Polycomb Repressive Complex 2: A Dimmer Switch of Gene Regulation in Calvarial Bone Development

Abstract

Purpose of Review

Epigenetic regulation is a distinct mechanism of gene regulation that functions by modulating chromatin structure and accessibility. Polycomb Repressive Complex 2 (PRC2) is a conserved chromatin regulator that is required in the developing embryo to control the expression of key developmental genes. An emerging feature of PRC2 is that it not only allows for binary ON/OFF states of gene expression, but can also modulate gene expression in feed-forward loops to change the outcome of gene-regulatory networks. This striking feature of epigenetic modulation has improved our understanding of musculoskeletal development.

Recent findings

Recent advances in mouse embryos unravel a range of phenotypes that demonstrate the tissue-specific, temporal, and spatial role of PRC2 during organogenesis and cell fate decisions in vivo. Here, we take a detailed view of how PRC2 functions during the development of calvarial bone and skin.

Summary

Based on the emerging evidence, we propose that PRC2 serves as a “dimmer switch” to modulate gene expression of target genes by altering the expression of activators and inhibitors. This review highlights the findings from contemporary research, that allow us to investigate the unique developmental potential of intramembranous calvarial bones.

Introduction

In embryonic development, multipotent stem cells require specific developmental cues to drive cell fate decisions, which propel these progenitor cells towards their terminal fates. Fundamental to this process are signaling molecules and transcription factors, which affect cell fate and differentiation by altering gene expression at the transcriptional level. Gene expression of signaling molecules and transcription factors can be modified in embryonic cells by epigenetic modification. Epigenetic modification is associated with various modifications such as acetylation, methylation, phosphorylation, or ubiquitylation of histone tails. Modification of DNA directly, or DNA-associated proteins initiates chromatin remodeling impacting gene expression. One significant group of epigenetic regulators is the Polycomb group (PcG) of proteins. Polycomb Repressive Complex 1 (PRC1) and Polycomb Repressive Complex 2 (PRC2) are two members of the PcG that play roles in monoubiquitylation or mono-/di-/tri-methylation of histones respectively[1,2]. Emerging evidence suggests that PRC2 plays a substantial role in regulating cell fate selection and differentiation in a temporally-restricted and cell-specific manner in vivo.

This review seeks to explore loss of function data from mouse embryonic calvarial bone and epidermis of the skin studies to introduce a framework for understanding the mechanisms and functions of PRC2-based regulation. The state of transcription factor expression is often viewed as a binary off or on in genetic programs (Figure 2A). However, recent data demonstrates that PRC2, and other similar epigenetic modifiers, can be viewed as “dimmer switches” or “rheostat controls” of gene expression (Figure 2C). Instead of binary activation or silencing of genes, in vivo PRC2 mutants produce graded effects in multiple tissues that reveal entirely new phenotypes. Epigenetic modifiers can therefore be manipulated in genetic experiments in order to tease out new and redundant functions of GRN components that were not revealed by traditional loss and gain of function approaches.

Figure 2. PRC2-mediated feed-forward control of transcription factors and signaling pathway components can serve as a “dimmer switch” of gene expression changes.

(A) Transcription factors serve as ON/OFF switches of target gene expression in a binary fashion. (B) PRC2 based gene regulation can modulate the expression of transcription activators and repressors that can subsequently modulate target genes in a feed-forward manner. (C) In conditional knockouts of PRC2 core components, simultaneous change in expression of numerous PRC2 target genes can serve as a “dimmer switch” of gene expression in GRN networks.

PRC2 composition and function

PRC2 is composed of four distinct protein subunits: EZH2, SUZ12, EED, and RbAp46/48 (RBBP4/7) [1,3]. EZH2 is the catalytic component of the complex, and is directly responsible for the methyltransferase activity that establishes a methylation mark on the 27th lysine of the 3rd histone (H3K27me3) [1,4,5]. SUZ12, and EED are both required indirectly for the function of this catalytic domain, while RbAP46/48 is required for the association of PRC2 with the histone tail during methylation [1,6–8]. Mutations in the human EZH2 gene lead to Weaver syndrome (OMIM ID 277590), which is characterized by craniofacial defects including domed head and smaller mandible [9,10]. PRC2-mediated gene repression in collaboration with other histone modifications provides positional identity of facial structures by transcriptionally poising genes required for craniofacial development and cell fate decisions [11–13].

Chromatin immunoprecipitation followed by high throughput genome-wide sequencing (ChIP-seq) of regions occupied by H3K27me3 and PRC2 core components can be used to identify PRC2 targets in cells. H3K27me3 is broadly deposited across the gene bodies and flanking regions of genes, which initiates chromatin compaction and transcriptional repression [14]. By combining global mRNA expression and H3K27me3 ChIP-seq data, studies show that PRC2-mediated H3K27me3 leads to transcriptional repression and plays a role in silencing of developmental genes [3,15–17] (Figure 1). Studies suggest unique functions for the other histone H3 lysine 27 modifications, H3K27me1 and H3K27me2. These marks may play a role in the “priming” of genomic locations for the catalytic conversion to H3K27me3 [18,19]. However, most PRC2 loss of function studies focus on H3K27me3 occupancy, which is the key histone modification for PRC2-mediated transcriptional repression.

Figure 1. PRC2 core and accessory proteins and downstream effects.

PRC2 consists of 4 main subunits, EZH2, SUZ12, EED, and RbAp46/48. Binding of accessory subunits is shown in the upper frame. PRC2 binds and methylates the 27th lysine on histone H3 resulting in chromatin compaction. Key transcription factors and cell signaling pathway components are potential targets of repression downstream of PRC2 binding and H3K27me3 modification.

In addition to the four main subunits that comprise it, PRC2 interacts directly with accessory proteins that have important functions in establishing the histone methylation mark. PRC2 does not encode DNA binding domains in any of its four core subunits in mammalian cells. Therefore, PRC2 must interact with accessory proteins JARID2 and AEBP2 in order to target the core subunits to specific genomic loci, and associate directly with histone H3 [20,21]. Additionally, accessory proteins such as MTF2, and PCL1/2/3--members of the polycomb-like (Pcl) family of proteins--have been found to play a role in maximizing the efficiency of trimethylation at H3K27, while also helping to regulate the binding of PRC2 to histones [22–24]. Overall, the emerging picture of PRC2 recruitment is that the N-terminus of SUZ12 interacts with these non-core subunits to mediate the histone binding necessary for establishing methylation [25–27]. Other PRC2 accessory proteins EPOP and PALI have been implicated in the shaping of specific H3K27 methylation patterns and/or influencing the PRC2 catalytic rate [28]. For these two subunits, as well as the other non-core subunits, it has been difficult to parse out their unique functions because of their functional redundancy throughout the PRC2 binding network. Deletion of PRC2 components and accessory proteins lead to subtle phenotypes, which demonstrate their functional redundancy [29]. Future research is needed to cement both the unique and overlapping roles of these accessory proteins in order to establish a clearer picture of PRC2 recruitment.

In mammals, a homologue of EZH2, EZH1, has also been shown to be associated with the PRC2 complex and possesses methyltransferase activity (Shen et al., 2008). However, Ezh1 null mutant mice exhibit no phenotype and are viable, suggesting that EZH2 is the primary methyltransferase during development [30]. Thus, in vivo mouse studies investigating PRC2 have focused on Ezh2 mutants and will be the focus of this review.

PRC2 controls gene expression temporally during development

As previously stated, mutations in the human EZH2 gene lead to Weaver syndrome highlighting the important role of PRC2 during embryonic development [9,10]. Functional studies of PRC2 reveal key roles in establishing cell fate and directing differentiation in stem cells in vitro and during embryonic development [29]. In vivo studies, demonstrate that PRC2 has temporally restricted roles in calvarial bone and skin development.

Mammalian calvaria is composed of eight bones that are derived from the cranial neural crest cells (CNCC) and the paraxial mesoderm (PM). The frontal bone progenitors originate from the CNCC and the parietal bone progenitors arise from the PM. In the mouse, the frontal and parietal bone form as a foci in the supraorbital arch region above the eye beneath the ectoderm by embryonic day (E) 10.5 and express early calvarial bone cell fate markers between E11.5–13.5. The calvarial bone progenitors begin to differentiate from E13.5 in a baso-apical direction and expand apically to cover the brain (reviewed in [31, 36]). In comparison to mouse mutants of other epigenetic regulators, Ezh2 loss of function mutants display the most severe calvarial bone phenotypes [31] (Figure 3). In mice, temporally restricted deletion of Ezh2 can result in a partial to complete loss of multiple skull bones. Schwarz et al., used Wnt1Cre to delete Ezh2 in the mouse pre-migratory CNCC by E8.5 and observed efficient depletion of H3K27me3 by E9.5 and complete loss of mineralized CNCC-derived frontal bones at E16.5 [32]. Conversely, deletion of Ezh2 in both the CNCC and the PM at E9.5 with Dermo1Cre or E10.5 using En1Cre results in comparable skull bone phenotype to the control [33,34]. In addition, loss of Ezh2 in the CNCC and PM after E9.5 using Dermo1Cre does not lead to depletion of H3K27me3 modification [33]. Ferguson et al., used an inducible Cre line, PDGFRα Cre-ER to initiate deletion of Ezh2 after E8.5 (E8.5CM^Ezh2) broadly in the cranial mesenchyme containing calvarial bone progenitors from CNCC and the PM. This mutant had diminished frontal bone and nearly absent parietal bones [35]. When Ferguson et al., initiated deletion of Ezh2 after E9.5 with PDGFRα Cre-ER (E9.5CM^Ezh2), the mutants had efficient loss of bulk H3K27me3 in the CM that was accompanied by premature fusion of the coronal suture between the frontal and parietal bones [35]. The specific cause of the dramatic phenotypic differences between each conditional Ezh2 mutant is not well understood. One possibility is the presence of stable H3K27me3 in calvarial bone progenitors by E9.5 may be still maintained at specific loci after deletion of Ezh2. Alternatively, EZH2 function is required preceding the induction of the calvarial bone initiation program of transcription factors such as Msx2, Runx2, and Osx/Sp7 [31,36,37] and loss of Ezh2 after the bone initiation program may have a different function. The conditional Ezh2 mutants support its developmental stage-specific role in affecting the calvarial bone lineage [32,35,38]. These recent data highlight the temporal role of Ezh2 and H3K27me3 function in calvarial bone development.

Figure 3: Summary of calvarial bone phenotypes in conditional Ezh2 mutants.

(A) Wnt1Cre-mediated deletion of Ezh2 in the pre-migratory CNCC before E8.5 leads to a complete loss of CNCC-derived craniofacial bones (blue). (B) Deletion of Ezh2 in the CNCC and mesoderm-derived cranial mesenchyme with PDGFRαCre-ER between E8.5–9.5 leads to diminished frontal bone and near loss of parietal bone. (C) Deletion of Ezh2 in the PM after E9.5 leads to craniosynostosis of the coronal suture between frontal, parietal bones (yellow). (D) Dermo1Cre-mediated deletion of Ezh2 with in the CNCC and PM after E10.0 results in normal skull bone phenotype. Modified from Ferguson et al., 2018 [31].

A transcriptional mechanism by which PRC2 regulates calvarial bone development in the CNCC is repression of non-osteogenic differentiation programs in the calvarial mesenchyme that may include anti-osteogenic factors such as Hox genes and transcription factor Hand2 [35,39–44]. The Hox cluster and Hand2 genomic loci are significantly enriched for the broad blanket pattern of H3K27me3 occupancy in cranial mesenchyme in vivo [35]. In wild-type mice, the cranial mesenchyme region lacks expression of Hox genes, which are known to repress of Runx2, the earliest calvarial bone lineage marker in vivo [40,45]. Consistently, in the queried conditional craniofacial Ezh2 mutants, multiple Hox genes are upregulated in both the cranial and facial mesenchyme cells in vivo [32,35,38]. Based on these data, deletion of Ezh2 prior to E10.0 enables up-regulation of anti-osteogenic factors. Interestingly, E8.5CM^Ezh2 mutants have dramatically elevated expression of Hoxc8, Hoxa2, and Hand2 in the cranial mesenchyme which are also responsive to Retinoic Acid (RA) signaling levels [35,39,46]. Small molecule antagonism of RA receptors, rescues the frontal and parietal bones in the E8.5CM^Ezh2 and downregulates the Hand2 mRNA and HOXC8 protein expression levels [35]. Thus, Ezh2 mutants highlight a level of complexity by which epigenetics may function during calvarial bone development. PRC2 has been associated with the simultaneous regulation of multiple signaling factors and transcription factors in a spatial and temporal manner [32,35,38,47] (Figure 2). Further genetic studies utilizing tissue-restricted and single cell genomic approaches will provide new insights into the complex dynamics of PRC2 in different calvarial bones.

In a different tissue, studies in embryonic skin also complement the emerging roles of PRC2 in calvarial bone development. In the epidermis, the basal cell keratinocytes exit the cell cycle and differentiate into various suprabasal layers in late fetal stages. In embryonic mouse epidermis, EZH2 expression and the H3K27me3 mark are highly enriched in basal cell keratinocytes at mid-gestation and become diminished in the late fetal stage [48]. In addition, EZH2 expression also diminishes as cells differentiate into the suprabasal layers [48]. This expression pattern suggests a temporally restricted role of PRC2 in basal keratinocytes during embryonic epidermal differentiation. Functional studies of Ezh2 in embryonic basal keratinocytes demonstrate the importance of the PRC2 in modulating gene expression, and introduce the idea that PRC2 can help regulate the terminal epidermal differentiation program in a step-wise fashion [48]. Conditional deletion of Ezh2 leads to derepression of its well-known target, Ink4b/Arf/Ink4a locus [49]. In a feed-forward manner, INK4A/4B represses the transcription of target genes such as cell cycle regulators leading to a decrease in proliferation of keratinocytes and in other cell types [48,49]. Importantly, Ezh2 deletion in embryonic basal keratinocytes leads to loss of H3K27me3 occupancy allowing for AP1 transcription factor to bind and transcriptionally activate the epidermal differentiation complex genes leading to premature differentiation of the epidermis [48]. Thus, in normal development as PRC2 diminishes over time in basal keratinocytes and differentiating suprabasal cells, H3K27me3 occupancy is depleted and genes of the epidermal differentiation complex become transcriptionally active. However, future studies deleting Ezh2 in a temporally controlled manner are required to conclusively demonstrate the temporal role in epidermal differentiation.

PRC2 controls gene expression in cell-context specific manner.

GRNs varies from tissue to tissue, and cell type to cell type, in a similar manner to the way that gene expression varies across temporal windows. Recent in vivo, loss of function work in mouse models has provided insights into the cell population specific control of PRC2 during embryonic development.

The deletion of Ezh2 in different mouse calvarial bone primordia leads to dramatically different phenotypes ranging from loss of calvarial bone to craniosynostosis of the coronal suture [32,35,38] (Figure 3). One possibility is that multiple histone modifications have been shown to work in tandem to coordinate stem cell identity and position [12]. The variations in the other histone H3 methylation modifications between the CNCC and PM could account for the differences in phenotype [12]. A second possibility is inherent developmental and GRN differences between the CNCC- and the PM-derived mesenchyme [37]. For instance, the posterior skull bones develop at a later developmental stage than the anterior bones as seen by the delay in the onset of the bone differentiation program initiated by Msx2 and expression of different signaling molecules and transcription factors [37,50]. A third possibility is the level and effect of derepression of anti-osteogenic factors could differentially impact various calvarial bone primordia (see above). Taken together, results with tissue specific conditional deletions of Ezh2 in vivo reveal a cell-context specific role for PRC2 and highlights that epigenetic regulation may be targets to specific GRNs [32,35].

Conditional ablation of PRC2 core components in the mouse basal keratinocytes give rise to similar phenotypes that are unique by cell context. For instance, phenotypes include a) premature differentiation and acquisition of a functional epidermal barrier, b) a change in cell fate toward Merkel cells, and c) defective postnatal hair follicle development [51]. In hair follicles, it has been demonstrated that defective development is partially due to upregulation of the Ink4b/Arf/Ink4a locus [51]. The elevated number of Merkel cells is correlated with an upregulation of the Merkel cell signature genes Isl1 and Sox2 [51]. The three distinct phenotypes that arise as a result of PRC2 knockout in basal keratinocytes, the epidermal progenitor population that gives rise to these three distinct lineages, accentuate the drastically different outcomes that loss of PRC2 can induce in different developmental lineages. These results from PRC2 loss of function experiments in mouse skin corroborate findings from mouse calvarial bone, and further highlight the cell type-specific role for PRC2 in vivo.

PRC2 controls gene expression through crosstalk with developmental signaling pathways

Recently, various studies demonstrate that PRC2 regulates--or is regulated by--multiple developmental signaling pathways including Wnt/β-catenin signaling and RA signaling, among others (Table1). Emerging in vitro data suggest that PRC2 has extensive interaction with the Wnt/β-catenin pathway, a well-studied cell signaling pathway known to regulate cell fate determination and cell proliferation during development. Some of the ways include the following: 1. PRC2 can regulate components of the Wnt/β-catenin signaling pathway, and vice-versa [47,52–54]. 2. β-catenin and PRC2 can cooperate with one another to enhance either Wnt signaling or PRC2 activity [55–58]. 3. β-catenin can physically interact with PRC2 components [55,56,58,59]; 4. both PRC2 and Wnt signaling are required for cell fate selection and commitment [3,60–62] (Table1). Consistently, in vivo experiments that genetically knock out Mtf2, the accessory protein necessary for the binding of PRC2 to histones, show that the PRC2-Mtf2 complex epigenetically represses Wnt signaling during vertebrate hematopoiesis [63]. Additionally, PRC2 effects can be influenced by MAPK/ERK signaling. Erk2 has been shown to bind to specific DNA sequence motifs typically accessed by PRC2. Inhibiting Erk1/2 binding leads to decreased occupancy of PRC2 and poised RNA Polymerase II at Erk2-PRC2 targeted developmental genes, indicating reversal of PRC2 gene repression [64].

Table 1.

Summary of PRC2 interactions with signaling pathways.

Wnt/β-catenin Signaling		RA Signaling		MAPK/ERK Signaling
Interaction	Citation	Interaction	Citation	Interaction	Citation
PRC2 Regulates components of Wnt/β-catenin signaling and vice-versa	Wang et al., 2010; Zemke et al., 2015; Mirzamohammadi et al., 2016; Yi et al., 2016	At-RA exposure to wild-type mice at E10.0 phenocopies the craniofacial defect seen in E8.5-CMEzh2	Jiang et al., 2002; Maclean et al., 2009; Ferguson et al., 2018	Inhibition of ERK1/2 binding can reverse PRC2 gene repression	Tee et al., 2014
PRC2 physically interacts with Wnt signaling components	Shi et al. 2007; Li et al., 2009; Jung et al., 2013; Hoffmeyer et al., 2017	RA can recruit PRC2 components and H3K27me3 modification to a Retinoic Acid Responsive Element (RARE) near the promoter of Fgf8 and HoxB1 leading to transcriptional repression	Kumar and Duester, 2014
PRC2 Cooperates with β-catenin to enhance Wnt signaling and PRC2 function	Shi et al., 2007; Jung et al., 2013; Kumar and Lassar, 2014; Hoffmeyer et al., 2017	Small molecule inhibition of RA signaling can rescue calvarial bone development in E8.5-CMEzh2	Ferguson et al., 2018
Both PRC2 and Wnt Signaling are required for cell fate selection	Lee et al., 2006; Sparmann and van Lohuizen 2006; Asp et al., 2011; Margueron and Reinberg, 2011	RA signaling and PRC2 function in an incoherent feed-forward loop to inhibit anti-osteogenic factors in calvarial bones.	Ferguson et al., 2018

PRC2 has also been tied to regulation of the RA signaling pathway. Recent in vitro data demonstrate that KDM5B, a histone lysine demethylase, works together with PRC2 to regulate the RA signaling cascade in a cooperative and orchestrated fashion [65]. PRC2 is an important epigenetic regulator in the head and face and requires RA for recruitment to specific genes [66]. Furthermore, in vivo evidence shows that RA signaling and EZH2 can function synergistically to guide calvarial bone progenitor commitment by regulating the suppression of anti-osteogenic factors [35]. Previous studies demonstrated a wide range of craniofacial phenotypes that result from positive and negative modulation of RA signaling [67,68]. Transient at-RA exposure to wild-type mice at E10.0 phenocopies the diminished frontal bone and absence of parietal bone phenotype in E8.5-CM^Ezh2 [31,68,69]. Interestingly, activation of RA-signaling regulates gene expression of PRC2 core components, Ezh2 and Suz12 in calvarial bone progenitors [35]. Ferguson et al., showed that E8.5-CM^Ezh2 mutant phenotype can be rescued by small molecule antagonism of the retinoic acid receptor (RAR), providing evidence that EZH2 in the calvarial bone progenitors is required for repressing the expression of anti-osteogenic factors that are poised to respond to RA signaling. Thus, a balance of RA-signaling and EZH2 expression is required to maintain calvarial bone lineage commitment. Consistent with this putative hypothesis, RA can recruit PRC2 components and H3K27me3 modification to a Retinoic Acid Responsive Element (RARE) near the promoter of Fgf8 and Hoxb1 leading to transcriptional repression [66]. In the context of calvarial bone lineage commitment, RA signaling may function with EZH2 protein in an incoherent feed-forward loop, where the two arms of the loop can function in opposition [70] (Figure 4). In one arm, RA recruits PRC2 to anti-osteogenic factors, resulting in H3K27me3 enrichment and transcriptional repression (Figure 4). In the absence of EZH2, RA can function as a transcriptional activator and induce the expression of anti-osteogenic factors on the other arm (Figure 4). Particularly, such incoherent feed-forward loops can generate a pulse effect which can be used to explain the transient and temporal role of RA signaling and EZH2 in the commitment of calvarial bone progenitors [70]. Thus, deletion of Ezh2 has revealed a new GRN and its function was not gleaned from traditional gain and loss of function studies of RA signaling.

Figure 4. Summary of the regulation of skull bone formation by EZH2 and at-RA.

(A, B) EZH2 maintains a balance of activation and suppression of anti-osteogenic genes. (C) Following administration of at-RA, the balance is shifted towards activation of anti-osteogenic genes. (D) Proposed model by which, in the absence of Ezh2, inhibition of RA-signaling prevents the activation of the anti-osteogenic genes restoring OSX expression and skull bone formation. Modified from Ferguson et al., 2018 [35].

Taken together, these studies underscore the important, and complex role that PRC2 plays in the coordination of developmental events during embryogenesis and organogenesis, especially by interplay with canonical signaling pathways and transcription factors.

New Concept:

From the studies detailed above, it is clear that epigenetic regulation by PRC2 is upstream of key GRNs that control cell differentiation, and guide the development of tissues. In order to use genetic engineering to recreate tissues, it is important that researchers understand and develop tools to manipulate the complex GRNs that govern the development of those tissues. Previous genetic research attempting to elucidate these GRNs has largely been focused on using loss-of-function or gain-of-function approaches to establish the roles for the genes and molecules that make up these pathways. These traditional approaches have elucidated the function of many GRN components, but they are not specialized enough to serve as therapeutic nodes of intervention.

Epigenetic modifiers like PRC2 provide researchers a new avenue for altering genetic programs of cell differentiation and cell-cell signaling to better understand the function and regulation of GRNs. PRC2 acts like a dimmer switch or rheostat of gene expression. The subtle phenotypes seen in conditional PRC2 knockouts are evidence of this phenomenon. To further the analogy, transcription factor or signaling molecule knockout is similar to a light switch. When deleting activators, you completely silence gene expression, and when deleting repressors, you completely activate gene expression (Figure 2A). When deleting epigenetic modifiers, instead of knocking out the function of one transcription factor, the levels of expression of multiple transcriptional activators and repressors are changed, which can create a dimmer effect (Figure 2B,C). In order to engineer complex tissues, it is vital to define the functions of key nodes in the vast GRNs in vivo. Using epigenetic modifiers as a tool to establish new functions for developmental genes is an additional way to demonstrate the function of these GRNs in developing tissues. Thus, PRC2 mutants provide new insights into the dynamic formation of calvarial bones highlighting the spatial and temporal differences in the successful progression of osteogenesis during intramembranous bone formation.

Significance.

In order to efficiently and effectively apply orthopedic stem cell therapies, an in depth understanding of specific genetic regulatory networks (GRN) in distinct musculoskeletal cell populations is essential. Epigenetic regulation is a “tool” that can be used to modulate regulation of GRN in order to alter cell differentiation programs. Manipulating epigenetic mechanisms has the potential to reveal new functions of GRN components that were not discernible by traditional loss and gain of function approaches. Thus, the new knowledge gained from investigating epigenetic regulation of developing calvarial bone program and other tissues’ programs will inform tissue engineering, and improve outcomes of genetic therapies in orthopedics.