Introduction

Adipose tissue has a key role in the regulation of energy homeostasis, not only through energy storage but also endocrine function such as adipokine secretion.1 Its dysfunction is closely associated with various metabolic disorders including obesity and type 2 diabetes. It is primarily composed of adipocytes, which differentiate from mesenchymal stem cells. The lineage commitment of mesenchymal stem cells to adipocytes is driven by the activation of master transcriptional regulators such as CCAAT/enhancer-binding proteins (C/EBPα, C/EBPβ and C/EBPδ) and peroxisome proliferator-activated receptor γ (PPARγ), resulting in subsequent expression of adipocyte-specific genes.2, 3 PPARγ is a member of the nuclear receptor superfamily of ligand-activated transcription factors,3, 4 cooperating with C/EBPα to act as a positive regulator in adipogenesis.5, 6 Heterozygous PPARγ-deficient mice showed the reduction in fat mass and adipocyte cell size,7 while overexpression of PPARγ in fibroblasts and preadipocytes was sufficient for adipogenesis, even in the absence of C/EBPα.8, 9 Consistently, pharmaceutical PPARγ ligands such as thiazolidinedione have been shown to have an impact on the regulation of gene expression in various metabolic pathways and function as an insulin sensitizer.10 These findings indicate that PPARγ-mediated control of adipogenesis may influence physiological body fat mass as well as metabolism and may be related with metabolic disorders such as obesity. Thus, understanding the molecular mechanisms by which PPARγ regulates adipogenesis may be critical to develop therapeutic strategies for the metabolic pathophysiology of obesity.

Recent evidence has shown that cell differentiation required proper epigenomic transition, besides serial coordination of adipogenic transcriptional factors.11 Epigenomic transition for cell differentiation and lineage decision takes place via changes of epigenetic patterns such as DNA methylation and histone modifications.11 DNA methylation is found in cytosines of the dinucleotide sequence CpG in mammals and has been demonstrated to be responsible for the stable maintenance of gene expression patterns during development or differentiation.12 Mammalian DNA methyltransferases (DNMTs) are responsible for the acquisition of DNA methylation during gametogenesis, embryogenesis and somatic tissue development,11, 13, 14 in which the pattern of DNA methylation changes and consequently differentiated cells develop a stable DNA methylation pattern that regulate tissue-specific gene transcription.15 Generally, DNA methylation regulates gene expression by recruiting proteins involved in gene repression or by inhibiting the binding of transcription factors to DNA.15, 16

Compared with establishment of DNA methylation, its removal process has not been well delineated. Recently, however, Tet methylcytosine dioxygenase (TET) protein family, composed of the three members TET1, TET2 and TET3, have the capacity to convert 5-methylcytosine (5-mC) to 5-hydroxymethylcytosine (5-hmC) of DNA that is an intermediate for oxidative demethylation.17, 18 They are Fe(II) and a-ketoglutarate (α-KG)-dependent dioxygenases that successively convert 5-mC to 5-hmC, 5-formylcytosine and 5-carboxycytosine of the DNA.17, 18 Genome-wide studies on embryonic stem (ES) cells have revealed that 5-hmC is enriched in gene bodies, promoter regions and CpG islands of actively transcribed genes in ES cells.19, 20

Since adipogenesis is one of the most dynamic processes of cellular differentiation, the analysis of genome-wide or gene-specific DNA demethylation pattern during this process may clarify its underlying mechanisms. The role of histone marks during adipocyte differentiation has been actively explored,21, 22, 23, 24, 25 but DNA methylation or demethylation remains poorly investigated. A few studies have indicated gene-specific DNA methylation changes during adipogenesis.26, 27 For example, the promoters of several adipogenic loci such as Pparγ, Leptin (Lep) and Glucose transporter 4 (Glut4) are hypomethylated during adipogenesis and becomes expressed.26, 27 How these genes become demethylated remains poorly investigated, but is likely mediated by 5-hmC. A genome-wide analysis of 5-hmC during adipogenesis suggested that the level of 5-hmC was more abundant in differentiated 3T3-L1 adipocyte compared to preadipocyte, and 5-hmC enrichment was closely associated with distal target regions of PPARγ.28 It suggests that TETs may have a crucial role in adipogenesis. In the current study, we closely investigate how the dynamics between DNA methylation and hydroxymethylation drives differentiation of 3T3-L1 cells to adipocytes.

Materials and methods

Cell culture and adipocyte differentiation

3T3-L1 preadipocyte cells (courtesy of Dr Yuri Kim lab at the Ewha Womans University) were maintained in Dulbecco’s modified Eagle’s medium (Welgene, Daegu, Korea) with 10% calf serum (Gibco, Gaithersburg, MD, USA), following the test for mycoplasma contamination. The cells were differentiated according to the protocol in Green et al.29 Briefly, the cells were continuously cultured for 2 days after reaching confluence and were stimulated with a differentiation cocktail containing 10 μg ml−1 insulin (Sigma-Aldrich, St Louis, MO, USA), 1 μM dexamethasone (Sigma-Aldrich) and 500 μM isobutyl-methyl xanthine (Sigma-Aldrich) in Dulbecco’s modified Eagle’s medium with 10% fetal bovine serum (ATLAS, Fort Collins, CO, USA), which were then incubated for 6–12 days with media replacement every 2 days.

Oil-Red-O staining

For Oil-Red-O staining, cells were fixed with 10% formalin and stained with Oil-Red-O solution (Sigma-Aldrich). For quantification of neutral triglycerides and lipids, Oil-Red-O staining was dissolved in 100% isopropanol (Junsei Chemical Co, Ltd, Tokyo, Japan) and the absorbance was measured at 510 nm.

Vitamin C and fumarate treatment

Confluent 3T3-L1 cells were treated with 100 μM diethyl fumarate (Sigma-Aldrich) or 50 μg ml−1 of ascorbic acid (Sigma-Aldrich). After 2 days, the cells were induced to differentiate into adipocyte with Dulbecco’s modified Eagle’s medium with 10% fetal bovine serum containing the differentiation cocktail. For ascorbic acid treatment, we applied the differentiation cocktail without insulin in order to slow down differentiation so that a possible difference caused by enhanced activities of TET proteins could be detected. Treatment of diethyl fumarate and ascorbic acids was maintained while changing the medium for differentiation.

Gene silencing using small interfering RNA

For small interfering RNA (siRNA) knockdown of Tet1 or Tet2, transfections were performed with Dharmafect1 (Dharmacon, Lafayette, CO, USA), using siRNAs targeting Tet1 (Dharmacon; M062861-01), Tet2 (Dharmacon; M058965-01) or non-targeting control (Dharmacon; D001810-10). 3T3-L1 cells were transfected with control siRNA and Tet1- or Tet2-specific siRNAs before adipocyte differentiation. After 2 days, cells were induced to differentiate with a cocktail of insulin, dexamethasone and IBMX (DMI), and kept in culture for 6 days. After the siRNA transfection, the levels of Tet1 and Tet2 were quantified by quantitative reverse transcriptase polymerase chain reaction (qRT-PCR).

Quantitative RT-PCR analysis

RNA was isolated using the RNeasy mini kit (QIAGEN, Hiden, Germany). After incubating with DNase I (Thermo Scientific, Waltham, MA, USA), complementary DNA was synthesized using RevertAid reverse transcriptase (Thermo Scientific). qPCR was performed using SYBR green mix (QIAGEN), and data were normalized to the Nono gene expression level, which has been reported to be a proper internal control during 3T3-L1 differentiation according to Arsenijevic et al. (2012).30 The primer pairs used are available in Table 1.

Table 1 Sequences of primer pairs used for quantitative RT-PCR analysis, MassARRAY and (h)MeDIP
Full size table

MassARRAY quantitative analysis

Genomic DNA from 3T3-L1 cells was isolated using the DNeasy Blood and Tissue Kit (QIAGEN). Bisulfite conversion of the genomic DNA was performed with the EZ DNA CT Conversion Reagent (Zymo Research Corporation, Irvine, CA, USA). PCRs were carried out with bisulfite-treated template DNA using Hotstart PCR polymerase (QIAGEN). Bisulfite amplification primers for specific regions of Pparγ, Cebpα and Fabp4 were designed using the EpiDesigner application (Sequenom Inc., San Diego, CA, USA). The primer pairs contain a T7-promoter tag to allow further in vitro transcription. Information of primer pairs used is available in Table 1. By sequencing analysis, partially methylated CpG sites were identified and quantified using the MassARRAY Compact System following the MassCLEAVE protocol (Sequenom Inc.). The resulting methylation calls were analyzed with the EpiTYPER software (Sequenom Inc.) to generate quantitative CpG methylation results, presented in the figures as the fraction of methylation. Importantly, DNA methylation analysis using bisulfite conversion,for example, applied in the MassARRAY methodology, does not distinguish between 5-mC and 5-hmC, but rather represents a summary of both marks. However, it is well documented that 5-mC is several orders of magnitude more common than 5-hmC.

Immunoprecipitation of methylated or hydroxymethylated DNA (MeDIP and hMeDIP)

Genomic DNA from cultured 3T3-L1 cells was isolated using the DNeasy Blood and Tissue Kit (QIAGEN) and sonicated to produce fragments ranging from 150 to 180 bp. Two micrograms of fragmented DNA were immunoprecipitated by methylated DNA-specific immunoprecipitation (MeDIP kit; Diagenode, Belgium, Europe) or hydroxymethylated DNA-specific immunoprecipitation (hMeDIP Kit; Diagenode). qPCR was used for quantifying the enrichment of methylated or hydroxymethylated DNA at the Pparγ locus. The sequence of the primer pair is available in Table 1.

Gas chromatography mass spectrometry detecting cytosine, methylcytosine and hydroxymethylcytosine

Isolated DNA (1 μg per sample) was hydrolyzed in 90% (V/V) formic acid at 100 °C for 1 h. Formic acid was removed under vacuum at 45 °C and samples were sealed air tight and stored at −80 °C until derivatization. Derivatization was performed as previously described.31 Briefly, dried samples were derivatized by adding 20 μl of BSTFA+TCMS (99:1; Sigma Aldrich; 3-3148, Lot: LC05232V) and 20 μl acetonitrile (VWR International, Darmstadt, Germany) at 30 °C for 30 min in a glass micro insert. Five microliters aliquots were directly subjected to gas chromatography mass spectrometry (GC–MS) analysis. The analysis was carried out on an Agilent 6890 GC/ 5973 MS System. Samples were injected into a HP-5 MS-fused silica column (30 m × 0.25 mm; 0.25 μm film thickness of the 5% phenyl 95% dimethylpolysiloxane stationary phase; Agilent J&W Scientific, Waldbronn, Germany) in pulsed splitless mode. The MS was operated in 70 eV electron impact selected ion monitoring mode for analysis. Dwell times for all ions were set to 200 ms. Characteristic ions of 5-hmC were determined by collecting a scan spectrum within the range of 50–400 m/z.

Statistical analysis

The data, except the results from GC–MS, were presented as means±s.d. values of the duplicates, triplicates and sextuplets for expression analysis, methylation analysis and Oil-Red-O staining, respectively, and were analyzed by two-sided Student’s t-test. P<0.05 was considered statistically significant. The statistical analysis for the results from GC–MS was omitted since the analysis was based on the pooled samples.

Results

DNA methylation gradually decreased at the Pparg locus during adipogenesis

First, we established the 3T3-L1 adipogenesis model according to the custom standard protocol32 and confirmed differentiation using the mRNA expression level of adipogenic molecular markers (Figures 1a–c). mRNA expression of Pparγ, Cebpα and Fabp4 genes was significantly induced compared to the preadipocyte at the latest as differentiation d3 (P<0.01). To monitor DNA methylation changes during adipogenesis, we first measured the global level of DNA methylation by quantifying DNA methylation at diagnostic repetitive elements (LINE, SINE B1, IAP and microsatellite), which represent a reliable surrogate measure of heterochromatic sites subject to extensive DNA methylation. Those loci did not show any significant change in DNA methylation during the process of adipogenic differentiation in 3T3-L1 cells (data not shown). Next, the gene-specific DNA methylation was analyzed using MassARRAY assays at major adipogenic genes including Cebpα, Pparγ and Fabp4. The locations of DNA methylation analysis were selected for putative regulatory sequences such as the putative promoter and enhancer based on genome-wide identification with histone markers. The DNA methylation at the putative enhancer of the Pparγ locus marginally decreased at differentiation d6 (P=0.08) and significantly decreased at differentiation d9 and d12 (Figure 1d, P<0.01). The average percentage of DNA methylation was 93% in preadipocyte, while that in adipocyte differentiation day 12 was decreased to 44%. The methylation of the Fabp4 gene also significantly decreased at differentiation d9 and d12 (Figure 1f, P<0.01); however, the magnitude of the changes was relatively small. The methylation of Cebpα, however, showed consistent levels during adipogenesis (Figure 1e).

Figure 1
figure 1

Changes in relative expression of Pparγ, Cebpα and Fabp4 genes and DNA methylation at their loci during adipogenic differentiation of 3T3 -L1 cells. DNA and RNA samples were collected after adipogenic differentiation of 3T3 -L1 cells at different time points. Relative expression levels of adipogenic markers, Pparγ, Cebpα and Fabp4 genes against Nono gene measured by qPCR in triplicates during differentiation (a–c). DNA methylation at the Pparγ, Cebpα and Fabp4 loci quantified by MassARRAY in duplicates during differentiation (d–f), presented as a fraction of methylation. The data were analyzed by Student's t-test and **P<0.01 indicates significance at days versus preadipocytes (pre); d0, d3, d6, d9, d12 indicate days after adipogenesis induction.

Full size image

Tet genes contribute to the demethylation of the Pparg locus during adipogenesis

To investigate which adipogenesis-related factor contributes to the demethylation of Pparγ region, we obtained a publically available gene expression data set upon 3T3-L1 differentiation from Serandour et al.28 The gene set enrichment and functional annotation analysis using DAVID bioinformatics resources (version 6.7) indicated Tet1 as one of the highly upregulated candidate genes in adipogenesis (data not shown). Among TET family proteins, TET3 is lowly expressed in adipose tissue, while TET1 and TET2 are highly expressed according to The Human Protein Atlas (www.proteinatlas.org). Gene expression of Tet1 and Tet2 genes was confirmed in our adipogenesis model with 3T3-L1. Both genes were significantly induced during adipocyte differentiation from differentiation d6 (P<0.05), similar to the pattern of other adipogenic molecular markers such as Pparγ or Cebpα (Figures 2a and b). As TET proteins convert 5-mC in DNA into 5-hmC, we quantified the global level of 5-hmC during adipogenesis using GC–MS. The global level of 5-hmC tended higher in differentiated cells compared to preadipocytes (Figure 2c), although the level was not directly correlated with the expression level of Tet genes. Since we observed a significant drop in crude DNA methylation at the Pparγ locus (Figure 1d, P<0.05), we then measured the 5-mC and the 5-hmC levels specific at the Pparγ locus with MeDIP and hMeDIP using 5-mC and 5-hmC antibodies, respectively. While 5-mC decreased (Figure 2d), 5-hmC significantly increased at the Pparγ locus after differentiation d12 (Figure 2e, P<0.01).

Figure 2
figure 2

Changes in relative expression of Tet1 and Tet2 genes and 5-mC and 5-hmC in the global and Pparγ locus-specific levels during adipogenic differentiation of 3T3-L1 cells. Relative expression of Tet1 and Tet2 genes against the Nono gene measured by qPCR in triplicates during differentiation (a, b). Global 5-hmC level during differentiation measured by GC–MS (c). A locus-specific 5-mC and 5-hmC at the Pparγ locus quantified by MeDIP (d) and hMeDIP (e), respectively, in duplicates during adipogenesis. The data were analyzed by Student's t-test, and *P<0.05 and **P<0.01 indicate significance at days versus preadipocytes (pre), preadipocyte; d0, d3, d6, d9, d12 indicate days after adipogenesis induction.

Full size image

TETs are required for adipogenesis

We continued by testing whether TET1 and TET2 are required for adipogenesis by depleting them with specific siRNAs in 3T3-L1 preadipocytes prior to differentiation. The knockdown of Tet1 or Tet2 expression led to significantly reduced levels of lipid droplet measured by Oil-Red-O staining, compared to cells with non-target control siRNA treatment (Figure 3a). To confirm knockdown efficiency, their expression levels were estimated by quantitative RT-PCR. As expected, the mRNA levels of Tet1 and Tet2 were markedly reduced down to less than 20% upon siRNA treatment (Figures 3b and c, P<0.01). To check whether the knockdown of Tet1 or Tet2 was sufficient to downregulate the expression of adipogenic genes, the mRNA levels of Pparγ and Fabp4 were also measured by quantitative RT-PCR analysis at the differentiated adipocyte stage (day 6). Indeed, the expression of Pparγ was significantly reduced both after the knockdown of Tet1 and Tet2; P<0.01 and P<0.05 respectively (Figure 3d). Furthermore, the decreased expression of Pparγ subsequently resulted in the significant decline of Fabp4 expression; P<0.01 (Figure 3e). The expression levels of Cebpα (which acts upstream of PPARγ) were not significantly changed throughout 6 days after the adipogenic induction in Tet1- or Tet2-silenced 3T3-L1 cells (data not shown).

Figure 3
figure 3

Impact of Tet1 and Tet2 knockdown on adipogenesis via 5-hmC-mediated Pparg regulation. The amount of triglycerides measured by Oil-Red-O staining in sextuplets upon Tet1 or Tet2 knockdown followed by the induction of adipogenesis (a). Tet1 and Tet2 knockdown efficiency based on Tet1 and Tet2 mRNA levels measured by qPCR in triplicates (b, c). Pparγ and Fabp4 expression levels, measured by qPCR in triplicates, in adipogenic induced 3T3-L1 cells after Tet1 and Tet2 knockdown (d, e). Global 5-hmC level during differentiation measured by GC–MS (f) and a locus-specific 5-mC and 5-hmC at the Pparγ locus quantified by MeDIP (g) and hMeDIP (h), respectively, in duplicates upon Tet1 or Tet2 knockdown followed by the induction of adipogenesis. The data were analyzed by Student's t-test, and *P<0.05 and **P<0.01 indicate significance between the knockdown versus the control; ‘ns’ presents ‘not significant’; d0, d3, d6, d9, d12 indicate days after adipogenesis induction.

Full size image

We next analyzed whether the effect of Tet1 and Tet2 knockdown during adipogenesis was via changes in the 5-hmC level rather than other indirect effects, both globally and at the Pparγ locus. Mass spectrometry analysis indicated that both Tet1 and Tet2 knockdown efficiently reduced the global 5-hmC level on day 6 (Figure 3f). More importantly, 5-hmC levels at the Pparγ locus (as measured by hMeDIP) markedly decreased upon the knockdown of Tet2 (P<0.05), and the reduction was inversely correlated with an increase in the 5-mC level at the locus (P<0.01), measured by MeDIP (Figures 3g and h). The knockdown of Tet1 did not significantly change the 5-hmC or 5-mC level at the locus.

Modulation of TET activity via metabolites influences adipogenesis

We next modulated TET activity with metabolites and monitored adipogenesis and Pparγ gene expression. Vitamin C enhances the activity of Fe (II) and α-KG-dependent dioxygenases, including the TET proteins, presumably by maintaining the ferrous state of iron.33 Fumarate, in contrast, leads to inhibition of the TET proteins by competing with α-KG, an essential TET co-factor.34 Here we investigated whether the treatment with vitamin C or fumarate altered adipogenesis by changing DNA methylation status during adipogenesis. 3T3-L1 cells were treated with 50 μg ml−1 ascorbic acid or 100 μM diethyl fumarate during adipogenesis for 6 days of induced adipogenesis. Lipid accumulation, a hallmark of adipocyte formation, in cells treated with vitamin C was significantly increased, while that with fumarate was significantly decreased when stained with Oil-Red-O staining (Figures 4a and e; P<0.05 and P<0.01, respectively). The mRNA level of Pparγ, measured by quantitative RT-PCR, significantly increased after vitamin C treatment (Figure 4b, P<0.05) and decreased after fumarate treatment (Figure 4f, P<0.05). The decreased expression of Pparγ subsequently resulted in the significant decline of Fabp4 expression in fumarate-treated cells (Figure 4f, P<0.05). We, furthermore, analyzed the global levels of 5-mC and 5-hmC to determine whether the treatment of those two metabolites was sufficient to alter TET protein activity. Upon vitamin C treatment, the level of 5-mC decreased and 5-hmC increased as measured by GC–MS (Figure 4c). In contrast, 3T3-Ll cells dramatically lost 5-hmC and gained 5-mC after fumarate treatment (Figure 4h). These results were confirmed by additional independent analysis using dotblots (data not shown). More importantly, we observed a significant decrease from of 5-mC at the Pparγ locus after vitamin C treatment (Figure 4d, P<0.01), and, in contrast, significantly increased after fumarate treatment (Figure 4h, P<0.01).

Figure 4
figure 4

Modulation of TETs via ascorbic acid and fumarate during adipogenesis. The amount of triglycerides measured by Oil-Red-O staining in sextuplets after adipogenesis with and without ascorbic acid, an activator of TETs, and fumarate, a repressor of TETs (a, e). Pparγ and Fabp4 expression levels, measured by qPCR in triplicates, in adipogenic induced 3T3-L1 cells with ascorbic acid (b) and with fumarate (f). The global 5-hmC level during differentiation measured by GC–MS (c, g) and a locus-specific 5-mC at the Pparγ locus quantified by MassARRAY in duplicates (d, h). The data were analyzed by Student's t-test, and *P<0.05 and **P<0.01 indicate significance between the treated samples versus the NT ones; d0, d3, d6, d9, d12 indicate days after adipogenesis induction. FT, fumarate-treated; NT, nontreated; VT, vitamin C-treated.

Full size image

Discussion

Adipogenesis is one of the most well-studied systems of cellular differentiation, and the understanding of how it is orchestrated is of major relevance to the development of human metabolic disease, including obesity and insulin resistance. During the pathological process, uncontrolled adipogenesis may be a major cause of altered uptake and storage of energy through an increase in both hyperplasia and hypertrophy of adipocytes. Previous studies have reported that several transcriptional factors including CEBPα and PPARγ have the primary role as the key drivers of adipogenesis.3, 4, 5, 6 In this study, we investigated whether the dynamics of DNA methylation drives differentiation of 3T3-L1 cells to adipocytes and how the DNA methylation status influence adipogenesis. We have demonstrated that expression of TETs are induced upon adipogenesis and are required for adipogenesis via locus-specific demethylation of the Pparγ locus, subsequently turning on its expression.

It is widely recognized that DNA methylation placement is essential to the chromatin during cellular differentiation.35 The acquisition and removal of DNA methylation at regulatory elements of the genome determine lineage fate by regulating the gene expression. A study using 5-aza-2'-deoxycytidine, an inhibitor of the DNA methyltransferase DNMT1, suggested that DNA methylation might be involved in adipogenesis in a differentiation stage-dependent manner.36 It is, however, known that the inhibitor is useful to reveal the active placement of DNA methylation, but not that of its removal. In recent years, oxidation of the methylated cytosines by the TET dioxygenases has been recognized as an intermediate step toward complete removal of methylated cytosines.37 Importantly, 5-hmC correlates with expression levels of pluripotency-related genes in undifferentiated ES cells.19, 20 A few studies on the relationship between 5-hmC and adipocyte differentiation have been reported.28, 38, 39 Genome-wide 5-hmC profiling identified high levels of 5-hmC in differentiated adipocytes, in particular at PPARγ-binding sites, for example, the Fabp4 locus.28 It has been speculated that PPARγ directed DNA demethylation around PPAR response elements via either poly(ADP-ribosyl)ation of PPARγ38 or by recruitment of CTCF.39 While these studies indicate that hydroxymethylation at PPARγ-targeted genes influenced the function of PPARγ protein during adipogenesis,37, 38 in the present study we instead investigated whether transcriptional activity of the Pparγ locus is directly regulated by DNA demethylation. First, we demonstrate the role of TETs in turning on the expression of Pparγ via hydroxymethylcytosine-mediated regulation of the Pparγ locus during induced adipogenesis. Increased expression of Tet1 and Tet2 in differentiated adipocytes resulted in high level of 5-hmC. In particular, loss of 5-mC and gain of 5-hmC were concerted at the Pparγ locus, but not at the Cebpα or Fabp4 loci.

The functional analysis using siRNA knockdown revealed that TET1 and TET2 might not target the same loci during adipogenesis. Knockdown of Tet1 and Tet2 led to a decreased 5-hmC at the global level (Figure 3f); Tet2 repression led to a stronger depletion of 5-hmC at the Pparγ locus than the Tet1 knockdown (Figure 3h). Both TET proteins carry an oxygenase domain, but only TET1 and not TET2 possess a CXXC domain that binds unmethylated CpG sequences,40 possibly differentiating targets in the genome. In mouse ES cells, TET1 and TET2 have differential roles in lineage determination during development, where Tet1 loss-of-function in ES cells results in developmental skewing toward endoderm/mesoderm lineage, whereas Tet2 loss-of-function maintains a tendency toward a neuroectoderm lineage.41 Moreover, the enrichment for 5-hmC by two TETs appears distinct genome wide.42 Likewise, Fujiki et al.38 observed that TET2 did not directly interact with PPARγ, unlike TET1. Taken together, TET1 and TET2 might have distinct and/or complementary roles in active demethylation at specific regions of the genome or during specific stages during adipogenesis.

Fewer studies were conducted regarding nutrients affecting DNA demethylation, whereas those regarding DNA methylation have been heavily investigated in the past decades. Vitamin C and citric acid cycle metabolites have been shown to have enhancing and inhibitory effect on TET proteins,34, 43 possibly modulating the availability of TET cofactors, Fe(II) and α-KG. Vitamin C treatment in ES cells resulted in global increase in 5-hmC in DNA by promoting the activity of TET enzymes.33 Fumarate also has been reported to induce the accumulation of α-KG by inhibiting α-KG dehydrogenase in citric acid cycle, which subsequently activates α-KG-dependent dioxygenase including TET enzymes.44 We applied vitamin C and fumarate as an activator and a repressor of TETs, and the results further confirmed that TET-mediated hydroxymethylcytosine was indeed a key factor to modulate adipogenesis.

In conclusion, our study has identified functional roles of TET1 and TET2 in transcriptional regulation of Pparγ-driving adipocyte differentiation. Further study is required to elucidate how TET proteins are directed to Pparγ, and the distinct roles TET1 and TET2 have in driving adipogenesis.