Three-dimensional quantitative structure-activity relationship study of nonsteroidal estrogen receptor ligands using the comparative molecular field analysis/cross-validated r2-guided region selection approach

A newly developed comparative molecular field analysis (CoMFA) technique, the cross-validated r2-guided region selection (CoMFA/q2-GRS) method, has been used to build a quantitative structure-activity relationship (3D-QSAR) for nonsteroidal estrogen receptor (ER) ligands. Ligands included in this study belong to a series of diethylstilbestrol (DES) and indenestrol analogues whose affinities for the mouse ER (mER) have been determined in our laboratory. The final model utilized 30 compounds and yielded a q2GRS (cross-validated r2, guided region selection) of 0.796, as compared to a q2 of 0.720 for conventional CoMFA, with a standard error of prediction of 0.594 at 3 principal components. This model was used to visualize steric and electrostatic features of the ligands that correspond with ER binding affinity. Results obtained from the CoMFA steric and electrostatic plots of this model have also been compared to information from the ER binding affinities of substituted estradiol analogues. This is in an effort to determine structural features of compounds in the CoMFA analysis that may correspond to those of the estradiol analogues and to further clarify the mode of binding of nonsteroidal ER ligands.     

Comparative molecular field analysis of a series of paclitaxel analogues

A series of 94 paclitaxel analogues exhibiting antitumor activity by promoting the assembly of microtubules and inhibiting the disassembly process of microtubules to tubulin were investigated using the comparative molecular field analysis (CoMFA) method. These compounds belonging to 10 structural classes were randomly divided into a training set of 80 compounds and a test set of 14 compounds. Since the three-dimension structure of ligand--receptor complex is unknown, from X-ray and NMR data we rationally selected the three-dimension structure of paclitaxel in a polar solution as the active conformation and starting structure for molecule modeling, the other molecules were aligned using this molecule model as the template. The most optimal CoMFA yielded a two-components model, with significant cross-validation r2cv of 0.640 and conventional r2 of 0.868. The predictive ability of training set model was tested on the test set of 14 compounds. The tests not only revealed the robustness of the CoMFA model but demonstrated that for our model r2pred based on the mean activity of test set compounds can accurately estimate external predictivity but r2pred based on the mean activity of training set compounds overestimated the model. The CoMFA model explained why the activity of taxoid is sensitive to the stereochemistry of the atoms at C-2' and C-3' positions and the presence of hydroxyl group at C-2' position. The other factors affecting activity were also elucidated according to standard coefficient contour maps of steric and electrostatic fields derived from the CoMFA model.     

A new procedure for improving the predictiveness of CoMFA models and its application to a set of dihydrofolate reductase inhibitors

A new automated procedure to improve the predictive quality of CoMFA models for both training and test sets is described. A model of greater consistency is generated by performing small reorientations of the underlying molecules for which too low activities are calculated. In order to predict activities of test compounds, the most similar molecules in the previously optimized model are identified and used as a basis for the prediction. This method has been applied to two independent sets of dihydrofolate reductase inhibitors (80 compounds each, serving as training sets), resulting in a significant increase of the cross-validated r2 value. For both models, the predictive r2 value for a test set consisting of 70 compounds was improved substantially.     

Synthesis, biochemical evaluation, and classical and three-dimensional quantitative structure-activity relationship studies of 7-substituted-1,2,3,4-tetrahydroisoquinolines and their relative affinities toward phenylethanolamine N-methyltransferase and the alpha2-adrenoceptor

7-Substituted-1,2,3,4-tetrahydroisoquinolines (7-substituted-THIQs) are potent inhibitors of phenylethanolamine N-methyltransferase (PNMT, EC 2.1.1.28), the enzyme involved in the biosynthesis of epinephrine. Unfortunately, most of these compounds also exhibit strong affinity for the alpha2-adrenoceptor. To design a selective (PNMT vs alpha2-adrenoceptor affinity) inhibitor of PNMT, the steric and electrostatic factors responsible for PNMT inhibitory activity and alpha2-adrenoceptor affinity were investigated by evaluating a number of 7-substituted-THIQs. A classical quantitative structure-activity relationship (QSAR) study resulted in a three-parameter equation for PNMT (PNMT pKi = 0.599pi - 0.0725MR + 1. 55sigmam + 5.80; n = 27, r = 0.885, s = 0.573) and a three-parameter equation for the alpha2-adrenoceptor (alpha2 pKi = 0.599pi - 0. 0542MR - 0.951sigmam + 6.45; n = 27, r = 0.917, s = 0.397). These equations indicated that steric effects and lipophilicity play a similar role at either active site but that electronic effects play opposite roles at either active site. Two binding orientations for the THIQs were postulated such that lipophilic and hydrophilic 7-substituents would not occupy the same region of space at either binding site. Using these two binding orientations, based on the lipophilicity of the 7-substituent, comparative molecular field analysis (CoMFA) models were developed that showed that the steric and electrostatic interactions at both sites were similar to those previously elaborated in the QSAR analyses. Both the QSAR and the CoMFA analyses showed that the steric interactions are similar at the PNMT active site and at the alpha2-adrenoceptor and that the electrostatic interactions were different at the two sites. This difference in electrostatic interactions might be responsible for the selectivity of THIQs bearing a nonlipophilic electron-withdrawing group at the 7-position. These QSAR and CoMFA results will be useful in the design of potent and selective (PNMT vs alpha2-adrenoceptor affinity) inhibitors of PNMT.     

Molecular modeling of phenothiazines and related drugs as multidrug resistance modifiers: a comparative molecular field analysis study

A set of 40 phenothiazines, thioxanthenes, and structurally related drugs with multidrug resistance modulating activity in tumor cells in vitro were selected from literature data and subjected to three-dimensional quantitative structure-activity relationship study using comparative molecular field analysis (CoMFA). More than 350 CoMFA models were derived and evaluated using steric, electrostatic, and hydrophobic fields alone and in combination. Four alignment strategies based on selected atom pairs or field fit alignment were compared. Several training and test sets were analyzed for both neutral and protonated drug forms separately. Each chemical class was trained and tested individually, and finally the classes were combined together into integrated models. All models obtained were statistically significant and most of them highly predictive. All fields contributed to MDR reversing activity, and hydrophobic fields improved the correlative and predictive power of the models in all cases. The results point to the role of hydrophobicity as a space-directed molecular property to explain differences in anti-MDR activity of the drugs studied.     

The local minima method (LMM) of pharmacophore determination: a protocol for predicting the bioactive conformation of small, conformationally flexible molecules

Software has been developed for potential energy surface analysis and the local minima method of pharmacophore determination. LMM is rigorous and systematic and employs multiple conformations which are the local minima from the potential energy surface of each compound in the data set. It produces a series of possible pharmacophores from a postulated set of pharmacophore elements. The best pharmacophore is then determined by performing a comparative molecular field analysis (CoMFA) on each one. The pharmacophore which produces the most self-consistent model is deemed the best. Local minima on the gas-phase potential energy surface are shown to be a reasonably close approximation to protein bound conformations, and these conformations can be found through systematic conformational searches followed by minimization of the local minima. LMM was used to develop a 3D-QSAR model for dopamine beta-hydroxylase (DBH) inhibitors which was highly predictive (predictive R2 = 0.71 and standard error of predictions = 0.41). The model predicted that the phenyl and thienyl series of inhibitors were acting as bioisosteres. Examination of compounds overlayed in the model indicated a possible hydrogen bond acceptor in the DBH active site. Three tyrosine residues previously labeled by mechanism based inhibitors may be acting as the acceptor and therefore represent excellent candidates for site-directed mutagenesis studies.     

Structural analysis by the comparative molecular field analysis method of the affinity of beta-adrenoreceptor blocking agents for 5-HT1A and 5-HT1B receptors

The affinities of 17 beta-adrenoreceptor antagonists for 5-HT1A and 5-HT1B receptors were evaluated in binding assays. A large range of Ki values (2-10,000 nM) was observed and ortho or meta substitution of the aromatic ring carrying the amino chain was implicated in the high affinity Ki values, whereas para substitution elicited a dramatic drop in activity. These variations were analyzed with two molecular design tools: the active analogue approach (AAA) and the new 3D-QSAR (quantitative structure activity relationship) method, comparative molecular field analysis (CoMFA). The AAA method emphasized, by superimposition of selected conformations of the molecules, the favorable and unfavorable volumes implicated in the receptor recognition. CoMFA generated a linear expression between the biological data and the different values of electrostatic and steric fields surrounding the molecules. It predicted the values of selected molecules but also those of new molecules not included in the study. The excellent accuracy of the prediction revealed the potential of the method for the design of new compounds. CoMFA demonstrated the important contribution of steric parameters, evaluated at 92%, compared to the electrostatic field (evaluated at 8%) to explain the affinity for 5-HT1A and 5-HT1B receptors. This study emphasizes also the importance of the occupancy of a hydrophobic pocket in the receptor site located near the area interacting with the aromatic moiety, and subsequently its use for the design of new, potent, specific antagonists of 5-HT1A and 5-HT1B receptors.     

Synthesis and comparative molecular field analysis (CoMFA) of antitumor 3-arylisoquinoline derivatives

In this study a series of 3-arylisoquinoline derivatives were synthesized and cytotoxicity against human melanoma tumor cell evaluated, and a three dimensional quantitative structure-activity relationship was investigated using the comparative molecular field analysis (CoMFA). The results suggested that the electrostatic, steric and hydrophobic factors of 3-arylisoquinolines were strongly correlated with the antitumor activity. Considerable predictive ability (cross-validated r2 as high as 0.721) was obtained through CoMFA.     

NMR and topochemical studies of peptidomimetic HIV-I protease inhibitors containing a cis-epoxide amide isostere

NMR and topochemical studies of irreversible HIV-1 protease inhibitors containing a cis-epoxide as amide isostere have been carried out to identify conformational preference of the inhibitors in solution. The inhibitors prefer to adopt extended conformations similar to the beta-strand in solution.     

Substrates and inhibitors of human T-cell leukemia virus type I protease

HTLV-I is an oncogenic retrovirus that is associated with adult T-cell leukemia. HTLV-I protease and HTLV-I protease fused to a deca-histidine containing leader peptide (His-protease) have been cloned, expressed, and purified. The refolded proteases were active and exhibited nearly identical enzymatic activities. To begin to characterize the specificity of HTLV-I, we measured protease cleavage of peptide substrates and inhibition by protease inhibitors. HTLV-I protease cleavage of a peptide representing the HTLV-I retroviral processing site P19/24 (APQVLPVMHPHG) yielded Km and kcat values of 470 microM and 0.184 s-1 while cleavage of a peptide representing the processing site P24/15 (KTKVLVVQPK) yielded Km and kcat values of 310 microM and 0.0060 s-1. When the P1' proline of P19/24 was replaced with p-nitro-phenylalanine (Nph), the ability of HTLV-I protease to cleave the substrate (APQVLNphVMHPL) was improved. Inhibition of HTLV-I protease and His-protease by a series of protease inhibitors was also tested. It was found that the Ki values for inhibition of HTLV-I protease and His-protease by a series of pepsin inhibitors ranged from 7 nM to 10 microM, while the Ki values of a series of HIV-1 protease inhibitors ranged from 6 nM to 127 microM. In comparison, the Ki values for inhibition of pepsin by the pepsin inhibitors ranged from 0.72 to 19.2 nM, and the Ki values for inhibition of HIV-1 protease by the HIV protease inhibitors ranged from 0.24 nM to 1.0 microM. The data suggested that the substrate binding site of HTLV-I protease is different from the substrate binding sites of pepsin and HIV-1 protease, and that currently employed HIV-1 protease inhibitors would not be effective for the treatment of HTLV-I infections.     

Coevolutionary analysis of resistance-evading peptidomimetic inhibitors of HIV-1 protease

We have developed a coevolutionary method for the computational design of HIV-1 protease inhibitors selected for their ability to retain efficacy in the face of protease mutation. For HIV-1 protease, typical drug design techniques are shown to be ineffective for the design of resistance-evading inhibitors: An inhibitor that is a direct analogue of one of the natural substrates will be susceptible to resistance mutation, as will inhibitors designed to fill the active site of the wild-type or a mutant enzyme. Two design principles are demonstrated: (i) For enzymes with broad substrate specificity, such as HIV-1 protease, resistance-evading inhibitors are best designed against the immutable properties of the active site-the properties that must be conserved in any mutant protease to retain the ability to bind and cleave all of the native substrates. (ii) Robust resistance-evading inhibitors can be designed by optimizing activity simultaneously against a large set of mutant enzymes, incorporating as much of the mutational space as possible.     

Comparative molecular field analysis of non-steroidal aromatase inhibitors: an extended model for two different structural classes

Aromatase is a cytochrome P450 isozyme, whose inhibition is known to be therapeutically relevant in the treatment of the breast cancer. A comparative molecular field analysis (CoMFA) has been carried out on a series of non-steroidal aromatase inhibitors belonging to two different structural classes. One subset of compounds consists of fadrozole analogues and was studied in a previous work, from which a 'local' 3-D quantitative structure-activity relationship (QSAR) model for the inhibition of aromatase was obtained. In the present paper, that model is extended to include a second subset of compounds bearing a tetralone nucleus and acting at the same enzyme site with the same mechanism as the azoles. The critical alignment step has been solved by using two different steroidal inhibitors of aromatase as rigid templates, on which the non-steroidal compounds have been superimposed. The final 3-D QSAR models are discussed in terms of predictivity and some implications regarding the steric and electronic requirements of steroidal and non-steroidal inhibitors are pointed out.     

Nonpeptide cyclic cyanoguanidines as HIV-1 protease inhibitors: synthesis, structure-activity relationships, and X-ray crystal structure studies

Comparison of the high-resolution X-ray structures of the native HIV-1 protease and its complexes with the inhibitors suggested that the enzyme flaps are flexible. The movement at the tip of the flaps could be as large as 7 A. On the basis of this observation, cyclic cyanoguanidines have been designed, synthesized, and evaluated as HIV-1 protease (PR) inhibitors. Cyclic cyanoguanidines were found to be very potent inhibitors of HIV-1 protease. The choice of cyclic cyanoguanidines over cyclic guanidines was based on the reduced basicity of the former. X-ray structure studies of the HIV PR complex with cyclic cyanoguanidine demonstrated that in analogy to cyclic urea, cyclic cyanoguanidines also displace the unique structural water molecule. The structure-activity relationship of the cyclic cyanoguanidines is compared with that of the corresponding cyclic urea analogues. The differences in binding constants of the two series of compounds have been rationalized using high-resolution X-ray structure information.     

Comparative molecular field analysis-based prediction of drug affinities at recombinant D1A dopamine receptors

Determination of quantitative structure-activity relationship (QSAR) for affinity at particular dopamine (DA) receptors has become an even greater priority with the cloning of five DA receptor subtypes. The use of agonist affinity at recombinant receptors selectively expressed in clonal cells as the dependent variable in QSAR presents a unique opportunity for accuracy and precision in measurement of biological values. Bound conformations of 11 agonists (for which both affinity data at the recombinant D1A DA receptor and stereochemical configurations were available) were determined by alignment with a template compound, SKF38393, which shows high affinity and selectivity for D1A receptors and is fairly rigid in structure. These aligned structures suggested a 3-point pharmacophore map (one cationic nitrogen and two electronegative centers) of the D1A DA receptor. This map shows both similarities and differences when compared with a previously reported D2 DA receptor pharmacophore map based on biological data from rat brain and with a recently published map of the native D1 DA receptor using several semirigid compounds. Log(1/K(d)) values at recombinant D1A DA receptors were used as the target property for a CoMFA (comparative molecular field analysis) of the 11 aligned structures. The resulting CoMFA model yielded a cross-validated r(2)(q(2)) value of 0.829 and a simple r(2) = 0.96. In contrast, when a CoMFA model was developed for 10 of these compounds using striatal D1 K(d) values, the q(2) value was reduced to 0.178. These results are consistent with the idea that drug affinity data obtained from clonal cells expressing recombinant receptors may be superior to that obtained using heterogeneous mixtures of native receptors prepared from brain membranes. The predictive utility of the CoMFA model was evaluated using several high-affinity dopamine agonists and m- and p-tyramine, two compounds with a single hydroxyl group on the aromatic ring. Predictions were fairly accurate for all compounds but the two tyramines.