Leptin Gene Expression in Human Preadipocytes Is Switched on by Maturation-induced Demethylation of Distinct CpGs in Its Proximal Promoter*

The peptide hormone leptin plays a major role in the regulation of energy intake and expenditure and is predominantly expressed in mature adipocytes but not in preadipocytes. Using bisulfite genomic sequencing, we found that 32 CpGs, distributed within a 317-bp sequence of the proximal leptin promoter, were highly methylated in human preadipocytes (73.4% (cid:1) 9.0%). During maturation toward terminally differentiated adipocytes, this promoter region was extremely demethylated (9.4% (cid:1) 4.4%). CpG methylation-dependent transcriptional activity of the promoter fragment was determined in transfection experiments using a set of 5 (cid:1) -truncated mock-, Hha I-, and Sss I-methylated promoter-reporter constructs. Whereas the methylated CpG within the CCAAT/enhancer-binding protein (cid:2) recognition site down-regulated reporter expression, methylated CpGs proximal to the TATA motif and/or in a further upstream region abrogated promoter activity completely. These distinct promoter CpG sequences were found unmethylated in leptin-expressing mature adipocytes. As evidenced by electrophoretic mobility shift assays, nuclear protein complexes were specifically formed on methylated oligonucleotide probes corresponding to the dedi-cated promoter sequences, indicating that methyl-CpG (cid:3) CG composition of

Terminal differentiation of precursor cells involves the coordinate expression of a subset of cell type-specific genes that gives rise to a new morphological and functional phenotype. Fibroblast-like cells derived from adipose tissue, which have undergone commitment to the adipose lineage, can be induced to differentiate into adipocyte phenotype by treatment with insulin, insulin-like growth factor-1, glucocorticoid, and triiodothyronine (1). Upon exposure to differentiation inducers, preadipocytes undergo several rounds of cell division (clonal expansion), become growth-arrested, and coordinately express adipocyte-specific genes (2)(3)(4)(5). This process is regulated by a cascade of transcriptional activators, most notably CCAAT/ enhancer binding proteins (C/EBPs) 1 (6 -8) and peroxisome proliferator-activated receptor ␥2 (9,10). Maintenance of the terminally differentiated state is ensured by sustained expression of C/EBP␣, which blocks mitosis (11,12) and trans-activates a number of adipocyte-specific genes (13)(14)(15). One of these, the leptin gene, is transcriptionally highly activated by C/EBP␣ (16,17). The sequence of the human leptin gene has been identified and spans about 18 kilobase pairs containing three exons. The 5Ј-flanking region includes a TATA motif, three GC boxes, an AP-2 binding site, and a C/EBP␣ motif (18).
The gene product, leptin, is a multifunctional 16-kDa peptide hormone, mainly involved in energy homeostasis and regulation of body weight (19 -21) as well as in reproduction (22), sympathetic nerve activation (23), hematopoiesis (24), angiogenesis (25), and immune response (26). In a previous study, we have shown that in LiSa-2 cells, a human liposarcoma cell line, the adipocyte-specific genes GLUT4, aP2, phosphoenolpyruvate carboxykinase, and SCD1 were inducibly expressed as the response to hormonal agents of adipocyte differentiation in the absence of leptin gene expression (27). However, treatment of these cells with DNA methyltransferase inhibitor 5-aza-deoxycytidine led to leptin gene expression. This observation argues for a role of DNA demethylation in leptin gene expression during the process of adipocyte differentiation. The promoter regions of many eukaryotic genes contain stretches of CpG-rich sequences known as CpG islands. These are considered to be usually unmethylated in transcriptionally active genes (28). By contrast, the inactivation of the X-chromosome and genomic imprinting are established consequences of DNA methylation (29 -31). Furthermore, it was shown that the extent of methylation of CpG islands in tissue-specific genes directly influenced their transcriptional activity (32)(33)(34). As the first evidence of a developmentally regulated DNA demethylation in human adipocyte differentiation, we show that a highly methylated CpG island within the proximal promoter sequence of the silent leptin gene of human preadipocytes is unmethylated in terminally differentiated adipocytes expressing the leptin gene. We further show that distinct methylated CpG motifs in this proximal promoter sequence act as specific sites for methyl-CpG binding proteins that are involved in leptin gene repression. and antibiotics (27). For adipose differentiation, cells were subcultured in serum-free DMEM/Ham's F-12 (1:1) medium supplemented with 10 g/ml transferrin, 15 mM NaHCO 3 , 15 mM HEPES, 33 M biotin, 17 M pantothenate, 100 units/ml penicillin, and 0.1 mg/ml streptomycin and further referred to as basal medium. Additionally, Lisa-2 cells were preincubated for 48 h in basal medium complemented with 0.5 M 5-aza-dC. To induce adipose differentiation, basal medium of subconfluent cultures was supplemented with 1 nM insulin, 20 pM triiodothyronine, and 1 M cortisol and further referred to as adipogenic medium. Media were changed twice a week. Cells were considered as differentiated adipocytes once their cytoplasm was loaded with lipid droplets.
Oligonucleotides-The oligonucleotides used in this study are listed in Table I. PCR primers were selected from the published human leptin gene promoter sequences (GenBank TM accession numbers U43589 and D62708) to amplify promoter fragments of different lengths. A sequence of 15 bases including the BglII restriction side (underlined) was attached to the 5Ј-end of the reverse primer, and 15 bases including the SacI restriction side were attached to the 5Ј-end of the forward primers to achieve suitable PCR products for subcloning into the promoterless pGL3basic vector. Primer pairs ϩ52-bsm and Ϫ258-bsm and 5Ј-deleted bsm primer pairs (deleted sequence is indicated as italic type) for nested PCR were designed to amplify a fragment of the sense strand of bisulfite-modified human leptin promoter DNA. PCR primer pairs for cDNA analysis were selected from the corresponding GenBank TM cDNA sequences using the GCG prime program package (Wisconsin Version 10.2, Genetics Computer Group (GCG), Madison, WI). Oligonucleotides for electrophoretic mobility shift assays (EMSA) were chosen from the human leptin gene promoter sequence. AGG sequences (underlined) were attached to the 5Ј-end for 32 P labeling. The positions of 5-methylcytosine instead of cytosine are indicated as *C. PCR primers were custom-synthesized (MWG-Biotech). The defined methylated oligonucleotides were synthesized by ThermoHybaid.
DNA Isolation and PCR-Cells growing in the log phase were washed twice with phosphate-buffered saline and lysed in digestion buffer, consisting of 10 mM Tris-HCl, pH 8, 100 mM NaCl, 25 mM EDTA, 0.5% SDS, and 0.1 mg/ml proteinase K at 50°C for 12 h. Following the phenol/chloroform extraction procedure, DNA was treated with DNasefree RNase (Roche Molecular Biochemicals), repeatedly phenol/chloroform-extracted, and ethanol-precipitated.
DNA Was Quantified by Spectrophotometry-PCR conditions were as follows: a mixture consisting of 100 ng of DNA, forward and reverse primers, 25 pmol each, 10 pmol of dNTPs, 5 l of 10ϫ PCR buffer, 25 pmol of MgCl 2 , and 2.5 units Taq polymerase (Qiagen) at a total of 50 l was hot started at 95°C and run for 5 min at 95°C, followed by 35 cycles of 95, 56, and 72°C for 1 min at each temperature and a final elongation step for 10 min at 72°C. PCR products were resolved in a 1% low melting agarose gel with ethidium bromide staining and sized by comparison with a 100-bp DNA ladder. Expected bands were excised and eluted with the Wizard PCR Preps DNA Purification System (Promega).
Bisulfite Modifying of Genomic DNA and Sequencing-Cellular DNA was treated with sodium bisulfite using the CpGenome DNA Modification Kit (Invitrogen) according to the manufacturer's protocol. PCRs were performed on modified DNA using the primers ϩ52r-bsm and Ϫ258f-bsm. Nested PCRs were performed on 10 l of the first PCR amplification product with the primers ϩ47r-bsm and Ϫ253f-bsm. PCR products were purified by agarose gel electrophoresis and inserted for sequencing into a pGEM-T Easy vector (Promega). Plasmids were isolated with the Qiagen plasmid kit and custom sequenced using standard sequencing primers (MWG Biotech).
RNA Isolation and Reverse Transcription-PCR-Total RNA was isolated from cultured cells with TRIzol reagent according to the manufacturer's instructions (Invitrogen) and treated with 1 g/l RNase-free DNase (Roche Molecular Biochemicals). 5 g of RNA were reverse transcribed using random hexamer primers and SuperScript II reverse transcriptase according to the manufacturer's protocol (Invitrogen). PCRs were performed as described upon 1 l of reverse transcription reaction using primer pairs listed in Table I.
Plasmid Leptin-Promoter Constructs-DNA fragments of the human leptin promoter corresponding to the region from ϩ59 to Ϫ258 and a set of 5Ј-truncated promoter fragments were made via PCR using the primers listed in Table I. The agarose-gel-purified PCR products were cleaved with a mixture of 20 units each of SacI and BglII restriction enzymes in 50 l of restriction buffer A (Roche Molecular Biochemicals) for 6 h at 37°C. The digests were purified by low melting agarose gel electrophoresis and ligated into the SacI/BglII site of the promoterless luciferase-reporter vector pGL3Basic (Promega).
The inserted promoter fragments were reamplified by PCR using plasmid-specific PCR primers, and the sequence was confirmed on the specific patterns of different restriction enzyme digests (data not shown). Leptin promoter-pGL3Basic constructs were referred to as, for example, P-258 (i.e. the insert is the PCR product amplified with primer pair Ϫ258f and ϩ59r).
An expressing plasmid for C/EBP␣ was made by inserting the coding sequence of the human C/EBP␣ gene into the cloning site of the pcDNA3.1/V5-Topo vector (Invitrogen).
Plasmid CpG Methylation-20 g of plasmid DNA were incubated for 24 h without (mock methylated) and with 20 units of SssI-or HhaI-methylase (New England Biolabs) in 400 l of methylase buffer supplemented with 160 M S-adenosylmethionine. After 4 h of incubation, 2 l of S-adenosylmethionine solution (32 mM) were repeatedly added. The reaction was terminated by heating (60°C, 10 min), and methylated plasmids were purified by the Wizard DNA Clean-Up System (Promega). The efficiency of CpG methylation of the inserted promoter constructs was tested as follows. 2 g of plasmid DNA were digested with SacI/BglII. The inserts were separated by agarose gel electrophoresis, and completeness of methylation was proven by the inhibition of the digest with 5-methyl-cytosine-sensitive restriction enzymes (data not shown).
Transient Transfection of LiSa-2 Cells-24 h prior to transfection, 5 ϫ 10 4 LiSa-2 cells in 0.5 ml of basal medium were seeded into 24-well tissue culture plates. For each well, 0.5 g of promoter-pGL3basic construct, 0.5 g of C/EBP␣ expression vector, and 25 ng of pRL-CMV vector were diluted in 25 l of serum-free DMEM medium and precomplexed with 2 l of LipofectAMINE reagent (Invitrogen) in a total of 50 l of DMEM and incubated for 30 min at 20°C. The culture medium was replaced by 200 l of serum-free DMEM, and the plasmid-Lipo-fectAMINE complexes were added to the cells. Following 5 h of incubation at 37°C, 400 l of medium containing 20% (v/v) fetal calf serum were added. 48 h after transfection, the cells were washed once with phosphate-buffered saline and lysed in passive lysis buffer (100 l/ well). Luciferase activity was determined using the dual-luciferase reporter assay system according to the manufacturer's instructions (Promega) in the Lumat LB 9507 luminometer (Berthold).
Quadruple independent experiments were made for the transfections. Luciferase activity was measured in doublets. The relative promoter activity is expressed as the ratio of light units of leptin promoterdriven fire fly luciferase gene and cytomegalovirus promoter-driven Renilla luciferase gene of the co-transfected pRL-CMV vector.
EMSA-Nuclear protein extract from rat liver (Geneka Biotechnology Inc.) or nuclear protein extract prepared according to the Dignam protocol from nondifferentiated Lisa-2 cells was used in the protein binding reactions (36). Complementary oligonucleotide probes (Table I) were annealed and end-labeled with [␣-32 P]dCTP using Klenow enzyme. The specific activity of the probes used in the assays was adjusted to 50,000 cpm/0.1 pmol DNA. TATA probes were incubated at 20°C for 30 min with 1 g of rat liver nuclear protein extract in a buffer consisting of 10 mM Tris-HCl, pH 7.5, 50 mM KCl, 5 mM dithiothreitol, 10% glycerol, 0.2 g/l poly(dI-dC), and 1 g/l bovine serum albumin, in a total volume of 20 l. For protein binding reactions on C/EBP probes, the binding buffer contained 20 mM HEPES, pH 7.9, 50 mM KCl, 1 mM EDTA, 1 mM dithiothreitol, 4% Ficoll, 0.2 g/l poly(dI-dC), and 1 g/l bovine serum albumin. Protein binding reactions on MeCP-200 probes were performed essentially as previously described (15), using nuclear extract from LiSa-2 cells. The methylated or unmethylated competitors were added in a 10-or 100-fold molar excess. Samples were loaded on 4.5% polyacrylamide gels and run at 10 V/cm for 2 h in 0.5ϫ TBE buffer. The gels were dried and exposed to x-ray film. Densitometric analysis of the DNA-protein complexes was performed on the captured images using the ImageMaster VDS software (Amersham Biosciences).

A CpG Island Is Located in the Human Proximal Leptin
Promoter-In this study, we used a 317-base pair DNA fragment, ranging from nt ϩ59 to Ϫ258 of the human leptin promoter, to correlate the transcriptional activity of the leptin gene with the level of CpG methylation. 32 CpG dinucleotides are distributed within this sequence, particularly located in the GC-box at Ϫ19, Ϫ95, and Ϫ100; proximal to the TATA-box at Ϫ33 and Ϫ38; within the C/EBP␣ binding site at Ϫ51; and within six CpG motifs at positions Ϫ183, Ϫ186, Ϫ188, Ϫ200, Ϫ202, and Ϫ204. Fig. 1 shows the result of a CpG plot calculation. This program defines a CpG island as a region where, over an average of 10 windows, the calculated percentage of CG composition is over 50%, and the calculated observed CpG distribution versus the expected is over 0.6 (37). The program recognizes in the leptin promoter fragment a CpG island, sized from nt ϩ3 to Ϫ211 with a CG composition of 73.8%, the average observed-to-expected ratio of CpGs being 0.87.
Adipose Differentiation Is Essential for Leptin Expression-Among other adipose tissue-specific proteins, leptin can be considered as a marker protein expressed in terminal differentiated adipocytes. To evaluate leptin mRNA expression and the presence of transcription factor mRNAs involved in the regulation of leptin promoter activity, reverse transcription-PCR experiments on preadipocyte and adipocyte RNA were carried out. The results are shown in Fig. 2. In preadipocytes, leptin mRNA was undetectable, C/EBP␣ message was found in traces, whereas AP-2␣ and Sp1 transcripts were found abundantly (gel image A). On culturing these cells for 20 days in adipogenic medium, the morphology changed to spheroid adipocytes loaded with lipid droplets, in which C/EBP␣ and leptin mRNAs were clearly detectable (gel image B).
Leptin Promoter Demethylation Occurs during Adipocyte Differentiation-In a recent study, we postulated an association between DNA demethylation and leptin expression during hormone-induced adipocyte differentiation of LiSa-2 cells (27). Induction of leptin mRNA in LiSa-2 cells was only observed after preculturing the cells in basal medium supplemented with 5-aza-desoxycytosine (5-aza-dC). To confirm these findings, we determined the methylation pattern of the leptin promoter region of cultured human preadipocytes, in vitro-differentiated adipocytes, and differentiated LiSa-2 cells with and without 5-aza-dC treatment. Cellular DNA was treated with bisulfite, and the C to U converted leptin promoter sequence was PCRamplified and sequenced. The promoter region investigated ranges from nt ϩ47 to Ϫ253 and includes 32 CpG dinucleotides. As shown in Fig. 3, following adipose differentiation, the level of CpG methylation in the leptin promoter DNA decreased greatly from 73.4 Ϯ 9.0% in leptin-nonexpressing preadipocytes to 9.4 Ϯ 4.4% in leptin-expressing adipocytes. In LiSa-2 cells, the effect on demethylation caused by 5-aza-dC resembled that observed during preadipocyte differentiation. Upon 5-aza-dC treatment, promoter methylation declined in differentiated LiSa-2 cells from 48.1 Ϯ 12.5 to 11.5 Ϯ 11.0%.
Two informative examples of sequencing chromatograms are shown in Fig. 4. Fig. 4 (left) represents the human leptin promoter region from nt Ϫ14 to Ϫ59, including the core promoter and the upstream C/EBP␣ site. To demonstrate the correctness of sequencing, the published sequence of the leptin gene is superimposed (Fig. 4 (left), line A) on the corresponding DNA sequence of preadipocytes ( Fig. 4 (left), line B). In Fig. 4 (left) line C, the sequencing chromatogram of bisulfite-modified preadipocyte DNA illustrates the CpG methylation status of this particular promoter region. The cytosines in the CpG dinucleotides at positions Ϫ19, Ϫ33, Ϫ38, and Ϫ51 remained and were therefore methylated in the original preadipocyte DNA. By contrast, in the modified DNA of differentiated adipocytes ( Fig. 4 (left), line D), all cytosines were converted to uracils (appearing after PCR as thymidines), indicating that the original promoter sequence was unmethylated. Fig. 4 (right) represents the leptin promoter region from nt Ϫ175 to Ϫ215, including two CpG triplets at nt Ϫ183 to Ϫ188 and at nt Ϫ200 to Ϫ204, all of which were originally methylated in preadipocyte-derived DNA as documented in Fig. 4 (right), line C. After terminal differentiation, these CpG motifs were unmethylated as evidenced in the sequencing chromatogram ( 5Ј-truncated promoter fragments, ranging from nt ϩ59 up to Ϫ258 of the leptin gene, were inserted upstream of the luciferase gene into pGL3Basic vectors. In addition to the mockmethylated plasmid constructs, total CpG-methylated constructs were made by the action of SssI methylase. GCGC sequences in the fragments at positions ϩ44, ϩ42, ϩ11, Ϫ51, Ϫ74, Ϫ85, Ϫ186, Ϫ188, Ϫ202, and Ϫ204 were methylated with HhaI methylase. Subsequently, plasmid constructs were transiently transfected into LiSa-2 cells. Fig. 5 shows leptin promoter activity in transfection experiments using construct P-258. Reporter expression increased 2-fold in cells cultured after transfection in adipogenic medium compared with those cultured in basal medium. Co-transfection with a C/EBP␣expressing vector markedly increased the reporter expression rate to about 30-fold, without an additional hormone effect. These results are in accordance with published data on cotransfection experiments to achieve leptin promoter expression in 3T3-L1 mouse preadipocytes (17) and indicate that in LiSa-2 cells ectopic C/EBP␣ expression is needed for stimulation of leptin promoter activity too. Fig. 6A shows the results of transient transfection experiments with the mock-methylated promoter-plasmid constructs or after methylation by HhaI-or SssI-methyltransferase. Construct P-106 containing the DNA binding sites for Sp1, C/EBP␣, and the TATA-box, achieved the highest reporter activity of 103.9 relative light units. The additional AP-2 binding site in promoter fragment P-149 had no enhancing effect on promoter activity. In accordance with published data, the promoter fragment in construct P-106 can be considered as the minimal leptin promoter (17). A decrease in promoter activity was observed using constructs, which were extended in their upstream sequences. The activity declined to about one-half of maximal activity when cells were transfected with P-174 and to about one-quarter using P-199. After lengthening the promoter fragment sequence up to nt Ϫ258, a 2-fold increase in reporter activity was observed compared with that of P-199, which approximated the levels of P-174 or P-60, respectively. Excluding the AP-2 and Sp1 binding sites by shortening the fragment sequence to nt Ϫ60, the promoter activity dropped to one-half of maximal activity. An additional, even more dramatic decrease in promoter activity was observed with construct P-35, which only consists of the TATA motif and downstream sequences. The decrease in promoter activity from 65.3 relative light units produced by P-60 compared with 15.9 produced by P-35 was a clear indication of the importance of the C/EBP␣ binding site for gene activation. The promoter activity of about 3 relative light units achieved by the TATA-less plasmid P-24, was just slightly above that produced by the promoterless pGL3Basic vector (Ͻ1.0 relative light units).
SssI methylation of the CpGs in the leptin promoter constructs P-258 and P-199 dramatically decreased the promoter activity to values of the promoterless basic vector. This silencing effect of SssI methylation was less extensively achieved by plasmids P-174 up to P-60. The promoter activity increased Except for promoter construct P-199, HhaI methylation of CpGs had a moderate effect on reporter gene expression. Using the construct P-258 and P-174 down to P-35, each methylated at the indicated CpGs of the leptin promoter region (Fig. 6C), the reporter expression declined to about one-third compared with that obtained by unmethylated constructs. Since these expression rates were approximately as high as that of the unmethylated construct P-35, in which the C/EBP␣ DNA binding site is absent, the reduced promoter activity was most likely due to the methylated CpG dinucleotide at Ϫ51 in the C/EBP␣ site. A striking response to HhaI-CpG methylation was observed by promoter construct P-199. Whereas the unmethylated construct achieved 32.6 relative light units, HhaI methylation caused a dramatic decrease of reporter expression to 0.2 units, similar to the 0.3 units in the case of SssI methylation. Surprisingly, reporter expression increased up to 10.1 light units, using the HhaI-methylated promoter construct P-258, which contains two additional HhaI methylation sites at Ϫ202 and Ϫ204. However, methylation of P-258 at positions Ϫ186, Ϫ188, Ϫ200, Ϫ202, and Ϫ204 (SssI methylation) decreased the expression to 1.1 light units (i.e. to values comparable with the promoterless vector). Collectively, the results of these transfection experiments indicate that methylation/demethylation of CpGs within three regions of the leptin promoter (i.e. at positions Ϫ33 and Ϫ38 proximal the TATA-box, at Ϫ51 within the C/EBP␣ motif, and upstream at positions Ϫ186, Ϫ188, and Ϫ200) plays an important part in the regulation of leptin gene transcription.
Methylated CpG Sites in the Leptin Gene Promoter Are Targets for Protein Binding-To prove methylation-dependent protein-DNA interactions in the leptin promoter, we performed EMSA with defined methylated DNA probes that correspond to the TATA-box region from position Ϫ19 to Ϫ42, including two CpGs at Ϫ33 and Ϫ38, the C/EBP␣ binding motif from Ϫ39 to Ϫ64, including two CpGs at Ϫ51 and Ϫ62, and a CpG dense upstream region from position Ϫ178 to Ϫ210, including two CpG triplets at Ϫ183 to Ϫ188 and Ϫ200 to Ϫ204 (MBD-200; Table I). EMSA was performed on rat liver nuclear protein extract on TATA DNA probes, which were both unmethylated and methylated at positions Ϫ33 and Ϫ38. The result is shown in Fig. 7A. Complexes were formed on both methylated (M) and unmethylated (U) probes (lanes 1 and 2, asterisk). In competition experiments, total displacement of the unmethylated probe was achieved by a 10-fold molar excess of unmethylated and methylated competitors (lanes 7-10). The methylated probe was completely displaced from the complex by the corresponding competitor (lanes 5 and 6). Even a 100-fold molar excess of unmethylated competitor led only to a partial displacement of the methylated probe (lanes 3 and 4). Additional protein-DNA complexes were formed on the methylated probe, as shown in Fig. 7A, lanes 2-4 (arrows). The methylationspecific binding of the probe was proven by total displacement with the methylated competitor ( lanes 5 and 6), whereas the unmethylated competitor was ineffective (lanes 3 and 4). The result of an EMSA carried out with DNA probes corresponding to the binding region of C/EBP␣ and rat liver nuclear protein extract is shown in Fig. 7B. DNA-protein complexes of virtually identical electrophoretic mobility, associated with C/EBPs-DNA complexes (7), were shown with both unmethylated and methylated probe (lanes 1 and 2, arrows). In comparison with the protein binding affinity of the unmethylated C/EBP␣ probe, the methylated counterpart exhibited a markedly reduced binding affinity. A 10-fold molar excess of both unmethylated and methylated competitors blocked probe binding. Confirming our results obtained in the transfection experiments, methylation of the CpG dinucleotide in the C/EBP␣ binding site obviously diminished binding affinity to its cognate element in the proximal leptin promoter. To investigate whether methylation of a CpG-dense region in the upstream promoter sequence of the human leptin gene affects binding of specific proteins, we performed EMSAs using nuclear protein extract derived from nonstimulated LiSa-2 cells and a probe (MBD-200) corresponding to the promoter sequence from nt Ϫ178 to Ϫ210. In the methylated probe, cytosine was replaced by 5-methylcytosine The loci numbers of the CpG sites within the sequence (start codon in exon 1 is counted as ϩ1) are shown above (19). All CG dinucleotides in the sequence are SssI methylation sites, and HhaI methylation sites (-G*CGC-) are indicated by the arrows.
at positions Ϫ186, Ϫ188, and Ϫ200. As shown in Fig. 7C, a strong protein complex is formed on both the unmethylated and methylated probe (lanes 1 and 2, asterisk), which was only partly blocked by a 100-fold molar excess of unmethylated or methylated competitor. An additional complex was formed, which was restricted to the methylated probe (lane 2, arrow). Binding of the methylated probe was totally displaced by the methylated competitor but not by the unmethylated one. Thus, this shifted protein complex appeared to be specifically formed with the methylated probe. Against the background of our transfection experiments, these data suggest that methyl-CpGbinding proteins are crucially involved in the regulation of the leptin gene expression.

Demethylation at Distinct CpG Sites in the Leptin Gene Promoter of Human Adipocytes Is Essential for Its Expression-
The connection between cellular differentiation and DNA methylation was established by Taylor and Jones, who converted 3T3 cells to new different phenotypes by treatment with 5-aza-dC and found cellular differentiation associated with specific demethylation on methylated DNA (38,39). The role of DNA methylation in transcriptional regulation of tissue-specific genes has received considerable attention and appears to be the primary silencing mechanism for genes with a CpG-rich promoter (40,41).
Terminal differentiation of preadipocytes induced by agents like insulin, glucocorticoids, and triiodothyronine follows a precisely orchestrated program of transcriptional regulatory events at the promoters of ubiquitous and adipose tissue-specific genes. Leptin, one example of a protein specifically expressed in adipose tissue, acts as a multifunctional hormone and is predominantly delivered by mature adipocytes. In a previous study, we investigated the expression profile of adipocyte-specific genes of the liposarcoma cell line, LiSa-2, following terminal differentiation (27). We found that leptin expression was only induced after preventing DNA methylation by 5-aza-dC. The leptin gene promoter fragment used in this study is associated with a CpG island which includes the regulatory elements for transcriptional activation. DNA derived from cultured human preadipocytes and terminal differentiated adipocytes was bisulfite-modified, whereupon the pattern of CpG methylation in the leptin promoter region was determined by sequencing. 73.4 Ϯ 9.0% of the included 32 CpG motifs were found to be methylated in preadipocytes. High frequency (Ͼ90%) methylated CpG dinucleotides were observed in the leptin promoter region of preadipocytes and in other non-leptin-expressing cells (LiSa-2, human fibroblasts; data not shown) at positions Ϫ33 and Ϫ38 proximal the TATA motif, within the cognate element of C/EBP␣ at Ϫ51, in the Sp1 sites at Ϫ19, Ϫ95, and Ϫ100, and at six CpG sites located in the promoter sequence from nt Ϫ183 to Ϫ204. Following terminal differentiation, the percentage of methyl-CpG in the promoter region dropped to 9.4 Ϯ 4.4%. In addition, the CpG sites at the distinct promoter loci mentioned above, turned out to be unmethylated in adipocytes as well as in 5-aza-dC-treated and differentiated LiSa-2 cells. Recently, Yokomori et al. (42) examined the methylation status of the mouse leptin promoter in 3T3-L1 preadipocytes and adipocytes from positions Ϫ54 to Ϫ159, including eight CpG dinucleotides. These authors found a variable degree of demethylation in seven CpG sites during differentiation, whereas the CpG at Ϫ54, located in the C/EPB␣ binding site of the mouse leptin promoter, remained highly methylated. The corresponding human leptin promoter fragment ranges from nt Ϫ51 to Ϫ160 and includes 13 CpG sites. At variance with the data reported by Yokomori et al. (42), in our experiments all CpG sites were dramatically demethylated upon differentiation. In the leptin promoter of mature human adipocytes, the CpG dinucleotide within the C/EBP␣ motif and the three CpGs within the Sp1 sites were unmethylated. Additional evidence that DNA demethylation acts as an essential process in adipocytic differentiation was provided by showing that, in 3T3-L1 preadipocytes, the expression of the insulin-responsive glucose transporter GLUT4 gene was activated by demethylation of the promoter during further maturation to adipocytes (43).

CpG Methylation at the C/EBP␣ Motif Down-modulates Leptin Promoter Activity, whereas Methylation at Sp1 Sites Has No
Effect-Methylation of CpG sites in the promoter region of tissue-specific genes is most crucial for their expression (41). One mechanism by which methylation may contribute to gene regulation is to reduce or inhibit the binding of transcription factors. However, a minority of known transcription factors have CpG sites in their recognition sequences. In the case of cAMP-response element-binding protein, AP2, E2F, and c-Myc, methylated CpGs in their recognition site preclude the binding of the transcription factor and thereby directly inhibit gene expression (44 -47). Sp1 binding at GC-rich elements was found to be unaffected by CpG methylation (48,49). However, the contribution of methylated CpG in the Sp1 recognition site to T1␣ gene expression was also shown (50). We examined the effect of CpG methylation on transcription factor binding and leptin promoter activity indirectly by transfection experiments using 5Ј-truncated leptin promoter-reporter constructs, which were mock-, HhaI-, or SssI-methylated. The sequence of the leptin promoter construct P-59 used in this study includes four HhaI methylation sites. One is present in the C/EBP␣ binding site. Methylation at this CpG site diminished promoter activity by about 50%. Nuclear protein binding assays on unmethylated and at this CpG site methylated C/EBP␣ probe yielded protein probe complexes with a diminished binding affinity to the methylated probe. These observations indicate that C/EBP␣binding is affected by CpG methylation at its cognate element and is associated with regulation of gene expression. The importance of the CpG sequence in the C/EBP␣ recognition site for factor binding and leptin expression was recently shown (51). Base change from CG to AT in the C/EBP␣ recognition sequence abolished factor binding and caused a sharp decrease in promoter activity.
In transfection experiments using the promoter construct P-106, we here confirmed that cooperative action of Sp1 and C/EBP␣ achieved maximal promoter activity. The expression markedly decreased to about 5% when the SssI-methylated reporter construct was used. Thus, we assume that the loss of promoter activity is not a consequence of inhibited Sp1 binding to methylated Sp1 recognition sites but is due to binding of methyl-CpG-sensitive repressor proteins on methylated CpGs proximal to the TATA motif.
Two Highly Methylated CpG Motifs in the Human Preadipocyte Leptin Promoter Recruit Methyl-CpG-binding Proteins, Which Silence Leptin Gene Expression-We have shown that leptin gene promoter activity, as determined by transfection experiments, is practically silenced using the SssI-methylated promoter-reporter construct P-258, the SssI-or HhaI-methylated construct P-199, and even the SssI-methylated core promoter construct P-35. In addition to the high frequency methylated Sp1 and C/EBP␣ binding sites in the human leptin promoter mentioned above, the CpGs at Ϫ33 and Ϫ38 proximal to the TATA motif and a region including the HhaI methylasesensitive CpGs at Ϫ186, Ϫ188, Ϫ202, and Ϫ204 as well as the CpG at Ϫ200 were highly methylated in preadipocytes. In sharp contrast, the corresponding CpGs were found to be unmethylated in mature adipocytes. In nuclear protein binding assays on unmethylated and specifically methylated DNA probes corresponding to the two distinct promoter regions, we found proteins with specific binding affinity toward the methylated promoter sites. These observations suggest that methyl-CpG-binding proteins are operative in leptin gene repression.
Considerable attention has been focused on proteins that bind to promoter-proximal methylated DNA and function as transcriptional repressors. Two chromosomal protein complexes with high affinity toward methylated DNA have been identified (52,53). Methyl-CpG-binding protein-1 is a protein complex composed of 10 major polypeptides. The MBD2 subunit binds specifically to a variety of methylated DNA sequences that contain at least 12 symmetrically methylated CpGs and represses transcription by recruiting nucleosomeremodeling histone deacetylase (54,55). Methyl-CpG-binding protein-2 is reported to bind specifically to a single methylated CpG pair and to be concentrated in the centromeric heterochromatin (56,57). It has been shown that the transcriptional repressing domain of methyl-CpG-binding protein-2 binds components of the mSin3A-histone deacetylase complex and represses transcription via histone deacetylation and remodeling of the chromatin structure (58). An alternative pathway of transcriptional repression, aside from the recruitment of histone deacetylase, was reported, in which the transcriptional repressing domain interacts with the TFIIB of the basal transcriptional machinery (59). This mechanism might also be operative in the repression of leptin promoter activity once CpGs proximal to the TATA motif are methylated. Furthermore, the methylated motif, -CGCGGCG-, was found in the Xist promoter, which binds a methyl-CpG-binding protein with high affinity (60). This motif is strikingly similar to the recognition FIG. 7. Details of EMSA on methylated (M) and unmethylated (U) 32 P-labeled DNA probes of human leptin promoter regions and corresponding competitors. A, EMSA was performed using rat liver nuclear protein extract and probes ranging from nt Ϫ19 to Ϫ43 of the human leptin promoter TATA-box region, including two CpGs at Ϫ33 and Ϫ38. Specific complexes formed on unmethylated and methylated probes are indicated by an asterisk, and specific complexes formed only on methylated probe are indicated by the arrows. B, EMSA was carried out using rat liver nuclear protein extract and probes ranging from Ϫ39 to Ϫ64 of the human leptin promoter including the C/EBP␣ recognition site and two CpGs at Ϫ51 and Ϫ62. Specific complexes formed on unmethylated and methylated probes indicated by the arrows. C, EMSA was carried out using nuclear protein extract derived from undifferentiated LiSa-2 cells and MBD-200 probes ranging from base pair Ϫ178 to Ϫ210, including six CpGs at Ϫ183, Ϫ186, Ϫ188, Ϫ200, Ϫ202, and Ϫ204. Specific complexes formed on unmethylated and methylated probes are indicated by an asterisk, and specific complexes formed only on methylated probe are indicated by an arrow. Competitions were run with a 10-or 100-fold molar excess of unlabeled unmethylated or methylated probe (competitor) in the reaction mixture. Lanes 1,7,8,9, and 10, EMSA with unmethylated probe; lanes 7 and 8, competition by adding unmethylated competitor; lanes 9 and 10, competition by adding methylated competitor. Lanes 2-6, EMSA with methylated probe; lanes 3 and 4, competition by adding unmethylated competitor; lanes 5 and 6, competition by adding methylated competitor.
site of Kaiso, a novel methylation-dependent transcriptional repressor. Kaiso is associated with methyl-CpG-binding protein-1 and requires two symmetrically methylated CpG dinucleotides in its recognition sequence (61). A quite similar sequence, -CGCGCCG-, is located at Ϫ182 to Ϫ188 in the upstream region of the human leptin promoter, which we define as a binding site for a methyl-CpG-binding protein. The biochemical characterization of the methyl-CpG-binding proteins and histone-modifying proteins probably involved in transcriptional regulation of the leptin gene in the preadipocyte/ adipocyte model will be addressed in future work.