ReviewVolume 5 | 2007
Cite this Article: Nuclear Receptor Signaling (2007) 5, e009.
The NR3B subgroup: an ovERRview
Department of Biochemistry, McGill University and Molecular Oncology Group, McGill University Health Centre, Montréal, Québec, Canada
Members of the NR3B group of the nuclear receptor superfamily, known as the estrogen-related receptors (ERRs), were the first orphan receptors to be identified two decades ago. Despite the fact that a natural ligand has yet to be associated with the ERRs, considerable knowledge about their mode of action and biological functions has emerged through extensive biochemical, genetic and functional genomics studies. This review describes our current understanding of how the ERRs work as transcription factors and as such, how they control diverse developmental and physiological programs.
4-OHT: 4-hydroxytamoxifen; AF-2: activation function 2; BMI: body mass index; CREB: cyclic AMP response element binding protein; DES: diethylstilbestrol; eNOS: endothelial nitric oxide synthase; ER: estrogen receptor; ERR: estrogen-related receptor; IFN-γ: interferon γ; NR: nuclear receptor; OPN: osteopontin; OXPHOS: oxidative phosphorylation; PDK4: pyruvate dehydrogenase kinase 4; PGC-1: PPARγ coactivator 1; PPAR: peroxisome proliferator activated receptor; RIP140: receptor interacting protein of 140 kDa; ROS: reactive oxygen species; SRC: steroid receptor coactivator; STAT3: signal transducer and activator of transcription 3
The nuclear receptor (NR) superfamily was originally defined as a group of structurally-related transcription factors controlling gene expression in response to binding to small lipophilic ligands best represented by the steroid and thyroid hormones, as well as vitamin D and the active derivatives of vitamin A [Evans, 1988]. However, it was soon realized that the number of nuclear receptors exceeded the number of known, classic lipophilic hormones, and receptors that could not be matched with a natural ligand were labeled as orphan nuclear receptors [Giguere, 1999; Giguere et al., 1988]. The emerging challenge was to perform “reverse endocrinology”, starting with a gene encoding a putative receptor and ending up with a corresponding natural ligand and/or the recognition of the developmental and physiological processes modulated by these receptor-like proteins [Kliewer et al., 1999]. Here, we will review the current knowledge on the NR3B subgroup of nuclear receptors, commonly known as estrogen-related receptors (ERRs), the first orphan nuclear receptors identified, and still in search of a natural ligand.
The NR3B subgroup includes three nuclear receptors referred to as ERRα (NR3B1, ERR1, ESRRA), ERRβ (NR3B2, ERR2, ESRRB) and ERRγ (NR3B3, ERR3, ESRRG), respectively. Members of the NR3B subgroup belong to the larger NR3 class of nuclear receptors that includes the classic steroid hormone receptors for estrogens, androgens, progesterone, aldosterone and cortisol. The first member of the subgroup (ERRα) was originally identified owing to the significant nucleotide and primary amino acid sequence similarities that it shared with the estrogen receptor α (NR2B1) gene and protein, while ERRβ was identified using the ERRα cDNA as a probe [Giguere et al., 1988]. Ironically, while ERRα was the first orphan nuclear receptor identified, ERRγ was the final addition to the superfamily [Eudy et al., 1998; Heard et al., 2000; Hong et al., 1999]. Although ERR homologs also exist in invertebrates such as Drosophila [Sullivan and Thummel, 2003] and amphioxus (Branchiostoma floridae) [Bardet et al., 2005], suggesting an ancient origin for the ERRs, it is not yet possible to identify the ancestral member of the NR3 group [Bardet et al., 2006]. The genomic organization of the three ERR loci also shares a structural characteristic unique among receptor isoforms within a subgroup of the superfamily. The exon encoding the amino terminal domain of the receptor also encodes the first zinc finger of the DNA binding domain, a genetic link between the two domains that could explain the unusual level of amino acid sequence identity present in the amino terminal domains of the three ERR isoforms. A single polypeptide of 423 amino acid residues encodes human ERRα, while several splice variants of ERRβ and γ have been identified in human (Figure 1A). The ERRβ2 variant contains an extended carboxy-terminal domain [Chen et al., 1999], ERRβ2Δ10 lacks exon 10 and encodes a different carboxy-terminal region [Zhou et al., 2006], the ERRγ2 splice variant possesses an additional 23 amino acids within its amino-terminal domain [Heard et al., 2000; Susens et al., 2000], while a new ERRγ splice variant lacking 39 amino acid residues of the second zinc finger of the DNA-binding domain has been found to be expressed in adipocytes and the prostate [Kojo et al., 2006]. The existence and relative abundance of most of these isoforms need to be confirmed by endogenous protein detection assays, while further studies are required to elucidate their putative physiological roles.
In general, the ERRs display a similar tissue distribution in both mice and humans. The ERR isoforms are ubiquitously expressed. However, ERRα is generally more abundantly expressed than ERRγ, which in turn is more abundant than ERRβ. The three isoforms are expressed at elevated levels in tissues subjected to high energy demand, such as the heart and kidneys. ERRα is also expressed at high levels in the intestine, brown adipose tissue and skeletal muscles, while ERRγ mRNA can be found in abundance in the brain stem and the spinal cord. ERRβ can be found at relatively high levels in a subset of extra-embryonic ectoderm in the developing placenta and undifferentiated trophoblast stem cell lines, as well as in the adult eyes, inner ear, heart and kidneys [Bookout et al., 2006; Chen and Nathans, 2007; Giguere et al., 1988; Luo et al., 1997; Pettersson et al., 1996; Tremblay et al., 2001b]. The graphical views of the tissue-specific mRNA expression patterns for the three mouse ERR isoforms are available at [Bookout et al., 2005].
Interestingly, it was shown that the expression of all three ERR isoforms displays distinct diurnal rhythmicity in tissues such as the liver, white adipose, skeletal muscle, uterus and bone [Horard et al., 2004; Yang et al., 2006], suggesting that the ERRs may serve as a molecular link between the circadian oscillator and energy metabolism (see below). In addition, physiological stress signals such as exposure to cold, exercise or fasting also induce ERRα expression in brown fat, skeletal muscles and liver, respectively [Cartoni et al., 2005; Ichida et al., 2002; Schreiber et al., 2003]. It has also been shown that ERRα expression in bone marrow-derived macrophages is activated by lipopolysaccharide, interferon γ (IFN-γ) and interleukin 4 [Barish et al., 2005; Sonoda et al., 2007]. In addition, a 2-fold increase of ERRα mRNA expression has been demonstrated during differentiation of human bone marrow-derived mesodermal progenitor cells into osteoblasts [Qi et al., 2003].
Little is known about how the expression of the three ERR genes is controlled. The main regulator of ERRα expression has been shown to be the ERRs themselves [Laganiere et al., 2004; Liu et al., 2005; Mootha et al., 2004]. The ESRRA promoter contains a polymorphic 23 base pair sequence (ESRRA23) that is present in 1-4 copies in human. Each ESRRA23 element contains one perfect ERRE, as well as an additional nuclear receptor half-site. Transient transfection experiments have demonstrated that induction of the ESRRA promoter by PGC-1α is dependent on the presence of the ESRRA23 element, and that the strength of the activation correlates with its dosage [Laganiere et al., 2004].
The three ERRs are constitutively active transcription factors. Their transactivation properties are independent of any exogenously added natural ligand and their relative potency as transcriptional activators varies in cell context- and promoter-dependent manners. The three ERR isoforms bind to a number of coregulator proteins, which they also share with other NRs. Their transcriptional activity is increased by members of the steroid receptor coactivator (SRC) family (SRC-1, TIF-2/SRC-2, AIB1/ACTR/SRC-3) [Hong et al., 1999; Xie et al., 1999; Zhang and Teng, 2000], the peroxisome proliferator-activated receptor γ coactivator-1 (PGC-1) α and β [Huss et al., 2002; Kamei et al., 2003; Laganiere et al., 2004; Schreiber et al., 2003; Sonoda et al., 2007], as well as the proline-rich nuclear receptor coregulatory protein (PNRC), PNRC2 and transducin-like enhancer of split 1 [Hentschke and Borgmeyer, 2003; Zhou and Chen, 2001; Zhou et al., 2000]. The transcriptional activity of the ERRs is also modulated by the nuclear receptor corepressor RIP140/Nrip1 [Augereau et al., 2006; Castet et al., 2006; Debevec et al., 2007; Sanyal et al., 2004]. The ERRs can interact with the orphan nuclear receptor small heterodimer partner (SHP; NR0B2), which represses their transcriptional activity [Sanyal et al., 2002]. SHP lacks a conventional DNA binding domain and interacts with several other members of the nuclear receptor superfamily to inhibit their receptor transcriptional activity. It is interesting to note that the mutations in SHP that have been associated with moderate obesity in humans prevent the inhibition of ERRγ activity [Sanyal et al., 2002].
Despite the absence of response to natural estrogens, it has been reported that the transcriptional activity of all ERR isoforms is inhibited by the synthetic estrogen analog diethylstilbestrol (DES) [Coward et al., 2001; Tremblay et al., 2001b]. The constitutive transcriptional activity of ERRβ and ERRγ, but not ERRα, can also be repressed by the selective estrogen receptor modulator (SERM) 4-hydroxytamoxifen (OHT), which behaves like a selective inverse agonist, causing the dissociation of coactivator protein [Coward et al., 2001; Tremblay et al., 2001a]. Recently, bisphenol A, a ubiquitous environmental contaminant with estrogenic activity, was shown to bind to ERRγ and antagonize the repression of ERRγ activity induced by OHT [Takayanagi et al., 2006]. Also, toxaphene and chlordane, two organochlorine pesticides with estrogen-like activity, have been identified as weak antagonists for ERRα [Yang and Chen, 1999]. However, mutation of phenylalanine 329, an amino acid crucial for the constitutive activity of the receptor, to an alanine residue, allowed toxaphene to act as an ERRα agonist instead [Chen et al., 2001]. In addition, the isoflavones genistein, daidzein, and biochanin A and the flavone 6,3',4'-trihydroxyflavone were identified as agonists of the ERRs by mammalian two-hybrid experiments under comparable conditions to those for the activation of ERα and ERβ [Suetsugi et al., 2003]. It should be noted that binding of the organochlorine pesticide molecules to ERRα could not be directly demonstrated [Tremblay et al., 2001b] and that structure-based predictions showed that flavone and isoflavone cannot be accommodated within the ERR binding pocket [Greschik et al., 2002]. Although these compounds constitute useful tools to study the regulation of the ERRs, they share the pitfall of being non-specific to the ERRs, as they modulate the activity of other nuclear receptors such as the ERs. However, several synthetic compounds have recently been characterized as ERR-specific ligands. XCT790 is a highly specific inverse agonist for ERRα that disrupts the interaction between ERRα and PGC-1α and has no effect on either the ERs or the other ERR isoforms [Busch et al., 2004; Willy et al., 2004], while GSK5182, a tamoxifen analog, showed improved inverse agonist selectivity for ERRγ [Chao et al., 2006]. On the other hand, the structurally-related phenolic acyl hydrazones GSK4716 and DY131 were reported to effectively and selectively activate ERRβ and ERRγ [Yu and Forman, 2005; Zuercher et al., 2005].
The possible existence of a natural ligand for the ERRs remains an unresolved question. The initial crystal structure of unliganded ERRγ showed that the ligand-binding pocket is very small, approximately 280 Å [Greschik et al., 2002]. Likewise, the unoccupied volume of the ligand-binding pocket of the unliganded ERRα bound to a PGC-1α coactivator peptide was reported to be ~100 Å [Kallen et al., 2004], suggesting that a natural agonist would have to be composed of at most four to five non-hydrogen atoms. However, the crystal structures of the GSK4716 agonist-bound ERRγ and of the cyclohexylmethyl-(1-p-tolyl-1H-indol-3-ylmethyl)-amine inverse agonist-bound ERRα revealed that these ligands can force the rearrangement of amino acid residues in the respective ligand binding domains that allows access of the ligands to a larger ligand-binding pocket. The results of these experiments thus suggest that the plasticity of the ERR ligand binding pockets could allow for larger compounds to act as natural ligands [Kallen et al., 2007; Wang et al., 2006]. The crystal structures of ERRα and ERRγ ligand-binding domains also provided strong evidence for ligand-independent transactivation by these receptors. In both cases, the apo-receptors are in a permanent active configuration and thus ready to interact with coregulatory proteins [Greschik et al., 2002; Kallen et al., 2007; Kallen et al., 2004; Wang et al., 2006]. Because binding of the agonist GSK4617 does not affect the orientation of the AF-2 helix, the mechanism by which GSK4617 activates ERRγ is currently unknown. In contrast, binding of DES and 4-OHT to ERRγ induces the displacement of the AF-2 helix to a position that interferes with the recruitment of coactivators [Greschik et al., 2004; Wang et al., 2006].
The physiological roles played by nuclear receptors can often be ascribed by investigating their target genes. Unbiased binding site selection and characterization of the first ERR-responsive genes defined the ERR response element (ERRE) as the consensus nucleotide sequence TCAAGGTCA [Johnston et al., 1997; Sladek et al., 1997a; Vega and Kelly, 1997]. Bioinformatic analysis of a large set of ERR target promoters identified using a combination of chromatin immunoprecipitation (ChIP) and genomic DNA arrays (ChIP-on-chip) confirmed that the TCAAGGTCA motif serves as the main ERR binding site in vivo (Figure 1B) [Dufour et al., 2007]. Although the consensus ERRE contains a single nuclear receptor core binding half-site, the ERRs can bind DNA either as monomers, homodimers or heterodimers (Figure 1C) [Barry et al., 2006; Dufour et al., 2007; Gearhart et al., 2003; Huppunen and Aarnisalo, 2004; Johnston et al., 1997; Sladek et al., 1997a; Vanacker et al., 1999a]. ERRα and ERRβ have also been shown to bind the estrogen response element (ERE) as homodimers [Vanacker et al., 1999b; Yang et al., 1996; Zhang and Teng, 2000], and two studies even suggested that ERα and ERRα could heterodimerize in vitro [Johnston et al., 1997; Yang et al., 1996]. However, convincing evidence that the ERRs can commonly bind to EREs and/or interact in a physiologically significant manner with the ERs in vivo is still lacking. A study of the TFF1 (pS2) gene showed that while the promoter contains both an ERE and an ERRE, regulation by the ERRs is lost only when the ERRE is ablated, suggesting that the ERRE is the ERRs’ legitimate binding site, at least on this promoter [Lu et al., 2001].
Biochemical purification of HeLa cell nuclear extract proteins that bind to the Simian Virus 40 (SV40) late promoter first identified ERRα as a repressor of the transcription of the SV40 late genes [Wiley et al., 1993; Zuo and Mertz, 1995]. In contrast, the ERRs have been shown to mainly activate the expression of cellular genes. Until recently, most ERR target genes had been identified through the discovery of a putative ERRE in the promoters of genes of interest. These included genes encoding the medium-chain acyl coenzyme A dehydrogenase (MCAD or Acadm) [Sladek et al., 1997a; Vega and Kelly, 1997], osteopontin (OPN) [Bonnelye et al., 1997; Vanacker et al., 1998b], the thyroid receptor α (TRα) [Vanacker et al., 1998a], aromatase [Yang et al., 1998], lactoferrin [Yang et al., 1996], the orphan nuclear receptor SHP [Sanyal et al., 2002], endothelial nitric oxide synthase (eNOS) [Sumi and Ignarro, 2003], PPARα [Huss et al., 2004], pyruvate dehydrogenase kinase 4 (PDK4) [Araki and Motojima, 2006; Wende et al., 2005; Zhang et al., 2006a], monoamine oxidase B (MAO-B) [Willy et al., 2004; Zhang et al., 2006b], ERRα itself [Laganiere et al., 2004; Liu et al., 2005; Mootha et al., 2004], apolipoprotein A4 (ApoA4) [Carrier et al., 2004], phosphoenolpyruvate carboxykinase (PEPCK) [Herzog et al., 2006], surfactant protein A (SP-A) [Liu et al., 2006], RIP-140/Nrip1 [Nichol et al., 2006], mitofusin 2 [Soriano et al., 2006], Polo-like kinase 2 (Plk2) [Park et al., 2007] and uncoupling protein 1 (UCP-1) [Debevec et al., 2007]. A second, more global approach to identifying ERR-responsive genes was to exploit the observation that ERRα transcriptional activity is highly stimulated in the presence of PGC-1α. Various cell lines were first infected with an adenovirus expressing either wild-type PGC-1α or a PGC-1α variant engineered to specifically interact with ERRα, and differential gene expression profiling was then carried out using DNA microarrays [Gaillard et al., 2006; Mootha et al., 2004; Rangwala et al., 2007; Schreiber et al., 2004]. A small subset of bona fide ERRα targets were further identified using a combination of computational biology, small inhibitory RNAs (siRNAs) against ERRα, a specific antagonist, as well as DNA binding assays and cotransfection of reporter genes. These targets included several genes involved in oxidative phosphorylation (OXPHOS) and mitochondrial biogenesis, such as ATP synthase b (ATPsynb), cytochrome c (CYCS), COX4, GABPA, adenine nucleotide translocator 1 (ANT1) and carnitine palmytoyltransferase 1A (CPT1A), thus suggesting that ERRα could serve as a conduit for PGC-1α action in mitochondrial biogenesis and the OXPHOS pathway.
A powerful approach to identify direct target genes of nuclear receptors is to determine their occupancy on a genome-wide scale by using the ChIP-on-chip technology [Carroll et al., 2005; Carroll et al., 2006; Laganiere et al., 2005; Odom et al., 2004]. This technique was recently used to appraise the role of ERRα and ERRγ in the adult heart [Dufour et al., 2007], of ERRγ in the newborn heart [Alaynick et al., 2007] and of ERRα in bone marrow-derived macrophages [Sonoda et al., 2007]. Together, these studies identified more than 500 promoters that considerably expanded the repertoire of direct ERR target genes encoding cytosolic and mitochondrial proteins involved in the control of energy metabolism. In addition, these studies showed that the ERRs control tissue-specific functions, such as fuel sensing and contractile work in the heart and bacterial clearance in macrophages. Further analyses of the ChIP-on-chip data demonstrated that ERRα and ERRγ target the same promoter as non-obligatory heterodimers and cooperate with other transcription factors, namely CREB and STAT3, to control of the expression of metabolic genes [Dufour et al., 2007].
Review of the data gathered on ERR coregulatory proteins and target genes described above also reveals an important emerging concept of ERRs functioning in a regulatory network. First, induction of PGC-1α expression by physiological stimuli leads to the upregulation of ERRα, which in turn stimulates its own expression, thereby generating a positive feedback loop [Laganiere et al., 2004; Mootha et al., 2004]. Second, ERRα can also stimulate the expression of the corepressor RIP140 and SHP, thereby providing an inhibitory feedback mechanism to control the expression of its target genes [Nichol et al., 2006; Sanyal et al., 2002]. Third, since the three ERR isoforms can regulate the same target genes as homo- or heterodimers, the expression of these genes can be differentially regulated depending on the levels of individual ERR isoform in each tissue. The existence of such networks has been shown in vivo where, in the ERRα null heart, the expression of RIP140 is downregulated, while that of PGC-1α is upregulated at a time when the expression of ERRγ is also elevated [Dufour et al., 2007]. The compensatory mechanism was particularly evident in HL-1 cardiomyocytes, where ERRγ was shown to play an important role in maintaining the expression of ERRα target genes. A similar compensatory mechanism was observed in the neonatal heart lacking ERRγ [Alaynick et al., 2007].