Bioinformatics
|
|
|
| |
|
 Characterization of Protein-Protein Binding Sites: a comparative study of Unbound versus Bound Protein complexes
Is the whole protein surface available for interaction with other proteins? or are specific sites pre-assigned according to their biophysical and structural character? Can these properties be detected in the unbound proteome? These questions are quantitatively addressed using a database, of 107 unique, non-homologous protein complexes involved in heteromeric, transient protein-protein interactions, from which the structures of both the unbound and bound states were determined for 76 of the proteins.
Surface properties were probed for circles of 10 Å radii, comparing proteins in their bound and unbound state in order to detect unique characteristics distinguishing interacting surfaces from non-interacting surfaces. In structural terms, the binding site is often located in concave or flat areas, with a rugged surface (as determined from the higher surface density). Binding sites are located more often on
An exiting outcome of interface identification (described above) and of using PARE to calculate the electrostatic attraction between proteins, is their potential as guidelines for docking of protein-protein complexes. The incorporation of these into the scoring function of a standard, shape complementarity docking algorithm achieved very promising results, which are currently being refined.
We are using the PARE algorithm, developed for the design of tighter and faster binding protein complexes, to probe the influence of electrostatics in the association of over 100 protein complexes for whom structures are available. Moreover, we have developed a computational tool which can simultaneously analyze the potential for electrostatic rate enhancement on many proteins in parallel. |
|
-sheets and not on a
helixeshelices, or on the edges of relatively long non-structured chains. Chemically, we found similar preferences for amino-acids as have been previously reported. Interestingly, charge-charge interactions are the only pairwise interactions that were found to be more prevalent then expected in interface regions. While the hydrophobicity of the interface is similar to the rest of the surface, hydrophobic residues tend to form larger clusters. Entropically, the temperature factor of the binding site is lower already in the unbound form. The emerging picture is logical from both a structural and a chemical point of view, and clearly suggests that potential interface regions are different from the rest of the protein surface. These unique characteristics are the basis for an interface prediction program developed in our lab. This program has the ability to predict the location of the binding site on the unbound protein in about 75% of cases checked.
