Comprehensive analysis of hydrogen bonds in regulatory protein DNA-complexes: in search of common principles.

A systematic analysis of hydrogen bonds between regulatory proteins and their DNA targets is presented, based on 28 crystallographically solved complexes. All possible hydrogen bonds were screened and classified into different types: those that involve the amino acid side-chains and DNA base edges and those that involve the backbone atoms of the molecules. For each interaction type, all bonds were characterized and a statistical analysis was performed to reveal significant amino acid-base interdependence. The interactions between the amino acid side-chains and DNA backbone constitute about half of the interactions, but did not show any amino acid-base correlation. Interactions via the protein backbone were also observed, predominantly with the DNA backbone. As expected, the most significant pairing preference was demonstrated for interactions between the amino acid side-chains and the DNA base edges. The statistically significant relationships could mostly be explained by the chemical nature of the participants. However, correlations that could not be trivially predicted from the hydrogen bonding potential of the residues were also identified, like the preference of lysine for guanine over adenine, or the preference of glutamic acid for cystosine over adenine. While Lys x G interactions were very frequent and spread over various families, the Glu x C interactions were found mainly in the basic helix-loop-helix family. Further examination of the side-chain-base edge contacts at the atomic level revealed a trend of the amino acids to contact the DNA by their donor atoms, preferably at position W2 in the major groove. In most cases it seems that the interactions are not guided simply by the presence of a required atom in a specific position in the groove, but that the identity of the base possessing this atom is crucial. This may have important implications in molecular design experiments.