Computational methods in molecular diversity and combinatorial chemistry.

Molecular diversity, combinatorial chemistry and automated synthesis are helping usher in a new age in medicinal chemistry. The tools and practices of computational chemistry and molecular modeling are rising to the challenges and opportunities presented by the current trends in drug discovery and design. Recent advances include a number of new and meaningful measures of molecular diversity and the use of genetic algorithms to help design diverse libraries.

Direct prediction of dissociation constants (pKa's) of clonidine-like imidazolines, 2-substituted imidazoles, and 1-methyl-2-substituted-imidazoles from 3D structures using a comparative molecular field analysis (CoMFA) approach.

The applicability of a comparative molecular field analysis (CoMFA) method to reproduce and predict the pKa values of 28 clonidine-like imidazoline analogues and 16 2-substituted imidazoles has been investigated with the GRID force field. Molecular fields calculated with an H+ probe and AM1 partial atomic charges produced a correlation with a small standard deviation and a high correlation coefficient with cross validation. It was concluded that the CoMFA treatment of electrostatic effects is suitable for predicting pKa values and thus for the examination of the electronic effects in 3D quantitative structure-activity relationships.

Designing combinatorial library mixtures using a genetic algorithm.

The design of combinatorial mixture libraries should take account of a number of factors. This paper describes the application of a genetic algorithm to optimizing the diversity of libraries while minimizing the effort that will be needed to deconvolute the biological hits by mass-spectroscopic techniques. It differs from previous applications of genetic algorithms to combinatorial library design in that each chromosome encodes an entire library with the result that properties of the library are optimized. Our method is such that it is easily extensible to optimizing the distributions of any number of physical or other properties of the library. The method allows for the combinatorial constraint inherent in mixtures that every substituent at each diversity site must occur in combination with every substituent at every other site. We present results showing that the genetic algorithm can produce good library designs in timely manner.

Theoretical model-based equations for the linear free energy relationships of the biological activity of ionizable substances. 1. Equilibrium-controlled potency.

Because of the ambiguities of how to treat ionization in empirical equations which relate biological activity to partition coefficient by use of a (log P)2 term, a theoretical approach to the problem is proposed. Based on a simplified view of assays of potency following in vitro or continuous infusion administration of drugs, equations have been derived from a combination of mass law, equilibrium, and extrathermodynamic assumptions. In general form the equations which relate potency to partition coefficient (P) and degree of ionization (alpha) are the following. If the neutral form reacts with the receptor, log (1/C) =-log [1 + SIGMAM(DIPci) + sigman[aj/Pb(1-alpha4y]] + X. If the ionic form reacts with the receptor, log (1/C) =-log [1 + (1 - Alphan)/(alphan)[sigmam(diPci) + sigman[aj/Pb(1-alphaj)]]] + X. In this generalized model there are m nonaqueous compartments and n aqueous compartments of different pH. The parameters a, b, c, and d can be interpreted in terms of the model. The shape of the log (1/C) vs. log P curve may be asymptotic, linear, or composed of two portions of unequal slope which meet at an optimum or a bend. With the use of these equations it is possible to examine whether the ion or the neutral form is the active species and whether there is hydrophobic bonding to the receptor and/or an inert compartment. The models may be further extended to include terms other than log P and alpha.

A general and fast scoring function for protein-ligand interactions: a simplified potential approach.

A fast, simplified potential-based approach is presented that estimates the protein-ligand binding affinity based on the given 3D structure of a protein-ligand complex. This general, knowledge-based approach exploits structural information of known protein-ligand complexes extracted from the Brookhaven Protein Data Bank and converts it into distance-dependent Helmholtz free interaction energies of protein-ligand atom pairs (potentials of mean force, PMF). The definition of an appropriate reference state and the introduction of a correction term accounting for the volume taken by the ligand were found to be crucial for deriving the relevant interaction potentials that treat solvation and entropic contributions implicitly. A significant correlation between experimental binding affinities and computed score was found for sets of diverse protein-ligand complexes and for sets of different ligands bound to the same target. For 77 protein-ligand complexes taken from the Brookhaven Protein Data Bank, the calculated score showed a standard deviation from observed binding affinities of 1.8 log Ki units and an R2 value of 0.61. The best results were obtained for the subset of 16 serine protease complexes with a standard deviation of 1.0 log Ki unit and an R2 value of 0.86. A set of 33 inhibitors modeled into a crystal structure of HIV-1 protease yielded a standard deviation of 0.8 log Ki units from measured inhibition constants and an R2 value of 0.74. In contrast to empirical scoring functions that show similar or sometimes better correlation with observed binding affinities, our method does not involve deriving specific parameters that fit the observed binding affinities of protein-ligand complexes of a given training set. We compared the performance of the PMF score, Böhm's score (LUDI), and the SMOG score for eight different test sets of protein-ligand complexes. It was found that for the majority of test sets the PMF score performs best. The strength of the new approach presented here lies in its generality as no knowledge about measured binding affinities is needed to derive atomic interaction potentials. The use of the new scoring function in docking studies is outlined.

Evaluation of PMF scoring in docking weak ligands to the FK506 binding protein.

A new knowledge-based scoring function (PMF-score), implemented into the DOCK4 program, was used to screen a database of 3247 small molecules for binding to the FK506 binding protein (FKBP). The computational ranking of these compounds was compared to the binding affinities measured by NMR. It was demonstrated that small, weakly binding molecules have, on average, higher computational scores than nonbinders and are enriched in the upper ranks of the computational scoring lists. In addition, the results obtained with the PMF scoring function were superior (by 30-120% larger enrichment factors) to those obtained with the standard force field score of DOCK4. The reliable ranking of small, weakly binding molecules offers new ways of designing building blocks in combinatorial libraries as well as SAR by NMR libraries with the increased chance of identifying suitable lead compounds for drug design.

PLS analysis of distance matrices to detect nonlinear relationships between biological potency and molecular properties.

Although the statistical method of partial least squares (PLS) is widely used for the analysis of the relationship between molecular properties and biological potency, it is recognized that PLS detects only linear relationships. We tested two types of properties: simulated univariate data and electrostatic molecular field as a function of Hammett sigma constants. In both cases we compared relationships in which the function is linear, asymptotic, or rises to an optimum and then falls. We found that PLS analysis of the matrix of the distances between every pair of compounds detects all three types of relationships with the same quality of cross-validation. The successful application of the method requires that the distance matrices be constructed such that each contains only information about one property (for example, the electrostatic field around the functional group of interest). Carbo and Hodgkin similarities perform less well than distances.

Mechanistic interpretation of the genotoxicity of nitrofurans (antibacterial agents) using quantitative structure-activity relationships and comparative molecular field analysis.

Quantitative structure-activity relationship (QSAR) and comparative molecular field analysis (CoMFA) have been applied to elucidate the mechanisms of genotoxicity (SOSIP) of nitrofuran derivatives on Escherichia coli PQ37. The following equation was developed: log SOSIP = -33.1qc2 + 1.00 log P - 1.50Isat - 1.19MR - 0.76I5,6 - 3.76; n = 40, r = 0.900, s = 0.475. The QSAR model clearly reveals three important factors, namely, electronic (qc2), hydrophobic (log P) and steric (MR, Isat, I5,6) contributing toward the genotoxic activity of this class of compounds. qc2, the charge on the c2 atom attached to the NO2 group, supports a furan ring opening mechanism in explaining the genotoxicity. The finding of the coefficient of 1 with log P conforms to our previous findings with several different classes of mutagens acting on different systems. CoMFA analysis clearly demonstrates its potential in unraveling the steric features of the molecules through contour maps. The CoMFA cross-validated model also supports the importance of the electronic factor. It could not reveal any hydrophobic influence because the interaction energies of the CH3 and H2O probes are collinear. QSAR (classical) and CoMFA, if used judiciously, may complement each other and enhance the applicability of SAR in drug design.

Regression analysis of the relationship between physical properties and the in vitro inhibition of monoamine oxidase by propynylamines.

Regression analysis of the potency of inhibition of monoamine oxidase by 47 propynylamines revealed that there are three determinants of inhibitory potency: (1) the smallest substituent on the nitrogen must be methyl or hydrogen in order for any activity to be observed; (2) potency is parabolically related to pKa-the optimum pKa is 6.2; and (3) ortho-substituted benzylamine analogs are ten times more potent than predicted on the basis of pKa values. The optimum pKa cannot be explained by differences in fraction ionized but rather in terms of the multistep sequence whereby these compounds inhibit MAO. A very slight positive effect of hydrophobicity on potency was found. The potency of several analogs not included in the original analysis was predicted.

Structural requirements for the inhibition of 5-lipoxygenase by 15-hydroxyeicosa-5,8,11,13-tetraenoic acid analogues.

The structural requirements for inhibition of RBL-1 (rat basophilic leukemia) 5-lipoxygenase by 15-hydroxyeicosa-5,8,11,13-tetraenoic acid (15-HETE, 1) were studied by systematic chemical modifications of the molecule at the hydroxyl and carboxyl groups, the double bonds, and the carboxylate and omega side chains. The most potent inhibitors were analogues that contained a 5,8-cis,cis-diene system and acted as alternate substrates for the enzyme. However, several analogues in which the 5,8-diene had been reduced were also found to inhibit the enzyme. Inhibition of 5-lipoxygenase by 15-hydroxyeicosa-11,13-dienoic acid (15-HEDE) analogues was optimal in compounds that generally contained a free carboxyl group, a carboxylate side chain of nine carbons, an omega side chain of five or six carbons, a cis,trans- or trans,cis-11,13-diene or 11,13-diyne system, and a 15-hydroxyl group. Conversion of 15-HEDE to its 16-membered lactone reduced but did not eliminate 5-lipoxygenase inhibitory activity. In contrast, a 3- to 10-fold enhancement of activity occurred when 5,15-diHETE (58) or 5-HETE (56) were cyclized to their respective delta-lactones. Molecular modeling of 15-HEDE analogues, modified in the C11-C15 region, showed that inactive analogues protrude into regions in space not occupied by active analogues. These structural studies indicate that multiple regions are important for 5-lipoxygenase inhibition by both 15-HETE and 15-HEDE analogues and that no single region plays a predominant role in inhibition.

Conformationally defined adrenergic agents. 2. Catechol imidazoline derivatives: biological effects at alpha 1 and alpha 2 adrenergic receptors.

The synthesis and pharmacology of 2-(5,6-dihydroxy-1,2,3,4-tetrahydro-1-naphthyl)imidazoline (A-54741, 4), a very potent alpha-adrenergic agonist, are described. The change in biological activity resulting from variation of the carbocyclic ring size of 4 from four through seven members (2-5) is presented, as well as an explanation that accounts for this change in activity by considering the "exactness of fit" of these compounds to both the alpha 1- and alpha 2-adrenergic receptors. Compound 4 was found in vitro to be a full agonist with greater potency at the alpha 2 receptor (ED50 norepinephrine (NE)/ED50 4 = 188 +/- 22) than at the alpha 1 receptor (ED50 NE/ED50 4 = 13 +/- 2).

[(Aminomethyl)arloxy]acetic acid esters. A new class of high-ceiling diuretics. 3. Variation in the bridge between the aromatic rings to complete mapping of the receptor.

Continued structural evaluation of the [(aminomethyl)aryloxy]acetic ester diuretics has produced a series of compounds in which the functional group that bridges the two aromatic rings has been varied. Diuretic screening of these analogues in rats indicates that the keto group can be effectively replaced with an ether or thio ether function with a slight increase in potency, whereas the methylene and sulfoxide linking groups lead to diminished saluretic potency. Replacement with either -SO2-, -COCO-, -CH2O-, -CONH- or direct bond results in a loss of activity. Although the series was designed according to QSAR criteria, the traditional linear free-energy properties of these compounds do not correlate with diuretic potency. However, conformational analysis of the series by potential energy calculations indicates that all active compounds have an accessible conformation that matches the bridge atom-carboxylate distance of the very potent dihydrobenzofuran analogue 56. Conformational calculations of several compounds in which the aminomethyl group was varied suggests that the active conformation is probably a low-energy conformation. Consideration of rotation about the bridge could not distinguish between two possible orientations of the aminomethyl ring in the active conformation. However, there is a quantitative negative linear correlation between diuretic potency and the protrusion into space of the group that bridges the two aromatic rings.

Novel 3-Pyridyl ethers with subnanomolar affinity for central neuronal nicotinic acetylcholine receptors.

Recent evidence indicating the therapeutic potential of cholinergic channel modulators for the treatment of central nervous system (CNS) disorders as well as the diversity of brain neuronal nicotine acetylcholine receptors (nAChRs) have suggested an opportunity to develop subtype-selective nAChR ligands for the treatment of specific CNS disorders with reduced side effect liabilities. We report a novel series of 3-pyridyl ether compounds which possess subnanomolar affinity for brain nAChRs and differentially activate subtypes of neuronal nAChRs. The synthesis and structure-activity relationships for the leading members of the series are described, including A-85380 (4a), which possesses ca.50 pM affinity for rat brain [(3)H]-(-)-cytisine binding sites and 163% efficacy compared to nicotine to stimulate ion flux at human alpha4beta2 nAChR subtype, and A-84543 (2a), which exhibits 84-fold selectivity to stimulate ion flux at human alpha4beta2 nAchR subtype compared to human ganglionic type nAChRs. Computational studies indicate that a reasonable superposition of a low energy conformer of 4A with (S)-nicotine and (-)-epibatidine can be achieved.

Comparison of calculated versus measured partition coefficients of some phenyl beta-D-glucopyranosides.

Experimentally determined octanol-water partition coefficient values of substituted phenyl beta-D-glucopyranosides are compared with the calculated values using the computer program CLOGP. The systematic deviation of the calculated values from the measured ones in this series suggests that caution is required when calculations are performed on classes of compounds where many of the partition coefficients have not been experimentally determined.

Inhibitors of immune complex-induced inflammation: 3-[1-(2-benzoxazolyl)hydrazino]propanenitrile derivatives.

The octanol-water partition coefficients (log P) and the dissociation constants (pKa) of 3-[1-(2-benzoxazolyl)hydrazino]propanenitrile analogues have been determined, and quantitative structure--activity relationships (QSAR) of the analogues as inhibitors of immune complex-induced inflammation have been studied. A significant correlation is observed between log P and pi substituent constants, and between pKa and inductive-field (F) and resonance (R) constants. The QSAR equations indicate that smaller substituents both at the 5-position and/or at the side chain tend to make the compound more potent, while an electron-withdrawing group at the side chain tends to make the compound less potent. The predicted potencies of 14 of 18 additional monosubstituted and all six disubstituted analogues agree reasonably well with the observed activities.

Inhibitors of immune complex-induced inflammation: 5-substituted 3-[1-(2-benzoxazolyl)hydrazino]propanenitrile derivatives.

A number of 5-substituted 3-[1-(2-benzoxazolyl)hydrazino]propanenitrile analogues have been studied as inhibitors of the rat pleural reverse passive Arthus reaction, and quantitative structure-activity relationships (QSAR) of these analogues have been examined. The QSAR equations indicate that hydrophilic substituents at the 5-position produce more potent compounds, while electron-releasing groups decrease activity. The results supplement QSAR data we previously obtained from the dermal reverse passive Arthus reaction.

Quantitative structure-activity relationships of 5-lipoxygenase inhibitors. Inhibitory potency of pyridazinone analogues.

Quantitative structure-activity relationships (QSAR) of 72 1-phenyltetrahydropyridazin-3(2H)-one (I) analogues are examined for the inhibitory potency (IC50) of 5-lipoxygenase in a broken cell. The potency is increased by lipophilic substituents at the 3'- and 4'-positions. Substituents with positive F value at the 4'-position also increase the potency, while substituents at the 3'-position with a positive R value decrease it. The potency also decreases as the size of the 2'- and/or 4'-substituents increases. Thioketone analogues are about 5 times more potent than the corresponding carbonyl analogues.

Quantitative structure-activity relationships of inhibitors of immune complex-induced inflammation: 1-phenyl-3-aminopyrazoline derivatives.

Quantitative structure-activity relationships (QSAR) of the 1-phenyl-3-aminopyrazoline analogues as inhibitors of immune complex-induced inflammation have been studied. The correlation suggests that the overall size of the phenyl substituents are of importance, and bulky groups have negative effects on potency. The negative steric effects are gradually increased from ortho to meta to para positions. The negative steric effects were sometimes altered by the electronic effects of the substituents. Electron-releasing groups on the phenyl ring increased potency, while electron-withdrawing groups decreased it. Ortho substituents, however, have unaccounted for additional deleterious effects described here with an indicator variable. The octanol-water partition coefficient (log P) and dissociation constants (pKa) of the 1-(m-trifluoromethylphenyl)-3-aminopyrazoline analogue have been experimentally determined.

Molecular diversity: how we measure it? Has it lived up to its promise?

Molecular diversity needs to be considered when designing a screening set, whether a set of individual compounds or a combinatorial library. We have found that traditional substructure descriptors encode information relative to the biological properties of molecules. Of the methods tested, Ward's clustering is most effective. Combinatorial library design also requires consideration of the method of deconvolution and the combinatorial constraint. We have developed a genetic algorithm solution to this problem. Lastly, predicting affinity of molecules for a macromolecular target has been addressed with the use of a potential of mean force. There continues to be an opportunity for innovative computational approaches to solving problems in medicinal chemistry.

An evaluation of structural descriptors and clustering methods for use in diversity selection.

An evaluation of the suitability of a number of structural descriptors and clustering methods for use in diversity selection is presented. The methods are compared by their success in simulated biological activity predictions. The results suggest that simple 2D structural descriptors are particularly effective and that hierarchical clustering methods are superior to the non-hierarchical methods traditionally used for diversity related tasks. Results are presented which suggest that the difference in the utility of the descriptors can be accounted for by the different extent to which each encodes information relevant to the forces of ligand-receptor binding.