Abstract
Glucocorticoid inducibility of the CYP3A23 gene is conferred by a multisite unit comprising binding sites for several members of the nuclear receptor superfamily of transcription factors, including the chicken ovalbumin upstream promoter-transcription factor COUP-TF, pregnane X receptor (PXR), and hepatocyte nuclear factor 4 (HNF-4). The presence of three binding sites, each of which interacts with more than one factor, contributes to the complexity of the CYP3A23 glucocorticoid-responsive region. Despite the glucocorticoid sensitivity of this gene, direct binding of ligand-activated glucocorticoid receptor (GR) to theCYP3A23 dexamethasone-responsive region (DexRE) is not required for induction. This study demonstrates that DexRE-2 is the key element within the CYP3A23 proximal promoter, conferring ligand sensitivity via its interaction with the PXR/RXRα heterodimer. The DexRE-1 and HNF-4 sites are not ligand-responsive, but are essential accessory elements required for full promoter inducibility. In addition to ligand-mediated activation of PXR, the overall induction response involves a GR-mediated stimulation of PXR and RXRα expression. Hence, the induction pathway can be divided into two stages. In stage one, maximal induction requires a GR-dependent increase in PXR and RXRα expression, and stage two is characterized by direct transcriptional activation of CYP3A23, which is dependent on ligand-activated PXR as well as accessory factors bound at the DexRE-1 and HNF-4 sites. Because multiple proteins bind at each element within the glucocorticoid-responsive region, factors not contributing to ligand responsiveness, such as chicken ovalbumin upstream promoter-transcription factor, may modulate the response through competitive interactions.
The cytochrome P450 (CYP) superfamily of heme-thiolate mono-oxygenases is involved in biosynthetic pathways (steroids, prostaglandins) as well as in the metabolism of both endogenous and foreign compounds (Nelson et al., 1996). Approximately 6 isoforms of the 40 found in humans metabolize most clinically administered drugs; therefore, direct substrate competition for a given isozyme presents a potential mechanism for drug interactions (Rendic and DiCarlo, 1997; Michalets, 1998). Expression of the xenobiotic metabolizing enzymes is regulated transcriptionally by several classes of compounds, including hormones, peroxisome proliferators, anticonvulsants, and polycyclic aromatic hydrocarbons; therefore, both substrates and compounds not metabolized by CYP can influence the cellular levels of these enzymes (Dogra et al., 1998). Glucocorticoid activation of the CYP3A subfamily of enzymes has long been a topic of clinical interest, because CYP3A4, the major human isoform in adults, metabolizes 50 to 60% of drugs (Wilkinson, 1996).
Studies aimed at elucidating the precise mechanism of CYP3Aactivation by glucocorticoids have predominantly focused on the rat isoforms, particularly the major glucocorticoid-inducible form,CYP3A23 (Schuetz and Guzelian, 1984; Schuetz et al., 1984;Gonzalez et al., 1985; Simmons et al., 1987; Ribeiro and Lechner, 1992;Cooper et al., 1993; Komori and Oda, 1994; Sidhu and Omiencinski, 1995). Insights gained from studying rat genes have led directly to a greater understanding of the regulation of CYP3A isoforms from various species (Wrighton et al., 1985; Barwick et al., 1996;Lehmann et al., 1998). Within the CYP3A23 proximal 5′-flanking region, three elements, dexamethasone response element 1 (DexRE-1), DexRE 2 (DexRE-2), and Site A, were shown to be essential for the full induction response (Huss et al., 1996). All of theCYP3A23 elements contain imperfect AGGTCA direct repeats and bind members of the nuclear receptor superfamily of ligand-activated transcription factors (Huss and Kasper, 1998; Quattrochi et al., 1998). Chicken ovalbumin upstream promoter-transcription factors (COUP-TFs) bind the two distal elements, DexRE-1 (−144 to −169) and DexRE-2 (−118 to −136). Hepatocyte nuclear factor 4 (HNF-4) binds to the proximal site, Site A (−85 to −110), and directly activates theCYP3A23 promoter in HeLa cells, which lack endogenous HNF-4 (Huss and Kasper, 1998). Interactions at each of the sites were demonstrated to correlate with functional activity in mutagenesis experiments; therefore, it is likely that these factors play a direct role in the response of CYP3A23 to glucocorticoids (Huss and Kasper, 1998). The glucocorticoid-responsive CYP3A genes from all other species examined contain homologous DexRE-1 elements. In fact, the DexRE-1 of the rabbit and human CYP3A6 andCYP3A4, respectively, confers dexamethasone responsiveness on a heterologous promoter (Barwick et al., 1996; Lehmann et al., 1998). The DexRE-2 and Site A elements, however, are present only in mouse and rat isoforms (Burger et al., 1992; Telhada et al., 1992;Hashimoto et al., 1993; Jounaı̈di et al., 1994; Barwick et al., 1996; Huss et al., 1996; Toide et al., 1997).
Interestingly, gene regulatory studies have ruled out the likelihood of direct glucocorticoid receptor (GR) involvement via binding to theCYP3A23 promoter (Quattrochi et al., 1995; Huss et al., 1996). However, whether GR plays an indirect role in the overall induction pathway is still a focus of debate (Schuetz and Guzelian, 1984; Schuetz et al., 1984; Burger et al., 1992; Huss et al., 1996). Early pharmacologic studies revealed an atypical profile forCYP3A23 induction compared with other glucocorticoid-responsive genes whose mechanisms are GR-dependent (Schuetz and Guzelian, 1984; Schuetz et al., 1984; Burger et al., 1992). These studies demonstrated that the response was specific for glucocorticoids, with the exception of pregnenolone 16-α carbonitrile (PCN); the CYP3A23 dexamethasone induction response was partially inhibited by RU486; a 100-fold higher glucocorticoid concentration was required to induce CYP3A23 than was necessary for maximal GR activation; and the potencies of various glucocorticoids for inducing CYP3A23 did not parallel their potencies for activating GR. This work provides a feasible model for GR involvement that is consistent with the pharmacologic characteristics of the response.
Recently, a novel orphan receptor isolated from mouse, the pregnane X receptor (PXR), was shown be activated by the CYP3A inducers PCN and dexamethasone (Kliewer et al., 1998). The PXR/RXRα heterodimer binds the dexamethasone-responsive region at DexRE-2 and mediates ligand-dependent activation of a thymidine kinase reporter construct containing multiple upstream copies of this element (Kliewer et al., 1998). This study investigates the role of PXR in activatingCYP3A23 promoter constructs in both CV-1 and H4IIE cells and examines the role of these response elements within the context of the wild-type promoter, rather than a heterologous system. These findings demonstrate, that although PXR plays a key role in mediating theCYP3A23 response to glucocorticoids, DexRE-1 and Site A, which bind other nuclear receptors, including COUP-TF and HNF-4, respectively, are also required for full induction. Although GR does not directly transactivate the CYP3A23 promoter (Quattrochi et al., 1995; Huss et al., 1996), GR is involved in the regulation ofPXR and RXRα gene expression. Finally, the capacity of the DexRE-1 and -2 elements from various species to bind PXR/RXRα is examined, and the nucleotide requirements for strong PXR/RXRα binding are precisely defined.
Experimental Procedures
Materials.
Oligonucleotides were synthesized by Integrated DNA Technologies, Inc. (Coralville, IA). Polymerase chain reactions were performed with a GeneAmp kit and Taqpolymerase (PE Applied Biosystems, Foster City, CA). Restriction enzymes, T4 ligase, and T4 polynucleotide kinase were from New England Biolabs (Beverly, MA). Radioactive isotopes, [α-32P]UTP (800 Ci/mmol) and [γ-32P]dATP (3000 Ci/mmol), were from NEN Life Science Products (Boston, MA). Luciferase reagents were from Promega (Madison, WI). Dexamethasone was purchased from Sigma (St. Louis, MO), and dexamethasone t-butylacetate from Research Plus, Inc. (Bayonne, NJ). RU486 and pregnenolone 16α-carbonitrile were from Biomol (Plymouth Meeting, PA).
Construction of Reporter Constructs.
Construction ofCYP3A23 deletion constructs has been described previously (Huss et al., 1996). The nucleotide numbering for CYP3A23constructs is based on a submitted CYP3A23 promoter sequence (Genbank accession number S82239). The mutant construct P3–175/TTG was generated by using the polymerase chain reaction overlap extension technique to make substitutions at positions −167 and −165 of the P3–175 deletion construct (Huss et al., 1996; Huss and Kasper, 1998). Construction of the heterologous promoter constructs has been described previously (Huss et al., 1996). To define the region of theCYP3A23 promoter cloned upstream of the thymidine kinase (TK) promoter (−110 to +50), the −170TK construct is identified by its 5′ base and has nucleotide −60 as its 3′ terminus. The other constructs are identified by their 5′/3′ termini. Constructs with multiple element copies were made using oligonucleotides designed such that the annealed oligonucleotides had overhanging BamHI andKpnI sites; these were ligated with theBamHI/KpnI-digested TK-Luc vector. All constructs were sequenced by the dideoxy chain termination method with a T7 Sequenase V2.0 DNA-sequencing kit or with BigDye sequencing reagents according to the manufacturer's protocol (PE Applied Biosystems).
Cell Culture, Transfection, and Luciferase Assays.
H4IIE cells were maintained, transfected, and treated as described previously (Huss et al., 1996; Huss and Kasper, 1998). In cotransfection experiments, 2 μg/ml of expression vector was used. CV-1 cells were generously provided by S. A. Kliewer (Glaxo-Wellcome); overnight (14 h) transfections of 2 μg/ml reporter and 0.1 μg/ml expression vector were performed using Lipofectin (Life Technologies, Grand Island, NY). Transfected CV-1 cells were treated for 24 h with inducers. Luciferase assays were carried out as described previously (Huss and Kasper, 1998). Activities from CV-1 experiments were normalized to β-galactosidase activity (0.4 μg/ml transfected plasmid).
Electrophoretic Mobility Shift Assays.
Probes were generated by annealing complementary oligonucleotides. Sequences forCYP3A23 DexRE-1 and DexRE-2 have been described (Huss and Kasper, 1998). Other probes included 3A2DexRE-1:5′AGAATGTTAGCTCAAGAAGGTCAAAGAAGCTGT3′; 3A2DexRE-2:5′TGTAGATGAACTTTATGA- ACTGTTTAGG3′, 5′TAAACAGTTCATAAAGTTCATCTACAGC3′; 3A6DexRE-1:5′CAGCACATGAACTCAGAGGAGGTCACCACGGAT- T3′; 3A4DexRE-1:5′GAATATGAACTCAAAGGAGGTCAGTGAGT3′; 3a11DexRE-1:5′CAGAATGTTAGCTCAAAGTAGGTCAAGTTGGG- CT3′; 3a11DexRE-2:5′CTGTGGATGAACTATACGAACTGCCTAG3′; and TTGDexRE-1:5′CCCAGAATTTGAACTCAAAGGAGGTCAAAA- TAG3′ (exact complementary oligonucleotides were used for opposite strands unless otherwise indicated). Nuclear extracts from H4IIE cells were prepared using a protocol described by Dignam, with modifications (Dignam et al., 1983; Ausubel et al., 1987). Briefly, cells are incubated in hypotonic buffer (10 mM HEPES, pH 7.6, 1.5 mM MgCl2, 0.5 mM dithiothreitol, 0.2 mM phenylmethylsulfonyl fluoride) and disrupted with a glass homogenizer. The nuclei are collected by centrifugation (3300g, 15 min), and proteins are extracted stepwise using low-salt buffer (20 mM HEPES, pH 7.6, 25% glycerol, 1.5 mM MgCl2, 20 mM KCl, 0.2 mM EDTA, 0.2 mM phenylmethylsulfonyl fluoride, 0.5 mM dithiothreitol) and high-salt buffer (substitute 1.2 M KCl in low-salt buffer). Recombinant PXR was synthesized from the pSG5-PXR cDNA clone (kindly provided by S. A. Kliewer) in rabbit reticulocyte lysate, using TNT in vitro coupled transcription/translation system (Promega). The GST-RXRα fusion protein (provided by J. E. Mertz, McArdle Laboratory, Madison, Wisconsin) was purified from 1 liter JM109 bacterial cultures, using glutathione-Sepharose 4B (Amersham Pharmacia Biotech), according to the manufacturer's protocol.
RNase Protection Assays.
Antisense probes were synthesized using T7 RNA polymerase and corresponded to the 151 to 1101 region of mouse PXR, the 850 to 1740 region of human COUP-TFII (cDNA clone provided by M. J. Tsai, Baylor College of Medicine, Houston, TX), the 1340 to 1764 region of rat HNF-4 (cDNA clone provided by F. M. Sladek, University of California-Riverside), and a 334-nucleotide mouse β-actin probe (provided in a MAXIscript kit; Ambion, Austin, TX) in the presence of [32P]UTP (β-actin probe was labeled at 50-fold lower specific activity). Full-length probes were purified by denaturing polyacrylamide gel electrophoresis. Isolated probes were incubated with 10 μg total RNA from 5 × 107 H4IIE cells isolated with a Qiagen RNeasy Midi kit. Overnight hybridization of probe and RNA at 42°C and RNase (A/T1) digestion (30 min) were performed according to the manufacturer's protocol (RPAII kit; Ambion). RNA digestion products were resolved on 5% denaturing polyacrylamide electrophoresis gels and visualized by phosphorimage analysis. Band intensities were measured by Imagequant software and normalized to the β-actin reaction included in each analysis.
Results
PXR/RXRα Binds to Distinct Elements in CYP3A Genes from Various Species.
Previous work has established that the dexamethasone response elements DexRE-1 and DexRE-2 are required for glucocorticoid activation of CYP3A23 (Huss and Kasper, 1998). DexRE-2 binds several nuclear receptors, including COUP-TFI and II, and the recently described pregnane X receptor (PXR) (Huss and Kasper, 1998; Kliewer et al., 1998; Quattrochi et al., 1998). PXR (also known as steroid and xenobiotic receptor (SXR) in humans) binds as a PXR/RXRα heterodimer to AGTTCA hexamers arranged as direct, inverted, or everted repeats (Bertilsson et al., 1998; Blumberg et al., 1998;Lehmann et al., 1998). Homologous elements are present in rodent, human, and rabbit CYP3A genes (Fig.1A) (Hashimoto et al., 1993; Tukey, 1995;Toide et al., 1997). A direct repeat of AGTTCA hexamers (DR3) comprises the DexRE-2 of the rodent isoforms, although the mouseCyp3a11 element differs by one mismatch. In contrast, the human CYP3A4 and rabbit CYP3A6 lack a DexRE-2 homology region. Their DexRE-1 elements, however, contain an AGTTCA everted repeat separated by six nucleotides (IR6) and have been shown to confer activation by glucocorticoids on a heterologous promoter (Barwick et al., 1996). The IR6 components of the rodent DexRE-1 elements are disrupted within their upstream hexamers, but their ability to support glucocorticoid activation or interact with PXR/RXRα has not been tested.
To further investigate the role of PXR in glucocorticoid regulation of these CYP3A isoforms, binding of PXR/RXRα to these elements was analyzed by gel shifts (Fig. 1B). The DexRE-2 elements of rat CYP3A2 and CYP3A23 and mouseCyp3a11 bind the PXR/RXRα heterodimer, whereasCYP3A4 and CYP3A6 bind PXR/RXRα at DexRE-1. The mouse DexRE-2 (Fig. 1B, lane 12) displayed a much weaker binding for PXR/RXRα compared with the rat DexRE-2 probes under identical conditions (Fig. 1B, lanes 4 and 8), presumably because of the single DR3 mismatch. The functional effects of this relatively weak PXR/RXRα binding on the Cyp3a11 induction response are not known. Although the upstream hexamer of the IR6 within CYP3A23DexRE-1 (AGTTAA) differs by only one mismatch from theCYP3A4 and CYP3A6 elements, PXR/RXRα binds less strongly to the CYP3A23 DexRE-1 (Fig. 1B, lane 2) than to the CYP3A4 and CYP3A6 DexRE-1 elements (Fig. 1B, lanes 14 and 16). The CYP3A23 DexRE-1 element also binds PXR/RXRα less strongly than CYP3A23 DexRE-2 (Fig. 1B, lane 4). As might be expected, the CYP3A2 DexRE-1, with two mismatches in the IR-6 upstream hexamer, has virtually no affinity for PXR/RXRα (Fig. 1B, lane 6).
A Single Nucleotide Mismatch Confers Differential Binding and Function on Human and Rat DexRE-1 Elements.
Mutagenesis experiments were performed to determine whether the single nucleotide mismatch within DexRE-1 that gave rise to the differences in PXR/RXRα binding between CYP3A4 and CYP3A23 also affected the glucocorticoid response. A mutant, designated mP3–175/TTG, was made that substituted a G for T at position −165 of the P3–175 construct, making the CYP3A23 IR6 sequence identical with that of CYP3A4 (Fig. 2). In the cloning process, the G at position −167 of CYP3A23, which is outside the IR6 consensus, was changed to T and therefore does not correspond to either CYP3A4 or CYP3A23 in that position. The mP3–175/TTG mutant displayed an enhanced dexamethasone induction response 2.5 times that of the 15-fold induction exhibited by wild-type constructs P3–175 and P3–210 and 15 times that of P3–144, which lacks the DexRE-1 (Fig. 2A). When binding was assessed, the mutant DexRE-1 oligonucleotide displayed a strong affinity for PXR/RXRα compared with the wild-type CYP3A23element (Fig. 2B, lanes 2 and 6). These results suggest that the single nucleotide difference between the DexRE-1 IR6 of CYP3A23 and that of CYP3A4 does account for the observed difference in PXR/RXRα binding affinity. Hence, in the case of wild-typeCYP3A23 DexRE-1, the imperfect IR6 renders the element unable to interact strongly with the ligand-receptor complex and, therefore, to directly mediate ligand activation (see below).
Multimerized CYP3A23 DexRE-2 Can Support Ligand Activation through PXR.
To determine whether PXR/RXRα binding to DexRE-2 correlated with PXR-dependent transcriptional activation, the CYP3A23elements were examined individually for their ability to confer glucocorticoid responsiveness on the thymidine kinase promoter. Figure3A shows the results of PXR cotransfection experiments in CV-1 monkey kidney cells, using constructs in which each of the three elements within theCYP3A23 dexamethasone-responsive region was multimerized. Consistent with previous observations (Kliewer et al., 1998), a 6-fold, PXR-dependent induction by 10 μM dexamethasonet-butylacetate (DtBu) was shown for the DexRE-2-containing construct. In contrast, neither the CYP3A23 DexRE-1 nor Site A constructs displayed responsiveness to DtBu in the presence of PXR. Site A supports basal and dexamethasone-induced activity through the binding of HNF-4 and has been shown to also bind the upstream stimulatory factor, a bHLH/leucine zipper transcription factor (Huss et al., 2000). Notably, the DexRE-1, despite the fact that it shows measurable binding affinity for PXR/RXRα, did not support PXR-dependent induction by DtBu in CV-1 cells.
The DexRE-1 and DexRE-2 multimer constructs were further characterized in H4IIE cells, which support the CYP3A glucocorticoid response, for their ability to respond to dexamethasone independent of PXR cotransfection (Fig. 3B). As shown previously, single copies of the DexRE-1 or DexRE-2 were unable to confer inducibility on the TK promoter (Huss et al., 1996). However, a construct containing multiple copies of DexRE-2 displayed more than an 11-fold induction response by dexamethasone, whereas no response was observed with a multimerized DexRE-1 construct. This further supports our hypothesis that DexRE-2, but not DexRE-1, directly mediates ligand-dependent activation ofCYP3A23 through an interaction with PXR/RXRα. The DexRE-1, which is necessary for full glucocorticoid induction ofCYP3A23, may contribute to the response as an accessory factor binding site.
Factors in Addition to PXR Are Required for FullCYP3A23 Induction in CV-1 Cells.
Thus far, the ability of PXR to directly mediate ligand-dependent activation has been studied using multimerized constructs (Kliewer et al., 1998) (Fig. 3A). These conditions do not reflect the natural organization of elements in the wild-type CYP3A genes, because CYP3A23 as well as the human (CYP3A4), rabbit (CYP3A6), and mouse (Cyp3a11) isoforms contain only single copies of their PXR/RXRα binding elements. Indeed, the high-magnitude induction response observed in hepatic cells was not obtained in CV-1 cells cotransfected with PXR and constructs containing single copies of the essential elements (Fig. 4). Compared with the multimerized DexRE-2 construct that is activated 6-fold by DtBu, a single-copy construct, 144/110TK, is activated by only 2-fold under the same conditions (Fig. 4). Similarly, −170TK, which contains the entire CYP3A23 glucocorticoid-responsive region (−60 to −170), mediates only a 2- to 3-fold PXR-dependent induction by DtBu. Hence, PXR/RXRα cannot mediate a full glucocorticoid induction response in the absence of additional transcription factors that bind at DexRE-1 and Site A, factors that are presumably absent from CV-1 cells. We predict that a higher magnitude response would be observed if the additional trans-acting proteins binding at Site A and DexRE-1 were present in the system.
PXR/RXRα Do Not Enhance the CYP3A23 Induction Response to Dexamethasone in H4IIE Cells.
To determine whether increased expression of PXR could enhance the level of glucocorticoid activation observed in a cell line replete with all necessarytrans-acting factors, cotransfection experiments were performed in H4IIE cells (Fig. 5). Cotransfection of PXR/RXRα activated CYP3A23 promoter constructs in the absence of added ligand. This so-called ligand-independent increase in activity required the presence of both the DexRE-1 and DexRE-2 elements (Fig. 5A) and, furthermore, required cotransfection of both PXR and RXRα (data not shown). The addition of 10 μM dexamethasone elicited a much lower fold induction over untreated PXR/RXRα cotransfected controls (3-fold) than the 15-fold induction typically observed in the absence of cotransfected PXR/RXRα (Fig. 5B). This graph, in which data are reported relative to P3–210 activity without PXR/RXRα or dexamethasone treatment, clearly shows a biphasic response, in which P3–210 is activated 6-fold by PXR/RXRα, then further induced 2.5-fold by the addition of 10 μM dexamethasone. The resulting activity level is equivalent to that observed on treatment of non-cotransfected cells with 10 μM dexamethasone. No activation by PXR/RXRα or dexamethasone is observed with the P3–110 construct, which lacks DexRE-1 and DexRE-2.
Expression of PXR and RXRα Is Increased by Dexamethasone Treatment in H4IIE Cells.
Because an increase in PXR/RXRα alone can mediate CYP3A23 activation, presumably by mediating the response to an endogenous ligand, the effect of glucocorticoids on PXR expression was evaluated. RNase protection assays were performed to measure PXR message levels in untreated and dexamethasone-treated H4IIE cells. For comparison, the effect of dexamethasone treatment on the expression of COUP-TFII, which binds to DexRE-1 and DexRE-2, and on HNF-4, which binds to Site A, was determined as well. Increased PXR expression was observed with 10 μM dexamethasone, which was attenuated by cotreatment with RU486, a GR antagonist (Fig.6A). Dose-response experiments revealed that peak PXR mRNA induction was achieved at 0.1 to 1.0 μM concentrations of dexamethasone (Fig. 6B). In addition, a time course experiment revealed that PXR message reached maximum levels after 7 h and remained high through 24-h dexamethasone treatment (Fig. 6B). None of the treatment levels caused a large change in COUP-TFII or HNF-4 message levels, although at the highest dexamethasone concentration, a 25% drop in COUP-TFII and HNF-4 message was observed (data not shown). Finally, PXR expression was unaffected by 10 μM pregnenolone16α-carbonitrile (PCN), a PXR agonist and CYP3A23 inducer, excluding the possibility of autoregulation through the activation of PXR (data not shown).
The effects of dexamethasone treatment on RXRα expression were also determined. Immunoblotting was performed with antibody specific for RXRα, using nuclear extracts from either control or dexamethasone-treated H4IIE cells. An obvious increase in RXRα protein was observed in cells treated with 10 μM dexamethasone (data not shown). These results are in agreement with the reported induction of RXRα (but not RXRβ or RXRγ) message levels in H4IIE cells treated with 0.5 μM dexamethasone (Wan et al., 1994; Steineger et al., 1997).
Differential Dose Response of CYP3A23 and Mouse Mammary Tumor Virus-LTR to Dexamethasone.
A well-characterized feature of the CYP3A23 response to glucocorticoids that distinguishes it from typical GR-mediated responses is the high concentration of dexamethasone required for maximal induction (Schuetz et al., 1984;Schuetz and Guzelian, 1984). The dose curves shown in Fig.7 demonstrate this point by displaying the CYP3A23 dose response relative to that of mouse mammary tumor virus-LTR (MMTV-LTR). At submicromolar dexamethasone concentrations, CYP3A23 was induced by less than 50% of the maximum, which is achieved at 10 μM dexamethasone. The CYP3A23 dose-response curve is clearly shifted to higher ligand concentrations when compared not only to that of MMTV-LTR (Fig.7), which is a GR-regulated promoter, but also to the dose curve reported for PXR message induction (Fig. 6B). At submicromolar dexamethasone concentrations, PXR message and MMTV-LTR transcriptional activity are maximally induced, suggesting a common mode of regulation. Hence these data support GR involvement in PXR induction and, therefore, an indirect role for GR in the CYP3A23transcriptional response to glucocorticoids.
Submicromolar Dexamethasone Concentrations Are Synergistic with 10 μM PCN.
The results implicating GR in the CYP3A23response directly address a key observation concerning the synergisticCYP3A23 response to combined PCN and dexamethasone treatment (Burger et al., 1992; Quattrochi et al., 1995). Cotreatment with 10 μM PCN and submicromolar dexamethasone (0.1 μM) resulted in transcriptional activity exceeding that reached with either individual treatment. This effect was demonstrated for the P3–210 construct in H4IIE cells (Fig. 8). When administered individually, 10 μM PCN and 0.1 μM dexamethasone each elicited an approximate 3-fold induction response, whereas when cells were cotreated with both inducers a synergistic activation (18-fold) was observed. The resulting level of activation by cotreatment is equivalent to that normally seen with 10 μM dexamethasone alone. One possible explanation for these results is that low concentrations of dexamethasone induce expression of PXR and RXRα, and, in turn, PXR is activated directly by PCN, at a high concentration. In CV-1 cotransfection experiments, DexRE-2 multimer constructs were strongly activated by PCN in the presence of PXR, verifying that PCN is a potent PXR ligand (Kliewer et al., 1998, and data not shown).
Discussion
This study advances our understanding of the multisite regulatory unit that mediates CYP3A23 glucocorticoid inducibility (Huss and Kasper, 1998) by defining precisely the roles of individual elements within the glucocorticoid-responsive region. DexRE-2, through its ability to bind PXR/RXRα, imparts ligand sensitivity to theCYP3A23 glucocorticoid-responsive region. This is supported by CV-1 experiments in which PXR-dependent activation by glucocorticoids was mediated solely through the DexRE-2. Neither DexRE-1 nor Site A supported or enhanced the response in this experimental system, presumably because CV-1 cells do not express the factors binding at these sites. We propose that DexRE-1 and Site A act with DexRE-2 as accessory factors to maximize the induction response. DexRE-1 and Site A bind, respectively, the nuclear receptors COUP-TF and HNF-4 but do not bind PXR/RXRα and cannot directly support the induction response. However, they are essential for the full response observed in H4IIE cells (Huss et al., 1996; Huss and Kasper, 1998). By acting together, the elements optimize interactions with general transcription factors, thereby conferring full activation in response to glucocorticoids. Additional support for these conclusions would be gained by showing that the full response can be reconstituted in CV-1 cells on the expression of transcription factors that bind at DexRE-1 and Site A (i.e., COUP-TF, complex B protein, HNF-4, etc.).
These findings, when considered with the GR-dependent regulation of PXR and RXRα, form the basis for the mechanistic model presented in Fig.9. According to this hypothesis, the response to glucocorticoids is divided into two steps: one mediated by submicromolar glucocorticoid concentrations and the other triggered at high concentrations of PXR ligands. At low dexamethasone levels, activation of GR causes an increase in cellular PXR and RXRα levels. This is predicted to increase the occupancy of DexRE-2 by the PXR/RXRα heterodimer to the exclusion of COUP-TF binding. The increased occupancy of DexRE-2 by the PXR/RXRα complex is predicted to elicit a moderate transcriptional activation, based on experimental results showing that PXR/RXRα cotransfection directly activated CYP3A23 promoter constructs in H4IIE cells. This activation in the absence of added ligand may, in fact, be due to the presence of endogenous PXR ligand in H4IIE cells. Previous work has established that the most likely endogenous candidates are pregnanes (Kliewer et al., 1998).
Based on several observations, we propose that the first stage is GR-dependent. First, the dose response for PXR message induction by dexamethasone paralleled that of an MMTV-LTR-driven reporter but was distinct from the CYP3A23 dose response. Second, cotreatment with 10 μM dexamethasone and 10 μM RU486 blocked the expected increase in PXR expression observed with dexamethasone alone. Finally, cellular levels of RXRα protein increased in response to dexamethasone.
The second stage of the mechanism involves direct activation of PXR by high concentrations (10 μM) of PXR-specific ligands, such as dexamethasone or PCN. The activated receptor, in cooperation with DexRE-1 and Site A bound factors, transactivates the downstream promoter. Previous characterization of the induction response in H4IIE cells, utilizing 10 μM dexamethasone, simultaneously activated both pathways and did not permit differentiation of the two stages (Huss and Kasper, 1998). To separate the GR-mediated stage of the pathway from the PXR-mediated stage, PCN, which is a ligand for PXR but not for GR, was used. Cotreatment of H4IIE cells with 0.1 μM dexamethasone activated GR, whereas 10 μM PCN directly activated PXR. Activation of both stages of the pathway resulted in a synergistic transcriptional response for CYP3A23 constructs containing the glucocorticoid-responsive region. Consistent with our hypothesis, the response to cotreatment was of the same magnitude as observed with high concentrations of dexamethasone alone. Interestingly, this synergistic response to low-dose glucocorticoids and high-dose PCN is a well-known characteristic of CYP3A23 regulation, and, until now, it was thought to serve as evidence that the two inducers acted through different mechanisms (Heuman et al., 1982; Schuetz and Guzelian, 1984;Quattrochi et al., 1995). Our model provides a feasible and testable hypothesis for how this synergy occurs.
Because multiple proteins bind to DexRE-1 and DexRE-2, competition for binding among trans-acting factors will likely occur, creating a potential for functional antagonism. This proposal is particularly relevant with respect to COUP-TF, because it binds at DexRE-1 along with another protein(s) in complex B, and at DexRE-2, which is also the PXR/RXRα binding site (Huss and Kasper, 1998). COUP-TF frequently represses transcription via several mechanisms, including direct competition with other nuclear receptors for their binding sites (Liu et al., 1993; Miyata et al., 1993; Galson et al., 1995; Leng et al., 1996; Tsai and Tsai, 1997). Because PXR/RXRα and COUP-TF bind at DexRE-2, the ability of PXR/RXRα to occupy the site and to activate CYP3A23 will be influenced by cellular levels of COUP-TF, PXR, and RXRα and the relative binding affinities of the complexes. COUP-TF binding would also be predicted to influence DexRE-1 function, because competition gel-shift experiments using recombinant COUP-TF showed that COUP-TF binding and complex B binding are mutually exclusive (data not shown). Although the relative importance of COUP-TF binding versus complex B binding at DexRE-1 has yet to be determined, binding of both complexes correlates with function (Huss and Kasper, 1998).
Considering that GR is indirectly involved in the CYP3A23response through PXR/RXRα induction and that PXR directly mediates ligand-dependent transcriptional responsiveness, results from early pharmacologic studies on the CYP3A induction response might be understood in light of this new model (Schuetz et al., 1984; Schuetz and Guzelian, 1984; Burger et al., 1992). For instance, the high concentration of dexamethasone or PCN typically required for transcriptional activation of CYP3A23 reflects the relatively low potency of these ligands for activating PXR (Kliewer et al., 1998). The rank order potencies of various glucocorticoids for inducing CYP3A23 do not parallel their potency for activating GR, but instead may parallel their relative abilities to activate PXR. With respect to the observation that RU486 partially inhibits dexamethasone induction of CYP3A23 (Burger et al., 1992), a GR antagonist would block the GR-dependent increase in PXR and RXRα expression; however, because RU486 does not antagonize stage two, a total inhibition would not be expected. Hence moderate activation would still occur through PXR/RXRα. Finally, the mechanism predicts that GR activation should cause moderate CYP3A23induction of the same magnitude as that caused by PXR and RXRα cotransfection in H4IIE cells. To bypass the use of glucocorticoids that might also act as PXR ligands, a GR mutant, rendered constitutively active by truncation at the ligand binding domain, was able to specifically activate a glucocorticoid-responsiveCYP3A23 reporter construct but not a noninducible construct in H4IIE cells (data not shown).
With respect to PXR regulation of other CYP3A family members, there are interesting species differences among genetic elements through which PXR acts. The abilities of homologous DexRE-1 or DexRE-2 elements from rabbit, human, mouse, and rat CYP3Agenes to interact with the PXR/RXRα complex were compared. A dichotomy exists between the rabbit (CYP3A6) and human (CYP3A4) isoforms, which bind PXR/RXRα at DexRE-1, and the rodent isoforms, CYP3A23, CYP3A2, andCyp3a11, which bind PXR/RXRα at their DexRE-2 elements. This differential binding reflects functional differences between the elements. The CYP3A23 DexRE-2 was required for ligand-dependent activation in PXR-expressing CV-1 cells, whereas the DexRE-1 could not support an induction response. In contrast, DexRE-1 of CYP3A4 and CYP3A6 directly mediates ligand-dependent induction (Barwick et al., 1996). In the human and rabbit genes, DexRE-1 contains the same imperfect DR4 asCYP3A23 (AACTCA(n)4AGGTCA), but more important is the IR6 (TGAACT(n)6AGGTCA), because the CYP3A23 DexRE-1 could be converted to a PXR/RXRα binding site by changing the upstream hexamer of the IR6 to match that of CYP3A4. The CYP3A23 reporter construct containing this mutation displayed enhanced responsiveness to glucocorticoids because of the creation of a PXR/RXRα binding site in addition to the native DexRE-2 site. Our results are consistent with the findings that PXR/RXRα binds to hexamers of AGTTCA rather than to typical AGGTCA elements but displays no strong preference for a specific hexamer arrangement (Blumberg et al., 1998; Lehmann et al., 1998). Therefore, evidence supports the hypothesis that PXR/RXRα is mediating ligand-dependent activation through DexRE-2 in the rodentCYP3A genes, but through DexRE-1 in the human and rabbit genes, but that in the rat CYP3A23 gene additional accessory sites are essential for full induction.
Acknowledgments
We thank Dr. Steven A. Kliewer for making the original PXR.1 mouse cDNA available and for his helpful comments, as well as Dr. Cary Weinberger (National Institutes of Health) for the pCMV-RXRα encoding mouse RXRα. In addition, we thank Dr. Nancy Thompson (McArdle Laboratory) for the gift of Tata-binding protein monoclonal antibody, and Dr. Janet Mertz (McArdle Laboratory) for providing the GST-RXRα construct. We thank Dr. Anna Shen, Dr. Kathleen O'Leary, and John Sheehan for critical evaluation of the manuscript. We are grateful to Kristen Adler and Mary Jo Markham for preparing the manuscript.
Footnotes
- Received December 20, 1999.
- Accepted March 20, 2000.
-
Dr. Charles B. Kasper, Department of Oncology, McArdle Laboratory for Cancer Research, 1400 University Ave., Madison, WI 53706. E-mail: kasper{at}oncology.wisc.edu
-
↵1 Present address: Center for Cardiovascular Research, Department of Internal Medicine, Washington University School of Medicine, Box 8086, 660 S. Euclid, St. Louis, MO 63110.
-
This work was supported by National Institutes of Health Grants CA22484 and CA0920. J.M.H. was supported by National Institutes of Health Grant T32-CA-09135.
Abbreviations
- CYP
- cytochrome P450
- CMV
- cytomegalovirus
- COUP-TF
- chicken ovalbumin upstream promoter-transcription factor
- CYP3A
- cytochrome P450 3A subfamily
- ds
- double-stranded
- DexRE-1
- dexamethasone response element 1
- DexRE-2
- dexamethasone response element 2
- FCS
- fetal calf serum
- GR
- glucocorticoid receptor
- HNF-4
- hepatocyte nuclear factor 4
- Me2SO
- dimethyl sulfoxide
- MMTV
- mouse mammary tumor virus
- PCN
- pregnenolone 16α-carbonitrile
- PXR
- pregnane X receptor
- RLU
- relative light units
- SV
- simian virus
- TK
- thymidine kinase
- RXR
- retinoid X receptor
- The American Society for Pharmacology and Experimental Therapeutics