The use of heterologously expressed drug metabolizing enzymes— state of the art and prospects for the future

Abstract

The first report of the functional, heterologous expression of a mammalian cytochrome P450 (CYP) enzyme occurred more than a decade ago. In the intervening years, these expression systems have been optimized with regard to the specific requirements for production of catalytically active enzymes. In this review, we discuss the strengths and limitations of heterologously expressed enzymes as they affect in vitro drug metabolism studies. Emphasis is given to new applications (screens for CYP inhibition and novel enzyme mixtures) that have been enabled by high level, functional expression of CYP enzymes.

Introduction

While many enzymatic systems metabolize drugs and other xenobiotics, this review focuses on the cytochrome P450 (CYP) mixed-function oxygenase system. The CYP system is a superfamily of membrane-bound, heme-containing mixed-function oxygenases that are the principal enzyme system for the metabolism of drugs. These enzymes are expressed in many tissues, but are found at the highest levels in liver. CYPs principally function to introduce oxygen into a molecule to increase the hydrophilicity of the product and hence, the ease with which the product can be eliminated from the body. A nomenclature for these enzymes has been developed based on similarities in amino acid sequences (Nelson et al., 1996).

The catalytic activities of CYPs require molecular oxygen and reducing equivalents from NADPH. The catalytic cycle of the microsomal CYPs requires electron transfer from NADPH via the flavoprotein NADPH:CYP oxidoreductase (OR). In addition, another membrane-bound, heme-containing protein, cytochrome b₅, stimulates the activity of some CYPs towards some substrates. Therefore, the catalytic activity of a particular CYP in a particular tissue, subcellular fraction, or expression system is determined not only by the abundance of the CYP, but also by the abundance of its electron transport partners.

Human populations exhibit considerable variability in CYP activity levels. This is due to the fact that some human CYP enzymes are polymorphic, with a significant percentage of populations being deficient in a specific enzyme (e.g., CYP2C19 and CYP2D6; Gonzalez et al. 1988, Wrighton et al. 1993) or having a functional enzyme with an altered amino acid sequence, which can change the kinetics of substrate metabolism (CYP2C9; Furuya et al., 1995). Expression levels of specific CYP enzymes vary substantially among individuals. The levels of other human CYP enzymes (e.g., CYP1A2 and CYP3A4) are induced by certain environmental exposures or drug treatments. If an enzyme, which is polymorphic or subject to environmental regulation, is rate-limiting for the elimination of a drug, then substantial interindividual variation in pharmacokinetics is often observed.

There are 11 xenobiotic-metabolizing CYPs that are expressed in a typical human liver (CYP1A2, CYP2A6, CYP2B6, CYP2C8/9/18/19, CYP2D6, CYP2E1, and CYP3A4/5). Comprehensive reviews of the properties of the enzymes comprising each of the CYP subfamilies have been published Ioannides 1996, Rendic & Di Carlo 1997. The reader is directed to these reviews for a detailed description of the individual CYP forms.

A relatively limited subset of these enzymes (CYP1A2, CYP2C9, CYP2C19, CYP2D6, and CYP3A4) appears to be most commonly responsible for the metabolism of drugs and associated drug-drug interactions (Spatzenegger & Jaeger, 1995). The relative importance of this subset of enzymes is due to both the mass abundance of these enzymes (e.g., CYP3A4 is the most abundant P450 in human liver at ∼30% of total P450) and the preference of these enzymes to bind and/or metabolize chemical structures commonly found in drugs (e.g., CYP2D6 preferentially binds and metabolizes drugs with basic amine functionalities).

Quantitative and qualitative aspects of CYP-mediated metabolism are important in drug development. Four of these aspects are as follows:

• The overall rate of metabolism of the drug or metabolic stability: The rate of metabolism influences pharmacokinetic properties, such as bioavailability and clearance, in both humans and preclinical animal species. It is currently common to perform in vitro analyses of metabolic stability in support of drug discovery programs. Isolated hepatocytes, liver slices, or tissue fractions (e.g., microsomes fortified with the necessary cofactors) are typically used as the active metabolic element (Houston, 1994).

• The metabolite profile: The profile of metabolites is important for the selection of appropriate species for preclinical safety assessment (i.e., are the metabolites formed in humans also formed in the toxicology animal species?). Metabolite profile is also of importance if the drug is administered as a prodrug or if pharmacologically active metabolites are formed. Like metabolic stability studies, it is common to perform in vitro metabolite profile analyses in support of drug discovery programs.

• The specific CYP enzyme(s) involved in metabolism (or reaction phenotype): The number and identity of the enzymes involved in metabolism influence the interindividual variability in metabolism (i.e., metabolism exclusively by polymorphic enzymes will lead to high levels of variability). The safe clinical use of drugs showing a high degree of interindividual variability requires a large therapeutic index or careful patient monitoring. In addition, identification of the enzyme(s) involved in metabolism can predict which co-administered drugs may inhibit metabolism and cause a drug-drug interaction. The involvement of more than one enzyme in a drug's metabolism will tend to decrease interindividual variability and mitigate any single-enzyme drug-drug interactions.

• The specific CYP enzymes inhibited by the drug: Metabolism by and inhibition of CYPs are distinct processes that should be analyzed separately. For example, a drug may inhibit the metabolism of a CYP that is not significantly involved in its metabolism. The majority of drug-drug interactions are metabolism-based, i.e., two or more drugs competing for metabolism by the same enzyme (Murray, 1992). If the drug is a potent CYP inhibitor, it may inhibit the metabolism of a co-administered medication and cause a drug-drug interaction. New chemical entities that cause drug-drug interactions may be more costly to develop, suffer from decreased market acceptance, or simply not survive the drug development and registration process.

There are several commonly used systems for the heterologous expression of CYPs. These include bacteria (Escherichia coli), yeast (Saccharomyces cerevisiae, and to a lesser extent Schizosaccharomyces pombe), insect cells, and mammalian cells. These systems can be used to express the CYP enzyme alone or in conjunction with one or both of its redox partners. Expression of some level of CYP oxidoreductase and cytochrome b₅ appears to be ubiquitous in mammalian cells, although the native levels may be rate-limiting for catalysis. The appropriate electron transport capabilities in the other, non-mammalian host cells may be minimal to nonexistent. The different expression systems for CYPs have been reviewed recently (Gonzalez & Korzekwa, 1995). In this review, we highlight some of the differences and similarities between these expression systems and the practical impact of these differences.

Due to the requirement for redox partners, the heterologous expression of only a CYP enzyme rarely results in a system with optimal catalytic activity. The CYP may be purified and reconstituted with its redox partners, or these redox partners may be co-expressed with the CYP. Co-expression of OR with CYP generally reduces the yield of spectral CYP enzyme Pritchard et al. 1998, Dong & Porter 1996, Chen et al. 1997. However, the catalytic activity of the preparation (expressed as pmol product per mg protein per min) is substantially increased. A detailed analysis in the insect cell/baculovirus system Gonzalez et al. 1991, Tamura et al. 1992, Chen et al. 1997 indicated that while the yield of spectrally active CYP was reduced by increasing the levels of co-expressed OR, the total amount of cDNA-derived CYP apoprotein (as determined by Western blot) was relatively unaffected. This observation implies that high levels of OR expression either attenuate the incorporation of heme into the CYP or decrease the stability of the heme in expressed protein. Additional investigations are needed to determine which of these processes is the mechanism for the loss of spectral P450.

Bacterial expression has the advantage that system development is rapid and that the culture medium is inexpensive relative to mammalian or insect cell systems. The bacterial system is ideal for the isolation of large quantities of enzyme for spectroscopic or structural studies. Bacterial expression has been performed using two approaches: modification of the N-terminal sequence of the protein Iwata et al. 1998, Barnes et al. 1991 or incorporation of the ompA targeting sequences that direct the CYP to the periplasmic space of the bacteria, where it is then removed by proteolytic cleavage (Pritchard et al., 1998). With either approach, appropriate choice of the host strain is essential for efficient cDNA expression. The former approach modifies the primary sequences of the CYP, and while the N-terminal region of the protein is not believed to be important for substrate recognition or catalysis, extensive validation of the kinetic properties of the expressed protein is prudent. In general, the reported rates of catalysis for isolated bacterial-expressed CYPs are lower than those observed for other systems. When mammalian OR is co-expressed with a CYP in bacteria, significant rates of catalysis are obtained in whole cells (Iwata et al., 1998). This offers the possibility of using bacterial cultures as bioreactors for the biosynthesis of metabolites.

Yeast expression shares many of the advantages of bacterial expression. The host organism grows rapidly on inexpensive media, and reasonable levels of expression and catalytic activity can be obtained with this system Oeda et al. 1985, Gonzalez & Korzekwa 1995. The stable co-transfection approach has been used successfully to co-express human OR and human b₅ in yeast (Peyronneau et al., 1992). A CYP3A4 turnover number of 1.9 min⁻¹ for testosterone 6β-hydroxylation, a 30-fold increase in turnover relative to that reached with only native yeast OR and b₅, was achieved by this method. Endogenous yeast P450s may contaminate microsome preparations and require appropriate control preparations.

Baculovirus expression offers high-level expression of both the CYP and OR (Lee et al., 1995). However, highest spectral P450 contents are obtained when the P450 is expressed in the absence of co-expressed OR (Buters et al., 1994). The development of the cDNA-bearing virus is relatively labor- and time-consuming. However, several convenient systems for virus development are commercially available. Baculovirus expression appears to offer the best combination of high CYP content and high catalytic activity per unit CYP enzyme. However, because the enzymes are produced transiently in the host cells, exact harvest time and other culture conditions can have a pronounced effect on the activity of the final preparation. Therefore, the expression process must be carefully controlled to minimize variability among preparations (Fig. 1).

Mammalian cell expression requires the highest level of resources for the development of the cDNA-expressing cell and the production of enzyme. Several host mammalian cells have been used for the expression of CYP cDNAs. These include V79 cells (Doehmer et al., 1988), Chinese hamster ovary cells (Ding et al., 1997), HepG2 cells Aoyama et al. 1990, Dai et al. 1993, NIH 3T3 cells (Battula et al., 1987), and human lymphoblast cells Crespi et al. 1990, Crespi et al. 1991a. At some level, most CYP enzymes appear to adversely affect the growth of the host mammalian cell. This effect on growth presents the upper limit on the level of expression that can be achieved. However, within this limitation, the system is quite reproducible, and there is the possibility to exploit the presence of native levels of co-enzymes and to avoid the need for co-expression or reconstitution (Fig. 1). Mammalian cell expression offers the possibility to couple CYP metabolite formation to toxicological endpoints (i.e., cytotoxicity, malignant transformation, or mutagenesis investigations). However, some toxicological endpoints can also be measured in bacterial and yeast systems.

For reasons that are poorly understood, some specific CYP cDNAs do not yield significant levels of expressed protein in some systems. For example, human CYP3A5 cDNA expresses very well in most systems, except the human lymphoblastoid cell system. In contrast, CYP2F1 expresses (albeit at a low level) in lymphoblastoid cells, but does not express in baculovirus.

Over the past 10+ years, the expression systems have been optimized, and for applications involving the in vitro study of drug metabolism, any of the above systems are useful sources of individually expressed CYPs. The recent trend has been toward both bacterial and baculovirus expression. This trend appears to be due to both the overall good yields of active enzyme and the widespread technical competence with these systems.

There is currently little need for dramatic technical improvements to these expression systems. There is a much greater need to develop novel applications of these materials in order to facilitate drug candidate selection.