Content uploaded by Felix Krueger
Author content
All content in this area was uploaded by Felix Krueger
Content may be subject to copyright.
Molecular Cell
Article
The Dynamics of Genome-wide DNA Methylation
Reprogramming in Mouse Primordial Germ Cells
Stefanie Seisenberger,
1
Simon Andrews,
2
Felix Krueger,
2
Julia Arand,
3
Jo
¨rn Walter,
3
Fa
´tima Santos,
1
Christian Popp,
1
Bernard Thienpont,
1,4
Wendy Dean,
1
and Wolf Reik
1,5,6,
*
1
Epigenetics Programme
2
Bioinformatics Group
The Babraham Institute, Cambridge CB22 3AT, UK
3
Department of Biological Sciences, Institute of Genetics/Epigenetics, University of Saarland, Campus Saarbru
¨cken,
66123 Saarbru
¨cken, Germany
4
Laboratory of Translational Genetics, Vesalius Research Center, VIB and KULeuven, 3000 Leuven, Belgium
5
Centre for Trophoblast Research, University of Cambridge, Cambridge CB2 3EG, UK
6
Wellcome Trust Sanger Institute, Cambridge CB10 1SA, UK
*Correspondence: wolf.reik@babraham.ac.uk
http://dx.doi.org/10.1016/j.molcel.2012.11.001
SUMMARY
Genome-wide DNA methylation reprogramming
occurs in mouse primordial germ cells (PGCs) and
preimplantation embryos, but the precise dynamics
and biological outcomes are largely unknown. We
have carried out whole-genome bisulfite sequencing
(BS-Seq) and RNA-Seq across key stages from E6.5
epiblast to E16.5 PGCs. Global loss of methylation
takes place during PGC expansion and migra-
tion with evidence for passive demethylation, but
sequences that carry long-term epigenetic memory
(imprints, CpG islands on the X chromosome, germ-
line-specific genes) only become demethylated upon
entry of PGCs into the gonads. The transcriptional
profile of PGCs is tightly controlled despite global
hypomethylation, with transient expression of the
pluripotency network, suggesting that reprogram-
ming and pluripotency are inextricably linked. Our
results provide a framework for the understanding
of the epigenetic ground state of pluripotency in the
germline.
INTRODUCTION
Epigenetic information in the mammalian genome is relatively
stable in differentiated cells of the soma but is reprogrammed
on a genome-wide scale in primordial germ cells (PGCs) and
early embryos (Reik et al., 2001;Surani et al., 2007;Sasaki and
Matsui, 2008;Smith et al., 2012). This includes the erasure of
DNA methylation and the large-scale reprogramming of histone
modifications and histone variants (Hajkova et al., 2002,2008;
Lee et al., 2002;Lane et al., 2003;Yamazaki et al., 2003;Seki
et al., 2005,2007;Popp et al., 2010;Guibert et al., 2012). A
recent interesting insight into reprogramming of histone modifi-
cations in PGCs was provided when it was shown that the
H3K27me3 demethylase Utx is responsible, at least in part, for
the erasure of H3K27me3 in PGCs and the transcriptional activa-
tion of some pluripotency genes (Mansour et al., 2012).
PGCs are first formed as a small cluster (around 40 cells) of
Prdm1-expressing cells in the proximal epiblast at around
E7.25, and their specification and further fate are dependent
on the transcriptional regulators Prdm1,Prdm14, and Tcfap2c
(Magnu
´sdo
´ttir et al., 2012). These transcriptional regulators
appear to be important for the suppression of somatic cell fate
in PGCs, and Prdm14 is at least in part responsible for the induc-
tion of epigenetic reprogramming (Yamaji et al., 2008). Early
PGCs also express Nanog,Oct4, and Sox2, and pluripotent
stem cells (embryonic germ cells, EGCs) can be derived from
them (Surani et al., 2007).
Blastocyst-stage embryos including the inner cell mass are
globally hypomethylated, as a result of epigenetic reprogram-
ming during preimplantation development, but upon implanta-
tion rapid de novo methylation occurs primarily in the epiblast
(Howlett and Reik, 1991;Santos et al., 2002;Borgel et al.,
2010;Smith et al., 2012). Recent work indicates that early
PGCs express de novo methyltransferases just as other epiblast
cells do (Kurimoto et al., 2008) and show evidence of high levels
of methylation at E8.0–E8.5 by immunofluorescence (IF) with
a 5-methylcytosine (5mC) antibody or by bisulfite sequencing
of some candidate loci (Seki et al., 2005;Guibert et al., 2012).
Based on a reduction of the IF signal after E8.5, it was proposed
that genome-wide loss of methylation occurs relatively early,
during the migration phase, in PGC development (Seki et al.,
2005). Other studies, however, have shown that many individual
sequences analyzed by bisulfite sequencing, including some
differentially methylated regions (DMRs) in imprinted genes,
were demethylated relatively late once PGCs were colonizing
the gonads (Hajkova et al., 2002;Lee et al., 2002;Maatouk
et al., 2006;Guibert et al., 2012;Hackett et al., 2012a). It is
thus unclear how the dynamics of demethylation are orches-
trated across the whole genome and potentially across different
stages of PGC development.
Knowledge of the mechanisms of demethylation during PGC
development is also still in its infancy. The de novo methyltrans-
ferases Dnmt3a and Dnmt3b as well as Np95 (also known as
Molecular Cell 48, 849–862, December 28, 2012 ª2012 Elsevier Inc. 849
D
0
20
40
60
80
100
% CG methylation
13.5
♀
13.5
♂
11.5
♂♀
6.5 epi
♂♀
10.5
♂♀
9.5
♂♀
whole genome
mat DMR
pat DMR
CGI
A
0
20
40
60
80
100
J1 13.5
♀
13.5
♂
16.5
♀
16.5
♂
11.5
♂♀
6.5 epi
♂♀
10.5
♂♀
9.5
♂♀
PGC
replicate 1
replicate 2
% CG methylation
B
75 - 100
50 - 75
25 - 50
0 - 25
% methylation
0
20
40
60
80
100
% of all probes N=400885
J1 13.5
♀
13.5
♂
16.5
♀
16.5
♂
11.5
♂♀
6.5
epi
♂♀
10.5
♂♀
9.5
♂♀
PGC
5kb probes 5kb probes
C
02468
0
20
40
60
80
% CG density
16.5 ♀
16.5 ♂
13.5 ♀
13.5 ♂
11.5 ♂♀
10.5 ♂♀
9.5 ♂♀
6.5 epi ♂♀
J1
% CG methylation
E
Igf2r
CGI
Location on chromosome 17
Airn
Nanog
% CG methylation
Location on chromosome 6
0
100
0
100
100
0
100
0
100
0
122656000
122657000
122658000
122659000
122660000
122661000
H19/MiR675
CGI
11.5 ♂♀
10.5 ♂♀
13.5 ♂
9.5 ♂♀
6.5 epi ♂♀
PGC
DMR DMR
12933000
12934000
12935000
12936000
12937000
0
100
0
100
100
0
100
0
100
0
Location on chromosome 7
149763000
149764000
149765000
149766000
149767000
149768000
0
100
0
100
100
0
100
0
100
0
5kb probes
Figure 1. Demethylation Dynamics in PGCs
(A) Global CG methylation levels for each data set assessed by 5 kb tiling probes. Open circles represent eac h data point; lines represent the median value for the
two samples per time point. Note that the two replicates of the J1 data point are technical replicates, and all others are biological replicates from pooled samples.
(legend continued on next page)
Molecular Cell
Resetting DNA Methylation in Primordial Germ Cells
850 Molecular Cell 48, 849–862, December 28, 2012 ª2012 Elsevier Inc.
Uhrf1), which targets Dnmt1 to the DNA replication machinery to
maintain methylation during mitosis (Bostick et al., 2007;Sharif
et al., 2007), are downregulated in early PGCs (Kurimoto et al.,
2008), and components of the active demethylation pathways
such as the hydroxylase Tet1 and members of the base excision
repair pathway are expressed (Hajkova et al., 2010). This is
consistent with genetic studies which show that deficiency of
the deaminase Aid (Popp et al., 2010) or the glycosylase Tdg
(Cortellino et al., 2011) results in defects in methylation erasure
in PGCs. Hence, the current thinking is that a combination of
passive and active demethylation pathways is probably oper-
ating in PGCs, possibly in a context-dependent manner (Feng
et al., 2010;Saitou et al., 2011;Hackett et al., 2012b).
The biological purposes and outcomes of epigenetic reprog-
ramming in the germline are also not fully understood. Clearly,
parental imprints need to be reprogrammed for normal develop-
ment to occur in the next generation. Is reprogramming in PGCs
really linked to pluripotency, and if so, why? Is most epigenetic
information erased in germ cells so as to prevent the inheritance
of acquired epigenetic information across generations? And are
transposons resistant to reprogramming, or conversely, widely
expressed in germ cells because of reprogramming?
We recently initiated studies for the genome-wide mapping
of DNA methylation in PGCs using unbiased BS-Seq (Popp
et al., 2010). Further optimization of the technique allowed
us to include earlier stages of PGCs, and here we describe
a systematic study of BS-Seq and RNA-Seq of key stages of
PCG development, which provides a framework for the under-
standing of epigenetic reprogramming, pluripotency, and trans-
generational epigenetic inheritance.
RESULTS
PGCs are induced by external signals in the epiblast around E6.5
and first arise as a small group of about 40 cells in the proximal
epiblast at E7.25 (Saitou, 2009). We therefore decided to profile
E6.5 epiblast cells, as these are expected to have a primed
epigenetic state characteristic of nascent PGCs. At E9.5, a small
population of about 200 PGCs starts to migrate through the
hindgut endoderm and reaches the gonadal anlagen at E10.5–
E11.5 (Saitou, 2009). Using an Oct4-Gfp transgene (on
a C57Bl/6J background) (Yoshimizu et al., 1999), we isolated
PGCs at E9.5, E10.5, E11.5, E13.5, and E16.5. For each time
point, PGCs from 10–30 embryos were pooled, and at E13.5
and E16.5, male and female PGCs were profiled separately.
BS-Seq libraries were prepared from two independent samples
of each time point, and two independent sequencing runs for
a J1 embryonic stem cell (ESC) (129S4/SvJae) BS-Seq library
were performed as well (Figure 1A). To assess bisulfite conver-
sion efficiency, we measured CHH (H = C/A/T) methylation levels
for 1 kb tiling probes across the genome and found that more
than 60% of all 1 kb probes for each sample revealed 100%
conversion, indicating high conversion efficiency (see Figure S1A
online). RNA-Seq libraries were prepared from one pooled
sample per time point. Table S1 summarizes the outcomes of
the Illumina sequencing runs of all BS-Seq and RNA-Seq
libraries.
Methylation Erasure Occurs in Two Distinct Phases
In the E6.5 epiblast, the overall methylation level at CG dinucle-
otides was 71% similar to the values observed for J1 ESCs (74%)
and to values reported for somatic and ESCs and the E6.5 post-
implantation embryo (Hajkova et al., 2002;Lister et al., 2009;
Laurent et al., 2010;Stadler et al., 2011;Smith et al., 2012)(Fig-
ure 1A). In E9.5 PGCs, methylation levels were already reduced
to 30%, which means that the bulk of methylation erasure in
PGCs occurs prior to E9.5. This is in line with previous reports
using IF and locus-specific bisulfite sequencing (Seki et al.,
2005;Guibert et al., 2012) but differs from the expectation that
global erasure of methylation marks occurs concomitantly with
imprint erasure from E11.5 to E13.5 (Reik et al., 2001). From
E9.5, methylation levels were reduced gradually to about 15%
in E11.5 PGCs, with a further drop to 14% and 7% in male and
female E13.5 PGCs, respectively. The global loss of methylation
affects all methylation levels (Figure 1B) and is mirrored by the
loss of the correlation between CG density and methylation
levels across all time points (Figure 1C). It is noteworthy that
no de novo methylation was observed between E6.5 and E13.5
in any of the PGC samples, indicating that global demethylation
is a unidirectional process (Figure S1B). In female PGCs the low
levels of methylation at E13.5 persist to E16.5 with cells being in
meiotic arrest (Saitou, 2009), while male E16.5 PGCs show
evidence of robust de novo methylation with an increase to
about 50% methylation (Figures 1A and 1B, and Figure S1B).
The early phase of methylation erasure prior to E9.5 is truly
global, affecting promoters, CpG islands (CGIs) (Deaton and
Bird, 2011;Jones, 2012), introns, exons, and intergenic
sequences (Figure 2). Promoters of genes that are expressed
early in PGC development, such as Nanog, are demethylated
during this phase (Figure 1E). However, there are a number of
distinct sequence classes in which methylation marks are largely
maintained during this early phase of methylation loss, and de-
methylation of these regions is only completed once PGCs enter
the genital ridges from E10.5 (late demethylaters). These include
Note that E9.5 PGCs are already fairly hypomethylated followed by a further gradual loss of methylation toward E13.5. De novo methylation is only observed at
E16.5 in male PGCs. See also Figure S1 and Table S1.
(B) Distribution of CG methylation levels across the genome (5 kb probes). Note that the loss of methylation in PGCs is observed across the entire percentile
distribution.
(C) Correlation between CG density and methylation levels.
(D) Methylation levels for the whole genome (5 kb probes), CGIs, and maternal and paternal DMRs (DMR coordinates were taken from E12.5 embryos [Tomizawa
et al., 2011]). The dashed line indicates the expected 50% methylation levels for a germline DMR. Note that maternal DMRs retain more methylation than the rest
of the genome during global loss of methylation. Outliers are not shown.
(E) Example plots for Nanog promoter (left), maternally methylated Igf2r DMR (middle), and paternally methylated H19 DMR (right). Each bar represents a single
CG dinucleotide. Note that while the Nanog promoter shows early demethylation kinetics, CGIs within imprint DMRs undergo delayed demethylation. See also
Figure S1.
Molecular Cell
Resetting DNA Methylation in Primordial Germ Cells
Molecular Cell 48, 849–862, December 28, 2012 ª2012 Elsevier Inc. 851
DMRs of imprinted genes and particularly the maternal ones
(Figures 1D, 1E, and 2, and Figure S1C). Closer inspection shows
that sequences surrounding the CGIs in DMRs are also deme-
thylated relatively early and that it is specifically the CGIs that
are resistant to demethylation until the late stages of PGC devel-
opment (Figure 1E and Figure S1C). This effect is less
pronounced for paternal DMRs, which may be connected to their
lower CG content (Schulz et al., 2010).
J1 13.5
♀
13.5
♂
16.5
♀
16.5
♂
11.5
♂♀
6.5 epi
♂♀
10.5
♂♀
9.5
♂♀
PGC
CGI-containing promoter
Non-
CG
I promoter
Non-promoter CGI
Exon
Intron
Intergenic region
IAP
Maternal DMR
Paternal DMR
LINE1
Methylation levels
100%
0%
Figure 2. Methylation Heatmap for Various
Genomic Features
Each line represents a single probe within the
indicated feature. High methylation levels are
shown in red, low methylation levels are shown in
green. Of all features analyzed, IAPs seem to retain
most methylation at all time points, while all other
features undergo substantial reprogramming.
Another class of late demethylater
CGIs is found on the X chromosome (Fig-
ure 3A) (Brockdorff, 2011). The X-linked
delayed demethylation is specific to
CGIs, as the demethylation dynamics
for the X chromosome as a whole mirror
those of the genome globally (Figure S2A).
It is noteworthy that CGIs on the X chro-
mosome show elevated methylation
levels in the E6.5 epiblast, as this is
a pooled sample from male and female
cells and thus includes a reduced but
undefined number of inactivated X chro-
mosomes contributed by female cells
(Figure 3A). In an exclusively female
epiblast, we expect 50% methylation for
X linked CGIs, and this was observed
for two female epiblast stem cell lines,
which are derived from female epiblast
(T. Hore, personal communication). The
delayed demethylation kinetics for
X-linked CGIs is significant because while
it is known that early PGCs inherit
a randomly inactivated X chromosome
from the epiblast (Sugimoto and Abe,
2007), whether this involves methylation
of CGIs with subsequent demethylation
was unknown. Our data suggest that
methylation at CGIs on the X chromo-
some is actively maintained during global
methylation loss and results in a slow and
gradual demethylation pattern, which is
consistent with the gradual reactivation
of X-linked genes over a prolonged
period from E7.5 to E14.5 (Sugimoto
and Abe, 2007).
We next identified a group of promoter
CGIs that were demethylated with the
same delayed kinetics as DMRs and re-
tained more than 25% methylation for
each time point prior to E13.5 (Figure 3B and Table S2,Table
S3,Table S4, and Table S5). This cutoff was selected as it
includes all of the DMRs, which are largely resistant to demethy-
lation until E11.5, but excludes the vast majority of genomic
CGIs, which exhibit less than 25% methylation at these time
points (Figure 1D and Figure S2B). Notably, this group is associ-
ated with genes specifically involved in meiosis and gamete
generation (Figure 3D), which in general are only transcribed in
Molecular Cell
Resetting DNA Methylation in Primordial Germ Cells
852 Molecular Cell 48, 849–862, December 28, 2012 ª2012 Elsevier Inc.
germ cells and methylated in most if not all somatic tissues (data
not shown; Maatouk et al., 2006;Borgel et al., 2010;Hackett
et al., 2012a). As observed for DMRs and X-linked CGIs, CGs
in the neighborhood of these CGI promoters became demethy-
lated in early PGCs, while the CGIs themselves retained methyl-
ation until E11.5 and became demethylated thereafter (Figure 3C,
Figure S2C).
Thus, it seems that a select group of CGIs actively maintain
methylation marks during the global loss of methylation that
occurs in early PGCs. This is reminiscent of how methylation at
some DMRs is maintained during global methylation loss in the
early embryo by the zinc finger protein Zfp57 (Li et al., 2008).
We found that late-demethylating CGI promoters were also
substantially enriched for Zfp57 binding sites in ESCs (Quenne-
ville et al., 2011)(Figure S2D); this preliminary finding suggests
that Zfp57 might play a role in maintaining methylation marks
at some CGIs during global loss of methylation in PGC develop-
ment as it does for some DMRs during global methylation
erasure in the early embryo.
Mechanisms for Transgenerational Epigenetic
Inheritance
PGCs are in an extremely hypomethylated state at E13.5;
however, a small amount of methylation is retained. We exam-
ined how the remaining methylation at E13.5 is distributed
across the genome. We confirmed that as a sequence class
only intracisternal A particles (IAPs) remained substantially meth-
ylated across all stages analyzed, while other elements such as
LINE1s as well as SINEs retain small amounts of methylation at
E13.5 but are largely reprogrammed (Figure 4A, Figure S3A).
The resistance of IAPs against demethylation is particularly
true for the consensus sequence of the monomer repeat within
the long terminal repeat (LTR) of IAP1 and IAP2, two distinct
classes of these aggressively transposing elements, while the
50UTR of LINE1Tf and LINE1A elements, which have been exten-
sively studied in ESCs (Ficz et al., 2011), undergoes significant
demethylation (Figure 4A).
We investigated if any single-copy regions in the genome were
resistant to demethylation. We identified resistant CGIs and non-
CGI promoters that remained methylated (with a cutoff of 25%
methylation) in male or female E13.5 PGCs (numbers are shown
in Figure S3B). CGIs located close to an IAP showed consistently
high methylation levels throughout all developmental stages,
while CGIs without an IAP showed more variable resistance to
erasure (Figure 4B, see below). It is noteworthy that resistant
CGIs with an IAP were rare (Figure S3B) and that IAPs were
more frequently found at resistant non-CGI promoters (Figures
S3C and S4A). In fact, these non-CGI promoters are resistant
to demethylation as a function of their distance from the IAP (Fig-
ure S4B). This is in line with previous reports (Guibert et al., 2012)
and suggests that the genomic context or chromatin environ-
ment of IAPs can confer resistance to erasure on neighboring
elements. Alternatively, prevention of demethylation of IAPs
including their surrounding sequences may be a protective
mechanism of the genome to avoid activation of these potentially
mutagenic elements in the germline.
CGIs (and non-CGI promoters) that were not located close to
an IAP were more variable in their resistance to demethylation
(Figure 4B and Figure S5). However, some of these variably
erased CGIs (VECs) remained methylated at all stages, including
in mature oocytes and sperm (Figure S5A), for example a CGI in
the Exoc4 gene which is associated with type 2 diabetes and
involved in insulin-stimulated glucose transport (Inoue et al.,
2003)(Figure 4C). We have extended this analysis to a number
of publicly available data sets from sperm, oocyte, two-cell
embryo, ICM, and ESCs (Stadler et al., 2011;Kobayashi et al.,
2012;Smith et al., 2012)(Figure S5B). Importantly, we observe
that a substantial proportion of VECs retain significant methyla-
tion levels in various data sets, suggesting that VECs might be
carriers of epigenetic inheritance transgenerationally. Interest-
ingly, more of these CGIs were found methylated in sperm
than in oocyte (Figure S5), implying that there may be a bias
for such VECs to escape reprogramming in the male germline.
These CGIs may be candidates for short-term transgenerational
inheritance in mammals, which seems variable in its persistence
and hence heritability (Jimenez-Chillaron et al., 2009;Carone
et al., 2010;Ng et al., 2010;Daxinger and Whitelaw, 2012).
Complex Mechanisms of Demethylation
The dynamics of global methylation erasure observed in our
BS-Seq data sets shows that demethylation takes place over
a prolonged period from before E9.5 to E13.5, during which
PGCs undergo several cell divisions and hence cycles of DNA
replication (Seki et al., 2007). Thus, we investigated if DNA
demethylation in PGCs could be the result of a passive loss of
methylation due to a lack of methylation maintenance at DNA
replication. In such a scenario, methylation marks on the parental
strand do not get copied onto the newly synthesized strand re-
sulting in a hemimethylated product, which then becomes
further diluted by continued replication and eventually results
in complete hypomethylation.
To gain more detailed molecular insights into the dynamics of
demethylation, we carried out hairpin bisulfite high-throughput
sequencing of the LINE1Tf 50UTR. Hairpin bisulfite sequencing
keeps the two original DNA strands together, allowing an
assessment of full versus hemimethylation and demethylation
at each CG (Arand et al., 2012). There was a substantial amount
of hemimethylated CG sites in PGCs at E9.5 and E10.5, which
was then reduced to the fully unmethylated state by E13.5 (Fig-
ure 5A). It is noteworthy that within sequences that were found to
be hemimethylated, methylated CG dinucleotides were almost
exclusively located on the same strand, and instances of
hemimethylated sequences with methylated CG dinucleotides
on both strands were rare (Figure 5B and Figure S6). Over the
time course analyzed, the number of methylated CGs is drasti-
cally reduced toward E13.5, but the strand bias is preserved in
all data sets (Figure S6). These dynamics are consistent with
a predominantly passive demethylation mechanism with a minor
contribution by active mechanisms.
The entire de novo methylation system including Dnmt3a,
Dnmt3b, and Dnmt3L is transcriptionally silenced during this
period (Hajkova et al., 2002;Kurimoto et al., 2008), consistent
with our observations revealing a complete lack of de novo
methylation until E16.5 in male PGCs, at which stage Dnmt3a
and Dnmt3L show a burst of transcription (Figure S6A). Further-
more, while Dnmt1 is expressed (Figure 5C) and localized in the
Molecular Cell
Resetting DNA Methylation in Primordial Germ Cells
Molecular Cell 48, 849–862, December 28, 2012 ª2012 Elsevier Inc. 853
DGene ontology categories for CGI promoters > 25% methylation in
sexual reproduction
gamete generation
male gamete generation
spermatogenesis
meiosis I
meiosis
M phase of meiotic cell cycle
meiotic cell cycle
oogenesis
female gamete generation
sexual reproduction
N= 496 promoters
E9.5
male gamete generation
sexual reproduction
gamete generation
spermatogenesis
meiosis
N= 256 promoters
E10.5
male meiosis
sexual reproduction
sex differentiation
gamete generation
N= 131 promoters
E11.5
C
11.5 ♂♀
10.5 ♂♀
13.5 ♂
9.5 ♂♀
6.5 epi ♂♀
PGC
% CG methylation
Rhox13
CGI
0
100
0
100
0
100
0
100
0
100
Dazl
CGI
50432000
50432500
50433000
50433500
50434000
Location on chromosome 17
35474000
35475000
35476000
35477000
Location on chromosome X
0
100
0
100
100
0
100
0
100
0
B
0
20
40
60
80
100
13.5
♀
13.5
♂
11.5
♂♀
6.5 epi
♂♀
10.5
♂♀
9.5
♂♀
CGI promoters >25% methylated in
E6.5 epi
E9.5
E10.5
E11.5
PGC
% CG methylation
AX CGIGenome CGI
% CG methylation
0
5
10
15
20
25
30
PGC
J1 6.5 epi
♂♀
9.5
♂♀
10.5
♂♀
11.5
♂♀
13.5
♂
13.5
♀
J1 13.5
♀
13.5
♂
11.5
♂♀
6.5 epi
♂♀
10.5
♂♀
9.5
♂♀
PGC
meiotic cell cycle
Figure 3. CGIs with Late Demethylating Kinetics
(A) Demethylation kinetics of CGIs across the genome (left) or on the X chromosome (right). Note that X-linked CGIs undergo slower demethylation than CGIs for
the rest of the genome. Outliers are not shown. See also Figure S2.
(legend continued on next page)
Molecular Cell
Resetting DNA Methylation in Primordial Germ Cells
854 Molecular Cell 48, 849–862, December 28, 2012 ª2012 Elsevier Inc.
nucleus (Hajkova et al., 2002)(Figure 5D), Np95 is transcription-
ally downregulated (Kurimoto et al., 2008;Figure 5C), and impor-
tantly we find that the remaining protein seems to be largely
(B) Demethylation kinetics for CGI-containing promoters selected with >25% methylation in E6.5 epiblast and E9.5, E10.5, and E11.5 PGCs. These promoters
show consistently higher methylation levels across all time points analyzed. See also Table S2,Table S3,Table S4, and Table S5.
(C) Example plots for an X-linked CGI (left) and a CGI-containing promoter with slow demethylation kinetics. Methylation marks seem to be retained especially
around the CGI. Each bar represents a single CG dinucleotide. See also Figure S2.
(D) Gene ontology categories with a Bonferroni-corrected p value < 0.05 for promoters selected for >25% methylation in E9.5, E10.5, and E11.5 PGCs. CGI
promoters of genes selected for higher methylation levels between E9.5 and E11.5 seem to have a functional connection. The number of genes in each groupis
indicated.
C
Exoc4
CGI
% CG methylation
0
100
0
100
0
100
0
100
0
100
100
0
11.5 ♂♀
10.5 ♂♀
6.5 epi ♂♀
PGC
13.5 ♂
9.5 ♂♀
IAP
CGI
33219000
33220000
33221000
33222000
33223000
Location on chromosome 6
0
100
0
100
0
100
0
100
0
100
16.5 ♂
3753500
3754000
3754500
3755000
3755500
3756000
3756500
3757000
Location on chromosome 8
100
0
A
J1 13.5
♀
13.5
♂
16.5
♂
11.5
♂♀
6.5
epi
♂♀
10.5
♂♀
16.5
♀
9.5
♂♀
PGC
IAP1
IAP2
0
20
40
60
80
100
% CG methylation
0
20
40
60
80
100
L1A
L1Tf
J1 13.5
♀
13.5
♂
16.5
♂
11.5
♂♀
6.5
epi
♂♀
10.5
♂♀
16.5
♀
9.5
♂♀
PGC
B
selected > 25% in E13.5 ♂
0
20
40
60
80
100
% CG methylation
0
20
40
60
80
100
16.5
♂
16.5
♀
13.5
♀
13.5
♂
11.5
♂♀
6.5 epi
♂♀
10.5
♂♀
9.5
♂♀
PGC
CGIs <2kb to IAP
CGIs without IAP
selected > 25% in E13.5 ♀
13.5
♀
13.5
♂
16.5
♂
11.5
♂♀
10.5
♂♀
16.5
♀
6.5 epi
♂♀
9.5
♂♀
PGC
Figure 4. Demethylation Resistance
(A) Methylation levels for CG dinucleotides within
the consensus sequence of IAP1 (top left), IAP2
(bottom left), LINE1A (top right), and LINE1Tf
(bottom right). Note that these elements retain
substantial levels of methylation across all
time points. Outliers are not shown. See also
Figure S3.
(B) Methylation levels for resistant CGIs selected
with >25% methylation in E13.5 male (left) and
female PGCs (right) without an IAP in close
proximity (top) or near an IAP (bottom). Note that
resistant CGIs with a distance of <2 kb to an IAP
show consistently higher methylation levels
across all time points, while CGIs without the
presence of an IAP show variable methylation
levels across all time points. See also Figures
S3–S5.
(C) Example figures for resistant CGIs without
an IAP in close proximity (left) or with an IAP
nearby (right). Each bar represents a single CG
dinucleotide.
excluded from the nucleus in replicating
PGCs, while Dnmt1 is not as confirmed
by EdU staining (Figure 5D and Fig-
ure S7B). In ESCs, Np95 and Dnmt1
are both located in the nucleus, while
control stainings in Np95 KO cells show
no background staining for Np95 (Fig-
ure S7C). The predominant cytoplasmic
localization of Np95 in PGCs was
observed for all time points analyzed
(Figure S7D) and is independent of the
cell-cycle stage of these cells. This
apparent retention of Np95, but not of
Dnmt1, in the cytoplasm of PGCs
suggests that the canonical somatic
pathway for methylation maintenance,
which involves Dnmt1 targeting to the
replication fork by Np95, may be
disabled. Together with the lack of de
novo methylation, this could contribute
to the global loss of methylation in early
PGCs. At the same time, the presence
(and most likely noncanonical targeting)
of Dnmt1 presumably allows main-
tenance of methylation at DMRs of
imprinted genes and other sequences that undergo late deme-
thylation, which is strongly reminiscent of the maintenance of
methylation marks at DMRs in the early embryo during global
Molecular Cell
Resetting DNA Methylation in Primordial Germ Cells
Molecular Cell 48, 849–862, December 28, 2012 ª2012 Elsevier Inc. 855
B
A
CD
Figure 5. Mechanisms for Demethylation
(A) Hairpin bisulfite heatmap of LINE1Tf. Total methylation levels are shown at the top. For each time point analyzed, each column represents one CG dyad along
the LINE1Tf consensus sequence, and each row represents one sequencing read. The bars next to the heatmap represent the average distribution of fully
methylated, hemimethylated, and unmethylated sites. Note that E9.5 and E10.5 PGCs have high levels of hemimethylated sites, which are then reduce d to
almost-complete hypomethylation at E13.5.
(B) Shown is the distribution of methylated CG dinucleotides (meCG) at hemimethylated sites across the top (x axis) and bottom (y axis) strands of the LINE1Tf
consensus sequence assessed by hairpin bisulfite sequencing. The LINE1Tf consensus sequence contains five CG dinucleotides, and the numbers 0–5 on the
axis refer to the amount of meCGs on each strand and contain no position information. The values in the heat diagram represent the number of instances with the
respective number of meCGs observed on the top and bottom strands. Shown is a simulation of the distribution of meCG within hemimethyl ated sites in the case
of passive DNA demethylation (left) and active demethylation (middle). Note that with passive demethylation, all meCGs are located on the top strand, while the
bottom strand is completely unmethylated and contains 0 meCGs and vice versa. For active DNA demethylation, a strand-independent distribution was
simulated that leads to methylated and unmethylated CGs randomly distributed across both strands. The hairpin bisulfite data for E9.5 PGCs are shown in the
right panel, and there is a strong strand bias for meCGs toward either top or bottom strand highly similar to the outcome for the simulation of passive DNA
demethylation. Instances with meCGs distributed across both strands are rare in E9.5 PGCs. See also Figure S6.
(C) Expression analysis of the DNA methylation machinery. Single-cell microarray data for ESCs (Vincent et al., 2011) and PGCs (Kurimoto et al., 2008) were
reanalyzed (left, see the Experimental Procedures for more detail). RNA-Seq data for ESC and embryoid body (EB) (Cloonan et al., 2008) and PGCs of various time
points are shown on the right. Whiskers represent the interquartile range of variation between replicates. Note that while Dnmt1 is continuously expressed, Np95
is transcriptionally downregulated in early PGCs. See also Figure S7.
(D) Immunofluorescence staining for DNA (blue), Dnmt1 (green), and Np95 (red). Shown are immunostainings and RGB profiles created with Zeiss LSM software.
Scale bars represent 10 mM in all images, and where RGB profiles are shown, the red line across a cell represents the midline along which the signal intensity is
traced for each pixel and the profile is plotted below. Shown are stainings for Np95 and Dnmt1 in E13.5 male PGCs and E14 ESCs. In cycling PGCs, Dnmt1
localizes to the nucleus while Np95 is preferentially located in the cytoplasm. In ESCs, both Dnmt1 and Np95 localize to the nucleus. This suggests that in ESCs,
the subcellular localization of Dnmt1 and Np95 is linked during S phase, while this dynamic pattern may be uncoupled in PGCs. See also Figure S7.
Molecular Cell
Resetting DNA Methylation in Primordial Germ Cells
856 Molecular Cell 48, 849–862, December 28, 2012 ª2012 Elsevier Inc.
loss of methylation (Saitou et al., 2011). Tet1, among other
factors involved in active demethylation, is expressed during
both early and late PGC development at low levels (Figure S7A)
(Hajkova et al., 2010), and our analysis does not exclude the
presence of additional active demethylation pathways.
Reprogramming the Transcriptional Landscape of PGCs
The erasure of most 5mC from the genome raises questions of
transcriptional regulation. Are there large-scale transcriptional
activation and promiscuity? And importantly, is there a global
link between epigenetic reprogramming and pluripotency?
We first looked at the complexity of the RNA-Seq transcrip-
tome in PGCs in comparison to ESCs and somatic cells and
found no fundamental shift in complexity, meaning that similar
numbers of genes had high, intermediate, and low levels of tran-
scription in all cell types including PGCs (Figure 6A). Hence the
global loss of methylation at promoters (Figure 2) does not result
in a profound shift in transcriptional regulation, indicating that
a mechanism independent of DNA methylation promotes tran-
scriptional control in reprogramming PGCs. Similarly, a loss of
methylation over exons and introns (gene bodies) (Figure 2)
was not accompanied by any shift in the transcriptional profile
(Figure 6A). De novo methylation at E16.5 in male PGCs also
seemed to be independent of transcriptional changes from
E13.5 to E16.5 as the promoters of genes that increased or
decreased in expression became de novo methylated at similarly
high levels (Figure S8A). However, gene body methylation was
positively correlated with transcription in E16.5 male PGCs (Fig-
ure S8B). This suggests that DNA methylation and transcription
are largely uncoupled during methylation erasure in PGCs but
show some degree of positive correlation when genome-wide
methylation is restored, suggesting that the relationship between
DNA methylation and transcription is complex (Jones, 2012).
Next we defined clusters of genes with a highly similar expres-
sion profile across the different stages of PGC development (see
the Supplemental Experimental Procedures for details). We
discovered 12 clusters of transcripts that changed in consistent
ways over the time course analyzed (data not shown). The two
largest clusters with 26 and 49 genes, respectively, revealed
interesting sets of genes with functional importance for PGC
development (Figure 6B and Table S6). The first cluster was
highly enriched for transcription factors of the pluripotency
network, which are fully expressed at E11.5 with a steep decline
toward E16.5 (Figure 6B). The second cluster begins to be tran-
scribed as the pluripotency network declines and corresponds
to meiosis network genes (Figure 6B). Expression of these tran-
scripts is particularly high in female PGCs from E13.5, which
arrest in meiotic prophase at that time (Bowles and Koopman,
2010)(Figure 6B). Notably, the pluripotency cluster that we iden-
tified is particularly enriched for Tet1 targets (based on transcrip-
tomics in Tet1 knockdown or knockout ESCs and on Tet1 ChIP-
seq data in ESCs) (Figure 6B) (Dawlaty et al., 2011;Ficz et al.,
% of all reads
0.000
0.005
0.010
0.015
0.020
0.025
L1A L1Tf Cnpy3 Pdia5
ESC
MLF
J1
TKO
11.5 ♂♀
13.5 ♂
13.5 ♀
16.5 ♂
16.5 ♀
PGCs
C
AVery high expression (Log2 RPKM > 4)
High expression (Log2 RPKM 0 to 4)
Low expression (Log2 RPKM -4 to 0)
No expression (Log2 RPKM < -4)
0
20
40
60
80
100
ES 13.5
♀
13.5
♂
16.5
♀
16.5
♂
11.5
♂♀
EB
% of all transcripts
(N=51723)
BMale
Female
−2
−1
0
1
2
13.5 16.511.5ESEB
PGC
Relative RPKM
Pluripotency network (*Tet1 target):
*Prdm14, Lefty2, Lefty1, Tcfap2c,
Sox2, *Gjb3, Lin28a, *Nanog,
Cdh1, *Ecat1, *Tet1, *Esrrb, *Tcl1
N=26
−2
0
2
4
13.5 16.511.5ESEB
PGC
N=49
Meiosis network:
Mei1, Rec8, Ccnb3, Prdm9, Stra8,
Msh5, Sycp2, Smc1b, Foxl2,
Tex15, Dmrtc2, Fst, Fgfr3, Dnahc8,
Figure 6. Dynamics of Transcriptomic
Reprogramming
(A) Distribution of expression values for the PGC
RNA-Seq data sets. ESC and EB data sets were
included for comparison (Cloonan et al., 2008).
Even the most hypomethylated samples with
E13.5 PGCs have an orderly expression program
similar to that of all other PGC samples and the
ESC and EB data sets.
(B) Expression profiles for the pluripotency cluster
(left) and the meiosis cluster (right). Error bars
represent the standard deviation of measure
across all probes within the cluster, and example
genes are shown underneath for each cluster. Tet1
targets within the pluripotency cluster are high-
lighted by an asterisk. See also Figure S8 and
Table S6.
(C) Expression of LINE1s. Shown is the
percentage of all RNA-Seq reads that map to the
LINE1Tf and LINE1A consensus sequence and
also as a comparison to the sequence of Cnpy3
and Pdia5, two single-copy genes that are ex-
pressed at constant levels across the time course.
Note that the two LINE1 elements have higher
expression levels at E16.5 in female PGCs than in
any other data set and also than the two single-
copy genes. Results for RNA-Seq data from
ESCs and mouse lung fibroblasts (MLFs) (Guttman
et al., 2010), J1 (Ficz et al., 2011), and Dnmt TKO
ESCs (KO for Dnmt1,Dnmt3a, and Dnmt3b [Karimi
et al., 2011]) are shown for comparison.
Molecular Cell
Resetting DNA Methylation in Primordial Germ Cells
Molecular Cell 48, 849–862, December 28, 2012 ª2012 Elsevier Inc. 857
2011;Williams et al., 2011;Wu et al., 2011). Interestingly, the
promoters of these transcripts associated with genes such as
Nanog and Prdm14 are methylated in E6.5 epiblast cells, and
all become demethylated in PGCs by E9.5 (Figure 1E and Fig-
ure S8C), indicating that demethylation of these promoters
may be connected with their activity in early PGCs. Transcription
of the pluripotency network is then collectively silenced as
female PGCs go into meiotic arrest and male PGCs into mitotic
arrest around E13.5 (Bowles and Koopman, 2010). Expression
of these genes is replaced by the network of meiosis- and
germ-cell-function-related genes further driving PGCs toward
germ cell fate (Figure 6B).
Lastly, we were interested to see if the substantial demethyla-
tion in LINE1 elements resulted in their transcriptional activation.
Surprisingly, demethylation did not lead to general transcrip-
tional activation of LINE1s in PGCs by E13.5 (Figure 6C).
However, there was a specific transcriptional burst of LINE1
elements exclusively in female E16.5 PGCs, consistent with
the possibility that LINE1 particles persist during oogenesis,
leading to transposition events in early embryos (Figure 6C)
(Kano et al., 2009). Nonetheless, it is unclear why this activation
does not take place at E13.5, as methylation levels in female
PGCs at E13.5 and E16.5 are similarly low. It seems that expres-
sion of repetitive elements does not consistently show an
inverse correlation to DNA methylation, and additional mecha-
nisms other than DNA methylation are in place to regulate
LINE1 expression.
DISCUSSION
We have carried out a systematic study of genome-wide DNA
methylation (BS-Seq) and transcription (RNA-Seq) across key
stages of PGC development during which epigenetic reprogram-
ming takes place. A similar study of methylation reprogramming
in preimplantation embryos using RRBS-Seq has been recently
published (Smith et al., 2012). Together these studies provide
an advanced framework for the understanding of the dynamics
of reprogramming in embryonic development and their biological
outcomes. Our work provides four key insights. First, it defines
two phases of demethylation in PGCs, global demethylation
occurring early during their migration with the methylation of
specific regions being actively maintained, and a second phase
which occurs upon entry into the genital ridges and affects
sequences carrying epigenetic memory. Second, global DNA
demethylation in PGCs is consistent with contribution from
a passive mechanism supplemented by active maintenance of
methylation in specific regions, which ceases upon arrival in
the gonads. Third, global erasure of methylation does not lead
to promiscuous transcription including that of retrotransposons;
instead the core pluripotency network is expressed at early
stages of PGC development and is then replaced by expression
of a meiosis and germ cell development network. Finally, we
identify VECs that may act as carriers of short-term transgenera-
tional epigenetic inheritance in mammals.
An important question that arises from the early demethylation
dynamics of PGCs is whether these cells have somatic methyla-
tion levels to begin with. Earlier work using an antibody against
5mC to visualize DNA methylation suggested that E8.0 PGCs
retain a signal intensity comparable to somatic cells, which
diminishes subsequent to this stage (Seki et al., 2005). Bisulfite
sequencing analysis of individual loci showed that E8.5 PGCs
retain high levels of methylation at certain loci (Guibert et al.,
2012). In addition, the presence of hemimethylated sites in
LINE1 elements at E9.5 implies that these elements have under-
gone demethylation and thus are likely to have started out from
epiblast-like methylation levels. This body of evidence strongly
suggests that the earliest PGCs emerging in the E7.25 epiblast
inherit a highly methylated genome characteristic of epiblast
cells.
Indiscriminate genome-wide loss of methylation occurs early
in PGC development and is accompanied by the transcriptional
downregulation of the de novo methyltransferases (Dnmt3a,b,L)
and also seems to involve the impairment of the methylation
maintenance factor Np95. By contrast, DMRs in imprinted
genes, CGI promoters of germ-cell-specific genes, and CGIs
on the X chromosome have their methylation largely maintained
during global methylation loss, and demethylation of these
sequences is only completed once PGCs have entered the
genital ridges. Interestingly, this suggests that the mechanisms
of demethylation in PGCs and in preimplantation embryos share
similarities including passive and active demethylation, with
perhaps a key difference being the continuing protection from
demethylation of imprinted DMRs by Zfp57 in preimplantation
embryos and ESCs, which is lacking in PGCs (Li et al., 2008;
Quenneville et al., 2011). Also, it is unclear at this point if methyl-
ation present in the PGC founder population is first converted
into 5-hydroxymethylcytosine (5hmC) and then lost by subse-
quent passive demethylation, as BS-Seq data sets do not distin-
guish between 5mC and 5hmC, and current techniques that
allow for this distinction require amounts of input material that
are not currently applicable to PGCs (Booth et al., 2012;Yu
et al., 2012). In addition, other demethylation mechanisms
involving factors such as Aid and Tdg have been shown to play
a role in DNA methylation reprogramming in PGCs (Popp et al.,
2010;Cortellino et al., 2011), suggesting that active and passive
mechanisms of demethylation work in concert to ensure robust
epigenetic reprogramming in PGCs (Feng et al., 2010;Saitou
et al., 2011;Hackett et al., 2012b).
Global demethylation in PGCs is not associated with promis-
cuous transcriptional activation. Indeed, LINE1 elements, which
have been substantially demethylated by E13.5, are not tran-
scribed at that stage, suggesting that other mechanisms for
transcriptional repression of retrotransposons are in place,
such as those provided by Setdb1 and Kap1 in ESCs (Rowe
et al., 2010;Karimi et al., 2011). Early PGCs transcribe Oct4,
Nanog, and slightly later, Sox2, consistent with the possibility
that they activate at least part of the pluripotency transcription
factor network (Surani et al., 2007). Indeed, our transcriptome
analysis shows that from E11.5 to E13.5 the core pluripotency
network is fully transcribed at similar levels as in ESCs, consis-
tent with the capability of deriving EGCs from these stages of
PGC development. Activation of the pluripotency network is
associated with promoter demethylation (from E6.5 epiblast
cells to E9.5 PGCs) and with demethylation of H3K27me3 by
the histone demethylase Utx (Mansour et al., 2012). Without
any change in genomic methylation patterns, this transcriptional
Molecular Cell
Resetting DNA Methylation in Primordial Germ Cells
858 Molecular Cell 48, 849–862, December 28, 2012 ª2012 Elsevier Inc.
program is extinguished by E16.5 and replaced by the meiosis
network, especially in female PGCs (which are in meiotic
prophase arrest). How the pluripotency network is silenced
and the meiosis network activated in such a coordinated fashion
remains to be elucidated.
Why is the full pluripotency network activated in PGCs when
these cells subsequently undergo a defined differentiation
program rather than the pluripotential one of the ICM cells?
We suggest that while epigenetic reprogramming is tightly con-
nected with the activation of the pluripotency network in PGCs
and early embryos, similarly the expression of the pluripotency
network may be linked to demethylation of some of the targets
described here. Notably, the pluripotency network expressed in
PGCs is enriched for Tet1 targets, and Tet1 itself could be
responsible for demethylation of these factors. This is sup-
ported by the fact that these promoters are almost completely
demethylated by E9.5, but further analysis of earlier stages is
needed to confirm if these promoters become demethylated
with even faster kinetics than the rest of the genome. These
factors may have evolved to become demethylated by faster
and more targeted mechanisms than passive loss of methyla-
tion. Hence pluripotency and reprogramming appear to be inex-
tricably linked in PGCs as suggested for ESCs (Ficz et al.,
2011).
While most DNA methylation is erased by E13.5, there
are some notable exceptions. First, IAPs are the class of
sequences most resistant to demethylation, as previously
observed (Lane et al., 2003;Guibert et al., 2012), consistent
with IAPs being the evolutionarily most recently acquired trans-
poson family in the mouse genome, which is still potentially very
active and hence needs to be suppressed by methylation in the
germline. This property explains the transgenerational epige-
netic inheritance of the viable yellow (A
vy
) and axin-fused
(Axin
Fu
) mutant alleles in the mouse, which have arisen by inser-
tion of an IAP LTR into the agouti or fused gene, respectively
(Morgan et al., 1999;Rakyan et al., 2003). Indeed, CGIs in the
neighborhood of an IAP (up to 2 kb away) are resistant to
erasure. Importantly, we identified a number (89 and 176 in
male and female PGCs, respectively) of CGIs outside of an
IAP context in which DNA methylation was incompletely erased
at E13.5 (Figure 4C shows the example of the Exoc4 gene
which is associated with type 2 diabetes and involved in
insulin-stimulated glucose transport [Inoue et al., 2003]). Most
of these CGIs are variably erased, meaning that their extent
of erasure differs between stages, the sexes, and potentially
between individuals. The molecular mechanism of transgenera-
tional epigenetic inheritance is not known, but there are several
examples of epigenetic heritability through the male germline
(Daxinger and Whitelaw, 2012), which might be consistent
with our observation that VECs are more resistant to erasure
in male than in female PGCs. Characteristically, this type of
epigenetic inheritance shows variable penetrance, and the
phenotype is frequently lost after a short number of generations
(Daxinger and Whitelaw, 2012). This makes VECs interesting
candidates for transgenerational epigenetic inheritance of
induced metabolic phenotypes, and perhaps more generally,
for variations in phenotype that are not predicted by genotype
(intangible variation).
EXPERIMENTAL PROCEDURES
Sample Collection
All embryonic samples for library preparation were collected from timed
matings of C57Bl/6J female mice. Embryos collected for the E6.5 epiblast
samples were isolated and mechanically dissected, separating away all extra-
embryonic tissues, and pooled prior to DNA and RNA isolation. PGCs were
isolated from timed mated females carrying the Oct4-Gfp transgene ex-
pressed in the developing gonad (Yoshimizu et al., 1999) on a C57Bl/6J back-
ground. PGCs from 10–30 embryos were pooled for each time point, and final
PGC numbers ranged from 800 to 40,000. For E13.5 and E16.5 PGCs, male
and female samples were collected separately, as gonads can be readily
distinguished morphologically from E13.5. PGC samples were collected
following collagenase digestion using a FACSAria cell sorter with >98% purity.
J1 ESCs (129S4/SvJae) were grown on feeder cells under standard conditions
as described previously (Ficz et al., 2011). Animal work carried out as part of
this study is covered by a project license (to W.R.) under the 1986 animal
(scientific procedures) act, and is further regulated by the Babraham Institute
Animal Welfare, Experimentation, and Ethics Committee.
BS-Seq Library Prep
The amount of input material for the BS-Seq libraries was between 5 ng and
50 ng genomic DNA. The input DNA was sonicated, and end repair and
A-tailing were performed using the NEB Next kit according to the manufac-
turers’ instructions. Illumina’s Early Access Methylation Adaptor Oligo Kit
was used for the adaptor ligation. The adaptor-ligated DNA was treated with
sodium-bisulfite using the Imprint DNA Modification Kit from Sigma-Aldrich
according to the manufacturer’s instructions for the two-step protocol.
Bisulfite-treated DNA was amplified using PfuTurbo Cx Hotstart DNA
Polymerase from Agilent Technologies with 14–18 cycles depending on the
input amount. Size selection was performed by gel extraction for DNA
fragments between 200 bp and 250 bp.
RNA-Seq Library Prep
Between 30 ng and 100 ng total RNA was DNase treated with Ambion’s DNA-
free Kit according to the manufacturer’s instructions. Enrichment for mRNA
was performed using Dynabeads Oligo(dT)
25
from Invitrogen in two subse-
quent steps of purification with fresh beads. The isolated mRNA was frag-
mented and converted into cDNA. For the library preparation, the NEB Next
kit was used according to the manufacturers’ instructions. Illumina PE
adapters were ligated onto the end-repaired and A-tailed cDNA. Libraries
were amplified with 12 cycles and size selected by gel extraction for fragments
between 200 bp and 250 bp.
Immunofluorescence
Antibody staining against Np95 (Th10, gift from Haruhiko Koseki) and Dnmt1
(sc-20701, Santa Cruz Biotechnology) was performed as previously described
(Santos et al., 2003) with modifications. PGCs were identified either by the
presence of Oct4-Gfp or by staining for Stella. EdU incorporation was
achieved by incubating gonads before staining (as per the manufacturer’s
instructions–Invitrogen). Single optical sections were captured with a Zeiss
LSM510 Meta microscope (633oil immersion objective).
Hairpin BS with PGC Collection
For the hairpin bisulfite analysis, PGCs were isolated from Oct4-Gfp transgenic
embryos (Yoshimizu et al., 1999) at the desired time points (E9.5–E13.5).
Isolated genital ridges were trypsinized, and single GFP-positive cells were
collected manually using inverted fluorescence microscope Zeiss AxioVert
200M and micromanipulators TransferMan NK2 (Eppendorf). Each sample
contained at least 40 PGCs. Hairpin bisulfite sequencing for LINE1Tf 50UTR
was carried out on a 454 sequencing platform as described previously (Arand
et al., 2012).
DNA Sequencing
Libraries were sequenced on either an Illumia GAIIx or an Illumina HiSeq using
the default RTA analysis software. See Table S1 for the outcomes of the
sequencing runs.
Molecular Cell
Resetting DNA Methylation in Primordial Germ Cells
Molecular Cell 48, 849–862, December 28, 2012 ª2012 Elsevier Inc. 859
Data Analysis
Computational methods are described in the Supplemental Experimental
Procedures.
ACCESSION NUMBERS
Raw data from all libraries were deposited with the European Nucleotide
Archive under the accession number ERP001953.
SUPPLEMENTAL INFORMATION
Supplemental Information includes eight figures, six tables, Supplemental
Experimental Procedures, and Supplemental References and can be found
with this article at http://dx.doi.org/10.1016/j.molcel.2012.11.001.
ACKNOWLEDGMENTS
We would like to thank Heather Burgess and Laura Biggins for help with the
analysis of the microarray expression data and the hairpin bisulfite data. We
also thank Haruhiko Koseki for the Th10 antibody. We also would like to thank
Miguel Branco, Tim Hore, and Heather Lee for suggestions on the manuscript
and Stephan Beck for experimental advice. We would like to acknowledge
Victoria Hansford, Sophie Messager, David Jackson, and Kristina Tabbada
for their help with Illumina sequencing and members of the FACS facility for
their support. We would like to thank all Reik lab members for discussion
and our funding bodies for their generous support: Boehringer Ingelheim
Fonds (S.S.), BBSRC, MRC, Wellcome Trust, EU EpiGeneSys and
BLUEPRINT (W.R.).
Received: August 22, 2012
Revised: October 4, 2012
Accepted: November 1, 2012
Published: December 6, 2012
REFERENCES
Arand, J., Spieler, D., Karius, T., Branco, M.R., Meilinger, D., Meissner, A.,
Jenuwein, T., Xu, G., Leonhardt, H., Wolf, V., and Walter, J. (2012). In vivo
control of CpG and non-CpG DNA methylation by DNA methyltransferases.
PLoS Genet. 8, e1002750. http://dx.doi.org/10.1371/journal.pgen.1002750.
Booth, M.J., Branco, M.R., Ficz, G., Oxley, D., Krueger, F., Reik, W., and
Balasubramanian, S. (2012). Quantitative sequencing of 5-methylcytosine
and 5-hydroxymethylcytosine at single-base resolution. Science 336,
934–937.
Borgel, J., Guibert, S., Li, Y., Chiba, H., Schu
¨beler, D., Sasaki, H., Forne
´, T.,
and Weber, M. (2010). Targets and dynamics of promoter DNA methylation
during early mouse development. Nat. Genet. 42, 1093–1100.
Bostick, M., Kim, J.K., Este
`ve, P.-O., Clark, A., Pradhan, S., and Jacobsen,
S.E. (2007). UHRF1 plays a role in maintaining DNA methylation in mammalian
cells. Science 317, 1760–1764.
Bowles, J., and Koopman, P. (2010). Sex determination in mammalian germ
cells: extrinsic versus intrinsic factors. Reproduction 139, 943–958.
Brockdorff, N. (2011). Chromosome silencing mechanisms in X-chromosome
inactivation: unknown unknowns. Development 138, 5057–5065.
Carone, B.R., Fauquier, L., Habib, N., Shea, J.M., Hart, C.E., Li, R., Bock, C., Li,
C., Gu, H., Zamore, P.D., et al. (2010). Paternally induced transgenerational
environmental reprogramming of metabolic gene expression in mammals.
Cell 143, 1084–1096.
Cloonan, N., Forrest, A.R.R., Kolle, G., Gardiner, B.B.A., Faulkner, G.J.,
Brown, M.K., Taylor, D.F., Steptoe, A.L., Wani, S., Bethel, G., et al. (2008).
Stem cell transcriptome profiling via massive-scale mRNA sequencing. Nat.
Methods 5, 613–619.
Cortellino, S., Xu, J., Sannai, M., Moore, R., Caretti, E., Cigliano, A., Le Coz, M.,
Devarajan, K., Wessels, A., Soprano, D., et al. (2011). Thymine DNA glycosy-
lase is essential for active DNA demethylation by linked deamination-base
excision repair. Cell 146, 67–79.
Dawlaty, M.M., Ganz, K., Powell, B.E., Hu, Y.-C., Markoulaki, S., Cheng, A.W.,
Gao, Q., Kim, J., Choi, S.-W., Page, D.C., and Jaenisch, R. (2011). Tet1 is
dispensable for maintaining pluripotency and its loss is compatible with
embryonic and postnatal development. Cell Stem Cell 9, 166–175.
Daxinger, L., and Whitelaw, E. (2012). Understanding transgenerational epige-
netic inheritance via the gametes in mammals. Nat. Rev. Genet. 13, 153–162.
Deaton, A.M., and Bird, A. (2011). CpG islands and the regulation of transcrip-
tion. Genes Dev. 25, 1010–1022.
Feng, S., Jacobsen, S.E., and Reik, W. (2010). Epigenetic reprogramming in
plant and animal development. Science 330, 622–627.
Ficz, G., Branco, M.R., Seisenberger, S., Santos, F., Krueger, F., Hore, T.A.,
Marques, C.J., Andrews, S., and Reik, W. (2011). Dynamic regulation of
5-hydroxymethylcytosine in mouse ES cells and during differentiation.
Nature 473, 398–402.
Guibert, S., Forne
´, T., and Weber, M. (2012). Global profiling of DNA
methylation erasure in mouse primordial germ cells. Genome Res. 22,
633–641.
Guttman, M., Garber, M., Levin, J.Z., Donaghey, J., Robinson, J., Adiconis, X.,
Fan, L., Koziol, M.J., Gnirke, A., Nusbaum, C., et al. (2010). Ab initio recon-
struction of cell type-specific transcriptomes in mouse reveals the conserved
multi-exonic structure of lincRNAs. Nat. Biotechnol. 28, 503–510.
Hackett, J.A., Reddington, J.P., Nestor, C.E., Dunican, D.S., Branco, M.R.,
Reichmann, J., Reik, W., Surani, M.A., Adams, I.R., and Meehan, R.R.
(2012a). Promoter DNA methylation couples genome-defence mechanisms
to epigenetic reprogramming in the mouse germline. Development 139,
3623–3632.
Hackett, J.A., Zylicz, J.J., and Surani, M.A. (2012b). Parallel mechanisms of
epigenetic reprogramming in the germline. Trends Genet. 28, 164–174.
Hajkova, P., Erhardt, S., Lane, N., Haaf, T., El-Maarri, O., Reik, W., Walter, J.,
and Surani, M.A. (2002). Epigenetic reprogramming in mouse primordial germ
cells. Mech. Dev. 117, 15–23.
Hajkova, P., Ancelin, K., Waldmann, T., Lacoste, N., Lange, U.C., Cesari, F.,
Lee, C., Almouzni, G., Schneider, R., and Surani, M.A. (2008). Chromatin
dynamics during epigenetic reprogramming in the mouse germ line. Nature
452, 877–881.
Hajkova, P., Jeffries, S.J., Lee, C., Miller, N., Jackson, S.P., and Surani, M.A.
(2010). Genome-wide reprogramming in the mouse germ line entails the
base excision repair pathway. Science 329, 78–82.
Howlett, S.K., and Reik, W. (1991). Methylation levels of maternal and
paternal genomes during preimplantation development. Development 113,
119–127.
Inoue, M., Chang, L., Hwang, J., Chiang, S.-H., and Saltiel, A.R. (2003). The
exocyst complex is required for targeting of Glut4 to the plasma membrane
by insulin. Nature 422, 629–633.
Jimenez-Chillaron, J.C., Isganaitis, E., Charalambous, M., Gesta, S.,
Pentinat-Pelegrin, T., Faucette, R.R., Otis, J.P., Chow, A., Diaz, R.,
Ferguson-Smith, A., and Patti, M.E. (2009). Intergenerational transmission
of glucose intolerance and obesity by in utero undernutrition in mice.
Diabetes 58, 460–468.
Jones, P.A. (2012). Functions of DNA methylation: islands, start sites, gene
bodies and beyond. Nat. Rev. Genet. 13, 484–492.
Kano, H., Godoy, I., Courtney, C., Vetter, M.R., Gerton, G.L., Ostertag, E.M.,
and Kazazian, H.H., Jr. (2009). L1 retrotransposition occurs mainly in embryo-
genesis and creates somatic mosaicism. Genes Dev. 23, 1303–1312.
Karimi, M.M., Goyal, P., Maksakova, I.A., Bilenky, M., Leung, D., Tang, J.X.,
Shinkai, Y., Mager, D.L., Jones, S., Hirst, M., and Lorincz, M.C. (2011). DNA
methylation and SETDB1/H3K9me3 regulate predominantly distinct sets of
genes, retroelements, and chimeric transcripts in mESCs. Cell Stem Cell 8,
676–687.
Kobayashi, H., Sakurai, T., Imai, M., Takahashi, N., Fukuda, A., Yayoi, O., Sato,
S., Nakabayashi, K., Hata, K., Sotomaru, Y., et al. (2012). Contribution of
Molecular Cell
Resetting DNA Methylation in Primordial Germ Cells
860 Molecular Cell 48, 849–862, December 28, 2012 ª2012 Elsevier Inc.
intragenic DNA methylation in mouse gametic DNA methylomes to establish
oocyte-specific heritable marks. PLoS Genet. 8, e1002440. http://dx.doi.
org/10.1371/journal.pgen.1002440.
Kurimoto, K., Yabuta, Y., Ohinata, Y., Shigeta, M., Yamanaka, K., and Saitou,
M. (2008). Complex genome-wide transcription dynamics orchestrated by
Blimp1 for the specification of the germ cell lineage in mice. Genes Dev. 22,
1617–1635.
Lane, N., Dean, W., Erhardt, S., Hajkova, P., Surani, A., Walter, J.R., and
Reik, W. (2003). Resistance of IAPs to methylation reprogramming may
provide a mechanism for epigenetic inheritance in the mouse. Genesis 35,
88–93.
Laurent, L., Wong, E., Li, G., Huynh, T., Tsirigos, A., Ong, C.T., Low, H.M.,
Kin Sung, K.W., Rigoutsos, I., Loring, J., and Wei, C.L. (2010). Dynamic
changes in the human methylome during differentiation. Genome Res. 20,
320–331.
Lee, J., Inoue, K., Ono, R., Ogonuki, N., Kohda, T., Kaneko-Ishino, T., Ogura,
A., and Ishino, F. (2002). Erasing genomic imprinting memory in mouse clone
embryos produced from day 11.5 primordial germ cells. Development 129,
1807–1817.
Li, X., Ito, M., Zhou, F., Youngson, N., Zuo, X., Leder, P., and Ferguson-Smith,
A.C. (2008). A maternal-zygotic effect gene, Zfp57, maintains both maternal
and paternal imprints. Dev. Cell 15, 547–557.
Lister, R., Pelizzola, M., Dowen, R.H., Hawkins, R.D., Hon, G., Tonti-Filippini,
J., Nery, J.R., Lee, L., Ye, Z., Ngo, Q.-M., et al. (2009). Human DNA methyl-
omes at base resolution show widespread epigenomic differences. Nature
462, 315–322.
Maatouk, D.M., Kellam, L.D., Mann, M.R.W., Lei, H., Li, E., Bartolomei, M.S.,
and Resnick, J.L. (2006). DNA methylation is a primary mechanism for
silencing postmigratory primordial germ cell genes in both germ cell and
somatic cell lineages. Development 133, 3411–3418.
Magnu
´sdo
´ttir, E., Gillich, A., Grabole, N., and Surani, M.A. (2012).
Combinatorial control of cell fate and reprogramming in the mammalian germ-
line. Curr. Opin. Genet. Dev. 22, 466–474.
Mansour, A.A., Gafni, O., Weinberger, L., Zviran, A., Ayyash, M., Rais, Y.,
Krupalnik, V., Zerbib, M., Amann-Zalcenstein, D., Maza, I., et al. (2012). The
H3K27 demethylase Utx regulates somatic and germ cell epigenetic reprog-
ramming. Nature 488, 409–413.
Morgan, H.D., Sutherland, H.G., Martin, D.I., and Whitelaw, E. (1999).
Epigenetic inheritance at the agouti locus in the mouse. Nat. Genet. 23,
314–318.
Ng, S.-F., Lin, R.C.Y., Laybutt, D.R., Barres, R., Owens, J.A., and Morris, M.J.
(2010). Chronic high-fat diet in fathers programs b-cell dysfunction in female
rat offspring. Nature 467, 963–966.
Popp, C., Dean, W., Feng, S., Cokus, S.J., Andrews, S., Pellegrini, M.,
Jacobsen, S.E., and Reik, W. (2010). Genome-wide erasure of DNA methyla-
tion in mouse primordial germ cells is affected by AID deficiency. Nature
463, 1101–1105.
Quenneville, S., Verde, G., Corsinotti, A., Kapopoulou, A., Jakobsson, J.,
Offner, S., Baglivo, I., Pedone, P.V., Grimaldi, G., Riccio, A., and Trono, D.
(2011). In embryonic stem cells, ZFP57/KAP1 recognize a methylated hexanu-
cleotide to affect chromatin and DNA methylation of imprinting control regions .
Mol. Cell 44, 361–372.
Rakyan, V.K., Chong, S., Champ, M.E., Cuthbert, P.C., Morgan, H.D., Luu,
K.V.K., and Whitelaw, E. (2003). Transgenerational inheritance of epigenetic
states at the murine Axin(Fu) allele occurs after maternal and paternal trans-
mission. Proc. Natl. Acad. Sci. USA 100, 2538–2543.
Reik, W., Dean, W., and Walter, J. (2001). Epigenetic reprogramming in
mammalian development. Science 293, 1089–1093.
Rowe, H.M., Jakobsson, J., Mesnard, D., Rougemont, J., Reynard, S., Aktas,
T., Maillard, P.V., Layard-Liesching, H., Verp, S., Marquis, J., et al. (2010).
KAP1 controls endogenous retroviruses in embryonic stem cells. Nature
463, 237–240.
Saitou, M. (2009). Germ cell specification in mice. Curr. Opin. Genet. Dev. 19,
386–395.
Saitou, M., Kagiwada, S., and Kurimoto, K. (2011). Epigenetic reprogramming
in mouse pre-implantation development and primordial germ cells.
Development 139, 15–31.
Santos, F., Hendrich, B., Reik, W., and Dean, W. (2002). Dynamic re-
programming of DNA methylation in the early mouse embryo. Dev. Biol. 241,
172–182.
Santos, F., Zakhartchenko, V., Stojkovic, M., Peters, A., Jenuwein, T., Wolf, E.,
Reik, W., and Dean, W. (2003). Epigenetic marking correlates with develop-
mental potential in cloned bovine preimplantation embryos. Curr. Biol. 13,
1116–1121.
Sasaki, H., and Matsui, Y. (2008). Epigenetic events in mammalian
germ-cell development: reprogramming and beyond. Nat. Rev. Genet. 9,
129–140.
Schulz, R., Proudhon, C., Bestor, T.H., Woodfine, K., Lin, C.-S., Lin, S.-P.,
Prissette, M., Oakey, R.J., and Bourc’his, D. (2010). The parental non-equiva-
lence of imprinting control regions during mammalian development and evolu-
tion. PLoS Genet. 6, e1001214. http://dx.doi.org/10.1371/journal.pgen.
1001214.
Seki, Y., Hayashi, K., Itoh, K., Mizugaki, M., Saitou, M., and Matsui, Y. (2005).
Extensive and orderly reprogramming of genome-wide chromatin modifica-
tions associated with specification and early development of germ cells in
mice. Dev. Biol. 278, 440–458.
Seki, Y., Yamaji, M., Yabuta, Y., Sano, M., Shigeta, M., Matsui, Y., Saga, Y.,
Tachibana, M., Shinkai, Y., and Saitou, M. (2007). Cellular dynamics associ-
ated with the genome-wide epigenetic reprogramming in migrating primordial
germ cells in mice. Development 134, 2627–2638.
Sharif, J., Muto, M., Takebayashi, S.-I., Suetake, I., Iwamatsu, A., Endo, T.A.,
Shinga, J., Mizutani-Koseki, Y., Toyoda, T., Okamura, K., et al. (2007). The SRA
protein Np95 mediates epigenetic inheritance by recruiting Dnmt1 to methyl-
ated DNA. Nature 450, 908–912.
Smith, Z.D., Chan, M.M., Mikkelsen, T.S., Gu, H., Gnirke, A., Regev, A., and
Meissner, A. (2012). A unique regulatory phase of DNA methylation in the early
mammalian embryo. Nature 484, 339–344.
Stadler, M.B., Murr, R., Burger, L., Ivanek, R., Lienert, F., Scho
¨ler, A., van
Nimwegen, E., Wirbelauer, C., Oakeley, E.J., Gaidatzis, D., et al. (2011).
DNA-binding factors shape the mouse methylome at distal regulatory regions.
Nature 480, 490–495.
Sugimoto, M., and Abe, K. (2007). X chromosome reactivation initiates in
nascent primordial germ cells in mice. PLoS Genet. 3, e116. http://dx.doi.
org/10.1371/journal.pgen.0030116.
Surani, M.A., Hayashi, K., and Hajkova, P. (2007). Genetic and epigenetic
regulators of pluripotency. Cell 128, 747–762.
Tomizawa, S.I., Kobayashi, H., Watanabe, T., Andrews, S., Hata, K., Kelsey,
G., and Sasaki, H. (2011). Dynamic stage-specific changes in imprinted
differentially methylated regions during early mammalian development
and prevalence of non-CpG methylation in oocytes. Development 138,
811–820.
Vincent, J.J., Li, Z., Lee, S.A., Liu, X., Etter, M.O., Diaz-Perez, S.V., Taylor, S.K.,
Gkountela, S., Lindgren, A.G., and Clark, A.T. (2011). Single cell analysis facil-
itates staging of Blimp1-dependent primordial germ cells derived from mouse
embryonic stem cells. PLoS ONE 6, e28960. http://dx.doi.org/10.1371/journal.
pone.0028960.
Williams, K., Christensen, J., Pedersen, M.T., Johansen, J.V., Cloos, P.A.C.,
Rappsilber, J., and Helin, K. (2011). TET1 and hydroxymethylcytosine in tran-
scription and DNA methylation fidelity. Nature 473, 343–348.
Wu, H., D’Alessio, A.C., Ito, S., Xia, K., Wang, Z., Cui, K., Zhao, K., Sun, Y.E.,
and Zhang, Y. (2011). Dual functions of Tet1 in transcriptional regulation in
mouse embryonic stem cells. Nature 473, 389–393.
Yamaji, M., Seki, Y., Kurimoto, K., Yabuta, Y., Yuasa, M., Shigeta, M.,
Yamanaka, K., Ohinata, Y., and Saitou, M. (2008). Critical function of
Molecular Cell
Resetting DNA Methylation in Primordial Germ Cells
Molecular Cell 48, 849–862, December 28, 2012 ª2012 Elsevier Inc. 861
Prdm14 for the establishment of the germ cell lineage in mice. Nat. Genet. 40,
1016–1022.
Yamazaki, Y., Mann, M.R.W., Lee, S.S., Marh, J., McCarrey, J.R.,
Yanagimachi, R., and Bartolomei, M.S. (2003). Reprogramming of primordial
germ cells begins before migration into the genital ridge, making these cells
inadequate donors for reproductive cloning. Proc. Natl. Acad. Sci. USA 100,
12207–12212.
Yoshimizu, T., Sugiyama, N., De Felice, M., Yeom, Y.I., Ohbo, K., Masuko, K.,
Obinata, M., Abe, K., Scho
¨ler, H.R., and Matsui, Y. (1999). Germline-specific
expression of the Oct-4/green fluorescent protein (GFP) transgene in mice.
Dev. Growth Differ. 41, 675–684.
Yu, M., Hon, G.C., Szulwach, K.E., Song, C.-X., Zhang, L., Kim, A., Li, X., Dai,
Q., Shen, Y., Park, B., et al. (2012). Base-resolution analysis of 5-hydroxyme-
thylcytosine in the mammalian genome. Cell 149, 1368–1380.
Molecular Cell
Resetting DNA Methylation in Primordial Germ Cells
862 Molecular Cell 48, 849–862, December 28, 2012 ª2012 Elsevier Inc.
Molecular Cell, Volume 48
Supplemental Information
The Dynamics of Genome-wide DNA Methylation
Reprogramming in Mouse Primordial Germ Cells
Stefanie Seisenberger, Simon Andrews, Felix Krueger, Julia Arand,
Jörn Walter, Fátima Santos, Christian Popp, Bernard Thienpont,
Wendy Dean, and Wolf Reik
Supplemental Information Inventory
Figure S1. Global methylation profiling, relates to Figure 1.
Figure S2. Promoters with slow demethylating kinetics, relates to Figure 3.
Figure S3. Resistance to demethylation, relates to Figure 4.
Figure S4. Local effect of IAPs on methylation resistance, relates to Figure 4.
Figure S5. Resistant CGIs in various datasets, relates to Figure 4.
Figure S6. Continued hairpin phasing analysis of the LINE1Tf consensus sequence,
relates to Figure 5.
Figure S7. Continued analysis of passive demethylation pathways, relates to Figure 5.
Figure S8. Correlation analysis for transcription and DNA methylation, relates to Figure
6.
Table S1: Details for all Illumina sequencing runs, relates to Figure 1.
Table S2: CGI promoters with >25% methylation in E6.5 epiblast, relates to Figure 3.
Table S3: CGI promoters with >25% methylation in E9.5 PGC, relates to Figure 3.
Table S4: CGI promoters with >25% methylation in E10.5 PGC, relates to Figure 3.
Table S5: CGI promoters with >25% methylation in E11.5 PGC, relates to Figure 3.
Table S6: Transcripts within the pluripotency and meiosis clusters, relates to Figure 6.
Supplemental Information: Figure legends, table legends, extended experimental procedures,
supplemental references, Figures S1 – S8.
10
Figure S1
11
Figure S2
12
Figure S3
13
Figure S4
14
Figure S5
15
Figure S6
16
Figure S7
17
Figure S8
Figure S1. Global methylation profiling
(A) Conversion efficiencies in BS-Seq datasets. Measured were levels of CHH
methylation with 1kb tiling probes across the genome and shown is the proportion of
all probes with varying CHH methylation levels. Note that the largest proportion for
each dataset shows complete conversion indicating that conversion efficiencies are
high. Two independent biological replicates (1 and 2) are shown for the embryonic
samples and two technical replicates of the same BS-Seq library is shown for the J1
samples (highlighted by the asterisk). It is indicated whether the libraries were
sequenced on the GIIAx or HiSeq platform.
(B) Global methylation dynamics. Methylation levels of 5kb tiling probes were
compared pair wise between two data points (indicated below) and shown is the
number of probes with a 25% and 50% increase (top) or decrease (bottom) in
methylation levels. Note that the biggest decrease in methylation levels occurs from
E6.5 epiblast to E9.5 PGCs and that virtually no de novo methylation is observed
outside of the comparison between E13.5 and E16.5 male PGCs.
(C) Example figures for the maternally methylated Kcnq1ot1 DMR (left), the paternally
methylated Rasgrf1 DMR (middle), and the Xist promoter region (right). Each bar
represents a single CG dinucleotide.
Figure S2. Promoters with slow demethylating kinetics
(A) Methylation levels for the entire genome (left) and across the X chromosome (right)
assessed by 5kb tiling probes. Note that the demethylation kinetics for the X
chromosome closely resembles those of the whole genome. Outliers are not shown.
(B) Methylation levels for CGIs across the whole genome.
(C) Example figure for a CGI-containing promoter with late demethylation kinetics.
Methylation levels seem to be retained especially around the CGI within the
promoter. Each bar represents a single CG dinucleotide.
(D) Percentage of promoters that overlap or are nearby (<1kb distance) to confirmed
Zfp57-binding sites (Quenneville et al., 2011). Note that there is an enrichment for
Zfp57 binding sites in late demethylating CGI promoters.
Figure S3. Resistance to demethylation
(A) Methylation levels for IAPs (left), LINEs (middle), and SINEs (right) across the
genome. Note that IAPs retain high levels of methylation whilst LINEs and SINEs
undergo significant methylation reprogramming.
(B) Numbers for non-CGI promoters and CGIs that were selected for > 25% methylation
at E13.5 in male (left) and female (middle) PGCs. The number of elements
overlapping between the lists for male and female PGCs is also indicated (right).
Note that resistant non-CGI promoters are more frequent than resistant CGIs and that
the overlap between male and female PGCs increases in the presence of an IAP
nearby.
(C) Methylation levels of non-CGI promoters that were selected with >25% methylation
in male (left) and female (right) PGCs of E13.5 without the presence of an IAP (top)
or with an IAP nearby (bottom). Note that non-CGI promoters show consistently
higher methylation levels in the presence of an IAP nearby. Outliers are not shown.
Figure S4. Local effect of IAPs on methylation resistance
(A) IAP frequencies found near (< 2kb) or overlapping with CGI-containing or non-CGI
promoters. Note that IAPs are rarely found near CGI-containing promoters and
frequencies increase drastically for non-CGI promoters selected for > 25%
methylation in E13.5 PGCs.
(B) Distance effect of IAPs. Shown are methylation levels for all non-CGI promoters
overlapping with or nearby (1kb – 10kb) an IAP in all datasets. Note that the
resistance effect is highest up to 2kb away from the IAP.
Figure S5. Resistant CGIs in various datasets.
(A) Methylation levels for CGIs in sperm and oocyte RRBS datasets (Smallwood et al.,
2011) selected for > 25% methylation in male (left) and female (right) PGCs of E13.5
without the presence of an IAP (top) or with an IAP nearby (bottom).
(B) VECs were identified as CGIs with > 25% methylation in either male (left) or female
(right) PGCs of E13.5 without the presence of an IAP nearby. Each datapoint
represents methylation levels for one CGI. Datasets included in this analysis were ES
cells (1), blastocyst, sperm, and oocyte (Kobayashi et al., 2012), ES cells (2) (Stadler
et al., 2011), 2-cell embryo and ICM (Smith et al., 2012).
Figure S6. Continued hairpin phasing analysis of the LINE1Tf consensus sequence.
Analysis of PGC samples of various time points. Note that in all datasets, there is a strong
strand bias for meCGs toward either top or bottom strand highly similar to the outcome for
the simulation of passive DNA demethylation (see Figure 5). The number of meCGs is
drastically reduced to E13.5 but the strand bias for meCG is preserved throughout the time
course.
Figure S7. Continued analysis of passive demethylation pathways.
(A) Expression analysis of the DNA methylation machinery, see Figure 5 for more detail.
Note that the expression of the de novo methylation machinery is downreglated in
early PGCs until E16.5, at which point Dnmt3a and Dnmt3L seem to increase in
expression. Tet1 is highly expressed in early PGCs, similar to ES cells, and becomes
trancriptionally downregulated toward E16.5.
(B) Staining for Np95 and Dnmt1 in replicating PGCs of E13.5 male embryos. Gonads
were cultured for 1 hour in the presence of EdU. Note that Np95 localizes
predominantly in the cytoplasm of replicating PGCs whilst Dnmt1 signal is strongest
in the nucleus. This suggests that whilst the staining pattern for Dnmt1 is in line with
the cell cycle stage, the Np95 pattern appears to be cell cycle independent.
(C) Control staining for the Np95 antibody in Np95 KO ES cells. Signal for Np95
staining is virtually absent and this shows high specificity of the antibody.
(D) Sub-cellular localization of Np95 in PGCs throughout development. During the time
course analyzed in this study, we found Np95 consistently localized preferentially to
the cytoplasm. This is the case for both replicating PGCs (E10.5 – E13.5) and cell
cycle arrested PGCs (E13.5 – E16.5). This suggests that uncoupling of the Np95
pattern from the cell cycle may be a common phenomenon throughout PGC
development.
Figure S8. Correlation analysis for transcription and DNA methylation.
(A) Methylation levels for CGI-containing (left) and non-CGI (right) promoters of
transcripts that change in expression levels from E13.5 to E16.5 in male PGCs. Note
that all promoters show increased methylation levels in E16.5 male PGCs and that
only CGI-containing promoters of transcripts with decreased expression show a
slightly lower increase in methylation levels than all other promoters.
(B) Methylation levels of gene bodies for transcripts that are expressed at low or high
levels in E16.5 male PGCs. Only transcripts with a length >25kb were selected and
transcripts within the bottom and top 5% of the RPKM were selected for low and
high expression, respectively. For each gene with multiple transcripts in this set, the
transcript with the highest expression level was selected. Note that gene bodies of
transcripts with low expression at E16.5 in male PGCs show lower methylation
levels than all other transcripts.
(C) Methylation example graphs for genes from the pluripotency cluster. Each bar
represents a single CG dinucleotide.
Supplemental Experimental Procedures
DNA/RNA purification
Genomic DNA and total RNA was extracted either in combination using the Qiagen All Prep
DNA/RNA Micro kit according to the manufacturers’ instructions or if DNA and RNA were
collected separately, Qiagen’s QIAmp kit was used for extracting DNA and Life Sciences’s
Picopure kit for extraction of total RNA according to the manufacturers’ instructions.
Annotations
CGI annotations were used based on pull down experiments (Illingworth et al., 2010).
Promoters were defined as the region -1kb to +100bp of the transcription start site as
annotated in NCBIM37. DMR coordinates were used from E12.5 embryos (Tomizawa et al.,
2011). Consensus sequences were used for the analysis of LINE1Tf (DeBerardinis and
Kazazian, 1999), LINE1A (Schichman et al., 1993), and IAP1 and IAP2 (Qin et al., 2010).
Repeat annotations were extracted from the UCSC RepeatMasker track (mm9 build).
BS-Seq Analysis
Raw sequence reads were trimmed to remove both poor quality calls and adapters using
TrimGalore (www.bioinformatics.babraham.ac.uk/projects/trim_galore/). Remaining sequences
were mapped to the mouse NCBIM37 genome using Bismark (Krueger and Andrews, 2011),
and CG methylation calls were extracted which excluded any duplicate calls from
overlapping read ends of short inserts. Read numbers varied greatly depending on the
sequencing platform used. For consistency, only the sample with the highest read number for
each time point was used in the subsequent analysis. Methylation over a region was
calculated for each CG in the region and then these individual values were averaged to give a
representative value for the region. Only regions where at least 5 CGs had been measured
were taken forward for subsequent analysis. Regions that were covered by a disproportionally
high read number and were most likely mapping artifacts were excluded from subsequent
analysis. For consensus repeat methylation analysis Bismark was used to map all reads
against a consensus sequence for each repeat class and the methylation calls from these
results were analysed directly.
RNA-Seq Analysis
RNA-Seq data was mapped to the mouse NCBIM37 genome assembly using TopHat in
conjunction with gene models from Ensembl release 61. Initial quantitation was made by
counting the number of reads per transcript corrected per million reads (RPKM). This was
adjusted by globally matching the count distributions at the 75th percentile, and then
adjusting counts to have a uniform distribution across all samples. For comparisons of
absolute expression the quantitated value was corrected by the length of the transcript in
kilobases. Differential expression was called by selecting transcripts which changed with a
significance of p<0.05 after Benjamini and Hochberg correction using a null model
constructed from the 1% of transcripts showing the closest average level of observation to
estimate experimental noise. Expression clusters were defined by performing hierarchical
clustering of transcripts based on a Pearson’s correlation across all samples, and selecting
groups which were related R>0.7 and which contained at least 10 transcripts. Alignments of
RNA-Seq data sets against repeat consensus sequences were carried out using Bowtie
(v0.12.8, default options) whereby the fraction of aligning reads was scored. Publically
available datasets for ESC and MLF (Guttman et al., 2010), J1 (Ficz et al., 2011), ES and EB
(Cloonan et al., 2008), and TKO ES cells (Karimi et al., 2011) were included in this analysis
where indicated.
Functional Enrichment Analysis (GO analysis)
Function enrichment was analysed by generating a non-redundant list of genes from an initial
transcript list and analysing this with the DAVID web tool, using a background of all mouse
genes. Groups with a significance of p<0.05 after correction were taken to be significant.
Microarray data analysis
The microarray data from early PGCs (Kurimoto et al., 2008) and ES cells (Vincent et al.,
2011) was analysed with R/Bioconductor using the package affy. The PGC expression data
and the ES data were normalised together using robust multi-array average (RMA) expression
measure. Normalised expression data for specific genes was extracted. Where genes had
multiple probes, the probe with the highest average (mean) expression was selected as it was
reported that this leads to best between-study consistency (Miller et al., 2011).
Supplemental References
Cloonan, N., Forrest, A.R.R., Kolle, G., Gardiner, B.B.A., Faulkner, G.J., Brown, M.K.,
Taylor, D.F., Steptoe, A.L., Wani, S., Bethel, G., et al. (2008). Stem cell transcriptome
profiling via massive-scale mRNA sequencing. Nat Meth 5, 613–619.
DeBerardinis, R.J., and Kazazian, H.H. (1999). Analysis of the promoter from an expanding
mouse retrotransposon subfamily. Genomics 56, 317–323.
Ficz, G., Branco, M.R., Seisenberger, S., Santos, F., Krueger, F., Hore, T.A., Marques, C.J.,
Andrews, S., and Reik, W. (2011). Dynamic regulation of 5-hydroxymethylcytosine in mouse
ES cells and during differentiation. Nature 473, 398–402.
Guttman, M., Garber, M., Levin, J.Z., Donaghey, J., Robinson, J., Adiconis, X., Fan, L.,
Koziol, M.J., Gnirke, A., Nusbaum, C., et al. (2010). Ab initio reconstruction of cell type–
specific transcriptomes in mouse reveals the conserved multi-exonic structure of lincRNAs.
Nat Biotechnol 28, 503–510.
Illingworth, R.S., Gruenewald-Schneider, U., Webb, S., Kerr, A.R.W., James, K.D., Turner,
D.J., Smith, C., Harrison, D.J., Andrews, R., and Bird, A.P. (2010). Orphan CpG islands
identify numerous conserved promoters in the mammalian genome. PLoS Genet 6, e1001134
Karimi, M.M., Goyal, P., Maksakova, I.A., Bilenky, M., Leung, D., Tang, J.X., Shinkai, Y.,
Mager, D.L., Jones, S., Hirst, M., et al. (2011). DNA Methylation and SETDB1/H3K9me3
Regulate Predominantly Distinct Sets of Genes, Retroelements, and Chimeric Transcripts in
mESCs. Stem Cell 8, 676–687.
Kobayashi, H., Sakurai, T., Imai, M., Takahashi, N., Fukuda, A., Yayoi, O., Sato, S.,
Nakabayashi, K., Hata, K., Sotomaru, Y., et al. (2012). Contribution of Intragenic DNA
Methylation in Mouse Gametic DNA Methylomes to Establish Oocyte-Specific Heritable
Marks. PLoS Genet 8, e1002440.
Krueger, F., and Andrews, S.R. (2011). Bismark: a flexible aligner and methylation caller for
Bisulfite-Seq applications. Bioinformatics 27, 1571–1572.
Kurimoto, K., Yabuta, Y., Ohinata, Y., Shigeta, M., Yamanaka, K., and Saitou, M. (2008).
Complex genome-wide transcription dynamics orchestrated by Blimp1 for the specification of
the germ cell lineage in mice. Genes Dev 22, 1617–1635.
Miller, J.A., Cai, C., Langfelder, P., Geschwind, D.H., Kurian, S.M., Salomon, D.R., and
Horvath, S. (2011). Strategies for aggregating gene expression data: The collapseRows R
function. BMC Bioinformatics 12, 322.
Qin, C., Wang, Z., Shang, J., Bekkari, K., Liu, R., Pacchione, S., McNulty, K.A., Ng, A.,
Barnum, J.E., and Storer, R.D. (2010). Intracisternal A particle genes: Distribution in the
mouse genome, active subtypes, and potential roles as species-specific mediators of
susceptibility to cancer. Mol. Carcinog. 49, 54–67.
Quenneville, S., Verde, G., Corsinotti, A., Kapopoulou, A., Jakobsson, J., Offner, S., Baglivo,
I., Pedone, P.V., Grimaldi, G., Riccio, A., et al. (2011). In Embryonic Stem Cells,
ZFP57/KAP1 Recognize a Methylated Hexanucleotide to Affect Chromatin and DNA
Methylation of Imprinting Control Regions. Mol Cell 44, 361–372.
Schichman, S.A., Adey, N.B., Edgell, M.H., and Hutchison, C.A. (1993). L1 A-monomer
tandem arrays have expanded during the course of mouse L1 evolution. Mol. Biol. Evol. 10,
552–570.
Smallwood, S.A., Tomizawa, S.-I., Krueger, F., Ruf, N., Carli, N., Segonds-Pichon, A., Sato,
S., Hata, K., Andrews, S.R., and Kelsey, G. (2011). Dynamic CpG island methylation
landscape in oocytes and preimplantation embryos. Nat Genet 43, 811–814.
Smith, Z.D., Chan, M.M., Mikkelsen, T.S., Gu, H., Gnirke, A., Regev, A., and Meissner, A.
(2012). A unique regulatory phase of DNA methylation in the early mammalian embryo.
Nature 484, 339–344.
Stadler, M.B., Murr, R., Burger, L., Ivanek, R., Lienert, F., Schöler, A., van Nimwegen, E.,
Wirbelauer, C., Oakeley, E.J., Gaidatzis, D., et al. (2011). DNA-binding factors shape the
mouse methylome at distal regulatory regions. Nature 480, 490–495.
Tomizawa, S.I., Kobayashi, H., Watanabe, T., Andrews, S., Hata, K., Kelsey, G., and Sasaki,
H. (2011). Dynamic stage-specific changes in imprinted differentially methylated regions
during early mammalian development and prevalence of non-CpG methylation in oocytes.
Development 138, 811–820.
Vincent, J.J., Li, Z., Lee, S.A., Liu, X., Etter, M.O., Diaz-Perez, S.V., Taylor, S.K.,
Gkountela, S., Lindgren, A.G., and Clark, A.T. (2011). Single Cell Analysis Facilitates
Staging of Blimp1-Dependent Primordial Germ Cells Derived from Mouse Embryonic Stem
Cells. PLoS ONE 6, e28960.