Current Pharmaceutical Biotechnology - Volume 9, Issue 2, 2008

The ultimate goal of pharmacology and biotechnology is to develop drugs that could prevent or cure human diseases. Despite of the enormous progress made in experimental techniques, still discovering a new drug is an expensive and lengthy procedure. Structure-based drug discovery techniques offer fast and efficient alternative to the experimental approaches. Since protein-protein interactions are essential for the function of the living cell, they are one of the primary subjects of pharmaceutical investigations. However, the success of structure-based drug discovery depends on the availability of 3D structures of the proteins and protein-protein complexes being targeted. Apparently vast majority of these structures have to be modeled in silico. This special issue describes the current state-of-art and the progress made in developing computational approaches in two major directions: (A) Predicting structural features such as 3D structures and interfaces of protein-protein complexes and the conformational changes induced by the binding; and (B) Using the 3D structures to calculate biophysical characteristics such as the binding affinity and the effect of pH and salt concentration, to design inhibitors and to evaluate the effect of disease-associated single nucleoside polymorphism. (A). Predicting Structural Features Such as 3D Structures and Interfaces of Protein-Protein Complexes and the Conformational Changes Induced by the Binding The recent success of the human genome project and the progress in sequencing other genomes has enormously increased the universe of known proteins at amino acid sequences level. However, the biological function and the molecular basis of protein-protein interactions cannot easily be revealed from the sequence alone. Specifically, for the aims of structure-based drug design, 3D structures of the proteins and their complexes are needed. It is unlikely that all these structures will be determined experimentally. To bridge this gap, Structural Genomics Initiatives (SGI) are intended to experimentally determine the 3D structures of carefully selected targets so they can later serve as templates for the maximum number of protein sequences with unknown 3D structures. Currently SGIs are in stage 2, the production phase, and it has been projected that the 3D structures of all monomeric proteins of interest will be predicted in feasible time. The next level of these initiatives is naturally extended toward predicting the 3D structure of the corresponding protein-protein complexes. There are two distinctive approaches of predicting the 3D structures of protein-protein complexes: ab-initio docking and template-based docking. The first one uses physical methods to dock the experimentally determined 3D structures or high quality models of the monomers. The second one, that does not require a priori knowledge of the monomeric structures, predicts the 3D structure of a complex based on the homology relations to another complex with known 3D structure. The last approach, perhaps will require Structural Proteomic Initiatives (SPI), such that significant carefully selected number of representative 3D structures of protein complexes will be experimentally determined and will be further used as templates to generate models for maximal number of complexes with unknown 3D structures. The achievements and perspectives in this important area are outlined in the manuscript “Predicting 3D structures of protein-protein complexes” by Ilya A. Vakser and Petras Kundrotas. The performance of both ab-initio and template-based docking methods can be significantly improved if the binding interfaces are successfully predicted. With respect to the ab-initio docking, this will reduce the sampling and will avoid wrong binding modes. With regard to template-based docking, especially if the interfaces are predicted on sequence level, this will allow for applying profile-to-profile alignment methods that emphasize on the alignment of interfacial residues and thus will contribute to better detection of appropriate templates. From point of view of drug discovery, the ability to predict putative binding interfaces on the 3D structures of monomeric proteins is of special interest. It has the potential to reveal protein-protein interaction networks and the interfacial residues, which can be targeted by inhibitors to alter the corresponding protein-protein interactions. The progress made in developing computational methods to predict protein-protein interfaces and corresponding interaction networks is reviewed in the manuscript “Characterization and prediction of protein interfaces to infer proteinprotein interaction networks”by Ozlem Keskin, Nurcan Tuncbag and Attila Gursoy.

The protein-protein docking problem is one of the focal points of activity in computational structural biology. Adequate computational techniques for structural modeling of protein interactions are important because of the growing number of known protein structures, particularly in the context of structural genomics. The protein docking methodology offers tools for fundamental studies of protein interactions and provides structural basis for drug design. The paper presents a critical review of the existing protein-protein docking approaches in view of the fundamental principles of protein recognition.

Complex protein-protein interaction networks govern biological processes in cells. Protein interfaces are the sites where proteins physically interact. Identification and characterization of protein interfaces will lead to understanding how proteins interact with each other and how they are involved in protein-protein interaction networks. What makes a given interface bind to different proteins; how similar/different the interactions in proteins are some key questions to be answered. Enormous amount of protein structures and experimental protein-protein interactions data necessitate advanced computational methods for analyzing and inferring new knowledge. Interface prediction methods use a wide range of sequence, structural and physico-chemical characteristics that distinguish interface residues from non-interface surface residues. Here, we present a review focusing on the characteristics of interfaces and the current status of interface prediction methods.

The ability to predict the three-dimensional structure of a protein complex starting from the isolated binding partners is becoming increasingly relevant. As our understanding of the molecular mechanisms behind protein-protein binding improves, so do the docking methods, however, it remains a challenge to adequately predict the unbound to bound transition. Side-chain flexibility is routinely handled and most docking methods allow for a certain degree of backbone flexibility, but systems undergoing moderate to large conformational changes can at present not correctly be modeled. The docking community is therefore putting an increased effort in the treatment of protein flexibility. Here we present a survey of the existing computational techniques to model protein flexibility in the context of protein-protein docking.

The functional capabilities of biological systems, such as enzyme catalysis, nutrient import, and cell signaling, depend crucially on specific molecular interactions. In addition, the effects of common drugs also act through a mechanism of binding to specific biomolecular targets. Models for the prediction of binding affinity are used in basic research to study the molecular basis of biological function as well as in applied research to study the development of new drugs. This review will address the biological importance of molecular recognition as well as its influence on the development of pharmaceuticals. Further, a broad overview of computational approaches used for the prediction of biological activity and specifically binding free energy will be presented.

Ionic strength- (or salt-) effects on the protein-protein binding free energy has been included in many computational studies, while comparatively fewer computational studies have looked at the corresponding effect of pH. The pH dependence can be very complex if several groups change protonation state, while the ionic strength dependence usually scales as ln(I), and the main challenge is to predict the magnitude of the correlation. However, there is now very strong indication that pH effects due to binding induced changes in protonation states make a non-negligible contribution to the binding energy of most protein-protein complexes. This observation, together with more efficient pKa prediction methods and the emergence of constant pH molecular dynamics simulations to model the protonation-dependent structural changes will spark more experimental and theoretical work in pH effects on protein-protein binding.

Protein-protein interactions (PPIs) have a pivotal role in many biological processes suggesting that targeting macromolecular complexes will open new avenues for the design of the next generation of therapeutics. A wide range of “in silico methods” can be used to facilitate the design of protein-protein modulators. Among these methods, virtual ligand screening, protein-protein docking, structural predictions and druggable pocket predictions have become established techniques for hit discovery and optimization. In this review, we first summarize some key data about proteinprotein interfaces and introduce some recently reported computer methods pertaining to the field. URLs for several recent free packages or servers are also provided. Then, we discuss four studies aiming at developing PPI modulators through the combination of in silico and in vitro screening experiments.

Almost all (99.9%) nucleotide bases are exactly the same in all people, however, the remaining 0.1% account for about 1.4 million locations where single-base DNA differences/polymorphisms (SNPs) occur in humans. Some of these SNPs, called non-synonymous SNPs (nsSNPs), result in a change of the amino acid sequences of the corresponding proteins affecting protein functions and interactions. This review summarizes the plausible mechanisms that nsSNPs may affect the normal cellular function. It outlines the approaches that have been developed in the past to predict the effects caused by nsSNPs with special emphasis on the methods that use structural information. The review provides systematic information on the available resources for predicting the effects of nsSNPs and includes a comprehensive list of existing SNP databases and their features. While nsSNPs resulting in amino acid substitution in the core of a protein may affect protein stability irreversibly, the effect of an nsSNP resulting to a mutation at the surface of a protein or at the interface of protein-protein complexes, could, in principle be, subject of drug therapy. The importance of understanding the effects caused by nsSNP mutations at the protein-protein and protein-DNA interfaces is outlined.