Oxygen- and Carbon-centered Free Radical Formation during Carbon Tetrachloride Metabolism OBSERVATION OF LIPID RADICALS IN VIVO AND IN VITRO*

Free radical reactions involved in the metabolism of carbon tetrachloride by rat liver have been considered to be a cause of at least part of the injury resulting from exposure to this halocarbon. In an earlier study employing electron spin resonance and spin-trapping techniques, we demonstrated that trichloromethyl (‘3*cc13) radicals are readily observed in rat liver microsomes metabolizing l3cCl4, and that the same radi- cal could be shown to form in vivo in the liver of intact rats given a single dose of I3CCl4. This report describes the production of lipid dienyl (La) and oxygen-centered lipid radicals (LO. or LOO-, or both) in in vitro systems metabolizing lSCCl4, and also the formation of lipid dienyl radicals (L.) in liver of intact animals exposed to CC14. The radicals appear to be produced in a se- quence of reactions governed among other things by the oxygen tension in the system. The lipid radicals (Le) which form in intact liver of CCl,-treated rats are apparently the result of an attack on lipids of the endoplasmic reticulum by ‘‘.Cc13 radicals formed by reductive cleavage to CCl, and are the initial inter- mediates in the process of lipid peroxidation. These investigations demonstrate that while the events oc- curring in liver microsomes in vitro appear to parallel those which take place in

PBN.' In the reconstituted system, the .CCL radical was observed only when a specific form of liver microsomal cytochrome P-450 (52,000 Da) was present (3). This particular cytochrome was quickly destroyed during the reaction. Loss of this 52,000-Da cytochrome P-450 appears to be the earliest demonstrable molecular change i n uivo in the liver of rats exposed to CC1, (4). Our initial observations were evaluated by Kalyanaraman et al. (5), who suggested that the radical detected was a lipid dienyl radical rather than the .CC13 radical. The question was resolved in our laboratories by the use of l3Ccl4 for both the in uitro and i n vivo systems (6). The 12-line spectrum obtained with l3CcI4 as compared to the 6line spectrum observed with '*CC14 clearly demonstrated that the . CC13 radical was being formed. These results using 13Clabeled carbon tetrachloride have recently been confirmed by Albano et al. (7) using the same spin-trapping agent, PBN. A more detailed analysis of the spectra obtained from I3Clabeling studies indicated that Kalyanaraman et al. (5) were also correct in the sense that it now appears that more than one type of radical intermediate is produced during the metabolism of carbon tetrachloride by liver microsomes.
The information presented in this article supports the conclusion that a sequence of detectable radicals are produced in vitro and apparently i n vivo also, and that this sequence is markedly influenced either by high or low levels of oxygen in the system. These results indicate that lipid dienyl radicals (L.) are formed during the metabolism of CCl, by rat liver microsomes metabolizing cc14. Both lipid dienyl radicals (L.) and lipid oxy (LO.) or lipid peroxy (LOO . ) radicals (or both) appear to be the primary species trapped initially in the i n vitro systems incubated under low oxygen-containing atmospheres. In all systems, however, the very stable trichloromethyl radical adduct signal is ultimately the major feature of the ESR spectrum. In the intact animal studies, evidence is presented that, in addition to .CC13 radicals, there is also formation of carbon-centered lipid radicals which can be trapped in uivo by ( MO)3PBN. These investigations are being carried out as part of a program to determine how the sequence of early molecular events occurring in the liver relates to the peroxidation of lipids in the endoplasmic reticulum which occurs as a consequence of CC1, exposure. This information in Liver should be useful in determining the possible role these processes may play in the hepatotoxicity of this halomethane (8).

MATERIALS AND METHODS
P-Nicotinamide adenine dinucleotide phosphate, reduced form (NADPH) and oxidized form (NADP), glucose 6-phosphate, and glucose-6-phosphate dehydrogenase were obtained from the Sigma Chemical Co., St. Louis, MO; phenobarbital sodium was purchased from Merck Co., Inc., Rahway, NJ; PBN was obtained from Eastman Organic Chemicals, Rochester, NY; l3CCI4 (99 atom % I3C) was from Stohler Isotope Chemicals and Merck Co., Inc., Rahway, NJ; AtmosBag (the flexible inflatable polyethylene chamber with built-in gloves; can be used to work in a totally isolated and controlIed environment) was purchased from Aldrich Chemical Co., Milwaukee, WI; nitrogen gas and oxygen were obtained from Liquid Air, Inc., San Francisco, CA argon gas was from Big Three Industries Inc., Houston, TX; crystalline (MO)3PBN was synthesized in our laboratory by the procedure described by Sommermeyer and Seiffert (9). ' The structure and purity of (M0)SPBN was confirmed by mass spectrometry and nuclear magnetic resonance in accordance with the criteria outlined by Sommermeyer and Seiffert (9). 2 Male Sprague-Dawley rats (300-400 g) that had been maintained on a commercial rat ration were injected with phenobarbital sodium intraperitoneally in 0.85% NaCl solution (75 mg/kg body weight) for 4 days. Treatment of rats with phenobarbital is not required to observe the effects seen but does enhance the intensity of ESR spectra observed. The rats were killed and the livers were removed. Liver microsomes were prepared by homogenizing the livers in 0.15 M potassium phosphate buffer, pH 7.4, and were subjected to centrifugation at 10,000 X g for 15 min. The supernatant fraction was then centrifuged at 105,000 X g for 1.0 h. The microsomal pellets obtained were kept frozen until ready for use. The thawed pellets were resuspended in 50 mM potassium phosphate buffer, pH 7.4, so that the microsomes from 1 g of liver were suspended in 1.0 ml of buffer. The protein concentration of microsomal suspensions were determined by the method of Lowry et al. (10). The ESR spectrometer settings were as follows: microwave power, 25 milliwatts; modulation amplitude, 1 G; time constant, 10 s; scan range, 100 G; and scan time, 16 or 30 min. The spectra were taken at room temperature, 25 "C, unless otherwise indicated.
The incubation systems not carried out in an air atmosphere were assembled in a sealed AtmosBag which was flushed with either argon gas, nitrogen gas, or oxygen. If argon gas or nitrogen gas was used, all the components of the reaction mixture were gassed slowly with argon gas or nitrogen gas for at least 5 min. The reaction mixture consisted of liver microsomes (3.3 mg of protein), 0.1 M PBN, 0.2 mM I3CCl4 (99 atom % 13C), and either 0.3 mM NADPH or an NADPHgenerating system which was composed of 5 p M glucose 6-phosphate, 0.3 g~ NADP, and 0.5 Kornberg unit of glucose-6-phosphate dehydrogenase/ml of reaction system. All the components of the system were contained in a buffer system composed of 0.05 M potassium phosphate, pH 7.4. The final volume of all in vitro systems was 1.0 ml. The incubation of the reaction mixture was carried out at various times at room temperature (25 "C) except in certain cases where different temperatures (0 and 37 "C) were employed. Depending on the nature of the experiment, the incubation systems were either incubated in test tubes first before transferring into the bottom of Pasteur pipettes with sealed tip ends, or else the systems were immediately transferred into the pipettes after assembly. The open top end of the Pasteur pipette was then closed and the tube was centrifuged for 3 min at 500 X g in a holder designed to prevent collapse of the thin part of the tube. This technique was used to sediment the microsomes toward the tip end of the pipette so that the particulates were within the magnetic field of the spectrometer. Microsomes which have been frozen prior to use sediment easily at the g force indicated.
In experiments in which PBN was added at various times after l3CCI4 metabolism was initiated, the experimental conditions were the same as described above. At the end of the initial incubation period at 25 "C, 13 mg of PBN was added to the reaction mixture (to achieve 0.1 M final concentration), and then the system was incubated for another 15 min at 25 "C. The ESR spectrum of the reaction E. G. Janzen and C. M. DuBose, manuscript to be submitted elsewhere which will contain more information on this type of spintrapping chemistry. system was then determined by placing the mixture in a Pasteur pipette as described above.
In the in vivo experiments, the rats were fasted for 24 h before being administered a mixture of '3cC14 and either PBN or (M0)3PBN. A mixture of I3Cc14 (80 g1/100 g of body weight) and 1.0 ml of a 0.05 M solution of (MO),PBN in glass-distilled water was homogenized in 0.3 ml of stripped corn oil. This emulsion was administered to the male Sprague-Dawley rats by stomach tube. After 15 min, the rats were killed, the livers were removed, and the liver lipids were immediately extracted by the method of Folch et al. (11). The concentrated lipid extracts were placed in the tip of Pasteur pipettes for scanning in an EPR spectrometer (Varian E-9). The separation of the spin adducts of '3.CC4 from the lipid radicals was accomplished as follows. The lipid extracts were shaken with 30 ml of acetonitrile. The mixture was centrifuged resulting in a layered separation of the lipid and the acetonitrile. The acetonitrile layer was recovered and concentrated. EPR spectroscopy of both the acetonitrile extract and the remaining lipid was performed.
The parameters used for the computer simulated EPR spectra are given in Table I.

RESULTS AND DISCUSSION
I n Vitro Production of Radicals by Rat Liver Microsomes I n vitro experiments were conducted under air and either oxygen-poor, or oxygen-rich conditions in which the gas phase of the chamber where the systems were assembled and incubated was thoroughly flushed either with O2 or with nitrogen or argon gas which contained about 25 and 1 ppm 02, respectively. Situating the reaction system at the bottom of a Pasteur pipette in which the thawed, resuspended liver microsomes sediment toward the bottom of the sealed tip provides for an extended diffusion gradient with respect to oxygen in the gas phase, not dissimilar to the gradient between the blood stream and the endoplasmic reticulum in intact liver tissue in terms of oxygen access. Depending on the gas phase employed, a distinctively different progression of ESR signals was observed during the course of the reaction. The signals were all dependent, however, on an NADPH-dependent rat liver microsomal enzyme activity. The in vitro studies described in the present report were conducted under conditions similar to those employed in our earlier reports (6) except for the variation in some cases of the time and temperature of incubation and the atmospheric composition of the systems.
Radical Production in Vitro under an Air Atmosphere-When rat liver microsomes, NADPH, PBN, and ' 3CC13 are incubated under an air atmosphere for 30 min to 1 h, a 12line spectrum due to the spin-trapping of the 13C-centered trichloromethyl radical adduct of PBN is always observed (Fig. 1). The 12-line spectrum makes this 13C-centered radical ( I = 1/ z for 13C) easily identifiable (aN = 13.9, a; = 1.5, a$-13 = 9.5, a: = 0.23 G ) since the "C-centered radical adduct of PBN has only a 6-line spectrum (aN = 14.1, a7 = 1.8 G).
However, if this system is observed at earlier time intervals, a different sort of spectrum is obtained. There is a rapid appearance (with 6 min) of a signal in the form of a 6-line spectrum shown in Fig. 2 (first trace). This spectrum is reproducible. Under the conditions used for these experiments, the spectrum intensity is relatively weak and the lines are broad. The latter feature must be due in part to the presence of oxygen in the system. However, it should be noted that the high field doublet is broadened more than the other doublets, indicating that the adducts generating the signal are bulky molecules. These initial radical adducts must be derived from membrane lipid radicals because the total signal is present in the microsomal pellet isolated from the system, and can be recovered from the thoroughly washed chloroform phase of a Folch extract. Total extraction of the adduct into 'g value shift in gauss relative to other spectra, e.g. 0.5 means center is shifted 0.5 G upfield.
1. EPR spectrum of the 's.CCls radical. This signal was observed after a 30-min incubation of rat liver microsomes with 13CClr, an NADPH-generating system, and PBN at 25 "C under an air atmosphere. The composition and conditions of incubation of the system as well as the manner in which the EPR observations were made are described under "Materials and Methods." A, spectrum assigned to 13. CC13 obtained with complete reaction system. B, spectrum of system in which NADPH was omitted. The presence of a triplet of doublets of much smaller amplitude was also observed in the spectrum and is explained in Fig. 2. The EPR parameters for the ".CCI3 signal are aN = 13.9, up" = 1.5, abJc.' = 9.5, at.' = 0.23 G. that phase would not be expected to occur if the spin-trapped radical had any solubility in the aqueous phase. Using only the spacings between the low field doublet and the center doublet, the following hyperfine splitting constants are ob-
Within 30 min after appearance of the LO. radical adduct, FIG. 2. Evidence for lipid oxy radical formation during "CCl. metabolism by rat liver microsomes. Time sequence of radical generation observed during CC, metabolism in the complete system is described in Fig. 1. Within 6 min of incubation, the EPR signal appeared which was assigned to a mixture of L. and LO. radicals. Note the diminished doublet at the high field end of the 6min spectrum indicating that a relatively bulky radical molecule (presumably a phospholipid) has been spin-trapped. At later time intervals, the signal assigned to L. and LO. diminishes and the 12line spectrum characterizing the 13.CC13 radical becomes the dominant feature (indicated by arrows in the 120-min spectrum). The doublets indicated by the asterisks are assigned to LO. since the splittingconstants are consistent with those of primary and secondary oxy radicals of PBN (aN = 13.8, up" = 2.2 G). the 1 3 . c c l 3 signal becomes detectable and grows rapidly during the next hour to become the dominant feature of the spectrum which is stable for several days. The initial 6-line spectrum continues to diminish.
The advantage of using '3C-labeled CCl, (other than for the definitive identification of the signal produced by the PBN adduct of l3.CC13) is that the effect of the isotope is to shift the -CC13 spectral peaks away from the position where other "C-c&tered and oxygen-centered radical adduct signals would appear. In order to obtain more information about the nature of these early signals, the oxygen content of the systems was decreased to a low level by flushing with either nitrogen or argon gas, or was increased by flushing the systems with 0,.
Radical Production in Vitro under Conditions of Low Oxygen Tension-When the microsomal metabolism of l3CCl, is carried out in the presence of PBN at room temperature for 15

Oxygen-and Carbon-centered Lipid Free Radicals in Liver
min under a nitrogen atmosphere containing approximately 25 ppm 02, there was simultaneous observation of a t least three radical species (Fig. 3A). In addition to the 12-line CCl, signal, there is a set of doublets with a larger P-H hyperfine splitting constant (aN = 14.4, up" = 3.25 G) characteristic of a cdrbon-centered radical (marked by triangles). In addition, shoulders on the low field and center doublets, plus the unusual cancellation of peaks within the high field doublet, clearly indicate that an oxygen-centered radical has also been trapped (marked by daggers). A computer simulation of these three components is also shown (Fig. 3B). The carbon in the carbon-centered radical is not I3C, and, therefore, not derived from "CCl,. The diminished signal at the high field end of the spectrum indicates a bulky radical, and, since all of the spin adducts in this system are extractable with the membrane lipids, it is highly likely that these are spin adducts of lipid radicals (L . ) and lipid oxy radicals (LO. ).
The sequence of radical production in a system in which air is flushed from both the aqueous and gas phase above the  Table I).
system using argon is shown in Fig. 4. The effect of incubation temperature is also shown. The 0, concentration of the system is reduced to approximately 1.0 ppm 02. At 25 and 37 "C, a 6-line signal characteristic of the L. adduct ( u N = 14.1 G, a: = 3.3 G ) appears very rapidly (within 1 min), and is much more intense at 37 and 25 "C than when the same reaction is carried out at 0 "C (ice bath). At all temperatures, trapping of Is. CCln is observed subsequently, but occurs much earlier at 0 than it does a t 25 and 37 "C. The spectrum observed a t 1 min is probably due to overlapping L. and LO. adduct signals since the intensity of the second pair of doublets is considerably greater than that of the first, especially at 37 "C (Fig. 4); that is, as in Fig. 3, the center doublet indicates summation of more than one radical species, and this is reflected as peak broadening in the other doublets of this spectrum. Fig. 5 shows that, after 40 min, a t all three temperatures, the I:'. CCln signal has now become the dominant feature of the EPR spectrum, while the initially observed 6-line signal has diminished, especially in the systems incubated at 25 and 37 "C where this signal was previously strong. All of these signals are extractable from the system by Folch extraction with chloroform-methanol.
Because the formation of spin adducts in systems where PBN is added a t time 0 provides no indication of the duration of production of a particular form of radical over the total period of the incubation, the addition of the spin-trapping agent was made a t different time intervals during the reaction. Essentially, only radicals formed from the time of addition onward would be subject to trapping. Microsomal systems metabolizing "CCl, under nitrogen were incubated for 30 min a t 25 "C before the addition of PBN, and then allowed to incubate for another 15 min. As long as 30 min after assembling the CC1,-metabolizing system, lipid radicals and some oxygen-centered radicals as well are still being generated in sufficient amounts to provide a substantial signal which is observed along with that of the I3.CCl3 adduct (Fig. 6). Addition of PBN 120 min after starting the reaction indicates that production of L. radicals has diminished to levels undetect.able by this technique. The formation of ".CC13 radicals, however, is still continuing. These results provide further indications that the generation times for LOradicals under conditions of very low oxygen tension is brief. Lradical formation continues for a longer time. For example, oxygencentered radicals were a major feature of the spectrum if PBN is added initially but not if PBN is added at 30 min. Also L.
can still be trapped after the reaction had proceeded for up to 1 h, but usually not if PBN was added after that time. The .CCln adduct is still observed if PBN is added 2 h after initiating the reaction.

Radical Production in Vitro under an Oxygen Atmosphere-
When this same microsomal system is incubated with 02, the carbon-centered lipid radical is not observed, but an adduct appears within 15 min which has the characteristics of the oxygen-centered radical ( U N = 13.88, a: = 2.17 G ) (Fig. 7A). This signal is most likely due primarily to the LO. adduct of PBN since it is also membrane-bound and totally lipid soluble. This signal predominates in the system at least for several hours. By 24 h, the ':3.CC13 radical has appeared and the intensity of the lipid oxy radical is diminishing (Fig. 7B). Thereafter, the much more stable ".CCl, radical adduct is the only signal detected. This suggests that the domain in which the ".CCl, radicals are formed can become locally depleted of both oxygen and lipids. Even though there is an abundance of oxygen in these systems, the geometry of the incubation vessels, with the reaction system sedimented a t the tip end of sealed Pasteur pipettes, could provide for loci sufficiently low in 0, to permit some adduct formation be-

Oxygen-and Carbon-centered Lipid Free Radicals in Liver 2139
FIG. 4. EPR spectrum of a microsomal system metabolizing ''Ccl4 under an argon atmosphere. The argon gas employed contained about 1 ppm of 02. The reaction system was assembled as in Fig. 3 except for incubation temperatures ire indicated. The incubations were carried out for 1.0 min at either 0, 25, or 37 "C. A, the 0 "C spectrum shows that the 12-line ".CCl3 radical is observed more readily at this early time than at 25 ( B ) or 37 "C (C) where the overlapping L. and LO. doublets predominate. This may reflect the decreased fluidity of membrane lipids at 0 "C which could hinder the availability of the lipids to react with the I3.CC13 radicals being formed. The low temperature together with the very low oxygen tension in the system, would enable PBN to compete for the l3.CCln radicals more effectively at 0 than at 25 ( B ) or 37 "C (C). In the latter two systems, L. radicals are dominant at 1.0 min. 5. EPR spectrum of a microsomal system metabolizing ''Ccl4 under an argon atmosphere. The reaction system and experimental conditions are the same as in Fig. 4 except that the incubation time in these experiments was 40 min. Incubations were done at 0 ( A ) , 25 ( B ) , and 37 'C (C). After 40 min at 0 "C, the spectrum displays the 12-line ".CCI3 signal but at a lesser intensity than the system incubated at 25 "C. The intensity of the L. and LO. radical signal is decreased by this time at both 25 and 37 "C but is essentially unchanged at 0 "C, indicating that the spin adducts of L. and LO. are relatively unstable at 25 "C or above. Whereas the ".CC13 signals were barely detectable at 1.0 min in the 25 and 37 "C systems (Fig. 4, B and C), this radical adduct is a major feature of the spectra of these systems at 40 min ( B and C).

FIG.
tween "-CC13 and PBN. In addition, during the 24-h period, the enzymic activity of the system which produces the radicals from "CCI, was determined to be lost, and the less stable oxygen-centered lipid radical adducts have largely decayed, leaving the much more stable 1 3 . CCln adducts as the dominant feature.
All of the above experiments indicate that the spin-trapping of ".CCln in vitro by PBN in liver microsomes metabolizing CCl, occurs earlier under very low oxygen concentrations (i.e. argon atmosphere) than it does under air or oxygen. When the amount of oxygen present in air or commercial N, is available to the system, the lipid radical can still be observed under the conditions of these experiments (for example, see Fig. 3), but under a 100% O2 atmosphere, it is not possible to observe L. (Fig. 7 A ) . Instead, only the oxygen-centered radical is observed.
The results can be accommodated by considering the following reaction scheme: Given that the reaction is initiated at some rate depending on the local concentration of CCl, and the appropriate cytochrome P-450 (assumed to be in the reduced form), the .CC13 radicals can either react with the lipid molecules in the immediate vicinity of the reaction site or with PBN. The rate constant for spin-trapping .CC13 appears to be rather slow (13) (estimated to be less than lo4 M-' s-'). In contrast, the addition of radicals like .CC13 to olefins in hydrocarbon solvents is faster (14) (e.g. 3 X lo6 M" s-' for . CF3 + CH2 = CH2 in cyclohexane or heptane). Since hydrogen atom abstraction is probably slower (this value is unknown but could be about IO5 M" s-I), the major route for the initial reactions of . CC13 radicals should be addition to the unsaturated sites of lipid molecules. Moreover, the organization of the lipid molecules in relation to the enzyme system producing the .CCln radicals may be such that the reaction of the radicals with the lipids in the immediate vicinity may be facilitated. However, as the accessible lipid molecules are depleted, the reaction of .CCln radicals with PBN begins to predominate (60-90 min in the experiment a t 25 "C).
Simultaneously, of course, L. can be trapped or react with oxygen. The latter reaction will always be much faster in the presence of even a small amount of oxygen since the rate constant of the reaction of carbon-centered radicals with oxygen is essentially the rate of diffusion of oxygen to the radical (109-10'0 M" s-'). In comparison, the rate constant for spin-trapping carbon-centered radicals by PBN is much smaller (15, 16) (1.3 X lo5 M-' s" for 1-hexenyl in benzene a t 40 "C; 6.8 X lo4 M" s-' for 6-hepten-2-yl in benzene a t 40 "C). Thus, L. can only be detected when oxygen is absent or completely consumed in the vicinity of the spin trap.
In regard to the possible trapping of the lipid peroxy radical,

SCHEME 2
Although very little is known about the reaction chemistry of .CCls and oxygen, it can be assumed that the rate would be very fast to produce the peroxy radical. The trichloromethyl peroxy radical would behave similarly to a lipid peroxy radical, although a greater reactivity might be expected because of the electron withdrawing nature of the trichloromethyl group. The rate constant for the addition of C13CO0 e to PBN has been estimated (13) (5.6 X lo5 M" s" in CCl,). However, this adduct does not appear to be stable at room temperature. The trapping of trichloromethoxy radical seems very unlikely since &cleavage to give chlorine atoms appears to be a rapid process? C. M. DuBose, unpublished observations. Thus, the spectral data are consistent with the concept that, initially, the reaction of the -CC& radical with 0 2 and with the microsomal lipid proceeds at a significantly greater rate than it does with PBN. The addition of CC1, metabolites to unsaturated sites of the lipid has previously been detected by using l4Cc1, (17). The resulting L. radicals would also rapidly react with O2 to form LOO.. As the limited oxygen supply is depleted, Lwould be the principal radical that is spin-trapped by PBN, producing the type of mixed signal seen in Figs   37 "C ( Fig. 5) appears to be a reflection of the mobility of the lipid molecules in the microsomal membrane which, in turn, could influence their availability for reaction with 13-CC13 within the domain of the radical formation. Decreased fluidity resulting in restricted availability of the lipids at that site at low temperatures would favor an earlier reaction of .CC13 with PBN, which is the effect that was observed.

TABLE I1 Coupling constants for spin adducts of trimethxy phenyl t-butyl nitrone and 2-hydroxy-4,6-dimthoxyphenyl t-butyl nitrone
Recently, Albano et al. (7) reported the spin-trapping of lipid radicals in peroxidizing fatty acids and in rat liver microsomes metabolizing CCl, using 2-methyl-2-nitrosopropane as the trapping agent. However, in view of the report by Mason et al. (18) that 2-methyl-2-nitrosopropane can react with double bonds of unsaturated fatty acids by a pseudo-Diels-Alder mechanism based on reactions of similar nitroso spin traps (19), the use of 2-methyl-2-nitrosopropane to detect radicals in biological systems should be done with that information in mind (20,21). I n Vivo Production of Radicals in Rat Liver Several years ago, the feasibility of spin-trapping radicals in livers of intact animals was reported from our laboratory after exposure of rats to "CCl, (2) and to halothane (22). In the case of l3Ccl4, we have only been able to detect the ".CC13 radical in the liver with PBN even though, in the in vitro investigations reported above, several other radical species were detected with this spin trap.
(MO)3PBN is a spin trap which forms radical adducts with quite different hyperfine splitting parameters than PBN (9). Specifically, the P-H-splitting constants are larger and, thus, it was thought that the L. and LO. adducts could be more easily distinguished in the spin adduct spectra: When (M0)3PBN was administered to animals in vivo along with "'CCl, the spectrum shown in Fig. 8 was observed. The signal was found in the lipid extract of the liver as in the case of 1 3 . c c l 3 radical trapped by PBN. No signal is observed if fresh liver tissue, (M0)3PBN, and ' 3cc14 are subjected to a Folch extraction together, demonstrating that the signal observed in the liver extract of animals treated with (M0)3PBN and CCl, in vivo is not an artifact of the extraction procedure. In order to ascertain if the small additional peaks observed in Fig. 8 were possibly due to the I3Cc13 adduct, the liver lipid containing the spin adducts was extracted with acetonitrile and concentrated to a small volume. Fig. 9A shows a weak but distinct l3Ccl3 adduct in the acetonitrile extract. Fig. 9B indicates that the adduct with the narrow doublet shown in Fig. 8 still remains in the lipid material.
The fact that triplets of doublets with relatively small 0-H splittings were observed was puzzling, since authentic adducts of (MO),PBN have P-H coupling constants in the range of 3.7-12.00 G., Further investigation of the spin-trapping chemistry of (MO)3PBN disclosed that certain carbon-centered adducts of (M0)3PBN demethylate in solutions exposed to air to give HO(M0)2PBN, for example, the n-butyl adduct: The mechanism of this reaction will be discussed elsewhere., The spin adducts of HO(MO),PBN give relatively small 8-H hyperfine coupling constants. A comparison of some carboncentered and oxygen-centered radical adducts of (MO)3PBN and HO(MO),PBN is shown in Table 11. The spectrum in Fig. 8 consists The spectrum in Fig. 9A is clearly due to a "CCCl.? adduct. Since the hyperfine coupling constants are similar to those of the "CCln adduct of PBN, the spectrum is assigned to the ". These results indicate that, depending on the spin-trapping agent employed in investigations on "CCl, metabolism in vivo, different radicals may be observed, and this suggests that a combination of spin-trapping agents may be useful in such experiments. The reason for not observing carbon-centered L. adducts in vivo with PBN, whereas L. adducts of this spin trap are readily observed in microsomes in vitro during CCl, metabolism, appears to be due to decay of the L e adducts during lipid extraction and concentration. The carbon-centered L. adducts of (MO),PBN are much more stable to such manipulation.
In vitro experiments with [(MO)sPBN] with rat liver microsomes have been performed and gave similar results to those performed in vivo. Spectra consisting of two components with very broad lines were obtained after 30 min (Fig. 10A) (see e.g. Fig. 10A after 30 min). Analysis of these spectra indicates that again an adduct (or adducts) of HO(MO),PBN  nitrogen hyperfine coupling constants obtained from the still broad lines after 24 h in Fig. 10B (uN = 14.94, a; = 2.15 G) fall between those obtained from oxygen-and carbon-centered spin adducts ( a~ = 13.8 and 16.09 G, respectively). Perhaps both LO. and L. adducts of HO(MO),PBN are present in the mixture. With the exception that the LO. radical has yet to be observed in vivo, the findings thus far indicate that the events occurring in vitro during CCl, metabolism by the liver also are occurring in vivo. It now appears possible to trap radicals formed during lipid peroxidation in tissues of intact animals.
There is substantial evidence indicating that lipid peroxidation must occur in the endoplasmic reticulum during the metabolism of CCl, in mammalian liver tissue in vivo (23).
studies are more difficult to assign at this time since the Detection of conjugated dienes in microsomal lipids (24). loss

Oxygen-and Carbon-centered Lipid Free Radicals in Liver 2143
of polyunsaturated fatty acids from the endoplasmic reticulum (25), and exhalation of pentane and ethane in CC1,-treated rats (26) constitute part of this evidence. Similar findings are observed in vitro with liver microsomal preparations (27).
Prior treatment of experimental animals with free radical scavenging compounds provides a degree of protection against the toxicity of CCl, (28). These compounds also inhibit lipid peroxidation in the endoplasmic reticulum (and in microsomes in vitro) during the metabolism of CCl, (27), but do not appear to inhibit the binding of CCl, to lipids or macromolecules (27). Hence, the possibility exists that the toxicity of CCl, to hepatocytes may be exerted to a significant degree as a consequence of the injury to the endoplasmic reticulum resulting from the peroxidative breakdown of lipids in that membrane.
The sequence of early radical events occurring during ccl4, metabolism in vivo as envisioned in this report is consistent with the following: CCI, enters the hepatocyte, diffuses to the endoplasmic reticulum and is reductively cleaved to .CC13 radicals by an NADPH-dependent cytochrome P-450 system requiring the 52,000-Da form of this group of heme proteins (3). Presumably, .CC13 radicals are produced at a locus vicinal to unsaturated phospholipids and addition to the double bonds can occur. Also, . CCl, radicals react with O2 to form C13COO-. Peroxidative degradation of the lipids is also initiated by reaction of the .CC13 and C1,COO. radicals with these lipids as indicated previously. The lipid radicals then react with oxygen to form peroxy radicals which initiate further peroxidation of other lipids via the oxy radical: LOO. + LH + LOOH + L.
The data indicate that .CCI3 must react with the microsomal lipid extensively before beginning to react with PBN since, as described above, the .CC& radical adduct is not observed during the initial stages of the reaction while lipid radicals are observed. The latter adduct formation could only be expected in domains where oxygen is essentially depleted. Both PBN and (MO),PBN are sufficiently lipid-soluble to provide for trapping of the lipid radicals so that the availability of the spin-trapping agent to the . CC13-forming site seems likely. Hence, oxygen and lipid (and probably other organic membrane components as well) must be competing with PBN for . CCl3 radicals during the early phase of the reaction. The reaction at this time appears to be occurring in highly localized sites in the endoplasmic reticulum since initially there is a very rapid, specific polymerization of the 52,000-Da cytochrome P-450 which appears to be the primary heme protein associated with the production of the C13C. radical in liver tissue (29). Since free radical scavengers provide a significant degree of protection against the lipid peroxidation and the membrane damage caused by CCl, to animals (27), it is reasonable to conclude that the free radical events under observation are causally related to the injury process. It has been claimed that lipid peroxidation in the hepatic endoplasmic reticulum was not a consistent feature of CC1,-induced liver damage because mice, which are very susceptible to CCl, poisoning, did not show evidence that lipid peroxida-tion occurred after CC1, administration (30). However, it has now been demonstrated that the mouse responds to CCl, treatment with an early peroxidation of lipids which is more marked than that occurring in the rat (31). Unpublished observations in our laboratory have also revealed that the chicken, which is very resistant to CCl, toxicity (32), produces only an extremely weak . CCl, EPR signal in the liver following CC1, treatment. CC1, treatment in the chicken also causes little or no lipid peroxidation in vivo (33). Hence, there is a correlation between . CCl, production and lipid peroxidation in the liver with the hepatotoxicity of CCl4 that suggests a causal relationship, but further investigation will be required to demonstrate an obligatory link between production of the radicals which have been observed in vivo and the subsequent liver damage.