DNA demethylation of PPARγ promoter in 3T3-L1 adipocyte differentiation
Since PPARγ is the key regulator of adipogenic differentiation, its transcription is highly restricted in most of cell types [4]. Real time reverse-transcription polymerase chain reaction (RT-PCR) of total PPARγ mRNA revealed that its expression is strongly inhibited in NIH/3T3 fibroblasts and somewhat less so in 3T3-L1 preadipocytes (Figure 1a). Expression of PPARγ mRNA in NIH/3T3 cells was approximately 3% of that observed in 3T3-L1 preadipocytes. In contrast, expression of PPARγ mRNA in differentiated 3T3-L1 adipocytes (day 6) was approximately 24 times greater than that observed in 3T3-L1 preadipocytes. To examine the contribution of epigenetic factors, such as DNA methylation, on the transcriptional regulation of PPARγ expression, we first compared the methylation status of the PPARγ promoter region in NIH-3T3 cells, preadipocytes (day 0) and adipocytes (day 6) using the bisulfite sequencing method. There are seven CpG methylation sites flanking the transcription start site (TSS) of the murine PPARγ2 gene (Figure 1b). We examined the methylation status of these seven sites in promoter fragments isolated from 24 cells of each cell type (Figure 1c). In NIH/3T3 cells, which express a low level of PPARγ mRNA, almost all of these sites were methylated. A significant fraction of these sites were also methylated in 3T3-L1 preadipocytes, although the extent of methylation was lower than that observed in NIH/3T3 cells. In contrast, the CpG sites located upstream of the TSS in 3T3-L1 adipocytes were mostly demethylated after differentiation to adipocytes. These results indicated that methylation of the CpG sites upstream of the TSS correlates inversely with PPARγ mRNA expression. Thus, CpG methylation of these sites in the PPARγ promoter might contribute to silencing of its expression. The two CpG sites located downstream of the TSS, at positions +89 and +158, were methylated in each cell type suggesting that methylation of these CpG sites does not contribute to the regulation of PPARγ mRNA expression.
Contribution of DNA methylation to PPARγ expression
To further evaluate a causal relationship between methylation of the PPARγ promoter and the expression of PPARγ mRNA, we examined the effect of 5'-aza-cytideine (5'-aza-C), an inhibitor of DNA methylation, on PPARγ mRNA expression in NIH/3T3 cells. After 48 h treatment with 5 and 10 μM 5'-aza-C, PPARγ expression was analyzed quantitatively by real time RT-PCR. As shown in Figure 2a, the expression of PPARγ mRNA increased following 5'-aza-C treatment in a dose-dependent manner. In contrast, treatment of cells with trichostatin A (TSA), an inhibitor of histone deacetylases, for 48 h, had no effect on PPARγ mRNA expression (Additional file 1). These results suggest that DNA methylation is an important mechanism of epigenetic regulation of the expression of PPARγ and that the role of DNA methylation is dominant to that of histone acetylation, at least under these experimental conditions tested here.
To further confirm the effect of DNA methylation on the PPARγ promoter region, we performed luciferase reporter assays in NIH/3T3 cells following transfection of a reporter plasmid comprising 1 kb of the 5'-upstream region of PPARγ containing five CpG sites (at positions -437 to -60, Figure 1b) placed upstream of a luciferase reporter gene. Transfection of the PPARγ reporter construct led to the expression of 3.7-fold more luciferase activity relative to that of the empty vector (Lacking a promoter upstream of the reporter gene) (Figure 2c). In contrast, in vitro DNA methylation of the reporter construct prior to transfection reduced the expression of the luciferase reporter to a level similar to that observed when using the empty vector (Figure 2c). These results further suggest that transcription of the PPARγ gene is regulated through CpG methylation of the promoter.
The same 5'-aza-C treatment and luciferase assays were next performed on 3T3-L1 preadipocytes (day 0) and adipocytes (day 4). In contrast to the results observed in NIH/3T3 cells, we observed no increase in PPARγ mRNA in 5'-aza-C-treated preadipocytes (Figure 2b). This result suggested that, in preadipocytes, DNA demethylation alone is not sufficient for the activation of the PPARγ promoter. Similar results were obtained from following luciferase reporter assays in preadipocytes (Figure 2c). The unmethylated PPARγ reporter construct did not give rise to increased luciferase expression and instead expressed a similar level of luciferase as the promoterless construct in preadipocytes. Luciferase expression from the promoterless construct in preadipocytes was similar to the level observed in NIH/3T3 cells. These results suggested the existence of an inhibitory mechanism that represses expression from the demethylated PPARγ promoter in preadipocytes, or alternatively, an activating mechanism that induces the expression of the demethylated PPARγ promoter in NIH/3T3 cells.
Luciferase expression from the in vitro methylated reporter construct was strongly repressed in preadipocytes (Figure 2c). Indeed, luciferase reporter expression from the methylated PPARγ construct was further reduced to 18% of that of the promoterless control construct, suggesting that preadipocytes inhibit transcription from the methylated PPARγ promoter more robustly than NIH/3T3 cells. These data further support the idea that the expression from the PPARγ promoter is under the control of its promoter methylation.
As was expected, luciferase reporter expression from the unmethylated PPARγ reporter vector was very efficient in differentiating adipocytes (day 4), in which the endogenous PPARγ gene is activated (Figure 2c). In contrast, luciferase reporter expression from the in vitro methylated promoter construct remained low in adipocytes, despite the presence of a suitable environment for activation of the transcription of the PPARγ promoter. These results further suggest that DNA methylation contributes to the regulation of expression of PPARγ mRNA. In addition, although there are structural differences between endogenous chromatin and exogenous reporter plasmids, DNA methylation of the PPARγ promoter may be a dominant regulatory mechanism that can override activation of the PPARγ promoter by other transcription factors during adipogenesis.
Kinetic analysis of PPARγ promoter demethylation
For further analysis of PPARγ methylation in adipocytes, we next investigated the time course of promoter demethylation during adipocyte differentiation of 3T3-L1 cells. Bisulfite-converted PCR amplicons of the PPARγ promoter of the cells were digested with the restriction enzyme, HpyCH4IV. The level of demethylation was estimated by the cutting efficiency every 24 h (see Methods for details). We could detect demethylation of the promoter immediately following induction of differentiation and the demethylation increased gradually until it reached a plateau on day 6 (Figure 3a).
The level of PPARγ mRNA was also monitored by real time RT-PCR (Figure 3b). Similar to the time course of demethylation of the promoter, the amount of PPARγ mRNA gradually increased until it reached a plateau on day 6, although the starting time of the increase in mRNA appeared to lag 1 to 2 days behind the start of promoter demethylation and differentiation induction. This synchronization supports the idea that transcriptional activation of the PPARγ gene is regulated by DNA methylation/demethylation. The observed lag time probably indicates that promoter demethylation is not the only factor controlling PPARγ expression. The differentiation stimulus induces the recruitment of several transcriptional regulators that increase the expression of PPARγ, including C/EBPs, sterol regulatory element binding protein (SREBP), and SWI/SNF family chromatin remodeling enzymes to the PPARγ promoter, and the observed lag in PPARγ expression might be due to a requirement to recruit these factors to the promoter so as to activate transcription following DNA demethylation [1, 19, 20].
During adipogenesis of 3T3-L1 cells, genomic DNA is newly synthesized during two cycles of mitosis termed the mitotic clonal expansion (MCE) [23]. The replication of methylated DNA produces hemimethylated CpG sites, which could cause loss of DNA methylation if those sites are not remethylated by Dnmt1, the maintenance DNA methyltransferase [17]. We next investigated whether the demethylation of the PPARγ promoter during differentiation is caused by passive demethylation during DNA replication through MCE. To do this, we compared cell growth and demethylation of the PPARγ promoter during differentiation (Figure 3c). The number of the cells increased immediately following induction of differentiation, expanding fourfold by day 2, suggesting that two cycles of cell division had occurred. In contrast, as shown in Figure 3a, DNA demethylation increased gradually up day 6, and we observed no immediate demethylation that corresponding to the increase in cell number on approximately day 2. This result demonstrates that PPARγ promoter demethylation is not a passive process caused by MCE, but that it is instead an active process of epigenetic regulation of the transcriptional activity of the PPARγ promoter.
Chromatin immunoprecipitation assays of the PPARγ promoter during 3T3-L1 adipocyte differentiation
Promoter DNA methylation represses the transcription of a downstream gene both directly and indirectly. Methylated CpG sites within the promoter itself can directly inhibit the binding of transcription factors due to steric hindrance and thereby repress transcription. Alternatively, methylated CpG binding domain (MBD) proteins, which specifically bind to methylated CpG residues, recruit other enzymes that modify histone tails or chromatin structure, so as to indirectly create a repressed state of chromatin [16–18]. To assess the protein interactions on the methylated PPARγ promoter region, we performed chromatin immunoprecipitation (ChIP) assays to evaluate the binding of certain MBD proteins to the PPARγ promoter. The ChIP assay revealed that methyl CpG binding protein 2 (MeCP2) was associated with the methylated PPARγ promoter in preadipocytes (day 0) (Figure 4). We were unable to detect the binding of other MBDs, such as MBD1 and MBD2a, to the promoter in preadipocytes (data not shown). In contrast, we detected the binding of MBD2a to the promoter in NIH/3T3 cells, suggesting that MBD binding to the PPARγ promoter differs between NIH/3T3 and 3T3-L1 cells (Additional file 2).
Dissociation of MeCP2 from the promoter was nearly complete within 1 to 2 days following the induction of differentiation (Figure 4), although promoter demethylation continued gradually until it reached a plateau on day 6 (Figure 3a). An increase in the level of histone H3 acetylation in the PPARγ promoter region has been reported until day 2 [20], and MeCP2 has been shown to recruit the protein complex including histone deacetylases to methylated CpG sites [18]. Thus, the increase in histone acetylation is likely caused by dissociation of MeCP2 on days 1 to 2 following the induction of differentiation. This result further suggests that DNA methylation plays a greater role than histone acetylation on the regulation of the expression of PPARγ mRNA.
We next evaluated other types of histone modifications in 3T3-L1 cells following the induction of differentiation. On day 0, we observed dimethylation of histone H3 on lysine 9 (H3K9me2), which represents a repressed state of chromatin, on the PPARγ promoter. This dimethylation disappeared within 1 day of the induction of differentiation (Figure 4). The demethylation of H3K9me2 occurred at a stage in the differentiation process that preceded DNA demethylation, the expression of PPARγ and even MCE (Figure 3), and so therefore was not synchronous with PPARγ expression and therefore was not responsible for the induction of its expression. This result suggests that demethylation of H3K9me2 is insufficient to induce the expression of PPARγ mRNA. In contrast, the level of dimethylation of lysine 4 on histone H3 (H3K4me2), which represents an activated state of chromatin, was low prior to the induction of differentiation, but increased gradually following induction (Figure 4). The kinetics of this gradual increase appeared to mirror that of the DNA demethylation and the increase in mRNA expression, suggesting that H3K4 dimethylation correlates well with DNA demethylation.
DNA methylation and mRNA expression of PPARγ in white adipose tissues of mouse models of diabetes
PPARγ not only acts to induce adipogenesis, but also to maintain the functional phenotype of adipocytes. Thus, the expression and epigenetic regulation of PPARγ mRNA might also be important for the maintenance of the adipocyte phenotype, and any defect in this regulation could become a pathogenic factor in metabolic syndromes. In other words, changes in the epigenetic status of the PPARγ gene might be observed in adipose tissues under pathogenic conditions. To test this possibility, we analyzed methylation of the PPARγ promoter in white adipose tissues (WAT) and compared the methylation profiles of the promoter in wild-type mice and mouse models of diabetes. WAT are classified into two main types: subcutaneous adipose tissue (SAT) and visceral adipose tissue (VAT). Visceral obesity has been linked strongly to diabetic insulin resistance [24]. Therefore, we compared the methylation status of the PPARγ promoter in both types of WAT.
We first compared promoter methylation in WAT isolated from 10-week-old wild-type (WT) and +Lepr
db
/+Lepr
db(db/db) mice. The db/db mouse is a well established model of diabetes that contains a homozygous mutation in the leptin receptor gene. This mouse exhibits a phenotype similar to type 2 diabetes mellitus owing to hyperphagia and disrupted metabolism of adipocytes. We prepared genomic DNA from the two types of WAT: the inguinal part of the SAT and the visceral, epididymal adipose tissues (EAT). The genomic DNA was treated with sodium bisulfite, and the methylation status of two CpG sites within the PPARγ promoter, at positions -437 bp and -247, was estimated based on the efficiency of HpyCH4IV restriction enzyme digestion (see Methods for detail) (Figure 5a, b). As shown in Figure 5a, the fraction of methylated CpG sites at both positions -437 bp and -247 in SAT was reduced in db/db mice compared to WT mice. The -247 site in particular was methylated to a much lesser extent in db/db mice (12.7%) than in WT mice (approximately 50%) (Figure 5a). Since the SAT in db/db mice was greatly enlarged relative to that in WT mice because of hyperphagia (approximately × 6.7 by weight, Additional file 3), the reduced level of methylation of the CpG sites in the PPARγ promoter in the SAT of db/db mice relative to the level in WT mice suggests a possible enrichment of differentiated, PPARγ-expressing adipocytes in the SAT of db/db mice.
Interestingly, despite the hypertrophy of the EAT in db/db mice (× 6.0 compared to WT), methylation of the CpG at position -437 in the PPARγ promoter was increased somewhat in db/db mice relative to the level observed in WT mice (Figure 5b). Approximately 36% of the cytosines at the site were methylated in WT mice, whereas 50% of such sites were methylated in db/db mice. The level of expression of PPARγ2 mRNA in the SAT and EAT differed between WT and db/db mice to an extent that mirrored qualitatively the differences in methylation (Figure 6a, b, WT and db/db). Whereas the expression of PPARγ2 mRNA in the SAT of db/db mice was four times greater than that in WT mice, the level of PPARγ mRNA in the EAT of db/db mice was about one-third of that observed in WT mice, which correlated inversely with the level of methylation within the PPARγ promoter, particularly with that at the CpG at -437 bp.
To our surprise, the hypertrophic EAT in db/db mice, which likely contains a large number of differentiated adipocytes, expressed a decreased level of PPARγ mRNA compared to that observed in the smaller EAT in WT mice, presumably due to the greater extent of methylation of the PPARγ promoter in the EAT of db/db mice. A reduction in the level and/or activity of PPARγ has been previously suggested to be linked to the development of diabetic symptoms such as insulin resistance [10, 11]. These results suggest that enhanced methylation of the PPARγ promoter and the concomitant reduction of PPARγ mRNA in the EAT may be linked causally to the diabetic phenotype induced by obesity in db/db mice.
To further characterize this correlation, we next analyzed methylation of the PPARγ promoter in adipocytes from a mouse model of diet-induced obesity (DIO), which are WT mice fed with a high-fat diet from 4 to 20 weeks old that subsequently present a type 2 diabetes-like phenotype. Although DIO mice also had an excessive amount of SAT owing to their diet (approximately × 7.6 in weight compared to the SAT of WT mice, Additional file 3), the level of methylation of the PPARγ promoter in the SAT of these mice was similar to that observed in the SAT of 20-week-old WT mice (Figure 5c). The level of expression of PPARγ2 mRNA was similar in the SAT of DIO and WT mice (Figure 6a).
However, cells of the hypertrophic EAT of DIO mice (approximately × 3.1 compared to the EAT of WT mice) also exhibited enhanced methylation of the CpG at position -437 than was observed in cells of the smaller EAT of WT mice (Figure 5d), as had been observed in cells of the EAT of db/db mice relative to that observed in WT mice (Figure 5b). Approximately 40% of the cytosines at the site were methylated in WT mice, whereas 56% of such sites were methylated in DIO mice. Consistent with these findings, the expression of PPARγ2 mRNA was similarly reduced in the EAT of DIO mice (Figure 6b). These results suggest that, as in the EAT of db/db mice, PPARγ mRNA expression is reduced owing to promoter DNA methylation in the enlarged EAT of mice fed the high-fat diet, and further support the possibility that increased methylation of the PPARγ promoter in the obese EAT contributes to the pathogenesis of diabetes.