Insights into epigenetic patterns in mammalian early embryos

A unique pattern in mouse oocytes and major resetting of H3K4me3 after fertilization

Histone modifications play critical roles in the spatiotemporal regulation of gene expression in mammals (Bannister and Kouzarides, 2011; Sadakierska-Chudy and Filip, 2015; Huang et al., 2019; Xia et al., 2019). Recently, low-input ChIP-seq methods were developed and utilized to profile the genome-wide landscape of histone modifications during ZGA and the first cell fate decision in mouse early embryos (Dahl et al., 2016; Liu et al., 2016b; Zhang et al., 2016) (Fig. 1). After fertilization, H3K4me3 (a mark of active promoters) in the paternal genome is rapidly depleted, but this mark is re-established during major ZGA. By contrast, the noncanonical form of H3K4me3 (ncH3K4me3), covering broad domains in both promoters and distal regions, is found in the maternal genome (making up ~ 22% of the genome) (Dahl et al., 2016; Zhang et al., 2016). ncH3K4me3 is readily established in mature oocytes and is not replaced by canonical H3K4me3 until the major ZGA stage. The establishment and removal of distal ncH3K4me3 is a unique characteristic of oocyte and early embryo genomes. However, the mechanism and function of ncH3K4me3 dynamics still need to be further investigated. DNA methylation in the maternal genome is anti-correlated with ncH3K4me3, as ∼ 18% versus ∼ 57% CpG methylation is observed in oocytes with or without ncH3K4me3 (Wang et al., 2014; Dahl et al., 2016). Active removal of broad H3K4me3 domains by the lysine demethylases KDM5A and KDM5B is essential for normal ZGA and early embryogenesis(Dahl et al., 2016), while Kdm5b overexpression results in transcriptome reactivation in mature oocytes (Zhang et al., 2016), suggesting that ncH3K4me3 may be responsible for the genome-wide silencing state. Transposable elements (TEs), including B1/B2/B4 and ERVL, show high overlap with ncH3K4me3 in distal regions, indicating that ncH3K4me3 in distal regions can be correlated with the activities of repeats (Zhang et al., 2016). ncH3K4me3 is reprogrammed upon ZGA at the late 2-cell stage, which requires both zygotic transcription and active demethylation, rather than passive dilution (Zhang et al., 2016). Maternal factors within the oocyte cytoplasm must direct the rapid change of the H3K4me3 landscape. Identifying key maternal factors may provide more insight into the mechanism of H3K4me3 remodeling. On the other hand, H3.3 is cooperated to the genome after fertilization and is essential for embryo development (Lin et al., 2013; Lin et al., 2014; Wen et al., 2014). The turnover of H3.3 is associated with histone modification changes such as H3K27me3 and H3K36me3 (Kraushaar et al., 2013; Lin et al., 2013), but the correlation between H3.3 and ncH3K4me3 is unclear and whether the replacement of H3.3 is responsible for the removal of ncH3K4me3 deserves further investigation.

The canonical form of H3K4me3 in the promoter region is established at the 2-cell stage. Interestingly, the width of H3K4me3 peaks is highly dynamic during mouse pre-implantation embryonic development and is positively correlated with gene expression level (Liu et al., 2016b). The dynamicity of the extent of H3K4me3 marks may provide a novel mechanism of epigenetic regulation in early cleavage embryos when most repressive makers are removed from the promoter region (Liu et al., 2016b). Broad H3K4me3 domains (> 5 kb) in the promoters are much more abundant in early embryos than in MII oocytes or derived cell lines. Lack of KDM5B can extend the width of promoter H3K4me3 domains and disrupt precise lineage differentiation, which indicates that broad promoter H3K4me3 marks may help maintain the transcription of key cell-type-specific factors in a steady state (Liu et al., 2016b). The mechanism of H3K4me3 reprogramming remains unknown, but a recent study showed that the establishment of broad H3K4me3 domains may be catalyzed by KMT2B, a deficiency of which results in slower ovulation and embryonic arrest (Andreu-Vieyra et al., 2010). However, which transcription factors are involved in this precisely controlled process requires further investigation.

Canonical pattern of H3K4me3 in GV oocytes and species-specific dynamics of H3K4me3 in humans

The CUT&RUN (cleavage under _targets and release using nuclease) method (Skene and Henikoff, 2017) reduces the requirement for materials for the genome-wide analysis of histone modifications. Thus, the reprogramming of key histone modifications in human pre-implantation embryo development is finally accessible to profiling (Fig. 2). Unexpectedly, H3K4me3 was found to show sharp peaks at promoters in human GV and MI oocytes (Xia et al., 2019), unlike the abundant broad ncH3K4me3 domains distributed in distal partially methylated regions in mouse mature oocytes. These differences indicate a different epigenetic mechanism of genome silencing in human oocytes that is regulated by ncH3K4me3 in mice. Notably, weaker (compared to promoter H3K4me3) but widespread distal H3K4me3 marks have been discovered in pre-ZGA embryos, which indicates the de novo deposition of H3K4me3 and characterizes a primed epigenetic state for subsequent transcriptional regulation (Xia et al., 2019). The function of these distal H3K4me3 marks remains unknown, and whether their function is similar to that of ncH3K4me3 in mice before ZGA is worthy of elucidation. The profiling of such major reprogramming of both promoter and distal H3K4me3 marks will pave the way for future investigations of parental-to-zygotic transition in human pre-implantation embryos.

Large-scale loss of promoter H3K27me3 marks after fertilization and pervasive ncH3K27me3 marks

H3K27me3 is a polycomb-based chromatin signature associated with gene repression (Bernstein et al., 2006; Simon and Kingston, 2009; Di Croce and Helin, 2013). H3K27me3 has been proved to be intergenerationally inherited from the maternal genome during early embryogenesis, regulating the activation of enhancers and lineage-specific genes (Inoue et al., 2017a; Zenk et al., 2017). During mouse early embryo development, H3K27me3 in promoter regions of both maternal and paternal alleles exhibits a widespread loss as early as in PN5 zygotes, followed by rapid dynamics during the transition from the morula to inner cell mass (ICM) and trophectoderm (TE) stages (Santos et al., 2005; Liu et al., 2016b; Zheng et al., 2016; Inoue et al., 2017a) (Fig. 1). This rapid loss is accomplished by the global erasure of H3K27me3 in the paternal genome and the selective depletion of promoter H3K27me3 in the maternal genome. In addition, dual-omics analysis indicate a negative correlation between H3K27me3 and DNA methylation in MII oocytes (Zheng et al., 2016) and some genomic regions in post implantation embryos (Yang et al., 2018b), while the anti-correlation is indistinct in cultured cells like ESCs and neural stem cells (NSCs) (Bartke et al., 2010; Wu et al., 2010; Brinkman et al., 2012), as well as mouse pre-implantation embryos, in which H3K27me3 is negatively correlated with H3K4me3 (Liu et al., 2016b; Zheng et al., 2016). On the other hand, H3K27me3 also appears in non-promoter regions, where it is highly pervasive and dramatically promiscuous. This supports the hypothesis that polycomb complexes constantly scan the whole genome for H3K27me3 deposition to compensate for the transcription repression function, which can also be realized by DNA methylation and H3K9me3. However, the functions and molecular mechanisms of such non-canonical H3K27me3 patterns remain to be verified.

In SCNT embryos, strong H3K27me3 signals are observed in all pseudopronuclei at the 1-cell stage (Zhang et al., 2009; Xie et al., 2016), which are highly distinct from the asymmetric H3K27me3 marks in parental genomes in normally fertilized embryos. In addition, ZGA failure in SCNT embryos at the 2-cell stage seems to be compatible with the defect of H3K27me3 erasure inherited from donor somatic cells, suggesting that H3K27me3 represents an epigenetic barrier to epigenome reprogramming (Zhang et al., 2009; Matoba et al., 2018). But simply eliminating H3K27me3 signals does little to promote the SCNT efficiency. A recent study shows that the overexpression of Kdm6a (an H3K27me3 demethylase) facilitates the expression of ZGA-related genes in SCNT embryos but results in no improvement of the birth rate or ntESC establishment efficiency. Conversely, knockdown of Kdm6b (another H3K27me3 demethylase) promotes both ZGA and SCNT efficiency (Yang et al., 2018a). H3K27me3-dependent imprinting genes are aberrant in SCNT embryos (Okae et al., 2014b; Inoue et al., 2017a) and may be amended by knockdown of Kdm6b. Kdm6b reduction also disrupts the ectopic expression of Xist, in which H3K27me3 serves as an imprinting mark and is correlated with abnormality of X chromosome inactivation in SCNT embryos (Inoue et al., 2010; Matoba et al., 2011; Inoue et al., 2017b; Matoba et al., 2018; Yang et al., 2018a).

Maternal H3K27me3-mediated imprinting has been discovered in early mouse embryos (Inoue et al., 2017a), as it can mediate silencing of DNA hypomethylated promoters, achieved by polycomb recruitment (Deaton and Bird, 2011). The H3K27me3 imprinting is probably established during oogenesis and maintained in pre-implantation embryos, which will be diluted in ICM and most of the H3K27me3-mediate imprinting is lost in the epiblast (EPI) of E6.5 embryos (Inoue et al., 2017a). Recent studies reveal that oocyte-derived H3K27me3-mediated imprinting will switch to being DNA methylation-dependent in extra-embryonic cells after implantation, in which both maternal EED and zygotic DNMT3A/B may take part (Smith et al., 2017; Inoue et al., 2018; Chen et al., 2019b), reflecting the complementary roles of H3K27me3 and DNA methylation in controlling imprinting. Recently, the complete loss of H3K27me3-dependent imprinting patterns in maternal allele-specific genes has been found in SCNT blastocysts, probably due to the absence of H3K27me3 marks at these loci in donor somatic cells (Okae et al., 2014b; Matoba et al., 2018). These epigenetic abnormalities of pre-implantation SCNT embryos are highly correlated with subsequent placenta overgrowth and embryonic lethality (Miri et al., 2013; Okae et al., 2014b; Inoue et al., 2017b). Intriguingly, the upregulation of clustered miRNAs of Sfmbt2, one of the H3K27me3-dependent imprinted genes (Matoba et al., 2018; Chen et al., 2020), has recently been identified as the major cause of placental hyperplasia in SCNT mice (Inoue et al., 2020). These observations highlight the crucial function of H3K27me3-imprinted genes in both pre-implantation and postimplantation embryos, although the detailed molecular mechanisms and consequences of intergenerational epigenetic inheritance remain elusive.

Global erasure of H3K27me3 at human ZGA

H3K27me3 in human GV oocytes is deposited in promoters of developmental genes and partially methylated domains, differing from the pattern in mouse oocytes (Xia et al., 2019). During human pre-implantation embryo development, the resetting of H3K27me3 is also different from that in mice. Human embryos at ZGA (8-cell stage) show almost no H3K27me3 signal, indicating the global erasure of H3K27me3 on both parental genomes (Xia et al., 2019) (Fig. 2). The absence of the core components of polycomb repression complex 2 (PRC2) in human embryos may be correlated with the loss of H3K27me3 (Saha et al., 2013). On the other hand, the global loss of H3K27me3 in human embryos predicts the absence of imprinting regulation, such as X chromosome inactivation (XCI), which is one of the critical events during early mouse embryogenesis (Petropoulos et al., 2016; Inoue et al., 2017b). Moreover, the H3K27me3-mediated imprinted genes identified in mouse early embryos, which have human orthologs, seems to experience de novo deposition of H3K27me3 (Xia et al., 2019). Further study is required to verify whether H3K27me3-controlled imprinting exist in human early embryos. Intriguingly, an analysis revealed asymmetric H3K27me3 patterning between ICM- and TE-specific genes in human pre-implantation embryos. Considering that ICM samples are a mixture of EPI and PE (primitive endoderm) cells, further studies are urgently needed to determine whether preferential H3K27me3 deposition also exists between EPI and PE cells (Xia et al., 2019).

Bivalent H3K4me3 and H3K27me3

Bivalent domains are considered to provide developmental regulators for later transcriptional activation upon differentiation (Vastenhouw and Schier, 2012). Although an H3K4me3/H3K27me3 bivalent state has been observed in lineage control genes in ESCs (Bernstein et al., 2006), zebrafish blastomeres (Vastenhouw et al., 2010), and primordial germ cells (PGCs) (Sachs et al., 2013), whether the associated regulatory characteristics differ remains unclear. It has been shown that the number of bivalent peaks is much lower in mouse pre-implantation embryos than in ESCs (Liu et al., 2016b), which indicates that bivalency may be more important in a stable cell line than in a transient stage. Bivalency is absent in developmental genes and is not established until the blastocyst stage, when lineage differentiation begins. Notably, most bivalent ICM and TE genes can be inherited by ESCs and TSCs and exhibit lower levels of expression (Liu et al., 2016b; Zheng et al., 2016). It is worth mentioning that EZH2 and SUZ12, the core components of PRC2, which are responsible for H3K27me3 deposition, _target most of the inherited bivalent genes (Margueron and Reinberg, 2011; Liu et al., 2016b). This finding may help us understand the regulatory pattern of PRC2-mediated H3K27me3 upon the exit from totipotency to differentiation in early embryo development.

There is an absence of H3K4me3 and H3K27me3 bivalency in developmental genes upon implantation (Zheng et al., 2016). Intriguingly, in the EPI on embryonic day 6.5 (E6.5), stronger bivalency, defined as “super bivalency”, is found in the promoters of developmental genes, which is also evident in the E7.5 ectoderm but is much weaker in the E6.5 visceral endoderm (VE), mESCs and somatic lineages (Xiang et al., 2020). This raises the question of the function of this transient super bivalent state. Evidence shows that “super bivalency” is correlated with lineage-specific gene activation at later stages, such as cortex or heart differentiation. In addition, these super bivalent genes in the E6.5 EPI exhibit a unique higher-order chromatin organization (Xiang et al., 2020). These findings raise the possibility that strong bivalency in the primed EPI can help key developmental genes to maintain a unique spatial distribution and remain in a poised state for activation. KMT2B (MLL2) is known to be a methyltransferase for ncH3K4me3 in mouse oocytes (Hanna et al., 2018) and is responsible for the deposition of H3K4me3 in bivalent regions. In Kmt2b knockout embryos, H3K4me3 is globally downregulated at bivalent promoters, which is associated with aberrant developmental gene activation. Unexpectedly, bivalency is partially restored in some fraction of developmental gene promoters in the E8.5 head, indicating that Kmt2b plays a critical role in the super bivalency of the E6.5 EPI, while compensation mechanisms are involved in later embryonic development (Xiang et al., 2020). Furthermore, DNA hypomethylation is also suggested to participate in bivalency maintenance, in that Tet1/2 double knockout (DKO) leads to increased DNA methylation levels, followed by a significant H3K4me3 decrease in bivalent developmental genes (Xiang et al., 2020). Taken together, these results reveal a unique chromatin state transition that specifically appears during development from pre-implantation embryos to cell fate-determined lineages.

H3K9me3 is a barrier to cell fate transition

H3K9me3-dependent heterochromatin is regarded as a barrier to cell fate changes, as it occludes the DNA from transcription factor binding (Tachibana et al., 2002; Burton and Torres-Padilla, 2010; Soufi et al., 2012; Becker et al., 2016). The principles and mechanisms of H3K9me3-dependent heterochromatin formation and function have recently been well reviewed elsewhere (Allshire and Madhani, 2018).

Early study has discovered gradual but incomplete demethylation patterns of H3K9me2 and H3K9me3 in SCNT embryos, in contrast to the asymmetric ones present in the parental genomes of fertilized embryos (Wang et al., 2007). The aberrant H3K9me3 reprogramming is deemed to directly causes ZGA failure, especially in SCNT embryos (Schultz, 2002; Matoba et al., 2014). At the 2-cell stage, reprogramming-resistant regions (RRRs) marked by H3K9me3 are defined, which are regions inherited from donor cells that fail to be successfully reprogrammed upon embryonic development (Matoba et al., 2014). Kdm4d (an H3K9 demethylase) overexpression in embryos and Suv39h1/h2 (H3K9me3 methyltransferases) knockdown in donor cells can rescue the transcription of ZGA-related genes impeded by H3K9me3, which dramatically improves the developmental rate of blastocysts. Notably, subsequent experiments using an embryo biopsy system and single-cell transcriptome sequencing identified Kdm4b, another H3K9me3 demethylase, as a key factor in the 2-cell arrest of SCNT embryos (Matoba et al., 2014; Liu et al., 2016a). Both studies emphasize the critical role of H3K9me3 reprogramming during early embryogenesis. The analysis of H3K9me3 in donor cells and 2-cell embryos showed a reduction in resistant H3K9me3 signals in Kdm4d-overexpressing SCNT embryos (Matoba et al., 2014). Moreover, overexpression of KDM4A also drastically increased the blastocyst rate of human SCNT and improves the derivation of human ntESC, with the similar mechanism as in mice (Chung et al., 2015). Our recent study also shows that H3K9me3 in donor cells also prevent the removal of topologically associated domains (TADs) during SCNT (Chen et al., 2020). These findings suggest that some H3K9me3, mainly associated with heterochromatin, is resistant to the cell conversion methods and presents as a barrier to reprogramming to pluripotency that impairs both the efficiency of reprogramming and the quality of embryogenesis.

On the other hand, H3K9me3 modification regulates the expression of repeats elements and some protein coding genes in mouse pre-implantation embryos (Hatanaka et al., 2015; Wang et al., 2018a). Kdm4b overexpression can improve cloning efficiency, but injection of high levels of Kdm4b mRNA have been proved to interrupt TE differentiation and an optimized Kdm4b mRNA injection dose can further increase the implantation and birth rates of cloned mice (Liu et al., 2016a). These indicate the level of H3K9me3 modification need to be well balanced during reprogramming and embryo development.

H3K9me3 regulates the proper expression of LTRs in early mouse embryos

Upon fertilization, a large fraction of repeat-rich sequences, including LTR retrotransposons, undergo dramatic demethylation and become accessible for highly active transcription, which has been shown to be critical for ZGA (Peaston et al., 2004; Wang and Dey, 2006; Zhang et al., 2019). LTRs must be properly regulated because they pose a risk to genome integrity through their potential for illicit recombination and self-duplication. Since major DNA demethylation occurs upon fertilization, the regulation of the transcription of LTRs requires a switch from DNA methylation to other types of epigenetic modifications, including repressive histone modifications, H3K9me3 and H3K27me3 (Wang et al., 2014; Becker et al., 2016). Previous research has shown that CAF-1 is responsible for the deposition of repressive histone marks, including H4K20me3 and H3K9me3, in retrotransposon regions and results in the silencing of retrotransposons in the mouse morula (Hatanaka et al., 2015; Ishiuchi et al., 2015), which provides a possible mechanism of retrotransposon regulation in early embryos. Recently, a high-resolution map for investigating the reprogramming of H3K9me3-dependent heterochromatin in mouse pre-implantation embryos was profiled (Wang et al., 2018a). As expected, it was found that the H3K9me3 peaks fall mainly within LTRs in early embryos and the number of H3K9me3-marked LTRs gradually increases, which remains high during pre-implantation development (Wang et al., 2018a). In contrast, promoter H3K9me3 marks are erased upon fertilization and will be established postimplantation, indicating the involvement of different regulatory mechanisms from those of H3K9me3 in LTRs (Wang et al., 2018a) (Fig. 1). The distribution characteristics of H3K9me3 echo its function in LTR regulation. Further analysis showed that increased H3K9me3 levels are correlated with the silencing of LTRs in the corresponding regions. The proper decoration of H3K9me3 by the CHAF1A complex in LTRs may be critical for blastocyst formation, cell fate decisions and the totipotency-pluripotency transition (Wang et al., 2018a). These results shed light on the molecular mechanisms of LTR silencing by H3K9me3-dependent heterochromatin. Further studies are required to reveal the molecular mechanisms of H3K9me3-dependent chromatin organization for transposable element (TE) regulation in mammalian embryos.

At the postimplantation stage, H3K9me3 marks are re-established in promoter regions. The expression level of lineage-specific genes is repressed owing to the formation of H3K9me3-decorated chromatin in their promoter regions (Wang et al., 2018a). Motif enrichment analysis indicated that Pou5f1, Sox12, Sox11, Lhx1, Zfp105 and Foxa2 are potential transcription factors contributing to the formation of epiblast-specific H3K9me3, while Zbed6, Elf4, Glis2, Creb3l2 and Ascl2 are involved in extraembryonic-specific H3K9me3 formation. Whether these transcription factors all play roles in chromatin organization during embryogenesis remains to be verified. Excitingly, endoderm-specific triple-knockout mutant (TKO) mice for all three histone H3K9 methyltransferases (SETDB1 and SUV39H1/H2) show a three-fold reduction in body weight and are much smaller in size compared to the control mice (Nicetto et al., 2019). In addition, the livers of the TKO mice show remarkable derepression of nonhepatic genes and fail to express mature hepatocyte genes (Nicetto et al., 2019). This highlights the significance of the precise regulation of H3K9me3 deposition at lineage-specific genes during mammalian embryo development.

H3K36me3

H3K36me3, catalyzed by SETD2, is associated with transcriptionally active chromatin (Edmunds et al., 2008). SETD2 can mediate RNA polymerase II interaction and couple H3K36me3 with transcript elongation (Kizer et al., 2005). Unlike H3K4me3, H3K36me3 positively correlates with DNA methylation at gene bodies by recruiting DNMT3A/B (Baubec et al., 2015), which is conserved in most mammalian cells (Hawkins et al., 2010). SETD2 depleted oocytes exhibit huge loss of H3K36me3 and leads to H3K4me3 and H3K27me3 invasions into the former H3K36me3 regions (Xu et al., 2019). In addition, oocytes lacking SETD2 results in an aberrant DNA methylome which includes loss of maternal imprints and aberrant H3K4me3 deposition instead of DNA methylation, especially at ICRs (Xu et al., 2019). Furthermore, SETD2 scarcity results in oocyte maturation defects and embryonic lethality (Xu et al., 2019). These observations emphasize the essential role of H3K36me3 in establishing and safeguarding the maternal DNA methylome during mouse oogenesis and early embryo development.

H3R26me2

Mouse ESCs and blastomeres in early embryos with increased H3R26me2, an activating mark, show higher expression of a subset of pluripotency genes, is highly correlated with cell fate decision and pluripotency (Torres-Padilla et al., 2007; Wu et al., 2009; Goolam et al., 2016; White et al., 2016). Overexpression of coactivator-associated-protein-arginine-methyltransferase 1 (CARM1) in mouse ESCs and embryos elevates expression of key pluripotent genes, like Oct4/Pou5f1 (Wu et al., 2009; Goolam et al., 2016), Nanog (Torres-Padilla et al., 2007; Wu et al., 2009) and Sox2 (Goolam et al., 2016; White et al., 2016), of which promoters display detectable levels of H3R17/26 methylation. Differentially expressed H3R26me2 between 4-cell blastomeres are reported to be mediated by the heterogeneous activity of CARM1, with CARM1-activated blastomeres are prone to develop into ICM rather than TE (Torres-Padilla et al., 2007; Parfitt and Zernicka-Goetz, 2010; Shi et al., 2015). Intriguingly, the earliest cell fate decision in mouse early embryos has recently been advanced to emerge as early as late 2-cell stage, when a long noncoding RNA, LincGET, is transiently and asymmetrically expressed in the nucleus from 2-cell to 4-cell stage (Wang et al., 2018b). Notably, LincGET physically binds to CARM1 (Wang et al., 2018b), promoting CARM1 to accumulate in nuclear granules that requires paraspeckle component NEAT1 and its partner P54NRB (Hupalowska et al., 2018). This further gives rise to the increasing of H3R26me2 level, activating ICM-specific gene expression, upregulating transposons, and increasing global chromatin accessibility (Wang et al., 2018b). The mechanisms of H3R26me2 deposition and its potential effect on nuclear organization and lineage allocation during early embryo development require further investigations.

Substantially distinct 3D structures in gametes and gradually established chromatin organization during embryogenesis

Recent studies utilizing the low-input Hi-C (genome-wide chromosome conformation capture) method revealed the dynamics of 3D genome architecture reprogramming in mouse early embryo development (Du et al., 2017; Ke et al., 2017) (Fig. 3). Notably, MII oocytes lacks TADs and compartments given its mitotic nature, while earlier stages of oocytes show Polycomb-associating domains (PADs), which is marked by H3K27me3 and briefly reappear on the maternal genome upon fertilization (Du et al., 2020). By contrast, sperm present both topologically associating domains (TADs) and A/B compartments (Du et al., 2017). The sperm genome also shows frequent extra-long-range interactions (> 4 Mb) and inter-chromosomal interactions(Battulin et al., 2015). Following fertilization, the higher-order structures of the chromatin of both parents are obscure at the zygotic and ZGA stages but are spatially separated from each other and show distinct compartmentalization. Such allele separation and compartmentalization are maintained until the 8-cell stage and coincide with accumulations of H3K27me3 ( Du et al., 2017; Borsos et al., 2019; Collombet et al., 2020). The gradual establishment of the parental chromatin organization occurs throughout the development of pre-implantation embryos, with slow consolidation of TADs and the A/B compartments. Notably, the re-establishment of chromatin compartments in the maternal genome appears to be much weaker than that in the paternal genome, which is compatible with the distinct 3D nature of the chromatin of oocytes and sperm to some extent (Du et al., 2017). In addition, the lamina-associated domains (LADs) (van Steensel and Belmont, 2017) of both parental genomes experience de novo but distinct establishment (Borsos et al., 2019), which precedes consolidation of TADs.

Dramatic higher-order chromatin structure reprogramming indicates the importance of epigenetic reorganization during mammalian early development. However, little is known about the molecular basis of the reprogramming of the 3D chromatin structure. The establishment of TADs depends on DNA replication rather than ZGA, which reveals the function of cell cycle transition in genome structure dynamics (Ke et al., 2017). Furthermore, the reprogramming of other epigenetic modifications is dramatically interrelated with the reorganization of the 3D genome during mouse early embryogenesis. Large-scale DNA demethylation, for instance, appears to preferentially occur in compartment A rather than compartment B, resulting in an increase in the accessibility of chromatin to DNA demethylases (Du et al., 2017; Ke et al., 2017).

Moreover, a recent study by our group revealed spatiotemporally dynamic chromatin reorganization in SCNT embryos (Chen et al., 2020). Both aberrant TADs and compartment A/B organization are observed throughout early SCNT embryo development. The overexpression of Kdm4b partially ameliorates abnormal 3D chromatin structures, suggesting that H3K9me3 modification in donor cells is a barrier to chromatin structure reprogramming. This indicates a correlation between the organization of the 3D genome architecture and histone modifications.

No TADs or CTCF-dependent 3D chromatin structure in human sperm

The dynamics of the 3D chromatin architecture during human embryogenesis have recently been revealed. Unlike mouse sperm, human sperm lack TADs and the expression of the chromatin regulator CTCF (Chen et al., 2019a) (Fig. 3). Consistent with findings in mice, TADs and A/B compartmentalization are gradually established during human embryogenesis. CTCF is highly expressed at the ZGA stage of human embryos, which coincides with the time at which TADs are observed, indicating the critical role of CTCF in the establishment of the higher-order chromatin structure during early human embryo development. Intriguingly, the establishment of TADs in human embryos is ZGA dependent, which leads to the hypothesis that the correlation between TAD formation, DNA replication and gene activation differs among species or cell types (Ke et al., 2017; Chen et al., 2019a).