Prediction of nucleoside-carcinogen reactivity. Alkylation of adenine, cytosine, guanine, and thymine and their deoxynucleosides by alkanediazonium ions

Abstract

MNDO semiempirical molecular orbital calculations for the SN2 alkylation of nucleic acid bases and deoxynucleosides by the methane-, ethane-, and propanediazonium ions are presented. An approximate correlation is demonstrated between the calculated relative activation enthalpies for attack at alternative base sites and the related experimental quantities for DNA modification by alkylnitrosoureas. The empirically observed shift from N- to O-alkylation with increasing complexity of the alkylating agent is reproduced by the calculations and rationalized by using an extension of a model worked out previously for the analogous reactions of simple nucleophiles. According to this model, the energetics of the related SN1 reactions, while not directly involved, have a profound influence on the SN2 transition-state geometries. For reactions in which the SN1 dissociation is unfavorable the forming bond to the incoming nucleophiles in the related SN2 transition state tends to be short and covalent interactions, which favor N-alkylation, play a significant role. When the SN1 reaction is more facile, the SN2 transition states are "looser" and the covalent interactions correspondingly smaller, leading to an overall shift away from N-alkylation. Consideration of the form of the electrostatic potential around the base, in conjunction with these ideas, provides a detailed explanation of the behavior of electrophiles toward the guanine N2-, 7-, and O6-positions. This model unifies much of the language already used in discussions of nucleic acid regiochemistry. At the same time it is consistent with the geometries and charge distributions in the transition states calculated for the gas-phase reaction processes.