Abstract
CYP3A4 has been subjected to random and site-directed mutagenesis to enhance peroxide-supported metabolism of several substrates. Initially, a high-throughput screening method using whole cell suspensions was developed for H2O2-supported oxidation of 7-benzyloxyquinoline. Random mutagenesis by error-prone polymerase chain reaction and activity screening yielded several CYP3A4 mutants with enhanced activity. L216W and F228I showed a 3-fold decrease in Km, HOOH and a 2.5-fold increase in kcat/Km, HOOH compared with CYP3A4. Subsequently, T309V and T309A were created based on the observation that T309V in CYP2D6 has enhanced cumene hydroperoxide (CuOOH)-supported activity. T309V and T309A showed a >6- and 5-fold higher kcat/Km, CuOOH than CYP3A4, respectively. Interestingly, L216W and F228I also exhibited, respectively, a >4- and a >3-fold higher kcat/Km, CuOOH than CYP3A4. Therefore, several multiple mutants were constructed from rationally designed and randomly isolated mutants; among them, F228I/T309A showed an 11-fold higher kcat/Km, CuOOH than CYP3A4. Addition of cytochrome b5, which is known to stimulate peroxide-supported activity, enhanced the kcat/Km, CuOOH of CYP3A4 by 4- to 7-fold. When the mutants were tested with other substrates, T309V and T433S showed enhanced kcat/Km, CuOOH with 7-benzyloxy-4-(trifluoromethyl)coumarin and testosterone, respectively, compared with CYP3A4. In addition, in the presence of cytochrome b5, T433S has the potential to produce milligram quantities of 6β-hydroxytestosterone through peroxide-supported oxidation. In conclusion, a combination of random and site-directed mutagenesis approaches yielded CYP3A4 enzymes with enhanced peroxide-supported metabolism of several substrates.
Mammalian cytochromes P450 comprise a ubiquitous superfamily of heme-containing enzymes, which perform a variety of oxidative reactions on a wide range of substrates including >90% of drugs and environmental pollutants (Coon, 2005). P450-derived metabolites are often biologically active themselves, and understanding their effects is crucial in evaluating a drug's efficacy, toxicity, and pharmacokinetics. Such studies, however, can require large quantities of the pure metabolites, which may be difficult to synthesize. An alternate approach is to use human P450s to generate the metabolites of drugs and drug candidates. Limitations include poor activity, stability, and expression of P450s in Escherichia coli. In addition, a troublesome reconstitution of P450s, which involves NADPH-cytochrome P450 reductase (CPR), often cytochrome b5 (b5), and phospholipids concurrent with an expensive cofactor, NADPH, further limits their application in synthesis. Therefore, designing xenobiotic-metabolizing P450s having enhanced activity, stability, and expression, and using an alternate oxygen donor such as hydrogen peroxide (H2O2) or cumene hydroperoxide (CuOOH), is highly desirable (Kumar and Halpert, 2005).
CYP3A4 is the most abundant P450 enzyme in human liver and metabolizes a wide variety of drugs (>50%) and carcinogens (Guengerich, 1999; Nebert and Russell, 2002). Because of its pharmacological and environmental significance, CYP3A4 has been the subject of a large number of structure-function studies involving X-ray crystallography, protein modeling, and site-directed mutagenesis (Domanski and Halpert, 2001; Hutzler and Tracy, 2002; Ekins et al., 2003; Tanaka et al., 2004; Williams et al., 2004; Yano et al., 2004; Park et al., 2005; Scott and Halpert, 2005). However, no attempt has been made to engineer CYP3A4 for enhanced catalytic activity, especially in a peroxide system. More recently, microsomal and E. coli-expressed CYP3A4-CPR with an NADPH generating system has been used to produce drug metabolites in milligram quantities (Vail et al., 2005).
Over the past decade, directed evolution has been used successfully to design industrial biocatalysts for enhanced catalytic efficiency, stability, and versatility (Schoemaker et al., 2003; Gillam, 2005). Exciting recent results with the bacterial enzyme P450 BM3 have illustrated the potential of directed evolution for engineering P450s that use artificial oxygen donors (Glieder et al., 2002; Meinhold et al., 2005). Recently, P450 BM3 has been engineered by directed evolution to produce the authentic human metabolites of propranolol (Otey et al., 2006). Directed evolution of several mammalian P450s for enhanced activity has been established (Kim and Guengerich, 2004) and their potential applications have been reviewed recently (Kumar and Halpert, 2005). In addition, we have developed a directed evolution approach to engineer P450 2B1 for enhanced H2O2-supported 7-ethoxy-4-trifluoromethylcoumarin O-deethylation (Kumar et al., 2005a). In the present study, we have engineered CYP3A4 using the combination of directed evolution and site-directed mutagenesis for enhanced utilization of peroxide. The most efficient enzyme, F228I/T309A, shows a >11-fold higher kcat/Km, CuOOH than CYP3A4 for 7-benzyloxyquinoline (7-BQ) debenzylation. In addition, an engineered enzyme, T433S, can metabolize testosterone more efficiently than CYP3A4 and has the potential to produce milligram quantities of 6β-OH testosterone.
Materials and Methods
Materials. 7-BQ, 7-BFC [7-benzyloxy-4-(trifluoromethyl)coumarin], 7-hydroxyquinoline, and 7-hydroxy-4-(trifluoromethyl)coumarin were purchased from Invitrogen (Carlsbad, CA). Polymyxin B sulfate, H2O2, CuOOH, and NADPH were bought from Sigma Chemical Co. (St. Louis, MO). [4-14C]Testosterone was obtained from GE Healthcare (Little Chalfont, Buckinghamshire, UK). Recombinant CPR and b5 from rat liver were prepared as described previously (Harlow and Halpert, 1997). Oligonucleotide primers for polymerase chain reaction (PCR) were obtained either from the University of Texas Medical Branch (UTMB) Molecular Biology Core Laboratory (Galveston, TX) or from Sigma Genosys (Woodlands, TX). Error-prone PCR and ligation kits were obtained from Roche (Indianapolis, IN). The QuikChange XL site-directed mutagenesis kit was obtained from Stratagene (La Jolla, CA). Nickelnitrilotriacetic acid affinity resin was purchased from QIAGEN (Valencia, CA). All other chemicals were of the highest grade available and were obtained from standard commercial sources.
Random Mutagenesis by Error-Prone PCR and Construction of Mutant Libraries. Random mutagenesis was essentially performed using errorprone PCR of the cDNA as described previously for P450 2B1 (Kumar et al., 2005a). Initially, error-prone PCR was standardized to ensure a mutation rate of 1 to 2 base pairs per P450 by using 25 mM MgCl2, 10 mM dCTP, 10 mM dTTP, and 2 mM each dATP and dGTP (Cirino et al., 2003). Subsequently, ligation and transformation were standardized to obtain >1000 clones per PCR by selecting suitable forward and reverse primers (Fig. 1), restriction sites (NcoI/KpnI), ligation conditions (overnight ligation kit), and transformation procedures (commercial XL-Blue E. coli).
Site-Directed Mutagenesis. T309A, T309I, and T309F were created as described previously (Domanski et al., 2001). Whereas T309V was created using CYP3A4 as a template, other multiple mutants were created using a single/double mutant as a template and appropriate forward and reverse primers as presented in Fig. 1. To confirm the desired mutation and verify the absence of unintended mutations, all constructs were sequenced at the University of Texas Medical Branch Protein Chemistry Laboratory (Galveston, TX).
Screening and Selection. Screening and selection of random mutants were essentially done as described previously for P450 2B1 (Kumar et al., 2005a). Liquid handling was performed using a multichannel robot Biomek 2000 (Beckman Coulter, Fullerton, CA). In brief, transformed E. coli was grown in a 300-μl, V-shaped 96-well microplate containing 150 μl of Luria broth media for 18 to 20 h at 30°C on a microplate shaker. Next, 20 μl of the Luria broth-grown cells were inoculated into a 300-μl, V-shaped 96-well microplate containing 180 μl of Terrific broth medium and grown for ∼3 h until the cell density reached O.D.600 = 1 to 1.5. δ-Aminolevulinic acid and isopropyl β-d-thiogalactoside were then added to the cells to final concentrations of 80 and 240 μg/ml, respectively. The cells were then grown for 48 h at 30°C on a microplate shaker. Cells were then harvested by centrifugation at 5000g for 10 min using an Allegra 25R microplate centrifuge (Beckman Coulter). The supernatant was removed by decanting, and the pellet was dried. The cells were resuspended in 150 μl of 0.1 M Hepes buffer, pH 7.4, which is termed whole cells.
Next, 50 μl of the substrate mixture (400 μM 7-BQ containing 4% methanol and 5 U/well polymyxin B sulfate, respectively) was incubated with 40 μl of whole cells for 5 min at room temperature. The background intensity was recorded at λex = 405 nm and λem = 510 nm using a fluorescence microplate reader (Fluoroskan Ascent; Thermo Electron Corporation, Waltham, MA). The reaction was initiated by the addition of 10 μl of H2O2 (10 mM final concentrations), and the formation of product was recorded at 2.5 min. Mutants with ≥2-fold higher activity than the average template activity were sequenced at the UTMB Protein Chemistry Laboratory and were further characterized as described below.
Expression and Purification of Wild-Type and Engineered Enzymes. CYP3A4 and mutants were expressed as His-tagged proteins in E. coli TOPP3 and purified using a nickel affinity column as described previously (Domanski et al., 2001). Protein concentrations were determined using the Bradford protein assay kit (Bio-Rad, Hercules, CA). The specific contents of wild-type and most random mutants were between 10 and 15 nmol of P450 per mg of protein. Upon purification, the yield, purity, and stability of the random mutant proteins were either similar to or higher than those of the wild-type enzyme. T309V, T309A/L216W, and F228I/F241V showed low P450 expression (<40 nM), high P420 (>75%), and low specific content (<4.0 nmol of P450 per mg of protein). L216W/T309V and F228I/T309V showed very low P450 expression (>10 nM) and were not purified for further study.
Enzyme Assays. H2O2- and CuOOH-supported enzyme activities were assayed as described previously with slight modifications (Kumar et al., 2005a,b). Substrate mixtures were prepared in 100 mM Hepes buffer, pH 7.4, with 2% as the final concentration of methanol (100 μl/well of a 96-well microplate). The substrate mixture was preincubated with 25 or 50 pmol of purified P450 enzyme at room temperature for 5 min. Reactions were initiated by the addition of H2O2 (10 mM) or CuOOH (2.5 mM). For steady-state kinetics with the peroxide, different concentrations of H2O2 (1–25 mM) or CuOOH (0.05–2.5 mM) were used at 200 μM 7-BQ. After 3 min of incubation, the reactions were stopped by adding 340 units (50 μl) of catalase. Subsequently, 50 μl of 100 mM Hepes buffer, pH 7.4, was added before recording the fluorescence intensity at λex = 405 nm and λem = 510 nm using an Ascent fluorescence plate reader (Thermo Electron). For 7-BFC O-deethylation and testosterone 6β-hydroxylation, 75 μM and 200 μM concentrations of the substrates, respectively, were incubated with the enzyme systems before initiating the reactions using appropriate concentrations of CuOOH as described above. Testosterone hydroxylation was assayed by thin-layer chromatography using radiolabeled substrate (Kumar et al., 2005a), whereas 7-BFC oxidation was assayed using a fluorescence method (Domanski et al., 2001). To determine the effect of b5, a 1:1 molar ratio of P450 and b5 at 0.25 μM P450 was incubated for 10 min before incubating with substrates.
The standard NADPH-dependent assays with 7-BQ, 7-BFC, and testosterone were essentially carried out as described previously (Domanski et al., 2001; Kumar et al., 2003). The reconstitution system contained CYP3A4, CPR, and b5 at a ratio of 1:4:2 and included 10 μg of dioleoyl phosphatidylcholine/100-μl reaction volume.
Steady-state kinetic parameters were determined by regression analysis using SigmaPlot (SPSS, Inc., Chicago, IL). The Km value was determined using the Michaelis-Menten equation, whereas the S50 and n values were determined using the Hill equation. Each kinetic experiment included CYP3A4 and mutants (presented in Tables 1, 2, 3, 4, 5, 6) simultaneously, using a high-throughput microplate assay for more accurate comparison of the data. The standard errors for fit to Michaelis-Menten or Hill equations were large when the Km, HOOH or S50, 7-BQ was higher than the maximum concentration of the reactants used.
P450 Heme-Depletion Assay. Determination of the kinetics of CYP3A4 heme depletion in the presence of CuOOH was done as described previously for H2O2 (Kumar et al., 2005b). In brief, the reaction was carried out at 25°C in 100 mM Hepes buffer, pH 7.4, in a 1-ml semi-microspectrophotometric cell with constant stirring. The reaction mixture contained 1 μM P450 and different concentrations of CuOOH (0.01–15 mM). Bleaching of the hemoprotein was followed by measuring a series of absorbance spectra using a Shimadzu-2600 spectrophotometer in the 340- to 700-nm range (Shimadzu, Kyoto, Japan). Fitting of the titration curves was performed by regression analysis using SigmaPlot (SPSS, Inc.). A simple pseudo-first order equation was used to determine the kinact values at different CuOOH concentrations, and the Michaelis-Menten equation was used to determine the KI, CuOOH value.
Results
Directed Evolution of CYP3A4. The most critical step in directed evolution is to find a simple, economical, and high-throughput activity screen. Recently, we have developed a facile assay method to circumvent steps involving CPR and b5 by using the alternate oxygen donor H2O2 for CYP3A4 (Kumar et al., 2005b). Because H2O2-supported activity with 7-BQ is very low in the wild type, several existing CYP3A4 mutants were examined. Only L211F/D214E and L373F showed 1.5-fold higher activity than CYP3A4. However, L211F/D214E showed abolished enzyme cooperativity and L373F showed low expression and stability (data not shown). Therefore, CYP3A4 was used as the template for directed evolution. The activity screen was optimized using whole cell suspensions as described earlier for P450 2B1 (Kumar et al., 2005a). A concentration of 10 mM H2O2 was used for screening mutants using whole cell suspensions because, above this concentration, the reaction is not linear up to 2.5 min (data not shown). A representative assay of multiple colonies of CYP3A4 indicated ∼70% colony-to-colony variation in activity (Fig. 2, open triangles) compared with the average CYP3A4 colony. The control showed very low relative intensity (Fig. 2, open diamonds). These data suggested that random clones having ≥2-fold higher activity than CYP3A4 would correspond to mutants with enhanced catalytic activity.
Representative data from an initial screen of several thousand random clones derived from directed evolution of CYP3A4 are shown in Fig. 2B, with the clones graphed in order of decreasing activity. Upon random mutagenesis, we selected four clones that showed ≥2.5-fold higher activity than the average of CYP3A4. DNA sequencing revealed a single new mutation in each case, L216W, F228I, F241V, and T433S. The mutants were then expressed in large-scale culture and purified. These single mutants either showed higher or similar expression levels and very low P420 compared with CYP3A4, suggesting that the random mutants that are selected using activity screen maintain stability unlike many site-directed mutants (Kumar et al., 2003).
Steady-State Kinetics at Varying 7-BQ Concentrations. To test whether the selected purified mutants also showed enhanced kcat, steady-state kinetic analysis of 7-BQ oxidation at 10 mM H2O2 and at varying 7-BQ concentrations was carried out. Above 10 mM, H2O2 accelerates heme depletion, and the reaction is not linear up to 5 min (Kumar et al., 2005b). L216W, F228I, and T433S showed 60 to 80% higher kcat values than did CYP3A4 (Table 1). However, F241V showed only a 20% higher kcat than did CYP3A4. In addition, F228I showed a significant increase in the S50, 7-BQ and decrease in the n value. Subsequently, to test whether the increased kcat in the H2O2-supported assay is also associated with an increased kcat in the NADPH-supported reaction, the mutants were assayed in the standard reconstituted system using NADPH. All the mutants showed 10 to 40% lower kcat than did CYP3A4 with a significant change in the S50 and n values (Table 1). In this assay, the accuracy of the S50 is limited by the highest 7-BQ concentration used. Overall, the results suggested that our screening method is limited to identification of CYP3A4 mutants with enhanced activity in the H2O2-supported system.
Steady-State Kinetics at Varying H2O2 Concentrations. Subsequently, we sought to determine whether the screening method is targeted to increased kcat and/or decreased Km, HOOH. Therefore, steady-state kinetic analysis at 200 μM 7-BQ and at varying H2O2 concentrations was carried out. At 20 mM H2O2, the rate of 7-BQ debenzylation is linear for up to 3 min (data not shown). CYP3A4 showed a very high Km, HOOH (61 mM), which is much higher than the H2O2 concentration used for the activity screening (10 mM) (Table 2). None of the mutants showed improvements in kcat, whereas L216W and F228I showed an ∼3-fold decrease in the Km, HOOH values, leading to >2.5-fold enhancement in kcat/Km, HOOH compared with CYP3A4. Compared with CYP3A4, F241V and T433S showed a modest increase in the kcat/Km, HOOH values as the result of decreased Km, HOOH (Table 2). The results suggested that the screening method is selective for mutants with decreased Km, HOOH. Furthermore, to test the hypothesis that the combination of single mutants would show a further enhancement in kcat/Km, HOOH, several multiple mutants were constructed. Unexpectedly, these multiple mutants did not show further improvement (Table 2). F228I/T433S and F241V/F228I showed similar kcat/Km, HOOH values to F228I, whereas most multiple mutants showed decreased kcat/Km, HOOH, mainly as the result of decreased kcat.
Steady-State Kinetics at Varying CuOOH Concentrations. Recently, T309V has been shown to display higher activity than CYP2D6 wild-type in a CuOOH-supported reaction (Keizers et al., 2005). Therefore, the existing CYP3A4 mutants T309A, T309I, and T309F were assayed for CuOOH-supported 7-BQ debenzylation. At 200 μM 7-BQ and 2.5 mM CuOOH, T309A showed a >3-fold higher activity, whereas T309I and T309F demonstrated lower activity than CYP3A4. Subsequently, T309V was constructed by site-directed mutagenesis and showed a >5-fold higher activity than CYP3A4. Therefore, steady-state kinetic analysis at 200 μM 7-BQ concentration and at varying CuOOH concentrations was carried out with T309A and T309V. At 2.5 mM CuOOH, the rate of 7-BQ debenzylation is linear for up to 5 min (data not shown). T309A and T309V exhibited a 4- and 6-fold higher kcat/Km, CuOOH, respectively, than CYP3A4 in the CuOOH-supported reaction, mainly as the result of increased kcat values (Table 3). When steady-state kinetic analysis was performed in the H2O2-supported reaction, T309V exhibited a 2-fold higher kcat/Km, HOOH than did CYP3A4, whereas T309A showed lower kcat/Km, HOOH than did CYP3A4 (data not shown). In addition, T309A and T309V showed lower kcat values in H2O2-supported than CuOOH-supported reactions. Furthermore, to test whether T309A and T309V in CYP3A4 follow a trend similar to T309V in CYP2D6, the standard NADPH-supported oxidation of 7-BQ was performed. T309A and T309V showed lower NADPH-supported activity than CYP3A4, as was found in the case of CYP2D6 (data not shown).
Subsequently, we tested individual single random mutants and multiple mutants for CuOOH-supported activity (Table 3). L216W and F228I respectively exhibited >3- and >4.5-fold higher kcat/ Km, CuOOH than did CYP3A4, whereas F241V and T433S exhibited ∼2-fold higher kcat/Km, CuOOH than did CYP3A4. The results indicated that the random mutants that were isolated for decreased Km, HOOH also showed enhanced utilization of CuOOH (kcat/Km, CuOOH), mainly as the result of increased kcat values compared with CYP3A4. As seen earlier with the H2O2-supported reaction, combination of these single random mutants did not improve the kcat/Km for CuOOH further. Therefore, T309A was added to the individual single random mutants to create several multiple mutants, and their activity was assayed in the CuOOH-supported reaction. Among the mutants with reasonable expression (see Materials and Methods), F228I/T309A showed the highest kcat/Km for CuOOH, which was 2-fold higher than that of T309A and 11-fold higher than that of CYP3A4 (Table 3). Interestingly, the Km for CuOOH was lower by more than an order of magnitude than the Km for H2O2 (∼1.0 mM versus 20–60 mM).
Effect of Cytochromeb5 on CuOOH-Supported Oxidation of 7-BQ. To investigate whether b5 also stimulates CuOOH-supported 7-BQ oxidation by CYP3A4, as shown earlier for H2O2 (Kumar et al., 2005b), steady-state kinetic analysis was performed at varying CuOOH concentrations and at 200 μM 7-BQ. Cytochrome b5 stimulated kcat/Km, CuOOH by 7-fold in CYP3A4, and by >5-fold in random mutants (Table 3 versus Table 4). However, the site-directed mutants T309A and T309V showed relatively less stimulation by b5. In F228I/T309A, b5 stimulated kcat/Km, CuOOH by 3-fold. Interestingly, b5 stimulated CuOOH-supported kcat/Km, CuOOH by both increasing the kcat and decreasing the Km, CuOOH values (Table 3 versus Table 4).
T309A and T309V Exhibit Enhanced P450 3A4 Activity with 7-BFC. The CYP3A4 random mutants were tested with 7-BFC to investigate whether the enhanced CuOOH-dependent activity with 7-BQ also was retained with a structurally distinct substrate. Whereas T309A and T309V showed significantly increased activity with 7-BFC (at 75 μM) by 1.5- and 3.5-fold, respectively, the random mutants showed lower activity than did CYP3A4 (data not shown). Therefore, steady-state kinetic analysis of 7-BFC O-deethylation with CYP3A4, T309A, and T309V was performed. The results are presented in Table 5. It is interesting that the Km, CuOOH for 7-BFC O-deethylation was 5-fold lower than that for 7-BQ debenzylation. T309V showed ∼2.5-fold higher kcat/Km, CuOOH for 7-BFC oxidation than did CYP3A4. Furthermore, b5 stimulated kcat/Km, CuOOH for the oxidation of 7-BFC by >7-fold for CYP3A4, whereas it stimulated kcat/Km, CuOOH by 5- and 3-fold for T309A and T309V, respectively (Table 5). The b5-mediated enhancement of kcat/Km, CuOOH with 7-BFC was accomplished as the result of increased kcat and decreased Km, CuOOH. As seen earlier with 7-BQ, T309A and T309V also showed decreased kcat in the NADPH-supported reaction with 7-BFC compared with CYP3A4 (data not shown).
T433S Exhibits Enhanced Activity with Testosterone. The CYP3A4 random mutants were also assayed with another structurally distinct substrate, testosterone. CYP3A4 yielded 6β-hydroxytestosterone as the major (92%) and 2β-hydroxytestosterone as a minor metabolite. The metabolite profile was unchanged in random mutants and T309A, whereas T309V yielded ∼30% 2β-hydroxytestosterone. Although most of the mutants showed decreased testosterone 6β-hydroxylation, T433S showed a >1.5-fold higher activity at 200 μM substrate than did CYP3A4. Therefore, steady-state kinetic analysis of testosterone 6β-hydroxylation by CYP3A4 and T433S was performed, and the results are presented in Table 6. It is interesting to note that the Km, CuOOH for testosterone hydroxylation was >3-fold lower than that for 7-BQ debenzylation. T433S showed a >3-fold higher kcat/Km, CuOOH than did CYP3A4 for testosterone 6β-hydroxylation as the result of increased kcat and decreased Km, CuOOH. Cytochrome b5 further stimulated kcat/Km, CuOOH by >5-fold by increasing the kcat and decreasing the Km, CuOOH values. In the NADPH system T433S showed a >2-fold higher kcat than did CYP3A4 with unaltered S50, 7-BQ and n values (Table 6). Other mutants, however, showed either similar or lower kcat than did CYP3A4 (data not shown). The results suggested that Thr433→ Ser substitution is unique in that it enhances CuOOH- and NADPH-supported reactions with testosterone, but not with 7-BQ or 7-BFC.
Synthesis of 6β-OH Testosterone by T433S in the Presence of Cytochromeb5. The decrease in the Km, CuOOH with testosterone T433S and upon b5 addition is very important, because CuOOH also depletes the heme in a time- and concentration-dependent manner. Therefore, it was critical to find the optimal CuOOH concentrations at which T433S could convert testosterone into 6β-OH testosterone. CuOOH concentration-dependent heme depletion was examined at different time intervals (1–60 min), and the rate constants for P450 inactivation were determined (Fig. 3). Subsequently, a re-plot of kinact and CuOOH concentrations using the Michaelis-Menten equation was carried out, which gave a KI, CuOOH of 6.5 mM (Fig. 3). The KI, CuOOH for heme depletion is >10-fold higher than the Km, CuOOH for testosterone hydroxylation. Therefore, lower concentrations (0.1–0.25 mM) of CuOOH should be suitable for longer incubations to achieve nearly complete product formation.
A time course experiment with CYP3A4 and T433S in the absence and presence of b5 containing 50 μM testosterone, 0.25 μM CYP3A4 and b5, and 0.2 mM CuOOH was carried out (Fig. 4). The results showed that T433S in the presence of b5 converts ∼75% of testosterone to 6β-OH testosterone in 2 h. Conversely, CYP3A4 in the absence of b5 converts only ∼5% to 6β-OH testosterone in 2 h. Furthermore, to test whether the percentage formation of 6β-OH testosterone was reduced when the reaction volume was increased, 0.1-, 1.0-, and 10-ml reactions were carried out in a container with a large surface area, and with occasional mixing. The results showed that there was only a modest decrease in the percentage conversion when the reaction volume was increased by 100-fold (data not shown).
Discussion
As a complement to the standard reconstituted system, which requires CPR, b5, and the expensive cofactor, NADPH, we engineered CYP3A4 using random and site-directed mutagenesis for peroxide-supported activity. The engineered enzymes can use CuOOH more efficiently than CYP3A4 wild-type to metabolize several substrates such as 7-BQ, 7-BFC, and testosterone. Random mutagenesis yielded mutants (L216W, F228I, F241V, and T433S) outside of the active site, whereas by site-directed mutagenesis, active site mutants T309A and T309V were produced (Fig. 5). This is the first report, to our knowledge, on the engineering of CYP3A4 for enhanced utilization of an alternate oxygen donor. Although combination of single random mutants did not enhance kcat/Km, CuOOH further, combination of a random and a site-directed mutant yielded F228I/T309A, which displays an 11-fold higher kcat/Km, CuOOH than does CYP3A4 for 7-BQ debenzylation. Whereas directed evolution yielded mutants without a significant effect on P450 expression or stability, site-directed mutants showed decreased P450 expression and stability. Therefore, one advantage of directed evolution from the standpoint of generating novel catalysts for industrial applications is that only mutants with good expression levels and stability are isolated in the whole-cell screens (Kumar and Halpert, 2005; Kumar et al., 2005a). In contrast, site-directed mutagenesis often yields mutants with decreased expression and/or stability, especially when multiple substitutions are combined (Kumar et al., 2003; Kumar and Halpert, 2005). Thus, directed evolution allows us to screen for mutants with enhanced activity that retain expression and stability.
The general mechanism by which peroxide supports P450-mediated substrate oxidation remains unclear, because significant activity requires an appropriate cytochrome P450, substrate, and peroxide. An X-ray crystal structure of CYP3A4 suggests that Arg-212 may provide general base catalysis of peroxide cleavage (Yano et al., 2004). Therefore, replacement by a nonbasic residue may decrease or even eliminate such activity. To explore this hypothesis we probed H2O2-supported oxidation of 7-BQ by R212A. However, there was no significant effect of this replacement on kcat (data not shown). Based on the crystal structure of P450eryF T252A, the conserved threonine is thought to be involved in proton delivery to the oxygen species (Raag et al., 1991). It has been hypothesized that P450 can deploy three different oxygenating species (peroxo, hydroperoxo, and oxenoid-iron), and they prefer certain types of reactions (Keizers et al., 2005). Furthermore, based on enhanced CuOOH-supported and decreased NADPH-supported activity by T309A in CYP2D6, it was suggested that the mutant predominantly forms the peroxo and/or hydroperoxo complexes as opposed to oxenoid-iron complex, which is mainly preferred by the NADPH-dependent reaction in the wild type. Similarly, the finding that T309V and T309A in CYP3A4 exhibited increased CuOOH- and decreased NADPH-supported activity compared with CYP3A4 is consistent with the hypothesis that the mutants predominantly form peroxo and/or hydroperoxo complexes. In contrast, the corresponding T302A in CYP2B4 and T303A in CYP2E1 showed decreased peroxide-dependent activity with one substrate but increased activity with another substrate (Vaz et al., 1996, 1998), which is consistent with our observations that CYP3A4 T309V showed decreased CuOOH-supported activity with testosterone, but not with 7-BQ or 7-BFC.
Another important observation in this study suggests that the Km, CuOOH is dependent on the substrate used. In the absence of b5, the Km, CuOOH is 1.2 mM, 0.22 mM, and 0.38 mM for the oxidation of 7-BQ, 7-BFC, and testosterone, respectively. The Km, CuOOH for P450 is usually associated with the accessibility of CuOOH to the active site heme pocket (Zhang and Pernecky, 1999). Removal of the NH2-terminal region in P450 2B4 caused a 5-fold decrease in Km, CuOOH for N-methylaniline demethylation, suggesting that the N-terminal region might mask the accessibility of CuOOH in the active site (Zhang and Pernecky, 1999). However, the mechanism by which substrates alter CuOOH accessibility is rather unclear. More recently, our laboratory has proposed that binding of a second molecule of substrate that shows homotropic cooperativity, such as 1-pyrenebutanol, induces a conformational transition in CYP3A4 that appears to close the heme pocket (Davydov et al., 2005; D. Davydov, unpublished observations). This phenomenon does not appear to occur when only one molecule of bromocriptine binds to CYP3A4. Therefore, we speculate that 7-BQ, which shows a higher extent of cooperativity than does 7-BFC or testosterone (Harlow and Halpert, 1998; Domanski et al., 2001) (Table 1), masks the active site for the access of CuOOH leading to an increased Km, CuOOH.
Our results from the peroxide-supported reaction suggest that L216W and F228I play a major role in enhancing catalytic efficiency of 7-BQ debenzylation. X-ray crystal structures (Fig. 5) and extensive site-directed mutagenesis do not predict a role of these nonactive site residues in substrate binding and/or metabolism (Domanski and Halpert, 2001; Hutzler and Tracy, 2002; Ekins et al., 2003; Tanaka et al., 2004; Williams et al., 2004; Yano et al., 2004; Park et al., 2005; Scott and Halpert, 2005). However, recent molecular dynamics simulations with the newly developed force field parameters for the heme-thiolate group and its dioxygen adduct show differences in structural and dynamic properties between CYP3A4 in the resting form and its complexes with the substrate progesterone and the inhibitor metyrapone (Park et al., 2005). The results indicated that the broad substrate specificity of CYP3A4 stems from the malleability of a loop (residues 211–218) that resides in the vicinity of the channel connecting the active site and bulk solvent. Interestingly, L216W (F-helix) and F228I (F-G loop) mutations are located in or close to this region, which may cause a conformational transition upon 7-BQ binding leading to enhanced peroxide-supported reactions. Phe-241 is located in the G-helix, and its role in substrate binding or metabolism is not clear. Thr-433 is located near the most conserved region, Cys-442, which ligates the heme iron as the fifth ligand.
Although the role of b5 as a source of electrons for P450 is well known (Schenkman and Jansson, 2003), increasing evidence points to a modulatory effect of b5 mediated by a conformational transition in P450 (Reed and Hollenberg, 2003; Yamaori et al., 2003; Yamaguchi et al., 2004). Recently, we have shown that b5 stimulates H2O2-supported 7-BQ debenzylation by CYP3A4 (Kumar et al., 2005b). In addition, an independent finding has shown that b5 directly induces positive cooperativity and enhances catalytic turnover of CYP3A4 with multiple substrates (Jushchyshyn et al., 2005). In the present study, an ∼3-fold increase in the kcat and ∼2-fold decrease in the Km, CuOOH for the oxidation of 7-BQ, 7-BFC, and testosterone further suggests that b5 plays a major role in modulating CYP3A4 activity. Earlier, it was proposed that b5 induces some reorganization in the heme moiety and substrate binding pocket of P450 2B4 that favors the binding of peroxide and benzphetamine (Aplentalina and Davydov, 1997). Therefore, we suggest that b5 reorganizes the CYP3A4 active site heme pocket, which leads to a decreased Km, CuOOH by favoring the binding of CuOOH. Because the random mutants reside outside of the active and b5 binding sites (Bridges et al., 1998; Schenkman and Jansson, 2003; Williams et al., 2004; Yano et al., 2004), these mutations do not primarily alter the b5-P450 interaction. In contrast, a reduced b5-stimulated P450 activity in T309V is similar to earlier observations that I301F and A305F, which are closest to the heme (Williams et al., 2004; Yano et al., 2004), show impaired b5-P450 interaction (Kumar et al., 2005b). To further test the hypothesis that the b5-P450 interaction is impaired in T309V, but not in random mutants, the Kd values for b5 were determined in CYP3A4, T309V, and F228I at 0.25 μM protein concentration as described previously (Kumar et al., 2005b). The Kd for b5 in T309V (0.30 μM) was increased by 6-fold compared with CYP3A4 (0.05 μM), whereas the Kd for b5 in F228I (0.04 μM) was similar to that of CYP3A4 (data not shown).
An interest in producing large quantities of P450-mediated drug metabolites prompted us to find a simple and cost-effective way to synthesize such compounds. In a recent study, microsomal CYP3A4 from human liver or recombinant sources and a NADPH regenerating system have been used to produce ∼50 mg of 6β-OH testosterone with 70% conversion to product (Vail et al., 2005). Consistent with the above studies, our data predict that in a 1-liter reaction with CuOOH, T433S in the presence of b5 can produce ∼10 mg of 6β-OH testosterone with 60% conversion to product. Although the yield using CuOOH is lower than the standard method, it is more cost-effective. More recently, CYP2D6 and CYP3A4 have been identified and further optimized to use a selected peroxide donor such as CuOOH with improved activity compared with the standard NADPH reconstitution system (Chefson et al., 2006).
In summary, the directed evolution approach created CYP3A4 mutants with enhanced utilization of peroxide, whereas site-directed mutants T309A and T309V specifically showed enhanced CuOOH-supported activity with 7-BQ and 7-BFC. Combination of the site-directed and random mutants created the most efficient enzyme, F228I/T309A, for peroxide-mediated oxidation of several substrates. Addition of b5 further stimulated the kcat/Km for CuOOH-supported CYP3A4 activity, and the CYP3A4 random mutant T433S showed enhanced CuOOH- and NADPH-supported activity with testosterone. The mutations found in this study may be useful for re-engineering other related mammalian P450s for enhanced utilization of peroxide. In addition, this approach may be used to engineer xenobiotic-metabolizing P450 enzymes for facile synthesis of milligram quantities of specific drug metabolites, agrochemicals, and food ingredients, and for bioremediation.
Acknowledgments
We thank Drs. Frances H. Arnold and Edgardo Farinas from California Institute of Technology (Pasadena, CA) for providing technical details of directed evolution and sharing their unpublished materials. We thank Dr. Dmitri Davydov (UTMB) forexpert suggestions, Dr. B. K. Muralidhara (UTMB) for generating Fig. 5, and You-Qun He (UTMB) for technical support in optimizing error-prone PCR. We also thank Dr. Richard Hodge, Synthetic Organic Chemistry Core Laboratory, UTMB, for synthesizing 7-BQ, and Dr. Kenneth Johnson, Pharmacology and Toxicology, UTMB, for allowing us to use the fluorescence plate reader (Fluoroskan Ascent).
Footnotes
-
This work was supported by National Institutes of Health Grants GM54995 and Center Grant ES06676. This work was partially presented at Experimental Biology 2006, April 1–5, 2006, San Francisco, CA, and the 5th SW P450 meeting, May 8–10, 2006, Navasota, TX.
-
Article, publication date, and citation information can be found at http://dmd.aspetjournals.org.
-
doi:10.1124/dmd.106.012054.
-
ABBREVIATIONS: P450, cytochrome P450; PCR, polymerase chain reaction; 7-BQ, 7-benzyloxyquinoline; 7-BFC, 7-benzyloxy-4-(trifluoromethyl)coumarin; CPR, NADPH-cytochrome P450 reductase; b5, cytochrome b5; HOOH, hydrogen peroxide; CuOOH, cumene hydroperoxide; OH, hydroxy; UTMB, University of Texas Medical Branch.
- Received July 17, 2006.
- Accepted September 13, 2006.
- The American Society for Pharmacology and Experimental Therapeutics