Liver X receptor β controls thyroid hormone feedback in the brain and regulates browning of subcutaneous white adipose tissue

Significance

Liver X receptors (LXRs) are nuclear receptors vital in controlling lipid and glucose metabolism through direct or indirect mechanisms. Here, we demonstrate for the first time, to our knowledge, that, in addition to changing the activity of classic brown adipose tissue, LXRs, especially LXRβ, are able to regulate the browning of white adipose tissue (WAT), activation of mitochondria, and increased energy expenditure. Deletion of LXRs removed their suppressive effect on thyrotropin releasing hormone expression in brain, thus activating the hypothalamic–pituitary–thyroid (HPT) axis and inducing the browning of WAT. Altogether, the data confirm the importance on LXRs in ameliorating obesity through a previously unidentified mechanism (i.e., increasing browning of WAT and energy dissipation).

Abstract

The recent discovery of browning of white adipose tissue (WAT) has raised great research interest because of its significant potential in counteracting obesity and type 2 diabetes. Browning is the result of the induction in WAT of a newly discovered type of adipocyte, the beige cell. When mice are exposed to cold or several kinds of hormones or treatments with chemicals, specific depots of WAT undergo a browning process, characterized by highly activated mitochondria and increased heat production and energy expenditure. However, the mechanisms underlying browning are still poorly understood. Liver X receptors (LXRs) are one class of nuclear receptors, which play a vital role in regulating cholesterol, triglyceride, and glucose metabolism. Following our previous finding that LXRs serve as repressors of uncoupling protein-1 (UCP1) in classic brown adipose tissue in female mice, we found that LXRs, especially LXRβ, also repress the browning process of subcutaneous adipose tissue (SAT) in male rodents fed a normal diet. Depletion of LXRs activated thyroid-stimulating hormone (TSH)-releasing hormone (TRH)-positive neurons in the paraventricular nucleus area of the hypothalamus and thus stimulated secretion of TSH from the pituitary. Consequently, production of thyroid hormones in the thyroid gland and circulating thyroid hormone level were increased. Moreover, the activity of thyroid signaling in SAT was markedly increased. Together, our findings have uncovered the basis of increased energy expenditure in male LXR knockout mice and provided support for targeting LXRs in treatment of obesity.

The metabolic syndrome is a constellation of related disorders (obesity, insulin resistance, dyslipidemia, fatty liver, hypertension, and atherosclerosis) (1–3). Of note, obesity, which is attributable to the chronic imbalance between energy intake and energy expenditure, is an epidemic, for which there is no effective therapy (4). A major challenge in battling this epidemic is to identify a target that can either decrease energy intake or increase energy expenditure. There is great research interest in brown adipose tissue (BAT), which is specialized for the dissipation of chemical energy in the form of heat (4, 5). BAT defends mammals against hypothermia, obesity, and type 2 diabetes; however, adult humans lack this thermogenic interscapular organ (6). Recently, studies have demonstrated that adult humans harbor a distinct cold-inducible depot of brown adipocytes that are expressed in WAT in the supraclavicular, paraaortic, and suprarenal regions (7–9). These cells, called beige or brite fat cells because of their beige color, undergo a browning process following cold stimulus and share some molecular, histologic and functional characteristics with beige adipocytes found in the subcutaneous white adipose tissue (SAT) of mice (7, 10, 11). The discovery of beige cells has raised clinical interest in the potential of these cells in the treatment of obesity.

Uncoupling protein-1 (UCP1), which dissipates the mitochondrial electrochemical gradient which is the key for ATP formation, mediates the thermogenic activity of brown and beige adipocytes (12). Cell death-inducing DNA fragmentation factor α-like effector A (CIDEA), a member of a novel family of proapoptotic proteins, is expressed abundantly in both BAT and beige cells (10). Despite the similarity in thermogenic function, multiple lines of evidence indicate that they have unique expression profiles and distinct characters that likely contribute to their tissue-specific functions (4). Several genes such as Prdm16 (PR domain containing 16), Tbx1 (T-box 1), Tmem26 (transmembrane 26), pRb (protein retinoblastoma), Foxc2 (forkhead box protein C2), and CD137 [also know as TNFRSF9 (tumor necrosis factor receptor superfamily, member 9)] are preferentially expressed in beige adipocytes and ablation of some of these genes or of beige cells make mice more prone to develop obesity and metabolic dysfunction (10, 13, 14). Although recent evidence suggested that beige and brown adipocytes are likely to function in the maintenance of energy balance and thermogenesis, the safety of therapeutic stimulation of the browning process in treatment of obesity has not been established partly because the mechanisms underlying this process are not understood.

Liver X receptors (LXRs) α and β are two members of the nuclear receptor family involved in multiple metabolic pathways, including insulin sensitivity; metabolism of glucose, lipid, and cholesterol; and energy expenditure (15). Our team has shown that LXR participates in regulation of key genes of energy pathways in the BAT in female rodents (16, 17). Genetic knock out of LXRs in both male and female mice provided them protection from diet-induced obesity, which was consistent with findings observed by other research groups using different LXR knockout mice (18, 19). These phenomena were explained by an ectopic expression of UCP1 in visceral white adipose and skeletal muscles or increased fat oxidation (18, 20). However, we have speculated that alteration of the browning process in LXR knockout mice could contribute to the metabolic protection against obesity and type 2 diabetes. We now present the evidence that this is the case.

We found that depletion of LXRs in male mice reduced fat content and body weight. This finding was associated with an increased browning of SAT and consequently increased energy expenditure. Meanwhile, activated TSH-releasing hormone (TRH) signaling in the paraventricular nucleus (PVN) area of the hypothalamus in LXRαβ^−/− mice increased the activity of the hypothalamic–pituitary–thyroid (HPT) axis, which ultimately led to the enhanced browning of SAT.

Results

Reduced Body Weight and Fat Content in LXRαβ^−/− Male Mice Attributable to Increased Browning of SAT.

In a previous study, we have reported that female LXRαβ^−/− mice were resistant to high-fat diet-induced obesity because of an increase in metabolic rate (16). We now report that in male mice, ablation of LXRβ or both LXRα and LXRβ reduced body weight (Fig. 1A), fat mass (Fig. 1B), SAT content (Fig. 1D), and gonadal adipose tissue content (Fig. 1E), while increasing lean mass (Fig. 1C). There was no significant change in the weight of BAT in mice on a normal chow diet (Fig. 1F). SAT was recently found to be a specific fat depot in which browning (increased UCP1 and heat production) could be induced (10). Histological examination of subcutaneous fat of LXRαβ^−/− mice fat showed a marked increase in UCP1-positive adipocytes (Fig. 2A). In LXRβ^−/− mice, there was also more scattered UCP1⁺ adipocytes in SAT than in WT mice, but these were very rare in LXRα^−/− mice (Fig. 2A, Lower). Quantitative PCR (qPCR) results confirmed that both thermogenic genes like UCP1 and Cidea were significantly increased in LXRαβ^−/− and LXRβ^−/− mice. However, surprisingly, there was no induction of PGC1α or PPARγ (Fig. 2B). Because LXRαβ^−/− mice showed the most robust browning phenotype, the following work focused on this genotype.

Fig. 1.

Reduced body weight and fat mass in LXRαβ^−/− mice. The body weight (A), fat mass (B), and SAT weight (D) were reduced in both LXRβ^−/− and LXRαβ^−/− mice compared with WT mice, whereas lean mass (C), conversely, was increased. LXRαβ^−/− mice showed more profound change than LXRβ^−/− mice. The gonadal adipose tissue weight tended to be decreased in LXRα^−/− and LXRβ^−/− mice, whereas LXRαβ^−/− showed a statistically significant decrease (E). There was no difference in BAT weight (F). *P < 0.05, **P < 0.01, ***P < 0.001: LXRαβ^−/− vs. WT; ##P < 0.01: LXRβ^−/− vs. WT.

Fig. 2.

Significant browning of SAT in LXRαβ^−/− mice. The H&E staining (A, upper row) of SAT in four different genotype mice. Both adipocyte sizes were smaller in LXRβ^−/− and LXRαβ^−/− mice than WT mice, along with multilocular lipid droplets appearing in LXRαβ^−/− mouse SAT. UCP1 immunostaining (A, lower row) in SAT was relatively stronger in LXRαβ^−/− mice, indicating an increase of mitochondrial uncoupling function. LN, lymph node. The red arrow illustrates the patchy distribution staining of UCP1. All pictures were 20× magnification. (B) The common thermogenesis gene expression in SAT in LXR knockout mice. Compared with WT mice, LXRαβ^−/− mice expressed the highest level of UCP1 and Cidea without a significant change in PGC1α or PPARγ. These parameters were also increased in LXRβ^−/− mouse SAT, although still less than LXRαβ^−/− mice. (C) Beige cell-specific gene expression in SAT of LXRαβ^−/− mice. Almost all of the beige-specific genes were increased in LXRαβ^−/− mice. (D) No differences in thermogenesis genes in interscapular BAT between different genotypes. *P < 0.05, **P < 0.01, ***P < 0.001: LXRαβ^−/− vs. WT; #P < 0.05: LXRβ^−/− vs. WT.

Recent characterization of the brown adipocyte-like cells in WAT depots has revealed that these cells (beige cells) are a novel cell type, distinct from classic brown adipocytes (11). Indeed, in LXRαβ^−/− mice, there was abundant expression of several beige cell-specific genes (Tbx1, CD137, Prdm16, and pRb) in SAT (Fig. 2C). Surprisingly, in BAT itself, there was no increase in UCP1 expression, although Cidea was slightly increased (Fig. 2D). This finding indicated that the increased energy expenditure in male LXRαβ^−/− mice (18, 21) was not attributable to BAT but instead to the emergence of beige cells in SAT.

Facilitated Action of Thyroid Hormone in SAT of LXRαβ^−/− Mice.

Thyroid hormone (TH) plays a key role in energy metabolism acting on peripheral tissues (BAT, WAT, and skeletal muscle), and it works synergistically with norepinephrine to generate a full thermogenic response (22). We have shown that TH receptors (TRs) manifest numerous similarities functionally or structurally with LXRs in both nutritional and developmental processes (23, 24). However, little is known about the cross-talk between these two receptor subfamilies with regard to the promotion of adaptive thermogenesis in peripheral tissues. We found that the serum free-triiodothyronine (T3) level in LXRαβ^−/− mice was higher than that in WT mice (Fig. 3A). The serum free-thyroxine (T4) level was comparable between the two genotypes (Fig. 3B). Thus, the ratio of free T3/free T4 was relatively higher in LXRαβ^−/− mice (Fig. 3C), indicating that the activity of TH in the circulatory system was elevated after depletion of LXRs. Not only was the TH level changed, but the genes associated with TH synthesis and transportation were affected in SAT as well. DIO2 is the enzyme which catalyzes the conversion of T4 to T3, and thus increases the active and functional form of TH (T3) (25). In SAT of LXRαβ^−/− mice, there was an approximately twofold increase of DIO2 gene expression (Fig. 3D). Meanwhile, there was an increase in TH transporters Mct8 and Mct10 but not Lat1 and Lat2 (Fig. 3D). Mct8 and Mct10 function as more specific TH transporters than Lat1 and Lat2, which are also in charge of transporting large neutral amino acids and act as secondary transporters for TH (26). TH functions through binding to its receptors and there are two subtypes of TRs, TRα and TRβ. These TRs are both expressed in the cell nuclei of SAT, as shown in Fig. 3E, and the amounts of both TRα and TRβ were increased in LXRαβ^−/− mouse SAT.

Fig. 3.

Facilitated action of TH in SAT of LXRαβ^−/− mice. (A–C) Increased free-T3 hormone level (A) and free-T3/free-T4 ratio (C) in the serum of LXRαβ^−/− mice but not free T4 (B). (D) Increased expression of genes associated with T3 synthesis (DIO2) and TRs (MCT8, MCT10). (E) In SAT, there were more TRα (left column) and TRβ (right column) expression in the cell nuclei in LXRαβ^−/− mice (lower row) than in WT mice (upper row). Inset pictures in the upper row are higher-power magnifications (×40) of TRα or TRβ staining in adipocyte nuclei. All of the pictures were 20× magnification unless specifically mentioned otherwise. *P < 0.05, **P < 0.01: LXRαβ^−/− vs. WT.

Depletion of LXRs Led to a Hyperactive State in Thyroid Gland.

To determine the reason for the increase in TH level in LXRαβ^−/− mice, we prepared sections from thyroid glands (the organs that synthesize and secrete TH) and analyzed their morphology. After quantitative examination of 90 thyroid follicles per genotype, we found that although the mean number of thyrocytes per follicle and whole-follicle areas were similar in the two genotypes, in LXRαβ^−/− mice, there was an increase in the average thyrocyte size and a reduction in colloid-containing areas (Fig. 4 A and B). An increase in size of thyrocytes usually reflects an active form of thyroid epithelial cell with a strong capacity for synthesis and secretion (27). The synthesis of TH is driven by thyroid-stimulating hormone (TSH) (or thyrotropin), which acts through binding to TSH receptor (TSHR) in thyroid gland (28). In LXRαβ^−/− mice, there was a significant increase in staining of TSHR in thyrocytes (Fig. 4 C and D), which partially explained the hyperactive state of thyroid gland in LXRαβ^−/− mice.

Fig. 4.

Histology of thyroid gland and TSHR expression in WT and LXRαβ^−/− mice. (A) Low-power (×20) (Upper) and high-power (×40) (Lower) view of H&E-stained sections from thyroid glands of 12-wk-old WT and LXRαβ^−/− mice. (B) Morphometric analysis of thyroid gland sections showing the number of cells per follicle, whole-follicle area, and colloid-containing area and average thyrocyte size (see Materials and Methods for details). The data were expressed relative to WT mice. (C) Low-power (×20) (Upper) and high-power (×40) (Lower) views of TSHR-stained sections from thyroid glands of 12-wk-old WT and LXRαβ^−/− mice. Note that the cytoplasmic signaling of TSHR staining was stronger in LXRαβ^−/− mice. (D) Semiquantification of TSHR staining in the thyroid gland of C. *P < 0.05, **P < 0.01: LXRαβ^−/− vs. WT.

The increased TH level was attributable to constitutive difference in the expression of the genes that were transcriptionally up-regulated by TSH and thyroid transcriptional factor 1 (Ttf1). Although there were no differences in expression of thyroperoxidase (TPO) (29) and thyroglobulin (Tg) (29), genes involved in TH synthesis, there was an increase in Ttf-1, Na⁺/I^– symporter (NIS) (30), and iodotyrosine deiodinase (Dehal1) (31) (Fig. 5A). As in SAT, Mct8 and Mct10 but not Lat1 and Lat2 were markedly increased in the thyroid gland (Fig. 5B).

Fig. 5.

Expression of genes associated with TH metabolism in the thyroid gland. (A) Expression of transcriptional factors (Ttf1) and TH synthesis genes in WT and LXRαβ^−/− mouse thyroid. (B) TH transporters Mct8 and Mct10 were increased, but not Lat1 and Lat2, in LXRαβ^−/− thyroid. *P < 0.05, **P < 0.01: LXRαβ^−/− vs. WT.

Depletion of LXRs Increased TRH Expression in the PVN Area, Which Mobilized the HPT Axis.

The PVN area in the hypothalamus controls the activity of the thyroid gland through the classic HPT axis. TRH-expressing neurons in the PVN area project to the portal system. TRH stimulates release of TSH from the anterior pituitary. TSH in turn binds to its receptor TSHR on the thyroid gland, where it stimulates TH synthesis (28). Multiple studies indicate that LXRs play a vital role in the development of brain and are involved in several severe neurodegenerative diseases (24, 32), such as Parkinson’s disease (33). Here, we discovered that the TSH staining in endocrine cells of anterior pituitary was increased in LXRαβ^−/− mice (Fig. 6A). In addition, there was more abundant TRH expression in the PVN area of LXRαβ^−/− than in WT mice, both at the protein level (Fig. 6B) and mRNA level (Fig. 6C).

Fig. 6.

Increase of thyroid signaling in the central nervous system. (A) TSH staining in the anterior pituitary. Graph on the right shows the semiquantification of TSH staining. (B) TRH expression in the PVN area (yellow line-labeled area) of hypothalamus. Inset pictures (red arrow) show the positive TRH staining in intracellular cytoplasm. III, the third ventricle. (C) Relative expression of TRH mRNA level in the hypothalamus area. *P < 0.05: LXRαβ^−/− vs. WT. (D) Abundant TRα and TRβ expression in the WT PVN area, which was almost lost in LXRαβ^−/− mice, especially for TRβ. The inset pictures were under higher power (×40), indicating the positive nuclear staining (red arrow) and negative staining (black arrow). In the LXRαβ^−/− PVN area, there were some nonnuclear and nonspecific TRα stainings. (E) Predominant LXRβ expression in the PVN area in WT compared with LXRα, which both disappeared in LXRαβ^−/− mice. Inset pictures were under higher power (×40), indicating the positive nuclear staining (red arrow) and negative staining (black arrow). All of the pictures were 20× magnification unless specifically mentioned otherwise.

Under normal conditions, TH signaling is controlled via a classic negative-feedback pathway at the level of the anterior pituitary and the hypothalamus through binding to TRs (28). The high expression of TRH and TSH in the face of high TH levels is an unexpected paradox. The surprising explanation for this is the following: TRα and TRβ were very markedly reduced in the PVN area of LXRαβ^−/− mice (Fig. 6D). Thus, depletion of LXRs caused the brain to escape from the control of negative feedback of TH. This abnormality of feedback loop could partially explain the continuously elevated TH level in LXRαβ^−/− mice. In addition, the expression of LXRβ in the PVN area was higher than that of LXRα in WT mice (Fig. 6E). The results reveal a very unexpected role for LXRβ in the feedback regulation of TH and explain why LXRβ^−/− mice were more sensitive to TH stimulus and exhibited more browning of WAT than LXRα^−/− mice (Fig. 7 A and B).

Fig. 7.

Schematic diagram of actions of LXRβ in controlling of TH feedback in the brain and browning of SAT. (A) TRH expressed by the neurons in the PVN area of the hypothalamus stimulates release of TSH from the anterior pituitary, which in turn stimulates TH synthesis at the thyroid gland. T4/T3 in the circulation enter their target organs such as SAT and activate the browning process. In reverse, the circulating THs negatively regulate their own production through targeting both pituitary and hypothalamus. LXRβ is able to transcriptionally inhibit the expression of TRH in the PVN area. (B) Genetic depletion of LXRs releases their transcriptional suppression on TRH and breaks the negative-feedback loop because of the lack of TRs in the PVN area, thus promoting TSH secretion in the pituitary and activating the synthesis of TH, which eventually increases the browning of SAT.

Discussion

Since 2002, our team has reported that activation of LXRs changes the expression patterns of key genes in the energy pathway in BAT (34). By using genome-wide gene expression profiling analysis, we found that treatment with synthetic LXR agonist of male mice down-regulated the expression of UCP1, as well as cytochrome c and mitochondrial ribosomal proteins, in classic BAT. Our findings were further confirmed by other research groups: Kalaany et al. demonstrated that in vivo knock out of both LXR genes prevented male mice from high fat and cholesterol diet-induced obesity (18). The phenotype was explained as abnormally increased energy dissipation resulting from uncoupled oxidative phosphorylation and ectopic expression of uncoupling proteins in muscle and white adipose other than in classic BAT. Later, Wang et al. showed that LXRα was a direct transcriptional inhibitor of β-adrenergic receptor-mediated, cAMP-dependent UCP1 gene expression, through LXRα’s binding to the critical enhancer region of UCP1 promoter (19). In line with the work of Kalaany et al., we delineated the specific roles of LXRα and LXRβ in regulation of energy homeostasis in female mice and concluded that both receptors could play crucial roles (16).

In the present study, we have shown that activation of the browning process in SAT contributes to increased energy dissipation in male LXRαβ^−/− mice. Adult human BAT, which is unlike the classic brown adipocytes found in the interscapular regions of rodents and human infants, shares many molecular, histologic, and functional characteristics with cold-inducible beige cells found in the SAT of rodents (6, 14). Because of these similarities, cold-induced (browning process) or hormone-induced recruitment and activation of rodent beige adipocytes was targeted for potential antidiabetic actions and the treatment of obesity. We have found that the LXRβ receptor (as shown in Figs. 2 A and B and 6E) functions as systemic repressor for browning process.

The classic HPT axis controls the concentration of THs in the bloodstream. Previous studies helped to establish the pivotal role of TRH neuron in the PVN area in the neuroendocrine regulation of the HPT axis (28). Interaction between LXR and TR has been described not only in the periphery but also in the central nervous system. Ghaddab-Zroud et al. reported that at the hypothalamic level, activation of LXR had a repressive action on TRH promoters (35). When both LXRs were knocked down by shRNA, the repression on TRH promoter transcriptional activity was lost (with or without ligand), resulting in an activation of transcription. Accordingly, in the PVN area of LXRαβ^−/− mouse brain, we observed that both TRH protein and gene levels were increased simultaneously because of the release from LXR’s transcriptional suppression, as shown in Fig. 6 B and C, which in turn promoted TSH expression in anterior pituitary (Fig. 6A) and activated synthetic function of the thyroid gland. The sizes of thyrocytes in LXRαβ^−/− mice were relatively bigger than in WT, reflecting a higher secretive function. Moreover, we found that TSH receptor, which mediates the effects of TSH, and several other genes regulated by TSH or involved in TH synthesis, such as Ttf-1, NIS, and Dehal1, were all increased (26). Of the studied TH transporters in the thyroid gland, Mct8 was the most abundantly expressed, followed by Mct10 (mean mRNA level 14% of Mct8), Lat1 (mean mRNA level 0.9% of Mct8), and Lat2 (mean mRNA level 0.5% of Mct8). Of these four transporters, Mct8 had the highest affinity for iodothyronines and was able to stimulate the efflux of TH, followed by Mct10 (26). In fact, the up-regulation of these two transporters in LXRαβ^−/− mice further confirmed the hyperactivity of the thyroid gland.

Local conversion of T4 to T3, by DIO2, provides negative feedback through binding to TRs at the level of thyrotrophs in the pituitary and tanycytes in the hypothalamus. This results in reduction in TRH and TSH secretion in response to adequate tissue level of TH (28). Unlike the cross-talk between LXR and TR in periphery such as SAT, loss of both LXRs in brain diminished the expression of TRs. Unlike what we have reported in cortical layers in the brain, where TRβ was partially capable of compensating for LXRβ’s function during postnatal brain development (32), the existence of LXRs seemed indispensable for maintaining TR’s expression and normal function in the hypothalamus.

In conclusion, our characterization of LXRαβ^−/− knockout mice has revealed a coordination between LXR regulation of the HPT axis and the browning of peripheral SAT. In the absence of LXR, even under ambient temperature, negative feedback of TH on the HPT axis was repressed, resulting in the browning of SAT and consequent energy dissipation. Included in the mechanism of loss of the feedback regulation on the HPT axis in LXRαβ^−/− mice was a loss of transcriptional suppression on the TRH gene by TH. Although loss of LXRβ leads to resetting of the HPT axis, it is unclear whether or not it will be possible to target LXRβ to treat obesity. LXRβ has many beneficial effects in the brain (24, 32, 33), and it does not appear to be beneficial to block LXRβ. However, some of the effects of loss of LXRβ on browning of SAT may be independent of the HPT axis. If loss of LXRβ induces DIO2 in the periphery, it would increase T3 and TR in SAT and up-regulate the effects of TH. If this is the case, a peripherally acting WAT-specific LXRβ antagonist could be valuable. To delineate the particular task of each cell type in the LXR-TH axis, LXRβ tissue-specific knockout mice will be invaluable and will inform us on the use of LXR as a target in the field of metabolic syndrome.

Materials and Methods

Animals.

Twelve-week-old male C57BL/6 WT, LXRα^−/−, LXRβ^−/−, and LXRαβ^−/− knockout mice as previously described (16) were housed under standard conditions and ambient temperature (23 °C) with free access to water and chow diet throughout the study. At the end of the study, animals were deeply euthanized with CO₂, and tissues were quickly harvested, weighed, and fixed or frozen for further study. The animal care and use program complied with the standards and recommendations set forth in the National Research Council’s 1996 Guide for the Care and Use of Laboratory Animals (36) and were in accordance with the University of Houston guidelines on the use of laboratory animals.

Body Composition Analysis.

One day before sacrifice, the mice were placed in a clear plastic holder without anesthesia or sedation and inserted into the EchoMRI device (Echo Medical Systems) to measure total body fat and lean mass in triplicate. The percentages of fat and lean mass were calculated as total mass divided by body weight.

Gene Expression Analysis.

BAT or SAT (50–100 mg) were used for total RNA extraction by using the Qiagen RNeasy lipid tissue mini kit following manufacturer’s protocol. For thyroid gland and pituitary RNA extraction, two different samples were mixed together for enrichment purposes and considered as one individual. Hypothalamus area was dissected freshly using microsurgical tweezer under microscope. cDNA was synthesized using random hexamer primer and a revertAid first-strand cDNA synthesis kit (Invitrogen). qPCR was performed using iTaq universal SYBR green supermix (Bio-Rad), and the primers applied are listed in Table S1.

Table S1.

qPCR primers

Genes	Forward (5′→3′)	Reverse (5′→3′)
UCP1	GATCCAAGGTGAAGGCCAGG	GTTGACAAGCTTTCTGTGG
Cidea	TGCTCTTCTGTATCGCCCAGT	GCCGTGTTAAGGAATCTGCTG
PGC1α	CGGAAATCATATCCAACCAG	TGAGAACCGCTAGCAAGTTTG
PPARγ	TGCAACCGTTACCCCATAGAA	TGCAACCGTTACCCCATAGAA
Tbx1	GGCAGGCAGACGAATGTTC	TTGTCATCTACGGGCACAAAG
Tmem26	ACCCTGTCATCCCACAGAG	TGTTTGGTGGAGTCCTAAGGTC
CD137	CGTGCAGAACTCCTGTGATAAC	GTCCACCTATGCTGGAGAAGG
Prdm16	CTTCTCCGAGATCCGAAACTTC	GATCTCAGGCCGTTTGTCCAT
pRb	CTGGCCTGTGCTCTTGAAGTT	CCACGGGAAGGACAAATCTGT
Foxc2	TCCATGGGAACCTTCTTCGA	GATCTCAAACTGAGCTGCGGATA
DIO2	ATGGGACTCCTCAGCGTAGA	GCACAGGCAAAGTCAAGAAG
TPO	TGCAACCGTTACCCCATAGAA	GCGGAGGAGCGGTAGAAAG
Tg	GCCCACCATCTGTGGACTTC	CATTCCCCTTTCACATCCCA
NIS	GTGGGCCAGTTGCTCAATTC	GTGCGTAGATCACGATGCCA
TTF-1	GGTGCTGGGACTGGGATGT	TCAAGATGTCAGACACTGAGAACG
Dehal1	ACACCGCCCCAGTTCTGAT	ACCGTCACTAGCCCTGCATT
Mct8	GTGCTCTTGGTGTGCATTGG	CCGAAGTCCCGGCATAGG
Mct10	GGCCGCATTGCTGACTATTT	CAATGGGCGCCATGATAGA
Lat1	CTGCTGACACCTGTGCCATC	GGCTTCTTGAATCGGAGCC
Lat2	CCAGTGTGTTGGCCATGATC	TGCAACCGTTACCCCATAGAA
TRH	GGAGGAAGGTGCTGTGACTC	TGTCCCCCTCATCTGACCAT

Immunohistochemistry.

All mice were transcardially perfused with heparinized saline followed by 4% (wt/vol) paraformaldehyde in 0.1 M PBS (pH 7.4). SAT, thyroid gland, pituitary, and brain were dissected and postfixed in the same fixative solution overnight at 4 °C. After fixation, tissues were transferred to 70% (vol/vol) ethanol and processed for paraffin sectioning (5-μM sections). Paraffin sections were deparaffinized in xylene, rehydrated, and processed for antigen retrieval with 10 mM citrate buffer (pH 6.0) for 10 min in a PT Module (Thermo Scientific). Cooled sections were then incubated with 3% H₂O₂ in PBS for 30 min at room temperature to quench endogenous peroxidase. Unspecific binding was blocked with incubation of 3% (wt/vol) BSA (Sigma-Aldrich) in PBS for 1 h at room temperature. For brain sections, a biotin blocking system (DAKO) was used to block endogenous biotin signaling. Then, these sections were subjected to immunohistochemistry staining at 4 °C overnight with different antibodies diluted in 1% BSA as follows: SAT: UCP1 [1:500; ab10983 (Abcam)], TRα [1:300; AP158118PU-N (Acris)], and TRβ [1:200; AP15818PU-N (Acris)]; thyroid gland: TSHR [1:100; HPA026680 (Sigma)]; pituitary: TSH β chain [1:100; 8922-6009 (Abd Serotec)]; and PVN area of the hypothalamus: TRH [1:100; BP5066 (Acris)], TRα [1:100; AP158118PU-N (Acris)], TRβ [1:100; AP15818PU-N (Acris)], LXRα (raised against the N-terminal amino acids 68–82 of mouse LXRα; 1:1,000; produced in J.-Å.G.’s laboratory), and LXRβ [raised against the N-terminal amino acids of 1–17 of mouse LXRβ; 1:1,000 (33)]. After washing, sections were incubated with biotinylated secondary antibody (1:200 dilution) for 1 h at room temperature, and then the Vectastain Avidin-Biotin Complex kit (Vector laboratories) was used according to the manufacturer’s instruction. After washing with PBS, peroxidase activity was visualized with 3,3-Diaminobenzidine staining (Thermo Scientific) and counterstained with Mayer’s hematoxylin (Sigma-Aldrich), dehydrated through an ethanol series to xylene, and mounted with Permount (Fisher Scientific).

Free Serum T3 and T4 Measurement.

Blood was collected, followed by centrifugation at 1,485 × g for 8 min, and serum was kept for further analysis. The levels of free T3 and T4 in sera were measured using Free T3/T4 ELISA kit respectively (ALPCO) according to the manufacturer’s protocol. Briefly, 50 μL of serum samples or reference was pipetted into the assigned wells, followed by T3/T4 enzyme-conjugated solution with 1 h of incubation. The mixture was removed, and the plate was washed several times with water; then, working substrate solution was added to each well for reaction, and after 20 min in the dark, the reaction was stopped by adding 3 M HCl. Absorbance was read at A at 450 nm. The corresponding sample free-T3/T4 concentrations were calculated using mean absorption values from the standard curve.

Histological Morphometry of Thyroid Gland.

Three randomly selected thyroid gland slides stained with H&E from three mice in each group were analyzed by National Institutes of Health (NIH) ImageJ software. Thirty thyroid follicles were measured from each slide, and the following parameters were quantified: colloid-containing area, whole-follicle area, thyrocyte area (whole-follicle area minus colloid area), number of thyrocytes, and average thyrocyte size (thyrocyte area divided by number of thyrocytes). For semiquantification of TSHR and TSH immunohistochemistry staining, three randomly chosen thyroid gland slides stained with TSHR from each genotype were analyzed by NIH ImageJ software as well. Data for TSHR were presented as the ratio of integrated optical density (positive cytoplasmic staining) to selected area, and TSH expression in the pituitary was semiquantified as TSH-positive cells/100 total cells.

Statistical Analysis.

Data were expressed as means ± SD. Statistical comparisons were made using Student’s t test or a one-way ANOVA, followed by a Newman–Keuls posttest. P < 0.05 was considered a significant difference.