Insulin and Sterol-regulatory Element-binding Protein-1c (SREBP-1C) Regulation of Gene Expression in 3T3-L1 Adipocytes

We evaluated the hypothesis of sterol-regulatory element-binding protein (SREBP)-1c being a general mediator of the transcriptional effects of insulin, with a focus on adipocytes, in which insulin profoundly influences specific gene expression. Using real time quantitative reverse transcriptase-PCR to monitor changes in the expression of about 50 genes that cover a wide range of adipocyte functions, we have compared the impact of insulin treatment with that of adenoviral overexpression of either dominant positive or dominant negative SREBP-1c mutants in 3T3-L1 adipocytes. As expected, insulin up-regulated, dominant positive stimulated, and dominant negative decreased previously characterized direct SREBP targets (FAS, SCD-1, and low density lipoprotein receptor). We also identified three novel SREBP-1c transcriptional targets in adipocytes, which were confirmed by run-on assays: plasminogen activator inhibitor 1, CCAAT/enhancer-binding protein δ (C/EBPδ), and C/EBPβ. Because most insulin-regulated genes were also modulated by SREBP-1c mutants, our data establish that 1) SREBP-1c is an important mediator of insulin transcriptional effects in adipocytes, and 2) C/EBPβ is under the direct control of SREBP-1c, as demonstrated by the ability of SREBP-1c to activate the transcription from C/EBPβ promoter through canonical SREBP binding sites. Thus, some of the effects of insulin and/or SREBP-1c in mature fat cells might require C/EBPβ or C/EBPδ as transcriptional relays.

Insulin is the main anabolic hormone in mammals and exerts its effects in liver, adipose tissue, and skeletal and cardiac muscle via the insulin receptor (for a review, see Ref. 1). The cellular mechanism underlying its action on carbohydrate, lipid, and protein metabolism has been the center of major interest for many years. Active research has led to the identification of the major steps of the insulin signal transduction pathway. These include a family of soluble scaffolding mole-cules, known as insulin receptor substrates, which initiate downstream signaling cascades involving the phosphatidylinositol 3-kinase/Akt pathway and the mitogen-activated protein kinase pathway (for reviews, see Refs. 1 and 2). In this cascade, rapid changes in the state of protein phosphorylation ultimately mediate many important actions of insulin (e.g. glucose transport, glycogen synthesis, lipogenesis, and antilipolysis).
It is also well known that, alongside these rapid nongenomic effects, important changes in gene expression play critical roles in insulin action in insulin-sensitive tissues (3). The transcriptional effects of insulin and the mechanisms by which insulin can relay signal to the nucleus have remained largely unknown until recently. As described (4), new light was shed by the identification of SREBP-1c 1 as a transcription factor capable of mediating some of the effects of the hormone on previously identified insulin target genes. Indeed, SREBP-1c was shown not only to regulate the expression of key genes of glucose, fatty acid, and triglyceride metabolism in fibroblasts, adipocytes, hepatocytes, and the livers of transgenic mice (5-7) but also to be able to substitute to insulin in inducing transcription of known insulin target genes like glucokinase or FAS in hepatocytes (8). SREBP-1c is particularly abundant in the adipose tissue and the liver, both of which are insulin-sensitive and display quite a restricted expression pattern compared with the ubiquitously expressed SREBP-2, the other SREBP isoform that is encoded by a separate gene (9). In agreement with their distinct expression pattern and regulation, SREBP-1c and SREBP-2 can also be distinguished in vivo by their ability to target different genes. Indeed, SREBP-1c and SREBP-2 assume different functions, SREBP-2 being more selective for activating genes involved in cholesterol homeostasis (reviewed in Ref. 10), whereas SREBP-1c actions are focused on lipid synthesis and glucose metabolism. From these studies, SREBP-1c thus appears as a strong candidate to be a general mediator of the metabolic actions of insulin via the regulation of gene expression.
The aim of the present study was to document further this hypothesis with a focus on adipocytes, in which specific gene expression is profoundly influenced by insulin. In the context of the adipose cell, several transcription factors that play interconnected roles ultimately determine the fully differentiated adipocyte gene expression profile. Among these factors is SREBP-1c, known also as ADD-1 (for adipocyte determination and differentiation factor-1 (11)), a member of the basic helixloop-helix-leucine zipper family of transcription factors. Other important adipocyte transcriptional regulators include the fatty acid derivative-activated nuclear receptor zinc finger peroxisome proliferator-activated receptor ␥ (PPAR␥) and several members of the basic leucine zipper family of CAAT/enhancer binding proteins (C/EBPs) (reviewed in Ref. 12). In particular, C/EBP␤ and C/EBP␦, when induced by appropriate stimuli, can initiate a transcriptional cascade that culminates in the induction of PPAR␥ and C/EBP␣ and the activation of the adipogenic program.
In this study, we have used mature 3T3-L1 adipocytes to investigate the impact of the overexpression of mutant SREBP-1c isoforms, either dominant positive or dominant negative, on a panel of 50 adipocyte-specific genes. The latter were selected to cover a number of aspects of key fat cell functions such as lipid storage, lipolysis, glucose metabolism, energy expenditure, adipocyte-gene transcription factors, and adipocyte-derived secreted products. Our results show that genes that were up-regulated by the dominant positive and downregulated by dominant negative SREBP-1c forms were also up-regulated by insulin, confirming that SREBP-1c is a major factor underlying the transcriptional effect of insulin. Moreover, we found out that SREBP-1c specifically mediated insulin action on some adipocyte genes, such as PAI-1, and the ␤ and ␦ C/EBP isoforms, which were known as insulin-sensitive but previously unrecognized as SREBP-1c targets. Finally, we provide evidence that SREBP-1c can directly transactivate the C/EBP␤ promoter through canonical SREBP binding sites. Moreover, those sites map the insulin response region of the C/EBP␤ promoter. Thus, this study demonstrates the existence of an insulin-SREBP-1c-C/EBP␤ axis in adipocytes and suggests that a transcriptional cascade might be initiated by SREBP-1c to mediate insulin effects on fully differentiated adipocyte gene regulation.

MATERIALS AND METHODS
Preparation of Recombinant Adenoviruses-The adenovirus vector containing the transcriptionally active dominant positive (DP) aminoterminal fragment (amino acids 1-403) of rat SREBP-1c, Ad.SREBP-1c DP, was constructed as previously described (8) with homologous recombination in BJ5183 bacteria using the shuttle vector pAdTrack-CMV containing the green fluorescent protein (GFP) (13). The recombinant adenovirus containing the dominant negative form of rat SREBP-1c, Ad.SREBP-1c DN, was described elsewhere (14). Both Ad.SREBP-1c DP and Ad.SREBP-1c DN were under control of a cytomegalovirus promoter. The adenovirus vector containing the major late promoter with no exogenous gene (Ad.null) was used as control. The adenoviral vectors were propagated in the HEK 293 cell line, purified by cesium chloride density centrifugation, and stored at Ϫ80°C until use. The efficiency of infection in 3T3-L1 adipocytes was assessed by visualizing GFP expression using a fluorescence microscope (Eclipse E800, Nikon). Measurements of SREBP target gene expression were also performed using various MOI (from 10 to 500) and various postinfection times from 24 to 72 h. Experiments were performed 4 -6 times. Various tested conditions were Ad.SREBP-1c DP with no insulin; Ad.SREBP-1c DN with insulin (100 nM); Ad.null with insulin (100 nM); and Ad.null with no insulin.
RNA Preparation and Real Time Quantitative RT-PCR-Total RNA was prepared as described (15). cDNA was synthesized from 5 g of total RNA in 20 l using random hexamers and murine Moloney leukemia virus reverse transcriptase (Invitrogen). The design of primers was done using either Primer Express (Applied Biosystems) or Oligo (MedProbe, Olso, Norway) software. Real time quantitative RT-PCR analyses for the genes described in Table I were performed starting with 50 ng of reverse transcribed total RNA (diluted in 5 l of 1ϫ Sybr Green buffer), with a 200 nM concentration of both sense and antisense primers (Genset) in a final volume of 25 l using the Sybr Green PCR core reagents in an ABI PRISM 7700 Sequence Detection System instrument (Applied Biosystems). Fluorescence is generated after laser excitation by bound Sybr Green to double-stranded DNA. Because we used Sybr Green in measurements of amplification-associated fluorescence for real time quantitative RT-PCR, it was important to verify that generated fluorescence was not overestimated by contaminations resulting from residual genomic DNA amplification (using controls without reverse transcriptase) and/or from primer dimers formation (controls with no DNA template nor reverse transcriptase). RT-PCR products were also analyzed on ethidium bromide-stained agarose to ensure that a single amplicon of the expected size was indeed obtained. To measure PCR efficiency, serial dilutions of reverse transcribed RNA (0.1 pg to 200 ng) were amplified, and a line was obtained by plotting cycle threshold (C T ) values as a function of starting reverse transcribed RNA, the slope of which was used for efficiency calculation using the formula E ϭ 10 ͉(1/slope)͉ -1 (16). Ribosomal 18 S RNA amplifications were used to account for variability in the initial quantities of cDNA. The relative quantitation for any given gene, expressed as -fold variation over control (untreated cells), was calculated after determination of the difference between C T of the given gene A and that of the calibrator gene B (GAPDH) in treated cells (⌬C T1 ϭ C T1A Ϫ C TB ) and control cells (⌬C T0 ϭ C T0A Ϫ C TB ) using the 2 Ϫ ⌬⌬C T(1-0) formula (16). GAPDH expression of a control cDNA was used as interplate calibrator. Variation over controls was determined using the above-mentioned formula as follows. The effect of insulin was calculated by comparing mean C T values obtained in the Ad.null with insulin condition and that obtained in the Ad.null with no insulin condition; the effect of Ad.SREBP-1c DP by comparing Ad.SREBP-1c DP with no insulin and the Ad.null with no insulin conditions; and the effect of Ad.SREBP-1c DN by comparing Ad.SREBP-1c DN with insulin and the Ad.null with insulin conditions. C T values are means of triplicate measurements. Experiments were repeated 4 -6 times. All primers are presented in Table I. In a given cDNA population, relative expression level between genes could be calculated based on individual C T , provided that PCR efficiencies were close to 1. The latter were calculated according to Ref. 16 and were 1.1 ϩ 0.07 (mean Ϯ S.E., n ϭ 21), indicating an approximate doubling of DNA at each PCR cycle, as theoretically expected. The percentage of relative expression between several genes of a given family (e.g. for the three C/EBP isoforms) was calculated as follows; mean C T of C/EBP␣, C/EBP␤, and C/EBP␦ in the absence of insulin were 24.9, 22.24, and 30.31, respectively. Using the 2 Ϫ ⌬C T formula, these could be expressed as two equations (C/EBP␤ ϭ 6.32 ϫ C/EBP␣ and C/EBP␤ ϭ 268 ϫ C/EBP␦) plus another equation as C/EBP␣ ϩ C/EBP␤ ϩ C/EBP␦ ϭ 100. It could thus be calculated that the percentage expression of the C/EBP␣, C/EBP␤, and C/EBP␦ isoforms was 13.6, 86.1, and 0.3%, respectively.
Nuclear Run-on Transcription Analysis-Differentiated 3T3-L1 cells were treated with insulin or were infected with the adenovirus encoding the DP SREBP-1c mutant. After 24 h, nuclei were prepared as previously described (17) and were incubated with [␣-32 P]CTP (3000 Ci/ mmol) for 45 min at 32°C.
Incubations were terminated by the addition of RNase-free DNase and proteinase K, and labeled RNA was extracted by phenol/chloroform. Labeled transcripts were hybridized for 72 h with plasmid cDNA immobilized on nylon membranes. Blots were washed to high stringency, and hybridized RNA was quantified with an optical scanner (Storm 860; Amersham Biosciences).
Promoter Analysis and Transfections-A 1.4-kb promoter fragment of the rat C/EBP␤ gene, cloned in front of the luciferase reporter in the p19-Luc vector, has been described elsewhere (18). From this construct, a series of 5Ј deletions was derived, encompassing regions from Ϫ441 to ϩ16, Ϫ183 to ϩ16, and Ϫ136 to ϩ16 relative to the transcription start site. Point mutations on sterol-regulatory element (SRE) sites in the 1.4-kb promoter fragment were introduced using the QuikChange multisite-directed mutagenesis kit (Promega) as recommended by the manufacturer. The sequences of the mutagenic primers were 5Ј-GGG-CGGAGGTCGTACCAGCTCAGCAfor the SRE1 site located at Ϫ1124 and 5Ј-AAGGTTGAGCAACGTACCACCAGCTTGCC for the SRE2 at Ϫ1064. In both cases, the disruption of the SRE motif was performed by replacing the ACC triplet by GTA as underlined in the sequences. Growing 3T3L1 preadipocytes in 60-mm dishes were transfected using the calcium phosphate precipitation method, with a mixture of plasmid DNA containing 1 g of a promoter luciferase construct, 1 g of Rous sarcoma virus-␤-galactosidase as an internal standard, and 50 ng of pSV Sport1-ADD1 expression vector encoding an active form of the rat SREBP-1c transcription factor (19). The total amount of DNA was kept constant in each experiment by adding empty pSV Sport1 when necessary. Reporter gene activities were assayed 24 h after transfection, and luciferase data were normalized to galactosidase.
Differentiated 3T3-L1 cells were placed in serum-free medium containing 2% bovine serum albumin and no insulin for 24 h and transfected by electroporation as described previously for mature adipocytes (20). Briefly, 1-2 ϫ 10 6 cells in 200 l were shocked electrically in the presence of 20 g of promoter luciferase constructs and 1 g of Rous sarcoma virus-chloramphenicol acetyltransferase internal control. Cells were then replated in Dulbecco's modified Eagle's medium containing 10% fetal calf serum in the presence or absence of 1 g/ml insulin. Reporter gene activities were measured after 24 h.
Western Blot Analysis-Nuclear extracts were obtained from differentiated 3T3-L1 adipocytes as described previously (20) and used for Western blotting with a commercially available antibody against C/EBP␤ (Santa Cruz Biotechnology, Inc., Santa Cruz, CA).
Statistical Analysis-For statistical analyses of real time RT-PCR experiments, results for a given gene were expressed as differences from the mean C T value obtained in the Ad.Null with no insulin condition. Statistical significance was assessed by analysis of variance followed by Newman-Keuls comparison tests (Statistica, StatSoft Inc.). In transfection experiments, statistical differences were assessed by Student's t test. A p value of Ͻ0.05 was considered as the threshold of statistical significance.

RESULTS AND DISCUSSION
Insulin Selectively Stimulated the SREBP-1c Isoform in Adipocytes-As a first step to study the impact of changes in SREBP-1c content in 3T3-L1 adipocytes and its potential correlation with insulin-induced gene expression profile, we assessed the ability of insulin to modulate endogenous SREBPs expression in these cells. In agreement with initial studies (21), we observed a stimulatory effect of insulin on SREBP-1c mRNA, increasing its levels by 3-fold (Fig. 1A). The effect of insulin was restricted to the SREBP-1c isoform, with no change in SREBP-1a or SREBP-2. Similar results were obtained in livers of streptozotocin-induced diabetic rats (7) and in cultured hepatocytes (22). Thus, insulin specifically targets SREBP-1c in adipocytes. It remains to be determined whether the isoform-specific effect of the hormone equally occurs in adipose in vivo and in other insulin-sensitive tissues (i.e. skeletal, cardiac muscle, and brown fat).
SREBP-1c Transcriptional Activity Can Be Efficiently Manipulated in 3T3-L1 Adipocytes-Having shown that insulin stimulated SREBP-1c, we examined the conditions to achieve optimal expression of SREBP-1c mutants (DP or DN forms) following adenoviral infection of differentiated 3T3-L1 adipocytes. First, using Ad.SREBP-1c DP, which co-expresses GFP, optimal conditions for transduction of 3T3-L1 adipocytes with recombinant adenoviruses were assessed. After 24 h, nearly 100% of GFP-expressing adipocytes was achieved with an MOI of 500 (Fig. 1B, bottom panel). Second, the steady state levels of SREBP-1 mRNA were monitored by real time RT-PCR using primers designed to differentially target endogenous SREBP-1c, endogenous SREBP-1a, or total SREBP-1 expression (in-cluding endogenous as well as adenovirus-mediated DN or DP mutant forms of SREBP-1c; see Table I for primer sequences). Ad.Null was used as control and exerted no effects on gene expression, whatever the MOI (not shown). Fig. 1B shows that the infection of adipocytes with increasing titers of adenoviruses encoding mutant forms of SREBP-1c (either dominant negative or positive) produced, as expected, a dose-dependent increase in total SREBP-1 mRNA expression, demonstrating significant transgene expression in adipocytes. Because we observed that the endogenous expressions of SREBP-1a and SREBP-1c were not altered by Ad.SREBP-1c DP or DN (Fig.  1B), the increase in total SREBP-1 observed following adenovirus infection was solely accounted for by an increase in the expression of the transgene. As shown in Fig. 1B, total SREBP-1 expression increased by ϳ15-fold in cells infected with the Ad.SREBP-1c DP (MOI of 500) and was stimulated to a similar extent (ϳ10-fold) using the same titer of Ad. SREBP-1c DN. This was confirmed by Western blot analysis with an anti-SREBP-1 antibody that showed a huge increase in SREBP transgene protein content in nuclear extracts prepared from cells infected with Ad.SREBP-1c DP (data not shown). To ensure that SREBP-mediated transcriptional activity was significantly altered in adipocytes infected with the adenoviruses encoding the DP or DN SREBP-1c mutants, we measured the steady state levels of known SREBP target genes such as FAS (5,14,23), SCD-1 (24), and LDL receptor (25). Fig. 1C shows that increasing titers of Ad.SREBP-1c DP mutant dose-dependently stimulated the expression of FAS, SCD-1, and LDL receptor genes (left panel). The induction of FAS, SCD-1, and LDL receptor gene expression started at 10 -100 plaque-forming units/cell, and a plateau was reached after 250, the 500 plaque-forming units/cell condition being optimal. The mRNAs encoding FAS and SCD-1 were increased with higher efficiencies (up to 10-fold) than that of the LDL receptor, which was stimulated only 3-fold. This agreed well with the ability of SREBP-1c to stimulate lipogenesis in preference to cholesterol uptake in vivo (26,27) (reviewed in Ref. 10). In reciprocal experiments, cells were infected with increasing titers of Ad. SREBP-1c DN mutant (Fig. 1C, right panel). We observed, as expected, a gradual decline in the steady state levels of FAS, SCD-1, and LDL receptor mRNAs. The observed changes in DN-expressing cells were of lesser magnitude than those in cells expressing the DP form. Since the dominant negative mutant inhibits SREBP-1c transcriptional activity by titrating endogenous SREBP-1c into inactive heterodimers (5), it is possible that Ad DN expression might not reach sufficient levels to completely inhibit endogenous SREBP-1c. Alternatively, because transcriptional activation of these genes requires in addition to SREBP other transcription factors such as NFY or Sp1 (24,28), it remained plausible that the presence of these untitrated factors or that of other co-activators allows sufficient residual transcriptional activity, thus obviating total inhibition of transcription. Taken together, all these results establish that SREBP-1c transcriptional activity can be efficiently manipulated in 3T3-L1 cells by means of adenovirus-mediated overexpression of dominant positive or negative SREBP-1c mutants.
SREBP-1c Mimics Most Insulin-induced Changes in Gene Expression 3T3-L1 Cells-The 3T3-L1 differentiated adipocyte cell system was used to compare the effects of insulin with that of SREBP-1c manipulations on the expression of various adipocyte genes. Specific primers were designed for real time fluorescent RT-PCR analyses of a panel of genes that covers a wide range of fat cell functions, i.e. lipid storage or lipolysis, transcriptional regulators of adipocyte differentiation, energy expenditure, and adipocyte-derived secreted factors (primers used are displayed in Table I). Table II shows the effects of FIG. 1. A, endogenous expression of SREBPs mRNAs in 3T3-L1 adipocytes. Fully differentiated 3T3-L1 cells (day 10 postconfluence) were shifted to a serum-free medium containing 5% free fatty acid bovine serum albumin, 25 mM glucose in the absence or presence of insulin for 48 h. Then the steady state mRNA levels of three SREBP isoforms were measured using real-time RT-PCR as described under "Materials and Methods." The primers used to distinguish SREBP-1c, SREBP-1a, and SREBP-2 mRNA are shown in Table I. B, efficiency of SREBPs transgene expression in 3T3-L1 adipocytes. Fully differentiated cells were infected with either Ad.SREBP-1c DP (open symbols) or Ad.SREBP-1c DN (black symbols) at various MOI from 0 to 500. After 24 h, RNA was prepared, and quantitative RT-PCRs were performed with primers designed to specifically target endogenous SREBP-1c expression (circles) and endogenous SREBP-1a expression (squares) or total SREBP-1 (triangles). After normalization to 18 S mRNA, values were expressed relative to that measured in control noninfected cells. Transduction efficiency was also evaluated visually insulin and that of overexpressed SREBP-1c DP and DN mutant forms on steady state levels of about 50 adipocyte mRNA species. Over the 47 genes presented here, we observed that the expression of 20 was significantly affected by insulin (see Table  II, first and third gene groups) confirming that insulin profoundly influenced the tone of adipose-specific gene expression. On the other hand, a total of 27 genes were found to be modulated by SREBP-1c mutants (see Table II; first through third gene groups). Importantly, among the 20 insulin-regulated transcripts presented in Table II, all but one (Akt/PKB) were sensitive to SREBP mutant overexpression. Noticeably, SREBP-1c, which is induced by insulin (Fig. 1A), is not subjected to autoregulation (Fig. 1B). Eight genes remained (see Table II, second gene group) that were sensitive to SREBP-1c but unaffected by insulin. We suppose that such a pattern can be explained by interactions of mutants with other transcription factors of the helix-loop-helix family or cofactors that might be important for basal expression of these genes. Collectively, these data indicate that most insulin-regulated genes in adipocytes can also be modulated by SREBP-1c and establish that SREBP-1c is an important mediator of insulin action in adipose tissue.
Identification of Novel Transcriptional SREBP-1c Target Genes-Among the 19 genes regulated by insulin and either dominant positive or dominant negative SREBP mutants, only 10 genes exhibited coordinate increased expression with both insulin treatment and overexpression of the SREBP-1c DP form and a reciprocal decrease in cells overexpressing the SREBP-1c DN mutant (Table II, first gene group). Because of such a coordinately regulated pattern of expression, we considered these genes as being strong candidates for insulin regulation through SREBP-1c (Fig. 2). These genes encode FAS, LDL receptor, HMG-CoA reductase, high density lipoprotein receptor SR-BI, SCD-1, GAPDH, GLUT1, PAI-1, and the ␤ and ␦ isoforms of C/EBP. The transcriptional control of FAS (29), GAPDH (30), GLUT1 (31), SCD-1 (32), PAI-1 (33), and C/EBP␤ and -␦ (34) by insulin has been demonstrated in adipocytes or in other cell types for the LDL receptor (35) and SR-BI (36).
following infection by Ad.SREBP-1c DP, which co-expresses GFP, using a fluorescence microscope (bottom images). C, effects of the dominant positive or dominant negative SREBP-1c mutants on the expression of SREBP target genes. 3T3-L1 adipocytes were infected with various MOI from 0 to 500 using either Ad.SREBP-1c DP (left panel) or Ad.SREBP-1c DN (right panel) and studied after 48 h (the optimal postinfection time as judged by modulation of gene expression, not shown). Steady-state levels of mRNA encoding FAS, SCD-1, and LDL receptor were quantified by real time RT-PCR, and -fold variations over control are presented (variations between ϩ1 and Ϫ1 corresponded to no changes). Ad.Null was used in controls. No difference between Ad.Null-treated (whatever the MOI (0 -500)) and noninfected cells was observed (not shown).

TABLE I Primer sequences of the selected genes involved in key pathways of adipose metabolism
The abbreviations of the genes, full name, accession number or locus, corresponding primer numbers, and 5Ј to 3Ј nucleotide sequences of the sense and antisense primers are presented.
Functional insulin-responsive sequences have also been described in the promoter region of GAPDH (37) and GLUT1 (38). Five of the insulin-regulated genes in Fig. 2, namely FAS (5,14,22), SCD-1 (24), LDL receptor (25), HMG-CoA reductase (39), and SR-BI (40), have been previously characterized as direct SREBP targets, underlying the importance of SREBP-1c for insulin effects on gene expression. It is worth mentioning, however, that insulin-regulated expression of SR-BI and LDL receptor was demonstrated in nonadipose cell types, thus raising the question of its physiological significance in fat cells. In particular, whether insulin notably influences the metabolism of cholesterol-rich lipoproteins in adipose tissue is an issue that has to be clarified. Interestingly, the present data point out that five other adipocyte genes were regulated in parallel by insulin and SREBP-1c (i.e. C/EBP␤, C/EBP␦, GLUT1, PAI-1, and GAPDH), suggesting that they might be new SREBP-1c targets. To assess the existence of a transcriptional control by SREBP-1c, we performed run-on transcription experiments on differentiated fat cells infected with the Ad.SREBP-1c DP or treated by insulin. Results in Table III clearly establish a positive effect of SREBP-1c on the transcription of FAS, a typical SREBP-1c target gene, but also on the C/EBP␤, C/EBP␦, and PAI-1 genes. It is noteworthy that the magnitude of the stimulation of the transcription of these genes by insulin and SREBP-1c were in a very similar range. This demonstrated that C/EBP␤, C/EBP␦, and PAI-1 are transcriptionally controlled by SREBP-1c in adipocytes. By contrast, although insulin stimulated the transcription of GAPDH and GLUT1, we   could not detect any direct effect of SREBP-1c overexpression on the transcription of these two genes. Noticeably, the two insulin-responsive elements of GAPDH localized between bases Ϫ480 and Ϫ269 (37) or that reported for GLUT1 at Ϫ2.7 kb within intron 2 (38) do not match with putative SREBP binding sequences that could be found using the TransFac data base (55). Thus, these results demonstrate that C/EBP␤, C/EBP␦, and PAI-1 are new transcriptional targets of SREBP-1c in the adipocytes. Some Insulin Genic Actions May Not Be Mediated through SREBP-1c-Some other genes (Table II, third gene group) were regulated by insulin but not in a coordinate manner by the SREBP-1c DP or DN mutants. This is the case for Akt/PKB and ␤3-AR, which were down-regulated by insulin, and for GAPDH, GLUT4, Id1, Id2, leptin, perilipin, PPAR␥, and resistin, which were up-regulated. Insulin-regulated gene expression in adipocytes has already been reported for GAPDH (41), leptin (21), and PPAR␥ (42), but not for Id1, Id2, Akt/PKB, and perilipin.
We are aware that the positive effect of insulin on GLUT4 observed in our study is not in accordance with previous reports showing an insulin-mediated decrease in GLUT4 mRNA (43). Given the importance of the C/EBP family of transcription factors in the regulation of GLUT4 expression (44), it has been postulated that the negative effect of insulin on GLUT4 in 3T3-L1 cells resulted from a decrease in C/EBP␣. In our study, for some unknown reasons, C/EBP␣ expression is not modified by insulin, which could explain why GLUT4 is not decreased by insulin. Then the observed increase in GLUT4 mRNA expression in the presence of insulin might be secondary to a SREBP-1c-mediated increase in C/EBP␤ and C/EBP␦. The group of genes in Table II, third gene group, which are differentially regulated by insulin and SREBP mutants, would suggest at first glance that not all insulin actions on gene expression might be mediated by SREBPs. However, such a conclusion cannot be readily drawn in the absence of experimental evidence demonstrating that insulin-regulated expression of these genes occurs at the level of transcription. Indeed, insulin was shown to stabilize GAPDH mRNA post-transcriptionally in the Ob17 adipose cell line (45). Thus, it is likely that insulin regulation of some of these genes results from other pleiotropric effects of the hormone and may be unrelated to a transcriptional regulation. It is noteworthy that this group includes the recently discovered resistin, which encodes a new adipocytederived secreted product (46). Very little is known yet of the regulation of the expression of the resistin gene. We show here a positive control of resistin mRNA by insulin. In agreement, a recent report has described a marked decrease of resistin mRNA in the adipose tissue of streptozotocin-induced diabetic rats, which was restored upon insulin administration (47). However, it remains to be established whether insulin action on the resistin gene is exerted transcriptionally. In the adipocytes that overexpress the SREBP-1c dominant negative SREBP-1c mutant, a significant decrease in resistin expression was noted. This might suggest that SREBP-1c is required for sustained expression of the resistin gene in the fat cell. However, the lack of stimulation observed with the dominant positive questions about the ability of SREBP-1c to mimic insulin regulation, raises the possibility that, for a limited number of genes, the effects of insulin might not be achieved through SREBP-1c only. Another example is the ␤ 3 -adrenoreceptor gene, whose transcription is repressed by insulin (48). Our study shows that indeed, the ␤ 3 -adrenoreceptor mRNA was down-regulated in cells treated by insulin as well as in those overexpressing the DP mutant of SREBP-1c. However, the DN mutant was not able to raise ␤ 3 -adrenoreceptor mRNA, possibly because it is already expressed at remarkably high levels in the 3T3-L1 cell line. Interestingly, the ␤ 3 -adrenoreceptor gene possesses putative SREBP binding sites distal to the coding sequence at positions 6473 and 6588 relative to ATG. Thus, whether the ␤ 3 -adrenoreceptor gene is a negative SREBP target, as reported for the microsomal triglyceride transfer protein gene (49) and phosphoenolpyruvate carboxykinase (50), remains to be firmly established. This would be relevant with an integrated role for SREBP-1c to promote overall energy storage by stimulating lipogenic enzymes and to favor antilipolysis by diminishing an important component of the lipolytic machinery.
SREBP-1c Directly Transactivates the C/EBP␤ Promoter-A significant finding of the present study is the observation that C/EBP␤, C/EBP␦, and PAI-1 can now be considered as new SREBP-1c transcriptional targets. This statement is based on data showing that the expression of these genes is insulinsensitive, stimulated by the forced expression of the dominant positive mutant, and down-regulated by the dominant negative  Table II.

TABLE III
Run-on transcription analysis of gene expression in 3T3 L1 adipocytes treated by insulin or infected by Ad.SREBP-1c DP Transcription rates of the indicated genes were measured in control cells, insulin-treated cells, or cells infected with Ad.SREBP-1c. Nuclei were pooled from at least 10 100-mm culture dishes. After in vitro elongation of initiated transcripts in the presence of [␣-32 P]CTP, labeled RNA was hybridized to the indicated plasmids and counted. All of the cDNAs have been described elsewhere and were cloned in pUC-derived vectors. Hormone-sensitive lipase, being not transcriptionally regulated by insulin, was used as negative control. Values were normalized for hybridization efficiency with signals generated by immobilized genomic DNA. Results are expressed as the ratio to basal transcription rates obtained in control, untreated cells. One typical experiment, representative of two, is shown. SREBP-1c mutant. Moreover, their transcription rates assessed by run-on experiments were stimulated in SREBP-1coverexpressing cells. The importance of C/EBP␤ and C/EBP␦ has been clearly established in the context of the adipocyte differentiation program. They act at an early stage of the adipocyte conversion, as suggested by their expression pattern that peaks early after cell confluence (51). When induced by appropriate drugs (52), C/EBP␤ and C/EBP␦ can initiate a transcriptional cascade that culminates in the induction of PPAR␥ and C/EBP␣, which in turn induces the expression of the mature adipocyte phenotype. In the present study of fully mature differentiated fat cells, C/EBP␤ and C/EBP␦ expression were supposed to have returned to basal levels. However, if one calculates the respective proportion of the mRNA encoding the three C/EBP isoforms in the differentiated fat cells, which can be done under the real time RT-PCR conditions of this study (see "Materials and Methods"), it can be found that the steady state levels of C/EBP␤ were still very high in the fully mature adipocyte and most abundant among the three C/EBP isoforms. Upon insulin treatment, C/EBP␤ mRNA was largely predominant, representing 82% of total C/EBP mRNA. C/EBP␣ and C/EBP␦ then account for 17 and 1%, respectively. Similarly, in the SREBP-1c DP-overexpressing cells, C/EBP␤ mRNA represented 98% of the total C/EBP transcripts. This relative abundance of the C/EBP␤ mRNA among the other isoforms suggests that the increase in the C/EBP␤ mRNA levels by insulin and SREBP-1c might play a significant role on the mature adipocyte gene transcription program.
To further establish the direct control of SREBP-1c on C/EBP␤ transcription, we performed cotransfection experiments in which was tested the ability of an SREBP-1c-expressing vector to activate the C/EBP␤ promoter controlling the expression of the luciferase reporter gene. Fig. 3 shows that in the context of the proliferating 3T3-L1 cells, in which SREBP-1c expression is virtually absent, basal C/EBP␤ promoter activity is low, with luciferase expression exceeding only by 2-fold that obtained with the promoterless pGL3 construct.
This fits with the known very low expression of C/EBP␤ in proliferating 3T3-L1 preadipocytes. At this stage, in the absence of endogenous SREBP-1c, ectopic expression of this transcription factor was able to transactivate the 1.4-kb C/EBP␤ promoter construct, with a 4-fold stimulation of luciferase reporter expression (Fig. 3A). However, transactivation by SREBP-1c could not be observed using shorter promoter constructs, indicating that the SREBP-1c-responsive region in the C/EBP␤ promoter was located upstream of Ϫ441. Using the TransFac data base (53) and the TFSearch algorithm aimed at searching transcription factor-binding sites (available on the World Wide Web at http://www.rwcp.or.jp/papia/), we identified two consensus sequences for potential SREBP binding sites in the 1.4-kb promoter sequence, one located at Ϫ1124 (SRE1) relative to the transcription start site and the other at Ϫ1064 (SRE2). These sites are located within the SREBP-1cresponsive fragments identified in the cotransfected experiments, suggesting that they might be involved in SREBP-1c responsiveness of the C/EBP␤ promoter. 50 bp upstream of these SRE sites, we also found a Sp1 binding sequence, the presence of which was shown to be required for efficient transcriptional activation by SREBP-1c (10). This further suggested that the SRE sequences identified here on the C/EBP␤ promoter might be functional. Fig. 3B shows the results of experiments in which point mutations in these SRE sites have been introduced. The mutation of the SRE1 sequence completely abolished the ability of cotransfected SREBP-1c to transactivate the 1.4-kb C/EBP␤ promoter. On the other hand, the response of the promoter was severely blunted when the SRE2 site was mutated. In addition, in a construct bearing a double mutation of both SRE1 and SRE2 sites, no transactivation by SREBP-1c could be observed. This clearly demonstrated that SREBP-1c responsiveness of the C/EBP␤ promoter relies on the presence of two identified SRE sites and suggests a predominant functional effect of the upstream SRE1. To further establish the direct control of C/EBP␤ promoter activity by SREBP-1c and by insulin, we addressed the question of the localization of the insulin-responsive region in C/EBP␤ promoter. For such purpose, we used fully differentiated and insulin-responsive 3T3-L1 cells, and we show in Fig. 4 that the full-length 1.4-kb promoter responded to the addition of insulin in the culture medium by a 2-fold stimulation of the luciferase reporter. We found that the insulin effect was highly reproducible among experiments, but of low magnitude, not exceeding a 2-fold increment, consistent with the results of nuclear run-on experiments (Table III). Fig. 4 also shows that the effect of insulin is abolished when cells were electroporated with promoter constructs bearing point mutations in one or both of the SRE binding sites. This demonstrates that these SRE sites are required for the insulin responsiveness of the C/EBP␤ promoter. Altogether, these results establish that SREBP-1c is able to transactivate the C/EBP␤ promoter through canonical SRE binding sites that coincide with the insulin-responsive region of the promoter, identifying C/EBP␤ as a new direct transcriptional target of SREBP-1c and insulin. Finally, having established that the effect of insulin on C/EBP␤ gene expression was mediated through a direct transcriptional control of SREBP-1c at the promoter level, we examined how SREBP-1c and insulin affected C/EBP␤ protein. Fig. 5 shows the results of Western blot analyses of C/EBP␤ protein contents in nuclear extracts from 3T3-L1 adipocytes exposed to insulin for 24 h or infected with the adenovirus encoding SREBP-1c. Two protein products are synthesized from the single C/EBP␤ messenger RNA: a transcriptionally active LAP form and a shorter naturally occurring dominant negative LIP that lacks the N-terminal transactivating domain of the protein. In agreement with the work of MacDougald et al. (34), which first described the stimulatory effect of insulin on C/EBP␤, we observed a 4-fold stimulation in total C/EBP␤ content (LIP and LAP) in cells treated for 24 h with insulin. A similar effect, although of higher magnitude, was obtained by infection of the cell with Ad SREBP-1c, showing that SREBP-1c, like insulin, was able to increase the intracellular levels of C/EBP␤ proteins. We also evaluated the relative proportion of LAP and LIP (Fig. 5, lower  panel) and observed that the molecular ratio of LIP/LAP increased from 1:9 in control cells to 3:7 in insulin-treated adipocytes and 5:5 in cells overexpressing SREBP-1c. This indicates that the effects of SREBP-1c overexpression on the repartition of the C/EBP␤ isoforms closely mimic that of insulin. This reinforced the conclusion that in fully differentiated adipocytes, insulin stimulates C/EBP␤ through SREBP-1c and favors the LIP form.
In conclusion, the present study provides experimental evidence that insulin activation of SREBP-1c in adipocytes initiates a transcriptional cascade involving C/EBP␤. Interestingly, the implication of C/EBP␤ as a transcriptional mediator of the effects of insulin has been recently published (54,55) in studies that focused on the regulation of the insulin-like growth factor-binding protein-1 gene in the liver. Unfortunately, whether SREBP-1c was involved in that regulation has not been investigated by the authors. Their data in addition to ours strongly suggest that C/EBP␤ might be an important relay for the transcriptional effects of insulin. Thus, a main finding in this study is the demonstration of a direct link between SREBP-1c and C/EBP␤. Interestingly, a recent publication by Farmer's group (56) argues for the existence of such a link. They demonstrated that the inhibition of C/EBP␤ renders the preadipocytes dependent on exogenous PPAR␥ ligands for their differentiation, suggesting that C/EBP␤ might be involved in the activation of PPAR␥ by triggering the production of ligands. The same role had been proposed for SREBP-1c by Kim et al. (57), who established that ADD1/SREBP-1c could control the adipocyte production of endogenous PPAR␥ ligand(s). In this regard, the present study, which demonstrates that C/EBP␤ expression is controlled by SREBP-1c, might explain why SREBP-1c and C/EBP␤ share a common ability to activate PPAR␥. Finally, in the mature adipocyte, since SREBP-1c can induce C/EBP␤, our results suggest that some SREBP-controlled mechanisms might involve a transcriptional relay through C/EBP␤. This might be particularly relevant in mediating the effects of insulin and in particular on the maintenance of the insulin-sensitive state that characterizes the differentiated adipocyte phenotype.