Structural
BioinformaticsLab

Despite the growing number of examples of small-molecule inhibitors that disrupt protein–protein interactions (PPIs), the origin of druggability of such targets is poorly understood. To identify druggable sites in protein–protein interfaces we combine computational solvent mapping, which explores the protein surface using a variety of small “probe” molecules, with a conformer generator to account for side-chain flexibility. Applications to unliganded structures of 15 PPI target proteins show that the druggable sites comprise a cluster of binding hot spots, distinguishable from other regions of the protein due to their concave topology combined with a pattern of hydrophobic and polar functionality. This combination of properties confers on the hot spots a tendency to bind organic species possessing some polar groups decorating largely hydrophobic scaffolds. Thus, druggable sites at PPI are not simply sites that are complementary to particular organic functionality, but rather possess a general tendency to bind organic compounds with a variety of structures, including key side chains of the partner protein. Results also highlight the importance of conformational adaptivity at the binding site to allow the hot spots to expand to accommodate a ligand of drug-like dimensions. The critical components of this adaptivity are largely local, involving primarily low energy side-chain motions within 6 Å of a hot spot. The structural and physicochemical signature of druggable sites at PPI interfaces is sufficiently robust to be detectable from the structure of the unliganded protein, even when substantial conformational adaptation is required for optimal ligand binding.

This study was published in PNAS,  was  featured in PNAS, and highlighted in Nature Reviews Drug Discovery. 

For more information, please read the paper:

Kozakov D, Hall DR, Chuang G-Y, Cencic R, Brenke R, Grove LE, Beglov D, Pelletier J, Whitty A, Vajda S.  2011.  Structural conservation of druggable hot spots in protein-protein interfaces. Proceedings of the National Academy of Sciences. 108(33):13528-13533.

CAPRI (Critical Assessment of Predicted Interactions) is a community-wide experiment devoted to the prediction of protein complexes based on the structures of the component proteins (http://www.ebi.ac.uk/msd-srv/capri/). The results for targets 29-42 were evaluated at the Fourth CAPRI Evaluation Meeting in Barcelona on December 9-11, 2009, for 63 predictor groups and 10 automated docking servers (http://mmb.pcb.ub.es/capri2009/). Our team was found to have the best performance in both categories. The complete set of slides by the assessors is available for download.

The summary of the results is also shown below. For each predictor group, the table shows the number of acceptable or better predictions, and among those the number of high quality models, indicated by three stars, as well as the number of medium quality solutions, indicated by two stars. According to this table, a number of groups made 6 successful predictions, and thus our group distinguished itself only by the slightly higher number of high or medium quality models. However, together with the performance of the ClusPro automated server this result represents a real breakthrough. By definition, servers can use only the atomic coordinates of two component proteins, and no literature or other data that are available to human predictors. Nevertheless, ClusPro outperformed not only the other 9 servers, but also 53 of the 63 participating groups! According to Dr. Shoshana Wodak, head of CAPRI assessment team, an automated server was able to compete with human predictors for the first time in CAPRI’s history. 

The best solutions provided by ClusPro were further refined higher accuracy, again without the use of additional information. Accordingly, ours was the only group obtaining good results for Target T39 on which incorrect experimental data have been published, misleading all other predictors. Although the present web-based version of ClusPro does not perform the computationally very demanding refinement step, it will be included when appropriate computational resources will be available. Thus, our results in the latest round of CAPRI indicate the feasibility of high accuracy ab initio protein-protein docking for a substantial fraction of proteins.  In addition, the high computational efficiency of the initial docking and the increasing number of protein structures will enable us to solve problems in systems biology such as the validation of putative protein-protein interaction networks. 

The ClusPro 2.0 protein-protein docking server is available at http://cluspro.bu.edu.

Assessment of CAPRI Predictions 2009

Barcelona, December 9-11, 2009

Predictors

Servers