Structural Biochemistry/Binding-sites Predictions assisting protein-protein Docking

Introduction
Protein binding sites are the region where proteins interact with each other. This region usually contains the specific part of the three-dimensional of the protein. If we can identify their biding sites, we can proceed to study their function and the protein-protein docking by docking algorithms.

Protein Data Bank (or PDB) functions as storage of protein complex structures. Biochemists always try to obtain the structure of specific proteins, but under experiment condition, protein structures are really hard to be obtained under the condition when it needs for crystallization. Because of the disadvantages of constructing the experiment, biochemist leads to the development of protein-protein docking.

Binding-site prediction and protein-protein docking
Protein-protein docking is a computational approaches to predict the three-dimensional structure of complex proteins. The success of this technique depends mostly on pre-knowledge of the protein-protein binding sites. In order to predict the structure, the computational approach must focus difference in binding sites between the interfaces of a set of proteins. Most of the time, there are some proteins interfacing at the same regions which then become a hotspots, whereas others might change.

With the requirement in the precision of the binding sites, biochemists developed the algorithm- which is used for predicting the protein binding sites by preserving the protein surface structure and the properties of the fundamental protein structures. We have to insert this algorithm to ProBiS which is a host to detect protein binding sites. The idea behind the algorithm is that most of the conversed parts of protein surface are somehow in accompanying with other proteins or ligands. In order to obtain the conserved part of protein surface, we have to find out the similar local surface between the concerned protein and other proteins.

To conduct the example of this method, we choose the two unbound interacting proteins: FKBP12 (immunophilin) and TBR-1 (a growth factor) with PDB codes of 1d6o and 1ias. Some of the proteins that seem to share the same similarities in structure with FKBP12 are: 1ix5, 1jvw, 1pbk, 1q6h, 1r9h, 1u79, 2awg, 2d9f, 2if4, 2ofn, 2pbc, 2uz5, and 3b7x; with TBR-1 are : 1ckj, 1kob, 1m17, 1o6k, 1o9u, 1u59, 1wak, 1yhv, 1yvj, 2b7a, 2bfy, 2csn, 2f4j, 2ivt, 2izs, 2j0l, 2jbo, 2pzy, 2qkw, 2qlu, 2qr7, 2v7o, 3bkb.

ProBis is now used to predict the binding sites. The fundamental protein has to interact with the polypeptide chain. Our goal is to find out the similar surfaces of these proteins, so we want to minimize the dissimilarities as much as we can. As you can see on the picture on the right, all the conversed regions are mapped over the other ones.

AutoDock 4.0 is then used for docking of protein FKBP12 to the protein TBR-1. This program requires computational interference since it workswith the whole protein structures, so it needs a precise image. The AutoDock uses a force field to give a stronger attraction between the atoms on predicted binding site. The success of docking depends on the comparison between the regions of predicted binding site residues with the corresponding ones. This force field affects the docking. As you can see in this chart, five time larger force field has the highest number of best docked structure.

This 5x force field has 9 different structures between the predicted and the actual binding site residues. The most preferable clustering also belongs to this one since it has the most best docked structure. This theory somehow states that the docking algorithm can be able to explain the structure of the complex protein.

Reference
Scientific Paper. Binding-sites Prediction Assisting Protein-protein Docking. Acta Chim. Slov. 2011, 58, 396–401