The Formation of a Novel Free Radical Metabolite from CCl, in the Perfused Rat Liver and in Viuo*

Electron spin resonance spectroscopy has been used to monitor free radicals formed during,CCl, metabolism by perfused livers from phenobarbital-treated rats. Livers were perfused simultaneously with the spin trap phenyl N-t-butylnitrone and with either 12CC14 or l3Ccl4. Perfusate samples and CHCla:CH30H extracts of perfusate and liver samples were analyzed for phenyl N-t-butylnitrone radical adducts of reactive free radicals. In the organic extracts, hyperfine cou- pling constants and 13C isotope effects observed in the ESR spectra indicated the presence of the radical ad- duct of the trichloromethyl radical. Surprisingly, an additional free radical signal about two orders of magnitude more intense than that of the phenyl N-t-butyl- nitrone/CCl; radical adduct was observed in the aqueous liver perfusate. This adduct was also detected by ESR in rat urine 2 h after intragastric addition of spin trap and CC14. This radical adduct had hyperfine coupling constants and 13C isotope effects identical with the radical adduct of the carbon dioxide anion radical (COX). Analysis of the pH dependence of the coupling constants yielded a pK, of the radical adduct formed either in the perfused liver or chemically. Carbon tetrachloride is converted into CClt by cytochrome P-450 through a was then centered in an E-231 TEloz pK. determination of the PBN/CO; radical adduct were conducted microwave cavity for analysis. Some preliminary experiments and the using an IBM ER-200 ESR spectrometer operating at 9.7 GHz with a 100-kHz modulation frequency and equipped with an ER-4103 TM microwave cavity. The simulations of ESR spectra were performed on a Hewlett-Packard HP 9835B computer equipped with a Varian data acquisition system.

Electron spin resonance spectroscopy has been used to monitor free radicals formed during,CCl, metabolism by perfused livers from phenobarbital-treated rats. Livers were perfused simultaneously with the spin trap phenyl N-t-butylnitrone and with either 12CC14 or l3Ccl4. Perfusate samples and CHCla:CH30H extracts of perfusate and liver samples were analyzed for phenyl N-t-butylnitrone radical adducts of reactive free radicals. In the organic extracts, hyperfine coupling constants and 13C isotope effects observed in the ESR spectra indicated the presence of the radical adduct of the trichloromethyl radical. Surprisingly, an additional free radical signal about two orders of magnitude more intense than that of the phenyl N-t-butylnitrone/CCl; radical adduct was observed in the aqueous liver perfusate. This adduct was also detected by ESR in rat urine 2 h after intragastric addition of spin trap and CC14. This radical adduct had hyperfine coupling constants and 13C isotope effects identical with the radical adduct of the carbon dioxide anion radical (COX). Analysis of the pH dependence of the coupling constants yielded a pK, of 2. 8 for the CO; radical adduct formed either in the perfused liver or chemically. Carbon tetrachloride is converted into CClt by cytochrome P-450 through a reductive dehalogenation. The trichloromethyl free radical reacts with oxygen to form the trichloromethyl perpxyl radical, CC1300', which may be converted into COCl and then trapped. This radical adduct would hydrolyze to the carboxylic acid form, which is detected spectroscopically. Alternatively, the carbon dioxide anion free radical could form through complete dechlorination and then react with the spin trap to give the COX radical adduct directly.
CCl, is dehalogenated reductively to the trichloromethyl free radical by cytochrome P-450 (1). The PBN1/CC1, radical adduct was identified in microsomal incubations containing NADPH, CC14, and PBN based on the similarity of its sixline ESR spectrum to that of the free radical formed by W photolysis of a CCl, solution containing PBN (2). Poyer et al. 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 t o indicate this fact.
invariant ESR spectrum was detected and eventually assigned to a PBN radical adduct of a lipid alkoxy radical (4).
Tomasi et al. (5) obtained similar results with extracts of microsomes or isolated hepatocytes incubated for 10 min with 13CCL and PBN, either aerobically or anaerobically. Again, a 12-line ESR spectrum indicative of the PBN/l3CC1; radical adduct was detected. The PBN/'3CC1; radical adduct has also been formed in uiuo and detected in liver extracts of rats given 13CCL and PBN orally (3,6). The failure to detect the PBN/ CCl30O' radical adduct under aerobic conditions, despite the high rate constant for the reaction of CC1; with 02, has been attributed to the high reactivity of CC1300' and the instability of the PBN-peroxyl adduct (5).
We have examined rat urine and effluent perfusate from rat liver for PBN adducts formed during CC1, metabolism. In addition to detecting the trichloromethyl radical in the liver and in extracts of effluent perfusate, a novel free radical metabolite of carbon tetrachloride was discovered in the effluent perfusate. This free radical intermediate is a product of the reaction of CCl; with oxygen.

MATERIALS AND METHODS
PBN, sodium formate, ferrous sulfate, bovine serum albumin, catalase, and ascorbate oxidase were purchased from Sigma and were used without modification. Hydrogen peroxide (30%) (American Chemical Society certified) and carbon tetrachloride (analytical reagent grade) were from Fisher. [13C]Sodium formate and [13C]carbon tetrachloride were from MSD Isotopes.
Female Sprague-Dawley rats (Zivic-Miller, 250-350 g) were treated with sodium phenobarbital (1 mg/ml) in drinking water for at least 5 days to induce cytochrome P-450 prior to perfusion experiments. Livers were perfused with Krebs-Henseleit bicarbonate buffer (pH 7.6,37 "C) saturated with 02:COz (95:5) in a non-recirculating system as described previously (7). The buffer was pumped into the liver via a cannula placed in the portal vein and out of the liver via a cannula in the inferior vena cava. The effluent perfusate flowed past a Teflonshielded, Clark-type O2 electrode and was collected in polyethylene bottles for ESR analysis. PBN (10 mM) was dissolved in the perfusate and carbon tetrachloride was bound to albumin by stirring with a 22.5% aqueous albumin solution for 16 h.
Liver samples (7 g) were homogenized in perfusion buffer (30 ml) and extracted with 30 ml of a CHCl,:CH,OH (2:l) solution. The mixture was centrifuged for 10 min at 2500 rpm. The organic layer was removed, dried with anhydrous sodium sulfate, and placed in a Pyrex sample container with a 3-mm outside diameter side arm for ESR analysis. Samples were degassed using standard vacuum techniques and stored in liquid nitrogen until ESR analysis. The aqueous layer of the extraction was bubbled with oxygen for 10 min and then with nitrogen for 5 min prior to ESR analysis. Aqueous perfusate samples were treated in a similar manner. Effluent perfusate was extracted using the same procedure except that 400 ml of perfusate was extracted with 15 ml of CHCI,:CH3OH (2:l).
For the analysis of PBN adducts in urine, rats were fasted for 24 h. Then they were given 0.5 ml of 0.05 M PBN solution and 0.3 ml of 2.2 M CCl, in corn oil intragastrically three times at 'h-h intervals.

4542
This is an Open Access article under the CC BY license.

Radical Metabolites of CC14 in Perfused Liver and in Vivo 4543
About 2 h after the last dose, rat urine was collected in a Petri dish and was washed into a small ampoule with an equal volume of the perfusion buffer. The urine sample was transferred to an ESR flat cell, and 4 pl containing 1 unit of ascorbate oxidase and 4.7 pl containing 1 unit of catalase were added. The solution was then bubbled with oxygen for 10 min, followed by nitrogen for 5 min.
A hydroxyl radical-generating Fenton system containing formate, the PBN/CO; radical adduct. The PBN/OH' radical adduct was a hydroxyl radical scavenger, was used to independently synthesize prepared by the addition of 50 pl of a 0.36 M ferrous sulfate solution to 5 ml of a pH 7.6 solution of perfusion buffer containing 3.3 mM hydrogen peroxide and 15 mM PBN. The PBN/CO; radical adduct was generated by the addition of ["CI-or ['3C]sodium formate (100 mM) to perfusion buffer prior to the addition of ferrous sulfate.
In the study of the pH dependence of the hyperfine couplings of the PBN/13CO; radical adduct, 200 p1 of the solution described above were added to 3 ml of perfusion buffer which was pH-adjusted by the addition of 1 N HC1 or NaOH. After mixing thoroughly, 2 ml of the sample was aspirated into the ESR cavity and its spectrum was recorded within 2 min. The final pH of the remaining portion of the solution was measured after the ESR spectrum was recorded. In the corresponding analysis of the radical from the liver perfusate, the portion of perfusate with the highest free radical concentration was identified first. The ESR spectrum and pH value of a mixture of 200 pl of this solution and 3 ml of pH-adjusted perfusate were then obtained.
The ESR spectra were obtained using a Varian E-109 spectrometer operating at 9.5 GHz with a 100-kHz modulation frequency. Aqueous samples were aspirated into a quartz flat cell centered in an E-238 TMllo microwave cavity. Organic solutions were transferred to the ESR sample tube side arm which was then centered in an E-231 TEloz pK. determination of the PBN/CO; radical adduct were conducted microwave cavity for analysis. Some preliminary experiments and the using an IBM ER-200 ESR spectrometer operating at 9.7 GHz with a 100-kHz modulation frequency and equipped with an ER-4103 TM microwave cavity. The simulations of ESR spectra were performed on a Hewlett-Packard HP 9835B computer equipped with a Varian data acquisition system.

PBN Radical Adducts in Liver
Extracts-Introduction of oxygenated perfusate containing PBN (10 mM) into a perfused liver resulted in an increase in oxygen uptake of about 15%, possibly due to the partial mixed-function oxidation of the spin trap (Fig. 1). The subsequent infusion of CC1, (1 mM) produced a small increase followed by a progressive decrease in O2 uptake for the next 30 min of perfusion ( As in previous studies of CC1, metabolism (3, 6), ESR spectra were taken of organic extracts of the liver after perfusion with cc14 and PBN. Experiments utilizing 12CC14 produced a stable six-line ESR spectrum due to the PBN radical adduct of the trichloromethyl radical characterized by its hyperfine coupling constants of aN = 14.45 G and aF = 1.85 G (Fig. 2 A ) . Confirmation of this spectral assignment was provided by the 12-line ESR spectrum obtained from the organic extract of a liver into which l3Ccl4 was infused. This spectrum exhibited an additional hyperfine coupling of 9.20 G attributable to the nuclear spin of 13C (Fig. 2B).
ESR analysis of the aqueous layer in the extract of a liver exposed to "CC14 yielded a stable siz-line spectrum (aN = 15.8 G and a; = 4.6 G) (Fig. 2C) which was not similar to PBN/ CC1,. Following infusion of 13CC1,, the corresponding ESR spectrum yielded a 12-line spectrum where 2 lines are nearly superimposed (Fig. 2 0 ) with an additional hyperfine coupling from 13C (a;"3 = 11.7 G). There was no evidence of this new radical in the organic phase.
PBN Radical' Adducts in Effluent Perfmute-Direct ESR analysis of aqueous perfusate also yielded the spectra attributable to the new radical adduct (Fig. 3, A  hours. Samples bubbled with oxygen for 10 min and then with nitrogen for 5 min exhibited a stable ESR signal that was identical with the spectrum of untreated samples allowed to sit at room temperature for several hours. Oxygen could either oxidize nitroxide-reducing species such as ascorbate or the sulfhydryl groups of proteins or oxidize the hydroxylamine form of the radical adduct directly to the nitroxide form (Equation 1). R R Subsequent bubbling with nitrogen decreased the concentration of dissolved oxygen to levels below which ESR linebroadening was not significant.
The amount of PBN/CCli radical adduct in the perfusate was below the sensitivity limits of the ESR spectrometer; therefore, it was concentrated by extraction. The organic layers from extractions with 0rganic:aqueous ratios of 1:40 produced ESR spectra that were composites of spectra from three free radical species (Figs. 4A and 5A). The ESR spectrum from the effluent perfusate of a liver exposed to l3Ccl4 clearly indicated the presence of the PBN/13CC1, radical adduct. Extracts of control samples collected during perfusion with PBN prior to CC1, exposure produced a distorted sixline spectrum most likely due to di-t-butyl nitroxide and an unassigned free radical adduct of PBN. This latter weak spectrum, also produced by the addition of PBN (10 mM) to perfusate followed by extraction, is probably due to an impurity and is not a radical adduct. These impurity spectra changed slightly in time, possibly due to exposure to fluorescent light and the high potential for radical adduct formation due to the high PBN concentration (approximately 0.4 M if all PBN was extracted into the organic phase). Computer simulation was necessary to resolve the three contributing spectra in the composite spectrum from the organic extract. After a satisfactory simulation of the spectrum from the experiment involving l3Ccl4 was obtained (Fig. 5B), the 13C coupling was deleted and relative amplitudes were adjusted slightly to yield a composite spectrum (  perfusate of a liver perfused with "CCL (Fig. 4A): In the future, studies of CCL metabolism involving ESR analysis of perfusate extracts will require the use of 13CC1, to minimize difficulties associated with these impurity species.
In Vitro Preparation of PBNICOI Radical Adduct-A Fenton system containing formate was used to generate the PBN/COT radical adduct independently (Fig. 6). The hydroxyl radical produced from the reaction of ferrous ion with hydrogen peroxide (Equation 2) abstracted the hydrogen atom from the formate ion (Equation 3) producing the carbon dioxide anion free radical which was trapped by PBN (Equation 4). Reasonably concentrated solutions (1 mM) of the PBN/COG radical adduct could be formed which were stable for several hours at 0 "C. The ESR hyperfine coupling constants for these PBN/CO; radical adducts (aN = 15.8 G, a : = 4.6 G, = 11.7 G) were identical with those obtained from aqueous perfusate.
The pH dependence of the ESR hyperfine couplings for the PBN/COG radical adduct was measured (Fig. 7). Identifica- tion of the pH at the midpoint of the pH-sensitive region of the hyperfine couplings gave a pK, value of 2.85 for this radical adduct produced from either the Fenton system or effluent perfusate (Fig. 7).

Radical Metabolites of CCl, in Perfused Liver and in Vivo
Formation of the PBNICO, Radical Adduct i n Vivo-The absence of any mention of the PBN/COT radical adduct in previous spin-trapping studies of CC14 metabolism i n vivo led to the conclusion that this species does not remain in the liver, the focus of previous work, but rather moves into the bloodstream and eventually is excreted in the urine. Indeed, the radical adduct was observed in rat urine collected 2 h after the rat had been treated with PBN and [13C]carbon tetrachloride (Fig. 8). Initial ESR spectra of urine samples exhibited strong ascorbate semidione free radical peaks which partially obscured the PBN/CO< radical adduct spectrum. Treatment of the sample with oxygen, ascorbate oxidase, and catalase to oxidize ascorbate to dehydroascorbate and to convert HzOz into H20 reduced the ascorbate free radical ESR peaks significantly (Fig. 8). The PBN/CO; radical adduct concentration was not increased by this procedure, indicating that the ascorbate semidione free radical does not significantly reduce this radical adduct in urine. The similarity of the ESR hyperfine couplings for the spectrum obtained from the rat urine Spectrometer settings were the same as in B.
( Fig. 8, Table I) to those obtained from the effluent perfusate and the Fenton system justifies the assignment of the free radical in the urine as the PBN/COT radical adduct. reported ESR hyperfine coupling constants for the PBN/ CCl, radical adduct and the radical adduct observed in organic extracts of livers perfused with CCl, (Table I). A substantial fraction of this species remained in the liver. To obtain measurable quantities of the PBN/CCI, radical adduct from the aqueous perfusate, it was necessary to use concentrating extractions. It is concluded that PBN/CC13 is formed from CCl, in the perfused liver as would be expected and is distributed based on its hydrophobicity. Evidence for a PBN Adduct of a Novel Radical Metabolite of CC1,"Using pulse radiolysis, the trichloromethyl radical has been shown to react with O2 to form the C13COO' peroxyl   free radical (8,Equation 5). Since Ccl4 is metabolized to phosgene (9) and COZ (10) in biological systems, it is reasonable to assume that the c1,cOo' radical is formed; however, direct evidence for this peroxyl radical from ESR spectroscopy is lacking. When CC1, was added to the perfused liver, 0, consumption increased 15-20 pmol/g/h (11). An increase in O2 uptake of approximately 10 pmol/g/h was observed in the presence of PBN (Fig. 1). Thus, it is possible that oxygencontaining free radicals are produced in perfused liyers exposed to CC1,. Indeed, a unique ESR signal with coupling constants distinctly different from the CCl; radical adduct was obtained in effluent perfusate from livers perfused with CCl,. Moreover, the species in the aqueous perfusate was between one and two orders of magnitude more intense than the CC1, radical adduct. Proof that this free radical arises from CC1, metabolism comes from the observation of an additional hyperfine coupling in the ESR spectra when a liver was exposed to 13CCL (Fig. 3).
The ESR hyperfine coupling constants for the radical ad-duct (Table I) in effluent perfusate correspond closely to values reported for the PBN/COT radical adduct generated photochemically (12) (Table I). Furthermore, the radical detected in the effluent perfusate had hyperfine couplings identical with those of the PBN/COX radical adduct produced in vitro from a Fenton system containing formate (Fig. 6). It is concluded, therefore, that this new species is the PBN/CO; radical adduct. This conclusion was supported by studies of the effect of pH on the ESR hyperfine coupling constants. A pK, of 2.85 was obtained when the PBN/COT radical adduct was generated either by the perfused liver or the Fenton system (Fig. 7).
Most previous ESR studies utilized organic extracts of tissue or hydrophobic membrane preparations to study radicals formed during CC1, metabolism. Since the pK, of the PBN/COX radical is less than 3, it is nearly completely ionized at physiological pH. Therefore, since this species is charged, it does not appear in organic extracts (Figs. 4 and 5), but is observed in the effluent perfusate (Fig. 3). This ionic character may explain, in part, why the PBN/COX radical adduct has been overlooked in past studies.
Pathways Responsible for the Formation of the PBNICOT Radical Adduct in Perfused Liver-It is possible that the PBN/COT radical adduct is formed from chloroform, phosgene, formate, or carbon dioxide. It is well established that small quantities of phosgene are produced from the metabolism of CCl4 (9, Equations 5 and 6). H.

+2e-+ZH+
-HCI + H20 Thus, the reaction of phosgene or a phosgene-derived free radical with PBN to form the PBN/COg radical adduct was considered. To evaluate this possibility, an experiment was undertaken using chloroform, which is metabolized to phosgene over 8 times faster than CC1, (9). However, no PBN/ CO; radical adduct was observed in the effluent perfusate of livers exposed to CHCL; therefore, the involvement of phosgene is unlikely (data not shown). Similarly, infusion of formate failed to produce the PBN/COI radical adduct, indicating that either this radical is not formed from formate produced from CC1, metabolism or formate is not absorbed. In experiments with 13C-labeled ccl4, the PBN/COG radical adduct was not diluted by any carbon source, such as carbon dioxide. Taken together, these data indicate that the PBN/COX radical does not arise from chloroform, phosgene, formate, or carbon dioxide. The possible dechlorination of PBN/CCl; to form PBN/CO.X is unlikely because PBN/ COT appears immediately in the perfusate and its concentration does not increase with time, whereas PBN/CCl, accumulates in the liver and would provide an increasing source of PBN/COZ. The reaction sequence most likely responsible for PBN/ CO; radical adduct formation involves the trichloromethyl peroxyl radical (ccl300*) (Equation 5). The trichloromethyl peroxyl radical is converted to the trichloromethoxy radical (CCbO * ) by a two-electron reduction followed by protonation and elimination of a water molecule (Equation 7). The trichloromethoxy radical then reacts with hydroxide ion to produce the chlorocarbonyl radical COC1, chloride ion, and a molecule of hypochlorous acid (Equation 7). C,3C00.+2e;+2H+ -Ct3CO. 5 ClcO + HOC1 +CI-171 -H20 Previous observations of a radical adduct of the chlorocar-bony1 radical from the photolysis of CCl, provide additional support that this species reacts with PBN (13, Equation 8).

191
The carbon dioxide anion radical then reacts with PBN to form the PBN/CO; radical adduct (Equation 9). The COX radical is known to reduce oxygen to superoxide with a nearly diffusion-limited rate, 2.4 X lo9 "' s-' (14). This reaction forms carbon dioxide, the final product of CC14 metabolism (10) and superoxide.
. c 0~+ 0 2 -c 0 2 + 0~ no1 The PBN/OZ radical adduct (aN = 14.8 G and ap" = 2.75 G) was not observed under any conditions. Apparently the abundant hepatic superoxide dismutase, which can totally suppress the formation of this radical adduct (E), disproportionated superoxide before it could be trapped in detectable concentrations. If PBN/COX is formed by the trapping of COX, then PBN must compete with O2 with its 2.4 X lo9 ~3 -l rate constant (14). Apparently, this is possible only because, at the site of reaction, the concentration of PBN is much greater than that of 02. It is noteworthy that CCl; is trapped by PBN in spite of the 3.3 X lo9 "'s" rate constant for its reaction with oxygen (16). Scheme 1 summarizes Equations 6-10. Although PBN/ COT is clearly CC1;-derived and presumably c c l 3 o 0 ' -derived, other aspects of the proposed mechanisms(s) are speculative. The mechanism(s) by which CC1; loses its three chlorine atoms to form PBN/COZ can only be suggested at this time (16).
Fate of PBNICOH in Vivo-This is the first report of detection of the adduct of a free radical metabolite in a body fluid of a living whole animal. Because of the charged nature of the PBN/COX radical adduct, we predicted that it should be filtered by the kidney and appear in the urine. Indeed, after treatment of a rat with PBN and 13CC&, an ESR spectrum identical with that characteristic of the PBN/COT was observed (Fig. 8). It was identical with the spectrum of the PBN/CO, radical adduct generated in vitro or detected in the effluent perfusate. Thus, the PBN/COX radical adduct is co2 indeed formed in viuo. Although phosgene is presently thought to be the precursor to carbon dioxide formed in uiuo from CC1, (Scheme l), the near diffusion-limited rate of air oxidation of the carbon dioxide anion radical is consistent with at least some of the carbon dioxide formed in vivo being the result of this alternate route (Scheme 1).
The formation of the PBN/CO; radical adduct in the perfused liver presumably arises from CC&OO' peroxyl radical formation. The direct evidence for CC1300. formation consists of in vitro kinetic (8,16) and ESR (17) studies of irradiated CCl,. Although the metabolism of CC1, to phosgene is thought to occur via C13COO', a non-free radical pathway is also possible. In view of these limitations and until the CC1300' radical can be detected in biological systems, the characterization of factors which influence PBN/COq formation may give insight into CC1,OO' formation in liver.