The Lipoprotein Lipase Inhibitor ANGPTL3 Is Negatively Regulated by Thyroid Hormone*

Whereas the role of thyroid hormone is clearly established in the regulation of cholesterol homeostasis, its involvement in the control of serum triglyceride (TG) levels remains largely debated. Angiopoietin-like proteins 3 and 4 have recently been characterized as potent lipoprotein lipase inhibitors and therefore as important components of plasma triglyceride homeostasis. In the present study, the role of thyroid hormone in the regulation of both ANGPTL4 and ANGPTL3 gene expression was investigated. In vivo studies revealed that thyroid hormone down-regulates ANGPTL3 but not ANGPTL4 gene expression in hypothyroid rats. Using thyroid hormone receptor (TR)-deficient mice, we show that thyroid hormone regulates ANGPTL3 gene expression in a TRβ-dependent manner. Transfection studies revealed that this inhibition occurs at the transcriptional level in a DNA binding-independent fashion and requires the proximal (–171 to +66) region of the ANGPTL3 gene promoter. Moreover, site-directed mutagenesis experiments indicate that the HNF1 site within this proximal region mediates this TRβ-dependent repression. Finally, co-transfection studies and electrophoretic mobility shift assays suggest that TRβ antagonizes the HNF1α signaling pathway by inhibiting its transcriptional activity without interfering with its DNA-binding capacity. Taken together, our results lead to the identification of ANGPTL3 as a novel TRβ target gene and provide a new potential mechanism to explain the hypotriglyceridemic properties of TRβ agonists in vivo.

Thyroid hormone (T3) 2 plays major roles in the development and adult functions of many organs and tissues. Most of the effects of the thyroid hormone are mediated by the thyroid hormone receptors (TRs), which are members of the nuclear receptor superfamily (1). Two distinct genes, THRA and THRB (encoding TR␣ and TR␤, respectively) produce various forms of TR proteins, including the functional receptors TR␣1, TR␤1, and TR␤2 (2). Whereas TR isoforms are detectable in almost every tissue, there is some isoform-specific pattern of distribution. TR␣1 is highly expressed in skeletal muscle and brown fat, and TR␣2 expression is mostly restricted to the brain. The almost ubiquitous TR␤1 is mainly expressed in the brain, liver, and kidney, whereas TR␤2 is exclusively present in the pituitary and hypothalamus in adults. All of the TR isoforms except TR␣2 bind T3 with a similar affinity (2). TRs regulate gene transcription through their binding to specific TR-response elements (TRE) located in the promoter regions of target genes. TRs can bind in the absence of T3 as homodimers but preferentially bind those TRE as heterodimers with another nuclear receptor, namely the retinoid X receptor. TR can positively or negatively regulate T3-responsive gene expression, depending on the nature of the TRE, the hormonal status, and the cellular environment (2). In the absence of T3, the TR/retinoid X receptor heterodimer is associated with corepressors at the TRE. Those corepressors interact with multisubunit protein complexes containing histone deacetylase (HDAC) activity that maintain the chromatin in a compact state repressing gene activation. Upon T3 binding, TR undergoes a conformational change, releasing corepressors and allowing coactivator binding, thereby promoting gene activation through histone acetylation. The transcription of any given gene is defined by the control of the interactions between TR and those co-regulatory proteins (for reviews, see Refs. 3 and 4).
The use of transgenic and knock-out mouse models revealed the existence of TR isoform-specific functions (see Ref. 5 for review). Whereas it has been shown that TR␣ plays a key role in cardiac function by regulating both heart rate and contractility, TR␤ has been characterized as a master regulator of the hypothalamus-pituitary-thyroid feedback and an important regulator of cholesterol homeostasis (5). In fact, the link between thyroid disease and lipid disorders has been established several decades ago (6). Hypothyroidism is characterized by elevated low density lipoprotein levels as well as an increased risk of atherosclerosis (7,8). Furthermore, administration of thyroid hormone efficiently lowers plasma cholesterol levels in rodent models and also in humans (9,10). T3 regulates cholesterol homeostasis by affecting the expression of key hepatic proteins, such as the low density lipoprotein receptor and the cholesterol 7␣-hydroxylase. Using transgenic mice, Gullberg and co-workers (11) have demonstrated that T3 regulates cholesterol homeostasis in mice in a TR␤-dependent manner. In addition, those authors showed that TR␤ regulates the transcription of CYP7A1, the rate-limiting step in bile acid biosynthesis (12). Recently, Johansson et al. (13) demonstrated that TR␤ activation by a synthetic thyromimetic (GC-1) results in a sharp reduction in total cholesterol levels, in agreement with previous findings (14 -16). This strong effect is explained, at least in part, by a stimulation of the reverse cholesterol transport: increase of high density lipoprotein selective uptake in the liver by SR-B1 and a subsequent increase in cholesterol catabolism through Cyp7A activation (13). Interestingly, TR␤ activation by thyromimetics has been shown to lower plasma triglyceride levels in various rodent models (14 -16). In hypothyroid patients, serum levels of TG-rich lipoprotein were reported to be increased (17). Moreover, Ito et al. (18) showed that administration of T4 to patients with hypothyroidism led to a strong decrease in low density lipoprotein-cholesterol and also triglyceride lev-els. This decrease was accompanied by a significant increase in lipoprotein lipase (LPL) activity. In another study, Nikkila and Kekki (19) found that postheparin LPL activity is increased in hyperthyroid state and decreased in hypothyroidism. Finally, Abrams et al. (17) observed that hyperthyroid patients have remarkable facility in clearing VLDL-TG without any significant changes in LPL activity. Altogether, these reports suggest that thyroid hormone influences triglyceride metabolism, but the precise molecular mechanism remains to be elucidated.
Recently, novel regulators of LPL activity have been identified. Angiopoietin-like proteins are proteins containing a coiled-coil domain and a fibrinogen-like domain similar to those found in angiopoietins (20). Six members of this growing family have been described so far (for a review, see Ref. 20). In addition to their angiogenic properties in the vasculature, members of this family have been found to play a role in metabolism (20). ANGPTL3 was identified by genetic analysis of the KK/San mice that present low plasma lipid levels despite being hyperinsulinemic and hyperglycemic (21). Administration or overexpression of ANGPTL3 using adenovirus elicited a marked increase in circulating plasma total cholesterol, nonesterified fatty acid levels and TGs (21). ANGPTL3 regulates VLDL-TG levels by inhibiting LPL activity (22). Another study also proposed that ANGPTL3 may be able to target adipocytes to activate lipolysis, thereby enhancing the release of free fatty acids and glycerol (23). More recently, a significant genetic association was found between ANGPTL3 and the development of atherosclerosis lesions in both mice and humans (24). The patients carrying the mutated ANGPTL3 allele had higher circulating TG levels (24). Another member of the Angiopoietin-like proteins, ANGPTL4, has also been found to inhibit LPL activity (25). This gene, also called PGAR (peroxisome proliferator-activated angiopoietin-related protein) or FIAF (fasting-induced adipose factor), is tightly regulated in the liver and in the adipose tissue after feeding and fasting, suggesting a role in fat metabolism (26). Furthermore, transgenic mice overexpressing ANGPTL4 in the heart showed a significant reduction in cardiac LPL activity and a subsequent inhibition in lipoprotein-derived free fatty acid delivery to the heart (27).
In this study, we investigated the potential involvement of thyroid hormone in ANGPTL3 and ANGPTL4 gene regulation. ANGPTL3, but not ANGPTL4, was found to be regulated specifically by TR␤ in vivo. Furthermore, we show that TR␤ negatively regulates this gene by, at least in part, antagonizing the HNF1␣ transcriptional pathway in a DNA binding-independent manner.
Plasmids-The plasmids pSG5-TR␤ and GAL4-TR␤ (containing the human TR␤ ligand-binding domain cloned in frame with the DBD of the yeast GAL4 transcription factor) were provided by Dr. J. Moore (GlaxoSmithKline, Research Triangle Park, NC). The pSG5 plasmid was purchased from Stratagene (La Jolla, CA). The G5-Luc reporter vector containing five copies of the GAL4 response element cloned upstream of the minimal thymidine kinase promoter was provided by Promega. The human ANGPTL3 promoter constructs (Ϫ929 to ϩ109, Ϫ296 to ϩ109, Ϫ171 to ϩ109, and Ϫ66 to ϩ109) were obtained by PCR amplification using human genomic DNA (Clontech) as template. The result-ing PCR products were inserted as BglII/HindIII fragments into pGL3 basic vector (Promega), yielding Ϫ929 to ϩ109, Ϫ296 to ϩ109, Ϫ171 to ϩ109, and Ϫ66 to ϩ109 ANGPTL3-Luc. The mutation of the LXR and HNF1 binding sites within the Ϫ929 to ϩ109 ANGPTL3 promoter were obtained by site-directed mutagenesis (Stratagene, La Jolla, CA). The GAL4-HNF1␣ construct was obtained by PCR amplification of fulllength HNF1␣ using a human liver cDNA library (Clontech) as template. The resulting PCR fragment was cloned in frame with the GAL4 DNA-binding domain into the pM vector (Clontech), yielding GAL4-HNF1␣. All constructs were verified by DNA sequence analysis.
Transient Transfection Assays-HepG2 cells, plated in 96-well plates at 50 -60% confluence in basic Eagle's medium supplemented with 10% fetal calf serum, were transiently transfected with reporter and receptor expression plasmids using Fugene 6 reagent (Roche Applied Science) as indicated in the legends for Figs. 3-7. The pSEAP2 expression plasmid (Clontech) was cotransfected to assess transfection efficiency. 24 h posttransfection, cells were refed with fresh medium containing 1% charcoal/dextran-treated serum and TR ligands or vehicle (Me 2 SO or 0.1% ethanol). 24 h later, cells were collected and assayed for luciferase and alkaline phosphatase activities. All experiments were repeated at least three times. Results are expressed as mean Ϯ S.E.
Animal Studies-Experimental protocols were approved by the GlaxoSmithKline Institutional Animal Care and Use Committee. Male Wistar rats (12 weeks old) were obtained from Charles River Laboratories. Hypothyroidism was induced by administration by gavage of a 10-mg/kg/day 6-n-propyl-2-thiouracil (PTU) aqueous solution (Sigma) for 3 weeks. On the last day of the treatment, half of the PTU-treated rats received an intraperitoneal injection of L-T3 (300 g/kg in 100 l of PBS). 6 h postinjection, animals were sacrificed, and livers were excised for RNA preparation. Adult 8 -10-week-old males TR␣ 0/0 (29) and TR␤ Ϫ/Ϫ (30) as well as wild type (WT) mice were used. Mice were housed and maintained with approval from the animal experimental committee of the Ecole Normale Supérieure de Lyon (Lyon, France). Briefly, for each genotype, control animals were fed standard mouse chow. TH deficiency was induced by feeding a low iodine diet supplemented with 0.15% PTU purchased from Harlan/Teklad and 0.05% methimazole into the drinking water. The day before the sacrifice (5 pm), one-half of the PTU-treated mice received an intraperitoneal injection of T3 (300 g/kg in 100 l of PBS). The PTU-and TH-treated mice were sacrificed at the end of the night cycle after an overnight fasting. At the conclusion of each experiment, blood was recovered for serum preparation, and the liver was quickly removed, frozen in liquid nitrogen, and used for RNA extraction.
RNA Analysis-Total RNA was extracted using TRIZOL reagent (Invitrogen) following the manufacturer's instructions. The RNA was treated with DNase I (Ambion Inc., Austin, TX) at 37°C for 30 min, followed by inactivation at 75°C for 5 min. Real time quantitative PCR assays were performed using an Applied Biosystems 7900 sequence detector. Total RNA (1 g) was reverse transcribed with random hexamers using a Taqman reverse transcription reagent kit (Applied Biosystems) following the manufacturer's protocol. RNA expression levels were determined by Sybr green assays as described (31). ␤-Actin transcript was used as an internal control to normalize the variations for RNA amounts. Gene expression levels are expressed relative to ␤-actin mRNA levels. All of the PCR primers used in the present study are available upon request.
Electrophoretic Mobility Shift Assay-Double-stranded oligonucleotides (5Ј-GCTTAATGATTAACT-3Ј) were end-labeled with [␥-32 P]ATP using T4 polynucleotide kinase according to standard procedures. Nuclear extracts were prepared as described (32,33), and protein concentration was determined using the BCA kit (Bio-Rad). 5 g of extracts were incubated with 100,000 cpm of labeled probe for 30 min at room temperature in 20 l of buffer containing 10 mM Tris (pH 7.5), 50 mM NaCl, 1 mM dithiothreitol, 1 mM EDTA, 5% glycerol, 0.3 g bovine serum albumin, and 2 g of poly(dI-dC). DNA-protein complexes were analyzed by electrophoresis in a 5% nondenaturing polyacrylamide gel with 0.5ϫ TBE buffer. The gel was then dried and exposed at Ϫ80°C for autoradiography.

RESULTS AND DISCUSSION
In order to study a potential involvement of thyroid hormone in ANGPTL3 and ANGPTL4 gene regulation, male Wistar rats were rendered hypothyroid using a 3-week PTU treatment. This treatment induced a dramatic fall in both T4 and T3 levels. Animals then received T3 by intraperitoneal injection or vehicle (PBS). Hepatic gene expression was monitored 6 h later by real time quantitative PCR (Fig. 1). As expected, T3 injection resulted in a significant induction of SPOT14 gene expression (3.5-fold) in the hypothyroid rats in agreement with previous reports (34). Interestingly, ANGPTL3, but not ANGPTL4, gene expression was markedly down-regulated by the treatment (Ϫ70%), suggesting a role for thyroid hormone in the control of ANGPTL3 expression in vivo.
To determine which TR isotype mediates this effect, we next measured ANGPTL3 gene expression in WT, TR␣ KO, and TR␤ KO mice that had been subjected to PTU treatment for 3 weeks. At the end of this treatment, hypothyroid mice received T3 (300 g/kg) by intraperitoneal injection. 16 h later, mice were sacrificed, and hepatic gene expressions were monitored by Sybrman analysis (Fig. 2). As expected, SPOT14 was strongly (60-fold) induced in WT and TR␣ KO mice, but not in TR␤ KO mice ( Fig. 2A), which is consistent with previous reports (34). ANGPTL3 mRNA levels were significantly diminished after T3 injec-tions in both WT and TR␣ KO mice. By contrast, this repressive effect of T3 on ANGPTL3 expression was not observed in TR␤-deficient animals (Fig. 2B). Finally, as a control, liver ANGPTL4 mRNA levels were not significantly affected by T3 injection (Fig. 2C). Taken together, these results demonstrate that ANGPTL3 is negatively regulated by T3 in vivo in a TR␤-dependent manner.
In order to understand by which mechanism TR␤ regulates ANGPTL3 gene expression, a 1-kb fragment of the promoter was cloned upstream of the luciferase gene. A functional analysis of this promoter was then carried out in HepG2 cells (Fig. 3). Preliminary experiments revealed that HepG2 cells express extremely low TR␤ levels, as demonstrated by Taqman analysis (data not shown) and in line with previous reports (35,36). In the absence of co-transfected TR␤, treatment with T3 or thyromimetics (GC-1 and CGS-23425) did not significantly affect ANGPTL3 promoter activity (Fig. 3A). Co-transfection of TR␤ in the absence of ligand resulted in a small but significant inhibition of the reporter gene activity. This effect was potentiated by the addition of the different agonists, resulting in a 10-fold repression of the promoter activity (Fig. 3A). Furthermore, this ligand-induced transcriptional inhibition was dose-dependent, as demonstrated on Fig. 3B. This inhibition is also specific, since co-transfection of another nuclear receptor, peroxisome proliferator-activated receptor ␣, in the presence or absence of its synthetic ligand, failed to suppress ANGPTL3 promoter activity (data not shown). Taken together, the results indicate that T3 regulates ANGTPL3 at the transcriptional level. Since T3 has also been shown to regulate gene expression via mRNA destabilization (37), mRNA half-life studies were performed in rat primary hepatocytes treated with actinomycin D in the presence of T3 (100 nM) or vehicle. In those experiments, T3 did not affect ANGPTL3 mRNA half-life (around 20 h) (data not shown), indicating again that T3 regulates ANGPTL3 gene expression at the transcriptional level.
To determine whether T3-mediated ANGPTL3 promoter repression requires TR␤ DNA binding, we next evaluated the influence of a TR␤ construct lacking the DBD on ANGPTL3 promoter activity. Co-transfection of increasing concentrations of TR␤ ligand-binding domain construct resulted in a dose-dependent inhibition of the ANGPTL3 promoter (Fig. 4B). This inhibition was further augmented by the addition of T3 similarly as reported for the full-length receptor (Fig. 4A). This finding strongly suggests that TR␤ negatively regulates ANGPTL3 promoter activity in a DNA binding-independent manner.
To localize which ANGPTL3 promoter region confers TR␤ responsiveness, 5Ј deletions of the promoter were transiently transfected into HepG2 cells. As a control, co-transfection of TR␤ resulted in a liganddependent inhibition of the 1-kb promoter construct as previously shown (Figs. 3 and 4). The Ϫ296 to ϩ109 and Ϫ171 to ϩ109 promoter  constructs were also strongly inhibited by the T3 addition when TR␤ was co-transfected (Fig. 5). By contrast, the Ϫ66 to ϩ109 promoter activity was not repressed by the liganded TR␤ (Fig. 5), indicating that the Ϫ171 to ϩ66 promoter region confers TR␤ responsiveness. A bioinformatic analysis revealed the presence of putative LXR and HNF1 binding sites within this region, in agreement with a previous report (38). Since TR␤ represses ANGPTL3 promoter activity in a DNA binding-independent manner, we hypothesized that liganded TR␤ could inhibit the transcription via a trans-repression mechanism. Therefore, both binding sites were mutated by site-directed mutagenesis, and the resulting promoter constructs were tested in HepG2 cells. The construct bearing the LXR binding site mutation was significantly inhibited by T3 treatment in TR␤-expressing HepG2 cells similarly as described for the wild type promoter (Fig. 6). By contrast, mutation of the HNF1 site completely abolished TR␤-mediated promoter repression (Fig. 6), indicating that this site is required for TR␤-mediated ANGPTL3 promoter regulation. Of note, this construct has a significant lower basal promoter activity compared with the WT promoter, which is consistent with HNF1 playing a role in ANGPTL3 gene regulation (data not shown).
Next, we investigated the molecular basis of this cross-talk between TR␤ and the HNF1 signaling. First, we tested whether TR␤ could repress HNF1␣ transcriptional activity independently of the promoter context. Therefore, the effect of TR␤ activation was analyzed on a GAL4-dependent reporter, activated by a GAL4-HNF1␣ chimera. Co-transfection of the GAL4-HNF1␣ chimera resulted in a strong transcriptional activation (170fold), which was significantly (Ϫ70%) inhibited by TR␤ co-transfection in a ligand-dependent manner (Fig. 7A). This result suggests that TR␤ can repress HNF1-driven transcription independently of the promoter context. As a control, TR␤ activation did not influence the activity of the GAL4 DBD alone (Fig. 7A), and co-transfection of peroxisome proliferator-activated receptor ␣ did not affect HNF1-driven promoter activity, demonstrating the specificity of the observed effects (data not shown). We then assessed   whether TR␤ activation resulted in a loss of HNF1␣ DNA binding using electrophoretic mobility shift assays. In a first set of experiments, we verified that HNF1 was binding to the predicted site we identified in ANGPTL3 promoter. Incubation of HepG2 nuclear extracts with this 32 P-radiolabeled oligonucleotide containing the HNF1 site resulted in a strong protein-DNA complex, which was significantly reduced by a preincubation with a polyclonal HNF1␣ antibody (Fig. 7B). HNF1␣ binding to this site was subsequently examined in HepG2 cells transfected with TR␤ or empty vector (pSG5). In pSG5-or TR␤-transfected cells, HNF1␣ DNA binding was not modified by T3 treatment (Fig. 7B), suggesting that TR␤ does not repress HNF1␣ transcriptional activity by preventing its DNA binding.
In the present study, we have shown that T3 down-regulates ANGPTL3 but not ANGPTL4 gene expression in vivo in a TR␤-dependent manner. The role of the ␤ isotype is not surprising, since it is the prevalent TR in the liver, representing 85% of the T3-binding activity (39). Using promoter mapping experiments, a region conferring T3 responsiveness was identified within the proximal promoter. Interestingly, this region contains an LXR response element, a DNA-binding site that can be shared by TR␤ (40). Recently, two groups suggested that unliganded TR represses LXR activation on an LXR response element by competing for the same binding site (41,42). In this study, TR␤ was found to inhibit ANGPTL3 promoter activity in a strictly ligand-dependent manner (Fig. 3). Furthermore, this inhibition occurs in a DNA binding-independent fashion (Fig. 4B). Finally, a construct mutated for the LXR response element was repressed by co-transfection of TR␤ in the presence of T3 (Fig. 6). Altogether, those results strongly suggest that TR␤ down-regulates ANGPTL3 gene expression in a LXR-independent manner.
This proximal promoter region contains also an HNF1 binding site whose presence is required for TR␤ to mediate this transcriptional repression. Transfection studies and electrophoretic mobility shift assays suggest that TR␤ can antagonize HNF1␣ transcriptional activity independently of the promoter context without interfering with its DNA-binding capacity. Interestingly, Caturla et al. (43) suggested that T3 down-regulates the mouse ␣-foetoprotein promoter activity by competing for DNA binding with HNF1 and CCAAT/enhancer-binding proteins. Several hypotheses could be made to explain how TR␤ affects HNF1␣-driven transcription. First, TR␤ could be recruited to the negatively regulated gene via an interaction with HNF1␣, thereby preventing the recruitment of coactivator(s) and/or allowing the binding of co-repressor(s), similar to what has been described for the AP-1/TR cross-talk (44,45). However, co-immunoprecipitation experiments did not allow us to detect any significant association between HNF1␣ and TR␤ (data not shown). In addition, electrophoretic mobility shift assays did not reveal any interference of HNF1 DNA binding in the presence of co-transfected TR␤ (Fig. 7B). Moreover, TR␤ does not repress all of the HNF1-driven gene expression, indicating some gene-specific mechanism. Altogether, these data strongly argue against a direct HNF1-TR␤ interaction. It has been shown that TR␤ is able to recruit corepressors and HDAC activity on certain gene promoters. For instance, TR␤ has been shown to recruit histone deacetylase in a ligand-dependent fashion on the thyrotropin ␤ gene (46). We tested a potential implication of HDAC in ANGPTL3 gene regulation. Whereas the potent HDAC inhibitor, trichostatin A, increased ANGPTL3 promoter activity in a dosedependent manner, it did not prevent TR␤-mediated transcriptional repression (data not shown), ruling out a potential involvement of HDACs in this transcriptional cross-talk. However, TR␤ could sequester a limiting co-activator or co-repressor specifically recruited by HNF1␣ on the ANGPTL3 promoter. This is the so-called squelching model (47,48). The identification of such factor(s) will require further experiments.
The central role of T3 in cholesterol homeostasis is well documented (6). By contrast, its involvement in the regulation of serum TG levels remains largely debated. Nevertheless, a recent study FIGURE 6. The HNF1 site within the ANGPTL3 promoter is required for TR␤-dependent repression. HepG2 cells were transfected with wild type or LXR response elementmutated or HNF1 response element-mutated promoter construct (40 ng) and TR␤ or empty vector (pSG5; 10 ng). 24 h later, cells were refed with fresh medium supplemented with 1% charcoal/dextran-treated FBS in the presence of T3 (100 nM) or vehicle (ethanol 0.1%) for an additional 24 h. Results are plotted as repression levels. *, p Ͻ 0.05. FIGURE 7. TR␤ antagonizes HNF1␣ transcriptional activity without interfering with its DNA binding. A, HepG2 cells were transfected with G5-Luc (40 ng), GAL4-HNF1␣ (5 ng), and TR␤ (5 ng) or empty vectors (GAL4 DBD and pSG5, respectively). 24 h later, cells were refed with fresh medium supplemented with 1% charcoal/dextran-treated FBS in the presence of T3 (100 nM) or vehicle (0.1% ethanol) for an additional 24 h. B, 5 g of nuclear extracts prepared from HepG2 cells transfected with pSG5 or TR␤ and treated for 24 h with T3 (100 nM) or vehicle (0.1% ethanol) were subjected to electrophoretic mobility shift assay using the HNF1 site radiolabeled probe. Shifted DNA-protein complexes were visualized by autoradiography. The HNF1-DNA complex was significantly reduced by incubating HepG2 nuclear extracts with 1 l of HNF1␣ antibody for 10 min. N.E., nuclear extracts. APRIL 28, 2006 • VOLUME 281 • NUMBER 17

TR␤ Down-regulates ANGPTL3 Expression
showed that administration of a TR␤-selective agonist results in a strong decrease of VLDL-TG in rodent models (13). These authors suggested that TR␤ activation leads to SREBP1c gene suppression, thereby inhibiting hepatic fatty acid and TG synthesis (13). Transcriptome analysis comparing TR␤ KO versus wild type mice revealed that T3-mediated fatty acid synthase and stearoyl-CoA desaturase gene regulation was rapid and transient (34). In this study, we were able to identify ANGPTL3 as a novel target gene for T3 in vivo. ANGPTL3 is a hepatic secreted protein with structural similarity to angiopoietins (50). ANGPTL3 was identified by genetic analysis of the KK/San mice that present low plasma lipid levels despite being hyperinsulinemic and hyperglycemic (21). Administration or overexpression of ANGPTL3 using adenovirus elicited a marked increase in circulating plasma total cholesterol, nonesterified fatty acid levels, and TGs (21). ANGPTL3 regulates VLDL-TG levels by inhibiting LPL activity (22). Another study also proposed that ANGPTL3 may be able to target the adipocytes and to activate lipolysis, thereby enhancing the release of free fatty acids and glycerol (23). More recently, a significant genetic association was found between ANGPTL3 and the development of atherosclerosis lesions in both mice and humans (24). The patients carrying the mutated ANGPTL3 allele had higher circulating TG levels (24). Another member of the angiopoietin-like proteins, ANGPTL4, has also been found to inhibit LPL activity (25). In this study, we found that T3 regulates ANGPTL3 but not ANGPTL4 gene expression in rats and also in mice ( Fig. 1 and 2). Since ANGPTL3 behaves as an LPL inhibitor, it is tempting to speculate that T3 and its synthetic analogues lower plasma TG levels by suppressing ANGPTL3 expression, thereby activating plasma LPL activity. Moreover, recent studies suggested that apolipoprotein A-5, which accelerates VLDL-TG hydrolysis via LPL activation (51,52), was up-regulated by T3-or TR␤-selective molecules in vitro and also in vivo (53). Altogether, these results suggest that T3 may increase LPL activity by different mechanisms, thereby resulting in a sharp reduction in plasma TG. To determine the biological relevance of T3-mediated ANGPTL3 regulation, it would be interesting to administer a TR␤ agonist to ANGPTL3-deficient mice and to measure TG levels during this experiment. However, ANGPTL3-deficient mice display already an increased LPL activity (9-fold higher compared with wild type animals) and also a severe hypotriglyceridemia in the fed state (49). Therefore, it will be difficult to assess the TG-lowering activity of TR␤ agonist in such animal models.
In conclusion, this study led to the identification of ANGPTL3 as a negatively regulated gene by T3 in vivo. In addition to its central role in cholesterol homeostasis, TR␤ seems to regulate an array of proteins involved in TG metabolism. Therefore, compounds targeting this receptor and devoid of the deleterious effects of T3 may be useful therapeutic agents for the treatment of lipid disorders and obesity.