Regulation of FGF21 Expression and Secretion by Retinoic Acid Receptor-related Orphan Receptor α*

Fibroblast growth factor 21 (FGF21) is a hormone produced by fat and the liver that plays an important role in lipid metabolism. FGF21 expression is induced by peroxisome proliferator-actived receptor α in response to physiological conditions requiring increased fatty acid oxidation. Retinoic acid receptor-related receptor α (RORα) is another nuclear receptor that plays a critical role in lipid metabolism as well as in regulation of the circadian rhythm. In this study we demonstrate that RORα directly regulates the expression and secretion of FGF21. A canonical ROR response element was identified in the proximal promoter of the FGF21 gene and shown to exhibit functional activity. Overexpression of RORα in HepG2 cells resulted in increased expression and secretion of FGF21. Suppression of RORα expression caused a decrease in FGF21 expression and secretion, suggesting that RORα contributes to the basal expression of FGF21. These data suggest that one mechanism by which RORα regulates lipid metabolism may be by modulation of FGF21 secretion. Furthermore, this study identifies a clear link between RORα, a key regulator of the mammalian clock, and FGF21, an important hormone regulating glucose and lipid homeostasis.

Fibroblast growth factor 21 (FGF21) is a hormone produced by fat and the liver that plays an important role in lipid metabolism. FGF21 expression is induced by peroxisome proliferatoractived receptor ␣ in response to physiological conditions requiring increased fatty acid oxidation. Retinoic acid receptorrelated receptor ␣ (ROR␣) is another nuclear receptor that plays a critical role in lipid metabolism as well as in regulation of the circadian rhythm. In this study we demonstrate that ROR␣ directly regulates the expression and secretion of FGF21. A canonical ROR response element was identified in the proximal promoter of the FGF21 gene and shown to exhibit functional activity. Overexpression of ROR␣ in HepG2 cells resulted in increased expression and secretion of FGF21. Suppression of ROR␣ expression caused a decrease in FGF21 expression and secretion, suggesting that ROR␣ contributes to the basal expression of FGF21. These data suggest that one mechanism by which ROR␣ regulates lipid metabolism may be by modulation of FGF21 secretion. Furthermore, this study identifies a clear link between ROR␣, a key regulator of the mammalian clock, and FGF21, an important hormone regulating glucose and lipid homeostasis.
The regulation of lipid metabolism is tightly controlled and several nuclear hormone receptors (NHRs) 2 play an essential role in maintaining metabolic homeostasis. The retinoic acid receptor-related orphan receptors ROR␣ (NR1F1) and ROR␥ (NR1F3) have been implicated in metabolic pathways, energy homeostasis, and thymopoiesis. In addition, RORs regulate the expression of several components of the circadian clock and the expression of several downstream genes involved in metabolism.
Like all NHRs, the RORs display a typical nuclear receptor domain structure consisting of four major functional domains: an amino-terminal (A/B) domain followed by a highly conserved DNA-binding domain, a hinge domain, and a carboxylterminal ligand-binding domain. RORs regulate gene expression by binding as monomers to specific "half-site" DNA response elements (ROREs) consisting of an AGGTCA sequence with a 5Ј AT-rich extension. RORs recruit coactivators resulting in constitutive activation of target gene transcription (1)(2)(3).
ROR␣ has been shown to regulate several genes involved in lipid metabolism. Detailed examination of staggerer mice (ROR␣ sg/sg ) in which there is a frameshift and a premature stop codon, rendering ROR␣ inactive, have revealed significant alterations in lipid metabolism including low levels of total plasma cholesterol, triglycerides, and apolipoprotein CIII (4). ROR␣ directly regulates the expression of genes for both apoA1 and apoCIII (5,6). Consistent with this phenotype, the staggerer mice are less susceptible to high fat diet-induced obesity and hepatic stetosis (7).
ROR␥-deficient mice exhibit normal levels of plasma cholesterol and triglycerides (8). However, when crossed with the staggerer mouse, the ROR␣/␥-deficient mice present with hypoglycemia, suggesting a role for these receptors in the maintenance of glucose metabolism (8). This study also demonstrated that ROR␣ and ROR␥ display significant redundancy in function, which is consistent with plasma glucose levels remaining unaffected unless both receptors were lost. Recently, a role for ROR␣ in glucose metabolism was established when Chopra et al. (9) found that loss of the NHR coactivator SRC-2 resulted in a murine phenotype similar to von Gierke disease which is associated with severe hypoglycemia and abnormal accumulation of glucose in the liver.
Fibroblast growth factor 21 (FGF21), a member of the FGF family, is a hepatic hormone that regulates peripheral glucose tolerance, torpor, and hepatic lipid metabolism (10 -12). In vivo administration of FGF21 in rodent models of diabetes lowers both plasma glucose and triglyceride levels and improves insulin sensitivity and glucose clearance (12). Transgenic mice overexpressing FGF21 in the liver display resistance to diet-induced obesity and have improved glycemic control (14). FGF21 administration in diabetic monkeys results in a similar phenotype, including improvements in lipoprotein profiles (13). The ability of FGF21 to improve and protect against metabolic disorders like diet-induced obesity makes it an attractive drug candidate for the treatment of obesity and other metabolic diseases (15)(16)(17). Here, we demonstrate that ROR␣ regulates both the expression and secretion of FGF21, providing another link between this NHR and lipid metabolism.

EXPERIMENTAL PROCEDURES
Plasmids and Viruses-The FGF21 promoter (Ϫ1738 to ϩ71) was amplified from genomic DNA of HepG2 cells and cloned into pTAL-Luc luciferase report vector (Clontech, CA). The FGF21 promoter mutant construct was made by using QuikChange II site-directed mutagenesis kit (Stratagene) according to the manufacturer's instructions. The RORE (Ϫ1564 to Ϫ1554) was muted from TTCTAGGTCA to TTCTAGGGGG. PGL4.73 reporters were from Promega. PTrex-ROR␣ and pTrex-ROR␥ were from Phenex Pharmaceuticals AG. ROR␣ was tagged with FLAG and subcloned into pAd/CMV/V5-DEST vector through Gateway TM technique (Invitrogen). The adenovirus with FLAG-ROR␣ was produced according to the manufacturer's instructions.
Site-directed Mutagenesis-The FGF21 promoter mutant construct was made using QuikChange II site-directed mutagenesis kit (Stratagene) according to the manufacturer's instructions. The RORE (Ϫ1564 to Ϫ1554) was mutated from TTCTAGGTCA to TTCTAGGGGG. The mutant primers targeting ROR binding site were: CACCTCTTGCCTTCTAG-GGGGCTTCTCACAATGTC (forward) and GACATTGTG-AGAAGCCCCCTAGAAGGCAAGAGGTG (reverse). The mutant primers were used to amplify mutant plasmid from pTAL-FGF21 reporter using PfuUltra HF DNA polymerase. The PCR product were treated with DpnI to select for mutation-containing synthesized DNA and then transformed into XL1-Blue supercompetent cells. Positive clones were picked and grown overnight in LB media. The plasmid were isolated using QIAprep spin miniprep kit (Qiagen). The mutant construct was verified by sequencing.
Cell Culture and Luciferase Assay-HEK293 cells were maintained in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum at 37°C under 5% CO 2 . 24 h prior to transfection, HEK293 cells were plated in 96-well plates at a density of 15 ϫ 103 cells/well. Transfections were performed using Lipofectamine TM 2000 (Invitrogen). 8 h post-transfection, the cells were treated with vehicle or ligands. 24 h posttreatment, the luciferase activity was measured using the Dual-Glo TM luciferase assay system (Promega).
Overexpression and Knockdown-The HepG2 cells were maintained in minimum essential medium supplemented with 10% fetal bovine serum at 37°C under 5% CO 2 . HepG2 cells were plated in 6-well plate 1 day before infection. The cells were infected with adenovirus for 24 h and then switched to regular growth media. 24 h later, the media were collected for ELISA, and the cells were harvested to isolate total RNA. For knockdown assay, the control siRNA, human ROR␣ siRNA (Thermo Scientific) were transfected with Lipofectamine TM RNAiMAX (Invitrogen) by using reverse transfection. After 24 h, the cells FIGURE 1. The FGF21 promoter contains an evolutionarily conserved RORE. A, schematic of the FGF21 promoter. The FGF21 promoter contains a RORE in its proximal promoter region and the two recently identified REV-ERB response elements (REVRE) in its distal region. B, the RORE is evolutionarily conserved between humans and mice. C, sequence of the human FGF21 promoter region. The first underlined sequence depicts the RORE that we have identified. The second and third underlined sequences are the two REV-ERB response elements that were recently identified by Estall et al. (24).
were harvested to perform quantitative PCR assay or Western blot. The media were subjected to ELISA.
Chromatin Immunoprecipitation (ChIP) Assay-ChIP assays were performed according to the manufacturer's instructions (Millipore) and as we have described previously (18 -21). HepG2 cells were plated in 100-mm plates 24 h before infection. The cells were infected with adenovirus (AD-lacZ or AD-ROR␣) for 24 h and then switched to regular growth media. After overnight incubation, the cells were cross-linked for 10 min with 1% formaldehyde and stopped with 125 mM glycine for 5 min at room temperature. The cells were washed twice with ice-cold phosphate-buffered saline and then harvested in 1 ml of phosphate-buffered saline with protease inhibitor. Cell pellets were obtained by centrifugation at 700 ϫ g at 4°C and resuspended in SDS lysis buffer (1% SDS, 10 mM EDTA, 50 mM Tris, pH 8.1) with protease inhibitor. Cross-linked chromatins were sheered to fragments in length between 200 and 1000 base pairs by sonication. The cell lysates were centrifuged at 12,000 ϫ g at 4°C for 10 min to remove insoluble material. Equal amounts of chromatin were diluted 10-fold in dilution buffer (0.01% SDS, 1.1% Triton X-100, 1.2 mM EDTA, 16.7 mM Tris-HCl, pH 8.1, 167 mM NaCl.) and precleared with protein G-agarose for 1 h at 4°C. Immunoprecipitation with the follow-ing antibodies was performed at 4°C overnight: Anti-mouse IgG (Millipore), anti-␣-acetyl histone H3 (Millipore), anti-FLAG (Sigma). Antibody-chromatin complexes were collected with protein G-agarose beads followed by sequential washes with low salt, high salt, lithium, and Tris-EDTA buffers. Chromatins were eluted and reverse cross-linked for 4 h at 65°C in the presence of NaCl. The proteins were digested with proteinase K, and the DNA fragments were column-purified. FGF21 ChIP primers used in PCR were GTAACGGGTGCCTT-CCCAGA (forward) and ACCAGGAAACAACCGGTGG (reverse).
ELISA-HepG2 cells were plated in 6-well plate 1 day before infection. The cells were infected with adenovirus for 24 h and then switched to regular growth media. 24 h later, the media were used to perform FGF21 ELISA according to the manufacturer's instructions (Thermo Scientific).
Western Blot-Control siRNA or human ROR␣ siRNA (Thermo Scientific) were transfected into HepG2 cells using Lipofectamine RNAiMAX (Invitrogen). After 24 h, cell medium was switched to minimum essential medium with 10% fetal bovine serum. 48 h post-transfection, HepG2 cells were washed once with phosphate-buffered saline and then incubated for 10 min at 4°C in 100 l of TNT lysis buffer (50 mM Tris-Cl, pH 7.5, 150 mM NaCl, and 1% Triton X-100) and a complete miniprotease inhibitor mixture (Roche Applied Science). The samples were then scraped and harvested into 1.5-ml microcentrifuge tubes, vortexed for 30 s, and then centrifuged (425 ϫ g for 10 min). Protein levels in the supernatants were determined using a Coomassie protein assay kit (Bio-Rad), and 20 g of protein from each sample was separated by SDS-PAGE (Bio-Rad; 10%) and then transferred to a polyvinylidene difluoride membrane (Millipore, Milford, MA) and immunoblotted with primary antibodies: ROR␣ (BioLegend) or ␣-tubulin (Sigma) and horseradish peroxidase-conjugated secondary antibodies (Jackson Immunoresearch). Detection of the bound antibody by enhanced chemiluminescence was performed according to the manufacturer's instructions (Santa Cruz).
Statistical Analysis-All of the individual cotransfection experiments were performed in quadruplicate wells, and each experiment was performed at least three times. Results from the cotransfection experiments were normalized and combined for presentation, and the mean Ϯ S.E. is indicated. The FGF21 secretion experiments were performed three times, and statistical analysis was performed on the combined data. Statistical analysis was performed using GraphPad Prism and Student's t test. The differences were considered statistically significant if p Ͻ 0.05. Representative blots/gels are shown for the ChIP experiment and Western blot experiments.

Identification of a Putative RORE in the FGF21 Promoter-
We analyzed the promoters of several genes involved in lipid metabolism for the presence of putative ROREs that are conserved between humans and mice using MAPPER (22,23). A site within the FGF21 promoter was identified (Fig. 1A) that displayed characteristics of a classical RORE with an AGGTCA "half-site" and an AT-rich 5Ј sequence (Fig. 1B). The position of the site was from Ϫ1564 to Ϫ1554 within the promoter 1359-bp upstream of the putative REV-ERB response element recently suggested to mediate REV-ERB responsiveness to the FGF21 gene (Fig. 1C) (24).
Characterization of a Functional RORE within the FGF21 Promoter-We cloned the FGF21 promoter (Ϫ1738 to ϩ71) into the pTAL-Luc luciferase reporter vector and assessed the ability of RORs to modulate FGF21 activity in a cotransfection assay. HEK293 cells were transfected with ROR␣ along with the FGF21 reporter construct. Fig. 2A demonstrates that ROR␣ can induce the transcription of a FGF21 promoter-driven reporter. Inclusion of ROR␣ resulted in an approximate 3-fold increase in FGF21 promoter-driven transcription. ROR␥ is structurally similar to ROR␣ and displays an overlapping pattern of expression with ROR␣, and these two receptors cooperate to regulate diverse physiological functions including glucose metabolism and Th17 cell development. We hypothesized that ROR␥ would also induce the expression of FGF21 similar to ROR␣. HEK293 cells were transfected with a FGF21 reporter construct and ROR␥. Fig. 2A demonstrates that, like ROR␣, ROR␥ can induce the expression of FGF21. To confirm that this event is ROR-specific, we mutated the RORE in the FGF21 promoter (Fig. 2B). This mutation rendered the FGF21 promoter unresponsive to both ROR␣ and ROR␥, demonstrating the critical role for this sequence in mediating ROR responsiveness (Fig. 2B).
ROR␣ Overexpression Increases FGF21 Expression-To investigate the potential role of ROR␣ in FGF21 regulation, we overexpressed adenoviral ROR␣ in HepG2 cells and measured the expression of FGF21 by reverse transcription-PCR. As seen in Fig. 3A, the addition of ROR␣ dramatically increased the expression of FGF21 compared with control adenovirus (ϳ5.5-fold). We examined whether ROR␣ was directly modulating FGF21 expression via binding to the RORE in its promoter sequence by ChIP in these HepG2 Fig. 3B, we detect significant FGF21 promoter occupancy by ROR␣.

cells. As illustrated in
ROR␣ Overexpression Stimulates FGF21 Secretion-HepG2 cells have been used as a model to examine the effects of fatty acids on FGF21 secretion (25); thus we used this model to determine whether overexpression of ROR␣ would result in an alteration of FGF21 secretion. ROR␣ was overexpressed in these cells using the ROR␣ adenovirus or control virus, and after 24 h the amount of FGF21 in the medium was measured using an ELISA. As illustrated in Fig. 3C, overexpression of ROR␣ led to a 10-fold increase in secreted FGF-21.
Knockdown of ROR␣ Results in Decreased FGF21 Expression-To further characterize the physiological role of ROR␣ on FGF21 expression, HepG2 cells were treated with either control siRNA or ROR␣ siRNA. As depicted in Fig. 4A, Western blot analysis confirmed that the ROR␣ siRNA significantly decreased ROR␣ protein expression. Furthermore, Fig. 4B demonstrates that a loss of ROR␣ expression leads to a signifi- cant 60% decrease in FGF21 gene expression as measured by reverse transcription-PCR. Thus, ROR␣ contributes to the basal expression of FGF21.
A ROR␣/␥ Inverse Agonist Represses FGF21 Promoter-driven Transcription-We recently published that the transcriptional activity of ROR␣ and ROR␥ is modulated by 7-oxygenated sterols. Specifically, we demonstrated that 7␣-hydroxycholestrol (7␣-OHC) acts as a ROR␣/␥ inverse agonist. Furthermore, 7␣-OHC-modulated expression of ROR target genes and suppressed glucose output from primary hepatocytes (21). Due to our recent findings, we hypothesized that 7␣-OHC-mediated inhibition of ROR␣ and ROR␥ transcriptional activity would in turn suppress FGF21 expression. HEK293 cells were transfected with the FGF21 promoter luciferase reporter construct and expression vectors for either ROR␣ or ROR␥. 24 h post-transfection, the cells were treated for 24 h with 7␣-OHC or vehicle control. Fig. 5 illustrates the ability of 7␣-OHC to suppress transcription driven by the FGF21 promoter presumably via repression of ROR␣ and ROR␥ activity. FGF21 and ROR␣ have been demonstrated to play an important role in lipid metabolism. Both are highly expressed in the liver, and mice with null mutations in either of these two genes have aberrant lipid metabolism (14,26). Here, we show that the FGF21 promoter contains a functional RORE and that ROR␣ contributes to the basal expression of FGF21. Furthermore, overexpression of ROR␣ leads to a significant increase in the secretion of FGF21. Another nuclear receptor subclass, the REV-ERBs (REV-ERB␣ (NR1D1) and REV-ERB␤ (NR1D2)) also recognize RORE sites and are often coexpressed with the RORs (27). As opposed to the RORs, the REV-ERBs are repressors of transcription. Until recently, the REV-ERBs were thought to be constitutive repressors of transcription. However, work from our lab as well as that from the Lazar lab has demonstrated that the porphyrin heme functions as a ligand for both REV-ERBs and is required for repression of transcription (18,28). RORs and REV-ERBs share a number of target genes, indicating that they coordinately regulate physiological processes. Interestingly, a very recent study demonstrated that PGC1␣ regulates hepatic FGF21 expression by modulation of the heme/REV-ERB␣ axis (24). PGC-1␣ modulates the levels of the REV-ERB␣ ligand, heme, by activation of transcription of the rate-limiting enzyme in heme synthesis, aminolevulinic acid synthase. Elevated levels of heme lead to increased REV-ERB␣ repressor activity directly at the FGF21 promoter. The putative REV-ERB response element (REVRE) identified by this group is a site distinct from the site we identified in this study. Our mutation studies indicate that the ROR␣ activity is mediated entirely by the upstream site. Although RORs and REV-ERBs typically bind to similar sites, it is conceivable that they recognize different sites on the FGF21 promoter. Ultimately, these data still indicate that ROR and REV-ERBs coordinately regulate the expression of FGF21. Furthermore, these data suggest that the heme (REV-ERB ligand) and oxysterol (ROR␣ ligand) synthetic pathways may regulate lipid and glucose metabolism via modulation of FGF21 secretion.

DISCUSSION
These two NHRs play an essential role in regulation of the mammalian clock that in turn coordinates various physiological processes (29 -34). Studies have suggested that FGF21 expression is regulated in a circadian manner (35), leading to the possibility that ROR␣ and REV-ERB␣ may be contributing to its circadian pattern of expression. The circadian rhythm is tightly coupled to metabolic regulation, and among several stimuli that can entrain the clock is food (36). This food-entrainable oscillator controls food-anticipatory behavior as well as other physiological alterations in anticipation of a meal.   levels are induced by fasting and suppressed by refeeding (10,11,37), and although another NHR, the peroxisome proliferator-activated receptor ␣, has been shown to play an essential role in regulation of FGF21 expression, especially as it relates to the fasting-induced expression of this hormone (10 -11, 35, 37-39), it is possible that there may also be a circadian regulatory component to FGF21 secretion mediated by ROR␣.
An aberrant circadian rhythm can also lead to development of metabolic syndrome. Clock mutant mice display hepatic steatosis, hyperlipidemia, high glucose levels, and low insulin levels (40). In humans, epidemiological data indicates that shift work is associated with increased risk for development of diabetes, obesity, and cardiovascular disease (41,42). A recent study in humans demonstrated that controlled misalignment of the circadian rhythm led to metabolic alterations, with some of the healthy subjects exhibiting postprandial glucose responses expected of individuals in a prediabetic state (43). Recent studies have shown administration of FGF21 to be beneficial in treating hyperglycemia and insulin resistance (16,44). Our recent work demonstrating that ROR␣ can be regulated by a synthetic ligand (20) suggests that one may be able to modulate FGF21 levels by targeting this NHR.