Tuberous Sclerosis Complex 1–Mechanistic Target of Rapamycin Complex 1 Signaling Determines Brown-to-White Adipocyte Phenotypic Switch

In mammals, there exist two different types of adipose tissues: white adipose tissues (WATs) and brown adipose tissues (BATs), which are made up of unilocular white adipocytes and multilocular (ML) brown adipocytes, respectively. White adipocytes contain all of the enzymatic machinery necessary to build triglycerides from fatty acids synthesized de novo or imported from circulating lipoproteins, while brown adipocytes comprise of abundant mitochondria that oxidize fatty acids and generate heat via uncoupling protein 1 (UCP1) in response to cold or excess feeding (1–3). WAT and BAT are traditionally considered distinct fat depots located in different anatomic sites. However, recent studies have shown that most of the fat depots contain both WAT and BAT, which are interconvertible (4). Differentiated WAT has been demonstrated to acquire certain BAT-like properties and can be converted into an energy-consumption organ under special conditions. Cold exposure or administration of β3-agonist has been reported to induce the emergence of ML brown adipocytes containing a high amount of mitochondria and UCP1 in WAT depots (5–7).

The molecular mechanism underlying the interconversion between WAT and BAT is currently under active investigation. Study with the transgenic mouse model suggests that C/EBPβ is critical for the brown adipocyte conversion. Transgenic knockin mouse that replaces the C/EBPα gene by C/EBPβ gene (denoted as C/EBP β/β) shows an increase in brown adipocyte-like cells characterized by cAMP accumulation, enhanced mitochondrial biogenesis, and UCP1 expression in WATs (8). In addition, several transcription-related molecules such as retinoblastoma protein (Rb), p107, transcriptional intermediary factor 2, 4E-binding protein 1, and RIP140 (9–15) have been reported to influence the brown or white fat phenotype selectivity. All of these transcriptional factors act to suppress the brown fat phenotype. Mice lacking p107, Rb, or RIP140 demonstrate a uniform replacement of WAT with BAT and elevated levels of peroxisome proliferator–activated receptor γ coactivator 1α (PGC1α) and UCP1 (13,14). In contrast, FoxC2, a member of the forkhead transcription factor family, induces the emergence of brown fat cells in WAT. WAT in mice overexpressing FoxC2 acquires certain BAT-like properties such as increased levels of PGC1α, UCP1, and two important cAMP pathway proteins: β3-adrenoceptor and the protein kinase A α regulatory subunit 1 (16). While these studies suggest that a group of transcriptional factors is critical for maintaining the WAT identity, the upstream signals or factors linking the extracellular signals with these transcriptional factors to initiate the conversion and therefore determine the fate of BAT versus WAT remain largely unknown.

The mechanistic target of rapamycin (mTOR), an evolutionarily conserved serine-threonine protein kinase, can sense the extracellular energy status to regulate the cell growth and proliferation in a variety of cells and tissues. Two distinct mTOR complexes, mTORC1 and mTORC2, have been characterized. mTORC1 is sensitive to rapamycin and is responsible for phosphorylation of ribosomal S6 kinase 1 (S6K1) and eukaryotic initiation factor 4E-binding protein 1. In contrast, mTORC2 is not inhibited by rapamycin (17,18). In adipocytes, mTOR is proposed to regulate protein synthesis, leptin synthesis and secretion, and adipogenesis (19–21). Inhibition of mTORC1 signaling by rapamycin attenuates the mTOR-induced adipogenesis (22–24). All of these data suggest that mTOR may serve as the critical intracellular molecule sensing the extracellular energy surplus to increase the energy storage capacity of adipocytes. Whether mTOR signaling controls the interconversion between brown and white adipocytes remains to be explored.

Tuberous sclerosis complex 1 (TSC1) and TSC2 form a heterodimer complex to negatively regulate mTORC1 activity, but might activate mTORC2 (18,25). In order to examine whether the TSC1-mTOR signaling is involved in the fate determination of BAT versus WAT, we generated a colony of transgenic mice Fabp4-Tsc1^−/− mice in which TSC1, the negative upstream regulator of mTORC1 signaling, is specifically deleted by fatty acid binding protein 4 (FABP4)-Cre (26). We report in this study that deletion of Tsc1 gene activates mTORC1 signaling in the BAT of Fabp4-Tsc1^−/− mice, which subsequently induces the conversion of brown adipocytes into cells with the morphology and gene expression pattern of white adipocytes.

Fabp4-Cre mice that express the Cre recombinase gene under the control of the FABP4 gene promoter, and Tsc1^lox/lox mice in which exons 17 and 18 of Ts1 gene are flanked by loxP sites by homologous recombination, were purchased from The Jackson Laboratory (Bar Harbor, ME). Fabp4-Tsc1^−/− mice were generated by breeding Tsc1^lox/lox mice with Fabp4-Cre mice. Control experiments were performed using littermate Tsc1^lox/lox animals (26). Mice were housed on a 12:12-h light/dark cycle. Regular chow and water were available ad libitum. The animals used in this study were handled in accordance with the Guide for the Care and Use of Laboratory Animals published by the National Institutes of Health (publication no. 85-23, revised 1996). All of the experimental protocols were approved by the Animal Care and Use Committee of Peking University.

Rapamycin (Santa Cruz Biotechnology) was initially dissolved in 100% DMSO, stored at −20°C, and further diluted in normal saline immediately before use. Pregnant mice were randomly divided into two groups: control and treatment. At the near-term pregnancy, mice were intramuscularly injected with rapamycin at a dose of 1 mg/kg body weight/day or vehicle for 3 consecutive days before the birth of pups.

Interscapular BAT was harvested from neonatal Tsc1^lox/lox mice and brown adipocytes isolated by digestion with collagenase and mechanistic dispersion as previously described (27). Isolated cells were plated in tissue-culture dishes in DMEM supplemented with 20% FBS. After 4 h of culture at 37°C, cells were rinsed twice with PBS, after which 70% of the initial cells were attached to the dish, forming a monolayer. To induce adipocyte differentiation, cells were cultured for 2 days in 10% FBS-DMEM supplemented with 20 nmol/L insulin, 1 nmol/L T3, 12.5 mmol/L indomethacin, 0.5 mmol/L isobutylmethylxanthine, and 2 mg/mL dexamethasone. On days 3–10, the induction medium was substituted by maintenance medium consisting of DMEM supplemented with 20 nmol/L insulin and 1 nmol/L T3. Fresh maintenance medium was added every 3 days until day 10.

The Cre adenoviruses were expanded, titrated in 293 cells, and purified by cesium chloride methods as described previously (28). For adenovirus-mediated gene transfer, primary brown fat preadipocytes were cultured to 30–50% confluence and infected with adenovirus for 48 h. Infection efficiency was generally >60%, as judged by green fluorescent protein (GFP) expression observed under the microscope. Infected primary brown fat preadipocytes were differentiated for 10 days and then harvested for subsequent analysis.

Total RNA was isolated using the TRIzol reagent. Mitochondrial DNA (mtDNA) was isolated by Mitochondrial DNA Isolation Kit (BioVision Inc., Milpitas, CA). Reverse transcription and quantitative real-time PCR were performed as previously described (29,30). PCR reactions were performed in duplicate, and each experiment was repeated three to five times. Primers used in this study were shown in Table 1.

Protein extracts of BAT tissues or cultured cells were immunoblotted with UCP1 (Santa Cruz Biotechnology), angiotensinogen (Agt; Epitomics), phosphorylated (p-)mTOR (Ser2448; Cell Signaling Technology, Beverly, MA), mTOR (Cell Signaling Technology), p-S6K1 (Thr389; Cell Signaling Technology), S6K1 (Cell Signaling Technology), p-S6 (Ser235/236; Cell Signaling Technology), S6 (Cell Signaling Technology), TSC1 (Cell Signaling Technology), TSC2 (Cell Signaling Technology), and β-actin (Cell Signaling Technology) as previously described (29,30). After incubation with IRDye-conjugated second antibody for 1 h, specific reaction was visualized using the Odyssey infrared imaging system (LI-COR Biosciences, Lincoln, NE).

The BATs were quickly removed and prepared for immunostaining. Slides were individually incubated with p-S6 (Ser235/236; Cell Signaling Technology) antibody (1:100) in a humid chamber at 4°C overnight. Secondary antibody staining was performed with a biotin-labeled horse anti-rabbit antibody (1:200) for 1 h at room temperature, followed by incubation with a streptavidin-biotin horseradish peroxidase complex (Vector Laboratories, Burlingame, CA). Immunoreactivity was detected using diaminobenzidine substrate (peroxide substrate kit, SK-4100; Vector Laboratories) for 2–5 min. Slides were then counterstained with Mayer hematoxylin before dehydration and mounting. The expression of p-S6 was evaluated by comparing the staining intensities between different samples incubated with specific primary antibody or control IgG under same conditions.

The BATs were fixed with 3% glutaraldehyde and 1% osmium in 0.1% cacodylate buffer, rinsed with a series of graded acetone solutions, and then embedded in epon 812. After ultrathin sectioning with an ultramicrotome (MT-7000; RMC Products, Tucson, AZ), specimens were stained with uranyl acetate and lead citrate and then observed under the JEM1230 electron microscope (Horiba Corp., Kyoto, Japan).

Data were expressed as mean ± SEM. Data analysis used GraphPad Prism software. One-way ANOVA, Student-Newman-Keul test (comparisons between multiple groups), or unpaired Student t test (between two groups) was used as appropriate. A P value <0.05 denotes statistical significance.

To explore the role of TSC1-mTORC1 signaling in the determination of brown versus white adipocytes, we generated the Fabp4-Tsc1^−/− transgenic mice in which the Tsc1 gene was specifically deleted by FABP4-Cre. As showed in Fig. 1A, levels of TSC1 were negligible in the BAT of Fabp4-Tsc1^−/− transgenic mice, whereas modest expression was detected in wild-type (WT) littermates. Levels of TSC2 remain unaltered in Fabp4-Tsc1^−/− transgenic mice relative to the WT animals. Similar reduction in TSC1 levels was also observed in the brain, skin and liver tissues, whereas no alteration was detected in heart, lung and kidney (Fig. 1B). These data indicate the successful deletion of the Tsc1 gene in the Fabp4-Tsc1^−/− transgenic mice. Since TSC1 is the upstream negative regulator of mTORC1 signaling, we next investigated the mTOR activity in the BAT derived from the Fabp4-Tsc1^−/− transgenic mice by examining p-mTOR and ribosomal protein S6K1, the downstream target of mTORC1. Relative to the WT mice, p-mTOR (Ser2448) and p-S6K1 (Thr389) signal intensity increased markedly in the BATs of Fabp4-Tsc1^−/− mice (knockout [KO]) (Fig. 1A). Further examination by immunohistochemistry showed that immunoreactivity of p-S6 (Ser235/236), the downstream signal of S6K1, increased significantly in brown adipocytes of Fabp4-Tsc1^−/− mice relative to the WT littermates (Fig. 1C). All of these observations demonstrate the activation of mTORC1 signaling in BAT of Fabp4-Tsc1^−/− mice.

Since Fabp4-Tsc1^−/− mice died prematurely within 48 h after birth (26), our capability to analyze the development and function of white adipocytes was limited. We thus focused on the BAT, which was already developed perinatally. As shown in Fig. 2A, cells resembling white adipocytes with more lipid vacuoles increased markedly in the BAT derived from Fabp4-Tsc1^−/− mice as compared with the WT littermates. These vacuolar structures were further validated to be lipid droplets by Oil Red O staining (Fig. 2B). The area and diameter of Oil Red O–positive lipid droplets were significantly increased in the BAT derived from Fabp4-Tsc1^−/− mice as compared with the WT littermates (Fig. 2C). These results suggest a phenotypic conversion of brown adipocytes into cells with morphological property of white adipocytes in the Fabp4-Tsc1^−/− mice.

As the major source for mammalian nonshivering thermogenesis, brown adipocytes contain abundant mitochondria. We thus examined the alteration of mitochondrial structure of BAT using the electron microscopy. In the adipocytes from WT mice, large and electron lucid mitochondria with well-developed cristae are observed, while cells from Fabp4-Tsc1^−/− mice revealed mitochondrial swelling and cristae rupture (Fig. 2D). The diameter of mitochondria increased significantly in Fabp4-Tsc1^−/− mice in comparison with WT littermates (Fig. 2E). Consistent with this observation, content of relative mtDNA decreased markedly in Fabp4-Tsc1^−/− mice relative to WT animals (Fig. 2F). These observations indicate the abnormal mitochondrial structure and reduced mtDNA content in Fabp4-Tsc1^−/− mice.

We next examined the gene expression profile in the BAT from Fabp4-Tsc1^−/− mice. With one exception, UCP2, all examined brown adipocyte markers such as UCP1, UCP3, and PGC1α decreased significantly by 62 ± 5, 41 ± 12, and 68 ± 7% in the BAT from Fabp4-Tsc1^−/− mice relative to the WT littermates, respectively (Fig. 3A). Meanwhile, markers of white adipocytes (31), such as Agt, phosphoserine aminotransferase (Psat), and endothelian receptor type A (Ednra), increased markedly by 90 ± 24.9, 50 ± 16, and 23 ± 8%, respectively (Fig. 3B). To further validate the changes at the protein levels, the typical brown adipocyte marker UCP1 and white adipocyte marker Agt were examined by Western blotting. As shown in Fig. 3C, expression of UCP1 protein was significantly decreased by 56 ± 15%, while Agt increased by 225 ± 42% in the BAT from Fabp4-Tsc1^−/− mice relative to the WT littermates.

To further explore the effect of TSC1-mTORC1 signaling in the brown-to-white adipocyte phenotypic conversion, we infected the cultured brown fat preadipocytes derived from the Tsc1^loxp/loxp mice with Cre adenovirus. As shown in Fig. 4A, reduction of TSC1 activated the mTORC1 signaling in cultured brown adipocytes. Consistent with the observations in Fabp4-Tsc1^−/− mice, brown adipocyte markers such as UCP1 and UCP3 decreased by 88 ± 3 and 73 ± 10%, respectively, while UCP2 remained unaltered in comparison with control cells infected with GFP adenovirus (Ad) (Fig. 4B). Contradicting the in vivo observations, PGC1α increased modestly. In contrast, white adipocyte markers Agt, Psat, and Ednra increased by 327 ± 13, 183 ± 19, 287 ± 45%, respectively (Fig. 4C). These results further demonstrate that activation of mTORC1 signaling by removal of Tsc1 gene induces the brown-to-white adipocyte phenotypic switch in vitro.

In order to investigate whether the effect of mTORC1 activation on the brown-to-white adipocyte phenotypic conversion is reversible, we used rapamycin, a selective inhibitor of mTORC1 signaling. Pregnant mice were randomly divided into two groups: control and treatment. At the near-term pregnancy, pregnant mice were intramuscularly injected with rapamycin at a dose of 1 mg/kg body weight/day or vehicle for 3 consecutive days before the birth of pups. As shown in Fig. 5A, activation of mTORC1 signaling measured by increase in the phosphorylation of S6K1 (Thr389) and S6 (Ser235/236) was detected in the BAT from Fabp4-Tsc1^−/− mice treated with control vehicle, whereas rapamycin treatment completely reversed the increase in the mTORC1 signaling. Furthermore, both the decrease in the brown adipocyte markers UCP1, UCP3, and PGC1α and the increase in the white adipocyte markers Agt, Psat, and Ednra in the BAT of Fabp4-Tsc1^−/− mice were reversed by maternal administration of rapamycin (Fig. 5B and C).

To confirm the specific effect of Tsc1-mTORC1 signaling on BAT, we then infected the primary brown adipocytes isolated from Tsc1^lox/lox mice with Ad-Cre and Ad-GFP. Cells were then treated with rapamycin to block the activation of mTORC1 signaling induced by deletion of Tsc1 (Fig. 5D). As shown in Fig. 5E and F, treatment of cells with rapamycin reversed the downregulation of BAT markers and the upregulation of WAT markers.

These data confirm that rapamycin is able to rescue the brown-to-white adipocyte phenotypic switch induced by deletion of Tsc1 gene in the BAT both in mice and cultured brown adipocytes.

We next investigated the effect of mTORC1 activation on the transcriptional factors related to the fate determination of BAT and WAT. The mRNA level of FoxC2, which decides the fate of BAT, decreased in the BAT from Fabp4-Tsc1^−/− mice, while treatment with rapamycin attenuated the downregulation of FoxC2 (Fig. 6A). In contrast, mRNA levels of Rb, p107, and RIP140, which inhibit the differentiation of BAT, were significantly upregulated relative to WT littermates (Fig. 6A). Again, rapamycin reversed the increase in the mRNA levels of Rb (Fig. 6A). Similar effects were observed in cultured brown adipocytes in which mTORC1 signaling was activated by removal of Tsc1 gene using the Cre recombinase. In comparison with control cells infected with GFP Ad, removal of the Tsc1 gene in brown adipocytes resulted in a significant decrease in FoxC2 and a marked increase in Rb, p107, and RIP140 mRNA levels (Fig. 6B). Removal of Tsc1 by Cre recombinase activated the mTORC1 signaling as measured by increase in the phosphorylation of mTOR, S6K1, and S6 (Fig. 4A). These results further demonstrate that activation of mTORC1 signaling by deletion of Tsc1 gene renders the brown adipocytes more prone to convert into white adipocytes.

Our present study demonstrates that TSC1-mTORC1 signaling regulates the conversion of brown adipocytes into cells with morphological and gene properties of white adipocytes. This conclusion is supported by the following observations: 1) deletion of TSC1 enhanced mTORC1 activity in BAT of Fabp4-Tsc1^−/− mice; 2) cells with morphological properties of white adipocytes increased markedly in the BAT from Fabp4-Tsc1^−/− mice; 3) white adipocyte typical transcripts increased significantly, whereas brown adipocyte typical genes reduced markedly in the BAT of Fabp4-Tsc1^−/− mice; 4) deletion of Tsc1 gene in cultured brown adipocytes induced the conversion into cells with gene properties of white adipocytes; 5) rapamycin rescued the brown-to-white adipocyte phenotypic switch both in the BAT of FABP4-Tsc1^−/− mice and in cultured brown adipocytes; and 6) activation of mTORC1 signaling altered the mRNA levels of FoxC2, Rb, p107, and RIP140 both in mice and cultured brown adipocytes, an effect reversible by treatment with rapamycin.

Evidence suggesting the interconversion between brown and white adipocytes has been emerging. Because of its potential as a promising strategy to counteract obesity by promoting energy expenditure, the conversion of white to brown-like adipocytes has been the focus of studies. Typically, browning involves the appearance of ML adipocyte clusters that are dispersed among unilocular WAT. The browning phenomenon varies among WAT depots and is strongly influenced by synergy between signaling and transcription factors that vary depending upon the environmental conditions. Whether brown adipocytes can directly convert into white adipocytes remain under debate. In an early tracing study, no direct switch from predominantly BAT to predominantly WAT in mice was observed during development (32). In contrast, our present studies provide evidence supporting the phenotypic conversion of brown adipocytes into cells with morphological and gene properties of white fat cells. Several previous reports also support the potential conversion from brown to white adipocytes. Studies in humans have found detectable BAT in up to 96% of healthy young adults but less often in older or obese subjects, suggesting that human adipocytes might convert from a brown into a white adipocyte phenotype during aging. Using immunoelectron microscopic analysis to trace the alteration of UCP in mitochondria, convertible adipose tissue was first identified in mice by Loncar (33). In response to cold stress, cells with morphological properties of brown adipocytes appear in inguinal fat mass. Further examination demonstrates that these brownlike adipocytes regain the properties of white adipocytes upon rewarming the animal to 28°C. Perhaps the most convincing evidence comes from the study by Rosenwald et al. (34). Using the lineage tracing approach in Ucp1-GFP mice, they demonstrate that cold-induced brown like cells are converted into cells with the morphology and gene expression pattern of white adipocytes. Moreover, these white-typical adipocytes can convert into brownlike adipocytes on additional cold stimulation. All of these studies suggest the interconversion between brown and white adipocytes. Our studies further support this concept. However, it is worth of noting that our study cannot exclude the contribution of mTOR signaling in other metabolically critical organs because FABP4 is not exclusively expressed in BAT and WAT (26,35,36). While our in vitro studies suggest that mTOR signaling may determines the interconversion between brown and white adipocytes, a portion of the whitening of brown fat in Fabp4-Tsc1^−/− transgenic mice could result from increased mTORC1 function in tissue other than adipose tissues such as brain, liver, and skeletal muscles.

Decrease of BAT in obese subjects indicates that long-term excess energy may be accountable for the loss of brown adipocytes, which are likely converted into the white fat cells based on previous observations described above. The molecular mechanism by which excess energy signals brown adipocytes to initiate the conversion remains unknown. The current study supports the concept that mTORC1 signaling is critical for the conversion of brown adipocytes into cells with white adipocyte properties. Activation of mTORC1 signaling may signal the brown adipocytes the oversupply of energy and therefore initiate the whitening of the BAT. This concept is consistent with previous studies demonstrating that mTORC1 can sense nutrient availability to regulate the differentiation of white adipocytes. The mTORC1 signaling has been reported to specifically regulate the activity of peroxisome proliferator–activated receptor-β, which is essential for the adipogenic gene expression program and thus critical for both initiating and maintaining adipogenesis. In contrast, inhibition of mTORC1 with rapamycin significantly reduces the expression of most adipocyte marker genes and the accumulation of intracellular lipid in white adipocytes.

Appearance of white adipocytes in the brown fat depots may also result from the switch of the progenitor cells to differentiate into white adipocytes. Several recent reports have identified adipocyte progenitor cells in adult adipose tissues that can give rise to adipocytes with either brown or white characteristics (37–40). Our data cannot exclude the contribution of precursor cells to the formation of white adipocytes in BAT. The appearance of large amount of cells with morphological and gene properties of white adipocytes supports the concept based on previous studies that differentiation of additional white adipocytes de novo from precursors may occur when the interconversion capacity is insufficient.

In summary, our study demonstrates that activation of mTORC1 signaling stimulates the conversion of brown to white adipocytes. If inhibition of mTORC1 signaling is able to block the conversion of brown to white adipocytes, it might be a promising strategy to target the mTORC1 signaling to increase the number of lipid-burning cells to counteract obesity by promoting energy expenditure.

Funding. This work was supported by grants from the National Natural Science Foundation of China (81330010, 81390354, 81370962, 81170795, and 81030012) and American Diabetes Associationhttp://dx.doi.org/10.13039/100001440 grant 1-13-BS-225.

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. X.X., Y.L., and W.Z. contributed to the design of the experiments, interpretation of data, and the writing of the manuscript; performed the experiments; and approved the final version of the manuscript. H.L., H.T., F.Y., Y.X., and J.Z. performed the experiments, analyzed data, and approved the final version of the manuscript. W.Z. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.