Automated generation of MCSS-derived pharmacophoric DOCK site points for searching multiconformation databases

All docking methods employ some sort of heuristic to orient the ligand molecules into the binding site of the target structure. An automated method, MCSS2SPTS, for generating chemically labeled site points for docking is presented. MCSS2SPTS employs the program Multiple Copy Simultaneous Search (MCSS) to determine target-based theoretical pharmacophores. More specifically, chemically labeled site points are automatically extracted from selected low-energy functional-group minima and clustered together. These pharmacophoric site points can then be directly matched to the pharmacophoric features of database molecules with the use of either DOCK or PhDOCK to place the small molecules into the binding site. Several examples of the ability of MCSS2SPTS to reproduce the three-dimensional pharmacophoric features of ligands from known ligand-protein complex structures are discussed. In addition, a site-point set calculated for one human immunodeficiency virus 1 (HIV1) protease structure is used with PhDOCK to dock a set of HIV1 protease ligands; the docked poses are compared to the corresponding complex structures of the ligands. Finally, the use of an MCSS2SPTS-derived site-point set for acyl carrier protein synthase is compared to the use of atomic positions from a bound ligand as site points for a large-scale DOCK search. In general, MCSS2SPTS-generated site points focus the search on the more relevant areas and thereby allow for more effective sampling of the target site.     

Similarity-driven flexible ligand docking

A similarity-driven approach to flexible ligand docking is presented. Given a reference ligand or a pharmacophore positioned in the protein active site, the method allows inclusion of a similarity term during docking. Two different algorithms have been implemented, namely, a similarity-penalized docking (SP-DOCK) and a similarity-guided docking (SG-DOCK). The basic idea is to maximally exploit the structural information about the ligand binding mode present in cases where ligand-bound protein structures are available, information that is usually ignored in standard docking procedures. SP-DOCK and SG-DOCK have been derived as modified versions of the program DOCK 4.0, where the similarity program MIMIC acts as a module for the calculation of similarity indices that correct docking energy scores at certain steps of the calculation. SP-DOCK applies similarity corrections to the set of ligand orientations at the end of the ligand incremental construction process, penalizing the docking energy and, thus, having only an effect on the relative ordering of the final solutions. SG-DOCK applies similarity corrections throughout the entire ligand incremental construction process, thus affecting not only the relative ordering of solutions but also actively guiding the ligand docking. The performance of SP-DOCK and SG-DOCK for binding mode assessment and molecular database screening is discussed. When applied to a set of 32 thrombin ligands for which crystal structures are available, SG-DOCK improves the average RMSD by ca. 1 A when compared with DOCK. When those 32 thrombin ligands are included into a set of 1,000 diverse molecules from the ACD, DIV, and WDI databases, SP-DOCK significantly improves the retrieval of thrombin ligands within the first 10% of each of the three databases with respect to DOCK, with minimal additional computational cost. In all cases, comparison of SP-DOCK and SG-DOCK results with those obtained by DOCK and MIMIC is performed.     

Incorporating replacement free energy of binding‐site waters in molecular docking

Binding‐site water molecules play a crucial role in protein‐ligand recognition, either being displaced upon ligand binding or forming water bridges to stabilize the complex. However, rigorously treating explicit binding‐site waters is challenging in molecular docking, which requires to fully sample ensembles of waters and to consider the free energy cost of replacing waters. Here, we describe a method to incorporate structural and energetic properties of binding‐site waters into molecular docking. We first developed a solvent property analysis (SPA) program to compute the replacement free energies of binding‐site water molecules by post‐processing molecular dynamics trajectories obtained from ligand‐free protein structure simulation in explicit water. Next, we implemented a distance‐dependent scoring term into DOCK scoring function to take account of the water replacement free energy cost upon ligand binding. We assessed this approach in protein targets containing important binding‐site waters, and we demonstrated that our approach is reliable in reproducing the crystal binding geometries of protein‐ligand‐water complexes, as well as moderately improving the ligand docking enrichment performance. In addition, SPA program (free available to academic users upon request) may be applied in identifying hot‐spot binding‐site residues and structure‐based lead optimization. Proteins 2014; 82:1765–1776. © 2014 Wiley Periodicals, Inc.     

Recognition templates for predicting adenylate-binding sites in proteins.

Recognition templates encapsulate the structural and energetic features for the specific recognition of a given ligand by a protein active site. These templates identify the major interactions used for specific recognition and may be used to find specific binding sites in proteins of unknown function. We present a grid-based method for deriving recognition templates for adenylate groups from a set of diverse nucleotide-binding proteins. The templates reveal the basis of specific binding of adenylate, including tight shape complementarity, specific hydrogen bonds, and underscoring the importance of a key steric contact for excluding guanylate from adenylate-specific sites. We demonstrate the utility of recognition templates in identifying specific adenylate-binding sites in a diverse set of dinucleotide-binding proteins.     

Role of structural water for prediction of cation binding sites in apoproteins

Structures of many metal-binding proteins are often obtained without structural cations in their apoprotein forms. Missing cation coordinates are usually updated from structural templates constructed from many holoprotein structures. Such templates usually do not include structural water, the important contributor to the ion binding energy. Structural templates are also inconvenient for taking into account structural modifications around the binding site at apo-/holo- transitions. An approach based upon statistical potentials readily takes into account structural modifications associated with binding as well as contribution of structural water molecules. Here, we construct a set of statistical potentials for Mg²⁺, Ca²⁺, and Zn²⁺ contacting with protein atoms of a different type or structural water oxygens. Each type of the cations tends to form tight contacts with protein atoms of specific types. Structural water contributes relatively more into the binding pseudo-energy of Mg²⁺ and Ca²⁺ than of Zn²⁺. We have developed PIONCA (Protein-Ion Calculator), a fast CUDA GPGPU-based algorithm that predicts ion-binding sites in apoproteins. Comparative tests demonstrate that PIONCA outperforms most of the tools based on structural templates or docking. Our software can be also used for locating bound cations in holoprotein structures with missing cation heteroatoms. PIONCA is equipped with an interactive web interface based upon JSmol.     

Ligand Binding Site Similarity Identification Based on Chemical and Geometric Similarity

The similarity comparison of binding sites based on amino acid between different proteins can facilitate protein function identification. However, Binding site usually consists of several crucial amino acids which are frequently dispersed among different regions of a protein and consequently make the comparison of binding sites difficult. In this study, we introduce a new method, named as chemical and geometric similarity of binding site (CGS-BSite), to compute the ligand binding site similarity based on discrete amino acids with maximum-weight bipartite matching algorithm. The principle of computing the similarity is to find a Euclidean Transformation which makes the similar amino acids approximate to each other in a geometry space, and vice versa. CGS-BSite permits site and ligand flexibilities, provides a stable prediction performance on the flexible ligand binding sites. Binding site prediction on three test datasets with CGS-BSite method has similar performance to Patch-Surfer method but outperforms other five tested methods, reaching to 0.80, 0.71 and 0.85 based on the area under the receiver operating characteristic curve, respectively. It performs a marginally better than Patch-Surfer on the binding sites with small volume and higher hydrophobicity, and presents good robustness to the variance of the volume and hydrophobicity of ligand binding sites. Overall, our method provides an alternative approach to compute the ligand binding site similarity and predict potential special ligand binding sites from the existing ligand targets based on the target similarity.     

The effect of different electrostatic potentials on docking accuracy: a case study using DOCK5.4

As a commonly used structure-based approach for virtual screening, molecular design and lead optimization, molecular docking can search the preferred orientation and conformation of a ligand for its optimal binding to a receptor or enzyme active site. In doing so, selecting an appropriate method to calculate the electrostatic potentials is critical. In the current report, nine different semi-empirical and empirical methods, including AM1, AM1-BCC, Del-Re, MMFF, Gasteiger, Hückel, Gasteiger-Hückel, Pullman and formal charges were investigated for their performance on the prediction of docking poses using the DOCK5.4 program. The results demonstrated that the AM1-BCC charges had the highest success rate.     

Direct identification of the agonist binding site in the human brain cholecystokininB receptor.

In investigating the agonist binding site of the human brain cholecystokininB receptor (CCKBR), we employed the direct protein chemical approach using a photoreactive tritiated analogue of sulfated cholecystokinin octapeptide, which contains the p-benzoylbenzoyl moiety at the N-terminus, followed by purification of the affinity-labeled receptor to homogeneity. This probe bound specifically, saturably, and with high affinity (KD = 1.2 nM) to the CCKBR and has full agonistic activity. As the starting material for receptor purification, we used stably transfected HEK 293 cells overexpressing functional CCKBR. Covalent labeling of the WGA-lectin-enriched receptor revealed a 70-80 kDa glycoprotein with a protein core of about 50 kDa. Identification of the agonist binding site was achieved by the application of subsequent chemical and enzymatical cleavage to the purified receptor. A radiolabeled peptide was identified by Edman degradation amino acid sequence analysis combined with MALDI-TOF mass spectrometry. The position of the radioactive probe within the identified peptide was determined using combined tandem electrospray mass spectrometry and peptide mapping. The probe was covalently attached within the sequence L52ELAIRITLY61 that represents the transition between the N-terminal domain and predicted transmembrane domain 1. Using this interaction as a constraint to orientate the ligand within the putative receptor binding site, a model of the CCK-8s-occupied CCKBR was constructed. The hormone was found to be placed in a binding pocket built from both extracellular and transmembrane domains of CCKBR with its N-terminus mainly interacting with residues Arg57 and Tyr61.     

Application of the message-address concept to the docking of naltrexone and selective naltrexone-derived opioid antagonists into opioid receptor models

A binding site model for the opioid family of G-protein coupled receptors (GPCRs) is proposed based on the message-address concept of ligand recognition. Using ligand docking studies of the universal opioid antagonist, naltrexone, the structural basis for ‘message’ recognition is explored across all three receptor types, μ, δ, and κ. The binding mode proposed and basis for selectivity are also rationalized using the naltrexone-derived ligands, naltrindole (NTI) and norbinaltorphimine (nor BNI). These ligands are docked to the receptor according to the common naltrexone core or message. The resulting orientation places key ‘address’ elements in close proximity to amino acid residues critical to selectivity among receptor types. Selectivity is explained by sequence differences in the μ, δ, and κ receptors at these recognition points. Support for the model is derived from site directed mutagenesis studies and ligand binding data for the opioid receptors and other related GPCRs. Special issue dedicated to Dr. Eric J. Simon     

The ligand binding site and transduction mechanism in the inositol-1,4,5-triphosphate receptor.

The inositol-1,4,5-triphosphate (InsP3) receptor consists of a homotetramer of highly conserved 313 kd subunits that contain multiple transmembrane regions in the C-terminal part of the protein. The receptor was expressed in COS cells and its domain structure was studied by mutagenesis. Deletion of the transmembrane regions from the receptor results in the synthesis of a soluble receptor protein that efficiently binds InsP3 but which instead of associating into homotetramers remains monomeric. This result suggests a role for the transmembrane regions in the association of the receptor subunits into tetramers but not in ligand binding. To localize the ligand binding site, further cDNAs encoding truncated receptor proteins were constructed. Assays of InsP3 binding to these truncated InsP3 receptors revealed that sequences in the N-terminal fourth of the InsP3 receptor are sufficient for ligand binding. Accordingly, each subunit of the InsP3 receptor homotetramer contains an independent ligand binding site that is located on the N-terminal ends of each subunit and is separated from the putative channel-forming transmembrane regions by greater than 1400 amino acids. Gel filtration experiments demonstrate a large conformational change of the receptor as a function of ligand binding, suggesting a mechanism by which ligand binding might cause channel opening.     

Knowledge-based scoring function to predict protein-ligand interactions

The development and validation of a new knowledge-based scoring function (DrugScore) to describe the binding geometry of ligands in proteins is presented. It discriminates efficiently between well-docked ligand binding modes (root-mean-square deviation <2.0 A with respect to a crystallographically determined reference complex) and those largely deviating from the native structure, e.g. generated by computer docking programs. Structural information is extracted from crystallographically determined protein-ligand complexes using ReLiBase and converted into distance-dependent pair-preferences and solvent-accessible surface (SAS) dependent singlet preferences for protein and ligand atoms. Definition of an appropriate reference state and accounting for inaccuracies inherently present in experimental data is required to achieve good predictive power. The sum of the pair preferences and the singlet preferences is calculated based on the 3D structure of protein-ligand binding modes generated by docking tools. For two test sets of 91 and 68 protein-ligand complexes, taken from the Protein Data Bank (PDB), the calculated score recognizes poses generated by FlexX deviating <2 A from the crystal structure on rank 1 in three quarters of all possible cases. Compared to FlexX, this is a substantial improvement. For ligand geometries generated by DOCK, DrugScore is superior to the "chemical scoring" implemented into this tool, while comparable results are obtained using the "energy scoring" in DOCK. None of the presently known scoring functions achieves comparable power to extract binding modes in agreement with experiment. It is fast to compute, regards implicitly solvation and entropy contributions and produces correctly the geometry of directional interactions. Small deviations in the 3D structure are tolerated and, since only contacts to non-hydrogen atoms are regarded, it is independent from assumptions of protonation states.     

Probabilistic Prediction of Contacts in Protein-Ligand Complexes

We introduce a statistical method for evaluating atomic level 3D interaction patterns of protein-ligand contacts. Such patterns can be used for fast separation of likely ligand and ligand binding site combinations out of all those that are geometrically possible. The practical purpose of this probabilistic method is for molecular docking and scoring, as an essential part of a scoring function. Probabilities of interaction patterns are calculated conditional on structural x-ray data and predefined chemical classification of molecular fragment types. Spatial coordinates of atoms are modeled using a Bayesian statistical framework with parametric 3D probability densities. The parameters are given distributions a priori, which provides the possibility to update the densities of model parameters with new structural data and use the parameter estimates to create a contact hierarchy. The contact preferences can be defined for any spatial area around a specified type of fragment. We compared calculated contact point hierarchies with the number of contact atoms found near the contact point in a reference set of x-ray data, and found that these were in general in a close agreement. Additionally, using substrate binding site in cathechol-O-methyltransferase and 27 small potential binder molecules, it was demonstrated that these probabilities together with auxiliary parameters separate well ligands from decoys (true positive rate 0.75, false positive rate 0). A particularly useful feature of the proposed Bayesian framework is that it also characterizes predictive uncertainty in terms of probabilities, which have an intuitive interpretation from the applied perspective.     

A fragment library based on Gaussian mixtures predicting favorable molecular interactions

Here, a protein atom-ligand fragment interaction library is described. The library is based on experimentally solved structures of protein-ligand and protein-protein complexes deposited in the Protein Data Bank (PDB) and it is able to characterize binding sites given a ligand structure suitable for a protein. A set of 30 ligand fragment types were defined to include three or more atoms in order to unambiguously define a frame of reference for interactions of ligand atoms with their receptor proteins. Interactions between ligand fragments and 24 classes of protein target atoms plus a water oxygen atom were collected and segregated according to type. The spatial distributions of individual fragment - target atom pairs were visually inspected in order to obtain rough-grained constraints on the interaction volumes. Data fulfilling these constraints were given as input to an iterative expectation-maximization algorithm that produces as output maximum likelihood estimates of the parameters of the finite Gaussian mixture models. Concepts of statistical pattern recognition and the resulting mixture model densities are used (i) to predict the detailed interactions between Chlorella virus DNA ligase and the adenine ring of its ligand and (ii) to evaluate the "error" in prediction for both the training and validation sets of protein-ligand interaction found in the PDB. These analyses demonstrate that this approach can successfully narrow down the possibilities for both the interacting protein atom type and its location relative to a ligand fragment.     

Mechanisms for ligand binding to GluR0 ion channels: crystal structures of the glutamate and serine complexes and a closed apo state

High-resolution structures of the ligand binding core of GluR0, a glutamate receptor ion channel from Synechocystis PCC 6803, have been solved by X-ray diffraction. The GluR0 structures reveal homology with bacterial periplasmic binding proteins and the rat GluR2 AMPA subtype neurotransmitter receptor. The ligand binding site is formed by a cleft between two globular alpha/beta domains. L-Glutamate binds in an extended conformation, similar to that observed for glutamine binding protein (GlnBP). However, the L-glutamate gamma-carboxyl group interacts exclusively with Asn51 in domain 1, different from the interactions of ligand with domain 2 residues observed for GluR2 and GlnBP. To address how neutral amino acids activate GluR0 gating we solved the structure of the binding site complex with L-serine. This revealed solvent molecules acting as surrogate ligand atoms, such that the serine OH group makes solvent-mediated hydrogen bonds with Asn51. The structure of a ligand-free, closed-cleft conformation revealed an extensive hydrogen bond network mediated by solvent molecules. Equilibrium centrifugation analysis revealed dimerization of the GluR0 ligand binding core with a dissociation constant of 0.8 microM. In the crystal, a symmetrical dimer involving residues in domain 1 occurs along a crystallographic 2-fold axis and suggests that tetrameric glutamate receptor ion channels are assembled from dimers of dimers. We propose that ligand-induced conformational changes cause the ion channel to open as a result of an increase in domain 2 separation relative to the dimer interface.     

SuperStar: a knowledge-based approach for identifying interaction sites in proteins.

An empirical method for identifying interaction sites in proteins is described and validated. The method is based entirely on experimental information about non-bonded interactions occurring in small-molecule crystal structures. These data are used in the form of scatterplots that show the experimentally observed distribution of one functional group (the "contact group" or "probe") around another. A template molecule (e.g. a protein binding site) is broken down into structure fragments and the scatterplots, showing the distribution of a chosen probe around these structure fragments, are superimposed on the corresponding parts of the template. The scatterplots are then translated into a three-dimensional map that shows the propensity of the probe at different positions around the template molecule. The method is illustrated for l -arabinose-binding protein, complexed with l -arabinose and with d -fucose, and for dihydrofolate reductase complexed with methotrexate. The method is validated on 122 X-ray structures of protein-ligand complexes. For all the binding sites of these proteins, propensity maps are generated for four different probes: a charged NH+3nitrogen, a carbonyl oxygen, a hydroxyl oxygen and a methyl carbon atom. Next, the maps are compared with the experimentally observed positions of ligand atoms of these types. For 74% of these ligand atoms (84% of the solvent-inaccessible ones) the calculated propensity of the matching probe at the experimental positions is higher than expected by chance. For 68% of the atoms (82% of the solvent-inaccessible ones) the propensity of the matching probe is higher than that of the other three probes. These results indicate that the approach generally gives good predictions for protein-ligand interactions. The potential applications of the propensity maps range from an aid in manual docking and structure-based drug design to their use in pharmacophore development.     

Construction of protein binding sites in scaffold structures

We have developed a strategy for grafting a protein-protein interface based on the known crystal structure of a native ligand and receptor proteins in a complex. The key interaction residues at the ligand protein binding interface are transferred onto a scaffold protein so that the mutated scaffold protein will bind the receptor protein in the same manner as the ligand protein. First, our method identifies key residues and atoms in the ligand protein, which strongly interact with the receptor protein. Second, this method searches the scaffold protein for combinations of candidate residues, among which the distance between any two candidate residues is similar to that between relevant key interaction residues in the ligand protein. These candidate residues are mutated to key interaction residues in the ligand protein respectively. The scaffold protein is superposed onto the ligand protein based upon the coordinates of corresponding atoms, which are assumed to strongly interact with the receptor protein. Complementarity between scaffold and receptor proteins is evaluated. Scaffold proteins with a low superposing rms difference and high complementary score are accepted for further analysis. Then, the relative position of the scaffold protein is adjusted so that the interfaces between the scaffold and receptor proteins have a reasonable packing density. Other mutations are also considered to reduce the desolvation energy or bad steric contacts. Finally, the scaffold protein is cominimized with the receptor protein and evaluated. To test the method, the binding interface of barstar, the inhibitor of barnase, was grafted onto small proteins. Four scaffold proteins with high complementary scores are accepted.     

Characterization of receptors with a new negative image: Use in molecular docking and lead optimization

The characterization of receptor binding sites is an important aspect of molecular docking, molecular recognition, and the structure-based design process. This characterization can take several forms: the receptor surface itself can be delineated or described, the space adjacent to the surface can be chemically mapped, or a negative image of the protein binding region can be generated. In this report, we describe a new method of constructing a negative image through generation of a set of spheres. These spheres lie along the receptor surface, and their centers represent possible ligand atom positions. By the method in which they are constructed, these spheres carry a limited amount of energetic and chemical information in addition to their primary geometric information. We test the accuracy of the image by comparing sphere positions to the positions of bound ligand atoms and propose a figure of merit for such tests. Then, we use the spheres to orient ligands in enzyme active sites and show how they can be used to generate low scoring configurations more efficiently than other approaches that search orientation space. In addition, two novel applications of these spheres are described: they are used to help identify structural differences among families of enzymes and to suggest points for ligand modification in analog design. Proteins 30:321–336, 1998. © 1998 Wiley-Liss, Inc.     

A method for localizing ligand binding pockets in protein structures

The accurate identification of ligand binding sites in protein structures can be valuable in determining protein function. Once the binding site is known, it becomes easier to perform in silico and experimental procedures that may allow the ligand type and the protein function to be determined. For example, binding pocket shape analysis relies heavily on the correct localization of the ligand binding site. We have developed SURFNET-ConSurf, a modular, two-stage method for identifying the location and shape of potential ligand binding pockets in protein structures. In the first stage, the SURFNET program identifies clefts in the protein surface that are potential binding sites. In the second stage, these clefts are trimmed in size by cutting away regions distant from highly conserved residues, as defined by the ConSurf-HSSP database. The largest clefts that remain tend to be those where ligands bind. To test the approach, we analyzed a nonredundant set of 244 protein structures from the PDB and found that SURFNET-ConSurf identifies a ligand binding pocket in 75% of them. The trimming procedure reduces the original cleft volumes by 30% on average, while still encompassing an average 87% of the ligand volume. From the analysis of the results we conclude that for those cases in which the ligands are found in large, highly conserved clefts, the combined SURFNET-ConSurf method gives pockets that are a better match to the ligand shape and location. We also show that this approach works better for enzymes than for nonenzyme proteins.