Protein Interaction Networks—More Than Mere Modules

It is widely believed that the modular organization of cellular function is reflected in a modular structure of molecular networks. A common view is that a “module” in a network is a cohesively linked group of nodes, densely connected internally and sparsely interacting with the rest of the network. Many algorithms try to identify functional modules in protein-interaction networks (PIN) by searching for such cohesive groups of proteins. Here, we present an alternative approach independent of any prior definition of what actually constitutes a “module”. In a self-consistent manner, proteins are grouped into “functional roles” if they interact in similar ways with other proteins according to their functional roles. Such grouping may well result in cohesive modules again, but only if the network structure actually supports this. We applied our method to the PIN from the Human Protein Reference Database (HPRD) and found that a representation of the network in terms of cohesive modules, at least on a global scale, does not optimally represent the network''s structure because it focuses on finding independent groups of proteins. In contrast, a decomposition into functional roles is able to depict the structure much better as it also takes into account the interdependencies between roles and even allows groupings based on the absence of interactions between proteins in the same functional role. This, for example, is the case for transmembrane proteins, which could never be recognized as a cohesive group of nodes in a PIN. When mapping experimental methods onto the groups, we identified profound differences in the coverage suggesting that our method is able to capture experimental bias in the data, too. For example yeast-two-hybrid data were highly overrepresented in one particular group. Thus, there is more structure in protein-interaction networks than cohesive modules alone and we believe this finding can significantly improve automated function prediction algorithms.     

Role of Ryanodine Receptors in the Assembly of Calcium Release Units in Skeletal Muscle

Abstract. In muscle cells, excitation–contraction (e–c) coupling is mediated by “calcium release units,” junctions between the sarcoplasmic reticulum (SR) and exterior membranes. Two proteins, which face each other, are known to functionally interact in those structures: the ryanodine receptors (RyRs), or SR calcium release channels, and the dihydropyridine receptors (DHPRs), or L-type calcium channels of exterior membranes. In skeletal muscle, DHPRs form tetrads, groups of four receptors, and tetrads are organized in arrays that face arrays of feet (or RyRs). Triadin is a protein of the SR located at the SR–exterior membrane junctions, whose role is not known. We have structurally characterized calcium release units in a skeletal muscle cell line (1B5) lacking Ry₁R. Using immunohistochemistry and freeze-fracture electron microscopy, we find that DHPR and triadin are clustered in foci in differentiating 1B5 cells. Thin section electron microscopy reveals numerous SR–exterior membrane junctions lacking foot structures (dyspedic). These results suggest that components other than Ry₁Rs are responsible for targeting DHPRs and triadin to junctional regions. However, DHPRs in 1B5 cells are not grouped into tetrads as in normal skeletal muscle cells suggesting that anchoring to Ry₁Rs is necessary for positioning DHPRs into ordered arrays of tetrads. This hypothesis is confirmed by finding a “restoration of tetrads” in junctional domains of surface membranes after transfection of 1B5 cells with cDNA encoding for Ry₁R.     

Identification of adherens junction-associated GTPase activating proteins by the fluorescence localization-based expression cloning

The junctional complex, including tight junctions (TJs), adherens junctions (AJs), and desmosomes, plays crucial roles in the structure and functions of epithelial cellular sheets. In this study, we evaluated the fluorescence localization-based retrovirus-mediated expression cloning (FL-REX) method as an approach to identify novel molecular components of TJs and AJs. Using an expression library of cDNA-GFP-fusions derived from mRNA of a mouse epithelial cell line, we confirmed that cDNAs for various known TJ- and AJ-components could be cloned in the FL-REX. Furthermore, cDNAs for ARHGAP12 and SPAL3, two putative GTPase activating proteins (GAPs) for small G proteins, were cloned as novel components of the junctional complex. Immunofluorescence staining using antibodies generated in-house demonstrated that these GAPs were localized at epithelial cell-cell junctions in various mouse tissues, and were specific to AJs when observed under confocal laser-scanning microscopy. These data suggest that FL-REX is a powerful tool to identify novel proteins localized at TJs and AJs.     

Stochastic Blockmodeling of the Modules and Core of the Caenorhabditis elegans Connectome

Recently, there has been much interest in the community structure or mesoscale organization of complex networks. This structure is characterised either as a set of sparsely inter-connected modules or as a highly connected core with a sparsely connected periphery. However, it is often difficult to disambiguate these two types of mesoscale structure or, indeed, to summarise the full network in terms of the relationships between its mesoscale constituents. Here, we estimate a community structure with a stochastic blockmodel approach, the Erdős-Rényi Mixture Model, and compare it to the much more widely used deterministic methods, such as the Louvain and Spectral algorithms. We used the Caenorhabditis elegans (C. elegans) nervous system (connectome) as a model system in which biological knowledge about each node or neuron can be used to validate the functional relevance of the communities obtained. The deterministic algorithms derived communities with 4–5 modules, defined by sparse inter-connectivity between all modules. In contrast, the stochastic Erdős-Rényi Mixture Model estimated a community with 9 blocks or groups which comprised a similar set of modules but also included a clearly defined core, made of 2 small groups. We show that the “core-in-modules” decomposition of the worm brain network, estimated by the Erdős-Rényi Mixture Model, is more compatible with prior biological knowledge about the C. elegans nervous system than the purely modular decomposition defined deterministically. We also show that the blockmodel can be used both to generate stochastic realisations (simulations) of the biological connectome, and to compress network into a small number of super-nodes and their connectivity. We expect that the Erdős-Rényi Mixture Model may be useful for investigating the complex community structures in other (nervous) systems.     

Junctional Adhesion Molecule,a Novel Member of the Immunoglobulin Superfamily That Distributes at Intercellular Junctions and Modulates Monocyte Transmigration

Tight junctions are the most apical components of endothelial and epithelial intercellular cleft. In the endothelium these structures play an important role in the control of paracellular permeability to circulating cells and solutes. The only known integral membrane protein localized at sites of membrane–membrane interaction of tight junctions is occludin, which is linked inside the cells to a complex network of cytoskeletal and signaling proteins. We report here the identification of a novel protein (junctional adhesion molecule [JAM]) that is selectively concentrated at intercellular junctions of endothelial and epithelial cells of different origins. Confocal and immunoelectron microscopy shows that JAM codistributes with tight junction components at the apical region of the intercellular cleft. A cDNA clone encoding JAM defines a novel immunoglobulin gene superfamily member that consists of two V-type Ig domains. An mAb directed to JAM (BV11) was found to inhibit spontaneous and chemokine-induced monocyte transmigration through an endothelial cell monolayer in vitro. Systemic treatment of mice with BV11 mAb blocked monocyte infiltration upon chemokine administration in subcutaneous air pouches. Thus, JAM is a new component of endothelial and epithelial junctions that play a role in regulating monocyte transmigration.     

Network-Based Elucidation of Human Disease Similarities Reveals Common Functional Modules Enriched for Pluripotent Drug Targets

Current work in elucidating relationships between diseases has largely been based on pre-existing knowledge of disease genes. Consequently, these studies are limited in their discovery of new and unknown disease relationships. We present the first quantitative framework to compare and contrast diseases by an integrated analysis of disease-related mRNA expression data and the human protein interaction network. We identified 4,620 functional modules in the human protein network and provided a quantitative metric to record their responses in 54 diseases leading to 138 significant similarities between diseases. Fourteen of the significant disease correlations also shared common drugs, supporting the hypothesis that similar diseases can be treated by the same drugs, allowing us to make predictions for new uses of existing drugs. Finally, we also identified 59 modules that were dysregulated in at least half of the diseases, representing a common disease-state “signature”. These modules were significantly enriched for genes that are known to be drug targets. Interestingly, drugs known to target these genes/proteins are already known to treat significantly more diseases than drugs targeting other genes/proteins, highlighting the importance of these core modules as prime therapeutic opportunities.     

Spatiotemporal organization and mechanosensory function of podosomes

Podosomes are small, circular adhesions formed by cells such as osteoclasts, macrophages, dendritic cells, and endothelial cells. They comprise a protrusive actin core module and an adhesive ring module composed of integrins and cytoskeletal adaptor proteins such as vinculin and talin. Furthermore, podosomes are associated with an actin network and often organize into large clusters. Recent results from our laboratory and others have shed new light on podosome structure and dynamics, suggesting a revision of the classical “core-ring” model. Also, these studies demonstrate that the adhesive and protrusive module are functionally linked by the actin network likely facilitating mechanotransduction as well as providing feedback between these two modules. In this commentary, we briefly summarize these recent advances with respect to the knowledge on podosome structure and discuss force distribution mechanisms within podosomes and their emerging role in mechanotransduction.     

Analysis and Prediction of the Metabolic Stability of Proteins Based on Their Sequential Features,Subcellular Locations and Interaction Networks

The metabolic stability is a very important idiosyncracy of proteins that is related to their global flexibility, intramolecular fluctuations, various internal dynamic processes, as well as many marvelous biological functions. Determination of protein''s metabolic stability would provide us with useful information for in-depth understanding of the dynamic action mechanisms of proteins. Although several experimental methods have been developed to measure protein''s metabolic stability, they are time-consuming and more expensive. Reported in this paper is a computational method, which is featured by (1) integrating various properties of proteins, such as biochemical and physicochemical properties, subcellular locations, network properties and protein complex property, (2) using the mRMR (Maximum Relevance & Minimum Redundancy) principle and the IFS (Incremental Feature Selection) procedure to optimize the prediction engine, and (3) being able to identify proteins among the four types: “short”, “medium”, “long”, and “extra-long” half-life spans. It was revealed through our analysis that the following seven characters played major roles in determining the stability of proteins: (1) KEGG enrichment scores of the protein and its neighbors in network, (2) subcellular locations, (3) polarity, (4) amino acids composition, (5) hydrophobicity, (6) secondary structure propensity, and (7) the number of protein complexes the protein involved. It was observed that there was an intriguing correlation between the predicted metabolic stability of some proteins and the real half-life of the drugs designed to target them. These findings might provide useful insights for designing protein-stability-relevant drugs. The computational method can also be used as a large-scale tool for annotating the metabolic stability for the avalanche of protein sequences generated in the post-genomic age.     

Whole-proteome genetic analysis of dependencies in assembly of a vertebrate kinetochore

Kinetochores orchestrate mitotic chromosome segregation. Here, we use quantitative mass spectrometry of mitotic chromosomes isolated from a comprehensive set of chicken DT40 mutants to examine the dependencies of 93 confirmed and putative kinetochore proteins for stable association with chromosomes. Clustering and network analysis reveal both known and unexpected aspects of coordinated behavior for members of kinetochore protein complexes. Surprisingly, CENP-T depends on CENP-N for chromosome localization. The Ndc80 complex exhibits robust correlations with all other complexes in a “core” kinetochore network. Ndc80 associated with CENP-T interacts with a cohort of Rod, zw10, and zwilch (RZZ)–interacting proteins that includes Spindly, Mad1, and CENP-E. This complex may coordinate microtubule binding with checkpoint signaling. Ndc80 associated with CENP-C forms the KMN (Knl1, Mis12, Ndc80) network and may be the microtubule-binding “workhorse” of the kinetochore. Our data also suggest that CENP-O and CENP-R may regulate the size of the inner kinetochore without influencing the assembly of the outer kinetochore.     

NIRF constitutes a nodal point in the cell cycle network and is a candidate tumor suppressor

In biological networks, a small number of “hub” proteins play critical roles in the network integrity and functions. The cell cycle network orchestrates versatile cellular functions through interactions between many signaling modules, whose defects impair diverse cellular processes, often leading to cancer. However, the network architecture and molecular basis that ensure proper coordination between distinct modules are unclear. Here, we show that the ubiquitin ligase NIRF (also known as UHRF2), which induces G₁ arrest, interacts with multiple cell cycle proteins including cyclins (A2, B1, D1 and E1), p53 and pRB, and ubiquitinates cyclins D1 and E1. Consistent with its versatility, a bioinformatic network analysis demonstrated that NIRF is an intermodular hub protein that is responsible for the coordination of multiple network modules. Notably, intermodular hubs are frequently associated with oncogenesis. Indeed, we detected loss of heterozygosity of the NIRF gene in several kinds of tumors. When a cancer outlier profile analysis was applied to the Oncomine database, loss of the NIRF gene was found at statistically significant levels in diverse tumors. Importantly, a recurrent microdeletion targeting NIRF was observed in non-small cell lung carcinoma. Furthermore, NIRF is immediately adjacent to the single nucleotide polymorphism rs719725, which is reportedly associated with the risk of colorectal cancer. These observations suggest that NIRF occupies a prominent position within the cell cycle network, and is a strong candidate for a tumor suppressor whose aberration contributes to the pathogenesis of diverse malignancies.Key words: NIRF, UHRF2, cell cycle network, systems biology, ubiquitin ligase, COPA, tumor suppressor, glioblastoma, non-small cell lung carcinoma, rs719725     

Drug Repositioning through Systematic Mining of Gene Coexpression Networks in Cancer

Gene coexpression network analysis is a powerful “data-driven” approach essential for understanding cancer biology and mechanisms of tumor development. Yet, despite the completion of thousands of studies on cancer gene expression, there have been few attempts to normalize and integrate co-expression data from scattered sources in a concise “meta-analysis” framework. We generated such a resource by exploring gene coexpression networks in 82 microarray datasets from 9 major human cancer types. The analysis was conducted using an elaborate weighted gene coexpression network (WGCNA) methodology and identified over 3,000 robust gene coexpression modules. The modules covered a range of known tumor features, such as proliferation, extracellular matrix remodeling, hypoxia, inflammation, angiogenesis, tumor differentiation programs, specific signaling pathways, genomic alterations, and biomarkers of individual tumor subtypes. To prioritize genes with respect to those tumor features, we ranked genes within each module by connectivity, leading to identification of module-specific functionally prominent hub genes. To showcase the utility of this network information, we positioned known cancer drug targets within the coexpression networks and predicted that Anakinra, an anti-rheumatoid therapeutic agent, may be promising for development in colorectal cancer. We offer a comprehensive, normalized and well documented collection of >3000 gene coexpression modules in a variety of cancers as a rich data resource to facilitate further progress in cancer research.     

Cingulin and paracingulin show similar dynamic behaviour,but are recruited independently to junctions

Abstract

Cingulin (CGN) and paracingulin (CGNL1) are structurally related proteins that regulate Rho family GTPases by recruiting guanine nucleotide exchange factors to epithelial junctions. Although the subcellular localization of cingulin and paracingulin is likely to be essential for their role as adaptor proteins, nothing is known on their in vivo localization, and their dynamics of exchange with the junctional membrane. To address these questions, we generated stable clones of MDCK cells expressing fluorescently tagged cingulin and paracingulin. By FRAP analysis, cingulin and paracingulin show a very similar dynamic behaviour, with recovery curves and mobile fractions that are distinct from ZO-1, and indicate a rapid exchange with a cytosolic pool. Interestingly, only paracingulin, but not cingulin, is peripherally localized in isolated cells, requires the integrity of the microtubule cytoskeleton to be stably anchored to junctions, and associates with E-cadherin. In contrast, both proteins require the integrity of the actin cytoskeleton to maintain their junctional localization. Although cingulin and paracingulin form a complex and can interact in vitro, the junctional recruitment and the dynamics of membrane exchange of paracingulin is independent of cingulin, and vice-versa. In summary, cingulin and paracingulin show a similar dynamic behaviour, but partially distinct localizations and functional interactions with the cytoskeleton, and are recruited independently to junctions.     

Claudin-1 and -2: Novel Integral Membrane Proteins Localizing at Tight Junctions with No Sequence Similarity to Occludin

Occludin is the only known integral membrane protein localizing at tight junctions (TJ), but recent targeted disruption analysis of the occludin gene indicated the existence of as yet unidentified integral membrane proteins in TJ. We therefore re-examined the isolated junction fraction from chicken liver, from which occludin was first identified. Among numerous components of this fraction, only a broad silver-stained band ~22 kD was detected with the occludin band through 4 M guanidine-HCl extraction as well as sonication followed by stepwise sucrose density gradient centrifugation. Two distinct peptide sequences were obtained from the lower and upper halves of the broad band, and similarity searches of databases allowed us to isolate two full-length cDNAs encoding related mouse 22-kD proteins consisting of 211 and 230 amino acids, respectively. Hydrophilicity analysis suggested that both bore four transmembrane domains, although they did not show any sequence similarity to occludin. Immunofluorescence and immunoelectron microscopy revealed that both proteins tagged with FLAG or GFP were targeted to and incorporated into the TJ strand itself. We designated them as “claudin-1” and “claudin-2”, respectively. Although the precise structure/function relationship of the claudins to TJ still remains elusive, these findings indicated that multiple integral membrane proteins with four putative transmembrane domains, occludin and claudins, constitute TJ strands.     

The Paracellular Pathway and Bile Formation

Choleretic infusions of taurocholate (40 μ moles for one hour) result in a significant increase in the number of lateral cell surface invaginations observed by scanning electron microscopy adjacent to the junctional complex of bile canaliculi in rat liver. Transmission electron microscopy indicates that these invaginations resemble “blisters” induced by osmotic gradients across epithelial tissues, a morphologic change which correlates with increases in ionic and hydraulic conductivity of the paracellular “shunt” pathway in such tissue. Since taurocholate infusions result in localization of ionic lanthanum chloride within hepatocyte junctional complexes, bile acids may also stimulate the movement of fluid and electrolytes across paracellular pathways during the process of bile formation.     

Fine-Scale Dissection of Functional Protein Network Organization by Statistical Network Analysis

Revealing organizational principles of biological networks is an important goal of systems biology. In this study, we sought to analyze the dynamic organizational principles within the protein interaction network by studying the characteristics of individual neighborhoods of proteins within the network based on their gene expression as well as protein-protein interaction patterns. By clustering proteins into distinct groups based on their neighborhood gene expression characteristics, we identify several significant trends in the dynamic organization of the protein interaction network. We show that proteins with distinct neighborhood gene expression characteristics are positioned in specific localities in the protein interaction network thereby playing specific roles in the dynamic network connectivity. Remarkably, our analysis reveals a neighborhood characteristic that corresponds to the most centrally located group of proteins within the network. Further, we show that the connectivity pattern displayed by this group is consistent with the notion of “rich club connectivity” in complex networks. Importantly, our findings are largely reproducible in networks constructed using independent and different datasets.     

Automated Network Analysis Identifies Core Pathways in Glioblastoma

Background

Glioblastoma multiforme (GBM) is the most common and aggressive type of brain tumor in humans and the first cancer with comprehensive genomic profiles mapped by The Cancer Genome Atlas (TCGA) project. A central challenge in large-scale genome projects, such as the TCGA GBM project, is the ability to distinguish cancer-causing “driver” mutations from passively selected “passenger” mutations.

Principal Findings

In contrast to a purely frequency based approach to identifying driver mutations in cancer, we propose an automated network-based approach for identifying candidate oncogenic processes and driver genes. The approach is based on the hypothesis that cellular networks contain functional modules, and that tumors target specific modules critical to their growth. Key elements in the approach include combined analysis of sequence mutations and DNA copy number alterations; use of a unified molecular interaction network consisting of both protein-protein interactions and signaling pathways; and identification and statistical assessment of network modules, i.e. cohesive groups of genes of interest with a higher density of interactions within groups than between groups.

Conclusions

We confirm and extend the observation that GBM alterations tend to occur within specific functional modules, in spite of considerable patient-to-patient variation, and that two of the largest modules involve signaling via p53, Rb, PI3K and receptor protein kinases. We also identify new candidate drivers in GBM, including AGAP2/CENTG1, a putative oncogene and an activator of the PI3K pathway; and, three additional significantly altered modules, including one involved in microtubule organization. To facilitate the application of our network-based approach to additional cancer types, we make the method freely available as part of a software tool called NetBox.     

Cingulin and paracingulin show similar dynamic behaviour, but are recruited independently to junctions

Cingulin (CGN) and paracingulin (CGNL1) are structurally related proteins that regulate Rho family GTPases by recruiting guanine nucleotide exchange factors to epithelial junctions. Although the subcellular localization of cingulin and paracingulin is likely to be essential for their role as adaptor proteins, nothing is known on their in vivo localization, and their dynamics of exchange with the junctional membrane. To address these questions, we generated stable clones of MDCK cells expressing fluorescently tagged cingulin and paracingulin. By FRAP analysis, cingulin and paracingulin show a very similar dynamic behaviour, with recovery curves and mobile fractions that are distinct from ZO-1, and indicate a rapid exchange with a cytosolic pool. Interestingly, only paracingulin, but not cingulin, is peripherally localized in isolated cells, requires the integrity of the microtubule cytoskeleton to be stably anchored to junctions, and associates with E-cadherin. In contrast, both proteins require the integrity of the actin cytoskeleton to maintain their junctional localization. Although cingulin and paracingulin form a complex and can interact in vitro, the junctional recruitment and the dynamics of membrane exchange of paracingulin is independent of cingulin, and vice-versa. In summary, cingulin and paracingulin show a similar dynamic behaviour, but partially distinct localizations and functional interactions with the cytoskeleton, and are recruited independently to junctions.     

The fine structure of identified electrotonic synapses following increased coupling resistance

Summary Gap junctions exist in the septa between the segments of the lateral giant axons in the ventral nerve cord of the crayfish Procambarus. A large increase in the resistance (uncoupling) of these gap junctions was brought about by mechanical injury to the axonal segments. Both thin sections and freeze-fracture preparations were used to monitor the morphological changes which occurred up to 45 min after injury.There was no apparent change in the organization (a loose polygonal array) of the intramembrane particles which make up the junctional complex up to 45 min after injury. In some instances, however, the intramembrane particles appeared to have moved away from the junctional area. Other junctional regions were internalized and appeared similar to what have been called annular gap junctions. Also at this time (20–25 min after injury), a dense cytoplasmic plug formed in uninjured axon near the junctional region. It is concluded that the gap junctions that exhibit a loose polygonal organization of the intramembrane particles may be either in a state of low resistance (coupled) or a state of high resistance (uncoupled).     

Global Analysis of the Human Pathophenotypic Similarity Gene Network Merges Disease Module Components

The molecular complexity of genetic diseases requires novel approaches to break it down into coherent biological modules. For this purpose, many disease network models have been created and analyzed. We highlight two of them, “the human diseases networks” (HDN) and “the orphan disease networks” (ODN). However, in these models, each single node represents one disease or an ambiguous group of diseases. In these cases, the notion of diseases as unique entities reduces the usefulness of network-based methods. We hypothesize that using the clinical features (pathophenotypes) to define pathophenotypic connections between disease-causing genes improve our understanding of the molecular events originated by genetic disturbances. For this, we have built a pathophenotypic similarity gene network (PSGN) and compared it with the unipartite projections (based on gene-to-gene edges) similar to those used in previous network models (HDN and ODN). Unlike these disease network models, the PSGN uses semantic similarities. This pathophenotypic similarity has been calculated by comparing pathophenotypic annotations of genes (human abnormalities of HPO terms) in the “Human Phenotype Ontology”. The resulting network contains 1075 genes (nodes) and 26197 significant pathophenotypic similarities (edges). A global analysis of this network reveals: unnoticed pairs of genes showing significant pathophenotypic similarity, a biological meaningful re-arrangement of the pathological relationships between genes, correlations of biochemical interactions with higher similarity scores and functional biases in metabolic and essential genes toward the pathophenotypic specificity and the pleiotropy, respectively. Additionally, pathophenotypic similarities and metabolic interactions of genes associated with maple syrup urine disease (MSUD) have been used to merge into a coherent pathological module.Our results indicate that pathophenotypes contribute to identify underlying co-dependencies among disease-causing genes that are useful to describe disease modularity.