Ligand-specific scoring functions: improved ranking of docking solutions1

Molecular docking is a powerful computational method that has been widely used in many biomolecular studies to predict geometry of a protein-ligand complex. However, while its conformational search algorithms are usually able to generate correct conformation of a ligand in the binding site, the scoring methods often fail to discriminate it among many false variants. We propose to treat this problem by applying more precise ligand-specific scoring filters to re-rank docking solutions. In this way specific features of interactions between protein and different types of compounds can be implicitly taken into account. New scoring functions were constructed including hydrogen bonds, hydrophobic and hydrophilic complementarity terms. These scoring functions also discriminate ligands by the size of the molecule, the total hydrophobicity, and the number of peptide bonds for peptide ligands. Weighting coefficients of the scoring functions were adjusted using a training set of 60 protein–ligand complexes. The proposed method was then tested on the results of docking obtained for an additional 70 complexes. In both cases the success rate was 5–8% better compared to the standard functions implemented in popular docking software.     

Morphological similarity: A 3D molecular similarity method correlated with protein-ligand recognition

Recognition of small molecules by proteins depends on three-dimensional molecular surface complementarity. However, the dominant techniques for analyzing the similarity of small molecules are based on two-dimensional chemical structure, with such techniques often outperforming three-dimensional techniques in side-by-side comparisons of correlation to biological activity. This paper introduces a new molecular similarity method, termed morphological similarity (MS), that addresses the apparent paradox. Two sets of molecule pairs are identified from a set of ligands whose protein-bound states are known crystallographically. Pairs that bind the same protein sites form the first set, and pairs that bind different sites form the second. MS is shown to separate the two sets significantly better than a benchmark 2D similarity technique. Further, MS agrees with crystallographic observation of bound ligand states, independent of information about bound states. MS is efficient to compute and can be practically applied to large libraries of compounds.     

Nanoparticle shape anisotropy dictates the collective behavior of surface-bound ligands

We report on the modification of the properties of surface-confined ligands in nanoparticle systems through the introduction of shape anisotropy. Specifically, triangular gold nanoprisms, densely functionalized with oligonucleotide ligands, hybridize to complementary particles with an affinity that is several million times higher than that of spherical nanoparticle conjugates functionalized with the same amount of DNA. In addition, they exhibit association rates that are 2 orders of magnitude greater than those of their spherical counterparts. This phenomenon stems from the ability of the flat, extended facets of nonspherical nanoparticles to (1) support more numerous ligand interactions through greater surface contact with complementary particles, (2) increase the effective local concentration of terminal DNA nucleotides that mediate hybridization, and (3) relieve the conformational stresses imposed on nanoparticle-bound ligands participating in interactions between curved surfaces. Finally, these same trends are observed for the pH-mediated association of nanoparticles functionalized with carboxylate ligands, demonstrating the generality of these findings.     

Distilling the essential features of a protein surface for improving protein-ligand docking,scoring, and virtual screening

For the successful identification and docking of new ligands to a protein target by virtual screening, the essential features of the protein and ligand surfaces must be captured and distilled in an efficient representation. Since the running time for docking increases exponentially with the number of points representing the protein and each ligand candidate, it is important to place these points where the best interactions can be made between the protein and the ligand. This definition of favorable points of interaction can also guide protein structure-based ligand design, which typically focuses on which chemical groups provide the most energetically favorable contacts. In this paper, we present an alternative method of protein template and ligand interaction point design that identifies the most favorable points for making hydrophobic and hydrogen–bond interactions by using a knowledge base. The knowledge-based protein and ligand representations have been incorporated in version 2.0 of SLIDE and resulted in dockings closer to the crystal structure orientations when screening a set of 57 known thrombin and glutathione S–transferase (GST) ligands against the apo structures of these proteins. There was also improved scoring enrichment of the dockings, meaning better differentiation between the chemically diverse known ligands and a 15,000-molecule dataset of randomly-chosen small organic molecules. This approach for identifying the most important points of interaction between proteins and their ligands can equally well be used in other docking and design techniques. While much recent effort has focused on improving scoring functions for protein-ligand docking, our results indicate that improving the representation of the chemistry of proteins and their ligands is another avenue that can lead to significant improvements in the identification, docking, and scoring of ligands.(These authors contributed equally to this work)     

An NMR Method To Pinpoint Supramolecular Ligand Binding to Basic Residues on Proteins

Targeting protein surfaces involved in protein–protein interactions by using supramolecular chemistry is a rapidly growing field. NMR spectroscopy is the method of choice to map ligand‐binding sites with single‐residue resolution by amide chemical shift perturbation and line broadening. However, large aromatic ligands affect NMR signals over a greater distance, and the binding site cannot be determined unambiguously by relying on backbone signals only. We herein employed Lys‐ and Arg‐specific H2(C)N NMR experiments to directly observe the side‐chain atoms in close contact with the ligand, for which the largest changes in the NMR signals are expected. The binding of Lys‐ and Arg‐specific supramolecular tweezers and a calixarene to two model proteins was studied. The H2(C)N spectra track the terminal CH₂ groups of all Lys and Arg residues, revealing significant differences in their binding kinetics and chemical shift perturbation, and can be used to clearly pinpoint the order of ligand binding.     

Pattern recognition strategies for molecular surfaces. II. Surface complementarity

Fuzzy logic based algorithms for the quantitative treatment of complementarity of molecular surfaces are presented. Therein, the overlapping surface patches defined in article I1 of this series are used. The identification of complementary surface patches can be considered as a first step for the solution of molecular docking problems. Standard technologies can then be used for further optimization of the resulting complex structures. The algorithms are applied to 33 biomolecular complexes. After the optimization with a downhill simplex method, for all these complexes one structure was found, which is in very good agreement with the experimental results.     

Ligand-specific scoring functions: improved ranking of docking solutions

Molecular docking is a powerful computational method that has been widely used in many biomolecular studies to predict geometry of a protein-ligand complex. However, while its conformational search algorithms are usually able to generate correct conformation of a ligand in the binding site, the scoring methods often fail to discriminate it among many false variants. We propose to treat this problem by applying more precise ligand-specific scoring filters to re-rank docking solutions. In this way specific features of interactions between protein and different types of compounds can be implicitly taken into account. New scoring functions were constructed including hydrogen bonds, hydrophobic and hydrophilic complementarity terms. These scoring functions also discriminate ligands by the size of the molecule, the total hydrophobicity, and the number of peptide bonds for peptide ligands. Weighting coefficients of the scoring functions were adjusted using a training set of 60 protein-ligand complexes. The proposed method was then tested on the results of docking obtained for an additional 70 complexes. In both cases the success rate was 5-8% better compared to the standard functions implemented in popular docking software.     

Electrostatic complementarity between proteins and ligands. 3. Structural basis

Summary Electrostatic potential complementarity between ligands and their receptor sites is evaluated by the superposition of the electrostatic potential, generated by the receptor, onto the ligand potential over the ligand van der Waals surface. We would like to examine which structural factors generate this pattern of superposition. Example studies suggest that in many ligand-protein pairs, there exist principal formal charges on each molecule, largely responsible for the electrostatic potential complementarity observed. Electrostatic potential complementarity depends on the relative disposition of these principal charges and the ligand van der Waals surface. Simple mathematical models were constructed to predict the complementarity solely from structural considerations. The essential conditions for electrostatic potential complementarity were elucidated. These can be used in ligand design strategies to obtain an electrostatically optimal ligand.     

Fast computation,rotation, and comparison of low resolution spherical harmonic molecular surfaces

A procedure that rapidly generates an approximate parametric representation of macromolecular surface shapes is described. The parametrization is expressed as an expansion of real spherical harmonic basis functions. The advantage of using a parametric representation is that a pair of surfaces can be matched by using a quasi-Newton algorithm to minimize a suitably chosen objective function. Spherical harmonics are a natural and convenient choice of basis function when the task is one of search in a rotational search space. In particular, rotations of a molecular surface can be simulated by rotating only the harmonic expansion coefficients. This rotational property is applied for the first time to the 3-dimensional molecular similarity problem in which a pair of similar macromolecular surfaces are to be maximally superposed. The method is demonstrated with the superposition of antibody heavy chain variable domains. Special attention is given to computational efficiency. The spherical harmonic expansion coefficients are determined using fast surface sampling and integration schemes based on the tessellation of a regular icosahedron. Low resolution surfaces can be generated and displayed in under 10 s and a pair of surfaces can be maximally superposed in under 3 s on a contemporary workstation. ©1999 John Wiley & Sons, Inc. J Comput Chem 20: 383–395, 1999     

On the Role of Flexibility in Protein–Ligand Interactions: the Example of p53 Tetramerization Domain

The recognition of protein surfaces by designed ligands has become an attractive approach in drug discovery. However, the variable nature and irregular behavior of protein surfaces defy this new area of research. The easy to understand “lock‐and‐key” model is far from being the ideal paradigm in biomolecular interactions and, hence, any new finding on how proteins and ligands behave in recognition events paves a step of the way. Herein, we illustrate a clear example on how an increase in flexibility of both protein and ligand can result in an increase in the stability of the macromolecular complex. The biophysical study of the interaction between a designed flexible tetraguanidinium‐calix[4]arene and the tetramerization domain of protein p53 (p53TD) and its natural mutant p53TD‐R337H shows how the floppy mutant domain interacts more tightly with the ligand than the well‐packed wild‐type protein. Moreover, the flexible calixarene ligand interacts with higher affinity to both wild‐type and mutated protein domains than a conformationally rigid calixarene analog previously reported. These findings underscore the crucial role of flexibility in molecular recognition processes, for both small ligands and large biomolecular surfaces.     

Ligand-biased ensemble receptor docking (LigBEnD): a hybrid ligand/receptor structure-based approach

Ligand docking to flexible protein molecules can be efficiently carried out through ensemble docking to multiple protein conformations, either from experimental X-ray structures or from in silico simulations. The success of ensemble docking often requires the careful selection of complementary protein conformations, through docking and scoring of known co-crystallized ligands. False positives, in which a ligand in a wrong pose achieves a better docking score than that of native pose, arise as additional protein conformations are added. In the current study, we developed a new ligand-biased ensemble receptor docking method and composite scoring function which combine the use of ligand-based atomic property field (APF) method with receptor structure-based docking. This method helps us to correctly dock 30 out of 36 ligands presented by the D3R docking challenge. For the six mis-docked ligands, the cognate receptor structures prove to be too different from the 40 available experimental Pocketome conformations used for docking and could be identified only by receptor sampling beyond experimentally explored conformational subspace.     

A computational tool to optimize ligand selectivity between two similar biomacromolecular targets

Summary Algorithms for a new computer program designed to increase ligand--receptor selectivity between two proteins are described. In this program ligand--receptor selectivity is increased by functional modifications to the ligand so as to increase the calculated binding affinity of it to one protein and/or decrease the calculated binding affinity of it to the other protein. The structure of the ligand is modified by selective replacement of atoms and/or functional groups in silico based on a specific set of steric and/or hydropathic complementarity rules involving atoms and functional groups. Relative binding scores are calculated with simple grid-based steric penalty, hydrogen bond complementarity, and with the HINT score model. Two examples are shown. First, modifying the structure of the ligand CB3717 is illustrated in a number of ways such that the binding selectivity to wild type L. casei thymidylate synthase or its E60Q mutant may be improved. Second, starting with a non-selective lead compound that had been co-crystallized with both plant and mammalian 4-hydroxyphenylpyruvate dioxygenases, new compounds (similar to selective ligands discovered by screening) to improve the selectivity of (herbicidal) inhibitors for the plant enzyme were designed by the program.     

Differential Epitope Mapping by STD NMR Spectroscopy To Reveal the Nature of Protein–Ligand Contacts

Saturation transfer difference (STD) NMR spectroscopy is extensively used to obtain epitope maps of ligands binding to protein receptors, thereby revealing structural details of the interaction, which is key to direct lead optimization efforts in drug discovery. However, it does not give information about the nature of the amino acids surrounding the ligand in the binding pocket. Herein, we report the development of the novel method differential epitope mapping by STD NMR (DEEP‐STD NMR) for identifying the type of protein residues contacting the ligand. The method produces differential epitope maps through 1) differential frequency STD NMR and/or 2) differential solvent (D₂O/H₂O) STD NMR experiments. The two approaches provide different complementary information on the binding pocket. We demonstrate that DEEP‐STD NMR can be used to readily obtain pharmacophore information on the protein. Furthermore, if the 3D structure of the protein is known, this information also helps in orienting the ligand in the binding pocket.     

An automated method for predicting the positions of hydrogen-bonding atoms in binding sites

Hydrogen bonds are the most specific, and therefore predictable of the intermolecular interactions involved in ligand–protein binding. Given the structure of a molecule, it is possible to estimate the positions at which complementary hydrogen-bonding atoms could be found. Crystal-survey data are used in the design of a program, HBMAP, that generates a hydrogen-bond map for any given ligand, which contains all the feasible positions at which a complementary atom could be found. On superposition of ligands, the overlapping regions of their maps represent positions of receptor atoms to which each molecule can bind. The certainty of these positions is increased by the incorporation of a larger number and diversity of molecules. In this work, superposition is achieved using the program HBMATCH, which uses simulated annealing to generate the correspondence between points from the hydrogen-bonding maps of the two molecules. Equivalent matches are distinguished on the basis of their steric similarity. The strategy is tested on a number of ligands for which ligand–protein complexes have been solved crystallographically, which allows validation of the techniques. The receptor atom positions of thermolysin are successfully predicted when the correct superposition is obtained.     

Protein‐specific force field derived from the fragment molecular orbital method can improve protein–ligand binding interactions

Accurate computational estimate of the protein–ligand binding affinity is of central importance in rational drug design. To improve accuracy of the molecular mechanics (MM) force field (FF) for protein–ligand simulations, we use a protein‐specific FF derived by the fragment molecular orbital (FMO) method and by the restrained electrostatic potential (RESP) method. Applying this FMO‐RESP method to two proteins, dodecin, and lysozyme, we found that protein‐specific partial charges tend to differ more significantly from the standard AMBER charges for isolated charged atoms. We did not see the dependence of partial charges on the secondary structure. Computing the binding affinities of dodecin with five ligands by MM PBSA protocol with the FMO‐RESP charge set as well as with the standard AMBER charges, we found that the former gives better correlation with experimental affinities than the latter. While, for lysozyme with five ligands, both charge sets gave similar and relatively accurate estimates of binding affinities. © 2013 Wiley Periodicals, Inc.     

Electrostatic complementarity between proteins and ligands. 2. Ligand moieties

Summary Drug design strategies consider factors governing intermolecular interactions to build up putative ligands. In many strategies, the ligand is constructed using fragments which are placed in the site sequentially. The optimization is then performed with each fragment. We would like to examine if this optimization strategy could generate ligands with optimal electrostatic interactions. The electrostatic complementarities between constituent moieties and the receptor site have been calculated. The whole-ligand complementarity does not appear to be the mathematical mean of the individual complementarities, nor have we found a simple relationship between the moiety and whole-ligand complementarities. The results demonstrate clearly that, using a simple model, it is very difficult to predict the electrostatic potential complementarity of the whole ligand from the complementarities of its constituent chemical moieties. This means that ligand design strategies must optimize the electrostatic complementarity globally, and not moiety by moiety.     

The relationship between the sequence identities of alpha helical proteins in the PDB and the molecular similarities of their ligands.

This paper considers the relationship between the percentage sequence identities of protein chains and the molecular similarities of the ligands they bind. Among a set of alpha helical proteins from the PDB, it is found that related proteins tend to bind similar ligands. Furthermore, the property of binding similar ligands can be used to define the categories of "like" and "unlike" pairs of protein chains, separated by an approximate cutoff at a sequence identity of, or somewhat above, 45%. Similarly, the property of binding related protein chains can be used to define "low" and "high" similarity pairs of ligand residues, with a cutoff at a Tanimoto score of 0.70. The ligands bound to two "like" protein chains are five times more likely to be of high similarity than would be expected if protein sequence identity and ligand molecular similarity were independent variables. Nonetheless, the nature of the PDB means that it is unclear whether the same conclusions would be reached with a data set representing an unbiased sample of all protein-ligand complexes in a living cell. The construction of an appropriate data set for such a study represents a significant challenge.     

Molecular-scale neural nets: an approach to the self-assembly of molecular networks

It is suggested that molecular self-assembly techniques may be utilised in molecular scale electronic systems. Taking a neural network as an example, it is shown how the affinity of proteins for their complementary ligand could be used to form the molecular scale network essential to such a system. The approach is based on the polymerisation of the tetrameric proteins avidin and streptavidin using biotinylated ligands. By synthesising a range of aromatic bisbiotin ligands, some simple ground rules are established for ensuring complexation of protein and ligand. Based on these findings, a tetrabiotinylated tetrapyridylporphyrin salt (TBPP) was synthesised with a view to forming a two-dimensional protein polymer. Gel chromatography experiments and a UV-visible investigation of solutions containing the protein/TBPP mixture provide strong evidence for complex formation. The specific immobilisation of streptavidin to a monolayer of TBPP formed at the air-water interface is also demonstrated using fluorescence microscopy. Based on the promising results of these initial studies, a number of suggestions are made for the further exploitation of the techniques described herein.     

Structures and Interligand Interaction Pattern of Phosphoramidite Pd Complexes by NMR Spectroscopy: Modulations in Extended Interaction Surfaces as Stereoselection Mode of a Privileged Class of Ligands

During the last decade, phosphoramidites have been established as a so‐called privileged class of ligands in various transition metal catalyses. However, the interactions responsible for their favorable properties have hitherto remained elusive. To address this issue, the formation trends, structural features, and interligand interaction patterns of several trans‐ and cis‐[PdLL′Cl₂] complexes have been investigated by NMR spectroscopy. The energetic contribution of their interligand interactions has been measured experimentally using the supramolecular balance for transition‐metal complexes. The resulting energetics combined with an analysis of the electrostatic potential surfaces reveal that in phosphoramidites not only the aryl groups but the complete (CH)CH₃Ph moieties of the amine side chains form extended quasi‐planar CH‐π and π‐π interaction surfaces. Application of the supramolecular balance has shown that modulations in these extended interaction surfaces cause energetic differences that are relevant to enantioselective catalysis. In addition, the energetics of these interligand interactions are quite independent of the actual structures of the complexes. This is shown by similar formation and aggregation trends of complexes with the same ligand but different structures. The extended quasi‐planar electrostatic interaction surface of the (CH)CH₃Ph moiety explains the known pattern of successful ligand modulation and the substrate specificity of phosphoramidites. Thus, we propose modulations in these extended CH‐π and π‐π interaction areas as a refined stereoselection mode for these ligands. Based on the example of phosphoramidites, this study reveals three general features potentially applicable to various ligands in asymmetric catalysis. First, specific combinations of alkyl and aryl moieties can be used to create extended anisotropic interaction areas. Second, modulations in these interaction surfaces cause energetic differences that are relevant to catalytic applications. Third, bulky substituents with matching complementary interaction surfaces should not only be considered in terms of steric hindrance but also in terms of attractive and repulsive interactions, a feature that may often be underestimated in asymmetric catalysis.