Protein–Protein Docking: Past,Present, and Future

The Protein Journal - The biological significance of proteins attracted the scientific community in exploring their characteristics. The studies shed light on the interaction patterns and functions...     

Integrating Cross-Linking Experiments with Ab Initio Protein–Protein Docking

Ab initio protein–protein docking algorithms often rely on experimental data to identify the most likely complex structure. We integrated protein–protein docking with the experimental data of chemical cross-linking followed by mass spectrometry. We tested our approach using 19 cases that resulted from an exhaustive search of the Protein Data Bank for protein complexes with cross-links identified in our experiments. We implemented cross-links as constraints based on Euclidean distance or void-volume distance. For most test cases, the rank of the top-scoring near-native prediction was improved by at least twofold compared with docking without the cross-link information, and the success rate for the top 5 predictions nearly tripled. Our results demonstrate the delicate balance between retaining correct predictions and eliminating false positives. Several test cases had multiple components with distinct interfaces, and we present an approach for assigning cross-links to the interfaces. Employing the symmetry information for these cases further improved the performance of complex structure prediction.     

A Second-generation Protein–Protein Interaction Network of Helicobacter pylori

Helicobacter pylori infections cause gastric ulcers and play a major role in the development of gastric cancer. In 2001, the first protein interactome was published for this species, revealing over 1500 binary protein interactions resulting from 261 yeast two-hybrid screens. Here we roughly double the number of previously published interactions using an ORFeome-based, proteome-wide yeast two-hybrid screening strategy. We identified a total of 1515 protein–protein interactions, of which 1461 are new. The integration of all the interactions reported in H. pylori results in 3004 unique interactions that connect about 70% of its proteome. Excluding interactions of promiscuous proteins we derived from our new data a core network consisting of 908 interactions. We compared our data set to several other bacterial interactomes and experimentally benchmarked the conservation of interactions using 365 protein pairs (interologs) of E. coli of which one third turned out to be conserved in both species.Helicobacter pylori is a Gram-negative, microaerophilic bacterium that colonizes the stomach, an unusual highly acidic niche for microorganisms. In 1983, Warren and Marshall found it to be associated with gastric inflammation and duodenal ulcer disease (1, 2). A chronic infection with H. pylori can lead to development of stomach carcinoma and MALT lymphoma (reviewed in (3)). Hence, the World Health Organization has classified H. pylori as a class I carcinogen (4). It is estimated that half of the world′s population harbors H. pylori but with large variations in the geographical and socioeconomic distribution while causing annually 700,000 deaths worldwide (reviewed in (5)).The pathogenesis of H. pylori has been extensively studied, including the effector CagA, cytotoxin VacA, its adhesins and urease (reviewed in (3, 5–7)). The latter allows the bacterium to neutralize the stomach acid through ammonia production. However, H. pylori is not a classical model organism and thus many gaps in our knowledge still exist.The genome of H. pylori reference strain 26695 was completely sequenced in 1997 (8) and encodes 1587 proteins of which about 950 (61%) have been assigned functions (excluding “putatives”; Uniprot, CMR (9)). These numbers indicate that a large fraction of the proteins of H. pylori has not been functionally characterized.Protein–protein interactions (PPIs)¹ are required for nearly all biological processes. Unbiased interactomes are helpful to understand proteins or pathways and how they are linking poorly or uncharacterized proteins via their interactions. For instance, our study of the Treponema pallidum interactome (10) has led to the characterization of several previously “unknown” proteins such as YbeB, a ribosomal silencing factor (11), or TP0658, a regulator of flagellar translation and assembly (12, 13). However, only a few other comprehensive bacterial interactome studies have been published to date, including Campylobacter jejuni (14), Synechocystis sp. (15), Mycobacterium tuberculosis (16), Mesorhizobium loti (17), and recently Escherichia coli (18). In addition, partial interactomes are available for Bacillus subtilis (19) and H. pylori (20). Most of them used the yeast two-hybrid (Y2H) screening technology (21) which allows the pairwise detection of PPIs. Furthermore, a few other studies (22–25) systematically identified protein complexes and their compositions in bacteria.In 2001, Rain and colleagues have established a partial interactome of H. pylori, the first published protein interaction network of a bacterium (20). In this study, 261 bait constructs were screened against a random prey pool library resulting in the detection of over 1500 PPIs. Although this network likely represents a small fraction of all PPIs that occur in H. pylori, many downstream studies were motivated by these results (see below).Recent studies have disproved the notion that Y2H data sets are of poor quality (26, 27). Similarly, a high false-negative rate can be avoided by multiple Y2H expression vector systems (28–30) or protein fragments as opposed to full-length constructs (31). The aim of this study was to systematically screen the H. pylori proteome for binary protein interactions using a complementary approach to that of Rain et al. to produce an extended protein–protein interaction map of H. pylori. As a result, we have roughly doubled the number of known binary protein–protein interactions for H. pylori in this study.     

Characterization of Protein–Protein Interfaces

We analyze the characteristics of protein–protein interfaces using the largest datasets available from the Protein Data Bank (PDB). We start with a comparison of interfaces with protein cores and non-interface surfaces. The results show that interfaces differ from protein cores and non-interface surfaces in residue composition, sequence entropy, and secondary structure. Since interfaces, protein cores, and non-interface surfaces have different solvent accessibilities, it is important to investigate whether the observed differences are due to the differences in solvent accessibility or differences in functionality. We separate out the effect of solvent accessibility by comparing interfaces with a set of residues having the same solvent accessibility as the interfaces. This strategy reveals residue distribution propensities that are not observable by comparing interfaces with protein cores and non-interface surfaces. Our conclusions are that there are larger numbers of hydrophobic residues, particularly aromatic residues, in interfaces, and the interactions apparently favored in interfaces include the opposite charge pairs and hydrophobic pairs. Surprisingly, Pro-Trp pairs are over represented in interfaces, presumably because of favorable geometries. The analysis is repeated using three datasets having different constraints on sequence similarity and structure quality. Consistent results are obtained across these datasets. We have also investigated separately the characteristics of heteromeric interfaces and homomeric interfaces.     

Computational Assessment of Protein–protein Binding Affinity by Reversely Engineering the Energetics in Protein Complexes

The cellular functions of proteins are maintained by forming diverse complexes. The stability of these complexes is quantified by the measurement of binding affinity, and mutations that alter the binding affinity can cause various diseases such as cancer and diabetes. As a result, accurate estimation of the binding stability and the effects of mutations on changes of binding affinity is a crucial step to understanding the biological functions of proteins and their dysfunctional consequences. It has been hypothesized that the stability of a protein complex is dependent not only on the residues at its binding interface by pairwise interactions but also on all other remaining residues that do not appear at the binding interface. Here, we computationally reconstruct the binding affinity by decomposing it into the contributions of interfacial residues and other non-interfacial residues in a protein complex. We further assume that the contributions of both interfacial and non-interfacial residues to the binding affinity depend on their local structural environments such as solvent-accessible surfaces and secondary structural types. The weights of all corresponding parameters are optimized by Monte-Carlo simulations. After cross-validation against a large-scale dataset, we show that the model not only shows a strong correlation between the absolute values of the experimental and calculated binding affinities, but can also be an effective approach to predict the relative changes of binding affinity from mutations. Moreover, we have found that the optimized weights of many parameters can capture the first-principle chemical and physical features of molecular recognition, therefore reversely engineering the energetics of protein complexes. These results suggest that our method can serve as a useful addition to current computational approaches for predicting binding affinity and understanding the molecular mechanism of protein–protein interactions.     

Computational Prediction of Protein–Protein Interactions

Recently a number of computational approaches have been developed for the prediction of protein–protein interactions. Complete genome sequencing projects have provided the vast amount of information needed for these analyses. These methods utilize the structural, genomic, and biological context of proteins and genes in complete genomes to predict protein interaction networks and functional linkages between proteins. Given that experimental techniques remain expensive, time-consuming, and labor-intensive, these methods represent an important advance in proteomics. Some of these approaches utilize sequence data alone to predict interactions, while others combine multiple computational and experimental datasets to accurately build protein interaction maps for complete genomes. These methods represent a complementary approach to current high-throughput projects whose aim is to delineate protein interaction maps in complete genomes. We will describe a number of computational protocols for protein interaction prediction based on the structural, genomic, and biological context of proteins in complete genomes, and detail methods for protein interaction network visualization and analysis.     

Structure–Critical Distribution of Aromatic Residues in the Fibronectin Type III Protein Family

Over a thousand individual Fibronectin type III (FnIII) domain sequences, extracted from more than 60 different FnIII-dependent protein super-structures, were downloaded from curated database resources. Three regions of extreme sequence conservation within the well-characterized FnIII β-sandwich structure were respectively defined by near absolute conservation of a tryptophan (Trp) in β-strand-B, tyrosines (Tyr) in both β-strand-C and β-strand-F, and a leucine (Leu) residue in the unstructured region immediately preceding β-strand-F. Employing these four conserved landmarks, the entire FnIII sequence dataset was vertically registered to align the three conserved regions, and the cumulative distribution of all other amino acid functionality was determined and plotted relative to these landmark residues. Conserved aromatic sites were each found to be flanked by aliphatic residues that assure localization of these sites to the inaccessible hydrophobic interface between major sheet structures. Mapping the location of conserved aromatic sites in numerous PDB structures demonstrated the consistent pair-wise co-localization of the indole side-chain of the conserved strand-B Trp site to within 0.35 nm of the phenolic side-chain of the strand-C Tyr site located 8–14 amino acids distal. Likewise, the side-chain of the strand-F Tyr site co-localized to within 0.45 nm of the aliphatic side-chain of the conserved Leu that uniformly precedes it by six residues. While classic hydropathy-based theories would deem the “burying” of Tyr and Trp side-chains and/or their association with hydrophobic FnIII core residues thermodynamically unnecessary, alternative contributions of conserved Trp and Tyr residues, and particularly the role of the absolutely conserved tyrosine phenolic –OH in native FnIII structure–function are considered. A more global role for conserved FnIII aromaticity is also discussed in light of the aromatic conservation observed in other well-established protein families.     

A Latent Eigenprobit Model with Link Uncertainty for Prediction of Protein–Protein Interactions

Protein–protein interactions (PPI’s) are of fundamental importance in biology and biomedicine. Identifying and characterizing protein interactions based on various genomic and proteomic data has become a canonical problem in computational biology. Approaching this task as a binary classification problem, we propose a hierarchical Bayesian probit-based framework, incorporating multiple sources of relational protein data as covariates, for modeling binary network topology. More importantly, this model has two distinctive features: (1) capturing the latent characteristics of nodes in the network by an eigenmodel, and (2) accounting for and correcting the link uncertainty in the training data, a well-known critical issue with protein interactions generated by high-throughput technology. We evaluate and compare the predictive performance of the proposed model with three submodels without one or both of these features. Results from two yeast functional subnetworks have demonstrated that both the latent eigenmodel and accounting for link uncertainty are important for better predictions, and the latter can yield substantial improvement in predictive precision.     

A Comprehensive Benchmark of Kernel Methods to Extract Protein–Protein Interactions from Literature

The most important way of conveying new findings in biomedical research is scientific publication. Extraction of protein–protein interactions (PPIs) reported in scientific publications is one of the core topics of text mining in the life sciences. Recently, a new class of such methods has been proposed - convolution kernels that identify PPIs using deep parses of sentences. However, comparing published results of different PPI extraction methods is impossible due to the use of different evaluation corpora, different evaluation metrics, different tuning procedures, etc. In this paper, we study whether the reported performance metrics are robust across different corpora and learning settings and whether the use of deep parsing actually leads to an increase in extraction quality. Our ultimate goal is to identify the one method that performs best in real-life scenarios, where information extraction is performed on unseen text and not on specifically prepared evaluation data. We performed a comprehensive benchmarking of nine different methods for PPI extraction that use convolution kernels on rich linguistic information. Methods were evaluated on five different public corpora using cross-validation, cross-learning, and cross-corpus evaluation. Our study confirms that kernels using dependency trees generally outperform kernels based on syntax trees. However, our study also shows that only the best kernel methods can compete with a simple rule-based approach when the evaluation prevents information leakage between training and test corpora. Our results further reveal that the F-score of many approaches drops significantly if no corpus-specific parameter optimization is applied and that methods reaching a good AUC score often perform much worse in terms of F-score. We conclude that for most kernels no sensible estimation of PPI extraction performance on new text is possible, given the current heterogeneity in evaluation data. Nevertheless, our study shows that three kernels are clearly superior to the other methods.     

7 The Molecular Biology of Tryptophan Synthase: A Model for Protein–Protein Interaction

Abstract

The use of plastic produced from non-renewable resources constitutes a major environmental problem of the modern society. Polylactide polymers (PLA) have recently gained enormous attention as one possible substitution of petroleum derived polymers. A prerequisite for high quality PLA production is the provision of optically pure lactic acid, which cannot be obtained by chemical synthesis in an economical way. Microbial fermentation is therefore the commercial option to obtain lactic acid as monomer for PLA production. However, one major economic hurdle for commercial lactic acid production as basis for PLA is the costly separation procedure, which is needed to recover and purify the product from the fermentation broth. Yeasts, such as Saccharomyces cerevisiae (bakers yeast) offer themselves as production organisms because they can tolerate low pH and grow on mineral media what eases the purification of the acid. However, naturally yeasts do not produce lactic acid. By metabolic engineering, ethanol was exchanged with lactic acid as end product of fermentation. A vast amount of effort has been invested into the development of yeasts for lactic acid production since the first paper on this topic by Dequin and process insight. If pH stress is used as basis for DNA microarray analyses, in order to improve the host, what exactly is addressed? Growth? Or productivity? They might be connected, but can be negatively correlated. A better growing strain might not be a better producer. So if the question was growth, the answer might not be what was initially intended (productivity).

A major task for the future is to learn to ask the right questions – a lot of studies intended to lead to better productivity, did lead to interesting results, but NOT to better production strains.

Taking together what we learned from lactic acid production with yeasts, we see a bright future for bulk and fine chemical production with these versatile hosts.     

Combining NMR Relaxation with Chemical Shift Perturbation Data to Drive Protein–protein Docking

The modeling of biomolecular complexes by computational docking using the known structures of their constituents is developing rapidly to become a powerful tool in structural biology. It is especially useful in combination with even limited experimental information describing the interface. Here we demonstrate for the first time the use of diffusion anisotropy in combination with chemical shift perturbation data to drive protein–protein docking. For validation purposes we make use of simulated diffusion anisotropy data. Inclusion of this information, which can be derived from NMR relaxation rates and reports on the orientation of the components of a complex with respect to the rotational diffusion tensor, substantially improves the docking results.     

A Crystal Structure of the Cyclic GMP-dependent Protein Kinase Iβ Dimerization/Docking Domain Reveals Molecular Details of Isoform-specific Anchoring

Cyclic GMP-dependent protein kinase (PKG) is a key mediator of the nitric oxide/cGMP signaling pathway and plays a central role in regulating cardiovascular and neuronal functions. The N-terminal ∼50 amino acids of the kinase are required for homodimerization and association with isoform-specific PKG-anchoring proteins (GKAPs), which target the kinase to specific substrates. To understand the molecular details of PKG dimerization and gain insight into its association with GKAPs, we solved a crystal structure of the PKG Iβ dimerization/docking domain. Our structure provides molecular details of this unique leucine/isoleucine zipper, revealing specific hydrophobic and ionic interactions that mediate dimerization and demonstrating the topology of the GKAP interaction surface.     

Environmental Impact of the Production of Mealworms as a Protein Source for Humans – A Life Cycle Assessment

The demand for animal protein is expected to rise by 70–80% between 2012 and 2050, while the current animal production sector already causes major environmental degradation. Edible insects are suggested as a more sustainable source of animal protein. However, few experimental data regarding environmental impact of insect production are available. Therefore, a lifecycle assessment for mealworm production was conducted, in which greenhouse gas production, energy use and land use were quantified and compared to conventional sources of animal protein. Production of one kg of edible protein from milk, chicken, pork or beef result in higher greenhouse gas emissions, require similar amounts of energy and require much more land. This study demonstrates that mealworms should be considered a more sustainable source of edible protein.     

A New Amylose Derivative for the Preparation of Protein–Carbohydrate Conjugates

N,N-Dimethylformamidase (DMFase) from Alcaligenes sp. strain KUFA-1, a bacterium that can grow on N,N-dimethylformamide (DMF) as the sole carbon and nitrogen source, catalyzes the first step of the DMF degradation. The DMFase gene dmfA1A2 was cloned in Escherichia coli, and its nucleotides were sequenced. The deduced amino acid sequence of the enzyme consisted of two α- and two β-subunits with 132 and 762 amino acids, respectively, and had little similarity to sequences in protein databases, including various amidases. The protein may be a new kind of amidase. DMFase activity was detected in E. coli cells transformed with an expression plasmid of the cloned DMFase gene. The properties of recombinant DMFase purified from E. coli were identical to those of Alcaligenes DMFase.     

Selection of Protein–Protein Interactions of Desired Affinities with a Bandpass Circuit

We have developed a genetic circuit in Escherichia coli that can be used to select for protein–protein interactions of different strengths by changing antibiotic concentrations in the media. The genetic circuit links protein–protein interaction strength to β-lactamase activity while simultaneously imposing tuneable positive and negative selection pressure for β-lactamase activity. Cells only survive if they express interacting proteins with affinities that fall within set high- and low-pass thresholds; i.e. the circuit therefore acts as a bandpass filter for protein–protein interactions. We show that the circuit can be used to recover protein–protein interactions of desired affinity from a mixed population with a range of affinities. The circuit can also be used to select for inhibitors of protein–protein interactions of defined strength.