Evaluation of quantitative structure-activity relationship methods for large-scale prediction of chemicals binding to the estrogen receptor

Structure-activity approach to the identification of environmental estrogens: the MCASE approach

A sizable number of environmental contaminants and natural products have been found to possess hormonal activity and have been termed endocrine-disrupting chemicals. Due to the vast number (estimated at about 58,000) of environmental contaminants, their potential to adversely affect the endocrine system, and the paucity of health effects data associated with them, the U.S. Congress was led to mandate testing of these compounds for endocrine-disrupting ability. Here we provide evidence that a computational structure-activity relationship (SAR) approach has the potential to rapidly and cost effectively screen and prioritize these compounds for further testing. Our models were based on data for 122 compounds assayed for estrogenicity in the ESCREEN assay. We produced two models, one for relative proliferative effect (RPE) and one for relative proliferative potency (RPP) for chemicals as compared to the effects and potency of 17beta-estradiol. The RPE and RPP models achieved an 88 and 72% accurate prediction rate, respectively, for compounds not in the learning sets. The good predictive ability of these models and their basis on simple to understand 2-D molecular fragments indicates their potential usefulness in computational screening methods for environmental estrogens.

A quantitative structure-antifungal activity relationship study of oxygenated aromatic essential oil compounds using data structuring and PLS regression analysis

Twenty two oxygenated aromatic essential oil compounds were chosen for the study of the antifungal activity against two wood-decaying fungi, the white-rot Trametes versicolor, which mainly metabolizes lignin, and the brown-rot Coniophoha puteana, which digests cellulose in plant cell walls. Minimal inhibitory concentrations (MICs) were determined by the agar dilution method, using dimethyl sulfoxide (DMSO) as the solvent for the selected compounds and potato-dextrose agar (PDA) as the growth medium for both fungi. The MICs were then used to generate a tree structure, which represents the structuring of the essential oil compounds by the nature and position of the substituents in their aromatic rings, and as dependent variables (log(1/MIC)) in the QSAR analysis. Data structuring proved that a relationship between the molecular structures of the essential oil compounds and their antifungal activity exists, and the hypotheses derived therefrom were complemented by performing a QSAR analysis using the partial least squares (PLS) method. Statistically significant PLS models were obtained with the 1-octanol-water partition coefficient (C log P), the energy of the highest occupied molecular orbital (E(HOMO)), and the number of hydrogen-bond donor atoms in the molecules of the compounds studied (Donor) for T. versicolor and with C log P and the fractional negative surface area (FNSA1) for C. puteana.

Novel topological index F based on incidence matrix

Based on incidence matrix W, the novel topological index F is defined by the matrices L, W, X as F = LWX. The properties, the chemical environments, and the interaction of the vertexes in a molecule are taken into account in this definition. Several good QSPR models in the hetero-atom-containing organic compounds and inorganic compounds are obtained. Moreover, based on the definition of F and the values C(i) of the vertexes or the values (m)F(ij) of the chemical bonds, we have obtained serial indices, (m)F(v), (m)F(b), and F(w). They are successfully applied to QSPR models and good correlation results have been obtained as well.

Comparative molecular field analysis (CoMFA) model using a large diverse set of natural, synthetic and environmental chemicals for binding to the androgen receptor

A large number of natural, synthetic and environmental chemicals are capable of disrupting the endocrine systems of experimental animals, wildlife and humans. These so-called endocrine disrupting chemicals (EDCs), some mimic the functions of the endogenous androgens, have become a concern to the public health. Androgens play an important role in many physiological processes, including the development and maintenance of male sexual characteristics. A common mechanism for androgen to produce both normal and adverse effects is binding to the androgen receptor (AR). In this study, we used Comparative Molecular Field Analysis (CoMFA), a three-dimensional quantitative structure-activity relationship (3D-QSAR) technique, to examine AR-ligand binding affinities. A CoMFA model with r2 = 0.902 and q2 = 0.571 was developed using a large training data set containing 146 structurally diverse natural, synthetic, and environmental chemicals with a 10(6)-fold range of relative binding affinity (RBA). By comparing the binding characteristics derived from the CoMFA contour map with these observed in a human AR crystal structure, we found that the steric and electrostatic properties encoded in this training data set are necessary and sufficient to describe the RBA of AR ligands. Finally, the CoMFA model was challenged with an external test data set; the predicted results were close to the actual values with average difference of 0.637 logRBA. This study demonstrates the utility of this CoMFA model for real-world use in predicting the AR binding affinities of structurally diverse chemicals over a wide RBA range.

HQSAR and CoMFA approaches in predicting reactivity of halogenated compounds with hydroxyl radicals

Two quantitative structure-activity relationship (QSAR) methods: hologram QSAR (HQSAR) and comparative molecular field analysis (CoMFa) were evaluated for predicting half-lives of the hydroxyl radicals reaction with substituted aromatic compounds. The HQSAR approach, which is topological in nature, results in a mathematical model which was more stable and has a greater predictive ability than the model derived on the 3-D CoMFA approach. Interpretations of the colour coded results of both methods are in good agreement with the proposed mechanism of the hydroxyl radical oxidation of halogenated aromatic compounds in the atmosphere.

Prediction of hERG potassium channel affinity by traditional and hologram qSAR methods

Traditional and hologram QSAR (HQSAR) models were developed for the prediction of hERG potassium channel affinities. The models were validated on three different test sets including compounds with published patch-clamp IC(50) data and two subsets from the World Drug Index (compounds indicated to have ECG modifying adverse effect and drugs marked to be approved, respectively). Discriminant analysis performed on the full set of hERG data resulted in a traditional QSAR model that classified 83% of actives and 87% of inactives correctly. Analysis of our HQSAR model revealed it to be predictive in both IC(50) and discrimination studies.

Modification of Wiener Index and its application

A novel topological index based on the Wiener Index is proposed as W* = 1/2 sigma (n)(i,j=1) S(*)(ij), the element S(*)(ij) of the distance matrix is defined either as S(*)(ij) = alpha x square root of I(i)I(j)/R(ij) (atoms i and j are adjacent) or as S(*)(ij) = = alpha x (j-i+1)square root of I(i) x x x x x I(j)/R(ij) (atoms i and j are not adjacent), where I(i) and I(j) represent the electronegativity of vertices i or j, respectively, R(ij)() is the sum of the bond length between the vertices i and j in a molecular graph, and alpha = (Z(i)/Z(j))(0.5), where Z(i) and Z(j) are the atomic numbers of the positive valence atom i and the negative valence atom j, respectively. The properties and the interaction of the vertices in a molecule are taken into account in this definition. That is why the application of the index W to heteroatom-containing and multiple bond organic systems and inorganic systems is possible. Correlation coefficients above 0.97 are achieved in the prediction of the retention index of gas chromatography of the hydrocarbons, the standard formation enthalpy of methyl halides, halogen-silicon, and inorganic compounds containing transition metals.

The signature molecular descriptor. 1. Using extended valence sequences in QSAR and QSPR studies

We present a new descriptor named signature based on extended valence sequence. The signature of an atom is a canonical representation of the atom's environment up to a predefined height h. The signature of a molecule is a vector of occurrence numbers of atomic signatures. Two QSAR and QSPR models based on signature are compared with models obtained using popular molecular 2D descriptors taken from a commercially available software (Molconn-Z). One set contains the inhibition concentration at 50% for 121 HIV-1 protease inhibitors, while the second set contains 12865 octanol/water partitioning coefficients (Log P). For both data sets, the models created by signature performed comparable to those from the commercially available descriptors in both correlating the data and in predicting test set values not used in the parametrization. While probing signature's QSAR and QSPR performances, we demonstrates that for any given molecule of diameter D, there is a molecular signature of height h </= D+1, from which any 2D descriptor can be computed. As a consequence of this finding any QSAR or QSPR involving 2D descriptors can be replaced with a relationship involving occurrence number of atomic signatures.

Molecular hologram derived quantitative structure-property relationships to predict physico-chemical properties of polychlorinated biphenyls

Polychlorinated biphenyls (PCBs) congeners with various degrees of chlorination and substitution patterns are among the most widespread and persistent man-made organic pollutants. They are toxic, lipophilic and tend to be bioaccumulated. The knowledge of the physico-chemical properties is very useful to explain the environmental behavior of PCBs and to perform an exposure assessment. In this paper, we have used a new molecular representation, the molecular hologram, to generate quantitative structure-property relationship models to predict the physico-chemical properties of biphenyl and all of its chlorinated congeners. The investigated properties include 1-octanol/water partition coefficient (logK(ow)), aqueous solubility (-logS(w)), aqueous activity coefficient (-logY(w)), Total molecular surface area, Henry's law constant (logH). The results show that this new quantitative structure-activity relationship approach presents highly predictive models for important physico-chemical properties of PCBs.

VSMP: a novel variable selection and modeling method based on the prediction

The use of numerous descriptors that are indicative of molecular structure and topology is becoming more common in quantitative structure-activity relationship (QSAR). How to choose the adequate descriptors for QSAR studies is important but difficult because there are no absolute rules to govern this choice. A variety of variable selection techniques including stepwise, partial least squares/principal component analysis (PLS/PCA), neural network, and evolutionary algorithm such as genetic algorithm have been applied to this common problem. All-subsets regression (ASR) is capable of finding out the best variable subset from among a large pool. In this paper, a novel variable selection and modeling method based on the prediction, for short VSMP, has been developed. Here two controllable parameters, the interrelation coefficient between the pairs of the independent variables (r(int)) and the correlation coefficient (q(2)) obtained using the leave-one-out (LOO) cross-validation technique, are introduced into the ASR to improve its performances. This technique differs from the other variable selection procedures related to the ASR by two main features: (1) The search of various optimal subset search is controlled by the statistic q(2) or root-mean-square error (RMSEP) in the LOO cross-validation step rather than the correlation coefficient obtained in the modeling step (r(2)). (2) The searching speed of all optimal subsets is expedited by the statistic r(int) together with q(2). A comparison of the results of the VSMP applied to the Selwood data set (n = 31 compounds, m = 53 descriptors) with those obtained from alternative algorithms shows the good performance of the technique.

Development and validation of an average mammalian estrogen receptor-based QSAR model

Development and evaluation of quantitative structure activity relationships (QSARs) for predicting estrogen receptor binding from chemical structure requires reliable algorithms for three-dimensional (3D) QSAR analysis and establishment of structurally diverse training sets of chemicals whose modes of action and measures of potency are well defined. One approach to selecting an appropriate training set is to minimize the biological variability in the model development, by using structurally restricted data sets. A second approach is to extend the structural diversity of chemicals at the cost of increased variability of biological assays. In this study, the second approach was used by organizing a training set of 151 chemicals with measured human alpha Estrogen Receptor (ERalpha), mouse uterine, rat uterine, and MCF7 cell Relative Binding Affinities (RBAs). The structurally augmented training set was submitted to a 3D pattern recognition analysis to derive a model for average mammalian ER binding affinity by employing the COmmon REactivity PAttern (COREPA) approach. Elucidation of this pattern required examination of the conformational flexibility of the compounds in an attempt to reveal areas in the multidimensional descriptor space, which are most populated by the conformers of the biologically active molecules and least populated by the inactive ones. The approach is not dependent upon a predetermined and specified toxicophore or an alignment of conformers to a lead compound. Reactivity patterns associated with mammalian ER binding affinity were obtained in terms of global nucleophilicity (E(HOMO)), interatomic distances between nucleophilic sites, and local nucleophilicity (charges or delocalizabilities) of those sites. Based on derived patterns, descriptor profiles were established for identifying and ranking compounds with RBA of > 150, 150-10, 10-1 and 1-0.1% relative to 17beta-estradiol. Specificity of reactivity profiles was found to increase gradually with increasing affinities associated with RBAs ranges under study. Using the results of this analysis, an exploratory expert system was developed for use in ranking relative mammalian ER binding affinity potential for large chemical data sets. The validity of the RBA predictions were confirmed by independent development and comparison with measured RBA values.

Influence of the structural diversity of data sets on the statistical quality of three-dimensional quantitative structure-activity relationship (3D-QSAR) models: predicting the estrogenic activity of xenoestrogens

Federal legislation has resulted in the two-tiered in vitro and in vivo screening of some 80 000 structurally diverse chemicals for possible endocrine disrupting effects. To maximize efficiency and minimize expense, prioritization of these chemicals with respect to their estrogenic disrupting potential prior to this time-consuming and labor-intensive screening process is essential. Computer-based quantitative structure-activity relationship (QSAR) models, such as those obtained using comparative molecular field analysis (CoMFA), have been demonstrated as useful for risk assessment in this application. In general, however, CoMFA models to predict estrogenicity have been developed from data sets with limited structural diversity. In this study, we constructed CoMFA models based on biological data for a structurally diverse set of compounds spanning eight chemical families. We also compared two standard alignment schemes employed in CoMFA, namely, atom-fit and flexible field-fit, with respect to the predictive capabilities of their respective models for structurally diverse data sets. The present analysis indicates that flexible field-fit alignment fares better than atom-fit alignment as the structural diversity of the data set increases. Values of log(RP), where RP = relative potency, predicted by the final flexible field-fit CoMFA models are in good agreement with the corresponding experimental values. These models should be effective for predicting the endocrine disrupting potential of existing chemicals as well as prospective and newly prepared chemicals before they enter the environment.

Fragmental approach in QSPR

Methodological problems of using fragmental descriptors for construction of QSAR/QSPR equations are considered, and the main achievements in this field are summarized and discussed. If a structure-property data set is sufficiently large to allow building statistically significant models, then any topological index can be replaced with a set of fragmental descriptors. Several examples of using the fragmental approach for predicting retention indices and the normal boiling points of organic compounds are considered. Advantages of using fragmental descriptors, namely a "transparency" and interpretability of QSAR/QSPR models, are exemplified.

Three-dimensional quantitative structure-activity relationship analysis of ligand binding to human sequence antidigoxin monoclonal antibodies using comparative molecular field analysis

The present study indicates that the newly generated human sequence antidigoxin monoclonal antibody (mAb), 1B3, binds digoxin with a different fine specificity binding than our previously obtained human sequence monoclonal antibodies (mAbs) (Ball, W. J.; et al. J. Immunol. 1999, 163, 2291-2298). Uniquely, 1B3 has a higher affinity for digitoxin than digoxin, the immunizing hapten, and a strong requirement for at least one sugar residue linked to the aglycone (-genin). By means of comparative molecular field analysis (CoMFA), the results of competition binding studies for 56 cardiotonic and hormonal steroids were employed to develop three-dimensional quantitative structure-activity relationship (3D-QSAR) models for ligand binding to 1B3 and to three additional human sequence mAbs, as well as the murine antidigoxin mAb 40-50 (Mudgett-Hunter, M.; et al. Mol. Immunol. 1985, 22, 447-488). All five 3D-QSAR models yielded cross-validated q(2) values greater than 0.5, which indicates that they have significant predictive ability. The CoMFA StDevCoeff contour plots, as well as the competition results, indicate that 1B3 binds ligands in a manner distinct from the other four mAbs. The CoMFA contour plots for 40-50 were also compared with the known X-ray crystallographic structure of the 40-50-ouabain complex (Jeffrey, P. D.; et al. J. Mol. Biol. 1995, 248, 344-360) in order to identify correlations between residues in the mAb binding site and specific contour plot regions. These 3D-QSAR models and their respective contour plots should be useful tools to further understand the molecular nature of antibody-antigen interactions and to aid in the redesign or enhancement of therapeutic antibodies.

New p-methylsulfonamido phenylethylamine analogues as class III antiarrhythmic agents: design, synthesis, biological assay, and 3D-QSAR analysis

Class III antiarrhythmic agents selectively delay the effective refractory period (ERP) and increase the transmembrane action potential duration (APD). Using dofetilide (2) as a template of class III antiarrhythmic agents, we designed and synthesized 16 methylsulfonamido phenylethylamine analogues (4a-d and 5a-l). Pharmacological assay indicated that all of these compounds showed activity for increasing the ERP in isolated animal atrium; among them, the effective concentration of compound 4a is 1.6 x 10(-8) mol/L in increasing ERP by 10 ms, slightly less potent than that of 2, 1.1 x 10(-8) mol/L. Compound 4a also produced a slightly lower change in ERP at 10(-5) M, DeltaERP% = 17.5% (DeltaERP% = 24.0% for dofetilide). On the basis of this bioassay result, these 16 compounds together with dofetilide were investigated by the three-dimensional quantitative structure-activity relationship (3D-QSAR) techniques of comparative molecular field analysis (CoMFA), comparative molecular similarity index analysis (CoMSIA), and the hologram QSAR (HQSAR). The 3D-QSAR models were tested with another 11 compounds (4e-h and 5m-s) that we synthesized later. Results revealed that the CoMFA, CoMSIA, and HQSAR predicted activities for the 11 newly synthesized compounds that have a good correlation with their experimental value, r(2) = 0.943, 0.891, and 0.809 for the three QSAR models, respectively. This indicates that the 3D-QSAR models proved a good predictive ability and could describe the steric, electrostatic, and hydrophobic requirements for recognition forces of the receptor site. On the basis of these results, we designed and synthesized another eight new analogues of methanesulfonamido phenylethyamine (6a-h) according to the clues provided by the 3D-QSAR analyses. Pharmacological assay indicated that the effective concentrations of delaying the ERP by 10 ms of these newly designed compounds correlated well with the 3D-QSAR predicted values. It is remarkable that the percent change of delaying ERP at 10(-5) M compound 6c is much higher than that of dofetilide; the effective concentration of compound 6c is 5.0 x 10(-8)mol/L in increasing the ERP by 10 ms, which is slightly lower than that of 2. The results showed that the 3D-QSAR models are reliable and can be extended to design new antiarrhythmic agents.

Homology modeling of the estrogen receptor subtype beta (ER-beta) and calculation of ligand binding affinities

Estrogen is a steroid hormone playing critical roles in physiological processes such as sexual differentiation and development, female and male reproductive processes, and bone health. Numerous natural and synthetic environmental compounds have been shown capable of estrogenic effects. This area has been the focus of significant fundamental and applied research due both to the potential detrimental effects of these compounds upon normal physiological processes and to the potential beneficial effects of tissue-selective estrogen agonists/antagonists for the prevention and treatment of numerous diseases. Genomic effects of the active form of estrogen, 17beta-estradiol, are mediated through at least two members of the steroid hormone receptor superfamily, estrogen receptor subtype alpha (ER-alpha) and estrogen receptor subtype beta (ER-beta). At the time of this work, the X-ray crystal structure of the ER-alpha had been elucidated, however, coordinates of the ER-beta were not publicly available. Based upon the significant structural conservation across members of the steroid hormone receptor family, and the high sequence homology between ER-alpha and ER-beta (>60%), we have developed a homology model of the ER-beta structure. Using the crystal structure of ER-alpha and the homology model of ER-beta, we demonstrate a strong correlation between computed values of the binding-energy and published values of the observed relative binding affinity (RBA) for a variety of compounds for both receptors, as well as the ability to identify receptor subtype selective compounds. Furthermore, using the recently available crystal structure of ER-beta for comparison purposes, we show that not only is the predicted homology model structurally accurate, but that it can be used to assess ligand binding affinities.

CoMFA and HQSAR of acylhydrazide cruzain inhibitors

An approach combining CoMFA and HQSAR methods was used to describe QSAR models for a series of cruzain inhibitors having the acylhydrazide framework. A CoMFA study using two alignment orientations (I and II), three different probe atoms and changes of the lattice spacing (1 and 2 A) was performed. Alignment II and an sp3 probe carbon atom yielded good cross-validation (q2=0.688) employing lattice spacing of 1 A. The best HQSAR model was generated using atoms, bond, and connectivity as fragment distinction and fragment size default (4-5) showing similar cross-validated value of CoMFA (q2=0.689). Based upon the information derived from CoMFA and HQSAR, we have identified some key features that may be used to design new acylhydrazide derivatives that may be more potent cruzain inhibitors.

Combined MEDV-GA-MLR method for QSAR of three panels of steroids, dipeptides, and COX-2 inhibitors

The MEDV-13, molecular electronegativity distance vector based on 13 atomic types, has at best 91 descriptors. It is impossible to indirectly use multiple linear regression (MLR) to derive a quantitative structure-activity relationship (QSAR) model. Although principal component regression (PCR) or partial least-squares regression (PLSR) can be employed to develop a latent QSAR model, it is still difficult how to determine the principal components (PCs) and depict the physical meaning of the PCs. So, a genetic algorithm (GA) is first employed to select an optimal subset of the descriptors from original MEDV-13 descriptor set. Then MLR is utilized to build a QSAR model between the optimal subset and the biological activities of three sets of compounds. For 31 benchmark steroids, a 5-descriptor QSAR model (M1) between the corticosteroid-binding globulin (CBG) binding affinity of the steroids and 5-descriptor subset is developed. The root-mean-square error of estimations (RMSEE) and the correlation coefficient of estimations (r) between the CBG binding affinity (BA) observed and the BA estimated by M1 are 0.422 and 0.9182, respectively. The root-mean-square error of predictions (RMSEP) and the correlation coefficient of predictions (q) between the BA observed and the BA predicted by leave-one-out cross validations are 0.504 and 0.8818, respectively. For 58 dipeptides inhibiting angiotensin-converting enzyme (ACE), a 5-variable QSAR model (M2) between the pIC(50) of peptides and 5-descriptor subset is derived. The M2 has a high quality with RMSEE = 0.339 and r = 0.9398 and RMSEP = 0.370 and q = 0.9280. For 16 indomethacin amides and esters (ImAE) inhibiting cyclooxygenase-2 (COX-2), a 6-variable QSAR model (M3) with RMSEE = 0.079 and r = 0.9839 and RMSEP = 0.151 and q = 0.9413 is built.

Xenoestrogenic gene expression: structural features of active polycyclic aromatic hydrocarbons

Estrogenicity was assessed using the Saccharomyces cerevisiae-based Lac-Z reporter assay and was reported as the logarithm of the inverse of the 50% molar beta-galactosidase activity (log[EC50(-1)]). In an effort to quantify the relationship between molecular structure of polycyclic aromatic hydrocarbons (PAHs) and estrogenic gene expression, a series of PAHs were evaluated. With noted exceptions, the results of these studies indicate that the initial two-dimensional structural warning for estrogenicity, the superpositioning of a hydroxylated aromatic system on the phenolic A-ring of 17-beta-estradiol, can be extended to the PAHs. This two-dimensional-alignment criterion correctly identified estrogenicity of 22 of the 29 PAHs evaluated. Moreover, the estrogenic potency of these compounds was directly related to the size of the hydrophobic backbone. The seven compounds classified incorrectly by this structural feature were either dihydroxylated naphthalenes or aromatic nitrogen-heterocyclic compounds; all such compounds were false positives. Results with dihydroxylated naphthalenes reveal derivatives that were nonestrogenic when superimposed on the phenolic A-ring of 17-beta-estradiol had the second hydroxyl group in the position of the C-ring or were catechol-like in structure. Structural alerts for nitrogen-heterocyclic compounds must take into account the position of the hydroxyl group and the in-ring nitrogen atom; compounds with the hydroxyl group and nitrogen atom involved with the same ring were observed to be nonactive.

An integrated "4-phase" approach for setting endocrine disruption screening priorities--phase I and II predictions of estrogen receptor binding affinity

Recent legislation mandates the US Environmental Protection Agency (EPA) to develop a screening and testing program for potential endocrine disrupting chemicals (EDCs), of which xenoestrogens figure prominently. Under the legislation, a large number of chemicals will undergo various in vitro and in vivo assays for their potential estrogenicity, as well as other hormonal activities. There is a crucial need for priority setting before this strategy can be effectively implemented. Here we report an integrated computational approach to priority setting using estrogen receptor (ER) binding as an example. This approach rationally integrates different predictive computational models into a "Four-Phase" scheme so that it can effectively identify potential estrogenic EDCs based on their predicted ER relative binding affinity (RBA). The system has been validated using an in-house ER binding assay dataset for 232 chemicals that was designed to have both broad structural diversity and a wide range of binding affinities. When applied to 58,000 chemicals identified by Walker et al. as candidates for endocrine disruption screening, some 9100 chemicals were predicted to bind to ER. Of these, only 3600 were expected to bind to ER at RBA values up to 100,000-fold less than that of 17beta-estradiol. The method ruled out 83% of the chemicals as non-binders with a very low rate of false negatives. We believe that the same integrated scheme will be equally applicable to endpoints of other endocrine disrupting mechanisms, e.g. androgen receptor binding.

Three-dimensional quantitative structure-activity relationship study of the inhibition of Na(+),K(+)-ATPase by cardiotonic steroids using comparative molecular field analysis

Na(+),K(+)-ATPase is a transmembrane protein that transports sodium and potassium ions across cell membranes during an activity cycle that uses the energy released by ATP hydrolysis. Cardiotonic steroids (digitalis) inhibit this activity and consequently produce a positive inotropic response in the heart. To identify the structural features of the steroids that are important for this inhibition, we have tested the inhibitory properties of 47 cardiotonic and hormonal steroids and developed a three-dimensional quantitative structure-activity relationship (3D-QSAR) model for the inhibition of Na(+),K(+)-ATPase using comparative molecular field analysis (CoMFA). We also developed a 3D-QSAR model for the binding of digoxin to the murine anti-digoxin monoclonal antibody (mAb) 26-10 because we have previously shown that the environment of the binding sites of 26-10 and the enzyme are similar (Kasturi et al. (1998) Biochemistry 37, 6658-6666). These statistically predictive 3D-QSAR models indicate that both binding sites are about 20 A long and have a close fit or complementarity about the beta side of the lactone ring of digitalis. Furthermore, steric bulk about the lactone ring and the alpha sugar may be critical for drug binding. However, the binding site of Na(+),K(+)-ATPase differs from that of mAb in that it has a greater number of electrostatic interactions along the alpha-sugar, steroid, and lactone moieties. In addition, the availability of the structure of the 26-10 Fab-digoxin complex (Jeffrey et al. (1993) Proc. Natl. Acad. Sci. U.S.A. 90, 10310-10314) enabled us to compare the CoMFA-derived contour maps with the known locations for amino acid residues comprising the mAb ligand binding site.

Elucidating the inhibiting mode of AHPBA derivatives against HIV-1 protease and building predictive 3D-QSAR models

The Lamarckian genetic algorithm of AutoDock 3.0 has been used to dock 27 3(S)-amino-2(S)-hydroxyl-4-phenylbutanoic acids (AHPBAs) into the active site of HIV-1 protease (HIVPR). The binding mode was demonstrated in the aspects of the inhibitor's conformation, subsite interaction, and hydrogen bonding. The data of geometrical parameters (tau(1), tau(2), and tau(3) listed in Table 2) and root mean square deviation values as compared with the known inhibitor, kni272,(28) show that both kinds of inhibitors interact with HIVPR in a very similar way. The r(2) value of 0.860 indicates that the calculated binding free energies correlate well with the inhibitory activities. The structural and energetic differences in inhibitory potencies of AHPBAs were reasonably explored. Using the binding conformations of AHPBAs, consistent and highly predictive 3D-QSAR models were developed by performing CoMFA, CoMSIA, and HQSAR analyses. The reasonable r(corss)(2) values were 0.613, 0.530, and 0.717 for CoMFA, CoMSIA, and HQSAR models, respectively. The predictive ability of these models was validated by kni272 and a set of nine compounds that were not included in the training set. Mapping these models back to the topology of the active site of HIVPR leads to a better understanding of vital AHPBA-HIVPR interactions. Structural-based investigations and the final 3D-QSAR results provide clear guidelines and accurate activity predictions for novel HIVPR inhibitors.

CoMFA and CoMSIA 3D-quantitative structure-activity relationship model on benzodiazepine derivatives, inhibitors of phosphodiesterase IV

Recently, we reported structurally novel PDE4 inhibitors based on 1,4-benzodiazepine derivatives. The main interest in developing bezodiazepine-based PDE4 inhibitors is in their lack of adverse effects of emesis with respect to rolipram-like compounds. A large effort has thus been made toward the structural optimization of this series. In the absence of structural information on the inhibitor binding mode into the PDE4 active site, 2D-QSAR (H-QSAR) and two 3D-QSAR (CoMFA and CoMSIA) methods were applied to improve our understanding of the molecular mechanism controlling the PDE4 affinity of the benzodiazepine derivatives. As expected, the CoMSIA 3D contour maps have provided more information on the benzodiazepine interaction mode with the PDE4 active site whereas CoMFA has built the best tool for activity prediction. The 2D pharmacophoric model derived from CoMSIA fields is consistent with the crystal structure of the PDE4 active site reported recently. The combination of the 2D and 3D-QSAR models was used not only to predict new compounds from the structural optimization process, but also to screen a large library of bezodiazepine derivatives.

The computational prediction of toxicity

Recent developments in the prediction of toxicity from chemical structure have been reviewed. Attention has been drawn to some of the problems that can be encountered in the area of predictive toxicology, including the need for a multi-disciplinary approach and the need to address mechanisms of action. Progress has been hampered by the sparseness of good quality toxicological data. Perhaps too much effort has been devoted to exploring new statistical methods rather than to the creation of data sets for hitherto uninvestigated toxicological endpoints and/or classes of chemicals.

3D-QSAR model of flavonoids binding at benzodiazepine site in GABAA receptors

With flavone as a structural template, three-dimensional quantitative structure-activity relationship (3D-QSAR) studies and ab initio calculations were performed on a series of flavonoids. A reasonable pharmacophore model was built through CoMFA, CoMSIA, and HQSAR analyses and electrostatic potential calculations. A plausible binding mode for flavonoids with GABA(A) receptors was rationalized. On the basis of the commonly recognized binding site, the specific S1 and S2 subsites relating to substituent positions were proposed. The different binding affinities could be explained according to the frontier orbitals and electrostatic potential (ESP) maps. The ESP could be used as a novel starting point for designing more selective BZ-binding-site ligands.

Classification of environmental estrogens by physicochemical properties using principal component analysis and hierarchical cluster analysis

A structurally diverse assortment of 60 environmental estrogens was divided into two main clusters ("A", "B") and a pair of subclusters ("C1", "C2") by applying principal component analysis to selected 1D and 2D molecular descriptors and subjecting the PCs to hierarchical cluster analysis. Although clustering was predicated solely on physicochemical properties, the dependence on particular physicochemical parameters of xenoestrogen binding affinities (pK(i)) to murine uterine cytosolic estrogen receptor (ER) proved greater for compounds within (sub)clusters than for compounds between (sub)clusters. Quantitative structure-binding affinity relationships derived using molecular descriptors and PCs suggested differences in the driving forces for xenoestrogen-ER binding for different (sub)clusters. The modeling power for xenoestrogen-ER binding affinities of a combination of TLSER and WHIM 3D indices was much greater than that of combinations of 1D and 2D molecular descriptors or the PCs derived therefrom. The clusterings obtained using PCs also proved applicable to the 3D-QSARs.

QSAR study of steroid benchmark and dipeptides based on MEDV-13

A molecular electronegativity distance vector based on 13 atomic types, called MEDV-13, is a descriptor for predicting the biological activities of molecules based on the quantitative structure-activity relations (QSAR). The MEDV-13 uses a modified electrotopological state (E-state) index to substitute for the relative eletronegativity (q) of non-hydrogen atoms in the molecule of interest in the MEDV and a topological distance for the relative distance (d) in the MEDV. For an organic molecule containing several chemical elements such as C, H, O, N, S, F, Cl, Br, I, and P, the MEDV-13 includes at best 91 descriptors. Then it is essential to employ a principal component regression (PCR) technique to derive a QSAR model relating the biological activities to the MEDV-13. The MEDV-13 is used to study the QSAR of the corticosteroid-binding globulin (CBG) binding affinity of the steroids and the activity inhibiting angiotensin-converting enzyme (ACE) of dipeptides, and resulting models have a comparable quality to the current three-dimensional (3D) methods such as CoMFA though the MEDV-13 is a descriptor based on two-dimensional topological information.

Structure-activity relationships for a large diverse set of natural, synthetic, and environmental estrogens

Understanding structural requirements for a chemical to exhibit estrogen receptor (ER) binding has been important in various fields. This knowledge has been directly and indirectly applied to design drugs for human estrogen replacement therapy, and to identify estrogenic endocrine disruptors. This paper reports structure-activity relationships (SARs) based on a total of 230 chemicals, including both natural and xenoestrogens. Activities were generated using a validated ER competitive binding assay, which covers a 10(6)-fold range. This study is focused on identification of structural commonalities among diverse ER ligands. It provides an overall picture of how xenoestrogens structurally resemble endogenous 17beta-estradiol (E(2)) and the synthetic estrogen diethylstilbestrol (DES). On the basis of SAR analysis, five distinguishing criteria were found to be essential for xenoestrogen activity, using E(2) as a template: (1) H-bonding ability of the phenolic ring mimicking the 3-OH, (2) H-bond donor mimicking the17beta-OH and O-O distance between 3- and 17beta-OH, (3) precise steric hydrophobic centers mimicking steric 7alpha- and 11beta-substituents, (4) hydrophobicity, and (5) a ring structure. The 3-position H-bonding ability of phenols is a significant requirement for ER binding. This contributes as both a H-bond donor and acceptor, although predominantly as a donor. However, the 17beta-OH contributes as a H-bond donor only. The precise space (the size and orientation) of steric hydrophobic bulk groups is as important as a 17beta-OH. Where a direct comparison can be made, strong estrogens tend to be more hydrophobic. A rigid ring structure favors ER binding. The knowledge derived from this study is rationalized into a set of hierarchical rules that will be useful in guidance for identification of potential estrogens.

QSAR models using a large diverse set of estrogens

Endocrine disruptors (EDs) have a variety of adverse effects in humans and animals. About 58,000 chemicals, most having little safety data, must be tested in a group of tiered assays. As assays will take years, it is important to develop rapid methods to help in priority setting. For application to large data sets, we have developed an integrated system that contains sequential four phases to predict the ability of chemicals to bind to the estrogen receptor (ER), a prevalent mechanism for estrogenic EDs. Here we report the results of evaluating two types of QSAR models for inclusion in phase III to quantitatively predict chemical binding to the ER. Our data set for the relative binding affinities (RBAs) to the ER consists of 130 chemicals covering a wide range of structural diversity and a 6 orders of magnitude spread of RBAs. CoMFA and HQSAR models were constructed and compared for performance. The CoMFA model had a r2 = 0.91 and a q2LOO = 0.66. HQSAR showed reduced performance compared to CoMFA with r2 = 0.76 and q2LOO = 0.59. A number of parameters were examined to improve the CoMFA model. Of these, a phenol indicator increased the q2LOO to 0.71. When up to 50% of the chemicals were left out in the leave-N-out cross-validation, the q2 remained significant. Finally, the models were tested by using two test sets; the q2pred for these were 0.71 and 0.62, a significant result which demonstrates the utility of the CoMFA model for predicting the RBAs of chemicals not included in the training set. If used in conjunction with phases I and II, which reduced the size of the data set dramatically by eliminating most inactive chemicals, the current CoMFA model (phase III) can be used to predict the RBA of chemicals with sufficient accuracy and to provide quantitative information for priority setting.

High-throughput and virtual screening: core lead discovery technologies move towards integration

In addition to high-throughput screening (HTS), the main lead discovery technology employed by most pharmaceutical companies today is virtual screening (VS). Although the two techniques have somewhat different philosophical origins, they contain many synergies that can potentially enhance the lead discovery process. Here, we describe many of the latest developments in VS technology with particular emphasis on their potential impact on HTS in, for example, focussed screening and data mining. In addition, we highlight key issues that need to be addressed before the potential of such efforts can be fully realized.

Classical QSAR and comparative molecular field analyses of the host-guest interaction of organic molecules with cyclodextrins

The application of classical QSAR and molecular modeling analysis using Comparative Molecular Field Analysis (CoMFA) to the complexation of some natural and modified cyclodextrins (CDs) with guest molecules was examined. For 1:1 complexation systems between natural beta-CD, modified alpha-, beta-, and gamma-CD that bear one p-(dimethylamino)benzoyl (DMAB) moiety (DMAB-alpha-, beta-, and gamma-CDs) and guest molecules of widely varying chemical structures and properties, the binding constants of the complexes were successfully fitted using multiple linear regression (MLR) with hydrophobic descriptor log P (the partition coefficient between 1-octanol and water phases) and molecular connectivity indices. A non-linear dependency of binding constants on the zero-th and/or first order molecular connectivity index as a measure of size becomes apparent. The modeling performance of the CoMFA models with steric/electrostatic fields to DMAB-alpha- and beta-CD systems was comparable to those of MLR models. However, statistically significant CoMFA models for gamma-CD systems which have higher conformational flexibility of the ring could not be obtained. The CoMFA models obtained for DMAB-alpha- and beta-CD systems showed that the predominant effects were steric for the DMAB-alpha-CD system and electrostatic for the DMAB-beta-CD system, respectively.

Understanding the antifungal activity of terbinafine analogues using quantitative structure-activity relationship (QSAR) models

Terbinafine and its analogues, which are a major class of non-azole antifungal agents, are known to act by inhibition of squalene epoxidase enzyme in fungal cells. We have performed a quantitative structure-activity relationship (QSAR) study on a series of 92 molecules using different types of physicochemical descriptors. Inhibitors were divided into five classes depending upon chemical structure. QSAR models were generated for correlation between antifungal activity against Candida albicans using genetic function approximation (GFA) technique. Equations were evaluated using internal as well as external test set predictions. Models generated for all these classes show that steric properties and conformational rigidity of side chains play an important role for the activity. The present QSAR analysis agrees with the results of the previously reported CoMFA study.

Pharmacophore fingerprinting. 1. Application to QSAR and focused library design

A new method of rapid pharmacophore fingerprinting (PharmPrint method) has been developed. A basis set of 10,549 three-point pharmacophores has been constructed by enumerating several distance ranges and pharmacophoric features. Software has been developed to assign pharmacophoric types to atoms in chemical structures, generate multiple conformations, and construct the binary fingerprint according to the pharmacophores that result. The fingerprint is used as a descriptor for developing a quantitative structure-activity relationship (QSAR) model using partial least squares. An example is given using sets of ligands for the estrogen receptor (ER). The result is compared with previously published results on the same data to show the superiority of a full 3D, conformationally flexible approach. The QSAR model can be readily interpreted in structural/chemical terms. Further examples are given using binary activity data and some of our novel in-house compounds, which show the value of the model when crossing compound classes.

Comparison of estrogen receptor alpha and beta subtypes based on comparative molecular field analysis (CoMFA)

A substantial body of evidence indicates that both humans and wildlife suffer adverse health effects from exposure to environmental chemicals that are capable of interacting with the endocrine system. The recent cloning of the estrogen receptor beta subtype (ER-beta) suggests that the selective effects of estrogenic compounds may arise in part by the control of different subsets of estrogen-responsive promoters by the two ER subtypes, ER-alpha and ER-beta. In order to identify the structural prerequisites for ligand-ER binding and to discriminate ER-alpha and ER-beta in terms of their ligand-binding specificities, Comparative Molecular Field Analysis (CoMFA) was employed to construct a three-dimensional Quantitative Structure-Activity Relationship (3D-QSAR) model on a data set of 31 structurally-diverse compounds for which competitive binding affinities have been measured against both ER-alpha and ER-beta. Structural alignment of the molecules in CoMFA was achieved by maximizing overlap of their steric and electrostatic fields using the Steric and Electrostatic ALignment (SEAL) algorithm. The final CoMFA models, generated by correlating the calculated 3D steric and electrostatic fields with the experimentally observed binding affinities using partial least-squares (PLS) regression, exhibited excellent self-consistency (r2 > 0.99) as well as high internal predictive ability (q2 > 0.65) based on cross-validation. CoMFA-predicted values of RBA for a test set of compounds outside of the training set were consistent with experimental observations. These CoMFA models can serve as guides for the rational design of ER ligands that possess preferential binding affinities for either ER-alpha or ER-beta. These models can also prove useful in risk assessment programs to identify real or suspected EDCs.