Coordinated transcriptional control of adipocyte triglyceride lipase (Atgl) by transcription factors Sp1 and peroxisome proliferator–activated receptor γ (PPARγ) during adipocyte differentiation

The breakdown of stored fat deposits into its components is a highly regulated process that maintains plasma levels of free fatty acids to supply energy to cells. Insulin-mediated transcription of Atgl, the enzyme that mediates the rate-limiting step in lipolysis, is a key point of this regulation. Under conditions such as obesity or insulin resistance, Atgl transcription is often misregulated, which can contribute to overall disease progression. The mechanisms by which Atgl is induced during adipogenesis are not fully understood. We utilized computational approaches to identify putative transcriptional regulatory elements in Atgl and then tested the effect of these elements and the transcription factors that bind to them in cultured preadipocytes and mature adipocytes. Here we report that Atgl is down-regulated by the basal transcription factor Sp1 in preadipocytes and that the magnitude of down-regulation depends on interactions between Sp1 and peroxisome proliferator–activated receptor γ (PPARγ). In mature adipocytes, when PPARγ is abundant, PPARγ abrogated transcriptional repression by Sp1 at the Atgl promoter and up-regulated Atgl mRNA expression. Targeting the PPARγ–Sp1 interaction could be a potential therapeutic strategy to restore insulin sensitivity by modulating Atgl levels in adipocytes.

The breakdown of TGs 4 into glycerol and FAs, called lipolysis, is a catabolic process that serves to maintain FFA homeo-stasis (1,2). Under normal physiological conditions, plasma levels of NEFAs are controlled via clearance of circulating FAs and storage as TGs in adipose tissue. After meals, the pool of plasma NEFAs is increased by as much as 50% by lipoprotein lipasemediated cleavage of newly circulating TGs, whereas basallevel mobilization of NEFAs by lipolysis in adipose tissue is impaired by the release of insulin (1,3).
Obese individuals display several altered characteristics in this process, including decreased ability of adipose tissue to mobilize stored lipids, loss of insulin-mediated inhibition of NEFA release, and a general increase in plasma NEFA levels. There remains substantial debate regarding the meaning of the correlation between increased hypertrophy of adipocytes and high plasma NEFA levels in obesity (1,3); however, as circulating lipids exceed the energy requirements of the body, higher plasma NEFA levels do contribute to the accumulation of fat in the liver and skeletal muscle, further aggravating insulin resistance and perpetuating this cycle (4,5). As FFAs are critical factors in promoting insulin resistance, it is crucial to understand the mechanisms that control the basal breakdown of fat.
Insulin is classically described as an inhibitor of lipolysis (6 -9). The initial and rate-limiting step in TG breakdown is mediated by the enzyme ATGL, which yields the metabolites FA and diacylglycerol. The predominant mechanism by which insulin regulates Atgl is via transcriptional control (6,10). Insulin activates the mTORC1 pathway and simultaneously stimulates transcription of the transcription factor Egr1 in a mTORC1-dependent manner. Egr1 suppresses mTORC1-mediated Atgl transcription and decreases its expression both in cultured adipocytes and in adipose tissue derived from animals fed a high-fat diet (11). Insulin also controls nucleo-cytoplasmic shuttling of the transcription factor FoxO1, primarily via Akt-mediated phosphorylation and nuclear exclusion of FoxO1, reducing FoxO1-mediated transcription of Atgl (12). In addition, the master regulator of adipogenesis PPAR␥ has consensus binding sites within 3 kb upstream of the Atgl transcrip-tion start site (13) that respond to insulin treatment to positively regulate Atgl transcription (14).
Insulin exhibits control over the expression of other various metabolically important genes via transcriptional regulation as well, including the insulin receptor (15)(16)(17). Transcriptional control of a significant cohort of its target genes is mediated by the Sp1 family of transcription factors (18). The Sp1 family binds to GC-rich motifs at promoters and subsequently drives or inhibits transcription (19). Sp1 has been demonstrated to dynamically modify promoter recognition of various genes in response to insulin (17,18) and plays diverse roles in the control of a wide array of cellular processes, including cell growth (20,21), apoptosis (21), angiogenesis (22), immune response (23), and differentiation (24). Sp1 has multiple functional domains to regulate gene expression by other mechanisms as well, including via protein-protein interaction with transcription factors such as PPAR␥ (25), Egr1 (26), Ets1 (27), c-myc (28), c-Jun (29), and Stat1 (30).
Transcription of Atgl is regulated by insulin, and insulinmediated gene expression can be mediated via binding of the transcription factor Sp1 to specific promoters, which is itself transcriptionally regulated by insulin (31). In addition, PPAR␥ mediates transcription of target genes via a functional interaction with Sp1 (25). This inspired us to investigate the involvement of insulin-responsive Sp1 in the regulation of Atgl and to explore the potential of the PPAR␥-Sp1 interaction to regulate Atgl gene expression during adipocyte differentiation.

Sp1 binding to the Atgl promoter negatively regulates Atgl transcription
Analysis of approximately 3 kb of the DNA sequence immediately upstream of the Atgl transcription start site by Genomatix and TRANSFAC matrices revealed a putative Sp1-binding site at the minimal promoter (Ϫ50 to Ϫ36 bp). This site is conserved across numerous species, including mice and humans, suggesting an important biological function (Fig. 1A). Therefore, to test the involvement and mechanism of Sp1 in regulating the expression of Atgl, 3T3-L1 preadipocytes were co-transfected with luciferase reporter constructs of varying 5Ј end deletions of the Atgl promoter and an Sp1 expression vector. Sp1 overexpression significantly decreased luciferase activity across all Atgl promoter constructs, indicating that Sp1 binding at the minimal promoter region (Ϫ192 to ϩ21) can negatively affect Atgl expression (Fig. 1B). We examined the predicted Sp1-binding site by selectively mutating five conserved nucleotides (Ϫ46 5Ј-CCGCC-3Ј Ϫ42), as shown in Fig.  1C. These mutations completely prevented the inhibitory effect of Sp1 on the minimal promoter of Atgl (Fig. 1D), confirming the involvement of this site in the transcriptional inhibition of Atgl. When 3T3-L1 preadipocytes were transfected with variable-length Atgl promoter-luciferase reporter constructs and incubated with mithramycin A, an Sp1 inhibitor, for 24 h, it also abrogated the inhibitory effect of endogenous Sp1 when transfected with the Ϫ192/ϩ21 and Ϫ373/ϩ21 constructs contain- Figure 1. Atgl promoter-driven luciferase expression is inhibited by an Sp1-binding site. A, identification of an evolutionarily conserved putative Sp1-binding site in the Atgl promoter. All sites shown were identified by JASPAR as putative Sp1-binding sites (p Ͻ 0.05), with a higher score indicative of a stronger prediction of Sp1 binding. B, 3T3-L1 preadipocytes were co-transfected with luciferase reporter constructs of varying 5Ј-end deletions of the Atgl promoter and an Sp1 expression vector (pcDNA_Sp1). C, diagram of the putative Sp1-binding site at the Atgl minimal promoter (Ϫ50 to Ϫ36 bp) and the mutation (MUT) generated in the Atgl promoter constructs to disrupt this binding site. D, 3T3-L1 preadipocytes were co-transfected with luciferase reporter constructs of varying 5Ј-end deletions of the Atgl promoter bearing mutations in the predicted Sp1-binding site and an Sp1 expression vector. E, 3T3-L1 preadipocytes were co-transfected with variable-length Atgl promoter-luciferase reporter constructs and incubated with mithramycin A for 24 h. pgl2, empty vector luciferase control. not significant, p Ͼ 0.05; *, p Ͻ 0.05; **, p Ͻ 0.01; ***, p Ͻ 0.001; ****, p Ͻ 0.0001.

Transcriptional control of Atgl by Sp1 and PPAR␥
ing the Sp1-binding site (Fig. 1E). This further demonstrates that Sp1 exerts inhibitory control over Atgl gene expression via the Sp1-binding site at Ϫ36 to Ϫ50.
The significance of Sp1 on endogenous Atgl gene expression was further investigated by Sp1 depletion and overexpression studies in 3T3-L1 cells. Knockdown of Sp1 in preadipocytes via a lentiviral vector producing shRNA targeting Sp1 increased the mRNA levels of Atgl relative to a control shRNA (Fig. 2, A  and B). Conversely, overexpression of Sp1 in cultured adipocytes by electroporation (Fig. 2, C and D) resulted in lower expression levels of ATGL mRNA and protein (Fig. 2, E and F) with a concomitant reduction in basal glycerol release in the medium (Fig. 2G). In addition, cultured preadipocytes treated with mithramycin A (10 M) showed significant increase in ATGL levels both in the presence and absence of insulin (Fig.  2H). Also, in the presence of insulin, mithramycin A decreased Sp1 mRNA levels and increased Atgl mRNA levels (Fig. 2G). Collectively, this demonstrates that Sp1 mediates the negative regulation of Atgl in preadipocytes. The data further support that Sp1 reduces glycerol release in adipocytes via mechanisms that involve Atgl.

Sp1-mediated negative regulation of Atgl is abrogated by PPAR␥ in adipogenesis
Levels of PPAR␥ protein increase dramatically during adipogenesis, correlating with an increase in Atgl mRNA and protein levels (supplemental Fig. 1). Bioinformatic analyses identified putative PPAR␥-binding sites in the Atgl promoter at Ϫ2424, Ϫ1674, and Ϫ1573 bp (14). In the absence of ChIP data in the literature (11)(12)(13), we speculated that the profound up-regulation of Atgl transcription is due to the binding of the PPAR␥/ RXR␣ heterodimer (32)(33)(34) at the abovementioned sites.
Of note, PPAR␥ and Sp1 proteins both functionally and physically interact with each other (25). Thus, we hypothesized that PPAR␥ may abrogate the Sp1-mediated negative control of Atgl transcription. To test this, PPAR␥ and Sp1 overexpression

Transcriptional control of Atgl by Sp1 and PPAR␥
plasmids were co-transfected with wild-type or Sp1 sitemutated Ϫ2979/ϩ21 bp Atgl promoter constructs in 3T3-L1 preadipocytes. PPAR␥ overexpression alone or in the presence of Sp1 increased the reporter activity compared with Sp1 overexpression alone in cells transfected with the wild-type Ϫ2979/ϩ21 Atgl luciferase promoter construct (Fig. 3A). Upon mutation of the Sp1-binding site in the full-length Atgl promoter, PPAR␥ overexpression no longer induced Atgl expression (Fig. 3A). These data suggest that the Sp1 site (Ϫ50 bp to Ϫ36 bp) contributes to the regulation of Atgl by PPAR␥ in the context of the Ϫ2979 to ϩ21 promoter construct. The Sp1mediated transcriptional repression of Atgl in preadipocytes allows transcriptional activation by PPAR␥ during adipocyte differentiation. Furthermore, the presence of the Sp1 site has a positive effect on PPAR␥-mediated transcriptional activation of the Atgl promoter. To further support this hypothesis, PPAR␥ overexpression alone or with Sp1 did not induce reporter activity of the shorter Atgl promoter fragment (Ϫ192/ϩ21) in either the presence or absence of the Sp1 mutation (Fig. 3B). Taken together, we speculate that induction of PPAR␥ expression during adipogenesis reverses the negative control of Sp1 over Atgl transcription, presumably via direct interaction with Sp1.
The fact that PPAR␥ increases luciferase activity in the presence of Sp1 in the full-length promoter but not in the minimal promoter fragment shows that PPAR␥ stimulates Atgl transcription via binding sites upstream of Ϫ192 and by interacting directly with the Sp1-binding site at Ϫ50 bp to Ϫ36 bp. We speculate that the Sp1-mediated transcriptional down-regulation that is reversed by PPAR␥ is a stoichiometric transcriptional phenomenon in which there is a reversal of the role of the transcription complex, depending upon the available transcription factor levels. To further test this hypothesis, we knocked down PPAR␥ in preadipocytes via shRNA (Fig. 3C, left  panel). PPAR␥ deficiency decreased Atgl mRNA levels (Fig. 3C, right panel), further supporting the conclusion that PPAR␥ is a positive transcriptional regulator of Atgl in preadipocytes via mechanisms that involve its interaction with Sp1 and inhibition of the Sp1-mediated negative regulation of Atgl transcription.

PPAR␥-mediated transactivation of Atgl is Sp1-dependent
As the data suggested that Sp1 is a negative regulator of Atgl transcription in preadipocytes and that PPAR␥ overrides this process, we further investigated the importance of the Sp1binding site on PPAR␥-mediated transactivation of Atgl tran-

Transcriptional control of Atgl by Sp1 and PPAR␥
scription. During adipogenesis, there is a decrease in the nuclear levels of Sp1 and an increase in PPAR␥ levels in 3T3-L1 cells (supplemental Fig. 1). We co-transfected 3T3-L1 preadipocytes with full-length Atgl luciferase reporter constructs containing mutations at both the Sp1 (Ϫ50 bp to Ϫ36 bp) and PPAR␥-binding sites (Ϫ2428 bp to Ϫ2408 bp). We did not observe an increase in reporter activity by overexpression of PPAR␥ or PPAR␥ and Sp1 combined (Fig. 4A). PPAR␥ alone did not increase luciferase activity either in the presence or absence of the PPAR␥ agonist rosiglitazone. This indicates that the effect of the PPAR␥-binding site is to positively drive transcription of Atgl.
In mature adipocytes, expression from the full-length Atgl promoter (Ϫ2979/ϩ21) containing the PPAR␥and Sp1-binding sites was significantly greater than that from the minimal promoter (Ϫ192/ϩ21) containing only the Sp1-binding site (Fig. 4B), suggesting a positive regulatory role of PPAR␥. Interestingly, mutation of the Sp1-binding site in the full-length promoter that contains the PPAR␥-binding site decreased Atgl reporter expression (Fig. 4B), suggesting a positive role of Sp1 in Atgl transcription in mature adipocytes, where PPAR␥ is abundant. This effect is lost in the minimal promoter lacking the PPAR␥-binding site (Fig. 4B) and is in contrast to the data from preadipocytes, in which mutation of the Sp1 site in the fulllength promoter increased expression of the Atgl reporter construct (Fig. 3A). Together, this suggests that PPAR␥-mediated transactivation of Atgl is partially driven by the Sp1-binding site and that PPAR␥ binding to the promoter is essential to positively drive transcription of Atgl in mature adipocytes. Interestingly, this regulation is cell-type specific, as a luciferase reporter assay performed using Sp1 overexpression vectors and the fulllength Atgl promoter construct in MEF and C2C12 cells did not result in the reduction in luciferase activity that was detected in preadipocytes (Figs. 1B and 4C).
To examine the physical interaction between Sp1 and PPAR␥ proteins, co-immunoprecipitations were performed from 3T3-L1 protein lysates; they demonstrated that Sp1 interacts with PPAR␥ in this cell type (Fig. 4E). In addition, ChIP

Transcriptional control of Atgl by Sp1 and PPAR␥
analysis of Sp1 binding to the minimal Atgl promoter identified Sp1 genomic occupancy at the Atgl1 promoter in both preadipoctyes and mature adipocytes (Fig. 4F). Taken together, these data support the hypothesis that Sp1 and PPAR␥ coordinately directly regulate Atgl transcription in a cell type-and stage-dependent manner that is governed by the changes in PPAR␥ abundance that occur during adipocyte differentiation (Fig. 5).

Discussion
It has been well-documented that active transcriptional complexes can adopt alternative functions when bound to additional co-activators or co-repressors (35)(36)(37). For example, Klf1 is a transcription factor that exhibits alternate functions in hematopoietic cells, depending on the stage of development. In erythroid cells, a primitive cell type, Klf1 functions as an activator of ␤-globin expression. However, in hematopoietic cells, a definitive cell type, Klf1 switches to function as a repressor of ␤-globin expression (38). As Sp1 is a prominent member of the KLF family of transcription factors, similar mechanisms could explain the transcriptional switch identified in this study.
Our results showed that the shortest fragment of the Atgl promoter (Ϫ192/ϩ21) contains an evolutionarily conserved Sp1-binding site that is capable of inhibiting Atgl transcription when Sp1 is overexpressed exogenously. This negative impact of Sp1 on Atgl transcription is modulated by PPAR␥. In addition, cell type-specific responses were observed. In MEFs, which produce negligible amounts of PPAR␥, a positive regulation of the Atgl promoter was observed when Sp1 was expressed exogenously, further supporting the critical nature of the cellular transcription factor milieu in Atgl gene regulation. In adipocytes, we found that Sp1 allows PPAR␥-mediated transactivation via a protein-protein interaction between Sp1 and PPAR␥ (Fig. 4E). PPAR␥-mediated transactivation of the Atgl promoter decreased when the conserved Sp1 overlapping binding site at the proximal promoter was mutated (Fig. 4B). We propose that, although Sp1 maintains a steady promoter binding at the minimal promoter, as identified by ChIP analysis (Fig. 4F), and PPAR␥ and Sp1 directly interact (Fig. 4E), there is a complete reversal of the repressive transcription complex to an activating transcription complex during adipogenesis that governs the fate of Atgl transcription (Fig. 5).
PPAR␥ heterodimerizes with RXR␣, binds to peroxisomal proliferator response elements, and exhibits ligand-induced transactivation of gene expression (39). Our results show transcriptional suppression of the Atgl promoter reporter by the Sp1-binding site at Ϫ192/ϩ21 in the presence of exogenous PPAR␥ (Fig. 3A). As consistent binding of Sp1 at this smaller fragment is observed in both preadipocytes and mature adipocytes (Fig. 4E), we propose that physical interaction of Sp1-PPAR␥ contributes to Sp1-mediated repression and PPAR␥mediated transactivation of the Atgl gene in adipocytes. A similar mechanistic role of PPAR␥-Sp1 interaction has been shown to regulate thromboxane gene expression (25). The increased abundance of PPAR␥ in mature adipocytes seems likely to explain this reversal of Sp1 function, warranting further investigation into potential Sp1-binding partners.
Sp1-nuclear hormone receptor interaction is found to enhance Sp1-induced gene transcription (40 -42). This study demonstrates the possibility that other interacting partners of Sp1 play a role in mediating this process, in this case PPAR␥. However, there are a number of other proteins that interact with Sp1, including Egr1. It has been determined that both Sp1 and Egr1 occupy overlapping binding sites and can compete with each other for DNA binding, with Egr1 displacing Sp1 when the local concentration is more than that of Sp1 (43). Therefore, additional cofactors, such as Egr1, may be involved in the repression of Sp1 target gene expression. Another potential interacting partner is C/ebp ␤, which, like PPAR␥, increases during adipogenesis and has been shown to functionally interact with Sp1 (44,45). In fact, the presence of additional transcription factors such as C/ebp ␤ in mature adipocytes may contribute to the finding that, in preadipocytes, mutation of the Sp1-binding site in the Atgl promoter increased transcription, even in the presence of PPAR␥ (Fig. 3A), whereas mutation of this site in mature adipocytes led to an overall decrease in transcription (Fig. 4B).
In conclusion, Sp1 negatively regulates Atgl transcription in a PPAR␥-dependent fashion. It is interesting to note how, being part of the same transcription complex, the increased abundance of PPAR␥ during adipogenesis leads to a reversal in the transcriptional action of Sp1, which acts as a repressor in prea-

Transcriptional control of Atgl by Sp1 and PPAR␥
dipocytes and as an activator in fully differentiated adipocytes (Fig. 5). Understanding the alternating functions of these transcription factors in regulating Atgl in adipocytes is relevant to the treatment of insulin resistance and type 2 diabetes. Obesity is a predominant risk factor for insulin resistance, and insulin action is highly impaired in adipocytes of individuals with metabolic syndrome (46 -48). In these individuals, there is a reduction in hormone-induced lipolysis and an increase in the basal lipolysis of adipocytes (6,49), most likely because of the changes in the levels of Atgl. Therefore, targeting the specific PPAR␥-Sp1 interaction could possibly modulate Atgl levels in metabolic syndrome and restore whole-body insulin sensitivity.

RNA isolation and RT-qPCR
Total RNA was isolated from 3T3-L1 cells using TRIzol reagent (Invitrogen). RNA was then reverse-transcribed using SuperScript III First-Strand Synthesis SuperMix (Invitrogen), and levels of cDNA were quantified by RT-qPCR using Veri-Quest SYBR Green qPCR Master Mix (Affymetrix) with the StepOnePlus RealTime PCR System (Applied Biosystem). Primers used in the study are listed in supplemental Table S1. Relative mRNA levels were normalized to Gapdh using the ⌬⌬Ct method.

Immunoblotting and nuclear localization
For total protein lysates, cells were washed with cold PBS twice and lysed with RIPA buffer (50 mM Tris-HCl (pH 7.5), 150 mM NaCl, 2 mM EDTA, 1% Nonidet P-40, and 0.1% SDS) supplemented with Complete EDTA-free protease inhibitor and PhosSTOP phosphatase inhibitor tablets (Roche, 1 tablet/10 ml of RIPA buffer). Lysates were sonicated and centrifuged at 13,000 rpm for 10 min at 4°C. Supernatants were collected, and the amount of protein was quantified using the DC Protein Assay Kit (Bio-Rad).
For nuclear fractionation experiments, nuclei were isolated as described previously (50). Briefly, cells were pelleted, washed in PBS, and lysed with a mild detergent buffer (10 mM HEPES, 10 mM KCl, 1.5 mM MgCl 2 , and 0.3% Nonidet P-40). Nuclei were collected by centrifugation, washed in this buffer without detergent, and resuspended in a high-salt buffer (20 mM HEPES, (pH 7.9), 0.45 M NaCl, 1 mM EDTA, and 0.5 mM DTT). Nuclei were lysed via sonication, and protein was quantified as described above.
Equal quantities of protein were loaded and separated by SDS-PAGE and transferred to Immobilon-P membranes (Millipore) following standard procedures. Membranes were blocked with 5% nonfat milk in Tris-buffered saline with Tween 20 (TBST) for 1 h, rinsed in TBST, and incubated with a primary antibody overnight at 4°C. Membranes were then probed for respective HRP-conjugated secondary antibody for 1 h at room temperature. A Western Lightning ECL substrate kit (PerkinElmer Life Sciences) was used for detection, and bands were quantified using densitometry by Adobe Photoshop CS 5.0.

JASPAR analysis of Sp1-binding site
The Sp1 consensus motif in the promoter region of Atgl in mice was aligned to other species using the LastZ track in the Ensembl database (51). Conserved sequences were identified in a number of species. Analysis of these sequences using the JAS-PAR database (52) identified putative Sp1 motifs in all species shown in Fig. 1A.

Transient transfection, electroporation, and luciferase reporter assay
3T3-L1 preadipocytes were transiently transfected with plasmids using X-tremeGENE HP transfection reagent (Roche) according to the instructions of the manufacturer. Briefly, cells were grown to 70 -80% confluence and transfected with 500 ng of Atgl luciferase constructs (14), 500 ng of a plasmid encoding Sp1, 500 ng of a plasmid encoding PPAR␥, and 50 ng of a plasmid encoding Renilla luciferase in a 6-well plate format. After 48 h of transfection, cells were lysed using passive lysis buffer (Promega). Luciferase activities were determined by Dual Glo Luciferase Reporter Assay Kit (Promega), with values expressed in relative light units and normalized to Renilla luciferase levels.
For electroporation, 3T3-L1 differentiated adipocytes were washed with calcium-and magnesium-free warm PBS and trypsinized. Post-trypsinization, cells were washed twice with PBS (containing Ca 2ϩ and Mg 2ϩ ) and centrifuged at 2000 rpm for 5 min. Cells were resuspended in 500 l of PBS and electroporated with 50 g DNA using the Gene Pulser Xcell electroporation system (Bio-Rad) with a pulse setting of 160 V and 950 microfarad. Following electroporation, cells were resuspended in growth medium in 6-well plates. 24 h after electroporation, cells were collected for the luciferase activity assay or immunoblot analysis.

Site-directed mutagenesis
The putative Sp1-binding site at the Atgl minimal promoter region was identified using TRANSFAC matrices. Five conserved residues in this putative Sp1-binding site were mutated to thymine residues using the following set of primers: forward, 5Ј-CGACC AGGCC ttttt CTCAC CCCGC ACTAA AACAC-3Ј; reverse, 5Ј-GCAGG GGGCA GGACC TGG-3Ј. A putative PPAR␥-binding site was also located Ϫ2408 to Ϫ2428 bp upstream of the transcriptional start site, and five conserved Transcriptional control of Atgl by Sp1 and PPAR␥ residues were mutated to thymine residues using the following set of primers: forward, 5Ј-CTGAG TTCGA ttttt GCCTG GTCTA CAATG TGAGT TC-3Ј; reverse 5Ј-AAATC CACCT GCCTC TGC-3Ј. Mutagenesis was carried out using the Q5 site-directed mutagenesis kit (New England Biolabs) following the recommended protocol. Mutations were confirmed by Sanger sequencing.

Lipolysis assay
Differentiated 3T3-L1 adipocytes were incubated with phenol red-free DMEM with 2% fatty acid-free BSA for 6 h. Glycerol release was measured as a function of NADH consumption via absorbance at 340 nm using the Free Glycerol Kit (Megazyme).

ChIP
3T3-L1 preadipocytes and differentiated adipocytes were cross-linked by adding formaldehyde to the culture medium at a final concentration of 1% and incubated for 10 min at 37°C in a CO 2 incubator. The medium was then aspirated and washed twice with cold PBS (pH 7.4) containing protease inhibitors (1 mM phenylmethylsulfonyl fluoride, 1 g/ml aprotinin, and 1 g/ml pepstatin A). Cells were collected, lysed with buffer containing 0.2% SDS, and then sonicated using a Bioruptor (Diagenode) with eight pulses of 15 s at high setting. Tubes were centrifuged for 10 min, and the supernatant was immunoprecipitated with 2 g of anti-Sp1 antibody (Cell Signaling Technology) using Dynabeads protein A (Invitrogen). DNA was obtained by reverse cross-linking, purified, and analyzed via RT-qPCR with primers to detect the Sp1-binding site (forward, 5Ј-CGGCG GAGGC GGAGA CGCT-3Ј; reverse, 5Ј-TCCCT GCTTG ATCCA GTTGG AT-3Ј) using Premix TaqDNA Polymerase, Hot Start (catalog no. R028A, Takara) with 0.1% input.

Statistical analysis
All data are presented as mean Ϯ S.E. To evaluate statistical significance, two tailed Student's t test or one-way analysis of variance (followed by post tests, Dunnett's multiple comparison test and Welch correction test) were performed using the GraphPad Prism software package. p Ͻ 0.05 was considered to be statistically significant and is presented as follows: not significant, p Ͼ 0.05; *, p Ͻ 0.05; **, p Ͻ 0.01; ***, p Ͻ 0.001; ****, p Ͻ 0.0001.