Preadipocytes from obese humans with type 2 diabetes are epigenetically reprogrammed at genes controlling adipose tissue function



Deterioration of the adipogenic potential of preadipocytes may contribute to adipose tissue dysfunction in obesity and type 2 diabetes (T2D). Here, we hypothesized that extracellular factors in obesity epigenetically reprogram adipogenesis potential and metabolic function of preadipocytes.


The transcriptomic profile of visceral adipose tissue preadipocytes collected from Lean, Obese and Obese with T2D was assessed throughout in vitro differentiation using RNA sequencing. Reduced Representation Bisulfite Sequencing was used to establish the genome-wide DNA methylation profile of human preadipocytes and 3T3-L1 preadipocytes treated by the inflammatory cytokine Tumour Necrosis Factor-α (TNF-α) or palmitate.


While preadipocytes from all obese subjects (Obese+Obese T2D), compared to those of Lean, were transcriptionally different in response to differentiation in culture, preadipocytes from Obese T2D showed impaired insulin signalling and a further transcriptomic shift towards altered adipocyte function. Cultures with a lower expression magnitude of adipogenic genes throughout differentiation (PLIN1, CIDEC, FABP4, ADIPOQ, LPL, PDK4, APOE, LIPE, FABP3, LEP, RBP4 and CD36) were associated with DNA methylation remodelling at genes controlling insulin sensitivity and adipocytokine signalling pathways. Prior incubation of 3T3-L1 preadipocytes with TNF-α or palmitate markedly altered insulin responsiveness and metabolic function in the differentiated adipocytes, and remodelled DNA methylation and gene expression at specific genes, notably related to PPAR signalling.


Our findings that preadipocytes retain the memory of the donor in culture and can be reprogrammed by extracellular factors support a mechanism by which adipocyte precursors are epigenetically reprogrammed in vivo. Epigenetic reprogramming of preadipocytes represents a mechanism by which metabolic function of visceral adipose tissue may be affected in the long term by past exposure to obesity- or T2D-specific factors.


During the course of obesity, adipose tissue undergoes hyperplasia and hypertrophy to allow net storage of energy excess into intracellular lipids. Preadipocytes contribute to adipose tissue growth by differentiating into metabolically mature adipocytes that store energy through insulin-mediated glucose uptake followed by biosynthesis of lipids. Extracellular factors like the inflammatory cytokine tumour necrosis factor (TNF)-α participate in the development of insulin resistance in human adipose tissue [1]. While mature adipocytes have been predominantly considered as primary targets for these extracellular mediators of insulin resistance [2], preadipocytes in the adipose tissue niche are also affected.

The hypothesis that extracellular factors can reprogram preadipocytes is supported by several studies. Preadipocytes exposed to high-glucose medium can be primed to become more inflammatory after undergoing adipogenesis [3]. Moreover, preadipocytes isolated from obese subjects have a decreased potential to take up lipids in response to an adipogenic cocktail [4,5,6]. Despite the fact that a role of genetic factors in the adipogenic memory of preadipocytes cannot be ruled out in this particular study, the observation that adipogenic capacity of human preadipocytes could be ameliorated by weight loss strongly suggests that epigenetic factors contribute to the decreased adipogenic potential of preadipocytes in obesity [7].

Adipogenesis is orchestrated by epigenetic factors that regulate chromatin conformation and accessibility of pro- and anti-adipogenic transcription factors to the DNA [8, 9]. We and other groups have previously established that the epigenome of adipose precursor cells is amenable to extracellular factors in vivo [3, 10,11,12,13]. A role of environmentally-driven epigenetic changes in cell differentiation has been suggested in pluripotent stem cells where developmental commitment and potential were influenced by the epigenetic state at key regulatory genes [14, 15].

Here, we sought to investigate the role of extracellular milieu on epigenetic reprogramming of preadipocytes and the effect on their adipogenic capacity. We hypothesized that preadipocytes from obese humans with or without type 2 diabetes (T2D) exhibit distinct epigenomic and transcriptomic responses to differentiation in culture.

Materials/subjects and methods

Study participants

The study was approved by the Ethics Committee from the Capital Region of Denmark (reference H-1-2011-077) and informed consent was obtained from all participants. This study included a total of 15 lean controls, 14 obese subjects with T2D according to ICPC-2-DK and 14 obese subjects with no history of diabetes. The participants were recruited from Surgical Gastrointestinal Department, Hvidovre Hospital, Denmark. The lean controls were subjects undergoing surgery for laparoscopic inguinal hernia repair. Individuals of both the Obese T2D and Obese groups were subjects to laparoscopic gastric bypass operation. The Obese groups were matched for weight. The cohorts were matched for age. Prior to surgery, all study participants were measured for height and weight. Exclusion criteria for all three groups were: alcohol consumption of more than 14 units/week, smoking, daily intake of medicine and presence of chronic/acute diseases. Lean men with diagnosed hypercholesterolaemia, hypertension and/or diabetes were excluded. Participants were fasted for at least 12 hrs and blood was drawn before undergoing anaesthetics. Blood was analysed at the Clinical Biochemistry Department, Hvidovre Hospital. Visceral adipose tissue was collected from the omental fat pat with laparoscopic surgery instruments under full narcosis during surgery.

Isolation and culture of human preadipocytes

Isolation of preadipocytes was performed as described [16]. Isolation and culture of human preadipocytes is further detailed in the ESM file.

Quantitative real-time polymerase chain reaction

Quantitative real-time polymerase chain reaction (qPCR) was performed using a protocol already described [17] and further detailed in the ESM file.

Transcriptomic analysis by RNA sequencing

RNA sequencing libraries were prepared using the Illumina TruSeq Stranded Total RNA with Ribo-Zero Gold protocol (Illumina) and performed as described [11]. Libraries were sequenced on a NextSeq500 instrument (Illumina) with 38-bp paired end. Reads were mapped to ENSEMBL [18] release 79 cDNA transcripts with transcript support level of 3 or less, using kallisto [19] with bias correction, stranded mapping and 100 bootstrap samples. Differentially expressed genes (DEG) were found using sleuth [20] aggregating transcripts to gene level. All models included a term to model individual variation. Genes that change throughout the differentiation were found by a likelihood ratio test comparing a model with only group information (Lean, Obese or Obese T2D) with a model with both group and differentiation state. Main effects of Obesity and T2D were detected by a Wald test in a model with differentiation state and group effects. All other comparisons were Wald tests using a model with differentiation state and groups nested within differentiation state.

Multiplexed reduced representation bisulfite sequencing

Cells undergoing reduced representation bisulfite sequencing (RRBS) were purified using magnetic activated cell sorting (MACS) to deplete the population for endothelial cells and leucocytes as previously described [21]. Quickly, cells were pelleted and resuspended in MACS buffer (Miltenyi Biotec) and incubated with FcR blocking reagent and mixed. Anti-CD31 (Miltenyi Biotec) and anti-CD45 (Miltenyi Biotec) antibodies were added to the tube for 15 min incubation at 4 °C. The suspension was washed once in MACS buffer and put through magnetic LD columns (Miltenyi Biotec). The flow-through was then collected and tested for purity using fluorescence-activated cell sorting.

Multiplexed RRBS was performed as described [22], with minor modifications. Briefly, genomic DNA was incubated with Msp1 restriction enzyme overnight for fragmentation. Adenylation was performed using dNTP (NEB) and Klenow fragment (NEB) followed by an AMpure Bead clean-up (Beckman Coulter) and ligation to TruSeq adapters (Illumina). Twelve ligated samples were pooled, and subjected to bisulfite conversion using EZ DNA Methylation Kit (Zymo Research). The library pool was amplified by PCR (2 min 95 °C (30 s 95 °C, 30 s 65 °C; 45 s 72 °C) × 20 cycles; 7 min 72 °C). PCR products were purified using the AMPure Bead clean-up. Libraries were quality-controlled by a Bioanalyzer instrument running the Agilent High Sensitivity DNA Kit. Sequencing libraries were sequenced on a HiSeq 2500 (Illumina) at the Danish National High-Throughput DNA Sequencing Centre, on 50 bp single-end sequencing mode. Fasta files were preprocessed with Trim Galore using the --rrbs flag. Alignment and CpG coverage statistics were computed using Bismark [23]. Differentially methylated regions (DMRs) were detected using BiSeq [24] using standard settings, except for the parameters min.sites, which was set to 5, and perc.samples which was set to 0.5. Precision was allowed to vary between conditions. DMRs were found with a FDR cut-off of 10%.

Lipid staining

Lipid accumulation was measured using Oil red O staining (Sigma-Aldrich) as previously described [4]. Cells were washed twice with PBS then fixed with 10% formaldehyde during 20 min at room temperature. Oil Red O (0.5%) in isopropanol was diluted 3/2 with distilled water (v/v), filtered and added to the fixed cells for 1 h at room temperature. Lipid staining was assessed under a light microscope. Stained cells were then rinsed with distilled water, eluted in 1 ml of ethanol and optical density for Oil Red O was measured at 540 nm on a Hidex Sense Microplate Reader.


Statistical analysis was performed using R, SigmaPlot or GraphPad Prism software. Data were tested for normality using Sigmaplot. Statistical difference between the groups for clinical parameters, Western blots, qPCR, cell count, BrdU, glucose uptake and Oil Red O staining, were analysed by one-way or two-way ANOVA setup with Turkey’s multiple-corrections test depending on experimental setup. A p-value <0.05 was considered statistically significant.


Preadipocytes from type 2 diabetic subjects display altered adipogenic potential

To compare the adipogenic potential of preadipocytes from lean humans to those from obese humans with various types of metabolic dysfunction, we isolated preadipocytes from visceral adipose tissue of lean subjects and obese subjects with or without T2D (Fig. 1a). Clinical characteristics of the study participants are presented in ESM Table 1. The study participants were matched for age at inclusion to avoid age-related epigenetic profiles, as previously reported [25]. As expected, all obese subjects, regardless of T2D, showed increased circulating HbA1c and C-peptide, with higher values in the T2D group. Conversely, total high-density lipoprotein was lower in the two obese groups, and further decreased in T2D. Expectedly, the clinical characteristics point at a more insulin-resistant phenotype in subjects with T2D compared to obese only group or compared to the lean group.

Fig. 1

Preadipocytes from Obese and Obese T2D show lower responsiveness to insulin and decreased expression of markers of adipogenesis. a Graphical illustration of the experimental setup. b, c Representative blot and quantification of the abundance of PPARG protein. d, e Expression of PPARG and FABP4 at confluence and at day 3 of differentiation. Quantification was made relative to the ribosomal RNA 18S. fi Insulin-induced phosphorylation of AKT and ERK as represented relative to total AKT and ERK, respectively. Representative Western blot image are shown. Data are shown as the mean + SEM. Time and group (Obese ± T2D) effects were evaluated by two-way ANOVA. * indicates p-value <0.05. ** indicates p-value <0.01

Morphological inspection of cultured preadipocytes by phase contrast microscopy at day 15 of differentiation showed marked inter-individual variation (ESM Fig. 1). While gene expression and protein abundance of the master regulator of adipogenesis peroxisome proliferator-activated receptor gamma (PPARG) were the same in Obese and Obese T2D compared to Lean (Fig. 1b–d), gene expression of fatty acid binding protein 4 (FABP4), another marker of adipocyte differentiation, was markedly lower in Obese T2D compared to the other groups (Fig. 1e). Although AKT phosphorylation was similar across groups (Fig. 1f, g), we detected a marked impairment in ERK-related insulin responsiveness in the obese groups, as showed by a failure of insulin to further increase phosphorylation of ERK compared to the basal condition (Fig. 1h, i).

Altered transcriptomic response to adipocyte differentiation in obesity

To study potential differences at the transcriptomic level, we performed RNA sequencing of the preadipocytes at the proliferative state, at confluence, and at 3 and 15 days of differentiation (ESM Table 2, Fig. 1a). Gene expression profiling of each individual sample by principal component analysis (PCA) and hierarchical clustering revealed a marked gene expression shift at day 3 and day 15, as well as between conditions at specific time points (ESM Fig. 2a, b). Across all subjects, the transcriptome was markedly remodelled throughout differentiation, with, respectively, 1311, 4348 and 4307 DEG at confluence, day 3 and day 15 of differentiation (Fig. 2a).

Fig. 2

Preadipocytes from Obese and Obese T2D have an altered transcriptome throughout differentiation. a Heatmap representing transcriptional clusters during differentiation. Groups (Lean, Obese and Obese T2D) and time points (at proliferation, confluence, day 3 or day 15) are indicated in the top panel with the colour codes showed in the legends. b KEGG pathway enrichment analysis at confluence, day 3 and day 15 for upregulated genes (red bars) and downregulated genes (blue bars). Numbers in parentheses after each pathway name indicate: total genes in pathway/genes annotated to pathway

To investigate enrichment for functional pathways, we attributed each DEG to Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways using the online web-based GEne SeT AnaLysis Toolkit (WEBGESTALT) (Fig. 2b). At confluence, we found pathways related to cell to cell interaction, cell cycle inhibition and switching of metabolism (Fig. 2b). At day 3, we found pathways related to the establishment of adipocyte metabolism being upregulated, whereas DEG with decreased expression were associated with pathways related to cytoskeletal organization and cell division (Fig. 2b). At day 15, we found similar pathways as compared to day 3 but with additional pathways involved in metabolism (Fig. 2b). Analysis of regulatory regions at proximity of DEG at day 3 returned binding sites for FOXO4, FREAC2, TATA, NFAT, LEF1, GFI1, CDC5, OCT1 and RP58 for upregulated genes, and AP1, E12, TEF1, SRF, E2F1DP1, CEPB and HFH4 for downregulated genes (ESM Table 3). At day 15, we found binding sites for FOXO2, SD1 and ERR1 for upregulated genes, and TEF1, LEF1, E12, AP1, SRF, E2F, E2F1DP1 and NFAT for downregulated genes. These analyses suggest that distinct transcription factors drive differentiation at day 3 compared to day 15.

Gene expression profiles were different in cells from the obese groups compared to the lean group (Fig. 3a, ESM Fig. 3a, b). Indeed, 172 genes were upregulated and 447 downregulated in both obese groups, with only 2 genes with opposite regulation between Obese and Obese T2D (Fig. 3a). Gene pathway analysis showed that genes downregulated and in common in the obese groups were enriched for the Ribosome pathway, with analysis of transcription factor binding sites at promoter regions enriched for MAZ, YY1 and SMAD. Interestingly, genes downregulated only in Obese T2D were enriched for pathways related to citrate cycle, fatty acid metabolism, PPAR signalling pathway, insulin signalling pathway and oxidative phosphorylation. Analysis of the transcription factor binding sites at promoter region of the downregulated genes returned CEBPB, SF1, HNF4, PPAR and OCT1, suggesting that a lower binding activity of these transcription factors could be involved in the altered transcription that we identified in T2D cells (Fig. 3b).

Fig. 3

Differential gene expression in Obese and Obese T2D compared to the Lean group. a Venn diagram showing overlapping genes that are differentially regulated in Obese vs. Lean and Obese T2D vs. Lean. b KEGG pathway analysis and regulatory binding factors at proximity of downregulated genes, specific for Obese T2D (upper panel) and common between Obese and Obese T2D (lower panel). c Heatmap for genes related to adipogenesis that are differentially regulated in Obese T2D vs. Lean (upper panel) compared to Obese vs. Lean. Gene ontologies are indicated at the top of the map. Numbers in parentheses after each pathway name indicate: total genes found in pathway/genes annotated to pathway

When we analysed differences between groups at each specific time point, we found 285, 230, 669 and 345 genes being differentially expressed at the proliferative state, confluence, day 3 and day 15, respectively, with a total of 886 unique genes for Obese subjects (without T2D). In cells from Obese T2D, we found 361, 848, 960 and 592 genes, respectively, with a total of 1399 unique genes (ESM Fig. 3a, b). KEGG pathway analysis of DEG at each time point returned more enriched terms by comparing Obese T2D to Lean than the Obese vs. Lean comparison (ESM Fig. 3c, d). Notably, we found an enrichment of pathways related to adipocyte function in Obese T2D, suggesting a more pronounced phenotype in obese humans with T2D (ESM Fig. 3d). Consistently with the KEGG pathway analysis, at day 15, cells from Obese T2D showed lower expression of key adipogenic genes such as PLIN1, LPIN3, ADIPOQ, FAS, AGPAT2, NR1H3, RORA, LIPE, SREBF1, STAT3 and STAT5A, as well as genes involved in metabolic processes like, amino acid metabolism, electron transport chain and fatty acid metabolism (Fig. 3c). Altogether, these results indicate that, while preadipocytes from all obese subjects have a distinct transcriptomic profile compared to the lean individuals, the specific transcriptomic response of preadipocytes from diabetic obese individuals to differentiation is further affected towards an impaired metabolism and oxidative phosphorylation in differentiated adipocytes.

DNA methylation is reprogrammed in preadipocytes from obese and T2D subjects

Next, we analysed genome-wide DNA methylation in preadipocytes at the proliferative state from the Lean and the Obese ± T2D groups at single-nucleotide resolution using RRBS, at GC-rich loci [22]. Global and region-specific methylation levels did not vary between groups and CpG location (ESM Fig. 4a–f). When comparing Lean and Obese, we found 132 DMRs (86 had decreased and 46 increased methylation compared to Lean) and 73 DMRs (37 had decreased and 36 increased methylation compared to Lean) between the Lean and the Obese T2D groups. When comparing Obese and Obese T2D we found 100 DMRs (53 had decreased and 47 had increased DNA methylation in Obese T2D, ESM Table 4, Fig. 4a–c). Thirteen genes with altered methylation were found in common in both obese groups compared to the Lean group. Most importantly, we found several DMRs at proximity of genes differentially expressed. In the Obese group, for example, NAT10, SHC1, PITX1, FRMPD4, THY1, ZBTB33, PA2G4 and DDX39B showed a positive correlation between expression and methylation, while SNRPE, FST, BCOR, KANSL1L and COL27A1 showed a negative correlation (ESM Table 5). In the Obese T2D group, RPL6 RNPS1, DDX39B and HNRNPD exhibited a positive correlation, and GTF3C3, PRSS12 and L1TD1, a negative correlation (ESM Table 5).

Fig. 4

Overview of the differentially methylated regions between the Obese, Obese T2D and Lean groups. Venn diagram showing overlapping regions that are differentially methylated in Obese vs. Lean, Obese T2D vs. Lean and Obese vs. Obese T2D (a) and a table of the direction of methylation change for the differentially methylated regions (b). Table of top 10 regions with most increased (dark colour) or decreased (light colour) methylation in Obese vs. Lean (Blue) and Obese T2D vs. Lean (Red) (c). Enriched KEGG pathways in d Obese and e Obese T2D

While we did not find any statistically significant enriched pathways for Obese vs. Lean, we noted a trend for pathways related to insulin signalling pathway (Fig. 4c). For Obese T2D vs. Lean, the cellular senescence pathway was enriched (Fig. 4d). Yet, these results point at a moderate association between transcriptional levels and DNA methylation.

DNA methylation signature of preadipocytes predicts adipogenesis potential

To investigate if the DNA methylation footprint of preadipocytes could predict adipogenic potential, we assigned an Adipogenic Score based on expression of the most highly expressed adipocyte-specific genes at day 15 of differentiation, in all of the three groups (PLIN1, CIDEC, FABP4, ADIPOQ, LPL, PDK4, APOE, LIPE, FABP3, LEP, RBP4 and CD36). A PCA based on Adipogenic Score revealed a large dispersion of subjects only at day 15 of differentiation, suggesting the Adipogenic Score is a good determinant for classifying terminal differentiation (Fig. 5a). We then analysed the RRBS dataset for DMRs associated with an alteration in Adipogenic Score. Independently of the group, we found that 3906 DMRs were associated to an increased Adipogenic Score, with 1901 DMRs with decreased methylation and 2005 regions with increased methylation (ESM Table 6, Fig. 5b, c). Notably, gene pathway analysis of the nearest genes to the DMRs showed enrichment for pathways related to insulin resistance, adipocytokine signalling, toll-like receptor signalling and signalling pathways regulating pluripotency of stem cells (Fig. 5d). Of interest, we found a significant enrichment between the nearest genes to the DMRs and DEG at day 15 (p = 7.6E-06). Since we did not find a correlation between DNA methylation and gene expression changes (ESM Fig. 5), this indicates that the directional effect of DNA methylation on gene expression is not uniform (ESM Table 7). When comparing Lean and Obese, we found 100 DMRs associated to an increased Adipogenic Score, and 305 DMRs between Lean and Obese T2D (Fig. 5b). Pathway analysis of genes located at close proximity of these DMRs only showed significant difference for the Obese T2D group, notably the Vitamin digestion and absorption pathway, which was also found significantly upregulated at day 15 of differentiation (Fig. 5e, f). Altogether, our results indicate that specific genes that are epigenetically reprogrammed in adipocytes from obese subjects are also differentially expressed, supporting epigenetic reprogramming of preadipocytes in obesity plays a functional role.

Fig. 5

Evidence for a link between DNA methylation and adipogenic potential. Principal component analysis for Adipogenic Score genes throughout differentiation (a). Groups (Lean, Obese and Obese T2D) and time points (At proliferation, confluence, day 3 or day 15) are indicated in the top panel with the colour codes showed in the legends. Differentially methylated genes for adipogenic score genes (SCORE), Obese vs. Lean and Obese T2D vs. Lean (b). Table of top 10 regions with most increased (dark colour) or decreased (light colour) methylation associated with an increase in adipogenic score (c). Enriched KEGG pathways for SCORE (d) Obese (e) and Obese T2D (f). p-Values and gene ratios (number of gene hits within the respective terms) are represented by colour gradient indicated in the figure

Extracellular milieu reprograms preadipocytes and alters adipogenesis

To identify factors in the extracellular milieu that reprogram preadipocytes in obesity and T2D, we tested the effect of short-term incubation of cultured 3T3-L1 preadipocytes with the inflammatory cytokine TNF-α or the free fatty acid palmitate, on later adipocyte differentiation, compared to cells grown in control medium (Fig. 6a). Preadipocytes in the proliferative state were incubated for 1 day with TNF-α or palmitate, then washed and grown in regular media. Cell proliferation, as measured by BrdU incorporation, was only transiently altered by palmitate, and cell number was similar across conditions 2 days after treatment removal (Fig. 6b, c). After differentiation, lipid content was reduced in TNF-α-, but not palmitate-treated cells (Fig. 6d). Both TNF-α- and palmitate-treated cells showed a reduced capacity to take up glucose in response to insulin (Fig. 6e). These results indicate that a short-term treatment with TNF-α or palmitate impairs the metabolic potential of preadipocytes.

Fig. 6

Short-term TNF-α and palmitate treatment in preadipocytes reprograms adipocyte function. a Experimental setup. b Growth curve for control, palmitate and TNF-α-treated cells. c BrdU incorporation 1 day after removing all treatments. d Quantification of the Oil red O staining at day 16 (day 8 of differentiation). e Insulin-induced glucose uptake at day 16. f DMRs in common between TNF-α- and palmitate-treated groups. g Number and example of genes differentially methylated in common between TNF-α and palmitate treatments. KEGG pathway enrichment analysis for TNF-α-treated (h) and palmitate-treated (i) cells. Data are shown as the mean + SEM. The time and group (Obese ± T2D) effects were evaluated by two-way ANOVA. * indicates p-value <0.05. ** indicates p-value <0.01

To determine if short-term incubation with TNF-α or palmitate reprogrammes the epigenome of preadipocytes, we performed RRBS on the treated preadipocytes at confluence. Treatment with TNF-α induced 70 DMRs whereas palmitate induced 62 DMRs (ESM Table 8, Fig. 6f). Interestingly, we observed an important overlap of DMRs between treatments, with a total of 36 regions: 21 hypomethylated, 10 hypermethylated and 4 with opposite regulation (Fig. 6g). Enrichment analysis of genes at close proximity to DMRs found gene pathways related to HIF-1 signalling, PPAR signalling and ribosome pathways enriched in both groups (Fig. 6h, i). Differential expression of Nr2f1, Rxra, Ctnnb1 and Terf1 showed negative correlation between DNA methylation and expression for Nr2f1 and positive correlation between Ctnnb1, Rxra and Terf1, suggesting that altered DNA methylation at specific regions is associated with a changed gene expression (ESM Fig. 6). These results support that extracellular factors can reprogram differentiation potential of preadipocytes and their metabolic function at the mature adipocyte level, through remodelling of DNA methylation at specific genomic regions.


In this study, we investigated the adipogenic potential of preadipocytes collected from obese humans with or without T2D. Analysis of the transcriptomic response to an adipocyte differentiation cocktail in culture allowed us to establish that preadipocytes from humans with obesity have distinct transcriptome and epigenome and an impaired metabolic potential. We identified a subset of candidate genes that undergo epigenetic reprogramming under TNF-α or palmitate exposure and which could participate in the altered adipogenic potential of preadipocytes from obese humans.

Our transcriptomic analysis provides resource information establishing the gene expression pattern throughout differentiation of human visceral fat. We found 12,651 transcripts varying over the course of adipogenesis. This magnitude is strikingly consistent with a previous study showing that expression of 11,830 transcripts was changed after differentiation of preadipocytes from subcutaneous adipose tissue [26]. Comparing the two datasets, we found that 16/20 gene pathways for increased gene expression, and 8/20 pathways for decreased expression at day 15 overlapped. These data suggest high similarities in gene expression programming during differentiation of visceral and subcutaneous adipocytes [27]. Our results support previous findings showing that depot-specific differences of adipose tissue is partly driven by preadipocyte differences [28, 29]. The subset of genes specifically activated during the differentiation of visceral adipocytes (e.g. genes related to the non-alcoholic fatty liver disease pathway) or the genes unique to subcutaneous adipocytes (e.g. related to the biosynthesis of unsaturated fatty acid pathway) likely represent gene networks setting the specific biology of each respective adipose tissue depot. Given the different contribution of each depot in the whole-body glucose and lipid metabolism, activation or inhibition of depot-specific genes could provide therapeutic entry points. When comparing the transcriptome of preadipocytes from the Obese and Obese T2D groups, we noted that numerous pathways (10 out of 18) that are upregulated at day 15 of differentiation were downregulated in Obese T2D. This is consistent with previous functional analysis of preadipocytes from T2D subjects differentiated in vitro and strongly indicates an altered adipogenic potential in preadipocytes from Obese T2D [4,5,6].

In the present study, we identified that insulin signalling is altered in preadipocytes from obese subjects. This is in line with previous studies showing distinct metabolic signatures [30], mitochondrial function and inflammation [31], adipogenesis [6, 32], tissue remodelling [33] and osteogenic capacity [34] in adipocytes from obese subjects with or without T2D. Taken collectively, these results strongly indicate a functional reprogramming of preadipocytes from obese subjects. Our transcriptomic results support that gene expression is reprogrammed during preadipocyte differentiation and provide insight into the specific altered pathways that could be responsible of the downstream impaired function of the differentiated adipocytes. Indeed, we report gene expression remodelling in pathways related to metabolic and mitochondrial function, inflammation, adipogenesis and osteogenesis, and tissue remodelling in obese individuals. It is interesting to note that some of these pathways were found by comparing adipose tissue from monozygotic twins with or without T2D [12]. These data support that preadipocytes are transcriptionally reprogrammed at specific gene pathways in obese humans.

In addition to the transcriptome, we found genome-wide alteration of the DNA methylation signature of preadipocytes from lean and obese subjects. Differential methylation in omental adipose tissue and adipocytes from obese humans before and after gastric bypass has been previously reported [10]. We did not find any overlap with our differentially methylated gene list. Nevertheless, in a comparison with another study where 5529 DMRs were reported in mature adipocytes from obese women [35], we identified an overlap of 61 genes and 39 genes, in our Obese and Obese T2D groups, respectively. These modest overlaps are likely to be caused by the nature of the cells investigated (either full tissue, purified adipocytes or, in our study, preadipocytes). To our knowledge, only few studies have investigated the DNA methylation pattern of preadipocytes in obese and low birthweight subjects [36, 37]. When methylation differences were reported, we however did not find a substantial overlap with our dataset. Given that we found differential methylation at gene pathways very specific to adipocyte function, the lack of overlap across studies is likely due to disparities in factors such as the DNA methylation assays, cell type composition and purity, sex, age and clinical characteristics of the respective cohorts, rather than caused by unspecific or noise signal.

To our knowledge, we are the first to show that the DNA methylation footprint of preadipocytes is highly related to their differentiation capacity. The adipogenic potential of mesenchymal stem cells from various depots is associated with different DNA methylation profiles in the same individual [38]. We have previously shown that TNF-α and palmitate can dynamically alter DNA methylation of skeletal muscle [39]. Here, we show that preadipocytes treated with TNF-α or palmitate retain an epigenetic 'memory' of the treatments, even after numerous cell divisions. These data support previous findings that prior cellular events can be epigenetically memorized, and that epigenetic reprogramming may play a role in the dysfunctional adipose tissue [3, 40, 41]. Resetting the epigenome of preadipocytes could be used as a therapeutic option to ameliorate adipose tissue biology in obesity.

In conclusion, we have identified that preadipocytes from obese humans are epigenetically reprogrammed. We established a link between DNA methylation changes and remodelling of gene expression during adipogenesis, which was further altered in T2D. Given the effect of TNF-α or palmitate on the epigenome of preadipocytes and the function of the differentiated adipocytes, we propose aetiological signals that are specific to obesity or T2D reprogram preadipocytes and alter adipocyte differentiation, thereby participating in the metabolic dysfunction of adipose tissue in obesity and T2D.


  1. 1.

    Hotamisligil GS, Arner P, Caro JF, Atkinson RL, Spiegelman BM. Increased adipose tissue expression of tumor necrosis factor-a in human obesity and insulin resistance. J Clin Invest. 1995;95:2409–15.

    CAS  Article  Google Scholar 

  2. 2.

    Hajer GR, Van Haeften TW, Visseren FLJ. Adipose tissue dysfunction in obesity, diabetes, and vascular diseases. Eur Hear J. 2008;29:2959–71.

    CAS  Article  Google Scholar 

  3. 3.

    Rønningen T, Shah A, Reiner AH, Collas P, Moskaug JØ. Epigenetic priming of inflammatory response genes by high glucose in adipose progenitor cells. Biochem Biophys Res Commun. 2015;467:979–86.

    Article  Google Scholar 

  4. 4.

    Isakson P, Hammarstedt A, Gustafson B, Smith U. Impaired preadipocyte differentiation in human abdominal obesity: role of Wnt, tumor necrosis factor-alpha, and inflammation. Diabetes. 2009;58:1550–7.

    CAS  Article  Google Scholar 

  5. 5.

    Permana PA, Nair S, Lee Y, Luczy-bachman G, De Courten BV, Tataranni PA, et al. Subcutaneous abdominal preadipocyte differentiation in vitro inversely correlates with central obesity. Am J Physiol Endocrinol Metab. 2004;286:958–62.

    Article  Google Scholar 

  6. 6.

    Lessard J, Laforest S, Pelletier M, Leboeuf M, Blackburn L, Tchernof A. Low abdominal subcutaneous preadipocyte adipogenesis is associated with visceral obesity, visceral adipocyte hypertrophy, and a dysmetabolic state. Adipocyte. 2014;3:197–205.

    CAS  Article  Google Scholar 

  7. 7.

    Rossmeislová L, Mališová L, Kračmerová J, Tencerová M, Kováčová Z, Koc M, et al. Weight loss improves the adipogenic capacity of human preadipocytes and modulates their secretory profile. Diabetes. 2013;62:1990–5.

    Article  Google Scholar 

  8. 8.

    Siersbæk MS, Loft A, Aagaard MM, Nielsen R, Schmidt SF, Petrovic N, et al. Genome-wide profiling of peroxisome proliferator-activated receptor γ in primary epididymal, inguinal, and brown adipocytes reveals depot-selective binding correlated with gene expression. Mol Cell Biol. 2012;32:3452–63.

    Article  Google Scholar 

  9. 9.

    Mikkelsen TS, Xu Z, Zhang X, Wang L, Gimble JM, Lander ES, et al. Comparative epigenomic analysis of murine and human adipogenesis. Cell. 2010;143:156–69.

    CAS  Article  Google Scholar 

  10. 10.

    Benton MC, Johnstone A, Eccles D, Harmon B, Hayes MT, Lea RA, et al. An analysis of DNA methylation in human adipose tissue reveals differential modification of obesity genes before and after gastric bypass and weight loss. Genome Biol. 2015;16:8:1–21.

    Google Scholar 

  11. 11.

    Nylander V, Ingerslev LR, Andersen E, Fabre O, Garde C, Rasmussen M, et al. Ionizing radiation potentiates high-fat diet–induced insulin resistance and reprograms skeletal muscle and adipose progenitor cells. Diabetes. 2016;65:3573–84.

    CAS  Article  Google Scholar 

  12. 12.

    Nilsson E, Jansson PA, Perfilyev A, Volkov P, Pedersen M, Svensson MK, et al. Altered DNA methylation and differential expression of genes influencing metabolism and inflammation in adipose tissue from subjects with type 2 diabetes. Diabetes. 2014;63:2962–276.

    Article  Google Scholar 

  13. 13.

    Rönn T, Volkov P, Davegårdh C, Dayeh T, Hall E, Olsson AH, et al. A six months exercise intervention influences the genome-wide DNA methylation pattern in human adipose tissue. PLoS Genet. 2013;9:e1003572.

    Article  Google Scholar 

  14. 14.

    Mikkelsen TS, Ku M, Jaffe DB, Issac B, Lieberman E, Giannoukos G, et al. Genome-wide maps of chromatin state in pluripotent and lineage-committed cells. Nature. 2007;448:553–60.

    CAS  Article  Google Scholar 

  15. 15.

    Reik W. Stability and flexibility of epigenetic gene regulation in mammalian development. Nature. 2007;447:425–32.

    CAS  Article  Google Scholar 

  16. 16.

    Arner P, Arner E, Hammarstedt A, Smith U. Genetic predisposition for type 2 diabetes, but not for overweight/obesity, is associated with a restricted adipogenesis. PLoS ONE. 2011;6:e18284.

    CAS  Article  Google Scholar 

  17. 17.

    Pattamaprapanont P, Garde C, Fabre O, Barrès R. Muscle contraction induces acute hydroxymethylation of the exercise-responsive gene Nr4a3. Front Endocrinol (Lausanne). 2016;7:1–9.

    Article  Google Scholar 

  18. 18.

    Yates A, Akanni W, Amode MR, Barrell D, Billis K, Carvalho-Silva D, et al. Ensembl 2016. Nucleic Acids Res. 2016;44:D710–6.

    CAS  Article  Google Scholar 

  19. 19.

    Bray NL, Pimentel H, Melsted P, Pachter L. Near-optimal probabilistic RNA-seq quantification. Nat Biotechnol. 2016;34:525–7.

    CAS  Article  Google Scholar 

  20. 20.

    Pimentel H, Bray NL, Puente S, Melsted P, Pachter L. Differential analysis of RNA-seq incorporating quantification uncertainty. Nat Methods. 2017;14:687–90.

    CAS  Article  Google Scholar 

  21. 21.

    Busser H, Najar M, Raicevic G, Pieters K, Velez Pombo R, Philippart P, et al. Isolation and characterization of human mesenchymal stromal cell subpopulations: comparison of bone marrow and adipose tissue. Stem Cells Dev. 2015;24:2142–57.

    CAS  Article  Google Scholar 

  22. 22.

    Boyle P, Clement K, Gu H, Smith ZD, Ziller M, Fostel JL, et al. Gel-free multiplexed reduced representation bisulfite sequencing for large-scale DNA methylation profiling. Genome Biol. 2012;13:R92.

    CAS  Article  Google Scholar 

  23. 23.

    Krueger F, Andrews SR. Bismark: a flexible aligner and methylation caller for Bisulfite-Seq applications. Bioinformatics. 2011;27:1571–2.

    CAS  Article  Google Scholar 

  24. 24.

    Hebestreit K, Dugas M, Klein H-U. Detection of significantly differentially methylated regions in targeted bisulfite sequencing data. Bioinformatics. 2013;29:1647–53.

    CAS  Article  Google Scholar 

  25. 25.

    Bjornsson HT, Sigurdsson MI, Fallin MD, Irizarry RA, Aspelund T, Cui H, et al. Intra-individual change over time in DNA methylation with familial clustering. JAMA. 2008;299:2877–83.

    CAS  Article  Google Scholar 

  26. 26.

    Broholm C, Olsson AH, Perfilyev A, Gillberg L, Hansen NS, Ali A, et al. Human adipogenesis is associated with genome-wide DNA methylation and gene-expression changes. Epigenomics. 2016;8:1601–17.

    CAS  Article  Google Scholar 

  27. 27.

    Van Harmelen V, Röhrig K, Hauner H. Comparison of proliferation and differentiation capacity of human adipocyte precursor cells from the omental and subcutaneous adipose tissue depot of obese subjects. Metabolism. 2004;53:632–7.

    Article  Google Scholar 

  28. 28.

    Pinnick KE, Nicholson G, Manolopoulos KN, McQuaid SE, Valet P, Frayn KN, et al. Distinct developmental profile of lower-body adipose tissue defines resistance against obesity-associated metabolic complications. Diabetes. 2014;63:3785–97.

    CAS  Article  Google Scholar 

  29. 29.

    Baglioni S, Cantini G, Poli G, Francalanci M, Squecco R, Franco A, et al. Functional differences in visceral and subcutaneous fat pads originate from differences in the adipose stem cell. PLoS ONE. 2012;7:e36569.

    CAS  Article  Google Scholar 

  30. 30.

    Böhm A, Halama A, Meile T, Zdichavsky M, Lehmann R, Weigert C, et al. Metabolic signatures of cultured human adipocytes from metabolically healthy versus unhealthy obese individuals. PLoS ONE. 2014;9:e93148.

    Article  Google Scholar 

  31. 31.

    Naukkarinen J, Heinonen S, Hakkarainen A, Lundbom J, Vuolteenaho K. Characterising metabolically healthy obesity in weight-discordant monozygotic twins. Diabetologia. 2014;57:167–76.

    CAS  Article  Google Scholar 

  32. 32.

    van Tienen FHJ, van der Kallen CJH, Lindsey PJ, Wanders RJ, van Greevenbroek MM, Smeets HJM. Preadipocytes of type 2 diabetes subjects display an intrinsic gene expression profile of decreased differentiation capacity. Int J Obes. 2011;35:1154–64.

    Article  Google Scholar 

  33. 33.

    Henninger AMJ, Eliasson B, Jenndahl LE, Hammarstedt A. Adipocyte hypertrophy, inflammation and fibrosis characterize subcutaneous adipose tissue of healthy, non-obese subjects predisposed to type 2 diabetes. PLoS ONE. 2014;9:1–8.

    Article  Google Scholar 

  34. 34.

    Oliva-olivera W, Gea AL, Lhamyani S, Coín-aragüez L, Torres JA, Bernal-lópez MR, et al. Differences in the osteogenic differentiation capacity of omental adipose-derived stem cells in syndrome. Endocrinology. 2015;156:4492–501.

    CAS  Article  Google Scholar 

  35. 35.

    Arner P, Sinha I, Thorell A, Rydén M, Dahlman-Wright K, Dahlman I. The epigenetic signature of subcutaneous fat cells is linked to altered expression of genes implicated in lipid metabolism in obese women. Clin Epigenet. 2015;7:1–13.

    Article  Google Scholar 

  36. 36.

    Broholm C, Olsson AH, Perfilyev A, Hansen NS, Schrölkamp M, Strasko KS, et al. Epigenetic programming of adipose-derived stem cells in low birthweight individuals. Diabetologia. 2016;59:2664–73.

    Article  Google Scholar 

  37. 37.

    Guénard F, Tchernof A, Deshaies Y, Pérusse L, Biron S, Lescelleur O, et al. Differential methylation in visceral adipose tissue of obese men discordant for metabolic disturbances. Physiol Genom. 2014;46:216–22.

    Article  Google Scholar 

  38. 38.

    Collas P. Programming differentiation potential in mesenchymal stem cells. Epigenetics. 2010;5:476–82.

    CAS  Article  Google Scholar 

  39. 39.

    Barrès R, Osler ME, Yan J, Rune A, Fritz T, Caidahl K, et al. Non-CpG Methylation of the PGC-1α promoter through DNMT3B controls mitochondrial density. Cell Metab. 2009;10:189–98.

    Article  Google Scholar 

  40. 40.

    Sharples AP, Polydorou I, Hughes DC, Owens DJ, Hughes TM, Stewart CE. Skeletal muscle cells possess a ‘memory’ of acute early life TNF-a exposure: role of epigenetic adaptation. Biogerontology. 2016;17:603–17.

    CAS  Article  Google Scholar 

  41. 41.

    Asterholm IW, Tao C, Morley TS, Wang QA, Delgado-Lopez F, Wang ZV, et al. Adipocyte inflammation is essential for healthy adipose tissue expansion and remodeling. Cell Metab. 2014;20:103–18.

    Article  Google Scholar 

Download references


OF was recipient of a research grant from the Danish Diabetes Academy supported by the Novo Nordisk Foundation. We would like to acknowledge The Danish National High-Throughput DNA Sequencing Centre, University of Copenhagen, for sequencing services. The Novo Nordisk Foundation Centre for Basic Metabolic Research is an independent research centre at the University of Copenhagen partially funded by an unrestricted donation from the Novo Nordisk Foundation.

Author contribution:

EA collected human samples, performed experiments, analysed the data and wrote the manuscript; LRI and AA performed bioinformatics analysis and generated figures; OF performed experiments, analysed the data and edited the manuscript; ID, SV, TB and VBK collected human samples, contributed to study design and edited the manuscript; DS provided expert advice and edited the manuscript; RB designed the study, analysed the data and wrote the manuscript. All authors approved the final version of the manuscript.

Author information



Corresponding author

Correspondence to Romain Barrès.

Ethics declarations

Conflict of interest:

The authors declare that they have no conflict of interest.

Electronic supplementary material

Rights and permissions

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Andersen, E., Ingerslev, L.R., Fabre, O. et al. Preadipocytes from obese humans with type 2 diabetes are epigenetically reprogrammed at genes controlling adipose tissue function. Int J Obes 43, 306–318 (2019).

Download citation

Further reading