| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Division of Pharmacology/Neurobiology, Biozentrum of the University of Basel, Basel, Switzerland
Abstract I. Introduction II. Drug-Mediated Induction of Cytochromes P450 A. Induction of CYP2Bs, CYP2Cs, and CYP3As by Drugs and Xenobiotics III. Drug-Response Elements in Inducible Cytochrome P450 Genes A. CYP102/106 in Bacillus megaterium B. CYP6 in Insects C. CYP2H1/2, CYP3A37, and CYP2C45 in Chicken D. CYP2Bs, CYP3As, and CYP2Cs in Mammals E. Other Mammalian Drug-Inducible Cytochromes P450 IV. Nuclear Receptors Involved in Drug Induction of Cytochromes P450 A. Constitutive Androstane Receptor B. Pregnane X Receptor C. The Evolution of Xenosensors: Lessons Learned from the Chicken Xenobiotic Receptor D. Structure of the Xenosensors E. Other Target Genes of Pregnane X Receptor and Constitutive Androstane Receptor V. Endogenous Roles of the Xenosensors A. Receptor Cross Talk in Hepatic Drug Induction VI. Clinical Relevance of Induction A. Altered Kinetics of Drugs B. Genetic Variants of Pregnane X Receptor and Constitutive Androstane Receptor VII. Open Questions A. Mechanisms of Constitutive Androstane Receptor Translocation and Activation B. Cofactors Involved in Pregnane X Receptor- and Constitutive Androstane Receptor-Mediated Signal Transduction C. The Xenosensors as Drug Targets D. The Mystery of How Cells Recognize Phenobarbital-Type Inducers VIII. Outlook
 |
Abstract |
|---|
 Top Next |
|---|
|
I. Introduction |
|---|
|
TopPreviousNext |
|---|
A gene superfamily of heme proteins, the cytochromes P450 (P4502), encodes for the main enzymatic system for metabolism of lipophilic substrates of diverse structures (Nelson et al., 1996
; Nebert and Russell, 2002). P450s are important in the oxidative, peroxidative, and reductive metabolism of numerous endogenous compounds including steroids, bile acids, fatty acids, prostaglandins, leukotrienes, biogenic amines, and retinoids (Waxman and Azaroff, 1992). Together with dehydrogenases, reductases, and oxidases, P450s belong to the group of enzymes in the hepatic detoxification system that are responsible for primary modifications of lipophilic compounds (phase I reactions) (Ziegler, 1994). With the help of reducing equivalents from NADPH cytochrome P450 oxidoreductase, P450s catalyze mono-oxygenase reactions of lipophilic compounds allowing subsequent use of the attached hydroxyl group as a reactive group that can be used by other so-called phase II enzymes for further modifications. Phase II reactions consist mainly of glucuronidation, sulfation, attachment of glutathione, methylation, N-acetylation, or conjugation with amino acids. In addition, esterases, amidases, imidases, epoxide hydratases, or other hydrolytic processes increase the hydrophilicity of xenobiotic compounds (Jakoby, 1994). Finally, the intracellular levels of both parent drugs and their metabolites are regulated by transporter proteins, sometimes called phase III enzymes, localized on the sinusoidal and the apical membrane of hepatocytes, the intestine, and the kidney (Stieger and Meier, 1998; Muller, 2000; Suzuki and Sugiyama, 2000; Bohan and Boyer, 2002).
In higher animals, the number of expressed P450s is in the range of 50 to 80 with 57 P450s and 19 P450 pseudogenes known in the human genome (Nelson et al., 1996; Nebert and Russell, 2002). In contrast to bacterial P450s, these enzymes are membrane-bound and located in the endoplasmic reticulum or the inner mitochondrial membrane. Since the biosynthesis of sterols, an important component of eukaryotic membranes, requires P450-catalyzed oxidation reactions, P450s are essential for eukaryotic life (Werck-Reichhart and Feyereisen, 2000). In most species, some of these P450s are important in the metabolism of a large number of xenobiotic substrates such as drugs, carcinogens, food additives, pollutants, pesticides, or environmental chemicals in addition to the metabolism of endogenous compounds (Waxman and Azaroff, 1992). Of the 57 human P450s, approximately 15 are involved in xenobiotic metabolism. The biotransformation of xenobiotics in most cases leads to pharmacologically inactive metabolites that are subsequently excreted. However, biotransformation may also activate so-called prodrugs to pharmacologically active products or even to toxic metabolites. Similarly, nontoxic procarcinogens can be activated by P450-catalyzed reactions and thus be turned into potent carcinogens (Nebert and Gonzalez, 1987). Since P450s play key roles in biosynthetic and catabolic pathways of a variety of compounds, their expression must be highly regulated. Some P450s are expressed only in some tissues and specific cells within this tissue. Similarly, the expression pattern of a number of P450s is different in developmental stages and in females and males.
|
II. Drug-Mediated Induction of Cytochromes P450 |
|---|
|
TopPreviousNext |
|---|
, 1972; Conney et al., 1960; Remmer and Merker, 1963). This increase in drug metabolism was subsequently attributed to PB-induced transcriptional activity of P450 genes (Adesnik et al., 1981; Gonzalez and Kasper, 1982). Later, it was found that PB, other barbiturates, and numerous other compounds that exhibit a similar induction pattern and therefore are called PB-type inducers, activate transcription of CYP2A, CYP2B, CYP2C, and CYP3A genes, the same P450s activated by the dexamethasone/rifampicin-type compounds (Waxman and Azaroff, 1992). As indicated in Fig. 1B, the effect of inducer drugs is not restricted to the regulation of P450s and other drug-metabolizing enzymes or drug transporters but involves a major pleiotropic response including the up- or down-regulation of numerous genes and physiological systems including proliferation of the smooth endoplasmic reticulum (Okey, 1990; Waxman and Azaroff, 1992).
|
In addition to PB-type and dexamethasone/rifampicin-type inducers, other prototypical classes of compounds are represented by aromatic hydrocarbons that mainly induce CYP1As and CYP1Bs, peroxisome proliferators elevating CYP4A levels, and ethanol that increases CYP2E1 (Table 1). The dexamethasone/rifampicin class of inducers affects the same P450s as the PB-type compounds but with different relative potencies. CYP3As are more efficiently induced than CYP2Cs and CYP2Bs by the dexamethasone/rifampicin-type compounds (Waxman and Azaroff, 1992; Denison and Whitlock, 1995; Meyer, 1996; Dogra et al., 1998).
|
In this review, we focus on the PB- and dexamethasone/rifampicin-type induction of CYP2Bs, CYP2Cs, and CYP3As, the major drug-metabolizing P450s (Meyer, 1996). The mechanisms underlying induction by the other inducer classes are briefly discussed, but interested readers are referred to the respective reviews. The elucidation of the mechanism of induction of CYP1As by polycyclic aromatic hydrocarbons has progressed more rapidly than the PB- and dexamethasone/rifampicin-type induction mechanism. With the help of genetic polymorphisms, high-affinity ligands, and inducible cell culture systems, the aromatic hydrocarbon receptor (AhR) and its binding partner, AhR nuclear translocator, could be identified. These findings, in addition to the discovery of AhR-response elements in the flanking regions of CYP1As were the basis for further characterization of this mechanism (Hankinson, 1995; Whitlock, 1999; Ma, 2001). Soon after the peroxisome proliferator-activated receptor (PPAR) was discovered in 1990, it also became clear that this orphan nuclear receptor plays a crucial role in induction of CYP4As by peroxisome proliferators and related compounds (Johnson et al., 1996; Simpson, 1997). In contrast, ethanol affects CYP2E1 at the post-translational level by stabilization of the enzyme not involving a receptor-dependent mechanism (Gonzalez et al., 1991; Lieber, 1997).
A. Induction of CYP2Bs, CYP2Cs, and CYP3As by Drugs and Xenobiotics
The major mystery in the induction of P450s by drugs and other chemicals for many years was how the cell recognizes these inducers and how the information is conveyed to the transcriptional machinery. Although CYP2B, CYP2C, and CYP3A induction by PB has been described decades ago, progress in this field has been hampered by four major peculiarities. First, the classes of PB- and dexamethasone/rifampicin-type inducer compounds constitute a variety of different substrates such as drugs, steroids, pesticides, pollutants, food additives, and many other chemicals that show no obvious quantitative structure/activity relationship, except that they are lipid-soluble molecules with a relatively low molecular weight. Moreover, most of these xenobiotics activate their target enzymes only at relatively high concentrations in the micro- to millimolar range. Any putative receptor would have to be able to accommodate all these different structures and would require considerable plasticity in its recognition site similar to the substrate binding sites of P450s (Okey, 1990; Waxman and Azaroff, 1992). Indeed, because the P450 substrate binding site exhibits a similar promiscuity toward different substrates, direct interaction of inducers with P450s and thereby release of an endogenous inducer or formation of chemically reactive, reduced oxygen species by uncoupling of the hydroxylation reactions were postulated as alternatives to the PB receptor theory (Fonné-Pfister and Meyer, 1987). Second, as an additional experimental drawback, PB-type induction of P450s is not commonly observed in primary hepatic cell culture systems where it is either qualitatively disturbed or completely absent. This might be due to the dedifferentiation process that occurs when generating continuously dividing cell systems, since drug induction and metabolism is a hallmark of highly differentiated, nondividing hepatocytes (Meyer and Hoffmann, 1999). When culture conditions were modified and included matrix components and other factors, primary rat hepatocytes that retained PB inducibility could be cultured (Waxman et al., 1990). This methodological breakthrough led to the discovery of DNA-enhancer elements that mediate induction (Trottier et al., 1995). However, despite these promising advances, the identified enhancer elements in various species apparently exhibited no obvious common features (Dogra et al., 1998). Third, in contrast to CYP1A induction, no animal models with genetic defects of induction were available that allowed mapping of important components of the PB induction machinery. Finally, the induction potency of several compounds is drastically different in different species, suggesting that multiple mechanisms or receptors may operate to produce this response (Denison and Whitlock, 1995). For example, the antibiotic rifampicin is one of the strongest inducers of human CYP3A4 but has very little effect on rodent CYP3As, whereas the antiglucocorticoid 5-pregnene-3
-ol-20-one-16
-carbonitrile (PCN) is a potent activator of rodent, but not human, CYP3As (Savas et al., 1999; Xie and Evans, 2001). Together, these features have delayed the elucidation of the mechanisms by which PB- and dexamethasone/rifampicin-type inducers change gene expression.
|
III. Drug-Response Elements in Inducible Cytochrome P450 Genes |
|---|
|
TopPreviousNext |
|---|
), whereas hepatocytes from mammals rapidly lose this ability if not cultured under special conditions (Waxman et al., 1990). Thus, the first breakthrough in isolating PB-responsive DNA elements was made by identifying a drug-responsive 4.8-kb enhancer in the flanking region of chicken CYP2H1 (Hahn et al., 1991). The first mammalian PB-responsive enhancer element was isolated from the rat CYP2B2 5' flanking region in 1995 (Trottier et al., 1995). Later, similar drug-responsive enhancer elements in other mammalian CYP2Bs, CYP3As, and CYP2Cs could be identified (Honkakoski and Negishi, 1998b; Sueyoshi and Negishi, 2001). Interestingly, the basic mechanism of PB induction in higher animals seems to be conserved, whereas bacteria apparently use different mechanisms to react to barbiturate exposure. A. CYP102/106 in Bacillus megaterium
Bacterial and eukaryotic P450s differ in several ways: whereas bacterial P450s are soluble, eukaryotic P450s are membrane bound. In bacteria, NADH is the predominant cofactor in contrast to NADPH in eukaryotes, and although some bacterial P450s are one-component systems, eukaryotic P450s depend on a reductase (Fulco, 1991). Despite these differences, induction of P450s by PB and other barbiturates is also observed in certain bacteria. In B. megaterium, PB induction of CYP102 and CYP106 was postulated to be mediated by PB removal of a repressor protein from a "barbie-box", a 17-bp DNA element with a conserved AAAG motif. The expression of the protein Bm1P1 is stimulated by the inducer and then perturbs the binding of the repressor protein Bm3R1 to the barbie-box (He and Fulco, 1991; Shaw and Fulco, 1993; Liang et al., 1995). However, this concept of PB-mediated de-repression has recently been challenged by results that show that neither mutations of the gene encoding for Bm1P1 nor mutations of the barbie-box affect PB induction of CYP106 (Shaw et al., 1998). On the contrary, Bm1P1 might even help to repress the CYP101 gene (Shaw et al., 2000). Thus, the molecular basis of PB induction in B. megaterium and the role of the barbie-box in this process remain controversial. Intriguingly, conserved barbie-boxes are also found in the proximal flanking regions of chicken and mammalian P450s. As discussed below, the discovery of nuclear receptors as mediators of drug induction in higher animals implies that bacteria use a different strategy to mediate PB gene expression, since nuclear receptor genes have exclusively been observed in metazoan genomes (Mangelsdorf et al., 1995).
In the fruit fly Drosophila melanogaster and the house fly Musca domesticus, P450s of the subfamilies CYP6A and CYP6D have been isolated and shown to be responsive to PB (Feyereisen, 1999). Analysis of the flanking region of the D. melanogaster CYP6A2 gene revealed PB induction to be mediated by sequences within the first 428 bp upstream of the transcriptional start site (Dunkov et al., 1997). No detailed analysis of drug response elements has been reported. In contrast, comparison of 13 members of the subfamily CYP6B from the closely related tiger swallowtail Papilio glaucus and Papilio canadensis, which are inducible by a number of compounds, revealed differences in the 5' flanking regions distal of -640 bp from the transcriptional start site. A response element to the xenobiotic xanthotoxin and to ecdysone as well as putative drug-responsive elements known to regulate vertebrate-inducible P450s are present in this region, including a binding site for AhR and an imperfect pregnane X receptor (PXR)-responsive element, which might suggest a conservation of drug-responsive elements in insects compared with those found in vertebrates (Li et al., 2002). Similar elements were recently described in the 5' flanking region of CYP6B1 of Papilio polyxenes (Petersen et al., 2003). However, transcription factors that may bind to these elements and are responsible for insect xenobiotic induction have not been reported. The sequence of the D. melanogaster genome has revealed a much lower number of predicted nuclear receptors compared with the human and the Caenorhabditis elegans genomes (Enmark and Gustafsson, 2001). Recently, it has been reported that aberrant transcription of the CYP6G1 gene in D. melanogaster confers resistance to DDT, and the respective mutation in the CYP6G1 locus is found worldwide (Daborn et al., 2002). Furthermore, genomic comparison of the three major enzyme families responsible for insecticide resistance, the carboxylesterases, glutathione transferases, and the P450s between D. melanogaster and Anopheles gambiae revealed an expansion of these enzyme families in the mosquito genome compared with the fruit fly (Ranson et al., 2002). Understanding the signaling mechanism responsible for insecticide-mediated induction of P450s and other genes could help to develop countermeasures for insecticide resistance.
C. CYP2H1/2, CYP3A37, and CYP2C45 in Chicken
In 1991, Hahn, Hansen, and May described the first drug-responsive enhancer sequence, a 4.8-kb fragment of DNA (-5.9 to -1.1 kb) in the flanking region of chicken CYP2H1 (Hahn et al., 1991). Following this report, it took several years to identify the functional elements within this large fragment (Fig. 2). The first 1.1 kb of DNA proximal to the CYP2H1 transcriptional start site were not contributing to PB induction unlike the elements found in bacteria (Dogra and May, 1997). In fact, the presence of this 1.1-kb fragment together with the 4.8-kb enhancer largely decreased the drug response in reporter gene assays. Other experiments in chicken primary hepatocytes using the protein synthesis inhibitor cycloheximide and puromycin suggested that the mechanism of PB induction in chicken and mammals may differ. Inhibition of protein synthesis caused a superinduction of CYP2H1 in chicken primary hepatocytes exposed to phenobarbital, but this superinduction did not occur in mammalian hepatocytes (Dogra et al., 1993; Denison and Whitlock, 1995; Sidhu and Omiecinski, 1998). Moreover, this argument was initially supported when 240-bp PB-responsive enhancer sequence in the 4.8-kb enhancer (-1640 to -1400 bp) did not reveal the typical hexamer repeats of mammalian PB-responsive elements (Dogra et al., 1999). This element was predominantly active in combination with additional DNA fragments resulting in a size of the responsive domain of 556 bp. In contrast, our own studies identified a 264-bp PB-responsive enhancer unit (PBRU) at -1657 to -1393 bp that overlaps with the 240-bp element of Dogra and coworkers, as well as an additional 240-bp PBRU (-5120 to -4881 bp) further upstream in the flanking region of CYP2H1, both harboring direct repeats of hexamer half-sites with a spacing of four nucleotides (DR-4). Both of these elements mediated PB induction in reporter gene assays in the chicken hepatoma cell line LMH, similar to those elements found in mammalian CYP2B PBRUs (Handschin and Meyer, 2000; Handschin et al., 2001a). A third PB-inducible fragment is present within the first 6 kb of flanking region between -5896 and -4528 bp (Dogra et al., 1999; Handschin and Meyer, 2000). First analysis of this region, however, failed to reveal conserved PB-responsive DNA elements, and this third drug-responsive enhancer awaits further examination (Handschin et al., 2001a). The mRNA levels of PB-induced CYP2H1 are about 10 times higher than those of the closely related CYP2H2. This is due to differences in the sequence of a hepatic nuclear factor 3 site in the CYP2H2 promoter that leads to lower expression of CYP2H2 compared with CYP2H1, whereas the enhancer regions are identical between these two genes (Davidson et al., 2001). Thus, DR-4 hexamer repeats are the common theme in PB-inducible enhancers in CYP2H1/2 as well as in a chicken PB-inducible member of the CYP3A family, CYP3A37, where a 159-bp PBRU has been located at -1159 to -1037 bp (Podvinec et al., 2002). Similar to the CYP3A37 enhancer, we have observed that mutagenesis of a DR-4 site in a 239-bp PBRU (-2435 to -2197 bp) of the chicken CYP2C45 abolishes induction by PB (Baader et al., 2002). In summary, all the currently known chicken drug-inducible P450s share a conserved arrangement of DNA elements that mediate induction by PB and other xenobiotics. Moreover, these elements show a striking conservation when compared with drug-responsive enhancers in mammals as discussed below.
|
|
A seminal breakthrough in identifying mammalian CYP2B PBRUs was reported in 1995 by Trottier and coworkers who isolated a 163-bp PB-responsive enhancer fragment in the rat CYP2B2 5' flanking region situated at -2318 to -2155 bp upstream of the transcription start site (Trottier et al., 1995). Drug induction of this PBRU in vivo was confirmed by in situ DNA injections in rat liver (Park et al., 1996). Soon thereafter, a DNA fragment located at -1404 to -971 sharing high similarity to the rat CYP2B2 PBRU was reported to regulate drug induction of mouse Cyp2b10 (Honkakoski et al., 1996). In both sequences, candidate transcription factor binding sites were predicted, most strikingly repeats of hexamer half-sites that resembled known nuclear receptor binding sites (Honkakoski and Negishi, 1997; Stoltz et al., 1998). The mouse Cyp2b10 enhancer could subsequently be reduced to a 51-bp PB-responsive enhancer module (PBREM) located at -2339 to -2289 bp, which responded to a variety of xenobiotics in reporter gene assays in mouse primary hepatocyte cultures (Honkakoski et al., 1998a). A characteristic of the mouse Cyp2b10 PBREM, the subsequently identified human CYP2B6 PBREM (Sueyoshi et al., 1999), and the rat CYP2B2 PBRU is a conserved arrangement of two DR-4 elements separated by a putative nuclear factor-1 (NF-1) binding site (Fig. 2). Site-specific mutations of the hexamers within the DR-4 sites dramatically decrease PB induction of these elements (Honkakoski et al., 1998b; Liu et al., 1998; Ramsden et al., 1999; Stoltz and Anderson, 1999; Paquet et al., 2000; Liu et al., 2001). Wang and coworkers (2003a) further analyzed the CYP2B6 5' flanking region and were able to isolate an additional PB-responsive element located 8.5-kb upstream of the transcriptional start site that contains a DR-4 element. The functional role of the NF-1 site in the PBREM is much less clear compared with the DR-4 elements. In transgenic mice that contain 2.5 kb of CYP2B2 flanking region, specific mutations of the NF-1 site abolished binding of NF-1 but retained full inducibility by PB, thus suggesting no functional role of NF-1 in drug induction of CYP2B2 (Ramsden et al., 1999). However, experiments using in vivo footprinting techniques revealed that the NF-1 binding site is protected under normal conditions and that this protected region is enlarged after PB treatment (Kim and Kemper, 1997; Kim et al., 2000). Moreover, NF-1 binding increased drug induction in reporter gene assay using Drosophila embryo extract to assemble chromatin (Kim et al., 2001). These and another report (Stoltz and Anderson, 1999) suggest that NF-1 contributes to drug induction mediated by these PBRUs. Recently, this configuration of two functional DR-4 elements separated by a NF-1 site has also been found in the chicken CYP2H1 264-bp PBRU (Podvinec et al., 2002). Finally, in vivo injection experiments have shown that additional sequences flanking the two DR-4 elements and the NF-1 sites are also contributing to drug responsiveness, namely an uncharacterized site at the 3' flank and an additional nuclear receptor binding element at the 5' flank (Rivera-Rivera et al., 2003).
Because of the presence of a barbie-box similar to that found in bacteria, regions proximal to mammalian CYP2B promoters were also analyzed for their ability to confer PB induction (Kemper, 1998). After PB treatment, increased binding of phosphorylated proteins to a positive element (-98 to -69 bp) in the CYP2B1/2 flanking region was observed in rat livers in vivo (Prabhu et al., 1995; Nirodi et al., 1996; Sultana et al., 1997). These proteins have eluded identification so far (Samudre et al., 2002). In contrast, in transgenic mice expressing either 800 bp or 19 kb of CYP2B2 flanking region, only the strain with the 19 kb showed responsiveness to PB (Ramsden et al., 1993, 1999). In other experiments, no specific protein binding to the barbie-box in the proximal promoter region was observed, and targeted disruption of the barbie-box did not affect PB inducibility of CYP2B genes (Kemper, 1998; Sueyoshi and Negishi, 2001). These findings from various laboratories provide compelling evidence that the distal enhancer elements harboring the DR-4 sites are the predominant regulatory DNA elements in drug induction of these P450s.
Mammalian CYP3A genes were initially analyzed to map regions responsive to both classical glucocorticoids and antiglucocorticoids (Quattrochi and Guzelian, 2001). The identified regions proved to be more heterogeneous compared with the highly conserved CYP2B PBRUs. In the proximal promoter between -170 and -140 bp, DR-3 elements in the rat CYP3A2, everted repeats with a spacing of six nucleotides (ER-6) in the human CYP3A4 and CYP3A7 and a DR-4 element in the rat CYP3A23 were identified, as shown in Fig. 2 (Miyata et al., 1995; Quattrochi et al., 1995; Barwick et al., 1996; Huss et al., 1996; Huss and Kasper, 1998; Pascussi et al., 1999; Bertilsson et al., 2001). Furthermore, when testing 13 kb of the CYP3A4 5' flanking region, an important 230-bp xenobiotic-responsive enhancer module (XREM) was discovered at -7836 to -7606 bp that apparently accounts for a major proportion of the drug induction response and harbors DR-3 and ER-6 sites that respond to both dexamethasone/rifampicin- and PB-type inducers (Goodwin et al., 1999, 2002a). This XREM and the upstream enhancer module found in the CYP2B6 flanking region are both essential for maximal induction of CYP3A4 and CYP2B6, respectively (Goodwin et al., 1999; Wang et al., 2003a). However, the exact contribution of the hexamer repeats near the promoter, the XREM, and an additional DR-3 at -7287 to -7273 bp is not known. Since mutations of each of these sites decrease reporter gene activity in the range of 20 to 50%, none of these sites seems to be responsible for mediating induction of CYP3As, and all of these elements apparently contribute to drug induction (Quattrochi and Guzelian, 2001; Sueyoshi and Negishi, 2001).
Of the mammalian drug-inducible CYP2Cs, PBRUs have been reported in human CYP2C9, CYP2C8, and CYP2C19 (Fig. 2). In the CYP2C9 flanking region, a PBRU located at -1856 to -1783 bp that contains a DR-4 site confers induction by PB and rifampicin (Gerbal-Chaloin et al., 2002) similar to the DR-4 in the chicken CYP2C45 (Baader et al., 2002). Recently, a more distal enhancer between -2900 and -2841 bp in the flanking region of CYP2C9 has been characterized, and two DR-5 sites were identified (Ferguson et al., 2002c). In the human CYP2C8 5' flanking region, two DR-4 sites have been identified in a 400-bp fragment that are responsive to preferentially dexamethasone/rifampicin-type inducer compounds (Ferguson et al., 2002b). The human CYP2C19 flanking region is very similar to that of CYP2C9. Thus, analysis of the two homologous drug-enhancer regions revealed that the more proximal element at -1874 bp is mainly responsible for drug induction and differs from the CYP2C9 element only by one nucleotide (Ferguson et al., 2002a).
E. Other Mammalian Drug-Inducible Cytochromes P450
Induction by PB- and dexamethasone/rifampicin-type compounds has been observed for a range of P450s other than those discussed above. Most strikingly, PB activates members of the CYP1A and CYP2A subfamily in mammals (Dogra et al., 1998; Kemper, 1998). Although no DNA response elements have been identified so far, PB induction of at least CYP1A2 seems to be independent of the presence or absence of the AhR (Zaher et al., 1998; Sakuma et al., 1999).
|
IV. Nuclear Receptors Involved in Drug Induction of Cytochromes P450 |
|---|
|
TopPreviousNext |
|---|
; Enmark and Gustafsson, 1996; Nuclear Receptors Nomenclature Committee, 1999). Nuclear receptors were prime candidates for mediating hepatic drug induction for several reasons (Waxman and Azaroff, 1992). First, their ligands are normally small and lipophilic, properties strikingly similar to those of xenobiotic and endobiotic inducer compounds such as steroids, bile acids, or fatty acids. Second, nuclear receptors bind to DNA elements consisting of repeats of hexamers in different kinds of arrangements such as those found in drug-responsive enhancers of P450s (Fig. 3B). Third, the tissue-specific expression of a subset of nuclear receptors is identical to the tissue specificity of drug induction. Finally, closely related members of the nuclear receptor subfamilies NR1I and NR1H (Fig. 3C) play key roles in many physiological processes where P450s are involved. These include steroid, vitamin D, cholesterol, lipid, or bile acid biosynthesis and metabolism (Waxman and Azaroff, 1992; Beato et al., 1995; Mangelsdorf and Evans, 1995; Mangelsdorf et al., 1995; Enmark and Gustafsson, 1996; Waxman, 1999; Honkakoski and Negishi, 2000).
A. Constitutive Androstane Receptor
Within the CYP2B PBRU structure, the two DR-4 sites, called NR1 and NR2, are not equivalent in terms of activation potency by drugs. The more distal DR-4 site (NR1) is more conserved among man, rodents, and chicken (Paquet et al., 2000; Zelko and Negishi, 2000). Thus, the NR1 site was used in affinity purifications for isolation of proteins binding to this sequence and mediating drug induction. This approach led to the identification of the murine nuclear receptor CAR to bind as a heterodimer with the retinoid X receptor (RXR) to the mouse Cyp2b10 NR1 but not to the minimally different, noninducing corresponding fragment from Cyp2b9 (Honkakoski et al., 1998b). Apart from mouse, CAR orthologs have also been described in man, monkey, and rat (Baes et al., 1994; Choi et al., 1997; Yoshinari et al., 2001). Moreover, binding of CAR to the NR1 site predominantly occurred in liver extracts of PB-treated mice and to a much lesser degree in untreated control animals. Subsequently, CYP2Bs in rat have also been shown to be regulated by the rat CAR ortholog but, in addition, require binding of the transcription factor Sp1 to the CYP2B1 proximal promoter (Muangmoonchai et al., 2001; Xiong et al., 2002). In transient transfection assays and in stably transfected HepG2 cells, CAR triggered high basal activity of reporter genes regulated by the mouse Cyp2b10 and the human CYP2B6 PBREMs (Sueyoshi et al., 1999), as expected by the initial reports describing CAR as a constitutively active receptor (Baes et al., 1994; Choi et al., 1997; Yoshinari et al., 2001). Thus, CAR activity after drug induction has to be regulated by additional mechanisms than just ligand binding. Different mechanisms of how CAR can be activated by drugs have been proposed so far, none of them explaining the whole process of signal transduction (Fig. 4). First, although CAR normally resides in the cytoplasm of untreated mouse liver and hepatocytes, it undergoes a cytosolic-nuclear translocation upon PB stimulation, at least in mouse liver and primary rat hepatocytes (Kawamoto et al., 1999; Maglich et al., 2003). This process is controlled by protein phosphorylation events and can be inhibited by using the protein phosphatase inhibitor okadaic acid. Furthermore, the translocation event appears to be mediated by a leucine-rich xenochemical response signal in the C-terminal part of CAR (Zelko et al., 2001). The composition of the protein complex in which CAR is retained in the cytoplasm has not been elucidated. Recent reports described that the nuclear receptor coactivator glucocorticoid receptor-interacting protein 1 (GRIP1) enhances CAR activity and increases cytoplasmic nuclear translocation of CAR in untreated mice (Min et al., 2002a). A second level of CAR activation has been observed in stably transfected HepG2 cells where CAR was located in the nucleus but could be inhibited by administration of certain androstanols (Sueyoshi et al., 1999). These androstanols have been found to work as inverse agonists of CAR activity, the inhibition being reversed by treatment with inducer compounds (Forman et al., 1998; Tzameli and Moore, 2001). However, it is unknown whether this reversal of inhibition is due to a direct interaction of inducers with CAR. In addition to derepression, direct activation of CAR by a few chemicals has been reported. The chemical 1,4-bis[2-(3,5-dichloropyridyloxy)]benzene (TCPOBOP) is one of the strongest inducers in mouse but hardly affects CYP2B levels in man. Accordingly, TCPOBOP strongly binds to and activates mouse CAR but not human CAR (Moore et al., 2000b; Tzameli et al., 2000). Differences in activation of CAR in mouse and man are most likely due to the divergent ligand binding domain of the CAR orthologs from these species (Moore et al., 2000b). Furthermore, CAR activity in the nucleus also seems to be under the regulation of protein phosphorylation events. Experiments using calcium/calmodulin kinase inhibitors revealed changes in CAR-mediated drug induction even in the case where CAR was located in the cell nucleus (Zelko and Negishi, 2000). Finally, an additional regulation of CAR mRNA and activity was reported to occur via the glucocorticoid receptor (GR), which induces CAR transcription via a distal GR-responsive element at -4.4 kb in the human CAR 5' flanking region (Pascussi et al., 2000b, 2003a).
|
CAR knockout mice reveal virtually absent induction of Cyp2b10 by TCPOBOP and PB in the liver and small intestine (Wei et al., 2000, 2002; Maglich et al., 2002; Ueda et al., 2002a). Moreover, TCPOBOP and PB induction of Cyp1a1, Cyp2a4, Cyp3a11, and a range of phase II enzymes and transporters is impaired in the livers of CAR knockout mice (Maglich et al., 2002; Ueda et al., 2002a; Wei et al., 2002). Comparison of wild-type and CAR-null mice also revealed complete absence of liver hypertrophy and hyperplasy as well as altered metabolism of different compounds resulting in altered sensitivity to toxins (Wei et al., 2000). Thus, the acetaminophen-metabolizing enzymes Cyp1a2, Cyp3a11, and glutathione S-transferase are activated in a CAR-dependent manner after treatment with acetaminophen in wild-type, but not in CAR knockout, mice (Zhang et al., 2002). This finding could be recapitulated in "humanized" mice where the endogenous CAR was ablated, and human CAR under the control of the albumin promoter was expressed in the liver (Zhang et al., 2002). The results obtained with the animal models clearly indicate a crucial role of CAR in mediation of drug induction of certain inducer compounds. However, the molecular mechanism of CAR-mediated signal transduction and the relative contribution of CAR to the total drug effect on gene expression remain enigmatic (Fig. 4). The recently described ligand and activator with high affinity for human CAR provides an opportunity to learn more about CAR signal transduction in human liver (Maglich et al., 2003).
PXR, alternatively called steroid and xenobiotic receptor or pregnane-activated receptor, has been independently discovered in mice and humans by three groups in 1998. These investigators used either homology cloning or database mining techniques (Bertilsson et al., 1998; Blumberg et al., 1998b; Kliewer et al., 1998; Lehmann et al., 1998). Later, PXR orthologs in rat, rabbit, dog, pig, and monkey have been cloned (Zhang et al., 1999; Jones et al., 2000; Savas et al., 2000; Moore et al., 2002). PXR has subsequently been shown to bind to the DR-3 and ER-6 elements found in CYP3A drug-responsive enhancers and to be activated by a variety of steroids, drugs, and other xenobiotics. Like CAR, PXR transcription is stimulated by activators of GR, and in addition, PXR expression is inhibited by interleukin-6 during acute-phase response, which might explain the observed down-regulation of drug-induced P450s in infections (Pascussi et al., 2000a, 2001, 2000c; Beigneux et al., 2002; Jover et al., 2002). In contrast to other members of the nuclear receptor superfamily, amino acid sequence comparison of the ligand binding domains of different PXR orthologs revealed an unusual high divergence (Jones et al., 2000). This divergence explains the species differences observed in P450 induction by different drugs as demonstrated by site-directed mutagenesis of the mouse PXR ligand binding domain. Four amino acids of the mouse sequence were changed into their corresponding human counterparts, which led to a typical "human" activation pattern (Watkins et al., 2001). Similarly, PXR knockout mice that express the human PXR as transgene exhibit a human-typical response to different inducer compounds (Xie et al., 2000a). As depicted in Fig. 4, in contrast to CAR, PXR is found exclusively in the nucleus (Sueyoshi and Negishi, 2001), and a direct correlation between ligand binding and receptor activation has been demonstrated (Jones et al., 2000). Interestingly, one of the most potent inducers of human PXR discovered so far is hyperforin, a component of extracts from the herb St. John's wort (Moore et al., 2000a; Wentworth et al., 2000). St. John's wort is only one example of the many herbal remedies which are widely used with the potential to interact with drugs and lead to unwanted herb-drug interactions (Zhou et al., 2003). It is thus of considerable importance to elucidate the molecular mechanisms underlying these interactions to prevent adverse effects of herbal remedies (Raucy, 2003).
In PXR knockout animals, induction of Cyp3a11 by PCN is impaired, and basal levels of this gene are increased (Xie et al., 2000a; Staudinger et al., 2001b). However, Cyp3a11 can still be activated by PB. Similarly, PCN induction of Cyp2b10 is abolished in liver and intestine. In contrast, PCN inhibition of Cyp7a1 is abolished in PXR-/- animals. Cyp7a1 is the first enzyme of cholesterol metabolism to bile acids in the liver. Similarly, the expression of Cyp1a1 in the intestine is also derepressed in PXR-null mice compared with PCN-treated wild-type animals (Maglich et al., 2002). Although neither the CAR- nor the PXR knockout animals show an overt phenotype under standard laboratory conditions, constitutive activation of PXR in a transgenic mouse line expressing PXR fused to a VP16-activator domain led to a severe phenotype characterized by growth retardation, hepatomegaly, and liver toxicity (Xie et al., 2000a). Obviously, PXR plays a key role in drug induction, and because of its direct activation by ligands, PXR constitutes an attractive drug target. Activators of PXR include calcium channel blockers, statins, antidiabetic drugs, human immunodeficiency virus protease inhibitors, and anticancer drugs among many other drugs (Kliewer et al., 1998, 2002; Jones et al., 2000; Drocourt et al., 2001; Dussault et al., 2001; Synold et al., 2001; Goodwin et al., 2002b; Kliewer and Willson, 2002; Liddle and Goodwin, 2002). Many of these drugs are clinically relevant inducers at therapeutic doses in humans (Fig. 1C).
C. The Evolution of Xenosensors: Lessons Learned from the Chicken Xenobiotic Receptor
The similarity between chicken and mammalian PBRUs led us to attempt to clone the avian orthologs of the mammalian xenosensors PXR and CAR. Surprisingly, only one nuclear receptor responsive to drugs, the chicken xenobiotic receptor (CXR), was identified. No additional avian receptors related to this receptor family were observed (Handschin et al., 2000). When comparing the amino acid sequences of CXR, PXRs, and CARs, we found that CXR is about equally related to the mammalian PXRs as it is to the mammalian CARs as depicted in Fig. 5 (Handschin et al., 2000). In regard to their function as xenosensors, the mammalian PXRs and CARs and the chicken CXR are interchangeable as shown by activation of mouse, rat, and human PBRUs in the drug-inducible chicken hepatoma cell line LMH and by the binding of PXR and CAR to the chicken CYP2H1 PBRU (Handschin et al., 2001b). Thus, despite the apparent difference in the number of xenosensors, the basic molecular mechanism of drug induction is conserved from birds to mammals. In a recent report, Dogra and coworkers (2003) described that the coactivator CBP/p300 increases the activity of CXR and stimulates PB-induced but not basal expression of CYP2H1. In their model, coactivator proteins such as CBP/p300 and p/CAF link factors binding to distal enhancer sites such as CXR with the proximal promoter upon drug stimulation and then promote chromatin acetylation and the subsequent increase in transcription of CXR target genes similar to proposed models in mammals.
|
When testing different drugs, steroids, xenobiotics, bile acids, and benzoates, CXR turned out to be one of the most promiscuous receptors, with a broad spectrum of drugs that activate or inhibit compared with the mammalian xenosensors (Moore et al., 2002). Interestingly, only one nuclear receptor related to PXR and CAR has been found in zebrafish (Moore et al., 2002), and when searching the recently published Fugu rubripes genome for PXR and CAR orthologs (M. Podvinec, unpublished observations). These receptors also are equally related to the mammalian PXRs and mammalian CARs (Fig. 5). Interestingly, even in the C. elegans genome, a single nuclear receptor related to CXR, PXR, and CAR called nhr-8 was found to be activated by different toxins and contributes to xenobiotic resistance (Lindblom et al., 2001). In mammals, cloning attempts on the basis of the mouse and human sequences were successful for the isolation of pig, dog, rabbit, and rat PXR; all have a very high similarity with the sequence of the mouse and human orthologs (Zhang et al., 1999; Jones et al., 2000; Savas et al., 2000; Moore et al., 2002). Thus, the single xenosensors found in nonmammalian species likely represent the ancestral genes that in mammals diverged into two receptors, PXR and CAR. The reason for this duplication of xenobiotic-sensing nuclear receptors in mammals is not clear, but it may reflect the specific challenges in diet and environment that the different species encountered. Moreover, the xenosensors found in nonmammalian species resemble more PXR-type receptors in terms of direct ligand activation. This raises the questions of why and how the unusual nuclear receptor CAR has evolved. Further comparative genomics of additional xenosensors from other species including D. melanogaster should shed more light on this issue. The exact roles of the related benzoate X receptors and identified in xenopus are not known (Blumberg et al., 1998a; Nishikawa et al., 2000). However, benzoate X receptors and are clearly pharmacologically distinct from the described xenosensors, and in addition, their expression pattern exhibits no similarities to those found for PXRs, CARs, and CXRs in mammals and chicken (Heath et al., 2000; Grün et al., 2002; Moore et al., 2002). Thus, xenobiotic-sensing nuclear receptors in amphibians remain to be cloned and characterized.
D. Structure of the Xenosensors
Several puzzles concerning drug induction were clarified by solving the crystal structures of the nuclear receptors involved in this process. The extreme structural variety of inducer compounds hardly fits with the hypothesis of a common receptor. However, when PXR was crystallized and the structure analyzed, it became clear that PXR not only has a much larger ligand binding domain compared with other nuclear receptors, it also was possible for the cocrystallized ligand SR12813, a synthetic biphosphonate, to bind to PXR in three different conformations (Watkins et al., 2001). Hyperforin, one of the psychoactive components of St. John's wort and a potent activator of PXR, induces a structural change in the PXR conformation and considerably increases the size of the ligand binding pocket (Watkins et al., 2003). Whether the possibility for PXR ligands to bind in different conformations also has an impact on their activation potential remains to be investigated (Ekins and Schuetz, 2002). Analysis of the 28 amino acids shaping the ligand binding pocket can, in principle, explain the species differences in drug induction (Watkins et al., 2001). In comparison with other known nuclear receptor structures, PXR shares the same general confirmation. However, the size of the ligand binding cavity is much larger and mostly coated with hydrophobic residues that can accommodate lipophilic inducer compounds. In addition to the 12 helices found in classical nuclear receptor ligand binding domains, PXR has a large, flexible loop that apparently provides additional flexibility when binding bulky ligands and further explains the promiscuity of this receptor (Gillam, 2001). Interestingly, mutation of a single histidine at position 407 in human PXR into an alanine resulted in high constitutive activity and dramatically increased basal expression of PXR-activated reporter gene assays (Ostberg et al., 2002). The insights about the structure of the PXR ligand binding domain could now help to predict PXR activators and ligands in drug discovery and development (Ekins and Erickson, 2002; Ekins et al., 2002).
For CAR, no crystal structure has been reported yet. Molecular modeling of the CAR ligand binding domain on the basis of other nuclear receptor structures combined with site-directed mutagenesis provided some insights into the function of CAR (Dussault et al., 2002; Xiao et al., 2002; Andersin et al., 2003; Jacobs et al., 2003; Moore et al., 2003). The foremost questions regarding CAR are whether CAR has a ligand binding domain similar in size compared with PXR and whether its structure reflects the constitutive activity. A three-dimensional model based on the related PXR crystal structure predicts that CAR lacks the flexible surface loop found in PXR and thus would be less promiscuous for direct ligand binding (Xiao et al., 2002). However, the volume of the ligand binding pocket of these two receptors seems to be similar, allowing CAR to putatively accommodate compounds of different structures as observed for PXR (Dussault et al., 2002; Xiao et al., 2002). Strikingly, several features found only in the CAR model may account for its constitutive activity. Between helix 11 and helix 12, site of the classical transactivation domain in nuclear receptors, CAR has a short loop and a C-terminal helix that fix the ligand binding domain in a conformation normally found in ligand-activated nuclear receptors even in absence of CAR ligands (Dussault et al., 2002). Moreover, charge-charge interactions between the C-terminal activation domain and helix 4 apparently favor ligand-independent activation, as verified by site-directed mutagenesis of key residues in this intramolecular interaction. In contrast to the charge clamp in classical endocrine nuclear receptors, three hydrophobic amino acids in the AF-2 domain were observed to be of more importance than the lysine in helix 3 and the glutamate in helix 12 for the interactions of CAR with coactivator proteins (Andersin et al., 2003). In summary, CAR uses some of the classical conserved motifs and coregulator proteins as described for other nuclear receptors, but its structure has differences which might account for its constitutive activity. Ligand-mediated repression of CAR may be caused by replacement of coactivator proteins by corepressors. These predicted structural features of CAR are strikingly different from classical nuclear receptors and open the discussion about the evolution of such a configuration. Hopefully, more decisive answers will be provided when the CAR crystal structure is solved. All of these interpretations have to be seen in regard to the fact that most inducers seem to activate CAR by an indirect mechanism leading to cytoplasmic-nuclear translocation not involving direct ligand activation. Predictions of the nature of compounds that trigger this translocation and activation therefore might not be achieved by knowing the structure of CAR and may require other experimental approaches.
E. Other Target Genes of Pregnane X Receptor and Constitutive Androstane Receptor
Although P450s have obviously been the primary focus in the characterization of xenosensor targets and are the primary focus of this review, numerous other genes have been reported to be regulated by these nuclear receptors. This makes sense, since inducer drugs have been known to increase the expression of not only phase I enzymes (functionalization reactions), but also phase II enzymes (conjugation reactions), drug-transporters, and related enzyme systems for endogenous substrates for these reactions (Table 2). Thus, a role of CAR and PXR has been proposed in the regulation of human bilirubin UDP-glucuronosyltransferase (Sugatani et al., 2001), dehydroepiandrosterone sulfotransferase, 3'-phosphoadenosine 5'-phosphosulfate synthetase 2 (an enzyme that is involved in the synthesis of the donor sulfate group) (Sonoda et al., 2002), hydroxysteroid sulfotransferase (Duanmu et al., 2002), and glutathione S-transferase (Falkner et al., 2001; Zhang et al., 2002). Transporters regulated by the xenosensors are mostly drug- or bile acid-transport proteins and include the multidrug-resistance proteins 2, 3, and 4 (Schuetz et al., 2001; Cherrington et al., 2002; Kast et al., 2002;; Xiong et al., 2002; Staudinger et al., 2003), the intestinal P-glycoprotein (Geick et al., 2001; Synold et al., 2001), and the organic anion transport protein 2 (Staudinger et al., 2001b; Guo et al., 2002). Other PXR, CAR, or CXR target genes were anticipated, such as the first and rate-limiting enzyme in heme biosynthesis, the 5-aminolevulinic acid synthase (Fraser et al., 2002). Other regulated genes were unexpected, for example the activation of expression of the human inducible nitric oxide synthase (Toell et al., 2002). To analyze the pleiotropic induction response (Fig. 1B), DNA-expression microarrays with cDNA derived from CAR- or PXR-deficient mice and from the humanized mice expressing human PXR recently have expanded the list of putative xenosenor target genes (Maglich et al., 2002; Ueda et al., 2002a; Rosenfeld et al., 2003), although the observed effects on mRNA expression may of course also represent secondary effects (Ueda et al., 2002a). Interestingly, expression of PXR and CAR themselves as well as of the AhR seems to be auto-regulated by these two xenosensors (Maglich et al., 2002). All of the genes analyzed so far are positively affected by the respective xenobiotic-sensing nuclear receptor, and a list of these genes can be found in Table 2. However, these drugs are also known to repress a number of genes (Frueh et al., 1997). Thus, the list of genes up- or down-regulated by CAR, PXR, and CXR is expected to grow in the future when additional genes are analyzed for their ability to be activated or repressed by drugs.
|
|
V. Endogenous Roles of the Xenosensors |
|---|
|
TopPreviousNext |
|---|
