Pharmacological modulation of RORα controls fat browning, adaptive thermogenesis, and body weight in mice

Abstract

Beiging is an attractive therapeutic strategy to fight against obesity and its side metabolic complications. The loss of function of the nuclear transcription factor RORα has been related to a lean phenotype with higher thermogenesis in sg/sg mice lacking this protein. Here we show that pharmacological modulation of RORα activity exerts reciprocal and cell-autonomous effect on UCP1 expression ex vivo, in cellulo, and in vivo. The RORα inverse-agonist SR3335 upregulated UCP1 expression in brown and subcutaneous white adipose tissue (scWAT) explants of wild-type (WT) mice, whereas the RORα agonist SR1078 had the opposite effect. We confirmed the reciprocal action of these synthetic RORα ligands on gene expression, mitochondrial mass, and uncoupled oxygen consumption rate in cultured murine and human adipocytes. Time course analysis revealed stepwise variation in gene expression, first of TLE3, an inhibitor of the thermogenic program, followed by a reciprocal effect on PRDM16 and UCP1. Finally, RORα ligands were shown to be useful tools to modulate in vivo UCP1 expression in scWAT with associated changes in this fat depot mass. SR3335 and SR1078 provoked the opposite effects on the WT mice body weight, but without any effect on sg/sg mice. This slimming effect of SR3335 was related to an increased adaptive thermogenesis of the mice, as assessed by the rectal temperature of cold-stressed mice and induction of UCP1 in scWAT, as well as by indirect calorimetry in presence or not of a β3-adrenoceptor agonist. These data confirmed that RORα ligands could be useful tools to modulate thermogenesis and energy homeostasis.

NEW & NOTEWORTHY The regulation of adipose tissue browning was not fully deciphered and required further studies explaining how the regulation of this process may be of interest for tackling obesity and related metabolic disorders. Our data confirmed the involvement of the transcription factor RORα in the regulation of nonshivering thermogenesis, and importantly, revealed the possibility to in vivo modulate its activity by synthetic ligands with beneficial consequences on fat mass and body weight of the mice.

INTRODUCTION

Activating nonshivering thermogenesis in brown adipose tissue (BAT) and white adipose tissue (WAT) to enhance the rate of energy expenditure is a new pharmacological target to protect against obesity. BAT burns glucose and fatty acids, thanks to UCP1 which dissociates substrate oxidation from ATP production (1), whereas WAT specializes in energy storage and mobilization. Brown and white adipocytes share several transcription factors critical for the development and maintenance of mature fat cells (2, 3). However, the ways used to generate the different adipocyte subtypes are still not fully understood. In brown adipocytes, PRDM16 and PGC1α interact with PPARγ, the master regulator of adipocyte differentiation, to activate the thermogenic program (4). Nevertheless, in white adipocytes, specific transcriptional effectors of white adipocyte gene expression are scarce (5).

Moreover, the ability of some white mature adipocytes to adopt a thermogenic phenotype in response to cold has led to the discovery of “beige adipocytes,” molecularly distinct from the brown ones. These beige adipocytes can emerge either from the differentiation of few “beige” progenitors hidden among “white” precursors present within the WAT or from the direct conversion of existing white adipocytes into beige fat cells (transdifferentiation) (6, 7). This latter hypothesis suggests the presence of transcriptional suppressors of the thermogenic program in white adipocytes and such molecular brakes have been described, including Rb, RIP 140, TLE3, or ZFP423 (reviewed in 8).

Elucidating the molecular mechanisms that switch on the beige phenotype could identify new molecular targets to be addressed in the fight against obesity and its metabolic complications. Although many extracellular effectors able to modulate WAT browning have been already reported (9), new gene candidates can still emerge from the phenotype of mutated mice. Given that the RORα-deficient (sg/sg) mice exhibited higher rate of energy expenditure both in chow diet (10) and high-fat diet (11), a relationship between RORα and thermogenesis has been suspected. Muscat’s group was the first to establish an inverse correlation between RORα and UCP1 expression and to suggest that enhanced UCP1 and thermogenic activity may be an important contributor to the lean, obesity-resistant phenotype of sg/sg mice (12, 13). We further confirmed an increased “browning” of all the fat depots in these mice, thus conferring to RORα an inhibitory role on the thermogenic program that could be counteracted ex vivo by the use of a synthetic RORα inverse-agonist (14). Indeed, we showed that a short treatment with this specific RORα ligand, in the absence of other external stimulation, was sufficient to induce UCP1 protein appearance in WAT explants, within a timing compatible with a potential browning process by transdifferentiation or revealing the presence of preexisting beige adipocytes.

RORα, β, and γ constitute a subfamily of nuclear receptors involved in many physiological processes. RORα is the major form expressed in WAT and is increased during adipogenesis (15). The recent characterization of endogenous ligands for these former orphan receptors has stimulated the development of synthetic ligands by pharmaceutical companies, as targeting these receptors could represent new opportunities to treat several diseases (16). Several oxysterols have been found to function as RORα inverse-agonists, inhibiting G6PC (glucose-6-phosphatase) and Bmal expression (17). The Scripps Research Institute designed several RORα ligands with, for some of them, an in vivo efficiency: SR3335 is a selective RORα inverse-agonist and SR1078 is an RORα agonist able to suppress and increase some hepatic target genes in vivo, respectively (18, 19).

The aim of this work was 1) to investigate the possibility of a pharmacologically-induced inverse relationship between RORα activity and UCP1 expression in adipose tissue explants by using either an RORα inverse-agonist or an RORα agonist; 2) to more precisely delineate the time course of the molecular events produced by these RORα ligands in murine adipocyte cell lines and in human adipocytes; and 3) to test the ability of in vivo administration of these RORα ligands on nonshivering thermogenesis and body weight in mice.

MATERIAL AND METHODS

Animals and Treatment

sg/sg mice (a spontaneous mutation in C57BL6/strain) were obtained by crossing heterozygous sg/+ mice, given by Pr. J. Mariani’s laboratory (IBPS, Paris, France), and compared with +/+ littermates. All animal care and use procedures were in accordance with the guidelines of the Charles Darwin Ethics Committee (Ce5/26092 and 26094). Mice were housed in a specific pathogen-free environment in a temperature-controlled room maintained at 24°C, with a 12 h/12 h light-dark cycle (lights on from 8 h–20 h). Water and food (A03, UAR, Epinay-sur-Orge, France) were provided ad libitum.

Given that age and body weight influence thermogenesis (12, 20) and that sg/sg mice are leaner than wild-type (WT), especially during the youth (10), the mice selected for this study were of a certain age (between 6 to 18 mo-old) to minimize their weight disparities. Nevertheless, there was still a 10%–14% difference in body weight between WT and sg/sg mice, even though the latter were 6 mo older than their WT counterparts. For in vivo studies, mice were intraperitoneally injected with SR3335 or SR1078 (15 mg/kg) or with the excipient [10% dimethyl sulfoxide (DMSO), and 10% Tween 80 solution], single-caged, then shifted to 6°C for 3 h for cold experiment. Body temperature was measured with a rectal probe (Physitemp, RET3, Clifton, NJ) and a reader (Physitemp, BAT-12, Clifton, NJ) every hour during the test. Mice were sacrificed by cervical dislocation. Perigonadal and inguinal (white) and interscapular (brown) adipose tissues were cautiously dissected, weighted, snap-frozen in liquid nitrogen and stored at −80°C before RNA extraction or protein homogenate preparation. For ex vivo studies, adipose tissues were minced into small pieces (5–10 mg) and incubated in DMEM containing 10 mM glucose, 0.5% bovine serum albumin (BSA), and 10 µM RORα ligands (or the excipient DMSO) for the times indicated in the figure legends.

Metabolic rate was measured by indirect calorimetry in TSE Phenomaster cages (Systems, Chesterfield, Missouri). Nineteen-week-old C57BL6/j mice were housed individually 1 wk before the experiment under a 12-h light/12 h-dark cycle at 22°C. Food (chow diet) and water were provided ad libitum. Six mice were injected daily with SR3335 (15 mg/kg) or with the excipient at the end of the light period (ZT11) for 4 days. On the morning of the fourth day (ZT5), each mouse was challenged with the β3-adrenoceptor agonist CL316,243 (1 mg/kg) to explore adaptive thermogenesis. Body composition was measured by nuclear magnetic resonance (NMR) at the beginning and the end of the experiment. Measurements in the metabolic chambers were made from D1 to D5.

Cell Culture

Murine 3T3-F442A adipocytes were differentiated as previously described (21). Human adipocytes have been obtained from adipose-derived stem cell (hASC) isolated from the abdominal subcutaneous adipose tissue of three healthy and premenopausal women [body mass index (BMI) < 25 kg/m²) in a context of plastic surgery and with the free and informed consent of the donors, in accordance with the ethical standards of the local ethical committee and the Declaration of Helsinki (1964). The study was approved by the French regulatory authorities (CPP Ile de France V, n° 14964). hASCs were isolated according to Zhu et al. (22) Briefly, tissue was digested with 0.2% collagenase (Sigma, Darmstadt, Germany). After centrifugation, the stroma vascular fraction was filtered on 100 µM pores, rinsed and seeded. Cells were cultured in αMEM supplemented with 10% fetal calf serum (FCS), 2 mM glutamine, 1% penicillin-streptomycin, and 145 nM basic fibroblast growth factor (Pepro Tech, Neuilly sur Seine, France). Subconfluent hASC cells were trypsinized and seeded at 10,000 cells/cm². The white differentiation process used was: 2 days post confluence, adipogenesis was initiated by culturing hASCs in DMEM with 10% FCS, 1 µM rosiglitazone, 1 µM dexamethasone, 500 µM 3-isobutyl-1methylxanthine, 1 µM insulin during 5 days, and then with 1 µM insulin, and 1 µM rosiglitazone for the following 9 days. The impact of repeated doses of RORα ligands during adipogenesis was tested at least twice with four points in each kind of adipocytes. Treatment with 10 µM RORα ligands in mature adipocytes was performed in DMEM containing 10 mM glucose and 0.5% BSA; three different experiments including duplicate points have been performed for each RORα ligand.

Confocal Immunofluorescence Microscopy

Adipocytes were grown and differentiated on 13-mm glass coverslips. They were fixed and permeabilized in cold methanol for 10 min at −20°C. Primary and secondary antibodies were applied to coverslips in PBS with 1% BSA for 15–30 min at room temperature. Confocal microscopy was performed using a Leica TCS-SP microscope (Lasertechnik, Berlin, Germany) equipped with a ×63 objective. Double fluorescence images were acquired in a sequential mode. Serial optical sections of 1 μm were taken. Selected paired sections were then processed to produce single-composite color-merged overlay images using Adobe Photoshop software (version 5.5). The antibody against PLIN1 (ab3526) was from Abcam, Paris, France.

RNA Extraction, cDNA Synthesis, and Quantitative PCR

Adipocyte and adipose tissue mRNA were extracted using the RNeasy Lipid Tissue Mini Kit (Qiagen, Courtaboeuf, France), and then reverse-transcribed using the High-Capacity cDNA Reverse Transcription Kit (Applied Biosystems Carlsbad, CA). Quantitative PCR of the genes of interest was performed using SYBR Green and a Light Cycler 480 Real-Time PCR System (Roche Diagnostics, Meylan, France) and the specific primers listed as followed: for mUcp1, ACAGAAGGATTGCCGAAAC (fwd) and AGCTGATTTGCCTCTGAATG (rev); for mPRDM16, ATGTGCTTAATTCCACCTTA (fwd) and GGAGAGGAGTGTCTTCAGAG (rev); for mTLE3, TGGTGAGCTTTGGAGCTGTT (fwd) and CGGTTTCCCTCCAGGAAT (rev); for mADRB3, ATCGTGTCCGCTGCCGT (fwd) and ATCTGCCCCTACACGCCAC (rev); for m36B4, GCTGATGGGCAAGAACACCA (fwd) and CCCAAAGCCTGGAAGAAGGA (rev); for mHPRT, AGGACCTCTCGAAGTGT (fwd) and TCAAATCCCTGAAGTACTCAT (rev); for PGC1α, CAACCGCAGTCGCAACAT (fwd) and TGGGAACCCTTGGGGTCA (rev); for mcox8b, GAACCATGAAGCCAACGACT (fwd) and GCGAAGTTCACAGTGGTTCC (rev); for mcox7a, CAGCGTCATGGTCAGTCTGT (fwd) and AGAAAACCGTGTGGCAGAGA (rev); for mCPT1b, ATCATGTATCGCCGCAAACT (fwd) and CCATCTGGTAGGAGCACATGG (rev); for mCS, GGACAATTTTCCAACCAATCTGC (fwd) and TCGGTTCATTCCCTCTGCATA (rev); for mERRα, GCAGGGCAGTGGGAAGCTA (fwd) and CCTCTTGAAGAAGGCTTTGCA (rev); for RIP140, TCAAAAGCCACACCATACCC (fwd) and TCGTCTCCACTGCTGTCATC (rev); for hCS, TAGTGCTTCCTCCACGAATTTG (fwd) and CCACCATACATCATGTCCACAG (rev); for hCOX2, TACGGCGGACTAATCTTCAA (fwd) and CCGGGAATTGCATCTGTTTT (rev).

Gene expression was normalized to 36B4 and HPRT, and data analysis was based on the ΔCCT method.

Protein Extraction and Western Blot Analysis

Frozen adipose tissue was homogenized on ice in tissue protein lysis buffer (Euromedex, Souffelweyersheim, France) and then centrifuged first at 10,000 g at 4°C for 8 min to remove lipids, and then at 20,000 g, at 4°C for 10 min to remove insoluble material. Supernatants were subjected to SDS-PAGE and Western blotted with antibodies against RORα (sc-28612), PPARγ (sc-7273), hUCP1 (sc-293418) (Santa-Cruz Biotechnology, Heidelberg, Germany), against PRDM16 (ab 106410), mUCP1 (ab 10938) (Abcam, Paris, France), against TLE3 (Proteintech 11372-1-AP, Manchester, UK), and against RIP140 (MABS 1917) and tubulin (T5168) from Sigma (Darmstadt, Germany). Bands were measured on unsaturated exposition by densitometry using the software “Image J” (NIH) and tubulin was used as loading control.

Mitochondrial Bioenergetics Analysis

Mitochondrial mass of the differentiated adipocytes were appreciated by the fluorescent labeling of the Mitotracker Red probe (M-7512, Molecular Probes, Eugene, OR). Measurement of oxygen consumption was performed using a Seahorse Bioscience XF96 Analyzer (Agilent, Santa Clara, CA), according to the manufacturer’s protocol. OCR measurement were obtained following sequential additions of oligomycin (10 µM), FCCP (6.3 µM), and antimycin A (AA)/rotenone (10 µM).

ROR Synthetic Ligands

SR3335 and SR1001 were purchased from Cayman (Interchim, Montigny le Bretonneux, France) and SR1078 from Calbiochem (Merck Chemicals, Nottingham, UK).

Statistical Analysis

Values are presented as means ± SE. Statistical analysis was performed by unpaired Student’s t test (Graph Pad Software, San Diego, CA). P < 0.05 was considered the limit for statistical significance.

RESULTS

UCP1 Expression is Modulated by RORα Inverse-Agonist or Agonist in scWAT and BAT Explants

A 16-h exposure of scWAT explants from WT mice to the RORα inverse-agonist SR3335 enabled a transient appearance of UCP1 protein (Fig. 1A); this effect was found to be specific of RORα because explants from RORα^−/− (sg/sg) mice, physiologically expressing UCP1, were unsensitive to SR3335 addition (histogram in Fig. 1A). SR3335 was also able to increase the UCP1 amount in the BAT of WT mice, but with a lower magnitude than in scWAT (×1.8 and ×3 in BAT and scWAT, respectively; Fig. 1B). To test if the reciprocity was true, BAT explants from both genotypes were treated with an RORα agonist, SR1078. Figure 1C confirms that this ligand-dependent increase of RORα activity led to a half-decrease of UCP1 amount in BAT. Expectedly, SR1078 had no effect on UCP1 expression in explants from sg/sg mice. Our data thus show that the amount of UCP1 in adipose tissue is ex vivo modulated by ligands of the nuclear receptor RORα in the absence of other culture manipulations.

Figure 1.

RORα inverse-agonist SR3335 and agonist SR1078 reciprocally modulated the amount of UCP1 protein in adipose explants. A: representative Western blot of UCP1 protein appearance in response to ex vivo treatment by SR3335 (10 µM) in scWAT explants from WT mice and averaged data obtained by 20 h SR3335 treatment in WT and sg/sg mice (side-histogram; n = 5, female and male). UCP1 expression follows circadian variation (14) that persists for a few days in explants. ZT = Zeitgeber time, ZT0 is the beginning of the light phase, and ZT12 is the beginning of the dark phase. The points 4 h and 8 h have been frozen at the same period than the controls (T0) at ZT10, and the points 16 h and 24 h have been frozen at the same period than the controls (T0) at ZT2. Effect of SR3335 (B) or SR1078 (C) ex vivo treatment on BAT explants from WT and sg/sg mice and averaged data after 20 h (side histogram; n = 5 female and male). Data are presented as means ± SE. *P < 0.05 versus excipient-treated WT explants as determined by unpaired two-tailed Student’s test. BAT, brown adipose tissue; scWAT, subcutaneous white adipose tissue; WT, wild type.

Beige Cells Can Arise from 3T3-F442A Adipocytes Treated With RORα Inverse-Agonists

Loss of RORα in mice promotes adipose tissue browning and its decreased activity in adipose explants leads to an increase of UCP1 expression (12, 14). This shows that RORα exerts a negative effect on UCP1 expression and suggests an indirect mechanism of action. Indeed, experiments of global run analysis have shown that, if the mechanism of gene induction by RORα was direct and transcriptional, the mechanism used by RORα to repress brown fat genes expression was indirect (23). We previously observed that in WAT of sg/sg mice, the higher expression level of brown genes was concomitant with the decreased level of two transcriptional suppressors of the thermogenic program, TLE3 and RIP 140 (14), making them putative direct target genes of RORα. We thus investigated whether and how RORα ligands were able to modulate the thermogenic pathway in a murine white adipocyte cell line, either in differentiating adipocytes or in mature adipocytes (Fig. 2).

Figure 2.

Beige cells can arise from 3T3-F442A adipocytes treated with RORα inverse-agonists: inverse effect by RORα agonist. A: representative Western blot of murine differentiating adipocytes treated by two different RORα inverse-agonists (SR3335 or SR1001) and by an RORα agonist (SR1078), from D4 to D14 post confluence, with isoproterenol being added 6 h before cells lysis; the sum of respective protein that accumulated along the adipogenic differentiation process was quantified versus that of tubulin in the side-histograms. B: representative Western blot of the time course analysis of a single dose of SR3335 (10 µM) on protein amount in mature 3T3-F442A adipocytes. C: time course analysis of one single dose of SR3335 (red) or SR1078 (green) (10 µM) on gene expression. D: estimation of the mitochondrial mass into the adipocytes [as measured by the mitotracker (MTK) dye expressed per ng DNA], either 17 h after a single dose of SR3335 or after several doses all along adipogenesis (D4–D14). E: relative gene expression 16 or 24 h after the RORα inverse-agonist (SR3335) or agonist (SR1078) addition on differentiated adipocytes. At least three different experiments were performed in duplicate. Data are presented as means ± SE. *P < 0.05, **P < 0.01 versus excipient-treated adipocytes as determined by unpaired two-tailed Student’s test. DMSO, dimethyl sulfoxide.

We first compared the effect of both kinds of RORα ligands (two inverse-agonists and one agonist) on the “white” or “brown” phenotype of 3T3-F442A cells treated during the course of adipose conversion from D4 to D14, and with isoproterenol for the past 6 h. Figure 2A shows representative Western blots and histograms averaging the respective protein amount accumulated into the adipocytes from D7 to D14. We first verified if these ligands were able to exert an effect on the RORα protein amount; indeed, ROR gene promoters contained several ROREs, suggesting that RORα could regulate its own transcription (24). Consistently, the content of RORα that accumulated during adipogenesis was found to be increased by the treatment with the RORα agonist SR1078 and decreased with the RORα inverse-agonist SR3335, whereas the other inverse-agonist SR1001 tested had no significant effect (Fig. 2A). The RORα inverse-agonist SR3335 was able to increase the UCP1 and PRDM16 protein amount and the mitochondrial mass of the adipocytes (Fig. 2D); in parallel, the number of multilocular lipid droplets into the adipocytes seemed to be enhanced (Supplemental Fig. S1A, red arrows; all Supplemental Figures are available at https://doi.org/10.6084/m9.figshare.12086457.v1) and the amount of RIP 140, a suppressor of the thermogenic program, to be decreased. At the opposite, the RORα agonist SR1078 decreased the UCP1 and PRDM16 protein amount, while increasing TLE3 amount, another suppressor of the thermogenic program. Finally, the RORα inverse-agonist SR1001 was not found to exert any significant variation on these proteins. Thus, our data show that the browning of murine adipocytes is modulated by the ligands of RORα, suggesting a putative role for this nuclear receptor in the maintenance of a “white” phenotype.

We next tested whether in fully mature 3T3-F442A adipocytes a single exposure to the RORα inverse-agonist SR3335 was also able to induce the onset of the thermogenic program. Figure 2B shows that SR3335 increased PRDM16 and UCP1 protein as soon as 7 h after its addition in the medium, with a maximal effect at 16 h and a return to almost initial levels after 24 h, in agreement with previous results in WAT explants (Fig. 1A). Concomitantly, we observed an increase of their mitochondrial mass (Fig. 2D), with some of these adipocytes already exhibiting a higher number of lipid droplets than in control (Supplemental Fig. S1A, red arrows). Time course analysis of the effect of SR 3335 on UCP1 and PRDM16 mRNA levels indicated that the increased gene expression preceded that of the related proteins (UCP1 and PRDM16 mRNA peaking at 9 h vs. proteins at around 16 h; Fig. 2C). Although no significant variation of the TLE3 repressor was detectable by Western blot (Fig. 2B), a transient decrease of TLE3 mRNA levels was observed 6 h after SR3335 treatment (Fig. 2C); reciprocally, the RORα agonist SR1078 was found to induce a transient peak in TLE3 mRNA levels after a 3-h exposure. The gene expression of the other suppressor of the thermogenic program, RIP 140, was rather similarly decreased by either the RORα agonist or inverse-agonist (Fig. 2C). Finally, we observed an increased gene expression of PGC1α, Cox8b, Cox7a, and CPT1b in F442A cells, suggesting an enhanced β-oxidation capacity in response to SR3335 (Fig. 2E). At the opposite, we noted a decrease of Cox7a gene expression by SR1078 in F442A cells (Fig. 2E).

These data show that a single dose of SR3335 was able to promote the browning of mature adipocytes. The time course of the events allows the hypothesis of an initial action of RORα on the level of TLE3 mRNA, rather than on the level of RIP140 mRNA, whose gene expression would be expected to be oppositely controlled by either the RORα agonist or inverse-agonist.

RORα Inhibition Induces the Emergence of Beige Characteristics in Human Adipocytes, Whereas RORα Activation Counteracts This Effect

To determine whether the regulation of the thermogenic program by RORα was also effective in human adipocytes, we analyzed the effect of RORα ligands on adipocytes derived from human adipose stem cells (hASCs) either by adding repeated doses all along the differentiation process or by using a single dose on mature adipocytes.

hASCs were differentiated in white adipogenic medium supplemented with 10 µM RORα ligands from D5 to D13 before being harvested at D14. Protein expression was monitored along the differentiation process and illustrated by a representative Western blot (Fig. 3A), and protein amount accumulated from D7 to D14 was compared with control cells in side-histograms. At D14, their mitochondrial mass was measured with the MTK dye by fluorimetry and by immunocytochemistry (Fig. 3B). The perilipin antibody was used to visualize the size of the lipid droplets (Supplemental Fig. S1B). Their oxygen consumption rate (OCR) was analyzed by the Seahorse technology. Data confirmed the results obtained in the murine 3T3-F442A cell line. When added during adipocyte differentiation, SR3335 enabled the emergence of a beige phenotype of human adipocytes with an increase of PRDM16 and UCP1 protein levels (Fig. 3A) and with the appearance of smaller multilocular lipid droplets as delineated by the perilipin immunostaining (Supplemental Fig. S1B). The mitochondrial mass was enhanced in parallel, as visualized by the Mitotracker staining (Fig. 3B), together with an increase of the uncoupled OCR (proton leak) and no variation of coupled respiration (ATP production; Fig. 3C). In addition, SR3335 decreased the amount of the TLE3 protein. Noteworthy, the RORα agonist SR1078 had the opposite effect on the TL3 protein (Fig. 3A) and on the mitochondrial mass (Fig. 3B). Finally, the amount of PPARγ was not affected by the RORα ligands (Fig. 3A).

Figure 3.

Effect of RORα inverse-agonist and agonist along the adipogenic process of human adipocyte stem cells (hASC). A: on the left, representative Western blot of human differentiating adipocytes treated by the RORα inverse-agonist (SR3335) or agonist (SR1078), from D5 to D14 post confluence; on the right, the sum of respective proteins that accumulated along the adipogenic differentiation process was quantified versus tubulin in hASC from two different donors. B: estimation of the mitochondrial mass into the adipocytes (as measured by the mitotracker dye expressed per ng DNA) and immunodetection of mitochondria by the mitotracker dye into the human mature adipocytes, pretreated or not with SR3335, and in presence or not of isoproterenol. C: the oxygen consumption rate (OCR) was measured by Seahorse analysis at D14 on mature adipocytes, pretreated or not with SR3335 or SR1078. On the left, representative Seahorse runs with and without isoproterenol 10 µM. On the right, basal and isoproterenol-stimulated uncoupled (proton leak) and coupled (ATP production) mitochondrial respiration under various RORα ligands treatment. Proton leak and ATP production were defined by the equations: OCRoligomycin −OCRAA/rotenone and OCRbasal − OCRoligomycin, respectively. At least three different experiments were performed in duplicate. Data are presented as means ± SE. *P < 0.05 and **P < 0.01 versus excipient-treated adipocytes as determined by unpaired two-tailed Student’s test. AA, antimycin A; DMSO, dimethyl sulfoxide.

Thereafter, we performed the time course analysis of a single dose of each RORα ligand on protein expression in differentiated human adipocytes (Fig. 4). Western blots from Fig. 4A illustrate representative data obtained with SR3335 or with SR1078, whereas kinetics shown in Fig. 4A averaged the data obtained by protein density analysis. RORα inverse-agonist SR3335 increased PRDM16 (from 3 h–9 h) and UCP1 (peaking at 16 h), while decreasing both TLE3 and RIP 140 (lowest level after a 9 h exposure; Fig. 4A). On the opposite, the TLE3 protein amount increased as soon as 3 h after the RORα agonist SR1078 addition, followed by the progressive decrease of PRDM16 and UCP1 (Fig. 4A). In the case of RIP140, we also observed an increase of its protein amount, but peaking only 16 h after the SR1078 addition. The expression of these two thermogenic repressors in response to the RORα ligands was thus examined at the mRNA levels in hASC (Fig. 4B); their respective kinetic profile looked like those obtained in 3T3-F442A adipocytes (Fig. 2C) and suggested an earlier increase of TLE3 transcription than of RIP 140 in response to the RORα agonist SR1078. Finally, we observed an increase of PGC1α, COX2, and citrate synthase (CS) mRNAs amount by SR3335 and not by SR1078, suggesting an enhanced β-oxidation capacity by SR3335. As observed in the murine preadipocyte cell line, these data confirmed that the ligand-dependent RORα inhibition enables the emergence of beige characteristics in human adipocytes as well, whereas RORα activation counteracts it. In hASC however, the two suppressors of thermogenesis TLE3 and RIP140 seemed to be involved in this regulation.

Figure 4.

RORα inhibition enables the emergence of beige characteristics in human differentiated adipocytes, whereas RORα activation counteracted it. A: on the left, representative Western blot of the time course effect of SR3335 or SR1078 on protein amount in differentiated human adipocytes. On the right, corresponding averaged protein amount; two or three different hASC donors have been used. B: time course analysis of one single dose of SR3335 (red) or SR1078 (green) (10 µM) on gene expression. At least three different experiments were performed in duplicate. Data are presented as means ± SE. *P < 0.05, **P < 0.01, and ***P < 0.005 versus excipient-treated adipocytes as determined by unpaired two-tailed Student’s test. hASC, human adipose stem cells.

In Vivo Pharmacological Modulation of UCP1 in scWAT by RORα Ligands Affects the mass of the fat Depot, Body Weight, and Energy Expenditure of the Mice

Because the two RORα ligands that we ex vivo used have been demonstrated to be active in vivo (18, 19), we then tested their ability to control UCP1 amount in mice adipose tissues after a single intraperitoneally injection of SR3335 or SR1078 (15 mg/kg). Given the average 16-h period necessary for UCP1 to rise in WAT explants and the short half-life of the compounds, we chose to test the effect of the ligands 18–20 h after injection.

The representative Western blot and the side-histograms (Fig. 5A) show that SR3335 induced a fourfold increase in UCP1 amount in scWAT of WT mice, rising an average level similar to that detected in sg/sg mice. Under the same conditions, no significant effect of SR3335 on UCP1 protein amount was observed in the BAT (Fig. 5B). By contrast, in vivo exposure to the RORα agonist exerted the opposite effect: SR1078 injection reduced by threefold the UCP1 amount in the scWAT of WT mice and by 1.3-fold in the BAT, but without any effect on sg/sg mice (Fig. 6, A and B). These data confirmed that RORα ligands could be useful tools to modulate UCP1 protein amount in adipose tissues in vivo.

Figure 5.

Injection of the RORα inverse-agonist (SR3335) enables the appearance of UCP1 in scWAT and decreases body weight and subcutaneous fat mass of mice. WT and sg/sg male mice, aged 6 and 12 mo respectively, have been intraperitoneally-injected with SR3335 (or DMSO) (15 mg/kg) 23 h before analysis. Data are presented as means ± SE. *P < 0.05 versus excipient-treated WT mice as determined by unpaired two-tailed Student’s test. On the left, representative Western blot showing UCP1 protein amount into the scWAT (A) and BAT (B) of the mice injected with SR3335 or DMSO; on the right, averaged data obtained with five mice of each genotype. C: Analysis of body weight variation of the mice expressed in % of initial weight (n = 5). D: Analysis of adipose tissue and liver masses expressed in % of body weight at the same period of the day (n = 5). BAT, brown adipose tissue; DMSO, dimethyl sulfoxide; scWAT, subcutaneous white adipose tissue; gWAT, gonadic adipose tissue; WT, wild type.

Figure 6.

Injection of the RORα agonist (SR1078) decreases UCP1 in the BAT and scWAT and increases body weight of mice. WT and sg/sg female mice, aged 12 and 18 mo respectively, have been intraperitoneally-injected with SR1078 (or DMSO) (15 mg/kg) 18 h before analysis. Data are presented as means ± SE. *P < 0.05, **P < 0.01, and ***P < 0.005 versus excipient-treated WT mice as determined by unpaired two-tailed Student’s test. On the right, representative Western blot showing UCP1 protein amount into the scWAT (A) and BAT (B) of the mice injected with SR1078 or DMSO; on the left, averaged data obtained with six mice of each genotype. C: analysis of body weight variation of the mice expressed in % of initial weight (n = 6). D: Analysis of adipose tissue and liver masses expressed in % of body weight at the same period of the day (n = 6). BAT, brown adipose tissue; DMSO, dimethyl sulfoxide; scWAT, subcutaneous white adipose tissue; gWAT, gonadic adipose tissue; WT, wild type.

We thus investigated whether these effects of RORα ligands on UCP1 expression have been associated with variations of whole body weight and/or adipose tissue depot mass of the mice. We found that SR3335 and SR1078 exerted the opposite effect on the mice body weight. A single injection of SR3335 was able to decrease their body weight (Fig. 5C), while that of SR1078 increased it (Fig. 6C), by ∼2.8% of their initial weight compared with control animals. Likewise, SR3335 decreased the scWAT relative mass (Fig. 5D) of the mice, whereas SR1078 induced a nonsignificant increase in the mass of this depot (Fig. 6D). The relative BAT and liver mass were not affected by the RORα ligands (Figs. 5D and 6D). Because RORα ligands had no effect on the body weight and fat mass in sg/sg mice, their specificity towards RORα was confirmed (Figs. 5 and 6).

To further functionally investigate the origin of the weight loss exerted by SR3335 on WT mice, we tested their energy expenditure by indirect calorimetry in response to the RORα agonist (Fig. 7). C57BL6/j mice were daily injected with SR3335 or with the excipient for 3 days. Figure 7A illustrated the pattern of data acquisition on day 3 and night 3, and Figure 7B showed the averaged data upon 48 h (days and nights 2 and 3). Data clearly showed that SR3335 was able to increase energy expenditure, O₂ consumption, and CO₂ production, especially during the night period, without affecting food and drink intake or the locomotor activity of the mice. Finally, given that the respiratory exchange rate (RER) was nightly decreased by SR3335, data suggests an increased lipid oxidation in this period of time at the expense of the other nutrients present in the chow diet.

Figure 7.

RORα inverse-agonist (SR3335) increases energy expenditure in C57BL6/j mice. Six 19- wk-old female mice were daily intraperitoneally injected with SR3335 (15 mg/kg) or the excipient at the end of the light period (ZT11) for 3 days. Energy expenditure, O₂ consumption and CO₂ production were measured by indirect calorimetry and related to the lean mass of the mice housed at 22°C. A: representative profile of data acquisition on day 3 and night 3. In red and black, mice were treated with SR3335 and the excipient, respectively. Black arrows indicate the time of SR3335 injection. B: average data obtained upon 48 h (days and nights 2 and 3) and presented per 12 h-period of daylight or night. Respiratory exchange ratio (RER) was calculated by the ratio VCo₂/Vo₂. Average food and water consumption were measured by differential weight-based sensors. Spontaneous locomotor activity was measured by infrared beam breaks. Data are presented as means ± SE. *P < 0.05, **P < 0.01 versus excipient-treated WT mice as determined by unpaired two-tailed Student’s test. DMSO, dimethyl sulfoxide.

Decreased Body Weight and Fat Mass Induced by the RORα Inverse-Agonist SR3335 is Associated with Increased Body Temperature and Browning of scWAT in Cold-Stressed Mice

To investigate if the SR3335-dependent increase of UCP1 in scWAT was functionally associated with a gain in adaptive thermogenesis, mice from both genotypes were treated or not for 24 h with a single dose of SR3335, and then challenged by a 6°C cold exposure for 3 h. Rectal temperature was measured every hour. Figure 8A confirmed that sg/sg mice maintain higher rectal temperature than WT mice during cold exposure, as previously shown by Lau et al. (12) Indeed, although WT mice lost 3.5°C after 2 h in the cold, sg/sg mice only lost 2°C. Strikingly, SR3335 injection partially prevented the decrease in the rectal temperature of WT mice that exhibited a rectal temperature pattern similar than sg/sg mice. The measurement of body weight before and after the cold challenge showed that the higher thermogenesis of sg/sg mice was associated with a higher weight loss (−7.5%) than that of WT mice (−5%). Remarkably, SR3335 injection allowed the WT mice to lose similar body weight than sg/sg mice (−8%) (Fig. 8B). As expected, SR3335 did not induce an additional weight loss in sg/sg mice (Fig. 8B).

Figure 8.

RORα inverse-agonist (SR3335) decreases body weight and subcutaneous fat by increasing body temperature and browning of scWAT in cold-stressed mice. WT and sg/sg male mice aged 10–18 mo have been intraperitoneally-injected with SR3335 (or DMSO) (15 mg/kg) 23 h before being cold-stressed for 3 h at 4°C in individual cages (n = 6 of each genotype and condition). Data are presented as means ± SE. *P < 0.05, **P < 0.01 versus excipient-treated WT mice as determined by unpaired two-tailed Student’s test. A: core body temperature appreciated by rectal measure every hour. B: body weight lost at the end of the cold exposure and expressed in % of initial weight. C: relative fat and liver masses of the mice at the end of the experiment. D: Western blot illustrating the UCP1 protein amount in the scWAT of some of the mice at the end of the cold exposure. E: analysis of some gene expression in the BAT of the mice at the end of the experiment. BAT, brown adipose tissue; DMSO, dimethyl sulfoxide; scWAT, subcutaneous white adipose tissue; WT, wild type.

Mice were sacrificed immediately after cold exposure and their fat depots collected. SR3335 injection in WT mice decreased scWAT relative mass by 30%, at a level comparable to the subcutaneous fat depot of the untreated sg/sg mice (Fig. 8C). The scWAT mass of sg/sg mice was unsensitive to SR3335 injection. Of note, the BAT and liver masses of the cold-stressed mice were not influenced by the genotype, or by the SR3335 treatment. The level of UCP1 amount in the scWAT depots of these mice fitted with their respective mass losses; it was increased by 1.7- ± 0.3-fold in the scWAT of all SR3335-treated versus vehicule-treated cold-stressed WT mice (Fig. 8D). Finally, the BAT of these mice was compared at the level of gene expression (Fig. 8E): UCP1 and ADBR3 mRNA levels were not different according to the genotype, and these only enhanced in WT mice in response to SR3335 and acute cold challenge.

Because the sg/sg mice used in our experiments were necessarily 6-mo older than WT mice to reduce their weight disparities, we investigated whether the failure of sg/sg response to RORα ligands was not because they are much older than WT mice. By examining the absolute weight loss in response to SR3335 injection in old age-matched (65-wk-old) mice from both genotypes after acute cold exposure (4 h at 4°C), we could observe that only WT mice lost more weight in response to SR3335 + cold versus cold alone (WT: −0.77 g vs. sg/sg −0.06 g). Cold-stressed sg/sg mice lost 1.22 ± 0.27 and 1.16 ± 0.24 g with and without SR 3335 injection, respectively, while WT mice lost 2.18 ± 0.34 and 1.41 ± 0.14 g with and without SR3335 pretreatment, respectively (P < 0.05; n = 4 for sg/sg and 5 for WT mice). These data thus confirm that aged WT mice still respond to SR3335, whereas aged sg/sg mice do not.

Taken together, these data show that during cold exposure, a single SR3335 injection enables WT mice to acquire the same performance than untreated sg/sg mice, and suggests a more efficient adaptive thermogenesis associated with an increased UCP1 expression and a specific loss of scWAT.

Browning Effect of RORα Inverse-Agonist SR3335 Affects the Nonshivering Thermogenesis in WT Mice

To unequivocally demonstrate that browning induced by pharmacological modulation of RORα affect nonshivering thermogenesis, we performed indirect calorimetry analysis in C57BL6/j mice in response to a β3-adrenoceptor agonist. Six mice having been pretreated in vivo by SR3335 or not for 3 days were challenged by a CL316,243 injection on D4. Figure 9, A and B, compared the data 6 h after the injection versus the same period of time on D3 and clearly showed that CL316,243 increased the energy expenditure and decreased the RER more importantly in mice pretreated with SR3335 than in untreated mice. In parallel, their body temperature was increased by 0.3°C (Fig. 9C) and at the end of the experiment, these mice have lost double of the fat mass loss owing to the CL316,243 injection alone (Fig. 9D).

Figure 9.

RORα inverse-agonist (SR3335) increases nonshivering thermogenesis in C57BL6/j mice in response to a β3-adrenoceptor agonist. Six 19-wk-old female mice were daily intraperitoneally injected with SR3335 (15 mg/kg) or the excipient for 4 days (at ZT11), and challenged with a dose of a β3-adrenoreceptor agonist CL316,243 (1 mg/kg) on the day 4 at ZT5. Energy expenditure, O₂ consumption and CO₂ production were measured by indirect calorimetry and related to the lean mass of the mice housed at 22°C. Energy expenditure (A) and RER (B) were measured upon the 6 h-period of time following CL316,243 injection; +CL referred to data measured on day 4 (between ZT5 and ZT11) and were compared with the data obtained on the corresponding period of time on day 3 (called –CL). In red and white, mice were treated with SR3335 or the excipient, respectively. C: the change in mice body temperature exerted by CL316,243 administration was the difference between rectal temperature measured on D4 and D3 evenings, thus after and before CL316,243 injection. D: referred to fat mass loss in SR3335 and DMSO-treated mice at the end of the experiment, including the 24 h-period of time after the CL316,243 injection. We compared the fat mass (measured by NMR) of each mouse between the end (D5) and the beginning (D1) of the indirect calorimetry experiment. Data are presented as means ± SE. *P < 0.05 referred to SR3335 effect (SR3335-treated vs. DMSO-treated mice, #P < 0.05 ##P < 0.01 referred to CL effect vs. corresponding pretreated mice as determined by unpaired two-tailed Student’s test). DMSO, dimethyl sulfoxide; NMR, nuclear magnetic resonance; RER, respiratory exchange ratio; WAT, white adipose tissue; WT, wild type.

Taken together, these data confirm that as the loss of RORα function in sg/sg mice, pharmacological antagonism of RORα by an inverse-agonist allows the browning of the scWAT depot of WT mice and potentiates the decrease in the mass of this fat pad subsequently to the activation of nonshivering thermogenesis that improves lipid oxidation.

DISCUSSION

The regulation of adipose tissue browning was not fully deciphered and required further studies explaining how the regulation of this process may be of interest for tackling obesity and related metabolic disorders. Our data confirmed the involvement of the transcription factor RORα in the regulation of nonshivering thermogenesis and importantly revealed the possibility to in vivo modulate its activity by synthetic ligands with beneficial consequences on fat mass and body weight of the mice.

The first evidence for the association between RORα signaling and UCP1 mRNA expression was provided by Muscat’s team. They observed a decrease in the expression of UCP1 mRNA in the BAT of an adipocyte-specific RORα gain of function mouse model (13), whereas on the opposite, an increased UCP1 expression was found in the BAT and WAT of the RORα-deficient sg/sg model (12) that we further confirmed (14). These findings pointed to RORα as a new transcription factor negatively involved in UCP1 expression. We previously showed that a synthetic RORα inverse-agonist was a useful tool for increasing UCP1 protein amount and oxygen consumption in scWAT explants of WT mice, thus pharmacologically ex vivo mimicking the effect of the in vivo RORα deficiency in WAT (14). We confirmed here a similar regulation by the RORα inverse-agonist into BAT explants and also demonstrated that reciprocally, an RORα agonist was able to downregulate their UCP1 protein amount.

Moreover, we could observe an increase of the COX4 protein in BAT explants treated by SR3335 (by 2.4-fold, data not shown). From these experiments emerged the possibility to ex vivo modulate UCP1 expression (and perhaps mitochondrial function) into adipose tissues by the use of synthetic ligands of the nuclear factor RORα. Moreover, the action of these ligands appeared to be specific of RORα, as no effect was detectable on adipose explants isolated from sg/sg mice.

We then investigated if this “up and down” browning effect of RORα ligands was reproducible (and examinable) in cultured adipocytes. We observed for the first time that the ligand-dependent RORα modulation can control the emergence of beige characteristics, not only in murine but also in human adipocytes; the RORα inverse-agonist increasing thermogenic genes and mitochondrial biogenesis and function, and the RORα agonist exerting the opposite effect.

The first protocol we used was several repeated doses added every 2–3 days along the differentiation process in a way to “optimize” the expected answer. It took into account the kinetics of expression of RORα that appears only significant from D2 or D4 of differentiation (36, 27). Thus, this approach was performed in a way to prevent any interference with the initial differentiation cocktail used for hASC, for example. Our goal was not to decipher the origin of the beige cells, i.e., whether the RORα ligands interfered with the differentiation of specific precursors, but rather to determine whether modulating RORα activity allowed their switch in an active or inactive thermogenic state (26). Our demonstration of a browning effect obtained only few hours after the addition of a RORα inverse-agonist on mature adipocytes would fit with a transdifferentiating effect (6, 7). At the opposite, the fact that RORα activation by agonists was still able to rapidly decrease some residual brown gene expression in the human adipocytes was compatible with such hypothesis. However, one cannot exclude the possibility that RORα ligands acted on preexisting beige adipocytes into the 3T3-F442A cell line or hASC. In that field, molecules reported as affecting beige activation and differentiation have been most often tested ex vivo on cell systems of differentiation (and by addition as soon as the beginning of the differentiation process) (9). Few studies have been performed on mature adipocytes (27–29) or on explants (30).

Our data revealed that, in both murine and human mature adipocytes, treatment with the RORα inverse-agonist SR3335 resulted in a rapid increase in PRDM16 and UCP1 protein amount, reaching a peak after ∼16 h before decreasing at 24 h. The appearance of the UCP1 protein as soon as 6–7 h after the treatment corroborated the time course observed in primary mouse adipocytes exposed to 31°C (31). The maximal increase in their respective mRNA, peaking at 9 h before decreasing at 16 h, fits with the pattern of UCP1 transcript induction by transretinoids (32). The transient effect of SR3335 on UCP1 expression is coherent with the reported half-lives of UCP1 transcript and protein (5 h and 10 h, respectively) in brown adipocytes treated with the β3-adrenoceptor agonist CL316,243 (33). Our results suggest transcriptional events activated by SR3335 or repressed by SR1078. Given that RORα activation exerts a negative effect on UCP1, and because it has been reported that negative gene regulation by RORα tends to be indirect (23), we focused on the pattern of expression of two different suppressors of the thermogenic program (25, 34). We previously found these suppressors to be decreased in sg/sg mice (14), suggesting they could be putative targets for RORα. As expected, TLE3 expression was reciprocally regulated by the RORα agonist or inverse-agonist; TLE3 mRNA and protein peaking as soon as 3 h after the addition of the RORα agonist SR1078 in the medium, and inversely decreasing maximally 6 h (mRNA) and 9 h (protein) after the addition of the RORα inverse-agonist SR3335. In the case of RIP140, its transcript was also early decreased by the RORα inverse-agonist SR3335 in both kinds of adipocytes, in accordance with the recent observation of the presence of a RORE in the RIP140 promoter (35). Surprisingly, the RIP140 mRNA level was also decreased by the RORα agonist in both murine and human adipocytes, but increased at the protein level in hASC, suggesting a potential enhancement of the protein time-life by SR1078.

These data suggest that the suppressor of the thermogenic program TLE3 would better fit as a candidate target of RORα transcriptional activity to repress an acute control of beiging, considering the time course of the events and also its ability to inhibit isoproterenol-induced UCP1 expression (34). On the other hand, RIP140, being “up and down” transcribed by RORα and Reverbα, respectively (35) and independently of the β3-signaling pathway (25), it could be more particularly involved in the circadian control of thermogenesis.

A key mechanism of SR3335 action could be to release the brake of RORα on the thermogenic pathway. TLE3 has been described as a white-selective cofactor that acts reciprocally with the brown-selective cofactor PRDM16 to specify thermogenic gene program. Indeed, TLE3 is able to disrupt the physical interaction between PRDM16 and PPARγ (34). Consistently, TLE3 is less abundant in BAT than in WAT (Fig. 2A and 34). The same distribution pattern between fat depots was observed for RORα expression (Fig. 2A), allowing to hypothesize that these two proteins, whose expression rises during white adipogenesis (36–38), could act in concert to maintain a “white” adipocyte phenotype. Finally, the RORα ligands did not affect the expression of the master adipogenic gene PPARγ (Fig. 3A), and thus, these did not interfere with the global adipogenic program, but rather acted downstream to switch the fat cell toward a white or beige phenotype.

There are limitations in the interpretation of our data:

1. We have not tested the expression of other thermogenic brakes than TLE3 and RIP140.

2. TLE3 transcript could not be strictly RORα-dependent, because it is not entirely absent from scWAT of sg/sg mice (14); however, comparing the circadian rhythms of the thermogenic program in the scWAT of sg/sg with those in the WT mice (14), we observed that the highest reciprocal variation of PRDM16 and TLE3 mRNA levels took place at the same period of the day (at the beginning of the dark period), suggesting a higher disproportion in the abundance of these reciprocal actors of thermogenesis in the absence of RORα than in its presence. This suggested that the stoichiometric equilibrium between these two factors was modified in sg/sg mice during the circadian cycle of gene expression. However, only experiments of TLE3 gain and function in the presence of RORα inverse-agonist would demonstrate the involvement of TLE3 in the regulation of UCP1 by RORα.

3. The specificity of synthetic ligands is always questionable. Although these RORα ligands were shown to have no activity at other nuclear receptors as defined in a Gal4-chimeric receptor assay (16), some possible interactions within the ROR subfamily existed for some of them. Indeed, while SR3335 is an RORα-specific inverse-agonist (18), SR1001 and SR1078 are able to bind both RORα and RORγ (16, 19). This is the reason why we previously verified the reciprocal effects of the inverse-agonist SR3335 and of the agonist SR1078 on the RORα protein amount in 3T3-F442A adipocytes and then excluded SR1001 because it was not found to decrease the amount of RORα (Fig. 2A). In the case of SR1078, its inhibitory effect on UCP1 expression in scWAT and BAT (Fig. 6) could be because of the inhibition of both RORα and RORγ; this is an option because a specific RORγ inverse-agonist (SR1555) has been reported to increase UCP1 expression into the BAT (39).

Another important point to discuss was whether the link between thermogenesis and RORα activity involved β3-adrenoceptor signaling. Cold being the natural stimulus for the activation of thermogenesis, the prevailing model was that it exerts its effect on adipocytes indirectly, i.e., via the central nervous system, relayed by the β3-adrenoceptor signaling. However, it was reported that the induction of thermogenic genes in the scWAT from β-adrenoceptor-deficient mice was largely preserved on cold exposure, suggesting that there is a β-adrenoceptor-independent pathway that controls thermogenic gene expression in scWAT (31). In sg/sg mice, there was no increase of the β3-adrenoceptor expression (ADBR3) in BAT and WAT versus WT mice (40; Fig. 8). Although increased levels of norepinephrine have been reported in the BAT of sg/sg mice (10), these mice displayed similar increase of oxygen consumption after intraperitoneal injection of the β3-adrenoceptor agonist CL316,243 than WT mice (12). Lau et al. (12) concluded that the increased thermogenic activity of sg/sg mice suggested that RORα deficiency results in a cell-autonomous effect that is not driven by the central nervous system. The ligand-dependent RORα modulation of thermogenesis we observed here confirmed a cell-autonomous process. Indeed, it was obtained ex vivo in the absence of other stimuli. However, during the treatment of differentiating 3T3-F442A adipocytes with the RORα agonist SR1078, we observed a higher inhibition of UCP1 in presence of the β-adrenoceptor agonist isoproterenol (Fig. 2A) than in its absence (data not shown). This suggests a potential interference between RORα activation and β3-adrenoceptor signaling. Nevertheless, our investigations on the possibility for the RORα ligands to directly impact the cAMP/protein kinase A signaling, and thus downstream the activation of β-adrenoceptors, have not been conclusive (data not shown). Interestingly, it was observed that TLE3 expression inhibited the ability of isoproterenol to induce UCP1 expression in PRDM16-expressing cells (34). Given that SR1078 was able to increase TLE3 protein amount in 3T3-F442A adipocytes that also expressed PRDM16 (Fig. 2A), our data confirmed a similar scenario, but the underlying molecular mechanisms remain to be explored. Finally, we observed that SR3335 was able to increase ADBR3 gene expression into the BAT of WT mice after acute cold challenge (Fig. 8), thus perhaps improving the local adrenergic tone that was reduced, but still upgradable by a β3-specific agonist in 13-month- aged mice (41).

On a physiological point of view, we show that RORα ligands could also be useful tools to modulate in vivo UCP1 expression in scWAT of WT mice with associated consequences on the mass of this fat depot. We found that SR3335 and SR1078 exert the opposite effect on the WT mice body weight, without any effect on sg/sg mice. Remarkably, a single injection of SR3335 was able to rapidly decrease their body weight (Fig. 5C), whereas that of SR1078 increased it (Fig. 6C) by ∼3% of their initial weight at the same period of the day. Indirect calorimetry allowed to demonstrate that SR3335 was in vivo able to promote an increase of energy expenditure associated with a decrease of RER, confirming an enhanced lipid oxidation in C57BL6/j mice. We then show that, during cold exposure, a single SR3335 injection enabled WT mice to acquire the same performance than untreated sg/sg mice, suggesting a more efficient adaptive thermogenesis associated with an increased UCP1 expression and a specific loss of scWAT. However, acute cold exposure was not sufficient to assess UCP1-dependent nonshivering thermogenesis, as other thermogenic processes, including shivering, participate in the maintenance of body temperature during the first hours following exposure to low temperatures. Thus, to unequivocally demonstrate that browning induced by pharmacological modulation of RORα affects nonshivering thermogenesis, we performed indirect calorimetry in response to β3-adrenoceptor-specific agonist. Data confirmed that SR3335 increased CL316,243-induced energy expenditure, lipid oxidation, and fat mass loss in C57BL6/j mice.

We thus demonstrate that, similarly to the loss of RORα function in sg/sg mice, pharmacological antagonism of RORα by an inverse-agonist improves the browning of the scWAT depot and potentiates the decrease in the mass of this fat pad in cold-stressed WT mice and in global fat mass in response to a β3-adrenoceptor-specific agonist. It was previously reported that sg/sg mice did not lose more body weight than WT mice after cold exposure (10°C), but they exhibited a higher body temperature (12). We thought that this discrepancy with our own data in sg/sg mice could be explained by the fact that the authors expressed their results in gram of weight lost (absolute value), but not as the percentage of variation of the initial body weight. Indeed, Lau et al. (12) mentioned that the differences in weight between sg/sg and WT mice were <15%; in our case, the significant difference was similar (10%–14%). Indeed, sg/sg mice are leaner than their WT counterparts, and we performed our experiments on aged mice to minimize their body weight differences and to also obtain enough adipose tissue explants to work with. Nevertheless, because age is known to have a deep impact on UCP1 expression that declines precipitously between 3 and 12 mo of age in C57BL6/j mice (41), we previously checked the potential effect of age in the differential response to RORα ligands between the two genotypes. We actually noticed a declined response toward in vivo SR3335 treatment in aged WT mice, but that was nevertheless conserved at the age of 52–65 wk; in 26-wk-old WT mice, the UCP1 protein amount in scWAT increased by approximately fourfold (Fig. 5), then by 1.82- ± 0.52-fold (n = 5) in 52–65 wk-old mice. In parallel, in 52- to 65-wk-old-matched sg/sg mice, UCP1 failed to response to SR3335 (×0.91 ± 0.3) (n = 5), and thus, the difference between the two genotypes remained significant. Aged mice have been shown to be defective in cold-induced beige adipocyte formation owing to a cellular aging senescence-like phenotype of beige progenitors (20). Our findings propose a new therapeutical perspective to activate in vivo the white to beige switch in aged individuals, with beneficial effect on body weight. This effect mainly concerned the scWAT in mice and was tested in only adipocytes differentiated from precursors isolated from human abdominal scWAT. However, it was reported that in humans, visceral adipocytes were more prone to beiging (42); thus, it would be interesting to test the effect of RORα ligands on human deep fat depots. Finally, our in vivo study did not allow to distinguish in which proportion the energy balance of the mice was affected by the SR3335-induced beiging of scWAT and by the SR3335-activation of BAT itself. Nevertheless, we observed a decreased weight of only WAT fat pads, and not of interscapular BAT, by the RORα agonist, perhaps in line with the activation of FFA oxidation in these fat depots. In humans, it is difficult to delineate if BAT activity declines with age or with obesity because these events are often linked. Anyway, it was believed there would be good reason to attempt (pharmacologically) to oppose this development and thus decrease the risk for middle-age obesity (43).

In conclusion, our data highlight insight into 1) a new potential mechanism of WAT beiging that will require more investigations for a better knowledge in the specific biology of white and brown adipose tissues and 2) a potential therapeutic target to stimulate adipose tissue beiging in the absence of cold exposure. The fact that RORα inverse-agonist was also able to decrease neoglucogenesis in the liver ex vivo (44) and blood glucose in diet-induced rodent models of obesity (18) confirmed the therapeutic potential of antagonizing RORα to lower body fat percentage and improve glucose homeostasis. However, it should be kept in mind that RORα is a ubiquitous nuclear factor, the inhibition of which will probably lead to adverse effects in other tissues. Further long-term investigations are mandatory to assess their efficiency and safety in animal models of metabolic diseases.

GRANTS

This work was supported by CNRS, Inserm, Sorbonne University and “la Fondation pour la Recherche Médicale” (EQU201903007868).

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

B.A. conceived and designed research; M.A., N.R., E.C. and B.A. performed experiments; B.A. analyzed data; M.A., E.C. and B.A. interpreted results of experiments; B.A. prepared figures; B.A. drafted manuscript; B.F. and B.A. edited and revised manuscript; M.A., B.F. and B.A. approved final version of manuscript.