PRIORITY PAPER Epoxidation of benzo[ a ]pyrene-7,8-dihydrodiol by human CYP1A1 in reconstituted membranes Effects of charge and nonbilayer phase propensity of the membrane Pyotr Kisselev 1, *, Dieter Schwarz 1 , Karl-Ludwig Platt 2 , Wolf-Hagen Schunck 3 and Ivar Roots 1 1 Institute of Clinical Pharmacology, University Medical Centrum Charite ´ , Humboldt University of Berlin, Germany; 2 Institute of Toxicology, University of Mainz, Germany; 3 Max Delbrueck Centrum for Molecular Medicine, Berlin, Germany Human cytochrome P 4501A1 (CYP1A1) is one of the key enzymes in the bioactivation of environmental pollutants such as benzo[a]pyrene (B[a]P) and other polycyclic aro- matic hydrocarbo ns. To ev aluate the effect of membrane properties a nd distinct phospholipids o n t he activity of human CYP1A1 purified insect cell-expressed human CYP1A1 and of human NADPH-P450 reductase were reconstituted into phospholipid vesicle membranes. Con- version rates of up to 36 pmolÆmin )1 Æpmol )1 CYP1A1 of the enantiomeric promutagens (–)- and (+)-trans-7,8-dihy- droxy-7,8-dihydro-B[a]P (7,8-diol) to the genotoxic diolep- oxides were achieved. The highest rates were obtained when negatively charged lipids such as phosphatidylserine and phosphatidylinositol and/or nonbilayer phospholipids such as phosphatidylethanolamine were present in the membrane together with neutral lipids. Both V max and K m values w ere changed. This suggests a rather complex mechanism of stimulation which might include altered substrate binding as well as more effective interaction between CYP1A1 and NADPH-P450 reductase. Furthermore, the ratio of r-7,t-8 - dihydroxy-t-9,10-epoxy-7,8,9,10-tetrahydro-B[a]P (DE2) to r-7,t-8-dihydroxy-c-9,10-epoxy-7,8,9,10-tetrahydro-B[a]P (DE1) f ormed f rom (–)-7,8-diol was significantly increased by the introduction of anionic lipids, but not by that of nonbilayer lipids. Thus, c harged lipids affect the stereose- lectivity of the epoxidation b y leading to the formation of a larger amount of the ultimate mutagen DE2 than of DE1, which is far less carcinogenic. These data suggest that membrane properties such as negative charge and nonbi- layer phase propensity are important for the efficiency and selectivity o f e nzymatic function of human CYP1A1. Keywords: human cytochrome P4501A1; vesicle r econstitu- tion; epoxidation; benzo[a]pyren e; benzo[ a]pyrene-7, 8-diol. Human cytochrome P 4501A1 (CYP1A1) is one of the key enzymes in the bioactivation of environmental pollutants. Benzo[a]pyrene (B[a]P) and other polycyclic aromatic hy- drocarbons acquire their mutagenic and carcinogenic prop- erties by its a ction. The first s tep of a ctivation catalyzed by CYP1A1 is the formation of 7R,8S-epoxy-7,8-dihydro- B[a]P. This is transformed via r egioselective hydrolysis by microsomal epoxide h ydrolase to (–)-7 R,8R-dihydroxy-7,8- dihydrobenzo[a]pyrene ((–)-7,8-diol) and then metabolized by CYP1A1 to the ultimately genotoxic r-7,t-8-dihydroxy-t- 9,10-epoxy-7,8,9,10-tetrahydro-B[a]P (so-c alled diolepox- ide-2 or antidiolepoxide, DE2) [1–3]. The last reaction appeared to be highly stereoselective as no or much less of the less carcinogenic r-7,t-8-dihydroxy-c-9,10-epoxy-7,8,9,10- tetrahydo-B[a]P (diolepoxide-1 or sy n-diolepoxid e, DE1) was p roduced from (–)-7,8-diol [4,5]. Racemic (+/–)-7,8-diol is mainly converted by human CYP1A1 to the DE2 [6,7]. CYP1A1-dependent activity can be reconstituted by mixing the basic components of the monooxygenase system, i.e. purified CYP1A1, NADPH-cytochrome P450 reductase, which transfers elect rons f rom NA DPH t o P 450, and dilaurylglycerophosphocholine (Lau 2 PtdCho) [8–10]. H ow- ever, this micellar system is not appropriate for studying the interactions between the components o f the system as some of its properties are unlike those of the endoplasmic reticulum m embrane, t he n atural enviro nment o f t he microsomal monooxygenase system. This membrane is a bilayer, and it is highly p robable that CYP1A1 and NADPH-cytochrome P450 reductase exhibit other impor- tant protein–lipid and protein–protein interactions there. Indeed, reconstitution systems using bilayer vesicles yielded higher rates of activity w ith rabbit liver CYP3A6 and CYP2B4 [11,12] and with human CYP3A4 [13] t han micellar Correspondence to D. Schwarz, Charite ´ , Humboldt University of Berlin, c/o M ax-Delbrueck-Centrum, Robert Roessle Str. 10, D-13125 Berlin, Germany. Fax: + 4 9 30 9406 3329, Tel.: + 49 3 0 9406 3711, E-mail: schwarz@mdc-berlin.de Abbreviations: P450, human cytochrome P4501A1 (CYP1A1); Lau 2 PtdCho, dilaurylglycerophosphocholine; Ole 2 PtdCho, diol- eoylglycerophosphocholine; Ole 2 PtdPEtn, dioleoylglycerophospho- ethanolamine; PtdCho, phosphatidylcholine; PtdEtn, phosphatidylethanolamine; Ela 2 PtdEtn, dielaidoylglycerophospho- ethanolamine; PtdSer, phosphatidylserine; PtdIns, phosphatidylinos- itol; P A, phosphatidic acid; B[a]P, benzo[a]pyrene; DE2, diolepoxide 2 (r-7,t-8-dihydroxy-t-9,10-epoxy-7,8,9,10-tetrahydro-B[a]P); DE1, diolepoxide 1 (r-7,t-8-dihydroxy-c-9,10-epoxy-7,8,9,10-tetrahydro- B[a]P); (+/–)-7,8-diol, (+/–)-trans-7,8-dihydroxy-7,8-dihydro-B[a]P]. *Present address: Institute of Bioorganic Chemistry, Academy of Sciences of Belarus, Minsk, Belarus. (Received 21 December 2 001, revised 14 February 2002, a ccepted 20 February 2 002) Eur. J. Biochem. 269, 1799–1805 (2002) Ó FEBS 2002 doi:10.1046/j.1432-1033.2002.02848.x Lau 2 PtdCho sys tems. There is s ome evidence that the membrane charge is an important determinant for P450- dependent activity [14–16]. Furthermore, the presence of specific lipids of the nonbilayer class was found to be essential for optimal activity of microsomal and mitochondrial P450s. For instance, rabbit liver CYP2B4 requires phosphatidy- lethanolamine ( PtdEtn) [ 12], and bovine a drenal CYP11A1 requires cardiolipin, PtdEtn or branched phosphatidylcho- line (PtdCho) [17–19] for optimal activity as well as for membrane reconstitution. Reconstitution of purified human CYP1A1 using phospholipid vesicles with CYP1A1 and P450-reductase incorporated into the membrane has not been reported to our knowledge. There are no data available which demonstrate how lipids influence the ster eo- and/or regioselectivity of a P 450-catalyzed r eaction. We investigated whether human CYP1A1 could be reconstituted into vesicle membranes effectively enough to study the e ffects of mem- brane properties and/or distinct phospholipid s on C YP1A1 activity. We c haracterized the i mpact o f negatively c harged and nonbilayer lipids on CYP1A1-dependent stereoselective epoxidation by using both (optically pure) enantiomeric promutagens (+)-7S,8S-dihydroxy-7,8-dihydrobenzo[a]py- rene ((+)-7,8-diol) and (–)-7 R,8R-dihydroxy-7,8-dihydro- benzo[a]pyrene ((–)-7,8-diol), as substrates. EXPERIMENTAL PROCEDURES Materials Ole 2 PtdCho, Ole 2 PtdEtn, E la 2 PtdEtn, a nd PtdSer were bought from Avanti Polar Lipids (Alabaster, AL, U SA). The mixture PtdCho/PtdEtn/PA (2 : 1 : 0.06, w/w/w) and PtdIns were bought from Lipid Products (Redhill, Surrey, UK), and the [ 14 C]Ole 2 PtdCho was from Amersham Pharmacia Biotech (Freiburg, Germany). For the preparation of (+)-7,8-diol and (–)-7,8-diol, racemic 7,8-diol was synthesized [20] and chromatographi- cally separated into the enantiomers using a chiral stationary phase [21]. Structure, assignment of absolute configuration and optical purity of t he two enantiomers were confirmed b y UV, CD a nd measurements of the rotation. UV sp ectra were recorded with a Shimadzu (Japan) UV 2401 PC, CD spectra were recorded with a Jasco J-720, and specific rotations with a Perkin-Elmer 241-MC automatic polarimeter. The specific rotations [a 20 D in acetone were )395 ° and +417 ° for (– )- 7,8-diol and (+)-7,8-diol, respectively. These are nearly identical with the data reported for these c ompounds when optically pure [ 22]. To prove the absolute configuration C D spectra were recorded f or both enantiomers. As expected, the CD curves of the enantiomeric pair were almost mirror images o f each other (not shown). The tetraols r-7,t-8,t-9, c-10-tetrahydroxy-7,8,9,10-tetra- hydro-B[a]P (RTTC), r-7,t-8,t-9,t-10-tetrahydroxy-7,8,9, 10-tetrahydro-B[a]P (RTTT), r-7,t-8, c-9,t-10-tetrahydroxy- 7,8,9,10-tetrahydro-B[a]P (RTCT), r-7,t-8,c-9,c-10- tetra- hydroxy-7,8,9,10-tetrahydro-B[a]P (RTCC) were obtained from NCI Chemical C arcinogen Repository, Midwest Research Institute, Kansas City, MI, USA. Preparation of enzymes and lipid vesicles Human CYP1A1 was heterologously expressed as C-terminal 6xHis-fusion protein in Spodoptera frugiperda insect cells using baculovirus [23]. P urification was performed with nickel-chelate chromatography, essential- ly as described for human CYP2D6 [24] but with the modification that a mixture of emulgen 913 (2%) and Na-cholate (0.2%) was used for solubilization. CYP1A1 was electrophoretically homogeneous and had a specific P450contentof11nmolÆmg )1 protein. Human NAD- PH cytochrome P450 reductase was purified from Spodoptera frugiperda insect cells as described previously [25]. Phospholipid vesicles were prepared essentially as described by Ingelman–Sundberg et al. [13] by cholate gel filtration. 5 mg of lipid or lipid mixture were d ried under nitrogen and resuspended in 1.25 mL of 50 m M Tris buffer, pH 7.5, containing 100 m M NaCl and 2% sodium cholate. 1 nmol P450 and 0.5 nmol human NADPH cytochrome P450 reductase were added to 250 lL o f t he suspension and incubated for 60 min at 4 °C (final cholate concentration: 1%). Cholate was removed by Sephadex G-50 gel filtration. The vesicular fractions were collected as void volume eluting from the column and were immediately used for the a ssays. The vesicular fractions were characterized in terms of P450, reductase, a nd p hospholipid content. The amount of CYP1A1 was determined by CO difference spectrometry using an extinction coefficient of 91 m M )1 Æcm )1 [26]. Rates of NADPH-cytochrome c r eduction by reductase were measured using an e xtinction coefficient o f 21 m M )1 Æcm )1 [27]. Lipid was quantitated by measuring the 14 C-radio- activity of the fractions by liquid scintillation counting using [ 14 C]-Ole 2 PtdCho as marker. The final vesicular preparations are characterized as follows: 1 l M CYP1A1, molar reductase/P450 stoichiometry of 0.9, a nd molar lipid/protein ratio of 1200. The intra- and interday degree of variation in relative P450, reductase, and lipid content did not exceed 10% in any of the vesicular preparations. Finally, standardized amounts of CYP1A1 were u sed for the e nzymatic assays, usually 5 or 10 pmol (vesicular) CYP1A1. Enzyme assays The epoxidation assays were perfo rmed as described for the racemic 7,8-diol [ 7] with the following modifications: incubations contained 50 m M Tris/HCl (pH 7.5), 100 m M NaCl with either (+)- or (–)-7,8-diol, and vesicles with 5 p mol CYP1A1 ( for (–)-7 ,8-diol) and 1 0 pmol CYP1A1 (for (+)-7,8-diol) in a final volume of 0.5 mL. Extraction and HPLC separation of the products were performed essentially as described earlier [7]. The rates of DE1 and DE2 f ormation were estimated from the accumulation of their hydrolysis products, the tetraols, as follows: RTCC + RTCT represent DE1 formation a nd RTT C + RTTT represent DE2 formation [28]. Kinetic constants were determined by nonlinear analysis of Michaelis–Menten kinetics using the computer program ENZFITTER (by J. R. L eatherbarrow, Elsevier-Biosoft). The data presented a re the m eans and s tandard d eviations of three separate experiments. Statist ical significance of results b etween lipid systems was analysed using one - way ANOVA software (GraphPad software, San Diego, CA, USA). 1800 P. Kisselev et al. (Eur. J. Biochem. 269) Ó FEBS 2002 RESULTS Effect of charged lipids on epoxidation rates We evaluated the kinetics of CYP1A1-dependent epoxida- tion of 7,8-diol in vesicles prepared from neutral Ole 2 Ptd- Cho and mixtures of Ole 2 PtdCho and anionic lipids (PtdSer, PtdIns, P A). PtdSer is the most frequently occur- ring negatively c harged microsomal lipid, whereas PtdIns and P A are minor constituents. The mixture P tdCho/ PtdEtn/PA with the two main lipid components of the microsomal membrane, PtdCho and PtdEtn, in a 2 : 1 ratio, and with a slightly negative charge i ntroduced by PA, roughly imitates the properties of the microsomal mem- brane [29]. This lipid mixture has already been used in studies of P450–lipid interaction ( e.g [12]). Figure 1A represents the effects of the phospholipids on the conversion of(+)-and(–)-7,8-dioltoDE1andDE2. Table 1 s ummarizes the results of kinetic a nalysis. Typical Lineweaver–Burk plots of the kinetic r ates of (–)-7,8-diol epoxidation to DE2 by human CYP1A1 are shown for selected lipid vesicle systems in Fig. 2. The results demonstrate a clear dependence on the phospholipid charge and the type of lipid used. For instance, V max for (–)-7,8-diol oxidation to D E2 was about 33 pmolÆmin )1 Æpmol )1 in vesicles containing PtdSer. This rate is very high for a CYP1A1-catalyzed reaction; it is more than 13 t imes higher than the r ate obtained with Ole 2 PtdCho, which i s a neutral lipid. The incorporation of PtdIns and PtdCho/PtdEtn/PA led to a sevenfold a nd twofold to threefold activation, respectively. Statistical a nalysis o f the data showed that th e differences mu st be considered very significant ( P <0.05) apart from the lipid system PtdCho/PtdEtn/PA (P >0.05). The activation of the CYP1A1-catalyzed oxidation of t he (+)-7,8-diol was less pronounced. Effects of nonbilayer lipids on the metabolic rates of epoxidation We analyzed the effects of a typical member of the nonbilayer c lass of phospholipids, namely PtdEtn [30]. Together with PtdCho, it belongs to the main components of the liver microsomal membrane [29]. Comparison of CYP1A1 act ivity in vesicles consist ing of Ole 2 PtdCho/ Ole 2 PtdEtn and Ole 2 PtdCho/Ela 2 PtdEtn revealed striking evidence for the importance of the hexagonal phase forming tendency of the membrane. The two di-18:1-acyl-PtdEtn, Ole 2 PtdEtn and E la 2 PtdEtn, differ only in their conforma- tion of the double bond, which is cis in Ole 2 PtdEtn and trans in Ela 2 PtdEtn, whereas both headgroup and chain length are identical. This difference r esults in a much higher bilayer-hexagonal phase transition temperature in Ela 2 Pt- dEtn (about 65 °C) than in Ole 2 PtdEtn (about 10 °C) and has been used to investigate the impact of the hexagonal phaseformingtendencybyYangandHwang[31].Datain Fig. 1A and in Table 1 show that the activity of CYP1A1 is significantly enhanced (P < 0.001, c onsidered e xtremely significant), e.g. about eightfold for the formation of the main product DE2 from (–)-7,8-diol in vesicles containing Ole 2 PtdEtn, whereas Ela 2 PtdEtn has almost no activation potential (P > 0.05, no significant activation). For the (+)- 7,8-diol metabolism t he stimulation b y the incorporation o f Ole 2 PtdEtn was less but also pronounced (four fold to fivefold) and statistically significant (P <0.001). These results clearly demonstrate a strong correlation between the activation of the CYP1A1-catalyzed epoxidation reaction of 7,8-diols a nd the enhanced nonbilayer phase propensi ty in membranes containing PtdEtn. Effect of lipids on the stereoselectivity of epoxidation The stereoselectivity o f the e poxidation reaction can be demonstrated by the ratio of the formation rates o f the two diol-epoxides, DE2 and DE1. These differ only in the conformation of the 9 ,10-epoxy-group, anti in DE2 and syn in DE1. The respective data in F ig. 1 B and Table 1 show Fig. 1. Effect of phospholipids on the total epoxidation (A) and the ratio of diolepoxide-2 t o diolepoxide-1 formation (B) of (–)- and (+)-7, 8-diol by human CYP1A1 in reconstituted vesicles. Vesicles consisting of pure Ole 2 PtdCho, or of a lipid mixture of Ole 2 PtdCho and the particular lipid in a weight ratio of 2 : 1 were prepared as d escribed under Experimental procedures. PtdCho/PtdEtn/PA is a lipid mixture of egg PtdCho, egg PtdEtn, and phosphatidic acid in a weight ratio of 2:1:0.06.RatesrepresentV max values determined by k inetic analysis based on data from three separate experiments. Ratios were calculated from these V max values. Ó FEBS 2002 Membrane-reconstituted human CYP1A1 (Eur. J. Biochem. 269) 1801 that charged lipids strongly affect the stereoselectivity of epoxidation. The presence of the anionic lipid PtdSer increased the formation of DE2 twice as much as that of DE1. Even the relatively small portion of the negatively charged PA in the PtdCho/PtdEtn/PA membrane led to a pronounced enhancement in the product ratio DE2/DE1. In both cases statistical analysis proved the enhancement in the ratio DE2/ DE1 to b e v ery s ignificant (P < 0.05). Actually, the inc orporation of PtdIns i nto Ole 2 PtdCho membrane also leaded to an increase but cannot be considered statistically significant higher compared to Ole 2 PtdCho (P > 0.05). By contrast, we found almost no influence of either Ole 2 PtdEtn or Ela 2 PtdEtn. The ratios determined for both systems are not significant different from that of Ole 2 PtdCho ( P > 0.05). Thus, t he results demonstrate that the charge of the membrane has an important influence on the ratio DE2/DE1, whereas the nonbilayer phase propensity has hardly any effect on the stereoselective formation of the DEs. DISCUSSION The results presented show that purified human CYP1A1 can be efficiently reconstituted i nto phospholipid vesicle membranes together with P450-reductase. T he main results can be summarized as follows: (a) the presence of negatively charged lipids in the membrane stimulates diol epoxidation by human CYP1A1 significantly, (b) negatively charged lipids affect the stereoselectivity of epoxidation by favouring the formation of DE2 to the detriment of DE1, and (c) nonbilayer lipids also lead to s trong activation probably by increasing the effective substr ate concentration in the membra ne. Thus, we found that in addition to the membrane charge, the nonbilayer phase propensity of the membrane is an important determinant for an effective reconstitution of CYP1A1-dependent epoxidation activity. The reason for the requirement of such a specific and complex membrane structure including charged and nonbilayer lipids for the maximum activity of CYP1A1 i s not known. It seems t hat the native f unction of human CYP1A1 requires a microso- mal membrane containing negatively charged as well as nonbilayer lipids. Kinetic analysis showed, that both kinetic parameters, V max and K m, were altered, as is clearly demonstrated by the Lineweaver–Burk plots of the rates o f (–)-7,8-diol epoxida- tion for th e main mutagenic product DE2 ( Fig. 2). With regard to dependence on membrane charge, this analysis supports the g eneral concept t hat the negative charge of the membrane not only improves t he electron transfer and interaction betw een reductase and P450 but also affects t he active site conformation of P450. This last conclusion is confirmed by the observation that charged lipids also strongly affect the stereoselectivity of the epoxidation reaction, whereas nonbilayer lipids do not. Considering all the data, the observed increase in the formation of DE2 is caused, a t least partially, by a lipid-induced conformational change. This change mediates more favourable active site spatial coordinates responsible for t he binding and produc- tive orientation between heme-bound oxygen and the acceptor 9,10-double bond of the (–)-7,8-diol. CYP3A4 is the only o ther human liver P450, with which similar h igh metabolic rates could be reached in a v esicle reconstitution system that includes charged lipids [13]. However, only V max was i ncreased whereas K m remained unchanged. This suggests that CYP3A4 activity is s timu- lated by a more effective interaction between P450 and reductase. Recently published data for rabbit liver CYP1A2 belonging to the same P450 subfamily also showed that anionic phospholipids (PA, PtdIns, PtdSer) present i n the membrane leaded to enhanced enzymatic activity. More- over, evidence by s tructural studies was p resented for considerable changes of the overall conformation of CYP1A2 coinciding with the increase of activity [32,33]. Table 1. Lipid dependence of 7,8-diol epoxidation by human CYP1A1: kinetic analysis for (–)-7,8-diol and (+)-7,8-diol. The rates of DE1 and DE2 formation were calculated from the accumulation of their hydrolysis products, the tetraols, as follows: RTCC + RTCT represent DE1 formation and RTTC + RTTT represent DE2 formation (28). Data are means ± SD of V max and K m values, determined by fitting experimental data from three s eparate experiments to Michaelis–Menten k inetics a s described under Experimental Procedures. DE2/DE1 d ata represent the ratio of the respective V max values. For preparation of l ipid vesicles see legend to Fig. 1. V max is in pmolÆmin )1 Æpmol P450 )1 , K m is in l M . Lipid mixture (w/w) DE2 DE1 Total rate (pmolÆmin )1 Æpmol P450 )1 ) V max K m V max K m DE2/DE1 (–)-7,8-diol Ole 2 PtdCho 2.5 ± 0.1 1.2 ± 0.2 0.6 ± 0.1 1.2 ± 0.3 4.2 3.1 Ole 2 PtdCho/PtdSer (2 : 1) 32.7 ± 6.3 25.1 ± 7.2 3.2 ± 0.4 17.3 ± 4.2 10.2 35.9 Ole 2 PtdCho/PtdIns (2 : 1) 18.0 ± 1.6 6.7 ± 1.4 2.3 ± 0.3 6.4 ± 1.5 7.8 20.3 PtdCho/PtdEtn/PA (2 : 1 : 0.06) 7.3 ± 0.7 4.9 ± 1.3 0.8 ± 0.1 6.0 ± 2.1 9.1 8.1 Ole 2 PtdCho/Ole 2 PtdEtn (2 : 1) 19.9 ± 2.2 9.7 ± 2.2 3.2 ± 0.3 7.5 ± 1.6 6.0 23.2 Ole 2 PtdCho/Ela 2 PtdEtn (2 : 1) 3.7 ± 0.1 1.6 ± 0.1 0.8 ± 0.1 2.7 ± 0.3 4.6 4.5 (+)-7,8-diol Ole 2 PtdCho 0.5 ± 0.1 11.4 ± 3.7 1.2 ± 0.1 7.4 ± 1.4 0.42 1.7 Ole 2 PtdCho/PtdSer (2 : 1) 0.9 ± 0.12 14.3 ± 3.2 2.8 ± 0.5 12.1 ± 3.7 0.32 3.7 Ole 2 PtdCho/PtdIns (2 : 1) 0.7 ± 0.04 5.7 ± 0.9 1.3 ± 0.08 3.7 ± 0.7 0.54 2.0 PtdCho/PtdEtn/PA (2 : 1 : 0.06) 0.4 ± 0.03 6.5 ± 0.9 0.7 ± 0.08 3.7 ± 1.2 0.57 1.1 Ole 2 PtdCho/Ole 2 PtdEtn (2 : 1) 2.2 ± 0.3 39.9 ± 8.0 6.0 ± 1.2 28.9 ± 8.7 0.37 8.2 Ole 2 PtdCho/Ela 2 PtdEtn (2 : 1) 0.7 ± 0.2 21.9 ± 7.5 1.2 ± 0.1 9.9 ± 1.9 0.58 1.9 1802 P. Kisselev et al. (Eur. J. Biochem. 269) Ó FEBS 2002 Note, the enzymatic activity of CYP1A2 was measured in the presence o f cumene h ydroperoxide in place of r eductase and NADPH proving the conclusion that the observed effects are indeed related to lipid-ind uced conformational changes of CYP1A2. So far we discussed only the effect of lipids on P450. However, it is also possible that lipids induce structural changes in the reductase which improve its interaction with P450 and t hereby enhance V max .Ina previous study, an increase in V max of CYP2B1-dependent O-dealkylation activity was ascribed to a PtdSer-induced conformational change of reductase [34]. We also observed a strong stimulation of C YP1A1 epoxidation activity by typical nonbilayer lipids such as PtdEtn, but the m echanism o f activation might be different from that discussed a bove for charged lipids. The striking difference in the activation capacities of Ole 2 PtdEtn and Ela 2 PtdEtn suggests that the hexagonal phase propensity i s probably the characteristic of the membrane which best explains these changes in activity. Nonbilayer lipids do not cause significant alterations in the metabolite profile, i.e. they do not alter the stereoselectivity of epoxidation. Obviously, there is no alteration of the active s ite confor- mation of P450. But the s triking parallelism of the curves in the Lineweaver–Burk plots with different lipid components (i.e. for Ole 2 PtdCho, Ole 2 PtdCho/Ela 2 PtdEtn, and Ole 2 Ptd- Cho/Ole 2 PtdEtn) i ndicates that K m and V max were c hanged by the same facto r. The most p robable e xplanation is an increase in the effective substrate concentration which is probably due to the redistribution of the substrate b etween the aqueous a nd the m embrane phase and i s brought about by a change in the nonbilayer p hase propensity of the membrane. In accordance with this hypothesis, it is now generally assumed that microsomal P450s apart f rom their N-terminal transmembrane domain have a dditional attach- ment region(s) associating t he protein partially buried into the membrane [35]. Th is would favour lipophilic substrates by placing the opening of the substrate access channel into the lipid bilayer. Thus the pool of substrate molecules accessible for P450 would c onsist of the substrate molecules in the membrane phase. The activity of s everal membrane enzymes, among which are protein kinase C, mitochondrial reductases and ATPases, mitochondrial cytochrome P450 (CYP11A1), and others [18,36,37], i s i ncreased by nonbilayer lipids. There i s s ome e vidence t hat t he latter influence the conformation of membrane-bound proteins by changing membrane properties, e.g. introduce curvature stress, that might mediate an optimal conformation of the protein [37–39]. Here, we propose an additional mechanism of action for nonbilayer lipids. The nonbilayer phase propen- sity of the membrane might lead to an enhancement of t he effective s ubstrate concentration i n the m embrane by redistributing t he substrate between the aqueous and membrane phase. H owever, other reasons for an improved substrate accessibility can not be excluded. We showed for the first time that the stimulation of catalytic CYP1A1 activity and its s tereoselectivity depend on the type of lipid present in the membrane. Anionic lipids, particularly PtdSer, favour the formation of the ultimate mutagen DE2 t o th e detrimen t of th at of the far less carcinogenic DE1. It has been reported that lipids affect several c atalytic activities of CYP3A4 [40,4 1]. Thus, it would be interesting to know whether other enzymatic activities of human CYP1A1 also depend on the membrane structure and/or lipids. ACKNOWLEDGEMENTS This work was s upported by grants of the G erman Research Foundation (DFG) to I. R. a nd D. S. (RO 1287/2-3), and to P. K . (436 WER 17/8/01), and the Volkswagen-Stif tung to D. S. (I/75 468). We are grateful to Dr F. J. Gonzalez for providing CYP1A1 cDNA and virus for reductase expression (National Cancer I nstitute, NIH , Bethesda, MD, USA). We thank Dr A. Chernogolov for protein purification, Dr D. Zirwer for CD measurements, A. Sternke f or her skilful cell culturing, and Dr H. Honeck and R. Zummach (all from Max D elbrueck Centrum for Molecu lar M edicine, Berlin-Buch, Germany) for assistance with HPLC. REFERENCES 1. Thakker, D.R., Yagi, H ., Lu, A .Y.H., Levin, W., Conney, A.H. & Jerina, D.M. (1976) Metabolism of benzo[a]pyrene: conversion of (+/–)-trans-7,8-dihydroxy-7,8-dihydrobenzo[a]pyrene t o h ighly mutagenic 7,8-diol-9,10-epoxides. Proc. Natl Acad. Sci. USA 73, 3381–3385. 2. Kapitulnik, J., Wislocki, P.G., Levin, W., Yagi, H., Jerina, D.M. & Conney, A.H. ( 1978) Tumorigenicity studies with diol-epoxides of benzo(a)pyrene which indicate that (+/-)-trans-7beta,8alpha-di- hydroxy-9alpha,10alpha-epoxy-7,8,9,10-tetrahydrobenzo(a)pyren e is an ultimate carcinogen in newborn mice. Cancer Res. 38, 3 54– 358. 3. Conney, A.H. (1982) Indu ction of microsomal enzymes by foreign chemicals and cancerogenesis by p olycyclic aromatic hydrocar- bons. Cancer Res . 42, 487 5–4917. Fig. 2. Lineweaver–Burk plots of the kinetic analysis of (–)-7,8-diol epoxidation to diolepoxide-2 by human CYP1A1 for selected lipid vesicle membranes. Vesicle membranes were reconstituted from CYP1A1, NADPH-P450 re ductase, and either Ole 2 PtdCho, Ole 2 PtdCho/ Ole 2 PtdEtn ( 2 : 1 , w/w), Ole 2 PtdCho/Ela 2 PtdEtn (2 : 1 , w/w), or Ole 2 PtdCho/PtdSer (2 : 1, w/w). Note the parallelism of the curves for Ole 2 PtdCho, Ole 2 PtdCho/Ole 2 PtdEtn, and Ole 2 PtdCho/Ela 2 PtdEtn. Ó FEBS 2002 Membrane-reconstituted human CYP1A1 (Eur. J. Biochem. 269) 1803 4. Yang, S.K., McCourt, D.W., Roller, P.P. & Gelboin, H.V. (1976) Enzymatic conversion of benzo[a]pyrene leading predominantly to the diol-epoxide r-7,t-8-dihydroxy-t-9,10-oxy-7,8,9,10-tetra- hydrobenzo[a]pyrene t hrough a single e na ntiomer of r-7,t-8- dihydroxy-7,8-dihydrobenzo[ a]pyrene. Proc. Natl Acad. Sci. USA 73, 2594–2598. 5.Gautier,J.C.,Lecoeur,S.,Cosme,J.,Perret,A.,Urban,P., Beaune, P. & Pompon, D. (1996) Contrib ution of human cyto - chrome P 450 to benzo[a]pyrene a nd benzo[a]pyrene-7,8-dihydro- diol me tabolism, as predicted from heterologous expression in yeast. Pharmacogenetics 6, 489–499. 6. Kim, J.H., Stansbury, K.H., Walker, N.J., Trush, M.A., Strick- land, P.T. & Sutter, T.R. (1998) Metabolism o f benzo[a]py re ne and benzo[a]pyrene-7,8-diol by human cytochrome P450 1B1. Carcinogenesis 19 , 1847–1853. 7. Schwarz, D., Kisselev, P., Cascorbi, I., Schunck, W H. & Roots, I. (2001) Differential metabolism of benzo[a]pyrene and ben- zo[a]pyrene-7,8-dihydrodiol by human CYP1A1 var iants. Carci- nogenesis 22 , 453–459. 8. Guo, Z., Gillam, E.M.J., Ohmori, S., Tukey, R.H. & Guengerich, F.P. (1994) Expression of modified human cytochrome P450 1A1 in Escherichia coli:effectsof5¢ substitution, stabilization, purifi- cation, spectral characterization, and c atalytic properties. Arch. Biochem. Bio phys. 312, 436–446. 9. Buters, J.T.M., Shou, M., Hardw ick, J.P., K orzekwa, K.R. & Gonzalez, F .J. (1995) cDNA- directed e xpression of human cytochrome P450 CYP1A1 using b aculovirus. Drug Metab. Dispos. 23, 6 96–701. 10. Zhang, Z.Y., Fasco, M.J., Huang, L., Guengerich, F.P. & Kaminsky, L.S. (1996) Characterization of purified human recombinant cytochrome P4501A1-Ile462 and –Val462: assess- mentofarolefortherarealleleincarcinogenesis.Cancer Res. 56, 3926–3933. 11. Ingelman-Sundberg, M. & Glaumann, H. (1977) Reconstitution of the liver microsomal hydroxylase system into liposomes. FEBS Lett. 78, 72–76. 12. Bo ¨ sterling, B., Stier, A., Hildebrandt, A.G., Dawson, J.H. & Trudell, J.R. (1979) Reconstitution of cytochrome P-450 and cytochrome P-450 reductase into phosphatidylcholine-phosphati- dylethanolamine bilayers: c haracterization of structure and metabolic activity. Mol. Pharmacol. 16, 332–342. 13. Ingelman-Sundberg, M., Hagbjo ¨ rk, A L., Ueng, Y F., Yama- zaki, H. & Guengerich, F.P. (1996) High rates of substrate hydroxylation by human cytochrome P450 3A4 in reconstituted membranous vesicles: Influence of membrane charge. Biochem. Biophys. R es. Commun. 221, 318–322. 14. Ingelman-Sundberg, M., H aaparanta, T. & Rydstro ¨ m, J. (1981) Membrane charge as effector of cytochrome P-450LM2 catalyzed reactions in reconstituted liposomes. Biochemistry 20, 4100–4106. 15. Blanck, J., Smettan, G., R istau, O., Ingelman-Sundberg, M. & Ruckpaul, K. ( 1984) Mechanism of r ate control of the NADPH- dependent reduction of cytochrome P-450 by lipids in recon- stituted phospholipid vesicles. Eur. J. Biochem. 144 , 509–513. 16. Imaoka, S., Ima i, Y., Shimada, T. & Funae, Y. (1992) Role of phospholipids in reconstituted cytochrome P4503A form and mechanism of their activation o f catalytic activity. Biochemistry 31, 6063–6069. 17. Lambeth, J.D. (1991) Cytochrome P-450SCC: Cardiolipin as an effector of activity of a mitochondrial cytochrome P-450. J. Biol. Chem. 256, 4757–4762. 18. Schwarz,D.,Kisselev,P.,Pfeil,W.,Pisch,S.,Bornscheuer,U.& Schmid, R.D . ( 1997) E vidence that nonbilayer p hase propensity of the membrane i s important for t he side chain c leavage activity of cytochrome P450SCC (CYP11A1). Bioc hemistry 36, 14262– 14270. 19. Kisselev, P., Wessel, R., Pisch, S., Bornscheuer, R.D., Schmid, D. & Schwarz, D. (1998) Branched phosphatidylcholines stimulate a ctivity of cytochrome P 450SCC (CYP11A1) i n phospholipid vesicles by enhancing cholesterol binding, membrane incorp oration, and protein exc hange. J. Biol. Chem. 273, 1380–1386. 20. Platt, K.L. & Oesch, F. (1983) Efficient synthesis of non-K-region trans-dihydro diols of polycyclic aromatic hydrocarbons from o-quinons and catechols. J. O rg. Chem. 48, 265–268. 21. Funk, M., Frank, H., Oesch, F . & Platt, K.L. (1994) De velopment of chiral stationary phases for the enantiomeric resolution of dihydrodiols of p olycyclic aromatic hydrocarbons by p-donar– acceptor interactions. J. Chromatogr. A 659, 57–68. 22. Yagi, H ., Akagi, H., Thakker, D.R., Mah, H.D., Koreeda, M. & Jerina, D.M. (1977) Absolute stereochemistry of the highly mutagenic 7,8-diol 9,10-epoxides derived from the potent carci- nogen trans-7,8-dihydroxy-7,8-dihydrobenzo[a]pyrene. J. Am. Chem. Soc. 99, 2358–2359. 23. Chernogolov, A., Schwarz, D., Schunck, W.H. & Roots, I. (2000) Purification and characterization of baculovirus-expressed CYP1A1.1, CYP1A1.2, and CYP1A1.4. 13th International Sym- posium on Microsomes and Drug Oxidation. 5, 111. 24. Modi, S ., Paine, M.J., Sutcliffe, M.J., L ian, L Y., Primrose, W .U., Wolf,C.R.&Roberts,G.C.K.(1996)Amodelforhumancyto- chrome P450 2D6 based on homology modeling and NMR stu- dies of sub strate binding. Biochemistry 35, 4540–4550. 25. Tamura, S., Korzekwa, K.R., Kimura, S., Gelboin, H.V. & Gonzalez, F.J. (1992) Baculovirus- mediated e xpression and functional characterization of human NADPH-P450 oxidor- eductase. Arch. Biochem. B iop hys. 293, 219–223. 26. Omura, T. & Sato, R. (1964) Th e carbon monoxide-binding pig- ment of liver microsomes. J. Biol. Chem. 23 9 , 2370–2378. 27. Yasukochi, Y. & Masters, B.S. (1976) Som e properties of detergent-solubilized NADPH-cytochrome c (cytochrome P450) reductase purified by biospecific affinity chromatography. J. Biol. Chem. 251, 5337–5344. 28. Shou, M., Ko rzekwa, K . R., C respi, C.L., Gonzalez, F.J. & Gel- boin, H.V. (1994) The role of 12 cDNA-expressed hum an, roden t, and rabbit cytochromes P450 in the metabolism of benzo[a]pyrene and benzo[a]pyrene-trans-7,8-dihydrodiol. Mol. Carcinogenesis 10, 159–168. 29. Depierre, J.W. & Dallner, G. (1975) Structural aspects of the membrane of the endoplasmic reticulum. Biochim. Biophys. Acta 415, 411–472. 30. Cullis, P.R. & De Kruijff, B. (1978) The polymorphic phase behaviour of phosphatidylethanolamines of natural and synthetic origin. Biochim. Biop hys. Acta 513, 31–42. 31. Yang, F.Y. & Hwang, F. (1996) Effect of non-bilayer lipids on the activity of membrane enzymes. Chem. Phys. Lipids 81, 197–202. 32. Ahn, T., Guengerich, F.P. & Yun, C H. (1998) Membrane insertion of cytochrome P450 1A2 promoted by anionic phos- pholipids. Biochemistry 37, 12860–12866. 33. Yun, C H., Song, M. & Kim, H. (1997) Conformational change of cytochrome P450 1A2 induced by ph ospholipid and d etergents. J. Biol. Chem. 272, 1 9725–19730. 34. Balvers, W.G., Boersma, M.G., Vervoort, J., O uwehand, A. & Rietjens, I.M. (1993) A specific interaction between NADPH- cytochrome reductase and p hosphatidylserine and phosphatidyli- nositol. Eur. J. Biochem. 218, 1021–1029. 35. Williams, P .A., Cosme, J., S ridhar, V ., Johnson, E.F. & McRee, D.E. (2000) Microsomal cytochrome P450 2C5: comparison to microbial P450s and unique features. J. Inorg. Biochem. 81 ,183– 190. 36. Stubbs, C.D. & Slater, S.J. (1996) The effects of n on-lamellar forming lipids on membrane protein–lipid interactions. Chem. Phys. Lipids 81, 185–195. 1804 P. Kisselev et al. (Eur. J. Biochem. 269) Ó FEBS 2002 37. Epand, R.M. (1996) The prope rties a nd biological role s of no n- lamellar forming lipids. Chem.Phys.Lipids81, 101–264. 38. Gruner, S.M., C ullis, P.R., H ope, M .J. & Tilcock, C.P.S. (1985) Lipid polym orphism: t he m olecular b asis o f n onbilayer phases. Annu. Rev. Biophys. C hem. 14, 211–238. 39. Hui, S.W. (1987) Non-bilayer-forming lipids: Why are they nec- essary in biomembranes? Comments Mol. Cell. Biophys. 4/5, 233– 248. 40. Eliasson, E., Mkrtchian, S., Halpert, J. & Ingelman-Sundberg, M. (1994) Substrate-regulated, cAMP-dependent phospho rylation, denaturation, and degradation of glucocorticoid-inducible rat liver cytochrome P45 0 3A1. J. Biol. Chem. 269, 18378–18383. 41. Shet, M.S., Fisher, C.W., Holmans, P.L. & Estabrook, R.W. (1993) Human cytochrome P450 3A4: enzymatic properties of a purified recombinant fusion protein containing NADPH-P450 reductase. Proc. Natl Acad. Sc i. USA 90 , 11748–11752. Ó FEBS 2002 Membrane-reconstituted human CYP1A1 (Eur. J. Biochem. 269) 1805 . PAPER Epoxidation of benzo[ a ]pyrene-7,8-dihydrodiol by human CYP1A1 in reconstituted membranes Effects of charge and nonbilayer phase propensity of the membrane Pyotr. used to investigate the impact of the hexagonal phaseformingtendencybyYangandHwang[31].Datain Fig. 1A and in Table 1 show that the activity of CYP1A1 is significantly