Development and application of a PBPK modeling strategy to support antimalarial drug development

As part of a collaboration between Medicines for Malaria Venture (MMV), Certara UK and Monash University, physiologically-based pharmacokinetic (PBPK) models were developed for 20 antimalarials, using data obtained from standardized in vitro assays and clinical studies within the literature. The models have been applied within antimalarial drug development at MMV for more than 5 years. During this time, a strategy for their impactful use has evolved. All models are described in the supplementary material and are available to researchers. Case studies are also presented, demonstrating real-world development and clinical applications, including the assessment of the drug-drug interaction liability between combination partners or with co-administered drugs. This work emphasizes the benefit of PBPK modeling for antimalarial drug development and decision making, and presents a strategy to integrate it into the research and development process. It also provides a repository of shared information to benefit the global health research community.

An in vitro toolbox to accelerate anti-malarial drug discovery and development

Background

Modelling and simulation are being increasingly utilized to support the discovery and development of new anti-malarial drugs. These approaches require reliable in vitro data for physicochemical properties, permeability, binding, intrinsic clearance and cytochrome P450 inhibition. This work was conducted to generate an in vitro data toolbox using standardized methods for a set of 45 anti-malarial drugs and to assess changes in physicochemical properties in relation to changing target product and candidate profiles.

Methods

Ionization constants were determined by potentiometric titration and partition coefficients were measured using a shake-flask method. Solubility was assessed in biorelevant media and permeability coefficients and efflux ratios were determined using Caco-2 cell monolayers. Binding to plasma and media proteins was measured using either ultracentrifugation or rapid equilibrium dialysis. Metabolic stability and cytochrome P450 inhibition were assessed using human liver microsomes. Sample analysis was conducted by LC–MS/MS.

Results

Both solubility and fraction unbound decreased, and permeability and unbound intrinsic clearance increased, with increasing Log D_7.4. In general, development compounds were somewhat more lipophilic than legacy drugs. For many compounds, permeability and protein binding were challenging to assess and both required the use of experimental conditions that minimized the impact of non-specific binding. Intrinsic clearance in human liver microsomes was varied across the data set and several compounds exhibited no measurable substrate loss under the conditions used. Inhibition of cytochrome P450 enzymes was minimal for most compounds.

Conclusions

This is the first data set to describe in vitro properties for 45 legacy and development anti-malarial drugs. The studies identified several practical methodological issues common to many of the more lipophilic compounds and highlighted areas which require more work to customize experimental conditions for compounds being designed to meet the new target product profiles. The dataset will be a valuable tool for malaria researchers aiming to develop PBPK models for the prediction of human PK properties and/or drug–drug interactions. Furthermore, generation of this comprehensive data set within a single laboratory allows direct comparison of properties across a large dataset and evaluation of changing property trends that have occurred over time with changing target product and candidate profiles.

Risk-Benefit Assessment of Ethinylestradiol Using a Physiologically Based Pharmacokinetic Modeling Approach

Current formulations of combined oral contraceptives (COC) containing ethinylestradiol (EE) have ≤35 μg due to increased risks of cardiovascular diseases (CVD) with higher doses of EE. Low‐dose formulations however, have resulted in increased incidences of breakthrough bleeding and contraceptive failure, particularly when coadministered with inducers of cytochrome P450 enzymes (CYP). The developed physiologically based pharmacokinetic model quantitatively predicted the effect of CYP3A4 inhibition and induction on the pharmacokinetics of EE. The predicted C_max and AUC ratios when coadministered with voriconazole, fluconazole, rifampicin, and carbamazepine were within 1.25 of the observed data. Based on published clinical data, an AUC_ss value of 1,000 pg/ml.h was selected as the threshold for breakthrough bleeding. Prospective application of the model in simulations of different doses of EE (20 μg, 35 μg, and 50 μg) identified percentages of the population at risk of breakthrough bleeding alone and with varying degrees of CYP modulation.

Effects of the moderate CYP3A4 inhibitor, fluconazole, on the pharmacokinetics of fesoterodine in healthy subjects

WHAT IS ALREADY KNOWN ABOUT THIS SUBJECT

Available data suggest that fesoterodine dosage should not exceed 4 mg once daily when taken concomitantly with potent CYP3A4 inhibitors, such as ketoconazole. Currently, no information is available on whether dose adjustment is necessary when fesoterodine is administered with a moderate CYP3A4 inhibitor.

WHAT THIS STUDY ADDS

This study shows that adjustment of fesoterodine dose is not warranted when co-administered with a moderate CYP3A4 inhibitor.

AIMS

To assess the effects of fluconazole, a moderate CYP3A4 inhibitor, on the pharmacokinetics (PK) and safety/tolerability of fesoterodine.

METHODS

In this open-label, randomized, two-way crossover study, 28 healthy subjects (18-55 years) received single doses of fesoterodine 8 mg alone or with fluconazole 200 mg. PK endpoints, including the area under the plasma concentration-time curve from 0 to infinity (AUC(0,∞)), maximum plasma concentration (C(max) ), time to C(max) (t(max) ), and half-life (t(1/2) ), were assessed for 5-hydroxymethyl tolterodine (5-HMT), the active moiety of fesoterodine.

RESULTS

Concomitant administration of fesoterodine with fluconazole increased AUC(0,∞) and C(max) of 5-HMT by approximately 27% and 19%, respectively, with corresponding 90% confidence intervals of (18%, 36%) and (11%, 28%). There was no apparent effect of fluconazole on 5-HMT t(max) or t(½) . Fesoterodine was generally well tolerated regardless of fluconazole co-administration, with no reports of death, serious adverse events (AEs) or severe AEs. Following co-administration of fesoterodine with fluconazole, 13 subjects (48%) experienced a total of 40 AEs; following administration of fesoterodine alone, six subjects (22%) experienced a total of 19 AEs. The majority of AEs were of mild intensity. There were no clinically significant changes in laboratory or physical examination parameters.

CONCLUSION

Fesoterodine 8 mg single dose was well tolerated when administered alone or with fluconazole. Based on the observed increase in 5-HMT exposures being within the inherent variability of 5-HMT pharmacokinetics, adjustment of fesoterodine dose is not warranted when co-administered with a moderate CYP3A4 inhibitor provided they are not also inhibitors of transporters.

The evolution of the OATP hepatic uptake transport protein family in DMPK sciences: from obscure liver transporters to key determinants of hepatobiliary clearance

Over the last two decades the impact on drug pharmacokinetics of the organic anion transporting polypeptides (OATPs: OATP-1B1, 1B3 and 2B1), expressed on the sinusoidal membrane of the hepatocyte, has been increasingly recognized. OATP-mediated uptake into the hepatocyte coupled with subsequent excretion into bile via efflux proteins, such as MRP2, is often referred to as hepatobiliary excretion. OATP transporter proteins can impact some drugs in several ways including pharmacokinetic variability, pharmacodynamic response and drug-drug interactions (DDIs). The impact of transporter mediated hepatic clearance is illustrated with case examples, from the literature and also from the Pfizer portfolio. The currently available in vitro techniques to study the hepatic transporter proteins involved in the hepatobiliary clearance of drugs are reviewed herein along with recent advances in using these in vitro data to predict the human clearance of compounds recognized by hepatic uptake transporters.

Case studies addressing human pharmacokinetic uncertainty using a combination of pharmacokinetic simulation and alternative first in human paradigms

PF-184298 ((S)-2,3-dichloro-N-isobutyl-N-pyrrolidin-3-ylbenzamide) and PF-4776548 ((3-(4-fluoro-2-methoxy-benzyl)-7-hydroxy-8,9-dihydro-3H,7H-pyrrolo[2,3-c][1,7]naphthyridin-6-one)) are novel compounds which were selected to progress to human studies. Discordant human pharmacokinetic predictions arose from pre-clinical in vivo studies in rat and dog, and from human in vitro studies, resulting in a clearance prediction range of 3 to >20 mL min⁻¹ kg⁻¹ for PF-184298, and 5 to >20 mL min⁻¹ kg⁻¹ for PF-4776548. A package of work to investigate the discordance for PF-184298 is described. Although ultimately complementary to the human pharmacokinetic data in characterising the disposition of PF-184298 in humans, these data did not provide any further confidence in pharmacokinetic prediction. A fit for purpose human pharmacokinetic study was conducted for each compound, with an oral pharmacologically active dose for PF-184298, and an intravenous and oral microdose for PF-4776548. This provided a relatively low cost, clear decision making approach, resulting in the termination of PF-4776548 and further progression of PF-184298. A retrospective analysis of the data showed that, if the tools had been available at the time, the pharmacokinetics of PF-184298 in human could have been predicted from a population based simulation tool in combination with physicochemical properties and in vitro human intrinsic clearance.

Simulation of human intravenous and oral pharmacokinetics of 21 diverse compounds using physiologically based pharmacokinetic modelling

BACKGROUND

The importance of predicting human pharmacokinetics during compound selection has been recognized in the pharmaceutical industry. To this end there are many different approaches that are applied.

METHODS

In this study we compared the accuracy of physiologically based pharmacokinetic (PBPK) methodologies implemented in GastroPlus™ with the one-compartment approach routinely used at Pfizer for human pharmacokinetic plasma concentration-time profile prediction. Twenty-one Pfizer compounds were selected based on the availability of relevant preclinical and clinical data. Intravenous and oral human simulations were performed for each compound. To understand any mispredictions, simulations were also performed using the observed clearance (CL) value as input into the model.

RESULTS

The simulation results using PBPK were shown to be superior to those obtained via traditional one-compartment analyses. In many cases, this difference was statistically significant. Specifically, the results showed that the PBPK approach was able to accurately predict passive distribution and absorption processes. Some issues and limitations remain with respect to the prediction of CL and active transport processes and these need to be improved to further increase the utility of PBPK modelling. A particular advantage of the PBPK approach is its ability to accurately predict the multiphasic shape of the pharmacokinetic profiles for many of the compounds tested.

CONCLUSION

The results from this evaluation demonstrate the utility of PBPK methodology for the prediction of human pharmacokinetics. This methodology can be applied at different stages to enhance the understanding of the compounds in a particular chemical series, guide experiments, aid candidate selection and inform clinical trial design.

An evaluation of ondansetron binding interactions with human cytochrome P450 enzymes CYP3A4 and CYP2D6

The results of an evaluation study of ondansetron binding to human cytochromes P450 CYP3A4 and CYP2D6 is reported. The methodology includes NMR spectroscopic measurements of substrate to heme iron distances together with molecular modelling of the enzyme-substrate interactions. It is shown that there is a generally good agreement between the experimental and calculated binding affinities for ondansetron towards CYP2D6 and CYP3A4 enzymes, based on interactive docking studies. Moreover, the modelled binding orientations for ondansetron in CYP2D6 and CYP3A4 are largely consistent with the NMR data and with the known routes for P450-mediated metabolism of this compound.

Quantitative structure–activity relationships (QSARs) in CYP3A4 inhibitors: The importance of lipophilic character and hydrogen bonding

The results of Quantitative Structure-Activity Relationship (QSAR) analyses on three series of CYP3A4 inhibitors are reported for enzyme inhibition expressed as Ki values. These include a small series of structurally related statins and two larger groupings of structurally diverse compounds, some of which display competitive inhibition of CYP3A4 whereas others act via heme iron ligation. In all cases, however, it is apparent that there are lipophilicity relationships associated with CYP3A4 inhibitory activity in the total of 46 compounds investigated. This is evidenced by linear correlations between inhibition of CYP3A4 and the octanol-water partition coefficient (P value) when expressed logarithmically (ie., log P). In the case of the statins, however, the distribution coefficient (D) at pH 7 is used due to the effect of compound ionization. Conversion of equilibrium constants (ie. Ki and P) to the corresponding free energy changes (deltaG values) facilitates exploration of the likely intermolecular forces of interaction between the inhibitors and the active site region of CYP3A4. In this respect, there appears to be good agreement between QSAR analyses and molecular modelling of the CYP3A4 enzyme itself, and both are consistent with the known mechanisms of inhibition displayed.

Lipophilicity Relationships in Inhibitors of CYP2C9 and CYP2C19 Enzymes

Quantitative structure-activity relationships (QSARs) within a series of cytochrome P450 2C9 (CYP2C9) and cytochrome P450 2C19 (CYP2C19) inhibitors are reported. In particular, it is noted that compound lipophilicity, in the form of log P values (where P is the octanol/water partition coefficient), is an important factor in explaining the variation in inhibitory potency within these series of compounds, many of which also act as substrates for the respective enzymes. In addition, there is a role for hydrogen bonding and pi-pi stacking interactions within the P450 active site which represent secondary factors in the binding processes of these compounds.

Quantitative structure-activity relationships within a homologous series of 7-alkoxyresorufins exhibiting activity towards CYP1A and CYP2B enzymes: molecular modelling studies on key members of the resorufin series with CYP2C5-derived models of human CYP1A1, CYP1A2, CYP2B6 and CYP3A4

1. The results of quantitative structure-activity relationships (QSARs) within a homologous series of 7-n-alkoxyresorufins are reported. They are consistent with homology modelling of the relevant P450s involved in their metabolism. 2. QSARs were generated for activities involving CYP1A and CYP2B enzymes with structural descriptors relating to compound planarity and other shape parameters, together with certain features of the n-alkoxyresorufin electronic structure, especially electron densities and superdelocalizabilities. 3. A quadratic relationship between compound lipophilicity and binding to CYP2B enzymes was apparent, and which indicated maximal interaction for 7-pentoxyresorufin. Such indications help to explain enzyme selectivity in terms of optimal alkyl chain length for fitting within the relevant active site region. 4. Calculation of the binding affinities for methoxy-, ethoxy-, pentoxy- and benzyloxy-resorufins towards either CYP1A2 or CYP2B6 enzymes, depending on the 7-alkoxyresorufin agree favourably with experimental values obtained from K(m) determinations.

Homology modelling of CYP3A4 from the CYP2C5 crystallographic template: analysis of typical CYP3A4 substrate interactions

1. The results of homology modelling of cytochrome P4503A4 (CYP3A4), which is a human enzyme of major importance for the Phase 1 metabolism of drug substrates, from the CYP2C5 crystal structure is reported. 2. The overall homology between the two protein sequences was generally good (46%) with 24% of amino acid residues being identical and a 22% similarity between matched pairs in the CYP3A4 and CYP2C5 aligned sequences, thus indicating that CYP2C5 represents a viable template for modelling CYP3A4 by homology. 3. The CYP3A4 model appears to show consistency with the reported findings from the extensive site-directed mutagenesis studies already published. 4. Typical CYP3A4 substrates, such as midazolam, testosterone, nifedipine and verapamil, are shown to fit the putative active site of the enzyme structure in a manner consistent with their known positions of metabolism.

Maraviroc: in vitro assessment of drug-drug interaction potential

AIMS

To characterize the cytochrome P450 enzyme(s) responsible for the N-dealkylation of maraviroc in vitro, and predict the extent of clinical drug-drug interactions (DDIs).

METHODS

Human liver and recombinant CYP microsomes were used to identify the CYP enzyme responsible for maraviroc N-dealkylation. Studies comprised enzyme kinetics and evaluation of the effects of specific CYP inhibitors. In vitro data were then used as inputs for simulation of DDIs with ketoconazole, ritonavir, saquinavir and atazanvir, using the Simcyptrade mark population-based absorption, distribution, metabolism and elimination (ADME) simulator. Study designs for simulations mirrored those actually used in the clinic.

RESULTS

Maraviroc was metabolized to its N-dealkylated product via a single CYP enzyme characterized by a K(m) of 21 microM and V(max) of 0.45 pmol pmol(-1) min(-1) in human liver microsomes and was inhibited by ketoconazole (CYP3A4 inhibitor). In a panel of recombinant CYP enzymes, CYP3A4 was identified as the major CYP responsible for maraviroc metabolism. Using recombinant CYP3A4, N-dealkylation was characterized by a K(m) of 13 microM and a V(max) of 3 pmol pmol(-1) CYP min(-1). Simulations therefore focused on the effect of CYP3A4 inhibitors on maraviroc pharmacokinetics. The simulated median AUC ratios were in good agreement with observed clinical changes (within twofold in all cases), although, in general, there was a trend for overprediction in the magnitude of the DDI.

CONCLUSION

Maraviroc is a substrate for CYP3A4, and exposure will therefore be modulated by CYP3A4 inhibitors. Simcyptrade mark has successfully simulated the extent of clinical interactions with CYP3A4 inhibitors, further validating this software as a good predictor of CYP-based DDIs.

Application of CYP3A4 in vitro data to predict clinical drug-drug interactions; predictions of compounds as objects of interaction

WHAT IS ALREADY KNOWN ABOUT THIS SUBJECT

Numerous retrospective analyses have shown the utility of in vitro systems for predicting potential drug-drug interactions (DDIs). Prediction of DDIs from in vitro data is commonly obtained using estimates of enzyme K(i), inhibitor and substrate concentrations and absorption rate for substrate and inhibitor.

WHAT THIS STUDY ADDS

Using a generic approach for all test compounds, the findings from the current study showed the use of recombinant P450s provide a more robust in vitro measure of P450 contribution (fraction metabolized, f(m)) than that achieved when using chemical inhibitors in combination with human liver microsomes, for the prediction of potential CYP3A4 drug-drug interactions prior to clinical investigation. The current study supported the use of SIMCYP(R), a modelling and simulation software in utilizing the in vitro measures in the prediction of potential drug-drug interactions.

AIMS

The aim of this study was to explore and optimize the in vitro and in silico approaches used for predicting clinical DDIs. A data set containing clinical information on the interaction of 20 Pfizer compounds with ketoconazole was used to assess the success of the techniques.

METHODS

The study calculated the fraction and the rate of metabolism of 20 Pfizer compounds via each cytochrome P450. Two approaches were used to determine fraction metabolized (f(m)); 1) by measuring substrate loss in human liver microsomes (HLM) in the presence and absence of specific chemical inhibitors and 2) by measuring substrate loss in individual cDNA expressed P450s (also referred to as recombinant P450s (rhCYP)) The fractions metabolized via each CYP were used to predict the drug-drug interaction due to CYP3A4 inhibition by ketoconazole using the modelling and simulation software SIMCYP.

RESULTS

When in vitro data were generated using Gentest supersomes, 85% of predictions were within two-fold of the observed clinical interaction. Using PanVera baculosomes, 70% of predictions were predicted within two-fold. In contrast using chemical inhibitors the accuracy was lower, predicting only 37% of compounds within two-fold of the clinical value. Poorly predicted compounds were found to either be metabolically stable and/or have high microsomal protein binding. The use of equilibrium dialysis to generate accurate protein binding measurements was especially important for highly bound drugs.

CONCLUSIONS

The current study demonstrated that the use of rhCYPs with SIMCYP provides a robust in vitro system for predicting the likelihood and magnitude of changes in clinical exposure of compounds as a consequence of CYP3A4 inhibition by a concomitantly administered drug.

Application of CYP3A4 in vitro data to predict clinical drug-drug interactions; predictions of compounds as objects of interaction

WHAT IS ALREADY KNOWN ABOUT THIS SUBJECT

Numerous retrospective analyses have shown the utility of in vitro systems for predicting potential drug-drug interactions (DDIs). Prediction of DDIs from in vitro data is commonly obtained using estimates of enzyme K(i), inhibitor and substrate concentrations and absorption rate for substrate and inhibitor.

WHAT THIS STUDY ADDS

Using a generic approach for all test compounds, the findings from the current study showed the use of recombinant P450s provide a more robust in vitro measure of P450 contribution (fraction metabolized, f(m)) than that achieved when using chemical inhibitors in combination with human liver microsomes, for the prediction of potential CYP3A4 drug-drug interactions prior to clinical investigation. The current study supported the use of SIMCYP(R), a modelling and simulation software in utilizing the in vitro measures in the prediction of potential drug-drug interactions.

AIMS

The aim of this study was to explore and optimize the in vitro and in silico approaches used for predicting clinical DDIs. A data set containing clinical information on the interaction of 20 Pfizer compounds with ketoconazole was used to assess the success of the techniques.

METHODS

The study calculated the fraction and the rate of metabolism of 20 Pfizer compounds via each cytochrome P450. Two approaches were used to determine fraction metabolized (f(m)); 1) by measuring substrate loss in human liver microsomes (HLM) in the presence and absence of specific chemical inhibitors and 2) by measuring substrate loss in individual cDNA expressed P450s (also referred to as recombinant P450s (rhCYP)) The fractions metabolized via each CYP were used to predict the drug-drug interaction due to CYP3A4 inhibition by ketoconazole using the modelling and simulation software SIMCYP.

RESULTS

When in vitro data were generated using Gentest supersomes, 85% of predictions were within two-fold of the observed clinical interaction. Using PanVera baculosomes, 70% of predictions were predicted within two-fold. In contrast using chemical inhibitors the accuracy was lower, predicting only 37% of compounds within two-fold of the clinical value. Poorly predicted compounds were found to either be metabolically stable and/or have high microsomal protein binding. The use of equilibrium dialysis to generate accurate protein binding measurements was especially important for highly bound drugs.

CONCLUSIONS

The current study demonstrated that the use of rhCYPs with SIMCYP provides a robust in vitro system for predicting the likelihood and magnitude of changes in clinical exposure of compounds as a consequence of CYP3A4 inhibition by a concomitantly administered drug.

Quantitative structure-activity relationships (QSARs) in inhibitors of various cytochromes P450: the importance of compound lipophilicity

The results of extensive quantitative structure-activity relationship (QSAR) analyses on 15 series of cytochrome P450 inhibitors, covering a total of 7 enzymes and 199 compounds, are reported. In general, it is found that lipophilicity represents the most important single factor in describing differences in inhibitory potency towards P450 enzymes. In two instances, this relationship is parabolic in nature but, by and large, the logarithm of inhibitory activity relates linearly with log P, where P is the octanol-water partition coefficient. On occasions, other parameters are involved in the QSAR expressions but there are many examples where either log P or its ionization-corrected equivalent, log D7.4, are the sole structural descriptors of inhibition. The correlations presented exhibit a range in R value from 0.85 to 0.99, where R is the correlation coefficient, and it is found that R is greater than 0.9 in 80% of the QSARs presented. It is apparent from these findings, therefore, that compound lipophilicity plays a major role in the ability of xenobiotics to inhibit enzymes of the cytochrome P450 superfamily, presumably due to the essentially hydrophobic nature of the active site region.

Induction of cytochromes P450

The induction of cytochromes P450 (CYPs) has been appreciated for some time but an understanding of the mechanisms involved has been poorly understood until recently. The discovery of the role of nuclear receptors such as the pregnane X receptor (PXR) and the constitutive androstane receptor (CAR) has provided a major trigger for research in this area. This work has provided an explanation for species differences in hepatic induction. The production of a PXR crystal structure in the presence and absence of known high affinity ligands has offered the possibility of predicting structures which may bind to the receptor and hence act as inducing agents in man. An improvement in the technology of hepatocyte culture, access to good quality human hepatocytes and the miniaturisation of cultured preparations has meant that the potential of this technique to predict induction in man has been realised. Molecular biological techniques have also proved essential in both the science and the quantitation of CYP induction. The use of transient transfection cell based systems coupled with reporter gene assays have meant that dose response curves can be generated for many chemicals. Assays have been developed to measure the increase of the corresponding CYP mRNAs in primary hepatocytes and some cell lines with a high degree of sensitivity and specificity (allowing the quantitation of closely related CYPs). Although CYP induction is not usually considered as a major drawback in drug development, the aim should be to eliminate or reduce the inducing effects of a new drug to a minimum. Thus, it is essential to increase our understanding of the complex mechanisms that regulate induction and to pay attention to both the dose and the physicochemical and structural properties of CYP inducing agents.

Compound lipophilicity for substrate binding to human P450s in drug metabolism

Compound lipophilicity is of key importance to P450 binding affinity and enzyme selectivity. Here, lipophilicity is discussed with reference to the human drug-metabolizing P450 enzymes of families CYP1, CYP2 and CYP3. From an extensive compilation of log P values for P450 substrates, and by analysis of relationships between partitioning energy and substrate-binding free energy, the relevance of lipophilicity and other factors pertaining to P450 binding affinity is explained, leading to the formulation of lipophilicity relationships within substrates of each human P450 enzyme involved in drug metabolism. Furthermore, log P values for P450 substrates appear to represent markers for enzyme selectivity. Together with the important roles of hydrogen bonding and pi-pi stacking interaction energies, the desolvation of the P450 active site makes a major contribution to the overall substrate-binding energy and, consequently, a good agreement with experimental information is reported based on this analysis.

Investigation of enzyme selectivity in the human CYP2C subfamily: homology modelling of CYP2C8, CYP2C9 and CYP2C19 from the CYP2C5 crystallographic template

Homology modelling of human CYP2C subfamily enzymes, CYP2C8, CYP2C9 and CYP2C19, based on the rabbit CYP2C5 crystal structure template is reported. The relatively high sequence homologies (75-80%) between the rabbit CYP2C5 and human CYP2C subfamily enzymes tend to indicate that the resulting structures should prove adequate models of these major catalysts of human drug metabolism. Selective substrates of all three human CYP2C enzymes are found to fit closely within the putative active sites in a manner which is consistent with site-directed mutagenesis experiments and known positions of substrate metabolism. The specific interactions between substrates and enzymes can be used to rationalize the variation in substrate binding affinity and generate QSAR models for both inhibition and metabolism via CYP2C family enzymes, yielding a generally good agreement with experimental binding data obtained from Km values, with correlation coefficients (R values) of between 0.97 and 0.99 depending on the QSAR equation produced.

Substrates of Human Cytochromes P4S0 from Families CYP1 and CYP2: Analysis of Enzyme Selectivity and Metabolism

A compilation of information relating to substrate metabolism via human cytochromes P450 (CYP) from the CYP1 and CYP2 families is reported. The data presented include details of preferred sites of metabolism and Km values (usually for the expressed enzymes) for each reaction for selected substrates of CYP1A2, CYP2A6, CYP2B6, CYP2C8, CYP2C9, CYP2C19, CYP2D6 and CYP2E1. Although other P450 databases are available, they do not provide such information as is collated here, and which can prove useful for comparing P450 substrate characteristics. This information can be employed in analysing the structural requirements for human P450 enzyme selectivity and for establishing various rules regarding preferred site of metabolism for selective P450 substrates. For example, in most cases it would appear that there is a set number of intervening 'heavy' atoms (atoms other than hydrogen) between sites of metabolism and key hydrogen bond acceptors (or donors) for human P450 substrates, with the number of intervening atoms being dependent upon the type of P450 involved.

Molecular modelling of CYP2B6 based on homology with the CYP2C5 crystal structure: analysis of enzyme-substrate interactions

The results of homology modelling of CYP2B6 based on the CYP2C5 crystal structure is described in terms of substrates and inhibitors binding within the putative active site. In general these results are in agreement with currently available evidence from substrate metabolism, mode of inhibitor action and site-directed mutagenesis experiments within the CYP2B subfamily of enzymes. Consequently, the model based on the CYP2C5 template represents an advance on those models produced from bacterial P450s, such as CYP101 and CYP102. Quantitative Structure-Activity Relationships (QSARs) for substrates binding to CYP2B6 indicate a key role for hydrogen bonding, and lipophilic character, as determined by the log P parameter (where P is the octanol/water partition coefficient), is also of importance for explaining the variation in experimental binding affinity for CYP2B6 substrates. It is possible to estimate the binding energies for typical CYP2B6 substrates based on their properties and interactions with the enzyme, which show good concordance with experimental data in the form of apparent Km values.

Homology modelling of CYP2A6 based on the CYP2C5 crystallographic template: enzyme-substrate interactions and QSARs for binding affinity and inhibition

The results of homology modelling of the human P450 enzyme CYP2A6, based on the CYP2C5 crystallographic template structure are reported. A substantial number of selective substrates of the CYP2A6 enzyme fit the putative active site in a manner that is consistent with their known metabolites. Moreover, the evidence from site-directed mutagenesis experiments is in accordance with the current model, particularly in relation to complementary amino acid contacts within the haem environment. The binding of substrates is rationalized in terms of QSAR analyses and from a consideration of the contributory factors affecting the binding affinity. The latter approach appears to represent a highly correlated (R=0.99) method for estimating the relative strength of enzyme-substrate binding within CYP2A6-selective compounds, albeit within a fairly limited dataset of substrates.

Homology modelling of human CYP1A2 based on the CYP2C5 crystallographic template structure

1. The results of homology modelling of human cytochrome P4501A2 (CYP1A2) based on the CYP2C5 crystal structure are reported. It exhibits improved sequence homology relative to that of CYP102. 2. It was demonstrated that many selective substrates for CYP1A2 could fit within the putative active site of the enzyme, and in orientations which agree with documented evidence for CYP1A2-mediated metabolism. 3. Furthermore, a number of amino acid residues lining the haem pocket have been shown, via site-directed mutagenesis, to have an influence on substrate metabolism, and these experimental findings from the literature are consistent with the modelled interactions for selective substrates. 4. The binding affinities of several CYP1A2 substrates have also been calculated from the CYP1A2 active site interactions and they agree closely with experimental values.

Homology modelling of human CYP2E1 based on the CYP2C5 crystal structure: investigation of enzyme-substrate and enzyme-inhibitor interactions

The construction of a homology model of human cytochrome P450 2E1 (CYP2E1) is reported, based on the CYP2C5 crystallographic template. A relatively high degree of primary sequence homology (identity=59%), as expected for proteins of the same CYP family, ensured a straightforward generation of the 3-dimensional model due to relatively few deletions and insertions of amino acid residues with respect to the CYP2C5 crystal structure. Probing the CYP2E1 model with typical substrates of the enzyme showed a good agreement with experimental information in the form of positions of metabolism for substrates, and with site-directed mutagenesis data on certain residues. Furthermore, quantitative relationships between substrate binding affinity and various structural parameters associated with the substrate molecules facilitated the formulation of a procedure for estimating relative binding energy and, consequently, K(m) or K(D) values towards the CYP2E1 enzyme. This method has been based on a consideration of the active site interactions between substrates and key amino acid residues lining the haem pocket, together with compound lipophilicity data from partition coefficients.

Homology modelling of the nuclear receptors: human oestrogen receptorbeta (hERbeta), the human pregnane-X-receptor (PXR), the Ah receptor (AhR) and the constitutive androstane receptor (CAR) ligand binding domains from the human oestrogen receptor alpha (hERalpha) crystal structure, and the human peroxisome proliferator activated receptor alpha (PPARalpha) ligand binding domain from the human PPARgamma crystal structure

We have generated by homology the three-dimensional structures of the ligand binding domain (LBD) of several interrelated human steroid hormone receptors (SHRs). These are the oestrogen receptor beta (hERbeta), the pregnane-X-receptor (PXR), the Ah receptor (AhR) and the constitutive androstane receptor (CAR). They were produced by homology modelling from the human oestrogen receptor alpha (hERalpha) crystallographic coordinates [Nature 389 (1997) 753] as a template together with the amino acid sequences for hERbeta [FEBS Lett. 392 (1996) 49], PXR [J. Clin. Invest. 102 (1998) 1016], AhR [Proc. Natl. Acad. Sci. U.S.A. 89 (1992) 815] and CAR [Nature 395 (1998) 612; Mol. Cell. Biol. 14 (1994) 1544], respectively. The selective endogenous ligand, in each case, was docked interactively within the putative ligand binding site using the position of oestradiol in hERalpha as a guide, and the total energy was calculated. In each receptor model a number of different ligands known to fit closely within the ligand binding site were interactively docked and binding interactions noted. Specific binding interactions included combinations of hydrogen bonding and hydrophobic contacts with key amino acid sidechains, which varied depending on the nature of the ligand and receptor concerned. We also produced the human peroxisome proliferator activated receptor alpha (PPARalpha) by homology modelling using the human PPARgamma (hPPARgamma) LBD crystallographic coordinates summarised in [Toxicol. In Vitro 12 (1998) 619] as a template together with the amino acid sequence for hPPARalpha [Toxicol. In Vitro 12 (1998) 619; Nature 395 (1998) 137]. The models will provide a useful tool in unravelling the complexity in the physiologic response to xenobiotics by examining the ligand binding interactions and differences between the steroid hormone receptors activation or inactivation by their ligands.

A MOLECULAR MODEL OF CYP2D6 CONSTRUCTED BY HOMOLOGY WITH THE CYP2C5 CRYSTALLOGRAPHIC TEMPLATE: INVESTIGATION OF ENZYME-SUBSTRATE INTERACTIONS

The results of homology modelling of CYP2D6 based on the mammalian P450 crystal structure of rabbit CYP2C5 are reported. It is found that many CYP2D6-selective substrates are able to fit closely within the putative active site of the enzyme where there are favourable contacts with complementary amino acid residues, including aspartate-301 which has been probed via site-directed mutagenesis. The homology model of CYP2D6 is consistent with available experimental evidence from selective substrate metabolism and site-specific mutation data. Quantitative structure-activity relationships (QSARs) with substrate binding affinity based on KD values and inhibition data (Ki values) demonstrate the importance of hydrogen bonding, pi-pi stacking and relative molecular mass in describing variations in avidity towards the CYP2D6 enzyme, although the compound lipophilicity (log D(7.4)) appears to be the most important single descriptor for CYP2D6 inhibition. Calculation of substrate binding affinity based on contributions from active site interactions and lipophilic character gives satisfactory agreement with experimentally determined KD values.

Substrate SARs in human P450s

Drug metabolism is now an integral part of the drug discovery process, and the cytochromes P450 (CYPs) are the most important family of enzymes involved in human drug metabolism. An increased understanding of the properties of the substrates for the major human CYPs is thus highly desirable. This article shows how key characteristics of CYP substrates, such as lipophilicity, molecular mass and hydrogen-bonding potential, govern selectivity towards individual CYPs. Importantly, the variation in binding affinities of 60 human CYP substrates can be explained by understanding the key physicochemical, structural and electronic characteristics that govern substrate binding to each isozyme.

Quantitative structure--activity relationships for inducers of cytochromes P450 and nuclear receptor ligands involved in P450 regulation within the CYP1, CYP2, CYP3 and CYP4 families

The results of quantitative structure-activity relationships (QSARs) are reported for several series of cytochrome P450 inducers, including those which also act as ligands for the various nuclear receptors involved in regulation of the relevant P450 genes, namely, CYP1, CYP2, CYP3 and CYP4. In several examples presented, the QSARs are consistent with homology modelling studies of the nuclear receptor ligand-binding domains (LBDs) based on available crystal structures of the oestrogen and peroxisome proliferator-activated receptors' LBDs. Good correlations (R=0.91-0.99) are found between various structural parameters and biological activity (either in the form of P450 induction or ligand-binding affinity) for the Ah receptor (AhR), human estrogen receptor alpha (hER alpha), human glucocorticoid receptor (hGR) and the rat peroxisome proliferator-activated receptor alpha (rPPAR alpha).

Molecular modelling of the peroxisome proliferator-activated receptor alpha (PPAR alpha) from human, rat and mouse, based on homology with the human PPAR gamma crystal structure

The generation of homology models of human, rat and mouse peroxisome proliferator-activated receptor alpha (PPAR alpha) are reported, based on the recently published crystal structure of the human PPAR gamma ligand-binding domain (LBD) with bound ligand, rosiglitazone. It is found that a template of peroxisome proliferating fibrate drugs and related compounds can fit within the putative ligand-binding site of rat PPAR alpha, via contacts with amino acid residues which are consistent with their biological potency for peroxisome proliferation, site-directed mutagenesis experiments and with quantitative structure-activity relationship (QSAR) analysis studies. The experimental binding affinity of leukotriene B(4) (LTB(4)) for the mouse PPAR alpha agrees closely with the calculated value based on the modelled interactions, whereas selective PPAR alpha ligands such as clofibric acid are able to fit the human PPAR alpha binding site in agreement with reported site-directed mutagenesis information.

Structure-activity relationship for human cytochrome P450 substrates and inhibitors

Criteria governing the avidity of substrate binding to human hepatic cytochromes P450 (CYP) associated with Phase 1 metabolism of drugs are described. The results of extensive quantitative structure-activity relationship (QSAR) analyses are reported for substrates of human P450s: CYPIA2, CYP2A6, CYP2B6, CYP2C9, CYP2C19, CYP2D6, CYP2E1, and CYP3A4, representing the enzymes exhibiting major involvement in the metabolism of drug substrates in Homo sapiens. In particular, it is shown that hydrogen bond properties in each class of enzyme-substrate complex are especially important factors in determining substrate binding affinity towards those human P450s which are involved in drug metabolism.

Quantitative structure-activity relationships (QSARs) within substrates of human cytochromes P450 involved in drug metabolism

The results of quantitative structure-activity relationship (QSAR) analyses are reported for structurally diverse series of chemicals which act as substrates or inhibitors for human hepatic microsomal cytochromes P450 (CYP). In particular, this study focuses on the major catalysts of drug metabolism in man, namely CYP1A2, CYP2B6, CYP2C9, CYP2C19, CYP2D6 and CYP3A4. It is found that good correlations (with correlation coefficients ranging from R = 0.94 to 0.99) with P450 binding affinity (Km and K(D)) or competitive inhibition (Ki) values are obtained in each case, especially when consideration of hydrogen bonding parameters are included in the QSAR analysis, together with the number of pi-pi stacking interactions.

Factors influencing rates and clearance in P450-mediated reactions: QSARs for substrates of the xenobiotic-metabolizing hepatic microsomal P450s

Various contributory factors associated with the kinetics of cytochrome P450-mediated catalytic activity and the metabolic clearance of drug substrates are discussed and evaluated, based on literature data and physicochemical parameters. Quantitative relationships between molecular structure and biological activity for several series of P450 substrates are presented which point to certain commonalities in P450-catalyzed reactions. In particular, it appears that frontier orbital energies are especially important for the estimation of reaction rates and clearance for many P450 substrates, although occasionally these have to be combined with other descriptors, such as compound lipophilicity (in the form of logP or logD(74)).

An interaction between the cytochrome P450 probe substrates chlorzoxazone (CYP2E1) and midazolam (CYP3A)

AIMS

The use of multiple probe substrates to evaluate the activity of drug metabolizing enzymes requires that there are no inter-substrate interactions. As part of a series of studies to develop a clinically useful collection of probe substrates that could be given alone or in any combination, we observed an interaction between midazolam (MDZ) and another component of the six-drug cocktail. Published data indicated that the interacting component was likely to be chlorzoxazone. This was investigated as part of a second study. The data relating to the interaction from both studies are reported here.

METHODS

Both studies were performed in 16 healthy subjects. All treatments were given orally after an overnight fast. In study 1, which was performed to a four-period, open, crossover design, subjects received on separate occasions MDZ 5 mg, diclofenac 25 mg, a four drug cocktail (caffeine 100 mg, mephenytoin 100 mg, debrisoquine 10 mg and chlorzoxazone 250 mg) and a six drug cocktail (caffeine 100 mg, mephenytoin 100 mg, debrisoquine 10 mg, chlorzoxazone 250 mg, diclofenac 25 mg and MDZ 5 mg). In study 2, which was performed to a two-period, open, crossover design, subjects received a five drug cocktail (as the six drug cocktail in the first study, but without chlorzoxazone and with diclofenac dose increased to 50 mg) and a six drug cocktail (as five drug cocktail, with chlorzoxazone 250 mg). In both studies, blood samples were taken for measurement of plasma MDZ and 1-hydroxy MDZ (1-OH MDZ) concentrations. In study 1, blood samples were taken up to 12 h post-dose while in study 2 a single sample was taken 2 h after dosing. In study 1, the potential interaction between MDZ and the other components of the six drug cocktail was assessed by comparing AUClast ratios (1-OH MDZ/MDZ) between the two treatments. Additionally, a single sampling timepoint of 2 h post-dose for determination of concentration, rather than AUC, ratios was established. The 2 h plasma concentration ratios from studies 1 and 2 were combined and a pooled analysis performed to compare ratios within each study (to determine the change in ratio when MDZ was dosed with and without chlorzoxazone) and between studies (to determine the consistency of the ratios when MDZ was given either as part of the two six drug cocktails or when given alone and as part of the five drug cocktail).

RESULTS

In study 1, both the AUClast ratio and the 2 h post-dose plasma concentration ratio were reduced when MDZ was given as part of the six drug cocktail in comparison with those for MDZ alone. This was the result of an increase in MDZ, rather than decrease in 1-OH MDZ, concentrations and was considered to result from a reduction in first pass metabolism of MDZ. The geometric mean AUClast values (with 95% CI) for MDZ were 95.6 (79.0, 115.7) and 160.4 (133.6, 192.6) microg l(-1) h when given alone and as part of the six drug cocktail, respectively. The corresponding values for 1-OH MDZ were 789.6 (697.6, 893.6) and 791.4 (701.7, 892.6) microg l(-1) h. The ratio of adjusted geometric mean AUClast ratios for the two treatments was 1.82 (90% CI 1.48, 2.23, P < 0.001). The pooled plasma 1-OH MDZ/MDZ ratio data from both studies showed that the differences in MDZ metabolism observed in study 1 were replicated in study 2. The adjusted geometric mean 1-OH MDZ/MDZ ratios when MDZ was given alone and as part of the six drug cocktail were 7.79 and 4.59, respectively, for study 1 (ratio 1.70, 95% CI 1.36, 2.11, P < 0.001) and 7.64 and 4.60 for study 2 (ratio 1.66, 95% CI 1.34, 2.06, P < 0.001). These data indicate that when given orally chlorzoxazone interacts with MDZ, increasing plasma MDZ concentrations. In contrast, there was no difference between the plasma 1-OH MDZ/MDZ ratios when MDZ was given alone and as part of the five drug cocktail indicating that there were no interactions between MDZ and any of the other components of that cocktail.

CONCLUSIONS

Chlorzoxazone appears to significantly influence the pharmacokinetics of oral MDZ, probably through inhibition of first pass metabolism by CYP3A in the GI tract. Data from these studies and literature evidence showing a further interaction between chlorzoxazone and CYP1A2 substrates and questions concerning the specificity of chlorzoxazone as a probe substrate for CYP2E1, indicate that the use of chlorzoxazone in multisubstrate probe cocktails should be avoided.

Quantitative structure-activity relationships (QSARs) within series of inhibitors for mammalian cytochromes P450 (CYPs)

The results of quantitative structure-activity relationship (QSAR) studies on series of P450 inhibitors are reported. Cytochrome P450 families CYP1, CYP2 and CYP51 have been investigated for QSAR analysis, including those of CYP2 subfamilies: CYP2A, CYP2B, CYP2C, CYP2D and CYP2E. The accumulated evidence indicates different structural descriptors being involved, depending on the P450 enzyme concerned, although compound lipophilicity in the form of either logP or logD(7.4) appears to represent a common factor in some cases. This is thought to represent desolvation of the P450 active site, although quadratic expressions in lipophilicity tend to suggest that membrane transport is important, especially for CYP2B and CYP2E isoforms. In general, there is close agreement (R = 0.95-0.99) between experimental pKi values and those calculated via QSAR analysis.

Cytochrome P450 substrate specificities, substrate structural templates and enzyme active site geometries

The structural characteristics of human cytochrome P450 substrates are outlined in the light of extensive studies on P450 substrate specificity. Templates of superimposed substrates for individual P450 isozymes are shown to fit the corresponding enzyme active sites, where contacts with specific amino acid residues appear to be involved in the interaction with each structural template. Procedures leading to the evaluation of likely P450 specificity, binding affinity and rate of metabolism are described in the context of key examples in which molecular modelling appears to rationalize experimentally observed findings.

Molecular modelling of human CYP2E1 by homology with the CYP102 haemoprotein domain: investigation of the interactions of substrates and inhibitors within the putative active site of the human CYP2E1 isoform

1. The construction of a three-dimensional model of human CYP2E1 is reported. It is based on homology with the haemoprotein domain of the unusual bacterial P450, CYP102, which is of known crystal structure. 2. Interactive docking of a number of human CYP2E1 substrates is consistent with their known positions of CYP2E1-mediated metabolism, where specific interactions with key active site amino acid side-chains appear to rationalize the binding and orientation of substrate molecules. 3. Amino acid residues within the putative active site of human CYP2E1, including those associated with the binding of substrates and inhibitors, are shown to correspond with those identified by site-directed mutagenesis experiments conducted on CYP2 family isoforms, and they are known to affect substrate metabolism regioselectivity. 4. Consequently, it was found that the CYP2E1 active site exhibits complementarity with the structural characteristics of known substrates and inhibitors of this enzyme, including their relatively low molecular weights and disposition of hydrogen bond-forming groups.

Molecular modelling of CYP1 family enzymes CYP1A1, CYP1A2, CYP1A6 and CYP1B1 based on sequence homology with CYP102

Molecular modelling of a number of CYP1 family enzymes from rat, plaice and human is described based on amino acid sequence homology with the haemoprotein domain of CYP102, a unique bacterial P450 of known structure. The interaction of various substrates and inhibitors within the putative active sites of rat CYP1A1, human CYP1A2, a fish CYP1 enzyme CYP1A6 (from plaice) and human CYP1B1, is shown to be consistent with P450-mediated oxidation in each example or, in the case of inhibitors, mechanism of inhibition. It is reported that relatively small changes between the enzymes' active site regions assist in the rationalization of CYP1 enzyme preferences for particular substrate types, and a template of superimposed CYP1A2 substrates is shown to fit the putative active site of the human CYP1A2 enzyme.