A computational study of the structures and base-pairing properties of pyrrolizidine alkaloid-derived DNA adducts

Pyrrolizidine alkaloids (PAs) are found in many plants worldwide, including some in Canada. PAs have been linked to losses in livestock populations and the development of human PA diseases. Four PA-derived dehydrosupinidine (DHP) adducts formed at the exocyclic amino groups of DNA purines have been found in tumor tissue and flagged as potential biomarkers for tumor formation (denoted DHP–G/A–3 and DHP–G/A–4). Four additional adducts (DHP–G/A–1 and DHP–G/A–2) have also been identified, which differ in the stereochemistry at the DHP–nucleobase linker or DHP ring carbon containing the hydroxy group, as well as the length of the DHP–nucleobase linker. Since the impact of these distinct chemical features on adduct mutagenicity is currently unclear, the present work uses density functional theory calculations to uncover the structures and base-pairing properties of these experimentally-observed DHP-derived purine adducts. Adduct Watson–Crick–Franklin base pairs involving the canonical partner are energetically and structurally feasible. However, the G/A–3 and G/A–4 adducts are also highly susceptible to mispairing with G. Indeed, the longer and more flexible DHP–nucleobase linker in these adducts affords interactions between the DHP moiety and the pairing G that are not possible for the G/A–1 and G/A–2 counterparts. Our data thus rationalize the experimental identification of this subset of adducts in liver tumor tissue, and provide key insights to guide future experimental and computational studies that investigate the replication and broader biological outcomes of DHP-derived lesions.