HepG2 cells with recombinant cytochrome P450 enzyme overexpression: Their use and limitation as in vitro liver model

1 Introduction

Regarding the investigation of questions related to liver biotransformation, the challenge is to generate an in vitro model that is as close to the human liver as possible and still practicable to use. Therapeutic efficiency of most compounds strongly depends on biotransformation which can lead to detoxification and excretion of the given drug, but also to bioactivation. Many drugs are converted into potentially reactive metabolites with much greater cytotoxicity, mutagenicity, or carcinogenicity [1, 2]. Therefore, an early risk assessment of drug-induced toxicity in the pharmaceutical industry is of great importance and requires suitable in vitro biotransformation models. Biotransformation mainly occurs in hepatocytes of the liver and consists of two reaction steps: phase I (functionalization) and phase II (conjugation). Phase I is mainly catalyzed by the cytochrome P450 (CYP) superfamily, which consists of 57 functional CYP protein coding genes and is classified into 18 CYP families [3]. Among them, the human CYP-families 1 to 4 are predominantly involved in the biotransformation of drugs [4, 5].

In the last few decades, a wide range of different strategies for in vitro human liver models have been developed to determine drug biotransformation by CYP enzymes [4, 79], with freshly isolated primary human hepatocytes (PHH) frequently been regarded as gold standard. However, the loss of CYP enzyme activity during cultivation of primary hepatocytes, poor availability of human liver tissue, high costs, short life-span in cell culture, and donor variability limit their usage for drug discovery at a large scale [7–9]. To overcome scarcity and donor variation we, as well as other research groups, have been using the hepatoblastoma cell line HepG2. Because of relatively low CYP expression in HepG2 [10, 11], attempts were made to genetically modify the cells by introducing CYP-encoding genes via transfection or transduction. In this review, reports on generation and application of HepG2 cells with transient or stable expression of CYPs are presented and their respective advantages and disadvantages are discussed.

2 The gold standard

Primary human hepatocytes are differentiated, fully competent metabolic cells [12] with high activity of drug-metabolizing enzymes and transporters [13, 14]. These cells are freshly isolated from liver tissue - mainly from patients undergoing partial liver resection - and remain viable for several days or up to few weeks in monolayer culture [4, 8].

Due to their ability to retain hepatic functions, primary human hepatocytes are the closest in vitro cell model to human liver [15] and are widely regarded as “gold standard” in drug discovery, liver metabolism and hepatotoxicity studies [9, 16] as well as for CYP induction analysis [17].

Although they have excellent properties, the use of primary hepatocyte cultures is limited due to the rare and irregular availability of suitable human liver tissue. In vitro, these cells dedifferentiate quickly and show a restricted ability to proliferate [18]. For example, CYP mRNA levels drastically decrease already during the isolation process and decline further within the first hours in monolayer culture [15, 19]. It was shown that primary hepatocytes retain no more than 20–40% of the enzyme activity in vivo after 48 h in culture, which only allows short-time experimental setups [15]. In addition, basal expression and functionality of CYP enzymes strongly depends on culture conditions such as medium composition, cell density and culture time [20], which hampers the generation of stable and reproducible results. To add another level of complexity to the difficult and time-consuming hepatocyte isolation and cultivation, it must be mentioned that different human donors display a wide variation of hepatic gene expression and enzyme activity. In summary, the phenotypic instability in vitro, the challenging culture conditions, as well as the high donor variability make primary hepatocytes unsuitable for routine test procedures [14, 21].

3 Cell lines as alternative

As an alternative to primary hepatocytes, human liver cell lines can be used as an in vitro model, of which HepG2 is the best characterized and most commonly used system [4]. HepG2 originated from an epithelial hepatoblastoma of a 15-year-old Caucasian male [22] and retained several hepatocyte-specific characteristics, such as synthesis and excretion of albumin and bile acids [23, 24]. Comparative gene expression profiles of HepG2 cells and primary hepatocytes exist [14, 26] and can help to decide individually whether HepG2 cells are a suitable alternative to primary cells for a given study design. Obvious advantages over primary hepatocytes are their unlimited availability and life-span, easy handling due to simple and inexpensive cell culture media, and a stable phenotype. These properties promise a high experimental reproducibility and appropriateness for standardized screening purposes [21, 27]. Since HepG2 cells have been shown to express several phase I enzymes at a basal level that can be increased using certain induction reagents such as such as 3-methylcholanthrene for CYP1A2 or phenobarbital for CYP2B6 and CYP3A4 [20, 29], they are also used for metabolism and toxicological studies [30–34].

The major drawback of HepG2 cells is their limited biotransformation activity [21], especially due to their low or even absent basal expression of CYP enzymes as compared to primary hepatocytes [4, 28]. Therefore, HepG2 CYP activity appears to be sufficient for toxicological analysis of certain compounds [31–33], but their usage can also lead to underestimation of toxic effects [28]. For example, we could show in our own study that HepG2 cells are less sensitive to the anti-cancer drug cyclophosphamide when compared to CYP3A4- or CYP2C19-overexpressing HepG2 cells, which thus reflect the in vivo situation better than the native HepG2 cells [80]. Phase I and II enzyme activities also depend on culture conditions, culture time and source of the HepG2 cells [35–38], which further complicates the interpretation of toxicological studies and the reproducibility of results.

Another widely accepted in vitro model as alternative to primary hepatocytes are HepaRG cells. They were derived from a hepatocellular carcinoma and have mRNA levels of drug metabolizing enzymes comparable to primary human hepatocytes [39, 40]. However, Gerets et al. have shown that despite higher gene expression levels of drug-metabolizing enzymes in HepaRG cells and a higher inducibility of CYPs, their sensitivity to hepatotoxic substances is not remarkably better than in HepG2 cells [41]. For drug metabolism and toxicity studies, a higher basal CYP activity may be required, which can be achieved by overexpression of CYPs in HepG2 cells.

4 Improving biotransformation in HepG2 cells: Transient overexpression of CYPs

Many strategies have been developed to overcome the lack of CYP activity in HepG2 cells by generating metabolically competent cells through gene transfer of expression vectors encoding for human CYPs. The principle behind it is to benefit from the unlimited availability and proliferation capacity of HepG2 cells and to genetically modify their CYP expression patterns, mainly by using gene transfection or viral transduction.

Adenoviral vectors have prevalently been used for the generation of HepG2 cells transiently expressing drug-metabolizing enzymes [27, 42–48]. Due to the rapid and efficient infection properties of adenoviruses, high expression rates of the transgene can be achieved in a broad range of host cell lines including HepG2 [43]. The fact that the method can be performed easily with high efficiency, accuracy and reproducibility in multi-well culture plates without toxic or mutagenic effects in the host cells, allows the usage of transient CYP expression for high-throughput drug toxicity studies [9, 49]. Co-transduction of several CYPs is well feasible in HepG2 cells in order to come as close as possible to the expression pattern of the in vivo human liver [27, 49]. It was claimed by some authors that it is possible to adjust the virus dose used for cell infection and thus to influence the transgene expression rate. This should enable the creation of individual and controllable CYP expression patterns as well as relative ratios to mimic the in vivo situation of certain human populations such as extensive or poor metabolizers [9, 27]. However, using transient adenoviral transduction it is hardly possible to exactly calculate from the used multiplicity of infection (MOI) the cellular load with functional and defective adenovirus particles, the latter being always present in virus preparations. Furthermore, most recombinant adenoviral vectors make use of strong heterologous promoters such as the CMV promoter and thus already few adenovirus copies per cell should lead to strong transgene expression. In addition, the intrinsic disadvantage of this method is that adenoviral genomes normally are maintained as episomes without any integration into the host genome. Since conventional adenoviral vectors do not replicate in their target cell, the virus will be diluted out along with cell proliferation. This inevitably results in a short-term (transient) expression of the enzyme of interest lasting for a few days only, so that a new adenoviral transduction is necessary for each experiment [27, 47].

5 Improving biotransformation in HepG2 cells: Stable overexpression of CYPs

In addition to plasmid-based transfection, retroviral and especially lentiviral vectors are suitable tools to obtain cell clones stably overexpressing transgenes by taking advantage of a virally encoded integrase that achieves efficient integration of DNA sequences into the mammalian genome. Several research groups have stably transfected HepG2 cells with expression vectors encoding individual human CYP isoforms by using transfection reagents [50–52]; we and others have transduced the cells using lentiviruses [53–55] resulting in a higher efficiency of transgene expression. In both methodological approaches, successful transfer of the gene of interest can be monitored by designing vectors encoding for resistance genes and/or fluorescent proteins. Finally, selected cell clones that stably express the CYP (or any other gene) of interest can be expanded and upscaled. It should be noted that retroviral vectors do not integrate specifically, but randomly within the target genome, which is associated with a mutagenic risk of alteration of important genetic regions. Compared with transient adenoviral transduction methods, the capacity of lentiviral vectors is much more limited regarding the size of the coding sequence(s) of interest [9]. Despite these limitations, lentivirally engineered cells represent an extremely robust in vitro model that is easy to handle and that is warranting stable phenotypic properties over many passages. These properties make cell clones stably expressing particular CYPs an indispensable tool allowing high experimental reproducibility in metabolic and toxicological studies.

6 Application of recombinant CYP-overexpressing HepG2 cells

It is crucial for the pharmaceutical industry to exclude drug candidates that can cause potentially harmful side effects as early as possible, i.e. in the early research phase and not just during expensive phases of animal or clinical studies. In general, it must be investigated whether potential new drug candidates are converted by biotransformation enzymes and how the respective metabolites would contribute to pharmacological and toxic effects [12]. Whether with stable or transient expression, HepG2 cells overexpressing CYPs have been shown to be a suitable in vitro model for such early toxicological investigations and risk assessment of drugs and chemicals [50, 56–60]. It is of special importance to use a physiologically relevant liver cell model for the characterization of new drug candidates, since hepatotoxicity, triggered by the drug itself or its metabolites, is the major reason for failure of new drugs during clinical testing, or it can even cause their withdrawal from the market [61]. Hashizume et al. demonstrated the use of ten HepG2 transformants stably expressing major human CYP isoforms as a valuable screening tool to test for possible genotoxicity of substances requiring bioactivation [57, 58]. The pharmacokinetic properties of drugs are of great importance in the optimization and selection of drug candidates. Donato et al. refers to HepG2 cells transduced with adenoviral vectors encoding a single CYP enzyme as promising in vitro system for clearance predictions of metabolized drugs [43]. In addition, such cell-based systems allow the investigation of drug action mechanisms in detail, for example their interaction with other proteins such as receptors [62], or possibly also their influence on gene expression as well as potential resistance mechanisms. HepG2 cells that overexpress one certain CYP are also ideal for screening CYP isoforms that might be involved in conversion and toxicity of an investigated substance [43, 63]. Such reaction phenotyping enables the pharmaceutical industry to choose for drug candidates that act independently of enzymes that are affected by polymorphisms, and thus to avoid strong inter-individual differences in drug effects [12].

With such in vitro liver models, it is also possible to quickly identify cytochrome P450 inhibitors [12, 65]. Especially for multidrug therapies, it is important to predict which drug has which influence on metabolizing enzymes, since such mechanisms can alter the drug’s own effect or that of other drugs. Due to the overlapping substrate specificity of CYP enzymes, two or more substances may compete for one CYP, or one of the compounds can act as a CYP inhibitor thereby reducing the effect of the other drug. Both scenarios can result in an over- or underdosage of co-administered drugs [12].

HepG2 cells expressing CYP enzymes can not only be used for toxicity testing of drugs, but also of other substances that are absorbed by the body. For example, ethanol is known to cause hepatotoxicity, as CYP2E1 contributes to the formation of reactive oxygen species (ROS) during ethanol metabolism. Therefore, a HepG2 cell line overexpressing CYP2E1 is often used as a model to study metabolic and pathological alterations in the liver caused by alcohol consumption [66, 67]. HepG2 cell clones stably overexpressing cytochrome P450 enzymes with a physiological enzyme activity were further proposed to be useful as a reference cell line for CYP enzyme activity during comparative studies of liver cell models [55].

7 Advantages and disadvantages of CYP-overexpressing HepG2 cells

HepG2 cells originated from a human liver and thus exhibit liver-specific characteristics; for example, they retain the activity of many phase II enzymes. HepG2 cells express and show activity of UDP-glucuronyltransferases (UGT), sulphate transferases (SULT) and glutathione S-transferase (GST), whereby the expression level of GST is comparable to that in human primary hepatocytes [11, 68–70]. The presence of phase II enzymes, which are responsible for conjugation reactions, and their cofactors are important in toxicological studies since they might cause both the deactivation or the activation of a drug [58, 71]. Treatment of cells with the substance of interest combined with phase II inhibitors might be helpful in identifying in which phase of biotransformation a substance becomes toxic or becomes inactivated [58]. In addition, electron carriers such as NADPH-cytochrome P450 reductase or cytochrome b5 are formed per se in HepG2 cells [9, 72]. These are essential for the function of CYPs and upon genetically introducing CYP enzymes into HepG2 cells, no co-transfection or transduction with such electron carriers is required for the enzymatic CYP function [50].

Due to their liver-specific transport and metabolism system, hepatic cell lines in general reflect the in vivo liver situation better than cell-free systems like microsomes. These subcellular fragments derived from the endoplasmic reticulum of hepatocytes are also used as an in vitro system for biotransformation research [4], since they contain almost only CYP and UGT enzymes. Microsomes from primary human hepatocytes have been co-cultivated together with HepG2 cells to bridge their lack of expression of metabolizing enzymes [17, 73]. However, important phase II enzymes and cofactors are missing in this system for a complete biotransformation and substances are metabolized extracellularly. Thus, metabolites have to pass through the membranes of co-cultured cells before effecting the target cell, which possibly leads to misinterpretation of observed effects [74] and requires a long half-life of the metabolites. Therefore, cell models such as CYP-overexpressing HepG2 are better suited for toxicity testing of highly reactive, short-lived metabolites that are formed and can act directly in the cell [9].

CYP-overexpressing HepG2 cells can exhibit physiologically relevant expression levels of the transduced CYP, which allows to study biotransformation reactions, and metabolites of compounds undergoing such reactions can be produced in sufficient, well detectable amounts without the need to enhance CYP expression by induction compounds [55, 75]. The continuous expression of CYPs in HepG2 cells opens up the possibility to perform long-time drug exposure experiments to study chemicals that are metabolized slowly or by minor CYPs, which is not feasible with primary human hepatocytes [12].

Nevertheless, such liver cell models cannot fully reflect the in vivo situation, and limitations of the model depending on the experimental goal should be known to every user [76]. For example, CYP-overexpressing HepG2 cells show only an increased activity of the genetically introduced enzymes, while the expression rates of other CYPs or biotransformation-relevant proteins such as conjugating enzymes or transporters are rather low. This imbalance of the factors involved in biotransformation makes it difficult to transfer results to the in vivo situation [9]. Due to possible low transporter expression, test substrates might not be able to enter or respective metabolites might not be able to leave the cell [77]. Many chemically reactive metabolites bind to intracellular macromolecules like proteins, nucleic acids or lipids [54], which can make their detection more difficult as in contrast to cell-free systems. It should be noted that HepG2 cells show a low basal expression of certain phase I enzymes. Thus, observed effects of a compound cannot be solely attributed to the CYP that is overexpressed in HepG2 cells and that is in the focus of the respective study. In general, it is difficult to assess the significance of findings when genes are expressed at an enhanced and probably non-physiological level [78].

Obviously, a cell culture model such as CYP-overexpressing HepG2 cells can never completely replace animal models in drug studies. In particular, it is not possible to assess complex questions like the absorption and bioavailability of drugs after oral administration with such a model system. However, cell-based in vitro systems can reduce the use of in vivo models as well as related costs and animal burden by identifying exclusion criteria, especially in the early stages of drug candidate characterization [76].

8 Conclusion

Primary human hepatocytes are still the gold standard and best in vitro model to study drug metabolism and other liver-specific features. However, due to the limitations of primary hepatocytes, HepG2 cells overexpressing defined CYP enzymes provide a cost-effective and easy to handle model system for studying certain liver functions in vitro. With their stable phenotype, such cells might become an indispensable option to generate highly reproducible results for specific research questions.