- Email: [email protected]
human-derived reagents in drug discovery and development: An industrial perspective
Environmental Toxicology and Pharmacology 21 (2006) 179–183
Utility of human/human-derived reagents in drug discovery and development: An industrial perspective Ruth Hyland ∗ , Barry Jones, Han van de Waterbeemd Pfizer Global Research & Development, PDM, Sandwich, Kent CT13 9NJ, UK Available online 10 August 2005
Abstract The shift to combinatorial chemistry and parallel synthesis in drug discovery has resulted in large numbers of compounds entering the lead seeking and lead development phases of the process. To support this, higher throughput computational (in silico) and in vitro approaches have become the forefront of the drug metabolism and pharmacokinetic (DMPK) input into drug discovery. This has been accompanied by a shift in focus from animal-derived data to human based studies, reflecting the realisation that extrapolation from animals to human has its limitations. In silico approaches may be regarded as human derived tools for DMPK, since models (template/pharmacophore and protein homology modelling), for example, for the human CYP enzymes, are widely used for identifying qualitatively enzyme/substrate interactions. Quantitative assessment of drug metabolism using human hepatocytes or sub-cellular fractions provide a valuable tool both for the screening out of high metabolic lability and in estimations of human intrinsic clearance. In terms of drug absorption, the human colon adenocarcinoma cell line, Caco-2, offers a versatile human derived system for measuring drug permeability, despite over expression of the efflux transporter P-glycoprotein (P-gp). The importance of P-gp can then be further assessed in recombinant systems expressing the human P-gp, where substrate affinity and inhibition potency can be measured, important factors when considering transporter mediated drug–drug interactions. The primary cause of pharmacokinetic-based drug–drug interactions (DDIs) is through enzyme inhibition or induction, with the CYP enzymes being of major importance. Human liver microsomes and hepatocytes are invaluable tools in assessment of DDI vulnerability of new chemical entities, having the capacity to identify enzymes responsible for specific routes of metabolism, and hence areas of vulnerability for a DDI. In addition, human-based screening tools can be used to identify the perpetrator of a DDI through enzyme inhibition/induction. Large differences in the nature of enzymes induced and the extent of induction when comparing animals to man are known. Thus, in vitro models allowing assessment of induction potential in human tissue, establishes some relevance to the clinical situation. © 2005 Elsevier B.V. All rights reserved. Keywords: Human reagents; Drug discovery; In silico ADME; Drug–drug interactions (DDI)
1. Introduction During the last 5–10 years there has been a dramatic change in the drug discovery paradigm. Combinatorial chemistry and parallel synthesis approaches have led to an explosion in terms of the numbers of compounds synthesised, in turn resulting in an increase in the number of compounds entering lead seeking and lead development. When trying to identify a good orally active drug candidate it is important that it possess good absorption properties and metabolic stability in addition to potency and selectivity for the biological target (Smith et al., 1996). In order to support the numbers of compounds from chemistry higher throughput drug metabolism and pharmacokinetic
∗
Corresponding author. Tel.: +44 1304 641386; fax: +44 1304 651987. E-mail address: [email protected] (R. Hyland).
1382-6689/$ – see front matter © 2005 Elsevier B.V. All rights reserved. doi:10.1016/j.etap.2005.07.007
(DMPK) studies have become an essential tool in drug discovery. In order to achieve the rapid feedback for new chemical synthesis, computational and in vitro approaches have become the forefront of DMPK drug discovery. In the field of DMPK there has been a major shift from the use of animal-derived data to human-based studies, the exception being comparative studies for cross species comparisons in order to support animal-based regulatory toxicology studies. This shift is driven in part by the desire to reduce the number of animalbased studies and an increase in the availability of human-based reagents. However, the main impetus for using human-derived reagents has been the appreciation of distinct species differences between animals and humans in terms of pharmacokinetic properties. For example, in terms of drug metabolism the rat is well recognised as possessing good N-acetylation capabilities, whilst the dog has none, and human lies somewhere between the two (Glinsukon et al., 1975; Poirier et al., 1963; Trepanier et al.,
180
R. Hyland et al. / Environmental Toxicology and Pharmacology 21 (2006) 179–183
1997). In terms of drug absorption the architecture of the gastrointestinal tract of the dog makes it more likely for drugs to be absorbed through the paracellular route. The larger pore size and greater frequency of pores (He et al., 1998) can result in the dog grossly over predicting oral bioavailability of drugs absorbed through the paracellular route (Chiou et al., 2000). Hence extrapolation of pharmacokinetic data from animals to human has its limitations. This review will concentrate on the utility of humanderived reagents in the prediction of metabolism, absorption and drug–drug interactions. A summary of human-derived reagents used in drug discovery and development and their potential utility is provided in Table 1. 2. Metabolism 2.1. In silico approaches Although not strictly human reagents, in silico models may be regarded as human-derived tools for DMPK. For example, in silico models for the human CYP enzymes have been available for many years and are continually being refined with the emergence of new data and science. The approaches used include template/pharmacophore models and protein homology modelling (de Groot and Ekins, 2002). Template or pharmacophore modelling involves the correlation of structural elements between a series of compounds known to interact with the enzyme either as substrates or inhibitors, in order to identify key structural features involved in the interaction with the enzyme. This approach is exemplified by the production of a template model for CYP2D6 (de Groot et al., 1999a, 1999b). Protein homology modelling involves the generation
of a model by alignment of the primary amino acid sequence for the protein of interest, with that of closely related proteins, for which the secondary structure is known based upon X-ray crystallography, and assuming the same secondary structure. Until recently the only CYP enzymes for which crystal structures were available were those from bacteria and yeast, which have low sequence homology compared to human forms. In the last 5 years, significant efforts have gone into obtaining crystal structures for mammalian CYPs. In 2000, the solution of the first mammalian crystal structure (rabbit CYP2C5/3) was published (Williams et al., 2000) and more recently the first human CYP2C9 and CYP3A4 crystal structures (Williams et al., 2003; Wester et al., 2004; Yano et al., 2004). The first step in any homology model is the sequence alignment that, although carried out on computer, is operator dependent, and hence two modellers may have a different alignment for the same protein. A comprehensive review of in silico methods and the resulting models for the cytochromes P450 is provided by de Groot et al. (2004). Other areas where in silico models, based upon human derived data can be beneficial is in predicting properties that will provide key information on dose size and dose frequency, such as oral absorption, bioavailability, clearance, volume of distribution and drug–drug interactions (van de Waterbeemd and Gifford, 2003; Dickins and van de Waterbeemd, 2004). For example, in silico techniques have been used to predict intrinsic clearance values. A number of different statistical techniques have been applied to the same data set resulting in the production of computational models for the prediction of hepatocyte intrinsic clearance. An initial study showed that, as might be expected, human clearance could be predicted from hepatocyte data (Schneider et al., 1999). A subsequent study
Table 1 Human derived reagents used in drug discovery and development Human-derived reagents
What’s required
Utility
In silico Simulation Protein homology or pharmacophore model
Various in vitro data Human CYP crystal structures
Prediction of absorption, DDI, PK Substrate/Inhibitor docking
Molecular modelling Expert systems
In vivo metabolite schemes
Prediction of site of metabolism Prediction of human metabolites
QSAR
In vitro data for training model
Oral absorption Volume of distribution Clearance PPB Bioavailability CYP inhibition BBB uptake P-gp substrate
Microsomes/cytosols/S9 Hepatocytes
Clearance, metabolic routes, enzyme inhibition Clearance, metabolic routes, enzyme inhibition/induction Absorption
In vitro Human cell and sub-cellular fractions
Caco-2 Recombinant human proteins
rCYP450 microsomes (and other drug metabolising enzymes) rP-gp membrane Transfected cell-lines (eg MDCK, HEK)
Clearance, CYP inhibition P-gp affinity and inhibition Uptake and Efflux assays (hepatic/renal/BBB transport)
R. Hyland et al. / Environmental Toxicology and Pharmacology 21 (2006) 179–183
took this further by attempting to identify a three-dimensional QSAR for the prediction of intrinsic clearance using molecular features such as hydrogen bond acceptors and hydrogen bond donors (Ekins and Obach, 2000). 2.2. In vitro approaches It is possible to use human in vitro hepatic systems, principally whole cell systems (hepatocytes) or sub-cellular fractions (S9, microsomes or cytosol), to rank relative metabolic liability and to predict the in vivo clearance of a given compound. In the latter case, the principles behind the use of these systems are the same, in that first the rate of metabolism of the compound is determined experimentally using the in vitro system. This rate can be determined from following the disappearance of the drug molecule from the system or by measuring the rate of formation of one or more of the compound’s metabolites. The rate is then scaled using physiological scaling factors appropriate to the particular system to give an estimate of the intrinsic clearance of the compound (Clint ). In the case of the liver, the total clearance of the compound can be estimated from the intrinsic clearance by application of a hepatic extraction model such as the ‘wellstirred’ or ‘parallel tube’ models. Of these, the simplest and often the most widely used is the ’well-stirred’ model, which relates intrinsic clearance to total clearance via the following equation: Cl =
Clint fu Q Clint fu + Q
where fu is the free fraction of the compound in blood and Q is liver blood flow. These methods for estimating the clearance of a compound from in vitro systems have been extensively studied and reviewed (Obach, 2001; Houston and Carlile, 1997; Iwatsubo et al., 1997) In general, if the appropriate experimental conditions are used, both hepatocytes and microsomes will give reasonable predictions of total clearance. However, the different systems have advantages and disadvantages. For example, microsomes can only be reliably used for compounds where most of the clearance is mediated by the cytochrome P450 system, which represents a large majority of drugs. However, it can be argued that hepatocytes are of greater utility since all the major enzyme systems involved in drug metabolism are present in the cells. Prediction of human clearance by microsomes and hepatocytes can be carried out in a high throughput automated mode. However, to make this process even faster in silico techniques have also been employed, as discussed earlier. 3. Absorption 3.1. Cell permeability The permeation rate of compounds through monolayers of the human colon adenocarcinoma cell line, Caco-2, is one of the most widely used systems for predicting absorption (Artursson et al., 1996). In this system, the cells are grown on a permeability
181
filter support and the passage of compounds is then measured across the monolayer from a donor to an acceptor compartment and gives an assessment of permeability. Correlations have been observed between permeability across Caco-2 monolayers and the extent of absorption in man (Molero-Monfort et al., 2001). For example, for a series of six non-peptidic compounds exhibiting between 5 and 100% absorption in man, permeability measured in the Caco-2 experiments ranged from 0.5 to >50 × 10−4 cm/min and all of the compounds that showed permeability >20 × 10−4 cm/min were completely absorbed. One complicating factor related to this system, which may limit its predictive ability in terms of absorption, is the apparent over expression of P-glycoprotein (P-gp). Caco-2 cell monolayers are polarised with apical and basal surfaces to the cell monolayers. The P-gp efflux pump is expressed on the apical side of the monolayer and has been shown to mediate the efflux of compounds such as cyclosporin A. Hence, substrates for this transport protein will be prevented from crossing the monolayer due to efflux back into the donor compartment. P-gp levels vary between Caco-2 cell preparations, introducing inter-experimental variability, and the transporter may also be saturated by some substrates. These variables clearly complicate extrapolation from the in vitro to in vivo situation. However, Caco-2 cells do still offer a versatile human derived system for measuring drug permeability. Although they require considerable time, expense and effort to establish and maintain the cell line, they do provide an extremely simple system to measure permeability with straightforward analysis of drug in the donor and acceptor buffer solutions. This system is also ideally suited for the measurement of passive, transcellular permeation, which is the route of absorption for the majority of drugs. However, for compounds where paracellular absorption may be involved, the absence of aqueous pores limits the applicability of the Caco-2 cell model. The presence of P-gp in Caco-2 cells raises the issue that the permeability of a compound in this system may be a function of two structure activity relationships (SARs): one for intrinsic permeability and one for interaction with, and efflux by, P-gp. These SARs can be deconvoluted by the use of assays that are designed to investigate these processes directly. 3.2. P-gp: affinity/binding To assess the extent of P-gp transport, binding and/or inhibition, a number of approaches are currently in use, most of which rely upon recombinant expression systems for the human P-gp protein. Indirect methods can be used, based on measurement of the effect of membrane transport using a P-gp binding assay, an ATPase assay, or a calcein AM test. The ATPase assays rely on the measurement of inorganic phosphate which is released when ATP is converted to ADP and so providing the energy for the transport of a drug molecule. Calcein AM is a non-fluorescent P-gp substrate, which is transformed into Calcein, a fluorescent non-substrate. Hence in the absence of a P-gp inhibitor cells will be non-fluorescent, but will show an increasing amount of fluorescence in the presence of increasing concentrations of a P-gp inhibitor.
182
R. Hyland et al. / Environmental Toxicology and Pharmacology 21 (2006) 179–183
A comparison has been made between monolayer efflux, ATPase and calcein-AM assays (Polli et al., 2001). Some of the compounds showed concordance across all the assays, while others revealed differences related to their apparent permeability. Transporters mediated drug–drug interactions (DDI) are becoming an important research topic and the above assays can help to quantify the potential for such interactions. Similar screens could be developed for other transporters. It should be possible to separate passive diffusion from efflux transport processes through a rational combination of these screens, and this would allow the study of structure-activity relationships for each of these processes separately. Much more still needs to be done to fully understand and predict the clinical relevance of P-gp binding and inhibition (Ayrton and Morgan, 2001). 4. Drug–drug interactions (DDI) The primary cause of a pharmacokinetic DDI is as a result of a modulation in enzyme activity, namely enzyme inhibition or enzyme induction (Li and Jurima-Romet, 1997). A compound may be associated with DDIs as either a perpetrator or victim. That is, the compound may be an enzyme inhibitor or inducer itself (perpetrator) thereby altering the pharmacokinetics of co-administered drugs. For example, the CYP3A4 inhibitor ketoconazole, is well documented as a perpetrator, is contra-indicated with many CYP3A4 substrates and can cause life-threatening drug–drug interactions (Honig et al., 1993). Enzyme induction will result in the more rapid clearance of substrates and thus whilst not having toxic consequence will result in sub-therapeutic drug concentrations. This is of particular importance for DDIs associated with the oral contraceptive, where a lack of contraceptive cover and thus unwanted pregnancies may arise (Shenfield, 1986). As a victim, the compound may be metabolised by enzymes whose activities are being modulated by a co-administered drug or through genetic factors. For this reason, it is important to identify substrates for the polymorphic CYP enzymes, since genotype can significantly influence plasma concentrations, and lead to accumulation (Grasmader et al., 2004). In vitro systems are routinely used to investigate drug–drug interactions, and since the prime goal of such experiments is to predict likely DDIs in the clinic, systems utilising humanderived reagents are clearly the method of choice. By identifying the major enzymes responsible for the metabolism of a compound several conclusions may be drawn. 1. A compound which is a substrate for several enzymes will have less problems resulting from enzyme inhibition (alternative pathways being able to compensate for a reduction in metabolic capacity of others). 2. A number of enzymes are polymorphically expressed (i.e. different alleles exist) and consequently there are different populations of people expressing the enzyme at different levels, or in some cases not at all. CYP2D6 and CYP2C19 are members of the CYP family where poor metabolisers exist at a frequency of approximately 8 and 2–5% in Caucasions,
respectively. A substrate for a polymorphic enzyme will be associated with large pharmacokinetic variability between normal and poor metabolisers. It is therefore important to identify not only the routes of metabolism of a compound, but also the enzymes responsible. One must first make a determination of what types of metabolic clearance pathways comprise the major clearance pathways. For example, it is unproductive to engage in CYP reaction phenotyping if the major route of metabolism is conjugative metabolism or oxidative metabolism catalyzed by a non-CYP oxidase. The most appropriate human-derived reagents for such initial studies in as ‘complete’ an in vitro system as is possible, either primary hepatocytes or liver homogenates. Follow-up studies can then be performed in the most appropriate sub-cellular fractions, for example, liver microsomes for CYPs and UGTs, liver cytosol for alcohol dehydrogenases, xanithine oxidase, SULTs. 4.1. Assessment of enzyme inhibition and induction Potency of enzyme inhibition can be established by determination of IC50 values in appropriate human-derived reagents, again depending on the enzymes under investigation. Typically human liver microsomes will be used to investigate CYP inhibition, whilst hepatocytes or recombinant systems utilising expressed human enzymes may be used for other enzymes, such as MAO. In vitro data derived from such studies can become a very powerful tool when used in conjunction with in silico approaches to simulate the extent of pharmacokinetic drug interactions. This approach is exemplified by the SimCYP® project at the University of Sheffield. This software package is capable of simulating drug interactions in man using information from in vitro studies. SimCYP® contains databases on human physiological, genetic and epidemiological information, in addition to information on the in vitro metabolism of drugs that are currently used clinically. This information is then integrated with human in vitro data, generated by the user, to predict the pharmacokinetic behaviour of new compounds and the simulation and prediction of the likely magnitude of drug–drug interactions, giving population distributions (as opposed to single values). In vitro models also allow a measurement of the induction potential in human tissue, thus establishing some relevance of a new chemical’s induction ability and help make predictions to the clinical situation. Primary human hepatocytes in culture have been used most extensively to investigate enzyme induction. A number of molecules have been observed to induce CYP in cultured human hepatocytes and in vivo studies have confirmed induction, resulting in a pharmacokinetic drug interaction (Worboys and Carlile, 2001). Human hepatocytes do show donor-to-donor variability and in order to over come this a number of immortalised cell lines derived from human tissue have been developed. Although human hepatocytes are usually considered the “gold standard” for determining enzyme induction, reporter gene systems using transient transfection of the PXR response element are becoming more popular as a more rapid screen of induction potential in drug discovery (El-Sankary et al., 2001).
R. Hyland et al. / Environmental Toxicology and Pharmacology 21 (2006) 179–183
5. Conclusions It is clear from this review that in vitro assays utilising human reagents are now well established in the key drug discovery areas of absorption, metabolism and drug–drug interactions. With the development of robust in vitro methodologies, and assay validation using carefully chosen test sets (where clinical data is available) there is a greater understanding of the utility of these assays in predicting human pharmacokinetics. Human-based in vitro assays are therefore invaluable in drug discovery and development. Data generated in such assays will increasingly be used in combination with in silico approaches, becoming a powerful tool in meeting the higher throughput needs of the drug discovery process. References Artursson, P., Palm, K., Luthman, K., 1996. Caco-2 monolayers in experimental and theoretical predictions of drug transport. Adv. Drug Deliv. Rev. 22, 67–84. Ayrton, A., Morgan, P., 2001. Role of transport proteins in drug absorption, distribution, and excretion. Xenobiotica 31, 469–497. Chiou, W.L., Jeong, H.Y., Chung, S.M., Wu, T.C., 2000. Evaluation of using dog as an animal model to study the fraction of oral dose absorbed of 43 drugs in humans. Pharm. Res. (N. Y.) 17, 135. de Groot, M.J., Kirton, S.B., Sutcliffe, M.J., 2004. In Silico methods for predicting ligand binding determinants of cytochromes P450, Cytochrome P450s in medicinal chemistry and drug discovery. Curr. Topics Med. Chem. 4, 1803. de Groot, M.J., Ekins, S., 2002. Pharmacophore modeling of cytochrome P450. Adv. Drug Deliv. Rev. 54, 367. de Groot, M.J., Ackland, M.J., Horne, V.A., Alex, A.A., Jones, B.C., 1999a. Novel approach to predicting P450 mediated drug metabolism. Development of a combined protein and pharmacophore model for CYP2D6. J. Med. Chem. 42, 1515. de Groot, M.J., Ackland, M.J., Horne, V.A., Alex, A.A., Jones, B.C., 1999b. A novel approach to predicting P450 mediated drug metabolism. CYP2D6 catalysed N-dealkylation reactions and qualitative metabolite predictions using a combined protein and pharmacophore model for CYP2D6. J. Med. Chem. 42, 4062. Dickins, M., van de Waterbeemd, H., 2004. Simulation models for drug disposition and drug interactions, DDT. BioSilico 2, 38. Ekins, S., Obach, R.S., 2000. Three-dimensional quantitative structure activity relationship computational approaches for prediction of human in vitro intrinsic clearance. J. Pharmacol. Exp. Ther. 295, 463–473. El-Sankary, W., Gibson, G.G., Ayrton, A., Plant, N., 2001. Use of reporter gene assay to predict and rank the potency and efficiency of CYP3A4 inducers. Drug Metab. Dispos. 29, 1454. Glinsukon, T., Benjamin, T., Grantham, P.H., Weisburger, E.K., Roller, P.P., 1975. Enzymic N-acetylation of 2,4-toluenediamine by liver cytosols from various species. Xenobiotica 5, 475. Grasmader, K., Verwohlt, P.L., Rietschel, M., Dragicevic, A., Muller, M., Hiemke, C., Freymann, N., Zobel, A., Maier, W., Rao, M.L., 2004. Impact of polymorphism of cytochrome-P450 isoenzymes 2C9, 2C19 and 2D6 on plasma concentrations and clinical effects of antidepressants in a naturalistic clinical setting. Eur. J. Clin. Pharmacol. 60, 329.
183
He, Y., Murby, S., Warhurst, G., Gifford, L., Walker, D., Ayrton, J., Eastmond, R., Rowland, M., 1998. Species differences in size discrimination in the paracellular pathway reflected by oral bioavailability of polyethylene glycol and D peptides. J. Pharm. Sci. 87, 626. Honig, P.K., Wortham, D.C., Zamani, K., Conner, D.P., Mullin, J.C., Cantilene, L.R., 1993. Terfenadine-ketoconazole interaction. Pharmacokinetic and electrocardiographic consequences. JAMA 269, 1513. Houston, J.B., Carlile, D.J., 1997. Prediction of hepatic clearance from microsomes, hepatocytes, and liver slices. Drug Metab. Rev. 29, 891–922. Iwatsubo, T., Hirota, N., Ooie, T., Suzuki, H., Shimada, N., Chiba, K., Ishizaki, T., Green, C.E., Tyson, C.A., Sugiyama, Y., 1997. Prediction of in vivo drug metabolism in the human liver from in vitro metabolism data. Pharmacol. Ther. 73, 5147–5171. Li, A.P., Jurima-Romet, M., 1997. Overview; pharmacokinetic drug–drug interactions. Drug Metab. Dispos. 29, 141. Molero-Monfort, M., Escuder-Gilabert, L., Villanueva-Camanoas, Sagrado, S., Medina-Hernandez, M.J., 2001. Biopartitioning micellar chromatography: an in vitro technique for predicting human drug absorption. J. Chromatogr. B Biomed. Sci. Appl. 753 (2), 225–236. Obach, R.S., 2001. The prediction of human clearance from hepatic microsomal metabolism data. Curr. Opin. Drug Disc. Dev. 4, 36–44. Poirier, L.A., Miller, J.A., Miller, E.C., 1963. The N- and ring-hydroxylation of 2-acetylaminofluorene and the failure to detect N-acetylation of 2aminofluorene in the dog. Cancer Res. 23, 790. Polli, J.W., Wring, S.A., Humphreys, J.E., Huang, L., Morgan, J.B., Webster, L.O., Serabjit-Singh, C.S., 2001. Rational use of in vitro P-glycoprotein assays in drug discovery. J. Pharmacol. Exp. Ther. 299, 620–628. Schneider, G., Coassolo, P., Lave, T., 1999. Combining in vitro and in vivo pharmacokinetic data for prediction of hepatic drug clearance in humans by artificial neural networks and multivariate statistical techniques. J. Med. Chem. 42, 5072–5076. Shenfield, G.M., 1986. Drug interactions with oral contraceptive preparations. Med. J. Aust. 144, 205. Smith, D.A., Jones, B.C., Walker, D.K., 1996. Design of drugs involving the concepts and theories of drug metabolism and pharmacokinetics. Med. Res. Rev. 16 (3), 243. Trepanier, L.A., Winand, N.J., Spielberg, S.P., Cribb, A.E., 1997. Cytosolic arylamine N-acetyltransferase (NAT) deficiency in the dog and other canids due to an absence of NAT genes. Biochem. Pharmacol. 54, 73. van de Waterbeemd, H., Gifford, E., 2003. ADMET in silico modelling: towards prediction paradise? Nat. Rev. Drug Disc. 2, 192. Wester, M.R., Yano, J.K., Schoch, G.A., Yang, C., Griffin, K.J., Stout, C.D., Johnson, E.F., 2004. The structure of human cytochrome P450 2C9 complexed with flurbiprofen at 2.0-A resolution. J. Biol. Chem. 279, 35630. Williams, P.A., Cosme, J., Sridhar, V., Johnson, E.F., McRee, D.E., 2000. Microsomal cytochrome P450 2C5: comparison to microbial P450s and unique features. J. Inorg. Biochem. 81, 183. Williams, P.A., Cosme, J., Ward, A., Angrove, H.C., Vinkovic, D.M., Jhoti, H., 2003. Crystal structure of human cytochrome P450 2C9 with bouind warfarin. Nature 424, 464. Worboys, P.D., Carlile, D.J., 2001. Implications and consequences of enzyme induction on preclinical and clinical drug development. Xenobiotica 31, 539. Yano, J.K., Wester, M.R., Schoch, G.A., Griffin, K.J., Stout, C.D., Johnson, E.F., 2004. The structure of human microsomal cytochrome P450 3A4 determined by X-ray crystallography to 2.05-A resolution. J. Biol. Chem. 279, 38091.
Utility of human/human-derived reagents in drug discovery and development: An industrial perspective Ruth Hyland ∗ , Barry Jones, Han van de Waterbeemd Pfizer Global Research & Development, PDM, Sandwich, Kent CT13 9NJ, UK Available online 10 August 2005
Abstract The shift to combinatorial chemistry and parallel synthesis in drug discovery has resulted in large numbers of compounds entering the lead seeking and lead development phases of the process. To support this, higher throughput computational (in silico) and in vitro approaches have become the forefront of the drug metabolism and pharmacokinetic (DMPK) input into drug discovery. This has been accompanied by a shift in focus from animal-derived data to human based studies, reflecting the realisation that extrapolation from animals to human has its limitations. In silico approaches may be regarded as human derived tools for DMPK, since models (template/pharmacophore and protein homology modelling), for example, for the human CYP enzymes, are widely used for identifying qualitatively enzyme/substrate interactions. Quantitative assessment of drug metabolism using human hepatocytes or sub-cellular fractions provide a valuable tool both for the screening out of high metabolic lability and in estimations of human intrinsic clearance. In terms of drug absorption, the human colon adenocarcinoma cell line, Caco-2, offers a versatile human derived system for measuring drug permeability, despite over expression of the efflux transporter P-glycoprotein (P-gp). The importance of P-gp can then be further assessed in recombinant systems expressing the human P-gp, where substrate affinity and inhibition potency can be measured, important factors when considering transporter mediated drug–drug interactions. The primary cause of pharmacokinetic-based drug–drug interactions (DDIs) is through enzyme inhibition or induction, with the CYP enzymes being of major importance. Human liver microsomes and hepatocytes are invaluable tools in assessment of DDI vulnerability of new chemical entities, having the capacity to identify enzymes responsible for specific routes of metabolism, and hence areas of vulnerability for a DDI. In addition, human-based screening tools can be used to identify the perpetrator of a DDI through enzyme inhibition/induction. Large differences in the nature of enzymes induced and the extent of induction when comparing animals to man are known. Thus, in vitro models allowing assessment of induction potential in human tissue, establishes some relevance to the clinical situation. © 2005 Elsevier B.V. All rights reserved. Keywords: Human reagents; Drug discovery; In silico ADME; Drug–drug interactions (DDI)
1. Introduction During the last 5–10 years there has been a dramatic change in the drug discovery paradigm. Combinatorial chemistry and parallel synthesis approaches have led to an explosion in terms of the numbers of compounds synthesised, in turn resulting in an increase in the number of compounds entering lead seeking and lead development. When trying to identify a good orally active drug candidate it is important that it possess good absorption properties and metabolic stability in addition to potency and selectivity for the biological target (Smith et al., 1996). In order to support the numbers of compounds from chemistry higher throughput drug metabolism and pharmacokinetic
∗
Corresponding author. Tel.: +44 1304 641386; fax: +44 1304 651987. E-mail address: [email protected] (R. Hyland).
1382-6689/$ – see front matter © 2005 Elsevier B.V. All rights reserved. doi:10.1016/j.etap.2005.07.007
(DMPK) studies have become an essential tool in drug discovery. In order to achieve the rapid feedback for new chemical synthesis, computational and in vitro approaches have become the forefront of DMPK drug discovery. In the field of DMPK there has been a major shift from the use of animal-derived data to human-based studies, the exception being comparative studies for cross species comparisons in order to support animal-based regulatory toxicology studies. This shift is driven in part by the desire to reduce the number of animalbased studies and an increase in the availability of human-based reagents. However, the main impetus for using human-derived reagents has been the appreciation of distinct species differences between animals and humans in terms of pharmacokinetic properties. For example, in terms of drug metabolism the rat is well recognised as possessing good N-acetylation capabilities, whilst the dog has none, and human lies somewhere between the two (Glinsukon et al., 1975; Poirier et al., 1963; Trepanier et al.,
180
R. Hyland et al. / Environmental Toxicology and Pharmacology 21 (2006) 179–183
1997). In terms of drug absorption the architecture of the gastrointestinal tract of the dog makes it more likely for drugs to be absorbed through the paracellular route. The larger pore size and greater frequency of pores (He et al., 1998) can result in the dog grossly over predicting oral bioavailability of drugs absorbed through the paracellular route (Chiou et al., 2000). Hence extrapolation of pharmacokinetic data from animals to human has its limitations. This review will concentrate on the utility of humanderived reagents in the prediction of metabolism, absorption and drug–drug interactions. A summary of human-derived reagents used in drug discovery and development and their potential utility is provided in Table 1. 2. Metabolism 2.1. In silico approaches Although not strictly human reagents, in silico models may be regarded as human-derived tools for DMPK. For example, in silico models for the human CYP enzymes have been available for many years and are continually being refined with the emergence of new data and science. The approaches used include template/pharmacophore models and protein homology modelling (de Groot and Ekins, 2002). Template or pharmacophore modelling involves the correlation of structural elements between a series of compounds known to interact with the enzyme either as substrates or inhibitors, in order to identify key structural features involved in the interaction with the enzyme. This approach is exemplified by the production of a template model for CYP2D6 (de Groot et al., 1999a, 1999b). Protein homology modelling involves the generation
of a model by alignment of the primary amino acid sequence for the protein of interest, with that of closely related proteins, for which the secondary structure is known based upon X-ray crystallography, and assuming the same secondary structure. Until recently the only CYP enzymes for which crystal structures were available were those from bacteria and yeast, which have low sequence homology compared to human forms. In the last 5 years, significant efforts have gone into obtaining crystal structures for mammalian CYPs. In 2000, the solution of the first mammalian crystal structure (rabbit CYP2C5/3) was published (Williams et al., 2000) and more recently the first human CYP2C9 and CYP3A4 crystal structures (Williams et al., 2003; Wester et al., 2004; Yano et al., 2004). The first step in any homology model is the sequence alignment that, although carried out on computer, is operator dependent, and hence two modellers may have a different alignment for the same protein. A comprehensive review of in silico methods and the resulting models for the cytochromes P450 is provided by de Groot et al. (2004). Other areas where in silico models, based upon human derived data can be beneficial is in predicting properties that will provide key information on dose size and dose frequency, such as oral absorption, bioavailability, clearance, volume of distribution and drug–drug interactions (van de Waterbeemd and Gifford, 2003; Dickins and van de Waterbeemd, 2004). For example, in silico techniques have been used to predict intrinsic clearance values. A number of different statistical techniques have been applied to the same data set resulting in the production of computational models for the prediction of hepatocyte intrinsic clearance. An initial study showed that, as might be expected, human clearance could be predicted from hepatocyte data (Schneider et al., 1999). A subsequent study
Table 1 Human derived reagents used in drug discovery and development Human-derived reagents
What’s required
Utility
In silico Simulation Protein homology or pharmacophore model
Various in vitro data Human CYP crystal structures
Prediction of absorption, DDI, PK Substrate/Inhibitor docking
Molecular modelling Expert systems
In vivo metabolite schemes
Prediction of site of metabolism Prediction of human metabolites
QSAR
In vitro data for training model
Oral absorption Volume of distribution Clearance PPB Bioavailability CYP inhibition BBB uptake P-gp substrate
Microsomes/cytosols/S9 Hepatocytes
Clearance, metabolic routes, enzyme inhibition Clearance, metabolic routes, enzyme inhibition/induction Absorption
In vitro Human cell and sub-cellular fractions
Caco-2 Recombinant human proteins
rCYP450 microsomes (and other drug metabolising enzymes) rP-gp membrane Transfected cell-lines (eg MDCK, HEK)
Clearance, CYP inhibition P-gp affinity and inhibition Uptake and Efflux assays (hepatic/renal/BBB transport)
R. Hyland et al. / Environmental Toxicology and Pharmacology 21 (2006) 179–183
took this further by attempting to identify a three-dimensional QSAR for the prediction of intrinsic clearance using molecular features such as hydrogen bond acceptors and hydrogen bond donors (Ekins and Obach, 2000). 2.2. In vitro approaches It is possible to use human in vitro hepatic systems, principally whole cell systems (hepatocytes) or sub-cellular fractions (S9, microsomes or cytosol), to rank relative metabolic liability and to predict the in vivo clearance of a given compound. In the latter case, the principles behind the use of these systems are the same, in that first the rate of metabolism of the compound is determined experimentally using the in vitro system. This rate can be determined from following the disappearance of the drug molecule from the system or by measuring the rate of formation of one or more of the compound’s metabolites. The rate is then scaled using physiological scaling factors appropriate to the particular system to give an estimate of the intrinsic clearance of the compound (Clint ). In the case of the liver, the total clearance of the compound can be estimated from the intrinsic clearance by application of a hepatic extraction model such as the ‘wellstirred’ or ‘parallel tube’ models. Of these, the simplest and often the most widely used is the ’well-stirred’ model, which relates intrinsic clearance to total clearance via the following equation: Cl =
Clint fu Q Clint fu + Q
where fu is the free fraction of the compound in blood and Q is liver blood flow. These methods for estimating the clearance of a compound from in vitro systems have been extensively studied and reviewed (Obach, 2001; Houston and Carlile, 1997; Iwatsubo et al., 1997) In general, if the appropriate experimental conditions are used, both hepatocytes and microsomes will give reasonable predictions of total clearance. However, the different systems have advantages and disadvantages. For example, microsomes can only be reliably used for compounds where most of the clearance is mediated by the cytochrome P450 system, which represents a large majority of drugs. However, it can be argued that hepatocytes are of greater utility since all the major enzyme systems involved in drug metabolism are present in the cells. Prediction of human clearance by microsomes and hepatocytes can be carried out in a high throughput automated mode. However, to make this process even faster in silico techniques have also been employed, as discussed earlier. 3. Absorption 3.1. Cell permeability The permeation rate of compounds through monolayers of the human colon adenocarcinoma cell line, Caco-2, is one of the most widely used systems for predicting absorption (Artursson et al., 1996). In this system, the cells are grown on a permeability
181
filter support and the passage of compounds is then measured across the monolayer from a donor to an acceptor compartment and gives an assessment of permeability. Correlations have been observed between permeability across Caco-2 monolayers and the extent of absorption in man (Molero-Monfort et al., 2001). For example, for a series of six non-peptidic compounds exhibiting between 5 and 100% absorption in man, permeability measured in the Caco-2 experiments ranged from 0.5 to >50 × 10−4 cm/min and all of the compounds that showed permeability >20 × 10−4 cm/min were completely absorbed. One complicating factor related to this system, which may limit its predictive ability in terms of absorption, is the apparent over expression of P-glycoprotein (P-gp). Caco-2 cell monolayers are polarised with apical and basal surfaces to the cell monolayers. The P-gp efflux pump is expressed on the apical side of the monolayer and has been shown to mediate the efflux of compounds such as cyclosporin A. Hence, substrates for this transport protein will be prevented from crossing the monolayer due to efflux back into the donor compartment. P-gp levels vary between Caco-2 cell preparations, introducing inter-experimental variability, and the transporter may also be saturated by some substrates. These variables clearly complicate extrapolation from the in vitro to in vivo situation. However, Caco-2 cells do still offer a versatile human derived system for measuring drug permeability. Although they require considerable time, expense and effort to establish and maintain the cell line, they do provide an extremely simple system to measure permeability with straightforward analysis of drug in the donor and acceptor buffer solutions. This system is also ideally suited for the measurement of passive, transcellular permeation, which is the route of absorption for the majority of drugs. However, for compounds where paracellular absorption may be involved, the absence of aqueous pores limits the applicability of the Caco-2 cell model. The presence of P-gp in Caco-2 cells raises the issue that the permeability of a compound in this system may be a function of two structure activity relationships (SARs): one for intrinsic permeability and one for interaction with, and efflux by, P-gp. These SARs can be deconvoluted by the use of assays that are designed to investigate these processes directly. 3.2. P-gp: affinity/binding To assess the extent of P-gp transport, binding and/or inhibition, a number of approaches are currently in use, most of which rely upon recombinant expression systems for the human P-gp protein. Indirect methods can be used, based on measurement of the effect of membrane transport using a P-gp binding assay, an ATPase assay, or a calcein AM test. The ATPase assays rely on the measurement of inorganic phosphate which is released when ATP is converted to ADP and so providing the energy for the transport of a drug molecule. Calcein AM is a non-fluorescent P-gp substrate, which is transformed into Calcein, a fluorescent non-substrate. Hence in the absence of a P-gp inhibitor cells will be non-fluorescent, but will show an increasing amount of fluorescence in the presence of increasing concentrations of a P-gp inhibitor.
182
R. Hyland et al. / Environmental Toxicology and Pharmacology 21 (2006) 179–183
A comparison has been made between monolayer efflux, ATPase and calcein-AM assays (Polli et al., 2001). Some of the compounds showed concordance across all the assays, while others revealed differences related to their apparent permeability. Transporters mediated drug–drug interactions (DDI) are becoming an important research topic and the above assays can help to quantify the potential for such interactions. Similar screens could be developed for other transporters. It should be possible to separate passive diffusion from efflux transport processes through a rational combination of these screens, and this would allow the study of structure-activity relationships for each of these processes separately. Much more still needs to be done to fully understand and predict the clinical relevance of P-gp binding and inhibition (Ayrton and Morgan, 2001). 4. Drug–drug interactions (DDI) The primary cause of a pharmacokinetic DDI is as a result of a modulation in enzyme activity, namely enzyme inhibition or enzyme induction (Li and Jurima-Romet, 1997). A compound may be associated with DDIs as either a perpetrator or victim. That is, the compound may be an enzyme inhibitor or inducer itself (perpetrator) thereby altering the pharmacokinetics of co-administered drugs. For example, the CYP3A4 inhibitor ketoconazole, is well documented as a perpetrator, is contra-indicated with many CYP3A4 substrates and can cause life-threatening drug–drug interactions (Honig et al., 1993). Enzyme induction will result in the more rapid clearance of substrates and thus whilst not having toxic consequence will result in sub-therapeutic drug concentrations. This is of particular importance for DDIs associated with the oral contraceptive, where a lack of contraceptive cover and thus unwanted pregnancies may arise (Shenfield, 1986). As a victim, the compound may be metabolised by enzymes whose activities are being modulated by a co-administered drug or through genetic factors. For this reason, it is important to identify substrates for the polymorphic CYP enzymes, since genotype can significantly influence plasma concentrations, and lead to accumulation (Grasmader et al., 2004). In vitro systems are routinely used to investigate drug–drug interactions, and since the prime goal of such experiments is to predict likely DDIs in the clinic, systems utilising humanderived reagents are clearly the method of choice. By identifying the major enzymes responsible for the metabolism of a compound several conclusions may be drawn. 1. A compound which is a substrate for several enzymes will have less problems resulting from enzyme inhibition (alternative pathways being able to compensate for a reduction in metabolic capacity of others). 2. A number of enzymes are polymorphically expressed (i.e. different alleles exist) and consequently there are different populations of people expressing the enzyme at different levels, or in some cases not at all. CYP2D6 and CYP2C19 are members of the CYP family where poor metabolisers exist at a frequency of approximately 8 and 2–5% in Caucasions,
respectively. A substrate for a polymorphic enzyme will be associated with large pharmacokinetic variability between normal and poor metabolisers. It is therefore important to identify not only the routes of metabolism of a compound, but also the enzymes responsible. One must first make a determination of what types of metabolic clearance pathways comprise the major clearance pathways. For example, it is unproductive to engage in CYP reaction phenotyping if the major route of metabolism is conjugative metabolism or oxidative metabolism catalyzed by a non-CYP oxidase. The most appropriate human-derived reagents for such initial studies in as ‘complete’ an in vitro system as is possible, either primary hepatocytes or liver homogenates. Follow-up studies can then be performed in the most appropriate sub-cellular fractions, for example, liver microsomes for CYPs and UGTs, liver cytosol for alcohol dehydrogenases, xanithine oxidase, SULTs. 4.1. Assessment of enzyme inhibition and induction Potency of enzyme inhibition can be established by determination of IC50 values in appropriate human-derived reagents, again depending on the enzymes under investigation. Typically human liver microsomes will be used to investigate CYP inhibition, whilst hepatocytes or recombinant systems utilising expressed human enzymes may be used for other enzymes, such as MAO. In vitro data derived from such studies can become a very powerful tool when used in conjunction with in silico approaches to simulate the extent of pharmacokinetic drug interactions. This approach is exemplified by the SimCYP® project at the University of Sheffield. This software package is capable of simulating drug interactions in man using information from in vitro studies. SimCYP® contains databases on human physiological, genetic and epidemiological information, in addition to information on the in vitro metabolism of drugs that are currently used clinically. This information is then integrated with human in vitro data, generated by the user, to predict the pharmacokinetic behaviour of new compounds and the simulation and prediction of the likely magnitude of drug–drug interactions, giving population distributions (as opposed to single values). In vitro models also allow a measurement of the induction potential in human tissue, thus establishing some relevance of a new chemical’s induction ability and help make predictions to the clinical situation. Primary human hepatocytes in culture have been used most extensively to investigate enzyme induction. A number of molecules have been observed to induce CYP in cultured human hepatocytes and in vivo studies have confirmed induction, resulting in a pharmacokinetic drug interaction (Worboys and Carlile, 2001). Human hepatocytes do show donor-to-donor variability and in order to over come this a number of immortalised cell lines derived from human tissue have been developed. Although human hepatocytes are usually considered the “gold standard” for determining enzyme induction, reporter gene systems using transient transfection of the PXR response element are becoming more popular as a more rapid screen of induction potential in drug discovery (El-Sankary et al., 2001).
R. Hyland et al. / Environmental Toxicology and Pharmacology 21 (2006) 179–183
5. Conclusions It is clear from this review that in vitro assays utilising human reagents are now well established in the key drug discovery areas of absorption, metabolism and drug–drug interactions. With the development of robust in vitro methodologies, and assay validation using carefully chosen test sets (where clinical data is available) there is a greater understanding of the utility of these assays in predicting human pharmacokinetics. Human-based in vitro assays are therefore invaluable in drug discovery and development. Data generated in such assays will increasingly be used in combination with in silico approaches, becoming a powerful tool in meeting the higher throughput needs of the drug discovery process. References Artursson, P., Palm, K., Luthman, K., 1996. Caco-2 monolayers in experimental and theoretical predictions of drug transport. Adv. Drug Deliv. Rev. 22, 67–84. Ayrton, A., Morgan, P., 2001. Role of transport proteins in drug absorption, distribution, and excretion. Xenobiotica 31, 469–497. Chiou, W.L., Jeong, H.Y., Chung, S.M., Wu, T.C., 2000. Evaluation of using dog as an animal model to study the fraction of oral dose absorbed of 43 drugs in humans. Pharm. Res. (N. Y.) 17, 135. de Groot, M.J., Kirton, S.B., Sutcliffe, M.J., 2004. In Silico methods for predicting ligand binding determinants of cytochromes P450, Cytochrome P450s in medicinal chemistry and drug discovery. Curr. Topics Med. Chem. 4, 1803. de Groot, M.J., Ekins, S., 2002. Pharmacophore modeling of cytochrome P450. Adv. Drug Deliv. Rev. 54, 367. de Groot, M.J., Ackland, M.J., Horne, V.A., Alex, A.A., Jones, B.C., 1999a. Novel approach to predicting P450 mediated drug metabolism. Development of a combined protein and pharmacophore model for CYP2D6. J. Med. Chem. 42, 1515. de Groot, M.J., Ackland, M.J., Horne, V.A., Alex, A.A., Jones, B.C., 1999b. A novel approach to predicting P450 mediated drug metabolism. CYP2D6 catalysed N-dealkylation reactions and qualitative metabolite predictions using a combined protein and pharmacophore model for CYP2D6. J. Med. Chem. 42, 4062. Dickins, M., van de Waterbeemd, H., 2004. Simulation models for drug disposition and drug interactions, DDT. BioSilico 2, 38. Ekins, S., Obach, R.S., 2000. Three-dimensional quantitative structure activity relationship computational approaches for prediction of human in vitro intrinsic clearance. J. Pharmacol. Exp. Ther. 295, 463–473. El-Sankary, W., Gibson, G.G., Ayrton, A., Plant, N., 2001. Use of reporter gene assay to predict and rank the potency and efficiency of CYP3A4 inducers. Drug Metab. Dispos. 29, 1454. Glinsukon, T., Benjamin, T., Grantham, P.H., Weisburger, E.K., Roller, P.P., 1975. Enzymic N-acetylation of 2,4-toluenediamine by liver cytosols from various species. Xenobiotica 5, 475. Grasmader, K., Verwohlt, P.L., Rietschel, M., Dragicevic, A., Muller, M., Hiemke, C., Freymann, N., Zobel, A., Maier, W., Rao, M.L., 2004. Impact of polymorphism of cytochrome-P450 isoenzymes 2C9, 2C19 and 2D6 on plasma concentrations and clinical effects of antidepressants in a naturalistic clinical setting. Eur. J. Clin. Pharmacol. 60, 329.
183
He, Y., Murby, S., Warhurst, G., Gifford, L., Walker, D., Ayrton, J., Eastmond, R., Rowland, M., 1998. Species differences in size discrimination in the paracellular pathway reflected by oral bioavailability of polyethylene glycol and D peptides. J. Pharm. Sci. 87, 626. Honig, P.K., Wortham, D.C., Zamani, K., Conner, D.P., Mullin, J.C., Cantilene, L.R., 1993. Terfenadine-ketoconazole interaction. Pharmacokinetic and electrocardiographic consequences. JAMA 269, 1513. Houston, J.B., Carlile, D.J., 1997. Prediction of hepatic clearance from microsomes, hepatocytes, and liver slices. Drug Metab. Rev. 29, 891–922. Iwatsubo, T., Hirota, N., Ooie, T., Suzuki, H., Shimada, N., Chiba, K., Ishizaki, T., Green, C.E., Tyson, C.A., Sugiyama, Y., 1997. Prediction of in vivo drug metabolism in the human liver from in vitro metabolism data. Pharmacol. Ther. 73, 5147–5171. Li, A.P., Jurima-Romet, M., 1997. Overview; pharmacokinetic drug–drug interactions. Drug Metab. Dispos. 29, 141. Molero-Monfort, M., Escuder-Gilabert, L., Villanueva-Camanoas, Sagrado, S., Medina-Hernandez, M.J., 2001. Biopartitioning micellar chromatography: an in vitro technique for predicting human drug absorption. J. Chromatogr. B Biomed. Sci. Appl. 753 (2), 225–236. Obach, R.S., 2001. The prediction of human clearance from hepatic microsomal metabolism data. Curr. Opin. Drug Disc. Dev. 4, 36–44. Poirier, L.A., Miller, J.A., Miller, E.C., 1963. The N- and ring-hydroxylation of 2-acetylaminofluorene and the failure to detect N-acetylation of 2aminofluorene in the dog. Cancer Res. 23, 790. Polli, J.W., Wring, S.A., Humphreys, J.E., Huang, L., Morgan, J.B., Webster, L.O., Serabjit-Singh, C.S., 2001. Rational use of in vitro P-glycoprotein assays in drug discovery. J. Pharmacol. Exp. Ther. 299, 620–628. Schneider, G., Coassolo, P., Lave, T., 1999. Combining in vitro and in vivo pharmacokinetic data for prediction of hepatic drug clearance in humans by artificial neural networks and multivariate statistical techniques. J. Med. Chem. 42, 5072–5076. Shenfield, G.M., 1986. Drug interactions with oral contraceptive preparations. Med. J. Aust. 144, 205. Smith, D.A., Jones, B.C., Walker, D.K., 1996. Design of drugs involving the concepts and theories of drug metabolism and pharmacokinetics. Med. Res. Rev. 16 (3), 243. Trepanier, L.A., Winand, N.J., Spielberg, S.P., Cribb, A.E., 1997. Cytosolic arylamine N-acetyltransferase (NAT) deficiency in the dog and other canids due to an absence of NAT genes. Biochem. Pharmacol. 54, 73. van de Waterbeemd, H., Gifford, E., 2003. ADMET in silico modelling: towards prediction paradise? Nat. Rev. Drug Disc. 2, 192. Wester, M.R., Yano, J.K., Schoch, G.A., Yang, C., Griffin, K.J., Stout, C.D., Johnson, E.F., 2004. The structure of human cytochrome P450 2C9 complexed with flurbiprofen at 2.0-A resolution. J. Biol. Chem. 279, 35630. Williams, P.A., Cosme, J., Sridhar, V., Johnson, E.F., McRee, D.E., 2000. Microsomal cytochrome P450 2C5: comparison to microbial P450s and unique features. J. Inorg. Biochem. 81, 183. Williams, P.A., Cosme, J., Ward, A., Angrove, H.C., Vinkovic, D.M., Jhoti, H., 2003. Crystal structure of human cytochrome P450 2C9 with bouind warfarin. Nature 424, 464. Worboys, P.D., Carlile, D.J., 2001. Implications and consequences of enzyme induction on preclinical and clinical drug development. Xenobiotica 31, 539. Yano, J.K., Wester, M.R., Schoch, G.A., Griffin, K.J., Stout, C.D., Johnson, E.F., 2004. The structure of human microsomal cytochrome P450 3A4 determined by X-ray crystallography to 2.05-A resolution. J. Biol. Chem. 279, 38091.