Mutagenesis by (+)-anti-benzo[a]pyrene-7,8-dihydrodiol-9,10-epoxide [(+)-anti-B[a]PDE], an important mutagenic/carcinogenic metabolite of benzo[a]pyrene (B[a]P), is being studied in order to understand the factors that influence mutagenesis both quantitatively and qualitatively. A new mutational system, which permits the selection of supF- mutations in an Escherichia coli plasmid, pUB3, was used. The work described herein is an extension of previous work, which involved plasmid adduction and then immediate transformation (Rodriguez & Loechler, 1993), and began with the observation that mutation frequency (MF) decreased approximately 2-fold when the (+)-anti-B[a]PDE-adducted plasmid pUB3 is either (1) frozen and then thawed prior to transformation or (2) heated at 80 degrees C for 10 min prior to transformation. Several results suggest that this decrease is not due to the loss of labile adducts. To begin to understand this phenomenon, the mutagenic spectra are compared for (+)-anti-B[a]PDE in supF for the unheated (187 mutants), the freeze/thawed (134 mutants), and the heated (254 mutants) samples. In general, freeze/thawing and heating cause a decrease in all classes of mutations. Considering substitution mutations at G.C base pairs, which predominate, the mutagenic specificity for the combined data sets is GC-->TA (57%), GC-->AT (23%), and GC-->CG (20%). This raises the question, how does (+)-anti-B[a]PDE generate this complex mutagenic specificity, which contrasts with the situation for, e.g., simple methylating agents? One factor is that mutagenic specificity at a particular guanine residue can be influenced by the base on its immediate 5'-side, most notably where mutations are virtually exclusively restricted to GC-->TA in 5'-TG-3' sequence contexts. One unexpected finding may provide additional insight. G115 in supF, which is the major hot spot for base-pairing mutagenesis, is the only site where the qualitative pattern of mutagenesis is significantly affected by heating the (+)-anti-B[a]PDE-adducted plasmid prior to transformation. Without heating, G115-->T mutations predominate, but following heating there is a statistically significant increase in the fraction of G115-->A and G115-->C mutations. The most likely model to explain this and other results is (1) a particular DNA adduct can adopt multiple conformations, (2) the conformation adopted by an adduct can be influenced by various factors, including DNA sequence context, as well as heating and freeze/thawing, and (3) each of these conformations can cause a different pattern of mutation.(ABSTRACT TRUNCATED AT 400 WORDS)