Nuclear receptors CAR and PXR: Molecular, functional, and biomedical aspects

Abstract

Nuclear receptors (NRs) are ligand-activated transcription factors sharing a common evolutionary history and having similar sequence features at the protein level. Selective ligand(s) for some NRs is not known, therefore these NRs have been named “orphan receptors”. Whenever ligands have been recognized for any of the orphan receptor, it has been categorized and grouped as “adopted” orphan receptor. This group includes the constitutive androstane receptor (CAR) and the pregnane X receptor (PXR). They function as sensors of toxic byproducts derived from endogenous metabolites and of exogenous chemicals, in order to enhance their elimination. This unique function of CAR and PXR sets them apart from the steroid hormone receptors. The broad response profile has established that CAR and PXR are xenobiotic sensors that coordinately regulate xenobiotic clearance in the liver and intestine via induction of genes involved in drug and xenobiotic metabolism. In the past few years, research has revealed new and mostly unsuspected roles for CAR and PXR in modulating hormone, lipid, and energy homeostasis as well as cancer and liver steatosis. The purpose of this review is to highlight the structural and molecular bases of CAR and PXR impact on human health, providing information on mechanisms through which diet, chemical exposure, and environment ultimately impact health and disease.

Introduction

Living organisms are severely xenophobes in that they have an aversion to xenobiotics (from Greek, ξενoς “foreign” and βιoς “life”). This term was coined to cover all organic compounds that were foreign in the organism under study. The history of xenobiotics is interwoven with the birth of organic chemistry. In the early 1800s, the prevailing belief was that the composition of the human body, and indeed of all living things, was the result of a “vital force” or “internal flame” and that mere mortals were incapable of understanding these workings and would be particularly unable to synthesize constituent compounds of the human body (Murphy, 2001). Faced with this daunting concept, scientists were discouraged from thinking about the processes that dominated everyday digestion and nutrition. In 1828, Wöhler was able to shatter this myth of synthesizing urea, one of the compounds that he had examined in his studies on the urinary waste products, and sent the following note to Berzelius “I must tell you that I can prepare urea without requiring a kidney of an animal, either man or dog” (Whöler and Frerichs, 1848). In modern-day terminology, this was a true “paradigm shift” and led to revised thinking on what chemists could accomplish with regard to “organic” compounds. Wöhler test led to the first metabolism experiments performed in 1841 by Ure who demonstrated the conjugation of benzoic acid with glycine in man (Ure, 1841, Keller, 1842). The transformation of compounds such as cinnamic acid (Erdmann and Marchand, 1842) and benzaldehyde (Whöler and Frerichs, 1848) to benzoic acid, further emphasized by the discovery of the conversion of benzene to phenol (Schultzen and Naunyn, 1867), indicated the ability of the body to oxidize ingested compounds. A rapid development of chemical industries occurred soon after World War II, resulting in the manufacturing of a large number of chemical products. The increased use of fertilizers, insecticides, and herbicides led to a dramatic increase in world food production. This coupled with improved medicine and pharmacological science helped to improve general public health rising life expectancy to an average of 75 years (Yu, 2001). Whereas many people of the world were enjoying the benefits of technological and economic expansion and higher living standards, others perceived that these extraordinary developments were not without cost. Indeed, the impact of global environmental changes on human health has become a great concern. As early as the 1950s and 1960s, many urban dwellers and residents in the vicinity of industrial factories began to recognize undesirable changes in the environment (Yu, 2001).

Now it is clear that after a chemical pollutant gets into a mammalian organism, chemical reactions occur within the body, altering its reactivity and structural properties. The tome “Detoxication Mechanisms” by Williams (1947) heralded the birth of xenobiotic/drug metabolism as a distinct branch of science, providing a systematic approach based on the organic chemistry classification. Williams also expanded on his concepts of the principal biochemical reactions whereby drugs and other foreign compounds are metabolized in the body. Most importantly, he proposed that foreign compounds were metabolized in two distinct phases: one including oxidation, reduction, and hydrolysis processes, and the other involving conjugation reactions.

In the early 1950s, biochemical research unveiled the mechanisms of a wide variety of transformation reactions. The pioneering studies of James and Elizabeth Miller and Axelrod revealed the subcellular localization of metabolism reactions and brought to the identification of enzymes responsible for catalytic transformations (Mueller and Miller, 1949, Axelrod, 1955).

A major class of catalytic oxidative transformations was initially characterized in 1955 by Hayashi and Mason. Since these reactions need an oxidant (molecular oxygen) and a reductant (NAD(P)H), the enzymes catalyzing these processes were trivially named “mixed-function oxidases” (Hayashi et al., 1955, Mason et al., 1955).

The understanding of the biochemical nature of chemical transformations occurring in living organisms grew out from early studies on the liver pigments by Garfinkel and Klingenberg in 1958. These researchers observed in liver microsomes an unusual carbon monoxide-binding pigment with an absorbance maximum at 450 nm (Garfinkel, 1958, Klingenberg, 1958). This pigment was ultimately characterized as a cytochrome by Omura and Sato (1962). The function of this unique cytochrome (called P450; CYP) was initially revealed in 1963 by Estabrook, Cooper, and Rosenthal, while investigating the role of microsomes from the adrenal cortex in the catalytic conversion (i.e., hydroxylation) of 17-hydroxyprogesterone to deoxycorticosterone (Estabrook et al., 1963). In the clinical setting, the discovery of the CYP polymorphism role in debrisoquine and sparteine metabolism emphasized the importance of individual response to drugs (Thomas et al., 1976, Mahgoub et al., 1977, Eichelbaum et al., 1978). At present, 18 CYP families have been identified in mammals, among which only three are primarily responsible for most xenobiotic metabolism. Determining the molecular bases for the xenobiotic-mediated induction of CYP gene expression has become one of the most challenging dilemmas of modern toxicology (Dickins, 2004, Xu et al., 2005, Xu et al., 2005, Nakata et al., 2006, Plant, 2007, Graham and Lake, 2008).

Thus, the threat represented by xenobiotics for the organism is the target of a complex strategy. Metabolism of drugs and other xenobiotics in the liver is the body primary defence against accumulation of potentially toxic lipophilic compounds. This strategy comprises xenosensors, dedicated to the recognition of xenobiotics and dangerous endogenous molecules, as well as xenobiotic metabolizing and transporters systems. The first step of the detoxification mechanism is the xenobiotic detection by the constitutive androstane receptor (CAR) and/or the pregnane X receptor (PXR), both belonging to the nuclear receptor (NR) super-family (Fuhr, 2000, Dixit et al., 2005, Chang and Waxman, 2006, Moreau et al., 2008), and/or by the aryl hydrocarbon receptor (AhR), a member of the Per/ARNT/Sim (PAS) family of transcription factors (Francis et al., 2003).

CAR and PXR are activated by a variety of endogenous (i.e., steroids and bile acid salts), and exogenous ligands including drugs, insecticides, pesticides, and nutritional compounds (Fig. 1). Currently, it remains uncertain whether these receptors display a high ligand specificity or instead function as more generalized steroid/xenobiotic sensors. These xenosensors coordinate the expression of several genes encoding CYPs. Both CAR and PXR are able to elicit alterations in xeno- and endo-biotic metabolism, regulating the expression of phases I and II metabolizing enzymes and phase III transporters (Fig. 2) (Matic et al., 2007, Plant, 2007, Timsit and Negishi, 2007, Tompkins and Wallace, 2007, Graham and Lake, 2008, Lim and Huang, 2008, Ma and Lu, 2008, Moreau et al., 2008, Pascussi et al., 2008, Teng and Piquette-Miller, 2008, Köhle and Bock, 2009, Zhou et al., 2009). Although CAR and PXR play physiological roles (e.g., regulate cholesterol elimination pathways), no physiological ligands have been identified, therefore they still remain orphan NRs (Timsit and Negishi, 2007).

In contrast, AhR ligands are the hydrophobic environmental pollutants of polyhalogenated aromatic hydrocarbons, such as polychlorinated dibenzodioxins, dibenzofurans and coplanar biphenyls or polycyclic aromatic hydrocarbons (e.g., benzo[a]pyrene, 3-methylcholanthrene, benzoflavones, and omeprazole) (Francis et al., 2003, Nguyen and Bradfield, 2008, Gomez-Duran et al., 2009, Hahn et al., 2009).

Human CAR (hCAR), initially identified as the orphan receptor MB67 (Baes et al., 1994), was found as a heterodimer partner for the retinoic acid X receptor (RXR) (see Tompkins and Wallace, 2007, Graham and Lake, 2008, Lim and Huang, 2008, Ma and Lu, 2008, Moreau et al., 2008, Pascussi et al., 2008, Teng and Piquette-Miller, 2008, Köhle and Bock, 2009, Zhou et al., 2009). hCAR was able to bind to retinoic acid response elements (REs), but the extent of activation was much lower than that mediated by RXR. The isolation and characterization of mouse CAR (mCAR) was reported in 1997 (Choi et al., 1997); at that time, it was hypothesized to target a subset of retinoid acid REs, but its role in CYP regulation has not yet been realized (Stanley et al., 2006).

Meanwhile, separate lines of investigation were leading to the identification of mouse PXR (mPXR) that was identified in 1998 by using an expressed sequence tag to screen a mouse liver library (Kliewer et al., 1998). mPXR was named “pregnane X receptor” since it was shown to be activated by derivatives of dexamethasone and pregnenolone (Kliewer et al., 1998).

At approximately the same time, the human steroid X receptor was cloned in studies focused on the identification of the Xenopus laevis benzoate X receptor homologues. The human steroid X receptor was shown to be related to CAR and to the vitamin D receptor (VDR) (Blumberg et al., 1998), and then it was established to be the human homologue of mPXR (Lehmann et al., 1998). A parallel computer modeling approach confirmed that the human steroid X receptor is the human homologue of mPXR (Bertilsson et al., 1998). According to the literature (Bertilsson et al., 1998), the human steroid X receptor has been renamed human PXR (hPXR).

One of the most intriguing hypotheses to justify the existence of NR xenosensors was built on the basis of studies dealing with the broad implications of these receptors in steroid hormone homeostasis. It is intriguing to hypothesize that an ancestral protein could act as the endogenous metabolite (endobiotics) sensor. Starting from this hypothetical receptor progenitor, evolution proceeded by ligand exploitation and serial genome expansions. If this hypothesis is true, the conceptual distance between specific endogenous hormones and exogenous environmental molecules is probably closer than previously thought (Buters et al., 1994, Zhai et al., 2007).

NRs form a super-family of ligand-activated transcription factors implicated in various physiological functions from development to homoeostasis (Escriva et al., 2004, Ascenzi et al., 2006). NRs share a common evolutionary history as revealed by their conserved structure and by their high degree of sequence conservation (Escriva et al., 2004). The phylogenetic analysis of the NR super-family brought to its sub-division into six sub-families of unequal size (Fig. 3): (i) the large sub-family I contains the thyroid hormone receptors (THRs), the retinoic acid receptors (RARs), VDR, the ecdysone receptor, the peroxisome proliferator-activated receptors (PPARs), as well as numerous orphan receptors including CAR and PXR; (ii) the sub-family II includes RXR, the chicken ovalbumin upstream promoter-transcription factor (COUP-TF), and the hepatocyte nuclear factor 4 (HNF4); (iii) the steroid receptor sub-family III includes the estrogen receptors (ERs), the estrogen-related receptors (ERRs), the glucocorticoid receptors (GRs), the mineralocorticoid receptor (MR), the progesterone receptor (PR), as well as the androgen receptors (ARs); (iv) the sub-family IV contains the nerve growth factor inducible I-B group of orphan receptors (NGF1B); (v) the sub-family V includes the steroidogenic factor-1 (SF-1) and the Drosophila melanogaster “fushi tarazu” factor-1 receptor (FTZ-F1); and (vi) the small sub-family VI contains only the germ cell nuclear factor-1 receptor (GCNF) (Laudet et al., 1999, Escriva et al., 2004, Ascenzi et al., 2006).

A hypothetical evolutionary path that might have been taken by the first NR in the early Metazoan is represented in Fig. 3. Since several sub-families were present in early Metazoan, it appears that the super-family underwent an “explosive expansion” during early Metazoan evolution. The diversification of the super-family followed two waves of gene duplication. The first wave before the Protostome/Deuterostome split, during the emergence of Metazoan, led to the acquisition of the present six sub-families and the various groups of receptors within each sub-family. The second wave occurred after the arthropod/vertebrate split, specifically in vertebrates, producing the paralogous groups within each sub-family (see Laudet et al., 1999, Escriva et al., 2004, Ascenzi et al., 2006). In support of NR paralogous through genome or block duplications, mapping studies have demonstrated the presence of extensive “paralogy groups” which include NRs on different chromosomes (Owen and Zelent, 2000).

The first NR arose as one unit consisting of the currently recognized DNA-binding domain (DBD) and ligand-binding domain (LBD), based on the finding that the two domains existed together in lower Metazoan. However, such observation does not entirely dismiss the possibility that the two regions existed independently earlier in evolution, despite the fact that strong similarity to either domain has not yet been observed outside the Metazoan kingdom (Owen and Zelent, 2000). The ancestor DBD may have been similar to that of modern receptors, the LBD being present but not possessing a transactivation domain or the ability to dimerize (Owen and Zelent, 2000). The possibility that the VDR could be a chimera is in accordance with its clustering into different families based on the phylogenetic trees compiled from either DBD or LBD sequence (Laudet et al., 1999). The DBD of the VDR closely resembles that of FTZ-F1, while the LBD is closer to that of the RAR grouping. However, the more recently characterized CAR, PXR, and liver X receptor (LXR), along with several arthropod receptors, cluster together with the VDR in phylogenetic trees based on both the DBD and LBD, thus throwing doubt on the VDR being a chimera (Owen and Zelent, 2000).

Convergent evolution may be in part responsible for this phenomenon which is observed in other receptors such as GCNF and NGF1B. The possibility that each of these receptors diverged from a common chimeric ancestor still remains and a mechanism for such an event, originally suggested by Laudet et al. (1999), is represented in Fig. 3. In constructing the evolutionary tree represented in Fig. 3, three assumptions have been made: (i), the earliest Metazoan possessed COUP-TF, RXR, and FTZ-F1, all of which are well conserved to the present day and thus are close to the NR ancestor of subfamilies II and V; (ii) the ability to heterodimerize with RXR arose once and has diversified through sub-families I, II, and IV; and (iii) there is no correlation between the ligand recognized by a given receptor and its position in the family tree (see Laudet, 1997). The presence of RXR in the earliest branches of the Metazoan enables the extrapolation that an archaic family II member gave rise to families I and IV, which can also dimerize with RXR. It could be assumed from data in Cnidaria that a direct RXR ancestor developed the ability to bind the ligand 9-cis-retinoic acid with a subset diverging in RAR, thus acquiring the ability to bind all trans-retinoic acids (Owen and Zelent, 2000).

The gene comparison among human, chimpanzee, mouse, and rat suggests that the high degree of conservation within members of the NR super-family may have arisen from negative evolution selection against changes in the protein amino acid sequence (Clark et al., 2003, Zhang et al., 2004a). Indeed, the amino acid sequence identity between human and mouse ortholog NR genes is typically greater than 95% and 85% in the DBD and in the LBD, respectively (Zhang et al., 2004a, Iyer et al., 2006). The only two exceptions of such conservation are CAR-LBD and PXR-LBD, both characterized by a high cross-species amino acid sequence difference, indicating divergence during the course of evolution (Fig. 3) (Jones et al., 2000, Moore et al., 2002, Zhang et al., 2004a, Krasowski et al., 2005). Indeed, CAR-LBD and PXR-LBD show considerable differences in ligand recognition (Watkins et al., 2001, Watkins et al., 2003a, Watkins et al., 2003b, Shan et al., 2004, Suino et al., 2004, Xu et al., 2004, Chrencik et al., 2005, Xue et al., 2007a, Xue et al., 2007b, Teotico et al., 2008, Wang et al., 2008).

The percentage of identity among CAR and PXR expressed in several vertebrate species has been calculated by aligning their complete amino acidic sequences, and are summarized in Table 1, Table 2, respectively. While hCAR and mCAR share approximately 93.9% amino acid identity in their whole sequence (Table 1), only approximately 70% amino acid identity has been observed in their LBDs (Escriva et al., 2002, Robinson-Rechavi et al., 2004). Moreover marked differences have been observed in their responses to xenobiotics. Indeed, clotrimazole is an efficacious deactivator of hCAR but has little or no effect on the mCAR activity. Conversely, 3,5-dichloro-2-{4-[(3,5-dichloropyridin-2yl)oxy]phenoxy}pyridine (TCPOBOP) is a potent activator of mCAR but lacks any activity on hCAR. The divergence in the amino acid sequence across CAR orthologs undoubtedly contributes to cross-species differences in the physiological effects of xenobiotics (Moore et al., 2000a).

The analysis of the complete amino acid sequence of PXR of different species indicates an identity of 76.5% and 76% between hPXR and mouse and rat PXRs, respectively (Table 2). Similarly to CAR, the human, rabbit, mouse, and rat PXRs are all roughly equally divergent, sharing only approximately 80% amino acid identity in their LBDs (Moore et al., 2002, Zhang et al., 2004a). Indeed, although both the rabbit and human PXR are activated by rifampicin, there are marked differences in their responsiveness to synthetic steroids such as dexamethasone and cyproterone, as indicated by the level of CYPs gene expression in vivo (Jones et al., 2000). The PXR divergence represents an important component of cross-species differences in the regulation of CYP3A expression by xenobiotics, and could correspond to either an adaptive response to different environmental xenobiotic challenges or the different specificities of natural PXR ligands within vertebrate species. Despite their divergence, several lines of evidence suggest that PXRs are orthologs and not closely related members of sub-family 1I of NRs (Jones et al., 2000).

The cross-species variation in CAR-LBD and PXR-LBD is even more striking when DNA sequences are compared; in fact, the ratio between the rate of non-synonymous (nucleotidic substitutions that result in amino acid replacement) and synonymous (nucleotidic substitutions that do not cause an amino acid replacement) nucleotide variation provides information concerning evolutionary selective forces acting on a given gene (Yang and Bielawski, 2000). The analysis of this parameter strongly suggests that the natural selection has favored sequence diversity in CAR-LBD and PXR-LBD, possibly to adapt to cross-species differences in ligand recognition. CAR and PXR may thus represent unusual examples of NR genes that have changed their ligand specificity across vertebrate species to adapt to cross-species differences in the recognition of exogenous and/or endogenous toxic compounds (Kliewer et al., 1998, Schuetz et al., 2001, Xie et al., 2001, Moore et al., 2002, Krasowski et al., 2005, Reschly and Krasowski, 2006).

The different CAR and PXR ligand recognition properties constitute an especially unusual finding in the NR super-family, representing the more extreme divergence of ligand-binding residues of any of the ligand-activated NRs in vertebrates (Krasowski et al., 2005, Reschly and Krasowski, 2006). Indeed, residues involved in ligand binding are strongly conserved in other NRs as demonstrated by VDR, an endocrine NR closely related to PXR; only four residues are not conserved across vertebrate species, ranging from sea lamprey to human (Reschly and Krasowski, 2006).

The origin and evolution of the NR1I sub-family (which currently includes VDR, CAR, and PXR) is still a matter of debate, although it appears probable that a single NR gene duplicated early in vertebrate evolution gave origin to VDR and PXR genes (Reschly and Krasowski, 2006). These two genes then diverged from each other, and additional duplications have resulted in multiple VDR and PXR genes in some non-mammals species. Interestingly, distinct CAR and PXR genes appear to be solely found in mammals (Reschly and Krasowski, 2006). Indeed, the chicken has only a single “xenobiotic-responsive” NR1I gene, currently classified as CAR, the product of which has properties similar to both CAR and PXR (Fig. 3) (Handshin et al., 2004). A single “xenobiotic-responsive” gene is also present in Danio renio (Fig. 3) and Xenopus tropicalis (Reschly and Krasowski, 2006). A likely explanation is that an ancestral gene, similar to that of chicken CAR, duplicated just prior to or early in mammalian evolution. The two genes then diverged from each other to become the modern-day CAR and PXR genes found in all mammalian genomes sequenced so far (including opossum, seal, dog, pig, mouse, rat, and rhesus monkey) (Reschly and Krasowski, 2006).

The different amino acid sequences of vertebrate CAR-LBD and PXR-LBD contrast with the fact that non-synonymous substitutions are rare in hCAR and hPXR (Lamba et al., 2005). Interestingly, the hCAR and hPXR sequencing of 70 and 100 individuals from three different ethnic groups, respectively, shows very low nucleotide diversity (lower than the genome-wide averages for human genes) and no non-synonymous substitutions in the LBD coded by either gene (Zhang et al., 2001, Thompson et al., 2005). Overall, NR1I sub-family members show little variation in the LBD between human individuals. In addition, the nucleotide divergence between human and chimpanzee CAR and PXR is lower than the average of the nucleotide divergence for other genes in the human genome (Ebersberger et al., 2002, Clark et al., 2003, Thompson et al., 2005). This suggests that CAR and PXR ligands, relevant at least in terms of influencing reproductive fitness, do not vary between humans, and perhaps not even between humans and primates, but do vary between different animals (Ebersberger et al., 2002, Clark et al., 2003, Thompson et al., 2005).