Transcriptional Regulation of the Mouse Uncoupling Protein-2 Gene DOUBLE E-BOX MOTIF IS REQUIRED FOR PEROXISOME PROLIFERATOR-ACTIVATED RECEPTOR- g DEPENDENT ACTIVATION*

, Uncoupling protein-2 (UCP2) is present in many tissues with relevance to fuel metabolism, and its expression is increased in fat and muscle in response to ele-vated circulating free fatty acids resulting from fasting and high fat feeding. We proposed a role for peroxisome proliferator-activated receptor- g (PPAR g ) as a mediator of these physiological changes in UCP2, because thiazolidinediones also increase expression of UCP2 in these cell types (1). To determine the molecular basis for this regulation, we isolated the 7.3-kilobase promoter region of the mouse UCP2 gene. The 2 7.3-kilobase/ 1 12-base pair fragment activates transcription of a reporter gene by 50–100-fold. Deletion and point mutation analysis, coupled with gel shift assays, indicate the presence of a 43-base pair enhancer ( 2 86/ 2 44) that is responsible for the majority of both basal and PPAR g -dependent transcriptional activity. The distal ( 2 86/ 2 76) part of the enhancer specifically binds Sp1, Sp2, and Sp3 and is indis-tinguishable from a consensus Sp1 element in com-petition experiments. Point mutation

Uncoupling protein-2 (UCP2) 1 was discovered in 1997 as a homologue of the brown fat UCP (UCP1) (2). Like UCP1, which is able to dissipate caloric energy as heat by uncoupling mitochondrial respiration from ATP production, UCP2 can similarly uncouple respiration, at least in vitro (2). This fact, coupled with other significant features, including tissue distribution and genetic location in a strong quantitative trait locus for hyperinsulinemia and leptin levels, made UCP2 an attractive candidate for regulation of resting metabolic rate and as an obesity/diabetes susceptibility gene. However, we recently reported that targeted disruption of the UCP2 gene does not lead to spontaneous or diet-induced obesity and diabetes (3), but changes in sensitivity to glucose-stimulated insulin secretion are observed. 2 Thus, the relationship between energy metabolism and UCP2 remains unclear.
Transcripts and protein expression of UCP2 are found in many tissues with relevance to fuel metabolism such as pancreatic ␤-cells, white and brown fat, skeletal muscle, and hypothalamus. In response to high fat diets, we have previously reported adipose-specific increases in expression of UCP2 in obesity-resistant mouse strains such as A/J and C57BL/Kalis, which are absent from the obesity-prone C57BL/6J strain of mice (2,4). Of mechanistic importance, blockade by nicotinic acid of both the fasting-induced rise in free fatty acids, as well as fatty acid transport into mitochondria (5), prevented the fasting-induced increase in UCP2 mRNA in muscle. The importance of changes in fat metabolism for regulation of UCP2 expression was further supported by the observation that fasting-induced increases in circulating free fatty acids led to a stimulation of UCP2 expression in adipose tissue and muscle, whereas subsequent refeeding suppressed UCP2 expression (6 -8). However, this down-regulation of UCP2 was not observed when the post-fasting diet contained a high percentage of calories from fat (9).
Although these findings provide compelling evidence for the involvement of fatty acids in regulation of UCP2 expression, the molecular mechanisms remain unknown. However, a PPARdependent pathway is one of the possible mechanisms that can * This work was supported in part by National Institutes of Health Grants R01-DK54024 (to S. C.) and F31-DK09812 (to S. K. S. and S. C.), Center National de la Recherche Scientifique (to D. R.), Human Frontier Science Program RG 0307 (to D. R.), Institut de Recherche Servier (to D. R.), and Institut National de la Santé et de la Recherche Médicale Grant 4P007E (to D. R.). 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.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBank TM  be considered to mediate the response of UCP2 to fatty acids, because it has been shown that at least some fatty acids can bind PPAR␥ specifically and may act as natural ligands for this transcription factor (10,11). It also has been shown that treatment with PPAR␥ agonists increases UCP2 mRNA levels in vivo and in vitro (1,12). PPAR␥ plays an important role in the control of the expression of many genes involved in energy metabolism by binding specific elements consisting of a direct repeat (DR-1) of a consensus sequence (AGGTCA) separated by one base, although functional peroxisome proliferator-responsive elements (PPREs) can deviate significantly from the consensus (13). The presence of functional PPREs in the regulatory sequences of such genes as the adipocyte fatty acid-binding protein, aP2 (14), phosphoenolpyruvate carboxykinase (15), acyl coenzyme A synthetase (16), and lipoprotein lipase (17) is consistent with the crucial role attributed to PPAR␥ in lipid metabolism. To further understand the mechanisms regulating transcription of the UCP2 gene, as well as a role of PPAR␥ in its regulation, we report here cloning and analysis of the transcriptional activity of a 7.3-kb fragment of mouse UCP2 promoter.

Cloning of Mouse UCP2 Promoter and Construction of Deletion and Point Mutants in Luciferase Reporter
Plasmids-A mouse genomic library (129 SVJ mouse DNA digested by Sau3A DNA and inserted into lambda FIX II vector, catalog number 946306; Stratagene, Palo Alto, CA) was screened using a 1,304-bp PCR probe encompassing intron 1 and exon II of the mouse UCP2 gene. The PCR probe was generated using sense 5Ј-TGCACTCCTGTGTTCTCCTG-3Ј and reverse 5Ј-GACTCTGGAGTTTCGTCGGA-3Ј primers and corresponding to intron 1 and exon II. The cloned genomic fragment (made of 7,452 bp) was inserted into the NotI site of pSPORT plasmid (Life Technologies, Inc.) and entirely sequenced (18). Based upon this sequence from the 129 SVJ mouse gene, the same region was isolated from the A/J, C57BL/6J, and C57Ks/J strains and entirely sequenced. To construct luciferase reporter plasmids, fragments of the A/J mouse UCP2 promoter (Gen-Bank TM accession number AF115319) were cloned into pGL2-Basic (Promega) using restriction fragments or amplified by PCR and cloned using the primer-introduced MluI and NheI restriction sites. Point mutations were made by direct cloning of chemically synthesized oligonucleotides or by PCR. The integrity and fidelity of all promoter-reporter constructs thus made were verified by DNA sequencing.
Cell Culture Transfections and Reporter Gene Assays-C3H10T1/2 (ATCC number CCL-226) and HIB-1B preadipocytes (gift from Dr. Reed Graves) and C2C12 (ATCC number CRL-1772) myoblasts were maintained in Dulbecco's modified Eagle's medium and D1 preadipocytes (19) on RPMI 1640 medium supplemented with 10% fetal calf serum, 100 units/ml penicillin, and 100 g/ml streptomycin (Life Technologies, Inc.) at 37°C in a humidified 5% CO 2 atmosphere. For transfection experiments cells were plated into 6-well dishes and grown overnight to 50 -60% of confluence. In cotransfection experiments plasmid pGEM3z was used to equalize amount of DNA per well. All transfections included a reporter plasmid containing a 2-kb promoter region of human ␤-actin and chloramphenicol acetyltransferase (CAT) as an internal control (20). Prior to transfection, cells were washed twice with calcium-free phosphate-buffered saline and then incubated with 1 ml/ well of transfection mixture, which contained 4 l of LipofectAMINE (Life Technologies, Inc.) and 1.5 g of total amount of DNA prepared in serum-free, antibiotic-free F-12 medium according to the manufacturer's protocol (Life Technologies, Inc.). After 5 h of incubation, transfection mixture was removed, and cells were supplemented with fresh medium containing 10% fetal bovine serum. For CAT and luciferase assays cells were harvested 48 h after transfection. Whole cell extracts were prepared according to CAT enzyme-linked immunosorbent assay kit protocol (Roche Molecular Biochemicals) and used in both CAT and luciferase assays. Luciferase activity of the UCP2 reporter constructs is normalized for CAT activity of the control plasmid. Where indicated, cells were treated with BRL49653 (GL237310x).
Electrophoretic Mobility Shift Assay-Nuclear extracts were prepared from cultured cells according to the method of Dignam et al. (21) with some modifications. Briefly, cells were washed twice in ice-cold phosphate-buffered saline, collected into microtubes, and centrifuged for 15 s in a microcentrifuge. The cell pellet was then resuspended in 1 ml of ice-cold, low salt buffer (10 mM HEPES, pH 7.9, 1 mM EDTA, 10 mM KCl, 1 mM phenylmethylsulfonyl fluoride, 0.5 mM dithiothreitol, and protease inhibitors). After incubation on ice for 20 min, Nonidet P-40 was added to a final concentration of 0.5%. The cell suspension was vigorously vortexed for 20 s followed by centrifugation for 30 s in a microcentrifuge to collect the nuclei. The nuclei were resuspended in 10 volumes of high salt buffer (10 mM HEPES, pH 7.9, 1 mM EDTA, 0.42 M NaCl, 10% glycerol, and protease inhibitors) and incubated on ice for 20 min followed by a 10-min centrifugation in a microcentrifuge at 4°C. The supernatants were collected and stored in aliquots at Ϫ70°C. Oligonucleotides for electrophoretic mobility shift assay were end-labeled with [␥-32 P]ATP (PerkinElmer Life Sciences) by T4 polynucleotide kinase New England Biolabs and purified with MicroSpin G-25 columns (Amersham Pharmacia Biotech). Approximately 0.2 ng (50,000 cpm) of the oligonucleotide probe and 0.5 g of nuclear extract were routinely used in the binding reaction in buffer containing 20 mM HEPES, pH 7.9, 50 mM NaCl, 1 mM EDTA, 1 mM phenylmethylsulfonyl fluoride, and 10% glycerol. To minimize nonspecific binding, 1 g of poly(dI-dC) and 0.2 g of salmon sperm DNA were included in the reaction. The binding reactions were carried out at room temperature for 25 min. To determine specificity and relative affinity, the reactions may contain a 100fold molar excess of unlabeled oligonucleotides as competitors. DNAprotein complexes were separated by 6% nondenaturing polyacrylamide gel electrophoresis at room temperature in 0.5ϫ Tris-borate EDTA, dried, and exposed to PhosphorImager plates for imaging (Storm680; Molecular Dynamics).
Isolation and Analysis of RNA-Total cellular RNA was prepared by the Tri reagent method according to the manufacturer's protocol (Molecular Research Center, Inc). For Northern blot hybridization, RNA was denatured by the glyoxal procedure, fractionated through 1.2% agarose gels, and blotted onto Biotrans (ICN) nylon membranes (22). Radiolabeled probes were prepared by random primer extension (Prime-It RmT; Stratagene) of the purified DNA fragments in the presence of [␣-32 P]dCTP to a specific activity Ͼ 2 ϫ 10 9 dpm/g DNA. The DNA fragments that were used as probes were obtained from the following sources. A fragment specific for the mouse UCP2 was prepared as described in Ref. 2. A rat cDNA probe for cyclophilin was used as an internal hybridization/quantitation standard. Blots were hybridized and washed as previously described (23,24). The intensity of hybridization signals was quantified by a phosphorimager (Image-Quant/Storm) and normalized to the values for cyclophilin.

PPAR␥ Stimulates Expression and Promoter
Activity of UCP2-We and others have previously shown that PPAR␥ agonists are able to increase UCP2 mRNA levels in adipocytes and muscle cells (1,12,25). Fig. 1A shows that UCP2 mRNA is present at relatively high levels in HIB-1B brown and D1 white preadipocyte and C2C12 myoblast cell lines. Treatment of these cells for 24 h with the thiazolidinedione BRL49653, a PPAR␥ ligand, and agonist (26) results in a 2-fold increase of UCP2 mRNA in all cell lines. Note that these results in HIB-1B cells recapitulate our previously reported findings (1).
To analyze the mechanisms responsible for the transcriptional regulation of the mouse UCP2 gene, and to test whether PPAR␥ affects the activity of the UCP2 promoter, a fragment of the mouse UCP2 promoter region from position Ϫ7359 to ϩ12 was isolated, sequenced, and subcloned into a luciferase reporter vector. This DNA fragment spans the entire region between the last exon of UCP3 and the first exon of UCP2 (4). As shown in Fig. 1B, and consistent with the Northern blot data of Fig. 1A, BRL49653 was able to stimulate transcriptional activity of the 7.3-kb fragment of the UCP2 promoter in all cell lines tested. Moreover, cotransfection of the 7.3-kb UCP2 promoter construct with an expression vector containing PPAR␥ (Fig.  1B) led to a similar 2-fold increase in UCP2 promoter activity. However, simultaneous agonist treatment and cotransfection with PPAR␥-expressing vector did not produce a significant cumulative effect, possibly indicating saturation of PPAR␥-dependent stimulatory activity on the UCP2 promoter. Such agonist-independent responses to PPAR␥ have been observed by others (27).
The Region between Ϫ86 and Ϫ44 Nucleotides of the UCP2 Promoter Is Important for Basal Transcriptional Activity and PPAR␥ Responsiveness-To identify regulatory sequences that are important for transcriptional control of the UCP2 gene, including the response to PPAR␥, we created a series of deletion constructs of the UCP2 promoter as shown in Fig. 2A. The transcriptional activity of these constructs was analyzed by transient transfection in D1, HIB-1B, and C2C12 cells (Fig. 2,  B-D). Fig. 2 shows that the largest fragment of the UCP2 promoter (Ϫ7.3 kb/ϩ12 bp) is able to increase luciferase activity of the reporter construct by 50 -150-fold depending on the cell line, whereas the activity of a minimal promoter (Ϫ44/ϩ12 bp) is only 1.5-2-fold above empty vector. Deletion of the region between Ϫ7.3 and Ϫ5 kb reduces promoter activity by 2-3-fold in HIB-1B and C2C12 (Fig. 2, B and D) but had no effect in D1 cells (Fig. 2C) or in C3H10T1/2 cells (data not shown). Although further deletions between positions Ϫ5 kb and Ϫ86 bp had no significant effect on activity, deletion of the region between positions Ϫ86 and Ϫ44 led to a dramatic 25-50-fold depression of activity, indicating that this region is indeed a major contributor to the overall expression of the UCP2 gene in all cell lines. Having established the pattern of basal activity of these deletion fragments, we next examined the activity of these regions in cells treated with BRL49653, as well as cotransfected with PPAR␥. As shown in Table I, deletion of the sequence up to Ϫ86 bp still retained PPAR␥-dependent stimulation of UCP2 promoter activity equivalent to the Ϫ7.3-kb fragment. However, deletion of the region between Ϫ86 and Ϫ44 bp resulted in a total loss of PPAR␥-dependent activity, indicating that this region of the UCP2 gene is necessary not only for basal activity but also for regulation by PPAR␥.
Identification of the Transcription Elements Located in the Ϫ86/Ϫ44 Region of the Mouse UCP2 Promoter-Because our data establish that the Ϫ86/Ϫ44 region of the mouse UCP2 promoter is a major locus of control in cell lines of fat and muscle origin, in this series of experiments we focused on identifying the elements and transcription factors involved in regulation of this region. Examination of the sequence of the Ϫ86/Ϫ44 region of the UCP2 promoter of mouse revealed that this region has no significant homology with any known PPAR␥ response element (see Fig. 4). Moreover, as shown in Fig. 3, PPAR␥ does not bind to the Ϫ86/Ϫ44 fragment in HIB-1B nuclear extracts, and this region does not compete for PPAR␥ binding versus a bona fide PPRE from the acyl-CoA oxidase promoter (AGGACAAAGGTCA) (28). The inability of this Ϫ86/ Ϫ44 UCP2 region to bind PPAR␥ indicates that stimulation of the UCP2 promoter by PPAR␥ must be indirect, utilizing other transcription factors or/and cofactors interacting with this region. As Fig. 4 shows, the Ϫ86/Ϫ44 region of the mouse UCP2 promoter contains several putative regulatory elements that are also found in active promoter regions of other genes involved in energy metabolism. These elements include Sp1 (Ϫ86/CTCCGCCTC/Ϫ76), sterol response element (SRE) (Ϫ74/ TCACGCCAC/Ϫ66), and a double E-Box-like motif (CACGCC) separated by 5 bases found in several genes critical for energy metabolism (reviewed in Ref. 29) and that can also be recognized by SRE-binding proteins (SREBPs) (30).
To determine whether these putative elements were capable of binding transcription factors we performed a series of gel shift assay experiments with in vitro-translated transcription factors or nuclear extracts. The results in Fig. 5A show that the Ϫ84/Ϫ76 element has transcription factor binding activity similar to an Sp1 consensus element. An oligonucleotide containing this region (Ϫ86/GGCTCCGCCTCGTCACG/Ϫ70) is able to compete efficiently with an Sp1 consensus oligonucleotide (CGCCCCGCCCCGATCGA), whereas mutation of the core region (Ϫ86/GGCTCTACCTCGTCACG/Ϫ70) disrupts this binding. Fig. 5B shows that antibodies against Sp1, Sp2, and Sp3 are able to supershift bound proteins, and a mixture of all three antisera resulted in complete supershift of this major binding species.
The transcription factor ADD1/SREBP1 has been shown to have the unique ability to bind two distinct regulatory elements, SRE and E-Box (31). The SRE (consensus sequence, 5Ј-TCACNCCAC-3Ј; reviewed in Ref. 30) is found in the promoter regions of several genes involved in lipid metabolism, including fatty acid synthase and the low density lipoprotein receptor (see Fig. 4). The Ϫ78/Ϫ62 region of the mouse UCP2 promoter, containing a putative SRE (Ϫ74/TCACGCCAC/Ϫ66), is able to specifically bind ADD1/SREBP1 as confirmed by the ability of anti-SREBP1 antibody to supershift the binding species (Fig. 6B). Surprisingly, the region Ϫ71/Ϫ44 containing two E-Box motifs does not show significant affinity for ADD1/ SREBP1. Several studies have established that double E-Box elements are associated with the so-called insulin/glucose response region of such genes as spot 14 (S14) and liver-specific pyruvate kinase and are recognized by upstream stimulatory factor (USF) family transcription factors (29). As shown in Fig.  7, an oligonucleotide comprising this region of the UCP2 promoter (Ϫ71/Ϫ44) specifically binds in vitro-translated USF1 and USF2 transcription factors and binds USF1 protein in the nuclear extracts from HIB-1B cells.
Having shown that these three motifs are capable of binding their factors in gel shift experiments, we next proceeded to determine which of these regions might be important for transcriptional activity of the UCP2 enhancer. Therefore, we introduced further deletions, as well as point mutations, in the Ϫ86/Ϫ44 region of the UCP2 promoter and used these constructs in transient transfection assays. Table II depicts the specific mutations made, and Fig. 8A presents these schematically. When transfected into all cell lines, deletion of the sequence between positions Ϫ86 and Ϫ76, containing the Sp1 motif, or a point mutation of two bases in this region (Ϫ86M1) reduces activity of the reporter gene by 70% (Fig. 8B). Further deletion of five more bases to position Ϫ71 disrupts SRE located between Ϫ76 and Ϫ67, but this mutation did not affect transcriptional activity of the UCP2 promoter. The region between Ϫ71 and Ϫ44 contains a direct repeat of two imperfect E-Boxes separated by five bases (see Fig. 4). Fig. 8 shows that point mutations in either the upstream or downstream E-Box motifs result in a significant loss of transcriptional activity. Complete deletion of the upstream E-Box to position Ϫ58 reduced activity to the level of the minimal UCP2 promoter (Ϫ44/ϩ12) indicating that both E-Boxes are necessary for promoter activity. Interestingly, mutations replacing two bases in the spacer region between the E-Boxes do not affect transcriptional activity. However, insertion of three additional bases between E-Boxes again reduces promoter activity, highlighting the importance of the precise positioning of the two parts of the repeat relative to each other.
Given our results in Table I that PPAR␥ responsiveness is retained within this small proximal promoter region, and the complex multipartite character of this enhancer, we next sought to narrow down which of these sequences might be important for PPAR␥-dependent stimulation. We again used our series of point mutations and deletions of the Ϫ86/Ϫ44 region. Fig. 9 demonstrates that despite deletion of the Ϫ86/ Ϫ72 region, PPAR␥-dependent stimulation was retained, indicating that PPAR␥ acts through elements in the Ϫ71/Ϫ44 region. However, point mutations disrupting either of the E-Boxes (mutants M1 and M2) led to a complete loss of UCP2 promoter responsiveness to PPAR␥. Identical results were obtained when the activity of these deletions was examined in the absence/presence of BRL49653 (data not shown). Thus, from these data we conclude that stimulation of the UCP2 gene by PPAR␥ requires transcription factors or coactivators interacting with the double E-Box-containing region. DISCUSSION PPAR␥ is best known as a key transcriptional regulator of adipocyte differentiation, stimulating transcription of many genes involved in glucose and lipid metabolism. Adipocyte differentiation is considered to be regulated by the interplay of   Fig. 2 with deletion mutant constructs of the UCP2 promoter. PPAR␥-dependent activation was tested by treatment with 1 M BRL49653 for 24 h (BRL) or cotransfection with expression plasmid for PPAR␥ (PPAR␥). Results are shown as -fold increase over unstimulated basal activity. three families of transcription factors, PPAR␥, C/EBP␣, and ADD1/SREBP1 (32), and like most tissues (33) it is characterized by a cascade activation of genes that includes other transcription factors and their targets. However, the powerful role of PPAR␥ to regulate other transcription factors is highlighted by the fact that PPAR␥-specific agonists alone can induce adipose differentiation and the expression of C/EBPs even in fibroblasts ectopically expressing PPAR␥ (34,35).
Our results demonstrate that PPAR␥ activation leads to increased UCP2 expression, and we defined the Ϫ86/Ϫ44 region of the promoter as responsible for stimulation by PPAR␥. However, as we show, PPAR␥ does not bind to this Ϫ86/Ϫ44 region, suggesting that its role is indirect and likely involves other transcription factors. Parenthetically, PPAR␥ lacking its DNA-binding domain was unable to stimulate UCP2 promoter activity (data not shown), indicating that at some level DNA binding of PPAR␥ is required. Instead, the Ϫ86/Ϫ44 region is composed of three overlapping putative elements, Sp1, SRE, and a double E-Box. Each of these sequences is capable of binding multiple proteins in nuclear extracts, including Sp1 members, SREBP, and the USFs, respectively.
Our point mutation analysis showed that the E-Box motif is required for PPAR␥ responsiveness. The ability of USF1 and USF2 to bind to the E-Box-containing region (Ϫ71/Ϫ44) in our experiments makes these transcription factors likely candidates for mediating the effect of PPAR␥ on the UCP2 gene.  Table II) used for transfection experiments in Fig. 8 However, we cannot rule out a role for SREBP because of the known dual specificity of SREBP to interact with both the SRE and E-Boxes (30). Another possibility that we must consider in future work is that the SRE frequently requires the presence of other active elements such as Sp1 or NF-Y in close proximity. For example, mutation of the Sp1 element, positioned next to the SRE site in the low density lipoprotein receptor promoter, abolished the response of this gene to sterol regulation (36,37). It has also been suggested that the double E-Box element of the mouse S14 gene is regulated by glucose through as yet unidentified factor(s) that were clearly not USFs (38).
Finally, coactivators must also be considered as candidate mediators of the PPAR␥ response on the UCP2 gene. Wu et al. (41) showed that over-expression of PGC1 led to increased expression of endogenous UCP2, along with increased oxygen consumption and uncoupled respiration. In the cells we have studied, expression of PGC1 is rapidly increased in response to PPAR␥ stimulation. Our preliminary studies, together with the data of Monsalve et al. (39), indicate that PGC1 increases expression of the UCP2 promoter. Considering the ability of PGC-1 to interact with a number of other nuclear receptors and coactivators (40,41) it is possible that PGC1, together with (an)other transcription factor(s), serves to mediate the PPAR␥ response on the UCP2 promoter. All of these possibilities remain to be explored.