CCAAT/Enhancer-binding Protein α Is Required for Transcription of the β3-Adrenergic Receptor Gene during Adipogenesis*

The β3-adrenergic receptor (β3AR) is expressed predominantly in adipocytes, and it plays a major role in regulating lipolysis and adaptive thermogenesis. Its expression in a variety of adipocyte cell models is preceded by the appearance of CCAAT/enhancer-binding protein α (C/EBPα), which has been shown to regulate a number of other adipocyte-specific genes. Importantly, it has been demonstrated that several adipocyte cell lines that fail to express C/EBPα exhibit reduced insulin sensitivity, despite an apparent adipogenic phenotype. Here we show that transcription and function of the β3AR correlates with C/EBPα expression in these adipocyte models. A 5.13-kilobase pair fragment of the mouse β3AR promoter was isolated and sequenced. This fragment conferred a 50-fold increase in luciferase reporter gene expression in adipocytes. Two putative C/EBP binding sites exist at −3306 to −3298 and at −1462 to −1454, but only the more distal site is functional. Oligonucleotides corresponding to both the wild-type and mutated −3306 element were inserted upstream of a thymidine kinase luciferase construct. When cotransfected in fibroblasts with a C/EBPα expression vector, reporter gene expression increased 3-fold only in the wild-type constructs. The same mutation, when placed into the intact 5.13-kilobase pair promoter, reduced promoter activity in adipocytes from 50-fold to <10-fold. Electrophoretic mobility shift analysis demonstrated that the site at −3306 generated a specific protein-oligonucleotide complex that was supershifted by C/EBPα antibody, while a probe corresponding to a putative site at −1462 did not. These results define C/EBPα as a key transcriptional regulator of the mouse β3AR gene during adipogenesis.

the ␤AR family because, unlike the ␤ 1 AR and ␤ 2 AR, it is expressed predominantly in adipocytes and regulates both lipolysis and nonshivering thermogenesis (reviewed in Ref. 1). In genetic and dietary models of obesity, progressive accumulation of adipose tissue is associated with defects in the ability of catecholamines to mobilize lipid stores (2)(3)(4). We have previously shown that the expression and function of the adipocyte ␤ARs are blunted in most models of obesity (5,6). Nevertheless, a curious aspect of ␤ 3 AR biology is that, despite defects in ␤ 3 AR expression and function, selective agonists for this receptor have been shown to prevent or reverse obesity (4,(7)(8)(9)(10). The efficacy of these drugs is related to increased brown adipose tissue thermogenesis and a restoration of expression of the ␤ 3 AR and ␤ 1 AR in white adipose tissue depots (4). For these reasons it is important to understand the tissue-specific and hormonal factors that regulate the expression of this receptor.
Two groups of transcription factors are known to be responsible for initiating and maintaining adipocyte differentiation: the CCAAT/enhancer-binding proteins (C/EBP) (11)(12)(13)(14)(15)(16)(17) and PPAR␥ (18 -20). From a large body of work in model adipocyte cell lines, such as 3T3-L1, it has been shown that the C/EBPs are expressed in a cascade-like fashion during the early stages of adipocyte differentiation, with C/EBP␤ and C/EBP␦ preceding the appearance of C/EBP␣ (15,21). More recent studies indicate that the expression of the adipogenic transcription factor PPAR␥ is partially under the control of the C/EBP family of transcription factors and vice versa (22,23). Additionally, insulin-sensitive glucose uptake has been shown to be impaired in adipogenic cells that lack C/EBP␣, due to deficits in insulin receptor and insulin receptor substrate-1 (22,24). Like other adipocyte-specific genes, the ␤ 3 AR is not expressed in preadipocytes, but appears late in the adipogenic program of both white and brown adipocytes (Refs. 25 and 26; this report). Because of the pivotal role of C/EBP␣ in activating many adipocyte-specific genes, the focus of our studies was to determine the role of C/EBP␣ in initiating ␤ 3 AR gene transcription during adipogenesis. Our results show that C/EBP␣ is required for the adipocyte-dependent expression of the mouse ␤ 3 AR gene, and we define the C/EBP binding site in the ␤ 3 AR promoter that is responsible for this regulation.
Isolation and Sequencing of Genomic Clones and Construction of * This work was supported by National Institutes of Health Grants R01DK46793 (to S. C.) and R01DK53092 (to S. C.) and a National Institutes of Health minority predoctoral award (to T. M. D.). 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 /EBI Data Bank with accession number(s) AF303739.
ʈ To whom correspondence should be addressed: Duke University Medical Center, Box 3557, Durham, NC 27710. Tel.: 919-684-8991; Fax: 919-684-3071; E-mail: colli008@mc.duke.edu. 1 The abbreviations used are: AR, adrenergic receptor; C/EBP, CCAAT/enhancer-binding protein; PPAR, peroxisome proliferator-activated receptor; aP2, adipocyte fatty acid-binding protein; bp, base pair(s); kb, kilobase pair(s); DMEM, Dulbecco's modified Eagle's medium; SFM, serum-free medium; PPRE, peroxisome proliferator-activated receptor ␥ response element. ␤ 3 AR Promoter-Luciferase Plasmids-A mouse genomic DNA library constructed in -DASH (Stratagene) was screened with a mouse ␤ 3 AR probe comprising the first 108 amino acids of the protein (27). Eight independent positive clones were isolated and analyzed by restriction enzyme mapping and Southern blot hybridization with the 108 aminoterminal probe and a with second probe corresponding to the first 600 nucleotides 5Ј to the initiator methionine. A 6641-nucleotide BamHI fragment was subcloned into the pGEM-4Z plasmid, and two isolates containing the insert in opposite orientations were sequenced along both strands by the Duke University Automated DNA Sequencing Facility. The sequence of the 5283-nucleotide BamHI-NarI fragment can be retrieved as GenBank accession no. AF303739. This BamHI-NarI fragment was subcloned into the luciferase reporter construct pGL3Basic (Promega) at the BglII/HindIII sites to generate m␤ 3 -Luc. A series of deletion mutants was created by digesting the m␤ 3 -Luc with the following enzymes to yield the indicated fragments: NheI and SpeI (3288 base pairs), KpnI (2079 base pairs), EcoRV and MluI (1107 base pairs), or MluI and BglII (557 base pairs). A site-directed mutation of the putative C/EBP response element at Ϫ3306 to Ϫ3298 was generated using the GeneEditor kit (Promega) such that the sequence TGGAG-CAAT was changed to GACTAGCCT. The C/EBP␣ expression vector has been described previously (28). Plasmids for transfections were purified using the Promega Megaprep system.
Cell Culture and Transfections-COS-7 and C3H10T1/2 (T1/2) clone 8 fibroblasts (American Type Culture Collection) were maintained in DMEM ϩ 10% fetal bovine serum. 3T3-L1 and NIH-3T3 cells were cultured in DMEM ϩ 10% calf serum. The NIH-C/EBP␣, Swiss-PPAR␥, and NIH-PPAR␥ cell lines were maintained and differentiated as described (13,14,24). T1/2 cells were seeded into six-well dishes and differentiation proceeded as described (29). For most cell lines, 2 g of DNA were transfected per well using 3 l of FuGENE 6 as outlined by the supplier (Roche Molecular Biochemicals). For preparation of nuclear extracts enriched in C/EBP␣, 20 g of C/EBP␣ expression vector were transfected into 10-cm dishes of COS-7 cells by calcium phosphate coprecipitation (30). The 3T3-L1 cells were cotransfected with 4 g of reporter vectors, 4 g of either the murine sarcoma virus-C/EBP␣ expression vector or the empty murine sarcoma virus expression vector, and 2 g of cytomegalovirus-␤-galactosidase using calcium phosphate coprecipitation. T1/2 cells were harvested for luciferase activity 72 h after transfection, while all others were harvested after 48 h.
Enzymatic Assays-Luciferase activity was determined in a TD 20/20 luminometer (Promega) using the luciferase assay kit (Promega). ␤-Galactosidase activity was determined by a colorimetric assay (absorbance at 570 nm) using chlorophenol red-␤-D-galactopyranoside as the substrate. Luciferase data were normalized by dividing the light units by ␤-galactosidase activity.
Oligonucleotides-Oligonucleotides used in this study were synthesized by Life Technologies, Inc. Nucleotide sequences are listed in Table  I.
Preparation of Nuclear Extracts and Electrophoretic Mobility Shift Assays-Nuclear extracts were prepared following the method of Schreiber (31). Protein determinations were made by the Bradford method (32). Double-stranded oligonucleotide probes were annealed and end-labeled with [␥-32 P]ATP (111TBq/mmol) by T4 polynucleotide kinase. Nuclear extracts (10 g) were incubated with labeled oligonucleotide probes for 30 min on ice before loading on pre-run polyacrylamide gels (Invitrogen) in 0.5ϫ TBE (44.5 mM Tris borate, 1 mM EDTA, pH 8.0). For competition studies, unlabeled nucleotides were incubated with nuclear extracts on ice for 15 min prior to addition of the labeled oligonucleotides. Rat liver nuclear extract was obtained from Geneka Biotechnology (Montréal, Quebec, Canada) and gel shifts were performed per manufacturer's instructions, using 2 g of extract/reaction. Reactions were resolved in 1ϫ TGE (50 mM Tris, 380 mM glycine, and 2 mM EDTA). Where indicated, antibodies used in supershift analysis that are specific to C/EBP␣ (sc-61x), C/EBP␤ (sc-150x), and C/EBP␦ (sc-151x) were obtained from Santa Cruz Biotechnology.
Isolation and Analysis of RNA-Total cellular RNA was prepared using TRI reagent according to the manufacturer's specifications (Molecular Research Center). RNA (40 g) was denatured by the glyoxal procedure, fractionated through 1.2% agarose gels, and blotted onto Biotrans (ICN) membranes as detailed previously (33). Radiolabeled probes were prepared by random primer synthesis (PrimeIT, Stratagene) of the purified DNA fragments in the presence of [␣-32 P]deoxy CTP (111TBq/mmol) 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 mouse ␤ 3 AR was prepared as described previously (5). Mouse PPAR␥ cDNA was a gift from Jurgenn Lehmann. The aP2 probe was a gift from Bruce Spiegelman (34). A cDNA fragment for rat C/EBP␣ was described previously (13). A rat cDNA probe for cyclophilin was used as an internal hybridization/ quantitation standard as described previously (35). Blots were hybridized and washed as described previously (5,36).
Adenylyl Cyclase Assay-Cells were grown in 12-well plates for cAMP assays according to published methods (37). Briefly, cells were preincubated in serum-free DMEM, 25 mM HEPES, pH 7.5 (SFM) for 20 min, followed by fresh SFM containing 0.25 mM isobutylmethylxanthine for 5 min. CL316,243 (5 M final) was added, and the plates were returned to the incubator for 20 min (determined to be an optimal time in pilot experiments). Treatment was terminated by rapidly aspirating the medium and adding cold 5% trichloroacetic acid (100 l/well). Two hundred l of 50 mM KPO 4 , pH 7.4, was added to each well to partially neutralize the trichloroacetic acid. Cyclic AMP in the cell lysate was determined by radioimmunoassay (38) using a polyclonal antiserum to iodinated cAMP (39).

RESULTS
Differentiation-dependent Expression of the ␤ 3 AR Gene-T1/2 fibroblasts have been shown to differentiate into white adipocytes in response to PPAR␥ ligands, and these cells can serve as a vehicle for examining the expression of transfected genes (29), but the expression of ␤ 3 AR in these cells as a function of differentiation was unknown. Fig. 1 presents the pattern and time course of expression of several adipogenic genes, including ␤ 3 AR, in these cells during differentiation. Total RNA was isolated every other day from day 2 after ligand treatment until day 12 and analyzed by Northern blotting (Fig.  1A). A progressive increase in the expression of C/EBP␣, PPAR␥, and aP2 preceded detectable levels of ␤ 3 AR transcripts. The internal control transcript, cyclophilin, was unchanged throughout this period. Fig. 1B depicts the concomitant increase in the ability of a selective ␤ 3 AR agonist to stimulate adenylyl cyclase activity during differentiation of the T1/2 cells. By day 3 of differentiation, there is a 2-fold increase in ␤ 3 AR-stimulated adenylyl cyclase activity. However, in subsequent days, there is a substantial increase in cAMP production such that, by day 10, ␤ 3 AR-stimulated cAMP production is enhanced 25-fold over basal.
C/EBP␣ Is Necessary for ␤ 3 AR Expression in Adipocytes-Previous studies have shown that ectopic expression of either PPAR␥ or C/EBP␣ in nonadipogenic fibroblasts stimulates adipogenesis (11,20). Coexpression of the two proteins (22) re- sults in a synergism of adipogenesis that is comparable to that seen in the well-studied 3T3-L1 model (20). Fig. 2 compares the expression of ␤ 3 AR and other relevant adipocyte genes in a collection of cell lines that either possess de novo adipogenic capacity in response to hormonal induction or were engineered to express various combinations of C/EBPs and PPAR␥ to assess their contribution to the adipocyte phenotype. ␤/␦39 cells are NIH-3T3 cells that were designed to express both C/EBP␤ and C/EBP␦ under a tetracycline-responsive promoter to ascertain the role of glucocorticoids and C/EBP␦ in the initiation of adipogenesis (13,14). When these cells are induced to differentiate, they accumulate lipid droplets and they express both PPAR␥ and aP2 (Fig. 2). However, despite obvious adipocyte morphology, these cells fail to express either C/EBP␣ or ␤ 3 AR.
In two other adipocyte models, ectopic retroviral expression of PPAR␥ induces C/EBP␣ and ␤ 3 AR expression in the Swiss-3T3 cells (24), but not in NIH-3T3 cells (20), again despite lipid accumulation and expression of aP2. Finally, we examined ␤ 3 AR expression in another cell line, NIH-␣, which ectopically express C/EBP␣ in the NIH-3T3 fibroblast background (24). As shown in the right-hand side of Fig. 2, when these cells are induced to differentiate they express ␤ 3 AR mRNA, in addition to PPAR␥ and aP2, as early as day 6.
In support of these mRNA expression studies, we assessed the functional activity of the ␤ 3 AR in these various adipocyte cell lines by the ability of the ␤ 3 AR-specific agonist, CL316,243, to stimulate adenylyl cyclase activity. Table II shows that there is no detectable increase in ␤ 3 AR stimulated cyclase activity in undifferentiated versus differentiated NIH-PPAR␥ cells, whereas there was a 5-fold increase in cAMP production in the NIH-␣ cells when comparing the differentiated and undifferentiated states. Together, these data strongly suggest that C/EBP␣ is necessary for ␤ 3 AR expression and function in adipocytes.
Structure and Activity of the Mouse ␤ 3 AR Promoter.-We sequenced a 5.28-kb BamHI-Nar I fragment containing 150 nucleotides of exon 1 and 5.13 kb of sequence 5Ј to the transcription start site (40) of the mouse ␤ 3 AR gene. Sequence analysis with SIGNALSCAN (GCG, University of Wisconsin, Madison, WI) revealed a number of putative transcription factor binding sites within the 5.13-kb promoter. As outlined in Fig. 3, two putative C/EBP protein binding sites between Ϫ3306 and Ϫ3298 and between Ϫ1462 and Ϫ1354 were identified. Several other putative regulatory elements, including glucocorticoid and AP-1 binding sites, were also identified (see GenBank accession no. AF303739).
In experiments designed to evaluate the tissue-specificity of the ␤ 3 AR promoter, the promoter region was subcloned into the pGL3-Basic luciferase vector (Promega). This construct was then transfected into differentiating T1/2 adipocytes or proliferating NIH-3T3 cells. Relative to the promoter-less parent vector, the 5.13-kb ␤ 3 AR fragment stimulated luciferase expression 50-fold in the adipocytes (Fig. 4A). By contrast, in NIH-3T3 cells, the same construct induced luciferase expres-FIG. 2. Expression of adipogenic transcripts in differentiating adipocyte cell line models. Forty g of total cellular RNA were isolated on the indicated days from either the ␤/␦39, NIH-P␥, or the Swiss-P␥ cells and were fractionated through 1.2% agarose gels and blotted as described under "Experimental Procedures." The blot was hybridized with ␣-32 P-labeled ␤ 3 AR, C/EBP␣, PPAR␥, aP2, and cyclophilin probes.  1. Differentiation-dependent expression of ␤ 3 AR-and agonist-stimulated adenylyl cyclase activity in T1/2 adipocytes. A, Forty g of total cellular RNA from differentiating T1/2 adipocytes were fractionated through 1.2% agarose gels and blotted as described under "Experimental Procedures." The blot was hybridized with ␣-32 P-labeled probes for ␤ 3 AR, C/EBP␣, PPAR␥, aP2, and cyclophilin. B, fully differentiated cells were incubated with the ␤ 3 AR-selective agonist CL316,243 (5 M) for 20 min, and the cAMP produced was measured by radioimmunoassay as described under "Experimental Procedures." The data are expressed as picomoles of cAMP produced/well/min. sion Ͻ2-fold. These results are consistent with the observation that ␤ 3 AR mRNA is primarily observed in adipocytes. To determine the importance of the putative C/EBP regions for the activity of the mouse ␤ 3 AR promoter, we evaluated the activity of several 5Ј-deletion constructs (Fig. 4B). The activity of the Ϫ3138 promoter truncation, which lacks the proposed C/EBP site at Ϫ3306, is decreased to less than 50% of the Ϫ5.13 kb promoter fragment. The Ϫ1929 bp deletion led to a further 30% decrease of luciferase activity, while deletion of the second putative C/EBP site at Ϫ1462, shown by the Ϫ957 bp truncation, had no further effect. These data support the hypothesis that C/EBP␣ is necessary for both the expression and function of the ␤ 3 AR in these adipocyte models.
C/EBP␣ Binds the ␤ 3 AR Ϫ3306 Element-To establish whether either of the putative C/EBP binding sites was capable of binding C/EBP proteins, we performed a series of gel shift and antibody supershift assays. For these experiments, nuclear extracts were prepared from COS-7 cells that had been trans-fected with a C/EBP␣ expression vector, as described previously (41). The oligonucleotides used in these experiments are shown in Table I. Using a consensus C/EBP oligonucleotide as the probe resulted in a major binding species (filled arrow) that was supershifted (open arrow) in the presence of antisera to C/EBP␣ (Fig. 5A, lanes 2 and 3). Addition of a 100-fold molar excess of the unlabeled ␤ 3 AR Ϫ3306 eliminated this band (lane 4). When the ␤ 3 AR Ϫ3306 element was used as a probe (Fig.  5A, lanes 5-8), several binding species were detected, with one major band that comigrated with the major C/EBP band in lane 2. This major band was completely supershifted (open arrow) by anti-C/EBP␣ antibody (lane 5 versus lane 6), but was eliminated when excess unlabeled oligonucleotide for a consensus C/EBP binding site (42) was included as a competitor (lane 7). In contrast, an oligonucleotide with mutations in the Ϫ3306 C/EBP was unable to affect the gel shift pattern of the labeled Ϫ3306 oligonucleotide (lane 8). In Fig. 5B, we again used the C/EBP consensus sequence (lanes 1-4) and the Ϫ3306 ␤ 3 AR element (lanes 5-8) as probes to further examine the specificity of binding. The major band (filled arrow) was eliminated in the presence of 100-fold molar excess of the Ϫ3306 ␤ 3 AR C/EBP element, but neither an excess of the Ϫ1462 C/EBP element nor of a Sp1 consensus affected binding. In lanes 5-9, the Ϫ3306 C/EBP element was used as a probe. A 100-fold molar excess of the C/EBP consensus oligonucleotide inhibited binding of the major band. The addition of the unlabeled C/EBP element at Ϫ1462 was able to partially inhibit binding of the Ϫ3306 C/EBP element while the addition of unlabeled Sp1 oligonucleotide or the mutant Ϫ3306 oligonucleotide had no effect. The relative affinity of C/EBP␣ for the Ϫ3306 C/EBP element versus the consensus sequence was assessed by including 10-, 50-, or 100-fold molar excess of unlabeled oligonucleotides (Fig. 5C). tide (lanes 2-4), while the mutant Ϫ3306 had no effect (lanes 5-7). As anticipated, addition of the unlabeled consensus C/EBP sequence completely blocked the appearance of the major band (lanes 8 -10). Finally, as shown in Fig. 5D, the oligonucleotide corresponding to the putative C/EBP site at Ϫ1462 produced gel shift bands with nuclear extracts, but these bands do not appear to bind C/EBP␣. First, the relative position of these faint bands did not correspond to C/EBP binding as observed for the Ϫ3306 sequence and identical to the consensus, nor was the band pattern affected when either the consensus or the Ϫ3306 C/EBP were used as competitors. More importantly, the addition of anti-C/EBP␣ antibody did not affect the abundance or position of these bands.
Finally, we wanted to determine whether binding of the Ϫ3306 element was specific for C/EBP␣. To address this, we performed gel shifts using rat liver nuclear extract, which contains multiple C/EBP isoforms, and antibodies specific for C/EBP␣, C/EBP␤, and C/EBP␦. As shown in Fig. 6, the Ϫ3306 C/EBP sequence produced strong gel shift bands when incubated with the rat liver extract (lane 2), while a consensus C/EBP oligonucleotide inhibited the appearance of these bands (lane 3). A C/EBP mutant had no effect (lane 4) (for sequences see Table I). The major band was nearly completely supershifted by antisera to C/EBP␣ (lanes 5, 8, and 9). Addition of anti-C/EBP␤ supershifted a lower, weaker band (lanes 6, 8, and  10), and anti-C/EBP␦ had no effect (lanes 7, 9, and 10). For comparison, the same consensus C/EBP sequence used in Fig.  5 (see Table I) was used as the probe in lanes 11 and 12. The major band was supershifted by the addition of C/EBP␣ antibody. From these data, we conclude that C/EBP␣ binds the Ϫ3306 C/EBP element within the mouse ␤ 3 AR promoter.
The ␤ 3 AR C/EBP␣ Element Confers Activity to a Heterologous Promoter-Having identified a potential role for C/EBP␣ in the regulation of the ␤ 3 AR by correlative expression and gel FIG. 6. The ؊3306 C/EBP site binds proteins present in rat liver nuclear extract. 32 P-Labeled oligonucleotides corresponding to either the Ϫ3306 C/EBP in the mouse ␤ 3 AR promoter or a C/EBP consensus were incubated with 2 g of nuclear protein. For competition assays, 100-fold molar excess of unlabeled oligonucleotide of a consensus C/EBP binding site or a mutant C/EBP site were used. In supershift assays 2 l of the indicated antibodies were added; in the lanes where there are two antibodies present, 1 l of each antibody was used.
FIG. 5. Electrophoretic mobility shift analysis with probes corresponding to putative C/EBP sites within the mouse ␤ 3 AR promoter. A and B, 32 P-labeled oligonucleotides corresponding to residues Ϫ3306 to Ϫ3298 of the mouse ␤ 3 AR promoter or a consensus C/EBP were incubated with nuclear extract from COS-7 cells transfected with a C/EBP␣ expression vector. Ten g of nuclear extract was used per sample. For competition assays, 100-fold molar excess of unlabeled oligonucleotide was used. Supershifting was performed by adding 1 l of C/EBP␣ antibody. C, a 32 P-labeled oligonucleotide corresponding to residues Ϫ3306 to Ϫ3298 was incubated with nuclear extract from COS-7 cells transfected with a C/EBP␣ expression vector. For competition assays, 10-, 50-, or 100-fold molar excess of unlabeled oligonucleotide was used. D, 32 P-labeled oligonucleotides corresponding to residues Ϫ1462 to Ϫ1454 or residues Ϫ3306 to Ϫ3298 of the mouse ␤ 3 AR promoter were incubated with nuclear extract from COS-7 cells transfected with a C/EBP␣ expression vector. For competition assays, 100-fold molar excess of unlabeled oligonucleotides were used; 1 l of C/EBP␣ antibody was used for supershift analysis.
shift analysis, we sought to determine whether the putative C/EBP at Ϫ3306 was functional. Two approaches were taken. First, wild-type and mutated versions of the Ϫ3306 C/EBP␣ element were transfected in the presence or absence of a C/EBP␣ expression vector. Fig. 7 shows that the insertion of one copy of the putative C/EBP␣ element results in a 3-fold increase in luciferase activity, while the presence of two tandem C/EBP elements results in a 9-fold enhancement. In contrast, the mutant Ϫ3306 C/EBP element abolished transactivation by C/EBP␣.
Next we mutated the C/EBP site at Ϫ3306 to Ϫ3298 from TGGAGCAAT to GACTAGCCT within the context of the intact 5.13-kb ␤ 3 AR promoter. This mutation is the same as the one used in the C/EBP␣ transactivation experiments in Fig. 7, as well as the gel shift assays in Fig. 5. As shown in Fig. 8, when the ␤ 3 AR promoter constructs containing wild-type and mutant C/EBP sites were introduced into differentiating T1/2 cells, luciferase activity from the mutant was reduced by more than half, to a level equivalent to that observed in the Ϫ2852 deletion (Fig. 4B). These data demonstrate that this site mediates the effect of C/EBP␣ on ␤ 3 AR gene expression. DISCUSSION In this study we have shown that the mouse ␤ 3 AR gene is specifically activated by C/EBP␣ and that this activation is correlated with the binding of C/EBP␣ to an element residing between Ϫ3306 and Ϫ3298 bp upstream of the ␤ 3 AR gene transcription start site. By utilizing several cell lines that contain various combinations of the C/EBPs and PPAR␥, we showed that only adipocytes expressing C/EBP␣ possess ␤ 3 AR transcripts and concomitant functional activity. A 5.13-kb promoter fragment of the mouse ␤ 3 AR gene containing two putative C/EBP binding sites, at Ϫ3306 to Ϫ3298 and at Ϫ1462 to Ϫ1454, confers robust expression of a luciferase reporter preferentially in adipocytes. This transcriptional activity is significantly decreased upon deletion of the more distal C/EBP element, while removal of the more proximal element had no further effect. We also showed that this C/EBP element at Ϫ3306 conveys transcriptional activity to C/EBP␣ in vitro. Finally, electrophoretic mobility shift assays provided evidence that C/EBP␣ interacts specifically with the element at Ϫ3306 and not the element at Ϫ1462 in the mouse ␤ 3 AR promoter. In every case, mutation of the Ϫ3306 C/EBP site eliminates these responses.
Our observation that expression of the ␤ 3 AR gene is positively regulated by C/EBP␣ is consistent with numerous reports showing the role of this transcription factor in adipogenesis. For example, ectopic expression of C/EBP␣ is sufficient to induce adipocyte differentiation in a number of cell lines, while the expression of an antisense C/EBP␣ construct in 3T3-L1 adipocytes blocks differentiation (43). Consistent with these in vitro studies, C/EBP␣ null mice fail to develop white adipose tissue (44).
In addition to C/EBP␣, C/EBP␤ and C/EBP␦ have also been shown to be critical regulators of adipocyte differentiation. C/EBP␤ and C/EBP␦ are transiently expressed and precede the appearance of C/EBP␣ (12). Interestingly, overexpression of C/EBP␤, but not C/EBP␦, in preadipocytes converts them to adipocytes, suggesting that C/EBP␤ can substitute for C/EBP␣ (15). However, the ␤/␦39 cells, which constitutively express C/EBP␤ and C/EBP␦, acquire an adipocyte phenotype, as evidenced by the presence of PPAR␥ and aP2, but fail to express either C/EBP␣ or ␤ 3 AR. As shown in our gel shift assays, it appears that C/EBP␤ is capable of binding the C/EBP element at Ϫ3306 in the mouse ␤ 3 AR promoter. Perhaps this is not surprising because C/EBP isoforms have been shown to bind to a common DNA consensus sequence (21,41). Despite this interaction, it is clear that C/EBP␤ or C/EBP␦ alone is insufficient to induce ␤ 3 AR.
Although we show a pivotal role for C/EBP␣ in transactivating the ␤ 3 AR gene during adipogenesis, tissues containing high levels of C/EBP␣, such as liver and lung, do not contain appreciable amounts of ␤ 3 AR. In addition, we show that the NIH-␣ cells do not express ␤ 3 AR in the preadipocyte state, despite the constitutive expression of C/EBP␣. Therefore, it is clear that some other transcription factor(s) are important in directing the adipocyte-specific expression of the ␤ 3 AR gene. Another key regulator of adipogenesis is PPAR␥. Constitutive expression of PPAR␥ in fibroblasts can induce the conversion to the adipocyte phenotype (20). PPAR␥ ligands, such as the thiazolidinediones, which are a class of insulin-sensitizing agents, convert fibroblasts and multipotential stem cells to adipocytes (13,29). As well, PPAR␥ (Ϫ/Ϫ) animals completely lack adipose tissue (45)(46)(47). Previous studies showed that, despite expression of PPAR␥ and development of an adipocyte morphology, a lack of C/EBP␣ results in decreased insulin-stimulated glucose transport (22,24). Despite this apparent cooperation between C/EBP␣ and PPAR␥ in adipocyte differentiation, our initial FIG. 7. The ؊3306 C/EBP site in the mouse ␤ 3 AR promoter confers transcriptional responsiveness to C/EBP␣. Complementary oligonucleotides corresponding to the C/EBP region at Ϫ3306 to Ϫ3298 (and a mutant version) were annealed and ligated into a thymidine kinase-luciferase construct as described under "Experimental Procedures." Four micrograms of the constructs were transfected into proliferating 3T3-L1 cells with or without 4 g of C/EBP␣ expression vector and 2 g of ␤-galactosidase vector. Shown is a representative experiment performed in triplicate.
FIG. 8. A mutation in the ؊3306 C/EBP-like region inhibits mouse ␤ 3 AR promoter activity. Differentiating T1/2 cells were transfected with either the pGL3 Basic vector, the Ϫ5.13 kb promoter, or the mut Ϫ5.13 kb. The mut Ϫ5.13 kb contains a mutation in the C/EBP site at Ϫ3306 in context of the original Ϫ5.13 kb construct. Shown is a representative experiment performed in triplicate. attempts to locate a DR-1 PPAR␥ response element (PPRE) within our mouse ␤ 3 AR promoter fragment or to demonstrate transactivation of the ␤ 3 AR by PPAR␥ have been unsuccessful. It is plausible that a PPRE lies outside of the promoter region that we have isolated and studied. However, it is known that PPREs in several PPAR␥ target genes deviate significantly from the consensus DR-1 site (18). For this reason we are currently investigating some regions that weakly resemble a PPRE half-site. To fully understand the role of the ␤ 3 AR in obesity, it will be important for us to determine not only what other critical transcription factor(s) is/are required for the regulation of the ␤ 3 AR gene, but how their effects on ␤ 3 AR expression may contribute to obesity.