Introduction
With the continuous improvement in living standards, obesity has become a worldwide public health problem. Unhealthy lifestyles and eating habits have caused the number of obese people to rapidly increase worldwide. From 2011 to 2012, 8.1% of children aged 0 to 2 in the United States were overweight. The proportion of obese children among children aged 2–19 years reached 16.9%, and the proportion of obesity among adults aged 20 and above reached 34.9% [
1]. Obesity is caused by energy intake exceeding energy expenditure. At the cellular level, obesity is the result of an increase in the number or volume of adipocytes [
2].
All adipocytes, along with osteoblasts, muscle cells, and chondrocytes, are derived from mesenchymal stem cells. This process of differentiation is called adipogenesis [
3]. The formation of mature adipocytes includes two stages, commitment and terminal differentiation [
4]. In the commitment stage, mesenchymal stem cells are committed to differentiate to preadipocytes. Preadipocytes can differentiate into adipocytes, but preadipocytes will not spontaneously undergo terminal differentiation without exogenous adipogenesis-stimulating factors [
5]. In vitro, when adipogenesis-stimulating factors, including glucocorticoids, cAMP agonists, and insulin, are added, preadipocytes are induced to differentiate into mature adipocytes. In the terminal stage, preadipocytes undergo lipid accumulation and morphological changes of turning spheral. A series of signaling pathways, transcription factors and related proteins are subsequently activated in this process [
6].
MicroRNAs (miRNAs) can interact with transcription factors and important signaling molecules related to adipocyte differentiation to regulate adipogenesis [
7]. PPARγ and C/EBPs are the most important transcription factors throughout the process of adipocyte differentiation. miRNAs can directly or indirectly interact with these transcription factors to regulate cell differentiation [
8]. Kim et al. [
9] and Lee et al. [
10] found that miR-27a and miR-130a bind with the 3′UTR of PPARγ to downregulate the expression of PPARγ. MiR-27a and miR-130a are downregulated during the differentiation of 3T3-L1 cells, which upregulates the expression of PPARγ. Yang et al. [
11] found that the levels of miR-138 were significantly decreased during the adipogenic differentiation of primary adipose stem cells. When miR-138 is overexpressed, adipogenic genes, such as PPARγ, C/EBPα, and FABP4, are inhibited; therefore, lipid droplet aggregation is reduced.
It has also been reported that the DNA methylation of several key genes affects their expression levels during adipogenesis [
12]. Leptin is a hormone that regulates energy homeostasis. The leptin promoter is rich in CpG sites and is a chromosomal tissue-specific methyl region that can be dynamically methylated in humans and mice. Detecting the changes in DNA methylation at CpG sites in the promoter region of the leptin gene before and after differentiation of 3T3-L1 cells confirmed that the degree of DNA methylation was reduced, and DNA demethylation promoted leptin gene expression in 3T3-L1 cells [
13]. Yokomori et al. [
14] found that the GLUT4 promoter region exhibited similar changes in DNA methylation during 3T3-L1 differentiation. Horii et al. [
15] screened a small G protein Rho family guanine nucleotide exchange factor (Rho guanine nucleotide exchange factor 19, ARHGEF19) gene using a methylation-sensitive endonuclease PCR method and demonstrated that this gene plays an important role in regulating adipocyte differentiation. Studies on PPARγ, the key regulator of adipocyte differentiation, found that 5-aza-2′-deoxycytidine (AzaD), a DNA methylation inhibitor, interferes with the normal differentiation of 3T3-L1 preadipocytes and inhibits fat accumulation [
16].
As aforementioned, the differentiation of preadipocytes into adipocytes is controlled by a regulatory network, which is responsible for the regulation of commitment, lipid accumulation, adipocyte phenotype development, and maturation [
17]. The regulatory mechanism can target both stages involved at the same time.
It is worth noting that the expansion of adipose tissue through de novo adipogenesis can neutralize the detrimental metabolic effects secondary to obesity. In addition, the balance of hypertrophy or expansion of existing adipocytes and adipogenesis within the individual has a profound impact on metabolic health. Studies have shown that the increase in adipocyte size was associated with an increased risk of systemic insulin resistance [
18]. Other studies have shown that small adipocytes are particularly crucial for suppressing obesity and related metabolic disorders [
19] and it was revealed that small adipocytes are usually correlated with reduced susceptibility to diabetes [
20]. As their size expands, adipocytes will experience more mechanical stress as their contact pressure with neighboring structures increases, and when they expand to a size close to the oxygen diffusion limit, deficiency of oxygen occurs. The increased mechanical and hypoxic stress of hypertrophic adipocytes can cause inflammation of adipose tissue [
21]. Many experimental observations have shown that, compared with smaller adipocytes, larger hypertrophic adipocytes may exhibit different biochemical properties, for example, increased lipolysis, strengthened secretion of inflammatory cytokines, and reduced anti-inflammatory adipokines secretion [
22].
These findings raise an intriguing hypothesis that obesity itself may not be responsible for obesity-related metabolic disorders, but the deficiency of adipose tissue to expand further. This can lead to adipocytes hypertrophy instead of hyperplasia and a continuous increase in plasma glucose and lipid levels that can accumulate in other tissues and cause insulin resistance. Based on this theory, promoting de novo adipogenesis or inhibiting adipocyte hypertrophy can be a feasible therapeutic approach for insulin resistance and chronic inflammation secondary to obesity. Consequently, the intervention strategy for adipogenesis should be different between early-stage (promotion) and late-stage (inhibition).
This study systematically analyzed and summarized the epigenetic regulatory network of adipocyte differentiation between the early and late stages through expression profiling, noncoding RNA, and methylation sequencing data derived from public databases. This study aimed to construct a miRNA and DNA methylation regulatory network and to screen out pivotal genes that may provide a more comprehensive understanding of the epigenetic regulation of adipogenesis and potential therapeutic targets in the early and late stages of adipogenesis for the treatment of metabolic diseases.
Discussion
The formation of mature adipocytes includes two stages, commitment and terminal differentiation. Once differentiation is initiated, preadipocytes will first enter the contact inhibition phase. When the adipogenesis stimulating factor is added, the preadipocyte morphology will enter the second phase. With the accumulation of lipid droplets, they will transform into mature adipocytes [
25]. This article primarily examined the process of adipocyte differentiation in molecular mechanisms and epigenetic regulation during adipogenesis. The miRNA microarray, DNA methylation microarray, and mRNA microarray data were systematically analyzed, and analyses between samples in the early and late stages of adipogenic differentiation were compared. Hub genes and key pathways involved in epigenetic regulation during adipogenic differentiation were screened out.
In this study, day 4 (96 h) of induction was regarded as the cutoff, based on which the cell cohorts were divided into early-stage and late-stage groups. For miRNA datasets, there were no significant probes screened out between days 7 and 13. In addition, for DNA methylation datasets, when 48 h was selected as the cutoff, only 33 significantly methylated CpG islets were screened out. Combined with the aforementioned information, 96 h was selected as the optimal cutoff value.
Through the overlapping strategy of DEG and DEM target genes, 101 upregulated genes with low miRNAs were screened out. KEGG analysis revealed that the insulin signaling pathway was enriched. The insulin signaling pathway has been widely investigated in fat metabolism [
26]. After insulin binds to the extracellular alpha subunit of the insulin receptor (InsR), it activates tyrosine kinase activity of the intracellular beta subunit, and the beta subunit undergoes autophosphorylation so that the tyrosine of the downstream substrate is phosphorylated and activated [
27]. Insulin promotes the synthesis of fat and glycogen by acting on target tissues and inhibits the decomposition rate of lipids and liver glycogen, strongly promoting the storage of substances in the body. If insulin resistance or insulin production defects occur in the body, it will lead to abnormal regulation of adipogenesis [
28].
Sixty-four lowly expressed genes binding to miRNAs with high expression were filtered by determining the intersection between DEGs and DEM target genes. Using GO analysis, it was found that these genes are primarily involved in cell cycle regulation and chromosome regulation. KEGG analysis revealed that these genes were primarily responsible for the cell cycle and proliferation. As illustrated in Fig.
4C of the miRNA-mRNA network, it is worth noting that a total of 7 genes were bound by multiple miRNAs, which may be key targets for the regulation of multiple miRNAs. For instance, hsa-miR-455-3p can bind abundant target genes, such as APOBEC3F, BAIAP2, CDKN2A, CLCC1, PFKP, PLPP4, POFUT2, SYT7, and TMEM43. Hsa-miR-455 enhances adipogenic differentiation of 3T3-L1 cells by targeting uncoupling protein-1 [
29]. Hsa-miR-455 activates AMPKα1 by targeting HIF1, and AMPK promotes the brown adipogenic program and mitochondrial biogenesis. Concomitantly, miR-455 also targets the adipogenic suppressors Runx1t1 and Necdin, initiating adipogenic differentiation [
30]. KEGG analysis of Hsa-miR-455-3p binding genes revealed that these genes were related to cell cycle regulation and CDC42-related signaling pathways. Recent in vitro experiments confirmed that Cdc42 promotes the differentiation of adipose-derived mesenchymal stem cells (ADSCs) into pancreatic β-like cells through the Wnt/β-catenin pathway [
31].
The pathway enrichment of 663 genes with low methylation and high expression was primarily concentrated in Alzheimer's disease, prion disease, diabetic cardiomyopathy, Parkinson's disease, and reactive oxygen species-related chemical carcinogenesis. Diabetic cardiomyopathy is one of the complications of T2DM. These cellular changes include enhanced adipogenesis of MSCs, as observed in both type 1 and 2 models of diabetes. Emerging evidence now implicates enhanced marrow adipogenesis and changes to cellular makeup of the marrow in a novel mechanistic link between various secondary complications of diabetes [
32]. The MCODE plug-in was used to screen the core module. Analysis of the core module showed that its function was primarily related to fatty acid metabolism, peroxisomal protein import, the PPAR signaling pathway, the citric acid cycle, respiratory electron transport, glyoxylate metabolism and glycine degradation, and mitochondrial transition initiation.
For 237 genes with high methylation and low expression, GO and KEGG enrichment analysis indicated that they were primarily enriched in the extracellular matrix organization, focal adhesion, and cell-substrate junctions. These genes are primarily related to the movement of cells and formation of the extracellular matrix. The core modules of hypermethylated-low expressed genes were screened out from the PPI network. The functions of the core modules were focused on elastic fiber formation, collagen formation, and constitutive signaling by aberrant PI3K. It was reported that type III collagen (ColIII) is required for 3T3-L1 preadipocyte adipogenesis as well as the formation of actin stress fibers [
33]. The above analysis suggests that during the late stage of adipogenesis, cytoskeletal components are reduced, which may be related to the activation of PI3K-related pathways.
Interestingly, miRNA and DNA methylation may work together to regulate the expression of certain genes in adipogenesis. Twenty-four genes, including ACACA, ALDH2, AP1G1, and ARIH1, were increased due to two types of epigenetic regulation. Under dual regulation, 17 genes, such as APOBEC3F, ATP8B2, BAIAP2, and CALU, were downregulated. These genes were primarily involved in apoptosis, the JAK-STAT signaling pathway, the PI3K-Akt signaling pathway, another type of O-glycan biosynthesis, and adherens junctions. For 24 genes upregulated by low miRNA and hypomethylation, KEGG analysis determined that these genes may be involved in the insulin signaling pathway, pyruvate metabolism, nitrogen metabolism, fatty acid biosynthesis, and protein processing in the endoplasmic reticulum. ALDH2 and GLUL were upregulated in hematopoietic and lymphoid tissue and soft tissue. On the other hand, FKBP14, OSMR, and DFFB were downregulated in hematopoietic and lymphoid tissue and soft tissue, indicating that they may be involved in the adipogenesis of fat cells in these tissues.
From the CMap database, 10 chemical substances were identified, including dolasetron, dopamine, everolimus, ivermectin, and atenolol, which may have potential pharmacological actions on adipogenesis. Dopamine is a substance with a wide range of effects, and studies have shown that it has a potential effect on adipogenic differentiation. Dopamine receptor domain 5 (Drd5) genes were previously suggested to contribute to the adipogenesis. Knockdown of dopamine receptor D2 upregulates the expression of adipogenic genes in mouse primary mesencephalic neurons [
34].
The MCODE plug-in was used to screen out core modules, including 4 central genes, CANX, HNRNPA1, MCL1, and PPIF. The promoter region of CANX predicts four CpG islands, which may bind a large number of transcription factors, including AP2, MyoD, and VDR. The encoded protein calnexin is a calcium-binding, endoplasmic reticulum (ER)-related protein that transiently interacts with newly synthesized N-linked glycoproteins to promote protein folding and assembly. By keeping misfolded protein subunits in the ER for degradation, it may also play a central role in the quality control of protein folding [
35]. HNRNPA1 contains five CpGs in its promoter region. This gene encodes members of the ubiquitously expressed heterogeneous ribonucleoprotein (hnRNP) family. These ribonucleoproteins are RNA-binding proteins associated with pre-mRNA in the nucleus that affects pre-mRNA processing, as well as mRNA metabolism and transport. The protein encoded by this gene is one of the most abundant core proteins of the hnRNP complex and plays a key role in the regulation of alternative splicing. Mutations in this gene have been observed in individuals with amyotrophic lateral sclerosis [
36]. MCL1 encodes an anti-apoptotic protein that is a member of the Bcl-2 family. The longest gene product (isotype 1) enhances cell survival by inhibiting apoptosis, while the shorter gene products (isotypes 2 and 3) that are alternately spliced promote apoptosis [
37,
38]. The protein encoded by PPIF is a member of the peptidyl-prolyl cis–trans isomerase (PPIase) family. PPIases catalyze the cis–trans isomerization of proline amide peptide bonds in oligopeptides and accelerate protein folding. This protein is part of the mitochondrial permeability transition pore in the inner mitochondrial membrane. Activation of this pore is thought to be related to the induction of apoptosis and necrotic cell death [
39].
The role of ceRNA in adipogenesis has been a research hotspot in recent years. For example, lncRNA H19 targets LCoR by interacting with the miR-188 sponge, thereby affecting the osteogenic and adipogenic differentiation process of mouse BMSCs. LncRNA ADNCR inhibits adipogenesis and differentiation by targeting miR-204 [
40]. Therefore, potential ceRNA networks have been predicted for the above four central genes.
In the past, it was widely assumed that inhibiting adipogenesis could be a potential anti-obesity approach. However, results from various experiments indicate that adipogenesis inhibitors are not a good choice for improving metabolic disorders because restricting adipocyte expansion may lead to insulin resistance. As reported by Danfour et al., the failure of adipocyte differentiation may lead to type 2 diabetes [
17]. The mouse model showed that the improvement of the metabolic health of obese animals can be induced by further healthy expansion of adipose tissue [
41]. In these mice, adipogenesis allows adipose tissue to expand in the way of hyperplasia while preventing hypoxia, chronic inflammation, and fibrosis caused by hypertrophy of adipocytes [
42]. In the present study, through filtering the pivotal genes and characterizing different epigenetic regulation between the early and late stage of adipogenesis, more precise therapeutic targets could be provided to intervene in the early stage ( promoting hyperplasia) or late stage of adipogenesis (inhibiting hypertrophy) for the treatment of metabolic disorders secondary to obesity. Besides, the genes differentially expressed in the early and late stages of adipogenesis could also be the potential biomarkers to evaluate the risk of metabolic disorders for the overweight population.
There is some inherent limitation in our study. Due to data availability, the study did not include the correlation analysis involving clinical parameters such as incidence rate of metabolic disorders. Besides, more experiments are needed to further validate the effect of specific methylation or miRNA expression changes on adipogenesis. As a consequence, in the future, clinical trials and experiments are planned to validate the effectiveness of the epigenetic regulation of these genes and to explore the feasibility to use certain biomarkers to predict the risk of metabolic disorders among overweight populations.
In summary, this study comprehensively analyzed abnormally methylated, miRNA-targeted, and differentially expressed genes involved in the process of adipogenesis that is epigenetically regulated. Twenty-four genes were upregulated in response to miRNA reduction and hypomethylation, while 17 genes were downregulated in response to miRNA increase and hypermethylation. Ten chemicals were identified as putative therapeutics for adipogenesis. In addition, among these dual-regulated genes, CANX, HNRNPA1, MCL1, and PPIF may be key biomarkers in the epigenetic regulation of adipogenesis and may serve as aberrant methylation or miRNA targeting biomarkers.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.