MIT Dept of biology, Pabo lab, room 68-580, MIT, Cambridge, MA 02139, USA
Structural studies of protein-DNA complexes have shown that there are many distinct families of DNA-binding proteins; these studies also have revealed a rich variety of side chain - base interactions. It is generally accepted that there is no simple "code" for recognition: The diversity of observed structures and contacts even makes it difficult to systematically analyze and compare the known complexes.
A possibility of side chain - base contact clearly depends on the relative position of the base and the peptide backbone. We can say that this relative position determines the "meaning" (i.e. the range of possible contacts) that a side chain might have in protein-DNA recognition. Geometric analysis of side chain - base spatial relationships is, therefore, a promising approach to classification of protein-DNA contacts, and eventually to the design.
To numerically express these spatial relationships for a given protein-DNA complex, we introduce local coordinate systems on segments of polypeptide backbone and on DNA bases. This approach also enables numeric comparisons of various amino acid - base spatial relationships. These pairwise comparisons, in turn, allow a global comparison of the overall docking arrangements in different protein-DNA complexes.
A set of programs was written, that
(a) parses all currently available high-resolution protein-DNA complexes
into a database of spatial relationships (expressed as affine
transformations);
(b) searches this database for existing contacts (or spatial relationships)
that are similar to known contacts. This may eventually facilitate
the design of novel DNA-binding proteins.
(c) performs quantitative comparison, analysis and classification of
overall protein-DNA docking arrangements.
Some unexpected similarities and differences, both in amino acid - base relative positions at contact sites, and in orientation of alpha-helices in the major groove of DNA, are revealed by this analysis.