SIRT1 Regulates Adiponectin Gene Expression through Foxo1-C/Enhancer-binding Protein α Transcriptional Complex*

Adiponectin is an adipose-derived hormone that plays an important role in maintaining energy homeostasis. Adiponectin gene expression is diminished in both obesity and type 2 diabetes. However, the mechanism underlying the impaired adiponectin gene expression remains poorly understood. Recent studies have indicated that forkhead transcription factor O1 (Foxo1) and silent information regulator 2 mammalian ortholog SIRT1 are involved in adipogenesis. Here we have shown that Foxo1 up-regulates adiponectin gene transcription through a Foxo1-responsive region in the mouse adiponectin promoter that contains two adjacent Foxo1 binding sites. Foxo1 interacts with CCAAT/enhancer-binding protein α (C/EBPα) to form a transcription complex at the mouse adiponectin promoter and up-regulates adiponectin gene transcription. Our study has revealed that C/EBPα accesses the adiponectin promoter through two Foxo1 binding sites and acts as a co-activator. Further, SIRT1 increases adiponectin transcription in adipocytes by activating Foxo1 and enhancing Foxo1 and C/EBPα interaction. Importantly, both Foxo1 and SIRT1 protein levels were significantly lower in epididymal fat tissues from db/db and high fat diet-induced obese mice compared with normal mice. We propose that low expression of SIRT1 and Foxo1 leads to impaired Foxo1-C/EBPα complex formation, which contributes to the diminished adiponectin expression in obesity and type 2 diabetes.

Adiponectin is an adipocyte-derived hormone that plays an important role in energy metabolism, immunological responses, and development of cardiovascular disease (1)(2)(3). Compelling evidence demonstrates that adiponectin enhances insulin sensitivity, improves fatty acid oxidation in skeletal muscle, and suppresses hepatic gluconeogenesis (4 -7). Although adiponectin is predominantly produced by adipose tissues, plasma adiponectin concentration and adiponectin gene expression are inversely correlated with adiposity (2).
However, information is limited regarding the underlying mechanisms that impair adiponectin gene expression in obesity and type 2 diabetes. Foxo1 2 is a member of the forkhead transcription factor class O family and is involved in adipocyte differentiation (8). Foxo1 gene haploinsufficiency leads to significant reduction of adiponectin gene expression in the adipose tissue (8). SIRT1 is an NAD ϩ -dependent protein deacetylase that is also involved in adipogenesis (9). Calorie restriction induces Foxo1 and SIRT1 expression, which mediate the resultant longevity effect in cells from yeast to mammals (10,11). The expression of Foxo1 and SIRT1 is also up-regulated during adipocyte differentiation (8,9). SIRT1 regulates Foxo1 transactivation activity by deacetylating three lysine residues within the forkhead DNA binding domain (12). These studies led us to hypothesize that Foxo1 and SIRT1 may regulate adiponectin gene expression.
Here, we report that overexpression of Foxo1 increased adiponectin gene expression in differentiated 3T3-L1 adipocytes. Our study has identified two Foxo1 responding elements at the mouse adiponectin promoter and demonstrated that Foxo1 interacts with C/EBP␣ and forms a transcriptional complex at the adiponectin promoter. Furthermore, overexpression of SIRT1 or knocking down endogenous SIRT1 increased or decreased adiponectin gene expression in 3T3-L1 adipocytes, respectively. Expression of SIRT1 synergistically increased Foxo1-and C/EBP␣-mediated adiponectin promoter activation. In addition, SIRT1 enhanced Foxo1-C/EBP␣ transcription complex formation. Therefore, we conclude that SIRT1 up-regulates adiponectin gene transcription via a Foxo1-C/ EBP␣ transcription complex. The study also suggests that decreased expression of SIRT1 and Foxo1 in obesity may play a causal role in the diminished adiponectin gene expression.

EXPERIMENTAL PROCEDURES
Experimental Animals-Male C57BL/6J and C57BL/ 6J-ϩ/ϩ Lepr db/db (db/db) mice were housed in a pathogen-free animal facility. Obesity of C57BL/6J mice was induced by 17 weeks of high fat diet feeding (60 kcal% as fat, Research Diets, New Brunswick, NJ). Control mice were fed a diet containing 10% fat.
Total Cellular Protein and Nuclear Protein Extraction and Western Blot Analysis-Total cell and nuclear proteins were extracted as described previously (14). Relative protein levels of SIRT1, Foxo1, C/EBP␣, and adiponectin were measured by Western blot using specific antibody (Santa Cruz Biotechnology and R & D Systems).
Luciferase Assay-Reporter constructs and expression plasmids were introduced into cells by electroporation or chemical reagent FuGENE 6 (Roche Applied Science). Differentiated 3T3-L1 adipocytes were transfected by the electroporation method (14). The medium was changed to a medium without fetal bovine serum 12 h after transfection.
GST Pulldown Assay-The plasmid encoding full-length GST-Foxo1 fusion protein was kindly provided by Dr. O'Brien (Vanderbilt University Medical School, Nashville, TN). A series of truncated DNA fragments of Foxo1 were subcloned into pGEX-4T-1 (Amersham Biosciences). The expression of GST fusion proteins was induced by 0.1 mM isopropyl 1-thio-␤-Dgalactopyranoside in BL21(DE3) at 25°C. BL21(DE3) sonicates were used as bait protein. Nuclear extracts from HEK293 cells transduced with C/EBP␣-encoding adenovirus vector were used as prey protein. Glutathione-Sepharose 4B beads (Amersham Biosciences) were preincubated with 200 g of HEK293 cell lysates for 1 h at 4°C to block nonspecific binding. The preblocked beads were washed three times with phosphatebuffered saline (pH 7.4). The GST fusion proteins were immobilized on the preblocked Sepharose beads at room temperature for 30 min. The beads were then incubated with 100 g of nuclear extract prey protein at 4°C for 2 h. After three times of intensive washing, the GST-fused bait and bound proteins were eluted by 5 min of boiling in 4ϫ loading buffer (250 mM Tris, 8% SDS, 40% glycerol, 20% mercaptoethanol, 2% bromphenol blue). The eluted proteins were separated by 12% SDS-PAGE followed by Western blotting using anti-C/EBP␣ antibody (Santa Cruz Biotechnology).
Data Analysis-Data are expressed as mean Ϯ S.D. Statistical analysis was performed using the Student's t test or analysis of variance followed by contrast testing with Tukey or Dunnett error protection. Differences were considered significant at p Ͻ 0.05.

Foxo1
Up-regulates Adiponectin Gene Expression in Differentiated Adipocytes-Foxo1 expression is induced in the early stages of adipocyte differentiation, but its activation is delayed until the end of the clonal expansion (8), which suggests that Foxo1 regulates adipogenesis in a stage-dependent manner. The main goal of this study was to investigate the underlying molecular mechanisms of adiponectin gene expression, which is turned on in the middle of adipocyte differentiation and fully expressed in mature adipocytes (19,20). We, therefore, focused our studies on the effect of Foxo1 on adiponectin gene expression in mature adipocytes. Eight days after differentiation, the 3T3-L1CAR⌬1 adipocytes were transduced with adenovirus vectors encoding wild type (Foxo1-WT), constitutively nuclear (Foxo1-ADA), or dominant negative Foxo1(Foxo1-⌬256) (21). The results clearly showed that adiponectin protein and RNA levels were significantly increased in adipocytes transduced by adenovirus vector encoding Foxo1-WT or Foxo1-ADA (p Ͻ 0.05) (Fig. 1, A and B). However, the magnitude of Foxo1-ADA-induced adiponectin expression was not significantly greater than that induced by Foxo1-WT overexpression. This finding suggests that Foxo1-WT and Foxo1-ADA are equally potent in up-regulating adiponectin gene expression in this cellular model. Surprisingly, adiponectin protein and RNA levels were also significantly elevated in adipocytes transduced with Foxo1-  DECEMBER 29, 2006 • VOLUME 281 • NUMBER 52 ⌬256-encoding adenovirus vectors (Fig. 1, A and B). Foxo1-⌬256 is a truncated protein containing the DNA binding domain but lacking the transactivation domain (8)(supplemental Fig. S1) Theoretically, Foxo1-⌬256 should compete with endogenous Foxo1 for DNA binding and impair Foxo1-mediated gene transcription. Indeed, Foxo1-⌬256 inhibits the transcription of several Foxo1 target genes (8,22,23). Our results seem contradictory to the design of the constitutively negative mutation. The following studies demonstrate that the transactivation activity of Foxo1 explains only part of the role of Foxo1 in adiponectin gene transcription. In addition, Foxo1 provides a docking site for C/EBP␣ at the adiponectin promoter. Nevertheless, these results demonstrate that Foxo1 up-regulates adiponectin gene expression in mature adipocytes.

Regulation of Adiponectin Gene Transcription
Foxo1 Activates the Adiponectin Promoter-To study whether and how Foxo1 up-regulates adiponectin gene transcription, a mouse adiponectin promoter-directed luciferase gene reporter construct was co-transfected with Foxo1-encoding plasmids in differentiated 3T3-L1 adipocytes. As expected, expression of Foxo1-WT and Foxo1-ADA dramatically increased adiponectin promoter-directed luciferase expression in 3T3-L1 adipocytes (Fig. 1C). Again, expression of Foxo1-⌬256 increased adiponectin promoter activity significantly (p Ͻ 0.05) (Fig. 1C). However, the magnitude of Foxo1-⌬256stimulated promoter activation is much less than the induction by Foxo1-WT or Foxo1-ADA (Fig. 1C). These results are consistent with the data from adenovirus vector-transduced adipocytes studies (Fig. 1, A and B). Although these studies did not provide any clue for dominant negative Foxo1-⌬256-enhanced adiponectin gene expression, the results indicate that Foxo1 up-regulates adiponectin expression at the transcription level.
To solve the puzzle of Foxo1-⌬256-mediated adiponectin gene expression, we repeated the transient transfection study in HEK293 cells, which have been used as a simplified cellular model for studying transcriptional regulation of adipocyte gene expression (14, 24 -26). Similar to the results in 3T3-L1 adipocytes, adiponectin promoter activities were elevated nearly 5-fold by wild type or Foxo1-ADA (p Ͻ 0.001) (Fig. 1D). These results further support our conclusion that Foxo1 up-regulates adiponectin gene transcription. In contrast to the result from adipocytes, ectopic expression of Foxo1-⌬256 reduced adiponectin promoter activity in HEK293 cells (p Ͻ 0.05) (Fig.  1D), raising the possibility that there is (are) adipocyte-specific protein(s) in adipocytes that prevail over the inhibitory effect of Foxo1-⌬256 and enhance adiponectin gene transcription.
C/EBP␣ and PPAR␥ are two master adipogenic transcription factors that regulate most adipose protein expression (14,27). We used transient transfection to determine the relationship of Foxo1, C/EBP␣, and PPAR␥ in the regulation of adiponectin gene transcription. Consistent with earlier studies (14,28,29), ectopic expression of C/EBP␣ or PPAR␥2 and its partner RXR␣ modestly increased mouse adiponectin promoter-directed luciferase activity (Fig. 1E). The results also showed that expression of C/EBP␣ and PPAR␥2/RXR␣ increased adiponectin promoter activation additively (Fig. 1E, last lane). Co-expression of Foxo1 and C/EBP␣ robustly increased the promoter activity nearly 20-fold (p Ͻ 0.001) (Fig. 1E), indicating that Foxo1 and C/EBP␣ synergistically increase adiponectin promoter activity.
Co-expression of Foxo1 did not show any synergistic effect on PPAR␥2-induced adiponectin promoter activation (Fig. 1E). Of note, it has been reported that C/EBP␣ up-regulates adiponectin expression through a C/EBP consensus sequence at the human adiponectin promoter (29). Although a putative C/EBP␣ binding site in the murine adiponectin promoter can be found by sequence scan, most studies did not demonstrate binding of C/EBP␣ to this site (14,26,30). Our study indicated that C/EBP␣ and Foxo1 up-regulate adiponectin promoter activity in a synergistic manner. Therefore, we hypothesized that Foxo1 interacts C/EBP␣ and forms a transcription complex at the adiponectin promoter.
To further test the physical interaction of Foxo1 and C/EBP␣, GST pulldown was employed. Several recombinant GST-Foxo1 constructs with truncated Foxo1 (Fig. 2, C and D) were also used to define the binding site of Foxo1 with C/EBP␣. As shown in Fig. 2D, bottom panel and last lane, there is physical interaction of Foxo1 and C/EBP␣. Moreover, the study indicated that the forkhead domain of Foxo1 is the binding site for C/EBP␣ (Fig. 2D). A positive GST pulldown result has been observed using GST-C/EBP␣ as bait (supplemental Fig. S2). These results demonstrated that Foxo1 interacts with C/EBP␣ through the forkhead domain.
Identification of Foxo1-responding Element in the Adiponectin Promoter-The mouse adiponectin promoter was examined to define the Foxo1-responding element. A Foxo1-responding region was located within Ϫ760 to Ϫ560 bp of the adiponectin promoter by 5Ј end deletion (supplemental Fig. S3). Scanning this region revealed three putative Foxo1 binding sites (core motif T(G/A)TTT) (Fig. 3A), which are usually named insulin response sequences (IRSs) (18,31). The results of the transient transfection study showed that mutation of IRS1 or IRS3 significantly decreased the basal adiponectin promoter activity (Fig.  3B). In contrast, mutation of IRS2 increased the basal promoter activity significantly (Fig. 3B). Mutation of either IRS1 or IRS3 almost completely abolished Foxo1-induced promoter activation (Fig. 3C).
To determine the binding of Foxo1 to these IRS sites, EMSA and supershift assays were conducted using nuclear proteins from adenovirus vector-transduced HEK293 cells. The results indicated that only IRS1 and IRS3 bind Foxo1 (Fig. 3D). The supershift assay revealed that Foxo1 was present in the protein complex bound to IRS1 and IRS3 probes (Fig. 3E). Taken together, these data demonstrate that IRS1 and IRS3 are Foxo1-responding elements. Although the mechanism by which IRS2 contributes to basal promoter activity remains to be determined, it appears that the three IRSs regulate adiponectin gene transcription in a functionally distinct manner. IRS1 and IRS3 are co-operative in mediating Foxo1-induced adiponectin promoter activation. A similar observation has been reported in the glucose-6-phosphatase promoter (18,32).
Interestingly, IRS1 or IRS3 mutation also dampened C/EBP␣-stimulated adiponectin promoter activation (Fig. 3C). However, there are no C/EBP binding consensus sequences in this region (Fig. 3A). The above studies have demonstrated that C/EBP␣ interacts with Foxo1 and Foxo1 binding to IRS1 and IRS3 sites (Figs. 2 and 3). In addition, C/EBP␣ protein was detected in the IRS1 or IRS3 protein complexes (Fig. 3E). Therefore, these results indicate that Foxo1 may provide a docking site for C/EBP␣ at the adiponectin promoter, where C/EBP␣ acts as a co-activator. However, our study does not rule out the possibility that C/EBP␣ regulates adiponectin gene transcription through its own responding elements at the proximal promoter.
SIRT1 is a nuclear protein that regulates target gene expression via NAD ϩ -dependent de-acetylation of histones or transcription factors. We used transient transfection and the HEK293 cellular model to study the effect of SIRT1 on adiponectin promoter activation. The results showed that overexpression of SIRT1 increased adiponectin promoter-directed luciferase expression (ϳ2-fold; p Ͻ 0.05) (Fig. 4C). Co-expression of SIRT1 and Foxo1 robustly increased adiponectin promoter activity (p Ͻ 0.001) (Fig. 4C). SIRT1 and Foxo1 appear to synergistically increase adiponectin promoter-directed luciferase expression.
C/EBP␣ and PPAR␥ are important for adiponectin gene transcription. As shown in Fig. 4C, co-expression of SIRT1 increased C/EBP␣-mediated adiponectin promoter activation in an additive manner (lanes 8th from the left). Overexpression of SIRT1 synergistically increased the adiponectin promoter activities in Foxo1 and C/EBP␣ co-transfected cells (Fig. 4C, 10th lane from the left). However, expression of SIRT1 and PPAR␥2/RXR␣ only increased the adiponectin promoter-directed luciferase expression additively (data not shown). Together, these data suggest that SIRT1 synergistically up-regulates Foxo1-and C/EBP␣-mediated adiponectin gene transcription. A study has reported that SIRT1 inhibits PPAR␥ activity in adipocytes (9). Apparently, our result does not agree with this report in the context of PPAR␥-induced adiponectin promoter activation. We do not have any explanation for this disparity. A most recent study reported that SIRT1 activator resveratrol not only increases SIRT1 expression but also increases PPAR␥ activities in Caco-2 cancer cells (33).
To determine whether the deacetylase activity of SIRT1 is required for the transcriptional enhancement of adiponectin gene expression, a deacetylase inactive mutant of SIRT1H363Y was used (Fig. 4C), which contains a base substitution that converts the invariant catalytically active histidine at amino acid residue 363 to a tyrosine (17). Expression of the inactive SIRT1H363Y did not further increase Foxo1-and C/EBP␣-mediated adiponectin promoter activation (Fig. 4C, 6th and 11th   FIGURE 3. Foxo1-responding elements in the adiponectin promoter. A, DNA sequences of mouse adiponectin promoter that respond to Foxo1. The putative Foxo1 binding sites are underlined and indicated as insulin response sequence (IRS). Plasmid containing the adiponectin promoter (Ϫ760 to ϩ1 nucleotide(s)) directing luciferase expression was transfected into HEK293 cells (B and C). The plasmids encoding wild type Foxo1 or C/EBP␣ were cotransfected (C). Luciferase activities were measured 24 h after transfection and normalized to ␤-galactosidase activity. The activities are presented as relative activity or fold activation. Each study was repeated in four independent experiments. *, p Ͻ 0.05 versus parental construct (B) or pcDNA transfected cells (C). **, p Ͻ 0.05 versus pcDNA transfected parental construct (C). D, association of Foxo1 and C/EBP␣ with Foxo1-responding elements at the adiponectin promoter. Protein binding to DNA was measured by EMSA. Positive control (PC) is the double-stranded oligonucleotide corresponding to the IRS site in the mouse G-6-Pase promoter. Competition assays were performed using 100-fold excess of unlabeled oligonucleotides (Comp ϫ100). For the supershifting assay, antibody or control IgG was added in the reaction (E).
lanes from the left). The results indicated that deacetylase activity is required for SIRT1-enhanced adiponectin gene transcription.
SIRT1 Enhances Foxo1 and C/EBP␣ Interaction with the Adiponectin Promoter-Three lysine residues (Lys-242, -245, and -262) in the Foxo1 forkhead DNA binding domain can be deacetylated by SIRT1 (34,35). Acetylation of these lysine residues in Foxo1 decreases its DNA-binding and enhances protein kinase B-mediated phosphorylation, which excludes Foxo1 from the nucleus (34). Our results support the idea that SIRT1mediated deacetylation of Foxo1 increases its transactivation activity and increases adiponectin promoter activity (supplemental Fig. S5).
To investigate the mechanism by which SIRT1 enhances Foxo1-and C/EBP␣-mediated adiponectin gene transcription, we used chromatin immunoprecipitation assays to determine whether SIRT1 alters Foxo1 and C/EBP␣ binding to the adiponectin promoter. As shown in Fig. 5, both Foxo1 and C/EBP␣ were detected at the adiponectin promoter. Overexpression of Foxo1 or C/EBP␣ increased not only binding of the individual proteins to the promoter but also significantly increased the binding of the partner proteins C/EBP␣ and Foxo1, respectively. Most importantly, SIRT1 knockdown robustly reduced the binding of C/EBP␣ or Foxo1 to the adiponectin promoter (Fig. 5). However, consistent with our supershift assay (Fig. 3E), we did not detect SIRT1 protein in the complex bound to the adiponectin promoter (data not shown). These results support our hypothesis that Foxo1 and C/EBP␣ interact and form a complex at the adiponectin promoter, and SIRT1 enhances this process.
We next studied whether SIRT1 alters the Foxo1 and C/EBP␣ interaction using co-immunoprecipitation of protein samples from transfected HEK293 cells (Fig. 6A). Overexpression of SIRT1 increased the association of Foxo1 and C/EBP␣ (Fig. 6A). However, overexpression of SIRT1H363Y did not increase Foxo1 and C/EBP␣ association, suggesting SIRT1 deacetylase activity is required for SIRT1-enhanced Foxo1 and C/EBP␣ interaction. SIRT1 deacetylates three lysine residues in the forkhead DNA binding domain of Foxo1 (34,35), which is the region that interacts with C/EBP␣, as we described above. To test the effect of the acetylation status of these three lysine residues on the interaction between Foxo1 and C/EBP␣, Foxo1 with acetylation or deacetylation mimic mutations (16) were used for coimmunoprecipitation. The Lys3 Arg substitution (K3R) prevents acetylation but keeps positive charges, thus mimicking the deacetylated form. The Lys3 Ala (K3A) and Gln (K3Q) substitutions mimic the constitutively acetylated form through the absence of positive charges (16). Greater amounts of C/EBP␣ were co-immunoprecipitated with the Foxo1 deacetylation mimic mutant protein than with either wild type or Foxo1 acetylation mimic-mutant proteins (Fig. 6B), which suggest that deacetylation of Foxo1 enhances Foxo1 and C/EBP␣ interaction. . SIRT1 up-regulates adiponectin expression. A, differentiated 3T3-L1 adipocytes were transfected by electroporation with SIRT1-encoding plasmid. 24 h after transfection, cells were lysed and SIRT1 and adiponectin protein were measured by Western blot. B, 3T3-L1CAR⌬1 adipocytes were transduced with adenovirus vectors encoding SIRT1 siRNA or control siRNA. 48 h after transduction, protein levels of SIRT1 and adiponectin were measured. C, SIRT1, Foxo1, and C/EBP␣ synergistically increase adiponectin promoter activity. The adiponectin promoter-directed luciferase gene reporter construct was co-transfected with plasmids encoding SIRT1, SIRT1H363Y, Foxo1, and C/EBP␣ in HEK293 cells. 24 h after transfection, luciferase activity was measured. Luciferase activities were normalized to ␤-galactosidase activity. The values are expressed as relative luciferase activities. The error bar indicates S.E. for four independent experiments. *, p Ͻ 0.05; **, p Ͻ 0.001 versus pcDNA-transfected cells. The above studies have demonstrated that SIRT1 up-regulates adiponectin gene transcription by increasing Foxo1 transactivation activity. To determine whether SIRT1 regulates adiponectin transcription through the Foxo1-C/EBP␣ complex, truncated Foxo1-⌬256 and C/EBP␣ were co-transfected with adiponectin promoter gene reporter construct into HEK293 cells (Fig. 6C). Expression of SIRT1 did not alter the luciferase activities in Foxo1-⌬256-expressed cells. However, SIRT1 increased the luciferase activity significantly in Foxo1-⌬256and C/EBP␣ co-transfected cells. Foxo1-⌬256 does not have transactivation activity but can bind with C/EBP␣ (Fig. 2, A and  D). Taken together with the results presented above, our study indicates that SIRT1 up-regulates adiponectin gene transcription not only by increasing Foxo1 transactivation activity but also by enhancing Foxo1-C/EBP␣ complex formation at the adiponectin promoter.
Low Level of SIRT1 and Foxo1 Expression in the Adipose Tissues from Obese and db/db Diabetic Mice-The studies described above were performed with cultured cells. Thereafter, we studied the physiological significance of these discoveries. Although Foxo1 gene haploinsufficiency decreased adiponectin expression in mice (8), it is not known whether Foxo1 and SIRT1 contribute to the diminished adiponectin gene expression in obesity and type 2 diabetes. We examined Foxo1 and SIRT1 protein levels in the epididymal fat tissues of high fat diet-induced obese and db/db diabetic mice. Mice were made obese by 17 weeks of high fat diet (60 kcal% as fat). The mice exhibited significantly reduced insulin sensitivity (data not shown). db/db mice have a spontaneous mutation in the leptin receptor gene. They are obese and widely used as a type 2 diabetes mouse model. Consistent with previous studies, serum adiponectin concentrations and adiponectin RNA levels in adipose tissues were severely low in both high fat diet-induced obese and db/db mice (supplemental Fig. S6, A and B). Parallel with decreased adiponectin expression, Foxo1 and SIRT1 protein levels were also significantly decreased in epididymal fat from high fat diet-induced obese and db/db diabetic mice (supplemental Fig. S6C).

DISCUSSION
Adiponectin is an adipose-derived hormone with a variety of beneficial biological functions, such as sensitizing insulin action and protecting against atherogenesis. However, the underlying mechanisms of the diminished adiponectin gene expression in obesity and type 2 diabetes are poorly understood. Here we show that Foxo1 directly up-regulates adiponectin gene transcription but also interacts and recruits C/EBP␣ to the promoter. SIRT1 enhances the interaction of these two transcription factors and coordinately increases Foxo1-C/EBP␣-mediated adiponectin promoter activation. Our data demonstrate that the Foxo1-C/EBP␣ transcription complex is critical in controlling adiponectin gene expression. Our study also suggests that decreased Foxo1 and SIRT1 expression and impaired Foxo1-C/EBP␣ transcription complex formation may contribute to the diminished adiponectin gene expression in obesity and type 2 diabetes.
Foxo1 regulates glucose and fat metabolism through its various target genes in the liver, muscle, adipose tissue, and pancreas. Current study is the first to identify its direct regulatory effect on adiponectin gene transcription. Nakae et al. (8) have reported that, with the initiation of adipocyte differentiation, Foxo1 expression is increased up to 6-fold over basal levels, reaching peak level at days 2-4. Overexpression of constitutive nuclear Foxo1 before clonal expansion prevents adipocyte dif-FIGURE 6. SIRT1 enhances adiponectin gene expression by increasing Foxo1 and C/EBP␣ interaction. A, HEK293 cells were co-transfected with FLAG-Foxo1-and C/EBP␣-encoding plasmids. SIRT1 or SIRT1H363Y was overexpressed in the indicated group. B, FLAG-fused wild type or mutant Foxo1 proteins and C/EBP␣ were expressed in HEK293 cells. Proteins were precipitated using FLAG antibody and probed with Foxo1 or C/EBP␣ antibodies. The autoradiograph is representative of three independent experiments (A and B). C, plasmids containing the adiponectin promoter (Ϫ1130 to ϩ1 nucleotide(s)) directing luciferase expression and plasmids encoding SIRT1, Foxo1-⌬256, or C/EBP␣ were transfected into HEK293 cells. Luciferase activities were measured 24 h after transfection and normalized to ␤-galactosidase activity. The error bar indicates S.E. for four independent experiments. *, p Ͻ 0.05 versus pcDNA-transfected cells.
ferentiation by altering the expression of genes involved in cell cycle control and adipogenesis (8). Interestingly, during clonal expansion (days 1 and 2), Foxo1 activity is inhibited through high level phosphorylation and resultant exclusion from the nucleus (8). By day 3 of differentiation, phosphorylation of Foxo1 is substantially decreased, and Foxo1 is located in the nucleus (8). Therefore, despite increased Foxo1 expression, Foxo1 does not suppress adipocyte differentiation physiologically at the early stage, because its transactivation activity is suppressed by protein kinase B-mediated phosphorylation. Both our results 3 and the Nakae et al. study (8) demonstrate that overexpression of wild type Foxo1 has no effect on adipocyte differentiation. Therefore, unlike Foxa2 and Foxc2, which suppress adipocyte differentiation by inhibiting PPAR␥ or increasing preadipocyte factor-1, respectively (36,37), Foxo1 provides an integrating function for hormone-activated signaling pathways in a transcriptional cascade that promotes adipogenesis (38). Adiponectin gene expression is turned on after clonal expansion (19,20). During the same time period transactivation of Foxo1 is activated (8). Furthermore, Foxo1 haploinsufficiency significantly reduces adiponectin expression without significant changes in fat tissue mass (8). Taken together, we conclude that Foxo1 plays an important role in regulating adiponectin gene expression.
There are two main components of obesity: increased adipocyte number (hyperplasia) and cell size (hypertrophy). Adipocyte size is closely related with adiponectin expression, with reduced adiponectin expression in larger adipocytes (20,39,40). Despite many factors that influence adipocyte size, ultimately lipid storage plays a determining role. Picard et al. (9) recently reported that SIRT1 increases lipolysis in differentiated adipocytes. They also found that upon calorie restriction, SIRT1 expression is induced and fatty acids are mobilized in white adipose tissues, indicating that SIRT1 regulates lipid metabolism and decreases adipocyte size (9). Our results demonstrate that SIRT1 increases adiponectin gene expression. Thus, our studies provide mechanisms and consolidate the observations of adipocyte size and adiponectin gene expression. In addition, the Picard et al. study suggests that SIRT1 is a nuclear nutrient sensor in mammalian adipocytes (9). A similar function of SIRT1 in sensing metabolic status and regulating hepatic glucose metabolism has been recently reported (15). We speculate that, with nutritional deprivation, SIRT1 expression is induced, which leads to mobilization of fat storage in adipocytes and increased adiponectin gene expression. Increased adiponectin further sensitizes cells to insulin and enhances fatty acid oxidation (2), leading to a new energy homeostasis. In contrast to calorie restriction, long term calorie overload is the main reason for obesity. Here we show that SIRT1 and Foxo1 protein levels are reduced in fat tissues from high fat diet-induced obese and type 2 diabetic mouse models. Furthermore, our study demonstrated that SIRT1 up-regulates adiponectin gene transcription through a Foxo1-C/EBP␣ complex. Therefore, we propose that decreased expression of SIRT1 and Foxo1 and impaired Foxo1-C/EBP␣ transcription complex formation contribute to the diminished adiponectin gene expression in obesity. Apparently, more in vivo studies are required to verify this hypothesis.