Arunachalam Vinayagam

A Directed Protein Interaction Network for Investigating Intracellular Signal Transduction

Effective prediction of the direction of signal flow in an interaction network enables modeling of signaling dynamics and identification of regulatory proteins. Finding More Pieces to the Signaling Puzzle Even well-studied pathways are likely to be incomplete in terms of our knowledge of all the components and their relationships, and the larger interconnected network that represents the true cellular regulatory landscape remains woefully unknown. Vinayagam et al. used an automated yeast two-hybrid interaction mating assay to identify protein-protein interactions (PPIs) among human proteins and then integrated that PPI data set with previously published data to create an undirected human PPI network connecting 9832 proteins through 39,641 interactions. The authors then applied a Bayesian learning strategy to assign direction to the interactions among the proteins. The resulting directed network enabled them to evaluate growth factor–induced protein phosphorylation dynamics and to identify previously unknown modulators of the extracellular signal–regulated protein kinase pathway, of which 18 were validated with cell-based assays. This strategy should prove useful in completing the puzzle of the cellular regulatory network. Cellular signal transduction is a complex process involving protein-protein interactions (PPIs) that transmit information. For example, signals from the plasma membrane may be transduced to transcription factors to regulate gene expression. To obtain a global view of cellular signaling and to predict potential signal modulators, we searched for protein interaction partners of more than 450 signaling-related proteins by means of automated yeast two-hybrid interaction mating. The resulting PPI network connected 1126 proteins through 2626 PPIs. After expansion of this interaction map with publicly available PPI data, we generated a directed network resembling the signal transduction flow between proteins with a naïve Bayesian classifier. We exploited information on the shortest PPI paths from membrane receptors to transcription factors to predict input and output relationships between interacting proteins. Integration of directed PPI with time-resolved protein phosphorylation data revealed network structures that dynamically conveyed information from the activated epidermal growth factor and extracellular signal–regulated kinase (EGF/ERK) signaling cascade to directly associated proteins and more distant proteins in the network. From the model network, we predicted 18 previously unknown modulators of EGF/ERK signaling, which we validated in mammalian cell-based assays. This generic experimental and computational approach provides a framework for elucidating causal connections between signaling proteins and facilitates the identification of proteins that modulate the flow of information in signaling networks.

HIPPIE: Integrating protein interaction networks with experiment based quality scores.

Protein function is often modulated by protein-protein interactions (PPIs) and therefore defining the partners of a protein helps to understand its activity. PPIs can be detected through different experimental approaches and are collected in several expert curated databases. These databases are used by researchers interested in examining detailed information on particular proteins. In many analyses the reliability of the characterization of the interactions becomes important and it might be necessary to select sets of PPIs of different confidence levels. To this goal, we generated HIPPIE (Human Integrated Protein-Protein Interaction rEference), a human PPI dataset with a normalized scoring scheme that integrates multiple experimental PPI datasets. HIPPIEs scoring scheme has been optimized by human experts and a computer algorithm to reflect the amount and quality of evidence for a given PPI and we show that these scores correlate to the quality of the experimental characterization. The HIPPIE web tool (available at http://cbdm.mdc-berlin.de/tools/hippie) allows researchers to do network analyses focused on likely true PPI sets by generating subnetworks around proteins of interest at a specified confidence level.

The Hippo Signaling Pathway Interactome

Dissecting Hippo Interactions The Hippo signaling pathway plays key roles in many processes, from cell proliferation and cell death to regulation of stem cells and cancer cells. Kwon et al. (p. 737, published 10 October) attempted to systematically identify all components of the pathway. A protein-protein interaction screen identified more than 200 interactions among approximately 150 proteins. A protein identified in the screen, Leash, restrained the activity of the transcriptional coactivator Yorkie, which regulates gene expression in response to Hippo signaling. A proteomics approach for protein-protein interactions reveals new components of a conserved cell signaling pathway. The Hippo pathway controls metazoan organ growth by regulating cell proliferation and apoptosis. Many components have been identified, but our knowledge of the composition and structure of this pathway is still incomplete. Using existing pathway components as baits, we generated by mass spectrometry a high-confidence Drosophila Hippo protein-protein interaction network (Hippo-PPIN) consisting of 153 proteins and 204 interactions. Depletion of 67% of the proteins by RNA interference regulated the transcriptional coactivator Yorkie (Yki) either positively or negatively. We selected for further characterization a new member of the alpha-arrestin family, Leash, and show that it promotes degradation of Yki through the lysosomal pathway. Given the importance of the Hippo pathway in tumor development, the Hippo-PPIN will contribute to our understanding of this network in both normal growth and cancer.

Proteomic and functional genomic landscape of receptor tyrosine kinase and ras to extracellular signal-regulated kinase signaling.

Interactome mapping and functional genomics in Drosophila reveal common and specific components of a mitogen-activated protein kinase pathway. Finding the Shared and the Specific Components Regulating MAPK Signals Even in extensively studied pathways, such as the extracellular signal–regulated kinase (ERK) pathway that is activated by receptor tyrosine kinases, there remain gaps in our knowledge. Friedman et al. combined protein-protein interaction screens with RNAi functional genomic screens in Drosophila cell lines to identify components of the ERK pathway downstream of two receptor tyrosine kinases. Their analysis suggested that these receptors may compete for some common components, in addition to using receptor-specific and cell-specific signal transduction pathways. Knockdown of several newly identified pathway regulators resulted in wing phenotypes in vivo, confirming these as components in the pathway. Detailed understanding of this pathway has clinical relevance because of its importance in both physiological and pathophysiological contexts, such as cell fate decisions and mechanisms of oncogenesis and resistance to chemotherapy. Characterizing the extent and logic of signaling networks is essential to understanding specificity in such physiological and pathophysiological contexts as cell fate decisions and mechanisms of oncogenesis and resistance to chemotherapy. Cell-based RNA interference (RNAi) screens enable the inference of large numbers of genes that regulate signaling pathways, but these screens cannot provide network structure directly. We describe an integrated network around the canonical receptor tyrosine kinase (RTK)–Ras–extracellular signal–regulated kinase (ERK) signaling pathway, generated by combining parallel genome-wide RNAi screens with protein-protein interaction (PPI) mapping by tandem affinity purification–mass spectrometry. We found that only a small fraction of the total number of PPI or RNAi screen hits was isolated under all conditions tested and that most of these represented the known canonical pathway components, suggesting that much of the core canonical ERK pathway is known. Because most of the newly identified regulators are likely cell type– and RTK-specific, our analysis provides a resource for understanding how output through this clinically relevant pathway is regulated in different contexts. We report in vivo roles for several of the previously unknown regulators, including CG10289 and PpV, the Drosophila orthologs of two components of the serine/threonine–protein phosphatase 6 complex; the Drosophila ortholog of TepIV, a glycophosphatidylinositol-linked protein mutated in human cancers; CG6453, a noncatalytic subunit of glucosidase II; and Rtf1, a histone methyltransferase.

A Regulatory Network of Drosophila Germline Stem Cell Self-Renewal

Stem cells possess the capacity to generate two cells of distinct fate upon division: one cell retaining stem cell identity and the other cell destined to differentiate. These cell fates are established by cell-type-specific genetic networks. To comprehensively identify components of these networks, we performed a large-scale RNAi screen in Drosophila female germline stem cells (GSCs) covering ∼25% of the genome. The screen identified 366 genes that affect GSC maintenance, differentiation, or other processes involved in oogenesis. Comparison of GSC regulators with neural stem cell self-renewal factors identifies common and cell-type-specific self-renewal genes. Importantly, we identify the histone methyltransferase Set1 as a GSC-specific self-renewal factor. Loss of Set1 in neural stem cells does not affect cell fate decisions, suggesting a differential requirement of H3K4me3 in different stem cell lineages. Altogether, our study provides a resource that will help to further dissect the networks underlying stem cell self-renewal.

Protein complex-based analysis framework for high-throughput data sets.

An analysis tool maps network dynamics at the protein complex level in multiple species. Complexes Reveal Signaling Dynamics Analysis of high-throughput data sets can provide information about changes in gene expression, protein abundance, and signaling pathway activity. However, current data mining approaches do not identify changes to functional protein complexes within a pathway over time, a critical aspect for network analysis. Vinayagam et al. developed an interactive Web tool called COMPLEAT, which uses raw genome-wide, RNA interference data to map protein complex dynamics during the cellular response to stimuli in humans, flies, and yeast. Using phosphorylated extracellular signal–regulated kinase as a marker for pathway activation, COMPLEAT identified the Brahma complex in the cellular response to insulin, a prediction that was validated in a Drosophila cell line. Analysis of high-throughput data increasingly relies on pathway annotation and functional information derived from Gene Ontology. This approach has limitations, in particular for the analysis of network dynamics over time or under different experimental conditions, in which modules within a network rather than complete pathways might respond and change. We report an analysis framework based on protein complexes, which are at the core of network reorganization. We generated a protein complex resource for human, Drosophila, and yeast from the literature and databases of protein-protein interaction networks, with each species having thousands of complexes. We developed COMPLEAT (http://www.flyrnai.org/compleat), a tool for data mining and visualization for complex-based analysis of high-throughput data sets, as well as analysis and integration of heterogeneous proteomics and gene expression data sets. With COMPLEAT, we identified dynamically regulated protein complexes among genome-wide RNA interference data sets that used the abundance of phosphorylated extracellular signal–regulated kinase in cells stimulated with either insulin or epidermal growth factor as the output. The analysis predicted that the Brahma complex participated in the insulin response.

Controllability analysis of the directed human protein interaction network identifies disease genes and drug targets

Significance Large-scale biological network analyses often use concepts used in social networks analysis (e.g. finding “communities,” “hubs,” etc.). However, mathematically advanced engineering concepts have only been applied to analyze small and well-characterized networks so far in biology. Here, we applied a sophisticated engineering tool, from control theory, to analyze a large-scale directed human protein–protein interaction network. Our analysis revealed that the proteins that are indispensable, from a network controllability perspective, are also commonly targeted by disease-causing mutations and human viruses or have been identified as drug targets. Furthermore, we used the controllability analysis to prioritize novel cancer genes from cancer genomic datasets. Altogether, we demonstrated an application of network controllability analysis to identify new disease genes and drug targets. The protein–protein interaction (PPI) network is crucial for cellular information processing and decision-making. With suitable inputs, PPI networks drive the cells to diverse functional outcomes such as cell proliferation or cell death. Here, we characterize the structural controllability of a large directed human PPI network comprising 6,339 proteins and 34,813 interactions. This network allows us to classify proteins as “indispensable,” “neutral,” or “dispensable,” which correlates to increasing, no effect, or decreasing the number of driver nodes in the network upon removal of that protein. We find that 21% of the proteins in the PPI network are indispensable. Interestingly, these indispensable proteins are the primary targets of disease-causing mutations, human viruses, and drugs, suggesting that altering a network’s control property is critical for the transition between healthy and disease states. Furthermore, analyzing copy number alterations data from 1,547 cancer patients reveals that 56 genes that are frequently amplified or deleted in nine different cancers are indispensable. Among the 56 genes, 46 of them have not been previously associated with cancer. This suggests that controllability analysis is very useful in identifying novel disease genes and potential drug targets.

Conserved Regulators of Nucleolar Size Revealed by Global Phenotypic Analyses

Loss-of-function analyses in yeast and flies identify molecular complexes that regulate ribosomal DNA transcription. Regulating Nucleolar Size The higher proliferation rate of cancer cells requires an increased rate of protein synthesis. Thus, cancer cells often show increased rates of ribosomal DNA (rDNA) transcription and have more ribosomes and larger nucleoli, which are nuclear structures that function in ribosome biogenesis. Neumüller et al. identified genes in yeast that, when ablated, resulted in smaller or larger nucleoli. A similar analysis in Drosophila enabled the identification of evolutionarily conserved molecular complexes that increase or decrease nucleolar size when the complex constituents were targeted by RNA interference. Understanding how cells regulate rDNA transcription could provide new therapeutic avenues for interfering with the unrestricted growth that occurs in cancer. Regulation of cell growth is a fundamental process in development and disease that integrates a vast array of extra- and intracellular information. A central player in this process is RNA polymerase I (Pol I), which transcribes ribosomal RNA (rRNA) genes in the nucleolus. Rapidly growing cancer cells are characterized by increased Pol I–mediated transcription and, consequently, nucleolar hypertrophy. To map the genetic network underlying the regulation of nucleolar size and of Pol I–mediated transcription, we performed comparative, genome-wide loss-of-function analyses of nucleolar size in Saccharomyces cerevisiae and Drosophila melanogaster coupled with mass spectrometry–based analyses of the ribosomal DNA (rDNA) promoter. With this approach, we identified a set of conserved and nonconserved molecular complexes that control nucleolar size. Furthermore, we characterized a direct role of the histone information regulator (HIR) complex in repressing rRNA transcription in yeast. Our study provides a full-genome, cross-species analysis of a nuclear subcompartment and shows that this approach can identify conserved molecular modules.

Combining genetic perturbations and proteomics to examine kinase-phosphatase networks in Drosophila embryos.

Connecting phosphorylation events to kinases and phosphatases is key to understanding the molecular organization and signaling dynamics of networks. We have generated a validated set of transgenic RNA-interference reagents for knockdown and characterization of all protein kinases and phosphatases present during early Drosophila melanogaster development. These genetic tools enable collection of sufficient quantities of embryos depleted of single gene products for proteomics. As a demonstration of an application of the collection, we have used multiplexed isobaric labeling for quantitative proteomics to derive global phosphorylation signatures associated with kinase-depleted embryos to systematically link phosphosites with relevant kinases. We demonstrate how this strategy uncovers kinase consensus motifs and prioritizes phosphoproteins for kinase target validation. We validate this approach by providing auxiliary evidence for Wee kinase-directed regulation of the chromatin regulator Stonewall. Further, we show how correlative phosphorylation at the site level can indicate function, as exemplified by Sterile20-like kinase-dependent regulation of Stat92E.

A computational framework for boosting confidence in high-throughput protein-protein interaction datasets.

Improving the quality and coverage of the protein interactome is of tantamount importance for biomedical research, particularly given the various sources of uncertainty in high-throughput techniques. We introduce a structure-based framework, Coev2Net, for computing a single confidence score that addresses both false-positive and false-negative rates. Coev2Net is easily applied to thousands of binary protein interactions and has superior predictive performance over existing methods. We experimentally validate selected high-confidence predictions in the human MAPK network and show that predicted interfaces are enriched for cancer -related or damaging SNPs. Coev2Net can be downloaded at http://struct2net.csail.mit.edu.