Binding modes and pharmacophore modelling of thermolysin inhibitors.

In the present paper 25 known thermolysin inhibitors were docked into thermolysin using the Internal Coordinate Mechanics (ICM) software. Pharmacophore models based on thermolysin binding modes and activity profiles were generated using the LigandScout program. The docking studies indicated that all 25 inhibitors coordinated the catalytic zinc in bidentate or monodentate geometry. A 'three-point' pharmacophore model was proposed which consisted of a hydrophobic group, a negative ionizable group and a hydrogen bond acceptor group. Finally the pharmacophore model has been tested against a small compound library containing 18 highly, moderately, less active as well as inactive compounds. The screening indicated that the pharmacophore model could, identify highly active compounds in front of inactive or less active ones.

Templates and models of monoamine transporter proteins.

X-ray crystallography, structural bioinformatics and computational chemistry have become important techniques in the discovery and development of effective and safe new drugs. From a drug discovery point of view, membrane proteins are among the most interesting molecular targets, but the current knowledge about detailed 3D structures of membrane proteins is sparse. Homology modeling techniques may provide structural knowledge about membrane proteins and their interactions with drugs and other molecules. The neurotransmitter sodium symporters (NSS) are the molecular targets of many pharmacologically active substances, and we have used three different secondary transporters as templates for modeling the NSS proteins DAT, NET and SERT. The first template was based on the electron density projection map of the Escherichia coli Na+/H+ antiporter (NhaA), while later the X-ray structure of Lac Permease (symporter) was used as a template. The helical architectures of these templates have a lot in common, and models based on both could contribute with structural explanations of several experimental studies in spite of low homology with NSS proteins. In 2005 the crystal structure of a bacterial homologue of the human monoamine neurotransmitter transporter Aquifex aeolicus (LeuTAa) was reported. This structure was the first experimental structure of a NSS family member, and represented a breakthrough for homology modeling of pharmacological important NSS proteins. Since then several X-ray structures LeuTAa in complex with pharmacologically important compounds have been published. Homology models of NSS proteins, combined with site-directed mutagenesis data, have identified ligand binding sites and contributed with important knowledge for new drug development.

Minireview: molecular structure and dynamics of drug targets.

Summary of lectures presented at the Czech and Slovak Pharmacological Meeting, Prague, September 2008.

Molecular model of the Escherichia coli Na1/H1 antiporter NhaA.

A three-dimensional electron density projection map of the ion-coupled membrane protein Escherichia coli Na+/H+ antiporter (NhaA) was recently published. Based on this projection map, and previous biophysical studies determining the assignment of the 12 transmembrane alpha-helices (TMHs), a three-dimensional molecular model of the NhaA was constructed, using interactive molecular graphics and energy calculations. The diuretic drug, amiloride, was docked into the model and putative interacting amino acids were identified. The model suggests that the pH dependent activity of NhaA may be explained by charge changes in the intracellular loop between TMH8 and TMH9 which alter the positions of TMHs 4, 5 and 11 relative to each other, such that a pore area of the transporter protein is opened.

Bioinformatics and receptor mechanisms of psychotropic drugs.

One important aspect in biotechnology is gene discovery and target validation for drug discovery. Information from the human genome (HUGO) project may be used to deduce the amino acid sequence of all proteins produced in the human body. However, knowing the amino acid sequence of a protein is not the same as knowing its function. Identification of novel molecular targets for discovery of new, safer and more efficient therapeutic drugs from the human genome sequences requires multidisciplinary research efforts, including proteomics, structural biology and bioinformatics. In addition to possible effects on gene expression, most of the currently used therapeutic drugs either have enzymes or membrane proteins as their molecular targets of action. These membrane proteins include transporters of small molecules across cell membranes, ion channels, or receptors that convey signals from one side of a membrane to the other. Our research group as well as others have used computational techniques, along with biotechnology, molecular biology and other experimental techniques, to construct detailed 3-dimensional models of transporter proteins and G-protein coupled receptors (GPCRs), which are the molecular targets of action of psychotropic drugs. The models have been used to simulate the molecular dynamics and study the ligand binding and signal transduction mechanisms of these receptors. The use of bioinformatics, as exemplified in our modelling of GPCRs, is only one of the key factors for success in post-genomic research for new targets for therapeutic drugs.

Molecular dynamics of buspirone analogues interacting with the 5-HT1A and 5-HT2A serotonin receptors.

Three-dimensional (3-D) models of the human serotonin 5-HT1A and 5-HT2A receptors were constructed, energy refined, and used to study the interactions with a series of buspirone analogues. For both receptors, the calculations showed that the main interactions of the ligand imide moieties were with amino acids in transmembrane helix (TMH) 2 and 7, while the main interactions of the ligand aromatic moieties were with amino acids in TMH5, 6 and 7. Differences in binding site architecture in the region of highly conserved serine and tyrosine residues in TMH7 gave slightly different binding modes of the buspirone analogues at the 5-HT1A and 5-HT2A receptors. Molecular dynamics simulations of receptor-ligand interactions indicated that the buspirone analogues did not alter the interhelical hydrogen bonding patterns upon binding to the 5-HT2A receptor, while interhelical hydrogen bonds were broken and others were formed upon ligand binding to the 5-HT1A receptor. The ligand-induced changes in interhelical hydrogen bonding patterns of the 5-HT1A receptor were followed by rigid body movements of TMH2, 4 and 6 relative to each other and to the other TMHs, which may reflect the structural conversion into an active receptor structure.

Ligand induced conformational states of the 5-HT(1A) receptor.

It has been shown that G-protein coupled receptors have seven transmembrane alpha-helices, but the structural changes occurring in a G-protein coupled receptor as a response on agonist stimulus and the molecular events leading to blockade of the signal transduction by antagonists are not well understood. In the present study, the AMBER 5.0 force field was used for comparative molecular dynamics simulations of a 5-HT(1A) receptor model in the absence of ligand, in complex with a 5-HT(1A) receptor agonist (R)-8-hydroxy-2-(di-n-propylamino)tetralin [(R)-8-OH-DPAT], in complex with a selective 5-HT(1A) receptor antagonist (S)-N-tert-butyl-3-[4-(2-methoxyphenyl)piperazin-1-yl ]-2-phenylpropanamide [(S)-WAY100135], and in complex with the partial agonist, buspirone. In the simulations, the agonist induced larger conformational changes into transmembrane helix 3 and 6 than into the other helices, while the main conformational differences between the agonist bound receptor and the antagonist bound receptor were in transmembrane helix 5 and 6. During the simulations, all the three ligands constrained the helical movements compared to those observed in the receptor without any ligand.

Effect of the glutathione/glutathione disulfide redox couple on thiopurine methyltransferase.

The susceptibility of recombinant human thiopurine methyltransferase (hTPMT) to thiol-disulfide exchange was investigated. The enzyme was incubated in buffers of the redox couple GSH and GSSG. The values of the chosen concentrations and concentration ratios of the redox couple equaled those expected to occur in vivo. Activity measurements of the enzyme over time in these buffers at 30 degrees C indicated that thiol-disulfide exchange may be a part of the posttranslational modulation of hTPMT activity. Activity varied between 5% and 100%, with the lowest activities in buffers of low [GSH]/[GSSG] concentration ratios and of low total concentration of the redox couple. A thiol-disulfide exchange mechanism involving a mixed disulfide was proposed. Titration of the protein thiol groups with Ellmann's reagent (5,5'-dithiobis[2-nitrobenzoic acid]) revealed that at least two protein thiols were readily accessible for conjugation with the reagent, while others were conjugated more slowly. The previous model of hTPMT constructed by our group was in accordance with the experimental results. Inspection of the model indicated that one of the protein thiols subject to slow thiol-disulfide exchange may be situated at the binding site of the co-substrate of the enzyme and thus be responsible for the glutathione/glutathione disulfide modulation of the activity of hTPMT.

Molecular dynamics of 5-HT1A and 5-HT2A serotonin receptors with methylated buspirone analogues.

In the present study experimentally determined ligand selectivity of three methylated buspirone analogues (denoted as MM2, MM5 and P55) towards 5-HT1A and 5-HT2A serotonin receptors was theoretically investigated on a molecular level. The relationships between the ligand structure and 5-HT1A and 5-HT2A receptor affinities were studied and the results were found to be in agreement with the available site-directed mutagenesis and binding affinity data. Molecular dynamics (MD) simulations of ligand-receptor complexes were performed for each investigated analogue, docked twice into the central cavity of 5-HT1A/5-HT2A, each time in a different orientation. Present results were compared with our previous theoretical results, obtained for buspirone and its non-methylated analogues. It was found that due to the presence of the methyl group in the piperazine ring the ligand position alters and the structure of the ligand-receptor complex is modified. Further, the positions of derivatives with pyrimidinyl aromatic moiety and quinolinyl moiety are significantly different at the 5-HT2A receptor. Thus, methylation of such derivatives alters the 3D structures of ligand-receptor complexes in different ways. The ligand-induced changes of the receptor structures were also analysed. The obtained results suggest, that helical domains of both receptors have different dynamical behaviour. Moreover, both location and topography of putative binding sites for buspirone analogues are different at 5-HT1A and 5-HT2A receptors.

Theoretical models of interactions between buspirone analogues and 5-HT1A and 5-HT2A serotonin receptor subtypes.

In present study the structure-selectivity relationship of buspirone and six of its analogues towards 5-HT1A and 5-HT2A serotonin receptors was investigated on molecular level. Molecular mechanics energy minimisation and advanced molecular dynamics (MD) simulations allowed us to perform a dynamic structural analysis of transmembrane helical domains of the human 5-HT1A and 5-HT2A receptors and investigate the ligand-induced changes of the entire structure of the ligand-receptor complex. The obtained results suggest, that helical and extracellular domains of both receptors have different topography of the putative binding sites and also different dynamical behaviour. The results of this study are consistent with experimental site-directed mutagenesis data and binding affinities of examined ligands towards both serotonin receptors.

The catalytic triad in Drosophila alcohol dehydrogenase: pH, temperature and molecular modelling studies.

Drosophila alcohol dehydrogenase belongs to the short chain dehydrogenase/reductase (SDR) family which lack metal ions in their active site. In this family, it appears that the three amino acid residues, Ser138, Tyr151 and Lys155 have a similar function as the catalytic zinc in medium chain dehydrogenases. The present work has been performed in order to obtain information about the function of these residues. To obtain this goal, the pH and temperature dependence of various kinetic coefficients of the alcohol dehydrogenase from Drosophila lebanonensis was studied and three-dimensional models of the ternary enzyme-coenzyme-substrate complexes were created from the X-ray crystal coordinates of the D. lebanonensis ADH complexed with either NAD(+) or the NAD(+)-3-pentanone adduct. The kon velocity for ethanol and the ethanol competitive inhibitor pyrazole increased with pH and was regulated through the ionization of a single group in the binary enzyme-NAD(+) complex, with a DeltaHion value of 74(+/-4) kJ/mol (18(+/-1) kcal/mol). Based on this result and the constructed three-dimensional models of the enzyme, the most likely candidate for this catalytic residue is Ser138. The present kinetic study indicates that the role of Lys155 is to lower the pKa values of both Tyr151 and Ser138 already in the free enzyme. In the binary enzyme-NAD(+) complex, the positive charge of the nicotinamide ring in the coenzyme further lowers the pKa values and generates a strong base in the two negatively charged residues Ser138 and Tyr151. With the OH group of an alcohol close to the Ser138 residue, an alcoholate anion is formed in the ternary enzyme NAD(+) alcohol transition state complex. In the catalytic triad, along with their effect on Ser138, both Lys155 and Tyr151 also appear to bind and orient the oxidized coenzyme.

Comparative molecular dynamics of mesophilic and psychrophilic protein homologues studied by 1.2 ns simulations.

It is well established that the dynamic motion of proteins plays an important functional role, and that the adaptation of a protein molecule to its environment requires optimization of internal non-covalent interactions and protein-solvent interactions. Serine proteinases in general, and trypsin in particular has been used as a model system in exploring possible structural features for cold adaptation. In this study, a 500 p.s. and a 1200 p.s. molecular dynamics (MD) simulation at 300 K of both anionic salmon trypsin and cationic bovine trypsin are analyzed in terms of molecular flexibility, internal non-covalent interactions and protein-solvent interactions. The present MD simulations do not indicate any increased flexibility of the cold adapted enzyme on an overall basis. However, the apparent higher flexibility and deformability of the active site of anionic salmon trypsin may lower the activation energy for ligand binding and for catalysis, and might be a reason for the increased binding affinity and catalytic efficiency compared to cationic bovine trypsin.

Molecular dynamics of NPY Y1 receptor activation.

A three-dimensional model of the human neuropeptide Y(NPY)Y1 receptor (hY1) was constructed, energy refined and used to simulate molecular receptor interactions of the peptide ligands NPY, [L31, P34]NPY, peptide YY (PYY) and pancreatic polypeptide (PP), and of the nonpeptide antagonist R-N2-(diphenylacetyl)-N-(4-hydroxyphenyl)methyl-argininamide (BIBP3226) and its S-enantiomer BIBP3435. The best complementarity in charges between the receptor and the peptides, and the best structural accordance with experimental studies, was obtained with amino acid 1-4 of the peptides interacting with Asp194, Asp200, Gln201, Phe202 and Trp288 in the receptor. Arg33 and Arg35 of the peptides formed salt bridges with Asp104 and Asp287, respectively, while Tyr36 interacted in a binding pocket formed by Phe41, Thr42, Tyr100, Asn297, His298 and Phe302. Calculated electrostatic potentials around NPY and hY1 molecules indicated that ligand binding is initiated by electrostatic interactions between a highly positive region in the N- and C-terminal parts of the peptides, and a negative region in the extracellular receptor domains. Molecular dynamics simulations of NPY and BIBP3226 interactions with the receptor indicated rigid body motions of TMH5 and TMH6 upon NPY binding as mechanisms of receptor activation, and that BIBP3226 may act as an antagonist by constraining these motions.

The ligand binding site of NPY at the rat Y1 receptor investigated by site-directed mutagenesis and molecular modeling.

The ligand binding site of neuropeptide Y (NPY) at the rat Y1 (rY1,) receptor was investigated by construction of mutant receptors and [3H]NPY binding studies. Expression levels of mutant receptors that did not bind [3H]NPY were examined by an immunological method. The single mutations Asp85Asn, Asp85Ala, Asp85Glu and Asp103Ala completely abolished [3H]NPY binding without impairing the membrane expression. The single mutation Asp286Ala completely abolished [3H]NPY binding. Similarly, the double mutation Leu34Arg/Asp199Ala totally abrogated the binding of [3H]NPY, whereas the single mutations Leu34Arg and Asp199Ala decreased the binding of [3H]NPY 2.7- and 5.2-fold, respectively. The mutants Leu34Glu, Pro35His as well as Asp193Ala only slightly affected [3H]NPY binding. A receptor with a deletion of the segment Asn2-Glu20 or with simultaneous mutations of the three putative N-terminal glycosylation sites, displayed no detectable [3H]NPY binding, due to abolished expression of the receptor at the cell surface. Taken together, these results suggest that amino acids in the N-terminal part as well as in the first and second extracellular loops are important for binding of NPY, and that Asp85 in transmembrane helix 2 is pivotal to a proper functioning of the receptor. Moreover, these studies suggest that the putative glycosylation sites in the N-terminal part are crucial for correct expression of the rY1 receptor at the cell surface.

The ligand-binding site of buspirone analogues at the 5-HT1A receptor.

A three-dimensional model of the 5-HT1A receptor in man was constructed by molecular-modelling techniques and used to study the molecular interactions of a series of buspirone analogues with the 5-HT1A receptor by molecular-mechanical-energy minimization and molecular-dynamics simulations. The receptor has seven trans-membrane alpha helices (TMHs) organized according to the electron-density-projection map of visual rhodopsin, and includes all loops between TMHs and the N- and C-terminal parts. The best fit between the buspirone analogues and the receptor model was obtained with the quinolinyl part of the ligand molecules interacting with amino acids in TMH6, the imide group interacting with amino acids in TMH2, TMH3 and TMH7, and the carbonyl groups hydrogen-bonded with Ser86 and Ser393. The ligand-binding rank order deduced from the experimentally determined inhibition constant was reproduced by calculation of receptor-binding energies of the buspirone analogues. The models suggest that steric hindrance and repulsive forces between the receptor and the imide group of the buspirone analogues are the most important determinants of ligand-binding affinity for discriminating between these ligands.

Mapping the suramin-binding sites of human neutrophil elastase: investigation by fluorescence resonance energy transfer and molecular modeling.

Neutrophil elastase (NE), a mediator of inflammation, binds with high affinity numerous anionic molecules including suramin, a polysulfated naphthylurea, which inhibits it with a Ki of 0.2 microM and a 4:1 suramin:NE stoichiometry and thus constitutes a potential therapeutic agent. In an attempt to locate the suramin molecules on NE, we investigated the NE-suramin interaction using steady-state and time-resolved fluorescence spectroscopy. The time-resolved intensity decay of NE, a protein with three Trp residues, in positions 27, 141, and 237 (chymotrypsin numbering system) was best described by a three-exponential function with lifetimes ranging from 0.22 to 2.28 ns. Comparison of the accessibility of the three lifetime classes to the fluorescence quenchers acrylamide and iodide with the computed solvent accessibility of the three Trp residues in the crystal structure of NE indicates that the main, if not the sole, contribution to the 2.28 ns lifetime class is brought about by the fully buried Trp 141 residue. The addition of suramin to NE induces a sharp decrease in NE fluorescence and a corresponding increase in suramin fluorescence due to an efficient fluorescence resonance energy transfer (FRET) between the Trp residues of NE, acting as donors, and the naphthalene rings of suramin, behaving as acceptors. From the fate of the longest lifetime class in the presence of variable suramin concentrations, we deduce that two suramins are bound at less than 17 A from Trp 141, whereas the two others are located at least 29 A from Trp 141. Moreover, neither the binding of suramin to NE nor the FRET process was modified when NE was complexed with a peptide chloromethylketone inhibitor, suggesting that suramin does not directly interfere with the substrate binding site of NE. These data were used as constraints to model the NE-suramin complex.

Molecular modelling of UH-301 and 5-HT(1a) receptor interactions.

A three-dimensional model of the human 5-HT(1a) receptor was constructed by molecular modelling, and the molecular and electronic structures of (R)- and (S)-5-fluoro-8-hydroxy-2-(dipropylamino)tetralin (UH-301) and of (R)- and (S)-8-hydroxy-2-(dipropylamino)tetralin (8-OH-DPAT) were examined by molecular mechanics and quantum mechanics calculations and molecular dynamics simulations. The receptor model has seven transmembrane alpha-helices (TMHs), organized according to a projection map of visual rhodopsin, and includes all loops between helices and the N- and C-terminal parts. Interactions of UH-301 and 8-OH-DPAT with the 5-HT(1a) receptor were examined by molecular dynamics simulations and energy minimization of receptor-ligand complexes. 8-OH-DPAT had lower electrostatic potentials around the hydroxyl group and stronger hydrogen bonding to the receptor model than had UH-301. The simulations indicated that the 5-HT(1a) receptor agonists, (R)- and (S)-8-OH-DPAT and (R)-UH-301, interacted with the receptor at a site closer to Asp82 in TMH2 than did (S)-UH-301, which is a 5-HT1a receptor antagonist. Simulations of receptor-ligand complexes indicated that Asp82, Asp116, Serl99, Thr2OO and Ile385 are essential for binding of both agonist and antagonist to the receptor.

Molecular dynamics of the 5-HT1a receptor and ligands.

A 3-D model of the human 5-HT1a receptor was constructed from its amino acid sequence by computer graphics techniques, molecular mechanics calculations and molecular dynamics simulations. The model has seven alpha-helical membrane spanning segments, which form a central core containing a putative ligand binding site. Electrostatic potentials 1.4 A outside the water accessible surface were mainly negative on the synaptic side of the receptor model and at the postulated ligand binding site, and positive in the cytoplasmic domains. The negative electrostatic potentials around the synaptic domains indicate that positively charged ligands are attracted to the receptor by electrostatic forces. Molecular dynamics simulations of the receptor model with serotonin, ipsapirone, R(-)-methiothepin or S(+)-methiothepin in the central core suggested that up to 22 different amino acid residues may form a ligand binding pocket, and contribute to the specificity of ligand recognition and binding.

Molecular dynamics of serotonin and ritanserin interacting with the 5-HT2 receptor.

A three-dimensional model of the serotonin (5-hydroxytrytamine; 5-HT) 5-HT2 receptor was constructed from the amino acid sequence by molecular graphics techniques, molecular mechanics energy calculations and molecular dynamics simulations. The receptor model has 7 alpha helical segments which form a membrane-spanning duct with a putative ligand binding site. Most of the synaptic domains and the ligand binding site were surrounded by negative electrostatic potentials, suggesting that positively charged ligands are attracted to the receptor by electrostatic forces. The cytoplasmic domains, except the C-terminal tail, had mainly positive electrostatic potentials. The molecular dynamics of the receptor-ligand complex was examined in 100 ps simulations with 5-HT or ritanserin at a postulated binding site. During the simulations the helices moved from an initial circular arrangement into a more oval arrangement, and became slightly tilted relative to each other. The protonated ligands neutralized the negative electrostatic potentials around Asp 120 and Asp 155 in the central core of the receptor. 5-HT had only weak interactions with Asp 155 but strong interactions with Asp 120 during the simulations, with the amino group of 5-HT tightly bound to the carboxylic side chain of Asp 120. Ritanserin showed similarly strong interactions with Asp 120 and Asp 155 during the simulations.