Interface-resolved network of protein-protein interactions

Integrating protein copy numbers with interaction networks to quantify stoichiometry in clathrin-mediated endocytosis

Proteins that drive processes like clathrin-mediated endocytosis (CME) are expressed at copy numbers within a cell and across cell types varying from hundreds (e.g. auxilin) to millions (e.g. clathrin). These variations contain important information about function, but without integration with the interaction network, they cannot capture how supply and demand for each protein depends on binding to shared and distinct partners. Here we construct the interface-resolved network of 82 proteins involved in CME and establish a metric, a stoichiometric balance ratio (SBR), that quantifies whether each protein in the network has an abundance that is sub- or super-stoichiometric dependent on the global competition for binding. We find that highly abundant proteins (like clathrin) are super-stoichiometric, but that not all super-stoichiometric proteins are highly abundant, across three cell populations (HeLa, fibroblast, and neuronal synaptosomes). Most strikingly, within all cells there is significant competition to bind shared sites on clathrin and the central AP-2 adaptor by other adaptor proteins, resulting in most being in excess supply. Our network and systematic analysis, including response to perturbations of network components, show how competition for shared binding sites results in functionally similar proteins having widely varying stoichiometries, due to variations in both abundance and their unique network of binding partners.

Improved sequence-based prediction of interaction sites in α-helical transmembrane proteins by deep learning

•

Fast and accurate prediction of transmembrane protein interaction sites.

•

First ever computational survey of interaction sites in membrane proteins.

•

10-30% of amino acid positions predicted to be involved in interactions.

Interactions between transmembrane (TM) proteins are fundamental for a wide spectrum of cellular functions, but precise molecular details of these interactions remain largely unknown due to the scarcity of experimentally determined three-dimensional complex structures. Computational techniques are therefore required for a large-scale annotation of interaction sites in TM proteins. Here, we present a novel deep-learning approach, DeepTMInter, for sequence-based prediction of interaction sites in α-helical TM proteins based on their topological, physiochemical, and evolutionary properties. Using a combination of ultra-deep residual neural networks with a stacked generalization ensemble technique DeepTMInter significantly outperforms existing methods, achieving the AUC/AUCPR values of 0.689/0.598. Across the main functional families of human transmembrane proteins, the percentage of amino acid sites predicted to be involved in interactions typically ranges between 10% and 25%, and up to 30% in ion channels. DeepTMInter is available as a standalone package at https://github.com/2003100127/deeptminter. The training and benchmarking datasets are available at https://data.mendeley.com/datasets/2t8kgwzp35.

Comparative dynamics of tropomyosin in vertebrates and invertebrates

Tropomyosin (Tpm) is an extended α-helical coiled-coil homodimer that regulates actinomyosin interactions in muscle. Molecular simulations of four Tpms, two from the vertebrate class Mammalia (rat and pig), and two from the invertebrate class Malacostraca (shrimp and lobster), showed that despite extensive sequence and structural homology across metazoans, dynamic behavior-particularly long-range structural fluctuations-were clearly distinct. Vertebrate Tpms were more flexible and sampled complex, multi-state conformational landscapes. Invertebrate Tpms were more rigid, sampling a highly constrained harmonic landscape. Filtering of trajectories by principle component analysis into essential subspaces showed significant overlap within but not between phyla. In vertebrate Tpms, hinge-regions decoupled long-range interhelical motions and suggested distinct domains. In contrast, crustacean Tpms did not exhibit long-range dynamic correlations-behaving more like a single rigid rod on the nanosecond time scale. These observations suggest there may be divergent mechanisms for Tpm binding to actin filaments, where conformational flexibility in mammalian Tpm allows a preorganized shape complementary to the filament surface, and where rigidity in the crustacean Tpm requires concerted bending and binding.

Match fitness landscapes for macromolecular interaction networks: Selection for translational accuracy and rate can displace tRNA-binding interfaces of non-cognate aminoacyl-tRNA synthetases

Advances in structural biology of aminoacyl-tRNA synthetases (aaRSs) have revealed incredible diversity in how aaRSs bind their tRNA substrates. The causes of this diversity remain mysterious. We developed a new class of highly rugged fitness landscape models called match landscapes, through which genes encode the assortative interactions of their gene products through the complementarity and identifiability of their structural features. We used results from coding theory to prove bounds and equalities on fitness in match landscapes assuming additive interaction energies, macroscopic aminoacylation kinetics including proofreading, site-specific modifiers of interaction, and selection for translational accuracy in multiple, perfectly encoded site-types. Using genotypes based on extended Hamming codes we show that over a wide array of interface sizes and numbers of encoded cognate pairs, selection for translational accuracy alone is insufficient to displace the tRNA-binding interfaces of aaRSs. Yet, under combined selection for translational accuracy and rate, site-specific modifiers are selected to adaptively displace the tRNA-binding interfaces of non-cognate aaRS-tRNA pairs. We describe a remarkable correspondence between the lengths of perfect RNA (quaternary) codes and the modal sizes of small non-coding RNA families.

PPI inhibitor and stabilizer development in human diseases

All processes in living organisms are regulated by, or at least influenced by, protein-protein interactions (PPI). Membrane proteins play a fundamental part in this class of interactions: by providing inter-cellular communication and sensing capabilities to the cell, they lead to downstream regulation signaling events. It is therefore not surprising that PPI modulators are of keen interest when developing drug-like molecules for a range of diseases and medical conditions. However, techniques for exploiting PPIs in meaningful ways have only recently become readily available. This review is meant to provide a brief overview of applied techniques for PPI elucidation, and present various case studies of PPI exploitation ranging from early discovery efforts to now-approved market drugs.

Stoichiometric balance of protein copy numbers is measurable and functionally significant in a protein-protein interaction network for yeast endocytosis

Stoichiometric balance, or dosage balance, implies that proteins that are subunits of obligate complexes (e.g. the ribosome) should have copy numbers expressed to match their stoichiometry in that complex. Establishing balance (or imbalance) is an important tool for inferring subunit function and assembly bottlenecks. We show here that these correlations in protein copy numbers can extend beyond complex subunits to larger protein-protein interactions networks (PPIN) involving a range of reversible binding interactions. We develop a simple method for quantifying balance in any interface-resolved PPINs based on network structure and experimentally observed protein copy numbers. By analyzing such a network for the clathrin-mediated endocytosis (CME) system in yeast, we found that the real protein copy numbers were significantly more balanced in relation to their binding partners compared to randomly sampled sets of yeast copy numbers. The observed balance is not perfect, highlighting both under and overexpressed proteins. We evaluate the potential cost and benefits of imbalance using two criteria. First, a potential cost to imbalance is that ‘leftover’ proteins without remaining functional partners are free to misinteract. We systematically quantify how this misinteraction cost is most dangerous for strong-binding protein interactions and for network topologies observed in biological PPINs. Second, a more direct consequence of imbalance is that the formation of specific functional complexes depends on relative copy numbers. We therefore construct simple kinetic models of two sub-networks in the CME network to assess multi-protein assembly of the ARP2/3 complex and a minimal, nine-protein clathrin-coated vesicle forming module. We find that the observed, imperfectly balanced copy numbers are less effective than balanced copy numbers in producing fast and complete multi-protein assemblies. However, we speculate that strategic imbalance in the vesicle forming module allows cells to tune where endocytosis occurs, providing sensitive control over cargo uptake via clathrin-coated vesicles.

Protein copy numbers are often found to be stoichiometrically balanced for subunits of multi-protein complexes. Imbalance is believed to be deleterious because it lowers complex yield (the dosage balance hypothesis) and increases the risk of misinteractions, but imbalance may also provide unexplored functional benefits. We show here that the benefits of stoichiometric balance can extend to larger networks of interacting proteins. We develop a method to quantify to what degree protein networks are balanced, and apply it to two networks. We find that the clathrin-mediated endocytosis system in yeast is statistically balanced, but not perfectly so, and explore the consequences of imbalance in the form of misinteractions and endocytic function. We also show that biological networks are more robust to misinteractions than random networks when balanced, but are more sensitive to misinteractions under imbalance. This suggests evolutionary pressure for proteins to be balanced and that any conserved imbalance should occur for functional reasons. We explore one such reason in the form of bottlenecking the endocytosis process. Our method can be generalized to other networks and used to identify out-of-balance proteins. Our results provide insight into how network design, expression level regulation, and cell fitness are intertwined.

Cytosolic proteins can exploit membrane localization to trigger functional assembly

Cell division, endocytosis, and viral budding would not function without the localization and assembly of protein complexes on membranes. What is poorly appreciated, however, is that by localizing to membranes, proteins search in a reduced space that effectively drives up concentration. Here we derive an accurate and practical analytical theory to quantify the significance of this dimensionality reduction in regulating protein assembly on membranes. We define a simple metric, an effective equilibrium constant, that allows for quantitative comparison of protein-protein interactions with and without membrane present. To test the importance of membrane localization for driving protein assembly, we collected the protein-protein and protein-lipid affinities, protein and lipid concentrations, and volume-to-surface-area ratios for 46 interactions between 37 membrane-targeting proteins in human and yeast cells. We find that many of the protein-protein interactions between pairs of proteins involved in clathrin-mediated endocytosis in human and yeast cells can experience enormous increases in effective protein-protein affinity (10–1000 fold) due to membrane localization. Localization of binding partners thus triggers robust protein complexation, suggesting that it can play an important role in controlling the timing of endocytic protein coat formation. Our analysis shows that several other proteins involved in membrane remodeling at various organelles have similar potential to exploit localization. The theory highlights the master role of phosphoinositide lipid concentration, the volume-to-surface-area ratio, and the ratio of 3D to 2D equilibrium constants in triggering (or preventing) constitutive assembly on membranes. Our simple model provides a novel quantitative framework for interpreting or designing in vitro experiments of protein complexation influenced by membrane binding.

In a multitude of cellular processes, including cell division and endocytosis, proteins must bind to one another to form large multi-protein complexes. To initiate the formation of these critical multi-protein assemblies at the right time and the right place, the constituent proteins must be present at sufficient concentrations. We show here that membrane localization offers a powerful way of controlling protein concentrations by reducing the dimensionality of the protein’s search space. We present a simple and practical analytical theory that determines the significance of membrane localization for triggering protein-protein interactions. We show that protein binding partners will often form substantially more complexes when both partners can localize to surfaces, and thus localization can regulate the timing of multi-protein assembly. We collect in vitro binding data and cellular concentrations of proteins and lipids involved in pathways including clathrin-mediated endocytosis to demonstrate how cellular proteins could exploit membrane localization to regulate assembly.

Properties that rank protein:protein docking poses with high accuracy

The development of docking algorithms to predict near-native structures of protein:protein complexes from the structure of the isolated monomers is of paramount importance for molecular biology and drug discovery.

Engineering strategies for rational polyketide synthase design

In this review, we highlight strategies in engineering polyketide synthases (PKSs). We focus on important protein–protein interactions that constitute an intact PKS assembly line.

Switch-like Arp2/3 activation upon WASP and WIP recruitment to an apparent threshold level by multivalent linker proteins in vivo

Actin-related protein 2/3 (Arp2/3) complex activation by nucleation promoting factors (NPFs) such as WASP, plays an important role in many actin-mediated cellular processes. In yeast, Arp2/3-mediated actin filament assembly drives endocytic membrane invagination and vesicle scission. Here we used genetics and quantitative live-cell imaging to probe the mechanisms that concentrate NPFs at endocytic sites, and to investigate how NPFs regulate actin assembly onset. Our results demonstrate that SH3 (Src homology 3) domain-PRM (proline-rich motif) interactions involving multivalent linker proteins play central roles in concentrating NPFs at endocytic sites. Quantitative imaging suggested that productive actin assembly initiation is tightly coupled to accumulation of threshold levels of WASP and WIP, but not to recruitment kinetics or release of autoinhibition. These studies provide evidence that WASP and WIP play central roles in establishment of a robust multivalent SH3 domain-PRM network in vivo, giving actin assembly onset at endocytic sites a switch-like behavior.

DOI:http://dx.doi.org/10.7554/eLife.29140.001

Actin is one of the most abundant proteins in yeast, mammalian and other eukaryotic cells. It assembles into long chains known as filaments that the cell uses to generate forces for various purposes. For example, actin filaments are needed to pull part of the membrane surrounding the cell inwards to bring molecules from the external environment into the cell by a process called endocytosis.

In yeast, a member of the WASP family of proteins promotes the assembly of actin filaments around the site where endocytosis will occur. To achieve this, WASP interacts with several other proteins including WIP and myosin, a motor protein that moves along actin filaments to generate mechanical forces. However, it was not clear how these proteins work together to trigger actin filaments to assemble at the right place and time.

Sun et al. addressed this question by studying yeast cells with genetic mutations affecting one or more of these proteins. The experiments show that WASP, myosin and WIP are recruited to sites where endocytosis is about to occur through specific interactions with other proteins. For example, a region of WASP known as the proline-rich domain can bind to proteins that contain an “SH3” domain. WASP and WIP arrive first, stimulating actin to assemble in an “all and nothing” manner and attracting myosin to the actin. Further experiments indicate that WASP and WIP need to reach a threshold level before actin starts to assemble.

The findings of Sun et al. suggest that WASP and WIP play key roles in establishing the network of proteins needed for actin filaments to assemble during endocytosis. These proteins are needed for many other processes in yeast and other cells, including mammalian cells. Therefore, the next steps will be to investigate whether WASP and WIP use the same mechanism to operate in other situations.

DOI:http://dx.doi.org/10.7554/eLife.29140.002

Protein-protein binding selectivity and network topology constrain global and local properties of interface binding networks

Protein-protein interactions networks (PPINs) are known to share a highly conserved structure across all organisms. What is poorly understood, however, is the structure of the child interface interaction networks (IINs), which map the binding sites proteins use for each interaction. In this study we analyze four independently constructed IINs from yeast and humans and find a conserved structure of these networks with a unique topology distinct from the parent PPIN. Using an IIN sampling algorithm and a fitness function trained on the manually curated PPINs, we show that IIN topology can be mostly explained as a balance between limits on interface diversity and a need for physico-chemical binding complementarity. This complementarity must be optimized both for functional interactions and against mis-interactions, and this selectivity is encoded in the IIN motifs. To test whether the parent PPIN shapes IINs, we compared optimal IINs in biological PPINs versus random PPINs. We found that the hubs in biological networks allow for selective binding with minimal interfaces, suggesting that binding specificity is an additional pressure for a scale-free-like PPIN. We confirm through phylogenetic analysis that hub interfaces are strongly conserved and rewiring of interactions between proteins involved in endocytosis preserves interface binding selectivity.

Gene Prioritization by Integrated Analysis of Protein Structural and Network Topological Properties for the Protein-Protein Interaction Network of Neurological Disorders

Neurological disorders are known to show similar phenotypic manifestations like anxiety, depression, and cognitive impairment. There is a need to identify shared genetic markers and molecular pathways in these diseases, which lead to such comorbid conditions. Our study aims to prioritize novel genetic markers that might increase the susceptibility of patients affected with one neurological disorder to other diseases with similar manifestations. Identification of pathways involving common candidate markers will help in the development of improved diagnosis and treatments strategies for patients affected with neurological disorders. This systems biology study for the first time integratively uses 3D-structural protein interface descriptors and network topological properties that characterize proteins in a neurological protein interaction network, to aid the identification of genes that are previously not known to be shared between these diseases. Results of protein prioritization by machine learning have identified known as well as new genetic markers which might have direct or indirect involvement in several neurological disorders. Important gene hubs have also been identified that provide an evidence for shared molecular pathways in the neurological disease network.

GreedyPlus: An Algorithm for the Alignment of Interface Interaction Networks

The increasing ease and accuracy of protein-protein interaction detection has resulted in the ability to map the interactomes of multiple species. We now have an opportunity to compare species to better understand how interactomes evolve. As DNA and protein sequence alignment algorithms were required for comparative genomics, network alignment algorithms are required for comparative interactomics. A number of network alignment methods have been developed for protein-protein interaction networks, where proteins are represented as vertices linked by edges if they interact. Recently, protein interactions have been mapped at the level of amino acid positions, which can be represented as an interface-interaction network (IIN), where vertices represent binding sites, such as protein domains and short sequence motifs. However, current algorithms are not designed to align these networks and generally fail to do so in practice. We present a greedy algorithm, GreedyPlus, for IIN alignment, combining data from diverse sources, including network, protein and binding site properties, to identify putative orthologous relationships between interfaces in available worm and yeast data. GreedyPlus is fast and simple, allowing for easy customization of behaviour, yet still capable of generating biologically meaningful network alignments.

Competition-cooperation relationship networks characterize the competition and cooperation between proteins

By analyzing protein-protein interaction (PPI) networks, one can find that a protein may have multiple binding partners. However, it is difficult to determine whether the interactions with these partners occur simultaneously from binary PPIs alone. Here, we construct the yeast and human competition-cooperation relationship networks (CCRNs) based on protein structural interactomes to clearly exhibit the relationship (competition or cooperation) between two partners of the same protein. If two partners compete for the same interaction interface, they would be connected by a competitive edge; otherwise, they would be connected by a cooperative edge. The properties of three kinds of hubs (i.e., competitive, modest, and cooperative hubs) are analyzed in the CCRNs. Our results show that competitive hubs have higher clustering coefficients and form clusters in the human CCRN, but these tendencies are not observed in the yeast CCRN. We find that the human-specific proteins contribute significantly to these differences. Subsequently, we conduct a series of computational experiments to investigate the regulatory mechanisms that avoid competition between proteins. Our comprehensive analyses reveal that for most yeast and human protein competitors, transcriptional regulation plays an important role. Moreover, the human-specific proteins have a particular preference for other regulatory mechanisms, such as alternative splicing.

NPPD: A Protein-Protein Docking Scoring Function Based on Dyadic Differences in Networks of Hydrophobic and Hydrophilic Amino Acid Residues

Protein-protein docking (PPD) predictions usually rely on the use of a scoring function to rank docking models generated by exhaustive sampling. To rank good models higher than bad ones, a large number of scoring functions have been developed and evaluated, but the methods used for the computation of PPD predictions remain largely unsatisfactory. Here, we report a network-based PPD scoring function, the NPPD, in which the network consists of two types of network nodes, one for hydrophobic and the other for hydrophilic amino acid residues, and the nodes are connected when the residues they represent are within a certain contact distance. We showed that network parameters that compute dyadic interactions and those that compute heterophilic interactions of the amino acid networks thus constructed allowed NPPD to perform well in a benchmark evaluation of 115 PPD scoring functions, most of which, unlike NPPD, are based on some sort of protein-protein interaction energy. We also showed that NPPD was highly complementary to these energy-based scoring functions, suggesting that the combined use of conventional scoring functions and NPPD might significantly improve the accuracy of current PPD predictions.

Burial Level Change Defines a High Energetic Relevance for Protein Binding Interfaces

Protein-protein interfaces defined through atomic contact or solvent accessibility change are widely adopted in structural biology studies. But, these definitions cannot precisely capture energetically important regions at protein interfaces. The burial depth of an atom in a protein is related to the atom's energy. This work investigates how closely the change in burial level of an atom/residue upon complexation is related to the binding. Burial level change is different from burial level itself. An atom deeply buried in a monomer with a high burial level may not change its burial level after an interaction and it may have little burial level change. We hypothesize that an interface is a region of residues all undergoing burial level changes after interaction. By this definition, an interface can be decomposed into an onion-like structure according to the burial level change extent. We found that our defined interfaces cover energetically important residues more precisely, and that the binding free energy of an interface is distributed progressively from the outermost layer to the core. These observations are used to predict binding hot spots. Our approach's F-measure performance on a benchmark dataset of alanine mutagenesis residues is much superior or similar to those by complicated energy modeling or machine learning approaches.

Calcineurin regulates the yeast synaptojanin Inp53/Sjl3 during membrane stress

During hyperosmotic shock, Saccharomyces cerevisiae adjusts to physiological challenges, including large plasma membrane invaginations generated by rapid cell shrinkage. Calcineurin, the Ca(2+)/calmodulin-dependent phosphatase, is normally cytosolic but concentrates in puncta and at sites of polarized growth during intense osmotic stress; inhibition of calcineurin-activated gene expression suggests that restricting its access to substrates tunes calcineurin signaling specificity. Hyperosmotic shock promotes calcineurin binding to and dephosphorylation of the PI(4,5)P2 phosphatase synaptojanin/Inp53/Sjl3 and causes dramatic calcineurin-dependent reorganization of PI(4,5)P2-enriched membrane domains. Inp53 normally promotes sorting at the trans-Golgi network but localizes to cortical actin patches in osmotically stressed cells. By activating Inp53, calcineurin repolarizes the actin cytoskeleton and maintains normal plasma membrane morphology in synaptojanin-limited cells. In response to hyperosmotic shock and calcineurin-dependent regulation, Inp53 shifts from associating predominantly with clathrin to interacting with endocytic proteins Sla1, Bzz1, and Bsp1, suggesting that Inp53 mediates stress-specific endocytic events. This response has physiological and molecular similarities to calcineurin-regulated activity-dependent bulk endocytosis in neurons, which retrieves a bolus of plasma membrane deposited by synaptic vesicle fusion. We propose that activation of Ca(2+)/calcineurin and PI(4,5)P2 signaling to regulate endocytosis is a fundamental and conserved response to excess membrane in eukaryotic cells.

Structure-based inhibition of protein-protein interactions

Protein-protein interactions (PPIs) are emerging as attractive targets for drug design because of their central role in directing normal and aberrant cellular functions. These interactions were once considered "undruggable" because their large and dynamic interfaces make small molecule inhibitor design challenging. However, landmark advances in computational analysis, fragment screening and molecular design have enabled development of a host of promising strategies to address the fundamental molecular recognition challenge. An attractive approach for targeting PPIs involves mimicry of protein domains that are critical for complex formation. This approach recognizes that protein subdomains or protein secondary structures are often present at interfaces and serve as organized scaffolds for the presentation of side chain groups that engage the partner protein(s). Design of protein domain mimetics is in principle rather straightforward but is enabled by a host of computational strategies that provide predictions of important residues that should be mimicked. Herein we describe a workflow proceeding from interaction network analysis, to modeling a complex structure, to identifying a high-affinity sub-structure, to developing interaction inhibitors. We apply the design procedure to peptidomimetic inhibitors of Ras-mediated signaling.

From proteomes to complexomes in the era of systems biology

Protein complexes carry out almost the entire signaling and functional processes in the cell. The protein complex complement of a cell, and its network of complex-complex interactions, is referred to here as the complexome. Computational methods to predict protein complexes from proteomics data, resulting in network representations of complexomes, have recently being developed. In addition, key advances have been made toward understanding the network and structural organization of complexomes. We review these bioinformatics advances, and their discovery-potential, as well as the merits of integrating proteomics data with emerging methods in systems biology to study protein complex signaling. It is envisioned that improved integration of proteomics and systems biology, incorporating the dynamics of protein complexes in space and time, may lead to more predictive models of cell signaling networks for effective modulation.

Rule-based modeling: a computational approach for studying biomolecular site dynamics in cell signaling systems

Rule-based modeling was developed to address the limitations of traditional approaches for modeling chemical kinetics in cell signaling systems. These systems consist of multiple interacting biomolecules (e.g., proteins), which themselves consist of multiple parts (e.g., domains, linear motifs, and sites of phosphorylation). Consequently, biomolecules that mediate information processing generally have the potential to interact in multiple ways, with the number of possible complexes and posttranslational modification states tending to grow exponentially with the number of binary interactions considered. As a result, only large reaction networks capture all possible consequences of the molecular interactions that occur in a cell signaling system, which is problematic because traditional modeling approaches for chemical kinetics (e.g., ordinary differential equations) require explicit network specification. This problem is circumvented through representation of interactions in terms of local rules. With this approach, network specification is implicit and model specification is concise. Concise representation results in a coarse graining of chemical kinetics, which is introduced because all reactions implied by a rule inherit the rate law associated with that rule. Coarse graining can be appropriate if interactions are modular, and the coarseness of a model can be adjusted as needed. Rules can be specified using specialized model-specification languages, and recently developed tools designed for specification of rule-based models allow one to leverage powerful software engineering capabilities. A rule-based model comprises a set of rules, which can be processed by general-purpose simulation and analysis tools to achieve different objectives (e.g., to perform either a deterministic or stochastic simulation).

MSEA: detection and quantification of mutation hotspots through mutation set enrichment analysis

Many cancer genes form mutation hotspots that disrupt their functional domains or active sites, leading to gain- or loss-of-function. We propose a mutation set enrichment analysis (MSEA) implemented by two novel methods,MSEA-clust and MSEA-domain, to predict cancer genes based on mutation hotspot patterns. MSEA methods are evaluated by both simulated and real cancer data. We find approximately 51% of the eligible known cancer genes form detectable mutation hotspots. Application of MSEA in eight cancers reveals a total of 82 genes with mutation hotspots,including well-studied cancer genes, known cancer genes re-found in new cancer types, and novel cancer genes.

Evolutionary pressure on the topology of protein interface interaction networks

The densely connected structure of protein-protein interaction (PPI) networks reflects the functional need of proteins to cooperate in cellular processes. However, PPI networks do not adequately capture the competition in protein binding. By contrast, the interface interaction network (IIN) studied here resolves the modular character of protein-protein binding and distinguishes between simultaneous and exclusive interactions that underlie both cooperation and competition. We show that the topology of the IIN is under evolutionary pressure, and we connect topological features of the IIN to specific biological functions. To reveal the forces shaping the network topology, we use a sequence-based computational model of interface binding along with network analysis. We find that the more fragmented structure of IINs, in contrast to the dense PPI networks, arises in large part from the competition between specific and nonspecific binding. The need to minimize nonspecific binding favors specific network motifs, including a minimal number of cliques (i.e., fully connected subgraphs) and many disconnected fragments. Validating the model, we find that these network characteristics are closely mirrored in the IIN of clathrin-mediated endocytosis. Features unexpected on the basis of our motif analysis are found to indicate either exceptional binding selectivity or important regulatory functions.