UCMB CP 160/16 , ave. F. Roosevelt, Universite Libre de Bruxelles, Belgium
The structural similarities and differences observed among the HTH DNA-binding domains, and the various modes of protein-DNA interactions, are examined and rationalized. Using an automatic classification procedure, all the domains of known structure are classed into families on the basis of the spatial arrangement of their helices, irrespective of the type of loops and the presence of beta-strands. It is found that the recognition helix and the preceding helix along the chain have always the same relative orientation. Structural differences arise when considering 3 helices, corresponding usually to the recognition helix and the 2 preceding ones, but sometimes to the recognition helix and the 2 flanking helices. Seven main families are obtained, whose members have in common the spatial arrangement of their 3 key helices, but have sometimes different topology and belong to different species. The structural divergence among these families and the existence of structural intermediates are analyzed. Searching these families systematically for recurrent motifs, leads to identify 2 specific turns, besides the HTH turn. They both link the 2 helices preceding the recognition helix and are each characteristic of a given family. Furthermore, the interaction patterns between amino acids and DNA within each family are analyzed and their conservation is examined with respect to the structural alignments. These patterns are found to be relatively well conserved within each family and to be different between the different families. The agreement of the structural classification and the patterns of protein-DNA contacts justify our approach, and suggests its applicability, in particular for modelling protein-DNA interactions.