Molecular modeling approaches for the prediction of the nonspecific binding of drugs to hepatic microsomes

Evidence-based strategies for the characterisation of human drug and chemical glucuronidation in vitro and UDP-glucuronosyltransferase reaction phenotyping

Enzymes of the UDP-glucuronosyltransferase (UGT) superfamily contribute to the elimination of drugs from almost all therapeutic classes. Awareness of the importance of glucuronidation as a drug clearance mechanism along with increased knowledge of the enzymology of drug and chemical metabolism has stimulated interest in the development and application of approaches for the characterisation of human drug glucuronidation in vitro, in particular reaction phenotyping (the fractional contribution of the individual UGT enzymes responsible for the glucuronidation of a given drug), assessment of metabolic stability, and UGT enzyme inhibition by drugs and other xenobiotics. In turn, this has permitted the implementation of in vitro - in vivo extrapolation approaches for the prediction of drug metabolic clearance, intestinal availability, and drug-drug interaction liability, all of which are of considerable importance in pre-clinical drug development. Indeed, regulatory agencies (FDA and EMA) require UGT reaction phenotyping for new chemical entities if glucuronidation accounts for ≥25% of total metabolism. In vitro studies are most commonly performed with recombinant UGT enzymes and human liver microsomes (HLM) as the enzyme sources. Despite the widespread use of in vitro approaches for the characterisation of drug and chemical glucuronidation by HLM and recombinant enzymes, evidence-based guidelines relating to experimental approaches are lacking. Here we present evidence-based strategies for the characterisation of drug and chemical glucuronidation in vitro, and for UGT reaction phenotyping. We anticipate that the strategies will inform practice, encourage development of standardised experimental procedures where feasible, and guide ongoing research in the field.

Prediction of pharmacokinetic clearance and potential Drug-Drug interactions for omeprazole in the horse using in vitro systems

Horses are exposed to various kinds of medication, however, there are limited determinations of plasma clearance (CL_p) for the drugs used due to the high cost of equine in vivo studies.Many of the CL_p values generated come from the equine sports industry for determining drug plasma screening limits in the control of medications at the time of competition.The kinetics of omeprazole metabolism were investigated in freshly isolated and cryopreserved equine hepatocytes and hepatic microsomes (n = 3 horses).The V_max, K_m and intrinsic clearance (CL_int) of omeprazole were determined via the substrate depletion method as well as K_m values for the formation of three metabolites.The CL_int values were extrapolated to in vivo hepatic plasma clearance (CL_H) using the well stirred and parallel tube models.Cl_p for omeprazole was successfully predicted using freshly isolated or cryopreserved equine hepatocytes, while microsomes under-predicted.Equine microsomes were used to perform a drug-drug interaction (DDI) study between omeprazole and chloramphenicol. The average inhibitor constant K_i, assuming competitive inhibition, was 15.4 ± 5 µM.To the authors' knowledge, this is the first report showing the successful extrapolation of drug CL_p in the horse using equine hepatocytes and the prediction of a DDI using microsomes.

Drug Distribution. Part 1. Models to Predict Membrane Partitioning

PURPOSE

Tissue partitioning is an important component of drug distribution and half-life. Protein binding and lipid partitioning together determine drug distribution.

METHODS

Two structure-based models to predict partitioning into microsomal membranes are presented. An orientation-based model was developed using a membrane template and atom-based relative free energy functions to select drug conformations and orientations for neutral and basic drugs.

RESULTS

The resulting model predicts the correct membrane positions for nine compounds tested, and predicts the membrane partitioning for n = 67 drugs with an average fold-error of 2.4. Next, a more facile descriptor-based model was developed for acids, neutrals and bases. This model considers the partitioning of neutral and ionized species at equilibrium, and can predict membrane partitioning with an average fold-error of 2.0 (n = 92 drugs).

CONCLUSIONS

Together these models suggest that drug orientation is important for membrane partitioning and that membrane partitioning can be well predicted from physicochemical properties.

A Fragment-Based Approach for the Computational Prediction of the Nonspecific Binding of Drugs to Hepatic Microsomes

Correction for the nonspecific binding (NSB) of drugs to liver microsomes is essential for the accurate measurement of the kinetic parameters Km and Ki, and hence in vitro-in vivo extrapolation to predict hepatic clearance and drug-drug interaction potential. Although a number of computational approaches for the estimation of drug microsomal NSB have been published, they generally rely on compound lipophilicity and charge state at the expense of other physicochemical and chemical properties. In this work, we report the development of a fragment-based hologram quantitative structure activity relationship (HQSAR) approach for the prediction of NSB using a database of 132 compounds. The model has excellent predictivity, with a noncross-validated r2 of 0.966 and cross-validated r2 of 0.680, with a predictive r2 of 0.748 for an external test set comprising 34 drugs. The HQSAR method reliably predicted the fraction unbound in incubations of 95% of the training and test set drugs, excluding compounds with a steroid or morphinan 4,5-epoxide nucleus. Using the same data set of compounds, performance of the HQSAR method was superior to a model based on logP/D as the sole descriptor (predictive r2 for the test set compounds, 0.534). Thus, the HQSAR method provides an alternative approach to laboratory-based procedures for the prediction of the NSB of drugs to liver microsomes, irrespective of the drug charge state (acid, base, or neutral).

The Nonspecific Binding of Tyrosine Kinase Inhibitors to Human Liver Microsomes

Drugs and other chemicals frequently bind nonspecifically to the constituents of an in vitro incubation mixture, particularly the enzyme source [e.g., human liver microsomes (HLM)]. Correction for nonspecific binding (NSB) is essential for the accurate calculation of the kinetic parameters Km, Clint, and Ki. Many tyrosine kinase inhibitors (TKIs) are lipophilic organic bases that are nonionized at physiologic pH. Attempts to measure the NSB of several TKIs to HLM by equilibrium dialysis proved unsuccessful, presumably due to the limited aqueous solubility of these compounds. Thus, the addition of detergents to equilibrium dialysis samples was investigated as an approach to measure the NSB of TKIs. The binding of six validation set nonionized lipophilic bases (felodipine, isradipine, loratidine, midazolam, nifedipine, and pazopanib) to HLM (0.25 mg/ml) was shown to be unaffected by the addition of CHAPS (6 mM) to the dialysis medium. This approach was subsequently applied to measurement of the binding of axitinib, dabrafenib, erlotinib, gefitinib, ibrutinib, lapatinib, nilotinib, nintedanib, regorafenib, sorafenib, and trametinib to HLM (0.25 mg/ml). As with the validation set drugs, attainment of equilibrium was demonstrated in HLM-HLM and buffer-buffer control dialysis experiments. Values of the fraction unbound to HLM ranged from 0.14 (regorafenib and sorafenib) to 0.93 (nintedanib), and were generally consistent with the known physicochemical determinants of drug NSB. The extensive NSB of many TKIs to HLM underscores the importance of correction for TKI binding to HLM and, presumably, other enzyme sources present in in vitro incubation mixtures.

In Silico Discovery of Novel Potent Antioxidants on the Basis of Pulvinic Acid and Coumarine Derivatives and Their Experimental Evaluation

A pigment from the edible mushroom Xerocomus badius norbadione A, which is a natural derivative of pulvinic acid, was found to possess antioxidant properties. Since the pulvinic acid represents a novel antioxidant scaffold, several other derivatives were recently synthetized and evaluated experimentally, along with some structurally related coumarine derivatives. The obtained data formed the basis for the construction of several quantitative structure-activity and pharmacophore models, which were employed in the virtual screening experiments of compound libraries and for the prediction of their antioxidant activity, with the goal of discovering novel compounds possessing antioxidant properties. A final prioritization list of 21 novel compounds alongside 8 established antioxidant compounds was created for their experimental evaluation, consisting of the DPPH assay, 2-deoxyribose assay, β-carotene bleaching assay and the cellular antioxidant activity assay. Ten novel compounds from the tetronic acid and barbituric acid chemical classes displayed promising antioxidant activity in at least one of the used assays, that is comparable to or even better than some standard antioxidants. Compounds 5, 7 and 9 displayed good activity in all the assays, and were furthermore effective preventers of oxidative stress in human peripheral blood mononuclear cells, which are promising features for the potential therapeutic use of such compounds.

Activity cliffs and activity cliff generators based on chemotype-related activity landscapes

Activity cliffs have large impact in drug discovery; therefore, their detection and quantification are of major importance. This work introduces the metric activity cliff enrichment factor and expands the previously reported activity cliff generator concept by adding chemotype information to representations of the activity landscape. To exemplify these concepts, three molecular databases with multiple biological activities were characterized. Compounds in each database were grouped into chemotype classes. Then, pairwise comparisons of structure similarities and activity differences were calculated for each compound and used to construct chemotype-based structure-activity similarity (SAS) maps. Different landscape distributions among four major regions of the SAS maps were observed for different subsets of molecules grouped in chemotypes. Based on this observation, the activity cliff enrichment factor was calculated to numerically detect chemotypes enriched in activity cliffs. Several chemotype classes were detected having major proportion of activity cliffs than the entire database. In addition, some chemotype classes comprising compounds with smooth structure activity relationships (SAR) were detected. Finally, the activity cliff generator concept was applied to compounds grouped in chemotypes to extract valuable SAR information.

Design, synthesis, and mechanistic investigations of bile acid-tamoxifen conjugates for breast cancer therapy

We have synthesized two series of bile acid tamoxifen conjugates using three bile acids lithocholic acid (LCA), deoxycholic acid (DCA), and cholic acid (CA). These bile acid-tamoxifen conjugates possess 1, 2, and 3 tamoxifen molecules attached to hydroxyl groups of bile acids having free acid and amine functionalities at the tail region of bile acids. The in vitro anticancer activities of these bile acid-tamoxifen conjugates show that the free amine headgroup based cholic acid-tamoxifen conjugate (CA-Tam3-Am) is the most potent anticancer conjugate as compared to the parent drug tamoxifen and other acid and amine headgroup based bile acid-tamoxifen conjugates. The cholic acid-tamoxifen conjugate (CA-Tam3-Am) bearing three tamoxifen molecules shows enhanced anticancer activities in both estrogen receptor +ve and estrogen receptor -ve breast cancer cell lines. The enhanced anticancer activity of CA-Tam3-Am is due to more favorable irreversible electrostatic interactions followed by intercalation of these conjugates in hydrophobic core of membrane lipids causing increase in membrane fluidity. Annexin-FITC based FACS analysis showed that cells undergo apoptosis, and cell cycle analysis showed the arrest of cells in sub G0 phase. ROS assays showed a high amount of generation of ROS independent of ER status of the cell line indicating changes in mitochondrial membrane fluidity upon the uptake of the conjugate that further leads to the release of cytochrome c, a direct and indirect regulator of ROS. The mechanistic studies for apoptosis using PCR and western analysis showed apoptotsis by intrinsic and extrinsic pathways in ER +ve MCF-7 cells and by only an intrinsic pathway in ER -ve cells. In vivo studies in the 4T1 tumor model showed that CA-Tam3-Am is more potent than tamoxifen. These studies showed that bile acids provide a new scaffold for high drug loading and that their anticancer activities strongly depend on charge and hydrophobicity of lipid-drug conjugates.

Application of the fluorescent probe 1-anilinonaphthalene-8-sulfonate to the measurement of the nonspecific binding of drugs to human liver microsomes

The fluorescence of 1-anilinonaphthalene-8-sulfonate (ANS) in the presence of human liver microsomes (HLMs) is altered by drugs that bind nonspecifically to the lipid bilayer. The present study characterized the relationship between the nonspecific binding (NSB) of drugs to HLMs as measured by equilibrium dialysis and the magnitude of the change in baseline ANS fluorescence. Fraction unbound in incubations of HLMs (f(u(mic))) was determined for 16 drugs (12 bases, 3 acids, and 1 neutral) with log P values in the range 0.1 to 6.7 at three concentrations (100, 200, and 500 μM). Changes in ANS fluorescence induced by each of the drugs in the presence of HLMs were measured by spectrofluorometry. Values of f(u(mic)) determined by equilibrium dialysis ranged from 0.08 to 1.0. Although NSB of the basic drugs tended to increase with increasing log P, exceptions occurred. Basic drugs generally caused an increase in ANS fluorescence, whereas the acidic and neutral drugs resulted in a decrease in ANS fluorescence. There were highly significant (p < 0.001) linear relationships between the modulus (absolute value) of the increment/decrement in ANS fluorescence and both f(u(mic)) (r = 0.90 to 0.96) and log(1 - f(u(mic))/f(u(mic))) (r = 0.85 to 0.92) at the three drug concentrations. Agreement between measured f(u(mic)) and that predicted by ANS fluorescence was very good (<10% variance) for a validation set of six compounds. The ANS fluorescence method provides an accurate measure of the NSB of drugs to HLMs. Physicochemical determinants other than log P and charge type influence the NSB of drugs to HLMs.

Comparative molecular field analysis (CoMFA) and comparative molecular similarity indices analysis (CoMSIA) of some benzimidazole derivatives with trichomonicidal activity

Trichomonosis is a common sexually transmitted infectious disease linked to reproductive health complications. Recently, the benzimidazole nucleus has emerged as a promising scaffold to develop new trichomonicidal agents. Despite the fact that large amounts of experimental data have been accumulated over the past eight years, no quantitative studies have yet been reported on this class of compounds. In our effort to develop new antiparasitic benzimidazole derivatives, we report in this paper CoMFA and CoMSIA studies with an initial set of 70 benzimidazole derivatives with trichomonicidal activity. Four CoMFA models and eight CoMSIA models were generated; ten of these models had values of r(2) > 0.6 and q(2) > 0.5. The best CoMFA model had r(2) = 0.936 and q(2) = 0.634, and the best CoMSIA model had r(2) = 0.858 and q(2) = 0.642. These models were generated by using two conformer selection methodologies (minimum energy conformations and 3D similarity), and three charge types (Mulliken, Gasteiger-Hükel and electrostatic potential atomic charges). The putative active tautomers of 1H-benzimidazole derivatives were selected using 3D-QSAR calculations. All models were validated via an external test set with 13 molecules. The best models satisfied additional validation criteria. The contour maps generated show the most important features that a benzimidazole derivative should have for trichomonicidal activity; they also, suggest that substituents at the 2- and 6-positions are important in the generation of derivatives with strong activity.

Lack of appreciable species differences in nonspecific microsomal binding

Species differences in microsomal binding were evaluated for 43 drug molecules in human, monkey, dog and rat liver microsomes, using a fixed concentration of microsomal protein. The dataset included 32 named drugs and 11 proprietary compounds encompassing a broad spectrum of physicochemical properties (11 acids, 24 bases, 8 neutral, c log D -1 to 7, MW 200 to 700 and free fraction <0.001 to 1). Free fractions (f(u,mic)) in monkey, dog, rat and human microsomes were highly correlated, with linear regression correlation coefficients greater than 0.97. The average fold-difference in f(u,mic) between monkey, dog, or rat, and human was 1.6-, 1.3-, and 1.5-fold, respectively. Species differences in f(u,mic) were also assessed for a range of microsomal protein concentrations (0.2-2 mg/mL) for midazolam, clomipramine, astemizole, and tamoxifen, drugs with low to high microsomal binding. The mean fold species-difference in f(u,mic) for midazolam, clomipramine, astemizole, and tamoxifen was 1.1-, 1.2-, 1.3-, and 2.0-fold, respectively, and was independent of normalized microsomal protein concentration. For a fixed concentration of microsomal protein, greater than 76% and 90% of drugs examined in this study had preclinical species f(u,mic) within 1.5- and 2-fold, respectively, of experimentally measured human values.

Aldosterone glucuronidation by human liver and kidney microsomes and recombinant UDP-glucuronosyltransferases: inhibition by NSAIDs

AIMS

To characterize: i) the kinetics of aldosterone (ALDO) 18beta-glucuronidation using human liver and human kidney microsomes and identify the human UGT enzyme(s) responsible for ALDO 18beta-glucuronidation and ii) the inhibition of ALDO 18beta-glucuronidation by non-selective NSAIDs.

METHODS

Using HPLC and LC-MS methods, ALDO 18beta-glucuronidation was characterized using human liver (n= 6), human kidney microsomes (n= 5) and recombinant human UGT 1A1, 1A3, 1A4, 1A5, 1A6, 1A7, 1A8, 1A9, 1A10, 2B4, 2B7, 2B10, 2B15, 2B17 and 2B28 as the enzyme sources. Inhibition of ALDO 18beta-glucuronidation was investigated using alclofenac, cicloprofen, diclofenac, diflunisal, fenoprofen, R- and S-ibuprofen, indomethacin, ketoprofen, ketorolac, meclofenamic acid, mefenamic acid, S-naproxen, pirprofen and tiaprofenic acid. A rank order of inhibition (IC(50)) was established and the mechanism of inhibition investigated using diclofenac, S-ibuprofen, indomethacin, mefenamic acid and S-naproxen.

RESULTS

ALDO 18beta-glucuronidation by hepatic and renal microsomes exhibited Michaelis-Menten kinetics. Mean (+/-SD) K(m), V(max) and CL(int) values for HLM and HKCM were 509 +/- 137 and 367 +/- 170 microm, 1075 +/- 429 and 1110 +/- 522 pmol min(-1) mg(-1), and 2.36 +/- 1.12 and 3.91 +/- 2.35 microl min(-1) mg(-1), respectively. Of the UGT proteins, only UGT1A10 and UGT2B7 converted ALDO to its 18beta-glucuronide. All NSAIDs investigated inhibited ALDO 18beta-G formation by HLM, HKCM and UGT2B7. The rank order of inhibition (IC(50)) of renal and hepatic ALDO 18beta-glucuronidation followed the general trend: fenamates > diclofenac > arylpropionates.

CONCLUSION

A NSAID-ALDO interaction in vivo may result in elevated intra-renal concentrations of ALDO that may contribute to the adverse renal effects of NSAIDs and their effects on antihypertensive drug response.

Structure-based prediction of the nonspecific binding of drugs to hepatic microsomes

For the accurate prediction of in vivo hepatic clearance or drug-drug interaction potential through in vitro microsomal metabolic data, it is essential to evaluate the fraction unbound in hepatic microsomal incubation media. Here, a structure-based in silico predictive model of the nonspecific binding (fu(mic), fraction unbound in hepatic microsomes) for 86 drugs was successfully developed based on seven selected molecular descriptors. The R(2) of the predicted and observed log((1 - fu(mic))/fu(mic)) for the training set (n = 64) and test set (n = 22) were 0.82 and 0.85, respectively. The average fold error (AFE, calculated by fu(mic) rather than log((1 - fu(mic))/fu(mic))) of the in silico model was 1.33 (n = 86). The predictive capability of fu(mic) for neutral drugs compared well to that for basic compounds (R(2) = 0.82, AFE = 1.18 and fold error values were all below 2, except for felodipine and progesterone) in our model. This model appears to perform better for neutral compounds when compared to models previously published in the literature. Therefore, this in silico model may be used as an additional tool to estimate fu(mic) and for predicting in vivo hepatic clearance and inhibition potential from in vitro hepatic microsomal studies.

In silico modeling of nonspecific binding to human liver microsomes

Estimation of unbound fraction of substrate in microsomal incubation media is important in accurately predicting hepatic intrinsic clearance and drug-drug interactions. In this study, the unbound fraction of 1223 drug-like molecules in human liver microsomal incubation media has been determined using equilibrium dialysis. These compounds, which include 27 marketed drug molecules, cover a much broader range of physiochemical properties such as hydrophobicity, molecular weight, ionization state, and degree of binding than those examined in previous work. In developing the in silico model, we have used two-dimensional molecular descriptors including cLogP, Kier connectivity, shape, and E-state indices, a subset of MOE descriptors, and a set of absorption, disposition, metabolism, and excretion structural keys used for our in-house absorption, disposition, metabolism, excretion, and toxicity modeling. Hydrophobicity is the most important molecular property contributing to the nonspecific binding of substrate to microsomes. The prediction accuracy of the model is validated using a subset of 100 compounds, and 92% of the variance is accounted for by the model with a root mean square error (RMSE) of 0.10. For the training set of compounds, 99% of variance is accounted for by the model with a RMSE of 0.02. The performance of the developed model has been further tested using the 27 marketed drug molecules with a RMSE of 0.10 between the observed and the predicted unbound fraction values.

Characterisation of the binding of cationic amphiphilic drugs to phospholipid bilayers using surface plasmon resonance

The interactions of three cationic amphiphilic drugs (CPZ, AMI, PROP) with phospholipid vesicles comprising DOPC, DMPC, or DSPC were investigated using surface plasmon resonance (SPR). Responses for CAD concentrations in the range 15.625 to 1500 microM were measured. The greatest uptake by each phospholipid bilayer occurred with CPZ. Inclusion of CAD concentrations between 750 and 1500 microM provided evidence for a second nonsaturable binding process, which may arise from intercalation of the drugs within the lipid bilayer. CAD binding was additionally shown to be dependent on membrane fluidity. Responses were initially fitted over a concentration range of 15.625 to 500 microM using a model which incorporated terms for a saturable binding site. This yielded very poor values of K(D) and nonsensible values of saturation responses. Subsequently, responses were fit to the expression for a model which incorporated terms for both a saturable binding site and second nonsaturable site. Measurable binding affinities (K(D) values ranged from 170 to 814 microM) were obtained for DOPC and DMPC bilayers which are similar to values reported previously. This work demonstrates that SPR studies with synthetic phospholipid bilayers provide a potentially useful approach for characterising drug-membrane binding interactions and for providing insight into the processes that contribute to drug-membrane binding.

Glucuronidation of fenamates: Kinetic studies using human kidney cortical microsomes and recombinant UDP-glucuronosyltransferase (UGT) 1A9 and 2B7

Mefenamic acid, a non-steroidal anti-inflammatory drug (NSAID), is used commonly to treat menorrhagia. This study investigated the glucuronidation kinetics of flufenamic, mefenamic and niflumic acid using human kidney cortical microsomes (HKCM) and recombinant UGT1A9 and UGT2B7. Using HKCM Michaelis-Menten (MM) kinetics were observed for mefenamic (K(m)(app) 23 microM) and niflumic acid (K(m)(app) 123 microM) glucuronidation, while flufenamic acid exhibited non-hyperbolic (atypical) glucuronidation kinetics. Notably, the intrinsic renal clearance of mefenamic acid (CL(int) 17+/-5.5 microL/minmg protein) was fifteen fold higher than that of niflumic acid (CL(int) 1.1+/-0.8 microL/minmg protein). These data suggest that renal glucuronidation of mefenamic acid may result in high intrarenal exposure to mefenamic acyl-glucuronide and subsequent binding to renal proteins. Diverse kinetics were observed for fenamate glucuronidation by UGT2B7 and UGT1A9. Using UGT2B7 MM kinetics were observed for flufenamic (K(m)(app) 48 microM) and niflumic acid (K(m)(app) 135 microM) glucuronidation and atypical kinetics with mefenamic acid. Similarity in K(m)(app) between HKCM and UGT2B7 suggests that UGT2B7 may be the predominant renal UGT isoform catalysing niflumic acid glucuronidation. In contrast, UGT1A9 glucuronidation kinetics were characterised by negative cooperativity with mefenamic (S(50) 449 microM, h 0.4) and niflumic acid (S(50) 7344 microM, h 0.4) while atypical kinetics were observed with flufenamic acid. Additionally, potent inhibition of the renal glucuronidation of the UGT substrate 'probe' 4-methylumbelliferone by flufenamic, mefenamic and niflumic acid was observed. These data suggest that inhibitory metabolic interactions may occur between fenamates and other substrates metabolised by UGT2B7 and UGT1A9 in human kidney.