Role of Sirtuin 1 in the Regulation of Hepatic Gene Expression by Thyroid Hormone*

Background: Sirtuin 1 elevates the expression of genes involved in hepatic fatty acid oxidation. Results: Sirtuin 1 modulates the thyroid hormone regulation of the cpt1a, pdk4, and srebp-1c genes. Conclusion: Sirtuin 1 coregulates the thyroid hormone receptor-mediated induction of gene expression. Significance: Activators of sirtuin 1 and thyroid hormone receptor β agonists could cooperatively stimulate fatty acid oxidation and inhibit lipogenesis. Sirtuin 1 (SIRT1) is a nuclear deacetylase that modulates lipid metabolism and enhances mitochondrial activity. SIRT1 targets multiple transcription factors and coactivators. Thyroid hormone (T3) stimulates the expression of hepatic genes involved in mitochondrial fatty acid oxidation and gluconeogenesis. We reported that T3 induces genes for carnitine palmitoyltransferase (cpt1a), pyruvate dehydrogenase kinase 4 (pdk4), and phosphoenolpyruvate carboxykinase (pepck). SIRT1 increases the expression of these genes via the activation of several factors, including peroxisome proliferator-activated receptor α, estrogen-related receptor α, and peroxisome proliferator-activated receptor γ coactivator (PGC-1α). Previously, we reported that PGC-1α participates in the T3 induction of cpt1a and pdk4 in the liver. Given the overlapping targets of T3 and SIRT1, we investigated whether SIRT1 participated in the T3 regulation of these genes. Resveratrol is a small phenolic compound whose actions include the activation of SIRT1. Addition of resveratrol increased the T3 induction of the pdk4 and cpt1a genes in hepatocytes. Furthermore, expression of SIRT1 in hepatocytes mimicked resveratrol in the regulation of gene expression by T3. The deacetylase activity of SIRT1 was required and PGC-1α was deacetylated following addition of T3. We found that SIRT1 interacted directly with T3 receptor (TRβ). Knockdown of SIRT1 decreased the T3 induction of cpt1a and pdk4 and reduced the T3 inhibition of sterol response element binding protein (srebp-1c) both in isolated hepatocytes and in rat liver. Our results indicate that SIRT1 contributes to the T3 regulation of hepatic genes.


Sirtuin 1 (SIRT1) is a nuclear deacetylase that modulates lipid metabolism and enhances mitochondrial activity. SIRT1 targets multiple transcription factors and coactivators. Thyroid hormone (T 3 ) stimulates the expression of hepatic genes involved in mitochondrial fatty acid oxidation and gluconeogenesis.
We reported that T 3 induces genes for carnitine palmitoyltransferase (cpt1a), pyruvate dehydrogenase kinase 4 (pdk4), and phosphoenolpyruvate carboxykinase (pepck). SIRT1 increases the expression of these genes via the activation of several factors, including peroxisome proliferator-activated receptor ␣, estrogen-related receptor ␣, and peroxisome proliferator-activated receptor ␥ coactivator (PGC-1␣). Previously, we reported that PGC-1␣ participates in the T 3 induction of cpt1a and pdk4 in the liver. Given the overlapping targets of T 3 and SIRT1, we investigated whether SIRT1 participated in the T 3 regulation of these genes. Resveratrol is a small phenolic compound whose actions include the activation of SIRT1. Addition of resveratrol increased the T 3 induction of the pdk4 and cpt1a genes in hepatocytes. Furthermore, expression of SIRT1 in hepatocytes mimicked resveratrol in the regulation of gene expression by T 3 . The deacetylase activity of SIRT1 was required and PGC-1␣ was deacetylated following addition of T 3

. We found that SIRT1 interacted directly with T 3 receptor (TR␤). Knockdown of SIRT1 decreased the T 3 induction of cpt1a and pdk4 and reduced the T 3 inhibition of sterol response element binding protein (srebp-1c) both in isolated hepatocytes and in rat liver. Our results indicate that SIRT1 contributes to the T 3 regulation of hepatic genes.
Thyroid hormone (T 3 ) 2 regulates the expression of numerous hepatic genes through both genomic and more rapid non-genomic actions (1,2). The genomic actions of T 3 are mediated by the binding of the T 3 receptor (TR) to T 3 response elements (TRE) in gene promoters and the subsequent recruitment of various coactivators by the liganded receptor. The primary mechanistic model for TR suggests that the unliganded receptor is bound to DNA and associated with corepressors and histone deacetylases, whereas the addition of ligand leads to the recruitment of coactivators (3). The two isoforms of TR include TR␣ and TR␤, which is the isoform most highly expressed in the liver (4). Previously, we reported that T 3 induced the expression of carnitine palmitoyltransferase 1a (cpt1a), phosphoenolpyruvate carboxykinase (pepck), and pyruvate dehydrogenase kinase 4 (pdk4) via TREs in their promoters (5,6). However, binding of the liganded TR␤ was not sufficient for full activation of gene expression by T 3 . Additional factors including the CCAAT enhancer binding protein ␤ (C/EBP␤), cAMP-responsive element-binding protein binding protein, and peroxisome proliferator-activated receptor ␥ coactivator (PGC-1␣) participate in the T 3 induction of pdk4, pepck, and cpt1a genes (6 -8). These genes are regulated by multiple nuclear receptors and transcription factors that recruit coactivators in a combinatorial manner (9).
CPT1a catalyzes the translocation of long chain acyl-CoA to acyl-carnitine and is an initiating step in mitochondrial fatty acid oxidation (10). Although the cpt1a TRE is located 3000 bp 5Ј to the start site of transcription, elements in the first intron are required for the T 3 induction (5). In part, these intronic elements act by the recruitment of PGC-1␣, which is needed for the T 3 response (7). The peroxisome proliferator activated receptor (PPAR␣)-induced genes involved in mitochondrial oxidation of fatty acids, including cpt1a (11,12). The hepatic gluconeogenic genes, pepck and glucose-6-phosphatase, are regulated by the forkhead transcription factor (FoxO1), hepatic nuclear factor 4, and the glucocorticoid receptor (13), whereas expression of the pdk4 gene is stimulated by the estrogen-re-* This work is supported by National Institutes of Health Grants DK0059368 lated receptor (ERR␣) and PPAR␣ (14,15). PGC-1␣, which stimulates pepck and pdk4, interacts with PPAR␣, FoxO1, glucocorticoid receptor, and ERR␣ (16).
The NAD-dependent deacetylase sirtuin 1 (SIRT1) regulates lipid and carbohydrate metabolism and provides resistance to metabolic diseases (17). There are seven human sirtuin genes from SIRT1 to SIRT7 (18). SIRT1 regulates metabolic processes by deacetylating important transcriptional regulators, including PGC-1␣ (19), FoxO1 (20,21), liver X receptor ␣ (22), farnesoid X receptor (23), ERR␣ (24), and sterol receptor element binding protein (SREBP-1c) (25). SIRT1 interacts with PPAR␣ and hepatocyte specific deletion of SIRT1 impairs PPAR␣ activity and subsequently decreases fatty acid oxidation in the liver (26). SIRT1 enhances the transcriptional activation of PGC-1␣ via the deacetylation of multiple lysines (19). Resveratrol activates SIRT1 although this activation may involve AMP kinase (AMPK) (27). Activation of SIRT1 by administration of resveratrol improved mitochondrial function, increased fatty acid oxidation, and decreased lipogenesis (17,28,29). SIRT1 and T 3 stimulate many common hepatic gene targets. In addition, PGC-1␣ is activated by SIRT1 and PGC-1␣ participates in the T 3 induction of several genes in the liver (19,30). In light of this potential relationship, we examined the role of SIRT1 in the T 3 induction of cpt1a, pdk4, and pepck. Our data indicate that SIRT1 enhanced the T 3 stimulation of these genes and that SIRT1 contributes to the regulation of hepatic gene expression by T 3 .

MATERIALS AND METHODS
Transient Transfection of Luciferase Vectors-Luciferase constructs were transfected into HepG2 or HeLa cells along with the mammalian expression vectors SV40-TR␤ and TK-Renilla by calcium phosphate precipitation as described previously (7). Cells were maintained in DMEM containing 5% calf serum and 5% fetal calf serum. After overnight incubation, the medium was replaced by DMEM without serum, and the cells were treated with T 3 or resveratrol for 24 h as indicated in the figure legend. Transfected cells were harvested in passive lysis buffer (Promega, Madison, WI). Luciferase and Renilla luciferase activity were measured with the dual luciferase reporter kit (Promega). Protein content in the lysates was determined by Pierce BCA protein assay kit (Thermo Scientific, Rockford, IL). Luciferase activity was normalized for protein content and Renilla luciferase activity.
Real-time PCR-The cDNA for real time PCR was prepared using RNA isolated from cells using RNA-Stat-60 (Tel-test, Friendswood, TX) (32) and cleaned with Qiagen RNEasy kit. The RNA concentration of each sample was measured with a Nano drop (Thermo Scientific). Two g of DNA-free RNA was converted to cDNA using Superscript reverse transcriptase III and random hexamers (Invitrogen). The parameters for real time PCR were as follows: 95°C for 5 min, 40  Adenoviral Infection-Adenovirus encoding shRNA specific for SIRT1 (AdshSIRT1) and scrambled shRNA (AdScr) and pADEasy Flag-SIRT1H355A plasmid were generous gifts of Dr. P. Puigserver (33). The adenovirus were amplified in HEK293 cells and purified by cesium chloride centrifugation (34). Rat hepatocytes were plated at a density of 3 ϫ 10 6 in a 60-mm dish in RPMI 1640 medium. Five hours after plating, the medium was changed, and the cells were infected with adenovirus Adsh-SIRT1 or AdScr. Similarly, cells were infected with adenovirus expressing SIRT1 (AdSIRT1), adenovirus expressing deacetylase dead SIRT1 mutant (AdSIRT1H355A) or nonspecific adenovirus (35). Medium was changed after 16 h of infection, and T 3 was added for 24 h in serum-free medium. Cells were harvested for RNA and protein.
For the adenoviral infection of rats, the rats were placed on a propylthiouracil containing diet (Teklad 95125) for 5 weeks. Rats were injected via the tail vein with 10 11 pfu/kg of adenovirus. Three days after infection, the rats were given T 3 injections intraperitoneally at a dose of 0.33 mg/kg for 2 days. RNA and protein were obtained from the harvested liver for further analysis.
Chromatin Immunoprecipitation Assay-The ChIP assays were done according to the protocol provided with Millipore magna ChIP kit . Rat hepatocytes were treated with 100 nM T 3 for 24 h in serum-free RPMI 1640 medium. Cells were cross-linked with 1% formaldehyde for 10 min at room temperature, and the cross-linking reaction was stopped with 0.125 M glycine for 5 min. The lysate was prepared and sonicated as described previously (6). Chromatin preparations were diluted with dilution buffer and protease inhibitor mixture and designated as input samples. Immunoprecipitation was done with antibodies against control mouse IgG (sc2025, Santa Cruz Biotechnology) and anti-SIRT1 (A21993, Invitrogen). Eluted DNA was purified with the PCR purification kit (Qiagen, 28104). DNA was subjected to 30 cycles of PCR using 3-5 l of DNA. PCR products were analyzed on 2% NuSieve 3:1 agarose (Lonza, Walkersville, MD) and visualized with MultiImage light cabinet with Alphaimager EP software. The following primers were used to amplify portions of the cpt1a promoter: upstream region (Ϫ6473/Ϫ6450) ctctggtgtcctgtaacctgtgg, (Ϫ6076/ Ϫ6099), gaaggctggaattacactggtcag, TRE (Ϫ3079/Ϫ3056) gacaggcagggtacatttcacag, (Ϫ2802/Ϫ2825) gaaggcagtgcttttccctac, first intron (ϩ674/ϩ693) agactgctcaaggtcgcgct, (ϩ907/ ϩ927) gagagtcctgagcctgattgt.
PGC-1␣ Acetylation-HeLa cells were transfected with Flag-PGC-1␣ expression vector either alone or with p300 expression vector. To determine the T 3 -mediated change in acetylation, TR␤ was cotransfected, and the cells were treated with T 3 for 16 h. 1 mg of protein lysate was immunoprecipitated with Flag-M2 beads (Sigma). The levels of acetylated PGC-1␣ were determined by Western blot analysis with anti-acetyl lysine antibody (9441, Cell Signaling) and total PGC-1␣ protein levels with PGC-1␣ antibody (sc-13067, Santa Cruz Biotechnology).

Resveratrol Enhanced the T 3 Induction of Gene Expression in
Rat Hepatocytes-We reported previously that T 3 induced the expression of the pdk4, cpt1a, and pepck genes (37, 38). Our initial experiments investigated whether the SIRT1 activator resveratrol would enhance T 3 induction of these genes. Rat hepatocytes were treated with resveratrol, T 3 , or both for 24 h. T 3 increased the mRNA abundance of pdk4 2.7-fold, cpt1a 3.6fold, and pepck 10-fold (Fig. 1, A-C). Resveratrol also stimulated the expression of these genes. When resveratrol was combined with T 3 , the expression of cpt1a and pdk4 was significantly enhanced. srebp-1c mRNA levels were decreased by both T 3 and resveratrol ϳ0.78and 0.6-fold, respectively. T 3 and resveratrol slightly increased fasn gene expression. Resveratrol did not alter the T 3 induction of pgc-1a (Fig. 1F). We asked whether T 3 would increase the abundance of SIRT1. The mRNA level of sirt1 was slightly elevated (1.8-fold) by T 3 , but the protein abundance of SIRT1 was not changed by T 3 in hepatocytes (Fig. 1H).
Resveratrol Increased the T 3 Induction in Human Hepatoma Cells-Next, we examined whether resveratrol had similar effects on the gene expression in HepG2 cells. In these cells, T 3 increased CPT1a and PDK4 gene expression by 2.5-and 2.2fold, respectively. Resveratrol alone induced these genes by 1.8and 3.8-fold and further enhanced the T 3 induction in HepG2 cells, suggesting that rat and human hepatocytes have a similar regulation of CPT1a and PDK4 gene expression (Fig. 2, A and  B). Curiously, we did not observe an effect of resveratrol on the T 3 induction of PEPCK. We observed no change in expression of FASN either with T 3 or with resveratrol. Our results suggest that resveratrol has an additive effect on the T 3 responsiveness of metabolic genes in HepG2 cells.

SIRT1 Enhances T 3 Responsiveness through Gene Promoters-
We wished to determine whether the effects of resveratrol on the T 3 regulation of the cpt1a and pdk4 genes transpired through the promoters of these genes. Previously, we had identified TREs in both genes (5,6). HepG2 cells were cotransfected with either Ϫ1256/ϩ78 pdk4 luciferase (pdk4-luc) or Ϫ4495/ ϩ1240 cpt1a luciferase (cpt1a-luc) along with an expression vector for TR␤ and then treated with T 3 , resveratrol, or both. For both pdk4 and cpt1a genes, T 3 stimulated the promoter and resveratrol enhanced the T 3 induction significantly (Fig. 3, A  and B). We observed that resveratrol also enhanced the PPAR␣ induction of cpt1a-luc (data not shown), suggesting that in addition to T 3 , resveratrol also affects the stimulation of genes by PPAR␣ agonists as has been reported previously (26). We tested whether cotransfection with SIRT1 would enhance the T 3 stimulation of cpt1a-luc. Although the TRE is located in the promoter, the first intron is needed for the T 3 induction of cpt1a. PGC-1␣ induces the cpt1a gene through elements in the first intron (7). SIRT1 enhanced the T 3 response of Ϫ4495/ ϩ1240 cpt1a-luciferase (Fig. 3C). Cotransfection of the deacetylase mutant SIRT1H363Y eliminated the enhancement of T 3 activity. SIRT1 modestly increased the T 3 induction of the Ϫ4495/ϩ19 cpt1a-luc without the first intron suggesting that SIRT1 may have several targets on the cpt1a gene. In addition, we transfected the Ϫ4495/ϩ1240 cpt1a-luc vector with a mutation in the TRE. This vector did not respond to T 3 or overexpression of SIRT1 (Fig. 3D). There was a slight increase in luciferase activity with T 3 treatment and overexpression of SIRT1, suggesting that SIRT1 acts not only through TR␤ but also on other factors.
Sirtuin 1 Interacts with TR␤-We next tested whether resveratrol could increase the T 3 induction through an isolated TRE. TREx2 SV40 luciferase containing two direct repeats of the AGGTCA motif separated by four nucleotides (DR4) ligated in front of the SV40 promoter was cotransfected in HepG2 cells along with a vector expressing TR␤. T 3 induced the TREx2 containing reporter, and resveratrol elevated the T 3 induction. These data suggested that resveratrol directly enhanced the transactivation capacity of TR␤ (Fig. 3E). To test whether overexpression of SIRT1 would enhance the T 3 response, we transfected the TREx2 luciferase vector with SIRT1. SIRT1 increased the T 3 induction of TREx2 luciferase, indicating that SIRT1 could directly enhance the transactivation through the TR␤ (Fig. 3F).
To determine whether SIRT1 interacts with TR␤, we conducted pulldown studies with purified His-tagged TR␤ protein and lysate from HeLa cells transfected with Flag-SIRT1. As shown in Fig. 4A, Flag-SIRT1 was precipitated by nickel resin when incubated with His-TR␤ as compared with nickel resin alone. Addition of T 3 did not change the amount of Flag-SIRT1 bound to TR␤ (data not shown). We also conducted the reciprocal experiment using purified His-tagged SIRT1 and lysates from HeLa cells expressing Flag-TR␤. These experiments yielded similar results (Fig. 4B). To test for interactions of purified SIRT1 and TR␤, we mixed SIRT1 with TR␤ and pulled down with either glutathione or nickel resin. GST-SIRT1 bound to glutathione-Sepharose beads pulled down His-TR␤ (Fig. 4C). Similarly, His-TR␤ bound to nickel resin pulled down SIRT1 (Fig. 4D). Next, we tested the ability of GST-SIRT1 to interact with the DNA binding domain or ligand binding domain of TR␤. Our data indicated that SIRT1 interacted with the ligand binding domain of TR␤ (Fig. 4E). To determine whether SIRT1 was associated with the cpt1a gene, we conducted ChIP assays. T 3 increased the association of SIRT1 with the TRE region of the cpt1a gene as well as the elements in the first intron that recruit PGC-1␣ (Fig. 4F). The upstream region  does not bind TR␤ or PGC-1␣ and was a negative control. These results indicate that SIRT1 is associated with the cpt1a promoter in a T 3 -dependent manner.

SIRT1 Enhances the T 3 Induction of Metabolic
Genes-Resveratrol has been reported to have several mechanisms of action besides the activation of SIRT1 (19). To determine whether SIRT1 was responsible for the enhanced T 3 induction of various metabolic genes, we examined the effect of SIRT1 expression in rat hepatocytes. Introduction of SIRT1 with adenoviruses increased SIRT1 mRNA and protein levels (Fig. 5,  F and G). SIRT1 elevated expression of the cpt1a, pdk4, and pepck genes (Fig. 5, A-C). fasn and srebp1c gene expression was FIGURE 3. SIRT1 enhances the T 3 stimulation through gene promoters. HepG2 cells were transiently transfected with 2 g of different luciferase constructs, 1 g of SV40-TR␤ and 0.1 g of TK-Renilla. T 3 and resveratrol were added at a concentration of 100 nM and 50 M, respectively, either separately or together for 24 h. Results are expressed as the relative induction by T 3 and resveratrol compared with the untreated cells: A, Ϫ4495/ϩ1240 cpt1a-luciferase; B, Ϫ1256/ϩ78 pdk4-luciferase; C, Ϫ4495/ϩ1240 and Ϫ4495/ϩ19 cpt1a-luc were cotransfected with SV40-TR␤, TK-Renilla, and with either Flag-SIRT1 or Flag-SIRT1H363Y. Cells were treated with T 3 as described above. D, Ϫ4495/ϩ1240 cpt1a-luc wild type or TRE mutant were cotransfected with Flag-SIRT1, SV40-TR␤ and TK-Renilla and treated with 100 nM T 3 for 24 h. E, HepG2 cells were transfected with TREX2 SV40-luciferase as above. The cells were treated with 100 nM T 3 or 50 M resveratrol. F, TREX2 SV40-luciferase vectors were transfected with Flag-SIRT1 and SV40-TR␤. Cells were treated with T 3 as described above. All transfections were performed in duplicate and repeated three to six times. Luciferase activity was corrected for both protein content and Renilla activity. The significance was determined by t test ( * , p ϭ 0.01 to 0.05; **, p ϭ 0.001 to 0.01). reduced to 0.13 and 0.23, respectively, by SIRT1 (Fig. 5, D and  E). T 3 induced pdk4 and cpt1a ϳ 2-fold and pepck ϳ 3-fold. SIRT1-expressing cells enhanced the T 3 induction of these genes significantly compared with control cells. fasn gene expression was unchanged, whereas SREBP1c was decreased to 0.73 by T 3 . These data suggest that SIRT1 affects the T 3 induction of cpt1a, pdk4, and pepck genes in rat hepatocytes.
To further demonstrate the requirement for deacetylation activity of SIRT1, we infected hepatocytes with AdSIRT1H355A and treated with T 3 . The changes in basal expression and T 3 induction of genes observed with AdSIRT1 virus infection in hepatocytes as above were not seen with the deacetylase dead mutant, which had expression pattern similar to control adenovirus (Fig. 5, A-C). These results are similar to the transfection data obtained in Fig. 3C. SIRT1 activates PGC-1␣ by deacetylation (19). We measured changes in the acetylation status of PGC-1␣ in response to T 3 . Cotransfection of HeLa cells with PGC1␣ and p300 increased PGC1␣ acetylation. The acetylation of PGC-1␣ was decreased by T 3 addition and by cotransfection with SIRT1 (Fig. 5H).
We introduced AdSIRT1 into HepG2 cells in the presence or absence of T 3 (Fig. 6). We observed that the effect of SIRT1 expression was highly similar to the addition of resveratrol (Fig.  2), suggesting that resveratrol acts by inducing SIRT1 activity.
Knockdown of SIRT1 Alters the T 3 Induction of Metabolic Genes-To demonstrate the requirement for SIRT1 in the T 3 induction of hepatic metabolic gene expression, we infected rat hepatocytes with adenovirus encoding shRNA to silence SIRT1 (AdshSIRT1) prior to treatment of cells with T 3 . Adenovirus encoding scrambled shRNA was used as a control (AdScr). sirt1 gene expression was reduced 60% in hepatocytes transduced with adenovirus expressing shSIRT1 (Fig. 7F). T 3 treatment induced the gene expression of cpt1a, pdk4, and pepck as before in cells transduced with AdScr. Knockdown of SIRT1 in hepatocytes reduced the T 3 induction of these genes, indicating that SIRT1 participates in the regulation of cpt1a, pdk4, and pepck gene expression by T 3 (Fig. 7, A-C). Interestingly, the knockdown of SIRT1 also reduced the inhibition of srebp-1c by T 3 (Fig. 7E). In Fig. 7G, we show that the SIRT1 protein is decreased by AdshSIRT1. . Sirtuin 1 and TR␤ physically interact. A, His-tagged TR␤ was prepared in BL21 E. coli. HeLa cells were transfected with a vector that expresses Flag-SIRT1, and cell lysates were prepared. His-TR␤ and cell lysates were allowed to interact, and the interacting proteins were precipitated with nickel (Ni) resin. The interacting proteins were resolved by Western analysis with antibodies for the Flag epitope, TR␤, and SIRT1. B, His-tagged SIRT1 was prepared in BL21 E. coli. HeLa cells were transfected with a Flag-TR␤ expression vector, and the protein interactions were tested as described above. C, His-TR␤ was allowed to interact with either purified GST-SIRT1 or GST protein, and the interacting proteins were pulled down with glutathione-Sepharose 4B beads. Proteins were detected with TR␤ and GST antibodies. D, purified GST-SIRT1 protein was mixed with His-TR␤; E, His-tagged DNA binding (DBD) and ligand binding domains (LBD) of TR␤. Proteins were pulled down with nickel (Ni) resin and assayed with SIRT1 and His tag antibodies. F, rat hepatocytes were treated with T 3 for 24 h and cross-linked. The chromatin was sheared and immunoprecipitated with IgG or SIRT1 antibodies. The immunoprecipitated DNA was amplified with primers for cpt1a TRE, first intron, and upstream region.  We also tested whether the T 3 requirement for SIRT1 was seen in vivo. Rats were made hypothyroid by the addition of propylthiouracil in the diet. Hypothyroid rats were transduced via the tail vein with either AdScr or AdshSIRT1. Three days later T 3 was administered for two consecutive days. As with the hepatocytes, we achieved a significant knockdown of both the SIRT1 mRNA and protein (Fig. 8, F-H). The T 3 induction of cpt1a, pdk4, and pepck was diminished by SIRT1 knockdown (Fig. 8, A-C). Also, the inhibition of srebp-1c was reduced (Fig.  8E). The induction of CPT1a protein was decreased following SIRT1 knockdown (Fig. 8H). These data indicate that the effect of SIRT1 is evident in vivo as well as in vitro.

DISCUSSION
In these studies, we explored the role of SIRT1 in the regulation of hepatic gene expression by T 3 . We found that resveratrol enhanced the T 3 induction of genes related to fatty acid oxidation and gluconeogenesis in the liver. Similar to resveratrol, overexpression of SIRT1 increased the stimulation of the cpt1a and pdk4 genes by T 3 . SIRT1 is recruited to the cpt1a gene in response to T 3 addition. Finally, knockdown of SIRT1 reduced the T 3 -mediated regulation of hepatic gene expression in isolated hepatocytes and in vivo. Overall, our data suggest that T 3 and SIRT1 activators can induce the activation of genes associated with elevated mitochondrial metabolism.
Several lines of evidence led us to consider the possibility that SIRT1 would enhance the T 3 regulation of hepatic gene expression. Many T 3 -responsive genes such as cpt1a and pdk4 are also induced by SIRT1 (39,40). SIRT1 regulates the expression of genes encoding mitochondrial proteins involved in hepatic lipid metabolism (26). As these genes are often T 3 targets, we speculated that SIRT1 and T 3 might co-regulate gene expression. We previously reported that PGC-1␣ participates in the T 3 induction of several hepatic genes (6,7). PGC-1␣ is activated by SIRT1 via the deacetylation of multiple lysines suggesting that SIRT1 might enhance the T 3 induction via activation of PGC-1␣ (39).
There have been a very limited number of studies linking T 3 with SIRT1. It was found that addition of T 3 to human peripheral blood mononuclear cells increased the abundance of SIRT1 after 24 h (41). We tested whether T 3 would increase SIRT1 abundance in the liver since other nuclear factors such as PGC-1␣ and C/EBP␤ are induced by T 3 (7,42). We did not observe any elevation of SIRT1 abundance by T 3 in rat hepatocytes ( Fig. 1) or in hypothyroid rats treated with T 3 (Fig. 8). Supporting our data were the observations that a T 3 analog, TRC150094, did not increase SIRT1 protein abundance, although there was elevated SIRT1 activity (43).
Our results provide a direct link between SIRT1 and TR␤. However, SIRT1 could impact T 3 actions in multiple ways. Using both transient transfections and adenoviral expression of deacetylase dead SIRT1, we observed that the deacetylase activity of SIRT1 is required. In addition, we found that PGC-1␣ is deacetylated in response to T 3 . SIRT1 enhanced the T 3 induc-FIGURE 8. SIRT1 knockdown reduces the T 3 induction of gene expression in rat liver. Hypothyroid rats were infected with 10 11 pfu/kg AdshSIRT1 and AdScr adenovirus via the tail vein. T 3 (0.33 mg/kg) was administered for two consecutive days before harvesting the liver. The mRNA abundance in liver was measured by real time PCR: A, cpt1a; B, pdk4; C, pepck; D, fasn; E, srebp-1c mRNA; and F, fold decrease of sirt1 mRNA by AdshSIRT1 adenovirus compared with untreated rats infected with AdScr adenovirus. G, Western blot is shown for SIRT1 and CPT1a protein abundance in AdshSIRT1-and AdScr-infected rat liver with and without T 3 treatment. H, the protein abundance of SIRT1; and I, CPT1a was measured by densitometry of the Western blots. The data are expressed as the mean of the fold induction by T 3 Ϯ S.E. of mRNA abundance relative to untreated rat liver. The induction by T 3 treatment compared with its respective control or the effect of SIRT1 knockdown is indicated as fold induction, and the inhibition is indicated in percent inhibition. (*, p ϭ 0.01 to 0.05; **, p ϭ 0.001 to 0.01; ***, p Ͻ 0.001).
tion of cpt1a-luciferase in transfections. Elements in the first intron of the cpt1a gene are required for the full T 3 and PGC-1␣ induction, even though the TRE is located in promoter (5). Removal of the first intron reduced the stimulation by SIRT1, indicating that proteins in addition to TR␤ might be targets for SIRT1. Furthermore, we showed by ChIP assay that SIRT1 is associated with the first intron of the cpt1a gene as well as the TRE in response to T 3 administration. Based on our data, we can speculate that at least for few genes, T 3 -dependent recruitment of SIRT1 to the promoter possibly leading to activation of PGC-1␣ could be one of the mechanisms by which SIRT1 mediates T 3 regulation. It was reported that SIRT1 regulates another nuclear receptor, PPAR␣ primarily through activation of PGC-1␣ (26).
Nuclear receptors are regulated by SIRT1 via varied mechanisms. For example, SIRT1 regulates the transcriptional activity of liver X receptor ␣ by marking it for ubiquitination and degradation (22). It was speculated that the clearing of liver X receptor ␣ from regulated genes allowed the initiation of the next round of liver X receptor ␣-mediated transcription (22). SIRT1 directly interacts with and activates ERR␣ (24). Inhibition of SIRT1 decreased the activity of estrogen receptor signaling (44). On the other hand, SIRT1 deacetylates and represses androgen receptor signaling (45). Finally, SIRT1 inhibits the glucocorticoid induction of the UCP3 gene both by reducing histone acetylation and by disrupting interactions of the glucocortocoid receptor and p300 (46). Our studies indicate that TR␤ is another nuclear receptor whose signaling is modulated bySIRT1.
Our observation, that SIRT1 directly interacts with TR␤ and increases TR␤ transcriptional activity, suggests that TR␤ may also be a direct SIRT1 target. It might either directly deacetylate TR␤, or alternatively, it is possible that SIRT1 acts as a bridging molecule to recruit other coactivators. Previous studies reported that TR␤ was acetylated and that this acetylation was associated with increased nuclear localization (47). A subsequent report found that TR␣ was acetylated by p300 and that this acetylation enhanced the activity of TR␣ (48). It has been proposed that p300 and SIRT1 are in a complex of coregulators associated with genes, suggesting that both factors could be present on T 3 -regulated genes (13). The fact that SIRT1 is a deacetylase and enhances thyroid hormone actions is not easily reconciled to the increased activity of thyroid hormone receptor by acetylation, suggesting that the role of SIRT1 in thyroid hormone regulation is quite complex. SIRT1 has numerous cellular targets, and the genes that we analyzed in this study are regulated by various SIRT1 target transcription factors. pdk4 is induced by FoxO1 and ERR␣ (32,49). The cpt1a gene is stimulated by PPAR␣, whereas the pepck is regulated by both hepatic nuclear factor 4 and FoxO1 (50,51). Thus, in addition to TR␤ and PGC-1␣, SIRT1 may target several factors to enhance the T 3 induction in a gene-specific manner.
Our initial studies used resveratrol as an activator of SIRT1, but recent studies found that resveratrol is an indirect activator of SIRT1 (27). Resveratrol activates AMPK and some of the actions of resveratrol on the T 3 induction could be mediated by AMPK (52). In our experiments, the expression of SIRT1 mimicked the effect of resveratrol and knockdown of SIRT1 decreased the T 3 response. These data suggest that resveratrol acts through the promoters and SIRT1 to induce these genes. With respect to SIRT1, we do not know whether the enhancement of the T 3 effect involves only a few genes or is a genome wide phenomenon. The T 3 induction of several genes such as fasn and pgc-1␣ was not altered by SIRT1. In addition, it is not known if this action of SIRT1 is confined to the liver or involves other tissues. We have observed that cpt1a and pgc-1␣ are increased by T 3 in the liver but not the heart (53). Therefore, SIRT1 might only enhance the T 3 actions in the liver. Further studies will be required to address these issues.
Finally, selective activators of TR␤ decrease serum cholesterol and lipid levels (54). These TR␤ agonists induce lipid metabolism and cholesterol clearance by the liver but do not have the detrimental effects of T 3 on the cardiovascular system. Resveratrol also has positive hepatic actions including the enhancement of mitochondrial metabolism and the reduction of hepatic steatosis (55). Therefore, it is interesting to speculate that selective TR␤ agonists and activators of SIRT1 might have additive beneficial actions in the liver with respect to the treatment of hepatic steatosis and insulin resistance.