The polycyclic aromatic hydrocarbons (PAHs) dibenzo[a,l]pyrene (DBP) and benzo[a]pyrene (BP) are widespread environmental contaminants and potent carcinogens. The fjord-region DBP is considerably more carcinogenic than the bay-region BP. This fact can be ascribed to differences in DNA binding efficiency of their ultimate carcinogenic diol epoxide (DE) intermediates, differences in structural features of the DNA adducts, and differences in DNA adduct recognition and the subsequent lesion removal by nucleotide excision repair (NER). We have compared the formation and removal of adducts as a function of time formed by the carcinogenic metabolites (-)-anti-DBPDE and (+)-anti-BPDE in A549 human epithelial lung carcinoma cells. Cells were exposed to 0.1 or 1.0 microM (-)-anti-DBPDE and (+)-anti-BPDE, respectively. Adducts were measured at various post-treatment times (up to 6 h) by enzymatic DNA hydrolysis and a HPLC procedure that allows monitoring of all cis- and trans-nucleoside adducts of dA and dG. Treatment with 0.1 microM (-)-anti-DBPDE resulted in an initial increase of adducts to a maximal level of 144 pmol adducts/mg of DNA after 1 h of incubation. This was followed by an apparent, although not statistically significant, slow removal of adducts. After 6 h of incubation, at least 80% seems to remain. In cells treated with 1.0 microM (+)-anti-BPDE, the maximal level of 140 pmol adducts/mg of DNA was reached within 20 min of exposure. The formation was followed by an initial rapid decline in the adduct level (1.54 pmol adducts/mg of DNA/min) and a later statistically significant slower rate (0.14 pmol adducts/mg of DNA/min) of adduct removal. After 1 h of incubation, about 45% of the adducts are removed followed by 75% at 6 h. The biphasic pattern of BPDE removal has been observed previously in mammalian cells and, at least in part, may reflect the action of transcription-coupled repair (TCR) and the subsequent global genomic repair (GGR). Comparing the rate of removal of adducts derived from BPDE with those of DBPDE, the latter are obviously more refractory to the NER-coupled repair than the former. Furthermore, the apparent resistance of adducts from DBPDE to be eliminated may reflect the ability of such adducts to escape recognition and/or the subsequent removal by the NER machinery. Further analysis of DNA adduct distribution as a function of incubation time reveals that the dA/dG adduct ratio for BPDE was independent of time (4% dA, 96% dG), whereas the corresponding ratio for DBPDE was significantly increased from 2.9 (74% dA, 26% dG) at 20 min to 4.0 (80% dA, 20% dG) after 6 h of incubation. The results presented here on DNA adduct removal in mammalian cells are in part consistent with recent results on NER-coupled activity on bay- and fjord-region DE-modified oligonucleotides in vitro and further substantiate the hypothesis that the high carcinogenicity of the nonplanar PAHs arise from the ability of the preferentially formed dA adducts to escape recognition by surveillance systems and the subsequent NER-coupled lesion removal.