The relationships between the different paralogs of RORs in vertebrates have remained unclear, in part due to the abundance of products for this gene family in teleost fish. We therefore set out to first clarify the ROR family relationships. In our phylogenetic analysis, rors were classified into four main groups, a new fourth group being created against the conventional ror subtypes, nr1f1–3. This additional subtype consists of rorca, rorcb, and rorb-like homologs. We propose to name this group rord or nr1f4.
On the basis of the phylogenetic tree, we propose that a proto-ortholog of ror in the cephalochordate ancestral genome gave rise to two paralogs, the common ancestor for both rora and rorc and its counterpart (rorb + rord), after an initial round of duplication. A total of four orthologs were generated thereafter following a second genome duplication event. Interestingly, orthologs belonging to nr1f4/rord were not found in mammalian taxa in any databases. Nr1f4/rord thus appears to have been lost in the mammalian lineage during evolution, as these orthologs were present in other tetrapods and in fish. In the latter group, the fourth ror subtype gave rise to two daughters, rord1 and rord2, after yet another round of genome duplication; this third duplication is a teleost-specific whole genome-duplication (TWGD) and is believed to have occurred around 320 million years ago [16]. Indeed, the presence of only a single nr1f4/rord1 ortholog in taxa including chicken, turtle, and the holostean (non-teleost) spotted gar reinforces the notion that rord1 and rord2 paralogs appeared as a consequence of TWGD. Although nr1f1, nr1f2 and nr1f3 could be expected to also have two paralogs after TWGD, just like nr1f4, there was no evidence to support this; two functional NR1F4s are seemingly needed in teleosts, while this subtype appears to be dispensable in mammals.
Among the tissues we examined, the eye showed the highest expression of rord1 mRNA in adult medaka at ZT7. Expression of ror subtypes in various organs, including the eye, has been documented in different species; for example, Rora mRNA was detected in moderate quantities in muscle and skin in mouse, but its principal expression site is the brain, particularly the thalamus and Purkinje cell layer of the cerebellum [17]. The Rorb gene is highly expressed in the pineal gland, the thalamus and the hypothalamus in the rat. In situ hybridization analysis revealed that the suprachiasmatic nucleus and the inner nuclear layer of the retina in particular are high-expression Rorb sites [18]. Similarly, Rorc expression has been reported in muscle and the thymus in mammals. In mouse, Rorγt, which is an alternative splicing form of Rorc in the thymus, was shown to be an essential factor for differentiation of naïve CD4+T (Th0) cells into Th17 cells, indicating an immunologically relevant function [19].
Unlike their mammalian counterparts, in non-mammalian species only a few studies on localization and function of rors/RORs have been reported. Using whole-mount in situ hybridization analysis, rora, rorb and rorc expression were detected in diencephalon, eyes, pineal gland, pharynx, heart, liver, gut, and somites of the zebrafish embryo [20]. The rora and rorc were strongly expressed in pituitary, brain and immune-related tissues in grass carp, suggesting that these subtypes are associated with the endocrine and immune systems [21]. In rainbow trout, rorc was reportedly highly expressed in muscle, other sites of expression being brain, heart, and skin [22]. In keeping with these findings on rors in fish, medaka rord1 was also detected in the brain. It is unclear whether rord1 is expressed universally in the eye, as there are no studies of tissue localization of this subtype in other species. In any case, its conspicuous expression in the eyes suggests that RORd1 may relate to visual functions; indeed, there are two earlier reports that implicate RORa and RORb in the regulation of the expression of opsins, suggestive of a photoreceptor function, in cone cells [23, 24]. Furthermore, RORd1 could be involved in yet-unknown physiological responses characteristic to teleost fish.
Rord1 mRNA abundance peaked during the middle of the light phase in eyes of medaka in the present study. It is thus tempting to speculate that RORd1 may be involved in synchronizing the circadian clock, or that it plays a role in regulation of rhythmicity. An example of the latter, via RORE, RORs are likely to control the expression level of bmal1 whose products constitute a core loop of the circadian rhythm [25, 26]. In a promoter assay performed using bmal1a-luc in this study, although 4OH-ATRA stimulated the expression of luciferase without RORd1, significantly higher expression of luciferase was detected in the presence of RORd1 than in the absence of RORd1. This suggest two possibilities: (1) the endogenous 4OH-ATRA receptor binding to bmal1a promoter domain is expressed in HEK293; and (2) the sequence that acts as a RORE resides in the bmal1a promoter domain of transfected bmal1a-luc. Thus, RORd1 can enhance bmal1a promoter activity. Given the periodic expression of RORd1, this receptor could therefore be involved in the regulation of circadian clocks, at least in the eyes. The finding that the mRNA expression pattern of medaka bmal1a is similar to that of rord1 in eyes supports this notion.
In the present study, we used a heterogeneous system consisting of a mammalian host cell and a Gal4-fusion receptor, which we believe properly represents the transactivational ability of medaka RORd1. It has been suggested that mammalian RORs can function as constitutively active receptors [27], and indeed, constitutive luminescence was observed in the absence of any ligand in the present study. However, forcing the cell line to express the medaka RORd1 resulted in threefold increases in luciferase activity in the presence of a suitable ligand – therefore, RORd1 is likely to act as a bona fide nuclear receptor.
Natural cholesterol-related compounds, which are known ligands for RORs in mammals [27], had varying effects on activity of the medaka RORd1. Among the cholesterol derivatives, 20α-OHC displayed the highest agonist activity for medaka RORd1 in this system. However, the potency of 20α-OHC for medaka RORd1 (EC50; > 1 μM) was substantially lower than that for human RORC (EC50; 20-40 nM) [5]. Interestingly, 7α-hydroxycholesterol, 7β-hydroxycholesterol and 7-ketocholesterol all showed agonistic activity for medaka RORd1. These findings contrast with their natural inverse agonistic effects on RORa and RORc in mammals, as assessed by decreased transactivation activity [28]. This prompts us to suggest a physiological function for RORd1 in medaka that is distinct from the ROR-mediated role in glucose metabolism in response to binding 7-oxygenated cholesterol in mammals; ROR subtypes other than RORd may perform this function in medaka.
Putative natural agonists for medaka RORd1 were identified in this study as retinoids, rather than cholesterols. A previous study documented that ATRA induced the expression of a Purkinje cell-specific gene ‘synergistically’ with RORa via RAR/RXR [29]. ATRA was further reported as an ‘antagonist’ for RORb [7], in contrast with our results which, for the first time, assign an agonistic effect to a retinoid for ROR. Among all-trans retinoids, the more oxidized it was, the higher its efficacy, reflected by ATRA > retinal > retinol. Meanwhile, the stereoisomers 9-cis retinoic acid and 13-cis retinoic acid affected RORd1 in comparable manner to ATRA, but induced higher reporter activity than did 20α-OHC.
We were particularly interested in the ligand activity of 11-cis-retinoid species, as these are generated only in the retina, which is relevant, as rord1 expression was highest in the eyes. However, 11-cis-isomers were no more effective in elevating reporter activity than ATRA. Admittedly, ATRA yielded high transactivation activity mediated by RORd1 but its effective concentration was not calculable (EC50 value >1 μM). Therefore, ATRA is too weak to act as a ligand for RORd1 in vivo. In contrast, 4OH-ATRA and 4 K–ATRA yielded EC50 values in the sub-μM range, while that of 5,6E–ATRA was not calculable. It is widely held that the predominant route of elimination of ATRA is through its oxidation in the 4-position of the β-ionone ring to generate 4OH-RA [30, 31]. Here, we show that ATRA metabolites may serve as bona fide ligands for RORd1, reflecting their high potency for this receptor.
Retinoic acid plays an important role as a morphogen during embryonic development [32]. However, catabolism of retinoic acid is also essential. The concentration of retinoic acid is regulated by Cyp26, a hydroxylase that degrades ATRA into 4OH-ATRA [33,34,35], highlighting the importance of this enzyme for proper development [36]. A recent study revealed the existence of a retinoic acid concentration gradient in the zebrafish embryo by 14 h after fertilization; retinoic acid levels were low in the head and tail, which was due to local expression of cyp26s (cyp26a1, cyp26b1 and cyp26c1) responsible for retinoic acid conversion into its metabolites [37]. In one previous study, rora, rorb and rorc message RNAs were detected in the head and other areas of the zebrafish embryo [19]; these ROR subtypes conceivably receive ATRA metabolites, as shown for medaka RORd1. It appears likely that RORs function as a key regulator in a positive feedback loop resulting in the rapid degradation of ATRA, as RORs serve as sensors for ATRA metabolites and upregulate the expression of these enzymes for catabolism. Further studies of the distribution of RORd1 in the medaka embryo are necessary to define the role of this receptor during development. Aside from its role in development, RORd1 may also be required for hematopoiesis or immunity, as our results indicate that this subtype is expressed at around 107 copies/μg-RNA in a suite of tissues in adult medaka. Our finding, obtained by qPCR, that cyp26a1 mRNA levels increase at nearly the same time with rord1 expression suggests that medaka RORd1 may primarily regulate the transactivation of cyp26a1 gene, and consequently induce ATRA metabolism, at least in the eye, at ZT7. The results of our luciferase reporter assay using cyp26a1p-luc support this possibility.
A promoter assay using cyp26a1p-luc demonstrated that one of the four ROR subtypes, RORd1, is activated by 4-oxygenated ATRA’s metabolites. It is unclear whether these metabolites also act as agonists for other ROR subtypes, prompting the need for further study to compare and define ligand selectivity and physiological function. In any case, a suit of endogenous compounds, including 17-hydroxyprogesterone, melatonin, estradiol, androstenedione, and T3, stimulated or inhibited the transcriptional activity in HEK293 cells expressing Gal4-RORd1 LBD. Despite modulation of transcriptional activity, however, we suggest that these compounds do not normally interact with RORd1 in vivo, given the significantly lower ED50 for RORd1 compared to that of ATRA metabolites.