The expanding role of the Ehmt2/G9a complex in neurodevelopment

ABSTRACT

Epigenetic regulators play a crucial role in neurodevelopment. One such epigenetic complex, Ehmt1/2 (G9a/GLP), is essential for repressing gene transcription by methylating H3K9 in a highly tissue- and temporal-specific manner. Recently, data has emerged suggesting that this complex plays additional roles in regulating the activity of numerous other non-histone proteins. While much is known about the downstream effects of Ehmt1/2 function, evidence is only beginning to come to light suggesting the control of Ehmt1/2 function may be, at least in part, due to context-dependent binding partners. Here we review emerging roles for the Ehmt1/2 complex suggesting that it may play a much larger role than previously recognized, and discuss binding partners that we and others have recently characterized which act to coordinate its activity during early neurodevelopment.

The Ehmt1/2 complex

One of the most fundamental processes that cells must undergo during development is lineage specification where progenitor cells are instructed to form the numerous varieties of cells found in the adult organism. Gene expression must be tailored in a cell type-specific manner to allow cells to undergo lineage commitment. Signaling pathways, transcription factors and chromatin modifiers cooperate to regulate transcriptional activity. Throughout development, not only are lineage-specific genes upregulated, but non-essential progenitor genes must also be silenced. An inability of cells to properly repress the progenitor regime can lead to various developmental syndromes and cancer. The euchromatin histone lysine methyltransferases (Ehmts) play an important role in this process. Ehmt1 (GLP) and Ehmt2 (G9a) are the major methyltransferases responsible for the mono- and dimethylation of lysine 9 on Histone 3 (H3K9me1/2),^1,2 and potentially monomethylation of both H3K27¹ and H3K56³ in some contexts. Its predominant histone mark, H3K9me2, is essential to alter gene activity from a state of active transcription to transcriptional repression.^2,4 However, the loci-specific targeting mechanisms of these protein complexes are poorly characterized.

Although the function of Ehmt1/2 to modify histone proteins is well established, co-factors of the Ehmt1/2 epigenetic complex required to fulfill its various functions are still not well defined. The Ehmt1 and 2 proteins belong to the SET containing family of epigenetic regulators,^1,5 a conserved domain responsible for catalyzing lysine methylation on histones among other proteins. At the protein level, both Ehmt1 and 2 are very similar, each containing an N-terminal activation domain,⁶ an Ankyrin repeat domain^7,8 and the aforementioned SET domain. Ehmt1 and Ehmt2 assemble in a stoichiometric 1:1 ratio into a larger complex and function together to regulate the mono- and dimethylation of H3K9.⁹ The molecular mechanisms that target the Ehmt1/2 complex to unmethylated chromatin sites is not well defined, but may ultimately rely on physical association with auxiliary cofactor proteins such as the zinc finger-containing proteins Wiz^10,11 and Znf518,¹² or in some cases non-coding RNA.¹³ Once recruited, to exert its repressive effects the Ehmt1/2 complex functions with co-repressors such as H3K4 demethylase Jarid1a,¹⁴ the histone deacetylase Hdac1,^15-17 the heterochromatin promoting HP1^18,19 and various DNMTs.^20-22

We, and others, have recently identified an additional cofactor necessary for proper ehmt1/2 recruitment to target genes: znf644.^11,23 It remains to be determined whether znf644 is sufficient to recruit ehmt2 to specific chromatin domains or whether it serves as a scaffold or adaptor protein required for the ehmt2 complex integrity and function. Another report describes a consensus sequence for ZNF644 binding sites based on ChIP-seq data.¹¹ However, additional biochemical data are necessary to further explore the possibility of direct sequence-specific DNA binding activity for znf644. Regardless of its precise functional role, it is clear that znf644 is nonetheless an essential member of the ehmt2 complex required during early neurogenesis, and it will be interesting to see if additional complex members are found playing a similar role in ehmt2 within other lineages.

In addition to being recruited to genomic loci, presumably through physical association with cofactor proteins, the Ehmt1/2 complex directly binds monomethylated proteins, including H3K9me1, via its Ank repeat domains.⁷ Intriguingly, although Ehmt1 and Ehmt2 contain nearly identical protein domains, there is evidence that they have distinct roles within the complex. When the Ehmt1 Ank repeat domains are inactivated by mutation, the Ehmt1/2 complex is no longer able to bind effectively to monomethylated H3K9.²⁴ However, when the SET domain of Ehmt1 is inactivated by mutation embryos are viable and survive to adulthood,²⁵ in contrast to Ehmt2 SET mutations, which are embryonic lethal.²⁶ This data implies that Ehmt1 plays a particularly important role in recruiting the complex to monomethylated sites, whereas Ehmt2 is required for the methyltransferase activity.

Epigenetic marks on chromatin occur in combinations, rarely acting alone to control transcriptional activity. Generally speaking, marks such as H3K4me3, H3K9ac and H3K27ac denote regions of active gene expression, and marks such as H3K9me2/3 and H3K27me3 denote regions of transcriptional repression (for review see refs. 27, 28). Among the many known repressive epigenetic modifiers, the Ehmt1/2 complex plays a central part in establishing repressive epigenetic marks. In embryonic stem cells, deposition of H3K9me2 coincides with deactivation of pluripotency genes, occurring often before the removal of activating marks such as H3K27ac.²⁹ There is evidence that Ehmt2 is also able to regulate both mono-methylation¹ and dimethylation of H3K27 via both direct interaction with PRC2³⁰ as well as indirect mechanisms.³¹ Furthermore, in Ehmt2 loss-of-function experiments, PRC2 occupancy at target promoters is significantly reduced.³⁰ Ehmt1/2 function is therefore required, at least in some circumstances, for PRC2 recruitment. Interestingly however, the Ehmt2 (H3K9me2) and PRC2 (H3K27me3) marks do not co-localize, instead repressing distinct ‘sets’ of genes.²⁹ It is possible, however, that the Ehmt2-dependent mark is removed by a specific demethylase before the PRC2-dependent mark is deposited to transition between chromatin states. Also, in addition to its role regulating histone methylation, Ehmt2 is required to regulate DNA methylation by recruiting Dnmt3 to sites of H3K9 methylation to methylate adjacent DNA.^21,32 This suggests that the Ehmt1/2 complex plays a central role in long-term epigenetic silencing of gene expression during lineage commitment and cell differentiation.

Canonical roles of Ehmt1/2 complex

As mentioned above the most well studied roles of the Ehmt1/2 complex are as a nuclear epigenetic regulator. Epigenetic repression is essential for the proper progression of several processes throughout development, such as: genomic imprinting, cell cycle regulation and exit, lineage commitment, and differentiation. Ehmt2 has long been known as a central regulator of these processes.

Ever since Ehmt2 was implicated in Prader-Willi syndrome,³³ Ehmt2-mediated epigenetic repression has been examined for its role in genomic imprinting. A great deal of attention has been paid to Ehmt2's role in defining large blocks of repressive chromatin,³⁴ however this has been called into question with multiple reports showing that H3K9me2 is not deposited in large blocks of chromatin during differentiation.^35,36 In vitro experiments suggest that Ehmt2 function might instead be essential for proper maintenance of imprinted alleles. For instance, Ehmt2 has been implicated in both embryonic stem cells³⁷ and embryonic fibroblasts³⁸ to maintain a repressed state of imprinted alleles. However, Ehmt2 and DNMTs remain bound and DNA methylation persists at imprinted loci even when the histone methyltransferase activity of Ehmt2 is inhibited.³⁷ The persistence of DNA methylation in the absence of H3K9me2 suggests that Ehmt2s role in imprinting is independent of its HMT activity, and rather that it may be required simply to recruit DNMTs to imprinted loci. No apparent requirement for imprinting in embryonic tissue has been found in vivo, however a role has emerged in maintaining repression of imprinted alleles within the placenta.^13,39 Creation of an Ehmt2 mutant lacking the C-terminal half of the protein, including the Ank repeat and SET domains, lead to a partial loss of imprinting at the Kcnq1 locus on chromosome 7 in the trophoblast, but not in the embryo proper.³⁹ Imprinting at the Kcnq1 locus is therefore more sensitive to a loss of Ehmt2 function in the trophoblast. The absence of a detected role for Ehmt2 in embryonic imprinting may be due to residual N-terminal protein function in this deletion strategy, or possibly even redundancy with Ehmt1 or other mechanisms in the embryo proper to guard against a loss of imprinting maintenance.

Ehmt2 function in the epigenetic control of gene expression during lineage commitment and differentiation has been extensively documented. Ehmt2 is absolutely required for early development as loss-of-function experiments result in delayed development and embryonic lethality.²⁶ It has historically been considered that as cells differentiate increasing amounts of methylation are deposited on histones acting to silence progenitor gene expression. This classical view has recently been brought into question with the observation that loss of Ehmt2 does not affect the adoption of lineage progenitor identity.²⁹ However, H3K9me2 has been implicated in the terminal commitment of blood, cardiac, retinal, neural, muscle and germline lineages.^{23,25,29,40-43} For example, H3K9me2 plays a dynamic role in retinal progenitor cells⁴⁴ and loss of Ehmt2 mediated epigenetic repression leads to profound defects in the early retina. Zebrafish ehmt2 morphants show severe decreases in crx, neuroD and irbp gene expression; markers of retinal cell differentiation.²¹ Similarly, we have shown that a loss of the ehmt2 cofactor znf644 shows prolonged expression of progenitor genes vsx2 and ccnd1,²³ as well as abnormal markers of differentiated cell lineages. We found marked reduction of H3K9me2 within the vsx2 and ccnd1 promoter regions suggesting that the ehmt2 complex was either not properly recruited or otherwise functionally unable to repress the progenitor regime and allow for proper cell cycle exit and lineage commitment. Similarly, in the mouse constitutive knockout, Ehmt2 loss in retinal progenitor cells lead to maintained progenitor gene expression (Chx10/Vsx2 and various cyclins) and concomitant differentiation defects.⁴² This maintenance of progenitor gene expression in the retina suggests that Ehmt2 is required in lineage progenitors for final commitment of mature retinal cells. Furthermore, in both zebrafish and mouse, loss of Ehmt2 mediated repression leads to prolonged cell cycling and an increase in apoptosis. Sustained proliferation is a hallmark of Ehmt2 loss in several tissues and suggests that it is required to control cell fate transitions during differentiation.

Methylation-independent transcriptional activating functions of Ehmt2

In depth protein analysis and domain-specific deletions of Ehmt2 have recently begun to elucidate a second, opposite role in transcriptional regulation. In addition to a requirement in epigenetic repression, reports have also surfaced illustrating a transcriptional co-activator function of Ehmt2.⁴⁵ Ehmt2 has been shown to be essential for activation of several genes critical to early development including p21⁴⁶ and β-Globin.⁴⁷ Unlike transcriptional repression, Ehmt2's co-activator role does not require either the SET or Ank repeat domains,^6,46,48 suggesting that this co-activator role is methylation-independent. Furthermore, Ehmt2 is dependent upon a different pool of transcription factors, such as Runx2⁴⁹ and Nfe2,⁴⁷ to recruit it to sites of transcriptional activation. Once recruited Ehmt2 in turn functions as a scaffold protein to enhance recruitment of further co-activators including Carm1,⁴⁵ p300,⁴⁸ the mediator complex¹⁴ and DNA Pol II.⁴⁷ In contrast, Ehmt1 has also been recently implicated in transcriptional activation during adipogenesis independent of Ehmt2.⁵⁰ Indeed Ehmt1 and Ehmt2 seem to play contradictory roles during adipogenesis.^50,51 This suggests that there may remain underappreciated functions of Ehmt1 and 2 independent of the Ehmt1/2 heterodimer.

The dual nature of Ehmt2 as both a repressor and activator could enable coordinated regulation of transitions in gene expression. For example, a switch occurs in the regulation of the globin locus during development. The predominant form of embryonic hemoglobin is fetal hemoglobin (HbF), which incorporates the developmentally expressed γ-globin protein. After gestation this switches to adult hemoglobin (HbA), which in turn incorporates β-globin (reviewed in ref. 52). Ehmt2 is required during this switching process to epigenetically repress the embryonic γ-globin, and activate transcription of the adult β-globin genes.⁴⁷ A similar role for Ehmt2 in neurogenesis has not yet been explored, but given that Ehmt2 knock down experiments show maintained proliferation markers and a loss of lineage-specific markers (see below), it remains possible that Ehmt2 is required for both negative and positive transcriptional regulation during early neurodevelopment.

While the activator/repressor duality of Ehmt2 function is becoming better established, the control between these 2 opposing functions is not yet clear. Multiple splice variants have been noted both in the human and mouse, denoted by inclusion/exclusion of a 5′ exon, as well as inclusion/exclusion of exon 10.^53-56 Although none of these splice variants has been reported to effect the overall HMT function of the Ehmt1/2 dimer, both alternatively spliced 5′ domain and exon 10 have independently been reported to be required for nuclear localization.^54,56 Furthermore, activation by the human EHMT2 protein has been found to be predominantly dependent of the hEHMT2-s variant which lacks part of the 5′ TAD, in at least some contexts.⁵³ It is therefore possible that alternative splicing the Ehmt2 transcript acts to balance both activator and repressor, as well as nuclear versus cytoplasmic functions.

Ehmt2 in cell cycle regulation

Proper control of the cell cycle and cell cycle exit is essential to development. Deregulation of cell cycle regulators can lead to differentiation defects, increased apoptosis and cancer – processes in which Ehmt2 has been implicated. Progression through the cell cycle is controlled predominantly by the functions of the CDK/cyclin family of genes (reviewed in refs. 57, 58). Recently, our laboratory and others have demonstrated a direct role of Ehmt2 as a cell cycle regulator, being required to epigenetically repress the expression of CyclinD genes upon cell cycle exit.^23,42 As discussed above, we found that when the ehmt2 co-factor znf644 was inhibited, ccnd1 expression was maintained and cells in the early retina remained in a proliferative progenitor state much longer than is normal. We thereby concluded that the ehmt2 complex was not functioning properly to repress the progenitor regime thereby blocking the cells ability to exit the cell cycle and leading to abnormal differentiation. In addition to cyclins however, there are various stages, or checkpoints, at which the cell cycle can be either paused or stopped entirely to allow the cell to cope with various cell stresses, such as DNA damage (reviewed in refs. 59, 60). Central to the control of the cell cycle is p53, a tumor suppressor gene that functions in response to DNA damage and can lead to senescence and/or apoptosis.^61-63 Once p53 becomes activated by phosphorylation,^64-67 it upregulates its own transcription, creating a positive feedback loop, as well as another central checkpoint gene p21 (CDKN1A) which acts to inhibit various Cdks, leading to cell cycle arrest.^68-73 In parallel, p53 can also activate Bax in the apoptosis pathway promoting programmed cell death.^74-77 Although not the only gene responsible for cell cycle control, p53 is a central player in the regulation of cell cycle progression and its mutation in a variety of cancers demonstrates its essential protective function.

Loss of Ehmt2 leads to pronounced growth defects during early development. This growth defect likely manifests from defects in the cell cycle machinery (i.e., Cyclin gene expression) and increased apoptosis as seen in tissue progenitor cells of the early retina.^23,42 It is therefore reasonable to hypothesize that the manifestation of the Ehmt2 phenotype is p53-dependent. Indeed Ehmt2 has been found to bind and methylate lysine 373 of the murine p53 directly.⁷⁸ The Ehmt2-dependent methylation of p53 correlates with its inactivation,⁷⁹ thereby suggesting that when Ehmt2 function is lost, p53 may become primed for activation. However, it has been recently shown that one splice variant of hEHMT2 is also necessary to attenuate p53's transcriptional activation of its target genes in vitro.⁵³ This demonstrates an added level of complexity in Ehmt2's regulation of p53, and that its function as a repressor or co-activator is likely context dependent. This context-dependent activity also extends to the p53 target gene p21, where Ehmt2 is required to both epigenetically repress,^16,43 and activate⁴⁶ expression. This suggests that Ehmt2 may be a core co-regulator of the p53 pathway, being involved in both repressing p53 when not activated, and a co-activator of the p53-dependent stress response pathway once stimulated.

In addition to regulating cell cycle checkpoints and cell stress genes, Ehmt2 has been demonstrated to have a distinct role during S-phase. H3K9me2 has been shown to mark distinct sets of late replicating genes; upon Ehmt2 depletion however there is no effect on the replication timing of these genes.^80-82 Via its role in regulating H3K56 mono-methylation, Ehmt2 plays a direct part in regulating the binding of Pcna to chromatin during late G1 to early S-phase.³ Loss of Ehmt2 in mouse embryonic stem cells leads to a delay in progression through S-phase. Alternatively, this S-phase delay may be in part due to a recently described role of the Ehmt2 complex at the replication fork. Znf644, Ehmt1 and Ehmt2 are present at the replication fork, and their removal resulted in phenotypes consistent with replication stress,⁸³ perhaps owing to impaired Pcna loading. While the role of the Ehmt1/2 complex at the replication fork is still not defined, it was suggested that it may be necessary to regulate non-histone protein function at the fork, or for transmission of H3K9me2 marks to newly synthesized chromatin. In either role, it is likely that the loss of Ehmt2 mediated methylation serves to enhance the activation of a p53-dependent cell cycle checkpoint and delayed progress through S-phase.

Emerging roles of Ehmt2

Ehmt2 has been shown to have roles outside of its function as a transcriptional regulator. These newly defined functions are diverse and regulate several processes throughout the cell. The 3 lesser known functions we will discuss are protein stability, chromatin architecture and genome instability.

As discussed above, there are several examples of Ehmt2 binding and methylating non-histone proteins. Various post-translational histone modifications, including methylation, have been implicated in altering protein stability (for review see ref. 84). Recently there have been several examples that demonstrate Ehmt2-dependent methylation may also impact protein stability. Ehmt2 regulates the protein levels of Sox2 in breast cancer cell lines, independent of its effect on transcription.⁸⁵ Similarly, treatment with BIX01294, a small molecule inhibitor of EHMT1/2's HMT function, decreased the half-life of HIF1α protein in HepG2 cells.⁸⁶ This suggests that Ehmt2 dependent methylation may be necessary to positively regulate protein stability of at least some proteins. However, in other contexts lysine methylation promotes protein degradation;⁸⁷ for example, methylation by Ehmt2 has been show to target MyoD for ubiquitination and proteasomal degradation.⁸⁸ Therefore, it is likely that Ehmt2's role in protein stability is also highly context dependant and that in some cases, Ehmt2s role in negatively regulating protein function may be in part due to its promoting target protein degradation.

It is well established that apoptosis and cancer, 2 phenotypes that Ehmt2 has been strongly associated with in vivo, can arise from abnormal chromosome content. There have been recent reports suggesting a requirement of Ehmt2 for genome integrity.^81,89 How a loss of Ehmt2 dependant methylation leads to chromosome instability is still not well understood; One intriguing possibility is suggested in the recent report that Ehmt2 participates in a multi-HMT complex that is required for maintaining peri-centromeric chromatin,^90-92 a structure that is essential for proper spindle attachment and chromosome segregation during mitosis (reviewed in ref. 93). Without Ehmt2, the multi-HMT complex is disrupted leading to a drastic reduction of H3 methylation in the pericentromeric regions.⁹⁰ Additionally, there is evidence that Ehmt2 loss also perturbs centrosome number and function. EHMT2 knockdown resulted in disruptions in centrosome biogenesis leading to increased genome instability in several cancer cell lines.⁸⁹ Interestingly, disruptions in centrosome biogenesis have been shown to lead to cell cycle exit defects, genome instability and increases in apoptosis reminiscent of the Ehmt2 loss-of-function phenotype in vivo.^94,95 Whether Ehmt2 is required to regulate the expression or activity of individual genes in the centrosome biogenesis pathway has not been studied, and thus the mechanism of centrosome disruption is not clearly defined. Nonetheless, it seems likely that the genome instability resulting from Ehmt2 loss could result from both spindle attachment defects and centrosome defects. Either of these effects on their own, or abnormal chromosome compliment in daughter cells could account for the increases in apoptosis in Ehmt2^−/− cells.

Conclusions and future perspectives

While the enzymatic activity of the Ehmt1/2 SET domains as well as the mono-methyl-lysine binding activity of the Ank repeat domains have been well established, there remains much to be learned regarding the function and regulation of the Ehmt1/2 complex. In particular, the precise contribution of the Ank repeat domains to gene repression needs to be determined. Likewise, the precise function and contribution of the many zinc-finger cofactors, such as Wiz and Znf644, and how they may confer target gene specificity to the methyltransferase complex needs to be elucidated. A deeper understanding of the biochemical action of the Ehmt1/2 complex will greatly assist in more thoroughly interpreting the phenotypic outcomes of dysregulated Ehmt1/2 function.

The epigenetic regulator Ehmt2 and its paralog Ehmt1, along with its cofactors Wiz and Znf644 clearly play an important role during neurogenesis. While delayed development, cell cycle defects, differentiation defects and widespread apoptosis are consistent among various reports in early neural development, molecular consequences of losing Ehmt1/2 function and the mechanism behind these phenotypic outcomes is less clear. Ehmt2 function is central to the epigenetic repression of progenitor genes for differentiation, cell cycle exit, DNA damage and cell stress response, DNA synthesis and mitosis in several cell types. Defects in these cell processes would lead to the overall early neural phenotypes we,²³ and others have reported. However, the impact of Ehmt1/2 on these various cell processes in early neural progenitors has not yet been fully explored.

Some of the newly defined roles for the Ehmt1/2 complex in regulating protein function and stability, as well as in the biogenesis of structures such as the centrosome provide a novel perspective in reinterpreting past results. Ehmt2's role in regulating cell division is surely complex, with potential impacts on cyclin gene expression, cell stress response, multiple inputs into DNA synthesis, and chromatin compaction and segregation. However, these results also suggest an intriguing question: if Ehmt2 plays a role in centrosome biogenesis, could it also be important for regulating or organizing the microtubule network. If so, this would have intriguing implications for early neurodevelopment as neural progenitors rely on microtubules for multiple cell signaling and trafficking events, as well as for cell movements and tissue morphogenesis. Whatever the case, the Ehmt1/2 complex seems to play a much larger role in early development than simple epigenetic repression at target promoters.

Finally, an intriguing possibility is that Ehmt2 may act to both repress neural progenitor gene expression while co-activating markers of differentiation as it does with the β-globin locus. Supporting evidence from studies in early retinal development show that Ehmt2 and Znf644 knockdown experiments result in cells that fail to undergo terminal differentiation.^21,23,42 This would place Ehmt2/Znf644 in the center of a genetic switch, turning off the progenitor gene cassette while activating the differentiation gene expression regime. Studies of Ehmt2's activator function in early neural progenitors is thus far lacking, and it will be interesting to see if future studies corroborate this hypothesis.

Disclosure of potential conflicts of interest

No potential conflicts of interest were disclosed.

Funding

Funding for research in the Tropepe lab is supported by CIHR.