Computational prediction of host-pathogen protein-protein interactions

MOTIVATION: Infectious diseases such as malaria result in millions of deaths each year. An important aspect of any host-pathogen system is the mechanism by which a pathogen can infect its host. One method of infection is via protein-protein interactions (PPIs) where pathogen proteins target host proteins. Developing computational methods that identify which PPIs enable a pathogen to infect a host has great implications in identifying potential targets for therapeutics. RESULTS: We present a method that integrates known intra-species PPIs with protein-domain profiles to predict PPIs between host and pathogen proteins. Given a set of intra-species PPIs, we identify the functional domains in each of the interacting proteins. For every pair of functional domains, we use Bayesian statistics to assess the probability that two proteins with that pair of domains will interact. We apply our method to the Homo sapiens-Plasmodium falciparum host-pathogen system. Our system predicts 516 PPIs between proteins from these two organisms. We show that pairs of human proteins we predict to interact with the same Plasmodium protein are close to each other in the human PPI network and that Plasmodium pairs predicted to interact with same human protein are co-expressed in DNA microarray datasets measured during various stages of the Plasmodium life cycle. Finally, we identify functionally enriched sub-networks spanned by the predicted interactions and discuss the plausibility of our predictions. AVAILABILITY: Supplementary data are available at http://staff.vbi.vt.edu/dyermd/publications/dyer2007a.html. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.     

Host pathogen protein interactions predicted by comparative modeling

Pathogens have evolved numerous strategies to infect their hosts, while hosts have evolved immune responses and other defenses to these foreign challenges. The vast majority of host-pathogen interactions involve protein-protein recognition, yet our current understanding of these interactions is limited. Here, we present and apply a computational whole-genome protocol that generates testable predictions of host-pathogen protein interactions. The protocol first scans the host and pathogen genomes for proteins with similarity to known protein complexes, then assesses these putative interactions, using structure if available, and, finally, filters the remaining interactions using biological context, such as the stage-specific expression of pathogen proteins and tissue expression of host proteins. The technique was applied to 10 pathogens, including species of Mycobacterium, apicomplexa, and kinetoplastida, responsible for "neglected" human diseases. The method was assessed by (1) comparison to a set of known host-pathogen interactions, (2) comparison to gene expression and essentiality data describing host and pathogen genes involved in infection, and (3) analysis of the functional properties of the human proteins predicted to interact with pathogen proteins, demonstrating an enrichment for functionally relevant host-pathogen interactions. We present several specific predictions that warrant experimental follow-up, including interactions from previously characterized mechanisms, such as cytoadhesion and protease inhibition, as well as suspected interactions in hypothesized networks, such as apoptotic pathways. Our computational method provides a means to mine whole-genome data and is complementary to experimental efforts in elucidating networks of host-pathogen protein interactions.     

Inferring protein-protein interactions through high-throughput interaction data from diverse organisms

MOTIVATION: Identifying protein-protein interactions is critical for understanding cellular processes. Because protein domains represent binding modules and are responsible for the interactions between proteins, computational approaches have been proposed to predict protein interactions at the domain level. The fact that protein domains are likely evolutionarily conserved allows us to pool information from data across multiple organisms for the inference of domain-domain and protein-protein interaction probabilities. RESULTS: We use a likelihood approach to estimating domain-domain interaction probabilities by integrating large-scale protein interaction data from three organisms, Saccharomyces cerevisiae, Caenorhabditis elegans and Drosophila melanogaster. The estimated domain-domain interaction probabilities are then used to predict protein-protein interactions in S.cerevisiae. Based on a thorough comparison of sensitivity and specificity, Gene Ontology term enrichment and gene expression profiles, we have demonstrated that it may be far more informative to predict protein-protein interactions from diverse organisms than from a single organism. AVAILABILITY: The program for computing the protein-protein interaction probabilities and supplementary material are available at http://bioinformatics.med.yale.edu/interaction.     

Prediction,assessment and validation of protein interaction maps in bacteria

High-throughput proteomics technologies, especially the yeast two-hybrid system, produce large volumes of protein-protein interaction data organized in networks. The complete sequencing of many genomes raises questions about the extent to which such networks can be transferred between organisms. We attempted to answer this question using the experimentally derived Helicobacter pylori interaction map and the recently described interacting domain profile pair (IDPP) method to predict a virtual map for Escherichia coli. The extensive literature concerning E.coli was used to assess all predicted interactions and to validate the IDPP method, which clusters protein domains by sequence and connectivity similarities. The IDPP method has a much better heuristic value than methods solely based on protein homology. The IDPP method was further applied to Campylobacter jejuni to generate a virtual interaction map. An in-depth comparison of the chemotaxis pathways predicted in E.coli and C.jejuni led to the proposition of new functional assignments. Finally, the prediction of protein-protein interaction maps across organisms enabled us to validate some of the interactions on the original experimental map.     

Prediction and comparison of Salmonella-human and Salmonella-Arabidopsis interactomes

Salmonellosis caused by Salmonella bacteria is a food-borne disease and a worldwide health threat causing millions of infections and thousands of deaths every year. This pathogen infects an unusually broad range of host organisms including human and plants. A better understanding of the mechanisms of communication between Salmonella and its hosts requires identifying the interactions between Salmonella and host proteins. Protein-protein interactions (PPIs) are the fundamental building blocks of communication. Here, we utilize the prediction platform BIANA to obtain the putative Salmonella-human and Salmonella-Arabidopsis interactomes based on sequence and domain similarity to known PPIs. A gold standard list of Salmonella-host PPIs served to validate the quality of the human model. 24,726 and 10,926 PPIs comprising interactions between 38 and 33 Salmonella effectors and virulence factors with 9,740 human and 4,676 Arabidopsis proteins, respectively, were predicted. Putative hub proteins could be identified, and parallels between the two interactomes were discovered. This approach can provide insight into possible biological functions of so far uncharacterized proteins. The predicted interactions are available via a web interface which allows filtering of the database according to parameters provided by the user to narrow down the list of suspected interactions. The interactions are available via a web interface at http://sbi.imim.es/web/SHIPREC.php.     

Proteome-wide prediction of overlapping small molecule and protein binding sites using structure

Small molecules that modulate protein-protein interactions are of great interest for chemical biology and therapeutics. Here I present a structure-based approach to predict 'bi-functional' sites able to bind both small molecule ligands and proteins, in proteins of unknown structure. First, I develop a homology-based annotation method that transfers binding sites of known three-dimensional structure onto protein sequences, predicting residues in ligand and protein binding sites with estimated true positive rates of 98% and 88%, respectively, at 1% false positive rates. Applying this method to the human proteome predicts 8463 proteins with bi-functional residues and correctly recovers the targets of known interaction modulators. Proteins with significantly (p < 0.01) more bi-functional residues than expected were found to be enriched in regulatory and depleted in metabolism functions. Finally, I demonstrate the utility of the method by describing examples of predicted overlap and evidence of their biological and therapeutic relevance. The results suggest that combining the structures of known binding sites with established fold detection algorithms can predict regions of protein-protein interfaces that are amenable to small molecule modulation. Open-source software and the results for several complete proteomes are available at http://pibase.janelia.org/homolobind.     

Integrated prediction of the helical membrane protein interactome in yeast

At least a quarter of all genes in most genomes contain putative transmembrane (TM) helices, and helical membrane protein interactions are a major component of the overall cellular interactome. However, current experimental techniques for large-scale detection of protein-protein interactions are biased against membrane proteins. Here, we define protein-protein interaction broadly as co-complexation, and develop a weighted-voting procedure to predict interactions among yeast helical membrane proteins by optimally combining evidence based on diverse genome-wide information such as sequence, function, localization, abundance, regulation, and phenotype. We use logistic regression to simultaneously optimize the weights of all evidence sources for best discrimination based on a set of known helical membrane protein interactions. The resulting integrated classifier not only significantly outperforms classifiers based on any single genomic feature, but also does better than a benchmark Na?ve Bayes classifier (using a simplifying assumption of conditional independence among features). Finally, we apply the optimized classifier genome-wide, and construct a comprehensive map of predicted helical membrane protein interactome in yeast. This can serve as a guide for prioritizing further experimental validation efforts.     

An evaluation of in vitro protein-protein interaction techniques: assessing contaminating background proteins

Determination of protein-protein interactions is an important component in assigning function and discerning the biological relevance of proteins within a broader cellular context. In vitro protein-protein interaction methodologies, including affinity chromatography, coimmunoprecipitation, and newer approaches such as protein chip arrays, hold much promise in the detection of protein interactions, particularly in well-characterized organisms with sequenced genomes. However, each of these approaches attracts certain background proteins that can thwart detection and identification of true interactors. In addition, recombinant proteins expressed in Escherichia coli are also extensively used to assess protein-protein interactions, and background proteins in these isolates can thus contaminate interaction studies. Rigorous validation of a true interaction thus requires not only that an interaction be found by alternate techniques, but more importantly that researchers be aware of and control for matrix/support dependence. Here, we evaluate these methods for proteins interacting with DmsD (an E. coli redox enzyme maturation protein chaperone), in vitro, using E. coli subcellular fractions as prey sources. We compare and contrast the various in vitro interaction methods to identify some of the background proteins and protein profiles that are inherent to each of the methods in an E. coli system.     

Shared molecular strategies of the malaria parasite P. falciparum and the human virus HIV-1

We augmented existing computationally predicted and experimentally determined interactions with evolutionarily conserved interactions between proteins of the malaria parasite, P. falciparum, and the human host. In a validation step, we found that conserved interacting host-parasite protein pairs were specifically expressed in host tissues where both the parasite and host proteins are known to be active. We compared host-parasite interactions with experimentally verified interactions between human host proteins and a very different pathogen, HIV-1. Both pathogens were found to use their protein repertoire in a combinatorial manner, providing a broad connection to host cellular processes. Specifically, the two biologically distinct pathogens predominately target central proteins to take control of a human host cell, effectively reaching into diversified cellular host cellular functions. Interacting signaling pathways and a small set of regulatory and signaling proteins were prime targets of both pathogens, suggesting remarkably similar patterns of host-pathogen interactions despite the vast biological differences of both pathogens. Such an identification of shared molecular strategies by the virus HIV-1 and the eukaryotic intracellular pathogen P. falciparum may allow us to illuminate new avenues of disease intervention.     

Topological structure analysis of the protein-protein interaction network in budding yeast

Interaction detection methods have led to the discovery of thousands of interactions between proteins, and discerning relevance within large-scale data sets is important to present-day biology. Here, a spectral method derived from graph theory was introduced to uncover hidden topological structures (i.e. quasi-cliques and quasi-bipartites) of complicated protein-protein interaction networks. Our analyses suggest that these hidden topological structures consist of biologically relevant functional groups. This result motivates a new method to predict the function of uncharacterized proteins based on the classification of known proteins within topological structures. Using this spectral analysis method, 48 quasi-cliques and six quasi-bipartites were isolated from a network involving 11,855 interactions among 2617 proteins in budding yeast, and 76 uncharacterized proteins were assigned functions.     

Building and analyzing protein interactome networks by cross-species comparisons

Background  

A genomic catalogue of protein-protein interactions is a rich source of information, particularly for exploring the relationships between proteins. Numerous systems-wide and small-scale experiments have been conducted to identify interactions; however, our knowledge of all interactions for any one species is incomplete, and alternative means to expand these network maps is needed. We therefore took a comparative biology approach to predict protein-protein interactions across five species (human, mouse, fly, worm, and yeast) and developed InterologFinder for research biologists to easily navigate this data. We also developed a confidence score for interactions based on available experimental evidence and conservation across species.     

Molecular dynamics shows complex interplay and long-range effects of post-translational modifications in yeast protein interactions

Post-translational modifications (PTMs) play a vital, yet often overlooked role in the living cells through modulation of protein properties, such as localization and affinity towards their interactors, thereby enabling quick adaptation to changing environmental conditions. We have previously benchmarked a computational framework for the prediction of PTMs’ effects on the stability of protein-protein interactions, which has molecular dynamics simulations followed by free energy calculations at its core. In the present work, we apply this framework to publicly available data on Saccharomyces cerevisiae protein structures and PTM sites, identified in both normal and stress conditions. We predict proteome-wide effects of acetylations and phosphorylations on protein-protein interactions and find that acetylations more frequently have locally stabilizing roles in protein interactions, while the opposite is true for phosphorylations. However, the overall impact of PTMs on protein-protein interactions is more complex than a simple sum of local changes caused by the introduction of PTMs and adds to our understanding of PTM cross-talk. We further use the obtained data to calculate the conformational changes brought about by PTMs. Finally, conservation of the analyzed PTM residues in orthologues shows that some predictions for yeast proteins will be mirrored to other organisms, including human. This work, therefore, contributes to our overall understanding of the modulation of the cellular protein interaction networks in yeast and beyond.     

Plasticity of the gene functions for DNA replication in the T4-like phages

We have completely sequenced and annotated the genomes of several relatives of the bacteriophage T4, including three coliphages (RB43, RB49 and RB69), three Aeromonas salmonicida phages (44RR2.8t, 25 and 31) and one Aeromonas hydrophila phage (Aeh1). In addition, we have partially sequenced and annotated the T4-like genomes of coliphage RB16 (a close relative of RB43), A. salmonicida phage 65, Acinetobacter johnsonii phage 133 and Vibrio natriegens phage nt-1. Each of these phage genomes exhibited a unique sequence that distinguished it from its relatives, although there were examples of genomes that are very similar to each other. As a group the phages compared here diverge from one another by several criteria, including (a) host range, (b) genome size in the range between approximately 160 kb and approximately 250 kb, (c) content and genetic organization of their T4-like genes for DNA metabolism, (d) mutational drift of the predicted T4-like gene products and their regulatory sites and (e) content of open-reading frames that have no counterparts in T4 or other known organisms (novel ORFs). We have observed a number of DNA rearrangements of the T4 genome type, some exhibiting proximity to putative homing endonuclease genes. Also, we cite and discuss examples of sequence divergence in the predicted sites for protein-protein and protein-nucleic acid interactions of homologues of the T4 DNA replication proteins, with emphasis on the diversity in sequence, molecular form and regulation of the phage-encoded DNA polymerase, gp43. Five of the sequenced phage genomes are predicted to encode split forms of this polymerase. Our studies suggest that the modular construction and plasticity of the T4 genome type and several of its replication proteins may offer resilience to mutation, including DNA rearrangements, and facilitate the adaptation of T4-like phages to different bacterial hosts in nature.     

Robust co-regulation of tyrosine phosphorylation sites on proteins reveals novel protein interactions

Cell signaling networks propagate information from extracellular cues via dynamic modulation of protein-protein interactions in a context-dependent manner. Networks based on receptor tyrosine kinases (RTKs), for example, phosphorylate intracellular proteins in response to extracellular ligands, resulting in dynamic protein-protein interactions that drive phenotypic changes. Most commonly used methods for discovering these protein-protein interactions, however, are optimized for detecting stable, longer-lived complexes, rather than the type of transient interactions that are essential components of dynamic signaling networks such as those mediated by RTKs. Substrate phosphorylation downstream of RTK activation modifies substrate activity and induces phospho-specific binding interactions, resulting in the formation of large transient macromolecular signaling complexes. Since protein complex formation should follow the trajectory of events that drive it, we reasoned that mining phosphoproteomic datasets for highly similar dynamic behavior of measured phosphorylation sites on different proteins could be used to predict novel, transient protein-protein interactions that had not been previously identified. We applied this method to explore signaling events downstream of EGFR stimulation. Our computational analysis of robustly co-regulated phosphorylation sites, based on multiple clustering analysis of quantitative time-resolved mass-spectrometry phosphoproteomic data, not only identified known sitewise-specific recruitment of proteins to EGFR, but also predicted novel, a priori interactions. A particularly intriguing prediction of EGFR interaction with the cytoskeleton-associated protein PDLIM1 was verified within cells using co-immunoprecipitation and in situ proximity ligation assays. Our approach thus offers a new way to discover protein-protein interactions in a dynamic context- and phosphorylation site-specific manner.     

Towards an eco‐phylogenetic framework for infectious disease ecology

Identifying patterns and drivers of infectious disease dynamics across multiple scales is a fundamental challenge for modern science. There is growing awareness that it is necessary to incorporate multi‐host and/or multi‐parasite interactions to understand and predict current and future disease threats better, and new tools are needed to help address this task. Eco‐phylogenetics (phylogenetic community ecology) provides one avenue for exploring multi‐host multi‐parasite systems, yet the incorporation of eco‐phylogenetic concepts and methods into studies of host pathogen dynamics has lagged behind. Eco‐phylogenetics is a transformative approach that uses evolutionary history to infer present‐day dynamics. Here, we present an eco‐phylogenetic framework to reveal insights into parasite communities and infectious disease dynamics across spatial and temporal scales. We illustrate how eco‐phylogenetic methods can help untangle the mechanisms of host–parasite dynamics from individual (e.g. co‐infection) to landscape scales (e.g. parasite/host community structure). An improved ecological understanding of multi‐host and multi‐pathogen dynamics across scales will increase our ability to predict disease threats.     

Protein-lipid interactions of bacteriophage M13 major coat protein

During the past years, remarkable progress has been made in our understanding of the replication cycle of bacteriophage M13 and the molecular details that enable phage proteins to navigate in the complex environment of the host cell. With new developments in molecular membrane biology in combination with spectroscopic techniques, we are now in a position to ask how phages carry out this delicate process on a molecular level, and what sort of protein-lipid and protein-protein interactions are involved. In this review we will focus on the molecular details of the protein-protein and protein-lipid interactions of the major coat protein (gp8) that may play a role during the infection of Escherichia coli by bacteriophage M13.     

Dissecting the specificity of protein-protein interaction in bacterial two-component signaling: orphans and crosstalks

Predictive understanding of the myriads of signal transduction pathways in a cell is an outstanding challenge of systems biology. Such pathways are primarily mediated by specific but transient protein-protein interactions, which are difficult to study experimentally. In this study, we dissect the specificity of protein-protein interactions governing two-component signaling (TCS) systems ubiquitously used in bacteria. Exploiting the large number of sequenced bacterial genomes and an operon structure which packages many pairs of interacting TCS proteins together, we developed a computational approach to extract a molecular interaction code capturing the preferences of a small but critical number of directly interacting residue pairs. This code is found to reflect physical interaction mechanisms, with the strongest signal coming from charged amino acids. It is used to predict the specificity of TCS interaction: Our results compare favorably to most available experimental results, including the prediction of 7 (out of 8 known) interaction partners of orphan signaling proteins in Caulobacter crescentus. Surveying among the available bacterial genomes, our results suggest 15～25% of the TCS proteins could participate in out-of-operon "crosstalks". Additionally, we predict clusters of crosstalking candidates, expanding from the anecdotally known examples in model organisms. The tools and results presented here can be used to guide experimental studies towards a system-level understanding of two-component signaling.     

Efficient Prediction of Co-Complexed Proteins Based on Coevolution

The prediction of the network of protein-protein interactions (PPI) of an organism is crucial for the understanding of biological processes and for the development of new drugs. Machine learning methods have been successfully applied to the prediction of PPI in yeast by the integration of multiple direct and indirect biological data sources. However, experimental data are not available for most organisms. We propose here an ensemble machine learning approach for the prediction of PPI that depends solely on features independent from experimental data. We developed new estimators of the coevolution between proteins and combined them in an ensemble learning procedure.We applied this method to a dataset of known co-complexed proteins in Escherichia coli and compared it to previously published methods. We show that our method allows prediction of PPI with an unprecedented precision of 95.5% for the first 200 sorted pairs of proteins compared to 28.5% on the same dataset with the previous best method.A close inspection of the best predicted pairs allowed us to detect new or recently discovered interactions between chemotactic components, the flagellar apparatus and RNA polymerase complexes in E. coli.     

Predicting protein--protein interactions from primary structure

MOTIVATION: An ambitious goal of proteomics is to elucidate the structure, interactions and functions of all proteins within cells and organisms. The expectation is that this will provide a fuller appreciation of cellular processes and networks at the protein level, ultimately leading to a better understanding of disease mechanisms and suggesting new means for intervention. This paper addresses the question: can protein-protein interactions be predicted directly from primary structure and associated data? Using a diverse database of known protein interactions, a Support Vector Machine (SVM) learning system was trained to recognize and predict interactions based solely on primary structure and associated physicochemical properties. RESULTS: Inductive accuracy of the trained system, defined here as the percentage of correct protein interaction predictions for previously unseen test sets, averaged 80% for the ensemble of statistical experiments. Future proteomics studies may benefit from this research by proceeding directly from the automated identification of a cell's gene products to prediction of protein interaction pairs.