Evolutionary adaptation of the Polycomb repressive complex 2

PRC2 complex composition in Drosophila and mammals

The early characterization of PRC2 in Drosophila has demonstrated that PRC2 is composed of a core (Fig. 1A), consisting of the histone methyltransferase E(z) and the proteins Su(z)12, Esc/Escl, and Caf1-55 [8]. The general composition of the PRC2 core is basically identical in mammals [7], with two paralogs of the methyltransferase, EZH1 and EZH2, and the proteins SUZ12, EED and either RBBP4 or RBBP7 (Fig. 1B). This catalytic core is sufficient to methylate histones in vitro [7, 8]. However, the PRC2 core does not act alone but in concert with additional proteins, which extend and customize the activity of the core complex. In Drosophila, the Polycomb-like protein (Pcl) is essential for efficient H3K27 methylation [19]. The core can further be associated with Jarid2 (also named Jumonji) and Jing [3]. In mammals, several independent studies [20,21,22,23,24,25] indicated that PRC2 typically contains either one of the Pcl homologs (PHF1, MTF2, PHF19), or the homologs of Jarid2 and Jing, JARID2 and AEBP2, demonstrating the establishment of two major PRC2 subcomplexes, commonly referred to as PRC2.1 and PRC2.2, respectively (Fig. 1B). Although it has not yet been conclusively confirmed that Drosophila PRC2 forms similarly distinct subcomplexes, it is likely that the PRC2 core preferentially interacts with either Pcl or Jarid2, which may explain their nonredundant functions [19, 26].

Fig. 1figure 1

Overview of the main PRC2 members and their competition in Drosophila (A) and mammals (B), as well as their evolutionary occurrences (C). Mammalian complexes with one of the Polycomb-like proteins (PHF1, MTF2, or PHF19) are named PRC2.1, while complexes including AEBP2 and JARID2 are referred to as PRC2.2. Related complex members in Drosophila and mammals are indicated by similarly colored groups. Members of the catalytic core are illustrated in green, whereas Pcl and the respective PRC2.1 subcomplex members are indicated in blue. Mammalian PRC2.1 can have additional proteins such as EPOP, SKIDA1, and PALI1/2 (yellow), which compete against each other, allowing multiple PRC2.1 complex compositions (see also Fig. 6). Jing, Jarid2, and the respective mammalian PRC2.2 subcomplex members AEBP2 and JARID2 are colored red. The mammalian-specific protein EZHIP (purple) interacts with PRC2 in a subcomplex-independent manner

Recent work demonstrated that mammalian PRC2.1 not only contains a Pcl homolog but can also be associated with additional proteins, for which no homologs are known in Drosophila (Fig. 1B, C). At least three proteins are associated mainly with the PRC2.1 subcomplex: EPOP, PALI1, and PALI2 [21, 22, 27] (Fig. 1B). The protein SKIDA1 (C10orf140), which shares the PRC2-binding domain with EPOP [28], was also found to be associated with PRC2 [20, 29] and is likely to associate specifically with PRC2.1 (Fig. 1B). PRC2.1 contains at least one of two methyltransferases (EZH1 or EZH2), and one of the three Polycomb-like proteins (PHF1, MTF2, or PHF19), resulting in already six possible combinations. It has also been shown that in PRC2.1, EPOP and PALI1 compete for binding [20, 21, 30]. The additional presence of one of either EPOP, SKIDA1, PALI1, or PALI2 therefore generates twenty-four further putative combinations. These are all referred to as “PRC2.1”. Thus, the term “PRC2.1” refers to a large family of similar, yet distinct, complexes in which the PCL proteins are common to all members, in combination with a wide range of signature subunits. Notably, in the absence of AEBP2, JARID2 and MTF2 were found to be present within the same complex [23]. This suggests that the distinction between PRC2.1 and PRC2.2 is not strict, and that hybrid PRC2 complexes can form when members of PRC2.1 and PRC2.2 target nonoverlapping interaction sites [31]. In addition to the proteins that preferentially are either part of PRC2.1 or PRC2.2, the PRC2-associated protein EZHIP directly interacts with EZH2 and influences the function of PRC2 in a subcomplex-independent manner [32,33,34,35] (Fig. 1B).

In mouse embryonic stem cells (mESCs) the members of PRC2.1 and PRC2.2 colocalize at the majority of PRC2 binding sites [36, 37]. In these cells, only the simultaneous removal of the associated proteins of both PRC2.1 and PRC2.2 completely abrogates PRC2 core chromatin recruitment, while single deletions merely lead to reduced recruitment [36, 37]. These results demonstrate that the efficient recruitment of PRC2 requires the associated factors of both PRC2 subcomplexes, suggesting a certain redundancy and synergism between PRC2.1 and PRC2.2. Nonetheless, recent work demonstrated that PRC2.1- and PRC2.2-associated factors have unique roles in chromatin binding and mouse ES cell differentiation [38,39,40,41]. In mESCs, PRC2.1 chromatin recruitment is mainly mediated by the DNA binding function of MTF2, while PRC2.2 chromatin recruitment is mediated by the affinity of JARID2 for the PRC1-associated histone mark H2AK119ub [38]. These alternative binding modes may explain why shifting the balance of PRC2 towards PRC2.1 using specific point mutations within SUZ12 leads to a strongly increased affinity of PRC2 for chromatin [40]. Consistent with this hypothesis, the removal of MTF2 or JARID2 in mESCs leads to the derepression of a distinct set of genes upon differentiation and consequently to altered lineage choices. MTF2 is associated with genes that remain repressed during differentiation [39] and its loss leads to enhanced commitment to all germ layers [41]. Conversely, JARID2-sensitive genes were active in mESCs and became newly repressed upon differentiation [39]. JARID2-deleted cells possess a more selective differentiation process towards early differentiated precursors and have a reduced efficiency towards mesendodermal lineages [41]. Thus, this recent work suggests that each of the PRC2 subcomplexes has a unique function, which cannot be reproduced by alternative complex assemblies. Further work is therefore required to unravel the specific functions of the PRC2-associated factors.

Evolution of the PRC2 core

As described above, some but not all components of PRC2 are represented in both Drosophila and mammals (Fig. 1). As expected, the proteins that make up the PRC2 catalytic core are generally well conserved during evolution. The structural proteins RBBP4/7 are more than 80% identical to the Drosophila homolog Caf1-55, suggesting that these proteins have evolved little from the common ancestor protein (Fig. 2). In contrast, the other core proteins show a lower sequence identity of approximately 50% for EZH1/2 and EED and less than 30% for SUZ12. This relatively low sequence identity demonstrates that these genes have considerably diverged during evolution. Most pronounced is the development of the methyltransferase E(z). The ancestral E(z) gene was duplicated during vertebrate evolution [42], leading to two genes in mammals, EZH1 and EZH2 (Fig. 1C). Both mammalian enzymes have retained H3K27me3 methylation activity [43], but the protein sequence of EZH2 is more closely related to the Drosophila homolog E(z). A large body of work suggests that EZH2 is the main PRC2 methyltransferase in mammals and is responsible for the deposition of most H3K27me3 marks [43, 44]. Consistently, EZH2 knockout leads to a strong reduction in H3K27me3 and is embryonically lethal [6, 43]. The absence of EZH2 also leads to a reduction in H3K27me2 and H3K27me1, but to a lesser extent than the reduction of H3K27me3, suggesting that the remaining EZH1 can compensate for the loss of EZH2 and can deposit H3K27me1 and H3K27me2 [44]. Nonetheless, EZH1 is generally considered less relevant for the function of PRC2 since it is not embryonically lethal [6, 42]. The establishment of two distinct methyltransferases within PRC2 indicates a special EZH1-related function of PRC2. Interestingly, biochemical studies suggest that PRC2 containing EZH1 can compact chromatin in a methyltransferase-independent manner [43, 45]. Additionally, recent structural work demonstrated that the EZH1-PRC2 has a substantially different conformation than EZH2-PRC2 and that the previously observed chromatin compaction is mainly facilitated by EZH1-PRC2 dimers [46]. Although EZH2-PRC2 can also establish dimeric states, this conformation appears to have other functions, such as enhanced chromatin and RNA binding [25, 47]. Thus, these preliminary studies suggest that EZH1-containing PRC2 may have functions different from EZH2-containing PRC2, which might explain, for example, the EZH1-specific biological role in hematopoiesis [48, 49]. More work will be required to better understand the mechanistic and biological function of EZH1-containing PRC2.

Fig. 2figure 2

Comparison of PRC2 members that exist in both Drosophila and humans. On the left side, the regions of the Drosophila proteins that are also present in human proteins are colored. On the right side, the identity and conservation of the retained regions are shown. Annotated domains are highlighted. BAM: β-addition motif; SANT: SANT (Swi3, Ada2, N-CoR, and TFIIIB) domain; CXC: cysteine-rich domain; SET: SET (Su(var)3–9, Enhancer-of-zeste and Trithorax) domain; WDB: WD40-repeat binding; ZnB: zinc finger binding helix; C2: C2 domain; Zn: zinc finger; VEFS: VEFS (VRN2-EMF2-FIS2-Su(z)12)-Box; WD: W40 repeat; PHD: plant homology domain; WH: winged-helix domain; JmjN: Jumonji N domain; JmjC: Jumonji C domain; ARID: A–T rich interaction domain; ChromoL: chromo-like domain

In addition to E(z), also the Su(z)12/SUZ12 proteins has diverged during evolution. Although the general domain structure of SUZ12 is conserved, the sequences of Drosophila and human proteins display only 30% identity (Fig. 2). Considering that the SUZ12 protein acts as a binding hub, required for the assembly of PRC2 [31], one could speculate that these rather substantial differences in the primary sequence of SUZ12 compared to Su(z)12 are possibly necessary to allow the binding of additional proteins unique to mammals, such as EPOP and PALI1. This idea is supported by recent structural work that demonstrated that the less-well conserved N-terminal part of SUZ12 (Fig. 2) acts as a binding platform for several associated proteins such as AEBP2, JARID2, PHF19, and EPOP [31].

In contrast to Su(z)12/SUZ12 and E(z), the changes to the core member Esc/EED are less pronounced. Drosophila Esc and the human EED proteins are more than 50% identical. Esc/EED consists of seven WD40 domains that form a beta propeller and function as a critical structural component of PRC2 [50]. Esc/EED deletion almost completely abrogates the activity of PRC2 in Drosophila [51] and mammals [52, 53], demonstrating the importance of this protein. Esc/EED interacts with the H3K27me3 mark [54], which is required for steric activation of PRC2 [50, 55] and thus the propagation of the repressive histone mark. This critical function is facilitated by the highly similar WD domains of Esc/EED (Fig. 2) and is thus conserved between Drosophila and mammals [54].

Notably, in Drosophila, a homolog of Esc exists, called Esc-like (Escl), which mainly differs from Esc at the N-terminus [56, 57]. It shows a distinct expression pattern, and it has been shown that PRC2 with Escl has reduced methyltransferase activity [56, 57]. In mammals, only one EED homolog exists, which, however, produces four alternative isoforms with varying N-termini [58]. PRC2 with an N-terminally truncated EED has reduced methyltransferase activity towards the noncanonical substrate H1K26, at least in vitro [58]. Thus, this work implies that alternative Esc/EED proteins may be involved in modulating the methyltransferase activity of PRC2. However, the biological relevance and molecular differences of the distinct Esc/EED proteins remain uncertain.

Importantly, PRC2 is not only allosterically activated upon binding of Esc/EED to the H3K27me3 mark, but also inhibited by active chromatin marks, such as H3K4me3, H3K36me2, and H3K36me3 [59]. In contrast to the in trans acting H3K27me3 [50], the inhibition by the active histone marks is established in cis, meaning that active histone marks on a specific histone tail inhibit PRC2 from depositing the H3K27me3 mark on the very same tail [59,60,61]. Mechanistically, it has been shown that the H3K36me3 mark occupies a critical position at the PRC2 nucleosome interaction interface, therefore directly influencing the activity of PRC2 [62]. Similar to the allosteric activation by H3K27me3, this inhibitory function of active histone marks is evolutionarily conserved [62].

Collectively, during evolution the PRC2 core has largely retained the ability to form a complex and facilitate H3K27me3 deposition. Both allosteric activation by H3K27me3 and inhibition by active histone marks are conserved between Drosophila and mammals.

Evolution of the PRC2.1 subcomplexThe Polycomb-like proteins (PCLs)

The Polycomb-like (Pcl) protein was first described in the Drosophila system to participate in a Polycomb-related function [63]. Further work confirmed that Pcl and its mammalian homologs are components of PRC2 [64, 65] and are required to generate high levels of H3K27me3 [19, 65]. In Drosophila, Pcl is part of a subset of PRC2 that does not contain Jarid2 [19], which is in line with the results from mammalian PRC2, where homologous Polycomb-like proteins (PHF1, MTF2, PHF19; also PCL1, PCL2, PCL3) have been linked to the PRC2.1 subcomplex [20]. Accordingly, the phenotype of mutants or knockout of Pcl or the respective mammalian homologs is weaker compared to PRC2 core proteins [19, 66, 67]. However, numerous studies have demonstrated the importance of all three mammalian PCL proteins in physiological and pathophysiological processes, such as DNA repair [68, 69] and cancer cell proliferation [70,71,72,73], supporting that they are essential regulators of the PRC2 function.

Structurally, the PCL proteins possess five globular domains, namely a Tudor domain, two PHD fingers, a winged-helix domain, and a C-terminal chromo-like domain. This general structural composition is conserved between the Drosophila Pcl and its vertebrate paralogs PHF1, MTF2 and PHF19. The C-terminal chromo-like domain was identified as necessary and sufficient for the interaction with PRC2 [31, 74,

留言 (0)

沒有登入
gif