The pharmacological response to a drug is primarily the result of its interaction with the target and the subsequent signal transduction inside cells (pharmacodynamics). However, the absorption, distribution, metabolism and excretion of the compound (pharmacokinetics) also play a determinant role in its efficacy by governing the concentration of the drug at the site of action. Inappropriate pharmacokinetics (e.g. low or inconsistent bioavailability due to different elimination rates or metabolism) result in an inadequate or variable clinical response to the drug, which frequently compromises its therapeutic usage. Drug metabolism is one of the major determinants of drug clearance and the factor that is most frequently responsible for interindividual differences in drug pharmacokinetics.[1] Hence, knowledge of major routes involved in metabolite formation, the enzymes involved and the potential enzyme-inhibiting or enzyme-inducing properties of the drug become key issues in selecting drug candidates. The process of selecting drug candidates is becoming much more rational, as studies on metabolism and kinetics of drug candidates are implemented earlier.

This article reviews the principles of the methods that are used to study and predict drug metabolism, with a focus on in vitro techniques.

1. The Cytochrome P450 (CYP) Superfamily

Most xenobiotics undergo chemical modifications in the body before they can be effectively eliminated. Biotransformation is the process by which lipophilic drugs are rendered more hydrophilic and, hence, more easily excreted. This is achieved by a series of reactions classified into phase I and phase II. Phase I metabolic reactions are mainly oxidative processes that generate metabolites more polar than the parent compound. However, this can also lead to the formation of more reactive (and possibly toxic) metabolites. Phase II comprises different types of reactions, including the conjugation of the parent molecule (or its phase I metabolites) with endogenous molecules such as glucuronic acid, glutathione and sulfate. The conjugates tend to be much more water-soluble and usually much less active (less toxic) than the nonconjugated compounds.

CYP enzymes, together with flavin monooxygenases, are major players in the oxidative metabolism of a wide range of structurally diverse xenobiotics (drugs, chemicals) and endobiotics (steroids, fatty acids, prostaglandins).[2] CYP enzymes are bound to membranes of the endoplasmic reticulum and are associated with other cytochromes (b5) as well as with cytochrome P450 reductase. CYP is predominantly expressed in the liver, although it is also present in extrahepatic tissues, such as the lung, kidney, gastrointestinal tract and placenta.[3–5]

The first attempts to classify CYP enzymes were based on the type of reaction catalysed. This was replaced in the late 1980s by criteria based on sequence homology of CYP genes,[6] which has allowed a meaningful classification of human and animal CYPs. The CYP gene superfamily is subdivided into a number of families and subfamilies according to the degree of nucleotide sequence homology (table I). In humans, 16 gene families and 29 subfamilies have been identified to date. Families 1, 2 and 3 are largely involved in the biotransformation of xenobiotics and account for most of the metabolism of pharmaceuticals.[2] CYP3A4 is the most abundantly expressed isoform and represents approximately 30–40% of the total CYP protein in human adult liver.[7] There is no specific chromosomal location of CYP genes (table I). The genes of CYP families 1 to 3 are found on chromosomes 2, 7, 10, 15, 19 and 22. Although they are largely present in liver tissue, some isoforms are typically extrahepatic (i.e. CYP 1A1 and CYP1B1).

Table I
figure Tab1

Major human drug-metabolising cytochrome P450s[8,9]

Full size image

Highly conserved regions are found in CYPs from different animals. Based on that, it has been possible to classify the genes and to establish equivalence across species (table II). The CYP1A family is present in all animals, CYP1A2 being the predominant isoform. This is not a common pattern. The predominant isoform within the same CYP family is usually different among animal species (for example, CYP3A4 in humans, CYP3A1 in the rat and CYP3A6 in the rabbit; CYP2C9 in humans, CYP2C11 in the rat and several minor isoforms in the dog, rabbit, pig and mouse). Many animals express CYP genes with no known equivalent in other animals or humans.[10] This is most evident in the CYP2 family (table II).

Table II
figure Tab2

Cytochrome P450 isoenzymes in humans and other species.[8,10] Bold type indicates the predominant forms

Full size image

The interindividual differences in drug metabolism are largely the consequence of the variability of CYP gene expression. CYP genes are regulated by genetic as well as by nongenetic factors (such as physiological/pathological states, intake of drugs, exposure to chemicals or dietary habits).[20–23] A number of human CYP genes are known to be polymorphically expressed (table I). These gene variants are Mendelian inherited and show differences in catalytic activities towards xenobiotics (different Michaelis constant [Km] and/or maximum rate [Vmax]). Polymorphism can result in great differences in pharmacokinetics and pharmacodynamics of a drug when it is largely metabolised by that CYP and/or metabolism is required for its pharmacological action.[13] Indeed, this variability in plasma concentration can result in either an exaggerated or a lack of pharmacological response as well as increased toxicity. The metabolism of a compound by a polymorphic enzyme tends to be interpreted as a potential drawback for a new drug. However, this should not necessarily stop its development; the final decision should be taken on the therapeutic margin of the drug and the contribution of the polymorphic enzyme to total metabolism.

A characteristic of many CYP isoenzymes is their inducibility by xenobiotics and, in particular, by drugs (table I). The phenomenon of induction also contributes significantly to the interindividual variation in CYP phenotype and hence the pharmacokinetics, efficacy and potential toxicity of a given drug.[24,25] Most human drug-metabolising CYPs are either polymorphic or inducible, which accounts for the large interindividual variability found in humans.

2. Tools to Investigate the Expression of Human CYPs

In the past two decades, great progress has been made in the characterisation of human CYPs. Thanks to the use of purified enzymes and the identification of specific substrates and inhibitors, considerable knowledge on the specificity of CYP enzymes has been gained. Thus, characterised microsomes from human livers became a suitable tool to anticipate human metabolism and drug-drug interactions. Cultured hepatocytes represent a more complex, yet more predictable, model of metabolism in vivo and are widely used. Human hepatocytes from liver transplant programmes and surgical waste are making the prediction of human metabolism based on in vitro techniques more predictable.[26]

The application of recombinant DNA technology to CYP genes (i.e. use of cell lines expressing individual CYP enzymes) has considerably simplified the identification of the human CYPs responsible for the metabolism of a given xenobiotic. As a result of this, it is now possible to gain information on the human metabolism of drug candidates and drug interactions before clinical trials start, resulting in more cost-effective and ethically acceptable studies.[27]

Because of the clinical implications of the high variability of CYP isoforms in humans, considerable interest has grown in developing reliable methods for CYP ‘phenotyping’. The characterisation of the CYP system can be assessed in terms of activity (using specific substrates/inhibitors), protein (antibody analysis) and mRNA (gel blotting or polymerase chain reaction [PCR] techniques). However, only some of these techniques can be applied both in vitro and in vivo.

2.1 Substrates and Inhibitors

CYP isoenzymes exhibit distinct but overlapping substrate specificities. Consequently, xenobiotics tend to be metabolised by several isoenzymes, and only a few compounds are exclusively metabolised by one enzyme. In past recent years, precise information on specific substrates (table III) and inhibitors (table IV) for each of the major human CYPs has been reported in the scientific literature. Most of these compounds are readily available from commercial sources and their use has become routine in the characterisation of human CYP activities. Although all substrates reported in table III are suitable for assays in microsomes, only some of them can also be used in assays with intact cells. Substrates to be used in intact cell assays should be noncytotoxic and easily transported into cells.

Table III
figure Tab3

Substrates for in vitro evaluation of major human hepatic cytochrome P450 isoenzymes

Full size image
Table IV
figure Tab4

Selective inhibitors for human cytochrome P450 isoenzymes

Full size image

In contrast to this, the number of well characterised CYP inhibitors is much lower. Compounds are known to act as inhibitors of several CYPs, but only few of these chemicals have been properly identified as specific inhibitors of a single CYP isoform (table IV). For an inhibitor to be of practical use, it must be highly selective (i.e. the concentration inhibiting 50% of the activity of a CYP isoenzyme [IC50] must be 100-fold lower than for the other isoenzymes). Reversible/competitive inhibition is one of the possible mechanisms involved. In such cases, inhibitors tend to bind effectively to the catalytic site of the enzyme (low Km/Ki) but are poorly metabolised, thus slowing down the activity of the enzyme. Irreversible or quasi-irreversible inhibitions are linked to the formation of intermediate reactive metabolites, during or subsequent to the oxidation step in CYP catalysis, that irreversibly bind to the enzyme and block its catalytic activity.[77]

Few CYP-specific substrates listed in table III can be used in vivo in humans. These in vivo probes have undergone previous validation and appropriate pharmacokinetic analysis (table V). The use of these compounds is the procedure of choice for the metabolic phenotyping of human beings, as well as to assess in vivo the inhibition or induction of specific CYP enzymes by another drug.[46,78]

Table V
figure Tab5

Substrates useful for in vivo estimation of specific cytochrome P450s

Full size image

Substrates that are metabolised by different CYP enzymes and produce isoenzyme-specific products are also useful for the simultaneous quantification of several CYP isoforms. Testosterone, (R)-warfarin and progesterone are regio- and stereo-selectively hydroxylated by individual human/animal CYPs, and provide a rapid method for assessing the activity of several CYP isoenzymes.[62,93–95]

The most recent strategy for multi-CYP activity assessment is the use of a mixture of several compounds, each one being specifically metabolised by a CYP isoform.[46,96,97] This ‘cocktail’ strategy (table VI) offers advantages for in vitro and in vivo hepatocyte phenotyping, since several CYP enzymes can be simultaneously assessed in the same experiment and in the same individual, thus reducing time, size of the sample and intrasubject variability. Sample analysis requires highly sensitive analytical equipment; the use of high performance liquid chromatography (HPLC) linked to tandem mass spectrometry (MS/MS) is the most suitable procedure.

Table VI
figure Tab6

In vivo and in vitro ‘substrate cocktails’ for characterisation of human cytochrome P450 isoenzymes

Full size image

2.2 Antibodies

Polyclonal or monoclonal antibodies raised against CYP isoforms are of great value for identification and semiquantitative measurement of CYP protein content.[98] Antibodies can be easily generated by immunisation with the pure protein isolated from liver or from cDNA-directed expression systems. The C-terminal 15–20 amino acids of most CYPs are, however, sufficiently antigenic to yield an effective and specific antibody response in animals. Although antibodies frequently show cross-reactivity to closely related isoforms, they can be used in Western blot analysis to identify individual isoforms. The commercial availability of specific antibodies has significantly increased. Several suppliers offer antibodies against human and animal CYPs (more information can be found at the following sites: http://antibodyresource.com; http://abcam.com; http://biotrend.com; http://www.gentest.com).

Inhibiting antibodies can be used for unequivocal identification of the CYPs involved in the metabolism of a particular compound.[99] The use of such antibodies in combination with liver micro-somes/reconstituted enzymes is of great value to demonstrate the quantitative role of a particular CYP in the metabolic pathway of a drug, in particular when several CYP enzymes are involved.[99–102] However, for most commercially available antibodies, binding of the antibody to the CYP isoform does not cause effective inhibition of its catalytic activity. Inhibiting antibodies are unfortunately not easy to obtain from commercial sources.

2.3 Qualitative and Quantitative Molecular Biological Measurements

The sensitivity of enzyme activity measurements with specific substrates and/or antibodies makes the characterisation of CYP expression in small biological samples difficult. Both catalytic and protein assays require a significant amount of sample, which in the case of human in vitro studies is a serious drawback, as the size of the liver sample/number of hepatocytes available is limited.

These problems can be overcome by the use of molecular biology-based amplification techniques. These techniques allow rapid, specific and sensitive detection of small amounts of mRNA/DNA which are hardly detectable by other analytical methods.

Analysis of DNA is the easiest and most straightforward method in identifying CYP polymorphisms.[12,13,103] These techniques are based on the isolation of genomic DNA, PCR amplification of the fragment containing the polymorphic sequence using appropriate primers, followed by restriction analysis of the amplified fragment.[103] Alternatively, the amplified fragment can be used as a template for a second PCR amplification using another set of primers, which will reveal the existence of a polymorphism (nested PCR), or by the use of primers with associated fluorochromes.[12,104]

Precise quantification of CYP mRNA is more troublesome. It involves, first, the extraction of good quality mRNAs from the samples, followed by high efficiency reverse transcription (RT) to produce cDNAs and, finally, a PCR to amplify a specific CYP cDNA. The quantification of the amplified cDNA can be performed by densitometry of an ethidium bromide stained gel of the PCR amplified sample,[105,106] or by fluorimetry using appropriate dyes (e.g. picogreen).[107,108] Although PCR-based techniques are well suited for qualitative measurements of nucleic acids, quantitative and reproducible determination of a specific mRNA requires a number of internal and external standards to minimise the deviations inherent to exponential amplification. In a recent paper, Rodríguez-Antona et al.[109] described a PCR-based method which makes use of these principles and allows simultaneous determination of the ten most important human CYPs (figure 1). The greatest advantage of this method is that it allows quantification of CYP mRNAs in very small samples (e.g. liver needle biopsies or human cell microcultures) or in cells having much lower CYP mRNA content than liver cells.

Fig. 1
figure 1

Measurement of cytochrome P450 (CYP) mRNA levels in human liver and hepatocytes by reverse transcriptase-polymerase chain reaction (RT-PCR). Total RNA was extracted from liver biopsies (a) or hepatocytes cultured for 24 hours (b), reverse-transcribed and amplified by PCR. Using internal (luciferase mRNA) and external (cDNA inserts) standards, the number of specific mRNA molecules was calculated.

Full size image

CYP mRNA levels may not always reflect the concentrations of the catalytically active enzyme present in cells.[106] Hence, the use of mRNA quantification assays for phenotyping CYP expression in human liver is not straightforward. Recent studies from our laboratory showed good correlation between CYP mRNA and activity for CYP1A1, 1A2, 3A4, 2D6 and 2B6 (figure 2), whereas no evident correlation was found for CYP2C9, 2A6 and 2E1.[110] Factors that could contribute to this lack of correlation are different translation rates, mRNA stability and enzyme stability. Changes in CYP mRNA levels are indicative of variation of expressed enzyme for those genes mainly regulated at the transcriptional level.

Fig. 2
figure 2

Relationship between cytochrome P450 (CYP) mRNAs and metabolic activities in human liver. The mRNA content of 12 liver samples was determined by quantitative reverse transcriptase-polymerase chain reaction (RT-PCR). The enzyme activities corresponding to each CYP were determined in microsomes obtained from the same livers using specific substrates (reproduced from Rodríguez-Antona et al.,[110] with permission).

Full size image

3. Assessment of the Role of CYPs in the Metabolism of a Drug

To speed up the identification of new drug candidates, the pharmaceutical industry increasingly makes use of different in vitro systems to investigate drug metabolism (table VII). Several issues need to be addressed:

  • the comparative metabolic profile of a drug (i.e. identification of stable and/or reactive metabolites);

  • knowledge of major metabolic routes involved in metabolite formation and the human enzymes involved;

  • potential enzyme-inhibiting or enzyme-inducing properties of the drug.

Table VII
figure Tab7

Scope of use of hepatic in vitro models for drug metabolism studies

Full size image

During preclinical development of a new drug, animal testing is done assuming the animal species selected will be a good predictive model for humans. Inherent in this assessment is that the metabolism of the drug in the animal model will be similar to that in humans. However, metabolic differences exist between humans and animals, not only because of the different CYPs expressed in humans and other species (table II) but also because of their relative abundance.[14,111,112] Thus, comparative studies using liver microsomes/cells from animals and humans are very useful for demonstrating species differences in the metabolism of a given drug candidate,[113–117] and are of great value in the judicious and justifiable selection of animals for later pharmacokinetic and toxicological studies. As it is not ethically acceptable to use humans in early drug development, the need for animal studies will continue. Hence, the choice of the model needs to be rationally justified so that the data obtained in animal studies can be safely extrapolated to humans. To this aim, in vitro comparisons between the metabolic profile in human and animal hepatic models is an appropriate strategy.

Due to the importance that CYP phenotype has for the pharmacokinetics, pharmacodynamics and potential toxicity of a given drug, the identification of CYP isoforms involved in its metabolism is also of relevance.[27] CYP-engineered cells (or microsomes from CYP-engineered cells; ‘supersomes’) have made the identification of the CYPs involved in the metabolism of a drug candidate straightforward and easier.

Once the major human CYP isoforms involved in the metabolism of a given drug have been identified, it is of relevance to know whether the drug can influence their expression and/or catalytic activity. Such phenomena are at the root of potential drug-drug interactions.

3.1 Metabolic Profile of a New Drug

There are two basic strategies for in vitro investigation of the metabolic profile of a drug:

  • incubation with subcellular fractions, i.e. microsomes;

  • incubation with differentiated cellular models such as primary cultures, encapsulated hepatocytes or tissue slices.

Microsomes are often the first hepatic model used in metabolism studies in the course of development of a drug. The metabolic stability of a new chemical entity and its metabolic profile can be easily investigated, in a first approach, by incubating the drug with hepatic microsomes followed by chromatographic analysis (e.g. HPLC-MS/MS) of the incubation mixtures.[113,115,118–120] These techniques have considerably simplified and speeded up the identification of metabolites as well the identification of the CYP involved in their production.

Microsomes are easily prepared from liver tissue by homogenisation and centrifugation, and can be stored at −80°C for years with little or no loss of CYP enzyme activities.[121,122] Microsomes from different animal species are commercially available and are well standardised. They are simple to use and allow an easy comparison of the metabolic profile of a drug across animal species (figure 3 a). The major limitation of microsomes is that they have very low phase II activities. Another drawback is that incubation can be performed only for a short time (usually less than 1 hour), and poorly metabolised drugs, as well as secondary metabolism, are hardly detected. As a consequence of this, it may happen that the results obtained in vitro are markedly different from those obtained in vivo.

Fig. 3
figure 3

Testosterone metabolism in different animal species. Relative abundance of the major hydroxylated metabolites (6β, 17, 2β, 15β, 16β, 16α, 2α and 7α) of testosterone in liver microsomes (a) or primary cultured hepatocytes (b) from different animal species.

Full size image

Liver slices are relatively simple to use and do not require complex equipment/installations. Their major drawback is the survival of cells within the sliced tissue. Several authors claim that tissue slices can metabolise drugs for about 24 hours, which is considerably longer than by microsome incubations.[31,123–125]

Hepatocytes have some advantages that make them the closest model for in vivo studies.[58,114,117,125,126] As intact cells are used, plasma membrane, metabolic pathways, levels of physiological cofactors and coenzymes and active gene expression are reasonably well maintained for several hours/days in culture. A major drawback is that fully differentiated cells (i.e. primary cultures) need to be used. Because of the inability of differentiated hepatocytes to grow efficiently in vitro, cell cultures need to be prepared each time from liver tissue. This has greatly hindered the widespread use of human hepatocytes because of the restricted accessibility to suitable liver samples. Satisfactory cryopreservation of adult hepatocytes has been achieved and frozen cells are now sold entrapped in polysaccharide matrixes (e.g. alginate).[127,128] The viability of the thawed cells is about 70–80% and their metabolic capacity of phase I and II enzymes is ≥60% of that of freshly isolated cells.[129,130] The metabolic functionality of cryopreserved hepatocytes seems to be quite acceptable for short-term assays.

Primary cultured hepatocytes show a gradual loss of CYP activity[58,131,132] that is preceded by a decrease in CYP mRNA expression.[133,134] CYP activity levels of isolated human hepatocytes show a 50–60% reduction during the first 24 hours in culture, probably due to the adaptation of cells to culture conditions, thereafter remaining quite stable for a few days. Hence, cultured human hepatocytes can be safely used in drug metabolism studies for up to 2–3 days. The situation is quite different in rat hepatocytes, where CYP activities rapidly decay during the first 48 hours in culture.[58]

Despite certain limitations, hepatocytes are, in many respects, more appropriate than any other in vitro model for the prediction of drug metabolism in vivo.[26,135–137] They are well suited for investigating the metabolic profile of a drug across species, helping to better select the most appropriate animal species for further pharmacokinetic/biodisposition studies. An example of this type of comparative study is illustrated in figure 3 b. For the compound investigated here, pig hepatocytes instead of rodent or dog hepatocytes would be a more appropriate species to mimic human hepatic metabolism.[126]

Hepatoma cell lines have been investigated as alternatives to primary hepatocyte cultures in drug metabolism studies.[113,138,139] Despite showing an unlimited life span and being simple to culture, all hepatomas currently available show very limited metabolic capacity due to a very low expression of CYP activities.[126,140,141] Hence, they do not constitute a real alternative to primary cultured hepatocytes. An exception to this assessment is the human hepatoma BC2, a cell line that retains contact growth inhibition and that undergoes differentiation in vitro. The activities expressed by human BC2 cells are much higher than those recorded in HepG2 cells, but are still lower than in hepatocytes, as we have recently reported.[36]

3.2 Identification of CYPs Involved in Metabolism

Incubation of a drug with microsomes from a well-characterised liver bank was the first procedure to be used for this purpose. Characterisation of liver microsomes (i.e. determination of all major CYPs) is currently done in terms of catalytic function (by using CYP-specific substrates) and protein content (immunoquantification by Western blot analysis or enzyme-linked immunoassay). The use of microsomes from a liver bank with different CYP isoenzyme content allows establishment of a correlation between the rate of metabolism of the compound of interest and CYP activities present in the microsomes. Regression analysis gives indications about the major isoform involved in the biotransformation of the drug.[94,120,142–144] In view of the known multiplicity and overlapping substrate specificity of human CYPs, saturating concentrations of substrate should be avoided in these analyses. This is particularly relevant when more than one enzyme is presumably involved. By using specific inhibitors (table IV) or inhibiting CYP antibodies,[99–102,119,143,145] confirmation of the role of a given CYP in the metabolism of a drug can be obtained.

The involvement of a CYP isoenzyme in the metabolism of a drug can easily be determined by incubating the compound with genetically manipulated cells expressing single CYP genes.[27,146–150] These cells have been permanently transfected with a cDNA construct driven by a strong promoter in order to attain high levels of expression of a functional enzyme. Metabolism of the drug by a given CYP is examined by incubating the drug with each cell line separately.[119,120,143–145] These models indicate (i) whether the drug can be a substrate for a given CYP, and (ii) whether a given metabolite can be produced by a certain CYP. As incubation with cells can be extended for a significant number of hours, these in vitro models are particularly useful for investigating slowly metabolised compounds as well as metabolism due to minor CYP isoforms.

CYP-engineered cells, in contrast to microsomes, do not provide quantitative input about the participation of a given isoenzyme in the overall metabolism of the drug.[100,101,145,151] The indicators for correct interpretation of kinetic data from cDNA-expression systems have been the subject of recent studies.[27,146,148,149] Relative concentrations of accessory proteins (NADPH-CYP reductase and cytochrome b5) or membrane lipid composition may differ in a heterologous expressing system compared with human hepatocytes/human liver and, hence, influence the results. Moreover, in cDNA-expressing systems a single CYP interacts with an electron-carrier/supplier protein, whereas in liver, hepatocytes or microsomes many CYPs can interact with such proteins. This can lead to incorrect predictions of the relative contributions of individual CYPs to the metabolism of a drug.[149]

Taking into account the limitations inherent to each experimental model, it is advantageous to use them in a complementary fashion.[27,119,120,144,145,152] Such an integrated approach enables one to exploit the strengths of each model system. Thus, by comparing (i) the kinetics of metabolism in CYP cell lines and human hepatocytes and (ii) the activity of each CYP isoenzyme both in the cell line and in hepatocytes, is possible to determine the degree of participation of a given CYP isoform in the overall metabolism of a drug.[153]

3.3 Drug-Drug Interactions

Multidrug therapy is not uncommon in clinical practice. Simultaneous administration of several drugs may result in metabolic drug-drug interactions having pharmacological and/or toxicological implications. A metabolic drug interaction occurs when the disposition of one drug is altered by the presence of another compound.[154] As drugs are metabolised by a limited number of enzymes, they can compete with each other as substrates for the same enzyme. Thus, competitive, noncompetitive or irreversible inhibition of a CYP by one of the drugs will result in elevations in plasma/tissue concentrations of the other drugs. For compounds with a narrow therapeutic index, this can lead to overdosage and/or toxicity.[77] As CYP inhibition is an undesirable feature for a drug candidate, this information should be obtained before a drug candidate is considered for the clinical stages of development.

Nowadays, evaluation of potential drug-drug interactions has become mandatory for the registration of new drugs by most regulatory agencies.[155] There are two basic questions to be answered: (i) does the new drug influence the metabolism of other compounds; and (ii) is the metabolism of the drug candidate altered by other compounds that could be administered concomitantly in clinical practice?

Although related, both questions require specific experimental approaches. The inhibitory effect of the drug on a CYP isoenzyme can be readily investigated by co-incubating several concentrations of the drug and CYP-selective substrates (table III). Kinetic measurements, for instance, will help to determine whether the tested drug acts as an inhibitor on a given CYP isoform, as well the type of inhibition caused.[156–160] Liver microsomes, reconstituted enzymes and/or microsomes from CYP-engineered cells are suitable tools for this purpose,[101,120,146] and allow an easy comparison of the inhibitory potency of related drugs. However, the question might be somewhat more difficult to answer if metabolites, rather than the parent molecule, are the chemical entities involved. For this purpose the use of metabolically competent cells is more appropriate.

Another limitation of microsome assays is that inhibition parameters obtained may not accurately reflect the situation in vivo, since the contribution of drug transport is not considered. Inhibition assays can be performed in genetically engineered cell lines expressing individual drug-metabolising enzymes. These cells, because of the presence of membrane barriers are, in some aspects, more predictive than isolated enzymes or microsomes from hepatocytes.[161]

The second question is addressed by examining the rate of metabolism of the new drug in the presence of other compounds that are likely to be administered concomitantly in clinical practice. It is highly desirable that the pharmacokinetics of a drug candidate not be influenced by the coadministration of other drugs. Hence, using liver microsomes or microsomes from CYP-expressing cells, experiments are designed to monitor the rate of catabolism of the new drug in the presence of variable amount of the other compounds. Again, the use of cells, instead of subcellular fractions, although adding more complexity to the assays, better reflects the drug transport processes that are taking place in the liver.

Since in the metabolism of a compound several isoenzymes may be involved, the use of cells expressing one single CYP isoenzyme may overestimate the inhibitory effect of a given drug. Hence, the best picture of the potential drug-drug interaction is obtained in metabolically competent human hepatocytes.

3.4 In Vitro-In Vivo Extrapolation

With the use of appropriate controls, primary cultured human hepatocytes can be a good model to quantitatively predict the in vivo metabolic profile of a drug. In a recent study, the metabolism of aceclofenac (a prescribed analgesic) was examined both in vitro and in vivo. Hepatocytes were isolated and cultured from liver samples obtained in the course of programmed surgery; the metabolic profile of the drug and its rate of metabolism was investigated and compared with that observed in vivo in the same individual (figure 4). Although variations in drug metabolism were observed in the various cell preparations, they actually reflected the interindividual variation among donors in vivo.[26]

Fig. 4
figure 4

Aceclofenac metabolism in vitro and in vivo. Human hepatocytes were incubated with aceclofenac, and the rate of metabolism was determined in terms of 4′-hydroxy-aceclofenac formation. 4′-Hydroxy-aceclofenac excreted in the urine from the same individuals was also determined. The figure shows the linear regression of in vitro and in vivo metabolism, excluding those shown as empty symbols (reproduced from Ponsoda et al.,[26] with permission).

Full size image

Prediction of in vivo metabolic clearance from in vitro data is still difficult and controversial. Despite well documented examples showing a good in vitro and in vivo correlation in humans [26,135,137] and other animal species,[115,124,162–165] no general consensus has been reached concerning the parameters and experimental strategies to be used.[162,166–168] In vivo extrapolation of in vitro data is firstly based on the determination of in vitro clearance (calculated from Km and Vmax values in vitro) and, secondly, on in vivo parameters such as hepatic mass, hepatic blood flow and unbound drug fraction in blood.[115,137,162,166,169]

Although it is generally acknowledged that in vitro data can be used to broadly classify the compounds as a function of their low, intermediate or high rate of metabolism,[168] it is not always possible to reasonably predict the in vivo clearance based solely on in vitro metabolic data. The in vitro experimental design, possible participation of extrahepatic metabolism, active transport by the liver, the existence of interindividual variability and nonlinear pharmacokinetics are considered major causes of discrepancy between in vitro and in vivo data.[115,162,166]

Pooled hepatic microsomes, although representing a simplification of the hepatic metabolic machinery, still are the most frequently used model to determine the in vitro clearance of a drug. In contrast to microsomes, where substrates can easily reach the enzymes, hepatocytes possess a membrane and transport limiting systems that better reflect the in vivo situation. Thus, the in vivo clearance of xenobiotics is better predicted in studies with hepatocytes rather than with microsomes.[136,169] Additional factors such as binding of the compound to protein, uptake of the drug by cells or in vitro enzyme instability must also be considered as source of discrepancies when comparing microsomal incubations with hepatocyte incubations.[170]

4. Induction of CYP

Upon repeated administration, certain drugs can alter their own metabolism, or that of other simultaneously or subsequently administered therapeutic agents, by changing the expression of drug-metabolising enzymes. Exposure to certain chemicals (pollutants, cigarette smoke, alcohol and dietary constituents) can also induce drug metabolising enzymes. Induction may result in rapid metabolism of the drug and lower plasma concentrations to levels that are no longer efficacious.[171] Enzyme induction also accounts for the onset of tolerance to some therapeutic agents. On the other side, a possible consequence of enzyme induction is the increased formation of pharmacologically or toxicologically active metabolites. Thus, enzyme induction significantly contributes to interindividual differences in drug metabolism and toxicity.[24]

The phenomenon of CYP induction was first discovered and studied in experimental animals, but it was soon recognised to occur in humans. All human CYPs can be influenced to a certain extent, some of them being clearly induced by xenobiotics.[24,172,173] Notably, there appears to be more variation in response to enzyme inducers among humans than in animals, probably due to varying genetics, lifestyles and dietary habits.[174,175]

Enzyme inducers are usually classified on the basis of their action on individual CYP isoenzymes. Table VIII shows representative inducers for human CYPs. Some compounds show CYP inductive potential across species (particularly for the CYP1A family). For other substances, significant differences exist in their inducing abilities in animals. Rifampicin (rifampin), for example, is a potent inducer in humans and rabbits but it is a poor inducer in the rat.[176] In contrast, pregnenolone 16α-carbonitrile, a potent inducer of CYP3A in the rat, is not an inducer in either rabbits or humans.[177]

Table VIII
figure Tab8

Inducers of human cytochrome P450 isoenzymes[24,178–181]

Full size image

4.1 Mechanisms of Induction

Different mechanisms are known to operate in CYP enzyme induction, but in general it involves transcriptional activation of CYP genes by a receptor-dependent mechanism (table IX), resulting in increased levels of specific CYP mRNAs.

Table IX
figure Tab9

Mechanisms of induction of cytochrome P450[25,182]

Full size image

The first discovered inducing receptor, the aryl hydrocarbon (Ah) receptor, belongs to the PAS family of transcription factors.[183,184] It stimulates transcription of CYP1A genes via direct interaction with the promoter region of the gene (figure 5). Upon binding of the inducer to cytosolic Ah receptor, the complex is translocated to the nucleus, where it heterodimerises with the nuclear factor Arnt (Ah receptor nuclear translocator protein), and binds to an enhancer/promoter DNA region of CYP1A genes.[183,185] This is a well-conserved mechanism across species and accounts for the consistent induction of CYP1A by polycyclic aromatic hydrocarbons in many cell types.[186]

Fig. 5
figure 5

Role of nuclear receptors in cytochrome P450 (CYP) gene induction. Three nuclear receptors belonging to the nuclear receptor/steroid superfamily are recognised players in the inductive mechanisms of the CYP 2, 3 and 4 families: the constitutive androstane receptor (CAR) mediates CYP2B induction by phenobarbital and ‘phenobarbital-like’ chemicals (a), the peroxisome proliferator activated receptor (PPAR) activates CYP4A genes in response to peroxisome proliferators (b) and the pregnane X receptor (PXR) is involved in CYP3A induction by steroids and other chemicals (c). The aryl hydrocarbon receptor (AhR) activates transcription of CYP1A genes upon binding to polycyclic aromatic hydroxarbons (d).[25,182] Arnt = Ah receptor nuclear translocator protein; C/EBP = CCAAT/enhancer binding protein; HNF = hepatocyte nuclear factor; Hsp = heat shock protein; LAP = liver-enriched activating protein; RXR = retinoid X receptor; Sp1 = stimulatory protein 1.

Full size image

Three nuclear receptors belonging to the nuclear receptor/steroid superfamily are recognised players in the inductive mechanisms of the CYP 2, 3 and 4 families. The constitutive androstane receptor (CAR) mediates CYP2B induction by phenobarbital and ‘phenobarbital-like’ chemicals; the pregnane X receptor (PXR) is involved in CYP3A induction by steroids and other chemicals; and the peroxisome proliferator activated receptor (PPAR) activates CYP4A genes in response to peroxisome proliferators (figure 5).[25,182] All these nuclear receptors share a common heterodimerisation partner, the retinoid X receptor (RXR), and are subject to cross-talk interactions with other receptors. A number of endogenous ligands for these nuclear receptors have been identified: androstanes (CAR ligands), pregnenolone derivatives and other steroids (PXR ligands) and certain polyunsaturated long-chain fatty acids and their metabolites (PPAR ligands).[187–189] Endogenous ligands stimulate receptor activity, with the exception of CAR ligands, which are inhibitory. CAR is a constitutively active receptor and endogenous ligands may bind directly to the DNA-bound CAR-RXR heterodimer in a manner that alters the conformation of nuclear receptor complex and prevents its interaction with SRC-1 (a general transducer between DNA-bound nuclear receptors and the basal transcriptional machinery[190]). In the presence of phenobarbital, however, the binding of androstanes to CAR is abolished and receptor activity is thereby derepressed.[172,191] The mechanisms of CYP induction mediated by PXR and PPAR are different. Upon binding to an exogenous/endogenous ligand, the nuclear receptor displays enhanced binding to DNA as a PXR-RXR or PPAR-RXR heterodimer, activating the transcription of the gene.[188,192,193]

Ethanol-type induction of CYP2E1 is a non-receptor-mediated mechanism. Regulation of CYP2E1 expression by xenobiotics and/or pathophysiological factors occurs at transcriptional, translational and post-translational levels.[24,194] At low ethanol concentrations, protein levels are increased without changes in mRNA by a mechanism mainly involving protein stabilisation after binding of the inducer to the active site of the enzyme.[195] One proposed mechanism for CYP2E1 stabilisation is protection of the protein from proteolytic degradation by the enzyme-bound substrate.[196] No changes in mRNA are generally observed. At higher concentrations, ethanol produces additional induction by increases in transcription.[197]

4.2 Assessment of Induction In Vitro

CYP induction was classically investigated in vivo by administering the drug to animals followed several days later by monitoring of sleeping time upon administration of a given dose of a barbiturate. However, because of the known interspecies differences in CYP induction, the convenience of examining CYP induction in humans was soon recognised. Human studies are limited, for obvious ethical reasons, to compounds that are at a late stage of clinical development, and not to drug candidates at the preclinical stage.

Screening for potential inducers cannot be done in microsomes, but requires the use of a cellular system fully capable of transcribing and translating CYP genes. Cultured hepatocytes do have these features, as they respond to CYP inducers,[34,58,131,198,199] and hence represent a suitable in vitro model for scrutinising chemicals of unknown CYP induction potential.

Enzyme induction can be easily monitored in vitro as an increase in catalytic activity of cells in response to a stimulus.[34,58,131,198] This is the direct consequence of an increased amount of enzyme present in cells. Upon incubation of hepatocytes for 24–72 hours with a compound, assessment of CYP activities is performed using specific substrates. The activity measured in induced cells is then compared with that in control, untreated, cells. In general, hepatocytes respond very well to CYP1A2 inducers. Typically, a 100-fold induction can be observed for CYP1A2, and a 2–10-fold induction for the other inducible CYP isoforms. Hepatocytes in culture respond moderately to phenobarbital, an otherwise potent in vivo inducer.[58,131] In addition to the measurement of catalytic activities using specific substrates, information about the pattern of isoenzyme induction can be obtained by immunodetection in cell homogenates using appropriate antibodies.[132,173,199,200]

Human hepatocytes in primary culture respond well to enzyme inducers during the first few days; this ability is lost thereafter. Rat hepatocytes are much more unstable, and soon become unresponsive to inducers. Culturing hepatocytes in a collagen matrix greatly improves their responsiveness to CYP inducers. In a paper by Gómez-Lechón et al.,[201] collagen-entrapped hepatocytes were able to induce CYP activities in response to 3-methylcholanthrene and ethanol even after 2 weeks in culture (figure 6). Conversely, hepatocytes cultured on plastic dishes were unable to respond.

Fig. 6
figure 6

Induction of cytochrome P450 (CYP) 1A2 and CYP2E1 in collagen-cultured hepatocytes incubated with methylcholanthrene (MC) and ethanol (EtOH). Hepatocytes were treated with 2 µmol/L methylcholanthrene (a) or 75 mmol/L ethanol (b) for 3 and 6 days, respectively, starting on a different day of culture. Activity values in hepatocytes entrapped in collagen gel (open bars) or cultured on plastic (hatched bars) are expressed as the fold variation respect to control. Changes in CYP1A1 and CYP2E1 mRNA levels were determined by semiquantitative reverse transcriptase-polymerase chain reaction.

Full size image

The use of cryopreserved human hepatocytes in enzyme induction studies is constrained by the fact that these cells respond poorly to enzyme inducers.[130] Future use of cryopreserved cells in induction studies will depend on improvements in the cryopreservation procedures. Liver slices could also be a valuable system for this type of studies. Despite their limited survival, preliminary data obtained with this model are very promising (G.M.M. Groothuis, P. Olinga and J.V. Castell, personal communication).

As enzyme induction is preceded by an increase of CYP transcription (except for CYP2E1), the use of quantitative RT-PCR has been proposed as a reliable and sensitive technique when a reduced amount of cells or tissue is available. By selecting appropriate primers, Rodríguez-Antona et al.[109] succeeded in developing a procedure that allows the quantitative measurement of the ten most relevant human CYPs in human hepatocytes incubated with model drug inducers. Differences compared with control in CYP mRNA are best observed after 48 hours of continuous incubation with the inducer. The small amount of cells required for this technique makes this procedure suitable for studies in liver needle biopsies (unpublished observations).

The use of RT-PCR techniques has enabled the evaluation of CYP induction in the less differentiated hepatoma cell lines.[202,203] The response of these cells to inducers in terms of enzyme activity is very poor, except for the non-hepatic isoform CYP1A1.[140] However, by making use of PCR quantification techniques some authors have claimed that it is possible to monitor the induction of some isoforms (CYP3A4 and CYP2C9).[202,203]

In a recent study, it was shown that the in vitro differentiated human hepatoma cell line BC2 is able to increase CYP activities in response to methylcholanthrene, phenobarbital, rifampicin and dexamethasone.[36] Although CYP activities in BC2 were markedly lower than in human hepatocytes, the effects elicited by some of these model inducers were comparable, in relative terms, to those observed in human cellular models (figure 7).

Fig. 7
figure 7

Induction of cytochrome P450 (CYP) in human hepatocytes and BC2 cells. Primary human hepatocytes (open bars) or in vitro differentiated BC2 cells (hatched bars) were exposed to inducers after 24 hours or 21 days of culture, respectively, and CYP activities were determined 4 days later. (a) Increases in 7-methoxyresorufin O-demethylase (MROD), 7-ethoxyresorufin O-deethylase (EROD) and 7-ethoxycoumarin O-deethylase (ECOD) induced by methylcholanthrene (MC); (b) effect of phenobarbital (PB) on 7-benzoxyresorufin O-debenzylase (BROD), 7-pentoxyresorufin O-depentylase (PROD) and ECOD; (c) changes in testosterone 6β-hydroxylation in cells treated with rifampicin (RIF), dexamethasone (DEX) and phenobarbital (PB) [reproduced from Gómez-Lechón et al.,[36] with permission].

Full size image

In summary, from a practical point of view, primary cultured hepatocytes are presently the most reliable in vitro model for global examination of the inductive potential of a drug.

5. Conclusions and Outlook

Characterisation of CYP genes can be done in terms of enzyme activity, protein content and gene analysis/expression. Precise information on specific substrates and inhibitors for each major human CYP is currently available, and their use has become a routine procedure for the metabolic phenotyping of hepatocytes. However, few CYP-specific substrates/inhibitors can be used in vivo as well as in vitro in humans. The most recent strategy for rapid CYP activity assessment in hepatocytes makes use of a mixture of several compounds, each one being specifically metabolised by a CYP isoform. This procedure requires high sensitivity detection techniques (HPLC-MS/MS), but may soon constitute a standard procedure for hepatocyte phenotyping.

Amplification-based PCR techniques, first applied to identify CYP polymorphisms, are being satisfactorily used to monitor CYP mRNA expression. Provided that the technique can be performed quantitatively, and the principles can be applied to cDNA-chip technology, the measurement of CYP gene expression in small biological samples could be considerably facilitated.

Pharmaceutical companies are making use of in vitro drug metabolism models to speed up the selection of new drugs. The comparative metabolic profile of a drug, the enzymes involved and the potential enzyme-inhibiting or enzyme-inducing properties of a compound are key issues that should be investigated at early stages of drug development to better select drug candidates. Although human hepatocytes are still the best model to gain a complete picture of how a compound will be metabolised in vivo, the currently used in vitro models (primary cultures, encapsulated hepatocytes or tissue slices) are not suitable for high throughput screening. By genetic manipulation of hepatocyte cell lines (transfection with vectors encoding key transcription/nuclear factors), it is possible to attain a significant degree of differentiation.[36] Such cells expressing enzyme activities and responding to inducers could become a real alternative to human hepatocytes.