Search

The reliable prediction of the affinity of candidate peptides for the MHC is important for predicting their potential antigenicity and thus influences medical applications, such as decisions on their inclusion in T cell–based vaccines. In this study, we present a rapid, predictive computational approach that combines a popular, sequence-based artificial neural network method, NetMHCpan 4.0, with three-dimensional structural modeling. We find that the ensembles of bound peptide conformations generated by the programs MODELLER and Rosetta FlexPepDock are less variable in geometry for strong binders than for low-affinity peptides. In tests on 1271 peptide sequences for which the experimental dissociation constants of binding to the well-characterized murine MHC allele H-2Db are known, by applying thresholds for geometric fluctuations the structure-based approach in a standalone manner drastically improves the statistical specificity, reducing the number of false positives. Furthermore, filtering candidates generated with NetMHCpan 4.0 with the structure-based predictor led to an increase in the positive predictive value (PPV) of the peptides correctly predicted to bind very strongly (i.e., Kd < 100 nM) from 40 to 52% (p = 0.027). The combined method also significantly improved the PPV when tested on five human alleles, including some with limited data for training. Overall, an average increase of 10% in the PPV was found over the standalone sequence-based method. The combined method should be useful in the rapid design of effective T cell–based vaccines.

Many proteins have small molecule-binding pockets that are not easily detectable in the ligand-free structures. These cryptic sites require a conformational change to become apparent; a cryptic site can therefore be defined as a site that forms a pocket in a holo structure, but not in the apo structure. Because many proteins appear to lack druggable pockets, understanding and accurately identifying cryptic sites could expand the set of drug targets. Previously, cryptic sites were identified experimentally by fragment-based ligand discovery, and computationally by long molecular dynamics simulations and fragment docking. Here, we begin by constructing a set of structurally defined apo-holo pairs with cryptic sites. Next, we comprehensively characterize the cryptic sites in terms of their sequence, structure, and dynamics attributes. We find that cryptic sites tend to be as conserved in evolution as traditional binding pockets, but are less hydrophobic and more flexible. Relying on this characterization, we use machine learning to predict cryptic sites with relatively high accuracy (for our benchmark, the true positive and false positive rates are 73% and 29%, respectively). We then predict cryptic sites in the entire structurally characterized human proteome (11,201 structures, covering 23% of all residues in the proteome). CryptoSite increases the size of the potentially “druggable” human proteome from ~40% to ~78% of disease-associated proteins. Finally, to demonstrate the utility of our approach in practice, we experimentally validate a cryptic site in protein tyrosine phosphatase 1B using a covalent ligand and NMR spectroscopy. The CryptoSite web server is available at http://salilab.org/cryptosite.

Selecting peptides that bind strongly to the major histocompatibility complex (MHC) for inclusion in a vaccine has therapeutic potential for infections and tumors. Machine learning models trained on sequence data exist for peptide:MHC (p:MHC) binding predictions. Here, we train support vector machine classifier (SVMC) models on physicochemical sequence-based and structure-based descriptor sets to predict peptide binding to a well-studied model mouse MHC I allele, H-2Db. Recursive feature elimination and two-way forward feature selection were also performed. Although low on sensitivity compared to the current state-of-the-art algorithms, models based on physicochemical descriptor sets achieve specificity and precision comparable to the most popular sequence-based algorithms. The best-performing model is a hybrid descriptor set containing both sequence-based and structure-based descriptors. Interestingly, close to half of the physicochemical sequence-based descriptors remaining in the hybrid model were properties of the anchor positions, residues 5 and 9 in the peptide sequence. In contrast, residues flanking position 5 make little to no residue-specific contribution to the binding affinity prediction. The results suggest that machine-learned models incorporating both sequence-based descriptors and structural data may provide information on specific physicochemical properties determining binding affinities.

Proteins are essential elements of biological systems, and their function typically relies on their ability to successfully bind to specific partners. Recently, an emphasis of study into protein interactions has been on hot spots, or residues in the binding interface that make a significant contribution to the binding energetics. In this study, we investigate how conservation of hot spots can be used to guide docking prediction. We show that the use of evolutionary data combined with hot spot prediction highlights near‐native structures across a range of benchmark examples. Our approach explores various strategies for using hot spots and evolutionary data to score protein complexes, using both absolute and chemical definitions of conservation along with refinements to these strategies that look at windowed conservation and filtering to ensure a minimum number of hot spots in each binding partner. Finally, structure‐based models of orthologs were generated for comparison with sequence‐based scoring. Using two data sets of 22 and 85 examples, a high rate of top 10 and top 1 predictions are observed, with up to 82% of examples returning a top 10 hit and 35% returning top 1 hit depending on the data set and strategy applied; upon inclusion of the native structure among the decoys, up to 55% of examples yielded a top 1 hit. The 20 common examples between data sets show that more carefully curated interolog data yields better predictions, particularly in achieving top 1 hits. Proteins 2015; 83:1940–1946. © 2015 The Authors. Proteins: Structure, Function, and Bioinformatics Published by Wiley Periodicals, Inc.

We describe methods and results for four new types of challenge in the Critical Assessment of PRedicted Interactions (CAPRI). Two new challenges asked predictors to create models related to protein interface design. The first of these was to distinguish binding interfaces from designed nonbinding interfaces. The second was to predict the effects of all single-point mutations on hemagglutinin binding to two small designed proteins. Two additional challenges asked predictors to submit high-resolution structures for interface-bound crystallographic waters and for binding heparin to a putative glycosylase.

Community-wide blind prediction experiments such as CAPRI and CASP provide an objective measure of the current state of predictive methodology. Here we describe a community-wide assessment of methods to predict the effects of mutations on protein-protein interactions. Twenty-two groups predicted the effects of comprehensive saturation mutagenesis for two designed influenza hemagglutinin binders and the results were compared with experimental yeast display enrichment data obtained using deep sequencing. The most successful methods explicitly considered the effects of mutation on monomer stability in addition to binding affinity, carried out explicit side-chain sampling and backbone relaxation, evaluated packing, electrostatic, and solvation effects, and correctly identified around a third of the beneficial mutations. Much room for improvement remains for even the best techniques, and large-scale fitness landscapes should continue to provide an excellent test bed for continued evaluation of both existing and new prediction methodologies.

Lactococcus lactis metabolizes glucose homofermentatively to lactate. However, after disruption of the gene coding for lactate dehydrogenase, LDH, a key enzyme in NAD^+ regeneration, the glycolytic flux shifts from homolactic to mixed-acid fermentation with the redirection of pyruvate towards production of formate, acetate, ethanol and CO"2. A mathematical model of the pyruvate metabolism pathway that enhances ethanol production was developed from in vivo Nuclear Magnetic Resonance (NMR) time-series measurements that describe the dynamics of the metabolites in L. lactis. Both Michaelis-Menten and S-system models capture the observed in vivo dynamics of the glycolysis pathway in L. lactis, while prior models describe only the in vitro dynamics. The models provide insight into the maximization of selectivity of ethanol with respect to acetate and CO"2 as undesired products in multiple reactions. High concentrations of NADH and acetyl-CoA and low concentrations of pyruvate and NAD appear to maximize ethanol selectivity.

chipD is a web server that facilitates design of DNA oligonucleotide probes for high-density tiling arrays, which can be used in a number of genomic applications such as ChIP-chip or gene-expression profiling. The server implements a probe selection algorithm that takes as an input, in addition to the target sequences, a set of parameters that allow probe design to be tailored to specific applications, protocols or the array manufacturer’s requirements. The algorithm optimizes probes to meet three objectives: (i) probes should be specific; (ii) probes should have similar thermodynamic properties; and (iii) the target sequence coverage should be homogeneous and avoid significant gaps. The output provides in a text format, the list of probe sequences with their genomic locations, targeted strands and hybridization characteristics. chipD has been used successfully to design tiling arrays for bacteria and yeast. chipD is available at http://chipd.uwbacter.org/.

The KFC Server is a web-based implementation of the KFC (Knowledge-based FADE and Contacts) model—a machine learning approach for the prediction of binding hot spots, or the subset of residues that account for most of a protein interface's; binding free energy. The server facilitates the automated analysis of a user submitted protein–protein or protein–DNA interface and the visualization of its hot spot predictions. For each residue in the interface, the KFC Server characterizes its local structural environment, compares that environment to the environments of experimentally determined hot spots and predicts if the interface residue is a hot spot. After the computational analysis, the user can visualize the results using an interactive job viewer able to quickly highlight predicted hot spots and surrounding structural features within the protein structure. The KFC Server is accessible at http://kfc.mitchell-lab.org.

Synaptotagmins contain tandem C2 domains and function as Ca2+ sensors for vesicle exocytosis but the mechanism for coupling Ca2+ rises to membrane fusion remains undefined. Synaptotagmins bind SNAREs, essential components of the membrane fusion machinery, but the role of these interactions in Ca2+-triggered vesicle exocytosis has not been directly assessed. We identified sites on synaptotagmin−1 that mediate Ca2+-dependent SNAP25 binding by zero-length cross-linking. Mutation of these sites in C2A and C2B eliminated Ca2+-dependent synaptotagmin−1 binding to SNAREs without affecting Ca2+-dependent membrane binding. The mutants failed to confer Ca2+ regulation on SNARE-dependent liposome fusion and failed to restore Ca2+-triggered vesicle exocytosis in synaptotagmin-deficient PC12 cells. The results provide direct evidence that Ca2+-dependent SNARE binding by synaptotagmin is essential for Ca2+-triggered vesicle exocytosis and that Ca2+-dependent membrane binding by itself is insufficient to trigger fusion. A structure-based model of the SNARE-binding surface of C2A provided a new view of how Ca2+-dependent SNARE and membrane binding occur simultaneously.

Motivation: Metal reduction kinetics have been studied in cultures of dissimilatory metal reducing bacteria which include the Shewanella oneidensis strain MR-1. Estimation of system parameters from time-series data faces obstructions in the implementation depending on the choice of the mathematical model that captures the observed dynamics. The modeling of metal reduction is often based on Michaelis–Menten equations. These models are often developed using initial in vitro reaction rates and seldom match with in vivo reduction profiles.Results: For metal reduction studies, we propose a model that is based on the power law representation that is effectively applied to the kinetics of metal reduction. The method yields reasonable parameter estimates and is illustrated with the analysis of time-series data that describes the dynamics of metal reduction in S.oneidensis strain MR-1. In addition, mixed metal studies involving the reduction of Uranyl (U(VI)) to the relatively insoluble tetravalent form (U(IV)) by S.alga strain (BR-Y) were studied in the presence of environmentally relevant iron hydrous oxides. For mixed metals, parameter estimation and curve fitting are accomplished with a generalized least squares formulation that handles systems of ordinary differential equations and is implemented in Matlab. It consists of an optimization algorithm (Levenberg–Marquardt, LSQCURVEFIT) and a numerical ODE solver. Simulation with the estimated parameters indicates that the model captures the experimental data quite well. The model uses the estimated parameters to predict the reduction rates of metals and mixed metals at varying concentrations.Contact: rl8q@cms.mail.virginia.eduSupplementary information: Supplementary data are available at Bioinformatics online.

Protein–protein interactions can be altered by mutating one or more “hot spots,” the subset of residues that account for most of the interface's binding free energy. The identification of hot spots requires a significant experimental effort, highlighting the practical value of hot spot predictions. We present two knowledge-based models that improve the ability to predict hot spots: K-FADE uses shape specificity features calculated by the Fast Atomic Density Evaluation (FADE) program, and K-CON uses biochemical contact features. The combined K-FADE/CON (KFC) model displays better overall predictive accuracy than computational alanine scanning (Robetta–Ala). In addition, because these methods predict different subsets of known hot spots, a large and significant increase in accuracy is achieved by combining KFC and Robetta–Ala. The KFC analysis is applied to the calmodulin (CaM)/smooth muscle myosin light chain kinase (smMLCK) interface, and to the bone morphogenetic protein-2 (BMP-2)/BMP receptor-type I (BMPR-IA) interface. The results indicate a strong correlation between KFC hot spot predictions and mutations that significantly reduce the binding affinity of the interface. Proteins 2007. © 2007 Wiley-Liss, Inc.

Coenzyme Q (CoQ) is an isoprenylated quinone that is essential for cellular respiration and is synthesized in mitochondria by the combined action of at least nine proteins (COQ1-9). Although most COQ proteins are known to catalyze modifications to CoQ precursors, the biochemical role of COQ9 remains unclear. Here, we report that a disease-related COQ9 mutation leads to extensive disruption of the CoQ protein biosynthetic complex in a mouse model, and that COQ9 specifically interacts with COQ7 through a series of conserved residues. Toward understanding how COQ9 can perform these functions, we solved the crystal structure of Homo sapiens COQ9 at 2.4 Å. Unexpectedly, our structure reveals that COQ9 has structural homology to the TFR family of bacterial transcriptional regulators, but that it adopts an atypical TFR dimer orientation and is not predicted to bind DNA. Our structure also reveals a lipid-binding site, and mass spectrometry-based analyses of purified COQ9 demonstrate that it associates with multiple lipid species, including CoQ itself. The conserved COQ9 residues necessary for its interaction with COQ7 comprise a surface patch around the lipid-binding site, suggesting that COQ9 might serve to present its bound lipid to COQ7. Collectively, our data define COQ9 as the first, to our knowledge, mammalian TFR structural homolog and suggest that its lipid-binding capacity and association with COQ7 are key features for enabling CoQ biosynthesis.

Onconase (ONC), an amphibian member of the bovine pancreatic ribonuclease A (RNase A) superfamily, is in phase III clinical trials as a treatment for malignant mesothelioma. RNase A is a far more efficient catalyst of RNA cleavage than ONC but is not cytotoxic. The innate ability of ONC to evade the cytosolic ribonuclease inhibitor protein (RI) is likely to be a primary reason for its cytotoxicity. In contrast, the non-covalent interaction between RNase A and RI is one of the strongest known, with the RI.RNase A complex having a K(d) value in the femtomolar range. Here, we report on the use of the fast atomic density evaluation (FADE) algorithm to identify regions in the molecular interface of the RI.RNase A complex that exhibit a high degree of geometric complementarity. Guided by these "knobs" and "holes", we designed variants of RNase A that evade RI. The D38R/R39D/N67R/G88R substitution increased the K(d) value of the pRI.RNase A complex by 20 x 10(6)-fold (to 1.4 microM) with little change to catalytic activity or conformational stability. This and two related variants of RNase A were more toxic to human cancer cells than was ONC. Notably, these cytotoxic variants exerted their toxic activity on cancer cells selectively, and more selectively than did ONC. Substitutions that further diminish affinity for RI (which has a cytosolic concentration of 4 microM) are unlikely to produce a substantial increase in cytotoxic activity. These results demonstrate the utility of the FADE algorithm in the examination of protein-protein interfaces and represent a landmark towards the goal of developing chemotherapeutics based on mammalian ribonucleases.

Cryo-EM and molecular dynamics simulations reveal unexpected flexibility in individual monomers and a stable interface between monomers in the spiral filaments formed by the ESCRT-III subunit Vps32/CHMP4B., The scission of biological membranes is facilitated by a variety of protein complexes that bind and manipulate lipid bilayers. ESCRT-III (endosomal sorting complex required for transport III) filaments mediate membrane scission during the ostensibly disparate processes of multivesicular endosome biogenesis, cytokinesis, and retroviral budding. However, mechanisms by which ESCRT-III subunits assemble into a polymer remain unknown. Using cryogenic electron microscopy (cryo-EM), we found that the full-length ESCRT-III subunit Vps32/CHMP4B spontaneously forms single-stranded spiral filaments. The resolution afforded by two-dimensional cryo-EM combined with molecular dynamics simulations revealed that individual Vps32/CHMP4B monomers within a filament are flexible and able to accommodate a range of bending angles. In contrast, the interface between monomers is stable and refractory to changes in conformation. We additionally found that the carboxyl terminus of Vps32/CHMP4B plays a key role in restricting the lateral association of filaments. Our findings highlight new mechanisms by which ESCRT-III filaments assemble to generate a unique polymer capable of membrane remodeling in multiple cellular contexts.

Snowflake Vitreoretinal Degeneration (SVD) is associated with the R162W mutation of the Kir7.1 inwardly-rectifying potassium channel. Kir7.1 is found at the apical membrane of Retinal Pigment Epithelial (RPE) cells, adjacent to the photoreceptor neurons. The SVD phenotype ranges from RPE degeneration to an abnormal b-wave to a liquid vitreous. We sought to determine how this mutation alters the structure and function of the human Kir7.1 channel. In this study, we expressed a Kir7.1 construct with the R162W mutation in CHO cells to evaluate function of the ion channel. Compared to the wild-type protein, the mutant protein exhibited a non-functional Kir channel that resulted in depolarization of the resting membrane potential. Upon co-expression with wild-type Kir7.1, R162W mutant showed a reduction of IKir7.1 and positive shift in '0' current potential. Homology modeling based on the structure of a bacterial Kir channel protein suggested that the effect of R162W mutation is a result of loss of hydrogen bonding by the regulatory lipid binding domain of the cytoplasmic structure.

In this study, we present the DNA-Binding Site Identifier (DBSI), a new structure-based method for predicting protein interaction sites for DNA binding. DBSI was trained and validated on a data set of 263 proteins (TRAIN-263), tested on an independent set of protein-DNA complexes (TEST-206) and data sets of 29 unbound (APO-29) and 30 bound (HOLO-30) protein structures distinct from the training data. We computed 480 candidate features for identifying protein residues that bind DNA, including new features that capture the electrostatic microenvironment within shells near the protein surface. Our iterative feature selection process identified features important in other models, as well as features unique to the DBSI model, such as a banded electrostatic feature with spatial separation comparable with the canonical width of the DNA minor groove. Validations and comparisons with established methods using a range of performance metrics clearly demonstrate the predictive advantage of DBSI, and its comparable performance on unbound (APO-29) and bound (HOLO-30) conformations demonstrates robustness to binding-induced protein conformational changes. Finally, we offer our feature data table to others for integration into their own models or for testing improved feature selection and model training strategies based on DBSI.