of the dissociation constant obtained by averaging the results of 3 individual experiments is 0.3 ± 0.15 μM. Similar to results with CYP3A4 WT [48], this value is considerably lower than the S₅₀ determined by the spectrophotometric titration (5.4 ± 0.9 μM, see Table 1) and suggests a multi-site binding mechanism. As reported previously with P450eryF and CYP3A4 WT [14, 40, 48], the binding of 1-PB to the high-affinity binding site detected in our FRET experiments with CYP3A4 FEW does not result in the displacement of the spin equilibrium of the enzyme.

Figure 3.

Changes in specific fluorescence of 1-PB upon dilution of a 1:1 mixture of CYP3A4 FEW and 1-PB. The signal was normalized based on protein concentration and corrected for the internal filter effect as described. Panel (a) represent the series of fluorescence spectra measured at 0.14, 0.28, 0.42, 0.70, 1.5, 2.4, 4.1, 6.1, and 10.2 μM enzyme. In all cases the increase in the concentration of enzyme-substrate mixture caused a decrease in the normalized intensity of emission. Panel (b) represents the titration curve obtained from these data using principal component analysis, as described before [48]. The solid line represents the approximation of the data set by the equation of bimolecular association with K_D1 of 0.19 μM. Conditions as indicated in Fig 1.

It should be noted, however, that the fitting of the results of our FRET experiments is only moderately sensitive to the possible heterogeneity of the enzyme discussed above. The curve in Fig. 3b represents the fitting based on the assumption that the whole CYP3A4 pool interacts with 1-PB. An increase in the apparent substrate-to-enzyme ratio to 2 based on the suggestion that only 1/2 of the enzyme participates in high-affinity binding increases the K_D to 0.7 μM. Therefore, the FRET experiments are consistent with the presence of two fractions of the enzyme inferred from the Job’s titration. Taken together the results suggest that the two 1-PB binding sites found in the wild-type enzyme are retained in at least some fraction of the CYP3A4 FEW despite the abolished cooperativity revealed in the 1-PB-induced spin shift.

Conformational heterogeneity of CYP3A4FEW revealed in the kinetics of dithionite-dependent reduction

Previous data from our laboratory on the kinetics of dithionite-dependent reduction showed that CYP3A4 oligomers in solution are represented by a mixture of three persistent (not interconvertible in the time frame of the experiment) conformers having different position of spin equilibrium and different rate constants of reduction [19]. In contrast, in CYP3A4 FEW the reduction by dithionite is considerably slower due to the loss of the fastest phase of the reaction and decreased rate constants of the two other phases (Figure 4, Table 2). Similar to results with the wild type enzyme, addition of BCT to CYP3A4 FEW decreases the reduction due to an increase in the amplitude of the slow phase from 32% in the absence of added substrate to 90% at 32 μM BCT (data not shown). Therefore, we may conclude that in FEW the distribution of conformers is altered, leading to biphasic as opposed to triphasic reduction kinetics and loss of the fraction that is unaffected by BCT binding in CYP3A4 WT. Furthermore, the decreased rate constants of dithionite-dependent reduction in CYP3A4 FEW suggest that the heme moiety is less accessible to the dithionite anion-radical than that in the wild type enzyme.

Figure 4.

Kinetics of dithionite-dependent reduction of CYP3A4 WT and FEW in solution. Conditions: 3 μM 3A4, 12.5 mM sodium dithionite, CO-saturated 0.1 M Na-HEPES buffer (pH = 7.4), 1 mM DTT, 1 mM EDTA, 25 °C. Spectra recorded in a stop-flow cell with 5 mm optical path length. Kinetics of reduction of CYP3A4 WT (circles) and CYP3A4 FEW (squares). The lines show the results of the fitting of the data to the equation of the sum of three (CYP3A4 WT) or two (CYP3A4 FEW) exponents, and the respective constants are given in Table 3. The data were scaled according to the fitting results to represent a percent decrease in the content of the respective states of the ferric enzyme. Inset shows the same data plotted in semilogarithmic coordinates.

Table 2.

Kinetic parameters of dithionite reduction of CYP3A4 WT and FEW^*

Protein	F1	F2	F3	k1, s−1	k2, s−1	k3, s−1
CYP3A4 WT	0.42 ± 0.03	0.29 ± 0.06	0.29 ± 0.03	5.7 ± 0.36	0.79 ± 0.04	0.14 ± 0.05
CYP3A4 FEW		0.68 ± 0.03	0.31 ± 0.04		0.38 ± 0.04	0.084 ± 0.001

^*The values given in the table were obtained by averaging the results of three individual measurements, and the ± values show the confidence interval calculated for p=0.05.

Removal of the apoprotein from CYP3A4 FEW

The single-column purification procedure used in the study of Domanski et al. [55], as well as in the experiments described above, resulted in preparations of CYP3A4 FEW with a specific content of 7.2 nmol/mg protein, which is 2 times less than the WT. However, SDS-PAGE indicates that the protein is >85% homogeneous (Fig 5a, Lane 2), suggesting the presence of a substantial amount of apoprotein. To probe the possible effect of this apoprotein we employed an additional ion-exchange chromatography step [57]. This yielded electrophoretically homogenous CYP3A4 FEW (Fig. 5a, lane 1) with a specific content of 14.5 nmol/mg, which is comparable to the value of 16.5 nmol/mg obtained with the wild type enzyme purified by the same procedure. Interestingly, this additional purification step resulted in a considerable displacement of the spin equilibrium of CYP3A4 FEW towards the high spin state, whereas no such effect of further purification on the spin state of CYP3A4 WT was observed (Fig. 5b,c).

Figure 5.

Purification and spectra of CYP3A4. Panel (a) represents the SDS-PAGE analysis. Holo-P450 (0.013 nmol) was loaded in each lane. LN 1, CYP3A4 FEW (2 columns), LN 2, CYP3A4 FEW (1 column), LN 3, CYP3A4 WT and LN 4, molecular weight standards. The gel was stained with Coomassie blue. Panel (b) panel represents spectra of CYP3A4 WT purified from one (dashed line) and two (solid line) columns. Panel (c) represents CYP3A4 FEW from one (dashed line) and two (solid line) columns. All spectra are normalized to 1 μM heme protein.

Interactions of second-column purified CYP3A4 FEW with 1-PB, ANF, and BCT monitored by absorbance spectroscopy

As shown in Fig. 6 and Table 1, further purification of CYP3A4 FEW restored the cooperativity of 1-PB binding. Averaging the results of 3 individual experiments gives an S₅₀ value and Hill coefficient of 8.0 ± 3.7 μM and 1.6 ± 0.2 respectively, very similar to previous results with CYP3A4 WT (Table 1). In contrast, the parameters of ANF or BCT binding were essentially unaltered by the removal of apoprotein (Table 1). With all three substrates, the substrate-induced increase in % high spin enzyme was considerably lower in the more highly purified preparation of CYP34 FEW, presumably because of the high intrinsic high spin content in the absence of substrate. These results indicate that the loss of cooperativity of 1-PB binding to CYP3A4 FEW does not reflect the properties of a single molecule of the enzyme, but is rather caused by changes in protein-protein interactions in P450 oligomers. We may hypothesize therefore that in CYP3A4 FEW the architecture of the oligomer is altered in such a way that the interactions of the apo- P450 with the holoenzyme (co-oligomerization of the apo- and holoenzyme) affects the spin state of the heme protein and changes the mechanism of its interactions with 1-PB, thus resulting in the loss of cooperativity.

Figure 6.

Interactions of CYP3A4 FEW (2 columns) with 1-PB monitored by the substrate-induced spin shift. (a) A series of absorbance spectra obtained at no substrate present and at 4.1, 8.2, 12, 19, 25, and 31 μM 1-PB. (b) The same data shown as a plot of percent high spin P450 versus the concentration of the substrate. The line shows the approximation of this data set by the Hill equation with S₅₀ = 11.6 μM, n = 1.6, and the maximal amplitude of the spin shift of 13%. Conditions as indicated in Fig 1.

Kinetics of testosterone and progesterone metabolism

Figure 7 and Table 3 show the kinetic analysis of testosterone and progesterone 6β-hydroxylation by CYP3A4 FEW purified by two different methods. CYP3A4 FEW purified by the one-column procedure shows hyperbolic behavior with testosterone and progesterone in the absence and presence of ANF. However, kinetic analysis of CYP3A4 FEW purified by the two-column procedure reveals sigmoidal behavior with both substrates in the absence of ANF and hyperbolic behavior in the presence of ANF. Thus the catalytic assays are completely consistent with the experiments involving the 1-PB induced spin shift and show that the loss of cooperativity with testosterone and progesterone [56] represent a specific feature of a mixed oligomer of the holo- and apoenzyme rather than being inherent to a single P450 molecule.

Figure 7.

Kinetic analysis of testosterone and progesterone 6β-hydroxylation. Closed circles represent the absence of ANF and open circles represent the presence of 25 μM ANF. Data were fitted with the Michaelis-Menton or the Hill equation. Panel a: testosterone hydroxylation by CYP3A4 FEW purified with one-column procedure in the absence (k_cat= 41.0 min⁻¹ and K_m= 57.2 μM) and in the presence of ANF (k_cat = 39.0 min⁻¹ and K_m 38.2 μM). Panel b: testosterone hydroxylation by CYP3A4 FEW purified with two-column precedure in the absence (k_cat = 21.5 min⁻¹, n = 1.6, and S₅₀= 55.2 μM) and in the presence of ANF (k_cat = 19.1 min⁻¹ and K_m= 50.4 μM). Panel c: progesterone hydroxylation by CYP3A4 FEW purified with one-column procedure in the absence (k_cat = 29.9 min⁻¹ and K_m= 70.3 μM) and in the presence of ANF (k_cat = 28.9 min⁻¹ and K_m= 61.1 μM). Panel d: progesterone hydroxylation by CYP3A4 FEW purified with two-column procedure in the absence (k_cat = 11.1 min⁻¹, n = 1.6, and S₅₀= 43.0 μM) and in the presence of ANF (k_cat = 16.6 min⁻¹ and K_m= 81.8 μM).

Table 3.

Kinetic constants for testosterone and progesterone 6β-hydroxylation catalyzed by CYP3A4 FEW in the absence and in the presence ANF^*

Protein	Substrate	Parameter	ANF
			none	25 μM
One-column purified CYP3A4 FEW	testosterone	kcat (min−1) Km (μM)	40 ± 1.5 56 ± 2.0	36 ± 6.3 37 ± 1.8
	progesterone	kcat (min−1) Km (μM)	30 ± 0.4 75 ± 8.6	26 ± 5.5 57 ± 8.4
Re-purified (2 columns) CYP3A4 FEW	testosterone	kcat (min−1) n S50, or Km (μM)	24 ± 4.7 1.6 ± 0.02 53 ± 5.1	22.0 ± 5.6 N/A 43 ± 15
	progesterone	kcat (min−1) n S50, or Km (μM)	12.5 ± 2.8 1.4 ± 0.3 52 ± 17	17 ± 0.6 N/A 77 ± 9.5

^*The values given in the table were obtained by averaging the results of two individual measurements, and the ± values show the confidence interval calculated for p=0.05.

Instability of the highly purified mutant

Prolonged incubation of the highly purified CYP3A4 FEW at ambient temperature enriched the sample in apoprotein and makes the high spin content in the sample similar to that characteristic to single-column purified enzyme (Fig 8a). As shown in Fig. 8b the total hemeprotein content slowly decreases with time due to apparent heme loss and formation of apoprotein in the aged sample. This process is concomitant with a corresponding decrease in the high spin fraction and increase in the content of both low-spin P450 and P420 fractions. This result corroborates the above conclusion that the interactions between holo- and apo-CYP3A4 FEW affects the properties of the enzyme.

Figure 8.

Changes in the absorbance of CYP3A4 FEW upon incubation. (a) A series of absorbance spectra obtained at 0, 5, 10, 25, 55, 250 and 600 minutes of incubation. (b) Changes in the concentration of the low-spin P450 (circles), high-spin P450 (triangles), P420 (squares) and the total heme protein (diamonds) during the incubation. Lines show the fitting of the data sets by bi-exponential equation. The reaction mixture contained 1.6 μM CYP3A4 in 0.1 M Na-HEPES buffer (pH = 7.4), 1 mM DTT, and 1 mM EDTA and was kept at 25 °C.

Conformational heterogeneity of CYP3A4 FEW observed by sedimentation velocity experiments

Previous results of sedimentation velocity experiments from our laboratory on CYP3A4 WT at 1-3 μM concentrations indicated the presence of large aggregates with sedimentation coefficients ranging from 8.2–12.6 S [19]. Application of this technique to monitor the aggregation state of CYP3A4 FEW purified by the two different methods is shown in Fig. 9a. CYP3A4 FEW purified from one column has larger aggregates than CYP3A4 WT, but both show bimodal peak distributions. Applying the second column removes these larger aggregates of CYP3A4 FEW and gives a sedimentation profile that is more homogeneous. Fig. 9b shows the sedimentation profiles of single-column purified CYP3A4 FEW, doubly purified CYP3A4 FEW, and the aged protein where the Soret maximum returns to 419 nm. The sedimentation profiles clearly show that the loss of the preferential high-spin state in the aged, doubly purified CYP3A4 FEW is accompanied by a substantial increase in heterogeneity of the aggregation state.

Figure 9.

Analytical ultracentrifugation data. (a) The sedimentation profiles for 1.0 μM CYP3A4 WT, 1.0 μM FEW (1), and 1.0 μM FEW (2). The buffer contained 0.05 M Na-HEPES buffer (pH 7.4), 0.5 mM DTT, 0.5 mM EDTA and 10% glycerol. The temperature was 20 °C, and the monitoring wavelength was 419 nm. (b) The sedimentation profiles for 0.5 μM FEW (1), 0.6 μM FEW (2), and 0.3 μM FEW (2) incubated at 25 °C overnight as shown in Fig. 6. The experiment was run at 20 °C, and the monitoring wavelength was 419 nm.

Discussion

The goal of the present investigation was to elucidate the basis of the loss of cooperativity in the previously described CYP3A4 mutant L211F/D214E/F304W. This construct was derived from our initial observation of abolished homotropic cooperativity of testosterone and progesterone 6β-hydroxylation and of testosterone binding in L211F/D214E [12]. Interestingly, however, whereas heterotropic activation of testosterone hydroxylation by ANF was lost, progesterone hydroxylation by L211F/D214E was still simulated by ANF. In order to prepare a mutant completely devoid of heterotropic activation of progesterone 6β-hydroxylation, F304W was added to the double mutant to create L211F/D214E/F304W (CYP3A4 FEW). This mutant exhibited hyperbolic progesterone 6β-hydroxylation kinetics that were unaffected by the presence of ANF and were indistinguishable from those of ANF-activated wild-type enzyme [55]. Accordingly, CYP3A4 FEW appeared to be a perfect mimic of the ANF-bound wild-type enzyme. One puzzling observation, however, was that the mutant retained cooperative binding of ANF (assessed from the Type I spectral shift) and actually exhibited an S₅₀ that was 3-fold lower than wild-type, although ANF 5,6-oxidation was decreased 8-fold. These results suggested a complex relationship between substrate and effector binding, oxidation, and action. However, further progress was not possible until the very recent development of advanced spectroscopic methods for investigating multiple ligand binding to CYP3A4 [19, 48] and the elucidation of ligand free and ligand bound x-ray crystal structures [11, 54, 66].

With the availability of methods to determine two separate constants for binding of 1-PB to CYP3A4 WT, we screened several mutants and found that FEW shows a loss of cooperativity of the 1-PB induced spin shift. We expected this to be a result of the blockage of the effector site, as proposed initially [12, 55]. To probe the stoichiometry of the binding we performed a Job’s titration of CYP3A4 FEW with 1-PB. At first glance the shape of the Job’s titration curve obtained here is consistent with 1:1 binding, as it reaches the maximal level at F=0.5. However, the atypical shape of this asymmetric curve (Fig. 3) indicates that 1-PB binding to CYP3A4 FEW is inconsistent with the underlying assumptions of the classical Job’s titration. This shape is likely to reflect heterogeneity of the enzyme, so that the total pool of CYP3A4 FEW diverges into two subpopulations with different stoichiometry of the interactions with 1-PB. However, when FEW was used in FRET experiments, we observed a high affinity site with a dissociation constant 18-fold lower than that obtained from the spin shift, demonstrating that at least some fraction of the mutant retains both sites. In addition, the dissociation constant derived from FRET experiments with FEW is similar to that observed in CYP3A4 WT, suggesting that the high affinity binding event remains unaltered in some population of the mutant despite the masked cooperativity.

To probe the conformational heterogeneity of the enzyme we studied the kinetics of reduction of CYP3A4 FEW by dithionite. In contrast to the three-exponential kinetics observed with the wild type enzyme, the reduction of CYP3A4 FEW obeys a bi-exponential equation. However, similar to the wild type enzyme, addition of BCT results in redistribution of the phases towards the slowest one, suggesting that the biphasic reduction of CYP3A4 FEW is due to loss of the fraction that is unaffected by BCT binding in the wild type enzyme. As the heterogeneity of dithionite-dependent reduction of CYP3A4 was shown to be caused by oligomerization of the enzyme [19], we may conclude that the subunit interactions in the oligomer are considerably altered in CYP3A4 FEW.

This conclusion was further corroborated by our experiments on the effect of apoprotein on the properties of CYP3A4 FEW. In our initial experiments as well as prior activity studies, we used a single column to purify the protein [12, 55], which yields a lower specific content than that of CYP3A4 WT, reflecting the presence of apoprotein as verified by SDS-PAGE. Removal of apoprotein on a CM-Sepharose as described recently [57] yields FEW protein with a high % of high spin enzyme, resulting in displacement of the Soret peak maximum to 396 nm. In contrast, no such change in the content of high spin P450 is observed following further purification of CYP3A4 WT. Intriguingly, the doubly purified mutant regained the cooperativity of 1-PB binding and of testosterone and progesterone 6β-hydroxylation. These results lead us to the conclusion that the high content of the high-spin heme protein in the absence of substrate is an inherent feature of the CYP3A4 FEW mutant, similar to P450 1A2 [67]. However, the interactions of CYP3A4 FEW holoenzyme with apoprotein result in the displacement of the spin equilibrium towards the low-spin state and abolish the cooperativity with 1-PB, testosterone, and progesterone.

To gain further insight into the role of apoprotein in the properties of FEW we used sedimentation velocity experiments to monitor the oligomeric state of CYP3A4 WT, single-column purified CYP3A4 FEW, and doubly purified mutant. Sedimentation velocity shows more heterogeneity and the presence of higher order oligomers in single-column purified protein than in doubly purified enzyme. These experiments demonstrated a distinct effect of apoprotein on the oligomerization of the enzyme in solution. These findings with CYP3A4 FEW are consistent with prior experimental data, which demonstrated an important impact of oligomerization on substrate binding and other functional properties of such microsomal P450 enzymes as CYP2B4, CYP1A2 and CYP3A4 [19, 20, 35, 38, 58].

The novel aspect of the FEW mutant is the possibility of correlating the altered function with the location of the modified residues in x-ray crystal structures. For this purpose, we examined an x-ray crystal structure of CYP3A4 with progesterone bound at a peripheral binding site [54] and a more recent structure with two ketoconazole molecules bound in the active site [11]. In the first structure, progesterone binds in the vicinity of L211 and D214 [54]. Here, similar to the substrate-free enzyme [54, 66], the side chains of these residues point away from the active site. L211 and D214 are close to a Phe cluster containing Phe-213, Phe-215, Phe-219, Phe-220, Phe-241, and Phe-304, which is suggested to comprise a flexible region and a dimer interface [68]. However, in the ketoconazole complex this hydrophobic cluster is broken up, and Leu-211 and Phe 304 point into the active site (see Fig. 3 in [11]),. It appears likely that the substitutions of Leu-211, Asp-214, and Phe-304 implemented in our FEW mutant may disrupt or modify this hydrophobic cluster by reorienting the side chains of residues 211 and 304. Such a modification of a large superficial hydrophobic patch is likely to affect the subunit interactions, which is apparently reflected in the abolished cooperativity in the mixed oligomers of the holo- and apo-CYP3A4 FEW found in the single-column purified enzyme. These changes make the formation of the heterooligomer of the apo- and holoenzyme capable of modulating the cooperativity. This interpretation is favored by the observation that homotropic cooperativity of FEW is restored upon removal of apoprotein.

In summary, our results on CYP3A4 FEW presented here show that the altered cooperativity in this mutant is not interpretable within any model involving a single molecule of the enzyme. Rather, we propose that the loss of cooperativity involves the interaction of holo-CYP3A4 with its heme-depleted apoprotein derivative. This result provides the first direct indication of the involvement of P450 oligomerization in the mechanisms of cooperativity. We suggest that in CYP3A4 FEW some of the locations in the P450 oligomer are preferable for binding of apoprotein. Such occupancy of one of the kinetically distinguishable conformers by apoprotein results in elimination of one of three phases of the dithionite-dependent reduction in single-column purified CYP3A4 FEW. The interaction between apo- and holoprotein is likely to affect the conformation and/or conformational dynamics of the CYP3A4 apoprotein resulting in a displacement of the spin equilibrium of CYP3A4 FEW and in the loss of its cooperativity with 1-PB, testosterone, and progesterone. Therefore, further insight into the mechanisms of cooperativity requires consideration of the interactions between several molecules of the enzyme. Furthermore, our results suggest an important role of the phenylalanine cluster (Phe213, Phe215, Phe219, Phe220, Phe241, and Phe304) in the mechanisms of cooperativity and in the interactions among CYP3A4 molecules in the oligomer.