Transcriptional networks and chromatin remodeling controlling ...

Review

Transcriptional networks and chromatin remodeling controlling adipogenesis Rasmus Siersbæk, Ronni Nielsen and Susanne Mandrup Department of Biochemistry and Molecular Biology, University of Southern Denmark, Odense, Denmark

Adipocyte differentiation is tightly controlled by a transcriptional cascade, which directs the extensive reprogramming of gene expression required to convert fibroblast-like precursor cells into mature lipid-laden adipocytes. Recent global analyses of transcription factor binding and chromatin remodeling have revealed ‘snapshots’ of this cascade and the chromatin landscape at specific time-points of differentiation. These studies demonstrate that multiple adipogenic transcription factors co-occupy hotspots characterized by an open chromatin structure and specific epigenetic modifications. Such transcription factor hotspots are likely to represent key signaling nodes which integrate multiple adipogenic signals at specific chromatin sites, thereby facilitating coordinated action on gene expression. Transcriptional waves controlling adipogenesis The molecular mechanisms controlling the conversion of preadipocytes to mature adipocytes (adipogenesis) have received much attention in the past decades due to the massive increase in the prevalence of obesity and obesityrelated diseases. Adipogenesis is controlled by a tightly regulated transcriptional cascade where the transcription factors activate or repress the expression of each other in a sequential manner. Key players in this transcriptional cascade include CCAAT/enhancer-binding protein (C/ EBP) family members (i.e. C/EBPa, C/EBPb, and C/EBPd) and the nuclear receptor peroxisome proliferator-activated receptor g (PPARg), which have been shown by both in vitro and in vivo studies to be important regulators of adipocyte differentiation [1]. The adipogenic cascade can be divided into at least two waves of transcription factors that drive the adipogenic program. The first wave is initiated by adipogenic stimuli that activate several early adipogenic factors including C/EBPb/d, Kru¨ppel-like factors (KLFs), cAMP response element binding protein (CREB), early growth response 2 (Krox20), and sterol regulatory element-binding protein 1c (SREBP-1c) (Figure 1; Box 1). These transcription factors in turn induce expression of the second wave of transcription factors, of which PPARg and C/EBPa are the most important, and these key adipogenic factors then induce expression of the gene program that leads to the mature adipocyte phenotype [1–3] (Figure 1; Box 1). Recently, genome-wide Corresponding author: Mandrup, S. ([email protected]).

56

studies have mapped the binding profiles of multiple adipogenic transcription factors of both the first and second transcription factor wave. These analyses revealed a novel direct crosstalk between transcription factors at shared genomic target sites. Here, we review recent advances in the understanding of chromatin remodeling events associated with adipocyte differentiation and discuss how these findings have provided novel insights into how adipogenic transcription factor networks orchestrate adipocyte differentiation. Early adipocyte differentiation A limited number of signaling pathways including bone morphogenic protein (BMP) and growth differentiation factor (GDF) signaling have been shown to regulate the commitment of mesenchymal stem cells to the adipocyte lineage and the differentiation of adipocyte precursor cells in vivo [4]. Thus, most of our knowledge about the molecular mechanisms that control adipocyte differentiation comes from studies of in vitro preadipocyte cell-line models or primary cell cultures. These model cell systems can be induced to differentiate into mature adipocytes using a standard hormonal cocktail containing insulin, a cAMPelevating agent, glucocorticoids (e.g. dexamethasone), and growth factors (e.g. from serum in the differentiation media) [5]. A multitude of studies using these in vitro model systems have revealed the core transcriptional cascade (i.e. the sequential induction of key adipogenic transcription factors) that controls adipocyte differentiation [3,5,6]. However, the precise molecular mechanisms through which these various transcription factors control different developmental stages of differentiation are only beginning

Glossary DHS site (DNase I hypersensitive site): genomic chromatin region that is hypersensitive to DNase I digestion compared to the surrounding DNA. Enhanceosome: enhancer-associated protein complex that drives transcriptional activation. HOT region (high-occupancy target region): genomic region that is bound by >8 transcription factors in whole animal studies. Hotspot: genomic region that is co-occupied by multiple transcription factors in a given cell (similar to MTL). MTL (Multiple transcription factor loci): genomic regions that are co-occupied by multiple transcription factors in a given cell (similar to hotspots). DR1 (direct repeat 1): two similar DNA sequences of 6 bp in the same orientation separated by 1 bp. PPRE (PPAR response element): a DR1 element where PPAR can bind directly and activate transcription.

1043-2760/$ – see front matter ß 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.tem.2011.10.001 Trends in Endocrinology and Metabolism, February 2012, Vol. 23, No. 2

Review

Trends in Endocrinology and Metabolism February 2012, Vol. 23, No. 2

Adipogenesis

Wnt10b Krox20

β-catenin

RAR

COUP-TFII

TR

GATA2/3

RORα

KLF4

C/EBPα

C/EBPβ

cAMP

Adipocyte phenotype

CREB CHOP10

C/EBPδ KLF5

PPARγ KLF15

GR Glucocorticoids Insulin

PPARγ ligand

FOXO1

KLF2 IRF4

STAT5A/B FBS SREBP-1c

LXR TRENDS in Endocrinology & Metabolism

Figure 1. Transcriptional cascade controlling adipocyte differentiation. Early adipogenic transcription factors (green) are activated directly by the adipogenic hormone cocktail consisting of glucocorticoids, insulin, a cAMP-elevating agent, and fetal bovine serum (FBS). This first wave of adipogenic factors directly induces the late adipogenic factors (blue), which coordinately induce genes linked to the adipocyte phenotype. Anti-adipogenic factors (orange) act on various adipogenic factors to inhibit adipocyte differentiation. Solid lines indicate regulation of gene expression, dashed lines indicate regulation of activity, and solid lines with arrowheads at both ends indicate that transcription factors are found in the same complex. Abbreviations: C/EBP homologous protein, CHOP10; retinoic acid receptor, RAR; RAR-related orphan receptor a, RORa; forkhead box O1, FOXO1; thyroid hormone receptor, TR; liver X receptor, LXR; chicken ovalbumin upstream promoter transcription factor II, COUP-TFII.

to be elucidated. Recently, global approaches have emerged as powerful tools to study transcription factor action at a genome-wide level. These studies have greatly increased our understanding of the molecular mechanisms through which adipogenic transcription factors regulate adipocyte differentiation at different developmental stages. Importantly, extensive crosstalk between

transcription factors is emerging as an important mechanism through which various signaling pathways are integrated at the genomic level. Adipogenic signaling pathways converge at transcription factor hotspots during early differentiation Addition of the standard hormone cocktail of adipogenic

Box 1. Pro-adipogenic transcription factors C/EBPs (CCAAT/enhancer-binding proteins): these belong to the basic-leucine zipper (bZIP) family of transcription factors and exist in six different isoforms, four of which [i.e. a, b, d and z (CHOP-10)] have been demonstrated to play a role in adipocyte differentiation [57]. C/EBP family members bind as homo- or heterodimers to DNA [57]. CREB (cAMP response element binding protein): belongs to the bZIP transcription factor family and binds to DNA as a homodimer or as a heterodimer with other members of the family. Is activated by phosphorylation upon adipogenic hormone stimulation and is required for differentiation [58,59]. GR (glucocorticoid receptor): member of the steroid receptor family of nuclear receptors. Binds as a homodimer to DNA. An important regulator of early adipogenesis [18,21]. KLFs (Kru¨ppel-like factors): zinc-finger proteins containing three zinc-fingers separated by conserved spacers of seven amino acids (TGEKP(Y/F)X) [60–62]. This family of transcription factors contains both repressors and activators of transcription [62]. Seven of the 17 known KLF proteins have been shown to play a role in different phases of adipocyte differentiation (i.e. KLF2 [11] and -3 [63] inhibit differentiation, whereas KLF4 [12], -5 [64], -6 [65], -9 [66], and -15 [67] stimulate differentiation) (Figure 1). Krox20 (early growth response 2, Egr2): a member of the Egr family of zinc finger transcription factors. Is required in the early phase of adipocyte differentiation [10].

LXRs (liver X receptors): members of the nuclear receptor superfamily. These exist in two isoforms, LXRa and -b, that bind to DNA as heterodimers with RXR. Regulate lipid metabolism in mature adipocytes [68–70]. PPARg (peroxisome proliferator-activated receptor g): member of the nuclear receptor superfamily. PPARg exists in two different isoforms (PPARg1 and PPARg2) that bind to DNA together with the retinoic X receptor (RXR) and are activated by fatty acids and their derivatives. Play a pivotal role in adipocyte differentiation both in vivo and in vitro [25,26]. SREBP-1c (sterol regulatory element-binding protein-1c): belongs to the basic helix-loop-helix leucine zipper (bHLH-Zip) family of transcription factors. Promotes adipocyte differentiation [71]. STAT5A/B (signal transducer and activator of transcription 5A/B): belong to the STAT family of transcription factors and bind to DNA as homo- or heterodimers. STAT5A is particularly important for efficient adipocyte differentiation whereas STAT5B plays a more auxiliary role [20,72,73]. TR (thyroid hormone receptor): member of the nuclear receptor superfamily. Binds to DNA either as a homodimer or as a heterodimer with RXR. Plays a role in adipocyte differentiation both in vitro and in vivo [74,75].

57

Review


Box 2. Chromatin remodeling and associated techniques Chromatin is inherently repressive of transcription due to the limited accessibility of the DNA. Transcription factor binding to DNA is therefore often associated with remodeling of the chromatin template to increase accessibility of DNA. This increase in chromatin accessibility associated with protein binding can be detected as an increase in the local susceptibility to nucleolytic attack. In addition to disrupting the local nucleosome structure, protein binding is usually also associated with ‘activating’ histone marks on nucleosomes in the vicinity. The advent of deep sequencing platforms in recent years has resulted in an almost explosive increase in the number of genomewide maps of different histone modifications in various cell types. This has also resulted in the identification of multiple distinct chromatin states with different functions [76–79]. DNase I hypersensitive (DHS) regions are ‘sites of action’ in the genome (e.g. active promoters and enhancers) that have an increased DNA accessibility compared to surrounding sequences due to local remodeling of the chromatin structure. DNase I analysis involves limited DNase digestion of isolated nuclei followed by analysis of digested DNA by southern blotting, qPCR, or more recently by deep sequencing (DHS-seq), which allows unbiased genome-wide mapping of all DHS sites in a given cell type [80–83].

inducers to preadipocyte cell cultures (e.g. 3T3-L1 and 3T3F442A cells) leads to the activation of multiple early proadipogenic and repression of anti-adipogenic transcription factors (Figure 1). C/EBPb and d have been widely recognized for a long time as key early pro-adipogenic transcription factors that are involved in initiating the adipogenic program [7–9]. More recently, Krox20 [10] and members of the KLF family [11,12] have been demonstrated to be important early regulators of adipogenesis as well through their ability to induce expression of C/EBPb and PPARg (Figure 1). Glucocorticoids are important adipogenic inducers at least in vitro [13] and alterations in glucocorticoid signaling in vivo are associated with changes in fat mass and adipocyte metabolism [14–17]. However, the precise role of the glucocorticoid receptor (GR) during the adipocyte differentiation process has remained largely unexplored. Lazar and coworkers recently showed that GR is indeed necessary for adipocyte differentiation in vitro [18]. Using ChIP-seq (Box 2) to map transcription factor binding sites at a genome-wide scale, they demonstrated that GR binds transiently and cooperatively with C/EBPb to several thousand genomic regions six hours after induction of differentiation. These regions are transiently marked by H3K9ac early in differentiation, possess enhancer-like features such as binding of the coactivators p300 (histone acetyltransferase) and Med1 (Mediator subunit), and function as enhancers in reporter assays, indicating that they are in fact transiently active enhancer regions. This suggests that GR and C/EBPb drive adipocyte differentiation in response to external adipogenic stimuli through cooperative binding to transient enhancer regions. STAT5A is activated by growth hormone-induced tyrosine phosphorylation within few hours after stimulation with adipogenic inducers, and it has been reported to be required for efficient adipocyte differentiation [19,20]. It was also reported that STAT5A forms a complex with GR in the nucleus in the early phase of differentiation [19], suggesting that STAT5A may be recruited to GR-associated putative enhancers. Using ChIP-seq analysis, we have recently shown that STAT5A and GR do in fact co-occupy 58

Formaldehyde-assisted isolation of regulatory elements (FAIRE) is an alternative and simpler method that can be used to study the chromatin structure within cells [84,85]. This method involves isolation of nucleosome-depleted chromatin regions using sequential phenol– chloroform extractions. FAIRE in combination with deep sequencing (FAIRE-seq) and DHS-seq identify distinct but overlapping profiles of open chromatin [86]; however, the structural and functional differences between FAIRE sites and DHS sites are currently unknown. Chromatin immunoprecipitation (ChIP) can be used to identify interactions between a protein of interest (e.g. a transcription factor) and chromatin or to study the location of specific histone modifications (e.g. H3K27ac). This method involves protein–DNA crosslinking (typically accomplished by fixing the cells in a formaldehyde solution) followed by DNA fragmentation and immunoprecipitation of the protein of interest using specific antibodies. DNA from the immunoprecipitated protein–DNA complexes can be purified and analyzed by various methods. Real-time qPCR can be used to analyze ChIP’ed DNA in a gene-by-gene manner, whereas global platforms such as microarray (ChIP-chip) and more recently deep sequencing (ChIP-seq) allow unbiased analyses of the ChIP’ed DNA resulting in genomewide binding profiles of the protein of interest.

multiple chromatin regions early in adipogenesis [21]. Remarkably, we identified nearly 1000 transcription factor hotspots (see Glossary), where STAT5A, GR, C/EBPb/d, and the nuclear receptor RXR bind cooperatively four hours after induction of differentiation [21]. Binding of multiple transcription factors at such regions was shown to be functionally correlated with expression of early induced genes, suggesting that these regions are likely to represent important regulatory nodes (i.e. enhancers) where multiple adipogenic signals converge early in adipogenesis (Figure 2). Most of the identified hotspots have a transiently open chromatin configuration suggesting that they are specifically active during the early phase of differentiation, where they regulate cell-cycle progression as well as late-acting transcription factors (e.g. PPARg) (Figure 3, left). Taken together, these findings indicate the existence of a previously unrecognized high degree of crosstalk between multiple early adipogenic transcription factors directly on the chromatin template, which may serve to integrate various adipogenic signals and thereby fine-tune the transcriptional program accordingly. C/EBPb is an important early transcriptional coordinator C/EBPb is known to be required for adipocyte differentiation both in vivo and in vitro [7,8]. Lane and coworkers have previously shown that C/EBPb is highly induced and phosphorylated by MAPK within a few hours after adipogenic hormone stimulation; however, subsequent phosphorylation of C/EBPb by GSK3b after a 14 h time-lag dramatically increases the affinity of C/EBPb for naked DNA [9]. Interestingly, using ChIP (Box 2) to study interactions between C/EBPb and selected chromatin regions within cells, Imbalzano and colleagues showed that both C/EBPb and -d are already able to associate with DNA in a chromatin context within the first few hours after the induction of differentiation [22], in other words before their reported phosphorylation by GSK3b. Genome-wide profiling of C/EBPb and -d binding during the first hours of adipogenesis have extended these findings and shown that both of these C/EBPs associate with several thousand

Review


Adipogenic signals

C/EBPβ STAT5A

GR GR

RXR:?

C/EBPδ

PIC RNAP RNAPI RNAPII NAP I

Hotspot

Promoter TRENDS in Endocrinology & Metabolism

Figure 2. Model of integration of signaling pathways at transcription factor hotspots. Multiple early adipogenic signals activate a variety of adipogenic transcription factors, which associate with transcription factor hotspots. This allows integration of adipogenic signals on the chromatin template. Transcription factors associated with such hotspots coordinate the assembly of a functional enhanceosome and activate transcription of target genes. Abbreviations: pre-initiation complex, PIC; RNA polymerase II, RNAPII.

chromatin regions at these early time-points as described above [18,21]. This suggests that particular features of chromatin (e.g. histone modifications) may assist binding of C/EBPs to DNA in a chromatin context within cells during the first hours of adipogenesis. Remarkably, profiling of C/EBPb binding in preadipocytes demonstrates that C/EBPb actually associates with multiple chromatin regions even before the induction of differentiation, suggesting that C/EBPb may also play a role in the

C/EBPβ/δ

preadipocyte state [18,21]. Interestingly, C/EBPb seems to mark a subset of early transcription factor hotspots before the initiation of differentiation, and before the binding of other transcription factors and full-scale chromatin remodeling of these hotspots [21]. By doing so it may act as a pioneering factor for the formation of these hotspots as well as for the binding of early adipogenic transcription factors to other sites. Consistent with this, knockdown of C/EBPb in preadipocytes interferes with

PPARγ:RXR

Preadipocyte

GR

C/EBPα

STAT5A

Protein X/Y

1. Wave

4-24 hours Early enhanceosome en

Early enhanceosome en

Enhancer

Enhancer

2. Wave

Cell cycle Ribosome Translation Cellular growth Glucose catabolism Late-acting TFs

Day 2-6 Late enhanceosome

Late enhanceosome

Enhancer

Enhancer

Lipid metabolism Glucose metabolism Lipid binding Mitochondria Transport TRENDS in Endocrinology & Metabolism

Figure 3. Model for establishment of early and late enhanceosomes. (Left) The first wave of adipogenic transcription factors coordinates the transient formation of early enhanceosomes at transcription factor hotspots. This is associated with a transiently open chromatin structure. (Middle) Early transcription factors recruit coregulators to induce chromatin remodeling and form early enhanceosomes at transcription factors hotspots. These putative enhancer regions are subsequently ‘inherited’ by late adipogenic factors, which coordinate the assembly of late enhanceosomes at the same hotspots. (Right) Late adipogenic transcription factors direct de novo chromatin remodeling and the assembly of late enhanceosomes at late hotspots. TFs, transcription factors.

59

Review


Early hotspots are ‘inherited’ by late adipogenic transcription factors Imbalzano and coworkers have previously demonstrated that the PPARg2 promoter is remodeled within a few hours after induction of adipocyte differentiation using the DNase I hypersensitive (DHS) assay (Box 2) [23]. This remodeling event was shown to be dependent on C/EBPb and cFos as well as PKA signaling. Interestingly, PPARg2 is not expressed until later in adipogenesis (days 1–2), suggesting that remodeling of the PPARg2 promoter takes place before transcriptional activation of this gene [23,24]. We have recently used DHS-seq (Box 2) to map chromatin remodeling events associated with 3T3-L1 adipocyte differentiation on a global scale [21]. Remarkably, the chromatin landscape is extensively remodeled during the first hours after initiation of the adipogenic program, and this coincides with the recruitment of multiple early transcription factors (e.g. C/EBPs, GR, STAT5A, and RXR). Chromatin remodeling events generally correlate well with transcriptional changes of nearby genes; however, we also identified numerous loci where chromatin remodeling precedes the transcriptional activation of nearby genes, as previously reported for PPARg2 [23,24]. These findings suggest that early chromatin remodeling primes genomic regions for late adipogenic signals (i.e. late-acting transcription factors, coregulators or post-translational modifications). Our finding that C/EBPb binds to many PPARg binding sites before PPARg suggests that pioneering complexes containing C/EBPb remodel the chromatin structure at putative enhancers to assist subsequent binding of PPARg in late adipogenesis [21]. This would indicate that early enhanceosomes directly affect late transcriptional events by regulating chromatin accessibility (Figure 3, middle).

differentiation using ChIP-seq [33] and in mature adipocytes by ChIP-chip [34,35], ChIP-PET [36] and ChIP-seq [37,38] demonstrated that PPARg binding is specifically enriched in the vicinity of most induced adipocyte genes (including genes linked to lipid and glucose metabolism). These findings indicate that PPARg, in addition to being a key regulator of adipogenesis per se, is also a direct activator of the entire genomic reprogramming important for acquisition of the mature adipocyte phenotype. Interestingly, de novo motif analyses revealed, in addition to the DR-1-type PPAR response element (PPRE) (see Glossary), a striking enrichment of the C/EBP binding motif at PPARg binding sites [33,34]. C/EBPa ChIP-chip [34] and ChIP-seq [38] confirmed extensive colocalization (35–60%) of PPARg and C/EBPa, suggesting that these two key adipogenic factors cooperate directly on the chromatin template to induce the adipocyte gene program. Intriguingly, one of the early adipogenic C/EBP family members, C/EBPb, is also expressed in mature adipocytes, although at a lower level than during early adipocyte development, and this C/EBP subtype showed a similar binding pattern to C/EBPa in the mature adipocyte [34]. Depletion of C/EBPa/b or PPARg in mature adipocytes demonstrated that both PPARg and C/EBPs are required for maintaining high expression of several adipocyte genes in the vicinity of shared PPARg–C/EBPa binding sites [34]. Consistent with these findings, shared binding sites were enriched in the vicinity of highly induced, compared to modestly induced, adipocyte genes, suggesting that PPARg–C/EBPa cooperativity results in a massive induction of nearby target genes, also on a global scale [39]. Taken together, these studies indicate extensive direct crosstalk between the two key adipogenic transcription factors, PPARg and C/EBPa, on chromatin. Interestingly, however, depletion of either PPARg or C/EBPa does not affect chromatin binding of the other factor to the shared sites that were tested [34], suggesting a limited mutual dependence between the two factors in binding to chromatin in the mature adipocyte. Cumulatively, these results suggest that PPARg and C/EBPa are part of a late adipogenic transcription factor network that participates in the assembly of cofactor complexes at these shared sites to cooperatively induce transcriptional activation of adipocyte genes (Figure 3, right).

The late adipogenic program Extensive crosstalk between the key late adipogenic transcription factors PPARg and C/EBPa The first wave of early transcription factors described above directly activates expression of the second wave of adipogenic factors, of which the most important players are PPARg and C/EBPa (Figure 1). Interestingly, multiple early transcription factor hotspots are found in the vicinity of the PPARg gene, suggesting that several adipogenic signaling pathways orchestrate the transcriptional induction of this important nuclear receptor during adipogenesis [21]. PPARg has been known for a long time to be essential for adipogenesis both in vitro and in vivo [25,26]; however, the molecular mechanisms underlying PPARg action during adipocyte differentiation have only been studied for a limited number of target genes [27–32]. Interestingly, genome-wide profiling of PPARg throughout adipocyte

Conservation of PPARg and C/EBPa binding events in adipocytes Comparative genome-wide profiling of PPARg and C/EBPa binding in mouse and human adipocytes revealed limited conservation of binding events between the two species (approximately 10–20% of mouse binding sites are conserved in human adipocytes) [37,38,40]. This low interspecies conservation of binding events is consistent with findings for other factors in other cell types [41]. Interestingly, however, there is extensive colocalization of PPARg and C/EBPa both in mouse and human adipocytes, and sites that are co-occupied by these transcription factors in mouse are more likely to be retained in human adipocytes compared to sites where only one of the factors bind [38]. These findings showed that the extensive cooccupancy of PPARg and C/EBPa at putative enhancer regions is conserved between mouse and human, and

the recruitment of GR, STAT5A and RXR to early target sites [21]. Taken together, these studies indicate that C/EBPb associates with specific putative enhancer regions both in the preadipocyte state and in the initial stage of differentiation and that C/EBPb, in addition to its wellestablished role as activator of PPARg and C/EBPa expression, plays an important role in the establishment of the early enhanceosomes.

60

Review suggest that co-occupancy between these factors is functionally important for adipocyte differentiation and function. Comparative analyses focusing on gene regulation furthermore suggest that the genes associated with PPARg binding are significantly more conserved between mouse and human than the actual binding events, indicating that PPARg targets the same gene programs controlling lipid and glucose metabolism in both species in part through non-conserved binding sites [37,40]. These findings are consistent with a high degree of turnover of PPARg binding sites (i.e. a new binding event replaces an old one during the course of evolution) in the vicinity of adipocyte genes [38]. Other late-acting regulators of adipogenesis In addition to PPARg and C/EBPa, several other transcription factors, including KLF15 and sterol regulatory element-binding protein (SREBP)-1c, have been shown to regulate the late phase of adipogenesis (Figure 1; Box 1). Even though no genome-wide binding profiles of any of these late transcription factors have yet been reported, it is highly likely that they bind to genomic chromatin regions that are hypersensitive to DNase I and that are modified by activating histone marks – as is the case for PPARg and C/EBPa. In addition to transcription-activating factors, repressive transcription factors are also likely to bind to remodeled sites. Interestingly, Rosen and colleagues recently employed DHS site analysis in combination with qPCR to identify novel late transcriptional regulators of adipocyte differentiation [42]. The authors determined the DNase I susceptibility of 268 conserved regions within 50 kb of 27 known adipocyte genes and identified 32 differentiation-dependent DHS sites (i.e. regions increasing in accessibility during adipogenesis). Based on motif analysis of the sequences in these regions, they identified interferon regulatory factors (IRFs) as novel negative regulators of differentiation that may function as a ‘brake’ on adipogenesis through binding to DHS sites in the later stages of differentiation. Consistently, in response to a high-fat diet, adipocyte-specific IRF4 knockout mice have significantly increased fat mass compared to wild-type mice as a result of adipocyte hypertrophy, and this is probably related to the function of IRF4 as a repressor and activator of lipogenesis and lipolysis, respectively [43]. Importantly, these findings also indicate that transcriptional repressors bind to DHS sites, as do activating transcription factors. This further highlights the potential of DHS site analysis to identify ‘sites of action’ in the genome without any a priori knowledge of the transcriptional events taking place at these sites. Crosstalk between late adipogenic transcription factors and activating histone marks Multiple epigenetic modifiers have been shown to regulate adipocyte differentiation at various stages of differentiation [44–47]. Even though it is currently unknown to what extent these effects on adipocyte differentiation are mediated entirely through the catalytic domain of the coregulators, these findings point to an important role of epigenetic modifications in regulating adipogenesis. For


extensive reviews on this topic we refer to recent reviews [48,49]. We will focus here on recent studies investigating global associations between epigenetic marks and the function of adipogenic transcription factors. Global analyses of the activating histone marks H3K27ac and H3K9ac during adipocyte differentiation and in mature adipocytes, respectively, revealed that both marks are highly enriched at PPARg binding sites [34,37,50]. Interestingly, ectopic expression of PPARg in preadipocytes increases H3K9ac levels at a few selected PPARg binding sites [50]. Similarly, acute ectopic expression of PPARg2 in NIH-3T3 cells induced H3 and H4 acetylation at an enhancer in the vicinity of the well-characterized PPARg target gene aP2 [51]. These findings indicate that PPARg can induce deposition at target sites of at least some activating histone marks. However, most PPARg binding sites in mature adipocytes are actually located in regions that are already enriched in H3K27ac in the preadipocyte state, well before PPARg induction [37]. This suggests that this mark may be important for directing PPARg binding. Importantly, however, PPARg binding sites that show fivefold increased levels of H3K27ac during differentiation are more highly correlated with induced genes than PPARg sites that are already modified before induction of differentiation [37]. This suggests that, even though most PPARg binding sites in mature adipocytes are marked by H3K27ac already in the preadipocyte state, the PPARg binding events that direct acetylation of H3K27 (and presumably of other histone tail residues such as H3K9) during differentiation may be functionally more important for the regulation of adipocyte genes. Sakai and coworkers recently demonstrated that PPARg directly induces expression of Setd8, a SET-domain-containing histone methyl transferase (HMT), which is required for adipogenesis and is expressed at elevated levels in two mouse models of obesity [35]. Setd8 monomethylates H4K20 at multiple adipocyte genes including the PPARg gene itself, and changes in the levels of this histone mark were found to correlate with changes in gene expression during differentiation. Based on these findings, the authors propose that adipocyte differentiation is controlled by a positive feedback loop in which PPARg induces expression of Setd8, which in turn activates PPARg expression by methylating H4K20. Taken together, these genome-wide studies indicate important functions of epigenetic marks in the transcriptional reprogramming of gene expression during adipogenesis. Detailed molecular studies are necessary to delineate the precise molecular mechanisms through which different histone marks regulate the binding and activity of adipogenic transcription factors. Transcription factor hotspots in other model systems The recent explosion in genome-wide maps of transcription factor binding has revealed several examples where different transcription factors associate with the same genomic regions. Pu and colleagues recently reported that multiple cardiac transcription factors co-occupy several thousand genomic regions in cardiomyocytes [52]. Interestingly, a subset of these putative enhancer regions were found to be 61

Review active in transient reporter assays both in vivo and in vitro, suggesting that they represent functionally important regulatory sites involved in coordinating cardiac gene expression. Multiple signaling pathways have also been found to converge at specific genomic regions termed multiple transcription factor-binding loci (MTL) in embryonic stem cells [53]. Several transcription factors seem to coordinate the assembly of enhanceosomes at these loci and regulate stem cell-specific gene expression. Taken together, these studies, and the recent studies on adipocyte differentiation described above, indicate that transcription factors do not work in isolated transcriptional programs but instead function as part of transcription factor networks that integrate multiple signaling pathways at specific genomic regions. This extensive co-occupancy of transcription factors in the genome may represent an important avenue to achieve cell type-specific transcriptional responses to activation of a particular transcription factor [50,54]. The modENCODE consortium, which has set out to characterize the genome of the model organisms Drosophila melanogaster and Caenorhabditis elegans, has recently identified genomic regions of approximately 400 bp termed high occupancy target (HOT) regions (see Glossary) that, at a whole-organism level, are occupied by at least 15 and 8 transcription factors, in C. elegans and D. melanogaster, respectively [55,56]. Notably, genes containing HOT regions in their promoter are ubiquitously expressed and have an essential developmental function, whereas expression of genes associated with ‘factor-specific’ binding sites (occupied by a subset of 1–4 of the investigated transcription factors) is generally restricted to a few tissues [55]. It should be noted that these results were obtained with whole animal samples and that the HOT regions therefore probably represent regulatory regions that are targeted by different factors in different cell types. By contrast, hotspots and MTL are genomic regions co-occupied by multiple transcription factors in the same cell at a specific time point. However, it is likely that many of the HOT regions will also turn out to be hotspots at a cellular level. Concluding remarks The adipocyte differentiation systems constitute excellent model systems for studying the establishment of enhanceosomes during development. The 3T3-L1 preadipocyte cell line is particular well-suited due to its highly synchronous and efficient differentiation into mature adipocytes in response to adipogenic stimuli. Given the extent of transcription factor co-occupancy described in this cell line, cooperation between multiple transcription factors in coordinating reprogramming of gene expression associated with adipocyte differentiation is likely to be functionally highly important. Thus, future studies should aim at identifying the molecular mechanisms underlying the establishment of enhanceosomes at adipogenic transcription factor hotspots to obtain a better understanding of the transcriptional control of adipogenesis and of the molecular mechanisms regulating enhancer activity in general. Specifically, it will be interesting to elucidate the sequence of transcription factor and coregulator binding to hotspots to identify the factor(s) that initiate formation of active 62


enhanceosomes at these sites. It will also be of great importance to determine the role of chromatin structure and histone modifications in the establishment of active enhanceosomes. Finally, little is known about which other transcription factors co-occupy chromatin together with PPARg and C/EBPa in the late adipogenic wave of transcription factors. We predict that these transcription factors, similar to the first wave of adipogenic transcription factors, cooperate in hotspots, and we propose that the action of each individual factor is highly dependent on cooperating factors. Given the clinical importance of PPARg in the treatment of insulin resistance, it will be interesting to identify these cooperating factors and determine whether modulation of their activities can be used to fine-tune PPARg activity in adipocytes as well as in other cell types. Acknowledgments We thank members of the Mandrup laboratory for comments to the manuscript. This work was supported by the Danish Natural Science Research Council and the Novo Nordisk Foundation.

References 1 Farmer, S.R. (2006) Transcriptional control of adipocyte formation. Cell Metab. 4, 263–273 2 Lefterova, M.I. and Lazar, M.A. (2009) New developments in adipogenesis. Trends Endocrinol. Metab. 20, 107–114 3 Rosen, E.D. and MacDougald, O.A. (2006) Adipocyte differentiation from the inside out. Nat. Rev. Mol. Cell Biol. 7, 885–896 4 Zamani, N. and Brown, C.W. (2011) Emerging roles for the transforming growth factor-beta superfamily in regulating adiposity and energy expenditure. Endocr. Rev. 32, 387–403 5 Hwang, C.S. et al. (1997) Adipocyte differentiation and leptin expression. Annu. Rev. Cell Dev. Biol. 13, 231–259 6 MacDougald, O.A. and Mandrup, S. (2002) Adipogenesis: forces that tip the scales. Trends Endocrinol. Metab. 13, 5–11 7 Tanaka, T. et al. (1997) Defective adipocyte differentiation in mice lacking the C/EBPbeta and/or C/EBPdelta gene. EMBO J. 16, 7432– 7443 8 Tang, Q.Q. et al. (2003) CCAAT/enhancer-binding protein beta is required for mitotic clonal expansion during adipogenesis. Proc. Natl. Acad. Sci. U.S.A. 100, 850–855 9 Tang, Q.Q. et al. (2005) Sequential phosphorylation of CCAAT enhancer-binding protein beta by MAPK and glycogen synthase kinase 3beta is required for adipogenesis. Proc. Natl. Acad. Sci. U.S.A. 102, 9766–9771 10 Chen, Z. et al. (2005) Krox20 stimulates adipogenesis via C/EBPbetadependent and -independent mechanisms. Cell Metab. 1, 93–106 11 Banerjee, S.S. et al. (2003) The Kru¨ppel-like factor KLF2 inhibits peroxisome proliferator-activated receptor-gamma expression and adipogenesis. J. Biol. Chem. 278, 2581–2584 12 Birsoy, K.I. et al. (2008) Transcriptional regulation of adipogenesis by KLF4. Cell Metab. 7, 339–347 13 Pantoja, C. et al. (2008) Glucocorticoid signaling defines a novel commitment state during adipogenesis in vitro. Mol. Biol. Cell 19, 4032–4041 14 Yu, C.Y. et al. (2010) Genome-wide analysis of glucocorticoid receptor binding regions in adipocytes reveal gene network involved in triglyceride homeostasis. PLoS ONE 5, e15188 15 Morton, N.M. et al. (2004) Novel adipose tissue-mediated resistance to diet-induced visceral obesity in 11beta-hydroxysteroid dehydrogenase type 1-deficient mice. Diabetes 53, 931–938 16 Kershaw, E.E. et al. (2005) Adipocyte-specific glucocorticoid inactivation protects against diet-induced obesity. Diabetes 54, 1023–1031 17 Masuzaki, H. et al. (2001) A transgenic model of visceral obesity and the metabolic syndrome. Science 294, 2166–2170 18 Steger, D.J. et al. (2010) Propagation of adipogenic signals through an epigenomic transition state. Genes Dev. 24, 1035–1044

Review 19 Baugh, J.E. et al. (2007) The modulation of STAT5A/GR complexes during fat cell differentiation and in mature adipocytes. Obesity 15, 583–590 20 Nanbu-Wakao, R. et al. (2002) Stimulation of 3T3-L1 adipogenesis by signal transducer and activator of transcription 5. Mol. Endocrinol. 16, 1565–1576 21 Siersbaek, R. et al. (2011) Extensive chromatin remodelling and establishment of transcription factor ‘hotspots’ during early adipogenesis. EMBO J. 30, 1459–1472 22 Salma, N. et al. (2006) Temporal recruitment of CCAAT/enhancerbinding proteins to early and late adipogenic promoters in vivo. J. Mol. Endocrinol. 36, 139–151 23 Xiao, H. et al. (2011) Chromatin accessibility and transcription factor binding at the PPARgamma2 promoter during adipogenesis is protein kinase A-dependent. J. Cell. Physiol. 226, 86–93 24 Salma, N. et al. (2004) Temporal recruitment of transcription factors and SWI/SNF chromatin-remodeling enzymes during adipogenic induction of the peroxisome proliferator-activated receptor gamma nuclear hormone receptor. Mol. Cell. Biol. 24, 4651–4663 25 Barak, Y. et al. (1999) PPAR gamma is required for placental, cardiac, and adipose tissue development. Mol. Cell 4, 585–595 26 Rosen, E.D. et al. (1999) PPAR gamma is required for the differentiation of adipose tissue in vivo and in vitro. Mol. Cell 4, 611–617 27 Tontonoz, P. et al. (1994) mPPAR gamma 2: tissue-specific regulator of an adipocyte enhancer. Genes Dev. 8, 1224–1234 28 Helledie, T. et al. (2002) The gene encoding the acyl-CoA-binding protein is activated by peroxisome proliferator-activated receptor gamma through an intronic response element functionally conserved between humans and rodents. J. Biol. Chem. 277, 26821–26830 29 Tomaru, T. et al. (2009) Adipocyte-specific expression of murine resistin is mediated by synergism between peroxisome proliferatoractivated receptor gamma and CCAAT/enhancer-binding proteins. J. Biol. Chem. 284, 6116–6125 30 Schoonjans, K. et al. (1996) PPAR alpha and PPAR gamma activators direct a distinct tissue-specific transcriptional response via a PPRE in the lipoprotein lipase gene. EMBO J. 15, 5336–5348 31 Shimizu, M. et al. (2004) Tissue-selective, bidirectional regulation of PEX11a and perilipin genes through a common peroxisome proliferator response element. Mol. Cell. Biol. 24, 1313–1323 32 Olswang, Y. et al. (2002) A mutation in the peroxisome proliferatoractivated receptor gamma-binding site in the gene for the cytosolic form of phosphoenolpyruvate carboxykinase reduces adipose tissue size and fat content in mice. Proc. Natl. Acad. Sci. U.S.A. 99, 625– 630 33 Nielsen, R. et al. (2008) Genome-wide profiling of PPARgamma:RXR and RNA polymerase II occupancy reveals temporal activation of distinct metabolic pathways and changes in RXR dimer composition during adipogenesis. Genes Dev. 22, 2953–2967 34 Lefterova, M.I. et al. (2008) PPARgamma and C/EBP factors orchestrate adipocyte biology via adjacent binding on a genome-wide scale. Genes Dev. 22, 2941–2952 35 Wakabayashi, K.I. et al. (2009) The peroxisome proliferator-activated receptor gamma/retinoid x receptor alpha heterodimer targets the histone modification enzyme PR-Set7/Setd8 gene and regulates adipogenesis through a positive feedback loop. Mol. Cell. Biol. 29, 3544–3555 36 Hamza, M.S. et al. (2009) De novo identification of PPARgamma/RXR binding sites and direct targets during adipogenesis. PLoS ONE 4, e4907 37 Mikkelsen, T.S. et al. (2010) Comparative epigenomic analysis of murine and human adipogenesis. Cell 143, 156–169 38 Schmidt, S. et al. (2011) Cross species comparison of C/EBPalpha and PPARgamma profiles in mouse and human adipocytes reveals interdependent retention of binding sites. BMC Genomics 12, 152 39 Siersbæk, R. et al. (2010) PPARgamma in adipocyte differentiation and metabolism - Novel insights from genome-wide studies. FEBS Lett. 584, 3242–3249 40 Soccio, R.E. et al. (2011) Species-specific strategies underlying conserved functions of metabolic transcription factors. Mol. Endocrinol. 25, 694–706 41 Dowell, R.D. (2010) Transcription factor binding variation in the evolution of gene regulation. Trends Genet. 26, 468–475


42 Eguchi, J. et al. (2008) Interferon regulatory factors are transcriptional regulators of adipogenesis. Cell Metab. 7, 86–94 43 Eguchi, J. et al. (2011) Transcriptional control of adipose lipid handling by IRF4. Cell Metab. 13, 249–259 44 Chatterjee, T.K. et al. (2011) Histone deacetylase 9 is a negative regulator of adipogenic differentiation. J. Biol. Chem. 286, 27836– 27847 45 van Beekum, O. et al. (2008) The adipogenic acetyltransferase Tip60 targets activation function 1 of peroxisome proliferator-activated receptor gamma. Endocrinology 149, 1840–1849 46 Musri, M.M. et al. (2010) Histone demethylase LSD1 regulates adipogenesis. J. Biol. Chem. 285, 30034–30041 47 Yadav, N. et al. (2008) CARM1 promotes adipocyte differentiation by coactivating PPAR gamma. EMBO Rep. 9, 193–198 48 Sugii, S. and Evans, R.M. (2011) Epigenetic codes of PPARgamma in metabolic disease. FEBS Lett. 585, 2121–2128 49 Koppen, A. and Kalkhoven, E. (2010) Brown vs white adipocytes: the PPARgamma coregulator story. FEBS Lett. 584, 3250–3259 50 Lefterova, M.I. et al. (2010) Cell-specific determinants of peroxisome proliferator-activated receptor gamma function in adipocytes and macrophages. Mol. Cell. Biol. 30, 2078–2089 51 Nielsen, R. et al. (2006) Peroxisome proliferator-activated receptor subtype- and cell-type-specific activation of genomic target genes upon adenoviral transgene delivery. Mol. Cell. Biol. 26, 5698–5714 52 He, A. et al. (2011) Co-occupancy by multiple cardiac transcription factors identifies transcriptional enhancers active in heart. Proc. Natl. Acad. Sci. U.S.A. 108, 5632–5637 53 Chen, X. et al. (2008) Integration of external signaling pathways with the core transcriptional network in embryonic stem cells. Cell 133, 1106–1117 54 Heinz, S. et al. (2010) Simple combinations of lineage-determining transcription factors prime cis-regulatory elements required for macrophage and B cell identities. Mol. Cell 38, 576–589 55 Gerstein, M.B. et al. (2010) Integrative analysis of the Caenorhabditis elegans genome by the modENCODE Project. Science 330, 1775–1787 56 Negre, N. et al. (2011) A cis-regulatory map of the Drosophila genome. Nature 471, 527–531 57 Ramji, D.P. and Foka, P. (2002) CCAAT/enhancer-binding proteins: structure, function and regulation. Biochem. J. 365, 561–575 58 Zhang, J.W. et al. (2004) Role of CREB in transcriptional regulation of CCAAT/enhancer-binding protein beta gene during adipogenesis. J. Biol. Chem. 279, 4471–4478 59 Reusch, J.E.B. et al. (2000) CREB activation induces adipogenesis in 3T3-L1 Cells. Mol. Cell. Biol. 20, 1008–1020 60 Dang, D.T. et al. (2000) The biology of the mammalian Kru¨ppel-like family of transcription factors. Int. J. Biochem. Cell Biol. 32, 1103–1121 61 Pearson, R. et al. (2008) Kru¨ppel-like transcription factors: a functional family. Int. J. Biochem. Cell Biol. 40, 1996–2001 62 Kaczynski, J. et al. (2003) Sp1- and Kru¨ppel-like transcription factors. Genome Biol. 4, 206 63 Sue, N. et al. (2008) Targeted disruption of the basic kruppel-like factor gene (Klf3) reveals a role in adipogenesis. Mol. Cell. Biol. 28, 3967–3978 64 Oishi, Y. et al. (2005) Kru¨ppel-like transcription factor KLF5 is a key regulator of adipocyte differentiation. Cell Metab. 1, 27–39 65 Li, D. et al. (2005) Kru¨ppel-like factor-6 promotes preadipocyte differentiation through histone deacetylase 3-dependent repression of DLK1. J. Biol. Chem. 280, 26941–26952 66 Pei, H. et al. (2011) Kruppel-like factor KLF9 regulates PPARg transactivation at the middle stage of adipogenesis. Cell Death Differ. 18, 315–327 67 Mori, T. et al. (2005) Role of Kru¨ppel-like Factor 15 (KLF15) in transcriptional regulation of adipogenesis. J. Biol. Chem. 280, 12867–12875 68 Juvet, L.K. et al. (2003) On the role of liver X receptors in lipid accumulation in adipocytes. Mol. Endocrinol. 17, 172–182 69 Darimont, C. et al. (2006) Liver X receptor preferentially activates de novo lipogenesis in human preadipocytes. Biochimie 88, 309–318 70 Stenson, B.M. et al. (2011) Liver X receptor (LXR) regulates human adipocyte lipolysis. J. Biol. Chem. 286, 370–379 71 Kim, J.B. and Spiegelman, B.M. (1996) ADD1/SREBP1 promotes adipocyte differentiation and gene expression linked to fatty acid metabolism. Genes Dev. 10, 1096–1107 63

Review 72 Jung, H.S. et al. (2011) Peroxisome proliferator-activated receptor gamma/signal transducers and activators of transcription 5A Pathway Plays a key factor in adipogenesis of human bone marrowderived stromal cells and 3T3-L1 preadipocytes. Stem Cells Dev. DOI: 10.1089/scd.2010.0591 73 Floyd, Z.E. and Stephens, J.M. (2003) STAT5A promotes adipogenesis in nonprecursor cells and associates with the glucocorticoid receptor during adipocyte differentiation. Diabetes 52, 308–314 74 Ying, H. et al. (2007) Impaired adipogenesis caused by a mutated thyroid hormone a1 receptor. Mol. Cell. Biol. 27, 2359–2371 75 Mishra, A. et al. (2010) Adipogenesis is differentially impaired by thyroid hormone receptor mutant isoforms. J. Mol. Endocrinol. 44, 247–255 76 Rada-Iglesias, A. et al. (2011) A unique chromatin signature uncovers early developmental enhancers in humans. Nature 470, 279–283 77 Creyghton, M.P. et al. (2010) Histone H3K27ac separates active from poised enhancers and predicts developmental state. Proc. Natl. Acad. Sci. U.S.A. 107, 21931–21936 78 Barski, A. et al. (2007) High-resolution profiling of histone methylations in the human genome. Cell 129, 823–837

64


79 Wang, Z. et al. (2008) Combinatorial patterns of histone acetylations and methylations in the human genome. Nat. Genet. 40, 897–903 80 Boyle, A.P. et al. (2008) High-resolution mapping and characterization of open chromatin across the genome. Cell 132, 311–322 81 Crawford, G.E. et al. (2006) Genome-wide mapping of DNase hypersensitive sites using massively parallel signature sequencing (MPSS). Genome Res. 16, 123–131 82 Hesselberth, J.R. et al. (2009) Global mapping of protein–DNA interactions in vivo by digital genomic footprinting. Nat. Methods 6, 283–289 83 John, S. et al. (2011) Chromatin accessibility pre-determines glucocorticoid receptor binding patterns. Nat. Genet. 43, 264–268 84 Nagy, P.L. et al. (2003) Genomewide demarcation of RNA polymerase II transcription units revealed by physical fractionation of chromatin. Proc. Natl. Acad. Sci. U.S.A. 100, 6364–6369 85 Hogan, G.J. et al. (2006) Cell cycle-specified fluctuation of nucleosome occupancy at gene promoters. PLoS Genet. 2, 1433–1450 86 Song, L. et al. (2011) Open chromatin defined by DNaseI and FAIRE identifies regulatory elements that shape cell-type identity. Genome Res. 21, 1757–1767