Atomistic modeling of enantioselection in chromatography.

A review of atomistic molecular modeling studies related to chromatographic separations of enantiomers is presented. Only those types of calculations where direct interactions between a selector and a selectand are involved are described in this review; omitted are regression models. An emphasis is placed on comparing methods used for sampling potential energy surfaces implementing different methodologies like quantum and molecular mechanics for energy calculations, and molecular dynamics and Monte Carlo sampling strategies for simulations. Type I-V chiral stationary phases and additives for capillary electrophoresis and ion-pair chromatography are covered in this review.

Comparative molecular field analysis of quinine derivatives used as chiral selectors in liquid chromatography: 3D QSAR for the purposes of molecular design of chiral stationary phases.

A comparative molecular field analysis (CoMFA) was carried out on a set of aligned quinine-based stationary phase molecules used in enantioselective chromatography. The best QSAR derived has a cross-validated (predictive) r(2)(cv) = 0.671 and a normal r(2) = 0.998. For CoMFAs using both steric and electrostatic fields as descriptors, the steric field descriptors explained more than 91% of the variance while the electrostatic descriptors explained less than 9% of the variance. It is concluded that the long-range electrostatic potential surrounding the positively charged CSPs are not enantiodiscriminating, while the van der Waals and local electrostatic surface features of these CSPs are highly discriminating. Quantum mechanical calculations back up this claim by showing a relatively symmetric electrostatic iso-contour surface. From the QSAR derived here, a region around the carbamate moiety was located where placement of steric bulk is predicted to enhance chiral discrimination. A set of possible synthetic target molecules is presented.

Atomistic modeling of enantioselective binding.

This Account focuses on computational studies related to chiral recognition. It begins with a description of potential energy surfaces and the computational tools used to explore such surfaces, describes approximations and assumptions made by researchers computing enantioselective binding, and then explains why differential free energies of binding can be computed so accurately. The review focuses on chiral recognition in chromatography, emphasizing binding and enantiodiscriminating forces responsible for chiral recognition. The Account also describes computational studies of chiral recognition in cyclodextrins, proteins, and synthetic receptors.

Enantiodiscrimination by a quinine-based chiral stationary phase: a computational study.

A detailed computational study of a derivatized quinine chiral stationary phase (CSP) interacting with enantiomeric 3, 5-dinitrobenzoyl derivatives of leucine was carried out to understand where and how chiral discrimination takes place. The most stable structure of the CSP derived from a conformer search gave a structure whose geometry agrees with an X-ray structure (rmsd 0.6 A). The computed retention order and enantiodiscriminating free energy differences also agree with chromatographic data. The location and characteristics of the analyte binding site were assessed. An evaluation of total energies and intermolecular energies responsible for complex formation and for chiral discrimination was performed. Molecular dynamics trajectories of those intermolecular forces as well as distributions of the stabilizing and destabilizing forces are presented. A partitioning of the CSP into molecular fragments and the role each fragment plays in complexation and chiral recognition is also described.

A computational model of the nicotinic acetylcholine binding site.

We have derived a model of the nicotinic acetylcholine binding site. This was accomplished by using three known agonists (acetylcholine, nicotine and epibatidine) as templates around which polypeptide side chains, found to be part of the receptor cavity from published molecular biology studies, are allowed to flow freely in molecular dynamics simulations and mold themselves around these templates. The resulting supramolecular complex should thus be a complement, both in terms of steric effects as well as electronic effects, to the agonists and it should be a good estimation of the true receptor cavity structure. The shapes of those minireceptor cavities equilibrated rapidly on the simulation time scale and their structural congruence is very high, implying that a satisfactory model of the nicotinic acetylcholine binding site has been achieved. The computational methodology was internally tested against two rigid and specific antagonists (dihydro-beta-erytroidine and erysoidine), that are expected to give rise to a somewhat differently shaped binding site compared to that derived from the agonists. Using these antagonists as templates there were structural reorganizations of the initial receptor cavities leading to distinctly different cavities compared to agonists. This indicates that adequate times and temperatures were used in our computational protocols to achieve equilibrium structures for the agonists. Overall, both minireceptor geometries for agonists and antagonists are similar with the exception of one amino acid (ARG209).

The nicotinic cholinergic receptor: a theoretical model.

Based on published affinity-labeling and mutagenesis experiments describing the effect of changes in specific amino acids in molecular biological studies on the nicotinic acetylcholinergic receptor (nAChR), we have identified 12 amino acids which are important in functioning at the nicotinic cholinergic receptor. The work presented here provides an atomistic model of this important receptor based on our molecular modeling studies. We found five of these amino acids (TRP86, ASP89, TYR93, ASP138, and THR191) to be associated with the cationic end of acetylcholine (ACh), which is electron-deficient. Three other amino acids (ARG209, TYR190, and TYR198) are associated with the ester end, where an enhanced electron density is present. After hydrogen bonding between the two oxygen atoms at the ester end, and two of the guanidinium hydrogen atoms in ARG209. ASP200 hydrogen bonds to the other two hydrogen atoms of the guanidinium group, thus forming a pseudo-ring. Two aromatic amino acids (TRP149 and TYR151) then enhance the binding at the pseudo-ring through additional hydrogen bonding and charge-transfer complexation, with THR150 functioning to further stabilize this evolving charge-transfer complex. We postulate that this latter process allows the ion channel to twist, thus opening it. From the published amino acid sequence in the polypeptides at the 5HT-3, GABA, and glycine receptors (Maricq et al.: Science 254:432-437, 1991), we also speculate on which amino acids are involved in these three receptors.

Glycine and GABA receptors: molecular mechanisms controlling chloride ion flux.

We have been able to show that the three clearly identified atoms common to the inhibitory neurotransmitters glycine and GABA, that we previously hypothesized to serve as attachment points at the glycinergic and gabanergic receptor, can indeed interact through both electrostatic and hydrogen bonding to several amino acids, which have been identified in molecular biological investigations as both present and critical in the physiological functioning of key polypeptides common to these inhibitory receptors. In addition, amino acids also involved in stabilizing the interaction between the antagonists strychnine and R5135 at the glycinergic and gabanergic receptors, respectively, have been shown to fit our complex model. We identify in detail molecular mechanisms to explain how glycine and GABA initiate chloride ion movement from extraneuronal fluid in the synaptic cleft to intraneuronal volume. In addition, we also identify the molecular mechanisms involved in the blocking of chloride ion movement by strychnine at the glycinergic receptor and by R5135 at the gabanergic receptor. We also present two computer-generated color prints, one for the glycine receptor and one for the GABA receptor, which show the quantum mechanically geometry optimized complex formed between receptor side chains, i.e., the part of the amino acids in the polypeptide that interacts with the zwitterionic inhibitory neurotransmitters. These computer-generated color figures also show a) the important electrostatic and hydrogen bonding in these interactions, b) a van der Waals model of this complex to illustrate that no steric repulsions exist, and c) the molecular electrostatic potential energy map showing the electrostatic potentials of neurotransmitter bound to the receptor model. Finally, we show with computer calculations that the pseudo-rings, formed between the positive quanidinium group in arginine and one of the oxygen atoms in the carboxyl group in both glycine or GABA, result in a positive planar region which appears to be involved in a charge-transfer complex with aromatic benzene groups in amino acids such as phenylalanine and tryosine.

Comparison of binding mechanisms at cholinergic, serotonergic, glycinergic and GABAergic receptors.

Employing computational methods and published data from molecular biological studies involving amino acid sequences in the polypeptide receptors, the authors studied and compared how two excitatory neurotransmitters, ACh and 5-HT, and two inhibitory neurotransmitters, glycine and GABA, can bind to their respective recognition sites at CNS receptors. Models for each neurotransmitter interaction with specific amino acids are described and identified. Molecular mechanisms are identified that can explain how the binding process initiates ion flow through channels located within the postsynaptic membrane such that if the neurotransmitter is inhibitory, hyperpolarization occurs, and if excitatory, depolarization occurs. Although the theoretical work described indicates that there is a difference in molecular mechanisms operative at the anionic and cationic channels, and provides an explanation why the former is more specific, the molecular modeling data and the similarities of specific amino acids in the sequence in all four receptor polypeptides used to construct the four models support ACh, 5-HT, glycine and GABA as being members of the same ligand-gated ion channel superfamily.

Identifying agonistic and antagonistic mechanisms operative at the GABA receptor.

Based on our molecular modeling investigations of the glycinergic receptor, we expanded our studies to similarly investigate the GABAergic receptor. New data suggest there may exist a slightly different agonistic mechanism for the molecules described herein as compared to glycine. The origin of this is undoubtedly the fact that, while glycine has a positive and two negative binding sites, it is significantly shorter than GABA and the other GABA agonists. Clearly, discovery of more glycine agonists is needed to further clarify this point. Moreover, we find a remarkedly different antagonistic mechanism exists for this phylogenetically newer inhibitory system in the central nervous system (CNS) than recently reported for strychnine and eight weaker glycine antagonists. We used GABA and six agonists (muscimol, dihydromuscimol, THIP, isoguvacine, trans-3-aminocyclopentane-1-carboxylic acid, piperidine-4-sulfonic acid) and five antagonists (bicuculline-N15-methobromide, R5135, pitrazepin, iso-THAZ and securinine) to derive our conclusions. We found that each of the agonists have three clearly defined atoms that can serve as attachment points at the GABAA receptor site. One of the three attachment atoms includes a carbonyl or carboxylate oxygen. The role of the carbonyl or carboxylate atom is very important. First, we theorize that a rapid two-point attachment occurs (one from the positive end and one from one of the other two negative atoms on the ligand) at the recognition site in the receptor where GABA or a GABAergic agonist binds. The positive end of the agonist perhaps associates through hydrogen bonding to a beta-carboxyl group in one of the aspartate molecules in the polypeptide. The negative attachment points perhaps bind through hydrogen bonding to arginine molecules in this polypeptide. The second negative site in the agonist immediately triggers a conformational change by pulling together the aforementioned groups by electrostatic attraction, and hence opening the chloride channel. We propose the carbonyl oxygen is partly responsible for triggering the opening by formation of a double hydrogen bond to arginine. We postulate that this attraction is the first step inducing the conformational change. In the case of the GABA antagonists investigated, a fourth attachment site was not found. In fact only two sites have been identified similar to the group II glycine antagonists. Our data support a hypothesis for GABAergic antagonist activity which suggests that the antagonist simply binds to the recognition site and blocks the neurotransmitter, GABA, from entering this site thereby preventing the opening of the chloride channel; it just stays closed. This mechanism is different from the mechanism proposed for the large number of Group I glycine antagonists (Aprison et al.: J Neurosci Res 41: 259-269, 1995).

On identifying a second molecular antagonistic mechanism operative at the glycine receptor.

We used molecular modeling techniques to examine six reported antagonists of glycine with varying Ki values against strychnine. We found the data suggest two groups operating with different mechanisms. In group 1 (strychnine, brucine, Pitrazepin, and bicuculline methobromide) the antagonist contains two or three sites that can electrostatically bind to the three comparable groups of opposite charge in the recognition site where the natural neurotransmitter binds, thus opening the chloride channel. In addition, when in this position, the antagonist is able to also block the now opened chloride channel with a different portion of its structure. In many cases, this involves an interaction between a carbonyl group on the antagonist and the guanidinium group of arginine which is part of the polypeptide segment of the outer mouth of the chloride channel (Grenningloh et al., Nature 330:25-26, 1987). In group 2 (R5135 and 1,5-diphenyl-3,7-diazaadamantan-9-ol) the antagonist contains charged sites but when one of these molecules attaches to the recognition site, the chloride channel is not opened. In addition, R5135 contains a carbonyl group which attaches to arginine as pointed out in the text, whereas 1,5-diphenyl-3,7-diazaadamantan-9-ol contains a phenyl group that can block the channel.

On a molecular comparison of strong and weak antagonists at the glycinergic receptor.

Using molecular modeling techniques, we studied nine glycine antagonists in order to try to identify the molecular descriptors that characterize strychnine as a strong antagonist and N,N-dimethyl-muscimol, iso-THIA, THIA, N-methyl-THIP, iso-THAZ, THAZ, iso-THPO, and iso-THAO (see Experimental for chemical names) as weak glycine antagonists. We confirm that all nine compounds have the three-atom regions (two negative and one positive) that we have postulated are necessary to permit such compounds to attach to the recognition site in the glycinergic synapse. Furthermore, in the case of antagonists we have postulated the presence of a fourth atom that can attach to the top of the chloride ion channel. Each of the nine antagonists has such a fourth negative atom and the latter property gives each of these compounds their antagonistic characteristic. Further, only in the case of strychnine is there evidence that at its positively charged end does the positive charge extend to cover a region that could bind through electrostatic domains to a tertiary carboxyl group in an amino acid like aspartate. Published molecular biological data show that such an amino acid is present in the portion of the polypeptides identified in the glycine receptor. The bidentate binding is superior to the single site attachment that is present in the other eight weak glycine antagonists. In addition, the two negative atom sites in each antagonist are also in a position to participate in electrostatic binding through bidentate involvement with the positively charged guanidinium group of arginine. The latter amino acid also has been identified in the portion of the polypeptide chain at the glycine receptor. Finally, our molecular data predict that after strychnine, the eight weak glycine antagonists listed above are in order of decreasing potency, i.e., N,N-dimethyl-muscimol is the best of the weak antagonists and iso-THAO should be the weakest.

Identification of a second glycine-like fragment on the strychnine molecule.

Strychnine is a complex molecule that inhibits the physiological actions of glycine, an important inhibitory neurotransmitter in the spinal cord, brain stem, and other areas of many vertebrates. Since 1987, we have employed atomistic molecular modeling tools to find an explanation at the molecular level for how this antagonism works. We have located a second glycine-like fragment in the strychnine molecule that, when compared to glycine in a three pair atom analysis, provides an excellent topological and electronic charge congruence. The topological congruence in the second glycine-like fragment is much better than with the first fragment reported in 1987 when using a truncated strychnine molecule in the quantum mechanical analysis. A fourth negative atom, a characteristic of antagonists which we reported earlier (Aprison and Lipkowitz: J Neurosci Res 30:442-446, 1991; Aprison and Lipkowitz: J Neurosci Res 31:166-174, 1992) was found in strychnine. This result follows the pattern reported recently for the three weak glycine antagonists N,N-dimethylmuscimol, N-methyl-THIP, and iso-THAO, a bicyclic 5-isoxazolol zwitterion.

Potential radioprotective agents--IV. Schiff bases.

Twelve Schiff bases were prepared using salicylaldehyde, one with 5-chlorosalicylaldehyde, one with benzaldehyde, and a series of anilines substituted in the m- or p-positions. They were assayed for radioprotective activity in male, Swiss mice irradiated with a nearly lethal dose (950 cGy) of 6 mV photons produced by a linear accelerator, and were compared with the parent amines. Schiff base formation reduced toxicity of the parent amines; its effect on radioprotective activity was erratic, increasing activity in some cases, decreasing activity in others, and having no effect in still others. Radioprotective activity appears to be unrelated to a number of molecular descriptors. The highest radioprotection (100%) was observed for mixtures of p-aminopropiophenone with its Schiff base, or with the Schiff base of 1-(p-aminophenyl)-1-propanol (95%).

Potential radioprotective agents. 2. Substituted anilines.

A series of substituted anilines was examined for radioprotective activity by injecting them ip into mice subjected to a near-lethal dose of 6 mV photons. Electronegative groups such as Br, NO2, CN, and acyl in the meta or para position gave rise to highly active compounds (80-100% protection), while p-amino, methyl, amide, hydroxy, and fluoro groups decrease activity. No general correlations could be developed, however, between biological activity and a wide variety of calculated molecular parameters. The highest activity was found with p-aminobenzophenone (1), p-aminopropiophenone (2), its ethylene ketal (3), 2-amino-5-chloropyridine (35), and 5-amino-2-chloropyridine (36).

Trifluoromethanesulfonamide anthelmintics. Protonophoric uncouplers of oxidative phosphorylation.

A series of trifluoromethanesulfonamides (TFMS) was synthesized and tested for uncoupling activity in rat liver mitochondria. With succinate as the mitochondrial substrate, and the respiratory control index (RCI) as an indicator of their uncoupling ability, we found that all of the TFMS tested were uncouplers of oxidative phosphorylation; the effective concentration (RCI I50) ranged from less than 1 microM to greater than 1000 microM. Correlation techniques were used to assess the strength of the relationship between the ability of a TFMS to uncouple oxidative phosphorylation and its ability to lower the electrical resistance of planar bimolecular lipid membranes. There was a highly significant (P < 0.001) positive linear relationship (r = 0.97) between the ability of a TFMS to uncouple oxidative phosphorylation and its ability to lower electrical resistance. These findings are consistent with the view that the TFMS are lipophilic protonophoric uncouplers of mitochondrial oxidative phosphorylation. Quantitative structure-activity relationship studies using experiment and semiempirical molecular orbital theory revealed that the hydrophobicity of a TFMS and its molecular dipole moment were the principal determinants of mitochondrial uncoupling activity within the pKa range examined.

Molecular modeling: a tool for predicting anthelmintic activity in vivo.

The structural and electronic features of the broad-spectrum benzimidazole anthelmintic mebendazole [MBZ, methyl 5-(benzoyl)-benzimidazole-2-carbamate] have been determined using a combination of quantum mechanics, molecular graphics, and molecular modeling techniques. Using conformational analyses and quantum mechanics, we found that the three-dimensional structure and electronic features of MBZ were consistent with those previously reported for highly active broad-spectrum benzimidazole anthelmintics and that in vivo drug efficacy against Hymenolepis diminuta depends upon the orientation of the benzoyl group at position 5 on the heterocyclic ring system, the magnitude of the molecular dipole moment, and the percentage of polar surface area. The chemotherapeutic actions of MBZ on H. diminuta in vivo were accompanied by marked changes in worm weight and chemical composition. Tapeworms recovered from rats that had received a therapeutically effective dose of MBZ 24 h earlier were significantly smaller and contained much less glycogen (as a percentage of the wet weight) than worms from untreated controls. In MBZ-treated worms, protein concentrations rose at a rate sufficient to offset the decline in glycogen concentration. Glycogen/protein ratios in MBZ-treated worms were considerably lower than the corresponding control values. Differences in the absolute amounts of glycogen between control and drug-treated worms were even more profound. Administration of a curative dose of MBZ to the rat host produced in H. diminuta another change, the onset of which coincided with the gross alterations in worm weight and chemical composition.(ABSTRACT TRUNCATED AT 250 WORDS)

Conformation of MgATP bound to nucleotidyl and phosphoryl transfer enzymes 1H-transferred NOE measurements on complexes of methionyl tRNA synthetase and pyruvate kinase.

The conformations of MgATP bound to a nucleotidyl transfer enzyme, methionyl tRNA synthetase and a phosphoryl transfer enzyme, pyruvate kinase, were studied by transferred NOE (TRNOE) measurements in 1H NMR. The experiments were performed on D2O solutions at 276 MHz and 300 MHz, and 10 degrees C in the presence of approximately a tenfold excess of substrate over the enzyme (sites). Selective inversion of chosen resonances was accomplished with an appropriately tailored DANTE sequence consisting of 100 phase-alternating hard 1.8 degree pulses. NOE measurements were made in terms of difference spectra (with and without inversion) at 6-8 delay times ranging from 10-500 ms following the DANTE sequence. A full complement of ten NOE build-up curves obtained for each enzyme complex was analyzed by using the complete relaxation-matrix method (which includes all the non-exchangeable protons in MgATP) suitably modified to include exchange between bound and free substrate. Molecular mechanics computations were used to examine the energetic implications of the NOE-determined structure. The final structures obtained for MgATP bound to the two enzymes were very similar to each other, with a 3'-endo sugar pucker and an anti conformation with a glycosidic torsional angle (O'4-C'1-N9-C8) of 39 degrees +/- 4 degrees. Both enzymes contain multiple binding sites for MgATP and hence the structure obtained in each case represents an average due to chemical exchange. However, TRNOE experiments performed on a tryptic fragment of methionyl tRNA synthetase which has a single MgATP binding site, show that the same structure fits these measurements as well. This evidence, coupled with the striking similarity of the structures deduced, for the two enzyme complexes, and the reciprocal sixth-power dependence of NOE on interproton distance, strongly suggests that the conformations at the individual binding sites of both the enzymes are virtually identical. This conclusion is in contrast with multiple conformations of MgATP bound to pyruvate kinase, proposed by Rosevear, P.R., Fox, T.L. & Mildvan, A.S. (1987) Biochemistry 26, 3487-3493.

Efficacy of albendazole and mebendazole against Hymenolepis microstoma and Hymenolepis diminuta.

An investigation of the chemotherapeutic effects of 2 anthelmintics, albendazole (ABZ, methyl 5-[propylthio]benzimidazole-2- carbamate) and mebendazole (MBZ, methyl 5-[benzoyl]benzimidazole-2-carbamate), on Hymenolepis microstoma and Hymenolepis diminuta in experimentally infected mice and rats is reported. Single (50 mg/kg) or multiple daily oral doses (50 mg kg-1 day-1 for 3 consecutive days) of MBZ had no effect on H. microstoma; at necropsy, the drug treated mice harbored appreciable numbers of the parasite in the bile duct and biliary passages. ABZ was also inactive when given as a single oral 50 mg/kg dose on day 27 PI. Better results were obtained when ABZ was administered at a dosage of 50 mg kg-1 day-1 for 3 consecutive days; the reduction in worm burden obtained with this treatment regimen was 50%. These results are in marked contrast to those obtained with the same anthelmintics against enteral H. diminuta in rats which succumbed at lower dosages. A review was made of the published reports on the pharmacokinetic behavior of these benzimidazole carbamate anthelmintics and a hypothesis for the inactivity of MBZ against H. microstoma is proposed.

Muscimol and N,N-dimethylmuscimol: from a GABA agonist to a glycine antagonist.

By using molecular modeling methods, a molecular mechanism was identified which can explain how the incorporation of two methyl groups in place of two hydrogen atoms on the terminal nitrogen atom of muscimol can not only convert this potent agonist at GABAnergic receptors to an inactive molecule at these receptors, but also can convert this new derivative to an antagonist of glycine at glycinergic receptors. This insight into the molecular mechanism operative in the conversion of physiological function provides a basis for understanding how a single molecule may be able to act at both the GABA- and glycine-inhibitory receptors.