Principles of stem cell biology and cancer: future applications and therapeutics. Edited by T. Regad, T. J. Sayers and R. C. Rees. John Wiley & Sons (2015)
Part I. Stem Cells
ESCs are derived from the inner cell mass (ICM) of the blastocyst. During the last 2 decades, a large effort has been dedicated to identifying the environmental cues that could allow researchers to freeze and preserve the two main properties of the ICM in a Petri dish; this would involve keeping cells in a self-renewing condition while retaining their ability to differentiate into virtually any cell type of the adult organism, a property known as pluripotency. In this way, ESCs have emerged as an unprecedentedly useful tool for the study of mechanisms of differentiation, as well as an endless source of differentiated tissues for regenerative medicine.
The identity of ESCs is maintained by a complex network of transcription factors, with Oct4, Sox2 and Nanog the most studied. These transcription factors maintain high levels of expression of self-renewal genes, at the same time as participating in repressing developmental regulators. The molecular basis of this duality is not fully understood. Self-renewal genes have very high rates of transcription, which seem to be supported by the presence of large enhancers called super-enhancers (which in some instances can span over 1 Mb of DNA) (Hnisz et al., 2013). Super-enhancers are fully loaded with pluripotency-related transcription factors like Oct4, Sox2 and Nanog, and strongly occupied by the Mediator complex, which facilitates the physical interaction between enhancers and promoters and the efficient recruitment of Pol II to support high rates of transcription (Whyte et al., 2013). They display high levels of H3K27 acetylation, which are dynamically regulated, since both histone acetyltransferases and histone deacetylases occupy these sites. Eventually, this dynamic regulation renders super-enhancer-regulated genes extremely vulnerable to perturbations of its components, providing a highly sensitive mechanism that will rapidly turn off the expression of these genes in response to differentiation signals.
The basis of pluripotency consists in critical developmental genes remaining silent but being ready for expression when differentiation signals arrive. Differentiated cells have lost this property, and therefore cannot change their identity in response to environmental signals. It has been proposed that a particular chromatin conformation is responsible for keeping developmental genes poised for activation in ESCs. This chromatin conformation consists in the simultaneous presence of both H3K27me3 and H3K4me3 at the regulatory regions of developmental genes and has been called ‘bivalent domains’ (Figure 2.1) (Bernstein et al., 2006). These antagonistic modifications are unlikely to occur in the same histone tail, but seem to co-exist asymmetrically in nucleosomes; that is, on opposite H3 tails (Voigt et al., 2012). It has been proposed that during differentiation, bivalent domains resolve into H3K4me3 only in genes that become activated and into H3K27me3 only in genes that become repressed (Bernstein et al., 2006).
Figure 2.1. Different complexes are involved in maintaining bivalent domains in ESCs. (A) Bivalent domain-containing genes in ESCs are CpG-rich and are hypomethylated or hydroxymethylated by Tet1. Hydroxymethylated CpG islands provide binding sites for the NURD complex through the MBD3 subunit. LSD1 is recruited as part of the NURD complex and counteracts the methylation of H3K4 mediated by MLL2. Different PRC1 complexes are recruited to these regions through the binding of Kdm2b to unmethylated CpG islands or the recognition of Ezh2-mediated H3K27me3 by CBX7. (B) Induction (red) or repression (green) of the expression of different chromatin factors and subunits contributes to the resolution of bivalent domains during differentiation. (C) Bivalent domains resolved after differentiation.
Different complexes mediate the modifications found at bivalent domains. Methylation of H3K27 is catalysed by the Polycomb group of proteins (PcG), which refers to two distinct complexes. Polycomb repressive complex 2 (PRC2) contains four core components, including the SET domain-containing EZH2 subunit capable of trimethylating H3K27. Polycomb repressive complex 1 (PRC1) is represented in mammals by several complexes with different subunit compositions. All PRC1 complexes contain the ubiquitin ligase Ring1b but they are recruited to chromatin through different mechanisms, depending on the presence of other subunits (Figure 2.1). For example, CBX-containing complexes are recruited through chromodomain-mediated methylated H3K27 interaction (Cao et al., 2002), while Kdm2b-containing complexes are recruited through Kdm2b-mediated recognition of nonmethylated CpG-rich areas (Farcas et al., 2012; He et al., 2013; Wu et al., 2013). Methylation of H3K4 in mammals can be catalyzed by at least six different methyltransferases, Set1A and Set1B and MLL1 to MLL4. MLL2 has been recently suggested to be responsible for the H3K4me3 found at bivalent promoters (Hu et al., 2013; Denissov et al., 2014).
Although the resolution of bivalent domains during differentiation seems clearly correlated with changes in gene expression at different stages of cellular commitment (Xie et al., 2013a), the role of bivalency in poising developmental genes remains controversial. Genes containing bivalent domains have also been located in differentiated cells, although in lower numbers (Mikkelsen et al., 2007). Treatment of mouse ESCs (mESCs) with ‘2i’ (inhibitors of Mek and GSK3 kinases) renders cells more naпve, with a lower expression of developmental regulators that correlates with reduced levels of H3K27me3 at their promoters, while Pol II appears sharply restrained at their TSSs (Marks et al., 2012). This argues against a major role of H3K27me3 in poising developmental genes and is coincident with the observation that, following the maternal – zygotic transition in zebrafish, a subset of developmental genes are kept repressed in the presence of H3K4me3 only (Vastenhouw et al., 2010). Additionally, mESCs can be derived from PRC2-deficient blastocysts and maintained in culture for many generations, although they show defects in differentiation (Pasini et al., 2007). All together, these data suggest that the presence of H3K27me3 at developmental genes might reflect a transitional state towards final resolution during differentiation. The role of H3K4me3 in poising genes also remains obscure, since mESCs null for MLL2 loose H3K4me3 at many bivalent genes, but this does not prevent their induction by retinoic acid, suggesting that this modification is not critical to priming genes for activation (Hu et al., 2013; Denissov et al., 2014).
The ultimate event that holds the key to pluripotency is the presence of Pol II restrained at the TSSs of relevant developmental regulators, ready to respond to differentiation signals and engage in productive rounds of transcription. Pol II stalled at bivalent gene promoters in ESCs is preferentially phosphorylated on Ser-5 residues (Stock et al., 2007), but its conformation appears to be different from that of the conventional paused Pol II found in differentiated cells, since it is confined to regions extremely close to the TSS (Min et al., 2011). This unique Pol II conformation has yet to be fully characterized, but it is usually referred to as ‘poised’ (Stock et al., 2007). Ablation of Ring1B, the PRC1 subunit that mediates ubiquitination of histone H2A, results in loss of ubiquitinated H2A at bivalent genes and de-repression of these genes without changes in H3K27me3 (Stock et al., 2007), suggesting that H2A ubiquitination is involved in restraining Pol II at these promoters. More recent data show that, in ESCs, the kinases Erk1/2 mediate the phosphorylation of Pol II at Ser-5, contributing to holding it back at the TSSs of developmental genes (Tee et al., 2014). Further analysis will be needed to elucidate how this very particular Pol II conformation is achieved at critical promoters in ESCs.
The epigenetic events that regulate the expression of developmental genes in ESCs appear to be linked to genetic features at the regulatory regions of such genes. Most of these genes contain CpG-rich regions, called ‘islands’, next to their TSSs. These CpG islands remain hypomethylated and facilitate the recruitment of both PRC1 and PRC2 (Wu et al., 2011, 2013). Occupancy by Tet1, which is highly expressed in ESCs, may both protect these sites against DNA methylation and mediate their hydroxymethylation (Wu et al., 2011). Hydroxythylated CpG islands are likely to provide binding sites for MBD3, a subunit of the NuRD repressive complex (Yildirim et al., 2011). The histone demethylase LSD1 is also recruited to these genes as part of this complex, where it finely regulates the levels of H3K4me2/3 (Adamo et al., 2011). Importantly, the recruitment of both activating and repressing complexes to these domains may keep them in a dynamic equilibrium that allows for rapid response during differentiation, merely by tipping the balance of these complexes (Figure 2.1).