Novel Function of Rev-erbα in Promoting Brown Adipogenesis

Brown adipose tissue is a major thermogenic organ that plays a key role in maintenance of body temperature and whole-body energy homeostasis. Rev-erbα, a ligand-dependent nuclear receptor and transcription repressor of the molecular clock, has been implicated in the regulation of adipogenesis. However, whether Rev-erbα participates in brown fat formation is not known. Here we show that Rev-erbα is a key regulator of brown adipose tissue development by promoting brown adipogenesis. Genetic ablation of Rev-erbα in mice severely impairs embryonic and neonatal brown fat formation accompanied by loss of brown identity. This defect is due to a cell-autonomous function of Rev-erbα in brown adipocyte lineage commitment and terminal differentiation, as demonstrated by genetic loss- and gain-of-function studies in mesenchymal precursors and brown preadipocytes. Moreover, pharmacological activation of Rev-erbα activity promotes, whereas its inhibition suppresses brown adipocyte differentiation. Mechanistic investigations reveal that Rev-erbα represses key components of the TGF-β cascade, an inhibitory pathway of brown fat development. Collectively, our findings delineate a novel role of Rev-erbα in driving brown adipocyte development, and provide experimental evidence that pharmacological interventions of Rev-erbα may offer new avenues for the treatment of obesity and related metabolic disorders.


Loss of Rev-erbα impairs BAT development.
To study the function of Rev-erbα in brown fat, we first examined its abundance in brown and white adipose tissue. Rev-erbα protein is highly enriched in BAT as compared to its low expression in WAT, at a level that is comparable to the liver (Fig. 1A). In contrast, the Rev-erbα target gene in the molecular clock circuit, Bmal1, display an opposite distribution pattern in adipose tissues with higher level in WAT than BAT. To investigate the in vivo function of Rev-erbα in brown fat development, we generated Rev-erbα-null (Rev-erbα −/− ) mice using gene-targeted ES cell clone obtained from the KOMP consortium, with the null allele identified as a 380 bp PCR amplification (Fig. 1B). The homozygote null mice are viable, but adult mice show reduced fertility, as reported previously 23 . The brown fat of Rev-erbα −/− mice display complete ablation of Rev-erbα (Fig. 1C), while mRNA level of its direct target gene, Bmal1, is markedly induced as compared to wild-type controls (WT), as expected. Next, we examined the effects of loss of Rev-erbα function on BAT development in embryonic and neonatal stages in WT, heterozygote and homozygote mutants. As BAT forms late during development with fully mature features detected after E16.5 24 , we first analyzed BAT formation at E18.5 embryos by haematoxylin and eosin staining. BAT in WT mice at this stage of development display fully organized, densely stained structures with three separate lobes in between the dermis and underlying muscle layer, as shown in Fig. 1D. In contrast, the size of interscapular BAT in Rev-erbα −/− mice is markedly reduced as compared to that of the WT littermates (Fig. 1D), with large part of the structure replaced by white adipose tissue. In addition, the structural organization of these brown fat pads with the underlying muscular layer is severely disrupted, with skeletal muscle interspersed within the brown fat. BAT of Rev-erbα +/− embryos are also smaller, but their juxtaposition with adjacent muscle is largely maintained. Therefore, the loss of Rev-erbα in vivo markedly impairs the formation, structural integrity and characteristics of BAT.
We further examined BAT formation in early postnatal development of day 5 (P5), which allows us access to tissues and quantitative analysis. Similar substantially reduced size and disrupted structure of BAT is evident in the Rev-erbα −/− mice through examination of one side of the interscapular fat pad ( Fig. 2A), with Rev-erbα +/− affected to a lesser extent. Reduction of the Rev-erbα −/− BAT size is accompanied by diminished mitochondrial-rich cytoplasmic staining characteristic of brown fat and its replacement by increased lipid droplets as compared to controls (Fig. 2B). Analysis of total protein in BAT reveals nearly absence of UCP-1 expression and marked reduction of C/EBPβ in these mice, corroborating the histological observation of loss of brown fat identity (Fig. 2C). Likely due to impaired brown fat development, the BAT mass as determined by its tissue weight (Fig. 2D), or tissue to body weight ratio (Fig. 2E) is ~16% and ~31% lower in neonatal Rev-erbα +/− and Rev-erbα −/− mice than that of the littermate controls, respectively. Their body weight at this developmental stage is not altered (Fig. 2F). Similar to what we found in embryonic development and neonatal mice, brown fat weight in 10 weeks old adult Rev-erbα +/− and Rev-erbα −/− mice remains to be significantly lower (Fig. S1A).
Interestingly, in addition to impaired BAT formation, adult Rev-erbα −/− display ~40% reduction of WAT mass relative to the WT, as indicated by measured epididymal fat pad weight (Fig. S1B) or NMR analysis of total body fat ( Fig. S1C & S1D). In addition, NMR analysis reveals a significant reduction of total lean mass in Rev −/− mice, but not in Rev +/− mice (Fig. S1E). The body weight of adult Rev-erbα −/− mice are also significantly lower than that of the WT controls (Fig. S1F), whereas their heart weights are not affected (Fig. S1G). As lean mass accounts for over 80% of total body weight, the reduced lean weight likely contributed to the lower body weight observed in Rev −/− mice together with lower fat mass. The reduced amount of white adipose tissue observed in adult Rev −/− mice is consistent with previously reported Rev-erba positive regulation of adipogenesis 20,25 , and shRNA silencing of Rev-erbα in 3T3-L1 cells further demonstrates this function (Fig. S2). Interestingly, despite the reduction of brown adipose tissue in adult Rev-erbα −/− mice, the H/E histology is largely comparable to that of the WT (Fig. S3A), and cold tolerance test reveals similar responses (Fig. S3B). Furthermore, gene expression analysis indicate induction of UCP-1 mRNA in Rev-erbα −/− mice, although brown adipogenic markers (Myf5, Pparg, Fabp4), and mitochondrial genes (Pgc1a and Cox7a1) are down-regulated (Fig. S3C). These intriguing findings suggest that despite the marked impairment of brown adipose tissue development during embryonic and neonatal stages, adult Rev-erbα −/− mice do generate functional brown fat.
Silencing of Rev-erbα inhibits mesenchymal precursor brown lineage commitment and differentiation. Based on the finding that Rev-erbα promotes BAT development, we next explored whether it directly participates in brown adipogenesis in a cell-autonomous manner using genetic lossand gain-of-function studies in brown adipocyte progenitors. Brown adipocytes arise from mesenchymal precursors that first commit to a brown progenitor fate and subsequently undergo terminal differentiation 26 . Initial analysis of Rev-erbα protein level in mesenchymal precursors and brown preadipocytes, as compared to white preadipocytes and myoblasts, indicate that it is more abundant in C3H10T1/2 (10T1/2) mesenchymal stem cells and HIB1B brown preadipocytes (Fig. 3A), two cell types that are capable of differentiating into mature brown adipocytes under specific induction conditions. Using stable  transfection of a retroviral short hairpin RNA (shRNA) construct targeting Rev-erbα (shRev), we generated stable knockdown clones of shRev-expressing 10T1/2 mesenchymal precursors, which results in near absence of Rev-erbα protein as compared to cells expressing scrambled control shRNA (shSC) (Fig. 3B). When subjected to a brown adipocyte induction cocktail, scrambled control cells display robust progression to mature adipocyte, with ~48% exhibiting typical round morphology and muti-locular lipid accumulation as indicated by phase-contrast images (Fig. 3C). In comparison, Rev-erbα silencing markedly reduces the percentage of cells with mature adipocyte morphology to ~22%. BODIPY staining of neutral lipids accumulation as an index of mature differentiation confirms the reduction of lipid droplets in shRev cells. Moreover, MitoTracker Red staining of functional mitochondria in live cells is substantially weaker in Rev-erbα-deficient cells than that of the shSC. The observed lower mitochondrial abundance suggests that loss of Rev-erbα impairs mitochondrial biogenesis associated with mature brown adipocyte differentiation. Furthermore, gene expression analysis at beginning, day 4 and 9 of differentiation demonstrates a markedly suppressed brown adipogenic gene program (Fig. 3D). The induction of brown-specific differentiation markers, Ucp-1 and Prdm16, are significantly attenuated in shRev cells as compared to their robust expression (~7-8 fold induction) in the controls. The lack of induction of adipogenic genes, Cebpβ, Cebpα, Pparγ and FABP4 during the differentiation time course indicates suppressed adipogenesis (Fig. 3D). We also found a moderate ~2-fold up-regulation of Rev-erbα mRNA level at day 9 of brown adipogenic induction, with significantly attenuated expression in the shRev cells. Collectively, these effects of Rev-erbα deficiency in 10T1/2 cells indicate that it modulates the lineage commitment and differentiation of mesenchymal precursor cells into brown adipocytes.
Rev-erbα promotes brown preadipocyte terminal differentiation. Next, we determined the role of Rev-erbα in terminal differentiation stage of brown adipogenesis through gain-and loss-of-function studies in HIB1B brown preadipocytes. Similar to the phenotype observed with stable Rev-erbα shRNA depletion in mesenchymal precursors, inhibition of Rev-erbα in committed brown preadipocytes markedly suppresses lipid accumulation and mitochondrial abundance at day 6 of differentiation, as indicated by reduced BODIPY and MitoTracker staining relative to scrambled control cells (Fig. 4A). This significantly attenuated differentiation is accompanied by marked suppression of brown specific marker gene expression (Ucp-1, Dio2 and Prdm16), and adipogenic factors (Cebpβ, Cebpα and Pparγ) (Fig. 4B). Stable Rev-erbα depletion in these cells achieved ~70% reduction of Rev-erbα transcript.
Using stable expression of a Rev-erbα cDNA plasmid (Rev-cDNA) in HIB1B cells, we tested the effect of Rev-erbα overexpression on brown adipocyte terminal differentiation. Forced expression of Rev-cDNA markedly elevated Rev-erbα protein level over vector control, pcDNA3 (Fig. 5A). Consistent with its protein expression, Rev-erbα transcript is increased ~6-fold in Rev-cDNA-expressing cells as indicated by RT-qPCR analysis (Fig. 5C). BODIPY and MitoTracker staining to assess mature differentiation demonstrate increased lipid accumulation and mitochondrial abundance in Rev-cDNA-expressing cells at day 6 of differentiation than controls, indicating that ectopic expression of Rev-erbα enhances terminal differentiation (Fig. 5B). Of note, at the same differentiation stage, the pcDNA3 vector control display less lipid accumulation and mitochondrial staining than scrambled shRNA control shown in Fig. 5B, suggesting a slight inhibitory effect of pcDNA3 vector on brown differentiation. Gene expression analysis of Rev-erbα-overexpressing HIB1B cells further demonstrates marked induction of brown adipocyte-specific markers over vector controls, together with moderate up-regulation of genes involved in adipogenesis (Fig. 5C).
To assess the impact of complete loss of Rev-erbα function on brown preadipocyte differentiation, we isolated primary brown preadipocytes from WT and Rev-erbα −/− mice and generated immortalized cell clones using SV40 large T antigen retroviral transduction. As shown in Fig. 6A,B immortalized WT primary brown preadipocytes exhibit rounded mature brown adipocyte morphology with robust lipid accumulation and mitochondrial formation upon brown adipogenic induction. In sharp contrast, differentiation of Rev-erbα −/− brown preadipocytes is severely blocked, as they fail to progress to mature adipocytes and display barely detectable amount of lipid and mitochondria staining. Nearly abolished brown adipogenic marker gene induction in Rev-erbα −/− cells relative to WT controls further confirms the severe defect in differentiation (Fig. 6C). Taken together, these findings indicate that Rev-erbα positively regulates terminal differentiation in lineage-committed brown preadipocytes.
Pharmacological interventions of Rev-erbα activity modulates brown adipogenesis. As a ligand-dependent nuclear receptor, the transcriptional activity of Rev-erbα is amenable to regulation by small molecule ligands, and both its agonist, SR9011, and antagonist, SR8278, display metabolic effects [27][28][29] . Therefore, we tested the effect of pharmacological activation and blockade of Rev-erbα activity on brown adipocyte differentiation using SR9011 and SR8278, respectively. In line with results from genetic interventions, Rev-erbα activation by SR9011 enhances, whereas its inhibition by SR8278 impairs brown preadipocyte terminal differentiation in comparison to DMSO vehicle control, as indicated by lipid and mitochondrial staining shown in Fig. 7A,B. Repression and induction of Bmal1 mRNA in response to SR9011 and SR8278, respectively, confirms the modulation of Rev-erbα activity by these ligands in brown adipocytes (Fig. 7B). SR8278 suppression of brown adipocyte marker genes expression further validates the findings on differentiation as assessed by morphological progression (Fig. 7B). Interestingly, although SR9011 significantly up-regulates brown-specific markers, its effect on adipogenic Scientific RepoRts | 5:11239 | DOi: 10.1038/srep11239 factors are not evident. Together, these results demonstrate that modulation of Rev-erbα activity by pharmacological means is sufficient to impact brown adipogenesis.
Interestingly, molecular clock components in epidermal stem cells regulate various signaling cascades, including the TGF-β pathway 33,34 , and a recently published ChIP-Seq dataset revealed Rev-erbα binding peaks on certain TGF-β pathway gene regulatory regions 28 . We thus tested whether Rev-erbα could modulate TGF-β signaling to impact brown adipogenesis. Using the TGF-β -responsive luciferase reporter, SBE4-Luc, to assess TGF-β pathway activity, we found significantly increased reporter activity in Rev-erbα-deficient 10T1/2 cells as compared to SC control upon TGF-β stimulation, indicating that loss of Rev-erbα function promotes TGF-β signaling transduction (Fig. 8A). Smad3 phosphorylation in response to TGF-β activation ultimately transduces TGF-β signaling to the nucleus to modulate gene transcription 35 . Therefore, we examined Smad3 phosphorylation in WT and Rev-erbα-null brown preadipocytes. Robust Smad3 phosphorylation is detected in WT cells only in response to TGF-β 1, as expected. In contrast, Rev-erbα-null cells already display a moderate level of Smad3 phosphorylation at basal condition that is enhanced by TGF-β 1 treatment (Fig. 8B). Non-responsiveness of Smad3 to BMP4 stimulation validates the specificity of TGF-β 1 signaling. Interestingly, total Smad3 protein level is increased in Rev-erbα-null preadipocytes, indicating that Rev-erbα impacts Smad3 expression. To address potential direct transcriptional regulation of Rev-erbα on genes of the TGF-β signaling pathway, we screened for RORE elements, in − 2 kb and first intron of their regulatory regions to identify putative Rev-erbα binding sites. Among the genes screened, including the ligands Tgfb1 and Tgfb2, the receptors Tgfbr1 and Tgfbr2, and signal transducer Smad3, Rev-erbα response elements were identified in promoter regions of Tgfbr2 and Smad3, as listed in Fig S3. Using specific primers flanking these identified sites (Fig. S4), chromatin immunoprecipitation qPCR (ChIP-qPCR) analysis detects Rev-erbα enrichment on Tgfbr2 and Smad3 promoters similar to, or higher than that of the known Rev-erbα target, Bmal1 (Fig. 8C). In line with this finding, absence of Rev-erbα in brown preadipocytes leads to a ~15-fold induction of Smad3 mRNA, and a tendency for higher Tgfbr2 expression (Fig. 8D). These results demonstrate that Rev-erbα is a bona-fide transcription repressor of TGF-β pathway genes and loss of its repression leads to enhanced TGF-β signaling in brown adipocyte progenitors.  Collectively, findings from our study suggest a model in which Rev-erbα transcriptional control of the TGF-β pathway suppresses TGF-β activity, and attenuation of this inhibitory signal of brown fat development ultimately promotes brown adipocyte lineage determination and terminal differentiation (Fig. 8E).

Discussion
Compare to the extensive knowledge of white adipocyte development, our current understanding of regulatory mechanisms governing BAT formation is only emerging. Employing genetic approaches and pharmacological interventions, our study uncovers a novel, cell-autonomous function of Rev-erbα as a key regulator of brown adipocytes lineage determination and terminal differentiation, which critically impacts brown adipose tissue development.
We found that Rev-erbα plays a critical role in promoting de novo brown adipocyte formation. In brown adipocytes, Rev-erbα exerts transcriptional repression of key genes of the TGF-β pathway, Tgfbr2 and Smad3. TGF-β signaling is a known inhibitory pathway of brown fat development [30][31][32] . Various components of this pathway, including various ligands, the receptors or signal transducers, have been shown to suppress brown fat development. Ablation of Smad3, the ultimate signaling mediator of the TGF-β pathway, selectively induces brown-specific markers gene expression, particularly Prdm16, and enhances mitochondrial function 32 . Inhibition of the Activin receptor IIB or myostatin, both signaling through Smad3, also promote brown adipogenesis 30,31,36,37 . Interestingly, circadian clock regulators, such as Bmal1 or the Period genes, have been demonstrated to modulate developmental signaling cascades, including the TGF-β pathway, in epidermal stem cells 33 . Thus, it is conceivable that the circadian clock regulatory circuit exerts coordinated temporal control on developmental signaling pathways to modulate stem cell activation and differentiation behaviors. Indeed, Smad3, the ultimate TGF-β signal transducer, is circadianly-regulated in mesenchymal stem cells 38 , while intrinsic diurnal phosphorylation of Smad3 are detected in suprachiasmatic nuclei 39 . Given the importance of TGF-β signaling in stem cell proliferation and differentiation 40 , Rev-erbα modulation of this pathway may confer specific temporal cues to fine-tune tissue development, such as in the brown fat. Our current study defines a specific role of Rev-erbα in brown adipocyte differentiation and development program. An intriguing aspect of its potential broader impact on additional TGF-β -regulated biological processes remains to be elucidated.
Rev-erbα has been shown to regulate white adipocyte differentiation [19][20][21] . Fontaine et al. reported that Rev-erbα promotes adipogenesis as a direct target of PPARγ 20 and Rev-erbα agonist was found to stimulate 3T3-L1 differentiation to a similar extent as Rosiglitazone 25 , while Wang et al. found its dynamic regulation and bifunctional role in this process 21 . Hence, the precise mechanisms mediating Rev-erbα actions in white adipocyte differentiation remains unclear. Nonetheless, in line with the reported involvement of Rev-erbα in promoting white adipocyte differentiation 20,21 , we observed a severe loss of white adipose depot in Rev-erbα −/− mice in addition to impaired BAT development (Fig. S1). Although Rev-erbα effect in other tissues could potentially contribute to this observed phenotype in WAT, our further analysis of stable silencing of Rev-erbα in 3T3-L1 found substantially suppressed adipogenesis (Fig. S2), indicating its cell-autonomous role in white adipocyte differentiation. TGF-β signaling is a potent inhibitor of adipogenesis 41 , and transgenic overexpression of TGF-β 1 in mice led to a severe lipodystrophy-like phenotype with severe reductions of white and brown adipose tissue 42 . Therefore, it is intriguing to postulate that Rev-erbα modulation of TGF-β pathway may also contribute, at least in part, to its observed effect on white adipose tissue formation. The new evidence of Rev-erbα regulation of brown adipogenesis and its established effect on white adipose tissue imply that Rev-erbα participates in the adipogenic cascade, namely, the sequential activation events of adipogenic factors CEBPβ , PPARγ and CEBPα , a shared feature of brown and white adipocyte differentiation. As Smad3, the signal transducer of the TGF-β pathway, is known to directly repress the transactivation activity of C/EBP family of adipogenic factors to suppress adipogenesis 43,44 , Rev-erbα negative regulation of TGF-β pathway genes we observed is, therefore, in line with this notion.
Our study reveals a previously unappreciated function of Rev-erbα as a positive regulator of brown fat development. Not only the absence of Rev-erbα significantly impairs the in vivo formation and structural organization of brown fat, it results in substantial loss of brown fat characteristics. Nonetheless, due to the global targeted deletion nature of our current model, Rev-erbα functions in other tissues could potentially confound findings in BAT. Therefore, we focused on effects of Rev-erbα on BAT development in order to minimize such issues. Additional in vitro analysis and ex vivo studies of Rev-erbα-deficiency in brown adipocyte differentiation lends further support to cell-autonomous functions of Rev-erbα in brown adipogenesis. However, based on analyses of BAT in adult Rev-erbα-null mice, the severe impairment of embryonic and neonatal development could represent a developmental delay, as adult mice are able to generate functional brown fat (Fig. S3).
We detected a paradoxical induction of UCP-1 mRNA in adult Rev −/− BAT, which was similarly reported by Gerhart-Hines et al previously 45 . However, due to the global ablation nature, we postulate that the significant lack of white adipose tissue observed in Rev −/− mice may induce cold stress at ambient temperature (22 °C) leading to UCP-1 induction via sympathetic overdrive. We thus examined norepinephrine levels in plasma (Fig. S4A) and 16-hour overnight urine collection (Fig. S4B), and found doubling of the normal WT values in the Rev −/− mice. This finding suggests activation of the sympathetic-adrenergic system and provides potential explanation why Ucp-1 is elevated in Rev −/− BAT despite decreased brown adipogenic markers. It is also possible, that the comparable cold tolerance response observed between WT and Rev −/− mice is a combined result of higher Ucp-1 expression together with reduced amount of brown fat in these mice. As Gerhart-Hines et al 45 performed their analyses at thermo-neutral conditions, the sympathetic overstimulation observed in adult Rev −/− mice in our study carried out under ambient facility temperature (22 °C) may account for certain differences in these independent studies. Given that the regulation of BAT formation and brown adipogenesis, which is the focus of our study, and acute modulation of its function, investigated by Gerhart-Hines et al 45 , both contributes to the total thermogenic capacity of BAT, these studies likely reflect distinct yet related aspects of Rev-erbα functions in BAT. Nonetheless, its specific role in brown fat activity will require future investigations using appropriate tissue-selective ablation models. Taken together, these newly discovered layers of Rev-erbα actions in BAT highlights its importance in fine-tuning thermogenic capacities during development and in response to functional demand.
As a major energy-dissipating organ, BAT activity critically regulates whole-body energy homeostasis 46 . As our study suggests, a Rev-erbα-controlled regulation of brown fat metabolic capacity could Scientific RepoRts | 5:11239 | DOi: 10.1038/srep11239 influence energy expenditure and metabolic homeostasis. Most importantly, as a ligand-activated nuclear receptor amenable to modulation by synthetic small molecule ligands 29 , Rev-erbα represents an ideal target for therapeutic interventions. Our elucidation of an important role of Rev-erbα in brown adipogenesis and the molecular mechanisms mediating its action may thus lead to discovery of new therapies against development of obesity and related metabolic disorders.

Methods
Animals. Mice were maintained in the Houston Methodist Hospital Research Institute mice facility at ambient temperature under a constant 12:12 light dark cycle, with light on at 7:00 AM (ZT0). All experimental protocols were approved by the IACUC animal care research committee of the Houston Methodist Research Institute, and carried out in accordance with the NIH Guidelines for the Care and Use of Laboratory Animals. Mice with global ablation of Rev-erbα were generated using gene targeted embryonic stem cells, Velocigene ES clone 11705A-E7, C57BL/6NTac origin, obtained from KOMP repository. Briefly, ES cell clones were injected into albino C57BL/6J-Tyr mouse blastocysts. The chimeric mice obtained were mated with C57BL/6J-Tyr mice for germ line transmission and the targeted Rev-erbα allele was confirmed by polymerase chain reaction (PCR) analysis using tail DNA. Offspring were backcrossed to C57BL/6J, and heterozygotes were bred to obtain homozygote null and littermate wild-type controls used in the study. The Rev-erbα homozygotes were born at a normal Mendelian ratio without obvious developmental defects. Tissue samples were obtained at ambient temperature at the same time of the day to ensure comparable circadian gene expression.
Brown adipogenic differentiation of C3H10T1/2 and HIB1B cells. Mesenchymal stem cell line, C3H10T1/2, and brown preadipocyte cell line, HIB1B, were obtained from ATCC and maintained at subconfluence between passages in DMEM with 10% FBS and antibiotics. C3H10T1/2 differentiation to brown adipocytes was conducted as described 47 without BMP7 pretreatment. Briefly, cells were grown to confluency and subjected to brown induction media of DMEM containing 10% FBS, insulin (20 nM), T3 (1 nM), isobutylmethylxanthine (0.5 mM), dexamethasone (5 mM), and Rosiglitazone (1 μ M) for 3 days. Cells were then switched to maintenance media of DMEM containing 10% FBS supplemented with only insulin and T3 for 9 additional days. Fully differentiated brown phenotype with significant lipid accumulation and Ucp-1 expression occur at 9 days (D9) after induction. Differentiation of HIB-1B and primary brown adipocytes were induced using similar cocktails with induction media for two days and maintenance media for four days.

Generation of stable knockdown and overexpression cell lines.
Rev-erbα shRNA construct in pGIPZ vector (V2LMM_25660) and scrambled control vector (RHS4346) were purchased from Open Biosystems, and mouse full-length Rev-erbα cDNA clone (MC203687 in pcDNA3 vector) was obtained from OriGene. C3H10T1/2 or HIB1B cell lines expressing Rev-erbα shRNA or cDNA were transiently transfected using FuGENE 6 reagent, and selection of stably transfected clones by Puromycin (for shRNA) or Neomycin (for pcDNA3) were started 48 hours following transfection and maintained for 7-10 days. The entire transfected plate was subcultured in order to minimize the effect of confluency on efficiency of differentiation, as described previously.
Primary brown adipocyte, preadipocyte isolation and immortalization. Primary brown adipocytes and preadipocytes were isolated from interscapular brown adipose tissue pad of 4-week-old mice, as described 48 . Briefly, tissues were digested by Type I collagenase digestion in the presence of 1% BSA at 37 °C for 30 minutes. The suspension was filtered, centrifuged, and the top fat layer was collected as adipocytes and pellet containing the stromal vascular fraction was resuspended and plated. Preadipocytes in the stromal vascular fraction were passaged once prior to adipogenic differentiation. Immortalization of isolated primary preadipocytes was performed using by retroviral SV-40 Large T antigen transformation and puromycin selection, as described 47 . BODIPY and Mitotracker staining. BODIPY 493/503 (Life Technologies) staining of neutral lipids was carried out at a concentration of 1 mg/ml after formaldehyde fixation and incubation for 30 minutes. MitoTracker Red (MitoTracker Deep Red FM, 100 nM, Life Technologies) staining was applied to live cells at 37°C for 30 minutes before fixation to stain functional mitochondria, according to manufacturer's protocol.

RNA extraction and quantitative reverse-transcriptase PCR analysis. RNeasy miniprep kits
(Qiagen) were used to isolate total RNA from snap-frozen tissues or cells. Tissues samples are collected at times as indicated and cell samples were obtained at non-synchronized normal culture conditions. cDNA was generated using q-Script cDNA Supermix kit (Quanta Biosciences) and quantitative PCR was performed using a Roche 480 Light Cycler with Perfecta SYBR Green Supermix (Quanta Biosciences), as described 49 . Relative expression levels were determined using the comparative Ct method to normalize target genes to 36B4 internal control or compared to controls as indicated.
Chromatin immunoprecipitation (ChIP)-qPCR analysis. Immunoprecipitation was performed using Bmal1 Rev-erbα antibody (PA5-29865) or control rabbit IgG plus protein A/G beads, as described 50 . Briefly, cells were fixed by formaldehyde, lysed and sonicated to shear the chromatin. The immunoprecipitated chromatin fragments were eluted, treated with proteinase K and purified using Qiaquick PCR purification kit (Qiagen). Real-time PCR using Perfecta SYBR Green Supermix (Quanta Biosciences) was carried out with an equal volume (4 ul) of each reaction with specific primers. Sequences of the specific primers flanking the identified RORE sites are listed in supplementary Fig. S3. TBP were included as negative controls, and the known Rev-erbα target Bmal1 as a positive control. Data was expressed as fold enrichment over IgG after normalization to 1% input.
Luciferase reporter assays. C3H10T1/2 cells were seeded to ~80% confluency overnight in 24 well plates. Transient transfection using FuGENE 6 (Roche) was carried out in four replicates, as described 39 . Per well, the transfection mixture contains 150 ng of TGF-β -responsive SBE4-Luc luciferase reporter (Addgene Plasmid 16495 51 ) together with 20 ng of Renilla luciferase (pRL-TK, Promega) as an internal control. TGF-β 1 (2 ng/ml) were added 16 hours after transfection and luciferase activity was measured using Dual-Glo luciferase assay system (Promega) 24 hours following ligand treatment. Reporter luciferase values were normalized to Renilla readings and expressed as fold of induction over controls.
Statistical analysis. Data is expressed as Mean ± SE. Statistical differences between mean values of two groups were assessed by two-tailed, unpaired Student's t test Student's t test. P ≤ 0.05 is considered as statistically significant.