Expression of CYP3A4 , CYP2B6 and CYP2C9 is regulated by the vitamin D receptor pathway in primary human hepatocytes

The fully active di-hydroxylated metabolite of vitamin D 3 (1 α ,25(OH) 2 D 3 ) induces the expression of CYP3A4 and to a lesser extend of CYP2B6 and CYP2C9 genes in normal differentiated primary human hepatocytes. Electrophoretic mobility shift assays and cotransfection in HepG2 cells using wild-type and mutated oligonucleotides revealed that the vitamin D receptor (VDR) binds and transactivates those xenobiotic-responsive elements (ER6, DR3 and DR4) previously identified in CYP3A4 , CYP2B6 and CYP2C9 promoters and shown to be targeted by the Pregnane X Receptor (PXR) and/or the Constitutive Androstane Receptor (CAR). Full VDR-response of various CYP3A4 heterologous/homologous promoter-reporter constructs requires both the proximal ER6 and the distal DR3 motifs, as observed previously with rifampicin-activated PXR. Co-transfection of a CYP3A4 homologous promoter-reporter construct (including distal and proximal PXR-binding motifs) and of PXR or CAR expression vectors in HepG2 cells revealed the ability of these receptors to compete with VDR for transcriptional regulation of CYP3A4 . In conclusion, this work suggests that VDR, PXR and CAR control the basal and inducible expression of several CYP genes through competitive interaction with the same battery of response elements. but also CYP2B6 and CYP2C9 in primary human hepatocytes. In addition, we show that VDR is able to bind and transactivate different motifs recognized by xenobiotic-activated PXR and CAR, in the promoter of these CYP genes.


INTRODUCTION
Cytochrome P450 enzymes (CYP) are mainly expressed in the liver and catalyze the metabolic conversion of xenobiotics, including environmental pollutants and drugs, to more polar and easily disposable derivatives (2,3). CYP genes from the CYP2 and CYP3 families are inducible by many xenobiotics including notably, barbiturates and rifampicin. Two nuclear receptors, the Pregnane X Receptor (PXR, NR1I2) and the Constitutive Androstane Receptor (CAR, NR1I3) have recently been shown to mediate CYP2 and CYP3 gene induction in animals and in man (4)(5)(6). Both PXR and CAR form heterodimers with the Retinoic Acid Receptor (RXR, NR2B1). PXR is activated by a wide spectrum of xenobiotics and steroids (4,7,8) and controls CYP3A4 and CYP3A7 induction by targeting two specific responsive elements present in the regulatory region of these genes (4,(7)(8)(9)(10)(11)(12). The first of these is the proximal PXR responsive element (pPXRE) located at -160. It consists of an everted repeat of the nuclear receptor half-site AGGTCA separated by 6 nucleotides (ER6); this element is necessary but not sufficient for full transactivation of the CYP3A4 promoter.
Indeed, full PXR-mediated induction requires the presence of a second distal xenobiotic response element (dPXRE), located between -7800 and -7200 (9). This element is composite and consists of two direct repeats separated by 3 nucleotides (DR3), encompassing an ER6 motif. In contrast to PXR, CAR is sequestered in the cytoplasm and translocates into the nucleus upon activation, notably in response to phenobarbital (6,13). Several groups have identified a complex phenobarbital-response element module (PBREM) which consists of two nuclear receptor binding sites (termed NR1 and NR2) and one NF1 binding site (12,14). Both NR1 and NR2 are imperfect DR4 motifs and essential for phenobarbital induction of CYP2B genes. In human CYP2B6, PBREM is located between -1684 and -1733, and has been shown to bind to and be transactivated by CAR and by PXR (12,15).
Previous reports revealed that 1α, 25-dihydroxy vitamin D3 (1α,25(OH) 2 D 3 ), the most active metabolite of vitamin D3, behaves as a transcriptional inducer of CYP3A4 in the colic carcinoma Caco-2 cell line and in human intestinal LS180 cell line (16,17). Vitamin D3 function is mediated through the vitamin D receptor (VDR, NR1I1) which, after binding 1α,25(OH) 2 D 3 with high affinity forms heterodimers with RXR (18)(19)(20). The heterodimer then binds to and transactivates the vitamin D response elements (VDRE) present in the regulatory region of target genes (21). The classical VDREs consist of a direct repeat of nuclear receptor half-sites separated by three nucleotides (DR3) (18). In the classic vitamin D-responsive organs including the intestine, bone, kidney and parathyroid gland, vitamin D3-activated VDR plays a central role in the regulation of calcium and phosphate homeostasis, bone mineralisation and resorption, inhibition of cell growth and parathyroid hormone synthesis (22). VDR is also expressed in many other non-classic vitamin D-responsive organs including the liver, muscle, skin, immune system, pancreas, brain and cancer cells (23) in which it controls a number of biological processes including immunomodulation, tissue regeneration, inhibition of cell growth and apoptosis, and cell differentiation (24)(25)(26).
In an exploratory part of this work we found that 1α,25(OH) 2 D 3 is an inducer of CYP3A4 in human hepatocytes, as previously observed by others in intestinal cell lines (16,17). We therefore thought that VDR might be able to target PXR-and/or CAR-responsive elements of CYP3A4. We further reasoned that, if the hypothesis is correct, 1α,25(OH) 2 D 3 could be an inducer of other CYP genes controlled by these receptors. The data presented here show that 1α,25(OH) 2 D 3 not only induces CYP3A4, but also CYP2B6 and CYP2C9 in primary human hepatocytes. In addition, we show that VDR is able to bind and transactivate different motifs recognized by xenobiotic-activated PXR and CAR, in the promoter of these CYP genes.

HepG2 cells (Human hepatocarcinoma) purshased from the European Collection of Cell
Cultures (Salysbury, England) were maintained in Dulbecco's modified Eagle's medium supplemented with 10% FCS, 100µg/ml penicillin and 100µg/ml streptomycin (Life Technology, Inc.). Transfection of plasmid DNA was performed in single batches with Fugene-6 (Roche Diagnostics Corporation, Indianapolis, USA) as recommended by the manufacturer. Transfections were performed using 100 000 cells, 250 ng of reporter plasmid and 50ng of pSG5-hVDR expression vector (provided by P. Balaguer, INSERM U439, Montpellier). For competition experiments, we used 500ng of reporter plasmid and 100ng of pSG5-hVDR expression vector. Cotransfection of hPXR or mCAR were performed using increasing concentrations (10, 50, 100, 300ng) of both expression vectors and pSG5 (Stratagene, La Jolla, CA) empty vector was added to normalize the total concentration of transfected plasmid DNA. As an internal control of transfection, 25ng of pSV-galactosidase (Progema Madison WI) was used in all experiments. After 16 hours, the medium was changed, and fresh medium containing 0.1% dimethylsulfoxide or inducers were added. Cells were harvested in reporter lysis buffer (Promega, Madison, WI) after 24 hours incubation, and cell extracts were analysed for luciferase and β-galactosidase activities as described elsewhere (11).

Liver samples and hepatocyte cultures
Hepatocytes were prepared from liver lobectomy segments resected from adult patients for medically required purposes unrelated to our research program. The use of these human hepatic specimens for scientific purposes has been approved by the French National Ethics Committee. Three different cultures from three different liver donors were made in this work :

RT-PCR experiments
Reverse transcription was performed from 1 µg of mRNA using the MMLV-RT Life Technology Kit (Gibco BRL, Cergy-Pontoise France) according to the manufacturer's instructions. One tenth of the RT reaction was then subjected to PCR.

In vitro translation and electrophoretic mobility shift assays
Electrophoretic mobility shift assays were performed using VDR and RXR prepared by in vitro translation using a transcription-translation coupled system (Promega). Proteins were incubated for 20 minutes at room temperature with 50,000 cpm of T4 polynucleotide kinaselabelled oligonucleotides in 10 mM Tris (pH 8), 6% glycerol, 1mM DTT, 1µg/µl poly-(dI-dC) Anti-RXRα (N197, sc 774 X, Santa Cruz Biotechnology) was used for the "supershift" assays. Autoradiography was carried out by exposing the dried gel to Kodak X-AR film. 15±2 for CYP3A4 ; 3.5±1 for CYP2B6 and 2.6±1 for CYP2C9 mRNAs. In comparison, rifampicin induction ratios were respectively: 50±15 for CYP3A4, 10±3 for CYP2B6 and 3.3±1.5 for CYP2C9 mRNAs. This last gene was recently shown to be positively regulated by rifampicin and phenobarbital through CAR/PXR activation (28,33). GAPDH mRNA levels used as quality controls of RNA preparations were not affected significantly by 1α,25(OH) 2 D 3 . The finding that 1α,25(OH) 2 D 3 induces CYP mRNAs within the nanomolar concentration range suggested a classical vitamin D3 receptor-mediated mechanism of induction. Although the consensus VDRE is a DR3 motif, this nuclear receptor has been shown to bind other motifs including DR4, DR6 and inverted palindromes (32,34). We therefore suspected that CYP2B6, CYP2C9 and CYP3A4 induction by 1α,25(OH) 2 D 3 could be mediated by VDR through the previously identified PXR and CAR responsive elements.

VDR-RXR heterodimer binds the PXR responsive elements of the CYP3A4 promoter
The PXR responsive elements of CYP3A4 consist of a proximal ER6 (-160), hereafter referred to as 3A4-pER6, and a distal enhancer (-7800/-7200) containing three nuclear receptor motifs, referred to hereafter as 3A4-5'dDR3, 3A4-dER6, and 3A4-3'dDR3 ( Figure   2). These elements correspond to dNR1, dNR2 and dNR3, identified by Goodwin et al. (9), respectively. dNR1 and dNR3 have been reported to be the key elements confering enhancer activity. Gel mobility shift assays were performed in order to determine whether VDR interacts with these elements.
First, we checked the binding of in vitro translated VDR-RXR heterodimer to a consensus VDRE oligonucleotide (rANF-DR3-type) by gel shift assay, as shown in Figure   3A. As expected, a retarded band was observed when both VDR and RXR were incubated with the target oligonucleotide (lane 4) but not when these receptors were incubated alone (lanes 2 and 3). Anti-RXR antibodies produced a supershift (lane 5) while an excess of unlabelled rANF DR3 oligonucleotide suppressed the retarded band (not shown). In addition, the specific VDR-RXR-DNA complex was suppressed, in a dose-dependent manner, when incubated in the presence of a 5-to 50-fold molar excesses of unlabelled 3A4-5'dDR3 (lanes [8][9] or 3A4-pER6 (lanes [10][11]. This suggests that these elements can be targeted by the VDR/RXR heterodimer. In contrast, excess of the 3A4-dER6 oligonucleotide did not produce any suppression of the VDR/RXR-VDRE complex (lanes 6-7).
In the next series of experiments, we used the same assay to investigate the binding of VDR to both 3A4-5'dDR3 and 3A4pER6 oligonucleotides. As expected, no complex was observed when the probes were incubated with VDR or RXR alone ( Figure 3B

CYP2C9 promoters
A 51 bp sequence termed the phenobarbital-responsive element (PBRE) has been shown to be necessary and sufficient for phenobarbital induction of mouse Cyp2b10 gene (35)(36)(37).
Sequence analysis of various CYP2B PBREs reveals the presence of two conserved imperfect DR4 motifs (NR1 and NR2) which appear to be essential for a full response to phenobarbital.
In the human CYP2B6 gene, these elements are oriented in opposite directions with respect to those of the mouse and the rat, and are located in the region -1733/-1684 (12). They are hereafter referred to as 2B6-3'DR4 and 2B6-5'DR4, respectively. Recently, we identified a functional CAR-responsive element (CAR-RE) in the -1856/-1783 region of human CYP2C9 (28). Sequence analysis revealed the presence of an imperfect DR4 motif, hereafter referred to as 2C9-DR4. This element was shown to bind to and be transactivated by CAR as well as by PXR, albeit to a lower extent.
As shown in  (Figure 4 lanes 4-5 and 11-12). Note, however, that the binding of the VDR-RXR heterodimer to 2C9-DR4 seems to be of much lower affinity compared with the binding to the other CYP3A4 and 2B6 PXR/CAR elements.
In sum, these observations show that the VDR-RXR heterodimer binds to the major PXR/CAR-responsive elements of CYP3A4, CYP2B6 and CYP2C9.

VDR transactivates the PXR responsive elements of CYP3A4
Transactivation (harboring element pER6), in front of the LUC reporter gene (Figure 2). This construct (CYP3A4-5'dDR3/dER6/3'dDR3/pER6-LUC, construct A) has been shown to be fully responsive to PXR (9) and this was confirmed in this work (see Figure 7A). Several deletions of this construct (constructs B and C, Figure 2) were then generated and their transcriptional activity was measured in response to 1α,25(OH) 2 D 3 -activated VDR. The results are presented in Figure 5C. Interestingly, when the proximal promoter of CYP3A4 (-262/+11) was replaced by a minimal thymidine kinase promoter (corresponding to the loss of the proximal ER6), the transcriptional activity of 5'dDR3/dER6/3'dDR3 (construct D) and of 5'dDR3/dER6 (construct E) was less than 50% of that measured with construct A, suggesting a cooperative interaction between the dPXRE region and the pER6 element as previously reported for PXRmediated transactivation of these elements (9). Finally, in control experiments, neither PXR nor CAR were activated by 1α,25(OH) 2 D 3 (not shown).
In sum, these results show that both the proximal region containing pER6 and the distal enhancer dPXRE containing the dDR3 motifs are necessary to confer full VDRresponse and that, in the context of the CYP3A4 homologous promoter, transactivation by 1α,25(OH) 2 D 3 -activated VDR parallels transactivation by xenobiotic-activated PXR.

Downloaded from
Similar experiments were carried out with the CAR/PXR responsive elements identified in CYP2B6 and CYP2C9. The results are shown in Figure 6. A modest but significant and reproducible activation of both 2B6-3'DR4 and 2C9-DR4 constructs was observed in the presence of 1α ,25(OH) 2 D 3 -activated VDR. Indeed, VDR-mediated transactivation of the major CYP3A4 responsive elements was much greater than the activation observed here. This is consistent with the finding that in primary hepatocytes the induction ratio of CYP3A4 mRNA in response to 1α,25(OH) 2 D 3 is much greater than that of both CYP2B6 and CYP2C9 mRNAs ( Figure 1B).

Competition of VDR-mediated CYP3A4 transactivation by PXR and CAR
Since VDR binds and transactivates PXR and CAR responsive elements, the next step activator recrutement is blocked, but it is still able to bind to its responsive element. These results suggest a competition between CAR and VDR for the CYP3A4 promoter elements.
Finally, these results are in agreement with the gel shift experiments showing that VDR can bind to PXR (Figure 3) and CAR (Figure 4) responsive elements and therefore confirm that, in the context of the CYP3A4 homologous promoter, the sites targeted by VDR overlap with those recognized by PXR and CAR.

DISCUSSION
In this study, we have shown that 1α,25(OH) 2 (19-190nM). Although it was considered in the past that VDR could be absent or expressed at very low level in the liver, it was recently demonstrated that this receptor is present in fetal, neonatal and adult rat liver, by RT-PCR and immunohistochemistry (38). Control experiments, using the inducible expression of FBP, previously shown to be controlled by VDR (31), have clearly shown that VDR is expressed and activated in our cultures after treatment with 1α,25(OH) 2 D 3 .
Although each nuclear receptor binds preferentially to a specific DNA sequence (39,40,1), there have recently been indications that a given receptor (whatever the family it belongs to) may bind to and transactivate different responsive elements. Thus for example, the steroid hormone receptors (NR3C subfamily) bind classically, and almost exclusively, as homodimers to palindromic sequences separated by 3 nucleotides . However, the glucocorticoid or the estrogen receptors have been shown to bind to DRs with different spacings between half sites (including DR2, DR5, DR6, DR9) as well as to an ER9, although binding to these motifs is weaker than to the palindrome (41). Zhou et al (42) reported that the androgen receptor may bind to a DR1 motif in addition to the classical palindrome. On the other hand, VDR, PXR and CAR belong to the NR1I subfamily which form heterodimers with RXR. Their responsive motifs consist of a hexanucleotide consensus sequence (AGGTCA), which can be configured into different motifs including direct repeats (DR), everted repeats (ER) and inverted repeats (IR). Several authors have reported that CAR and PXR can transactivate CYP2 or CYP3 genes via the same response elements in a xenobioticdependent manner. Thus, for example PXR is able to transactivate CYP2B genes via recognition of the phenobarbital responsive DR4 element (43) and reciprocally, CAR is able to transactivate human CYP3A4 through the PXR-response elements, pER6 and dDR3 (15).
The existence of a possible cross-talk between these two nuclear receptor signaling pathways has accordingly been suggested. This apparent versatility in the ability of a given nuclear receptor to target similar but distinct DNA sequences is believed to result from the flexibility of either the ligand and/or DNA binding domains, the intervening linker region, or the DNA template itself.
The results presented here suggesting that VDR binds to and transactivates DR4 and ER6 motifs, in addition to the more classical DR3 elements, are therefore not surprising and clearly offer another example of this nuclear receptor versatility. Indeed, other VDRE motifs have been previously identified including a DR4 (for which VDR exhibited an higher affinity than for DR3), a DR6 and an inverted palindrome IP9 (32,44). In addition, sequence comparison with other members of the nuclear receptor family shows that VDR and PXR isoforms share the greatest similarity (64%) in their DNA binding domain (4).  (16,17) showing that vitamin D affects CYP gene expression increase the list of those physiological compounds able to interfere with the metabolism of xenobiotics. Actually, our results suggest that, in the absence of xenobiotic, the basal expression of CYP2 and CYP3 genes may be, at least in part, controlled through VDR activation. In the presence of xenobiotics able to activate either PXR or CAR, these receptors then will compete efficiently with VDR (see Figure 7) on CYP gene promoter response elements. In this respect, it has to be noted that the extent of CYP3A4 and CYP2B6 mRNA induction in primary human hepatocytes, was much greater in response to rifampicin than in response to 1α,25(OH) 2 D 3 .
Finally, although the results presented here suggest that the implication of VDR on CYP3A4 basal expression is substantial at physiological concentrations of vitamin D, its implication on CYP2C9 and CYP2B6 appears to be relatively modest, so that, the physiological significance of vitamin D effects on these genes is less clear. In this respect it is worth emphasizing that, CYP2C9 appears to be a primary glucocorticoid receptor-responsive gene (28)  Vitamin D can be obtained from different sources (22). A few components of the diet including fish oils, egg yolks, milk and liver contain naturally significant amounts of vitamin D3, while some plants contain vitamin D2. Many other foods are fortified with these vitamins.
Another source is the skin in which ultraviolet light induces the photoconversion of 7dehydrocholesterol to previtamin D3, followed by thermal isomerisation to vitamin D3. It is therefore possible that interindividual differences in dietary and/or light exposure habits may partly account for interindividual variations in CYP2/CYP3 basal expression and related processes such as drug and xenobiotic metabolism as well as prodrug and procarcinogen activation. These considerations provide another reasonable basis for the occurrence of xenobiotic-dietary compound interactions.
Finally, the reason why these genes are controlled by VDR is unclear. CYP2B6, CYP2Cs and CYP3A4 have not been shown to be involved in the metabolism of vitamin D (49)(50)(51). However, it has been observed that prolonged therapy with rifampicin can cause vitamin D deficiency (52). In 8 healthy subjects, rifampicin treatment reduced circulating levels of 25 hydroxy vitamin D and 1α,25(OH) 2 D 3 by 34 and 23% respectively. In addition, rifampicin and phenobarbital are two of the drugs most frequently associated with osteomalacia, a metabolic bone disease characterized by a defect of bone mineralization frequently due to an alteration of vitamin D metabolism (53). This suggests that CAR and/or PXR might be involved in the control of genes involved in vitamin D synthesis or catabolism.
In conclusion, this work suggests that VDR, PXR and CAR control the basal and inducible expression of several CYP genes through competitive interaction with the same battery of response elements (ER6, DR3, DR4). In consequence, we suggest that the expression of VDR-controlled genes might be affected by xenobiotics such as rifampicin through the PXR and/or CAR pathway. This possibility is under current evaluation in our laboratory.