p300 coactivates the adipogenic transcription factor CCAAT/enhancer-binding protein alpha.

Despite the knowledge that CCAAT/enhancer-binding protein alpha (C/EBPalpha) plays an important role in preadipocyte differentiation, our understanding of how C/EBPalpha interacts with nuclear proteins to regulate transcription is limited. Based on the hypothesis that evolutionarily conserved regions are functionally important and likely to interact with coactivators, we compared the amino acid sequence of C/EBPalpha from different species (frog to human) and identified four highly conserved regions (CR1-CR4) within the transactivation domain. A series of amino-terminal truncations and internal deletion constructs were made creating forms of C/EBPalpha which lack single or multiple conserved regions. To determine which regions of the C/EBPalpha transactivation domain are important in its ability to induce spontaneous differentiation of 3T3-L1 preadipocytes, we infected preadipocytes with expression vectors encoding the C/EBPalpha conserved region mutants and observed their ability to induce differentiation. We found that CR2 fused to the DNA binding domain is able to induce spontaneous differentiation independent of the other conserved regions. However, CR2 was not necessary for the adipogenic action of C/EBPalpha because a combination of CR1 and CR3 can also induce adipogenesis. Because the transcriptional coactivator p300 participates in the signaling of many transcription factors to the basal transcriptional apparatus, we examined whether functional interaction exists between C/EBPalpha and p300. Cotransfection of p300 with p42C/EBPalpha results in a synergistic increase in leptin promoter activity, indicating that p300 acts as a transcriptional coactivator of C/EBPalpha. Analyses using C/EBPalpha conserved region mutants suggest that multiple regions (CR2 and CR3) of the C/EBPalpha transactivation domain functionally interact with p300.

Adipose tissue plays multiple roles in the mechanisms controlling homeostasis. Adipocytes serve not only as an energy reservoir by storing excess calories as triacylglycerol, but they also have endocrine and immune functions (1)(2)(3)(4). The importance of understanding adipocyte biology is emphasized further by the complications that arise from either too much or too little adipose tissue. Obesity and its associated disorders, such as type 2 diabetes and cardiovascular disease, are an epidemic in the developed world today (5). Conversely, lipoatrophy, the lack of adipose tissue, is also associated with diabetes and a number of other metabolic abnormalities (6). Understanding the factors that govern the transcriptional control of adipogenesis will aid in our understanding of how these disorders arise.
A number of the key factors controlling the adipocyte differentiation cascade have been identified (for review, see Refs. 3 and 7-9), including the transcription factors CCAAT/enhancer binding protein ␣ (C/EBP␣) 1 and peroxisome proliferator-activated receptor ␥ (PPAR␥). Although the structure, function, and regulation of PPAR␥ have been studied extensively (10 -13), similar analyses have yet to be performed upon C/EBP␣. Consistent with C/EBP␣ being a key component in the adipogenic cascade, mice with the c/ebp␣ gene deleted have a deficiency in lipid accumulation in both white and brown adipose tissue (14). Likewise, 3T3-L1 preadipocytes (15)(16)(17) are unable to differentiate after hormonal induction if C/EBP␣ antisense RNA is expressed in the cells (18). Moreover, the enforced expression of C/EBP␣ in 3T3-L1 fibroblasts is sufficient to induce spontaneous adipogenesis (19,20) and overcome the effects of repressors of adipogenesis such as Wnt-1 (21). C/EBP␣ is a prototypical basic region/leucine zipper (bZIP) transcription factor consisting of a well characterized carboxylterminal leucine zipper that confers dimerization ability and a neighboring DNA binding domain and nuclear localization signal, which are rich in basic residues (22)(23)(24). The remaining amino-terminal 273 amino acids, termed the transactivation domain, contains insulin-responsive sites of phosphorylation (25)(26)(27) and regions that affect the transcriptional activity of C/EBP␣ (28 -30). Previous studies have arbitrarily divided the transactivation domain and identified regulatory regions within it by examining the ability of these C/EBP␣ constructs to transactivate reporter genes under the control of the liver serum albumin promoter (29,30) or an artificial Gal4-responsive promoter (28). In contrast, the current study approaches the functional analyses of the C/EBP␣ transactivation domain by targeting regions of C/EBP␣ which are highly conserved across several species (31). We examined the role of these conserved regions (CR) in the ability of C/EBP␣ to activate the genes necessary to induce spontaneous preadipocyte differentiation. Furthermore, we demonstrate that the nuclear coactivator p300 is able to potentiate C/EBP␣-mediated transcription of the leptin (ob) promoter through multiple conserved regions within the C/EBP␣ transactivation domain.
To create retroviral expression vectors, pLXSN (34) was linearized with HindIII, blunt ends were created using DNA polymerase I, large (Klenow) fragment (New England Biolabs) and ligated with T4 DNA ligase (New England Biolabs). The resulting plasmid was digested with HpaI, and a HindIII linker (d(pCAAGCTTG), New England Biolabs) was ligated in to form pNH2. The full-length, p30, CR1, and CR1/3/4 isoforms of C/EBP␣ were each excised from pcDNA3.1(Ϫ) with EcoRI and HindIII and subcloned into the EcoRI-HindIII sites of pNH2. CR2/ 3/4 and CR2 in pcDNA3.1(ϩ) were first digested with BamHI. The resulting fragment was filled in with DNA polymerase I, large (Klenow) fragment and ligated with an EcoRI linker (d(pCGGAATTCCG), New England Biolabs). This vector was subsequently digested with EcoRI plus HindIII, and the resulting insert was subcloned into pNH2.
An expression vector encoding a chimeric protein consisting of the Gal4 DNA binding domain and the C/EBP␣ transactivation domain was constructed by subcloning a KpnI-SacI fragment of pSER23 into pSG424 (35). This construct was subsequently digested with BamHI and KpnI, and the BamHI-KpnI fragment from pSER28 was inserted.
Human embryonic kidney 293T cells (100-mm plates) were transiently transfected with 20 g of total DNA by calcium phosphate coprecipitation, including 1 g of luciferase reporter gene, 500 ng of CMV-␤-galactosidase (␤Gal), and 10 g of sheared herring sperm DNA. Additional plasmid DNA amounts varied based upon experimental conditions and are documented in the figure legends. A constant amount of CMV promoter was maintained in all conditions to control for the potential squelching of the transcriptional machinery. After precipitation for 4 -5 h, cells were shocked with 12.5% glycerol in phosphatebuffered saline (157 mM NaCl, 2.7 mM KCl, 1.5 mM KH 2 PO 4 , 5.5 mM Na 2 HPO 4 -H 2 O, pH 7.3) for 3 min. Cells were incubated for 24 -48 h in 10% calf serum and Dulbecco's modified Eagle's medium before they were lysed in 1 ϫ reporter lysis buffer (100 mM KH 2 PO 4 , 0.2% (v/v) Triton X-100, 1 mM dithiothreitol). Samples were vortexed and subsequently centrifuged in a microcentrifuge for 30 s at 16,000 ϫ g. To assay the samples for luciferase activity, 100 l of the supernatant was mixed with 360 l of luciferase buffer 2 (25 mM glycylglycine (Fisher Biotech), pH 7.8, 30 mM MgSO 4 , 4 mM EGTA, pH 8.0, 0.0027% (v/v) Triton X-100, 15 mM KH 2 PO 4 , 2 mM ATP, 1 mM dithiothreitol) in glass test tubes. The tubes were then placed in an Optocomp II luminometer (MGM Instruments, Hamden, CT) and injected with 100 l of luciferase buffer 3 (25 mM glycylglycine, 30 mM MgSO 4 , 4 mM EGTA, pH 8.0, 200 nM luciferin (Promega), 2 mM dithiothreitol), and the relative light units emitted were measured and normalized against the relative light units obtained from a chemiluminescent ␤-galactosidase assay. ␤-galactosidase activity was measured by taking 10 l of cell lysate and inoculating 100 l of ␤-galactosidase reaction buffer (100 mM KH 2 PO 4 , 1 mM MgCl 2 , 1% (v/v) Galacton (Tropix)) and incubating for 45 min to 1 h at room temperature. Chemiluminescence was measured after the injection of 100 l of light emission accelerator (10% (v/v) Emerald enhancer (Tropix) in 0.2 N NaOH (39)).

Retroviral Infection and Oil Red-O Staining of 3T3-L1 Preadipocytes
293T cells were transfected as described above with the retroviral C/EBP␣ expression vectors and the viral packaging vectors SV-E-MLVenv and SV-E-MLV (7.5 g of each) (21,34). Virus-containing medium was collected at three 12-h intervals post-transfection and at each interval passed through a 0.45-m syringe filter. 8 g/ml filter-sterilized polybrene (hexadimethrine bromide; Sigma) was added to the virus-loaded medium. This medium was then applied to preconfluent (ϳ40%) 3T3-L1 preadipocytes. The infection protocol was repeated two additional times. After the third round of infection, 3T3-L1 preadipocytes were trypsin treated and replated, on multiple plates, in Dulbecco's modified Eagle's medium supplemented with 10% calf serum and Functional Interactions between p300 and C/EBP␣ 400 g/ml Geneticin (Life Technologies, Inc.). Cells were then allowed to proliferate to confluence, at which point the cells were fed with Dulbecco's modified Eagle's medium containing only 10% calf serum. For 14 days postconfluence the cells were observed for evidence of spontaneous differentiation. To detect cytoplasmic lipid accumulation, retrovirus-infected 3T3-L1 cells were stained by Oil Red-O, essentially as outlined previously (40).

Immunoblot Analysis
Protein expression levels in samples used for reporter gene assays were determined by immunoblot analysis. For detection of p300, supernatant from cells lysed in 1 ϫ reporter lysis buffer was combined with 4 ϫ SDS loading buffer (4% (v/v) SDS, 350 mM ␤-mercaptoethanol, 240 mM Tris, pH 6.8, 40% (v/v) glycerol, 0.01% (w/v) bromphenol blue) and electrophoresed on an SDS-polyacrylamide gel (5%). Proteins were transferred to a polyvinylidene difluoride (Osmonics) membrane, and immunoblotting was performed with mouse monoclonal p300 antibody (BD PharMingen). To determine the expression of C/EBP␣, pellets from reporter gene samples were transferred to new tubes, resuspended in Western lysis buffer (1% SDS, 60 mM Tris, pH 6.8), and sonicated. Lysate was mixed with 4 ϫ SDS loading buffer and separated on an SDS-polyacrylamide gel (11.5%), transferred to polyvinylidene difluoride, and immunoblotted with an affinity-purified polyclonal C/EBP␣ antibody generated against a synthetic polypeptide corresponding to amino acids 253-265 (32). Immunoblot analyses of proteins from infected 3T3-L1 preadipocytes and adipocytes were performed as described previously (18,26). The adipocyte marker 422/aP2 was detected using a polyclonal antibody received from Dr. David Bernlohr (University of Minnesota).

Multiple Conserved Regions Contribute to the Adipogenic
Action of C/EBP␣-Conserved regions of the C/EBP␣ transactivation domain were identified by aligning the primary amino acid sequences of human, bovine, mouse, chicken, and frog C/EBP␣ (31). Analysis of the alignment revealed four regions within the transactivation domain with a high degree of sequence homology (Fig. 1A). Homology among CR1, 2, 3, and 4 from the mouse and chicken isoforms of C/EBP␣ is 66, 87, 87, and 94%, respectively.
To determine the region within the C/EBP␣ transactivation domain capable of inducing differentiation, retroviral expression vectors encoding the C/EBP␣ isoforms depicted in Fig. 1A were infected into 3T3-L1 preadipocytes. The adipogenic activity of the various isoforms was measured by the expression of the adipocyte marker 422/aP2 (Fig. 1B) and Oil Red-O staining of cytoplasmic lipid accumulation (Fig. 1C). Consistent with the findings of other investigators (19,20), p42C/EBP␣ induces spontaneous differentiation, but the p30C/EBP␣ isoform that contains only CR3 and CR4 does not (Fig. 1). Based upon these results and our previous results that a C/EBP␣ deletion mutant containing only CR1 and CR2 is able to induce differentiation at a level comparable to p42C/EBP␣ (31), we examined the contribution of CR1 and CR2 to the adipogenic action of C/EBP␣. The C/EBP␣ construct CR2/3/4, which lacks CR1, was able to induce 422/aP2 expression (Fig. 1B) and lipid accumulation (data not shown) comparable to p42C/EBP␣, suggesting that CR1 is not required for C/EBP␣-induced adipogenesis and that CR2 is the adipogenic region. CR1 alone was not sufficient to induce 422/aP2 expression (Fig. 1B) nor promote cytoplasmic lipid accumulation (Fig. 1C), whereas CR2 alone was able to induce spontaneous differentiation. This was consistent with CR2 being sufficient to mediate C/EBP␣-induced adipogenesis. However, despite lacking CR2, CR1/3/4 is also able to induce adipogenesis (Fig. 1B). Thus it appears that multiple conserved regions within C/EBP␣ are involved in the adipogenic effect. CR2 is able to contribute independently of the other regions, whereas CR1 and CR3 work in combination.
p300 Coactivates C/EBP␣-mediated Transcription of the Leptin Promoter-p300 is a nuclear coactivator that has been shown to interact with transcription factors that are important for a number of differentiation paradigms (for review, see Ref. 41). We performed a series of experiments to investigate whether p300 coactivates C/EBP␣ through the adipogenic domains. To determine if p300 is a limiting component and func- tions as a coactivator of C/EBP␣ transcription from an adipocyte-specific promoter, reporter gene assays were performed using a C/EBP␣-responsive, leptin-luciferase reporter gene (obluc). 293T cells, which do not contain endogenous C/EBP␣, were transfected with a constant amount of expression vector encoding C/EBP␣ and an increasing amount of p300 expression vector. The ob-luc activity showed a p300-dependent increase (Fig. 2, top) in the presence of a constant level of C/EBP␣, as seen by immunoblotting (Fig. 2, bottom). p300 had no effect on ob-luc activity in the absence of C/EBP␣. Similar results were seen in assays using the PPAR␥ 2 promoter (data not shown). To ensure that the coactivational effects observed were the result of C/EBP␣ binding the ob promoter and not nonspecific or non-DNA binding events, we performed the reporter gene assay with an ob-luciferase reporter gene with a mutation in the C/EBP binding site (mob-luc). This mutation prevents C/EBP␣ from binding to the promoter (36) and abrogates C/EBP␣-dependent transactivation. Consistent with p300 coactivation being mediated through the C/EBP␣ binding site, coactivation was disrupted in reporter assays utilizing mob-luc (Fig. 3, solid  bars). The ability of p300 alone to activate both ob-luc and mob-luc is the result of utilizing high amounts of p300 expression vector (5 g) in this experiment and is not seen at lower plasmid concentrations (Figs. 2, 4, and 6). These data strongly suggest that p300 is a rate-limiting component in the C/EBP␣ transcriptional machinery.
Functional Interaction between C/EBP␣ and p300 Is Inhibited by Adenovirus E1A-p300 was first cloned as an adenovirus E1A-associated protein (42). It was subsequently shown that E1A prevents the association of p300 with a number of transcription factors (38,43,44) and inhibits the interaction required for coactivation. We examined whether E1A could inhibit coactivation of C/EBP␣ transactivation by p300. Cotransfection of an expression vector encoding wild-type 12S E1A repressed C/EBP␣-mediated ob-luc activity in both the basal and the p300-coactivated states (Fig. 4). The E1A-RG2 mutant (45), which has a decreased affinity for p300, fails to repress (Fig. 4). Expression of C/EBP␣ remained constant as shown by immunoblotting. These results confirm that p300 acts as a coactivator of C/EBP␣. Furthermore, the inhibition of basal C/EBP␣ activity by E1A is consistent with endogenous p300 coactivating C/EBP␣ transcription.
Multiple Regions of the C/EBP␣ Transactivation Domain Functionally Interact with p300 -Having firmly established that p300 functions as a coactivator of C/EBP␣ transcription, we then investigated the region of the C/EBP␣ transactivation domain which interacts with p300. To determine if coactivation of C/EBP␣ by p300 is independent of the bZIP domain we created a chimeric protein consisting of the Gal4 DNA binding domain and the C/EBP␣ transactivation domain (Gal4-C/ EBP␣). We then performed a luciferase reporter gene assay using a Gal4-responsive reporter construct and tested the ability of p300 to coactivate the chimeric protein (Fig. 5). Cotransfecting p300 with the reporter gene alone or with the Gal4 DNA binding domain generated minimal reporter gene activity. Cotransfection of Gal4-C/EBP␣ resulted in a robust induction in reporter gene activity, whereas cotransfection with p300 potentiates the basal Gal4-C/EBP␣ transactivation by ϳ36-fold. This response confirmed our hypothesis that p300 functionally in- teracts with the transactivation domain of C/EBP␣.
To determine which conserved regions of the C/EBP␣ transactivation domain mediate the interaction with p300, we created C/EBP␣ transactivation domain truncations (Fig. 6A, top) and conserved region deletion mutants (Fig. 6B, top). We then tested the ability of p300 to coactivate transcription of ob-luc by the deletion mutants. Immunoblot analysis was used to ensure that expression levels of the C/EBP␣ mutants were similar (data not shown). Truncation of the transactivation domain revealed that the removal of CR1 did not have an effect on basal transcription or on coactivation by p300 (Fig. 6A, CR2/3/ 4). Further truncation of the amino terminus showed that upon the loss of CR2 there was a substantial decrease in the basal activation, but there was no change in the potentiation by p300 (ϳ2.6-fold in both CR2/3/4 and p30). Removal of CR3 resulted in a further decrease in basal transcription and the loss of coactivation by p300 (Fig. 6A, p30 versus CR4). These results suggest that CR3 is a region within the C/EBP␣ transactivation domain which interacts with p300. However, when a construct containing only CR1 and CR2 (Fig. 6B, CR1/2) was assayed, p300 cotransfection potentiated the transcriptional activity, indicating that CR3 is able to interact functionally with p300 but is not required for p300 interaction with C/EBP␣. Based upon the observation that the deletion of CR1 had a minimal effect on coactivation by p300 (Fig. 6A, CR2/3/4), we suspected that CR2 was also able to interact functionally with p300. Consistent with this hypothesis, a C/EBP␣ con-struct lacking both CR2 and CR3 (Fig. 6B, CR1/4) was not coactivated by p300. Furthermore, deletion of all of the conserved regions in the C/EBP␣ transactivation domain except CR2 (Fig. 6B, CR2) revealed that like CR3 (data not shown), CR2 alone can be coactivated by p300. Thus it appears that multiple regions of the C/EBP␣ transactivation domain interact functionally with p300. DISCUSSION In this study we find that multiple conserved regions of the C/EBP␣ transactivation domain are able to stimulate adipogenesis and interact functionally with p300. C/EBP␣, ␤, and ␦ are related transcription factors in the C/EBP family of genes which play a role in adipogenesis (for review, see Ref. 7). These proteins display a high degree of homology between their bZIP domains (93% homology between C/EBP␣ and C/EBP␤) and recognize the same consensus DNA binding site (46). However, their sequences diverge markedly within the transactivation domain. Work by Elberg et al. (47) demonstrated the important role of the transactivation domain by showing that despite the similarities in the DNA binding domains, C/EBP␤ homodimers are unable to bind to and activate the PPAR␥ 2 promoter. C/EBP␣ is able both to bind and transactivate the PPAR␥ 2 promoter. These investigators found that fusing the C/EBP␣ transactivation domain to the C/EBP␤ bZIP allows transactivation, whereas fusing the C/EBP␤ transactivation domain to the C/EBP␣ bZIP blocks transactivation (47). Furthermore, . Expression levels of the various mutants were similar to that of p42 and were not altered with p300 cotransfection (data not shown). Data for control, p42, p30, and bZIP isoforms were obtained in parallel, whereas data used for CR2/3/4 and CR4 were from separate experiments where a p42-positive control was assayed in parallel, and the basal transactivation and p300 potentiation were similar to that displayed. B, schematic diagram of C/EBP␣ conserved region deletion mutants (top). An experiment similar to that in panel A was performed using expression vectors for p42, CR1/2, CR1/4, CR2/4, CR2, and CR1/3/4 isoforms of C/EBP␣ (bottom). Data for control, p42, CR1/4, and CR2/4 were obtained from assays performed in parallel. Data for CR1/2, CR2, and CR1/3/4 are from independent experiments where a p42-positive control was assayed in parallel, and the basal transactivation and p300 potentiation were similar to that displayed. All reporter gene results are representative of at least three independent experiments. because of the use of alternative translation start sites (32,48), C/EBP␣ (p42/p30) and C/EBP␤ (LAP/LIP) have isoforms that vary in the composition of their transactivation domain. The ratio of expression between these isoforms is critical for normal 3T3-L1 differentiation (49), again implicating the C/EBP␣ transactivation domain as a key component in understanding the control of adipogenesis. In our study, we have identified highly conserved regions of the C/EBP␣ transactivation domain which mediate C/EBP␣-induced differentiation of 3T3-L1 preadipocytes. Of the four conserved regions in the transactivation domain, CR2 is the only region capable of inducing spontaneous differentiation alone when fused to the C/EBP␣ bZIP (Fig. 1). Although CR1 and CR3 are incapable of inducing spontaneous differentiation when placed individually next to the bZIP, a construct containing both regions but lacking CR2 (CR1/3/4) promotes cytoplasmic lipid accumulation and the expression of adipocyte markers.
The ability of CR2 to act as a strong activation domain and induce differentiation independent of the other conserved regions may be the result of its ability to interact with multiple coactivators and components of the basal transcription apparatus. The retinoblastoma protein (pRb) has been shown to coactivate C/EBP␣ transcription (50) and is thought to interact with a site within CR2 (51). Furthermore, TBP and transcription factor IIB (TFIIB) have been shown to interact within the CR2 region of C/EBP␣ (52) to promote C/EBP␣-mediated transcription. The coactivator p300 is also able to potentiate C/EBP␣ transactivation ( Fig. 2 and Ref. 43); and based upon our results, this is mediated, in part, by functional interaction with CR2 (Fig. 6B). Spontaneous adipogenesis induced by CR1/ 3/4 ( Fig. 1), which lacks the putative pRb interaction site (51), suggests that C/EBP␣ can induce differentiation in 3T3-L1 preadipocytes independent of pRb. This is in contrast with the observation that C/EBP␣ fails to induce differentiation of pRb Ϫ/Ϫ 3T3 fibroblasts (50). The conflicting observations may be the result of differences in the cell models or the presence of a previously unpredicted pRb-interacting region within the C/EBP␣ transactivation domain.
We show that p300, in addition to interacting functionally with CR2, is able to interact with CR3 (p30). This interaction of p300 with multiple regions has been shown for a number of transcription (53,54). Attempts to identify regions of physical interaction between C/EBP␣ and p300, utilizing a variety of techniques (glutathione S-transferase affinity precipitation, coimmunoprecipitation and electrophoretic mobility shift assays), by ourselves and others 2 have been unsuccessful. It may be that p300 does not interact directly with C/EBP␣ but is a limiting component in a higher order complex that must form to allow C/EBP␣ transactivation. An alternative hypothesis is that interactions between C/EBP␣ and nuclear factors, like p300, are very low affinity interactions and are disrupted using standard techniques to demonstrate physical interactions. This is consistent with the measures taken to demonstrate C/EBP␣ interaction with TBP and TFIIB (52). Interaction between C/EBP␣ and both of these proteins was demonstrated in vitro using glutathione S-transferase affinity precipitation, but TFIIB could not be coimmunoprecipitated with C/EBP␣, whereas TBP could only be coimmunoprecipitated under very low stringency conditions (52). We report in a separate manuscript 3 that a blue fluorescent protein (BFP)-C/EBP␣ fusion protein localizes to punctate regions within the nucleus when transfected into GHFT1-5 pituitary progenitor cells or 3T3-L1 preadipocytes. This is similar to the pattern exhibited by en-dogenous C/EBP␣ when localized using immunofluorescence (55). In contrast, green fluorescent protein (GFP)-CREB-binding protein (CBP), a functional homolog of p300, is expressed diffusely within the nucleus. When BFP-C/EBP␣ and GFP-CBP are cotransfected, GFP-CBP is recruited to discreet regions of the nucleus occupied by BFP-C/EBP␣. This study suggests that p300 and C/EBP␣ interact with one another within the cell. This observation in combination with those presented in our study indicates that p300 is acting as a component of the C/EBP␣ transcriptional machinery and suggests that p300 plays a role in mediating the effects of C/EBP␣ on adipogenesis.