A chromatin perspective of adipogenesis

Abstract

The transcriptional cascade governing adipogenesis has been thoroughly examined throughout the years. Transcription factors PPARγ and C/EBPα are universally recognized as the master regulators of adipocyte differentiation and together they direct the establishment of the gene expression pattern of mature adipose cells. However, this familiar landscape has been considerably broadened in recent years by the identification of novel factors that participate in the regulation of adipogenesis, either favoring or inhibiting it, through their effects on chromatin. Epigenetic signals and chromatin-modifying proteins contribute to adipogenesis and, through regulation of the phenotypic maintenance of the mature adipocytes, to the control of metabolism. In this review we intend to summarize the recently described epigenetic events that participate in adipogenesis and their connections with the main factors that constitute the classical transcriptional cascade.

Introduction

The concept of epigenetics currently refers to the study of cellular mechanisms that control somatically, and sometimes transgenerationally, inheritable gene expression states that are established in the absence of a change in the DNA sequence itself.1 Epigenetic mechanisms comprise but are not limited to an array of molecular modifications to both DNA and its associated proteins that control the “transcriptional state” of the cells.2 Consequently, epigenetic processes play a crucial role in the establishment and maintenance of cell identity by selectively activating or repressing transcription of subsets of tissue-specific genes.

Adipogenesis is one of the most studied and best characterized differentiation processes. The increased prevalence of obesity in present day society, and the current view of the adipose tissue as one of the most critical regulators of energy homeostasis, metabolism and immunity have warranted a sustained interest in the study of the mechanisms controlling its formation and regulation.3 The transcriptional cascade governing adipogenesis has been extensively studied for many years and is reviewed elsewhere, 4,5 but the underlying role of epigenetic processes in the development of fat cells has only begun to attract attention.

The molecular mechanisms that mediate epigenetic regulation are principally DNA methylation, chromatin remodeling, the post-translational modifications of the histones, the control of the higher-level organization of chromatin within the nucleus and the regulation by non-codings RNAs (reviewed in ref. 2). DNA methylation, which takes place at the carbon-5 position of cytosine in CpG dinucleotides, is usually associated with gene silencing and is the only known modification that targets DNA itself.6 But the most studied epigenetic mechanism to date is the covalent modification of the histone tails.7 Several transcriptional coregulators have the capacity to enzymatically modify histones by acetylation, methylation, phosphorylation, sumoylation or ubiquitylation8,9 and numerous reports have shown a clear link between the pattern of histone modifications in the chromatin of a given gene and its transcriptional state. Thus, histone lysine acetylation is usually related to gene activation10,11 whereas lysine methylation results in different outcomes, depending on the modified residue.9 Global analysis of histone lysine methylation by means of chromatin immunopecipitation assays coupled to microarray hybridization or direct sequencing have shown that methylation of lysine 4 of histone 3 (H3K4) usually correlates with gene activation12–14 whereas methylation of H3K9 or H3K27 associate with transcriptional silencing.15,16 Adding a layer of complexity, a number of histone-modifying enzymes also target nonhistone proteins, such as sequence-specific transcription factors or RNA polymerase II, thus broadening their range of influence upon gene activity.8

In this review we intend to summarize the role that different epigenetic signals and chromatin-modifying proteins play in adipogenesis by their interactions with the classical transcriptional cascade regulating the differentiation of adipose cells.

The Adipose Tissue

There are in mammals two types of adipose tissue (reviewed in refs. 17 and 18). The white adipose tissue (WAT) is constituted by highly specialized cells that store high quantities of lipids and respond to environmental cues by releasing a wide variety of bioactive molecules known as adipocytokines that participate in the regulation of insulin sensitivity, energy balance, vascular haemostasis or inflammation.19 White adipocytes are found in stereotypical depots throughout the body, displaying distinct molecular and physiological properties depending on their location.20,21 Brown adipose tissue (BAT), on the other hand, is predominantly localized in the interescapular region in both humans and rodents. The distinctive feature of brown adipocytes is their high mitochondrial content and expression of uncoupling protein 1 (UCP1), that make these cells opt for energy burning instead of storage since they can metabolize fatty acids and generate heat.22,23 Thus, brown adipocytes play a crucial role in the regulation of nonshivering thermogenesis, essential in newborn mammals, but until recently considered only of marginal significance in adults. However, several recent works have reported the existence of defined regions of functionally active BAT in adult humans.24,25 Interestingly, the amount of BAT inversely correlates with body-mass index, especially in older people, suggesting a role for brown fat in the protection against obesity.26 These findings have emphasized the necessity to better understand the subtle balance in the regulation of white and brown adipose tissue development and identify which are the environmental cues and molecular basis upon which each cell chooses its fate.

The adipose tissue, together with muscle and bone, is considered to emerge from the mesenchymal stem cells (MSCs) derived from the mesodermal layer of the embryo. However, lineage-tracking studies have shown that at least a subset of adipocytes originates from MSCs derived from the neuroectoderm.27 Comprehensive lineage-tracking studies are required to clarify this question, but a mixed origin of the different fat deposits would explain their distinctive molecular and biological features.18,21

An extracellular cue recruits MSCs to commit to a particular lineage (Fig. 1). This step results in the conversion of the pluripotent stem cells into preadipocytes, which can not be distinguished morphologically from their precursors but have lost the capacity to differentiate into other cell types. This is the first phase of adipocyte differentiation itself and is known as determination. In the second phase, known as terminal differentiation, the preadipocytes acquire all the features of mature adipocytes including the necessary machinery for lipid biosynthesis and transport, and the secretion of adipocyte-specific proteins such as leptin or adiponectin.

The murine 3T3-L1 or 3T3-F442A cell lines have been extensively used to study the differentiation of committed preadipocytes into adipocytes, because they are thought to recapitulate faithfully the main steps of the in vivo differentiation process.28,29 To study the earlier steps of the process, that is, the commitment of pluripotent precursors into preadipocytes, other cell types can be used such as totipotent embryonic stem cells (ESCs) that can give rise to any cell type, or multipotent MSCs that can develop into fat, muscle, cartilage or bone.30,31 By using these tools, the main transcription factors involved in the differentiation of adipocytes have been and are still being identified. However, in recent years this well-known scenario has been considerably widened by the identification of novel factors that participate in the regulation of adipogenesis, either favoring or inhibiting it, through their effects on chromatin (Table 1).

Commitment of Pluripotent Precursors into Preadipocytes

Chromatin dynamics in pluripotent cells.

The chromatin of pluripotent cells possesses a number of features of its own. Pluripotent cells display highly dynamic and heterogeneous chromatin, with a high level of decondensation and diffuse binding of heterochromatin protein HP1.32,33 Interestingly, these features were observed in the pluripotent MSC line 10T1/2, which can give rise to preadipocytes, myoblasts and chondroblast, but not in the still undifferentiated but already committed C2C12 myoblasts, 32 probably reflecting the genetic flexibility of pluripotent cells and how it gets restricted as differentiation proceeds and the number of cell fate choices diminishes. In order to maintain chromatin plasticity, pluripotent cells express a wide range of chromatin regulators, and their loss usually results in aberrant differentiation. Thus, ESCs lacking the BAF250a subunit of the SWI/SNF chromatin remodeling complex are unable to differentiate into adipocytes or cardiomyocytes, but are still capable of differentiating into neurons.34 These data suggest that the role of particular chromatin regulators in differentiation is lineagedependent, probably reflecting the subset of promoters that they regulate.

In accordance with this highly dynamic state of their chromatin, pluripotent cells are enriched in epigenetic signals associated with gene activity, such as histone acetylation or H3K4 methylation, that can be found even in silenced genes.35 Most strikingly, the histones of developmentally important genes display a bivalent epigenetic mark in ESCs.36 The promoters of genes that regulate cell fate decisions, including several markers of adipogenesis37,38 present a large region of the silencing mark H3K27me3 and smaller regions of the activation-related mark H3K4me3.35–37 This bivalent signal ensures that developmentally important genes remain silent but poised for transcription. H3K27 methylation is mediated by the Polycomb group (PcG) proteins, a family of transcriptional repressors that play a crucial role in the maintenance of pluripotency by repressing differentiation-related genes.37,39 In the absence of the H3K27 methyltransferase Eed or the H3K4 methylation regulator PTIP, resulting in alteration of these histone signals, developmental genes are expressed aberrantly and spontaneous differentiation occurs.35,40 On the other hand, induction of adipogenic differentiation results in H3K27 demethylation specifically occurring in adipogenic promoters, thus establishing the definitive activated state of this subset of genes by erasing this silencing mark.38

Developmentally important genes are also marked by DNA methylation in their promoter regions, a complementary mechanism to ensure gene repression.39,41 In fact, most of the silenced genes presenting the H3K27me3/H3K4me3 bivalent histone mark display either PcG binding or DNA methylation as an additional silencing mechanism.39,42 Once differentiation is induced, lineage-specific genes are DNA demethylated whereas pluripotency genes are methylated correlating with their silencing.42 Several lines of evidence support the fact that DNA demethylation plays a significant role in the cell fate decision of MSCs, at least in part by inducing expression of genes responsible for lineage commitment. Thus, treatment of the pluripotent cell line 10T1/2 with the methyltransferase inhibitor 5-azacytidine results in random DNA demethylation and cell differentiation. Interestingly, adipogenesis is associated in this model with DNA demethylation of the bmp4 gene.43 The secreted factor BMP4 favors commitment of MSCs to the adipogenic lineage, and its increased expression correlates with the capacity of these cells to undergo adipogenesis.44 Similarly, under 5-azacytidine treatment, commitment of 10T1/2 cells to the myogenic lineage correlates with DNA demethylation of the promoter region of the myogenic transcription factor myod gene.45 In accordance with all these data, the promoters of adipogenic genes such as pparg2 or leptin are unmethylated in isolated adipose tissue stromal cells enriched in preadipocytes, whereas myogenic and endothelial promoters are methylated,46 indicating that these cells have already committed to an adipogenic fate and additionally supporting a role for DNA methylation on lineage selection and cell differentiation. Moreover, increased levels of pparg2 promoter methylation have been described in visceral adipose tissue of several mouse models of diabetes47 indicating that the gene needs to remain unmethylated in order to maintain the mature characteristics of the cells. Surprisingly, DNA demethylation in the exon 1 of the Rho guanine nucleotide exchange factor WEGF takes place during adipogenesis concurrently with its progressive silencing,48 demonstrating that the relationship between DNA methylation and transcription may not be as straightforward as usually assumed.

Higher-order chromatin structure also regulates gene expression and thus cell differentiation. During differentiation of porcine MSCs into adipocytes, adipogenic genes such as those coding for transcription factors PPARγ (peroxisome proliferator activated receptor-γ) and C/EBPα(CCAAT/enhancer binding protein-α) are repositioned from the nuclear periphery to the nuclear interior coinciding with their increased transcription.49 Interestingly, we have previously observed a similar change in the nuclear pattern of H3K4me2 during adipogenesis.50 In 3T3-L1 preadipocytes, H3K4me2, which was observed to be enriched in the promoters of still silent adipogenic genes such as the hormone adiponectin, was found strongly enriched along the nuclear border in inmature cells and dramatically rearranged towards the interior nucleoplasma in differentiated cells.50 The region along the nuclear envelope is occupied by mid-to-late replicating chromatin in 3T3-L1 cells, and mostly includes silent genes.51 Taking together all these results it is tempting to speculate that histone modifications and nuclear location collaborate in regulating transcription of adipogenic genes during adipogenesis. These data are in accordance with the idea that chromatin marks regulate and/or reflect the higher-order chromatin structure.52,53

Factors regulating commitment to the adipose lineage.

Several factors involved in cell commitment and differentiation have been identified and characterized (Fig. 1). Regulators that induce commitment into one lineage often inhibit the differentiation along the other, for example the nuclear receptor PPARγ is a major inducer of adipogenesis whose activation inhibits osteoblastogenesis. Inducers of osteoblastogenesis such as the canonical Wnt/β-catenin pathway suppress the trans-activation and expression of PPARγ in MSCs,54 by activating the histone methyltransferase (HMT) SETDB1.55,56 SETDB1 together with NLK (Nemo-like kinase) and CHD7 (chromodomain helicase DNA binding protein-7) constitutes a corepressor complex that represses PPARγ trans-activation through H3K9 methylation of its target promoters. At the same time, Wnt-5a activates the transcription factor Runx2, a known inductor of ostoeblastogenesis. 55,56 Similarly, in vitro studies have shown that attenuation of PPARγ activity by overexpression of the histone deacetylase (HDAC) Sirt1 in MSCs also blocks adipogenesis while favoring osteoblastogenesis.57,58 Sirt1 binds to the promoter of PPARγ target genes together with corepressors NCoR/SMRT,58 resulting in gene silencing.

On the other hand, a family of tension-responsive factors, referred to as tension-induced/inhibited proteins (TIPs), is involved in the determination of the adipogenic versus myogenic lineage from MSCs.59 The TIP family is composed by at least eight members, all of whom possess a SANT (switching-defective protein 3, adaptor 2, nuclear receptor corepresor and transcription factor IIIB) domain,60 characteristic of proteins with chromatin-modifying activities. Therefore, these proteins possibly exert their effects through chromatin remodeling of their target promoters. TIP-1 is induced by stretch and promotes smooth muscle differentiation by binding to the promoter of myogenic transcription factor Serum Response Factor (srf ). TIP-3, on the contrary, is suppressed by stretch and promotes adipogenesis in nonstretched MSCs by binding to the pparg gene promoter.59 A recently identified member of the family, TIP-6 was shown to bind to the pparg promoter in NIH-3T3 fibroblasts and induce adipogenesis by recruiting the HAT p300 through interaction with the SANT domain, resulting in increased histone acetylation and expression of pparg.60

Origins of brown adipose tissue.

The existence of a common adipose precursor cell for brown and white fat cells was supported by the ability of white adipocytes to transdifferentiate into BAT cells under certain conditions.61–64 Several in vitro studies support the differentiation of white preadipocytes into brown adipocytes and aim to identify the factors implicated in this switch in cell fate. Thus, functional inactivation of the retinoblastoma protein (pRb) in mouse embryo fibroblasts (MEFs) and in white preadipocytes results in increased differentiation of adipose cells with a gene expression pattern consistent with that of brown adipocytes.65 On the other hand, receptor-interacting protein 140 (rip140) knockout mice exhibit a lean phenotype due to an increase in BAT-specific genes.66 RIP140 nucleates a repressive complex containing histone deacetylases HDAC1 and HDAC3, the H3K9 methyltransferase G9 and DNA methyltransferases Dnmt1, Dnmt3a/3b, that is recruited to the ucp1 promoter during white adipocyte differentiation to mediate ucp1 gene repression by DNA methylation and histone hypoacetylation, maintaining a white phenotype.67,68

Interestingly, a series of recent works have demonstrated that brown and white preadipocytes arise from different precursor cells, with brown preadipocytes being strikingly related to myoblasts. 69–71 Seale et al.70 have shown by means of lineage tracking studies that skeletal muscle progenitor cells can give rise to either muscle or brown fat cells, but not white fat cells. The decision of this precursor cell to differentiate into muscle or brown adipocytes is controlled by the expression of PRMD16, a 140 kD PR (PRD1-BF-1RIZ1 homologous)-domain containing protein70,72 that activates a number of brown adipocyte genes, such as UCP1 or the cofactor PGC1α, when expressed in cultured white preadipocytes or in white fat depots in vivo. Moreover, PRMD16 selectively represses white fat-specific genes such as the hormone resistin by interacting with the corepressors C-terminal-binding protein-1 and 2 (CtBP1, CtBP2). Recruitment of PGC1γ to the PRMD16 complex displaces CtBP allowing PRMD16 to activate brown adipocyte-specific genes.73 Recently it has been shown that adipogenic transcription factor C/EBPγ is the critical partner in the PRDM16 transcriptional complex,74 although the presence of histone modifying enzymes or chromatin-associated proteins has not been described yet.

However, an alternative origin of brown fat cells in vivo is also documented, as lineage tracking studies show that brown adipocytes in mice adapted to cold have never expressed the myogenic marker myf5 contrary to what happens with adipocytes in the conventional BAT deposits,70 thus giving support to the traditional idea that white adipocytes may differentiate into brown fat cells.

Differentiation of Preadipocytes

Chromatin remodeling in adipogenesis.

Taking the 3T3-L1 cell line as a model of white adipocyte differentiation, the transcriptional cascade consists in a reduced number of transcription factors acting in time-controlled fashion. The increased expression and protein accumulation of C/EBPγ and C/EBPδ in the first hours of differentiation stimulate expression of C/EBPα and PPARγ, which in turn maintain the expression of each other.4,5 C/EBPα and PPARγ are considered the master regulators of adipogenesis and directly control transcription of most adipogenic genes.75,76 However, PPARγ is the only known factor that is both necessary and sufficient to induce adipogenesis77–79 and there are no other factors described to date that promote adipogenesis in its absence. Consistent with that, most pro-adipogenic factors and cofactors seem to act precisely through the induction of PPARγ expression or activity, whereas most inhibitors of adipogenesis may function in part by blocking PPARγ.

In preadipocytes, C/EBPγ and C/EBPδ are already bound to the cepba and pparg promoters,80 while PPARγ is bound to the promoters of its target genes in association with a repressive complex composed of pRb and HDAC3,81 (Fig. 2). When differentiation is induced, recruitment by C/EBPγ of the chromatin remodeling complex SWI/SNF promotes transcription of the pparg gene.82 In the case of the cebpa promoter, C/EBPγ is initially blocked by a repressive complex constituted by mSin3A/HDAC1, but once PPARγ protein accumulates, it is able to target HDAC1 to the 26S proteasome, causing its degradation and allowing activation of cebpa expression.83 Glucocorticoids have also been shown to exert the same effect,84 in this case mediated by the HAT GCN5, which acetylates C/EBPβ within a cluster of lysine residues, resulting in dissociation of the interaction between C/EBPβ and HDAC1. Once C/EBPα accumulates in the nucleus of differentiating cells, it displaces C/EBPβ from the promoters of late adipogenic genes, such as leptin, adiponectin or pparg itself.75,76,80,82 Like C/EBPβ, C/EBPα promotes transcription of its target genes by recruiting the SWI/SNF complex.85 At the same time, C/EBPα promotes cell cycle exit by repressing E2F-dependent promoters also through its interaction with the SWI/SNF complex.86,87 In agreement with its pleiotropic role in adipogenesis, knockdown of the snf5/Ini1 subunits of the SWI/SNF complex blocks adipogenic differentiation of 3T3-L1 and MSCs.88

Histone modifications in adipogenesis.

Given the preponderance of HAT/HDAC interactions with adipogenic regulators, it is easy to recognize the key role that histone acetylation plays in adipogenesis. Histone acetylation increases at the promoters of adipogenic marker genes during adipocyte differentiation, correlating with their increased expression,50 whereas a decrease in HDAC expression takes place throughout this process.89 Furthermore, blocking HDAC activity with general inhibitors such as trichostatin A increases adipogenesis, whereas overexpression of HDAC1 impairs it.81,84,89 Unexpectedly, a negative effect of HDAC inhibitors on adipogenesis has also been described by some works.90,91 Moreover, treatment of mature adipocytes with the HDAC inhibitor apicidin but not other inhibitors such as sodium butyrate, induced dedifferentiation of mature adipocytes. 90 This apparent contradiction may be due to the different specificities of the HDAC inhibitors used, as particular inhibitors present different affinity for the different families of HDACs that do exist. Also, these data may suggest that HDACs may play a key role in silencing anti-adipogenic genes during adipogenesis and in mature cells. In this sense, transcription factor KLF6 promotes adipogenesis by silencing the preadipocyte marker gene dlk1 through HDAC3 recruitment to the dlk1 promoter.92 Accordingly, we have observed that the promoter region of dlk1 is acetylated in 3T3-L1 fibroblasts, when the gene is highly expressed, becoming progressively deacetylated throughout differentiation as the gene is gradually silenced.50

On the initial phases of adipogenesis, PPARγ is blocked by its association with a repressive pRb-HDAC3 complex resulting in the recruitment of deacetylase activity to PPARγ target promoters and their consequent repression.81 When pRb is inactivated by phosphorylation upon induction of differentiation, this complex is dissociated and PPARγ is then free to interact with HATs CBP/p300 and activate transcription.81,93,94 Downregulation of CBP/p300 expression significantly reduces adipocyte differentiation. 93,95 Similarly, coexpression of HAT SRC1 with PPARγ enhances the transcriptional activity of the factor and adipogenesis, whereas coexpression of corepressor NCoR suppresses it.96 In 3T3-L1 cells, addition of PPARγ ligands breaks the PPARγ/NCoR complex and results in activation of PPARγ transcriptional activity.97,98 At the same time, binding to a ligand also enhances interaction of PPARγ with CBP/p300, although this interaction may also happen in the absence of ligand.94

Cyclin D1 blocks PPARγ activity by recruiting HDAC1 and HDAC3 as well as the H3K9 HMT SUV39H1 to PPARγ target promoters, resulting in decreased expression of adipogenic genes.99 Cyclin D3 and its kinase partner CDK6, on the other hand, have a positive role in differentiation by binding to and phosphorylating PPARγ resulting in transcriptional activation of the factor.100 Thus, a relay of cyclins occurs during adipogenesis. At the beginning of differentiation elevated levels of Cyclin D1 block PPARγ transcriptional activity, whereas after mitotic clonal expansion, the decrease of Cyclin D1 and increase of Cyclin D3 ensure the activation of the factor.

Together with acetylation, histone methylation has been shown to participate actively in the regulation of adipocyte differentiation. Mouse embryos deficient for the histone arginine methyltransferase CARM1 display reduced BAT and silencing of the enzyme in 3T3-L1 cells results in impaired adipogenesis due to decreased PPARγ trans-activation.101 Deletion of the H3K9 demethylase JMJD1A in mice results in obesity and metabolic syndrome.102,103 JMJD1A is recruited to the PPRE of the ucp1 gene, resulting in H3K9 demethylation and increased transcription. 103 Decreased UCP1 expression in absence of JMJD1A may contribute to defective thermogenesis and fat burning in the knockout mice. In contrast, targeted inactivation of the H3K4 HMT MLL3 in mouse results in PPARγ-dependent impaired adipogenesis.104 MLL3 or its paralogue MLL4 form the ASCOM complexes that regulate the transcriptional activity of PPARγ by being recruited to PPARγ target promoters, such as that of lipid transporter aP2, and increasing H3K4 methylation and consequently transcription.104

We have observed that the promoters of key adipogenic genes, such as the hormone apm1, present significant levels of H3K4me2 in preadipocytes, when these genes are not expressed. Moreover, this signal coincides with significant loading of RNA Pol II at the same regions, suggesting that this mark is related to developmentally poised chromatin.50 At the same time the apm1 promoter has been shown to present detectable levels of H3K9me2 in preadipocytes,105 thus indicating the existence of a bivalent mark in this promoter in precursor cells. During terminal differentiation, significant increases in both histone H3 acetylation and H3K4me3 coincide with upregulation of the main adipogenic genes, whereas a concomitant decrease in histone acetylation and H3K4 trimethylation is observed in the promoter of the progressively silenced preadipocyte-specific gene dlk1.50 Accordingly with this increase in H3K4me3 during adipogenesis, a recent report has shown that PTIP, a protein that associates with H3K4 HMTs MLL3 and MLL4 and regulates histone methylation, is essential for adipogenesis.106,107 Deletion of PTIP impairs the enrichment of H3K4me3 and the loading of RNA Pol II on pparg and cebpa promoters, consequently impairing induction of PPARγ and C/EBPα expression.

Monomethylation of H4K20 also increases in adipogenic genes during adipogenesis and has been shown to play an important role in this process.108 In a feedback loop, the H4K20 HMT Set7/Setd8 is upregulated by PPARγ during adipogenesis, at the time that this enzyme positively regulates the expression of PPAR? and its target genes trough H4K20 monomethylation.

The role of miRNAs in adipogenesis.

In recent years a number of works have demonstrated a key role of microRNAs in cell differentiation, including adipogenesis. MicroRNAs are single-stranded RNAs of 19–25 nucleotides that originate from endogenous hairpin-shaped transcripts. These miRNAs interact with their targets by base-pairing and result in silencing by either translational repression of cleavage of the target RNA. A wide number of miRNAs, such as miR-103 and miR-143 increase during in vivo and in vitro adipocyte differentiation and their ectopic expression increases adipogenesis.109 Interestingly, these miRNAs were found to be downregulated in mouse models of obesity, thus suggesting that they play a role in phenotypic maintenance and function of mature adipose cells. Another report showed downregulation of miR-31, miR-125b-5p and miR-326 during adipocyte differentiation from adipose-derived stem cells, suggesting a role for these miRNAs in this process.110 miR-27 has been shown to downregulate PPARγ and C/EBPα expression and subsequently plays a key role in adipogenesis.111 The expression of miR-27 decreases during adipogenesis and, interestingly, was found to be increased in adipose tissue of obese mouse. Finally, significant correlation of the levels of differents miRNAs with metabolic parameters such as fasting plasma glucose or circulating leptin and adiponectin has also been reported,112 highlighting the importance of these factors in the regulation of metabolism.

Concluding Remarks

Epigenetics has become an aspect that can not be overlooked when attempting to understand the regulation of gene expression. In this review, we have attempted to put into perspective recent studies highlighting the interplay between the chromatin environment and transcriptional regulation in the context of adipocyte differentiation. A number of works have revealed the importance of epigenetics in the transition from undifferentiated to differentiated cells. Adipogenesis appears to involve the combined regulation of chromatin structure, histone modifications and the regulation by miRNAs. All these mechanisms are employed by sequence-specific transcription factors to coordinately regulate expression of sets of genes that either maintain pluripotency, direct cells towards a particular lineage, or maintain the differentiated state of the mature cells. The continuing study and identification of these epigenetic signals and the factors that are responsible of their establishment and turnover would ensure the discovery of new targets for the treatment of obesity and its associated diseases.

Abbreviations

BAT	=	brown adipose tissue
BMP	=	bone morphogenetic protein
C/EBP	=	CCAAT/enhancer binding protein
CtBP	=	C-terminal-binding protein
ESCs	=	embryonic stem cells
HAT	=	histone acetyltransferase
HDAC	=	histone deacetylase
HMT	=	histone methyltransferase
MEFs	=	mouse embryo fibroblasts
MSCs	=	mesenchymal stem cells
NCoR	=	nuclear corepressor
PGC1	=	PPARγ coactivator 1
PPAR	=	peroxisome-proliferator associated receptor
PPRE	=	PPAR response element
RIP140	=	receptor interacting protein 140
SANT	=	switching-defective protein 3, adaptor 2, nuclear receptor corepresor, and transcription factor IIIB domain
SRF	=	serum response factor
TIP	=	tension-induced/inhibited protein
UCP1	=	uncoupling protein 1
WAT	=	white adipose tissue

Figures and Tables

Figure 1 Cell fate decision of MSCs. Commitment of MSCs into the chondro-osteogenic lineage has been shown to rely on the expression of the master transcription factor Runx2.114–116 Sox proteins are the major transcription factors that regulate chondrogenic differentiation, whereas Runx2/Cbfa1 and Osterix/Osx participate in both osteocyte and chondrocyte differentiation. The Wnt canonical pathway activates Runx2 and induces an osteogenic fate,117 at the expense of the adipogenic lineage.55,56 On the other hand, BMP2 induces the expression of Sox9 but represses Runx2, thus driving MSCs into chondrogenesis. Commitment to the myogenic lineage and myogenesis are governed by muscle-specific transcription factors that include Myf5, MyoD, myogenin and MRF4,118,119 as well as the induction of TIP-1 expression, whereas expression of TIP-3 directs the progenitors to the adipogenic lineage.59 The myogenic precursors can give rise to either myocytes or brown adipocytes, depending on the absence or presence of adipogenic PRDM16.70,72 However, white adipocytes can also be transdifferentiated into brown adipocytes under different circumstances.

Figure 2 Epigenetic events during adipocyte differentiation. The schematics show the main epigenetic events that take place during adipogenesis represented on a hypothetical promoter incorporating characteristics from different adipogenic genes. In preadipocytes, C/EBPβ is already bound to the promoters of its target genes including pparg and cebpa75,76,80,82 but remains inactive by its interaction with HDAC1/mSin3A.83,84 Similarly, PPARγ is bound to its target promoters in association with a number of repressors including HDACs and histone methyltransferases.57,58,75,76,80–82,97,99 At this stage, most adipogenic genes are characterized by bivalent histone marks, such as methylation of H3K27, H3K9 and H3K4.35–38,50 Methylation of DNA is also found120 as well as recruitment of a still inactive RNA Pol II.50 Upon induction of differentiation, pRB phosporylation results in its inactivation and HDAC1 is dislodged from PPARγ which is now free to bind CBP/p300 and a number of other coactivators.81,94,95,100 In the early phases of differentiation, increased amounts of PPARγ displace HDAC1 from C/EBPβ allowing the factor to recruit the SWI/SNF remodeling complex and stimulate transcription of cebpa.83 Glucocorticoids have the same effect by stimulating C/EBPβ acetylation by GCN5.84,113 High levels of PPARγ and C/EBPα direct terminal differentiation. 75,76 Decrease of H3K27/H3K9 methylation is observed in adipogenic promoters, accompanied by increased histone acetylation and H3K4/H3K20 methylation.38,50,108 Another characteristic feature during terminal differentiation of adipocytes is the spread of di and tri-methylation of H3K4, acetylation of H3 and the Pol II to the coding region of the adipogenic genes, coinciding with high levels of transcription of those genes.38,50

Table 1 Histone modifying enzymes involved in adipogenesis

Histone Acetyltransferases (HATs)
Gcn5/PCAF family
Gcn5	Acetylates and activates C/EBPβ	113
CBP/p300 family
CBP/p300	Activate pparγ expression Coactivate PPARγ	60, 81, 93, 94
Others
SRC1	Activates PPARγ	96
Histone deacetylases (HDACs)
Class I HDACs
HDAC1	Inhibits PPARγ activity and PPARβ-target gene expression Inhibits C/EBPβ activity.	67, 83, 99
HDAC3	Inhibits expression of PPARγ-target genes ucp1 gene repression	67, 81, 92, 99
Class III HDACs (Sirtuins)
SIRT1	Attenuation of PPARγ activity	57, 58
Histone methyltransferases (HMTs)
SET domain lysine HMTs
MLL3-4	Activate PPARγ activity Stimulate pparg and cebpa expression	104, 106
SUV39H1	Blocks PPARγ-target gene expression	99
SETDB1	Represses PPARγ trans-activation	55
G9a	ucp1 gene repression	67
Set7/Setd8	Activates expression of pparγ and its target genes	108
Arginine HMTs
CARM1	Coctivates PPARγ activity	101
Histone demethylases (HDMs)
JmjC family demethylases
JMJD1A	Activates ucp1 expression	103

Acknowledgements

This work was supported by grants BFU2006-14251/BMC and SAF2006-07382 from the Spanish Ministry of Science and Innovation (MICINN) awarded to M.P. and R.G. respectively. We are grateful to Rosa Gasa, Perla Kaliman and Felicia Hanzu for critical reading of the manuscript.