Specific CpG hyper-methylation leads to Ankrd26 gene down-regulation in white adipose tissue of a mouse model of diet-induced obesity

Epigenetic modifications alter transcriptional activity and contribute to the effects of environment on the individual risk of obesity and Type 2 Diabetes (T2D). Here, we have estimated the in vivo effect of a fat-enriched diet (HFD) on the expression and the epigenetic regulation of the Ankyrin repeat domain 26 (Ankrd26) gene, which is associated with the onset of these disorders. In visceral adipose tissue (VAT), HFD exposure determined a specific hyper-methylation of Ankrd26 promoter at the −436 and −431 bp CpG sites (CpGs) and impaired its expression. Methylation of these 2 CpGs impaired binding of the histone acetyltransferase/transcriptional coactivator p300 to this same region, causing hypo-acetylation of histone H4 at the Ankrd26 promoter and loss of binding of RNA Pol II at the Ankrd26 Transcription Start Site (TSS). In addition, HFD increased binding of DNA methyl-transferases (DNMTs) 3a and 3b and methyl-CpG-binding domain protein 2 (MBD2) to the Ankrd26 promoter. More importantly, Ankrd26 down-regulation enhanced secretion of pro-inflammatory mediators by 3T3-L1 adipocytes as well as in human sera. Thus, in mice, the exposure to HFD induces epigenetic silencing of the Ankrd26 gene, which contributes to the adipose tissue inflammatory secretion profile induced by high-fat regimens.

Ankrd26 was recently identified as a gene involved in the regulation of the feeding behavior and in the development of both obesity and T2D in mice [13][14][15] . ANKRD26 maps at chromosome 10p12 in humans, a region associated with certain forms of hereditary obesity 16 . In mice, Ankrd26 gene is highly expressed in both the hypothalamus and WAT and its partial inactivation induces marked hyperphagia, severe obesity and diabetes in vivo 13,14 . In addition, in vitro evidence indicates Ankrd26 as a regulator of adipogenesis 17,18 . A methylome analysis of mouse epididymal WAT (eAT) 19 , the largest and easy accessible VAT depots in rodents 20,21 , has identified promoter hyper-methylation of Ankrd26 gene in HFD-fed compared to age-and sex-matched chow diet-fed mice, suggesting that Ankrd26 gene is amenable to nutritionally-induced epigenetic modifications. In this study, we aimed at establishing whether and how HFD modulates Ankrd26 gene expression in VAT in vivo through epigenetic processes.

HFD affects body weight, glucose homeostasis and insulin sensitivity in mice. HFD-fed mice
were heavier than standard chow diet (STD)-fed mice and reached a 50% increase of body weight compared with controls after 22 weeks of diet regimens (Table 1). These mice also exhibited increased fasting blood glucose levels, impaired glucose tolerance upon glucose loading and reduced insulin sensitivity after insulin injection compared with control mice (Supplementary Fig. S1).

HFD impairs Ankrd26 expression in mice.
To establish whether HFD modulates in vivo Ankrd26 expression, mRNA and protein levels were measured in eAT. Treatment with HFD for 22 weeks led to a significant decrease in both Ankrd26 mRNA (p < 0.001) and protein (p < 0.01) levels in obese mice compared with controls ( Fig. 1a and b). Similarly, HFD lowered Ankrd26 mRNA levels in mesenteric VAT ( Supplementary Fig. S2). HFD treatment for 4 additional weeks did not elicit any further decrease in Ankrd26 mRNA expression in the HFD-fed mice (34 week-old STD, Ankrd26 mRNA: 2.29 × 10 −3 ± 0.11 × 10 −3 AU; 34 week-old HFD, Ankrd26 mRNA: 1.36 × 10 −3 ± 0.19 × 10 −3 AU; p < 0.001). Differently from the long-term treatment, both Ankrd26 mRNA and protein levels showed no differences between HFD-and STD-fed mice upon 8 weeks of diet regimens (Supplementary Fig. S3a and b). Next, the Ankrd26 gene expression was assessed in vitro by exposing 3T3-L1 adipocytes to either palmitate or oleate, representing saturated and unsaturated fatty acid species, which are abundant in the HFD, or alternatively to leptin, whose levels raise through obesity development 22 . Quantitative real-time PCR (qPCR) analysis showed that palmitate, but not oleate or leptin, reduced Ankrd26 expression by about 25% (Supplementary Fig. S4a and b), suggesting that, at least in part, excess of saturated fats accounts for HFD-induced Ankrd26 gene repression.

HFD induces DNA methylation at the Ankrd26 promoter in mice.
To discover whether HFD induces DNA methylation changes at the Ankrd26 promoter and 5′ -untraslated region (5′ UTR), we performed Methylated DNA Immunoprecipitation (MeDIP) assay on pooled eAT genomic DNA from STD-and HFD-fed mice. This analysis revealed a 2-fold increase in DNA methylation at a segment of the promoter region (S1; − 462 bp/− 193 bp) in HFD-fed mice, while no DNA methylation enrichment was observed in a second segment (S2; − 158 bp/ + 140 bp; Fig. 1c). Consistently, palmitate but not oleate or leptin, enhanced S1 DNA methylation at the Ankrd26 promoter in 3T3-L1 adipocytes (Supplementary Fig. S4c and d), as showed by MeDIP assay. To further determine the specific HFD-induced DNA methylation profile occurring at 9 CpGs located at − 436 and − 221 bp from the Ankrd26 TSS, we adopted bisulfite sequencing analysis. High CpG methylation density was detected in obese mice compared with controls in 2 close cytosine residues at − 436 and − 431 bp from the Ankrd26 TSS (Fig. 1d). The combined percentage of methylation at these sites was inversely related to the amount of Ankrd26 mRNA (Fig. 1e). In parallel with mRNA expression, mice fed HFD or STD for 8 weeks showed no difference in the Ankrd26 DNA methylation state ( Supplementary Fig. S3c). In addition, in 16 week-old and 30 week-old STD-fed mice, no difference in both Ankrd26 mRNA levels (16 week-old STD, Ankrd26 mRNA: 2.10 × 10 −3 ± 0.20 × 10 −3 AU; 30 week-old STD, Ankrd26 mRNA: 1.96 × 10 −3 ± 0.23 × 10 −3 AU; p = 0.126) and DNA methylation (16 week-old STD, DNA methylation: 51.7 ± 2.9%; 30 week-old STD, DNA methylation: 51.3 ± 4.8%; p = 0.900) were observed. All together, these data indicate that the long-term exposure to calorie overload, rather than aging, affects eAT Ankrd26 expression and DNA methylation in mice.   containing the CpGs − 436, − 431 and − 391 bp. The un-methylated Ankrd26 promoter induced a 2.5-fold increase in the luciferase activity compared with the empty vector (Fig. 2a), indicating that this selected fragment is sufficient to mediate promoter activity. In addition, methylation of the Ankrd26 promoter caused a 35% decrease of luciferase activity compared with the un-methylated Ankrd26 promoter (Fig. 2a), indicating that methylation of this region has a negative impact on Ankrd26 gene expression. Next, to define whether the CpGs − 436 or − 431 bp or both is/are responsible for the regulation of the Ankrd26 promoter activity, luciferase assays were performed in NIH-3T3 cells transfected with un-methylated site-specific mutagenized vectors. The un-methylated Ankrd26 promoter mutagenized at the − 436 bp CpG site (Ankrd26− 436unme), similarly to the wild type (Wt) methylated Ankrd26 promoter, showed a 40% reduction of the luciferase activity compared with the un-methylated Ankrd26 Wt fragment (Fig. 2b). Similar data were obtained when the Ankrd26 promoter was mutagenized at the − 431 bp CpG site (Fig. 2b). Conversely, when the Ankrd26 promoter was mutagenized at the − 391 bp CpG site, the luciferase activity of the un-methylated Ankrd26− 391 promoter was comparable to the un-methylated Wt promoter (Fig. 2b). Thus, specific methylation at the − 436 and − 431 bp CpGs in the Ankrd26 promoter modulates Ankrd26 gene expression in vitro. DNA methylation often induces gene silencing by inhibiting transcriptional activator binding to promoters 23 . TFBIND analysis of the Ankrd26 promoter region spanning the CpGs − 436 and − 431 bp, predicted a consensus sequence (− 442 bp/− 429 bp) for the histone acetyltransferase/transcriptional coactivator p300 24,25 . The recruitment of p300 to this putative binding site and its relevance to the regulation of the Ankrd26 gene expression was therefore investigated. Chromatin Immunoprecipitation (ChIP) analysis showed a 40% decrease in p300 binding to the Ankrd26 promoter in eAT of obese compared with lean mice (Fig. 3a). In addition, Electrophoretic Mobility Shift Assay (EMSA) with a double-stranded labeled probe containing the p300 consensus sequence on the Ankrd26 promoter revealed that the addition of p300 antibody to the probe/Nuclear Extract (NE) mix super-shifted one of the complexes formed by interaction of the probe with the NE proteins (Fig. 3b). Also, the presence of an un-methylated competitor to the probe/NE mix effectively displaced p300 binding to the probe, while the probe/p300 complex was not affected by the addition of both a methylated or a mutagenized competitor (Fig. 3c). The in vitro over-expression of p300 in NIH-3T3 cells caused a 3-fold increase of the un-methylated Ankrd26 promoter activity ( Fig. 3d). At variance, when p300 was over-expressed, the luciferase activity of the methylated Ankrd26 promoter was 60% lower compared with the un-methylated Ankrd26 promoter (Fig. 3d). All together, these data indicate that p300 binding to Ankrd26 promoter regulates Ankrd26 gene expression and is dependent on the methylation state of the CpGs − 436 and − 431 bp.
HFD induces hyper-methylation of the Ankrd26 promoter through DNMT3a and DNMT3b in mice. The molecular events upstream and downstream methylation-induced displacement of p300 binding to the Ankrd26 promoter were subsequently analyzed. ChIP analysis revealed that binding of DNMT3a and DNMT3b, but not of DNMT1, to the Ankrd26 promoter was increased in HFD-fed mice compared to controls HFD affects histone acetylation, nucleosome positioning and RNA Pol II binding at the Ankrd26 promoter in mice. Further analysis of the Ankrd26 promoter using the NuPoP software predicted 2 nucleosomes (Nuc), Nuc-2 (− 288 bp/− 132 bp) and Nuc-1 (− 105 bp/ + 41 bp), positioned between the p300 consensus sequence and the Ankrd26 TSS. Micrococcal Nuclease (MNase) treatment of eAT chromatin from HFD-and STD-fed mice followed by qPCR revealed that HFD rendered the Ankrd26 promoter less sensitive to nuclease digestion, increasing Nuc-2 and Nuc-1 positioning (Fig. 4c). Consistently, ChIP analysis showed that HFD feeding significantly lowered histone H4 acetylation at both nucleosomes in the HFD-compared to STD-fed mice (p < 0.001; Fig. 4d). The RNA Pol II binding to the Ankrd26 TSS was also significantly lower in HFD-fed mice (p < 0.001; Fig. 4e). All together, these data indicate that methylation at the CpGs − 436 and − 431 bp and the subsequent p300 displacement from this region silenced Ankrd26 expression through nucleosome remodeling at the Ankrd26 promoter. Ankrd26 silencing promotes secretion of pro-inflammatory chemokines by cultured adipocytes. To assess the functional consequences of the HFD-induced epigenetic silencing of the Ankrd26 gene, Ankrd26 mRNA was reduced by about 35% by transfecting Ankrd26-specific siRNA in 3T3-L1 adipocytes (Fig. 5a). Silenced adipocytes showed enhanced secretion of the pro-inflammatory chemokines, Keratinocyte-derived Cytokine/Interleukine 8 (KC/IL-8), Eotaxin, Monocyte chemotactic protein 1 (MCP1) and Rantes (Table 2). These changes were accompanied by increased mRNA levels of Eotaxin and Mcp1 with no change in Kc/Il-8 and Rantes mRNAs (Fig. 5b,c,d and e). It appeared therefore that Ankrd26 physiologically controls the adipocyte pro-inflammatory secretion profile through effects occurring at different levels.
ANKRD26 expression negatively correlates with Body Mass Index (BMI) and inflammation markers in humans. mRNA expression of ANKRD26 in VAT was further examined in human obese subjects (Supplementary Table S1) in relation to BMI and inflammatory parameters. Interestingly, in obese subjects with normal glucose tolerance (NGT), ANKRD26 expression in VAT was found to negatively correlate with BMI ( Fig. 6a), with serum levels of the pro-inflammatory chemokines, IL-8 and RANTES and with serum levels of the inflammatory markers, IL-6 and C-reactive protein (CRP) (Fig. 6b,c,d and e). Altogether, these data indicate that, in obese humans, the reduction of ANKRD26 gene expression is associated with increased body weight and with a pro-inflammatory status.   Discussion Epigenetic modifications represent a common mechanism through which both genetic and environmental exposures impact on the susceptibility to obesity and T2D [26][27][28] . Recent evidence has underlined the potential importance of epigenetic regulation of gene expression and function in obesity 26 . Methylation changes at the promoter of several genes have been identified in both human and rodent obesity 29 . Additionally, the exposure to high calorie diets, which promotes DNMT expression and enzymatic activities, impacts on DNA methylation profiles both at specific genes and at genome-wide level 22,30 , suggesting that DNA methylation changes play a role in the responses to fat and high-calorie diets 31 .
Our study revealed HFD-induced methylation of the Ankrd26 promoter. We have no clue at the moment on the detailed mechanisms causing the increased methylation of Ankrd26 in HFD-fed mice, but our results indicate that the HFD-induced hyper-methylation at Ankrd26 promoter was concomitant with enhanced binding of de novo DNMT3a and DNMT3b to the same Ankrd26 promoter region. These changes were followed by down-regulation of Ankrd26 expression in the eAT. The epigenetic silencing of Ankrd26 gene in eAT appear to depend, at least in part, on saturated fats, abundant in the HFD. Indeed, we found increased promoter DNA methylation and down-regulation of Ankrd26 in 3T3-L1 adipocytes upon exposure to palmitate, a major component of the HFD, but not to oleate. At the variance with palmitate, cell exposure to leptin, whose serum concentration increases in relation to obesity 22 , showed no effect on Ankrd26 expression and methylation. These findings suggest that specific nutritional components of the HFD may contribute to the epigenetic silencing of Ankrd26 gene.
The HFD-induced changes in DNA methylation at Ankrd26 promoter and gene expression result from a long-term diet exposure. Indeed, cytosine hyper-methylation at the Ankrd26 promoter and gene silencing appeared in eAT from obese mice after a prolonged HFD feeding, while no evident difference was observed at earliest time-point. This time effect was associated with the eAT compensatory remodeling occurring in response to HFD. Indeed, eAT, along with other VAT depots, contributes to the inflammatory and metabolic complications in murine obesity 32 , and responds to HFD through different time-dependent changes 33,34 . Early upon HFD exposure (8-12 weeks), eAT expansion is accompanied by a major increase in adipocyte size. Upon more prolonged HFD exposure (20 weeks), however, eAT expansion is mainly sustained by increased adipogenesis and accompanied by enhanced secretion of inflammatory mediators, including Tumor necrosis factor alpha (TNF-α), IL-6, MCP1 and Rantes [33][34][35] . The mechanisms triggering this compensatory response in eAT have not been clarified yet but the present work now shows that they may involve HFD-induced Ankrd26 down-regulation. Along with its role in feeding behavior and body fat accumulation [13][14][15] , Ankrd26 has been identified as a regulator of adipogenesis in vitro 17,18 . Firstly, adipogenesis of 3T3-L1 cells is enhanced by selective silencing of the Ankrd26 gene with an Ankrd26-specific shRNA 18 . Secondly, Mouse Embryonic Fibroblasts from Ankrd26 mutant mice (MEFs Ankrd26 −/− ) have a higher rate of adipocyte differentiation. Indeed, the mRNA expression of the master regulator genes of differentiation process, CCAAT enhancer-binding protein α (C/ebpα), and Peroxisome proliferator-activated receptor γ (Pparγ), are up-regulated in MEFs Ankrd26 −/− , indicating that this gene is involved in regulating both the pre-adipocyte commitment and differentiation 17 .
In this work, we further demonstrated enhanced expression and/or secretion of the pro-inflammatory chemokines Eotaxin, MCP1, KC/IL-8, and Rantes by 3T3-L1 adipocytes whose Ankrd26 expression was silenced to levels similar to those occurring in response to HFD. These cytokines have been reported to contribute to adipose tissue inflammation 36,37 . Since secretion of Eotaxin, MCP1, KC/IL-8, and Rantes by the eAT also increases upon prolonged exposure to HFD 33,34 , our findings suggest the involvement of Ankrd26 down-regulation in raising and/or sustaining the low-grade inflammatory response which occursin the eAT after long-termHFD feeding and is implicated in the development of insulin resistance and T2D [33][34][35] . This might represent a mechanism by which environmental cues are integrated at specific genomic loci, contributing to the metabolic disorder. The relevance of these observations to humans is supported by our further findings in obese individuals with normal glucose tolerance, revealing that the reduction of ANKRD26 expression in VAT negatively correlates with the serum concentrations of inflammatory markers and pro-inflammatory chemokines, which are associated to obesity in humans 38,39 and whose increased levels predict occurrence of T2D [38][39][40][41][42][43] . Cardamone et al. have recently shown in the adipose tissue the relevance of the cytosolic function of the Ankrd26 partner GPS2 (G protein pathway suppressor 2) to the prevention of uncontrolled activation of inflammatory programs 44 . Even though this issue deserves further mechanistic investigation, we suggest that Ankrd26 might work as a molecular regulator of inflammatory signaling pathways, at least in part, by facilitating the cytoplasmic localization of its interacting partner GPS2 18 . Therefore, ANKRD26 down-regulation might represent an early event triggering chronic low-grade inflammatory response in human adipose tissue.
Detailed methylation analysis of the Ankrd26 promoter showed that HFD induced specific methylations at − 436 and − 431 bp CpGs, thereby exerting a suppressive effect on Ankrd26 promoter activity. Similar to DNA methylation, mutagenesis at the Ankrd26 promoter showed that introduction of C → T mutation at − 436 or − 431 bp CpGs significantly reduces Ankrd26 promoter activity. These results provide evidence that i., these cytosine residues have functional significance to the Ankrd26 gene expression; and ii., the DNA methylation at these specific CpGs plays a functional role in the epigenetic repression of Ankrd26 gene.
Current evidence supports a role for epigenetic changes in the regulation of metabolic diseases and in some cases, as in our study, it has been demonstrated that small methylation changes are associated with gene expression variability with significant effects on the phenotype [45][46][47][48] . In support of this concept, Barrès et al. 46 have recently shown that hyper-methylation of the Peroxisome proliferator-activated receptor γ coativator 1 α (PGC1α) promoter modulates PGC1α expression, implying a mechanism for decreased mitochondrial content in skeletal muscle from T2D patients. Also, using a gene reporter assay, these same authors have demonstrated that the in vitro methylation of a single cytosine residue at the PGC1α promoter is responsible for the reduction of gene activity 46 . CpG methylation generally affects transcription directly, by blocking the binding of transcriptional activators 24,49 , or indirectly, by recruiting DNA-binding proteins and co-repressor complexes that occupy the methylated promoters and facilitate the formation of heterochromatin 50 . In this study, we have further demonstrated that i., in vitro, the histone acetyltransferase/transcriptional coactivator p300 directly binds the consensus sequence at the Ankrd26 promoter, containing the methylation sensitive cytosines − 436 and − 431 bp; ii., hyper-methylation of these sites affects p300 binding and activity in vivo and in vitro. p300 regulates gene expression by acetylating both histones and transcriptional factors and plays a key role in modulating chromatin structure and function 25,26 . In this paper, we have also reported that HFD reduces histone H4 acetylation, increases nucleosome occupancy at the Ankrd26 promoter, and impairs RNA Pol II binding to the Ankrd26 TSS, suggesting that the HFD-dependent p300 displacement from the Ankrd26 promoter silences Ankrd26 gene. These findings are consistent with recent studies demonstrating that CpG methylation suppresses transcription of several genes by direct inhibition of p300 binding to their promoter sequences [51][52][53] . In conjunction with the inhibition of p300 binding, HFD induced the binding of MBD2 to the Ankrd26 promoter in mice. MBD2 is a methyl-CpG binding protein and causes gene silencing by recruiting histone deacetylase at the methylated promoter regions 31,50 . Accordingly, we propose that the specific CpG methylation at the Ankrd26 promoter leads to HFD-induced epigenetic gene silencing by triggering a cascade of events which involves DNA-associated regulatory proteins, such as p300 and MBD2, and changes in chromatin structure.
The potential relevance to humans of the findings reported in the present work is supported by our further evidence that VAT ANKRD26 mRNA levels were negatively correlated with BMI in humans. Consistent with our results, very recent computational data from a genome-wide DNA methylation analysis in human adipose tissue have revealed that ANKRD26 DNA methylation and mRNA expression correlate with BMI 53 . Thus, epigenetic regulation of ANKRD26 gene may occur in humans as well.
In conclusion, our work reveals that the methylation of specific CpGs at the Ankrd26 promoter occurs in mice during HFD treatment and causes the down-regulation of Ankrd26 expression, at least in part, by impairing p300 binding to its promoter. We propose that the epigenetic silencing of the Ankrd26 gene contributes to VAT inflammation following unhealthy dieting.  Supplementary Table S2. Body weight was recorded weekly throughout the study. The glucose tolerance test (GTT) and insulin tolerance test (ITT) were performed as described 14,16 . Blood glucose levels were measured using a glucometer (One Touch Lifescan, Milan, Italy). Mice were killed by cervical dislocation. eAT was collected from each mouse, snap frozen in liquid nitrogen and stored at − 80 °C.
Methylated DNA Immuoprecipitation (MeDIP). MeDIP assay was performed as described 19 . DNA methylation enrichment was evaluated on genomic DNA isolated from eAT of STD-and HFD-fed mice and from 3T3-L1 adipocytes. Sonicated pooled genomic DNA from eAT or cells was immuno-precipitated using anti-5meCpG (#ab10805, Abcam, Cambridge, MA) or mouse IgG (#I8765, Sigma-Aldrich) with anti-mouse IgG beads (Life Technologies, Carlsbad, CA). DNA methylation enrichment on recovered DNA was evaluated by qPCR. Samples were normalized to their respective input using the 2 −ΔCT method. Cloning strategy, site-direct mutagenesis and in vitro methylation. Ankrd26 promoter (− 733 bp/− 344 bp) was amplified by PCR. The purified PCR fragment was cloned into the firefly luciferase reporter pCpGfree-promoter-Lucia vector (Invivogen, Toulouse, France). The following site-specific mutated constructs were generated by PCR-based mutagenesis: pCpG-Ankrd26-436, pCpG-Ankrd26-431, pCpG-Ankrd26-391. The wilde type (Wt) pCpG-Ankrd26 vector, used as template, was removed from the PCR reaction by DpnI digestion (New England BioLabs, Ipswich, MA). Wt and mutated (mut) vectors were amplified into E. coli GT115 cells (Invivogen). Site-specific mutagenesis of each construct was validated by sequencing. In vitro methylation Scientific RepoRts | 7:43526 | DOI: 10.1038/srep43526 was performed using the M.SsI CpG methyltransferase following manufacturer's protocol (New England BioLabs). Un-methylated DNA was obtained in the absence of M.SsI. Methylation was confirmed by resistance to HpyCH4IV digestion (New England BioLabs).
Luciferase assay. NIH-3T3 cells were transfected with methylated or un-methylated Wt or mutagenized pCpG-Ankrd26 vector and Renilla control vector (Promega) by lipofectamine (Life Technologies), following manufacturer's instructions. Where indicated, cells were co-transfected with pCl.p300 expression vector (Promega). Firefly luciferase activity of each transfection was normalized for transfection efficiency against Renilla luciferase activity.
Electrophoretic mobility shift assay (EMSA). Protein-DNA complexes were detected using unlabeled or biotin end-labeled double-stranded DNA probes by annealing complementary oligonucleotides. Biotin 3′ -end oligonucleotides, spanning the Ankrd26 promoter sequence from − 455 bp to − 425 bp relative to the Ankrd26 TSS, were from Sigma-Aldrich and, where indicated, were synthesized to incorporate methylated cytosines ( me C). Binding reactions, consisting of biotin-labeled probe and NE, were performed using the LightShift kit (Thermo Fisher Scientific, Waltham, MA) following manufacturer's instructions. Biotin-labeled probe (20 fmol) was added and the reaction was allowed to incubate for 20 min at room temperature. In the competition experiments, the nuclear extracts were preincubated with 200 molar excess of unlabeled probes for 20 min on ice. In super-shift experiments, 2 μ g of p300 antibody (#SC-585, Santa Cruz Biotechnology) or 2 μ g of rabbit IgG (#I8140, Sigma-Aldrich) was preincubated with nuclear extracts for 60 min on ice. Protein-DNA complexes were separated on native polyacrylamide gel, transferred onto nylon membrane and detected by the LightShift Chemiluminescent EMSA kit (Thermo Fisher Scientific) following manufacturer's procedure.
Primer Sequences. The list of oligonucleotides used for PCR, qPCR, MeDIP, bisulfite sequencing, ChIP, MNase, EMSA can be found as Supplementary Table S3.
Fatty Acid/BSA complex solution preparation. Palmitate and oleate have been conjugated to fatty acid-free BSA (2:1 molar ratio Fatty Acid/BSA) as described in ref. 57. Briefly, a stock solution of palmitate (100 nM) was dissolved at 70 °C in 50% ethanol in a shaking water bath. In parallel, a fatty acid-free BSA solution was prepared at 55 °C in NaCl in a shaking water bath. Finally, the palmitate and the fatty acid-free BSA solutions were complexed at 55 °C in a shacking water bath, cooled to room temperature and sterile filtered. Oleate was complexed to the fatty acid-free BSA solutions following the same protocol. For fatty acid cell treatment, control adipocytes were treated with diluent only, corresponding concentrations of BSA and ethanol.
Cell culture and transfection. 3T3-L1 cells were grown and allowed to differentiate in mature adipocytes as described 19 . Mature adipocytes were i. silenced with 25 nmol/l of scrambled-siRNA or Ankrd26-siRNA for 48 h, or ii. treated with palmitate (0.250 mM; Sigma-Aldrich), or oleate (0.250 mM; Sigma-Aldrich) or corresponding vehicle (diluent solution with the same concentrations of BSA and ethanol of the Fatty Acid/BSA complex solution) for 96 h, or iii. treated with leptin (100 nM; R&D Systems, Minneapolis, CDN) or corresponding vehicle (20 mM Tris-HCl, pH 8.0) for 24 h. Adipokines were assayed in media from silenced cells by Bio-Plex Pro Mouse Cytokine Immunoassay following the manufacturer's protocol (Bio-Rad, Hercules, CA). Ankrd26 promoter methylation and gene expression were analyzed as previously described in this section.
Patient enrollment and tests. Abdominal VAT biopsies and serum samples were from patients undergoing bariatric surgery. Eleven normal glucose tolerance (NGT) obese subjects were selected. Population characteristics are in Supplementary Table S1. Participants with metabolic and endocrine disorders, inflammatory diseases, previous or current malignancies, and/or treated with drugs able to interfere with the epigenome were excluded. Secreted mediators were assayed in serum samples by Bioplex multiplex Human Cytokine, Chemokine and Growth factor kit (Bio-Rad) following manufacturer's protocol. ANKRD26 gene expression was analyzed in VAT as previously described in this section.
Ethics statement. This study adhered to the Declaration of Helsinki and has been reviewed and approved by the Ethics Committee of the "Federico II" University of Naples (Ethics Approval Number: No. 225_2013). Informed consent was obtained from all of enrolled individuals.
Scientific RepoRts | 7:43526 | DOI: 10.1038/srep43526 Statistical procedures. The area under the curve (AUC) was calculated using the trapezoidal rule. Data are expressed as mean ± SD. Comparison between groups were performed using Student's t-test or the one-way analysis of variance, as appropriate, using GraphPad Software (version 6.00 for Windows, La Jolla, CA). Correlation between two variables was calculated using the parametric Pearson r-test. p < 0.05 was considered statistically significant.