Among the main DNA-reactive metabolites of 1,3-butadiene (BD), both 1,2:3,4-butadiene diepoxide (BDE) and 1,2-epoxy-3-butene (BME) have been reported in mice and rats exposed to BD, but blood and tissue levels of these metabolites are much higher in mice than in rats under similar exposure conditions. BDE, being more reactive and genotoxic than BME, is thought to be responsible for the greater susceptibility of mice to BD carcinogenicity. While BDE is a DNA-alkylating agent and some BDE adducts have been characterized, no sufficiently sensitive method has been reported for studying BDE-DNA binding in vivo. In the present investigation, a modified dinucleotide/monophosphate version of the 32P-postlabeling assay was applied to detect BDE-DNA adducts, which were prepared by reacting BDE with calf thymus DNA or deoxyribooligonucleotides [(AC)10, (AG)10, (CCT)7 and (GGT)7] in vitro or with skin DNA of mice in vivo upon topical treatment. Optimal resolution by 2-D PEI-cellulose TLC of the highly polar 5'-monophosphate adducts was achieved at +4 degrees C using 0.3 M LiCI (DI) and 0.4 M NaCl, 0.04 M H3BO3, pH 7.6 (D2). The profiles of the 32P-postlabeled adducts were similar for calf thymus and skin DNA, with 3 major spots being detected. Adducts obtained in in vitro and in vivo experiments were compared by re- and cochromatography in 4 or 5 different solvents, and these experiments provided evidence that corresponding BDE adducts, for the most part, were identical and represented adenine derivatives. Guanine adducts were not detected by this method although literature data indicate their formation. Quantitatively, the assay responded linearly to adduct concentration, as shown in an experiment where BDE-modified skin DNA was serially diluted up to 81-fold with control DNA. The limit of detection was approximately 1 adduct in 10(8) normal nucleotides. Further, in an in vivo dosimetry study, skin DNA from groups of 8 individual mice treated with different doses of BDE (1.9, 5.7, 17, 51 and 153 mumol/mouse) for 3 days exhibited a linear relationship (r > or = 0.992) between adduct levels and dose. The results suggest that the 32P-postlabeling assay described herein will have utility in mechanistic studies and biomonitoring of DNA adduct formation from BDE and possibly other polar epoxides.