Challenges predicting ligand-receptor interactions of promiscuous proteins: the nuclear receptor PXR.

Transcriptional regulation of some genes involved in xenobiotic detoxification and apoptosis is performed via the human pregnane X receptor (PXR) which in turn is activated by structurally diverse agonists including steroid hormones. Activation of PXR has the potential to initiate adverse effects, altering drug pharmacokinetics or perturbing physiological processes. Reliable computational prediction of PXR agonists would be valuable for pharmaceutical and toxicological research. There has been limited success with structure-based modeling approaches to predict human PXR activators. Slightly better success has been achieved with ligand-based modeling methods including quantitative structure-activity relationship (QSAR) analysis, pharmacophore modeling and machine learning. In this study, we present a comprehensive analysis focused on prediction of 115 steroids for ligand binding activity towards human PXR. Six crystal structures were used as templates for docking and ligand-based modeling approaches (two-, three-, four- and five-dimensional analyses). The best success at external prediction was achieved with 5D-QSAR. Bayesian models with FCFP_6 descriptors were validated after leaving a large percentage of the dataset out and using an external test set. Docking of ligands to the PXR structure co-crystallized with hyperforin had the best statistics for this method. Sulfated steroids (which are activators) were consistently predicted as non-activators while, poorly predicted steroids were docked in a reverse mode compared to 5alpha-androstan-3beta-ol. Modeling of human PXR represents a complex challenge by virtue of the large, flexible ligand-binding cavity. This study emphasizes this aspect, illustrating modest success using the largest quantitative data set to date and multiple modeling approaches.

Chemoinformatic methods for predicting interference in drug of abuse/toxicology immunoassays.

BACKGROUND: Immunoassays used for routine drug of abuse (DOA) and toxicology screening may be limited by cross-reacting compounds able to bind to the antibodies in a manner similar to the target molecule(s). To date, there has been little systematic investigation using computational tools to predict cross-reactive compounds. METHODS: Commonly used molecular similarity methods enabled calculation of structural similarity for a wide range of compounds (prescription and over-the-counter medications, illicit drugs, and clinically significant metabolites) to the target molecules of DOA/toxicology screening assays. We used various molecular descriptors (MDL public keys, functional class fingerprints, and pharmacophore fingerprints) and the Tanimoto similarity coefficient. These data were then compared with cross-reactivity data in the package inserts of immunoassays marketed for in vitro diagnostic use. Previously untested compounds that were predicted to have a high probability of cross-reactivity were tested. RESULTS: Molecular similarity calculated using MDL public keys and the Tanimoto similarity coefficient showed a strong and statistically significant separation between cross-reactive and non-cross-reactive compounds. This result was validated experimentally by discovery of additional cross-reactive compounds based on computational predictions. CONCLUSIONS: The computational methods employed are amenable toward rapid screening of databases of drugs, metabolites, and endogenous molecules and may be useful for identifying cross-reactive molecules that would be otherwise unsuspected. These methods may also have value in focusing cross-reactivity testing on compounds with high similarity to the target molecule(s) and limiting testing of compounds with low similarity and very low probability of cross-reacting with the assay.

Molecular similarity methods for predicting cross-reactivity with therapeutic drug monitoring immunoassays.

Immunoassays are used for therapeutic drug monitoring (TDM), yet may suffer from cross-reacting compounds able to bind the assay antibodies in a manner similar to the target molecule. To our knowledge, there has been no investigation using computational tools to predict cross-reactivity with TDM immunoassays. The authors used molecular similarity methods to enable calculation of structural similarity for a wide range of compounds (prescription and over-the-counter medications, illicit drugs, and clinically significant metabolites) to the target molecules of TDM immunoassays. Utilizing different molecular descriptors (MDL public keys, functional class fingerprints, and pharmacophore fingerprints) and the Tanimoto similarity coefficient, the authors compared cross-reactivity data in the package inserts of immunoassays marketed for in vitro diagnostic use. Using MDL public keys and the Tanimoto similarity coefficient showed a strong and statistically significant separation between cross-reactive and non-cross-reactive compounds. Thus, 2-dimensional shape similarity of cross-reacting molecules and the target molecules of TDM immunoassays provides a fast chemoinformatics methods for a priori prediction of potential of cross-reactivity that might be otherwise undetected. These methods could be used to reliably focus cross-reactivity testing on compounds with high similarity to the target molecule and limit testing of compounds with low similarity and ultimately with a very low probability of cross-reacting with the assay in vitro.

Using molecular similarity to highlight the challenges of routine immunoassay-based drug of abuse/toxicology screening in emergency medicine.

BACKGROUND: Laboratory tests for routine drug of abuse and toxicology (DOA/Tox) screening, often used in emergency medicine, generally utilize antibody-based tests (immunoassays) to detect classes of drugs such as amphetamines, barbiturates, benzodiazepines, opiates, and tricyclic antidepressants, or individual drugs such as cocaine, methadone, and phencyclidine. A key factor in assay sensitivity and specificity is the drugs or drug metabolites that were used as antigenic targets to generate the assay antibodies. All DOA/Tox screening immunoassays can be limited by false positives caused by cross-reactivity from structurally related compounds. For immunoassays targeted at a particular class of drugs, there can also be false negatives if there is failure to detect some drugs or their metabolites within that class. METHODS: Molecular similarity analysis, a computational method commonly used in drug discovery, was used to calculate structural similarity of a wide range of clinically relevant compounds (prescription and over-the-counter medications, illicit drugs, and clinically significant metabolites) to the target ('antigenic') molecules of DOA/Tox screening tests. These results were compared with cross-reactivity data in the package inserts of immunoassays marketed for clinical testing. The causes for false positives for phencyclidine and tricyclic antidepressant screening immunoassays were investigated at the authors' medical center using gas chromatography/mass spectrometry as a confirmatory method. RESULTS: The results illustrate three major challenges for routine DOA/Tox screening immunoassays used in emergency medicine. First, for some classes of drugs, the structural diversity of common drugs within each class has been increasing, thereby making it difficult for a single assay to detect all compounds without compromising specificity. Second, for some screening assays, common 'out-of-class' drugs may be structurally similar to the target compound so that they account for a high frequency of false positives. Illustrating this point, at the authors' medical center, the majority of positive screening results for phencyclidine and tricyclic antidepressants assays were explained by out-of-class drugs. Third, different manufacturers have adopted varying approaches to marketed immunoassays, leading to substantial inter-assay variability. CONCLUSION: The expanding structural diversity of drugs presents a difficult challenge for routine DOA/Tox screening that limit the clinical utility of these tests in the emergency medicine setting.

Molecular characterization of CYP2B6 substrates.

CYP2B6 has not been as fully characterized at the molecular level as other members of the human cytochrome P450 family. As more widely used in vitro probes for characterizing the involvement of this enzyme in the metabolism of xenobiotics have become available, the number of molecules identified as CYP2B6 substrates has increased. In this study we have analyzed the available kinetic data generated by multiple laboratories with human recombinant expressed CYP2B6 and along with calculated molecular properties derived from the ChemSpider database, we have determined the molecular features that appear to be important for CYP2B6 substrates. In addition we have applied 2D and 3D QSAR methods to generate predictive pharmacophore and 2D models. For 28 molecules with K(m) data, the molecular weight (mean +/- SD) is 253.78+/-74.03, ACD/logP is 2.68+/-1.51, LogD(pH 5.5) is 1.51+/-1.43, LogD(pH 7.4) is 2.02+/-1.25, hydrogen bond donor (HBD) count is 0.57 +/-0.57, hydrogen bond acceptor (HBA) count is 2.57+/-1.37, rotatable bonds is 3.50+/-2.71 and total polar surface area (TPSA) is 27.63+/-19.42. A second set of 15 molecules without K(m) data possessed similar mean molecular property values. These properties are comparable to those of a set of 21 molecules used in a previous pharmacophore modeling study (Ekins et al., J Pharmacol Exp Ther 288 (1), 21-29, 1999). Only the LogD and HBD values were statistically significantly different between these different datasets. We have shown that CYP2B6 substrates are generally small hydrophobic molecules that are frequently central nervous system active, which may be important for drug discovery research.

Treating chemical diversity in QSAR analysis: modeling diverse HIV-1 integrase inhibitors using 4D fingerprints.

A set of 213 compounds across 12 structurally diverse classes of HIV-1 integrase inhibitors was used to develop and evaluate a combined clustering and QSAR modeling methodology to construct significant, reliable, and robust models for structurally diverse data sets. The trial-descriptor pool for both clustering- and QSAR-model building consisted of 4D fingerprints and classic QSAR descriptors. Clustering was carried out using a combination of the partitioning around medoids method and divisive hierarchical clustering. QSAR models were constructed for members of each cluster by linear-regression fitting and model optimization using the genetic function approximation. The 12 structurally diverse classes of integrase inhbitors were partitioned into five clusters from which corresponding QSAR models, overwhelmingly composed of 4D fingerprint descriptors, were constructed. Analysis of the five QSAR models suggests that three models correspond to structurally diverse inhibitors that likely bind at a common site on integrase characterized by a common inhibitor hydrogen-bond donor, but involving somewhat different alignments and/or poses for the inhibitors of each of the three clusters. The particular alignments for the inhibitors of each of the three QSAR models involve specific distributions of nonpolar groups over the inhibitors. The two other clusters, one for coumarins and the other for depsides and depsidones, lead to QSAR models with less-defined pharmacophores, likely representing an inhibitor binding to a site(s) different from that of the other nine classes of inhibitors. Overall, the clustering and QSAR methodology employed in this study suggests that it can meaningfully partition structurally diverse compounds expressing a common endpoint in such a manner that leads to statistically significant and pharmacologically insightful composite QSAR models.

Human pregnane X receptor antagonists and agonists define molecular requirements for different binding sites.

The pregnane X receptor (PXR) is an important transcriptional regulator of the expression of xenobiotic metabolism and transporter genes. The receptor is promiscuous, binding many structural classes of molecules that act as agonists at the ligand-binding domain, triggering up-regulation of genes, increasing the metabolism and excretion of therapeutic agents, and causing drug-drug interactions. It has been suggested that human PXR antagonists represent a means to counteract such interactions. Several azoles have been hypothesized to bind the activation function-2 (AF-2) surface on the exterior of PXR when agonists are concurrently bound in the ligand-binding domain. In the present study, we have derived novel computational models for PXR agonists using different series of imidazoles, steroids, and a set of diverse molecules with experimental PXR agonist binding data. We have additionally defined a novel pharmacophore for the steroidal agonist site. All agonist pharmacophores showed that hydrophobic features are predominant. In contrast, a qualitative comparison with the corresponding PXR antagonist pharmacophore models using azoles and biphenyls showed that they are smaller and hydrophobic with increased emphasis on hydrogen bonding features. Azole antagonists were docked into a proposed hydrophobic binding pocket on the outer surface at the AF-2 site and fitted comfortably, making interactions with key amino acids involved in charge clamping. Combining computational and experimental data for different classes of molecules provided strong evidence for agonists and antagonists binding distinct regions on PXR. These observations bear significant implications for future discovery of molecules that are more selective and potent antagonists.

QSAR analyses of skin penetration enhancers.

QSAR models for four skin penetration enhancer data sets of 61, 44, 42, and 17 compounds were constructed using classic QSAR descriptors and 4D-fingerprints. Three data sets involved skin penetration enhancement of hydrocortisone and hydrocortisone acetate. The other data set involved skin penetration enhancement of fluorouracil. The measure of penetration enhancement is the ratio of the net permeation of the penetrant with and without a common fixed concentration of enhancer. Significant QSAR models could be built using multidimensional linear regression fitting and genetic function model optimization for all four data sets when both classic and 4D-fingerprint descriptors were used in the trial descriptor pool. Reasonable QSAR models could be built when only 4D-fingerprint descriptors were employed, and no significant QSAR models could be built using only classic descriptors for two of the four data sets. Comparison analyses of the descriptor terms, and their respective regression coefficients, across the pairs of the best QSAR models of the four skin penetration enhancer data sets did not reveal any significant extent of similar terms. Overall, the QSAR models for the penetration-enhancer systems appear meaningfully different from one another, suggesting that there are distinct mechanisms of skin penetration enhancement that depend on the chemistry of both the enhancer and the penetrant.

Prediction and mechanistic interpretation of human oral drug absorption using MI-QSAR analysis.

Membrane-interaction [MI]-QSAR analysis, which includes descriptors explicitly derived from simulations of solutes [drugs] interacting with phospholipid membrane models, was used to construct QSAR models for human oral intestinal drug absorption. A data set of 188 compounds, which are mainly drugs, was divided into a parent training set of 164 compounds and a test set of 24 compounds. Stable, but not highly fit [R2 = 0.68] MI-QSAR models could be built for all 188 compounds. However, the relatively large number [47] of drugs having 100% absorption, as well as all zwitterionic compounds [11], had to be eliminated from the training set in order to construct a linear five-term oral absorption diffusion model for 106 compounds which was both stable [R2 = 0.82, Q2 = 0.79] and predictive given the test set compounds were predicted with nearly the same average accuracy as the compounds of the training set. Intermolecular membrane-solute descriptors are essential to building good oral absorption models, and these intermolecular descriptors are displaced in model optimizations and intramolecular solute descriptors found in published oral absorption QSAR models. A general form for all of the oral intestinal absorption MI-QSAR models has three classes of descriptors indicative of three thermodynamic processes: (1) solubility and partitioning, (2) membrane-solute interactions, and (3) flexibility of the solute and/or membrane. The intestinal oral absorption MI-QSAR models were compared to MI-QSAR models previously developed for Caco-2 cell permeation and for blood-brain barrier penetration. The MI-QSAR models for all three of these ADME endpoints share several common descriptors, and suggest a common mechanism of transport across all three barriers. A further analysis of these three types of MI-QSAR models has been done to identify descriptor-term differences across these three models, and the corresponding differences in thermodynamic transport behavior of the three barriers.

Functional evolution of the pregnane X receptor.

The pregnane X receptor (PXR; NR1I2) is a nuclear hormone receptor (NR) that transcriptionally regulates genes encoding transporters and drug-metabolising enzymes in the liver and intestine. PXR activation leads to enhanced metabolism and elimination of xenobiotics and endogenous compounds such as hormones and bile salts. Relative to other vertebrate NRs, PXR has the broadest specificity for ligand activators by virtue of a large, flexible ligand-binding cavity. In addition, PXR has the most extensive sequence diversity across vertebrate species in the ligand-binding domain of any NR, with significant pharmacological differences between human and rodent PXRs, and especially marked divergence between mammalian and nonmammalian PXRs. The unusual properties of PXR complicate the use of in silico and animal models to predict in vivo human PXR pharmacology. Research into the evolutionary history of the PXR gene has also provided insight into the function of PXR in humans and other animals.

Constructing optimum blood brain barrier QSAR models using a combination of 4D-molecular similarity measures and cluster analysis.

A new method, using a combination of 4D-molecular similarity measures and cluster analysis to construct optimum QSAR models, is applied to a data set of 150 chemically diverse compounds to build optimum blood-brain barrier (BBB) penetration models. The complete data set is divided into subsets based on 4D-molecular similarity measures using cluster analysis. The compounds in each cluster subset are further divided into a training set and a test set. Predictive QASAR models are constructed for each cluster subset using the corresponding training sets. These QSAR models best predict test set compounds which are assigned to the same cluster subset, based on the 4D-molecular similarity measures, from which the models are derived. The results suggest that the specific properties governing blood-brain barrier permeability may vary across chemically diverse compounds. Partitioning compounds into chemically similar classes is essential to constructing predictive blood-brain barrier penetration models embedding the corresponding key physiochemical properties of a given chemical class.

Predicting blood-brain barrier partitioning of organic molecules using membrane-interaction QSAR analysis.

PURPOSE: Membrane-interaction quantitative structure-activity relationship (OSAR) analysis (MI-QSAR) has been used to develop predictive models of blood-brain barrier partitioning of organic compounds by, in part, simulating the interaction of an organic compound with the phospholipid-rich regions of cellular membranes. METHOD: A training set of 56 structurally diverse compounds whose blood-brain barrier partition coefficients were measured was used to construct MI-QSAR models. Molecular dynamics simulations were used to determine the explicit interaction of each test compound (solute) with a model DMPC monolayer membrane model. An additional set of intramolecular solute descriptors were computed and considered in the trial pool of descriptors for building MI-QSAR models. The QSAR models were optimized using multidimensional linear regression fitting and a genetic algorithm. A test set of seven compounds was evaluated using the MI-QSAR models as part of a validation process. RESULTS: Significant MI-QSAR models (R2 = 0.845, Q2 = 0.795) of the blood-brain partitioning process were constructed. Blood-brain barrier partitioning is found to depend upon the polar surface area. the octanol/water partition coefficient, and the conformational flexibility of the compounds as well as the strength of their binding" to the model biologic membrane. The blood-brain barrier partitioning measures of the test set compounds were predicted with the same accuracy as the compounds of the training set. CONCLUSION: The MI-QSAR models indicate that the blood-brain barrier partitioning process can be reliably described for structurally diverse molecules provided interactions of the molecule with the phospholipids-rich regions of cellular membranes are explicitly considered.