Discriminating physiological from non-physiological interfaces in structures of protein complexes: A community-wide study

Reliably scoring and ranking candidate models of protein complexes and assigning their oligomeric state from the structure of the crystal lattice represent outstanding challenges. A community-wide effort was launched to tackle these challenges. The latest resources on protein complexes and interfaces were exploited to derive a benchmark dataset consisting of 1677 homodimer protein crystal structures, including a balanced mix of physiological and non-physiological complexes. The non-physiological complexes in the benchmark were selected to bury a similar or larger interface area than their physiological counterparts, making it more difficult for scoring functions to differentiate between them. Next, 252 functions for scoring protein-protein interfaces previously developed by 13 groups were collected and evaluated for their ability to discriminate between physiological and non-physiological complexes. A simple consensus score generated using the best performing score of each of the 13 groups, and a cross-validated Random Forest (RF) classifier were created. Both approaches showed excellent performance, with an area under the Receiver Operating Characteristic (ROC) curve of 0.93 and 0.94, respectively, outperforming individual scores developed by different groups. Additionally, AlphaFold2 engines recalled the physiological dimers with significantly higher accuracy than the non-physiological set, lending support to the reliability of our benchmark dataset annotations. Optimizing the combined power of interface scoring functions and evaluating it on challenging benchmark datasets appears to be a promising strategy.     

Prediction of interface residues in protein-protein complexes by a consensus neural network method: test against NMR data

The number of structures of protein-protein complexes deposited to the Protein Data Bank is growing rapidly. These structures embed important information for predicting structures of new protein complexes. This motivated us to develop the PPISP method for predicting interface residues in protein-protein complexes. In PPISP, sequence profiles and solvent accessibility of spatially neighboring surface residues were used as input to a neural network. The network was trained on native interface residues collected from the Protein Data Bank. The prediction accuracy at the time was 70% with 47% coverage of native interface residues. Now we have extensively improved PPISP. The training set now consisted of 1156 nonhomologous protein chains. Test on a set of 100 nonhomologous protein chains showed that the prediction accuracy is now increased to 80% with 51% coverage. To solve the problem of over-prediction and under-prediction associated with individual neural network models, we developed a consensus method that combines predictions from multiple models with different levels of accuracy and coverage. Applied on a benchmark set of 68 proteins for protein-protein docking, the consensus approach outperformed the best individual models by 3-8 percentage points in accuracy. To demonstrate the predictive power of cons-PPISP, eight complex-forming proteins with interfaces characterized by NMR were tested. These proteins are nonhomologous to the training set and have a total of 144 interface residues identified by chemical shift perturbation. cons-PPISP predicted 174 interface residues with 69% accuracy and 47% coverage and promises to complement experimental techniques in characterizing protein-protein interfaces. .     

Structural templates for comparative protein docking

Structural characterization of protein‐protein interactions is important for understanding life processes. Because of the inherent limitations of experimental techniques, such characterization requires computational approaches. Along with the traditional protein‐protein docking (free search for a match between two proteins), comparative (template‐based) modeling of protein‐protein complexes has been gaining popularity. Its development puts an emphasis on full and partial structural similarity between the target protein monomers and the protein‐protein complexes previously determined by experimental techniques (templates). The template‐based docking relies on the quality and diversity of the template set. We present a carefully curated, nonredundant library of templates containing 4950 full structures of binary complexes and 5936 protein‐protein interfaces extracted from the full structures at 12 Å distance cut‐off. Redundancy in the libraries was removed by clustering the PDB structures based on structural similarity. The value of the clustering threshold was determined from the analysis of the clusters and the docking performance on a benchmark set. High structural quality of the interfaces in the template and validation sets was achieved by automated procedures and manual curation. The library is included in the Dockground resource for molecular recognition studies at http://dockground.bioinformatics.ku.edu . Proteins 2015; 83:1563–1570. © 2014 Wiley Periodicals, Inc.     

Unlocking the secrets to protein–protein interface drug targets using structural mass spectrometry techniques

Protein–protein interactions (PPIs) drive all biologic systems at the subcellular and extracellular level. Changes in the specificity and affinity of these interactions can lead to cellular malfunctions and disease. Consequently, the binding interfaces between interacting protein partners are important drug targets for the next generation of therapies that block such interactions. Unfortunately, protein–protein contact points have proven to be very difficult pharmacological targets because they are hidden within complex 3D interfaces. For the vast majority of characterized binary PPIs, the specific amino acid sequence of their close contact regions remains unknown. There has been an important need for an experimental technology that can rapidly reveal the functionally important contact points of native protein complexes in solution. In this review, experimental techniques employing mass spectrometry to explore protein interaction binding sites are discussed. Hydrogen–deuterium exchange, hydroxyl radical footprinting, crosslinking and the newest technology protein painting are compared and contrasted.     

DockTrina: Docking triangular protein trimers

In spite of the abundance of oligomeric proteins within a cell, the structural characterization of protein–protein interactions is still a challenging task. In particular, many of these interactions involve heteromeric complexes, which are relatively difficult to determine experimentally. Hence there is growing interest in using computational techniques to model such complexes. However, assembling large heteromeric complexes computationally is a highly combinatorial problem. Nonetheless the problem can be simplified greatly by considering interactions between protein trimers. After dimers and monomers, triangular trimers (i.e. trimers with pair‐wise contacts between all three pairs of proteins) are the most frequently observed quaternary structural motifs according to the three‐dimensional (3D) complex database. This article presents DockTrina, a novel protein docking method for modeling the 3D structures of nonsymmetrical triangular trimers. The method takes as input pair‐wise contact predictions from a rigid body docking program. It then scans and scores all possible combinations of pairs of monomers using a very fast root mean square deviation test. Finally, it ranks the predictions using a scoring function which combines triples of pair‐wise contact terms and a geometric clash penalty term. The overall approach takes less than 2 min per complex on a modern desktop computer. The method is tested and validated using a benchmark set of 220 bound and seven unbound protein trimer structures. DockTrina will be made available at http://nano‐d.inrialpes.fr/software/docktrina . Proteins 2014; 82:34–44. © 2013 Wiley Periodicals, Inc.     

BIPSPI+: Mining Type-Specific Datasets of Protein Complexes to Improve Protein Binding Site Prediction

Computational approaches for predicting protein-protein interfaces are extremely useful for understanding and modelling the quaternary structure of protein assemblies. In particular, partner-specific binding site prediction methods allow delineating the specific residues that compose the interface of protein complexes. In recent years, new machine learning and other algorithmic approaches have been proposed to solve this problem. However, little effort has been made in finding better training datasets to improve the performance of these methods. With the aim of vindicating the importance of the training set compilation procedure, in this work we present BIPSPI+, a new version of our original server trained on carefully curated datasets that outperforms our original predictor. We show how prediction performance can be improved by selecting specific datasets that better describe particular types of protein interactions and interfaces (e.g. homo/hetero). In addition, our upgraded web server offers a new set of functionalities such as the sequence-structure prediction mode, hetero- or homo-complex specialization and the guided docking tool that allows to compute 3D quaternary structure poses using the predicted interfaces. BIPSPI+ is freely available at https://bipspi.cnb.csic.es.     

An orientation-dependent hydrogen bonding potential improves prediction of specificity and structure for proteins and protein-protein complexes

Hydrogen bonding is a key contributor to the specificity of intramolecular and intermolecular interactions in biological systems. Here, we develop an orientation-dependent hydrogen bonding potential based on the geometric characteristics of hydrogen bonds in high-resolution protein crystal structures, and evaluate it using four tests related to the prediction and design of protein structures and protein-protein complexes. The new potential is superior to the widely used Coulomb model of hydrogen bonding in prediction of the sequences of proteins and protein-protein interfaces from their structures, and improves discrimination of correctly docked protein-protein complexes from large sets of alternative structures.     

Atomic contact vectors in protein-protein recognition

The ability to analyze and compare protein-protein interactions on the structural level is critical to our understanding of various aspects of molecular recognition and the functional interplay of components of biochemical networks. In this study, we introduce atomic contact vectors (ACVs) as an intuitive way to represent the physico-chemical characteristics of a protein-protein interface as well as a way to compare interfaces to each other. We test the utility of ACVs in classification by using them to distinguish between homodimers and crystal contacts. Our results compare favorably with those reported by other authors. We then apply ACVs to mine the PDB for all known protein-protein complexes and separate transient recognition complexes from permanent oligomeric ones. Getting at the basis of this difference is important for our understanding of recognition and we achieved a success rate of 91% for distinguishing these two classes of complexes. Although accessible surface area of the interface is a major discriminating feature, we also show that there are distinct differences in the contact preferences between the two kinds of complexes. Illustrating the superiority of ACVs as a basic comparison measure over a sequence-based approach, we derive a general rule of thumb to determine whether two protein-protein interfaces are redundant. With this method, we arrive at a nonredundant set of 209 recognition complexes--the largest set reported so far.     

Quantitative prediction of protein–protein binding affinity with a potential of mean force considering volume correction

Quantitative prediction of protein–protein binding affinity is essential for understanding protein–protein interactions. In this article, an atomic level potential of mean force (PMF) considering volume correction is presented for the prediction of protein–protein binding affinity. The potential is obtained by statistically analyzing X‐ray structures of protein–protein complexes in the Protein Data Bank. This approach circumvents the complicated steps of the volume correction process and is very easy to implement in practice. It can obtain more reasonable pair potential compared with traditional PMF and shows a classic picture of nonbonded atom pair interaction as Lennard‐Jones potential. To evaluate the prediction ability for protein–protein binding affinity, six test sets are examined. Sets 1–5 were used as test set in five published studies, respectively, and set 6 was the union set of sets 1–5, with a total of 86 protein–protein complexes. The correlation coefficient (R) and standard deviation (SD) of fitting predicted affinity to experimental data were calculated to compare the performance of ours with that in literature. Our predictions on sets 1–5 were as good as the best prediction reported in the published studies, and for union set 6, R = 0.76, SD = 2.24 kcal/mol. Furthermore, we found that the volume correction can significantly improve the prediction ability. This approach can also promote the research on docking and protein structure prediction.     

Protein–protein alternative binding modes do not overlap

Proteins often bind other proteins in more than one way. Thus alternative binding modes is an essential feature of protein interactions. Such binding modes may be detected by X‐ray crystallography and thus reflected in Protein Data Bank. The alternative binding is often observed not for the protein itself but for its structural homolog. The results of this study based on the analysis of a comprehensive set of co‐crystallized protein–protein complexes show that the alternative binding modes generally do not overlap, but are spatially separated. This effect is based on molecular recognition characteristics of the protein structures. The results are also in excellent agreement with the intermolecular energy funnel size estimates obtained previously by an independent methodology. The results provide an important insight into the principles of protein association, as well as potential guidelines for modeling of protein complexes and the design of protein interfaces.     

Using physical potentials and learned models to distinguish native binding interfaces from de novo designed interfaces that do not bind

Protein–protein interactions are a fundamental aspect of many biological processes. The advent of recombinant protein and computational techniques has allowed for the rational design of proteins with novel binding capabilities. It is therefore desirable to predict which designed proteins are capable of binding in vitro. To this end, we have developed a learned classification model that combines energetic and non‐energetic features. Our feature set is adapted from specialized potentials for aromatic interactions, hydrogen bonds, electrostatics, shape, and desolvation. A binding model built on these features was initially developed for CAPRI Round 21, achieving top results in the independent assessment. Here, we present a more thoroughly trained and validated model, and compare various support‐vector machine kernels. The Gaussian kernel model classified both high‐resolution complexes and designed nonbinders with 79–86% accuracy on independent test data. We also observe that multiple physical potentials for dielectric‐dependent electrostatics and hydrogen bonding contribute to the enhanced predictive accuracy, suggesting that their combined information is much greater than that of any single energetics model. We also study the change in predictive performance as the model features or training data are varied, observing unusual patterns of prediction in designed interfaces as compared with other data types. Proteins 2013; 81:1919–1930. © 2013 Wiley Periodicals, Inc.     

Detecting the native ligand orientation by interfacial rigidity: SiteInterlock

Understanding the physical attributes of protein‐ligand interfaces, the source of most biological activity, is a fundamental problem in biophysics. Knowing the characteristic features of interfaces also enables the design of molecules with potent and selective interactions. Prediction of native protein‐ligand interactions has traditionally focused on the development of physics‐based potential energy functions, empirical scoring functions that are fit to binding data, and knowledge‐based potentials that assess the likelihood of pairwise interactions. Here we explore a new approach, testing the hypothesis that protein‐ligand binding results in computationally detectable rigidification of the protein‐ligand interface. Our SiteInterlock approach uses rigidity theory to efficiently measure the relative interfacial rigidity of a series of small‐molecule ligand orientations and conformations for a number of protein complexes. In the majority of cases, SiteInterlock detects a near‐native binding mode as being the most rigid, with particularly robust performance relative to other methods when the ligand‐free conformation of the protein is provided. The interfacial rigidification of both the protein and ligand prove to be important characteristics of the native binding mode. This measure of rigidity is also sensitive to the spatial coupling of interactions and bond‐rotational degrees of freedom in the interface. While the predictive performance of SiteInterlock is competitive with the best of the five other scoring functions tested, its measure of rigidity encompasses cooperative rather than just additive binding interactions, providing novel information for detecting native‐like complexes. SiteInterlock shows special strength in enhancing the prediction of native complexes by ruling out inaccurate poses. Proteins 2016; 84:1888–1901. © 2016 Wiley Periodicals, Inc.     

Protein models docking benchmark 2

Structural characterization of protein–protein interactions is essential for our ability to understand life processes. However, only a fraction of known proteins have experimentally determined structures. Such structures provide templates for modeling of a large part of the proteome, where individual proteins can be docked by template‐free or template‐based techniques. Still, the sensitivity of the docking methods to the inherent inaccuracies of protein models, as opposed to the experimentally determined high‐resolution structures, remains largely untested, primarily due to the absence of appropriate benchmark set(s). Structures in such a set should have predefined inaccuracy levels and, at the same time, resemble actual protein models in terms of structural motifs/packing. The set should also be large enough to ensure statistical reliability of the benchmarking results. We present a major update of the previously developed benchmark set of protein models. For each interactor, six models were generated with the model‐to‐native C^α RMSD in the 1 to 6 Å range. The models in the set were generated by a new approach, which corresponds to the actual modeling of new protein structures in the “real case scenario,” as opposed to the previous set, where a significant number of structures were model‐like only. In addition, the larger number of complexes (165 vs. 63 in the previous set) increases the statistical reliability of the benchmarking. We estimated the highest accuracy of the predicted complexes (according to CAPRI criteria), which can be attained using the benchmark structures. The set is available at http://dockground.bioinformatics.ku.edu . Proteins 2015; 83:891–897. © 2015 Wiley Periodicals, Inc.     

Computational alanine scanning with linear scaling semiempirical quantum mechanical methods

Alanine scanning is a powerful experimental tool for understanding the key interactions in protein–protein interfaces. Linear scaling semiempirical quantum mechanical calculations are now sufficiently fast and robust to allow meaningful calculations on large systems such as proteins, RNA and DNA. In particular, they have proven useful in understanding protein–ligand interactions. Here we ask the question: can these linear scaling quantum mechanical methods developed for protein–ligand scoring be useful for computational alanine scanning? To answer this question, we assembled 15 protein–protein complexes with available crystal structures and sufficient alanine scanning data. In all, the data set contains ΔΔGs for 400 single point alanine mutations of these 15 complexes. We show that with only one adjusted parameter the quantum mechanics‐based methods outperform both buried accessible surface area and a potential of mean force and compare favorably to a variety of published empirical methods. Finally, we closely examined the outliers in the data set and discuss some of the challenges that arise from this examination. Proteins 2010. © 2010 Wiley‐Liss, Inc.     

Development of unified statistical potentials describing protein-protein interactions

A residue-based and a heavy atom-based statistical pair potential are developed for use in assessing the strength of protein-protein interactions. To ensure the quality of the potentials, a nonredundant, high-quality dimer database is constructed. The protein complexes in this dataset are checked by a literature search to confirm that they form multimers, and the pairwise amino acid preference to interact across a protein-protein interface is analyzed and pair potentials constructed. The performance of the residue-based potentials is evaluated by using four jackknife tests and by assessing the potentials' ability to select true protein-protein interfaces from false ones. Compared to potentials developed for monomeric protein structure prediction, the interdomain potential performs much better at distinguishing protein-protein interactions. The potential developed from homodimer interfaces is almost the same as that developed from heterodimer interfaces with a correlation coefficient of 0.92. The residue-based potential is well suited for genomic scale protein interaction prediction and analysis, such as in a recently developed threading-based algorithm, MULTIPROSPECTOR. However, the more time-consuming atom-based potential performs better in identifying near-native structures from docking generated decoys.     

A comparison of SCOP and CATH with respect to domain-domain interactions

The analysis and prediction of protein-protein interaction sites from structural data are restricted by the limited availability of structural complexes that represent the complete protein-protein interaction space. The domain classification schemes CATH and SCOP are normally used independently in the analysis and prediction of protein domain-domain interactions. In this article, the effect of different domain classification schemes on the number and type of domain-domain interactions observed in structural data is systematically evaluated for the SCOP and CATH hierarchies. Although there is a large overlap in domain assignments between SCOP and CATH, 23.6% of CATH interfaces had no SCOP equivalent and 37.3% of SCOP interfaces had no CATH equivalent in a nonredundant set. Therefore, combining both classifications gives an increase of between 23.6 and 37.3% in domain-domain interfaces. It is suggested that if possible, both domain classification schemes should be used together, but if only one is selected, SCOP provides better coverage than CATH. Employing both SCOP and CATH reduces the false negative rate of predictive methods, which employ homology matching to structural data to predict protein-protein interaction by an estimated 6.5%.     

Artificial intelligence based methods for hot spot prediction

Proteins interact through their interfaces to fulfill essential functions in the cell. They bind to their partners in a highly specific manner and form complexes that have a profound effect on understanding the biological pathways they are involved in. Any abnormal interactions may cause diseases. Therefore, the identification of small molecules which modulate protein interactions through their interfaces has high therapeutic potential. However, discovering such molecules is challenging. Most protein–protein binding affinity is attributed to a small set of amino acids found in protein interfaces known as hot spots. Recent studies demonstrate that drug-like small molecules specifically may bind to hot spots. Therefore, hot spot prediction is crucial. As experimental data accumulates, artificial intelligence begins to be used for computational hot spot prediction. First, we review machine learning and deep learning for computational hot spot prediction and then explain the significance of hot spots toward drug design.     

Charged residues at protein interaction interfaces: unexpected conservation and orchestrated divergence

Protein-protein interactions play an essential role in the functioning of cell. The importance of charged residues and their diverse role in protein-protein interactions have been well studied using experimental and computational methods. Often, charged residues located in protein interaction interfaces are conserved across the families of homologous proteins and protein complexes. However, on a large scale, it has been recently shown that charged residues are significantly less conserved than other residue types in protein interaction interfaces. The goal of this work is to understand the role of charged residues in the protein interaction interfaces through their conservation patterns. Here, we propose a simple approach where the structural conservation of the charged residue pairs is analyzed among the pairs of homologous binary complexes. Specifically, we determine a large set of homologous interactions using an interaction interface similarity measure and catalog the basic types of conservation patterns among the charged residue pairs. We find an unexpected conservation pattern, which we call the correlated reappearance, occurring among the pairs of homologous interfaces more frequently than the fully conserved pairs of charged residues. Furthermore, the analysis of the conservation patterns across different superkingdoms as well as structural classes of proteins has revealed that the correlated reappearance of charged residues is by far the most prevalent conservation pattern, often occurring more frequently than the unconserved charged residues. We discuss a possible role that the new conservation pattern may play in the long-range electrostatic steering effect.     

Cation–π, amino–π, π–π, and H‐bond interactions stabilize antigen–antibody interfaces

The identification of immunogenic regions on the surface of antigens, which are able to stimulate an immune response, is a major challenge for the design of new vaccines. Computational immunology aims at predicting such regions—in particular B‐cell epitopes—but is far from being reliably applicable on a large scale. To gain understanding into the factors that contribute to the antigen–antibody affinity and specificity, we perform a detailed analysis of the amino acid composition and secondary structure of antigen and antibody surfaces, and of the interactions that stabilize the complexes, in comparison with the composition and interactions observed in other heterodimeric protein interfaces. We make a distinction between linear and conformational B‐cell epitopes, according to whether they consist of successive residues along the polypeptide chain or not. The antigen–antibody interfaces were shown to differ from other protein–protein interfaces by their smaller size, their secondary structure with less helices and more loops, and the interactions that stabilize them: more H‐bond, cation–π, amino–π, and π–π interactions, and less hydrophobic packing; linear and conformational epitopes can clearly be distinguished. Often, chains of successive interactions, called cation/amino–π and π–π chains, are formed. The amino acid composition differs significantly between the interfaces: antigen–antibody interfaces are less aliphatic and more charged, polar and aromatic than other heterodimeric protein interfaces. Moreover, paratopes and epitopes—albeit to a lesser extent—have amino acid compositions that are distinct from general protein surfaces. This specificity holds promise for improving B‐cell epitope prediction. Proteins 2014; 82:1734–1746. © 2014 Wiley Periodicals, Inc.