Germ cells possess the extraordinary and unique capacity to give rise to a new organism and create an enduring link between all generations. To acquire this property, primordial germ cells (PGCs) transit through an unprecedented programme of sequential epigenetic events that culminates in an epigenomic basal state that is the foundation of totipotency. This process is underpinned by genome-wide DNA demethylation, which may occur through several overlapping pathways, including conversion to 5-hydroxymethylcytosine. We propose that the epigenetic programme in PGCs operates through multiple parallel mechanisms to ensure robustness at the level of individual cells while also being flexible through functional redundancy to guarantee high fidelity of the process. Gaining a better understanding of the molecular mechanisms that direct epigenetic reprogramming in PGCs will enhance our ability to manipulate epigenetic memory, cell-fate decisions and applications in regenerative medicine.
Call Girls Ludhiana Just Call 9907093804 Top Class Call Girl Service Available
Parallel mechanisms of epigenetic reprogramming in the germline
1. Review
Parallel mechanisms of epigenetic
reprogramming in the germline
Jamie A. Hackett, Jan J. Zylicz and M. Azim Surani
Wellcome Trust/Cancer Research UK Gurdon Institute and Department of Physiology, Development and Neuroscience,
University of Cambridge, Cambridge, CB2 1QN, UK
Germ cells possess the extraordinary and unique capac- [5,6]. This process is initiated in response to localised sig-
ity to give rise to a new organism and create an enduring nals, including BMP4 and WNT3, which direct activation of
link between all generations. To acquire this property, the key transcriptional regulators B-lymphocyte-induced
primordial germ cells (PGCs) transit through an unprec- maturation protein 1 (Blimp1) and PR domain containing
edented programme of sequential epigenetic events that 14 (Prdm14) in competent epiblast cells [1,7–9]. Lineage-
culminates in an epigenomic basal state that is the restricted PGCs then embark on an orchestrated sequence
foundation of totipotency. This process is underpinned of reprogramming that culminates in a basal epigenetic
by genome-wide DNA demethylation, which may occur state. The number of early PGCs is highly restricted (ap-
through several overlapping pathways, including con- proximately 40 by E7.25) so it is crucial that the complex
version to 5-hydroxymethylcytosine. We propose that series of epigenetic events be robust to ensure that most, if
the epigenetic programme in PGCs operates through not all, cells efficiently transit through the process [10].
multiple parallel mechanisms to ensure robustness at Reprogramming must also proceed rapidly because of strict
the level of individual cells while also being flexible
through functional redundancy to guarantee high fideli- Glossary
ty of the process. Gaining a better understanding of the
5-hydroxymethylcytosine (5hmC): oxidation of methylated cytosines (5mC) by
molecular mechanisms that direct epigenetic repro- TET proteins generates 5hmC, which may be an intermediate during DNA
gramming in PGCs will enhance our ability to manipu- demethylation and can be further converted to 5caC and 5fC. 5hmC is enriched
late epigenetic memory, cell-fate decisions and in pluripotent and some neuronal cell types, but its precise functional
consequences in the genome and its role in DNA demethylation are unclear.
applications in regenerative medicine. Base excision repair (BER): a cellular mechanism for repair of nonhelix-
distorting base mutations or lesions in the genome. BER is initiated by a DNA
Reprogramming PGCs towards totipotency glycosylase (e.g. TDG) that removes inappropriate bases and forms an
Development from the zygote to adulthood is characterised apurinic/apyrimidinic (AP) site, which is then cleaved by an AP endonuclease
and repaired by specific lyases and polymerases. BER may function to remove
by a progressive restriction of cellular potential that gives downstream derivatives of 5mC, such as 5caC, and mediate repair to
rise to all the differentiated somatic cell types. A unique unmodified C [60].
Basal epigenetic state: the unique epigenetic state of PGCs following
exception to this unidirectional process occurs in the germ- reprogramming. By E13.5, the PGC epigenome has undergone extensive
line, where an unprecedented reprogramming event in reorganisation of histone modifications and is stripped of genome-wide DNA
PGCs (see Glossary) reverses epigenetic barriers to plas- methylation, rendering it at its most basal level during the mammalian life
cycle.
ticity and resets genomic potential. Reprogramming in Bisulfite sequencing: a technique used to determine the pattern of allelic DNA
PGCs results in chromatin remodelling, erasure of genomic methylation (5mC) at specific genomic regions. Bisulfite sequencing cannot
imprints and extensive DNA demethylation [1]. This pro- distinguish 5mC from 5hmC [33]. Additionally, 5caC is indistinguishable from
unmodified C by bisulfite sequencing [60].
cess represents the most comprehensive erasure of epige- DNA demethylation: the removal of a methyl group from position 5 of a
netic information in the mammalian life cycle and cytosine base (5mC), which usually resides within a CpG genomic context, to
underpins the totipotent state. Therefore, unravelling generate an unmodified C. DNA demethylation may occur through either a
‘passive’ mechanism that relies on replication-dependent dilution or an ‘active’
the mechanisms that drive reprogramming, particularly process driven by enzymatic replacement independently of DNA replication. As
DNA demethylation, in the unique context of PGCs is of DNA methylation is associated with transcriptional silencing, DNA demethyla-
great interest. tion can generate a transcriptionally competent state.
Epigenetic reprogramming: genome-wide reorganisation of epigenetic mod-
In mice, PGCs are specified from a subset of posterior ifications that overcomes stable epigenetic barriers and enables acquisition of
proximal epiblast cells at approximately embryonic day (E) genomic potential. During the mammalian life cycle, epigenetic reprogram-
ming occurs in PGCs and in early zygotic development.
6.25, resulting in the establishment of a founder population
Genomic imprints: genomic sequences that exhibit differences in CpG methyla-
of PGCs at E7.25 [1,2]. These nascent PGCs subsequently tion according to the parent of origin. These differentially methylated regions
migrate to the genital ridges by approximately E10.5 and, (DMRs) can influence the allele-specific expression of one or more genes.
Primordial germ cell (PGC): the precursors of mature germ cells that are
from E12.5 onwards, they undergo sex-specific development specified during post-implantation development. In vivo, PGCs are restricted
in the gonads [3,4]. Because mammalian PGCs are specified as a unipotent lineage and only give rise to gametes, which generate the
from cells that are already primed towards a somatic fate, totipotent state upon fertilisation. Early PGCs also possess an underlying
genomic plasticity, as evidenced through their capacity to form pluripotent EG
nascent PGCs must both repress the ongoing somatic pro- cells upon in vitro culture.
gramme and activate the germ cell transcriptional network Totipotency: the ability of a cell to give rise to all the cell types of the
embryonic and extra-embryonic lineages. By contrast, pluripotency refers to
the capacity of a cell to generate all the cell types of the embryo.
Corresponding author: Surani, M.A. (a.surani@gurdon.cam.ac.uk).
164 0168-9525/$ – see front matter ß 2012 Elsevier Ltd. All rights reserved. doi:10.1016/j.tig.2012.01.005 Trends in Genetics, April 2012, Vol. 28, No. 4
2. Review Trends in Genetics April 2012, Vol. 28, No. 4
temporal constraints imposed by the entry of male germ sperm or oocytes in vivo, during migration (E8.5–E11.5),
cells into mitotic arrest and female germ cells into meiotic PGCs show multiple transcriptional and epigenetic simi-
arrest at approximately E13.5. To overcome these con- larities to pluripotent ES cells. Indeed, E8.5–E11.5 PGCs
straints, epigenetic reprogramming in PGCs is probably can form pluripotent embryonic germ (EG) cells in vitro,
directed by multiple parallel mechanisms that ensure the which resemble ES cells rather than the post-implantation
fidelity, flexibility and efficiency of this fundamental pro- epiblast cells from which PGCs were originally specified
cess. As such, events like genome-wide erasure of DNA [19,20]. It is possible that this epigenetic state of PGCs is
methylation in PGCs may occur through several intercon- an underlying requirement for the initiation of meiosis
nected mechanisms, including both active and passive path- and the eventual acquisition of totipotency in the zygote.
ways, that collectively confer redundancy and hence Interestingly, migrating PGCs additionally exhibit upre-
robustness to PGC reprogramming and development. gulation of histone H2A/H4 arginine 3 symmetrical meth-
Here, we discuss the epigenetic events in PGCs and ylation (H2A/H4R3me2 s), which is catalysed by protein
the mechanisms that may operate to drive epigenetic arginine methyltransferase 5 (PRMT5) [21]. This modifi-
reprogramming. We suggest that an integrated process cation may contribute to maintaining PGCs in a unipotent
involving parallel systems is at play and consider the state in vivo and to repression of the somatic programme
potential pathways of DNA demethylation. We also high- [22].
light some of the potential roles and developmental Changes in histone modifications occur in parallel with
processes that epigenetic reprogramming contributes to a reported reduction in global levels of DNA methylation
in PGCs. (5mC) in migrating PGCs from approximately E8.0 [11].
Any loss of 5mC might reflect the effects of BLIMP1 and
Epigenetic events in PGCs PRDM14, which repress both DNA (cytosine-5)-methyl-
At the point at which PGCs are specified from post-im- transferase 3a and 3b (Dnmt3a and Dnmt3b) and ubiqui-
plantation epiblast cells, they are epigenetically indistin- tin-like, containing PHD and RING finger domains 1
guishable at the global level from their neighbours, which (Uhrf1), which are essential components of the de novo
are destined for a somatic fate [9,11]. Therefore, nascent and maintenance methylation machinery, respectively
PGCs inherit stable epigenetic states, including DNA (Figure 1) [7,8,23–26]. Additionally, repression of GLP
methylation and X-inactivation, which constitute an epi- may directly affect DNA methylation through a parallel
genetic barrier against the eventual acquisition of totipo- mechanism that is both dependent and independent of
tency [1,12]. It is thus an important early step in PGC H3K9me2 [27,28]. However, most analysed genomic
development to initiate a process of reprogramming that regions, including transposable elements, imprinted loci
erases these stable epigenetic blocks. The first gross epi- and single-copy genes, apparently retain DNA methylation
genetic changes in PGCs entail a reciprocal loss of histone at CpG sites in PGCs until at least E10.5 [29–32], although
H3 lysine 9 dimethylation (H3K9me2) from E7.75 and, in this does not exclude the possibility that there is conver-
most PGCs, a global increase of H3 lysine 27 trimethyla- sion of 5mC to 5-hydroxymethylcytosine (5hmC) [33].
tion (H3K27me3) by E9.5 (Figure 1) [11,13,14]. The ge- Therefore, it is unclear to what extent DNA demethylation
nome-wide depletion of H3K9me2 is potentially a contributes to the early stages of epigenetic reprogram-
consequence of the downregulation of GLP, a methyltrans- ming (from E8.0) in PGCs and whether conversion to 5hmC
ferase that forms a heteromeric complex with G9a (also plays a role. Conversely, it is unclear how 5mC (or 5hmC) is
known as EHMT2) that is required for deposition of H3K9 maintained at analysed genomic regions in migrating
mono- and dimethylation, and the parallel upregulation of PGCs given that essential components of the de novo
specific lysine demethylases ([15] and unpublished obser- and maintenance machinery, particularly UHRF1, are
vations). By contrast, the mechanisms responsible for the absent [7]. One possibility is that UHRF2, which is con-
increased H3K27me3 levels in PGCs remain unclear, al- served with UHRF1 at the sequence level and preferen-
though notably enhancer of zeste homologue 1 (Ezh1), tially binds hemi-methylated DNA associated with
which has H3K27me3 methyltransferase activity, is upre- H3K9me3, can compensate for UHRF1 to maintain DNA
gulated in PGCs [16]. Because H3K9me2 and H3K27me3 methylation in PGCs, either globally or at specific loci [34].
marks are both associated with transcriptional repression, In support of this, UHRF2 is specifically upregulated
it has been postulated that the loss of H3K9me2 is com- during PGC specification, at least in vitro [7,35]. Addition-
plemented by the gain of H3K27me3 to maintain a repres- ally, unlike H3K9me2, global H3K9me3 levels are main-
sive chromatin state in PGCs [17]. However, the precise tained in migrating PGCs [14]. The in vitro PGC-like cells
genomic location and relationship between these epigenet- (PGCLC) generated in an elegant recent study may enable
ic changes remains to be determined. Nonetheless, the further mechanistic insights into epigenetic events in na-
global enrichment of H3K27me3 and loss of H3K9me2 scent PGCs at a higher resolution [36].
establishes a chromatin environment in PGCs that is The early stages of epigenomic reorganisation in PGCs
grossly similar to that in pluripotent embryonic stem are followed by, and are probably a prerequisite for, the
(ES) cells and is coupled to upregulation of pluripotency dramatic genome-wide erasure of DNA methylation and
genes, such as Nanog and SRY Sox2 [1]. Additionally, extensive chromatin remodelling subsequent to entry into
unlike global H3K27me3 levels, the inactive X-chromo- the genital ridges at approximately E10.5. Because gross
some (Xi) exhibits a protracted decline in H3K27me3 in DNA demethylation seems to occur rapidly at a distinct
female PGCs, which is linked to initiation of X-reactivation time point, while most PGCs are in G2 phase, it is held that
[18]. Thus, while PGCs are unipotent and only form either this process is an ‘active’ event occurring independently
165
3. Review Trends in Genetics April 2012, Vol. 28, No. 4
(a) Chromatin dynamics
5mC H3K27me3
Relative levels
H3K9
me2
E5.5 E6.5 E8.0 E10.5 E11.5
(b) Specification and chromatin-modifying factors
Blimp1
Relative expression
Prdm14
G9a
Glp
E5.5 E6.5 E8.0 E10.5 E11.5
(c) DNA methylation factors
Uhrf2
Relative expression
Dnmt1
Dnmt3a
Dnmt3b Uhrf1
E5.5 E6.5 E8.0 E10.5 E11.5
(d) Tet dioxygenases
Relative expression
Tet1
Tet2
Tet3
E5.5 E6.5 E8.0 E10.5 E11.5
Spec. R1 R2
TRENDS in Genetics
Figure 1. Chromatin and transcriptional changes in primordial germ cells (PGCs). Following specification (Spec.), PGCs transit through two sequential phases of
reprogramming during migration (R1) and subsequent to entry into the genital ridge (R2). (a) Epigenomic reorganisation in PGCs. At R1, PGCs exhibit global erasure of
histone H3 lysine 9 dimethylation (H3K9me2), upregulation of H3 lysine 27 trimethylation (H3K27me3) and some loss of the DNA methylation signal. At R2, there is a further
dramatic loss of DNA methylation, which includes erasure of imprints. This correlates with a transient reorganisation of H3K27me3, loss of heterochromatic chromocentres
and remodelling of other chromatin modifications, including histone H2A/H4 arginine 3 symmetrical methylation (H2A/H4R3me2; not shown). (b–d) Transcriptional changes
during PGC development [13]. (b) Specification and chromatin-modifying factors. Upregulation of B-lymphocyte-induced maturation protein 1 (Blimp1) and PR domain
containing 14 (Prdm14) is necessary for specification of PGC fate, whereas the parallel downregulation of glucagon-like peptide (Glp)/G9a and upregulation of lysine
demethylases (not shown) contribute to erasure of H3K9me2 at R1. (c) DNA methylation proteins. Downregulation of DNA (cytosine-5)-methyltransferase 3b (Dnmt3b) and
ubiquitin-like, containing PHD and RING finger domains 1 (Uhrf1) may account for DNA demethylation at R1. UHRF2 may partially compensate for the absence of UHRF1 to
maintain DNA methylation until R2. (d) Ten-eleven translocation gene Tet dioxygenases. Tet1 and Tet2 are expressed in PGCs by R2, whereas Tet3 cannot be detected [50].
Dashed line indicates putative levels. Abbreviation: E, embryonic day.
from DNA replication, although the precise mechanism is [29,32]. The process of DNA methylation erasure also
yet to be elucidated [14]. Reprogramming in PGCs at this initiates a cascade of chromatin remodelling in PGCs at
stage results in almost full erasure of DNA methylation by approximately E11.5. This includes a transient reorgani-
E13.5 [37], with complete stripping of parental imprints sation of linker histone H1, H3K27me3 and H3K9me3 and
and promoter CpG methylation at germline-specific genes stable remodelling of global H3K9ac and H2A/H4R3me2 s
166
4. Review Trends in Genetics April 2012, Vol. 28, No. 4
[14]. Some sequences, including intracisternal A particle replace 5mC during global DNA demethylation in PGCs
(IAP) retrotransposons, partially evade DNA demethyla- [50].
tion in PGCs, although the mechanism that protects these Mechanistically, conversion of 5mC to 5hmC could lead
elements remains to be determined [31]. Nevertheless, the to unmodified cytosine through several routes in PGCs,
erasure of DNA methylation and extensive chromatin including through providing a substrate for base excision
reorganisation at approximately E11.5 renders PGCs in repair (BER)-mediated active demethylation [40]. Several
a basal epigenetic state that represents an epigenomic recent studies have linked BER components to demethyl-
nadir during mammalian development. It is of note that ation during zygotic reprogramming and at specific loci in
the apparent biphasic nature of demethylation in PGCs (at somatic contexts [50–52]. As BER components are also
approximately E8.0 and approximately E11.5) may poten- upregulated in PGCs relative to somatic neighbours during
tially represent a continuous process of 5mC erasure. In epigenetic erasure, genome-wide demethylation in PGCs
any case, reprogramming of the PGC genome as a whole is may also occur, at least partially through BER. Indeed, the
distinguished from reprogramming during zygotic devel- presence of chromatin-associated XRCC1 and active poly[-
opment, where imprints, DNA methylation at multiple loci ADP-ribose] polymerase 1 (PARP1) in PGCs at approxi-
and polycomb-based chromatin modifications remain; this mately E11.5, which signify single-strand DNA breaks
implies that reprogramming in PGCs represents a more associated with BER, further support this possibility [50].
extensive process [32]. Although both reprogramming One potential involvement of BER-mediated DNA de-
events probably share some mechanistic similarities, it methylation in PGCs is that TET-generated 5hmC is further
is also likely that both employ unique systems to achieve processed by deamination to 5-hydroxymethyluridine
distinct levels of epigenomic resetting. (5hmU) by the activation-induced cytidine deaminase/apo-
lipoprotein B mRNA-editing, enzyme-catalytic, polypeptide
Erasure of DNA methylation and cellular identity (AID/APOBEC) family of deaminases, and subsequently
DNA methylation (5mC) within a CpG context is a highly excised by a glycosylase and repaired to unmodified C.
heritable epigenetic mark that is associated with tran- Consistent with this, loss of AID impairs global demethyla-
scriptional repression and that contributes to stable line- tion in PGCs by E13.5, indicating that 5mC erasure in PGCs
age commitment [38–41]. In this respect, global erasure of depends, at least in part, on AID [37]. Indeed, AID also
DNA methylation during PGC development is a fundamen- enhances locus-specific active demethylation mediated by
tal event towards the acquisition of totipotency. The ca- TET1 in somatic cells and has been reported to interact with
pacity of global DNA demethylation to alter cellular thymine DNA glycosylase (TDG), which can excise the
identity, for example in the course of derivation of induced deamination product 5hmU [52,53]. Moreover, AID has
pluripotent stem cells (iPS) from somatic cells, makes been proposed to be required for DNA demethylation of
unravelling the mechanisms that mediate this process of octamer-binding transcription factor 4 (Oct4) and Nanog
great importance [42,43]. Although extensive studies have during somatic cell reprogramming induced by heterokary-
established how and where 5mC is and can be introduced, on formation with ES cells as well as in demethylation in
it remains enigmatic precisely how DNA methylation is zebrafish embryos [54,55]. However, murine PGCs still
removed from the genome [44]. The global DNA demethyl- undergo extensive demethylation by E13.5 in the absence
ation during PGC development represents a unique in vivo of AID; global 5mC is reduced to 22% and 20% in male and
event in that it results in near-complete stripping of CpG female mutant PGCs, respectively, as opposed to 16% and
methylation at almost all genomic loci, including imprints. 8% in wild-type PGCs and compared with approximately
The magnitude of DNA demethylation makes PGCs an 74% in E13.5 embryonic soma. Moreover, AID-mutant mice
unparalleled system to understand the process of 5mC are both viable and fertile, as are APOBEC 1- and APOBEC
erasure in an in vivo context. 2/3-null mice [56–58]. It is also unclear whether AID is
expressed to any significant degree in PGCs at the time of
Active demethylation in PGCs gross demethylation (approximately E11.5), as it has only
The recent identification of three 5mC-dioxygenases [ten- been detected from E12.5 in PGCs, suggesting that AID has
eleven translocation gene 1, 2 and 3 (TET1, TET2 and a role in PGC demethylation either after global 5mC erasure
TET3)], which can convert 5mC to 5-hydroxymethylcyto- (approximately E12.5) or perhaps at an earlier stage (ap-
sine, presented a potential solution to the longstanding proximately E8.5), when the expression of AID is unknown
debate regarding the mechanism of DNA demethylation [50,59]. Likewise, TDG is not detectable in PGCs between
[45,46]. Conversion of 5mC to 5hmC enables several alter- E10.5 and E13.5, as judged by immunofluorescence studies
native but partially overlapping routes to generate an [50]. However, TDG-null PGCs accumulate biallelic CpG
unmodified cytosine (C) residue independently of, or de- methylation by E11.0 at the insulin-like growth factor 2
pendent on, DNA replication (Figure 2). In the zygote, the receptor (Igf2r)-imprinted differentially methylated region
rapid loss of 5mC from the male pronucleus correlates with (DMR), suggesting that TDG has a role in maintaining
a concomitant gain in 5hmC, implying that 5mC is con- a methylation-free state at this locus, presumably by de-
verted to 5hmC in this context and has a fundamental role methylation, although this event may occur prior to PGC
in reprogramming [47,48]. Indeed, loss of TET3, which is specification [52]. Although further clarification of TDG
the only 5mC-dioxigenase significantly expressed in function and expression in PGCs is necessary, its putative
zygotes, led to a failure to erase 5mC and neonatal lethality absence at E11.5 argues against a direct AID- and/or TDG-
[49]. Although TET3 is not detectable in PGCs, TET1 and mediated active demethylation reaction in PGCs, as has
TET2 are present in these cells, suggesting that 5hmC may been reported in other contexts [52]. Therefore, alternative
167
5. Review Trends in Genetics April 2012, Vol. 28, No. 4
(a)
C
OBE 5mC
/ AP
D
AI TET1/2
C
T BE 5hmC
APO
D/ TET1/2
AI
5hmU 5fC
TET1/2
5caC
?
BE
ive
R
ss
Pa
(b) C
5mC
Relative level
Pa
ss
ive
Ac
ti
ve
5hmC
No conversion
E5.5 E6.5 E8.0 E10.5 E11.5
Spec R1 R2
TRENDS in Genetics
Figure 2. Multiple parallel mechanisms of DNA demethylation. (a) Methylated cytosine (5mC) can be demethylated through several overlapping pathways, including
passive and active mechanisms, which may occur in parallel. Passive demethylation (right) can occur through direct replication-dependent depletion of 5mC owing to an
absence or reduction of DNA methyltransferase activity. Alternatively, TET oncogene (TET) proteins can catalyse oxidation of 5mC to 5-hydroxymethylcytosine (5hmC) or
further conversion to 5-formylcytosine (5fC) and 5-carboxylcytosine (5caC), which may all lead to passive demethylation. Active erasure of 5mC (left) can occur through
deamination of either 5mC or 5hmC to thymidine (T) or 5-hydroxymethyluridine (5hmU), respectively, which can act as a substrate for base excision repair (BER) to
unmodified C. Further conversion of 5hmC to 5fC or 5caC may also be actively removed by BER or 5caC can putatively be removed via a direct decarboxylation reaction
(centre). (b) Putative temporal pattern of 5mC conversion to 5hmC during primordial germ cell (PGC) development. The depletion of the 5mC signal at R1 may occur as a
result of conversion to 5hmC or through other mechanisms. Similarly, 5mC may be converted to 5hmC at R2, which may subsequently be removed through passive and/or
active mechanisms. Abbreviations: AID/APOBEC, activation-induced cytidine deaminase/apolipoprotein B mRNA-editing, enzyme-catalytic, polypeptide; R1 and R2, the two
sequential phases of reprogramming during migration (R1) and subsequent to entry into the genital ridge (R2); Spec. specification.
168
6. Review Trends in Genetics April 2012, Vol. 28, No. 4
deaminase–glycosylase complexes may operate down- 5hmU or further oxidation to 5fC and 5caC, all of which are
stream of 5hmC (or 5mC) in PGCs. Investigation of potential BER substrates, ultimately leading to repair-driven de-
candidates, such as other members of the APOBEC family, methylation.
and methyl-CpG-binding domain protein 4 (MBD4), single-
strand selective monofunctional uracil DNA glycosylase Passive demethylation in PGCs
(SMUG1), nth endonuclease III-like 1 (NTH1) or nei endo- There is accumulating evidence that genome-wide DNA
nuclease VIII-like 1 (NEIL1) glycosylases, may shed light on demethylation events include at least a partial ‘passive’
the issue. component. Conversion of 5mC to 5hmC in the zygote has
An alternative possibility is that TET-mediated hydro- recently been shown to lead to replication-dependent pas-
xymethylation has a deamination-independent role in ac- sive demethylation, rather than to active removal of 5hmC
tive demethylation in PGCs, as TET proteins can further or its derivatives [68]. Thus, chromosomes originating in
oxidise 5hmC to 5-formylcytosine (5fC) and subsequently to the paternal pronucleus retain 5hmC through successive
5-carboxylcytosine (5caC) [60,61]. Interestingly, these new cleavage divisions, whereas newly synthesised sister chro-
C derivatives could also be targets for BER excision [60]. matids are devoid of 5mC and 5hmC, leading to a progres-
Indeed, 5fC and 5caC are substrates for TDG, and poten- sive dilution of DNA modifications during development.
tially other glycosylases [60,62]. 5caC could also theoreti- The passive loss of DNA modification is consistent with the
cally be ‘actively’ removed from the genome by a BER- fact that the human maintenance DNA (cytosine-5)-
independent pathway that would involve decarboxylation methyltransferase 1 (DNMT1) cannot recognise 5hmC
by an as yet unknown enzyme. It will be important to and there are no other known 5hmC maintenance mecha-
determine the prevalence of 5fC and 5caC in PGCs during nisms (although notably UHRF1 does efficiently recognise
the key stages of reprogramming to establish whether they 5hmC) [69,70]. Nevertheless, bisulfite sequencing has in-
contribute to demethylation through any mechanism. A dicated there may be an additional active mechanism in
further possibility is that 5hmC is directly targeted for zygotes that is dependent on BER and operates prior to
BER-mediated excision by 5hmC-specific glycosylases with- DNA replication, which may target specific genomic land-
out the requirement for processing by deamination or fur- marks [51]. Additionally, the high sensitivity of the 5hmC
ther oxidation. Indeed, 5hmC glycosylase activity has been antibody may mask partial or locus-specific active erasure
reported in calf thymus extract [63]. Taken together, a of 5hmC on the paternally derived pronuclear chromatids
mechanism based on BER may play at least a partial role [71]. However, it seems probable that the bulk of paternal
in PGC demethylation, although it is currently unclear pronuclear demethylation occurs passively following
which protein complexes direct the process or whether there 5hmC conversion. It remains unclear how the maternal
is a degree of redundancy. Likewise, it is not clear which pronucleus, which does not undergo 5mC to 5hmC conver-
5hmC or 5mC derivative substrate [5hmC, 5hmU, 5fC, 5caC sion, undergoes concomitant passive demethylation, as
or thymidine (T)] might be targeted by the BER machinery DNMT1 is present and sufficient to maintain maternal
for active DNA demethylation in PGCs. imprints [68]. Nonetheless, these studies indicate that
It is also possible that alternative mechanisms, indepen- conversion of 5mC to 5hmC coupled with passive demeth-
dent of 5hmC, contribute to, or drive, active demethylation ylation is feasible and could operate in PGCs.
in PGCs. In Arabidopsis thaliana, demethylation occurs via In murine PGCs, although there is an overall loss of
direct excision of 5mC by the specific bifunctional DNA DNA methylation over a relatively short period at approx-
glycosylase/lyases, DEMETER (DME) and REPRESSOR imately E11.5, it is important to reconsider whether era-
OF SILENCING 1 (ROS1), followed by recruitment of sure of imprinted genes can occur over a protracted period
BER machinery [64,65]. Although no mammalian orthologs of 2 days [30], which would not be compatible with a single
of DME–ROS1 or 5mC-specific glycosylases have been de- genome-wide wave of ‘active’ demethylation model [14].
scribed, it remains a formal possibility that 5mC is directly Indeed, erasure of DNA methylation from imprinted loci
excised from the genome in PGCs. Another possibility is that and repeat elements in porcine PGCs occurs over an ex-
5mC is deaminated directly to T followed by excision by a tended period of up to 20 days, which suggests that ‘active’
glycosylase that recognises the T–G mismatch, such as TDG DNA demethylation does not apply in all mammalian
or MBD4, and repaired to unmodified cytosine by BER. It is PGCs, possibly including those in mice [72]. Similarly to
additionally possible that a demethylation route based on zygotes, conversion of 5mC to 5hmC (or 5fC and 5caC) in
radical sterile alpha motif (SAM) enzymes operates, al- PGCs would facilitate passive replication-dependent de-
though experimental evidence for this is lacking in PGCs methylation (Figure 2). Such a conversion of 5mC to 5hmC
[66]. Similarly, active 5mC erasure has been reported to be would account for the dramatic loss of 5mC staining ob-
linked to the nucleotide excision repair (NER) pathway, and served in PGCs at E11.5 and the protracted nature of
this could contribute to DNA demethylation during PGC imprint erasure, given that bisulfite conversion would
migration (at approximately E8.5–E10.5) [67]. However, it identify hydroxymethylated loci as methylated but 5mC
remains unclear whether such a mechanism could operate antibodies would not [14,30,33]. Interestingly, conversion
in PGCs at approximately E11.5 owing to the absence of of 5mC to 5hmC may also skew bisulfite sequencing
several key NER components between E10.5 and E12.5 [50]. results, as polymerases appear to amplify unmethylated
Thus, the consensus of available data points towards a alleles preferentially over 5hmC-modified alleles after bi-
putative active DNA demethylation mechanism in PGCs sulfite treatment, which may falsely imply demethylation
at approximately E11.5 being based on initial TET-mediat- has occurred rather than conversion to 5hmC [73]. Addi-
ed hydroxylation of 5mC and either deamination towards tionally, TET-mediated conversion to 5hmC may proceed
169
7. Review Trends in Genetics April 2012, Vol. 28, No. 4
to 5caC in PGCs, which is read as unmethylated by bisul- [49]. This suggests that rapid epigenetic reprogramming is
fite sequencing, erroneously indicating the occurrence of involved in diminishing developmental barriers that some
active demethylation [60]. Although functional conse- cells can nevertheless still overcome in a stochastic man-
quences of 5hmC, 5caC or other C derivatives are as yet ner. Similarly, Tet1-null mice sire reduced litter sizes and
unclear, the inherent biases and lack of derivative speci- mutant pups have a smaller body size at birth, potentially
ficity of bisulfite sequencing mean that observing an owing to a delay in overcoming epigenetically imposed
unmethylated allele via this technique does not necessarily barriers [77]. The presence of multiple means of 5mC
signify the presence of unmodified C or the implied active erasure could compensate for the loss of function of Tet1
demethylation. As such, conversion to 5hmC or down- through the use of alternative routes in a stochastic man-
stream C derivatives could at least partially account for ner. For example, given that key maintenance-methylation
the apparent active ‘replication-independent’ demethyla- enzymes, including UHRF1, are absent in PGCs, it is
tion that has been reported in PGCs. possible that direct passive demethylation of 5mC can
An alternative to passive erasure through 5hmC con- partly offset the absence of any putative 5hmC-mediated
version is that PGCs undergo direct passive dilution of mechanisms, a possibility that might also affect interpre-
5mC, based on an absence or exclusion of maintenance tation of the demethylation that still occurs in AID-defi-
components, such as DNMT1 or UHRF1. As the doubling cient PGCs by E13.5 [7,37].
time of PGCs at this stage is approximately 16 hr [13,74], In addition to contributing to a robust system, parallel
direct replication-dependent dilution could account for a mechanisms of 5mC erasure could function to target spe-
significant reduction in global 5mC in PGCs between E10.5 cific genomic loci in PGCs. For example, passive demeth-
and E11.5. Because the 5mC antibody (33D3) exhibits ylation could account for the bulk of demethylation,
relatively weak avidity (e.g. as compared to the 5hmC analogously to the zygote, whereas erasure of imprinted
antibody [71]) and 5mC levels at E10.5 may already be regions may be targeted by specific active mechanisms that
partially depleted relative to somatic neighbours, this are absent in the zygote, perhaps based on BER. Interest-
passive reduction of 5mC may appear as a largely complete ingly, targeting specific loci with distinct demethylation
and, therefore, active erasure of global DNA methylation, mechanisms may also lead to particular functional out-
while there would also be a significant loss of methylation comes, perhaps based on the specific C derivative (5hmC,
by bisulfite sequencing. Although it is not clear precisely 5fC, 5caC, etc.) that directed demethylation. This could
how a passive demethylation mechanism might be trig- have an important functional role during meiosis, similarly
gered, the existence of such a system would circumvent the to the role of inherited histone modifications at develop-
potential genetic damage that may result from genome- mental loci in the zygote [81]. Active DNA demethylation
wide BER. mechanisms could also operate in parallel to a passive
system to initiate DNA repair-driven chromatin remodel-
Parallel mechanisms of epigenetic reprogramming ling [14]. Notably, other phases of epigenetic reprogram-
Although cumulative evidence indicates that several mo- ming in PGCs also utilise parallel systems. For example,
lecular pathways of DNA demethylation may operate, erasure of H3K9me2 in early PGCs may occur through
the precise mechanism(s) that mediate the comprehen- both downregulation of the methyltransferase GLP and
sive methylation erasure in PGCs remain to be clarified upregulation of specific lysine demethylases at approxi-
[44,75,76]. Indeed, DNA demethylation and chromatin mately E8.5, which potentially contribute to erasure of this
remodelling in PGCs may occur through several comple- modification in a redundant manner. Similarly, the histone
mentary parallel pathways, including active and passive replacement that is associated with reorganisation of glob-
systems, which would provide a degree of redundancy al H3K27me3 and H3K9me3 marks in PGCs at approxi-
and confer robustness and flexibility to the programme mately E11.5 also occurs in parallel with upregulation of
(Figure 3). This seems necessary because of the funda- specific lysine demethylases, which may contribute to the
mental importance of epigenetic reprogramming to germ efficient erasure. Indeed, both Tet1 and Tet2 are also
cell development. Furthermore, there are very limited upregulated at this stage, indicating a possible parallel
numbers of PGCs and most, if not all, of them must role [50]. Thus, we propose that the unique importance of
transit through the process efficiently and prior to the reprogramming in PGCs necessitates a system of multiple
entry of female germ cells into meiosis shortly after epigenetic erasure mechanisms to confer efficiency, fidelity
reprogramming. and robustness to the process.
A robust system based on multiple parallel mechanisms
and inbuilt redundancy may account for the observation Potential roles of epigenetic reprogramming in PGCs
that Tet1-null and Tet2-null mice are viable and fertile, The fundamental role of epigenetic reprogramming in
despite a probable role for 5hmC in demethylation in PGCs PGCs is to overcome multiple epigenomic barriers to the
[77–80]. Although functional redundancy coupled with eventual acquisition of totipotency acquired by epiblast
genetic background effects may explain this observation, cells during early development. This is necessary because
it is also possible that impeded DNA demethylation in mammals utilise an inductive mechanism of germ cell
TET-deficient PGCs would not cause complete loss of the specification (they specify the germline from cells primed
cells but rather a reduction in their numbers. In support of towards a somatic fate) rather than acquire germ cell fate
this, impeded demethylation of the paternal genome in through an inherited germplasm. This essential role of
TET3-deficient zygotes is only associated with a severe reprogramming in mammals is evident in mice lacking the
developmental phenotype in a subset of mutant embryos transcriptional regulator PRDM14. Nascent PGCs in
170
8. Review Trends in Genetics April 2012, Vol. 28, No. 4
t 3 a / b , U h r f 1 d o wn r e
Dnm gu l a
t io n
T ET 1 / 2 a c t i v i t y
Deamination
5mC 5mC 5mC
Me
e
Pa s
M
sive
Ac
BE R
Ac
e Me
M
Reprogramming
DNA demethyla
M
e
Ac
Primed state
tion
H M T/
Histo
HDM
HA
ne e
/HD
Td
xch
AC
ow
an
nr
up
eg
ge
reg
ul
ati
ula
on
on
ti
His
ton
em
ark r
eprog M
Ac
ramming e
Progress
ion toward
Basal sta s
te
TRENDS in Genetics
Figure 3. Parallel routes towards the basal epigenetic state. Post-implantation epiblast cells acquire epigenetic modifications that constitute a barrier against their reversion
to a pluripotent state but remain primed to respond to signals that specify a primordial germ cell (PGC) fate. Upon specification, PGCs embark on a process of
reprogramming that overcomes the epigenetic barrier and establishes a basal epigenetic environment that underpins totipotency. Multiple parallel mechanisms contribute
to epigenetic reprogramming in PGCs to ensure robustness and redundancy to the process. Global erasure of DNA methylation (upper routes) can occur through passive or
active mechanisms directed through several interconnected processes. Likewise, reorganisation of chromatin architecture (lower routes) can occur through the parallel
upregulation of histone demethylases (HDM)/deacetylases (HDAC), downregulation of chromatin-modifying enzymes, including histone methyltransferases (HMT)/
acetyltransferases (HAT), and dynamic histone exchange. Abbreviations: 5mC, methylated cytosine; Ac, acetylation; BER, base excision repair; Dnmt3a/b, DNA (cytosine-5)-
methyltransferase 3a/b; Me, methylation; TET, ten-eleven translocation gene; Uhrf1, ubiquitin-like, containing PHD and RING finger domains 1.
Prdm14-mutant mice fail to undergo early reprogramming RNA Xist necessary to complete the process [18,82]. The
of chromatin and DNA methylation or to activate the germ wave of global DNA demethylation in PGCs at approxi-
cell genetic programme and consequently exhibit severely mately E11.5 seems to be a trigger for several other
impaired PGC development and are sterile [8]. Evidence important processes, including the erasure of parental
shows that PGC specification and epigenetic reprogram- imprints to enable establishment of new methylation
ming are intimately linked. Furthermore, within the broad marks according to the sex of the embryo [17,29]. The
remit of establishing a state of ‘underlying totipotency’, significant impact that aberrant imprints have on mam-
epigenomic reprogramming has many distinct roles that malian development, foetal growth and behaviour, and in
collectively contribute to a continuous process of PGC human disease, has led to the suggestion that the whole
development (Figure 4). process of genome-wide demethylation in PGCs reflects a
An important event driven by reprogramming is reacti- byproduct of the necessity to reset locus-specific imprints.
vation of Xi in female PGCs. This requires the integration Indeed, as imprinted loci are resistant to demethylation
of multiple genetic and epigenetic systems; with early during zygotic reprogramming [82], they may have intrin-
chromatin reorganisation initiating X-reactivation and sic or epigenetic properties that preclude erasure except
DNA demethylation and repression of the long noncoding under exceptional cellular conditions, such as exists in
171
9. Review Trends in Genetics April 2012, Vol. 28, No. 4
(a)
Removal of epigenetic barriers to totipotency
(b) X-chromosome reactivation
(c)
Epimutation erasure
(d) Germline gene reactivation
(e) Imprint erasure
(f) Retrotransposon reactivation
(g) Setting for meiosis
R1 R2
~E7.75–9.0 ~E10.5–12.5
TRENDS in Genetics
Figure 4. Potential functions of epigenetic reprogramming in primordial germ cells (PGCs). The distinct roles of epigenetic reprogramming in PGCs and the time point at
which each occurs (rectangles). (a) Epigenetic reprogramming events during PGC migration (R1) and post-migration (R2) overcomes the layers of epigenetic modifications
acquired by epiblast cells, which form a barrier to acquisition of totipotency. (b) In female cells, reprogramming contributes to X-chromosome reactivation. (c) Erasure of
epigenetic marks in PGCs also prevents perpetuation of epimutations through the germline. (d) Global DNA demethylation at R2 triggers activation of stably silenced genes
necessary for germ cell development (e) A fundamental role of reprogramming is to erase parent-of-origin imprints, thereby enabling subsequent sex-specific methylation
marks to be established during gametogenesis (f) Erasure of methylated cytosine (5mC) may permit transient expression of retrotransposons, which are subsequently
targeted for stable transcriptional repression by small RNAs, thereby ensuring that potentially harmful genetic elements are strongly silenced in the germline. (g)
Reprogramming in PGCs contributes to establishing a chromatin and transcriptional environment that is primed for entry into meiosis. Abbreviation: E, embryonic day.
PGCs. Global DNA demethylation in PGCs has other transient activation of TEs during reprogramming
important functions, including the activation of stably enables silencing mechanisms to target repressive chro-
silenced genes that are required for germ cell development matin, possibly directed by small RNAs derived from
[32,83]. Genes such as deleted in azoospermia-like (Dazl) expressed TEs [86]. This putative mechanism would en-
and synaptonemal complex protein 3 (Sycp3) are robustly sure that harmful genetic elements that may have escaped
silenced by methylated CpG-dense promoters during post- repressive mechanisms are rendered transcriptionally
implantation development, potentially to prevent ectopic quiescent throughout germ cell maturation. Notably,
activation that may drive malignant tumour growth, but the transient erasure of the PGC epigenome may also
are activated by demethylation specifically in the germline provide an opportunity for epigenetic writers to access
[32,39,84,85]. Another key role of reprogramming in PGCs chromatin and establish new modifications that direct
is to erase aberrant epigenetic information or epimuta- events that are crucial for meiosis. Indeed, the absence
tions. Erasure of such modifications in the zygote and of some chromatin-modifying proteins, including G9a,
particularly in PGCs prevents their inheritance and per- MLL2, SUV39H1, SUV39H2, LSH and PRDM9, only
petuation through successive generations with potentially manifests in germ cells during meiosis [87–91]. One in-
detrimental effects. triguing possibility is that epigenetic erasure is necessary
Global DNA demethylation may generate an epigenetic to promote chiasmata formation at recombination hot-
environment susceptible to expression of harmful trans- spots during meiosis. Here, reprogramming may enable
posable elements (TE), yet paradoxically it is possible that the lysine methylase PRDM9 access to target H3K4me3
172
10. Review Trends in Genetics April 2012, Vol. 28, No. 4
marks specifically to 13-mer recognition sites, which act as 16 Shen, X. et al. (2008) EZH1 mediates methylation on histone H3 lysine
27 and complements EZH2 in maintaining stem cell identity and
beacons for recombination [92–95]. Thus, it is probable
executing pluripotency. Mol. Cell 32, 491–502
that functionally, epigenetic reprogramming in PGCs is a 17 Sasaki, H. and Matsui, Y. (2008) Epigenetic events in mammalian
protracted event (E7.75–E12.5) that contributes to myriad germ-cell development: reprogramming and beyond. Nat. Rev. Genet. 9,
interconnected processes that are collectively essential for 129–140
germ cell potency and development. 18 Chuva de Sousa Lopes, S.M. et al. (2008) X chromosome activity in
mouse XX primordial germ cells. PLoS Genet. 4, e30
19 Tada, T. et al. (1998) Epigenotype switching of imprintable loci in
Concluding remarks embryonic germ cells. Dev. Genes Evol. 207, 551–561
Considerable advances have been made in understanding 20 Matsui, Y. et al. (1992) Derivation of pluripotential embryonic stem
how new epigenetic information can be introduced, but less cells from murine primordial germ cells in culture. Cell 70, 841–847
is known about the mechanisms that can erase existing 21 Ancelin, K. et al. (2006) Blimp1 associates with Prmt5 and directs
histone arginine methylation in mouse germ cells. Nat. Cell Biol. 8,
modifications. Studies on PGCs provide an unprecedented 623–630
opportunity to unravel the role of key factors and their 22 Tee, W.W. et al. (2010) Prmt5 is essential for early mouse development
combined roles in resetting the epigenome. Fundamental and acts in the cytoplasm to maintain ES cell pluripotency. Genes Dev.
knowledge gained from these studies may potentially find 24, 2772–2777
wider application. Somatic cells are prone to epimutations 23 Bostick, M. et al. (2007) UHRF1 plays a role in maintaining DNA
methylation in mammalian cells. Science 317, 1760–1764
through ageing and environmental factors, whereas the 24 Sharif, J. et al. (2007) The SRA protein Np95 mediates epigenetic
germ line can reset the epigenome with the potential to inheritance by recruiting Dnmt1 to methylated DNA. Nature 450,
‘rejuvenate’ adult cells. Insights from research on early 908–912
germ cells may provide knowledge and tools for applica- 25 Kaneda, M. et al. (2004) Essential role for de novo DNA
methyltransferase Dnmt3a in paternal and maternal imprinting.
tions in ageing-related diseases and generally improve
Nature 429, 900–903
ability to manipulate cell fates for applications in repro- 26 Kato, Y. et al. (2007) Role of the Dnmt3 family in de novo methylation of
ductive and regenerative medicine. imprinted and repetitive sequences during male germ cell development
in the mouse. Hum. Mol. Genet. 16, 2272–2280
27 Tachibana, M. et al. (2008) G9a/GLP complexes independently mediate
Acknowledgements
H3K9 and DNA methylation to silence transcription. EMBO J. 27,
We would like to thank Roopsha Sengupta and Harry Leitch for critical
2681–2690
reading of the manuscript and to apologise to the authors whose work
28 Dong, K.B. et al. (2008) DNA methylation in ES cells requires the lysine
could not be cited here owing to space constraints. JAH was funded by a
methyltransferase G9a but not its catalytic activity. EMBO J. 27,
Wellcome Trust Grant and JJZ is the recipient of a Wellcome Trust PhD
2691–2701
Scholarship. MAS is supported by The Wellcome Trust (RG49135).
29 Hajkova, P. et al. (2002) Epigenetic reprogramming in mouse
primordial germ cells. Mech. Dev. 117, 15–23
References 30 Lee, J. et al. (2002) Erasing genomic imprinting memory in mouse clone
1 Surani, M.A. et al. (2007) Genetic and epigenetic regulators of embryos produced from day 11.5 primordial germ cells. Development
pluripotency. Cell 128, 747–762 129, 1807–1817
2 Ginsburg, M. et al. (1990) Primordial germ cells in the mouse embryo 31 Lane, N. et al. (2003) Resistance of IAPs to methylation reprogramming
during gastrulation. Development 110, 521–528 may provide a mechanism for epigenetic inheritance in the mouse.
3 McLaren, A. (1984) Meiosis and differentiation of mouse germ cells. Genesis 35, 88–93
Symp. Soc. Exp. Biol. 38, 7–23 32 Maatouk, D.M. et al. (2006) DNA methylation is a primary mechanism
4 Kocer, A. et al. (2009) Germ cell sex determination in mammals. Mol. for silencing postmigratory primordial germ cell genes in both germ cell
Hum. Reprod. 15, 205–213 and somatic cell lineages. Development 133, 3411–3418
5 Saitou, M. et al. (2002) A molecular programme for the specification of 33 Nestor, C. et al. (2010) Enzymatic approaches and bisulfite
germ cell fate in mice. Nature 418, 293–300 sequencing cannot distinguish between 5-methylcytosine and 5-
6 Yabuta, Y. et al. (2006) Gene expression dynamics during germline hydroxymethylcytosine in DNA. Biotechniques 48, 317–319
specification in mice identified by quantitative single-cell gene 34 Pichler, G. et al. (2011) Cooperative DNA and histone binding by Uhrf2
expression profiling. Biol. Reprod. 75, 705–716 links the two major repressive epigenetic pathways. J. Cell. Biochem.
7 Kurimoto, K. et al. (2008) Complex genome-wide transcription 112, 2585–2593
dynamics orchestrated by Blimp1 for the specification of the germ 35 West, J.A. et al. (2009) A role for Lin28 in primordial germ-cell
cell lineage in mice. Genes Dev. 22, 1617–1635 development and germ-cell malignancy. Nature 460, 909–913
8 Yamaji, M. et al. (2008) Critical function of Prdm14 for the establishment 36 Hayashi, K. et al. (2011) Reconstitution of the mouse germ cell
of the germ cell lineage in mice. Nat. Genet. 40, 1016–1022 specification pathway in culture by pluripotent stem cells. Cell 146,
9 Ohinata, Y. et al. (2009) A signaling principle for the specification of the 519–532
germ cell lineage in mice. Cell 137, 571–584 37 Popp, C. et al. (2010) Genome-wide erasure of DNA methylation in
10 McLaren, A. and Lawson, K.A. (2005) How is the mouse germ-cell mouse primordial germ cells is affected by AID deficiency. Nature 463,
lineage established? Differentiation 73, 435–437 1101–1105
11 Seki, Y. et al. (2005) Extensive and orderly reprogramming of genome- 38 Mohn, F. et al. (2008) Lineage-specific polycomb targets and de novo
wide chromatin modifications associated with specification and early DNA methylation define restriction and potential of neuronal
development of germ cells in mice. Dev. Biol. 278, 440–458 progenitors. Mol. Cell 30, 755–766
12 Bao, S. et al. (2009) Epigenetic reversion of post-implantation epiblast 39 Borgel, J. et al. (2010) Targets and dynamics of promoter DNA
to pluripotent embryonic stem cells. Nature 461, 1292–1295 methylation during early mouse development. Nat. Genet. 42,
13 Seki, Y. et al. (2007) Cellular dynamics associated with the genome- 1093–1100
wide epigenetic reprogramming in migrating primordial germ cells in 40 Wu, S.C. and Zhang, Y. (2010) Active DNA demethylation: many roads
mice. Development 134, 2627–2638 lead to Rome. Nat. Rev. Mol. Cell Biol. 11, 607–620
14 Hajkova, P. et al. (2008) Chromatin dynamics during epigenetic 41 Hodges, E. et al. (2011) Directional DNA methylation changes and
reprogramming in the mouse germ line. Nature 452, 877–881 complex intermediate states accompany lineage specificity in the adult
15 Tachibana, M. et al. (2005) Histone methyltransferases G9a and GLP hematopoietic compartment. Mol. Cell 44, 17–28
form heteromeric complexes and are both crucial for methylation of 42 Mikkelsen, T.S. et al. (2008) Dissecting direct reprogramming through
euchromatin at H3-K9. Genes Dev. 19, 815–826 integrative genomic analysis. Nature 454, 49–55
173
11. Review Trends in Genetics April 2012, Vol. 28, No. 4
43 Surani, M.A. and Hajkova, P. (2010) Epigenetic reprogramming of 69 Valinluck, V. and Sowers, L.C. (2007) Endogenous cytosine damage
mouse germ cells toward totipotency. Cold Spring Harb. Symp. Quant. products alter the site selectivity of human DNA maintenance
Biol. 75, 211–218 methyltransferase DNMT1. Cancer Res. 67, 946–950
44 Ooi, S.K. and Bestor, T.H. (2008) The colorful history of active DNA 70 Frauer, C. et al. (2011) Recognition of 5-hydroxymethylcytosine by the
demethylation. Cell 133, 1145–1148 Uhrf1 SRA domain. PLoS ONE 6, e21306
45 Tahiliani, M. et al. (2009) Conversion of 5-methylcytosine to 5- 71 Ficz, G. et al. (2011) Dynamic regulation of 5-hydroxymethylcytosine in
hydroxymethylcytosine in mammalian DNA by MLL partner TET1. mouse ES cells and during differentiation. Nature 473, 398–402
Science 324, 930–935 72 Hyldig, S.M. et al. (2011) Epigenetic reprogramming in the porcine
46 Ito, S. et al. (2010) Role of Tet proteins in 5mC to 5hmC conversion, germ line. BMC Dev. Biol. 11, 11
ES-cell self-renewal and inner cell mass specification. Nature 466, 73 Huang, Y. et al. (2010) The behaviour of 5-hydroxymethylcytosine in
1129–1133 bisulfite sequencing. PLoS ONE 5, e8888
47 Wossidlo, M. et al. (2011) 5-Hydroxymethylcytosine in the mammalian 74 Tam, P.P. and Snow, M.H. (1981) Proliferation and migration of
zygote is linked with epigenetic reprogramming. Nat. Commun. 2, 241 primordial germ cells during compensatory growth in mouse
48 Iqbal, K. et al. (2011) Reprogramming of the paternal genome upon embryos. J. Embryol. Exp. Morphol. 64, 133–147
fertilization involves genome-wide oxidation of 5-methylcytosine. Proc. 75 Bhutani, N. et al. (2011) DNA demethylation dynamics. Cell 146,
Natl. Acad. Sci. U.S.A. 108, 3642–3647 866–872
49 Gu, T.P. et al. (2011) The role of Tet3 DNA dioxygenase in epigenetic 76 Law, J.A. and Jacobsen, S.E. (2010) Establishing, maintaining and
reprogramming by oocytes. Nature 477, 606–610 modifying DNA methylation patterns in plants and animals. Nat. Rev.
50 Hajkova, P. et al. (2010) Genome-wide reprogramming in the mouse Genet. 11, 204–220
germ line entails the base excision repair pathway. Science 329, 78 77 Dawlaty, M.M. et al. (2011) Tet1 is dispensable for maintaining
51 Wossidlo, M. et al. (2010) Dynamic link of DNA demethylation, DNA pluripotency and its loss is compatible with embryonic and
strand breaks and repair in mouse zygotes. EMBO J. 29, 1877–1888 postnatal development. Cell Stem Cell 9, 166–175
52 Cortellino, S. et al. (2011) Thymine DNA glycosylase is essential for 78 Li, Z. et al. (2011) Deletion of Tet2 in mice leads to dysregulated
active DNA demethylation by linked deamination-base excision repair. hematopoietic stem cells and subsequent development of myeloid
Cell 146, 67–79 malignancies. Blood 118, 4509–4518
53 Guo, J.U. et al. (2011) Hydroxylation of 5-methylcytosine by TET1 79 Moran-Crusio, K. et al. (2011) Tet2 loss leads to increased
promotes active DNA demethylation in the adult brain. Cell 145, hematopoietic stem cell self-renewal and myeloid transformation.
423–434 Cancer Cell 20, 11–24
54 Bhutani, N. et al. (2010) Reprogramming towards pluripotency 80 Quivoron, C. et al. (2011) TET2 inactivation results in pleiotropic
requires AID-dependent DNA demethylation. Nature 463, 1042–1047 hematopoietic abnormalities in mouse and is a recurrent event
55 Rai, K. et al. (2008) DNA demethylation in zebrafish involves the during human lymphomagenesis. Cancer Cell 20, 25–38
coupling of a deaminase, a glycosylase, and gadd45. Cell 135, 1201–1212 81 Hammoud, S.S. et al. (2009) Distinctive chromatin in human sperm
56 Muramatsu, M. et al. (2000) Class switch recombination and packages genes for embryo development. Nature 460, 473–478
hypermutation require activation-induced cytidine deaminase (AID), 82 Morgan, H.D. et al. (2005) Epigenetic reprogramming in mammals.
a potential RNA editing enzyme. Cell 102, 553–563 Hum. Mol. Genet. 14, R47–R58 (Spec No 1)
57 Morrison, J.R. et al. (1996) Apolipoprotein B RNA editing enzyme- 83 Rodic, N. et al. (2005) DNA methylation is required for silencing
deficient mice are viable despite alterations in lipoprotein metabolism. of ant4, an adenine nucleotide translocase selectively expressed
Proc. Natl. Acad. Sci. U.S.A. 93, 7154–7159 in mouse embryonic stem cells and germ cells. Stem Cells 23,
58 Mikl, M.C. et al. (2005) Mice deficient in APOBEC2 and APOBEC3. 1314–1323
Mol. Cell. Biol. 25, 7270–7277 84 Janic, A. et al. (2010) Ectopic expression of germline genes drives
59 Morgan, H.D. et al. (2004) Activation-induced cytidine deaminase malignant brain tumor growth in Drosophila. Science 330, 1824–1827
deaminates 5-methylcytosine in DNA and is expressed in 85 Simpson, A.J. et al. (2005) Cancer/testis antigens, gametogenesis and
pluripotent tissues. J. Biol. Chem. 279, 52353–52360 cancer. Nat. Rev. Cancer 5, 615–625
60 He, Y.F. et al. (2011) Tet-mediated formation of 5-carboxylcytosine and 86 Seisenberger, S. et al. (2010) Retrotransposons and germ cells:
its excision by TDG in mammalian DNA. Science 333, 1303–1307 reproduction, death, and diversity. F1000 Biol. Rep. 2
61 Ito, S. et al. (2011) Tet proteins can convert 5-methylcytosine to 5- 87 Tachibana, M. et al. (2007) Functional dynamics of H3K9 methylation
formylcytosine and 5-carboxylcytosine. Science 333, 1300–1303 during meiotic prophase progression. EMBO J. 26, 3346–3359
62 Maiti, A. and Drohat, A.C. (2011) Thymine DNA glycosylase can 88 Peters, A.H. et al. (2001) Loss of the Suv39 h histone
rapidly excise 5-formylcytosine and 5–carboxylcytosine: potential methyltransferases impairs mammalian heterochromatin and
implications for active demethylation of CpG sites. J. Biol. Chem. genome stability. Cell 107, 323–337
286, 35334–35338 89 Glaser, S. et al. (2009) The histone 3 lysine 4 methyltransferase, Mll2,
63 Cannon, S.V. et al. (1988) 5-Hydroxymethylcytosine DNA glycosylase is only required briefly in development and spermatogenesis.
activity in mammalian tissue. Biochem. Biophys. Res. Commun. 151, Epigenetics Chromatin 2, 5
1173–1179 90 De La Fuente, R. et al. (2006) Lsh is required for meiotic chromosome
64 Choi, Y. et al. (2002) DEMETER, a DNA glycosylase domain protein, is synapsis and retrotransposon silencing in female germ cells. Nat. Cell
required for endosperm gene imprinting and seed viability in Biol. 8, 1448–1454
arabidopsis. Cell 110, 33–42 91 Kota, S.K. and Feil, R. (2010) Epigenetic transitions in germ cell
65 Gong, Z. et al. (2002) ROS1, a repressor of transcriptional gene development and meiosis. Dev. Cell 19, 675–686
silencing in Arabidopsis, encodes a DNA glycosylase/lyase. Cell 111, 92 Myers, S. et al. (2010) Drive against hotspot motifs in primates
803–814 implicates the PRDM9 gene in meiotic recombination. Science 327,
66 Okada, Y. et al. (2010) A role for the elongator complex in zygotic 876–879
paternal genome demethylation. Nature 463, 554–558 93 Baudat, F. et al. (2010) prdm9 is a major determinant of meiotic
67 Schmitz, K.M. et al. (2009) TAF12 recruits Gadd45a and the nucleotide recombination hotspots in humans and mice. Science 327, 836–840
excision repair complex to the promoter of rRNA genes leading to active 94 Parvanov, E.D. et al. (2010) Prdm9 controls activation of mammalian
DNA demethylation. Mol. Cell 33, 344–353 recombination hotspots. Science 327, 835
68 Inoue, A. and Zhang, Y. (2011) Replication-dependent loss of 95 Hayashi, K. et al. (2005) A histone H3 methyltransferase
5-Hydroxymethylcytosine in mouse preimplantation embryos. controls epigenetic events required for meiotic prophase. Nature
Science 334, 194 438, 374–378
174