Analysis of homodimeric protein interfaces by graph-spectral methods

Network Re-Wiring During Allostery and Protein-Protein Interactions: A Graph Spectral Approach

The process of allostery is often guided by subtle changes in the non-covalent interactions between residues of a protein. These changes may be brought about by minor perturbations by natural processes like binding of a ligand or protein-protein interaction. The challenge lies in capturing minute changes at the residue interaction level and following their propagation at local as well as global distances. While macromolecular effects of the phenomenon of allostery are inferred from experiments, a computational microscope can elucidate atomistic-level details leading to such macromolecular effects. Network formalism has served as an attractive means to follow this path and has been pursued further for the past couple of decades. In this chapter some concepts and methods are summarized, and recent advances are discussed. Specifically, the changes in strength of interactions (edge weight) and their repercussion on the overall protein organization (residue clustering) are highlighted. In this review, we adopt a graph spectral method to probe these subtle changes in a quantitative manner. Further, the power of this method is demonstrated for capturing re-ordering of side-chain interactions in response to ligand binding, which culminates into formation of a protein-protein complex in β₂-adrenergic receptors.

Comprehensive review and empirical analysis of hallmarks of DNA-, RNA- and protein-binding residues in protein chains

Proteins interact with a variety of molecules including proteins and nucleic acids. We review a comprehensive collection of over 50 studies that analyze and/or predict these interactions. While majority of these studies address either solely protein–DNA or protein–RNA binding, only a few have a wider scope that covers both protein–protein and protein–nucleic acid binding. Our analysis reveals that binding residues are typically characterized with three hallmarks: relative solvent accessibility (RSA), evolutionary conservation and propensity of amino acids (AAs) for binding. Motivated by drawbacks of the prior studies, we perform a large-scale analysis to quantify and contrast the three hallmarks for residues that bind DNA-, RNA-, protein- and (for the first time) multi-ligand-binding residues that interact with DNA and proteins, and with RNA and proteins. Results generated on a well-annotated data set of over 23 000 proteins show that conservation of binding residues is higher for nucleic acid- than protein-binding residues. Multi-ligand-binding residues are more conserved and have higher RSA than single-ligand-binding residues. We empirically show that each hallmark discriminates between binding and nonbinding residues, even predicted RSA, and that combining them improves discriminatory power for each of the five types of interactions. Linear scoring functions that combine these hallmarks offer good predictive performance of residue-level propensity for binding and provide intuitive interpretation of predictions. Better understanding of these residue-level interactions will facilitate development of methods that accurately predict binding in the exponentially growing databases of protein sequences.

Prediction of Protein Hotspots from Whole Protein Sequences by a Random Projection Ensemble System

Hotspot residues are important in the determination of protein-protein interactions, and they always perform specific functions in biological processes. The determination of hotspot residues is by the commonly-used method of alanine scanning mutagenesis experiments, which is always costly and time consuming. To address this issue, computational methods have been developed. Most of them are structure based, i.e., using the information of solved protein structures. However, the number of solved protein structures is extremely less than that of sequences. Moreover, almost all of the predictors identified hotspots from the interfaces of protein complexes, seldom from the whole protein sequences. Therefore, determining hotspots from whole protein sequences by sequence information alone is urgent. To address the issue of hotspot predictions from the whole sequences of proteins, we proposed an ensemble system with random projections using statistical physicochemical properties of amino acids. First, an encoding scheme involving sequence profiles of residues and physicochemical properties from the AAindex1 dataset is developed. Then, the random projection technique was adopted to project the encoding instances into a reduced space. Then, several better random projections were obtained by training an IBk classifier based on the training dataset, which were thus applied to the test dataset. The ensemble of random projection classifiers is therefore obtained. Experimental results showed that although the performance of our method is not good enough for real applications of hotspots, it is very promising in the determination of hotspot residues from whole sequences.

Specificity and stability of transient protein-protein interactions

Remarkable features that are achieved in a protein-protein complex to precise levels are stability and specificity. Deviation from the normal levels of specificity and stability, which is often caused by mutations, could result in disease conditions. Chemical nature, 3-D arrangement and dynamics of interface residues code for both specificity and stability. This article reviews roles of interfacial residues in transient protein-protein complexes. It is proposed that aside from hotspot residues conferring stability to the complex, a small set of 'rigid' residues at the interface that maintain conformation between complexed and uncomplexed forms, play a major role in conferring specificity. Exceptionally, 'super hotspot' residues, which confer both stability and specificity, are attractive sites for interaction with small molecule inhibitors.

Protein-Protein Interface and Disease: Perspective from Biomolecular Networks

Protein-protein interactions are involved in many important biological processes and molecular mechanisms of disease association. Structural studies of interfacial residues in protein complexes provide information on protein-protein interactions. Characterizing protein-protein interfaces, including binding sites and allosteric changes, thus pose an imminent challenge. With special focus on protein complexes, approaches based on network theory are proposed to meet this challenge. In this review we pay attention to protein-protein interfaces from the perspective of biomolecular networks and their roles in disease. We first describe the different roles of protein complexes in disease through several structural aspects of interfaces. We then discuss some recent advances in predicting hot spots and communication pathway analysis in terms of amino acid networks. Finally, we highlight possible future aspects of this area with respect to both methodology development and applications for disease treatment.

Characterization of Protein-Protein Interfaces through a Protein Contact Network Approach

Anthrax toxin comprises three different proteins, jointly acting to exert toxic activity: a non-toxic protective agent (PA), toxic edema factor (EF), and lethal factor (LF). Binding of PA to anthrax receptors promotes oligomerization of PA, binding of EF and LF, and then endocytosis of the complex. Homomeric forms of PA, complexes of PA bound to LF and to the endogenous receptor capillary morphogenesis gene 2 (CMG2) were analyzed. In this work, we characterized protein–protein interfaces (PPIs) and identified key residues at PPIs of complexes, by means of a protein contact network (PCN) approach. Flexibility and global and local topological properties of each PCN were computed. The vulnerability of each PCN was calculated using different node removal strategies, with reference to specific PCN topological descriptors, such as participation coefficient, contact order, and degree. The participation coefficient P, the topological descriptor of the node’s ability to intervene in protein inter-module communication, was the key descriptor of PCN vulnerability of all structures. High P residues were localized both at PPIs and other regions of complexes, so that we argued an allosteric mechanism in protein–protein interactions. The identification of residues, with key role in the stability of PPIs, has a huge potential in the development of new drugs, which would be designed to target not only PPIs but also residues localized in allosteric regions of supramolecular complexes.

PPCheck: A Webserver for the Quantitative Analysis of Protein-Protein Interfaces and Prediction of Residue Hotspots

BACKGROUND

Modeling protein-protein interactions (PPIs) using docking algorithms is useful for understanding biomolecular interactions and mechanisms. Typically, a docking algorithm generates a large number of docking poses, and it is often challenging to select the best native-like pose. A further challenge is to recognize key residues, termed as hotspots, at protein-protein interfaces, which contribute more in stabilizing a protein-protein interface.

RESULTS

We had earlier developed a computer algorithm, called PPCheck, which ascribes pseudoenergies to measure the strength of PPIs. Native-like poses could be successfully identified in 27 out of 30 test cases, when applied on a separate set of decoys that were generated using FRODOCK. PPCheck, along with conservation and accessibility scores, was able to differentiate 'native-like and non-native-like poses from 1883 decoys of Critical Assessment of Prediction of Interactions (CAPRI) targets with an accuracy of 60%. PPCheck was trained on a 10-fold mixed dataset and tested on a 10-fold mixed test set for hotspot prediction. We obtain an accuracy of 72%, which is in par with other methods, and a sensitivity of 59%, which is better than most existing methods available for hotspot prediction that uses similar datasets. Other relevant tests suggest that PPCheck can also be reliably used to identify conserved residues in a protein and to perform computational alanine scanning.

CONCLUSIONS

PPCheck webserver can be successfully used to differentiate native-like and non-native-like docking poses, as generated by docking algorithms. The webserver can also be a convenient platform for calculating residue conservation, for performing computational alanine scanning, and for predicting protein-protein interface hotspots. While PPCheck can differentiate the generated decoys into native-like and non-native-like decoys with a fairly good accuracy, the results improve dramatically when features like conservation and accessibility are included. The method can be successfully used in ranking/scoring the decoys, as obtained from docking algorithms.

Aurora-A-Dependent Control of TACC3 Influences the Rate of Mitotic Spindle Assembly

The essential mammalian gene TACC3 is frequently mutated and amplified in cancers and its fusion products exhibit oncogenic activity in glioblastomas. TACC3 functions in mitotic spindle assembly and chromosome segregation. In particular, phosphorylation on S558 by the mitotic kinase, Aurora-A, promotes spindle recruitment of TACC3 and triggers the formation of a complex with ch-TOG-clathrin that crosslinks and stabilises kinetochore microtubules. Here we map the Aurora-A-binding interface in TACC3 and show that TACC3 potently activates Aurora-A through a domain centered on F525. Vertebrate cells carrying homozygous F525A mutation in the endogenous TACC3 loci exhibit defects in TACC3 function, namely perturbed localization, reduced phosphorylation and weakened interaction with clathrin. The most striking feature of the F525A cells however is a marked shortening of mitosis, at least in part due to rapid spindle assembly. F525A cells do not exhibit chromosome missegregation, indicating that they undergo fast yet apparently faithful mitosis. By contrast, mutating the phosphorylation site S558 to alanine in TACC3 causes aneuploidy without a significant change in mitotic duration. Our work has therefore defined a regulatory role for the Aurora-A-TACC3 interaction beyond the act of phosphorylation at S558. We propose that the regulatory relationship between Aurora-A and TACC3 enables the transition from the microtubule-polymerase activity of TACC3-ch-TOG to the microtubule-crosslinking activity of TACC3-ch-TOG-clathrin complexes as mitosis progresses. Aurora-A-dependent control of TACC3 could determine the balance between these activities, thereby influencing not only spindle length and stability but also the speed of spindle formation with vital consequences for chromosome alignment and segregation.

The construction of an amino acid network for understanding protein structure and function

Amino acid networks (AANs) are undirected networks consisting of amino acid residues and their interactions in three-dimensional protein structures. The analysis of AANs provides novel insight into protein science, and several common amino acid network properties have revealed diverse classes of proteins. In this review, we first summarize methods for the construction and characterization of AANs. We then compare software tools for the construction and analysis of AANs. Finally, we review the application of AANs for understanding protein structure and function, including the identification of functional residues, the prediction of protein folding, analyzing protein stability and protein-protein interactions, and for understanding communication within and between proteins.

Ranking the quality of protein structure models using sidechain based network properties

Determining the correct structure of a protein given its sequence still remains an arduous task with many researchers working towards this goal. Most structure prediction methodologies result in the generation of a large number of probable candidates with the final challenge being to select the best amongst these. In this work, we have used Protein Structure Networks of native and modeled proteins in combination with Support Vector Machines to estimate the quality of a protein structure model and finally to provide ranks for these models. Model ranking is performed using regression analysis and helps in model selection from a group of many similar and good quality structures. Our results show that structures with a rank greater than 16 exhibit native protein-like properties while those below 10 are non-native like. The tool is also made available as a web-server ( http://vishgraph.mbu.iisc.ernet.in/GraProStr/native_non_native_ranking.html), where, 5 modelled structures can be evaluated at a given time.

Combined computational design of a zinc-binding site and a protein-protein interaction: one open zinc coordination site was not a robust hotspot for de novo ubiquitin binding

We computationally designed a de novo protein-protein interaction between wild-type ubiquitin and a redesigned scaffold. Our strategy was to incorporate zinc at the designed interface to promote affinity and orientation specificity. A large set of monomeric scaffold surfaces were computationally engineered with three-residue zinc coordination sites, and the ubiquitin residue H68 was docked to the open coordination site to complete a tetrahedral zinc site. This single coordination bond was intended as a hotspot and polar interaction for ubiquitin binding, and surrounding residues on the scaffold were optimized primarily as hydrophobic residues using a rotamer-based sequence design protocol in Rosetta. From thousands of independent design simulations, four sequences were selected for experimental characterization. The best performing design, called Spelter, binds tightly to zinc (Kd  < 10 nM) and binds ubiquitin with a Kd of 20 µM in the presence of zinc and 68 µM in the absence of zinc. Mutagenesis studies and nuclear magnetic resonance chemical shift perturbation experiments indicate that Spelter interacts with H68 and the target surface on ubiquitin; however, H68 does not form a hotspot as intended. Instead, mutation of H68 to alanine results in tighter binding. Although a 3/1 zinc coordination arrangement at an interface cannot be ruled out as a means to improve affinity, our study led us to conclude that 2/2 coordination arrangements or multiple-zinc designs are more likely to promote high-affinity protein interactions.

JC virus agnoprotein enhances large T antigen binding to the origin of viral DNA replication: evidence for its involvement in viral DNA replication

Agnoprotein is required for the successful completion of the JC virus (JCV) life cycle and was previously shown to interact with JCV large T-antigen (LT-Ag). Here, we further characterized agnoprotein's involvement in viral DNA replication. Agnoprotein enhances the DNA binding activity of LT-Ag to the viral origin (Ori) without directly interacting with DNA. The predicted amphipathic α-helix of agnoprotein plays a major role in this enhancement. All three phenylalanine (Phe) residues of agnoprotein localize to this α-helix and Phe residues in general are known to play critical roles in protein-protein interaction, protein folding and stability. The functional relevance of all Phe residues was investigated by mutagenesis. When all were mutated to alanine (Ala), the mutant virus (F31AF35AF39A) replicated significantly less efficiently than each individual Phe mutant virus alone, indicating the importance of Phe residues for agnoprotein function. Collectively, these studies indicate a close involvement of agnoprotein in viral DNA replication.

PROTEIN STRUCTURE: INSIGHTS FROM GRAPH THEORY

The sequence and structure of a large body of proteins are becoming increasingly available. It is desirable to explore mathematical tools for efficient extraction of information from such sources. The principles of graph theory, which was earlier applied in fields such as electrical engineering and computer networks are now being adopted to investigate protein structure, folding, stability, function and dynamics. This review deals with a brief account of relevant graphs and graph theoretic concepts. The concepts of protein graph construction are discussed. The manner in which graphs are analyzed and parameters relevant to protein structure are extracted, are explained. The structural and biological information derived from protein structures using these methods is presented.

Protein interaction hotspot identification using sequence-based frequency-derived features

Finding good descriptors, capable of discriminating hotspot residues from others, is still a challenge in many attempts to understand protein interaction. In this paper, descriptors issued from the analysis of amino acid sequences using digital signal processing (DSP) techniques are shown to be as good as those derived from protein tertiary structure and/or information on the complex. The simulation results show that our descriptors can be used separately to predict hotspots, via a random forest classifier, with an accuracy of 79% and a precision of 75%. They can also be used jointly with features derived from tertiary structures to boost the performance up to an accuracy of 82% and a precision of 80%.

Prediction of protein-protein interactions: unifying evolution and structure at protein interfaces

The vast majority of the chores in the living cell involve protein-protein interactions. Providing details of protein interactions at the residue level and incorporating them into protein interaction networks are crucial toward the elucidation of a dynamic picture of cells. Despite the rapid increase in the number of structurally known protein complexes, we are still far away from a complete network. Given experimental limitations, computational modeling of protein interactions is a prerequisite to proceed on the way to complete structural networks. In this work, we focus on the question 'how do proteins interact?' rather than 'which proteins interact?' and we review structure-based protein-protein interaction prediction approaches. As a sample approach for modeling protein interactions, PRISM is detailed which combines structural similarity and evolutionary conservation in protein interfaces to infer structures of complexes in the protein interaction network. This will ultimately help us to understand the role of protein interfaces in predicting bound conformations.

Dr. PIAS: an integrative system for assessing the druggability of protein-protein interactions

Background

The amount of data on protein-protein interactions (PPIs) available in public databases and in the literature has rapidly expanded in recent years. PPI data can provide useful information for researchers in pharmacology and medicine as well as those in interactome studies. There is urgent need for a novel methodology or software allowing the efficient utilization of PPI data in pharmacology and medicine.

Results

To address this need, we have developed the 'Druggable Protein-protein Interaction Assessment System' (Dr. PIAS). Dr. PIAS has a meta-database that stores various types of information (tertiary structures, drugs/chemicals, and biological functions associated with PPIs) retrieved from public sources. By integrating this information, Dr. PIAS assesses whether a PPI is druggable as a target for small chemical ligands by using a supervised machine-learning method, support vector machine (SVM). Dr. PIAS holds not only known druggable PPIs but also all PPIs of human, mouse, rat, and human immunodeficiency virus (HIV) proteins identified to date.

Conclusions

The design concept of Dr. PIAS is distinct from other published PPI databases in that it focuses on selecting the PPIs most likely to make good drug targets, rather than merely collecting PPI data.

Graphlet kernels for prediction of functional residues in protein structures

We introduce a novel graph-based kernel method for annotating functional residues in protein structures. A structure is first modeled as a protein contact graph, where nodes correspond to residues and edges connect spatially neighboring residues. Each vertex in the graph is then represented as a vector of counts of labeled non-isomorphic subgraphs (graphlets), centered on the vertex of interest. A similarity measure between two vertices is expressed as the inner product of their respective count vectors and is used in a supervised learning framework to classify protein residues. We evaluated our method on two function prediction problems: identification of catalytic residues in proteins, which is a well-studied problem suitable for benchmarking, and a much less explored problem of predicting phosphorylation sites in protein structures. The performance of the graphlet kernel approach was then compared against two alternative methods, a sequence-based predictor and our implementation of the FEATURE framework. On both tasks, the graphlet kernel performed favorably; however, the margin of difference was considerably higher on the problem of phosphorylation site prediction. While there is data that phosphorylation sites are preferentially positioned in intrinsically disordered regions, we provide evidence that for the sites that are located in structured regions, neither the surface accessibility alone nor the averaged measures calculated from the residue microenvironments utilized by FEATURE were sufficient to achieve high accuracy. The key benefit of the graphlet representation is its ability to capture neighborhood similarities in protein structures via enumerating the patterns of local connectivity in the corresponding labeled graphs.

Identification of computational hot spots in protein interfaces: combining solvent accessibility and inter-residue potentials improves the accuracy

MOTIVATION

Hot spots are residues comprising only a small fraction of interfaces yet accounting for the majority of the binding energy. These residues are critical in understanding the principles of protein interactions. Experimental studies like alanine scanning mutagenesis require significant effort; therefore, there is a need for computational methods to predict hot spots in protein interfaces.

RESULTS

We present a new intuitive efficient method to determine computational hot spots based on conservation (C), solvent accessibility [accessible surface area (ASA)] and statistical pairwise residue potentials (PP) of the interface residues. Combination of these features is examined in a comprehensive way to study their effect in hot spot detection. The predicted hot spots are observed to match with the experimental hot spots with an accuracy of 70% and a precision of 64% in Alanine Scanning Energetics Database (ASEdb), and accuracy of 70% and a precision of 73% in Binding Interface Database (BID). Several machine learning methods are also applied to predict hot spots. Performance of our empirical approach exceeds learning-based methods and other existing hot spot prediction methods. Residue occlusion from solvent in the complexes and pairwise potentials are found to be the main discriminative features in hot spot prediction.

CONCLUSION

Our empirical method is a simple approach in hot spot prediction yet with its high accuracy and computational effectiveness. We believe that this method provides insights for the researchers working on characterization of protein binding sites and design of specific therapeutic agents for protein interactions.

AVAILABILITY

The list of training and test sets are available as Supplementary Data at http://prism.ccbb.ku.edu.tr/hotpoint/supplement.doc.

SUPPLEMENTARY INFORMATION

Supplementary data are available at Bioinformatics online.

Insights into protein-DNA interactions through structure network analysis

Protein–DNA interactions are crucial for many cellular processes. Now with the increased availability of structures of protein–DNA complexes, gaining deeper insights into the nature of protein–DNA interactions has become possible. Earlier, investigations have characterized the interface properties by considering pairwise interactions. However, the information communicated along the interfaces is rarely a pairwise phenomenon, and we feel that a global picture can be obtained by considering a protein–DNA complex as a network of noncovalently interacting systems. Furthermore, most of the earlier investigations have been carried out from the protein point of view (protein-centric), and the present network approach aims to combine both the protein-centric and the DNA-centric points of view. Part of the study involves the development of methodology to investigate protein–DNA graphs/networks with the development of key parameters. A network representation provides a holistic view of the interacting surface and has been reported here for the first time. The second part of the study involves the analyses of these graphs in terms of clusters of interacting residues and the identification of highly connected residues (hubs) along the protein–DNA interface. A predominance of deoxyribose–amino acid clusters in β-sheet proteins, distinction of the interface clusters in helix–turn–helix, and the zipper-type proteins would not have been possible by conventional pairwise interaction analysis. Additionally, we propose a potential classification scheme for a set of protein–DNA complexes on the basis of the protein–DNA interface clusters. This provides a general idea of how the proteins interact with the different components of DNA in different complexes. Thus, we believe that the present graph-based method provides a deeper insight into the analysis of the protein–DNA recognition mechanisms by throwing more light on the nature and the specificity of these interactions.

The interaction of proteins with DNA is crucial for several cellular processes. Some insights into the mode of interaction can be obtained from the analysis of the complexed structures. Conventional analyses are based on the identification of pairwise interactions. However, a collective representation of the network of interactions and the analyses of such networks provide valuable information, which is not easy to obtain from pairwise analyses. Although the protein structure networks have been described in the literature, this is the first time that a network representation of protein–DNA is described. Construction and analysis of such networks have given valuable information on protein–DNA interactions in terms of network parameters, such as clusters of interacting residues and hubs, which are highly connected residues. Furthermore, the results also represent both the protein- and the DNA-centric viewpoints, because the analysis is carried out on combined networks. The methodology developed here can lead to predictions, such as important residues responsible for stabilizing protein–DNA interactions, and will be of interest to experimentalists.

Statistical analysis of physical-chemical properties and prediction of protein-protein interfaces

We have developed a fully automated method, InterProSurf, to predict interacting amino acid residues on protein surfaces of monomeric 3D structures. Potential interacting residues are predicted based on solvent accessible surface areas, a new scale for interface propensities, and a cluster algorithm to locate surface exposed areas with high interface propensities. Previous studies have shown the importance of hydrophobic residues and specific charge distribution as characteristics for interfaces. Here we show differences in interface and surface regions of all physical chemical properties of residues as represented by five quantitative descriptors. In the current study a set of 72 protein complexes with known 3D structures were analyzed to obtain interface propensities of residues, and to find differences in the distribution of five quantitative descriptors for amino acid residues. We also investigated spatial pair correlations of solvent accessible residues in interface and surface areas, and compared log-odds ratios for interface and surface areas. A new scoring method to predict potential functional sites on the protein surface was developed and tested for a new dataset of 21 protein complexes, which were not included in the original training dataset. Empirically we found that the algorithm achieves a good balance in the accuracy of precision and sensitivity by selecting the top eight highest scoring clusters as interface regions. The performance of the method is illustrated for a dimeric ATPase of the hyperthermophile, Methanococcus jannaschii, and the capsid protein of Human Hepatitis B virus. An automated version of the method can be accessed from our web server at http://curie.utmb.edu/prosurf.html.

Peptide segments in protein-protein interfaces

An important component of functional genomics involves the understanding of protein association. The interfaces resulting from protein-protein interactions - (i) specific, as represented by the homodimeric quaternary structures and the complexes formed by two independently occurring protein components, and (ii) non-specific, as observed in the crystal lattice of monomeric proteins - have been analysed on the basis of the length and the number of peptide segments. In 1000 A2 of the interface area, contributed by a polypeptide chain, there would be 3.4 segments in homodimers, 5.6 in complexes and 6.3 in crystal contacts. Concomitantly, the segments are the longest (with 8.7 interface residues) in homodimers. Core segments (likely to contribute more towards binding) are more in number in homodimers (1.7) than in crystal contacts (0.5), and this number can be used as one of the parameters to distinguish between the two types of interfaces. Dominant segments involved in specific interactions, along with their secondary structural features, are enumerated.

Association of putative concave protein-binding sites with the fluctuation behavior of residues

Here, we propose a binding site prediction method based on the high frequency end of the spectrum in the native state of the protein structural dynamics. The spectrum is obtained using an elastic network model (GNM). High frequency vibrating (HFV) residues are determined from the fastest modes dynamics. HFV residue clusters and the associated surface patch residues are tested for their likelihood to locate at the binding interfaces using two different data sets, the Benchmark Set of mainly enzymes and antigen/antibodies and the Cluster Set of more diverse structures. The binding interface is identified to be within 7.5 A of the HFV residue clusters in the Benchmark Set and Cluster Set, for 77% and 70% of the structures, respectively. The success rate increases to 88% and 84%, respectively, by using the surface patches. The results suggest that concave binding interfaces, typically those of enzyme-binding sites, are enriched by the HFV residues. Thus, we expect that the association of HFV residues with the interfaces is mostly for enzymes. If, however, a binding region has invaginations and cavities, as in some of the antigen/antibodies and in cases in the Cluster data set, we expect it would be detected there too. This implies that binding sites possess several (inter-related) properties such as cavities, high packing density, conservation, and disposition for hotspots at binding surfaces. It further suggests that the high frequency vibrating residue-based approach is a potential tool for identification of regions likely to serve as protein-binding sites. The software is available at http://www.prc.boun.edu.tr/PRC/software.html.

Binding Hot Spots in the TEM1–BLIP Interface in Light of its Modular Architecture

Proteins bind one another in aqua's solution to form tight and specific complexes. Previously we have shown that this is achieved through the modular architecture of the interaction network formed by the interface residues, where tight cooperative interactions are found within modules but not between them. Here we extend this study to cover the entire interface of TEM1 beta-lactamase and its protein inhibitor BLIP using an improved method for deriving interaction maps based on REDUCE to add hydrogen atoms and then by evaluating the interactions using modifications of the programs PROBE, NCI and PARE. An extensive mutagenesis study of the interface residues indeed showed that each module is energetically independent on other modules, and that cooperativity is found only within a module. By solving the X-ray structure of two interface mutations affecting two different modules, we demonstrated that protein-protein binding occur via the structural reorganization of the binding modules, either by a "lock and key" or an induced fit mechanism. To explain the cooperativity within a module, we performed multiple-mutant cycle analysis of cluster 2 resulting in a high-resolution energy map of this module. Mutant studies are usually done in reference to alanine, which can be regarded as a deletion of a side-chain. However, from a biological perspective, there is a major interest to understand non-Ala substitutions, as they are most common. Using X-ray crystallography and multiple-mutant cycle analysis we demonstrated the added complexity in understanding non-Ala mutations. Here, a double mutation replacing the wild-type Glu,Tyr to Tyr,Asn on TEM1 (res id 104,105) caused a major backbone structural rearrangement of BLIP, changing the composition of two modules but not of other modules within the interface. This shows the robustness of the modular approach, yet demonstrates the complexity of in silico protein design.

Interaction interfaces of protein domains are not topologically equivalent across families within superfamilies: Implications for metabolic and signaling pathways

Using a data set of aligned protein domain superfamilies of known three-dimensional structure, we compared the location of interdomain interfaces on the tertiary folds between members of distantly related protein domain superfamilies. The data set analyzed is comprised of interdomain interfaces, with domains occurring within a polypeptide chain and those between two polypeptide chains. We observe that, in general, the interfaces between protein domains are formed entirely in different locations on the tertiary folds in such pairs. This variation in the location of interface happens in protein domains involved in a wide range of functions, such as enzymes, adapters, and domains that bind protein ligands, or cofactors. While basic biochemical functionality is preserved at the domain superfamily level, the effect of biochemical function on protein assemblies is different in these protein domains related by superfamily. The divergence between proteins, in most cases, is coupled with domain recruitment, with different modes of interaction with the recruited domain. This is in complete contrast to the observation that in closely related homologous protein domains, almost always the interaction interfaces are topologically equivalent. In a small subset of interacting domains within proteins related by remote homology, we observe that the relative positioning of domains with respect to one another is preserved. Based on the analysis of multidomain proteins of known or unknown structure, we suggest that variation in protein-protein interactions in members within a superfamily could serve as diverging points in otherwise parallel metabolic or signaling pathways. We discuss a few representative cases of diverging pathways involving domains in a superfamily.

Small-world network approach to identify key residues in protein-protein interaction

We show that protein complexes can be represented as small-world networks, exhibiting a relatively small number of highly central amino-acid residues occurring frequently at protein-protein interfaces. We further base our analysis on a set of different biological examples of protein-protein interactions with experimentally validated hot spots, and show that 83% of these predicted highly central residues, which are conserved in sequence alignments and nonexposed to the solvent in the protein complex, correspond to or are in direct contact with an experimentally annotated hot spot. The remaining 17% show a general tendency to be close to an annotated hot spot. On the other hand, although there is no available experimental information on their contribution to the binding free energy, detailed analysis of their properties shows that they are good candidates for being hot spots. Thus, highly central residues have a clear tendency to be located in regions that include hot spots. We also show that some of the central residues in the protein complex interfaces are central in the monomeric structures before dimerization and that possible information relating to hot spots of binding free energy could be obtained from the unbound structures.

Insights into the quaternary association of proteins through structure graphs: a case study of lectins

The unique three-dimensional structure of both monomeric and oligomeric proteins is encoded in their sequence. The biological functions of proteins are dependent on their tertiary and quaternary structures, and hence it is important to understand the determinants of quaternary association in proteins. Although a large number of investigations have been carried out in this direction, the underlying principles of protein oligomerization are yet to be completely understood. Recently, new insights into this problem have been gained from the analysis of structure graphs of proteins belonging to the legume lectin family. The legume lectins are an interesting family of proteins with very similar tertiary structures but varied quaternary structures. Hence they have become a very good model with which to analyse the role of primary structures in determining the modes of quaternary association. The present review summarizes the results of a legume lectin study as well as those obtained from a similar analysis carried out here on the animal lectins, namely galectins, pentraxins, calnexin, calreticulin and rhesus rotavirus Vp4 sialic-acid-binding domain. The lectin structure graphs have been used to obtain clusters of non-covalently interacting amino acid residues at the intersubunit interfaces. The present study, performed along with traditional sequence alignment methods, has provided the signature sequence motifs for different kinds of quaternary association seen in lectins. Furthermore, the network representation of the lectin oligomers has enabled us to detect the residues which make extensive interactions ('hubs') across the oligomeric interfaces that can be targetted for interface-destabilizing mutations. The present review also provides an overview of the methodology involved in representing oligomeric protein structures as connected networks of amino acid residues. Further, it illustrates the potential of such a representation in elucidating the structural determinants of protein-protein association in general and will be of significance to protein chemists and structural biologists.

Oligomeric protein structure networks: insights into protein-protein interactions

BACKGROUND

Protein-protein association is essential for a variety of cellular processes and hence a large number of investigations are being carried out to understand the principles of protein-protein interactions. In this study, oligomeric protein structures are viewed from a network perspective to obtain new insights into protein association. Structure graphs of proteins have been constructed from a non-redundant set of protein oligomer crystal structures by considering amino acid residues as nodes and the edges are based on the strength of the non-covalent interactions between the residues. The analysis of such networks has been carried out in terms of amino acid clusters and hubs (highly connected residues) with special emphasis to protein interfaces.

RESULTS

A variety of interactions such as hydrogen bond, salt bridges, aromatic and hydrophobic interactions, which occur at the interfaces are identified in a consolidated manner as amino acid clusters at the interface, from this study. Moreover, the characterization of the highly connected hub-forming residues at the interfaces and their comparison with the hubs from the non-interface regions and the non-hubs in the interface regions show that there is a predominance of charged interactions at the interfaces. Further, strong and weak interfaces are identified on the basis of the interaction strength between amino acid residues and the sizes of the interface clusters, which also show that many protein interfaces are stronger than their monomeric protein cores. The interface strengths evaluated based on the interface clusters and hubs also correlate well with experimentally determined dissociation constants for known complexes. Finally, the interface hubs identified using the present method correlate very well with experimentally determined hotspots in the interfaces of protein complexes obtained from the Alanine Scanning Energetics database (ASEdb). A few predictions of interface hot spots have also been made based on the results obtained from this analysis, which await experimental verification.

CONCLUSION

The construction and analysis of oligomeric protein structure networks and their comparison with monomeric protein structure networks provide insights into protein association. Further, the interface hubs identified using the present method can be effective targets for interface de-stabilizing mutations. We believe this analysis will significantly enhance our knowledge of the principles behind protein association and also aid in protein design.

A network representation of protein structures: implications for protein stability

This study views each protein structure as a network of noncovalent connections between amino acid side chains. Each amino acid in a protein structure is a node, and the strength of the noncovalent interactions between two amino acids is evaluated for edge determination. The protein structure graphs (PSGs) for 232 proteins have been constructed as a function of the cutoff of the amino acid interaction strength at a few carefully chosen values. Analysis of such PSGs constructed on the basis of edge weights has shown the following: 1), The PSGs exhibit a complex topological network behavior, which is dependent on the interaction cutoff chosen for PSG construction. 2), A transition is observed at a critical interaction cutoff, in all the proteins, as monitored by the size of the largest cluster (giant component) in the graph. Amazingly, this transition occurs within a narrow range of interaction cutoff for all the proteins, irrespective of the size or the fold topology. And 3), the amino acid preferences to be highly connected (hub frequency) have been evaluated as a function of the interaction cutoff. We observe that the aromatic residues along with arginine, histidine, and methionine act as strong hubs at high interaction cutoffs, whereas the hydrophobic leucine and isoleucine residues get added to these hubs at low interaction cutoffs, forming weak hubs. The hubs identified are found to play a role in bringing together different secondary structural elements in the tertiary structure of the proteins. They are also found to contribute to the additional stability of the thermophilic proteins when compared to their mesophilic counterparts and hence could be crucial for the folding and stability of the unique three-dimensional structure of proteins. Based on these results, we also predict a few residues in the thermophilic and mesophilic proteins that can be mutated to alter their thermal stability.

Protein subunit interfaces: heterodimers versus homodimers

Protein dimers are either homodimers (complexation of identical monomers) or heterodimers (complexation of non-identical monomers). These dimers are common in catalysis and regulation. However, the molecular principles of protein dimer interactions are difficult to understand mainly due to the geometrical and chemical characteristics of proteins. Nonetheless, the principles of protein dimer interactions are often studied using a dataset of 3D structural complexes determined by X-ray crystallography. A number of physical and chemical properties govern protein dimer interactions. Yet, a handful of such properties are known to dominate protein dimer interfaces. Here, we discuss the differences between homodimer and heterodimer interfaces using a selected set of interface properties.

Determinants of quaternary association in legume lectins

It is well known that the sequence of amino acids in proteins code for its tertiary structure. It is also known that there exists a relationship between sequence and the quaternary structure of proteins. The question addressed here is whether the nature of quaternary association can be predicted from the sequence, similar to the three-dimensional structure prediction from the sequence. The class of proteins called legume lectins is an interesting model system to investigate this problem, because they have very high sequence and tertiary structure homology, with diverse forms of quaternary association. Hence, we have used legume lectins as a probe in this paper to (1) gain novel insights about the relationship between sequence and quaternary structure; (2) identify the sequence motifs that are characteristic of a given type of quaternary association; and (3) predict the quaternary association from the sequence motif.

Topology of small-world networks of protein-protein complex structures

The majority of real examples of small-world networks exhibit a power law distribution of edges among the nodes, therefore not fitting into the wiring model proposed by Watts and Strogatz. However, protein structures can be modeled as small-world networks, with a distribution of the number of links decaying exponentially as in the case of this wiring model. We approach the protein-protein interaction mechanism by viewing it as a particular rewiring occurring in the system of two small-world networks represented by the monomers, where a re-arrangement of links takes place upon dimerization leaving the small-world character in the dimer network. Due to this rewiring, the most central residues at the complex interfaces tend to form clusters, which are not homogenously distributed. We show that these highly central residues are strongly correlated with the presence of hot spots of binding free energy.

CONTACT

ao-mesa@fujirebio.co.jp

SUPPLEMENTARY INFORMATION

http://www.fujirebio.co.jp/support/index.php (under construction).

Protein--protein recognition: juxtaposition of domain and interface cores in immunoglobulins and other sandwich-like proteins

Structural analysis of a non-redundant data set of 47 immunoglobulin (Ig) proteins was carried out using a combination of criteria: atom--atom contact compatibility, position occupancy rate, conservation of residue type and positional conservation in 3D space. Our analysis shows that roughly half of the interface positions between the light and heavy chains are specific to individual structures while the other half are conserved across the database. The tendency for conservation of a primary subset of positions holds true for the intra-domain faces as well. These subsets, with an average of 12 conserved positions and a contact surface of 630 A(2), delineate the inter- and intra-domain core, a refined instrument with a reduced target for analysis of sheet--sheet interactions in sandwich-like proteins. Employing this instrument, we find that a majority of Ig interface core positions are adjoined in sequence to domain core positions. This was derived independent of geometric considerations, however beta-sheet side-chain geometry clearly dictates it. The geometric wedding of the domain and interface cores supports the concept of a rigid-like substructure on the protein surface involved in complex formation and indicates a close relationship between surface determinants and those involved in protein folding of Ig domains. The definitions developed for the Ig interface and domain cores proved satisfactory to extract first-approximation cores for a group of 24 non-Ig sandwich-like proteins, treated as individual structures due to their diverse strand topologies. We show that the same rule of positional connectivity between the rigid domain core and interface core extends generally to sandwich-like proteins interacting in a sheet--sheet fashion. The non-Ig structures were used as templates to analyze sandwich-like interfaces of unresolved homologous proteins using a database merging structure and sequence conservation.

Interaction of DNA with clusters of amino acids in proteins

Protein-DNA interactions facilitate the fundamental functions of living cells and are universal in all living organisms. Several investigations have been carried out, essentially identifying pairs of interactions between the amino acid residues in proteins and the bases in DNA. In the present study, we have detected the recognition motifs that may constitute a cluster of spatially interacting residues in proteins, which interact with the bases of DNA. Graph spectral algorithm has been used to detect side chain clusters comprising Arg, Lys, Asn, Gln and aromatic residues from proteins interacting with DNA. We find that the interaction of proteins with DNA is through clusters in about half of the proteins in the dataset and through individual residues in the rest. Furthermore, inspection of the clusters has revealed additional interactions in a few cases, which have not been reported earlier. The geometry of the interaction between the DNA base and the protein residue is quantified by the distance d and the angle theta. These parameters have been identified for the cation-pi/H-bond stair motif that was reported earlier. Among the Arg, Lys, Asn and Gln residues, the range of (d, theta) values of the interacting Arg clearly falls into the cation-pi and the hydrogen bond interactions of the 'cation-pi/H-bond' stair motif. Analysis of the cluster composition reveals that the Arg residue is predominant than the Lys, Asn and Gln residues. The clusters are classified into Type I and Type II based on the presence or absence of aromatic residues (Phe, Tyr) in them. Residue conservation in these clusters has been examined. Apart from the conserved residues identified previously, a few more residues mainly Phe, Tyr and Arg have also been identified as conserved and interactive with the DNA. Interestingly, a few residues that are parts of interacting clusters and do not interact directly with the DNA have also been conserved. This emphasizes the importance of recognizing the protein side chain cluster motifs interacting with the DNA, which could serve as signatures of protein-DNA recognition in the families of DNA binding proteins.

Characterization of the amino acid contribution to the folding degree of proteins

The folding degree index (Estrada, Bioinformatics 2002;18:697-704) is extended to account for the contribution of amino acids to folding. First, the mathematical formalism for extending the folding degree index is presented. Then, the amino acid contributions to folding degree of several proteins are used to analyze its relation to secondary structure. The possibilities of using these contributions in helping or checking the assignation of secondary structure to amino acids are also introduced. The influence of external factors to the amino acids contribution to folding degree is studied through the temperature effect on ribonuclease A. Finally, the analysis of 3D protein similarity through the use of amino acid contributions to folding degree is studied by selecting a series of lysozymes. These results are compared to that obtained by sequence alignment (2D similarity) and 3D superposition of the structures, showing the uniqueness of the current approach.

Application of a novel graph-theoretic folding degree index to the study of steroid-DB3 antibody binding affinity

A novel folding degree index, together with other macromolecular descriptors, is used to study steroid-DB3 antibody interactions. This index is based on graph spectral moments of a matrix representing the dihedral angles of a protein backbone. The causes influencing the different order of binding affinity of steroids to DB3 antibody are identified. It is shown that the changes in the chain compactness of the DB3 antibody with respect to its center of mass (radius of gyration) is compensated by a change in the folding degree index in the contrary sense. In fact, the increment in compactness of chain L and the lower increment in the folding degree index of chain H are able to explain the variations in affinity for DB3 of the steroids studied. Consequently, the highest binding affinities are reached by increasing the compactness of chain L in DB3 at the same time that producing the smallest increment in the folding degree of chain H. This study shows the possibilities of application for the graph-theoretic folding degree index in studying drug-protein interactions.

Protein-protein interactions: structurally conserved residues distinguish between binding sites and exposed protein surfaces

Polar residue hot spots have been observed at protein-protein binding sites. Here we show that hot spots occur predominantly at the interfaces of macromolecular complexes, distinguishing binding sites from the remainder of the surface. Consequently, hot spots can be used to define binding epitopes. We further show a correspondence between energy hot spots and structurally conserved residues. The number of structurally conserved residues, particularly of high ranking energy hot spots, increases with the binding site contact size. This finding may suggest that effectively dispersing hot spots within a large contact area, rather than compactly clustering them, may be a strategy to sustain essential key interactions while still allowing certain protein flexibility at the interface. Thus, most conserved polar residues at the binding interfaces confer rigidity to minimize the entropic cost on binding, whereas surrounding residues form a flexible cushion. Furthermore, our finding that similar residue hot spots occur across different protein families suggests that affinity and specificity are not necessarily coupled: higher affinity does not directly imply greater specificity. Conservation of Trp on the protein surface indicates a highly likely binding site. To a lesser extent, conservation of Phe and Met also imply a binding site. For all three residues, there is a significant conservation in binding sites, whereas there is no conservation on the exposed surface. A hybrid strategy, mapping sequence alignment onto a single structure illustrates the possibility of binding site identification around these three residues.