Regulatory Role of Glycogen Synthase Kinase 3 for Transcriptional Activity of ADD1/SREBP1c* □ S

Adipocyte determination- and differentiation-depend-ent factor 1 (ADD1) plays important roles in lipid metabolism and insulin-dependent gene expression. Because insulin stimulates carbohydrate and lipid synthesis, it would be important to decipher how the transcriptional activity of ADD1/SREBP1c is regulated in the insulin signaling pathway. In this study, we demonstrated that glycogen synthase kinase (GSK)-3 negatively regulates the transcriptional activity of ADD1/SREBP1c. GSK3 inhibitors enhanced a transcriptional activity of ADD1/ SREBP1c and expression of ADD1/SREBP1c target genes including fatty acid synthase (FAS), acetyl-CoA carboxylase 1 (ACC1), and steroyl-CoA desaturase 1 (SCD1) in adipocytes and hepatocytes. In contrast, overexpression of GSK3 (cid:1) down-regulated the transcriptional activity of ADD1/SREBP1c. GSK3 inhibitor-medi-ated ADD1/SREBP1c target gene activation did not require de novo protein synthesis, implying that GSK3 might affect transcriptional activity of ADD1/SREBP1c at the level of post-translational modification. Addition-ally, we demonstrated that GSK3 efficiently phosphorylated ADD1/SREBP1c in vitro and in vivo . Therefore, these data suggest that GSK3 inactivation

between Wnt and Frizzled receptor inhibits GSK3, which results in dephosphorylation of ␤-catenin leading to the nuclear accumulation of ␤-catenin and activation of ␤-catenin/T cell factor target genes (4, 6 -9). Also, GSK3 plays an important role in the insulin signaling pathway and it phosphorylates and inhibits glycogen synthase in the absence of insulin (10). In insulin-sensitive tissues, insulin-derived GSK3 inactivation is mediated by phosphatidylinositol 3-kinase and protein kinase B/Akt (PKB/Akt), which phosphorylate at the N-terminal serine residue of GSK3 (serine 21 and serine 9 in GSK3␣ and -3␤, respectively) (5,11,12). Thus, recent reports suggest a potential role for GSK3 as a negative regulator of insulin signaling (13). Consistent with this notion, one of the GSK3 inhibitors, lithium chloride (LiCl), shows insulin-like effects including increase of glucose uptake, activation of glycogen synthase, and stimulation of glycogen synthesis in various tissues (13)(14)(15). Additionally, inhibition of GSK3 alters gene expression profiles in insulin-responsive cell lines. For example, LiCl selectively reduces expressions of glucose-6-phosphatase and phosphoenolpyruvate carboxykinase genes, whose expression is also inhibited by insulin (16). Furthermore, it has been demonstrated that GSK3 can phosphorylate several transcription factors, including p53, HSF-1, and C/EBP␣ and regulate their transcriptional activities (17)(18)(19).
Adipocyte determination-and differentiation-dependent factor 1 (ADD1/SREBP1c) is a bHLH-ZIP transcription factor that is involved in the stimulation of many lipogenic genes (20 -22), such as fatty acid synthase (FAS) (23), acetyl-CoA carboxylase (ACC) (24), and steroyl-CoA desaturase (SCD) (25). In insulinsensitive tissues including fat, liver, and muscle, insulin stimulates the expression of ADD1/SREBP1c mRNA and its target genes via phosphatidylinositol 3-kinase and PKB/Akt (26 -29). Therefore, it has been suggested that ADD1/SREBP1c plays a key role in orchestrating insulin-dependent lipid and glucose metabolism (22,30). Furthermore, several groups have investigated the post-translational modifications of ADD1/SREBP1c that affect its transcriptional activity. Brown and Goldstein (31)(32)(33)(34) have elegantly revealed that cholesterol-dependent proteolytic cleavage of SREBP family proteins is tightly controlled by SREBP cleavage-activating protein and Insigs. In addition, Kotzka et al. (35) reported that ADD1/SREBP1c is phosphorylated at the N-terminal end of the protein through the MAP kinase pathway, and such phosphorylation of ADD1/ SREBP1c stimulates its transcriptional activity (35). However, PD98059, an inhibitor of MAP kinase pathway, did not antagonize the effect of insulin on the expression of ADD1/SREBP1c target genes in cultured hepatocytes (28,36,37), implying that the MAP kinase pathway might not be a major player for ADD1/SREBP1c target gene expression by insulin treatment.
Given the importance of GSK3 as a negative regulator of insulin signaling and ADD1/SREBP1c as a mediator of insulinresponsive gene expression, it would be important to understand the molecular mechanism between inhibition of GSK3 and increase of ADD1/SREBP1c transcriptional activity by insulin. To address this question, we examined the roles of GSK3 on the regulation of transcriptional activity of ADD1/SREBP1c. Here, we demonstrate that GSK3 inhibitors enhance the transcriptional activity of ADD1/SREBP1c and promote the expression of ADD1/SREBP1c target genes without de novo protein synthesis. Additionally, we reveal a phosphorylation event of ADD1/SREBP1c by GSK3. Taken together, these results propose a novel regulatory role of GSK3 in the insulin-dependent activation of ADD1/SREBP1c in adipocytes and hepatocytes.

EXPERIMENTAL PROCEDURES
Reagents and Antibodies-LiCl and cycloheximide were obtained from Sigma and SB-216763 was purchased from Tocris. GSK3 antibody was purchased from BD Sciences. Polyclonal antibody against ADD1/ SREBP1c was previously described (21). Mammalian expression vector for GSK3␤ was thankfully provided from Dr. Junichi Sadoshima (New Jersey Medical School) (38).
Transient Transfections and Luciferase Assays-HEK293 cells and HepG2 hepatocytes were transfected with several DNA constructs 1 day before confluence by the calcium phosphate method described previously (22,39). The mammalian expression vector for ADD1/SREBP1c contains 1-403 amino acids, which was engineered by in-frame fusion with the Myc His tag derived from pCDNA3.1 (Invitrogen). After being starved with low glucose DMEM containing 0.1% bovine serum albumin for 12 h, cells were treated with LiCl, NaCl, or insulin and incubated for 24 h. Total cell extracts were prepared with lysis buffer (25 mM Tris phosphate (pH 7.8), 2 mM dithiothreitol, 2 mM 1,2-diaminocyclohexane-N,N,NЈ,NЈ-tetraacetic acid, 10% glycerol, and 1% Triton X-100), and the activities of ␤-galactosidase and luciferase were measured according to the manufacturer's instructions (Promega). Rat1-IR cells were transfected with the same method as described earlier (22,39).
Northern Blot Analysis-Total RNA was isolated with TRIzol reagent (Invitrogen) according to the manufacturer's protocol. Twenty-five g of each RNA were denatured in formamide and formaldehyde buffer, and separated by electrophoresis on formaldehyde-containing agarose gels and transferred to nylon membranes (Schleicher and Schuell). After cross-linking, membranes were hybridized and washed as described by the manufacturer's instructions. DNA probes were labeled by random priming using the Klenow fragment of DNA polymerase I (MBI Fermentas) and [␣-32 P]dCTP (Amersham Biosciences). The cDNAs used as probes were: FAS, ACC1, SCD1, ADD1/SREBP1c, and 36B4.
Immunoprecipitation and Assay of Protein Kinases (IP Kinase Assay)-Equal amounts of total cell extract were incubated with GSK␤ antibody (Transduction Laboratories). The immunoprecipitates collected with protein A-Sepharose (Sigma) were pelleted and washed twice with 1 ml of NETN buffer (20 mM Tris-HCl (pH 8.0), 1 mM EDTA, 0.5% Nonidet P-40, and 100 mM NaCl). Activity of immunoprecipitated GSK3 complexes were determined by counting counts/min or autoradiography as mentioned previously.

Lithium Chloride Stimulates the Transcriptional Activity of ADD1/SREBP1c and Its Target Gene Expression-To investi-
gate whether the transcriptional activity of ADD1/SREBP1c is regulated by GSK3, LiCl, an inhibitor of GSK3, was used in luciferase reporter assays. Previous studies revealed that LiCl inhibits GSK3 activity with an IC 50 of 2.5 mM and a 90% loss of GSK3 activity is observed with 32 mM LiCl treatment (41). FAS is one of the best characterized target genes of ADD1/SREBP1c and its promoter region contains the binding sites for ADD1/ SREBP1c (22,42,43). As shown in Fig. 1A, LiCl stimulated the FAS promoter activity in the presence of ADD1/SREBP1c in a dose-dependent manner. However, it had no influence on the basal promoter activity in the absence of ADD1/SREBP1c in HEK293 cells. When 20 mM sodium chloride (NaCl) was treated, we did not observe any changes in promoter activity, suggesting that transcriptional activity of ADD1/SREBP1c is specifically modulated by LiCl. Similar results were obtained in Rat1-IR cells that express insulin receptor (Fig. 1B).
Next, to investigate the alteration of target gene expression of ADD1/SREBP1c by GSK3, we performed a Northern blot analysis using insulin-responsive cells, such as differentiated 3T3-L1 adipocytes and FAO hepatocytes in the absence or presence of GSK3 inhibitor. Consistent with the results from reporter assays, LiCl treatment remarkably increased the mRNA expression of ADD1/SREBP1c target genes, such as FAS, ACC1, and SCD1 (Fig. 2, A and B). However, LiCl had little effect on the expression level of nuclear SREBP1 protein (Fig. 2C). Together, these observations indicate that inhibition of GSK3 clearly stimulates transcriptional activity of ADD1/ SREBP1c and enhances its target gene expression in adipocytes and hepatocytes.
Inhibition of GSK3 by SB-216763 Stimulates ADD1/ SREBP1c Target Gene Expression-Because LiCl has several targets other than GSK3, such as the phosphatidylinositol signaling pathway via inhibition of inositol monophosphatase (44), we performed similar experiments using a selective GSK3 inhibitor, SB-216763 (45,46). As shown in Fig. 3, A and B, SB-216763 evidently enhanced FAS mRNA expression in 3T3-L1 adipocytes and HepG2 cells in a dose-dependent manner, similar to LiCl. The nuclear form of the SREBP1 protein was not significantly changed by SB-216763 treatment (Fig. 3C).
To clarify whether SB-216763 efficiently inhibits GSK3 kinase activity, IP kinase assays were conducted. After HepG2 cells were treated with or without GSK3 inhibitors, total cell lysates were immunoprecipitated with GSK3 antibody. Then, immune complexes were incubated with phospho-eIF2B peptide, one of the well known substrates of GSK3 (40,47), and GSK3 kinase activities were measured. The kinase activity of GSK3 was decreased (about 50% reduction) by LiCl or SB-216763 treatment at concentrations of 10 mM or 10 M, respectively (Fig. 3D, lanes 3 and 4). These observations implicate that LiCl-or SB-216763-stimulated target gene expression of ADD1/SREBP1c is closely associated with the inhibition of GSK3 kinase activity.
Overexpression of GSK3␤ Inhibits Transcriptional Activity of ADD1/SREBP1c-To confirm the effect of GSK3 on the transcriptional activity of ADD1/SREBP1c, we investigated the effects of GSK3␤ overexpression. Luciferase reporter assays revealed that overexpression of GSK3␤ decreased the transcriptional activity of ADD1/SREBP1c in a dose-dependent fashion (Fig. 4A). Furthermore, we examined the effect of ADD1/SREBP1c target gene expression in the absence or presence of GSK3␤ overexpression. As shown in Fig. 4B, ectopic expression of GSK3␤ in HepG2 cells concomitantly inhibited FAS mRNA levels, whereas SB-216763 treatment reversed GSK3␤-mediated down-regulation of FAS mRNA expression. These results suggest that GSK3␤ plays a negative regulator for ADD1/SREBP1c.

LiCl-and SB-216763-induced Target Gene Expression of ADD1/SREBP1c Does Not Require de Novo Protein Synthesis-
Although it has been demonstrated that an increase of ADD1/ SREBP1c mRNA expression is important for insulin-stimulated gene expression (30,48), GSK3 inhibitors shows little effect on the mRNA level of ADD1/SREBP1c and protein level of nuclear SREBP1 (Figs. 2 and 3). Thus we hypothesized that the changes of the ADD1/SREBP1c target gene expression with GSK3 inhibitors can occur at the level of post-translational modification such as phosphorylation. To test the above idea, cycloheximide, an inhibitor of de novo protein synthesis (49), was treated in the absence or presence of LiCl or SB-216763. As shown in Fig. 5, A and B, cyclohexamide did not alter the induction of FAS mRNA expression by LiCl or SB-216763 treatment. These results suggest that GSK3-mediated regulation of the ADD1/SREBP1c transcriptional activity does not require additional protein synthesis, and that modulation of ADD1/ SREBP1c transcriptional activity by GSK3 is presumably accomplished by post-translational modification of ADD1/SREBP1c. Lithium Chloride Does Not Completely Mimic Insulin-stimulated Gene Expression-It has been reported that LiCl mimics insulin-stimulated glucose metabolism including an increase of glycogen synthesis and glucose uptake and regulates several hepatic gluconeogenic gene expressions, including glucose-6phosphatase and phosphoenolpyruvate carboxykinase (16). However, it has not been clearly understood whether LiCl sufficiently mediates insulin-dependent gene expression. To clarify the idea whether LiCl is able to substitute for insulin action in regulation of several gene expression, we performed a luciferase reporter assay in Rat1-IR cells in the presence or absence of LiCl and/or insulin. Treatment of insulin or LiCl stimulated the transactivation of FAS promoter by ADD1/SREBP1c (Fig. 6A, lanes 4 and 6), and simultaneous treatment of LiCl and insulin showed additive effects (Fig. 6A, lane 8). Likewise, cotreatment of LiCl and insulin increased FAS mRNA expression on 3T3-L1 adipocytes (Fig. 6B). But LiCl had minor effects on the expression level of ADD1/ SREBP1c, whereas insulin dramatically stimulated ADD1/ SREBP1c expression (Fig. 6C). Together, these results implicate that LiCl does not completely mimic insulin effects, at least, for the activation of ADD1/SREBP1c, suggesting that LiCl and insulin might have cooperative effects to stimulate ADD1/ SREBP1c activity.
GSK3␤ Phosphorylates ADD1/SREBP1c in Vitro and in Vivo-Because inhibition of GSK3 by LiCl or SB-216763 increased the transcriptional activity of ADD1/SREBP1c without de novo protein synthesis, we decided to examine whether GSK3 is directly able to phosphorylate ADD1/SREBP1c protein. GSK3 has two isoforms, GSK3␣ and -␤ share 98% homology in their catalytic domain and have similar substrate specificity (50,51). Because GSK3␤ is highly expressed in adipocytes and hepatocytes, we used GSK3␤ in the following experiments. To test the possibility that ADD1/SREBP1c is a novel substrate of GSK3␤, we purified recombinant GSK3␤ proteins from E. coli or insect cells. When in vitro kinase assays were conducted with recombinant GSK3␤ proteins from E. coli, ADD1/SREBP1c was strongly phosphorylated as well as GSK3␤ itself, implying that ADD1/SREBP1c is a putative substrate of GSK3␤, at least, in vitro (Fig. 7A). We obtained the same results using GSK3␤ proteins purified from insect cells (Fig. 7B). Furthermore, to validate the effect of purified recombinant GSK3␤ protein, we examined whether the kinase activity of recombinant GSK3␤ is directly inhibited by LiCl or SB-216763. Treatment of LiCl caused a significant reduction of kinase activity of GSK3␤, whereas treatment of SB-216763 completely abolished the kinase activity of GSK3␤ in vitro (Fig.  7C). Next, to test whether GSK3␤ phosphorylates ADD1/ SREBP1c in vivo, endogeneous GSK3␤ was obtained by IP, and IP kinase assays were carried out with ADD1/SREBP1c as substrate. As shown in Fig. 7D, ADD1/SREBP1c was clearly phosphorylated by endogenous GSK3␤ in both HepG2 hepatocytes and 3T3-L1 adipocytes, implying that ADD1/SREBP1c is a novel substrate for GSK3.
In addition, we performed in vivo orthophosphate labeling in the absence or presence of LiCl or insulin. Consistent with the in vitro kinase assay (Fig. 7), the basal phosphorylation level of ADD1/SREBP1c protein was decreased by LiCl and SB-216763 treatment (Fig. 8A). Previously, it has been reported that insulin might induce serine phosphorylation of ADD1/SREBP1c (22). We also observed that total phosphorylation of ADD1/ SREBP1c was moderately increased by insulin (Fig. 8B, lane  3), and insulin treatment partially rescued LiCl-mediated reduction (Fig. 8B, lane 5), implying that GSK3 might phosphorylate ADD1/SREBP1c in resting status. Thus, it appears that GSK3 and other kinase(s) would be involved in the regulatory phosphorylation of ADD1/SREBP1c to mediate insulin-dependent gene expression. DISCUSSION ADD1/SREBP1c is a member of key transcription factors that regulate genes involved in cholesterol and fatty acid metabolism. Three isoforms of the SREBP family have been identified in several mammalian species. SREBP-1a and ADD1/ SREBP1c are generated by using alternative promoters from a single gene. SREBP-2 is derived from a different gene with 47% similarity to SREBP-1 (52). Among the three SREBP isoforms, the transcript level of ADD1/SREBP1c is sensitively controlled by insulin and nutritional status, which are associated with lipid and glucose metabolism (30,48). As a major anabolic hormone, insulin enhances the transcriptional activity of ADD1/SREBP1c at the level of its own transcription and posttranslational modifications (22,26,28,35,48,53). With this regulatory mechanism, ADD1/SREBP1c induces expression of several lipogenic genes, which leads to fatty acid and triglyceride synthesis and regulates expression of several genes involved in glucose metabolism (28,30,54,55), suggesting that All transfection experiments were performed in duplicate and repeated at least three times. B, Northern blot analysis. After serum starvation, 3T3-L1 adipocytes were treated with LiCl or insulin as in panel A for 12 h. Total RNA was isolated and analyzed using Northern blot. C, a semiquantitative reverse transcriptase-PCR experiment was carried out with ADD1/ SREBP1c-specific primers (see details under "Experimental Procedures"). Glyceraldehyde-3-phosphate dehydrogenase (GAPDH) was used to normalize each sample. ADD1/SREBP1c plays a critical role in insulin-stimulated gene expression for whole body energy homeostasis. Indeed, ADD1/ SREBP1c-specifc knockout mice reveal not only a severe decrease of hepatic and plasma triglyceride with reduced fat mass but also elevation of plasma glucose levels (56).
Several lines of recent evidence indicate that GSK3 is involved in insulin resistance, which is defined as the status where peripheral tissues cannot respond to insulin and is closely associated with obesity and type II diabetes. Inhibition of GSK3 improves insulin resistance not only by an increase of glucose disposal rate in Zucker diabetic rats but also by inhibition of gluconeogenic genes such as phosphoenolpyruvate carboxykinase and glucose-6-phosphatase in hepatocytes (16,57,58). Furthermore, selective GSK3 inhibitors potentiate insulin-dependent activation of glucose transport and utilization in muscle in vitro and in vivo (58). GSK3 also directly phosphorylates serine/ threonine residues of insulin receptor substrate-1, which leads to impairment of insulin signaling (59,60).
In this study, we identified a novel role of GSK3 in the regulation of ADD1/SREBP1c transcriptional activity. Treatment of GSK3 inhibitors remarkably enhanced the transcriptional activity of ADD1/SREBP1c and stimulated the expression of its target genes such as FAS, SCD1, and ACC1, which are known to be key players for fatty acid metabolism as well as insulin-sensitive genes, in adipocytes and hepatocytes (Figs.  1-3). Accordingly, we observed that the levels of ADD1/ SREBP1c target gene expression by GSK3 inhibitors were similar to those by insulin (Fig. 6), implicating that stimulation of ADD1/SREBP1c by insulin appears to be mediated, at least partly, by GSK3 inhibition. We also observed that insulin augmented the effects of the GSK3 inhibitor on the transcriptional activity of ADD1/SREBP1c. In addition to the role of GSK3 in insulin signaling, these results might provide a clue to explain why expression of sev-  2-4) was transfected into HEK293 cells. After incubating with [ 32 P]orthophosphate in phosphate-free DMEM medium, cells were treated with LiCl (10 mM) or SB-216763 (10 M), and ADD1/SREBP1c protein was isolated by immunoprecipitation. Phosphorylation of ADD1/SREBP1c was quantified with phosphorimager (BAS1500; Fuji, Japan). B, similar experiments were carried out using LiCl (10 mM) or insulin (100 nM). eral lipogenic genes is reduced in fat tissues of insulin-resistant animals (61). Compared with lean mice, the levels of GSK3 and its enzyme activity are elevated in fat tissues or skeletal muscles of obese and diabetic mice models (62,63), proposing that an increase of GSK3 might be associated with abnormal energy metabolism by disrupting insulin action. Under these circumstances, it appears that transcriptional activity of ADD1/ SREBP1c might be repressed by increased GSK3 activity, leading to the reduced expression of lipogenic genes in fat tissues of obese and diabetic mice.
Previously, Roth et al. (64) demonstrated that MAP kinase phosphorylates SREBP-1a at serine 117 in vitro and mutation of this serine residue to alanine abolishes responsiveness of SREBP-1a to insulin and platelet-derived growth factor. However, the same for ADD1/SREBP1c was not confirmed and inhibitors of the MAP kinase pathway did not inhibit insulinstimulated ADD1/SREBP1c target gene expression (28,36,37). Rather, we suggest that ADD1/SREBP1c can be phosphorylated by GSK3, which is a downstream kinase of phosphatidylinositol 3-kinase in the insulin signaling pathway. With in vitro and in vivo assays, we showed that GSK3 is able to phosphorylate ADD1/SREBP1c (Fig. 7), and the basal level of ADD1/SREBP1c phosphorylation was drastically decreased by the GSK3 inhibitor (Fig. 8), clearly indicating that GSK3 might induce negative phosphorylation of ADD1/SREBP1c in resting cells. Additionally, we observed that GSK3 was colocalized with the mature form of ADD1/SREBP1c in the nucleus (data not shown). Therefore, it is likely that inactivation of GSK3 might be an important regulatory step that stimulates the transcriptional activity of ADD1/SREBP1c.
To identify the phosphorylation residue(s) of ADD1/ SREBP1c by GSK3␤, we performed phosphoamino acid analysis. As a result, we observed that ADD1/SREBP1c was predominantly phosphorylated at serine residues (Supplemental Materials Fig. 1). Furthermore, by using different truncated ADD1/SREBP1c recombinant proteins, we observed that the N-terminal end of ADD1/SREBP1c was phosphorylated by GSK3␤ (Supplemental Materials Fig. 2). Although we identified at least six putative target sites of GSK3␤ in the Nterminal portion of the ADD1/SREBP1c protein, the exact site(s) of phosphorylation has not been identified yet. We are currently investigating the phosphorylation site(s) of ADD1/ SREBP1c by GSK3␤ and the functional roles of those sites in regulating the transcriptional activity of ADD1/SREBP1c. Taken together, these data suggest that GSK3 plays a role in regulating the activity of ADD1/SREBP1c and that insulinmediated induction of ADD1/SREBP1c target gene expression might be associated with the regulation of GSK3. Consistent with this view, inhibitors of GSK3, both LiCl and SB-216763, stimulate ADD1/SREBP1c target gene expression by altering its transcriptional activity. Therefore, GSK3 appears to negatively regulate the transcriptional activity of ADD1/SREBP1c by phosphorylating ADD1/SREBP1c and this repression process is probably relieved when GSK3 is suppressed in the presence of the insulin signaling pathway.