Cation -pi/H-bond stair motifs are recurrently found at the binding interface between protein and DNA. They involve two nucleobases and an amino acid side chain, and encompass three different types of interactions: nucleobase stacking, nucleobase-amino acid H-bond and nucleobase-amino acid cation-pi interaction. The interaction energies of the 77 stair motif geometries identified in a data set of 52 high-resolution protein-DNA complexes were investigated by means of ab initio quantum chemistry calculations. Using the standard 6-31G* basis set, we first establish the value of the Gaussian alpha(d)-exponent of d-polarization functions on heavy atoms, which optimizes the MP2 interaction energies. We show that, although the default value of alpha(d) = 0.8 is appropriate to minimize the total MP2 energy of a system, the value of alpha(d) = 0.2 is optimal for the three types of pairwise interactions studied and yields MP2 interaction energies quite similar to those calculated with more extended basis sets. Indeed, the more diffuse nature of the alpha(d) = 0.2 basis functions allows a spatial overlap between the orbitals of the interacting partners. Such functions are also shown to improve the multipole electric moments in the interaction region, which results in a stabilizing polarization effect and a better description of the dispersive energy contributions. Using the MP2 computation level and the 6-31G* basis set with alpha(d) = 0.2 instead of alpha(d) = 0.8, we computed the interaction energies of the 77 observed stair motif geometries and found that, in a vacuum, the cation-pi energy is much less favorable, about 3 times, than the H-bond energy and of the same order of magnitude as the pi-pi stacking energy. Furthermore, the convergence of the MP perturbation theory expansions was analyzed by computing the MP3 and MP4 corrections on simplified complexes. These expansions exhibited an oscillatory behavior, where MP2 seems to provide a satisfactory approximation, albeit slightly overestimated, to the interaction energy.