In Vivo Detection of Free Radical Metabolites by Spin Trapping

BY R. P. MASON, K. R. MAPLES AND K. T. KNECHT

1 Introduction

1.1 The Problem.-In vivo detection of free radical metabolites is a very challenging task that has only recently been undertaken. One reason for the late development of this area is that most biochemicals, as opposed to drugs and industrial chemicals, are not easily metabolized through free radical intermediates. In addition, detection of something as ephemeral as a free radical inside a whole animal is inherently not easy. Production rates of free radicals in animals are slow in comparison to chemical systems; therefore, the highest possible sensitivity is of paramount importance. water, with its high dielectric constant, is the worst solvent for ESR spectroscopy in that only very small samples can be studied. This decreases the molar sensitivity of biological samples just when sensitivity is needed most. However, unless free radical metabolites can be demonstrated with a whole animal, there will always be some question as to their actual existence in biology.

1.2 Approaches.-Over the last thirty years, several approaches have been tried to circumvent the problem of working with aqueous samples. Freezing water lowers its dielectric constant so that larger samples can be studied. The freeze quench technique is useful for enzymes where solutions can be frozen in milliseconds. Frozen tissues, however, must be ground to fit into ESR sample tubes, and this leads to mechanically induced radicals or artifacts. In addition, the resulting powder spectra are poorly resolved, and their interpretation in complex biological systems is very difficult, if not impossible. Lyophilized, or freeze-dried tissue is plagued by the same problems of artifacts and poor resolution. Low-frequency ESR enables the study of larger samples, and perhaps even small animals could be studied directly, that is, with in vivo spectroscopy. Unfortunately, sensitivity is strongly dependent on frequency, and low-frequency instruments are unlikely to achieve the molar sensitivity of the standard X-band instruments. Spin trapping, in that it ideally integrates free radicals formed over time, appears to be the most attractive approach to the detection of free radicals in vivo. Since the concentration of naturally occurring radicals in body tissues is generally near the sensitivity limit of ESR spectroscopy, the spintrapping technique is not limited by background signals.

2 Spin Trapping In Vivo

2.1 Spin Trapping.-The technique of spin trapping involves the addition of a reactive, primary free radical across the double bond of a diamagnetic compound, the spin trap, to form a more persistent, secondary free radical, the radical adduct. This technique allows the indirect detection of primary free radicals that cannot be directly observed by conventional ESR due to low steady-state concentrations or to very short radical relaxation times, which lead to very broad lines.

To date, all in vivo spin trapping investigations have used the nitrone spin traps, phenyl-tert-butylnitrone (PBN), α-2,4,6-trimethoxy-PBN ((MeO)3PBN) and 5,5-dimethyl-l-pyrroline N-oxide (DMPO). In most cases, radical adducts of nitrone spin traps produce six-line ESR spectra. The hyperfine splittings arise from the nitrogen and β-hydrogen of the spin trap rather than from atoms of the primary radical. Identification of the trapped radical species, therefore, depends on a careful comparison of hyperfine splitting values with those of reference nitroxides analyzed under exactly the same experimental conditions. If the radical is trapped at a nucleus with nonzero spin such as 14N, then additional hyperfine splittings occur which greatly facilitate the identification of the radical adduct. A very useful approach for the radical adducts of C-and a-centered free radicals is isotopic substitution at this center in the radical precursor with 13C or 17O, respectively.

2.2 Difficulties.-Production of a radical adduct stable enough to be detected in biological samples is a major difficulty, but other factors must also be considered. When using the spin-trapping technique in whole animals, spin traps may interfere with the experimental system by inhibiting or stimulating enzymes, or by producing toxicity. The latter possibility has not seemed to affect in vivo work to date, although this issue has not been directly addressed in detail. There are several good reviews of spin trapping, which address in vitro applications of this technique. For discussions of spin traps as enzyme inhibitors or enzyme substrates, or of more general problems of spin trapping such as artifacts, the reader is referred to these reviews.

3 Folch Extraction

3.1 Halogenated Carbon-centered Radicals.-The first in vivo experiment was reported in 1979 when the spin trap PBN and carbon tetrachloride were given to a rat through a stomach tube. After 2 hours the liver was extracted with a mixture of methanol and chloroform (the Felch extraction for lipids), and the ESR spectrum of the chloroform layer was taken. The spectrum was assigned by comparing experimental hyperfine coupling constants with those of the PBN/CCl3 radical adduct generated in a microsomal system or by photolysis of carbon tetrachloride. More definitive identification of this species as the PBN/·CCl3 radical adduct was later made using 13C carbon tetrachloride, which produces an additional splitting in the radical adduct spectrum.

Carbon-centered radical adducts from other halogenated hydro-carbons have also been detected in the organic extracts of livers from treated animals. Of clinical interest have been studies with the volatile anesthetic halothane, which produces hepatitis in humans. Under hypoxic conditions, halothane produced both liver damage in phenobarbital-pretreated rats and free radicals, which were trapped by PBN and extracted from the liver. The trapped radical species has not yet been identified unambiguously, but reductive debromination is probably responsible for the reported carbon-centered free radical formation. Chloroform, iodoform, bromoform, and bromodichloromethane are other compounds that are converted to free radicals in vivo. Administration of carbon tetrachloride and PBN to gerbils results in detectable free radical adducts in Felch extracts of liver, kidney, heart, lung, testis, brain, and blood, with signal intensities of spectra decreasing in the order given. In Felch extracts of liver, PBN/·CCl3 was identified, but no assignment of radical identity in other tissues could be made.

3.2 Lipid-derived Radicals.-Halocarbon free radical formation results in lipid peroxidation, and it is logical to expect that the spin adducts detected after administration of carbon tetrachloride, which do not exhibit 13C hyperfine coupling from 13CCl4, are derived from carbon-or oxygen-centered lipid radicals. Such lipid-derived radical adducts have been detected in vitro by several investigators. The extracted livers of rats treated with (CH3O)3PBN and carbon tetrachloride yield a carbon tetrachloride-dependent, non-carbon tetrachloride-derived species assigned to a lipid-derived radical. However, 0-demethylation of the spin trap precludes a simple explanation of these latter studies. Definitive identification of lipid radicals is inherently difficult, since 13C-labelled fatty acids are unavailable and chromatography or mass spectroscopy of these heterogeneous radical adducts is a formidable task.

Lipid peroxidation is itself an important toxicological process, and lipid radical adducts may be the only evidence of initiating radical species that do not form stable adducts themselves. For instance, 3-methylindole is metabolized in vitro to form a PEN/nitrogen-centered free radical. In vivo, however, Folch extracts of lungs from goats treated with 3-methylindole and PBN yielded only a carbon-centered lipid radical adduct. Radical adduct detection varied inversely with in vivo levels of GSH, as was suggested with experiments with the GSH precursor cysteine and the GSH-depleting agent diethylmaleate. In a similar manner, carbon-centered lipid-derived radical adducts have been extracted from the brain, spleen, liver, and heart, but not the kidney, of rats dosed with PBN and exposed to a non-lethal burst of gamma-irradiation. An ethanol and high fat diet has also been used to produce the (CH3O)3PBN/lipid radical adduct in Folch extracts of hearts and livers from animals treated in vivo.

3.3 Advantages and Limitations.-In all of the in vivo investigations of carbon-centered free radicals from halocarbons and lipids, the Folch (1:2 methanol:chloroform) extraction was used. The greatest advantage of the Folch extraction is that the radical adduct is removed from the high dielectric biological tissues, enabling the use of larger sample volumes. In addition, the sample can be easily concentrated by evaporation of the chloroform layer. The greatest limitation of organic extraction is that only nonpolar radical adducts such as those of ·CCl3 and lipid-derived radicals can be detected.

4 Biological Fluids

4.1 New Approach.-Our approach to in vivo spin trapping has been to examine biological fluids directly for spin adducts using the TM110 ESR cavity and a 17 millimeter flat cell which gives the largest possible aqueous sample size in the active region of the cavity, about 100 µl. This approach gives us high molar sensitivity and high resolution. Oxygen solubility in water is only about 250 µM, so degassing of samples is not usually necessary, but sometimes will narrow the sharpest lines. No background signals other than the ascorbate semidione doublet and Mn2+ (only in bile) have been detected. The biological fluids are not extracted, and the difficult question of what happens during extraction other than a physical separation is avoided. The detection of free radical metabolites in urine, blood or bile is little different from the detection of the products of drug metabolism by HPLC as practiced in pharmacology departments or by the pharmaceutical industry.

4.2 Urine.-When this new approach was used with rats administered carbon tetrachloride and PBN, a novel radical adduct, PBN/'CO2-, was detected in the urine of living rats treated with carbon tetrachloride and PBN. Use of 13c carbon tetrachloride proved that this radical adduct, like PBN/'CCl3, was carbon tetrachloride-derived. PBN/·CO2- was also detected in the effluent perfusate of livers which were perfused with carbon tetrachloride and PBN, and in the urine of rats after administration of bromotrichloromethane and PBN. Bromotrichloromethane is metabolized by the same pathway and to the same metabolites as carbon tetrachloride, but is more readily dehalogenated due to the relative weakness of the C-Br bond.

4.3 Perfusate.-Hepatocellular necrosis in perfused liver, as measured by LDH release from lysed cells, occurs after the infusion of CCl4 or CBrcCl3 and follows the appearance of PBN/'CO2-. Perfusion of the liver with nitrogen-saturated instead of oxygen-saturated buffer accelerates this LDH release. The concentration of PBN/·CO2- in perfusate at the beginning of lysis is statistically correlated with the amount of time required until LDH is detected. The higher the radical concentration the shorter the time to lysis. Correlation does not imply causation; therefore, the stable trapped product PBN/·CO2- can only be considered a marker for some more reactive species actually responsible for membrane damage. Nevertheless, the detection of a radical adduct does imply production of reactive free radical metabolites in vivo, and the correlation of radical adduct production with an index of toxicity is consistent with a free radical-mediated pathology. In these experiments, PBN (5 mM) did not appear to prevent membrane-damaging free radical reactions from occurring, but this can be explained by the very low rates of radical trapping characteristic of PBN.

The U.S. Food and Drug Administration and Environmental Protection Agency require companies to identify the urinary metabolites derived from drugs and pesticides. As the above studies suggest, such requirements may be extended to the spin-trapped products of free radical metabolism as well. Unlike the stable product of detoxification, which can be detected by conventional analytical techniques, the detection of radical adducts proves the formation of highly reactive intermediates. Free radicals can clearly react as easily with cellular constituents as with spin traps and thus could cause damage in vivo.

4.4 Bile.-Both PBN/·CCl3 and PBN/·CO2- were detected in bile samples collected at multiple timepoints after treatment of living rats with PBN intraperitoneally and carbon tetrachloride intragastrically. In vivo manipulations such as low oxygen tension of inspired air or phenobarbital pretreatment, known to increase the toxicity of carbon tetrachloride, also produce qualitative and quantitative changes in the ESR signals detected. Either hypoxia or phenobarbital induction was required for the detection of PBN/'CO2-. Both treatments also increased the biliary concentration of PBN/·CCl3. In principle, ionic, polar, and nonpolar radical adducts can be detected in bile, because, in addition to the aqueous phase, the biliary micelles provide a hydrophobic environment.

4.5 Blood.-The reaction of oxyhemoglobin with phenylhydrazine and hydrazine-based drugs within red blood cells induces a series of processes which leads to destruction of the cell and results in hemolytic anemia. Considerable evidence obtained from in vitro ESR investigations implicates free radicals in the events contributing to red blood cell hemolysis.

An immobilized radical adduct (aNzz = 31.8 G and aNzz = 9.5 G) is formed in the blood of rats which received an intraperitoneal injection of DMPO followed by an intragastric dose of phenylhydrazine. This immobilized radical adduct is detected when phenylhydrazine is administered at a dose of only 1 mg/kg. Hydrazine itself gives a weaker spectrum of the same species. The immobilized radical adduct co-chromatographs with oxyhemoglobin and can be detected in vitro using purified rat hemoglobin, phenylhydrazine, and DMPO. The sulfhydryl reagents, iodoacetamide, maleimide, and N-ethylmaleimide all inhibit phenylhydrazine-dependent radical adduct formation when whole rat blood is treated in vitro. This sulfhydryl-dependent radical adduct has been assigned to a DMPO/hemoglobin thiyl radical adduct. This is the first report of free radical formation from a biological macromolecule formed as a consequence of free radical metabolism. In addition, PBN could replace DMPO in vivo to yield the PEN/hemoglobin thiyl radical adduct, aNzz = 30.8 G. The DMPO/phenyl radical adduct was also detected in chloroform extracts of whole blood in accord with in vitro results. In subsequent work, the DMPO/hemoglobin thiyl radical adduct was detected in the blood of rats following the administration of some hydrazine-based drugs. The drugs examined were hydralazine, iproniazid, isoniazid, and phenelzine. Of the four drugs, only iproniazid and phenelzine were able to induce DMPO/hemoglobin thiyl radical adduct formation in vivo, whereas only hydralazine and phenelzine were able to form this adduct in vitro. The in vivo iproniazid-induced radical adduct formation was decreased by pretreating the rats with bis-para-nitrophenylphosphate, an arylamidase inhibitor. These results support the argument that iproniazid is hydrolyzed in the liver to a more reactive metabolite, most likely isopropyl hydrazine, which is subsequently released into the blood stream. In contrast, phenylhydrazine and phenelzine react directly with red blood cells to yield the DMPO/hemoglobin thiyl radical adduct. As hydralazine did not yield this adduct in vivo, we proposed that hydralazine is metabolized in vivo into a less reactive compound, possibly via acetylation. In summary, a DMPO/hemoglobin thiyl radical adduct has been detected in vivo. This species is formed by the reaction of phenylhydrazin, and some hydrazine-based drugs with oxyhemoglobin.