Abstract
Etheno (ε)-adducts, e.g., 1,N2-ε-guanine (1,N2-ε-G) and 1,N6-ε-adenine (1,N6-ε-A), are formed through the reaction of DNA with metabolites of vinyl compounds or with lipid peroxidation products. These lesions are known to be mutagenic, but it is unknown how they lead to errors in DNA replication that are bypassed by DNA polymerases. Here we report the structural basis of misincorporation frequencies across from 1,N2-ε-G by human DNA polymerase (hpol) η. In single-nucleotide insertions opposite the adduct 1,N2-ε-G, hpol η preferentially inserted dGTP, followed by dATP, dTTP, and dCTP. This preference for purines was also seen in the first extension step. Analysis of full-length extension products by LC-MS/MS revealed that G accounted for 85% of nucleotides inserted opposite 1,N2-ε-G in single base insertion, and 63% of bases inserted in the first extension step. Extension from the correct nucleotide pair (C) was not observed, but the primer with A paired opposite 1,N2-ε-G was readily extended. Crystal structures of ternary hpol η insertion-stage complexes with nonhydrolyzable nucleotides dAMPnPP or dCMPnPP showed a syn orientation of the adduct, with the incoming A staggered between adducted base and the 5'-adjacent T, while the incoming C and adducted base were roughly coplanar. The formation of a bifurcated H-bond between incoming dAMPnPP and 1,N2-ε-G and T, compared with the single H-bond formed between incoming dCMPnPP and 1,N2-ε-G, may account for the observed facilitated insertion of dGTP and dATP. Thus, preferential insertion of purines by hpol η across from etheno adducts contributes to distinct outcomes in error-prone DNA replication.