Evaluation of protein docking predictions using Hex 3.1 in CAPRI rounds 1 and 2

This article describes and reviews our efforts using Hex 3.1 to predict the docking modes of the seven target protein-protein complexes presented in the CAPRI (Critical Assessment of Predicted Interactions) blind docking trial. For each target, the structure of at least one of the docking partners was given in its unbound form, and several of the targets involved large multimeric structures (e.g., Lactobacillus HPr kinase, hemagglutinin, bovine rotavirus VP6). Here we describe several enhancements to our original spherical polar Fourier docking correlation algorithm. For example, a novel surface sphere smothering algorithm is introduced to generate multiple local coordinate systems around the surface of a large receptor molecule, which may be used to define a small number of initial ligand-docking orientations distributed over the receptor surface. High-resolution spherical polar docking correlations are performed over the resulting receptor surface patches, and candidate docking solutions are refined by using a novel soft molecular mechanics energy minimization procedure. Overall, this approach identified two good solutions at rank 5 or less for two of the seven CAPRI complexes. Subsequent analysis of our results shows that Hex 3.1 is able to place good solutions within a list of 相似文献   

Mapping mutations in influenza A virus resistant to norakin

To elucidate the mode of action of norakin against influenza A virus we sequenced the hemagglutinin gene of 11 norakin-resistant mutants. Resistance was coupled with 1-3 amino acid exchanges. The majority of mutations was localized in the HA2 polypeptide and was mostly associated with changes in charge or polarity of the amino acids. The amino acid substitutions are discussed in the context of the 3D structure of X31 hemagglutinin considered to be representative of the influenza hemagglutinins. Most of the mutations appear to destabilize the pH 7.0 structure by distorting or destroying hydrogen bonds as well as salt-bridges which are responsible for intra- and intersubunit contacts, while others destabilize the location of the fusion peptide, facilitating conformational changes in the presence of the inhibitor.     

Correlation of the sweetness of variants of the protein brazzein with patterns of hydrogen bonds detected by NMR spectroscopy

In sequence-function investigations, approaches are needed for rapidly screening protein variants for possible changes in conformation. Recent NMR methods permit direct detection of hydrogen bonds through measurements of scalar couplings that traverse hydrogen bonds (trans-hydrogen bond couplings). We have applied this approach to screen a series of five single site mutants of the sweet protein brazzein with altered sweetness for possible changes in backbone hydrogen bonding with respect to wild-type. Long range, three-dimensional data correlating connectivities among backbone 1HN, 15N, and 13C' atoms were collected from the six brazzein proteins labeled uniformly with carbon-13 and nitrogen-15. In wild-type brazzein, this approach identified 17 backbone hydrogen bonds. In the mutants, altered magnitudes of the couplings identified hydrogen bonds that were strengthened or weakened; missing couplings identified hydrogen bonds that were broken, and new couplings indicated the presence of new hydrogen bonds. Within the series of brazzein mutants investigated, a pattern was observed between sweetness and the integrity of particular hydrogen bonds. All three "sweet" variants exhibited the same pattern of hydrogen bonds, whereas all three "non-sweet" variants lacked one hydrogen bond at the middle of the alpha-helix, where it is kinked, and one hydrogen bond in the middle of beta-strands II and III, where they are twisted. Two of the non-sweet variants lack the hydrogen bond connecting the N and C termini. These variants showed greater mobility in the N- and C-terminal regions than wild-type brazzein.     

Contribution of hydrogen bonds to protein stability

Our goal was to gain a better understanding of the contribution of the burial of polar groups and their hydrogen bonds to the conformational stability of proteins. We measured the change in stability, Δ(ΔG), for a series of hydrogen bonding mutants in four proteins: villin headpiece subdomain (VHP) containing 36 residues, a surface protein from Borrelia burgdorferi (VlsE) containing 341 residues, and two proteins previously studied in our laboratory, ribonucleases Sa (RNase Sa) and T1 (RNase T1). Crystal structures were determined for three of the hydrogen bonding mutants of RNase Sa: S24A, Y51F, and T95A. The structures are very similar to wild type RNase Sa and the hydrogen bonding partners form intermolecular hydrogen bonds to water in all three mutants. We compare our results with previous studies of similar mutants in other proteins and reach the following conclusions. (1) Hydrogen bonds contribute favorably to protein stability. (2) The contribution of hydrogen bonds to protein stability is strongly context dependent. (3) Hydrogen bonds by side chains and peptide groups make similar contributions to protein stability. (4) Polar group burial can make a favorable contribution to protein stability even if the polar groups are not hydrogen bonded. (5) The contribution of hydrogen bonds to protein stability is similar for VHP, a small protein, and VlsE, a large protein.     

Thermodynamic analysis of micellization in PEO–PPO–PEO block copolymer solutions from the hydrogen bonding point of view

A thermodynamic approach is suggested to study the micellization mechanism of poly(ethylene oxide)–poly(propylene oxide)–poly(ethylene oxide) (PEO–PPO–PEO) triblock copolymers solutions from the hydrogen bonding point of view. Using Flory–Huggins theory, an association model is presented to describe hydrogen bonded (HB) chains, which are bridged by hydrogen bonds between water molecules and segments of the copolymer. The entropic change due to hydrogen bonding is formulated and the individual contribution of EO–water (EO–W) and PO–water (PO–W) hydrogen bonding to the micellization are derived respectively. Fourier transform infrared (FTIR) spectroscopy is applied to obtain the information of hydrogen bonds. During the temperature-dependent micellization of P105 block copolymer solutions, rapid disruption of PO–W hydrogen bonds is observed by FTIR and the calculated entropy relating to PO–W hydrogen bonds increases drastically compared with that of EO–W hydrogen bonds. The results demonstrate that PO–W hydrogen bonds play a dominant role in micellization.     

Computational studies of H5N1 hemagglutinin binding with SA-alpha-2, 3-Gal and SA-alpha-2, 6-Gal

For influenza H5N1 hemagglutinin, a switch from SA-alpha-2, 3-Gal to SA-alpha-2, 6-Gal receptor specificity is a critical step leading to the conversion from avian-to-human to human-to-human infection. Therefore, the understanding of the binding modes of SA-alpha-2, 3-Gal and SA-alpha-2, 6-Gal to H5N1 hemagglutinin will be very important for the examination of possible mutations needed for going from an avian to a human flu virus. Based on the available H5N1 hemagglutinin crystal structure, the binding profiles between H5N1 hemagglutinin and two saccharide ligands, SA-alpha-2, 3-Gal and SA-alpha-2, 6-Gal, were investigated by ab initio quantum mechanics, molecular docking, molecular mechanics, and molecular dynamics simulations. It was found that SA-alpha-2, 3-Gal has strong multiple hydrophobic and hydrogen bond interactions in its trans conformation with H5N1 hemagglutinin, whereas the SA-alpha-2, 6-Gal only shows weak interactions in a different conformation (cis type).     

Strong and weak hydrogen bonds in drug-DNA complexes: A statistical analysis

A statistical analysis of strong and weak hydrogen bonds in the minor groove of DNA was carried out for a set of 70 drug-DNA complexes. The terms ‘strong’ and ‘weak’ pertain to the inherent strengths and weakness of the donor and acceptor fragments rather than to any energy considerations. The dataset was extracted from the protein data bank (PDB). The analysis was performed with an in-house software, hydrogen bond analysis tool (HBAT). In addition to strong hydrogen bonds such as O—H⋯O and N—H⋯O, the ubiquitous presence of weak hydrogen bonds such as C—H⋯O is implicated in molecular recognition. On an average, there are 1.4 weak hydrogen bonds for every strong hydrogen bond. For both categories of interaction, the N(3) of purine and the O(2) of pyrimidine are favoured acceptors. Donor multifurcation is common with the donors generally present in the drug molecules, and shared by hydrogen bond acceptors in the minor groove. Bifurcation and trifurcation are most commonly observed. The metrics for strong hydrogen bonds are consistent with established trends. The geometries are variable for weak hydrogen bonds. A database of recognition geometries for 26 literature amidinium-based inhibitors of Human African Trypanosomes (HAT) was generated with a docking study using seven inhibitors which occur in published crystal structures included in the list of 70 complexes mentioned above, and 19 inhibitors for which the drug-DNA complex crystal structures are unknown. The virtual geometries so generated correlate well with published activities for these 26 inhibitors, justifying our assumption that strong and weak hydrogen bonds are optimized in the active site.     

Contribution of a single heavy chain residue to specificity of an anti-digoxin monoclonal antibody.

Two distinct spontaneous variants of the murine anti-digoxin hybridoma 26-10 were isolated by fluorescence-activated cell sorting for reduced affinity of surface antibody for antigen. Nucleotide and partial amino acid sequencing of the variant antibody variable regions revealed that 1 variant had a single amino acid substitution: Lys for Asn at heavy chain position 35. The second variant antibody had 2 heavy chain substitutions: Tyr for Asn at position 35, and Met for Arg at position 38. Mutagenesis experiments confirmed that the position 35 substitutions were solely responsible for the markedly reduced affinity of both variant antibodies. Several mutants with more conservative position 35 substitutions were engineered to ascertain the contribution of Asn 35 to the binding of digoxin to antibody 26-10. Replacement of Asn with Gln reduced affinity for digoxin 10-fold relative to the wild-type antibody, but maintained wild-type fine specificity for cardiac glycoside analogues. All other substitutions (Val, Thr, Leu, Ala, and Asp) reduced affinity by at least 90-fold and caused distinct shifts in fine specificity. The Ala mutant demonstrated greatly increased relative affinities for 16-acetylated haptens and haptens with a saturated lactone. The X-ray crystal structure of the 26-10 Fab in complex with digoxin (Jeffrey PD et al., 1993, Proc Natl Acad Sci USA 90:10310-10314) reveals that the position 35 Asn contacts hapten and forms hydrogen bonds with 2 other contact residues. The reductions in affinity of the position 35 mutants for digoxin are greater than expected based upon the small hapten contact area provided by the wild-type Asn. We therefore performed molecular modeling experiments which suggested that substitution of Gln or Asp can maintain these hydrogen bonds whereas the other substituted side chains cannot. The altered binding of the Asp mutant may be due to the introduction of a negative charge. The similarities in binding of the wild-type and Gln-mutant antibodies, however, suggest that these hydrogen bonds are important for maintaining the architecture of the binding site and therefore the affinity and specificity of this antibody. The Ala mutant eliminates the wild-type hydrogen bonding, and molecular modeling suggests that the reduced side-chain volume also provides space that can accommodate a congener with a 16-acetyl group or saturated lactone, accounting for the altered fine specificity of this antibody.     

Proposed lead molecules against Hemagglutinin of avian influenza virus (H5N1)

Human infection with avian influenza H5N1 is an emerging infectious disease characterized by respiratory symptoms and a high fatality rate. Hemagglutinin and neuraminidase are the two surface proteins responsible for infection by influenza virus. Till date, neuraminidase has been the major target for antiviral drugs. In the present study we chose hemagglutinin protein as it mediates the binding of the virus to target cells through sialic acid residues on the host cell-surface. Hemagglutinin of H5 avian influenza (PDB ID: 1JSN) was used as the receptor protein. Ligands were generated by structure-based de novo approach and virtual screening of ZINC database. A total of 11,104 conformers were generated and docked into the receptor binding site using 'High Throughput Virtual Screening'. We proposed potential lead molecules against the receptor binding site of hemagglutinin based on the results obtained from in silico docking and hydrogen bond interaction between the ligand and the 1JSN protein molecule. We found sialic acid derivative 1 to be the lead molecules amongst the ligands generated by structure based de novo approach. However the molecules obtained from ZINC database were showing better docking scores as well as conserved hydrogen bond interactions. Thus we proposed ZINC00487720 and ZINC00046810 as potential lead molecules that could be used as an inhibitor to the receptor binding site of hemagglutinin. They could now be studied in vivo to validate the in silico results.     

Completely buried, non-ion-paired glutamic acid contributes favorably to the conformational stability of pyrrolidone carboxyl peptidases from hyperthermophiles

Pyrrolidone carboxyl peptidases (PCPs) from hyperthermophiles have a structurally conserved and completely buried Glu192 in the hydrophobic core; in contrast, the corresponding residue in the mesophile protein is a hydrophobic residue, Ile. Does the buried ionizable residue contribute to stabilization or destabilization of hyperthermophile PCPs? To elucidate the role of the buried glutamic acid in stabilizing PCP from hyperthermophiles, we constructed five Glu192 mutants of PCP-0SH (C142S/C188S, Cys-free double mutant of PCP) from Pyrococcus furiosus and examined their thermal and pH-induced unfolding and crystal structures and compared them with those of PCP-0SH. The stabilities of apolar (E192A/I/V) and polar (E192D/Q) mutants were less than PCP-0SH at acidic pH values. In the alkaline region, the mutant proteins, except for E192D, were more stable than PCP-0SH. The thermal stability data and theoretical calculations indicated an apparent pKa value > or = 7.3 for Glu192. Present results confirmed that the protonated Glu192 in PCP-0SH forms strong hydrogen bonds with the carbonyl oxygen and peptide nitrogen of Pro168. New intermolecular hydrogen bonds in the E --> A/D mutants were formed by a water molecule introduced into the cavity created around position 192, whereas the hydrogen bonds disappeared in the E --> I/V mutants. Structure-based empirical stability of mutant proteins was in good agreement with the experimental results. The results indicated that (1) completely buried Glu192 contributes to the stabilization of PCP-0SH because of the formation of strong intramolecular hydrogen bonds and (2) the hydrogen bonds by the nonionized and buried Glu can contribute more than the burial of hydrophobic groups to the conformational stability of proteins.     

Antigenic determinants of measles virus hemagglutinin associated with neurovirulence.

The biological activity of monoclonal antibodies specific for the hemagglutinin protein of measles virus strain CAM recognizing six epitope groups according to their binding properties to measles virus strain CAM/R401 was investigated in vivo in our rat model of measles encephalitis. When injected intraperitoneally into measles virus-infected suckling rats, some monoclonal antibodies modified the disease process and prevented the necrotizing encephalopathy seen in untreated animals. The analysis of measles virus brain isolates revealed emergence of variants that resisted neutralization with the passively transferred selecting monoclonal antibody but not with other monoclonal antibodies. Monoclonal antibody escape mutants were also isolated in vitro, and their neurovirulence varied in the animal model. Sequence data from the hemagglutinin gene of measles virus localize a major antigenic surface determinant of the hemagglutinin protein between amino acid residues 368 and 396, which may be functionally important for neurovirulence. The data indicate that the interaction of antibodies with the measles virus H protein plays an important role in the selection of neurovirulent variants. These variants have biological properties different from those of the parent CAM virus.     

PRL-Dock: protein-ligand docking based on hydrogen bond matching and probabilistic relaxation labeling

Protein-ligand docking is widely applied to structure-based virtual screening for drug discovery. This article presents a novel docking technique, PRL-Dock, based on hydrogen bond matching and probabilistic relaxation labeling. It deals with multiple hydrogen bonds and can match many acceptors and donors simultaneously. In the matching process, the initial probability of matching an acceptor with a donor is estimated by an efficient scoring function and the compatibility coefficients are assigned according to the coexisting condition of two hydrogen bonds. After hydrogen bond matching, the geometric complementarity of the interacting donor and acceptor sites is taken into account for displacement of the ligand. It is reduced to an optimization problem to calculate the optimal translation and rotation matrixes that minimize the root mean square deviation between two sets of points, which can be solved using the Kabsch algorithm. In addition to the van der Waals interaction, the contribution of intermolecular hydrogen bonds in a complex is included in the scoring function to evaluate the docking quality. A modified Lennard-Jones 12-6 dispersion-repulsion term is used to estimate the van der Waals interaction to make the scoring function fairly "soft" so that ligands are not heavily penalized for small errors in the binding geometry. The calculation of this scoring function is very convenient. The evaluation is carried out on 278 rigid complexes and 93 flexible ones where there is at least one intermolecular hydrogen bond. The experiment results of docking accuracy and prediction of binding affinity demonstrate that the proposed method is highly effective.     

Free energy simulations and MM-PBSA analyses on the affinity and specificity of steroid binding to antiestradiol antibody

Antiestradiol antibody 57-2 binds 17beta-estradiol (E2) with moderately high affinity (K(a) = 5 x 10(8) M(-1)). The structurally related natural estrogens estrone and estriol as well synthetic 17-deoxy-estradiol and 17alpha-estradiol are bound to the antibody with 3.7-4.9 kcal mol(-1) lower binding free energies than E2. Free energy perturbation (FEP) simulations and the molecular mechanics-Poisson-Boltzmann surface area (MM-PBSA) method were applied to investigate the factors responsible for the relatively low cross-reactivity of the antibody with these four steroids, differing from E2 by the substituents of the steroid D-ring. In addition, computational alanine scanning of the binding site residues was carried out with the MM-PBSA method. Both the FEP and MM-PBSA methods reproduced the experimental relative affinities of the five steroids in good agreement with experiment. On the basis of FEP simulations, the number of hydrogen bonds formed between the antibody and steroids, which varied from 0 to 3 in the steroids studied, determined directly the magnitude of the steroid-antibody interaction free energies. One hydrogen bond was calculated to contribute about 3 kcal mol(-1) to the interaction energy. Because the relative binding free energies of estrone (two antibody-steroid hydrogen bonds), estriol (three hydrogen bonds), 17-deoxy-estradiol (no hydrogen bonds), and 17alpha-estradiol (two hydrogen bonds) are close to each other and clearly lower than that of E2 (three hydrogen bonds), the water-steroid interactions lost upon binding to the antibody make an important contribution to the binding free energies. The MM-PBSA calculations showed that the binding of steroids to the antiestradiol antibody is driven by van der Waals interactions, whereas specificity is solely due to electrostatic interactions. In addition, binding of steroids to the antiestradiol antibody 57-2 was compared to the binding to the antiprogesterone antibody DB3 and antitestosterone antibody 3-C4F5, studied earlier with the MM-PBSA method.     

Investigation of the stereoselectivity of an anti-amino acid antibody using molecular modeling and ligand docking

The structure of the binding site of the stereoselective anti-D-amino acid antibody 67.36 was modeled utilizing web antibody modeling (WAM) and SWISS-MODEL. Although docking experiments performed with an aromatic amino acid as model ligand were unsuccessful with the WAM structure, ligand binding was achieved with the SWISS-MODEL structure. Incorporation of side-chain flexibility within the binding site resulted in a protein structure that stereoselectively binds to the D-enantiomer of the model ligand. In addition to four hydrogen bonds that are formed between amino acid residues in the binding site and the ligand, a number of hydrophobic interactions are involved in the formation of the antibody-ligand complex. The aromatic side chain of the ligand interacts with a tryptophan and a tyrosine residue in the binding site through pi-pi stacking. Fluorescence spectroscopic investigations also suggest the presence of tryptophan residues in the binding site, as ligand binding causes an enhancement of the antibody's intrinsic fluorescence at an emission wavelength of 350 nm. Based on the modeled antibody structure, the L-enantiomer of the model ligand cannot access the binding site due to steric hindrance. Additional docking experiments performed with D-phenylalanine and D-norvaline showed that these ligands are bound to the antibody in a way analogous to the D-enantiomer of the model ligand.     

Kinetics, in silico docking, molecular dynamics, and MM-GBSA binding studies on prototype indirubins, KT5720, and staurosporine as phosphorylase kinase ATP-binding site inhibitors: the role of water molecules examined

With an aim toward glycogenolysis control in Type 2 diabetes, we have investigated via kinetic experiments and computation the potential of indirubin (IC?? > 50 μM), indirubin-3'-oxime (IC?? = 144 nM), KT5720 (K(i) = 18.4 nM) and staurosporine (K(i) = 0.37 nM) as phosphorylase kinase (PhKγtrnc) ATP-binding site inhibitors, with the latter two revealed as potent inhibitors in the low nM range. Because of lack of structural information, we have exploited information from homologous kinase complexes to direct in silico calculations (docking, molecular dynamics, and MMGBSA) to predict the binding characteristics of the four ligands. All inhibitors are predicted to bind in the same active site area as the ATP adenine ring, with binding dominated by hinge region hydrogen bonds to Asp104:O and Met106:O (all four ligands) and also Met106:NH (for the indirubins). The PhKγtrnc-staurosporine complex has the greatest number of receptor-ligand hydrogen bonds, while for the indirubin-3'-oxime and KT5720 complexes there is an important network of interchanging water molecules bridging inhibitor-enzyme contacts. The MM-GBSA results revealed the source of staurosporine's low nM potency to be favorable electrostatic interactions, while KT5720 has strong van der Waals contributions. KT5720 interacts with the greatest number of protein residues either by direct or 1-water bridged hydrogen bond interactions, and the potential for more selective PhK inhibition based on a KT5720 analogue has been established. Including receptor flexibility in Schr?dinger induced-fit docking calculations in most cases correctly predicted the binding modes as compared with the molecular dynamics structures; the algorithm was less effective when there were key structural waters bridging receptor-ligand contacts.     

Structural, phylogenetic and docking studies of D-amino acid oxidase activator (DAOA), a candidate schizophrenia gene

Background

Schizophrenia is a neurodegenerative disorder that occurs worldwide and can be difficult to diagnose. It is the foremost neurological disorder leading to suicide among patients in both developed and underdeveloped countries. D-amino acid oxidase activator (DAOA), also known as G72, is directly implicated in the glutamateric hypothesis of schizophrenia. It activates D-amino acid oxidase, which oxidizes D-serine, leading to modulation of the N-methyl-D-aspartate receptor.

Methods

MODELLER (9v10) was utilized to generate three dimensional structures of the DAOA candidate gene. The HOPE server was used for mutational analysis. The Molecular Evolutionary Genetics Analysis (MEGA5) tool was utilized to reconstruct the evolutionary history of the candidate gene DAOA. AutoDock was used for protein-ligand docking and Gramm-X and PatchDock for protein-protein docking.

Results

A suitable template (1ZCA) was selected by employing BLASTp on the basis of 33% query coverage, 27% identity and E-value 4.9. The Rampage evaluation tool showed 91.1% favored region, 4.9% allowed region and 4.1% outlier region in DAOA. ERRAT demonstrated that the predicted model had a 50.909% quality factor. Mutational analysis of DAOA revealed significant effects on hydrogen bonding and correct folding of the DAOA protein, which in turn affect protein conformation. Ciona was inferred as the outgroup. Tetrapods were in their appropriate clusters with bifurcations. Human amino acid sequences are conserved, with chimpanzee and gorilla showing more than 80% homology and bootstrap value based on 1000 replications. Molecular docking analysis was employed to elucidate the binding mode of the reported ligand complex for DAOA. The docking experiment demonstrated that DAOA is involved in major amino acid interactions: the residues that interact most strongly with the ligand C₂₈H₂₈N₃O₅PS₂ are polar but uncharged (Gln36, Asn38, Thr 122) and non-polar hydrophobic (Ile119, Ser171, Ser21, Ala31). Protein-protein docking simulation demonstrated two ionic bonds and one hydrogen bond involving DAOA. Lys-7 of the receptor protein interacted with Lys-163 and Asp-2037. Tyr-03 interacted with Arg-286 of the ligand protein and formed a hydrogen bond.

Conclusion

The predicted interactions might serve to inhibit the disease-related allele. It is assumed that current bioinformatics methods will contribute significantly to identifying, analyzing and curing schizophrenia. There is an urgent need to develop effective drugs for schizophrenia, and tools for examining candidate genes more accurately and efficiently are required.     

Residue conservation information for generating near-native structures in protein-protein docking

MOTIVATION: Protein-protein docking algorithms typically generate large numbers of possible complex structures with only a few of them resembling the native structure. Recently (Duan et al., Protein Sci, 14:316-218, 2005), it was observed that the surface density of conserved residue positions is high at the interface regions of interacting protein surfaces, except for antibody-antigen complexes, where a lesser number of conserved positions than average is observed at the interface regions. Using this observation, we identified putative interacting regions on the surface of interacting partners and significantly improved docking results by assigning top ranks to near-native complex structures. In this paper, we combine the residue conservation information with a widely used shape complementarity algorithm to generate candidate complex structures with a higher percentage of near-native structures (hits). What is new in this work is that the conservation information is used early in the generation stage and not only in the ranking stage of the docking algorithm. This results in a significantly larger number of generated hits and an improved predictive ability in identifying the native structure of protein-protein complexes. RESULTS: We report on results from 48 well-characterized protein complexes, which have enough residue conservation information from the same 59 benchmark complexes used in our previous work. We compute conservation indices of residue positions on the surfaces of interacting proteins using available homologous sequences from UNIPROT and calculate the solvent accessible surface area. We combine this information with shape-complementarity scores to generate candidate protein-protein complex structures. When compared with pure shape-complementarity algorithms, performed by FTDock, our method results in significantly more hits, with the improvement being over 100% in many instances. We demonstrate that residue conservation information is useful not only in refinement and scoring of docking solutions, but also helpful in enrichment of near-native-structures during the generation of candidate geometries of complex structures.     

Structure-based tailoring of compound libraries for high-throughput screening: discovery of novel EphB4 kinase inhibitors

High-throughput docking is a computational tool frequently used to discover small-molecule inhibitors of enzymes or receptors of known three-dimensional structure. Because of the large number of molecules in chemical libraries, automatic procedures to prune multimillion compound collections are useful for high-throughput docking and necessary for in vitro screening. Here, we propose an anchor-based library tailoring approach (termed ALTA) to focus a chemical library by docking and prioritizing molecular fragments according to their binding energy which includes continuum electrostatics solvation. In principle, ALTA does not require prior knowledge of known inhibitors, but receptor-based pharmacophore information (hydrogen bonds with the hinge region) is additionally used here to identify molecules with optimal anchor fragments for the ATP-binding site of the EphB4 receptor tyrosine kinase. The 21,418 molecules of the focused library (from an initial collection of about 730,000) are docked into EphB4 and ranked by force-field-based energy including electrostatic solvation. Among the 43 compounds tested in vitro, eight molecules originating from two different anchors show low-micromolar activity in a fluorescence-based enzymatic assay. Four of them are active in a cell-based assay and are potential anti-angiogenic compounds.     

The Role of Backbone Hydrogen Bonds in the Transition State for Protein Folding of a PDZ Domain

Backbone hydrogen bonds are important for the structure and stability of proteins. However, since conventional site-directed mutagenesis cannot be applied to perturb the backbone, the contribution of these hydrogen bonds in protein folding and stability has been assessed only for a very limited set of small proteins. We have here investigated effects of five amide-to-ester mutations in the backbone of a PDZ domain, a 90-residue globular protein domain, to probe the influence of hydrogen bonds in a β-sheet for folding and stability. The amide-to-ester mutation removes NH-mediated hydrogen bonds and destabilizes hydrogen bonds formed by the carbonyl oxygen. The overall stability of the PDZ domain generally decreased for all amide-to-ester mutants due to an increase in the unfolding rate constant. For this particular region of the PDZ domain, it is therefore clear that native hydrogen bonds are formed after crossing of the rate-limiting barrier for folding. Moreover, three of the five amide-to-ester mutants displayed an increase in the folding rate constant suggesting that the hydrogen bonds are involved in non-native interactions in the transition state for folding.     

Hydrogen bonding penalty upon ligand binding

Ligand binding involves breakage of hydrogen bonds with water molecules and formation of new hydrogen bonds between protein and ligand. In this work, the change of hydrogen bonding energy in the binding process, namely hydrogen bonding penalty, is evaluated with a new method. The hydrogen bonding penalty can not only be used to filter unrealistic poses in docking, but also improve the accuracy of binding energy calculation. A new model integrated with hydrogen bonding penalty for free energy calculation gives a root mean square error of 0.7 kcal/mol on 74 inhibitors in the training set and of 1.1 kcal/mol on 64 inhibitors in the test set. Moreover, an application of hydrogen bonding penalty into a high throughput docking campaign for EphB4 inhibitors is presented, and remarkably, three novel scaffolds are discovered out of seven tested. The binding affinity and ligand efficiency of the most potent compound is about 300 nM and 0.35 kcal/mol per non-hydrogen atom, respectively.