ES cells and nuclear dynamics
From the chromosome territory occupation and genome distribution inside the nucleus
, it is clear that the epigenome is dynamic and, that among other processes, it contributes to gene expression and cell differentiation [
3–5]. ES cells present a different nuclear architecture and dynamics than differentiated cells [
6], indicating that ES cells experience drastic and progressive changes during the differentiation process.
ES cell nuclei are larger than those of differentiated cells, globally, ES cells have a more relaxed chromatin configuration and particular epigenetic features
. When differentiation programs are turned on, a gradual and organized redistribution of the genome occurs inside the nucleus, resulting in a rapid reorganization of large areas of the genome that acquire heterochromatin conformation [
7]. Indeed, it has been proposed that, the regulated formation of heterochromatin is one of the most critical signals for differentiation [
8].
Then, ES cells’ chromatin is globally more de-condensed as compared with differentiated cells and has particular epigenetic features (see below).
Chromatin modifications: histone modifications and histone variants
The recent development of genome-scale chromatin analyses, in particular for a large set of histone covalent modifications, has changed our vision about the chromatin structure forming the skeleton of genes and surrounding intergenic regions, including regulatory elements [
9]. Such modifications contribute to the establishment of the ES cell global chromatin configuration and impact on gene expression regulation; ES cell self-renewal and differentiation [
10,
11].
Indeed, the capacity of ES cells to respond to differentiation stimuli and acquire a particular cell fate might be determined by a very specific epigenetic trait known as bivalent chromatin. Bivalent chromatin domains are enriched in histone H3 tri-methylated and di/tri-methylated at lysines 4 and 27 (H3K27me3 and H3K4me2/me3), respectively [
12–15]. H3K27me3 and H3K4me are marks associated with transcriptionally inactive and active chromatin, respectively (). These opposing marks are thought to provide bivalent genes, which are expressed at basal levels in ES cells, with the plasticity to reach full expression potential or be repressed upon activation of specific differentiation programs. Indeed, many of the genes in bivalent domains encode for transcription factors directing tissue-specific differentiation programs. This chromatin organization suggests that histone modifiers inducing H3K27me3 and H3K4me3 have a key function in maintaining pluripotency [
16,
17]. Importantly, bivalent chromatin is not the only epigenomic trait associated with ES cells
().
Epigenetic silencing associated with histone lysine 9 methylation also contributes to the ES cell maintenance. It is known that globally, H3K9me2 and H3K9me3 histone marks, associated with repressive chromatin, are maintained at low levels in ES and they become enriched in differentiated cells () [
6]. Ng and collaborators showed that the H3K9me demethylases Jmjd1a and Jmjd2c are important for ES cell self-renewal [
18]. Notoriously, Oct4 positively regulates the expression of these histone demethylases, which maintain the
Tcl1 and
Nanog genes (two key transcription factors for self-renewal in ES cells) in an open chromatin configuration by H3K9me2 and H3K9me3 demethylation, respectively [
18]. Furthermore, the down regulation of Oct4 during differentiation favors decreased
Jmjd1a and
Jmjd2c transcription, facilitating the incorporation of H3K9me2 and H3K9me3 and the epigenetic silencing of pluripotency-associated genes. Thus, histone demethylases play a key function in ES cell pluripotency maintenance and differentiation.
Another relevant aspect of ES cell epigenetics is the incorporation of histone variants. Allis and collaborators recently demonstrated that the histone variant H3.3 interacts with active and repressed genes in ES cells, in a HIRA-dependent manner [
19]. HIRA is a histone chaperone specific for histone H3.3 that mediates replication-independent nucleosomes assembly [
20] and appears to limit ES cell differentiation [
6], suggesting that indeed H3.3 might influence the ES cell status. Other complexes have been found to deposit H3.3 in ES cells. The death domain-associated protein (Daxx) and the α-thalassemia X-linked mental retardation protein (ATRX) deposit H3.3 at constitutive heterochromatin in murine ES cells [
21]. However, if ATRX-Daxx and its associated deposition of H3.3 have a function in pluripotency remains to be addressed.
The human ES cell DNA methylome
DNA methylation is important for establishing the dynamic chromatin configuration of the genome in pluripotent ES cells, and for coordinating genomic reorganization during cell differentiation. DNA methylation and Polycomb-repressive proteins (PcG) are both required for pluripotency; they impede premature expression of differentiation regulators [
22]. However, although DNA methylation is critical in early embryonic differentiation, cellular memory and development [
23], its function in stem cell pluripotency and differentiation remains a topic of intense discussion.
ES cells apparently tolerate loss of both
de novo and maintenance DNA methyltransferases [
24,
25]. With just 0.6% methylation of CpG dinucleotides, Dnmt3a
−/− and Dnmt3b
−/− ES cells cannot initiate differentiation efficiently, but remain viable and pluripotent, as indicated by the presence of alkaline phosphatase and Oct4 expression [
24]. Similarly, a triple knockout ES cell line for Dnmt1, Dnmt3a, and Dnmt3b grew robustly and maintained its undifferentiated characteristics [
25]. These observations suggest that DNA methylation is not essential for ES cell pluripotency, but rather for ES cell differentiation. In addition, active promoters in murine ES cells are heavily methylated and 36% of genes with methylated promoters can still be expressed. Promoters bound by Nanog or Oct4 are examples of this trait [
26]. Thus, DNA methylation by itself does not suffice for gene repression in ES cells and pluripotency maintenance.
More recently, genome-wide DNA methylation analyses uncovered distinct and dynamic epigenetic profiles in stem cells as compared with differentiated cells [
27,
28]. For instance, DNA methylation is associated with the majority (87%) of repressed genes that do not overlap with bivalent chromatin domains in ES cells [
29]. Thus, DNA methylation constitutes a relevant repressive mechanism for genes not influenced by bivalent chromatin in ES cells.
DNA methylation is also necessary for the epigenetic silencing of key pluripotency transcription factors needed for ES cell differentiation (). Indeed, pluripotency-associated genes like
Nanog1 and
Zfp42/ are unmethylated and expressed in ES cells, while they are silenced and methylated in mouse fibroblasts [
29]. Furthermore, the DNA methyltransferases Dnmt3a and Dnmt3b target the
Oct3/4 and
Nanog promoters in differentiated ES cells [
30].
More recently a single-base resolution map of DNA methylation in human ES cells was generated [
31]. An unexpected result was the significant methylation of non-CpG-enriched DNA, with varied distributions of methyl marks on mCHG or mCHH (where H represents C, T or A). Moreover, non-CpG DNA methylation represents ~25% of the ES cell DNA methylation and is underrepresented in binding sites for Nanog, Sox2 and Oct4 transcription factors, enriched in exons, introns and 3′-untranslated regions [
31]. Importantly, the DNA methylation distribution in ES cells is different from that in differentiated cells, in which non-CpG methylation is lost, suggesting that non-CpG methylation may participate in cell differentiation and that it might be a signature for pluripotency [
31].
DNA methylation is linked to PcG-complexes-mediated repression. However, evidence suggests that this is not always the case. For example a genomic scale comparison of genes targeted by PcG complexes and those enriched on DNA methylation showed that both sets of genes were not strongly associated in ES cells [
29]. Thus, DNA methylation and PcG-mediated repression can act as independent silencing mechanisms. However, this is still in debate. In cancer cells, DNA methylation is linked to PcG components [
32]. For instance, EZH2 acts in concert with DNA methyltransferases. In contrast, other reports suggest that DNA methylation and PcG complexes act independently [
33,
34]. These results and others suggest that EZH2 is not the main means for DNA methylation recruitment in cancer cells [
35]. Indeed, the majority of the H3K27me3 occupied genes lack DNA methylation. Moreover, recent studies determined that targeting of EZH2 to a defined genomic site is sufficient for recruitment of Dnmt3a, but not
de novo DNA methylation [
35]. In conclusion, at this point the mechanisms targeting DNA methylation in undifferentiated cells are poorly understood. Identifying targets in which repression is associated with PcG-dependent or PcG-independent DNA methylation in ES cells would further our understanding of the function of different repressive chromatin configurations in establishing the pluripotency transcriptional network, as well as in determining cell lineages.
ATP-dependent chromatin remodeling complexes in embryonic stem cells
The ATP-dependent chromatin remodeling complexes are multiprotein complexes of variable compositions. Using energy from ATP hydrolysis, they relocate nucleosomes through sliding mechanisms and nucleosome eviction [
36], induce changes in nucleosomes conformation and favor the interchange of canonical histones by histone variants [
20,
37]. By these activities, chromatin-remodeling complexes contribute to gene expression activation or repression and label defined sectors of the genome through the incorporation of histone variants. ATP-dependent chromatin remodeling complexes are mainly grouped in the SWI/SNF, ISWI, CHD and INO80 families [
38].
In addition to DNA methylation and PcG-mediated regulation, the ATP-dependent chromatin remodeling complexes participate in regulating the ES cell chromatin structure (), self-renewal capacity and differentiation. In ES cells these complexes cooperate with pluripotency factors in gene expression regulation [
1]. A large
-scale RNA interference screen against regulatory factors and chromatin components relevant for ES cell maintenance identified Brg1, which is the ATPase of the SWI/SNF complex. Indeed Brg1 knockdown results in loss of the capacity of ES cells to self-renew [
39,
40]. Furthermore, Brg1 interacts and co-localizes with the pluripotency transcription factors Nanog, Oct4 and Sox2 at their target genes [
41,
42]. Interestingly, Brg1 binds to a significant number of lineage-associated genes that have bivalent histone marks in ES cells, suggesting that the repressive activity of Brg1 is relevant for cell fate determination [
42]. In support of this notion, Brg1 depletion impairs ectodermal and mesodermal determination [
43]. In addition, Baf250a or Baf250b, which are subunits of the SWI/SNF complex known as Baf (Brg1 associated factor), are also important for ES cell maintenance and differentiation [
44]. Several other remodeling complexes are necessary for stem cell pluripotency. For instance, Chd1 (chromodomain-helicase-DNA-binding protein 1), a component of the mammalian ISWI complex, maintains an open chromatin conformation and is required for pluripotency maintenance and induction [
45,
46].
The NURD (Mi-2/nucleosome remodeling and deacetylase) complex is associated with ATP-dependent nucleosome remodeling and histone deacetylation activities that mediate gene repression [
47]. The NURD component MBD3 (methyl-CpG-binding domain protein) is indispensable for silencing of pluripotency-associated factors, ES cell commitment into developmental lineages [
48] and embryo development [
49].
In summary, ATP-dependent chromatin remodeling enzymes are required for ES cell self-renewal, pluripotency and cell differentiation into particular lineages. Whether ATP-dependent chromatin remodelers perform hierarchical functions in the remodeling of ES cell chromatin would insight into the epigenetic control of pluripotency. Determining the genome-wide occupation and target genes of different ATP-dependent chromatin remodelers in ES cells and during induction and progression of differentiation should help resolve this issue.
Polycomb group of proteins in human embryonic stem cells
Transcriptional repression via the Polycomb repressor complexes (PRC) is important for maintaining the pluripotent state. PRCs are mostly conserved from
Drosophila to human [
50]. The PRC2 is recruited to genomic sites via interaction with DNA-binding factors (like YY1) [
51] and mainly catalyzes H3K27me3. This histone mark provides the recognition signal for PRC1 incorporation, resulting in induction of a repressive chromatin configuration that can be segregated through cell generations [
52]. In ES cells, the PRC2 complex occupies bivalent chromatin domains (). Thus, an important function of PRC2 is to keep cell differentiation regulators repressed to maintain pluripotency [
53]. At the same time genes repressed by PRC2 are marked by H3K4me3 and remain poised for activation upon differentiation induction ().
Accumulated evidence supports a dual function for PcG in ES cells. PcG proteins are required to maintain pluripotency and progenitor stem cells populations, in part, by epigenetically regulating key genes linked to the cell-cycle control and cell proliferation, such as
p16INK4a and
p19ARF [
54]. The resolution of bivalent histone marks upon cell differentiation induction implies that histones have to be demethylated either at K4 or K27 in a regulated manner. In this regard, JMJD3 and UTX have been identified as H3K27me2/3 demethylases that might counteract Polycomb-mediated epigenetic silencing and favor transcriptional activation of lineage-specific groups of genes [
55].
Similar to what happens upon loss or reduction of DNA methylation, lack of the polycomb members Ring1B or Eed in ES cells results in lineage-specific gene derepression [
56–58]. These transcriptional changes destabilize ES cells, but surprisingly, they do not affect their self-renewal properties. In addition, ES cells lacking members of either PRC1 or PRC2 can differentiate
in vitro. Similarly, Eed-deficient ES cells retain pluripotency, as they form teratomas in mice [
59]. Thus, members of the PcG of proteins appear to be dispensable for maintaining the ES cell state and for ES cell differentiation. Yet, these complexes contribute to establishing the global chromatin environment in ES cells, raising the possibility that PcG proteins act in concert with other epigenetic mechanisms in pluripotency regulation.
The PRC1 complex mono-ubiquitylates H2A at lysine 119 and induces gene repression [
60]. Despite the fact that PRC2-mediated H3K27me3 serves as docking site for PRC1, whether ubiquitylated H2A has a function in maintenance of bivalent domains, pluripotency maintenance or cell fate acquisition remains to be explored.
Another function of the PcG, which is poorly explored in the context of ES cells, is the formation of high-order structures through multiple long-range chromatin interactions or looping that occlude access of regulatory factors to their target sequences [
61].
In addition to PcG, the Trithorax (TrxG) group of proteins participates in the epigenetic regulation of ES cells. To some extent, this group of proteins antagonizes the activity of PcG proteins. TrxG forms a complex in which the histone methyltransferase MLL1 induces H3K4me3 methylation, which is an open chromatin mark. The activity of MLL1 complexes in transcriptional activation is complemented by the SWI/SNF or the NURF ATP-dependent chromatin remodeling complexes [
62].
A critical aspect of the action of the Polycomb and Trithorax (TrxG) groups of proteins is how such regulatory complexes are recruited to their genomic target regions in ES cells. Although there are no clear proposals, three possibilities have been discussed. The first one and the less documented, is the existence of highly specific DNA binding elements analogous to the
Drosophila Polycomb response elements (PREs), which might be recognized by PcG members. To our knowledge, only two mammalian PRE sequences have been identified [
63,
64], but whether PcG proteins occupy such sites in ES cells or not is an open question. The second proposal is that PcG is recruited via interaction with transcription factors and associated co-factors. One of the most studied PcG protein is YY1, the vertebrate homolog of
Drosophila PHO, which is a transcription factor that can recruit PcG complexes. However, this function of YY1 remains controversial, as its capabilities as recruiter cannot be generalized [
55]. Two other factors that interact with DNA have been associated with PRC2 recruitment in ES cells. JARID2, a histone demethylase, binds DNA through its C-terminal domain and co-occupies genomic regions with PRC2 complexes. Moreover, depletion of JARID2 negatively affects the interaction of PRC2 with its target genes. In one proposal, JARID2, which is catalytically inactive [
65], acts as enhancer or attenuator of the activity of the PcG complexes [
55]. The other factor, PCL2/MTF2 (Polycomb-like 2/metal response element-binding transcription factor 2), is the homolog of
Drosophila Polycomb-like (dPcl) and associates with the PRC2 complex in ES cells [
66,
67]. Like JARID2, PCL2 co-localizes with PRC2 in a subset of PcG target genes in ES cells and promotes H3K27me3, suggesting that PCL2 might function in regulating the pluripotency transcriptional network. Interestingly, the pluripotent transcription factors Oct4 and Nanog interact with the
Pcl2 gene promoter in ES cells and the Pcl2 relative abundance decreases upon differentiation [
67]. Finally, the third component recruiting PcG and TrxG proteins to their target sites along the genome are the non-coding RNAs [
68]. HOTAIR (Hox antisense Intergenic RNA), the most striking example, corresponds to a 2.2-kb non-coding RNA, which is transcribed from the
HOXC locus in the human chromosome 12 in fibroblasts and recruits PRC2 to the
HOXD locus on chromosome 2 via interaction with the PRC2 member SUZ12 [
69]. A recent report demonstrated that HOTAIR over-expression promotes cancer metastasis [
70]. Thus, HOTAIR or related non-coding RNAs might contribute to gene repression in ES cells by recruiting PcG proteins.
Two novel non-coding RNAs that can recruit the PcG complexes to specific locations have been recently described.
ANRIL, a 30–40 kb long non-coding RNA, expands over the
INK4a/ARF/INK4b locus [
71]. The association of
ANRIL with PcG of proteins is mediated by CBX7 (chromobox 7), a component of the PRC1 complex that binds
ANRIL. CBX7 and
ANRIL are expressed at elevated levels in prostate cancer tissues. The other non-coding RNAs are small RNAs of 50–200 nucleotides, which are transcribed from the 5′-non-transcribed region of Polycomb target genes in primary T cells and ES cells [
72]. A stem–loop structure formed by these short-RNAs interacts with the PRC2 complex through SUZ12, mediating repression of Polycomb target genes. Importantly, such short RNAs are depleted from polycomb target genes upon initiation of cell differentiation and transcriptional activation [
72].
MicroRNAs are essential for controling pluripotency
. Indeed, ES cells lacking proteins that mediate microRNA biogenesis exhibit defects in proliferation and differentiation [
73]. On the other hand epigenetic regulators target microRNAs in ES cells. Indeed, the H3K27 histone methyltransferase and PcG member EZH2 represses the expression of miR-214 in skeletal muscle and ES cells [
74]. Interestingly, once miR-214 is expressed a negative feedback loop is created; in which miR-214 targets the EZH2 3′-UTR, reducing EZH2 levels and promoting ES cell differentiation [
74].
In summary, the PcG complexes perform diverse functions over a varied number of target genes in ES cells. This underscores the requirement for better understanding how PcG complexes are recruited in a regulated manner to specific locations in the genome in order to unveil the epigenetic mechanisms of pluripotency, cell fate acquisition and cell differentiation.