Prediction of SAMPL-1 hydration free energies using a continuum electrostatics-dispersion model.

The SAMPL-1 hydration free energy blind prediction challenge data set includes 63 compounds that are more chemically diverse, polyfunctional, drug-like, and with examples of transfer free energies and molecular weights larger than ever before seen in previously tabulated data sets of neutral compounds. For the prospective SAMPL-1 study, we employed a continuum model including a boundary element solution of the Poisson equation to describe electrostatic solvation, a molecular surface area-based cost of cavity formation in water, and a continuum Lennard-Jones potential to account for dispersion-repulsion solute-solvent effects. For the latter contribution, continuum van der Waals atom-type coefficients were calibrated and validated on previously available hydration data sets. In the prospective study, this continuum hydration model yielded SAMPL-1 predictions highly correlated with experimental data, albeit with a slope of slightly above 0.5, suggesting a valid model but with a systematic error. Analysis of the major outliers, all overestimating the experimental hydration data, highlights a common structural theme as a possible cause of the prediction errors: densely polar and hydrogen-bond-capable structures, featuring primarily substituted (sulfon)amide groups, often in conjugated systems. By examining analog pairs within the SAMPL-1 data set, it was also noted that certain solvation trends are captured neither by chemical sense nor by our hydration model, which seem too additive. A retrospective analysis of model transferability between hydration data sets as a function of its parameters and complexity indicates that the electrostatic component of the model is fairly transferrable across data sets, but the nonelectrostatic terms are less so. For the chemical space covered in SAMPL-1, absolute prediction errors indicate that the simpler transferrable electrostatics-only model outperforms the more complex model including cavity and continuum dispersion terms. Possible directions to further improve this continuum hydration model are proposed.

Solvated interaction energy (SIE) for scoring protein-ligand binding affinities. 1. Exploring the parameter space.

We present a binding free energy function that consists of force field terms supplemented by solvation terms. We used this function to calibrate the solvation model along with the binding interaction terms in a self-consistent manner. The motivation for this approach was that the solute dielectric-constant dependence of calculated hydration gas-to-water transfer free energies is markedly different from that of binding free energies (J. Comput. Chem. 2003, 24, 954). Hence, we sought to calibrate directly the solvation terms in the context of a binding calculation. The five parameters of the model were systematically scanned to best reproduce the absolute binding free energies for a set of 99 protein-ligand complexes. We obtained a mean unsigned error of 1.29 kcal/mol for the predicted absolute binding affinity in a parameter space that was fairly shallow near the optimum. The lowest errors were obtained with solute dielectric values of Din = 20 or higher and scaling of the intermolecular van der Waals interaction energy by factors ranging from 0.03 to 0.15. The high apparent Din and strong van der Waals scaling may reflect the anticorrelation of the change in solvated potential energy and configurational entropy, that is, enthalpy-entropy compensation in ligand binding (Biophys. J. 2004, 87, 3035-3049). Five variations of preparing the protein-ligand data set were explored in order to examine the effect of energy refinement and the presence of bound water on the calculated results. We find that retaining water in the final protein structure used for calculating the binding free energy is not necessary to obtain good results; that is the continuum solvation model is sufficient. Virtual screening enrichment studies on estrogen receptor and thymidine kinase showed a good ability of the binding free energy function to recover true hits in a collection of decoys.

Impaired helix 12 dynamics due to proline 892 substitutions in the androgen receptor are associated with complete androgen insensitivity.

Structural studies of the ligand-binding domain (LBD) of several steroid receptors have revealed that the dynamic properties of the C-terminal helix 12 (H12) are the major determinant of the activation mode of these receptors. H12 exhibits high mobility and different conformations in the absence of ligand. Upon ligand binding, H12 is stabilized in a precise position to seal the ligand-binding pocket and finalize the assembly of the activation function (AF-2) domain. In this study, we investigated the role of the conserved proline 892 of the androgen receptor (AR) in directing the dynamic location and orientation of the AR-H12. We used a combined approach including kinetic and biochemical assays with molecular dynamic simulations to analyze two substitutions (P892A and P892L) identified in individuals with complete androgen insensitivity syndrome. Our analyses revealed distinct mechanisms by which these substitutions impair H12 function resulting in severely defective receptors. The AR-P892A receptor exhibited reduced ligand binding and transactivational potential because of an increased flexibility in H12. The AR-P892L substitution renders the receptor inactive due to a distorted, unstructured and misplaced H12. To confirm the mutants' inability to stabilize H12 in an active position, we have developed a novel in vivo assay to evaluate the accessibility of the H12-docking site on the AR-LBD surface. An extrinsic AR-H12 peptide was able to interact with wild-type and mutant LBDs in the absence of ligand. Ligand-induced proper positioning of the intrinsic H12 of wild-type AR prevented these interactions, whereas the misplacement of the mutants' H12 did not. Proline at this position may be critical for H12 dynamics not only in the AR, but also in other nuclear receptors where this proline is conserved.

Improved mapping of protein binding sites.

Computational mapping methods place molecular probes--small molecules or functional groups--on a protein surface in order to identify the most favorable binding positions by calculating an interaction potential. Mapping is an important step in a number of flexible docking and drug design algorithms. We have developed improved algorithms for mapping protein surfaces using small organic molecules as molecular probes. The calculations reproduce the binding of eight organic solvents to lysozyme as observed by NMR, as well as the binding of four solvents to thermolysin, in good agreement with x-ray data. Application to protein tyrosine phosphatase 1B shows that the information provided by the mapping can be very useful for drug design. We also studied why the organic solvents bind in the active site of proteins, in spite of the availability of alternative pockets that can very tightly accommodate some of the probes. A possible explanation is that the binding in the relatively large active site retains a number of rotational states, and hence leads to smaller entropy loss than the binding elsewhere else. Indeed, the mapping reveals that the clusters of the ligand molecules in the protein's active site contain different rotational-translational conformers, which represent different local minima of the free energy surface. In order to study the transitions between different conformers, reaction path and molecular dynamics calculations were performed. Results show that most of the rotational states are separated by low free energy barriers at the experimental temperature, and hence the entropy of binding in the active site is expected to be high.

Identification of substrate binding sites in enzymes by computational solvent mapping.

Enzyme structures determined in organic solvents show that most organic molecules cluster in the active site, delineating the binding pocket. We have developed algorithms to perform solvent mapping computationally, rather than experimentally, by placing molecular probes (small molecules or functional groups) on a protein surface, and finding the regions with the most favorable binding free energy. The method then finds the consensus site that binds the highest number of different probes. The probe-protein interactions at this site are compared to the intermolecular interactions seen in the known complexes of the enzyme with various ligands (substrate analogs, products, and inhibitors). We have mapped thermolysin, for which experimental mapping results are also available, and six further enzymes that have no experimental mapping data, but whose binding sites are well characterized. With the exception of haloalkane dehalogenase, which binds very small substrates in a narrow channel, the consensus site found by the mapping is always a major subsite of the substrate-binding site. Furthermore, the probes at this location form hydrogen bonds and non-bonded interactions with the same residues that interact with the specific ligands of the enzyme. Thus, once the structure of an enzyme is known, computational solvent mapping can provide detailed and reliable information on its substrate-binding site. Calculations on ligand-bound and apo structures of enzymes show that the mapping results are not very sensitive to moderate variations in the protein coordinates.

Algorithms for computational solvent mapping of proteins.

Computational mapping methods place molecular probes (small molecules or functional groups) on a protein surface to identify the most favorable binding positions by calculating an interaction potential. We have developed a novel computational mapping program called CS-Map (computational solvent mapping of proteins), which differs from earlier mapping methods in three respects: (i) it initially moves the ligands on the protein surface toward regions with favorable electrostatics and desolvation, (ii) the final scoring potential accounts for desolvation, and (iii) the docked ligand positions are clustered, and the clusters are ranked on the basis of their average free energies. To understand the relative importance of these factors, we developed alternative algorithms that use the DOCK and GRAMM programs for the initial search. Because of the availability of experimental solvent mapping data, lysozyme and thermolysin are considered as test proteins. Both DOCK and GRAMM speed up the initial search, and the combined algorithms yield acceptable mapping results. However, the DOCK-based approaches place the consensus site farther from its experimentally determined position than CS-Map, primarily because of the lack of a solvation term in the initial search. The GRAMM-based program also finds the correct consensus site for thermolysin. We conclude that good sampling is the most important requirement for successful mapping, but accounting for desolvation and clustering of ligand positions also help to reduce the number of false positives.

Computational mapping identifies the binding sites of organic solvents on proteins.

Computational mapping places molecular probes--small molecules or functional groups--on a protein surface to identify the most favorable binding positions. Although x-ray crystallography and NMR show that organic solvents bind to a limited number of sites on a protein, current mapping methods result in hundreds of energy minima and do not reveal why some sites bind molecules with different sizes and polarities. We describe a mapping algorithm that explains the origin of this phenomenon. The algorithm has been applied to hen egg-white lysozyme and to thermolysin, interacting with eight and four different ligands, respectively. In both cases the search finds the consensus site to which all molecules bind, whereas other positions that bind only certain ligands are not necessarily found. The consensus sites are pockets of the active site, lined with partially exposed hydrophobic residues and with a number of polar residues toward the edge. These sites can accommodate each ligand in a number of rotational states, some with a hydrogen bond to one of the nearby donor/acceptor groups. Specific substrates and/or inhibitors of hen egg-white lysozyme and thermolysin interact with the same side chains identified by the mapping, but form several hydrogen bonds and bind in unique orientations.