CCAAT/Enhancer Binding Protein and Nuclear Factor-Y Regulate Adiponectin Gene Expression in Adipose Tissue

Mouse adiponectin promoter spanning −1,138 to 30 bp was amplified by PCR and ligated with luciferase vector to generate pGLB-A-1138. 5′-Serial deletion constructs (pGLB-A-727, pGLB-A-502, pGLB-A-202, pGLB-A-134, and pGLB-A-57) were generated by deletion using restriction enzymes. Mutant constructs (pGLB-A-134m1, pGLB-A-134m2, pGLB-A-134m3, pGLB-A-134m4, pGLB-A-134m5, and pGLB-A-134m45) were produced by site-directed mutagenesis. Plasmid pmYA (Δ4YA13 m29), which encodes the dominant negative form of the A subunit of NF-Y, was provided by Dr. R. Mantovani (University of Milan, Milan, Italy) (33). Expression plasmids pCMV-rC/EBP-α and pCMV-mC/EBP-β were gifts from Dr. J.R. Cardinaux (Centre de Neurosciences Psychiatriques, Lausanne, Switzerland).

3T3-L1 preadipocytes were maintained in Dulbecco’s modified Eagle’s medium (DMEM) that contained 10% FBS. For inducing differentiation, 2-day postconfluent cells (designated day 0) were fed DMEM that contained 10% FBS, 0.5 mmol/l 3-isobutyl-1-methyl-xanthine, 1 μmol/l dexamethasone, and 1.67 μmol/l insulin until day 2. Cells then were fed DMEM supplemented with FBS/insulin for 2 days, after which they were fed every other day with DMEM that contained FBS. For investigating the effect of glucose or fatty acid, day 6–8 cells were maintained with DMEM that contained low glucose (5.5 mmol/l) and 10% FBS 24 h before treatment.

Total RNA was extracted using TRIzol reagent. Ten micrograms of total RNA of each group was subjected to electrophoresis and transferred to nylon membrane for hybridization. For RT-PCR, 5 μg of total RNA was reverse transcribed, and the resulting cDNAs were used as a template for PCR. The sequences used for RT-PCR are as follows: adiponectin, 5′-AGCCT GGAGA AGCCG CTTAT-3′, 5′-TTGCA GTAGA ACTTG CCAGT GC; resistin, 5′-TCAAC TCCCT GTTTC CAAAT GC-3′, 5′-TCTTC ACGAA TGTCC CACGA; aP2, 5′-GATCG GATCC ACCAT GTGTG ATGCC TTTGT GGGAA CC-3′, 5′-GATCC TCGAG TCATG CCCTT TCATA AACTC TTGT-3′; glyceraldehyde-3-phosphate dehydrogenase, 5′-ACCAC AGTCC ATGCC ATCAC-3′, 5′-TCCAC CACCC TGTTG CTGTA-3′.

Nuclear extract preparation protocol was modified from the description by Lavery et al. (34). Rat epididymal adipose tissues were homogenized with 0.3 mol/l sucrose buffer (15 mmol/l HEPES [pH 7.9], 60 mmol/l KCl, 15 mmol/l NaCl, 0.15 mmol/l spermine, 0.5 mmol/l spermidine, 0.3 mol/l sucrose, 0.5 mmol/l dithiothreitol [DTT], 0.5 mmol/l phenylmethylsulfonyl fluoride [PMSF], and 5 μg/ml protease inhibitor cocktail). The cell lysate was filtered through gauze and added on the 0.9-mol/l sucrose buffer for nuclear isolation. The nucleus was pelleted by centrifugation at 3,000 rpm for 5 min and resuspended in nuclear storage buffer (20 mmol/l Tris-HCl [pH 7.9], 75 mmol/l NaCl, 0.5 mmol/l EDTA, 0.85 mmol/l DTT, and 0.125 mmol/l PMSF). After a brief centrifugation, the pellet was resuspended in 1× NUN buffer (1 mol/l urea, 0.3 mol/l NaCl, 1% Nonidet P-40, 25 mmol/l HEPES [pH 7.6], and 1 mmol/l DTT) and shaken for 30 min for extraction of nuclear protein. The protein concentration was determined by the Bradford method (35). For extracting nuclear proteins from cultured cells, the harvested cells were resuspended in Buffer A (10 mmol/l HEPES [pH 7.9], 10 mmol/l KCl, 0.1 mmol/l EDTA, 0.1 mmol/l EGTA, 1 mmol/l DTT, and 0.5 mmol/l PMSF). After 15 min of incubation on ice, 1:20 volume of 10% Nonidet P-40 was added, vortexed, and then centrifuged. The next steps were the same as those for adipose tissue.

The promoter region of the adiponectin gene covering from −256 to 30 bp was labeled and bound with 60 μg of nuclear extract. After 30 min of incubation on ice, DNase I was treated for 2 min. The reactions were purified and analyzed on a 6% denatured polyacrylamide gel. A G+A ladder was prepared by using the same labeled fragment and was run on the same gel.

The probes corresponding to region AII were generated as follows: 10 pmol of single-stranded sense oligonucleotides were labeled and annealed with 5 mol/l excess of antisense oligonucleotides. The probes encompassing −121 to −63 of promoter used in the mutation study were generated by PCR, using ³²P-labeled antisense primer (−80 to −63) and unlabeled sense primer (−121 to −104) and pGLB-A-134 and corresponding mutant constructs as templates. Probes were incubated with nuclear extract for 15 min on ice, and the DNA-protein complexes were resolved on 4% polyacrylamide gel. Competitive and supershift assays were performed by adding 100 mmol/l excess of unlabeled oligonucleotides or 2 μg of antibodies, respectively. The oligonucleotides used as competitors are as follows: GATA, 5′-CACTT GATAA CAGAA AGTGA TAACT CT-3′; LDL-R sterol-response element (SRE), 5′-TTTGA AAATC ACCCC ACTGC AAAC-3′; NF-Y, 5′-AGACC GTACG TGATT GGTTA ATCTC TT-3′; C/EBP, 5′-TGCAG ATTGC GCAAT CTGCA-3′; and AP-1, 5′-CGCTT GATGA GTCAG CCGGA A-3′.

Fifteen micrograms of nuclear extract was mixed in sample buffer, incubated at 95°C, and separated by 12% SDS-PAGE. The specific proteins were detected by enhanced chemiluminescence for 1 min. Equal protein loading was confirmed by staining with Coomassie Brilliant G-250.

The chromatin immunoprecipitation (ChIP) assay protocol was modified from the description by Latasa et al. (36). Briefly, mice were perfused and fixed with formaldehyde, and adipose tissues were homogenized and sonicated. The resulting supernatant was divided into aliquots for 10-fold dilution in ChIP dilution buffer and precleared with protein A-agarose that contained salmon sperm DNA for 1 h. The antibodies were added and incubated for 12–18 h at 4°C, and the complexes were collected with 50% slurry of protein A-agarose for 3 h. The beads were washed, and chromatin complexes were eluted from the beads. After reverse of the cross-linking, DNA was purified, and 4 μl of input control or ChIP samples was used as a template in PCR using the primer sets for region AII (5′-AGAAG CTCTA CTTGG CTTCC CAGAC CCAAG CTGGA-3′, 5′-GCAGA CCCCA GCTTA CCA-3′) or for exon 3 region (5′-AGCCT GGAGA AGCCG CTTAG-3′, 5′-TTGCA GTAGA ACTTG CCAGT GC-3′), respectively. In the experiments using 3T3-L1 cells, ∼1 × 10⁶ differentiated adipocytes were fixed with formaldehyde, harvested, and then sonicated as described above.

Transient transfections were performed using Lipofectamine reagent (Invitrogen) according to the manufacturer’s protocol. Luciferase activities were measured using 10 μl of cell extract and 50 μl of luciferase assay reagent. β-Galactosidase activity was measured for normalization.

To investigate the regulation of adiponectin gene expression, we first examined the adiponectin mRNA level during adipocyte differentiation of 3T3-L1 fibroblasts. Total RNA was extracted from different stages of the cells after differentiation and subjected to RT-PCR. As shown in Fig. 1A, adiponectin mRNA was not detected in preadipocyte but started to be induced at 4 days after induction of differentiation and elevated further until 6 days, which is consistent with the previous reports (6,9). Meanwhile, resistin appeared earlier than adiponectin and continued to increase until 6 days after differentiation. Another adipocyte-specific gene, aP2, was slightly expressed in the preadipocyte, and the expression level almost reached the maximum level at day 2. This indicates that adiponectin is expressed only in fully differentiated adipocytes compared with other adipocyte-specific genes, aP2 and resistin. On the basis of this result, we used differentiated adipocytes of at least day 6 for the next experiment to investigate nutritional regulation.

Next, we examined the effect of glucose on the expression of adiponectin. Differentiated 3T3-L1 adipocytes were maintained in media that contained 5.5 mmol/l glucose for 24 h and then were treated with 27.5 mmol/l glucose. Northern blot analysis showed that treatment with a high concentration of glucose for 24 h drastically upregulated adiponectin expression (Fig. 1B), suggesting that transcription of adiponectin is induced by glucose. This result led us to investigate whether adiponectin expression could be regulated in response to diet in living organisms. Mice were fasted and then refed a standard diet for 2–8 h, and the adiponectin expression was examined by RT-PCR. As shown in Figs. 1C and D, the expression was induced and peaked 4–6 h after refeeding, suggesting that the adiponectin expression is regulated by diet at the transcriptional level in the early phase of refeeding.

To identify cis-acting elements responsible for the regulation of adiponectin gene expression, we performed DNase I footprinting analysis using nuclear extract from white adipose tissues. The DNA fragments covering the proximal −500-bp region of the promoter were subcloned and used as probes. When liver nuclear extract was used as control because of its abundance of transcription factors, several protein binding site regions (AO, AI, and AII; Fig. 2A) were protected. However, when nuclear extract from adipose tissue was used, only region AII was clearly protected. Region AII has one CCAAT box and two inverted Y boxes, and, interestingly, potential C/EBP binding sites are located between them. Figure 2B shows the sequence of AII with potential nuclear protein binding sites. When it was compared with the human sequence, only the proximal CCAAT box and C/EBP site showed high sequence similarity between species.

For investigating the binding of protein to region AII of the adiponectin promoter, a electrophoretic mobility shift assay (EMSA) was performed using nuclear extract from white adipose tissue. The oligonucleotide probe of the region AII sequence formed two major complexes, slow-migrating bands and a fast-migrating band (Fig. 3A). For determining which transcription factors are involved in these DNA-protein complexes, unlabeled competitor oligonucleotides for NF-Y, GATA, C/EBP, SRE, and AP-1 were added to the reaction. As shown in Fig. 3A, the addition of NF-Y and C/EBP competitors abolished the binding of slow- and fast-migrating bands, respectively, whereas unlabeled GATA, SRE, and AP-1 oligonucleotides did not affect the complex formation. The binding of NF-Y and C/EBP to region AII was further investigated by a supershift assay. As shown in Fig. 3B, the DNA-protein complexes were supershifted by anti-NF-YA, NF-YB, C/EBP-α, and C/EBP-β, indicating that both NF-Y and C/EBP bind to region AII of the adiponectin promoter.

For further investigation of precise binding sites of C/EBP and NF-Y, several mutant oligonucleotide probes were generated (Fig. 4A) and EMSA was carried out. Given that 293T cells express only a trace amount of endogenous C/EBPs, nuclear extracts from 293T cells that overexpress C/EBP-α or C/EBP-β were used to examine specific binding activities of C/EBP-α and C/EBP-β to adiponectin promoter. The overexpression of each transcription factor was confirmed by Western blot (Fig. 4C). As shown in Fig. 4B, C/EBP-α (lane 7) and C/EBP-β (lane 13) bound to the wild-type oligonucleotide probe. However, binding activities of C/EBP-α and C/EBP-β diminished when m4 mutated oligonucleotide was used as a probe (lanes 11 and 17). With nuclear extract from 293T cells, which do not express C/EBP, only NF-Y binding was observed (lane 1); however, this binding was almost diminished when the m5 probe was used (lanes 6, 12, and 18). It is interesting that the m4 mutation also caused the attenuated binding of NF-Y (lane 5), and, conversely, C/EBP binding was greatly affected by the mutation of the m5 region. It is acceptable because the potential C/EBP binding site expands the m4 and m5 regions, but it is not excluded that a certain form of the interaction is required for full binding of two transcription factor families from this result. When unlabeled mutated oligonucleotides were used as competitors, binding of C/EBP-α and C/EBP-β decreased by the addition of the m5 competitor (Fig. 4D, lane 3) as well as the m4 competitor (lane 4). These findings suggest that C/EBP-α and C/EBP-β bound to adiponectin promoter via the m4 and m5 regions, whereas NF-Y bound via the m5 region, and also suggest that NF-Y binding is required for the full binding activities of C/EBPs.

For determining whether NF-Y and C/EBP are able to regulate adiponectin promoter activity, the adiponectin promoter (−1,138/30) linked to the luciferase reporter vector was constructed and its responsiveness to C/EBPs and NF-Ys was tested. The reporter construct was transiently transfected into CV-1 cells in the presence or absence of expression plasmids of NF-YA, NF-YB, NF-YC, C/EBP-α, or C/EBP-β. As shown in Fig. 5A, C/EBP-α and C/EBP-β activated adiponectin promoter activities, whereas NF-Ys could not. Because NF-Y is constitutively expressed in all eukaryotic cells and is composed of three different subunits, YA, YB, and YC, which are collectively required to bind to the sequence (37), it was thought that overexpression of individual NF-Y subunits probably did not affect the promoter activities. So we tested the response to the dominant negative mutant of NF-YA. The promoter activity was markedly inhibited to <20% of basal promoter activity when cotransfected with the dominant negative form. Taken together, these results suggest that C/EBPs upregulate adiponectin promoter, whereas NF-Y has a crucial role in maintaining basal adiponectin gene expression.

Next, we generated a series of plasmid constructs that contained 5′ serial deletion constructs shown in Figs. 6A and transfected into 3T3-L1 adipocytes. As shown in Fig. 6B, the adiponectin promoter activity decreased dramatically when up to −57 bp was deleted, demonstrating that the −134/−57 region is critical and sufficient for the promoter activity. Moreover, the C/EBP responsiveness was significantly decreased between −134 and −57 (Fig. 6C), so we concluded that the most proximal 134 bp of the promoter has a functional C/EBP site, taken together with the mutation EMSA study shown in Fig. 4. We also investigated the influence of the mutations used for the EMSA study on the C/EBP transactivation property. When the m4 or m5 mutation, which proved to be important for the C/EBP binding on the EMSA, was introduced into the −134 construct, C/EBP responsiveness was greatly reduced (Fig. 6D). This effect was more apparent when m4 and m5 mutations were introduced simultaneously, suggesting that the C/EBP and NF-Y sites on region AII are functional and necessary for the stimulation of the adiponectin promoter.

Because the adiponectin expression was regulated by diet as shown in Fig. 1, we examined whether the binding of NF-Y and C/EBP could be changed according to the dietary status. Nuclear extracts from white adipose tissues of fasted or refed rats were prepared and examined by EMSA. As shown in Fig. 7A, the dietary intakes increased the complex formation on the region AII probe. To confirm the binding activities of C/EBP, we incubated nuclear extracts at 72°C for 20 min before the binding reaction because C/EBP has heat-stable characteristics (38). The result clearly showed the increase of binding of the heat-stable factor, supposed to be C/EBP, to region AII (Fig. 7A). Therefore, it is likely that C/EBP is responsible, in part, for the increased expression of the adiponectin gene ∼6 h after refeeding, as shown in Fig. 1 by the increased binding to region AII.

Next, we tested whether C/EBP binding is increased in response to glucose in differentiated 3T3-L1 adipocytes. Nuclear extracts were prepared from fully differentiated adipocytes that were maintained in media that contained 5.5 mmol/l glucose for 24 h and then treated with 27.5 mmol/l glucose. As shown in Fig. 7B, C/EBP binding was apparently increased 12 h after glucose treatment, similar to the result in refed animals (Fig. 7A). Thus, taken together with the result shown in Fig. 1B, it is concluded that the C/EBP binding is responsible for the glucose induction of adiponectin in 3T3-L1 adipocytes.

To correlate the change of C/EBP and NF-Y levels, we conducted Western blot analysis. As illustrated in Figs. 7C and D, C/EBP-α and C/EBP-β were markedly increased by diet or glucose, whereas NF-Y and other proteins were not changed significantly, suggesting that dietary status regulates the levels of C/EBP proteins, resulting in changes in their binding activities.

To confirm whether NF-Y or C/EBP binds directly to the adiponectin promoter in vivo, we used chromatin immunoprecipitation analysis for the next step. The DNA encompassing region AII was amplified by PCR from chromatin immunoprecipitated with antibodies against C/EBP-α, C/EBP-β, NF-YA, NF-YB, and NF-YC, whereas exon 3, which is located 10 kb downstream of region AII, was not amplified (Fig. 8B). It is interesting that the cyclic AMP response element binding protein (CBP), known as one of the coactivators of C/EBP, also occupied the adiponectin promoter. These results indicate that NF-Y and C/EBP bind directly to region AII of the adiponectin proximal promoter in vivo and also suggest the involvement of CBP as a coactivator.

Previous reports and our results in Fig. 1A showed that adiponectin expression is induced 4 days after adipocyte differentiation. For investigating whether the binding of C/EBPs and NF-Y is adipocyte specific, a ChIP assay of preadipocyte and adipocyte was also performed. As shown in Fig. 8C, C/EBP-α and NF-YA occupied adiponectin promoter in adipocytes but not in preadipocytes. These results suggested that the drastic induction of adiponectin expression during adipocyte differentiation might be explained by the recruitment of C/EBPs and NF-Y into the adiponectin promoter. In addition, the glucose treatment of differentiated adipocytes gave rise to the increased C/EBP-α and C/EBP-β binding in the ChIP assay (Fig. 8D), also suggesting that these transcription factors are involved in the regulation of adiponectin gene expression in response to nutritional status as well.

For many years, a number of studies have reported the linkage between adiposity and insulin resistance. As it became apparent that adipose tissue plays special roles for body metabolism by secreting a set of molecules, recent research has been focused on investigating how these “adipokines” can act as link molecules between obesity and insulin resistance (4). Adiponectin is one of the adipokines secreted by fully differentiated adipocytes and thought to be associated with some metabolic disease because plasma adiponectin levels are decreased under conditions of obesity and insulin resistance (10–12). Moreover, the administration of adiponectin causes glucose-lowering effects and improves insulin resistance in mice (17,39,40). Therefore, adiponectin is considered as an “insulin enhancer” linking adipose tissue and whole-body glucose metabolism. In this regard, if adiponectin plays a role as insulin enhancer, then it is likely that adiponectin could be regulated to control body homeostasis by the changes of metabolic stimulation such as the fasting-feeding cycle.

The present study clearly suggests that adiponectin is regulated in response to nutrition. We showed that glucose seemed to regulate adiponectin gene expression at the transcriptional level (Fig. 1B). Most important, adiponectin gene expression was induced in a relatively short period (4–6 h) after refeeding of mice (Figs. 1C and D). This finding in refed mice is interesting because several studies have failed to show the relationship between the adiponectin expression and fasting-refeeding cycle. It was reported that adiponectin levels are relatively constant and do not dramatically change postprandially in human subjects (11). It is possible that the plasma adiponectin levels might not represent acute changes of transcriptional levels because adipocytokines may function as autocrine or paracrine or the regulation at the protein level might appear in the late period after diet. In addition, we cannot exclude the possibility that the secretion pathway of the adiponectin protein participates in another way of regulation when the animals are subjected to a fasting-refeeding cycle (41). Although the regulation observed in this study cannot fully explain adiponectin biology, it is noteworthy that the induction profile of the adiponectin gene transcription (Fig. 1C) seemed to be similar to that of DNA-protein binding on region AII on the promoter (Fig. 7). Thus, it is likely that the expression of adiponectin is induced shortly after feeding, at least at the transcription level.

The present study revealed that C/EBPs and NF-Y are involved in the regulation of adiponectin gene expression. C/EBP-α and C/EBP-β bound to the adiponectin gene in in vitro and in vivo assays (Figs. 4 and 8) and activated adiponectin promoter in vitro (Figs. 5 and 6). In addition, NF-Y binds to the adiponectin promoter and plays a vital role in regulation of adiponectin gene expression. It is interesting that although transient transfection assays showed that the adiponectin promoter was activated by both C/EBP-α and C/EBP-β, the fold increase by C/EBP-β was larger than that by C/EBP-α (Fig. 5), indicating that C/EBP-β had more crucial roles in regulation of adiponectin expression than C/EBP-α. In this regard, it should be noted that a recently published study failed to show C/EBP-α responsiveness in the adiponectin promoter (42). It is unclear why the C/EBP-α response is small compared with that of C/EBP-β, despite its increased protein expression by nutritional status (Figs. 7C and D). One possible explanation is that in our EMSA findings C/EBP-α protein alone could not lead to the fasting migrating band (Fig. 4B), which is apparent in natural nuclear extract from white adipose tissue (Fig. 3). Meanwhile, the addition of C/EBP-β formed the fasting migrating band (Fig. 4B), suggesting that C/EBP-β, in association with C/EBP-α, plays an important role in the physiological regulation of adiponectin expression. Moreover, we showed that both C/EBP-α and C/EBP-β are important for adiponectin expression as shown by in vitro (EMSA) and in vivo (ChIP) assays along with the overexpression study. Thus, it can be concluded that C/EBP-α, in association with C/EBP-β and/or NF-Y, has a role in the adiponectin gene expression.

As summarized in the introduction, C/EBP has been identified as one of the major adipogenic factors controlling the transcription of the adipocyte-specific gene. The regulation of C/EBP gene expression has been examined in adipose tissue and liver. In liver cells, glucocorticoids induced C/EBP-α and C/EBP-β at the level of transcription and translation (43,44). Conversely, in adipose tissue, glucocorticoids inhibited C/EBP-α expression (45). Meanwhile, the effects of insulin on C/EBP-α expression have generally been shown to be inhibitory in adipose tissue (46,47). In our study, C/EBP-α and C/EBP-β protein levels were elevated in the refed state; thus, it is within the realm of possibility that the elevated adiponectin mRNA level from dietary nutrients is mediated by C/EBP-α or C/EBP-β.

The interaction between NF-Y and other transcription factors on several gene promoters has been reported. NF-Y is required for thyroid hormone regulation of liver S14 gene transcription (48), and NF-Y also participated in the regulation of the ATP citrate-lyase gene under the control of SREBP1c (49). Although NF-Y is unlikely to be regulated by metabolic status as shown in this study, it is interesting to note that NF-Y is needed not only for basal promoter activity but also for C/EBP responsiveness (Figs. 5 and 6). The region AII of the adiponectin promoter has three potential binding sites for NF-Y (represented as m1, m2, and m5 in Fig. 4); only mutation m5 abolished the binding of NF-Y. Comparing this sequence with human adiponectin gene, most nucleotides, including m4 and m5 but not m1/m2, were highly conserved between species (Fig. 2). Therefore, we can conclude that NF-Y could be another important regulator for the proper expression of the adiponectin gene. In this regard, the binding of NF-Y along with C/EBP might be required for adiponectin gene expression during adipocyte differentiation as shown in the ChIP assay (Fig. 8).

In summary, we found that transcription of adiponectin is regulated by diet at the early stage of the refed state and that C/EBP and NF-Y play roles in this process. It remains to be clarified how the nutrients regulate these transcription factors in adipose tissue. Also, whether the binding of C/EBP or interaction with NF-Y is dysregulated in obesity or insulin resistance must be investigated; however, it is clear that C/EBP and NF-Y are important for the regulation of the expression of adiponectin in response to nutrients and in the course of adipocyte differentiation as well.

This study was supported by a grant from the Korea Health 21 R&D Project, Ministry of Health & Welfare, Republic of Korea (02-PJ1-PG10-20908-0005).

We thank Dr. J.R. Cardinaux for generously providing expression plasmids pCMV-rC/EBP-α and pCMV-mC/EBP-β and Dr. R. Mantovani for kindly providing Δ4YA13 m29 plasmid.