Computational study of substituent effects on the interaction energies of hydrogen-bonded watson-crick cytosine: Guanine base pairs

The substituent effects on interaction energies of hydrogen-bonded DNA Watson-Crick base pairs in the gas phase were captured in a model using ab initio descriptors (at the B3LYP/6-311+G(2d,p) level). While forming a noncovalently bonded complex with unsubstituted guanine (G), cytosine (C) carried 42 possible substituents both at the C6 position (C6X:G) and at the C5 position (C5X:G). We rationalize why complexes possessing a more strongly electron-withdrawing group in CX form less stable base pairs. Multivariate linear regression constructed the quantitative relationships between the interaction energies of the complexes and the descriptors, which were drawn from quantum chemical topology (QCT). For the C6X dataset, the best model yielded r2 = 0.93 and a root-mean-square (rms) energy of 0.53 kJ/mol for the 28 complexes in the training set. This model was evaluated by an external test set (14 complexes), yielding an r2 value of 0.96 and an rms error of 0.42 kJ/mol. For the C5X dataset, the QCT descriptors generated a linear model, with r2 values of 0.92 and 0.97 and rms values of 1.69 and 1.24 kJ/mol for the training set (31 compounds) and the external test set (11 compounds), respectively. The models built here could therefore be useful for the assessment of the interaction energy of C6X:G and C5X:G purely from monomelic data. © 2008 American Chemical Society.