Rosiglitazone Induction of Insig-1 in White Adipose Tissue Reveals a Novel Interplay of Peroxisome Proliferator-activated Receptor (cid:1) and Sterol Regulatory Element-binding Protein in the Regulation of Adipogenesis*

Insulin-induced gene 1 (INSIG-1) is a key regulator in the processing of the sterol regulatory element-binding proteins (SREBPs). We demonstrated that Insig-1 is regulated by peroxisome proliferator-activated receptor (cid:1) (PPAR (cid:1) ) providing a link between insulin sensitization/ glucose homeostasis and lipid homeostasis. Insig-1 was identified as a PPAR (cid:1) target gene using microarray analysis of mRNA from the white adipose tissue of diabetic ( db/db ) animals treated with PPAR (cid:1) agonists. In-sig-1 was induced in subcutaneous (9-fold) and epididymal (4-fold) fat pads from db/db mice treated for 8 days with the PPAR (cid:1) agonist rosiglitazone (30 mg/kg/day). This in vivo response was confirmed in differentiated C3H10T1/2 adipocytes treated with rosiglitazone. To elu-cidate the molecular mechanisms regulating INSIG-1 Reporter Genes— The human LDL receptor promoter-luciferase con- struct has been described previously (22). The human INSIG-1 (cid:1) 1233 promoter ( (cid:1) 1233 to (cid:3) 23, relative to the transcription start site) was amplified from human genomic DNA using a 5 (cid:5) primer that contained an NheI restriction site followed by the nucleotides corresponding to (cid:1) 1233 to (cid:1) 1214 (5 (cid:5) -gagagctagcgaaaggtgccctgtccatta-3 (cid:5) ) and the 3 (cid:5) primer that contained a BglII site and the nucleotides corresponding to (cid:3) 5 to (cid:3) 23 (5 (cid:5) -gagaagatctggcccgggacggccgcagc-3 (cid:5) ) of the published genomic sequence (14). The PCR product was cloned into NheI/BglII sites of the pGL3 basic vector (Promega) to generate hINSIG-1 (cid:1) 1233 ( (cid:1) 1233 to (cid:3) 23). The hINSIG-1 promoter-truncation constructs were generated using the following 5 (cid:5) primer and the common 3 (cid:5) primer as mentioned above. The hINSIG-1 (cid:1) 861 ( (cid:1) 861 to (cid:3) 23) construct was generated using a 5 (cid:5) primer with nucleotides corresponding to (cid:1) 861 to (cid:1) 841 (5 (cid:5) -gagagctagcgcgtgtcatcctcaggaaat-3 (cid:5) ). The hINSIG-1 (cid:1) 752 con- struct was generated using a 5 (cid:5) primer corresponding to nucleotides (cid:1) 752 to (cid:1) 738 (5 (cid:5) -gagagctagcgccacaaccccagg-3 (cid:5) ). The hINSIG-1 (cid:1)

Ligands of the peroxisome proliferator-activated receptor ␥ (PPAR␥) 1 promote adipogenesis, stimulate glucose disposal in skeletal muscle, and depress glucose production from the liver (1)(2)(3). The molecular mechanisms and target tissues through which these ligands (such as the thiazolidinediones) exert their antidiabetic activity are not entirely clear, but this activity is thought to have a direct effect in adipose tissue, where PPAR␥ is abundant, and an indirect effect in muscle and liver, where levels of PPAR␥ are much lower. The synthetic PPAR␥ ligands clinically used to treat type 2 diabetes have the problematic side effects of inducing weight gain and edema, with the potential for congestive heart failure (4,5). Safer PPAR␥ modulators that retain the insulin sensitization properties while minimizing side effects are desirable. However, it is not clear whether the adipogenic activity (by lowering circulating free fatty acids) is also responsible for insulin sensitization in other tissues.
PPREs have been identified in many genes encoding enzymes involved in fatty acid oxidation and synthesis (6,7). Microarray and related technologies have allowed identification of a number of other genes in which expression is modulated by PPAR␥ ligands (8 -10). We have identified insulininduced gene 1 (Insig-1) as one of the genes highly induced in the adipose tissue of db/db mice treated with rosiglitazone. Insig-1/Cl-6 was originally isolated as the most abundant immediate early transcript in both insulin-treated hepatoma cells and in regenerating rat liver (11,12). It was also found to be induced both in epididymal fat in a diet-induced obesity model (13) and late in adipocyte differentiation of 3T3-L1 fibroblasts treated with dexamethasone, insulin, and isobutylmethylxanthine (14,15). Recently, the co-immunoprecipitation of IN-SIG-1 with SREBP cleavage-activating protein (SCAP) implicated involvement of INSIG-1 in cholesterol and lipid homeostasis (16).
SREBPs are transcription factors whose activity is modulated by intracellular sterol levels. Full-length SREBP is a membrane-bound protein located in the endoplasmic reticulum where it is associated with SCAP. When intracellular sterol levels are depleted, SCAP and SREBP migrate to the Golgi where SREBP is cleaved twice, releasing the transcriptionally active amino-terminal polypeptide (nSREBP). nSREBP translocates to the nucleus where it activates the transcription of a number of genes involved in cholesterol metabolism (notably LDL receptor and hydroxymethylglutaryl-CoA reductase) and fatty acid and triglyceride synthesis (e.g. fatty-acid synthase and phosphoglycerol acyltransferase) (17). Two genes code for SREBP-1 and SREBP-2, and there are alternate splice forms as well (17). The major forms in liver are SREBP-1c (primarily influencing fatty acid metabolism and lipogenesis) and SREBP-2 (central to cholesterol homeostatic mechanisms). A third form, SREBP-1a, differs from SREBP-1c by containing additional acidic amino acids at the amino terminus, which confer greater transactivational activity. When the INSIG-1 protein is overexpressed in cultured cells, it tethers the SCAP⅐SREBP complex in the endoplasmic reticulum and prevents translocation to the Golgi. This blocks SREBP maturation to nSREBP and reduces SREBP-dependent transcription (16).
SREBP-1c/ADD1 is also integral to the process of adipogenesis and is believed to mediate insulin-induced changes in gene expression in adipose tissue, reflecting the nutritional state of the organism (18 -20). Li et al. (15) have proposed that INSIG-1 serves as a "brake" for lipogenesis in fat, limiting the deposition of triglycerides in individual cells and possibly also influencing the recruitment of preadipocytes. What emerges is a highly integrated network of transcription factors (CCAAT enhancerbinding proteins or C/EBPs, PPARs, and SREBPs) and IN-SIG-1 as a check on the system, ensuring energy balance in the healthy state.
We present evidence that SREBP and PPAR␥ directly influence expression of INSIG-1 via sterol-responsive and PPARresponsive elements in the INSIG-1 promoter. The PPAR␥ ligand rosiglitazone induces endogenous Insig-1 in adipose tissue of db/db mice and in differentiated adipocytes in vitro. These data further implicate PPAR␥ as a central player in adipose tissue homeostasis.

EXPERIMENTAL PROCEDURES
Animals and Treatments-All in vivo procedures were approved by the Ligand Institutional Animal Care and Use Committee. Male db/db mice and lean littermates were purchased from Harlan Teklad (Indianapolis, IN) and housed on a 12-h light/12-h dark lighting regimen (lights on at 6:00 a.m.). The animals had free access to food (Purina, 5008) and water throughout the study. After 2 weeks of acclimation, the 7-week-old animals were weighed, bled via the tail vein in the nonfasted state, and sorted into treatment groups based on glucose levels and initial body weights. From the following day, the animals received a single daily oral treatment by gavage with vehicle or a suspension of compound in vehicle. Vehicle was 1% carboxymethylcellulose, 0.25% Tween 80 or 0.085% povidone, 1.5% lactose, 0.026% Tween 80, and 0.2% v/v Antifoam. No vehicle-dependent differences in the potency or efficacy of compounds were observed. Fenofibrate was purchased from Sigma. Rosiglitazone was synthesized at Eli Lilly, and LG268 was synthesized at Ligand Pharmaceuticals. The molecular structure and purity of both were confirmed by NMR prior to use in vivo. Following the terminal dose, food was withheld, and 6 h later, the animals were anesthetized with isoflurane, blood was sampled by cardiac puncture, and the animals were killed by CO 2 asphyxiation. Tissues were harvested, rapidly frozen in liquid nitrogen, and stored at Ϫ80°C.
Preparation of RNA and cDNA from Liver, White Adipose Tissue, and C3H10T1/2 Cells-Frozen tissues (subcutaneous and epididymal fat) were homogenized in TRIzol reagent (Sigma), and total RNA was isolated according to the manufacturer's protocol. RNA concentration was determined by absorbance at 260 nm. Each RNA sample was reverse transcribed using a Superscript preamplification system for first strand cDNA synthesis (Invitrogen) and oligo(dT) primers, according to the manufacturer's protocol. RNA was stored at Ϫ80°C and cDNAs at Ϫ20°C. C3H10T1/2 cells were plated at 5 ϫ 10 5 cells/10 cm plate in Dulbecco's modified Eagle's medium ϩ 10% fetal calf serum. Two days later when the cells were reaching confluency, the medium was changed to Dulbecco's modified Eagle's medium ϩ 10% fetal calf serum ϩ 10 g/ml insulin Ϯ 1 M rosiglitazone. The cells were washed in phosphate-buffered saline and lysed in 1 ml of TRIzol reagent 24, 72, or 120 h later, and total RNA and cDNA were prepared as for animal tissues.
Real-time PCR Assays-Relative mRNA levels of target genes and an invariant transcript, 36B4, were determined by real-time PCR using a single cDNA preparation for tissues from individual animals or cell culture conditions. For each transcript sequence to be quantified, a TaqMan probe and flanking primers (listed in Table I) were chosen using Primer Express software (PerkinElmer Life Sciences). Amplicons were positioned within 2 kb of the polyadenylation sequence. Individual PCR reactions contained cDNA prepared from 10 ng of total RNA and a master mix such that final concentrations were: 100 nM probe, 100 -900 nM primers optimized for each target, and 1ϫ TaqMan universal PCR master mix (PerkinElmer Life Sciences, proprietary). PCR was carried out and monitored by the ABI PRISM 7700 sequence detection system (PerkinElmer Life Sciences) for 40 cycles of amplification at 95°C for 15 s and 60°C for 1 min. One cDNA from each experiment was designated as a standard, and serial dilutions were prepared in nuclease-free water. Each 96-well reaction plate contained: duplicate standards of five concentrations covering a 2-log range, triplicate determinations of each unknown, as well as no template and no reverse transcriptase controls. Quantification of unknowns was performed by Sequence Detection Systems version 1.6 software (PerkinElmer Life Sciences) and was relative to the arbitrary standard of similar composition. The same standard dilution series was used for all target gene assays within a given in vivo experiment. SREBP-1a and SREBP-1c transcripts were measured by Sybr green incorporation using primer sequences published previously (21).
Reporter Genes-The human LDL receptor promoter-luciferase construct has been described previously (22). The human INSIG-1 Ϫ1233 promoter (Ϫ1233 to ϩ23, relative to the transcription start site) was amplified from human genomic DNA using a 5Ј primer that contained an NheI restriction site followed by the nucleotides corresponding to Ϫ1233 to Ϫ1214 (5Ј-gagagctagcgaaaggtgccctgtccatta-3Ј) and the 3Ј primer that contained a BglII site and the nucleotides corresponding to ϩ5 to ϩ23 (5Ј-gagaagatctggcccgggacggccgcagc-3Ј) of the published genomic sequence (14). The PCR product was cloned into NheI/BglII sites of the pGL3 basic vector (Promega) to generate hINSIG-1 Ϫ1233 (Ϫ1233 to ϩ23). The hINSIG-1 promoter-truncation constructs were generated using the following 5Ј primer and the common 3Ј primer as mentioned above. The hINSIG-1 Ϫ861 (Ϫ861 to ϩ23) construct was generated using a 5Ј primer with nucleotides corresponding to Ϫ861 to Ϫ841 (5Ј-gagagctagcgcgtgtcatcctcaggaaat-3Ј). The hINSIG-1 Ϫ752 construct was generated using a 5Ј primer corresponding to nucleotides Ϫ752 to Ϫ738 (5Ј-gagagctagcgccacaaccccagg-3Ј). The hINSIG-1 Ϫ375 (Ϫ375 to ϩ23) construct was generated using a 5Ј primer corresponding to nucleotides Ϫ375 to Ϫ362 (5Ј-gagagctagccaagcccgccattg-3Ј).
Cell Culture, Transient Transfections, and Reporter Gene Assays-HepG2 cells were maintained in minimum Eagle's medium with 10% fetal bovine serum and were transiently transfected in 96-well plates using the FuGENE 6 transfection reagent (Roche Applied Science). Unless otherwise indicated, the cells were transfected with reporter plasmid (40 ng/well) and pCMX-hPPAR␥ (15 ng/well) with pCMX-RXR␣ (1.5 ng/well) and pCMV-␤-galactosidase (10 ng/well). Each well received a total of 75 ng of DNA. The cells were transfected in minimum Eagle's medium ϩ 10% charcoal-stripped fetal bovine serum for 12 h and treated with ligands for an additional 24 h before harvesting and assaying for luciferase and ␤-galactosidase activity. GW9662 (23), an irreversible PPAR␥ antagonist was purchased from VWR International.

INSIG-1 Is Induced by a PPAR␥ Agonist in Adipose
Tissue of db/db Mice-We originally identified INSIG-1 mRNA in microarray experiments as being increased compared with vehicle-treated animals in epididymal fat from db/db mice 6 h following a single dose of either the RXR agonist LG268 or the PPAR␥ agonist rosiglitazone (Incyte Genomics). The signal remained elevated 6 h after a third daily dose of either compound. These results were confirmed by real-time PCR using cDNA prepared from pooled RNAs from the same experiment (Fig.  1A). Rosiglitazone treatment was more effective at inducing Insig-1 mRNA compared with LG268 or the PPAR␣ agonist fenofibrate. Induction by rosiglitazone was confirmed in two independent in vivo experiments at time points up to 14 days (Fig. 1, B and C). Fig. 1C shows that rosiglitazone-treated animals had significantly higher levels of Insig-1 mRNA than vehicle-treated controls at days 3 and 8 (n ϭ 5, p Ͻ 0.05) in both epididymal (4-fold) and subcutaneous (9-fold) fat depots. For comparison, in subcutaneous fat, C/EBP␣ was induced 2.6-fold (p Ͻ 0.05), and fatty-acid-CoA ligase, long-chain 2 (FACL2) was induced 5.2-fold (p Ͻ 0.05) at day 8 (data not shown). Compared with vehicle-treated db/db mice, Insig-1 mRNA levels in the fat of lean littermates were 1.6-fold higher (p Ͻ 0.05) in subcutaneous fat and not significantly different in epididymal fat (data not shown). Insig-1 was also induced in liver following rosiglitazone treatment but to a lesser extent than in fat.
Insig-1 Is Induced during Adipogenesis in Vitro -Next we tested for induction of Insig-1 mRNA in cultured preadipocytes. C3H10T1/2 preadipocytes were plated and grown to confluence, at which time insulin was added in the presence or absence of rosiglitazone. Insulin alone did not induce Insig-1 at any time point. By 5 days, rosiglitazone in the presence of A, epididymal fat was removed from male db/db mice fasted for 6 h following 1 or 3 daily doses of 100 mg/kg/day fenofibrate (Feno), 30 mg/kg/day rosiglitazone (or BRL49653 (BRL)), 30 mg/kg/day LG268, or Vehicle. Relative levels of Insig-1 mRNA were determined by real-time PCR of cDNAs prepared from RNA pooled from each treatment group (n ϭ 10). Levels of Insig-1 mRNA were normalized to 36B4 and then expressed relative to vehicle-treated animals at the same time point. B, epididymal fat was removed from male db/db mice fasted for 6 h following a naïve or 14th daily dose of 300 mg/kg/day fenofibrate, 10 mg/kg/day rosiglitazone, or 30 mg/kg/day LG268. Relative levels of Insig-1 mRNA were determined by real-time PCR of cDNAs prepared from RNA pooled from each treatment group (n ϭ 7) as in A. C, epididymal and subcutaneous fat were removed from male db/db mice fasted for 6 h following 1, 3, or 8 daily doses of 30 mg/kg/day rosiglitazone. cDNAs were prepared from individual animal RNAs, and relative levels of Insig-1 mRNA were determined as in A. Data are presented as mean Ϯ S.E. and were analyzed by unpaired Student's t test. A p value of Ͻ0.05 was considered statistically significant. *, p Ͻ 0.05 (n ϭ 5).
insulin had induced Insig-1 mRNA nearly 10-fold (Fig. 2). We also measured transcripts for other markers of adipogenesis. At 24 h, PGAR/angptl4 was induced 14-fold by rosiglitazone. PGAR, SREBP-1a, and SREBP-1c transcripts were maximally induced by 3 days, whereas C/EBP␣ and Insig-1 continued to rise from day 3 to day 5. PPAR␥2 mRNA appears to approach a plateau at day 5. Although Insig-1 was originally identified as a target of insulin signaling in liver, it is apparently not regulated by the insulin treatment of preadipocytes. Insig-1 is induced by a PPAR␥ agonist relatively late in the adipogenesis program. Therefore, we wanted to determine whether Insig-1 is directly regulated by PPAR␥ or indirectly through other transcriptional activities of PPAR␥.
Regulation of INSIG-1 Promoter-Reporter Genes in Response to Co-expression of PPAR␥ -To determine the molecular mechanisms involved in the regulation of INSIG-1 gene expression, we transfected cells with a luciferase reporter gene under the control of the proximal promoter of the human INSIG-1 gene. Co-transfection of the hINSIG-1 reporter construct with PPAR␥1 or PPAR␥2 and RXR␣ into HepG2 cells stimulated the transcriptional activity ϳ3.5-fold in the absence of ligand (Fig.  3B). There was a further increase (2-fold) in expression upon the addition of the PPAR␥ ligand rosiglitazone (7-fold over basal). The agonist effect of rosiglitazone (but not LG268) was attenuated when the co-transfection was repeated with a dominant negative form of PPAR␥ (Fig. 3C). In addition, PPAR␥/ RXR␣ induction of the hINSIG-1 promoter in response to rosiglitazone is repressed in a dose-dependent manner in the presence of an irreversible PPAR␥ antagonist, GW9662 (Fig.  3D). This repression was also observed for a reporter gene under the control of three copies of the PPRE from the acyl-CoA oxidase gene (3x-AOx). The high basal activity of these reporter genes in the presence of PPAR␥ (but in the absence of exogenous ligand) likely results from an endogenous agonist.
We analyzed the human proximal promoter of INSIG-1 to determine the cis-acting elements necessary for the induction of INSIG-1 in response to PPAR␥ ligands. We identified two putative PPAR␥/RXR␣ binding sites, designated PPRE1 and PPRE2 (Fig. 3A). A series of truncation reporter constructs were designed that systematically deleted the PPREs. Deletion of PPRE2 had no effect on the transactivation of hINSIG-1 by PPAR␥ and RXR␣ (Fig. 3E). However, deletion of PPRE1 resulted in loss of activation by PPAR␥/RXR␣. Therefore, we conclude that PPRE1 (but not PPRE2) is necessary for PPAR␥ regulation of hINSIG-1.

Regulation of INSIG-1 Promoter-Reporter Genes in Response to Co-expression of SREBP-Based upon previous studies, IN-
SIG-1 mRNA levels are induced when nuclear SREBP levels increase (24 -26). We sought to determine whether the hIN-SIG-1 promoter-reporter construct could be directly transactivated by SREBP. pGL3-hINSIG-1 (Ϫ1233 to ϩ23) was transiently transfected into HepG2 cells with plasmids expressing transcriptionally active SREBP-1a, SREBP-2, or a dominant negative form of SREBP-1a (Fig. 4A). At equivalent plasmid concentrations, the hINSIG-1 construct was more highly activated by SREBP-1a than SREBP-2. The dominant negative SREBP-1a plasmid (Fig. 4A, DN SREBP-1a) is able to bind to an SRE (27). However, as a result of the deletion of 90 amino acids at the amino-terminal end, dominant negative SREBP-1a is unable to transactivate genes (27). The dominant negative SREBP-1a plasmid repressed the basal expression of reporter genes under the control of promoters derived from either the hINSIG-1 or the LDL receptor genes. Titration experiments in HepG2 cells using the hINSIG-1 promoter reporter (Ϫ1233 to ϩ23) and an increasing concentration of SREBP-1a or SREBP-2 ( Fig. 4B) demonstrate that SREBP-1a is a more potent activator of transcription of the hINSIG-1 promoter than SREBP-2 at all doses tested.
To further define the cis element of the INSIG-1 promoter necessary for SREBP-dependent induction, we utilized the hINSIG-1 truncation constructs. The data of Fig. 4C suggest that an SRE is located between SREBP-1a Ϫ375 and Ϫ752. Subsequent analysis revealed an SRE at position Ϫ389 to Ϫ380, relative to the transcriptional start site. This putative SRE is similar to the well characterized SRE in the human LDL receptor promoter differing by only two nucleotides (Fig. 4D).
PPAR␥/RXR␣ Binds Directly to the hINSIG-1 Promoter-Based upon the transient transfection studies shown in Fig.  3, we hypothesized that PPRE1 was necessary for activation in response to PPAR␥ ligands. We used recombinant human PPAR␥ and RXR␣ proteins in gel shift assays to determine whether the PPAR␥/RXR␣ heterodimer binds directly to PPRE1 and/or PPRE2. A protein⅐DNA complex was formed between PPAR␥/RXR␣ and PPRE1 (Fig. 5B). No such complex was observed with PPRE2 (Fig. 5B). The PPRE from the AOx promoter (Fig. 5, ACOA)  PPRE1 appears to be substantially weaker than to the AOx PPRE. DISCUSSION We demonstrated that Insig-1 mRNA levels are induced by PPAR␥ agonists in a diabetic mouse model (db/db). This induction was observed in white adipose tissue (epididymal fat and subcutaneous fat) by day 3 of agonist treatment, and activation was maintained for as long as 14 days (Fig. 1). Similarly, Way et al. (8) demonstrate that Insig-1 mRNA expression increases in adipose tissue (epididymal white adipose tissue and interscapular brown adipose tissue) of Zucker diabetic fatty rats treated with the PPAR␥ agonist GW1929 for 7 days. In this rat model, they observe a greater induction of INSIG-1 mRNA levels in brown adipose tissue compared with white adipose tissue (8). In support of our in vivo observations, we showed that Insig-1 is induced by PPAR␥ agonists in C3H10T1/2 cells following 3 days of treatment (Fig. 2). In agreement with our  (Ϫ1233 to ϩ23), hINSIG-1 Ϫ861 (Ϫ861 to ϩ23), hINSIG-1 Ϫ752 (Ϫ752 to ϩ23), hINSIG-1 Ϫ375 (Ϫ375 to ϩ23) (20 ng of each reporter). The cells were transfected with or without PPAR␥1 (6 ng/well) and RXR␣ (0.75 ng/well) and a plasmid encoding ␤-galactosidase. The transfected cells were treated with either Me 2 SO or the PPAR␥1 agonist rosiglitazone (1 M). The cells were lysed 48 h later, and the luciferase activities were determined and normalized for ␤-galactosidase activity. Each construct is divided by vehicle control and reported as the fold induction over the vector control. All transfections were done in triplicate in three separate experiments. studies, Li et al. (15) establish, using 3T3-L1 cells, that Insig-1 expression is induced by day 9 of adipocyte differentiation. We demonstrated that in the presence of rosiglitazone, this induction occurs much earlier (day 3) and is more robust when compared with insulin-treated cells (10-fold, Fig. 2). Fig. 2 illustrates that prior to the activation of Insig-1 (observed at day 5) both PPAR␥1 and SREBP-1c are maximally induced as early as 72 h. This time course, when taken together with our transfection data (which demonstrated that both transcription factors directly activate the INSIG-1 promoter), suggests that they are involved in the activation of Insig-1 transcription during adipocyte differentiation.
We established that PPAR␥ directly transactivates the human INSIG-1 promoter (Figs. 3 and 5). A dominant negative form of PPAR␥ was completely inactive, and the INSIG-1 transactivation was repressed when a PPAR␥ antagonist was added (Fig. 3D). These data indicate that activation of the INSIG-1 promoter is mediated by the ligand-gated transactivation function (AF-2) in the ligand-binding domain of PPAR␥. Truncation reporter constructs localized the region through which PPAR␥/RXR␣ mediated this induction to ϳ90 base pairs in the proximal promoter (Fig. 3E). Analysis of this region revealed a PPRE, to which the PPAR␥/RXR␣ heterodimer binds directly in vitro (Fig. 5B).
INSIG-1 has recently been identified as a modulator of SREBP activity (16). INSIG-1 appears to tether the SCAP⅐SREBP complex in the endoplasmic reticulum in the presence of sterols. Interestingly, Insig-1 mRNA is induced in the livers of nSREBP-1a and nSREBP-2 transgenic animals when compared with wild-type littermate controls (26). In addition, Janowski (24) determined that INSIG-1 mRNA expression is inhibited by oxysterols, which also inhibit the expression of SREBP. When the oxysterol-mediated effect is reversed (using a hypocholesterolemic agent), SREBP processing increases, and the INSIG-1 gene is induced (24). Utilizing the hINSIG-1 promoter-reporter construct (Ϫ1233 to ϩ23), we were able to determine that SREBP-1a potently transactivates the promoter (Fig. 4). Through the use of truncation constructs, the sterol response element was localized to a 400-base pair region in the proximal promoter of INSIG-1. A putative SRE identified within the region is similar to that identified for the LDL receptor, differing by only two nucleotides.
The current studies, as well as previous work, have demonstrated that there are at least three mechanisms by which INSIG-1 expression is altered. Insig-1 mRNA levels can be induced in the liver by increasing insulin levels (12,28), in adipose and cultured cells by activated PPAR␥/RXR␣ (Figs. 1-3 and Ref. 8), as well as by SREBPs in the cultured liver cells (Fig. 4 and Refs. 24 and 26). Although we defined the PPRE and SRE, the region of the INSIG-1 promoter necessary for insulin responsiveness has not been examined. Future studies using FIG. 4. Regulation of hINSIG-1 promoter-reporter genes by co-expression of SREBP-1a or SREBP-2. A, the hINSIG-1 promoterreporter construct (pGL3-hINSIG-1 Ϫ1233 to ϩ23), the LDL receptor reporter construct, or pGL3 were transfected into HepG2 cells in triplicate along with a plasmid encoding ␤-galactosidase. The cells were co-transfected with plasmids encoding SREBP-1a, SREBP-2, or dominant negative SREBP-1a. After 24 h, the cells were lysed, and the normalized luciferase values were determined. All transfections are representative of three separate experiments. B, HepG2 cells in a 96-well dish were co-transfected with hINSIG-1 (Ϫ1233 to ϩ23) and increasing concentrations of the nuclear form of SREBP-1a and SREBP-2 as indicated. The cells were lysed 24 h later and were normalized for luciferase activity. C, HepG2 cells were transiently transfected with hINSIG-1 Ϫ1233 (Ϫ1233 to ϩ23), hINSIG-1 Ϫ861 (Ϫ861 to ϩ23), hINSIG-1 Ϫ752 (Ϫ752 to ϩ23), hINSIG-1 Ϫ375 (Ϫ375 to ϩ23), and the LDL receptor. The cells were transfected for 24 h with or without SREBP-1a and a plasmid encoding ␤-galactosidase. On the following day, the cells were lysed, and the luciferase activities were determined and normalized for ␤-galactosidase activity. RLUs, relative light units. D, schematic illustration of the human LDLr and human INSIG-1 proximal promoters. the truncation-reporter constructs will be useful in analyzing the mechanism by which insulin confers this regulation within the context of the liver. One possibility is that insulin mediates this effect on INSIG-1 via SREBP. Using 3T3L1 cells, Le Lay et al. (29) identifies a subset of genes (such as fatty-acid synthase and the LDL receptor) within the adipocyte that are regulated by insulin via SREBP.
What becomes apparent is a complex network of coordinated regulatory events in which SREBP, PPAR␥, and INSIG-1 modify the expression and/or activity of one another. We demonstrated that transcriptionally active SREBP and ligand-activated PPAR␥/RXR␣ increase INSIG-1 at the mRNA level (Fig. 6). In turn, INSIG-1 inhibits the processing of SREBPs, thus providing a feedback mechanism by which INSIG-1 regulates lipid homeostasis as well as indirectly modifying its own expression. Li et al. (15) describe the reduction of both SREBP-1c and PPAR␥2 transcripts in 3T3L1 adipocytes, which overexpressed INSIG-1. Similarly, we observed reciprocal mRNA expression between Insig-1 and SREBP-1c in C3H10T1/2 cells (Fig. 2). A more rigorous time course might also reveal further changes in PPAR␥2 levels. Finally, within the adipocyte, SREBP in the presence of C/EBP is thought to generate PPAR␥ agonists (29,30). Therefore, what we describe is a mechanism that links the insulin sensitization observed with the PPAR␥ ligands to lipid metabolism that is observed as a result of altering the processing of SREBP.
These observations suggest a complex network with multiple checkpoints to couple insulin signaling with lipid homeostasis. Although insulin signaling via SREBP and Insig-1 is compromised in diabetic animal models, treatment with rosiglitazone activates Insig-1 via PPAR␥. This convergence of the PPAR␥ pathway reestablishes lipid homeostasis and insulin sensitivity within the adipose tissue. In future studies, it will be important to determine whether the observations made in white adipose tissue are also true in brown adipose tissue. In addition, it will be of interest to understand the impact of PPAR␣ and PPAR␥ ligands on the hepatic expression of Insig-1 and its role in modifying the processing of SREBPs. A thorough understanding of the interplay between INSIG-1, SREBPs, and PPAR␥ will be valuable in understanding the full spectrum of thiazolidinedione-induced antidiabetic activities. This will facilitate attempts to design improved insulin sensitizers without the negative side affects (such as edema and weight gain) that often affect individuals treated with thiazolidinediones.  -1 (marked by a thick vertical arrow). In addition, nSREBP and C/EBP␣ activities lead to the generation of fatty acids, which serve as ligands for PPAR␥. Thus, either thiazolidinediones (TZD) or fatty acids activate PPAR␥, which in turn increases the mRNA expression levels of INSIG-1 (thick horizontal arrow). INSIG-1 protein ultimately inhibits the processing of SREBP by interacting with the SCAP⅐SREBP FL complex (depicted by the dashed line), generating a feedback mechanism to control lipogenesis.