Accuracy of functional surfaces on comparatively modeled protein structures

Identification and characterization of protein functional surfaces are important for predicting protein function, understanding enzyme mechanism, and docking small compounds to proteins. As the rapid speed of accumulation of protein sequence information far exceeds that of structures, constructing accurate models of protein functional surfaces and identify their key elements become increasingly important. A promising approach is to build comparative models from sequences using known structural templates such as those obtained from structural genome projects. Here we assess how well this approach works in modeling binding surfaces. By systematically building three-dimensional comparative models of proteins using Modeller, we determine how well functional surfaces can be accurately reproduced. We use an alpha shape based pocket algorithm to compute all pockets on the modeled structures, and conduct a large-scale computation of similarity measurements (pocket RMSD and fraction of functional atoms captured) for 26,590 modeled enzyme protein structures. Overall, we find that when the sequence fragment of the binding surfaces has more than 45% identity to that of the template protein, the modeled surfaces have on average an RMSD of 0.5 Å, and contain 48% or more of the binding surface atoms, with nearly all of the important atoms in the signatures of binding pockets captured.     

Predicting Protein Function and Binding Profile via Matching of Local Evolutionary and Geometric Surface Patterns

Inferring protein functions from structures is a challenging task, as a large number of orphan protein structures from structural genomics project are now solved without their biochemical functions characterized. For proteins binding to similar substrates or ligands and carrying out similar functions, their binding surfaces are under similar physicochemical constraints, and hence the sets of allowed and forbidden residue substitutions are similar. However, it is difficult to isolate such selection pressure due to protein function from selection pressure due to protein folding, and evolutionary relationship reflected by global sequence and structure similarities between proteins is often unreliable for inferring protein function. We have developed a method, called pevoSOAR (pocket-based evolutionary search of amino acid residues), for predicting protein functions by solving the problem of uncovering amino acids residue substitution pattern due to protein function and separating it from amino acids substitution pattern due to protein folding. We incorporate evolutionary information specific to an individual binding region and match local surfaces on a large scale with millions of precomputed protein surfaces to identify those with similar functions. Our pevoSOAR method also generates a probablistic model called the computed binding a profile that characterizes protein-binding activities that may involve multiple substrates or ligands. We show that our method can be used to predict enzyme functions with accuracy. Our method can also assess enzyme binding specificity and promiscuity. In an objective large-scale test of 100 enzyme families with thousands of structures, our predictions are found to be sensitive and specific: At the stringent specificity level of 99.98%, we can correctly predict enzyme functions for 80.55% of the proteins. The overall area under the receiver operating characteristic curve measuring the performance of our prediction is 0.955, close to the perfect value of 1.00. The best Matthews coefficient is 86.6%. Our method also works well in predicting the biochemical functions of orphan proteins from structural genomics projects.     

Insights from modeling the 3D structure of NAD(P)H-dependent D-xylose reductase of Pichia stipitis and its binding interactions with NAD and NADP

NAD(P)H-dependent d-xylose reductase is a homodimeric oxidoreductase that belongs to the aldo-keto reductase superfamily. The enzyme has the special function to catalyze the first step in the assimilation of xylose into yeast metabolic pathways. Performing this function via reducing the open chain xylose to xylitol, the xylose reductase of Pichia stipitis is one of the most important enzymes that can be used to construct recombinant Saccharomyces cerevisiae strain for utilizing xylose and producing alcohol. To investigate into the interaction mechanism of the enzyme with its ligand NAD and NADP, the 3D structure was developed for the NAD(P)H-dependent d-xylose reductase from P. stipitis. With the 3D structure, the molecular docking operations were conducted to find the most stable bindings of the enzyme with NAD and NADP, respectively. Based on these results, the binding pockets of the enzyme for NAD and NADP have been explicitly defined. It has been found that the residues in forming the binding pockets for both NAD and NADP are almost the same and mainly hydrophilic. These findings may be used to guide mutagenesis studies, providing useful clues to modify the enzyme to improve the utilization of xylose for producing alcohol. Also, because human aldose reductases have the function to reduce the open chain form of glucose to sorbitol, a process physiologically significant for diabetic patients at the time that their blood glucose levels are elevated, the information gained through this study may also stimulate the development of new strategies for therapeutic treatment of diabetes.     

Identification of protein functional surfaces by the concept of a split pocket

The function of a protein is often fulfilled via molecular interactions on its surfaces, so identifying the functional surface(s) of a protein is helpful for understanding its function. Here, we introduce the concept of a split pocket, which is a pocket that is split by a cognate ligand. We use a geometric approach that is site‐specific. Specifically, we first compute a set of all pockets in the protein with its ligand(s) and a set of all pockets with the ligand(s) removed and then compare the two sets of pockets to identify the split pocket(s) of the protein. To reduce the search space and expedite the process of surface partitioning, we design probe radii according to the physicochemical textures of molecules. Our method achieves a success rate of 96% on a benchmark test set. We conduct a large‐scale computation to identify ～19,000 split pockets from 11,328 structures (1.16 million potential pockets); for each pocket, we obtain residue composition, solvent‐accessible area, and molecular volume. With this database of split pockets, our method can be used to predict the functional surfaces of unbound structures. Indeed, the functional surface of an unbound protein may often be found from its similarity to remotely related bound forms that belong to distinct folds. Finally, we apply our method to identify glucose‐binding proteins, including unbound structures. Our study demonstrates the power of geometric and evolutionary matching for studying protein functional evolution and provides a framework for classifying protein functions by local spatial patterns of functional surfaces. Proteins 2009. © 2009 Wiley‐Liss, Inc.     

Computational approaches for identification of conserved/unique binding pockets in the A chain of ricin

MOTIVATION: Specific and sensitive ligand-based protein detection assays that employ antibodies or small molecules such as peptides, aptamers or other small molecules require that the corresponding surface region of the protein be accessible and that there be minimal cross-reactivity with non-target proteins. To reduce the time and cost of laboratory screening efforts for diagnostic reagents, we developed new methods for evaluating and selecting protein surface regions for ligand targeting. RESULTS: We devised combined structure- and sequence-based methods for identifying 3D epitopes and binding pockets on the surface of the A chain of ricin that are conserved with respect to a set of ricin A chains and unique with respect to other proteins. We (1) used structure alignment software to detect structural deviations and extracted from this analysis the residue-residue correspondence, (2) devised a method to compare corresponding residues across sets of ricin structures and structures of closely related proteins, (3) devised a sequence-based approach to determine residue infrequency in local sequence context and (4) modified a pocket-finding algorithm to identify surface crevices in close proximity to residues determined to be conserved/unique based on our structure- and sequence-based methods. In applying this combined informatics approach to ricin A, we identified a conserved/unique pocket in close proximity (but not overlapping) the active site that is suitable for bi-dentate ligand development. These methods are generally applicable to identification of surface epitopes and binding pockets for development of diagnostic reagents, therapeutics and vaccines.     

Shape variation in protein binding pockets and their ligands

A common assumption about the shape of protein binding pockets is that they are related to the shape of the small ligand molecules that can bind there. But to what extent is that assumption true? Here we use a recently developed shape matching method to compare the shapes of protein binding pockets to the shapes of their ligands. We find that pockets binding the same ligand show greater variation in their shapes than can be accounted for by the conformational variability of the ligand. This suggests that geometrical complementarity in general is not sufficient to drive molecular recognition. Nevertheless, we show when considering only shape and size that a significant proportion of the recognition power of a binding pocket for its ligand resides in its shape. Additionally, we observe a "buffer zone" or a region of free space between the ligand and protein, which results in binding pockets being on average three times larger than the ligand that they bind.     

Computer modelling of the NAD binding site of ADP-ribosylating toxins: active-site structure and mechanism of NAD binding

Five ADP-ribosylating bacterial toxins, pertussis toxin, cholera toxin, diphtheria toxin, Escherichia LT toxin and Pseudomonas exotoxin A, show significant homology in selected segments of their sequence. Site-directed mutagenesis and chemical modification of residues within these regions cause loss of catalytic activity and of NAD binding. On the basis of these results and of molecular modelling based on the three-dimensional structure of exotoxin A, the geometry of an NAD binding site common to all the toxins is deduced and described in the paper. For diphtheria toxin, sequence similarity with exotoxin A is such that its preliminary structure can be computed by molecular modelling, whereas for the other toxins similarity appears to be restricted to the NAD binding site. Moreover, an analysis of molecular fitting of the NAD molecule into its binding cavity suggests a new model for the conformation of the bound NAD that better accounts for all available experimental information.     

NADP-Dependent enzymes. II: Evolution of the mono- and dinucleotide binding domains

Nicotinamide adenine dinucleotides [NAD and NADP with both referred to as NAD(P)] are among the more diffuse redox cofactors. Despite their stereochemical similarity where the only difference is a phosphomonoester on the ribose near the adenine of NADP, they show different biochemical reactivities with NAD behaving as an oxidant and NADP as a reductant. NAD(P)-dependent enzymes generally share a common open α/β fold with few exceptions only recently structurally characterized. This study of the molecular evolution of the NAD(P) binding domains, possible given the large number of known molecular structures, addresses two main questions: 1) can a common fold exist in different biological systems (divergent evolution) and 2) does a relationship exist among similar biological systems that display different folds (convergent evolution)? Both the structures of mono- and dinucleotide binding domains have been classified by cluster analysis based on the similarity evaluated by their main chain Cα superposition. Moreover, the cofactor conformations and the stereochemical characteristics of their pockets have also been classified by analogous methods on the basis of the published tertiary structures. Two primary results appear: 1) the classification of the mononucleotide binding domains is different from that of the dinucleotide binding folds and 2) both divergent and convergent evolutionary pathways can be hypothesized, the latter less frequently observed and less pronounced but nevertheless evident. The generally accepted hypothesis that dinucleotide binding domains have evolved by gene duplication of primordial genes coding for the smaller mononucleotide binding domains is acceptable but the two halves of the resulting dinucleotide binding domains are evolutionarly uncorrelated. The NH₂-terminal mononucleotide binding domain is less variable than the COOH-terminal half, probably because it involves the binding of the ADP moiety of NAD(P) invariant in all examined systems. There is evidence to postulate that evolutionary pathways for NAD(P)-dependent enzymes are both divergent and convergent. In fact, nearly all combinations of similarity/dissimilarity in overall fold, cofactor conformation, and cofactor binding pocket structural characteristics for each enzyme pair examined are possible. The NAD(P)-dependent enzymes apparently provide a canonical example of an evolutionary principle that “anything goes.” © 1997 Wiley-Liss Inc.     

Local conformational changes induced by successive nicotinamide adenine dinucleotide binding to dissociable tetrameric D-glyceraldehyde-3-phosphate dehydrogenase. Quantitative analysis of a two-step dissociation process

Covalent binding of FITC up to 2 mol/mol of tetrameric enzyme does not affect the enzymatic activity and dissociation properties of pig muscle D-glyceraldehyde-3-phosphate dehydrogenase (GAPD). The binding of NAD to dehydrogenase-FITC complex partially reverts the quenching caused by the binding of dye to apo-GAPD. This phenomenon, as well as the formation of a characteristic absorption difference spectrum caused by the binding of NAD, makes it possible to follow the NAD-induced local conformational changes near the dye-binding region. The time course of NAD-induced spectral changes shows biphasic kinetics: a burst and a slow phase. The amplitude of burst phase as a function of NAD equivalents has sigmoidal shape due to the cooperative interaction between subunits. The same conclusion could be drawn from fluorescence anisotropy measurements. In the presence of excess NAD a slow conformational change can be detected, the amplitude of which is a function of NAD concentration. This phenomenon can be attributed to the binding of further NAD molecules to the holoenzyme. The slow phase follows first-order kinetics, and the rate constant depends on enzyme concentration. The specific fluorescence intensity and the fluorescence anisotropy of fluorescent dye labeled apo-GAPD and GAPD saturated with NAD are also dependent on enzyme concentration. We suggest that NAD binding induces major changes in the steric structure of tetrameric enzyme without influencing remarkably the interacting forces between the contact surfaces of subunits. Data are quantitatively interpreted in terms of a two-step dissociation model.     

Identification of the NAD(+)-binding fold of glyceraldehyde-3-phosphate dehydrogenase as a novel RNA-binding domain

There is growing evidence that metabolic enzymes may act as multifunctional proteins performing diverse roles in cellular metabolism. Among these functions are the RNA-binding activities of NAD(+)-dependent dehydrogenases. Previously, we have characterized the glycolytic enzyme glyceraldehyde-3-phosphate dehydrogenase (GAPDH) as an RNA-binding protein with preference to adenine-uracil-rich sequences. In this study, we used GST-GAPDH fusion proteins generated by deletion mutagenesis to search for the RNA binding domain. We established that the N-terminal 43 amino acid residues of GAPDH, which correspond to the first mononucleotide-binding domain of the NAD(+)-binding fold is sufficient to confer RNA-binding. We also provide evidence that this single domain, although it retains most of the RNA-binding activity, loses sequence specificity. Our results suggest a molecular basis for RNA-recognition by NAD(+)-dependent dehydrogenases and (di)nucleotide-binding metabolic enzymes that had been reported to have RNA-binding activity with different specificity. To support this prediction we also identified other members of the family of NAD(+)-dependent dehydrogenases with no previous history of nucleic acid binding as RNA binding proteins in vitro. Based on our findings we propose the addition of the NAD(+)-binding domain to the list of RNA binding domains/motifs.     

Predicting nucleic acid binding interfaces from structural models of proteins

The function of DNA‐ and RNA‐binding proteins can be inferred from the characterization and accurate prediction of their binding interfaces. However, the main pitfall of various structure‐based methods for predicting nucleic acid binding function is that they are all limited to a relatively small number of proteins for which high‐resolution three‐dimensional structures are available. In this study, we developed a pipeline for extracting functional electrostatic patches from surfaces of protein structural models, obtained using the I‐TASSER protein structure predictor. The largest positive patches are extracted from the protein surface using the patchfinder algorithm. We show that functional electrostatic patches extracted from an ensemble of structural models highly overlap the patches extracted from high‐resolution structures. Furthermore, by testing our pipeline on a set of 55 known nucleic acid binding proteins for which I‐TASSER produces high‐quality models, we show that the method accurately identifies the nucleic acids binding interface on structural models of proteins. Employing a combined patch approach we show that patches extracted from an ensemble of models better predicts the real nucleic acid binding interfaces compared with patches extracted from independent models. Overall, these results suggest that combining information from a collection of low‐resolution structural models could be a valuable approach for functional annotation. We suggest that our method will be further applicable for predicting other functional surfaces of proteins with unknown structure. Proteins 2012. © 2011 Wiley Periodicals, Inc.     

Detection of multiscale pockets on protein surfaces using mathematical morphology

Detection of pockets on protein surfaces is an important step toward finding the binding sites of small molecules. In a previous study, we defined a pocket as a space into which a small spherical probe can enter, but a large probe cannot. The radius of the large probes corresponds to the shallowness of pockets. We showed that each type of binding molecule has a characteristic shallowness distribution. In this study, we introduced fundamental changes to our previous algorithm by using a 3D grid representation of proteins and probes, and the theory of mathematical morphology. We invented an efficient algorithm for calculating deep and shallow pockets (multiscale pockets) simultaneously, using several different sizes of spherical probes (multiscale probes). We implemented our algorithm as a new program, ghecom (grid‐based HECOMi finder). The statistics of calculated pockets for the structural dataset showed that our program had a higher performance of detecting binding pockets, than four other popular pocket‐finding programs proposed previously. The ghecom also calculates the shallowness of binding ligands, R_inaccess (minimum radius of inaccessible spherical probes) that can be obtained from the multiscale molecular volume. We showed that each part of the binding molecule had a bias toward a specific range of shallowness. These findings will be useful for predicting the types of molecules that will be most likely to bind putative binding pockets, as well as the configurations of binding molecules. The program ghecom is available through the Web server ( http://biunit.naist.jp/ghecom ). Proteins 2010. © 2009 Wiley‐Liss, Inc.     

Protein-RNA interactions: exploring binding patterns with a three-dimensional superposition analysis of high resolution structures

MOTIVATION: The recognition of specific RNA sequences and structures by proteins is critical to our understanding of RNA processing, gene expression and viral replication. The diversity of RNA structures suggests that RNA recognition is substantially different than that of DNA. RESULTS: The atomic coordinates of 41 protein-RNA complexes have been used to probe composite nucleoside binding pockets that form the structural and chemical underpinnings of base recognition. Composite nucleoside binding pockets were constructed using three-dimensional superpositions of each RNA nucleoside. Unlike protein-DNA interactions which are dominated by accessibility, RNA recognition frequently occurs in non-canonical and single-strand-like structures that allow interactions to occur from a much wider set of geometries and make fuller use of unique base shapes and hydrogen-bonding ability. By constructing composites that include all van der Waals, hydrogen-bonding, stacking and general non-polar interactions made to a particular nucleoside, the strategies employed are made readily visible. Protein-RNA interactions can result in the formation of a glove-like tight binding pocket around RNA bases, but the size, shape and non-polar binding patterns differ between specific RNA bases. We show that adenine can be distinguished from guanine based on the size and shape of the binding pocket and steric exclusion of the guanine N2 exocyclic amino group. The unique shape and hydrogen-bonding pattern for each RNA base allow proteins to make specific interactions through a very small number of contacts, as few as two in some cases. AVAILABILITY: The program ENTANGLE is available from http://www.bioc.rice.edu/~shamoo     

The binding of advanced glycation end products to cell surfaces can be measured using bead-reconstituted cellular membrane proteins

Advanced glycation end products (AGEs) that arise from the reaction of sugars with protein side chains are supposed to be involved in the pathogenesis of several diseases and therefore the effects of AGEs on cells are the objective of numerous investigations. Although different cellular responses to AGEs can be measured in cell culture studies, knowledge about the nature of AGE-binding and the involved cell surface receptors is poor. The measurement of AGE-binding to cell surfaces bears the potential to gain a deeper understanding about the nature of AGE-binding to cell surface proteins and could be applied as a preliminary test before performing cell culture studies on AGE effects. Herein, a new material and method for the detection of AGE-binding to cell surfaces is introduced, which has the potential to facilitate the detection of binding. In the present paper, the detection of AGE-binding to cell surface proteins using an artificial system of cellular membrane proteins reconstituted on beads (TRANSIL CaCo-2) is described. The binding of a BSA-AGE derived from a 37 °C incubation with 500 mM Glc (BSA-Glc 500) and the corresponding control to this artificial system was compared with the binding to intact cells and was found to be in good agreement. Additionally, the K_d for the binding of the BSA-Glc 500 used in the study to CaCo-2 surfaces was determined using FITC-labelled samples in a flow cytometric approach. Competitive binding studies were performed using a set of non-labelled BSA-AGEs to compete with FITC-labelled BSA-Glc 500 for the cell surface binding sites. The binding was found to be inhibited to different extends, virtually depending on the degree of arginine modifications within the modified protein used for competition. Additionally, the effects of all AGEs used in the study on CaCo-2 cells was measured using the detection of reactive oxygen species (ROS), which are known to be induced as a primary result of AGE-receptor binding. The induction of ROS was found to linearly correlate to the capacity of the individual AGE to displace FITC-labelled BSA-Glc 500 in competitive binding studies. Therefore, the data indicate, that at least in case of CaCo-2 cells the detection of cell surface binding can serve as a reliable preliminary test for a potential cell-damaging effect of AGEs.     

Basis for half-site ligand binding in yeast NAD(+)-specific isocitrate dehydrogenase

Yeast NAD(+)-specific isocitrate dehydrogenase is an allosterically regulated octameric enzyme composed of four heterodimers of a catalytic IDH2 subunit and a regulatory IDH1 subunit. Despite structural predictions that the enzyme would contain eight isocitrate binding sites, four NAD(+) binding sites, and four AMP binding sites, only half of the sites for each ligand can be measured in binding assays. On the basis of a potential interaction between side chains of Cys-150 residues in IDH2 subunits in each tetramer of the enzyme, ligand binding assays of wild-type (IDH1/IDH2) and IDH1/IDH2(C150S) octameric enzymes were conducted in the presence of dithiothreitol. These assays demonstrated the presence of eight isocitrate and four AMP binding sites for the wild-type enzyme in the presence of dithiothreitol and for the IDH1/IDH2(C150S) enzyme in the absence or presence of this reagent, suggesting that interactions between sulfhydryl side chains of IDH2 Cys-150 residues limit access to these sites. However, only two NAD(+) sites could be measured for either enzyme. A tetrameric form of IDH (an IDH1(G15D)/IDH2 mutant enzyme) demonstrated half-site binding for isocitrate (two sites) in the absence of dithiothreitol and full-site binding (four sites) in the presence of dithiothreitol. Only one NAD(+) site could be measured for the tetramer under both conditions. In the context of the structure of the enzyme, these results suggest that an observed asymmetry between heterotetramers in the holoenzyme contributes to interactions between IDH2 Cys-150 residues and to half-site binding of isocitrate, but that a form of negative cooperativity may limit access to apparently equivalent NAD(+) binding sites.     

Defining A Global Map of Functional Group-based 3D Ligand-binding Motifs

Uncovering conserved 3D protein–ligand binding patterns on the basis of functional groups (FGs) shared by a variety of small molecules can greatly expand our knowledge of protein–ligand interactions. Despite that conserved binding patterns for a few commonly used FGs have been reported in the literature, large-scale identification and evaluation of FG-based 3D binding motifs are still lacking. Here, we propose a computational method, Automatic FG-based Three-dimensional Motif Extractor (AFTME), for automatic mapping of 3D motifs to different FGs of a specific ligand. Applying our method to 233 naturally-occurring ligands, we define 481 FG-binding motifs that are highly conserved across different ligand-binding pockets. Systematic analysis further reveals four main classes of binding motifs corresponding to distinct sets of FGs. Combinations of FG-binding motifs facilitate the binding of proteins to a wide spectrum of ligands with various binding affinities. Finally, we show that our FG–motif map can be used to nominate FGs that potentially bind to specific drug targets, thus providing useful insights and guidance for rational design of small-molecule drugs.     

Interaction of c(60)-fullerene and carboxyfullerene with proteins: docking and binding site alignment

The unique properties of fullerenes have raised the interest of using them for biomedical applications. Within this framework, the interactions of fullerenes with proteins have been an exciting research target, yet little is known about how native proteins can bind fullerenes, and what is the nature of these interactions. Moreover, though some proteins have been shown to interact with fullerenes, up to date, no crystal structure of such complexes was obtained. Here we report docking studies aimed at examining the interactions of fullerene in two forms (C60 nonsubstituted fullerene and carboxyfullerene) with four proteins that are known to bind fullerene derivatives: HIV protease, fullerene-specific antibody, human serum albumin, and bovine serum albumin. Our work provides docking models with detailed binding pockets information, which closely match available experimental data. We further compare the predicted binding sites using a novel multiple binding site alignment method. A high similarity between the physicochemical properties and surface geometry was found for fullerene's binding sites of HIV protease and the human and bovine serum albumins.     

Divergence of Pumilio/fem-3 mRNA binding factor (PUF) protein specificity through variations in an RNA-binding pocket

mRNA control networks depend on recognition of specific RNA sequences. Pumilio-fem-3 mRNA binding factor (PUF) RNA-binding proteins achieve that specificity through variations on a conserved scaffold. Saccharomyces cerevisiae Puf3p achieves specificity through an additional binding pocket for a cytosine base upstream of the core RNA recognition site. Here we demonstrate that this chemically simple adaptation is prevalent and contributes to the diversity of RNA specificities among PUF proteins. Bioinformatics analysis shows that mRNAs associated with Caenorhabditis elegans fem-3 mRNA binding factor (FBF)-2 in vivo contain an upstream cytosine required for biological regulation. Crystal structures of FBF-2 and C. elegans PUF-6 reveal binding pockets structurally similar to that of Puf3p, whereas sequence alignments predict a pocket in PUF-11. For Puf3p, FBF-2, PUF-6, and PUF-11, the upstream pockets and a cytosine are required for maximal binding to RNA, but the quantitative impact on binding affinity varies. Furthermore, the position of the upstream cytosine relative to the core PUF recognition site can differ, which in the case of FBF-2 originally masked the identification of this consensus sequence feature. Importantly, other PUF proteins lack the pocket and so do not discriminate upstream bases. A structure-based alignment reveals that these proteins lack key residues that would contact the cytosine, and in some instances, they also present amino acid side chains that interfere with binding. Loss of the pocket requires only substitution of one serine, as appears to have occurred during the evolution of certain fungal species.