Cytochrome P450 3A4 (CYP3A4) is widely recognized to be a key enzyme responsible for the biotransformation of more than 50% of clinically used drugs (Li et al., 1995). CYP3A4, the predominant P450¹ isoform constitutively expressed in adult human liver, is induced by a variety of structurally diverse compounds, including rifampicin, phenytoin, and phenobarbital (Pichard et al., 1990).

Several years ago, it was discovered that human pregnane X receptor (hPXR, NR1I2), a member of the nuclear receptor family of ligand-activated transcriptional factors, is a key regulator of CYP3A4 gene transcription (Bertilsson et al., 1998; Blumberg et al., 1998; Kliewer et al., 1998; Lehmann et al., 1998) and is activated by compounds that are known CYP3A4 inducers, including drugs, steroids, and environmental chemicals (Jones et al., 2000; Moore et al., 2000; Luo et al., 2002). Ligand-activated hPXR binds to proximal PXR response element (prPXRE) and distal xenobiotic-responsive enhancer module (XREM) in regulatory regions of the CYP3A4 gene as a heterodimer with the 9-cis retinoic acid receptor-α (NR2B1).

Structurally diverse compounds have been reported as ligands of hPXR, and hPXR is also activated in cell-based reporter assays by a broad spectrum of xenobiotics (Jones et al., 2000; Ekins and Erickson, 2002; Luo et al., 2002). Crystal structure analyses suggest that the ligand-binding domain (LBD) of hPXR is large, hydrophobic, and flexible (Watkins et al., 2001, 2003), allowing promiscuity in accepting structurally diverse ligands. However, little information is available regarding the critical substituents of compounds involved in the activation of hPXR.

To obtain a better understanding of the structure-activity relationships, we examined hPXR-mediated transactivation in cell-based reporter assays by several series of barbiturates (amobarbital, barbital, cyclobarbital, hexobarbital, mephobarbital, pentobarbital, phenobarbital, and primidone), hydantoins [ethotoin, hydantoin, hydantoin 5-acetic acid, 5-(4-hydroxyphenyl)-5-phenylhydantoin (HPPH), 5-methyl-5-phenylhydantoin, 5-(4-methylphenyl)-5-phenylhydantoin (MPPH), mephenytoin, and phenytoin], and macrolide antibiotics (clarithromycin, erythromycin, josamycin, midecamycin, oleandomycin, troleandomycin, and spiramycin).

Materials and Methods

Chemicals. Mephobarbital was a gift from Mitsubishi Pharma Corp. (Osaka, Japan). Mephenytoin was kindly donated by Dr. A. Küpfer (University of Berne, Berne, Switzerland). Amobarbital, clarithromycin, erythromycin, phenobarbital, spiramycin, and troleandomycin were purchased from Sigma-Aldrich (St. Louis, MO). Hydantoin 5-acetic acid, 5-(4-hydroxyphenyl)-5-phenylhydantoin (HPPH), 5-methyl-5-phenylhydantoin, and 5-(4-methylphenyl)-5-phenylhydantoin (MPPH) were obtained from Aldrich Chemical Co. (Milwaukee, WI). Cyclobarbital and phenytoin were obtained from Tokyo Kasei Kogyo Co. (Tokyo, Japan). Ethotoin, hexobarbital, and primidone were obtained from Dainippon Pharmaceutical Co. Ltd. (Osaka, Japan). Barbital, hydantoin, josamycin, midecamycin, oleandomycin phosphate, pentobarbital, and rifampicin were obtained from Wako Pure Chemical Industries Ltd. (Osaka, Japan).

Plasmids. The open reading frame of hPXR was amplified from a liver 5′-stretch plus cDNA library (BD Biosciences Clontech, Palo Alto, CA) using the specific primers (5′-CCA GAA GCC AAC ATG GAG GTG AGA CC-3′ and 5′-CAG CCG CTC AGC TAC CTG TGA TGC CG-3′). The PCR fragment was cloned into pGEM-T Easy vector (Promega, Madison, WI). After EcoRI digestion, the fragment was ligated into the EcoRI site of pTarget vector (Promega), resulting in pTarget-hPXR. Correct insert orientation was confirmed by DNA sequence analysis.

A chimeric CYP3A4 luciferase reporter plasmid was prepared as follows. A CYP3A4 5′-flanking fragment (-374 to +40) containing prPXRE was generated by PCR from human genomic DNA as a template and cloned into pGL3-basic vector containing the firefly luciferase reporter gene (Promega) to generate a prPXRE-luciferase plasmid. A chimeric CYP3A4 luciferase reporter plasmid containing XREM and prPXRE (pGL3-XREM/prPXRE) was generated by inserting the fragment (-7849 to -7211) containing XREM amplified by PCR from human genomic DNA into prPXRE-luciferase plasmid digested with NheI and BglII.

Cell Culture and Transfection Assay. FLC7 cells (Kawada et al., 1998), a human hepatocellular carcinoma cell line, were kindly provided by Dr. S. Nagamori (Kyorin University, Tokyo, Japan). FLC7 cells were cultured in Dulbecco's modified Eagle's medium/F-12 supplemented with 10% fetal bovine serum. Cells (1.8 × 10⁵/well) were plated in 24-well plates 1 day before transfection. Plasmid pGL3-XREM/prPXRE (200 ng/well) was cotransfected with pTarget-hPXR (10 ng/well) and phRL-TK Vector (4 ng/well) into FLC7 cells by lipofectAMINE reagent and PLUS reagent (Invitrogen, Carlsbad, CA). At 3 h after transfection, the medium was replaced by Dulbecco's modified Eagle's medium/F12 supplemented with 10% fetal bovine serum, and cells were treated for 24 h with chemicals at the indicated concentrations. Stock solutions of the chemicals were prepared in dimethyl sulfoxide (DMSO), and the final concentration of the solvent was 0.1% (v/v) except for phenytoin, clarithromycin, and troleandomycin, which were used at the final concentration of DMSO of 0.3%. Control cultures received vehicle (0.1% DMSO) alone. Luciferase reporter activities were measured using a Dual-Luciferase reporter assay system (Promega) and Turner Designs Luminometer TD-20/20 (Turner Designs, Inc., Sunnyvale, CA) according to the manufacturer's instructions. The Renilla luciferase activity of the control plasmid phRL-TK was used to normalize the results of the firefly luciferase activity of the reporter plasmid pGL3-XREM/prPXRE. Since FLC7 cells used in this study showed undetectable or much lower metabolic capacity than HepG2 cells (data not shown), biotransformation of the compounds tested in this study was regarded to be negligible during the 24-h assay.

Determination of Three-Dimensional Distance. Three-dimensional distance between atoms and log P values was calculated by using CS Chem 3D Pro (CambridgeSoft Corporation, Cambridge, MA) after performing geometry optimization and energy minimization using the AM₁ of the semiempirical method. An RMS gradient of 0.1 kcal/Å mol was used.

Statistics. Results are expressed as mean ± S.D. Differences between two groups were analyzed using two-sample t test.

Results

Previous experiments in several laboratories have shown that rifampicin induced efficient activation of hPXR in cell-based reporter assays at the concentration of 5 or 10 μM (Bertilsson et al., 1998; Goodwin et al., 1999; Jones et al., 2000; Moore et al., 2000; Geick et al., 2001). Therefore, in the present study, the transcriptional activation induced by this concentration of rifampicin (5 μM) was used as the reference for quantifying the activity of other compounds tested. The magnitude of rifampicin response (about 10-fold) observed in this study is in agreement with previously reported results (Bertilsson et al., 1998; Goodwin et al., 1999; Jones et al., 2000). In this study, activation of the CYP3A4-luciferase reporter gene construct was investigated in FLC7 cells. Our preliminary study showed that the mRNA levels of hPXR and constitutive androstane receptor in FLC7 cells were much lower, but the heterodimerization partner retinoid X receptor is expressed in sufficient amounts. Therefore, it is considered that activation of the CYP3A4-luciferase reporter gene construct used in this study by rifampicin is a valid model for examination of hPXR-mediated CYP3A4 transcriptional activation.

Figure 1 shows the effects of eight barbiturates on hPXR-mediated transcriptional activation in cell-based reporter gene assays. hPXR was activated strongly (>10-fold) by mephobarbital at the concentration of 100 μM, moderately (5- to 10-fold) by pentobarbital and hexobarbital, and weakly (<5-fold) by cyclobarbital, amobarbital, and phenobarbital. Barbiturate and primidone did not show dose-dependent activation of hPXR at the concentrations tested (up to 300 μM).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 1.

Effects of barbiturates on hPXR-mediated transactivation.

Constructs pTarget-hPXR (10 ng/well) and pGL3-XREM/prPXRE (200 ng/well) were transiently transfected into FLC7 cells. Cells were treated with water (vehicle for phenobarbital), 0.1% DMSO (vehicle for the other compounds), 5 μM rifampicin, or 3, 10, 30, 100, and 300 μM concentrations of each of the compounds listed in Table 1 for 24 h. Data are expressed as mean ± S.D. of at least three experiments. ***, p < 0.001, **, p < 0.01, and *, p < 0.05 compared with vehicle control.

Figure 2 shows the effects of eight hydantoins on hPXR-mediated transcriptional activation in cell-based reporter gene assays. Mephenytoin showed the strongest activation of hPXR (>10-fold), followed by ethotoin (>5-fold) at the concentration of 300 μM. MPPH and phenytoin also enhanced the hPXR-mediated transactivation at the concentration of 100 μM (about 3-fold), whereas the activation at the concentration of 300 μM was similar to or weaker than that at the concentration of 100 μM. This limited maximal activation by MPPH might be due to cytotoxicity, since the decrease of Renilla luciferase activity and cell numbers was observed. 5-Methyl-5-phenylhydantoin, HPPH, hydantoin 5-acetic acid, and hydantoin had little or no effect on the hPXR-mediated transactivation.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 2.

Effects of hydantoins on hPXR-mediated transactivation.

Constructs pTarget-hPXR (10 ng/well) and pGL3-XREM/prPXRE (200 ng/well) were transiently transfected into FLC7 cells. Cells were treated with 0.1% DMSO (vehicle), 5 μM rifampicin, or 3, 10, 30, 100, and 300 μM concentrations of each of the compounds listed in Table 2 for 24 h. Data are expressed as mean ± S.D. of at least three experiments. ***, p < 0.001, **, p < 0.01, and *, p < 0.05 compared with vehicle control.

Figure 3 shows the effects of seven macrolide antibiotics on hPXR-mediated transcriptional activation in cell-based reporter gene assays. hPXR was significantly activated by troleandomycin (3.5-fold). Oleandomycin, clarithromycin, erythromycin, josamycin, midecamycin, and spiramycin did not significantly activate hPXR at the concentrations tested (up to 100 μM).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 3.

Effects of macrolide antibiotics on hPXR-mediated transactivation.

Constructs pTarget-hPXR (10 ng/well) and pGL3-XREM/prPXRE (200 ng/well) were transiently transfected into FLC7 cells. Cells were treated with 0.1% DMSO (vehicle), 5 μM rifampicin, or 3, 10, 30, and 100 μM concentrations of each of the compounds for 24 h. Data are expressed as mean ± S.D. of at least three experiments. ***, p < 0.001, **, p < 0.01, and *, p < 0.05 compared with vehicle control.

Discussion

To clarify the general structure elements that lead to optimal activation of hPXR, structure-activity relationships of hPXR-mediated transactivation by several series of barbiturates, hydantoins, and macrolide antibiotics were examined by cell-based reporter assays.

In the case of barbiturates, the hPXR-mediated transactivation in cell-based reporter assays was enhanced by mephobarbital, pentobarbital, hexobarbital, cyclobarbital, amobarbital, and phenobarbital but not by barbital and primidone (Fig. 1). These results are in good agreement with previously reported results showing that treatment with cyclobarbital, hexobarbital, phenobarbital, and mephobarbital, but not treatment with barbital, induced CYP3A (Murayama et al., 1996). Since rat primary hepatocytes were used in the study, it is necessary to examine the inductive effects of these barbiturates on CYP3A4 in human primary hepatocytes. As shown in Table 1, chemical structures of barbiturates that enhanced the hPXR-mediated transactivation indicate that the R3-substitution group at the 5-position is an alkyl-group, cyclohexenyl-group, or phenyl-ring, possessing more than five carbon atoms (Table 1, log P = 1.87–2.63). On the other hand, the R3-substitution group of barbital, which showed less effect on the hPXR-mediated transactivation, is an ethyl-chain, showing low hydrophobicity (Table 1, log P = 1.26). Crystal structure analyses suggest that the LBD of hPXR is large and hydrophobic (Watkins et al., 2001). These results suggest that the presence of a hydrophobic moiety in the R3-substitution group at the 5-position in the structure of barbiturates is important for activation of hPXR. In addition, the extent of the hPXR-mediated transactivation by mephobarbital was greater than that by phenobarbital. Mephobarbital has a structure similar to that of phenobarbital except for the presence of a methyl group in the R1-substitution. Therefore, the presence of a hydrophobic moiety in the R1-substitution group in the structure of barbiturates might enhance the activation of hPXR. Not only barbital but also primidone failed to enhance the hPXR-mediated transactivation, even though primidone has a structure similar to that of phenobarbital. However, unlike phenobarbital, primidone is missing a hydrogen-bond acceptor, a C = O moiety (X-substitution) at the 2-position (Table 1). According to pharmacophore for hPXR ligands derived by 12 hPXR ligands, the model consisted of one hydrogen-bond acceptor and four hydrophobic groups (Ekins and Erickson, 2002). In addition, the distances between the hydrophobic groups and the hydrogen-bond acceptor ranged from 3.6 to 7.6 Å (Ekins and Erickson, 2002). These features predicted by this pharmacophore are in close agreement with the X-ray-determined structure of the hPXR LBD (Watkins et al., 2001). In the case of phenobarbital, which enhanced the hPXR-mediated transactivation, there is a three-dimensional distance of about 7 Å between the phenyl ring (R3) and C = O moiety (X substitution) at the 2-position of phenobarbital (Fig. 4, left). On the other hand, primidone has no substitution group at the 2-position, and the three-dimensional distance between the phenyl ring (R3) and C = O moiety at the 4- or 6-position of promidone is about 5.3 Å (Fig. 4, right). These findings suggest that a distance of 7 Å between the phenyl ring and C = O moiety (i.e., X-substitution at the 2-position) in addition to the presence of a hydrophobic moiety of the R3-substitution group at the 5-position is necessary for the activation of hPXR by barbiturates.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 4.

Three-dimensional structures of phenobarbital and primidone.

Three-dimensional distance between the phenyl ring and oxygen atom of the carbonyl group was calculated as described under Materials and Methods. Oxygen atoms are shown in red, carbon atoms in gray, nitrogen atoms in blue, and hydrogen atoms in light blue.

Next, we examined structure-activity relationships of hPXR-mediated transactivation by hydantoins. Among the hydantoins studied, mephenytoin showed the strongest activation of hPXR, followed by ethotoin. Mephenytoin and ethotoin, but not other hydantoins studied, have an alkyl chain as an R1-substitution group at the 3-position. Therefore, it was thought that the presence of an alkyl chain at the 3-position in the structure of hydantoin is an important factor for activation of hPXR. In addition, phenytoin and MPPH, which contain a phenyl or methylphenyl group as both R2- and R3-substitution groups at the 5-position, also activated hPXR, as did mephenytoin and ethotoin at the concentration of 100 μM, although the extent of activation was weaker than that by mephenytoin and ethotoin at the concentration of 300 μM. On the other hand, hydantoin, hydantoin 5-acetic acid, and 5-methyl-5phenylhydantoin, which contain no or only one phenyl group as R2- and R3-substitution groups at the 5-position, did not activate hPXR. Although HPPH contains phenyl and hydroxyphenyl groups at the R2- and R3-positions, it did not activate hPXR. These results suggest that bulky and hydrophobic moieties of both R2- and R3-substitution groups at the 5-position are responsible for the activation of hPXR. Therefore, the alkyl chain of the R1-substitution group at the 3-position and the bulky and hydrophobic moieties of both R2- and R3-substitution groups at 5-positions in the structure of hydantoin play an important role in hPXR-mediated transcriptional activation.

Among the macrolide antibiotics studied, only troleandomycin showed dose-responsible activation of hPXR (Fig. 3). Oleandomycin also showed a tendency to activate hPXR, but the activation was not significant. Troleandomycin is a triacetate ester of oleandomycin (Fig. 5). Delaforge et al. (1983) reported that total P450 content in rat liver was induced by erythromycin estolate but not by erythromycin. Since erythromycin estolate is an ester compound of erythromycin, they suggested that the ester form is an important factor for induction of total P450 content. Since hydrophobicity of the ligand is important for hPXR activation (Watkins et al., 2001), it was suggested that the activation of hPXR by troleandomycin is dependent on the increase in hydrophobicity by triacetate esterification of oleandomycin. On the other hand, activation of hPXR by clarithromycin, erythromycin, josamycin, midecamycin, and spiramycin was not observed (Fig. 3). Although the reason is not clear, these macrolide antibiotics might have three-dimensional structures that make it difficult for them to be ligands of hPXR. Further computational studies such as docking the macrolides in crystallographic structures of PXR are necessary to determine whether or not they are likely to fit in the binding site of hPXR.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 5.

Structures of macrolide antibiotics used in this study.

Watkins et al. (2001) reported that the ligand-binding cavity of the hPXR-LBD is largely hydrophobic and is lined by 28 amino acid residues consisting of 20 hydrophobic, 4 polar, and 4 charged residues. The fact that a small number of polar residues are spaced throughout the smooth and hydrophobic ligand-binding pocket of hPXR appears to permit binding of ligands that are hydrophobic and contain a small number of polar groups. The findings obtained in this study suggest that hydrophobicity of ligands and adequate distance between the hydrogen-bond acceptor and the hydrophobic group are important for hPXR activation. This observation was in good agreement with the prediction by crystal structure analyses (Ekins and Erickson, 2002). Since activation of hPXR leads to induction of not only drug-metabolizing enzymes but also drug transporters (Kliewer et al., 1998; Geick et al., 2001; Schuetz and Strom, 2001; Synold et al., 2001; Goodwin et al., 2002), an hPXR ligand is capable of leading numerous drug-drug interactions. Therefore, our findings regarding the structural features of compounds that activate hPXR as ligands provide important insights into how hPXR detects xenobiotics and may be useful in predicting and preventing drug-drug interactions. In addition, further computational studies such as docking the compounds in crystallographic structures of hPXR will provide useful information in determining whether or not they are likely to fit in the binding site of hPXR.