Predicting protein-ligand binding affinities using novel geometrical descriptors and machine-learning methods

Inspired by the concept of knowledge-based scoring functions, a new quantitative structure-activity relationship (QSAR) approach is introduced for scoring protein-ligand interactions. This approach considers that the strength of ligand binding is correlated with the nature of specific ligand/binding site atom pairs in a distance-dependent manner. In this technique, atom pair occurrence and distance-dependent atom pair features are used to generate an interaction score. Scoring and pattern recognition results obtained using Kernel PLS (partial least squares) modeling and a genetic algorithm-based feature selection method are discussed.     

A molecular mechanics approach to modeling protein-ligand interactions: relative binding affinities in congeneric series

We introduce the "Prime-ligand" method for ranking ligands in congeneric series. The method employs a single scoring function, the OPLS-AA/GBSA molecular mechanics/implicit solvent model, for all stages of sampling and scoring. We evaluate the method using 12 test sets of congeneric series for which experimental binding data is available in the literature, as well as the structure of one member of the series bound to the protein. Ligands are "docked" by superimposing a common stem fragment among the compounds in the series using a crystal complex from the Protein Data Bank and sampling the conformational space of the variable region. Our results show good correlation between our predicted rankings and the experimental data for cases in which binding affinities differ by at least 1 order of magnitude. For 11 out of 12 cases, >90% of such ligand pairs could be correctly ranked, while for the remaining case, Factor Xa, 76% of such pairs were correctly ranked. A small number of compounds could not be docked using the current protocol because of the large size of functional groups that could not be accommodated by a rigid receptor. CPU requirements for the method, involving CPU minutes per ligand, are modest compared with more rigorous methods that use similar force fields, such as free energy perturbation. We also benchmark the scoring function using series of ligands bound to the same protein within the CSAR data set. We demonstrate that energy minimization of ligands in the crystal structures is critical to obtain any correlation with experimentally determined binding affinities.     

Biophysical limits of protein-ligand binding

In classic work, Kuntz et al. (Proc. Nat. Acad. Sci. USA1999, 96, 9997-10002) introduced the concept of ligand efficiency. Though that study focused primarily on drug-like molecules, it also showed that metal binding led to the greatest ligand efficiencies. Here, the physical limits of binding are examined across the wide variety of small molecules in the Binding MOAD database. The complexes with the greatest ligand efficiencies share the trait of being small, charged ligands bound in highly charged, well buried binding sites. The limit of ligand efficiency is -1.75 kcal/mol·atom for the protein-ligand complexes within Binding MOAD, and 95% of the set have efficiencies below a "soft limit" of -0.83 kcal/mol·atom. On the basis of buried molecular surface area, the hard limit of ligand efficiency is -117 cal/mol·?(2), which is in surprising agreement with the limit of macromolecule-protein binding. Close examination of the most efficient systems reveals their incredibly high efficiency is dictated by tight contacts between the charged groups of the ligand and the pocket. In fact, a misfit of 0.24 ? in the average contacts inherently decreases the maximum possible efficiency by at least 0.1 kcal/mol·atom.     

A quantum mechanics/molecular mechanics study of the protein-ligand interaction for inhibitors of HIV-1 integrase

Human immunodeficiency virus type-1 integrase (HIV-1 IN) is an essential enzyme for effective viral replication. Diketo acids such as L-731,988 and S-1360 are potent and selective inhibitors of HIV-1 IN. In this study, we used molecular dynamics simulations, within the hybrid quantum mechanics/molecular mechanics (QM/MM) approach, to determine the protein-ligand interaction energy between HIV-1 IN and L-731,988 and 10 of its derivatives and analogues. This hybrid methodology has the advantage that it includes quantum effects such as ligand polarisation upon binding, which can be very important when highly polarisable groups are embedded in anisotropic environments, as for example in metal-containing active sites. Furthermore, an energy decomposition analysis was performed to determine the contributions of individual residues to the enzyme-inhibitor interactions on averaged structures obtained from rather extensive conformational sampling. Analysis of the results reveals first that there is a correlation between protein-ligand interaction energy and experimental strand transfer into human chromosomes and secondly that the Asn-155, Lys-156 and Lys-159 residues and the Mg(2+) ion are crucial to anti-HIV IN activity. These results may explain the available experimental data.     

Fully automated molecular mechanics based induced fit protein-ligand docking method

We describe a method for docking a ligand into a protein receptor while allowing flexibility of the protein binding site. The method employs a multistep procedure that begins with the generation of protein and ligand conformations. An initial placement of the ligand is then performed by computing binding site hotspots. This initial placement is followed by a protein side-chain refinement stage that models protein flexibility. The final step of the process is an energy minimization of the ligand pose in the presence of the rigid receptor. Thus the algorithm models flexibility of the protein at two stages, before and after ligand placement. We validated this method by performing docking and cross docking studies of eight protein systems for which crystal structures were available for at least two bound ligands. The resulting rmsd values of the 21 docked protein-ligand complexes showed values of 2 A or less for all but one of the systems examined. The method has two critical benefits for high throughput virtual screening studies. First, no user intervention is required in the docking once the initial binding site selection has been made in the protein. Second, the initial protein conformation generation needs to be performed only once for a given binding region. Also, the method may be customized in various ways depending on the particular scenario in which dockings are being performed. Each of the individual steps of the method is fully independent making it straightforward to explore different variants of the high level workflow to further improve accuracy and performance.     

Information theory-based scoring function for the structure-based prediction of protein-ligand binding affinity

The development and validation of a new knowledge based scoring function (SIScoreJE) to predict binding energy between proteins and ligands is presented. SIScoreJE efficiently predicts the binding energy between a small molecule and its protein receptor. Protein-ligand atomic contact information was derived from a Non-Redundant Data set (NRD) of over 3000 X-ray crystal structures of protein-ligand complexes. This information was classified for individual "atom contact pairs" (ACP) which is used to calculate the atomic contact preferences. In addition to the two schemes generated in this study we have assessed a number of other common atom-type classification schemes. The preferences were calculated using an information theoretic relationship of joint entropy. Among 18 different atom-type classification schemes "ScoreJE Atom Type set2" (SATs2) was found to be the most suitable for our approach. To test the sensitivity of the method to the inclusion of solvent, Single-body Solvation Potentials (SSP) were also derived from the atomic contacts between the protein atom types and water molecules modeled using AQUARIUS2. Validation was carried out using an evaluation data set of 100 protein-ligand complexes with known binding energies to test the ability of the scoring functions to reproduce known binding affinities. In summary, it was found that a combined SSP/ScoreJE (SIScoreJE) performed significantly better than ScoreJE alone, and SIScoreJE and ScoreJE performed better than GOLD::GoldScore, GOLD::ChemScore, and XScore.     

Estimating protein-ligand binding affinity using high-throughput screening by NMR

Many of today's drug discovery programs use high-throughput screening methods that rely on quick evaluations of protein activity to rank potential chemical leads. By monitoring biologically relevant protein-ligand interactions, NMR can provide a means to validate these discovery leads and to optimize the drug discovery process. NMR-based screens typically use a change in chemical shift or line width to detect a protein-ligand interaction. However, the relatively low throughput of current NMR screens and their high demand on sample requirements generally makes it impractical to collect complete binding curves to measure the affinity for each compound in a large and diverse chemical library. As a result, NMR ligand screens are typically limited to identifying candidates that bind to a protein and do not give any estimate of the binding affinity. To address this issue, a methodology has been developed to rank binding affinities for ligands based on NMR screens that use 1D (1)H NMR line-broadening experiments. This method was demonstrated by using it to estimate the dissociation equilibrium constants for twelve ligands with the protein human serum albumin (HSA). The results were found to give good agreement with previous affinities that have been reported for these same ligands with HSA.     

Top-down ESI-ECD-FT-ICR mass spectrometry localizes noncovalent protein-ligand binding sites

Mass spectrometry (MS) with electrospray ionization (ESI) has the capability to measure and detect noncovalent protein-ligand and protein-protein complexes. However, information on the sites of ligand binding is not easily obtained by the ESI-MS methodology. Electron capture dissociation (ECD) favors cleavage of covalent backbone bonds of protein molecules. We show that this characteristic of ECD translates to noncovalent protein-ligand complexes, as covalent backbone bonds of protein complexes are dissociated, but the noncovalent ligand interaction is retained. For the complex formed from 140-residue, 14.5 kDa alpha-synuclein protein, and one molecule of polycationic spermine (202 Da), ECD generates product ions that retain the protein-spermine noncovalent interaction. Spermine binding is localized to residues 106-138; the ECD data are consistent with previous solution NMR studies. Our studies suggest that ECD mass spectrometry can be used to determine directly the sites of ligand binding to protein targets.     

A searchable database for comparing protein-ligand binding sites for the analysis of structure-function relationships

The rapid expansion of structural information for protein-ligand binding sites is potentially an important source of information in structure-based drug design and in understanding ligand cross reactivity and toxicity. We have developed a large database of ligand binding sites extracted automatically from the Protein Data Bank. This has been combined with a method for calculating binding site similarity based on geometric hashing to create a relational database for the retrieval of site similarity and binding site superposition. It contains an all-against-all comparison of binding sites and holds known protein-ligand binding sites, which are made accessible to data mining. Here we demonstrate its utility in two structure-based applications: in determining site similarity and in aiding the derivation of a receptor-based pharmacophore model. The database is available from http://www.bioinformatics.leeds.ac.uk/sb/.     

Quantum mechanical map for protein-ligand binding with application to beta-trypsin/benzamidine complex

We report full ab initio Hartree-Fock calculation to compute quantum mechanical interaction energies for beta-trypsin/benzamidine binding complex. In this study, the full quantum mechanical ab initio energy calculation for the entire protein complex with 3238 atoms is made possible by using a recently developed MFCC (molecular fractionation with conjugate caps) approach in which the protein molecule is decomposed into amino acid-based fragments that are properly capped. The present MFCC ab initio calculation enables us to obtain an "interaction spectrum" that provides detailed quantitative information on protein-ligand binding at the amino acid levels. These detailed information on individual residue-ligand interaction gives a quantitative molecular insight into our understanding of protein-ligand binding and provides a guidance to rational design of potential inhibitors of protein targets.     

A water-swap reaction coordinate for the calculation of absolute protein-ligand binding free energies

The accurate prediction of absolute protein-ligand binding free energies is one of the grand challenge problems of computational science. Binding free energy measures the strength of binding between a ligand and a protein, and an algorithm that would allow its accurate prediction would be a powerful tool for rational drug design. Here we present the development of a new method that allows for the absolute binding free energy of a protein-ligand complex to be calculated from first principles, using a single simulation. Our method involves the use of a novel reaction coordinate that swaps a ligand bound to a protein with an equivalent volume of bulk water. This water-swap reaction coordinate is built using an identity constraint, which identifies a cluster of water molecules from bulk water that occupies the same volume as the ligand in the protein active site. A dual topology algorithm is then used to swap the ligand from the active site with the identified water cluster from bulk water. The free energy is then calculated using replica exchange thermodynamic integration. This returns the free energy change of simultaneously transferring the ligand to bulk water, as an equivalent volume of bulk water is transferred back to the protein active site. This, directly, is the absolute binding free energy. It should be noted that while this reaction coordinate models the binding process directly, an accurate force field and sufficient sampling are still required to allow for the binding free energy to be predicted correctly. In this paper we present the details and development of this method, and demonstrate how the potential of mean force along the water-swap coordinate can be improved by calibrating the soft-core Coulomb and Lennard-Jones parameters used for the dual topology calculation. The optimal parameters were applied to calculations of protein-ligand binding free energies of a neuraminidase inhibitor (oseltamivir), with these results compared to experiment. These results demonstrate that the water-swap coordinate provides a viable and potentially powerful new route for the prediction of protein-ligand binding free energies.     

Thermodynamics of buried water clusters at a protein-ligand binding interface

The structure of the complex of cyclophilin A (CypA) with cyclosporin A (CsA, 1) shows a cluster of four water molecules buried at the binding interface, which is rearranged when CsA is replaced by (5-hydroxynorvaline)-2-cyclosporin (2). The thermodynamic contributions of each bound water molecule in the two complexes are explored with the inhomogeneous fluid solvation theory and molecular dynamics simulations. Water (WTR) 133 in complex 1 contributes little to the binding affinity, while WTR6 and 7 in complex 2 play an essential role in mediating protein-ligand binding with a hydrogen bond network. The calculations reveal that the rearrangement of the water molecules contributes favorably to the binding affinity, even though one of them is displaced going from ligand 1 to 2. Another favorable contribution comes from the larger protein-ligand interactions of ligand 2. However, these favorable contributions are not sufficient to overcome the unfavorable desolvation free energy change and the conformational entropy of the hydroxylpropyl group of ligand 2 in the complex, leading to a lower binding affinity of ligand 2. These physical insights may be useful in the development of improved scoring functions for binding affinity prediction.     

Molecular mechanics force-field development for amino acid zwitterions

Understanding the conformational flexibility of amino acid zwitterions (ZWs) and their associated conformational energies is crucial for predicting their interactions in biological systems. Gas-phase ab initio calculations of ZWs are intractable. Molecular mechanics (MM), on the other hand, is able to handle large systems but lacks the necessary force field parameters to model ZWs. To develop force field parameters that are able to correctly model ZW geometries and energetics we used a novel combinatorial approach: amino acid ZWs were broken down structurally into key functional components, which were parameterized separately. M?ller-Plesset second-order perturbation calculations on small carboxylates, on the glycine cation, and on novel hydrogen bonded systems, coupled with available experimental data, were used to generate MM3(2000) ZW parameters (Allinger N. L.; Yuh, Y. H.; Lii, J.-H. J Am Chem Soc 1989, 111, 8551). The MM3 results from this combinatorial approach gave geometries that are in good agreement with neutron diffraction experiments, plus their frequencies and energies appear to be reasonably modeled. Current limitations and future development of MM force fields are discussed briefly.     

CSAR scoring challenge reveals the need for new concepts in estimating protein-ligand binding affinity

The dG prediction accuracy by the Lead Finder docking software on the CSAR test set was characterized by R(2)=0.62 and rmsd=1.93 kcal/mol, and the method of preparation of the full-atom structures of the test set did not significantly affect the resulting accuracy of predictions. The primary factors determining the correlation between the predicted and experimental values were the van der Waals interactions and solvation effects. Those two factors alone accounted for R(2)=0.50. The other factors that affected the accuracy of predictions, listed in the order of decreasing importance, were the change of ligand's internal energy upon binding with protein, the electrostatic interactions, and the hydrogen bonds. It appears that those latter factors contributed to the independence of the prediction results from the method of full-atom structure preparation. Then, we turned our attention to the other factors that could potentially improve the scoring function in order to raise the accuracy of the dG prediction. It turned out that the ligand-centric factors, including Mw, cLogP, PSA, etc. or protein-centric factors, such as the functional class of protein, did not improve the prediction accuracy. Following that, we explored if the weak molecular interactions such as X-H...Ar, X-H...Hal, CO...Hal, C-H...X, stacking and π-cationic interactions (where X is N or O), that are generally of interest to the medicinal chemists despite their lack of proper molecular mechanical parametrization, could improve dG prediction. Our analysis revealed that out of these new interactions only CO...Hal is statistically significant for dG predictions using Lead FInder scoring function. Accounting for the CO...Hal interaction resulted in the reduction of the rmsd from 2.19 to 0.69 kcal/mol for the corresponding structures. The other weak interaction factors were not statistically significant and therefore irrelevant to the accuracy of dG prediction. On the basis of our findings from our participation in the CSAR scoring challenge we conclude that a significant increase of accuracy predictions necessitates breakthrough scoring approaches. We anticipate that the explicit accounting for water molecules, protein flexibility, and a more thermodynamically accurate method of dG calculation rather than single point energy calculation may lead to such breakthroughs.     

A general mass spectrometry-based assay for the quantitation of protein-ligand binding interactions in solution

A new method that utilizes matrix-assisted laser desorption/ionization (MALDI) mass spectrometry and exploits the hydrogen/deuterium (H/D) exchange properties of proteins was developed for measuring the thermodynamic properties of protein-ligand complexes in solution. Dissociation constants (Kd values) determined by the method for five model protein-ligand complexes that included those with small molecules, nucleic acids, peptides, and other proteins were generally in good agreement with Kd values measured by conventional methods. Important experimental advantages of the described method over existing methods include: the ability to make measurements in a high-throughput and automated fashion, the ability to make measurements using only picomole quantitities of protein, and the ability to analyze either purified or unpurified protein-ligand complexes.     

Using silico methods predicting ligands for orphan GPCRs

The G-protein coupled receptor (GPCR) superfamily is one of the most important drug target classes for the pharmaceutical industry. The completion of the human genome project has revealed that there are more than 300 potential GPCR targets of interest. The identification of their natural ligands can gain significant insights into regulatory mechanisms of cellular signaling networks and provide unprecedented opportunities for drug discovery. Much effort has been directed towards the GPCR ligand discovery study by both academic institutions and pharmaceutical industries. However, the endogenous ligands still remain unknown for about 150 GPCRs in the human genome. It is necessary to develop new strategies to predict candidate ligands for these so-called orphan receptors. Computational techniques are playing an increasingly important role in finding and validating novel ligands for orphan GPCRs (oGPCRs). In this paper, we focus on recent development in applying bioinformatics approaches for the discovery of GPCR ligands. In addition, some of the data resources for ligand identification are also provided.     

Methods for the prediction of protein-ligand binding sites for structure-based drug design and virtual ligand screening

Structure Based Drug Design (SBDD) is a computational approach to lead discovery that uses the three-dimensional structure of a protein to fit drug-like molecules into a ligand binding site to modulate function. Identifying the location of the binding site is therefore a vital first step in this process, restricting the search space for SBDD or virtual screening studies. The detection and characterisation of functional sites on proteins has increasingly become an area of interest. Structural genomics projects are increasingly yielding protein structures with unknown functions and binding sites. Binding site prediction was pioneered by pocket detection, since the binding site is often found in the largest pocket. More recent methods involve phylogenetic analysis, identifying structural similarity with proteins of known function and identifying regions on the protein surface with a potential for high binding affinity. Binding site prediction has been used in several SBDD projects and has been incorporated into several docking tools. We discuss different methods of ligand binding site prediction, their strengths and weaknesses, and how they have been used in SBDD.     

Study of protein-ligand binding effects by direct chromatographic on-column injection.

The direct zonal on-column injection method was applied to a high-performance liquid chromatographic study of pollutant-protein binding interactions in solution. The protein and the protein-ligand complex are excluded on the basis of the size from the diol support, and the free ligand penetrates into the pores and is more retained. The pattern of the ligand elution profile depends on the protein-ligand dissociation constant. This effect was quantitatively analysed by developing a numerical simulation algorithm in which the column is divided into slices of given thickness. The column length, flow-rate and shape of the injection signal are given as input parameters. A global dispersion coefficient accounts for peak broadening. A rapid equilibrium is assumed with the hypothesis that a monovalent ligand interacts with a single binding site on the protein. The interaction of bovine serum albumin with pentachlorophenol was studied, and an apparent dissociation constant for the protein-ligand complex was determined by fitting the theoretical profile to the experimental one. The effect of the acetonitrile content in the solvent was studied. An important decrease of the dissociation constant is observed that affects the chromatographic elution pattern.