Chemistry and biology of spin-trapping radicals associated with halocarbon metabolism in vitro and in vivo.

The spin-trapping method is introduced and discussed. Some chemistry of nitroxides and nitrones is reviewed. Pattern recognition of ESR spectra of nitroxides is outlined. Factors controlling the magnitude of hyperfine splitting constants are mentioned. Methods of assigning spin adducts are listed. Review articles in the literature are referenced. Results in the electrochemical reduction of halocarbons are presented and some parallels with superoxide chemistry shown. Various speculative reactions are given. The in vitro and in vivo experiments where halocarbon radicals have been detected by spin trapping are reviewed and some new results reported. A comparison for different animals is added.


Introduction
The possible involvement of free radicals in biochemical reactions has been suggested for a long time. However direct evidence for such intermediates has been lacking for many years. This is because typical concentrations of reactive free radicals in biochemical systems are usually too low for direct ESR detection. The method of spin-trapping was developed to help with this problem. Although the use of this technique is becoming quite popular it is also possible to make erroneous deductions with this method. Caution must be exercised and the literature consulted before conclusions are reached. Nevertheless, spin trapping appears to be the only approach to the detection of short-lived reactive free radicals at low concentrations in a biological environment and considerable promise is already evident for the use of this technique in in vivo experiments. In this application the method is unique.
The Spin-Trapping Method Nitroxides The spin-trapping reaction takes advantage of the *Department of Chemistry and Biochemistry, University of Guelph, Guelph, Ontario, NlG 2W1, Canada. tBiomembrane Research Laboratory, Oklahoma Medical Research Foundation, Oklahoma City, OK 73104. well known stability of the nitroxyl free radical function: O0 Nitroxides* are "stable" free radicals because the unpaired electron is resonance stabilized in the monomer and strong bonds cannot be formed in the dimer:  (1) *The IUPAC recommended name is aminoxyl instead of nitroxide; this nomenclature was selected to facilitate naming of nitroxides on the basis of the secondary amine precursor, e.g., dimethylaminoxyl.
The most stable nitroxides have "inert" groups attached to the nitrogen atom, typically methylated carbon atoms: H3C 0" CH3 (2) H3! !H3 The complete replacement of neighboring hydrogens with methyl groups serves two purposes: the bulky groups reduce what small tendency there might be for dimerization and the rate of disproportionation to the hydroxylamine and nitrone is reduced, e.g., Unsubstituted nitroxides are actually very reactive and disproportionate rapidly, e.g. 0 0 R-N-R -* R-N + R+ (7) R + H30-* ROH + H+ (8) If nitroxides are warmed to above room temperature, e.g., during heating in GC equipment or while recrystallizing or distilling, dissociation to the nitroso compound can occur: 0o 0 heat II R N-R -t R-N + R. (9) Nitrones The nitrone function is the N-oxide of an imine. The simplest nitrone is not stable, apparently polymerizing rapidly to unknown products. The smallest isolable nitrone is N-tert-butyl nitrone but even For example, ascorbic acid reduces nitroxides rapidly and sulfhydryl groups are capable of the same reaction in the presence of transition metal ions. Oxidation produces the nitroxonium ion with strong oxidizing agents: Replacing one of the methylene hydrogen with an alkyl or aryl group, or incorporating the nitronyl function into a ring produces compounds of reasonable stability (BBN, PBN, DMPO).  (11) peracid, which after heating produces the nitrone. PBN and DMPO are commercially available. It should be noted that hydrolysis of nitrones does occur slowly at pH 7, and the rate increases at lower pH. Hydrolysis is the reverse of the above reaction and produces two reactive products, the aldehyde and the hydroxylamine, both of which can become involved in free radical reactions.

Spin-Trapping
The spin-trapping technique is an analytical method for the detection of reactive short-lived free radicals (which are not highly resonance-stabilized) by the addition reaction to chemical traps e.g. nitrones, which give relatively persistent radical addition products [Eq. (12) Substituted benzyl tert-butyl nitroxide a spin adduct When the spin trap reacts with radicals, the persistent spin adduct nitroxide accumulates until the detection limit of the ESR spectrometer is exceeded and a signal due to the nitroxide is recorded. If the spin trap concentration is high enough and the rate constant of spin trapping is fast, the detection of R can be quantitative. The major challenge in spin-trapping is assigning the structure of the spin adduct correctly because only in this way can the identity of R be ascertained. Since until now such assignments most commonly have been made on the basis of the ESR spectra obtained some discussion on the ESR spectroscopy of nitroxides will be necessary.

ESR Spectroscopy of Nitroxides
The ESR spectrum of the unpaired electron alone consists of one peak. The interaction of the unpaired electron with a nucleus with no spin produces no change i.e. the spectrum is still one line. However, when the unpaired electron interacts with a nucleus which itself has spin (e.g., nuclear spin, I = 1/2, 1, 3/2, etc.), the single peak is replaced by a set of equally intense lines, where the number of lines in the set equals 2I + 1 (see  Illustrative ESR spectra of chemical species (e.g., atoms, molecules, or ions) that possesses one unpaired electron as well as one nearby nucleus (i.e., H, N, Na, or Al) with spin.
In nitroxides the unpaired electron is located in a irorbital on the nitrogen and oxygen atoms. The spin of the unpaired electron will interact with both nuclei, but since the oxygen nucleus has no spin, only the interaction with the nitrogen nucleus will be noticed (for 14N, I = 1). Thus a spectrum of three equally spaced equally intense lines is expected for a nitroxide with groups attached consisting of atoms with no nuclear spin, e.g., If the groups attached have atoms with nuclear spin but these are attached too far away for significant interaction, the spectrum will still be only three lines: e.g., di-tert-butyl nitroxide (since for "2C, I = 0, but for 1H, H3C CH3 (where hydrogens are three bonds away). Nuclei with spin three bonds or more away from the nitroxyl function usually do not contribute extra lines to the ESR spectra of nitroxides, but a small unresolved interaction in tert-butyl nitroxides from the nine -y-hydrogens does increase the line-width of the peaks obtained. It is to be expected that when a group or atom with nuclear spin is attached to the nitrogen such nitroxides would give spectra reflecting the interaction of the unpaired electron with this nucleus. An example is tertbutyl hydronitroxide, the radical obtained from hydrogen atom abstraction from tert-butyl hydroxylamine: The hydrogen is one bond away from the nitroxyl function (a-position), and the unpaired electron interacts significantly with this atom. The three lines of the nitroxide spectrum are further "split" into two lines giving a six-line pattern in total (Fig. 2). The appearance of the splitting pattern will depend on the magnitude of the interaction of the extra nucleus. In this case the splitting of the hydrogen is approximately equal to that of the nitrogen and the spectrum consists of four lines with intensity 1:2:2:1, or depending on the solvent, six lines with intensity 1:1,1:1,1:1 where the intensities separated by commas may be only partially resolved (Fig.  2).
Atoms with nuclear spin located two bonds away also interact significantly with the unpaired electron on the nitroxyl function (p-position). This interaction is thought to be due to hyperconjugation [Eq. (14)  The methyl hydrogens are in the 3-position and will interact with the unpaired electron on the nitroxyl function. All hydrogens will be equivalent. The splitting pattern (Fig. 3) can be predicted on the basis of this equivalence (see Fig. 4). Also the number of lines expected can be deduced from the expression 2nI + 1 where n = number of magnetically equivalent nuclei with spin. The spectrum of methyl tert-butyl nitroxide will consist of 12 lines, since each of the lines in the nitrogen triplet will be split into 1:3 H2 N(CH3)3 two hydrogens are left in the 1-position which are expected to be magnetically equivalent. Then the splitting pattern anticipated would be a triplet of 1:2:1 triplets according to the splitting diagram shown in Figure 5. The ESR spectrum of benzyl tert-butyl nitroxide, in fact consists of nine lines with the nitrogen triplet split into FIGURE 4. Computer-simulated ESR spectra of various substituted N-tert-butyl nitroxides. (Top two spectra) The first illustrative spectrum exhibits a relatively large nitroxide nitrogen triplet splitting with respect to the ,B-hydrogen triplet splitting. The second illustrative spectrum is a triplet of quartets again with a relatively large nitroxide nitrogen splitting with respect to the ,B-hydrogen splitting. (Bottom three spectra) In descending order are shown the anticipated ESR spectra for methyl, benzyl, and 1-phenylethyl) tert-butyl nitroxide, respectively. It should be noted that the ESR patterns for these nitroxides are those expected in organic solvents (e.g., benzene). The ESR patterns may change appreciably in solvents of different polarity (e.g. water). 1:2:1 triplets by the splitting from the methylenic hydrogens (Fig. 4). If another hydrogen in methyl tert-butyl nitroxide is replaced with a group with no nuclear spin in the (3position, for example methyl, CHS the remaining (-hydrogen causes the nitrogen triplet to be split into 1:1 doublets and a total of six lines result (see Fig. 4).
The interesting feature of these three spectra is that the magnitude of the ,B-hydrogen splitting is markedly different and decreases with substitution in the (3-position. Thus the (3-hydrogen splitting is 11.3, 8.25, and 3.73 G respectively for methyl, benzyl, and 1-phenylethyl tert-butyl nitroxide (Fig. 4)! If the groups were all freely rotating the (3-hydrogen splitting should be essentially the same. The conclusion of course is that these groups are not freely rotating and that the tert-butyl nitroxyl function presents substantial hindrance to free rotation of the other groups attached. Thus the bulk of the R group in the spin adduct is expected to influence the rotation of all the groups attached to the nitroxyl function including the (-hydrogen so that a unique splitting should be realized for every different Rgroup encountered. This prediction is usually demonstrated by consideration of a Newman projection. Looking down the C-N bond it is obvious that the largest groups will be oriented so that the distance between them is maximized (Fig. 6). 0 bulk of R: equal to phenyl Moreover the magnitude of the ,B-hydrogen splitting depends on the "dihedral" angle between the C-H bond and the p-orbital on nitrogen. When the angle is large the splitting is small; conversely when the dihedral angle is small the splitting is large. This relationship is accommodated by the expression where 0 is the dihedral angle defined in Figure 7, PN is the spin density on the nitrogen atom of the nitroxyl function, B is a proportionality constant for each radical, and Bo is the same constant when 0 = 900. For nitrox-idesBo00, PN~0.5, andB -52G. It follows that when the bulk of R is equal to that of phenyl in the above examples the (-H splitting should be small because 0 -900 (actually the C-H bond rocks above and below the plane of the nitroxyl ir-bond). This is found to be the case (Fig. 8).
For smaller R groups, the (3-hydrogen splitting is smaller than phenyl larger than phenyl  expected to increase; it does when R is hydrogen ( Fig.  9). For R groups larger than phenyl, the n-hydrogen splitting also increases as expected (Fig. 10). It would be nice to say that all other R groups fall into the same pattern, but this is not the case. In fact, almost no other examples can be predicted with confidence and the reason for this fact is still a subject of research. In general, polar groups with electron-withdrawing atoms cause a decrease in the nitrogen splitting and usually (but not always) the 3-hydrogen splitting is smaller as well. However, predictions for unknown spectra based on splitting constants can be hazardous and other routes need to be used to make assignments of structure. This is because changes in solvent polarity in general produce larger effects on the splitting of the nitroxyl nitrogen and the 3-hydrogen than changes in structure.
As expected from consideration of the two resonance structures for the nitroxyl function shown earlier, polar solvents lead to an increase in the nitrogen splitting since the "fraction" of the unpaired electron (spin density) increases on nitrogen when the solvent stabilizes the dipolar resonance form. Since the magnitude of the 3-hydrogen splitting depends on the spin density on nitrogen, the increase in solvent polarity should also cause an increase in the P-hydrogen splitting. This is found to be true for some but not all cases. Thus it is advisable to test the spectrum of a nitroxide in a number of solvents of differing polarity before assignments can be considered firm.

Methods for Assigning Spin Adduct Spectra
A number of methods have been used to assign spin adduct spectria.
Oxygenation of the Appropriate Secondary Amine Precursor. This method is excellent but tedious and will usually not be suitable for functionalized R groups.
Phenyl radical spin adduct of PBN (15) Oxidation of the Hydroxylamine or Hydroxylamine Anion of the Spi'n Adduct Nitroxide. A convenient way to make this intermediate is to add the organolithium or organo-Grignard to the nitrone being used as a spin trap: Di-tert-butyl peroxalate, an excellent thermal source of (16) tert-butoxyl radicals, can often be used to generate the radicals needed to add to 2-methyl-2-nitrosopropane to produce a PBN spin adduct since the addition of tertbutoxyl radicals to the nitroso compounds gives a nitroxide with quite a different nitrogen splitting.
Resolution of -y-Splitting in the R Group Added. In some cases, long range splitting can identify the radical trapped.
Resolution of long-range splitting can be improved by deuteration of the tert-butyl group: Sources. This method is frequently used when all previous methods are not suitable or other attempts have failed. If reactions are carefully selected this approach is quite dependable.

Chemistry of Halocarbon Radical Formation Electrochemical Reduction
When interest developed in the possibility that the spin trapping technique could be used to detect trichloromethyl radicals in biological systems from carbon tetrachloride, and halocarbon radicals from other halocarbons, a method needed to be developed to provide authentic spin adduct spectra for assignment purposes. The electrochemical generation of halocarbon radicals in the presence of the spin trap, PBN, was found to be successful in most cases where three (and sometimes two) halogens are attached to the same carbon atom: CX4 + e -*CX3 + X-CHX3 + e -* -CHX2 + X-(24) (25) where X = Cl, Br.

Spin-Trapping Literature
The spin-trapping literature includes numerous reviews on the subject. The first few written by Perkins (1), Lagercrantz (2), and Janzen (3) appeared around 1970. A number of specialized articles on this topic by Janzen appeared later (4,5). Additional reviews covering the physical organic aspects of spin trapping were published by Perkins (6) and Evans (7). More recently, spin trapping applications in biological chemistry have been extensively reviewed and these references should be consulted: Janzen (8), Finkelstein, Rosen, and Rauckman (9), Kalyanaraman (10), Buettner (11) and Mason (12). Some useful chapters have been written on spin trapping by a number of authors in two recently published. books (13,14) and a special issue of the Canadian Journal of Chemistry is devoted to papers in spin trapping and nitroxyl radical chemistry (15). Tabulations of spectroscopic parameters for spin adducts (up to 1978).are available, listed among all other nitroxide radicals (16) and an updated set of tables where entries are listed in increasing order of empirical formula ofthe radical trapped will be available shortly (17). A chapter is being prepared with emphasis on pitfalls and artifacts in spin-trapping (18).
However, HPLC methods were needed to ascertain the structure of the radical produced when both chlorine and bromine were present in the same molecule. Because reasonably polar solvents are needed for electrochemical studies, all investigations had to be made in acetonitrile (MeCN) or N,N-dimethylformamide (DMF).
The first question of interest was whether free radicals could be detected in the electrochemical reduction of halocarbons using the spin trapping method. These results could be important in comparison to biological systems, because the reduction of carbon tetrachloride, for example, by the reduced form of cytochrome P450, could conceivably proceed by a one-electron or two-electron process.
One-electron process: CCl4 + e *CC13 + Cl-Two-electron processes: (27) -CCl3-* :CCl2 + Cl- (28) In the electrochemical experiment, the trichloromethyl radical would have to diffuse away from the electrode surface into the body of the solution and be detected by the spin trap by addition. 0- (29) In an analogous fashion the trichloromethyl radical would have to diffuse away from the enzyme active site after formation and find the spin trap before further reduction to the anion could occur. Moreover, in both the electrochemical and biochemical systems the additional possibility that the spin adduct might be reduced to the hydroxylamine anion also existed.

IC1
Recently completed studies have shown that the electrochemical reduction of the following halocarbons gives the corresponding radicals by loss of chloride or bromide ion at potentials before reduction of PBN (19): -CCI3 from CCI4, -CBr3 from CBr4, -CHBr2 from CHBr3, -CHCl2 from CHCl3, -CH2Br from CH2Br2. These radicals could be trapped by PBN and ESR spectra assigned either by inspection or by HPLC methods. The presence of the halide ion as a product of reduction could be determined electrochemically. Methylene chloride (CH2CI2), methyl chloride (CH3Cl), and methyl bromide (CH3Br), could not be reduced before reduction of spin trap or solvent. Thus we conclude that a biochemical reduction pathway is quite feasible for production of halocarbon radicals from certain halocarbons. Whether a given reductive process is likely will depend on the redox potentials available.
The next question of interest was concerned with which halide would be lost in the reduction if both chlorine and bromine exist in the halocarbon. It was known that the toxic activity of the halocarbon increases in the order Br > Cl > F, but no information was available as to which halide bond was broken first in the metabolism of the halocarbon. The electrochemical reduction showed consistently that the bromine is lost first (as bromide) in all cases containing bromine, chlorine, and fluorine: -CC13 from BrCC13, -CHCl2 from BrCHCl2, -CHBrCI from Br2CHCI, -CHCI-CF3 from CHCIBr-CF3.
Moreover, the ease of reduction increases with the replacement of chlorine with bromine. Thus, a reductive dissociation can accommodate the observed increase in toxicity of the chloro and bromo halocarbons. Also it appears quite feasible that the various chloro-and bromosubstituted radicals could be trapped by PBN in the biological sample before -further reduction or decomposition, since the lipid environment is not likely as hostile to the radicals, at least with respect to reduction as the region around the electrode in the electrochemical experiment. (30) Finally it was of interest to investigate which halocarbons would reduce before reduction of the spin trap itself takes place e.g., CCI4, CHC13, CH2C12,CH3C1. The reduction of the spin trap gives a benzyl tert-butyl nitroxide in the presence of protons: 0-+He <>H-;H{---QHB)e + e~< HgN -0C(CH8)8

Superoxide Anion Reduction
We have recently initiated studies concerned with the reaction chemistry of superoxide and halocarbons (19,20). When fresh commercial potassium superoxide is exposed to CC14 containing PBN (powdered K02 is insoluble in CC14 but a surface reaction seems to occur), the spin adduct of the trichloromethyl radical is produced. This reaction is quite slow. However when the same method is used with BrCCl3, the reaction is violent with fresh potassium superoxide! We assume the reaction is simple electron transfer: 02 + CC14-* *CC13 + 02 + Cl O2 + BrCCi3 ,-CC13 + 02 + Br-(32) (33) The chemistry of trichloromethyl radicals with oxygen is not known. It is assumed that the combination of trichloromethyl radicals with oxygen is very fast and diffusion-controlled: *CC13 + 02 -C13COO- (34) What the rate and products of the reaction of trichloromethyl radicals with superoxide might be is not known: *CC13 + 02 V: C13COO- (35) The chemistry of the trichloromethylperoxyl radical or the trichloromethyl hydroperoxide or its anion is also a puzzle. In the fast reaction between BrCCI3 and potassium superoxide, and in the reaction of trichloromethyl radicals (produced in chloroform from chloroform and tert-butoxyl radicals) with oxygen in the presence of PBN, the spin adduct resulting from trapping chlorine atoms was detected along with the trichloromethyl adduct. Other signals were detected, perhaps due to the spin adduct of trichloromethylperoxyl or trichloromethoxyl radicals (20,21). We suggest the reactions mechanism [Eqs. (36) and (37)] to account for these results: 2Cl3COO -Cl3COOOOCCl3 -2Cl3CO-+ 02 0 C13. --* Cl-+ Cl--Cl by superoxide or dissociation to the trichloromethyl radical which combines rapidly with oxygen in the solvent cage to produce the trichlormethylperoxyl radical [Eq. (41)].
Since we have found that the trichloromethyl radical is easily detected by PBN we prefer a simple electrontransfer step.
It is of interest to speculate on a mechanism which predicts the stoichiometry and products reported by Sawyer. The scheme shown in Eqs. The proposal is completely analogous to known reactions of tert-butylperoxyl radicals which are believed to combine to the tetroxide before dissociating to tert-butoxyl radicals (22-24).
2(CH3)3COO-(CH3)3COOOOC(CH3)3 (38) (CH3)3COOOOC(CH3)3 --2(CH3)3CO + 02 (39) 0 (CH3)3CO -> CH3-+ CH3-C-CH3 (40) Superoxide can be made electrochemically by reduction of oxygen and subsequently reacted with the halocarbon (19,25,26). Sawyer and co-workers (5,25,26) have published some studies on compounds with the trichloromethyl group performed in this manner. End products and the stoichiometry of the reaction were given. In the case of carbon tetrachloride in DMF, the products detected were chloride, bicarbonate, oxygen, and hydrogen peroxide. However, since water was added to the final mixture on work-up, peroxyformates or percarbonates could have decomposed to the bicarbonate. No mechanism for the formation of these products has been offered other than that the first step is assumed to be electron transfer through the "chlorine atom complex" followed by a concerted replacement of chloride  Sawyer's stoichiometry is: CC14 + 50 + H20--4C1-+ 7/202 + HCO3-+ 1/2H202 Other possibilities are obvious, but the chemistry of the products expected is unknown; for example the peroxide of trichloromethane, and the trichloromethyl perester of chloroformate: All ofthese compounds react with superoxide to produce the appropriate ethylene: Aryl2CH-CCI3-Aryl2C=CC12 + HCI (58) Of biological relevance might be the suggestion that trichloromethylperoxyl radicals could produce chlorine atoms either by the route shown earlier or through reaction with lipid peroxyl radicals, (LOO) probably present in the system: C13COOOOL --C13CO0 + LO + 02 (57) Indeed, this may be the method whereby lipidoxyl radicals (LO0) are produced which are detected by spin trapping when carbon tetrachloride is metabolized (27).
The reaction of superoxide with a variety of chlorinated ethanes and ethylenes (DDT, methoxychlor, F-DDT) was also studied by Sawyer and co-workers (25,26,28). The ethylene reacts further when the aryl group is pchlorophenyl to produce the appropriate benzophenone (which is isolated) and presumably phosgene. Sawyer's stoichiometry (25,26) is given as by Eq. (59).
It should be emphasized that these reactions are slow, at least 100 x slower than for carbon tetrachloride and not likely to be important in the biological system in comparison to direct oxygenation by cytochrome P-450.

Oxygenation Reactions
It is quite commonly accepted that cytochrome P-450 are able to oxygenate a wide variety of compounds, including the iodine atom in iodobenzene (29).
We have considered the possibility that such a mechanism pertains in the case of carbon tetrachloride metabolism but no evidence is available to support this notion at this time (Sawyer found all of the chlorine in carbon tetrachloride converted to chloride by superoxide): Cl 0 Cl C f>"!_jCl_C{51 + C10. ?

Biochemistry and Halocarbon Radical Formation
In Vitro Investigation with Spin-Trapping Agents Although the method of spin-trapping has been used for some time in organic chemistry to investigate mechanism of radical reactions, the use of spin-trapping agents to investigate free-radical processes in biological systems is relatively new and has been pursued in only a few laboratories. The methods developed have focused primarily on identifying the radicals trapped when the biological system is exposed to foreign chemicals, e.g., carbon tetrachloride.
Quantitation of radicals produced is still a problem, because the number of radicals trapped is almost certainly less than the number of radicals actually formed. Many competing reactions are assumed to occur in the biological milieu, e.g., methylenic hydrogens and double bonds which are present in the membrane lipids can react with free radicals by abstraction and addition reactions, respectively. It is this feature of biological membranes which allows the process of lipid peroxidation to be initiated and propagated so easily.
However, notwithstanding the problem of quantitation valuable information about free radical reactions in biological systems can be obtained with the use of spintrapping agents. Since the integrated ESR signal intensity is directly proportional to the concentration of spin adduct obtained (30), the spin-trapping assay can be a useful and sensitive tool to probe not only the existence of radicals per se, but also determine whether a given manipulation of the system producing the radicals causes an increase or a decrease in radical production. In addition, the duration of radical production in a living system can be assessed by timed introduction of the spin-trapping agent into the system. Thus, if treatment of an animal with a certain substance initiates free radical formation in a target organ (carbon tetrachloride metabolism in the liver, for instance), the du-ration of radical production following the treatment can be estimated by administering the spin-trapping agent to a series of animals at different times following the treatment. Finally, the identity of radicals produced in biological systems can often be determined by the spintrapping procedure.
Thus far, the results which have been reported from the laboratories where spin-trapping agents are being used as probes for detecting radicals in biological systems have been encouraging (8,31). An early report of the detection of a radical in a biological system by spintrapping was made by Harbour and Bolton, who were investigating radical production in spinach chloroplasts exposed to light (32). In 1977, Saprin and Piette trapped a radical in a system composed of liver microsomes, iron, pyrophosphate (33). This radical appears to have been derived from the buffer, Tris, through a reaction of a reactive species (presumably a radical) with the Tris molecule (8). Nevertheless it demonstrated the feasibility of radical trapping in a system containing particulate biological material as well as the likely identification of the radical that was trapped. The following year, Lai and Piette reported that they were able to trap hydroxyl radicals (HO ) in a system containing microsomes, NADPH, and an EDTA-Fe2 chelate, in which the microsomal membrane was undergoing lipid peroxidation (34). Although the spin-trapping agent used, DMPO, inhibited lipid peroxidation in this system, the relationship of the production of the hydroxyl radicals to the process of lipid peroxidation in this reaction system was not clear. During this period, at least two different free radicals were demonstrated to be generated when a nitrosamine was incubated with liver microsomes (35).
The first ESR data demonstrating that trichloromethyl radicals are produced in liver microsomes during carbon tetrachloride metabolism was reported in 1978 by Poyer et al. (36) who used the PBN. The radical adduct obtained exhibited a spectrum of a triplet of doublets (aN = 14.1, d: = 1.8 G) (36). Although these results were questioned at first (37), the radicals were later unequivocally identified as trichloromethyl adducts of PBN by use of 13C-labeled carbon tetrachloride (38). The strategy for using 13CC14 lay in the fact that in the ESR spectrum of the spin adduct obtained with '3CCI3 radicals a characteristic doubling of each line occurs because the nuclear spin of "3Cis 1/2. Since the line doubling does not occur when 12CC14 is being metabolized (Fig. 11), it is clear that the radical responsible for this effect is trichloromethyl.
These results have now been confirmed by other laboratories (39,40). It was also demonstrated that the trichloromethyl radicals apparently are generated in rat liver microsomes by only one form of cytochrome P-450 (41,42). This form of cytochrome P-450 has a minimum molecular weight of 52,000 daltons and is inducible by phenobarbital. This particular cytochrome is the first polypeptide component of rat liver microsomes to show alternation following exposure of rats to carbon tetrachloride (41). Following treatment of rats with CCl4, the amount of 52,000-dalton cytochrome decreases very (Bottom) ESR spectrum of the "3C-labeled trichloromethyl (130CC13) spin adduct of PBN, aN = 13.9, a'< = 1.5, a'-3 = 9.5, acy" = 0.23 G. The above six-line spectrum is split into a twelveline spectrum because the nuclear spin of 13C is one-half. This verifies that in both cases that spin-trapped radical is indeed trichloromethyl (13CC13). rapidly and essentially disappears within 1 hr. Eventually, other cytochromes decrease also, but this appears to be due to secondary damage to the microsomal membrane caused by lipid peroxidation which is initiated by the formation of the trichloromethyl radicals. It has been well established that prior induction of rats with phenobarbital markedly enhances the toxicity of a given dose of carbon tetrachloride. In theory, at least, increasing the level of this cytochrome should facilitate the production of a greater number of -CCl3 radicals du,ring carbon tetrachloride metabolism and may account for the increased toxicity of this halocarbon in the induced state. Further evidence for the involvement of this particular cytochrome P-450 (52 k dalton) component in the generation of -CC13 radicals is the observation that metyrapone (2-methyl-1,2-di-3-pyridyl-1-propanone) inhibits the amount of trapped radicals up to 90% when the animals are treated with this pyridine derivative (43). Metyrapone is known to inhibit phenobarbital-induced drug metabolism involving the same form of cytochrome c (44). The radicals which are trapped are all located within the microsomal membrane. This was shown as follows. Rat liver microsomes were allowed to metabolize carbon tetrachloride in the presence of a spin trapping agent (PBN). The reaction system was then subjected to centrifugation at 100,000 g to form a firm pellet. The supernatant layer was separated from the pellet and both fractions were assayed for the trichloromethyl radical adduct signal. All of the detectable signal was contained in the pellet. None has been observed in the soluble fraction. Since the radicals are formed in the microsomes, they are apparently trapped there and remain in the membrane, perhaps at or near the site of their formation.
It has now become apparent that when carbon tetrachloride is being metabolized by rat liver microsomes, a sequence of different radicals are trapped as a function of time (27). This sequence is shown in Figure 12. The spectra shown in Figure 12 clearly indicate that a progression of radical trapping has occurred during a 2-hr. period. Initially, the majority of the radicals being trapped are probably lipid oxy radicals (LO-). The factors responsible for this observation are believed to be as follows. When carbon tetrachloride enters the microsomes, it is reductively cleaved by the mixed function oxidase system to trichloromethyl radicals and chloride anions. These promptly react with available oxygen in the membrane at an essentially diffusion-limited rate to form trichloromethyl peroxy radicals. Some of the -CCIl radicals as well as some of the peroxy radicals will react with membrane lipids. In this event, the radicals may either abstract methylenic hydrogens from polyunsaturated fatty acid groups in the phospholipids, or they may add directly to double bonds to form covalentlylinked derivatives. sec). Thus, the spectrum obtained during the initial minutes of the reaction is primarily that of the spin-trapped LOradicals (see the 6-min. line of Fig. 12). When oxygen is depleted, L-radicals can partake significantly in the slower reaction with PBN: In either case, the reaction results in the formation of a lipid radical (L*) which also reacts with oxygen resulting eventually in the formation of lipid oxy radicals. L +O., 2---LO (69) The reaction rate for 'CC13 with unsaturated lipid structures is approximately 3 x 106/M-sec. The net result is that initially, essentially all the CC13 radicals are converted to OOCC13. These can react both with membrane lipids and with the spin trap (PBN). If the latter forms an adduct, however, it is unstable and does not persist long enough for detection by ESR spectroscopy. The LOradicals also react with PBN, and in this case relatively stable adducts are formed (the reaction rate has been estimated to be approximately 5.5 x 106/M- (70) After 21 min, the signal intensity of the lipid oxy radical (LO-) begins to diminish because the PBN adduct of that radical, while relatively stable in a chloroform extract, is not stable in the milieu of the microsomal membrane. The signal of the PBN adduct of *CCI3 becomes the dominant feature of the spectrum within 36 min (see Figure 12). If this reaction is performed under conditions where most of the oxygen in the system has been removed, the formation of LOradicals becomes limited and most of the radicals which are trapped in the first 15 min of the reaction are L (Fig. 13). Note that the third peak in the spectrum has a distinct shoulder on the right side indicated by the inverted delta. The splitting constants of the peaks marked with the inverted delta throughout the spectrum are: aN = 13.5, a'?j = 2.0 G. These values, together with computer simulation analyses, indicate that this portion of the spectrum is due to an oxygen-centered lipid radical. The spectral peaks marked with daggers have the following splitting constants: aN = 14.5, C = 3.25 G and have also been shown by similar analyses to be due to a carbon-centered lipid radical. Thus, both LO* and L-spin adducts are clearly present at this time as well as the *CC13 spin adduct. The latter quickly becomes the dominant feature of the spectrum. Presumably, this is due to reduction of the LO and Lradical adducts while formation of the much more stable -CCI3 adducts continue to exceed the rate of reduction for at least another 30 Figure 13A.
From these experiments, and from what is known from the work of others, several deductions are supported. In any system where -CC13 radicals are being formed, they will react preferentially with 02 as long as the latter is available to form Cl3COO. Both of these radicals will attack lipids and proteins in the vicinity of the enzymic production of the trichloromethyl radicals resulting in the loss of cytochrome P-450 activity and initiation of lipid peroxidation. The latter process also causes destruction of cytochrome P-450 which potentiates the overall loss of the total microsomal cytochrome content. Significant trapping of the -CCI3 radicals by PBN will only begin when the availability of both oxygen and membrane lipids becomes limiting in the domain of radical production. Trapping of the -CCI3 radicals then appears to continue until the process of lipid peroxidation and direct radical attack on the 52 k dalton cytochrome destroys remaining activity.
The oxygen tension in biological systems may be critical in determining the degree of toxicity resulting from exposure to halogenated hydrocarbons. In vitro studies have shown that the partial pressure of oxygen in the system can determine the extent of radical damage to membranous structures in which the halogenated hydrocarbon is being metabolized (45). When liver microsomes were incubated with NADPH and halothane, membrane damage, (determined by the extent of membrane lipid peroxidation in the system, measured by malondialdehyde formation) was observed to increase with the partial pressure of 02-Malondialdehyde formation increased until the partial pressure reached about 3 mm Hg. But when the pressure was increased above 3 mm Hg, the extent of peroxidation decreased (45). The authors interpreted the results as indicating that the oxygen tension must be low enough to permit reductive metabolism of the halocarbon, but not so low as to make the rate of lipid peroxidation insignificant.

In Vivo Investigations with Spin-Trapping Agents
Steady-state oxygen tension in liver tissue, a metabolically active tissue, is apparently low enough to permit lipid peroxidation to occur during the metabolism of carbon tetrachloride in vivo. That lipid peroxidation does occur in vivo as a consequence of CCl4 metabolism has been considered likely for a number of years (46). Direct evidence has now been obtained for the formation of lipid radicals in intact animals exposed to CC14 through the use of spin-trapping agents (27). When '3C-labeled CC14 was orally administered to rats along with PBN, detectable levels of trichloromethyl radicals were observed to form in the hepatic endoplasmic reticulum within 15 min (47). However, no other types of radicals were detected with that particular spin trap even though both lipid and lipid oxy radicals were trapped with PBN in the in vivo investigations described above. However, when a modified fonn of the spin trap (trimethoxy phenyltert-butyl nitrone (MO3PBN) was administered to rats along with 13CC14, lipid extracts of livers removed from the rats displayed a strong electron spin resonance signal which was identified as a carbon-centered radical adduct of the spin trap (27). The adduct which was recovered was as a modified form of MO3PBN, since one of the methoxy groups is demethylated during the events leading to the isolation of the spin adduct (27). The carbon-centered radical is lipid-soluble, *nd the char-acteristics ofits ESR spectrum indicate that the trapped radical is a large molecule, probably a phospholipid. At present there is no explanation why PBN should preferentially trap CCI3 radicals while MO3PBN appears to trap carbon-centered lipid radicals preferentially during CCl4 metabolism in vivo. The most likely explanation is that the positioning of the particular spin trap in the lipid domain of the endoplasmic reticulum determines its accessibility to the site of -CCI3 formation at the cytochrome P-450 which generates these radicals. The structure, lipid solubility, and reaction rates with particular lipids of the various spin trapping agents must all interact to determine which radicals will be trapped during the sequence of events initiated by the reductive cleavage of carbon tetrachloride to the trichloromethyl radical. Thus MO3PBN may penetrate the cytochrome P-450 domain less well than PBN, and, as a result, be unable to compete with lipids in that domain for -CCl3 radicals. The propagation of lipid peroxidation away from that domain may then provide ample opportunity for MO,PBN to trap lipid radicals. Whatever the case of the selectivity, the results with MO3PBN confirm that initiation of lipid peroxidation occurs in the hepatic endoplasmic reticulum as consequence of carbon tetrachloride metabolism. Since lipid peroxidation is thought to be involved in other types of diseases and toxic responses to environmental factors (48,49), this particular radical probe may be useful in determining the extent to which peroxidation may be involved.

Spin-Trapping Halothane Radicals
The halogenated hydrocarbon anesthetic, halothane (2-bromo-2-chloro-1,1,1-trifluoroethane), used in surgical and other clinical procedures, might be expected to be metabolized in the endoplasmic reticulum of the liver to a free radical product as is CC14 (38). Numerous attempts have been made to determine a cause-andeffect relationship between halothane anesthesia and "halothane hepatitis," since reports of postoperative halothane-induced hepatitis began appearing in 1958, the year that the anesthetic was released for clinical use (50).
When a rat was given halothane in breathing air for 2 hr immediately following oral administration of PBN, a lipid extract of the liver presented a strong ESR signal as shown in Figure 14A (51). This spectrum was found to depend on the prior exposure of the animal to halothane. Administration of the spin-trapping agent alone yielded no signal in the lipid extract. When halothane was incubated with PBN, rat liver microsomes, and a NADPH-generating system, an EPR signal also was observed (Fig. 14B) which had the same characteristics as that obtained with the extract from the livers of animals administered halothane in vivo. The finding of the spin-trapping of a halothane radical in vivo (52,53) and in vitro (54,55) has been confirmed by others. The halothane free radical formed in the liver endoplasmic reticulum is probably the 1-chloro-2,2,2-trifluoroethyl radical. Trudell et al. have shown that, when halothane was incubated with vesicles of dioleoylphosphatidylcholine and egg phosphatidylethanolamine containing cytochrome P-450-HA-2, cytochrome P-450 reductase, cytochrome b5, and NADPH, under argon, a dioleoylphosphatidylcholine product with a 1-chloro-2,2,2-trifluoroethyl group bound to either the 9or 10carbon atom of the fatty acid was formed (56). Gas chromatography-mass spectroscopy techniques were used to assign the 1-chloro-2,2,2-trifluoroethyl adduct (56). The halothane radical which is formed in vivo appears to be identical to that formed by the metabolism of halothane by liver microsomes and is presumably a product of the drug-metabolizing system in the liver of the intact animal. This is consistent with the observation that like in vivo experiments done with CC14, nearly all of the spin-trapped radical is found in the microsomal fraction of the liver of animals administered halothane and PBN in vivo (Fig. 14C) (51). Also, as is the case for CC14 which forms 0CC13 radicals, metyrapone inhibited the formation of halothane radicals up to 90% when animals were treated with metyrapone prior to exposure to halothane (43). This indicates that the same phenobarbital-inducible cytochrome P-450 (44) also generates the halothane radicals from halothane. About 0.4 pmole of halothane radicals per gram of liver are trapped in the in vivo experiment (51), and this is about one-fourth the quantity of radicals compared to the amount trapped when the rats are administered an equivalent dose of CC14 (57). This lesser formation of halothane radicals, together with the fact that the reactivity of the 1-chloro-2,2,2-trifluoroperoxyl radical (forned from the reaction of the halothane radical with oxygen) with fatty acids is only one-fifth that of the trichloromethylperoxyl radical (58) probably explains halothane being much less hepatotoxic than CC14.
Although no relationship between the formation of this radical and the etiology of halothane hepatitis has been established, the similarity of this reaction to that of CC14 in the liver suggests that, under some undefined conditions, a halothane radical associated with halothane metabolism may be toxic.
Differences between Rat, Chicken, and Turkey When chicken liver microsomes were treated with CC14 in presence of PBN and an NADPH-generating system, only a weak, indistinct ESR signal was observed as compared to rat liver microsomes. Also, if the chicken was administered CC14 and PBN in vivo in the same manner as described for the in vivo rat studies, no appreciable ESR signal could be observed in a total lipid extract from the liver. These results are in accordance with the low levels of CC14-catalyzed lipid peroxidation observed in chicken liver microsomes (59), and the resistance of the chicken to CC14 liver damage (60). However, when turkey liver microsomes were incubated with PBN, CCl4, and an NADPH-generating system, as was done for the chicken, a strong ESR signal was observed. This indicates a fundamental difference in the capability of the drug metabolizing system of these animals to metabolize halocarbons to highly reactive radicals.
Rat liver microsomes, when incubated with PBN, CHBr3, and an NADPH-generating system as in the case of CC14, produced an ESR signal (aN = 14.6, aCP = 2.2 G) which may be the PBN-trapped -CHBr2 radical. When chicken or turkey microsomes were incubated with PBN, CHBr3 and an NADPH-generating system, ESR signals were observed for both cases. These results contrast markedly with the observation that the chicken does not metabolize CC14 to the *CC13 radical, while the turkey does.
In Vitro Studies with Rat Liver Microsomes and Sodium Dithionite Sodium dithionite (Na2S204) has been used as a nonenzymic reducing agent in the study of the spectral binding characteristics of halogenated hydrocarbons to reduced cytochrome P-450 (61) and in the dehalogenation of several halocarbons by liver microsomes (62,63). During studies involving the formation and detection of trichloromethyl free radicals formed during the metabolism of CC14, it was found that a rat liver microsomal incubation system utilizing sodium dithionite as an electron donor for cytochrome P-450, in place of the enzymically active NADPH + NADPH cytochrome P-450 reductase, was very effective in producing CC13 radicals (64).
Trapped radicals were observed when CC14, CHC13, CH2Cl2, CHBr3, CH2Br2, 1,1,1-trichloroethane, hexachloroethane, 1,l,1-trichloro-2,2,2-fluoroethane, the Freon CC13F, DDT, and the anesthetic halothane were used as substrates. Studies with 13CC14 and CC13F (Freon 11) indicate that the free-radical species trapped by the PBN are halocarbon radicals and not some other free radicals derived from the incubation system. Although only the phenobarbital-inducible "C" form of cytochrome P-450 (65) is believed to be the enzymically active form metabolizing halogenated hydrocarbons to free-radical products (42,43), both the purified "A" and "B" forms (65) were found to be capable of forming trichloromethyl free radicals in this system. That is, when dithionite was used as reducing agent, the microsomal system did not dependent on any enzymic activity, since heat-denatured microsomes could function as well as undenatured microsomes. Also, when microsomes were treated with 0.2% sodium dodecyl sulfate (SDS) to convert the endogenous cytochrome P-450 to cytochrome P-450 (observed spectrophotometrically), they were still capable of forming the 'CC13 radical from CCl4 and dithionite. It is presumed that the sodium dithionite reduces the cytochrome P-450 to the ferrous form and that the ferrous form donates an electron to the halogenated hydrocarbon to form a free radical and a halide anion: Cyt. P-450(Fe2+) + R-X --Cyt. P-450(Fe 3) + R-+ X- (71) This nonenzymic system may be of value in determining the capability of halogenated hydrocarbons to be metabolized to free radical products in the liver. The metabolic conversion of CC14 to the trichloromethyl radical is thought to occur by an electron transfer (donation) from the ferrous form of cytochrome P-450 directly to the halocarbon to form the -CCI3 radical and a chloride ion. Therefore, it was of interest to determine what other electron-transfer (donation) reactions could produce -CCI3 radicals from CC14. It was found that aqueous solutions of thallous or ferrous ions, ferrous or cuprous ions chelated with 2,2-dipyridyl, Mn2+ protoporphyrin IX, and the one-electron reduced freeradical form of methyl viologen can all donate electrons to CC14 to form -CC13 radicals, observed after spin-trapping with PBN. Several mechanisms for the metabolic formation of free radicals and other products from halogenated hydrocarbons have been proposed and are given in a review by McDonald (66).
Electron transfer from a hemoprotein, such as cytochrome P-450, to a halocarbon probably takes place through an "inner sphere" electron transfer process (67,68), by which the halocarbon becomes axially bonded to the Fe2+ of the hemoprotein and accepts an electron from the Fe2+ of the heme to break a halogen-carbon bond and form a carbon-centered free radical and a halide anion. The half-wave reduction potential of the halocarbon would have to be sufficiently positive for the halocarbon to accept an electron from the hemoprotein (69)(70)(71).
For the electron donation process to take place, the axial ligands of the hemoprotein would have to be accessible to the halocarbon and capable of being displaced by the halocarbon, so that the inner sphere axial electron transfer could take place. In keeping with the table: by Castro (72) these requirements are apparently not met by cytochrome c, and it does not donate electrons in the reduced Fe2" state to CCI4 to form CCI3 radicals.
However, as predicted by the same table, hemoglobin, myoglobin, carbon monoxymyoglobin, catalase and horse radish peroxidase, all in the Fe2" state, will all donate electrons to CCI4 to form CCI3 radicals from CCI4.
Since several hemoproteins will form free radical products from halogenated hydrocarbons, it is implied that organelles containing these hemoproteins would be susceptible to some injury when exposed to some halogenated hydrocarbons, although not as severe as that occurring in the liver of animals exposed to CCl4.
Research described in this paper is being supported by the Natural Sciences and Engineering Research Council of Canada and National Jnstitute of Health Grant numbers ES02812 and ES03067.
In addition the authors are very grateful to Uwe Oehler, who provided computer drawn figures, and to Larry Haire who proofed the final manuscript.