You are on page 1of 17

TICB 1381 No.

of Pages 17

Review

Epigenome in Early Mammalian


Development: Inheritance, Reprogramming
and Establishment
Qianhua Xu1 and Wei Xie1,*

Drastic epigenetic reprogramming takes place during preimplantation devel- Trends


opment, leading to the conversion of terminally differentiated gametes to a Fertilization triggers drastic epige-
nomic reprogramming that converts
totipotent embryo. Deficiencies in remodeling of the epigenomes can cause
fully differentiated gametes to totipo-
severe developmental defects, including embryonic lethality. However, how tent embryos.
chromatin modifications and chromatin organization are reprogrammed upon
Early embryos show noncanonical epi-
fertilization in mammals has long remained elusive. Here, we review recent genomes compared with somatic cells
progress in understanding how the epigenome is dynamically regulated during and embryonic stem cells.
early mammalian development. The latest studies, including many from
Allele- and locus-specific inheritance
genome-wide perspectives, have revealed unusual principles of reprogram- and erasure of parental epigenetic
ming for histone modifications, chromatin accessibility, and 3D chromatin information ensure a successful transi-
architecture. These advances have shed light on the [805_TD$IF]regulatory network con- tion from the parents to the early
embryos.
trolling the earliest development [806_TD$IF]and maternal-zygotic transition.
Subsequent establishment of proper
zygotic epigenome is essential for
Introduction embryogenesis.
The epigenome carries information for cell identities encoded by histone modifications, DNA
Chromatin reprogramming during
methylation, chromatin accessibility, 3D chromatin organization, and small RNAs [1–3]. Epi-
early development includes multilevel
genetic marks, such as DNA methylation and histone modifications, can regulate the stable reorganization of chromatin accessibil-
inheritance of cellular memory during mitotic division [4,5]. However, dramatic epigenomic ity, chromatin modifications, and
reprogramming is required when life begins, that is, when two terminally differentiated gametes higher-order chromatin architecture.
fuse to form a totipotent zygote followed by early embryogenesis. This process effectively
transforms parental epigenomes to the zygotic epigenome, and a failure to do so may cause
early embryonic lethality or disease [6–8]. Thus, gametes and early embryos serve as an ideal
model to study the fundamental principles controlling the reprogramming, inheritance, and re-
establishment of epigenomes. The dynamics of DNA methylation during early mammalian
development has been extensively studied and reviewed elsewhere [9,10]. The global DNA
methylome is largely erased after fertilization except in selected regions, such as genomic
imprints, before it is then re-established in postimplantation embryos [9–11]. However, it is less
clear how other types of parental epigenetic information are reprogrammed and whether some
can be inherited by the next generation in mammals. Moreover, when and how the zygotic
epigenome is subsequently established in early embryos also remains elusive. Deciphering 1
Center for Stem Cell Biology and
these processes is crucial for understanding embryonic development and the molecular basis Regenerative Medicine, MOE Key
Laboratory of Bioinformatics, THU-
for pluripotency and totipotency [12]. Nevertheless, previous efforts in mammals were largely PKU Center for Life Sciences, School
hindered by the scarcity of experimental samples available for early embryos. Thanks to recent of Life Sciences, Tsinghua University,
developments in low-input chromatin analysis technologies (Table 1), [807_TD$IF]research has begun to Beijing 100084, China

illuminate the dynamic chromatin organization during early mammalian development [13–21].
Here, we review recent progress in our understanding of [73_TD$IF]chromatin patterning and its under- *Correspondence:
lying molecular mechanisms [738_TD$IF]in early embryogenesis. xiewei121@tsinghua.edu.cn (W. Xie).

Trends in Cell Biology, Month Year, Vol. xx, No. yy https://doi.org/10.1016/j.tcb.2017.10.008 1


© 2017 Elsevier Ltd. All rights reserved.
TICB 1381 No. of Pages 17

Table 1. Representative Low-Input Genomewide Chromatin Analysis Technologies


Chromatin [72_TD$IF]feature Technology Minimal [723_TD$IF]cell No.c Experimental procedure/features Coverage Refs

Chromatin Single cell ATAC-seq (scATAC-seq) 1 DNaseI + microfluidics Low for [134]
accessibility/ each cell
Single-cell combinatorial indexed 1 DNaseI [725_TD$IF]+ combinatorial indexing [135]
nucleosome
ATAC-seq (sciATAC-seq)
phasing
ATAC-seq Hundred Tn5, high mtDNA contamination High [100]

ATAC-seq + CARM Hundred Tn5 + Cas9-assisted mtDNA depletion [20]b[724_TD$IF]

Fast-ATAC Hundred Tn5, low mtDNA [726_TD$IF][151]

Omni-ATAC Hundred Tn5, low mtDNA, optimized for frozen tissues [136]

scDNase-seq 1 DNaseI Low for [137]


each cell

Low input DNAse-seq 30 DNaseI High [19]b


(liDNAase-seq)

scCOOL-seq/scNOMe-seq 1 GpC methyltransferase Low for [17]b [98]


each cell

Low-input MNase-seq 1000 MNase [79_TD$IF]High [101]b[796_TD$IF]

Histone Drop-ChIP 1 MNase digestion + microfluidics + indexing Low for [798_TD$IF][138]b


modification each cell

STAR ChIP-seq 200 MNase + TELP library preparation High [21,31]b[79_TD$IF]2

mChIP-seq 500 Crosslinking + sonication [13]b

ULI-NChIP 500 MNase [29,140,141]b[730_TD$IF]

iChIP 500 Crosslinking + sonication + indexing [142]

Small-scale ChIP 1000 Crosslinking + sonication + WGA [80_TD$IF][143,144]b

1000-cell ChIP-seq 1000 Crosslinking + sonication + microfluidics [139]b[801_TD$IF]

LinDA ChIP-seq 5000 Crosslinking + sonication + single-tube [145]


linear DNA amplification

MicroChIP 5000 Crosslinking + sonication [30]b[802_TD$IF]

Small-cell-number ChIP-seq 10 000 Crosslinking + sonication [146]

Low cell ChIP 12 500 Crosslinking + sonication [80]b[831_TD$IF]

Native ChIP-seq 20 000 MNase [730_TD$IF][147]

3D chromatin Single cell Hi-C 1 Ligation in nuclei (in situ), can scale Low for [804_TD$IF][119,148]
architecture up to thousands of cells each cell

Single cell Hi-C 1 Ligation in nuclei (in situ) [149]

Single-nucleus HiC (snHi-C) 1 Optimized in situ Hi-C without biotin [16]b


enrichment

Single cell combinatorial indexed 1 Combinatorial indexing, can scale up [150]


Hi-C (sciHi-C) to tens of thousands of cells

Small-scale in situ HiC (sisHi-C) Hundred In situ Hi-C optimized for minimal sample loss Medium [15]b[731_TD$IF]

Optimized low-input in situ Hi-C [731_TD$IF]Hundred In situ Hi-C optimized for minimal sample loss [18]b

a
A list of representative low-input (<20 000 cells) chromatin analysis methods for profiling chromatin accessibility, histone modifications, and 3D chromatin architecture.
For single cell methods, the coverage for each cell is usually low, and the coverage improves when using pooled single cells. [73_TD$IF]Abbreviations: mtDNA, mitochondrial DNA[734_TD$IF];
WGA, whole-genome amplification; TELP, Tailing-Extension-Ligation-PCR.
b
Studies that applied the methods to study germ cells or early embryos[735_TD$IF].
c
Some experiments may start with reactions with more than single cells, before individual cells are isolated or barcoded[736_TD$IF].

2 Trends in Cell Biology, Month Year, Vol. xx, No. yy


TICB 1381 No. of Pages 17

Establishment of [70_TD$IF]the Epigenome in Gametes


Establishing the correct epigenome in germ cells is critical for normal gametogenesis and
embryonic development [9,22,23]. In mice, gamete formation begins with the specification of
primordial germ cells (PGCs) in postimplantation embryos at embryonic day (E)7.25 [24]. PGCs
undergo extensive global DNA demethylation during development, reaching [739_TD$IF]an almost unme-
thylated state around E13.5 [25–27]. The following de novo methylation proceeds differently in
male and female germ cells. In mouse male germ cells, de novo methylation initiates shortly
after E13.5 and is completed before birth [28]. By contrast, methylation in oocytes is acquired
during the follicular growth phase after birth. In [740_TD$IF]the latter case, the DNA methylome is
established in a transcription-dependent manner, resulting in high levels of DNA methylation
in gene bodies [28]. Intergenic regions and nontranscribing gene bodies are poorly methylated,
forming partially methylated domains (PMDs). The asynchronous de novo methylation in female
and male germ cells indicates the highly sex-specific regulation of DNA methylation during
gametogenesis.

Similar to DNA methylation, histone modifications also display distinct patterns between female
and male germ cells. Newly developed low-input ChIP-seq methods (Table 1) were used [741_TD$IF]by
several groups to investigate the genomewide locations of histone modifications in mouse
oocytes [13,21,29–31]. A [742_TD$IF]pioneering study examined histone modifications in primary (E18.5)
and growing oocytes (P10, growing oocyte I, Figure 1), including H3K4me2/H3K4me3 (marks
of active promoters) and H3K36me3 (a mark of active gene bodies), which appear to display
largely canonical patterns similar to other cell types [30]. Surprisingly, histone modifications
adopt highly noncanonical patterns as oogenesis proceeds. For example, at later stages of
oocyte growth (growing oocyte II), H3K4me3 tends to appear as a broad domain at both
promoters and distal sites, including many intergenic regions, forming a noncanonical pattern
of H3K4me3 (ncH3K4me3) (Figure 1). Such ncH3K4me3 becomes widespread in full-grown
oocytes and metaphase II (MII) oocytes, in which the genome transits to a silenced state
[743_TD$IF][13,21,32]. Importantly, several pieces of evidence suggest that H3K4me3 is involved in
genome silencing. Maternal deficiency in KMT2B, the major H3K4me3 histone methyltransfer-
ase, leads to defects in genome silencing in oocytes [33]. The resulting mutant oocytes cannot
develop beyond the four-cell stage after fertilization. Consistently, overexpression of KDM5B
(also known as JARID1B), a histone demethylase for H3K4me3, in silenced full-grown oocytes
resulted in reactivated transcription [21]. How ncH3K4me3 executes such a repressive function
in mouse oocytes remains unknown. Given that H3K4me3 is known to interact with transcrip-
tion regulators, such as TAF3 [34–36], it was proposed that ncH3K4me3 at distal sites function
as ‘sponges’ that attract transcription factors (TFs) and regulators, therefore diluting transcrip-
tion resources away from promoters [21].

The unusual patterning of histone modifications in oocytes is not limited to H3K4me3.


H3K27me3, a repressive histone mark, is also present in a noncanonical form during oogene-
sis. H3K27me3 is typically [74_TD$IF]placed at promoters of developmental genes by the [80_TD$IF]Polycomb
Repressive Complex (PRC) [37,38]. While the PRC2 complex deposits H3K27me3, the PRC1
complex can bind to H3K27me3 and functions as an effector of gene silencing by modulating
chromatin compaction [39,40]. Maternal deficiency in Ring1/Rnf2, two components of PRC1,
led to developmental arrest at the two-cell stage, indicating a crucial role of Polycomb proteins
during mouse early embryonic development [41]. Surprisingly, H3K27me3 in mouse oocytes
shows relative low or no enrichment at promoters of developmental genes. By contrast, large,
broad H3K27me3 domains reside in intergenic regions and gene deserts, therefore also
forming a noncanonical pattern (ncH3K27me3) [31] (Figure 1). During the early stages of
growing oocytes (Figure 1, Growing oocyte I), H3K27me3 is found in most regions without

Trends in Cell Biology, Month Year, Vol. xx, No. yy 3


TICB 1381 No. of Pages 17

Canonical H3K4me3 Noncanonical H3K27me3 H3K36me3


Noncanonical H3K4me3 DNA methylaƟon H3K36me3 (predicted)

Growing oocyte I

Growing oocyte II

Full-grown oocyte

MII oocyte

H3K4me3 only H3K27me3 only H3K4me3/H3K27me3


PMD PMD tandem PMD

Figure 1. E [708_TD$IF] stablishment of the Epigenome during Mouse Oogenesis. A schematic model showing the dynamic
reprogramming of epigenetic marks during mouse oogenesis. In early-stage growing oocytes (growing oocyte I), the
noncanonical form of H3K27me3 is broadly distributed at low levels genome wide in regions lacking transcription. By
contrast, H3K4me3 and H3K36me3 are present largely as canonical forms, occurring as sharp peaks at promoters and
broad domains in transcribing gene bodies, respectively. At later stages of growing oocytes (growing oocyte II), DNA
methylation is established in a transcription-dependent manner, leaving untranscribed regions poorly methylated, forming
partially methylated domains (PMDs). Noncanonical H3K4me3 begins to appear and becomes prevalent in mature
oocytes, including full-grown oocytes and metaphase II (MII) oocytes. Both noncanonical H3K4me3 and H3K27me3 are
broadly deposited in PMDs but preferentially in nonoverlapping subregions, forming H3K4me3-only PMDs, H3K27me3-
only PMDs, or H3K4me3/H3K27me3 tandem PMDs. The state of H3K36me3 is unknown at later stages and is speculated
to be present in active gene bodies (broken yellow lines). GO I (growing oocyte I, with size 30–65 mm), GO II (growing
oocyte [709_TD$IF]II, 50–70 mm), and FGO (full-grown oocyte[710_TD$IF], >70 mm) represent approximately the major oocyte populations
collected at postnatal day 7–10, day 14, and week 8, respectively.

transcription [31]. Towards the later stages of oocyte growth (Figure 1, Growing oocyte I to MII
oocyte), H3K27me3 is gradually restricted to [745_TD$IF]a portion of inactive regions. Analyses comparing
DNA methylation, H3K4me3, and H3K27me3 showed that both distal ncH3K4me3 and
ncH3K27me3 preferentially overlap with oocyte PMDs (which are nontranscribing regions)
in full-grown oocytes and MII oocytes. The mutually exclusive presence is in line with the
antagonism between DNA methylation and H3K4me3 [42,43], or DNA methylation and
H3K27me3 [44–49]. However, ncH3K4me3 and ncH3K27me3 tend to reside in different
subregions within PMDs. This is especially evident in MII oocytes [31] (Figure 1). It is not clear
what determines the deposition of H3K4me3 versus H3K27me3 in these subregions. These
findings [746_TD$IF]raise a question of whether there is a causal relationship between the reprogramming
of DNA methylation and histone modifications. In fact, in growing oocytes, a subset of CpG
islands (CGIs) becomes methylated at a later stage of oocyte maturation, and this is associated
with increased H3K36me3 and reduced H3K4me2/3 [30]. Oocytes deficient in H3K4 deme-
thylase KDM1A or KDM1B showed increases in H3K4me3 and impaired DNA methylation at
CGIs [30]. However, it remains to be determined whether histone modifications regulate
reprogramming of DNA methylation or vice versa outside CGIs and promoters, where
ncH3K4me3 and ncH3K27me3 prevail.

4 Trends in Cell Biology, Month Year, Vol. xx, No. yy


TICB 1381 No. of Pages 17

Reprogramming of histone modifications is also essential for spermatogenesis. For example,


KMT2B is required for both embryonic development and normal spermatogenesis [50].
Similarly, ablation of the H3K27me3 methyltransferases EZH1 and EZH2 led to meiotic arrest
in spermatocytes [51]. Although most histones are replaced by protamines during spermato-
genesis in mammals, modifications are widely detected on [74_TD$IF]residual histones that are retained in
sperm [52–54]. Compared with oocytes, histone modifications in sperm appear to be largely in
canonical forms [52–54] (Figure 2A, Sperm). However, sperm-specific features also exist. For
example, DNA methylation appears to be largely uncorrelated with transcription in human and
mouse spermatocytes or spermatids, and can even coexist with H3K4me3 at a subset of
promoters, as confirmed by ChIP-BisSeq [52]. These ‘atypical’ promoters usually have a low
CG content, are marked by 5hmC but not by H2A.Z, and are highly transcribed. Given that
H3K4me3 is known to antagonize DNA methylation [42,43], how these promoters manage to
bear both marks in the presence of active transcription remains intriguing. Collectively, these
data demonstrate the distinct patterning of histone modifications in sperm [748_TD$IF]and oocyte.

Erasure and Brief Inheritance of Parental Marks in Early Embryos


Given that zygotes initially inherit epigenetic marks from terminally differentiated gametes,
extensive remodeling of the parental epigenome is necessary for subsequent early embryo
development. For instance, H3K9me3 was shown to be a key epigenetic barrier to epigenetic
reprogramming during somatic cell nuclear transfer (SCNT) [55,56]. Similar results were
reported for H3K27me3 in porcine SCNT [57]. Indeed, dynamic changes of histone modifi-
cations in the parental pronuclei after fertilization were widely reported using immunofluores-
cence (IF) in different species, including mouse, rabbit, and bovine [749_TD$IF][58–63]. Notably, many
epigenetic marks on the paternal genome appear to be more rapidly erased compared with
those on the maternal genome, including H3K4me3, H3K9me2/3, and H3K27me3 [58,60,62].
Intriguingly, H3K4me3 and H3K27me3 increase in [750_TD$IF]mouse paternal pronuclei by the end of the
first cell cycle, indicating possible de novo histone methylation [59,62]. These observations
were further corroborated by recent genomewide studies [13,21,29,31]. ChIP-seq data
showed that most sperm-originated H3K4me3 and H3K27me3 peaks are [809_TD$IF]likely lost after
fertilization in mouse embryos. Yet, the paternal genome appears to acquire broad, yet weak
H3K4me3 and H3K27me3 domains [21,31] (Figure 2A, sperm, MII, and zygotes) [810_TD$IF]in gene-rich
regions and gene deserts, respectively [21,31]. The broad paternal domains of H3K4me3 are
removed after zygotic genome activation (ZGA) at the late two-cell stage, whereas those for
H3K27me3 are maintained until the blastocyst stage [Figure 2A, Late two-cell and Inner cell
mass (ICM)]. These results raise an interesting question of whether the same set of maternal
enzymes, which remove canonical H3K4me3 and H3K27me3 and establish their noncanonical
forms in oocytes, act similarly on the sperm genome after fertilization. One major difference
between maternal and paternal noncanonical histone modifications is that the maternal
H3K4me3 and H3K27me3 largely fall into oocyte PMDs, while those on the paternal genome
are much larger and broader. This coincides with the presence of oocyte PMDs on the maternal
genome, while the paternal genome undergoes extensive DNA demethylation after fertilization.
Whether these paternal broad domains of H3K4me3 and H3K27me3 are functional is unclear.
Interestingly, loss of H3K4 methylation was shown to affect minor ZGA predominantly in the
paternal pronucleus, although the effect was mainly attributed to H3K4me1 (instead of
H3K4me3) [64]. PRC1 is required to repress pericentric major satellite transcription in the
paternal genome. Nevertheless, PRC1 binding appears to precede the appearance of
H3K27me3 [65]. Future studies are needed to fully elucidate the significance of such paternal
H3K4me3/H3K27me3 domains.

Trends in Cell Biology, Month Year, Vol. xx, No. yy 5


TICB 1381 No. of Pages 17

ChromaƟn Higher-order
(A) Histone modificaƟon (B) (C)
accessibility chromaƟn structure

TAD (priming state)


H3K4me3 Open chromaƟn
TAD (mature state)
H3K27me3 Open chromaƟn
Closed chromaƟn
Weak compartmentalizaƟon

Strong compartmentalizaƟon
MII
oocyte

Sperm
E TSS TES

Zygote

Early
two-cell MERVL

Late
two-cell E TSS TES

Eight-cell
E TSS TES

ICM
E TSS TES

Epiblast

Developmental
gene

Figure 2. R [795_TD$IF] eprogramming of the Epigenome during Mouse Preimplantation Development. A schematic model illustrating the reprogramming of the
epigenome during mouse preimplantation development. (A) Reprogramming of histone modifications. During gametogenesis, distinct epigenomes are established in
female and male gametes. H3K4me3 and H3K27me3 are largely present in canonical forms in sperm but exist in noncanonical forms in oocytes. After fertilization,
maternal noncanonical H3K4me3 from the oocyte is inherited by early embryos, before it is removed upon the onset of zygotic genome activation (ZGA) at the late two-
cell stage. Oocyte ncH3K27me3 at the promoters of developmental genes is specifically erased after fertilization, while ncH3K27me3 at distal sites is maintained to as
late as the inner cell mass (ICM). By contrast, sperm H3K4me3 and H3K27me3, which are largely in canonical form, are likely to be rapidly removed after fertilization,
followed by re-establishment of very broad domains at low levels genome wide. Such de novo paternal domains are briefly maintained during early development before

(See figure legend on the bottom of the next page.)

6 Trends in Cell Biology, Month Year, Vol. xx, No. yy


TICB 1381 No. of Pages 17

By contrast, maternal ncH3K4me3 is inherited by zygotes and is maintained before zygotic


genome activation (Figure 2A, MII to early [23_TD$IF]two-cell) [21]. Similar to [752_TD$IF]broad paternal ncH3K4me3,
maternal ncH3K4me3 is also dramatically reduced in late-two cell embryos (Figure 2A, Late
two-cell), and is almost completely erased by the four-cell stage [13,21]. Meanwhile, canonical
H3K4me3 appears at promoters after ZGA (Figure 2A, Late two-cell to ICM) [21]. Such
transition appears to depend on transcription but not on DNA replication [21], suggesting
that the erasure of ncH3K4me3 involves active demethylation. Indeed, H3K4 demethylases
KDM5B and KDM5A (known as Jarid1a) are highly induced during ZGA at the late two-cell
stage. Knocking down both demethylases resulted in a high level of H3K4me3 in the late two-
cell embryos, which showed defective ZGA and failure to reach the blastocyst stage [13].
Genetic ablation of Kdm5b [81_TD$IF]resulted in early embryonic lethality, whereas Kdm5a-depleted mice
were viable and fertile [66]. Furthermore, depletion of Kdm5b also impaired porcine preimplan-
tation development [67]. Interestingly, KDM5B can demethylate H3K4me3 in gene bodies in
both mouse embryonic stem cells (mESCs) [68,69] and early embryos [29]. It is possible that
KDM5B also demethylates intergenic ncH3K4me3 when it is highly induced at the two-cell
stage. Collectively, these data indicate that KDM5B has an important role in regulating
[812_TD$IF]H3K4me3 reprogramming and early development [13,29,66].

Reprogramming of H3K27me3 on the maternal allele adopts a distinct pattern compared with
that of H3K4me3. After fertilization, maternal H3K27me3 is specifically removed from the
promoters of developmental genes (Figure 2A, MII and Zygote). The absence of promoter
H3K27me3 persists for most of preimplantation development, before weak H3K27me3 starts
to appear at canonical Polycomb target promoters in the blastocyst (Figure 2A, Zygote to ICM).
Strong promoter H3K27me3 at developmental genes is readily found in postimplantation
epiblasts [31] (Figure 2A, Epiblast). Notably, these data are in line with reduced H3K27me3
at developmental gene promoters in naïve mESCs [70], the in vitro counterpart of ICM.
Interestingly, most developmental genes remain silenced despite the absence of
H3K27me3 in preimplantation embryos, suggesting either the presence of additional repres-
sive mechanisms or the absence of activators. Currently, it is not clear why these promoters are
specifically devoid of H3K27me3 in these early stages. By contrast, maternal H3K27me3 in
distal regions is inherited by zygotes and persists through preimplantation development,
although the global level is decreased at the blastocyst stage [29,31] (Figure 2A, MII to
ICM). Such distal H3K27me3 is no longer detected in postimplantation embryos, which
may be due to the passive loss caused by rapid cell division during peri- and postimplantation
development (Figure 2A, Epiblast). Importantly, a recent study [753_TD$IF]showed that H3K27me3 on the
maternal genome regulates allele-specific gene expression through a DNA methylation-inde-
pendent manner [71]. While allelic expression of these genes is only transient, with biallelic

they are erased at either the late two-cell (H3K4me3) or epiblast (H3K27me3) stage. At promoters of developmental genes, H3K4me3/H3K27me3 bivalent marks are
largely absent after fertilization and are re-established in the epiblast. (B) Reprogramming of chromatin accessibility. Open chromatin exists at promoters and putative
regulatory regions in sperm [97]. In zygotes, a few thousand loci are preferentially accessible, with some specific for each parental genome [71]. At the early two-cell
stage, strong open chromatin occurs at the transposable element MERVL. After ZGA, accessible chromatin is mainly found at promoters and putative enhancers. It also
occurs at transcription end sites (TESs) from the late two-cell to eight-cell stages [20]. [71_TD$IF]Abbreviations: E, enhancer; TSS, transcription start site; TES, transcription end
site. (C) Reprogramming of higher-order chromatin architecture. [712_TD$IF]Metaphase II (MII) oocytes exhibit uniform chromatin configuration, lacking [713_TD$IF]TADs and compartments.
By contrast, [714_TD$IF]TADs and compartments exist in mature sperm. TADs are weak in zygotes, existing in a priming state ([715_TD$IF]unfilled orange circles). They become consolidated
and mature at late stages ([716_TD$IF]filled orange circles). Long-distance chromatin interactions and chromatin compaction also increase at late developmental stages (indicated
by shortened horizontal lengths). Chromatin compartments (A/B) are present as early as in zygotes, preferentially on the paternal genome. As development proceeds,
chromatin compartments become increasingly segregated on both alleles. Open and closed chromatin are indicated as green and brown loops, respectively. Weak and
strong chromatin compartmentalization are represented by the degree of [71_TD$IF]striped and filled background respectively.

Trends in Cell Biology, Month Year, Vol. xx, No. yy 7


TICB 1381 No. of Pages 17

expression observed in later embryonic lineages, at least a few genes retain allele-specific
expression in extraembryonic tissues. This coincides with a study in flies, which showed that
maternal H3K27me3 can restrict the functions of enhancers in early embryos [72]. In sum,
these data indicate that the majority of studied parental histone marks are ultimately erased in
an allele- and mark-specific manner during preimplantation development. However, some
marks, many of which are from the maternal genome, are briefly inherited in embryos and can
have important functions during early development. It remains to be determined whether any
parental histone modifications can be inherited by somatic cells.

Establishment of the Zygotic Epigenome


Once parental marks are largely erased, the zygotic epigenome needs to be established to
accommodate the cleaving embryos. For instance, upon the onset of ZGA, canonical
H3K4me3 starts to form at promoters [13,21,29] (Figure 2A, Late two-cell). Consequently,
H3K4me3 becomes largely symmetric between the two alleles after ZGA [21]. Consistent with
the close relationship between histone acetylation and transcription, H3K27ac also appears at
promoters and putative enhancers near ZGA genes upon genome activation [13,20]. Gao and
colleagues reported a high frequency of broad promoter H3K4me3 in early mouse embryos
after ZGA at genes associated with high transcription and cell identity [29]. [754_TD$IF]Notably, the broad
promoter H3K4me3 is distinct from ncH3K4me3 observed in oocytes, which occurs before
ZGA at both promoters and distal regions [75_TD$IF](including gene deserts[756_TD$IF]) and is associated with
genome silencing. [754_TD$IF]Instead, the broad domains of promoter H3K4me3 after ZGA echo those
previously found in other cell types at genes important for cell identity [73,74] or tumor
suppressor genes [75]. In these cases, broad H3K4me3 colocalizes with superenhancers
and has strong enhancer activities [74,75]. Shortening of these H3K4me3 domains in cancer
cells is correlated with the repression of tumor suppressor [75]. By contrast, knockdown of
Kdm5b widens the H3K4me3 domain in mouse early embryos, which is associated with
impaired blastocyst formation [29]. Currently, it is not clear whether the phenotype of Kdm5a/b
knockdown is related to the failure of ncH3K4me3 removal, to the establishment of promoter
broad H3K4me3, or both. Nevertheless, these data indicate that the reprogramming of
H3K4me3 has an important role during early embryonic development [13,29].

Notably, developmental gene promoters preferentially harbor bivalent marks with both active
H3K4me3 and repressive H3K27me3 [37]. Bivalent marks are considered to [813_TD$IF]poise the expres-
sion of developmental genes, permitting timely activation upon developmental cues [37,76–
79]. Bivalent marks are broadly present in many cell types, including PGCs and sperm
[8,52,80]. The fact that they are present in sperm has attracted significant interest, because
it [75_TD$IF]raises the question of whether these marks can be inherited by early embryos and regulate the
activation of key developmental regulators (Figure 2A, sperm). In MII oocytes, H3K27me3 is
weakly enriched at promoters of developmental genes, where H3K4me3 is absent or present at
low levels [31] (Figure 2A, MII). However, sperm-originated bivalent marks [814_TD$IF]seem to be rapidly
removed from these promoters after fertilization and do not reappear until the postimplantation
stage [31] (Figure 2A, Zygote to Epiblast). Such global ‘erase-and-rewrite’ mechanisms may be
more efficient than locus-specific modification for large-scale reprogramming, [815_TD$IF]especially con-
sidering that more than half of the Polycomb targets in gametes differ from those in epiblasts
[31]. These data also [816_TD$IF]raise the possibility that bivalent marks may primarily function during
spermatogenesis[817_TD$IF]. In addition, an intriguing question is how bivalent mark establishment is
instructed in postimplantation embryos. CpG islands are known to attract Polycomb and the
trithorax complex to establish H3K[81_TD$IF]27me3 and H3K4me3, respectively [48,81,82]. However,
other factors, such as cell type-specific TFs, are 819_TD$IF]l[ ikely required to modulate locus specificity,

8 Trends in Cell Biology, Month Year, Vol. xx, No. yy


TICB 1381 No. of Pages 17

because not all CGIs are modified. Alternatively, other histone marks or factors may exist at
these promoters to direct the re-establishment of bivalent marks. Notably, the appearance of
bivalent domains coincides with the [758_TD$IF]timepoint at which ICM transits from the pluripotent ground
state to lineage specification. These data suggest a possible role of bivalent marks in pluri-
potency exit and lineage specification [83,84].

Aside from the canonical histones H2A, H2B, H3, and H4, histone variants endow chromatin
with special functions in a locus-specific manner. The roles of histone variants in development
have been well reviewed elsewhere [85]. Notably, mounting evidence has revealed a critical role
of the H3.3 variant in chromatin remodeling during the transition from oocytes to embryos
[86–90]. As a sperm enters an oocyte, protamines are swiftly replaced by maternal histones
before DNA replication [91,92]. H3.3, rather than H3.1/H3.2, is incorporated in the paternal
chromatin. In the absence of H3.3 or its chaperone Hira, canonical histones failed to be
incorporated into the nucleosome [87,93]. Maternal depletion of Hira results in a failure [759_TD$IF]of
paternal genome chromatinization, DNA replication, and [760_TD$IF]development beyond the zygote
stage. Notably, Hira is also required for developmental competence of the female genome.
This is attributed to the Hira/H3.3-dependent transcription of rRNA, which is essential for the
first cleavage via ribosome assembly [93]. These data strongly support a critical role of H3.3 in
embryonic development and chromatin remodeling after fertilization. The functions of other
histone variants in early development await further investigation.

Reprogramming of Chromatin Accessibility


Accessible chromatin can expose key regulatory DNA sequences to transcription regulators,
allowing gene activation [94]. Therefore, proper regulation of chromatin accessibility is essential
for transcription and development [95]. Low-input or single cell assays to interrogate chromatin
accessibility have been developed over the past few years, including DNase-seq, assay for
transposase accessible chromatin followed by high-throughput sequencing (ATAC-seq) and
nucleosome occupancy and methylome sequencing (NOMe-seq) (Table 1). Using these
methods, several recent studies explored the dynamics of chromatin accessibility in mouse
gametes and/or early embryos [17,19,20,71,96,97]. These studies revealed highly dynamic
chromatin landscapes during early development (Figure 2B, Zygote to ICM). Chromatin of
sperm and MII oocytes is often difficult to interrogate due to its compact nature. However,
Corces and colleagues successfully performed ATAC-seq on [820_TD$IF]mouse sperm [97]. Intriguingly,
accessible chromatin in sperm shares many features with ESCs or somatic cells, residing
preferentially in promoters, putative enhancers, and CTCF-binding sites [97]. Using DNase-
seq, a few thousand open loci were detected in pronuclei at the one-cell stage, including those
that were specific for each parental genome [71]. Larger numbers of accessible loci were
detected in embryos after ZGA, when chromatin accessibility is approximately symmetrical
between the two alleles [19,20] (Figure 2B, Late two-cell). Notably, ATAC-seq appears to
detect [821_TD$IF]more open chromatin regions, especially in distal sites, than does DNase-seq in two-cell
and four-cell embryos [19,20], indicating these two approaches may have different prefer-
ences, particularly in early-stage embryos. Analysis of these chromatin accessibility data
identified putative cis-regulatory elements in the genomes of early embryos. These elements
revealed the transcription network of early development that includes TFs [82_TD$IF]regulating ZGA (such
as NFYA) [19] or lineage circuitry (such as NR5A2) [20]. A third strategy to interrogate genome-
wide chromatin accessibility [823_TD$IF]includes single cell multiomics sequencing technology (scCOOL-
seq) [17] [762_TD$IF]and scNOMe-seq [98]. By adapting NOME-seq [99], these methods use GpC
methyltransferase to probe chromatin accessibility. Interestingly, Tang and colleagues reported
a more dynamic chromatin state, with chromatin accessibility increasing from gametes to

Trends in Cell Biology, Month Year, Vol. xx, No. yy 9


TICB 1381 No. of Pages 17

zygotes, followed by a decrease after the late zygote stage, before increasing again in the four-
cell embryo [17]. Notably, both NOMe-seq and ATAC-seq can also provide information about
nucleosome positioning or phasing [763_TD$IF][17,99,100]. Such information can be similarly investigated
by MNase-seq, as recently demonstrated in porcine SCNT and zygote embryos [101]. Thus,
ATAC-seq, DNase-seq, scCOOL-seq/scNOMe-seq, and MNase-seq interrogate chromatin
states from different angles. Integrating these data sets may provide valuable insights for
investigating chromatin states during early development.

Interestingly, these chromatin maps also revealed embryo-specific chromatin features. For
example, MERVL is a class of retrotransposable elements that are highly active during early
mouse development [102]. Promiscuous transcription of MERVL was shown to occur in the
mouse minor ZGA, which proceeds far downstream of these elements [102–104]. Such
transcription is associated with strong and broad ATAC-seq enrichment [824_TD$IF][20] (Figure 2B, Early
two-cell). These data indicate that chromatin before major ZGA is likely in a permissive
chromatin state, at least in regions near MERVL [20,103,105]. The relaxed chromatin state
is also supported by high histone mobility [106] and dispersed chromatin revealed by electron
spectroscopic imaging [107]. Recent high-throughput chromosome conformation capture (Hi-
C) analyses of early embryos also support a highly relaxed chromatin state after fertilization
[15,18] (see below). The exact mechanisms underlying such permissive chromatin in early
embryos are not fully understood. However, several factors have been shown or implicated to
be involved in this process, including chromatin assembly factor CAF-1 [108], transcription
factor DUX4 [109–111], and retrotransposons, such as LINE1 [96]. Notably, timely activation
and repression of LINE1 are important for both decompaction and subsequent recompaction
of chromatin, respectively, which regulate the developmental potential of embryos [96]. Taken
together, these data suggest that dynamic regulation of chromatin accessibility is essential for
early development.

Reprogramming of High-Order Chromatin Structure


In eukaryotic cells, DNA is packaged into a highly organized 3D structure [2,112]. Studies of
higher-order chromatin structure [764_TD$IF]using high-throughput approaches, such as 5C-seq or Hi-C,
revealed that large stretches of DNA are often packaged into self-interacting units, or topologi-
cally associating domains (TADs) [113,114]. TADs [765_TD$IF]are suggested to provide insulation for cis-
regulatory elements to restrain their functions within local regions [113,115,116]. Hi-C analyses
also showed that the genome is partitioned into two types of spatially segregated [76_TD$IF]regions,
namely compartments A and B, which largely comprise active and inactive genomic loci,
respectively [117]. Such regions tend to interact with other loci within the same compartment
class but not with those from different compartment class. Interestingly, TADs and chromatin
compartments A/B are restricted to interphase chromatin, because they are both absent in
mitotic cells [118]. Instead, mitotic chromatin adopts a configuration that is consistent with a
linearly compressed array of consecutive chromatin loops. A recent study using single cell Hi-C
revealed even more dynamic chromatin organization during the cell cycle [119].

Using various versions of Hi-C (Table 1), several studies revealed distinct chromatin organiza-
tions in gametes [15,16,18,97,120]. These [76_TD$IF]work showed that TADs and compartments in
sperm appear to be largely similar to those in other mammalian cells, such as mESCs or
somatic cells [97,120] (Figure 2C, Sperm). This is consistent with the notion that structural
proteins, such as CTCF and cohesins, are still bound on sperm chromatin [768_TD$IF][97,121]. Never-
theless, the sperm genome has more long-range contacts and interchromosomal contacts
than do somatic cells [18,120]. Several histone marks, such as H3K27me3, showed different

10 Trends in Cell Biology, Month Year, Vol. xx, No. yy


TICB 1381 No. of Pages 17

compartment enrichment compared with mESCs [97]. Despite the presence of these sperm-
specific features, the higher-order chromatin structure of sperm largely resembles that of ESCs
or somatic cells. Surprisingly, current Hi-C data do not appear to identify structure patterns that
may correlate with the differences between histone and protamine packaging [122,123]. This
might be due to the relatively low resolution of Hi-C assays for more [769_TD$IF]refined chromatin
structures. High-resolution chromatin-mapping approaches, such as microC [124] or ChIP-
PET/Hi-ChIP/PLAC-seq [125–127], may be necessary to delineate the fine-scale chromatin
structure of sperm.

A recent study also examined the 3D chromatin structure in mouse GV oocytes and zygotes
using single-nucleus Hi-C [16]. At the antral follicle stage, GV oocytes can be divided into two
classes: surrounded nucleolus (SN) oocytes, in which chromatin forms a rim surrounding the
nucleolus, and nonsurrounded nucleolus (NSN) oocytes [32]. During the later stage of oocyte
growth, NSN oocytes transit into a SN state, which is accompanied by genomewide tran-
scription silencing and chromatin compaction [128]. Such a transition was suggested to be
important for acquiring meiotic competence [128]. Loops and TADs are present in GV oocytes,
but vary between individuals [16]. Compared with NSN oocytes, mature SN oocytes display
more long-range (>400 kb) contacts, but with reduced strength for loops, TADs, and compart-
ments, which may be related to transcription silencing and the detachment of chromatin from
the nuclear envelope [129,130]. Recently, the reprogramming of higher-order chromatin
structures in MII oocytes and preimplantation embryos was recently reported [15,18]. Inter-
estingly, TADs and chromatin compartments are not observed in MII oocytes, a feature that is
likely due to the fact that they are arrested at metaphase [118] (Figure 2C, MII). These data may
also indicate that typical higher-order chromatin structures, at least from the maternal genome,
may not be passed on to the next generation [118]. Surprisingly, the re-establishment of TADs
after fertilization appears to be slow compared with that of the regular mitotic cycle in somatic
cells [15,18] (Figure 2C), indicating an absence of factors for establishing TADs or the presence
of inhibitory factors. During development, the consolidation of TAD proceeds with growing
kinetics during each cell cycle. This occurs even in the presence of transcription inhibitor alpha-
amanitin [15,18], although global chromatin interaction patterns are still somewhat altered (Du
and Xie, unpublished results[70_TD$IF]). This is consistent with a study in flies, which showed that TADs
are weak before zygotic genome activation [131]. The genome-wide establishment of TADs
coincides with zygotic genome activation. Notably[71_TD$IF], in Drosophila, TAD boundary insulation
appears to be independent of transcription but requires Zelda, a key transcription activator of
the zygotic genome[72_TD$IF], for locus-specific boundary insulation [131,132]. However, transcription
inhibition results in the reduction of intra-TAD interactions and an increase in inter-TAD
interactions [131]. A recent study in fly Kc167 cells demonstrated an even stronger effect
on chromatin domains by transcription inhibition and suggested that interaction domains are
related to RNA polymerase (Pol)-II binding in a dose-dependent manner [133]. Thus, these data
revealed a complex relationship between 3D chromatin organization and global transcription.
However, whether such a relationship is present in mammalian cells, including early mouse
embryos, remains to be [825_TD$IF]further examined. For example, it would be necessary to determine the
extent to which Pol II binding is affected in mouse early embryos treated with alpha-amanitin
[15,18].

Similar to DNA methylation and histone modifications, the 3D chromatin architecture also
undergoes allele-specific reprogramming. In zygotes, chromatin compartments appear to be
absent or weak on the maternal genome, but are evident on the paternal genome [15,16,18]
(Figure 2C, Zygote). This correlates with the lack of chromatin compartmentalization in MII
oocytes and the presence of strong compartments in sperm. Differential chromatin structures

Trends in Cell Biology, Month Year, Vol. xx, No. yy 11


TICB 1381 No. of Pages 17

between the two parental genomes can be found as late as the eight-cell stage [15] (Figure 2C). Outstanding Questions
Taken together, these data indicate that the chromatin structure becomes highly relaxed after [781_TD$IF]What is the biological significance 782_TD$IF]o
[ f
fertilization, followed by slow chromatin compaction during early development. These results parent-specific reprogramming of
epigenomes?
are consistent with a study using electron spectroscopic imaging [107], which showed that
chromatin in one-cell embryos is highly dispersed. Chromatin becomes less uniformly dis- [783_TD$IF]During gametogenesis and early
persed and tends to organize into large compact domains at the two-cell stage, with an development, what are the key events
increased concentration at the nuclear envelope. Such a relaxed chromatin organization may that initiate the ‘[784_TD$IF]erasing’ and ‘rewriting’
of the epigenomes at each stage[785_TD$IF]?
also contribute to promiscuous transcription and broad open chromatin domains near MERVL
transposable elements before ZGA [20,103,105]. One critical question is what guides chro-
How is zygotic epigenome estab-
matin folding during early development after fertilization. Although sperm carry chromatin with lished? Are there parental epigenetic
conventional structures, such as TADs and chromatin compartments, it is unlikely that the memories that instruct its
paternal genome helps direct the refolding of maternal genome, because these two sets of establishment?
chromosomes are [826_TD$IF]partially separated for an extended period even after fusion of the pronuclei
What is the 786_TD$IF]r[ ole of epigenetic reprog-
in zygotes [15]. Given that TADs and chromatin compartments are absent in mitosis but can
ramming in zygotic genome
rapidly reappear afterwards [118,119], refolding of such chromatin architecture may not rely on activation?
memory from previous chromatin structure. [827_TD$IF]However, it is possible that such refolding is guided
by persisting [82_TD$IF]architectural proteins. It remains to be determined whether other factors, such as [78_TD$IF]What is the molecular base of the
histone modifications, have roles in this process. transgenerational inheritance of epige-
netic information?

Concluding Remarks
The development of low-input epigenetic analysis technologies has greatly advanced our
understanding of epigenetic reprogramming during early development. Based on the results
of the latest studies on DNA methylation, histone modifications, chromatin accessibility, and
higher-order chromatin structure, a general theme is beginning to emerge. This includes the
extensive erasure of a large portion of parental epigenetic marks, including DNA methylation
[74_TD$IF]and histone modifications, followed by re-establishment of the zygotic epigenome. Interest-
ingly, the relaxation of the chromatin structure after fertilization that coincides with these
changes is also consistent with the ‘erasure’ of gametic chromatin organization (Figure 3).
The removal of parental epigenetic marks indicates that the epigenomes of gametes may be[75_TD$IF], at
least for a significant part, a product of gametogenesis. Such an ‘erase-and-rewrite’ approach
may efficiently convert gametes into totipotent embryos (Figure 3). In addition, it may help
prevent accidental inheritance of spontaneously acquired epigenetic marks by the next gener-
ation, some of which could be detrimental for progeny survival. By contrast, certain epigenetic
marks are inherited by early embryos, as exemplified by genomic imprints. Histone modifica-
tions can be briefly inherited by embryos in a genome-wide manner, as with maternal H3K4me3
and H3K27me3 (Figure 3). Importantly, such inherited marks can be functional [13,21,33,71]. It
remains to be determined whether a small subset of parental histone modifications may be
inherited to a later stage, such as adult tissues in normal or diseased cells. Finally, how the
dynamic reprogramming of higher-order chromatin structures during preimplantation devel-
opment might [76_TD$IF]regulate the reprogramming of DNA methylation, histone modifications, and
chromatin accessibility, or vice versa[7_TD$IF], awaits further investigation. Future studies are needed to
fully integrate multilayer information to decipher the molecular mechanisms underlying the
reprogramming of the epigenomes and their functions in early development [78_TD$IF](see Outstanding
Questions). Understanding such mechanisms will [79_TD$IF]also shed light on the fundamental principles
underlying [780_TD$IF]cell fate conversion and facilitate the generation of induced pluripotent cells for
therapeutic purposes.

12 Trends in Cell Biology, Month Year, Vol. xx, No. yy


TICB 1381 No. of Pages 17

Gametogenesis PreimplantaƟon Post-


implantaƟon

GO I GO II FGO MII

Zygote Early Late 8-cell Blastocyst Epiblast


Round Elongated 2-cell 2-cell
Spermatocyte SpermaƟd Sperm
H3K4me3 ncH3K4me3 H3K4me3
Bivalency K27me3 K4me3
Histone modificaƟon

ncH3K27me3 H3K27me3

No bivalency Bivalency
Weak bivalency
chromaƟn

MERVL LINE 1 TES


Open
structure

No TAD/C Priming TAD/Compartment (C) Mature TAD/C


3D

TAD/C
compacƟon
ChromaƟn

Compacted Relaxed Compacted

Canonical paƩern (somaƟc cell etc. ) Non-canonical paƩern

Figure 3. E [83_TD$IF]71920 pigenetic Reprogramming during the Transition from Gametes to Embryos in Mice. An outline of gametogenesis and early embryonic
development is represented schematically in the upper lane. For lower panels, canonical patterns of all epigenetic features are in gray. Other colors represent
noncanonical patterns, with the color intensity reflecting their extent. Thick broken lines indicate unknown patterns. During gametogenesis, both H3K4me3 and
H3K27me3 exist as relatively canonical patterns in the paternal genome. In the maternal genome, canonical H3K4me3 occurs at the initial phase of gametogenesis
(Growing oocyte I, GO I) but switches to noncanonical H3K4me3 (ncH3K4me3) at later stages of oogenesis. H3K27me3 appears in a noncanonical form in the maternal
genome as early as in growing oocytes. After fertilization, maternal ncH3K4me3 and ncH3K27me3 are briefly inherited by early embryos, while paternal H3K4me3 and
H3K27me3 are quickly replaced by noncanonical forms after fertilization. Both noncanonical forms of H3K4me3 and H3K27me3 are then converted back to canonical
forms at either the late two-cell (ncH3K4me3) or epiblast (ncH3K27me3) stage. Strong H3K4me3/H3K27me3 bivalent promoters are present during spermatogenesis
but are relatively weaker during oogenesis. Bivalency is absent during preimplantation development and is re-established in epiblasts. Accessible chromatin in sperm
preferentially occurs as a canonical pattern at promoters and putative regulatory elements. Accessible chromatin during minor zygotic genome activation (ZGA) (early
two-cell) is preferentially found near transcribing MERVL. LINE-1 expression initiates at the zygotic stage, peaks at the two-cell stage, and regulates chromatin
accessibility during developmental stages. From the two-cell to eight-cell stage, open chromatin is found not only at promoters and putative enhancers, but also near
transcription end sites (TESs). For higher-order chromatin structure, topologically associating domains (TADs) and compartments are observed in sperm but not in MII
oocytes. TADs and compartments exist in a priming state after fertilization and become more mature from the eight-cell stage onward. Compartments appear earlier on
the paternal genome than on the maternal genome in zygotes. Overall, chromatin likely exists in a relatively compact state in sperm and MII oocyte, and becomes more
relaxed or permissive after fertilization, before becoming compact again during early development. GO I (with size 30–65 mm), GO II (50–70 mm) and FGO (full-grown
oocytes, >70 mm) represent approximately the major oocyte populations collected at postnatal day 7–10, day 14, and week 8, respectively. Abbreviation: mESC,
mouse embryonic stem cells.

Trends in Cell Biology, Month Year, Vol. xx, No. yy 13


TICB 1381 No. of Pages 17

[78_TD$IF]Acknowledgments
We thank members of the Xie laboratory for [789_TD$IF]comments and Yunlong Xiang, Hui Zheng, Bofeng Liu, Bingjie Zhang, and
Jingyi Wu for help with the table and figures. This review included only selected studies as an illustration of the recent
progress in understanding chromatin reprogramming in early development; we apologize to all those authors whose work
was not cited because of space limitations. This work was supported by the National Key R&D Program of China
(2016YFC0900301), the National Basic Research Program of China (2015CB856201), the National Natural Science
Foundation of China (31422031), the THU-PKU Center for Life Sciences, and Beijing Advanced Innovation Center for
Structural Biology.

References
1. Bernstein, B.E. et al. (2007) The mammalian epigenome. Cell 23. Leitch, H.G. et al. (2013) Primordial germ-cell development and
128, 669–681 epigenetic reprogramming in mammals. Curr. Top. Dev. Biol.
2. Gibcus, J.H. and Dekker, J. (2013) The hierarchy of the 3D 104, 149–187
genome. Mol. Cell 49, 773–782 24. Matsui, Y. and Okamura, D. (2005) Mechanisms of germ-cell
3. Rivera, C.M. and Ren, B. (2013) Mapping human epigenomes. specification in mouse embryos. Bioessays 27, 136–143
Cell 155, 39–55 25. Guibert, S. et al. (2012) Global profiling of DNA methylation
4. Bird, A. (2002) DNA methylation patterns and epigenetic mem- erasure in mouse primordial germ cells. Genome Res. 22,
ory. Genes Dev. 16, 6–21 633–641

5. Kouzarides, T. (2007) Chromatin modifications and their func- 26. Kobayashi, H. et al. (2013) High-resolution DNA methylome
tion. Cell 128, 693–705 analysis of primordial germ cells identifies gender-specific
reprogramming in mice. Genome Res. 23, 616–627
6. Dean, W. et al. (2005) DNA methylation in mammalian develop-
ment and disease. Birth Defects Res. C Embryo Today 75, 98–111 27. Seisenberger, S. et al. (2012) The dynamics of genome-wide
DNA methylation reprogramming in mouse primordial germ
7. Marcho, C. et al. (2015) Epigenetic dynamics during preimplan-
cells. Mol. Cell 48, 849–862
tation development. Reproduction 150, R109–R120
28. Stewart, K.R. et al. (2016) Establishment and functions of DNA
8. Vastenhouw, N.L. and Schier, A.F. (2012) Bivalent histone mod-
methylation in the germline. Epigenomics 8, 1399–1413
ifications in early embryogenesis. Curr. Opin. Cell Biol. 24,
374–386 29. Liu, X. et al. (2016) Distinct features of H3K4me3 and
H3K27me3 chromatin domains in pre-implantation embryos.
9. Lee, H.J. et al. (2014) Reprogramming the methylome: erasing
Nature 537, 558–562
memory and creating diversity. Cell Stem Cell 14, 710–719
30. Stewart, K.R. et al. (2015) Dynamic changes in histone mod-
10. Smith, Z.D. and Meissner, A. (2013) DNA methylation: roles in
ifications precede de novo DNA methylation in oocytes. Genes
mammalian development. Nat. Rev. Genet. 14, 204–220
Dev. 29, 2449–2462
11. Bartolomei, M.S. and Ferguson-Smith, A.C. (2011) Mammalian
31. Zheng, H. et al. (2016) Resetting epigenetic memory by reprog-
genomic imprinting. Cold Spring Harb. Perspect. Biol. 3,
ramming of histone modifications in mammals. Mol. Cell 63,
a002592
1066–1079
12. Paranjpe, S.S. and Veenstra, G.J. (2015) Establishing pluripo-
32. Zuccotti, M. et al. (1995) Chromatin organization during mouse
tency in early development. Biochim. Biophys. Acta 1849,
oocyte growth. Mol. Reprod. Dev. 41, 479–485
626–636
33. Andreu-Vieyra, C.V. et al. (2010) MLL2 is required in oocytes for
13. Dahl, J.A. et al. (2016) Broad histone H3K4me3 domains in
bulk histone 3 lysine 4 trimethylation and transcriptional silenc-
mouse oocytes modulate maternal-to-zygotic transition. Nature
ing. PLoS Biol. 8, e1000453
537, 548–552
34. Lauberth, S.M. et al. (2013) H3K4me3 interactions with TAF3
14. Dirks, R.A. et al. (2016) Genome-wide epigenomic profiling for
regulate preinitiation complex assembly and selective gene acti-
biomarker discovery. Clin. Epigenetics 8, 122
vation. Cell 152, 1021–1036
15. Du, Z. et al. (2017) Allelic reprogramming of 3D chromatin
35. van Ingen, H. et al. (2008) Structural insight into the recognition
architecture during early mammalian development. Nature
of the H3K4me3 mark by the TFIID subunit TAF3. Structure 16,
547, 232–235
1245–1256
16. Flyamer, I.M. et al. (2017) Single-nucleus Hi-C reveals unique
36. Vermeulen, M. et al. (2007) Selective anchoring of TFIID to
chromatin reorganization at oocyte-to-zygote transition. Nature
nucleosomes by trimethylation of histone H3 lysine 4. Cell
544, 110–114
131, 58–69
17. Guo, F. et al. (2017) Single-cell multi-omics sequencing of
37. Bernstein, B.E. et al. (2006) A bivalent chromatin structure
mouse early embryos and embryonic stem cells. Cell Res.
marks key developmental genes in embryonic stem cells. Cell
27, 967–988
125, 315–326
18. Ke, Y. et al. (2017) 3D Chromatin structures of mature gametes
38. Margueron, R. and Reinberg, D. (2011) The Polycomb complex
and structural reprogramming during mammalian embryogene-
PRC2 and its mark in life. Nature 469, 343–349
sis. Cell 170, 367–381
39. Di Croce, L. and Helin, K. (2013) Transcriptional regulation by
19. Lu, F. et al. (2016) Establishing chromatin regulatory landscape
polycomb group proteins. Nat. Struct. Mol. Biol. 20, 1147–1155
during mouse preimplantation development. Cell 165,
1375–1388 40. Simon, J.A. and Kingston, R.E. (2009) Mechanisms of polycomb
gene silencing: knowns and unknowns. Nat. Rev. Mol. Cell Biol.
20. Wu, J. et al. (2016) The landscape of accessible chromatin in
10, 697–708
mammalian preimplantation embryos. Nature 534, 652–657
41. Posfai, E. et al. (2012) Polycomb function during oogenesis is
21. Zhang, B. et al. (2016) Allelic reprogramming of the histone
required for mouse embryonic development. Genes Dev. 26,
modification H3K4me3 in early mammalian development.
920–932
Nature 537, 553–557
42. Ooi, S.K. et al. (2007) DNMT3L connects unmethylated lysine 4 of
22. Kimmins, S. and Sassone-Corsi, P. (2005) Chromatin remodelling
histone H3 to de novo methylation of DNA. Nature 448, 714–717
and epigenetic features of germ cells. Nature 434, 583–589

14 Trends in Cell Biology, Month Year, Vol. xx, No. yy


TICB 1381 No. of Pages 17

43. Zhang, Y. et al. (2010) Chromatin methylation activity of Dnmt3a 66. Catchpole, S. et al. (2011) PLU-1/JARID1B/KDM5B is required
and Dnmt3a/3L is guided by interaction of the ADD domain with for embryonic survival and contributes to cell proliferation in the
the histone H3 tail. Nucleic Acids Res. 38, 4246–4253 mammary gland and in ER+ breast cancer cells. Int. J. Oncol.
44. Bartke, T. et al. (2010) Nucleosome-interacting proteins regu- 38, 1267–1277
lated by DNA and histone methylation. Cell 143, 470–484 67. Huang, J. et al. (2015) Impairment of preimplantation porcine
45. Brinkman, A.B. et al. (2012) Sequential ChIP-bisulfite sequenc- embryo development by histone demethylase KDM5B knock-
ing enables direct genome-scale investigation of chromatin and down through disturbance of bivalent H3K4me3-H3K27me3
DNA methylation cross-talk. Genome Res. 22, 1128–1138 modifications. Biol. Reprod. 92, 72

46. Murphy, P.J. et al. (2013) Single-molecule analysis of combina- 68. Kidder, B.L. et al. (2014) KDM5B focuses H3K4 methylation
torial epigenomic states in normal and tumor cells. Proc. Natl. near promoters and enhancers during embryonic stem cell self-
Acad. Sci. U. S. A. 110, 7772–7777 renewal and differentiation. Genome Biol. 15, R32

47. Reddington, J.P. et al. (2013) Redistribution of H3K27me3 upon 69. Xie, L. et al. (2011) KDM5B regulates embryonic stem cell self-
DNA hypomethylation results in de-repression of Polycomb renewal and represses cryptic intragenic transcription. EMBO J.
target genes. Genome Biol. 14, R25 30, 1473–1484

48. Wachter, E. et al. (2014) Synthetic CpG islands reveal DNA 70. Marks, H. et al. (2012) The transcriptional and epigenomic
sequence determinants of chromatin structure. Elife 3, e03397 foundations of ground state pluripotency. Cell 149, 590–604

49. Wu, H. et al. (2010) Dnmt3a-dependent nonpromoter DNA 71. Inoue, A. et al. (2017) Maternal H3K27me3 controls DNA meth-
methylation facilitates transcription of neurogenic genes. ylation-independent imprinting. Nature 547, 419–424
Science 329, 444–448 72. Zenk, F. et al. (2017) Germ line-inherited H3K27me3 restricts
50. Glaser, S. et al. (2009) The histone 3 lysine 4 methyltransferase, enhancer function during maternal-to-zygotic transition.
Mll2, is only required briefly in development and spermatogen- Science 357, 212–216
esis. Epigenetics Chromatin 2, 5 73. Benayoun, B.A. et al. (2014) H3K4me3 breadth is linked to cell
51. Mu, W. et al. (2017) EZH1 in germ cells safeguards the function identity and transcriptional consistency. Cell 158, 673–688
of PRC2 during spermatogenesis. Dev. Biol. 424, 198–207 74. Suzuki, H.I. et al. (2017) Super-enhancer-mediated RNA proc-
52. Hammoud, S.S. et al. (2014) Chromatin and transcription tran- essing revealed by integrative microRNA network analysis. Cell
sitions of mammalian adult germline stem cells and spermato- 168, 1000–1014
genesis. Cell Stem Cell 15, 239–253 75. Chen, K. et al. (2015) Broad H3K4me3 is associated with
53. Hammoud, S.S. et al. (2009) Distinctive chromatin in human increased transcription elongation and enhancer activity at
sperm packages genes for embryo development. Nature 460, tumor-suppressor genes. Nat. Genet. 47, 1149–1157
473–478 76. Azuara, V. et al. (2006) Chromatin signatures of pluripotent cell
54. Lesch, B.J. et al. (2016) Parallel evolution of male germline lines. Nat. Cell Biol. 8, 532–538
epigenetic poising and somatic development in animals. Nat. 77. Cui, K. et al. (2009) Chromatin signatures in multipotent human
Genet. 48, 888–894 hematopoietic stem cells indicate the fate of bivalent genes
55. Liu, W. et al. (2016) Identification of key factors conquering during differentiation. Cell Stem Cell 4, 80–93
developmental arrest of somatic cell cloned embryos by com- 78. Hattori, N. et al. (2013) Visualization of multivalent histone modi-
bining embryo biopsy and single-cell sequencing. Cell Discov. 2, fication in a single cell reveals highly concerted epigenetic
16010 changes on differentiation of embryonic stem cells. Nucleic
56. Matoba, S. et al. (2014) Embryonic development following Acids Res. 41, 7231–7239
somatic cell nuclear transfer impeded by persisting histone 79. Mikkelsen, T.S. et al. (2007) Genome-wide maps of chromatin
methylation. Cell 159, 884–895 state in pluripotent and lineage-committed cells. Nature 448,
57. Xie, B. et al. (2016) Histone H3 lysine 27 trimethylation acts as an 553–560
epigenetic barrier in porcine nuclear reprogramming. Reproduc- 80. Sachs, M. et al. (2013) Bivalent chromatin marks developmental
tion 151, 9–16 regulatory genes in the mouse embryonic germline in vivo. Cell
58. Burton, A. and Torres-Padilla, M.E. (2010) Epigenetic reprog- Rep. 3, 1777–1784
ramming and development: a unique heterochromatin organi- 81. Mendenhall, E.M. et al. (2010) GC-rich sequence elements
zation in the preimplantation mouse embryo. Brief. Funct. recruit PRC2 in mammalian ES cells. PLoS Genet. 6, e1001244
Genomics 9, 444–454 82. Orlando, D.A. et al. (2012) CpG island structure and trithorax/
59. Lepikhov, K. and Walter, J. (2004) Differential dynamics of polycomb chromatin domains in human cells. Genomics 100,
histone H3 methylation at positions K4 and K9 in the mouse 320–326
zygote. BMC Dev. Biol. 4, 12 83. Kalkan, T. and Smith, A. (2014) Mapping the route from naive
60. Lepikhov, K. et al. (2008) Evidence for conserved DNA and pluripotency to lineage specification. Philos. Trans. R. Soc.
histone H3 methylation reprogramming in mouse, bovine and Lond. B Biol. Sci. 369, 20130540
rabbit zygotes. Epigenetics Chromatin 1, 8 84. Leeb, M. et al. (2010) Polycomb complexes act redundantly to
61. Santenard, A. et al. (2010) Heterochromatin formation in the repress genomic repeats and genes. Genes. Dev. 24,
mouse embryo requires critical residues of the histone variant 265–276
H3.3. Nat. Cell Biol. 12, 853–862 85. Gaume, X. and Torres-Padilla, M.E. (2015) Regulation of reprog-
62. Santos, F. et al. (2005) Dynamic chromatin modifications char- ramming and cellular plasticity through histone exchange and
acterise the first cell cycle in mouse embryos. Dev. Biol. 280, histone variant incorporation. Cold Spring Harb. Symp. Quant.
225–236 Biol. 80, 165–175
63. van der Heijden, G.W. et al. (2005) Asymmetry in histone H3 86. Akiyama, T. et al. (2011) Dynamic replacement of histone H3
variants and lysine methylation between paternal and maternal variants reprograms epigenetic marks in early mouse embryos.
chromatin of the early mouse zygote. Mech. Dev. 122, 1008–1022 PLoS Genet. 7, e1002279
64. Aoshima, K. et al. (2015) Paternal H3K4 methylation is required 87. Inoue, A. and Zhang, Y. (2014) Nucleosome assembly is
for minor zygotic gene activation and early mouse embryonic required for nuclear pore complex assembly in mouse zygotes.
development. EMBO Rep. 16, 803–812 Nat. Struct. Mol. Biol. 21, 609–616
65. Puschendorf, M. et al. (2008) PRC1 and Suv39h specify parental 88. Jang, C.W. et al. (2015) Histone H3.3 maintains genome integ-
asymmetry at constitutive heterochromatin in early mouse rity during mammalian development. Genes Dev. 29,
embryos. Nat. Genet. 40, 411–420 1377–1392

Trends in Cell Biology, Month Year, Vol. xx, No. yy 15


TICB 1381 No. of Pages 17

89. Jullien, J. et al. (2012) HIRA dependent H3.3 deposition is 113. Dixon, J.R. et al. (2012) Topological domains in mammalian
required for transcriptional reprogramming following nuclear genomes identified by analysis of chromatin interactions. Nature
transfer to Xenopus oocytes. Epigenetics Chromatin 5, 17 485, 376–380
90. Torres-Padilla, M.E. et al. (2006) Dynamic distribution of the 114. Nora, E.P. et al. (2012) Spatial partitioning of the regulatory
replacement histone variant H3.3 in the mouse oocyte and landscape of the X-inactivation centre. Nature 485, 381–385
preimplantation embryos. Int. J. Dev. Biol. 50, 455–461 115. Lupianez, D.G. et al. (2015) Disruptions of topological chromatin
91. McLay, D.W. and Clarke, H.J. (2003) Remodelling the paternal domains cause pathogenic rewiring of gene-enhancer interac-
chromatin at fertilization in mammals. Reproduction 125, 625–633 tions. Cell 161, 1012–1025
92. Nonchev, S. and Tsanev, R. (1990) Protamine-histone replace- 116. Valton, A.L. and Dekker, J. (2016) TAD disruption as oncogenic
ment and DNA replication in the male mouse pronucleus. Mol. driver. Curr. Opin. Genet. Dev. 36, 34–40
Reprod. Dev. 25, 72–76 117. Lieberman-Aiden, E. et al. (2009) Comprehensive mapping of
93. Lin, C.J. et al. (2014) Hira-mediated H3.3 incorporation is long-range interactions reveals folding principles of the human
required for DNA replication and ribosomal RNA transcription genome. Science 326, 289–293
in the mouse zygote. Dev. Cell 30, 268–279 118. Naumova, N. et al. (2013) Organization of the mitotic chromo-
94. Tsompana, M. and Buck, M.J. (2014) Chromatin accessibility: a some. Science 342, 948–953
window into the genome. Epigenetics Chromatin 7, 33 119. Nagano, T. et al. (2017) Cell-cycle dynamics of chromosomal
95. Ong, C.T. and Corces, V.G. (2012) Enhancers: emerging roles in organization at single-cell resolution. Nature 547, 61–67
cell fate specification. EMBO Rep. 13, 423–430 120. Battulin, N. et al. (2015) Comparison of the three-dimensional
96. Jachowicz, J.W. et al. (2017) LINE-1 activation after fertilization organization of sperm and fibroblast genomes using the Hi-C
regulates global chromatin accessibility in the early mouse approach. Genome Biol. 16, 77
embryo. Nat. Genet. 49, 1502–1510 121. Carone, B.R. et al. (2014) High-resolution mapping of chromatin
97. Jung, Y.H. et al. (2017) Chromatin states in mouse sperm packaging in mouse embryonic stem cells and sperm. Dev. Cell
correlate with embryonic and adult regulatory landscapes. Cell 30, 11–22
Rep. 18, 1366–1382 122. Miller, D. (2015) Confrontation, consolidation, and recognition:
98. Pott, S. (2017) Simultaneous measurement of chromatin acces- the oocyte’s perspective on the incoming sperm. Cold Spring
sibility, DNA methylation, and nucleosome phasing in single Harb. Perspect. Med. 5, a023408
cells. Elife 6, e23203 123. Mudrak, O. et al. (2009) Reorganisation of human sperm nuclear
99. Kelly, T.K. et al. (2012) Genome-wide mapping of nucleosome architecture during formation of pronuclei in a model system.
positioning and DNA methylation within individual DNA mole- Reprod. Fertil. Dev. 21, 665–671
cules. Genome Res. 22, 2497–2506 124. Hsieh, T.H. et al. (2015) Mapping nucleosome resolution chro-
100. Buenrostro, J.D. et al. (2013) Transposition of native chromatin mosome folding in yeast by micro-C. Cell 162, 108–119
for fast and sensitive epigenomic profiling of open chromatin, 125. Fang, R. et al. (2016) Mapping of long-range chromatin inter-
DNA-binding proteins and nucleosome position. Nat. Methods actions by proximity ligation-assisted ChIP-seq. Cell Res. 26,
10, 1213–1218 1345–1348
101. Tao, C. et al. (2017) Dynamic reorganization of nucleosome 126. Fullwood, M.J. et al. (2009) An oestrogen-receptor-alpha-
positioning in somatic cells after transfer into porcine enucleated bound human chromatin interactome. Nature 462, 58–64
oocytes. Stem Cell Rep. 9, 642–653
127. Mumbach, M.R. et al. (2016) HiChIP: efficient and sensitive
102. Peaston, A.E. et al. (2004) Retrotransposons regulate host analysis of protein-directed genome architecture. Nat. Methods
genes in mouse oocytes and preimplantation embryos. Dev. 13, 919–922
Cell 7, 597–606
128. De La Fuente, R. (2006) Chromatin modifications in the germinal
103. Abe, K. et al. (2015) The first murine zygotic transcription is vesicle (GV) of mammalian oocytes. Dev. Biol. 292, 1–12
promiscuous and uncoupled from splicing and 30 processing.
129. Bouniol-Baly, C. et al. (1999) Differential transcriptional activity
EMBO J. 34, 1523–1537
associated with chromatin configuration in fully grown mouse
104. Macfarlan, T.S. et al. (2012) Embryonic stem cell potency fluc- germinal vesicle oocytes. Biol. Reprod. 60, 580–587
tuates with endogenous retrovirus activity. Nature 487, 57–63
130. Miyara, F. et al. (2003) Chromatin configuration and transcrip-
105. Aoki, F. et al. (1997) Regulation of transcriptional activity during tional control in human and mouse oocytes. Mol. Reprod. Dev.
the first and second cell cycles in the preimplantation mouse 64, 458–470
embryo. Dev. Biol. 181, 296–307
131. Hug, C.B. et al. (2017) Chromatin architecture emerges during
106. Boskovic, A. et al. (2014) Higher chromatin mobility supports zygotic genome activation independent of transcription. Cell
totipotency and precedes pluripotency in vivo. Genes Dev. 28, 169, 216–228 e219
1042–1047
132. Liang, H.L. et al. (2008) The zinc-finger protein Zelda is a key
107. Ahmed, K. et al. (2010) Global chromatin architecture reflects activator of the early zygotic genome in Drosophila. Nature 456,
pluripotency and lineage commitment in the early mouse 400–403
embryo. PLoS One 5, e10531
133. Rowley, M.J. et al. (2017) Evolutionarily conserved principles
108. Ishiuchi, T. et al. (2015) Early embryonic-like cells are induced by predict 3D chromatin organization. Mol. Cell 67, 837–852
downregulating replication-dependent chromatin assembly.
134. Buenrostro, J.D. et al. (2015) Single-cell chromatin accessibility
Nat. Struct. Mol. Biol. 22, 662–671
reveals principles of regulatory variation. Nature 523, 486–490
109. De Iaco, A. et al. (2017) DUX-family transcription factors regulate
135. Cusanovich, D.A. et al. (2015) Multiplex single cell profiling of
zygotic genome activation in placental mammals. Nat. Genet.
chromatin accessibility by combinatorial cellular indexing. Sci-
49, 941–945
ence 348, 910–914
110. Hendrickson, P.G. et al. (2017) Conserved roles of mouse DUX
136. Corces, M.R. et al. (2017) An improved ATAC-seq protocol
and human DUX4 in activating cleavage-stage genes and
reduces background and enables interrogation of frozen tis-
MERVL/HERVL retrotransposons. Nat. Genet. 49, 925–934
sues. Nat. Methods 14, 959–962
111. Whiddon, J.L. et al. (2017) Conservation and innovation in the
137. Jin, W. et al. (2015) Genome-wide detection of DNase I hyper-
DUX4-family gene network. Nat. Genet. 49, 935–940
sensitive sites in single cells and FFPE tissue samples. Nature
112. Bickmore, W.A. (2013) The spatial organization of the human 528, 142–146
genome. Annu. Rev. Genomics Hum. Genet. 14, 67–84

16 Trends in Cell Biology, Month Year, Vol. xx, No. yy


TICB 1381 No. of Pages 17

138. Rotem, A. et al. (2015) Single-cell ChIP-seq reveals cell sub- 145. Shankaranarayanan, P. et al. (2011) Single-tube linear DNA
populations defined by chromatin state. Nat. Biotechnol. 33, amplification (LinDA) for robust ChIP-seq. Nat. Methods 8,
1165–1172 565–567
139. Shen, J. et al. (2015) H3K4me3 epigenomic landscape derived 146. Adli, M. et al. (2010) Genome-wide chromatin maps derived
from ChIP-Seq of 1,000 mouse early embryonic cells. Cell Res. from limited numbers of hematopoietic progenitors. Nat. Meth-
25, 143–147 ods 7, 615–618
140. Brind’Amour, J. et al. (2015) An ultra-low-input native ChIP-seq 147. Gilfillan, G.D. et al. (2012) Limitations and possibilities of low cell
protocol for genome-wide profiling of rare cell populations. Nat. number ChIP-seq. BMC Genomics 13, 645
Commun. 6, 6033 148. Nagano, T. et al. (2013) Single-cell Hi-C reveals cell-to-cell
141. Liu, S. et al. (2014) Setdb1 is required for germline development variability in chromosome structure. Nature 502, 59–64
and silencing of H3K9me3-marked endogenous retroviruses in 149. Stevens, T.J. et al. (2017) 3D structures of individual mammalian
primordial germ cells. Genes Dev. 28, 2041–2055 genomes studied by single-cell Hi-C. Nature 544, 59–64
142. Lara-Astiaso, D. et al. (2014) Immunogenetics. Chromatin state 150. Ramani, V. et al. (2017) Massively multiplex single-cell Hi-C. Nat.
dynamics during blood formation. Science 345, 943–949 Methods 14, 263–266
143. Ng, J.H. et al. (2013) In vivo epigenomic profiling of germ cells 151. Corces, M.R. et al. (2016) Lineage-specific and single-cell chro-
reveals germ cell molecular signatures. Dev. Cell 24, 324–333 matin accessibility charts human hematopoiesis and leukemia
144. Zylicz, J.J. et al. (2015) Chromatin dynamics and the role of G9a evolution. Nat. Genet. 48, 1193–1203
in gene regulation and enhancer silencing during early mouse
development. Elife 4, e09571

Trends in Cell Biology, Month Year, Vol. xx, No. yy 17

You might also like