Thyroid Hormone Receptor-α Gene Knockout Mice Are Protected from Diet-Induced Hepatic Insulin Resistance

Abstract

Nonalcoholic fatty liver disease (NAFLD) is the most frequent chronic liver disease in the United States and is strongly associated with hepatic insulin resistance. We examined whether the thyroid hormone receptor-α (Thra) would be a potential therapeutic target to prevent diet-induced NAFLD and insulin resistance. For that purpose, we assessed insulin action in high-fat diet-fed Thra gene knockout (Thra-0/0) and wild-type mice using hyperinsulinemic-euglycemic clamps combined with ³H/¹⁴C-labeled glucose to assess basal and insulin-stimulated rates of glucose and fat metabolism. Body composition was assessed by ¹H magnetic resonance spectroscopy and energy expenditure by indirect calorimetry. Relative rates of hepatic glucose and fat oxidation were assessed in vivo using a novel proton-observed carbon-edited nuclear magnetic resonance technique. Thra-0/0 were lighter, leaner, and manifested greater whole-body insulin sensitivity than wild-type mice during the clamp, which could be attributed to increased insulin sensitivity both in liver and peripheral tissues. Increased hepatic insulin sensitivity could be attributed to decreased hepatic diacylglycerol content, resulting in decreased activation of protein kinase Cε and increased insulin signaling. In conclusion, loss of Thra protects mice from high-fat diet-induced hepatic steatosis and hepatic and peripheral insulin resistance. Therefore, thyroid receptor-α inhibition represents a novel pharmacologic target for the treatment of NAFLD, obesity, and type 2 diabetes.

Nonalcoholic fatty liver disease (NAFLD) is now the most frequent chronic liver disease in the United States, affecting one in four adults, and is a major risk factor for the development of type 2 diabetes (1). Current pharmacologic treatment of NAFLD is disappointing, relying mostly on weight loss (2–4), although insulin-sensitizing agents, such as thiazolidinediones, have been shown to decrease hepatic steatosis by promoting fat redistribution to the sc adipose tissue (5, 6).

Thyroid hormone plays a role in diverse important metabolic pathways in lipid and glucose metabolisms and regulation of body weight (7). Thyroid hormone acts predominantly through its nuclear receptors, thyroid hormone receptors α and β, which differ in their tissue distribution (8). Although thyroid hormone therapy for the treatment of obesity and NAFLD would be deleterious in euthyroid patients due to associated cardiovascular side effects, such as tachycardia and hypertension, selective thyroid receptor agonists are being developed to stimulate specific metabolic pathways and thus avoid these toxicities (9). In support of this novel therapeutic approach, mice lacking the thyroid hormone receptor-α gene (Thra-0/0) are leaner and are less sensitive to high-fat diet-induced obesity (10).

We therefore hypothesized that Thra-0/0 mice could also be protected from high-fat diet-induced hepatic steatosis and associated hepatic insulin resistance. To examine this hypothesis, we assessed whole-body and tissue-specific effects of insulin in awake mice using the hyperinsulinemic-euglycemic clamp technique combined with ³H/¹⁴C-labeled glucose. In addition, we also assessed liver lipid intermediates that have been associated with insulin resistance, such as triglycerides and diacylglycerol (DAG) (11–13) as well as signaling events typically associated with an increase in liver DAG content, i.e. protein kinase Cε (PKCε) activation as well as potential alterations in insulin signaling downstream of the insulin receptor kinase (14). Finally, we also assessed the effects of thyroid hormone receptor-α gene ablation on relative rates of hepatic glucose and fat oxidation in vivo using a novel proton-observed carbon-edited nuclear magnetic resonance technique.

Materials and Methods

Animals

Male Thra-0/0 mice and wild-type (WT) littermates were generated as previously described (15) and individually housed under controlled temperature (23 C) and lighting (12-h light, 12-h dark cycle, lights on at 0700 h) with free access to water and food. After 1 wk of acclimatization, a high-fat diet (TD 93075; Harlan Teklad, Madison, WI) was started and continued for 3 wk. The proportions of calories derived from nutrients were as follows: 54.8% fat, 24% carbohydrate, 21.2% protein, energy density 4.8 Kcal/g, and trace amount of cholesterol (0.007% wt/wt). Body composition was assessed by ¹H magnetic resonance spectroscopy using a Bruker Minispec analyzer (Bruker, The Woodlands, TX). Metabolic parameters and physical activity were measured using the Oxymax system from Columbus Instruments (Columbus, OH). All experiments were done in overnight-fasted animals (16 h, from 1800 to 1000 h). The studies were conducted at the Yale Mouse Metabolic Phenotyping Center. All procedures were approved by the Yale University Animal Care and Use Committee.

Plasma assays

Blood samples were collected by cardiac puncture in heparinized syringes and centrifuged at 12,000 rpm for 2 min. Plasma was then either directly used or frozen at −20 C for further analyzes. Plasma glucose was measured by a glucose oxidase method on a Beckman Glucose Analyzer II (Beckman Coulter, Brea, CA). Plasma fatty acids were determined with the NEFA C kit (Wako Pure Chemical Industries, Osaka, Japan). Plasma insulin was measured by a RIA kit (Millipore, Billerica, MA). Cholesterol panel was analyzed using COBAS Mira Plus (Roche, Indianapolis, IN). Plasma bile acids were measured as previously described (16).

Liver lipid intermediates measurements

Tissue triglycerides were extracted using the method of Bligh and Dyer (17) and measured using a DCL Triglyceride Reagent (Diagnostic Chemicals Ltd., Oxford, CT). For DAG extraction, livers were homogenized in a buffer (20 mm Tris-HCl, 1 mm EDTA, 0.25 mm EGTA, 250 mm sucrose, and 2 mm phenylmethylsulfonylfluoride) containing a protease inhibitor cocktail (Roche, Basel, Switzerland), and samples were centrifuged at 100,000 × g for 1 h. The supernatants containing the cytosolic fraction were collected. DAG levels were then measured as previously described (18). Total cytosolic DAG contents are expressed as the sum of individual species. All lipids measurements were done in animals fasted overnight and under basal conditions (i.e. without insulin stimulation).

Hyperinsulinemic-euglycemic clamp studies

Jugular venous catheters were implanted 6–7 d before the hyperinsulinemic-euglycemic clamp experiments. To assess basal whole-body glucose turnover, [3-³H]-glucose (high-performance liquid chromatography purified; PerkinElmer Life Sciences, Boston, MA) was dehydrated, reconstituted in saline, and infused at a rate of 0.05 μCi/min for 2 h. After the basal period, the hyperinsulinemic-euglycemic clamp was conducted in conscious mice for 140 min with a 4-min primed infusion (29 mU/kg) followed by a continuous [3 mU/kg · min] infusion of human insulin (Novolin; Novo Nordisk, Princeton, NJ), a continuous infusion of [3-³H]-glucose (0.1 μCi/min), and a variable infusion of [1-¹³C]glucose (99% ¹³C enriched, 20 g/dl) was used to maintain euglycemia (100–120 mg/dl). Plasma samples were obtained from the tail at 0, 30, 50, 65, 80, 90, 100, 110, 120, 130, and 140 min. The tail incision was done at least 2 h before the first blood sample was taken to allow for acclimatization, according to standard operating procedures (19). To measure tissue-specific glucose uptake, 10 μCi of 2-deoxy-D-[1-¹⁴C]-glucose (PerkinElmer Life Sciences) were injected as a bolus at 85 min. At the end of the clamp, mice were anesthetized with pentobarbital sodium injection (150 mg/kg), and all tissues were taken within 4 min, snap frozen in liquid nitrogen using aluminum tongs, and stored at −80 C for subsequent analysis. Biochemical analysis and calculations for the hyperinsulinemic-euglycemic clamps were performed as previously described (20). Pyruvate dehydrogenase flux (V_PDH)/mitochondrial tricarboxylic acid flux (V_TCA) flux was estimated from the [4-¹³C]glutamate/[3-¹³C]alanine enrichments in liver extracts after the hyperinsulinemic-euglycemic clamp studies as previously described (21).

Liver insulin signaling

PKCε membrane activation was assessed in liver protein extracts under basal conditions as previously described (22).

Immunoblot analysis

Immunoblots were done as previously described (23). For fibroblast growth factor (FGF)15 immunoblot, 3 μl of plasma were homogenized in 47 μl of radioimmunoprecipitation assay buffer. Subsequently, 25 μl of this solution were mixed with 25 μl of Laemmli buffer and submitted to electrophoresis, assuming similar total plasma protein content between genotypes. Membranes were incubated overnight with primary antibodies for phospho (p)-serine-threonine kinase (Akt) (Cell Signaling Technology, Danvers, MA), phosphoenolpyruvate caroxykinase (Abcam, Cambridge, MA), pyruvate carboxylase (1:1000 dilution; Abcam), sterol regulatory element-binding protein 1c (SREBP1c) (1:500 dilution; Santa Cruz Biotechnology, Inc., Santa Cruz, CA), p-acetyl-coenzyme A (CoA) carboxylase (ACC)2 (Cell Signaling Technology) (1:500 dilution), or FGF15 (1:200 dilution; Santa Cruz Biotechnology, Inc., Santa Cruz, CA). After further washings, membranes were incubated with horseradish peroxidase-conjugated secondary antibody (Bio-Rad, Hercules, CA) (1:3000 dilution, 1:2000 for SREBP1c, pACC2, and ACC2 and 1:2500 for FGF15) and visualized by enhanced chemiluminescent substrate (Pierce, Rockford, IL). Membranes were stripped and reblotted with antitotal-Akt antibody (Cell Signaling Technology), glyceraldehyde-3-phosphate dehydrogenase (1:1000 dilution; Santa Cruz Biotechnology, Inc., Santa Cruz, CA) or ACC2 (1:500 dilution; Cell Signaling Technology). Bands were then quantified using ImageJ (National Institutes of Health, Bethesda, MD).

Total RNA preparation, real-time quantitative PCR analysis

Total RNA was extracted from frozen livers using RNeasy 96 kit (QIAGEN, Valencia, CA), and then 1 μg of RNA was reverse transcribed into cDNA with the use of the Quantitect RT kit (QIAGEN) as per manufacturer's protocol. The abundance of transcripts was assessed by real-time PCR on a 7500 Real-Time PCR System (Applied Biosystems, Foster City, CA) with a SYBR Green detection system. Samples were run in duplicate for both the gene of interest and cyclophilin, and data were normalized for the efficiency of amplification, as determined by a standard curve included on each run. Primers used are available upon request.

Statistical analysis

Data are expressed as means ± sem. Results were assessed using two-tailed unpaired Student's t test or one-way ANOVA (GraphPad Prism 5, La Jolla, CA). A P value less than 0.05 was considered significant.

Results

Thra-0/0 mice are protected from high-fat diet-induced obesity

Thra-0/0 mice fed a high-fat diet were lighter and leaner than their WT littermates after 3 wk of high-fat diet, despite similar body weights before high-fat diet consistent with previous studies (10). Thra-0/0 mice had lower plasma total cholesterol, high-density lipoprotein cholesterol and triglycerides concentrations compared with the WT littermate mice (Table 1). Thra-0/0 mice had an approximately 30% increase in energy expenditure (Fig. 1A), oxygen consumption (VO₂) (Fig. 1B), and carbon dioxide production (VCO₂) (Fig. 1C) compared with WT littermates, and these increases remained significantly different when expressed in ml × h⁻¹ × 100 g^−0.75 (Table 2), which corrects energy expenditure for the surface area to volume ratio (24), because Thra-0/0 mice are smaller. Importantly, the increase in energy expenditure was present without any difference in food consumption or locomotor activity (Table 2).

Table 1.

Plasma analyses

	WT	Thra-0/0
Total cholesterol (mg/dl)	134.9 ± 5.1	113.1 ± 7.9a
HDL cholesterol (mg/dl)	66.9 ± 2.1	57.8 ± 4.1a
Triglycerides (mg/dl)	130.5 ± 4.5	87.6 ± 7.5b
Bile acids (μmol/liter)	1.4 ± 0.4	1.1 ± 0.3
Fasting insulin (μU/ml)	12.8 ± 2.3	6.8 ± 1.6a
Clamp insulin (μU/ml)	56.0 ± 11.7	57.3 ± 4.2
Fatty acids (fasting) (mmol/liter)	0.89 ± 0.05	0.79 ± 0.06
Fatty acids (clamp) (mmol/liter)	0.58 ± 0.04	0.33 ± 0.04b
Insulin suppression of fatty acids (%)	32.7 ± 7.9	57.4 ± 4.4a
Lactate (mmol/liter)	2.1 ± 0.1	3.1 ± 0.5a

HDL, High-density lipoprotein. n = 6–9 mice per group. Data are represented as mean ± sem.

^aP < 0.05.

^bP < 0.001 vs. WT.

Fig. 1.

Metabolic cage data. A, Energy expenditure. B, VO₂. C, VCO₂ (n = 4–5 per group). White circles and bars, WT; black circles and bars, Thra-0/0 mice. Data are mean ± sem. *, P < 0.05; **, P < 0.01; and ***, P < 0.001 vs. WT.

Table 2.

Physiologic parameters

	WT	Thra-0/0
Body weight before HFD (g)	21.5 ± 1.1	20.1 ± 0.5
Body weight after 3 wk HFD (g)	28.2 ± 0.6	25.6 ± 0.6a
Fat mass (% of body weight)	17.0 ± 1.3	11.5 ± 1.1b
Fat mass (g)	4.8 ± 0.5	3.0 ± 0.3b
Lean mass (% of body weight)	67.2 ± 1.0	71.6 ± 1.0b
Lean mass (g)	18.9 ± 0.3	18.4 ± 0.5
Energy expenditure (ml × h−1 × 100 g−0.75)	1.09 ± 0.03	1.26 ± 0.06a
VO2 (ml × h−1 × 100 g−0.75)	227 ± 7	263 ± 11a
VCO2 (ml × h−1 × 100 g−0.75)	183 ± 5	211 ± 8a
Caloric intake (kcal/kg · h)	14.0 ± 2.2	11.9 ± 1.7
Locomotor activity (counts/h)	100 ± 13.8	172 ± 31.6

HFD, High-fat diet. Except for body weight before HFD, all measurements were done 3 wk after HFD. n = 6–9 mice per group. Data are represented as mean ± sem.

^aP < 0.05.

^bP < 0.001 vs. WT.

Thra-0/0 mice are protected from high-fat diet-induced insulin resistance

To assess hepatic and peripheral insulin sensitivity, we performed hyperinsulinemic-euglycemic clamps 6–7 d after surgery. Although basal endogenous glucose production was approximately 25% higher in Thra-0/0 mice compared with the WT littermate mice (Fig. 2A), glucose infusion rates required to maintain euglycemia (Fig. 2B) during the hyperinsulinemic-euglycemic clamps were approximately 2-fold higher in Thra-0/0 mice compared with WT mice, demonstrating increased whole-body insulin sensitivity in Thra-0/0 compared with WT mice (Fig. 2C). Endogenous glucose production during the hyperinsulinemic-euglycemic clamp was approximately 2-fold lower in Thra-0/0 mice (Fig. 2D), reflecting improved hepatic insulin sensitivity. Although basal plasma insulin concentrations were approximately 50% lower in Thra-0/0 mice (Table 1), plasma insulin concentrations were matched between groups at the end of the clamp (Table 1), which is important for groups comparison. Insulin-stimulated peripheral glucose disposal (Fig. 3A) was approximately 35% higher in Thra-0/0 mice compared with the WT littermate mice. This increase in insulin-stimulated peripheral glucose disposal could be attributed, at least in part, to an approximately 2-fold increase in insulin-stimulated 2-deoxy-D-[1-¹⁴C]-glucose uptake in both white adipose tissue (WAT) (Fig. 3B) and brown adipose tissue (BAT) (Fig. 3C). In contrast, there was no difference in insulin-stimulated skeletal muscle (gastrocnemius, including soleus) (Fig. 3D) glucose uptake. Basal plasma fatty acids concentrations were similar between groups but approximately 60% lower after insulin stimulation in Thra-0/0 mice (Table 1). Therefore, the ability of insulin to suppress lipolysis during the hyperinsulinemic-euglycemic clamp was also significantly increased in Thra-0/0 mice. Fasting plasma lactate concentrations were approximately 50% higher in Thra-0/0 mice (Table 1).

Fig. 2.

Thra-0/0 mice are protected from high-fat diet-induced insulin resistance. A, Endogenous glucose production. B, Plasma glucose time course during the hyperinsulinemic-euglycemic clamp. C, Glucose infusion rate required to maintain euglycemia during the hyperinsulinemic-euglycemic clamp. D, Endogenous glucose production during the hyperinsulinemic-euglycemic clamp (n = 8–9 per group). White circles and bars, WT; black circles and bars, Thra-0/0 mice. Data are mean ± sem. **, P < 0.01 and ***, P < 0.001 vs. WT.

Fig. 3.

Whole-body insulin-stimulated glucose disposal and glucose uptake in Thra-0/0 mice. A, Whole-body insulin-stimulated glucose disposal. B–D, 2-Deoxy-D-[1-¹⁴C]-glucose uptake in WAT (B), BAT (C), and skeletal muscle (gastrocnemius including soleus) (D) (n = 8–9 per group). White bars, WT; black bars, Thra-0/0 mice. Data are mean ± sem. *, P < 0.05 and **, P < 0.01 vs. WT.

Thra-0/0 mice have a decrease in hepatic lipid intermediates content and increased insulin signaling in liver

Liver triglyceride content (Fig. 4A) and cytosolic DAG content (Fig. 4B) were significantly decreased in Thra-0/0 mice. The decrease in hepatic cytosolic DAG content was associated with a significant decrease in PKCε membrane activity (Fig. 4C). Associated with these changes, insulin-stimulated p-Akt/Akt ratio was increased by approximately 2-fold (Fig. 4D) in Thra-0/0 mice compared with WT mice.

Fig. 4.

Hepatic lipid intermediates and insulin signaling. A, Triglyceride content. B, Cytosolic DAG content. C, PKCε membrane activation. D, p-Akt/Akt ratio after insulin stimulation (n = 3–8 per group). White bars, WT; black bars, Thra-0/0 mice. Data are mean ± sem. *, P < 0.05 and **, P < 0.01 vs. WT.

Thra-0/0 mice have decreased hepatic lipogenesis

Thra-0/0 mice had a decrease in hepatic lipogenic genes expression, as reflected by a decrease in SREBP1c mRNA expression (Fig. 5A) and its downstream targets, ACC1 and fatty acid synthase (FAS) (Fig. 5A). Also, steaoryl-CoA desaturase 1 (SCD1) mRNA expression, another gene downstream of liver X receptor (LXR) gene, was decreased by approximately 35% (Fig. 5A), although this did not reach statistical significance (P = 0.15). Consistent with these findings, the ratio of mature over precursor protein levels of SREBP1c (Fig. 5B) was significantly reduced in Thra-0/0 mice. Interestingly, although most LXR target genes were down-regulated, LXR mRNA expression was similar between groups (Fig. 6A). There was no difference in FGF21, carnitine palmitoyltransferase 1, acyl-CoA oxidase, and 70-kDa peroxisomal membrane protein mRNA levels (Fig. 5C), which all encode proteins that are involved in lipid oxidation, confirming the predominance of decreased hepatic lipogenesis in Thra-0/0 mice. Moreover, protein expression of ACC2, a protein involved in fatty acid oxidation, was similar between genotypes (Fig. 5D). In accordance with these data, we found no difference in V_PDH/V_TCA flux between WT and Thra-0/0 mice, suggesting no difference in relative rates of glucose and fat oxidation between Thra-0/0 and WT mice (Fig. 5E). Interestingly, mRNA expression of cholesterol 7α-hydroxylase (CYP7A1), the rate-limiting enzyme in bile acids biosynthesis, was significantly decreased in Thra-0/0 mice (Fig. 6A). In contrast, bile acids (Table 1) were not different between groups, but FGF15 (assessed by plasma immunoblot analysis) was significantly decreased in Thra-0/0 mice (Fig. 6B). We then assessed farnesoid X receptor, small heterodimer partner, hepatic nuclear factor-4, and pregnane X receptor mRNA expression, all related to CYP7A1, and found no difference between groups (Fig. 6A).

Fig. 5.

Hepatic lipogenesis and lipid oxidation. A, SREBP1c, ACC1, FAS, and SCD1 mRNA expression. B, SREBP1c protein levels in basal livers (i.e. without insulin stimulation). C, FGF21, carnitine palmitoyltransferase 1, acyl-CoA oxidase 1, and 70-kDa peroxisomal membrane protein mRNA levels. D, ACC2 protein levels in basal livers (i.e. without insulin stimulation). E, Hepatic V_PDH/V_TCA flux at the end of the hyperinsulinemic-euglycemic clamp (n = 3–6 per group). White bars, WT; black bars, Thra-0/0 mice. Data are mean ± sem. *, P < 0.05 vs. WT. GAPDH, Glyceraldehyde-3-phosphate dehydrogenase; CPT1; carnitine palmitoyltransferase 1; ACOX, acyl-CoA oxidase; PMP70, 70-kDa peroxisomal membrane protein.

Fig. 6.

CYP7A1 and related genes expression and FGF15 immunoblot. A, mRNA levels of CYP7A1, LXR, farnesoid X receptor (FXR), small heterodimer partner (SHP), hepatic nuclear factor-4 (HNF-4), and pregnane X receptor (PXR). B, Plasma levels (measured by immunoblot) of FGF15 (n = 4–6 per group). White bars, WT; black bars, Thra-0/0 mice. Data are mean ± sem. *, P < 0.05 vs. WT.

Discussion

Thyroid hormone is a key metabolic regulator in glucose and lipid metabolisms. Selectively modulating thyroid receptor α or β is gaining clinical interest and may be a new approach in the treatment of obesity, NAFLD, and type 2 diabetes. In this regard, Thra-0/0 mice have been shown to be protected from high-fat diet-induced obesity (10). The present study is the first to show that Thra-0/0 mice are also protected from high-fat diet-induced hepatic insulin resistance. Improved whole-body insulin sensitivity in the Thra-0/0 mice could be attributed to increased hepatic insulin sensitivity as reflected by increased suppression of endogenous glucose production during the hyperinsulinemic-euglycemic clamp. Improved hepatic insulin sensitivity could be attributed to a more than 50% decrease in hepatic DAG content resulting in decreased PKCε membrane activation and increased insulin signaling as reflected by increased insulin-stimulated p-Akt/Akt (25). Interestingly, these findings corroborate a recent study showing that T₃ promotes insulin-induced glucose uptake in 3T3-L1 adipocytes by enhancing Akt phosphorylation (26). The insulin-sensitive state of Thra-0/0 mice was also due to an increase in insulin-stimulated whole-body glucose disposal, which was, at least partly, due to improved peripheral insulin sensitivity secondary to an increase in insulin-stimulated WAT and BAT glucose uptake. Our results are also consistent with a different mouse model knockout for the thyroid receptor-α (induced by a dominant negative mutation R384C) that was also protected from high-fat diet-induced obesity and had improved glucose tolerance compared with WT mice (27). Interestingly, in this model (27), insulin stimulation led to an increase in muscle glucose uptake, which we did not see in our Thra-0/0 mice, but the reasons for these discrepancies remain unclear. In this study by Sjögren et al. (27), the mutant thyroid receptor-α mediated hypermetabolism by interfering with sympathetic signaling. Because Thra-0/0 mice are whole-body knockout mice, it is impossible to discriminate between primary or secondary effects of the absence of thyroid receptor-α on insulin sensitivity. Notably, the role of thyroid receptor-α on sympathetic signaling to the periphery cannot be specifically addressed in this mouse model. Therefore, mice with tissue-specific deletion of thyroid receptor-α would be of interest to better assess the role of this receptor in peripheral metabolism.

Interestingly, Thra-0/0 also had a slight, but significant, increase in basal rates of endogenous glucose production compared with the WT littermate mice. Although the mechanism responsible for this increase in endogenous glucose production is unclear, it is possible that the decrease in basal plasma insulin concentrations in Thra-0/0 mice was a contributing factor. It is also possible that increased Cori cycling contributed to the increased endogenous glucose production in Thra-0/0 mice, as reflected by the observed approximately 50% increase in plasma lactate concentrations, leading to an increase in hepatic substrate delivery and an increase in hepatic gluconeogenesis. Consistent with this possibility, we did not observe any increases in mRNA or protein expression of the key gluconeogenic enzymes (Supplemental Fig. 1, A–C, published on The Endocrine Society's Journals Online web site at http://endo.endojournals.org).

Thra-0/0 mice were lighter and leaner than their WT littermates, which was consistent with an increase in energy expenditure, even when normalized for different body size. This increase in energy expenditure appears to be related to temperature homeostasis and disappears at thermoneutrality (30 C for mice) (15). The same is true for the lower sensitivity to diet-induced obesity that also disappears at thermoneutrality (10). Interestingly, the survival of Thra-0/0 mice below thermoneutrality relies on continuous food delivery. Indeed, without food, they become rapidly hypothermic when exposed at 4 C and die within 10–20 h (10). Also, we found that Thra-0/0 mice have a higher BAT glucose uptake, which is consistent with increased heat production by this tissue and therefore the increased energy expenditure seen in these mice.

To examine the mechanism responsible for the reduced hepatic triglyceride and DAG content in the Thra-0/0 mice, we assessed hepatic lipogenic and oxidative gene expression. Hepatic lipogenic gene expression was significantly decreased in Thra-0/0 mice, without increase in lipid oxidative genes expression. SREBP1c is a master regulatory transcription factor in lipid synthesis (28). Its mRNA expression and its protein ratio of mature over precursor forms were significantly decreased in Thra-0/0 mice, as was the expression of ACC1, FAS, and, although not significantly, SCD1, which are all LXR target genes. Although we did not detect any alteration in LXR mRNA expression, this may be due to the fact that LXR, as well as other nuclear factors, is largely regulated at the level of ligand availability. LXR signaling is indeed known to have a relationship with the thyroid axis, because LXR null mice show aberrant production of thyroid hormone in liver secondary to increased expression of deiodinases, are defective in hepatic lipid metabolism, and are protected from high-fat diet-induced obesity (29). Because Thra-0/0 mice are known to have increased hepatic deiodinase activity (10) and present a similar phenotype to LXR null mice, i.e. protection from diet-induced obesity and insulin resistance, it is highly probable that LXR activity is down-regulated in Thra-0/0 mice. SREBP1c forms a homodimer that recognizes a sterol response element, which is required along with a thyroid response element to fully activate ACC1 (30), providing another potential explanation of the inhibition of genes downstream of SREBP1c in the absence of the thyroid receptor-α. No difference was found in the mRNA expression of several genes that promote lipid oxidation, confirming the predominance of decreased hepatic lipogenesis in Thra-0/0 mice. Moreover, protein levels of ACC2 were similar between groups, confirming the absence of increased lipid oxidation at a posttranslational level. To further examine whether alterations in hepatic fat oxidation might have contributed to the reduced hepatic triglyceride content in the Thra-0/0 mice, we also examined the relative fluxes of hepatic V_PDH to hepatic V_TCA in vivo by proton-observed carbon-edited nuclear magnetic resonance (21). Consistent with our expression data of oxidative genes in liver, we observed no differences in V_PDH/V_TCA flux, reflecting no differences in the relative rates of hepatic glucose and fat oxidation in Thra-0/0 mice compared with WT mice. Taken together, these findings suggest that decreased hepatic lipogenesis in Thra-0/0 mice is most likely due to decreased expression of SREBP1c and downstream target genes, presumably secondary to LXR down-regulation. A cross talk between SREBP1c and thyroid receptor-α at the DNA level has been demonstrated (31, 32), suggesting that knocking down thyroid receptor-α may result in decreased SREBP1c and downstream targets gene expression and therefore inhibition of lipogenesis. However, SREBP1c null mice are not resistant to high-fat diet-induced obesity (29). Thus, a potential interaction between thyroid receptor-α and SREBP1c does not seem to be a likely explanation to the protection from high-fat diet-induced hepatic insulin resistance in Thra-0/0 mice. Of note, the lower fasting plasma insulin concentrations found in the Thra-0/0 mice may also have contributed to decreased lipogenesis, because insulin is known to induce SREBP1c (33). Moreover, the induction of SREBP1c by insulin is known to be mostly LXR dependent (34), thus emphasizing a potential major role of this gene in the phenotype of the Thra-0/0 mice. It is not clear how lacking thyroid receptor-α leads to decreased plasma insulin concentrations, but recent evidence suggests that liganded thyroid receptor-α leads to β-cell proliferation and insulin secretion (35).

In conclusion, the present study demonstrates that mice specifically lacking the thyroid receptor-α gene are protected from high-fat diet-induced hepatic and peripheral insulin resistance. The former can be attributed to a decrease in hepatic lipogenesis with a subsequent decrease in hepatic DAG content, leading to decreased PKCε activation and improved insulin signaling. Taken together, these findings support the hypothesis that specific inhibition of thyroid receptor-α in the liver may be a promising therapeutic approach to treat NAFLD, hepatic insulin resistance, and type 2 diabetes.