Hydroxyl Radical-mediated, Cytochrome P-450-dependent Metabolic Activation of Benzene in Microsomes and Reconstituted Enzyme Systems from Rabbit Liver*

The mechanism of benzene oxygenation in liver mi- crosomes and in reconstituted enzyme systems from rabbit liver was investigated. It was found that the NADPH-dependent transformation of benzene to water-soluble metabolites and to phenol catalyzed by cytochrome P-450 LM2 in membrane vesicles was inhibited by catalase, horseradish peroxidase, superox- ide dismutase, and hydroxyl radical scavengers such as mannitol, dimethyl sulfoxide, and catechol, indicat- ing the participation of hydrogen peroxide, superoxide anions, and hydroxyl radicals in the process. The cy- tochrome P-450 LM2-dependent, hydroxyl radical-mediated destruction of deoxyribose was inhibited con- comitantly to the benzene Oxidation. Also the microsomal benzene metabolism, which did not exhibit Mi- chaelis-Menten kinetics, was effectively inhibited by six different hydroxyl radical scavengers. Biphenyl was formed in the reconstituted system, indicating the cytochrome P-450-dependent production of a hydrox- ycyclohexadienyl radical as a consequence of interac-tions between hydroxyl radicals and benzene. The for- mation of benzene metabolites covalently bound to protein was efficiently inhibited by radical scavengers but not by epoxide hydrolase. The results indicate that the microsomal cytochrome P-450-dependent oxidation of benzene is mediated by hydroxyl radicals formed in a modified Haber-Weiss reaction between hydrogen peroxide and superoxide anions

The mechanism of benzene oxygenation in liver microsomes and in reconstituted enzyme systems from rabbit liver was investigated. It was found that the NADPH-dependent transformation of benzene to water-soluble metabolites and to phenol catalyzed by cytochrome P-450 LM2 in membrane vesicles was inhibited by catalase, horseradish peroxidase, superoxide dismutase, and hydroxyl radical scavengers such as mannitol, dimethyl sulfoxide, and catechol, indicating the participation of hydrogen peroxide, superoxide anions, and hydroxyl radicals in the process. The cytochrome P-450 LM2-dependent, hydroxyl radicalmediated destruction of deoxyribose was inhibited concomitantly to the benzene Oxidation. Also the microsomal benzene metabolism, which did not exhibit Michaelis-Menten kinetics, was effectively inhibited by six different hydroxyl radical scavengers. Biphenyl was formed in the reconstituted system, indicating the cytochrome P-450-dependent production of a hydroxycyclohexadienyl radical as a consequence of interactions between hydroxyl radicals and benzene. The formation of benzene metabolites covalently bound to protein was efficiently inhibited by radical scavengers but not by epoxide hydrolase. The results indicate that the microsomal cytochrome P-450-dependent oxidation of benzene is mediated by hydroxyl radicals formed in a modified Haber-Weiss reaction between hydrogen peroxide and superoxide anions and suggest that any cellular superoxide-generating system may be sufficient for the metabolic activation of benzene and structurally related compounds.
The liver microsomal benzene monooxygenase is known to have unique properties (7). It is not inactivated by metyrapone and is activated upon treatment of the membranes with detergent, in contrast to ordinary microsomal monooxygenase activities. It is known that hydroxyl radicals may react with the benzene ring at a rate of 3 X lo9 M-' s-l (13), yielding a hydroxycyclohexadienyl radical, which in the presence of *This work was supported by grants from Svensha Tobaks AB and the Swedish Medical Research Council. The costs of publication of this article were defrayed in part by the payment of page charges, This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
We present here results indicating that benzene monooxygenase (71, constituting the first enzymatic step in the metabolic activation process, mediates hydroxylation by hydroxyl radicals produced in an iron-catalyzed Haber-Weiss reaction between superoxide ions and hydrogen peroxide. The results may indicate that any cellular system generating superoxide anions will be sufficient for metabolic activation of benzene and structurally related compounds.

EXPERIMENTAL PROCEDURES
Materials-Catalase (11,500 units/mg), catechol, dihydroxyfumaric acid, D-mannitol, superoxide dismutase (2,410 units/mg), and hydroquinone were purchased from Sigma. Dimethyl sulfoxide was obtained from Merck, and ["Clbenzene (specific activity, 57 mCi/ mmol) was from New England Nuclear. Microsomal phospholipids were extracted from liver microsomes obtained from phenobarbitaltreated rabbits according to Bligh and Dyer (18) and stored under nitrogen in sealed tubes at -20 "C. Cytochrome P-450 LM2I and NADPH-cytochrome P-450 reductase was purified to apparent homogeneity from liver microsomes of phenobarbital-treated rabbits essentially as described by Haugen and Coon (19) and Yasukochi and Masters (20), respectively. The cytochrome P-450 LM2 preparations used had specific contents of 10.5-13 nmol/mg, and the specific content of flavin in the NADPH-cytochrome P-450 reductase preparations was 20-22 nmol/mg, when flavin was determined according to Iyanagi and Mason (21). Epoxide hydrolase (specific activity, 450 nmol of styrene oxide hydrolyzed per mg/min) was purified from liver microsomes of phenobarbital-treated rabbits as described by Halpert et al. (22).
Methods-Unilamellar phospholipid vesicles containing cytochrome P-450 LM2 and NADPH-cytochrome P-450 reductase were prepared by the cholate gel filtration method (23). The vesicles contained cytochrome P-450, reductase, and phospholipids in a molar ratio of about 1:0.3:1300. Incubations with benzene were performed with phospholipid vesicles corresponding to about 0.5 nmol of cytochrome P-450 in a final volume of 1 ml of 50 mM potassium phosphate buffer, pH 7.4, essentially according to Tunek and Oesch (7). The liposomes were incubated at 37 "C in stoppered glass tubes containing NADPH (0.5 mM) and 17 +M [14C]benzene dissolved in 10 p1 of acetone. The incubations were stopped after 20 or 60 min by adjusting the pH to 3-4 with 1 M HCl. Sodium chloride (100 pl, 9%, w/v) and 3 ml of ethyl acetate were immediately added and after mixing, the samples were centrifuged at 3400 X g for 10 min. The ethyl acetate extraction was repeated, and the combined organic phases were dried with 0.9 g of water-free sodium sulfate. The volume was reduced under a stream of nitrogen to about 100 pl, and the residue was applied on TLC plates which were developed using either ethyl acetate or to1uene:chloroform:ethyl acetate (l:l:l, by volume) as mobile 7312 Benzene Oxygenation phase. The TLC plates were autoradiographed, and radioactive zones were scraped off the plates and counted in a liquid scintillator (Intertechnique SL-30) using Aquoluma Plus as scintillator liquid.
Determination of Covalently Bound Benzene Metabolites-Incubations with benzene were performed as described above. After termination and extraction with ethyl acetate, 1 volume of ethanol was added to the water phase. The samples were rigorously shaken and centrifuged at 5000 X g for 15 min. The precipitates were washed with 3 ml of acetone:hexane (2510, v/v) and 3 ml of methanol (cf. Jollow et al. (24)). The pellets were dissolved in 0.3 ml of 1 M NaOH at 80 "C for 20 min. After neutralization with 1 volume of 1 M HCI, an aliquot was subjected to liquid scintillation.
Assay Methods-Hydroxyl radical-mediated destruction of deoxyribose was detected spectrofluorometrically as described by Halliwell and Gutteridge (25). Vesicles correspondingto 0.1 nmol of cytochrome P-450 LM2 were preincubated at 37 "C for 3 min in 1 ml of 50 mM potassium phosphate buffer, pH 7.4, containing 0.7 mM 2-deoxyribose. The incubations were started by the addition of NADPH to 0.1 mM and terminated after 15 min by addition of 1 ml of 1% (w/v) 2thiobarbituric acid and 1 ml of 2.8% (w/v) trichloroacetic acid. The samples were boiled for 10 min and subsequently centrifuged for 20 min at 20,000 X g. The amount of fluorescent material in the supernatant was determined using 532 nm for excitation and 553 nm for emission. Control incubations were performed in the absence of either NADPH, deoxyribose, or membrane vesicles. The thiobarbituric acid fluorescence is expressed as relative fluorescence against a standard of rhodamine G M = 100 units). 0-Demethylation of paranitroanisole was determined spectrophotometrically at 417 nm according to Netter and Seidel (26).
A.  (17 p~) in the presence of NADPH with reconstituted membrane vesicles containing NADPHcytochrome P-450 reductase and cytochrome P-450 LM2 resulted in the production of several lipid-and water-soluble metabolites, phenol constituting the major product ( Fig. 1). At least 11 different metabolites were formed, and among the most predominant ones were phenol, hydroquinone, and catechol. As shown in Fig. 1, the formation of both phenol and the other metabolites was strongly inhibited by the addition of the hydroxyl radical scavenger Me2S0 or of catalase to the incubation system. This indicates that hydrogen peroxide and hydroxyl radicals participate in the cytochrome P-450-dependent metabolic transformation of benzene. In the absence of cytochrome P-450 or NADPH, no or negligible metabolism of benzene was observed.

Incubations of benzene
Halliwell and Gutteridge (25) recently described a simple assay for hydroxyl radicals involving detection of fluorescent products formed by reaction of thiobarbituric acid with radical-destroyed deoxyribose. The cytochrome P-450 LM2-dependent conversion of benzene to water-soluble products, including the formation of covalently protein-bound benzene metabolites, was compared with respect to the inhibition profile obtained by addition of scavengers of hydrogen peroxide, superoxide anions, and hydroxyl radicals with the hydroxyl radical-mediated destruction of the deoxyribose. As shown in Fig. 2, concomitant inhibition of benzene metabolism and deoxyribose destruction was observed following the addition of either catalase, superoxide dismutase, M e 8 0 or mannitol to the system. By contrast, an ordinary cytochrome P-450 LM2-catalyzed reaction, the 0-demethylation of puranitroanisole, was not affected by the additions.
As shown in Table I, also the formation of the primary benzene metabolite, phenol, was concomitantly inhibited upon the addition of these scavengers to the incubation system. In addition, also catechol and horseradish peroxidase inhibited phenol production and the formation of watersoluble metabolites. The scavengers mannitol and catechol  ited the formation of covalently bound benzene metabolites Incubations were performed with liposomes corresponding to 0.4 in liver microsomes (Fig. 3). nmol  about 60% of the amount of protein-bound adducts were isolated compared to vesicles depleted of cytochrome bs (data not shown). In reconstituted phospholipid vesicles containing cytochrome bs instead of cytochrome P-450 LM2, the metabolic transformation of benzene was 0.58 pmol/nmol of cytochrome bs/min using otherwise similar conditions, whereas in vesicles only containing equivalent amounts of P-450 reductase, 0.37 pmol of benzene metabolites was formed per 0.3 nmol of reductase/min.

TABLE I1 Effect of hydroxyl radical scavengers on the NADPH-dependent oxidation of benzene to phenol and water-soluble metabolites in liver
microsomes from control rabbits Incubation conditions were as described under "Experimental Procedures" and in the legend to Fig. 3.

Scavenger used
To water tabolites   (Table I). mM potassium phosphate buffer, pH 7.4, with 17 PM benzene, containing 1 pCi of ["CJbenzene for 30 min at 37 "C. The products were In addition, liver microsomal metabolism of benzene was analyzed as described under "Experimental procedures," one thoureduced bY scavengers of hYhOXYl radjcals, i.e. the production sand cpm correspond to the formation of 0.2 pmol of metabolites/mg of phenol and water-soluble benzene metabolites was inhibof protein/min. The results hitherto presented indicate that the NADPHdependent metabolic transformation of benzene in liver microsomes is mediated by hydroxyl radicals as the active oxygenating species. It seems plausible that the radicals are formed in a iron-catalyzed Haber-Weiss reaction between hydrogen peroxide and superoxide anions. If this is the case, incubation of microsomes with increasing benzene concentrations would not result in normal Michaelis-Menten kinetics of the product formation. As shown in Fig. 4, elevated amounts of benzene in the incubation mixtures resulted in a proportionally increased rate of the production of water-soluble metabolites from benzene by the liver microsomes; no saturation of the activity was reached despite 10 mM benzene concentration.
Dihydroxyfumarate is known to autooxidize spontaneously in water, thereby generating superoxide anions (27). Incubal 0 O 0 h tion of benzene (17 p~) with 10 mM dihydroxyfumarate for 30 min at 37 "C resulted in the conversion of 2.7% of the added benzene to water-soluble metabolites, i.e. 7.7 pmol of products/min (Fig. 5). In the presence of superoxide dismutase, inhibition of the reaction was achieved; only 0.91 pmol of water-soluble products was formed per min.
Interaction of benzene with hydroxyl radicals will lead to the formation of hydroxycyclohexadienyl radicals, which in the presence of oxygen is converted to phenol (14,15) or in the presence of itself is converted to biphenyl (15). However, this type of radical is very reactive, and it may be expected that this intermediate may be responsible for covalent binding to proteins. In order to exclude the possibility that benzene oxide is the reactive metabolite binding to protein, reconstituted vesicles containing cytochrome P-450 LM2 were incubated with benzene in the presence of epoxide hydrolase. As shown in Table 111, the introduction of the hydrolase did not have any protective effect against the formation of water- High pressure liquid chromatographic analysis of benzene metabolites formed by cytochrome P-460 LM2-containing membrane vesicles. Reconstituted liposomes corresponding to 1 nmol of cytochrome P-450 LM2 were incubated for 1 h at 37 "C in 1 ml of 50 mM potassium phosphate buffer, pH 7.4, containing 1.6 pCi of ["Clbenzene (10 mM) and 0.5 mM NADPH. Biphenyl (5 pg) was added as carrier to the ethyl acetate phase obtained after extraction, and this phase was taken to dryness under nitrogen. The residue was subsequently dissolved in a small amount of methanol and injected onto the high pressure liquid chromatography column. Separation was achieved using a pBondapak CIS column (1.2 X 25 cm) and two pumps (Waters M 45 and 6000). Elution (1.2 ml/min) was performed using a linear gradient (7 min) of 60% (v/v) methanol in water to 100% methanol. Fractions were collected every 20 sec directly into counting vials to an Intertechnique SL-30 spectrometer. Counting was performed using Aquoluma Plus as scintillator liquid. Control incubations were performed in the absence of NADPH or membrane vesicles.
Benzene Oxygenation 7315 soluble benzene metabolites or metabolites covalently bound to protein, whereas mannitol completely inhibited both processes.
In order to obtain further evidence for the formation of the hydroxycyclohexadienyl radical as a product of hydroxyl radicals and benzene, experiments were designated to evaluate any formation of biphenyl in the reconstituted enzyme system. Using a concentration of benzene of 10 mM and reversed phase high pressure liquid chromatography for detection of benzene metabolites (Fig. 6), it was found that biphenyl was formed at the rate of 6.2 pmol/nmol of cytochrome P-450/ min.

DISCUSSION
Based upon the results obtained, one may postulate the following mechanism for the cytochrome P-450-dependent benzene monooxygenase (Fig. 7). Ferric cytochrome P-450 is reduced by NADPH-cytochrome P-450 reductase and upon binding of oxygen, the oxycytochrome P-450 complex is formed (Fe3+. 0;) which upon autooxidation releases superoxide anions (28)(29)(30). These will spontaneously dismutate to hydrogen peroxide but also be able to reduce ferric iron to a ferrous complex or to a species like (Fe . O#+ active in cleaving hydrogen peroxide into OH-and .OH-. The result will be a Haber-Weiss reaction The hydroxyl radicals generated will interact with benzene and produce the hydroxycyclohexadienyl radical (13) which in the presence of oxygen will give phenol. At high concentrations, the radical may also interact with itself, yielding biphenyl after elimination of water (15). This radical-induced reaction will probably not be specific for cytochrome P-450; any system producing superoxide anions containing trace amounts of iron should be capable of mediating the process. Accordingly, superoxide anions generated from dihydroxyfumarate were able to metabolize benzene to water-soluble products. In line with this argumentation, it may not be surprising that following injection of radioactively labeled benzene into mice, the main part of the covalently bound benzene metabolites in liver was found in mitochondria (31), known to produce substantial amounts of 0; (32,33), rather than in microsomes or cytosol (37).
Irons et al. (34) found that upon administration of benzene to perfused rat bone marrow, phenol, hydroquinone, and catechol as well as covalent binding of benzene metabolites to tissue constituents could be detected, indicating the capability of the bone marrow cells to metabolize benzene. However, the hydroxylase capacity of the microsomal system in bone marrow is very low (8) and may not explain the pronounced retention and covalent binding of benzene metabolites observed in the bone marrow (6,31,35,36). The superoxide anions released from leukocytes and granulocytes (37) may therefore be of importance in this respect.
Recently, it was found that benzene has the capability to induce the same type of cytochrome P-450 in rabbit liver microsomes as is observed following exposure of ethanol to animals (38,39)? This may be of certain interest since ethanol has previously been described to be metabolized by cytochrome P-450 LM2 according to the same mechanism as found presently for the benzene monooxygenase (17). Accumulating evidence suggest that the ethanol-and benzeneinducible form of cytochrome P-450 partly utilizes a similar reaction mechanism for ethanol oxidation as cytochrome P-450 LM2 (38,39).' This may indicate that the enhanced rate of microsomal benzene monooxygenase after exposure of benzene to animals (7, 39) may be inherent in an induction of this or similar types of microsomal cytochrome P-450 participating in the metabolism of benzene according to this radicalmediated oxygenation mechanism.
Originally, benzene oxide was proposed to be the reactive metabolite of benzene (3). However, recent studies have failed to show covalent binding to macromolecules or toxic effects of this compound (10,12). Since, in contrast to benzene, benzene metabolites such as phenol, benzene dihydrodiol, and hydroquinone or metabolites formed from these compounds, do not produce hemotoxic effects (40, 41), it seems plausible that another primary metabolite of benzene is toxic to the cell. Our results indicate that the hydroxycyclohexadienyl radical is the intermediate responsible for the covalent binding and the harmful effects.