Abstract
Assessing the inducibility of CYP3A4 by various xenobiotics can predict potential drug interactions. In the present investigation, human hepatoma cells were stably integrated with either the CYP3A4 enhancer region and a luciferase reporter gene or the CYP3A4-luciferase construct and the human pregnane X receptor (PXR). Several colonies containing one to three copies of luciferase per cell were identified by Southern blot analysis. Those transformants producing high luciferase activity in response to rifampicin were used to standardize a 96-well plate screening system with minimal inter- and intraplate variability. Standardization also consisted of assessing viability of cells cultured in medium containing various serum concentrations. In cells maintained for 48 h in medium with less than 5% serum, a significant (p < 0.01) decline was observed in viability accompanied by altered induction. A defined serum-free medium also produced less viable cells but did not alter the inductive response. Treatment of transformants with various concentrations of rifampicin produced a dose-response curve with maximal induction at 10 μM (5.6 ± 0.18- and 2.1 ± 0.3-fold above dimethyl sulfoxide (DMSO)-treated cells in transformants with and without PXR, respectively). Of additional agents examined for their ability to induce CYP3A4, omeprazole (200 μM) was the most potent inducer (12.8 ± 1.9- and 2.4 ± 0.2-fold above DMSO-treated cells in transformants with and without PXR, respectively). Mifepristone and mevastatin produced modest induction (∼3-fold) in the cell line containing exogenous PXR, but produced less than 1.2-fold increases in cells lacking PXR. Thus, only potent inducers can be identified in the cell line without PXR. In contrast, cells containing the receptor can be used to rank CYP3A4 induction. Because a high volume of chemicals can be readily and accurately screened for their ability to induce CYP3A4 with this format, such a system could be valuable in the initial stages of preclinical drug development.
One of the most abundant P450 enzymes present in human liver is CYP3A4, which can account for 30 to 60% of hepatic P450s, depending on the individual (Wrighton et al., 2000). Due to this wide range of hepatic content of CYP3A4, variability in metabolism occurs that can lead to unpredictable therapeutic outcomes. Individual differences in expression of hepatic CYP3A4 can be caused by several factors, including prolonged exposure to certain xenobiotics. As a result of elevated CYP3A4 concentrations, therapeutics metabolized by this P450 exhibit either lower or higher efficacy if metabolism results in inactivation or activation, respectively, of the substrate. Therefore, it is important to individualize drug therapies by identifying agents possessing the ability to induce CYP3A4. Unfortunately, it is difficult to predict which agents will induce this P450 because its expression is enhanced by a variety of structurally unrelated chemicals. For this reason, it is important to test new drug candidates in biological systems for their ability to increase expression of CYP3A4 and other drug-metabolizing enzymes. Assessing the inducibility of candidates before marketing can alleviate the liability associated with these types of drug interactions that lead to either a lack of efficacy or toxicity.
Recent studies have demonstrated that activation of CYP3A4 by certain xenobiotics results from orphan nuclear receptors, including the pregnane X receptor (PXR; NR1I2) (Kliewer et al., 1998a,b; Lehmann et al., 1998). This receptor interacts with certain chemicals and binds to specific DNA elements in the CYP3A genes (Barwick et al., 1996; Huss et al., 1999). These elements contain two nuclear receptor half-sites of the consensus sequence AGTTCA, organized as either a direct repeat with a three-nucleotide spacer or an everted repeat with a spacer of six nucleotides. To induce CYP3A4, PXR interacts with a number of natural and synthetic steroids such as dexamethasone, antiglucocorticoids such as pregnenolone 16α-carbonitrile (PCN) and mifepristone (RU486), and the macrolide antibiotic rifampicin (Desai et al., 2002; Jones et al., 2000). In addition to these agents, chemicals that activate PXR include the antifungal clotrimazole, the antidiabetic troglitazone, the human immunodeficiency virus protease inhibitor ritonavir, and the herbal St. John's wort (Kliewer and Willson, 2002). The receptor also plays a role in the activation of other genes, including the drug transporters MDR1 (Synold et al., 2001) and MRP2 (Kast et al., 2002). In addition to activating several genes, PXR is predominately expressed in liver and intestine where the majority of drug metabolism occurs and is present in all mammals examined to date (Lehmann et al., 1998). Among those studied, rat, rabbit, mouse, and human PXR share >90% amino acid identity in the DNA binding domain. However, there is marked divergence in the ligand binding region, which may account for species differences in response to various agents such as PCN and rifampicin (Barwick et al., 1996;Kliewer and Willson, 2002). Rodent CYP3A and not rabbit or human responds to PCN, whereas human and rabbit CYP3A, but not rat, responds to rifampicin.
Because CYP3A4 exhibits species differences with regard to induction, testing drug candidates for their ability to induce this enzyme is generally performed by the use of in vitro human systems. The availability of primary cultures of human hepatocytes has afforded the pharmaceutical industry the ability to obtain clinically relevant in vitro drug interaction data. There are however, many disadvantages associated with the use of primary cultures, including difficulty in obtaining liver specimens and significant variability among samples and culture conditions. Thus, other in vitro systems are being investigated. Transcriptional activation has been performed in vitro for a number of years to investigate changes in gene expression of P450 enzymes by chemicals (Plant et al., 2000). The most common type is use of transient transfections of a reporter gene construct into a suitable cell line. However, within this system there may be biological and experimental variation. More recently, high-throughput PXR activation and ligand binding assays to identify drug candidates that induce CYP3A gene expression have been developed (Moore and Kliewer, 2000). This system provides a robust assay with readily interpretable results, but it does not assess function and is limited to chemicals that bind PXR.
The in vitro system described herein is also designed to detect induction of the human CYP3A4. The technique detects transcriptional activation by foreign compounds of the CYP3A4 enhancer and a reporter gene that have been stably transfected into human hepatoma cells. It is a standardized microtiter plate assay and results can be obtained within 3 days. The advantage of this in vitro transcription system over isolated human hepatocytes or transient transfections lies in its consistency and reproducibility. Furthermore, interindividual variability, culture conditions, and transfection efficiencies that can influence results and add to varied responses are eliminated from this system. It is a high-volume screening system using 96-well plates and will accurately predict induction of the CYP3A4 gene through its proximal enhancer. Using this system could reduce the number of compounds tested in human hepatocytes, thereby reducing cost and time associated with drug discovery. Furthermore, the potential for drug interactions would be identified at the initial phases of the development process.
Materials and Methods
Construction of Plasmids for Transfections.
The full-length coding region of human PXR was derived by reverse transcription-polymerase chain reaction from RNA obtained from a human liver sample (UC9203). The forward and reverse oligonucleotide sequences were 5′-ATG GAG GTG AGA CCC AAA GAA-3′ and 5′-CTC AGC TAC CTG TGA TGC CGA-3′, respectively. The polymerase chain reaction conditions consisted of denaturation at 94°C for 4 min, followed by 30 cycles of 94°C for 45 s, 55°C for 1 min, and 72°C for 2 min with a final extension at 72°C for 7 min. The 1300-bp amplified product was cloned into pCR2.1 (Invitrogen, Carlsbad, CA) and subjected to sequence analysis. The sequences obtained agreed over the entire coding region with that described previously (Lehmann et al., 1998). The hPXR cDNA was then excised from pCR2.1 by digestion with BamHI andNotI and cloned into analogous sites of a pIRESneo vector (CLONTECH, Palo Alto, CA) containing a neomycin selection cassette.
Forward and reverse primers were made to a 5′-flanking region of CYP3A4, known to contain the pregnane X receptor response element (PXRE) (−568 to −110 bp) (Quattrochi et al., 1995). The forward and reverse oligonucleotide sequences were 5′-GAG CTC ACC TCT GTT CAG GGA AA-3′ and 5′-CAC CTT GGA AGT TGG C-3′, respectively. This 458-bp region was amplified by polymerase chain reaction from genomic DNA isolated from a sample of human liver (UC9203). The amplifier was cloned into pCR2.1 and sequenced. The enhancer region was then liberated from pCR2.1 with EcoRI, blunt-ended, and subsequently cloned into the SmaI site of the pGL3-promoter (simian virus 40) vector (Promega, Madison, WI). A double stranded oligonucleotide (75 bp) containing an EcoRI site flanked by two additional PXREs with XhoI sites on the 5′ and 3′ ends was synthesized and inserted into the XhoI site at the 3′ end of the CYP3A4 enhancer. Sequence analysis of the entire enhancer region containing the 75-bp oligonucleotide verified its identity (Quattrochi et al., 1995) and that the oligonucleotide had been inserted.
Generation of Stable Cell Lines Expressing CYP3A4 Response Elements.
HepG2 cells were harvested at approximately 50% confluence and seeded in six-well dishes at 5 × 105 cells/well in DMEM containing 10% FBS. After 24 h recovery, cells were transfected with the following combinations: CYP3A4 enhancer/pGL3 promoter (p3A4) and PXR/pIRESneo (pPXR) at a ratio of 5:1 (6 μg of total DNA/well), p3A4 and pIRESneo (5:1 ratio, 6 μg of DNA/well), and pGL3 promoter and pPXR (5:1 ratio, 6 μg of DNA/well) using the calcium phosphate coprecipitation procedure (Ausubel et al., 1990). The control cells were those that received plasmid DNA containing pGL3 promoter + pPXR. After 16 h of exposure to the precipitated DNA, the culture medium was removed, cells washed twice with DMEM, and fresh media containing 10% FBS were added. After an additional 24 h, media were replaced with that containing 400 μg/ml geneticin (G418; Invitrogen). Media were changed every 2 days for 3 weeks, to ensure that an adequate concentration of G418 was present in the medium, until small colonies were visible. These colonies were then tested for luciferase activity and positive colonies randomly selected. A microplate dilution technique was performed to purify the selected transformants. Confluent T75 flasks of the purified colonies were trypsinized and used to seed 96-well plates to measure rifampicin-induced luciferase response of individual clones to test for the presence of recombinants as described below.
Luciferase Assays.
Luciferase assays were performed as specified by the manufacturer (LucLite system; Packard Bioscience, Meriden, CT). Activity was determined using the Lumistar galaxy luminometer (BMG Labtechnologies, Offenburg, Germany), and results were expressed as relative light units or fold increase above control (DMSO-treated cells).
Treatment of Transformed Cultures.
The HepG2-derived transformed cell lines were grown as monolayers in medium comprised of DMEM (Invitrogen), 50 U/ml penicillin, 100 μg/ml streptomycin, 0.1 mM nonessential amino acids (Invitrogen), 400 μg/ml G418 (Invitrogen), and 10% FBS (Hyclone Laboratories, Logan, UT) and maintained in an atmosphere of 5% CO2 and 95% air at 37°C. Those lacking exogenous PXR also received 10−7 M dexamethasone in the media (Pascussi et al., 2000). Cells were seeded in T75 flasks and grown to confluence and passage was limited to 30. Upon 75% confluence, cells were removed from flasks by trypsinization and replated in 96-well plates at a density of 7.5 × 104 cells/well in DMEM medium containing 10% FBS and G418. After a 3-h recovery, medium was replaced with fresh medium that was serum free (HMM; Clonetics, San Diego, CA) (Raucy et al., 2002), DMEM containing 10% dextran-treated FBS, or various amounts of dextran-treated FBS ranging from 0.1 to 5% and cells treated with 0.1% DMSO (control) or inducer dissolved in DMSO at various concentrations. That cells contained the p3A4 was verified by comparing results with control cells transfected with pPXR and pGL3 promoter. Screening for cells containing the pGL3/3A4 enhancer was performed by treatment with 10 μM rifampicin and 0.1% DMSO. Those cells exhibiting greater than 3-fold increases in luciferase activity above cells containing control plasmids were considered transformed with the correct plasmids and were used for further testing or frozen in liquid nitrogen for future use. Finally, copy number of p3A4 integrated into the genome of the HepG2 cells was verified by Southern blot analysis as described below.
To test the cell lines considered positive for the pPXR and p3A4 or the p3A4 alone, time-course studies were performed. Cells were treated with 10 μM rifampicin for 24 to 72 h, with analysis of response determined at 24-h intervals. In addition, dose-response curves were constructed for various known CYP3A4 inducers to assess sensitivity of the cell lines and a noninducer of CYP3A4, TCDD, to confirm the specificity of the response element. The dose-response curves consisted of concentrations ranging from 0.1 to 200 μM, depending on the agent. The compounds tested were mifepristone (RU486; BIOMOL Research Laboratories, Plymouth Meeting, PA), mevastatin (BIOMOL Research Laboratories), rifampicin (Sigma-Aldrich, St. Louis, MO), omeprazole (Astra, Hässle, Mölndal, Sweden) and TCDD (Chemsyn Science Laboratories, Lenexa, KY). Cells were exposed to each compound for 48 h. All agents were dissolved in sterile DMSO (Sigma-Aldrich) and this solvent was added to control cells at 0.1%.
RNA Isolation and Northern Blot Analysis.
Total RNA from transformed HepG2 cells, human hepatocytes, or liver was isolated using TRIzol reagent (Invitrogen) or QIAGEN column purification (QIAGEN, Valencia, CA) and quantified by measuring absorbance at 260 nm; purity was assessed by determining the 260/280-nm ratio and integrity of the 28S and 18S bands on agarose gels. The human liver sample was from subject UC9203 and the hepatocytes were from subject 928. The latter subject (928) was a Caucasian male, 53 years old with a history of smoking one pack of cigarettes per day and a drug history that included sertraline. Subject UC9203 was a 24-year-old Caucasian female with no drug, alcohol, or tobacco history. Northern blot analysis was performed by electrophoresis of total RNA (10 μg) through a 1% agarose-2.2 M formaldehyde gel, followed by blotting onto a nylon membrane (Qiabrane; QIAGEN) (Shih et al., 1999). RNA was cross-linked to the membranes using a UV Crosslinker (Stratagene, La Jolla, CA) and the membranes hybridized to random-primed cDNA probe encoding human CYP3A4 or PXR. The cDNA probe for human CYP3A4 spanned a 263-bp region between 110 and 383 bp and that for PXR was 367 bp in length and analogous to a region between 990 to 1357 bp of the full-length clone. A probe for human 18S rRNA probe (Ambion, Austin, TX) was used to normalize for RNA quantity. Hybridization of blots was performed as described previously (Allen et al., 2001). Autoradiographs of Northern blots were quantified by scanning autoradiograms with a ScanMaker II (Microtek, Redondo Beach, CA) and digitizing with Un-Scan-It software (Silk Scientific, Orem, UT). Exposure times used were in the linear range for the film Kodak XAR-5 (Sigma-Aldrich).
Southern Blot Analysis.
The number of copies of each plasmid DNA incorporated into G418-resistant colonies was determined by Southern blot analysis of genomic DNA isolated from HepG2 cells and several transfectants. Total cellular DNA was extracted from HepG2 cells using the Genomic DNA preparation system (QIAGEN), and 15 μg was digested for 36 h with the restriction enzyme MfeI. A segment of luciferase DNA (456 bp) spanning 203 to 659 bp was used to establish a standard curve and as a probe for determining plasmid DNA incorporation. For the standard curve, digested DNA from nontransformed HepG2 cells was mixed with various amounts, 1.1 to 33.3 pg, of the luciferase cDNA fragment and loaded onto a 0.8% agarose gel. A 100-bp DNA marker and Mfe I digested DNA (15 μg) isolated from several transfectants were also subjected to agarose gel electrophoresis. The gel was then rinsed in 0.1 M HCl for 2 × 15 min, washed in distilled water, placed in denaturing buffer (1.0 M NaCl and 0.5 M NaOH) for 2 × 30 min, and rinsed again with distilled water. After this rinse the gel was exposed to a neutralizing buffer (1 M Tris pH 8.0 and 0.5 M NaCl) for 2 × 30 min. The DNA samples were then transferred onto a nylon membrane (Brightstar plus; Ambion) overnight in 20× standard saline citrate. After transfer, the membrane was hybridized with the 32P-labeled luciferase fragment as described above for Northern blots and exposed to Kodak XAR-5 film (Sigma-Aldrich). The developed film was scanned and quantified and the number of plasmid copies incorporated into the transformants determined from the standard curve of luciferase fragments.
Viability and Protein Determination of HepG2 Cells in Microtiter Plate.
Protein quantification was by a modification of a method described previously (Ziccardi et al., 2000) using bovine serum albumin as a standard. The bovine serum albumin was added to a row in the microtiter plates in varying amounts ranging from 15 to 105 μg/well. Fluorescamine (500 μg/ml) (Molecular Probes, Eugene, OR) in 100 μl of acetonitrile was added to each well. Fluorescence was quantified using a microtiter plate fluorometer (Packard Bioscience) set at an excitation of 400 nm and an emission of 460 nm. Results were calculated from the standard curve.
Cellular viability was assessed by using CellTiter-Glo Luminescent Cell Viability assay (Promega). This assay is designed for use with multiwell plate formats and detects a luminescent signal proportional to the amount of cellular ATP that is present. The amount of ATP is directly proportional to the number of viable cells in culture. Briefly, 100 μl of CellTiter-Glo reagent was added to each well and contents mixed for 2 min. The plates are then incubated at ambient temperatures for 10 min and luminescence measured.
Data Analysis.
Student's t test was used for the statistical analysis of data. Statistical significance was defined at a level of p < 0.01. Data are expressed as the mean ± S.D.
Results
Stable cell lines were developed by transforming the plasmids, p3A4, pPXR, and control vectors into HepG2 cells and selecting for G418 resistance. After a 6- to 8-week period, resistant colonies were identified for p3A4, p3A4 + pPXR, and pPXR + control vector, pGL3 promoter. Randomly selected colonies were tested for basal or inducible luciferase activity by treatment with 0.1% DMSO or 10 μM rifampicin in DMSO. Transfection of p3A4 into HepG2 cells, followed by G418 selection, resulted in the isolation of several G418-resistant colonies, of which 13 were tested for basal or inducible luciferase activity. Eleven G418-resistant colonies were able to support basal-level luciferase expression and six colonies supported inducer-mediated luciferase activity when treated with 10 μM rifampicin for 48 or 72 h. Of the 40 colonies screened that contained the stably integrated p3A4 and pPXR plasmids, 36 displayed inducible luciferase activity when treated with rifampicin. Three control antibiotic-resistant colonies harboring the pPXR + pGL3 plasmids exhibited basal level luciferase activity. After microplate dilution experiments to purify clones, we chose the p3A4 + pPXR and p3A4 transformants containing the highest inducible luciferase activity for further studies and these cell lines were referred to as C1F1 and C13-2 cells, respectively. Southern blot analysis of total cellular DNA from several transformants, including the latter two colonies, confirmed the presence of stably integrated p3A4 (Fig.1) with copy numbers ranging from one to three. The C1F1 cell line exhibited one copy per cell of the p3A4 plasmid, and C13-2 cells had two copies per cell of the luciferase-containing CYP3A4 enhancer plasmid.
Validation that PXR was stably integrated into cells receiving this plasmid was by Northern blot analysis. Blots containing total RNA from human liver, human hepatocytes treated with DMSO or rifampicin, and from the C1F1 and C13-2 cell lines were probed with a PXR cDNA. All samples displayed two transcripts at 2.6 and 4.3 kb (Fig.2), consistent with previous observations (Lehmann et al., 1998). In the liver sample, the 2.6-kb transcript was more abundant than that at 4.3 kb. Similarly, the C13-2 cells exhibited greater concentrations of the 2.6- than the 4.3-kb transcript with the former being 50% of that in the liver sample. Hepatocyte expression of the smaller transcript was also less than that in human liver (15% of hepatic levels) but unlike liver, the 4.3-kb transcript in hepatocytes was more abundant. Treatment of hepatocytes with 10 μM rifampicin did not increase either transcript above levels observed in DMSO-treated cells. In the C1F1 cell line, concentrations of the 2.6-kb transcript were similar to those in liver or C13-2 cells, whereas the 4.3-kb mRNA was highly expressed (7-fold) compared with RNA from C13-2 cells (Fig.2). Overexpression of the larger transcript in C1F1 cells was due to both exogenous PXR (1300 bp) and also to neomycin phosphotransferase (antibiotic resistance marker, 3000 bp). Because the pIRES vector containing the PXR was designed to contain an internal ribosome entry site of the encephalomyocarditis virus, it permits the translation of two open reading frames from one mRNA, hence, the larger band. However, when translated, two separate proteins were synthesized, PXR and neomycin phosphotransferase. Coincidentally, the overexpressed PXR and antibiotic resistance marker mRNA migrated in the same region as the 4.3-kb PXR transcript observed in hepatocytes or liver. Functionality of the stably integrated PXR was determined by treating several PXR cell lines with 10 μM rifampicin and measuring HepG2 CYP3A4 mRNA levels. For all PXR containing colonies tested, HepG2 CYP3A4 mRNA was 200 to 800% of that in DMSO-treated cells (data not shown).
Initial experiments were performed to determine optimal inducer exposure periods. In addition, the recovery time, i.e., that period between cell plating and inducer treatment, was assessed. Immediately after plating, luciferase activities were variable and background levels (DMSO-treated cells) were higher in either stable transformant, suggesting that cells were responding to stress associated with trypsinization and replating. However, 12 to 24 h after plating and 24 to 72 h of DMSO exposure produced lower luciferase activities that were within 10 to 15% variation. For the time-response curves, exposure to 10 μM rifampicin ranged from 0 to 72 h and were performed in both 24- and 96-well culture dishes. Curves generated for rifampicin induction were found to be similar in 24- or 96-well plates for either the C13-2 or C1F1 cell lines. In either transformant or culture plate format, rifampicin (10 μM) produced increases in luciferase activity at 24 and 48 h of exposure in a time-dependent manner and activity was maximal at 48 h (data not shown). However at 72 h, induction began to decline. We also tested the effect of DMSO and rifampicin exposure at various times on basal-level luciferase activity using a control transformant containing the pPXR + pGL3 promoter vectors. Variable exposure times to DMSO or rifampicin did not alter luciferase activities in the cells harboring control vectors. Based on these time course results, all subsequent experiments used 12-h recoveries and 48-h exposures to inducers.
Because serum constituents can vary among manufacturers and even between different lots of FBS and certain components, including hormones, may affect PXR-mediated transcriptional activation of CYP3A4, further characterization consisted of testing lower concentrations of serum in DMEM (ranging from 0.1 to 10%). A specifically defined medium used for maintenance of human hepatocytes (HMM) was also used here as the serum-free (SF) medium (Runge et al., 2000). Cell viability was assessed by determining mitochondrial ATP levels, and comparisons were made between medium in the presence of various concentrations of serum and in SF medium. We found that by adjusting serum concentrations below 5% FBS in DMEM, C13-2, and C1F1 cells treated with DMSO or media alone exhibited a substantial decline in viability by 60 h in culture (12-h recovery and 48-h exposure). Indeed, when serum concentrations were 0.1% in DMEM, viability was low with ATP values 6% (9,689 ± 443 relative light units) of those at 10% FBS in DMEM (161,489 ± 7,382 relative light units) (Table 1). At a serum concentration of 5%, ATP values indicated that viability in DMSO-treated cells or those exposed to medium only was the same (100%) as that at 10% FBS in DMEM. At 0.5% serum in DMEM, viability was significantly less (p < 0.01) than that at 10% serum with ATP values 80% of those at 10%. Thus, culturing stably transformed HepG2 cells in serum concentrations below 5% causes significant cell death. This cell death was accompanied by alterations in response to the inducer rifampicin. Surprisingly, rifampicin or omeprazole stabilized cells against declines in viability. Only those cells exposed to DMSO or media only (control) were less viable and produced minimum luciferase activity. This decline in luciferase light units in control cells produced artificially high induction (10–80-fold). In addition, greater well-to-well variability occurred in the DMSO treated cells and was dependent on the number of viable cells remaining in each well. Thus, lowering serum concentrations below 5% can produce greater induction but the induction is highly dependent upon the ability of the inducer to protect against cell death.
Results obtained in the SF media (HMM) were also compared with those in medium containing 10% FBS (Table 1). After 60 h in culture, 12-h recovery, and 48-h treatment, viability was significantly (n = 12, p < 0.01) better (16%) for the C1F1 cells in medium containing 10% serum (131,848 ± 5,013 RLUs) than in SF medium (110,133 ± 4,446 RLUs). For the C13-2 cells, viability was also significantly (n = 12,p < 0.01) better (16%) for those cultured in the presence of 10% FBS (161,489 ± 7,382 RLUs) than in SF medium (135,212 ± 7,308 RLUs). Interestingly, in both cell lines, rifampicin-mediated induction was slightly higher in SF medium (Fig.3A). Indeed, in the C1F1 and C13-2 cells rifampicin-mediated induction in HMM was 5.97 ± 0.3- and 2.08 ± 0.3-fold above DMSO-treated cells, respectively, compared with 5.56 ± 0.18- and 1. 67 ± 0.1-fold in C1F1 and C13-2 cells cultured in DMEM containing 10% FBS, respectively. However, the difference in rifampicin-mediated induction between SF medium and that containing 10% FBS was not significantly different (p< 0.01) in either the C13-2 or C1F1 cells. As a negative control, TCDD (1 nM) was examined in both cell lines and in the SF and 10% FBS-containing media (Fig. 3B). TCDD did not increase luciferase activity in either medium or cell line.
A factor that might cause variability between plates (interplate error) and within a single plate (intraplate error) is differing cell numbers. Given this potential for variation, a method was established to standardize the assay and maximize reliability. First, the number of C13-2 and C1F1 cells cultured in DMEM with 10% FBS added to each well was varied to optimize the assay. Increasing cell numbers ranging from 3.8 × 103 to 105 were added to each well of a 96-well microtiter plate and exposed to 10 μM rifampicin or 0.1% DMSO for 48 h. With increasing cell number above 50 μl/well, there was also an increase in rifampicin-mediated luciferase activity accompanied by an enhanced activity in DMSO-treated cells. However, at a cell density of 5.7 × 104 cells/well (150 μl/well) there was a greater fold increase in the rifampicin-treated cells compared with those treated with 0.1% DMSO. Moreover, 150 μl/well produced signals that were on the linear portion of the curve (data not shown).
Cellular proliferation after 60 h in culture could result in intraplate variability; therefore, protein concentrations in each well were determined postanalysis. Protein quantification revealed that equivalent amounts of cellular protein (50.6 μg/well ± 6.6) were present in each well after the 48-h exposure period, which consisted of a medium change at 24 h. To correct for interplate variability, a normalization correction factor was established for rifampicin induction. This factor was generated from the measurement of induction by 10 μM rifampicin in 15 separate analyses using the standardized conditions and is the mean fold increase above DMSO-treated cells. In medium containing 10% FBS, the correction factor was set at 1.7-fold increase above DMSO-treated cells for C13-2 cells and 5.6-fold for the C1F1 cell line. In SF media the normalization factor for 10 μM rifampicin in the C1F1 cells was set at 6-fold above DMSO-treated cells and in C13-2 cells the factor was 2.1-fold greater than control (DMSO-treated) cells.
Using the 96-well plate high throughput format and standardized conditions including 48-h exposures in SF medium and that containing 10% serum, we examined known inducers of CYP3A4, namely, rifampicin, mifepristone, and omeprazole (OMP). Dose-response curves were generated for each agent in addition to a noninducer, TCDD, and an agent that had not previously been shown to be a CYP3A4 inducer, mevastatin. Both cell lines, C1F1 and C13-2, were examined for their inductive properties. Enhanced luciferase activity above DMSO-treated cells occurred at doses as low as 10 μM OMP and increased in a dose-dependent manner up to 200 μM in either cell line (Fig. 4). In SF and 10% media, omeprazole (200 μM) produced a 16.3 ± 2.9- and 12.8 ± 1.9-fold difference, respectively, in C1F1 cells between OMP-treated and DMSO-treated cells (Fig. 4A). In C13-2 cells, the magnitude of induction by 200 μM OMP was 2.97 ± 0.2-fold above DMSO-treated cells in SF medium, whereas that in the medium containing 10% FBS was 2.38 ± 0.2-fold above control (Fig. 4B). Less induction occurred in medium with FBS for both cell lines. Also noted in either cell line and in the absence of inducer, was a 2-fold higher luciferase activity in both cell lines cultured in the SF medium when compared with those in 10% FBS in DMEM. Thus, promoter activation occurred by constituents present in the SF medium. However, this activation only slightly affected the fold induction for each agent tested and this affect was not significant (p < 0.01).
Rifampicin also induced luciferase activity in both cell lines in a dose-dependent manner up to 10 μM and then declined at 50 μM, and the difference between treated and DMSO-treated cells was slightly greater in SF medium (Fig. 5). In C13-2 cells (Fig. 5B), rifampicin-mediated induction at 10 μM was 2.06 ± 0.5-fold above DMSO-treated cells in SF medium (n = 3), whereas that in C1F1 cells (Fig. 5A) in the same media were 5.5 ± 0.7-fold above control. When a dose-response curve was generated for mifepristone in the two cell lines, we found that in the C13-2 cells, this agent did not produce induction in SF medium and only slightly increased luciferase activity at 5 and 10 μM (1.2 ± 0.07-fold above control) in DMEM with 10% FBS. In C1F1 cells, mifepristone produced the greatest induction in medium containing serum, but even at the highest concentration (10 μM), only a 3.72 ± 0.23-fold increase above control was noted (Fig.6). TCDD produced minimal luciferase activity in either cell line, < 1.1-fold above control, over the concentration range examined herein (0.1, 1, and 2 nM) (data not shown). Interestingly, an agent not previously identified as a CYP3A4 inducer, mevastatin, also increased CYP3A4-mediated luciferase activity in C13-2 cells cultured in SF medium only and at 50 μM was 1.3 ± 0.2-fold above DMSO (Fig. 7). However, in the same cell line cultured in DMEM with 10% FBS, increases above DMSO-treated cells were not observed. The C1F1 cell line also exhibited an increase in luciferase activity in SF medium (2.7 ± 0.6-fold above control, n = 3) at 50 μM mevastatin and in DMEM with 10% FBS (2.8 ± 1.1-fold above control, n = 3). Based on results obtained with these cell lines, both mevastatin and mifepristone are moderate CYP3A4 inducers, whereas OMP and rifampicin are more potent inducers. These results are consistent with those obtained in primary cultures of human hepatocytes (J. L. Raucy, unpublished observations).
Discussion
In the present investigation, we examined the response of the human CYP3A4 enhancer to chemical inducers. Plasmids were generated using the promoter region (Quattrochi et al., 1995, 1998; Barwick et al., 1996) and two additional PXREs fused to the luciferase gene as a reporter for transcriptional activation. These plasmids were stably integrated into HepG2 cells. This approach permits specificity of induction because the response measured, luciferase activity, is the result of transcription from the human CYP3A4 enhancer. The results of this integration were cell lines that could be effectively used in a standardized HTS to identify potential inducers of CYP3A4. Standardization of the HTS consisted of testing various serum concentrations in DMEM and also a defined serum-free media (HMM). Several known CYP3A4 inducers were tested in the system. In addition, we demonstrated that human PXR mRNA exhibited two transcripts with the most prominent one, 2.6 kb, being expressed in liver and HepG2 cells. These results agree with previous findings (Lehmann et al., 1998) that demonstrate a prominent 2.6-kb transcript in human tissues with less abundant messages at 4.3 and 5 kb.
Stable human hepatoma cell lines that were engineered to predict inducibility of CYP3A4 included cells with and without exogenous PXR. We found that those lacking exogenous PXR, C13-2 cells, also responded to known CYP3A4 inducers but with more modest increases. This is not the first report to demonstrate that HepG2 cells support CYP3A4 response element-driven luciferase activity. Indeed, the xenobiotic response element module (XREM) of CYP3A4 produced activation of the luciferase reporter in transiently transfected HepG2 cells in the absence of exogenous PXR (Goodwin et al., 1999). We also found that C13-2 cells produced dose-response curves for two inducers tested herein, namely, rifampicin and omeprazole, that were similar to those curves produced in the C1F1 cells (Figs. 4 and 5). Characterization of both stably integrated cell lines, C1F1 and C13-2, revealed that treatment by the inducers, including omeprazole, mimicked induction of CYP3A4 in cultures of human hepatocytes (Pichard et al., 1990; Daujat et al., 1991; Savas et al., 1999). Indeed, agents examined herein enhanced hepatocyte CYP3A4 mRNA accumulation and produced dose-response curves that were similar to those generated in the C1F1 and C13-2 cells (J. L. Raucy, unpublished observations). Moreover, those that were potent activators of luciferase were also potent inducers of CYP3A4 mRNA in primary cultures while those that were moderate inducers in the cell lines were moderate inducers of CYP3A4 in human hepatocytes.
Additional similarities between the two cell lines existed and stemmed from their ability to be used in a standardized high-throughput system. Indeed, a 96-well plate format using either cell line exhibited minimal (≤15%) inter- and intraplate variability, demonstrating the feasibility of this system to examine the effects of many different chemicals on CYP3A4 induction in a single assay. Another approach to such a system was developed that involved screening chemicals for induction by assessing their ability to bind the PXR ligand binding domain (Jones et al., 2000; Moore and Kliewer, 2000; Moore et al., 2000). Although this is effective, there are limitations associated with this type of assay. Both agonists and antagonists can bind to a receptor. However, the antagonist may not exhibit an expected biological response. An example of such an interaction is the ligand α-naphthoflavone, which binds the aryl hydrocarbon receptor with high affinity but cannot initiate receptor transformation and nuclear localization (Gasiewicz and Rucci, 1991; Merchant et al., 1992; Henry et al., 1999). Therefore this aryl hydrocarbon receptor ligand does not induce CYP1A1 or CYP1A2. Thus, one drawback to receptor binding assays is that they may not dependably predict induction of CYP3A4. An additional limitation associated with the ligand binding assay is that only chemicals having direct interaction with PXR are considered inducers of CYP3A4 (Jones et al., 2000). This would eliminate drugs such as phenobarbital that can induce CYP3A4 through the PXR response element but in an indirect manner, namely, through constitutive androstane receptor (CAR) (Xie et al., 2000). Alternatively, another advantage to the cell lines described in this report is the “cross talk” between CAR and PXR (Xie et al., 2000), allowing chemicals that induce CYP3A4 through either receptor to be identified. Indeed, HepG2 cells contain appreciable levels of CAR (Moore et al., 2002) and therefore, those that harbor the CYP3A4 enhancer such as the C1F1 or C13-2 cells, can be used to identify activators of this receptor. We found that treatment of the C1F1 and C13-2 cells with 1 mM phenobarbital enhanced luciferase activity 9.1- and 7.7-fold above DMSO-treated cells, respectively (J. L. Raucy, unpublished observations).
The primary difference between the two cell lines developed in this investigation was that those cells without exogenous PXR, C13-2, have limited utility. This cell line may be used to identify potent CYP3A4 inducers, including rifampicin or omeprazole. Both agents produced 1.7-fold or greater luciferase activities over DMSO-treated cells. However, the C13-2 cells may not identify weak or moderate inducers such as mifepristone or mevastatin (Figs. 6 and 7). In the case of the latter agents, only mifepristone produced any increase above DMSO-treated cells in serum-containing medium and that increase was less than 1.2-fold above control. In contrast, the C1F1 cells may be useful in identifying agents that behave as weak, moderate, or potent CYP3A4 inducers. Chemicals that are potent inducers would produce luciferase activities 5-fold or greater over DMSO-treated cells and those that are moderate would produce activities 3-fold or greater above control. Weak CYP3A4 inducers would produce luciferase activities less than 3-fold over DMSO-treated cells. Thus, both cell lines could be used to screen potential inducers of CYP3A4 but in different capacities.
A recent report (El-Sankary et al., 2000) demonstrated a 26-fold induction by rifampicin in HepG2 cells transiently transfected with a CYP3A4 plasmid containing the proximal promoter and hPXR. That induction by rifampicin in the C1F1 cells was only 5- to 6-fold and dissimilar to that previously reported, may be for several reasons. The first explanation may be because cells were late passage (27–30). We found that with earlier passage cells (<20), 10 μM rifampicin produced an 8- to 10-fold increase in luciferase activity above DMSO-treated cells (data not shown). A second reason may be that in the study by El-Sankary et al. (2000), background luciferase activity produced by the control vector was deducted from values obtained for rifampicin- and DMSO-treated cells. We also found that deducting background levels markedly increased luciferase activity above control (30-fold) in the C1F1 cells treated with 10 μM rifampicin. However, in results reported herein, luciferase values for DMSO and rifampicin were directly compared. Direct comparisons alleviate additional experimental steps and data preparation, making the assays less subject to variability and more conducive to high throughput. Finally, the lower induction observed in our investigation may be because the distal enhancer element described previously (Goodwin et al., 1999) was not incorporated in our plasmid (p3A4). The additional enhancer has been shown to increase the responsiveness to CYP3A4 inducers.
The effects of culturing stably integrated HepG2 cells in various concentrations of serum and in a defined serum-free medium were also investigated. Herein, we found that cells cultured for more than 48 h in DMEM with serum concentrations below 5% exhibited a significant decline (p < 0.01) in cell viability unless a defined serum-free medium was used. This decline in variability was accompanied by altered inductive responses and greater well-to-well variability. These findings suggest caution in interpreting results generated from cells used in transfection assays or stably integrated cells maintained in DMEM with serum concentrations below 5%. HepG2 cells cultured in SF medium also exhibited a decline in cell viability when cultured for 60 h (Table 1). However, this modest 16% decline did not seem to affect the response of cells to inducers. Induction by the agents examined here was slightly higher in SF medium, although, not significantly greater (p < 0.01) than that in cells cultured in DMEM containing 10% FBS. The only exception to this finding was mifepristone, which produced greater increases in activity with medium containing 10% serum in C1F1 and C13-2 cells. The reason for higher luciferase activities in cells maintained in SF medium than in those cultured in DMEM with 10% serum and the discrepancy among inducers is unclear at present, but may be due to endogenous serum constituents that bind to the hPXR. Indeed because CYP3A4 expression is mediated by hPXR, there is the likelihood that competition exists between endobiotics and xenobiotics. Thus, the role of serum in culture media needs to be further evaluated for its effects on xenobiotic receptor-mediated induction of P450s.
To our knowledge, this is the first report to describe the generation of stably integrated cells lines harboring the CYP3A4 enhancer. The significance of this cell line lies in its ability to be used in a standardized high-throughput system that permits the ability to simultaneously screen numerous agents for induction of this human enzyme. More importantly, this system will allow for screening of inducers at the initial stages of drug development rather than later in preclinical studies. Prescreening compounds with the HTS described herein permits identification of potent inducers of human P450s that can be essentially “screened out” of development. In summary, by using standardized HTS systems containing enhancers from human P450 genes and an easily monitored reporter, information regarding the potential for a drug candidate to induce a human P450 can be rapidly, reliably, and reproducibly obtained early in the preclinical stages of drug testing. This would reduce development cost and time and allow the emphasis to rapidly shift to candidate compounds devoid of inducing properties.
Acknowledgments
We thank Dr. Mei Hsu for excellent technical advice during the generation of the stable cell lines.
Footnotes
-
This research was supported by National Institutes of Health Grant GM58287 (to M.-F.Y.).
-
DOI: 10.1124/jpet.102.038653
- Abbreviations:
- P450
- cytochrome P450
- PXR
- pregnane X receptor
- PCN
- pregnenolone 16α-carbonitrile
- bp
- base pair(s)
- hPXR
- human pregnane X receptor
- DMSO
- dimethyl sulfoxide
- PXRE
- pregnane X response element
- DMEM
- Dulbecco's modified Eagle's medium
- FBS
- fetal bovine serum
- HMM
- hepatocyte maintenance media
- TCDD
- 2,3,7,8-tetrachlorodibenzo-p-dioxin
- SF
- serum-free media
- RLU
- relative light unit
- OMP
- omeprazole
- HTS
- high throughput screening
- CAR
- constitutive androstane receptor
- Received May 9, 2002.
- Accepted June 20, 2002.
- The American Society for Pharmacology and Experimental Therapeutics