Thyroid hormone (TH) has a profound effect on energy metabolism and systemic homeostasis. Adipose tissues are crucial for maintaining whole-body homeostasis; however, whether TH regulates systemic metabolic homeostasis through its action on adipose tissues is unclear. Here, we demonstrate that systemic administration of triiodothyronine (T3), the active form of TH, affects both inguinal white adipose tissue (iWAT) and whole-body metabolism. Taking advantage of the mouse model lacking adipocyte TH receptor (TR) α or TRβ, we show that TRβ is the major TR isoform that mediates T3 action on the expression of genes involved in multiple metabolic pathways in iWAT, including glucose uptake and use, de novo fatty acid synthesis, and both UCP1-dependent and -independent thermogenesis. Moreover, our results indicate that glucose-responsive lipogenic transcription factor in iWAT is regulated by T3, thereby being critically involved in T3-regulated glucose and lipid metabolism and energy dissipation. Mice with adipocyte TRβ deficiency are susceptible to diet-induced obesity and metabolic dysregulation, suggesting that TRβ in adipocytes may be a potential target for metabolic diseases.
How thyroid hormone (TH) achieves its diverse biological activities in the regulation of metabolism is not fully understood.
Whether TH regulates systemic metabolic homeostasis via its action on white adipose tissue is unclear.
Adipocyte TH receptor (TR) β mediates the triiodothyronine effect on multiple metabolic pathways by targeting glucose-responsive lipogenic transcription factor in white adipose tissue; mice lacking adipocyte TRβ are susceptible to high-fat diet–induced metabolic abnormalities.
TRβ in white adipocytes controls intracellular and systemic metabolism and may be a potential target for metabolic diseases.
Introduction
The prevalence of type 2 diabetes (T2D) is increasing worldwide. Insulin resistance is a major pathogenic factor; however, the role of individual peripheral tissues in the pathogenesis of T2D is not fully understood. Although skeletal muscle is the main site of insulin-stimulated glucose uptake, the importance of adipose tissue in insulin action and glucose homeostasis has been recognized (1). Studies on major insulin-responsive glucose transporter 4 (GLUT4) and glucose-responsive lipogenic transcription factor (ChREBP) in adipocytes have greatly advanced our understanding of the deleterious effects of altered adipocyte glucose flux on T2D development and the beneficial effects of increased de novo lipogenesis in adipose tissue on whole-body metabolic function (2–5). Evidence suggests that impaired adipose tissue function can contribute to the risk of developing insulin resistance and T2D. Nevertheless, more efforts are needed to better understand the metabolic regulation in adipocytes.
White adipose tissues (WATs) contain beige adipocytes, which display an oxidative phenotype with energy dissipation mediated by either a UCP1-dependent or -independent mechanism (6,7). Recent studies demonstrated that the interconversion between white and beige adipocytes is accompanied by metabolic reprogramming involving both anabolic and catabolic pathways that control the intracellular fates of glucose and fatty acids (FAs) (8). The interplays between glucose and FA metabolism and substrate cycles are essential for energy dissipation in adipocytes (6,7). Although the browning process has been extensively studied, the underlying mechanisms and consequences are not fully understood. Identification of novel regulators and the interplay between metabolic pathways in adipocytes in different physiological states will generate a more comprehensive picture of browning and underpin the development of novel adipocyte-based strategies for metabolic diseases.
Thyroid hormones (THs) are essential in energy expenditure (EE) and lipid metabolism and exhibit therapeutic potential (9,10). To avoid undesired cardiovascular effects, tissue-selective and/or TH receptor (TR) isoform-specific TH mimetics have been developed and found to offer many metabolic benefits not only on lipid profile and body weight (BW) but also on glycemic control and hepatic fat (11). Evidence from studies of TH mimetics, TRs, and the downstream effectors or cofactors of TRs suggests that the regulation by TH is multilayered and more complex than previously suggested (12–15). Nonredundant functions of TRα and TRβ have been suggested; for example, TRα is indispensable for norepinephrine signaling, and TRβ is sufficient but not strictly required to stimulate UCP1 (15,16). Recent studies established TH and its agonists as a class of browning agents (17–19). Because most of the available TH mimetics are TRβ selective and/or liver targeted, and systemic TH treatment affects metabolism in tissues other than adipose depots, the cell-autonomous effect of TH on adipose tissue and the importance of adipose TH action in systemic metabolic homeostasis remain unclear.
Here, we demonstrate that systemic treatment of triiodothyronine (T3), the active form of TH, has a profound effect on WAT and systemic homeostasis. By using mice with adipose tissue–specific deletion of TRα (ATRαKO) and TRβ (ATRβKO), we show that TRβ is the major TR isoform that mediates the effect of T3 on multiple metabolic pathways in inguinal WAT (iWAT). Further analysis illustrates the essential role of ChREBP in T3-mediated regulation of glucose uptake and use, FA anabolism and catabolism, and thermogenesis in iWAT. ATRβKO mice are susceptible to diet-induced obesity and metabolic dysregulation, indicating that TRβ in iWAT may serve as a therapeutic target.
Research Design and Methods
Animal Studies
ATRαKO and ATRβKO mice were generated by crossing TRα floxed mice (20) and TRβ floxed mice (20,21) with adiponectin-Cre mice, a gift from Liu Yong (Wuhan University). Mice with adipose tissue–specific deletion of ChREBP (AChRKO) were generated in the laboratory of author W.J.Z. (Naval Medical University) by crossing ChREBP floxed mice with adiponectin-Cre mice (22). Male C57BL/6J mice and transgenic mice on a C57BL/6 background, both 8–12 weeks old, were used and maintained at 23°C ± 1°C under a 12-h light/dark cycle in a specific pathogen-free animal facility. All animal experiments were performed in accordance with guidelines of the Ethics Committee of Shanghai Institute of Nutrition and Health (SINH) and approved by the Animal Care Committee of SINH (approval no. SIBS-2019-YH-1 and SINH-2020-YH-1).
To explore the T3 action in vivo, mice were rendered hypothyroid with methimazole (MMI mice), followed by daily T3 intraperitoneal injections (0.25 μg/g BW) for 5 days (MMI+T3 mice), as previously described (Supplementary Fig. 1A) (23). Some mice were acclimatized to thermoneutrality (30°C) for 2 weeks prior to T3 injection. To induce obesity, 8-week-old ATRβKO mice were fed a high-fat diet (HFD; 60 kcal% fat) for 2–3 months. For glucose tolerance tests (GTTs), mice were fasted overnight and then given a glucose intraperitoneal injection at 2 g/kg BW. For insulin tolerance tests (ITTs), mice were fasted for 6 h and then given an insulin intraperitoneal injection at 1 unit/kg BW. Minispec TD-NMR Analyzers were used to evaluate mouse body composition. Oxygen consumption and EE were measured by using metabolic cages (Columbus Instruments) and LabMaster Software.
RNA and Protein Analysis and Histology
Real-time PCR was performed on a QuantStudio 6 Flex Real-Time System (Thermo Fisher) and 18s rRNA was used for normalization. RNA-sequencing (RNA-seq) libraries were prepared using Illumina TruSeq Stranded mRNA Library Prep (Illumina). Samples were pooled for deep sequencing on NovaSeq 6000 platforms. Differential expression analysis was performed using DESeq2. Data were analyzed with the Majorbio Cloud Platform (www.majorbio.com). Functional enrichment analysis was performed with the Database for Annotation, Visualization, and Integrated Discovery (DAVID; https://david.ncifcrf.gov/). Primer and antibody information is provided in Supplementary Tables 1 and 2. For histology, tissues were fixed in 4% paraformaldehyde and embedded in paraffin. Sections were stained with hematoxylin and eosin. For immunofluorescence, paraffin sections were incubated with primary antibodies anti-UCP1 (Abcam; 1:200) or anti-GLUT4 (Abcam; 1:200), followed by detection with Alexa Fluor 594–conjugated goat anti-rabbit secondary antibody (Invitrogen; 1:1,000). Images were captured by laser scanning confocal microscope (Carl Zeiss LSM 880).
Cell Culture and Chromatin Immunoprecipitation Assay
Preadipocytes were collected, induced to differentiate, treated with T3, or were subjected to oxygen consumption rate assay or chromatin immunoprecipitation assay, as described in other reports (20,24,25). Flag-tagged mouse TRβ was cloned into pCDH-puro between the XbaI and EcoRI sites. An shRNA against mouse ChREBP (NM_021455) at +2044/2064 with sequence of 5′-AGAAGAGGCGGTTCAATATTA-3′ was cloned into pLKO.1-puro between the EcoRI and AgeI sites. These constructs and packaging plasmids PMD2.G and PSPAX were cotransfected into HEK293T cells to obtain virus particles. Preadipocytes were infected with lentivirus of TRβ-Flag or/and shRNA lentivirus of ChREBP. After differentiation for 6 days, cells were treated with 100 nmol/L T3 for 24 h in TH-deficient medium. Immunoprecipitation was performed using an anti-histone H3 (acetyl K27; Abcam) or anti-Flag (Cell Signaling Technology) antibody. Rabbit IgG (Cell Signaling Technology) was used as a control. Primer sequences for chromatin immunoprecipitation assay are listed in Supplementary Table 3.
Statistical Analysis
Data were analyzed using GraphPad Prism (version 8.0.2). Data are represented as mean ± SEM. The statistical significance was assessed by the Student t test or two-way ANOVA; P < 0.05 was considered statistically significant.
Data and Resource Availability
All data generated or analyzed during this study are included in the published article and its online supplementary files. The RNA-seq data generated in this study have been deposited in the Gene Expression Omnibus database under accession code GSE205520. The RNA-seq data from other publications used in this study are available under accession code GSE132885.
Results
MMI+T3 mice were compared with hypothyroid mice 24 h after the last T3 injection. We found that systemic T3 administration ameliorated glucose intolerance in MMI-treated mice, accompanied by both morphological change (decrease in adipocyte size) and phenotypic switch (browning) in the iWAT (Supplementary Fig. 1B–D). A decrease in adipocyte size was also noticed in epididymal WAT (eWAT) after T3 treatment, although the change was less evident (Supplementary Fig. 1C). Moreover, increased mRNA levels of key oxidative genes (CPT1B and VLCAD) and thermogenic genes (UCP1, PGC1α, COX8B, and CIDEA) were observed in the iWAT of MMI+T3 mice as compared with MMI mice (Supplementary Fig. 1E). These data imply that the T3 action in the WAT might contribute to T3’s beneficial effects on systemic homeostasis.
Deletion of Adipocyte TRβ Blunts T3 Effect on iWAT and Systemic Homeostasis
No significant difference in BW was observed between adipocyte-specific TR knockout (KO) mice and their corresponding TRflox/flox (floxed) mice (Supplementary Fig. 1F and G). Although no obvious abnormality in iWAT morphology was found in either ATRαKO or ATRβKO mice, the effect of systemic T3 treatment on iWAT morphology, particularly the adipocyte size, was greatly attenuated in mice deficient in adipocyte TRβ, but not TRα (Fig. 1A and Supplementary Fig. 1H), suggesting that TRβ is the primary TR isoform responsible for the T3 effect we observed in iWAT.
Because WAT metabolism is critically involved in systemic homeostasis (6), we investigated whether the adipocyte TRβ contributes to the beneficial effects of T3 on systemic homeostasis. Glucose tolerance was not significantly changed in ATRβKO mice (Fig. 1B). A significant decrease in glucose tolerance was observed in ATRβKO mice after T3 treatment (Fig. 1C), suggesting that T3-induced improvement of glucose metabolism is dependent on TRβ. Similarly, the deleterious effects of TRβ loss on oxygen consumption and EE were observed in T3-treated ATRβKO mice but not in untreated ATRβKO mice (Fig. 1D–G and Supplementary Fig. 1I and J). These results suggest that adipocyte TRβ is required for the T3 action on systemic homeostasis.
Loss of Adipocyte TRβ Influences T3 Effect on Multiple Metabolic Pathways in iWAT
RNA-seq analysis by using the iWAT of Floxed and ATRβKO mice with or without T3 treatment showed that 533 genes were upregulated by T3 in the iWAT of floxed mice, of which 355 genes were downregulated by adipocyte TRβ deficiency (Fig. 2A), indicating that the expression of these 355 genes was regulated by T3 in an adipocyte TRβ-dependent manner. Kyoto Encyclopedia of Genes and Genomes (KEGG) analysis of these 355 genes revealed that numerous pathways related to glucose and FA metabolism seemed to be tightly controlled by T3/TR signaling in iWAT (Fig. 2B and Supplementary Fig. 2A).
Ablation of Adipocyte TRβ Influences T3 Effect on Glucose Use in iWAT
In line with our RNA-seq data, qPCR analysis showed that the mRNA levels of adipose GLUT4 (also known as Slc2a4) were significantly elevated by T3 in the iWAT of floxed mice but not ATRβKO mice (Fig. 2C and Supplementary Fig. 2B). Similar results were observed in cultured adipocytes derived from the stromal vascular fraction (SVF) of iWAT (iWAT-SVF) (Fig. 2D). Accordingly, ablation of adipocyte TRβ markedly attenuated the T3 effect on the immunostaining intensity of GLUT4 in iWAT (Fig. 2E).
A similar expression pattern was also observed for genes involved in glycolysis (Supplementary Fig. 2B and C). These data indicate that T3 promotes glucose consumption by stimulating glucose uptake and subsequent glycolysis in adipocytes in a TRβ-dependent manner. The mRNA levels of pyruvate dehydrogenase kinase 4 (PDK4) were upregulated by T3 in either the iWAT or iWAT-SVF-derived adipocytes from floxed mice but not from ATRβKO mice (Fig. 2F and G and Supplementary Fig. 2B), indicating that adipocyte TRβ might mediate the effect of T3 on PDK4 mRNA expression. Because PDK4 redirects intracellular glucose from oxidation toward triglyceride (TAG) synthesis (8), we speculated that T3 invokes a shift in intracellular fuel selection, not favoring the use of glucose for mitochondrial oxidation, in iWAT.
Loss of Adipocyte TRβ Impairs T3 Action on FA Metabolism in iWAT
Because FA metabolism plays a role in maintaining membrane fluidity and systemic insulin sensitivity (26,27), we examined the key enzymes involved in these processes. In line with our RNA-seq data, qPCR analysis showed that the mRNA levels of de novo lipogenic genes (ACLY, ACC1, and FASN) were significantly elevated by T3 in the iWAT of floxed mice but not ATRβKO mice (Fig. 3A and Supplementary Fig. 3A). A similar expression pattern was observed for ELOVL6 and SCD1, two key enzymes involved in FA elongation and desaturation, respectively (Fig. 3B and Supplementary Fig. 3A). Consistently, after T3 treatment, the ACLY, ACC, and FASN protein levels were all significantly lower in the iWAT of ATRβKO mice than in that of floxed mice (Fig. 3C and Supplementary Fig. 3B). Similar results were obtained in the iWAT-SVF–derived adipocytes from ATRβKO and floxed mice (Fig. 3D and E and Supplementary Fig. 3C). We thus speculated that T3 promotes de novo FA synthesis in these adipocytes through TRβ.
In addition, the adipocyte TRβ–mediated T3 effects on other pathways controlling lipid anabolism and catabolism were also investigated. The mRNA levels of ATGL (also known as Pnpla2, the rate-limiting enzyme that catalyzes TAG breakdown to release FAs), DGAT2 (an enzyme involved in FA esterification), and genes involved in FA oxidation (PPARα, CPT1B, and VLCAD) were all elevated by T3 in the iWAT of floxed mice but not that of ATRβKO mice (Fig. 3F and Supplementary Fig. 3A, D, and E). Similar results were observed in cultured adipocytes derived from iWAT-SVF (Fig. 3G and Supplementary Fig. 3E).
The maximal oxygen consumption rate was significantly lower in the iWAT-SVF–derived adipocytes from ATRβKO mice than those from floxed mice (Fig. 3H). These results agree with the data obtained from oxygen consumption analysis (Fig. 1D and E), suggesting that T3 promotes overall fuel oxidation in iWAT through TRβ.
Deficiency of Adipocyte TRβ Impairs T3 Action on Thermogenesis in iWAT
Consistent with our RNA-seq data, qPCR analysis showed that the UCP1 mRNA levels were significantly elevated by T3 in the iWAT of floxed but not ATRβKO mice (Fig. 4A and Supplementary Fig. 3F). Accordingly, both Western blot analysis and immunostaining showed that, after T3 treatment, the UCP1 protein levels were lower in the iWAT of ATRβKO mice than in that of floxed mice (Fig. 4B and C). The TRβ-mediated T3 effect on UCP1 mRNA and protein levels also was observed in vitro (Fig. 4D–F). The mRNA levels of glycerol kinase (GYK) were increased by T3 in the iWAT or iWAT-SVF–derived adipocytes of floxed but not ATRβKO mice (Fig. 4G and H and Supplementary Fig. 3G). These results, together with the findings for de novo lipogenic genes ATGL, PDK4, and DGAT2 (Figs. 2F and 3A and Supplementary Fig. 3D and E), suggest that substrate cycle (TAG–FA cycling) might also contribute to the energy dissipation controlled by T3 and TRβ.
In contrast, the mRNA expression of genes (e.g., Slc6a8, Gatm, Ckmt2) involved in creatine cycling was negatively regulated by T3 and TRβ (Supplementary Fig. 3G), suggesting that creatine cycling is not involved. Additionally, the mRNA expression of genes (e.g., Itpr, Atp2a2, Ryr2) involved in Ca2+ cycling was unlikely under the control of T3 and TRβ (Supplementary Fig. 3G), indicating that Ca2+ cycling is also not involved.
Given that the effects of either T3 treatment or loss of adipocyte TRβ at room temperature (Figs. 2–4 and Supplementary Fig. 4A) were also observed at thermoneutrality (Supplementary Fig. 4B–I), we speculated that the T3 action we observed does not require the adrenergic stimulation.
Loss of Adipocyte TRβ Has Differential Effects on T3-Regulated Gene Expression in Different Fat Depots
The TRβ-dependent T3 effects were also observed for the mRNA levels of ACLY, ACC1, FASN, SCD1, ELOVL6, CPT1B, PDK4, and UCP1 but not for PPARα, VLCAD, GLUT4, and GYK in visceral eWAT (Supplementary Fig. 5A). These results indicate that, although much similarity was observed for the effects of either T3 treatment or loss of adipocyte TRβ, differential effects exist between subcutaneous iWAT and visceral eWAT. The mRNA levels of ACLY, GLUT4, CPT1B, UCP1, and GYK were slightly downregulated, although some changes did not reach statistical significance, rather than upregulated after T3 treatment in interscapular brown adipose tissue (iBAT), and the loss of adipocyte TRβ had no additional effect on T3-caused repression (Supplementary Fig. 5B). In addition, T3 treatment had no effects on the mRNA levels of ACC1, FASN, SCD1, ELOVL6, PDK4, PPARα, and VLCAD in iBAT (Supplementary Fig. 5B). We thus speculated that T3 treatment had minimal effects on these metabolic genes in iBAT.
ChREBP Is Regulated by T3 in a TRβ-Dependent Manner in iWAT
Although TH action is believed to be achieved via TR-mediated transcriptional regulation, indirect mechanisms are also crucial (12,28,29). Further analysis of our RNA-seq data revealed that, in accordance with the expression of GLUT4 and lipogenic genes (ACLY, ACC1, FASN, ELOVL6, and SCD1), the expression of ChREBP was regulated by T3 in a TRβ-dependent manner in WAT (Fig. 5A). Consistently, qPCR analysis revealed that T3 could upregulate the expression of either ChREBPα or ChREBPβ in the iWAT of floxed but not ATRβKO mice, regardless of housing conditions (Fig. 5B and Supplementary Fig. 6A). Accordingly, the ChREBPα protein levels were decreased in the iWAT of ATRβKO mice after T3 treatment (Fig. 5C). Similar results were observed in vitro (Fig. 5D). Consistently, T3 increased the levels of H3K27ac, a marker correlated with active transcription, and the recruitment of TRβ in the promoter region of ChREBP in cultured adipocytes (Supplementary Figure 6B). TRβ-mediated regulation of ChREBP mRNA expression by T3 was observed in eWAT but not iBAT (Supplementary Fig. 6C and D). In contrast, loss of adipocyte TRα had no effect on the regulation of ChREBP mRNA expression by T3 (Supplementary Fig. 6E and F). Furthermore, elevated ChREBP mRNA and protein levels were observed in the iWAT of MMI+T3 mice (Supplementary Fig. 6G and H). Notably, the mRNA levels of SREBP1C were not under the control of T3 and TRβ (Supplementary Fig. 6G and I). Collectively, these data suggest that ChREBP expression is under the control of T3 in iWAT, whereas TRβ is the primary TR isoform that mediates the T3 effect on ChREBP.
ChREBP Mediates T3 Action on Glucose and Lipid Metabolism in iWAT
ChREBP and GLUT4 play essential roles in governing not only intracellular fate of glucose in adipocytes but also systemic homeostasis (2,3,5,30). Considering that T3 treatment and adipocyte TRβ deficiency mimicked the effect of adipocyte ChREBP or GLUT4 overexpression and deletion, respectively, and ChREBP could mediate T3-induced lipogenesis in liver (31), we hypothesized that adipocyte ChREBP might also be critically involved in the T3-regulated adipose metabolism and systemic homeostasis. We tested our hypothesis using iWAT-SVF–derived adipocytes from AChRKO mice and found that those adipocytes had reduced GLUT4 mRNA levels (Fig. 6A and B). The T3 effect on GLUT4 was totally abolished in these adipocytes from AChRKO mice (Fig. 6B), whereas the T3 effects on the mRNA levels of genes involved in FA anabolism (i.e., ACLY, ACC1, FASN, and SCD1, but not ELOVL6) were either abolished or largely attenuated in adipocytes from AChRKO mice (Fig. 6C–E). Moreover, the T3 effect on the mRNA expression of genes involved in fuel selection (PDK4), TAG breakdown (ATGL), FA esterification (DGAT2), and oxidation (PPARα, CPT1B, and VLCAD) was blocked or greatly attenuated in these adipocytes from AChRKO mice (Fig. 6F and G and Supplementary Fig. 7A and B). Furthermore, the T3-induced upregulation of UCP1 and GYK mRNA expression was diminished in adipocytes from AChRKO mice (Fig. 6H and Supplementary Fig. 7C).
Additionally, knockdown of ChREBP by specific shRNA reduced the mRNA levels of GLUT4, ACLY, ACC1, FASN, PDK4, and UCP1 in vitro (Supplementary Fig. 7D–F). In line with the data showing an increase in GYK mRNA levels in the adipocytes from AChRKO mice in the absence of T3 (Supplementary Fig. 7C), the shChREBP treatment increased the GYK mRNA levels (Supplementary Fig. 7G), indicating that the elevation of GYK mRNA levels observed in adipocytes with either KO or inhibition of ChREBP might be attributed to a compensatory mechanism.
Consistent with qPCR data, T3 increased H3K27ac levels in promoter regions of ACLY, ACC1, FASN, SCD1, GLUT4, PDK4, and UCP1 (Supplementary Fig. 7H and I), which contain putative TH response elements reported previously or predicted in silico (32–38), indicating that T3 regulates these genes at the transcriptional level. In agreement with our proposed model that ChREBP mediates the T3 action on these genes, knockdown of ChREBP greatly attenuated the T3 effect on H3K27ac levels (Supplementary Fig. 7H and I). The results of ANOVA analysis also support the notion that T3-induced H3K27ac in promoter regions of these genes is ChREBP dependent (Supplementary Fig. 7H and I).
T3 treatment also increased the promoter occupancy of TRβ for ChREBP, ACC1, and UCP1, but not for ACLY, PDK4, and SCD1 (Supplementary Figs. 6B and 7J), suggesting that T3-induced increases in TRβ recruitment is not a universal mechanism (39). Moreover, knockdown of ChREBP could abolish the T3-induced TRβ recruitment for ACC1 and UCP1, indicating that ChREBP facilitates the TRβ binding. Furthermore, knockdown of ChREBP attenuated the TRβ recruitment to the ACLY promoter but not to the SCD1 promoter regardless of the presence of ligands. Surprisingly, knockdown of ChREBP had an opposite effect on TRβ recruitment to the PDK4 promoter in the presence and absence of ligand, suggesting a more complicated mechanism exists. The TRβ recruitment to promoter regions of GLUT4 and FASN could not be detected by gel electrophoresis analysis, suggesting that the regulation by T3 might not need TRβ binding but relies on ChREBP (Supplementary Fig. 7K).
Collectively, our data suggest that adipocyte ChREBP mediates the T3 effects on many genes involved in glucose and lipid metabolism and thermogenesis in WAT. We speculated that ChREBP might act as a critical downstream effector of T3 and mediate the T3 effect on many aspects of glucose and lipid metabolism, thereby redirecting the intracellular glucose from oxidation toward TAG synthesis and favoring the use of intracellular FAs as a direct energy source for oxidation and thermogenesis in WAT.
Ablation of Adipocyte TRβ Worsens Diet-Induced Obesity and Metabolic Disorder
Because adipose TRβ mRNA expression decreases in obese patients (40), to determine which altered metabolic pathways may contribute to the deleterious effects due to the downregulation of TRβ, we compared the downregulated genes in ATRβKO mice in response to T3 treatment with those genes downregulated by HFD feeding reported previously (accession no. GSE132885) (41). We found 70 overlapped genes regulated by either deletion of adipocyte TRβ or HFD feeding. Gene ontology analysis of these 70 overlapped genes revealed multiple metabolic pathways regulated by either loss of adipocyte TRβ or overfeeding (Supplementary Fig. 8A). Notably, FA biosynthetic process was ranked first among these pathways (Supplementary Fig. 8A).
As expected, ATRβKO mice were more prone to HFD-induced obesity, as evident from increased BW, fat mass, and adipocyte size in ATRβKO mice (Fig. 7A–E). Moreover, deletion of adipocyte TRβ exacerbated glucose intolerance and insulin insensitivity, and reduced oxygen consumption and EE in HFD-fed mice (Fig. 7F–K). Given that the expression of ChREBP, ACLY, ACC1, FASN, ELOVL6, SCD1, and GLUT4 was dysregulated in the iWAT of either mice lacking adipocyte TRβ in response to T3 (Fig. 5A) or mice upon HFD feeding (Supplementary Fig. 8B), we speculated that dysregulated glucose uptake and de novo lipogenesis due to impaired T3 signaling and subsequent altered ChREBP expression in WAT might be responsible for metabolic defects observed. To test our hypothesis, we performed RNA-seq analysis and found that 2,913 genes responded abnormally to HFD feeding in the iWAT of ATRβKO mice (Fig. 8A). KEGG analysis of the 2,230 downregulated genes due to adipocyte TRβ deficiency revealed that multiple pathways, particularly those involved in the pathogenesis of metabolic diseases, were dysregulated in the iWAT of HFD-fed ATRβKO mice (Fig. 8B).
Consistent with the RNA-seq data, qPCR analysis revealed decreases in mRNA levels of ChREBP and GLUT4, and of key enzymes involved in de novo lipogenesis (namely, ACLY, ACC1, and FASN), FA desaturation (SCD1), and oxidation (CPT1B) in the iWAT of HFD-fed ATRβKO mice (Fig. 8C–H and Supplementary Fig. 9A). On the basis of these results, we speculated that defects in adipocyte TRβ–mediated T3 signaling might contribute to the downregulation of glucose uptake and lipogenesis in response to HFD feeding, thereby being involved in the pathogenesis of HFD-induced obesity and metabolic dysfunction. In contrast, the downregulation of the mRNA levels of PDK4 and UCP1 by deletion of adipocyte TRβ was observed in the iWAT of mice fed a HFD (Fig. 8I and Supplementary Fig. 9B), which was not consistent with the results observed in vitro (Figs. 2G and 4D), suggesting that the regulation of these genes by the deletion of adipocyte TRβ was lost upon HFD feeding and other mechanisms might be involved and be more dominant here.
Discussion
TH is essential for maintaining systemic homeostasis. Tissue-specific distribution of TR isoforms provides a means to achieve diversity and selectivity of TRs in transcriptional regulation. Because resmetirom (MGL-3196), a TRβ agonist, exhibited efficacy and safety in a phase 3 nonalcoholic steatohepatitis study, interest in potential therapeutic options for nonalcoholic fatty liver disease, even dyslipidemia or metabolic symptoms, targeting TRβ continues to grow. In addition to liver, adipose tissue is another TH target tissue. However, whether and how adipocyte TR isoforms mediate the TH effect on intracellular metabolism, thereby getting involved in the regulation of systemic metabolic homeostasis, are unclear. Our findings in the present study suggest that adipocyte TRβ may be primarily responsible for the TH action on adipose tissue metabolism (Supplementary Fig. 9C), may mediate some of the beneficial effects of TH on systemic metabolism, and that activation of adipose TRβ may be a strategy for treating certain metabolic diseases. To our knowledge, this is the first study of the metabolic effect of adipocyte-specific KO of TRs.
Because systemic TH treatment affects metabolic pathways in many tissues, it is difficult to elucidate the cell-autonomous effect of TH on certain tissues and the importance of local TH action in systemic homeostasis. Here, we used mice with adipose tissue–specific deletion of TR and demonstrated the cell-autonomous effect of TH on adipocyte metabolism. We found that adipocyte TRβ, but not TRα, is important for the regulation of iWAT metabolism by TH. Moreover, we hypothesized that, in iWAT, T3, via its receptor TRβ, increases glucose utilization through stimulating glucose uptake and glycolysis, and meanwhile redirects glycolytic metabolite from oxidation toward TAG synthesis and enhances both FA anabolism (de novo lipogenesis) and catabolism (FA oxidation), thereby favoring the use of FAs to fuel the tricarboxylic acid cycle. We also propose that T3, via TRα, increases energy dissipation by promoting white to beige adipocyte conversion through both UCP1-dependent mechanism and UCP1-independent futile metabolic cycling (e.g., TAG–FA cycling). Notably, because the T3-induced changes in the mRNA expression of metabolic genes we studied were much more evident in WAT than those in iBAT and were relatively more evident in iWAT than those in eWAT, we speculated that iWAT might make a greater contribution to whole-body metabolism after T3 treatment. However, we did not intend to exclude the contributions of eWAT and iBAT, the latter of which did not display obvious alteration in mRNA levels of these metabolic genes involved in glucose and lipid metabolism and energy metabolism.
Adipose tissues are critical in whole-body metabolic homeostasis (6,42). Increased de novo lipogenesis in WAT is associated with enhanced glucose tolerance and insulin sensitivity (3,30). Adipocyte ChREBP not only regulates de novo FA synthesis but also modulates insulin action and glucose homeostasis. Here, our data suggest that ChREBP is essential for the transcription of many T3-regulated genes, although the underlying mechanisms may vary. For example, T3 may regulate these genes either directly, which requires the regulatory role of ChREBP in chromatin remodeling and accessibility, or indirectly via ChREBP, which relies on the transcription activity of ChREBP. We speculated that adipocyte ChREBP might act as a downstream effector of T3 as it does in liver (31) and mediate the T3 effect on many aspects of glucose and lipid metabolism, thereby redirecting the intracellular glucose from oxidation toward TAG synthesis and favoring the use of intracellular FAs as a direct energy source for oxidation and thermogenesis in WAT (Supplementary Fig. 9C). Because adipocyte ChREBP is regulated by T3 in a TRβ-dependent manner, we speculated that adipocyte TRβ could modulate the sensing and coordinating responses to changes in nutrient availability and integrate adipocyte and whole-body metabolic function by targeting ChREBP. We also proposed that defects in adipocyte TH signaling would lead to downregulation of ChREBP and its regulated metabolic pathway, contributing to diet-induced metabolic dysfunction, whereas the beneficial TH effects on systemic homeostasis may result from the regulation of glucose and FA metabolism through the TH action on ChREBP in adipocytes.
This article contains supplementary material online at https://doi.org/10.2337/figshare.21977249.
Y.M. and S.S. contributed equally to this work.
Article Information
Acknowledgments. The authors thank Zhonghui Weng from the Institutional Center for Shared Technologies and Facilities of the Shanghai Institute of Nutrition and Health, Chinese Academy of Sciences, for technical assistance.
Funding. This work was supported by National Key Research and Development Program of China (grant 2021YFA1100500 to H.Y.); National Natural Science Foundation of China (grants 91957205 to H.Y., 82070821 to J.J., 81970748 to L.Z., and 81870541 to Z. Lu); Pujiang Talent Program from the Science and Technology Commission of Shanghai Municipality (grant 21PJ1416100 to Yuying Li); Youth Innovation Promotion Association (grant CAS 2021261 to Yuying Li); Laboratory for Marine Drugs and Bioproducts of Pilot National Laboratory for Marine Science and Technology (Qingdao) (grant LMDBKF-2019-04 to H.Y.); and National Health Commission Key Laboratory of Food Safety Risk Assessment (grant 2020K02 to Y.W.).
Duality of Interest. No potential conflicts of interest relevant to this article were reported.
Author Contributions. Y.M., S.S., and H.Y. designed the experiments and analyzed the data, Y.M. carried out most of the experiments, S.S. mainly contributed to the revision of the manuscript, Y.Y., S.Z., S.L., Z.T., J.Y., M.M., Z.N., Z. Li, Y.W., C.W., and W.J.Z. provided the technical assistance, analyzed, and interpreted the data, L.Z., Z. Lu, Y.X., Q.Z., Yu Li, C.H., Yuying Li, and J.J. contributed to the discussion and supervised the project, Y.M., Yuying Li, J.J., and H.Y. wrote the manuscript. H.Y. 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.