A GC-rich Region Containing Sp1 and Sp1-like Binding Sites Is a Crucial Regulatory Motif for Fatty Acid Synthase Gene Promoter Activity in Adipocytes IMPLICATION IN THE OVERACTIVITY OF FAS PROMOTER IN OBESE ZUCKER RATS*

We have previously shown that the proximal 2-kb se- quence of the fatty acid synthase (FAS) promoter transfected into rat adipocytes was highly sensitive to the cellular context, displaying an overactivity in obese ( fa / fa ) versus lean Zucker rat adipocytes. Using deletional analysis, we show here that FAS promoter activity mainly depends on a region from (cid:50) 200 to (cid:50) 126. This sequence exerts a strong negative effect on FAS pro- moter in adipocytes from lean rats but not in those from obese rats, resulting in a marked overtranscriptional activity in the latter cells. This region, fused to a heterologous promoter, the E1b TATA box, induced differen- tial levels of gene reporter activity in lean and obese rat adipocytes, indicating it harbors fa -responsive ele- ment(s). Whatever the rat genotype, adipocyte nuclear proteins were shown to footprint the same protected sequence within the fa -responsive region, and super-shift analysis demonstrated that Sp1 or Sp1-like pro- teins were bound to this DNA subregion. Compelling evidence that the Sp1 binding site contained in this sequence was implicated in the differential promoter activity in lean versus obese rats, was provided by the observation that mutations at this Sp1 site induced a 2.5-fold increase in FAS promoter activity in adipocytes from lean rats, whereas they had no effect in adipocytes from obese rats. polymerase chain reaction product of the FAS promoter region ligated in the Sal I site of pBLCAT2. The internal deletion the 318 118 by

Fatty acid synthase (FAS) 1 is a multifunctional enzyme that catalyzes all the reaction steps in the conversion of acetyl-CoA and malonyl-CoA to palmitate. FAS plays a central role in de novo lipogenesis, and its level of expression is a key determinant of the maximum capacity of a tissue to synthesize fatty acids. In mammals, FAS is expressed at especially high levels in liver and adipose tissue, where it greatly contributes to the regulation of triglyceride-rich lipoprotein production and to appropriate fat storage. In these tissues, FAS enzyme concentrations are under strict nutritional and hormonal control, and many studies using animal models or established cell lines have determined that the regulation of FAS activity is exerted mainly at the transcriptional level of gene expression. In particular, insulin (1), thyroid hormone (2), and glucose (3) act as positive regulators, whereas cAMP (1) and polyunsaturated fatty acids (4) are able to suppress FAS gene transcription in an independent manner. In physiological conditions, these signals and presumably others, yet to be defined, are likely to operate in an interactive manner to elicit an integrated control of FAS transcription. It has been known for many years that pathological situations such as obesity are characterized by elevated FAS activity in adipose tissue and liver (5,6). In particular ob/ob, db/db mice and fa/fa rats in which obesity is due to inherited defects in the newly discovered leptin regulatory pathway (7)(8)(9) all share enhanced lipogenic capacity in adipose tissue. Previously, we had observed that adipocyte FAS hyperactivity in genetically obese fa/fa rats was due to proportionate changes in FAS protein and mRNA levels. In addition, run-on analysis using adipocyte nuclei isolated from lean or obese rats has led to the conclusion that FAS gene transcription was the altered step in obese rats (10). Recently the development of a system of transiently transfected rat adipocytes, shown to be well suited for the study of FAS promoter activity, enabled us to demonstrate that the first 2 kb of the 5Ј-flanking region of the rat FAS gene were able to direct overtranscription in obese rat adipocytes, mimicking the in vivo fatty genotype effect (11). Similarly, by using transgenic mice, the same 2-kb promoter region was shown to be sufficient for the hormonal and nutritional regulation of FAS gene transcription (12). The objectives of the present study were 1) to delineate the sequences within the 2-kb promoter region that play a role in FAS gene transcription in the context of the mature unilocular rat adipocyte, and 2) to identify the region(s) involved in the responsiveness of FAS promoter to adipocyte fa-dependent transcription factors. We highlighted a 74-bp (Ϫ200 to Ϫ126) functional motif as the critically important control region for defining genotype-specific differences in the level of FAS gene transcription. This motif was found to exert a strong negative control on FAS promoter activity in lean but not in obese rat adipocytes. Sp1 was characterized as one of the proteins interacting with this DNA sequence, and we provide evidence of a differential role of the Sp1 binding site in lean and obese rat adipocytes.

MATERIALS AND METHODS
Plasmids and Constructs-The plasmid p(Ϫ2187ϩ65)FAS-CAT was obtained from S. Clarke and contained a XhoI fragment from the rat FAS promoter (13) cloned into the SalI site of promoterless pUC-CAT basic (Promega). From this plasmid, the authentic transcription start site was previously verified by primer extension (11). Deletions with 5Ј end points at Ϫ1009, Ϫ318, Ϫ200, Ϫ126, Ϫ118, and Ϫ35 were generated by digestion with appropriate restriction endonucleases. Ends were filled in with Klenow fragment before religation. The Ϫ269ϩ16 construct was a linker-mediated polymerase chain reaction product of the FAS promoter region ligated in the SalI site of pBLCAT2. The internal deletion of the Ϫ318Ϫ118 region was produced by cutting * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Ϫ2187/ϩ65 FAS-CAT with XmaIII and SmaI, followed by plasmid recircularization. The Ϫ269 to Ϫ118 fragment of the FAS promoter was excised from Ϫ269/ϩ16 FAS-CAT by SphI and SmaI digestion and subcloned into a SphI-XbaI-treated BCAT vector (14) from which CATgenerated activity was controlled by the adenovirus E1b TATA box. Ϫ200/ϩ65 FAS-CAT was used as a template to produce Ϫ200/ϩ65 Mut-Sp1FAS-CAT using the quick change mutagenesis kit (Stratagene). Mutagenic oligonucleotides were as follows (only the upper strand is shown): 5Ј-GCAGGGTCCCGGCTGGGTACGGCGCGC-3Ј. The sequence and orientation of each construct were controlled by using the dideoxy chain termination method in the T7 sequencing kit (Pharmacia Biotech Inc.).
Transient Transfection of DNA and Reporter Assays-Unilocular adipocytes isolated from 4-week-old lean (Fa/fa) or obese (fa/fa) Zucker rats were electroporated as described previously (11) using 20 g of FAS-CAT DNA and 5 g of RSV-␤-galactosidase DNA as an internal control. Schneider cells (15) were maintained in M3 medium (Life Technologies) supplemented with 10% fetal calf serum, 2 mM glutamine, and antibiotics at 25°C. One day prior to transfection, cells (4 ϫ 10 6 ) were plated onto 30-mm multiwell dishes. Transfection was performed using the calcium phosphate precipitation method with 6 g of FAS-CAT reporter construct, 1.5 g of RSV-␤-galactosidase, and 1 g of previously described (16) expression vector pADH-Sp1 containing or not containing (pADH) the Sp1 coding region under the control of the Drosophila melanogaster ADH promoter. 48 h after transfection, triplicate dishes of mature adipocytes or Schneider cells were harvested, and whole cell lysates were prepared by three rounds of freezing and thawing in 0.25 M Tris-HCl, pH 8, 5 mM dithiothreitol. CAT and ␤-galactosidase activities were assayed as described previously (11). Protein concentrations and reaction times in the CAT assay were adjusted to bring the extent of CAT conversion into a linear range. CAT activities were expressed as the percentage of conversion of chloramphenicol to its acylated products per hour and normalized to ␤-galactosidase activities to correct for variations in transfection efficiency. Each experiment was repeated at least three times using at least two different CsCl-purified plasmid preparations.
Preparation of Adipocyte Nuclear Extracts-The procedure for the preparation of nuclear extracts from isolated fat cell nuclei was modified from that described by Dignam et al. (17) and was published previously (11). Protein content of the samples was measured according to Bradford et al. (18), using ovalbumin as the standard.
DNase I Footprinting Analysis-The Ϫ2187ϩ65 FAS-CAT plasmid was linearized by XbaI digestion, and the recessed 3Ј ends were filled in with the Klenow fragment of DNA polymerase in the presence of [␣-32 P]dCTP and [␣-32 P]dGTP. This fragment was recut with EagI and gel-purified to yield a 383-bp fragment corresponding to the Ϫ318ϩ65 region of the FAS promoter labeled on the antisense strand. Probes labeled on the sense strand were also produced by labeling at EagI ends. DNase I footprinting was performed using adipocyte nuclear extracts as described previously (11) or using purified human Sp1 or AP-2 according to the supplier's protocol (Santa Cruz Biotechnology). End points of protected regions were determined according to G, GϩA, and C ladders of the Maxam-Gilbert sequencing reaction run in parallel (19).

Deletional Analysis of FAS Promoter Activity in Primary Cultured Adipocytes: Evidence of a Strong Negative Regulatory
Region between Ϫ200 and Ϫ126 -We have previously shown that the 5Ј-flanking 2-kb region of the rat FAS gene is able to direct a high level of reporter gene activity in transfected rat adipocytes through correctly initiated transcription (11). In order to identify the sequences essential for transcription of the FAS gene in these cells, progressively shorter fragments of the 5Ј-flanking region were inserted in front of the coding region of the CAT gene and transfected into rat adipocytes. Co-transfection with RSV-␤-galactosidase gene reporter plasmids was routinely carried out in order to correct for transfection efficiency, and CAT activities were normalized to ␤-galactosidase activities. As seen in Fig. 1, transcription doubled upon deletion of the sequence between Ϫ2000 and Ϫ1009, suggesting the presence of a weak negative regulatory element in this region. Deletions to Ϫ318 and to Ϫ269 did not cause any further significant changes in CAT activity. In contrast, the deletion of the sequence between Ϫ269 and Ϫ118 led to a very significant 4-fold increase in transcriptional activity, pointing to a very strong negative regulatory element localized in this region. This conclusion was further ascertained by the observation that the 2-kb promoter deleted from the region Ϫ318 to Ϫ118 was four times more active than the intact promoter ( Fig. 2A). Finally, upon the truncation to Ϫ35, a dramatic loss of CAT activity occurred, indicating the presence of a crucial positive regulatory element within the proximal region of the promoter. Fig. 1 also illustrates FAS promoter activity in transfected adipocytes from obese fa/fa rats. In these cells, CAT expressions driven by Ϫ35 and Ϫ118 promoter fragments were similar to those measured in adipocytes from lean animals. In contrast, the region from Ϫ269 to Ϫ118 did not exhibit any negative effect on promoter activity in adipose cells from obese rats. Therefore, a 4 -6-fold increase in CAT activity was observed with the Ϫ269 to ϩ16 fragment and with longer promoter constructs, in adipocytes from obese as compared with lean rats. The crucial role of the region extending from Ϫ269 to Ϫ118 in the differential activity of FAS promoter in lean and obese rat adipocytes was further established by two sets of observations. First, the removal of the sequence Ϫ318 to Ϫ118 from the 2-kb promoter abolished the differences between the two genotypes ( Fig. 2A). Second, the fusion of the Ϫ269 to Ϫ118 sequence to a heterologous promoter, the TATA box of E1b adenovirus, induced a 3-4-fold increase in CAT activity in obese rat adipocytes, whereas it had no effect in lean rat adipocytes (Fig. 2B). This observation defines the Ϫ269 to Ϫ118 fragment as the fa-responsive region.
To further delineate the functional sequence harboring responsive elements to fa-dependent transcription factors, two additional CAT deletion constructs were created with 5Ј end points at Ϫ200 or Ϫ126. Results in Fig. 2C show that the negative regulatory region that acts to suppress FAS transcription in adipocytes from lean rats is restricted to a region spanning from Ϫ200 to Ϫ126 bp. This localized the fa-responsive element to within this 74-base pair sequence.
FAS Promoter Contains Multiple Nuclear Protein Binding Sites-In an attempt to identify the transcription factors reg-ulating the FAS promoter we next performed DNase I footprinting over the first 300 bp of the FAS promoter using increasing amounts of nuclear extracts from lean rat adipocytes (Fig. 3A). The first region (FPI) protected by adipocyte nuclear

FIG. 3. DNase I footprinting analysis of FAS promoter with adipocyte nuclear proteins (A and C) or purified human AP-2 and Sp1 (B)
. 10 fmol of a Ϫ318/ϩ65 FAS promoter fragment, labeled on the antisense strand (15,000 cpm/ng) were subjected to DNase I digestion after incubation with increasing (50, 70, and 100 g) amounts of nuclear proteins from adipocytes or with increasing (0.1, 1, 2, and 5 footprint-forming units) amounts of purified human Sp1 or AP-2 (1, 2, and 5 footprint-forming units). The migration of Maxam-Gilbert sequencing ladders from the same DNA fragment is shown on the right. Boxes protected by adipocyte nuclear extracts are indicated in the middle, and the position of consensus binding sites for Sp1 and AP2 is shown on the left. Panel C represents protection patterns of nuclear extracts from lean or obese rat adipocytes on FPIV using a Ϫ318 to ϩ65 FAS promoter fragment labeled on the sense strand as a probe. Increasing amounts (50, 70, and 100 g) of nuclear proteins from adipocytes of lean or obese rats were used.

FIG. 2.
Functional importance of the ؊269 to ؊118 and ؊200 to ؊126 FAS promoter fragments on CAT expression in rat adipocytes. 20 g of FAS-CAT constructs represented as diagrams on the left were cotransfected with 5 g of RSV-␤-galactosidase in adipocytes isolated from lean (white bars) or obese (black bars) rats. The bar graphs on the right show ␤-galactosidase-normalized promoter activities expressed either as a function of the Ϫ2000 CAT value (panel A) or the E1b TATA box-driven CAT value (panel B) in adipocytes from lean rats set arbitrarily to 1 or as absolute values (panel C). Results were obtained from at least four independent experiments performed in triplicate. extracts, from Ϫ10 to Ϫ40, overlaps with the TATA-box sequence (13) and might therefore correspond to the binding of members of the transcription initiators. The second protected region (FPII), extending from Ϫ42 to Ϫ64, coincides with the position where liver nuclear extracts were shown to bind to the insulin-responsive sequence (20), suggesting that adipocytes, like hepatocytes, contain insulin-responsive sequence-binding protein(s). The third protected box (FPIII), from Ϫ75 to Ϫ96, matches with a site recently characterized by Rangan et al. (21) as a cAMP-responsive element. The fourth region (FPIV), ranging from Ϫ135 to Ϫ180, is located within the fa-responsive region. A fifth protected region (FPV), from Ϫ210 to Ϫ255, i.e. upstream of the fa-responsive region, was detected.
Since several potential binding sites for Sp1 and AP-2 are present on this promoter fragment (schematically shown on the left in Fig. 3), the probe was footprinted with purified human Sp1 and AP-2 (Fig. 3B). Footprinting with purified AP-2 revealed a true AP-2 binding site, centered at Ϫ122, that did not appear to be occupied by adipocyte nuclear proteins. In addition, purified AP-2 also protected the FPIV region and part of the FPV. The Sp1 binding site at Ϫ90 was footprinted by human Sp1, in agreement with a recent report of Bennett et al. (22). In addition, two other Sp1 binding sites were detected, in FPIV and FPV.
Identical band protection patterns were observed with adipocyte nuclear proteins from lean and obese rats as shown in Fig. 3C for the fa-responsive region.
Sp1 and Sp1-like Proteins Bind the Ϫ180 to Ϫ135 Region (FPIV) of the FAS Promoter-The next step in our study was to characterize adipocyte nuclear protein binding to the fa-responsive region of the FAS promoter. For this purpose, gel mobility shift assays were performed using the Ϫ180 to Ϫ135 footprinted region IV as a probe (Fig. 4). Four specific retarded complexes, C1, C2, C3, and C4, competed away with a 30-fold molar excess of unlabeled FPIV but not with an unrelated oligonucleotide (YY1 consensus binding site), were detected. Since FPIV contains potential binding sites for Sp1, Egr-1, and AP-2 (schematically represented in the lower panel of Fig. 4), competitions with consensus oligonucleotides were performed. Fig. 4A shows that a 100-fold excess of Egr-1 oligonucleotide was unable to disrupt the binding of any complex on the FPIV probe, suggesting that this factor is not involved in complex formation on this region. Moreover, no specific binding of adipocyte nuclear proteins was observed using a labeled Egr-1 consensus oligonucleotide (data not shown). Binding of nuclear extracts on the FPIV probe could not be competed away by a 10or 30-fold molar excess of AP-2 consensus oligonucleotide, although a 300-fold molar excess produced a weak displacement. This observation together with the inability of anti-AP-2 antibody to produce supershifted bands, suggests that AP-2 is not involved in complex formation of adipocyte nuclear proteins with FPIV. By contrast, C1, C2, and C3 were efficiently competed away by a molar excess of Sp1 consensus oligonucleotide but not by a mutated Sp1 oligonucleotide (data not shown), strongly suggesting that these complexes contain Sp1 or Sp1related proteins. Moreover, purified Sp1 binding to the FPIV probe leads to a complex that co-migrates with C1 (Fig. 4B). Finally, the identification of C1 and C2 as authentic Sp1 binding complexes was further established by supershifting with anti-Sp1 antibody. C3, not supershifted by anti-Sp1 antibody, might contain other members of the multigene Sp1 family that exhibit the same binding properties as Sp1 (23,24). Altogether, these experiments demonstrate that Sp1 is present in C1 and C2 binding complexes and that Sp1-like proteins are part of C3. The composition of C4 remains totally unknown.
Sp1 Binding Site in FPIV Is Functional-Gene transfer experiments into the D. melanogaster Schneider cell line (SL2) that lacks endogenous Sp1 and nuclear factor 1 transcription factors were performed in order to further substantiate the possibility that Sp1 or related members of the Sp1 family Electrophoretic mobility shift assays were performed with a double-stranded 32 P-labeled FPIV probe (the sequence of only one strand is shown). The probe was incubated with 10 g of adipocyte nuclear proteins (A) or with 1 footprint-forming unit of purified human Sp1 (B). Competition with molar excess of unlabeled double-stranded oligonucleotides is indicated. In some binding reactions, antibodies raised against Sp1 (Sp1 Ab) or AP-2 (AP2 Ab) were used, as indicated under "Materials and Methods." The positions of complexes C1, C2, C3, and C4 are indicated on the left.
control the level of transcription of the FAS promoter (Fig. 5). In SL2 cells co-transfected with the full-length FAS promoter construct and the control vector pADH in which the sequence encoding Sp1 has been deleted, we were able to measure significant amounts of FAS-driven CAT activities, indicating that the expression of FAS promoter is not strictly Sp1-dependent. The presence of nuclear factor 1 is also not necessary for FAS expression, in agreement with a previous study using an in vitro transcription assay (25). In SL2 cells transfected with the full-length FAS-CAT construct, forced expression of Sp1 elicited a 3-fold increase in CAT expression, demonstrating the presence of functional Sp1 site(s) in the FAS promoter. A similar effect of Sp1 was observed with the Ϫ318 to ϩ65 FAS-CAT construct, suggesting that the functional Sp1 binding sites are present within this region. The removal of the Sp1 site located within the FPV fragment (by deletion of the region from Ϫ318 to Ϫ200) did not substantially alter the ability of Sp1 to stimulate CAT activity. In contrast, a complete loss of Sp1 stimulation was observed upon transfection of cells with the Ϫ118 to ϩ65 FAS-CAT promoter construct or with the full-length promoter in which the Ϫ318 to Ϫ118 fragment had been deleted. This establishes that Sp1 stimulation of promoter-driven CAT activity arises exclusively from the site located within the Ϫ118 to Ϫ200 region.

The Mutation of Sp1 Binding Site in FPIV Affects FAS Promoter Activity in Adipocytes-The
Ϫ200ϩ65 FAS-CAT construct in which the Ϫ161 Sp1 binding site had been mutated was transfected in rat adipocytes (Fig. 6). The results show that the Sp1 mutation elicited a 2.5-fold increase in CAT expression in adipocytes from lean rats, suggesting that the Sp1 binding site is partly responsible for the negative effect of the Ϫ200 to Ϫ126 region on FAS gene transcription. In contrast, the mutation did not modify reporter gene expression in adipocytes from obese animals. The differential effect of the mutated Sp1 binding site in lean and obese rat adipocytes was not explained by a change in Sp1 expression level as assessed by immunoblot analysis using a specific anti-Sp1 antibody (data not shown). DISCUSSION The cellular context is of critical importance for the study of regulated gene expression. This study provides the first dele-tional analysis of FAS promoter activity in primary cultured rat adipocytes, a question that has been previously documented only in 3T3-L1 cells, either fibroblasts or differentiated adipocytes (26), and in Hep G2 cells (27). Our data show that a major negative regulatory region contributes to FAS transcriptional activity in mature unilocular adipocytes, a distinct feature that has not been observed in cell lines. This difference underscores the importance of using physiologically relevant cell systems in this type of studies. The negative region has been delineated to a 74-bp sequence (Ϫ200 to Ϫ126) and shown by mutational analysis to contain a Sp1 binding site that is partly responsible for its regulatory feature. Sp1 is unlikely to be the perpetrator acting through this Sp1 binding site, since Sp1 is a well known activator of gene transcription (16) and exerts a positive effect on FAS gene transcription in SL2 cells. Our in vitro binding studies identified Sp1 and Sp1-like transcription factors as some of the components of the multiprotein complexes linked to this DNA segment. Thus, proteins acting through the Sp1 binding site such as Sp3 (28) or p74 (29), which are negative regulators of Sp1 activation, might contribute to the activity of this region.
The importance of this 74-bp region was further established by our observation that its negative effect on FAS gene transcription is lost in adipocytes from genetically obese fa/fa rats in which FAS is overexpressed. Importantly, this region fused to a heterologous promoter was able to direct a differential level of CAT activity in lean and obese rat adipocytes, thus demonstrating that it contained sequences responsive to fa-dependent transcription factors. Previously, fa-responsive regions have been delineated on the promoters of two other obesity-regulated genes, namely GAPDH (11) and Glut4 (30).The presence of GC-rich putative Sp1 binding sites on both of these gene promoters and the presence of a functionally active site for Sp1/Sp1-like proteins demonstrated here on the FAS promoter, suggest that the Sp1 binding site is a common trait of the fa-responsive regions. As shown here, mutation of the Sp1 binding site in the fa-responsive region does not alter the capacity of adipocytes from obese rats to direct high transcriptional activity of the FAS promoter. Thus, it is possible that the down-regulating factor(s) acting through the Sp1 binding site might be inactive in obese rat adipocytes. Alternatively, although not detected under the present experimental conditions, adipocytes from obese rats might specifically express a potent activator overriding the negative effect of the Sp family members.
The recent discovery that the db mutation, the mouse homolog of fa, invalidates the leptin receptor (8,9) raises new questions about the molecular pathogenesis of obesity in Zucker rats. In view of the fact that leptin receptor expression in mice is not restricted to the central nervous system but is instead expressed ubiquitously, including in adipose tissue, we cannot rule out the possibility of a direct control of adipocyte gene expression by leptin. The structure of the leptin receptor deduced from the cDNA sequence defines it as a new member of the cytokine receptor superfamily (31). Since the signals transduced via cytokine family receptor complexes are able to modulate the expression of a wide variety of genes, we can speculate that the fa-responsive region of the FAS gene might be an ultimate target of the leptin receptor signaling pathway. Thus, the crucial regulatory region defined here on the FAS promoter could provide a tool for unraveling the leptin signaling pathway.