Nuclear receptors (NRs) are a superfamily of transcription factors, which are ligand-dependent and homologous to steroid hormone receptors. NRs are widely distributed and have important physiological functions in cell development and differentiation, circadian rhythm, metabolism and immune regulation. NRs consist of three components: the steroid hormone receptors, non-steroid hormone receptors and the orphan nuclear receptor family. Steroid and non-steroid hormone receptors have specific ligands, including steroid hormones, thyroid hormones, retinoic acids and fatty acids. Ligands for orphan NRs have not yet been determined. Retinoic acid-related orphan receptors (RORs), also known as nuclear receptor subfamily 1 group F members (NR1F), are specified by gene sequences, which are homologous to retinoic acid receptor (RAR) and retinoid X receptor (RXR) which belong to non-steroid receptors (1,2). RORs include RORα, RORβ and RORγ, which are also referred to as RORA, RORB and RORC or NR1F1-3, respectively, and have been cloned from different mammalian species. The molecular mechanisms and physiological functions of RORα and RORγ are well established. However, the function of RORβ needs to be further elucidated. This review focuses on the structure of RORs and aims to provide an overview of the present studies on the functions that RORβ has in several biological and pathological processes, particularly in terms of circadian rhythm abnormalities and tumorigenesis.

RORα contains isoforms RORα 1–4, with only RORα1 and RORα4 being present in mice. Two isoforms of RORβ are found in mice and humans (RORβ1 and RORβ2), although studies have reported that only RORβ1 exists in humans. RORγ also has two isoforms (RORγ1 and RORγ2). RORγ2, originally found in the immune system, is often regarded as RORγt and is important in thymocyte development, with its expression being highly restricted to the thymus. The two isoforms of RORγ are found in mice and humans (3,4).

RORs share a common modular structure composed of four functional domains including an amino-terminal A/B domain, a DNA-binding domain (DBD), a hinge region and a carboxy-terminal ligand-binding domain (LBD). The A/B domains are highly variable in sequence between different ROR isoforms (5,6). The DBD consists of two highly conserved zinc finger motifs involved in the recognition of ROR response elements (ROREs) that contain the consensus motif AGGTCA preceded by a 5-bp AT-rich sequence. The RORs bind to ROREs as a monomer to regulate the transcription of target genes (7,8). The hinge region has shorter sequences and is thought to be a flexible domain that connects the DBD with the LBD. Its main roles are to maintain the structural stability of RORs and influence intracellular trafficking and subcellular distribution. The multifunctional domain LBD, consisting of 12 classic α helices (H1-12) and two additional α helices (H2′ and H11′), plays multiple roles in ligand binding, nuclear localization, receptor dimerization, and serves as an interface for the interaction with co-activators and co-repressors (9). RORs have two activation domains, including the ligand-independent activation function-1 domain (AF-1), which is localized in the N-terminal domain and the ligand-dependent AF-2, which resides in the C-terminal domain. The transcriptional activation of AF-1 is commonly very weak, but it can synergize with AF-2 to produce a more robust upregulation of gene expression. AF-2, located in H12 and consisting of the PLYKELF motif, is 100% conserved among RORs and plays a dominant role in transcriptional regulation. The study on the crystal structure of RORα suggests that deletion or point mutation in H12, in particular Y507A, may result in a decrease in transcriptional activity and a dominant negative RORα (1).

RORs are conservative during evolution. Orthologs of RORs have been identified in some lower species, such as Drosophila hormone receptor 3 (DHR3) in Drosophila melanogaster, caenorhabditis hormone receptor 3 (CHR3) in Caenorhabditis elegans and manduca hormone receptor 3 (MHR3) in Manduca sexta (10–12). The DBDs of RORs are highly conserved and the DBD of RORγ exhibits a 92 and 75% identity with that of RORβ and RORα, respectively (13). Although the LBD sequence is moderately conserved and does not have a high degree of homology (63% for RORα and RORβ, respectively, and 58% for RORα and RORγ, respectively) (5,14), their secondary structure is very similar and always contains 12α-helices (H1-H12). H12 is 100% conserved among RORs and contains the AF2 consensus motif ΦΦXE/DΦΦ (Φ denotes a hydrophobic amino acid and X denotes any amino acid) (15), indicating that RORs may likely have similar molecular functions.

RORα is widely distributed in multiple tissues including brain, liver, pancreas, kidney, thymus, skeletal muscle, testis, ovary, lung, skin and fat tissue, and is most highly expressed in the cerebellum and hypothalamus (16–18). RORγ is mainly expressed in the thymus, kidney, skeletal muscle, heart, liver, pancreas and testis, and is particularly highly expressed in immune cells (4,19). RORβ was initially identified as a member of the orphan nuclear receptor family by Carlberg et al (3). It has been mapped to human chromosome 9q21.13, the RORβ gene, which comprises 10 exons and spans approximately a region of 188 kb of the genome (Table I). RORβ1 and RORβ2 share the common DBD and LBD, but are characterized by a different A/B domain, which, respectively, contains 2 and 13 amino acids. The N-terminal 2–13 amino acids of RORβ1 are replaced by an arginine. RORβ1 and RORβ2 consist of 459 and 470 amino acids, respectively (Fig. 1). It is likely that RORβ1 and RORβ2 originate from the same gene by either alternative splicing or transcription from an alternative promoter. RORβ was primarily detected by northern blot analysis and its expression was restricted to the central nervous system, in particular in regions involved in the modulation of circadian rhythms, such as suprachiasmatic nucleus (SCN), pineal gland, and retina (13,20–22). Recently, with the increasing sensitivity of detection, RORβ has been found outside the nervous system such as normal bone tissue, pancreatic and endometrial cancer, and uterine leiomyosarcoma (23). The expression of RORβ in normal intestinal epithelial cells and intestinal tumors has been identified using the methods of mRNA view and qPCR. In addition, intestinal tumor tissues show a lower level of RORβ when compared with paralleled adjacent intestinal tissues. In the pathological state, the expression profile of RORβ is altered, indicating that the distribution of RORβ may be more widespread than is currently known. RORβ1 and RORβ2 differ from each other as regards their distribution. RORβ1 is highly expressed in cerebral cortex layer IV, thalamus, and hypothalamus, while comparatively low expressed in retina and pineal gland. By contrast, RORβ2 is the predominant isoform in retina and pineal gland, and little is found in the cerebral areas. When compared with RORβ1, the level of RORβ2 mRNA oscillates robustly with true circadian rhythmicity and is elevated to reach its maximal value at night. RORβ1 plays a leading role in serving sensory input integration, while RORβ2 mainly affects circadian rhythmicity. The different A/B domain may contribute to their different functions.

RORβ is a ligand-dependent transcription factor, although the identification of its functional ligands has not been determined. Melatonin was initially presumed as a natural ligand of RORβ due to its rhythmic synthesis and activity. RORβ can regulate circadian rhythms, and the mRNA level of RORβ coincides with melatonin production in the retina and pineal gland (22,24). However, subsequent studies have not confirmed this finding. Masana et al (25) found that the level of pineal melatonin exhibited a robust and significant diurnal rhythm with low levels during the day and high levels during the night in RORβ (C3H)^+/+, RORβ (C3H)^+/− and RORβ (C3H) ^−/− mice, indicating that RORβ does not affect the production of pineal melatonin and melatonin is not involved in the role of RORβ-regulating circadian rhythm. In the study on the X-ray structure of LBD of RORβ, stearic acid was found to bind to RORβ and act as a filler and stabilizer rather than a functional ligand. Later studies (1,26) demonstrated that all-trans-retinoic acid (ATRA) and synthetic retinoic acid (ALRT) 1550 reversibly combined to LBD of RORβ with a high affinity, which effectively reduced the transcriptional activity mediated by GAL4-RORβ in vitro. They are functionally equivalent to an inverse agonist. However, the effect of ATRA on RORβ activity is cell-type-dependent. In the neuronal cell lines HT22 and Neuro2A, ATRA antagonized transactivation by RORβ, while in cells such as NIH3T3, 293, and P19, ATRA showed no effect. ATRA and ALRT 1550 are believed to be bona fide ligands for RORβ, but whether they can regulate the transcriptional activity of full-length RORβ and its target genes remains to be proven. Furthermore, ATRA and ALRT 1550 can also bind to RORγ and inhibit its transcriptional activity, but not to RORα. Cholesterol, identified as a putative RORα ligand, neither binds to nor modulates RORβ activity (26). Based on the finding of the crystal structure study that the AF-2 side of ligand-binding pocket (LBP) is strictly hydrophobic, it was predicted that hydrophobic molecules with a carboxylic head are more likely to be ligand candidates for RORβ (9). On the issue of synthetic ligands, significant progress has been made as regards RORα and RORγ, while little is known with regard to RORβ. Studies on its specific ligands may provide insight into our understanding of RORβ (15).

Nuclear receptor transcriptional activation domain AF-1 and AF-2 are important in protein-protein interactions (27). Co-activators such as the RNA co-activator SRA and the RNA-binding DEAD-box protein (p72/p68) interact with AF-1 (28). The co-activators acting on AF-2 including the BRG1 complex, the p160 steroid receptor co-activator (SRC) family, CBP/p300, and the TRAP/DRIP/ARC complex enhance the receptor's transcriptional activity through histone acetylation or methylation. By contrast, co-repressors such as the nuclear receptor co-repressor (NCoR) and its closely associated protein, the silencing mediator for retinoid and thyroid hormone receptor (SMRT) repress gene transcription through deacetylation by histone deacetylases (HDACs) (29). Transcriptional regulation by RORs is mediated via an interaction with co-repressors such as NCoR or co-activator complexes such as SRC-1, indicating that RORs can act as repressors and activators of transcription (30). RORβ, which contains AF-1 and AF-2, also has those molecular functions. It was also found that RORβ proteins in humans and mouse have acetylation sites K27 and K98, respectively. Thus, the regulation of target genes by RORβ is likely to be more complex.

To determine the target genes regulated by RORβ, Greiner et al (31) cloned a 27-kDa protein consisting of 229 amino acid residues from a cDNA library derived from mouse brain by using the method of yeast two-hybrid screen and designated it as neuronal interacting factor X1 (NIX1). NIX1 was originally identified as a neuronal-specific cofactor and is exclusively expressed in the brain including in neurons of the dentate gyrus, amygdala, thalamus, hypothalamus and several brainstem nuclei. A subgroup of NRs such as RAR and TR but not RXR or steroid hormone receptors can interact with NIX1. Furthermore, RAR and TR only interact with NIX1 in the presence of their cognate ligands, whereas NIX1 does not bind to RXR in the absence or presence of the ligand. There is no similarity of protein domains of NIX1 with other nuclear receptor cofactors except for two nuclear receptor-binding LXXLL motifs, which are also identified as nuclear location signals. One is located within the C terminus (amino acids 192–196) and another is in an opposite orientation within the central part of the protein (amino acids 87–91) (31). Previous findings have demonstrated that LXXLL is required for the ligand-dependent binding of transcriptional co-activators to NRs (32,33). Greiner et al (31) found that NIX1 directly combined RORβ in vitro and in vivo, and specifically inhibited its activity in a dose-dependent manner. They also identified a minimal protein fragment spanning amino acids 61–99, which is both necessary and sufficient for the interaction between NIX1 and RORβ. Additionally, the fragment contains a nuclear receptor-binding motif in an inverted orientation (LLQAL, amino acids 87–91), which is required for the binding of NRs by NIX1. The AF-2 core motif of RORβ located at amino acid residues 445–451 is necessary for the interaction with NIX1. This finding is consistent with studies showing that elimination or deletion of AF-2 may cause the molecule to behave as a constitutive repressor or inactivate the protein (27,31,34).

NRIP2 is a protein derived from humans and it is homologous to NIX1. Compared with NIX1, NRIP2 has an extra 50 amino acids on the N terminal and is mainly distributed in the cytoplasm, which is different from NIX1. It has been found that NRIP2 and NIX1 belong to the aspartyl protease family. They are aspartyl endopeptidases and homologous to DNA damage-inducible protein (Ddi). Ddi1-related aspartyl proteases are believed to contain human homologues such as Ddi1 and Ddi2 and neuron-specific NRs NIX1, NRIP2 and NRIP3. However, the difference is that Ddi1 possesses three domains: a retroviral aspartyl-protease domain (RVP), an NH2-terminal ubiquitin-like domain (UBL), and a COOH-terminal ubiquitin-associated domain (UBA) and it is a ubiquitin receptor (35). NRIP2 lacks the three domains and contains only one conservative D-S/T-G fingerprint. The abovementioned findings suggest that NRIP2 functionally may be irrelevent to its ubiquitinated target proteins. RORβ is likely to be one of the major substrates for aspartyl endopeptidases NRIP2.

It is predicted that there are numerous interacting proteins for RORβ via UniProtKB, MINT, STRING and I2D, such as Nm23-H1(NME1), Nm23-H2(NME2), NCOA1 and MAP6. Nm23 and MAP6 are associated with cytoskeletal movement, suggesting that RORβ may be involved in the regulation of cytoskeletal movement.

Although RORs have similar structures comprising four homologous functional domains, there are obvious differences in physiological functions, as well as in the expression between RORs. RORα, RORβ and RORγ regulate circadian rhythms and RORα plays the central role (36). RORα has a key role in the development of the cerebellum, particularly in the regulation of the maturation and survival of Purkinje cells and the formation of bone. In RORα^−/− mice, Purkinje and granule cells are decreased and this results in cerebellar atrophy (17). RORγ is required for lymph-node organogenesis and the formation of multiple lymphoid tissues such as Peyer's patches, crypto patches, and isolated lymphoid follicles in the intestine. RORγ^−/− mice are deficient in lymph nodes. Moreover, RORγ promotes the differentiation of T cells and the development of lymphoid tissue-inducer (LTi) cells (37). RORβ is necessary for the proliferation and differentiation of retinal cells in addition to the maintenance of normal circadian rhythms. At birth, the retina of RORβ^−/− mice appears very similar to that of wild-type mice with regard to morphology, but in adulthood it is disorganized and lacks the normal layer structure. The degeneration of retina occurs during the first weeks after birth and eventually results in blindness (24). RORβ^−/− mice also exhibit behavioral changes such as reduced anxious behaviors, increased exploratory activities, changes in motor function occurring such as ‘duck gait’ decline in male reproductive capacity, and olfactory dysfunction (25). Recent studies have found that RORβ plays a role outside the neural system (38–41).

RORβ, as an important member of orphan nuclear receptor family, plays important regulatory roles in the maintenance of a variety of physiological processes and physiological rhythms. Circadian rhythm abnormalities are increasingly being considered as a novel incentive for tumorigenesis. Therefore, to elucidate the molecular mechanisms of RORβ, regulating tumor-associated circadian rhythm abnormalities may yield important clinical benefits on effective intervention and blocking of abnormal circadian rhythm-induced tumorigenesis.