It was previously shown that CYP3A4 is induced in the human intestinal Caco-2 cell model by treatment with 1α,25-dihydroxy vitamin D3 (1,25-D3). We demonstrate the vitamin D analog, 19-nor-1α,25-dihydroxy vitamin D2, is also an effective inducer of CYP3A4 in Caco-2 cells, but with half the potency of 1,25-D3. We report that treatment of LS180 cells, a human intestinal cell line, with 1 to 10 nM 1,25-D3 dose dependently increased CYP3A4 protein and CYP3A4 mRNA expression.CYP3A4- and CYP3A23-promoter-Luciferase reporter constructs transiently transfected into LS180 cells were transcriptionally activated in a dose-dependent manner by 1,25-D3, whereas mutation of the nuclear hormone receptor binding motif (ER6) in the CYP3A4 promoter abrogated 1,25-D3 activation of CYP3A4. Although theCYP3A4 ER6 promoter element has been shown to bind the pregnane X receptor (PXR), this receptor does not mediate 1,25-D3 induction of CYP3A4 because a) PXR is not expressed in Caco-2 cells; b) PXR mRNA expression is not induced by 1,25-D3 treatment of LS180 cells; and c) the ligand binding domain of human PXR was not activated by 1,25-D3. 1,25-D3 uses the vitamin D receptor to induce CYP3A4 because a) the vitamin D receptor (VDR)-retinoid X receptor (RXR) heterodimer binds specifically to the CYP3A4 ER6; b) selective mutation of the CYP3A4 ER6 disrupted the binding of VDR-RXR; and c) reporter constructs containing only three copies of the CYP3A4 ER6 linked to a TK-CAT reporter were activated by 1,25-D3 only in cells cotransfected with a human VDR expression plasmid. These data support the hypothesis that 1,25-D3 and VDR induce expression of intestinal CYP3A by binding of the activated VDR-RXR heterodimer to theCYP3A PXR response element and promoting gene transcription.
The CYP3A subfamily of enzymes catalyzes the metabolic transformation of numerous drugs and some endogenous molecules. Although CYP3A can be detected in several tissues of the body (Wrighton and Thummel, 2000), the highest specific content is found in parenchymal cells of the liver and in mature enterocytes lining mucosal villi of the small intestine (Kolars et al., 1994b). Accordingly, CYP3As have been implicated directly in the first-pass metabolism of cyclosporine (Kolars et al., 1991) and midazolam (MDZ) (Pine et al., 1996) and probably contribute to the low oral bioavailability of these and several other CYP3A substrates, including lovastatin, felodipine, saquinavir, and buspirone.
One of the most confounding and poorly understood aspects of CYP3A-dependent drug metabolism, is the significant interindividual variability in hepatic and intestinal CYP3A content observed in the human population. The two dominant isoforms in adults, CYP3A4 and CYP3A5, can vary 10-fold or more (Lown et al., 1994; Shimada et al., 1994; Paine et al., 1997), even when obvious causes of high (induction) or low (degradation) specific content are excluded. Although interindividual differences in observed CYP3A levels may be attributed to variable rates of CYP3A degradation, particularly after the ingestion of certain drugs or dietary products (Wrighton and Thummel, 2000), recent research suggests that constitutive differences can arise from variable mRNA levels and translation (for review, see Guengerich, 1999).
Interindividual differences in hepatic and intestinal CYP3A5mRNA levels and polymorphic enzyme expression have recently been attributed to an intronic point mutation that leads to aberrant splicing and production of a truncated protein (Kuehl et al., 2001). Differences in hepatic CYP3A4 mRNA levels have been attributed to multiple factors, including variable expression and activation of the pregnane X receptor/steroid and xenobiotic receptor (PXR/SXR) by endogenous and exogenous ligands, and variable expression of P-glycoprotein (Schuetz et al., 1996b; Blumberg et al., 1998;Lehmann et al., 1998). Human PXR is found preferentially in liver and intestine (Blumberg et al., 1998; Lehmann et al., 1998), as is CYP3A4 (Wrighton and Thummel, 2000). Moreover, expression of constitutively active human PXR in transgenic mouse liver leads to an up-regulation of hepatic CYP3A (Xie et al., 2000). Although several steroids bind to the receptor (Blumberg et al., 1998; Lehmann et al., 1998), the exact nature of the endogenous, high-affinity PXR ligand is unknown.
Although it is plausible to assume that PXR will also mediate constitutive intestinal CYP3A4 expression, several lines of evidence suggest an additional regulatory mechanism in this organ. First, hepatic and intestinal CYP3A4 specific contents are not concordant (Lown et al., 1994; Thummel et al., 1996; Paine et al., 1997; Gorski et al., 1998). A patient may exhibit high levels of hepatic CYP3A and low intestinal CYP3A levels and vice versa. Moreover, the relative level of PXR mRNA in the colon is greater than or equal to the level found in small intestine (Bertilsson et al., 1998; Lehmann et al., 1998), yet CYP3A4 expression is greatest in mucosa of the proximal small intestine (Paine et al., 1997) and very low in the colon (DeWaziers et al., 1990;Gervot et al., 1996).
A second nuclear hormone, the vitamin D receptor (VDR) is found at higher levels in the intestine than in the liver (Berger et al., 1988;Stumpf, 1995). Intestinal VDR plays a major role in calcium absorption and systemic calcium homeostasis (Schellens et al., 1988). In the rat, there seems to be a higher level of VDR in differentiated villus cells compared with cells of the Crypts of Lierberhün (Chan and Atkins, 1984), and ligand binding studies indicate a higher number of receptors in the proximal small intestinal mucosa, than in the colonic mucosa (Feldman et al., 1979). These patterns of heterogenous VDR expression mirror the heterogenous expression of CYP3A4 in human intestine (DeWaziers et al., 1990; Paine et al., 1997).
The regulation of intestinal CYP3A4 expression by VDR and 1α,25-dihydroxy vitamin D3 was first suggested by studies showing that treatment of Caco-2 cells with vitamin D analogs led to significant enhancement of CYP3A4 mRNA and protein levels and increased midazolam hydroxylation activity (Schmiedlin-Ren et al., 1997). Moreover, the rank order of CYP3A4 induction by vitamin D analogs (1α,25-dihydroxy vitamin D3 > 25-hydroxy vitamin D3 > unhydroxylated vitamin D3) was consistent with a mechanism of increasedCYP3A4 gene transcription that involved hormone binding to the vitamin D receptor. Thus, it was suggested that intestinal expression of CYP3A4 in vivo might also be mediated by a VDR signaling pathway (Schmiedlin-Ren et al., 1997). However, the CYP3A4gene lacks a vitamin D receptor binding element (VDRE) consensus sequence—an imperfect direct repeat of a core hexanucleotide sequence, [G/A] GGT[G/C]A, with a three-nucleotide spacer (i.e., DR3). Therefore, we hypothesized that if VDR affects CYP3A4transcription, it must be mediated through cross-talk with a different response motif such as the PXR response element (PXRE) (i.e., ER6). In this article, we report on results from in vitro and in vivo experiments that support this hypothesis.
Cos-7 monkey kidney cells, human embryonic kidney 293 cells, and LS180 cells [American Type Culture Collection (ATCC), Manassas, VA] were cultured in ATCC-recommended medium (Invitrogen, Carlsbad, CA) supplemented with 10% fetal calf serum and 1% Pen/Strep solution and maintained at 37°C in a humidified incubator. Caco-2 cells (American Type Culture Collection) were seeded onto semipermeable laminin-coated inserts at 105cells/cm2 and cultured as described previously (Schmiedlin-Ren et al., 1997). After reaching confluence, cells were fed for 1 to 3 weeks with Dulbecco's modified Eagle's medium containing 0.1 mM nonessential amino acids, 100 U/ml sodium penicillin, 100 μg/ml streptomycin, 0.1 M sodium selenite, 3 μM zinc sulfate, 45 nM dl-α-tocopherol, 5% heat-inactivated FBS, and 1α,25-dihydroxy vitamin D3 (calcitriol) or 19-nor-1α,25-dihydroxy vitamin D2(paricalcitol) as indicated for individual experiments. Calcitriol and paricalcitol were kindly provided by Abbott Laboratories (Chicago, IL).
The hVDR expression plasmid was a gift from Dr. J. Wesley Pike (University of Cincinnati, OH). pSG5-hPXRΔATG andCYP3A4-(ER6)3-TK-CAT were generously provided by Dr. Steve Kliewer (GlaxoSmithKline, Research Triangle Park, NC; Lehmann et al., 1998). pCMX-GAL-l-SXR and tk(MH100)4-LUC (Blumberg et al., 1998) were generously provided by Dr. Bruce Blumberg (University of California, Irvine). CYP3AP1-LUC [bp −266 to +17 relative to the start site of transcription in CYP3AP1 and previously thought to be the CYP3A5 promoter and that contains a PXRE that cannot bind PXR (Blumberg et al., 1998)] was described previously (Schuetz et al., 1996c). CYP3A23-TK-LUC (bp −220 to −56 relative to the start site of CYP3A23) contained both the proximal DR3 (bp −134 to −120, TGAACTtcaTGAACT) and distal ER6 (bp −166 to −149, TTAACTcaaaggAGGTCA) overlapping a DR4 (bp −164 to −149, AACTCAaaggAGGTCA). CYP3A4-SV40-LUC (bp −220 and −56, relative to the start site of CYP3A4 transcription) contained the ER6 (TGAACTCAAAGGAGGTCA). Mutant (mtCYP3A4-SV40-LUC) with a mutated ER6 (TGAACTCAAAGGAATACA; underlined bases indicate receptor binding sites and bold bases were changed to disrupt nuclear hormone receptor binding) was created directly fromCYP3A4-SV40-LUC by PCR with a mutant primer and the QuikChange Site-Directed Mutagenesis Kit (Stratagene, La Jolla, CA). The mutant was sequenced to confirm base changes.
Midazolam Hydroxylation Activity in Caco-2 Cells Treated with 1,25-D3 or 19-nor-1,25-D3.
Caco-2 cells (passage 23) were seeded on laminin-coated polyethylene terephthalate inserts (0.9 cm2). Upon reaching confluence, cells were cultured in differentiation medium supplemented with vehicle (ethanol, ∼ 1%) or 0.05, 0.1, 0.25, 0.5, or 1.0 μM 1,25-D3. Medium was changed every other day. On day 15, medium was changed again (FBS excluded), and cells were incubated with midazolam (2 μM applied to the apical compartment) for 60 min (n = 3 inserts per dose group) to determine 1′-hydroxymidazolam formation activity. Media from the apical and basolateral compartments and cultured cells were collected and stored at −80°C for analysis.
For a second experiment, confluent cells were cultured in differentiation medium supplemented with vehicle (ethanol, ∼ 1%), 0.1, or 1.0 μM of either 1,25-D3 or 19-nor-1,25-D2. Medium was changed every other day. On days 3, 7, 15, and 21, medium was changed again (FBS excluded), and cells were incubated with midazolam (2 μM) for 60 min to determine 1′-hydroxymidazolam formation activity (n = 3 inserts per group). Media from the apical and basolateral compartments and cultured cells were collected and stored at −80°C for analysis.
To assess CYP3A4 catalytic activity in Caco-2 cell culture, the extent of 1′-hydroxymidazolam (MDZ) formation was monitored using liquid chromatography combined with tandem mass spectrometry. Analysis was performed on a Micromass Quattro II tandem quadruple mass spectrometer (Micromass Ltd., Manchester, UK) coupled to a Shimadzu LD-10AD solvent delivery system and SIL-10AD autosampler (Shimadzu Scientific Instruments, Inc., Columbia, MD). Separations were achieved on an XBD-C8, 5-μm, 2.1- × 50-mm column (Zorbax; Agilent Technologies, Palo Alto, CA) using a water-methanol gradient that contained 0.1% acetic acid. The column effluent was subjected to mass spectral analysis by positive electrospray ionization. Using multiple reaction monitoring the transitions of m/z 341.9 to 202.9 andm/z 345.9 to 206.9, corresponding to a common fragment loss for the35Cl-d0-1′-OH MDZ and37Cl-d2-1′-OH MDZ (internal standard), respectively, were monitored. Peak area ratios of (1′-OH MDZ/d2-1′-OH MD) for samples were compared with those of the standard curve and unknown concentrations were interpolated. Statistical analysis of results from the different treatment regimens was performed using analysis of variance, followed by Bonferroni's pair-wise comparisons.
Immunoblot Analysis of CYP3A4 and Pgp in LS180 Cells.
LS180 cells growing at high density on tissue culture plastic were treated with various concentrations of 1,25-D3 for 48 h and then removed from the plate by scraping into phosphate-buffered saline. The cells were pelleted at 10,000g, and resuspended in storage buffer (100 mM potassium phosphate, pH 7.4, 1.0 mM EDTA, 20% glycerol, 1 mM dithiothreitol, 20 μM butylated hydroxytoluene, and 2 mM phenylmethylsulfonyl fluoride), and cell lysates were generated by sonication. Cell lysates (50 or 100 μg) were resolved on 6% Laemmli polyacrylamide gels for Pgp and CYP3A4, respectively, immunoblotted to nitrocellulose and developed with antibodies to human CYP3As and Pgp (Schuetz et al., 1996a) and appropriate secondary antibodies and developed with enhanced chemiluminescence (Amersham Pharmacia Biotech, Piscataway, NJ).
Northern Blot Analysis.
Total RNA was extracted from LS180 cells (Schuetz et al., 1988) and 20 μg was analyzed by Northern blot (Schuetz et al., 1996c). Equivalent loading of RNA samples and uniform transfer was assessed by analysis of ethidium bromide staining of the 18S and 28S ribosomal RNAs before and after transfer to membrane. The membrane was hybridized with a CYP3A cDNA (Schuetz et al., 1996a)
Electrophoretic Mobility Shift Assay.
Double-stranded,32P-labeled oligonucleotides representing the CYP3A4 PXR DNA binding sequence, an everted repeat with 6-bp spacer (hereafter referred to as CYP3A4-ER6) 5′-GATCAATATGAACTCAAAGGAGGTCAGTG-3′ or a mutant CYP3A4-ER6 5′-GATCAATATGAACTCAAAGGAATACAGTG-3′ (bolded bases disrupt PXR binding half-site) or a consensus VDRE 5′-GGCAGGTCATGGAGGTCAGTTC-3′ or mutated VDRE 5′-GGCAGAACATGGAGAACAGTTC-3′ was incubated with vitamin D receptor nuclear extracts in the presence or absence of VDR-specific antibody or approximately 200-fold molar excess of unlabeled oligonucleotides, according to the manufacturer's instructions. Electrophoretic mobility shift assay reagents were purchased from Geneka (Montreal, Quebec, Canada). Complexes were resolved by electrophoresis through a nondenaturing 4% polyacrylamide gel and analyzed using a Storm 860 PhosphoImager (Molecular Dynamics, Sunnyvale, CA).
Transient Transfection Studies.
LS180 cells (120 × 105 cells per 1.6-cm well, plated on day 1) were transfected on day 2 by calcium phosphate coprecipitation with 2 μg of CYP3A reporter plasmids. Cos-7 cells (300 × 105 cells per 2.2-cm well, plated on day 1) were cotransfected on day 2 by LipofectAMINE (Invitrogen) (4 μl/well) with 600 ng of TK(MH100)4-LUC and 400 ng of vector or pCMX-GAL-l-SXR plasmid, or were transfected with 200 ng of CYP3A4(ER6)3-TK-CAT, 100 ng TK-LUC, and 66 ng of pCMV-hVDR or pcDNA3 vector. 293 cells (200 × 105 cells per 1.6-cm well, plated on day 1) were cotransfected on day 2 by calcium phosphate coprecipitation with 500 ng of CYP3A4(ER6)3-TK-CAT, 100 ng TK-LUC, and 150 ng of pCMV-hVDR or pcDNA3 vector reporter plasmids. All cells were washed and incubated with fresh medium with and without 1,25-D3 18 h after transfection. Reporter activities were determined (Thottassery et al., 1999) 24 h later and normalized to either TK-LUC or cell protein. CAT activities were normalized to luciferase activity or cell protein.
Reverse Transcription-PCR: hPXR and hVDR.
Total RNA (5 μg) from LS180 cells treated for 48 h and Caco-2 cells treated for 2 weeks with or without 0.25 μM 1,25-D3 was reverse transcribed according to the manufacturer's instructions (Invitrogen). hPXR cDNA was amplified from first-strand cDNA by using oligonucleotides hPXR-U 5′-CAAGCGGAAGAAAAGTGAACG-3′ and hPXR-L 5′-CTGGTCCTCGATGGGCAAGTC-3′ under conditions described previously (Dotzlaw et al., 1999). hVDR cDNA was amplified using primers from bp 310 to 764 (S) 5′-CGGGAGATGATCCTGAAGCG-3′ and (AS) 5′-GTGAGGTCTCTGAATCCTGG-3′. The pSG5-hPXRΔATG cDNA and hVDR cDNA served as positive controls for amplification.
Concentration-Dependent Induction of CYP3A4 in LS180 and Caco-2 Cells.
Two types of immortalized human colon carcinoma cell lines were used for these studies. Treatment of LS180 cells with 1,25-D3 for 2 days resulted in a concentration-dependent increase in cellular CYP3A4 protein (Fig.1a) and CYP3A4 mRNA (Fig. 1b). The inductive response was detectable with 1 nM 1,25-D3, and this effect increased up to the maximum concentration employed—100 nM. Treatment of Caco-2 cells with 50 to 1000 nM 1,25-D3 for 2 weeks gave a similar hormone concentration-dependent response with a parallel increase in CYP3A4 protein (not shown) and in situ midazolam 1′-hydroxylation activity. An apparent maximum inductive effect was observed with 250 nM 1,25-D3 (Table 1). Interestingly, immunochemical detection of P-glycoprotein in LS180 cells (Fig. 1a) and in Caco-2 cells (not shown) increased in parallel to the change in CYP3A4 over the same respective 1,25-D3 concentration range.
Time-Course of Effect of Paricalcitol and Calcitriol on CYP3A4 in Caco-2 Cells.
1,25-D3 binds to the vitamin D receptor with a higher (∼3-fold) affinity than 19-nor-1,25-D2 (Takahashi et al., 1997). Caco-2 monolayers were cultured with 0.1 and 1 μM 1,25-D3 or 19-nor-1,25-D2for 3, 7, 15, and 21 days, and then midazolam 1′-hydroxylation activity was measured. Results presented in Table2 demonstrate induction of CYP3A catalytic activity by both hormones, and that 1,25-D3 was approximately 2-fold more effective than 19-nor-1,25-D2 at the 0.1 μM concentration but was equally effective at the higher 1 μM concentration. The inductive effect after treatment with 0.1 μM hormone seemed to be maximal by week 2 but was still increasing at week 3 with the 1 μM dose.
Transcriptional Activation of CYP3A-Luciferase Constructs by 1,25-D3.
LS180 cells were transiently transfected withCYP3A promoter-Luciferase, or SV-40-LUC or TK-LUC control plasmids. The CYP3A4 promoter contains a proximal ER-6 element (Barwick et al., 1996). The CYP3A23 promoter contains both ER-6 and DR-3 elements (Barwick et al., 1996). TheCYP3AP1 promoter [previously believed to represent theCYP3A5 promoter (GenBank S74700 and S74699)] (Finta and Zaphiropoulos, 2000), contains a PXRE with several nucleotide substitutions that render it incapable of binding PXR (Blumberg et al., 1998). Treatment of CYP3A4 and CYP3A23 promoter constructs with 1,25-D3 for 18 h resulted in 5- and 8-fold increases in luciferase activity above that of untreated cells (Fig. 2a). In contrast, the same 1,25-D3 treatment produced no change in luciferase activity above untreated controls in cells containing either the CYP3AP1 promoter construct containing the mutant PXRE, the SV40 construct, or the TK construct. Transcriptional activation of the CYP3A4 promoter construct was 1,25-D3 concentration-dependent with a significant 25% increase detected after incubation of 1 nM 1,25-D3 and a maximal 225% increase in activity observed after treatment with 10 nM 1,25-D3 (Fig.2b). Selective mutation of three sequential bases within the proximal ER-6 element of CYP3A4 (Fig. 2b, wild-type, TGAACT CAAAGG AGGTCA → mutant, TGAACT CAAAGG A ATA CA) completely abrogated the inductive effect of 1,25-D3 (Fig. 2c).
1,25-D3 Does Not Induce CYP3A by Modulating PXR Expression and VDR Is Expressed in LS180 and Caco-2 Cells.
Because the CYP3A4 ER6 promoter element has been shown to bind the PXR, we determined whether the effect of 1,25-D3on CYP3A4 transcription could be mediated indirectly by 1,25-D3 inducing PXR expression. PXR expression was compared in Caco-2 and LS180 cells cultured in the presence or absence of 1,25-D3. PXR mRNA was readily detected in LS180 cells, but its expression was unchanged by treatment with 1,25-D3 (Fig. 3a). PXR mRNA was undetectable in Caco-2 cells, regardless of whether the cells had been treated with hormone (Fig. 3a). In addition, a combined 2-week treatment of Caco-2 cells with 0.1 μM dexamethasone, a concentration shown to up-regulate the expression of PXR in primary human hepatocytes, and 10 μM rifampin, a potent activator of PXR, had no effect on midazolam 1′-hydroxylation activity (106 ± 5.6 and 99 ± 3.2%, respectively). We simultaneously examined VDR expression in RNA from these same cells and found that LS180 cells expressed at least 2-fold more VDR than Caco-2 cells, and that the amount of VDR mRNA was unchanged by treatment with 1,25-D3 in either cell line (Fig. 3b).
1,25-D3 Does Not Activate PXR.
To further rule out the possibility that 1,25-D3 was inducing CYP3A4 in LS180 cells by a non–VDR-dependent mechanism involving PXR, we determined whether this hormone could ligand activate hPXR. We cotransfected Cos-7 cells with a GAL-PXR/SXR construct containing only the ligand binding domain of PXR/SXR fused to the DNA binding domain of the yeast transcription factor GAL-4, along with a reporter containing the GAL4 response element fused to a luciferase reporter. 1,25-D3, at concentrations ranging from 0.1 to 100 nM, was incapable of ligand activating PXR and transcriptionally activating the CYP3A4 PXREs, whereas 10 μM rifampin readily increasedCYP3A4 transcriptional activity in a PXR-dependent fashion (Fig. 4).
Specific Binding of the VDR-RXR Heterodimer to theCYP3A4 PXRE-ER6.
To determine whether the VDR bound directly to the CYP3A4 PXRE-ER6, we performed electrophoretic mobility shift experiments.32P-labeled oligonucleotides containing either the consensus or mutated VDRE or wild-type or mutated PXR response element (CYP3A4-ER6) were incubated with nuclear extracts containing VDR and RXR in the absence or presence of competitor unlabeled oligonucleotides. VDR formed a specific complex with the labeled CYP3A4 PXRE-ER6 and with labeled consensus VDRE, but not with the labeled mutant CYP3A4-ER6 or mutant VDRE (Fig.5). Unlabeled oligonucleotides for the consensus VDRE and CYP3A4 PXRE-ER6, but not mutant VDR-response element and mutant CYP3A4 PXRE-ER6, reduced the intensity of labeled PXRE-VDR-RXR and VDRE-VDR-RXR complexes. Anti-VDR antibody produced a supershift of theCYP3A4- ER6-VDR-RXR complex, but had no effect on the nonspecific banding pattern associated with labeled mutated PXRE and VDR nuclear extracts (Fig. 5).
Cotransfection of VDR and CYP3A4 PXRE-Luc in Cos-7 or K293 Cells.
Because Cos-7 cells (a derivative of CV-1 cells) and K293 cells express RXRα, but are deficient in other nuclear hormone receptors, they are useful for studies of nuclear hormone receptor ligands (Lu et al., 2000). These cells were cotransfected with hVDR or pcDNA3 and a CYP3A4 (ER6)3-CAT reporter containing three copies of the CYP3A4 ER6 and no other CYP3A4 promoter sequences. Addition of 10 or 100 nM 1,25-D3 but not 20 μM rifampin to transfected Cos-7 cells produced a 70-fold increase in CYP3A4(ER6)3-CAT reporter activity, but only in those cells cotransfected with the human vitamin D receptor plasmid (Fig.6a). A similar requirement of the VDR for 1,25-D3 to induce the CYP3A4(ER6)3-CAT reporter was seen in transfected K293 cells (Fig. 6b) with a significant effect (60%) observed with 1 nM 1,25-D3, and maximal induction (390%) after incubation with 10 nM the hormone.
The specific content of CYP3A4 in both liver and small intestine is subject to considerable interindividual variability. In addition, interorgan expression of CYP3A4 is discordant. The cause of these apparent interindividual and interorgan differences in CYP3A4 regulation is largely unknown. Our studies demonstrate that the vitamin D nuclear receptor is an important signaling pathway for expression of CYP3A4 in mucosal villi of the small intestine. Several lines of evidence implicate the VDR as necessary for 1,25-D3 to induce CYP3A4. First, 1,25-D3 treatment was shown to increaseCYP3A4 mRNA content, CYP3A4 protein content and CYP3A-dependent midazolam hydroxylation in two immortalized cell lines that are accepted surrogate models for the human enterocyte. Moreover, CYP3A4 induction in LS180 cells was observed at total extracellular concentrations of 1,25-D3 (1–10 nM) similar to that needed to induce calcium transport in cell culture (Giuliano et al., 1991) although somewhat higher than the ∼0.1 nM concentration of 1,25-D3 that might be expected in vivo in arterial blood perfusing the small intestine (Bikle et al., 1984;Ebeling et al., 1992). Second, there is good correlation between the biologic potency of 1,25-D3 to induce CYP3A4 transcription, mRNA and protein production in LS180 cells, and its reported capacity to bind to the VDR. In addition, 1,25-D3 and 19-nor-1,25-D2induced CYP3A in Caco-2 cells with potencies directly related to their affinities for the VDR (affinity of 19-nor-1,25-D2 is one-third that of 1,25-D3) (Takahashi et al., 1997). Third, results from electrophoretic mobility shift assay and CYP3A4promoter transfection experiments demonstrated the specific binding of VDR-RXR to an ER-6 domain (nuclear hormone response element) of theCYP3A4 promoter and that this signal activated gene transcription. An additional line of evidence implicating the VDR as necessary for 1,25-D3 to induce CYP3A comes from transient transfection experiments of Cos-7 and 293 cells. These experiments demonstrated an absolute requirement for the hVDR for 1,25-D3 to transcriptionally activate theCYP3A4 ER6 elements. Again, a significant, stimulatory effect was detected at a relevant hormone concentration shown to increase VDR-mediated calcium uptake in cell culture (Giuliano et al., 1991).
These studies conclusively ruled out any role for PXR in mediating 1,25-D3 induction of CYP3A4 because, first, Caco-2 cells lack PXR and are resistant to the PXR inducing effects of dexamethasone, yet respond to 1,25-D3 with induction of CYP3A. Second, up-regulation of PXR by 1,25-D3 cannot be indirectly activating CYP3A4 expression in LS180 cells because 1,25-D3 had no effect on PXR protein expression in these cells. Third, 1,25-D3 was incapable of activating theCYP3A4 PXRE reporter in cells cotransfected with PXR.
Caco-2 cells were clearly less sensitive to the inductive effect of 1,25-D3 in comparison with LS180 cells. This discrepancy may reflect the dramatic difference in the number of 1,25-D3 binding sites within the two cell types. Indeed, VDR message levels were dramatically lower in the Caco-2 cells compared with the LS180 cells (2-fold higher band intensity in LS180 cells). Thus, in Caco-2 cells, the VDR B maxvalue at low 1,25-D3 concentrations may be insufficient to yield a detectable signal (increased transcription of CYP3A4 protein and activity). By analogy, the magnitude of glucocorticoid receptor transcriptional activation is directly proportional to the number of receptor molecules per cell (Vanderbilt et al., 1987). In addition, Caco-2 cells may show the lag in CYP3A induction (relative to LS180) because they require several days of 1,25-D3 treatment to fully differentiate into mature enterocytes. Indeed, the maximum number of 1,25-D3 binding sites is 2.6-fold greater in differentiated Caco-2 cells compared with undifferentiated ones (Giuliano et al., 1991) and increased vitamin D receptor levels enhance 1,25-D3-mediated gene expression and calcium transport in these cells (Shao et al., 2001). Because cells interpret signals in a context-dependent fashion, the observed cell-dependent dose response to 1,25-D3 may also be the result of a difference in expression of a critical coactivator or corepressor protein, such as the YY1 transcription factor shown to modulate the genomic response to 1,25-D3 (Raval-Pandya et al., 2001). It is also possible that the induction response in LS180s at high concentrations and some of the induction in Caco-2 cells involves nongenomic effects of 1,25-D3 (Dusso and Brown, 1998). Nevertheless, the rapid transcriptional activation of the endogenous CYP3A4 gene by near physiological concentrations of 1,25-D3 in LS180 cells strongly supports direct transactivation by VDR as the basis for increased CYP3A4 observed in vivo (Takahashi et al., 1997).
Our data demonstrate that the VDR-RXR complex binds to theCYP3A4 PXRE (ER6) to induce CYP3A transcription. Although the most common VDR binding sites are direct repeats of AGGTCA separated by three nucleotides (DR3), there is precedence for the VDR-RXR heterodimer to bind to a DR6 (Kahlen and Carlberg, 1994). Alternatively, the VDR-RXR heterodimer could be recognizing theCYP3A4 PXRE as an imperfect DR4. Although theCYP3A4 PXRE is an imperfect ER6 (TGAACT CAAAGGAGGTCA), it is also an imperfect DR4 (AACTCAAAGG AGGTCA), and VDR-RXR heterodimers bind even more avidly to DR4s than two DR3s (Quack and Carlberg, 2000). In addition, although we tested only the potential for the VDR-RXR complex to bind to the CYP3A4 proximal ER6, there is a distal enhancer module containing nuclear hormone receptor sites including a DR3 and adjacent ER6 at nucleotides −7273/−7286, −7733/−7719, and −7689/7672, respectively, in the CYP3A4 gene (Goodwin et al., 1999) that probably cooperates in mediating maximal VDR induction of CYP3A4.
The proposed regulation of intestinal CYP3A4 by VDR is complementary to the apparent role of PXR in regulating hepatic CYP3A4. VDR expression is more prominent in the human and rat small intestine, compared with liver (Berger et al., 1988; Stumpf, 1995), and the converse is true for PXR expression (Blumberg et al., 1998; Lehmann et al., 1998). Indeed, PXR was not involved in activation of the CYP3A4 promoter after treatment of Cos-7 and 293 cells with 1,25-D3, or in the 1,25-D3induction of CYP3A4 expression in Caco-2 cells that lacked PXR. Moreover, our results demonstrate that 1,25-D3does not ligand activate PXR.
The absence of PXR in Caco-2 cells helps explain the previously observed failure to induce CYP3A4 after treatment with rifampin and other known CYP3A4 inducers (Schmiedlin-Ren et al., 1997). Rifampin did not activate hVDR and CYP3A4 transcription in cotransfected Cos-7 and 293 cells (Fig. 4); thus, no effect should be expected. The deficiency of PXR in Caco-2 cells limits their use for studying induction phenomena, including other PXR targets such as MDR-1 (Synold et al., 2001). However, our data from LS180 and Caco-2 cell culture experiments (Fig. 1; Table 1) demonstrates that the MDR-1 gene and production of P-glycoprotein is also induced by 1,25-D3treatment. Presumably, transcriptional activation occurs through the binding of the VDR-RXR complex with PXR response elements in the 5′-flanking sequences of MDR-1 (Geick et al., 2001).
Our results demonstrate conclusively that formation of a 1,25-D3–VDR complex is sufficient for activation of CYP3A4 transcription and, thus, identify the vitamin D receptor as potentially important regulator of intestinal CYP3A4 expression. Therefore, interindividual differences in the expression of intestinal CYP3A4 could arise from known mutations in the vitamin D receptor (Hughes et al., 1988) and from differences in intestinal VDR content and circulating unbound 1,25-D3. Why does the vitamin D receptor induce CYP3A4 in intestinal cells? CYP3A4 is not involved in the metabolism of vitamin D and thus should not participate in an autoregulatory process. Perhaps it contributes to a protective enzymatic barrier to the influx of potentially toxic xenobiotic molecules. Indeed, vitamin D has been shown to enhance differentiation of intestinal cells (Halline et al., 1994). Moreover, one typically thinks of induction of CYP3A as a mechanism that has evolved to protect the organism from toxic chemicals, including toxins in the diet, by enhancing their elimination. For example, it has been speculated that enterocyte metabolism of aflatoxin-B1 by CYP3A4 may actually protect from dietary aflatoxins (Kolars et al., 1994a). An inverse association has been identified between intake of vitamin D and risk of colorectal cancer (Martinez et al., 1996). Thus, the role of vitamin D in protection from colorectal cancer could include the induction of CYP3A4 and the accelerated detoxification of toxins found in the diet.
- Received June 21, 2001.
- Accepted September 11, 2001.
This work was supported in part by National Institutes of Health Grants GM60346, ES08658, GM32165, and P30-CA21745 and by the American Lebanese Syrian Associated Charities and Abbott Laboratories.
- pregnane X receptor
- human pregnane X receptor
- steroid and xenobiotic sensor
- retinoid X receptor
- 1α,25-dihydroxy vitamin D3
- 19-nor-1α,25-dihydroxy vitamin D2
- vitamin D receptor binding element
- pregnane X receptor response element
- fetal bovine serum
- base pair(s)
- human vitamin D receptor
- simian virus 40
- thymidine kinase
- The American Society for Pharmacology and Experimental Therapeutics