EcID. A database for the inference of functional interactions in E. coli

Computational Network Inference for Bacterial Interactomics

Since the large-scale experimental characterization of protein–protein interactions (PPIs) is not possible for all species, several computational PPI prediction methods have been developed that harness existing data from other species. While PPI network prediction has been extensively used in eukaryotes, microbial network inference has lagged behind. However, bacterial interactomes can be built using the same principles and techniques; in fact, several methods are better suited to bacterial genomes. These predicted networks allow systems-level analyses in species that lack experimental interaction data. This review describes the current network inference and analysis techniques and summarizes the use of computationally-predicted microbial interactomes to date.

Co-complex protein membership evaluation using Maximum Entropy on GO ontology and InterPro annotation

Motivation

Protein–protein interactions (PPI) play a crucial role in our understanding of protein function and biological processes. The standardization and recording of experimental findings is increasingly stored in ontologies, with the Gene Ontology (GO) being one of the most successful projects. Several PPI evaluation algorithms have been based on the application of probabilistic frameworks or machine learning algorithms to GO properties. Here, we introduce a new training set design and machine learning based approach that combines dependent heterogeneous protein annotations from the entire ontology to evaluate putative co-complex protein interactions determined by empirical studies.

Results

PPI annotations are built combinatorically using corresponding GO terms and InterPro annotation. We use a S.cerevisiae high-confidence complex dataset as a positive training set. A series of classifiers based on Maximum Entropy and support vector machines (SVMs), each with a composite counterpart algorithm, are trained on a series of training sets. These achieve a high performance area under the ROC curve of ≤0.97, outperforming go2ppi—a previously established prediction tool for protein-protein interactions (PPI) based on Gene Ontology (GO) annotations.

Availability and implementation

https://github.com/ima23/maxent-ppi

Supplementary information

Supplementary data are available at Bioinformatics online.

Emerging roles of allosteric modulators in the regulation of protein-protein interactions (PPIs): A new paradigm for PPI drug discovery

Protein-protein interactions (PPIs) are closely implicated in various types of cellular activities and are thus pivotal to health and disease states. Given their fundamental roles in a wide range of biological processes, the modulation of PPIs has enormous potential in drug discovery. However, owing to the general properties of large, flat, and featureless interfaces of PPIs, previous attempts have demonstrated that the generation of therapeutic agents targeting PPI interfaces is challenging, rendering them almost "undruggable" for decades. To date, rapid progress in chemical and structural biology techniques has promoted the exploitation of allostery as a novel approach in drug discovery. By attaching to allosteric sites that are topologically and spatially distinct from PPI interfaces, allosteric modulators can achieve improved physiochemical properties. Thus, allosteric modulators may represent an alternative strategy to target intractable PPIs and have attracted intense pharmaceutical interest. In this review, we first briefly introduce the characteristics of PPIs and then present different approaches for investigating PPIs, as well as the latest methods for modulating PPIs. Importantly, we comprehensively review the recent progress in the development of allosteric modulators to inhibit or stabilize PPIs. Finally, we conclude with future perspectives on the discovery of allosteric PPI modulators, especially the application of computational methods to aid in allosteric PPI drug discovery.

CLMSVault: A Software Suite for Protein Cross-Linking Mass-Spectrometry Data Analysis and Visualization

Protein cross-linking mass spectrometry (CL-MS) enables the sensitive detection of protein interactions and the inference of protein complex topology. The detection of chemical cross-links between protein residues can identify intra- and interprotein contact sites or provide physical constraints for molecular modeling of protein structure. Recent innovations in cross-linker design, sample preparation, mass spectrometry, and software tools have significantly improved CL-MS approaches. Although a number of algorithms now exist for the identification of cross-linked peptides from mass spectral data, a dearth of user-friendly analysis tools represent a practical bottleneck to the broad adoption of the approach. To facilitate the analysis of CL-MS data, we developed CLMSVault, a software suite designed to leverage existing CL-MS algorithms and provide intuitive and flexible tools for cross-platform data interpretation. CLMSVault stores and combines complementary information obtained from different cross-linkers and search algorithms. CLMSVault provides filtering, comparison, and visualization tools to support CL-MS analyses and includes a workflow for label-free quantification of cross-linked peptides. An embedded 3D viewer enables the visualization of quantitative data and the mapping of cross-linked sites onto PDB structural models. We demonstrate the application of CLMSVault for the analysis of a noncovalent Cdc34-ubiquitin protein complex cross-linked under different conditions. CLMSVault is open-source software (available at https://gitlab.com/courcelm/clmsvault.git ), and a live demo is available at http://democlmsvault.tyerslab.com/ .

Genomewide landscape of gene-metabolome associations in Escherichia coli

Metabolism is one of the best-understood cellular processes whose network topology of enzymatic reactions is determined by an organism's genome. The influence of genes on metabolite levels, however, remains largely unknown, particularly for the many genes encoding non-enzymatic proteins. Serendipitously, genomewide association studies explore the relationship between genetic variants and metabolite levels, but a comprehensive interaction network has remained elusive even for the simplest single-celled organisms. Here, we systematically mapped the association between > 3,800 single-gene deletions in the bacterium Escherichia coli and relative concentrations of > 7,000 intracellular metabolite ions. Beyond expected metabolic changes in the proximity to abolished enzyme activities, the association map reveals a largely unknown landscape of gene-metabolite interactions that are not represented in metabolic models. Therefore, the map provides a unique resource for assessing the genetic basis of metabolic changes and conversely hypothesizing metabolic consequences of genetic alterations. We illustrate this by predicting metabolism-related functions of 72 so far not annotated genes and by identifying key genes mediating the cellular response to environmental perturbations.

Large-Scale and Targeted Quantitative Cross-Linking MS Using Isotope-Labeled Protein Interaction Reporter (PIR) Cross-Linkers

Quantitative measurement of chemically cross-linked proteins is crucial to reveal dynamic information about protein structures and protein-protein interactions and how these are differentially regulated during stress, aging, drug treatment, and most perturbations. Previously, we demonstrated how quantitative in vivo cross-linking (CL) with stable isotope labeling of amino acids in cell culture (SILAC) enables both heritable and dynamic changes in cells to be visualized. In this work, we demonstrate the technical feasibility of proteome-scale quantitative in vivo CL-MS using isotope-labeled protein interaction reporter (PIR) cross-linkers and E. coli as a model system. This isotope-labeled cross-linkers approach, combined with Real-time Analysis of Cross-linked peptide Technology (ReACT) previously developed in our lab, enables the quantification of 941 nonredundant cross-linked peptide pairs from a total of 1213 fully identified peptide pairs in two biological replicate samples through comparison of MS¹ peak intensity of the light and heavy cross-linked peptide pairs. For targeted relative quantification of cross-linked peptide pairs, we further developed a PRM-based assay to accurately probe specific site interaction changes in a complex background. The methodology described in this work provides reliable tools for both large-scale and targeted quantitative CL-MS that is useful for any sample where SILAC labeling may not be practical.

In vivo protein interaction network analysis reveals porin-localized antibiotic inactivation in Acinetobacter baumannii strain AB5075

The nosocomial pathogen Acinetobacter baumannii is a frequent cause of hospital-acquired infections worldwide and is a challenge for treatment due to its evolved resistance to antibiotics, including carbapenems. Here, to gain insight on A. baumannii antibiotic resistance mechanisms, we analyse the protein interaction network of a multidrug-resistant A. baumannii clinical strain (AB5075). Using in vivo chemical cross-linking and mass spectrometry, we identify 2,068 non-redundant cross-linked peptide pairs containing 245 intra- and 398 inter-molecular interactions. Outer membrane proteins OmpA and YiaD, and carbapenemase Oxa-23 are hubs of the identified interaction network. Eighteen novel interactors of Oxa-23 are identified. Interactions of Oxa-23 with outer membrane porins OmpA and CarO are verified with co-immunoprecipitation analysis. Furthermore, transposon mutagenesis of oxa-23 or interactors of Oxa-23 demonstrates changes in meropenem or imipenem sensitivity in strain AB5075. These results provide a view of porin-localized antibiotic inactivation and increase understanding of bacterial antibiotic resistance mechanisms.

Identification of 2-oxohistidine Interacting Proteins Using E. coli Proteome Chips

Cellular proteins are constantly damaged by reactive oxygen species generated by cellular respiration. Because of its metal-chelating property, the histidine residue is easily oxidized in the presence of Cu/Fe ions and H₂O₂ via metal-catalyzed oxidation, usually converted to 2-oxohistidine. We hypothesized that cells may have evolved antioxidant defenses against the generation of 2-oxohistidine residues on proteins, and therefore there would be cellular proteins which specifically interact with this oxidized side chain. Using two chemically synthesized peptide probes containing 2-oxohistidine, high-throughput interactome screening was conducted using the E. coli K12 proteome microarray containing >4200 proteins. Ten interacting proteins were identified, and successfully validated using a third peptide probe, fluorescence polarization assays, as well as binding constant measurements. We discovered that 9 out of 10 identified proteins seemed to be involved in redox-related cellular functions. We also built the functional interaction network to reveal their interacting proteins. The network showed that our interacting proteins were enriched in oxido-reduction processes, ion binding, and carbon metabolism. A consensus motif was identified among these 10 bacterial interacting proteins based on bioinformatic analysis, which also appeared to be present on human S100A1 protein. Besides, we found that the consensus binding motif among our identified proteins, including bacteria and human, were located within α-helices and faced the outside of proteins. The combination of chemically engineered peptide probes with proteome microarrays proves to be an efficient discovery platform for protein interactomes of unusual post-translational modifications, and sensitive enough to detect even the insertion of a single oxygen atom in this case.

Novel function discovery with GeneMANIA: a new integrated resource for gene function prediction in Escherichia coli

MOTIVATION

The model bacterium Escherichia coli is among the best studied prokaryotes, yet nearly half of its proteins are still of unknown biological function. This is despite a wealth of available large-scale physical and genetic interaction data. To address this, we extended the GeneMANIA function prediction web application developed for model eukaryotes to support E.coli.

RESULTS

We integrated 48 distinct E.coli functional interaction datasets and used the GeneMANIA algorithm to produce thousands of novel functional predictions and prioritize genes for further functional assays. Our analysis achieved cross-validation performance comparable to that reported for eukaryotic model organisms, and revealed new functions for previously uncharacterized genes in specific bioprocesses, including components required for cell adhesion, iron-sulphur complex assembly and ribosome biogenesis. The GeneMANIA approach for network-based function prediction provides an innovative new tool for probing mechanisms underlying bacterial bioprocesses.

CONTACT

gary.bader@utoronto.ca; mohan.babu@uregina.ca

SUPPLEMENTARY INFORMATION

Supplementary data are available at Bioinformatics online.

Practical aspects of protein co-evolution

Co-evolution is a fundamental aspect of Evolutionary Theory. At the molecular level, co-evolutionary linkages between protein families have been used as indicators of protein interactions and functional relationships from long ago. Due to the complexity of the problem and the amount of genomic data required for these approaches to achieve good performances, it took a relatively long time from the appearance of the first ideas and concepts to the quotidian application of these approaches and their incorporation to the standard toolboxes of bioinformaticians and molecular biologists. Today, these methodologies are mature (both in terms of performance and usability/implementation), and the genomic information that feeds them large enough to allow their general application. This review tries to summarize the current landscape of co-evolution-based methodologies, with a strong emphasis on describing interesting cases where their application to important biological systems, alone or in combination with other computational and experimental approaches, allowed getting new insight into these.

XLink-DB: database and software tools for storing and visualizing protein interaction topology data

As large-scale cross-linking data becomes available, new software tools for data processing and visualization are required to replace manual data analysis. XLink-DB serves as a data storage site and visualization tool for cross-linking results. XLink-DB accepts data generated with any cross-linker and stores them in a relational database. Cross-linked sites are automatically mapped onto PDB structures if available, and results are compared to existing protein interaction databases. A protein interaction network is also automatically generated for the entire data set. The XLink-DB server, including examples, and a help page are available for noncommercial use at http://brucelab.gs.washington.edu/crosslinkdbv1/ . The source code can be viewed and downloaded at https://sourceforge.net/projects/crosslinkdb/?source=directory .

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.

Proteome-wide prediction of protein-protein interactions from high-throughput data

In this paper, we present a brief review of the existing computational methods for predicting proteome-wide protein-protein interaction networks from high-throughput data. The availability of various types of omics data provides great opportunity and also unprecedented challenge to infer the interactome in cells. Reconstructing the interactome or interaction network is a crucial step for studying the functional relationship among proteins and the involved biological processes. The protein interaction network will provide valuable resources and alternatives to decipher the mechanisms of these functionally interacting elements as well as the running system of cellular operations. In this paper, we describe the main steps of predicting protein-protein interaction networks and categorize the available approaches to couple the physical and functional linkages. The future topics and the analyses beyond prediction are also discussed and concluded.

Inferring a protein interaction map of Mycobacterium tuberculosis based on sequences and interologs

BACKGROUND

Mycobacterium tuberculosis is an infectious bacterium posing serious threats to human health. Due to the difficulty in performing molecular biology experiments to detect protein interactions, reconstruction of a protein interaction map of M. tuberculosis by computational methods will provide crucial information to understand the biological processes in the pathogenic microorganism, as well as provide the framework upon which new therapeutic approaches can be developed.

RESULTS

In this paper, we constructed an integrated M. tuberculosis protein interaction network by machine learning and ortholog-based methods. Firstly, we built a support vector machine (SVM) method to infer the protein interactions of M. tuberculosis H37Rv by gene sequence information. We tested our predictors in Escherichia coli and mapped the genetic codon features underlying its protein interactions to M. tuberculosis. Moreover, the documented interactions of 14 other species were mapped to the interactome of M. tuberculosis by the interolog method. The ensemble protein interactions were validated by various functional relationships, i.e., gene coexpression, evolutionary relationship and functional similarity, extracted from heterogeneous data sources. The accuracy and validation demonstrate the effectiveness and efficiency of our framework.

CONCLUSIONS

A protein interaction map of M. tuberculosis is inferred from genetic codons and interologs. The prediction accuracy and numerically experimental validation demonstrate the effectiveness and efficiency of our method. Furthermore, our methods can be straightforwardly extended to infer the protein interactions of other bacterial species.

Short Co-occurring Polypeptide Regions Can Predict Global Protein Interaction Maps

A goal of the post-genomics era has been to elucidate a detailed global map of protein-protein interactions (PPIs) within a cell. Here, we show that the presence of co-occurring short polypeptide sequences between interacting protein partners appears to be conserved across different organisms. We present an algorithm to automatically generate PPI prediction method parameters for various organisms and illustrate that global PPIs can be predicted from previously reported PPIs within the same or a different organism using protein primary sequences. The PPI prediction code is further accelerated through the use of parallel multi-core programming, which improves its usability for large scale or proteome-wide PPI prediction. We predict and analyze hundreds of novel human PPIs, experimentally confirm protein functions and importantly predict the first genome-wide PPI maps for S. pombe (∼9,000 PPIs) and C. elegans (∼37,500 PPIs).

Identification of lactoferricin B intracellular targets using an Escherichia coli proteome chip

Lactoferricin B (LfcinB) is a well-known antimicrobial peptide. Several studies have indicated that it can inhibit bacteria by affecting intracellular activities, but the intracellular targets of this antimicrobial peptide have not been identified. Therefore, we used E. coli proteome chips to identify the intracellular target proteins of LfcinB in a high-throughput manner. We probed LfcinB with E. coli proteome chips and further conducted normalization and Gene Ontology (GO) analyses. The results of the GO analyses showed that the identified proteins were associated with metabolic processes. Moreover, we validated the interactions between LfcinB and chip assay-identified proteins with fluorescence polarization (FP) assays. Sixteen proteins were identified, and an E. coli interaction database (EcID) analysis revealed that the majority of the proteins that interact with these 16 proteins affected the tricarboxylic acid (TCA) cycle. Knockout assays were conducted to further validate the FP assay results. These results showed that phosphoenolpyruvate carboxylase was a target of LfcinB, indicating that one of its mechanisms of action may be associated with pyruvate metabolism. Thus, we used pyruvate assays to conduct an in vivo validation of the relationship between LfcinB and pyruvate level in E. coli. These results showed that E. coli exposed to LfcinB had abnormal pyruvate amounts, indicating that LfcinB caused an accumulation of pyruvate. In conclusion, this study successfully revealed the intracellular targets of LfcinB using an E. coli proteome chip approach.

Stress response regulators identified through genome-wide transcriptome analysis of the (p)ppGpp-dependent response in Rhizobium etli

BACKGROUND

The alarmone (p)ppGpp mediates a global reprogramming of gene expression upon nutrient limitation and other stresses to cope with these unfavorable conditions. Synthesis of (p)ppGpp is, in most bacteria, controlled by RelA/SpoT (Rsh) proteins. The role of (p)ppGpp has been characterized primarily in Escherichia coli and several Gram-positive bacteria. Here, we report the first in-depth analysis of the (p)ppGpp-regulon in an α-proteobacterium using a high-resolution tiling array to better understand the pleiotropic stress phenotype of a relA/rsh mutant.

RESULTS

We compared gene expression of the Rhizobium etli wild type and rsh (previously rel) mutant during exponential and stationary phase, identifying numerous (p)ppGpp targets, including small non-coding RNAs. The majority of the 834 (p)ppGpp-dependent genes were detected during stationary phase. Unexpectedly, 223 genes were expressed (p)ppGpp-dependently during early exponential phase, indicating the hitherto unrecognized importance of (p)ppGpp during active growth. Furthermore, we identified two (p)ppGpp-dependent key regulators for survival during heat and oxidative stress and one regulator putatively involved in metabolic adaptation, namely extracytoplasmic function sigma factor EcfG2/PF00052, transcription factor CH00371, and serine protein kinase PrkA.

CONCLUSIONS

The regulatory role of (p)ppGpp in R. etli stress adaptation is far-reaching in redirecting gene expression during all growth phases. Genome-wide transcriptome analysis of a strain deficient in a global regulator, and exhibiting a pleiotropic phenotype, enables the identification of more specific regulators that control genes associated with a subset of stress phenotypes. This work is an important step toward a full understanding of the regulatory network underlying stress responses in α-proteobacteria.

Evolution of the RpoS regulon: origin of RpoS and the conservation of RpoS-dependent regulation in bacteria

The RpoS sigma factor in proteobacteria regulates genes in stationary phase and in response to stress. Although of conserved function, the RpoS regulon may have different gene composition across species due to high genomic diversity and to known environmental conditions that select for RpoS mutants. In this study, the distribution of RpoS homologs in prokaryotes and the differential dependence of regulon members on RpoS for expression in two gamma-proteobacteria (Escherichia coli and Pseudomonas aeruginosa) were examined. Using a maximum-likelihood phylogeny and reciprocal best hits analysis, we show that the RpoS sigma factor is conserved within gamma-, beta-, and delta-proteobacteria. Annotated RpoS of Borrelia and the enteric RpoS are postulated to have separate evolutionary origins. To determine the conservation of RpoS-dependent gene expression across species, reciprocal best hits analysis was used to identify orthologs of the E. coli RpoS regulon in the RpoS regulon of P. aeruginosa. Of the 186 RpoS-dependent genes of E. coli, 50 proteins have an ortholog within the P. aeruginosa genome. Twelve genes of the 50 orthologs are RpoS-dependent in both species, and at least four genes are regulated by RpoS in other gamma-proteobacteria. Despite RpoS conservation in gamma-, beta-, and delta-proteobacteria, RpoS regulon composition is subject to modification between species. Environmental selection for RpoS mutants likely contributes to the evolutionary divergence and specialization of the RpoS regulon within different bacterial genomes.

Inference of functional relations in predicted protein networks with a machine learning approach

Background

Molecular biology is currently facing the challenging task of functionally characterizing the proteome. The large number of possible protein-protein interactions and complexes, the variety of environmental conditions and cellular states in which these interactions can be reorganized, and the multiple ways in which a protein can influence the function of others, requires the development of experimental and computational approaches to analyze and predict functional associations between proteins as part of their activity in the interactome.

Methodology/Principal Findings

We have studied the possibility of constructing a classifier in order to combine the output of the several protein interaction prediction methods. The AODE (Averaged One-Dependence Estimators) machine learning algorithm is a suitable choice in this case and it provides better results than the individual prediction methods, and it has better performances than other tested alternative methods in this experimental set up. To illustrate the potential use of this new AODE-based Predictor of Protein InterActions (APPIA), when analyzing high-throughput experimental data, we show how it helps to filter the results of published High-Throughput proteomic studies, ranking in a significant way functionally related pairs. Availability: All the predictions of the individual methods and of the combined APPIA predictor, together with the used datasets of functional associations are available at http://ecid.bioinfo.cnio.es/.

Conclusions

We propose a strategy that integrates the main current computational techniques used to predict functional associations into a unified classifier system, specifically focusing on the evaluation of poorly characterized protein pairs. We selected the AODE classifier as the appropriate tool to perform this task. AODE is particularly useful to extract valuable information from large unbalanced and heterogeneous data sets. The combination of the information provided by five prediction interaction prediction methods with some simple sequence features in APPIA is useful in establishing reliability values and helpful to prioritize functional interactions that can be further experimentally characterized.

Systems-level approaches for identifying and analyzing genetic interaction networks in Escherichia coli and extensions to other prokaryotes

Molecular interactions define the functional organization of the cell. Epistatic (genetic, or gene-gene) interactions, one of the most informative and commonly encountered forms of functional relationships, are increasingly being used to map process architecture in model eukaryotic organisms. In particular, 'systems-level' screens in yeast and worm aimed at elucidating genetic interaction networks have led to the generation of models describing the global modular organization of gene products and protein complexes within a cell. However, comparable data for prokaryotic organisms have not been available. Given its ease of growth and genetic manipulation, the Gram-negative bacterium Escherichia coli appears to be an ideal model system for performing comprehensive genome-scale examinations of genetic redundancy in bacteria. In this review, we highlight emerging experimental and computational techniques that have been developed recently to examine functional relationships and redundancy in E. coli at a systems-level, and their potential application to prokaryotes in general. Additionally, we have scanned PubMed abstracts and full-text published articles to manually curate a list of approximately 200 previously reported synthetic sick or lethal genetic interactions in E. coli derived from small-scale experimental studies.

Global functional atlas of Escherichia coli encompassing previously uncharacterized proteins

One-third of the 4,225 protein-coding genes of Escherichia coli K-12 remain functionally unannotated (orphans). Many map to distant clades such as Archaea, suggesting involvement in basic prokaryotic traits, whereas others appear restricted to E. coli, including pathogenic strains. To elucidate the orphans' biological roles, we performed an extensive proteomic survey using affinity-tagged E. coli strains and generated comprehensive genomic context inferences to derive a high-confidence compendium for virtually the entire proteome consisting of 5,993 putative physical interactions and 74,776 putative functional associations, most of which are novel. Clustering of the respective probabilistic networks revealed putative orphan membership in discrete multiprotein complexes and functional modules together with annotated gene products, whereas a machine-learning strategy based on network integration implicated the orphans in specific biological processes. We provide additional experimental evidence supporting orphan participation in protein synthesis, amino acid metabolism, biofilm formation, motility, and assembly of the bacterial cell envelope. This resource provides a "systems-wide" functional blueprint of a model microbe, with insights into the biological and evolutionary significance of previously uncharacterized proteins.