CCl4 has been shown previously to be metabolized to the trichloromethyl radical (.CCl3) and to a novel oxygen-containing carbon dioxide anion radical (.CO2-) in the perfused rat liver and in vivo. Since the role of free radicals in CCl4-induced hepatotoxicity is unclear, these studies were designed to determine if a relationship between .CO2- formation and halocarbon-induced hepatotoxicity exists. CCl4 or bromotrichloromethane (CBrCl3) was infused into livers from control or phenobarbital-treated rats perfused with either nitrogen- or oxygen-saturated Krebs-Henseleit bicarbonate buffer. Samples of effluent perfusate and chloroform/methanol extracts of liver were analyzed by ESR spectroscopy for free radical adducts following infusion of halocarbon and the spin trap, phenyl-t-butylnitrone (PBN). Hyperfine coupling constants and 13C-isotope effects observed in the ESR spectra of organic extracts of liver demonstrated the presence of the PBN radical adduct of .CCl3 from both halocarbons. Radical adducts in aqueous extracts of liver and effluent perfusate had hyperfine coupling constants and 13C-isotope effects identical to those of PBN/.CO2- generated chemically from formate. The PBN/.CO2- radical adduct was also observed in urine following the intragastric administration of CBrCl3 and PBN. Detection of PBN/.CO2- adducts in the effluent perfusate was decreased 3- to 4-fold by DIDS (0.2 mM), an inhibitor of the plasma membrane anion transport system. The rate of formation of PBN/.CO2- was decreased 2- to 3-fold following inhibition of cytochrome P-450-dependent monooxygenases by metyrapone (0.5 mM) and was increased about 2-fold by induction of cytochrome P-450 by phenobarbital pretreatment. Toxicity of halocarbons in the perfused liver was assessed by measuring the release of lactate dehydrogenase (LDH) into the effluent perfusate in livers from phenobarbital-treated rats under conditions identical to those employed to detect radical adducts (i.e., during the infusion of CCl4 or CBrCl3 into livers perfused with either nitrogen- or oxygen-saturated perfusate). Under all conditions studied, PBN/.CO2- was detected in the effluent perfusate within 2-4 min. Metabolism of halocarbons to PBN/.CO2- was 6- to 8-fold faster during perfusion with nitrogen-saturated rather than with oxygen-saturated perfusate. Concomitantly, liver damage detected from LDH release occurred much sooner during halocarbon infusion in the presence of nitrogen-saturated rather than oxygen-saturated perfusate. A good correlation between the rate of formation of PBN/.CO2- and the time of onset of LDH release following halocarbon infusion was observed.(ABSTRACT TRUNCATED AT 400 WORDS)