Metabolism by cytochrome P450 (P450) is a major route of detoxification for a large number of xenobiotics, including pharmaceuticals (Guengerich, 1990). It is now well established that expression of P450 genes can be regulated by a range of chemicals (LeCluyse, 2001). Consequently, any changes in the expression of these enzymes are of importance to scientists in the field of drug development, especially if the changes are brought about by treatment with a drug candidate.

An inducing drug may lead to a drug-drug interaction through induction of an alternate enzyme responsible for the clearance of a coadministered therapy, thereby resulting in a decrease in exposure of this drug (Worboys and Carlile, 2001). Another possible consequence of P450 induction is that blood concentrations of drugs that induce their own metabolism may fall to subtherapeutic levels (a so-called autoinduction), negating any beneficial effect (Simonsson et al., 2003). Consequently, an understanding of the potential for a new chemical entity to elicit an induction response is critical to avoid costly errors in the clinic.

Traditionally, the approach taken was to repeatedly dose animals with the test compound, followed by analysis of liver P450 enzymes ex vivo. This approach in drug discovery is cumbersome because it requires large numbers of animals and a large amount of test substance, and it is very labor intensive, with the additional caveat of lack of relevance to the human situation. Therefore, a number of in vitro human models have been developed, such as the use of in vitro assays based on expression of specific reporter gene constructs for orphan nuclear receptors such as the pregnane X receptor (Ogg et al., 1997). Although these assays have the potential of screening large numbers of compounds, they run the risk of false negatives because of the multiple mechanisms that may contribute to an induction phenomenon and serve to highlight the importance of hepatocytes as a tool.

Primary human hepatocytes are considered the “gold standard” for in vitro testing of the induction potential of drug candidates (Li et al., 1997; Silva et al., 1998; Kostrubsky et al., 1999). However, several disadvantages are associated with the use of this model system. Hepatocytes for primary cultures have to be isolated from the liver every time they are required. Since human liver availability is limited and unpredictable, planning of experiments is rendered difficult. In addition, it is difficult to assess the quality of hepatocyte preparations that are to be used as primary cultures. One possibility to overcome these difficulties is to develop long-term culture systems that additionally would provide flexibility in the timing of experiments.

Several systems for long-term culture have been described in the last 20 years. These include coculture of hepatocytes with a rat liver epithelial cell line (Guillouzo et al., 1985), the use of extracellular matrix components (Dunn et al., 1991), and the use of a specific additive such as dimethyl sulfoxide (Isom et al., 1985). The long-term culture system used in this study has the advantage of not requiring the use of additive cells or a matrix component preparation and is, therefore, as easy to use as the conventional culture systems. This system has been shown to maintain liver phenotypic characteristics for at least 35 days (Lanford et al., 1989; Ferrini et al., 1997). Previous work using long-term cultured human hepatocytes suggested that human hepatocytes maintained in the modified long-term Lanford's medium for several weeks were suitable for investigating xenobiotic metabolism (Guyomard et al., 1990). The use of long-term cultured hepatocytes to investigate P450 gene expression was previously reported (Pichard-Garcia et al., 2002). In this study we have extended this approach by investigating the usefulness of long-term cultured human hepatocytes as a high-throughput in vitro model to study the induction of P450 in drug discovery. To account for any idiosyncratic differences, hepatocytes obtained from each donor were split into two batches for short- and long-term culture on 96-well collagen-coated plates. CYP1A1, 1A2, 2B6, 2C9, and 3A4 mRNA levels were measured using quantitative PCR after treatment of the hepatocytes for 24 h with a range of prototypical P450 inducers and MRL compounds. The study showed that induction of these P450 isoforms can be measured from as few as 30,000 cells per well. It also demonstrated that long-term culture is a useful tool for investigating P450 induction of new chemical entities (NCEs) in the drug discovery process.

Materials and Methods

Chemicals. Collagen I-coated 96-well supports were obtained from AES (Combourg, France). Williams E culture medium, glutamine, penicillin, and streptomycin for the isolation and culture of the hepatocytes were obtained from Invitrogen (Cergy, France). Bovine insulin and hydrocortisone succinate were obtained from Sigma (St Quentin Fallavier, France), and fetal calf serum was obtained from Dutscher (Brumath, France). Williams E culture medium, l-glutamine, and penicillin/streptomycin for hepatocyte treatment were obtained from Invitrogen (Paisley, UK). Gentamycin, dexamethasone, rifampicin (RIF), phenobarbital, pregnenolone 16α-carbonitrile (PCN), clotrimazole, omeprazole, and sterile dimethyl sulfoxide (DMSO) were obtained from Sigma (Poole, UK). Nucleic acid purification lysis solution was obtained from Applied Biosystems (Warrington, UK).

Isolation of Hepatocytes and Culture Conditions. Human hepatocytes were isolated from surgical wastes by Biopredic (Rennes, France) using the two-step dissociation method as described previously (Guguen-Guillouzo et al., 1986). The cells were seeded on collagen I 96-well coated supports at a density of 0.17 × 10⁶ cells per cm² (Fig. 1). Hepatocytes were plated in Williams medium E supplemented with glutamine, 100 U/ml penicillin, 0.1 mg/ml streptomycin, 1 μM bovine insulin, and 10% (v/v) fetal calf serum (short-term culture medium). Cells were allowed to attach for 24 h, at which time the medium was replaced with either a long-term culture medium (Lanford et al., 1989; Ferrini et al., 1997) or a short-term culture medium composed of Williams E medium supplemented as described above, except that the fetal calf serum was replaced with 5 × 10^–5 M hydrocortisone hemisuccinate. Short-term culture hepatocytes were then shipped at room temperature to the Neuroscience Research Centre (Harlow, UK) for treatment. Long-term culture hepatocytes were maintained in the long-term culture medium for 2 weeks before shipping. Before shipping of the hepatocytes, phenacetin deethylase and nifedipine oxidase activities were measured according to published methods (Wortelboer et al., 1990). Long- and short-term hepatocytes were cultured from three different individuals as shown in Fig. 1.

Hepatocyte Treatment Protocol. Upon arrival of the cells at the Neuroscience Research Centre, the medium was replaced with Williams medium E containing 2 mM l-glutamine, 100 U/ml penicillin, 0.1 mg/ml streptomycin, 50 μg/ml gentamycin, 100 nM dexamethasone, and 1 μM insulin. The cells were maintained in an atmosphere of 95% air and 5% CO₂ at 37°C for an additional 24 h. Hepatocytes were then treated (n = 3 wells/treatment) with prototypical inducers including 10 to 50 μM RIF, 100 to 1000 μM phenobarbital, 5 to 10 μM PCN, 25 to 100 μM dexamethasone, 2.5 to 10 μM clotrimazole, and 25 to 100 μM omeprazole for 24 h. The hepatocytes were also treated with novel Merck compounds (A to E) at concentrations ranging from 0.1 to 10 μM (Fig. 1). All compounds were dissolved in sterile DMSO, which was added to the culture medium at a final concentration of 0.1%. DMSO was added as one treatment to act as a dose vehicle control. At the end of the induction period, the medium was aspirated and replaced with 125 μl/well lysis solution for storage at –80°C before analysis for P450 mRNA levels.

Extraction of Total RNA. Total RNA extraction from hepatocytes was performed in a 6700 Nucleic acid Workstation using the manufacturer's solutions and protocols (PE Applied Biosystems, UK). On extraction, treatment with an absolute RNA wash was included to eliminate DNA and inhibitory proteins.

TaqMan Quantitative RT-PCR. All primers and probes were submitted to the National Center for Biotechnology Information for nucleotide comparison using the basic logarithmic alignment search tool (BLASTn) to search for short, nearly exact sequences to ensure specificity. Primers and probes were synthesized by Applied Biosystems (UK), where probes were 5′- and 3′-labeled with the FAM and TAMRA reporter dyes, respectively. Additional primers and probes were generated as shown in Table 1. The human 18s primer/probe set was purchased from Applied Biosystems (UK) and used according to the manufacturer's instructions. A two-step RT-PCR was conducted by reverse-transcribing an aliquot of total RNA (∼50 ng) to cDNA on the Workstation using the high-capacity cDNA archive kit as per the manufacturer's instructions (PE Applied Biosystems, UK). Using the Workstation, samples were diluted 1:20 and final plates were prepared for analysis on a 9700 real-time quantitative PCR instrument (PE Applied Biosystems, UK).

Primer and probe concentrations were 300 and 200 nM, respectively, for all isoforms. PCR was performed using TaqMan Universal PCR Master Mix and TaqMan ribosomal RNA control kits according to the manufacturer's recommended protocols (PE Applied Biosystems, UK). RT-quantitative PCR was performed using the ABI PRISM 9700 Sequence Detector instrument and Sequence Detector v.2.1.1 software (PE Applied Biosystems, UK). PCR amplification conditions were as follows: 50°C, 2 min, followed by 95°C, 10 min, then 40 cycles at 95°C, 15 s; and 40 cycles at 60°C, 1 min. PCR amplified cDNAs were detected by real-time fluorescence on an ABI PRISM 9700 Sequence Detection System (Applied Biosystems, UK). Quantitation of the target cDNAs in all samples was normalized to ribosomal 18s (18s; Ct_target – Ct_18s =ΔCt), and the effects of each compound on the target cDNA was expressed to the amount in the dimethyl sulfoxide (vehicle) control sample (ΔC_tcompound –ΔCt_DMSO =ΔΔCt). -Fold changes in target gene expression were determined by taking 2 to the power of the ΔΔCt value (2^–ΔΔCt) as per the user bulletin (PE Applied Biosystems, UK).

Statistical Analysis. Results are presented as the mean ± S.D. from three separate experiments. Statistical analysis was performed using Student's t test, accepting p < 0.05 as significant.

Results and Discussion

Xenobiotic metabolism in mammals is a process mediated largely by the cytochrome P450 superfamily of enzymes. Induction of P450 has been shown to have considerable implications for drug therapy, namely, autoinduction and drug-drug interactions (Worboys and Carlile, 2001; Simonsson et al., 2003). Therefore, it is important to understand the induction potential of a new chemical entity early in the drug discovery process.

Here, we described the usefulness and the implementation of a P450 induction assay based on previous studies that used long-term cultured hepatocytes (Pichard-Garcia et al., 2002). The main advantages of this type of approach are that it is a high-throughput assay, which facilitates the investigation of P450 induction issues earlier in the drug discovery process, and it provides flexibility in the timing of the experiments.

The nifedipine oxidase (CYP3A4) and phenacetin deethylase (CYP1A2) activities of both the short-term cultured and long-term cultured hepatocytes were measured in intact cells before treatment with prototypical inducers for quality control purposes (Table 2). There was small interindividual variability in the levels of phenacetin deethylase and nifedipine oxidase activities in both short- and long-term cultured hepatocytes, reflecting the inherent variability within human donors (LeCluyse, 2001). The long-term cultured hepatocytes of each individual showed a decrease in phenacetin deethylase and nifedipine oxidase activities compared with the corresponding short-term cultured hepatocytes, except for individual number 2, who showed the same levels of nifedipine oxidase in both types of culture. However, the activity values obtained for both phenacetin deethylase and nifedipine oxidase in the long-term cultured hepatocytes remained within normal historical ranges of short-term cultured hepatocytes (LeCluyse, 2001; Madan et al., 2003). One of the advantages of using long-term cultured hepatocytes is that the quality control activity measurement could be done by suppliers before shipment of the cells. This could potentially save time and resources to investigators if the human hepatocytes were not good for further studies.

Prototypical inducers of the major P450 isoforms were incubated for 24 h in both short- and long-term cultured hepatocytes isolated from the same donors (Fig. 1). This method compensated for the inherent variability between human donors and, thus, allowed the effect of the long-term culture conditions on P450 induction to be assessed directly. Our results showed that rifampicin, phenobarbital, clotrimazole, dexamethasone, and omeprazole were able to induce CYP3A4 mRNA levels in both short- and long-term cultured hepatocytes as reviewed by Luo et al. (2004) (Fig. 2). Results showed differences in CYP3A4 mRNA -fold changes between short- and long-term cultured hepatocytes for most of the drugs tested in this study, although these differences were not statistically significant (Fig. 2A). An alternative way to report data is to do so as percentage of positive control to avoid the -fold change variability observed in response to the different inducers. Figure 2B shows that when data are expressed as percentage of positive control (RIF at 10 μM), induction values become more comparable between the two types of cultures and individuals. Pregnenolone 16α-carbonitrile was used as a negative control. PCN is known to induce CYP3A1 in rat but does not induce CYP3A4 in humans (Kocarek et al., 1995). As expected, PCN did not cause any induction of CYP3A4 mRNA in either the short-term or long-term cultured hepatocytes.

CYP1A1 and 1A2 mRNA levels were significantly increased in the presence of omeprazole (Fig. 3, A and B). This result was consistent with previous reports in the literature using primary human hepatocytes (Lu and Li, 2001). The level of induction of CYP1A mRNA by omeprazole was not statistically different between the short-term and long-term cultured hepatocytes. None of the other drugs tested in this study induced CYP1A1 and CYP1A2 mRNA levels (data not shown).

Increased interest in the CYP2B6 enzyme has been stimulated by the discovery of polymorphic and ethnic differences in CYP2B6 expression (Lang et al., 2001), identification of additional substrates for CYP2B6, and evidence for cross-regulation with CYP3A4 expression (Makinen et al., 2002). As previously reported, several prototypical CYP3A4 inducers also induced the expression of CYP2B6 mRNA (Gervot et al., 1999; LeCluyse et al., 2000; Sahi et al., 2000; Goodwin et al., 2001). Rifampicin and phenobarbital were shown to be potent inducers of CYP2B6 mRNA in both short- and long-term cultures, with phenobarbital exhibiting a marked dose-response effect (Fig. 3C). Clotrimazole and dexamethasone also showed a dose-response effect resulting in moderate induction of CYP2B6 mRNA in both culture types. Additionally, omeprazole was shown to be a potent inducer of CYP2B6, which, to our knowledge, has not been reported previously in the literature. The ability of omeprazole to potently induce CYP1A and CYP2B6 is perhaps not surprising based on the numerous reports of cross talk between the different nuclear receptors.

CYP2C9 is also important in the metabolism of numerous clinically used drugs (Goldstein and de Morais, 1994; Miners and Birkett, 1998; Goldstein, 2001; Lee et al., 2002). As such, the induction of CYP2C9 mRNA was investigated in this study. Rifampicin, phenobarbital, and dexamethasone were all shown to induce CYP2C9 mRNA in both culture types (Fig. 3D). For phenobarbital and dexamethasone, the CYP2C9 mRNA -fold change was equivalent in both the short- and long-term cultured hepatocytes. However, rifampicin showed a greater -fold change in the short-term cultured hepatocytes compared with the long-term cultured hepatocytes at 25 μM. The magnitude of the -fold change values of CYP2C9 mRNA after treatment with prototypical inducers was very small, ranging from approximately 1.5- to 3.5-fold. This is consistent with previous reports, which suggest that the small -fold change values are attributed to high basal CYP2C9 mRNA levels (Chen et al., 2004).

A variable induction response between hepatocytes from different individuals was observed in this study. This variation has previously been described and is believed to be due to a variety of factors including hepatocyte quality, together with health, diet, medication, polymorphisms, and basal mRNA expression of the donor (Kostrubsky et al., 1998; Silva et al., 1998; LeCluyse et al., 2000). A method to circumvent interindividual variability is to report percentage induction relative to a standard inducer, with the standard inducer included in every experiment (Silva et al., 1998). Our results from this study showed that variability between individuals was greatly reduced by expressing data as percentage induction rather than -fold induction. Therefore, this method would enable us to compare the induction potential of NCEs in hepatocytes from different individuals.

To demonstrate the utility of the long-term cultured hepatocytes for induction screening of drug candidates, we investigated the induction potential of five MRL compounds in both the short- and long-term cultured hepatocytes. The induction liability of these MRL compounds was expressed as percentage of positive control as shown in Fig. 4. Both culture types allowed the test compounds to be ranked in the same order. For compounds that exhibited a dose response in the short-term cultures, this finding was also observed in the long-term cultured hepatocytes. In addition, MRL-E, dosed at 10 μM, showed a decrease in CYP3A4 and 18s mRNA levels in both culture types, which was probably due to cell toxicity (data not shown). The same MRL compounds had also been tested in different short-term cultured hepatocytes from different sources, and a similar percentage of positive control was reported (data not shown).

In summary, our results showed that induction of a range of P450 isoforms can be measured in primary human hepatocytes cultured on a 96-well plate with 30,000 cells/well only. The long-term cultured hepatocytes have proven to be comparable to conventional culture systems with the addition of several advantages. Long-term cultured hepatocytes can be preserved from previous isolations, meaning that a fairly regular supply to the researcher can be maintained. The quality of long-term cultured hepatocytes can also be assessed before use, saving time and resources to investigators. This quality assessment, combined with the ability to culture hepatocytes in a 96-well format, and using TaqMan RT-PCR analysis, demonstrated that long-term cultured hepatocytes are an ideal tool for the assessment of induction potential of NCEs in a drug discovery environment.